Вы находитесь на странице: 1из 318

Tropical Forestry

Tropical Forestry
Volumes Already Published in this Series

Tropical Forest Genetics by Finkeldey, R., Hattemer, H. 2007

ISBN: 3-540-37396-9

Sampling Methods, Remote Sensing and GIS Multiresource Forest

Inventory by Khl, M. Magnussen, S., Marchetti M. 2006
ISBN: 3-540-32571-9

Tropical Forest Ecology - The Basis for Conservation and Management by

Montagnini, F., Jordan C. 2005
ISBN: 3-540-23797-6
Reiner Finkeldey Hans H. Hattemer

Tropical Forest

With 44 Figures and 19 Tables

Professor Dr. Reiner Finkeldey
Professor Dr. Hans H. Hattemer
Institute of Forest Genetics and Forest Tree Breeding
Bsgenweg 2
D-37077 Gttingen

ISSN: 1614-9785
ISBN-10 3-540-37396-9 Springer-Verlag Berlin Heidelberg New York
ISBN-13 978-3-540-37396-4 Springer-Verlag Berlin Heidelberg New York

Library of Congress Control Number: 2006931487

This work is subject to copyright. All rights are reserved, whether the whole or part of the material is con-
cerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, repro-
duction on microfilm or in any other way, and storage in data banks. Duplication of this publication or parts
thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965, in its
current version, and permission for use must always be obtained from Springer. Violations are liable to pros-
ecution under the German Copyright Law.

Springer-Verlag is a part of Springer Science+Business Media


Springer-Verlag Berlin Heidelberg 2007

The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply,
even in the absence of a specific statement, that such names are exempt from the relevant protective laws and
regulations and therefore free for general use.

Editor: Dr. Dieter Czeschlik, Heidelberg

Desk Editor: Anette Lindqvist, Heidelberg
Production: SPi
Typesetting: SPi
Cover Design: Design & Production, Heidelberg

Printed on acid-free paper 3/3152-HM 5 4 3 2 1 0


This book is intended to provide information on fundamental genetic processes

in tropical forests. It is based on lecture notes by the authors prepared for post-
graduate students, mainly from tropical countries, of the M.Sc. course Tropical
and International Forestry at the Faculty of Forest Sciences and Forest Ecology,
Georg August University Gttingen, Germany. The intended readership is stu-
dents, researchers and practitioners interested in the genetic variation of species,
in particular forest trees, living in complex forest ecosystems in the tropics.
Particular emphasis is placed on the human impact on forest genetic resources
in the tropics.
Readers should be familiar with basics of classical and molecular genetics
such as the structure and function of DNA (double-helix structure, replica-
tion), polypeptide synthesis (transcription and translation), and the transmis-
sion of genetic information during sexual reproduction (Mendels rules).
This knowledge is easily available from recently published genetics textbooks;
we recommend the book by Griffiths et al. (2000).
The development of biochemical and molecular marker techniques and
their application to species of tropical forests have greatly improved our knowl-
edge of genetic variation patterns of wild plants in the tropics during the last
two decades. Many examples in this book are based on gene marker techniques.
We discuss the results of selected molecular studies and their implications for
the development of sustainable management strategies for forest genetic
resources. However, we refrain from providing details on laboratory techniques
and statistical data analyses. Readers should consult the cited references or the
book by Weising et al. (2005) if they require additional information in this
regard. No advanced knowledge concerning modern molecular or statistical
methods for the analysis of genetic variation patterns is necessary to follow the
argumentation of this book.
Comprehensive data on DNA sequences are becoming available for more
and more plants, including forest trees. The first project aimed at sequencing
the full genome of a forest tree (Populus trichocarpa) has recently been com-
pleted (Brunner et al. 2004). However, only very few genomic data are currently
VI Preface

available for tropical forest trees; tropical forest genetics is still in a prege-
nomic era. Thus, the focus of this book in on population genetics processes in
natural and managed tropical forests rather than on genomics and related,
newly emerging fields of research in forest genetics.


The authors are grateful to Oliver Gailing for critically reading Sect. 12.5.4 to
Barbara Vornam and Martin Ziehe for assistance in literature search and in
word processing, and to Wolfgang Tambour for preparing Figures 7.1, 7.2, 7.5,
and 12.4.

July 2006 Reiner Finkeldey

Hans H. Hattemer

1 Introduction Genetics of Tropical Forests 1

Part A Genetic Processes in Tropical Forests 3

2 Population Genetics an Overview 5

2.1 Introduction 5
2.2 The Population 5
2.3 Variation at Gene Loci 6
2.3.1 The Molecular Basis of Genetic Variation 6 Molecular Markers 7 Biochemical Markers Isozymes 12
2.3.2 The Gene As a Unit of Heredity 13
2.3.3 The Mode of Inheritance 14
2.3.4 Definition and Classification of Gene Markers 15
2.4 Genetic Structures Within Populations 16
2.4.1 Allelic and Genotypic Structures 16
Example 2.1: Genetic Structures at an Isozyme Gene Locus in
Dalbergia sissoo 17
2.4.2 Variation at Uniparentally Inherited Markers 18
Example 2.2: Diversity of cpDNA Haplotypes in D. sissoo 18
2.5 Evolution and Evolutionary Factors 19
2.6 Phenotypic Variation 21
2.7 Recommended Literature 22

3 Genetic Variation of Tropical Forest Plants 23

3.1 Introduction 23
3.2 Genetic Inventories 23
3.3 Measurement of Genetic Variation 24
3.3.1 Genetic Variation Within Populations 24 Genetic Multiplicity 25 Genetic Diversity 25
3.3.2 Genetic Differentiation Among Populations 27 Genetic Distances 27 Genetic Differentiation 27
VIII Contents

3.4 Genetic Variation in Tropical Forest Species General Trends 28

3.5 Case Studies on Genetic Variation Patterns 31
Example 3.1: Genetic Variation at Isozyme Gene Loci
in Natural Populations of Acacia auriculiformis 31
Example 3.2: Genetic Differentiation Among Populations
of Eugenia dysenterica from the Brazilian Cerrado 33
Example 3.3: Genetic Variation of Swietenia macrophylla
Across the Brazilian Amazon 33
Example 3.4: Genetic Variation of Endangered Australian
Species of the Genus Fontainea Assessed with RAPD Markers
and by DNA Sequences 34
Example 3.5: Genetic Variation of Shorea leprosula and
Shorea parvifolia in Indonesia Assessed at AFLP Loci 35
Example 3.6: Genetic Variation of Cedrela odorata at cpDNA 37
3.6 Recommended Literature 39

4 Sexual and Asexual Reproduction in Tropical Forests 41

4.1 Introduction 41
4.2 Sexual Reproduction 41
4.2.1 Sexual Types and Sexual Systems 42
4.2.2 Sexual Systems 44
4.2.3 Sexual Structures and Sexual Function 47
4.3 Asexual Reproduction 49
4.3.1 Vegetative Reproduction 50
4.3.2 Apomixis 51
4.4 Recommended Literature 52

5 Gene Flow and Migration 53

5.1 Introduction 53
5.2 Gene Flow Through Pollen 54
5.2.1 Pollination and Fertilization 54
5.2.2 Pollen Vectors in Tropical Forests 54 Pollination by Wind (Anemogamy) 54 Pollination by Animals (Zoogamy) 55
5.2.3 Pollen Dispersal Assessed by Marker-Based Studies 60 Dispersal of Rare Alleles 60
Example 5.1: Gene Dispersal in a Clonal Seed Orchard
of Teak (Tectona grandis) 60 Paternity Analyses 61
Example 5.2: Gene Flow Through Pollen in Neobalanocarpus
heimii and Dipterocarpus tempehes (Dipterocarpaceae) 62
Example 5.3: Pollen Dispersal in Three Neotropical
Tree Species with Different Spatial Distribution Patterns on
Barro Colorado Island 63
Contents IX Model-Based Estimates of Pollen Dispersal 64

Example 5.4: Pollen Dispersal in Dinizia excelsa (Fabaceae) 64
5.2.4 Efficiency of Pollen Vectors for Gene Flow 65
5.3 Migration of Seeds 65
5.3.1 Seed Vectors 65 Abiotic Seed Dispersal 65 Biotic Seed Dispersal 66
5.3.2 Efficiency of Seed Dispersal 66
5.4 Long-Distance Gene Flow and Migration in Tropical Forest Species 67
5.5 Recommended Literature 68

6 Mating Systems 69
6.1 Introduction 69
6.2 Random Mating and Panmixis 70
6.2.1 Heterogeneity of Pollen Allele Frequencies 70
6.2.2 Panmixis and HardyWeinberg Structures 72
6.3 Selfing and Outcrossing Rates 73
6.3.1 Estimates of Selfing Rates Based on Rare Alleles 74
Example 6.1: Estimates of Selfing Rates in Teak (Tectona grandis)
Populations 74
6.3.2 Estimates of Selfing Rates Based on a Mixed Mating Model 76
6.3.3 Estimates of Selfing Rates Based on Nonmaternal Alleles 76
6.4 Inbreeding and Inbreeding Depression 78
6.4.1 Genetic Consequences of Inbreeding 78
6.4.2 Inbreeding Depression 80
Example 6.2: Inbreeding Depression in Eucalypts 80
6.5 Incompatibility and Self-Sterility 82
6.5.1 Incompatibility 82
6.5.2 Self-Sterility 84
6.6 Environmental Effects on Mating 84
6.7 Recommended Literature 85

7 Adaptation and Coevolution 87

7.1 Introduction 87
7.2 Physiological and Evolutionary Adaptation 87
Example 7.1: Viability Selection During Early Life Stages
of Platypodium elegans 89
7.3 Species Interactions and Coevolution 90
Example 7.2: Figs and Their Pollinators 91
7.4 Recommended Literature 97

8 Phylogenies and Evolution Above the Species Level 99

8.1 Introduction 99
8.2 The Evolution of Species Diversity in the Tropics 99
X Contents

8.2.1 Species Diversity in Tropical Forests 99

8.2.2 Evolution Above the Species Level 101
8.3 Molecular Phylogenies 102
Example 8.1: A Molecular Phylogeny of Indonesian
Dipterocarpoideae 103
Example 8.2: Evolution of the Genus Inga 104
8.4 Recommended Literature 106

Part B Applications of Genetics to Tropical Forestry 109

9 Fragmentation of Forests 111

9.1 Introduction 111
9.2 The Genetic Status of Fragmented Tree Populations 113
Example 9.1: Genetic Impact of Fragmentation on a Rain
Forest Canopy Tree 116
Example 9.2: Connectivity of Population Fragments of a
Medium-Sized Dry-Forest Tree 119
9.3 Genetic Preconditions for Restoration and Persistence 123
9.4 Recommended Literature 127

10 Genetic Aspects of the Management of Natural

Tropical Forests 129
10.1 Introduction 129
10.2 Selective Logging 130
10.2.1 Effective Population Density After Logging 131
10.2.2 The Question of Dysgenic Selection 135
10.3 Natural Regeneration 136
10.4 Genetic Aspects of the Manipulation of Dynamic
Processes in Natural Forests 137
10.5 Genetic Aspects of Sustainability in Natural Tropical Forests 140
10.6 Recommended Literature 146

11 Provenance Research 147

11.1 Introduction 147
11.2 Definitions 148
11.3 Historical Development 150
11.4 Provenance Tests 151
11.4.1 Types of Field Experiments in Provenance Research 153
11.4.2 Traits Studied in Field Trials 155
11.4.3 Design, Conduct, and Analysis of Provenance Trials 156 Collection of Material for Provenance Trials 156 Production of Planting Stock 159 Experimental Design 159
Contents XI Choice of Experimental Sites 161 Number and Distribution of Locations of an
Experimental Series 162 Observation Period of Field Experiments 163
11.5 Provenance Differentiation and Geographic Variation Pattern 163
11.6 Choice of Provenances in Tropical Forestry 165
Example 11.1: Provenances of Eucalyptus camaldulensis
and Their Growth Performance 169
Example 11.2: International Provenance Trials in Teak
(Tectona grandis) 170
11.7 Recommended Literature 172

12 Domestication and Breeding of Tropical Forest Trees 173

12.1 Introduction 173
12.2 Domestication 174
12.3 Genetic Controlledness of Phenotypic Traits 176
12.3.1 Individuals Related by Descent 176
12.3.2 Pair Comparisons in the Field 177
12.3.3 Response to Natural Viability Selection 179
12.4 Linear Model of Genetic Effects on a Phenotypic Trait 179
12.4.1 Phenotypic Trait Expressions 179
12.4.2 Genetic Variance Components and Heritability 182
12.5 Estimation of Breeding Parameters; Progeny Testing 183
12.5.1 Progenies of Open-Pollinated Trees 183
12.5.2 Progeny Tests After Controlled Pollination 184
12.6 Methods of Selection 186
12.6.1 Selection of Plus Trees 187
12.6.2 Selection in Progeny Tests 188
12.6.3 Selection of Clones 189
12.6.4 Marker-Assisted Selection 189
12.7 Heterosis Breeding in Forest Trees 192
Example 12.1: Breeding Eucalypts in Aracruz, Brazil 194
12.8 Propagation of Breeding Products 196
12.8.1 Clonal Seed Orchards 196
Example 12.2: Clonal Seed Orchards of Teak (Tectona grandis) 198
12.8.2 Seedling Seed Orchards 201
12.8.3 Mass Multiplication of Clones 204
12.9 Multiple Population Breeding 205
12.10 Breeding Strategy 207
12.11 Genetic Consequences of Domestication and Breeding 209
12.11.1 General Considerations 209
12.11.2 Monitoring of Breeding Projects by Use of Genetic Markers 210
12.12 Recommended Literature 211
XII Contents

13 Genetic Aspects of Plantation Forestry in the Tropics 213

13.1 Introduction 213
13.2 Plantations of Exotic Tree Species 215
Example 13.1: The Origin of Early Plantations of
Acacia mangium in Sabah 217
13.3 Plantations of Indigenous Species 218
13.4 Basic and Reproductive Material 220
13.5 Production and Collection of Seed 222
13.5.1 Seed Production Areas 222
13.5.2 Provenance Resource Stands 223
13.5.3 Seed Orchards 224
13.6 Collection and Storage of Seed 226
13.7 Planting Stock Production 227
13.7.1 Seedlings 227
13.7.2 Clonal Multiplication 228
13.8 Establishment and Development of Plantations 229
13.9 Natural Regeneration of Plantations 230
13.10 Use of Breeding Products 231
13.11 Recommended Literature 235

14 Conservation of Genetic Resources in Tropical Forest Trees 237

14.1 Introduction 237
14.2 Development of Tree-Conservation Genetics 239
14.3 Defining Priorities 240
14.4 Conservation Objectives 241
14.4.1 Objective 1: Preservation of the Potential for Particular
Trait Expressions 241
14.4.2 Objective 2: Preservation of Maximum Variation 242
14.4.3 Objective 3: Preservation of Adaptability 243
14.5 Selection of Genetic Resources 243
14.5.1 Inventory of Genetic Marker Loci 245
14.5.2 Inventory of Adaptive Trait Expressions and Adaptive Markers 247
14.6 Conservation Methods 250
14.6.1 Dynamic Conservation in Situ 251
14.6.2 Dynamic Conservation ex Situ in Man-Made Forests 253
14.6.3 Conservation of Seeds in Gene Banks 255
14.6.4 Vegetative Propagation for the Conservation of Forest
Genetic Resources 256
14.6.5 Biotechnological Methods of Preservation 256
14.7 Regeneration of Forest Genetic Resources 257
Example 14.1: Conservation of the Genetic Resources of
Pinus merkusii in Thailand 259
14.8 Programs for the Protection of Forest Genetic Resources 264
14.9 Recommended Literature 265

References 267
Index 307
Introduction Genetics of Tropical Forests 1

. . . the significance of forest genetics lies not only in squeezing extra gain out of mar-
ginal investments, but more importantly in understanding the genetics of forests, in
appreciating what forest ecosystems are and how they operate, and in better stewards of
forest resources (Namkoong 1989).

The importance of forest ecosystems for environmental protection, global bio-

diversity, and human welfare can hardly be overestimated. Genetic resources are
widely recognized as the basis of biodiversity and the main components of
ecosystems. Conservation and utilization of genetic resources are crucial aspects
of sustainable forest management practices. However, our current understand-
ing of forest genetic resources is poor at best. Genetic variation patterns have
been studied for a small fraction of forest trees only. Evolutionary processes
resulting in variation within species are difficult to study for long-living organ-
isms such as forest trees, and human impact on forest genetic resources is often
Thus, it is hardly surprising that misconceptions about forest genetic
resources and implications of human impact on genetic resources are com-
monplace. This holds specifically true for tropical forests, which deserve par-
ticular attention for the following reasons. First, tropical forests are centers of
biological diversity at least at the level of species diversity. Second, human
impact on forest ecosystems is particularly severe in the tropics. Alterations of
land use, notably deforestation, endanger forest genetic resources in most trop-
ical countries. Common silvicultural practices such as selective cutting systems
based on target diameters threaten genetic resources in managed forests. On
the other hand, many tropical countries have launched ambitious reforestation
and afforestation programs. The choice of reproductive material used for plan-
tation establishment is a matter of vital importance for the success of refor-
estation programs. Utilization of genetic resources (tree improvement) plays
an important role in this context.
This book is aimed at familiarizing readers with the fundamental principles
related to genetic resources in tropical forests. We will concentrate our discussion
on forest trees since trees are by definition the main structural and functional
2 C HAPTER 1 Introduction Genetics of Tropical Forests

components of forest ecosystems. However, other organisms will also be consid-

ered. This holds in particular for animals interacting with trees by distributing
their genetic information through the dispersal of pollen or seeds. Priority is
given to species of considerable economic importance, mainly trees grown in
plantations for wood production (e.g., eucalypts) and keystone species serving
vital functions in particular ecosystems (e.g., figs in many Asian forests).
The analysis of patterns of genetic variation is in the center of our discus-
sion. The dynamics of genetic structures of selected forest taxa without human
interference is discussed in Part A of this book. The role of the evolutionary
factors for shaping genetic variation patterns within and among populations is
described with emphasis on the reproduction of forest trees (sexual systems,
gene flow, and mating systems).
Part B is devoted to the human impact on forest genetic resources in the
tropics. Both deliberate changes of genetic structures owing to tree improve-
ment and genetic implications of other human activities, including forest
destruction, fragmentation, and silvicultural practices, are considered.
We describe mutual dependencies between environmental conditions and
evolutionary change in primary and secondary forests as well as in plantations.
Genetic processes are not seen in isolation but are discussed in an ecological
context following the fundamental work of Klaus Stern, the founder of the
authors institute (Stern and Roche 1974). Genetic variation is regarded as a key
aspect of biodiversity and as the basis for evolutionary adaptability. The devel-
opment of management and conservation strategies for natural and man-made
forests in the tropics requires the consideration of genetic aspects.
The main focus of the book is on mutual dependencies between environ-
mental conditions in the tropics and genetic variation patterns of selected
species, in particular forest trees. The species richness of many tropical forests
has numerous implications in this regard.
Genetic Processes in Tropical Forests Part A

The enormous species richness of tropical forests is mirrored by multiple

strategies to ensure the survival of species in tropical forests. Survival implies
adaptedness to prevailing environmental conditions, but also, at least in the
long run, adaptability to environmental change. Genetic variation is the basis
for evolutionary adaptive change. In this part our current understanding of the
dynamics of genetic variation in time and space is reviewed for species living
in tropical forests.
Basics of population genetics and the concept of gene marker loci are briefly
introduced to readers not familiar with the analysis of genetic variation (Chap. 2).
Surprisingly high levels of genetic variation within populations have been
reported for the majority of, but not all, tropical forest tree species (Chap. 3). The
maintenance of genetic variation during reproduction (Chap. 4) is due to partic-
ular features of gene flow and migration mechanisms (Chap. 5), and the mating
system (Chap. 6) of tropical forest trees. Changes of genetic structures due to
selection are difficult to monitor for long-living organisms growing in complex
forest ecosystems, but are crucial for patterns of adaptive variation (Chap. 7).
Finally, evolution and genetic differentiation above the species level, i.e., within
genera, families, or even higher taxonomic units, are discussed (Chap. 8).
Population Genetics an Overview 2

Population geneticists spend most of their time doing one of two things: describing the
genetic structure of populations or theorizing on the evolutionary forces acting on
populations. On a good day, these two activities mesh and true insights emerge
(Gillespie 1998).


Population genetics provides the framework for an understanding of the

dynamics of genetic structures or evolution. The following considerations are
neither specific for nor confined to forest species. Most aspects mentioned in
this chapter are as relevant for tropical forest species as for any other higher-
plant or animal species. However, the population genetics approach taken
throughout this book to discuss genetic processes in tropical forests requires a
basic understanding of the most important fundamentals concerning genetic
structures and their dynamics in time and space, and introduces the concept of
genetic markers as well as the most important molecular and biochemical gene

The Population

The genetic information or genotype of an organism does not change through-

out its lifetime. It is necessary to observe and to compare genetic information
of many plants or animals in order to describe genetic variation and its dynam-
ics in time. The fundamental units for studies of genetic variation are collec-
tives of individuals exchanging their genetic information among each other for
the sexual production of the next generation. Such units of (sexual) reproduc-
tion are defined as populations.
A species rarely consists of one single population. It is usually composed of
several more or less isolated populations. Isolation means the absence of mating
6 C HAPTER 2 Population Genetics an Overview

events between plants (or animals) from different populations in this context.
However, an assessment of reproductive isolation usually requires a fairly
detailed understanding of the reproduction system of a species. Furthermore,
the definition of a population is operational only if an occasional (but rare)
exchange of genetic information among populations is taken into considera-
tion. Thus, interfertile, spatially clustered plants are often regarded as belonging
to the same population, while spatially separated plants are regarded as parts of
different populations.
The delineation of populations is particularly difficult for immobile organ-
isms with poorly described spatial distribution patterns and largely unknown
means of gene dispersal through seed and pollen. In most cases, it is far from
obvious which trees in a species-rich tropical forest belong to the same popu-
lation and how big tree populations are. This has far-reaching consequences for
levels of genetic variation of tropical forest trees, as will be discussed in later
Analyses of spatial patterns of genotypes without a priori information on the
delineation of populations or partially isolated subpopulations have recently
been proposed as an alternative to the traditional approach of dividing the
overall variation into components within and among demes or populations
(Diniz-Filho and Telles 2002; Manel et al. 2003). Spatially explicit analyses of
genetic variation patterns might be particularly informative for plant species
occurring in low density in tropical forests; however, currently available infor-
mation is mainly confined to comparatively small areas (Ng et al. 2004).

Variation at Gene Loci

The Molecular Basis of Genetic Variation

The genetic information of all organisms is stored as a sequence of the four

bases or nucleotides adenine, thymine, cytosine, and guanine in their DNA. It
is multiplied by the process of replication and translated to a sequence of
amino acids by the processes of transcription (from DNA to messenger RNA,
mRNA) and translation (from mRNA to polypeptides) (Griffiths et al. 2000).
A cistron is a part of the DNA coding for a particular polypeptide, for exam-
ple, an enzyme. From this point of view, the variation at a particular gene or
gene locus can be assessed by observing differences in the DNA sequence
within a cistron. The total number of genes is unknown for any tropical for-
est plant; however, it may be assumed to be at least on the order of more than
25,000 estimated for the first fully sequenced flowering plant, the herbaceous
2.3 Variation at Gene Loci 7

Arabidopsis thaliana (The Arabidopsis Genome Initiative 2000). The number

of genes of the first sequenced tree species (Populus trichocarpa) is estimated to
be even higher (http://genome.jgi-psf.org/Poptr1/Poptr1.home.html).
The observation of variation at the most basic level of DNA sequences is rou-
tine for tropical trees just as for other organisms, but it is time-consuming and
costly if sequences of hundreds of organisms need to be compared. To date, only
a few population studies have been conducted based on the observation of vari-
ation of DNA sequences within a single or a few closely related tropical forest
species (but see, e.g., Ishiyama et al. 2003). Sequencing is more frequently
applied to clarify taxonomic subdivisions and phylogenies (Chap. 8).
Numerous alternatives to sequencing are available. The choice of the most
appropriate method depends on several factors, including available resources
and the purpose of a study. A comprehensive overview of currently available
molecular tools to assess patterns of genetic variation in plants has been given
by Weising et al. (2005).
Molecular Markers

The most important methods to assess patterns of genetic variation within and
among species rely on the amplification of short DNA fragments by means of
polymerase chain reaction (PCR). PCR allows short fragments of the DNA of
a target organism to be selectively amplified (multiplied) for further analyses.
The sequence of two short oligonucleotides (primers), each usually between
ten and 25 base pairs in length, is decisive for the amplified DNA region. The
following marker types, all based on the PCR method, have gained particular
importance in the study of genetic variation patterns of tropical forest trees.

(Partial) Gene Sequences and Single Nucleotide Polymorphisms

Selective primers can be designed to amplify the DNA coding for a particular
gene (cistron). DNA regions translated into a polypeptide (exons) and non-
translated regions (introns) of structural genes as well as regulatory genes can
be analyzed. It is possible to sequence the amplified PCR products, often after
a cloning step. For example, Kamiya et al. (2005) investigated the phylogeny of
the species-rich genus Shorea (Dipterocarpaceae) on the basis of a (partial)
sequence of the PgiC gene (Example 8.1).
An investigation of genetic variation is particularly rewarding at gene loci
with known or putative function and a directly observable effect on phenotypic
traits responsible for the adaptation to particular environmental conditions
(candidate genes). Populations can be screened for single nucleotide polymor-
phisms (SNPs) without the need for repeated sequencing (Morin et al. 2004).
8 C HAPTER 2 Population Genetics an Overview

The characterization of expressed sequence tags, studies of molecular varia-

tion at the level of expressed genes, and analyses of variation at SNPs are
expected to greatly improve our understanding of adaptive genetic variation
and its dynamics in space and time. However, these recently developed marker
types are not discussed in detail owing to the lack of published reports for trop-
ical forest species (but see Sect. 7.2).

Microsatellites (Simple Sequence Repeats)

Microsatellites or simple sequence repeats (SSRs) are short DNA fragments
of usually only two or three base pairs in length which are repeated several
times in a particular location of the DNA (Fig. 2.1). SSRs can be studied by the
development of primers in conserved DNA regions flanking the microsatel-
lite. The development of primers is rather costly and time-consuming
(Squirrell et al. 2003). Nuclear microsatellites developed for a particular
species can only be transferred to closely related species, usually within the
same genus.
Nuclear microsatellites are important genetic markers owing to their usually
high variability and codominant mode of inheritance (Sect. 2.3.2, Chap. 3).
The variation of microsatellite fragment sizes is due to a different number of
repeated motifs resulting in slightly different sizes of amplified fragments
(Fig. 2.2). The availability of highly variable markers is particularly useful for
analyses of the mating system and gene flow (Chaps. 5, 6). Microsatellites have
been developed for a number of tropical forest tree species, including
Pithecellobium elegans (Chase et al. 1996a), Swietenia humilis (White and Powell
1997), Gliricidia sepium (Dawson et al. 1997), Shorea curtisii (Ujino et al. 1998),
Caryocar brasiliense (Collevatti et al. 1999), Neobalanocarpus heimii (Iwata et al.
2000), and Prosopis spp. (Mottura et al. 2005).
Microsatellite motifs are not only found in DNA of the nucleus, but also in
DNA of chloroplasts (cpDNA; Sect. 2.3.2). Single nucleotide repeats are partic-
ularly common in cpDNA. Universal primers are available to investigate varia-
tion at cpSSRs in many different species of angiosperms (Weising and Gardner
1999) and gymnosperms (Vendramin et al. 1996).




Fig. 2.1. Three different DNA sequences of Prosopis chilensis containing the microsatel-
lite repeat motif AC in nine (lower and middle) or ten (upper) copies (underlined).
A adenine, C cytosine, G guanine, T thymine. (From Mottura, unpublished)
2.3 Variation at Gene Loci 9

195 200 205 210 215 220 225 230 235 240 245 250 255 260









Fig. 2.2. Length variation for eight Prosopis spp. trees at a microsatellite gene locus
(Mottura et al. 2005) separated in an automated sequencer (ABI 3100). Each line refers
to one tree. Different sizes of amplified fragments are visualized as peaks at different
positions in the chromatogram. Trees with two amplified fragments are heterozygous;
tree 5 is homozygous with only one amplified fragment. (From Mottura, unpublished)

Random Amplified Polymorphic DNA and Amplified Fragment

Length Polymorphism
Random amplified polymorphic DNA (RAPD; Newbury and Ford-Lloyd 1993;
Fig. 2.3) and amplified fragment length polymorphism (AFLP; Vos et al. 1995;
Fig. 2.4) are genetic fingerprinting techniques resulting in more or less complex
DNA patterns. Unlike most other techniques including microsatellites no previ-
ous sequence information is necessary for RAPD and AFLP studies. The presence
(+ or 1) or absence ( or 0) of a DNA fragment of a particular size is scored for
each sample plant investigated. This information is the basis for a matrix of size
n m (n is the number of plants investigated; m is the number of DNA fragments
observed) with elements 1 (fragment present) or 0 (fragment absent). RAPDs
and AFLPs are usually interpreted as dominant markers (see later). No
10 C HAPTER 2 Population Genetics an Overview

Fig. 2.3. Low variation at a random amplified polymorphic DNA marker in teak
(Tectona grandis). (From Finkeldey, unpublished)

information on the function of a particular DNA fragment or its mode of inher-

itance is available; hence, RAPDs and AFLPs are anonymous markers.
The RAPD technique is a simple molecular tool to assess genetic variation
within and among populations which has been widely used for tropical forest
plants such as Gliricidia spp. (Chalmers et al. 1992), Cedrela odorata (Gillies
et al. 1997), Caesalpinia echinata (Cardoso et al. 1998), Prunus africana
(Dawson and Powell 1999), and many others. The reproducibility of the RAPD
technique has been under dispute (Rabouam et al. 1999).
The AFLP method is more demanding with regard to laboratory equipment
and experience, but allows more DNA fragments to be investigated from a sin-
gle PCR reaction, and shows higher reproducibility in comparison with
RAPDs. The potential of the AFLP technique to assess genetic variation
patterns in tropical trees was recently reviewed by Kremer et al. (2005). AFLP
studies to assess genetic variation within tropical tree species were conducted,
for example, for Moringa oleifera (Muluvi et al. 1999), Calycophyllum
spruceanum (Russel et al. 1999), and Acer skutchii (Lara-Gomez et al. 2005).

Restriction Fragment Length Polymorphism

A simple method to observe variation among DNA fragments from different
plants is to digest the fragments into smaller pieces by restriction enzymes.
Restriction enzymes cut DNA at a particular sequence. These recognition
sequences are usually between four and six base pairs in length. The restric-
tion fragment length polymorphism (RFLP) technique is based on the use of
restriction enzymes to observe genetic variation. For example, nuclear RFLPs
2.3 Variation at Gene Loci 11

220 222 224 226 228 230 232








Fig. 2.4. Variation at amplified fragment length polymorphisms in seven Shorea

parvifolia trees after separation in an automated sequencer (ABI 3100). Only fragments
in the range 220232 bp are shown. The fragment with a size of 220 bp (arrows) is
diagnostic for the population Berau (Borneo), since it was not observed in six other
populations. (From Cao, unpublished)

were developed for Acacia mangium to assess genetic diversity and differentiation
among populations (Butcher et al. 1998) and with the objective to incorporate
marker-based approaches in breeding programs (Butcher et al. 2000; Butcher
More frequently, the RFLP techniques is combined with a PCR step and is
used to assess variation of cpDNA. For example, Tsumura et al. (1996) and
Indrioko et al. (2006) (Fig. 2.5) analyzed phylogenetic relationships among
Southeast Asian dipterocarps using the PCR-RFLP technique (Example 8.1),
and Cavers et al. (2003) observed strong genetic differentiation among five
cpDNA haplotypes of Cedrela odorata in Central America (Example 3.6).
12 C HAPTER 2 Population Genetics an Overview

M 1 2 3 4 5 6 7 8 9 10 11 12 Fig. 2.5. Polymerase chain reaction restric-

tion fragment length polymorphisms of
Indonesian Dipterocarpaceae after amplifi-
cation with the primer pair TrnLTrnF and
restriction with TaqI. M marker (size
standard), 1 Hopea celebica, 2 Vatica rassak,
3 V. pauciflora, 4 H. bancana,
5 Dipterocarpus oblongifolius, 6 S. javanica,
7 V. granulata, 8 Upuna borneensis,
9 S. mecistopteryx, 10 D. rigidus,
11 S. mecistopteryx, 12 H. grifithii. (From
Indrioko, unpublished)
Biochemical Markers Isozymes

The sequence of nucleotides of the DNA is converted to a sequence of amino

acids of polypeptides by the processes of transcription and translation.
Enzymes are the most important group of polypeptides catalyzing all kinds of
biochemical reactions in the metabolism; thus, there is a close relation between
enzymes and controlling genetic information according to the one gene one
polypeptide hypothesis. Isozymes are enzymes with similar or even identical
functions. Electrophoresis of isozymes followed by biochemical staining is a
simple way to observe differences with regard to their electric load and/or spa-
tial structure (Rothe 1994; Fig. 2.6). It is often possible to directly relate the
variation observed by isozyme electrophoresis to genetic variation (Bergmann
and Hattemer 1998). A formal inheritance study (Sect. 2.3.2) is strongly rec-
ommended prior to the use of isozymes as genetic markers.
Isozyme gene loci became the first widely used markers to assess patterns of
genetic variation of tropical forest species at single gene loci. The ease of isozyme
2.3 Variation at Gene Loci 13

Tree: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
A2A2 A2A2 A1A2 A2A2 A1A2 A1A2 A1A2 A2A2
A1A2 A2A2 A1A2 A1A2 A1A1 A1A2 A1A2 A2A2

Fig. 2.6. Zymogram showing variation of 16 Dalbergia sissoo trees from Nepal at the
enzyme system alcohol dehydrogenase. The variation is controlled by a gene locus
ADH-A. The inferred genotypes of the respective trees are indicated. (From Pandey,

inventories and the comparatively low costs have made them useful tools for
population genetics studies, although they were often substituted by various
types of molecular DNA markers during the last decade.

The Gene As a Unit of Heredity

Genetics as the science of heredity dates back to the experiments of the

Austrian monk Gregor Mendel (1866), if not earlier. The definition of a gene
as a unit of heredity was introduced in the early twentieth century, long before
the role of DNA as the material basis of genetic information was recognized.
A gene in this sense is identified by the observation of segregation within prog-
enies of particular parents.
The basic idea for the identification of a gene as a unit of heredity has
remained unchanged since the experiments Mendel (1866). He investigated
segregation ratios in progenies after controlled pollination at simple morpho-
logical traits such as the color of petals in peas.
If a plant is heterozygous at a controlling gene locus, it will transmit only one
of the two different alleles to a particular progeny (the terms heterozygosity and
allele are described later). Thus, segregation is expected in progenies if at least
one parent was heterozygous at a controlling gene locus. It is possible to com-
pare the observed segregation in a sample of progenies after controlled pollina-
tion with an expectation based on the assumption of a simple Mendelian
inheritance of the trait. The controlling gene or the gene locus is identified, if the
14 C HAPTER 2 Population Genetics an Overview

differences between the observed and expected values are not significant
(Hattemer 1991). Suitable statistical tests are, for example, described by Sokal
and Rohlf (1998, p. 686 ff.; c2 test or goodness-of-fit test, G test).
The formal identification of a genetic marker locus by an observation of seg-
regation is possible only for environmentally stable, genetic traits. It is also
advisable for the identification of gene loci based on the observation of bio-
chemical or molecular markers. For example, Moran and Bell (1983) observed
segregation at the isozyme system GOT in progenies of Eucalyptus regnans after
controlled pollination and found no evidence to reject the hypothesis of the
genetic control of the variation by a single gene locus.
Forest trees, in particular those of the tropics, are extraordinarily difficult
to cross owing to their large size, short viabilities of pollen, unknown repro-
ductive biology, and other obstacles. Thus, the identification of a gene locus
by inheritance studies is preferentially conducted on the basis of alternative
approaches such as the observation of segregation in the haploid megagame-
tophyte (the endosperm) of gymnosperms (Bergmann 1974) or the obser-
vation of segregation in the progenies of putatively heterozygous seed trees
after open pollination (Gillet and Hattemer 1989). The first method has been
used to identify enzyme gene loci in Pinus merkusii from Thailand
(Changtragoon and Finkeldey 1995b); the latter approach was used to clarify
the inheritance of isozymes in Pterocarpus indicus from the Philippines
(Finkeldey et al. 1998).

The Mode of Inheritance

The analysis of variation patterns at environmentally stable, genetic traits

(Chap. 3) requires an understanding of the transmission of the trait from
parents to progenies and from one generation to the next. Two considerations
are of particular importance in this context: the transmission of genetic infor-
mation from both parents or from a single parent only, and the impact of a
single or two alleles at a gene locus on the observed trait expression (the
Most of the DNA in a plant cell is located in the nucleus. Nuclear DNA
(nDNA) is inherited by both parents. For diploid organisms, one set of chro-
mosomes is transmitted by each of the two parents to their common progenies;
thus, nDNA is typically biparentally inherited. In addition, DNA of plants is
located in two types of plastids: mitochondria (mitochrondrial DNA, mtDNA)
and chloroplasts (cpDNA). DNA in plastids, both mitochondria and chloro-
plasts, is usually transmitted by a single parent to its progenies (uniparental
inheritance). mtDNA is mainly inherited from the female or seed parent only
2.3 Variation at Gene Loci 15

(maternal inheritance). The same holds for cpDNA in angiosperms. However,

cpDNA of gymnosperms is typically paternally inherited, i.e., only the genetic
information of the pollen parent is transferred to a progeny. cpDNA was
investigated in most studies on genetic variation of tropical forest plants based
on uniparentally inherited markers (Example 3.6). Maternally inherited
mtDNA was studied in several animal species such as Central American frogs
(Crawford 2003) and the giant panda in China (Lu et al. 2001).
The different types of genetic information at any biparentally inherited gene
locus are called alleles. Since two sets of chromosomes are inherited (one from
the seed parent, the other from the pollen parent), each nuclear gene is repre-
sented in two copies in each progeny of a diploid species. Polyploids are not
discussed in this introduction. A tree is homozygous (e.g., A1A1) at a particu-
lar locus A, if two identical alleles (both A1) were inherited from its parents.
A heterozygous tree (e.g., A1A2) received two different alleles (A1 from the seed
parent and A2 from the pollen parent or vice versa) from its parents.
An allele is dominant, if its possession in a single copy is sufficient to express
a particular phenotype. If A1 is dominant, the phenotypic expressions of trees
with the genotypes A1A1, A1A2, A1A3, etc. are identical. The other alleles (A2 and
A3 in the example) are recessive. If it is possible to distinguish all heterozygotes
(e.g., A1A2) from all homozygotes (A1A1 and A2 A2), the alleles are defined as
codominant. A simple observation of genetic structures (Sect. 2.4) is only
possible at codominant marker loci.

Definition and Classification of Gene Markers

On the basis of the previous considerations, it is possible to define a gene

marker in different contexts:

A gene marker (synonyms gene locus and marker locus) in a wide sense is
an environmentally stable trait; thus, the variation is determined by genetic
factors only.
A gene marker in a narrow sense is an environmentally stable, biparentally
inherited, codominant trait.

Although many different traits may be regarded as gene markers in a wide

sense, only the molecular and biochemical markers briefly described in Sect.
2.3.1 have considerable practical importance to assess genetic variation pat-
terns of tropical forest species. A rough classification of important, currently
available marker types with regard to their mode of inheritance and their vari-
ability is presented in Table 2.1.
16 C HAPTER 2 Population Genetics an Overview

Table 2.1. Classification of genetic markers in a broad sense commonly used in studies
on genetic variation of tropical forest species with regard to their mode of inheritance
and their variability
Variation Uniparental Anonymousa Codominantb

High Chloroplast simple sequence Amplified Microsatellites

repeats; sequences of fragment length (simple sequence
chloroplast DNA polymorphisms repeats)
Moderate Polymerase chain reaction Random Isozymes
to low restriction fragment length amplified
polymorphisms of chloroplast polymorphic
DNA and mitochondrial DNA DNAs
Mode of inheritance usually unknown; interpretation as dominant markers
Gene markers in a narrow sense

Genetic Structures Within Populations

The final result from a laboratory study to assess the genetic constitution of a
plant at a particular biochemical or molecular gene marker system is a specific
pattern, which can be used:

To infer a genotype in case of gene markers in a narrow sense (e.g., isozymes,

To identify a haplotype in the case of uniparentally inherited markers (e.g.,
To assign a genetic fingerprint usually consisting of many anonymous marker
loci to the plant (e.g., AFLPs)

Genetic structures are frequency distributions of such genetic types in popula-

tions. The most important frequency distributions to assess levels of genetic
variation within and among populations are allelic and genotypic structures.

Allelic and Genotypic Structures

Genetic variation is assessed by an investigation of all plants or a (random)

sample of plants from a population. At a gene marker in a narrow sense, the
genotypes of these plants are known, and it is possible to calculate the (relative)
frequency of a particular genotype Ai Aj as
Pij = Nij /N, (2.1)
2.4 Genetic Structures Within Populations 17

where Nij is the (absolute) frequency of genotype AiAj in the population (or
sample) and N is the population size (or sample size).
The frequency vector of all (relative) frequencies of genotypes is defined as
the genotypic structure of the population at the respective marker gene locus.
It obviously holds that // Pij = 1.
i j

Each genotype of a diploid species consists of two alleles; thus, the total
number of alleles in a population (or a sample) is twice the number of plants
investigated, and the frequency of a particular allele Ai can be calculated as
p i = Ni / 2N, (2.2)
where Ni is (absolute) frequency of allele Ai in the population (or sample) and
N is the population size (or sample size). The respective allele Ai is counted
twice in case of homozygous plants AiAi.
The frequency vector of all (relative) frequencies of alleles is defined as the
allelic structure of the population at the respective marker gene locus. Again,
it holds that / p i = 1.
If the genotypic structure of a population is known, it is easily possible to
compute the allelic structure at the respective gene locus. The frequency of
allele Ai (pi) can be computed by calculating the sum of the frequency of the
corresponding homozygote (Ai Ai ) and half of the frequencies of all heterozy-
gotes where the respective allele occurs (Ai Aj , Ai Ak , . . .). However, the calcula-
tion of genotypic structures based on allele frequencies requires additional
information or assumptions (e.g., Sect. 6.1.2).

Example 2.1: Genetic Structures at an Isozyme Gene Locus in Dalbergia sissoo

The observed variation at the isozyme gene locus ADH-A is illustrated for 16
Dalbergia sissoo trees in Fig. 2.6. Three different genotypes were observed. Six trees
(nos. 1, 3, 4, 7, 15, and 16) show the homozygote genotype A2A2, tree 10 is the only
tree with genotype A1A1, and the remaining nine trees exhibit the heterozygous
genotype A1A2. Thus, the genotypic structure of the sample is as follows:
Eleven of the 32 observed alleles (each tree has two alleles) are A1, the
remaining 21 alleles are of type A2. The allelic structure may also be computed
as follows:
p1 = P11 + 0.5P12 = 0.0625 + 0.5625/2 = 11/32 = 0.34375,
p2 = P22 + 0.5P12 = 0.375 + 0.5625/2 = 21/32 = 0.65625.
18 C HAPTER 2 Population Genetics an Overview

Variation at Uniparentally Inherited Markers

The uniparental inheritance of cpDNA and mtDNA implies the absence of

recombination among genes. Since cpDNA and mtDNA are usually transmit-
ted without any changes in maternal or, in the case of cpDNA of gym-
nosperms, paternal lineages, it is possible to interpret variation patterns
observed at different loci together and to assign particular cpDNA or mtDNA
haplotypes to different plants. The interpretation of genetic diversity within
and among populations (Chap. 3) is conventionally based on these haplotypes
rather than on variation at particular loci.

Example 2.2: Diversity of cpDNA Haplotypes in Dalbergia sissoo

Six to ten seeds from randomly selected, different seed trees were investigated
in each of ten populations of D. sissoo in Nepal. The inventory was conducted
in five regions, each represented by one natural population and a neighboring
plantation. Variation was detected by the PCR-RFLP technique (Sect. 2.3.1) by
restriction of the amplified fragment Trn K1/K2 with the enzymes AluI and
RsaI, and by amplification of the chloroplast microsatellites (cpSSRs) ccmp6
and ccmp7 (Weising and Gardner 1999). A total of eight haplotypes with two
or three haplotypes in each population were observed (Table 2.2). Haplotypes
13 were only observed in natural populations, while haplotypes 48 occurred
mainly in plantations (Pandey et al. 2004).

Table 2.2. Variation of chloroplast DNA haplotypes in five natural populations and
five plantations of Dalbergia sissoo in Nepal. (Adapted from Pandey et al. 2004)
Region Population N Chloroplast DNA haplotypes
1 2 3 4 5 6 7 8

Natural populations
A Hetauda N 10 9 1
B Shivapur 10 8 2
C Hattisar 6 5 1
D Godawari 10 4 6
E Pipariya 10 9 1
A Hetauda P 10 4 6
B Surai 10 1 7 2
C Thakurdwara 10 4 3 3
D Attaria 10 6 4
E Shuklaphanta 10 5 5
N sample size
2.5 Evolution and Evolutionary Factors 19

Evolution and Evolutionary Factors

Todays biodiversity is the result of the biological evolution which started at

least three and a half billion years ago on earth. The number of existing species
is still unknown, since a large proportion of the species living on our planet
have not yet been described. The majority of unknown species are insects liv-
ing in the canopy of tropical forests (Erwin 1988). Current estimates of the
global species diversity vary from approximately two million to 30 or even 50
million species (Erwin 1997).
The long history of evolution on earth does not preclude the study of its
effects in short periods. Evolution is permanently ongoing and it is possible to
observe its effects even in long-living organisms such as forest trees.
Furthermore, the majority of evolutionary change is within existing species
and does not immediately result in the creation of new taxa.
The main principle of evolution was recognized and described by Charles
Darwin (1859): Biological evolution is based on heritable variation within
species or populations. Darwin recognized the key role of natural selection in
this context. Fundamental observations, which led to the development of the
selection theory, were made both by Darwin and Alfred Russel Wallace, who
independently developed a similar theory, after extended studies in tropical
areas (Lefvre 1984; Sect. 8.2).
Our advanced knowledge about the process of inheritance allow us to define
evolution more precisely today that during Darwins era. Evolution is a change
of the genetic (allelic or genotypic) structure of a population at one or several
gene loci. Only changes improving the adaptedness of a population to particu-
lar environmental conditions were regarded as evolution by Dobzhansky et al.
(1977); however, Kimura (1983) proposed including all changes of genetic
structures including those at selectively neutral loci into the concept.
Evolutionary factors are the causes of changes of genetic structures (Hedrick
2000). The significance of evolutionary factors for the dynamics of genetic
structures of tropical forest plants is the main topic in later chapters of this
book; thus, only a brief description is given here:

Mutation is the prerequisite for any genetic variation. Mutations are ran-
dom changes of genetic information of an organism which might affect a
single nucleotide (a spontaneous change of a single nucleotide resulting in
a SNP; Sect. 2.3.1), the chromosomal structure (e.g., fissions, translocations,
inversions), or the number of genes (e.g., duplications, polyploidy). Most
mutations are detrimental or neutral for an organism. Only few advanta-
geous mutations increase the fitness, i.e., the capacity to produce offspring.
20 C HAPTER 2 Population Genetics an Overview

Mutations are rare events; mutation rates are estimated to be in the range
between 106 and 108 for most genes. However, the frequency of mutations
is not uniform. For example, considerably higher estimates of mutation
rates were reported for microsatellites (Goldstein and Pollock 1997).
Gene flow and migration increase the genetic multiplicity of a population,
if genes which were previously restricted to certain populations migrate to
other populations. Transport of pollen (gene flow in a narrow sense) and
seeds (migration) are the main processes involved in the dispersal of genes
for most plant species (Chap. 5). By definition, gene flow and migration
among populations are rare events.
The mating system is decisive for the fusion of male and female gametes.
Gene flow through pollen is an obvious prerequisite for mating between
two seed plants. Thus, gene flow and the mating system are closely con-
nected and are the two most important aspects of the reproduction system
of a plant species. The mating system decides on the combination of alle-
les to genotypes. The genotypic structure of the progeny generation is
influenced by the mating system, which primarily does not cause changes
of allelic structures. Inbreeding due to selfing or mating among relatives is
a particularly important aspect of the mating system (Chap. 6).
Random fluctuations of genetic structures are described as genetic drift.
Genetic drift is a consequence of limited population size and becomes
stronger in small populations. The strength of genetic drift can be described
as a function of the population size, although its final outcome is unpre-
dictable. Alleles, in particular rare alleles, are likely to completely disappear
from populations as a result of genetic drift, and previously polymorphic
(variable) gene loci may become monomorphic (fixed). A sudden reduction
of the population size for one or several generations is often described as a
genetic bottleneck. Bottlenecks result in reduced genetic variation even
after later population expansion owing to the effects of genetic drift while
populations are small. Genetic differentiation among partially or completely
isolated populations is expected to increase as a result of genetic drift.
Selection is a consequence of the differing abilities of organisms to produce
offspring (fitness). Selection does not increase multiplicity, but often con-
tributes to the maintenance of genetic diversity by stabilizing polymor-
phisms, for example, owing to a selective advantage of heterozygous
genotypes. In general, selection results in an improved adaptedness (Chap.
7). The basis of selection is the different survival of genotypes (viability
selection) and the different contributions of genotypes to the formation of
a progeny generation (fertility selection).

Population geneticists are concerned with the effect of these evolutionary

factors on patterns of genetic variation within and among populations
2.6 Phenotypic Variation 21

(Gillespie 1998). Genetic structures result from complicated interactions

between the different evolutionary factors in natural and managed populations.
The role of different evolutionary factors for the creation and maintenance of
the extraordinary species richness of tropical forests has been widely discussed
in the past. A high frequency of selfing and restricted gene flow and migration
resulting in small populations, which become differentiated from each other
mainly owing to the effects of genetic drift, have been proposed as the main
forces for the evolution of the enormous species richness in tropical forests by
Fedorov (1966). This view was challenged by Ashton (1969), who argued in
favor of more efficient means of gene flow through seed and pollen than previ-
ously thought. The evolution of species diversity is promoted by selective
differentiation in partially isolated subpopulations which eventually develop
reproductive barriers. The latter proposition is greatly supported by maker-
based assessments on genetic variation patterns in tropical forest species
(Chap. 3) and on the reproduction system of tropical forest plants (Chaps. 5, 6).

Phenotypic Variation

The vast majority of directly observable traits of trees are an outcome of the
interaction between the genetic constitution of a plant at a large number of gene
loci (genotype, G) and the environmental conditions (E). Thus, it holds for the
phenotype (P) that P= GE, where symbolizes the interaction between envi-
ronmental and genotypic effects, which often differs from simple additivity of
both components.
Phenotypic traits deserve particular interest if they are important for the
adaptation of plants to their environment (adaptive traits) or for the value of a
tree (economic traits). Adaptive traits such as the tolerance against biotic
(pests) or abiotic (e.g., draught) stress and economic traits such as volume
growth are controlled by both genetic and environmental factors for most
plant species. The analysis of phenotypic traits in a breeding context will be
discussed in Chap. 12 in more detail. It is also described for forest trees in other
textbooks (Wright 1976; Zobel and Talbert 1984; Williams et al. 2002).
The majority of the gene markers described in Sect. 2.3.1 are unlikely to have
strong, recognizable effects on adaptive traits or traits of economic importance.
They are regarded as neutral or nearly neutral markers with regard to an
adaptation to the environment; however, this does not preclude their applica-
tion to monitor adaptive processes (Ziehe et al. 1999; Sect. 7.2).
In line with the definition of genetic structures, the frequency distribution
of particular phenotypes is defined as the phenotypic structure of a popula-
tion. Direct and indirect effects of the environment on the phenotypic struc-
ture are illustrated in Fig. 2.7. Apart from the direct impact of the environment
22 C HAPTER 2 Population Genetics an Overview

Mode of gene structure
Migration Genotypic

Viability selection

Mating Environmental influence

system on mating

Population size
Fertility selection
Gene flow Allelic

Fig. 2.7. The environmental impact on phenotypic structures. (Adapted from Hattemer
and Mller-Starck 1990)

on the phenotype (P = GE), there are numerous impacts of the environment

on genotypic structures. The environmental conditions influence mutation
rates, are crucial for population sizes and, hence, the importance of genetic
drift, cause changes of allelic structures owing to fertility selection and geno-
typic structures owing to viability selection (Chap. 7), and have an impact on
the mating system (Chap. 6). Thus, manifold human alterations of environ-
mental conditions do not only directly effect phenotypes of forest plants, but
also change the genotypic structure of populations.

Recommended Literature

The molecular basics of genetics is covered in many recently published text-

books, such as the one written by Griffiths et al. (2000). A useful and detailed
description of genetic markers and their application to study plant genetics is
given by Weising et al. (2005). Altukhov and Salmenkova (2002) provide a com-
prehensive survey on currently available markers to study genetic variation of
plants. The need for inheritance studies in order to identify gene markers was
emphasized by Hattemer (1991). The basics of population genetics are covered
by Hedrick (2000). The book of Altukhov (2006) represents another introduc-
tion to population genetics putting genetic variation in an ecological context.
Genetic Variation of Tropical Forest Plants 3

. . . the main difficulty one faces in a subject like population genetics (. . .) is not the
mathematics itself, or the biology itself: it is how to fit them together (p. ix in Maynard
Smith 1998).


A main application of genetic markers is the observation of genetic variation

patterns within and among populations. Variation at phenotypic traits is influ-
enced both by genetic and by environmental factors. Thus, it is impossible to
conclude on levels of genetic variation at single loci on the basis of the obser-
vation of phenotypes at adaptive traits or traits of economic importance,
although an assessment of variance components in field trials allows the degree
of genetic control of a phenotypic trait to be estimated (p. 49 ff. in Eriksson and
Ekberg 2001). The application of gene markers has greatly enhanced our
understanding of genetic variation of tropical forest plants.

Genetic Inventories

Genetic inventories aim at describing extant patterns of genetic variation

within a single or, more frequently, within and among several populations.
These patterns are the result of the impact of the evolutionary factors acting on
populations in the past (Sect. 2.5). On the other hand, the evolutionary adap-
tive potential of populations depends on genetic variation patterns (Sect. 7.2).
For practical reasons, it is only possible to observe a small and very often pre-
sumably not representative set of marker gene loci for a given species. Variation
at most of the tens of thousands of gene loci remains unexplored. Thus, an
assessment of evolutionary adaptive potentials of populations is likely to be
unreliable if it is based on the small set of loci investigated only. This holds irre-
spective of the type of marker loci investigated. Still, the limited variation
24 C HAPTER 3 Genetic Variation of Tropical Forest Plants

observed at marker loci is often useful to elucidate the evolutionary past of

populations and species. This argument is particularly valid if genetic struc-
tures have been shaped by evolutionary factors acting on all genes in princi-
pally the same way. For example, a bottleneck in the evolutionary past of a
population is expected to reduce genetic variation due to drift at all loci, and
particular types of mating systems, for example, strong inbreeding, result in
characteristic genotypic structures (Chap. 6). On the other hand, it is difficult
to assess the impact of selection since its effects are restricted to the selected
and associated loci (Sect. 7.2). The conclusions drawn on the evolutionary past
of populations and species can be used to assess their evolutionary adaptive
potential (Finkeldey and Mtys 1999).

Measurement of Genetic Variation

A detailed description of the various methods to analyze genetic variation

patterns is beyond the scope of this book. The book by Nei (1987) should be
consulted for a comprehensive account. Berg and Hamrick (1997) and Weising
et al. (2005, p. 207ff.) present brief overviews. Only simple measures and
methods frequently used to analyze patterns of genetic variation of tropical for-
est species are briefly introduced in this section. Statistical problems related to the
estimation of genetic variation based on samples rather than on full inventories
of all plants are not discussed in detail. The focus is on the population genetics
interpretation of allelic and genotypic structures at gene markers in a narrow
sense, i.e., nuclear, codominant markers such as microsatellites or isozymes.
The most complete and comprehensive information on genetic variation
patterns at a particular marker locus is the genotypic and allelic structures at
that locus (Sect. 2.4); however, it is difficult to compare variation levels among
loci or populations on the basis of these genetic structures. Thus, the genotypic
and allelic frequency distributions are condensed to measures characterizing
particular aspects of variation within or among populations. This is particu-
larly important if the aim is to investigate and to compare genetic variation at
several gene loci. The gene pool is the average or mean of genetic variation esti-
mates from several single loci.

Genetic Variation Within Populations

The genetic variation within populations is characterized by measures of gene-

tic multiplicity, which only take into account the occurrence of different
genetic types, but not their frequencies, and measures of genetic diversity, which
also consider the frequency of different genetic types (alleles or genotypes).
3.3 Measurement of Genetic Variation 25
Genetic Multiplicity

Proportion of Polymorphic Loci

The simplest estimate of genetic multiplicity is the proportion of polymorphic
(variable) loci (PPL). It is calculated by dividing the number of polymor-
phic gene loci (PL) by the number of all loci investigated, which also includes
loci showing no variation (monomorphic loci, ML):
PPL = PL / (PL + ML). (3.1)
The PPL is often reported as a percentage after multiplication by 100. Some
authors regard a locus as polymorphic only if the frequency of the most fre-
quent allele is below 95% (or 99%). Other authors regard a locus as polymor-
phic if at least two types (irrespective of their frequencies) occur.

Number of Alleles per Locus and Allelic Richness

Another simple measure of genetic multiplicity is the average number of alle-
les per gene locus. It is calculated by counting the number of alleles (ni) at each
of L gene loci, summing over all loci investigated, and dividing this total num-
ber of alleles by the number of loci L.
/n i

Number of alleles per locus = i = L1 (3.2)

Since the probability of observing a (rare) allele in a population strongly

depends on sample size, comparisons of the average number of alleles among
populations possibly lead to false conclusions if the number of sampled plants
differs among populations. Effects of different sample sizes are taken into con-
sideration by the computation of the allelic richness (Petit et al. 1998; Comps
et al. 2001).
Genetic Diversity

The genetic diversity of a population does not only depend on the number of
genetic types (alleles or genotypes), but also on their (relative) frequencies (pi , pij).

Effective Number of Alleles

The effective number of alleles at a gene locus (Crow and Kimura 1970) has
also been defined as allelic diversity v (Gregorius 1978) of a population at that
locus. If n is the total number of alleles at a locus and pi are the relative fre-
quencies of the alleles, the effective number of alleles v is calculated as follows:
26 C HAPTER 3 Genetic Variation of Tropical Forest Plants

1#v= 1 # n. (3.3)
/p i

The effective number of alleles v reaches its maximum value for a given n if
the frequencies of all alleles are identical (1/n). In this case, v = n. The allelic
diversity v decreases with increasing frequency of one allele and consequently
decreasing frequencies of the remaining alleles.
The effective number of alleles of the gene pool is the harmonic mean of the v
at single loci.
The effective number of alleles from Example 2.1 is
v= 1 = 1 = 1 . 1.82.
p 12 + p 22 0.34375 2 + 0.65625 2

Expected Heterozygosity
The expected heterozygosity He is the most frequently used measure of genetic
diversity within populations. The expectation is based on the assumption
that alleles are randomly combined to genotypes, i.e., that HardyWeinberg
structures are realized (Sect. 6.2.2). In this case, it is possible to calculate the
frequency Pii of a particular homozygote with genotype Ai Ai as pi2 (pi is the fre-
quency of allele Ai). The frequency of the expected homozygosity is calculated
by summing over all alleles (pi2), the remainder being the expected heterozy-
gosity. Accordingly, the expected heterozygosity is calculated as
H e = 1 - / p i2 . (3.4)
Gene pool values of the expected heterozygosity are the arithmetic mean of
single locus values.
The expected heterozygosity from Example 2.1 is
H e = 1 - (p 12 + p 22 ) = 1 - (0.34375 2 + 0.65625 2 ) . 0.451.

Observed Heterozygosity
A simple measure of genetic diversity is the proportion of all heterozygotes
from all plants (homozygotes and heterozygotes) investigated in a population:
n n n
Ho = / / Pij = 1 - / Pii . (3.5)
i = 1j = 1 i = 1
i ! j

This observed heterozygosity equals the expected heterozygosity if the popu-

lation exhibits HardyWeinberg structures. The gene pool value of the observed
heterozygosity is calculated as the arithmetic mean of Ho from single loci.
The observed heterozygosity of the Dalbergia sissoo population in Example 2.1
is Ho= P12 = 1 (P11 + P 22) = 0.5625.
3.3 Measurement of Genetic Variation 27

It is also possible to calculate the degree of individual heterozygosity of a

single plant by dividing the number of heterozygous gene loci by the number
of observed loci.

Genetic Differentiation Among Populations

Genetic differences between two populations are quantified by measures of

genetic distance; the differences between an arbitrary number of populations
are measured as genetic differentiation.
Genetic Distances

Out of a large choice of commonly used genetic distances only the minimum
genetic distance according to Nei (1973) is mentioned here. For two popula-
tions k and l it is defined as
d min (k , l ) = 12 / 7 p i (k) - p i (l )A ,

where pi(k) and pi(l ) are the frequencies of the ith allele in populations k and l,
respectively. This distance reaches its minimum value if the allelic structures of
both populations are identical [pi(k)=pi(l) for all alleles i], and has a maximum
value of 1 if the populations are fixed on different alleles. More commonly used
distance measures such as Neis standard genetic distance are described by Nei
If more than two populations are investigated, it is possible to compute a
matrix with all pairwise genetic distances, and to use this information as a basis
for the illustration of genetic differentiation patterns by means of cluster analy-
ses (e.g., Fig. 3.4). Populations are grouped on the basis of their genetic distances
by different algorithms. A simple and commonly used method for grouping in
population genetics is the unweighted pair-group method with arithmetic
means (UPGMA) method (Sneath and Sokal 1973). Similar methods are used
to illustrate the relatedness of higher taxa in molecular phylogenies (Sect. 8.3).
Genetic Differentiation

The standard method to assess genetic differentiation among an arbitrary

number of populations is the calculation of GST (or FST) (Nei 1973). The total
gene diversity in a population composed of several subpopulations is calcu-
lated as HT. HT is the expected heterozygosity (Eq. 3.4) with the frequencies of
28 C HAPTER 3 Genetic Variation of Tropical Forest Plants

alleles (pi) calculated on the basis of their occurrence in the total population.
The average gene diversity within populations (HS) is calculated as mean gene
diversity (He) within the subpopulations. It can be shown that the difference
HT HS equals the average minimum genetic distance of Nei (see before)
between all population pairs (DST = HT HS) (Finkeldey 1994). GST (or FST) is
the relative contribution to the total gene diversity which is due to differences
among populations: GST = FST = DST /HT = (HT HS)/HT.

Genetic Variation in Tropical Forest Species General Trends

Levels and patterns of genetic variation at single loci depend on numerous fac-
tors. Apart from species-specific or population-specific variability, the type of
the marker(s) investigated is of crucial importance. An assessment of genetic
variation levels is meaningful only if results are related to those of comparable
studies. Most available data on patterns of genetic variation of tropical forest
plants were based on isozyme inventories, although the number of studies
using molecular markers is constantly increasing.
In view of the multitude of tropical forest plants it is hardly surprising that
levels and patterns of genetic variation considerably vary among species
(Table 3.1). In tropical forests, plant species with low, intermediate, and high
levels of genetic diversity within populations are encountered.
Mean levels of genetic variation of woody plants, mainly forest trees, from
different climatic regions were compared by Hamrick et al. (1992) (Table 3.2).
At isozyme gene loci, the measures of genetic multiplicity (PPL; number of
alleles per locus) are on average slightly lower for woody plants from the trop-
ical region as compared with temperate and boreal species. However, this does
not hold for the average expected heterozygosity (He) as the most commonly
used measure of genetic variation. Thus, there is no evidence for fundamental
differences with regard to levels of genetic variation of tropical and temperate
trees, even though the majority of tropical trees occur in low density, while
many temperate and boreal species form large, pure stands.
The impact of density on levels of genetic variation was studied in more
detail by Hamrick and Murawski (1991). Genetic variation of 16 uncommon
tropical forest tree species (density less than 0.5 trees per hectare) occurring on
Barro Colorado Island, an artificial island in the Panama Channel, was assessed
at isozyme gene loci, and compared with variation levels of 16 common
species (density more than four trees per hectare). Variation levels within
uncommon species varied from low (PPL = 14.3%, He = 0.026 for Tricanthera
gigantea) to high (e.g., PPL = 76%, He = 0.257 for Myrospermum fructescens).
The average variation of the uncommon species is lower than the variation of
more common species (Table 3.3); thus, there is evidence for losses of genetic
3.4 Genetic Variation in Tropical Forest Species General Trends 29

Table 3.1. Examples of tropical tree species showing low and high levels of genetic
variation within populations, respectively
Species Country/region Marker type He Reference

Low variation
Acacia mangium Australia, Papua Isozyme 0.017 Moran et al. (1989a)
New Guinea
Acer skutchii Mexico SSR 0.112 Lara-Gomez et al. (2005)
Moringa oleifera India, East Africa AFLP 0.067 Muluvi et al. (1999)
Pentaclethra Costa Rica Isozyme 0.074 Hall et al. (1994a)
Pinus merkusii Thailand Isozyme 0.058 Changtragoon and
Finkeldey (1995a)
High variation
Acacia albida West Africa Isozyme 0.442 Joly et al. (1992)
(=Faidherbia albida)
Calycophyllum Peru AFLP 0.249 Russel et al. (1999)
spruceanum 0.349
Gliricidia sepium Central America Isozyme 0.238 Chamberlain et al. (1996)
Pinus caribaea Caribbean Islands Isozyme 0.26 Zheng and Ennos (1999)
Pterocarpus Thailand Isozyme 0.246 Liengsiri et al. (1995)
Shorea leprosula Malaysia Isozyme 0.406 Lee et al. (2000)
Swietenia macrophylla Brazil SSR 0.781 Lemes et al. (2003)
He average expected heterozygosity within populations, SSR simple sequence repeat, AFLP amplified
fragment length polymorphism

Table 3.2. Average genetic variation at isozyme gene loci in long-lived woody species
from different climatic regions. (Adapted from Hamrick et al. 1992)
Climatic zone No. of species Average no. PPL (%) A/L He
of populations

Boreal temperate 26 8.2 82.5 2.58 0.206

Temperate 122 12.2 63.5 2.27 0.166
Temperate tropical 5 16.6 62.2 1.89 0.169
Tropical 38 21.3 57.9 1.87 0.191
PPL proportion of polymorphic gene loci, A/L average number of alleles per locus, He expected

variation due to small reproduction effective population sizes resulting in

genetic drift for very uncommon tree species in tropical forests. However, the
average variation of uncommon tree species on Barro Colorado Island is
higher than the average variation of 486 plant species investigated at isozyme
gene loci and is roughly comparable to the average variation found for 115
long-lived woody perennial species, mainly trees (Table 3.3).
Most of the genetic variation within a species is usually found within popu-
lations, with only a minor fraction of the overall diversity found among
30 C HAPTER 3 Genetic Variation of Tropical Forest Plants

Table 3.3. Mean genetic variation at isozyme gene loci for 16 uncommon (rare;
density less than 0.5 trees per hectare) and 16 common (density more than four trees
per hectare) tree species on Barro Colorado Island in comparison with average
variation of 486 plants and 115 long-lived woody species. (Adapted from Hamrick and
Murawski 1991)
Group No. of species Average no. PPL (%) He
of gene loci

Uncommon species 16 25.8 41.8 0.142

Common species 16 27.2 60.9 0.211
All plant species 486 16.5 34.2 0.113
Long-lived woody plants 115 17.0 50.0 0.149
He expected heterozygosity

populations. The mean population differentiation (GST) for 37 tropical woody

plant species included in a survey by Loveless (1992) was 0.109; thus, on aver-
age, less than 11% of the total genetic variation is due to differences among
populations. This result is not confined to the majority of tropical forest plants,
but has also been found in plants, in particular forest trees, of other climatic
regions (Austerlitz et al. 2000). Differentiation among populations at isozyme
gene loci is even lower for temperate (GST = 0.092) and boreal (GST = 0.038)
species as compared with 26 tropical species (GST = 0.119) included in the
review by Hamrick et al. (1992). Low differentiation at single gene loci does not
preclude considerable differentiation with regard to genetic components of the
variation observed at phenotypic traits of adaptive or economic significance
(Kremer et al. 1997; Chap. 11).
In summary, the following conclusions can be drawn from the multitude of
marker-based studies on genetic variation of tropical forest trees:

Forest trees are among the genetically most variable of all organisms. This
result was first obtained for a few species of the temperate and boreal zone
and later confirmed for tropical tree species. The genetic variation within
populations at isozyme gene loci is similar for temperate and tropical tree
species (Table 3.2).
Several important tropical tree species are characterized by low levels of
genetic variation within populations, while other species are highly diverse
(Table 3.1). It is not possible to reliably predict genetic variation levels for
a particular species on the basis of its life-history characteristics.
Endemic species are, on average, less variable within populations than
species with a wide distribution range (Loveless 1992).
Uncommon species, i.e., those occurring in very low population densities
in tropical forests, are, on average, less variable than species occurring in
3.5 Case Studies on Genetic Variation Patterns 31

higher density (Table 3.3). However, the expected heterozygosity is high for
many tree species even if the population density is low.
The degree of genetic differentiation among populations (GST) is slightly
higher for tropical forest tree species than for tree species of temperate
forests. Tropical tree species with abiotic means of seed dispersal (baro-
chorous and anemochorous species) show, on average, a much higher dif-
ferentiation among populations (GST = 0.138) than zoochorous species
(GST = 0.050) (Sect. 5.3). Seed dispersal by animals is usually very efficient
and results in small genetic differentiation among populations of zoo-
chorous tree species (Loveless 1992).

Case Studies on Genetic Variation Patterns
Since predictions of genetic variation patterns are very unreliable for a partic-
ular species, experimental investigations on the distribution of genetic
variation within and among populations are mandatory for a detailed under-
standing of variation patterns of a target species. Six case studies were selected
as examples to illustrate genetic variation patterns in tropical forest species out
of a wide variety of published reports. The case studies were chosen since they
represent different:

Marker types: isozymes, microsatellites, random amplified polymorphic

DNA (RAPDs), amplified fragment length polymorphisms (AFLPs), varia-
tion of chloroplast DNA (cpDNA)
Regions of the tropics (continents and climatic conditions)
Species with regard to succession stage (pioneerclimax species), dispersal
mechanism of seeds, abundance (rare or common), and other life-history

The focus is on genetic variation patterns and variation levels in forest tree pop-
ulations without or with only minor human impact. Genetic effects of human
disturbance of tropical forest ecosystems will be discussed in later chapters.

Example 3.1: Genetic Variation at Isozyme Gene Loci in Natural Populations

of Acacia auriculiformis
Acacia auriculiformis A. Cunn. Ex Benth. is one of the most important Acacia
species for plantation establishment worldwide. The natural distribution of the
species is confined to Australia, Papua New Guinea, and a few Indonesian
32 C HAPTER 3 Genetic Variation of Tropical Forest Plants

islands (Fig. 3.1). A genetic inventory was conducted at 22 isozyme gene loci
coding for 17 enzyme systems in 18 natural populations of the species from
Papua New Guinea and Australia (Wickneswari and Norwati 1993). Clear
differences with regard to the level of genetic variation measured as expected
heterozygosity were observed (Fig. 3.1). Populations from Papua New Guinea
revealed the highest level of genetic variation (average He = 0.133 or 13.3%),
which is comparable to the average for long-lived woody plant species
(Table 3.3). Intermediate levels of genetic diversity were observed in popula-
tions from Queensland, and low levels in populations from the Northern
Territory of Australia. Similar results were also reported for other measures of
genetic variation (PPL; alleles per locus; data not shown).
Thus, Papua New Guinea is a center of genetic diversity of the species.
Wickneswari and Norwati (1993) explained the observed patterns of genetic
variation by the evolutionary history of the species. They suspect that the
species survived less favorable climatic conditions in small, scattered refuges in
Papua New Guinea, and expanded only recently via land bridges connecting
Papua New Guinea and Australia until the end of the last glaciation. According
to this hypothesis, only a few trees colonized Queensland and, later, the
Northern Territory. Thus, the lower levels of genetic variation in Australia as
compared with Papua New Guinea are explained by bottleneck effects (genetic
drift) during recent recolonization (From Wickneswari and Norwati 1993).

Papua New Guinea He

Natural distribution Bensbach 14.3 %
Morehead R. Rouku 10.3 %
Investigated populations North Mibini 18.0 %
Mibini Swamp 13.0 %
Old Tonda Village 11.9 %
Northern Territory He Mai Kussa River 12.0 %
Manton River 5.0 % mean (Papua New Guinea) 13.3 %
Reynolds River 0.7 %
Douglas River 7.5 %
Gerowie Creek 3.0 %
South Alligator River 0.2 % Queensland He
East Alligator River 2.7 % Wenlock River 6.5 %
Goomadeer River 8.7 % South Coen 9.1 %
mean (Northern Territory) 4.0 % Morehead River 9.1 %
Mount Molloy 5.5 %
Kings Plain 9.4 %
mean (Queensland) 7.9 %

Fig. 3.1. Natural distribution and genetic variation of Acacia auriculiformis. He expected
heterozygosity. (Adapted from Wickneswari and Norwati 1993)
3.5 Case Studies on Genetic Variation Patterns 33

Example 3.2: Genetic Differentiation Among Populations of Eugenia dysenter-

ica from the Brazilian Cerrado
Eugenia dysenterica DC. is a fruit tree species growing in the Cerrado region of
Brazil. Ten populations were investigated for variation at eight isozyme gene
loci (Telles et al. 2003). The species proved to be highly polymorphic with an
expected heterozygosity between 0.223 and 0.377 in the populations and an
average number of alleles per locus ranging from 2.4 to 3.6. However, levels of
observed heterozygosity (Ho) were considerably lower than the respective val-
ues for He in all populations. Accordingly, positive fixation indices (F; inbreed-
ing coefficients; Sect. 6.4.1) were observed at all marker loci.
The overall genetic differentiation among populations was estimated as
0.154, indicating considerable genetic differences between populations. A
highly significant correlation between the genetic distance (Nei 1972) and the
geographical distance between populations was observed (Fig. 3.2). Two spa-
tially isolated populations are clearly distinct from the remaining eight popu-
lations. The two genetically distinct groups should be treated as independent
units for conservation (From Telles et al. 2003).

Example 3.3: Genetic Variation of Swietenia macrophylla Across the Brazilian

Mahogany (Swietenia macrophylla King) is one of the most important tropical
timber species worldwide. Timber is harvested in plantations and logged in
natural populations. Seven natural populations from the southern arc of the
Brazilian Amazon basin were investigated at eight microsatellite (simple
sequence repeat, SSR) gene loci. Variation was very high, with a mean number

r 0.725
Genetic distance (Nei)





0 50 100 150 200 250
Geographic distance (km)
Fig. 3.2. Comparison of genetic distances (Nei 1972) and geographic distances between
ten populations of Eugenia dysenterica. (From Telles et al. 2003)
34 C HAPTER 3 Genetic Variation of Tropical Forest Plants

of 18.4 alleles per locus observed in all populations and a mean expected het-
erozygosity of 0.781 (Table 3.4). The inbreeding coefficient was positive and
significantly different from 0 in four populations, indicating a slight excess of
homozygotes relative to HardyWeinberg proportions (Sect. 6.4.1). The total
genetic differentiation among populations was estimated as 0.097. This is only
slightly lower than the average reported for tropical forest trees (GST = 0.109;
Sect. 3.4) and may be regarded as a moderate level of genetic differentiation
among populations.
High levels of genetic variation were observed since microsatellites (SSRs)
were chosen as marker types. SSRs are hypervariable markers showing higher
levels of genetic variation in comparison with other codominant markers such
as isozymes (Sect. 2.3.1). However, genetic variation in the Amazon basin was
higher than in a directly comparable study on mahogany in Central America
(Novick et al. 2003). On the other hand, the differentiation among populations
was slightly higher among Central American populations. The significant
genetic differentiation among populations in both regions stresses the impor-
tance of conserving as many populations as possible rather than of concentrat-
ing on one or a few populations in a region to be conserved in situ (Chap. 14)
(From Lemes et al. 2003).

Example 3.4: Genetic Variation of Endangered Australian Species of the

Genus Fontainea Assessed with RAPD Markers and by DNA Sequences
Fontainea oraria Jessup and Guymer is an extremely rare and endangered,
dioecious, small tree species known to exist on a single site, a littoral rain for-
est in New South Wales (Australia) (Rossetto et al. 2000). Only ten adult plants
and 52 seedlings grow on a 600-m-long strip. Sequencing of a cpDNA frag-
ment and of the nuclear ITS2 region (Sect. 8.3) revealed a close relation to

Table 3.4. Genetic variation at eight microsatellite gene loci in seven populations of
Swietenia macrophylla from the Brazilian Amazon. (Adapted from Lemes et al. 2003)
Population A/L He Ho F Deviationa

A. Azul 8.4 0.761 0.753 0.012

Cach. A 10.7 0.785 0.781 0.005
Maraj 9.2 0.793 0.740 0.068
P. Lacerda 9.7 0.812 0.812 0.004
C. Mendes 10.6 0.754 0.709 0.060
Cach. E 10.2 0.810 0.780 0.042
P. Bueno 7.6 0.754 0.680 0.100
All populations 9.5 0.781 0.750 0.038
He expected heterozygosity, Ho observed heterozygosity, F inbreeding coefficient
Significant deviation from HardyWeinberg proportions at *P < 0.05, **P < 0.01, ***P < 0.001
3.5 Case Studies on Genetic Variation Patterns 35

other species of the genus Fontainea and in particular to F. australis, another

endangered species of the genus restricted in distribution to four sites.
It was possible to distinguish all F. oraria plants (adults and progenies) by an
investigation based on seven RAPD primers which amplified a total of 57
RAPD bands. The level of genetic variation at RAPD markers was similar for
F. oraria and F. australis; however, variation was lower in the progenies of F. oraria
as compared with the parental population. An uneven contribution of adult
plants to the production of the seedlings (uneven fitness) is likely. There is
evidence that a single tree dominates the production of progenies while other
adults do not participate in reproduction.
The example shows that even an extremely low population size does not
preclude the observation of genetic variation in a species (From Rossetto et al.

Example 3.5: Genetic Variation of Shorea leprosula and Shorea parvifolia in

Indonesia Assessed at AFLP Loci
Shorea leprosula Miq. and Shorea parvifolia Dyer are common and widespread
emergents of lowland dipterocarp forest in Indonesia and other Southeast
Asian countries. They are very important timber trees and are the main sources
of light red meranti wood. Genetic variation was assessed by AFLP markers in
12 natural populations (six for each species) from Sumatra and Kalimantan
(Borneo), and for S. leprosula in addition in one plantation from Java (Fig. 3.3)
(Cao et al. 2006). Levels of genetic variation (He; Table 3.5) were estimated
based on the assumption of HardyWeinberg structures in both populations.
Variation is higher for S. leprosula in comparison with S. parvifolia (Table 3.5).
Low variation was observed in both S. parvifolia populations sampled in Borneo.
The population Asialog from Sumatra clearly showed the highest variation for
both species. Variation levels measured as He decreased in the same order for the
five locations where both species were sampled (Asialog > Pasir Mayang > Bukit
Tiga Puluh National Park > Sari Bumi Kusuma > Nanjak Makmur).
Considerable differentiation was not only observed among species, but also
among populations. The proportion of the total genetic variation within species
due to differentiation among populations is high for S. leprosula (GST = 0.20)
and even higher for S. parvifolia (GST = 0.31). Populations of S. leprosula from
Borneo are clearly separated from populations of the same species from
Sumatra (Fig. 3.4). The cluster diagram (Fig. 3.4) suggests the origin of the
planted population (Haurbentes, Java) as Sumatra.
Conservation of genetic resources of dipterocarps needs to take into account
the strong genetic differentiation among populations at presumably neutral
AFLP loci. The presence of numerous diagnostic characters occurring in high
frequency in only one or a few populations suggests that conservation in as
36 C HAPTER 3 Genetic Variation of Tropical Forest Plants

Sumatra Kalimantan (Borneo)

(S. lepr.
S. parv.)
(S. parv.)

TS (S. lepr.
(S. lepr. S. parv.)
S. parv.)
(S. lepr.
S. parv.)
(S. lepr. (S. lepr.)
S. parv.)
(S. lepr.)

Fig. 3.3. Location of sampled populations of Shorea leprosula and S. parvifolia in
Indonesia. Abbreviations as in Table 3.5. (From Cao et al. 2006)

Table 3.5. Genetic variation at AFLP loci in two dipterocarp species (Shorea leprosula
and Shorea parvifolia) in Indonesia. (Adapted from Cao et al. 2006)
Species Population Abbreviation Location PPL (%) He

S. leprosula Haurbentes HA Java 60.71 0.186

Tering TB Borneo 64.29 0.146
Sari Bumi Kusuma SB Borneo 60.71 0.174
Asialog AS Sumatra 82.14 0.258
Pasir Mayang PS Sumatra 62.50 0.178
Bukit Tiga Puluh TS Sumatra 60.71 0.177
National Park
Nanjak Makmur NS Sumatra 50.00 0.151
Mean 61.93 0.177
Total 92.86 0.222
S. parvifolia Batu Ampar BB Borneo 33.93 0.115
Sari Bumi Kusuma SB Borneo 46.43 0.124
Asialog AS Sumatra 73.21 0.221
Pasir Mayang PS Sumatra 60.71 0.151
Bukit Tiga Puluh TS Sumatra 46.43 0.127
National Park
Nanjak Makmur NS Sumatra 50.00 0.119
Mean 54.57 0.150
Total 82.14 0.206
He expected heterozygosity
3.5 Case Studies on Genetic Variation Patterns 37

S. lepr._AS
S. lepr._HA
S. lepr._NS
S. lepr._PS
S. lepr._TS
S. lepr._TB
S. lepr._SB
S. parv._SB
S. parv._BB
S. parv._TS
S. parv._AS
S. parv._PS
S. parv._NS

Fig. 3.4. Unweighted pair-group method with arithmetic means tree based on Neis
standard genetic distance illustrating genetic differentiation among populations of two
dipterocarps from Indonesia. Abbreviations as in Table 3.5. (From Cao et al. 2006)

many populations as possible is important. The identification of centers of

genetic diversity of dipterocarps seems feasible since ordering of populations
with regard to their variation (He) revealed the same trend for both species, with
a single location (Asialog) clearly showing the highest diversity for S. leprosula
and S. parvifolia (From Cao et al. 2006).

Example 3.6: Genetic Variation of Cedrela odorata at cpDNA

Cedrela odorata is an important timber species of the family Meliaceae with a
wide natural distribution from Mexico to northern Argentina. A study of vari-
ation of cpDNA in Central American populations revealed five different hap-
lotypes (Cavers et al. 2003). Populations are extremely differentiated with only
three from 29 populations investigated showing polymorphism (Table 3.6);
96% of the total variation is distributed among populations (GST = 0.96).
Maternal inheritance of cpDNA frequently results in strong differentiation of
cpDNA haplotypes among populations.
Repeated migration from South America to Central America is a likely cause
of the strong differentiation. The ancestors of trees with the northern haplo-
types 1 and 2 found in Guatemala and Mexico might have migrated to Central
America early, possibly even before the land bridge connecting South America
and North America was formed. The differentiation between populations with
the central haplotype 3 and the southern haplotypes 4 and 5 coincides with
considerable differences of climatic conditions and morphological and physio-
logical traits. Likewise, investigations of anonymous RAPD markers revealed
strong differentiation among these ecotypes. Thus, repeated colonization of
Central America during eras with different climatic conditions is a likely cause
for the observed differentiation.
38 C HAPTER 3 Genetic Variation of Tropical Forest Plants

Table 3.6. Variation of chloroplast DNA haplotypes in

Central American populations of Cedrela odorata.
(Adapted from Cavers et al. 2003)
Population Haplotype
1 2 3 4 5

Gualaca 5
Las Lajas 5
San Francisco 1 4
Costa Rica
Horizontes 20
Puriscal 20
Canas 20
Palo Verde 20
Hojancha 20
Jimenez 20
Talamanca 20
Upala 19 1
Pacifico Sur 20
Perez Zeledon 20
Ometepe 20
Masatepe 20
Wabule 20
La Trinidad 20
Meambar 20
Taulabe 20
Comayagua 20
La Paz 20
Los Esclavos 4 1
El Idolo 5
Tikal 5
San Jose 5
Calakmul 20
Bacalar 20
Zona maya 20
Escarcega 20
Total 99 1 259 106 9

The strong differentiation among populations from different regions offers

good opportunities to apply molecular markers to control the origin of repro-
ductive material (seeds or seedlings) (From Cavers et al. 2003).
3.6 Recommended Literature 39

Recommended Literature

A comprehensive discussion of methods to analyze genetic variation within

and among populations is given by Nei (1987). Short introductions to the
analysis of genetic variation based on isozymes and several nuclear DNA mark-
ers, respectively, are provided by Berg and Hamrick (1997) and Nybom (2004).
Weising et al. (2005, p. 207 ff) describe the analysis of genetic variation at
marker loci. Older literature on isozyme variation of tropical forest trees has
been reviewed by Loveless (1992). Useful reviews on genetic variation within
and among plant populations at isozyme loci were also presented by Hamrick
and Godt (Hamrick and Godt 1989, 1996, 1997; Hamrick et al. 1992).
Sexual and Asexual Reproduction in Tropical Forests 4

Flowering plants have evolved one of the most complex and sexiest life cycles on
earth (Armstrong 1996).


Reproduction is an obvious prerequisite for the survival of species and popula-

tions. Reproduction in tropical forests has attracted much interest from ecologists,
botanists, and conservation biologists since an understanding of the processes tak-
ing place during reproduction is fundamental for an analysis of the evolution and
maintenance of the high (species) diversity in these forests (Sect. 7.2).
A population is defined as a unit of reproduction (Sect. 2.2), and the recom-
bination of genes during (sexual) reproduction implies that changes of genetic
structures are particularly likely during the reproduction phase. Thus, an
understanding of genetic processes during reproduction is crucial for the
analysis of the temporal dynamics of genetic structures within populations, in
particular for long-living organisms such as forest trees.
The basic genetic principles connected to sexual and asexual reproduction in
tropical forests are reviewed in this chapter. The distribution of genetic informa-
tion during reproduction is discussed in more detail in Chap. 5, and the new
combination of genes to genotypes by the mating system is the topic of Chap. 6.
Chapters 46 are mainly concerned with the reproduction process in natural
forests. The implications of human impact on the reproduction system of tropi-
cal forests plants and the production of forest reproductive material for enrich-
ment planting and plantation establishment are discussed in Part B of this book.

Sexual Reproduction

Sexual reproduction is characterized by the fusion of two gametes of opposing

sex to a zygote. The formation of gametes is preceded by meiosis, i.e., a (sexual)
cell division resulting in gametes with only half of the chromosome number as
42 C HAPTER 4 Sexual and Asexual Reproduction in Tropical Forests

compared with the chromosome number in somatic cells of the parental plant.
The consequences of this process for genetic structures of a progeny generation
derived from a parental stand (t) and the possible causes of deviations of the
genetic structures in the progeny generation (t +1) from the population of adults
are illustrated in Fig. 4.1. Trees of generation t produce female and male gametes.
Differences in the ability to produce gametes result in fertility differences among
trees (fertility selection). Male gametes (pollen) are distributed during the flow-
ering period and external pollen from other populations is carried in the popu-
lation (pollen influx). Mating preferences or incompatibilities possibly result in
differences between allelic structures of the pollen cloud reaching a stigma and
the fertilizing or effective pollen, for example, owing to differences in pollen tube
growth. Viability selection starts early after fertilization and results in differences
of genetic structures between fertilized ovules and fully developed seeds. Viability
selection continues during seed germination, seedling growth, and throughout
stand development. For practical reasons, it is not easily possible to observe the
genetic structures of female and male gametes (pollen and ovules) or zygotes.
Thus, changes of genetic structures are often monitored by comparing the adult
population (generation t) with the mature seeds (generation t +1; comparison 1 in
Fig. 4.1), by comparing the seeds with the population of seedlings, saplings, or
adults (comparison 2 in Fig. 4.1; Example 7.1), or by comparing two generations
of the same or, in case of long-living organisms like forest trees more frequently,
of different age (comparison 3 in Fig. 4.1).

Sexual Types and Sexual Systems

Only gametes of opposite sex can fuse and jointly produce the zygote which
will further develop by mitotic cell divisions into an embryo, a seedling, and
eventually into an adult plant capable of producing gametes itself. The pro-
duction of these two different types of gametes (female and male) is a require-
ment for successful sexual reproduction in a population. The sexual type of an
organism depends on the types of gametes it produces: It is either cosexual
(female and male gametes are produced) or unisexual female or male. Plants
not capable of producing any functioning gametes are sterile.
The determination of the sexual type of an organism is usually under strict
genetic control although an environmental influence on the sexual type has
been observed for some animals. The sexual type of most animals and some
plants is determined by particular chromosomes (gonosomes, i.e., sexual chro-
mosomes). Females often have two identical gonosomes: they are homoga-
metic. Males of such species have two different gonosomes: they are
heterogametic. An example of sex determination by gonosomes are the X and
4.2 Sexual Reproduction 43

production of
and gametes
fertility selection distribution of
flowering over time
trees of pollen transport
pollen influx
generation t


mating preferences
comparison 1


comparison 3 viability selection

(self-sterility etc.)
mature seeds of
generation t + 1

viability selection
(germination, mortality)

comparison 2 stand
n development
c tio
l ity

trees of
generation t + 1
Fig. 4.1. Causes for changes of genetic structures during and after reproduction in
consecutive generations. (From Hattemer 1999)

Y chromosomes of humans. Females have two X chromosomes; males one X

and a smaller Y chromosome.
The sex of some insects mostly belonging to the Hymenoptera (e.g., bees
and wasps) is determined by the ploidy level. Unfertilized, haploid eggs develop
into males. Diploid, fertilized eggs develop into females (Cook 1993).
Flowering plants are principally cosexual; however, the separation of
sexes (dioecy) repeatedly evolved in plants (Sect. 4.2.2). Heteromorphic sex
44 C HAPTER 4 Sexual and Asexual Reproduction in Tropical Forests

chromosomes have been identified for a few dioecious plants, for example, in
the herbaceous genera Silene and Rumex (Dellaporta and Calderon-Urrea
1993). A particular region on the LG1 chromosome has been identified as
carrying sex-determining genes in papaya (Carica papaya) (Charlesworth
2004). Sex expression is regulated by hormones in many dioecious plants
(Dellaporta and Calderon-Urrea 1993), but the mechanism of sex determi-
nation is still unknown for most woody plant species and for virtually all
tropical forest species.
A possible determination of the sex by age (or size) of trees has been
described for the neotropical tree Iryanthera macrophylla. The plants of this
species produce only one type of flower (female or male) in a particular repro-
duction period; however, Ackerly et al. (1990) found evidence that small, young
trees of this species flower male, but later produce female flowers. Thus, the sex
expression may not be determined by genetic factors, but by the age or size of
plants, and the species may not be dioecious, but diphasic (Ackerly et al. 1990).

Sexual Systems

Since the formation of a progeny generation depends on the production of

gametes of either sex, only the following combinations of the sexual types of
individual plants to the sexual system of a species are possible:

Cosexuality: all plants produce both types of gamete (female and male)
Dioecy: plants produce only one type of gamete they are female (female)
or male
Trioecy: some plants produce both types of gamete (female and male),
others are female or male
Gynodioecy: some plants produce both types of gamete (female and male),
others are female (no males)
Androdioecy: some plants produce both types of gamete (female and
male), others are male (no females)

Cosexuality is the archetypal and most common sexual system for plants. Two
forms of cosexuality are distinguished: Hermaphroditic plants have complete
flowers producing both types of gametes in a single flower. Monoecious plants
produce both types of gamete in different flowers. Thus, single flowers are
either male or female in monoecious species, but these flowers are produced by
the same plant (Table 4.1).
Different types of pollination can be distinguished for hermaphroditic species
depending on the origin of the fertilizing pollen (Fig. 4.2). Autogamy is the
4.2 Sexual Reproduction 45

Table 4.1. Sexual systems of tropical plants with examples

Sexual system Sexual types Examples

Cosexuality: ( ) Majority of tropical tree species, e.g.,

hermaphroditism in single flowers Dipterocarpaceae, Tectona, Eucalyptus
Cosexuality: monoecy ( ) in Pinus, Castanopsis, and other
different flowers on Fagaceae
single plants
Dioecy ( ), ( ) Calamus spp., Diospyros spp., most
Trioecy ( ), Pachycereus pringlei, Fraxinus
( ), ( ) excelsior a
Gynodioecy ( ), Rhizophora mangle, Ocotea tenera,
( ) Ficus spp.b
Androdioecy ( ), Castilla elastica, Bauhinia pauletia,
( ) Bauhinia ungulata
Temperate tree species
Gynodioecious figs are functionally dioecious (Example 7.2)

Autogamy Geitonogamy

Single flower

Selfing Outcrossing

Fig. 4.2. The origin of fertilizing pollen for a flower for a hermaphroditic tree species.
(Adapted from Hattemer et al. 1993)

fertilization of an ovule by pollen from the same flower. It is just like geito-
nogamy, the fertilization with pollen from a different flower of the same tree,
a form of allogamy resulting in selfing (self-pollination; Sect. 6.3). Xenogamy is
pollination from a different plant and results in outcrossing. Autogamy is not
possible for monoecious species, and no form of allogamy, neither autogamy
nor geitonogamy, is possible for dioecious species, which are, in consequence,
obligatory outcrossed.
46 C HAPTER 4 Sexual and Asexual Reproduction in Tropical Forests

Hermaphroditism is the most common sexual system in tropical forests, fol-

lowed by dioecy. Monoecious species are widespread in temperate and boreal
forests, but are rarely encountered in tropical forests (Table 4.2). The domina-
tion of monoecious species in temperate forests reported for North America in
Table 4.2 is also observed in central Europe: the dominating species of the fam-
ilies Fagaceae (Fagus, Quercus) and Pinaceae (Pinus, Picea, Abies) are monoe-
cious. All other sexual systems (trioecy, gynodioecy, androdioecy) are
extremely rare, both in tropical and in temperate forests.
The sexual system of figs (Ficus spp.), which are either monoecious or
gynodioecious, is explained in detail in Example 7.2. Bauhinia ungulata is
pollinated by bats. It is morphologically a hermaphrodite, but seems to be func-
tionally androdioecious. Trees produce flowers with either short or long pistils,
but seeds do not develop on trees with short pistils (Ramirez et al. 1984). The
floristic composition of a forest has an influence on the frequency of sexual sys-
tems since particular sexual systems are common in certain families. For exam-
ple, many species of the Salicaceae [willows (Salix spp.) and poplars (Populus
spp.)] are dioecious. The proportion of this family has a strong impact on the
frequency of dioecy in a community of woody plants in a temperate climate.
Tropical tree families with many dioecious species are, for example, the
Meliaceae and the Ebenaceae. The high proportion of dioecious species in a
South Nigerian rain forest (Table 4.2) is explained by the high frequency of these
two families in the area investigated.
The high frequency of dioecious species in species-rich forests of the central
tropics is surprising in view of the low densities of many populations, since
dioecious species are obligatory outcrossing. Thus, pollen vectors need to trans-
port pollen from male to female plants for the production of sexual progenies.
The advantages of outcrossing as compared with selfing (Sect. 6.4.2) are

Table 4.2. Frequency (%) of common sexual systems in tropical and temperate forests.
(Adapted from Bawa and Opler 1975)
Location Cosexual: hermaphroditic Cosexual: monoecious Dioecious

Costa Rica 68 10 22
South Nigeria 47 13 40
Sarawak 60 14 26
Florida 61 12 27
Temperate zone
Ohio 6 83 11
Ohio 27 60 13
Arkansas 0 83 17
Kentucky 15 70 15
New Hampshire 13 81 6
4.2 Sexual Reproduction 47

assumed to be the main reason for the evolution of dioecy for many originally
self-compatible species of tropical forests (Bawa and Opler 1975).
The frequency of sexual types in a forest is related to main pollinators (Sect.
5.2.2). Monoecious species are mainly wind-pollinated. The complete flowers
of hermaphroditic species are predominantly pollinated by animals.

Sexual Structures and Sexual Function

The sexual structure of a population is the frequency distribution of its differ-

ent sexual types. In the case of a cosexual (hermaphroditic or monoecious)
species, it simply holds that P = 1, where P is the relative frequency of
cosexual type from all f lowering plants. For a dioecious species, an even fre-
quency of sexual types (P = P = 0.5) is often assumed, where P (P ) is the
relative frequency of female (male) plants.
The frequencies of male and female flowering plants were calculated for var-
ious dioecious species in Costa Rica (Table 4.3) (Opler and Bawa 1978). The
23 species investigated belong to 13 different families. This supports the hypoth-
esis that dioecy evolved independently during the evolution of angiosperms in

Table 4.3. Sexual structures of tree species in a tropical forest in Costa Rica. Test for
significant deviation from an even sex ratio (male-to-female ratio of 1; com. X2) and
test for significant heterogeneity of the sex ratio among populations (int. X2). (Adapted
from Opler and Bawa 1978)
Family Species No. of No. of No. of Male-to- Com. Int.
Popu- males females female X2 X2
lation ratio

Anacardiaceae Spondias nigrescens 5 57 58 0.98 NS NS

** **
Boranginaceae Cordia collococca 6 127 85 1.50
Cordia panamensis 6 110 52 2.11 NS
Burseraceae Bursera simarouba 9 114 95 1.20 NS NS
Bursera tomentosa 5 74 60 1.23 NS NS
Erythroxylaceae Erythroxylon 1 30 14 2.14
Euphorbiaceae Bernardia 2 23 22 1.04 NS
Meliaceae Trichilia anisopleura 5 58 42 1.38 NS NS
Trichilia cuneata 5 115 93 1.24 NS NS
Moraceae Cecropia peltata 9 104 100 1.04 NS NS
Chlorophora tinctoria 7 79 64 1.23 NS NS
Nyctaginaceae Pisonia 5 61 45 1.36 NS NS
48 C HAPTER 4 Sexual and Asexual Reproduction in Tropical Forests

Table 4.3. Sexual structures of tree species in a tropical forest in Costa Rica. Test for
significant deviation from an even sex ratio (male-to-female ratio of 1; com. X2) and
test for significant heterogeneity of the sex ratio among populations (int. X2). (Adapted
from Opler and Bawa 1978)Contd
Family Species No. of No. of No. of Male-to- Com. Int.
Popu- males females female X2 X2
lation ratio
** **
Polygonaceae Coccoloba caracasana 6 55 136 0.41
Coccoloba floribunda 1 23 27 0.85 NS
Ruprechtia costata 1 21 30 0.70 NS
Triplaris americana 8 157 220 0.71 NS
* **
Sapindaceae Allophyllus 4 74 50 1.48
Simaroubaceae Simarouba glauca 4 111 102 1.08 NS NS
Rubiaceae Albertia edulis 4 83 49 1.69 NS
Genipa caruto 6 76 77 0.99 NS NS
** **
Randia spinosa 9 154 57 2.70
Randia subcordata 7 157 77 2.04 NS
Zanthoxylaceae Zanthoxylum 4 115 60 1.92 NS
NS not significant
P < 0.05; **P < 0.01

various taxa. A significant deviation from an expected even ratio of male and
female plants was observed for ten out of the 23 species (com. X2 in Table 4.3).
Significantly more males than females (male-to-female ratio greater than 1
and significant com. X2) were observed for eight species. A significant excess of
females (male-to-female ratio less than 1 and significant com. X2) was observed
for only two species. Significant heterogeneity of the sex ratio among popula-
tions (int. X2) was observed for four species with P < 0.01.
Deviations from a balanced sex ratio were also observed in populations of
functionally dioecious figs (Ficus spp.) and rattan (Calamus spp.). In most
cases an excess of male plants has been reported (Table 4.3). Possible reasons
for an unbalanced sex ratio are:

Mode of inheritance of sex. Sex determination is unknown for most tropical

forest plants (Sect. 4.2.1).
Gametic selection favoring one sex. Pollen carrying the information for a
male (or female) progeny is significantly more successful in fertilization
that pollen carrying the information for the opposite sex of the progeny.
Viability selection. Female (or male) plants show significantly lower survival
as compared with survival rates of the opposite sex. Different resource allo-
cation with regard to vegetative growth and investment in reproduction is
4.3 Asexual Reproduction 49

likely for many dioecious plants. Female plants need to supply energy and
nutrients for the production of female flowers and seeds which are often
embedded in large fruits. These resources are not available for vegetative
growth. This possibly results in reduced viability as compared with that of
male plants producing only pollen in male flowers.
Earlier maturity of one sex, usually males, as compared with the opposite
sex, or higher frequency of flowering of one sex. For example, annual flow-
ering for males but a biannual cycle for females results in an excess of males
in each flowering period.
Preferential asexual reproduction of one sex. Strict genetic control of the sex
results in identical sex expression of all plants from vegetative reproduction
of a particular (female or male) tree.

The sexual function of a plant is the proportion of successful female gametes

from all successful gametes (female and male) produced by a plant. A gamete
is regarded as successful or effective if it has contributed its genetic informa-
tion to a zygote. There are two sexual functions for dioecious species. Female
plants have a sexual function of 1, since they only produce female gametes. The
sexual function of males is 0, since they never produce female gametes. If the
female and male gametes of a cosexual plant are successful in even proportions,
its sexual function is 0.5.
Observation of flowering intensities in populations strongly suggests devi-
ations from a sexual function of 0.5 for many monoecious species. Some
trees flower predominantly or, at least in a particular flowering season, exclu-
sively male, while others flower predominantly female. Most of these investi-
gations were conducted for monoecious temperate forest trees such as Pinus
sylvestris, Fagus sylvatica, or the trioecious Fraxinus excelsior (p. 161 ff. in
Hattemer et al. 1993). Clones in a Pinus sylvestris seed orchard strongly dif-
fered with regard to their male and female reproductive success (Mller-
Starck and Ziehe 1984).
Hermaphroditism precludes the estimation of sexual functions based on an
observation of flowering, since male and female flowering cannot be easily dis-
tinguished. Thus, the variation of sexual functions in populations of tropical
trees is largely unknown owing to the predominance of hermaphroditism as a
sexual system in the tropical forests.

Asexual Reproduction

Many plant species and some animal species have the potential for asexual
reproduction. In this case, offspring is produced from tissue of a single organ-
ism by mitotic cell divisions only; meiosis and recombination do not occur.
50 C HAPTER 4 Sexual and Asexual Reproduction in Tropical Forests

Thus, the genotypes of asexually produced offspring are identical copies of the
genotype of the original plant. Obviously, asexually produced progenies of a
single plant are also genetically identical among each other. A collective of
genetically identical organisms is called a clone.

Vegetative Reproduction

Vegetative reproduction is not only an important tool for the production of

forest reproductive material such as cuttings (Hartmann and Kester 1983), but
is also frequently encountered in natural and secondary forests of the tropics.
The importance of vegetative propagation for tree improvement, plantation
forestry, and conservation of forest genetic resources will be discussed in later
chapters. The following remarks mainly refer to vegetative reproduction under
natural conditions.
Not all plants which are commonly vegetatively propagated by humans are
capable of vegetative reproduction in natural forests. Root suckers, i.e., the
development of sprouts from roots, are the most common form of vegetative
reproduction in natural forests. Coppicing may be regarded as a form of vege-
tative growth or of reproduction; thus, coppice shoots exemplify the transition
from vegetative growth to vegetative multiplication.
Many tropical forest plants use vegetative reproduction as a complementary
method in addition to sexual reproduction via seeds. Root suckers are an
important means of reproduction for Cordia alliodora. They are particularly
common in dry and semidry forests of the tropics. For example, Prosopis
juliflora and other legumes in dry forests frequently reproduce by root suckers.
Bamboos are another important species group which predominantly repro-
duces vegetatively. Flowering of bamboos is a rare event, resulting in the death
of a flowering plant after seed production for many species (Janzen 1976).
Bambusa vulgaris is a commonly cultivated species throughout the tropics. It
very rarely flowers and the production of viable seeds has never been observed
for this species (Banik 1995); thus, B. vulgaris is one of the very few taxa which
exclusively multiplies vegetatively.
Vegetative reproduction is not only common for tropical trees, but also for
other plants growing in forests of the lower latitudes. For example, vegetative
reproduction was expected in clusters of the neotropical epiphytic orchid
Laelia rubescens Lindley. Clonal reproduction within clusters was confirmed
by marker-based inventories, but the degree of sexual reproduction and,
hence, genetic diversity within clusters was surprisingly high (Trapnell
et al. 2004).
4.3 Asexual Reproduction 51


A sexual embryo develops in the seed of flowering plants after fertilization of

the ovule and fusion of the genetic information of two gametes (female and
male) to a zygote. The development of an embryo without the processes of
fertilization and fusion of gametes is called agamospermy or, more commonly,
apomixis in a narrow sense.
Typically, the embryo develops from a somatic cell of the seed parent
(apospory). In this case, it is genetically identical to the seed parent. All seeds
containing an embryo from apopory from the same seed parent are clones. The
development of an embryo from a macrospore mother cell after a meiotic cell
division is called diplospory. The resulting embryos from the same plant are
not necessarily genetically identical; hence, diplospory is not a form of asexual
reproduction in a strict sense.
Apospory is a method to propagate and distribute genetically identical
organisms (clones) through seeds. This is particularly advantageous if pollen
transport from other trees (xenogamy) is either absent or very limited, for
example, owing to a large spatial distance between trees growing in low density.
Seeds developing in the absence of foreign pollen after selfing usually show
strong inbreeding depression (Sect. 6.4.2). Inbreeding is avoided and individ-
ual variation (heterozygosity) maintained if apomictic seeds are produced.
The potential for apomixis is particularly advantageous for a plant if it is
combined with the ability to produce sexual progenies if foreign pollen is avail-
able (facultative apomixis). For example, the walnut (Juglans regia) produces
sexual offspring if pollen is available, but is capable of apomixis in the case of
a lack of pollination (p. 47 in Stern and Roche 1974).
Trees are usually sterile if meiotic cell divisions are disturbed owing to an
unbalanced chromosome number (e.g., triploidy) since no functioning
gametes are produced; however, such trees may produce viable seeds by means
of apomixis. Thus, many apomictic species are polyploid; the somatic cells of
those plants contain more than two full sets of chromosomes. For example,
triploid Casuarina nana plants reproduce through apomixis (Barlow 1958).
The West African species Pachira oleaginea (Bombacaceae) is a hermaphro-
dite pollinated by bats. It is a polyploid species with 72 chromosomes in
somatic cells. Seeds are produced only after pollination and fertilization, but
self-fertilization is possible; thus, Pachira oleaginea seems to be self-fertile
although other species of the genus Pachira have an incompatibility system
(Sect. 6.5.1). A sexual proembryo develops in the ovule after fertilization, but
up to 15 adventive embryos start to develop almost simultaneously in the same
ovule from tissue of the nucellus. The mature seeds contain only two to five
52 C HAPTER 4 Sexual and Asexual Reproduction in Tropical Forests

embryos. Only one or two of these develop into seedlings. In most cases, the
sexual embryo dies during seed development since adventive embryos grow
faster. Thus, seed production requires regular fertilization, but embryos of
most mature seeds have an apomictic origin (Baker 1960). This form of seed
development (pseudogamy) requiring pollination but resulting in apomictic
embryos has also been described for fruit species of the genus Citrus.
At least facultative apomixis has been described for several dipterocarps,
including Stemonoporus oblongifolius (Murawski and Bawa 1994), Hopea odor-
ata (Wickneswari et al. 1995), and the autotetraploid Shorea ovalis ssp. sericea
(Ng et al. 2004). Apomixis is suggested to be an evolutionary strategy allowing
the maintenance of high genetic diversity in populations of Shorea ovalis ssp.
sericea (Ng et al. 2004). Apomixis in H. odorata is possibly related to the occur-
rence of triploid trees in this species (Jong 1976).
Kaur et al. (1978) suggested that apomixis may be common not only for
dipterocarps, but also for other tropical forest plants. The observation of abun-
dant seed crops of spatially isolated trees of dioecious or self-incompatible
species suggests apomixis as a likely explanation. However, marker-based stud-
ies rarely confirmed apomixis in tropical forests. In many cases, long-distance
transport of effective pollen seems to be responsible for the production of out-
crossed seeds even for trees isolated by distances of several hundred meters
from other flowering trees of the same species serving as potential pollen
donors (Sect. 5.2.3).

Recommended Literature

Early studies of Bawa and Opler (Bawa 1974, 1980; Bawa and Opler 1975;
Opler and Bawa 1978) on the sexual system of neotropical trees revealed the
high frequency of dioecy and self-incompatibility. Kaur et al. (1978) suspected
that apomixis may be widespread among tropical forest plants. This idea stip-
ulated in-depth studies which, however, only rarely supported their hypothesis.
Gene Flow and Migration 5

The level of gene exchange between populations is determined by a multitude of

factors, including the size, density, and shape of the donor and recipient populations,
plant height and breeding system, characters of the surrounding vegetation, terminal
velocity of pollen and seeds, pollen and seed prodction, and the foraging behavior of
pollen and seed vectors, as well as the distance between populations. To further con-
found matters, many of these factors vary in both time and space (Levin and
Kerster 1974).


Genetic information is transported within and among populations of tropical

forest species. The main focus of the following discussion is on plants, but ani-
mals are also considered since they are important vectors for the dispersal of
pollen and seeds, and since gene flow is crucial for the spatial distribution of
genetic information both in plants and in animals.
Higher plants and their female gametophytes, which are part of the gynoe-
cium, are immobile; thus, the principal methods for the transportation of
genetic information in plants are the movement of the male gamete (pollen),
and of seeds. For some species, vegetative reproduction results in short-
distance (e.g., root suckers) or long-distance (e.g., bulbils) transport of genetic
information. Seedlings are dispersed in rare cases such as in viviparous
mangrove species.
The terms gene flow and migration are in common use for the trans-
portation of genetic information in plants. In a broad sense, both terms refer
to any transportation of genetic information, either by pollen or by seed move-
ment. We will use the term gene flow in its narrow sense only for the movement
of pollen, and will describe the transportation of seeds as migration.
54 C HAPTER 5 Gene Flow and Migration

Gene Flow Through Pollen

Pollination and Fertilization

Sexual reproduction is based on the fusion of a male and a female gamete. The
transportation of the male gametophyte to the female plant is required in seed
plants. This requires a series of events in angiosperms: First, pollen is trans-
ported to the stigma of the same (autogamy) or a different (geitonogamy or
xenogamy; Fig. 4.2) flower. The pollination is followed by the growth of a
pollen tube through the style to the ovule. Then, after entering the ovule
through the micropyle, fertilization of the embryo sac takes place (p. 32 ff. in
Proctor et al. 1996).
The genotypic structures of the progeny generation are influenced both by
the dispersal of pollen and by genetic differences of the pollen with regard to
germination, pollen tube growth, and fertilization success. Pollen dispersal
from anthers to stigmas or, in the case of gymnosperms, to the micropyle is dis-
cussed in this chapter. Aspects related to the processes of the germination of
pollen, pollen tube growth, and fertilization are covered in Sect. 6.5.1.

Pollen Vectors in Tropical Forests

A detailed description of pollination and pollen transport by biotic and abiotic

vectors was given by Proctor et al. (1996). Prance (1985) reviewed the pollina-
tion and pollinators of Amazonian plants. Only the most important consider-
ations related to pollen transportation in tropical forests are briefly reviewed
Pollen is dispersed by abiotic or biotic means. Wind is the principal abiotic
factor facilitating pollen dispersal; different animals are pollen dispersal agents
of most tropical forest plants.
Pollination by Wind (Anemogamy)

Wind is the most important pollen vector of trees in the temperate and boreal
zone. For example, the principal stand-forming species in central Europe belong-
ing to the families Fagaceae (Fagus sylvatica, Quercus spp.) and Pinaceae (Pinus
sylvestris, Picea abies) are wind-pollinated (anemogamous). Wind pollination is
rare in the humid central tropics (Regal 1982); only 12% of plants in the central
5.2 Gene Flow Through Pollen 55

tropics disperse their pollen by the wind (Bawa 1990). However, this proportion
increases in semidry and dry tropical forests. Tropical pines (Pinus spp.) and
many other gymnosperms such as Podocarpus species are wind-pollinated (Regal
1982). Pioneers and species commonly found in secondary forests such as some
Cecropia spp. are also anemogamous. Wind has been suggested as a pollen vector
for some eucalypts, but these statements are not supported by experimental
evidence (p. 190 in Eldridge et al. 1994).
The success of pollen parents in pollinating a particular seed tree depends
on numerous factors, including the spatial structure and the distance between
trees, environmental factors, including the preferential wind direction, velocity,
and competing vegetation acting as a pollen filter, species-specific criteria
such as pollen morphology and pollen size, and individual characteristics,
including the number of pollen produced during a flowering period. For exam-
ple, there is evidence for limited pollen dispersal and preferential matings
among near neighbors for the wind-pollinated Araucaria angustifolia (Sousa
and Hattemer 2003; Sect. 11.4.3).
A large number of studies conducted on wind-pollinated tree species in
temperate forests revealed both preferential pollen transport from directly
neighboring trees (e.g., for Pinus sylvestris; Mller 1977) and a large proportion
of successful long-distance pollination (e.g., for Quercus macrocarpa; Dow and
Ashley 1998).
Observations in clonal seed orchards (Sect. 12.8.1) of wind-pollinated
conifers in temperate and boreal climates proved a high frequency of contam-
ination by external pollen (Wheeler and Jech 1992); thus, spatial isolation of
clonal seed orchards from conspecific forest stands does frequently not result
in reproductive isolation owing to efficient long-distance dispersal of pollen by
the wind. A high frequency of matings after long-distance pollen dispersal has
been suggested by Lindgren et al. (1995) for Scots pine (Pinus sylvestris). The
authors assume that viable pine pollen can be transported over hundreds of
kilometers before becoming effective in the boreal climate of Scandinavia.
The high species diversity and, hence, the low population density is likely
to be the main reason for the rarity of wind-pollination in evergreen moist
tropical forests, since wind-pollination is usually not as effective as animal
pollination where inter-individual distances are large (Regal 1982).
Pollination by Animals (Zoogamy)

Zoogamy is dominant in tropical forests, and numerous animals are involved

in pollen dispersal. This is in sharp contrast to temperate and boreal forests,
where fewer species are pollinated by animals, and where only certain insects
play a significant role in pollination.
56 C HAPTER 5 Gene Flow and Migration

Only small flying vertebrates (birds and bats) are important pollinators for
tropical plants. Nonflying mammals such as opossums and monkeys have been
reported to act as pollinators for a few plant species with robust flowers
(Janson et al. 1981), but most flowers are severely damaged by visiting
nonflying mammals.

Many Bombacaceae such as the baobab (Adansonia digitata) and Ceiba
pentranda (Lobo et al. 2005), but also some Parkia spp. (Mimosaceae) and
Bauhinia spp. (Caesalpinaceae; Heithaus et al. 1974) are predominantly pol-
linated by bats. Flowers visited by bats are usually very big and white or light
yellow. They open at dawn or during the night, smell strong, and produce large
quantities of nectar (p. 36 in Whitmore 1975). Inflorescences either stand
upright over the crown or hang from long twigs. Sometimes they are produced
by modified twigs originating directly from the stem (cauliflory, e.g., the can-
nonball tree Couroupita guianensis; Fig. 5.1). The production of flowers
suitable for bat pollination is energy-demanding for a tree. On the other hand,
most bats fly long distances. Accordingly, they are efficient vectors for the
transport of pollen over long distances. The efficiency of bats as pollinators
critically depends on the behavior of each species as well as preferred resting
areas (Heithaus et al. 1974; Stuart and Marshall 1976).

Fig. 5.1. Flowers of the cauliflorous cannonball tree Couroupita guianensis (Lecthidaceae).
(Photo: R. Finkeldey)
5.2 Gene Flow Through Pollen 57

Only comparatively few small bird species act as pollinators for tropical forest
plants. Hummingbirds are well-known pollinators of forest plants in tropical
America. Their function is fulfilled by various other species belonging to the
Zosteropidae, Dicaeidae, Nectariniidae, and Meliphagidae in other parts of
the tropics.
Many flowers visited by birds are bright red and big. They produce large quan-
tities of nectar, which is the main energy source for the pollinators, but the flow-
ers rarely have strong smells. Many inflorescences are brushlike with long
filaments. The conspicuous flowers of Erythrina spp. and Bromeliaceae are polli-
nated by birds. Heliconia spp. (Musaceae) are frequently planted ornamentals
originating from America. Their beautiful flowers are adapted to pollination by
birds. Several taxa, such as species of the genus Calliandra, are pollinated by birds,
bats, and possibly also insects. The production of bird-flowers is as energy-
consuming as the production of bat-flowers. The size and the behavior of birds
suggests that they are efficient vehicles for long-distance transport of pollen;
however, the actual distances between two trees which can be bridged by birds
critically depends on the sizes of their territories.

Pollination by Insects
Insects are the most important pollen vectors in the tropics. Many generalist
species visit flowers of a multitude of plants. Other pollinating insects coe-
volved with particular plant species. The obligatory symbiosis between figs
(Ficus spp.) and their pollinators (fig wasps; Agaonidae) is a particular well-
known example of a close coevolution between tropical plants and insects
(Janzen 1979a; Example 7.2).

The Hymenoptera (bees, wasps, bumble bees, etc.) are an extraordinary
species-rich group of pollinators. Although most bees forage on flowers of
numerous plant species, they often prefer a particular species at a given time.
The flight behavior of bees and hence their efficiency as pollinators depends on
numerous factors, including their size, their sex (male or female; Ackerman
et al. 1982), and their social or solitary life cycle.
There are few uniform features of flowers which are exclusively or pre-
dominantly pollinated by Hymenoptera. The flower morphology greatly
varies even within a single tree family such as the neotropical Lecythidaceae,
which is predominantly pollinated by bees (Prance 1985). Some plant species
attract numerous bees and wasps. For example, more than 70 different bee
species were counted visiting the flowers of a single Andira inermis tree (Bawa
58 C HAPTER 5 Gene Flow and Migration

Bees prefer to fly short distances between visited flowers. Accordingly, most
visits are within the crown of a single tree, and geitonogamy is expected to be
common (Sect. 4.2.2). Strategies to minimize geitonogamy and hence self-
fertilization include the evolution of incompatibility mechanisms (Sect. 6.5.1)
and a particular flowering phenology (see Example 6.1 for teak).

Butterflies and Moths

Flowers mainly pollinated by butterflies typically combine bright pink or red
colors with a sweet scent (Proctor et al. 1996). Nectar is usually accessible
through a slender corolla. Night-active moths are attracted by a heavy sweet
smell; flowers pollinated by moths are often pale.
A close coevolution has been described for Heliconius spp. butterflies and
vines belonging to the genera Anguria and Gurania in South America (Prance
1985). The butterflies learn the routes between plants widely scattered in the
forest and repeatedly visit male flowering plants on their way. The dioecious
plants depend on the butterflies as pollinators for reproduction, and the
pollen is an important source of protein for the butterflies (p. 82 in Proctor
et al. 1996).
Moths are less conspicuous than butterflies, but they are important pollina-
tors of many trees of the Rubiaceae, Meliaceae, Mimosaceae, and Apocynaceae
and other families.

Flies and Beetles

Flies (Diptera) and beetles (Coleoptera) are ubiquitous, extraordinary species-
rich groups of insects, but they play an important role in pollination of woody
plants only in the tropics.
Flowers predominantly pollinated by flies are often sweet-scented.
Occasionally, they have an unattractive, excremental scent. Some flowers adapted
to pollination by flies are adapted to trap pollination and sapromyiophily, i.e.,
pollination associated with decaying organic material (p. 294 ff. in Proctor et al.
1996). Long-term imprisonment of flies in flowers is common in several genera
such as Aristolochia (Aristolochiaceae) and Ceropegia (Asclepiadaceae). The huge
flowers of Rafflesia arnoldii, a flowering plant parasitic on climbers in tropical
Southeast Asia, are pollinated by flies.
Many angiosperms belonging to the Annonaceae and other primitive fam-
ilies are pollinated by beetles. Thus, this type of pollination seems to be arche-
typal for angiosperm plants. Beetle pollination has also been reported for
species of the gymnosperm family Cycadaceae, and for younger families such
as many palms (Palmae) (Eguiarte et al. 1993; Listabarth 1996). Flowers pol-
linated by beetles often attract other insects as well. Various beetles destroy
flowers or inflorescences at least partially by feeding on them.
5.2 Gene Flow Through Pollen 59

Other Insects
Other, frequently inconspicuous groups of insects serve important functions
for forest ecosystems since they are the main pollinators of keystone plant
species. Tiny thrips, cicadas, and bugs belong to this group. Dipterocarps from
the central tropics show the phenomenon of simultaneous mass flowering in
irregular intervals. Many of them, in particular Shorea and Hopea spp. with
small flowers, are mainly pollinated by tiny thrips (Thysanoptera) (Ashton
1988). Another example of the enormous diversity of pollinating insects is the
climber Uvaria elmeri (Annonaceae), which is pollinated by cockroaches
(Nagamitsu and Inoue 1997).

Pollen Vectors: General Patterns and Trends

The species diversity of tropical forests is mirrored by a great diversity of pol-
lination mechanisms. Some plants, for example, most figs (Ficus spp.), depend
on a single animal species and evolved an obligatory symbiosis to its pollinator
(Example 7.2). Other forest plants are pollinated by a great variety of animals.
Pollination by birds, bats, opossums, and large bees has been described for
Mabea fistulifera (Euphorbiaceae) in Brazil (Viera and de Carvalho-Okano
The relative importance of different main groups of pollinators has been
estimated for evergreen rain forests by Bawa (1990) (Table 5.1). Distinct pat-
terns appear for species of the canopy and of the undergrowth. For example,
birds and beetles are important pollinators chiefly for species of the under-
growth. A significant role of bees as main pollinators is evident both for canopy
trees and for the undergrowth.

Table 5.1. Relative frequency of different pollen vectors in the canopy

and the undergrowth of an evergreen tropical rain forest. (Adapted
from Bawa 1990)
Pollen vector Canopy Undergrowth

No. of species 52 220

Bats 3.8% 3.6%
Birds 1.9% 17.7%
Large bees 44.2% 21.8%
Small bees 7.7% 16.8%
Beetles 15.5%
Butterflies 1.9% 4.5%
Moths 13.5% 7.3%
Wasps 3.8% 1.8%
Various small insects 23.1% 7.7%
Wind 3.2%
60 C HAPTER 5 Gene Flow and Migration

Pollen Dispersal Assessed by Marker-Based Studies

The physical distribution of pollen can be studied by observation of the behav-

ior of pollinators. Pollinators can be repeatedly trapped and marked, for exam-
ple, with color stains, to reveal their mobility (Ackerman et al. 1982). The
pollen itself can be marked by fluorescent dyes, and its distribution can be
studied (Webb and Bawa 1983). Minimum pollination distances can be esti-
mated for dioecious species as the distance between a seed-bearing female
plant and its nearest male neighbor (Midgley 1989).
The observation of physical pollen transport can be used to estimate the
potential gene flow by pollen since transportation of pollen is an obvious
requirement for successful pollination and fertilization; however, the actual
gene flow by pollen also depends on the success of male gametes. Only gametes
incorporated into zygotes and eventually seeds and seedlings contribute to the
movement of genetic information. The measurement of actual gene flow
through pollen among plants gives estimates of distances between plants in
mating contact; hence, it provides crucial information for population delin-
eation. Modern methods to estimate actual gene flow through pollen are based
on the observation of genetic markers, often in two consecutive generations,
i.e., adults and their progenies.
Dispersal of Rare Alleles

A straightforward method to directly assess the gene dispersal of a male flow-

ering tree which is the only carrier of a unique allele is the observation of the
distribution of this allele in progenies of surrounding trees (Example 5.1). The
estimation is feasible for unique alleles at biparentally inherited, codominant
loci and rests on the assumption of regular segregation in the case of a het-
erozygous potential pollen parent. If the potential pollen tree is homozygous
for the rare allele, its success as a pollen parent can be reliably assessed by the
observation of the allele in the heterozygous state in the progeny of a seed tree
in the vicinity. In the case of heterozygosity of the potential pollen parent for
the unique allele, only half of its progenies carry the marker allele.

Example 5.1: Gene Dispersal in a Clonal Seed Orchard of Teak (Tectona grandis)
A tree in a clonal seed orchard of teak (Tectona grandis) comprising 16 clones
was found to be the unique carrier of a rare allele A2 at the isozyme gene locus
PGM (genotype A1A2). Since no other tree in the investigated compartment of
the seed orchard exhibited this allele, it is unlikely that the tree was a ramet of
5.2 Gene Flow Through Pollen 61


Relative success of marker tree as


1996 and 1997

pollen parent



20 40 60 80 100
Distance from marker tree (m)
Fig. 5.2. Success of a teak (Tectona grandis) tree as a pollen parent for progenies of seed
trees growing at different distances in a clonal seed orchard in Thailand. (Adapted from
Finkeldey 2006)

a selected clone. The distribution of the allele in progenies of surrounding seed

trees was observed in two consecutive years (1996 and 1997) and the mating
success of the marker tree was estimated for each seed tree. It was found that
the intensely flowering tree was successful as a pollen parent in its immediate
neighborhood, but that the success of pollination was low or absent in
seed trees growing a distance of more than 20 m from the marker tree
(Fig. 5.2). The estimate of the selfing rate of the marker tree in two consecutive
years (distance assumed to be 0 m in Fig. 5.2) is discussed in Example 6.1
(From Finkeldey 2006).
Paternity Analyses

The observation of unique alleles allows the paternity of progenies harvested

from a tree to be assigned to a particular pollen parent; thus, it is a particularly
simple case of paternity assignment, applicable only if rare, ideally unique, alle-
les are observed in reproducing populations. A more general approach is the
exclusion of paternity on the basis of the simultaneous observation of geno-
types of seed trees, their progenies, and a set of potential pollen parents at mul-
tiple gene loci. Ideally, all but one of the potential pollen parents can be
excluded for a particular progeny owing to an impossible combination of its
genotype, the genotype of the seed parent, and the genotypes of the potential
pollen parents. A potential pollen tree is excluded as an actual pollen parent of
62 C HAPTER 5 Gene Flow and Migration

a particular progeny if the tree does not carry an allele which was paternally
transmitted to the progeny at any gene locus. Obviously, the reliability of pater-
nity exclusion procedures increases with an increasing number of observed
gene loci and with increasing variability at the observed loci. Hypervariable,
codominant microsatellites (simple sequence repeats; Sect. 2.3.1) are partic-
ular suitable tools for paternity analyses (Example 5.2). The calculation of
paternity likelihoods by special software such as FAMOZ (Gerber et al. 2003)
is often preferable to simple exclusion procedures.

Example 5.2: Gene Flow Through Pollen in Neobalanocarpus heimii and

Dipterocarpus tempehes (Dipterocarpaceae)
Thirty potentially flowering trees of the dipterocarp Neobalanocarpus heimii and
248 of their progenies growing in the Pasoh Forest Reserve, Malaysia, were inves-
tigated at four microsatellite gene loci (Konuma et al. 2000). Seedlings and
saplings were collected within a radius of 10 m around five putative seed parents.
The putative seed parent was excluded as an actual seed parent for 16% of the
progenies; thus, seeds were also dispersed by an unknown vector. Gene flow by
pollen was observed on the basis of paternity exclusion procedures. The estimated
minimum average distance between the five seed trees and the successful pollen
parents was 524 m. All pollen parents were excluded for 71 out of the 248 proge-
nies investigated (28.6%); thus, these progenies originated from matings with
pollen parents out of the plot studied. N. heimii is mainly pollinated by stingless
bees (Trigona spp.) and honey bees (Apis spp.). The results of the paternity analy-
ses point towards the efficiency of bees to bridge large distances of, on average,
more than 500 m between pollen parents and seed trees (Konuma et al. 2000).
Pollen flow of Dipterocarpus tempehes (Dipterocarpaceae) was studied in two
years in Lambir Hills National Park, Sarawak, Malaysia (Kenta et al. 2004). Mass
flowering was observed in 1996; 1998 was a year of less-intensive flowering.
A total of 277 trees in four spatially distinct subpopulations in a 70-ha area as
well as 147 (from seed year 1996) and 188 (from seed year 1998) progenies,
respectively, were investigated at six simple sequence repeat loci. Average selfing
rates were low in both years (7% and 4% for 1996 and 1998, respectively), and
mean distances of pollen dispersal were similar and lower than the estimates for
N. heimii (222 m in 1996 and 192 m in 1998). Pollen flow among subpopulations
was slightly higher in 1996 (44%) than in 1998 (33%). The main pollinators in
1996 were giant honey bees (Apis dorsata), while several moths were suspected of
being chiefly responsible for pollen transport in 1998. These differences in the
main pollinators did not result in fundamental differences of the pollen disper-
sal patterns and outcrossing rates; however, the mean relatedness of progenies
from the same seed parent was slightly lower in 1996 than in 1998 (From
Konuma et al. 2000; Kenta et al. 2004).
5.2 Gene Flow Through Pollen 63

Example 5.3: Pollen Dispersal in Three Neotropical Tree Species with Different
Spatial Distribution Patterns on Barro Colorado Island
The impact of population density and population structure on pollen disper-
sal was studied for three insect-pollinated species growing on Barro Colorado
Island, an artificial island in the Panama Canal (Stacy et al. 1996). Paternity
exclusion was conducted for each species on the basis of the observation of
variation at five to nine polymorphic isozyme gene loci. All adult or subadult
trees in an area of 50 or 84 ha were sampled, and seeds were collected under
reproducing trees. Minimum estimates of pollen dispersal were calculated
assuming that the nearest not-excluded potential pollen parent according to
the paternity analysis was the true pollen parent.
No selfing was observed for any of the populations investigated, and con-
siderable distances between pollen parents and seed trees were bridged by the
pollinators; however, the pollination patterns differed among the species
studied and depended on the spatial distribution of trees.
The distribution of the 25 adult trees of Calophyllum longifolium
(Clusiaceae)was rather uniform throughout the area investigated (84 ha). The
nearest neighbor was rarely identified as a pollen parent in the case of spatially
isolated seed trees, and the successful pollen parent was separated from the seed
tree by a distance of at least 210 m for 62% of the progenies investigated. If
trees were widely separated, pollen flow ranged up to several hundred meters.
However, cross-pollination seemed to be frequent within the only cluster of
two directly neighboring trees, which could not be excluded as a pollen parent
for 72 and 73% of the progenies investigated from the other tree in this cluster,
Fourteen adults of Spondias mombin (Anacardiaceae) were counted in the
same area of 84 ha. Pollination was mainly within clusters, and the nearest
neighbor could not be excluded as a pollen parent for roughly 90% of the prog-
enies investigated for this species. On the other hand, a low level of gene flow
through pollen over distances of at least 300 m was also confirmed, and pollen
flow from outside the area studied was confirmed by the occurrence of rare
alleles in progenies which were not observed in any adult in the plot.
Only five of the 30 potentially reproducing trees of Turpinia occidentalis
(Staphyleaceae) in the 50-ha plot studied produced abundant seeds in the sea-
son studied (1993). These five trees occurred in two small clusters of two and
three trees with a diameter of 35 and 57 m, respectively. The clusters were sep-
arated by a distance of approximately 235 m. Pollen flow was mainly within
clusters, and pollen flow over a distance of more than 130 m was confirmed for
only one out of 115 progenies investigated.
These results point towards a considerable importance of the spatial population
structure on pollen dispersal by small insects in tropical forests. Predominance of
64 C HAPTER 5 Gene Flow and Migration

pollination among near neighbors was observed mainly within clusters and
small groups of trees. Pollen dispersal is more far-reaching and not restricted
to nearest neighbous in the case of a more random distribution of reproducing
adults in the forest (From Stacy et al. 1996).
Model-Based Estimates of Pollen Dispersal

The investigation of pollen dispersal on the basis of paternity analyses requires

the availability of highly polymorphic marker systems and a full inventory of
all potential pollen parents in the selected area. Alternative marker-based
approaches such as the neighborhood model (Adams 1992) and the TWOGENER
method (Smouse et al. 2001; Austerlitz et al. 2004) are less demanding with
regard to the degree of polymorphism and the sampling design. However, esti-
mates of pollen dispersal based on these methods rest on model assumptions
which are difficult to test and often unlikely to be fulfilled.
The investigation of a limited number of seed trees and of a sample of their
separately harvested and investigated progenies allows the assessment of the
allelic structures in successful pollen contributions. The heterogeneity among
pollen allele frequencies of single seed trees is not only a suitable measure of
deviation from random mating (Sect. 6.2.1), but is also used by the TWOGENER
method to estimate pollen dispersal parameters.

Example 5.4: Pollen Dispersal in Dinizia excelsa (Fabaceae)

Pollen dispersal was estimated on the basis of an analysis of five microsatellite
loci and the TWOGENER method for the Amazonian endemic tree Dinizia
excelsa (Fabaceae) in Brazil (Dick et al. 2003). Single-tree progenies were har-
vested in a highly disturbed, fragmented landscape dominated by pastures, and
in an undisturbed continuous forest. Selfing rates were slightly higher in dis-
turbed (s =14%) than in undisturbed (s =10%) conditions. Average pollination
distances were estimated to be 212 m in undisturbed forests, where Dinizia
excelsa occurs in very low effective densities (approximately 0.1 trees per
hectare). The estimate for average pollination distances was considerably
higher in the fragmented landscape (1,264 m).
Native stingless bees and beetles are the main pollinators in undisturbed
forests. Introduced African honey bees (Apis mellifera scutellata) are the prin-
cipal pollinators in the disturbed landscape. The different types of pollinators
result in considerably modified pollination and mating patterns. African honey
bees are efficient pollen vectors linking remnant trees in open areas with trees
growing in forest fragments (From Dick et al. 2003).
5.3 Migration of Seeds 65

Efficiency of Pollen Vectors for Gene Flow

The examples of marker-based studies on pollen transport and pollination dis-

tances point towards the efficiency of pollinators to bridge distances of at least
several hundred meters between reproducing trees in species-rich tropical
forests. Similar results were observed in many other comparable studies. For
example, Nason et al. (1998) reported large estimates of pollination distances
for central American figs (Ficus spp.; Example 7.2).
In summary it appears that pollination by animals (zoogamy) is an efficient
mechanism to facilitate contacts among widely dispersed, spatially, but not
reproductively isolated trees. Thus, population sizes may be large even though
the population density is low for the majority of plant species occurring in
tropical forests.

Migration of Seeds

Regeneration and colonization of new habitats requires the movement of dias-

pores. These processes also involve transportation of genetic information usu-
ally by means of seed migration both for gymnosperms and for angiosperms.
Diaspores are either the seeds themselves or complete fruits or parts of fruits
containing seeds. The seedling develops from one or several embryos within a

Seed Vectors

Seeds are dispersed by biotic and abiotic means. Water and wind are the prin-
cipal abiotic dispersal agents. Biotic seed dispersal is facilitated by animals.
Humans play an enormous role in todays distribution of plants and patterns
of genetic variation within species.
Abiotic Seed Dispersal

The diaspores of many seed plants contain structures promoting the dispersal
of seeds by the wind. The name of the tree family Dipterocarpaceae refers to
two wings attached to the fruits of many dipterocarps. However, some diptero-
carps have more than two wings, and the wings of several species are only
66 C HAPTER 5 Gene Flow and Migration

rudimentarily developed. The hairs attached to the light seeds of many

Bombacaceae promote their dispersal by the wind.
Another important abiotic means for the dispersal of seeds is water.
Hydrochory requires special adaptations to ensure the survival of seeds in
water. Seeds of trees growing along the coastline, along rivers, and in swamp or
peat forests are potentially dispersed by water.
The weight of diaspores is of crucial importance for their ability to be trans-
ported by the wind. Trees with light diaspores such as Eucalyptus spp. and
Casuarina spp. have a wide seed shadow, i.e., they disperse their seeds in a
large area. Barochorous trees produce heavy diaspores and have no efficient
means of abiotic or biotic dispersal; their seed shadow is narrow since dias-
pores are dispersed by gravity alone. Typically, pioneer species have a larger
seed shadow than species of later successional stages. Most barochorous species
belong to climax communities.
Biotic Seed Dispersal

Vertebrates, in particular birds and mammals, are by far the most important
biotic seed dispersers of tropical forest plants (Loiselle et al. 1996).
Endozoochorous seed dispersal requires a passage of seeds through the gas-
trointestinal tract of an animal. Otherwise, dispersal of seeds by animals is exo-
zoochorous. Seed dispersal of tree species with fleshy, juicy fruits is typically
endozoochorous. Exozoochorous dispersal is typical for nuts.
Most bird species and both flying (bats, flying foxes) and nonflying (rodents,
pigs, primates, and others) mammals disperse seeds. The seeds of the small
mountain tree Dunalia arborescens (Solanaceae) are dispersed by at least 16
different bird species in Jamaica (Cruz 1981).

Efficiency of Seed Dispersal

The type of seed vector and, in the case of biotic seed dispersers, their behav-
ior are crucial factors for the seed shadow of forest plants. A comparison
between seed dispersal agents in temperate and tropical American forests
revealed a dominance of endozoochorous seed dispersal only in the tropics.
Endozoochorous, exozoochorous, and wind dispersal are of similar impor-
tance in temperate forests of North America (Howe and Smallwood 1982). In
general, endozoochorous seed dispersal results in a wider dispersal of seeds as
compared with that from other mechanisms since the fruits containing seeds
are digested before the seeds are defecated. Endozoochorous seed dispersal may
be wide but clustered, for example, under the rest trees of seed-dispersing birds
5.4 Long-Distance Gene Flow and Migration in Tropical Forest Species 67

and bats, and may result in nonrandom dispersal patterns (Loiselle et al. 1996).
The comparatively low differentiation among populations of the rare
European wild service tree (Sorbus torminalis; Rosaceae) at maternally inher-
ited chloroplast DNA (cpDNA) has been attributed to the endozoochorous
dispersal of seeds by birds resulting in frequent migration among spatially sep-
arated stands (Oddou-Muratorio et al. 2001).
Limited dispersal of seeds results in complex relations between spatial dis-
tribution patterns and kinship relations. In general, neighboring trees are
expected to be more closely related to each other and hence to be genetically
more similar than spatially separated trees of the same population. The exis-
tence of family structures can be studied if spatially explicit genetic data are
available. Ng et al. (2004) found a significant spatial aggregation in small- and
medium-diameter classes for two dipterocarp species (Shorea leprosula and
Shorea ovalis ssp. sericea) in a 50-ha plot in the Pasoh Forest Reserve, Malaysia.
They explained the result by limited dispersal of both seed and pollen for
Shorea leprosula, and an apomictic mode of reproduction in the tetraploid
Shorea ovalis ssp. sericea.
Strong genetic differentiation among populations separated by less than 10
km from each other has been found for the Amazonian tree Corythophora alta
at cpDNA haplotypes (Hamilton 1999). Seed dispersal is very restricted in
Corythophora alta, resulting in areas as large as 10 ha founded by a single
maternal lineage.

Long-Distance Gene Flow and Migration in Tropical
Forest Species

Genetic structures of populations are the outcome of evolutionary processes

acting over long time periods. Global changes of environmental conditions in
the past did not only affect the distribution of species, but also patterns of
genetic variation within species. The most fundamental global events with con-
siderable implications for the distribution of species and their genetic variation
patterns were the Quaternary glaciations. The ice ages did not only have a far-
reaching impact on plants in areas directly affected by growing ice sheets and
glaciers, but changed the climate and the sea water levels worldwide (Sect. 8.2).
The impact of glacial refugia and postglacial recolonization on genetic varia-
tion patterns has been studied for numerous temperate species (Hewitt 2000),
including forest trees such as oaks (Quercus spp.; Petit et al. 2002). While it is
likely that the Quaternary climatic variability had a considerable impact on the
distribution of the regional flora and plant species also in the tropics (Spichiger
et al. 1995; Sect. 8.2), much less is known about migration patterns of single
species and the resulting genetic variation patterns in the central tropics.
68 C HAPTER 5 Gene Flow and Migration

Maternally inherited markers are particularly useful to analyze migration

patterns of plants. The distribution of maternally inherited cpDNA haplotypes
among Central American populations of Cedrela odorata is related to its migra-
tion history (Example 7.6). Strong differentiation of cpDNA haplotypes among
populations has also been observed in brazilwood (Caesalpinia echinata Lam.;
Lira et al. 2003). The species occurs in three spatially isolated regions, all char-
acterized by particular cpDNA haplotypes. This pattern points towards the
existence of separate glacial refugia as an explanation for the spatial isolation
of distribution areas even prior to the beginning of anthropogenic forest
fragmentation (Lira et al. 2003).
Very restricted migration between orangutan (Pongo pygmaeus) orang-utan
populations on the two different sides of the Kinabatangan river in Sabah,
Malaysia, was estimated on the basis of the observation of variation at nuclear
microsatellites (Goossens et al. 2005). Thus, the river is an efficient natural bar-
rier for the gene flow for the endangered great ape. However, genetic structures
point towards frequent migration between distant locations on the same side
of the river, indicating relatively free and long movement of orang-utans.

Recommended Literature

Gene flow in seed plants has been thoroughly reviewed by Levin and Kerster
(1974), Ellstrand (1992) and, with a focus on forest trees, Adams (1992). The
book by Proctor et al. (1996) describes many aspects of the reproductive biol-
ogy of plants with regard to pollination.
The availability of uniparentally inherited markers and hypervariable gene
loci has greatly promoted experimental studies in this field as pointed out by
Cruzan (1998), by McCauley (1995) for cpDNA polymorphisms, and by Chase
et al. (1996a) for biparentally inherited microsatellites with emphasis on trop-
ical forest trees. The impact of palaeoclimatic conditions on migration and
patterns of genetic variation has been reviewed by Hewitt (2001).
Mating Systems 6

The importance of the generation and use of scientifically based information

for genetic conservation is well illustrated by changes in views of mating patterns in
tropical trees (Boshier 2000).


Mating is possible only between individuals producing gametes of opposite

sex. The sexual system (Sect. 4.2.1) describes the possible mating events.
Mating among conspecific plants requires the movement of the male gamete to
the female gamete and, hence, gene flow (Sect. 5.2). Both the sexual system and
gene flow are closely related to the analysis of the actual mating events in a pop-
ulation. The genotypic structure of a progeny generation is crucially affected by
the successful mating events in a population. Random mating is a useful refer-
ence, although it is rarely achieved for plant species occurring in tropical
forests (Sect. 6.2). Selfing (Sect. 6.3) and mating among relatives result in
inbreeding; inbred forest trees frequently suffer from strong inbreeding depres-
sion (Sect. 6.4). The statement at the beginning of this chapter by Boshier
(2000) refers to the accumulation of scientific evidence that inbreeding and in
particular selfing is not the rule for tropical forest trees, as it was assumed prior
to the conduct of numerous case studies on mating patterns of tropical forest
trees. Most forest trees are either dioecious or self-incompatible (Sect. 6.5), and
even the majority of self-compatible trees in tropical forests are predominantly
outcrossed. The mating system of a species is not uniform, but can vary among
different populations and among years. The impact of the population density
and other environmental factors on the mating systems of tropical forest plants
is mentioned in Sect. 6.6.
70 C HAPTER 6 Mating Systems

Random Mating and Panmixis

Ideally, a population consists of a group of individuals producing progenies at

random, i.e., without any preferences for mating events between particular
plants or groups of plants. Random mating is a useful reference even though in
actual plant populations mating preferences for neighboring plants or plants
with overlapping flowering periods are likely to exist.
A prerequisite for random mating is the even distribution and success of male
gametes in the population. Thus, all seed parents receive homogeneous pollen
contributions in a population with a random mating system. The degree of the
deviations from random mating can be assessed by measuring the heterogeneity
among pollen contributions to single seed plants (pollen allele frequencies).

Heterogeneity of Pollen Allele Frequencies

The constitution of the successful paternal contributions to the progenies of a

particular seed plant is its genetically effective pollen cloud. The term
pollen cloud is illustrative for wind-pollinated species, but may also be used
for zoogamous species. Genetic differentiation among pollen allele frequen-
cies results from an unbalanced success of trees as pollen parents for the prog-
enies of particular seed trees as illustrated by a fictitious example in Fig. 6.1.
The pollen allele frequency of a seed tree is assessed by estimating the allelic
structure of successful pollen after an investigation of a sample of its progenies
(seeds). If the haploid megagametophyte and the corresponding embryo in the
seed of a gymnosperm are simultaneously analyzed, the pollen contribution
can be reliably assessed for each progeny (Mller 1976). Maximum-likelihood
estimates of pollen allele frequencies can be obtained if progenies from
angiosperm plants are analyzed (Gillet 1997).
The separate investigation of progenies from several seed trees of a single
population allows the differentiation among them to be assessed with methods
used for an analysis of population differentiation (Sect. 3.3.2). The differentia-
tion among pollen allele frequencies is the most important parameter used to
estimate pollen dispersal with the TWOGENER method (Sect. 5.2.3). This illus-
trates the close affinities between gene flow and mating system analyses.
Considerable differences with regard to the heterogeneity of pollen allele
frequencies at isozyme gene loci appear between populations occurring in pure
stands, populations occurring in intermediate density in tropical forests, and
rare species distributed in low density in tropical forests (Finkeldey 2002;
Fig. 6.2). Extraordinary low differentiation among pollen clouds was observed
6.2 Random Mating and Panmixis 71


Fig. 6.1. Mating of two trees in a population. Only conspecific trees in the forest are
shown. Tree I is fertilized by only two other trees contributing effective pollen in some-
what unequal proportions; therefore, the effective number of its male mating partners
is close to unity and the majority of its offspring are full-sibs. Neither can the effective
pollen cloud of tree II be considered representative of that of the virtual pollen cloud
of the population. The two trees share only one common male mating partner. Their
effective pollen clouds are expected to be markedly differentiated

in three pure Norway spruce (Picea abies) plantations in Germany (Finkeldey

1995). Even this low level of differentiation was statistically significant at most
loci in one or several populations, but no strong deviations from random mat-
ing were observed. Differentiation was slightly higher in three populations of
teak (Tectona grandis) in Thailand, all characterized by comparatively high
density (Finkeldey 2006).
Most studies on genetic differentiation among pollen allele frequencies were
conducted on Barro Colorado Island (Murawski and Hamrick 1991).
Significant, but moderate heterogeneity was observed for most tropical forest
species occurring in intermediate density (approximately one to ten trees per
hectare). The highest differentiation among pollen allele frequencies and, thus,
the strongest deviations from random mating were observed for species occur-
ring in low density (less than one flowering tree per hectare) in tropical forests
(Murawski and Hamrick 1991).
72 C HAPTER 6 Mating Systems







Picea abies I

Picea abies II

Picea abies III

Tectona grandis Nat. Pop.

Tectona grandis Plant.

Beilschmedia pendula

Quararibea asterolepis

Sorocea affinis I

Sorocea affinis II

Pterocarpus indicus

Trichilia tuberculata

Brosimum alicastrum

Cavanilessia platanifolia I

Cavanilessia platanifolia II

Cavanilessia platanifolia III

Ceiba pentandra

Platypodium elegans I

Platypodium elegans II

Tachigali versicolor
>50 Ind/ha 1-10 Ind/ha <1 Ind/ha

Fig. 6.2. Differentiation (FST = GST) among pollen allele frequencies for species occur-
ring in different densities in temperate and topical forests. (Adapted from Finkeldey

In summary, analyses of pollen allele frequencies confirm the expectation

that deviations from random mating are stronger for species and popula-
tions occurring in low density in tropical forests in comparison with
populations growing in pure forests.

Panmixis and HardyWeinberg Structures

A population is panmictic if random mating is realized and if no other evolu-

tionary factors (drift, selection, mutation, gene flow, and migration) have an
impact on the genetic structures at a particular locus. In this case, alleles are
randomly combined to genotypes and genotype frequencies are a simple func-
tion of allele frequencies according to the following equations (see Sect. 2.4.1
for a definition of Pij and pi):
Pii = pi2 for homozygote genotypes (Ai Ai ), (6.1)
Pij = 2 pi pj for heterozygote genotypes (Ai Aj ). (6.2)
Genotypic structures conforming to these equations are defined as
HardyWeinberg structures. A population is in HardyWeinberg equilibrium
at a particular gene locus, if the genotype frequencies at the locus can be com-
puted according to Eq. 6.1 for homozygotes and Eq. 6.2 for heterozygotes.
6.3 Selfing and Outcrossing Rates 73

Numerous effects may cause significant deviation from HardyWeinberg

structures. For example, a uniform excess of homozygotes as compared with
HardyWeinberg structures (Pii > pi2 for all alleles i at several gene loci) is
expected to result from inbreeding (Sect. 6.4).
Slightly more heterozygotes than expected under HardyWeinberg condi-
tions were observed at isozyme gene loci in adult trees of a subdivided popula-
tion of the pioneer species Cecropia obtusifolia in Mexico (Alvarez-Buylla et al.
1996); however, homozygotes exceeded the expectation in the seeds collected
after distribution, and in seeds collected from the soil. The proportion of
homozygotes decreased from seedlings to juveniles and adults. Mixing of par-
tially differentiated subpopulations owing to seed dispersal explains the
homozygote excess in seeds after dispersal better than inbreeding for this dioe-
cious species. Selection favoring heterozygotes possibly accounts for the
increase of heterozygosity with age (Alvarez-Buylla et al. 1996). A similar tem-
poral dynamics of genotypic structures with a significant excess of homozy-
gotes in particular in seeds and young seedlings has also been reported for
several other tropical tree species such as Ardisia escallonioides (Pascarella
1997), Symphonia globulifera (Aldrich et al. 1998), and Platypodium elegans
(Hufford and Hamrick 2003; Example 7.1).
The frequency of deviations from HardyWeinberg structures in popula-
tions of tropical forest trees is another indication that random mating is a
useful reference, but not the rule in tropical forests.

Selfing and Outcrossing Rates

Cosexual (hermaphroditic and monoecious) species produce gametes of both

sexes on a single plant; thus, the pollination of a female gamete by the male
gamete of the same plant is possible (Sect. 4.2.2; Fig. 4.2). The relative fre-
quency of these selfings among all successful mating events of a single plant
or a population is the selfing rate s, and its complement is the outcrossing rate
t = 1 s.
The selfing rate is a fundamental aspect of the mating system of a popula-
tion. Many studies were conducted to estimate selfing rates in populations of
tropical forest species (Table 6.2).
Most experimental studies aimed at the estimation of selfing rates were
based on the observation of biochemical or molecular gene markers.
Occasionally, other marker types have been used to estimate selfing rates of
trees. For example, the mangrove species Rhizophora mangle is a pre-
dominantly selfing species according to a study based on the observation of
chlorophyll deficiency as a color trait marker (Lowenfeld and Klekowski 1992).
74 C HAPTER 6 Mating Systems

Estimates of Selfing Rates Based on Rare Alleles

Selfing can be reliably assessed if a tree is homozygous at a biparentally inher-

ited, codominant gene locus for a unique allele Ax, i.e., if no other tree in the
population carries this allele. In this case, all progenies from selfing are homozy-
gous Ax Ax, and all progenies from outcrossing are heterozygous Ax Ay (with
y = i, j, k, . . .). However, the homozygous occurrence of a rare allele in a single
tree is an extremely unlikely and hence rare event. Usually, trees are heterozy-
gous for unique variants x (genotype Ax Ay). In this case, only a fraction of the
progenies from selfing can be reliably identified: All progenies homozygous for
the rare allele (genotype Ax Ax) arise from selfing. If regular segregation and ran-
dom fusion of gametes are assumed, the expected segregation within the selfings
Ax Ay Ax Ay is 0.25Ax Ax , 0.5Ax Ay , 0.25Ay Ay. Hence, the proportion of geno-
types Ax Ax (Pxx) among a sample of progenies investigated equals one quarter
of all selfings, and the selfing rate of the tree can be estimated as s = 4 Pxx.

Example 6.1: Estimates of Selfing Rates in Teak (Tectona grandis) Populations

A tree heterozygous for a unique allele A2 at the isozyme gene locus PGM-A was
identified in a clonal seed orchard of teak (Example 5.1). Genotypic structures of
progenies harvested from this tree were observed in two consecutive years (Table
6.1). On the basis of these observations, the selfing rate was estimated as s = 4
P22 = 4 0.060 = 0.240 in 1996 and s = 4 0.94 = 0.376 in 1997.
The method was also applied to estimate selfing rates of 24 teak trees carry-
ing a rare allele at a marker locus in natural and planted populations. Estimates
of the selfing rate range from s = 0 to s = 0.62 for single trees. Population esti-
mates based on a mixed mating model were between s = 0.07 and s = 0.17
(Finkeldey 2006).
Teak flowers are complete (Fig. 6.3a). A single flower is receptive and releases
its pollen during a few hours of a single day; however, flowering of an inflores-
cence (Fig. 6.3b) usually extends over several weeks since only a few flowers are

Table 6.1. Genotypic structures at the gene locus PGM-A in progenies of a teak tree
(genotype A 1A2) in a clonal seed orchard in Thailand in 1996 and 1997. (Adapted
from Finkeldey 2006)
N P11 P12 P22

1996 383 0.400 0.540 0.060

1997 245 0.429 0.477 0.094
N sample size
6.3 Selfing and Outcrossing Rates 75

a b
Fig. 6.3. Single flower (a) and inflorescence (b) of teak (Tectona grandis). (Photos:
R. Finkeldey)

receptive each day. The extended flowering of teak inflorescences reduces the
probability of selfing due to geitonogamy since pollinators find only a limited
number of receptive flowers available within single inflorescences and hence
are forced to move among inflorescences, and possibly trees.
Observations of the mating system in natural and breeding populations
confirmed that teak is a predominantly outcrossing species. The existence of a
system of self-incompatibility has been suggested for teak (Sect. 6.5.1), but
considerable levels of selfing were observed in all populations studied in
Thailand. High variation of the selfing rate of individual trees was observed.
Some trees are (almost) completely outcrossing, while surprisingly high esti-
mates of selfing rates were estimated for other trees. Significantly different self-
ing rates were estimated for the same tree in different seed years, suggesting an
impact of the environmental conditions during flowering (temperature,
activity of pollinators, etc.) on selfing rates in teak (Finkeldey 2006).
76 C HAPTER 6 Mating Systems

Estimates of Selfing Rates Based on a Mixed Mating Model

Selfing is associated with characteristic changes of genotypic structures in the

progenies, since it is the strongest form of inbreeding (Sect. 6.4). Thus, it is
possible to estimate selfing rates on the basis of an investigation of the prog-
enies from a sample of seed parents in a population. The estimation is based
on the assumption that each progeny results either from selfing or from ran-
dom outcrossing. Random outcrossing is rarely expected to be realized in
particular for tropical trees occurring in low density (Sect. 6.2). Accordingly,
assumptions of the mixed mating model are often violated, and results are
possibly biased. This applies in particular to estimates based on single marker
loci and individual seed trees. Multilocus estimates for populations are
expected to be more robust against violations of model assumptions.
Estimates of the multilocus outcrossing rate tm are most frequently reported
(Ritland and Jain 1981).
Most estimates of outcrossing rates of tropical trees were obtained from
studies based on moderately variable isozyme gene markers using the mixed
mating model. Population estimates of tm are shown in Table 6.2 for selected
tropical trees species. If more than a single population was investigated in a
particular study, the minimum and maximum estimates of tm are reported.
Outcrossing rates are close to 1 (or 100%) for the majority of the tropical tree
species investigated. Thus, selfing rates (s = 1 tm) are usually low, and even
self-compatible species with a mixed mating system are mainly outcrossing.
Exceptions are the self-compatible Cavanillesia platanifolia (Sect. 6.6), some
populations of Ardisia escallonioides (Pascarella 1997), and Pinus merkusii in
Thailand. Extraordinary low estimates of outcrossing rates for Pinus merkusii
were estimated in a few populations only. Overaging, degradation, and lack of
flowering synchrony are likely causes for a presumably low availability of for-
eign pollen in these populations (Changtragoon and Finkeldey 1995a). High
outcrossing rates were estimated for Pinus merkusii in a natural stand and seed
orchards in Indonesia (Siregar and Hattemer 2001).

Estimates of Selfing Rates Based on Nonmaternal Alleles

The observation of a nonmaternal allele Ak (k i, j) in a progeny of a seed

parent with genotype Ai Aj at any biparentally inherited gene locus is suffi-
cient to exclude selfing; thus, it is possible to calculate a minimum estimate
of the outcrossing rate by computing the proportion of progenies with a
6.3 Selfing and Outcrossing Rates 77

Table 6.2. Estimates of multilocus population outcrossing rates tm based on a mixed

mating model for tropical forest tree species
Species tm Reference

Acacia auriculiformis 0.6730.951 Wickneswari and Norwati (1995)

Ardisia escallonioides 0.3901.051 Pascarella (1997)
Astrocaryum mexicanum 0.9331.050 Eguiarte et al. (1992)
Beilschmelia pandula 0.918 Murawski and Hamrick (1991)
Bertholletia excelsa 0.849 OMalley et al. (1988)
Brosimum alicastrum 0.875 Murawski and Hamrick (1991)
Calophyllum longifolium 1.0301.031 Stacy et al. (1996)
Carapa guianensis 0.9670.986 Hall et al. (1994b)
Carapa procera 0.78 Doligez and Joly (1997)
Cavanillesia platanifolia 0.2130.661 Murawski and Hamrick (1992)
Cedrela odorata 0.969 James et al. (1998)
Ceiba pentandra 0.689 Murawski and Hamrick (1991)
Cordia alliodora 0.966 Boshier et al. (1995)
Dipterocarpus cf. condorensis 0.680.81 Luu (2005)
Eucalyptus citriodora 0.86 Moran and Bell (1983)
E. delegatensis 0.79 Moran and Bell (1983)
E. grandis 0.84 Moran and Bell (1983)
E. kitsoniana 0.77 Moran and Bell (1983)
E. obliqua 0.76 Moran and Bell (1983)
E. pauciflora 0.63 Phillips and Brown (1977)
E. regnans 0.69 Moran and Bell (1983)
E. rhodantha 0.580.60 Sampson et al. (1989)
E. saligna 0.77 Moran and Bell (1983)
E. stellulata 0.77 Moran and Bell (1983)
E. stoatei 0.82 Moran and Bell (1983)
Jacaranda copaia 0.982 James et al. (1998)
Pterocarpus indicus 0.908 Finkeldey et al. (1999)
Pterocarpus macrocarpus 0.719 0.959 Liengsiri et al. (1998)
Pinus merkusii 0.0170.843 Changtragoon and Finkeldey
Platypodium elegans 0.8980.924 Murawski and Hamrick (1991)
Quararibea asterolepis 1.008 Murawski and Hamrick (1991)
Shorea congestiflora 0.874 Murawski et al. (1994a)
Shorea megistophylla 0.710.87 Murawski et al. (1994b)
Shorea trapezifolia 0.5190.602 Murawski et al. (1994a)
Sorocea affinis 0.9691.089 Murawski and Hamrick (1991)
Spondias mombin 0.9891.304 Stacy et al. (1996)
Stemmadenia donnel-smithii 0.896 James et al. (1998)
Stemonoporus oblongifolius 0.84 Murawski and Bawa (1994)
Tachigali versicolor 0.937 Murawski and Hamrick (1991)
Trichilia tuberculata 1.077 Murawski and Hamrick (1991)
Turpinia occidentalis 1.0061.071 Stacy et al. (1996)
78 C HAPTER 6 Mating Systems

nonmaternal allele at any gene locus investigated. The estimation is a special

case of paternity exclusion (Sect. 5.2.3) with the seed parent regarded as the
only potential pollen parent. The reliability of the estimate increases with the
number of gene loci investigated and with the variability at these loci. Thus,
the method is preferably based on an investigation of several hypervariable
loci; nuclear microsatellites (simple sequence repeats) are particularly useful
to identify nonmaternal alleles owing to their codominance and high vari-
ability (Example 7.1).
High outcrossing rates were estimated for a Dipterocarpus tempehes popula-
tion (t = 0.93 and 0.96 in two different seed years) on the basis of paternity
exclusion and taking possible genotyping errors into account (Kenta et al.
2004; Example 5.2). Comparable results were obtained for the outcrossing rate
in an undisturbed natural forest of another dipterocarp, Shorea curtisii
(t = 0.96); however, estimates of individual outcrossing rates were considerably
lower for five Shorea curtisii tress in a selectively logged forest (t = 0.52;
Obayashi et al. 2002).

Inbreeding and Inbreeding Depression

Plants (and animals) are genetically related, if they have at least one common
ancestor. If two related plants mate, they possibly transmit a copy of the same
allele, which both of them may have inherited from their common ancestor, to
their progeny. In result, mating among relatives leads to a certain probability
that two alleles in the progeny are identical by descent (p. 179 ff. in Hedrick
2000); thus, mating among relatives results in an increased probability that the
two alleles transmitted to a progeny are identical, i.e., that the progeny is
homozygous at the respective gene locus. The same applies in the case of a
progeny resulting from gametes produced by the same parent after selfing.
Inbreeding is defined as mating among relatives, including selfing, which is
regarded as the strongest form of inbreeding. It results in increased homozy-
gosity as compared with matings among nonrelated plants. The increase in
average homozygosity due to inbreeding depends on the degree of relatedness
of the parents.

Genetic Consequences of Inbreeding

The average level of inbreeding in a population is frequently quantified by a com-

parison of the heterozygosity He expected under HardyWeinberg conditions
6.4 Inbreeding and Inbreeding Depression 79

(Sect. 6.2.2) and the observed heterozygosity Ho (see Sect. 3.3.1 for a computa-
tion of He and Ho). It holds for the inbreeding coefficient F
F = 1 Ho/He. (6.3)
In case of HardyWeinberg structures, Ho equals He, resulting in F = 0. If the
observed heterozygosity exceeds the expected heterozygosity, F becomes nega-
tive. Negative values of F cannot be explained by inbreeding but have other
causes, for example, selection favoring heterozygotes at the gene locus investi-
gated. If fewer heterozygotes were observed than expected for corresponding
HardyWeinberg structures, F becomes positive. If positive inbreeding coeffi-
cients F are observed at several gene loci, and if the values for F are homoge-
neous across loci, inbreeding is a plausible explanation for the observed
genotypic structures.
Examples of inbreeding coefficients F observed in populations of tropical
forest plants are shown in Table 6.3. Weak or no impact of inbreeding on geno-
typic structures is suggested by low inbreeding coefficients for Pterocarpus indi-
cus, D. condorensis, Dipterocarpus cf. condorensis, and Swietenia macrophylla.
The inbreeding coefficient was low, but significantly different from zero in four
out of seven populations of Swietenia macrophylla studied (Lemes et al. 2003;
Sect. 3.5.3; Table 3.4). The genotypic structures of Ancistrocladus korupensis
(Foster and Sork 1997) and the five Pinanga species studied by Shapcott (1999)
are strongly influenced by inbreeding. Very high selfing rates and preferential
mating among neighboring, closely related trees are likely to account for the
high inbreeding coefficients observed for both species.

Table 6.3. Average inbreeding coefficients F = 1 Ho/He observed in populations of

tropical forest plants
Species Plant group Marker type F Reference

Ancistrocladus korupensis Liana Isozyme 0.455 Foster and Sork

Dipterocarpus cf. Timber tree Isozyme 0.051 Luu (2005)
Pinanga spp. (5 species) Palm Isozyme 0.6400.856 Shapcott (1999)
Pterocarpus indicus Timber tree Isozyme 0.015 Finkeldey et al.
Pterocarpus macrocarpus Timber tree Isozyme 0.099 Liengsiri et al.
Swietenia macrophylla Timber tree Microsatellite 0.038 Lemes et al.
80 C HAPTER 6 Mating Systems

Inbreeding Depression

The loss of heterozygosity as a consequence of inbreeding is often reflected at

phenotypic traits of adaptive or economic significance. Many complex traits,
including survival, growth rate, and quality traits such as stem straightness, are
negatively influenced by inbreeding of predominantly outcrossing species. The
negative impact of inbreeding on phenotypes and the fitness is defined as
inbreeding depression. Predominantly or exclusively selfing species, including
many agricultural crops such as rice and soy bean, do not suffer from inbreed-
ing depression since inbreeding does not further reduce the level of heterozy-
gosity for these species.
Inbreeding depression can be explained by a slight superiority of heterozy-
gotes in comparison with homozygotes at many gene loci (overdominance
hypothesis; Ziehe and Roberds 1989) or by a strong inferiority of particular
homozygotes for deleterious alleles at a few adaptive gene loci (dominance
hypothesis; p. 220 ff. in Allard 1960). Deleterious recessive alleles seem to be a
more important cause of inbreeding depression in plants than overdominance
(Ritland 1996a).
Most tropical forest tree species are highly variable (Sect. 3.4; Tables 3.2, 3.3)
and predominantly outcrossing (Sect. 6.3; Table 6.2); thus, the observed het-
erozygosity is usually high, and strong inbreeding depression is expected for
most trees regardless of the exact cause of the negative impact of low levels of
(average) heterozygosity on fitness and phenotypic traits (Ledig 1986b).
Inbreeding depression acts on numerous traits affecting fitness throughout
the life cycle of plants. Inbreeding effects have been studied particularly well for
eucalypts (Example 6.2) and a few other species grown in plantations. It is dif-
ficult to investigate inbreeding depression in natural and naturally regenerated
forests (but see Example 7.1), but a negative effect of inbreeding on fitness is
likely to be a main cause for the evolution of efficient pollen dispersal mecha-
nisms (Sect. 5.2) and of incompatibility systems (Sect. 6.5) for many tropical
forest plants. Thus, the disruption of mating contact among nonrelated, widely
separated plants poses a main threat to populations since it results in increased
inbreeding (Finkeldey 2002).

Example 6.2: Inbreeding Depression in Eucalypts

Eucalypts (Eucalyptus spp.) are hermaphroditic, self-compatible trees. A mixed
mating system is typical for eucalypts. Outcrossing usually dominates, but self-
ing rates of 2040% are frequently encountered (Table 6.2). Inbreeding depres-
sion of eucalypts has attracted considerable interest since it directly affects the
performance of eucalypt plantations.
6.4 Inbreeding and Inbreeding Depression 81

Inbreeding depression during an early life stage was studied for E. dele-
gatensis (Moran and Brown 1980). Fruits of this species remain on seed trees
for several years, allowing the investigation of seed cohorts of different age. The
outcrossing rate tm was estimated in 1-, 2, and 3-year-old seed cohorts on the
basis of isozyme studies and a mixed mating model with the following result:
tm = 0.66 (1-year-old seeds), tm = 0.78 (2-year-old seeds), tm = 0.85 (3-year-old
seeds). Thus, outcrossing seems to increase and, hence, selfing to decrease with
increasing age. The most likely reason for the effect is early mortality of inbred
seeds. Preferential survival of outcrossed progenies is a plausible explanation
for the apparent increase of the outcrossing rate.
The growth of progenies from controlled selfing, open pollination, and con-
trolled outcrossing was compared for seed parents of E. regnans in a field trial
in Australia. Survival and height growth were only slightly affected by selfing
during the early development of the trial, but strong and highly significant dif-
ferences were observed 12 years after planting. Only 31.2% of the selfed trees
survived; their average diameter was 20.4 cm. This compares with 71.9% sur-
vival with an average diameter of 24.2 cm for the open-pollinated trees. The
completely outcrossed family was even superior to the trees from open polli-
nation (Griffin and Cotterill 1988).
Many studies proved strong inbreeding depression in E. grandis. Inbreeding
results in inferior growth of young plants and has a negative impact on quality
traits like stem form (Hodgson 1976a). Inbreeding depression in E. grandis is
not restricted to selfing, but has also been described for matings among full-
and half-sibs (van Wyk 1981). Its strength depends on the relatedness of the
parent trees.
The growth performance of E. grandis progenies from single seed trees after
open pollination is correlated to the outcrossing rate tm of the seed parents esti-
mated by a mixed mating model on the basis of isozyme investigations
(Burgess et al. 1996). The mean family heights of 20 open-pollinated families
with high estimates of outcrossing rates (tm) were quite uniform for 2.5-year-
old plants (Fig. 6.4). Some families with low estimates of outcrossing rates
reached comparable mean heights, but the mean height growth of most fami-
lies with low tm was considerably below the mean for the reference populations
with high tm. Thus, the mean growth rate of a family is negatively affected by a
low outcrossing rate, presumably owing to inferior growth of selfed progenies.
The strong inbreeding depression of eucalypts has considerable practical
importance. Eucalypt plantations are established worldwide on a large scale
(Eldridge et al. 1994), and the origin of reproductive material is often
unknown. Most eucalypts flower and fruit prolifically and regularly even if they
grow as isolated individuals or in small stands. Uncontrolled harvest of seeds
from isolated trees or small stands poses a risk of producing inbred reproduc-
tive material which is expected to suffer from strong inbreeding depression.
82 C HAPTER 6 Mating Systems

Families with low outcrossing rates Families with high outcrossing rates

Mean family height (m)


0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Family outcrossing rate
Fig. 6.4. Family mean height of 2.5-year-old Eucalyptus grandis seedlings in relation to
an estimate of the family outcrossing rate tm for 40 families with contrasting estimates
of tm. (From Burgess et al. 1996)

Negative effects from inbreeding depression are particularly likely to occur if

seeds are harvested from well-performing, but isolated trees or small stands
close to villages for the establishment of small-scale plantations in rural areas.

Incompatibility and Self-Sterility

Since most tropical forest trees are hermaphroditic, selfing by means of auto-
gamy and geitonogamy is possible. However, selfing is the strongest form of
inbreeding and is associated with strong inbreeding depression. In consequence,
mechanisms preventing selfing offer considerable evolutionary advantage. The
most important mechanisms acting prior to fertilization are incompatibility sys-
tems. Cosexual species without incompatibility reactions often evolve systems of
self-sterility which result in early abortion of embryos after fertilization.


The genotypes of two plants which are able to produce a common progeny are
compatible. If plants belong to different biological species, they are incompatible
6.5 Incompatibility and Self-Sterility 83

since they cannot produce common progenies. On the other hand, incompati-
bility also often occurs between gametes produced by a single cosexual plant. For
example, controlled pollination of Gmelina arborea (Verbenaceae) is not
successful if pollen from the seed parent is used (Bolstad and Bawa 1982). This
suggests that G. arborea is completely self-incompatible. Observations of the
flower development after pollination suggest that teak (T. grandis Tectona
grandis), another important timber species of the Verbenaceae, is also at least
partially self-incompatible (Tangmitcharoen and Owens 1997); however, con-
trolled self-pollination in teak results in low success rates (Bryndum and
Hedegart 1969), and marker-based analyses of the mating system after open
pollination revealed that teak is a mixed mating species (Example 6.1).
Self-incompatibility is controlled by a single locus or only a few incompati-
bility loci (S-loci). The incompatibility usually results from the inability of
pollen to germinate on the pistil of particular plants, or from inhibition of the
growth of the pollen tube. The incompatibility system does not only exclude
self-pollen from successful fertilization since the genetic information at the
S-locus is crucial for the reaction.
Two main types of incompatibility systems are distinguished (Boshier
2000). In the case of gametophytic self-incompatibility, the genetic constitu-
tion of the male gametophyte at the S-locus is decisive for the incompatibility
reaction. Pollen is excluded, if it carries any of the two alleles of the seed par-
ent. Thus, self-pollination is impossible just like pollination between two
related or unrelated trees with identical genotypes at the S-locus. If two trees
share only one allele at the S-locus, they are semicompatible since half of their
pollen is compatible but the other half is excluded from pollination success.
Obviously, all plants are heterozygous at the S-locus, and populations need at
least three alleles at this locus in order to be able to produce progenies. Usually,
the allelic diversity at the S-locus is very high.
The genotypic information of the pollen parent at the S-locus, and not only
the allelic constitution of haploid pollen, is decisive for the incompatibility
reaction in the case of sporophytic systems. Two plants with identical geno-
types at the S-locus are incompatible; thus, selfing is obviously impossible.
Sporophytic systems are often heteromorphic with regard to flower morphol-
ogy. For example, two types of flowers are observed in case of distyly. Flowers
have either a long style and short stamina, or they are short-styled with long
stamina. Pollen from short stamina is excluded in long-styled flowered and vice
versa. Since plants produce only one of the two flower morphs, they are self-
incompatible. Many other mating combinations are also excluded. Distyly is
controlled by a sporophytic incompatibility system with only two alleles and
complete dominance of one of the alleles.
An incompatibility system is regarded as cryptic if selfing and other forms of
incompatible matings are disadvantageous, but not completely excluded. For
84 C HAPTER 6 Mating Systems

example, cryptic incompatibility has been reported for Brazil nut (Bertholetia
excelsa) (Moritz and Ldders 1993). Cryptic incompatibility results in a strong
dominance of outcrossing, if foreign, fully compatible pollen from other trees
is available, but does not preclude the production of seeds by sexual processes in
the case of the absence of foreign pollen.
Incompatibility systems are very common among hermaphroditic plants in
tropical forests, but the genetic control has rarely been investigated. Gametophytic
incompatibility has been reported for Inga spp. (Koptur 1984) and Acacia retinodes
(Kenrick and Knox 1985). Cordia pringlei and Cordia dentata are distylous (Opler
et al. 1975). Other species of the genus Cordia (Boraginaceae) are dioecious
(C. inermis, C. collococca, and C. panamensis), and Cordia alliodora is self-compat-
ible, but predominantly outcrossing (Boshier et al. 1995; Table 6.2). The system of
self-incompatibility is particularly complicated in cacao (Theobroma cacao).
Several loci are involved; the genetic control is both sporophytic and gametophytic.
Most trees in the South American region of origin of cacao are self-incompatible,
but there are also self-compatible varieties (Warren et al. 1995).
Bawa (1974) estimated a frequency of 54% of self-incompatible hermaph-
rodites among 130 species investigated in a tropical lowland rain forest in Costa
Rica. Since 22% of the species in this community are dioecious, 76% are likely
to be obligatory outcrossing. The remaining trees are either self-compatible
hermaphrodites (14%) or monoecious (10%).


Postzygotic effects of selfing have been studied in detail for many conifers
(p. 67 f. in Stern and Roche 1974; Savolainen et al. 1992). The flower morphol-
ogy of gymnosperms does not permit the selective discrimination of particular
pollen types. The early abortion of zygotes from selfing is not controlled by a
single locus or a few gene loci, but is the effect of many interacting genes. Thus,
self-sterility may be regarded as a very early form of inbreeding depression act-
ing on young zygotes and embryos and resulting in strong viability selection
against inbred genotypes. The dominance of angiosperms and the frequent
observation of incompatibility systems in tropical forests suggest that systems of
self-sterility are of lesser importance.

Environmental Effects on Mating

The mating system is not fixed or exclusively determined by genetic factors

within a species, but it is also influenced by environmental conditions; thus,
6.7 Recommended Literature 85

human impact on environmental conditions, for example, by means of forest

fragmentation and silvicultural management, has considerable implications for
the realized mating systems (Finkeldey 2002). Environmental variation is com-
mon both in managed and in unmanaged forests and hence the mating system
is likely to vary both in time and in space.
An effect of population density on the mating system is evident from com-
parisons of the heterogeneity of pollen allele frequencies among species occur-
ring in different density in tropical forests (Sect. 6.2.1). In general, low
population density results in strong deviation from random mating and strong
differentiation among pollen allele frequencies of single seed trees owing to a
restricted number of effective pollen parents contributing to the production of
progenies from single seed trees.
The reproduction effective population density has an influence on selfing
rates of the self-compatible Cavanillesia platanifolia (Murawski and Hamrick
1992). The outcrossing rate of a population studied on Barro Colorado Island
in Panama was very low in a year with only a few flowering trees (32% of cen-
sused trees that flowered: tm= 0.213; estimates based on isozyme variation and
a mixed mating model, see Sect. 6.3.1), but increased when more trees partici-
pated in reproduction in other seed years (49% of censused trees that flowered:
tm= 0.347; 74% of censused trees that flowered: tm= 0.569). The estimate of the
outcrossing rate was tm= 0.661 in a reference population with higher density.
No obvious impact of population density and stand structure on outcross-
ing rates was observed in five populations of the endemic Dipterocarpus cf. con-
dorensis in Vietnam, but the populations varied with regard to the estimates for
the effective number of pollen parents fertilizing single seed trees (Luu 2005).
An impact of different spectra of pollinators on gene flow and the mating
system may be assumed for species with a wide spatial distribution and species
occurring in different ecosystems. However, very little is known about the
consequences of different natural pollination environments on the mating
system of tropical forest trees.

Recommended Literature

The contribution of Boshier (2000) to the book edited by A. Young et al. (2000)
gives a concise introduction to many aspects of plant mating systems, includ-
ing inbreeding and incompatibility. The genetic consequences of inbreeding in
plants were discussed by Ritland (1996a). Inbreeding depression has been
reviewed with emphasis on conifers by Williams and Savolainen (1996) and
with a focus on tropical forest trees by Griffin (1990). Incompatibility is
discussed in detail in the book of de Nettancourt (2001).
Adaptation and Coevolution 7

The results of natural selection, the evolutionary force most responsible for adaptation
to the environment, are evident everywhere, yet it is remarkably difficult to observe the
time course of changes brought about by selection (p. 49 in Gillespie 1998).


Plants are able to survive and to reproduce in different environments. The

longevity and immobility of trees implies that they need to withstand different
environmental conditions during their lifetime. Thus, adaptation of trees to
their environment is ubiquitous and necessary for survival.
Adaptedness implies a high likelihood of survival and reproduction in a
given environment. Adaptation is the process leading to a state of adaptedness
in response to environmental variability. The adaptive potential or adaptability
describes the ability to achieve a state of adaptedness to a range of different
environmental conditions (p. 59 in Eriksson and Ekberg 2001).

Physiological and Evolutionary Adaptation

Since adaptation is a general aspect in the life cycle of plants it is important to

distinguish different strategies and levels for adaptive processes. Each plant
reacts to changing environmental conditions by adjustment of its metabolism.
These physiological adaptations or phenotypic plasticity (Schlichting 2002)
of plants is eventually manifested in morphological features such as sun- and
shadow-leaves. Many plants can be cloned and copies of the same genotype can
be grown in different environments. This allows the phenotypic response of
genotypes to different environments (their norms of reaction) to be assessed
and compared. Clonal tests revealed large differences of physiological adapt-
ability among genotypes for many tropical trees (Foster and Bertolucci 1994).
88 C HAPTER 7 Adaptation and Coevolution

Different norms of reactions are not only observed among clones, but are also
the rule among genetically heterogeneous demes such as families or prove-
nances. The use of genotypes in plantation forestry requires their adaptedness
to the respective site conditions. Thus, traditional tree improvement (Chap. 12)
is based on differences of the physiological adaptedness among demes com-
pared in field trials although the domestic fitness rather than the natural fit-
ness of genotypes and demes matters in applied tree breeding (p. 94 ff. in
Eriksson and Ekberg 2001).
Drastic environmental change results in the loss of adaptedness of particu-
lar genotypes. Preferential survival or higher reproductive capacities of
adapted genotypes result in changes of genetic structures owing to viability
selection or fertility selection. Evolutionary adaptations are caused by selective
changes of genetic structures; hence, the population, and not the single plant,
is the basic entity to investigate evolutionary adaptations, which can take place
only in genetically diverse populations.
There is no doubt that evolutionary changes of genetic structures due to via-
bility and fertility selection contributed to the adaptation of trees to their
respective environments in boreal, temperate, and tropical forests. Most tropi-
cal tree species are highly variable, and most of the genetic variation at single
loci resides within populations (Chap. 3). Thus, evolutionary adaptations were
not only involved in the creation of the high species diversity in tropical forests,
but they continue to play an important role for the reactions of forest ecosys-
tems to environmental change. The urgent need to conserve genetic resources
of tropical forest trees (Chap. 14) is motivated by the importance to maintain
evolutionary adaptive potentials in tropical forests.
Selection primarily acts on single gene loci; thus, it is much more difficult to
investigate in natural ecosystems than evolutionary factors affecting all gene loci
in principally the same way. For example, drift results in losses of genetic varia-
tion at all loci, and inbreeding reduces heterozygosity at all biparentally inher-
ited, nuclear genes. The dynamics of genetic structures at putatively neutral
gene loci can be used to monitor selective processes due to associations between
selected loci and marker loci (Finkeldey and Ziehe 2004). Comparisons of
genetic structures at isozyme gene loci between tolerant and sensitive
adult beech (Fagus sylvatica) trees in environmentally stressed forests in
Germany suggest a selective response of trees to drastic environmental change
(Mller-Starck 1985; Ziehe et al. 1999).
The potential for drastic changes of genetic structures due to viability selec-
tion without encountering critically low population sizes is highest during early
life stages of trees since most trees produce a large number of progenies.
Changes of genetic structures were observed during the development of beech
seedlings in greenhouses and under field conditions (Ziehe et al. 1999). Viability
selection was also observed during early life stages of the tropical tree
7.2 Physiological and Evolutionary Adaptation 89

Platypodium elegans (Fabaceae; Example 7.1); however, the selection is mainly

against inbred progenies (Sect. 6.4.2) and, thus the observed changes of genetic
structures essentially result from inbreeding depression (Hufford and Hamrick
2003). These investigations suggest that it is in principle feasible to monitor
adaptive processes at putatively neutral markers even in long-living organisms
like forest trees.
Attempts to identify single genes of adaptive significance (Remington and
Purugganan 2003) and to relate genetic variation at these genes to phenotypic
trait expressions in field trials and to different environmental conditions at the
growing sites of populations are hampered by a large number of candidate
genes and a poor understanding of their physiological significance for most for-
est trees. The knowledge of the complete genome sequence of poplar (Populus
spp.) (Brunner et al. 2004) and a rapidly growing number of expressed gene
sequences in other trees (e.g., Pinus spp., Cryptomeria japonica; Strabala 2004)
is an important basis to significantly improve our understanding of adaptive
processes in forest trees. A draft genomic sequence of Eucalyptus camaldulensis
might become available in 2007 (Poke et al. 2005). The first results on sequence
variation in candidate genes of putative adaptive significance have been pub-
lished (Krutovsky and Neale 2005). Simple markers such as single nucleotide
polymorphisms will be developed characterizing the variation in genes with
putatively adaptive function in model species and related taxa (Morin et al.
2004; Sect. 2.3.1).
Unfortunately, no comparable data are currently available with regard to
adaptive changes of genetic structures at single gene loci for most tropical for-
est trees. Genomic resources will only be available for few tropical tree taxa like
eucalypts (Poke et al. 2005) in the foreseeable future. Our current understand-
ing of evolutionary adaptation in tropical forest trees mainly relies on the
observation of phenotypic variation in field trials; thus, it is restricted to a
rather low number of plantation species, and refers to the phenotypic conse-
quences rather than to the genetic basics of adaptation. Many trials have
proved strong differentiation among provenances, families, and clones with
regard to important adaptive traits, including survival (Chap. 12).

Example 7.1: Viability Selection During Early Life Stages of Platypodium elegans
Genetic structures at four hypervariable microsatellite loci were observed at
three different early life stages (aborted fruits, mature seeds, and seedlings from
the same cohort) of progenies from 12 P. elegans (Fabaceae) trees and a popu-
lation of adults. The trees grow on Barro Colorado Island, an artificial island in
the Panama channel. A slight increase in heterozygosity was observed from
mature seeds to seedlings at two loci. The average observed heterozygosity
increased from Ho = 0.750 in aborted fruits to Ho = 0.761 in mature seeds,
90 C HAPTER 7 Adaptation and Coevolution

Ho = 0.768 in seedlings, and Ho = 0.797 in adult trees. This suggests a slight

selective advantage of heterozygosity. The inbreeding coefficient F (Sect. 6.4.1)
decreased with increasing observed heterozygosity. The outcrossing rate t was
estimated for each cohort on the basis of the observation of nonmaternal alle-
les (Sect. 6.3.1). It increased from t = 0.75 for aborted fruits to t = 0.78 for
mature seeds and t = 0.87 for young seedlings.
Comparisons of genetic structures during early life stages of P. elegans
revealed selection against homozygotes. However, the observed increase of het-
erozygosity with increasing age is likely due to the selective disadvantage of
inbred progenies and not related to a direct heterozygote advantage at the
observed microsatellite loci. Thus, inbreeding depression results in changes of
genetic structures due to selection in particular during the phase from mature
seeds to the establishment of seedlings. The observed genotypic structures in
adults suggest that the selective disadvantage of inbred progenies continues
and leaves an essentially completely outcrossed adult population with
HardyWeinberg structures (Sect. 6.2.2). (From Hufford and Hamrick 2003)

Species Interactions and Coevolution

Plants do not only need to be adapted to their chemical and physical environ-
ment, but also to interacting organisms. The biological environment is partic-
ularly complex for plant species in species-rich tropical forests. It consists of
other higher plants, animals, fungi, and microorganisms which interact either
positively (symbioses) or negatively (pests and diseases) with plants.
Interactions with animal pollinators are particularly important from a genetic
point of view (Chap. 5). The coevolution between tropical figs (Ficus spp.) and
fig wasps (Agaonidae) is an interesting and well-studied example of a coevolu-
tion between plants and their animal pollinators (Example 7.2).
The impact of pests and diseases on the distribution of forest trees and their
regeneration has been suggested as one of the main explanations for the high
species diversity in tropical forests (Baker 1970; Janzen 1970; Hill and Hill
2001). Breeding for disease resistance is an important goal in tree improvement
(Namkoong 1991). It is based on genetic variation with regard to the suscepti-
bility to particular diseases, which is commonly observed in forest trees. The
preservation of disease-resistant genotypes is an important objective in con-
servation of forest genetic resources (Byrne 2000).
Symbiotic associations between trees and mycorrhiza constitute an important
aspect of biotic interactions in tropical forests and possibly explain the domi-
nance of particular species in some low-diversity tropical forests (Connel and
Lowman 1989). Little is known about mutual preferences between genotypes of
7.3 Species Interactions and Coevolution 91

plants and their mycorrhiza in tropical forests. A comparison of two orchid

species revealed different preferences for particular mycorrhizal fungi. Tolumnia
variegata is a generalist in its association with mycorrhiza; Ionopsis utricularoides
is more specialized with regard to its fungal symbionts (Otero et al. 2004).

Example 7.2: Figs and Their Pollinators

The pantropical genus Ficus (Moraceae) is one of the most species-rich genera
of woody tropical plants, with more than 800 species. The distribution center
is in Asia, with more than 500 species.

Figs are Keystone Species

Tropical figs produce fruits throughout the year and are an important diet of
many animal species. Figs are an indispensable part of the nutrition of many
birds and mammals such as bats and primates particularly during periods of low
availability of fruits from other plants (Janzen 1979a); thus, figs are regarded as
keystone species since their disappearance would have a profound impact on the
structure and function of many ecosystems (Terborgh 1986). Case studies proved
the crucial importance of figs for ecosystems in particular in Asia (Leighton and
Leighton 1983) but not in African forests (Gautier-Hion and Michaloud 1989).

Morphology of Figs and Their Pollinators

Figs evolved many different growth habits: free-standing trees, shrubs,
climbers, epiphytes, and hemiepiphytes (strangling figs). The root systems of
strangling figs germinating on a particular host tree tend to fuse; thus, many
adult strangling fig trees are composed of several distinct genotypes they are
chimeras (Thomson et al. 1991). Population sizes of strangling figs are often
underestimated since a single tree may be a small population (Hamrick and
Murawski 1991).
The inflorescence of figs (syconium) is fleshy and folded inside to form a
cave (Fig. 7.1). Many simple, male or female flowers are inserted on the inner
wall inside the cave, which is closed by apical bracts leaving a small entrance for
pollinating insects, the ostiole (Verkerke 1989). Figs produce two types of
female flowers: short-styled gall flowers and long-styled flowers.
With only a few exceptions (Molbo et al. 2004) each fig species lives in oblig-
atory symbiosis with a single wasp species, which is the sole pollinator (Fig. 7.2);
thus, the number of species of fig wasps (Agaonidae) roughly equals the num-
ber of fig species. Molecular phylogenies based on chloroplast and mitochondr-
ial genes revealed similarities between the evolutionary relationships among figs
and among their pollinators, suggesting a true coevolution and a high stability
of the symbiosis (Herre et al. 1996). Similar phylogenies were also found for pol-
linating fig wasps and associated, nonpollinating wasps (Machado et al. 1996).
92 C HAPTER 7 Adaptation and Coevolution

Fig. 7.1. Inflorescence (A) and single male flowers (B1), short-styled gall flowers (B2),
and long-styled female flowers (B3) of a monoecious fig (Ficus spp.). (Adapted from
Hill 1967a)

Fig. 7.2. Female and male Blastophaga quadraticeps,

a fig wasp, the pollinator of Ficus spp. (Adapted from
Barth 1982)
7.3 Species Interactions and Coevolution 93

The Reproduction Cycle of Figs and the Life Cycle of Fig Wasps
The reproduction cycle of figs and the life cycle of the pollinators are closely
connected (Janzen 1979a; Wiebes 1979; Bronstein and McKey 1989; Fig. 7.3).
Female wasps are attracted by pheromones and enter the syconia of a recep-
tive, female flowering syconium, through the ostiole. They fertilize the
long-styled female flowers with pollen brought from another fig of the same
species and lay eggs through the style of short-styled gall flowers. Afterwards
they usually die within the syconium; only a few wasps manage to leave a syco-
nium they entered (Giebernau et al. 1996). The emerging larvae feed on the
developing seeds of their gall flowers, thereby destroying it. The heterostyly
within syconia allows the simultaneous development of seeds and larvae of
wasps in different flowers of the same inflorescence.
A new generation of wasps develops within a few weeks. Male insects appear
first, fertilize the females, and die shortly thereafter inside the syconium. The male
fig flowers shed their pollen only after the female fig wasps have emerged. Thus,
figs are distinctly protogynous, i.e., female flowering is earlier than male flowering.
The fertilized female fig wasps take up pollen actively or are passively loaded with
pollen. They leave the syconium and start searching for another syconium in the
receptive phase. The seeds continue to develop after the female wasps have left
the syconium. The shape and the taste of a syconium change when the seeds are
ripe. The seeds are endozoochorously dispersed by mammals and birds.

Fig Wasp

flower flower gall flower wasp wasp

receptive developed enters syconium

pollination, wasp
fertilization lays eggs

wasp eggs, larvae


imago imago


flowering pollen uptake

seeds leaves syconium,


Fig. 7.3. The reproduction cycle of monoecious figs, and the life cycle of fig wasps
94 C HAPTER 7 Adaptation and Coevolution

Flower Phenology
The flowering cycle of a single fig plant is usually strictly synchronized; all syco-
nia of a plant are in the same stage of floral development. Only a few fig species
growing in a seasonal climate show a certain degree of individual asynchrony
(Cook and Power 1996). Thus, emerging wasps do not find receptive syconia on
the same plant but have to search for other trees of the same species. The wasps
survive only for a few days outside of syconia. In consequence, asynchronous
flowering of the fig plants of a population is a necessary requirement for the
preservation of the pollinators (Fig. 7.4). Asynchronous flowering results in the
production of fruits throughout the year. The importance of figs as a more or
less continuously available energy source for vertebrates is explained by the need
to produce flowers as breeding grounds for their pollinators, and hence also
fruits, throughout the year.

Gynodioecious Figs
The previous description of the life cycle of figs refers to monoecious species.
The eatable fig F. carica and many Asian fig species are gynodioecious (Berg
1989). In these species, the cycle for the preservation of the pollinating wasps
is confined to male and female flowering plants. All female flowers in these
plants are short-styled gall flowers; hence, the pollinators are able to lay eggs in
all female flowers, and no mature seeds develop in this type of syconia. In con-
sequence, gynodioecious figs are functionally dioecious.



flowering phase wasp flight

development crop abortion

flowering phase wasp extinction

Fig. 7.4. A hypothetical example on the phenology of figs and its impact on pollinat-
ing wasps. (Adapted from Bronstein et al. 1990)
7.3 Species Interactions and Coevolution 95

If a wasp enters the syconium of a solely female flowering plant, the flowers
are fertilized by the pollen carried by the insect. All female flowers in these
syconia are long-styled; thus, the wasp cannot deposit its eggs and seeds
develop from all fertilized flowers. Since the wasp is irreversibly trapped in the
syconium, it will not be reproductively successful if it enters a solely female
flowering inflorescence without gall flowers (Fig. 7.5).

Genetic Consequences of the Symbiosis

Self-pollination of monoecious figs is usually excluded owing to the strong
protogyny and the synchronized flowering of single plants. Asynchronous
development of a few syconia is observed in some species (Cook and Power
1996). It helps to preserve the pollinators, but is likely to result in selfed
offspring of figs. Gynodioecious figs are functionally dioecious and hence
obligatory outcrossing.

Fig. 7.5. Reproduction of gynodioecious figs. (Adapted from Hill 1967b). Further
explanation in the text
96 C HAPTER 7 Adaptation and Coevolution

Mating preferences among figs are likely to be strongly influenced by their

complicated flowering phenology, and to a lesser degree by spatial proximity as
long as the distance can be bridged by pollinators. Investigations of seven
monoecious strangler fig species in Central America revealed that their seeds
were fertilized by a surprisingly high number of pollen parents (Nason et al.
1996, 1998). Owing to the low density of figs, the average movement of pollen
was estimated to be between 5.8 and 14.2 km. Breeding units are as large as
106632 km2, which is a considerably larger area than for any other plant
species investigated (Herre 1996). The surprisingly efficient long-distance
transport of fig pollen is possibly facilitated by passive wind-mediated disper-
sal of fig wasps flying high above the canopy (Compton et al. 2000). The wasps
are likely to be attracted by volatile chemicals if they come close to a receptive
fig tree.
These results indicate that fig populations can be maintained in fragmented
landscapes as long as the pollinators are capable of bridging the fragments;
thus, even widely isolated remnants of forests may be important for the main-
tenance of figs and their pollinators. The potential for local adaptation of figs
to specific sites is limited owing to the effective gene flow and migration by
pollen and seeds.
The genetic structures of consecutive-age cohorts of single seed trees are
likely to be strongly differentiated from each other owing to changing pollen
parents. The flowers of a single syconium are frequently fertilized by a single
wasp only (Janzen 1979b). The progenies produced in such a syconium are
full-sibs (Nason et al. 1998).
The sex of fig wasps is, like the sex of most Hymenoptera, determined by
a system of haplodiploidy. Male insects develop from unfertilized eggs and
are consequently haploid (Godfray 1988). Female insects are able to control
the sex ratio by facultative admittance of sperms to fertilize their egg cells
(Frank 1985).
Female fig wasps are fertilized before they leave the syconium; thus, the
number of females entering and laying eggs in a syconium is crucial for the
degree of genetic relatedness of the offspring produced. Matings between sib-
lings are frequent since syconia are often entered by only a single wasp and
rarely by more than four wasps. Inbreeding is frequent in wasp populations
(Molbo et al. 2004).
The very important conservation of tropical figs is only possible if their pol-
linators are also preserved. Minimal population sizes for the preservation of fig
populations should be based on models which allow the population dynamics
of the pollinators as the most important criterion to be assessed (Bronstein
et al. 1990). Actual population sizes critically depend on phenology and
the migration behavior of female fig wasps. Experimental results suggest
7.4 Recommended Literature 97

conserving even widely isolated fig trees which possibly belong to viable
populations owing to the high efficiency of fig wasps as pollinators.

Recommended Literature

The concept of physiological adaptation or phenotypic plasticity of plants has

been reviewed by Schlichting (1986, 2002). Evolutionary adaptations and the
role of genetic variation for adaptability have been analyzed by Dobzhansky
(1968) and Dobzhansky et al. (1977).
Plantpollinator interactions in tropical forests were reviewed by Bawa
(1990). The reproduction of figs was concisely described by Janzen (1979a).
Phylogenies and Evolution Above the Species Level 8

The wealth of species in humid tropical and subtropical forests has always fascinated
botanists and not only taxonomists but also ecologists, geneticists, and students of
evolution. Much research remains to be done in these forests,. . . (p. 210 in Stern and
Roche 1974).


One of the most obvious and fundamental differences between tropical and
temperate forests refers to their floral diversity and in particular the diversity
of tree species (Whitmore 1975). The number of tree species in a single
hectare in a tropical lowland forest is often considerably higher than the num-
ber of trees native to central Europe. Public and scientific discussions are often
focused on the conservation of this diversity rather than its evolution (Lugo
1988; Raven 1988; Sayer and Whitmore 1990). Conservation of biological
diversity continues to deserve the highest priority in view of the undiminished
speed of forest destruction in many parts of the tropics (FAO 2002;
http://www.fao.org/forestry/site/fra/en). However, improved knowledge
of the evolution of the high species diversity in tropical forests and its
maintenance in undisturbed forest ecosystems will help us to develop sound
strategies for its conservation.

The Evolution of Species Diversity in the Tropics

Species Diversity in Tropical Forests

The enormous floral diversity was noticed during early expeditions to all tropi-
cal regions and quickly became a field of intensive research (e.g., von Humboldt
1808). The rich biological diversity in the tropics stipulated evolutionary
100 C HAPTER 8 Phylogenies and Evolution Above the Species Level

thinking of researchers during the nineteenth century (Lefvre 1984). Alfred

Wallace and Charles Darwin independently developed the theory of evolution
by natural selection after personal observations of the diversity of life in the
Floristic diversity is not evenly distributed among continents and regions.
The overall number of flowering plants is considerably higher in tropical
America (approximately 90,000 species) than in Asia (approximately 40,000
species) and Africa (approximately 35,000 species; estimates from Thomas
1999). Regional centers of floral diversity are presumably related to the cli-
matic variability throughout the Quaternary (Cook et al. 1990) and possibly
represent refugial areas during eras with a less favorable climate (see later).
The identification of regional centers of species diversity is of obvious impor-
tance for the development of conservation strategies (Lugo 1988; Wilson
1988; Gentry 1992). Variation of species diversity has also been described at
the local level by detailed floristic inventories. Extraordinary high numbers of
tree species per hectare were counted in local inventories in the Neotropics
(Gentry 1986) and in tropical lowland forests in Southeast Asia (p. 5 ff. in
Whitmore 197).
Simulation models predict a positive correlation between levels of genetic
diversity within species (Sect. 3.4) and species diversity (Vellend 2005). However,
no empirical studies have been conducted in tropical forests, and it is unknown
at which spatial scale correlations possibly exist. Even the type of the expected
correlation is a matter of speculation: Refugial areas might be hotspots both of
species diversity and of diversity within species. On the other hand, a high diver-
sity of tree species implies low population densities for most species, and genetic
variation may be lost owing to drift, if low population density results in low
population sizes.
Not all natural tropical forests are characterized by a high species diversity.
The number of species declines with decreasing temperature in mountainous
areas and extended dry seasons in marginal regions of the tropics (Richards
1981). Azonal edaphic conditions also frequently result in decreasing species
diversity (Whitmore 1975). Forests with a low diversity of tree species occur
even in the humid central tropics (Hart et al. 1989; Hart 1990). Biotic interac-
tions possibly play an important role for their existence and stability (Connel
and Lowman 1989).
The floral diversity represents only a small part of the overall biodiversity in
tropical forests. Estimates concerning the number of existing species on earth
vary greatly since only a small proportion of the most species-rich group, the
insects, have been described (Adis 1990). The diversity of insects in the canopy
of tropical forests constitutes the biggest share of global species diversity
(Erwin 1988).
8.2 The Evolution of Species Diversity in the Tropics 101

Evolution Above the Species Level

The floral diversity of tropical forests is fairly well described; however, its evo-
lution and the mechanisms ensuring its maintenance are only poorly under-
stood. The most important theories trying to explain the species diversity of
tropical forests were reviewed by Baker (1970) and, more recently, by Hill and
Hill (2001) and Montagnini and Jordan (2005, p. 27 ff.). Most of these theories
are chiefly concerned with the ecological requirements for the coexistence of
many species in a given area (ecological niches, environmental heterogeneity,
disturbance) and species interactions, including competition and predation.
Two hypotheses with a focus on the factors ruling the emergence of the high
species diversity in tropical forests are concerned with the duration of periods
of an undisturbed ecosystem development.
A long period of cumulative evolution undisturbed by annual periods of
unfavorable environmental conditions such as low temperatures and long dry
seasons is regarded as an important prerequisite for the diversification of species.
A more or less constant climate for extended periods of millions of years is likely
to be even more important (museum model). The evolution of angiosperms in
tropical forests lasted at least since the beginning of the Tertiary for approxi-
mately 65 million years.It is likely that in some at least of the regions where Rain
forest is now the climatic climax, similar vegetation has existed uninterruptedly
since a very remote geological period. (p. 14 in Richards 1981). Thus, even a
slight excess of newly evolving species in comparison with presumably low
species extinction rates resulted in the accumulation of many species during this
long era. According to Richards (1981, p. 16), the floristic richness of tropical
forests is largely due to its great antiquity.
Multiple evidence suggests a strong impact of the climatic variability during
the Pleistocene not only on the flora in temperate regions, but also on tropical
rain forest vegetation (Haffer and Prance 2001). Allopatric speciation in iso-
lated refugia during dry periods possibly increased species diversity since popu-
lations isolated from each other in separate refugia eventually developed
reproductive barriers (Hill and Hill 2001); however, the size and the location of
pleistocene refugia are under dispute. The fossil pollen record suggests stability
and continuity of the forest cover over the Amazonian lowlands (Colinvaux and
De Oliveira 2001). Sequence variation patterns at the mitochondrial
cytochrome b gene in different mammals from North America and Amazonia
suggest rapid population growth for the North American species after the end
of the last glaciation, but no or only minor demographic expansion for small
mammals in Amazonia (Lessa et al. 2003). This result also suggests only a lim-
ited impact of climatic variability on the population sizes of small mammals in
102 C HAPTER 8 Phylogenies and Evolution Above the Species Level

the Neotropics and hence contradicts the refugia hypothesis. However, a molec-
ular phylogeny of the species-rich genus Inga suggests a recent speciation,
possibly as a consequence of the climatic variability during the pleistocene
(Richardson et al. 2001; Example 8.2).
Ecological research on the dynamics of tropical forests, including their regen-
eration, greatly profited from the establishment and repeated analysis of inten-
sively studied plots on Barro Colorado Island (Panama), Pasoh Forest Reserve
(Malaysia; Lee 1995), and in other tropical regions (Hubbell and Foster 1992).
Numerous genetics studies were conducted in these plots; several of them are
described in this book (Table 3.3, Fig. 6.2; Examples 5.2, 5.3). However, the focus
of these investigations was on the maintenance of species diversity, the ecological
and genetic consequences of a high species diversity, and the role of species inter-
actions in this context. The limited time for observations does not permit the evo-
lution of species diversity for long-living organisms like forest trees to be studied.
Taxonomic studies are the most important basis for the description of the
biodiversity in tropical forests. The relationships among existing species as
revealed by the differentiation of morphological traits of taxonomic significance
offer opportunities not only to classify taxa, but also to understand their origin.
During the past few decades, traditional taxonomic surveys were complemented
by molecular approaches to reveal phylogenies. Early caryological studies used
chromosome polymorphisms and in particular polymorphisms of chromo-
some numbers to infer relationships and the evolution of plants in the tropics
(Mehra and Bawa 1969). More recently, molecular genetics markers based on
variation of DNA became commonly applied tools not only to reveal genetic
variation within species (Sect. 3.4), but also to clarify phylogenetic relations
among more or less closely related taxa occurring in tropical forests (Sect. 8.3).

Molecular Phylogenies

The evolutionary history or phylogeny of a group of organisms can be analyzed

by investigating similarities at particular characters, which possibly reflect
relatedness or a common origin. Thus, if two taxa A and B are more similar to
each other than both are to taxon C, a more recent common origin and, hence,
a higher degree of relatedness of taxa A and B may be assumed in comparison
with taxon C. Similarities need to be measured at particular traits or characters,
and the observation of different traits often results in contradicting conclu-
sions. Various reasons can account for similarity apart from a common origin.
Convergent evolution in very distantly related organisms eventually results in
similarities at morphological traits in response to particular environmental
conditions. Frequently cited examples for plants are a herbaceous growth habit,
8.3 Molecular Phylogenies 103

which repeatedly evolved from woody ancestors, dioecy (Sect. 4.2.3), or leafless
plants in adaptation to a dry environment in Neotropical Cactaceae and
Euphorbiaceae of the Old World. Thus, similar trait expression may not only
be due to a common origin (homology), but characters may also be similar
although they are not derived from a common ancestor (homoplasy).
The observation of variation at molecular traits offers obvious advantages in
comparison with morphological characters. This applies in particular to variation
of DNA, which is not influenced by environmental conditions. Thus, some of the
methods briefly described in Sect. 2.3.1 are not only commonly used to investigate
genetic variation within species, but are also used to clarify differentiation among
species and to elucidate evolutionary relationships by phylogenetic analyses.
The observation of variation at nonrecombining chloroplast DNA (cpDNA)
is of particular importance for plants. Low mutation rates of cpDNA are respon-
sible for, in general, low variation within species, but ample and easily inter-
pretable differentiation among species at various levels. Sequences of cpDNA
were used to investigate phylogenetic relationships among plants at virtually all
taxonomic levels. Variation of cpDNA is informative to study splits during the
earliest evolution of land plants (Goremykin and Hellwig 2005), and also to
reveal phylogenetic relations within single genera (Gernandt et al. 2005).
Phylogenies are not only based on cpDNA, but also on nuclear DNA (nDNA)
and anonymous markers. Commonly used nDNA sequences for phylogenetic
analyses are ribosomal DNA internal transcribed spacer (ITS) regions. For exam-
ple, molecular phylogenies of the pioneer tree genus Macaranga (Euphorbiaceae)
were based on ITS sequence data (Blattner et al. 2001) and amplified fragment
length polymorphisms (Bnfer et al. 2004). The evolution and phylogeny of
flowering plants has recently been reviewed largely based on molecular data (The
Angiosperm Phylogeny Group 2003).
Two approaches are commonly used to construct phylogenetic trees based
on DNA variation (Hall 2001). The most important algorithmic methods are
the unweighted pair-group method with arithmetic means (UPGMA) and the
neighbor joining (NJ) method. Both are distance methods based on the com-
putation of a matrix of all pairwise differences between samples (Sect. 3.3.2).
Parsimony is the most commonly used tree-searching method. Many trees are
constructed and compared. The most likely (maximum parsimonious) tree is
the one requiring the lowest number of evolutionary steps, including
homoplasies, to explain the data. The reliability of phylogenetic trees is often
statistically tested by bootstrapping (p. 58 ff. in Hall 2001).

Example 8.1: A Molecular Phylogeny of Indonesian Dipterocarpoideae

The family Dipterocarpaceae consists of three subfamilies; the Asian
Dipterocarpoideae is by far the most important and most diverse subfamily,
104 C HAPTER 8 Phylogenies and Evolution Above the Species Level

consisting of about 470 species in 13 genera. Nine genera with more than 250
species are native to Indonesia. The Dipterocarpoideae are commonly divided
into two tribes (Ashton 1982). Tribe Dipterocarpae, with a basic chromosome
number of x = 11, consists of the genera Anisoptera, Cotylelobium, Dipterocarpus,
Upuna, and Vatica. Tribe Shoreae (x = 7) consists of Dryobalanops, Hopea,
Parashorea, and the species-rich genus Shorea. Genus Dryobalanops has a basic
chromosome number of x = 7, but shows intermediate characters between the
tribes concerning flower morphology (sepal aestivation) and wood anatomical
features (vessels, resin canals).
Variation of cpDNA was studied by means of polymerase chain reaction
restriction fragment length polymorphisms and chloroplast simple sequence
repeats (Sect. 2.3.1). A total of 129 samples from 58 species were investigated.
The African dipterocarp Monotes kerstingii (subfamily Monotoideae) was used
as an outgroup to root the phylogenetic tree (Indrioko et al. 2006).
No variation was observed within species. The differentiation among species
at cpDNA is in good agreement with the classification based on morphological
characters (Fig. 8.1). One major clade consists of Upuna, Cotylelobium,
Anisoptera, Vatica, Dipterocarpus (tribe Dipterocarpae), and Dryobalanops (tribe
Shoreae) in a basal position. A second clade consists of Hopea, Parashorea, and
Shorea (tribe Shoreae). The genera are, in general, well resolved and strongly
supported. The intermediate position of genus Dryobalanops is confirmed by
the analyses. Parashorea groups with Shorea in a single big, poorly resolved
clade. Both genera have close botanical affinities and the phylogenetic
relevance of their distinction is questionable.
Similar results were obtained by previous studies based on cpDNA variation
of dipterocarps (Tsumura et al. 1996; Kajita et al. 1998; Kamiya et al. 1998;
Gamage et al. 2003). Phylogenetic analyses based on cpDNA failed to consis-
tently group the species-rich genus Shorea into well-supported clades. However,
the variation at the nuclear PgiC gene locus is consistent with the distinction of
different timber groups (balau, red meranti, white meranti, yellow meranti) and
sections according to Ashton (1982) within genus Shorea (Kamiya et al. 2005).
(From Indrioko et al. 2006)

Example 8.2: Evolution of the Genus Inga

Molecular phylogenies can be used to test the alternative hypotheses of specia-
tion according to the museum model or recent speciation in response to
environmental variation during the past few million years (Sect. 8.2.2). The
slow evolution of species diversity as a result of a long period of favorable
climatic conditions (museum model) is expected to result in a well-resolved
phylogeny, each species distinguished from others by numerous mutations
(nucleotide substitutions at the level of DNA). Recent allopatric speciation in
8.3 Molecular Phylogenies 105

Cotylelobium lanceolatum
Upuna borneensis
Anisoptera costata
Anisoptera marginata
68 99 Anisoptera reticulata
Vatica bantamensis
Vatica bella
99 Vatica granulata x=11;
83 Vatica pauciflora valvate
94 Vatica rassak Diptero-
Vatica venulosa carpeae
Dipterocarpus grandiflorus
Dipterocarpus oblongifolius
71 Dipterocarpus retusus
100 Dipterocarpus rigidus
Dipterocarpus tempehes
Dryobalanops aromatica sub
99 Dryobalanops lanceolata valvate
Hopea bancana
Hopea celebica
85 Hopea odorata
Hopea sangal
Hopea dryobalanoides
Hopea nigra
Hopea griffithii
66 Hopea mengarawan
100 Parashorea lucida
Parashorea globosa
Shorea blumutensis
Shorea guiso
Shorea montigena
Shorea scaberrima
Shorea seminis
Shorea acuminata
Shorea andulensis
Shorea javanica x=7;
67 Shorea johorensis imbricate
Shorea leprosula Shoreae
95 Shorea macroptera
69 Shorea mecistopteryx
63 Shorea ovalis
Shorea palembanica
Shorea parvifolia
Shorea platyclados
Shorea xanthophylla
Shorea balangeran
59 Shorea selanica
Shorea splendida
Shorea macrophylla
55 Shorea pinanga
Shorea stenoptera
Shorea acuminatissima
Shorea dasyphylla
77 Shorea faguetiana
Shorea multiflora
Shorea fallax
98 Shorea materialis
60 Shorea virescens
Monotes kerstingii valvate x=7

Fig. 8.1. Strict consensus tree of Dipterocarpoideae (outgroup Monotes kerstingii)

Numbers at nodes indicate bootstrap values. Typical characters (valvate, subvalvate or
imbricate sepal aestivation; basic chromosome number x ) and tribes are indicated.
(From Indrioko et al. 2006)

refugia should be reflected in few nucleotide substitutions differentiating

species, and a poorly resolved phylogeny (Fig. 8.2).
The genus Inga (Fabaceae; Mimosoideae) contains about 300 species grow-
ing in neotropical rain forests. Sequence variation was observed for 32 species
at the ITS of nuclear ribosomal DNA and for 31 species at a cpDNA fragment.
A total of 45 species representing most sections of the genus were analyzed
(Richardson et al. 2001). Only a few substitutions were observed and the
phylogenies were unresolved, conforming Fig. 8.2b. The result is consistent
106 C HAPTER 8 Phylogenies and Evolution Above the Species Level

a Fig. 8.2. a A well-resolved phylogeny

due to the accumulation of mutations
b over long periods and b a poorly
c resolved phylogeny due to a low number
of mutations differentiating species
(ah) indicating recent speciation.
(Adapted from Richardson et al. 2001)

a h




b h

with a recent speciation within the genus Inga. Rough estimates of the diver-
gence time suggest that speciation occurred within the last ten million years,
and approximately 30% of the species diverged from their closest relative dur-
ing the Pleistocene (within the last two million years).
Shorea (Dipterocarpaceae) contains only slightly fewer species than Inga.
The molecular phylogeny within the genus Shorea is also poorly resolved
(Example 8.1), likewise suggesting comparatively recent speciation within this
genus, at least in Southeast Asia. However, the diversity of tropical forests does
not only result from a few species-rich genera like Inga, Shorea, Ficus, and
others, but is also reflected in a large number of genera, families, and higher
taxonomic units present in most tropical forests (Bermingham and Dick 2001).
Thus, the species diversity of tropical forests is presumably a consequence of
both their great antiquity (museum model) and recent speciation for certain
taxa like Inga. (From Richardson et al. 2001)

Recommended Literature

The causes of the species richness of many tropical forests were reviewed
by Baker (1970), Hill and Hill (2001), and Montagnini and Jordan (2005,
p. 27 ff.).
8.4 Recommended Literature 107

Hart (1990) pointed out that not all tropical forests are characterized by a
high species diversity.
Phylogenetic methods are described among other topics related to phylogeog-
raphy in the book by Avise (2000). Their application to plants was reviewed by
Schaal et al. (1998). An update of the classification of the orders and families of
angiosperms largely based on molecular phylogenetic methods has recently been
given by The Angiosperm Phylogeny Group (2003). The rationale behind
molecular phylogenies and the commonly used software PAUP for phylogenetic
analyses are described in the book by Hall (2001).
Applications of Genetics to Tropical Forestry Part B

The majority of tropical tree species possess high genetic diversity (Sect. 3.4).
The elements of their genetic systems that are responsible for the evolution of
this genetic variation and its temporal and spatial dynamics under natural con-
ditions were the topics of Part A.
Part B is oriented towards application. It concentrates on both genetic impli-
cations of human activities and on planned utilization of genetic variation.
Tropical forests are being cleared, burned, fragmented, logged, and overhunted
at rates that lack historical precedent (Laurance and Bierregaard 1997a). The
consequences of the large-scale conversion of forests to other types of land use
and the fragmentation of contiguous forests require analysis and the develop-
ment of guidelines for appropriate treatment of remnant forests (Chap. 9).
Forestry involves logging and the silvicultural management of diverse types of
forests. Silvicultural management has far-reaching consequences for the
genetic system of tree species and their genetic structures (Chap. 10).
Tree breeding including provenance selection (Chap. 11) aims at changing
genetic structures of populations in order to increase the plantation yield. Tree
improvement and planting activities in tropical forestry commenced a long
time before population genetic research on the genetic system of tropical for-
est trees was taken up (Chaps. 12, 13).
Last but not least, in view of the genetic impact of human interference, pro-
jects on the conservation of tree genetic resources are an imperative application
of genetics. Genetic information is not a renewable resource. These projects
require concomitant research in order to be successful (Chap. 14).
In most tropical countries forest genetics plays only a minor role in forestry
education. In most curricula emphasis is put on tree improvement, if subjects
related to genetics are discussed at all. In courses concentrating on breeding,
students become acquainted with working methods of quantitative genetic
research and relevant statistical methodology. Basics of population genetics, an
analysis of the genetic systems of tropical forest trees, the role of genetics in sil-
viculture, and the conservation of forest genetic resources have conventionally
been regarded as topics of minor importance as compared with tree improve-
ment. Not too much at variance from the conditions in the temperate zones,
110 PART B Applications of Genetics to Tropical Forestry

forest genetics research is occasionally used as a synonym for the formulation

of breeding goals and the conduct of breeding projects. Owing to this miscon-
ception, the fundamental importance of an analysis of the genetic system of
tropical forest trees by use of population genetics methodology is overlooked.
Nonetheless, tree improvement is but an application of forest genetics and
must be accompanied by the conservation of forest genetic resources.
Fragmentation of Forests 9


Following the advance of civilization, large forests are completely eliminated

and converted to other types of land use on all continents. However, small
pieces of forest are left on areas that are of little use owing to their adverse topo-
graphy or drainage, or other reasons. They are embedded in a largely agricultural
matrix (James et al. 1998). They are also looked at as islands surrounded by a sea
of modified habitats (Laurance and Bierregaard 1997a). Darby (1956)
described the history of the reduction of forests in Europe and documented
their increasing fragmentation on a continent-wide scale. When discussing the
human impact on forests, Ledig (1992) and Ledig and Kitzmiller (1992) listed
fragmentation of forests as next in line after complete destruction (see also
Clarke and Young 2000). Habitat fragmentation of trees and other living
organisms as an almost inevitable consequence of large-scale forest loss has
emerged as a global concern. The size distribution of the fragments, their spa-
tial pattern in landscapes, and the proportion of the area covered by forest
affect both the total number of persisting species and the relation between
them. Along with the forests, the distribution ranges of tree species and their
populations are fragmented. Depending on the distribution pattern of species,
fragmentation of forests may disrupt the range of one species but not so of
others. This condition is encountered not only in the tropics but also in other
climatic zones; however, in species-rich forests of the tropics it is of particular
concern to ecologists and forest geneticists. Owing to the species richness par-
ticularly of tropical rain forests with a high proportion of endemics (Viana
et al. 1997), the number of species being strongly affected or becoming extinct
is larger than in other biomes. Owing to local rarity or pronounced clumping
of species occurrence, many species may be missing from any particular frag-
ment simply because they never occurred there (Laurance et al. 2002). The
same applies to genotypes within species. Also, many species have disappeared
even from larger fragments because they are sensitive to disturbance.
112 C HAPTER 9 Fragmentation of Forests

Another condition often goes along with fragmentation. Owing to raised

pressure of the local human population on those remnants of forest, the trees
of the same species occur with reduced overall density. They also have a patchy
distribution, because the original density of trees varied with the species and
with microsite variation in a heterogeneous environment. Locally, the density
may be dramatically reduced. The landscape may not be completely denuded
of forest but remaining trees may grow in rows along roads and watercourses
or as solitary trees on pastures or abandoned fields. The question arises, to
what extent these trees form possibly persistent populations in the genetic
sense (Sect. 2.2). The implications of low population density and restricted
population size for the reproduction system of trees were discussed in Chaps.
5 and 6. In the present chapter, the very small remnant populations of reduced
density are addressed also from the management point of view.
The conversion of large areas of contiguous tropical forests to other land uses
leaves forest fragments of various sizes. For instance, Sanchez-Azofeifa et al.
(1998) reported on deforestation in lowlands of Costa Rica, where more than
80% of the primary forests had been eliminated since 1984. During a 10-year
period the annual loss rate was more than 2% and the number of forest frag-
ments doubled, while their average size decreased from 95 to 25 ha. The total
area of fragments of less than 100 km2 size or less than 1 km away from an edge
in lowland rain forest in Brazilian Amazonia increased from 20.8 106 ha in
1978 to 50.8 106 ha less than 20 years thereafter (Whitmore 1998). These
exemplary data indicate that fragmentation in a largely deforested country is
still in progress on many spatial scales and will presumably lead to fewer and
smaller fragments and to the extirpation of many plant and animal species. The
genetic impact of this process will become ever more severe. While considering
the genetic effects of isolation, it must not be overlooked that the amount of
available habitat is greatly reduced. The smaller the fragments, the more
important become edge effects that penetrate deep into the forest remnants
(Laurance et al. 2002). In a case study of the decline of deciduous and semide-
ciduous tropical forest in Bolivia, the increase in the number and in the area of
both large and small fragments is documented over a 20-year period. The
microclimatic edge effects such as increases in light, temperature, and wind
speed affect larger part of the remnant forest (Steininger et al. 2001). The mor-
tality among trees increases (Laurance et al. 2002). In an early paper on forest
fragmentation, Curtis (1956) documented the clearing of forest in a certain
part of Wisconsin, USA, during a 120-year period. This author pointed out the
decrease of forest area, the initially increasing and subsequently decreasing
number of fragments, their average size, their distance from one another, and
their total periphery. He also described the various ecological correlates of this
man-induced process. Loyn and MacAlpine (2001) analyzed different types of
forest fragmentation and found their ecological impacts somewhat different.
9.2 The Genetic Status of Fragmented Tree Populations 113

The Genetic Status of Fragmented Tree Populations

With some time-lag, the extinction of at least some tree populations or species
in fragmented tropical forest is likely in view of their reproduction system.
Most tropical trees are characterized by predominant outcrossing (Sect. 6.3),
strong inbreeding depression (Sect. 6.4.2), and reproductive strategies relying
on the presence of other tree and animal species (Ashton 1981). As already
indicated, if a forest has been fragmented by human action, its tree populations
are suddenly put into an environment to which they are not necessarily fully
adapted; therefore, it is the formerly widespread and common plant species
that suffer most from (complete) fragmentation of their populations (den Njis
and Oostermeijer 1997). The remnant forests possess less biodiversity. They
represent disturbed ecosystems comprising merely part of the tree species
belonging to the respective forest community. The moderate numbers of
individuals of the former keystone species form populations of reduced
reproduction-effective size. Even if tree species were intimately mixed, a certain
minimum area is required for all species to occur. Consequently, many species
are locally lost if the respective fragment is small. The same is true with
genotypes belonging to one species. The relict populations are in danger of
lacking formerly existing genetic variants due to genetic drift and those losses
are no longer compensated by the influx of external effective pollen. Isolation
is likely to result in increased genetic differentiation among relict populations
owing to nonadaptive changes of genetic structures. It must be emphasized
that reduced genetic variation in small populations is presumably a symptom
of endangerment rather than its cause (Namkoong et al. 1996; Holsinger 2000).
With progressing disturbance of forest ecosystems coadapted other species
of plants and of animals are eventually no longer present. Pollen vectors may
not survive or may no longer visit remote forest fragments (but see Example
8.2). However, some dispersers of pollen and seed respond positively to forest
clearings and to the edge effects implied by fragmentation (Laurance et al.
2002). Pollen vectors support a critical portion of the life cycle of tropical trees
and must have a permanent food basis. It is interrupted, if the flower rewards
are not offered all-year round. If also the seed vectors have become rare or
absent, the occupation of patches of the former habitat becomes increasingly
difficult for tree species. First of all, rare widely dispersed tree species will suf-
fer from altered pollinatorplant assemblages (James et al. 1998); therefore,
one has to anticipate complex, cryptic responses to fragmentation (Aldrich and
Hamrick 1998). The trees in those relict forests are less adapted to the new
environment, since they have evolved in obligate associations with groups of
pollen and seed dispersers that either are now missing or have greatly changed.
114 C HAPTER 9 Fragmentation of Forests

The decline of dispersers is followed by the decline of the respective tree

species. Changes in the guild of pollinators are not just responsible for pollen
limitation and lower seed set. As an environmental factor influencing the mat-
ing system (Sect. 6.6), they also affect genetic structures of the next generation
of trees. Adaptive changes of genetic structures are induced but they may
require generations until the former level of adaptedness is reached; hence, pop-
ulation fragments are genetically vulnerable. In one example from New Zealand,
a shift in the guild of pollinators was not accompanied by consistent changes in
the mating systems of the partially self-fertilized tree Metrosideros excelsa
(Schmidt-Adam et al. 2000). Other birds with similar behavior have presumably
taken over the role of two other species of pollinators. As Whelan et al. (2000)
found in several species of the protaceous genus Grevillea in fragmented
Australian populations, there may be strong interactions between pollinator
guilds such as vertebrates and insects depending on the natural or preferred
mating system of trees. In the completely self-fertile species G. macleayana the
proportion of self-fertilization ranged between 20 and more than 90%. In this
type of species, changes due to fragmentation are expected to be less intensive.
However, the majority of tropical tree species are all but fully self-fertile.
Introduced honey bees eventually change the realized mating systems and
the concomitant mating success under the now given conditions. These insects
have largely displaced honeyeaters as pollinators from some of their food
plants, thus contributing to their decline and leading to pollen-limitation in
Grevillea spp. and other species (Paton 2000). The disruption of mutualisms
between trees and their native pollinators is sometimes compensated by the
genetic rescue of remnant trees through introduced African honey bees. They
outnumbered native pollinators of the canopy-emergent tree Dinizia excelsa
(Fabaceae) in disturbed Brazilian habitats and were the only pollinator of a
lone pasture tree of this species, transferring pollen from large distances (Dick
2001). Almost 80% of the canopy-forming trees in Samoa have been found to
depend on flying foxes as dispersers of pollen or seed. It is still unknown
whether the role of these declining animal species with their vast foraging area
can be filled by introduced organisms (Cox and Elmqvist 2000). Studying
insect-pollinated perennials in a South African shrub community, Donaldson
et al. (2002) reported complex reactions to fragmentation. Ecological charac-
teristics of fragments rather than their size and their distance to a contiguous
shrub community had a stronger overall impact on pollinator diversity and
plant reproductive success; however, the conditions varied greatly between
plant species possessing different reproduction systems.
The remaining pollen vectors are able to transfer only reduced amounts of
pollen among the trees. Eventually, the distances between trees approach the
range of the feeding flights of the pollinators, so the mating contact among
the trees is impeded. It may even be interrupted, if tree distances exceed the
9.2 The Genetic Status of Fragmented Tree Populations 115

activity radius of pollinators. Fragments have then lost their connectivity.

Historically old population isolation in Caesalpinia echinata Lam. in coastal
Brazilian forest has led to reduced variation in local populations and their strong
differentiation at maternally inherited markers (Lira et al. 2003; Sect. 5.4).
Depending on the composition and the behavior of the pollinator guild, the
amount of remote pollen transferred inevitably, though not continuously,
decreases with increasing distance. Consequently, also the remnant groups of
trees within the forest fragments are at least partially isolated. They can no
longer be considered coherent populations of the type that once prevailed on
the respective sites. Reduced mating contact also within population fragments
implies an increase in self-pollination. Unless there is no absolute barrier
against self-fertilization, some of the arising offspring show inbreeding depres-
sion, a decline of fitness in terms of retarded or reduced fertility, reduced
growth, and reduced longevity. If also the seed vectors have become less abun-
dant, the trees in a neighborhood that are potential mating partners tend to be
related. This biparental inbreeding is to be expected wherever the dispersal of
pollen and seed is not fully effective. The genetic effects described are more or
less severe depending on the reproduction system of the respective species. In
low-density populations or low-density patches the seed shadows of individu-
als possibly do not overlap, so neighboring seedlings are more closely related,
because they have at least their seed parent in common. This family structure
is less likely to arise if seed dispersal is very efficient. An initial spatial struc-
ture eventually fades with progressing life stages owing to mortality and
In an ecological sense, a tree is solitary if it has no neighboring trees as com-
petitors; however, it is not necessarily isolated in the genetic sense. Contrary to
human perception, spatial distance to the nearest neighbors does not necessar-
ily reflect their actual gene flow pattern. If a solitary tree produces seed by
cross-fertilization or contributes pollen to the progeny of other trees, it is part
of their population. A solitary tree that does not reproduce in either way might
still possess a vital ecological function, since its flowers support pollinator
movement between trees. Tall free-standing intervening trees may also serve as
roosts for fruit dispersers and thus direct seeds to nearby patches (Nason et al.
1997; Galindo-Gonzlez et al. 2000; Dick 2001).
From the second generation following fragmentation, another source of
inbreeding is likely to occur depending on the reproduction-effective sizes of
population fragments. These inbreeding effects accumulate if the reproductive
isolation of small fragments persists for several generations (Ledig 1992). Since
not all members of a population contribute balanced numbers of gametes to
the produced progeny, the effective number of individuals is reduced to less
than their census number. The size of the parental pool may even drastically
differ from the census number. With help of highly variable genetic markers at
116 C HAPTER 9 Fragmentation of Forests

least the effective size of the parental pool relative to the census number may
be estimated. The diminished size is due to imbalances in the production of
effective gametes (review by Hattemer in 2006). Stern and Gregorius (1972)
have shown that in monoecious populations a further reduction is due, since
the individuals may produce gametes of either sex, which represents a further
source of kinship. Depending on the degree of the imbalance of gamete contri-
butions and the sexual function of the individual trees, the effective number is
more or less strongly reduced. The reproduction-effective population size has
two aspects, as Kjr (1996) discussed with reference to tree populations. For
instance, a population fragment of 50 trees with an inbreeding-effective size of
half the census number is expected to produce the same number of inbred pro-
geny as a panmictically reproducing population of size 25. Furthermore, a collec-
tive of 50 trees with a variance-effective number of half the census number is
expected to produce progeny with the same amount of nonadaptive gene fre-
quency change as a panmictically reproducing population of size 25. Although
the assumption of panmictic reproduction raises the impression of being highly
unrealistic, the panmictic population still represents a useful reference (Sect.
6.2.2). Depending on the number of progeny produced, the variation in the effec-
tive number of female and male gametes, and the proportion of self-fertilization,
the two types of effective population size may differ but they are mostly less than
the census number. Kjr (1996) indicated also methods of estimation. The the-
ory of the two parameters introduced was explained and reviewed under a range
of circumstances by Caballero (1994). These parameters possess relevance also for
breeding and are addressed in Chap. 13. A situation where presumably both
parameters are drastically reduced is described in Example 9.1.

Example 9.1: Genetic Impact of Fragmentation on a Rain Forest Canopy Tree

Aldrich and Hamrick (1998) and Aldrich et al. (1998) studied the Neotropical
Symphonia globulifera (Clusiaceae), a late-successional, bird-pollinated canopy
tree of Costa Rican rain forests. On the study site, a population had been frag-
mented 1030 years before. The conditions in remnant forest patches and on
pasture were compared with those in continuous forest at a control site. Seed set
was low owing to low pollination, high abortion rates, and rapid removal by fru-
givores. Seedlings did not recruit on pasture but had higher density in remnant
than in contiguous forest. The proportion of seedlings arising from self-fertil-
ization increased from about 1% in continuous forest to 11% in remnant forest
and to even 25% in pasture trees. The few pasture trees had produced more of
the younger than the older progeny. This indicates a sequential reduction in the
number of reproductively effective trees. A variance-effective population size of
Symphonia globulifera of 17 was estimated both in continuous and in remnant
forest; however, accounting for migration led to estimates of 9.4 in continuous
9.2 The Genetic Status of Fragmented Tree Populations 117

forest and 2.8 in fragmented forest owing to the lower seed migration rate in the
latter. These estimates are extremely low. The reproductive imbalance becomes
even clearer if also the respective area sizes are accounted for: the effective
population size was much smaller on 38 ha of remnant forest than on only 1 ha
of continuous forest. Parentage analysis by means of microsatellite markers
revealed that in remnant forest by far most of the progeny were produced by a
few large trees left on nearby pastures. The large reproductive success of the
pasture trees was positively correlated with their increased crown size owing to
the release from competition following fragmentation. This was interpreted by
a shift in the foraging pattern of the humming-bird pollinators from traplining
towards more pronounced territoriality, i.e., from moving among tree crowns
towards moving within the same large tree crown for the sake of saving energy.
The size difference between saplings and seedlings reflected their difference in
age, which helped in classifying progeny individuals as seedlings or saplings. The
adult trees did not show any homozygote excess (F 0; see Sect. 6.4.1 for an
explanation of the inbreeding coefficient F ). Saplings in the remnant forest had
but a slight excess of homozygotes. The moderate inbreeding coefficient of the
saplings which was presumably due either to mating events prior to fragmenta-
tion or to viability selection since then was greatly exceeded by that of the
seedlings (F = 0.21). Bats dispersed fruits of the pasture trees over a considerable
distance into the remnant forest, thus adding notably to the reproductive imbal-
ance but little, if anything at all, to the variance-effective population size and the
genetic variation in the remnant forest. It must be added that the seedlings were
established in a suboptimal habitat, and that they belonged to a small number
of families, i.e., they were closely related or even inbred. This is consistent with
statements made by Ledig (1992) on the decrease of heterozygosity to be
expected in future generations assuming restricted population size. (From
Aldrich and Hamrick 1998; Aldrich et al. 1998)

Hall et al. (1996) have pointed out that the effective population size at a
given time may be diminished by variation in flowering phenology and may
thus become critical in small fragments. This expectation was supported by
results of Fuchs et al. (2003) for bat-pollinated Pachira quinata (Bombacaceae).
These authors found no reduction in reproductive output, because the lower
percentage of flowers developing into fruit was overcompensated by the larger
number of flowers produced by trees in fragments. However, they also esti-
mated lower outcrossing rates and a lower number of sires, leading to a higher
degree of relatedness of the seed produced in fragmented forest, particularly by
asynchronously flowering trees.
The degree of isolation cannot necessarily be judged on a geographic scale
alone (Nason and Hamrick 1997). Rather, dispersal delimits the patches. It
should have become clear by now that population fragments are reproductively
118 C HAPTER 9 Fragmentation of Forests

isolated only if they are separated by distances that greatly impede the transfer
of effective pollen or that cannot even be bridged by pollen or seed vectors at
all (Chap. 5). Their behavior and abundance ultimately maintains links among
forest fragments. To some degree, low tree density expands pollinator foraging
ranges and pollen dispersal distances. It has been shown that tree pollen can
travel far even in fragmented terrain (Chase et al. 1996b; Dawson et al. 1997;
Nason and Hamrick 1997; Aldrich and Hamrick 1998; Aldrich et al. 1998; James
et al. 1998; White and Boshier 2000; Dick 2001; Example 7.2). Nonetheless, one
has to assume absolute limits beyond which no effective pollen can be trans-
ferred. However, as was shown both in wind-pollinated (Starke 1996; Wang
2001) and in insect-pollinated tree species of the northern temperate zone,
fragmentation may also facilitate pollen dispersal. Ballal et al. (1994) reported
up to 17% influx of pollen into small population fragments of Acer saccharum
close to contiguous forest. This species has a dual system of pollen transfer,
because in the absence of pollinators wind-transferred pollen is sufficient for
normal seed set (Gabriel and Garrett 1984). Surveying seedlings in samples
from a natural park and in patches (comprising 90646 individuals), Young
et al. (1993) found no evidence of reduced genetic variation, reduced gene flow,
or increased selfing in the latter, although fragmentation had taken place more
than a century before. These authors explain superior diversity of population
fragments by increased gene flow. Also Young et al. (1996) came to the conclu-
sion that increased gene flow may counteract if not even compensate or
overcompensate the effect of reduced population size.
Nason et al. (1997) discussed the results of numerous studies of pollen dis-
persal in tropical trees and concluded that the absolute reach of transfer
distances in species adapted to low population density in species-rich commu-
nities is several hundred meters to some kilometers (Sect. 5.2). As Wright (1931)
derived from his island model, few migrants or effective pollen per generation
are sufficient to counteract loss of genetic diversity and drift-related genetic dif-
ferentiation. The probability of mating between trees at a certain distance in
adjoining fragments is higher than expected on the basis of that distance in a
contiguous population. The reason is the absence of pollen-producing trees
growing in-between. Bierregaard et al. (1992) reported that in Amazonian
forest breaks as little as 80 m wide may diminish interfragment movement
of some insects and birds. There must clearly be a certain limit to the gap-
crossing ability of any pollinators; however, this gap may be rather wide depend-
ing on several circumstances, as shown for figs in Example 7.2 and Swietenia
humilis in Example 9.2. This study also shows that fragmentation-induced
changes in the foraging pattern of dispersal agents and the concomitant genetic
processes in a tree population can differ among fragmented habitats. Apsit et al.
(2001) described another situation of substantial pollen immigration into a
population fragment of the insect-pollinated Enterolobium cyclocarpum (Jacq.)
Griseb. (Fabaceae). A small number of trees received pollen from many outside
9.2 The Genetic Status of Fragmented Tree Populations 119

pollen donors growing hundreds of meters away. Similar findings were reported
by Cascante et al. (2002) studying isolated trees of Samanea saman
(Mimosaceae). Isolated trees received pollen from far-distant trees; however, the
styles of these trees received lower quantities of pollen and a greater proportion
of the pollen tubes were arrested at the bottom of the style, presumably owing
to genetic incompatibility. In spite of high reproductive capacity of the isolated
trees, they received pollen from fewer pollen donors, which led to an increase in
the degree of the relatedness among their progeny.

Example 9.2: Connectivity of Population Fragments of a Medium-Sized

Dry-Forest Tree
In a Honduran study region in seasonally dry forest the population of the
largely self-incompatible species Swietenia humilis (Meliaceae) is highly frag-
mented and comprises small forest remnants and scattered trees in pastures
and along the borders of agricultural fields (Fig. 9.1). A large proportion of the
trees was present prior to fragmentation; therefore, the present pattern largely
reflects the previous more continuous forest. Pollination is by small generalist
insects; the seeds are wind-dispersed. The species flowers for about 1 month
during the dry season with some individual variation.

Fig. 9.1. Locations of Swietenia humilis trees sampled. Each tree is denoted by a dot; the
17 trees selected for progeny analysis are circled (five in Las Tablas, three in Jiote, six in
Butus/Jicarito, and the isolated tree 501). No trees were sampled in the spotted enclosed
area adjoining the Las Tablas site marked by U. (From White and Boshier 2000)
120 C HAPTER 9 Fragmentation of Forests

The total number of trees was 98 in the largest fragment, Las Tablas, and was
smaller in Butus/Jicarito (44), Jiote (22), and Tablas Plains (7). Owing to
immediate losses of rare alleles of highly polymorphic simple sequence repeats,
their numbers of low-frequency alleles encountered in the trees amounted to
54, 29, 19, and 8, respectively.
The simple sequence repeats provided for complete paternity exclusion except
for one, so paternal assignment by exclusion was 100%. Observed maximum
pollen transport distances were at least 0.6 km for Las Tablas, 0.9 km for Tablas
Plains, 1.5 km for Jiote, 3.6 km for Butus/Jicarito, and 4.5 km for tree 501. The
true maximum distances may have been underestimated. As impressive as reports
on maximum reach of animal-mediated pollen transfer may ever , the totality
of the distribution of the quantities of transferred effective pollen is crucial.
Pollen flow had a contrasting pattern. Besides the predominance of near-neighbor
matings within 300-m distance, a large proportion of the pollen donors grew
outside each fragment in the study area. In Las Tablas, the largest fragment, 64%
of the matings were between near neighbors. Within-fragment matings domi-
nated also in Jiote (62%) and Butus/Jicarito (53%). In the smallest fragment,
Tablas Plains, only 31% of the effective pollen was produced within that
fragment. The lone tree 501 received less than 30% of its effective pollen from
distances between 1.2 and 2.4 km but over 70% of its pollen from unidentified
male partners at least 4.5 km away It turned out that restricted mating between
near neighbors was due to their asynchronous flowering. The flowers of the
Meliaceae are as a rule functionally monoecious, i.e., either the male or the
female organs are rudimentary (Styles 1972). The male flowering period of a
given tree may be much shorter, so one of two synchronously flowering trees
may fertilize the other but not vice versa. This reduces the number of near
potential mates over and above asynchronous flowering but opens the way to
some far-distance pollen becoming effective. Both the identity and the number
of pollen donors varied greatly among seed trees.
The number of individuals producing seed increased with increasing distur-
bance in this species. In general, a large proportion of the effective pollen came
from the largest fragment rather than the nearest one. Long-distance pollen
flow increased with a decrease in population size of the fragments, leading to
restoration of alleles lost in the original fragmentation, if not even to an
increase in the level of genetic variation within the local subpopulations. This
was facilitated by the ability of pollinators to move between the spatially
isolated stands.
The proportion of pollen flow was governed by the size of the pollen source
(large crowns or many trees, respectively) rather than its distance to the seed
tree. The latter was also true for a remnant tree in pasture. The levels of
heterozygosity among the seeds produced in the various fragments were only
little influenced by the sizes of the respective fragments. This was interpreted by
9.2 The Genetic Status of Fragmented Tree Populations 121

an interbreeding network of trees with sufficient gene exchange to compensate

for genetic drift in the parent generation. The widely spaced distribution of
Swietenia humilis did not serve as a barrier to gene flow. Consequently, the seed
was also very little differentiated between the fragments. Pairwise genetic dis-
tances were not correlated with geographic distances. The authors stress the
need for measuring gene flow between fragments in order to see whether reduc-
tion of the size of fragments is really followed by theorized genetic depletion.
In view of the rather efficient pollen dispersal, moderate distances of seed
dispersal rather than persistent biparental inbreeding must be responsible for
the existing spatial genetic structure of the old trees. Autocorrelation analysis
revealed a significant excess of relatives (presumably half-sibs) in the near
neighborhood (up to 50 m) of trees.
The production of seeds increased with increased degree of disturbance.
Since fragmentation does not necessarily have severe consequences for the con-
nectivity of fragments, also remnant stands have long-term conservation value.
Lone trees in pastures act as focal points or stepping-stones for pollinator
movement. The results for their progeny contrasted the prediction made by
Murawski and Hamrick (1991) that isolated trees receive pollen from fewer
donors and are more likely to deviate from random mating. The authors con-
clude that a system of nonspecialist insect pollinators is less susceptible to
habitat disturbance but stress the hypothesis that tropical tree species possess
the capacity of adaptation to fragmentation of their populations (From White
et al. 1999; White and Boshier 2000).

Increased atmospheric turbulence over a treeless area supports both the

transfer of airborne pollen and pollinator movement. It is, therefore, danger-
ous to extrapolate from data on pollen transfer in undisturbed populations to
draw conclusions about the impacts under fragmentation (White and Boshier
2000). The large size of breeding units of figs (Example 7.2) suggests that
forest fragments isolated by several kilometers can be part of a large viable
population at least of these species. As Law and Lean (1999) have shown, bats
as pollinators in Australian rain forest trees were highly mobile in a frag-
mented landscape. Their estimated home ranges reached 1,800 ha and might
cover several forest fragments. Since hand-pollinations using pollen from dif-
ferent fragments yielded more seed than those within the same fragment, it
may be inferred that the pollen transferred by bats between fragments is more
effective, since the mating partners are unrelated. The transfer of genes from
other subpopulations thus contributes to the genetic rescue of a subpopu-
lation as defined by Richards (2000) by increasing the probability of its sur-
vival. Unfortunately, the future density of bats in the fragmented landscape is
uncertain because of increased predation by owls. In Panama, Nason et al.
(1997) reported reduced reproductive success of Spondias mombin, a canopy
122 C HAPTER 9 Fragmentation of Forests

tree, from small islands. Seed viability was up to more than 30% less than in
contiguous forest.
The use of genetic markers is indispensable in measuring tree population
size as determined by the activity range of insects, birds, and bats (Nason et al.
1997; Sect. 5.2.3). In the case of only recent fragmentation, the search for
private alleles as indicators of reproductive isolation makes little sense. The
fragments are expected to share most genetic variants except the rarest ones,
unless they are very small and short-distance gene flow prevailed before human
interference. In general, genetic divergence may be due to pre-existing spatial
genetic structure (Aldrich et al. 1998).
Forest fragmentation does not only affect tree populations. For instance, the
decline in the number of animal species with decreasing size of forest frag-
ments has been studied in four Malagasy forest birds. However, owing to the
relatively large size of the fragments surveyed and the high migratory potential
of the birds studied, human interference was not reflected in genetic differen-
tiation of populations (Andrianarimisa et al. 2000).
Autopolyploidy is expected to partially offset the risk of genetic erosion and
the decay of heterozygosity. The incidence of polyploidy is high in some trop-
ical tree families such as the Bombacaceae (Bawa 1973). It is, however, not
known how many of these species are autopolyploids.
Several authors presented empirical data on the inverse relationship
between the size of isolated conspecific natural tree populations and the
amount of their genetic variation. Nason et al. (1997) focused our attention on
the differential short-term and long-term genetic consequences of bottlenecks.
Under the assumption of equal initial variation in the past, smaller genetic
multiplicity due to the loss of rare genes is to be expected in smaller popula-
tions. After several generations, overall genetic diversity of smaller populations
is affected more strongly (Nei et al. 1975). It has to be awaited whether empir-
ical studies will provide the evidence for these genetic changes in completely
isolated fragments. These authors also analyzed the quality of pre-existing
spatial genetic structure and of the spatial scale of fragmentation. Data presented
by Moran and Hopper (1983), McClenaghan and Beauchamp (1986), and
Billington (1991) largely confirmed the negative relationship between popula-
tion size and genetic variation. However, the establishment of a strict correlation
takes many generations of continuously small size of the reproduction-efficient
population until the results of multiple bottlenecks become manifest in diver-
sity, because in every generation with finite size a renewed size-dependent risk
exists that genetic variants will become lost. This risk also depends on the ini-
tial frequency structure of those variants. One may derive from this that the
sizes of man-made fragments of a spatially structured population are moder-
ately appropriate predictors of their genetic variation, as was found by Young
et al. (1996), Sork et al. (1998), and Aldrich et al. (1998). The findings of
9.3 Genetic Preconditions for Restoration and Persistence 123

Young et al. (1999) at allozyme gene loci in a perennial herb are consistent with
those in trees. Further to tropical trees, an inverse relationship was also found
by Hall et al. (1996) on the basis of their study of Pithecellobium elegans and by
White and Boshier (2000) studying highly variable microsatellites of Swietenia
humilis in recent forest fragments (Example 9.2). Little impact of fragmenta-
tion on the genetic system of another tree of the Meliaceae family, Swietenia
macrophylla, was reported by Cspedes et al. (2003). Also Prober and Brown
(1994) found moderately strict relationships between diversity and population
size; however, after discussing incomplete isolation and other reasons, they
considered the relationships surprisingly strong. Be the relationship called
strong or only moderate, the tendency involved in the results is beyond any
doubt. Among natural populations with more or less unknown past history,
the estimated degree of determination of genetic variation by population size
cannot possibly be expected to be as high as in artificial populations of model
insects in the laboratory.
When summarizing the results of the genetic impact of forest fragmenta-
tion, it has to be stated that there exists, of course, no empirical information on
the genetic change during a full tree generation. Nonetheless, according to the
estimates of various population genetics parameters, the reproduction system
of some tree populations is subject to major change with regard to mating sys-
tems, gene flow, and migration. Spatial genetic structures are affected by frag-
mentation both on a landscape scale and on a regional scale.

Genetic Preconditions for Restoration and Persistence

The attempt to precisely restore a forested area to its presumed original vege-
tation would certainly be all but promising. This applies even more to the
genetic structure of tree populations. However, taking account of some basic
preconditions for an adequate population size of tree species is helpful in
making decisions.
Since fragments of tropical tree populations are in a critical situation, man-
agers have to consider the size of a population that is worthwhile for preserva-
tion and possible later enlargement. The persistence of tree populations in
tropical forests over generations has certain preconditions, such as a minimum
population size, since small populations are prone to extinction sooner or later.
Besides the reduction of evolutionary adaptability due to the loss of genetic
variation (Sect. 7.2) caused by drift, the accumulation of detrimental inbreed-
ing effects is another genetic reason for increased vulnerability of small isolated
populations. Information on the size and structure of neighboring fragments
and their locations is important for making decisions on the restorability of
124 C HAPTER 9 Fragmentation of Forests

a fragment or a group of fragments (Loyn and MacAlpine 2001). Two

approaches towards conceiving population sizes adequate for persistence of
population fragments shall be briefly presented.
If mating is assumed to be random and selection to be absent, heterozygos-
ity is expected to decrease at a rate that depends exclusively on population size
(Crow and Kimura 1970). If the reproduction-effective size of the population
is 50, the decay of heterozygosity is expected to amount to just 1% per genera-
tion; hence, 50 is the minimum population size that keeps the speed of this
decay equal to or below 1% at least over short periods of time. Considering the
number 50 as operationally adequate is simplistic, since it does not account for
other factors such as genetic load and selection that influence the dynamics of
heterozygosity (Sherwin and Moritz 2000). Particularly in trees, the density
and the spatial distribution of the population members play a role in setting
the scene for inbreeding, as was described in Sect. 9.2. This number is still
referred to as the basic rule of conservation (short-term rule).
The tenfold population size of 500 individuals is not only expected to keep
the decay of heterozygosity below 0.1%. It is also expected to provide for a bal-
ance between the loss of genetic variants and their origination through muta-
tion. A population size of 500 individuals (long-term rule) is said to be
required to avoid immediate loss of diversity and to ensure the long-term
preservation of genetic variation and thus of evolutionary adaptability
(Franklin 1980; Frankel and Soul 1981). Experimental investigations in caged
Drosophila populations showed that the loss of genetic variation in populations
of 500 reproducing insects due to genetic drift is compensated by mutations;
however, this finding does not take account of the identity of the genetic vari-
ants and their adaptational relevance in nature. Furthermore, the time factor
has to be accounted for, since the generation period is much longer in trees
than in insects. The much larger number of 500 individuals can also be seen as
a gross safety measure. The 50/500 rule is weakly supported by experimental
studies on feral animal species and even more so on plant species but is still fre-
quently reported in the relevant literature (Frankel and Soul 1981). Last but
not least, the reproduction system of plants is much more complex than that of
animals. It has to be stressed that besides population size, imbalances in the
mating success of individual trees have to be accounted for, as was outlined in
Sect. 9.2. Consequently, reproduction-effective population sizes have to be con-
sidered when applying the basic rules.
Fragments of species-rich tropical forests of only a few hectares in size are
likely to be insufficient for the preservation of numerous tree species, if the
50/500 rule is adopted for forest tree populations. The long-term persistence of
species in tropical forests is only possible if certain additional preconditions
according to a different concept are fulfilled. The estimation of minimum viable
population (MVP) sizes as first conceived by Shaffer (1981) have attracted
considerable interest among conservation biologists (Soul 1987). The MVP
9.3 Genetic Preconditions for Restoration and Persistence 125

concept refers to two aspects of populations. One is the totality of critical factors
of the genetic structure of populations above a level below which inbreeding
depression and the loss of adaptive capacity would become a problem for their
continued survival. The other one is the size of an isolated population of a given
species in a given habitat that keeps the risk of extinction low. The MVP is to
ensure at some acceptable level of probability that the isolated population
persists in a viable condition for a given interval of time. Some authors have
specified these parameters as 95% probability and 1,000 years, respectively. This
period is rather extended at first sight; however, it is equivalent to a small
number of generations of forest trees.
Apart from genetic considerations, other factors such as demographic sto-
chasticy, i.e., random fluctuations of the sizes and perturbations of the age
structures of populations without associated changes in their genetic structure,
may also pave the way towards extinction (Lande 1988). The long-term growth
rate of a population is determined by the geometric mean of the growth rates
over time. The geometric mean is less than the arithmetic mean. Consequently,
the long-term growth of a population may be negative, although it presently
shows positive growth. Several nongenetic factors share in the determination of
the expected persistence time of small populations. Shrinking populations
require constant support and management. Also deterministic effects such as
the incidence of herbivores, pests and disease, competition, and the coloniza-
tion by invasive nonindigenous species represent threats to persistence. The
long-term conservation of forest fragments is more promising, the better the
habitat quality of pollinators and frugivores as seed dispersers (Nason et al.
1997). Reduced density of fruit-eating animals in small forest fragments leads
to lower recruitment of large-seeded animal-dispersed trees, although the trees
bear fruit (Cordeiro and Howe 2001). If the animal disperser of a tree species
is absent owing to subsistence hunting in fragmented forest, the seeds are only
gravity-dispersed. This leads to a closer relationship of the seedlings under
parental canopies. Therefore, defaunation also has an impact on the dynamics
of spatial genetic structures in trees which will become crucial at reproduction
(Pacheco and Simonetti 2000).
Fragmentation of tropical forests creates conditions that are similar to a
metapopulation structure. In unfragmented forests, the habitats of extinct local
subpopulations can be recolonized and mating contact between temporarily
isolated local subpopulations can sometimes be regained (Cspedes et al.
2003). In permanently or temporarily isolated subpopulations an increase of
inbreeding is expected, and genetic drift causes local losses of genetic variation
and evolutionary adaptability. Cspedes et al. (2003) concluded that metapop-
ulation models could be used to describe the genetic structure of plants in
highly fragmented landscapes. However, as Gilpin (1991) pointed out, not only
the sizes of the subpopulations but also the size of the total metapopulation
matter greatly. The question of MVP is then extended to the question of the
126 C HAPTER 9 Fragmentation of Forests

minimum number of interconnected local populations necessary for the long-

term persistence of the total metapopulation. Both of these quantities are likely
to be strongly reduced by man-induced fragmentation. Genetic variation
decays faster in the smaller subpopulations and these are hardly the source of
genetic variants that were lost in adjacent larger subpopulations (Aldrich et al.
1998). Therefore, both recolonization and gene flow originating from the
larger donor subpopulations are crucial for the persistence of patterns of par-
tially isolated population fragments (Hall et al. 1996). Habitat corridors may
provide avenues for gene dispersal between patches and first of all large trees
may contribute to the local genetic dynamics of the adjoining patches (Nason
et al. 1997). Only little pollen influx from genetically differentiated adjoining
fragments through corridors is sufficient to counteract drift-related genetic
change. Strip lines such as revegetated margins of streams and watercourses, live
fences, and regrowth on abandoned fields or pastures may support pollinator
movement (Law and Lean 1999). Restoration efforts aimed at increasing the
number of keystone species should therefore include the establishment of those
links between fragments (Nason et al. 1997; Galindo-Gonzlez et al. 2000; White
and Boshier 2000). If the effectiveness of corridors for connectivity of local pop-
ulations of various tree species can be proven to be valid in general, many tree
species may then persist as metapopulations, which often makes forest fragments
worthy of conservation. This is particularly true in regions that have been heav-
ily degraded. Few scattered big trees left in pastures serve as traps for genetic
information, as pollen from distant trees may sire their seeds and enrich the local
variation. Therefore, the favorable conditions offered on abandoned pastures for
survival and growth of seedlings and saplings of some species are augmented by
favorable genetic perspectives, although opposite effects were described for
Symphonia globulifera in Example 9.1. Remote-sensing data may help to select
priority areas and to design forest reserves that will have a good chance of con-
serving biodiversity (Saatchi et al. 2001). Some techniques of restoration in vari-
ous tropical regions were covered by Lamb et al. (1997).
As already expressed in this section, our methods of defining minimum
population sizes are admittedly rather restricted. Nevertheless, the considera-
tions of the type presented may help in making decisions. Since the 50/500 rule
does obviously not account for various nongenetic menaces to population
fragments of a given species in a given environment, minimum sizes much
larger than 50 should be maintained. If long-term preservation of forest with a
large spectrum of tree species is concerned, population sizes at least of all of the
keystone species must be much above 500. If a given population fragment falls
short of this lower bound, it may still be eligible for short-term maintenance.
Decisions of this type are supported by the presence of tree species that are
important in terms of ecology and/or economy. It has to be kept in mind that
numbers such as 50 or 500 must refer to the reproduction-efficient population
9.4 Recommended Literature 127

size introduced in Sect. 9.2 rather than the census number of reproducing trees.
In contiguous populations of anemophilous species this parameter can be
estimated from the variation in flower counts (Kjr 1996). More reliable in
tropical forest fragments is counting effective gametes on the basis of the
reconstruction of descent of seeds or even seedlings by means of highly vari-
able genetic markers such as microsatellites (Aldrich et al. 1998). This requires
much effort, since relevant genetic surveys have to be repeated in order to
account for variation in time (Murawski and Hamrick 1992). Eventually it is
sufficient to sample progeny of a restricted number of adult trees and to esti-
mate the size of the reproduction-effective population relative to the census
size. This ratio may serve as an appropriate reduction factor.
In view of the multitude of unforeseeable events, crisis-oriented immediate
decisions on long-term viability and preservation are required. Uncertainty
must be tolerated and managers be prepared to make best guesses on available
data (Menges 1991). Although the MVP concept is easier to apply to animal
populations, it is still useful also in connection with plants for several reasons
(Gilpin and Soul 1986). It reminds us that the focus of conservation must be
on populations rather than individual trees, and that only minimum require-
ments for persistence during a minimum period of time can be planned. Still
today, not all managers consider genetics an important factor in planning
preservation of fragmented forests. This is partly due to ignorance in a field of
biology conventionally felt to be difficult.

Recommended Literature

In his classic paper, Ledig (1992) discussed the various types of human impact
on forests and addressed their genetic implications. During the past 25 years
many books on conservation have analyzed the impact of fragmentation of
plant and animal populations. Six introductory chapters of the book edited by
Young and Clarke (2000) provide an overview of genetic and demographic
issues related to fragmentation. Two among the subsequent chapters covering
case studies deal with tropical forest trees. In the book edited by Brigham and
Schwartz (2003), many authors analyzed the preconditions for persistence in
various plant populations and designed management procedures. This book
demonstrates that the biological characteristics of the respective species must be
accounted for, if planning is to be reliable. In the book edited by Laurance and
Bierregaard (1997b), many aspects of the peculiarities of tropical forest frag-
ments and their appropriate management are viewed and discussed. A contri-
bution of Young and Boyle (2000) to the book edited by A. Young et al. (2000)
presents a brief but rigorous discussion of the genetic aspects of fragmentation.
Genetic Aspects of the Management 10
of Natural Tropical Forests


Research that is somehow related to the genetics of tropical forest trees has long
been largely confined to breeding of a few tree taxa that are suitable for plan-
tation forestry. Owing to the restriction of forest genetics to breeding work, the
genetic aspects of the management of natural forests in the tropics have
received only little attention and relevant studies are scarce. Numerous reports
on progress achieved through breeding of these taxa have been published but
only a few publications deal with genetic implications of the management of
natural forests.
Research on silvicultural systems for the sustainable management of tropi-
cal forests and their conversion to ecologically valuable species-rich production
forests has for several decades been supported by, for instance, German devel-
opment agencies (Lamprecht 1986). However, genetic aspects have been
neglected in this context. Although genetic diversity is one of the keystone
components of biological diversity, forest genetics research is still considered to
be of secondary importance. Only fairly recently has the preservation of bio-
logical diversity been accepted as an integral part of sustainable management
systems for tropical forests (p. 119 ff. in Deutscher Forstverein 1986; p. 50 f. in
Enquete Kommission Vorsorge zum Schutz der Erdatmosphre des deut-
schen Bundestages 1990; Ministerial Conferences on the Protection of Forests
in Europe, Strasbourg 1990: Resolution S2; Helsinki 1993: Resolution H2).
Since then, there has been growing public awareness of the fact that the
destruction of tropical forests leads to an irreversible loss of genetic informa-
tion of many species. Few studies on the genetic impact of the management of
natural forests were made in view of the need for the conservation of genetic
resources of tropical tree species (Chap. 14).
More recently, the number of genetic research projects on an international
level has increased. Some of the results will be addressed in the present chapter.
The subject deserves our interest, since much the surviving part of a tropical
forest is likely to be timber-production forest (Lamprecht 1986; Whitmore 1998).
130 C HAPTER 10 Genetic Aspects of the Management of Natural Tropical Forests

Selective Logging

A widely practiced silvicultural management technique in tropical forests is

various ways of selective logging and in particular target-diameter felling, i.e.,
removing all trees above a certain diameter limit (between 45 and 70 cm) with
or without restrictions, such as a sufficient number of adolescent trees left and
sufficient numbers of regenerates (p. 131 ff. in Montagnini and Jordan 2005).
It is, of course, problematic to define which numbers are sufficient from the
genetic point of view (Wickneswari and Boyle 2000).
Although the amounts of harvested wood volume are relatively low (Chap. 2
in Bruenig 1996; Chaps. 7, 10 in Whitmore 1998), the extraction of a propor-
tion of trees of merchantable size has a twofold genetic impact. It reduces the
number and density of reproducing trees in the populations of the respective
species and may imply dysgenic selection. This impact is stronger, if logging is
repeated in several successive cycles (polycyclic systems of logging; p. 136 f. in
Montagnini and Jordan 2005), and the shorter the intervals are in connection
with a lower diameter limit (p. 179 ff. in Bruenig 1996). As Murawski and
Hamrick (1992) have shown, the density of flowering trees in a population
varies over time. Much at variance from this is the effect of repeated logging of
a species, since its population density cannot recover anymore. Planting in
overlogged forest that once appeared to be suitable for enrichment planting is
problematic. Summarizing various pertinent silvicultural projects, Bruenig
(1996, p. 205 ff.) came to the conclusion that the results were by no means
encouraging. This precludes genetic enrichment at least in tropical rain forests.
From the genetics aspect, natural recovery might be a slow process. Things are,
of course, different in extremely overlogged forests. Selective logging has been
practiced for a long time on a small-scale basis but has drastically increased
with the increased demands for timber and with the progress made in improv-
ing logging technology (p. 87 ff. in Bruenig 1996).
Wickneswari et al. (1999) and Wickneswari and Boyle (2000) reviewed the
results of a comprehensive study in Peninsular Malaysia. In two regenerated
stands with 14 and 41% disturbance there was no adverse change of genetic
diversity after a single low-intensity logging event in comparison with unlogged
lowland dipterocarp forest. However, in a ridge forest the average loss of allelic
variants in five species with different life-history strategies after logging ranged
from 8 to 25% and their genetic diversity was reduced to values ranging from
5 to 23%. This reduction closely corresponded to the reduction in tree numbers
caused by tree harvesting itself. The loss of 25% of the alleles and 9% of genetic
diversity owing to logging was higher in species of low abundance (Shorea
leprosula) than in those of high abundance (Scaphium macropodum) with 8% of
the alleles and 5% of the diversity. One of the observed species, the climbing
10.2 Selective Logging 131

palm Daemonorops verticillaris, exhibited 14% allele loss owing to mortality

caused by the logging operation and the trampling of saplings.
In Central Africa Malcolm and Ray (2000) have found that the area of skid-
ding trails, secondary roads (for removing locally skidded logs), and primary
roads (frequently traveled by vehicles) themselves was still smaller than the area
of adjacent forest where the canopy was damaged. The diversity of tree species
among the saplings in the adjacent forest was reduced in comparison with that
in unlogged forest. Canopy damage was identified to be more hazardous than
clearing and eventually grading the ground for roads. With increasing time
since the logging operation, edge effects continued. These authors also devised
reduced-impact procedures in logging operations and road construction in order
to minimize damage to the remaining tree vegetation. Sinuous roads help to take
advantage of existing gaps and to avoid toppling trees. Dispersed felling helps to
avoid tangling of trees and to prevent too-large gaps. Last but not least, reducing
the number of trees harvested per hectare involves obvious genetic benefits,
although this procedure has economic constraints. Besides describing the eco-
logical impact of logging, Bruenig (1996, Chap 2) and Montagnini and Jordan
(2005, p. 139 ff.) have shown ways towards considerate procedures for this type
of rain forest use. Among several elements of low-impact logging Whitmore
(1998, p. 137) recommends the preplanning of snig-tracks, directional felling to
minimize damage to the canopy and facilitate log removal, and winching.
Logging is the major form of human disturbance, but it is still just one
form. A gradual reduction in vigor may result from practices that affect tree
health without tree removal, such as harvesting bark or resin (in dipterocarps
or in conifers such as Pinus and Agathis). The genetic impact strongly depends
on the product collected. Wickneswari and Boyle (2000) analyzed the impact
of harvesting wood and collecting other forest products such as fruits and seed.
Collection of fruits, nuts, and other nontimber products has long been prac-
ticed by the indigenous human populations (Prance 1994). This author also
pointed out that in certain dioecious species felling the trees for fruit collection
leaves only male trees.

Effective Population Density After Logging

The reduction of tree numbers causes an immediate reduction of genetic vari-

ation. Just as adverse is another aspect. Owing to the reduced density of indi-
viduals in the reproductive age class, the mating contact between the trees in
the population and even more so that between more remote parts of a popula-
tion is expected to decrease (Finkeldey 2002).
Eventually the effective population density is only loosely related to the
physical density, since the number of trees per unit area is hardly appropriate
132 C HAPTER 10 Genetic Aspects of the Management of Natural Tropical Forests

to describe the totality of conditions for the dispersal of sufficient pollen (Sect.
5.2). This is more relevant in species-rich tropical moist forests than in other
climatic zones, because the pollen of tropical trees is dispersed by pollinators
adapted to food trees occurring at a certain density. Increased distances to con-
specific neighbors are then less likely to be bridged by pollinators. Logging is
also likely to have an effect on the species spectrum of pollen vectors at least of
the keystone species, since these are the primary target of logging operations.
Logging is expected to have an influence on the mating system of the
remaining trees in general, since the balance between cross-pollination and
self-pollination is changed in favor of the latter. If the genetic barriers
against self-pollination are not fully efficient, self-fertilization as the most
severe form of inbreeding is implied. Since the largest trees produce most flow-
ers, their extraction has the most severe impact on the quantity of seed pro-
duced and its genetic structure.
Reduced mating contact between trees within populations due to selective
cutting results in decreased fertility of dioecious and self-incompatible species
(Ghazoul et al. 1998) and promotes inbreeding in species with a mixed mating
system. This is not just an expectation, as Murawski et al. (1994b) have shown
in Shorea megistophylla, a canopy-emergent species of the Sri Lankan diptero-
carp forest. These authors compared the estimated outcrossing rate in undis-
turbed forest with a density of ten trees of 20 cm or greater breast height
diameter per hectare and on plots that were logged 20 years before with a den-
sity of only two trees per hectare. This reduction is equivalent to an increase in
average tree distance from about 30 to about 70 m. It was unclear whether trees
that were left at the time of logging have since then entered the reproducing part
of the population. A genetic survey of seed at several polymorphic allozyme loci
provided for a comparison of an important parameter of the mating system on
the two plots. The multilocus estimate of the outcrossing rate t (Sect. 6.3.2)
amounted to 0.86 in the primary and 0.71 in the less dense forest; hence, the
outcrossing rate was significantly reduced owing to the disturbance.
Lee (2000) studied the impact of reducing tree density on the mating system
of another canopy-emergent dipterocarp species, Dryobalanops aromatica, in
two different types of forest in Peninsular Malaysia. In response to lower den-
sity and greater distance to the nearest conspecific neighbor (Table 10.1), the

Table 10.1. Mating system parameters of Dryobalanops aromatica Gaertn. f. in

primary and logged forest. (Adapted from Lee 2000)
Population Average distance tm rp
density (ha1) to nearest neighbor (m)

Primary forest 15 26 0.92 0.39

Logged forest 7 38 0.77 0.11
10.2 Selective Logging 133

outcrossing rate in the logged forest was remarkably reduced. Since even less
cross-fertilization in this species was estimated elsewhere (Chap. 13), it may be
concluded that self-incompatibility is presumably only partial, so the difference
of roughly 15% more self-fertilization after logging reflects the effect of less-
efficient pollen dispersal. If there once existed a family structure in the primary
forest, the consanguineous neighbors have been extracted, so after logging less
biparental inbreeding arises in comparison with that in the primary forest. Part
of the inbreeding due to the increased proportion of self-fertilization would
then be compensated by this condition. The striking difference in rp, the prob-
ability that seeds produced by the same seed tree are full-sibs, is difficult to
explain. The reduced density would be expected to reduce also the number of
male mates of the seed trees (i.e., the reciprocal of rp) instead of raising it from
less than 3 to 10 (Table 10.1). Eventually the increase in atmospheric turbu-
lence was favorable to pollinator movement and thus pollen dispersal. The
author also pointed out that the density of the flowering trees rather than that
of all trees had a major influence.
Kitamura et al. (1994) did not find a significant reduction in the outcross-
ing rate of this species in Borneo. These authors reported estimates of t = 0.86
in primary forest as compared with t = 0.79 in secondary forest, where the trees
larger than 60 cm in diameter had been logged 20 years before. However, the
physical population density may not simply be taken as the effective density, as
was indicated at the beginning of this subsection. Consequently, the estimates
of tm vary among populations in response to their numerous other determi-
nants. The authors interpreted the results by indicating high flowering density,
although in the secondary forest this was around 7 ha1, i.e., half of that in the
primary forest.
The results of Lee (2000) show that the increased intertree distances implied
by logging disturb the sensitive balance between mating system parameters at
least in particular situations. Besides pollination possibly being less complete,
the proportion of biparental inbreeding is supposedly lower but the proportion
of self-fertilization is raised. In a heavily logged population of Pinus caribaea,
Zheng and Ennos (1997) did not only find a marked reduction in the out-
crossing rate (tm = 0.894) but also substantial incidence of consanguineous
biparental mating. This was obviously due to highly irregular spacing of the
remaining trees.
The impact of logging was also studied by Dayanandan et al. (1999) in three
Costa Rican populations of Carapa guianensis, a species of the Meliaceae. In a
biological reserve used as a reference this species had a density of ten repro-
ductively mature trees per hectare. In two managed forests approximately four
and three large trees, respectively, had been removed about 10 years earlier. The
density of medium-sized trees (diameter at breast height greater than 35 cm)
after logging was still 11 and 15 per hectare, respectively. One of these two
134 C HAPTER 10 Genetic Aspects of the Management of Natural Tropical Forests

managed forests was surrounded by degraded forests and pastures; the other
was part of a large contiguous block of forest. Adult genotyped trees had an age
of more than 100 years; saplings were estimated to be between 5 and 10 years
old. In all three populations the genetic distance between the adult and sapling
cohorts was small; however, the average distances among the three sapling
cohorts (D = 0.15) was twice that among the adult cohorts (D=0.08), indicat-
ing reduced influx of external effective pollen prior to the establishment of the
sapling cohorts. Since no other reason for an allele frequency change of
microsatellites is thinkable, the hypothesis of reduced gene flow must be
accepted. It indicates incipient genetic divergence. The degree of logging must
be considered mild, anyway. There was no evidence of inbreeding in any of the
three populations.
Once the density of a population has been strongly reduced, the distances
among the trees approach or even exceed the distances of seed dispersal. This
induces spatial genetic structures in the form of family structures, i.e., neighbor-
hoods of related trees. The existence of family structures is most likely to exist in
barochorous species (Sect. 5.3.1) with their inefficient mechanisms of seed
dispersal. Hence, by logging close neighbors are removed. However, new and
more pronounced family structures are expected to arise, since seed shadows do
not overlap. The seeds in the same seed shadow are also sired by fewer pollen
parents and are expected to be closely related. This implies future inbreeding
owing to preferential mating of related neighboring trees. Indications of a fam-
ily structure were found in the dioecious species Altingia excelsa on the basis of
surveys of morphological traits and isozymes (Sakai 1985).
The negative consequences of inbreeding on phenotypic trait expressions,
i.e., inbreeding depression (Sect. 6.4.2), are difficult to observe in natural forests;
however, there is no doubt that inbreeding affects naturally regenerated popu-
lations just as much as it affects planted species. Inbreeding depression was
demonstrated by numerous studies also in tropical trees. The mating contact
among unrelated trees of predominantly outcrossing species must not be
unduly interrupted by management practices. This obviously requires not only
keeping minimum sizes but also keeping minimum densities of tree popula-
tions and preserving the food basis of pollen vectors. Owing to our restricted
knowledge of the mating system and of patterns of gene flow in most tropical
tree species, it is difficult to estimate threshold sizes and threshold densities of
populations. The long-term genetic effects of lowered density during subse-
quent generations is largely unknown. More research in this field is urgently
needed. Wickneswari and Boyle (2000) stress the desirability of research on the
response of tree species with different life-history strategies to logging disturbance
in general. Bruenig (1996, p. 166 ff.), Montagnini and Jordan (2005, p. 133 ff.), and
Wickneswari and Boyle (2000) describe various logging regimes mainly
developed in Southeast Asia that presumably differ in their genetic impact.
10.2 Selective Logging 135

The Question of Dysgenic Selection

Under regular silvicultural management the economically most desirable trees

are kept in the stand in order to improve the economic value of the stand incre-
ment. This practice hopefully involves also some genetic improvement in view
of subsequent natural regeneration.
Harvestable trees in target-diameter felling systems are selected on the basis
of a certain minimum diameter and stem form, so that during exploitation the
most vigorous, straight trees are removed. This means that from the time of
logging onwards, trees with superior phenotypes are excluded from reproduc-
tion. Conversely, the remaining population consists mainly of young or
slow-growing trees with inferior value for use. Their reproduction is favored;
therefore, the concomitant mass selection is dysgenic. Depending on the degree
of the heritability of traits this contributes to inferior expressions of the selected
traits in subsequent generations. Both the ecological and the economic value of
the population regenerated from the remaining trees may still be reduced.
Ziehe and Hattemer (2002) reported that target-diameter felling as prac-
ticed in old beech stands in central Europe had little effect on genetic structures
at marker gene loci. However, owing to heterozygote advantage at some gene
loci the most heterozygous trees were removed first. This had a minor effect on
the loss of rare alleles. However, at a gene locus involved in controlling diame-
ter growth target-diameter felling would substantially diminish the potential
for diameter growth in subsequent generations. The effect of dysgenic selection
for growth rate and stem form has often been feared but has rarely been veri-
fied by using rigid methodology (Finkeldey and Ziehe 2004).
This condition is expected to persevere over generations as pointed out by
Lindquist (1946, Chap. 2) in his early book on the role of genetics in Swedish
silviculture. Phenotypic trait expressions may not only be potentially affected
by inbreeding depression due to selective logging in tropical forests but also by
unintentional selection.
The heritability of economically important phenotypic traits is presumably
low in natural tropical forests owing to high environmental heterogeneity.
Furthermore, the correlation between the age and the size of trees is frequently
weak in natural tropical forests, i.e., the tallest trees are eventually but not nec-
essarily the oldest. Thus, if fast-growing trees are preferentially removed from
the population by selective logging, the risk of dysgenic selection is not neces-
sarily high. The magnitude of the damage due to dysgenic selection is still open
to debate. Whitmore (1998, p. 134) describes it as anybodys guess.
In order to minimize the risk of dysgenic selection, trees with superior phe-
notypic trait expressions should be able to contribute to the production of the
136 C HAPTER 10 Genetic Aspects of the Management of Natural Tropical Forests

progeny generation in proportions that are at least equivalent to their frequency

in the parent generation. Repeated logging involves repeated selection, provided that
notably many trees of the respective species are extracted during subsequent cycles.
The potential danger of dysgenic selection depends on the time of regener-
ation. If the progeny generation was established prior to the main logging
operation and natural regeneration was kept protected during extraction of the
logs, its genetic structures are not directly influenced. In this case the genetic
information of the harvested trees is represented in the progeny generation.

Natural Regeneration
The assumed ideal of naturally regenerated forest being maximally stable, maximally
diverse, and minimally susceptible or maximally resistant to unanticipated threats, is a
commonly held paradigm. The three properties are assumed to be simultaneously
achieved by the normal processes of forest regeneration without human intervention
(Namkoong 1999).
The sporadic exploitation of a forest is usually not followed by artificial regener-
ation. Even though a logged or exploited forest may be naturally regenerated, it
is still not a natural forest anymore. By definition, natural forests have never been
influenced by humans. Since this refers in particular to their reproduction, natu-
ral forests have to be discriminated from those that were naturally regenerated
after silvicultural treatment, particularly after selective logging. Since the inten-
sity of human influences varies, we have to consider degrees of naturalness.
Other human influences exerted as man-made environmental change of the
global water, energy, and nitrogen budget should not be considered marginal.
Namkoong (1999) and Wickneswari and Boyle (2000) state that virtually all
forests in the world are directly or indirectly affected by human activities.
However, it is difficult to decide, whether a given forest has escaped human
influence completely. The three properties mentioned in the prefatory quota-
tion of the present subsection are likely to be exacerbated with increases in
human intervention. Therefore, both climax forests and secondary forests are
still considered natural, if they were predominantly or completely regenerated
without planting or sowing. They are then likely to possess a family structure
(Sect. 5.3.2). Loyn and MacAlpine (2001) used the term natural to indicate
lack of discernible impact of modern technological society. Of particular
importance is that many if not all mature trees have contributed their share to
the subsequent generation. Since not all trees reproduce at a given time, the
extension of the regeneration period over several years up to several decennia
helps to achieve this condition. However, maximizing diversity does not neces-
sarily lead to maximizing stability (Altukhov 2006).
10.4 Genetic Aspects of the Manipulation of Dynamic Processes in Natural Forests 137

Since only few tree populations might reproduce panmictically, genetic

equilibria might be largely absent, so even without human interference some
genetic change during reproduction represents the rule rather than the excep-
tion (p. 175 ff. in Hattemer et al. 1993). The genetic structures of natural pop-
ulations have usually been shaped by adaptational processes over many
generations, even though periodic disruptions of benign environmental condi-
tions and shifts in the selective environment may be required for regeneration
(Namkoong et al. 2000). Nevertheless, the generation cycle is completed on the
same site. This stands for some continuity and is much at variance from artifi-
cial regeneration that is normally linked to a transfer of reproductive material
of the population.
A population that has never been transferred by humans and that has
undergone an uninterrupted sequence of natural regeneration is called
autochthonous. Such a population is expected to be adapted to the locally pre-
vailing environmental conditions. This does not mean that autochthonous
populations could not lack adaptational optima. Conversely, if a population
has ever been transferred by humans (in the seed stage), it is called allochtho-
nous. It is difficult to decide whether a given stand is autochthonous. It is nor-
mally not expected to represent optimum conditions for current or future
generations. If it does not fit into a geographic variation pattern of other,
hopefully natural stands (Chap. 12), its autochthony is doubtful. If artificial
regeneration of forests has created a mosaic of autochthonous and allochtho-
nous stands, the attribute of being autochthonous has to be understood quan-
titatively in view of the influx of genes from allochthonous neighboring
stands. Genetic aspects of artificial regeneration of forests will be addressed in
Chap. 13 in more detail.

Genetic Aspects of the Manipulation of Dynamic Processes
in Natural Forests

Von Gadow and Kleinn (2005) distinguish just two archetypes of forest man-
agement besides sporadic exploitation. Under rotation forest management the
totality of the standing volume is removed at the end of the rotation period
(Fig. 10.1). The stand is then artificially regenerated (Sect. 11.7), whereby
genetic structures are changed completely. This system is typical of intensive
silviculture as found primarily in the southern hemisphere.
Under a continuous-cover system the standing volume is never harvested
completely (Fig. 10.1). The age of the stand is undefined and forest develop-
ment does not follow a cyclic harvest-and-regeneration pattern. Instead, it
oscillates around a certain level. Natural regeneration is possibly going on peri-
odically depending on overwood retention.
138 C HAPTER 10 Genetic Aspects of the Management of Natural Tropical Forests

V RFM systems V CCF systems

age time
0 R R t1 t2
Fig. 10.1. Development of biomass (v) over age or time for two archetypical systems
of forest management. Left: rotation management (RFM) systems; R denotes the end
of a rotation period. Right: continuous-cover forestry (CCF); the biomass is largely a
function of the time elapsed since the most recent thinning or harvest. (From von
Gadow and Kleinn 2005)

Both archetypes may be sustainable (Sect. 10.5), although more risks are
involved in rotation forestry. A wide variety of silvicultural regimes have been
developed for the two systems. A forest development path or treatment sched-
ule (von Gadow and Kleinn 2005) comprising a sequence of operations related
to thinning, harvesting, and regeneration on the basis of growth models or
experience can be designed for any forest type or individual forest (Bruenig
1996). Continuous-cover forestry in species-rich tropical forests challenges
refined methods of controlling density, the species composition, the allocation
of trees belonging to the various species, and the regeneration at least of the
keystone species (Sect. 7.6 in Whitmore 1998).
Thinning as such in conifer stands without applying any phenotypic criteria
of tree removal such as described by Buchert et al. (1997) and El-Kassaby
(1999, 2000) is expected to lead to a reduced number of genetic variants and
reduced diversity. This reflects only the decrease in the number of individuals
left on a given area. As Hosius (1993) has shown in Picea abies, thinning from
above and thinning from below may have different implications for the
genotypic structure of the remaining stand.
Density and mating system parameters such as outcrossing rates are to be
considered sensitive indicators of a changed mating system owing to the
removal of trees for regeneration. Referring to western North American
conifers, El-Kassaby (1999) reported the surprising result that the incidence of
mating among relatives was higher in stands treated under the shelterwood sys-
tem (around 200 trees per hectare and 30% of the basal area are removed) than
in old growth (used as a control). One should have expected that the removal
of possibly related neighbors would force mating among less related trees.
Under this system also the correlation of paternity (rp) exceeded that of old
growth. However, under a patch-cut system with patch sizes ranging from 1.5
to 2 ha the incidence of correlated mating was typically reduced. Later analyses
of the same replicated field experiment (El-Kassaby et al. 2003) yielded the
result of small, significant treatment effects in only a few cases, although the
10.4 Genetic Aspects of the Manipulation of Dynamic Processes in Natural Forests 139

most drastic reduction was to fewer than 25 adult trees per hectare. Number,
density, and spatial arrangement of the seed-bearing trees are important deter-
minants of genetic parameters of the offspring.
Glaubitz et al. (1999) found no difference of genetic diversity and multiplic-
ity between native stands of Eucalyptus sieberi and their natural regeneration
achieved under the shelterwood system with 10% of the adult trees left. In a
study of regeneration in the closely related E. considensiana Glaubitz et al.
(2003) concluded that clear-felling and aerial resowing with local seed was more
efficient in genetic diversity conservation than natural regeneration under the
seed tree system. The number of seed trees must have been much smaller than
the number contributing genetic information to the seed collected.
According to Finkeldey (2002) continuous-cover forestry has the prime
impact on the population density of target species and on kinship of conspecific
neighboring trees. Experimental results from forests in both the temperate zone
and the humid tropics suggest that the latter are more sensitive because of their
a priori lower population densities; therefore, the consequences for inbreeding
and lowered heterozygosity are more pronounced in tropical forests.
Information on threshold densities is urgently needed.
Silvicultural management practice may considerably change environmental
conditions to an extent that populations lose their adaptedness. Adaptational
processes are then required to ensure the survival of the population. Evolutionary
adaptations are a manifestation of changed genetic structures (Sect. 7.2). Genetic
structures of populations are likely to be affected by silvicultural practices
although it is often difficult to prove adaptive or nonadaptive changes of genetic
structures in response to the human impact on forests.
The genetic dynamics during the reproduction of species-rich tropical forests
is rather complex. Frequently a succession of different dominant plant species is
observed (p. 377 ff. in Richards 1981). The reoccupation of a habitat by species
of late successional stages is only possible from surrounding populations if early
stages of forest regeneration are dominated by species of primary forests. The
life cycle of tropical forests is characterized by the disappearance not only of
trees but also of local populations. This holds in particular if forests are simul-
taneously regenerated naturally in a large area, for example, after a catastrophic
event. Extinction and recolonization are natural events both for pioneer species
and for species of later successional stages, including species of the climax
forest, although some species with less efficient systems of seed dispersal and
without a good seed or seedling bank are less tolerant to local extinctions.
Partially isolated populations experiencing a dynamics of local extinctions
followed by recolonization are defined as metapopulations (Hanski and
Gilpin 1991).
The population genetics consequences of a metapopulation structure cannot
be analyzed if only single populations are observed (McCauley 1991). A habitat
140 C HAPTER 10 Genetic Aspects of the Management of Natural Tropical Forests

is often recolonized by progeny of only a few individuals. Effects of founder

events were studied in the Neotropical tree Corythophora alta on the basis of
frequencies of haplotypes of maternally inherited chloroplast DNA (Hamilton
1999; Sect. 5.3.2). Reproduction-effective population sizes are usually small,
resulting in genetic drift, loss of genetic variation, and inbreeding (Gilpin 1991).
Genetic drift shaped genetic structures in a disturbed habitat of the tree
Symphonia globulifera owing to the reproductive dominance of a few pasture
trees (Aldrich and Hamrick 1998; Example 9.1). The genetic differentiation
among populations is likely to increase owing to drift effects. Metapopulation
structure retards adaptive genetic differentiation of subpopulations in response
to the selection by microsite conditions but favors the capacity of the total pop-
ulation to adapt to an environmentally heterogeneous habitat. The functioning
of a metapopulation depends on the relative sizes and densities of the subpopu-
lations as well as on their pattern, since these two factors have a strong impact on
the intensity and the directions of gene flow.
The genetic consequences of a metapopulation structure have rarely been
experimentally investigated in natural or managed forests of the tropics. The
temporal dynamics of genotypic structures of Cecropia obtusifolia, a pioneer
species possessing a population structure resembling a metapopulation
(Alvarez-Buylla et al. 1996), were discussed in Sect. 6.2.2.
Selective logging results in the extinction of local populations of at least
some species in species-rich forests of the humid tropics. Recolonization of an
area via migration starts out from surrounding populations possessing a
metapopulation structure. If under rotation forestry all the mature trees are
harvested before reproduction, artificial regeneration is obligatory. Since this is
mostly done by using reproductive material collected elsewhere, an artificial
process of migration is induced. Its genetic implications will be discussed in
Sects. 11.5 and 12.6.

Genetic Aspects of Sustainability in Natural Tropical Forests

With increasing knowledge of the complex structure and functioning of forest

ecosystems, the meaning of the term sustainable forestry has changed repeatedly.
This is due to frequent redefinitions of human relationships with nature. Also,
definitions of sustainability presented in the literature depend very much on the
view taken by the respective authors. Principles derived for sustainable manage-
ment must be ecologically viable, economically feasible, and socially desirable
(Salwasser et al. 1993). Bass (2001) considers sustainable forestry in terms of the
security of specific forest goods and services. The book by Ferguson (1996) rep-
resents a successful attempt to bring ecology and economy together. Gregorius
10.5 Genetic Aspects of Sustainability in Natural Tropical Forests 141

(2001) explained the sustainable treatment of resources under the aspect of

systems theory and drew conclusions on some basic forestry operations.
Essential elements of the definition presented by Bruenig (1996, p. 270) are
Sustainability is the capacity of a system in its entirety to persist and survive.
Within the time dimension of human perception, it relates to permanence and
steadiness of states and processes in contradiction to the fact that there is no
steady-state in ecosystems. In forestry, sustainability is a comprehensive ethical
principle, substantiated according to agreed criteria which are monitored by
agreed indicators of certain forest properties and judged by recognized stan-
dards of ecological, technical, economic and sociological state or flow vari-
ables. The capacity to persist and survive implies the capacity of
self-regulation and adaptive change.
This author also addresses the complexity of the relation between sustain-
ability and biodiversity. The latter term refers to . . . the variety and variability
among living organisms and the ecological complexes in which they occur.
Diversity can be defined as the number of different items and their relative
frequency. For biological diversity, these items are organized at many levels,
ranging from complete ecosystems to the chemical structures that are the
molecular basis of heredity. Thus, the term encompasses different ecosystems,
species, genes and their relative abundance.
This comprehensive definition made in 1987 by the US Governments Office
of Technology Assessment has been discussed by Boyle (1994). The need for
the integration of genetics and ecology was stressed by Hattemer and Gregorius
(1996), because regulatory adaptation and thus the stability of an ecosystem
are unthinkable without the genetic variation of the species spectrum (Noss
1993). Guidelines for sustainable management can be derived from the defini-
tion of sustainability.
Ecological sustainability has been overshadowed by the aspect of sustained
production in view of human demands of wood, timber, and other forest prod-
ucts. With regard to the future of tropical rain forest ecosystems, Honadle
(1993) analyzed causes of tropical deforestation and discussed ways to get away
from current human predation on tropical forests. This author proposed
strengthening of certain nongovernmental organizations and international
processes in order to promote sustained tropical forestry.
Absolute requirements for sustainable management of tropical rain forests
are effective measures to diminish the impact of uncontrolled logging practice
(p. 222 f. in Whitmore 1998). Bruenig (1996, Chap. 3) explained the concept of
sustainable tropical rain forests and derived strategies for their sustainable
management. Bruenig (1996, p. 131 f.) and von Gadow (2005, Chap. 4.5)
described the historical roots of the scientific concept of sustainable forestry in
central Europe. However, there is no doubt that among forest-dwelling people
in traditional societies the attitude towards the forest has been sustainability-
142 C HAPTER 10 Genetic Aspects of the Management of Natural Tropical Forests

oriented the world over. Unfortunately, the rural poor in the respective parts of
the world suffer most from the forest decline. Today, the ethical imperative
of intergenerational fairness and justice deserves most weight among the
motivations for sustainable forest management. This anthropocentric vision
has been accepted by the Brundtland Commission (World Commission on
the Environment and Development 1987). The present generation of men and
women should not impair the potential of future generations to cover their
demands. The idea of intergenerational justice might well be extended to
intercontinental justice, since the per capita consumption of natural resources
still varies greatly among continents but the worlds climate relying strongly
on globally sustainable forestry is shared by all living beings. Humans are not
only an integral part of forest ecosystems but play an ever more prominent
role in them.
Primack and Corlett (2005) analyzed the decline of tropical rain forests and
identified its complex causes. They described the international economic con-
nections and stressed the responsibility of countries outside the tropics for the
decline. Among possible solutions the authors discussed also certification of
forests and forest products (see also p. 223 in Whitmore 1998; p. 144 f. in
Montagnini and Jordan 2005; Bennett 2001; Putz and Romero 2001). In future,
genetic fingerprinting methods might play an important role in the prevention
of illegal felling in tropical forests, forest certification, and observation of the
chain-of-custody. The development of molecular genetics tools to test state-
ments concerning the origin of tropical wood from certified forest enterprises
is ongoing (Finkeldey et al. 2006).
Recent definitions of sustainable forest management account for the dynam-
ics of the systems and try to consider all factors necessary for the long-term
preservation of a balanced status. An assessment of the stability of a forest ecosys-
tem and the sustainability of wood and timber production requires not only
detailed analysis of nutrient cycling (Ulrich 1981) but also of the information
budget of tree populations as triggered by their genetic system. Sustainable man-
agement of forests implies not only the maintenance of a balanced nutritional
status and other chemical aspects. The functioning of a forest ecosystem rests also
on information, which is predominantly stored as genetic information of the
animals and plants that are part of the ecosystem.
Sustainability does not imply the preservation of a fixed status, for example,
of certain genetic structures. The objective is the continued existence of the
ecosystem itself rather than the preservation of certain components such as
individuals, genetic variants, or species. A necessary condition for the long-
term persistence of an ecosystem is the preservation of the evolutionary adapt-
ability of its keystone species (Sect. 7.2). Thus, apart from economic necessities,
the sustainable management of a forest requires the conservation of the genetic
resources at least of some keystone species (Ziehe et al. 1989; Chap. 14).
10.5 Genetic Aspects of Sustainability in Natural Tropical Forests 143

Most tropical tree species harbor considerable amounts of genetic variation

(Sect. 3.4) and are thus capable of evolutionary change. Preliminary experi-
mental results indicate flexibility of the reproduction systems and temporal
dynamics of genetic structures of tropical forest trees similar to those of tree
species of the temperate and boreal zone. Therefore, sustainable silvicultural
management techniques must not reduce the genetic variation and the adap-
tive potential of keystone species.
Prabhu et al. (2001) explained the value of a useful concept for defining sus-
tainable forest management as well as for assessing and for monitoring the
sustainability of forest management systems. Criteria and indicators are to out-
line conditions that should be met if forest management is to be deemed
sustainable (Boyle 2000). Criteria, standards by which a thing is judged, are not
direct measures but refer to large-scale combinations of information. An exam-
ple may be processes including human intervention that contribute to the
maintenance of genetic variation.
Indicators are variables used to infer attributes of the sustainability of a for-
est and its utilization. They must be relevant to the goal of assessment and must
deliver precise, meaningful, and reliable information. They must also be easy to
detect. They may be either absolute, rankings, or conditions relative to a
threshold. An example may be ongoing directional change of genotype or
allele frequencies due to selection. This indicator can be verified by repeated
genetic surveys (Boyle 2000). Another example, distances between forest frag-
ments can be bridged by pollen transfer, can be verified by the presence of
nonlocal private alleles in the progeny of neighboring populations.
It is difficult to assess the sustainability of tropical forest ecosystems and
their adequate management, particularly if economic and also sociological
aspects are considered; therefore, the description of a condition by a few indi-
cators was once criticized for being reductional. On the other hand, as Loyn
and MacAlpine (2001) put it, sustainable forest management demands that we
manage to sustain the whole ecosystem, and monitor appropriate subjects
which indicate how well we are doing. We have no choice but to use indicators
(Raison et al. 2001). If only genetic sustainability of ecosystems and their man-
agement is concerned, things become easier. Because of the widespread imple-
mentation of population genetics models that are appropriate for causal
interpretation of empirical findings, forest geneticists are better off.
Nonetheless, genetics may only contribute its share to the analysis of sustain-
ability jointly with other fields.
Adaptive and nonadaptive changes of genetic structures are caused by the evo-
lutionary factors drift, gene flow and migration, selection, and mutation, and the
mating system. As a rare event, mutation is not an essential factor during over-
seeable time periods and is neglected in this context. Selective logging, grazing,
forest fires, utilization of forest by-products, and conversion of forest to other
144 C HAPTER 10 Genetic Aspects of the Management of Natural Tropical Forests

land uses are examples of human influences on forests that are also expected to
change genetic structures (Namkoong et al. 2002). These authors proposed the
following four general indicators in the context of genetic variation as the basis
of adaptability: genetic drift, selection (in the widest sense), migration, and the
mating system. Some relationships are presented in Fig. 10.2.
The most fundamental ecological indicator is the amount of available habi-
tat in the landscape. Prevalent forms of forest destruction are conversion for
food production by clearing and fire. They are not considered selective in the
genetic sense. Loyn and MacAlpine (2001) have proposed the formulation of
ecological indicators of fragmentation. Their discussion is very helpful for
describing the physical conditions for gene flow with due detail. The mating
system, although part of the reproduction system, cannot be observed directly
but must be reconstructed from findings on the latter by routine methodology.
It includes also selection processes. Therefore, the closeness of the reproduc-
tion system to panmixia is presented as an indicator of the mating system. The
difference between the census number of populations and their parental pool
size was addressed in more detail in Chap. 9. Sexuality refers to sexual function
of monoecious genotypes and is thus used as a genetic verifier. On the other

Human Genetic Verifiers

Indicators genetic demographic
influences processes

Directional Gene Phenotypic

Forest Selection
genetic change frequencies means
Population Age/size class
Habitat differentiation distribution
Multiplicity, Census no.
Level of
Drift Parental
variation Diversity etc.
pool size

Gene flow Physical

Migration Mating (Effective isolation
Logging isolation pollen transfer)

Sexuality Sex ratio

Closeness of
Genotype frequencies Pollinator
Non-timber Mating reproduction
Correlated matings abundance
harvesting system system to
Homozygote excess Seed
Spatial genetic germination
structures Mortality
Fig. 10.2. Genetic processes in tree populations as influenced by forest events, their
indicators, and their verifiers. Dashed arrows represent only weak relationships. For
some further explanations see the text. (Adapted from Namkoong et al. 2002)
10.5 Genetic Aspects of Sustainability in Natural Tropical Forests 145

hand, the morphological sex ratio of dioecious populations is used as a demo-

graphic verifier (Sect. 4.2.3). Not all relationships are indicated in Fig. 10.2. The
genetic verifiers presented are but examples. Demographic verifiers are difficult
to show in their relation to indicators by this type of presentation. For instance,
seed germination and mortality have a close relationship to selection rather
than to the mating system.
The genetic indicators and their response to forest management practices
can be assessed by periodic surveys of the variation at gene markers and of
quantitative traits. Since this genetic monitoring using appropriate markers
(Weising et al. 2005) requires much effort, only exemplary surveys can be made
in some populations. Additional indicators are necessary in order to assess the
success of measures aimed at the conservation or restoration of genetic varia-
tion by forest management practices. Naturally, genetic indicators are valuable
for monitoring only to the extent that the respective processes can be managed
(Loyn and MacAlpine 2001).
According to their economic and/or ecological value, keystone tree species
possessing a variety of genetic characteristics and life-history traits should be
selected for monitoring. They may be suitable indicators representative of
many other organisms of the ecosystem. The assessment of a whole group of
species is worthwhile in view of the cost and the amount of information to
be gained (Loyn and MacAlpine 2001). Quantitative phenotypic traits as well
as biochemical and molecular genetics markers should be observed in order to
assess the dynamics of genetic variation of different parts of the genome
at adaptive and nonadaptive gene loci.
Continuous refinement of methods in response to the increase in knowledge
gained is imperative. The set of indicators presented in Fig. 10.2 may be specified
and improved considerably. If the results of periodic assessments of indicators
are treated as repetitions of an observational study, they may be integrated into
something like adaptive management (Prabhu et al. 2001).
The lay public seriously devoted to biological conservation looks at any
management of tropical forest as a disturbance. It is true that any human
interference somehow disturbs natural processes but not every disturbance is
alarming in view of the sustainability of forests. Also, tree species and forest
ecosystems differ in their ability to regenerate. We may not forget that even
though selective harvests of one or a few species may leave the forest environ-
ment physically undisturbed, mating partners for the harvested species may be
so diminished that regeneration is at least locally unlikely (Namkoong et al.
2000). Nevertheless, disturbances occur also in nature and challenge adaptive
processes in populations. Furthermore, in view of immense human pressure on
forests all over the globe, the appropriate management of forests is a necessity
just as is their utilization. Unless at least some regional economic value is
assigned to forests, human societies will hardly be convinced that forests are
146 C HAPTER 10 Genetic Aspects of the Management of Natural Tropical Forests

something worth conserving at all. Not much tropical forest will survive that is
not logged sooner or later. Therefore, the integration of appropriate logging
regimes into forest management will be of more help to forests than mere con-
cern about the inevitable. However, the avoidance of major disturbances is
imperative. If the new environments differ much from the predisturbance
environments, then adaptation to novel environments requires a large adaptive
capacity. This can be carried only in a large pool of potential parents available
for reproduction. In this case, natural selection may still not be able to increase
ecological security or economic productivity rapidly enough (Namkoong et al.
It should have become clear by now that it is not justified to consider any
change in genetic structures at any marker gene locus induced by a certain
management practice as a violation of the principle of genetic sustainability
(Finkeldey and Ziehe 2004). Genetic sustainability in the sense of Mller-
Starck (1996) is impaired only if the state of adaptedness or the adaptive capac-
ity (Ziehe et al. 1999) is endangered.

Recommended Literature

Stern and Roche (1974, p. 233 ff.) have presented a fairly general description of
how humans affect forest ecosystems. Finkeldey and Ziehe (2004) reviewed a
large number of studies made on the genetic impact of forestry in both temper-
ate and tropical forest. The focus of their paper is on forestry operations such as
artificial regeneration, soil amelioration, thinning and harvesting, and the regu-
lation of forest density and species composition. They come to the conclusion
that forestry induces most severe and detrimental genetic change in tropical
forests. Finkeldey (2002) has formulated a basic hypothesis on the genetic impli-
cations of continuous-cover forestry. Its implied assumptions are discussed and
some pieces of experimental evidence from both moderate and tropical forests
are presented. This hypothesis applies to any type of continuous-cover forest.
A joint test of all of these assumptions in one or several representative examples
is encouraged.
Provenance Research 11


During many generations the genetic structures of autochthonous populations

have developed under the influence of the evolutionary factors in conjunction
with historical events such as catastrophes. Since the totality of these effects is
likely to differ between the habitats of conspecific populations, the genetic
structures of these populations are likely to diverge. This differentiation is
more or less pronounced depending on the intensity of those effects, particu-
larly selection, and the length of the time period during which they have been
The model-supported study of the genetic differentiation of populations at
marker gene loci has opened the way to its causal interpretation. However, pop-
ulations usually differ only little at biparentally inherited marker loci. This con-
trasts with often much higher differentiation at adaptive and economically
important traits (Kremer et al. 1999). The study of the notable phenotypic dif-
ferentiation of populations is closely connected to their adaptation to the differ-
ent environmental conditions of their native habitats. Much regional variation in
day length, average temperature, and precipitation enforcing adaptation exists in
the temperate and boreal zone. In the tropics and subtropics the rainfall regime
varies more and takes precedence (p. 55 ff. in Morgenstern 1996).
The main objective of ecological genetics is to learn about the evolutionary
genetic change in plant populations brought about by ecological factors in their
natural habitat (Sect. 2.5). Provenance research has also a direct application: not
only the expression of adaptive traits but also other traits determining the value
of populations for use may be tested in experiments. On the basis of the results,
populations may be transferred to be planted elsewhere in order to increase sta-
bility and production. Hence, provenance research can also be seen as a first step
towards breeding and domestication. Only those populations can be bred that
justify the subsequent investment. In quite a few instances, arbitrary reproduc-
tive material has been introduced to other countries. Only later it was found out
that by prior systematic sampling and testing, large vain expenditure may have
148 C HAPTER 11 Provenance Research

been avoided. Owing to the rapid destruction of natural forests, there is an

urgent need to accelerate provenance collections (Ladrach 1998).


The geographic site where an autochthonous population has undergone its

evolutionary development is called its origin. With reference to origin, the
autochthonous population itself is commonly called a provenance and a test
for measuring its phenotypic traits in comparison with those of other popula-
tions and in a controlled environment is, consequently, called a provenance
test. The term provenance is also used for a sample (drawn at the seed stage)
from an autochthonous population and the planting material raised therefrom
to be included in a provenance test. Unfortunately, the term provenance is
further used for the origin of an autochthonous population. In legislation on
forest reproductive material the term is used even more generally for a place
in which any stand of trees is growing (EU 2000). This latter definition does
not discriminate whether this place is not only the growth site but also the ori-
gin of the respective population that is thus autochthonous or whether it is
allochthonous there. Although the latter usage is certainly appropriate for a
legal text, it has led and still leads to regrettable misunderstandings in the dis-
cussion of its genetic foundations. Zobel and Talbert (1984, p. 80) pointed to
the existing ambivalence and proposed denoting by provenance a place and
by seed source the trees growing there and from which seed is collected. Also
Zobel and van Buijtenen (1989, p. 33) and Evans and Turnbull (2004, p. 138)
used this definition that avoids misunderstandings about a site and the trees
growing there. This distinction helps to cure but part of the problem, since
provenance neither indicates the origin of a population nor indicates seed
source, whether a population is autochthonous or not (Jones and Burley
1973). Ladrach (1998) emphasized the need for clear definitions in this field.
An allochthonous population is regarded as introduced or exotic, if it grows
outside the natural distribution range of the respective species. An exotic plant
species is also denoted as a neophyte. A population that has been transferred by
humans within its natural distribution range is neither exotic nor autochtho-
nous anymore.
Unfortunately, the attribute indigenous is not helpful in relation to
autochthony. For instance, in EU (2000) it denotes a stand or seed source that
is either autochthonous or that has been raised artificially from seed, the ori-
gin of which is situated in the same region of provenance. The latter term
refers to a group of areas subject to sufficiently uniform ecological conditions,
in which seed sources show similar phenotypes or genetic characters. This term
11.2 Definitions 149

is again spongy under a genetic aspect and may illustrate the need for clear
definitions that might help to advance the discussion on the choice of appro-
priate forest reproductive material.
A race is an aggregate of several similar populations that are differentiated
from other races of the same species genetically or by particular expressions of
genetically controlled phenotypic traits. As an intraspecific division, races are
still connected by common descent. Just as populations, also races are difficult
to delineate. This is particularly true if they occupy the same habitat and are
thus sympatric. Owing to the effect of gene flow, also allopatric races occupy-
ing different territories present problems of clear-cut delineation. The pheno-
typic traits used for the description of races should be under strict genetic
control, i.e., their expression should show no or only minor environmental
influence. In plant taxonomy, races are often described morphologically as sub-
species or varieties. In the northern temperate and boreal zones, tree races
often originated in glacial refuges and have persisted during their postglacial
reimmigration. Also strains of animals or plants developed under the control
of humans are called races (Sect. 12.2).
An allochthonous population that after its transfer to its new growth site has
developed particular genetically controlled trait expressions owing to adaptation
is commonly called a land race (p. 89 in Zobel and Talbert 1984). The genetic
change is attributed to natural selection, artificial selection of trees with the best
performance as in breeding, or silvicultural selection. Eldridge et al. (1994,
p. 251) consider it essential that the respective population is exotic. Owing to the
low number of generations since then, it is hardly justified to expect substantial
adaptation of the genetic structure of a land race to its new environment. Zobel
and Talbert (1984), Eldridge et al. (1994), and Ladrach (1998) consider a popu-
lation with the quantity of genetic change as achieved during even only one gen-
eration a land race. Since a population originating from seed collected in its
progenitor population elsewhere can then have achieved improved adaptedness
only through viability selection, the customary definitions imply strong effects
of this type of selection that are linked to high mortality. The aforementioned
authors stress that the quality of the provenance used in the first planting is
crucial for the quality of the developing land race. This means that virtually all
allochthonous populations, particularly all plantations of exotic tree species,
deserve the attribute land races. For instance, first-generation stands of North
American conifers such as Douglas fir (Pseudotsuga menziesii) and eastern
white pine (Pinus strobus) planted early in the twentieth century in central
Europe have been called land races if they show good growth and their off-
spring show better health. In field tests with Pinus sylvestris, an exotic species in
the Netherlands, Squillace et al. (1975) found that the progeny of trees from
foreign sources planted before 1900 grew faster and were more resistant against
the needle cast disease than the progeny of trees from Dutch sources planted
150 C HAPTER 11 Provenance Research

later. A geographic pattern also existed, since offspring of trees in the northwest
of the country grew faster and were more resistant than those from the south.
The authors considered it highly likely that the combined effect of natural
adaptation and artificial selection was responsible for this gradual geographic
gradient. Eldridge et al. (1994, p. 31 f.) discussed the status of exotic eucalypt
plantations and stated that there is not much evidence in writing about the
improvement in eucalypts and other trees one generation after introduction
from natural stands. They attributed the superior growth of offspring of
Eucalyptus urophylla planted in the Congo in comparison with the growth of
the offspring of the progenitor population in Timor to the eventuality that the
neighborhood inbreeding effects of the natural stand had been broken down in
the plantation. Ladrach (1998) discussed several reports on land races in trop-
ical tree species. The critical factor in their formation seems to have been arti-
ficial selection rather than natural adaptation. Consequently, some land races
that have arisen on one continent showed superiority over natural provenances
elsewhere. In some instances, the poor seed production at the new growing site
had to be compensated by intervention of humans.
In future the importance of land races will certainly increase. The close
connection between their emergence and domestication will be discussed in
Sect. 12.2. Processes inducing immediate adaptedness may without doubt yield
arguments for the development and the existence of land races in trees.

Historical Development

The performance of forest tree plantations critically depends on the origin of the
reproductive material used for their establishment. Uncritical use of reproductive
material regardless of its origin has led to adverse experiences made with conifers
planted in several European countries. This insight gained considerable attention
from scientists and foresters concerned with the establishment of conifer planta-
tions. The foundation of the Verein deutscher forstlicher Versuchsanstalten
(Association of German Forestry Research Institutes) comprising the German,
Austrian, and Swiss forestry research institutes in 1892 and later, in 1929, the for-
mation of the International Union of Forestry Research Organizations (IUFRO)
have been intimately linked to research on the growth of different provenances of
economically relevant conifer species. From its very beginning this research field
has strongly relied on international co-operation and the exchange of research
material among countries. The history of the research on geographic variation in
forest trees was described by Morgenstern (1996, Chap. 1).
The first internationally co-ordinated provenance trials for tropical forest
tree species were established during the 1960s. Their main focus again was on
11.4 Provenance Tests 151

conifers. For example, the Oxford Forestry Institute co-ordinated a worldwide

series of provenance trials of Pinus caribaea (Barnes and Burley 1990).
Provenance trials and breeding populations were also established for other
tropical pines from Central America and Southeast Asia as well as for some
other conifers, e.g., Araucaria spp. (see Burley and Nikles 1972, 1973 for
At the same time the Food and Agriculture Organization (FAO) of the
United Nations organized and co-ordinated provenance trials of Eucalyptus
species in co-operation with Australian agencies (Example 11.1). Extensive
work has been done on E. camaldulensis. A series of teak (Tectona grandis)
provenance trials followed in 1973 and 1974 (Example 11.2). The Danish
DANIDA Forest Seed Centre co-ordinated these trials (Hedegart 1976; Kjr
et al. 1995). More internationally co-ordinated trials were established following
changed priorities for preferred plantation species in tropical countries.
Provenance trials were established for Gmelina arborea, Acacia spp., Dalbergia
sissoo, Casuarina equisetifolia, Cedrela spp., Leucaena spp. and other Central
American legumes, Calycophyllum spruceanum, and neem (Azadirachta indica)
(Burley 1973; Awang 1994; Kundu and Tigerstedt 1998; Sotelo Montes et al.
2003). Endo (1994) and Ladrach (1998) reported an impressve number of
provenance studies in Middle America made by CAMCORE. Many more insti-
tutions and organizations have been active in provenance research and subse-
quent breeding of tropical trees or have supported projects in this field, but
they cannot possibly be listed here.

Provenance Tests

A survey of genetic marker gene loci covering different tree populations can
readily be made using plant parts or seeds collected in those populations them-
selves. Comparisons between populations in phenotypic trait expressions are
not as easy to accomplish. The differentiation in phenotypic traits, particularly
the differential adaptedness, of provenances can be tested either in long-term
field experiments or in short-term tests under controlled conditions in phy-
totrons, greenhouses, or nurseries. The results o the latter type of experiments
are valid only if some overriding environmental factors can be readily simu-
lated in the controlled environment. For instance, Bekessy et al. (2003) tested
Chilean and Argentinean provenances of Araucaria araucana in the greenhouse
and found much higher drought resistance in those from the drier regions east
of the Andes.
Field trials require much more time, cost, and effort. They are nevertheless
indispensable experiments for genetically efficient selection of populations
152 C HAPTER 11 Provenance Research

suitable for the establishment of plantations and for the utilization of genetic
variation within populations by breeding. The main practice-oriented objec-
tive of the single field trial is to compare average trait expressions of the exper-
imental entries in economically relevant traits in a uniform environment. As
will be further described in Sect. 11.4.3, exposure of the experimental entries
to a variety of environmental conditions increases the information gained con-
siderably. Provenance tests represent also a collection of material that can be
utilized for more intensive breeding purposes.
Provenance trials are only one type of genetically motivated field trials.
Their establishment is an integral part of most breeding programs (Chap. 12).
Their methodology is not much different from that in forest tree breeding, and
many basic considerations of planning apply to any type of genetically moti-
vated field trials. Methods for their design, establishment, and analysis have
been covered more thoroughly in books such as those by Wright (1976) and
Williams et al. (2002). The latter publication contains worked examples and
reports on several experiments in tropical trees. Davidson (1995) prepared a
useful manual for this type of experimentation with tropical trees.
It is open to debate and to empirical research with model plants whether
part of the observed differentiation of provenances is the result of imprinting,
since the seed used for the experiments was produced under possibly very dif-
ferent environmental conditions. In experiments requiring great effort,
Skrppa and Johnsen (1999) found that offspring of Scandinavian Norway
spruce, Picea abies, possessed considerably less tolerance to frost if the seeds
were produced in a heated greenhouse (Johnsen et al. 2005a, b) . Although this
species is adapted to a cool climate, it obviously loses its adaptedness owing to
raised temperatures during seed maturation. Conversely, it acquires cold
resistance only during embryo development under lower temperatures in the
field. Hence, the behavior under different temperature regimes is controlled
by the expression of the genes responsible rather than the accumulation of
adaptive genes during many generations. The observed change is achieved
during a few months at a crucial stage of the life cycle. This result is undoubt-
edly relevant also for trees in warmer climates, because drought stress
(Vornam et al. 2003) had a similar effect as frost stress. Since corresponding
experiments with tropical trees have not yet been made, genetic variation has
still to be considered the main basis of their evolutionary adaptability. The
conservation of adapted, genetically variable and thus adaptable populations
is of particular importance, since rapid and pronounced environmental
change is predicted. Examples of expected environmental change in forest
ecosystems are air pollution and emissions in general, global warming, the
conversion of pristine forests to production forests, and other forestry prac-
tices. The conservation of genetic variation within and among populations is
11.4 Provenance Tests 153

a necessary component of sustainable forestry in view of the unforeseeable

future of forest ecosystems.
Provenances of most tree species that have been tested so far differ in the
expression of many phenotypic traits, however, this differentiation is rarely par-
alleled by findings on genetic structures at many marker loci (Bekessy et al. 2003).
A wealth of results of various provenance trials in tropical tree species have been
published during the past 30 years. They confirm that strong differentiation
among populations is the rule for almost all species studied in many economically
relevant traits. These results seem to contradict surveys of marker gene loci that
frequently have proved only little differentiation among populations. Typically,
the proportion of the total genetic diversity due to differentiation among popu-
lations (FST = GST; Sect. 3.3.2) at marker loci is less than 20% (Sect. 3.3). However,
one has to consider the different modes of gene action in genetic markers and in
those involved in the control of growth traits (Lewontin 1984). Also, different
evolutionary factors might have shaped the genetic structures at currently avail-
able marker gene loci and at those gene loci that contribute to particular trait
expressions. The only slight differentiation of tree populations at marker loci is
attributed to their life-history traits and, in particular, widespread gene flow
(Austerlitz et al. 2000; Chap. 5). The contrasting condition in phenotypic traits
must be explained by selection; however, selection was shown to cause strong dif-
ferentiation at an adaptive trait but only low differentiation at the controlling
gene loci if many genes, each with only minor effects on the phenotype, are
involved and if gene flow and diversifying selection are high (Kremer et al. 1999;
Le Corre and Kremer 2003). If selection resulted in pronounced differentiation in
spite of ongoing gene flow, the selection against immigrants is expected to be
strong. These immigrants are presumably maladapted and represent a severe
genetic load in the receptor populations. This condition has some bearing on the
choice of provenances for the establishment of plantations (Sect. 11.6). It has also
some bearing on the choice of populations as genetic resources (Sect. 14.5).

Types of Field Experiments in Provenance Research

In view of the cost, provenance trials aimed at the study of ecological genetics
problems might also serve practical goals. It has to be kept in mind that prove-
nance tests are planted at regular, wide spacing on cleared land without the
protection of a canopy in order to create an environment as uniform as possi-
ble. These highly artificial conditions are exactly the same as in plantations;
therefore, the results are expected to yield information primarily on the behav-
ior of trees in plantations rather than in the forest communities to which they
belong. Although provenance research ranks high in priority, many foresters
154 C HAPTER 11 Provenance Research

are only little aware that the implementation of its results ultimately leads to
stands to be established with allochthonous populations.
Several types of practice-oriented field trials have evolved:

1. The objective of a species trial is to assess the general suitability of a species

as a plantation species for a specific region. This practice-oriented type of
experiment is also called a species-elimination trial, since it is aimed at test-
ing a large number of species potentially suitable for artificial regeneration.
It serves also for the exclusion of those from further experiments that
would obviously fail in plantations. Owing to the fast juvenile growth of
tropical tree species, some rough information can be gained after only 34
years, so the duration of such trials can be rather short (p. 62 ff. in
Davidson 1995). Ideally, several species are identified as potentially suitable
for plantation establishment at the end of the trial. These species will be
included in further test plantations. Species elimination trials are pilot
studies for practice-oriented provenance research.
2. The performance of several populations of species that have been identi-
fied as potential plantation species is compared in provenance trials. Their
objective is the identification and selection of provenances that are partic-
ularly suitable for plantations on the basis of the expression of economi-
cally relevant quantitative traits. In first-generation tests populations
covering possibly a large part of the natural distribution range of a species
are included. Occasionally, locally well tried land races are also included as
a reference. Populations that proved to be most suitable as seed sources for
plantations are entered into restricted second-generation provenance tri-
als. All of these experiments are practice-oriented. Barnes and Simons
(1994) even state that before embarking on a breeding project both the full
natural range and the exotic range of the species should be sampled. In fact,
the progeny of exotic plantations performed among the best-growing
entries in several provenance experiments.
3. Some types of field trials may be combined in order to maximize infor-
mation gain and to save time. For example, species-provenance trials
have been established containing a few species each represented by sev-
eral provenances. These are particularly useful if decisions on the intro-
duction of exotic species are to be made. Also progenies of single seed
trees may be kept separate in provenance trials, unless the number of
experimental entries exceeds given limits. If the identities of single tree
progenies are maintained in a provenance trial, information on both
genetically based differentiation among populations and genetic varia-
tion within populations may be obtained. In tropical countries some
combined provenance-progeny trials have been established (Kanowski
and Nikles 1989; Chap. 12).
11.4 Provenance Tests 155

Traits Studied in Field Trials

In view of the cost involved in field work, periodically several traits determin-
ing plantation yield might be observed simultaneously. Several traits are usu-
ally recorded that characterize growth speed, such as tree dimensions reached
during given periods of time. At a young age the most important trait is plant
height. In later stages of the experiment, tree diameter (at breast height) is
recorded for the sake of easier measurement. Viability and tree growth indicate
harvestable yield per unit area, if not even stem volume or the total biomass
produced above ground can be measured directly.
Occasionally traits characterizing the health condition of the trees such as
leaf or needle color or the resistance against an insect or pathogen are relevant.
The presence or absence of symptoms of disease makes sense only if there is
notable infection pressure. The same applies to resistance against insect attack.
Other traits codetermine the use of provenances for commercial planta-
tions, i.e., the quality of the harvested product. Examples are morphological
traits such as stem straightness and branching habit (number, size, or angle of
branches). Depending on the planned utilization of plantations, wood charac-
teristics such as fiber length and cellulose content are measured. Sprouting
ability is an important trait if future plantations are to be coppiced (Quaile and
Mullin 1984). Palmer (1994) has presented a refined catalogue of desirable
properties of tropical forest trees.
Only some of these traits are measured on a continuous scale and are quan-
titative in the close sense of the word. Most traits characterizing growth (height
and diameter growth etc.) are quantitative; however, the viability is a trait with
only two possible trait expressions in the individual tree (survived or dead).
Many traits characterizing the health status are also qualitative with two
expressions only (infected or not) or are recorded in discrete classes. Some
traits such as growth form and stem straightness are actually not measured but
scored and are therefore subject to personal bias.
Some of the traits recorded in field trials are closely associated with the state
of adaptedness of the trees to the environmental conditions at the trial site.
This obviously holds true for viability and for traits characterizing tree health,
particularly under environmental stress or exposure to infection. Growth traits
are not necessarily positively correlated to fitness. Other economically impor-
tant traits, for example, stem straightness and length of branch-free bole, are
hardly associated with fitness. Trees showing superior growth may even possess
lower viability. Fitness components connected to fertility and reproduction are
only rarely observed in field trials. Thus, the provenances that are identified as
superior in field trials are not necessarily the best-adapted to the environmen-
tal conditions at the experimental site.
156 C HAPTER 11 Provenance Research

Design, Conduct, and Analysis of Provenance Trials
Collection of Material for Provenance Trials

The modalities of seed collection are a largely neglected element of the method-
ology of provenance trials. Principles of seed collection must be observed in
order to prevent the induction of undesirable genetic change.
Many details of the reproduction system that are of relevance for seed collec-
tion have been studied by means of genetic marker investigations. Above all, the
genetic structures of the seeds produced by a tree population vary over time. In
order to achieve a seed sample that possesses genetic structure representative of
its basic material, seeds produced in several years have to be mixed. This in turn
would require seed storage until material from all seed sources could be collected
repeatedly, an impossible undertaking. The seed of many tropical tree species, for
example, the dipterocarps, cannot be stored for long since they are recalcitrant
(Schmidt 2000). Consequently, seed should be collected in a year of abundant
seed production or just after most of the trees have flowered abundantly, as
Hodgson (1977) recommended with regard to seed orchards (Sect. 12.8.1).
The seed lots collected from individual seed trees are genetically differenti-
ated. This is due to the variation among the individual genotypes of the seed
trees themselves and to the differentiation of their individual effective pollen
clouds (Sect. 6.2.1). The less the seed lots differ, the easier it is to arrive at a rep-
resentative sample of the genetic structure of the reproductive output of the
seed source.
Figure 11.1 shows factors responsible for the differentiation of the individ-
ual effective pollen clouds. The higher the density of the population, the more
efficient the system of pollen dispersal, and the more homogeneous the physi-
cal structure of the population are, the larger is the expected effective number
of trees involved in cross-fertilization and the less are effective pollen clouds
expected to be differentiated (Sect. 6.2.1). Excessive self-fertilization that might
have an impact on phenotypic trait expressions may easily be tested in small
seed samples before a major collection. Genetic incompatibility (Sect. 6.5.1)
and the degree of self-sterility (Sect. 6.5.2) are beyond human control.
However, sufficient size of the seed source helps to capture many S-alleles and
provides for high fertility of stands to be derived from a seed source. The effect
of the genetic differentiation of neighborhoods conditioned by spatial genetic
structure, correlated matings, and hence a low number of effective male mates
of seed trees contributes directly to genetic differentiation among the seed lots.
The more pronounced the mass flowering of the trees, the less may flower phe-
nology contribute to this differentiation.
11.4 Provenance Tests 157

Effective distance to Population density

potential male Wind speed and
mating partners wind directions
Pollinator behavior and
sedimentation speed
of windborne pollen

Physical stand
Differentiated structure
Influx of external
effective pollen individual effective
pollen clouds
Spatial genetic
Flower phenology

Incompatibility and matings
self-sterility Excessive
Fig. 11.1. Causes of genetic differentiation of the pollen that becomes effective in indi-
vidual trees of a population (individual effective pollen clouds)

In forests consisting of just one dioecious tree species the differentiation of

the seed lots produced by single trees can be just as pronounced as in species-
rich forests. For instance, the 35 seed lots sampled in each of two populations
of the dioecious, wind-pollinated conifer Araucaria angustifolia shared only
70% of their alleles at six enzyme gene loci (Sousa et al. 2005; Sect. 5.2.2). On
the basis of correlated matings, the average effective number of male mating
partners was estimated to be only 3 or 4. Obviously, the female trees repre-
sented by half of the population greatly impair pollen dispersal. However, the
sedimentation speed of the pollen in this species is the highest hitherto found
in conifers, which may make this species unique. The differentiation of the seed
lots produced by single trees calls for seed collection from many trees.
If seed collected from individual trees is bulked arbitrarily, as is done in
many population genetics surveys, genetic structures depend very much on
the incidental amounts contributed by individual trees to the total sample.
Consequently, it is hard to predict whether a close or only loose genetic
relationship between a population and its reproductive output is detected.
Furthermore, the modalities of sampling play a role in whether panmictic
equilibrium (Sect. 6.2.2) is reached.
Approximately equal shares of the individual seed trees are desirable. Since the
vastly different amounts of seed produced by individual trees are primarily due
to their different crown sizes reflecting their different competition situations
158 C HAPTER 11 Provenance Research

rather than genetic effects, it hardly makes sense to collect amounts that are
roughly proportional to their seed production in the given year.
A minimum distance between the seed trees (at least double tree height or
100 m) helps to avoid harvesting seeds from related trees. This recommenda-
tion is particularly important if the gene flow system suggests the existence of
a family structure in a native population. Seed collection from close neighbors
also increases the risk of sampling progeny from reciprocal matings.
Owing to scarce reproduction in parts of the distribution range of the
species, the seed required for a provenance trial cannot be collected in 1 year.
Storage until completion is not possible for many tropical tree seeds. An expe-
dient solution of this conflict is to split up the experiment and compare sub-
sets of the provenances on the basis of the same age reached in different years.
The storage of seeds as practiced for many tree species of the temperate zones
is impossible for many tropical trees with recalcitrant seeds. One has to be
aware that the losses of germinability that are unavoidable even in conifers of
the temperate and boreal zones lead to genetic change of the stored seed. The
mortality during seed storage is selective as shown by Melchior (1986a) for
aspen (Populus tremula) seeds.
Very common in the past have been mail-order collections. In these cases
foresters in the region of origin of the seed sources were contacted and were
asked for shipment of local seed lots. The colleagues sent a sample of seed,
mentioning at best the number of seed trees, the distances between those trees,
and the proportions of seed collected from them. Many provenance experi-
ments with conifers of the northern hemisphere started out from seed lots
representing unspecified mixtures of seed from normally at least ten trees. It
is easy to comprehend that trait expressions observed in such provenances
are possibly biased considerably. The moderate demand for seed tempts us into
seed collection from a small number of trees only.
After all, seed must be harvested from a minimum number of trees. Several
authors recommended at least ten to 25 seed trees as sufficient to represent the
genetic structures of a population (p. 30 in Eldridge et al. 1994; p. 10 in
Williams et al. 2002; p. 84 in Davidson 1995). Population genetics considera-
tions suggest that these numbers are insufficient to obtain a representative
sample of the genetic structure of a population, since not only the genotypes of
the seed trees but also the genetic structures of their effective pollen clouds are
differentiated. Certain moderately frequent alleles and allele combinations are
expected to be severely underrepresented or overrepresented if seeds are har-
vested from only a small number of seed parents. Consequently, these numbers
represent absolutely lower bounds. However, as Williams et al. (2002) note,
even 25 seed trees might be hard to find in small local populations. It is then
open to debate whether populations of this size are eligible for entry into
provenance experiments.
11.4 Provenance Tests 159

Selection of the seed trees on the basis of phenotypic superiority should be

avoided. The attempt to combine seed collection for a provenance experiment
with something like plus-tree selection (Sect. 12.5) may lead to a biased sam-
ple. As a measure of precaution, small trees should not be favored as seed par-
ents, although it may be easier to collect seeds from them.
Further recommendations apply to the necessary documentation of all
operations and the deposition of copies of all records in several safe places.
Particularly, the information on the geographic location of the seed sources is
important. Good test results of a provenance will be exploited by later mass
collection of seed. In this case the seed source sampled for the test must closely
correspond to that of later mass collection. For short-term tests it is worthwhile
mapping and possibly labeling the seed trees. In most tropical tree species, seed
is harvested directly from standing trees. In view of major differences in ger-
minability, keeping single-tree seed lots separate until sowing is advisable.
Separation of single-tree progenies is, of course, necessary if the provenance
test is to be combined with a progeny test.
Last but not least, the seed collected from an autochthonous seed source is
more representative of its genetic structure, the less the external influx from
large, presumably differentiated, allochthonous stands in the vicinity is to be
Production of Planting Stock

In view of the environmental influence on planting stock the seed must be

sown out in the same year in the same nursery. It is an absolute precondition
of experiments that all experimental material has received equal pretreatment.
Particular care is devised during the germination phase. Since the plant
material of the experimental entries must receive the same nursery treatment,
seed lots and later the seedlings must be carefully kept separate. Losses of plant
material have to be minimized, since any such loss might be selective. The
resulting genetic change induced by environmental conditions that are atypical
of those that the trees are later exposed to may bias the data not only on
survival. In some provenance experiments aftereffects of differential nursery
treatment have been observed.
Experimental Design

Regrettably enough, even many years after the principles of the planning of
experiments as conceived by R.A. Fisher had been introduced in research,
costly provenance trials with forest trees have been established without paying
160 C HAPTER 11 Provenance Research

attention to appropriate design. Although older provenance trials were laid out
without adequate design, they are still an important source of information on
the growth of trees of different origin. This is particularly true for the first
provenance trials set out by P.A. de Vilmorin in Pinus sylvestris as early as
between 1820 and 1840, and early trials established with tropical trees such as
teak (T. grandis) (Coster and Eidmann 1934; Coster and Hardjowasono 1935).
However, small differences between provenances can hardly be entrusted and
generalization of the results requires caution.
Only some general suggestions for the design and the analysis of genetically
motivated field trials in forestry will be made. This topic is covered with due
detail in other publications (e.g., Williams et al. 2002).
Principal elements of experimental design now considered indispensable are
replication of the entries, the random layout of the experimental units (plots),
and the control of the experimental error. The plots may comprise just one tree
but mostly comprise many more trees. Single-tree plots are suitable for short-
time experiments only. Larger plots provide for later thinning when the indi-
vidual trees require more growth space. They are the method of choice for
medium-term tests. Commonly used are square plots with 4, 9, 16, or 25 trees.
Larger square plots reduce the effect of competition between adjacent plots
with possibly widely different growth speed of the trees growing there. In four-
tree plots every tree has more neighbors in other plots than in its own plot. In
a square plot with 25 trees, nine of them have no neighbor belonging to an
adjacent plot. In row plots the effects of competition between adjacent plots are
Four replicates of an experiment with a plot size of 25 trees leads to 100 trees
per provenance. Cotterill (1990) found that for estimating the provenance
means with sufficiently low error a total number of 200 would be required.
Depending on the heritability of the trait (Chap. 12), five to ten families should
be represented among those. Considerations of genetic representativity call for
larger numbers of families.
The randomized complete block design as proposed by R.A. Fisher 80 years
ago has become the most frequently used design for genetically motivated field
experiments in forestry, if not even for just any experiments. In this design,
a block contains just one plot of every entry. The spatial arrangement of the
plots of the various entries within the block is random; therefore, the design is
called random and complete. The design comprises several complete blocks (or
replicates). Statistical considerations call for a certain number of replicates.
If p provenances are to be tested in r replicates, the experiment contains pr
plots. If on every plot t trees are planted at a spacing of k meters, say, the
experiment requires an area of prtk2 square meters plus the space for one or
two border rows. If, for instance, only 20 provenances are to be compared in
a short-term experiment containing four replicates with nine trees per plot at
11.4 Provenance Tests 161

a spacing of 2 m 2 m, an area of 2,880 m2 is required. After including two

border rows, the area is 0.34 ha. Testing the provenances in five replicates with
16 trees planted in every plot requires an area of a little more than 0.7 ha.
Testing larger numbers of provenances soon leads to areas that are difficult to
get with adequate homogeneity. The same is true with larger plot sizes planned
for a longer duration of the experiment and accounting for thinnings.
Larger plot sizes reduce the risk of losing plots owing to mortality. Trials in
randomized complete blocks as the simplest design are insensitive to the
implied increase in complexity and have also for this reason been used widely.
The statistical evaluation of other designs requires complicated methodology,
if data are missing. However, the recent improvement in computational facili-
ties allows us to analyze also incomplete data, so incidentally missing plots do
no longer represent a problem. Now as before, the conformity with the princi-
ples of experimental design is a precondition for adequate analysis.
The blocks of an experiment may contain only part of the entries. In these
incomplete block designs the size required for one block is reduced, which pro-
vides for better control of the experimental error than the design using complete
blocks. There exist large numbers of incomplete block designs. They are recom-
mended by Williams et al. (2002) as efficient standard designs for experiments
with forest trees.
Choice of Experimental Sites

The choice of an appropriate experimental site is frequently the most difficult

practical problem. The research institutions conducting field experiments
often have no access to land that is most desirable for test plantations. The site
should be as homogeneous as possible in order to keep experimental error low.
The size should not comprise a systematic trend in edaphic or climatic condi-
tions as they occur on a slope. The consequences of a systematic change of
environmental conditions can only partly be reduced by an appropriate modi-
fication of the experimental design (p. 51 f. in Eldridge et al. 1994). Also sites
with obvious heterogeneity of growth conditions such as small-scale variation
are unsuitable for experimentation.
The trial site should represent the ecological conditions of the region of
future mass cultivation of the selected provenances. If the test environment is
atypical of the environmental conditions of future use, the transfer of the
experimental results to practical forestry is less efficient or may even lead to
failure. The establishment of field trials on inadequate sites is one of the most
frequent and foreseeable reasons for disappointing results or even complete
failures of experiments in tropical countries. The site should also be easily
accessible and be safe from destruction.
162 C HAPTER 11 Provenance Research
Number and Distribution of Locations of an Experimental Series

The ecological conditions prevailing at a given site induce certain growth dif-
ferences between the provenances. Apart from the genetically conditioned
growth potential of the provenances, these differences may be larger or smaller,
since the provenances respond to the environmental conditions in different
ways. Therefore, it is strongly recommended to plant an experiment on more
than one site as an experimental series. Statistical considerations suggest that
the design of the various trials should be identical.
According to experience with many series of experiments, not only the
degree of differentiation changes with the planting site but also the ranking of
the provenances. For instance, a provenance displaying fastest growth in one
location may be among the slow-growing provenances in another (Example
11.1). This phenomenon is described as interaction between provenances and
environments. The existence of interaction between genetically defined exper-
imental entries and the environment or G E interaction is commonly
experienced. Interaction in the close sense, i.e., changes in the ranks, calls for
the selection of different provenances at different locations. The importance of
interactions relative to the variation between provenance means can be meas-
ured by the relative sizes of variance components estimated by analysis of vari-
ance. Since only changes in the ranks are crucial, the consistency of the
rankings may be measured by the concordance statistic (Hattemer 1969). In
recent years the evaluation of interactions has been considerably improved
with the help of refined statistical models. Raymond and Lindgren (1990) have
discussed procedures of deploying provenances over planting sites. They pro-
posed a new approach based on the reaction of provenances to given changes
in ecological severity between the origin and the planting site.
The locations of the experimental series should have climatic and edaphic
conditions covering the typical spectrum of those in the region of future plant-
ing. In order to avoid serious setbacks, only those provenances with good per-
formance on several sites should be recommended for planting. The superior
performance of such provenances is considered environmentally stable.
Recommendations of provenances for planting involve a high risk of failure if
they are based on only one or a few field trials. This obviously holds even more
true, the more the environmental conditions at future planting sites differ from
the environmental conditions at the experimental sites.
The establishment of an experimental series requires more planting stock
and more area; however, the number of replicates in the single experiments can
be reduced. Experimental series require also much more effort in coordination
and conduct.
11.5 Provenance Differentiation and Geographic Variation Pattern 163
Observation Period of Field Experiments

The effort required for recording the data from a single tree can be low (e.g.,
measurement of breast-height diameter) or considerable (e.g., certain physio-
logical variables such as water-use efficiency). In most cases data are recorded
repeatedly in order to get both early results and information on their temporal
dynamics. The estimated means of the various traits are used for ranking the
provenances. The ranks may change considerably during the duration of a trial.
For example, in a trial in northern Thailand a land race of Pinus merkusii from
Papua New Guinea proved to grow much faster than provenances from
Thailand (Example 14.1). However, with increasing age of the trial this
land race lost its superiority in height growth over provenances from north-
easternThailand. No significant relation between the height of trees at the age
of 3 and 13 years was observed in this experiment (Changtragoon 1984). Thus,
with increasing age of the trial, conclusions on provenances with superior
growth had to be repeatedly revised.
The length of the observational period depends on the objective of the
experiment, the species, and the traits to be assessed. Species-elimination trials
in tropical countries are often restricted to a duration of 34 years. This time is
considered sufficient to assess the suitability of a species for plantation forestry.
Provenance trials should have a duration of at least half of the planned rotation
period. The fast juvenile growth of most tropical plantation species and their
short rotation periods allow the length of the observation period to be reduced
considerably in comparison with that for field trials with temperate or
even boreal species. For instance, in the context of selecting provenances of
E. grandis for high pulpwood production, Ladrach (1998) mentions a rotation
period of 6 years and a test period of 3 years.

Provenance Differentiation and Geographic Variation Pattern

Two basic types of geographic variation have been described (Langlet 1963;
Stern 1963; p. 89 ff. in Stern and Roche 1974). Since the differences in trait
expressions mainly reflect the results of adaptation to their habitats (origins),
the pattern of selective environmental factors has to be considered. Some of
these factors, such as day length and average annual temperatures, vary con-
tinuously. Edaphic or biotic factors such as the occurrence of competing
species may vary more or less abruptly. The totality of these effects finds its
common expression in the variation pattern of the populations.
164 C HAPTER 11 Provenance Research

Clinal variation, a term coined by Huxley (1938), describes a continuous

gradient in the frequency of genes or in average expressions of a trait in popu-
lations that closely follow an ecological gradient. For example, Langlet (1936)
described a close relationship between the relative dry matter content of the
needles of 1-year-old seedlings of Swedish Pinus sylvestris provenances and the
latitude and elevation of their origin. The relative dry matter content increases
with increasing latitude and elevation of the origin and thus with the gradually
decreasing temperatures.
Ecotypic variation, a term proposed by Turesson (1922), denotes discrete trait
expressions of populations or races of a species in adaptive traits following a dis-
continuous variation pattern of the selecting environmental conditions.
Different ecotypes may be found in proximity to each other. A relationship
between the spatial proximity of populations and their genetic similarity does
not necessarily exist.
Clinal variation is commonly described by regressions of trait expressions of
the different provenances on the expression of ecological factors of their ori-
gins. In some parts of the world ecological factors are closely correlated to the
geographic coordinates. Average temperatures and rainfall may be correlated
with latitude and altitude. Day lengths are determined by latitude; therefore,
the geographic coordinates are commonly used as a replacement for the long-
term averages of meteorological data, since no measurement is required. Both
types of variation patterns are often simultaneously observed and may overlap
in forest tree species (Example 11.1).
The evaluation of the geographic variation pattern and its interpretation by
ecological data requires much more intensive, preferably rangewide, sampling of
populations. For practice-oriented provenance tests the sampling might concen-
trate on a region with environmental conditions similar to those for the prospec-
tive use of well-performing provenances. Garnier-Gr and Ades (2001) used
several easily measured variables to statistically explain the variation in average
diameter growth of more than 60 provenances of E. delegatensis on seven
Australian sites. Two groups of variables were (1) bioclimatic data such as solar
radiation, temperature, and precipitation and (2) data on the ecological site qual-
ity of the origin. The models explained high percentages of both the variation
among provenances and the provenance by environment interactions. They rep-
resent an excellent means for locating adapted provenances for given sites.
Interaction between provenances and environments may cause heterogene-
ity of the geographic variation patterns estimated in different locations. To
expect regressions describing clines also in land races would mean overesti-
mating the speed of response to selection. A geographic pattern fitting into one
of the types is not always found.
Differentiation patterns among provenances of E. camaldulensis in econom-
ically relevant traits are described in Example 11.1. The use of this species in
plantation forestry has been greatly promoted by results of provenance trials.
11.6 Choice of Provenances in Tropical Forestry 165

Example 11.2 describes a series of internationally co-ordinated provenance

trials in teak.
Pronounced geographic variation has been found in some other eucalypt
species (Eldridge et al. 1994). Notable differentiation of provenances was
also found in Casuarina equisetifolia (Pinyopusarerk and Williams 2000).
Superiority of land races of Casuarina equisetifolia was found in at least some
traits related to growth or growth form. However, Chanshama et al. (1999)
reported that land races of E. tereticornis ranked low in provenance tests.
Kundu and Tigerstedt (1998) found notable physiological differentiation of
ten provenances of neem (Azadirachta indica). Laboratory tests revealed three
major groups related to annual rainfall at the origin. Offspring of exotic plan-
tations were among the experimental entries with medium rainfall. The pat-
tern of differentiation among four provenances of neem did not correspond
with that of the allelic structures at six allozyme loci (Kundu 1999).

Choice of Provenances in Tropical Forestry

If large-scale plantations of a species are planned in a region, recommenda-

tions are best based on the results of a series of experiments. The establishment
of experimental series admittedly takes time, effort, and money. The long-term
observation of field trials is nonetheless indispensable for making reliable deci-
sions on provenance choice.
The results of provenance tests can be implemented in different ways. Mass
collection of seed for plantations in the field test forbids itself because of too
few parent trees and their lack of isolation. However, Ladrach (1998) described
the conversion of a provenance test into a seed production area (see also Sect.
13.5) as a rapid procedure of using the test results. Given that the experimen-
tal trees were planted at a spacing of 3 m 3 m in six-tree row plots, 30 evalu-
ation trees were left unless mortality exceeded 15% at the end of the
observation period of 3 years. At this age the best trees are marked as seed trees.
If half of the provenances are rejected, about 100 trees are expected to be left at
a fairly regular spacing. Presumably the bulk of the seed produced there arises
from matings between trees of different provenances. Provenance hybrids are
likely to occur if the flowering periods of predominantly outcrossing species
overlap. The proportion of hybrids thus depends on the flowering phenology
and other parameters of the mating system. Phenotypic trait expressions of
provenance hybrids are usually unknown. Such a seed production area was
considered to be the beginning of a land race.
Another way is collection of more seed in those stands where the material
for the provenance tests had been collected. However, according to common
experience the parent stands often no longer exist when the provenance test is
166 C HAPTER 11 Provenance Research

evaluated. This is particularly true with provenances of exotic species. Once a

provenance has been tested in the field, the material under test may be inte-
grated into a base population for breeding.
The choice of the right provenance is of particular relevance in exotic
species. In this case reproductive material is transferred into possibly different
environmental conditions. However, the choice of provenances should always
be considered carefully at the planning stage.
Several criteria for the selection of provenances are used. They differ accord-
ing to the status of knowledge on patterns of genetic variation and differentia-
tion. If no information on geographic patterns is available from field tests, seed
from autochthonous populations growing in a similar environment close to
the planned growth site is recommended. These populations are likely to be
adapted to the prevailing environmental conditions. Their recommendation is
much more reliable than that of collecting seed just in any, possibly allochtho-
nous, stands in the vicinity.
The choice of reproductive material is often done on the basis of ecological
similarity of the seed collection site and the prospective planting site alone.
It may be more reliable to compare the prospective planting site with the
environment at the origin of the respective population. The environment at the
origin of a population is at least as relevant for its behavior at new sites than
only the phenotype of the autochthonous collection stands. To many forest
managers provenance has become the overriding concern in the establish-
ment of plantations. Unfortunately, many foresters do not distinguish between
the growing site and the origin of populations. They are in fact concerned only
about the site of seed collection. However, knowing where seed was collected
without making reference to the origin of the respective population does not
make a better plantation. Unfortunately, the origin of populations of tropical
trees is only rarely documented properly.
Ladrach (1998) reported some general guidelines for provenance choice
derived by Zobel et al. (1987). They might apply to seed collections in
autochthonous populations. Collection at high elevations for plantations at
low elevations is advised against. Also a shift of reproductive material from
maritime origins to more inland regions should be avoided. Reproductive
material from seed sources in regions with uniform climate should not be
transferred to regions with strong fluctuations. Also, seed of populations on
basic soil should not be chosen for planting on acid soils. Likewise, the trans-
fer from sandy soils to clay soils is not recommended. When reporting on expe-
rience with plantations of various exotic pine species, Marsh (1970) came to
the conclusion that the origin and the growing site should match as closely as
The study of environmental matching in terms of the temperature regime,
the amount and the seasonal distribution of rain fall, etc. (Davidson 1995)
11.6 Choice of Provenances in Tropical Forestry 167

might be completed by considering also edaphic and biotic conditions (Zobel

et al. 1987). Particularly in exotic species the suitability of a plantation for seed
collection cannot be assessed on the basis of a few trees of unknown origin and
unknown history (Jones and Burley 1973; Hattemer 1987).
When discussing the adaptedness of populations of E. globulus in various
parts of the world, Eldridge et al. (1994, p. 31 f.) raised the question, whether
one should be satisfied with a land race as a current source of seed and as the
genetic base for future selection. Mentioning casual introductions of seed to
South Africa as early as before 1860, the authors doubt that the land race is bet-
ter than newly introduced seed of the best provenances. Comparisons between
the offspring of land races and their natural progenitor populations in
designed field trials are needed for an objective judgment. In view of the situ-
ation in eucalypts, Eldridge et al. (1994, p. 2) also state that the collection of
seed from natural populations has led to considerably more improvement than
the uncritical use of land races. Seed collection in small plantations for the
establishment of other small plantations opens a sequence of populations with
decreasing genetic variation without the option of improving yield.
The choice of a certain provenance is often determined by the availability of
sufficient quantities of reproductive material in tropical plantation forestry;
however, there is no reason to believe that incidentally available material is also
the most suitable for the prospective growing sites.
Clines existing in economically relevant traits greatly facilitate the selection
of provenances. On the basis of the studies by Langlet (1936) on clinal varia-
tion, Eriksson et al. (1980) and Mtys and Yeatman (1992) have even formu-
lated transfer rules of provenances on the basis of the ecological difference
between the origin and the planting site. Westfall (1992) developed them
further to the construction of seed transfer zones. Two methods were discussed
by Raymond and Lindgren (1990).
In some tropical countries the establishment of field trials has become an
integral part of more comprehensive programs for plantation forestry. Most
plantation species in tropical countries have been exotics (Chap. 13). The
majority of eucalypt and pine plantations as well as acacia plantations in the
humid and semihumid tropics are outside the natural range of the species.
T. grandis, Swietenia spp., Leucaena spp., and other agroforestry and multipur-
pose tree species are planted on all continents. Only recently the awareness of
the multiple uses of indigenous species has grown (Leakey and Newton 1994a;
Margraf and Milan 2004).
Unfortunately, it is often difficult to obtain sufficient quantities of seed from
autochthonous or even natural stands of a species. Seed centers responsible for
the collection, storage, and deployment of tree seed do not exist in all coun-
tries. The situation is different for Australian tree species. Small quantities of
Eucalyptus and Acacia seed for field testing provenances can be procured
168 C HAPTER 11 Provenance Research

through the Australian Tree Seed Centre (Arnold and Midgley 1995). The
availability of reproductive material for both field-testing and large-scale plant-
ing has significantly contributed to the wide use of Australian tree species as
exotics throughout the tropics. Evans and Turnbull (2004 p. 113 ff.) discussed
some examples of successful species introductions.
Programs for afforestation and plantation forestry have often been pro-
moted by international organizations and/or bilateral development agencies
that also supported the establishment of field trials (Barnes and Burley 1990;
Palmberg and Esquinas-Alcazar 1990; Arnold and Midgley 1995; Ladrach
1998). The respective provenance trials and other long-term experiments were
often initiated with considerable effort and input (Examples 11.1, 11.2).
However, the long-term maintenance of the experiments and their adequate
analysis sometimes became impossible, when the respective projects expired.
However, meaningful outcomes take their time. Provenance research in tropi-
cal forest tree species is more complex than in trees of the temperate zone even
though rotation and observation periods are shorter.
Conifer plantations have not achieved the important role once forecasted by
the FAO and other international agencies during the 1960s. Thus, the consid-
erable amount of results available on provenance differentiation of pines is of
limited use. However, for certain species of pines and some eucalypts as well as
a few valuable hardwood species such as teak, results of long-term observation
of provenance trials are available.
Provenance research is closely associated with plantation forestry and con-
centrates on important plantation species and a few others with potential
importance. Practice-oriented provenance research in a tree species should
therefore be made only if it is of prospective use as a plantation species. New
provenance trials have recently been established for agroforestry species and
multipurpose tree species. There exists a strong argument for exploring prove-
nance variation in regular tests before major plantation projects are taken up.
If provenances of a species are differentiated in spite of inferred widespread
gene flow, immigrants must be part of the genetic load. Provenances that are
not adapted to the planting sites do not only imply the risk of failures. If they
are able to reproduce, the respective populations increase the genetic load of
the surrounding autochthonous populations through their emitted effective
pollen. Therefore, the large-scale transfer of arbitrary, possibly maladapted,
reproductive material adversely interferes with ecological genetics principles.
Field trials with several temperate conifers yielded the information that the
local autochthonous source is not always the fastest-growing or otherwise
best provenance (Namkoong 1969). For example, local sources of Pinus taeda
and Pinus palustris were frequently inferior to provenances transferred over
large distances (Wells and Wakeley 1966, 1970). A frequent experiencee in
reciprocal transplant experiments is that the local seed source is always among
11.6 Choice of Provenances in Tropical Forestry 169

the top-ranking but is not necessarily the best-adapted nor the most produc-
tive. There may exist obvious adaptation lags. However, limits of seed transfer
may be derived from mortality data of transferred provenances (Mtys and
Yeatman 1992).

Example 11.1: Provenances of Eucalyptus camaldulensis and Their Growth

The natural distribution of E. camaldulensis is confined to Australia but the
species covers the largest area of all eucalypts on the continent (Fig. 11.2). The
species is the most important plantation species for dry and semidry climates.
At least 500,000 ha of forest of this species have alone been established in the
Mediterranean region, mostly in Spain and Morocco. The species is also
increasingly used in semiarid regions of the tropics with summer rain.
E. camaldulensis grows naturally in proximity to rivers and lakes. The species
is closely related to E. tereticornis, E. brassiana, and E. exserta. Natural hybrids
between these species and some other eucalypts of the subgenus Symphyomyrtus
occur. Morphological and physiological differences between northern and
southern races of E. camaldulensis are obvious. The two races are even regarded
as subspecies by some authors.
Genetically controlled variation among provenances of E. camaldulensis has
been described for the following traits: growth, wood properties, salt tolerance,


Lake Albacutya

Fig. 11.2. Natural distribution of Eucalyptus camaldulensis and approximate locations

of two better-known provenances. (Adapted from Eldridge et al. 1994)
170 C HAPTER 11 Provenance Research

drought tolerance, frost tolerance, lignotubers and coppicing, and concentra-

tion of leaf oils (p. 60 ff. in Eldridge et al. 1994).
Provenance trials established in areas with predominant rainfall in summer
proved the superiority of northern provenances. The observed variation is pre-
dominantly clinal; however, this general pattern is blurred by ecotypic varia-
tion on a smaller geographic scale. For example, the provenance Petford
performed well in almost all experimental locations with tropical summer rain.
The difference in volume production between the best-growing and the poor-
est-growing among 16 provenances tested in a trial in Nigeria was 300%
(Otegbeye 1985).
Provenance trials in a Mediterranean climate with winter rain showed a supe-
riority of southern provenances but also ecotypic variation was superimposed
on the general clinal pattern. Some provenances proved to be fast-growing in
most trials established in this type of climate. For example, provenance Lake
Albacutya is slow-growing in locations with tropical summer rain but performs
well in a Mediterranean climate. The difference in volume production among
provenances was up to 800% at an age of 10 years in a trial established in Israel.
The differences in the growth potential of provenances are striking. Prove-
nances from northern Australia, an area characterized by tropical summer
rains, perform well in a similar climate throughout the tropics. Provenances
from the southern winter rain region perform better in a Mediterranean cli-
mate. Particularly fast growing provenances have been identified for both
climatic regions.
The geographic variation in other traits is frequently ecotypic. Owing to the
focal interest in growth they have been studied with less intensity. The obser-
vation made in provenance trials that certain populations are particularly suit-
able for areas characterized by tropical summer rain and a long dry season
greatly contributed to the current importance of the species for afforestation in
dry and semidry climates of the tropics (From Pryor and Byrne 1969; Midgley
et al. 1989; p. 60 ff. in Eldridge et al. 1994).

Example 11.2: International Provenance Trials in Teak (Tectona grandis)

Teak is an important tree species in monsoon forests of South Asia and
Southeast Asia. Its natural distribution ranges from India to Myanmar,
Thailand, and Laos. Whether teak forests in Java are indigenous is doubtful.
The total area of natural forests dominated by teak is more than 30 106 ha.
Teak is also a valuable species for plantation forestry. Most plantations exist in
Asia, but teak has been successfully planted in all main regions of the tropics
(Tewari 1992). The artificial regeneration of teak forests has a particularly long
tradition on Java.
The wood of teak has excellent properties. It is suitable for furniture mak-
ing, interior decoration, and as a veneer and plywood. Thus, the value of teak
11.6 Choice of Provenances in Tropical Forestry 171

plantations depends not only on growth rates but also on the quality of the
stems produced.
Recognizing the importance of teak as a plantation species, the DANIDA
Forest Seed Centre organized a series of international provenance trials (Kjr
and Verapong 1995). Forty-eight provenance trials were established in
19731974. A total of 75 provenances and land races from India, Laos,
Thailand, Indonesia, Africa, and Latin America were tested. The provenances
were assigned to eight main provenance regions. In the single trials only sub-
sets of provenances were represented. Unfortunately, material from Myanmar
was not available for the trials.
Only 21 out of the 48 once-planted trials were included in a first interna-
tionally co-ordinated evaluation at an age of 79 years. A second evaluation
was made at the age of 17 years (Kjr and Verapong 1995). By that time the
number of trials evaluated had decreased to only 8.
The main planting region for teak is tropical Asia. However, only one trial
in Asia, located at Pah Nok Kau in northeastern Thailand, was evaluated at the
age of 17 years. This trial comprised 25 provenances from seven provenance
regions. No significant differences among provenance regions were observed
in survival and growth (measured as basal area) at the age of 9 and 17 years.
Small but significant differences were observed in stem form and stem quality
at age 17. The stem form of provenances from Thailand is slightly superior to
that from most other provenances. Provenances from Indonesia showed
inferior stem form.
The results of African and Latin American provenance trials indicated
preferable provenance regions. However, pertinent recommendations must be
specified for the observed traits. Since fast-growing provenances may have
poor stem form, it is impossible to generally recommend certain provenances
on the basis of these international trials. No particular provenances or prove-
nance regions seem to be superior to others in the main planting regions.
The results of the series of internationally co-ordinated provenance trials on
teak are of limited value for immediate application for the following reasons:

Provenances from the large, continuous natural distribution area in

Myanmar were not available for the trials. Approximately half of the
30 106 ha of natural forests dominated by teak were found in Myanmar
(Tewari 1992). Even though this area is declining owing to ongoing defor-
estation, Myanmar is frequently regarded as a diversity center for teak, and
teak from Myanmar is often believed to be superior to that from other
sources (Gyi 1993).
Only eight trials out of the originally planted 48 experiments were evalu-
ated at the age of 17 years, i.e., less than one third of the regular rotation
period of teak plantations. At this stage the other trials either did not exist
172 C HAPTER 11 Provenance Research

anymore or could not be evaluated for some reason. The remaining trials
that could be evaluated are distributed unevenly. Three trials are located on
Puerto Rico, and only one represents the main growing region of teak in
Southeast Asia. No trials were evaluated at the age of 17 years in southern
Asia (India).
In many traits no significant differences among provenances or prove-
nance regions were found. Statistically significant patterns of variation
seem to be ecotypic. Neighboring provenances often showed contrasting
trait expressions.
The recommendation of superior provenances or provenance regions is
complicated by the contrasting variation patterns found in economically
relevant traits. Provenances showing good survival and growth are often
inferior in quality traits such as stem quality (From Kjr et al. 1995).

Recommended Literature

Stern and Roche (1974) analyzed the genetic variation in tree populations and
their genetic systems mainly in the context of their adaptation to the environ-
ment. Also the genetic implications of the influences of humans on tropical,
temperate, and boreal forest ecosystems have found the interest of those
authors. The book by Wright (1976) on forest genetics covers all aspects of this
field as seen 30 years ago, with emphasis on breeding. Methods and results of
provenance research are presented in three of the 19 chapters. The book by
Eldridge et al. (1994) deals with provenance research, breeding, propagation,
and resource conservation in eucalypt species. Morgenstern (1996) has pro-
duced the first book to concentrate on geographic variation of tree populations
in phenotypic traits. The results for the amount of variation and variation pat-
terns are implemented in silviculture and breeding in the major climatic
regions of the world, although the emphasis is on temperate and boreal species.
Williams et al. (2002) concentrated on the design, conduct, and analysis,
including data processing, of experiments in provenance research and breed-
ing. Much other literature on this subject has been published in languages
other than English.
Domestication and Breeding of Tropical 12
Forest Trees


Genetic variation is not only expressed as differentiation among populations,

but is also observed within populations. The objective of breeding is to exploit
genetic differences between individuals in populations in order to change the
expressions of economically relevant traits for increasing plantation yield. The
basic approach to conventional forest tree breeding is artificial selection of
individuals and progeny families. These are subsequently propagated sexually
or vegetatively by the breeder.
Breeding of tropical forest tree species has been and will be confined to a
comparatively small number of taxa that are suitable for plantations. Main
species and genera are those for which large-scale provenance trials have been
established (Chap. 11). Provenance testing as a lower-intensity activity yields
the basic material that close-sense breeding projects can build upon. Since even
provenances of good performance contain trees of low value for use, only
further breeding is likely to achieve the quantum leap in the quality of plan-
tations (Mesn et al. 1994). The most comprehensive breeding programs have
so far been conducted on tropical eucalypts and pines.
Most of the traits to be improved by breeding are influenced by both genetic
and environmental factors. Many of these traits that were mentioned previ-
ously (Sect. 11.3.2) are metric, i.e., they can be measured on a continuous scale.
This holds specifically for traits influencing quantity and quality of yield.
The conventional concentration of research and training in tropical forest
genetics on tree improvement has led to some successful breeding projects;
however, it has also contributed to failures and diverted interest from essential
characteristics of tropical forest trees. Costly and time-consuming breeding
projects must be grounded on previous studies of key aspects of the genetic
system of the respective species or must be accompanied by such studies. An
approach aimed at a short-term increase of plantation yield harbors consider-
able risks if it is not accompanied by population genetics studies of the genetic
174 C HAPTER 12 Domestication and Breeding of Tropical Forest Trees

system of the target species. Breeding of forest tree species is still an important
application of forest genetics research.
The decision to embark on a breeding project can best be substantiated if
adequately sized, well-adapted, and productive populations with known his-
tory are available as base populations. A breeding goal must then be defined.
Starting selection right away in a population that happens to be locally avail-
able is not necessarily promising and hardly justifies the substantial future
investment. Other preconditions are trained personnel, sufficient land, and
continued financing.
Species of agricultural crop plants have long been subjected to breeding.
Inadvertent selection practiced by humans has been an integral part of their
evolution. Today, more deliberate and refined methods are in use. The main
principles of breeding were successfully used long before the molecular basis of
inheritance was discovered and even before genetics became a science. It is
likely that humans have influenced the genetic structures of forest tree popula-
tions for several thousand years. However, systematic breeding of forest trees is
a much more recent development. European and North American conifers and
poplars were the first forest trees subjected to intensive breeding in the twenti-
eth century. Methods involving gene transfer and other biotechnology cannot
be addressed in this introduction.


Publications on the domestication of forest trees mainly deal with the genetic
consequences rather than the process itself (El-Kassaby 2000). Harvesting seeds
from the wild, natural state, germinating them, producing planting stock, and
planting stands is just the first step. Further and more crucial steps follow in
In the introduction to his comprehensive account of domestication Darwin
(1875) wrote, From a remote period in all parts of the world, man has sub-
jected many animals and plants to domestication or culture. In the past,
foresters have taken seeds from the wild and used them without selection,
thereby practicing inadvertent sampling of tree populations. Libby (1973) con-
sidered as essential elements of domestication the selection of desired trees in
variable populations and their mainly vegetative propagation. Forest trees have
been subjected to a series of selection stages as the essential phases of the
domestication process. These involve an increase in frequency of desirable
physical attributes with economic relevance such as those listed in Sect. 11.4.2.
Certain procedures of seed collection, seed extraction, and storage and, last but
not least, the methods of planting stock production are linked to breeding.
12.2 Domestication 175

These procedures are necessary elements of bringing trees under the control of
humans. Allard (1960, p. 28) defines domestication as the bringing of a wild
species under the management of humans. Sooner or later the natural repro-
duction system is brought under control and converted into something highly
Darwin (1875) did not mention changes in the reproductive system in our
modern sense. He emphasized that the principle of selection is very important:
No doubt man selects varying individuals, sows their seeds, and again selects
their varying offspring. But the initial variation on which man works, and with-
out which he can do nothing, is caused by slight changes in the conditions of
life, which often must have occurred under nature. Man, therefore, may be said
to have been trying an experiment on a gigantic scale; and it is an experiment
which nature during the long lapse of time has incessantly tried. Hence it fol-
lows that the principles of domestication are important for us. and continues:
Domestic races of animals and cultivated races of plants often exhibit an
abnormal character, as compared with natural species; for they have been mod-
ified not for their own benefit, but for that of man.
One may doubt that domestication of forest trees will ever progress to a
stage where they totally rely on humans and would become extinct if left alone,
although some authors consider the inability to survive in natural ecosystems
as part of the definition of full domestication (Clement and Villachica 1994).
These authors proposed also a nomenclature of domestication. Only semido-
mesticates, although they are significantly modified from the wild state, can
survive if abandoned. The lowest degree of domestication is that of species that
are only managed. Darwin (1875, Chap. 23) emphasized, Length of time is all
important; for as each character, in order to become strongly pronounced, has
to be augmented by the selection of successive variations of the same kind, this
can be affected only during a long series of generations. This judgment is sup-
ported when considering the long way from species-elimination trials to prove-
nance trials of the promising species, breeding experiments, and the
development of methods for propagation and mass cultivation.
The natural system of reproduction of populations is replaced step by step
by artificial matings following certain designs (Sect. 12.5) and the propagation
of the individuals selected. De Vecchi Pellati (1970) made the important point
that in conjunction with artificial selection just this change is an essential ele-
ment involved in domestication. She also stated that the new populations are
normally propagated in isolation, while native populations receive easily new
contributions from different original populations.
Hence, three crucial elements in domestication and breeding to increase fit-
ness for purpose or domestic fitness (p. 94 f. in Eriksson and Ekberg 2001) are
sampling, selection, and the change of the reproduction system. The domesti-
cation process has a strong social involvement. For this reason, Evans (1980)
176 C HAPTER 12 Domestication and Breeding of Tropical Forest Trees

defined it as the improvement of the harvest index, the proportion of humanly

useful material of plants.

Genetic Controlledness of Phenotypic Traits

The success of artificial selection for a phenotypic trait critically depends on its
genetic control. This is an absolute biological precondition for starting a breed-
ing project and one may only then predict some response of the population to
artificial selection. If trait expression proved to be different among provenances
of a species in a provenance experiment, then we may safely conclude that this
trait is under genetic control. If the test material has received equal treatment,
we cannot explain the observed phenotypic differentiation other than by dif-
ferentiation at the controlling gene loci. Planned experiments are the standard
procedure for inferring genetic controlledness. The genetic controlledness of
adaptive traits is also crucial in the context of genetic conservation (Chap. 14).
The statement of genetic controlledness alone is reliable only if the experi-
ment was based not only on an environmental design (Sect. 11.4) but also on
a genetic design (Sect. 12.5). Since an experiment takes effort and time, we have
a problem in testing the genetic controlledness of traits that are expressed only
late in the ontogeny of long-lived forest trees. Hattemer and Ziehe (1997) have
presented a review of pertinent procedures.

Individuals Related by Descent

Similarity between parents and their progeny has been extensively exploited in
the theory and practice of breeding. The enormous breeding progress made by
early humans was hardly based on comparison tests with a sophisticated design
but rather on the experience that a plant or animal with desirable properties
gave rise to progeny with the same or even better value for use. The phenotypic
similarity between parents and progeny must be based on the transmittance of
genes and implies genetic control; however, only little can be said about the
mode of inheritance.
Regression of offspring on a parent can be tested in seed trees and their prog-
eny from natural regeneration in the forest. Any observed similarity in trait
expression is then primarily due to the contributions of the female gametes to
the progeny. The seed parents and their progeny are compared in the same local
environment where reproduction took place. However, even if the peculiarities
of seed chorology admit the attribution of young trees to only one seed parent,
it is essential to check the descent of the progeny by genetic markers.
12.3 Genetic Controlledness of Phenotypic Traits 177

Pair Comparisons in the Field

Whether a phenotypic trait observed within a stand has a substantial genetic

component may not be directly testable by parentoffspring regression. If geno-
types and environmental conditions are associated, the variation of phenotypic
trait expressions may be controlled uniquely by environmental conditions, and
genetic patterns could also be correlated with phenotypic patterns. Hence, a pre-
condition for reliable conclusions is stochastic independence between genotypes
and environmental situations. This is safely fulfilled in artificial stands but it is
not necessarily so in stands that originated by natural regeneration. Most artifi-
cial stands are also even-aged. This is relevant if age modifies trait expression.
Gregorius (1989) devised a method to find out whether one or several phe-
notypic traits are under genetic control. For purposes of comparison, pairs of
trees are chosen in the forest depending on the expression of a phenotypic trait
(Fig. 12.1). When a tree with phenotype X is found, its nearest neighbor with
phenotype Y is chosen. This sampling procedure is supposed to guarantee local
stochastic independence between genotypes and environmental situations.
Herewith, it is essential that the partners could have emerged from the same

or not
infested comparison
Fig. 12.1. The method of pair-
wise sampling. In this example, of genetic
stress symptoms are tested for sensitive
genetic controlledness. (From or
Ziehe and Gregorius 1996) infested
178 C HAPTER 12 Domestication and Breeding of Tropical Forest Trees

seed pool. Hence, the contrasting phenotypes should be located within seed
transport distances from a single potential seed parent.
After genotyping many pairs of trees at a set of marker gene loci, a statisti-
cal test of homogeneity between the genetic structures of phenotypes X and Y
is undertaken. If the null hypothesis is rejected, phenotypes X and Y are asso-
ciated with different genotypes and the trait must be under genetic control;
however, gene loci with significant differences are not necessarily involved in
controlling this trait they may merely be stochastically associated with con-
trolling gene loci. Such stochastic associations can originate under a system of
mixed self-fertilization and random mating, which results in a correlation of
heterozygosity over the gene loci (Ziehe 2006). The difference in genetic struc-
ture between the phenotypic groups is quantified by genetic distances. Gene
loci that are stochastically associated with controlling gene loci are more likely
to show substantial distances if they carry a larger amount of variation.
If carriers of contrasting phenotypes tend to show similar frequency devia-
tions in different populations, we may call the genetic differences consistent.
Then either stochastic associations with the controlling gene loci are of a sim-
ilar type, or the marker gene locus surveyed is involved in the genetic control
of the trait. In predominantly outcrossing species, an increasing number of
populations with different genetic backgrounds can substantially strengthen
this evidence.
The extension of pair comparisons to traits with more than two expressions
is straightforward. Several slight modifications of the method were discussed
by Hattemer et al. (1993, Sect. 16.4). The genetic control of traits in trees at any
age can efficiently be tested in the forest environment even after the stand has
been thinned repeatedly. Since the method may also be applied to nonrepro-
ducing trees, it has essential practical advantage.
The method has been applied to European beech (Fagus sylvatica L.) but did
not support the hypothesis that certain stem forms of this species are under
genetic control (Ziehe and Hattemer 2002). In various instances, as reviewed
by Geburek (2000), the method helped to present evidence that air pollution
induced selection in European forest trees. In stress-related traits also degrees
of heterozygosity in the tolerant and sensitive groups were compared.
This method is straightforward and simple to apply but does not yield any
information on the rigidity of the genetic control. Ritland (1996b) has devised
a different approach to inferring genetic control of phenotypic traits. The
method is based on phenotypic similarity of pairs of (not necessarily neigh-
boring) individuals measured in the field. The phenotypic similarities of tree
pairs are regressed on the degrees of their pairwise relationship as estimated
from the fraction of alleles shared by the two individuals at a set of marker gene
loci. This yields a quantitative measure of genetic control (narrow-sense heri-
tability; Sect. 12.4).
12.4 Linear Model of Genetic Effects on a Phenotypic Trait 179

Response to Natural Viability Selection

A trait that possesses special adaptive relevance and, therefore, responds to

natural selection is early survival. If early mortality affects genetic structures
at marker loci, its variation can also be attributed to the control by either
these marker gene loci or stochastically associated gene loci. However, if the
response of several populations to similar selection regimes is consistent,
then the observed marker gene locus may be considered to be among the
gene loci that control the trait under selection. Since the trait considered is
viability, the controlling gene loci must then be considered to be adaptive
(Ziehe et al. 1999).

Linear Model of Genetic Effects on a Phenotypic Trait
Phenotypic Trait Expressions

Both genotype and environment contribute to phenotypic trait expressions (Sect.

2.6), and a simple linear additive model postulates the following interrelation
P = G + E, (12.1)
where P denotes the phenotypic and G the genotypic value of an individual,
and E is an environmental deviation. The supposed additivity of genotypic and
environmental effects on the expression of a phenotypic trait is intuitively
appealing and has proven its heuristic value in many empirical studies; how-
ever, there is no obvious biological justification for the presumed additivity.
Multiplicativity or other, more complex interrelations may also be used to
explain phenotypic trait expressions apart from simple additivity. The envi-
ronment does not only modify the expression of phenotypic traits. It also
determines the fitness of genotypes (Gregorius and Kleinschmit 2001).
The phenotypic trait expressions of different genotypes may respond to dif-
ferent environmental conditions in dissimilar ways. This observation has been
made in many experimental series with genetically defined entries in different
environments. Consequently, the simple additive model is expanded by an
interaction term IG E in order to account for this fact:
P = G + E + IG E. (12.2)
However, the separation of genotypic and environmental effects is not pos-
sible by introducing an interaction term. The supposed interactions between
180 C HAPTER 12 Domestication and Breeding of Tropical Forest Trees

genotypic and environmental effects are now no longer independent variables

and cannot be measured separately. The study of reaction norms of genotypes
helps to solve part of this dilemma (Gregorius and Namkoong 1986). It was
also addressed by Hattemer et al. (1993, Sect. 16.2).
The production of sexual offspring is required in most breeding strategies.
Thus, the impact of a genotype on the phenotypic trait expression of the indi-
vidual itself is of less importance for the trait expression of its progeny.
However, genotypes are not directly inherited. Only gametes are transmitted
from parents to offspring following segregation and recombination.
Interdependencies of the expression of the alleles at a single gene locus are the
reason for the occurrence of dominance deviation D. In a similar way, interac-
tions between gene loci are responsible for phenotypic effects on certain traits
that cannot simply be explained by additivity of the effects of several control-
ling loci. These nonadditive effects of gene loci are called epistatic. Epistatic
effects are responsible for an interaction deviation I that just like D influence
the genotypic value of an individual. Neither D nor I can be transmitted to the
These considerations make it necessary to expand the linear model by
splitting up the genotypic value into three components, i.e., the additive
effect or breeding value A, dominance deviation D, and interaction term I, in
order to account for epistatic effects of gene loci controlling the phenotype
such as
G = A + D + I. (12.3)
In Eq. 12.3 the quantities D and I depend also on gene frequencies; there-
fore, they are not only measures of dominance and epistasis but are also prop-
erties of the population (pp. 116, 119 in Falconer and MacKay 1996).
This conclusion is of crucial importance in quantitative genetics. The genetic
component to phenotypic trait expressions is determined by genotypes at a
number of gene loci. In diploid organisms each genotype is composed of two
alleles. Figure 12.2 illustrates the phenotype of a metric trait as controlled by a
single gene locus in the absence of environmental or epistatic effects. The aver-
age trait expression of the heterozygous genotype is not necessarily exactly half
way between the average trait expressions of the two different homozygotes.
d measures the deviation of the average trait expression of the heterozygote
from the intermediate value of the average trait expressions of the two homozy-
gotes. Intermediarity is characterized by d = 0. Positive or negative dominance
deviations are defined as partial dominance (d < a) or overdominance (d > a).
The dominance is complete if the heterozygote possesses the same genotypic
value as A1A1 and d = a, for instance.
From Fig. 12.2 we may derive some simple conclusions for the increasing
complexity of artificial selection for greater trait expression with increasing d:
12.4 Linear Model of Genetic Effects on a Phenotypic Trait 181




A1 A1 A 1 A2 A2 A2 Genotypes

Fig. 12.2. Partitioning of the genotypic value. m is considered the average phenotypic
value of the trait produced by the homozygous genotypes in the absence of environ-
mental effects. Phenotypic values can therefore be called genotypic values. a is the effect
of the genotype possessing the advantageous allele A1 in homozygous condition.
Consequently, a measures the effect of the genotype possessing only A2. d is the domi-
nance deviation. Note that a and d are defined relative to m. (Adapted from Becker 1993)

Dominance, partial (0 < d < a) or absent (d = 0). If the breeder selects indi-
viduals with the greatest trait expression and excludes all others from
reproduction, the population will contain only allele A1 in the subsequent
generation. Owing to complete selection against A2 the advantageous allele
is fixed after one generation. The breeding goal has been reached.
Complete dominance A1 > A2 or d = a. In this situation the selected part of
the population contains genotypes A1A1 and A1A2. In order to find out
which individuals with the greatest trait expression have the desirable
genotype A1A1 and which are only A1A2, the breeder has to cross them with
individuals possessing the smaller trait expression and to test their progeny.
This involves a certain effort and also a serious setback, since all progenies
produced contain the undesirable allele A2.
Overdominance or d > a. Reproduction of the selected part of the popula-
tion will consist of heterozygotes only and will contain A1 and A2 with
equal frequency. The subsequent generations will contain also homozy-
gotes with nonoptimal phenotypes. The balanced selection equilibrium
prevents further breeding progress. A possible solution is to vegetatively
propagate individuals with the greatest trait expression.

It is also easy to see in Fig. 12.2 that any environmental effects would blur the
phenotypic differences between the three genotypes, so also individuals with
genotype A2A2 may be selected, although they possess an undesirable pheno-
type. This can only reduce breeding progress.
182 C HAPTER 12 Domestication and Breeding of Tropical Forest Trees

Genetic Variance Components and Heritability

For the following considerations we admit environmental effects but do not

account for the existence of genotype environment interaction. If we modify
Eq. 12.1 by partitioning the genotypic value G into its components (Eq. 12.3),
we get
P = A + D + I + E. (12.4)
If we now consider random individuals in the population, the quantities in
Eq. 12.1 become random variables. Assuming their mutual independence, we
may derive an important property of their variances,
VP = VG + VE, (12.5)
and, according to Eq. 12.4,
VP = VA + V D + V I + V E. (12.6)
VA is the component of the genetic variation that can be captured also under
sexual reproduction and its concomitant recombination. VD measures the vari-
ation due to the effect of alleles masking others that have no apparent effect. It
is possible to estimate these variation components from appropriate experi-
ments following an adequate statistical design involving related individuals
with known pedigree. These may have both parents in common and are then
full-sibs, or may have only one parent in common and are then half-sibs (Sect.
12.5). The phenotypic similarity among relatives is the basis for estimating the
rigidity of genetic control. In fact, the proportion of the phenotypic variation
in the population attributable to genetic differences among the individuals or
VG/VP is the heritability in a broad sense or hw2 . As can be seen from Eq. 12.6, it
is never smaller than the heritability in a narrow sense or h2 = VA/VP (p. 123 in
Falconer and MacKay 1996) that measures the degree of genetic control by
additive effects only. The term degree of genetic control must be understood
in the sense of statistics rather than causation.
The heritability in the narrow sense is of main interest for breeders, since it is
involved in quantifying the expected change of average trait expression by mass
selection. This heritability quantifies the relative amount of useful variation
that may be exploited by breeding, because only the fraction of the genotypic
values due to additive effects is inherited. Heritabilities range from zero to unity.
A high heritability value h2 implies rapid improvement of trait expressions by
breeding methods based on sexually produced offspring. h2 = 0 means that no
genetic variation exists which can be used in breeding, or that breeding is not
possible for the observed trait. The maximum value of unity would mean that
12.5 Estimation of Breeding Parameters; Progeny Testing 183

the observed phenotypic variation is completely determined by additive genetic

variation. In consequence, breeding is expected to be highly efficient. The effi-
ciency of breeding methods and the breeding progress or genetic gain to be
expected from artificial selection for a trait increase with increasing values of
phenotypic variation, narrow-sense heritability, and the intensity of selection.
Although the analysis of variance requires some assumptions to be fulfilled,
it is a widely used standard method for the interpretation of experimental
results in the breeding of animals and plants in spite of these restrictions. It is
also in frequent use in forest tree breeding and has proven its practical value in
forest trees just as for other plants.
The results of an experiment strongly depend on the environmental condi-
tions at the site (Nienstaedt and Riemenschneider 1985), the experimental
design, and the sample of genotypes studied (p. 332 in Hattemer et al. 1993).
Estimators of trait heritability frequently exhibit also a temporal dynamics
(Nienstaedt and Riemenschneider 1985; Jayasekera et al. 1994).

Estimation of Breeding Parameters; Progeny Testing

Progeny tests have a dual function. They open the way to estimating genetic
components of the variation in economically relevant phenotypic traits.
Furthermore, it is possible to evaluate selected trees and to select them on the
basis of the performance of their offspring, or to select families and/or indi-
viduals within those families (Sect. 12.5.2). However, many related individuals
are produced by progeny testing. The degree of their relationship (usually full-
sibs or half-sibs) must be known for interpretations in terms of genetic vari-
ance components.
A variety of mating designs have been worked out and are being practiced.
They differ in efforts to be taken, and in the type and amount of information on
breeding parameters (van Buijtenen 1976; Otegbeye 1998). Otegbeye (1998)
included descriptions of the (environmental) design of the respective field tests.

Progenies of Open-Pollinated Trees

The simplest form of a progeny test is the comparison of progenies of single

seed trees after open pollination. This used to be the most frequently applied
procedure. If seed trees are assumed to have mated randomly, the allelic struc-
ture of their individual effective pollen clouds represents that of the population;
hence, they are not differentiated. The variation between the families is then due
to the variation in the breeding values of the seed trees.
184 C HAPTER 12 Domestication and Breeding of Tropical Forest Trees

In this basic experimental design the progeny means can be used to estimate
the breeding values of the seed trees, i.e., the means of their offspring under the
conditions of random mating. Analysis of variance of the trait observations
allows then estimation of the additive genetic variance and heritabilities.
Finally, the breeding progress expected from selection of the seed trees or that
expected from the selection of trees within the progenies can be estimated.
However, these ideal conditions are rarely realized.
Before sufficiently variable genetic markers were available, the mating sys-
tem of trees could not possibly be analyzed in detail; therefore, it was consid-
ered random for the sake of convenience. However, the pollen clouds differ,
because only moderate numbers of male mating partners become effective for
particular seed trees (Sect. 6.2.1; Fig. 6.1), which induces a drift effect. If also a
family structure exists in the population, the differentiation of effective pollen
clouds is enforced by differentiation of the neighborhoods (Sect. 11.5).
Under random mating all offspring of a family are ideally sired by different
pollen parents; hence, they share only their female parent and may be consid-
ered half-sibs that are expected to hold one quarter of their genes in common.
However, in many species the families might contain also offspring from self-
fertilization. Owing to restrictions in the effective number of male mating part-
ners, certain portions of the family members are in fact full-sibs, since they share
both of their parents. Hence, the progeny of a tree after open pollination is a
mixture with different but unknown degrees of relationship. This condition is
highly likely to be the rule for forest trees. Thus, the genetic relationship among
the offspring of single trees is closer, i.e., they share more than one quarter of
their genes (Squillace 1974). Consequently, the variation among the families is
inflated. If it is used for estimating VA, heritabilities are overestimated. The esti-
mates of VA and h2 are also not reliable. Selecting the trees on the basis of the
performance of their open-pollinated progeny would be only a rough approxi-
mation, since the offspring of a seed tree have a more or less unique group of
male parents. Dvorak and Wright (1994) took the complicated kinship relations
of open-pollinated progenies into account by assuming the average degree of
relatedness to be one third. Eldridge et al. (1994, p. 203) reported problems in
interpreting the results of testing open-pollinated progeny.

Progeny Tests After Controlled Pollination

In order to define sibship more precisely, trees must be control-pollinated fol-

lowing a mating design. For instance, in an experiment denoted as polycross,
seed trees are artificially pollinated by a pollen cocktail, i.e., a mixture of the
pollen of many trees. Since a great deal of the production of full-sibs is pre-
12.5 Estimation of Breeding Parameters; Progeny Testing 185

vented, the degree of relationship is then more homogeneous and closer to what
is expected from half-sib families. If pollen of the respective seed trees is not
contained in the otherwise identical pollen mix, self-fertilization is excluded. It
has turned out that the male mating partners contributing equal amounts to the
pollen mix still do not have equal proportions of offspring. The progenies aris-
ing from this experiment are, nonetheless, expected to yield unbiased estimates
of breeding values of the seed trees and the additive genetic variation.
Many other mating schemes have been developed (Otegbeye 1998). The
production of full-sib and half-sib families in the same experiment offers man-
ifold options for analysis and use of genetic variation.
In a factorial mating design (Fig. 12.3, top) a moderate number n of seed trees
are pollinated by a larger number m of other trees used as males. The objective is
to estimate the general combining abilities of those m trees, i.e., the average per-
formance of their offspring when mated with the three or four testers. Also
specific combining ability can be estimated. It reflects the ability of the trees to
transmit genotypic effects when crossed with specific other trees. The group of n
full-sib progenies of each tree are half-sibs among themselves, since they have just
their male parent in common. Consequently, the variation among all m half-sib
families yields an estimate of 0.25VA. The means of these groups reflecting their
general combining abilities are estimates of their breeding values in this experi-
mental population. Furthermore, the variation among the families having both
parents in common yields an estimate of 0.5VA + 0.25VD, so also the dominance
variance can be estimated separately and can be exploited by selection.

1 2 3 4 ... m
1 ...
2 ...

n ...

1 2 3 4 ... n
1 ...
2 ...
3 ...
Fig. 12.3. Complete mating schemes for 4 ...
progeny testing. Top: Factorial design com- .
prising m trees used as males and n trees
used as testers. Bottom: Half-diallel com-
prising n trees without selfs and reciprocal
crosses n 1
186 C HAPTER 12 Domestication and Breeding of Tropical Forest Trees

In the diallel cross all possible progenies are produced among a group of n
cosexual trees. In Fig. 12.3 (bottom) a variant of this design is shown (a half-
diallel) that includes neither selfed progenies nor the progenies resulting from
reciprocal crosses. If, for instance, trees 3 and 4 have to be crossed, either tree 3
is pollinated by tree 4 or tree 4 is pollinated by tree 3, or the cross is produced
in either direction and the seeds are pooled. Just as in the factorial design both
VA and VD as well as general combining abilities can be estimated.
Pollination in most zoophilous tree species requires much effort. Various
methods used in eucalypts have been described by Hodgson (1976b), van Wyk
(1977), and Eldridge et al. (1994, p. 205 f.). Species with different floral struc-
ture require very different methods (Sedgley et al. 1992; Oudraogo 1997).
Since work in tree crowns is an additional obstacle, pollinations are best
made in a clone bank or seed orchard. In view of the large number of proge-
nies to be produced [n m in the factorial design and n (n 1)/2 in the half-
diallel], there exist numerous incomplete designs requiring less effort (p. 234 ff.
in Namkoong 1981). Progeny tests on the basis of complete designs are hardly
applicable in just any tropical forest tree species. If not all crosses are success-
ful at the same time, missing families require advanced computational meth-
ods for evaluation. The same is true if mortality in the field experiments leads
to missing plot values or even family means. The single-pair mating design is
recommended for avoiding relationships between families (Otegbeye 1998).
The expected breeding progress is much faster if a large number of full-sib
families are tested as compared with the test of single tree progenies (Wright
1976, p. 176). Among tropical tree species, this approach has been confined to
the intensively bred genera Eucalyptus (Bouvet and Vigneron 1996) and Pinus
(Nikles 1995).

Methods of Selection
The decision for a method of selection is based on experimental evidence of its
breeding efficiency and on whether it fits into the given breeding strategy. For
instance, on the basis of the results of a progeny test, the genetic parameters are
estimated and used for exploring the breeding progress to be expected from
various selection regimes such as the selection among families or of single trees
within families. If the variation within families is much larger than that
between them, it is advised to retain more families and select with higher inten-
sity within families in a combined procedure. The variant maximizing expected
gain (Shelbourne 1992) is finally chosen as the most promising one.
For estimating the breeding progress expected from various selection regimes,
the heritabilities mentioned in Sect. 12.3 have to be adjusted to the particular
12.6 Methods of Selection 187

experimental design and the particular selection procedures. McKinley and van
Buijtenen (1998) have concisely reviewed the methods of selection in tree popu-
The closer the age at selection is to the rotation age, the more reliable is
selection in view of the desired target expression of the trait. However, the per-
formance at earlier ages, possibly at half of the rotation age, is often sufficiently
closely correlated to the performance at the rotation age. Periodic measure-
ments yield information on ongoing changes in the ranks. For breeding pro-
grams in several tree species of the northern temperate zone continued over
several generations, the selection age has been optimized.

Selection of Plus Trees

The selection of plus trees is a method of mass selection, i.e., the selection of
individuals in populations in the field for their subsequent propagation. The
selection intensity may be high.
Together with provenance testing, selection of trees in the wild (Ledig 1973)
represents the first step in many breeding projects. It differs much from the
selection in plantations. In natural forests the phenotype hardly indicates the
genotype of old trees closely (Sect. 10.2.2). Heritabilities are supposed to be low
owing to environmental heterogeneity (competition, etc.). Plantations consist
of more or less equally spaced individuals possessing the same age and prima-
rily belonging to the target tree species; therefore, the neighbors can be used as
comparison trees. In natural tropical forests, at best a regional average may be
used as a reference.
However, plus-tree selection is defined as a method based on the observa-
tion of the phenotypes of the selected trees themselves rather than those of
their progenies. The types of traits used for plus-tree selection are basically
those already mentioned in Sect. 12.4.2. Trees are selected on the basis of their
superiority in certain traits, for example, because of their good stem form
and/or volume growth. Practical guidelines for the identification of plus trees
have been summarized by Schmidt (1993a). The selection is often based on
indices which combine several traits characterizing the growth and quality of
trees (Zabala 1994; McKinley and van Buijtenen 1998).
Studies in European temperate conifers proved the efficiency of positive
plus-tree selection for increased height growth and even more so volume
growth and stem form (Cornelius 1994; Simpson 1998). Other investigations
on North American conifers detected only a small effect or failed to disclose any
effect of plus-tree selection (Wright 1976, p. 165). Eldridge et al. (1994) attrib-
ute the spectacular improvement of eucalypts to the selection of outstanding
188 C HAPTER 12 Domestication and Breeding of Tropical Forest Trees

individuals in plantations in combination with their large-scale vegetative

propagation (Sect. 12.8.3).
Selection of individual trees is a very simple procedure also practiced in
standard silvicultural management systems. The previously described selec-
tion of plus trees is positive since trees with superior phenotypes are
selected. Selection can also be negative, i.e., trees with unwanted phenotypes
are removed from a population first and the remaining trees form the repro-
ducing population. Thus, selective thinning favoring superior phenotypes by
removing inferior ones is a form of phenotypic selection (Wright 1976, p. 165)
with low intensity. It is the reverse of dysgenic selection as discussed in Sect.
10.2.2. The efficiency of this type of selection is likely to be low, i.e., the yield
of forest tree plantations is supposed to be only moderately increased and/or
improved. This holds in particular if most of the observed phenotypic varia-
tion is environmentally caused by marked heterogeneity of environmental
conditions. The observed variation of growth traits is also caused by a varia-
tion of the developmental stage, i.e., the different age of trees. An analysis
of stand structure, its history, and environmental variation gives hints on
the importance of genetic variation for an observed phenotypic variation
(Davidson 1995, p. 9).

Selection in Progeny Tests

In view of the generation interval of trees, selection of individuals according to

the performance of their offspring is a clumsy procedure. That the parents
cannot be selected before their offspring have been measured is a serious draw-
back. It is helped by proceeding with the propagation of the selected individu-
als and later removing those with poor progeny. Much more efficient is the
selection of families or of individuals within families, whereby the family mean
is used as a reference. Most efficient is the combination of these two.
Family selection is considered efficient also for traits with low heritability,
since unlike plus-tree selection the average performance of entire families is
evaluated. In view of the conservation of genetic information, a major disad-
vantage is the elimination of entire families or even all offspring of certain par-
ents. If this operation is repeated during subsequent generations, it leads to an
increase in inbreeding; therefore, it is operated with low intensity, so that only
the poorest families are eliminated. Moderate numbers of families contained in
a field test clearly restrict selection intensity. Selection of individual trees disre-
garding their family membership is the opposite of family selection. Inbreeding
increases at a slower rate but smaller breeding progress may be expected.
Therefore, individual trees are selected in conjunction with family selection.
A general disadvantage of progeny testing and the selection in these tests is the
existence of a large proportion of related individuals, as can be derived from
12.6 Methods of Selection 189

Fig. 12.3. Following evaluation, the field test may be converted into a seedling
seed orchard (Sect. 12.8.2).

Selection of Clones

Clonal propagules derived from an individual are not progeny in the close
sense. They can nevertheless be tested in field experiments. Selection of clones
in field tests still leads to single individuals but their performance can be eval-
uated much more efficiently than that of plus trees, even though only a small
number of ramets per clone are required. During clonal propagation the mul-
tilocus genotypes are conserved. Breeding progress is thus faster than in con-
ventional selection based on sexual progeny, because in the selection of clones
also all nonadditive genetic variation, i.e., dominance and epistatic effects
(D and I; Sect. 12.4), is used. The advantages of clonal selection become more
prominent with increasing significance of nonadditive genetic variation
(Burdon 1989). However, only one or a few selection steps are possible in a
given finite aggregate, since the number of individuals would get smaller at
every step. The selection of clones is part of the most successful breeding pro-
grams with eucalypts that also involve species and provenance trials, hybridiza-
tion among species, and progeny testing.

Marker-Assisted Selection

Certain types of highly variable molecular markers can be identified in large

numbers (Sect. 2.3.1). These markers are useful for the construction of genetic
maps and for localizing the chromosomal position of gene loci with a direct
impact on economically important phenotypic traits. These latter gene loci are
called quantitative trait loci, commonly abbreviated as QTLs. Particularly suit-
able for such studies are simple sequence repeats, restriction fragment length
polymorphisms, random amplified polymorphic DNA s (RAPDs), amplified
fragment length polymorphisms, and single nucleotide polymorphisms (p. 282
ff. in Becker 1993; Chap. 21 in Falconer and MacKay 1996).
Prior to the identification of QTLs, a linkage map with a large number of
variable gene loci is constructed. Genes located on different chromosomes are
inherited independently. The genes located on one chromosome form a linkage
group. They tend to be transmitted to the offspring jointly in the gametes.
However, parts of homologous chromosomes are occasionally exchanged at
meiosis owing to crossing over. The probability of crossing over between two
genes located on the same chromosome, i.e., their recombination rate, increases
with increasing distance between the genes. Thus, the probability of joint trans-
mission of two genes is higher if the genes are located close to each other.
190 C HAPTER 12 Domestication and Breeding of Tropical Forest Trees

The most common strategy is to identify a large number of highly variable

DNA markers in one or a few segregating full-sib families. The frequencies of
certain multilocus genotypes allow linkage groups to be identified. The num-
ber of these linkage groups is expected to equal the gametic number of the
chromosomes. Further studies would allow the linkage groups to be assigned
to certain chromosomes; however, this step is not required for QTL identifica-
tion. Within each linkage group the relative position of many marker gene loci
can be estimated on the basis of their recombination frequencies. The result is
a genetic map which illustrates the relative position of polymorphic gene loci
to each other (see an example in Fig. 12.4).
Ideally, many markers distributed regularly over the complete genome at
approximately equal distances allow a saturated map to be set up. At least 100
polymorphic gene markers and preferably many more are required. Variation
in quantitative traits must be observed on the same individuals within the




Fig. 12.4. Linkage map and localization of quantitative trait loci in a Eucalyptus uro-
phylla clone. (Adapted from Grattapaglia et al. 1995)
12.6 Methods of Selection 191

family. Statistically significant associations between quantitative traits and the

variation at certain gene markers can then be found.
A frequently used method is to relate the genetic markers of individuals with
contrasting phenotypic trait expressions such as particularly fast or slow growth.
A QTL showing an association to particular trait expressions is then presumed to
be located close to those markers. Finally, the approximate location of genes
influencing economically relevant traits can be identified on the map (Fig. 12.4).
It is highly unlikely that any of the gene markers is part of the nucleotide
sequence of the QTL itself. Thus, the function and mode of gene action of this
locus are not characterized. Only the location of the QTL can be inferred.
In marker-assisted selection, individuals are then no longer evaluated only
on the basis of the traits themselves but also on the basis of the variants at
closely linked marker gene loci. Once identified, these loci can be observed
more easily and at a much younger age. However, the applicability of marker-
assisted selection in forest tree breeding is restricted for some reasons:

The results for linkage among gene loci clearly apply to the respective
species but the results for the association of certain alleles at the marker loci
and QTLs apply to the particular full-sib family only. Since forest trees are
selected within populations rather than within one family only, it must be
possible to transfer the results to other families or even populations. Since
in outcrossing tree species linkage disequilibria even within genes rapidly
decrease with physical distance, even the transfer between families would
be possible only for closely linked markers.
The effect of single gene loci on quantitative trait expressions is often small
and difficult to detect by QTL mapping techniques. For instance, the suit-
ability of the markers must be tested by regressing the expressions of the phe-
notypic trait on the presence of no, one, or two marker alleles. Other test
procedures involve the phenotypic difference between carriers and non-
carriers of certain allelic variants. Also the phenotypic differentiation
between offspring carrying a codominant allelic variant in a homozygous or
heterozygous condition can be tested for significance. These methods were
used by Gailing et al. (2005) in several families of central European oaks for
a study of QTLs for flushing date, an adaptive trait in temperate climates. In
two species of the family Fagaceae, Quercus robur L. and Castanea sativa
Mill., QTL positions for this trait but not for the likewise adaptive isotope
discrimination were conserved (Casasoli et al. 2006). Putative candidate
genes for bud burst were at least located on the same chromosomes.
As follows from the test procedure, the identification of QTLs is compli-
cated if the observed quantitative variation is mainly due to nonadditive
gene effects.
192 C HAPTER 12 Domestication and Breeding of Tropical Forest Trees

Linkage maps have been developed for many agricultural species and breeding
techniques based on marker-assisted selection are in use. Saturated linkage
maps have also been developed for a few temperate conifers. A genetic map on
the basis of RAPDs has been developed for some eucalypt families within the
framework of the intensive breeding program of Aracruz Florestal (Grattapaglia
and Sederoff 1994). A linkage map was developed from the full-sib family of a
hybrid cross Eucalyptus urophylla E. grandis (Grattapaglia et al. 1995). Eleven
linkage groups corresponding to the gametic chromosome number of eucalypts
were identified. QTLs were mapped for three different quantitative traits: three
QTLs located in different linkage groups have an influence on the rooting per-
centage of cuttings, four QTLs influence the fresh weight of micropropagated
shoots, and two QTLs have an impact on the sprouting ability of stumps. Since
full-sib families are difficult to produce in some forest trees, also open-pollinated
progenies have been successfully used for detecting QTLs influencing the
volume yield of E. grandis families (Grattapaglia et al. 1996).
More recently, a linkage map was constructed for E. globulus, and candidate
genes (Sect. 7.2) for wood, fiber, and floral traits were mapped (Thamarus et al.
2002). The identification of QTLs and efforts to achieve marker-assisted selec-
tion are also ongoing for tropical acacias (Butcher and Moran 2000; Butcher
The individuals selected in this way are integrated into a population for
further breeding operations or for propagation.

Heterosis Breeding in Forest Trees

Heterosis or hybrid vigor is defined as the superior trait expression of het-

erozygous genotypes in comparison with either corresponding homozygote. It
is the opposite of inbreeding depression. Positive effects of heterozygosity are
widespread in forest trees (Namkoong 1985; Ziehe and Hattemer 1998).
There are various ways to achieve increased heterozygosity, the precondition
for heterosis. Within a population, genetic differentiation of individuals can be
inferred from a large genetic distance between their multilocus genotypes at a
large number of markers. If these markers are representative of the totality of
the genome, the individuals may be expected to possess different alleles also at
the gene loci controlling economically relevant traits; hence, after crossing
those individuals, a large proportion of heterozygotes may be expected.
Villaincourt et al. (1995) found in E. globulus that genetic distance estimated by
RAPDs was of little value in predicting heterosis; however, these authors stated
that genetic distance may be helpful in predicting inbreeding depression from
mating closely related trees.
12.7 Heterosis Breeding in Forest Trees 193

A method of heterosis breeding that is widely practiced in annual crop

plants is recurrent self-pollination until the degree of heterozygosity is sub-
stantially reduced. After three generations it amounts to at most 6.25% and
after three more generations to less than 1% at any gene locus. On crossing
those selfed lines, a certain proportion of highly heterozygous individuals
show heterosis in traits related to yield and stress resistance. This procedure is
hardly applicable in tree species owing to the length of their generation cycle,
the increase of mortality, and the retarded onset of reproduction with
increased homozygosity. Experiments of Durel and Kremer (1995) in Pinus
pinaster, a species less sensitive to inbreeding depression, showed no heterosis
in vigor after crossing trees that arose from first-generation selfing.
Hybridization in the true sense is the formation of progeny by the fusion of
two gametes produced by individuals belonging to different botanical species.
Interspecific hybrids can be produced if there is a certain degree of mating
propensity between the respective (mostly allopatric) species. Species hybrids
of several forest tree genera proved to be particularly fast growing or show
other beneficial trait expressions, for instance, improved resistance against
pathogens (Khurana and Khosla 1998). Hybridization between species belong-
ing to the genera Larix, Pinus, Populus, and Salix gained practical importance
in forestry of the northern temperate region. Most experience regarding
species hybrids of forest trees that are suitable for a tropical climate has been
gained from eucalypts (Barnes and Simons 1994; Example 12.1) and North
American pines (Francis et al. 1984; Nikles 1989).
Meanwhile, hybridization among species also attracts attention as part of
breeding programs of other tropical tree taxa. Hybridization has been proposed
as a way to overcome Heteropsylla attacks on Leucaena leucocephala (Hughes
1989; p. 323 in Smith et al. 1992; Sect. 12.11). Hybridization is also widely used
in breeding of Acacia species that are suitable for the humid and semihumid
tropics. Good performance of the species hybrid Acacia mangium Acacia
auriculiformis has been observed (Sedgley et al. 1992).
Hybridization in connection with vegetative propagation has gained practi-
cal importance for poplars and willows in the temperate zone and for eucalypts
in the tropics. The vegetative propagation of eucalypt hybrids has strongly con-
tributed to the success of breeding programs in Aracruz, Brazil (Example 12.1)
and in the Republic of Congo. It has also proved to be efficient in several other
programs (see van Wyk et al. 1989 for an example from South Africa).
Some species hybrids show heterosis, others do not. The genetic causes of
superior trait expressions in hybrids between certain forest tree species are
still dubious. A combination of genetic information responsible for favorable
trait expressions realized only in species hybrids is a likely cause (p. 122 in
Namkoong et al. 1988). An increased individual heterozygosity of hybrids
at some particular or many gene loci must also have a positive impact on
194 C HAPTER 12 Domestication and Breeding of Tropical Forest Trees

economically relevant traits owing to overdominance (p. 57 f. in Namkoong

et al. 1988.
In some instances progenies are regarded as hybrids, although the parent trees
belong to the same biological species but to different subspecies, varieties, or
provenances. Interprovenance hybrids of several conifer species (e.g., Araucaria
cunninghamii, Pinus pinaster, Pinus sylvestris) were produced. However, inter-
provenance hybridization currently plays a minor role in forest tree breeding
(Nikles 1993). Bolstad et al. (1991) reported on intermediate growth of inter-
provenance hybrids in Pinus banksiana.
The expression of phenotypic traits in provenance hybrids could not only
show whether there exists heterosis that could be used in breeding, but also
whether a reduction in yield has to be expected in the subsequent generation.
This question has gained importance, since plantations containing mixtures of
populations (Stern 1963) are used for seed production.

Example 12.1: Breeding Eucalypts in Aracruz, Brazil

More than 130,000 ha of eucalypt plantations have been established since 1967
in Aracruz, Brazil, by the private company Aracruz Florestal, a subsidiary of
Aracruz Cellulose. The wood produced is used for the production of pulp and
paper in the paper mill of the parent company.
During the early years of the program, an average annual volume yield of
12.4 m3 ha1 was achieved under favorable climatic and edaphic situations.
These first plantations were established from seedlings and were repeatedly har-
vested by coppicing. Since 1979, the original plantations have successively been
replaced by clonal plantations. The clonal plantations grow extraordinarily fast.
Their average annual volume production amounts to 60 m3 ha1. Air-dried pulp
yield increased from 3.3 (in 1969) to 15.9 t ha1 year1 in 1990. Furthermore, the
wood produced is particularly well suited for pulp production and the clones
are highly resistant against insects and diseases. Clonal plantations have been
harvested since 1986. Apart from the increase in volume yield, the cost of har-
vesting and wood extraction as well as the amount of wood required per unit
weight of pulp was reduced.
The first seedlings for plantation establishment were Brazilian land races of
the species E. grandis, E. saligna, and E. alba. The origin of these land races is
unknown. In 1973 a comprehensive breeding program was initiated using both
sexual and vegetative propagation. During 1989 more than 1,500 provenances
or families of 55 eucalypt species were under test in Aracruz.
Provenances of E. grandis from Queensland and E. urophylla from Flores
and Timor (Indonesia) proved to be particularly suitable for plantation estab-
lishment in Aracruz. More than 130 ha of seed production areas and over 50 ha
of seed orchards were established for the production of sexual offspring.
12.7 Heterosis Breeding in Forest Trees 195

Full-sib families of E. grandis and the hybrid E. grandis E. urophylla were pro-
duced. Controlled pollination for artificial hybridization required the development
of techniques for the long-term storage of pollen, because different eucalypt species
flower at different times in Aracruz. Partial diallels were used as mating schemes.
Breeding success considerably increased by the vegetative propagation of
elite trees. Seventy-five clones originating from crosses between E. grandis,
E. urophylla, and E. pellita were selected in 1989. These clones proved to be
completely resistant against cancerous diseases and, when coppiced, showed
high volume growth and desirable wood properties. The trees performing best
in progeny tests are continuously selected according to a set of criteria and are
mass-multiplied in clonal archives in order to steadily increase the number of
clones used in the wood-producing plantations. The multiple-population con-
cept (Sect. 12.9) has been integrated with 20 subpopulations of E. grandis and
E. urophylla consisting of 20 trees each. The best-performing trees in ten full-
sib families produced in each subpopulation are used for making hybrid
crosses and selecting among those hybrids for clonal testing.
Principles of the breeding strategy of Aracruz Florestal are presented in
Fig. 12.5. Breeding is supported by research on in vitro techniques for the

Introduction of
species and V Vegetative propagation

Best species and


Seed collection
Best trees in best provenances

Progeny test
Unimproved plantations
(with hybrids)
Selected trees
Best trees in best families
Individual tree
V selection
Controlled hybrids Breeding orchard Seed orchard
1st generation V

Progeny test Progeny test Seed orchard Clonal test

1 1/2 generation V

Individual tree Selected trees Seed orchard Clonal multiplication

Clone bank
selection Best trees in best families 2nd generation area

Clonal test V
Clonal multiplication Routine
Clone bank area plantation

Fig. 12.5. Breeding strategy for eucalypts in Aracruz. (Adapted from Campinhos and
Ikemori 1989)
196 C HAPTER 12 Domestication and Breeding of Tropical Forest Trees

multiplication of clones and by the use of molecular markers to identify QTLs

(Sect. 12.6).
The breeding strategy of Aracruz Florestal is integrated into a production
system which also comprises components such as pest management, mainte-
nance of balanced plant nutrition by minimizing nutrient extraction, and, if
necessary, fertilization, careful logging operations, and preservation of biolog-
ical diversity. More than 70,000 ha of natural forests (i.e., 35% of the total area)
has been preserved between the eucalypt plantations for the maintenance of
biodiversity (From Campinhos and Ikemori 1989; Campinhos 1991, 1993a, b,
1999; Foster and Bertolucci 1994).

Propagation of Breeding Products

A breeding project includes both the selection of superior trees and their sub-
sequent multiplication in order to capitalize on the breeding progress achieved;
hence, a certain progress achieved in breeding makes sense only if it is possible
to mass-propagate the selected material with as little genetic change as possi-
ble. Several options for sexual and vegetative propagation exist in tropical tree
species that were hitherto subjected to domestication. Seed production areas
are addressed in Sect. 13.5.

Clonal Seed Orchards

Propagating plus trees by harvesting their seed is hardly adequate for pro-
ducing the quantity and quality of reproductive material required for the
establishment of plantations. For the breeding progress involved in plus-tree
selection to become manifest, the individuals have to be brought into mat-
ing contact in a clonal seed orchard. The fertilization of ovules by external
pollen (pollen contamination) is avoided by maintaining sufficient distance
to conspecific populations in the region and by planting large orchards.
Remote pollen is then highly unlikely to prevail even if the density of the
orchard is low. In young seed orchards one has to envisage incomplete pol-
lination by the orchard clones, since the trees do not yet produce abundant
pollen and are planted with wide spacing. Consequently they are eventually
flooded by foreign pollen produced by unselected trees in surrounding
The reproduction in seed orchards is rarely random (Sect. 6.2) just as is the
case with forest stands. However, an equilibrium resulting in HardyWeinberg
structures (Sect. 6.2.2) and allele frequencies representing the frequencies
12.8 Propagation of Breeding Products 197

within the deme of the plus trees selected may be approximated by some pro-
visions made in the design of the orchard.
The clones are as a rule represented by the same number of copies (ramets).
In order to overcome implicit setbacks due to high variation in reproductive
output of the clones, additional ramets for the less prolific clones have been
recommended; however, this condition possibly induces additional kinship
among the seed produced depending on the variation in male mating success.
Vegetative propagation by cuttings is difficult if not impossible because of the
usually advanced physiological age of the plus trees selected (Hong 1975). The
trees selected are usually first incorporated in a clonal garden, a multiplication
garden, or a clonal archive. The scions for grafting are taken therefrom. Many
techniques for the production of grafts are in use (van Wyk 1977; p. 399 ff. in
Hartmann and Kester 1983). It must be checked repeatedly that the grafts in fact
develop from the scion rather than the unselected root stock (Example 11.2).
The number of clones represented in the orchard is important. In order to
maximize breeding progress, the selection intensity is high and the number of
retained clones moderate. However, if the orchard contains only a few clones,
the risk of phenological mismatch among them increases. The risk of uninten-
tional genetic change or even loss of alleles is obvious (Hattemer et al. 1982).
The size of an orchard is best determined by the number of clones rather
than the number of ramets per clone. Many old seed orchards contain only 20
or even fewer clones. Meanwhile, larger numbers of clones (up to 100) and,
consequently, lower numbers of ramets per clone are common (cf. Example
11.2). Problems related to losses of genetic variants due to small numbers of
orchard clones can be solved by reducing the intensity of plus-tree selection
that per se is very high. Ideal would be the reduction of the number of ramets
per clone to a few or just one grafted scion.
Each clone is represented with a limited number of ramets (grafts). Increasing
the number of ramets of a clone is genetically equivalent to the enlargement of
the crown of an individual tree; thus, the chance for self-pollination increases.
With reference to the conditions in a seed orchard, we have to consider pollina-
tor movement not only within but also between the crowns of the ramets; hence,
both individual and clonal self-fertilization are possible unless prevented by a
system of incompatibility. Both types are identical with regard to the genetic
consequences (Sect. 6.4).
The necessary spatial isolation from other populations of the same species
obviously depends on the system of pollen dispersal, and in particular on the
efficiency of the pollen vectors (Sect. 5.2). Clonal seed orchards are designed in
order to maximize the number and proportion of offspring from outcrossing
with pollen from other selected clones. Keeping the average distance between
ramets of the same clone large helps to lower the risk of clonal self-fertilization.
Many orchards are designed in completely randomized blocks. Some subsequent
198 C HAPTER 12 Domestication and Breeding of Tropical Forest Trees

corrections are made in order to prevent close proximity of ramets of the same
clone. Certain systematic designs are even more efficient in decreasing the pro-
portion of clonal self-fertilization (Nester 1994).
The density of seed orchards is much lower than in normal plantations in
order to get larger tree crowns. The trees are often sheared in order to develop
a short trunk and a wide crown to facilitate easy seed collection. Early and
gregarious flowering is promoted.
Clonal seed orchards were established on a large scale for conifers of the
northern hemisphere (Pinus spp., Picea spp., and Pseudotsuga spp.) and for
pines, eucalypts, and teak in the tropics (Eldridge 1975; Feilberg and Segaard
1975; Example 13.2).
The mating systems of clonal seed orchards have been studied in both tem-
perate and boreal conifers. Estimates of the proportion of self-fertilization in
conifer orchards are usually low. They are thus comparable to the situation in
plantations and natural populations (Muona and Harju 1989). Deviations
from random mating were proven in orchards of Pinus sylvestris (Mller-
Starck 1982) and of teak (Finkeldey 2006). El-Kassaby and Askew (1998)
described some genetic characteristics of the reproductive process in clonal
seed orchads.
An excess of self-fertilization and other deviations from random mating as
well as pollen contamination have an impact on the genetic structures of the
progeny produced; thus, the genetic gain attainable by a production of seeds in
orchards is likely to be overestimated if it is based on the assumption of ran-
dom mating among the clones and their isolation (Mller-Starck 1982, 1991).
Estimates of the contamination rate with foreign pollen from outside the seed
orchard are often surprisingly high (Paule et al. 1993; Pakkanen et al. 2000;
Sect. 5.2.2).
In South Korea, Pinus rigitaeda, the hybrid between Pinus rigida and Pinus
taeda, has been mass-produced by handcrosses made on grafts. The equally
successful hybrid between Larix europaea and Larix kaempferi can be mass-
produced in specially designed seed orchards containing many grafts of a self-
sterile clone and many clones of the other species. Seed is then collected only
from the self-sterile clone.

Example 12.2: Clonal Seed Orchards of Teak (Tectona grandis)

Among tropical tree species the largest area of clonal seed orchards has been
established for teak. More than 1,800 ha was planted from 1966 to 1978 in
Thailand (Fig. 12.6). More than 1,000 ha exists in India, more than 500 ha in
Indonesia, and smaller areas in Bangladesh, Myanmar, China, and several other
The establishment and maintenance of the clonal seed orchards of teak in
Thailand is under the supervision of the Teak Improvement Center established
12.8 Propagation of Breeding Products 199

Fig. 12.6. A clonal seed orchard of teak (Tectona grandis) in Thailand. (Photo:
R. Finkeldey)

in Lampang in northern Thailand. In total more than 300 clones were selected
on the basis of the phenotypic superiority of plus trees. A clonal multiplication
garden for the multiplication of selected trees was established at the Teak
Improvement Center. Single buds were grafted on stumps (Fig. 12.7). This
grafting method (budding) is quick and yields a high proportion of successful
grafts (Keiding and Boonkird 1960). Planting distances varied from 3 m 3 m to
12 m 12 m. The oldest orchard contains only 16 clones. Younger seed orchards
contain up to 100 clones; some of the clones are planned to be selectively
removed depending on the results of clonal and progeny tests.
200 C HAPTER 12 Domestication and Breeding of Tropical Forest Trees

a b
Fig. 12.7. Grafted teak (T. grandis) plants produced by the budding technique. (Photos:
R. Finkeldey)

Many seed orchards were established far from other teak forests in order to
ensure their reproductive isolation; however, several of the sites chosen were
unsuitable for growing teak. Thus, not all of the orchards established survived
and are productive. Seed harvest is below expectation in the 1830-year-old
orchards. In all orchards physiological incompatibilities between scion and
root stock are frequently observed on older grafts (Fig. 12.8). The low amounts
of fruits currently harvested in clonal seed orchards provide only a minor pro-
portion of the total reproductive material for teak plantations in Thailand.
Genetic marker studies based on isoenzyme gene loci were used to check the
clonal identity of the ramets in the clonal multiplication garden of the Teak
Improvement Center and in two clonal seed orchards. Clonal identity has not
been maintained for many trees of the multiplication garden and, consequently,
neither has it been maintained in the seed orchards. Possible explanations are a
mix-up of ramets during the establishment of the clonal garden and the seed
orchards as well as the occurrence of adventitious sprouts from the stump
which eventually develop into the main stem of the supposed graft. High levels
12.8 Propagation of Breeding Products 201

Fig. 12.8. Physiological incompatibility between root stock and scion in a grafted teak
(T. grandis) tree. (Photo: R. Finkeldey)

of self-fertilization (Example 6.1) and restricted pollen dispersal (Example 5.1)

were observed in the seed orchards (From Hedegart et al. 1975; Finkeldey 2006).

Seedling Seed Orchards

Also seedling seed orchards established with sexually produced trees serve
the production of seeds. In most cases seedling seed orchards arise from the
conversion of a progeny test aimed at the identification of superior families.
202 C HAPTER 12 Domestication and Breeding of Tropical Forest Trees

Those families showing inferior trait expressions are rogued by selective

thinning that stimulates also crown enlargement. Also inferior trees within
well-performing families are removed from the field test. After this com-
bined selection step, seeds are harvested that are used for the establishment
of plantations.
More details related to the establishment of seedling seed orchards have
been summarized by Schmidt (1993b). Seedling seed orchards were success-
fully established for many tropical tree species. Examples are Cordia alliodora
in Costa Rica (Boshier and Mesn 1989), several eucalypt species (Eldridge
1975), and Acacia mearnsii (Jones and Burley 1975).
Inbreeding depression is likely to result from a large proportion of mating
within families which may counteract the positive effects of selection on phe-
notypic traits. Selection among families must not be very intensive in order to
leave a sufficient number of families to avoid strong inbreeding.
If later conversion of a progeny test is planned, certain precautions in the
layout contribute to the success of producing superior-quality seed. The test
should be planted so as to maintain some degree of isolation. This is not very
easy to accomplish, since the progeny tests must be planted on typical sites of
the respective species where conspecific stands are possibly common. Both plot
size and spacing should allow sufficient numbers of trees to be kept during their
reproductive phase. As Fig. 12.9 shows, the layout in conventional plots may lead
to a highly uneven distribution of the remaining trees unless the selection inten-
sity is fairly low. Randomization of all trees within the blocks (single-tree plots)
helps to cure this problem and also helps to avoid an immediate neighborhood
of full-sibs. Since it is then very difficult to keep track of the trees belonging to
certain families, the production of a map and permanent tagging of the trees
during the experimental phase is recommended. Ladrach (1998) warns that
errors in installation and labeling of trees may ruin an experiment.
In tree species pollinated by either wind or animals the age, size, and density
of an orchard determine the pollen production and hence the risk of the
orchard being flooded by the pollen of the surrounding forest. Of course,
distance isolation is important. In contrast to clonal orchards, the system of
effective pollen dispersal can be analyzed in this type of orchard where every
tree has another genotype, so paternities of particular trees and not only clones
can be assigned to offspring (Sect. 5.2.3).
A 15-year-old field test with 12 open-pollinated progenies of E. grandis in
Madagascar had been planted with single-tree plots in 40 replicates (Chaix
et al. 2003). Following evaluation, it was converted into a seed orchard com-
prising 349 trees. The orchard trees and seedlings raised from their seed were
genotyped at six microsatellite loci. Thirty-nine percent of the effective pollen
was not produced inside the orchard or by trees in the two border rows.
Effective pollen received by 30 seed trees was transferred by introduced honey
12.8 Propagation of Breeding Products 203

Fig. 12.9. Conversion of a progeny test with 16 families (left) into a seedling seed
orchard (right). One block containing four trees per family is shown. The eight families
with the numbers 411 are selected. In a two trees and in b only one tree per four-tree
plot is retained. In c and d the plots are not contagious and the trees are randomized
within the block. In c two trees per family are retained and in d only one tree per
family is retained. Note the uneven distribution of the remaining trees in a and b

bees over distances ranging between 32 and 85 m with no mating preference

for neighboring trees. The outcrossing rate of 96.7% was much higher than
reported from natural forests (Table 6.2; Example 6.2). Seed trees of this species
growing close to the edge of the orchard showed the strongest deviations of the
realized matings from those expected. The reproductive success of the families
deviated moderately from equilibrium expectations. Pertinent ratios ranged
from 0.32 to 2.02, i.e., the families contributed between one third of and double
the expected number of genes to the seed produced.
204 C HAPTER 12 Domestication and Breeding of Tropical Forest Trees

When comparing the mating system in an isolated seedling seed orchard

with that in primary or logged forest of Dryobalanops aromatica (Sect. 10.2.1),
Lee (2000) found much reduced outcrossing rates (ts = 0.436) and a moderately
reduced number of male mates for single seed trees (rp = 0.428 vs. 0.389 in pri-
mary forest). The author attributed this finding to a lack of pollinators and the
transfer of much pollen by gravity alone, since the orchard was established with
only one species. This may have created problems with the subsistence of
pollinators in this irregularly flowering tree species. The difference between the
multilocus (tm) and the single-locus (ts) outcrossing rate is an estimator of
the level of inbreeding due to mating other than selfing, i.e., mating among rel-
atives or consanguineous mating. A large difference tm ts = 0.115 indicated the
use of related planting stock when the orchard was planted 70 years earlier.
This result was all the more surprising, since the density of the orchard with 20
trees per hectare exceeded that in the primary forest with 15 tress per hectare.
In all, this seed orchard showed increased inbreeding by both self-fertilization
and consanguineous biparental mating.

Mass Multiplication of Clones

During clonal propagation the entire genotypes of the trees selected are cap-
tured for the establishment of plantations, since no recombination is involved.
This provides for one-to-one exploitation of an achieved level of breeding and
is a particular advantage for rare outstanding trees. Zobel and Talbert (1984,
Chap. 10) and Zobel (1992) strongly advocated operational vegetative propa-
gation in tropical forest trees. Park et al. (1998) reviewed the advantages and
explained the integration of clonal selection and vegetative propagation into
breeding programs.
Many forest tree species are easy to propagate vegetatively (Sect. 4.3.1).
Thus, single or a few superior trees may be selected in the forest or in a field
test and identically mass-multiplied. Asexual propagation is the basis of clonal
forestry (Chaperon 1984). Among practical advantages, clonal propagation is
useful, since the reproductive phase need not be awaited; however, the matura-
tion phase must be arrested by hedging. Clonal forestry decreases the hetero-
geneity of chemical and physical wood properties. A homogeneous product
can be beneficial in itself for certain end uses such as pulp and paper making.
Greater uniformity of plantations might be of particular value, where wood is
used in industry. Also, specific management procedures may be developed for
individual clones (Burdon 1989). Particularly clonal forestry critically depends
on specific production and management systems. The establishment of large-
scale clonal plantations requires intensive forest management techniques,
including forest protection and the maintenance of soil fertility, for example,
by fertilization (Example 12.1).
12.9 Multiple Population Breeding 205

The advantages listed previously once gave rise to clonal forestry. In contrast
to sexual propagation, clonal propagation reduces the genotypic multiplicity of
plantations drastically unless the heterozygosity of both parents is extremely
low. It is highly unlikely that two trees even of the same full-sib family share an
identical multilocus genotype. The recombination during sexual reproduction
results in a very large number of multilocus genotypes. In clonal forestry the
number of multilocus genotypes equals the number of clones represented in a
plantation. Although the spectacular yields of selected clones in plantations
have created much enthusiasm about clonal forestry in tropical countries, the
loss of genotypic variability within plantations is associated with certain risks
(Sect. 12.11).
Also in vitro propagation has received much interest (Khuspe et al. 1994).
These authors report on adapting techniques to a variety of tropical trees and
to the unpredictably flowering bamboos.
Summarizing experience with eucalypts in clonal forestry, Eldridge et al.
(1994, p. 2) stated that the mass vegetative propagation of plus trees selected in
plantations has been most successful.
Clonal forestry based on an intensive breeding program may be comparable
to the breeding of high-yielding varieties of agricultural crop plant species dur-
ing the 1960s and 1970s. The success of the green revolution has not simply
been due to isolated breeding programs but resulted from their integration into
intensified agricultural production systems in tropical countries.
In species hybrids vegetative propagation has a particular advantage, since
recombination during sexual reproduction would reduce an effect of heterosis.
Some species hybrids proved to be particularly suitable for vegetative mass
multiplication and the establishment of clonal plantations. Probably the most
successful breeding program for tropical trees was initiated in Aracruz, Brazil
(Example 12.1). A similarly comprehensive breeding and multiplication pro-
gram which provides clones for large-scale plantations of eucalypt hybrids was
initiated in the Republic of Congo with French support during the 1970s
(Delwaulle 1989). Clonal eucalypt plantations owned by private enterprises
exist also in India and other countries. The production and selection of clones
in these plantations is reported to be based on less intensive breeding programs
than those in Aracruz and the Republic of Congo (Lal 1993).

Multiple Population Breeding

The hitherto mentioned breeding methods are hierarchical since certain indi-
viduals or genotypes are selected from a base population, and these individuals
or their offspring are assembled to a breeding population of again smaller size,
etc. This approach has been adopted with certain modifications from traditional
206 C HAPTER 12 Domestication and Breeding of Tropical Forest Trees

breeding methods for agricultural crop plants. The expected breeding progress
is accompanied by a gradual loss of genetic variation in the breeding population
which does not only affect the gene loci selected but owing to possible drift
effects the totality of the genome.
The multiple population breeding concept of Namkoong is nonhierarchical
and aims at the preservation and utilization of genetically differentiated sub-
populations (Namkoong et al. 1980, 1988, p. 70 ff.; Namkoong 1984). In 1994,
Namkoong was awarded the Marcus Wallenberg Prize for his pathbreaking
contributions to quantitative population genetics, tree breeding, and manage-
ment of genetic resources, which form a solid scientific basis for the mainte-
nance of biological diversity in forests all over the world.
The objective is no longer breeding of a single population accompanied by
a reduction of genetic variation or the selection of only a few genotypes that
are expected to produce superior phenotypes in different environments.
Instead, the breeding population consists of a variable number of separate sub-
populations. Multiple population breeding is directed towards different objec-
tives, i.e., different traits are selected at different sites. Even if the selection
criteria are the same, sets of controlling genes are expected to be differentiated.
Both the adaptedness and the adaptability to different and changing environ-
mental conditions are promoted by the maintenance or enhancement of a
differentiated population structure. The concept sacrifices the maximization of
breeding progress in a few economically important traits. An important objec-
tive is the enhancement of the adaptedness of different subpopulations to
diverse environmental conditions. Adaptability is maintained by the resulting
differentiated population structure. The concept is highly flexible but requires
skill for managing the genetically effective population size, including pedigree
control. It is possible to adjust the breeding program to changed economic pri-
orities since different breeding objectives are defined in different subpopula-
tions. Both the number of replicate populations and the number of parents in
populations, their diversity and their differentiation can be used to control the
total genetic variation available for managers to use in deploying trees in
plantations (Namkoong 1999). Now as before, the total size of the subdivided
breeding population is crucial under the aspect of inbreeding avoidance.
However, the genetic variation can be managed more effectively.
Multiple population breeding is particularly suited for potentially important
plantation species of the tropics which still are in an early state of domestica-
tion. A main advantage of the concept is its flexibility both with regard to the
necessary expenditure and the possibility for modifications during its imple-
mentation (Namkoong et al. 1980). For instance, also heterosis breeding can
easily be integrated. Furthermore, breeding and conservation objectives are
regarded as equally important and may be profitably combined in a single
program (Namkoong 1984).
12.10 Breeding Strategy 207

Hierarchical breeding strategies lead to a reduction of population sizes and

thus to increased genetic drift in shaping genetic structures. Genetic drift
affects not only selected loci, but principally all gene loci. The main effects of
genetic drift are an increased risk for the loss of genetic variation, random fluc-
tuations of genetic structures at single loci, and modified associations of alleles
at many loci.
The breeding strategy of several tree species in Zimbabwe has followed the
multiple population breeding concept since the early 1980s (Barnes and Mullin
1989). Although the concept has hitherto gained only limited practical impor-
tance in spite of its many advantages, it is hopefully going to be implemented
in more projects in future. Simons et al. (1994) have developed concepts of a
strategy shaped to nonindustrial tropical trees.

Breeding Strategy

The breeding progress made in a single generation is hardly ever substantial

enough to justify the expenditure for a single selection cycle. Depending on
what has been achieved, selection continues to a second, third, etc. generation.
For instance, Reddy and Rockwood (1989) reported on breeding progress
made in tree volume during four generations of selection in a land race of E.
grandis. Volume per tree increased from 7.04 dm3 in the first generation to 21.5,
25.9, and 40.2 dm3 in the subsequent generations.
The ultimate goal is the long-term breeding gain. In the framework of a
breeding strategy, it is planned how to manage those various populations
(White 1987), the methods of selection and propagation, and the introduction
of unrelated material. The first elements of planning are the evaluation of
material available as a base population and its size. Quite a few breeding proj-
ects were taken up that were too small from the very beginning. Others were
taken up without properly sampling the distribution range of the species in
order to set up a suitable base population (Barnes and Simons 1994).
In the first generation, mass selection prevails. Later operations include a
combination of family and individual selection. The available information on
genetic parameters increases and may be integrated. Libby (1973) has set up
and discussed a fairly general strategy for tree breeding.
Important items are comparisons of expected and realized breeding progress,
the monitoring of the genetic variation at marker gene loci, and not too intensive
or even aggressive selection (Sect. 12.11). Intensive selection reduces the popula-
tion size and favors the increase of the inbreeding coefficient. A solution can be
found by selection within several breeding groups. Subsequently more diverse
outcrossed progeny for deployment are produced in combined orchards.
208 C HAPTER 12 Domestication and Breeding of Tropical Forest Trees

The main objective of forest tree breeding is the modification of average trait
expressions of economically relevant traits in plantations. However, breeding and
propagation of breeding products is usually done in small populations. Thus, it
is meaningful to distinguish the following populations as part of a breeding
strategy (Libby 1973; van Buijtenen 1975; Kang et al. 1998):

Both natural populations and plantations may be used as base populations

from which individual trees are selected both at the beginning of the breed-
ing program and during later stages. It must neither be modified nor be
manipulated by artificial selection or even replaced. Its genetic structure
can still not be maintained in its original condition but is subject to evolu-
tionary change. Its production function has low priority.
Selected trees or their progenies are merged into breeding populations
such as progeny tests or clonal tests. They are mainly used for evaluating
and selecting individuals and must allow for future responses to selection.
They represent the core of breeding projects, they must be carefully
designed and monitored, and they must be able to persist.
Propagation populations contain subsets of the trees of the breeding pop-
ulations and serve the multiplication of selected genetic information (not
necessarily genotypes). Examples are seed production areas and seed
The breeding products harvested from propagation populations are used
to establish production populations (plantations).

The populations serve different functions. For instance, a progeny test (breed-
ing population) may later be converted to a seedling seed orchard (propagation
population). Under a multiple population breeding strategy, the base, breed-
ing, and propagation populations are primarily separated, although there are
relationships between the populations (Kang et al. 1998). The breeding pro-
gram has the improvement of yield (in the widest sense) and the conservation
of genetic variation as two goals possessing equal priority. It is integrated into
the domestication process of the respective tree species.
The decrease of additive genetic variation (Young and Boyle 2000) in
breeding populations as closed systems is inevitable. One may think of a
somewhat simple relationship requiring little theory of genetic variances and
selection plateaus. In a given population a selection trait is assumed to be con-
trolled by several biallelic gene loci without dominance or epistatic effects. If
the favorable alleles possess low frequency in the base population, the genetic
diversity is going to increase under selection. Maximum diversity is reached
when the alleles possess equal frequency. The more the breeding goal is fur-
ther approached, the smaller becomes the diversity. The situation is, of course,
12.11 Genetic Consequences of Domestication and Breeding 209

fundamentally different if new individuals, families, or even other populations

with other allelic variants are integrated into the breeding population.
Breeding of forest trees is time-consuming even in tropical species, since
only a long-term approach to breeding is successful. The progress expected
from breeding can only be higher than that from operations such as plus-tree
selection alone. A breeding strategy combines several methods of breeding and
propagation in order to both satisfy the immediate need for reproductive mate-
rial and reach long-term breeding objectives. The breeding strategy of Aracruz
Florestal for the breeding of eucalypts was discussed in Example 12.1.
The fundamental importance of improving our knowledge of the genetic
system of tropical forest trees and the equal priority of breeding and conserva-
tion objectives are not sufficiently appreciated by conventional breeding hier-
archical strategies. Management plans for the genetic resources of a species will
put more emphasis on population genetics research to understand and possi-
bly manipulate the genetic system of a species. The final objective will be to
integrate breeding and conservation objectives into a coherent and flexible
strategy. The multiple population breeding concept of Namkoong (Sect. 12.9)
deserves attention in this context.

Genetic Consequences of Domestication and Breeding

General Considerations

The global process of domestication in conjunction with breeding serves the

welfare of mankind. It neither shall be nor can be hindered. On the other hand,
at some stages of this process there exists the chance to retard the tendency
towards impoverishment of genetic variation. Domestication may also help to
rescue genetic variants.
Artificial selection changes genetic structures of populations in manifold
ways. Successful selection for a trait changes the allelic structures at the con-
trolling gene loci and has some correlated effects on other traits.
Artificial selection is hardly identical to natural selection both with regard to
the traits concerned and with regard to the preferred trait expressions. For this
reason the adaptedness of a population or its adaptability to environmental
change (Sect. 7.2) can be adversely affected by breeding. Guries (1990) put is
this way, Human desires may be counter to those that have guided evolution
over a long period of time, and compromises may have to be struck between
the intensity of breeding activities and biological factors that work to ensure
210 C HAPTER 12 Domestication and Breeding of Tropical Forest Trees

the long-term survival of the species. However, the actual danger of reduced
adaptedness through breeding has not been proven by experimental studies in
tropical forest trees. Some results have been reported for temperate and boreal
species. For example, selection for fast growth during early stand development of
Norway spruce (Picea abies) negatively affects the resistance against wind-break
at later age (p. 399 in Hattemer et al. 1993), an important element of stability. In
this case, breeding would lead to instability, i.e., the departure of realized from
predicted performance (Namkoong 1999). Among all organisms subjected to
breeding work, the environment of forest trees can be controlled least.
The less intensive the finite breeding populations are selected, the less is
population size reduced. Part of the breeding progress should be relinquished,
for instance, by retaining a few individuals also of culled families in the breed-
ing populations and the propagation populations such as seedling seed
orchards. The breeder must find a best compromise between selection objec-
tives and selection intensity.
Another form of renunciation is restricting the time during which seed is col-
lected from a given seed orchard. In view of the money spent for establishment,
there is a tendency to use the offspring for planting on vast areas. These consid-
erations may look uneconomic and paradoxical at first sight but they represent
sustainability imperatives. Immediate maximum economic benefit from tree
breeding implies setbacks during future generations of both trees and humans.
The technologies of breeding and selection offer opportunities for manag-
ing the genetics and guiding the evolution of forests that neither decrease nor
increase risks over that of uncontrolled, natural regeneration. Technology
involved in domestication and breeding is neither the problem nor the solu-
tion. It is a tool that humans have available. It can be used well or poorly
(Namkoong 1999).

Monitoring of Breeding Projects by Use of Genetic Markers

There is ample opportunity for the use of gene markers in breeding programs
(Yeh 1989; Forrest 1994; Butcher et al. 1999). A highly rewarding application is
the verification of the validity of crosses, as an example may show. A tree breeder
had established a clonal seed orchard. When it started to flower, crosses among
the clones were made for progeny testing. Years later it was detected by the appli-
cation of genetic markers that the progeny test was a complete mess, because
a muddle had happened already during production of the grafts. Part of the
financial loss may have been avoided by checking the clonal identity of the ram-
ets and by analyzing seed samples in the laboratory before sowing. Since possibly
12.12 Recommended Literature 211

high investment is made into pedigree breeding continued over generations,

keeping track of ancestries and coancestries is highly recommended.
These markers are widely unspecific and help to infer and monitor variation
in the genome, i.e., the genetic background of, for instance, breeding popula-
tions. It is not necessarily likely that the moderate number of markers that we
are able to identify will allow inference of the adaptive variation or even the
possession of adaptive variants at certain gene loci.
Goto et al. (2001) developed an efficient method to check the identity of
ramets in clonal orchards on the basis of RAPDs. According to Park et al.
(1998), keeping track of clonal identities also in vegetative mass propagation is
of minor importance but might depend on the management of deployment.
Applications of isoenzymes were discussed by El-Kassaby and Ritland
(1998), Bergmann and Hattemer (1998), and Mitton (1998). Acher et al.
(2004) and Eliott et al. (2005) gave examples illustrating that meanwhile all of
these objectives can be reached by the use of more variable or uniparentally
inherited markers.

Recommended Literature

The textbook by Wright (1976) covers forest genetics and tree breeding, includ-
ing provenance research. An introduction to quantitative genetics that has
found wide circulation is the book by Falconer and MacKay (1996). It deals
with population genetics and all aspects of quantitative genetics, including
simultaneous selection for multiple traits and the use of QTLs. Namkoong
et al. (1980, 1988) presented comprehensive accounts of breeding forest trees
on the basis of population and quantitative genetics. The book by Zobel and
Talbert (1984) covers all aspects of tree breeding, such as methods and results.
The book by Zobel and van Buijtenen (1989) is on wood and deals with the
genetic aspects, such as variation and breeding. The book edited by Fins et al.
(1992) covers all aspects of forest tree breeding and presents numerous exam-
ples from temperate species. Eldridge et al. (1994) cover breeding, with empha-
sis on eucalypts. In the book edited by Mandal and Gibson (1998), topics with
significant current interest in forest genetics and tree breeding are discussed. As
stated at the end of Chap. 11, the book by Williams et al. (2002) deals with
experimental design and analysis also in close-sense tree breeding.
Genetic Aspects of Plantation Forestry 13
in the Tropics


In tropical countries ever more land is becoming available for tree plantations.
This is due to the destruction and degrading of natural forests or failures of
their management and growing problems of access to existing forest.
Plantations represent either reforestation, i.e., changing deforested land back to
forest land, or afforestation, i.e., creating forest where it historically did not
exist. The high productivity of tropical and subtropical plantations and their
numerous ecological, social, and economic benefits make plantations highly
attractive. Much attention has recently been devoted to research and planting
of nitrogen-fixing hardwoods such as Leucaena spp. (Brewbaker and Sorensson
1994) and Sesbania spp. (Owino et al. 1994). Planting is urgently required on
eroded sites or on sites threatened by erosion. It is also practiced with the
objective of restoring forest ecosystems. Planting is done by private companies,
in the framework of special developmental projects, and in national afforesta-
tion programs. Planting projects are also promoted by private companies and
nongovernmental organizations (Evans and Turnbull 2004). The total area of
forest tree plantations in tropical and subtropical countries has been estimated
to be close to 100 106 ha in 2000 (Table 13.1) and may have exceeded this area
since then. The annual loss of forest area is still estimated to exceed by far that
of newly established plantations (FAO 2002).
A comprehensive discussion of tropical tree plantations has to take into
account ecological, genetic, silvicultural, and socioeconomic aspects (Evans and
Turnbull 2004). The genetic aspects of man-made forests established by plant-
ing or sowing are centered around their establishment. Therefore, the most sig-
nificant genetic aspect is the choice of reproductive material used. Plantations
are much less competitive than natural forests. The competition among adult
trees for light, nutrients, and water is not accompanied by as much mortality as
during the early phases of the life cycle in natural populations. These critical
phases are replaced under human intervention by procedures designed towards
214 C HAPTER 13 Genetic Aspects of Plantation Forestry in the Tropics

Table 13.1. Areas of planted forests including woodlots in tropical and subtropical
regions and their average annual increase from 1965 to 2000. Numbers refer to
thousands of hectares. (Adapted from Evans and Turnbull 2004)
Region 1965 1980 1990 2000 Annual

Africa 1,378 2,724 3,773 4,566 170

Asia (including southern China) 421 13,046 29,245 73,444 2,110
Northern Australia and Pacific Islands 70 269 420 480 16
Central America and Caribbean 219 486 786 1,311 44
South America 579 4,448 8,470 8,634 263
Total 6,667 20,973 42,694 88,435 2717

maximum amounts of planting stock to be derived from a given quantity of

seed. This is typical of all forms of artificial regeneration.
Many tropical plantations consist of very few if not only one tree species and
are therefore perceived as ecologically unstable moncultures. However, the real
cause for concern should rather be the eventual genetic uniformity of the
respective species that often accompanies monoculture (Guries 1990). This is
due to the restricted number of parent individuals of the planting stock and is
most pronounced in uniclonal plantations. Visual uniformity of even-aged
plantations established at regular spacing is attractive to some people and
unnatural to others but of minor importance.
A moderate number of tropical tree species are used in plantations (Table
13.2). The species differ much in growth and ecological demands. Their choice
depends on the purpose of the plantations. A large proportion of them are used

Table 13.2. Species used in tropical planted forest. Percentages refer to the main
groups only. Adapted from Evans and Turnbull (2004, p. 37)
Genus/group Percentage Species

Eucalyptus 50 grandis, camaldulensis, globulus, saligna, deglupta,

tereticornis, robusta, citriodora, exserta, urophylla, hybrids,
and others
Acacia 17 nilotica, mangium, auriculiformis, crassicarpa
Tectona 10 grandis
Pinus 23 caribaea, patula, elliotti, oocarpa, kesyia, merkusii,
massoniana, and others
Other conifers Cunninghamia lanceolata, Araucaria cunninghamii,
Araucaria angustifolia, Cupressus lusitanica, and others
Other hardwoods Gmelina, Terminalia, Albizzia, Leucaena, Grevillea,
Prosopis, Triplochiton, Meliaceae spp. (e.g., Swietenia,
Azadirachta), Paraserianthes, Cordia, Casuarina, and others
13.2 Plantations of Exotic Tree Species 215

as exotics. The relative plantation area of tropical pines is likely to decrease,

while several fast-growing hardwoods (e.g., Acacia spp., Gmelina arborea, several
Meliaceae) are gaining importance. Teak is the only plantation species planted
predominantly within its natural distribution range; however, significant exotic
plantations of teak also exist.
As Evans and Turnbull (2004, p. 11) noted, the distinction between forestry
and agricultural tree crops is not clear. For instance, rubber plantations are
classed as forests, while plantations of Acacia mearnsii that are also used for
bark contents are not. Numerous nonindustrial tree species are planted in
agroforestry, the mixture of tree-growing with food crops or livestock. These
species are grown for timber, oils, fruits, etc.

Plantations of Exotic Tree Species

When an exotic species is to be planted, the forester has the choice among
a very wide spectrum of species that have the advantage of being free of their
native herbivores and parasites. It may take some time until local insects or
fungal pathogens become adapted to the new food source, or until the native
pests and diseases are transferred.
At present, exotic tree species are of prime importance for plantation
forestry throughout the tropics. Managers can build on international experi-
ence with these well-known species that have been successfully used for plan-
tations worldwide. Pure plantations using only a single or a few commercial
species are the rule. Particularly various fast-growing, mainly exotic, pioneer
tree species such as eucalypts, acacias, Gmelina, and pines are grown in single-
species plantations. The plantations existing in more than 100 countries are
used for pulp, woodchips, sawtimber, fuel wood, and other products.
Owing to extremely short rotation periods, ecological risks are expected to
be low. If plantations are grown on suitable sites, the wood production is
extremely high. However, the demand for nutrients and water may also be
high. More and more such plantations are established with breeding products
(Chap. 12), which provides for sometimes spectacular rentability and makes
plantation forestry attractive. These plantations cannot simply be considered
to be substitutes of natural forest (Chap. 8 in Bruenig 1996). Instead of replac-
ing existing primary or secondary forest, they should be planted on deforested
land and remove pressure from natural forest.
The introduction of exotic species has frequently happened in several steps.
The widest distribution originated from locations far away from natural pop-
ulations. For example, both the coffee (Coffea arabica) and rubber (Hevea
brasiliensis) trees were originally planted as exotics in fairly small populations
216 C HAPTER 13 Genetic Aspects of Plantation Forestry in the Tropics

in botanical gardens in Europe. Large-scale exotic plantations of these crops

were later established throughout the tropics using material from only a few
plants growing in botanical gardens (Monaco 1977; pp. 22ff., 213 f. in Smith
et al. 1992). Botanical gardens have also played an important role in the artifi-
cial deployment of some forest trees.
The provenance or even the origin of reproductive material for exotic
plantations is often poorly documented. Maternally inherited markers strongly
suggest long-distance seed transfer prior to the establishment of Dalbergia
sissoo plantations in Nepal (Pandey et al. 2004; Example 2.2). Uncontrolled
transfer of seeds of forest trees even among continents has been practiced for
centuries. Small test plantations were established and reproductive material
was harvested in large quantities from those small test plantations or purely
incidental introductions, once the species turned out to be suitable for planta-
tion establishment in a region. Other species, such as Leucaena leucocephala in
tropical Asia, proved to be well adapted to the environmental conditions pre-
vailing at their numerous exotic planting sites. They regenerate naturally and
quickly disseminate their genetic information as colonizing species.
Mahogany (Swietenia spp.) has become one of the most important timber
species in the Philippines. Several documents suggest that no reproductive
material has ever been imported directly from tropical America. The first gen-
eration of mahogany on the Philippines was probably established with material
imported from a few exotic plantations in Sri Lanka.
The uncontrolled distribution of reproductive material and the small num-
ber of founder individuals that sometimes were even related have several genetic
implications. The planted material does not always originate from populations
most suitable for plantation establishment in a region. The distribution of
species and provenances by humans is more often an outcome of random events
rather than following a sound plan for the evaluation of genetic resources.
Inbreeding is a likely result of harvesting sexually produced progeny in small
trial plantations of a newly introduced exotic species and may further increase
in advanced generations. A decrease of vitality, viability, and growth observable
in the second and later generations is a likely consequence of inbreeding
depression (Example 13.1).
Another negative effect of small sizes of founding populations has been the
loss of genetic variation due to drift. Many exotic plantations exhibit reduced
levels of genetic variation as compared with levels in natural populations
because of random loss of genes. The genetic base of the plantations is said
to be narrow. Low levels of genetic variation impair processes of evolutionary
adaptation to changing environments. Populations with low genetic variation
are not suitable as base populations for breeding programs.
Many attempts to introduce plantation species have failed owing to envi-
ronmental mismatch, pests or diseases, and the disregard of some simple
principles of provenance choice.
13.2 Plantations of Exotic Tree Species 217

Example 13.1: The Origin of Early Plantations of Acacia mangium in Sabah

Acacia mangium is currently one of the preferred species for plantation estab-
lishment in the humid and subhumid tropics. The plantation area has rapidly
increased, in particular in Southeast Asia.
Acacia mangium was first introduced to Sabah (Malaysia) in 1967. The first
introduction was the offspring of a single open-pollinated seed tree in Australia.
Two small stands of 34 and approximately 300 trees, respectively, of its open-
pollinated progeny were planted. More than 15,000 ha of plantations was
established in Sabah during the period from 1971 to 1984. Seeds for the estab-
lishment of these plantations originated exclusively from these first plantations
and their advanced generations. Fructification of Acacia mangium is prolific a
few years after planting. Up to five consecutive generations of Acacia mangium
were produced until 1984. Not to forget, the ancestors of all these trees are the
initially introduced progeny.
A simple nursery trial was established comparing seedlings from the first to
third generations, a phenotypically selected second generation, and, for compar-
ison, a later-introduced provenance mix (Table 13.3). The growth of seedlings
harvested from the second and third generations was reduced compared with
that for the first generation and the provenance mix. Accumulating inbreeding
effects in advanced generations are a likely explanation for this result.
The seeds harvested from the first-generation stands must have been
affected by inbreeding. However, seedling growth was identical to that of the
provenance mix. A possible explanation is the generally low level of heterozy-
gosity of Acacia mangium (Table 3.1), leading to only moderate inbreeding
depression. Furthermore, the provenance mix was likely to contain some geno-
types that are not adapted to the conditions in the test.
The fastest-growing seedlings were harvested in a stand of the second gen-
eration which was transformed to a seed production area by phenotypic mass
selection. Sim (1984) explained this result by the development of a land race
particularly adapted to the conditions in Sabah, and by the efficiency of mass
selection. (From Sim 1984)

Table 13.3. Average height of Acacia mangium seedlings

in a nursery trial in Sabah. For further explanations see
the text. (Adapted from Sim 1984)
Seed harvest in stand Seedling height (cm)

1st generation 32.5

2nd generation 20.7
2nd generation, selected 35.2
3rd generation 18.1
Provenance mix 32.5
218 C HAPTER 13 Genetic Aspects of Plantation Forestry in the Tropics

Plantations of Indigenous Species
Given the threat to the forest gene pool, the promise of a renewed private planting pro-
gramme, and our ignorance of propagation techniques and growth potential of many
potentially important species, it is more than ever necessary to co-ordinate national
germplasm collection and breeding efforts on native hardwood trees . . . (Lawson 1994).
The species spectrum of indigenous species of a country is, of course, not as
broad as the worldwide spectrum of exotic species. Indigenous species still pos-
sess the advantage of being adapted to the regional climatic conditions. Local
consumers are familiar with the properties of their wood and the various non-
timber products that they yield. The utilization of species for plantation forestry
is frequently impeded by the low level of knowledge of their basic biological,
ecological, and silvicultural features. Evans and Turnbull (2004, Chap. 8) dis-
cussed arguments for the choice of a given species and presented examples of
indigenous species that have recently been used for plantations. Many of these
species are rare in natural ecosystems. Their use in plantations containing a few
or only one species involves a drastic change in ecological conditions, particu-
larly the pressure of infestation by fungi and insects.
The importance of indigenous species is still low in plantation forestry of the
tropics (Table 13.2). However, the anticipated increase of the plantation area is
expected to be based on a larger number of species, including indigenous species
and lesser-known species. An increased number of species used for plantation
forestry promotes the establishment of mixed plantations (FAO 1992).
The growth of species in natural forests of the region gives some indication
of their value for use. Subsequent species-elimination trials allow us to assess
whether a species is suitable for being grown in plantations. The observation of
a few trees, for example, in a botanical garden or an arboretum alone, is not a
sufficient basis for the identification of potentially important plantation species.
A test of the suitability of a species or population for plantation establish-
ment is only meaningful if the population from which the seeds were har-
vested is conserved at least until the end of the trial. This represents a genetic
resource as a necessary component of any program for the domestication of a
Procedures of seed procurement of exotics used during the early days of
tropical forest plantations may easily be avoided in indigenous species. This
provides for more genetic variation being retained in plantations. Nevertheless,
some basic principles of seed harvest and of experimental design must be fol-
lowed. Above all, seeds should never be harvested from isolated trees since
inbreeding is likely to occur and the genetic structures differ from those of nat-
ural populations.
13.3 Plantations of Indigenous Species 219

The most important exotic plantation species are characterized by their gre-
garious production of seeds. The copious small seeds of species such as pines,
eucalypts, acacias, and teak are easy to harvest in large quantities and to store
for several months or even years, i.e., the seeds are orthodox. This feature and
the ease of the production of seedlings in nurseries contributed to the impor-
tance of the species for plantation forestry.
Problems in seed procurement for afforestation projects are frequently
underestimated. It is difficult to produce and store suitable reproductive mate-
rial in sufficient quantities. This is due to irregular and sporadic seed produc-
tion and/or recalcitrant seeds, i.e., seeds which cannot be stored for long
periods. Other species have a seed dormancy that is difficult to break for large-
scale production of seedlings. The main problem with those species is frequently
not the choice of genetically suitable reproductive material but rather the pro-
duction of sufficiently large quantities of planting stock. Seed orchards (Sect.
12.8) help to solve some of these problems. Other help is seen in vegetative
Leakey et al. (1982) emphasized the potential use of numerous species for
plantation forestry in the tropics. They described the options for an improved
seed procurement and relevant research in this context using the example of
the West African species Triplochiton scleroxylon, a species with an erratic
pattern of reproduction. By adequate treatment and the use of appropriate
facilities, the storage period of seeds could be extended. Besides the develop-
ment of methods for controlled pollination, an integrated procedure allows for
vegetative propagation and the early selection of tree clones with good stem
form. Ladipo et al. (1991) reported further on this research.
In West Africa, where the discrepancy between annual deforestation and
total tree plantations is particularly wide, Lawson (1994) identified an urgent
need for domestication of indigenous tree species such as Terminalia ivorensis,
Terminalia superba, Triplochiton scleroxylon, and Lovoa trichilioides. The clonal
propagation option is preferred in view of the problems involved in the pro-
curement of seed of good quality, and the increased yield. This may be an effi-
cient measure where forests disappear fast. Milimo et al. (1994) listed species of
semiarid West Africa that are worthy of breeding. In view of various difficulties
involved in propagation by seed, these authors stressed the importance of veg-
etative propagation and reported on experimental results. In their comprehen-
sive review of domestication of the genus Swietenia and other mahoganies,
Newton et al. (1994) described the development of molecular genetics tech-
niques, provenance research, selection, and the development of both conven-
tional and in vitro propagation. On this basis, appropriate silvicultural systems
accounting for the Hypsipyla problem could be developed.
The worldwide use of eucalypts may create the impression that these are the
only important tree species in Australia. Nevertheless, projects in provenance
220 C HAPTER 13 Genetic Aspects of Plantation Forestry in the Tropics

research, breeding, and vegetative propagation of lesser-known species of

Casuarina, Melaleuca, Grevillea, and Sesbania are of importance also to countries
where these species are of potential use as exotics (Booth and Turnbull 1994).
Enrichment planting, underplanting valuable species in partially cleared
natural forest, is an efficient means of conserving the target species and of
introducing tested planting stock of the respective indigenous species (Lawson
1994). Research on domestication of agroforestry tree species is unfortunately
concentrated on a few exotic species, although the number of indigenous
species that are suitable for cultivation in this agricultural system is very large
(Sinclair et al. 1994).
As for the establishment of mixed forests, increased planting of indigenous,
frequently denoted as lesser-known, species is desirable. The quality of repro-
ductive material of lesser-known species is often poor: genetic differentiation
patterns are unknown and consequently no information is available concerning
the choice of optimum provenances for plantation establishment in a region
(Chap. 11). Also the physiological quality of harvested seeds is frequently poor.
Seeds are often infected by pathogens because they were picked up from the
ground rather than from standing trees or because they were stored using
inappropriate facilities (Kamra 1986).
Leakey and Newton (1994b) chose the name Cinderella tree species for these
species that have not contributed significantly to reafforestation, possibly because
they are rare in natural forests. Many of them yield also fruit, nuts, and other
nontimber products (Clement and Villachica 1994; Okafor and Lamb 1994;
Prance 1994). These species do not necessarily grow less fast than other species
(Maghembe et al. 1994). Many fruits and nuts have been collected from these
species in the wild but numerous species are also well suited for agroforestry.
Trees of indigenous species planted in agroforestry ecosystems do not necessar-
ily possess less variation than those in natural forests (Hollingsworth et al. 2005).

Basic and Reproductive Material

The success of artificial regeneration of forests critically depends on the choice

of suitable reproductive material. Seeds, seedlings produced in nurseries or col-
lected in forests (wildings), or vegetative propagules (e.g., stem or root cut-
tings) are reproductive material. Reproductive material must be adapted or
adaptable to the environmental conditions prevailing at the planting sites. The
objective is to establish a sufficient number of trees that are capable of surviv-
ing until the end of the rotation period.
Farmers attribute much weight to the choice of suitable varieties of agri-
cultural crops; however, foresters involved in the establishment of plantations
13.4 Basic and Reproductive Material 221

frequently show a surprising lack of interest in the selection of suitable repro-

ductive material. Once unsuitable material has been chosen, the choice cannot
be rectified later. The consequences of selecting inadequate reproductive
material become obvious only after a few years. Forestry is a low-input enter-
prise. It can hardly compete with tree plantations for producing export crops
such as oil palm, or rubber. Forest plantations are characterized by large
temporal and spatial heterogeneity of the environment. Thus, the identifica-
tion of optimum planting stock is of more crucial importance than in most
agricultural crops. The situation is aggravated by the still poor knowledge of
the genetic systems and genetic variation patterns of most tree species.
Detailed legal regulations on the marketing of forest reproductive material
exist in the member countries of the OECD (Nanson 2001) and the EU (2000),
i.e., mostly countries of the temperate and boreal zones. The forest owner is, of
course, free to produce reproductive material in his or her own forest in order
to meet his or her own demand. Comparable regulations do not exist or are of
only little practical importance in most tropical countries. Designated author-
ities control collection, processing, storage, seed testing, and marketing of
forest reproductive material. Genetic methods of certification are not included.
The legislation refers to forest reproductive material but contains very little
on the reproductive material itself. The bulk of the regulations deals with basic
material. The implicit assumption is that reproductive material possesses the
same genetic structure as the basic material that it was derived from (Hattemer
1987). Types of basic material are stands, seed orchards, and clones.
There are several categories of reproductive material. It may be classified as
selected if it was derived from basic material that meets criteria such as supe-
rior phenotypic quality, uniformity, and isolation. The stands must be officially
approved for seed collection. This accounts for the certainly most widely prac-
ticed method of selection in tree species, the selection of stands without prog-
eny testing. Foresters the world over collect seed for routine plantations in
stands selected for their superior phenotype. In the member countries of the
OECD and the EU this is even a legal precondition for the marketing of
the respective reproductive material. This method is not based on the estima-
tion of genetic parameters such as expected breeding progress. It has neither
been proven whether the plantations established with selected reproductive
material outperform other stands. Not even exemplary progeny tests have been
made in support of this procedure. For several reasons it is unlikely that this
procedure leads to appreciable, if any, breeding progress. The selection is done
at moderate intensity among stands scattered over large regions where the
effects of the environment and possible silvicultural treatment are unknown
but presumably large. According to what we know on the reproduction system
of stands, genetic equilibrium is at best realized approximately. Last but not
least, since in EU and OECD countries forest reproductive material to be put
222 C HAPTER 13 Genetic Aspects of Plantation Forestry in the Tropics

on the market is restricted to approved basic material, only about 1% of the area
covered by the main tree species contributes offspring to plantings in some
member countries. The area where seed is collected changes with time, so there
is some change in the collective of approved basic materials. The percentage of
the area covered by the respective species is nonetheless very small. However, the
given degree of isolation hardly prevents influx of effective pollen. If no addi-
tional cost is involved, seed may be collected in selected stands, although expec-
tations of far-higher value for use are not necessarily justified.
Source-identified reproductive material does not meet the criterion of being
derived from selected base material. It is assured to have been produced in a
given region of provenance. Qualified reproductive material must be produced
by parent trees or in a seed orchard the components of which were selected on
an individual basis. It may also be propagules of a clone or clonal mixture.
Tested material is not just derived from phenotypically superior basic mate-
rial but has been tested for improved value for use in experiments that meet
certain criteria of design. The improved value is defined as the superiority over
officially prescribed standard entries included in the experiments. Until the
present, only a very small fraction of reproductive material has been marketed
in this category.
The amounts of reproductive material collected in selected, qualified ,or
tested base materials are not restricted. The advantage of the legislation is cer-
tainly the concomitant documentation of the seed collection site, even if this
may not represent relevant information on reproductive material (Jones and
Burley 1973).

Production and Collection of Seed

Natural forests as a seed source may present problems for seed collection.
Particularly the collection from tall trees in less prolific species is laborious in
spite of modern equipment. Fruits of many species are dehiscent. Fruit and
seed parasitism and fruit-eating animals mitigate harvested seed. Melchior
(1986b) described problems related to seed procurement for three important
and promising South American tree species in Peru and Venezuela (Cedrelinga
catenaeformis, Juglans neotropica, Bombacopsis quinata).

Seed Production Areas

Seed production areas arise either by establishment of a stand with the objec-
tive of seed production or by phenotypically superior stands being converted
13.5 Production and Collection of Seed 223

to seed production areas by thinning. The main objective of the thinning is to

enhance pollen and seed production of the remaining trees. Since competing
trees belonging to other species are likely to impede pollen dispersal of the tar-
get species, they are removed just as phenotypically minor trees. This procedure
is similar to that leading to the approval of stands for seed collection in OECD
and EU member countries. If the converted stands were chosen according to
phenotypic criteria only, it is recommended including the progeny in prove-
nance tests in order to get some indication as to their merit. Also, the conversion
to seed production areas is recommended in regions that have been proved to
be suitable for seed procurement on the basis of the results of provenance trials.
Practical considerations regarding seed production areas were summarized by
Schmidt (1993c).
If stands possess some breeding status, they are really selected in the close
sense of the word and may be used for establishing seed production areas.
These stands may be used for the production of better-quality seed and may
only later be replaced by seed orchards or other more sophisticated techniques
for the deployment of a breeding product. Wellendorf and Kaosa-Ard (1988)
described the role of seed production areas in the breeding program for teak in
Thailand (Example 11.2).
Species-rich tropical natural forests with low density of the target species are
more difficult to convert but are still widely used as seed production areas. In
temperate conifers the density of seed production areas is of no or minor rele-
vance (El-Kassaby et al. 2003). Zheng and Ennos (1997) compared the out-
crossing rate in a seed production area with that of a natural population of
Pinus caribaea in a mixed stand. The two estimates were absolutely the same
but in the seed production area there were also indications of biparental
inbreeding. These authors made the interesting point that the milder form of
inbreeding implies more inbreeding depression in the plantations established
with this seed, since the offspring arising from self-fertilization would hardly
germinate at all. They recommended properly designed clonal seed orchards
that lack a family structure. In an orchard of this type neither form of inbreed-
ing was observed. In a seed production area of Eucalyptus citriodora in Brazil,
Yeh et al. (1983) found outcrossing rates around 0.82, i.e., markedly above
those in most natural stands of eucalypts (Table 6.2).

Provenance Resource Stands

Provenance resource stands are a particular type of seed production area. They
are frequently established simultaneously with the provenance trials, i.e.,
during an early stage of a breeding program. They represent a mixture of the
progeny of numerous seed trees of a single promising population and should
224 C HAPTER 13 Genetic Aspects of Plantation Forestry in the Tropics

be reproductively isolated. A main objective of such stands is the dynamic

ex situ conservation of genetic resources (Chap. 14). When matured, stands of
superior provenances are used for the production of seeds for plantations.
The genetic structure of the progeny also of provenance resource stands is not
necessarily identical to that of the trees representing a provenance in the field trial
but is similar to it. The cost is much less than that for establishing the field
trial, since no design is required. This type of stand is highly recommended but
rarely practiced. Nikles and Newton (1984) described their establishment with
several Pinus caribaea var. hondurensis provenances in Queensland, Australia.

Seed Orchards

Seed orchards are primarily established in order to produce large quantities of

seed from a limited number of selected trees (Sect. 12.8). Selected trees are
cloned and copies assembled to a propagation population. The trees may also
just have been scattered remnant individuals in degraded forest so that mating
contact can be reestablished. Once an orchard exists, seed collection is much
cheaper than anywhere else. In the present context, increased seed production
and greater ease of seed collection is of prime importance.
Not only grafting but also rooted cuttings may be used for the establishment
of clonal seed orchards, provided that they flower early at convenient height
above the ground. Seeds of B. quinata can easily be produced in clonal seed
orchards, since a simple method for the vegetative propagation of the species
was found. Melchior (1965) observed how the indigenous population erected
living fences. They used long and heavy branches rather than cuttings of the
size of a pencil. The resulting multisprout bushes flowered abundantly in the
second year. Just as in the crowns of adult trees, certain bees operated as nectar
thieves during the day.
The comparison of the genetic quality of seed produced in natural and
domesticated populations is complex and implies the risk of oversimplifica-
tion. El-Kassaby (1999) pointed out the genetic variation in orchard seed may
easily be higher than in the seed collected from a stand, if many clones were
derived from trees selected in several genetically differentiated populations.
Care is required when choosing the stands to be used as a reference. A well-
founded example was presented by Godt et al. (2001): 47 and 48 trees were
sampled in seven and five populations of Picea glauca and Pinus banksiana,
respectively, scattered over a vast area of forest in Canada. Forty and 31 seed
orchard clones selected in this area, respectively, showed only minor reductions
in variation assayed at 18 allozyme loci. Although numerous rare alleles were
not encountered in the seed orchards, the genetic distance was negligible.
13.5 Production and Collection of Seed 225

Plus trees are possibly older and more heterozygous at certain gene loci than
other trees in the same population (Bergmann and Ruetz 1991). They may have
more variable progeny. Comparisons of the inbreeding coefficient of the seed are
possibly less problematic. Since any comparisons are still subject to the vagaries
of sexual reproduction, repeated sampling of the seed output is required.
Also the less expensive seedling seed orchards are mostly linked to breeding
programs. Lee (2000) pointed to the aspect of pollination in seed orchards.
When studying the mating system of Dryobalanops aromatica, a dipterocarp
species with irregular flowering, he found an indication for a lack of pollina-
tors in a seed orchard being responsible for increased self-fertilization (see also
Sects. 10.2.1, 12.8.2). If a seed orchard does not consist of selected families to
be propagated under isolation, precautions may be taken for improving the
situation by locating orchards where they are surrounded by extensive forests
of dipterocarps or nondipterocarps providing pollinators. The appropriate
spatial configuration can help to minimize other consanguineous matings. The
incidence of the same species might also provide genes. Lee (2000) also warned
against placing seed orchards into other ecosystems in order to avoid undesir-
able molecular imprinting and maladaptation. Therefore, this author proposed
delineating planting zones on the basis of biological and ecological data and
establishing plantations only in their own planting zone. This procedure is
practiced in many countries of the northern temperate zone. It represents an
alternative to the principle of transfer rules as mentioned in Sect. 11.6.
The aspect of pollination was also discussed by Moncur et al. (1995). In seed
orchards of several eucalypt species these authors observed a considerable
improvement in both the amount and the quality of seed after setting up bee-
hives. The average outcrossing rate was raised from 0.76 to 0.91. In Europe,
beehives are also placed in remote stands of rare trees of the rose family in
order to increase the seed crop.
Also Moran et al. (1989b) and Chaix et al. (2003) (Sect. 12.8.2) reported
considerably lower rates of self-fertilization in seedling seed orchards than in
natural stands of eucalypts. The former authors emphasized the reduced
neighborhood inbreeding by less spatial clumping of relatives. In a seed
orchard containing both grafts and open-pollinated progeny of plus trees in
Douglas fir, Ritland and El-Kassaby (1985) found only little self-fertilization
but indications of consanguineous matings primarily due to the large and
varying size of the families. In a seedling seed orchard of Pinus merkusii Siregar
and Hattemer (2001) observed a relatively low outcrossing rate (0.87) with lit-
tle incidence of consanguineous biparental mating; however, during the
extended flower period the phenology differed between families and the indi-
vidual effective pollen clouds were genetically differentiated. In a far-distant
natural stand the outcrossing rate of 0.98 was similar to that of other pines and
226 C HAPTER 13 Genetic Aspects of Plantation Forestry in the Tropics

much higher than in natural stands of the species in Thailand (Changtragoon

and Finkeldey 1995a; Sect. 6.3.2).
If the orchard produces an excess of seed, collection should not be concen-
trated on part of the clones or families. Better, although more costly, is to collect
less seed from all trees so that more effective neighborhoods are represented in
the seed.

Collection and Storage of Seed

Seed is either collected off the trees or picked up from the ground. Seed collec-
tion by climbing is expensive and dangerous in spite of modern equipment.
The seed should be collected in an appropriate way so as to minimize losses of
genetic variants. The genetic implications of seed collection from trees were
discussed in Chap. 11. Too few trees might be available for seed collection from
the crowns of felled trees. Seeds of other species are caught in nets laid out on
the forest floor. If a restricted number of nets are laid out under highly prolific
trees only, a spatial genetic structure becomes manifest in the seed crop and
induces genetic change. In stands with a family structure this seed is expected
to be markedly different from seed picked up by hand from the total area of the
stand as was shown by Ziehe et al. (1998) in a study of beech (Fagus sylvatica).
The seed of many tropical trees such as Triplochiton scleroxylon or Araucaria
spp. is also collected from the ground.
Also when collecting wildings, their genetic variation should be accounted
for. A patchy distribution of saplings indicates that they might represent off-
spring of a restricted number of trees.
The recalcitrant seeds of many tropical species are sensitive to desiccation
(mangrove species, neem, Araucaria angustifolia and other Araucariaceae, most
dipterocarps) and are thus rather short-lived. Seeds of many tree species of the
temperate zones are orthodox and can be stored for a considerable time. The
seeds of most Acacia spp., Eucalyptus spp., and teak are orthodox. Seed treat-
ment such as the careful reduction of the moisture content and the storage in
a strictly controlled environment can be used to extend the life span of the
seeds of most species (Schmidt 2000).
Air-conditioned cold-storage rooms are used for the long-term storage of
seeds in gene banks. Optimum conditions for long-term storage depend on the
seed physiology of the species. Seeds of some tropical species retain germinabil-
ity even at temperatures below the freezing point of water and thus can be
frozen without damage. The life span of seeds of other species is longest at low
temperatures just above the freezing point of water. The humidity in storage
rooms is also controlled and usually reduced compared with that outside. More
13.7 Planting Stock Production 227

details on optimum storage conditions for the seeds of tropical trees have been
summarized, for example, by Albrecht (1993), Tompsett (1994), and Schmidt
(2000, p. 223 ff.).
The germinability of many recalcitrant seeds cannot be extended to more
than a few months. The principal distinction between orthodox and recalcitrant
seeds is maintained even under optimum storage conditions. The long-term
storage of seeds is not an option for the conservation of genetic resources of
most species with recalcitrant seeds.
In any event, the loss of germinability represents a reduction of the popula-
tion size. As some studies in temperate tree species have shown, this reduction
is also connected with selection (Melchior 1986a; p. 340 f. in Hattemer et al.
1993; El-Kassaby 1999). The selection going on during germination and nurs-
ery operations rids the planting stock of adaptively weak individuals arising
from inbreeding. Seeds of certain families survive for different periods, so the
variation decays during storage. It is an open question whether this also implies
adaptation to the storage conditions. Evans and Turnbull (2004, p. 134),
Schmidt (2000), and Midgley (1996) reported new developments in seed col-
lection and seed technology of tropical trees. Research on seed germination of
fruit-bearing trees is under way (Maghembe et al. 1994).

Planting Stock Production

Artificial regeneration does not rely on the natural seed fall and the concomi-
tant seed dispersal. Seeds are collected off the trees, processed, and eventually
stored. The seed is either sown out directly or planting stock is raised in a nurs-
ery or greenhouse under intensive treatment so that mortality and thus the
chance of adaptive viability selection are minimized. Outplanting or seeding is
hardly ever done on the site of the parent stand. The implied habitat change
interrupts the long-term process of adaptation and enforces selection with
changed intensity and direction. The emerging artificial stand no longer has a
family structure.


The seeds are germinated in nurseries or greenhouses under favorable condi-

tions in order to achieve the maximum planting stock. The prevention of mor-
tality implies that only little viability selection can take place at this early stage
of the life span. In tree species of the northern temperate zone, it was shown
that the early viability selection represents an important element of the genetic
228 C HAPTER 13 Genetic Aspects of Plantation Forestry in the Tropics

system of trees (Ziehe et al. 1999). This type of selection is then postponed to
the period after outplanting.
Also grading of plant material has a genetic implication. Under highly favor-
able conditions of planting stock production, Eriksson and Lindgren (1975)
found less inbreeding depression in both Picea abies and Pinus sylvestris and
inferred that there may be more inbred offspring among tall plants. In experi-
ments done by Konnert and Schmidt (1996), the tallest plants in planting stock
of Picea abies turned out to possess the lowest degree of heterozygositya prop-
erty that is otherwise an advantage for field survival. In European white fir the
result was the reverse. In both species, the taller and shorter plants were genet-
ically differentiated.

Clonal Multiplication

Plants are regenerated directly from propagules without generative processes

being involved (Sect. 4.3). The complexities of the natural systems of sexual
reproduction in tropical trees (Sect. 4.2) are avoided. So are problems involved
in parts of the production of planting stock. Consequently, ever more tropical
plantations are established with clonal material. Some authors (Felker 1994)
advocate clonal propagation for the sole reason of the self-sterility of a species
and the implied difficulty in capturing the genotype of individuals with desir-
able properties in sexual progeny.
Numerous techniques are used for the vegetative propagation of tropical
forest trees. Smits et al. (1994) have reported on the production of planting
stock by cutting propagation of dipterocarps, a family with erratic flowering,
recalcitrant seed, and much seed parasitism. The different techniques for
macrovegetative and microvegetative propagation of forest trees are not part of
forest genetics research and cannot be discussed in this introduction.
Laboratories and culture rooms for microvegetative propagation are not avail-
able everywhere.
Besides the capture of breeding progress (Sect. 12.8.3) the use of vegetative
propagation techniques solves problems in seed supply. The selection of supe-
rior clones for plantation establishment is often of only secondary importance
(cf. the example of Triplochiton scleroxylon described by Leakey et al. 1982; Sect.
13.3). The erratic flowering and seed production of dipterocarps of the humid
tropics was the principal motivation for the development of vegetative propa-
gation techniques for this important family (Smits 1993).
Some tropical tree species are potentially important plantation species
owing to certain silvicultural features or wood properties; however, seeds are
available only sporadically and/or in insufficient quantities. The seeds of these
species are often recalcitrant, and the production of seedlings is sometimes
13.8 Establishment and Development of Plantations 229

difficult. The inevitable loss of genetic variation associated with the shift from
sexual to asexual propagation techniques is accepted in view of the pragmatic
advantages of vegetative multiplication techniques such as the uniformity of
the plantations. Seed production of other species such as most eucalypts is
prolific, seeds are orthodox, and sexual progenies, i.e., seedlings, are easily pro-
duced using standard nursery techniques. In these species asexual propagation
is more laborious and expensive. Vegetative propagation is applied only for the
mass multiplication of certain selected trees. The term clonal forestry is asso-
ciated with an anticipated increase of the yield by planting these selected clones.
The risks associated with the unavoidable reduction of genetic variation in
clonal plantations occur, of course, independently from the motivation for the
use of vegetative propagation techniques. The often-neglected genetic implica-
tions of clonal plantations are addressed in Sect. 13.9.
However, vegetatively propagated plants are not necessarily superior to gen-
eratively produced progenies. For example, the growth of vegetatively propa-
gated E. grandis plants was compared with that of seedlings from the same trees
after open pollination in Brazil. For a variety of reasons, the growth of the
seedlings was superior to that of the vegetative offspring 30 months after plant-
ing (Kageyama and Kikuti 1989).

Establishment and Development of Plantations

From the genetic point of view the choice of the reproductive material is the
most important decision before plantation establishment; however, genetic
structures of artificially regenerated forest tree populations may change in time
during the reduction of the number of individuals owing to viability selection.
A temporal dynamics of genetic structures was observed in experimental popu-
lations of some temperate forest trees during early development of the seedlings
(Kim 1985; Ziehe et al. 1999). In order to provide ample room for adaptive
change, large numbers of individuals should be planted (Langner 1966; Ledig
and Kitzmiller 1992; Mller-Starck 1996). In rotation forestry with short rota-
tion periods, this safeguard is of less importance than in long-lived tree species.
Forest reproductive material can be transferred over large distances. The
capacity of artificial migration is obviously unlimited. The ecological gradients
overstepped by this transfer matter greatly. Sometimes reproductive material is
totally ill adapted. Uncritical use of just any reproductive material has long
been and still is hazardous to forestry; therefore, the careful choice of popula-
tions must be integral part of adequate forest management.
The number of plants per area or tree spacing is an important aspect of plant-
ing. In tropical plantations mostly fewer than 1,000 trees are planted depending
230 C HAPTER 13 Genetic Aspects of Plantation Forestry in the Tropics

on the purpose of the plantation. The potential for selective change of genetic
structures and the purging of physiologically ill adapted individuals depends on
the genetic variation of the planted population, the number of trees planted, and
subsequent mortality. Evolutionary adaptations (Sect. 7.2) avoiding critical
population sizes are more likely in plantations established with a large number
of different genotypes planted per unit area. No evolutionary adaptation on the
basis of viability selection is possible if only a single clone is planted.
The temporal dynamics of the genetic structures of a population is not only
influenced by natural mortality. Silvicultural operations and particularly selec-
tive thinning have also the potential to change genetic structures. Different
thinning methods affect genetic structures in different ways as was shown for
Norway spruce (Picea abies) by Hosius (1993). Comparable experimental stud-
ies for plantations of tropical forest trees are not yet available.
The temporal dynamics of genetic structures in planted forests deserves par-
ticular attention if it is planned to harvest reproductive material in a plantation
either after its conversion to a seed production area or if it was planned to be nat-
urally regenerated. Changes of genetic structures in production populations that
will never give rise to a subsequent generation are of no evolutionary significance
although they may have an impact on economically relevant trait expressions.

Natural Regeneration of Plantations

If an indigenous or introduced plantation species turns out to grow obviously

well and to be adapted to the new site and sets seed freely, it can be considered
naturalized (p. 11 in Evans and Turnbull 20041). Even after natural regener-
ation it is still a man-made forest. It may be on the way to becoming a land race,
since at least a whole generation cycle has been closed under field conditions
on the new site.
Land races with particular morphological and physiological trait expres-
sions have possibly evolved from exotic plantations within a few generations
(Sect. 11.2). The genetic differentiation of land races from natural populations
is likely to increase the adaptedness to environmental conditions prevailing at
the new planting sites; however, genetic differentiation between exotic and nat-
ural populations may also be due to genetic drift in the exotic plantations and
thus a consequence of evolutionary change that is not necessarily adaptive.
At the time of planting it may be left open whether a plantation might be
regenerated naturally in future. Plantations are not expected to have a spatial
genetic structure except through selection (Eldridge et al. 1994). The density
should not fall short of a critical lower limit below which pollination is incom-
plete and outcrossing is reduced in favor of self-fertilization. The advantage of
13.10 Use of Breeding Products 231

the natural seed fall is the larger number of seeds and seedlings produced dur-
ing the regeneration period comprising several years. In ever-more countries a
mosaic of autochthonous and allochthonus stands exists, so the incidence of
hybridization between trees from stands of different origin must be expected to
increase. Particular features of natural regeneration were dealt with in Sect. 10.3.

Use of Breeding Products

All plantations are perceived to be in danger to some degree. They are threat-
ened by a massive buildup of pests and diseases (p. 387 in Evans and Turnbull
2004). Examples of devastating outbreaks have been reported from the tropics,
the temperate zone, and the boreal zone. Plantations are still of growing impor-
tance. They are the only means of deploying breeding products and obtaining
return from earlier investment. It was pointed out in Sect. 12.11 that the col-
lection of unrestricted quantities of seed in a given seed orchard or seed pro-
duction area implies that the progeny of the respective trees cover an area
exceeding that of the collection stand by a factor of several hundred. This is the
experience in countries with official approval of seed collection stands and
provenance recommendations. It is more costly to establish many orchards and
many seed production areas and to collect only moderate amounts of seeds
there; however, it is more in line with sustainability principles.
Even the selection of provenances is a form of breeding that usually precedes
more intensive selection steps. The large-scale planting of introduced species,
provenances, or other breeding products counteracts the preservation of locally
adapted populations that may contain elsewhere rare alleles or even unique alle-
les or allele combinations in high frequencies. Thus, important genetic resources
for future adaptation processes and breeding programs can get lost. The loss of
local genetic resources can be either direct by substitution or indirect through
gene flow from breeding products and other allochthonous populations into
surrounding natural populations of the same species.
The dangers associated with the loss of locally adapted populations through
their substitution by a few high-yielding varieties were first recognized for cul-
tivated species. They have been the main motivation for the establishment of
gene banks of many agricultural crop species. Tree plantations using breeding
products currently cover only a small part of tropical forests. The danger of the
loss of locally adapted forest tree populations is still much less severe than
damage caused by forest destruction.
Reduced genetic variation is a consequence of most breeding programs. This
increases the risk of severe attacks by insects and diseases that may rapidly
adapt to a homogeneous source of nutrients offered to them during several
232 C HAPTER 13 Genetic Aspects of Plantation Forestry in the Tropics

generations. A well-known example is the destruction of most plantations of

Leucaena leucocephala by cicadas of the genus Heteropsylla. The attacks were
particularly severe in tropical Asia during the late 1980s and virtually termi-
nated the use of the species for industrial plantation establishment (p. 323 in
Smith et al. 1992). The calamity was promoted by the rapid large-scale expan-
sion of plantations using a few varieties of giant leucaena. Recent breeding
work in Leucaena is directed towards improving the resistance against
Heteropsylla. This requires the test of genetically more variable Leucaena leuco-
cephala varieties and hybridisztion with related species.
Extraordinarily low amounts of genetic variation are also the likely cause for
the destruction of Paulownia taiwaniana plantations in Taiwan. Paulownia tai-
waniana plantations were established on a large scale during the 1970s. Just like
other species of this genus, Paulownia taiwaniana is mainly propagated by root
cuttings and other forms of vegetative propagation. Genetic variation within
and among these plantations was extremely low (Finkeldey 1992). In the late
1970s almost all plantations were severely attacked by a witches-broom disease
caused by mycoplasma-like organisms.
There exists empirical evidence that long-term risks can be raised by breed-
ing for resistance to present threats. By breeding and propagation, the number
of genotypes required for buffering the population against unknown future
threats is likely to be reduced. The consequence would be increased instability
owing to recurrent artificial selection (Namkoong 1999).
The risk of a large-scale planting of breeding products with reduced genetic
variation is particularly high in clonal forestry; however, Park et al. (1998)
pointed out that the diversity of progeny from intensively rogued seed orchards
is not too different from that of clones that were derived from intensive selec-
tion in families produced by certain regular mating schemes. The production
of genetically identical trees either by conventional cutting propagation or
advanced biotechnology such as somatic embryogenesis and their planting
offer chances and bear risks that must be balanced by appropriate management
systems. Clones occur also in nature as the result of adaptation. Also, the culti-
vation of clones in Cryptomeria japonica, Cunninghamia lanceolata, and other
conifers has a long history in eastern Asia.
Further to Sect. 12.8.3, some guidelines are presented that may help to alle-
viate the problems associated with the large-scale planting of clones or other
material of low genetic variation. They are admittedly based on hypotheses and

In view of increased vulnerability of less variable plantations as production

populations, the use of clonal mixtures is recommended. Hundreds of
hectares planted with a single clone are all but desirable. The variation in a
mixture of given clones is maximized if all clones are represented in equal
13.10 Use of Breeding Products 233

proportions. This need not imply loss of uniformity of the wood and tim-
ber produced. The effect of mixture is the larger, the more the clones are
genetically differentiated from each other at adaptive gene loci or in resist-
ance. This quality of the mixture is more likely to be achieved if the clones
are of diverse descent. This spreads the risk in view of adverse environ-
mental factors and retards contagion during infestation by diseases and by
defoliating or other insects.
The question of the necessary number of components in mixtures has been
tackled by several authors. For instance, Hhn (1986) considered the risk
of losing the plantation through the loss of an unacceptable fraction q of
trees prior to final harvest. The risk is a function of this fraction, the num-
ber n of clones, the susceptibility x, and the intensity a of contagion. For
given q and a the risk decreases with increasing n and decreasing x. There
are also special cases of x where the risk increases with increasing n. In
many situations the risk function is asymptotic for larger n. Increasing this
n beyond 3040 equally represented clones is of little effect. Libby (1982)
used a binomial approach for constructing the risk function and recom-
mended between seven and 25 clones.

The genetic control of susceptibility was not considered in these models. In the
approach taken by Roberds et al. (1990) a single agent active in the plantation
and a single biallelic locus controlling resistance was assumed; hence, the fre-
quency of either a dominant or a recessive allele for susceptibility could be inte-
grated. These authors also assumed a particular mode of selecting given
numbers of clones from progenies of the base population for propagation and
found rather complex behavior of the risk function. It depends on the mode of
gene action, since under dominance the proportion of susceptible genotypes is
always larger. It also depends on the intensity of infestation whether increasing
numbers of clones reduce the risk. If the intensity of pest attack increases, the
number of clones necessary decreases. Simulation studies by Bishir and Roberds
(1999) revealed that situations requiring more than 40 clones are not prevalent.
Campinhos (1999) reported the use of 80 clones per production population in
a given cycle of clonal mass cultivation in Aracruz (Example 12.1). Several EU
and OECD countries have issued regulations on minimum numbers.
One of the strong simplifying assumptions for the validity of these models is
the static nature of susceptibility, i.e., the clones are either susceptible or not and
this property does not change during the single rotation period considered.
However, it is our experience that insects or pathogens, organisms with shorter
generation cycles and more flexible genetic systems, evolve mechanisms to
exploit the homogeneous substrate offered by genetically uniform plantations
for years. The clonal aggregates are in turn hardly able to develop evolutionary
adaptation in response to these attacks. This condition introduces uncertainty.
234 C HAPTER 13 Genetic Aspects of Plantation Forestry in the Tropics

Consequently, if biotic agents are concerned, the problem of damaging

agents cannot be solved in the long term by planting resistant clones. The
evolving nature of the agents themselves cannot be incorporated into breeding
trees for immediate resistance. It is not known whether the resistance breeder
is always able to keep a step ahead or whether coevolving organisms with a
more flexible genetic system are likely to take over (Namkoong 1999). It is still
open to debate whether stability of hostpathogen and hostherbivore rela-
tionships can be achieved under mass cultivation of resistant clones.

Planting of clones and sexually produced offspring in intimate mixture or

on the scale of medium-sized plots helps to increase overall variation. Since
planting the clones in intimate mixture may still interfere with manage-
ment and industrial use, their arrangement in a mosaic of monoclonal
blocks is the preferred layout in Aracruz. In view of the dispersal of
pathogens and insects, the establishment of such a mosaic in large con-
nected tracts of plantations may be of minor relevance for contagion. The
question of intimate mixture or monoclonal blocks does, of course, have
relevance for yield and management (Foster and Bertolucci 1994).
The use of clones might be restricted to fractions of the total forest area of
a region. Campinhos (1999) reported on large tracks of native vegetation
in-between clonal eucalypt plantations in Brazil (Example 12.1).
Different performance of clones under different environmental conditions
calls for the mass cultivation of different clonal mixtures in different regions
as practiced by Aracruz Florestal. This procedure also increases total diversity.
Shorter rotation cycles shorten the time available for herbivore and para-
site populations to become genetically adapted to particular clones, unless
the plantation is coppiced or the area is replanted by the same clonal aggre-
gate. Short rotation periods are the great advantage of tropical tree planta-
tions if compared with trees of other climatic zones.
Many genotypes should be preserved in large clone banks (clonal archive),
so that the spectrum of the mass-cultivated clones can be changed perma-
nently. It is crucial to keep track of the identities of the clones deployed by
means of highly variable genetic markers.
The absolute maximum number of copies of a single clone ever deployed
should be limited. In Sweden this number has been limited by law. On the
other hand, the owners of small estates tend to propagate trees on their
neighbors fields, making their choice on grounds of good growth, thereby
selecting for rootability. This is hardly possible to change by legislation but
leads to widespread deployment of few clones if not even a single clone.

Risk considerations are not intended to question or even disparage the mag-
nificent success of clone breeding in the tropics and subtropics. Some of the
13.11 Recommended Literature 235

worlds most productive tree plantations, primarily poplars and eucalypts, use
clones. Crucial advantages of clonal forestry are the independence of seed
shortages, the uniform stands, and the ease of their management. Problems are
seen in the deployment of clones. The concern is motivated by the growing
need for intensive measures of plant protection in agricultural crop plants, not
least in perennial crops such as avocado, banana, citrus, coffee, coconut,
mango, oil palm, papaya, rubber, tea, and vine. Sooner or later in all plant cul-
tures with reduced genetic variation a slow but steady increase in the incidence
of insects and diseases has to be envisaged and the pressure on tree plantations
will increase. Since forest trees are not a cash crop, intensive chemical control of
pests and diseases rules itself out for economic, not to speak for ecological rea-
sons; therefore, the matter is of importance both economically and environ-
mentally. It is therefore preferable to rely on prevention. Foster and Bertolucci
(1994) have discussed several independent approaches to the maintenance of
some level of genetic variation in large-scale clonal plantations.
For the time being, intensive monitoring is advised and is done (Campinhos
1999). The economic returns from highly productive plantations allow for this
active monitoring of pests and diseases (Evans and Turnbull 2004).
Repeated reference made to the clonal program of Aracruz Florestal should
not be perceived as hidden criticism. This program is under way in the most
widespread exotic tree genus and is also the most advanced and successful. In
1984, Leopoldo G. Brandao, Edgard Campinhos Jr., Ney M. Dos Santos, and
Yara K. Ikemori were awarded the Marcus Wallenberg Prize for their pioneer-
ing work leading to significant scientific and technological breakthroughs in
developing commercial forests based on cloned Eucalyptus. Their methods
have stimulated world-wide emulation. The high productivities achieved will
reduce pressures on natural forests.

Recommended Literature

The book by Evans and Turnbull (2004) covers all aspects of tropical tree plan-
tations. It represents comprehensive information on this subject and in Part III,
devoted to plantation silviculture, gives answers related to genetics that can
only be touched on in the present publication. Subjects related to genetics are
also covered in the book by Zobel et al. (1987).
Conservation of Genetic Resources 14
in Tropical Forest Trees

There is ample evidence that genetic impoverishment of many tree species has
occurred and is occurring as a result of the destruction of forest ecosystems of which
these species are a part. Furthermore, it is certain that in many countries forest genet-
ics and tree improvement programs will not be operative in time and on a sufficient
scale to insure the preservation in clone banks and seed orchards of an adequate sam-
ple of the spectrum of variation within the important tree species. Numerous foresters
are aware of this situation, and of the need for the conservation of forest gene resources
by insurance of the survival of the forest ecosystems in which the species are found, and
by the establishment of seed banks, clone banks, and seed orchards. Furthermore,
united action is occurring at an international level that has a direct bearing on this
problem (p. 274 in Stern and Roche 1974).


About one quarter of the earths land surface has been converted to food
production. Only 10% of the original landscapes is estimated to have
remained unchanged. This condition and other forms of human impact on
plant and animal populations have led to the conservation imperative
accepted by ever-more people. The importance of the conservation of bio-
logical variation is undisputed. Genetic variation is a crucial component of
biological diversity, since it is a necessary precondition for the continuation
of biological evolution.
Furthermore, genetic variation within populations is a preconditon for het-
erozygosity, which according to empirical studies is correlated to fitness at the
population level. Losses of heterozygostity have a deleterious effect on fitness
(Reed and Frankham 2003). Although this correlation depends on the methods
of assessment used and is only moderately strict, it can be considered a second
argument for the commitment to genetic conservation as maintaining the basis
of both the long-term stability and the short-term productivity of forest
238 C HAPTER 14 Conservation of Genetic Resources in Tropical Forest Trees

ecosystems. Genetic conservation is clearly a strong contributor to the conser-

vation of general biodiversity and according to Altukhov (2006) implies pre-
serving genetic diversity of population systems that still exist but are
degraded by utilization (e.g., logging) and artificial propagation (nonexhaust-
ing management of natural resources), restoring systems whose structures have
already been damaged, and creating new population systems in regions with
appropriate natural, historical, and economic conditions. It is self-evident that
conservation genetics has a population genetics basis.
Practical recommendations for the conservation of genetic resources are
based on the understanding that only certain types of genetic information
qualify as a genetic resource. A gene resource is defined as the biological mate-
rial either known to or expected to contain either specific or extensively vari-
able genetic information (Ziehe et al. 1989). Therefore, the ultimate target of
genetic conservation is neither an individual tree nor a taxonomic unit but the
variation of nucleotide sequences in a genetic resource. Genetic conservation is
understood as the preservation of genetic resources in a condition allowing for
their regeneration.
The need for the conservation of plant genetic resources was first recog-
nized for agricultural species. The Russian plant geneticist N.I. Vavilov
(18871943) identified centers of diversity possessing a wealth of varieties
(genes) of cultivated plants during the 1920s and 1930s (Vavilov 1992, 1996;
Altukhov 2006). Tropical forests are centers of diversity of wild species.
Namkoong et al. (2002) and Boyle (2000) analyzed the rationale of conserv-
ing forest genetic resources and developed a system of making pertinent deci-
sions. The discussion for a conservation strategy in this chapter is based on the
sequence of operations as suggested by Ziehe et al. (1989). The concept has
been previously presented by Finkeldey and Hattemer (1993), Hattemer
(1995, 1997), and Finkeldey (1996).
There exists a wide range of motivations for conserving forest genetic
resources. Some authors committed to resource conservation of widespread
forest trees of the northern temperate zone feel it is their general responsibil-
ity to maintain variable and adaptable tree populations. To other authors
actively involved in the domestication of tropical forest trees, resource conser-
vation is primarily an integral element of their efforts in domestication
(Leakey and Newton 1994b). Domestication of trees implies that they are
assigned higher value. Different genetic systems of trees and different goals
pursued in different parts of the world call for a variety of measures to be
Common species are usually not endangered. Many rare (endemic) species
are threatened, particularly those occurring with a unique population. An
increasing number of tropical forest tree species are at the fringe of extinction.
Their loss would be final, since extinction is forever.
14.2 Development of Tree-Conservation Genetics 239

Once certain priorities (Sect. 14.3) have been agreed upon, a conservation
project for a given tree species starts with the clear identification of the objec-
tives (Sect. 14.4). Genetic resources are subsequently selected mainly on the
basis of the available knowledge of spatial patterns of genetic variation (Sect.
14.5). The choice of the method of physical preservation of the genetic infor-
mation in the selected organisms (Sect. 14.6) is related to the final step, the
regeneration of the resource (Sect. 14.7).

Development of Tree-Conservation Genetics

Since 1960s, programs for the genetic conservation of a few species have been ini-
tiated in the framework of planned or ongoing breeding programs. Conservation
of forest genetic resources was considered an item of breeding strategies also in
more recent publications (Cossalter 1989; Eldridge 1990). However, the signifi-
cance of conserving genetic resources of tropical forest tree species not (yet)
incorporated in breeding programs was also realized (Kemp et al. 1976).
Scientific results and practical guidelines for the conservation of forest genetic
resources were first summarized in a report edited by Roche (1975).
During the 1980s, conservation biology emerged to provide the underlying
science needed to slow down the process of extinction by conserving biolog-
ical diversity at all levels, including the genetic diversity within species. This
crisis-oriented discipline (Soul 1991) builds on the fields of ecology, biogeog-
raphy, and genetics applied to small and declining populations. The crucial role
of tropical forests for the global conservation of biological diversity has been
repeatedly pointed out by conservation biologists (Myers 1986). Genetic
resources of forest trees are endangered for several reasons, including forest
destruction and fragmentation, global climate change, inappropriate manage-
ment practices, selective logging in natural forests, the conversion of species-
rich forests to plantations of low diversity, and inappropriate methods of
breeding. The destruction of tropical forests due to unsustainable management,
shifting cultivation, and the conversion to other land uses, for example, for graz-
ing, are regarded as the most serious threat to biological diversity on a global
scale. Estimates of the annual loss of tropical forest area are alarmingly high
(Chap. 10.8 in Whitmore 1998; FAO 2002). The extinction of tree species as a
result of forest destruction is poorly documented (Sayer and Whitmore 1990).
The causes of the destruction of tropical forests by humans are manifold,
complex, and closely related to the demographic and socioeconomic develop-
ment of human societies in tropical countries. Measures for the conservation
of biological diversity will be successful only if attention is paid to the socio-
economic situation of the local human population as the main target group for
240 C HAPTER 14 Conservation of Genetic Resources in Tropical Forest Trees

long-term beneficial effects of conservation programs. Local people must

receive compensation for the loss of their utilization rights in protected forests
and should possibly benefit from a conservation project on a community basis.
Conservation of genetic resources and utilization of forest products are not
incompatible. The objective of genetic conservation is to preserve genetic
information rather than just any material of a species. The individual carriers
of genetic information, the trees, are transient and may be harvested as long as
their genetic information itself is conserved. Thus, the establishment of
reserves in situ does not rule out the management of the respective forest,
including harvest of forest products (Cossalter 1989). Intensive production
systems of low biological diversity such as monocultures, including clonal
plantations, may also have potential benefits for the conservation of genetic
resources. When genetically diverse forests are conserved in a given region,
urgently needed products can be produced elsewhere (Gladstone and Ledig
1990). The ability of tree populations to evolve is a foundation stone of sus-
tained forest management (Hattemer and Melchior 1993; Boyle 2000).

Defining Priorities

Owing to their rarity or importance, certain types of genetic information are

in more urgent need of conservation than others. Biodiversity conservation
targets exist at three independent levels: ecosystems, species, and genes
(Sherwin and Moritz 2000). The choice of priority targets for conservation
programs is based on the ecological and/or economic importance of popula-
tions, species, or species groups and the potential threats to their gene pools
and their hierarchical, historically formed structures (p. 371 ff. in Altukhov
2006). Present thinking aims at the conservation of species and concentrates on
efforts to conserve as many of the huge number of endemic species in tropical
forests as possible (Andersen et al. 1997). A brief description of important and
particularly endangered trees and shrubs has been published by the FAO
(1986). This list is periodically updated by the Panel of Experts on Forest
Genetic Resources of the FAO. In the present text the emphasis is on the con-
servation of genetic variation within the taxa. Conservation measures for a
species are initiated because of its economic and silvicultural importance or its
ecological significance as a keystone species for the stability of an ecosystem.
In view of the species richness in tropical forests, intensive conservation pro-
grams can be conducted only for a certain proportion of species. The intensity
of a conservation program must be adjusted to the importance of a species
(Namkoong 1986). An intensive genetic management including co-ordinated
measures for the evaluation, conservation, and utilization of genetic resources
is primarily possible for the commercially important target species. A multiple-
14.4 Conservation Objectives 241

population breeding strategy combining breeding goals and the ex situ conser-
vation of genetic resources is particularly recommended for plantation tree
species of the tropics (Sect. 12.9).
The principal objective of a genetic management for the vast number of
tree species without immediate commercial importance is the preservation of
their genetic variation. This requires an improved understanding of the
genetic system of at least some keystone or indicator species. Most rare and
threatened species exist as small isolated populations. It is urgent that these
species be conserved. Viana et al. (1997) listed some nongenetic indicators of
vulnerable species that conservation efforts should be targeted at. Bierregaard
et al. (1997) extended this to a brief discussion of important aspects of frag-
mented tropical ecosystems and relevant research priorities. Species that were
once common approach this condition and become increasingly fragmented;
therefore, considerate handling of important features of the genetic system
and of sufficient population size (Sect. 9.3) is devised for any tree species
exposed to human interference (Young et al. 1999; A. Young et al. 2000). That
their populations are eventually harbored in small forest fragments might
help managers to adopt a holistic approach rather than limit their attention to
large forest tracts or simply the most pristine areas (p. 279 in Laurance and
Bierregaard 1997b).

Conservation Objectives

The objectives of a genetic conservation program depend on the target species

and their ecological and/or economic importance. The objective of any con-
servation measure should be clear from the beginning (Eriksson et al. 1993).
The following objectives are similar to those distinguished by Ledig (1986a).

Objective 1: Preservation of the Potential for Particular Trait Expressions

The preservation of the genetic information controlling superior expressions

of economically relevant traits supports breeding projects. It involves uncer-
tainties, since quantitative traits are controlled by multiple gene loci, the
expression of which is modified by environmental conditions. During drastic
environmental change due to management or climate change, the mainte-
nance of trait expressions themselves is not certain even if the genetic struc-
ture of a population is preserved unchanged. In view of the predicted
environmental change on a local, regional, and global scale, the universal
significance of this objective is therefore questionable, although the objective
is widely practiced.
242 C HAPTER 14 Conservation of Genetic Resources in Tropical Forest Trees

Objective 2: Preservation of Maximum Variation

The conservation of genetic multiplicity is the main objective of gene conser-

vation of agricultural crop plants in gene banks. Even a single copy (allele) at
a gene locus is of potential value for breeding. Genetic diversity is considered
a value in itself. An important allele can be incorporated into breeding popu-
lations by a variety of mating designs and can thus be multiplied indefinitely.
In its strict sense the term gene conservation applies to conserving a maxi-
mum number of alleles in a genetic resource. According to Marshall and
Brown (1975), genetic conservationists pursuing this goal are primarily inter-
ested in preserving at least one copy of each of the different alleles of the
target species and concentrate on the preservation of maximum genetic mul-
tiplicity in the totality of the genome comprising genes for both production
and adaptation. The preservation of specific rare or even unique genes is
expected to gain further importance in view of molecular methods for the
transfer of single genes. The transfer of single genes from widely separated
taxa to trees has become routine in modern molecular genetics laboratories
(Rautner 2001).
For plant species, including forest trees, that are not as intensely cultivated
as agricultural crop plants the preservation of a maximum number of alleles is
less important. Very rare or even unique alleles do not enhance the capacity of
populations to survive in a particular environment or after rapid environmen-
tal change (Finkeldey 1993). Furthermore, methods for the incorporation of
single rare alleles into breeding populations of forest trees will hardly become
available in the foreseeable future.
A prominent example of this objective is genetic incompatibility condi-
tioned by an S-locus (Young et al. 1999; A.G. Young et al. 2000). Allele loss at
an S-locus represents an immediate threat to population viability. For instance,
assuming random pollination in a population possessing a gametophytic sys-
tem of incompatibility (Sect. 6.5.1) with n equally frequent alleles, a fraction of
2/n of the pollen is expected to be eliminated in the styles. The term self -
incompatibility, is misleading, because also a fraction of cross-pollen gets
lost and for small n this fraction may prevail over the pollen involved in self-
pollination. This means severe mate limitation and a fertility reduction if n is
small; therefore, the conservation of a maximum of variation is devised.
There exists experimental evidence that incompatibility as a postpollination
event occurs also in tropical forest trees (Boshier 2000; Sect. 6.5.1). Since it
takes much effort and time to identify its genetic control and to survey varia-
tion of populations at the S-locus, one must trust that the larger the popula-
tion, the more S-alleles may be encountered.
14.5 Selection of Genetic Resources 243

Objective 3: Preservation of Adaptability

Trees experience large variation of environmental conditions both in time and

in space. In consequence, forest trees must tolerate high environmental varia-
tion and forest tree populations must be able to produce a large number of
genotypes capable of survival and reproduction under various environmental
conditions. Reduced variation may compromise the evolutionary potential,
particularly the ability of a population to respond to changing selection pres-
sure. Since forestry is a typical low-input production system, the physical,
chemical, and biological environment of forest trees cannot be controlled as
strictly as in the more intensive agricultural production systems.
Genetic variation is a prerequisite of processes of evolutionary adaptation.
The main ecological function of genetic variation is the ability of populations to
respond to changed and unpredictable environmental conditions with adaptive
changes of their genetic structures (Sect. 7.2). The preservation of adaptability
or adaptive capacity has been pointed out as the main objective for the conser-
vation of forest genetic resources (Gregorius 1991). Genetic variation within
keystone species is of particular importance for the maintenance of the struc-
ture and function of an ecosystem if rapid or strong environmental change
occurs or is expected.
In contrast to objective 2, the target is the adaptable population rather than
the single gene. Hardly much evidence exists on the decline of tree species for
lack of adaptive capacity. This, however, is presumably due to the fact that in
most documented cases of species extinction human impact was much faster
than the gradual loss of species occurring with genetic impoverishment.
Genetic variation at adaptively relevant gene loci and the possession of genetic
variants with positive adaptive value are still an insurance for a population to
persist in a changing world. It must be hoped that at least some of those genetic
variants will contribute to adaptation also in the future. The analysis of genet-
ically controlled adaptive phenotypic traits can only be helpful in planning
conservation programs pursuing this goal.

Selection of Genetic Resources

The first step in any conservation project is to find out how the diversity we
wish to conserve is distributed in space (Bawa and Krugman 1990). We can
but preserve adequate samples of the total diversity and ecosystems, including
both centers of diversity and endemites. Since they run the risk of remaining
244 C HAPTER 14 Conservation of Genetic Resources in Tropical Forest Trees

undetected in our sampling, it is of fundamental importance to find popula-

tions most worthy of conservation. This means giving priority to natural pop-
ulations. Populations should not be regarded as a genetic resource if their
origin is uncertain or if genetic bottlenecks that happened during the recent
past are inferred. If the observed disturbance had little or no effect on varia-
tion, they may be considered equivalent (Soonhuae et al. 1995).
The selection of genetic resources of a species must also consider the pro-
tection of the respective ecosystem. Unfortunately, nature reserves receiving
high-priority protection are not necessarily established where the populations
are the most desirable from the conservation point of view. Species-rich sites
possess advantages for species conservation because of saving on costs, but are
not necessarily adequate for genetic resource conservation. It is not known
whether geographic regions with high species diversity can also be expected to
host much genetic diversity within those many species.
Pursuing a certain conservation objective refers to particular genetic infor-
mation that is preferably to be conserved. It is involved in the expression of
superior phenotypes in economically relevant quantitative traits (objective 1),
is particularly diverse (objective 2), or is important for the preservation of evo-
lutionary adaptability (objective 3). Some knowledge of spatial patterns of
genetic variation is required for selecting genetic resources on a biologically
sound basis. For instance, peripheral populations exposed to different ends of
the ecological amplitude of the species range under certain conditions possess
conservation value when pursuing objective 3 (Lesica and Allendorf 1995).
Genetic data help to make better use of the available populations by maximiz-
ing the evolutionary-response potential of a set of genetic resources. Thus,
genetic inventories are essential components of conservation strategies.
A nongenetic approach by Burgman et al. (2001) allows target areas and
their size to be identified on the basis of types and degrees of threat. It takes
much less effort and is highly suitable if urgent decisions are required.
Spatial patterns of genetic variation are complex and cannot be described by
any of the previously mentioned methods alone. The combination of several
methods for the assessment of variation patterns is recommended for the selec-
tion of genetic resources. If the objective is the preservation of superior trait
expressions (objective 1), hints on particularly valuable genetic resources are
mainly obtained from field trials. Genetic marker studies and possibly early
testing are particularly recommended for capturing maximum variation or
evolutionary adaptability (objectives 2 and 3; p. 154 in Finkeldey 1993).
A genetic resource must justify later efforts spent on its conservation.
Choosing arbitrary populations or stands for their incidental phenotypic supe-
riority in volume yield or stem form (as in approval for the production of
selected reproductive material) may be common practice in many countries but
does not necessarily serve the purpose of genetic conservation. The phenotypic
14.5 Selection of Genetic Resources 245

appearance of stands or their variation in phenotypic traits can at best be the

basis for the selection of genetic resources for pursuing objective 1. The close
connection to breeding goals is evident. However, the success of this selection
criterion is doubtful unless the degree of genetic control of the observed traits
is known. The selection mainly or even exclusively by phenotypic traits is not
suitable for conservation objectives 2 and 3.
In most instances the choice among several candidate collectives requires
our decision. In some extremely rare species only one option of conservation
exists, i.e., the rescue of what there is. Genetics and demography are then
challenged (Schwartz et al. 2000; Yates and Broadhurst 2002). No sampling is
relevant in these cases. It becomes very important to keep track of identities
and consanguinities (Krauss et al. 2002) during the process of physical con-
servation. This is particularly true with tree species that have gone extinct in
the wild and survive only as some ex situ resources. Remnants of tree popu-
lations spared because of poor site quality for agricultural use no longer rep-
resent the species as such (Brown and Hardner 2000; Young and Boyle 2000).
The identity of the species in a genetic sense is likely to be lost under these

Inventory of Genetic Marker Loci

Most efforts towards defining evolutionarily significant units have focused on genetic
distance measures based on markers to determine when populations are sufficiently dis-
tinct to warrant separate conservation. These methods, however, work poorly for many
tree species, as they are based on selectively neutral markers and thus reflect mutation,
genetic drift and migration, but not adaptive differences (Namkoong et al. 2000).

It has become common practice to assess variation at numerous markers and

average those estimates for making comparisons between populations that
might be eligible as gene resources because of their variation.
Inventories of molecular genetic marker loci provide some insight into muta-
tion events (Vornam et al. 2004), gene flow (Chap. 5), the reproduction system
(Chaps. 4, 6), metapopulation structure, as well as past and present population
size (Chap. 9). Many molecular markers are unlikely to be of any relevance for
fitness. Past bottlenecks can be traced (Example 3.1), since the risk of allele loss
is the same regardless of adaptive significance of the respective alleles. Although
there is no strict relationship between the size of a population and its genetic
variation (Sect. 9.3), the size of a population is still a crucial selection criterion.
Very small populations have to be assumed genetically depauperate.
Neutral markers are not the target of selection and may therefore not be con-
sidered surrogates of the traits in which variation is adaptive. They are not
246 C HAPTER 14 Conservation of Genetic Resources in Tropical Forest Trees

suitable tools for the detection of a geographic genetic pattern indicating the
selection response of tree populations; therefore, genetic variation of adaptive
significance may not be appropriately reflected at neutral marker loci. When
discussing the properties of alloenzyme loci, Finkeldey and Mtys (1999)
found that even though some of these gene loci under certain environmental
conditions possess relevance for fitness (Mller-Starck 1985; Ziehe et al. 1999;
Geburek 2000), they are still unsuitable for directly inferring the adaptive capac-
ity of populations and species. Reed and Frankham (2001) consider molecular
markers including alloenzyme loci unsuitable for inference on adaptive varia-
tion and differentiation (see also Navarro et al. 2005 for an example in Cedrela
odorata and Yang et al. 1996 for an example in a temperate pine species).
Two main objectives are evident with regard to the application of genetic
marker studies for the selection of genetic resources:

1. Centers of genetic diversity or populations or regions with large amounts of

genetic variation can be identified. The identification of a center of genetic
diversity of Acacia auriculiformis in Papua New Guinea (Wickneswari and
Norwati 1993) was described in Example 3.1.
2. It is possible to identify populations containing otherwise rare or even
unique alleles in high frequency. Also the occurrence of localized common
alleles points towards valuable genetic resources (Brown 1978).

Genetic variation within populations and their differentiation has long been
observed most easily and reliably at alloenzyme loci. Their limited number is a
disadvantage for the selection of genetic resources. Many surveys have shown
remarkably low amounts of genetic differentiation among populations at these
loci (Sect. 3.4), although genetic differences in quantitative, adaptive traits is
suggested by the results of field trials (Chap. 11). These contrasting patterns of
genetic diversity are frequently explained by the presumed selective neutrality
of variation at all alloenzyme loci (Muona 1989).
Recently developed molecular tools allow the study of more types of genetic
markers (Sect. 2.3.1). Not all types of molecular markers are useful for the
selection of genetic resources (Szmidt and Wang 2000). Essential advantages
are access to large numbers of gene loci, the potential use of different markers
with different modes of inheritance (uniparental, biparental), and different
degrees of selective relevance. Approaches to relate variation at molecular
markers to adaptive variation patterns were mentioned in Sect. 7.2. In view of
the existing scarcity of available experimental data, an assessment of the gen-
eral potential of molecular genetics methods for the selection of forest genetic
resources would be premature.
A very large number of parameters have been developed that are suitable
for condensing the information gained from genetic inventories (Sect. 3.3).
14.5 Selection of Genetic Resources 247

Different parameters express different properties of the variation in popula-

tions and their differentiation at a gene locus. None is alone suitable for analy-
sis, as ongoing research shows. Petit et al. (1998) have discussed some of these
parameters in the context of selecting populations as genetic resources.
Unfortunately, existing software tends to give preference to a certain type of
marker only. Of importance is also the way of combining the data from several
markers for statements on the genome.
Although neutral markers are unlikely to be involved in future processes
of adaptation, they are appropriate for inferring the recent evolutionary past
of populations, which in turn represents key information on their adaptive
Just as in seed collection for provenance experiments, seed samples for
assessing variation must not only be large enough but must be collected in a
judicious way (Sect. 11.4.3). The same applies to sampling adult trees them-
selves for genotyping (Murillo Gamboa 1997) and to collecting seeds for con-
servation stands (Sect. 14.6.2). Kim et al. (1994) devised a method of designing
sample sizes by finding the point where the cumulative frequency curves of
detected genic variants flatten out.
Favoring allozyme loci, Brown and Hardner (2000) derived appropriate
sample sizes that are relevant under various conditions. These authors also
stress the need for sampling a relatively large number of sites that are distrib-
uted according to ecological conditions of the range and the local frequencies
of a species. If seed is analyzed, the results of repeated sampling are much more

Inventory of Adaptive Trait Expressions and Adaptive Markers

Phenotypic traits are expressions of genetic information at the level of the tran-
scribed genes. Phenotypic variation deliberately excludes variety at lower levels
of organization, such as within DNA base sequences where these differences
have no effect on phenotype.
The observation of phenotypic traits characterizing the health status, growth,
and reproduction of a population in a field experiment allows us to draw con-
clusions on its adaptedness to the given environment (Sect. 7.2). The adaptive
potential, i.e., the spectrum of environments that a population is able to
get adapted to, can be assessed in a series of field trials. In both situations a well-
performing population is usually considered adapted or adaptable, respectively,
although adaptation may be of a purely physiological nature. Strictly speaking,
inference on adaptedness in environments not included in the experimental
series cannot be made. Neither can adaptability to evolutionarily completely
novel environmental factors be foreseen. Nevertheless, populations performing
248 C HAPTER 14 Conservation of Genetic Resources in Tropical Forest Trees

well in experimental series are inferred to possess not only advantageous genetic
variants at relevant gene loci but possibly also enough variants that may locally
function as a genetic load but are advantageous under changed environmental
Several options exist for assessing the genetic controlledness of phenotypic
traits (Sect. 12.3). Populations become prime targets for genetic conservation
measures if their superior average expressions of economically relevant traits
has a genetic basis. Traditional field trials in forestry are time-consuming.
Results from planned or recently established trials are not available for the
selection of genetic resources if the initiation of conservation measures is
urgent. Early tests in the greenhouse, climate chamber, laboratory, or nursery
are conducted to observe quantitative variation of seedlings. The conditions
can be modified as required. Environmental variation can be minimized or
even controlled. These experiments usually take only several months. It is usu-
ally impossible to identify the gene loci controlling a quantitative trait, their
mode of gene action, and their genetic structures; therefore, only very few
marker genes with relevance for survival or reproduction have so far been ana-
lyzed (Sherwin and Moritz 2000).
Some authors consider it possibly misleading to use data on neutral genetic
variation as the exclusive basis of decisions on selecting populations as gene
resources (Bekessy et al. 2003). The following case studies support this view:

1. Larsen (1986) observed quantitative variation patterns of seedlings of

European silver fir (Abies alba) provenances and identified a center of
genetic diversity in the southernmost region of the distribution range in
Calabria in southern Italy. The assumed refuge population could not go on
postglacial northward migration and was therefore not forced through a
bottleneck when remigrating around the Alps that run from east to west.
This hypothesis was later confirmed by Bergmann et al. (1990) by genetic
surveys. Since Calabrian firs were never confronted with the climate even
at medium elevations in central Europe, they suffer from frost when trans-
ferred. However, apart from the large variation at alloenzyme loci, they
have very valuable properties (Larsen and Mekic 1991; Geburek 2000) even
when they are planted far north of the natural distribution range of the
species. Their consideration as a gene resource would be based both on
their history and on laboratory tests.
2. Bekessy et al. (2003) studied the variation of random amplified polymor-
phic DNA (RAPD) markers and two phenotypic traits related to drought
resistance (root mass ratio and isotyope discriminating) in Araucaria arau-
cana, a conifer endemic in the zone of prevailing westerly winds in South
America. Its distribution covers a wide elevational range from 600 to 2,000 m.
Annual precipitation decreases from 4,000 mm west of the Andes in
14.5 Selection of Genetic Resources 249

Chile to less than 200 mm in their wind shade in Argentina. In a green-

house, seedlings of nine study populations showed significant variation
and differentiation that was uncorrelated between markers and quantita-
tive traits. The populations east and west of the Andean range clustered
together only in the quantitative traits, whereas clustering based on RAPD
data indicated grouping of populations across the Andes. It is fairly obvi-
ous that the eastern populations had acquired higher water use efficiency
regulated by the selective rainfall gradient. However, only the level of dif-
ferentiation in RAPDs was significantly correlated to geographic distance
(r = 0.56), indicating that this was shaped through limited gene flow and
genetic drift. Once an adaptive trait has been identified, it may serve as a
target variable in the selection of populations. However, not just the varia-
tion (including the genetic load) alone is crucial but also the possession of
favorable variants in notable frequencies that may increase the chance for
persistence of the populations in the foreseeable future.
3. Pastorino et al. (2004) studied Austrocedrus chilensis, another dioecious
South American conifer occurring a little further south. Several hundred
trees from 15 Argentinean populations sampled along three east west
transects were surveyed at 12 alloenzyme loci, half of which were virtually
monomorphic. The populations differed moderately with a tendency
towards the distribution of genetic variation along a latitudinal gradient,
the northerly populations being more variable. The most variable were
found in the steppe. This condition may reflect the effect of remigration
from several glacial refuges in more northerly locations. Again, genetic and
geographic distances ranging from 10 to 450 km were significantly corre-
lated (r = 0.47). Although mean annual precipitation ranged from 2,500
mm in the Andes to 330 mm in the steppe, this was not reflected by the
geographic variation pattern. The selection pressure exerted by the steep
precipitation gradient was not paralleled by alloenzyme variation. In other
instances such a selection pressure might not be quite as obvious.
4. In contrast to the two South American conifers possessing a small range
and more or less patchy distribution, Eucalyptus delegatensis (Sect. 11.5) is
widespread in southeastern Australia and Tasmania and forms many large
populations. Garnier-Gr and Ades (2001) found a close correlation
between diameter growth and survival in this species and studied geo-
graphic variation of this adaptive trait in a comprehensive provenance
experiment. The variation between populations and their interaction with
seven experimental locations could be explained by various properties of
the physical environment at the origins and biological qualities of the
respective forest communities in analyses of covariance (factorial regres-
sion). Data on solar radiation allowed adaptedness and adaptability to
the experimental sites to be predicted with high degrees of determination.
250 C HAPTER 14 Conservation of Genetic Resources in Tropical Forest Trees

The variations in temperature and rainfall were typically better predictors

of adaptability than the respective averages. The advantage of analyzing
geographic patterns is the predictability of the conditions at nonsampled
sites. Adaptedness clearly means adaptedness to past selection pressure.

In summary, it can be stated that genetic marker variation provides insight into
the genetic system, the Quarternary past of populations, and their centers of
variation. This indirect information of adaptability hardly possesses a causal
association with environmental variation but still makes a valuable contribu-
tion to the development of conservation strategies and might be even more
useful in combination with data on adaptive traits derived from experiments
done in judiciously chosen environments (Eriksson et al. 1993; Eriksson 1995;
Ennos 1996). Inventories of the variation at marker gene loci should then be
made if objective 2 deserves the highest consideration. Inventories of the vari-
ation in adaptive phenotypic traits should be made if objective 3 is to be pur-
sued (McKay and Latta 2002).
The integration of genetic and phenotypic information takes time and
effort; therefore, intensive studies have to be confined to species of prime eco-
logical or/and economic relevance. Results may be transferred to less impor-
tant species with similar life-history traits.
Immediate action for the protection of the genetic resources of a species
often becomes urgent in the absence of any data on spatial patterns of genetic
variation and without the possibility to conduct any of the experiments out-
lined previously. This situation applies to many lesser-known species in tropi-
cal forests. The conservation of at least one autochthonous population in each
of the main distribution areas of a species is recommended in this case.
Populations growing under both optimum and extreme environmental condi-
tions should be preserved. Both the geographic center of the species distribu-
tion and marginal populations should be represented in order to conserve
genetic information of populations that are possibly differentiated from each
other at adaptive loci.
Once a certain population has been found to be worthy of conservation, it
must have a sufficient size to be maintained over generations.

Conservation Methods

Appropriate methods used for conservation are required to preserve genetically

variable populations. The survival of particular individuals is usually of minor
14.6 Conservation Methods 251

Genetic resources can be preserved in situ in natural, autochthonous popu-

lations. Ex situ conservation applies to all methods requiring the transfer of
reproductive material to preservation sites outside the natural environment.
Another fundamental distinction of conservation measures refers to the
temporal dynamics of the genetic structures of a resource. Methods that allow
for changes of genetic structures or that even promote adaptive changes of
genetic structures in response to newly experienced or modified environmen-
tal conditions are called dynamic. The in situ preservation of genetic resources
is by definition a dynamic conservation method. Conservation is static, if it is
aimed at keeping the genetic structure of the resource unchanged. The ex situ
conservation of genetic resources uses both dynamic and static approaches.
By monitoring using appropriate markers or trait expressions, it can be
found out whether the criteria applied at selection are still fulfilled. If these
genetic surveys require too much effort, at least the maintenance of an appro-
priate reproduction-efficient population size should occasionally be checked.

Dynamic Conservation in Situ

Both conceptual and practical considerations call for an important role of

in situ conservation of tropical forest trees because of its dynamic character.
This is different from conservation of many agricultural species (Sect. 14.6.3).
Their genetic resources are preserved mainly in gene banks and by other static
methods. If the environment changes drastically and rapidly, the preservation
of evolutionary adaptability can be achieved only by dynamic conservation
methods. The dynamics is induced by natural mortality and by the process of
natural regeneration.
The genetic resources of tropical tree species that depend on interactions with
other species at least during particular stages of their life cycle can best be con-
served by dynamic methods. Their conservation is usually most easy in situ. For
example, the static preservation of many figs (Ficus spp.) would not pose any
problems owing to the ease of vegetative propagation. However, static conserva-
tion is likely to result in the (local) extinction of the pollinating wasps (Example
7.2). A sexually produced progeny generation cannot be produced in this case
and, by definition, the conservation method would fail to preserve a population.
Although species interactions in tropical forests are often complex and poorly
understood, they are likely to be of much more significance for the conservation
of genetic resources in tropical forests than in temperate ecosystems.
The protection of certain forested areas is necessary for the in situ conser-
vation of genetic resources of trees. Thus, in situ gene reserves must be incor-
porated into overall plans for the establishment of nature reserves and other
252 C HAPTER 14 Conservation of Genetic Resources in Tropical Forest Trees

protected areas of a region or a country. The enforcement of protection meas-

ures and the acceptance by local communities is more difficult and important
than the legal declaration of a protected area in most tropical countries. Tree
populations in nature reserves, national parks, etc. are not necessarily typical
genetic resources in the close sense, since they were not chosen for their
genetic variation. These reserves are still important sites for the genetic con-
servation of tree populations in their natural forest community. They may be
considered more or less undisturbed and, last but not least, they are protected
against loss. Restrictions on the removal of plants or plant parts may represent
a hindrance of the occasional utilization of genetic resources and of silvicul-
tural management. An important advantage of in situ conservation is the pos-
sibility to combine the conservation of the genetic resources of a target species
with the conservation of other, associated species in the same block or belt of
conservation forest (Cossalter 1989). In situ conservation is the only feasible
option for the conservation of genetic resources of species of minor silvicul-
tural importance with unknown reproductive biology.
However, protection of a certain area is not necessarily in accordance with
the specific requirements for the genetic conservation of a target species.
Silvicultural management may be needed to direct the dynamics of the ecosys-
tem or to counteract human influence. Management techniques are as simple
as the establishment of buffer zones and firebreaks and the removal of com-
peting vegetation, or more laborious, such as the promotion of regeneration by
sowing or planting locally harvested material. An increasing number of in situ
resources are located in somehow managed landscapes, which adds one more
dimension to the complexity involved in maintaining resources (Boyle 2000).
The presence of humans in forest reserves may have positive effects on protec-
tion (Schwartzman et al. 2000). Forest genetic resources have nothing in com-
mon with museums.
Many tree species of tropical forests experience strong fluctuations of their
population size or density in an area of given size. This applies primarily to
species confined to particular successional phases. Local extinction is a natural
process for species characterized by a metapopulation structure (Sect. 9.2).
However, the fragmentation and isolation of protected areas may prevent their
recolonization by many species (Ashton 1981). Human intervention by silvi-
cultural management such as vine control and enrichment plantings is
unavoidable in order to ensure the survival of populations of the target species
and thus the conservation of the genetic resource in many small gene reserves.
It has still to be found out which interventions are incompatible with the con-
servation function of forest (Putz et al. 2001).
Complexity induced by allochthonous populations within reach of the
pollen transfer distance of gene resources (Ennos 1996; Lindgren and Lindgren
1996; Leinemann and Hattemer 2006) raises problems. If the resource is large
14.6 Conservation Methods 253

enough, the delineation of buffer zones helps to prevent undue contamination

in natural regeneration. A small resource may be rescued by vegetative propa-
gation (Sect. 14.6.4).

Dynamic Conservation ex Situ in Man-Made Forests

Dynamic ex situ conservation of genetic resources is the preservation of popu-

lations in man-made forests established outside their natural habitat with gen-
erative planting stock. It is an important complementary method (Li et al. 2002)
for the conservation of genetic resources primarily of plantation species (objec-
tives 1 and 3). In ex situ conservation, more management options may exist as
to size, density, and design. The same principles as described in Sect. 9.3 apply
to ex situ conservation. The mating system and gene flow are influenced by the
regulation of population density, stand structure, and the spatial arrangement
of the trees at planting. Planting stock from several populations of a region may
be combined to a single ex situ conservation stand or kept separate.
The establishment of conservation stands under different environmental
conditions eventually leads to adaptive changes of genetic structures. Changes
including losses of genetic variation due to genetic drift can be minimized by
maintaining large effective population sizes in conservation stands.
Provenance resource stands can be regarded as conservation stands. They
can be combined with other forests for the production of reproductive mate-
rial or converted to seed production areas (Guldager 1975).
In principle, the dynamic ex situ conservation of genetic resources is com-
patible with breeding. Conservation of genetic resources is mainly achieved in
base populations. Since both breeding and propagation populations are
selected for their own superior phenotype or that of their progeny, they should
not be considered genetic resources for the preservation of the evolutionary
adaptability of a species.
A nonhierarchical multiple-population breeding strategy (Sect. 12.9) coin-
cides with the establishment and management of ex situ conservation stands
(p. 143 ff. in Finkeldey 1993). Genetically differentiated populations are estab-
lished or maintained in different environments with the aim of enhancing their
adaptive differentiation. Variable selective regimes with both regard to selection
intensity and breeding goals are applied. Directional selection and genetic drift
potentially lead to a loss of genetic variation in single populations; however, no
or only slight losses of genetic multiplicity at adaptive loci are expected within
the overall resource composed of all populations. Genetic monitoring is recom-
mended in order to assess the loss of genetic variation within populations and
to take appropriate countermeasures by promoting gene flow through the
controlled exchange of planting stock.
254 C HAPTER 14 Conservation of Genetic Resources in Tropical Forest Trees

Genetic variation within populations can be maintained by a moderate level

of gene flow and/or migration among genetically differentiated, semi-isolated
subpopulations, thus adding dynamic processes involving several subpopula-
tions. The evolutionary adaptability of the populations comprising a complex of
ex situ conservation stands can be maintained by limited gene flow even if the
sizes of the subpopulations are small. Thus, population sizes of single subpopu-
lations may be chosen to prevent excessive inbreeding and very strong drift.
Both colonization processes and gene flow by pollen in artificial metapopula-
tions might include still-existing populations. It depends on the respective
species to what extent this system can be fragmented. In any event, populations
are ephemeral. The natural regeneration of these resources is difficult unless
they are colonized by a disperser fauna providing for mating contact and for
mixing genotypes in the subsequent generation. Conservation forests have a
long-term perspective only if the system of reproduction becomes effective as in
forests comprising several complementary species of trees and animals. They
require more than just minimal management.
On the basis of a cluster analysis of larger populations of E. albens at several
alloenzyme loci, Prober and Brown (1994) devised a conservation strategy for a
species without major geographic differentiation. They recommended a wide
geographic spread of reserved populations scattered across the region and their
augmentation with conservation stands for maximizing overall genetic diversity.
The seed for these stands was to be collected from at least ten trees in a large local
population with at least 500 reproductive individuals. This number appears to be
low (Sect. 9.3). The efficiency of different sampling strategies in two small and
one large population of Pinus tecunumanii was assessed by Dvorak et al. (1999).
They concluded that seeds from ten to 20 seed trees are sufficient to sample all
alleles with frequencies of more than 5% in each population.
It is true that under favorable conditions of pollen flow these few seed trees
sample most of the alleles in the population. Since the seed trees contribute at
least half of the genetic information of the seed, their low number might imply
an undue shift in allele frequencies.
Practical advantages of ex situ conservation stands are not limited to the
options of a genetic management as outlined earlier. The location for the estab-
lishment of ex situ conservation forests can be chosen freely. This allows ex situ
conservation stands to be protected more easily and efficiently as compared
with in situ gene reserves.
The decision about planting either a single large or several small (SLOSS)
stands ex situ is largely academic (Hattemer 2006; Sect. 9.3). Forest genetic
resources can only be established where appropriate land is available.
Consequently there is little room for this type of planning. The distribution of
given material over several sites has the advantage in view of safety of the
resource. The emphasis must be on the larger of those copies. However, if some
14.6 Conservation Methods 255

planning can be made, both the size and the design of genetic resources can be
chosen. It has to be kept in mind that the establishment of several small
resources is related to habitat fragmentation that might enhance the risk of los-
ing the resource (Wilcox and Murphy 1985); therefore, the issue is not splitting
up a given resource but rather establishing several of them in only partial isola-
tion. Long-term conservation focuses on persistence at spatial scales larger than
that of the single population (Thrall et al. 2000). As mentioned in Sect. 9.3,
a system of interconnected populations of moderate to large size has an optimal
chance of persistence. Persistence of a resource over generations must account
for other tree species providing a continuous food basis for pollen and seed dis-
persers, so that natural regeneration works and currently unoccupied habitat
fragments in the vicinity may be colonized. Thrall et al. (2000) reviewed both
the theory of metapopulations and its application to genetic conservation.

Conservation of Seeds in Gene Banks

Seeds can be transported easily and large numbers of genotypes can be stored
in a small facility. Their long-term storage in gene banks is the most important
static method for the conservation of crop plant genetic resources. The conser-
vation of seeds is static inasmuch as the genetic structures of the resource do
not change during storage.
The seed physiology of a tree species sets limits to the potential of long-term
storage of seeds in gene banks. Long-term storage of forest seeds as a genetic
resource is often combined with the establishment of forest seed centers in
tropical countries. The main task of seed centers is the procurement and pro-
vision of seeds for afforestation (Albrecht 1993). A concentration of tasks
related to forest tree seed handling and storage is meaningful in view of the
limited personal and financial resources available.
Long-term storage is impossible for recalcitrant seeds. The seed physiology
and optimum storage conditions of many tropical tree species are unknown.
In comparison with the lifetime of trees or their rotation periods, the maxi-
mum duration of seed storage is short even for species with orthodox seeds.
The seeds of annual agricultural crop plants are periodically sown out and the
seed of the new generation is put back into the gene bank. This cannot be
done for trees, so the germinability decays, which induces genetic change.
Since the main objective of a conservation programs for forest trees is the
preservation of evolutionary adaptability, static conservation methods are less
suitable than dynamic methods. Establishment and operation of a gene bank
are costly. Last but not least, gene banks are sensitive to short-term interrup-
tions of their operation, such as failure of the electricity network, which is
frequent in some areas.
256 C HAPTER 14 Conservation of Genetic Resources in Tropical Forest Trees

The long-term storage of seeds is a complementary method to dynamic con-

servation of forest genetic resources for some important tree species. The main
advantage is the preservation of a large number of genotypes in a fairly small

Vegetative Propagation for the Conservation of Forest Genetic Resources

Vegetative propagation allows us to identically reproduce and multiply geno-

types. It is a method to rescue moderate numbers of remnant trees of rare
species in danger of extinction and thus to preserve moderate numbers of indi-
vidual genotypes. It is therefore not necessarily suitable for the conservation of
forest genetic resources in general. Since in asexual propagation recombination
is excluded, the number of conserved genotypes is usually much lower than in
sexually reproducing populations. Nevertheless, vegetative propagation is of
great interest in domestication and represents an important technique for con-
servation programs for tropical forest trees (Longman 1976; Ladipo et al. 1991;
Smits 1993; Khuspe et al. 1994; Leakey et al. 1994; Newton et al. 1994; Smits et al.
1994). Trees and their genotypes become mobile by vegetative propagation. This
mobility is of particular importance for emergency measures to protect highly
endangered genetic resources (Finkeldey and Hattemer 1993). Many popula-
tions or even species are facing immediate extinction in their natural habitat, for
example, owing to large-scale forest destruction. Vegetative propagation offers
the opportunity to escape extinction and conserve genetic resources by moving
genotypes to areas which can more easily be protected than the natural habitat.
Disturbance of forests possibly interrupts the mating contact among plants
by decreasing population densities below a critical threshold. Vegetative propa-
gation allows artificial populations to be established by planting together
propagules from remnant trees scattered over the landscape and provides for
their reproduction with tolerable levels of inbreeding. Assembling vegetatively
propagated genotypes in clonal conservation orchards is a typical emergency
measure to protect genetic resources from immediate extinction (Rotach 2000).
Clonal plantations established as part of a breeding program such as clonal
seed orchards, clonal archives, and clonal gardens preserve the genetic infor-
mation of particular phenotypically superior genotypes, although they con-
tribute only little to the preservation of genetic variation and evolutionary
adaptability (objective 3).

Biotechnological Methods of Preservation

The distinction between traditional and modern biotechnological methods is

somehow arbitrary. Macrovegetative propagation of plants is not regarded as a
14.7 Regeneration of Forest Genetic Resources 257

biotechnology but tissue culture is a typical biotechnological method. However,

the genetic consequences of macrovegetative and microvegetative propagation
techniques are very similar.
Biotechnological methods have so far been of minor importance for the
conservation of tropical forest genetic resources. The application of modern
and sophisticated technologies fascinates research workers in many countries.
However, better understanding of the spatial and temporal dynamics of genetic
structures of the target tree species should have research priority. The applica-
tion of biotechnological methods such as in vitro culture, slow-growth tech-
niques, and cryopreservation to the physical conservation of tree material is
less urgent and can hardly be considered a main focus of forest genetic research
(p. 151 ff in FAO 1994).
Tissue culture allows us to conserve and vegetatively propagate trees in vitro.
It is an important alternative to the conservation of seed in particular for
species with recalcitrant seeds. Tissue cultures of many species can be main-
tained with moderate efforts over long periods of time if kept in suitable con-
ditions, i.e., at low temperature and light and in appropriate media. The
long-term in vitro preservation of genotypes is practiced for some agricultural
crops such as cassava but hardly for tropical forest trees (Engelmann 1994; p. 31 f.
in FAO 1994).
A virtually indefinite preservation of numerous plantlets in a small facility
is feasible by cryopreservation, i.e., the preservation of biological tissue at a
temperature of 196C in liquid nitrogen. Embryos, pollen, and other tissue
from temperate forest trees were frozen in liquid nitrogen and could be suc-
cessfully regenerated (Jrgensen 1990; Park and Son 1996).
Molecular genetics techniques allow us to isolate, multiply, and conserve
fragments of DNA outside of living cells of the respective species. A gene
library, gene bank, or genomic bank is a collection of cloned DNA fragments
which represent parts or even the complete genome of an organism. The frag-
ments are inserted into plasmids as vectors that are preserved in living cells of
a host organism. The host organisms, usually bacteria, are capable of multipli-
cation by cell division. Not only the host DNA, but also the foreign DNA is
transmitted during cell division. DNA fragments of any organism can be
preserved and amplified indefinitely in genomic banks.

Regeneration of Forest Genetic Resources

A program for the conservation of a genetic resource is frequently considered

successful if genetic information has been preserved for a certain time period.
However, the main objective of conservation programs for forest genetic
resources is not the preservation as such. It is rather the establishment of a
258 C HAPTER 14 Conservation of Genetic Resources in Tropical Forest Trees

population that is both adapted to the environmental conditions prevailing

during the regeneration of the resource and adaptable to future environmental
change. The success of a conservation program can be assessed only after the
resource has been regenerated; thus, the regeneration of the resource is an inte-
gral part of the conservation strategy and has to be planned during an early
stage of a conservation program. This applies, for example, to the natural
regeneration of an in situ resource. If static conservation methods have been
employed and, in particular, if the resource has been conserved by biotechno-
logical methods, the regeneration is strictly separated from the conservation
Genetic resources conserved in situ are regenerated naturally. Natural regen-
eration depends on numerous ecological, silvicultural, and socioeconomic
conditions in tropical forests (see the case study by Neil 1990 on Agathis macro-
phylla in Vanuatu). If natural regeneration is insufficient, it may be completed
by planting stock raised from local seed.
Dynamic evolutionary processes are promoted if an ex situ conservation
stand is regenerated naturally. After several generations of natural regeneration
a land race is likely to develop. Alternatively, the resource can be regenerated by
planting or sowing material that was harvested there. It is possible to mix
reproductive material from several genetically differentiated ex situ conserva-
tion stands derived from the same population in order to simulate migration
and to increase genetic multiplicity.
Germinability of seeds is obviously required for the regeneration of a resource
after storage of material in a gene bank. The planting stock derived from
these seeds is used either for the establishment of ex situ conservation stands or
for enrichment planting. A general weakness of static conservation methods
becomes evident during the regeneration of a resource. No adaptive changes of
genetic structures are possible during the storage period. A genetic resource
may have lost its adaptedness if drastic and large-scale environmental change
occurred during this phase. In this case evolutionary adaptation is necessary dur-
ing the regeneration. The adaptation process is likely to be accompanied by
strong viability selection and consequently a severe reduction of population size.
Hence, the survival of the population or at least the establishment of a genetically
variable ex situ conservation stand is at risk (Ziehe et al. 1989).
Genetically variable progeny is produced in clonal conservation orchards.
Seeds harvested there may be used for the establishment of ex situ conservation
stands or enrichment planting in logged forests. Too small numbers of clones
induce losses of genetic multiplicity due to genetic drift. In anemophilous tree
species, stored pollen may be used to support seed formation.
If biotechnological methods were used for the physical preservation of
material, particular techniques and abilities are required to regenerate com-
plete plants. Regeneration from tissue cultures requires hormonal treatment,
14.7 Regeneration of Forest Genetic Resources 259

different types of media, and modified culture conditions. The regeneration of

material conserved by cryopreservation in liquid nitrogen is even more diffi-
cult and requires in most cases an in vitro phase. The regeneration of a com-
plete plant from the DNA fragments of a gene library is likely to be impossible
even in the distant future. Molecular genetics methods do not have practical
importance for the physical conservation of genetic material. The application
of other biotechnological methods is impeded by difficulties to regenerate trees
from the conserved material or to establish a genetically variable population.
Genetic monitoring at marker loci of a resource is indicated during the regen-
eration in order to periodically assess genetic structures. Countermeasures are
indicated if genetic variation is lost.

Example 14.1: Conservation of the Genetic Resources of Pinus merkusii in

On the Southeast Asian mainland, the natural distribution of P. merkusii
stretches from Myanmar to Thailand, Laos, Cambodia, and Vietnam. The
species is also naturally distributed in Indonesia on Sumatra and in the
Philippines on Luzon and Mindoro. Large-scale plantations have been estab-
lished in Indonesia, mainly on Java, with provenances from Sumatra. The main
distribution area of the species in Thailand is split into three isolated regions in
northern, central, and northeastern Thailand.
In Thailand natural stands of P. merkusii (Fig. 14.1) are affected by forest
destruction, illegal logging, and resin tapping. P. merkusii is regarded as a
potential plantation species in Thailand and other countries on the Southeast
Asian mainland in particular for resin production. However, plantation estab-
lishment is difficult owing to the occurrence of the grass stage (Fig. 14.2), a
long-lasting growth depression of seedlings as an adaptation to periodic
ground fires (Fig. 14.3).
A provenance trial revealed fast growth, good stem form, and a compara-
tively short duration of the grass stage in provenances from northeastern
Thailand (Sect. 11.4.3). The average level of genetic variation at isozyme gene
loci is low for the species in Thailand (He = 0.058; Table 3.1). Extraordinarily
low outcrossing rates were observed in some stands of P. merkusii in Thailand
(Sect. 6.3.2) and were explained by a combination of low density, overaging,
and poorly synchronized flowering (Changtragoon and Finkeldey 1995a).
Genetic resources of P. merkusii are primarily conserved in two in situ con-
servation areas in northeastern Thailand since these provenances appear to be
most suitable for seed harvest for plantation establishment. Logging is strictly
forbidden in the gene reserves, firebreaks were laid out, and the reserves are
frequently patrolled by the local forest service. However, the protection of the
two in situ conservation areas is extremely difficult owing to the pressure of
260 C HAPTER 14 Conservation of Genetic Resources in Tropical Forest Trees

Fig. 14.1. Natural Pinus

merkusii forest in northern
Thailand. (Photo: R. Finkeldey)

the local population on the few remaining forests. Almost all pines in the gene
reserves are severely damaged by illegal resin tapping and/or the removal of
parts of the lower stem for the production of charcoal (Fig. 14.4). Ground fires
are frequent because the rural population wants to enhance the growth of
grasses for forest pasture.
Seed production of P. merkusii is poor in the gene reserves owing to overag-
ing, poor health, and low population density. Natural regeneration is almost
completely absent in the gene reserves. Young seedlings rarely survive owing to
yearly ground fires during the long dry season. In view of the absence of natu-
ral regeneration, seeds were harvested and sown in temporary nurseries. Three-
year-old seedlings of local origin were planted in the gene reserves. These
plants survive light ground fires (Fig. 14.3).
14.7 Regeneration of Forest Genetic Resources 261

Pinus kesiya

Pinus merkusii

Fig. 14.2. An approximately 3-year-old seedling of P. merkusii in the grass stage (front).
In the background is a seedling of P. kesiya, another pine native to Thailand. (Photo:
R. Finkeldey)

The establishment of at least two more in situ gene conservation areas in

northern and central Thailand, respectively, is recommended in addition to the
already existing gene reserves. Because of the less dense human population,
protection of an area is easier in these regions. Spontaneous natural regenera-
tion is abundant in the north. Light felling and other silvicultural techniques
can be used to naturally regenerate a gene reserve in northern Thailand.
Ex situ conservation measures are important complementary activities to
conserve the genetic resources of P. merkusii in Thailand. Seeds cannot be
stored long at ambient temperatures and facilities for the long-term storage of
seeds are not available. Vegetative propagation by means of cuttings is difficult.
The establishment of dynamic ex situ conservation areas is currently the only
feasible and meaningful measure in addition to the in situ gene reserves.
262 C HAPTER 14 Conservation of Genetic Resources in Tropical Forest Trees

Fig. 14.3. A P. merkusii seedling after a ground fire. The long, fleshy needles are
burned, but the terminal bud survived and a new sprout has developed. (Photo:
R. Finkeldey)

Ex situ conservation areas should be established in each of the three main

distribution areas of P. merkusii in Thailand. Changtragoon and Finkeldey
(2000) recommend establishing one ex situ conservation area in each region.
Areas should be selected according to the general suitability for the species and
the ease of protection. Material should be harvested from at least three differ-
ent stands in each region. At least ten trees should be harvested in each stand
following the general recommendations for seed harvest in provenance trials as
14.7 Regeneration of Forest Genetic Resources 263

Fig. 14.4. Heavily damaged P. merkusii tree in an in situ gene reserve in Thailand.
(Photo: R. Finkeldey)

outlined in Sect. 11.4.3. Seeds from the same region may be bulked, but the
reproductive material should not be exchanged among regions. The ex situ
conservation stands should be planted as pure forests in high density.
Outcrossing is expected to be promoted in the plantations owing to the high
population density and strong flowering of young trees after reaching maturity.
The double purpose of the plantations will be to conserve the genetic resources
of P. merkusii in Thailand and to serve as seed production areas for the
production of seeds eventually required for commercial plantings (From
Sa-Ardavut et al. 1989; Changtragoon and Finkeldey 1995a, 2000).
264 C HAPTER 14 Conservation of Genetic Resources in Tropical Forest Trees

Programs for the Protection of Forest Genetic Resources

National programs for the conservation of forest genetic resources are usually
planned and executed jointly by scientists and practical foresters trained in tree
breeding. Few researchers are experienced in the application of population
genetics and conservation-specific issues in tropical forests. Conservation of
forest genetic resources is often mistaken as subordinate elements of tree
improvement programs and the complex biological background is overlooked.
Apart from tree breeding, the conservation of genetic resources in tropical
forests is closely connected to the silvicultural management of forests (e.g., nat-
ural regeneration), nature conservation (conservation areas), legal regulations
(declaration of protected areas), political and socioeconomic issues (execution
of protection measures), and overall development planning. The biological and
genetic considerations outlined earlier must be incorporated into these subject
matters in order to initiate promising conservation programs. This applies in
particular to all planned activities for the in situ conservation of forest genetic
The local population is not only directly affected by many conservation
measures, but also by the main target group for the long-term beneficial effects
of a conservation program; thus, the rural populations must be integrated in
programs for the protection of forest genetic resources. Land is needed for
in situ and ex situ gene reserves and the utilization of forest products from con-
servation areas is restricted. The planning of conservation measures must take
into account legal and customary rights of forest owners and traditional users.
At an international level, several documents from the United Nations
Conference on Environment and Development (UNCED) during 1992 (the
Rio Summit) cover aspects of relevance for the conservation of tropical for-
est genetic resources. The Forest Principles, the Convention on Biological
Diversity, and Agenda 21 are of particular importance in addition to the 27
general principles of the Rio Declaration (Finkeldey 1996).
The importance of international scientific co-operation and financial aid for
the protection of natural forests in developing countries is stressed in the
Forest Principles and the Convention on Biological Diversity. They emphasize
the sovereign right of countries to utilize their genetic resources [see Article
8(g)]. The potential use of genetic resources by modern biotechnologies was a
controversial issue during the formulation of the Convention of Biological
Diversity. Technology transfer to developing countries is mentioned as being of
particular importance in this field.
Also Chap. 11 (Combating Deforestation), Chap. 15 (Conservation of
Biological Diversity), and Chap. 16 (Environmentally Sound Management of
14.9 Recommended Literature 265

Biotechnology) of Agenda 21 deal with forest genetic resources. The particular

role of forests for the conservation of biological diversity is recognized, the sig-
nificance of the of in situ conservation of forest genetic resources is stressed,
the complementary sides of the conservation and utilization of genetic
resources are emphasized, and the importance of regional and international
co-operation for the conservation of forest genetic resources is highlighted.
However, it is still a long way ahead before the ambitious objectives of the Rio
Documents concerning the conservation and sustainable utilization of forest
genetic resources are achieved.

Recommended Literature

A systematic treatment of the principles of genetic conservation is given in the

book by Frankel and Soul (1981). Two books with various contributions on
animals and plants, including some on tropical forests, are those edited by
Schonewald-Cox et al. (1983) and Soul (1986). Nowadays, various journals
and many books are devoted to genetic conservation.
The book edited by A. Young et al. (2000) is devoted to forest trees. In sev-
eral contributions reference is made to tropical trees. Very recently, Geburek
and Turok (2006) have edited a book on forest genetic resource conservation.
The emphasis is on European trees but many contributions deal with items of
general relevance to genetic resources of trees, including tropical forest trees.
The booklet edited by the FAO (1993) gives a general treatment of genetic con-
servation in tropical forest trees and presents some case studies. Three booklets
edited by the International Plant Genetic Resources Institute (Anonymous
2004ac) describe both general principles and methods of genetic conservation
in woody plant species.

Acher V, Faivre Rampant P, Pques LE, Prat D (2004) Chloroplast and mitochondrial
molecular tests identify European Japanese larch hybrids. Theor Appl Genet
Ackerly DD, Rankin-de-Merona JM, Rodriguez WA (1990) Tree densities and sex ratios
in breeding populations of dioecious Central Amazonian Myristicaceae. J Trop Ecol
Ackerman JD, Mesler MR, Lu KL, Montalvo AM (1982) Food-foraging behaviour of male
Euglossini (Hymenoptera: Apidae): vagabonds or trapliners? Biotropica 14:241248
Adams WT (1992) Gene dispersal within forest tree populations. New For 6:217240
Adis J (1990) Thirty million arthropod species too many or too few? J Trop Ecol
Albrecht J (1993) Forest seed handling. In: Pancel L (ed) Tropical forestry handbook,
vol 1. Springer, Berlin Heidelberg New York, pp 381462
Aldrich PR, Hamrick JL (1998) Reproductive dominance of pasture trees in a frag-
mented tropical forest mosaic. Science 281:103105
Aldrich PR, Hamrick JL, Chavarriaga P, Kochert G (1998) Microsatellite analysis of
demographic genetic structure in fragmented populations of the tropical tree
Symphonia globulifera. Mol Ecol 7:933944
Allard RW (1960) Principles of plant breeding. Wiley, New York
Altukhov YP (2006) Intraspecific genetic diversity. Monitoring, conservation, and
management. Springer, Berlin Heidelberg New York
Altukhov YP, Salmenkova EA (2002) DNA polymorphism in population genetics.
Russ J Genet 38:9891008
Alvarez-Buylla ER, Chaos , Piero D, Garay AA (1996) Demographic genetics of a
pioneer tropical tree species: patch dynamics, seed dispersal, and seed banks.
Evolution 50:11551166
Andersen M, Thornhill A, Koopowitz H (1997) Tropical forest disruption and stochas-
tic biodiversity losses. In: Laurance WF, Bierregaard RO (eds) Tropical forest rem-
nants: ecology, management, and conservation of fragmented communities.
University of Chicago Press, Chicago, pp 281291
Andrianarimisa A, Bachmann L, Ganzhorn JU, Goodman SM, Tomiuk J (2000) Effects
of forest fragmentation on genetic variation in endemic understory forest birds in
central Madagascar. J Ornithol 141:152159
Anonymous (2004a) Forest genetic resource conservation and management: vol 1:
overview, concepts and some systematic approaches. FAO-FGR-FLD-IPGRI, Rome
Anonymous (2004b) Forest genetic resource conservation and management: vol 2: in
natural forests and protected areas (in situ). FAO-FGR-FLD-IPGRI, Rome
268 References

Anonymous (2004c) Forest genetic resource conservation and management: vol 3: in

plantations and gene banks (ex situ). FAO-FGR-FLD-IPGRI, Rome
Apsit VJ, Hamrick JL, Nason JD (2001) Breeding population size of a fragmented pop-
ulation of a Costa Rican dry forest tree species. J Hered 92:415420
Armstrong WP (1996) Plant sexuality and political correctness. Waynes word 4.
Arnold RJ, Midgley SJ (1995) Conserving biodiversity the work of the Australian tree
seed centre. Commonw For Rev 74:121128
Ashton PS (1969) Speciation among tropical forest trees: some deductions in the light
of recent evidence. Biol J Linn Soc Lond 1:155196
Ashton PS (1981) Techniques for the identification and conservation of threatened
species in tropical forests. In: Synge H (ed) The biological aspects of rare plant con-
servation. Wiley, Chichester, pp 155164
Ashton PS (1982) Dipterocarpaceae. In: van Steenis CGGJ (ed) Flora Malesiana, series
1, Spermatophyta, vol 9, part 2. Nijhoff, The Hague pp 237552
Ashton PS (1988) Dipterocarp biology as a window to the understanding of tropical
forest structure. Annu Rev Ecol Syst 19:347370
Austerlitz F, Mariette S, Machon N, Gouyon P-H, Godelle B (2000) Effects of coloniza-
tion processes on genetic diversity: differences between annual plants and tree
species. Genetics 154:13091321
Austerlitz F, Dick CW, Dutech C, Klein EK, Oddou-Muratorio S, Smouse PE, Sork VL
(2004) Using genetic markers to estimate the pollen dispersal curve. Mol Ecol
Avise JC (2000) Phylogeography. The history and formation of species. Harvard
University Press, Cambridge
Awang K (1994) F/FREDs species network: results, constraints, achievements In:
Finkeldey R, Havmoller P (eds) Proceedings of the regional workshop on species
improvement networks (SPINs), Chiang Mai, Thailand, July 1994. RAS/91/004 field
document no 19. UNDP/FAO Forest Tree Improvement Project, Los Baos, pp 1018
Baker HG (1960) Apomixis and polyembryony in Pachira oleaginea (Bombacaceae).
Am J Bot 47:296302
Baker HG (1970) Evolution in the tropics. Biotropica 2:101111
Ballal SR, For SA, Guttman SI (1994) Apparent gene flow and genetic structure of Acer
saccarum subpopulations in forest fragments. Can J Bot 72:13111315
Bnfer G, Fiala B, Weising K (2004) AFLP analysis of phylogenetic relationships among
myrmecophytic species of Macaranga (Euphorbiaceae) and their allies. Plant Syst
Evol 249:213231
Banik RL (1995). Domestication and breeding of bamboos. RAS/91/004 working paper
no 9. UNDP/FAO Forest Tree Improvement Project, Los Baos
Barlow BA (1958) Tetraploid twins and apomixis in Casuarina nana Sieb. Aust J Bot
Barnes RD, Burley J (1990) Tropical forest genetics at the Oxford Forestry Institute:
exploration, evaluation, utilization and conservation of genetic resources. For Ecol
Manage 35:159169
Barnes RD, Mullin LJ (1989) The multiple population breeding strategy in Zimbabwe.
Five year results. In: Gibson GL, Griffin AR, Matheson AC (eds) Breeding tropical
trees: population structure and genetic improvement strategies in clonal and seedling
References 269

forestry. Proceedings of the IUFRO conference, Pattaya, Thailand, November 1988.

Oxford Forestry Institute, Oxford, and Winrock International, Arlington, pp 148158
Barnes RD, Simons ASJ (1994) Selection and breeding to conserve and utilise tropical
tree germplasm. In: Leakey RRB, Newton AC (eds) Tropical trees: the potential for
domestication and the rebuilding of forest resources. Her Majestys Stationery
Office, London, pp 8490
Barth FG (1982) Biologie einer Begegnung. Dioe Partnerschaft der Insekten und
Blumen. Stuttgart, Deutsche Verlagansanstalt
Bass S (2001) Policy inflation, capacity constraints: can criteria and indicators bridge
the gap? In: Raison RJ, Brown AG, Flinn DW (eds) Criteria and indicators for sus-
tainable management. CABI, Wallington, pp 1937
Bawa KS (1973) Chromosome numbers of tree species of a lowland tropical commu-
nity. J Arnold Arbor 54:422434
Bawa KS (1974) Breeding systems of tree species of a lowland tropical community.
Evolution 28:8592
Bawa KS (1980) Evolution of dioecy in flowering plants. Annu Rev Ecol Syst 11:1539
Bawa KS (1990) Plant-pollinator interactions in tropical rain forests. Annu Rev Ecol
Syst 21:399422
Bawa KS, Krugman SL (1990) Reproductive biology and genetics of tropical trees in
relation to conservation and management. In: Gomez-Pompa A, Whitmore TC,
Hadley M (eds) Rain forest regeneration and management. UNESCO, Paris, and The
Parthenon Publishing Group, Carnforth, pp 119136
Bawa KS, Opler PA (1975) Dioecism in tropical forest trees. Evolution 29:167179
Becker H (1993) Pflanzenzchtung. Ulmer, Stuttgart
Bekessy SA, Ennos RA, Burgman MA, Newton AC, Ades PK (2003) Neutral DNA
markers fail to detect genetic divergence in an ecologically important trait. Biol
Conserv 110:267275
Bennett EL (2001) Timber certification: where is the voice of the biologist? Conserv
Biol 15:308310
Berg CC (1989) Classification and distribution of Ficus. Experientia 45:605611
Berg EE, Hamrick JL (1997) Quantification of genetic diversity at allozyme loci. Can J
For Res 27:415424
Bergmann F (1974) The genetics of some isoenzyme systems in spruce endosperm
(Picea abies). Genetika 6:353360
Bergmann F, Hattemer HH (1998) Isozymes in forest genetic research. In: Mandal AK,
Gibson GL (eds) Forest genetics and tree breeding. CBS, New Delhi, pp 227238
Bergmann F, Ruetz W (1991) Isozyme genetic variation and heterozygosity in random
tree samples and selected orchard clones from the same Norway spruce population.
For Ecol Manage 46:3947
Bergmann F, Gregorius HR, Larsen JB (1990) Levels of genetic variation in European
silver fir (Abies alba). Are they related to species decline? Genetica 82:110
Bermingham E, Dick C (2001) The Inga newcomer or museum antiquity? Science
Bierregaard RO, Lovejoy TE, Kapos V, dos Santos AA, Hutchings RW (1992) The bio-
logical dynamics of tropical rain forest fragments. Bioscience 42:859866
Bierregaard RO, Laurance WF, Sites JW, Lynam AJ, Didham RK, Andersen M, Gascon
C, Tocher MD, Smith AP, Viana VM, Lovejoy TE, Sieving KE, Kramer EA, Restrepo C,
270 References

Moritz C (1997) Key priorities for the study of fragmented tropical ecosystems. In:
Laurence WF, Bierregaard RO (eds) Tropical forest remnants: ecology, management,
and conservation of fragmented communities. University of Chicago Press, Chicago,
pp 515525
Billington HL (1991) Effect of population size on genetic variation in a dioecious
conifer. Conserv Biol 5:115119
Bishir J, Roberds JH (1999) On numbers of clones needed for managing risks in clonal
forestry. For Genet 6:149155
Blattner FR, Weising K, Bnfer G, Maschwitz U, Fiala B (2001) Molecular analysis of phy-
logenetic relationships among myrmecophytic Macaranga species (Euphorbiaceae).
Mol Phylogenet Evol 19:331344
Bolstad PV, Bawa KS (1982) Self incompatibility in Gmelina arborea L. (Verbenaceae).
Silvae Genet 31:1921
Bolstad SB, Kang H, Guries RP, Marty TL (1991) Performance of inter- and intraprove-
nance crosses of jack pine in central Wisconsin. Silvae Genet 48:261266
Booth TH, Turnbull JW (1994) Domestication of lesser-known tropical tree species:
the Australian experience. In: Leakey RRB, Newton AC (eds) Tropical trees: the
potential for domestication and the rebuilding of forest resources. Her Majestys
Stationery Office, London, pp 189194
Boshier D, Mesn JF (1989) Breeding population of Cordia alliodora in Costa Rica. In:
Gibson GL, Griffin AR, Matheson AC (eds) Breeding tropical trees: population struc-
ture and genetic improvement strategies in clonal and seedling forestry. Proceedings of
the IUFRO conference, Pattaya, Thailand, November 1988. Oxford Forestry Institute,
Oxford, and Winrock International, Arlington, pp 406407
Boshier DH (2000) Mating systems. In: Young A, Boshier D, Boyle T (eds) Forest con-
servation genetics. Principles and practice. CSIRO, Collingwood, and CABI, Oxford,
pp 6379
Boshier DH, Chase MR, Bawa KS (1995) Population genetics of Cordia alliodora
(Boraginaceae), a neotropical tree. 2. Mating system. Am J Bot 82:476483
Bouvet JM, Vigneron P (1996) Variance structure in Eucalyptus hybrid populations.
Silvae Genet 45:171177
Boyle TJ (2000) Criteria and indicators for the conservation of genetic diversity. In:
Young A, Boshier D, Boyle T (eds) Forest conservation genetics. Principles and prac-
tice. CSIRO, Collingwood, and CABI, Oxford, pp 239251
Boyle TJB (1994) Introduction to the NATO advanced research workshop. In: Boyle
TJB, Boyle CEB (eds) Biodiversity, temperate forest ecosystems and global change.
Springer, Berlin Heidelberg New York, pp 17
Brewbaker JL, Sorensson CT (1994) Domestication of lesser-known species of the
genus Leucaena. In: Leakey RRB, Newton AC (eds) Tropical trees: the potential for
domestication and the rebuilding of forest resources, Her Majestys Stationery
Office, London, pp 196204
Brigham CA, Schwartz M (2003) Population viability in plants. Springer, Berlin
Heidelberg New York
Bronstein JL, McKey D (1989) The fig/pollinator mutualism: a model system for com-
parative biology. Experientia 45:601604
Bronstein JL, Gouyon P-H, Kjellberg F, Michaloud G (1990) The ecological conse-
quences of flowering asynchrony in monoecious figs: a simulation study. Ecology
References 271

Brown AHD, Hardner CM (2000) Sampling the gene pools of forest trees for ex situ
conservation. In: Young A, Boshier D, Boyle T (eds) Forest conservation genetics.
Principles and practice. CSIRO, Collingwood, and CABI, Oxford, pp 185196
Brown AHD (1978) Isozymes, plant population genetic structure and genetic conser-
vation. Theor Appl Genet 52:145157
Bruenig EF (1996) Conservation and management of tropical rainforests. CABI,
Brunner AM, Busov VB, Strauss SH (2004) Poplar genome sequence: functional
genomics in an ecologically dominant plant species. Trends Plant Sci 9:4956
Bryndum K, Hedegart T (1969) Pollination of teak (Tectona grandis L.). Silvae Genet
Buchert GP, Rajora OP, Hood JV, Dancik BP (1997) Effects of harvesting on genetic diver-
sity in old-growth eastern white pine in Ontorio, Canada. Conserv Biol 11:747758
Burdon RD (1989) When is cloning on an operational scale appropriate? In: Gibson
GI, Griffin AR, Matheson, AC (eds), Breeding tropical trees: population structure
and genetic improvement strategies in clonal and seedling forestry. Proceedings of
the IUFRO conference, Pattaya, Thailand, November 1988. Oxford Forestry
Institute, Oxford, and Winrock International, Arlington, pp 927
Burgess IP, Williams ER, Bell JC, Harwood CE, Owen JV (1996) The effect of out-
crossing rate on the growth of selected families of Eucalyptus grandis. Silvae Genet
Burgman MA, Possingham H, Lynch AJJ, Keith DA, McCarthy MA, Hopper SD, Drury
WL, Passioura JA, Devries RJ (2001) A method for setting the size of plant conser-
vation target areas. Conserv Biol 15:603616
Burley J (1973) Sources and distribution of seedlots in the CFI international prove-
nance trial of Cedrela odorata (including C. mexicana and C. tubiflora). In: Burley J,
Nikles DG (eds) Tropical provenance and progeny research and international coop-
eration. Commonwealth Forestry Institute, Oxford, pp 234237
Burley J, Nikles DG (eds) (1972) Selection and breeding to improve some tropical conifers,
vol 1. Commonwealth Forestry Institute, Oxford, and Department of Forestry, Brisbane
Burley J, Nikles DG (eds) (1973) Selection and breeding to improve some tropical
conifers, vol 2. Commonwealth Forestry Institute, Oxford, and Department of
Forestry, Brisbane
Butcher PA (2004) Genetic mapping in acacias. In: Kumar S, Fladung M (eds)
Molecular genetics and breeding of forest trees. Haworth, New York, pp 411427
Butcher PA, Moran GF (2000) Genetic linkage mapping in Acacia mangium.
2. Development of an integrated map from two outbred pedigrees using RFLP and
microsatellite loci. Theor Appl Genet 101:594605
Butcher PA, Moran GF, Perkins HD (1998) RFLP diversity in the nuclear genome of
Acacia mangium. Heredity 81:205213
Butcher PA, Glaubitz JC, Moran GF (1999) Applications for microsatellite markerks in
the domestication and conservation of forest trees. For Genet Resour 27:3442
Butcher PA, Moran GF, Bell R (2000) Genetic linkage mapping in Acacia mangium.
1. Evaluation of restriction endonucleases, inheritance of RFLP loci and their
conservation across species. Theor Appl Genet 100:576583
Byrne M (2000) Disease threats and the conservation genetics of forest trees. In: Young A,
Boshier D, Boyle T (eds) Forest conservation genetics. Principles and practice. CSIRO,
Collingwood, and CABI, Oxford, pp 159166
272 References

Caballero A (1994) Developments in the prediction of effective population size. Heredity

Campinhos E (1991) Plantations of fast-growing species for tropical areas. Rev For Fr
Hors Ser 5:111120
Campinhos E (1993a) A Brazilian example of a large-scale forestry plantation in a trop-
ical region: Aracruz. In: Davidson J (ed) Proceedings of the regional symposium on
recent advances in mass clonal multiplication of forest trees for plantation pro-
grammes. RAS/91/004 field document no 4. UNDP/FAO Forest Tree Improvement
Project, Los Baos, pp 4659
Campinhos E (1993b) The use/place of clonal multiplication in tree breeding and
propagation programmes. Prerequisites for success and reasons for failure. In:
Davidson J (ed) Proceedings of the regional symposium on recent advances in
mass clonal multiplication of forest trees for plantation programmes. RAS/91/004
field document no 4. UNDP/FAO Forest Tree Improvement Project, Los Baos,
pp 930
Campinhos E (1999) Sustainable plantations of high yield Eucalyptus trees for produc-
tion of fiber: the Aracruz case. New For 17:129143
Campinhos E, Ikemori YK (1989) Selection and management of the basic population
Eucalyptus grandis and E. urophylla established at Aracruz for the long term breed-
ing programme. In: Gibson GI, Griffin AR, Matheson, AC (eds), Breeding tropical
trees: population structure and genetic improvement strategies in clonal and
seedling forestry. Proceedings of the IUFRO conference, Pattaya, Thailand,
November 1988. Oxford Forestry Institute, Oxford, and Winrock International,
Arlington, pp 169175
Cao CP, Finkeldey R, Siregar I, Siregar U, Gailing O (2006) Genetic diversity within and
among populations of Shorea leprosula Miq. and Shorea parvifolia Dyer
(Dipterocarpaceae) in Indonesia detected by AFLPs. Tree Genet Genomics (in press)
Cardoso MA, Provan J, Powell W, Ferreira PCG, De Oliveira DE (1998) High genetic
differentiation among remnant populations of the endangered Caesalpinia echinata
Lam. (Leguminosae Caesalpinioideae). Mol Ecol 7:601608
Casasoli M, Derory J, Morera-Dutrey J, Brendel O, Porth J, Guehl JM, Villani F, Kremer
A (2006) Comparison of quantitative trait loci for adaptive traits for oak and chest-
nut based on an expressed sequence tag consensus map. Genetics 172:533546
Cascante A, Quesada M, Lobo JJ, Fuchs EA (2002) Effects of dry tropical forest frag-
mentation on the reproductive success and genetic structure of the tree Samanea
saman. Conserv Biol 16:137147
Cavers S, Navarro C, Lowe AJ (2003) Chloroplast DNA phylogeography reveals colo-
nization history of a Neotropical tree, Cedrela odorata L., in Mesoamerica. Mol Ecol
Cspedes M, Gutierrez MV, Holbrook NM, Rocha OJ (2003) Restoration of genetic
diversity in the dry forest tree Swietenia macrophylla (Meliaceae) after pasture aban-
donment in Costa Rica. Mol Ecol 12:32013212
Chaix G, Gerber S, Razafimaharo V, Vigneron P, Veghaegen D, Hamon S (2003) Gene
flow estimation with microsatellites in Malagasy seed orchard of Eucalyptus grandis.
Theor Appl Genet 107:705712
Chalmers KJ, Waugh R, Sprent JI, Simons AJ, Powell W (1992) Detection of genetic
variation between and within populations of Gliricidia sepium and G. maculata
using RAPD markers. Heredity 69:465472
References 273

Chamberlain JR, Galway NW, Simons AJ (1996) Population structure in Gliricidia

sepium (Leguminosae) as revealed by isozyme variation. Silvae Genet 45:112118
Chamshama SAO, Mugasha AG, Wate PA (1999) Variation in performance of Eucalyptus
tereticornis provenances at Michafutene, Mozambique. Silvae Genet 48:261266
Changtragoon S (1984). Variation in growth, seed and pollen charcteristics of Pinus
merkusii Jungh. and de Vriese. Masters thesis, Kasetsart University, Bangkok
Changtragoon S, Finkeldey R (1995a) Patterns of genetic variation and characteriza-
tion of the mating system of Pinus merkusii in Thailand. For Genet 2:8797
Changtragoon S, Finkeldey R (1995b) Inheritance of isozymes phenotypes of Pinus
merkusii. J Trop For Sci 8:167177
Changtragoon S, Finkeldey R (2000) Der Nutzen von Genmarkern fr die Erhaltung
forstgenetischer Ressourcen in Thailand. For Snow Landscape Res 75:153162
Chaperon H (1984) Influence of propagation by cuttings on the breeding strategy of
forest trees. In: Barnes RD, Gibson, GL (eds) Provenance and genetic improvement
strategies in tropical forest genetics. Commonwealth Forestry Institute, Oxford, and
Forest Research Centre, Harare, pp 135148
Charlesworth D (2004) Plant evolution: modern sex chromosomes. Curr Biol
Chase M, Kesseli R, Bawa K (1996a) Microsatellite markers for population and conser-
vation genetics of tropical trees. Am J Bot 83:5157
Chase MR, Moller C, Kesseli R, Bawa KS (1996b) Distant gene flow in tropical trees.
Nature 383:398399
Clarke G, Young AG (2000) Introduction: genetics, demography and conservation of
fragmented populations. In: Young AG, Clarke GM (eds) Genetics, demography and
viability of fragmented populations. Cambridge University Press, Cambridge, pp 16
Clement CR, Villachica H (1994) Amazonian fruits and nuts: potential for domestica-
tion in various agroecosystems. In: Leakey RRB, Newton AC (eds) Tropical trees: the
potential for domestification and the rebuilding of forest resources. Her Majestys
Stationery Office, London, pp 230238
Colinvaux PA, De Oliveira PE (2001) Amazon plant diversity and climate through the
Cenozoic. Palaeogeogr Palaeoclimatol Palaeoecol 166:5163
Collevatti RG, Brondani RV, Grattapaglia D (1999) Development and characterization
of microsatellite markers for genetic analysis of a Brazilian endangered tree species
Caryocar brasiliense. Heredity 83:748756
Comps B, Gmry D, Letouzey J, Thibaut B, Petit RJ (2001) Diverging trends between
heterozygosity and allelic richness during postglacial colonization in the European
beech. Genetics 157:389397
Compton SG, Ellwood MDF, Davis AJ, Welch K (2000) The flight heights of chalcid
wasps (Hymenoptera, Chalcidoidea) in a lowland Bornean rain forest: fig wasps are
the high fliers. Biotropica 32:515522
Connel JH, Lowman MD (1989) Low-diversity tropical rain forests: some possible
mechanisms for their existence. Am Nat 134:88119
Cook AG, Janetos AC, Hinds WT (1990) Global effects of tropical deforestation:
towards an integrated perspective. Environ Conserv 17:201212
Cook JM (1993) Sex determination in the Hymenoptera: a review of models and evi-
dence. Heredity 71:421435
Cook JM, Power SA (1996) Effects of within-tree flowering asynchrony on the dynam-
ics of seed and wasp production in an Australian fig species. J Biogeogr 23:487493
274 References

Cordeiro NJ, Howe HF (2001) Low recruitment of trees dispersed by animals in

African forest fragments. Conserv Biol 15: 17331741
Cornelius J (1994) The effectiveness of plus-tree selection for yield. For Ecol Manage
Cossalter C (1989) Genetic conservation: a cornerstone of breeding strategies. In:
Gibson GI, Griffin AR, Matheson AC (eds) Breeding tropical trees: population struc-
ture and genetic improvement strategies in clonal and seedling forestry. Proceedings
of the IUFRO conference, Pattaya, Thailand, November 1988. Oxford Forestry
Institute, Oxford, and Winrock International, Arlington, pp 2838
Coster C, Eidmann FE (1934) Selectie-onderzoek van den Djati (Tectona grandis L.f.).
Tectona 27:145
Coster C, Hardjowasono MS (1935) Selectieonderzoek van den Djati (Tectona grandis
L.f.) II. De groei gedurende het tweede levensjaar. Tectona 28:321
Cotterill PP (1990) Numbers of families and progeny required for provenance testing.
Silvae Genet 39:8283
Cox PA, Elmqvist T (2000) Pollinator extinction in the Pacific islands. Conserv Biol
Crawford AJ (2003) Huge populations and old species of Costa Rican and Panamanian
dirt frogs inferred from mitochondrial and nuclear gene sequences. Mol Ecol
Crow JF, Kimura M (1970) An introduction to population genetics theory. Harper and
Row, New York
Cruz A (1981) Bird activity and seed dispersal of a montane forest tree (Dunalia
arborescens) in Jamaica. Reprod Bot 3444
Cruzan MB (1998) Genetic markers in plant evolutionary ecology. Ecology 79:400412
Curtis JT (1956) The modification of midland-grasslands and forests by man. In:
Thomas WL (ed) Mans role in changing the face of the earth. University of Chicago
Press, Chicago, pp 721736
Darby HC (1956) The clearing of the woodland in Europe. In: Thomas WL (ed) Mans
role in changing the face of the earth. University of Chicago Press, Chicago, pp
Darwin C (1859) On the origin of species by means of natural selection. Murray, London
Darwin C (1875) The variation of animals and plants under domestication. Two vol-
umes. (2nd edn, Pickering and Chatto, London, 1988). Murray, London
Davidson J (1995) Training manual on tree breeding and propagation. RAS/91/004
field document no. 21. UNDP/FAO Forest Tree Improvement Project, Los Baos
Dawson IK, Powell W (1999) Genetic variation in the Afromontane tree Prunus
africana, an endangered medicinal species. Mol Ecol 8:151156
Dawson IK, Waugh R, Simons AJ, Powell W (1997) Simple sequence repeats provide a
direct estimate of pollen-mediated gene dispersal in the tropical tree Gliricidia
sepium. Mol Ecol 6:179183
Dayanandan S, Dole J, Bawa K, Kesseli R (1999) Population structure delineated with
microsatellite markers in fragmented populations of a tropical tree, Carapa guia-
nensis (Meliaceae). Mol Ecol 8:15851592
Dellaporta SL, Calderon-Urrea A (1993) Sex determination in flowering plants. Plant
Cell 5:12411251
Delwaulle JC (1989) Plantation clonales an Congo point des recherches sur le choix des
clones dix ans aprs les premires plantation. In: Gibson GL, Griffin AR, Matheson
References 275

AC (eds) Breeding tropical trees: population structure and genetic improvement

strategies in clonal and seedling forestry. Proceedings of the IUFRO conference,
Pattaya, Thailand, November 1988. Oxford Forestry Institute, Oxford, and Winrock
International, Arlington, pp 431434
de Nettancourt D (2001) Incompatibility and incongruity in wild and cultivated
plants. Springer, Berlin Heidelberg New York
den Njis HCM, Oostermeijer JGB (1997) Reproductive biology and gene flow in a
fragmented landscape. Bocconea 7:153165
Deutscher Forstverein (1986) Erhaltung und nachhaltige Nutzung tropischer
Regenwlder. Weltforum, Munich
De Vecchi Pellati E (1970) Evolution and importance of land races in breeding. In:
Proceedings of the 2nd world consultation on forest genetics and forest tree breed-
ing. FAO, Rome, pp 12661278
Dick CW (2001) Genetic rescue of remnant tropical trees by an alien pollinator. Proc
R Soc Lond Ser B 268:23912396
Dick CW, Etchelecu G, Austerlitz F (2003) Pollen dispersal of tropical trees (Dinizia
excelsa:Fabaceae) by native insects and African honeybees in pristine and frag-
mented Amazonian rainforest. Mol Ecol 12:753764
Diniz-Filho JAF, Telles MPC (2002) Spatial autocorrelation analysis and the
Identification of operational units for conservation in continuous populations.
Conserv Biol 16:924935
Dobzhansky T (1968) Adaptedness and fitness. In: Lewontin RC (ed) Population biol-
ogy and evolution. Syracuse University Press, Syracuse, pp 109121
Dobzhansky T, Ayala FJ, Stebbins GL, Valentine JW (1977) Evolution. Freeman, San
Doligez A, Joly HI (1997) Mating system of Carapa procera (Meliaceae) in the French
Guiana tropical forest. Am J Bot 84:461470
Donaldson J, Nnni I, Zachariades C, Kemper J (2002) Effects of habitat fragmentation
on pollinator diversity and plant reproductive success in renosterveld shrublands of
South Africa. Conserv Biol 16:12671276
Dow BD, Ashley MV (1998) High levels of gene flow in bur oak revealed by paternity
analysis using microsatellites. J Hered 89:6270
Durel CE, Kremer A (1995) Hybridization after self-fertilization: a novel perspective
for maritime pine breeding. For Genet 2:117120
Dvorak WS, Wright JA (1994) Offspring-parent correlation for wood density of Pinus
tecumanii between native and exotic environments. Can J For Res 24:15931596
Dvorak WS, Hamrick JL, Hodge GR (1999) Assessing the sampling efficiency of ex situ
gene conservation efforts in natural pine populations in central America. For
Genetics 6:2128
Eguiarte LE, Perez-Nasser N, Piero D (1992) Genetic structure, outcrossing rate and
heterosis in Astrocaryum mexicanum (tropical palm): implications for evolution and
conservation. Heredity 69:217228
Eguiarte LE, Burquez A, Rodriguez J, Martinez-Ramos M, Sarukhan J, Piero D (1993)
Direct and indirect estimates of neighborhood and effective population size in a
tropical palm, Astrocaryum mexicanum. Evolution 47:7587
Eldridge KG (1975) Broadleaved Seed Orchards. Part C Eucalyptus species. In:
Faulkner R (ed) Seed orchards. Her Majestys Stationery Office, London,
pp 134139
276 References

Eldridge KG (1990) Conservation of forest genetic resources with particular reference

to Eucalyptus species. Commonw For Rev 69:4553
Eldridge KG, Davidson J, Harwood C, van Wyk G (1994) Eucalypt domestication and
breeding. Oxford University Press, Oxford
Eliott F, Shepherd M, Henry RJ (2005) Verification of interspecific pine hybrids using
paternally inherited chloroplast microsatellites. For Genet 12:8187
El-Kassaby YA (1999) Impacts of industrial forestry on genetic diversity of temperate
forest trees. In: Mtys C (ed) Forest genetics and sustainability. Kluwer, Dordrecht,
pp 155169
El-Kassaby YA (2000) Effects of forest tree domestication on gene pools. In: Young A,
Boshier D, Boyle T (eds) Forest conservation genetics. Principles and practice.
CSIRO, Collingwood, and CABI, Oxford, pp 197213
El-Kassaby YA, Askew GR (1998) Seed orchards and their genetics. In: Mandal AK,
Gibson GL (eds) Forest genetics and tree breeding. CBS, New Dehli, pp 103111
El-Kassaby YA, Ritland K (1998) Isozymes: techniques, interpretation and application
to forest genetics. In: Mandal AK, Gibson GL (eds) Forest genetics and tree breed-
ing. CBS, New Dehli, pp 213226
El-Kassaby YA, Dunsworth BG, Krakowski J (2003) Genetic evaluation of alternative
silviculture systems in coastal montane forests: western hemlock and amabilis fir.
Theor Appl Genet 107:598610
Ellstrand NC (1992) Gene flow among seed plant populations. New For 6:241256
Endo M (1994) Twelve years of contribution to reforestation in the Andean region of
Colombia. For Ecol Manage 63:219233
Engelmann F (1994) In vitro systems for the conservation and utilization of germplasm.
In: Leakey RRB, Newton AC (eds) Tropical trees: the potential for domestication and
the rebuilding of forest resources. Her Majestys Stationery Office, London, pp 4952
Ennos RA (1996) Utilising genetic information in plant conservation programmes. In:
Hochberg ME, Clobert J, Barbault R (eds) Aspects of the genesis and maintenance
of biological diversity. Oxford University Press, Oxford, pp 278291
Enquete Kommission Vorsorge zum Schutz der Erdatmosphre des deutschen
Bundestages (1990) Schutz der Tropenwlder. Eine internationale Schwerpunktauf-
gabe. Economia und Mller, Bonn
Eriksson G (1995) Which traits should be used to guide sampling for gene resources?
In: Baradat P, Adams WT, Mller-Starck G (eds) Population genetics and genetic
conservation of forest trees. SPB, Amsterdam, pp 349358
Eriksson G, Ekberg I (2001) An introduction to forest genetics. SLU Uppsala, Uppsala
Eriksson G, Lindgren D (1975) Ngra genetiska reflexioner kring plantsortering. Sver
Skogsvrdsfrb Tidskr 73:413426
Eriksson G, Andersson S, Eiche V, Ifver J, Persson A (1980) Severity index and transfer
effects on survival and volume production of Pinus sylvestris in Northern Sweden.
Stud For Suec 156:532
Eriksson G, Namkoong G, Roberds JH (1993) Dynamic gene conservation for uncer-
tain futures. For Ecol Manage 62:1537
Erwin TL (1988) The tropical forest canopy. The heart of biotic diversity. In: Wilson
EO, Peter FM (eds) Biodiversity. National Academy Press, Washington, pp 123129
Erwin TL (1997) Biodiversity at its utmost: tropical forest beetles. In: Reaka-Kudla ML,
Wilson DE, Wilson EO (eds) Biodiversity II. Understanding and protecting our bio-
logical resources, Henry, Washington, pp 2740
References 277

EU (2000) Council Directive 1999/105/EC of 22 December 1999 on the marketing of

forest reproductive material. Official Journal of the European Communities L 11/17
of January 15, 2000
Evans J, Turnbull JW (2004) Plantation forestry in the tropics. The role, silviculture,
and use of planted forests for industrial, social, environmental, and agroforestry pur-
poses. Oxford University Press, Oxford
Evans LT (1980) The natural history of crop yield. Am Sci 68:388397
Falconer DS, MacKay TFC (1996) Introduction to quantitative genetics. Pearson,
FAO (1986) Databook on endangered tree and shrub species and provenances. FAO
forestry paper no 77. FAO, Rome
FAO (1992) Mixed and pure forest plantations in the tropics and subtropics. FAO
forestry paper no 103. FAO, Rome
FAO (1993) Conservation of genetic resources in tropical forest management.
Principles and concepts. FAO forestry paper no 107. FAO, Rome
FAO (1994) Biotechnology in forest tree improvement with special reference to devel-
oping countries. FAO forestry paper no 118. FAO, Rome
FAO (2002) Global forest resources assessment 2000. Main report. FAO forestry paper
no 140. FAO, Rome
Fedorov AA (1966) The structure of the tropical rain forest and speciation in the
humid tropics. J Ecol 54:111
Feilberg L, Segaard B (1975) Historical review of seed orchards. In: Faulkner R (ed)
Seed orchards. Her Majestys Stationery Office, London, pp 18
Felker P (1994) Capturing and managing the genetic variation in Prosopis spp. for eco-
nomically useful characters. In: Leakey RRB, Newton AC (eds) Tropical trees: the
potential for domestication and the rebu