Microsatellite Dna in Fishes

Reviews in Fish Biology and Fisheries 7, 331±363 (1997)
Microsatellite DNA in ®shes

M I C H A E L O ' C O N N E L L and J ONATH A N M . WR I GH T
Marine Gene Probe Laboratory, Department of Biology, Dalhousie University, Halifax, NS, Canada B3H 4J1
Contents
Abstract page 331
General introduction 332
Minisatellites 333
Historical overview of minisatellites and ®sheries research
Application of minisatellites to ®sheries research
Disadvantages=problems associated with minisatellites
Microsatellites 336
Introduction
Isolation protocols, potential problems and technical considerations
Applications of microsatellites
Population genetics
Parentage and kinship analysis
Genome mapping
Potential problems=issues
Mutation models
Potential for scoring error
Null alleles
Sample size required
Concluding remarks 354
Acknowledgements 356
References 356
Abstract
For the last 30 years, attempts have been made to discriminate among ®sh populations by
using molecular markers. Although some techniques have proved successful in certain
circumstances, the consistent trend to newer markers among ®shery geneticists highlights
the general lack of resolving power observed with older technologies. The last decade
has seen the increasing use of satellite DNA in investigations of genetic variability and
divergence. Applications to ®sh and ®sheries-related issues initially concentrated on
minisatellite single-locus probes. Although minisatellites have successfully addressed a
number of ®shery-related questions, this class of satellite DNA has not been widely
adopted by ®shery geneticists. Most of the current research effort is concentrated on
another class of satellite DNA called microsatellites. The large interest in microsatellite
loci is largely due to the very high levels of variability that have been observed and the

Author to whom correspondence should be addressed. E-mail: jmwright@is.dal.ca
0960±3166 # 1997 Chapman & Hall
332 O'Connell and Wright
ability to investigate this variation using PCR technology. The isolation and application
of microsatellites to research ®elds as diverse as population genetics, parentage analyses
and genome mapping are reviewed. Despite the undisputed advantages that the marker
possesses, there are a number of potential problems associated with investigating
variation at microsatellite loci. Statistical considerations (e.g. appropriate sample sizes,
number of loci and the mutation model assumptions on which the estimate is based) have
not been considered in detail yet and the problems are often exacerbated in ®sh species,
as some species show very large numbers of alleles at microsatellite loci. These issues
and others, e.g. null alleles, are reviewed and possible solutions are proposed.
General introduction
The conservation of genetic variation is an essential component of many species
management programmes. Ultimately, it is genetic variation that allows species to adapt
to changing environmental conditions and respond to selection=breeding programmes. To
manage any biological resource effectively, researchers must identify the level of genetic
variation within and among populations. Although some genetic markers are available,
for many species additional informative loci are required for population genetics studies
and for long-term endeavours such as genome mapping, localization of quantitative trait
loci and the implementation of marker-assisted selection regimes.
The potential of molecular markers to ®sheries management has long been
recognized (review: Utter, 1991). The early appreciation of the value of genetic
markers is in a large part due to the challenges of observing behaviour and migration
patterns in an aquatic=marine environment. Early studies on the molecular phenotypes
used blood group polymorphisms to discriminate between spatially discrete populations
of ®sh (Sick, 1961 cited in Ward and Grewe, 1994). However, the patterns were
dif®cult to interpret and the variation identi®ed could not be attributed to any single
locus. Because of the problems of interpretation, research turned to speci®c
histochemical stain procedures (Hunter and Markert, 1957). Staining for speci®c
proteins was used in association with starch gel electrophoresis and permitted the
detection of allozyme variation (Harris and Hopkinson, 1976). Allozyme electrophoresis
was adopted enthusiastically by ®shery scientists whose main objective was to
investigate population structuring (Moller, 1970; Avise and Smith, 1974).
In order to increase the number of variable genetic markers available, workers turned
to a direct investigation of DNA sequence in the late 1970s and early 1980s. Initially,
this work focused on the mitochondrial DNA (mtDNA) molecule using restriction
fragment length polymorphism (RFLP) analysis (Lansman et al., 1981), with later
research concentrating on sequencing speci®c areas of the molecule (e.g. Bernatchez et
al., 1992). Randomly ampli®ed polymorphic DNA (RAPDs) (Welsh and McClelland,
1990; Williams et al., 1990) and RFLP analyses of nuclear DNA (Pogson et al., 1995)
have also been applied to ®sheries. However, the last decade has seen an increasing
emphasis on the use of variable number tandem repeat loci (VNTRs) (Tautz and Renz,
1984; Jeffreys et al., 1985a,b; Tautz, 1989) which consist of short, tandemly repeated
DNA sequences, usually distributed at random throughout the genome. Initial studies
using nuclear markers concentrated on multilocus DNA ®ngerprinting, which gave
complex banding patterns (pro®les simliar to bar codes) unique to each individual.
Dif®culties in interpretation and between-gel comparisons prompted the move to single-
VNTR DNA and ®sheries research 333
locus VNTR markers. J.B. Taggart and colleagues (Queen's University, Belfast, UK) and
workers at the Marine Gene Probe Laboratory (Dalhousie University, Halifax, Canada)
were the ®rst to develop this technology for ®sheries applications (Taggart and
Ferguson, 1990; Bentzen et al., 1991; Prodohl et al., 1994a,b; Ferguson et al., 1995).
Attention has more recently focused on a second class of VNTR loci, microsatellites,
that are becoming the tool of choice for many wildlife and captive management
applications (Wright and Bentzen, 1994).
The application of microsatellites to ®sheries research issues is the major focus of
this review. The paper concentrates on reported literature in both capture ®sheries and
aquaculture. The application of minisatellite data and the potential problems of this
approach are discussed initially to provide a background to the microsatellite data. The
review discusses some of the problems=issues associated with interpreting microsatellite
data, e.g. null alleles, sample sizes, and mutation models. More detailed descriptions of
alternate molecular tools, e.g. allozymes, mtDNA, single copy nuclear RFLPs, are
available elsewhere (Carvalho and Hauser, 1994; Ward and Grewe, 1994; Skibinski,
1994).
Minisatellites
H I S TO R I C A L OV E RV I E W O F M I N I S AT E L L I T E S A N D F I S H E R I E S R E S E A R C H
Jeffreys et al. (1985a) ®rst used minisatellites as molecular markers, following their
initial discovery by Wyman and White (1980). The simultaneous investigation of
variation at many minisatellite loci, i.e. multilocus DNA ®ngerprinting, generates
complex banding patterns speci®c to an individual (Fig. 1). Multilocus ®ngerprinting
studies of pedigrees have demonstrated that this variation is inherited in a Mendelian
fashion (Wright, 1993). Multilocus ®ngerprinting was ®rst used in forensic applications
and in paternity assessment in humans (Jeffreys et al., 1985b; Burke and Bruford, 1987).
However, the technique was soon applied to a range of ®sh species and ®shery-related
questions (Carter et al., 1991; Harris et al., 1991; O'Reilly and Wright, 1995 and
references therein). Nevertheless, as the bands cannot be assigned to any particular locus
(see Amos et al., 1991, for notable exception), it is generally not possible to test mating
patterns within populations against Hardy±Weinberg expectations. Moreover, problems
were often reported with reproducibility across gels (e.g. Bentzen et al., 1991).
In an effort to offset the dif®culties in interpreting multilocus ®ngerprints, work
concentrated on developing single-locus minisatellite probes (Fig. 1) (Wong et al.,
1987). To develop these probes, DNA must be isolated and cloned from the study
organism (unless probes from closely related species are available). Taggart and
Ferguson (1990) and Bentzen et al. (1991) were the ®rst to report on the development
of single-locus probes for salmonid and tilapia species. Taggart and Ferguson (1990)
and Taggart et al. (1995a) reported very high numbers of alleles in Atlantic salmon
(Salmo salar, Salmonidae) not previously observed using other marker systems. Later
studies include Harris and Wright (1995) and Prodohl et al. (1994a,b), who reported on
the development of single-locus probes for Nile tilapia (Oreochromis niloticus,
Cichlidae) and brown trout (Salmo trutta, Salmonidae) respectively. The technical
demands and costs associated with developing and assaying single-locus minisatellite
DNA variation can be appreciated when one considers that very few groups have
published in this area (e.g. Taggart et al., 1995a; Taylor, 1995; McGregor et al., 1996).
(a)
1 2 3 4 5 6 7 8
(b)
1 2 3 4 5 6 7 8
Fig. 1. (a) Diagram showing the type of complex banding patterns observed using multilocus
®ngerprinting. (b) Diagram illustrating the type of variation observed using primers or a probe
developed for a single locus. Note that there are ®ve alleles represented in the illustration. Lane 1
reveals a homozygous pattern. All other lanes reveal heterozygous genotypes.
Moreover, owing to technical problems, most of the published studies have used only a
subset (one to two) of the probes available.
A P P L I C AT I O N O F M I N I S AT E L L I T E S TO F I S H E R I E S R E S E A R C H
Despite the technical problems associated with using minisatellite probes, they have
proved very successful in discriminating among ®sh populations in a number of cases.
Taggart et al. (1995b), in a survey of Atlantic salmon populations from throughout
Europe and North America, demonstrated signi®cant genetic differentiation between
Atlantic salmon from both sides of the North Atlantic. Ferguson et al. (1995)
demonstrated that single-locus probes can be very useful in cases where species have
undergone population bottlenecks, e.g. Icelandic brown trout populations where other
marker systems have revealed limited levels of variation. Taylor (1995) investigated
minisatellite DNA variability within and among populations of resident rainbow trout
(Oncorhynchus mykiss, Salmonidae) and sea-run steelhead (O. mykiss) from British
Columbia and Alaska. Allelic diversity was moderate to high for the two probes used
(61% for Ssa1 (a multilocus probe); 80% for T34), temporal stability of allele
frequencies was reported over two years and signi®cant heterogeneity in allele
frequencies was demonstrated among regions. In contrast to allozyme surveys of similar
size (e.g. Reisenbichler and Phelps, 1989), signi®cant differentiation was also observed
for the minisatellite data among tributaries within a catchment, which demonstrated the
increased resolution provided by minisatellites over other non-VNTR-based marker

systems.
Atlantic salmon populations have also been examined on a ®ne geographic scale
using single-locus minisatellite DNA markers. Galvin et al. (1995a) investigated 1252
Atlantic salmon from the river Shannon in Ireland and used the genotypes generated
over three minisatellite loci to trace the tributary of origin for a further 250 returning
adults. An initial analysis compared the minisatellite data with the allozyme data and
demonstrated the minisatellite data to be more accurate (de®ned as the difference
between the known value and the value predicted by mathematical simulations) than the
allozyme data for a given sample size. The study also revealed that a large portion of
the upper catchment area did not contribute to the adult run. This level of resolution
was previously not possible or feasible using less polymorphic marker systems and
demonstrates the obvious potential of this type of analysis to ®sheries management.
Another area where the increased variability of single-locus minisatellites is being
exploited is the assignment of parents to offspring. Ferguson et al. (1995) reported on
an experiment where the reproductive success of farmed Atlantic salmon in the wild
was compared with that of wild ®sh and of a third group derived from crosses between
farmed and wild ®sh. By using six minisatellite loci, 86% of individuals could be
assigned to individual families and 91% to their study group. Research into the
reproductive success of Atlantic salmon mature parr using six single-locus minisatellite
probes has also been carried out (P. Moran, Univ. of Oviedo, Spain, pers. comm.). This
research revealed that, in the presence of adult female and male couples, precocious
parr fertilized 24.7±89.3% of the eggs. This high level of resolution demonstrates the
utility of this marker system to captive management=selection programmes where the
identi®cation of parentage can be quickly assessed using genetic data.
D I S A DVA N TAG E S =P RO B L E M S A S S O C I AT E D W I T H M I N I S AT E L L I T E S
Standard single-locus minisatellite analyses require blotting and therefore, reasonable
quantities of high-quality DNA are required. To circumvent these restrictions, appropriate
minisatellite loci have been adapted for PCR analysis. Galvin et al. (1995b,c)
demonstrated in preliminary analyses how this approach could be used to survey
variation in gadoids. Signi®cant differences among Atlantic cod (Gadus morhua,
Gadidae) populations were observed at the locus investigated. Moreover, the level of
heterozygosity reported (94% heterozygosity) was extremely high. The polymerase chain
reaction (PRC) analysis involves the ampli®cation of the minisatellite of interest and
running the ampli®cation products in an agarose gel. The products are scored after
staining the gel in ethidium bromide (Galvin et al., 1995c). One problem with this
approach is that larger alleles will amplify poorly, if at all (large allele dropout), in the
PRC reaction. However, Galvin et al. (1995b) suggest this is not a major problem, at
least for the locus they investigated, as their study revealed the sample populations to be
in Hardy±Weinberg equilibrium. Another limitation is the complex mutation processes at
minisatellite loci (Wright, 1993; O'Reilly and Wright, 1995). The complexity leads to
alleles thay may not differ in size from each other by a discrete unit length and this can
make accurate scoring of alleles very dif®cult, if not impossible. Furthermore, the larger
size of the alleles observed at minisatellite loci can also lead to an inadvertent lumping
of alleles and=or an increase in scoring error. This has prompted some workers to
deliberately group (bin) alleles within discrete size ranges (Bentzen et al., 1991).
Nevertheless, existing data demonstrate that single-locus minisatellites can provide

valuable data for parentage determination and population analyses.
Microsatellites
I N T RO D U C T I O N
In an effort to avoid some of the problems associated with minisatellites, research on
VNTR loci has turned increasingly to microsatellites. Microsatellites or simple sequence
repeats (SSRs) consist of short (1±6 base pair, bp) tandem arrays (Tautz and Renz,
1984; Tautz, 1989). SSRs have generally been investigated by radioactively labelling one
of two primers that are complementary to the ¯anking sequence on either side of the
repeat unit array. These primers are used in a PCR reaction to amplify the repeat unit
array. Length differences, due to the variable number of repeats among samples, are
resolved by running out the ampli®cation products on a sequencing gel. The gel is then
dried, exposed to X-ray ®lm and usually developed overnight. Reference to standards
(e.g. M13 sequence and samples with known genotype) and=or an allelic ladder
commonly determine allele sizes.
These loci appear to be highly abundant and dispersed throughout the genome, in
contrast to some reports for minisatellites which tend to be clustered in telomeric
regions of chromosomes (Jeffreys et al., 1987, 1991; Royle et al., 1988). SSRs are
thought to occur approximately once every 10 kbp while minisatellite loci occur once
every 1500 kbp in ®sh species (Wright, 1993), which suggests that SSRs may be more
useful for genome mapping studies. However, these distribution estimates are based on
broad assumptions and likely underestimate the real distribution of SSRs because the
numbers reported in the literature are usually based on a single repeat unit type, e.g.
(GT), and do not consider the interspersion of other types of microsatellite arrays. SSRs
are relatively easy to isolate compared with minisatellites, sample DNA can be isolated
quickly because labour-intensive phenol=chloroform steps can generally be eliminated
in favour of a simpler form of DNA extraction (e.g. McGregor et al., 1996), and minute
quantities of tissue can be used for most PCR assays. Moreover, there is potential for
signi®cant increases in the number of samples that can be genotyped in a day (without
adversely affecting quality and accuracy) using automated ¯uorescent sequencers (e.g.
ABI 373A, VISTRA, LICOR or ALFexpressTM (Phamacia)) and imaging systems (e.g.
Molecular Dynamic's STORM TM or the Hitachi FMBIO system) (O'Reilly and Wright,
1995; O'Reilly et al., 1996). For applications where a large number of loci are required,
such as genome mapping or identi®cation of quantitative trait loci (QTL), micro-
satellites offer a powerful alternative to other marker systems. The remainder of the
review will focus on microsatellites, their isolation, and application to studies in pop-
ulation genetics, genome mapping, paternity assessment and captive management=
selection programmes. The review also discusses the potential problems (in inter-
pretation and development) associated with microsatellites.
I S O L AT I O N P ROTO C O L S , P OT E N T I A L P RO B L E M S A N D T E C H N I C A L C O N S I D E R AT I O N S
Descriptions of protocols used in the isolation and characterization of dinucleotide repeat
loci are common (review: O'Reilly and Wright, 1995). Brie¯y DNA is digested, size
fractionated on an agarose gel and a particular size range is excised from the gel (usually
around 300±600 bp). The recovered DNA is ligated to a vector, e.g. pUC18. This is used
to transform cells and construct a partial genomic library which is screened with a repeat
unit probe, e.g. GT(n) . The clones that hybridize to the GT(n) probe are sequenced and
primers are designed to complement unique sequences on either side of the array (Fig. 2).
Although the methodology is simple, there are a few key steps that can greatly increase
the number of microsatellites isolated from any library. The following list illustrates
technical considerations and problems associated with constructing microsatellite
libraries.
1. Use good-quality (puri®ed using an organic extraction protocol) DNA when

constructing a library.
2. Construct a large library (.20 000 colonies) if possible. Note: Libraries can be made
from 50 ìg of DNA. However, digesting twice this amount can greatly reduce the
Vector e.g. pUC18
Transformation e.g. DH5 α cells
Pick positives (black circles)

and sequence
ACGT ACGT ACGT
∗ ∗ ∗ Sequencing
gel
Design primers from sequence

flanking the repeat unit array
Fig. 2. Generalized diagram illustrating the isolation of microsatellites. The oblong arrays of
rectangles represent the microsatellites within the initial genomic digest. This DNA is cloned into a
vector and used to transform competent cells. The transformed colonies are lifted onto nylon
membranes and the target sequences are identi®ed with a probe, e.g. (GT15 ), represented by the black
circles. The DNA from the positively hybridizing colonies is sequenced and primers (illustrated by the
arrows) adjacent to the array (represented by ) are designed.
time taken to develop primers for microsatellite loci (particularly if the genomic
DNA has been digested with only a single enzyme, e.g. MboI).
3. Check the genomic digest with the individual enzymes to be used for multiple copy
elements, e.g. AluI elements, before proceeding with cloning and screening.
4. Keep PCR product size to a minimum: . 220 bp alleles are often fuzzy and faint.
This is particularly important if dealing with old or degraded tissue, e.g. otoliths or
scales, where products 120 bp often fail to amplify.
5. Most applications (excluding mapping) should concentrate on isolating tri- and
tetranucleotide primers if possible.
6. Signature (stuttering) patterns and, more seriously, the ability to amplify loci depends
on quality of primer synthesis. Thus, when a dependable supplier has been identi®ed,
changes should not be made unless this is completely unavoidable.
7. Invest time and resources at the development stage to identify loci that amplify
consistently. Avoid the temptation to use the ®rst polymorphic markers identi®ed.
In an effort to reduce scoring error and produce PCR products that can be easily
scanned=analysed by automated scoring systems, workers are increasingly switching to
loci with increased repeat unit length. These loci generally `stutter' considerably less
than dinucleotide repeat-based loci, which facilitates the accurate typing of alleles. The
majority of dinucleotide microsatellite alleles are observed as a series of bands, and not
as a single discrete band. Stuttering is the term used to describe this phenomenon,
which is thought to occur through slipped-strand mis-pairing (O'Reilly and Wright,
1995). There are a number of enrichment protocols outlining how to isolate tri- and
tetranucleotide based loci (Ostrander et al., 1992; Karagyozov et al., 1993; Kijas et al.,
1994; Armour et al., 1995; Edwards et al., 1995; Waldbieser, 1995). Although all
require considerably more work than the general `shotgun cloning' approach used to
isolate dinucleotide repeat loci, the bene®ts accrued from using tri- and tetranucleotide
loci greatly outweigh any initial developmental dif®culties (Figs 3, 4).
The application of tri- and tetranucleotide loci will probably shortly overtake
dinucleotide loci (excluding mapping studies) because of the ease of scoring and greater
potential for automation. Related to this is the reduced technical demands and cost of
assaying variation at this class of loci once isolated. Tetranucleotide loci can usually be
investigated using ethidium bromide, silver staining or other non-radio-isotopic staining
techniques, e.g. SYBR greenTM (FMC BioProducts, Rockland, ME), which can be used
in conjunction with high-resolution agarose, e.g. MetaphorTM agarose (FMC Bio-
Products, Rockland, ME).
A P P L I C AT I O N S O F M I C RO S AT E L L I T E S
Population genetics
Prudent management decisions should be based on an understanding of the stock
structure of the biological resource in question. However, delimiting the putative
boundaries of such stocks in marine=aquatic environments is a complex and dif®cult task
(see Carvalho and Hauser, 1994, for an excellent discussion of this area). Genetic
markers have been used to de®ne the underlying stock structure of many ®sh species, in
particular salmonids. The very high levels of variability associated with microsatellites,
the speed of processing and the potential to isolate large numbers of loci provides
Ligation of artificial
linkers to genomic
digest
Genomic DNA digest

with attached artificial Nylon membrane with
linkers fixed tri-, tetranucleotide
target sequence(s)
Hybridization of target
sequence to bound probe
Wash to remove unbound DNA.

Target/repeat sequences
remain bound to probe
Recover target DNA and

amplify for enriched sequence
using primers complementary
to synthetic linkers
Clone into vector, construct library,

re-screen and sequence positives
Fig. 3. A schematic of the enrichment procedure described by Armour et al. (1995). The oblong
tracks represent the microsatellites within the initial genomic digest while the black rectangles
represent the arti®cial linkers, which are ligated to the termini of the digest fragments.
a marker system capable of detecting differences among closely related populations

(Table 1).
One of the ®rst species to be investigated for within and among population variability
using microsatellites was rainbow trout (Nielsen et al., 1994). This survey revealed
similar patterns of differentiation for mtDNA and the microsatellite locus employed.
Populations of potadromous rainbow trout from Lake Ontario have also been
investigated using microsatellite loci and mtDNA (Dueck, 1994; O'Connell et al.,
1996a). A comparison of marker sets revealed that the number of mtDNA haplotypes
was similar to the number of alleles observed at microsatellite loci, although the
single and combined microsatellite loci data revealed signi®cantly higher levels of
Variable length inserts

cloned into vector
Universal primers used

in asymmetric PCR to
generate single-stranded
fragments
Hybridized to tri-,
tetranucleotide probe
Washed to release
non-homologous DNA fragments
Alkaline denatured to
release complementary
fragments
Re-amplify to get
double-stranded
product
Clone into vector and sequence

(secondary screen optional)
Fig. 4. An illustration describing an enrichment procedure using an oligonucleotide probe bound to

magnetic particles (Kijas et al., 1994). The array of tiny rectangles represents the target sequence, e.g.
(GATA), the black rectangles represent the universal primer sites, the hatched circles show the biotin
molecules attached to the oligonucleotide probe. The black circles denote streptavidin-coated magnetic
beads.
differentiation. The higher level of genetic divergence observed for microsatellite data
in this case is consistent with sex-biased dispersal and=or the reproductive success of
precocious male parr (as mtDNA is inherited primarily through the female line),
although such a result most likely re¯ects a random or chance event. An example of a
random event would be where founding individuals in a new population shared allele
frequencies at some loci with an adjacent population. Although the allele frequencies at
the majority of loci would differ between populations, this would not be apparent if
only a few loci were investigated. This type of random event can usually be excluded
by sampling several loci. Studies using more than one data set to describe population
structure have the potential to describe population interactions more fully than would be
possible by using a single marker set. However, it is important to consider the levels of
variability of the marker systems employed. If mtDNA is less variable than
microsatellites, an increased level of microsatellite DNA structuring, relative to mtDNA
VNTR DNA and ®sheries research
Table 1. Comparative advantages and disadvantages of some commonly applied molecular tools
Technique Technical requirements Cost Storage Potential for
requirements
Development Screening Development Screening Genome Parentage Population
mapping assessment genetics
Allozymes Low Low Low Moderate High Low Low Moderate/high
MtDNA1 Low Low Nil Moderate Low Nil Low/moderate Moderate/high
RAPDs2 Low Low Nil Low High High Moderate Low
ScnDNA3 High High High Moderate/high Moderate High Low Moderate
ESTs4 High High High Moderate Moderate High Low Low
MLF 5 Moderate Moderate Low High Moderate Nil High Low
SLPs6 Moderate/high High High Moderate Moderate High High High
SSRs7 Moderate/high High High High Low High High High
SSRs8 High Low High/moderate Low Low High High High
1
PCR-ampli®ed mitochondrial DNA.
2
Randomly ampli®ed polymorphic DNA.
3
Single-copy nuclear DNA restriction fragment length polymorphisms.
4
Expressed sequence tags.
5
Multilocus ®ngerprinting.
6
Single-locus minisatellite probes.
7
Dinucleotide simple sequence repeats (microsatellites).
8
Tri-, tetranucleotide simple sequence repeats (microsatellites).
341
structuring, could simply re¯ect the increased mutation rate of the former and not
represent a demographic phenomenon. Caution also has to be exercized when drawing
conclusions about differences between mtDNA and nuclear DNA data because random
events cannot be eliminated as a possible major determinant of the mtDNA patterns of
variability.
Atlantic salmon has also been investigated for microsatellite variability. McConnell et
al. (1995a), in a preliminary survey of populations from both sides of the Atlantic,
revealed signi®cant allele frequency differences between European and North American
populations. This result was not unexpected, given that other molecular techniques such
as allozymes, mtDNA and minisatellite data revealed a similar pattern of variation
(Bermingham et al., 1991; Taggart et al., 1995b). Microsatellites have also been used to
distinguish among populations within Nova Scotia and Newfoundland (McConnell et
al., 1996a). This study revealed that populations within the Bay of Fundy, believed to
be a discrete offshore feeding area for some populations, could be discriminated from
other populations which predominantly migrate to feeding grounds off Greenland.
Allozyme studies also revealed population differentiation but did not discriminate
populations within the Bay of Fundy region from other Atlantic salmon populations
within Eastern Canada (Verspoor, 1986). Studies of Atlantic salmon populations on a
more limited geographical scale have also been carried out. Tessier et al. (1995)
examined two adjacent populations from Lac St Jean, Canada, and revealed signi®cant
mtDNA and microsatellite differentiation between populations, with the mtDNA marker
proving more diagnostic than any single microsatellite locus (Table 2).
Atlantic salmon from sites within the river Dee (Wales) have also been studied using
microsatellites (O'Connell et al., 1996d). Previous surveys on the same ®sh using
allozyme variability suggested that the populations were largely isolated from each
other (O'Connell, unpublished data). The microsatellite data showed an FST value (an
index of differentiation) twice that of the value generated from the allozyme survey.
This is not surprising given that the allozymes identi®ed only 9% of the total number
of alleles revealed at the microsatellite loci. Microsatellite data were also more accurate
than the allozyme data in determining how precisely individuals from each population
could be assigned to their population of origin (using the computer package SPAM 1.01
± Anon., 1995) although both data sets performed poorly. One potential problem with
using microsatellite loci in GSI (genetic stock identi®cation±determining individual
stock proportions in a mixed sample) is the very high number of alleles detected. This
phenomenon, also observed with single-locus minisatellite probes, can confound the
analysis although binning or grouping together of alleles that fall within a particular
size range can eliminate or reduce the problem (discussed in detail later). The very high
levels of variability also suggest that some microsatellites will not be useful tools for
comparing the levels of genetic diversity among populations. Comparisons of allelic
diversity will prove more informative than estimates of heterozygosity, although the
very high levels of allelic diversity in some cold-water ®shes (Brooker et al., 1994;
McConnell et al., 1995a) suggest that microsatellites may be insensitive to all but the
most severe population bottlenecks. Thus, for comparisons of diversity in populations
that have not undergone severe bottlenecks, alternative, less variable data sets such as
mtDNA will likely prove more informative.
Population surveys in marine species, as opposed to anadromous and freshwater
species, are still rare. A microsatellite survey of Paci®c herring (Clupea harengus,
Clupeidae) from the Gulf of Alaska and the Bering Sea revealed signi®cant genetic
structuring among populations at all loci (O'Connell et al., 1996b), which supported
previously collected allozyme data for the region (Grant and Utter, 1984). However, in
contrast to the allozyme, the microsatellite data also revealed structuring between
populations within Prince William Sound. The most intensively surveyed marine ®sh
species to date for microsatellite data is the Atlantic cod. Brooker et al. (1994) ®rst
described microsatellite loci for this species and demonstrated very high levels of
variability (Table 3). These markers were then used to investigate structuring in
populations from the North-west Atlantic (Bentzen et al., 1996; Ruzzante et al., 1996).
In contrast to surveys of mtDNA (Arnason et al., 1992) and allozymes (Pogson et al.,
1995; although see Cross and Payne, 1978, for exception), signi®cant structuring of
microsatellite variation among populations was demonstrated (Table 2) (Bentzen et al.,
1996). Ruzzante et al. (1996), in a separate study, have shown that variation within the
North-west Atlantic appears stable over a 2 year period, which suggested that the cod
Table 2. Averaged estimates of differentiation using FST (Weir and Cockerham, 1984) or analogous
differentiation coef®cients, reported for different marker systems in a number of ®sh species
Species Allozymes NRFLPs mtDNA Minisatellite Microsatellite
DNA DNA
Atlantic cod,
Gadus morhua 0.0141 0.0691 NS2 0.0303 0.0354
Atlantic salmon,
Salmo salar 0.0645,6,7 No data 0.1065 0.0368 0.0289,10
Rainbow trout,
Oncorhynchus mykiss 0.08011 No data Not reported 0.04912 0.01613
Brown trout,
Salmo trutta 0.19014 No data 0.30914 0.17414 No data
Paci®c herring,
Clupea harengus 0.01415 No data No data No data 0.02316
Brook charr,
Salvelinus fontinalis 0.23217 No data 0.12817 No data 0.51518
1
Pogson et al. (1995), based on six North Atlantic populations.
2
Arnason et al. (1992), based on 20 populations from Iceland and North-west Atlantic populations.
3
Galvin et al. (1995a), based on ®ve North-east Atlantic populations.
4
Bentzen et al. (1996), based on six North-west Atlantic populations.
5
O'Connell et al. (1995), based on seven tributary populations from three drainages.
6
Stahl (1987), based on 31 populations from throughout the North Atlantic.
7
Jordan et al. (1992), based on 22 tributary populations from 16 drainages.
8
Galvin et al. (1995a), based on nine tributary populations.
9
O'Connell et al. (1996b), based on ®ve tributary populations.
10
McConnell et al (1997a), data based on 15 populations.
11
Hershberger (1992), based on 38 populations from western North America.
12
Taylor (1995), based on eight north-eastern Paci®c populations.
13
O'Connell et al. (1996a), based on ®ve Lake Ontario populations.
14
Ferguson et al. (1995), based on 29 populations for allozymes, 25 populations for minisatellites and 14
populations for minisatellites.
15
Grant and Utter (1984), based on 21 North Paci®c populations.
16
O'Connell et al. (1996b), based on seven North-east Paci®c populations.
17
Jones et al. (1996), based on 34 Eastern Canadian populations.
18
Angers et al. (1995), based on ®ve populations from two drainages.

NS, not signi®cant.
complex within the North-west Atlantic is made up of reproductively discrete

units=stocks.
A review of these studies revealed no consensus on which test statistics best
discriminated between the populations sampled. In a comparison of inshore=offshore
cod populations, Ruzzante et al. (1996) observed a signi®cant partitioning of the
genetic variation with the RST coef®cient (index of differentiation, which describes the
variance in allele size within populations relative to the total variance in allele size)
(Slatkin, 1995). The analogous FST statistic, which is based on allele frequencies alone,
failed to discriminate between the populations. However, in a more extensive survey of
cod populations within the North-west Atlantic, Bentzen et al. (1996) observed RST
values to be insigni®cant in comparison with statistically signi®cant FST estimates based
on the same data. The RST coef®cient is based on a stepwise mutation model which
takes into account the size difference between alleles and has been proposed as a more
appropriate statistic for microsatellite data in some cases (Slatkin, 1995; discussed
later). The large range in allele sizes commonly observed in cold-water ®sh species, e.g.
Atlantic salmon (O'Reilly et al., 1996), may lead to more variable and, thus, less
statistically meaningful RST values when compared with more traditional indices such
as FST . Furthermore, population structure estimates biased for allele size are more likely
to be lower than those based solely on allele frequencies in populations that have
undergone recent stocking and=or other mixing events.
Another potential problem, which is not generally considered when deciding how to
analyse microsatellite data, is the occurrence of allele size shifts. Changes in mean
allele sizes have been observed in comparisons of closely related species and these
shifts are frequently the result of insertion=deletion events in the ¯anking sequence
(Estoup et al., 1995). Large shifts in allele size have also been observed within species,
e.g. among Atlantic salmon populations from both sides of the Atlantic (unpubl. data).
These shifts might simply re¯ect differences in repeat number. However, insertion=
deletion events within the ¯anking sequence could also explain the allele size shifts.
Alternatively, more subtle changes within the ¯anking sequence or within the array
itself could lead to a completely different slippage pattern during DNA replication,
resulting in changes in the allele distribution. These potential mechanisms for large
changes in mean allele size suggest that a careful consideration of the assumptions of
different population genetics coef®cients and sequencing of some alleles is required to
establish why shifts in allele distributions occur.
Parentage and kinship analysis

The highly variable nature of microsatellite loci (Table 3) makes these markers
particularly suited for the investigation of kinship relationships and paternity analysis.
Such analyses can be useful for the management of captive populations and in
understanding mating patterns in the wild. Herbinger et al. (1996b) investigated cod
populations from the North-west Atlantic at six microsatellite loci. The objectives of the
study were to determine if natural selection could be observed directly in a cohort of
larvae and to estimate a minimum effective population size for the spawning aggregation
that produced the larval cohort. Kinship relationships among individuals within the study
population were assessed using a likelihood ratio method based on the observed and
expected number of shared alleles in an idealized situation (details: Thompson, 1991;
Herbinger et al., 1996b). This analysis permitted the authors to conclude that there was
VNTR DNA and ®sheries research
Table 3. Average sample sizes, estimates of variability, number of loci used and Hardy±Weinberg (H±W) analyses in a range of ®sh microsatellite
studies
Species Average Average Expected No. of H±W deviations/ Reference
sample number of heterozygosity variable potential for
size/population alleles/locus loci null allele1
Atlantic salmon, Salmo salar 32 24 0.730 8 Het. de®cit2 , Ssa4 S.K. McConnell (per.
comm.)
33 4 0.351 5 No data Tessier et al. (1995)
59 37 0.890 5 Het. de®cit, Ssa197, O'Connell et al. (unpubl.
Ssa4 data)
Rainbow trout, Oncorhynchus 9 20 0.7303 1 Het. de®cit, Omy77 Nielsen et al. (1994)
mykiss
307 9 0.7503 2 No data Morris et al. (1996)
29 18 0.864 5 Het. de®cit O'Connell et al. (1996a)
Atlantic cod, Gadus morhua 60 20 0.864 5 None reported Ruzzante et al. (1996)
186 46 0.900 6 None reported Ruzzante et al. (1996)
54 41 0.898 6 Het. de®cit, Gmo141 Bentzen et al. (1996)
127 25 0.7573 6 None reported Brooker et al. (1994)
Paci®c herring, Clupea harengus 50 33 0.889 5 Het. de®cit, Cha123 O'Connell et al. (1996b)
Sea bass, Dicentrarchus labrax 24 8 0.860 7 No data Garcia de Leon et al.
(1995)
Chinook salmon, Oncorhynchus 28 5 0.244 12 Het. de®cit, One ì13 Scribner et al. (1996)
tschawytscha
Brook charr, Salvelinus fontinalis 20 11 0.710 4 No data Angers et al. (1995)
Bluegill sun®sh, Lepomis 27 5 0.4993 7 No data Colbourne et al. (1996)
macrochirus
1
Describes whether populations were observed to be out of H±W proportions and which locus was responsible. Often reported as possible null allele.
2
Heterozygote de®ciency.
3
Observed heterozygosity.
345
no evidence of family structure within the larval cohort. This was the ®rst time that
family structure in a natural assemblage of larvae from a marine animal in the wild had
been determined. This ®nding is important as it has been proposed that some marine
animals may have an effective population size considerably smaller than the actual
population size, with relatively few crosses succeeding through sweepstakes evolution
(Hedgecock, 1994). The role of natural selection, however, could not be assessed as
sibship was not detected in this study, although the inbreeding effective population size
of the cohort investigated (in sensu Hartl and Clark, 1989) was calculated to be 2800
individuals. This is the ®rst time a naturally mating ®sheries population has had a
minimum effective population size estimated without invoking assumptions about genetic
marker mutation rates. In the past, to estimate effective population size, researchers had
to make unrealistic assumptions regarding the data, e.g. no selection, no migration,
randomly mating populations, and=or required temporal data (discussion: Waples, 1989;
Bartley et al., 1992). Thus, the estimates were likely to be subject to greater errors than
the allele-sharing likely ratio method of Herbinger et al. (1996b). The importance of the
effective population size Ne to many aspects of population and conservation biology
suggests that this analytical method, used in conjunction with microsatellite data, will
prove invaluable to the conservation management of ®shes.
Colbourne et al. (1996) examined parentage in the bluegill sun®sh (Lepomis
macrochirus, Centrarchidae) from eggs deposited within a natural nest using just two
microsatellite loci and demonstrated the reproductive success of precocious males
(cuckolders). The study also demonstrated that more than one female contributed to the
brood deposited within the nest. In a similar study, Kellogg et al. (1995) investigated
paternity in broods from seven species of lekking cichlids. This study revealed multiple
paternity in the majority of cases and in one extreme case revealed that the progeny
from one brood was derived from at least six males. In the past, this level of analysis
was impossible because of the high level of allele sharing commonly observed with less
variable markers. These types of analyses will allow the ®tness of different life-history
patterns in ®shes, which display the greatest diversity in evolutionary strategies amongst
vertebrates, to be evaluated.
In captive management programmes, pedigree data can also be highly useful to farm
managers (Harris et al., 1991). DNA ®ngerprinting could be used to elucidate the
pedigree of mixed family groups reared in communal environments. Elucidating
pedigrees of mixed family groups reared in a common environment is generally not
possible without DNA markers because physical tags cannot be applied to newborn ®sh.
Progeny groups have to be reared in different tanks until physical tagging is possible.
Selection experiments have generally been restricted to laboratories with specialized
breeding facilities as opposed to commercial farm operations. A DNA pro®ling
approach would permit the design of increased selection regimes with a minimal loss of
genetic variability on farms without specialized facilities. An understanding of the
genetic relationships among chosen broodstock is important as it allows managers to
reduce inbreeding (Kincaid, 1976). Herbinger et al. (1996a) examined the possibility of
determining pedigrees in a mix of farmed populations of rainbow trout and assessed the
potential of selection programmes based on microsatellite data to hatchery management.
Progeny from 100 crosses were reared in a common environment for one year. The
largest and smallest individuals were screened using four microsatellite loci and over
91% could be traced back to one or two crosses out of the 100 potential crosses.
Similarly, Kamonrat (1996) identi®ed successfully the parentage of communally reared

progeny in hatchery populations of the Thai silver barb (Puntius gonionotus,
Cyprinidae). These studies demonstrate that marker-assisted selection programmes are
bene®cial to relatively small farm operations, avoiding the need for separate tanks or
tags. Moreover, the conclusions drawn from the selection regime are more likely to
show reduced environmental bias, in that potential environmental effects are reduced
because the progeny from all crosses have been reared in a single tank.
Genome mapping
Aquaculture is a vital contributor to the economy of many countries worldwide. However,
the vast majority of species and strains reared globally are relatively unimproved for
commercially important traits such as growth rate, disease resistance and age of sexual
maturation. Thus, the potential for genetic improvement in ®sh species, compared with
domestic livestock, is very high. Most traits of commercial importance are polygenic, i.e.
the trait is controlled by more than one locus. These loci are commonly referred to as
quantitative trait loci (QTL). The ®rst detection of a QTL was for bunt and rust
resistance in wheat (Sax, 1923). Since this initial description, QTL have been mapped in
a wide variety of species including plants (Bradshaw et al., 1995), insects (Coyne and
Charlesworth, 1986) and livestock (Andersson et al., 1994). Marker-assisted selection
(MAS) can be carried out with an understanding of the linkage relationships between
QTLs and markers. This type of selection is particularly useful for characters that are
expressed at=after the onset of maturity, because it eliminates the need to rear animals to
this late stage. MAS can signi®cantly increase the intensity of selection when
complemented with traditional approaches. However, in order to map QTLs, a detailed
linkage map is required, with variable markers distributed throughout the genome.
Linkage between a marker locus and a segregating QTL allele can only be reliably
detected at or below a map distance of 20 centimorgans (cM) ((Soller et al., 1976),
although the generation of higher-density maps does little to increase the power of QTL
detection (Haley and Knott, 1994)). Given the tetraploidization, and consequently, the
large size of many ®sh genomes (Allendorf and Thorgaard, 1984), a large number of
variable, widely distributed markers are required for many ®sh genome mapping efforts,
e.g. about 200±300 evenly spaced markers are required for salmonids. Poopuang and
Hallerman (1996), in an excellent review of mapping and QTL, also reported that the
number of QTLs detected is more responsive to the degree of marker polymorphism than
to the number of individuals screened. Both these observations suggest that
microsatellites will prove to be an ideal tool for genome mapping.
To date, ®sh linkage maps have generally been constructed from allozyme data
(Morizot and Siciliano, 1983; May and Johnson, 1993; Postlethwait et al., 1994).
However, the number of loci and level of variability are insuf®cient for the mapping of
QTL. Randomly ampli®ed polymorphic DNA (RAPD) analyses are cheap and
technically simple. This marker set has been successfully used to generate maps in
®sh species including the zebra®sh (Brachydanio rerio, Poeciliidae) (652 markers and
25 linkage groups identi®ed; Postlethwait et al., 1994) and blue tilapia (Oreochromis
aureus, Cichlidae) (94 markers and 10 linkage groups identi®ed; Naish, Leamon,
Beynon and Skibinski, unpublished data, cited in McConnell et al., 1997b). However,
the susceptibility of the RAPD PCR to artefacts can complicate interpretation of
inheritance (Ferguson, 1994; Danzmann, O'Connell and Fishback, unpublished data)
and make cross-laboratory comparisons dif®cult. Furthermore, variation in primer sites

can lead to RAPD loci that are speci®c to a single strain or cross (Mitchell-Olds, 1995),
which limits the map to particular strains and restricts the potential of such maps in
comparative mapping applications (Montgomery et al., 1995).
Microsatellites have been quickly adopted for mapping applications in ®sh including
tilapia, zebra®sh, Atlantic salmon, rainbow trout and carp. In tilapia, 60 microsatellite
loci have been reported from the University of New Hampshire (http://tilapia.unh.edu ±
Lee and Kocher, 1996). Segregating loci (73% of variable loci) have been mapped to
15 linkage groups. Additional microsatellite loci are also being generated by other
groups (Ambali, 1996) and this collaborative approach is the most ef®cient method for
long-term mapping projects. To the best of our knowledge, the ®rst recorded QTL
mapped in a ®sh species is in rainbow trout (Jackson, 1995). In a map generated from
allozyme, RAPD and microsatellite data, a signi®cant association between upper
temperature tolerance and alleles at two microsatellite loci was observed in more than
one family, and the application of these data should prove especially useful in
broodstock selection. The testing of loci on more than one family is important because
without these data, observations on possible linkage relations are limited. Furthermore,
in the salmonids, which are currently `diploidizing', a process which involves centric
and pericentric fusions and ®ssions, variation in chromosome number has been reported
(Allendorf and Thorgaard, 1984). Thus, linkage relationships among some genes may
differ among strains. However, with the large number of progeny available from a
single cross in ®sh species, it is potentially very tempting to screen progeny from single
families. Workers interested in literature and programs in this general area are directed
to the http://chuck.agsci.colostate.edu website.
P OT E N T I A L P RO B L E M S =I S S U E S
Mutation models
Two models of mutation have basically been proposed to describe variation at
microsatellite loci, the in®nite allele mutation model (IAM) and the stepwise mutation
model (SMM). The IAM predicts that mutation will lead only to new allelic states and
may involve any number of repeat units. In contrast, the SMM predicts that mutation
occurs through the gain or loss of a single repeat unit, e.g. (GT). This means that some
mutations will generate alleles already present in the population. The importance of
assessing which model provides a better ®t to microsatellite data is that the assessment
can provide more accurate estimates of population size and structuring events (Estoup et
al., 1995).
The early work on mutation models showed that the SMM more accurately predicted
the type of variation commonly observed at microsatellite loci. Weber and Wong (1993)
observed, in direct studies of human families, that most mutations led to new alleles
which differed from the `parental' allele by only one or two repeat units. Thus, Valdes
et al. (1993) suggested that the SMM, originally proposed to account for variation at
allozyme loci (Kimura and Weiss, 1964), provided the best ®t to the microsatellite data
available at the time. Shriver et al. (1993) investigated different types of microsatellite
arrays and observed that the variability at trinucleotide and tetranucleotide loci was
more similar to the SMM relative to dinucleotide and minisatellite loci. Garza et al.
(1995), in an investigation of the variance in allele size at microsatellite loci in chimps
and humans, observed no signi®cant difference in the size variance between the two
groups, which suggested that there was a constraint on allele size. The authors proposed
a modi®ed SMM in which mutation at larger alleles tended to reduce allele size and
vice versa, although this concept remains untested to date (Jarne and Lagoda, 1996).
The basic SMM model was also re®ned by DiRienzo et al. (1994), who described
mutation events leading to an increase or decrease in allele size of one repeat unit as
being dominant, but with occasional mutations of several repeat units. The SMM
approach is attractive in that it permits workers to assess genetic divergence on the
basis of differences in allele size as well as frequency, and a number of distance
estimates have been developed to exploit this `phylogenetic' information, e.g. äì2
(Goldstein et al., 1995); RST (Slatkin, 1995).
However, for many ®sh species the number of alleles and heterozygosity values are
considerably larger than those observed in mammals, which have provided most of the
empirical data on which the mutation models have been based. For example, in
commercially important species such as rainbow trout, Atlantic salmon, Atlantic cod
and Paci®c herring, the number of alleles can be in excess of 50 and many alleles
within the total size range are not represented (although the latter observation may
result from low sample sizes). The only way to determine the most accurate model of
microsatellite mutation is to investigate mutation directly in large numbers of progeny
groups. However, this is not practical in most instances, and a more feasible way to test
the mutation models is to test how accurately they predict population parameters such
as heterozygosity and allele number, although this method does assume random mating
and mutation=drift equilibrium (details: Jarne and Lagoda, 1996). Although the ®rst
assumption can be easily tested, e.g. by Hardy±Weinberg test, mutation=drift equili-
brium often may not be assumed for commercially important species which are subject
to stocking and exploitation pressures. O'Connell et al. (1996a) observed that the
predictions of the IAM gave a consistently better ®t than the predictions of the SMM to
microsatellite variability data using uninterrupted SSR loci from rainbow trout. Estoup
et al. (1995), in a similar analysis of microsatellite variability at smaller loci in bees
(Apis mellifera), also observed that the IAM provided a more consistent ®t with the
data using loci that had interruptions (non-repeat units) embedded within the array.
Thus, array length=type alone does not likely appear to in¯uence the pattern of
mutation at microsatellite loci. The mode of mutation at microsatellite loci will likely
be case=population and locus speci®c.
The type of mutation per se at microsatellite loci may, however, not be relevant to
many ®sheries studies because of the often high level of arti®cial mixing (due to
stocking events) and the relatively high susceptibility of many migratory species, e.g.
salmonids, to local catastrophic events. These events would often tend to arti®cially bias
the observed patterns of microsatellite variability to the predictions of the IAM. Given
the uncertainty regarding the relative role of mutation models at present, we would
recommend that workers using microsatellite variability in ®sheries (in the absence of
testing and especially in species exhibiting a large number of alleles) take a
conservative approach and use conventional F-statistics (Weir and Cockerham, 1984).
Potential for scoring error

An important issue that must be addressed before microsatellite data can be routinely
applied to population genetics and wildlife forensic questions is the dif®culty in
accurately typing alleles. The stuttering at dinucleotide microsatellite loci can often
occlude adjacent alleles. Although this is not a problem in mapping studies where the
number and type of possible alleles are known a priori, it is a cause of concern in
population genetics where scoring errors can lead to an arti®cial excess of homozygotes.
Accurate scoring is also essential for parentage determination and family reconstruction
from wild populations. The problem is exacerbated in ®sh that have large microsatellites,
as the level of stuttering is generally higher in microsatellites with larger repeat arrays.
One possible approach to avoid this potential problem is to select for tetranucleotide loci.
These loci are easier to score because of the greater distance between alleles and reduced
stutter (Fig. 5; O'Reilly and Wright, 1995). A second method for reducing the potential
scoring dif®culties is to use dinucleotide loci with a reduced product size (, 120 bp).
Although this will likely reduce the level of variability detected, it will increase the
accuracy of scoring, as these loci tend to stutter less and their smaller size makes them
physically easier to separate during electrophoresis. Furthermore, the level of variability
that is required for most population genetics applications is often considerably less than
the level commonly observed at microsatellite loci.
However, if these alternatives remain unfeasible (and even if they do not), it is vital
to include several sets of size standards on all gels. The M13 sequencing ladder has
become a common reference. It is also important to incorporate a number of control
samples into each gel to take into account minor variations within and between gels
and between people preparing the samples, e.g. loading several pooled samples
frequently across the gel considerably eases scoring.
Fig. 5. This gel has been used to separate the ampli®cation products from two tetranucleotide loci.
The picture is ¯anked on either side and in the middle by the M13 sequence ladder, which is used as
a size reference. Note the general lack of large numbers of stuttering (shadow) bands that can lead to
scoring errors.
Null alleles
There has been an increasing realization that null alleles may contribute signi®cantly to
the patterns of variation observed at microsatellite loci. For the purposes of this review, a
null allele is de®ned as any allele at a microsatellite locus that is only weakly ampli®ed
or not visible after ampli®cation and separation. Callen et al. (1993) ®rst reported on the
incidence and origin of null alleles at dinucleotide microsatellite loci in humans. There
has subsequently been an increase in the number of null alleles reported in the literature
(Allen et al., 1995; Paetkau and Strobeck, 1995; Pemberton et al., 1995). Point mutations
within the primer site are thought to be generally responsible for the null allele
observations (Fig. 6). However, large insertion=deletion events between the array and the
primer site, poor DNA preparation and=or mutations within the array, leading to large
changes in product size, may also be responsible.
The incidence of null alleles may be detected through an excess of homozygotes,
relative to those expected under the Hardy±Weinberg theorem. If the null occurs at a
Product size is 120 bp
Insert size is 30 bp
Base point mutation

within primer site
No product
(Null allele)
Fig. 6. Diagram shows the effect of insertions and deletions in the ¯anking regions. The oblong track
represents the microsatellite array. The thickened lines denote the primer sites in which the inward-
facing arrows represent the primers. The short double arrowhead represents a 30 bp insertion adjacent
to the array. The double diagonal lines represent a deletion of DNA near to the array. The downward-
pointing arrow in the bottom diagram represents a base point mutation within the primer site, which
adversely affects annealing of the primer and can lead to null alleles. Note that the diagram is not
drawn to scale.
high frequency, it may also be observed through an increase in the number of PCR
`failures'. If a null allele is present in a population at a frequency p, the number of
individuals that would have to be sampled on average to observe one null allele
homozygote equals E(N ) 1= p2 . Therefore, if p 10ÿ2 , E(N ) 10 000 individuals,
although the heterozygote de®cit (inbreeding coef®cient, f (Weir and Cockerham,
1984)) produced by the null allele would be 0.02. This method should only be
considered a rough approximation as it assumes that the population tested is in Hardy±
Weinberg equilibrium. Furthermore, it should be noted that a null allele might be
scorable but weak. Owing to the PCR process, a null allele in a heterozygote may not
be evident, as the product of the non-null allele will dominate the PCR process to the
detriment of the null allele. However, if an individual is a homozygote for the null
allele, the product may be seen, albeit weakly, because no other PCR product is present
to `compete' with the null product in the reaction. This type of `partial null allele'
phenomenon may turn out to be a common phenomenon in cold-water ®sh
microsatellites, which on the basis of their larger size may be susceptible to large
allele dropout. `Large allele dropout' simply refers to the increased likelihood of PCR
failures for larger products and is often related to poor template quality. The ability to
investigate microsatellite DNA using PCR has facilitated the use of old=preserved
tissues, e.g. otoliths, scales and preserved specimens, as a source of DNA and rapid
DNA isolation protocols. However, shearing and protein contaminants can retard and=or
prevent the PCR ampli®cation of larger alleles relative to smaller alleles. Bentzen et al.
(1996) suggested that a null allele may be responsible for the signi®cant de®cit of
heterozygotes at Gmo141 in Atlantic cod from a number of populations from the
western Atlantic. However, the signi®cant de®cit of heterozygotes was observed only in
those populations where DNA was extracted from muscle tissue as opposed to blood,
which permitted a technically simpler and faster DNA extraction protocol to be applied
(D. Cook, MGPL, pers. comm.). Thus, in studies investigating old samples or where the
quality of the DNA isolated is thought to be poor, only smaller and well-characterized
loci (, 150 bp) should be investigated to reduce the likelihood of PCR size bias.
Workers should also be especially cautious when using loci originally developed for
alternate species, because the likelihood of changes in the ¯anking sequence, another
probable cause of null alleles, is increased in these cases and sequence characterization
of alleles in the alternate species is recommended.
Sample size required

One of the most important decisions which has yet to be made by ®sheries geneticists
and managers is the minimum sample size required to assay microsatellite variability and
make meaningful interpretations of the data. It is vital that this consideration be
addressed promptly so as to avoid the poor sampling regimes that have commonly been
observed in ®sheries with the introduction of other new genetic markers in the past.
Given the very high numbers of alleles at microsatellite loci in the majority of cold-water
®shes investigated to date, e.g. Atlantic cod with . 50 alleles (Ruzzante et al., 1996),
some increase in the sample size is required from that used to investigate genetic
variability using less variable markers. It is evident that the majority of studies carried
out to date have not included adequate sample sizes. This is more apparent when we
consider the number of alleles relative to sample size (Table 3), which demonstrates that
the majority of allele frequencies reported most likely occurred at less than 5%.
Furthermore, ®sh population surveys have shown that the majority of individual
populations are described with less than 50% of the possible allelic states being observed
(assuming a unimodal allele distribution pattern) (Table 3). Thus, the majority of studies
carried out to date can make no assumptions regarding private alleles (alleles that occur
at only one location). Numerical re-sampling techniques, e.g. Monte Carlo simulations
(Roff and Bentzen, 1989) and the Markov Chain approach (Guo and Thompson, 1992),
help in the analyses of `low' numbers of samples. However, increases in sample sizes are
required to improve the robustness of the data reported and produce more accurate
estimates of population differentiation through the reduction of the con®dence limits
associated with these estimates. The actual sample number required will depend largely
on the species=type of locus investigated, e.g. studies in which large numbers of alleles
are reported will necessarily require larger sample sizes (discussion: Chakraborty, 1992).
One possible strategy, which can be used to alleviate the problem of relatively low
sample sizes (and consequently large con®dence limits on data), is the grouping or
binning of alleles within a certain size class (Bentzen et al., 1991). This approach leads
to the loss of information and the particular size range and numbers of bins selected is
highly subjective. However, it may be the only way to overcome the very high levels of
variability observed at some loci, e.g. . 50 alleles in Atlantic cod (Ruzzante et al.,
1996). Alleles observed at low frequencies only need to be grouped, and these are
typically those alleles occurring at either end of the size range. If the SMM is
considered, binning of alleles solely on the basis of size similarity might also be
considered appropriate. This statement is based on the model's predictions that similar-
sized alleles are separated by fewer mutation events when compared with alleles with
greater size differences that are considered more evolutionarily distant. The precedent
for allele binning in population studies has already been established by using other
marker sets, both deliberately and inadvertently (e.g. binning all alleles except the most
common, or scoring alleles of identical size, but different sequence, as a single allele).
However, grouped data would be less useful when trying to investigate family structures
within populations (Herbinger et al., 1996a). Thus, modelling data and=or simulations
with different types of empirical data are urgently required to address this issue.
It is also important to point out that it is unrealistic to expect sample sizes, for
highly variable loci, that will allow the majority of alleles at a locus to be represented
®ve or more times. This would involve very large sample sizes, in some instances more
than would be present in the population, and in these cases the only solution may be to
adopt numerical re-sampling techniques, e.g. bootstrapping, which do provide estimates
of how reliable the data are (through con®dence limits). A sample regime in which
each population is sampled at two points in time might prove useful in determining the
within-population variance. The within-population variance could then be used to
calibrate the between-population component of microsatellite variation. Although it is
impossible to provide a single optimum sample size for microsatellites, a minimum
sample size of 50 individuals per population should be considered for loci showing
between ®ve and ten alleles (and larger sample sizes would be more appropriate).
A problem related to that of sample size is the number of loci required to investigate
microsatellite variability. The number of loci required depends very much on the
biological questions being addressed and the variability of the loci being used. If the
purpose of a study is to investigate population differentiation, one or two loci may
suf®ce in demonstrating a signi®cant level of structuring among populations.
Nevertheless, the re-sampling techniques that calculate the variance and con®dence
limits associated with a particular data set are performed over loci. Thus, estimates
obtained from a small number of loci will be dubious. Goudet (1996) reports that at
least ®ve loci need to be used to obtain meaningful con®dence limits and estimates of
variance for F statistics (Weir and Cockerham, 1984).
A related problem associated with using too few loci is that the robustness of nodes,
generated through bootstrapping across loci when constructing genetic distance trees,
may also be arti®cially high. The potential for arti®cially high values arises because the
number of possible permutations may be considerably less than the number of re-
sampling events (often 1000), and may consequently lead to an arti®cially small
variance estimate. Thus, re-sampling a large number of times with a `small' number of
loci can lead to a false interpretation of the robustness of the nodes identi®ed. It is also
important to examine several loci to investigate variability within populations, because
factors such as null alleles, and scoring=technical problems that may be associated with
some loci, can lead to unduly biased estimates of Hardy±Weinberg (dis)equilibria.
Comparisons of multilocus and single-locus analyses are essential if false interpretations
of population data are to be rejected. Scribner et al. (1996) investigated patterns of
microsatellite variability in chinook salmon (Oncorhynchus tschawytscha, Salmonidae)
and excluded population admixture as the likely agent responsible for the signi®cant
de®cit of heterozygotes at one locus through comparisons of single-locus and multilocus
results (Table 3). Similarly, Bentzen et al. (1996) observed a signi®cant de®cit of
heterozygotes at only one locus (out of six examined in Atlantic cod) and suggested a
null allele as the most likely explanation (see null allele section). To the best of our
knowledge, all cases of Hardy±Weinberg deviations at microsatellite loci in ®sh studies
have been caused by a de®cit of heterozygotes. This ®nding suggests that null alleles
may be quite common (Jarne and Lagoda, 1996; Table 3), although this might also be
due to heterozygote versus homozygote scoring problems at stuttering loci. Therefore,
despite the high information content of most microsatellite loci, it is recommended that
more than ®ve loci be used to investigate population genetic questions.
Concluding remarks
A large number of molecular techniques that can identify genetic varation within species
are now available. This review has described the application of VNTR DNA and
microsatellites in particular to ®sh studies, although other techniques are available
(Greider, 1991; Klein et al., 1993; Carvalho and Hauser, 1994; Ferguson, 1994; Ward and
Grewe, 1994; Ferguson et al., 1995). The different molecular techniques now available
have particular applications for which they are most powerful. For instance, the rapid
development of a framework genetic map may be readily achieved with RAPD markers.
An initial survey of genetic variation for a species in which no DNA development work
has been carried out is most easily realized with an allozyme survey. Nevertheless, the
tool with the greatest overall potential to address the most common applications of
molecular markers to ®sh and ®sheries-related questions (excluding cross-species
phylogenetic applications) is microsatellites.
There are, however, a number of potentially serious problems associated with
generating and interpreting microsatellite data, including null alleles, poor resolution of
larger alleles, designing appropriate sampling regimes, and deciding on which statistics
to employ (FST versus RST ). Most of these problems can be offset, or at least reduced,
by carefully considering the biological questions which are to be addressed, using
tetranucleotide loci and keeping the product size to a minimum. Although smaller loci
may reveal lower numbers of alleles, they amplify more reliable and are considerably
easier to score. As the allele size range in smaller loci is reduced, more than one locus
can also be frequently ampli®ed in a single reaction, i.e. multiplexing (Fig. 5), which
considerably reduces the costs associated with investigating microsatellite variability.
Microsatellites are being rapidly characterized for many teleosts, although most of the
research emphasis to date has concentrated on the commercially important salmonids
and cichlids (Table 4). However, the ability to amplify presumably homologous loci in
related species (McConnell et al., 1995a; Morris et al., 1996; Rico et al., 1996) can
help reduce the development costs associated with related, but less commercially
important, species.
Microsatellites will probably have the greatest impact on genome mapping and
parentage assessment applications. At present, there are a number of large collaborative
programmes dedicated to mapping the genomes of commercially important species.
Microsatellite markers have provided the ®rst practical tool that can be used to generate
robust high-density linkage maps (in species other than humans). The pace of
development of these maps is prodigious and maps are now extended further in months
than would have been possible with years of work before the advent of microsatellite-
based analyses. Parentage assessment and determining kinship relationships have also
been successfully addressed using microsatellites. These types of studies allow more
intensive surveys of populations and mating patterns than were previously possible.
Finally, as tri- and tetranucleotide loci do not require a high level of technical
Table 4. A species list showing the number of ®sh groups for which microsatellites have been
published1
Species Source
Atlantic cod, Gadus morhua Brooker et al. (1994), Ruzzante et al. (1996)
Atlantic salmon, Salmo salar Slettan et al. (1993, 1995a,b), McConnell et al.
(1995b), O'Reilly et al. (1996)
Bluegill sun®sh, Lepomis macrochirus Colbourne et al. (1996)
Brown trout, Salmo trutta Estoup et al. (1993)
Brook charr, Salvelinus fontinalis Angers et al. (1995)
Paci®c salmonids, Oncorhynchus spp. Sakamoto et al. (1994a,b,c), Jackson (1995), Morris et
al. (1996), O'Connell et al. (1996a), Scribner et al.
(1996)
Paci®c herring, Clupea harengus pallasi O'Connell et al. (1996b)
Common carp, Cyprinus carpio K.-A. Naish, pers. comm.; MPGL ± unpubl. data
Sea bass, Dicentrarchus labrax Garcia de Leon et al. (1995)
Silver barb, Puntius gonionotus Kamonrat (1996)
Stickleback, Gasterosteus aculeatus Rico et al. (1993)
Tilapia, Oreochromis spp. Ambali (1996), Lee and Kocher (1996)
Whiting, Merluccius merluccius Rico et al. (1996)
Zebra®sh, Brachydanio rerio Goff et al. (1992)
1
Note: Information on unpublished sequence data may also be obtained from GenBank and related databases.
expertise once isolated, the transfer of this technology to less developed countries is
possible. It is to be hoped that this will allow those countries, which often have limited
®nancial resources, to address population and conservation genetics issues. This is
important, as these countries are frequently rich in biological diversity but have been
unable to investigate genetic variability owing to the high technical demands and
running costs associated with the employment of other types of DNA markers.
Acknowledgements
We are grateful to friends and colleagues at the Marine Gene Probe Laboratory for their
comments and criticisms on an earlier draft of the manuscript, including Pat O'Reilly,
Doug Cook, Lorraine Hamilton and Mary Dillon. Many of the approaches and
conclusions reported here were generated through group discussions, although any errors
are our own. We are also grateful to the editor, to Dr Paul Galvin and to two anonymous
reviewers, whose critical comments greatly improved the quality and focus of the
manuscript. We acknowledge the support of the Natural Sciences and Engineering
Research Council to JMW during the writing of this review.
References
Allen, P.J., Amos, W., Pomeroy, P.P. and Twiss, S.D. (1995) Microsatellite variation in grey seals
( Halichoerus grypus) show evidence of genetic differentiation between two British breeding
colonies. Mol. Ecol. 4, 653±662.
Allendorf, F.W. and Thorgaard, G.H. (1984) Tetraploidy and the evolution of salmonid ®shes. In
Turner, B.J., ed. Evolutionary Genetics of Fishes. New York: Plenum Press, pp. 1±53.
Ambali, A.J.D. (1996) The relationship between domestication and genetic diversity of Oreochromis
species in Malawi: Oreochromis shiranus shiranus (Boulenger) and Oreochromis shiranus chilwae
(Trewavas). PhD thesis, Dalhousie Univ., Canada. 204 pp.
Amos, B., Barrett, J. and Dover, G.A. (1991) Breeding behaviour of pilot whales revealed by DNA
®ngerprinting. Heredity 67, 49±55.
Andersson, L., Haley, C.S., Ellegren, H., Knott, S.A., Johansson, M., Andersson, K., Andersson-
Eklund, L., Edfors-Lilja, I., Fredholm, M., Hansson, I., Hakansson, J. and Lundstrom, K. (1994)
Genetic mapping of quantitative trait loci for growth and fatness in pigs. Genetics 263,
1771±1774.
Angers, B., Bernatchez, L., Angers, A. and Desgroseillers, L. (1995) Speci®c microsatellite loci for
brook charr reveal strong population subdivision on a microgeographic scale. J. Fish Biol. 47,
177±185.
Anon. (1995) SPAM 1.01 ± Statistics Program for Analyzing Mixtures. Alaska Dept of Fish and
Game, Division of Commercial Fisheries Management and Development, Genetics Lab, 333
Raspberry Rd, Anchorage, AK 99518.
Armour, J.A.L., Neumann, R., Gobert, S. and Jeffreys, A.J. (1995) Isolation of human simple sequence
repeat loci by hybridization selection. Hum. Mol. Genet. 3, 599±605.
Arnason, E., Palsson, S. and Arason, S. (1992) Gene ¯ow and lack of population differentiation in
Atlantic cod, Gadus morhua L., from Iceland and a comparison of cod from Norway and
Newfoundland. J. Fish Biol. 40, 751±770.
Avise, J.C. and Smith, M.H. (1974) Biochemical genetics of sun®sh. I. Geographic variation and sub-
speci®c intergradation in the bluegill, Lepomis macrochirus. Evolution 28, 42±56.
Bartley, D., Bagley, M., Gall, G. and Bentley, B. (1992) Use of linkage disequilibrium data to estimate
effective population size of hatchery and natural ®sh populations. Cons. Biol. 6, 365±375.
Bentzen, P., Harris, A.S. and Wright, J.M. (1991) Cloning of hypervariable minisatellite and simple
sequence microsatellite repeats for DNA ®ngerprinting of important aquacultural species of
salmonids and tilapia. In Burke, T., Dolf, G., Jeffreys, A.J. and Wolf, R., eds. DNA
Fingerprinting Approaches and Application. Basel, Switzerland: Birkhauser Verlag, pp. 243±262.
Bentzen, P., Taggart, C.T., Ruzzante, D.E. and Cook, D. (1996) Microsatellite polymorphism and the
population structure of Atlantic cod (Gadus morhua) in the Northwest Atlantic. Can. J. Fish.
Aquat. Sci. 53, 2706±2721.
Bermingham, E., Friedland, K., Forbes, S. and Pla, C. (1991) Discrimination between Atlantic salmon
(Salmo salar) of North American and European origin using restriction analyses of mitochondrial
DNA. Can. J. Fish. Aquat. Sci. 48, 884±893.
Bernatchez, L., Guyomard, R. and Bonhomme, F. (1992) DNA sequence variation of mitochondrial
control region among geographically and morphologically remote European brown trout (Salmo
trutta) populations. Mol. Ecol. 1, 161±174.
Bradshaw, H.D.J., Wilbert, S.M., Otto, K.G. and Schemske, D.W. (1995) Genetic mapping of ¯oral
traits associated with reproductive isolation in monkey¯owers (Mimulus). Nature 376, 762±765.
Brooker, A.L., Cook, D. Bentzen, P., Wright, J.M. and Doyle, R.W. (1994) Organization of
microsatellites differs between mammals and cold-water teleost ®shes. Can. J. Fish. Aquat. Sci.
51, 1959±1966.
Burke, T. and Bruford, M.W. (1987) DNA ®ngerprinting in birds. Nature 327, 149±152.
Callen, D.F., Thompson, A.D., Shen, Y., Phillips, H.A., Richards, R.I., Mulley, J.C. and Sutherland,
G.R. (1993) Incidence and origin of ``null' alleles in the (AC)n microsatellite markers. Am. J.
Hum. Genet. 52, 922±927.
Carter, R.E., Mair, G.C., Skibinski, D.O.F., Parkin, D.T. and Beardmore, J.A. (1991) The application of
DNA ®ngerprinting in the analysis of gynogenesis of tilapia. Aquaculture 95, 41±52.
Carvalho, G.R. and Hauser, L. (1994) Molecular genetics and the stock concept in ®sheries. Rev. Fish
Biol. Fish. 4, 326±350.
Chakraborty, R. (1992) Sample size requirements for addressing the population genetics issues of
forensic use of DNA typing. Hum. Biol. 64, 141±159.
Colbourne, J.K., Neff, B.D., Wright, J.M. and Gross, M.R. (1996) DNA ®ngerprinting of bluegill
sun®sh (Lepomis macrochirus) using (GT)n microsatellites and its potential for assessment of
mating success. Can. J. Fish. Aquat. Sci. 53, 342±349.
Coyne, J.A. and Charlesworth, B. (1986) Location of an X-linked factor causing sterility in male
hybrids of Drosophila simulans and D. mauritiana. Heredity 57, 243±246.
Cross, T.F. and Payne, R.H. (1978) Geographic variation in Atlantic cod, Gadus morhua, off eastern
North America: a biochemical systematics approach. J. Fish. Res. Bd Can. 35, 117±123.
DiRienzo, A., Peterson, A.A., Garza, J.C., Valdes, A.M., Slatkin, M. and Freimer, N.B. (1994)
Mutational processes of simple sequence repeat loci in human populations. Proc. Natl Acad. Sci.
USA 91, 3166±3170.
Dueck, L.A. (1994) Population divergence of introduced rainbow trout (Oncorhynchus mykiss) in the
Lake Ontario watershed, based on the mitochondrial genome. MSc thesis, Univ. of Guelph,
Canada. 173 pp.
Edwards, K.J., Barker, J.H.A., Daly, A., Jones, C. and Karp, A. (1995) Microsatellite libraries enriched
for several microsatellites sequences in plants. Biotechniques 20, 758±760.
Estoup, A., Presa, P., Krieg, F., Vaiman, D. and Guyomard, R. (1993) (CT)n and (GT)n microsatellites:
a new class of genetic markers for Salmo trutta L. (brown trout). Heredity 71, 488±496.
Estoup, A., Tailliez, C., Cornuet, J.-M. and Solignac, M. (1995) Size homoplasy and mutational
processes of interrupted microsatellites in two bee species, Apis mellifera and Bombus terrestris
(Apidae). Mol. Biol. Evol. 12, 1074±1084.
Ferguson, A., Taggart, J.B., Prodohl, P.A., McMeel, O., Thompson, C., Stone, C., McGinnity, P. and
Hynes, R.A. (1995) The application of molecular markers to the study and conservation of ®sh
populations, with special reference to Salmo. J. Fish Biol. 47 (Suppl. A), 103±126.
Ferguson, M. (1994) The role of molecular genetic markers in the management of cultured ®shes. Rev.
Fish Biol. Fish. 4, 351±373.
Galvin, P., McGregor, D. and Cross, T. (1995a) A single locus minisatellite as a hypervariable genetic
marker in gadoid species. Aquaculture 137, 31±40.
Galvin, P., McKinnell, S., Taggart, J.B., Ferguson, A., O'Farrell, M. and Cross, T.F. (1995b) Genetic
stock identi®cation of Atlantic salmon using single locus minisatellite DNA pro®les. J. Fish Biol.
47 (Suppl. A), 186±199.
Galvin, P., Sadusky, T., McGregor, D. and Cross, T. (1995c) Population genetics of Atlantic cod using
ampli®ed single locus minisatellite VNTR analysis. J. Fish Biol. 47 (Suppl. A), 200±208.
Garcia de Leon, F.J., Dallas, J.F., Chatain, B., Canonne, M., Versini, J.J. and Bonhomme, F. (1995)
Development and use of microsatellite markers in sea bass, Dicentrarchus labrax (Linnaeus,
1758) (Perciformes: Serranidae). Mol. Biol. Biotech. 4, 62±68.
Garza, J.C., Slatkin, M. and Freimer, N.B. (1995) Microsatellite allele frequencies in humans and
chimpanzees, with implications for constraints on allele size. Mol. Biol. Evol. 12, 594±603.
Goff, D.J., Galvin, K., Katz, H., Wester®eld, M., Lander, E.S. and Tabin, C.J. (1992) Identi®cation
of polymorphic simple sequence repeats in the genome of the zebra®sh. Genomics 14,
200±202.
Goldstein, D.B., Linares, A.R., Cavilli-Sforza, L.L. and Feldman, M.W. (1995) An evaluation of
genetic distances for use with microsatellite loci. Genetics 139, 463±471.
Goudet, J. (1996) F-STAT (1.2), a program for IBM PC compatible to calculate Weir and Cockerham's
(1984) estimators of F-statistics. Institute de Zoologie et d'Ecologie Animale, Batiment de
Biologie, Universite de Lausanne, CH±1015 Dorigny, Switzerland.
Grant, W.S. and Utter, F.M. (1984) Biochemical genetics of Paci®c herring (Clupea pallasi). Can. J.
Fish. Aquat. Sci. 41, 856±864.
Greider, C.W. (1991) Telomeres. Curr. Opinion Cell Biol. 3, 444±451.
Guo, S.W. and Thompson, E.A. (1992) Performing the exact test of Hardy±Weinberg proportion for
multiple alleles. Biometrics 48, 361±372.
Haley, C.S. and Knott, S.A. (1994) Interval mapping. In Proc. 5th World Congress on Genetics
Applied to Livestock Production, Vol. 21. Guelph, Ontario: Dept of Animal and Poultry Sciences,
Univ. of Guelph, pp. 25±32.
Harris, A.H. and Wright, J.M. (1995) Nucleotide sequence and genomic organization of cichlid ®sh
minisatellites. Genome 38, 177±184.
Harris, A.H., Bieger, S., Doyle, R.W. and Wright, J.M. (1991) DNA ®ngerprinting of tilapia,
Oreochromis niloticus, and its application to aquaculture genetics. Aquaculture 92, 157±163.
Harris, H. and Hopkinson, D. (1976) Handbook of Enzyme Electrophoresis in Human Genetics. New
York: American Elsevier.
Hartl, D.L. and Clark, A.G. (1989) Principles of Population Genetics. Sunderland, MA: Sinauer
Assoc.
Hedgecock, D. (1994) Does variance in reproductive success limit effective population sizes of marine
organisms? In Beaumont, A.R., ed. Genetics and Evolution of Aquatic Organisms. London:
Chapman and Hall, pp. 122±134.
Herbinger, C.M., Doyle, R.W., Pitman, E.R., Paquet, D., Mesa, K.A., Morris, D.B., Wright, J.M. and
Cook, D. (1996a) DNA ®ngerprint based analysis of paternal and maternal effects on offspring
growth and survival in communally reared rainbow trout. Aquaculture 137, 245±256.
Herbinger, C.M., Doyle, R.W., Taggart, C.T., Lochmann, S.E., Brooker, A.L., Wright, J.M. and Cook,
D. (1996b) Family relationships and effective population size in a natural cohort of cod larvae.
Can. J. Fish. Aquat. Sci. 54 (Suppl. 1), 11±18. [1997].
Hershberger, W.K. (1992) Genetic variability in rainbow trout populations. Aquaculture 100, 51±71.
Hunter, R.L. and Markert, C.L. (1957) Histochemical demonstration of enzymes separated by zone
electrophoresis in starch gels. Science 125, 1294±1295.
Jackson, T.R. (1995) Linkage analysis of molecular markers and a search for quantitative trait loci for
upper temperature tolerance in rainbow trout (Oncorhynchus mykiss). PhD thesis, Univ. of
Guelph, Canada. 81 pp.
Jarne, P. and Lagoda, P.J.L. (1996) Microsatellites, from molecules to populations and back. Trends
Ecol. Evol. 11, 424±429.
Jeffreys, A.J., Wilson, V. and Thein, S.L. (1985a) Hypervariable ``minisatellite'' regions in human
DNA. Nature 314, 67±73.
Jeffreys, A.J., Wilson, V. and Thein, S.L. (1985b) Individual-speci®c `®ngerprints' of human DNA.
Nature 316, 76±9.
Jeffreys, A.J., Wilson, V., Kelly, R., Taylor, B.A. and Bul®eld, G. (1987) Mouse DNA ``®ngerprint''
analysis of chromosome localization and germ-line stability of hypervariable loci in recombinanat
inbred strains. Nucleic Acids Res. 15, 2823±2837.
Jeffreys, A.J., Royle, N.J., Patel, I., Armour, J.A.L., MacCleod, A., Collick, A., Gray, I.C., Neumann,
R., Gibbs, M., Crossier, M., Hill, M., Signer, E. and Monckton, D. (1991) Principles and recent
advances in human DNA ®ngerprinting. In Burke, T., Dolf, G., Jeffreys, A.J. and Wolf, R., eds.
DNA Fingerprinting Approaches and Application. Basel, Switzerland: Birkhauser Verlag,
pp. 1±19.
Jones, M.W., Clay, D. and Danzmann, R.G. (1996) Conservation genetics of brook trout (Salvelinus
fontinalis): population structuring in Fundy National Park, New Brunswick, and eastern Canada.
Can. J. Fish. Aquat. Sci. 53, 2776±2791.
Jordan, W.C., Youngson, A.F., Hay, D.W. and Ferguson, A. (1992) Genetic protein variation in natural
populations of Atlantic salmon (Salmo salar): temporal and spatial variation. Can. J. Fish. Aquat.
Sci. 49, 1863±1872.
Kamonrat, W. (1996) Spatial genetic structure of Thai Silver barb, Puntius gonionotus, (Bleeker)
populations in Thailand. PhD thesis, Dalhousie Univ., Canada. 193 pp.
Karagyozov, L., Kalcheva, I.D. and Chapman, V.M. (1993) Construction of random small insert
genomic libraries highly enriched for simple sequence repeats. Nucleic Acids Res. 21,
3911±3912.
Kellogg, K.A., Markert, J.A., Stauffer, J.R. and Kocher, T.D. (1995) Microsatellite variation
demonstrates multiple paternity in lekking cichlid ®sh from Lake Malawi, Africa. Proc. R.
Soc. Lond. 260B, 79±84.
Kijas, J.M.H., Fowler, J.C.S., Garbett, C.A. and Thomas, M.R. (1994) Enrichment of microsatellites
from the citrus genome using biotinylated oligonucleotide sequences bound to streptavidin-coated
magnetic particles. Biotechniques 16, 656±662.
Kimura, M. and Weiss, G.H. (1964) The stepping stone model of population structure and the
decrease of genetic correlation with distance. Genetics 49, 561±576.
Kincaid, H.L. (1976) Inbreeding in rainbow trout (Salmo gairdneri). J. Fish. Res. Bd Can. 33,
2420±2426.
Klein, D., Ono, H., O'hUigin, C., Vincek, V., Goldschmidt, T. and Klein, J. (1993) Extensive MHS
variability in cichlid ®shes. Nature 364, 330±334.
Lansman, R.A., Shade, R.O., Shapiro, J.F. and Avise, J.C. (1981) The use of restriction endonucleases
to measure mitochondrial DNA sequence relatedness in natural populations III. Techniques and
potential applications. J. Mol. Evol. 17, 214±226.
Lee, W.-J. and Kocher, T.D. (1996) Microsatellite DNA markers for genetic mapping in Oreochromis
niloticus. J. Fish Biol. 49, 169±171.
McConnell, S.K., Hamilton, L., Morris, D., Cook, D., Pacquet, D., Bentzen, P. and Wright, J.M.
(1995a) Isolation of salmonid microsatellite loci and their application to the population genetics
of Canadian east coast stocks of Atlantic salmon. Aquaculture 137, 19±30.
McConnell, S.K., O'Reilly, P., Hamilton, L., Wright, J.M. and Bentzen, P. (1995b) Polymorphic
microsatellite loci from Atlantic salmon (Salmo salar): genetic differentiation of North American
and European populations. Can. J. Fish. Aquat. Sci. 52, 1863±1872.
McConnell, S.K., Ruzzante, D.E., O'Reilly, P.T., Hamilton, L. and Wright, J.M. (1997a) Microsatellite
loci reveal highly signi®cant genetic differentiation between Atlantic salmon (Salmo salar L.)
populations from the east coast of Canada. Mol. Ecol. (in press).
McConnell, S.K., Franck, J.P.C. and Wright, J.M. (1997b) Molecular genetic markers and their
applications to the Tilapias. In Mair, G., ed. Reviews in Applied Genetics of Tilapia. Manila:
ICLARM (in press).
McGregor, D., Galvin, P., Sadusky, T. and Cross, T. (1996) PCR ampli®cation of a polymorphic
minisatellite VNTR locus in whiting (Merlangius merlangius L.). Anim. Genet. 27, 49±51.
May, B. and Johnson, K. (1993) Composite linkage map of salmonid ®shes (Salvelinus, Salmo,
Oncorhynchus). In O'Brien, S., ed. Genetic Maps: Locus Maps of Complex Genomes. Cold
Spring Harbor, NY: Cold Spring Harbor Laboratory Press. pp. 151±159.
Mitchell-Olds, T. (1995) The molecular basis of quantitative genetic variation in natural populations.
Trends Ecol. Evol. 10, 324±328.
Moller, D. (1970) Transferrin polymorphism in Atlantic salmon (Salmo salar). J. Fish Res. Bd Can.
27, 1617±1625.
Montgomery, G.W., Penty, J.M., Henry, H.M. Sise, J.A., Lord, E.A., Dodds, K.G. and Hill, D.F. (1995)
Sheep linkage mapping: RFLP markers for the comparative mapping studies. Anim. Genet. 26,
249±259.
Morizot, D.C. and Siciliano, M.J. (1983) Comparative gene mapping in ®shes. In Rattazzi, M.C.,
Scandalios, J.G. and Whitt, G.S., eds. Isozymes: Current Topics in Biological and Medical
Research. New York: Alan R. Liss Inc., pp. 261±285.
Morris, D.B., Richard, K.R. and Wright, J.M. (1996) Microsatellites from rainbow trout
(Oncorhynchus mykiss) and their use for genetic studies of salmonids. Can. J. Fish. Aquat.
Sci. 53, 120±126.
Nielsen, J.L., Gan, C.A., Wright, J.M., Morris, D.B. and Thomas, W.K. (1994) Biogeographic
distributions of mitichondrial and nuclear markers for southern steelhead. Mol. Mar. Biol.
Biotech. 3, 281±293.
O'Connell, M., Skibinski, D.O.F. and Beardmore, J.A. (1995) Mitochondrial DNA and allozyme
variation in Atlantic salmon (Salmo salar) populations in Wales. Can. J. Fish. Aquat. Sci. 52,
171±178.
O'Connell, M., Danzmann, R.G., Cornuet, J.-M., Wright, J.M. and Ferguson, M.M. (1996a)
Differentiation of rainbow trout populations in Lake Ontario and the evaluation of the stepwise
mutation and in®nite allele mutation models using microsatellite variability. Can. J. Fish. Aquat.
Sci. 54 (in press).
O'Connell, M., Dillon, M.C. and Wright, J.M. (1996b) Development of microsatellite markers for
genetic discrimination of Prince William Sound herring populations. Report to the Alaska Dept of
Fish and Game. 31 pp.
O'Connell, M., Skibinski, D.O.F. and Beardmore, J.A. (1996c) Allozyme and mtDNA divergence
between Atlantic salmon populations in North Wales. J. Fish Biol. 48, 1023±1026.
O'Reilly, P. and Wright, J.M. (1995) The evolving technology of DNA ®ngerprinting and its
application to ®sheries and aquaculture. J. Fish Biol. 47 (Suppl. A), 29±55.
O'Reilly, P.T., Hamilton, L.C., McConnell, S.K. and Wright, J.M. (1996) Rapid analysis of genetic
variation in Atlantic salmon (Salmo salar) by PCR multiplexing of dinucleotide and
tetranucleotide microsatellites. Can. J. Fish. Aquat. Sci. 53, 2292±2298.
Ostrander, E.A., Jong, P.M., Rine, J. and Duyck, G. (1992) Construction of small-insert genomic DNA
libraries highly enriched for microsatellite repeat sequences. Proc. Natl Acad. Sci. USA 89,
3419±3423.
Paetkau, D. and Strobeck, C. (1995) The molecular basis and evolutionary history of a null allele in
bears. Mol. Ecol. 4, 519±520.
Pemberton, J.M., Slate, J., Bancroft, D.R. and Barrett, J.A. (1995) Non-amplifying alleles at
microsatellite loci: a caution for parentage and ®ngerprinting studies. Mol. Ecol. 4, 519±520.
Pogson, G.H., Mesa, K.A. and Boutillier, R.G. (1995) Genetic population structure and gene ¯ow in
the Atlantic cod Gadus morhua: a comparison of allozyme and nuclear RFLP loci. Genetics 139,
375±385.
Poopuang, S. and Hallerman, E.M. (1996) Towards detection of quantitative trait loci and genetic
marker assisted selection in ®sh. Rev. Fish. Sci. (in press).
Postlethwait, J.H., Johnson, S.L., Midson, C.N., Talbot, W.S., Gates, M., Ballinger, E.W., Africa, D.,
Andrews, R., Carl, T., Eisen, J.S., Horne, S., Kimmel, C.B., Hutchinson, M., Johnson, M. and
Rodriguez, M.A. (1994) A genetic linkage map for the zebra®sh. Science 264, 699±703.
Prodohl, P.A., Taggart, J.B. and Ferguson, A. (1994a) Cloning of highly variable minisatellite DNA
single locus probes for brown trout (Salmo trutta L.) from a phagemid library. In Beaumont,
A.R., ed. Genetics and Evolution of Aquatic Organisms. London: Chapman and Hall,
pp. 263±270.
Prodohl, P.A., Taggart, J.B. and Ferguson, A. (1994b) Single locus inheritance and joint segregation
analysis of minisatellite (VNTR) DNA loci in brown trout (Salmo trutta L.). Heredity 73,
556±566.
Reisenbichler, R.R. and Phelps, S.R. (1989) Genetic variation in steelhead trout (Salmo gairdneri)
from the North coast of Washington. Can. J. Fish. Aquat. Sci. 46, 66±73.
Rico, C., Zadworny, D., Kuhnlein, U. and FitzGerald, J.G. (1993) Characterization of microsatellite
loci in the three-spine stickleback, Gasterosteus aculeatus. Mol. Ecol. 2, 271±272.
Rico, C., Rico, I. and Hewitt, G. (1996) 470 million years of conservation of microsatellite loci among
®sh species. Proc. R. Soc. Lond. 263, 549±557.
Roff, D.A. and Bentzen, P. (1989) The statistical analyses of mitochondrial DNA polymorphisms: ÷ 2
and the problem of small sample sizes. Mol. Biol. Evol. 6, 539±546.
Royle, N.J., Clarkson, R.E., Wong, Z. and Jeffreys, A.J. (1988) Clustering of hypervariable
minisatellites in the proterminal regions of human autosomes. Genomics 3, 352±360.
Ruzzante, D.E., Taggart, C.T., Cook, D. and Goddard, S. (1996) Genetic differentiation between
inshore and offshore Atlantic cod (Gadus morhua L.) off Newfoundland: microsatellite DNA
variation and anti-freeze level. Can. J. Fish. Aquat. Sci. 53, 634±645.
Sakamoto, T., Okamoto, N., Ikeda, Y., Nakamura, Y. and Sato, T. (1994a) Dinucleotide repeat
polymorphism in DNA of rainbow trout and its application in ®shery science. J. Fish Biol. 44,
1093±1096.
Sakamoto, T., Okamoto, N. and Ikeda, Y. (1994b) Dinucleotide repeat polymorphism of rainbow trout,
FGT2. J. Anim. Sci. 72, 2765.
Sakamoto, T., Okamoto, N. and Ikeda, Y. (1994c) Dinucleotide repeat polymorphism of rainbow trout,
FGT3. J. Anim. Sci. 72, 2765.
Sax, K. (1923) The relation between chromosome number, morphological characters and rust
resistance in segregates of partially sterile wheat hybrids. Genetics 8, 301±321.
Scribner, K.T., Gust, J.R. and Fields, R.L. (1996) Isolation and characterization of novel salmon
microsatellite loci: cross-species ampli®cation and population genetics applications. Can. J. Fish.
Aquat. Sci. 53, 833±841.
Shriver, M.D., Jin, L., Chakraborty, R. and Boerwinkle, E. (1993) VNTR allele frequency distribution
under the stepwise mutation model. Genetics 134, 983±993.
Sick, K. (1961) Haemoglobin polymorphism in ®shes. Nature, Lond. 192, 894±896.
Skibinski, D.O.F. (1994) DNA technology and genetics of aquatic invertebrates. In Beaumont, A.R.,
ed. Genetics and Evolution of Aquatic Organisms. London: Chapman and Hall, pp. 177±199.
Slatkin, M. (1995) A measure of population subdivision based on microsatellite allele frequencies.
Genetics 139, 457±462.
Slettan, A., Olsaker, I. and Lie, O. (1993) Isolation and characterization of (GT)n repetitive sequences
from Atlantic salmon, Salmo salar L. Anim. Genet. 24, 195±197.
Slettan, A., Olsaker, I. and Lie, O. (1995a) Atlantic salmon, Salmo salar, microsatellites at the
Ssosl25, Ssosl85, Ssosl311, Ssosl417 loci. Anim. Genet. 26, 281±282.
Slettan, A., Olsaker, I. and Lie, O. (1995b) A polymorphic dinucleotide repeat microsatellite in
Atlantic salmon, Salmo salar (Ssosl436). Anim. Genet. 26, 368.

Soller, M., Brody, T. and Genizi, A. (1976) On the power of experimental designs for the detection of
linkage between marker loci and quantitative trait loci in crosses between inbred lines. Theor.
Appl. Genetic. 47, 35±39.
Stahl, G. (1987) Genetic population structure of Atlantic salmon. In Ryman, N. and Utter, F., eds.
Population Genetics and Fishery Management. Seattle, WA: University of Washington Press,
pp. 121±140.
Taggart, J.B. and Ferguson, A. (1990) Hypervariable minisatellite DNA single locus probes for
Atlantic salmon, Salmo salar L. J. Fish. Biol. 37, 991±993.
Taggart, J.B., Prodohl, P.A. and Ferguson, A. (1995a) Genetic markers for Atlantic salmon (Salmo
salar L.); single locus inheritance and joint segregation analysis of minisatellite (VNTR) DNA
loci. Anim. Genet. 26, 13±20.
Taggart, J.B., Verspoor, E., Galvin, P., Moran, P. and Ferguson, A. (1995b) A minisatellite DNA
marker for discriminating between European and North American Atlantic salmon (Salmo salar
L.). Can. J. Fish. Aquat. Sci. 52, 2305±2311.
Tautz, D. (1989) Hypervariability of simple sequences as a general source for polymorphic DNA
markers. Nucleic Acids Res. 17, 6463±6471.
Tautz, D. and Renz, M. (1984) Simple sequences are ubiquitous repetitive components of eukaryotic
genomes. Nucleic Acids Res. 12, 4127±4138.
Taylor, E.B. (1995) Genetic variation at minisatellite DNA loci among North Paci®c populations of
steelhead and rainbow trout (Oncorhynchus mykiss). Heredity 86, 354±363.
Tessier, N., Bernatchez, L., Presa, P. and Angers, B. (1995) Gene diversity analysis of mitochondrial
DNA, microsatellites and allozymes in landlocked Atlantic salmon. J. Fish Biol. 47 (Suppl. A),
156±163.
Thompson, E.A. (1991) Estimation of relationships from genetic data. In Rao, C.R. and Chakraborty,
R., eds. Handbook of Statistics, Vol. VIII. Amsterdam, Holland: Elsevier, pp. 255±269.
Utter, F.M. (1991) Biochemical genetics and ®shery management: an historical perspective. J. Fish
Biol. 39 (Suppl. A), 1±20.
Valdes, A.M., Slatkin, M. and Freimer, N.B. (1993) Allele frequencies at microsatellite loci: the
stepwise mutation model revisited. Genetics 133, 737±749.
Verspoor, E. (1986) Spatial correlations of transferrin allele frequencies in Atlantic salmon, Salmo
salar, populations from North America. Can. J. Fish. Aquat. Sci. 43, 1074±1078.
Waldbieser, G.C. (1995) PCR-based identi®cation of AT-rich tri- and tetranucleotide repeat loci in an
enriched plasmid library. Biotechniques 19, 742±744.
Waples, R.S. (1989) A generalized method for estimating population size from temporal changes in
allele frequency. Genetics 121, 379±391.
Ward, R.D. and Grewe, P. (1994) Appraisal of molecular genetic techniques in ®sheries. Rev. Fish
Biol. Fish. 4, 300±325.
Weber, J.L. and Wong, C. (1993) Mutation of human tandem repeats. Hum. Mol. Genet. 2,
1123±1128.
Weir, B.S. and Cockerham, C.C. (1984) Estimating F-statistics for the analysis of populations
structure. Evolution 38, 1358±1370.
Welsh, J. and McClelland, M. (1990) Fingerprinting genomes using PCR with arbitrary primers.
Nucleic Acids Res. 18, 7213±7218.
Williams, J.G.K., Kubelik, A.R., Livak, K.J., Rafalski, J.A. and Tingey, S.V. (1990) DNA
polymorphisms ampli®ed by arbitrary primers are useful genetic markers. Nucleic Acids Res.
18, 6531±6535.
Wong, Z., Wilson, V., Patel, I., Povey, S. and Jeffries, A.J. (1987) Characterization of a panel of highly
variable minisatellites cloned from human DNA. Ann. Hum. Genet. 51, 269±288.
Wright, J.M. (1993) DNA ®ngerprinting in ®shes. In Hochachka, P.W. and Mommsen, T., eds.
Biochemistry and Molecular Biology of Fishes, Vol. 2. Amsterdam: Elsevier, pp. 58±91.
Wright, J.M. and Bentzen, P. (1994) Microsatellites: genetic markers for the future. Rev. Fish. Biol.
Fish. 4, 384±388.
Wyman, A. and White, R. (1980) A highly polymorphic locus in human DNA. Proc. Natl Acad. Sci.
USA 77, 6754±6758.
Accepted 29 May 1997

Microsatellite Dna in Fishes

Загружено:

Сведения о документе

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Microsatellite Dna in Fishes

Загружено:

Авторское право:

Доступные форматы

Reviews in Fish Biology and Fisheries 7, 331±363 (1997)

Microsatellite DNA in ®shes

increased resolution provided by minisatellites over other non-VNTR-based marker

Nevertheless, existing data demonstrate that single-locus minisatellites can provide

1. Use good-quality (puri®ed using an organic extraction protocol) DNA when

Vector e.g. pUC18

Transformation e.g. DH5 α cells

Pick positives (black circles)

Design primers from sequence

Genomic DNA digest

Wash to remove unbound DNA.

Recover target DNA and

Clone into vector, construct library,

a marker system capable of detecting differences among closely related populations

Variable length inserts

Universal primers used

Clone into vector and sequence

Fig. 4. An illustration describing an enrichment procedure using an oligonucleotide probe bound to

complex within the North-west Atlantic is made up of reproductively discrete

Parentage and kinship analysis

Similarly, Kamonrat (1996) identi®ed successfully the parentage of communally reared

and make cross-laboratory comparisons dif®cult. Furthermore, variation in primer sites

Potential for scoring error

Product size is 120 bp

Product size is 150 bp

Product size is 100 bp

Base point mutation

Sample size required

Atlantic salmon, Salmo salar (Ssosl436). Anim. Genet. 26, 368.

Accepted 29 May 1997

Вам также может понравиться