BRIEF COMMUNICATIONS

Evolution, 59(11), 2005, pp. 2473–2479

ORIGIN OF THE SERPENTINE-ENDEMIC HERB LAYIA DISCOIDEA FROM THE WIDESPREAD

L. GLANDULOSA (COMPOSITAE)

BRUCE G. BALDWIN
Department of Integrative Biology and Jepson Herbarium, University of California, Berkeley, California 94720
E-mail: bbaldwin@berkeley.edu

Abstract. Phylogenetic analyses of nuclear rDNA sequences uphold Gottlieb et al.’s hypothesis that Layia discoidea,
a morphologically unusual, serpentine-endemic herb of narrow distribution in central California, ‘‘budded off’’ recently
(less than one million years ago) from a nearby lineage of the widespread L. glandulosa, which occurs on sandy soils
across much of far western North America. Although L. discoidea and L. glandulosa retain complete interfertility,
nuclear rDNA data for the two species are almost free of evolutionary noise, without evidence of gene flow between
them; allopatric divergence of L. discoidea cannot be ruled out. Molecular data are consistent with a hypothesis of
accelerated morphological evolution of L. discoidea and Gottlieb et al.’s suggestion that the closest relatives of L.
discoidea are populations of L. glandulosa with yellow, rather than white, ray corollas, in accord with Clausen, Keck,
and Hiesey’s evidence of a gene for yellow ray coloration in the rayless L. discoidea.

Key words. Edaphic endemics, parapatric speciation, peripatric speciation, progenitor-derivative relationships, quan-
tum speciation, reproductive isolation, serpentine.

Received March 16, 2005. Accepted September 6, 2005.

Evolutionary origins of edaphic endemics, that is, species
or lineages restricted to particular soils, have received much
attention by plant biologists (see reviews by Kruckeberg
2002; Rajakaruna 2004). In general, edaphic endemics have
been regarded either as relicts (paleoendemics) that under-
went contraction of geographic or ecological ranges with
changing climatic or biotic conditions or as products of re-
cent, in situ evolution (neoendemics), with divergence at least
partly accompanying colonization of the new substrate (e.g.,
Raven and Axelrod 1978; Kruckeberg 1986). Potential for
accelerated plant evolution on harsh soils has been long ev-
ident from theoretical (e.g., Raven 1964) and empirical (e.g.,
Antonovics et al. 1971) considerations. However, unequiv-
ocal phylogenetic evidence for rapid divergence of edaphic
endemics has been elusive, notwithstanding experimental
data demonstrating simple genetic bases for ecological or
morphological differences between some closely related spe-
cies that occur on highly contrasting substrates (Macnair
1983; Ford and Gottlieb 1990).
In a review of (31) putative examples of recent plant spe-
ciation involving ‘‘budding off’’ of a highly distinctive new
lineage from a presumably paraphyletic assemblage of lin-
eages of relatively uniform characteristics, Gottlieb (2004)
focused on three particularly well-studied species pairs that
have been examined using cytogenetic; electrophoretic; mor-
phological; and, in two of the three pairs, Mendelian genetic
analyses. One of the three pairs, studied here, comprises spe-
cies of annual herbs in the tarweed genus Layia (Compositae,
Madiinae) that occur in distinct edaphic situations, with the
presumed ‘‘progenitor’’ (L. glandulosa) on sandy soils across
much of far western North America (Fig. 1A, B) and the
presumed ‘‘derivative’’ (L. discoidea) on serpentine barrens
of the Inner South Coast Ranges of California, in the New
Idria–San Benito Mountain region (Fig. 1C). On serpentines,
severe challenges to plant growth, such as low levels of es-
sential nutrients (e.g., nitrogen, phosphorus, potassium), a
2473
q 2005 The Society for the Study of Evolution. All rights reserved.
low calcium-to-magnesium ratio, and potentially toxic levels
of heavy metals (e.g., nickel, chromium), have been impli-
cated in the evolution and assembly of distinctive floras; for
example, about 9% of endemic minimum-rank plant taxa in
the California Floristic Province are restricted to serpentines
(see Kruckeberg 1984, 1992, 2002).
As noted by Gottlieb (2004), molecular phylogenetic data
confirming ‘‘progenitor-derivative (P-D) relationships’’ in
plants are mostly lacking. Preliminary phylogenetic data from
a broad-scale analysis of 18S–26S nuclear ribosomal DNA
(nrDNA) internal transcribed spacer (ITS) sequences of Layia
that included two populations of L. glandulosa and one pop-
ulation of L. discoidea (Baldwin 2003) provided provisional
support for Gottlieb et al.’s (1985) hypothesis that the ser-
pentine endemic L. discoidea diverged recently from within
L. glandulosa. More detailed study of the origin of L. dis-
coidea is desirable to test further a prominent, putative ex-
ample of a P-D relationship involving rapid evolution of
morphology and edaphic specialization, as pursued here by
phylogenetic analyses of nrDNA ITS and external transcribed
spacer (ETS) sequences.
MATERIALS AND METHODS
Total DNAs were extracted from fresh, air-dried, or silica-
gel-dried leaves of one to 10 (pooled) plant(s) representing
populations that span the geographical and ecological dis-
tributions of Layia discoidea (five collections) and L. glan-
dulosa (21 collections), and one or two populations each of
other species of Layia and the sister genus, Lagophylla. Sam-
pling of L. glandulosa included white-rayed and yellow-rayed
populations, which are sometimes treated as distinct subspe-
cies or varieties; sampling of species other than L. discoidea
and L. glandulosa was based on broad-scale phylogenetic
analyses of Layia and subtribe Madiinae (Baldwin 2003). The
DNA extraction protocol was a modification of Doyle and
Doyle’s (1987) CTAB procedure (adding a phenol extraction,
2474 BRIEF COMMUNICATIONS

FIG. 1. Distributions, edaphic (soil) settings, and floral characteristics of lineages of Layia glandulosa and L. discoidea resolved from
nuclear ribosomal DNA external and internal transcribed spacer (ETS and ITS) region sequences. Numbers with symbols on distribution
maps for (A) western North America, (B) California, and (C) the region where L. discoidea grows correspond to numbered lineages of
L. glandulosa (Lg) in the ETS 1 ITS tree (D). The tree topology was obtained under maximum likelihood (ML; 2ln likelihood 5
2782.54156) and is identical to the strict consensus of both maximum parsimony (MP) trees and to one of the two MP trees (aligned
sequence matrix and trees are in TreeBase, study S1357). Branches are scaled to relative time; rate constancy of molecular evolution
across lineages could not be rejected (22ln likelihood ratio 5 16.2, 15 df, P , 0.5; test value 5 25.0 for m 5 0.05). Bootstrap values
under maximum parsimony (MP; above branches) and maximum likelihood (ML) are from analyses with outgroup included (before
slash) and outgroup excluded (after slash). Asterisks indicate branches where two parallel substitutions (C to T transitions at site 5 in
the 5.8S gene) were mapped with ACCTRAN; the remaining 48 (of 50) substitutions are without homoplasy among lineages of L.
discoidea and L. glandulosa (for ingroup only, consistency index 5 0.98; retention index 5 0.97). The outgroup (L. carnosa, L. gail-
lardioides, L. pentachaeta) is not shown; bootstrap support for monophyly of the ingroup lineage (L. discoidea plus L. glandulosa) is
100% under ML and MP. Broad-scale phylogeography of L. glandulosa will be addressed elsewhere.
 BRIEF COMMUNICATIONS 2475

TABLE 1. Resolved ETS 1 ITS lineages of Layia glandulosa (Lg) and number of collections representing each lineage (N), HKY85
distances (D) and number of nucleotide substitutions (based on ACCTRAN optimization, ML-tree topology) between each lineage and
L. discoidea (Ld), estimated maximum time since divergence from the most recent common ancestor (MRCA) with L. discoidea, in
millions of years (Ma), and probability of monophyly of L. glandulosa or subsets of lineages of L. glandulosa (i.e., excluding earlier
diverging lineages listed above an entry), based on results of Shimodaira-Hasegawa tests (1000 replicate RELL bootstrap; one-tailed
test). Locality and voucher data accompany sequences in GenBank for L. glandulosa (see below), L. discoidea (DQ188068–DQ188072),
and outgroup taxa (L. carnosa, DQ188045; L. gaillardioides, DQ188044; and L. pentachaeta, DQ188046).

HKY85 D (and Maximum divergence

Lineages of Lg no. of substitutions) time (Ma) from MRCA Monophyly of Lg or subsets
(N) from Ld with Ld of Lg lineages (P) GenBank accession nos.
Lg 1 (4) 0.0241 (26) 2.0 6 0.5 0.049 (Lg1–9 sister to Ld) DQ188047–DQ188050
Lg 2 (1) 0.0160 (17) 0.9 6 0.3 0.071 (Lg2–9 sister to Ld) DQ188051
Lg 3 (1) 0.0064 (7) 0.6 6 0.2 0.416 (Lg3–9 sister to Ld) DQ188052
Lg 4 (1) 0.0074 (8) 0.6 6 0.2 0.416 (Lg3–9 sister to Ld) DQ188053
Lg 5 (1) 0.0112 (12) 0.6 6 0.2 0.416 (Lg3–9 sister to Ld) DQ188054
Lg 6 (7) 0.0122 (13) 0.6 6 0.2 0.416 (Lg3–9 sister to Ld) DQ188055–DQ188061
Lg 7 (1) 0.0036 (4) 0.6 6 0.2 0.735 (Lg7–9 sister to Ld) DQ188062
Lg 8 (2) 0.0046 (5) 0.6 6 0.2 0.735 (Lg7–9 sister to Ld) DQ188063–DQ188064
Lg 9 (3) 0.0046 (5) 0.6 6 0.2 0.735 (Lg7–9 sister to Ld) DQ188065–DQ188067

RNase digestion, and two ethanol precipitations). Amplifi-
cation, cloning (for B. G. Baldwin 650; see Results), and
sequencing of the ITS region, that is, ITS-1, 5.8S gene, and
ITS-2, of nrDNA were conducted following methods of Bald-
win and Wessa (2000), with use of primer ITS5 (White et
al. 1990) rather than ITS-I for sequencing of most polymerase
chain reaction (PCR) products. A 474–674 bp segment of the
nrDNA ETS, immediately upstream of the 18S gene, was
amplified and sequenced following methods of Baldwin and
Markos (1998).
ITS and ETS sequences were aligned manually, without
uncertainty involving variable sites. Sequence matrices were
pruned to include only one representative of each set of iden-
tical sequences. Phylogenetic analyses using maximum par-
simony (MP) and maximum likelihood (ML) were conducted
using PAUP* 4.0 beta version 10 (Swofford 2002), with 20
random-addition sequences for each heuristic search. Choice
of molecular evolutionary models for ML analyses was de-
termined by likelihood-ratio (LR) tests using Modeltest, ver-
sion 3.06 (Posada and Crandall 1998). Reliability of clades
was assessed in PAUP* using nonparametric bootstrapping,
with MP (500 replicates; 20 random-addition sequences per
replicate) and ML (100 replicates; 10 random-addition se-
quences per replicate). Rate constancy of molecular evolution
across lineages was tested using a tree-wide LR test (Fel-
senstein 1988), as implemented in PAUP*. A maximum-age
calibration of 15 million years ago for the tree node repre-
senting the most recent common ancestor of Layia was based
on the assumption that diversification of the genus, with all
species native (and all but L. glandulosa endemic) to the
summer-dry California Floristic Province, would not have
preceded onset of summer drying in western North America
at mid-Miocene (e.g., Axelrod 1992; Flower and Kennett
1994; see Baldwin and Sanderson 1998). Based on that ex-
ternal calibration, maximum nodal ages were estimated from
ML branch lengths using r8s version 1.70 (Sanderson 2003),
and compared to nodal ages estimated using minimum and
maximum ITS nucleotide substitution rates reported for other
angiosperms (Richardson et al. 2001). Standard errors on
maximum divergence-time estimates were obtained using a
nonparametric bootstrap procedure (see Baldwin and San-
derson 1998). Shimodaira–Hasegawa (SH) tests (Shimodaira
and Hasegawa 1999; see Goldman et al. 2000), with RELL
optimization and 1000 bootstrap replicates (one-tailed test),
were conducted in PAUP* to test monophyly of L. glandu-
losa.
RESULTS
Phylogenetic analyses of ETS 1 ITS sequences for L. dis-
coidea, L. glandulosa, and three outgroup species yielded two
MP trees, one of which is identical to the ML tree; rate
constancy of molecular evolution across lineages could not
be rejected for either tree (Fig. 1). Based on accelerated-
transformation (ACCTRAN) mapping of ingroup character-
states on the resolved trees, only one of 50 nucleotide sub-
stitutions appears to have occurred twice, in parallel; the other
48 substitutions are without homoplasy, as reflected by an
ingroup consistency index of 0.98 and retention index of 0.97.
Four samples of L. discoidea and eight of L. glandulosa were
excluded from analyses because of complete identity with an
included ETS 1 ITS sequence. Another sample, morpholog-
ically identical to L. glandulosa (B. G. Baldwin 650), from
eastern San Luis Obispo County, California, was excluded
because some cloned ITS sequences from that DNA (yielding
highly polymorphic ITS sequences from pooled PCR prod-
ucts; clones and collection data in GenBank, accession num-
bers DQ188034–DQ188043) were referable to L. pentachae-
ta, which occurs nearby and retains moderate interfertility
with L. glandulosa (Clausen et al. 1941; B. G. Baldwin, un-
publ. data.). Monophyly of L. glandulosa was rejected using
the SH test (P 5 0.049; see Table 1), in keeping with the
ML and MP trees (Fig. 1).
All ETS 1 ITS sequences of L. discoidea were identical
and differed from those of L. glandulosa by a minimum of
four nucleotide substitutions and a maximum of 26 substi-
tutions, based on ACCTRAN mapping of states on the tree
in Figure 1. Using the same mapping approach, nonidentical
ETS 1 ITS sequences of L. glandulosa differed by a minimum
of one and a maximum of 29 substitutions. Pairwise HKY85
distances (G-shape parameter 5 0.3117, from Modeltest)
ranged from 0.4 to 2.4% in comparisons between sequences
2476 BRIEF COMMUNICATIONS

of L. discoidea and L. glandulosa and from 0 to 2.6% in

comparisons between sequences of L. glandulosa. Rejection
of rate-constancy of molecular evolution across lineages for
ETS 1 ITS data spanning all major lineages of Layia (data
not shown) led to an analysis confined to ITS sequences,
which appeared to undergo clocklike evolution in an earlier
genus-wide analysis (Baldwin 2003).
Phylogenetic analyses of ITS sequences of L. glandulosa,
L. discoidea, representatives of all lineages resolved as basal
to those species within Layia (Baldwin 2003), and two out-
group species in Lagophylla yielded one MP tree, identical
in topology to the ML tree; rate-constancy of molecular evo-
lution across lineages could not be rejected for that tree (Fig.
2). Estimated maximum time of divergence for the lineage
representing L. discoidea is 0.6 6 0.2 million years (Ma),
based on the 15 Ma external calibration (Fig. 2). The same
tree topology and age estimate for L. discoidea were obtained
when all distinctive ITS lineages resolved among the 14 rec-
ognized species in Layia were included in otherwise identical
analyses (B. G. Baldwin, unpubl. data). Applying minimum
and maximum rates of ITS evolution reported for other an-
giosperms (3.2 3 10210 to 9.0 3 1029 substitutions/site/year)
by Richardson et al. (2001) to the clock-constrained tree in
Figure 2, with branch-lengths including only ITS-1 and ITS-
2 substitutions, yielded divergence time estimates for L. dis-
coidea ranging from Late Pleistocene to Holocene ( #40,000
years ago).

DISCUSSION
Recent Descent of Layia discoidea from L. glandulosa
High bootstrap support for nested placement of all (iden-
tical) rDNA sequences of L. discoidea among rDNA lineages
of L. glandulosa (Figs. 1, 2) and rejection of monophyly of
L. glandulosa based on the conservative SH test (P 5 0.049;
Table 1) corroborate Gottlieb et al.’s hypothesis (Gottlieb et
al. 1985; Gottlieb 2004) that the evolution of L. discoidea
represents an example of budding off of an ecologically and
morphologically distinctive lineage (L. discoidea) from with-
in a set of lineages that otherwise have undergone relatively
conservative change, that is, a ‘‘P-D relationship’’ sensu Got-
tlieb (2004). A maximum, estimated divergence time for L.
discoidea of ,1 million years ago (Ma; Fig. 2; Table 1),
comparable to estimated divergence times for conspecific lin-
eages in closely related members of Layia (e.g., L. penta-
chaeta, see Fig. 2), upholds Gottlieb et al.’s (1985) sugges-
tion, based in part on genetic considerations, that L. discoidea
evolved rapidly. Late Pleistocene to Holocene divergence-
time estimates for L. discoidea (#40,000 years ago), based
on rates of ITS nucleotide substitution reported for other
angiosperms (Richardson et al. 2001), may be more realistic
than the maximum estimated divergence time (0.6 6 0.2 Ma)
obtained from basal calibration of Layia at 15 Ma, a date
which defensibly could be applied to a deeper node in trees
corresponding to subtribe Madiinae or tribe Madieae (see
Baldwin 2003).
Clausen et al. (1947), who originally suggested that L.
glandulosa is the closest relative of L. discoidea, considered
morphological differences between the two groups to be suf-
ficiently great to suggest that L. discoidea is a ‘‘relict . . .
FIG. 2. Maximum-likelihood (ML) tree (2ln likelihood 5
2213.25695) from analysis of ITS-region sequences for Layia dis-
coidea, L. glandulosa, representatives of all basally divergent lin-
eages of Layia, and the outgroup (Lagophylla minor and Lagophylla
glandulosa; not shown). The tree is identical topologically to the
single maximum-parsimony (MP) tree (aligned sequence matrix and
tree are in TreeBase, study S1357). Branch lengths correspond to
time; rate constancy of molecular evolution could not be rejected
(22ln likelihood ratio 5 14.7, 15 df, P , 0.5; test value 5 25.0
for m 5 0.05). Maximum divergence times shown are based on a
15 million years (Ma) calibration of the node corresponding to the
most recent common ancestor of Layia (see text). Bootstrap values
under MP (above branches) and ML (below branches) are presented
for lineages not represented in Figure 1. Numbered lineages of Layia
glandulosa correspond to those in Figure 1. Locality and voucher
data accompany ITS sequences in GenBank for collections of Layia
chrysanthemoides (DQ188076–DQ188077), L. heterotricha
(AF229313, DQ188075), L. pentachaeta (DQ188078–DQ188079),
L. septentrionalis (DQ188080–DQ188082), and Lagophylla
(AF229311, DQ188073–DQ188074). As expected, the estimated
maximum divergence time for L. discoidea is older than the Late
Pleistocene to Holocene divergence times for that species (#40,000
years ago) obtained using evolutionary rates of ITS substitution
reported for other angiosperms (see text).
that in the past may have been much more widely distributed
than now,’’ (p. 121) although they pointed out that deter-
mining whether L. discoidea or L. glandulosa evolved first
was not possible at the time. Based on evidence available
now, L. discoidea appears to be a serpentine neoendemic,
rather than a relict or paleoendemic, in keeping with Raven
and Axelrod’s (1978) suggestion for Californian serpentine-
 BRIEF COMMUNICATIONS
endemic herbs in general (but see Mayer et al. 1994). Local,
rapid divergence of L. discoidea is in keeping with biogeo-
graphic and geological considerations, as well; lineages of
L. glandulosa resolved as most closely related to L. discoidea
also occur in the South Coast Ranges of California, where
serpentine exposures are relatively young and correspond-
ingly less rich in endemic plants than exposures in north-
western California, a possible consequence of the limited
time for evolution of South Coast Range serpentine floras
(Harrison et al. 2004, 2006).
Morphological Evolution
Evidence for rapid divergence of Layia discoidea from an
ancestor in L. glandulosa conforms well with Gottlieb et al.’s
(1985) and Ford and Gottlieb’s (1989, 1990) detailed inter-
pretations of morphological evolution in L. discoidea. Layia
discoidea differs greatly in morphology from all other mem-
bers of Layia—most strikingly in lacking any ray florets or
true involucral bracts—and was to have been described orig-
inally in a new genus (‘‘Roxira’’) until Clausen et al. (1947)
determined that the newly discovered plants were completely
interfertile with L. glandulosa. Mendelian genetic analyses
by Clausen et al. (1947) led them to conclude that only two
major genes are responsible for presence or absence of ray
florets (and associated bracts) and that L. discoidea, although
lacking ray florets, has genes that modify ray floret number,
color, and size. Ford and Gottlieb (1989, 1990) confirmed
the general findings of Clausen et al. (1947) and greatly ex-
tended understanding of qualitative and quantitative trait dif-
ferences between L. discoidea and L. glandulosa. Such dif-
ferences were statistically significant for 17 morphological
and phenological characters and far exceeded interpopula-
tional differences in vegetative and floral traits within each
species (Gottlieb et al. 1985); origin of L. discoidea evidently
involved accelerated morphological evolution.
Ford and Gottlieb’s (1990) conclusion that a change at the
R locus, from homozygous for dominant alleles (RR) to ho-
mozygous for recessive alleles (rr), would be sufficient to
result in evolutionary loss of ray florets and associated bracts
underscores the potential for rapid evolution of even the most
conspicuously distinctive characteristics of L. discoidea.
However, their demonstration of independence of various
vegetative and reproductive characters that differ between L.
discoidea and L. glandulosa (Ford and Gottlieb 1989) is con-
sistent with a sufficiently long evolutionary history of L.
discoidea to allow for accumulation of (four) diagnostic
rDNA mutations, reported here. Gottlieb (2004) suggested
that L. discoidea may be somewhat older than two other prom-
inent examples of recent speciation in plants, Clarkia lin-
gulata and Stephanomeria malheurensis, based in part on the
number of morphological and ecological differences between
L. discoidea and L. glandulosa.
A phylogenetic perspective on morphological and ecolog-
ical variation among lineages of L. glandulosa reinforces and
extends Gottlieb et al.’s conclusions concerning the imme-
diate ancestry of L. discoidea. Gottlieb et al. (1985) suggested
that L. discoidea may be more closely related to yellow-rayed
than to white-rayed populations of L. glandulosa based in
part on considerations of morphological similarity and geo-
2477
graphic proximity. In addition, Gottlieb et al. (1985) noted
Clausen et al.’s (1947) evidence, confirmed subsequently by
Ford and Gottlieb (1990), that a gene in L. discoidea imparts
yellow ray coloration in hybrids with white-rayed L. glan-
dulosa, in keeping with an ancestrally yellow-rayed condition
in (rayless) L. discoidea. Based on the ETS 1 ITS tree (Fig.
1), the closest relatives of L. discoidea include representatives
of all sampled, uniformly yellow-rayed populations of L.
glandulosa, to the exclusion of all but one white-rayed pop-
ulation, discussed below.
Samples of L. glandulosa resolved among the closest rel-
atives of L. discoidea include two with morphological char-
acteristics that conceivably reflect transitional stages in the
evolution of L. discoidea. The yellow-rayed samples of L.
glandulosa include two collections (D. W. Taylor 9188, 9575,
in lineage Lg9; Fig. 1), from sites downstream (along Clear
Creek) from sampled populations of L. discoidea (B. G. Bald-
win 703, 1314), that approach L. discoidea in diminutive stat-
ure and have either unusually small ray florets or aberrant
raylike florets that may correspond to the ‘‘gibbous’’ mor-
phology seen by Ford and Gottlieb (1990) in recombinants
between L. discoidea and L. glandulosa. Absence of any in-
dication of L. discoidea ancestry in ETS and ITS sequences
of the small, yellow-rayed L. glandulosa in lower Clear Creek
is consistent with the possibility that those plants may rep-
resent forms transitional to L. discoidea morphology rather
than products of hybridization between L. discoidea and L.
glandulosa.
Evolutionary Transition to Serpentine Soils
The nested phylogenetic position of L. discoidea in L. glan-
dulosa confirms Gottlieb et al.’s (1985) hypothesis that an
ecological shift from sandy to serpentine soils occurred in
the evolutionary history of L. discoidea. Unusual ecological
settings of some of the populations of L. glandulosa resolved
as most closely related to L. discoidea may reflect some level
of preadaptation (exaptation) to serpentine conditions in the
recent ancestry of L. discoidea, as Kruckeberg (1986) sug-
gested to be a necessary precondition for the evolutionary
transition to serpentine soils or, alternatively, may reflect a
propensity for evolutionary shifts to serpentine in the lineage
including L. discoidea and its closest relatives. The same two
collections of L. glandulosa from lower Clear Creek that
appear transitional to L. discoidea in morphology (see above)
were reportedly from sites with some serpentine influence.
Similarly, the only collection of white-rayed L. glandulosa
placed among the closest relatives of L. discoidea in the ETS
1 ITS trees (D. W. Taylor 12376; lineage Lg7 in Fig. 1)
represents the only population of white-rayed L. glandulosa
to my knowledge from a site bordering directly on serpentine,
well to the north of the known distribution of L. discoidea
and without any rDNA evidence of past hybridization with
L. discoidea. To understand the ecological components of
evolutionary divergence of L. discoidea, inclusion of the
above edaphically unusual populations of L. glandulosa in
experimental studies may be critical.
Mode of Evolutionary Divergence
The origin of L. discoidea does not conform precisely to
most previously proposed models of allopatric speciation in-
2478 BRIEF COMMUNICATIONS
volving rapid budding off of peripheral isolates. Rapid chro-
mosome evolution or genetic changes leading to intersterility
with ancestors, integral to concepts of quantum evolution
(Simpson 1944) or quantum speciation (Grant 1963, 1981;
Lewis 1966; see, e.g., Raskina et al. 2004), are not evident
in L. discoidea, which retains complete interfertility with L.
glandulosa (Clausen et al. 1947; Ford and Gottlieb 1989,
1990). Mayr’s (1982) conception of peripatric speciation, al-
though generally invoking founder effects, allowed for the
possibility of rapid divergence of peripheral isolates based
on strong ecological selection alone, without genomic reor-
ganization, and selection has come to be widely regarded as
central to accelerated divergence of marginal populations
(e.g., Barton and Charlesworth 1984; Coyne 1992).
At a more general level, Coyne and Orr (2004, p. 124)
have noted that phylogenetic resolution of a sister-group re-
lationship between a narrowly distributed edaphic endemic
(‘‘. . . restricted to a small patch of aberrant habitat’’) and a
species in an adjoining area would constitute strong evidence
for parapatric, rather than allopatric, speciation. Data pre-
sented here for recent divergence of a serpentine endemic
from a paraphyletic species widely associated with sandy
soils leaves the directionality of ecological evolution in even
less doubt than for sister species that occur in contrasting
edaphic settings. Obstacles to a hypothesis of parapatric spe-
ciation for L. discoidea include lack of documented contact
zones with L. glandulosa, lack of homoplasy or other evi-
dence of hybridization in the ETS 1 ITS data for the two
species (even among most sampled lineages of L. glandu-
losa), and sufficiently large unbroken expanses of serpentine
habitat in the range of L. discoidea to conceivably allow for
geographic isolation (i.e., allopatric divergence cannot be
ruled out). Regardless of whether L. discoidea diverged in
parapatry or allopatry with L. glandulosa, experimental data
(not yet available) on relative fitness of L. discoidea, L. glan-
dulosa, and hybrids grown under environmental conditions
of each of the two species should yield critical insights about
the potential role of ecological factors in the origin and re-
productive isolation of L. discoidea.
Retention of complete interfertility between recognized
species of annual plants is unusual (although common in
perennials, see Grant 1981) and may lead to questions about
whether the origin of L. discoidea constitutes speciation.
Keck (1957, pp. 79–80), who described L. discoidea at species
rank, noted that the ‘‘ecological separation is apparently com-
plete’’ between L. glandulosa and L. discoidea, which he
regarded as having ‘‘separate phylogenetic histories without
having developed fertility barriers between them.’’ Unam-
biguous divergence of L. discoidea from L. glandulosa with-
out evidence of hybridization in the ETS 1 ITS tree (Fig. 1)
is consistent with Keck’s (1957) interpretation (although his-
tories of the two taxa cannot be regarded as completely ‘‘sep-
arate’’ in the sense of a sister-group relationship). From those
perspectives, recognition of L. discoidea at species rank con-
forms to phylogenetic species concepts and may be recon-
cilable with a biological species concept that regards eco-
logical isolation as a sufficient barrier to gene flow across
lineages (see Wheeler and Meier 2000). More precise knowl-
edge of the actual extent of ecological barriers between L.
discoidea and L. glandulosa may emerge from detailed studies
of edaphic factors and possible differences in pollination bi-
ology associated with loss of (showy) ray florets and reduc-
tion of head size in L. discoidea (see Gottlieb 2004), which
was not associated with a shift to self-compatibility—L. dis-
coidea is strongly self-incompatible (Clausen et al. 1947;
Ford and Gottlieb 1990). Potential importance of such eco-
logical factors in reproductive isolation of L. discoidea and
L. glandulosa may be reflected by ITS evidence for past hy-
bridization between L. glandulosa (B. G. Baldwin 650, see
Results) and L. pentachaeta, which diverged from a common
ancestor with L. glandulosa long before the origin of L. dis-
coidea (Fig. 2) but more closely resembles L. glandulosa in
inflorescence morphology and edaphic setting (see Clausen
1951).
Rarity of phylogenetic confirmation of P-D relationships
in plants may be a consequence of insufficient variation in
sequences under study, gene flow within the widespread
‘‘progenitor’’ subsequent to divergence, or extinction or lack
of sampling of lineages critical for resolving paraphyly of
the progenitor, in addition to the difficulty of finding evo-
lutionary examples at the appropriate stages of divergence.
Budding off of distinctive lineages has been regarded as po-
tentially common and important in plants (Rieseberg and
Brouillet 1994), as has evolutionary change at local geo-
graphic scales in general (e.g., Ehrlich and Raven 1969; Levin
1993; but see Rieseberg and Burke 2001). Groups that have
undergone such evolutionary change can present taxonomists
with difficult options, such as recognition of paraphyletic
species, dissection of paraphyletic species into cryptic mono-
phyletic species, or abandonment of rank-based taxonomy
(see Wheeler and Meier 2000). Layia glandulosa and L. dis-
coidea appear to represent a particularly well-established ex-
ample of a recent P-D relationship and, as suggested by Got-
tlieb (2004), a promising experimental system for determin-
ing genetic and functional characteristics associated with the
evolutionary transition to serpentine habitats.
ACKNOWLEDGMENTS
This paper is dedicated to L. D. Gottlieb and V. S. Ford
for their contributions to plant evolutionary biology through-
out remarkable careers. This research was supported in part
by the National Science Foundation (DEB-9458237) and R.
B. Park and other Friends of the Jepson Herbarium. I thank
L. Ahart, B. Cassidy, B. Guggolz and the late J. Guggolz,
K. Heise, R. Kelley, S. Markos, R. Raiche, D. W. Taylor, V.
Yadon, and the University of California Botanical Garden
for field assistance and providing collections; S. J. Bainbridge
for assistance with collecting and mapping of collections; P.
Beardsley, J. L. Strother, and an anonymous reviewer for
helpful comments on the manuscript, K. Klitz for assistance
with preparing Figure 1; C. A. Meacham for providing dis-
tributional data from the unpublished Synthesis of North
American Flora 2.0 (J. Kartesz and C. Meacham, 2004); R.
L. Moe for technical help; M. J. Sanderson for assistance
with r8s; and B. L. Wessa for extensive laboratory assistance.
LITERATURE CITED
Antonovics, J., A. D. Bradshaw, and R. G. Turner. 1971. Heavy
metal tolerance in plants. Adv. Ecol. Res. 7:1–85.
 BRIEF COMMUNICATIONS
Axelrod, D. I. 1992. Miocene floristic change at 15 Ma, Nevada to
Washington, U.S.A. Palaeobotanist 41:234–239.
Baldwin, B. G. 2003. A phylogenetic perspective on the origin and
evolution of Madiinae. Pp. 193–228 in S. Carlquist, B. G. Bald-
win, and G. D. Carr, eds. Tarweeds and silverswords: evolution
of the Madiinae. Missouri Botanical Garden Press, St. Louis.
Baldwin, B. G., and S. Markos. 1998. Phylogenetic utility of the
external transcribed spacer (ETS) of 18S–26S rDNA: congru-
ence of ETS and ITS trees of Calycadenia (Compositae). Mol.
Phylogenet. Evol. 10:449–463.
Baldwin, B. G., and M. J. Sanderson. 1998. Age and rate of di-
versification of the Hawaiian silversword alliance (Compositae).
Proc. Natl. Acad. Sci. USA 95:9402–9406.
Baldwin, B. G., and B. L. Wessa. 2000. Origin and relationships
of the tarweed-silversword lineage (Compositae-Madiinae). Am.
J. Bot. 87:1890–1908.
Barton, N. H., and B. Charlesworth. 1984. Genetic revolutions,
founder effects, and speciation. Annu. Rev. Ecol. Syst. 15:
133–164.
Clausen, J. 1951. Stages in the evolution of plant species. Cornell
Univ. Press, Ithaca, NY.
Clausen, J., D. D. Keck, and W. M. Hiesey. 1941. Experimental
taxonomy. Carnegie Inst. Wash. Yearbook 40:160–170.
———. 1947. Heredity of geographically and ecologically isolated
races. Am. Nat. 81:114–133.
Coyne, J. A. 1992. Genetics and speciation. Nature 355:511–515.
Coyne, J. A., and H. A. Orr. 2004. Speciation. Sinauer Associates,
Sunderland, MA.
Doyle, J. J., and J. L. Doyle. 1987. A rapid DNA isolation procedure
for small quantities of fresh leaf tissue. Phytochem. Bull. 19:
11–15.
Ehrlich, P. R., and P. H. Raven. 1969. Differentiation of popula-
tions: gene flow seems to be less important in speciation than
the neo-Darwinians thought. Science 165:1228–1232.
Felsenstein, J. 1988. Phylogenies from molecular sequences: in-
ference and reliability. Annu. Rev. Genet. 22:521–566.
Flower, B. P., and J. P. Kennett. 1994. The middle Miocene climatic
transition: east Antarctic ice sheet development, deep ocean cir-
culation and global carbon cycling. Palaeogeogr. Palaeoclimatol.
108:537–555.
Ford, V. S., and L. D. Gottlieb. 1989. Morphological evolution in
Layia (Compositae): character recombination in hybrids between
L. discoidea and L. glandulosa. Syst. Bot. 14:284–296.
———. 1990. Genetic studies of floral evolution in Layia. Heredity
64:29–44.
Goldman, N., J. P. Anderson, and A. G. Rodrigo. 2000. Likelihood-
based tests of topologies in phylogenetics. Syst. Biol. 49:
652–670.
Gottlieb, L. D. 2004. Rethinking classic examples of recent spe-
ciation in plants. New Phytol. 161:71–82.
Gottlieb, L. D., S. I. Warwick, and V. S. Ford. 1985. Morphological
and electrophoretic divergence between Layia discoidea and L.
glandulosa. Syst. Bot. 10:484–495.
Grant, V. 1963. The origin of adaptations. Columbia Univ. Press,
New York.
———. 1981. Plant speciation. 2nd ed. Columbia Univ. Press, New
York.
Harrison, S., H. Safford, and J. Wakabayashi. 2004. Does the age
of exposure of serpentine explain variation in endemic plant
diversity in California? Int. Geol. Rev. 46:235–242.
Harrison, S., H. D. Safford, J. B. Grace, J. H. Viers, and K. F.
Davies. 2006. Regional and local species richness in an insular
environment: serpentine plants in California. Ecol. Monogr. In
press.
2479
Keck, D. D. 1957. Trends in systematic botany. Surv. Biol. Prog.
3:47–107.
Kruckeberg, A. R. 1984. California serpentines: flora, vegetation,
geology, soils, and management problems. Univ. of California
Press, Berkeley.
———. 1986. An essay: the stimulus of unusual geologies for plant
speciation. Syst. Bot. 11:455–463.
———. 1992. Plant life of western North American ultramafics.
Pp. 31–73 in B. A. Roberts and J. Proctor, eds. The ecology of
areas with serpentinized rocks: a world view. Kluwer Academic
Publishers, Dordrecht, The Netherlands.
———. 2002. Geology and plant life: the effects of landforms and
rock types on plants. Univ. of Washington Press, Seattle.
Levin, D. A. 1993. Local speciation in plants: the rule not the
exception. Syst. Bot. 18:197–208.
Lewis, H. 1966. Speciation in flowering plants. Science 152:
167–172.
Macnair, M. R. 1983. The genetic control of copper tolerance in
the yellow monkey flower, Mimulus guttatus. Heredity 50:
295–302.
Mayer, M. S., P. S. Soltis, and D. E. Soltis. 1994. The evolution
of the Streptanthus glandulosus complex (Cruciferae): genetic
divergence and gene flow in serpentine endemics. Am. J. Bot.
81:1288–1299.
Mayr, E. 1982. Processes of speciation in animals. Pp. 1–19 in C.
Barigozzi, ed. Mechanisms of speciation. Alan R. Liss, New
York.
Posada, D., and K. A. Crandall. 1998. Modeltest: testing the model
of DNA substitution. Bioinformatics 14:817–818.
Rajakaruna, N. 2004. The edaphic factor in the origin of plant
species. Int. Geol. Rev. 46:471–478.
Raskina, O., A. Belyayev, and E. Nevo. 2004. Quantum speciation
in Aegilops: molecular cytogenetic evidence from rDNA cluster
variability in natural populations. Proc. Natl. Acad. Sci. USA
101:14818–14823.
Raven, P. H. 1964. Catastrophic selection and edaphic endemism.
Evolution 18:336–338.
Raven, P. H., and D. I. Axelrod. 1978. Origin and relationships of
the California flora. Univ. Calif. Publ. Bot. 72:1–134.
Richardson, J. E., R. T. Pennington, T. D. Pennington, and P. M.
Hollingsworth. 2001. Rapid diversification of a species-rich ge-
nus of Neotropical rain forest trees. Science 293:2242–2245.
Rieseberg, L. H., and L. Brouillet. 1994. Are many plant species
paraphyletic? Taxon 43:21–32.
Rieseberg, L. H., and J. M. Burke. 2001. The biological reality of
species: gene flow, selection, and collective evolution. Taxon
50:47–67.
Sanderson, M. J. 2003. R8s: inferring absolute rates of molecular
evolution and divergence times in the absence of a molecular
clock. Bioinformatics 19:301–302.
Shimodaira, H. and M. Hasegawa. 1999. Multiple comparisons of
log-likelihoods with applications to phylogenetic inference. Mol.
Biol. Evol. 16:1114–1116.
Simpson, G. G. 1944. Tempo and mode in evolution. Columbia
Univ. Press, New York.
Swofford, D. L. 2002. PAUP*. Phylogenetic analysis using parsi-
mony (*and other methods). 4.0 beta. Sinauer Associates, Sun-
derland, MA.
Wheeler, Q. D., and R. Meier, eds. 2000. Species concepts: a debate.
Columbia Univ. Press, New York.
White, T. J., T. Bruns, S. Less, and J. Taylor. 1990. Amplification
and direct sequencing of fungal ribosomal RNA genes for phy-
logenetics. Pp. 315–322 in M. Innis, D. Gelfand, J. Sninsky, and
T. White, eds. PCR protocols: a guide to methods and appli-
cations. Academic Press, San Diego, CA.
Corresponding Editor: M. Geber