ORIGINAL ARTICLE
doi:10.1111/j.1558-5646.2008.00457.x
ECOLOGICAL REPRODUCTIVE ISOLATION
OF COAST AND INLAND RACES
OF MIMULUS GUTTATUS
David B. Lowry,1,2,3
R. Cotton Rockwood,4
and John H. Willis1,2
1
University Program in Genetics and Genomics, Box 3565 Duke University Medical Center, Durham, North Carolina 27710
2
Department of Biology, Box 90338, Duke University, Durham North Carolina 27707
3
E-mail: david.lowry@duke.edu
4
Bodega Marine Laboratory, University of California, Davis, 2099 Westside Rd., Bodega Bay, California 94923
Received October 30, 2007
Accepted June 12, 2008
Adaptive divergence due to habitat differences is thought to play a major role in formation of new species. However it is
rarely clear the extent to which individual reproductive isolating barriers related to habitat differentiation contribute to total
isolation. Furthermore, it is often difficult to determine the specific environmental variables that drive the evolution of those
ecological barriers, and the geographic scale at which habitat-mediated speciation occurs. Here, we address these questions
through an analysis of the population structure and reproductive isolation between coastal perennial and inland annual forms
of the yellow monkeyflower, Mimulus guttatus. We found substantial morphological and molecular genetic divergence among
populations derived from coast and inland habitats. Reciprocal transplant experiments revealed nearly complete reproductive
isolation between coast and inland populations mediated by selection against immigrants and flowering time differences, but not
postzygotic isolation. Our results suggest that selection against immigrants is a function of adaptations to seasonal drought in
inland habitat and to year round soil moisture and salt spray in coastal habitat. We conclude that the coast and inland populations
collectively comprise distinct ecological races. Overall, this study suggests that adaptations to widespread habitats can lead to the
formation of reproductively isolated species.
KEY WORDS: Adaptation, drought, flowering time, population structure, salt tolerance, speciation.
It is commonly observed that recently diverged sister species
tend to inhabit environments with different ecological characteristics,
and possess habitat-specific adaptations (Mayr 1947;
Clausen 1951; Schemske 2000; Coyne and Orr 2004; Angert
2006; Nakazato et al. 2008). What is not always clear is the extent
to which such habitat differentiation contributes to the process of
speciation because adaptation to different habitats can influence
several potential prezygotic and postzygotic reproductive isolating
barriers (Via et al. 2000; Ogden and Thorpe 2002; Rundle
2002; McKinnon et al. 2004; Nosil 2007; for reviews see Schluter
2001; Coyne and Orr 2004; Lexer and Fay 2005; Rundle and Nosil
2005; Hendry et al. 2007). For example, if the different habitats
exist only in disparate geographic regions, then the habitat-related
adaptations may affect the overall distribution and range limits of
sister species, such that the species rarely if ever come into contact
with each other (Schemske 2000; Coyne and Orr 2004; Angert and
Schemske 2005). Such environmental imposition of an allopatric
distribution on sister species is often termed ecogeographic isolation
(Mayr 1947; Clausen 1951; Schemske 2000; Ramsey et al.
2003; Husband and Sabara 2004; Kay 2006). If environmental
variation exists on a finer spatial scale, then habitat adaptation
can cause reproductive isolation between species if immigrants
have low viability or fertility (e.g., Nosil et al. 2005). The strength
of immigrant inviability and ecogeographic isolation can be tested
2196
C 2008 The Author(s). Journal compilation C 2008 The Society for the Study of Evolution.
Evolution 62-9: 2196–2214
ECOLOGICAL ISOLATION IN MIMULUS
through reciprocal transplant experiments (Coyne and Orr 2004;
Nosil et al. 2005). Indeed, numerous reciprocal transplant experiments
conducted over the last 75 years support the role of local
adaptation in restricting gene flow between species (Turresson
1922; Clausen and Heisey 1958; Linhart and Grant 1996; Wang
et al. 1997; Nagy and Rice 1997; Angert and Schemske 2005;
Hall and Willis 2006; Rieseberg and Willis 2007).
In addition to ecogeographic isolation and immigrant inviability,
adaptation to alternative habitats may contribute to other
components of reproductive isolation. It is well known, for example
that in plants, sister species or even “ecotypes” adapted to different
habitats often evolve differences in flowering time (Clausen
1951; Clausen and Heisey 1958; Grant 1981). These differences
in the timing of reproduction can cause temporal isolation, potentially
a strong premating barrier between species (Coyne and Orr
2004; Martin and Willis 2007; Martin et al. 2007; Pascarella 2007;
Yang et al. 2007). Even if populations or species adapted to different
habitats occasionally interbreed, hybrids may be maladapted
to either parental habitat, resulting in extrinsic postzygotic isolation
(Wang et al. 1997; Hatfield and Schluter 1999; Rundle and
Whitlock 2001; Rundle 2002; Forister 2005; Campbell and Waser
2007). Because of these complex and multifaceted relationships
between habitat adaptation and reproductive isolation, the importance
of habitat adaptation relative to other potential prezygotic
and postzygotic barriers has rarely been quantified (but see Nosil
et al. 2005; Nosil 2007; Lowry et al. 2008). Furthermore, the specific
ecological factors driving ecological reproductive isolation
are often unknown. The geographic scale at which different stages
of speciation occur has remained poorly understood, especially
because geographically widespread population genetic analysis
is rarely conducted in conjunction with the quantification of reproductive
isolating barriers (Levin 1993; Coyne and Orr 2004;
Rundle and Nosil 2005).
In this study, we address these issues by examining the extent
of genetic divergence, habitat adaptations, population structure,
and reproductive isolation between coastal perennial and inland
annual forms of the wildflower Mimulus guttatus. Mimulus guttatus
(yellow monkeyflower; Phrymaceae, historically Scrophulariaceae,
order Lamiales) is a highly variable species within the even
more diverse M. guttatus species complex (Vickery 1978), which
is distributed over western North America. Different populations
often appear adapted to a multitude of elevational, climatic, and
edaphic habitats (Vickery 1978; Wu et al. 2008). Throughout the
lower elevation inland meadows, seeps, and rocky outcrops, from
the Pacific coastal mountain range of the United States and Canada
east to the Rockies, M. guttatus populations are typically composed
of spindly spring flowering annual plants (Vickery 1952;
Clausen and Heisey 1958; Wu et al. 2008). In adjacent fog-bound
cliffs, sand dunes, and coastal terrace that occur within a couple
of kilometers of the Pacific Ocean, M. guttatus populations
are composed of large, robust, late summer flowering perennial
plants with compressed internodes (Vickery 1952; Clausen and
Heisey 1958; Hitchcock and Cronquist 1973; Hall and Willis
2006). Based on field collections, the coastal perennial populations
have frequently been assigned a separate taxonomic status:
M. guttatus var. grandis Greene (Hitchcock and Cronquist 1973)
and M. guttatus ssp. litoralis Pennell (Abrams and Ferris 1951).
However, common garden experiments are needed to demonstrate
that the morphological distinctness of coastal populations is not
due to phenotypic plasticity. Inland perennial forms of M. guttatus
are also found around permanently moist springs, creeks,
and lakes throughout western North America (Vickery 1952;
Clausen and Heisey 1958), but they are not included in the present
study.
A recent reciprocal transplant between a pair of coast and
inland M. guttatus populations from Oregon provides evidence of
strong local adaptation to their respective habitats (Hall and Willis
2006). The inland habitat of the coast ranges of California and
Oregon experiences a very long (∼6 month), hot summer drought,
in which moist areas (seeps, creeks, arroyos) that contain annual
M. guttatus populations completely dry out. Coastal habitat also
receives little or no rainfall during summer months, but persistent
fog and cooler temperatures along the coast maintain year-round
soil moisture and reduce the rate of plant transpiration (Corbin
et al. 2005). However, coastal plants must contend with ocean
salt spray, which is a major stress that is known to restrict the
distribution of many plant species (Boyce 1954; Barbour 1978;
Humphreys 1982; Wilson and Skyes 1999).
Here we report on a series of experiments to understand
the evolution of reproductive isolation between coast and inland
populations of M. guttatus. Using molecular genetic markers and
common garden experiments, we ask whether multiple, broadly
sampled coast and inland populations constitute two distinct morphological
and genetically structured groups. Using additional reciprocal
transplant experiments, we determine the extent to which
coast and inland populations are locally adapted to their respective
habitats. Finally, we combine results from field and greenhouse
experiments to estimate the strength of multiple prezygotic and
postzygotic barriers between coast and inland populations, and determine
the role of adaptations to specific environmental factors
in limiting hybridization and introgressive gene flow.
Materials and Methods
FIELD COLLECTIONS
In the spring of 2005, plants were collected from 14 latitudinal
pairs of coast and inland populations (two extra inland populations,
30 populations total, 20–30 haphazardly collected plants
per population) distributed from central California to northern
Oregon (Fig. 1, Table 1). Inland populations used in this study
EVOLUTION SEPTEMBER 2008 2197
DAVID B. LOWRY ET AL.
Figure 1. Map of western United States showing locations of coast (black) and inland (gray) populations used in this study. Arrows
indicate coast/inland population pairs used in common garden greenhouse experiment. Yin-yang symbol indicates locations of reciprocal
transplant experiments in this study (California) and that of Hall and Willis (2006) in Oregon. Stars represent population pairs used to
test for salt tolerance differences. Ovals denote population pairs used for analysis of intrinsic postzygotic isolation. Numbers correspond
to populations listed in Table 1.
were located 4.8–31.4 (mean = 16.7) kilometers from the Pacific
Ocean. Coastal plants were all collected within 500 m of the
ocean. Collected plants were shipped overnight to Durham, NC,
where they were potted, raised to flowering, and self-pollinated
in the Duke University greenhouses. Selfed seeds and tissue for
DNA extraction were collected from these plants. Tissue was
placed into 96-well Costar Plates and immediately deposited into
an −80◦
C freezer.
MORPHOLOGICAL POPULATION STRUCTURE
To determine the extent of genetically based phenotypic divergence
between coastal and inland populations; a common garden
greenhouse experiment was conducted using six pairs of coast
and inland populations and one extra coast population, for a total
of 13 populations. Replicates within each population consisted of
12 selfed individuals, each descended from separately collected
maternal plants. Some seeds failed to germinate, resulting in less
than 12 replicates for some populations.
Seeds were sown individually in Fafard-4P soil in 4 in. square
pots and were stratified in a dark room at 4◦
C for one week.
Pots were then moved to the Duke University greenhouses for
seed germination and subsequent growth. Greenhouse conditions
consisted of 18-h days at 21◦
C with supplemental high-pressure
sodium lights and 6-h nights at 16◦
C. Relative humidity was
maintained at 30%. The location of all plants was fully randomized
on the greenhouse bench.
2198 EVOLUTION SEPTEMBER 2008
ECOLOGICAL ISOLATION IN MIMULUS
Table 1. Coast and inland populations of M. guttatus used in this study. For each population longitude and latitude, the number of
samples (N), the expected heterozygosity (HS), mean number of alleles per locus (Na), mean allelic richness (RS), and mean inbreeding
coefficient (FIS) are listed. Ten codominant markers were used to calculate statistics in FSTAT 2.9.3.2 (Goudet 2001). “Num” corresponds
to the label of populations in Figure 1.
Race Pop Num Location Latitude Longitude N Na HS RS FIS
ID (N) (W)
M. guttatus OSW 1 Oswald West SP, Tillamook Co., OR 45◦
45 39 123◦
57 56 20 2 0.263 1.80 0.236
(Coast) CKI 2 Cape Kiwanda SP, Tillamook Co., OR 45◦
14 31 123◦
58 07 16 1.9 0.252 1.75 0.246
HEC 3 Heceta Lighthouse, Lane Co., OR 44◦
08 06 124◦
07 22 14 2.1 0.297 1.88 0.308
OPB 4 Otter Point SP, Curry Co., OR 42◦
27 50 124◦
25 22 20 2.6 0.227 2.05 0.056
GBM 5 Gold Bluffs Marsh, Humbolt Co., CA 41◦
22 43 124◦
04 10 15 1.6 0.096 1.36 0.283
CMD 6 Cape Mendocino, Humbolt Co., CA 40◦
24 33 124◦
23 31 16 2.6 0.350 2.16 0.382
USB 7 Usal Beach, Mendocino Co., CA 39◦
49 55 123◦
50 57 20 2.6 0.320 2.14 0.275
NAV 8 Navarro River, Mendocino Co., CA 39◦
11 12 123◦
45 26 16 2.3 0.277 2.04 0.008
SWB 9 Irish Beach, Mendocino Co., CA 39◦
02 09 123◦
41 25 20 1.8 0.234 1.61 0.260
SRN 10 Sea Ranch, Sonoma Co., CA 38◦
44 07 123◦
29 23 16 2.1 0.339 2.00 0.436
MRR 11 Russian River, Sonoma Co., CA 38◦
27 23 123◦
08 27 14 2.6 0.353 2.28 0.283
DAV 12 Davenport Beach, Santa Cruz Co., CA 37◦
01 29 122◦
13 02 20 2.6 0.374 2.20 0.155
BCB 13 Big Creek Reserve, Monterey Co., CA 36◦
03 46 121◦
35 31 16 2.5 0.277 2.10 0.431
ORO 14 Montana de Oro, San Luis Obisbo Co., CA 35◦
16 24 120◦
53 20 16 3 0.384 2.41 -0.028
M. guttatus SAM 15 Saddle Mountain SP, Clatstop Co., OR 45◦
57 33 123◦
40 46 19 5.5 0.505 3.72 -0.016
(Inland) LIN 16 Little Nestuca River, Tillamook Co., OR 45◦
08 09 123◦
53 46 16 5 0.534 3.60 0.300
SWC 17 Mapleton, Lane Co., OR 43◦
57 34 123◦
54 08 11 4.2 0.655 3.77 0.169
RGR 18 Rouge River, Curry Co., OR 42◦
29 21 124◦
12 30 16 4.3 0.561 3.30 0.377
BHI 19 Bald Hills, Humbolt Co., CA 41◦
09 17 123◦
53 22 20 4 0.446 2.99 0.312
BSR 20 Rio Dell, Humbolt Co., CA 40◦
31 46 124◦
09 46 16 4.1 0.482 3.07 0.165
ANR 21 Angelo Reserve, Mendocino Co., CA 39◦
44 12 123◦
37 51 20 5.2 0.565 3.58 0.232
SDA 22 Boonville, Mendocino Co, CA 39◦
01 05 123◦
19 08 14 4.9 0.672 3.81 0.359
RNC 23 Rancheria Creek, Mendocino Co, CA 38◦
54 40 123◦
14 42 18 8.3 0.701 5.06 0.115
LMC 24 Yorkville, Mendocino Co., CA 38◦
51 50 123◦
05 02 18 6.4 0.537 4.10 0.179
USK 25 Skaggs-Springs, Sonoma Co., CA 38◦
40 20 123◦
12 36 16 6.3 0.668 4.47 0.114
GUA 26 Gualala River, Sonoma Co., CA 38◦
40 05 123◦
18 41 12 4.3 0.536 3.43 0.008
OAE 27 Occidental, Sonoma Co., CA 38◦
24 40 122◦
57 34 12 3.6 0.507 3.04 0.309
LOR 28 Boulder Creek, Santa Cruz Co., CA 37◦
06 30 122◦
07 04 15 4.6 0.634 3.64 0.180
CAN 29 Big Creek Reserve, Monterey Co., CA 36◦
04 08 121◦
33 05 16 1 0.000 1.00 SLO
30 Morro Road, San Luis Obispo Co., CA 35◦
27 38 120◦
44 24 - - - - For
each plant we recorded the number of days from seed
germination to first flowering and at the same time measured 12
morphological traits using digital calipers (first and second internode
length, first internode length, first and second leaf width
and length, leaf thickness, maximal corolla width and length,
plant height, and rosette diameter). We also counted the number
of flowers 10 days after the first flower opened. To determine
whether there is overall multivariate morphological divergence
between coast and inland populations, a multivariate analysis of
variance (MANOVA) was implemented with the 13 morphological
traits and flowering time as the dependent variables and habitat
(coast or inland) as the independent variable. To visually assess
morphological divergence between coast and inland populations,
principle components analysis was conducted using the 14 traits,
and subsequently PC1 was plotted against PC2. To determine
how morphological variation is partitioned between coast/inland
habitats (fixed effect), among populations within habitats (random
effect), and among individuals (error) within populations, a
REML mixed-model nested analysis of variance (ANOVA) was
conducted individually for each of the first two PCs and the 14
measured traits. To estimate the percent of variation partitioned
among the hierarchical levels, the REML-nested ANOVA was
rerun treating habitat as a random effect. All morphological analyses
were implemented in JMP 6.0 (SAS Institute, 2005, Cary,
NC).
MOLECULAR GENETIC POPULATION STRUCTURE
To determine if coast and inland populations are differentiated
at loci located throughout the nuclear genome, molecular population
genetic analysis was conducted with all 30 collected
EVOLUTION SEPTEMBER 2008 2199
DAVID B. LOWRY ET AL.
populations. Genomic DNA was extracted from plant tissue using
a modified hexadecyl trimethyl-ammonium bromide chloroform
extraction protocol (Kelly and Willis 1998). We used a total of
10 codominant markers for population genetic analysis, including
six microsatellites (Kelly and Willis 1998) and four markers that
reveal length polymorphisms in the introns of single copy nuclear
genes (as described in Fishman and Willis 2005; Sweigart
et al. 2006). Primer sequences are available in the online Supplementary
Table S1. All PCR products were subjected to capillary
electrophoresis and fragment analysis on an ABI 3730×l DNA
Analyzer. Size of the amplified fragments was scored automatically
by the program GENEMARKER (SoftGenetics, 2005, State
College, PA) and was confirmed by eye. The chromosomal location
of all loci has been established by ongoing mapping studies.
Although some markers were located on the same linkage
group, they were never closer than 36.5 cM apart from each other
(MgSTS423 and AAT230). Linkage disequilibrium is highly improbable
at such large distances, but analysis with FSTAT 2.9.3.2
(Goudet 2001) was carried out to confirm that this is the case.
For all populations, we calculated the observed number of alleles
(Na), private alleles restricted to coast or inland habitat (Pa),
observed heterozygosity (HO), expected heterozygosity (HS), total
gene diversity (HT), allelic richness (RS), level of inbreeding
(FIS), and overall genetic differentiation (FST). Genetic differentiation
among all pairs of populations was also quantified as
pairwise FST (Weir and Cockerham 1984). All summary statistics
were calculated with FSTAT (Goudet 2001). In addition, FSTAT
was implemented for 1000 permutations to test for significant
differences in HS, HO, FST, and FIS between coast and inland
populations.
To test for population structure between coast and inland
populations we employed three different methods. Analysis of
molecular variation (AMOVA, Excoffier et al. 1992) was implemented
to determine how genetic variation is partitioned between
coast/inland habitats, among populations within habitats,
and among individuals within populations. AMOVA was implemented
in the program Arlequin 3.11 (Excoffier et al. 2005).
The program STRUCTURE is a multilocus model-based
clustering method that assigns individuals to a predefined number
of populations (K) and detects admixed/migrant individuals
(Pritchard et al. 2000). STRUCTURE was run for three iterations
for K-values from 2 to 35, using the admixture model, independent
allele frequencies, λ = 1, with a 200,000 burnin and 200,000
MCMC. We reran STRUCTURE 50 times at K = 2 and combined
the result of those runs with the program CLUMMP (Jakobsson
and Rosenberg 2007) and visualized this combined data with
DISTRUCT (Rosenberg 2004).
The program POPULATION GRAPH uses graph theory
techniques to determine the topological relationship among populations
that may currently be exchanging genes (Dyer and Nason
2004). This method is free of a priori assumptions about population
geographic arrangements, unlike AMOVA, and works by
simultaneously determining the high-dimensional covariance relationships
among all populations using the genetic marker data.
The program then determines the minimum set of edges (connections)
that sufficiently explain the total among population
covariance structure of all of the populations. The network of
population connections can then be analyzed by various post hoc
analyses. POPULATION GRAPH was implemented on the web
(http://dyerlab.bio.vcu.edu/wiki/index.php/) using the population
genetic dataset. A test for distinct clustering of coast and inland
populations was conducted post hoc using the methods outlined
by Dyer and Nason (2004).
RECIPROCAL TRANSPLANT EXPERIMENTS
To test for local adaptation and reproductive isolation between
coast and inland populations, reciprocal transplant experiments
were conducted in California. Two sets of reciprocal transplants
experiments, at different latitudes, were used to test the generality
of trends and to compare to the results of a previous transplant
experiment in Oregon (Hall and Willis 2006). The field sites for
the experiment were located in Mendocino County (Experiment
1) and on the University of California, Big Creek Reserve in
Monterey County (Experiment 2). The Mendocino coastal field
site was located on seepy cliffs, just north of Manchester Beach
state park (coast site 1, N39◦
00 29 , W123◦
41 38 ), whereas the
Mendocino inland field site was located in hilly oak savanna habitat
near Boonville, CA (inland site 1, N38◦
59 13 , W123◦
21 03 ,
28.7 km from the ocean). The Big Creek coastal site was
located near the reserve’s bridge (coast site 2, N36◦
04 12 ,
W 121◦
36 00 ), and the inland Big Creek site was located at
Shakemaker Meadow in mix grassland/chaparral habitat (inland
site 2, N36◦
03 41 , W121◦
33 16 , 3.5 km from the ocean). See
online Supplementary Figure S1 for location of field sites and
seed source populations.
Plants growing at reciprocal transplant field sites might be
adapted to highly local environmental factors instead of common
features of inland or coastal habitats. To control for highly local
adaptation and eliminate the possibility of genetic contamination
of source populations, field sites away from seed source populations
were used for field experiments (online Supplementary Fig.
S1). Seeds for experiment 1 were derived from the SWB (coast)
and LMC (inland) populations, whereas the seeds for experiment
2 were derived from the BCB (coast) and CAN (inland) populations
(see online Supplementary Fig. S1, Table 1 for location of
these populations). Field sites in experiment 1 were located within
existing M. guttatus populations. Field sites in experiment 2 were
placed into habitat that appeared suitable for M. guttatus because
UC Reserve restrictions forbid conducting transplant experiments
within existing plant populations of the same species. All seeds
2200 EVOLUTION SEPTEMBER 2008
ECOLOGICAL ISOLATION IN MIMULUS
were outbred and were derived from lines that had been grown
in the greenhouse at least one full generation to reduce maternal
effects. Four to six independent outbred full-sibling families were
planted from each population, with the intention to capture much
of the genetic variation within each population. At each field site,
100 inland parental plants, 100 coast parentals, and 100 F1s were
planted. F1s were derived from crosses between coast and inland
parentals. In sum, 300 plants were planted at each field site, 600
plants per reciprocal transplant experiment, and 1200 plants total.
Because of concern that Mimulus seeds would be displaced
in the field, plants were transplanted at the seedling stage. To
try to capture variation of germination conditions mediated by
site-specific environmental factors, seeds were germinated in the
soil of their eventual destination in December 2005. This timing
allowed for the transplantation of individuals at the four-leaf stage,
which was comparable to the developmental stage of native plants
found in situ. Plants were grown in soil-filled flats at the Bodega
Marine Laboratory greenhouses until transplantation in January
2006. Rosette diameter, plant height, flowering time, leaf damage,
and number of flowers produced were recorded for all plants every
two to three weeks beginning on April 20, 2006 and continuing
through December 27, 2006. Flowers and immature seedpods
were removed from plants to reduce genetic contamination of
local populations. This procedure prevented the collection of data
on the number of seeds produced and thus, it was not possible to
calculate full lifetime fitness.
ANALYSIS OF LOCAL ADAPTATION AND HYBRID
PERFORMANCE
To test for local adaptation and to analyze the performance of
hybrids relative to native parental plants, data were analyzed with
the program ASTER (Geyer et al. 2007). ASTER is a maximumlikelihood
method for analysis of multiple fitness components.
This method represents an improvement over previous methods
(e.g., ANOVA) for the analysis of reciprocal transplant experiments
because it accounts for dependencies among different components
of fitness and properly incorporates variables that have
different probability distributions (Geyer et al. 2007).
For all analyses, ASTER was used to combine two dependent
components of fitness, (1) survival to flowering and (2) number of
flowers produced by plants that survived to flower. We fit multiple
nested models to each set of data and compared these models
using likelihood-ratio tests. The models listed here are the ones
that best fit the data. A two-way analysis with ASTER was used
to test for local adaptation of coast and inland populations. The
model included genotype (coast or inland), family, and field site
as independent variables. In this case, local adaptation is defined
as a significant genotype × site interaction.
A separate analysis was conducted to assess extrinsic postzygotic
isolation because this analysis involves the direct comparison
of hybrids to local parentals. For the model that best fits
this data, genotype (local parental or hybrid) was the independent
variable and performance (survival to flowering combined with
number of flowers produced) was the dependent variable.
SALT SPRAY TOLERANCE
Oceanic salt spray is known to cause leaf necrosis (Boyce 1954).
To determine if coast, inland, and F1 hybrid plants differed in salt
spray tolerance in the field, we assessed percentage leaf damage
during each visit to coastal field sites. Percentage leaf damage
from April 26, 2006 was used for analysis of salt tolerance at
coast site 1 because enough plants were still alive to make this
comparison. Data were analyzed with a one-way ANOVA, where
type of plant was the independent variable and percent salt damage
was the dependent variable. Tukey-Kramer HSD tests were used
for post hoc comparisons of means (JMP 6.0).
Greenhouse experiments were used to more directly determine
salt spray tolerance of coast and inland populations. Plants
from three latitudinal pairs of inland/coast populations were used
in this experiment (LMC/SWB-California, RGR/OPB-Southern
Oregon, SAM/OSW-Northern Oregon, 25 individuals per population).
In addition, salt spray tolerance was assessed for 25
F1 hybrids of a cross between the LMC and SWB populations,
which are the same populations used in reciprocal transplant experiment
1. Plants were grown under greenhouse conditions as
described in the common garden experiment (above). All plants
were sprayed with ∼15 mL of 500 mM NaCl solution (approximately
same concentration as ocean water) every other day starting
four weeks after germination. Salt spray application leads to
accumulation of salt directly on leaves as well as a build up in
the soil, and in this way mimics conditions in the field at coastal
sites. Percentage leaf damage and date of death (complete leaf
necrosis) were recorded every other day. To compare the rate
of accumulation of leaf damage between coast/inland habitats,
among populations within habitats, and among individuals within
populations, data were analyzed with a nested repeated-measures
MANOVA. Here, damage at each time point was treated as a
separate dependent variable. Time till death was compared with
a nested ANOVA, where population was nested within coast or
inland type (JMP 6.0). Tukey-Kramer HSD tests were used for
post hoc comparisons of mean time till death of LMC (inland)
and SWB (coast) plants, as well as F1 hybrids between these two
populations.
INTRINSIC POSTZYGOTIC REPRODUCTIVE ISOLATION
To assess the level of intrinsic postzygotic isolation between coast
and inland populations, crosses were made among four pairs of
coast and inland populations from central California to northern
Oregon (Fig. 1). Four to ten plants from each population were
used in these crosses. Five different crosses were made for each
EVOLUTION SEPTEMBER 2008 2201
DAVID B. LOWRY ET AL.
coast/inland population pair, for a total of 20 crosses. All F1
hybrids were then intercrossed to produce F2 hybrids. To screen
for hybrid lethality/inviability, 86 F1 hybrids representing all 20
crosses, as well as 613 F2s hybrids, were grown under common
garden greenhouse conditions (above). Hybrid lethality and inviability
were scored for all crosses. For purposes of this study,
hybrid lethality is defined as death of plants at or before the fourleaf
stage. Hybrid inviability is defined as dwarfing, necrosis,
and the inability to survive to flower under standard greenhouse
conditions (as described above).
STRENGTH OF REPRODUCTIVE ISOLATING BARRIERS
The strength of reproductive isolating barriers between coastal
and inland population was calculated from the reciprocal transplant
experiments using methods modified from previous studies
(Ramsey et al. 2003; Kay 2006; Martin and Willis 2007; Lowry
et al. 2008). Potential for gene flow into inland habitat was calculated
separately from potential gene flow into coastal habitat for
reproductive isolating barriers that act asymmetrically. Data from
both inland field sites were combined for these analyses, but not
for coastal field sites (see Results).
To assess the potential for hybridization via pollen dispersal
between habitats, we determined the approximate flowering
phenologies of experimental plants growing in their native habitat
by conducting censuses every 2–3 weeks and recording the total
number of flowers produced since the previous census. To calculate
a crude, conservative measure of reproductive isolation due
to less than complete flowering time overlap, we assumed that
pollinators disperse pollen at random among flowers within and
between the two habitats despite allopatry, that the abundances
of native flowers at the two habitats are equal during any flowering
overlap period, and that all flowers produce equal seeds and
pollen grains. With these admittedly unrealistic assumptions, the
frequency of hybrids expected with complete flowering overlap
is 0.5 at each site. Let the proportion of coastal flowers that were
open during the overlap period be FC and that of inland flowers
be FI. Given the above assumptions, the expected frequency
of hybrids produced as seeds at the coastal site, qC, is equal to
FC /2, and that for seeds produced at the inland site is qI = FI /2.
Following the reasoning outlined in Martin and Willis (2007),
the reproductive isolation at the coastal site due solely to flowering
phenology differences between plants in allopatric habitats
is RIFA,C = 1 − [qC /(1 − qC )], with an analogous formula for
isolation at the inland site.
Gene flow can also occur between species through seed dispersal.
However, seeds that disperse between habitats must germinate,
survive, and flower in the nonnative habitat in order for
hybridization to potentially occur. Thus, habitat-mediated selection
against immigrants, which may prevent survival and limit
reproduction, can act as an impediment to gene flow (Nosil et al.
2005). We calculated the strength of reproductive isolation due
to selection against immigrants as RII = 1 − ¯wi /¯wn, where ¯wi is
the mean number of flowers produced by immigrant individuals,
and ¯wn is the mean number of flowers produced by individual
plants native to the habitat of the field site.
For immigrants that survive to flower, hybridization cannot
occur if immigrants flower at different times than native plants.
To determine if flowering time differences restricts hybridization
between immigrants and native plants in artificial sympatry, we
calculated the probability that an individual immigrant flowered
at the same time as any members of the native population and
estimated this form of temporal reproductive isolation between
immigrants and natives in sympatry, say at the coastal site, as
RI FS,C = 1 − F M,I where F M,I is the probability that a migrant
from the inland population flowers whereas any native plants are
flowering at the coastal site.
Postzygotic isolation can occur through intrinsic genetic incompatibilities.
To determine the strength of intrinsic postzygotic
isolation, the mean seedling to adulthood viability of F1 hybrids,
¯vF1, was compared to the mean viability of parental populations.
We did not include data from the F2 generation in our calculation
as it is unclear what the relative frequency of F2 versus backcross
hybrids would be in a natural situation. Because we are interested
here in intrinsic barriers, we focused on viability in the relatively
benign environment of our greenhouse experiments. Because all
parental individuals survived from seedling to adulthood in these
experiments, we calculated intrinsic postzygotic reproductive isolation
as RII P = 1 − ¯vF1/¯vP .
Postzygotic isolation could also occur, even where there is no
intrinsic postzygotic isolation, if the hybrids are less fit than native
parental plants due to extrinsic ecologically based inviability or
sterility (Hatfield and Schluter 1999). The strength of extrinsic
postzygotic reproductive isolation was calculated as RIE P = 1 −
¯wh/¯wn, where ¯wh is the mean lifetime number of flowers produced
by F1 hybrids in the field, and ¯wn is the mean lifetime number of
flowers produced by native parental plants.
Results
MORPHOLOGICAL POPULATION STRUCTURE
Striking morphological differences were found between coast and
inland populations in the common garden greenhouse experiment.
Joint analysis of 13 morphological traits and flowering time
showed substantial quantitative genetic divergence between coast
and inland habitats (MANOVA, F13,83 = 61.20, P < 0.0001).
In general, the coastal populations flowered later, had thicker
stems, shorter internodes, and larger flowers than inland populations
(Fig. 2). Plotting of PC1 and PC2 showed clear differences
between coast and inland individuals (Fig. 2C; see online Supplementary
Table S2 for trait loadings on PC1 and PC2). Most of
2202 EVOLUTION SEPTEMBER 2008
ECOLOGICAL ISOLATION IN MIMULUS
Figure 2. Morphological divergence of coast and inland populations.
(A) Morphological differences between coast (left) and inland
(right) populations grown in a common garden greenhouse
environment. (B) Flowering time was significantly different between
coast and inland races (F = 49.55, P < 0.0001). See Table 2
for analysis of all traits. Error bars denote one standard error. (C)
Principle components analysis (PC1 plotted against PC2) of individuals
from coast (closed circle) and inland populations (open circle)
using morphological data (14 traits, N = 12 populations, 107 individuals).
Data for tetraploid SLO population removed from this
analysis.
the variation in flowering time (71.9%) and PC1 (88.6%), but not
PC2 (6.1%), was partitioned between groups (coast vs. inland) in
the REML-nested ANOVA (see Table 2 for other traits). It should
be noted that we excluded the tetraploid (see below) inland SLO
population for all morphological analysis.
MOLECULAR GENETIC POPULATION STRUCTURE
All 10 codominant markers were highly polymorphic and successfully
amplified alleles in all 30 populations. No linkage disequilibrium
was detected among any of the 10 loci using FSTAT
2.9.3.2 (Goudet 2001). Two populations (SLO, CAN) had aberrant
molecular signatures and were removed from the subsequent
analyses. All individuals (N = 16) of the CAN population were
completely homozygous for one allele at each of the 10 loci.
Plants of the SLO population had more than two alleles at multiple
loci. Follow up analysis with flow cytometry revealed that
plants of the SLO population are tetraploid (J. Modliszewski, pers.
comm.).
Genetic divergence was high in all pair-wise population
comparisons (mean pairwise FST = 0.48, online Supplementary
Table S3). Genetic variation was greater within inland populations
than within coastal populations of M. guttatus. For all 10
loci, observed heterozygosity (HO) and expected heterozygosity
(HS) were significantly greater for the inland populations than
the coastal populations, whereas genetic divergence (FST ) was
significantly greater for coastal populations than inland populations
(FSTAT: N = 28 populations, 1000 permutations, P <
0.0001 for all tests, online Supplementary Fig. S2A; see online
Supplementary Table S4 for locus-specific summary statistics).
In contrast, the level of inbreeding (FIS) did not differ (P =
0.486) between coast and inland populations (online Supplementary
Fig. S2A). The inland populations collectively harbored far
more habitat-restricted private alleles than coastal populations
(online Supplementary Fig. S2B and Table S4). Across all loci,
64.13% of the alleles found in inland populations were restricted
to the inland habitat, whereas only 17.57% of the alleles found
in coastal populations were restricted to the coast habitat (online
Supplementary Fig. S2B, N = 28 populations, 2 habitats, 10 loci,
F = 39.34, P < 0.0001).
We found strong evidence for genetic divergence between
coast and inland populations with all three analyses employed.
Analysis of molecular variation (AMOVA) detected structure
(1023 permutations, FCT = 0.0845, P < 0.001) between the coast
and inland habitats. Even so, only 8% of the genetic variation was
partitioned between habitats, whereas 39% of the variation was
among populations within habitats, and the remaining 52% was
among individuals within habitats (Table 3).
There was a striking dearth of admixture among populations
in our analyses with STRUCTURE, such that all individuals
within a given population were assigned to the same cluster
EVOLUTION SEPTEMBER 2008 2203
DAVID B. LOWRY ET AL.
Table 2. Results of REML mixed-model nested ANOVA of first two principle components, flowering time, and 13 measured morphological
traits for 132 individual plants from coast (N=7) and inland (N=5) populations. Plants were grown in a common garden greenhouse
environment so observed differences should be genetically based. Estimates for the percent of variation (%var) distributed between
habitats, among populations within habitats, and among individuals within populations (error) estimated by treating all hierarchical
levels as random effects. ∗P<0.05, ∗∗P<0.01, ∗∗∗P<0.001.
Source of variation
Habitats Populations Individuals
(Error)
Trait Coast Inland
Mean (SE) Mean (SE) F-ratio %var χ2
Var %var Var %var
comp comp
PC1 1.86 (0.13) −2.87 (0.20) 181.24∗∗∗
88.6 4.7∗
0.22 1.7 1.23 9.7
PC2 −0.14 (0.16) 0.36 (0.25) 1.58 6.1 63.6∗∗∗
1.42 57.6 0.89 36.3
Flowering time (days) 33.00 (0.57) 23.25 (0.40) 49.55∗∗∗
71.9 15.0∗∗∗
4.31 6.5 14.27 21.6
Stem thickness 1 (mm)1
6.23 (0.13) 2.99 (0.17) 63.11∗∗∗
78.2 18.5∗∗∗
0.40 5.8 1.08 15.9
Internode length (mm) 23.28 (2.10) 59.49 (4.21) 9.88∗
48.4 70.8∗∗∗
355.38 29.7 262.34 21.9
Leaf length 1 (mm)3
57.12 (1.60) 39.00 (1.86) 14.32∗∗
45.1 20.3∗∗∗
56.27 15.9 138.07 39.0
Leaf width 1 (mm)3
33.01 (0.98) 27.50 (1.36) 3.24 11.6 18.5∗∗∗
22.74 23.7 61.86 64.7
Leaf thickness (mm) 0.439 (0.01) 0.401 (0.02) 0.55 0.0 53.5∗∗∗
0.01 49.9 0.01 50.1
Corolla length (mm) 37.17 (0.59) 25.80 (0.69) 80.27∗∗∗
71.1 3.3∗
2.45 2.7 23.91 26.2
Corolla width (mm) 31.38 (0.51) 20.55 (0.62) 67.46∗∗∗
73.3 7.7∗∗
3.31 4.2 17.5 22.4
Number of flowers 27.37 (1.60) 33.40 (2.96) 0.75 0.0 24.5∗∗∗
98.53 31.8 27.33 68.2
Stem thickness 2 (mm)2
6.25 (0.14) 1.99 (0.11) 137.53∗∗∗
88.0 16.1∗∗∗
0.299 2.9 0.946 9.1
Leaf length 2 (mm)4
71.47 (2.14) 25.43 (1.83) 85.30∗∗∗
79.4 8.8∗∗
50.52 3.8 225.09 16.8
Leaf width 2 (mm)4
45.88 (0.94) 24.06 (1.14) 127.08∗∗∗
78.5 2.1 4.90 1.6 59.91 19.9
Rosette diameter (mm) 100.3 (6.62) 29.11 (2.88) 14.31∗∗
50.1 36.5∗∗∗
913.01 19.0 1488.2 31.0
Height (mm) 326.40 (14.0) 279.21 (15.8) 0.83 0.0 67.3∗∗∗
8334.27 55.9 6451.20 44.1
1
Thickness of the first internode.
2
Thickness of the second internode.
3
First true leaf.
4
Second true leaf.
regardless of K-value. Most populations were correctly assigned
(at K = 2) to the habitat from which they were collected (Fig. 3A).
However, three populations (ORO, BSR, and ANR) were misassigned
on different runs. BSR was consistently misassigned on
all runs, whereas ORO was misassigned in 68% of the runs and
ANR in 10% of the runs (50 total runs). Interestingly, ANR and
BSR were both collected from along the Eel River in Northern
California (see Discussion). At greater K values (K = 3–35),
the partitioning of populations by STRUCTURE was much more
Table 3. Results of analysis of molecular variation (AMOVA) preformed in Arlequin 3.11. The nested model included habitats
(coast/inland), populations within habitats, and individuals within populations (∗∗∗P<0.001).
Source of df Sum of Variance % of
variation squares components variation
Among coast and inland habitats 1 195.29 0.312 8.45∗∗∗
Among populations within habitats 26 1299.71 1.456 39.48∗∗∗
Within populations 898 1724.93 1.921 52.08∗∗∗
Total 925 3219.93 3.689
complicated and population assignments were much less consistent
among runs (online Supplementary Fig. S3). Out of all of the
models tested with STRUCTURE, the model with K = 29 had
the highest likelihood. In addition, we observed little evidence of
isolation by distance, except for densely sampled populations in
central California (analysis not shown, but see online Supplementary
Fig. S3).
Our analysis with POPULATION GRAPH found that the
coast and inland populations cluster as two distinct groups. There
2204 EVOLUTION SEPTEMBER 2008
ECOLOGICAL ISOLATION IN MIMULUS
Figure 3. Genetic population structure of coast and inland populations. (A) Analysis with the program STRUCTURE (Pritchard et al. 2000)
run 50 times at K = 2. Results of multiple runs combined with CLUMMP (Jakobsson and Rosenberg 2007) and visualized with DISTRUCT
(Rosenberg 2004). (B) Analysis with the program POPULATION GRAPH (Dyer and Nason 2004). There were significantly more edges within
coast (blue lines) and inland (orange lines) groups than between (pink lines) groups (P < 0.0001). The diameter of each sphere is equal
to the expected heterozygosity (HS) of the population, which was calculated in FSTAT (Goudet 2001). For both analyses: N = 14 coast
populations, 14 inland populations, 479 individuals, 10 codominant loci.
EVOLUTION SEPTEMBER 2008 2205
DAVID B. LOWRY ET AL.
were 29 edges among inland populations, 21 edges among coast
populations, but only 10 edges between coast and inland populations
(Fig. 3B). This represents highly significant genetic differentiation
between coast and inland groups (N = 28 populations,
10 loci, P < 0.0001).
Two marker loci (AAT217 and MgSTS278) are located on
linkage group 8 near two previously identified quantitative trait
loci (QTLs), which are partly responsible for the phenotypic divergence
between a pair of coastal and inland populations (Hall
et al. 2006). Because selection at those QTLs may result in divergence
of linked markers, we reran our analyses of population
structure without those loci. Similar levels of clustering were observed
using the eight remaining loci for both STRUCTURE (data
not shown) and POPULATION GRAPH (23 edges among inland
pops, 23 edges among coast pops, and 13 edges between coast
and inland pops, P < 0.0001).
RECIPROCAL TRANSPLANT EXPERIMENTS
Early season survival was high at three of the four field sites
whereas all experimental plants at coast site 2 died early in the
season (online Supplementary Fig. S1, Table 4). As a result, we restricted
our analyses to the three remaining field sites. Landslides
destroyed three of the ten blocks at coast site 1. Therefore, the
sample size at coast site 1 is lower than at the two inland sites. At
the three field sites with survivors, local plants consistently outperformed
immigrant plants (Fig. 4, Table 4). In addition, there
was very little overlap in flowering time between coast and inland
plants (Fig. 4A).
ANALYSIS OF LOCAL ADAPTATION AND HYBRID
PERFORMANCE
Selection was very strong against immigrants between habitats
in both directions of migration. At the two inland sites, only one
Table 4. Summary of results from reciprocal transplant field experiments. The percentage of plants surviving (Early Survival) and rosette
diameter (mm) in late April are listed for inland, coast, and F1 hybrid plants at the three field sites. Survival to flowering is the amount
of plants that were planted in the field as seedlings that survived to flower. Number of flowers is the number of flowers produced by
plants that survived to flower. Means ± standard errors are list with sample sizes in parentheses.
Early season traits Fitness components
Site Type
Early Rosette Survival to Number of
survival diameter flowering flowers
Mendocino Inland (LMC) 68% (100) 36.00±2.06 (68) 57% (100) 3.53±0.90 (57)
Inland site 1 Coast (SWB) 62% (99) 14.48±2.17 (61) 1% (99) 1.00 (1)
F1 Hybrid 85% (98) 41.36±1.86 (83) 45% (98) 2.755±0.90 (44)
Mendocino Inland (LMC) 49% (69) 31.56±4.54 (34) 17% (69) 2.83±5.85 (12)
Coast site 1 Coast (SWB) 90% (69) 42.77±3.36 (62) 38% (68) 10.23±3.98 (26)
F1 Hybrid 87% (68) 45.44±3.44 (59) 72% (68) 15.20±2.90 (49)
Big Creek Inland (CAN) 52% (85) 18.07±2.21 (40) 22% (85) 3.37±2.08 (19)
Inland site 2 Coast (BCB) 51% (84) 15.36±2.24 (39) 0% (84) 0.00 (0)
F1 Hybrid 49% (79) 26.24±2.27 (38) 13% (79) 7.00±1.51 (10)
coastal plant of 183 (0.5%) survived to flower and this plant only
produced one flower. In contrast, 41% of inland plants at the two
inland field sites survived to flower, and survivors produced an
average of over three flowers. At coast site 1, over twice as many
coast plants survived to flower than inland plants. In addition,
coast plants that survived to flower at coast site 1 produced ∼3.5
times as many flowers as inland plants at coast site 1 (Table 4).
Overall, the population × site interaction (Fig. 4B,C) in the reciprocal
transplant between coast site 1 and inland site 1 was
highly significant (ASTER, df = 11, z = −4.170, P < 0.0001).
Because all plants died prematurely at coast site 2, analysis of a
site × genotype interaction was not possible for experiment 2.
Therefore, we restricted our analysis to a one-way comparison of
the performance of coast and inland plants within inland site 2.
Due to the fact that none of the coast (BCB) plants survived to
flower at inland site 2, performance of inland plants (CAN) was
significantly greater than the performance of coast plants at this
site (Fig. 4D,E; ASTER, df = 3, z = 2.459, P = 0.0139).
At the inland site 1, the native inland (LMC) plants survived
to flower at ∼1.25 times the rate of F1 hybrids, and these inland
plants produced ∼1.25 times as many flowers as the F1 hybrids
(Table 4, ASTER, df = 3, z = 2.568, P = 0.0102). At inland site
2, there was no difference in fitness between hybrids and local
inland (CAN) plants (ASTER, df = 3, z = −0.10, P = 0.92).
At coast site 1, hybrids greatly outperformed native coast (SWB)
plants (ASTER, df = 3, z = 5.259, P < 0.0001) as ∼2.5 times
more hybrids survived to flower and these survivors produced
∼1.5 times more flowers (Table 4).
SALT SPRAY TOLERANCE
In the field, leaf damage (presumably due to salt spray) significantly
differed among coast, inland, and F1 hybrids at coast site
2206 EVOLUTION SEPTEMBER 2008
ECOLOGICAL ISOLATION IN MIMULUS
Figure 4. Results of reciprocal transplant experiment in Northern California. (A) Flowering time differences between coast and inland
populations was assessed through a reciprocal transplant experiment between coast site 1 and inland site 1. There was no overlap in
flowering time between inland (LMC) plants at inland site 1 (dashed line) and coast (SWB) plants at coast site 1 (black line). When placed
into artificial sympatry, there was a slight overlap between inland immigrants (LMC) at coast site 1 (gray line) with coast plants at this
site (black line). Only one coast (SWB) immigrant survived to flower at inland site 1 and it overlapped in flowering with 4% of the inland
plants at this site (dashed line). (B) Survival to flowering of coast (black), inland (gray), F1 hybrids (dashed) between sites in experiment
1. (C) Number of flowers produced by plants that survived to flower between sites in experiment 1. (D) Survival to flowering of coast
(BCB, black), inland (CAN, light gray), and F1 hybrids (dark gray) at inland site 2. (E) Number of flowers produced by plants that survived
to flower at inland site 2. All error bars denote one standard error.
1 in early spring (April 26, 2006, F2,148 = 23.08, P < 0.0001,
Fig. 5A). Leaf damage was an order of magnitude greater for
inland (LMC) plants than coast (SWB) plants. Both coast and
F1 hybrids had significantly less leaf damage than inland (LMC)
plants in the post hoc analysis (P < 0.05, Fig. 5A). Further, 36%
(13 out of 36) of the inland (LMC) plants alive in late April were
subsequently killed by leaf damage. Leaves and flowers were
severely wilted on all of the inland that survived to flower (N =
12) at coast site 1.
The greenhouse salt tolerance experiment confirmed that
coastal populations are more tolerant to salt spray than inland populations.
Inland plants accumulated salt spray damage at nearly
three times the rate of coast plants (F1,148 = 244.69, P < 0.0001,
Fig. 5C). Further, coastal plants survived salt spray treatment
about twice as long as inland plants (F1,149 = 155.65, P < 0.0001,
Fig. 5D). There was also significant variation in leaf damage
(F4,148 = 19.71, P < 0.0001) and time to mortality (F4,149 =
12.46, P < 0.0001) among populations within coast and inland
habitats. F1 hybrids between coast (SWB) and inland (LMC) populations
were significantly more tolerant than LMC, but just as
salt tolerant as SWB in a Tukey-Kramer post hoc analysis (P <
0.05, Fig. 5C,D).
INTRINSIC POSTZYGOTIC REPRODUCTIVE ISOLATION
Seeds germinated in all 20 of the interpopulation crosses, and 84
of 86 of the F1 progeny were fully viable. Only one cross of 20
resulted in inviable F1 hybrids, and none of the crosses led to
hybrid lethality. The cross leading to inviability was conducted
between the OSW (coast) and the SAM (inland) populations. The
inviability appears to be a form of hybrid necrosis (Bomblies and
Weigel 2007) as plants were dwarfed with most of the leaves
turning brown. Overall, the affected F1 family contained two
inviable hybrids and three fully viable hybrids. A cross between
viable members of this F1 family and another independent F1
family also resulted in inviable plants (3 out of 20) in the F2
generation. The inviability was more severe in the F2 generation
EVOLUTION SEPTEMBER 2008 2207
DAVID B. LOWRY ET AL.
Figure 5. Salt spray tolerance differed among coast (black), inland (light gray), and F1 hybrids (dark gray, cross between LMC and SWB).
(A) In the field, leaf damage was significantly greater for inland plants than coastal plants and F1 hybrids at coast site 1 (F2,148 = 23.08,
P < 0.0001, Tukey-Kramer comparison of means with α = 0.05). In the greenhouse: (B) Inland plants (right) had lower tolerance to 500
mM NaCl solution than coast plants (left). (C) Accumulation of leaf damage, from application of 500 mM NaCl solution, was significantly
different between coast and inland populations (F1,148 = 244.69, P < 0.0001). (D) Days until mortality was also significantly different
between coast and inland populations as well (F1,149 = 155.65, P < 0.0001). All error bars denote one standard error.
than the F1 generation, and F2 plants were severely dwarfed.
We did not observe any additional inviable hybrids in the F2
generation of other population crosses.
THE STRENGTH OF REPRODUCTIVE ISOLATING
BARRIERS
The level of reproductive isolation between coast and inland populations
was near complete, with prezygotic barriers much stronger
than postzygotic barriers in their contribution to overall reproductive
isolation. For each barrier listed in Table 5, reproductive
isolation ranged from ≤ 0 (unrestricted gene flow) to 1 (complete
reproductive isolation). Because there was no overlap in flowering
time between coast plants in coastal habitat and inland plants in
inland habitat, there is complete (RIF,A = 1.000) temporal flowering
isolation between habitats. However, it should be kept in
mind that this barrier only applies to gene flow through pollen
movement, not seed dispersal. Selection against immigrants was
near complete (RII,I = 0.999) for coastal immigrants moving into
inland habitat, and was also strong for inland immigrants moving
into coastal habitat (RII,C = 0.874). The one immigrant coast
(SWB) plant that survived to flower in inland habitat flowered during
the last 4% of the flowering period of the native inland plants.
However, because this immigrant’s only flower was opened while
native plants were still flowering we calculated that there was no
2208 EVOLUTION SEPTEMBER 2008
ECOLOGICAL ISOLATION IN MIMULUS
Table 5. The strength of reproductive isolating barriers between
coast and inland populations. Data calculated from results of reciprocal
transplant studies and controlled crossing experiments. Barriers
range from negative values (unrestricted gene flow) to 1 (complete
reproductive isolation). The first column is the strength of
the barrier reducing gene flow into coastal populations, whereas
the second column is the restriction on gene flow into inland pop-
ulations.
Strength of barrier
Isolating barrier
Coast Inland
Temporal flowering isolation 1.000 1.000
among habitats (RIFA)
Selection against immigrants (RII) 0.874 0.999
Temporal flowering isolation 0.895 0.001
in sympatry (RIFS)
Intrinsic postzygotic isolation (RIIP) 0.023 0.023
Extrinsic postzygotic isolation (RIEP) −1.801 0.233
1
This was calculated for one surviving coast plant in inland habitat.
reproductive isolation (RIFS,I = 0.000). At coast site 1, 12 inland
plants survived to flower, but only 10.5% of the total immigrant
(LMC) flowers were open at the same time as native coast (SWB)
plants (RIFS,C = 0.895). Extrinsic postzygotic isolation or ecological
selection against hybrids provided a small barrier to gene
flow into inland habitat (RIEP,I = 0.233), but was nonexistent in
coastal habitat (RIEP,C = −1.801), where hybrids outperformed
local plants. The strength of intrinsic postzygotic isolation between
coast and inland populations is very low (RIIP = 0.023).
Discussion
Our results indicate that coastal perennial and inland annual populations
of M. guttatus comprise two distinct morphologically and
molecular genetically diverged groups. Nearly complete prezygotic
isolation through a combination of geography, selection
against immigrants, and flowering time isolation likely maintains
the genetic differentiation of these coast and inland groups. In
the inland habitat, the onset of the summer drought prevents the
successful immigration of late flowering coastal perennial plants,
whereas early flowering annual life-history (Hall and Willis 2006)
and low salt tolerance of inland plants inhibit their immigration
into coastal habitat. Overall, these results are consistent with the
role of habitat-dependent natural selection in the formation of
widespread reproductively isolated species.
PATTERNS OF MORPHOLOGICAL AND MOLECULAR
GENETIC DIFFERENTIATION
Our analysis of morphological data clearly demonstrates that M.
guttatus populations derived from coast and inland habitats are
genetically distinct (Fig. 2; Table 2). This pattern is consistent
with the findings of other botanists, who have long recognized
a common suite of morphological differences between coast and
inland plant populations of a variety of species (Turresson 1922;
Stebbins 1950; Clausen 1951; Clausen and Heisey 1958; Grant
1981). Clausen (1951) argued that the consistent distinctness of
coastal populations in species such as Layia platyglossa, Potentilla
glandulosa, and Achillea borealis suggest that morphologically
distinct coastal populations should be collectively classified
as ecological races. Ecological races of plants are somewhat analogous
to host races in insects (Berlocher and Feder 2002; Dres
and Mallet 2002; Funk et al. 2002; Nosil 2007), and are defined
as a large set of populations that are restricted to a particular type
of habitat by abiotic and/or biotic factors (Clausen 1951; Clausen
and Heisey 1958).
Our analyses of the highly variable molecular genetic markers,
using STRUCTURE and POPULATION GRAPH, are also
consistent with the hypothesis that coast and inland populations
of M. guttatus constitute distinct ecological races. Our analysis
implies that geographically distant coastal populations (>1000
km apart from each other) are more closely related to each other
than they are to adjacent inland populations, which are often only
a few kilometers away. Despite the clear overall divergence between
coast and inland races, only a relatively small proportion
of the total molecular variation is partitioned between races in the
AMOVA. Interestingly, far more of the variation in morphology
(PC1), flowering time, and nine other traits (Table 2) is partitioned
between coast and inland groups than genetic variation (Table 3).
A high level of quantitative trait divergence coupled with modest
levels of genetic divergence is consistent with habitat-mediated
selection driving morphological evolution (Spitze 1993; McKay
and Latta 2002). It should be noted that our common garden experiment
was performed after only one generation in a common
environment and thus, does not properly control for maternal effects.
However, individual lines grown for multiple generations
under common growth conditions maintain morphological distinctness
between coast and inland habitats (data not shown).
Average pairwise FST for inland populations is high in both
this study (0.32) and a previous study (0.32, Awadalla and Ritland
1997). Even greater FST was observed among coastal populations
(0.55). Because FST depends inversely on within population diversity
(Nei 1973, 1987; Charlesworth et al. 1997), the elevated
among coast population FST is likely the result of consistently
low within coast population heterozygosity. It should be noted
that rare long-distance migration likely occurs among M. guttatus
populations through dispersal by water (over 4.5 km in a single
season, Levine 2001), deer (over a kilometer in a season, Vickery
1986), and birds (Lindsay 1964). Even so, restricted migration
among all population of M. guttatus may increase the partitioning
of molecular genetic variation among populations and individuals
EVOLUTION SEPTEMBER 2008 2209
DAVID B. LOWRY ET AL.
while diminishing the between group (coast vs. inland) partition
of genetic variation in the AMOVA analysis.
Although most populations were correctly assigned to coastal
or inland habitats using STRUCTURE, three of the populations
(ANR, BSR, and ORO) did not cluster with other populations
from their respective habitats. One possible explanation is that
these populations are derived from the admixture of coast and inland
populations. The two misassigned inland populations (ANR
and BSR) are located along the Eel River in Northern California.
ANR appears to be admixed based on STRUCTURE results and
BSR may have been colonized by coastal plants from the nearby
tidal estuary region of the Eel River. Further, BSR contained two
different sized morphotypes, which may indicate that it is a mixed
coast and inland population. Interestingly, the morphology of the
ANR population is much like coastal plants in that these plants
have many lateral branches and adventitious roots. More detailed
sampling along the Eel River will be necessary to determine the
evolutionary history of populations in this region. The misassignment
of the ORO (coast) population may simply be the result of
being located at the very southern end of our sampling area.
HABITAT ADAPTATION AND REPRODUCTIVE
ISOLATION BETWEEN COAST AND INLAND
ECOLOGICAL RACES
The results of this study and a previous reciprocal transplant experiment
(Hall and Willis 2006) clearly demonstrate that total
reproductive isolation between coast and inland populations of
M. guttatus is nearly complete as a result of local adaptation.
The allopatric distribution of coast and inland populations may
be a byproduct of ecological range limits of these two races and
suggests that ecogeographic isolation (as defined by Schemske
2000) may be the key barrier to gene flow. Even so, as explained
above, rare dispersal of pollen or seed probably occurs between
coast and inland populations because they are often in about as
close proximity to each other as are different populations within
habitats. Even if rare dispersal occurs, the high estimates of reproductive
isolation due to immigrant inviability and flowering
time differences will sharply limit the opportunity for gene flow.
Intrinsic postzygotic isolation between coast and inland populations
appears to be insignificant. One caveat of our analysis of
intrinsic postzygotic isolation is that we did not assess the strength
of crossing barriers, hybrid sterility, or levels of transmission ratio
distortion (but see Hall and Willis 2005). However, we were able
to successfully generate F2 hybrids in all intercrosses among F1
progeny. Therefore, the rate of hybrid sterility is likely to have
limited effect on gene flow.
Extrinsic postzygotic isolation is thought to be a common
byproduct of local adaptation of different species (Schluter 2001;
Rundle and Whitlock 2001; Nosil et al. 2005). However, heterosis
is known to offset the effect of intrinsic incompatibilities in early
generation hybrids (Rhode and Cruzan 2005), and may offset extrinsic
effects as well (Rundle and Whitlock 2001; Lowry et al.
2008). Indeed, the high levels of coast/inland F1 hybrid performance
in the field indicate that hybrids could actually facilitate
gene flow into coastal habitats, while at most it would act as a
weak barrier to gene flow into inland habitats. The elevated fitness
of F1 hybrids in coastal habitat may be a product of heterosis
combined with high F1 salt tolerance (Fig. 5). The effects of heterosis
may be mitigated in inland habitat, where flowering time
is key to fitness, because F1s flower later than inland parentals
(Fig. 2B; Table 5).
Although extrinsic postzygotic isolation is insignificant for
coast/inland F1 hybrids, extrinsic postzygotic isolation may act
on advance generation hybrids, where the effect of heterosis will
be diminished (Burke and Arnold 2001; Rundle and Whitlock
2001). Even so, the patchy distribution of M. guttatus populations
means that a large number of hybrids would be backcrosses to
local individuals. Such backcrosses would be composed of genetic
segregrants that are 50% homozygous for locally adapted alleles
and heterozygous at the rest of their loci. Backcrosses to locally
adapted populations typically perform as well as parent species
under field conditions (Burke and Arnold 2001; Johnston et al.
2001; Rundle 2002; Lexer et al. 2003a,b,c). This also appears
to be the case in a previous reciprocal transplant between coast
and inland M. guttatus populations, where backcrosses to locally
adapted populations performed just as well as locally adapted
parentals (Hall and Wills 2006). Therefore, it appears unlikely
that extrinsic postzygotic isolation plays a major role in restricting
gene flow between coast and inland populations.
Even though extrinsic postzygotic isolation may not be strong
overall, particular genomic regions may not be able to introgress
between habitats due to selection in advanced generation hybrids,
whereas neutral loci introgress more readily (Wu 2001; Turner
et al. 2005). Although it appears from our marker data that most
genomic regions show at least some divergence, alternative alleles
at loci involved in flowering time or salt tolerance may show
even more restriction between habitats. We are currently in the
process of creating near-isogenic lines with flowering time and
salt tolerance QTLs to determine if habitat-dependent selection
can restrict the spread of adaptive loci.
Although it is clear that prezygotic barriers result in essentially
complete reproductive isolation between these ecological
races, we did not calculate cumulative total isolation or proportional
contribution of each barrier to total isolation, unlike several
previous studies that have quantified reproductive isolating barriers
(Ramsey et al. 2003; Coyne and Orr 2004; Nosil et al. 2005;
Kay 2006). Such calculations, which are based on multiplicative
functions, are not appropriate in most cases because sequential
barriers are often not independent (Martin and Willis 2007). Further,
our analysis of reproductive isolation was only based on a few
2210 EVOLUTION SEPTEMBER 2008
ECOLOGICAL ISOLATION IN MIMULUS
field sites and reproductive isolation may be weaker in other geographic
locations. Such context-dependent reproductive isolation
could explain the apparently admixed populations (BSR, ANR,
ORO) in the STRUCTURE analysis. We also did not measure all
possible reproductive isolating barriers, which is important for a
comprehensive understanding reproductive isolation (Lowry et al.
2008). Finally, it is not clear for this system how much weight
should be given to ecogeographic isolation versus temporal isolation
and isolation due to immigrant inviability because the frequency
of rare dispersal events is notoriously difficult to study.
THE ORIGIN, MAINTENANCE, AND REPRODUCTIVE
ISOLATION OF ECOLOGICAL RACES
Ecological races have long been thought of as an intermediate
stage in the process of plant speciation (Clausen 1951; Clausen
and Heisey 1958; Grant 1981; Barrett 2001). Although our results
demonstrate essentially complete prezygotic isolation and suggest
that the coastal perennial and inland annual races of M. guttatus
are in fact distinct biological species, the process by which most
ecological races form, maintain their genetic distinctness, and
accumulate further reproductive isolating barriers remains poorly
understood.
Ecological races may be the product of a single evolutionary
event or races may be derived from multiple geographically disjunct
parallel evolution (speciation) events driven by repeated evolution
of the same reproductive isolation mechanisms (Schluter
and Nagel 1995; Rundle et al. 2000; Nosil et al. 2002; Rajakaruna
et al. 2003). Our molecular genetic results suggest that the coastal
populations of M. guttatus may be the result of a single evolutionary
origin because coastal populations consistently had a
lower allelic diversity than inland populations. Further, populations
throughout the coastal range appear to have a subset of
the alleles of the inland race (online Supplementary Fig. S2 and
Table S4). Of course the hypothesis of parallel origins of ecological
races is difficult to reject because gene flow among parallel
lineages can wipe out the molecular signal of such a history. Further,
the low diversity of coastal populations of M. guttatus may
also be the result of a lower effective population size due to the
narrow band of suitable habitats along the Pacific coast, and their
perennial, potentially clonal life history.
Ecological races are thought to maintain their genetic integrity
as a result of habitat-mediated natural selection (Clausen
1951; Clausen and Heisey 1958; Schemske 2000). Local adaptation
can directly reduce gene flow through selection against
immigrants between individual populations (Nosil et al. 2005;
Rundle and Nosil 2005; Nosil 2007) or through species-wide
ecogeographic reproductive isolation as a result of the evolution
of range limits of species (Mayr 1947; Clausen 1951; Schemske
2000; Ramsey et al. 2003; Husband and Sabara 2004; Kay 2006).
However, studies of other ecological races will be necessary to
draw any general conclusions about the relative importance of
different types of reproductive isolating barriers.
Although the coastal and inland races of M. guttatus appear to
show approximately complete reproductive isolation, the process
by which ecological races become good species remains unclear.
If ecological races actually are intermediates in the process of
speciation, then there must be a mechanism by which additional
reproductive isolating alleles spread between races to complete
the speciation process. Levin (1993, 1995) argues that most intrinsic
incompatibility alleles are at best mildly deleterious, such
as those derived from underdominant chromosomal rearrangements,
and will only be fixed in local populations through drift
(Lande 1979, 1985). Therefore, Levin (1993, 1995) concludes that
plant speciation must be initiated and completed in local populations.
However, recent studies suggest that genic incompatibilities
are frequently involved in intrinsic postzygotic isolation between
plant species (Fishman and Willis 2001; Sweigart et al. 2006;
Bomblies et al. 2007; Moyle 2007; Sweigart et al. 2007; Case and
Willis 2008; reviewed in Bomblies and Weigel 2007 and Lowry
et al. 2008) and may be driven by natural selection or genomic
conflict (Macnair and Christie 1983; Ting et al. 1998; Presgraves
et al. 2003; Orr et al. 2007). If incompatibilities in plants are indeed
driven by natural selection or genomic conflict, then incompatibility
alleles may readily spread between widespread ecological
races (Kane and Rieseberg 2007) and facilitate the conversion
of ecological races into good species. Future research is clearly
needed to resolve this issue.
Over a half century ago, Clausen (1951) envisioned that future
studies and comparative analysis from the local population
through ecological races to good species would facilitate a general
understanding of how ecology and geography interact to create
new species. The rapid development of modern molecular techniques
and expansion of genomic resources to many taxa make
these prospects only brighter.
ACKNOWLEDGMENTS
The authors wish to thank K. Wright for assistance with collections and
countless insightful conversations, the Duke University greenhouse staff
and L. Bukovnik for making our experiments possible, as well as K.
Merg and J. Lowry who provided wisdom, lodging, and transportation.
P. Bradford, S. Fraser, D. Real, and the UC Big Creek Reserve generously
provided use of their land for field experiments. D. Burge, K.
Dexter, D. Garfield, T. Juenger, J. Hereford, M. Hickman, M. Johnson,
W. Morris, Y.W. Lee, D. Levin, T. Mitchell-Olds, M. Rausher, C. Riginos,
D. Schemske, B.H. Song, J. Tung, G. Wray supplied numerous
useful discussions and comments. Comments from A. Bouck, A. Cooley,
R. Hopkins, J. Modliszewski, M. Noor, C. Wessinger, C. Wu, and five
anonymous reviewers greatly improved this manuscript. The University
of California Reserve System, the State Parks of Oregon and California, as
well as the US National Park System provided permission for collections.
Funding was provided by the National Science Foundation, through a
FIBR grant (DBI-0328636), a Doctoral Dissertation Improvement Grant
(DEB-0710094), and an Environmental Genomics Grant (EF-0723814).
EVOLUTION SEPTEMBER 2008 2211
DAVID B. LOWRY ET AL.
Funding was also provided by a National Institute of Health Graduate
Student Fellowship and a Duke University Travel Grant.
LITERATURE CITED
Abrams, L., and R. Ferris. 1951. Illustrated flora of the Pacific States, Vol. 3.
Stanford Univ. Press, Palo Alto, CA.
Angert, A. L. 2006. Growth and leaf physiology of monkeyflowers with
different altitude ranges. Oecologia 148:183–194.
Angert, A. L., and D. W. Schemske. 2005. The evolution of species’ distributions:
reciprocal transplants across the elevation ranges of Mimulus
cardinalis and M. lewisii. Evolution 59:1671–1684.
Awadalla, P., and K. Ritland. 1997. Microsatellite variation and evolution in
the Mimulus guttatus species complex with contrasting mating systems.
Mol. Biol. Evol. 14:1023–1034.
Barbour, M. G. 1978. Salt spray as a microenvironmental factor in the distribution
of beach plants at Point Reyes, California. Oecologia 32:213–224.
Barrett, S. C. H. 2001. The life and times of plant species: from metapopulaitons
to mutational meltdown. Evolution 55:641–646.
Berlocher, S. H., and J. L. Feder. 2002. Sympatric speciation in phytophagous
insects: moving beyond controversy? Annu. Rev. Entomol. 47:773–815.
Bomblies, K., and D. Weigel. 2007. Hybrid necrosis: autoimmunity as a
potential gene-flow barrier in plant species. Nat. Rev. Genet. 8:382–
393.
Bomblies, K., J. Lempe, P. Epple, N. Warthmann, C. Lanz, J. L. Dangl, and D.
Weigel. 2007. Autoimmune response as a mechanism for a DobzhanskyMuller-type
incompatibility syndrome in plants. PLoS Biol. 5:1962–
1972.
Boyce, S. G. 1954. The salt spray community. Ecol. Monogr. 24:29–67.
Burke, J. M., and M. L. Arnold. 2001. Genetics and the fitness of hybrids.
Annu. Rev. Genet. 35:31–52.
Campbell, D. R., and N. M. Waser. 2007. Evolutionary dynamics of an Ipomopsis
hybrid zone: confronting models with lifetime fitness data. Am.
Nat. 169:298–310.
Case, A. L., and J. H. Willis. 2008. Hybrid male sterility in Mimulus guttatus
(Phrymaceae) is associated with geographically-restricted mitochondrial
rearrangement. Evolution 62:1026–1039.
Charlesworth, B., M. Nordborg, and D. Charlesworth. 1997. The effects of local
selection, balanced polymorphism and background selection on equilibrium
patterns of genetic diversity in subdivided populations. Genet.
Res. 70:155–174.
Clausen, J. 1951. Stages in the evolution of plant species. Cornell Univ. Press,
Ithaca, NY.
Clausen, J., and W. M. Heisey. 1958. Experimental studies on the nature of
species. IV. Genetic structure of ecological races. Carnegie Institution
of Washington, Washington, DC.
Corbin, J. D., M. A. Thomsen, T. E. Dawson, and C. M. D’Antonio. 2005.
Summer water use by California coastal prairie grasses: fog, drought,
and community composition. Oecologia 145:511–521.
Coyne, J. A., and H. A. Orr. 2004. Speciation. Sinauer Associates, Inc, Sunderland,
MA.
Dres, M., and J. Mallet. 2002. Host races in plant-feeding insects and their
importance in sympatric speciation. Philos. T. Roy. Soc. B 357:471–492.
Dyer, R. J., and J. D. Nason. 2004. Population graphs: the graph theoretic
shape of genetic structure. Mol. Ecol. 13:1713–1727.
Excoffier, L., G. Laval, and S. Schneider. 2005. Arlequin v. 3.0: an integrated
software package for population genetics data analysis. Evol. Bioinform.
Online 1:47–50.
Excoffier, L., P. E. Smouse, and J. M. Quattro. 1992. Analysis of molecular
variance inferred from metric distances among DNA haplotypes:
applications to human mitochondrial DNA restriction data. Genetics
131:479–491.
Fishman, L., and J. H. Willis. 2001. Evidence for Dobzhansky-Muller incompatibilities
contributing to the sterility of hybrids between Mimulus
guttatus and M. nasutus. Evolution 55:1932–1942.
———. 2005. A novel meiotic drive locus almost completely distorts segregation
in Mimulus (monkeyflower) hybrids. Genetics 169:347–353.
Forister, M. L. 2005. Independent inheritance of preference and performance
in hybrids between host races of Mitoura butterflies (Lepidoptera : Lycaenidae).
Evolution 59:1149–1155.
Funk, D. J., K. E. Filchak, and J. L. Feder. 2002. Herbivorous insects:
model systems for the comparative study of speciation ecology. Genetica
116:251–267.
Geyer, C. J., S. Wagenius, and R. G. Shaw. 2007. Aster models for life history
analysis. Biometrika 94:415–426.
Goudet, J. 2001. FSTAT: a program to estimate and test gene diversities and fixation
indices (version 2.9.3), www2.unil.ch/popgen/softwares/fstat.htm.
Grant, V. 1981. Plant speciation. Columbia Univ. Press, New York, NY.
Hall, M. C., and J. H. Willis. 2005. Transmission ratio distortion in intraspecific
hybrids of Mimulus guttatus: implications for genomic divergence.
Genetics 170:375–386.
———. 2006. Divergent selection on flowering time contributes to local
adaptation in Mimulus guttatus populations. Evolution 60:2466–2477.
Hall, M. C., C. J. Basten, and J. H. Willis. 2006. Pleiotropic quantitative
trait loci contribute to population divergence in traits associated with
life-history variation in Mimulus guttatus. Genetics 172:1829–1844.
Hatfield, T., and D. Schluter. 1999. Ecological speciation in sticklebacks:
environment-dependent hybrid fitness. Evolution 53:866–873.
Hendry, A. P., P. Nosil, and L. H. Rieseberg. 2007. The speed of ecological
speciation. Funct. Ecol. 21:455–464.
Hitchcock, C. L., and A. Cronquist. 1973. Flora of the Pacific Northwest.
Univ. of Washington Press, Seattle, WA.
Humphreys, M. O. 1982. The genetic basis of tolerance to salt spray in populations
of Festuca rubra L. New Phytol. 91:287–296.
Husband, B. C., and H. A. Sabara. 2004. Reproductive isolation between
autotetraploids and their diploid progenitors in fireweed, Chamerion
angustifolium (Onagraceae). New Phytol. 161:703–713.
Jakobsson, M., and N. A. Rosenberg. 2007. CLUMPP: a cluster matching and
permutation program for dealing with label switching and multimodality
in analysis of population structure. Bioinformatics 23:1801–1806.
Johnston, J. A., D. J. Grise, L. A. Donovan, and M. L. Arnold. 2001.
Environment-dependent performance and fitness of Iris brevicaulis, I.
fulva (Iridaceae), and hybrids. Am. J. Bot. 88:933–938.
Kane, N. C., and L. H. Rieseberg. 2007. Selective sweeps reveal candidate
genes for adaptation to drought and salt tolerance in common sunflower,
Helianthus annuus. Genetics 175:1823–1834.
Kay, K. M. 2006. Reproductive isolation between two closely related
hummingbird-pollinated neotropical gingers. Evolution 60:538–552.
Kelly, A. J., and J. H. Willis. 1998. Polymorphic microsatellite loci in Mimulus
guttatus and related species. Mol. Ecol. Notes 7:769–774.
Lande, R. 1979. Effective deme sizes during long-term evolution estimated
from rates of chromosomal rearrangement. Evolution 33:234–251.
———. 1985. The fixation of chromosomal rearrangements in a subdivided
population with local extinction and colonization. Heredity 54:323–
332.
Levin, D. A. 1993. Local speciation in plants: the rule not the exception. Syst.
Biol. 18:197–208.
———. 1995. Metapopulations: an arena for local speciation. J. Evol. Biol.
8:635–644.
Levine, J. M. 2001. Local interactions, dispersal, and native and exotic plant
diversity along a California stream. Oikos 95:397–408.
2212 EVOLUTION SEPTEMBER 2008
ECOLOGICAL ISOLATION IN MIMULUS
Lexer, C., and M. F. Fay. 2005. Adaptation to environmental stress: a rare or
frequent driver of speciation? J. Evol. Biol. 18:893–900.
Lexer, C., M. E. Welch, J. L. Durphy, and L. H. Rieseberg. 2003a. Natural selection
for salt tolerance quantitative trait loci (QTLs) in wild sunflower
hybrids: implications for the origin of Helianthus paradoxus, a diploid
hybrid species. Mol. Ecol. 12:1225–1235.
Lexer, C., M. E. Welch, O. Raymond, and L. H. Rieseberg. 2003b. The
origin of ecological divergence in Helianthus paradoxus (Asteraceae):
selection on transgressive characters in a novel hybrid habitat. Evolution
57:1989–2000.
Lexer, C., R. A. Randell, and L. H. Rieseberg. 2003c. Experimental hybridization
as a tool for studying selection in the wild. Ecology 84:1688–1699.
Lindsay, D. W. 1964. Natural dispersal of Mimulus guttatus. Proc. Utah Acad.
Sci. Art. Lett. 41:327–341.
Linhart, Y. B., and M. C. Grant. 1996. Evolutionary significance of local
genetic differentiation in plants. Annu. Rev. Ecol. Syst. 27:237–277.
Lowry, D. B., J. L. Modliszewski, K. M. Wright, C. A. Wu, and J. H.
Willis. 2008. The strength and genetic basis of reproductive isolating
barriers in flowering plants. Philos. T. Roy. Soc. Lond. B. doi:
10.1098/rstb.2008.0064.
Macnair, M. R., and P. Christie. 1983. Reproductive isolation as a pleiotropic
effect of copper tolerance in Mimulus guttatus? Heredity 50:295–302.
Martin, N. H., and J. H. Willis. 2007. Ecological divergence associated with
mating system causes nearly complete reproductive isolation between
sympatric Mimulus species. Evolution 61:68–82.
Martin, N. H., A. C. Bouck, and M. L. Arnold. 2007. The genetic architecture
of reproductive isolation in Louisiana irises: flowering phenology.
Genetics 175:1803–1812.
Mayr, E. 1947. Ecological factors in speciation. Evolution 1:263–288.
McKay, J. K., and R. G. Latta. 2002. Adaptive population divergence: markers,
QTL and traits. Trends Ecol. Evol. 17:285–291.
McKinnon, J. S., S. Mori, B. K. Blackman, L. David, D. M. Kingsley, L.
Jamieson, J. Chou, and D. Schluter. 2004. Evidence for ecology’s role
in speciation. Nature 429:294–298.
Moyle, L. C. 2007. Comparative genetics of potential prezygotic and postzygotic
isolating barriers in a Lycopersicon species cross. J. Hered. 98:123–
135.
Nagy, E. S., and K. J. Rice. 1997. Local adaptation in two subspecies of
an annual plant: implications for migration and gene flow. Evolution
51:1079–1089.
Nakazato, T., M. Bogonovich, and L. C. Moyle. 2008. Environmental factors
predict adaptive phenotypic differentiation within and between two wild
Andean tomatoes. Evolution 62:774–792.
Nei, M. 1973. Analysis of gene diversity in subdivided populations. Proc.
Natl. Acad. Sci. USA 70:3321–3323.
———. 1987. Molecular evolutionary genetics. Columbia Univ. Press, New
York.
Nosil, P. 2007. Divergent host plant adaptation and reproductive isolation between
ecotypes of Timema cristinae walking sticks. Am. Nat. 169:151–
162.
Nosil, P., B. J. Crespi, and C. P. Sandoval. 2002. Host–plant adaptation drives
the parallel evolution of reproductive isolation. Nature 417:440–443.
Nosil, P., T. H. Vines, and D. J. Funk. 2005. Perspective: reproductive isolation
caused by natural selection against immigrants from divergent habitats.
Evolution 59:705–719.
Ogden, R., and R. S. Thorpe. 2002. Molecular evidence for ecological speciation
in tropical habitats. Proc. Natl. Acad. Sci. USA 99:13612–13615.
Orr, H. A., J. P. Masly, and N. Phadnis. 2007. Speciation in Drosophila: from
phenotypes to molecules. J. Hered. 98:103–110.
Pascarella, J. B. 2007. Mechanisms of prezygotic reproductive isolation between
two sympatric species, Gelsemium rankinii and G. sempervirens
(Gelsemiaceae), in the southeastern United States. Am. J. Bot. 94:468–
476.
Presgraves, D. C., L. Balagopalan, S. M. Abmayr, and H. A. Orr. 2003. Adaptive
evolution drives divergence of a hybrid inviability gene between two
species of Drosophila. Nature 423:715–719.
Pritchard, J. K., M. Stephens, and P. Donnelly. 2000. Inference of population
structure using multilocus genotype data. Genetics 155:945–
959.
Rajakaruna, N., B. G. Baldwin, R. Chan, A. M. Desrochers, B. A. Bohm, and
J. Whitton. 2003. Edaphic races and phylogenetic taxa in the Lasthenia
californica complex (Asteraceae: Heliantheae): an hypothesis of parallel
evolution. Mol. Ecol. 12:1675–1679.
Ramsey, J., H. D. Bradshaw, and D. W. Schemske. 2003. Components of
reproductive isolation between the monkeyflowers Mimulus lewisii and
M. cardinalis (Phrymaceae). Evolution 57:1520–1534.
Rhode, J. M., and M. B. Cruzan. 2005. Contributions of heterosis and epistasis
to hybrid fitness. Am. Nat. 166:E124–E139.
Rieseberg, L. H., and J. H. Willis. 2007. Plant speciation. Science 317:910–
914.
Rosenberg, N. A. 2004. DISTRUCT: a program for the graphical display of
population structure. Mol. Ecol. Notes 4:137–138.
Rundle, H. D. 2002. A test of ecologically dependent postmating isolation
between sympatric sticklebacks. Evolution 56:322–329.
Rundle, H. D., and P. Nosil. 2005. Ecological speciation. Ecol. Lett. 8:336–
352.
Rundle, H. D., and M. C. Whitlock. 2001. A genetic interpretation of ecologically
dependent isolation. Evolution 55:198–201.
Rundle, H. D., L. Nagel, J. W. Boughman, and D. Schluter. 2000. Natural
selection and parallel speciation in sympatric sticklebacks. Science
287:306–308.
Schemske, D. W. 2000. Understanding the origin of species. Evolution
54:1069–1073.
Schluter, D. 2001. Ecology and the origin of species. Trends Ecol. Evol.
16:372–380.
Schluter, D., and L. M. Nagel. 1995. Parallel speciation by natural selection.
Am. Nat. 146:292–301.
Spitze, K. 1993. Population structure in Daphnia obstusa quantitative genetic
and allozymic variation. Genetics 135:367–374.
Stebbins, G. L. 1950. Variation and evolution in plants. Columbia Univ. Press,
New York, NY.
Sweigart, A. L., L. Fishman, and J. H. Willis. 2006. A simple genetic incompatibility
causes hybrid male sterility in Mimulus. Genetics 172:2465–
2479.
Sweigart, A. L., A. R. Mason, and J. H. Willis. 2007. Natural variation for
a hybrid incompatibility between two species of Mimulus. Evolution
61:141–151.
Ting, C. T., S. C. Tsaur, M. L. Wu, and C. I. Wu. 1998. A rapidly evolving
homeobox at the site of a hybrid sterility gene. Science 282:1501–
1504.
Turner, T. L., M. W. Hahn, and S. V. Nuzhdin. 2005. Genomic islands of
speciation in Anopheles gambiae. PLoS Biol. 3:1572–1578.
Turresson, G. 1922. The genotypic response of the plant species to habitat.
Hereditas 3:211–350.
Via, S., A. C. Bouck, and S. Skillman. 2000. Reproductive isolation between
divergent races of pea aphids on two hosts. II. Selection against migrants
and hybrids in the parental environments. Evolution 54:1626–1637.
Vickery, R. K. 1952. A study of the genetic relationships in a sample of the
Mimulus guttatus complex. Ph.D Dissertation, Stanford Univ., Califor-
nia.
———. 1978. Case studies in the evolution of species complexes in Mimulus.
Evol. Biol. 11:405–507.
EVOLUTION SEPTEMBER 2008 2213
DAVID B. LOWRY ET AL.
———. 1986. Seed dispersal in Mimulus guttatus by wind and deer. Am.
Midl. Nat. 116:206–208.
Wang, H., D. MacArthur, S. C. Sanderson, J. H. Graham, and D. C. Freeman.
1997. Narrow hybrid zone between two subspecies of big sagebrush
(Artemisia tridentata: Asteraceae). IV. Reciprocal transplant experiments.
Evolution 51:95–102.
Weir, B., and C. Cockerham. 1984. Estimating F-statistics for the analysis of
population structure. Evolution 38:1358–1370.
Wilson, J. B., and M. T. Sykes. 1999. Is zonation on coastal sand dunes
determined primarily by sand burial or by salt spray? A test in New
Zealand dunes. Ecol. Lett. 2:233–236.
Wu, C. I. 2001. The genic view of the process of speciation. J. Evol. Biol.
14:851–865.
Wu, C. A., D. B. Lowry, A. M. Cooley, K. M. Wright, Y. W. Lee, and J. H.
Willis. 2008. Mimulus is an emerging model system for the integration
of ecological and genomic studies. Heredity 100:220–230.
Yang, C. F., R. W. Gituru, and Y. H. Guo. 2007. Reproductive isolation of
two sympatric louseworts, Pedicularis rhinanthoides and Pedicularis
longiflora (Orobanchaceae): how does the same pollinator type avoid
interspecific pollen transfer? Biol. J. Linn. Soc. 90:37–48.
Associate Editor: D. Schoen
Supporting Information
The following supporting information is available for this article:
Table S1. Primers used in this study for population genetic analysis.
Table S2. Individual eigenvector trait loadings on the first two principle component axes.
Table S3. Pair-wise FST among all populations.
Table S4. Populations genetic summary statistics for SSR and intron markers.
Figure S1. Map of location of field sites (square) and seed source populations (circle) used in the reciprocal transplant
experiments.
Figure S2. Comparison of population genetic summary statistics and habitat restricted alleles.
Figure S3. The program STRUCTURE run on K-values from 2 to 6.
Supporting information may be found in the online version of this article.
Please note: Blackwell Publishing is not responsible for the content or functionality of any supporting materials supplied
by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.
2214 EVOLUTION SEPTEMBER 2008