O R I G I NA L A RT I C L E
doi:10.1111/evo.13248
Divergence in style length and pollen size
leads to a postmating-prezygotic
reproductive barrier among populations
of Silene latifolia
Amanda N. Brothers1 and Lynda F. Delph1,2
1
Department of Biology, Indiana University, Bloomington, Indiana 47405
2
E-mail: ldelph@indiana.edu
Received November 27, 2016
Accepted March 27, 2017
A central tenet of speciation research is the need to identify reproductive isolating barriers. One approach to this line of research
is to identify the phenotypes that lead to reproductive isolation. Several studies on flowering plants have shown that differences
in style length contribute to reproductive isolation between species, leading us to consider whether style length could act as
a reproductive barrier among populations of a single species. This could occur if style length varied sufficiently and pollen size
covaried with style length. Populations of Silene latifolia exhibit variation in flower size, including style length, that is negatively
correlated with annual precipitation. We show that this divergence in style length has a genetic basis and acts as a reproductive
barrier: males from small-flowered populations produced relatively small pollen grains that were poor at fertilizing ovules when
crossed to females from large-flowered populations, leading to a significant reduction in seed production. Manipulating the
distance pollen tubes had to travel revealed that this failure was purely mechanical and not the result of other incompatibilities.
These results show that style length acts as a postmating-prezygotic reproductive barrier and indicate a potential link between
ecotypic differentiation and reproductive isolation within a species.
KEY WORDS:
Common garden, flower size, gametic isolation, postmating-prezygotic barriers, reproductive isolation.
Identifying the first reproductive barriers to arise in the speciation process is fundamental to understanding how new species are
formed (Sobel et al. 2010; Shaw and Mullen 2011), and investigating divergent populations within a species is a useful approach
to identifying these barriers (Nosil 2007; Via 2009; Jennings et al.
2014; Rose et al. 2014; Ostevik et al. 2016). Studies from both
plants and animals have demonstrated that postmating-prezygotic
reproductive barriers (gametic isolation) exist among populations
of a single species and are among the first barriers to arise (e.g.,
Gregory and Howard 1994; Brown and Eady 2001; Fricke and
Arnqvist 2004; Howard et al. 2009; Nista et al. 2015; Ostevik
et al. 2016). Nevertheless, the traits that are divergent among populations and that also contribute to a reduction in fertilization
postmating can be difficult to identify because of the inherently
internal processes of fertilization (Dobzhansky 1951; Eady 2001;
C
1532
Chang 2004; Jennings et al. 2014). Identifying the factors that
cause trait divergence is important in understanding phenotypic
evolution; however, in terms of the evolution of reproductive isolation, a key first step is to identify which divergent phenotypic
traits affect reproductive outcomes (Ramsey et al. 2003; Jolivet
and Bernasconi 2007; Sobel et al. 2010; Shaw and Mullen 2011).
One form of gametic isolation between animal species involves poor fertilization ability and can arise via a variety of
sperm-related causes. Conspecific sperm precedence has been
shown to severely limit the production of interspecific hybrids
in crosses between species (e.g., Howard et al. 1998; Chang
2004), sometimes in an asymmetric way (Cramer et al. 2016).
Intriguingly, problems with fertilization have also been shown in
intraspecific studies involving crosses between allopatric populations of Drosophila (Alipaz et al. 2001; Knowles and Markow
C 2017 The Society for the Study of Evolution.
2017 The Author(s). Evolution Evolution 71-6: 1532–1540
R E P RO D U C T I V E BA R R I E R W I T H I N A S P E C I E S
2001; Jennings et al. 2014), beetles (Brown and Eady 2001),
walking sticks (Nosil and Crespi 2006), guppies (Ludlow and
Magurran 2006), and stalk-eyed flies (Rose et al. 2014). Although
these studies do not pinpoint the specific phenotypes leading to
gametic isolation, they show that gametic isolation can act as a
reproductive barrier among populations, in line with the premise
that traits that are “features of mating pairs” are important speciation phenotypes (Kliman et al. 2000).
In plants, a recurring theme involving gametic isolation has
been shown across many different genera. Style-length differences
among species have repeatedly been shown to lead to asymmetric
crossing success, such that fertilization of long-styled species
by pollen from short-styled species is impeded (Buchholz et al.
1935; Grant 1966; Smith 1970; Levin 1978; Williams and Rouse
1988; Sorensson and Brewbaker 1994; Carney and Arnold 1997;
Diaz and Macnair 1999; Tiffin et al. 2001; Kay and Schemske
2008; Lee et al. 2008; Field et al. 2010; Montgomery et al. 2010;
Nista et al. 2015). Notably, style length and pollen size have
been found to be positively correlated at the interspecific level
(Roulston et al. 2000; Jürgens et al. 2012), congruent with the
hypothesis that this pattern might be caused by the failure of
relatively small pollen to reach the ovules of relatively longstyled species. The importance of variation in style length in
repeatedly causing postmating-prezygotic reproductive isolation
between plant species motivated us to test whether style length
and pollen could vary enough within a species to lead to similar
barriers to mating. In other words, does natural selection leading
to phenotypic divergence among populations in these traits have
pleiotropic consequences for reproductive isolation?
We used a common herbaceous, dioecious flowering plant,
Silene latifolia. Several aspects of S. latifolia made it ideally suited
to the study. First, style length is known to act as a postmatingprezygotic reproductive barrier between S. latifolia, which has
relatively long styles, and two other closely related Silene species
with shorter styles (Rahmé et al. 2009; Montgomery et al. 2010;
Nista et al. 2015). Second, while previous studies have indicated that between-population crosses with S. latifolia differ from
within-populations crosses in terms of postpollination fertilization success, the processes involved in this variation have not
been identified (Hathaway et al. 2009). Third, S. latifolia shows
extreme divergence in flower size across its native range in Europe, affording us the variation in style length needed to test
the hypothesis that style-length differences among populations
operate as a mechanical barrier to fertilization success. Lastly,
because the styles are receptive to pollen along their entire length,
we were able to vary the stylar placement of pollen. This allowed us to distinguish between the effects of style length per
se from potential additional pollen-pistil incompatibilities, such
as those that are known to cause fertilization barriers between
species (Swanson et al. 2004; Dresselhaus and Franklin-Tong
2013). We identified six geographically widespread populations
of S. latifolia–-three large-flowered populations and three smallflowered populations–-from across Europe to address the following questions. Does among-population divergence in style length
covary with pollen grain size? Where are pollen grains deposited
along the style by native pollinators? Does the among-population
divergence in style length affect fertilization ability or seed
mass?
Given that the styles of S. latifolia are receptive along their
entire length, we determined where the pollinators deposit pollen
along the style in six natural populations. The underlying premise
was that style length would act as more of a barrier if pollen were
deposited near the tips of the styles. We grew plants from the
same six study populations in a common garden (greenhouse),
and measured floral parts, including style length and pollen size.
We then performed reciprocal crosses at the base versus the tip
of the styles between all six populations followed by seed counts.
Our goal here was to determine whether the among-population
divergence in style length has a genetic basis, covaries with pollen
size, and acts as a postmating-prezygotic barrier to fertilization.
We hypothesized that if style length and pollen size affected fertilization, then when relatively long-styled females were pollinated
with males from populations with short-styles, pollinations performed at the tips of the styles would result in either failure of a
flower to set fruit or in fewer ovules fertilized (and hence fewer
seeds produced). In addition, comparing seed production from
among- versus within-population crosses performed at the base
of the style (which effectively eliminated style length as a potential
barrier), allowed us to evaluate whether other incompatibilities existed among populations that led to prezygotic isolation. Lastly,
we determined which ovules were fertilized in “mis-matched”
pollinations (long-styled females pollinated by males from shortstyled populations) by observing which ovules were developing
into seeds within the ovary.
Methods
STUDY SYSTEM AND POPULATIONS
Silene latifolia is a dioecious, short-lived perennial that occurs
naturally in disturbed habitats across Europe (Goulson and Jerrim
1997; Bernasconi et al. 2009). Six populations were chosen for
study, from the western most tip of continental Europe, eastward
into Hungary, and up into northern France (Fig. 1). Populations
were chosen to incorporate both a broad expanse of the European
range of the species, as well as long- versus short-styled populations. We catergorized three populations as being long-styled
(Budapest, Hungary; Cova Negra, Spain; Cabo da Roca, Portugal)
and three as short-styled (Lille, France; Viero do Minho, Portugal;
Zagreb, Croatia) (Fig. 1). Data on local rainfall during the growing
season (months with nights above freezing) for these populations
EVOLUTION JUNE 2017
1533
A . N . B ROT H E R S A N D L . F. D E L P H
A
LIL
= short-styled populations
= long-styled populations
24 m, 50 mm
BDA
155 m, 39 mm
VDM
365 m, 73 mm
CNS
175 m, 36 mm
ZAG
150 m, 61 mm
N
CRC
150 m, 36 mm
VDM
ZAG
BDA
CNS
LIL
B
CRC
Figure 1. European location and representative flowers from the
study populations. (A) Map showing the locations of the six popu-
lations of S. latifolia used in this study: Budapest, Hungary (BDA);
Cova Negra, Spain (CNS); Cabo da Roca, Portugal (CRC); Lille, France
(LIL); Vieira do Minho, Portugal (VDM); Zagreb, Croatia (ZAG).
Numbers given under each location are altitude (m) and mean
monthly average rainfall during the growing season (mm). (B)
Flowers from females from all six populations taken from plants
grown in a common garden. To the left are whole flowers and to
the right are flowers with their calyx and petals removed to show
the styles and ovary.
was obtained from meteoblue (http://www.meteoblue.com),
which provides month-by-month averages for the past 30 years.
FLOWER AND POLLEN-GRAIN SIZE
Seeds from several female plants from the six study populations
were collected or obtained from other researchers. To minimize
maternal effects, we grew these wild-collected seeds for one
or more generations in the greenhouse, crossing among families to avoid inbreeding (i.e., no brother–sister matings). Seeds
from five families per population were sown in seed trays for
the experiments outlined below and seedlings were transplanted
into 5”-diameter clay pots in a greenhouse. As each plant came
into flower, we measured calyx length, calyx width, petal-claw
length (the lower portion of the petal that attaches to the base of
the flower), and petal-limb length (the portion of the petal that
1534
EVOLUTION JUNE 2017
extends perpendicular to the calyx when the flower is open; see
Fig. 1) of the second, third, and fourth flower to characterize
flower size for both sexes, and averaged these for analysis (for 1–
3 plants per family per population). We also measured style length
for flowers on the females. To control for the developmental stage
of the flowers, measurements were taken on the second day after
each flower opened. These measurements were made to determine whether there was a genetic basis to flower size, as well as
to categorize the populations as being either large-flowered/longstyled versus small flowered/short-styled. A MANOVA was run
separately for each sex to test for differences among populations
in the size of the calyx (width and length) and petals (length
of the claw and limb), followed by a posteriori contrasts between
the three populations with the largest flowers and the three with
the smallest flowers. We harvested pollen from the males and
determined pollen size for 61 individuals (8–12 per population).
Undehisced anthers from open flowers were collected and placed
in vials. After the anthers dehisced, we quantified their size using an Elzone II 5390 (Micromeritics) particle counter/sizer. We
determined whether style length and pollen size differed significantly among populations using ANOVA, and whether the two
traits were correlated. We also determined whether style length
and petal-claw length (which determines the length of the the
floral tube) were correlated.
POLLEN DEPOSITION ON WILD-COLLECTED STYLES
Silene latifolia is primarily pollinated by night-flying moths in
the genus Hadena throughout its native range (Wolfe 2002; Bopp
and Gottsberger 2004; Jürgens et al. 1996; Kephart et al. 2006;
Brothers and Atwell 2014). These moths visit the flowers to collect
nectar and lay their eggs on the flowers of female plants—they act
as both pollinators and seed predators. One scenario for pollen
deposition is that Hadena deposits pollen near the tips of the
style—the portion that extends beyond the floral tube formed by
the calyx and petal claws (Fig. 1). Based on our own pollinator
observations in the field, the petals of S. latifolia provide a landing
surface for the moths while they extend their proboscis down to
the base of the floral tube to reach the nectar that accumulates at
the base of the flower or their ovipositor to reach the ovary. While
the exerted portion of the styles is likely where the majority of
the pollen is deposited, this needed to be tested given that both
their proboscis and ovipositor are exerted into the floral tube. A
minimum of 20 female flowers were collected from each of the
six study populations during the time of flowering and stored
in a 70% ethanol solution. Flowers were chosen based on slight
wilting or the presence of a moth egg, as both are indicators that
a flower has been pollinated. In the lab, a single excised style was
randomly selected from each flower and softened with 4 M NaOH
for four hours. Styles were then placed on glass microscope slides
and viewed under a dissecting scope. Using digital calipers, we
R E P RO D U C T I V E BA R R I E R W I T H I N A S P E C I E S
measured the total length of the style and the distance from where
at least 98% of pollen was deposited to the base of the style to
estimate the minimum distance most of the pollen would have
needed to grow to reach the ovary (Fig. S1). These measurements
were the dependent variable in a oneway ANOVA with population
as the response variable, followed by a Tukey–Kramer HSD test.
EFFECT OF STYLE LENGTH ON SEED PRODUCTION
To test whether style-length differences among populations affected seed production, we placed pollen either near the ovary
(base of the style) or on the portion of the style extended beyond
the floral-tube opening (tip of the style) (Fig. S1). We performed
reciprocal crosses among all six populations with both placement
(base vs tip) treatments. On each female, two flowers that opened
on the same day were randomly assigned to receive pollen from
a single male at the base or tip of the styles. Choosing flowers
in this way minimized any resource-driven differences in seed
production. The two flowers on the female were marked with a
small jeweler’s tag indicating the cross. Four flowers with dehiscent anthers from the male to be used as the pollen donor were
collected, and two male flowers per cross were used to saturate
the styles with pollen such that seed production was not pollen
limited. Base pollinations were performed by gently moving apart
the unfused petals of the floral tube to expose the base of the styles
and anthers were brushed along the basal portion of the style. The
other two flowers were used for tip pollinations, by brushing the
anthers across the portion of the styles exerted beyond the floral
tube. These two types of pollinations were performed on the five
females per population using a unique male from each of the six
populations (intrapopulation crosses were done using males from
another family to prevent inbreeding), giving a total of 12 crosses
per female, and an overall total of 360 crosses (two treatments ×
five females × six males × six populations). Any fruits that set
were allowed to mature fully, and then seed number was counted
for each fruit, and five seeds per fruit (if present) were weighed
together to determine seed mass. We recorded any crosses that
failed to set fruit, scored seed number as zero, and then replicated
the pollination to determine if this failure could be replicated (to
control for possible failure caused by damage, etc.).
Two analyses were performed on seed number resulting from
these crosses. Using the number of seeds produced by base (B)
and tip (T) pollinations for each female, we calculated a linear
reproductive-isolation metric modified from equation 4A in Sobel
and Chen (2014): RI = 1 – 2(T/T+B). With this metric complete
reproductive isolation (i.e., seeds are produced from base but not
tip pollinations) results in a value of 1, equal seed production
results in a value of 0, and seed production from tip but not base
pollinations results in a value of –1. This metric was our response
variable in a model that included the population of the female (six
categories), whether the male was from a long- or short-styled
population (two categories), and their interaction, followed by
Tukey’s tests. If style length and pollen size covary and affect
fertilization, then there should be a significant interaction between
female population and the category of the male population (shortvs long-styled). Second, to determine whether reduced seed set or
seed mass occurs for among-population versus within-population
crosses, the number of seeds produced by base pollinations, and
their mass, was compared for among- versus within-population
crosses. By using the base pollinations we eliminated the effect
of style length and by comparing crosses within each female we
eliminated the effect of ovule-number variation. If we found that
within-population crosses resulted in greater seed numbers or
heavier seeds than among-population crosses, this would suggest
that incompatibilities based on genetic or pollen-style chemical
interactions exist among populations.
LOCATION OF OVULES FERTILIZED
The ovules of S. latifolia are arranged along a central placenta,
with those at the top being closest to the bottom of the styles
and the first to be fertilized (Fig. S1). This prompted us to test
which ovules were fertilized in ‘mismatched’ crosses involving
a long-styled population (CRC) and males from the three shortstyled populations (LIL, VDM, ZAG), by viewing ovules post
fertilization. We performed tip pollinations on a female from CRC
using pollen from a male from the same population and males from
the three short-styled populations. Five days after pollination the
developing fruit were harvested and stored in 70% ethanol. The
ovary was dissected out of each fruit and photographed under a
dissecting microscope. Fertilized ovules appear swollen and white
or light brown at this stage, while unfertilized ovules are dark and
shrunken. The location of the fertilized ovules was noted, and the
number of fertilized ovules present in the photograph was counted
for each fruit and compared with a t-test.
Results
FLOWER AND POLLEN SIZE
We grew all six populations for at least two generations in a common garden (greenhouse) and found evidence of a genetic basis
for floral-size differences among the six populations (Table 1;
males: Wilk’s lambda = 0.10, F15,182.6 = 15.67, P < 0.0001; females: Wilk’s lambda = 0.05, F15,204.7 = 27.8, P < 0.0001). A
posteriori contrasts revealed that the overall size of the calyces
and petals of both males and females in the small-flower category were significantly smaller than those in the large-flower
category (males: F3,66 = 2.67, P < 0.0001; females: F3,74 =
3.67, P < 0.0001). In addition, large-flowered populations had,
on average, longer styles (F5,55 = 65.09, P < 0.0001). Petalclaw length (which determines the length of the floral tube) was
EVOLUTION JUNE 2017
1535
A . N . B ROT H E R S A N D L . F. D E L P H
Table 1.
Means (± SE) for style measures taken from naturally pollinated flowers from the six study populations.
Population
Population category
Style length
BDA
CNS
CRC
LIL
VDM
ZAG
Long
Long
Long
Short
Short
Short
19.1 ± 0.60 A
21.3 ± 0.46 A
19.3 ± 0.79 A
14.6 ± 0.70 B
14.0 ± 0.70 B
12.8 ± 0.49 B
Distance 98% of pollen had identity
to travel to reach base of style
13.8 ± 0.90 A
14.9 ± 0.84 A
12.9 ± 1.06 A
8.6 ± 0.44 B
8.5 ± 0.67 B
7.4 ± 0.54 B
See Figure 1 for the key to population identity. Population category refers to long- versus short-styled categories. Means not followed by the same letter
are significantly different.
CRC
Pollen size (µm)
CNS
the longest styles and VDM the shortest styles. Style length was
not correlated with altitude (r = –0.20, P = 0.71). However, style
length was relatively longer in dry regions (Fig. 1), such that
populations with lower rainfall during the growing season produced flowers with significantly longer styles (N = 6, r = 0.86,
P = 0.026).
BDA
ZAG
VDM
LIL
Style length (mm)
Figure 2. Means and standard errors for style length (mm) and
pollen size (µm) for the six populations, showing a significant
positive correlation between the two. Measurements were taken
from plants in the common-garden planting. Tukey’s post-hoc differences among the populations are as follows (style length in
capitals and pollen in lowercase): CRC – A, d; CNS – B, cd; BDA
– B, bc; LIL – C, a; VDM – C, ab; ZAG – C, ab). See Figure 1 for
the key to specific populations, and note that open circles represent small-flowered and filled-in circles represent large-flowered
populations.
significantly correlated with style length (N = 6, r = 0.97,
P < 0.0001). Large-flowered populations had larger pollen grains
as compared to small-flowered populations (F5, 55 = 11.68,
P < 0.0001) (Fig. 2). Style length and pollen-grain size were
positively correlated (N = 6, r = 0.82, P = 0.047).
Style length did not vary by geographic region (e.g., eastwest) as illustrated by the two closest populations (both in Portugal) showing the greatest divergence in style length—CRC had
1536
EVOLUTION JUNE 2017
POLLEN DEPOSITION IN THE WILD
We measured style length of flowers collected in the field and
identified where along the style native pollinators placed pollen.
Styles in the wild from the populations categorized as long-styled
were significantly longer than those in the short-styled populations (F5, 80 = 41.8, P < 0.0001; Table 1). In all six populations,
the majority of pollen was deposited on the “tip” portion of the
style, which we defined as the portion of the style extending
above the corolla tube (Fig. S1). The distance required for 98%
of the pollen tubes to grow to reach the base of the style in the
long-styled populations averaged almost 1.7 times longer (13.9 vs
8.2 mm) than the distance pollen tubes were required to
travel in the short-styled populations, a significant difference
(F5, 80 = 18.8, P < 0.0001; Table 1).
EFFECT OF STYLE LENGTH ON SEED PRODUCTION
For each female, we calculated a reproductive-isolation metric, RI,
based on the number of seeds produced per fruit following pollination at the base of the style by a particular male and the number
produced following pollination at the tip of the style by the same
male, to determine whether tip pollinations resulted in reproductive isolation (i.e., fewer seeds being produced). A significant interaction was found between the population of the female (six populations) and the category of the male population (large-flowered
population vs small-flowered population) for the difference in the
number of seeds produced (F5, 167 = 9.7, P < 0.0001). The RI for
females from short-styled populations was not significantly different following pollinations from males from large- versus smallflowered populations (mean RI = –0.03 and 0.02, respectively;
Fig. 3). In contrast, females from long-styled populations had an
R E P RO D U C T I V E BA R R I E R W I T H I N A S P E C I E S
A A
0.6
A
0.4
0.2
BC
C C
C C
C C C
C
LxL
Figure 3.
LxS
SxL
ZAG x Small
VDM x Small
LIL x Small
ZAG x Large
VDM x Large
LIL x Large
CRC x Small
CNS x Small
BDA x Small
CRC x Large
-0.2
CNS x Large
0
BDA x Large
Reproductive Isolation
0.8
SxS
Means (± 1 SE) of the reproductive-isolation metric,
RI = 1 – 2(T/T+B), based on the number of seeds produced for
each cross type from tip (T) versus base (B) pollinations. Positive
values indicate some degree of reproductive isolation (i.e., lower
seed production from tip versus base pollinations), whereas values
close to zero indicate no difference in seed production between
tip versus base pollinations. Crosses are grouped by type, with
females first and males second (e.g., L × L = long-styled females
crossed with large-flowered males). A two-way ANOVA including
the population of the female (six populations), the size of the
flowers for the males (large vs small), and their interaction indicated that as well as both factors having a significant effect on
seed number (female population: F5,167 = 16.65, P < 0.0001; male
flower size: F1,167 = 65.29, P < 0.0001), the interaction was also
significant (F5,167 = 9.68, P < 0.0001). The letters for each group
indicate significant differences based on Tukey’s post-hoc comparisons. See Figure 1 for the key to the populations.
order of magnitude greater mean RI when crossed with males from
the small-flowered populations compared to crosses with males
from the large-flowered populations (mean RI = 0.45 and 0.04,
respectively). The number of seeds produced following base pollinations did not differ significantly between within-population
versus among-population crosses (within-population crosses:
314 ± 25.7, among-population crosses: 279 ± 10.3; t = –1.28,
P = 0.203), nor did the mass of the five seeds weighed per fruit
(within-population crosses: 4.8 mg ± 0.33; among-population
crosses: 4.9 ± 0.13; t = 0.29, P = 0.775).
LOCATION OF OVULES FERTILIZED
We found that when the long-styled females from CRC were
pollinated at the tip with pollen from males from the same population, ovules were fertilized all the way from the top to the
bottom of the ovary (see example in Fig. 4). In contrast, when
Two ovaries are shown from a flower from a largeflowered population (CRC; Cabo da Roca, Portugal) pollinated (five
Figure 4.
days prior to dissection) at the tip of the style, where pollen is normally deposited in the wild. The ovary on the left was pollinated
with pollen from a male of the same population and the ovary on
the right was pollinated with pollen from a small-flowered male
(LIL; Lille, France). Fertilized ovules appear plump, while unfertilized ovules are dark and shrunken.
these same females were pollinated using males from the three
small-flowered populations, some of the ovules at the top of the
ovary were fertilized, but those at the base of the ovary were
not. As a consequence, the number of fertilized ovules present in
the two-dimensional photograph ( half of the total number) was
significantly greater for ovaries from the same-population pollinations compared to the small-flowered pollinations (mean ± SE =
143 ± 14.4 vs 40 ± 14.4, respectively; t = 5.1, P = 0.007).
Discussion
Research on speciation has shifted from a mainly geographical
focus (e.g., allopatry vs sympatry) to one that encourages a focus
on phenotypic divergence and how this divergence leads to the
severing of genetic connections (The Marie Curie SPECIATION
Network 2011). Here, we took advantage of among-population
divergence in style length (the equivalent of the female reproductive tract in internally fertilizing animal species) and the ability
to vary the placement of pollen along the style to investigate reproductive isolation. We determined whether variation in style
length and pollen size covaried, and whether this variation was
sufficient to lead to a mechanical barrier to fertilization between
genetically divergent populations of S. latifolia. Our focus on
style length was predicated on previous research showing that in
interspecific crosses, short-styled species from multiple genera,
including Silene, were largely unable to achieve fertilization of
EVOLUTION JUNE 2017
1537
A . N . B ROT H E R S A N D L . F. D E L P H
closely related long-styled species (see Introduction). We found
clear evidence for style length acting as a reproductive-isolating
barrier, as well as covariation in style length and pollen size that
has a genetic basis. Nevertheless, the postmating-prezygotic reproductive isolation among these populations was asymmetric,
as has been shown for such barriers among populations of other
species (e.g., Jennings et al. 2014).
As previously suggested in animal studies, the earliest reproductive barriers may be cryptic in that they arise postmating but
prior to fertilization (e.g., Ludlow and Magurran 2006; Howard
et al. 2009; Manier et al. 2013; Jennings et al. 2014). These barriers can arise in part because successful fertilization requires complex, multistep interactions between male and female gametes.
In animals these include sperm storage, sperm motility, recognition, and binding between sperm and egg, the acrosome reaction,
sperm penetration, and fusion with the egg plasma membrane
(Eady 2001; Vieira and Miller 2006). In flowering plants, interactions between pollen and pistil that lead to successful fertilization
include pollen adhesion, pollen hydration, pollen germination,
pollen-tube growth down the style to the ovules, and orientation
of the pollen to the micropyle of the ovule (Swanson et al. 2004;
Higashiyama et al. 2006; Dresselhaus and Franklin-Tong 2013;
Moyle et al. 2014; Takeuchi and Higashiyama 2016; Wang et al.
2016). In other words, mate discrimination that thwarts fertilization can occur at several points postmating. By varying the
placement of the pollen along the style, we were able to rule out
other processes and show that style length and covarying pollen
size per se are responsible for the fertilization barrier.
All components of our study support this assertion. Shortstyled females produced similar proportions of seeds regardless
of where the pollen was placed on the style (tip vs base) or which
type of population the pollen was from (long- vs short-styled). In
contrast, this was not the case for females from long-styled populations. When long-styled females were fertilized at the tips of
their styles, which we showed to be where pollen is normally deposited in the wild, they produced a lower proportion of seeds than
when the pollen was placed at the base of their styles. However,
this reduction only occurred when pollen of males from smallflowered populations was used. Given that the ability of pollen
from short-styled populations to fertilize most of the ovules in females from long-styled populations depended on where the pollen
was placed along the style, together with the uniform ability for
fertilization in short-styled females, we conclude that there is a
mechanical limitation to fertilization rather than limitation based
on genetic incompatibilities. In further support of this premise,
when pollen was added to the base of the styles, effectively eliminating the difference in style length among populations, which
population the pollen came from did not affect seed production.
Moreover, the mass of the seeds produced was not significantly
affected by the population that the male used in the cross came
1538
EVOLUTION JUNE 2017
from. Lastly, the location of the unfertilized ovules also supports
style length as mechanical barrier. We found that ovules at the
base of the ovary, that is those that pollen tubes would have had
to grow the longest to reach, were left unfertilized in long-styled
female × small-flowered male crosses.
Our results clearly show asymmetry in reproduction isolation. Asymmetry in reproductive isolating barriers is pervasive
across taxa for both animals (Martin-Coello et al. 2009; Schrader
et al. 2013; Manier et al. 2013; Cramer et al. 2016) and plants
(Sorensson et al. 1994; Tiffin et al. 2001; Turelli and Moyle 2001),
and is especially common in the early stages of reproductive isolation (Runquist et al. 2013). Given the positive correlation between
style length and pollen size we observed, together with the results
from the crossing experiments, it appears that this asymmetry may
have occurred because small pollen grains were not sufficiently
provisioned to reach all of the ovules in long-styled plants. This
hypothesis is bolstered by the finding that pollen protein content (and pollen grain size) were positively correlated with style
length across 83 genera of flowering plants (Roulston et al. 2000).
Indeed, a positive association between style length and pollen
size is widespread across species, including the family to which
S. latifolia belongs (Jürgens et al. 2012). Moreover, pollen size
within a species typically varies less than other floral traits (Cresswell 1998), so the variation we found in our common-garden
planting is notable. Together, these two patterns suggest that
pollen size is commonly under selection, which might limit its
variation in size at the within-species level. Nevertheless, variation does occur within some species and artificial selection has
been shown to result in correlated changes between pollen grain
size and style length (Sarkissian and Harder 2001). Sarkissian and
Harder (2001) point out that the correlated responses might occur
via gametic-phase disequilibrium, because of nonrandom mating
whereby seeds from plants with long styles are sired by pollen
carrying alleles for large size. A process similar to this might be
occurrring in S. latifolia.
We have shown that the mechanism that causes postmatingprezygotic isolation between closely related plant species leads
to a similar barrier within a species. While we have found style
length, which covaries with flower size, to be a barrier to reproduction, we do not yet know why genetically based differences in
flower size exists among populations (see also Delph et al. 2002);
that is what factors caused the divergence? Based on previous
research with S. latifolia and our results here, one selective factor
might be related to climate, specifically water availability during
the flowering season. We found that flower size, including style
length and pollen size, was correlated with precipitation, wherein
plants produced large flowers in arid regions and small flowers
in areas of higher rainfall. While this might seem counterintuitive
(more resources spent per flower in drier areas), it follows from the
flower size-number tradeoff that has been shown for this species
R E P RO D U C T I V E BA R R I E R W I T H I N A S P E C I E S
(Delph et al. 2004; Steven et al. 2007). Moreover, ecophysiological traits are genetically integrated with flower size: artificial
selection to increase flower size resulted in correlated responses
that resulted in less water use, including thicker leaves, longer leaf
lifespan, and lowered leaf physiology (Delph et al. 2005). In other
words, given the genetic integration of these traits, selection to
reduce water loss would indirectly lead to the evolution of fewer,
larger flowers. If style length and pollen size, which we show here
to affect crossing success between populations, diverged because
of abiotic factors, this would provide a link between ecotypic
divergence and reproductive barriers that mirrors patterns seen
among species.
AUTHOR CONTRIBUTIONS
ANB and LFD conceived of the study, performed the experiments, performed the analyses, and wrote the manuscript.
ACKNOWLEGMENTS
We thank P. Touzet and L. Weingartner for help with flower collections, I.
Saruhashi for help with flower measurements, D. Castillo, D. Jacobsen, B.
Robeson, and L. Weingartner for helpful discussions and suggestions, and
R. Hopkins and L. Moyle for comments on earlier versions. This work
was supported by funding from NSF (DEB-1405737 to) and Indiana
University.
DATA ARCHIVING
The doi for our data is 10.5061/dryad.0f8j8.
LITERATURE CITED
Alipaz, J. A., C. I. Wu, and T. L. Karr. 2001. Gametic incompatibilities between
races of Drosophila melanogaster. Proc. R Soc. Lond. B 268:789–795.
Bernasconi, G., J. Antonovics, A. Biere, D. Charlesworth, L. F. Delph, D.
Filatov, T. Giraud, M. E. Hood, G. A. B. Marais, D. McCauley, et al.
2009. Silene as a model system in ecology and evolution. Heredity
103:5–14.
Bopp, S., and G. Bottsberger. 2004. Importance of Silene latifolia ssp. alba
and S. dioica (Caryophyllaceae) as host plants of the parasitic pollinator
Hadena bicuris (Lepidoptera, Noctuideae). Oikos 105:221–228.
Brothers, A. N., and J. W. Atwell. 2014. The role of pollinator-mediated
selection in the divergence of floral traits between two closely related
plant species. Int. J. Plant Sci. 175:287–295.
Brown, D. V., and P. E. Eady. 2001. Functional incompatibility between the
fertilization systems of two allopatric populations of Callosobruchus
maculatus (Coleoptera: Bruchidae). Evolution 55:2257–2262.
Buchholz, J. T., L. F. Williams, and A. F. Blakeslee. 1935. Pollen-tube growth
of ten species of Datura in interspecific pollinations. Proc. Natl. Acad.
Sci. USA 21:651–656.
Carney, S. E., and M. L. Arnold. 1997. Differences in pollen-tube growth rate
and reproductive isolation between Louisiana irises. J. Hered. 88:545–
549.
Chang, A. S. 2004. Conspecific sperm precedence in sister species of
Drosophila with overlapping ranges. Evolution 58:781–789.
Cramer, E. R. A., M. Ålund, S. E. McFarlane, A. Johnsen, and A. Qvarnström.
2016. Females discriminate against heterospecific sperm in a natural
hybrid zone. Evolution 70:1844–1855.
Cresswell, J. E. 1998. Stabilizing selection and the structural variability of
flowers within species. Ann. Bo. 81:463–473.
Delph, L. F., J. L. Gehring, A. M. Artnz, M. Levri, and F. M. Frey. 2005.
Genetic correlations with floral display lead to sexual dimorphism in the
cost of reproduction. Am. Nat. 166:S31–S41.
Delph, L. F., J. L. Gehring, F. M. Frey, A. M. Artnz, and M. Levri. 2004.
Genetic constraints on floral evolution in a sexually dimorphic plant
revealed by artificial selection. Evolution 58:1936–1946.
Delph, L. F., F. N. Knapczyk, and D. R. Taylor. 2002. Among-population
variation and correlations in sexually dimorphic traits of Silene latifolia.
J. Evol. Biol. 15:1011–1020.
Diaz, A., and M. R. Macnair. 1999. Pollen tube competition as a mechanism
of prezygotic reproductive isolation between Mimulus nasutus and its
presumed progenitor M. guttatus. New Phytol. 144:471–478.
Dobzhansky, T. 1951. Genetics and the origin of species. 3rd ed. Columbia
Univ. Press, New York.
Dresselhaus, T., and N. Franklin-Tong. 2013. Male–female crosstalk during
pollen germination, tube growth and guidance, and double fertilization.
Mol. Plant 6:1018–1036.
Eady, P. E. 2001. Postcopulatory, prezygotic reproductive isolation. J. Zool.
Lond. 253:47–52.
Field, D. L, D. J. Ayre, R. J. Whelan, and A. G. Young. 2010. Patterns of
hybridization and asymmetrical gene flow in hybrid zones of the rare
Eucalyptus aggregata and common E. rubida. Heredity 106:841–853.
Fricke, C., and G. Arnqvist. 2004. Divergence in replicated phylogenies: the
evolution of partial post-mating prezygotic isolation in bean weevils. J.
Evol. Biol. 17:1345–1354.
Goulson, D., and K. Jerrim. 1997. Maintenance of the species boundary between Silene dioica and S. latifolia (red and while campion). Oikos
79:115–126.
Grant, V. 1966. Selective origin of incompatibility barriers in plant genus
Gilia. Am. Nat. 100:99–118.
Gregory, P. G., and D. J. Howard. 1994. A post-insemination barrier to fertilization isolates two closely related ground crickets. Evolution 48:705–
710.
Hathaway, L., S. Andersson, and H. C. Prentice. 2009. Experimental crosses
within European Silene latifolia (Caryophyllaceae): intraspecific differentiation, distance effects, and sex ratio. Botany 87:231–240.
Higashiyama, T., R. Inatsugi, S. Sakamoto, N. Sasaki, T. Mori, H. Kuroiwa,
T. Nakada, H. Nozaki, T. Kuroiwa, and A. Nakano. 2006. Species preferentiality of the pollen tube attractant derived from the synergid cell of
Torenia fournieri. Plant Physiol. 142:481–491.
Howard, D. J., S. R. Palumbi, L. Birge, and M. K. Manier. 2009. Sperm
and speciation. Pp. 367–403 in T. R. Birkhead, D. J. Hosken, and S.
Pitnick, eds. Sperm biology: An evolutionary perspective. Academic
Press, London.
Jennings, J. H., R. R. Snook, and A. Hoikkala. 2014. Reproductive isolation
among allopatric Drosophila montana populations. Evolution 68:3095–
3108.
Jolivet, C., and G. Bernasconi. 2007. Molecular and quantitative genetic differentiation in European populations of Silene latifolia (Caryophyllaceae).
Ann. Bot. 100:119–127.
Jürgens, A., T. Witt, and G. Gottsberger. 1996. Reproduction and pollination
in central European populations of Silene and Saponaria. Bot. Acta
109:316–324.
Jürgens, A., T. Witt, and G. Gottsberger. 2012. Pollen grain size variation in
Caryophylloideae: a mixed strategy for pollen deposition along styles
with long stigmatic areas? Plant Syst. Evol. 298:9–24.
Kay, K. M., and D. W. Schemske. 2008. Natural selection reinforces speciation in a radiation of neotropical rainforest plants. Evolution 62:
2628–2642.
EVOLUTION JUNE 2017
1539
A . N . B ROT H E R S A N D L . F. D E L P H
Kephart, S., R. J. Reynolds, M. T. Rutter, C. B. Fenster, and M. R. Dudash.
2006. Pollination and seed predation by by moths on Silene and allied
Caryophyllaceae: evaluating a model system to study the evolution of
mutualisms. New Phytol. 169:667–680.
Kliman, R. M., P. Andolfatto, J. A. Coyne, F. Depaulis, M. Kreitman, A. J.
Berry, J. McCarter, J. Wakeley, and J. Hey. 2000. The population genetics of the origin and divergence of the Drosophila simulans complex
species. Genetics 156:1913–1931.
Knowles, L. L., and T. A. Markow. 2001. Sexually antagonistic coevolution
of a postmating-prezygotic reproductive character in desert Drosophila.
Proc. Natl. Acad. Sci. USA 98:8692–8696.
Lee, C. B., L. E. Page, B. A. McClure, and T. P. Holtsford. 2008. Postpollination hybridization barriers in Nicotiana Section Alatae. Sex. Plant
Reprod. 21:183–195.
Levin, D. A. 1978. The origin of isolating mechanisms in flowering plants.
Evol. Biol. 11:185–317.
Ludlow, A. M., and A. E. Magurran. 2006. Gametic isolation in guppies
(Poecilia reticulata). Proc. R Soc. Lond. B 273:2477–2482.
Nosil, P. 2007. Divergent host-plant adaptation and reproductive isolation
between ecotypes of Timema cristinae. Am. Nat. 169:151–162.
Nosil, P., and B. J. Crespi. 2006. Ecological divergence promotes the evolution
of cryptic reproductive isolation. Proc. R Soc. Lond. B 273:991–997.
Manier, K. M., S. Lüpold, J. M. Belote, W. T. Starmer, K. S. Berben, O.
Ala-Honkola, W. F. Collins, and S. Pitnick. 2013. Postcopulatory sexual
selection generates speciation phenotypes in Drosophila. Curr. Biol.
23:1853–1862.
Martin-Coello, J., J. Benavent-Corai, E. R. Roldan, and M. Gomendio. 2009.
Sperm competition promotes asymmetries in reproductive barriers between closely related species. Evolution 63:613–623.
Montgomery, B. R., D. M. Soper, and L. F. Delph. 2010. Asymmetrical
conspecific seed-siring advantage between Silene latifolia and S. dioica.
Ann. Bot. 105:595–605.
Moyle, L. C., C. P. Jewell, and J. L. Kostyun. 2014. Fertile approaches to
dissecting mechansims of premating and postmating prezygotic reproduction isolation. Curr. Opin. Plant Biol. 18:16–23.
Nista, P., A. N. Brothers, and L. F. Delph. 2015. Differences in style length
confer prezygotic isolation between two dioecious species of Silene in
sympatry. Ecol. Evol. 5:2703–2711.
Nosil, P. 2007. Divergent host-plant adaptation and reproductive isolation
between ecotypes of Timema cristinae. Am. Nat. 169:151–162.
Ostevik, K. L., R. L. Andrew, S. P. Otto, and L. H. Rieseberg. 2016. Multiple reproductive barriers separate recently diverged sunflower ecotypes.
Evolution 70:2322–2335.
Rahmé, J., A. Widmer, and S. Karrenberg. 2009. Pollen competition as an
asymmetric reproductive barrier between two closely related Silene
species. J. Evol. Biol. 22:1937–1943.
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.
Rose, E. G., C. L. Brand, and G. S. Wilkinson. 2014. Rapid evolution of
asymmetric reproductive incompatibilities in stalk-eyed flies. Evolution
68:384–396.
Roulston, T. H., J. H. Cane, and S. L. Buchmann. 2000. What governs protein
content of pollen: pollinator preferences, pollen-pistil interactions, or
phylogeny? Ecol. Mono. 70:617–643.
Runquist, R. B. D., E. Chu, J. L. Iverson, J. C. Kopp, and D. A. Moeller. 2014.
Rapid evolution of reproductive isolation between incipient outcrossing
and selfing Clarkia species. Evolution 68:2885–2900.
Sarkissian, T. S., and L. D. Harder. 2001. Direct and indirect responses to
selection on pollen size in Brassica rapa L. J. Evol. Biol. 14:456–
468.
Schrader, M., R. C. Fuller, and J. Travis. 2013. Differences in offspring size
predict the direction of isolation asymmetry between populations of a
placental fish. Biol. Lett. 9:20130327.
Shaw, K. L., and S. P. Mullen. 2011. Genes versus phenotypes in the study of
speciation. Genetica 139:649–661.
Smith, E. B. 1970. Pollen competition and relatedness in Haplopappus section
Isopappus (Compositae). Am. J. Bot. 57:874–880.
Sobel, J. M., and G. F. Chen. 2014. Unification of methodsd for estimating
the strength of reproductive isolation. Evolution 68:1511–1522.
Sobel, J. M., G. F. Chen, L. R. Watt, and D. W. Schemske. 2010. The biology
of speciation. Evolution 64:295–315.
Sorensson, C. T., and J. L. Brewbaker. 1994. Interspecific compatibility among
15 Leucaena species (Leguminosae: Mimosoideae) via artificial hybridizations. Am. J. Bot. 81:240–247.
Steven, J. C., L. F. Delph, and E. D. Brodie III. 2007. Sexual dimorphism in
the quantitative-genetic architecture of floral, leaf, and allocation traits
in Silene latifolia. Evolution 61:42–57.
Swanson, R., A. F. Edlund, and D. Preuss. 2004. Species specificity in pollenpistil interactions. Annu. Rev. Genet. 38:793–818.
Takeuchi, H., and T. Higashiyama. 2016. Tip-localized receptors control
pollen tube growth and LURE sensing in Arabidopsis. Nature 531:245–
248.
The Marie Curie SPECIATION Network. 2011. What do we need to know
about speciation? Trends Ecol. Evol. 27:27–39.
Tiffin, P., M. S. Olson, and L. C. Moyle. 2001. Asymmetrical crossing barriers
in angiosperms. Proc. R Soc. Lond. B 268:861–867.
Turelli, M., and L. C. Moyle. 2007. Asymmetric postmating isolation: Darwin’s corollary to Haldane’s rule. Genetics 176:1059–1088.
Via, S. 2009. Natural selection in action during speciation. Proc. Natl. Acad.
Sci. USA 106:9939–9946.
Vieira, A., and D. J. Miller. 2006. Gametic interactions: is it species-specific?
Mole. Reprod. Devel. 73:1422–1429.
Wang, T., L. Liang, Y. Xue, P.-F. Jia, W. Chen, M.-X. Zhang, Y.-C. Wang,
H.-J. Li, and W.-C. Yang. 2016. A receptor heterometer mediates
the male perception of female attractants in plants. Nature 531:241–
244.
Williams, E. G., and J. L. Rouse. 1988. Disparate style lengths contribute to
isolation of species in Rhododendron. Aust. J. Bot. 36:183–191.
Wolfe, L. M. 2002. Why alien invaders succeed: support for the escape-fromenemy hypothesis. Am. Nat. 160:705–711.
Associate Editor: A. Sweigart
Handling Editor: P. Tiffin
Supporting Information
Additional Supporting Information may be found in the online version of this article at the publisher’s website:
Table S1. Size of floral parts (means ± SE) by population as measured in the common garden.
Figure S1. Drawing of a S. latifolia flower from a female and photo of a style with naturally deposited pollen grains.
1540
EVOLUTION JUNE 2017