Annals of Botany 105: 595 –605, 2010
doi:10.1093/aob/mcq013, available online at www.aob.oxfordjournals.org
Asymmetrical conspecific seed-siring advantage between Silene
latifolia and S. dioica
Benjamin R. Montgomery*, Deanna M. Soper and Lynda F. Delph
Department of Biology, 1001 East Third Street, Indiana University, Bloomington, IN 47405, USA
* For correspondence. E-mail [email protected]
Received: 24 September 2009 Returned for revision: 4 December 2009 Accepted: 22 December 2009 Published electronically: 10 February 2010
† Background and Aims Silene dioica and S. latifolia experience only limited introgression despite overlapping
flowering phenologies, geographical distributions, and some pollinator sharing. Conspecific pollen precedence
and other reproductive barriers operating between pollination and seed germination may limit hybridization.
This study investigates whether barriers at this stage contribute to reproductive isolation between these species
and, if so, which mechanisms are responsible.
† Methods Pollen-tube lengths for pollen of both species in styles of both species were compared. Additionally,
both species were pollinated with majority S. latifolia and majority S. dioica pollen mixes; then seed set, seed
germination rates and hybridity of the resulting seedlings were determined using species-specific molecular
markers.
† Key Results The longest pollen tubes were significantly longer for conspecific than heterospecific pollen in both
species, indicating conspecific pollen precedence. Seed set but not seed germination was lower for flowers pollinated with pure heterospecific versus pure conspecific pollen. Mixed-species pollinations resulted in disproportionately high representation of nonhybrid offspring for pollinations of S. latifolia but not S. dioica flowers.
† Conclusions The finding of conspecific pollen precedence for pollen-tube growth but not seed siring in S. dioica
flowers may be explained by variation in pollen-tube growth rates, either at different locations in the style or
between leading and trailing pollen tubes. Additionally, this study finds a barrier to hybridization operating
between pollination and seed germination against S. dioica but not S. latifolia pollen. The results are consistent
with the underlying cause of this barrier being attrition of S. dioica pollen tubes or reduced success of heterospecifically fertilized ovules, rather than time-variant mechanisms. Post-pollination, pre-germination barriers
to hybridization thus play a partial role in limiting introgression between these species.
Key words: Conspecific pollen precedence, hybridization, pollen tubes, reproductive isolation, Silene dioica,
Silene latifolia.
IN T RO DU C T IO N
Many plant species maintain distinct traits in zones of sympatry
despite being interfertile. In such instances, it is of interest to
understand what pre- and postzygotic isolating mechanisms contribute to the maintenance of species boundaries (Arnold, 1994;
Ramsey et al., 2003). Prezygotic isolating mechanisms include
differences in ecogeographic distributions (Ramsey et al.,
2003), flowering phenology and periodicity (Levin, 1978;
Arnold, 1994), reliance on different pollinators (Grant, 1949;
Meléndez-Ackerman and Campbell, 1998) and spatial structure
minimizing pollen exchange (Wesselingh and Arnold, 2000).
However, species that overlap in flowering time and share pollinators are likely to receive mixed pollen loads on their stigmas,
in which case conspecific pollen precedence may be another
important isolating mechanism (Darwin, 1859; Mangelsdorf
and Jones, 1926; Smith, 1970; Howard, 1999).
Conspecific pollen precedence occurs when conspecific
pollen achieves fertilization of ovules more successfully than
heterospecific pollen because of faster pollen germination,
faster pollen-tube elongation or an increased ability to reach
the micropyle and effect fertilization (Smith and Clarkson,
1956; Grant, 1963; Chen and Gibson, 1972; Howard, 1999).
This mechanism and its consequences closely resemble
conspecific sperm precedence, which has been detected in a
variety of animals (Howard, 1999; Dean and Nachman,
2009). Faster conspecific pollen-tube elongation has been
observed or inferred in multiple studies (Smith and
Clarkson, 1956; Ascher and Peloquin, 1968; Jaynes, 1968;
Chen and Gibson, 1972; Song et al., 2002), although other
studies have not consistently detected differences (Kearney,
1932; Rieseberg et al., 1995; Diaz and Macnair, 1999), or
have found faster heterospecific elongation in at least some
cases (Buchholz et al., 1935; Carney et al., 1994; Emms
et al., 1996; Husband et al., 2002). Similarly, several studies
with mixed pollen treatments have detected disproportionate
siring of seeds by conspecific pollen (Smith, 1968, 1970;
Carney et al., 1994; Rieseberg et al., 1995; Emms et al.,
1996; Klips, 1999; Song et al., 2002; Campbell et al., 2003),
but other studies have detected no difference (Alarcón and
Campbell, 2000), conspecific advantage for only a subset of
species (Hauser et al., 1997; Diaz and Macnair, 1999; Wolf
et al., 2001; Ramsey et al., 2003; Figueroa-Castro and
Holtsford, 2009), or a heterospecific advantage for some
species (Carney and Arnold, 1997; Wang and Cruzan, 1998).
The proportion of hybrid seeds produced after application
of different quantities of pollen mixtures can be used to
# The Author 2010. Published by Oxford University Press on behalf of the Annals of Botany Company. All rights reserved.
For Permissions, please email: [email protected]
596
Montgomery et al. — Asymmetrical conspecific seed-siring advantage in Silene
disentangle the mechanisms that cause disproportionate seed
siring (Cruzan and Barrett, 1996; Wang and Cruzan, 1998).
Higher attrition rates of heterospecific pollen tubes, faster
pollen-tube elongation, and higher abortion rates of heterospecifically fertilized ovules can all lead to conspecific pollen
siring disproportionately more seeds. However, effects of
application of different pollen quantities on disproportionate
seed siring differ depending on whether the advantage
results from mechanisms based on speed at which fertilization
is achieved (time-variant), or mechanisms unrelated to the
timing of events (time-invariant; Cruzan and Barrett 1996).
For time-variant mechanisms, such as differential pollen-tube
elongation, receipt of larger mixed pollen loads allows conspecific pollen tubes to fertilize a greater number of ovules before
heterospecific pollen tubes reach ovules, thereby resulting in
disproportionately low hybridization rates compared with
smaller pollen loads of similar composition. In contrast, for
time-invariant mechanisms, such as different frequencies of
pollen-tube attrition or different post-fertilization abortion frequencies, then a similar proportion of ovules is expected to be
fertilized by conspecific pollen regardless of the pollen load.
Finally, in contrast to other mechanisms, abortion of hybrid
embryos caused by incompatibilities may result in reduced
seed set with increased proportions of heterospecific pollen.
However, abortion of hybrid embryos because of resource
allocation to nonhybrid embryos would not necessarily lead
to a change in seed set. These predictions are summarized
in Table 1.
Post-pollination differences in seed siring ability may contribute to maintenance of the species boundary between two
Silene species, S. dioica and S. latifolia. These species are
closely related (Desfeux and Lejeune, 1996), and both are
dioecious, with homogametic females (XX) and heterogametic
(XY) males (Grant et al., 1994). They have broadly overlapping ranges across much of Europe and are interfertile
(Baker, 1947a, b; Prentice, 1978) with no reduction in seed
set for interspecific pollinations (Goulson and Jerrim, 1997).
Rahmé et al. (2009) found evidence of a post-pollination
barrier to hybridization acting against S. dioica, but not
S. latifolia pollen in interspecific crosses, but their study did
not examine the mechanisms responsible for this pattern.
Although this study focuses on post-pollination barriers to
hybridization between S. dioica and S. latifolia, reproductive
TA B L E 1. For flowers pollinated with conspecific pollen, mixed
pollen or heterospecific pollen, predictions for (A) change in
proportions of seeds sired by more successful pollen upon
increasing the total pollen application, and (B) change in seed
set with increased proportions of the more successful pollen type
applied to stigma
Mechanism
Prezygotic, time-variant
Prezygotic, time-invariant
or post-zygotic resource
reallocation
Differential post-zygotic
abortion due to inviability
(A) Proportion of seeds sired by
more successful pollen
(B) Seed set
Increased proportions
No change in proportions
No change
No change
No change in proportions
Increased
seed set
isolation also results from differences in habitat, timing of
flowering, and floral syndromes. Silene dioica tends to occur
in partly shaded environments (Baker, 1947b) whereas
S. latifolia typically occurs in unshaded habitats with drier
soils, and more disturbance, including pastures and fallow
fields (Baker, 1947b; Willmot and Moore, 1973; Prentice,
1979; Karrenberg and Favre, 2008). However, the species
grow together where their habitats meet and in intermediate
habitats (Baker, 1947b). Flowering phenology overlaps incompletely between the species, with peak flowering for S. dioica
occurring earlier in the season than for S. latifolia (Baker,
1947a, b; Bopp and Gottsberger, 2004). Additionally, the
periodicity of flowering differs, with S. latifolia flowers primarily opening and being visited at night in contrast to
S. dioica flowers, which open and are visited more diurnally
(Jurgens et al., 1996; Goulson and Jerrim, 1997; van Putten
et al., 2007). Finally, S. dioica flowers, which are pink and
tubular, are most often visited by bumblebees whereas
S. latifolia flowers, which are white and tubular, are most
often visited by moths (Baker, 1947a, b; Jurgens et al.,
1996; Goulson and Jerrim, 1997; van Putten et al., 2007).
Nonetheless, direct and indirect evidence indicates that
S. dioica and S. latifolia share pollen. In mixed plantings
pollinators have been found to move pollen analogues
between species, albeit less often than to conspecific
flowers (Goulson and Jerrim, 1997; van Putten et al., 2007).
Additionally, in contact zones, widespread introgression has
been inferred from the rarity of species-specific markers in
studies of allozymes (Goulson and Jerrim, 1997) and AFLPs
(Minder et al., 2007; Karrenberg and Favre, 2008). Early generation hybrids are rarely observed in these same studies,
suggesting a pattern of prolonged contact with occasional
hybridization events followed by assortative backcrossing
toward one parental species. Conspecific pollen precedence
is one hypothesis proposed for the rarity of early generation
hybrids (Minder et al., 2007; Karrenberg and Favre, 2008).
This study was an investigation into whether interactions
occurring between pollination and seed germination limit
hybridization between these two Silene species. This was
done by measuring if conspecific pollen tubes elongate
further than heterospecific pollen tubes in a set time period
in styles of both species and by determining whether conspecific pollen sires a disproportionate share of seeds in flowers
pollinated with different quantities of pure or mixed pollen.
M AT E R I A L S A N D M E T H O D S
Seeds for Silene dioica (L.) Clairv. (red campion) and
S. latifolia Poir. (white campion) were collected in the field
near Alençon, France, from allopatric populations of each
species. Neither field population showed morphological evidence of introgression. Plants were grown in a mix of commercial potting medium, compost and mineral soil, with
supplemental lighting, weekly fertilization, and occasional
chemical pest-control for spider mites in a greenhouse that
excluded pollinators. Male and female plants were spatially
separated.
Montgomery et al. — Asymmetrical conspecific seed-siring advantage in Silene
597
Pollen tubes
Reproductive success and hybrid status
Pollen tubes of both species were measured in styles of nine
S. dioica and nine S. latifolia plants. To select flowers to pollinate, newly opened flowers were marked the preceding day
and one treatment flower per plant was selected from among
the marked flowers. Flowers have five styles each. Two nonadjacent styles were selected per flower to ensure no crosscontamination occurred; one selected style was randomly
assigned to receive S. dioica pollen and the other S. latifolia
pollen. Pollinations were performed on the same plants on
each of three days, and plants lacking appropriately aged
flowers on a treatment day were omitted from that round of
treatments. Four pollen mixes were made per species on
each day, with three male plants randomly selected for each
pollen mix from among 14 S. dioica and ten S. latifolia
males. Pollen was collected from equal numbers of flowers
from each male and mixed on a glass slide. Silene dioica
and S. latifolia pollen mixes were paired, and each pair of
mixes was used to pollinate up to two flowers of each
species, each on a different plant. Pollen was applied to the
apical tip of the style using a new single horse-hair bristle.
Styles were labelled by noting paternal identity on the adjacent
petal lobe with a marker. Stigmas were collected at their base
and fixed in 70 % ethanol after 3 h, which preliminary studies
indicated was sufficient time for pollen tubes to elongate
through much of the style without yet reaching the ovary.
To view pollen tubes, excised styles were soaked in 4 M
NaOH for several hours to overnight, then stained overnight
in a 0.1 N K2HPO4 solution with 0.1 % aniline blue dye.
Styles were viewed and photographed at 40 power using fluorescent microscopy with a DAPI excitation filter. Multiple
photographs were required to record each style, and photographs were later reassembled by matching landmarks using
ImageJ (Rasband, 1997 – 2007). The length of the style, as
well as the lengths of the three longest pollen tubes, was
measured using the freehand tool in ImageJ, and it was
noted whether each reached the base of the style.
The average length of the two hand-pollinated styles was
determined for each flower and was compared between
species with ANOVA with individual plants nested as a
random effect within species. The lengths of the longest three
pollen tubes were averaged for each style, excluding styles
with fewer than three pollen tubes. The average lengths of conspecific pollen tubes were then compared with heterospecific
pollen tubes independently for each species using ANOVA
with random effects of individual flowers nested within plant
and plants nested within species. ANOVA was used to test for
an interaction between donor and recipient species on average
pollen-tube length, including independent effects of donor
species, recipient species, and their interaction, with random
effects of individual flowers nested within plants and plants
nested within species. Finally, to determine whether the difference in conspecific and heterospecific pollen-tube lengths differed between species, the log-response ratio of average
conspecific versus heterospecific length was calculated for
each flower, and compared between species with ANOVA,
including a random effect of plant nested within species. The
log-response ratio was selected for this comparison because of
its favourable statistical properties (Hedges et al., 1999).
To investigate the effects of pollen competition on hybridization rates between S. dioica and S. latifolia, hand-pollinations
were performed with varied pollen composition and density on
flowers of greenhouse-grown individuals. For each species, five
female plants with abundant flowers were selected. For pollen
donors, three male S. dioica and four male S. latifolia plants
were selected. Pollen from flowers of all three S. dioica donors
were always included together. For S. latifolia donors, pollen
from three plants was also included in each mix, but which
donors were included was varied so as to exclude male siblings
of the female being pollinated.
Pollinations were performed with four different compositions at two quantities. Pollen mixes were made to the following four ratios by mass (S. dioica:S. latifolia): 0 : 1; 1 : 3;
3 : 1; 1 : 0. The 1 : 3 and 3 : 1 ratios were chosen with the
goal of allowing each species to be well represented, representing a situation in which an indiscriminate pollinator delivers
pollen of both species. Each species was made more abundant
in one mix because pollen relative abundance may affect
hybridization rates (Carney et al., 1994), though it does not
necessarily do so (Rieseberg et al. 1995; Alarcón and
Campbell, 2000). To determine the relative abundance of
pollen grains in mixed-pollen treatments, a mix of pollen
from several flowers was collected separately and weighed
for each of seven S. latifolia and six S. dioica males, including
most individuals whose pollen was included in pollen mixes.
Subsequently, to determine pollen density, pollen counts and
size distributions were determined for each sample of known
mass with a particle-size analyser (Micromeritics Elzone 2,
Norcross, GA, USA). Pollinations were performed in two
replicates per plant with each set corresponding to the eight
combinations of composition (four treatments) and quantity
(two treatments), with each set initiated on different days.
Two flowers for each pollen composition by quantity combination were pollinated on each female plant, and five
females were treated per species, for a total of 160 flowers
(four compositions two quantities two replicates five
plants two species). To ensure that each pollen mixture
was applied to the same numbers of flowers of each species,
S. dioica and S. latifolia females were paired, and initially
the same pollen mix was used for each member of the pair.
To create pollen mixes, pollen was collected into separate
microcentrifuge tubes for each species from an equal number
of flowers from each of three donor plants. The pure pollen
loads were subsequently weighed within 10 mg on a
Sartorius micro scale. Pollen was then combined as required
and mixed on a glass slide with a needle. Pollen mixes were
stored on slides in the shade in plastic containers containing
moistened filter paper to maintain humidity. Pollen was
initially collected in mid-morning (0900– 1030 h), and pollinations were completed by mid-afternoon (1430 h).
Treatments within a replicate were randomly assigned
among open flowers with elongated styles, indicating that the
flowers had been open for at least 1 d. If at least eight
flowers were available simultaneously, all treatments within
a replicate were performed on the same day. Otherwise, treatments were performed over multiple days with new pollen
mixes from the same males on each day. Treatments were
598
Montgomery et al. — Asymmetrical conspecific seed-siring advantage in Silene
repeated for aborted flowers, but the fruit set of replacement
flowers was excluded when analysing fruit set.
To perform pollinations, pollen mixes were applied to each
stigma within a flower on the tip of a new single horse-hair
bristle. Low quantity treatments were intended to be low
enough that pollen competition would be weak, and high quantity treatments were intended to supply excess pollen, such that
pollen competition would be important. For low-quantity treatments, the bristle’s tip was tapped perpendicularly into the
pollen on the slide’s surface. For high-quantity treatments,
1 mm of the bristle’s tip was pulled along the slide, parallel
to the surface. Then, regardless of treatment, the bristle was
rubbed against the apical end of the stigma, which in both
species extends along the inner side of style over much of
its length. To test whether these treatments succeeded in delivering different pollen loads, one of the five styles was collected from most flowers (n ¼ 145) into 70 % ethyl alcohol 1
d later. Preliminary experiments demonstrated that pollen
tubes reached the base of the style within several hours, so
removal of one style after this longer time period is unlikely
to affect fertilization. Pollen grains adhering to each style
were subsequently stained with Alexander’s stain then
counted, and dislodged pollen was counted by centrifuging
the collection tubes then mounting the collected pollen in
basic fuchsin gel (Kearns and Inouye, 1993). Similarities in
pollen morphology and overlapping size distributions
between species precluded identifying pollen to species.
Fruiting status was monitored and fruits were collected at
maturity. Silene dioica retain seeds in the upright capsule following maturation, so fruits were collected upon opening.
Silene latifolia fruits were not always upright at maturation
so, prior to opening, fruits were partially enclosed in waxed
bags. Fruits were stored in coin envelopes at room temperature.
To determine seed germination rates and hybrid status, a
subset of seeds were planted in several rounds in the greenhouse. For fruits from the first replicate of pollinations, six
seeds were planted in the initial round and four seeds were
planted in the next round. For the second replicate, four
seeds were planted in each of two rounds. Within each replicate, combined data from the two rounds of plantings were
used to estimate seed germination rates. Additional seeds
were later planted for those fruits whose initial plantings
yielded less than six seedlings from which DNA could be
extracted, but these additional plantings were excluded from
estimations of seed germination rates because not all treatments were equally represented. Seeds were planted into a sterilized mix of commercial potting medium, compost and
mineral soil. Germination, as indicated by emergence of cotyledons, was monitored and final germination was tallied
approx. 6 weeks after planting.
DNA extractions were performed on true leaves of seedlings
using Qiagen DNeasy Plant Mini kits. Markers used to determine
paternity depended on the sex of the seedling and the species of
the mother. To identify the sex of seedlings, PCR amplification
for the Y-linked SlAP3Y MADS box gene was performed
using the forward primer 50 -GGCATGGAGATCTCCTCA
TGGATC-30 and reverse primer 50 -TATATTCGAGACAAC
ATGGCCTGG-30 (Matsunaga et al., 2003). Preliminary amplifications confirmed that these primers amplified a product in males
but not in females of both species. For crosses with S. dioica
mothers, DNA from female seedlings was amplified with PCR
for the MADS box gene SlAP3A, using primers 50 -GGC
ATGGAGATCTCCTCATGGATC-30 and 50 -ATACTGGAGA
TAACACAGCCTAGT-30 . Preliminary PCR indicated that
these primers amplified similarly sized products in parental
plants of both sexes of S. latifolia but not S. dioica. Although
Matsunaga et al. (2003) report SlAP3A to be autosomal, the
marker in this study was present in female but not male F1
hybrids resulting from female S. dioica by male S. latifolia
crosses, indicating X-linkage. Consequently, a different marker
was used for male seedlings involving S. dioica seed parents.
Species paternity for these crosses was determined by amplification with PCR for the Y-linked SCAR marker SCQ14 using the
primers 50 -GGACGCTTCATGACCCAT-30 and 50 -GGACGC
TTCAGCGGGCGG-30 . This marker amplifies a Y-linked band
in S. latifolia but not S. dioica males (Zhang et al., 1998).
For crosses with S. latifolia mothers, DNA from female
seedlings was amplified with PCR for SlAP3A’s homologue
in S. dioica, SdAP3A, using the same forward primer as for
SlAP3A and the reverse primer 50 -GGTCGCAAACCAC
TAGTTTATACTC-30 (Matsunaga et al., 2003). This primer
combination was found to amplify a band in parental plants
of both sexes for S. dioica but not S. latifolia. As for
SlAP3A, X-linkage of SdAP3A was inferred from successful
amplification of the marker for F1 female but not male
hybrids with S. dioica seed parents. DNA from male seedlings
was amplified with PCR for the MADS box gene SdAP3Y,
using the forward primer 50 -CGAACAGAGGAACTACGGC
GG-30 and the reverse primer 50 -GGCCATCACGAAC
TAATCACA-30 . Preliminary PCRs indicated that SdAP3Y
amplified a band in males of S. dioica but not S. latifolia.
PCR was performed with Taq DNA polymerase from
Bioline or with Promega Green Master Mix. For reactions
with Bioline Taq polymerase, the following conditions were
used: 1.0 mL 10 KCl buffer (500 nM KCl, 100 mM Tris –
HCl pH 8.8, 15 mM MgCl2, 1 % Triton 100), 0.4 mL of
dNTP (2.5 mM each of dATP, dCTP, dGTP and dTTP),
0.2 mL each of forward and reverse primer (10 mM), 0.52 mL
MgCl2 (15 mM), 0.08 mL Taq DNA polymerase (5 U mL21),
1.0 mL of DNA and 6.6 mL of water. For reactions with
Promega Green Master Mix, the following conditions were
used: 5 mL 2 master mix (Taq DNA polymerase, 400 mL
each of dATP, dCTP, dGTP and dTTP, 3 mM MgCl2),
3.0 mL water, and the same volumes of primers, MgCl2, and
DNA as for reactions with Bioline Taq polymerase. Cycling
parameters were one cycle of 94 8C for 1 min; 30 cycles of
94 8C for 1 min; 63 8C for 1 min; and 72 8C for 2 min; and
one cycle of 72 8C for 5 min. PCR products were run on 1 %
agarose gels and stained with ethidium bromide.
PCR runs included seedlings from pure heterospecific and
pure conspecific pollinations, as well as seedlings from
mixed pollinations, in order to act as positive and negative
controls and allow assessment of error rates in determining
hybrid status. In total, 852 seedlings were genotyped.
Data analysis
To determine whether the numerical ratio of pollen in pollen
mixes of the two species differed from the ratio of pollen
masses, pollen diameter and density (counts per milligram)
Montgomery et al. — Asymmetrical conspecific seed-siring advantage in Silene
from anther-collected pollen were compared between species
with two-tailed t-tests. To determine whether more pollen
was in fact received in the high-quantity treatment, stigma
pollen counts were square-root transformed to improve normality, then tested with ANOVA, including a random effect
of plant within species, pollen quantity, and an interaction
between species and pollen quantity. One positive outlier
was excluded from the large quantity treatment, which
served to reduce the difference between treatments, thus providing a conservative estimate of differences.
To determine if pollination with heterospecific pollen
decreased reproductive success relative to conspecific pollen,
fruit and seed set and seed germination rates were analysed
across treatments. Fruit set, excluding replacement flowers
for initial unsuccessful pollinations, was analysed in S-Plus
7.0 using generalized estimating equations (GEE) with an
independent correlation structure to account for correlated
responses among flowers within a plant, and a binomial distribution. The statistical model included independent effects of
species, proportion of conspecific pollen, pollen quantity
(large or small) and interactions between species and pollen
type and between proportion of conspecific pollen and
pollen quantity (other interactions could not be fit as a result
of a lack of variation within or between some treatment combinations). Seed set, square-root transformed, was analysed
with ANOVA including a random effect of plant within
species, as well as pollen type, proportion of conspecific
pollen, and interactions among all three variables. Seed
germination was compared between pure heterospecific and
conspecific pollinations with a GEE with a binomial distribution. The numbers of germinated and ungerminated seeds
per flower were included as a two-vector response variable,
and maternal species as well as pollen type (heterospecific or
conspecific) were included as independent variables with an
interaction term.
Following Hauser et al. (1997), a statistical model was constructed to investigate whether conspecific pollen was disproportionately successful in siring seeds, after accounting
for differences in seed germination rates and seedlingidentification success. The data for the model included
counts of the number of seedlings identified as hybrid or conspecific for each treated flower. For pollen mixes, the relative
abundance of pollen from each species was estimated based on
the relative mass of each species’ pollen and the different
pollen densities.
The expected proportion of hybrid seeds was calculated as
599
where gi.hyb and gi.con are the germination rates of hybrid and
conspecific seeds, respectively, for maternal species i. Finally,
to account for the probability of misidentifying hybrid or conspecific seedlings, the proportions of hybrid seedlings
expected to be observed were calculated as
rijk ¼ qijk ð1 di:hyb Þ þ ð1 qijk Þdi:con
ð3Þ
where di.hyp and di.con are the probabilities of misidentifying
the paternity of a hybrid or conspecific seedling, respectively,
of maternal species i.
Consequently, the probability of observing nijk hybrid seedlings out of Nijk seedlings scored in total for pollen mix j
applied at quantity k to a flower l of species i is the binomial
probability with parameter rijk. This results in a likelihood
function with a logarithmic transformation of
ln L ¼
2 X
4 X
2 X
L
X
N n
Nijk nijk ln
1 rijk ð ijk ijk Þ
rijk
nijk
i¼1 j¼1
k¼1 l¼1
ð4Þ
The parameters for seed germination rates, gi.hyb and gi.con,
and for seedling misidentification, di.hyp and di.con were estimated from the pure heterospecific and pure conspecific treatments for each maternal species. The parameter sijk was
estimated by maximum likelihood estimation with the Solver
Add-in of Microsoft Excel version 11.5. Because the parameter sijk was estimated after accounting for differences in
seed germination and misidentification rates, as determined
by pure conspecific and heterospecific crosses, resulting differences in the value of sijk reflect differences in pre-germination
processes. The differences in sijk between maternal species,
between majority S. dioica and majority S. latifolia mixes,
and between high and low pollen quantities were tested for significance with the likelihood ratio test: D ¼ – 2 (loge Lsimplified
– loge Lsaturated), with Lsaturated referring to the full model and
Lsimplified referring to models in which sijk was variously constrained to not differ between pollen quantities, pollen compositions, or maternal species. D approximately follows a
chi-square distribution with degrees of freedom equal to the
difference in degrees of freedom between the models being
compared (Hosmer et al., 1989).
R E S ULT S
zijk ¼ sijk pij ð1 pij þ sijk pij Þ1 i ¼ 1; 2; j ¼ 1; . . . ; 4; k ¼ 1; 2
ð1Þ
where pij is the proportion of heterospecific pollen in pollen
mix j being applied to a flower of species i, and sijk is the relative competitive ability of heterospecific pollen when applied
in quantity k (high or low dose), with the competitive ability
of conspecific pollen defined to equal one.
To account for different seed germination rates, the expected
proportion of hybrid seedlings was calculated as
qijk ¼ zijk gi:hyb ½ð1 zijk Þgi:con þ zijk gi:hyb 1
ð2Þ
Pollen tubes
Average style length was significantly longer for S. latifolia
than S. dioica (F1,16 ¼ 80.30, P , 0.001), with averages
(+ s.e.) of 9.0 + 0.2 mm for S. dioica and 15.8 + 0.5 mm
for S. latifolia. The average length of the longest three
pollen tubes was greater for conspecific than heterospecific
pollen tubes for both S. dioica (F1,14 ¼ 12.71, P ¼ 0.003)
and S. latifolia (F1,14 ¼ 8.29, P ¼ 0.017; Fig. 1). In the analysis including both species, pollen tubes were significantly
longer in S. latifolia than S. dioica styles, and there was a significant interaction between the species identity of the pollen
and style (Table 2). The interaction is attributable to the
Montgomery et al. — Asymmetrical conspecific seed-siring advantage in Silene
S. dioica
S. latifolia
8
6
4
2
0
S. dioica
S. latifolia
Maternal species
F I G . 1. Length of S. dioica and S. latifolia pollen tubes in styles of S. dioica
and S. latifolia. Error bars + 1 s.e.
TA B L E 2. ANOVA testing effects of (A) flower species on style
length and (B) flower species, pollen species and their interaction
on pollen-tube lengths
Source
(A) Style length
Flower species
Plant (flower species)
Error
(B) Pollen-tube lengths
Flower species
Plant (flower species)
Pollen species
Flower species pollen species
Error
d.f.
SS
1
16
21
454.1
90.48
10.11
1
16
1
1
28
513950
1007312
3644.7
382125
560017
F
P
80.30
,0.001
8.164
0.182
19.106
0.01
for S. dioica and 5.8 + 0.3 pollen grains 105 mg21 for
S. latifolia. Based on this, pollen mixes with 25 % and 75 %
S. dioica pollen by mass correspond to 31.8 % and 80.7 %
by quantity, respectively. Counts of pollen applied to styles
differed significantly between high and low pollen quantities
(quantity term from ANOVA: F1,131 ¼ 323, P 0.001), and
neither the maternal species nor the interaction was significant
(both P 0.10). Single styles in the high and low treatments
received an average (+s.e.) of 344 (+14) and 95 (+ 5)
pollen grains, respectively.
Among flowers pollinated with pure heterospecific or pure
conspecific pollen, there was a trend toward lower fruit set
for flowers of S. latifolia than S. dioica, and fruit set did not
vary significantly with pollen type, pollen quantity, or any
measured interactions (Fig. 2A and Table 3). Seed set was significantly higher for large than small pollen quantities and
tended to increase with increasing proportions of conspecific
pollen, but did not significantly differ between species or
with any interactions (Fig. 2B and Table 4).
Germination rates were significantly higher for hybrid than
conspecific seeds over both maternal species; however, germination rates did not vary significantly among maternal species,
and there was not a significant interaction between seed hybridity and maternal species (Tables 5 and 6). Identification of seedlings resulting from pollinations with only conspecific or
heterospecific pollen revealed error rates in assigning species
paternity of between 2 –6 %, depending on maternal and
A 1·0
0·8
0.67
, 0.001
length of S. latifolia pollen tubes responding more strongly
than that of S. dioica to the species identity of the style
(Fig. 1). For S. dioica, pollen tubes reached similar lengths
regardless of the style’s identity, but S. latifolia pollen tubes
were longer in S. latifolia than S. dioica styles. The relative
length of conspecific versus heterospecific pollen tubes did
not differ significantly between species (F1,16 ¼ 1.11, P ¼
0.31). However, the longest pollen tube was not always conspecific in either species. In S. dioica styles, the longest
pollen tube was not conspecific more often than expected by
chance (eight of 15 trials, binomial test, P ¼ 1), whereas in
S. latifolia the longest pollen tube was conspecific more
often than expected by chance (13 of 15 trials, binomial test,
P ¼ 0.007).
Reproductive success and hybrid status
Average pollen diameter was significantly smaller for
S. dioica than S. latifolia pollen (two-tailed t-test, t12 ¼ 9.95,
P , , 0.01), with an average diameter (+ s.e.) of 35.7 +
0.4 mm for S. dioica and 41.8 + 0.5 mm for S. latifolia.
There were significantly more pollen grains per milligram
for S. dioica than S. latifolia pollen (t12 ¼ 2.8, P ¼ 0.02),
with an average of 8.1 + 0.7 pollen grains 105 mg21
Fruit set
Pollen tube length (mm)
10
0·6
0·4
0·2
0·0
B 80
60
Seeds
600
40
20
0
0
20
40
60
80
100
Heterospecific pollen (%)
F I G . 2. (A) Fruit set and (B) seed set for pollinations with pure heterospecific
or conspecific pollen, or pollen mixes composed of majority S. dioica or
S. latifolia pollen, applied to flowers of S. latifolia (squares) or S. dioica
(circles), with pollen applied in large quantities (open symbols) or small quantities (closed symbols); error bars + 1 s.e. Curve fits shown separately for large
and small pollen quantities irrespective of species.
Montgomery et al. — Asymmetrical conspecific seed-siring advantage in Silene
601
TA B L E 3. Regression coefficients of GEE with binomial
distribution testing effects of species, as well as proportion of
heterospecific pollen, pollen quantity (high or low) and their
interaction on fruit set
TA B L E 6. Binomial generalized linear model of effects of
maternal species, hybrid status (pure species or hybrid) and
their interaction on germination rates
Source
Source
Estimate
s.e.
Z
P
2.365
– 1.584
0.401
– 0.316
– 0.522
0.927
0.970
0.939
0.828
1.102
2.55
– 1.63
0.43
– 0.38
– 0.47
0.01
0.10
0.67
0.70
0.64
0.707
0.782
0.9
0.37
Intercept
Species
Proportion of heterospecific pollen
Quantity
Species proportion of heterospecific
pollen
Proportion of heterospecific
pollen pollen quantity
Intercept
Species
Hybrid status
Species hybrid
status
Num.
d.f.
Dev.
exp.
Den.
d.f.
Resid.
dev.
P
1
1
1
1.12
10.12
0.24
79
78
77
76
166.69
165.57
155.45
155.21
0.29
0.001
0.62
TA B L E 4. ANOVA of effects of maternal species, proportion of
heterospecific pollen, pollen quantity and interactions on
seed set
Source
Species
Plant (species)
Proportion of heterospecific pollen
Pollen quantity
Species proportion of heterospecific pollen
Species pollen quantity
Proportion of heterospecific pollen pollen
quantity
Species proportion of heterospecific
pollen pollen quantity
Residuals
d.f.
SS
F
P
1
8
1
1
1
1
1
12.60
122.10
20.64
35.48
1.37
4.00
0.64
0.826
0.39
3.535
6.074
0.235
0.685
0.110
0.06
0.02
0.63
0.41
0.74
1
13.11
2.245
0.14
64
373.82
TA B L E 5. Parameter estimates (means) for pollen competition
model, including seed germination rates (g), misidentification
rates (d ) for both maternal species and for conspecific and
heterospecific pollen, and relative competitive ability (s) of
heterospecific pollen on maternal species indicated for majority
S. latifolia or S. dioica pollen mixes and large or small pollen
quantities
Variable
Seed germination rate (%) (s.e. þ/ –)†
Conspecific
Heterospecific
Misidentification rate (%)
Conspecific
Heterospecific
Heterospecific pollen fitness (s)
Overall‡
Majority S. latifolia mix
Large
Small
Majority S. dioica mix
Large
Small
S. latifolia
S. dioica
78.4 (4.3/5.1)
89.2 (3.6/5.2)
75.7 (3.4/3.8)
85.0 (3.5/4.4)
2.7
6.0
1.7
6.0
0.22**
1.26 n.s.
0.28
0.28
1.15
0.71
0.26
0.14
2.41
1.19
†
Germination means and standard errors back-transformed from binomial
regression with logit link.
‡
Significance levels: n.s., result does not significantly differ from one;
** P , 0.001.
Proportion hybrid seeds
1·0
0·8
0·6
0·4
0·2
0·0
0·0
0·2
0·4
0·6
0·8
1·0
Proportion heterospecific pollen
F I G . 3. Proportion of hybrid seeds observed for pollinations with pure heterospecific or conspecific pollen, or pollen mixes composed of majority S. dioica
or S. latifolia pollen, applied to flowers of S. latifolia (squares) or S. dioica
(circles), with pollen applied in large quantities (open symbols) or small quantities (closed symbols). Line represents best fit for observed proportion of
hybrid seeds for treatments with pure heterospecific or conspecific pollen
only. The best-fit line was combined between species, and the slight deviation
from the origin and 1 : 1 slope results from the misidentification rate of pure
conspecific and heterospecific seedlings. For mixed pollen loads, deviations
from the best-fit line indicate an excess or lack of hybrids; error bars + 1 s.e.
paternal species (Table 5). Error was largely attributable to PCR
failure, as indicated by subsequent successful amplification of a
relevant hybrid marker in most reamplifications of putative
hybrids that initially failed to amplify either marker (because
only putative hybrids were retested, results were not modified
to reflect retrial results). Because the estimates of error were
incorporated in calculating relative competitive ability, the estimates of relative competitive ability should not be affected by
the error rate.
In mixed pollinations, S. latifolia generally sired seeds on
S. dioica flowers in proportion to its pollen’s relative abundance,
while S. dioica sired disproportionately few seeds relative to its
pollen’s relative abundance on S. latifolia flowers (Fig. 3). The
log-transformed maximum likelihood function was evaluated
for several models to determine whether competitive ability differed between quantities and between species. The model fit was
significantly improved by allowing the relative competitive
ability, s, of S. dioica pollen on S. latifolia flowers to differ
from 1.0, a value indicative of no competitive difference (x2 ¼
56.3, P , 0.001). Furthermore, the model fit was significantly
better when s was allowed to vary between maternal species compared with a model in which s was constrained to be the same
for both maternal species (x2 ¼ 43.8, P , 0.001). These results
602
Montgomery et al. — Asymmetrical conspecific seed-siring advantage in Silene
indicate, respectively, that S. dioica pollen is significantly
outperformed by S. latifolia pollen in siring seeds on
S. latifolia and that S. dioica pollen is significantly less competitive at siring seeds on S. latifolia than S. latifolia is on S. dioica
(Table 5). For pollen of S. dioica applied to S. latifolia, the
model fit was not significantly improved either by allowing s to
differ between the majority S. dioica and majority S. latifolia
pollen mixes (x2 ¼ 0.56, P ¼ 0.45), or by allowing s to differ
between high and low quantities (x2 ¼ 1.17, P ¼ 0.28).
The overall siring success of S. latifolia pollen applied
to S. dioica flowers did not differ significantly from 1.0
(x2 ¼ 1.88, P ¼ 0.17). However, irrespective of pollen quantity,
the relative competitive ability of S. latifolia pollen in majority
S. dioica mixes (s ¼ 1.76) was significantly higher than in
majority S. latifolia mixes (s ¼ 0.89; x2 ¼ 4.51, P ¼ 0.03).
Additionally, irrespective of pollen composition, the relative
competitive ability of S. latifolia in large quantities (s ¼ 1.88)
was significantly greater than in small quantities (s ¼ 0.88;
x2 ¼ 4.37, P ¼ 0.04).
DISCUSSION
In this study, it was found that S. latifolia pollen sires seeds in
proportion to its pollen’s relative abundance on S. dioica
flowers, but S. dioica pollen sires disproportionately few
seeds on S. latifolia flowers. This result indicates that mechanisms operating between pollination and seed set limit of introgression of S. dioica into S. latifolia, while not limiting
introgression of S. latifolia into S. dioica. Our methods differed from the related study of Rahmé et al. (2009) in
several respects: most notably, our study included majority
S. dioica or majority S. latifolia mixes, whereas theirs included
an equal pollen mix; and our study included two pollen quantities, the larger of which was about 10 % smaller than the
quantity they applied. Despite these differences, both studies
found an asymmetrical barrier to hybridization of similar
strength operating against S. dioica pollen, suggesting that
this is a general result across different populations and
conditions.
Although only S. latifolia exhibited a conspecific seed-siring
advantage, both species exhibited conspecific advantage in our
measures of pollen-tube growth. Conspecific advantage in
pollen-tube growth is unlikely to result from heterospecific
pollen-tube attrition because tubes of both species in heterospecific styles were on average shorter than conspecific
styles, indicating that pollen tubes had not yet achieved the
length necessary for fertilization of conspecific flowers.
Interactions between pollen tubes and stylar tissue are known
to affect pollen-tube elongation, although the mechanisms
involved and the role of these interactions in mediating interspecific fertilizations is a matter of ongoing research (Wheeler
et al., 2001). The findings that conspecific pollen tubes were
longer in both crosses and that S. latifolia pollen tubes
elongated further in conspecific than heterospecific stigmas
suggest that pollen – pistil interactions affected growth rates.
The lack of a conspecific siring advantage for S. dioica
despite its conspecific advantage in pollen-tube elongation
could indicate that other mechanisms, such as differential
pollen attrition over longer time intervals, were more important in determining paternity.
The conflict between the pollen-tube growth and hybridization results may have arisen because the pollen-tube study
measures the performances of only the few longest pollen
tubes, while the hybridization study assesses performance
over all pollen grains that sire viable seeds. For example,
pollen-tube elongation is influenced by interactions between
stylar tissue and nearby pollen tubes (Cruzan, 1990). Thus,
conspecific pollen tubes may initially grow faster through
stylar tissue, but such growth could facilitate the passage of
later tubes (Visser and Verhaegh, 1980), potentially reducing
or eliminating conspecific advantage. In this study, measurements were restricted to the longest few pollen tubes because
shorter pollen tubes were not discernable. Additionally, the
discrepancy between pollen-tube lengths and paternity may
have occurred because pollen tubes were measured at only
one time interval, and growth rates may vary with location
in the style (Mulcahy and Mulcahy, 1983, Douglas and
Cruden, 1994). By virtue of its larger pollen-grain size
S. latifolia may have maintained a faster growth rate than
S. dioica in S. dioica styles after the 3-h time period,
thereby eliminating S. dioica’s initial advantage in conspecific
styles. Measures of pollen tubes may have more accurately predicted seed-siring success for S. dioica, the shorter-styled
species, because pollen tubes were closer to the style’s base
in this species at the time of measurement.
Other studies investigating pollen-tube growth and seed
hybridity have also found conflicting results. Studies by
Emms et al. (1996) and Carney et al. (1994) both found cases
in irises of a conspecific seed-siring advantage despite a heterospecific pollen-tube growth advantage. Rieseberg et al. (1995)
did not detect conspecific pollen-tube growth advantage in sunflowers, but they did detect conspecific advantage in siring
seeds. In contrast, Husband et al. (2002) and Song et al.
(2002) found agreement between patterns of tube elongation
and seed hybridity in fireweed and rice, respectively. These
results indicate that studies of pollen-tube growth rates are valuable in investigating one mechanism of conspecific pollen precedence, but that such studies alone are not sufficient to
predict conspecific pollen precedence overall.
Previous studies have found that when differential pollentube elongation or pollen-tube attrition occurs, pollen from
the longer-styled species may perform better in interspecific
crosses than pollen from the shorter-styled species in reciprocal crosses (Buchholz et al., 1935; Smith, 1970; Carney et al.,
1996; but see Emms et al., 1996; Diaz and Macnair, 1999;
Wolf et al., 2001). Silene latifolia generally has larger
flowers and pollen than S. dioica (Baker, 1947b; Prentice,
1978), and styles were longer for S. latifolia than S. dioica
at the time of pollination in our pollen-tube study. It has
been hypothesized that the longer-styled species may generally
be less disadvantaged in interspecific crosses because its
pollen tends to be larger (Diaz and Macnair, 1999) and its
pollen tubes are adapted to elongate faster and to greater
lengths (Smith, 1970, Carney et al., 1996). The present paternity results are consistent with this hypothesis, although the
pollen-tube results are not. However, for aforementioned
reasons, paternity success is a better indicator of overall
success in this study.
The siring success of Silene dioica pollen on S. latifolia
flowers was similar, whether pollen mixes were applied in
Montgomery et al. — Asymmetrical conspecific seed-siring advantage in Silene
small or large quantities. This outcome suggests that conspecific
pollen on S. latifolia flowers has a pre-zygotic time-invariant
advantage, such as lower rates of pollen-tube attrition, which
could result from the longer style length in S. latifolia. Seed
set tended to be higher for flowers pollinated with conspecific
than heterospecific pollen, indicating that hybrid inviability
may also contribute to S. dioica’s seed siring disadvantage
(Table 1). Differences in seed germination rates were controlled
for in the analysis, so do not influence the calculated differences
in siring success. Thus, a combination of pollen-tube attrition
and post-fertilization abortion appears the most likely explanation for the conspecific pollen advantage in S. latifolia.
This outcome suggesting the lack of a time-variant advantage for conspecific pollen in S. latifolia flowers is surprising
because conspecific pollen tubes were longer, suggesting competitive superiority, and the difference in pollen quantity was
large, being 3-fold higher than in the low quantity treatment.
One explanation for this pattern could be that pollen limited
seed set at both small and large quantities, such that pollentube competition was unimportant and most pollen tubes
reaching the ovary succeeded in fertilization. On average,
total pollen receipt ranged from nearly 500 pollen grains per
flower (95 grains per style 5 styles) in the low treatment
to .1700 grains (344 grains per style) in the high treatment.
Flowers of greenhouse grown S. dioica and S. latifolia
flowers from the present study populations contain nearly
300 and 400 ovules, respectively (L. F. Delph, unpubl. res.).
Thus, flowers in the high pollination treatment received on
average at least 4-fold more pollen grains than they had
ovules. Nonetheless, average seed set was considerably lower
than ovule counts for both pollen quantities, and higher seed
set in the high- than low-pollen quantity treatment indicates
pollen limitation for the low-quantity treatment. Thus, it is
possible that the excess pollen was not enough for competition
to substantially impact the results.
The siring success of S. latifolia was greater when mixes
were applied to S. dioica flowers in large quantities, suggesting
a time-variant heterospecific advantage for S. latifolia pollen
(Table 1). However, the lack of an overall statistically significant advantage to S. latifolia pollen on S. dioica flowers indicates that the advantage is small or compensated for by a
time-invariant disadvantage. The decreased siring success of
S. latifolia in majority S. latifolia pollen mixes relative to
majority S. dioica mixes may have resulted from the increased
importance of intraspecific relative to interspecific competition
among pollen grains when S. latifolia pollen was common.
Increased intraspecific competition could occur if the two
species of pollen tend to reach ovules in different parts of
the ovary, as has been inferred in other studies (Manglesdorf
and Jones, 1926; Smith and Clarkson, 1956; Carney et al.,
1996; but see Kearney, 1932). The same pattern could also
be created if the rate of maternal provisioning to hybrid
ovules decreased as the numbers of such ovules increased.
Patterns of chloroplast and nuclear markers in the field
suggest that S. latifolia is usually the seed parent when introgression occurs, an outcome that could result from
S. latifolia producing more seeds than S. dioica or from
S. dioica pollen achieving more fertilizations on S. latifolia
flowers than S. latifolia pollen on S. dioica flowers (Minder
et al. 2007). The data indicate that S. dioica pollen is not
603
more competitive than S. latifolia in siring viable seeds on
the other species’ flowers; thus, the pattern of introgression
is more likely explained by either different levels of seed production or by S. dioica pollen reaching S. latifolia flowers at a
higher rate than S. latifolia pollen reaching S. dioica flowers.
Consistent with the latter hypothesis, Goulson and Jerrim
(1997) found that for arrays with equal numbers of both
species, a higher proportion of S. latifolia flowers received
S. dioica pollen than vice versa. The pattern could also
result from higher survival under field conditions of hybrids
with S. latifolia than S. dioica seed parents. Karrenberg and
Favre (2008) argue that differential introgression from
S. dioica to S. latifolia may occur because contact sites tend
to be located in conditions better suited for S. latifolia than
S. dioica.
Parsing the relative contributions of various barriers to
hybridization is a challenge. In a study designed to estimate
the sequential effects of multiple factors contributing to reproductive isolation between two Mimulus species, Ramsey et al.
(2003) found that, despite strong conspecific pollen precedence, this mechanism accounted for only a small proportion
of total isolation. As with Mimulus, this and other studies of
these Silene species suggest that multiple barriers contribute
to isolation and the strength of each barrier is not necessarily
reciprocal for both species (Goulson and Jerrim, 1997;
Minder et al., 2007, Karrenberg and Favre, 2008).
Conclusions
This research demonstrates that conspecific pollen tubes
elongate farther in the measured time period in both species.
However, conspecific pollen was only more successful than
heterospecific pollen in siring viable seeds in one species,
S. latifolia. Our finding in S. dioica of a conspecific pollentube advantage at an intermediate point in the style but no conspecific advantage in seed siring suggests that pollen-tube
growth rates may differ with location in the style. The siring
advantage of S. latifolia pollen in its own styles may result
from increased time-invariant pre-fertilization advantages or
post-hybridization advantages in viability or sequestering of
resources. Overall, these results indicate that net barriers to
hybridization operating between pollination and seed germination contribute to reproductive isolation for S. latifolia but not
S. dioica seed parents.
ACK NOW LED GE MENTS
We thank C. Herlihy and S. Scarpino for provision of seeds.
Comments of anonymous reviewers improved this manuscript.
This work was supported by National Science Foundation
grant DEB 0210971 to L.F.D.
L I T E R AT U R E CI T E D
Alarcón R, Campbell DR. 2000. Absence of conspecific pollen advantage in
the dynamics of an Ipomopsis (Polemoniaceae) hybrid zone. American
Journal of Botany 87: 819– 824.
Arnold ML. 1994. Natural hybridization and Lousiana irises. Bioscience 44:
141–147.
604
Montgomery et al. — Asymmetrical conspecific seed-siring advantage in Silene
Ascher PD, Peloquin SJ. 1968. Pollen tube growth and incompatibility following intra- and inter-specific pollinations in Lilium longiflorum.
American Journal of Botany 55: 1230– 1234.
Baker HG. 1947a. Melandrium album (Mill.) Garcke. Journal of Ecology 35:
274–282.
Baker HG. 1947b. Melandrium dioicum (L. emend.) Coss and Germ. Journal
of Ecology 35: 283–292.
Bopp S, Gottsberger G. 2004. Importance of Silene latifolia ssp. alba
and S. dioica (Caryophyllaceae) as host plants of the parasitic
pollinator Hadena bicruris (Lepidoptera, Noctuidae). Oikos 105:
221–228.
Buchholz JT, Williams LF, Blakeslee AF. 1935. Pollen-tube growth of
ten species of Datura in interspecific pollinations. Genetics 21:
651–656.
Campbell DR, Alarcón R, Wu CA. 2003. Reproductive isolation and hybrid
pollen disadvantage in Ipomopsis. Journal of Evolutionary Biology 16:
536–540.
Carney SE, Arnold ML. 1997. Differences in pollen-tube growth rate and
reproductive isolation between Louisiana irises. Journal of Heredity 88:
545–549.
Carney SE, Cruzan MB, Arnold ML. 1994. Reproductive interactions
between hybridizing irises: analyses of pollen tube growth and fertilization success. American Journal of Botany 81: 1169– 1175.
Carney SE, Hodges SA, Arnold ML. 1996. Effects of differential
pollen-tube growth on hybridization in the Lousiana irises. Evolution
50: 1871– 1878.
Chen CC, Gibson PB. 1972. Barriers to hybridization of Trifolium repens
with related species. Canadian Journal of Genetics and Cytology 14:
381–389.
Cruzan MB. 1990. Pollen–pollen and pollen– style interactions during pollen
tube growth in Erythronium grandiflorum (Liliaceae). American Journal
of Botany 77: 116–122.
Cruzan MB, Barrett SCH. 1996. Postpollination mechanisms influencing
mating patterns and fecundity: an example from Eichhornia paniculata.
American Naturalist 147: 576–598.
Darwin C. 1859. On the origin of species. London: John Murray.
Dean MD, Nachman MW. 2009. Faster fertilization rate in conspecific versus
heterospecific matings in house mice. Evolution 63: 20–28.
Desfeux C, Lejeune B. 1996. Systematics of Euromediterranean Silene
(Caryophyllaceae): evidence from a phylogenetic analysis using ITS
sequences. Comptes Rendus de l’Academie des Sciences – Serie III:
Sciences de la Vie 319: 351– 358.
Diaz A, Macnair MR. 1999. Pollen tube competition as a mechanism of prezygotic reproductive isolation between Mimulus nasutus and its presumed
progenitor M. guttatus. New Phytologist 144: 471–478.
Douglas KL, Cruden RW. 1994. The reproductive biology of Anemone
canadensis (Ranunculaceae): breeding system and facilitation of sexual
selection. American Journal of Botany 81: 314–321.
Emms SK, Hodges SA, Arnold ML. 1996. Pollen-tube competition, siring
success, and consistent asymmetric hybridization in Louisiana irises.
Evolution 50: 2201–2206.
Figueroa-Castro DM, Holtsford TP. 2009. Post-pollination mechanisms in
Nicotiana longiflora and N. plumbaginifolia: pollen tube growth rate, offspring paternity and hybridization. Sexual Plant Reproduction 22:
187–196.
Goulson D, Jerrim K. 1997. Maintenance of the species boundary between
Silene dioica and S. latifolia (red and white campion). Oikos 79:
115–126.
Grant S, Houben A, Vyskot B, et al. 1994. Genetics of sex determination in
flowering plants. Developmental Genetics 15: 214–230.
Grant V. 1949. Pollination systems as isolating mechanisms in angiosperms.
Evolution 2: 82–97.
Grant V. 1963. The origin of adaptations. New York, NY: Columbia
University Press.
Hauser TP, Jorgensen RB, Ostergard H. 1997. Preferential exclusion of
hybrids in mixed pollinations between oilseed rape (Brassica napus)
and weedy B. campestris (Brassicaceae). American Journal of Botany
84: 756–762.
Hedges LV, Gurevitch J, Curtis PS. 1999. The meta-analysis of response
ratios in experimental ecology. Ecology 80: 1150–1156.
Hosmer DW, Jovanovic B, Lemeshow S. 1989. Best subsets logistic
regression. Biometrics 45: 1265– 1270.
Howard DJ. 1999. Conspecific sperm and pollen precedence and speciation.
Annual Review of Ecology and Systematics 30: 109– 132.
Husband BC, Schemske DW, Burton TL, Goodwillie C. 2002. Pollen
competition as a unilateral reproductive barrier between sympatric
Chamerion angustifolium. Proceedings of the Royal Society of London
Series B – Biological Sciences 269: 2565– 2571.
Jaynes RA. 1968. Interspecific crosses in Kalmia. American Journal of Botany
55: 1120–1125.
Jurgens A, Witt T, Gottsberger G. 1996. Reproduction and pollination in
central European populations of Silene and Saponaria species. Botanica
Acta 109: 316– 324.
Karrenberg S, Favre A. 2008. Genetic and ecological differentiation in the
hybridizing campions Silene dioica and S. latifolia. Evolution 62:
763– 773.
Kearney TH. 1932. Pollen antagonism in cotton. Journal of Agricultural
Research 44: 191– 226.
Kearns CA, Inouye DW. 1993. Techniques for pollination biologists. Niwot,
CO: University of Colorado Press.
Klips RA. 1999. Pollen competition as a reproductive isolating mechanism
between two sympatric Hibiscus species (Malvaceae). American
Journal of Botany 86: 269–272.
Levin DA. 1978. The orgin of isolating mechanisms in flowering plants. In:
Hecht MW, Steere WC, Wallace B eds. Evolutionary biology, Vol. II.
New York, NY: Appleton Century Crofts, 185– 317.
Mangelsdorf PC, Jones DF. 1926. The expression of Mendelian factors in the
gametophyte of maize. Genetics 11: 423– 455.
Matsunaga S, Isono E, Kejnovsky E, et al. 2003. Duplicative transfer of a
MADS box gene to a plant Y chromosome. Molecular Biology and
Evolution 20: 1062–1069.
Meléndez-Ackerman E, Campbell DR. 1998. Adaptive significance of
flower color and inter-trait correlations in an Ipomopsis hybrid zone.
Evolution 52: 1293–1303.
Minder AM, Rothenbuehler C, Widmer A. 2007. Genetic structure of hybrid
zones between Silene latifolia and Silene dioica (Caryophyllaceae):
evidence for introgressive hybridization. Molecular Ecology 16:
2504–2516.
Mulcahy GB, Mulcahy DL. 1983. A comparison of pollen tube growth in
bi-and trinucleate pollen. In: Mulcahy DL, Ottaviano E eds.
Pollen:biology and implications for plant breeding. New York, NY:
Elsevier Biomedical, 29– 33.
Prentice HC. 1978. Experimental taxonomy of Silene section Elisanth
(Caryophyllaceae): crossing experiments. Botanical Journal of the
Linnean Society 77: 203–216.
Prentice HC. 1979. Numerical analysis of infraspecific variation in European
Silene alba and Silene dioica (Caryophyllaceae). Botanical Journal of the
Linnean Society 78: 181–212.
van Putten WF, Elzinga JA, Biere A. 2007. Host fidelity of the pollinator
guilds of Silene dioica and Silene latifolia: possible consequences for
sympatric host race differentiation of a vectored plant disease.
International Journal of Plant Sciences 168: 421– 434.
Rahmé J, Widmer A, Karrenberg S. 2009. Pollen competition as an asymmetric reproductive barrier between two closely related Silene species.
Journal of Evolutionary Biology 22: 1937–1943.
Ramsey J, Bradshaw HD, Schemske DW. 2003. Components of reproductive
isolation between the monkeyflowers Mimulus lewisii and M. cardinalis
(Phrymaceae). Evolution 57: 1520–1534.
Rasband WS. 1997–2007. ImageJ. Bethesda, MD, USA: US National
Institutes of Health.
Richardson TE, Stephenson AG. 1991. Effects of parentage, prior fruit set
and pollen load on fruit and seed production in Campanula americana
L. Oecologia 87: 80–85.
Rieseberg LH, Desrochers AM, Youn SJ. 1995. Interspecific pollen
competition as a reproductive barrier between sympatric species of
Helianthus (Asteraceae). American Journal of Botany 82:
515– 519.
Smith EB. 1968. Pollen competition and relatedness in Haplopappus in
section Isopappus. Botanical Gazette 129: 371–373.
Smith EB. 1970. Pollen competition and relatedness in Haplopappus section
Isopappus (Compositate). II. American Journal of Botany 57: 874–880.
Smith FH, Clarkson QD. 1956. Cytological studies of interspecific
hybridization in Iris, subsection Californicae. American Journal of
Botany 43: 582–588.
Montgomery et al. — Asymmetrical conspecific seed-siring advantage in Silene
Song ZP, Lug BR, Zhu YG, Chen JK. 2002. Pollen competition between
cultivated and wild rice species (Oryza sativa and O. rufipogon). New
Phytologist 153: 289– 296.
Visser T, Verhaegh JJ. 1980. Pollen and pollination experiments. 2. The
influence of the 1st pollination on the effectiveness of the 2nd one in
apple. Euphytica 29: 385–390.
Wang JQ, Cruzan MB. 1998. Interspecific mating in the Piriqueta caroliniana
(Turneraceae) complex: effects of pollen load size and composition.
American Journal of Botany 85: 1172– 1179.
Wesselingh RA, Arnold ML. 2000. Pollinator behaviour and the evolution
of Louisiana iris hybrid zones. Journal of Evolutionary Biology 13:
171– 180.
605
Wheeler MJ, Franklin-Tong VE, Franklin FCH. 2001. The molecular
and genetic basis of pollen–pistil interactions. New Phytologist 151:
565–584.
Willmot A, Moore PD. 1973. Adaptation to light intensity in Silene alba and
Silene dioica. Oikos 24: 458–464.
Wolf PG, Campbell DR, Waser NM, Sipes SD, Toler TR, Archibald JK.
2001. Tests of pre- and postpollination barriers to hybridization
between sympatric species of Ipomopsis (Polemoniaceae). American
Journal of Botany 88: 213– 219.
Zhang YH, Di Stilio VS, Rehman F, Avery A, Mulcahy D, Kesseli R. 1998.
Y chromosome specific markers and the evolution of dioecy in the genus
Silene. Genome 41: 141– 147.