Int. J. Plant Sci. 160(3):577–584. 1999.
q 1999 by The University of Chicago. All rights reserved.
1058-5893/99/6003-0015$03.00
SELF-STERILITY IN THE UNDERSTORY HERB CLINTONIA BOREALIS (LILIACEAE)
M. E. Dorken1 and B. C. Husband
Department of Botany, University of Guelph, Guelph, Ontario N1G 2W1, Canada
In the understory perennial Clintonia borealis, self-pollination yields lower seed number per fruit than crosspollination. To measure the magnitude of this partial self-sterility and identify the pre- and postzygotic mechanisms on which it is based, we conducted self- and outcross-pollinations in 34 patches in a natural population
and estimated the effects on seed and fruit production, frequency of fertilization, and postzygotic endosperm
development. Plants in nine of the 34 patches set no fruit after pollination. In the remaining patches, 22% of
self-pollinated flowers and 54% of outcrossed flowers set fruit. Mean seed set per fruit was 0.38 (SE 5
0.04) for outcrossed flowers compared with 0.09 (SE 5 0.04) after self-pollination. The self : outcross ratio
of seed set ranged from 1.88 to 0.0 among patches, but the effects of selfing did not differ significantly among
patches. Observations of cleared ovules from 1, 2, 3, 4, 6, 8, and 10 d after pollination showed that fertilization
does not occur until day 3. The proportion of ovules fertilized increased with time and, at 10 d, was 0.44
(SE 5 0.18) for outcross-pollinations and 0.14 (SE 5 0.06) for selfs. In addition, the number of endosperm
nuclei in fertilized ovules increased more rapidly and the size of full seeds was larger after outcross-pollination
than self-pollination. Our results indicate that prezygotic mechanisms preventing fertilization can account for
much of the self-sterility observed in C. borealis in the field. However, some postzygotic differences were also
evident, which may influence the number and size of selfed versus outcrossed offspring.
Keywords: reproductive biology, self-sterility, partial incompatibility, fertilization frequency, prezygotic, postzygotic, endosperm nuclei, inbreeding depression.
Introduction
pression, on the other hand, is an expression of the genetic
constitution of the zygote. Reduced fitness of selfed offspring
due to inbreeding depression generally results from increased
homozygosity and the expression of deleterious, recessive alleles (Lande and Schemske 1985; Charlesworth and Charlesworth 1987; Husband and Schemske 1996). Although the two
mechanisms of self-sterility are conceptually distinguishable,
they are sometimes difficult to distinguish empirically, particularly early-acting inbreeding depression and late-acting SI. In
most cases, identifying such post- and prezygotic processes
requires direct observation of pollen tube growth, fertilization
events, or embryo development (Seavey and Bawa 1986;
Seavey and Carter 1994, 1996; Manicacci and Barrett 1996).
Although the immediate outcomes of both SI and inbreeding
depression are to reduce the number of selfed offspring, these
processes have different fitness consequences for the maternal
plant. Active rejection of self-pollen in the stigma or style, as
observed in self-incompatibility, may occur sufficiently early
so that self-pollination would not reduce the number of ovules
available for cross-fertilization (but see Waser and Price 1991;
Scribailo and Barrett 1994). In contrast, postzygotic inbreeding
depression is potentially more costly to the maternal parent in
that ovules are usurped by self-fertilization and energy
“wasted” on the early development of selfed zygotes (Lewis
1979; Beach and Kress 1980; Barrett et al. 1996). Alternatively,
it has been suggested that postponing the blockage of selfpollen may increase fitness by extending the period of time
over which pollen genotypes may be “evaluated” by the maternal parent (Seavey and Bawa 1986). The contrasting implications of pre- and postzygotic self-sterility are related not
Most flowering plants are hermaphroditic and thus potentially self-pollinating (Fryxell 1957; Richards 1997). Despite
the opportunities for self-fertilization, many plant species are
completely or partially self-sterile, producing few or no offspring after self-pollination (Darwin 1876; de Nettancourt
1977, 1997). Self-sterility has important implications for the
design of breeding programs and cultivation practices for domesticated plants, and it represents one of the most important
selective forces governing the evolution of pollination systems
in natural populations. However, the underlying proximal
causes of self-sterility and their evolutionary origins are still
poorly understood (Seavey and Bawa 1986; Barrett 1988; Husband and Schemske 1996).
The genetic causes of self-sterility can be divided into two
categories: physiological self-incompatibility and inbreeding
depression. In systems of self-incompatibility (SI), self-pollen,
or pollen sharing the same incompatibility alleles as occur in
the pollen recipient are recognized and rejected on the stigma,
in the style, or in the ovary of the maternal parent (Brock
1954; Seavey and Bawa 1986; Barrett 1988; Sage et al. 1994).
However, some cases of postzygotic self-sterility have been interpreted as late acting, or ovarian SI (Seavey and Bawa 1986;
Sage and Williams 1991; Sage et al. 1994). Inbreeding de1
Author for correspondence and reprints. Current address: Department of Botany, University of Toronto, Toronto, Ontario M5S
3B2, Canada; e-mail [email protected].
Manuscript received April 1998; revised manuscript received December 1998.
577
578
INTERNATIONAL JOURNAL OF PLANT SCIENCES
only to the timing of their effects but also to the mechanisms
for their origin. Whereas inbreeding depression is the genetic
by-product of mutation and inbreeding on zygote fitness, incompatibility is a prezygotic mechanism that may have evolved
as a response to inbreeding depression (Charlesworth and
Charlesworth 1979).
The reproductive biology of the understory perennial Clintonia borealis has been examined in numerous studies, and,
in all cases, it has exhibited low seed set upon self-pollination
compared with cross-pollination (Galen et al. 1985; Galen
and Weger 1986; Barrett and Helenurm 1987). Barrett and
Helenurm (1987) found that self-pollination resulted in 57%
fewer seeds per fruit than cross-pollination, and Galen et al.
(1985) reported self-fertility as one-third of that from outcrossing. Moreover, seed production upon selfing appears to
be highly variable, with some individuals exhibiting complete
self-fertility and others, complete sterility (Galen et al. 1985;
Galen and Weger 1986). This apparent quantitative variation
in fertility may be the result of inbreeding depression caused
by variation in genetic load among individual clones. Alternatively, systems of partial self-incompatibility may exhibit
similar patterns of seed production (Cooper and Brink 1940;
Waser 1993).
Our goal was to estimate the magnitude of self-sterility in
a natural population of C. borealis and to examine the roles
of pre- and postzygotic mechanisms. To address this, we conducted cross- and self-pollinations on 34 clones in a southern
Ontario population and examined seed set, frequency of fertilization, and endosperm development. Specifically, we addressed the following questions: Do seed set and seed size differ
between offspring from self- and cross-pollination? Does the
degree of self-sterility, measured as the relative production of
selfed seeds, vary among clones within a population? Are there
differences in rates of fertilization between self- and crosspollinations? Finally, are there differences in the rate of postzygotic endosperm development between self- and crosspollinations?
Material and Methods
Natural History and Study Site
Clintonia borealis (Ait.) Raf. (Liliaceae) is a perennial, forest
understory herb found throughout the northeastern United
States and Canada (Fernald 1950). It forms long-lived clones,
the ramets of which are connected via rhizomes (Pitelka et al.
1985). Ramets live for 1 yr, and those that bloom produce
from one to six weakly protogynous yellow-green flowers (Galen and Weger 1986). Pollination is performed primarily by
bumblebees (Galen et al. 1985; Barrett and Helenurm 1987).
Our data were collected from a population of C. borealis in
the Little Tract Agreement Forest near Guelph, Ontario, during
the summers of 1994 and 1995.
The population of C. borealis at our study site formed loose
to dense patches in the forest understory. A preliminary electrophoretic survey indicated that most patches were fixed for
separate genotypes. In 1994, leaf material for cellulose acetate
electrophoresis was sampled from 10 ramets within each of
34 patches. We found that 85.3% of all surveyed patches (29
of 34) were composed of single electrophoretic phenotypes for
the enzyme phosphoglucose isomerase (PGI). All patches were
dominated by a single genotype, although one of the 34 patches
contained more than two electrophoretic phenotypes. A total
of 12 distinct electrophoretic phenotypes were detected (M. E.
Dorken and B. C. Husband, unpublished data).
Fruit and Seed Production
To examine the effects of selfing on fruit and seed production, we hand-pollinated flowers in the field in 1994. Because
of the difficulty in finding inflorescences with sufficient numbers of flowers, we replicated the hand-pollination treatments
on a pair of neighboring inflorescences within each patch. Prior
to flowering, pairs of inflorescences separated by !0.5 m were
located within 34 patches in the population and covered with
mesh bags to exclude pollinators. Two flowers on each inflorescence within a pair were randomly assigned one of two
pollination treatments: self-pollination and outcross-pollination. For self-pollinated flowers, pollen was acquired from anthers within the target flower or from another flower on the
same inflorescence. Pollen for the outcross treatment was gathered from two flowers at least 10 m away from each other
and the target flower. Our choice of a minimum distance of
10 m was based on findings by Galen et al. (1985), who
showed no difference in seed set between outcross-pollinations
with pollen collected 10 or 200 m away.
A third flower on each of the paired inflorescences was used
for two different control pollinations. On one randomly chosen inflorescence of the pair, the third flower was left unpollinated to check for inadvertent pollen transfer. On the remaining inflorescence, the flower was pollinated using pollen
from the other inflorescence in the pair to confirm that the
paired inflorescences belonged to the same genet. If the two
inflorescences were ramets from the same genet, as the electrophoretic screening indicated, seed set in this treatment
should not differ from the self-pollination treatment. All treatments were randomly assigned to the first three flowers within
an inflorescence. Excess flowers were removed immediately
after performing the last pollination. Flowers were emasculated prior to all non-self-pollinations. In addition to the inflorescences chosen for hand pollinations, 34 inflorescences,
one near each experimental pair, were tagged and left to be
open pollinated.
Following fruit maturation, berries were harvested, and
the number of full seeds and ovules within each berry were
counted. A berry was classified as a fruit if it had at least
one full seed. Full seeds were readily identified as they were
larger and plumper than undeveloped or unfertilized ovules
and had a well-developed seed coat. Seed set per fruit was
measured as the number of full seeds relative to the total
number of ovules produced. Differences in seed set among
self-, outcross-, and control pollinations were analyzed using a two-way, mixed-model ANOVA with patch membership (random effect) and pollination treatment (fixed effect)
as main effects. Means for the pollination treatments were
compared using a Tukey-Kramer HSD test. Only patches
DORKEN & HUSBAND—SELF-STERILITY IN CLINTONIA
with replicate values for each treatment were used in the
analysis. Results based on arcsine-transformed data were
the same as for untransformed data, so the latter are presented here. In addition, seed set in unpollinated flowers
(controls) was compared to zero using a one-tailed t-test.
Finally, to determine the effect of pollination on postzygotic
development, we measured the length and width of all full
seeds within the fruits from the experimental pollinations
and compared their means using a Tukey-Kramer HSD test.
Frequency of Fertilization and Endosperm
Development
To examine whether self-sterility was associated with preor postzygotic factors, we compared the frequency of fertilization of ovules and endosperm development in zygotes after
self- and cross-pollination. Seventy-one flowering ramets with
at least three flowers were chosen in 1995; on each, flowers
were randomly assigned one of three pollination treatments:
outcross-pollen, self-pollen, and no pollen. Pollinations, emasculations, and removal of extra flowers were performed as in
the previous year.
To examine temporal variation in fertilization, 10 ovaries
were collected at 1, 2, 3, 4, 6, 8, and 10 d following pollination.
Each inflorescence was randomly assigned to one of these harvest times. Upon collection, fruits were fixed in 50% FAA and
transferred to 50% ethanol after 24 h. Ovules were cleared
for light microscopy following methods in Stelly et al. (1984),
with modifications used by Scribailo and Barrett (1994). Each
ovary was opened and ovules were placed on a glass slide in
methyl salicylate and then examined under 100# magnification using a Leica DMRB Brightfield microscope.
We examined all ovules in 6–10 fruits per pollination treatment (self and outcross) from each harvest date (total n 5
2055 ovules) under the microscope and classified them as unfertilized or fertilized. Development of the embryo sac in C.
borealis is tetrasporic. The mature embryo sac is 5–6 nucleate
composed of an egg cell, two synergids, one upper polar nucleus, and one or two abortive nuclei at the chalazal end. The
lower polar nucleus is absent, and no antipodal cells are
formed (Walker 1944; Pahuja and Kumar 1971). Ovules were
classified as fertilized when cell divisions in the polar nucleus
and egg nucleus were observed. In some ovules in which the
egg cell could not be observed clearly, divisions of the polar
nucleus were assumed to indicate double fertilization.
To determine whether selfed and outcrossed embryos differed in their early development, we estimated the number of
endosperm nuclei in all observed fertilized ovules. Postfertilization endosperm development was scored by counting numbers of endosperm nuclei in fertilized ovules. Differences in
fertilization frequency and postfertilization endosperm development between self- and outcross-pollination treatments were
investigated using a Model I, two-way ANOVA with pollination type and elapsed time since pollination as main effects.
No fertilizations or endosperm development were observed in
unpollinated flowers, and therefore unpollinated flowers were
not included in the following analyses. Fertilization frequencies
were arcsine transformed before analysis; however, untransformed data are presented in the “Results.” All statistical anal-
579
yses were conducted using JmP 3.0 (SAS Institute 1994) statistical software.
Results
Fruit and Seed Production
Of the 34 patches used for experimental pollinations in
1994, three were damaged prior to fruit set and thus were
excluded from further analyses. In 25 of the remaining 31
patches, fruit set occurred on at least one of the two ramets.
Across the 25 patches, mean fruit set was 0.22 (SE 5 0.05)
for self-pollinated flowers and 0.54 (SE 5 0.06) for outcrosspollinated flowers. Mean fruit set in unpollinated controls was
0.02 (SE 5 0.03).
On average, Clintonia borealis fruits contained 12.5
(SE 5 0.30) ovules, yet in unpollinated control flowers, seed
set averaged 0.01 (SE 5 0.06). Seeds were produced in only
one of the control flowers, and the mean was not significantly
different from zero in a one-tailed t-test (t 5 1.0, df 5 25,
P 5 0.16). The ANOVA on seed set indicated highly significant
differences among pollination treatments and marginally significant differences among patches (table 1). Mean seed set for
intrapatch-pollination (mean 5 0.19, SE 5 0.06) was approximately twice that from self-pollination (mean 5 0.09, SE 5
0.04); however, there was no significant difference between
these treatments (Tukey-Kramer HSD, Q 5 2.8, P 1 0.05 ;
fig. 1). Seed set per fruit from outcross-pollinations was more
than four times higher (mean 5 0.38, SE 5 0.04) than in selfpollinations (fig. 1). Open-pollinated seed set averaged 0.45
(SE 5 0.05), significantly higher than all but outcrosspollinations. Mean seed set ranged from 0 to 0.94 per patch
for outcrosses and from 0 to 0.89 for self-pollinations (fig. 2).
Fifteen of the 25 patches that produced at least one fruit had
no selfed fruit. For those patches with selfed and outcrossed
fruit, the self : outcross ratio of seed set varied from 1.88 to
0.10 among patches. Despite this wide variation, the interaction between pollination treatment and patch was not significant (table 1).
Seed size, calculated as the product of seed length and width,
differed significantly among pollination treatments. Self- and
intrapatch-pollinations produced seeds of similar size (self
mean 5 0.44 mm2, SE 5 0.02; intrapatch mean 5 0.40 mm2,
SE 5 0.03). Outcrossed pollinations produced significantly
Table 1
Two-Way, Mixed-Model ANOVA on Seed Set in Clintonia borealis
Effect
SS
df
F
P
Patch . ... ... ... ... .... .. .... .. ...
Pollination type .. ... .... .. .... ..
Patch # pollination type ... ...
Residual .. .. ..... . ..... . .... ... ..
4.30
1.88
2.40
3.19
30
1
30
44
1.79
23.65
1.10
!0.0001
0.06
0.38
Note. The main effect “patch” tests for variation in seed production among 31 patches; “pollination type” tests for the effects of
outcross- vs. self-pollinations. Patch was treated as a random effect,
pollination type as a fixed effect. Means for pollination type are presented in fig. 1.
580
INTERNATIONAL JOURNAL OF PLANT SCIENCES
sac averaged 3.9 (SE 5 1.30) in self-pollinations, nearly half
of that observed in outcrosses (mean 5 8.0, SE 5 1.99). A significant interaction between pollination type and time elapsed
following pollination was observed, owing to much higher
endosperm nuclei production 8 and 10 d following pollination
in outcrosses.
Discussion
Fig. 1 Mean (5SE) seed set after outcross-, self-, intrapatch
self-, control, and open pollinations for 31 patches of Clintonia borealis. Statistical analysis of these data is presented in table 1. Results
of Tukey-Kramer HSD multiple comparison are shown; treatments
sharing the same lowercase letter were not significantly different.
larger seeds (mean 5 0.60 mm2, SE 5 0.03) than all other pollination treatments, based on a Tukey-Kramer HSD comparison of means (Q 5 2.6; P ! 0.05).
Frequency of Fertilization and Embryo Development
The frequency of fertilization differed significantly among
sampling times (day effect), and among pollination treatments
(table 2). Fertilization was first observed on the third day after
pollination and increased with time, from 0.0 on day 1 to 0.30
(SE 5 0.10) on day 10 (fig. 3a). On day 10, 0.14 (SE 5 0.06)
of all ovules from selfed flowers and 0.44 (0.18) of all ovules
from outcrossed flowers were fertilized. Averaged over all days,
the fertilization frequency for self-pollinations was 0.11
(SE 5 0.03) compared with 0.35 (SE 5 0.07) for outcrosses.
Differences in fertilization frequency between self- and outcross-pollinated flowers were detected by 6 d after pollination
(table 2). Although the frequency of fertilization in outcrosspollinations continued to increase until 8 d after pollination,
fertilization in self-pollinations did not increase after the fourth
day (fig. 3a). There was a significant interaction between the
effects of pollination treatment and elapsed time since pollination, which indicates that the effect of pollination was not
uniform among days. This result was primarily due to the high
fertilization frequency of outcross-pollinations collected 8 and
10 d after pollination.
Postfertilization development of endosperm also differed significantly among days and among pollination treatments (table
2). Among fertilized ovules, the number of endosperm nuclei
per embryo sac increased from 0.0 at day 1, to 0.98 (SE 5
0.63) on day 3, to a maximum of 14.7 (SE 5 4.3) on day 10
(fig. 3b). The mean number of endosperm nuclei per embryo
The magnitude of self-sterility observed in this study was
greater than in previous investigations of Clintonia borealis
(Galen et al. 1985; Barrett and Helenurm 1987). Seed set upon
selfing was 0.09 in our study, which is !25% of outcrossed
seed set (0.38) and less than half the estimates of selfed seed
set (0.22 and 0.32) reported by Barrett and Helenurm (1987)
and Galen et al. (1985), respectively. These differences in the
strength of self-sterility may reflect variation in both environmental and genetic determinants of seed production among
populations (Lee 1988). Differences among populations in the
impact of self-pollination may arise because of variation in
composition of incompatibility alleles and differences in the
strength of the incompatibility reaction among those alleles.
Variation in genetic load may also be important, particularly
among populations that differ in inbreeding history as influenced by mating system, pollination, population size, and colonization history (Schemske and Lande 1985). These genetic
factors may operate singly or in combination to account for
variation in seed set upon selfing among populations of C.
borealis.
Environmental differences, such as variation in resource and
pollen availability, may also be contributing to interpopulation
variation in self-sterility. High fruit and seed set in openpollinated flowers indicate that pollen is not limited in C. borealis and that there is a sufficient amount of outcross-pollen
for adequate seed set. This is supported by Galen et al. (1985),
who showed that the addition of pollen to open-pollinated
stigmas of C. borealis did not increase seed set. Similarly, Barrett and Helenurm (1987) found no evidence of pollen limi-
Fig. 2 Mean seed set for selfed and outcrossed flowers in 25
patches of Clintonia borealis. Patches are ordered by the magnitude
of self : outcross ratio of seed set per patch.
DORKEN & HUSBAND—SELF-STERILITY IN CLINTONIA
Table 2
ANOVA for Frequency of Fertilization and the Number of
Endosperm Nuclei Observed in Fertilized Ovules of
Clintonia borealis
Source of variation
df
Fertilization
frequency
No. of endosperm
nuclei
Pollination . . .. ... .... ...
Day . ... ... ... .. . .. .. .....
Pollination # day ... ...
1
4
4
14.94∗∗∗
6.54∗∗∗
4.81∗∗
7.37∗
13.52∗∗∗
12.67∗∗∗
Note. The values presented are F-ratios from a two-way, Model
I ANOVA. Flowers were hand pollinated using either self- or outcrosspollen (pollination effect). Ovaries were collected 1, 2, 3, 4, 6, 8, and
10 d following pollination (day effect). For fertilization frequency and
number of endosperm nuclei, both pollination type and day were
treated as fixed effects. F-tests for both fertilization frequency and
number of endosperm nuclei used MSresidual as the denominator. The
degrees of freedom of the residual mean square were 81 and 27 for
the analysis of fertilization and endosperm nuclei, respectively. Means
are presented in figs. 2 and 3.
∗
P ! 0.05.
∗∗
P ! 0.01.
∗∗∗
P ! 0.001.
tation, as seed set was higher in open-pollinated flowers than
in hand-cross-pollinated ones. However, Galen et al. (1985)
demonstrated that seed production was artificially elevated by
14% when plants were supplemented with both resources and
pollen. This result indicates that differences in resources may
be an important determinant of seed set in C. borealis. Low
seed set upon selfing in our study may therefore be a result of
low resource availability. That differences in resource levels
exist among the populations of C. borealis that have been
studied is supported by the observations that the number of
ovules per flower and seed set in outcrossed flowers were variable among studies and consistently lowest in the population
examined in this study.
The evidence based on frequencies of fertilization in selfpollinated and outcrossed flowers clearly supports the presence
of a prezygotic mechanism of self-sterility. Fertilization occurred less often in self-pollinated flowers than in outcrossed
flowers. Furthermore, the mean values of fertilization frequency (self mean 5 0.11, outcross mean 5 0.35) were similar
to mean seed-set proportions (self mean 5 0.09, outcross
mean 5 0.38), which indicates that additional mechanisms are
not required to explain the observed patterns of seed set. No
previous studies have directly tested for prezygotic mechanisms
in C. borealis, although Galen et al. (1985) and Barrett and
Helenurm (1987) considered that self-sterility may be the result
of partial self-incompatibility. Our result is also consistent with
the observation that self-incompatibility is common in the family Liliaceae (Fryxell 1957). In Lilium longiflorum and Lilium
martagon, for example, rejection of self-pollen is prezygotic
and occurs in the style (Dickinson et al. 1982; Lunqvist 1991).
The specific nature of self-incompatibility and the location of
pollen rejection in C. borealis is beyond the scope of this study
and would require a quantitative analysis of pollen tube
growth rates.
We used divisions of the polar nucleus and zygote as the
581
primary indication that fertilization of the egg cell had occurred in C. borealis embryo sacs. In a small number of cases,
fertilization was assessed indirectly based on divisions of the
polar nucleus. This approach was necessary because it was not
always possible to clearly view the egg (or zygote) in all ovules.
Although this approach is widely used in studies of reproductive development, it may lead to inaccurate estimates of fertilization rates, particularly if the polar nucleus can divide in
the absence of double fertilization. However, for ovules in
which both endosperm and zygote were observed, we found
that the number of nuclei in the zygote was positively correlated with the number of nuclei in the endosperm (r 5 0.65,
P 1 0.0001 ). Moreover, there is no previous evidence to indicate that nuclear divisions in the endosperm commonly occur
without fertilization in C. borealis. In fact, in previous studies
of self-sterility, the reverse was true: some embryos in selfpollinated flowers were found to develop without endosperm
Fig. 3 Temporal variation in (a) mean frequency of ovules fertilized and (b) mean number of endosperm nuclei in developing ovules
after hand outcross- and self-pollinations in a natural population of
Clintonia borealis. Points are means 5 1 SE. Statistical analyses are
presented in table 2.
582
INTERNATIONAL JOURNAL OF PLANT SCIENCES
division (Manasse and Pinney 1991). As a result, it is likely
that measures of the frequency of fertilization based on endosperm division provide an accurate minimum estimate of
the frequency of fertilization in the ovary.
Two pieces of evidence indicate that postzygotic processes
are also acting to reduce seed set upon selfing in C. borealis.
First, we found differences in the rate of endosperm division
between selfed and outcrossed seeds. Although we have no
data for this trait beyond 10 d after pollination, reduced endosperm development may result in selfed seeds that are
smaller and more likely to be aborted early in development
(Galen et al. 1985). Hence, it seems likely that low rates of
endosperm development after fertilization has contributed to
self-sterility in C. borealis. Second, full seeds from outcrossed
flowers were larger than those from selfed flowers. This difference parallels the observed patterns of endosperm development and will also have important consequences for offspring fitness (Stanton 1984).
While differences in endosperm development and seed size
between self- and cross-fertilized ovules are manifested postzygotically, we cannot exclude the possibility that they are the
indirect effects of prezygotic mechanisms rather than the direct
result of postzygotic processes (inbreeding depression or lateacting incompatibility). For example, seed enlargement can
occur in the absence of fertilization and therefore may be independent of the genetic quality of the zygote (Brock 1954).
In partially self-incompatible plants, the development of selfed
fruits and their seeds may be slower because they have lower
sink strength for maternal resources, as a result of having fewer
seeds. These prezygotic differences may persist and negatively
affect offspring at several life-history stages (Becerra and Lloyd
1992). Although the pathways through which prezygotic processes may affect seed development are clear, few experimental
studies have quantified the relative importance of prezygotic
events on postzygotic fruit and seed development.
Despite the evidence supporting the existence of a postzygotic component to self-sterility, the underlying cause, either
late-acting incompatibility or inbreeding depression, is less
clear. Seavey and Bawa (1986) suggest several criteria that
could be used to distinguish the rejection response of classic
incompatibility systems from the expression of deleterious mutations in inbreeding depression. Among those, they suggest
that embryo failure occurring at a number of stages could likely
be attributed to inbreeding effects, since inbreeding depression
is expected to arise owing to rare mutations at a large number
of genes. For the same reasons, inbreeding depression should
result in a variable degree of self-sterility among genotypes,
assuming that individuals exhibit random variation in the
number of deleterious alleles. However, uniform failure of zygotes, both in time and among genotypes, may indicate active
rejection as expected with physiological self-incompatibility.
The pattern of endosperm division observed in self-fertilized ovules of C. borealis is consistent with an active
incompatibility response. The number of divisions rarely
exceeded two, and there was no evidence that endosperm
in even a portion of the selfed ovules continued to divide
as one would predict under an inbreeding model. This pattern of ovule development is in accord with studies on Asclepias syriaca (Sparrow and Pearson 1948; Kahn and
Morse 1991), Rhododendron (Williams et al. 1984), Chor-
isia, Tabebuia (Gibbs and Bianchi 1993), and Pseudowintera (Sage et al. 1999) that demonstrate only minor endosperm divisions in self-fertilized ovules. Additional
observations of endosperm development beyond the 10-d
interval are necessary to confirm the result for C. borealis.
The wide variation in seed set among clones observed in C.
borealis appears consistent with the effects of inbreeding depression. Seed set in selfed flowers ranged from 0 to 0.89
among clones, with 15 of 25 producing no seed. Expressed
relative to seed set in outcrossed flowers, seed set in selfed
flowers was continuously distributed and ranged from 1.88 to
0.10 among clones. This result is in accord with many studies
of inbreeding depression that demonstrate high inbreeding depression in the early stages of embryo development and significant variation in the effects of selfing among plants (Husband and Schemske 1995, 1996). However, even these results
are not without some ambiguity. First, despite the wide range
of values, differences among clones in self-sterility were not
statistically detectable, although the lack of a significant
patch # pollination interaction may be the result of low power,
attributable to low number of clones and high variation among
flowers within a clone. Second, variation among selfed
flowers in the proportion of fertilized ovules per ovary
(CV 5 117.5 5 11.8 SE) was higher than for seed set (CV 5
89.3 5 11.3 SE), which indicates that variation in seed set may
be associated with variation in fertilization rates and the
strength of self-incompatibility. Because of these uncertainties
in the interpretation of the seed set data, we feel it is likely
but perhaps premature to conclude that inbreeding depression
is an important factor causing postzygotic self-sterility in C.
borealis. A crossing experiment involving plants of different
degrees of relatedness would provide a more definitive test of
the role of inbreeding depression in postzygotic self-sterility
(Seavey and Carter 1994).
The presence of both post- and prezygotic mechanisms of
self-sterility in plants is rarely reported in the literature, perhaps because both are rarely examined in the same plants.
Most studies of self-compatible species focus on inbreeding
depression (Husband and Schemske 1996), despite the fact that
prezygotic factors such as cryptic and partial self-incompatibility may also influence seed production and offspring fitness.
Microscopic studies of ovule development and fertilization,
infrequently used in studies of inbreeding depression, are more
likely to encounter both forms of sterility. For example, Manasse and Pinney (1991) found lower fertilization and higher
abortion in selfed flowers of Crinum erubescens, although they
attributed most of the self-sterility to postzygotic effects. Similarly, studies of Delphinium (Waser and Price 1993, 1994)
have revealed both prezygotic as well as postzygotic mechanisms of sterility in inbred matings. Despite the rarity of such
a result, the presence of both post- and prezygotic effects on
seed production may not be surprising. A species that is partially incompatible would, depending on the history of inbreeding, be expected to maintain relatively high genetic load
and hence exhibit postzygotic patterns of self-sterility (Levin
1984; Levin and Bulinska-Radomska 1988). Alternatively,
many species that are self-compatible may produce fewer seeds
under selfing and thus are partly self-sterile. In C. borealis,
prezygotic effects may be the dominant mechanism underlying
partial self-sterility, but postzygotic effects exist and may con-
DORKEN & HUSBAND—SELF-STERILITY IN CLINTONIA
tribute to the wide variation among individuals. A better understanding of the magnitude of pre- and postzygotic mechanisms and their interactions may provide further insight into
the basis and evolution of partial sterility in other plants, an
advance that is likely to arise only by direct examination of
pollination, double fertilization, and embryo development.
583
Acknowledgments
We thank P. Kron for technical assistance, N. M. Waser and
T. Sage for helpful reviews of the manuscript, and the Natural
Sciences and Engineering Research Council of Canada for financial support.
Literature Cited
Barrett SCH 1988 The evolution, maintenance, and loss of selfincompatibility systems. Pages 98–124 in J Lovett Doust, L Lovett
Doust, eds. Plant reproductive ecology: patterns and strategies. Oxford University Press, New York.
Barrett SCH, K Helenurm 1987 The reproductive biology of boreal
forest herbs. I. Breeding systems and pollination. Can J Bot 65:
2036–2046.
Barrett SCH, DG Lloyd, J Arroyo 1996 Stylar polymorphisms and
the evolution of heterostyly in Narcissus (Amaryllidaceae). Pages
339–376 in DG Lloyd, SCH Barrett, eds. Floral biology: studies on
floral evolution in animal-pollinated plants. Chapman & Hall, New
York.
Beach JH, WJ Kress 1980 Sporophyte versus gametophyte: a note on
the origin of self-incompatibility in flowering plants. Syst Bot 5:1–5.
Becerra JX, DG Lloyd 1992 Competition-dependent abscission of
self-pollinated flowers of Phormium tenax (Agavaceae): a second
action of self-incompatibility at the whole flower level? Evolution
46:458–469.
Brock RD 1954 Fertility in Lilium hybrids. Heredity 8:409–420.
Charlesworth D, B Charlesworth 1979 The evolution and breakdown
of self-incompatibility systems. Heredity 43:41–55.
——— 1987 Inbreeding depression and its evolutionary consequences. Annu Rev Ecol Syst 18:237–268.
Cooper DC, RA Brink 1940 Partial self-incompatibility and the collapse of fertile ovules as factors affecting seed formation in alfalfa.
J Agric Res 60:453–472.
Darwin C 1876 The effects of cross and self-fertilization in the vegetable kingdom. J. Murray, London.
de Nettancourt D 1977 Incompatibility in angiosperms. Springer,
Berlin.
——— 1997 Incompatibility in angiosperms. Sex Plant Reprod 10:
185–199.
Dickinson HG, L Moriarty, J Lawson 1982 Pollen-pistil interaction
in Lilium longiflorum: the role of the pistil in controlling pollen tube
growth following cross- and self-pollinations. Proc R Soc Lond B
215:45–62.
Fernald ML 1950 Gray’s manual of botany. 8th ed. American Book
Company, New York.
Fryxell PA 1957 Mode of reproduction in higher plants. Bot Rev 23:
135–233.
Galen C, RC Plowright, JD Thomson 1985 Floral biology and regulation of seed set and seed size in the lily, Clintonia borealis. Am
J Bot 72:1544–1552.
Galen C, HG Weger 1986 Re-evaluating the significance of correlations between seed number and size: evidence from a natural population of the lily Clintonia borealis (Ait.) Raf. Am J Bot 73:
346–352.
Gibbs PE, M Bianchi 1993 Post-pollination events in species of Chorisia (Bombacaceae) and Tabebuia (Bignoniaceae) with late-acting
self-incompatibility. Bot Acta 106:64–71.
Husband BC, DW Schemske 1995 Magnitude and timing of inbreeding depression in a diploid population of Epilobium angustifolium
(Onagraceae). Heredity 75:206–215.
——— 1996 Evolution of the magnitude and timing of inbreeding
depression in plants. Evolution 50:54–70.
Kahn AP, DH Morse 1991 Pollinium germination and putative ovule
penetration in self- and cross-pollinated common milkweed Ascelpias syriaca. Am Midl Nat 126:61–67.
Lande R, DW Schemske 1985 The evolution of self-fertilization and
inbreeding depression in plants. I. Genetic models. Evolution 39:
24–40.
Lee TD 1988 Patterns of fruit and seed production. Pages 179–202
in J Lovett Doust, L Lovett Doust, eds. Plant reproductive ecology:
patterns and strategies. Oxford University Press, New York.
Levin DA 1984 Inbreeding depression and proximity-dependent
crossing success in Phlox drummondii. Evolution 38:116–127.
Levin DA, Z Bulinska-Radomska 1988 Effects of hybridization and
inbreeding on fitness in Phlox. Am J Bot 75:1632–1639.
Lewis D 1979 Sexual incompatibility in plants. Studies in Biology no.
110. Arnold, London.
Lunqvist A 1991 Four-locus S-gene control of self-incompatibility
made probable in Lilium martagon (Liliaceae). Hereditas 114:
57–63.
Manasse RS, K Pinney 1991 Limits to reproductive success in a partially self-incompatible herb: fecundity depression at serial life-cycle
stages. Evolution 45:712–720.
Manicacci D, SCH Barrett 1996 Fertility differences among floral
morphs following selfing in tristylous Eichhornia paniculata (Pontedericeae): inbreeding depression or partial incompatibility. Am J
Bot 83:594–603.
Pahuja AN, V Kumar 1971 Embryo sac development, cytology, and
systematic position of Clintonia. Phytomorphology 20:97–102.
Pitelka LF, SB Hansen, JW Ashmun 1985 Population biology of
Clintonia borealis. I. Ramet and patch dynamics. J Ecol 73:
169–184.
Richards AJ 1997 Plant breeding systems. Allen & Unwin, London.
Sage TL, RI Bertin, EG Williams 1994 Ovarian and other late-acting
self-incompatibility systems. Pages 116–140 in EG Williams, AE
Clarke, RB Knox, eds. Genetic control of self-incompatibility and
reproductive development in flowering plants: advances in cellular
and molecular biology of plants. Vol 2. Kluwer, Dordrecht.
Sage TL, FB Sampson, P Bayliss, MG Gordon, EG Heij 1999 Selfsterility in the Winteraceae. Pages 317–328 in SJ Owens, P Rudell,
eds. Reproductive biology in systematics, conservation and economic botany. Royal Botanic Gardens, Kew.
Sage TL, EG Williams 1991 Self-incompatibility in Asclepias. Plant
Cell Incompatibility Newsl 23:55–57.
SAS Institute 1994 JmP user’s guide. 3d ed. SAS Institute, Cary, N.C.
Schemske DW, R Lande 1985 The evolution of self-fertilization and
inbreeding depression in plants. II. Empirical observations. Evolution 39:41–52.
Scribailo RW, SCH Barrett 1994 Effects of prior self-pollination on
outcrossed seed set in tristylous Pontederia sagittata (Pontederiaceae). Sex Plant Reprod 7:273–281.
Seavey SR, KS Bawa 1986 Late-acting self-incompatibility in angiosperms. Bot Rev 52:15–219.
584
INTERNATIONAL JOURNAL OF PLANT SCIENCES
Seavey SR, SK Carter 1994 Self-sterility in Epilobium obcordatum
(Onagraceae). Am J Bot 81:331–338.
——— 1996 Ovule fates in Epilobium obcordatum (Onagraceae).
Am J Bot 83:316–325.
Sparrow FK, NL Pearson 1948 Pollen compatibility in Asclepias syriaca. J Agric Res 77:187–199.
Stanton ML 1984 Seed variation in wild radish: effect of seed size on
components of seedling and adult fitness. Ecology 65:1105–1112.
Stelly DM, SJ Peloquin, RG Palmer, CF Crane 1984 Mayer’s
hemalum-methyl salicylate: a stain-clearing technique for observation within whole ovules. Stain Technol 59:155–161.
Walker RI 1944 Chromosome number, megasporogenesis and development of the embryo sac of Clintonia. Bull Torrey Bot Club 71:
529–535.
Waser NM 1993 Population structure, optimal outbreeding and as-
sortative mating. Pages 173–199 in NW Thornhill, ed. The natural
history of inbreeding and outbreeding: theoretical and empirical
perspectives. University of Chicago Press, Chicago.
Waser NM, MV Price 1991 Reproductive costs of self-pollination in
Ipomopsis aggregata (Polemoniaceae): are ovules usurped? Am J
Bot 78:1036–1043.
——— 1993 Crossing distance effects on prezygotic performance in
plants: an argument for female choice. Oikos 68:303–308.
——— 1994 Crossing distance effects in Delphinium nelsonii: outbreeding and inbreeding depression in progeny fitness. Evolution 48:
842–852.
Williams EG, V Kaul, JL Rouse, RB Knox 1984 Apparent selfincompatibility in Rhododendron ellipticum, R. championae and R.
amamiense: a post-zygotic mechanism. Plant Cell Incompatibility
Newsl 16:10–11.