SEARCH

SEARCH BY CITATION

Keywords:

  • gynodioecy;
  • inbreeding depression;
  • Silene vulgaris;
  • sex determination;
  • population structure

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Growing conditions
  6. Crossing design
  7. Fitness measurements and offspring gender
  8. Data analysis
  9. Results
  10. Offspring fitness
  11. Offspring gender
  12. Discussion
  13. Acknowledgments
  14. References

Abstract In gynodioecious plants, hermaphrodite and female plants co-occur in the same population. In these systems gender typically depends on whether a maternally inherited cytoplasmic male sterility factor (CMS) is counteracted by nuclear restorer alleles. These restorer alleles are often genetically dominant. Although plants of the female morph are obligatorily outcrossing, hermaphrodites may self. This selfing increases homozygosity and may thus have two effects: (1) it may decrease fitness (i.e. result in inbreeding depression) and (ii) it may increase homozygosity of the nuclear restorer alleles and therefore increase the production of females. This, in turn, enhances outcrossing in the following generation. In order to test the latter hypothesis, experimental crosses were conducted using individuals derived from four natural populations of Silene vulgaris, a gynodioecious plant. Treatments included self-fertilization of hermaphrodites, outcrossing of hermaphrodites and females using pollen derived from the same source population as the pollen recipients, and outcrossing hermaphrodites and females using pollen derived from different source populations. Offspring were scored for seed germination, survivorship to flowering and gender. The products of self-fertilization had reduced survivorship at both life stages when compared with the offspring of outcrossed hermaphrodites or females. In one population the fitness of offspring produced by within-population outcrossing of females was significantly less than the fitness of offspring produced by crossing females with hermaphrodites from other populations. Self-fertilization of hermaphrodites produced a smaller proportion of hermaphroditic offspring than did outcrossing hermaphrodites. Outcrossing females within populations produced a smaller proportion of hermaphrodite offspring than did crossing females with hermaphrodites from other populations. These results are consistent with a cytonuclear system of sex determination with dominant nuclear restorers, and are discussed with regard to how the mating system and the genetics of sex determination interact to influence the evolution of inbreeding depression.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Growing conditions
  6. Crossing design
  7. Fitness measurements and offspring gender
  8. Data analysis
  9. Results
  10. Offspring fitness
  11. Offspring gender
  12. Discussion
  13. Acknowledgments
  14. References

Gynodioecious plant species are those in which functionally female and hermaphroditic individuals co-occur in the same population. One condition thought to be necessary for the evolutionary maintenance of gynodioecy is that females must have a higher fitness through seed than hermaphrodites (e.g. Lewis, 1941; Lloyd, 1974; Charlesworth & Charlesworth, 1978; Frank, 1989; Couvet et al., 1990). Greater female fitness through seed can arise by two basic mechanisms. Females can either produce a larger quantity of seeds, often because of higher ovule number, or they can produce higher quality seeds. This latter case would be expected if the hermaphrodite morph was capable of self-fertilization. If so, seeds produced by self-fertilization would be more likely to suffer from inbreeding depression when compared with seeds produced by females, which must outcross. A female advantage in fitness through seed has been shown in a number of gynodioecious plant species, owing to either higher average seed production, or more vigorous offspring, or both (Frank, 1989; Couvet et al., 1990). Several recent studies have focused specifically on inbreeding depression following self-fertilization as an explanation of hermaphrodite – female fitness differences, and have shown an advantage to females that results from obligatory outcrossing (e.g. Maki, 1993; Sakai et al., 1997; Thompson & Tarayre, 2000).

Inbreeding in gynodioecious plants could also influence the maintenance of genetic load and the magnitude of inbreeding depression through its effect on offspring gender. Sex determination in gynodioecious plants is usually genetically based and can be either cytonuclear or purely nuclear (Frank, 1989; Couvet et al., 1990). In cytoplasmic systems a maternally inherited cytoplasmic male sterility (CMS) factor causes a plant to be functionally female, unless the effects of the CMS factor are countered by one or more restorer alleles present in the nuclear genome. In that case, male function is restored and the plant is a hermaphrodite. In both the cytonuclear and purely nuclear cases the coexistence of females and hermaphrodites requires genetic polymorphism at one or more of the nuclear loci that influence sex expression. One observation made of sex determination in gynodioecious species (whether nuclear or cytonuclear in nature) is that the nuclear alleles associated with the hermaphroditic morph are usually dominant to alleles associated with the female morph (e.g. Charlesworth & Laporte, 1998). This leads to the prediction that females should be more common in the offspring of hermaphrodites that self-fertilize, relative to the offspring of hermaphrodites that outcross. This is interesting because it could result in a negative feedback between the selfing rates in successive generations which, in turn, could reduce the rate that harmful recessive or semirecessive alleles are purged from the population.

In gynodioecious plants bi-parental inbreeding is also possible in either sex, especially when the species maintains a high degree of local population structure owing to limited gene flow. Indeed, several recent studies of gynodioecious and nongynodioecious plant species have shown increased vigour resulting from crossing among local populations, relative to outcrossing within populations (e.g. Van Treuren et al., 1993; Richards, 2000; Thompson & Tarayre, 2000; Keller & Waller, 2002). This result suggests that occasional gene flow events can have important demographic consequences for small or isolated populations and contribute to their persistence. It is also possible that bi-parental inbreeding within local gynodioecious populations influences the sex ratio through its effects on heterozygosity at the sex determining loci. If so, crossing among populations should, on average, produce both more vigorous offspring and a greater proportion of hermaphroditic offspring relative to outcrossing within populations. This prediction is counter to the suggestion in the literature that among-population crosses should produce a greater proportion of female offspring than within-population crosses owing to the mismatch of locally coadapted CMS factors and restorers (Gigord et al., 1998).

Understanding how inbreeding depression and the genetics of sex determination might interact to influence sex ratio evolution in gynodioecious species requires the evaluation of the magnitude of inbreeding effects on both offspring fitness and gender. However, these two phenomena are rarely evaluated simultaneously. This is a study of the consequences of inbreeding for offspring fitness and gender in Silene vulgaris, a gynodioecious plant known to have cytonuclear sex determination (Charlesworth & Laporte, 1998; Taylor et al., 2001). Previous studies have shown that hermaphrodites of this species are capable of self-fertilization and that the products of selfing exhibit reduced fitness, at least to the point of seed germination (Jolls & Chenier, 1989; Pettersson, 1992). An ongoing study conducted in south-west Virginia (USA) has revealed several features of the population biology of S. vulgaris relevant to questions relating to inbreeding, fitness and gender. The average selfing rate by hermaphrodites ranges from 0.19 to 0.61, depending on the population (Emery, 2001). In addition, S. vulgaris shows considerable variation among local Virginia populations in allozyme allele frequencies, chloroplast and mitochondrial DNA haplotype frequencies, and the local sex ratio (McCauley, 1998; McCauley et al., 2000; Taylor et al., 2001; Olson & McCauley, 2002). Females are known to enjoy higher fitness through seed than hermaphrodites, although this effect has been shown to depend on the local sex ratio (McCauley et al., 2000).

Here, we report on the results of three types of greenhouse crosses – self-fertilization of hermaphrodites, crosses of hermaphrodites or females with other hermaphrodites derived from the same local population, and crosses of hermaphrodites or females with hermaphrodites derived from nearby populations – in order to address two related questions. First, we focus on two components of fitness through seed, and ask how the fitness of individuals produced by selfing hermaphrodites compares with the fitness of females and hermaphrodites crossed with members of their own or other populations. Secondly, by determining the gender of the surviving offspring, we also test the hypothesis that self-fertilization increases the proportion of female offspring produced by hermaphrodites, as well as the hypothesis that crossing among populations increases the proportion of hermaphroditic offspring. In general, the effect of inbreeding on offspring gender is less well known than the effect of inbreeding on offspring fitness. This study represents an opportunity to investigate how the mating system interacts with the genetics of sex expression to influence the evolution of inbreeding depression.

Growing conditions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Growing conditions
  6. Crossing design
  7. Fitness measurements and offspring gender
  8. Data analysis
  9. Results
  10. Offspring fitness
  11. Offspring gender
  12. Discussion
  13. Acknowledgments
  14. References

Seeds were collected from four S. vulgaris populations found in the vicinity of Mountain Lake Biological Station (Giles County, VA, USA) during the summer of 1996. Natural populations of S. vulgaris found in this area are well-defined patches of 5–100+ individuals, separated from the nearest neighbouring population by between 100 m and several kilometres. Local sex ratios range from 75% female to 100% hermaphrodite (McCauley et al., 2000). The specific populations selected for this study span the range of sex ratios and population sizes found in this species (Table 1).

Table 1.  Properties of the natural populations of Silene vulgaris that were the source of the grandparents of the plants used in experimental crosses. All populations are located along roadsides in Giles or Craig Counties, Virginia, USA. Isolation refers to the distance to the nearest known neighbour population. Sex ratio refers to the proportion of individuals that were hermaphroditic.
DesignationPopulation sizeSex ratioIsolation
HO80.25  5 km
CH50+0.800.3 km
TA141.000.2 km
GS50+0.960.2 km

Five fruits were collected from up to 15 maternal plants in each of the four populations. Twenty seeds from each maternal plant were each planted in 10 cubic inch ‘Cone-tainers’ (Stuewe and Sons, Corvallis, OR, USA), and grown in the greenhouse at Vanderbilt University during the autumn of 1997. Conditions were maintained at near 24 h of light to encourage growth and flowering.

Crossing design

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Growing conditions
  6. Crossing design
  7. Fitness measurements and offspring gender
  8. Data analysis
  9. Results
  10. Offspring fitness
  11. Offspring gender
  12. Discussion
  13. Acknowledgments
  14. References

During the spring of 1998, two to seven plants were haphazardly selected from each of three to five maternal lines in each population and transferred to 4′′ pots. These plants (both females and hermaphrodites) were crossed with other plants from the same population but from different maternal lines. This kind of ‘outcrossing’ would minimize the degree of prior inbreeding because of selfing within the population of origin, but maintain other population structure effects. Crosses were repeated over a 3-month period until five to seven flowers on each plant were successfully setting seed. Maturation of seeds took approximately 2 weeks. When mature, fruits were collected and stored individually in small coin envelopes until the actual experimental crosses could be conducted.

In spring 1999, approximately seven seeds per fruit from this initial round of outcrossing were planted (one seed/Cone-tainer) and grown under the same conditions outlined above. Experimental crosses were begun in the autumn of 1999. At least three plants from each maternal line were randomly selected for participation in the three cross treatments: selfing of hermaphrodites (S), within population outcrosses (W), and among population outcrosses (A). For the outcross treatments, a pollen donor was determined by randomly selecting a hermaphrodite from a different population (treatment A), or a hermaphrodite from the same population (treatment W). Both hermaphrodite and female plants were used as pollen recipients in these two outcrossing treatments. For all three treatments five to seven flowers from each pollen recipient were fertilized by the same pollen donor. This resulted in one to seven fruits per mother. Under this design the genetic variation contained within the set of seeds originally collected from each natural population was distributed randomly among experimental cross treatments involving that population. Some mothers used in a particular Population × Cross combination shared one set of grandparents or one grandmother, but most were unrelated.

Fitness measurements and offspring gender

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Growing conditions
  6. Crossing design
  7. Fitness measurements and offspring gender
  8. Data analysis
  9. Results
  10. Offspring fitness
  11. Offspring gender
  12. Discussion
  13. Acknowledgments
  14. References

Seed germination and survivorship from germination to first flowering were assayed as measures of offspring fitness. In addition, a cumulative measure of fitness was based on the survivorship from planting until flowering. Seeds were planted in one of four blocks over a period of 8 months. One hundred fruits, selected at random from among all treatments, were included into each block. A maximum of 10 seeds per fruit were randomly selected and planted in smaller (3 cubic inch) Cone-tainers (one per Cone-tainer) and grown in the greenhouse using the same conditions described above. Plants were rearranged every other day to control for a position effect on the misting bench. Cone-tainers were checked every other day in order to monitor survivorship. The gender of all plants surviving to flowering was noted in order to test for the effects of cross type on offspring gender.

Data analysis

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Growing conditions
  6. Crossing design
  7. Fitness measurements and offspring gender
  8. Data analysis
  9. Results
  10. Offspring fitness
  11. Offspring gender
  12. Discussion
  13. Acknowledgments
  14. References

The main goals of this study were (1) to compare the products of self-fertilization with the products of within population outcrossing in terms of offspring fitness and gender (S vs. W comparisons), (2) to compare the products of within population outcrossing to the products of among population outcrossing in terms of offspring fitness and gender (A vs. W comparisons), and (3) to compare population specific responses to the cross treatments.

In order to accomplish these goals, three factor analysis of variance (anova) was used to evaluate the effects of cross treatment, population of maternal origin, and block on the relevant variables. All interaction effects were also calculated. Females and hermaphrodites were analysed separately in order to maintain a balanced design, as females could not be self-fertilized. However, in the GS population there were no female plants produced after the first round of outcrossing (Population GS contains nearly 100% hermaphrodites), and in the CH population only two females were produced. Therefore, only the TA and HO populations were included in the analysis of the progeny of females.

Each analysis was nested in that the individual mother was considered the unit of replication when testing Cross and Population effects, given that fruits within individuals are not independent observations, and recognizing that no individual mother was used in more than one Cross × Population combination. The Population and Cross treatments were considered fixed effects because crosses were under experimental control and populations were selected for use in this study based on the sex ratio and population size at the time that the original seeds were collected. Block and Individual were treated as random effects. anovas were calculated using JMP IN software and the restricted maximum likelihood (REML) option (SAS Institute, Sall et al., 2001; p. 336). In this design the error mean squares used to calculate F ratios for Cross and Population main effects and Cross–Population interactions are derived from variation among individuals within groups. The variables that were analysed were the proportion of each progeny array that survived to each life stage, or the proportion of flowering offspring scored as hermaphrodite. Proportions were transformed to arcsine square roots in order to achieve normality (Sokal & Rohlf, 1995).

For hermaphrodite mothers, two three-factor anovas were conducted for each variable. One included the S vs. W crosses and the other the W vs. A crosses. This approach aided in the interpretation of the Population–Cross interaction terms. The critical P-value for each test was reduced from the standard P = 0.05 to 0.025 to account for using the data from the W crosses in both analyses. For female pollen recipients only one anova was needed for each variable as there were only two cross treatments (A and W).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Growing conditions
  6. Crossing design
  7. Fitness measurements and offspring gender
  8. Data analysis
  9. Results
  10. Offspring fitness
  11. Offspring gender
  12. Discussion
  13. Acknowledgments
  14. References

The results are based on the progeny of 146 individual mothers (111 hermaphrodites and 35 females) distributed across treatments. Fitness components were measured in 3629 seeds, taken from a total of 392 fruits. Gender was assayed from the 2061 seeds that survived to flowering. Results are presented as treatment averages and standard errors in Figs 1–4. Statistical analysis is presented as factorial anovas (Tables 2–5). The summary of the results below will focus on Cross treatment main effects and Population–Cross interactions.

image

Figure 1. Mean and standard error (SE) proportion seed germination for individuals whose ancestry traces to one of four populations of Silene vulgaris . Results are presented for self-fertilization (S), within population outcrossing (W) and among population outcrossing (A) treatments. The number of individual mothers is given above the SE cross bar for each group. Results for hermaphrodite (a) and female (b) pollen recipients are presented separately.

Download figure to PowerPoint

image

Figure 2. Mean and standard error (SE) proportion surviving from seed germination to flowering for individuals whose ancestry traces to one of four populations of Silene vulgaris . Results are presented for self-fertilization (S), within population outcrossing (W) and among population outcrossing (A) treatments. The number of individual mothers is given above the SE cross bar for each group. Results for hermaphrodite (a) and female (b) pollen recipients are presented separately.

Download figure to PowerPoint

image

Figure 3. Mean and standard error (SE) proportion cumulative survival from seed to flowering for individuals whose ancestry traces to one of four populations of Silene vulgaris . Results are presented for self-fertilization (S), within population outcrossing (W) and among population outcrossing (A) treatments. The number of individual mothers is given above the SE cross bar for each group. Results for hermaphrodite (a) and female (b) pollen recipients are presented separately.

Download figure to PowerPoint

image

Figure 4. Mean and standard error (SE) proportion hermaphroditic offspring for individuals whose ancestry traces to one of four populations of Silene vulgaris . Results are presented for self-fertilization (S), within population outcrossing (W) and among population outcrossing (A) treatments. The number of individual mothers is given above the SE cross bar for each group. Results for hermaphrodite (a) and female (b) pollen recipients are presented separately.

Download figure to PowerPoint

Table 2.  Analysis of variance contrasting either the S & W crosses with hermaphrodite mothers, the A & W crosses with hermaphrodite mothers, or the A & W crosses with female mothers. The response variable is [arcsine v (per cent seed germination)]. Population (Pop) and Cross (CR) are fixed effects and Block (Blk) is a random effect.
SourceS & W (Herms.)A & W (Herms.)A & W (Females)
d.f.F -ratio d.f.F -ratio d.f.F -ratio
  1. A significant F-ratio is denoted by *(0.05 > P > 0.025 – significant for females only), or ***(P < 0.01).

Pop33.0630.4010.01
CR113.28***11.4810.40
Blk32.7731.8933.28*
Pop * CR31.1532.5610.06
Pop * Blk90.4290.6230.73
CR * Blk32.0430.1330.13
Pop * CR * Blk91.5590.8431.48
Individual610.25640.18311.03
Error90 57 67 
Table 3.  Analysis of variance contrasting either the S & W crosses with hermaphrodite mothers, the A & W crosses with hermaphrodite mothers, or the A & W crosses with female mothers. The response variable is [arcsine v (per cent survivorship between seed germination and flowering)]. Population (Pop) and Cross (CR) are fixed effects and Block (Blk) is a random effect.
SourceS & W (Herms.)A & W (Herms.)A & W (Females)
d.f.F -ratio d.f.F -ratio d.f.F -ratio
  1. A significant F-ratio is denoted by *(0.05 > P > 0.025 – significant for females only), **(0.025 > P > 0.01), or ***(P < 0.01).

Pop33.40**30.5915.23*
CR16.86**12.0113.04
Blk34.77***38.56***37.17***
Pop * CR31.2831.6117.86***
Pop * Blk90.7191.4931.26
CR * Blk30.8934.09**30.99
Pop * CR * Blk91.3392.99***30.04
Individual600.17620.05311.12
Error83 49 59 
Table 4.  Analysis of variance contrasting either the S & W crosses with hermaphrodite mothers, the A & W crosses with hermaphrodite mothers, or the A & W crosses with female mothers. The response variable is [arcsine v (per cent total survivorship to flowering)]. Population (Pop) and Cross (CR) are fixed effects and Block (Blk) is a random effect.
SourceS & W (Herms.)A & W (Herms.)A & W (Females)
d.f.F -ratio d.f.F -ratio d.f.F -ratio
  1. A significant F-ratio is denoted by *(0.05 > P > 0.025 – significant for females only), or ***(P < 0.01).

Pop36.76***30.5211.46
CR124.04***12.3310.40
Blk35.03***39.31***35.94***
Pop * CR31.6731.6214.81*
Pop * Blk90.7890.9730.86
CR * Blk31.9831.6630.16
Pop * CR * Blk92.1091.9230.30
Individual600.24620.18311.08
Error88 54 64 
Table 5.  Analysis of variance contrasting either the S & W crosses with hermaphrodite mothers, the A & W crosses with hermaphrodite mothers, or the A & W crosses with female mothers. The response variable is [arcsine v (per cent hermaphroditic offspring)]. Population (Pop) and Cross (CR) are fixed effects and Block (Blk) is a random effect.
SourceS & W (Herms.)A & W (Herms.)A & W (Females)
d.f.F -ratio d.f.F -ratio d.f.F -ratio
  1. A significant F-ratio is denoted by *(0.05 > P > 0.025 – significant for females only), **(0.025 > P > 0.01), or ***(P < 0.01).

Pop31.6832.3311.34
CR111.27***10.5415.72**
Blk30.0330.0230.32
Pop * CR31.1130.2910.40
Pop * Blk90.4092.3132.43
CR * Blk30.4230.8832.71
Pop * CR * Blk90.7790.7430.72
Individual600.18630.79311.03
Error88 54 64 

Offspring fitness

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Growing conditions
  6. Crossing design
  7. Fitness measurements and offspring gender
  8. Data analysis
  9. Results
  10. Offspring fitness
  11. Offspring gender
  12. Discussion
  13. Acknowledgments
  14. References

The effects of cross type on seed germination are presented in Fig. 1 and Table 2. The S vs. W Cross treatment had a significant main effect on the germination success of seeds from hermaphrodites. The Population–Cross interaction was not significant, as the germination rate of seeds produced by within population outcrossing was greater than that of those produced by self-fertilization in all populations. There was no significant difference between the W and A Cross treatment main effects for hermaphrodite mothers, although the Population–Cross interaction was borderline significant. The among population cross mean was greater than the within population cross mean in three of four populations. There was no significant difference in per cent seed germination for the Cross or Cross–Population interaction for females.

Seedling survivorship from germination to flowering is summarized in Fig. 2 and Table 3. Among hermaphrodites, Cross main effects were significant in the S vs. W comparison, but the Population–Cross interaction was not significant. Post-germination survivorship was consistently higher in the outcross treatment. In the W vs. A analysis, Cross main effects were not significant, nor was the Population–Cross interaction. For female pollen recipients the Cross main effects were not significant, but the Population–Cross interaction was highly significant. Among population outcrossing increased fitness when pollen recipients were from the HO population.

Total survivorship, calculated as the number of individuals that flowered divided by the number of seeds from that fruit that were planted, is summarized in Fig. 3 and Table 4. There was a significant Cross main effect for total survivorship in hermaphrodites when the S vs. W contrast was considered, but the Population–Cross interaction was not significant. Seedlings produced by self-fertilization had consistently lower overall survivorship than seedlings from within population crosses. However, the main effects of within and among population outcrossing (W vs. A) were not significantly different, nor was the Population–Cross interaction. In females, the Cross main effect was not significant, but the Population–Cross interaction was. Crossing females with hermaphrodites from other populations increased fitness of pollen recipients from HO, but not from the TA population.

Offspring gender

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Growing conditions
  6. Crossing design
  7. Fitness measurements and offspring gender
  8. Data analysis
  9. Results
  10. Offspring fitness
  11. Offspring gender
  12. Discussion
  13. Acknowledgments
  14. References

The effect of the Cross treatment on the sex ratio of offspring (proportion hermaphroditic) is summarized in Fig. 4 and Table 5. The S vs. W contrast was strongly significant for hermaphrodite mothers, and the Population–Cross interaction was not. Self-fertilized hermaphrodites had a consistently lower average proportion of hermaphrodite offspring than did the hermaphrodites that were crossed within populations. There was not a significant difference between the proportion of hermaphrodite offspring produced by the within- and among-population outcrossed hermaphrodites, nor was there a Population–Cross interaction. Female mothers who crossed outside their population produced a significantly higher proportion of hermaphroditic offspring than did those who crossed within their population of origin. As this effect was found when mothers were from either the TA or HO population, the Population–Cross interaction was not significant.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Growing conditions
  6. Crossing design
  7. Fitness measurements and offspring gender
  8. Data analysis
  9. Results
  10. Offspring fitness
  11. Offspring gender
  12. Discussion
  13. Acknowledgments
  14. References

The goal of this study was to compare the immediate effects of inbreeding, as evidenced by offspring fitness, with the delayed effect of inbreeding that would accrue if a greater production of female children increased the likelihood of outbred grandchildren. Inbreeding was evaluated in two ways, by comparing progeny produced by self-fertilization of hermaphrodites with the progeny of hermaphrodites or females who received nonself pollen from within their own population (the S vs. W treatment contrasts), and by comparing within and among population outcrossing (the W vs. A contrasts). The following discussion will first consider the selfing vs. within population outcrossing comparison and then the within vs. among population outcrossing comparison.

Self-fertilization of hermaphrodites had two effects. Not surprisingly, it reduced offspring fitness, relative to hermaphrodites that were crossed with other members of their population. It also produced a greater proportion of female offspring, relative to offspring produced by outcrossed hermaphrodites. These female offspring would presumably enjoy high fitness in the next generation owing to obligatory outcrossing. In our formal statistical analysis, females and hermaphrodites were analysed separately for logistical reasons. In order to interpret these results in terms of the long-term consequences of selfing, the fitness and gender of offspring produced by hermaphrodites must be further compared with those of offspring produced by females. In order to facilitate this, Table 6 presents a comparison of the ratio of the grand means for the three possible within population comparisons of self-fertilized hermaphrodites, outcrossed hermaphrodites, and outcrossed females with regard to the cumulative measure of inbreeding depression and to offspring gender.

Table 6.  Comparison of Cross treatment main effects calculated as relative means (see text). Comparisons are self-fertilized hermaphrodites (Sh) relative to hermaphrodites outcrossed within populations (Wh), as well as each of these treatment means relative to females outcrossed within populations (Wf). Comparisons are made for cumulative seed to flowering survivorship and for offspring gender (% hermaphrodite offspring).
 Cumulative survivorshipOffspring gender
Sh/Wh0.700.73
Wh/Wf0.931.54
Sh/Wf0.651.13

Clearly, the offspring of self-fertilized hermaphrodites are considerably less fit than those of females, and outcrossed hermaphrodites are more likely to produce hermaphroditic offspring than are females. Note from Table 6 that outcrossed hermaphrodites are nearly equivalent to females with regard to offspring fitness, whereas selfed hermaphrodites are nearly equivalent to females with regard to offspring gender. Thus, the negative consequences of self-fertilization in terms of offspring fitness are indeed somewhat tempered by an increased likelihood that more fit grandchildren will be produced when those female offspring outcross.

The increase in the proportion of female offspring of self-fertilized hermaphrodites has interesting implications with regard to the maintenance of inbreeding depression. Generally, one would expect that frequent self-fertilization would reduce the genetic load that contributes to inbreeding depression, because the increased frequency of expression of harmful recessive alleles in homozygotes exposes them to purifying selection. In gynodioecious species, the population-wide frequency of self-fertilization is limited by the presence of obligatorily outcrossing females. When self-fertilization increases the proportion of female offspring of hermaphrodite mothers, it should reduce the population-wide frequency of self-fertilization in the next generation. Thus, the consequences of self-fertilization for offspring gender could act as a self-limiting mechanism for purging of genetic load. This effect would be less important if selfing also limited the availability of pollen to females, i.e. if there was pollen discounting (Holsinger, 1988).

The most likely explanation for the increase in female offspring following self-fertilization of hermaphrodites would be a genetic control of sex expression in which the hermaphrodite phenotype is dominant to the female phenotype. As the genetic control of sex expression in S. vulgaris is cytonuclear (Charlesworth & Laporte, 1998; Taylor et al., 2001), this would imply that nuclear restorer alleles act in a dominant fashion. The increased proportion of recessive females in the offspring of self-fertilized hermaphrodites then requires that a substantial proportion of the hermaphrodites that acted as mothers in this study were heterozygous at the restorer loci, as might be expected, because there was a round of outcrossing before the experimental crosses. Still, the magnitude of the effect is surprising. Consider a model with a single restorer locus with a restorer allele dominant (R) to a nonrestorer allele (r). As females are rr, the only way a hermaphrodite × hermaphrodite cross can produce a female is by a Rr × Rr cross. Selfing increases the proportion of female offspring because it reduces the fraction of RR × Rr crosses. Proportionally, self-fertilization has the greatest impact on offspring gender when the hermaphrodite population consists of an equal proportion of RR and Rr genotypes. In that case one would expect selfing all hermaphrodites to produce 12.5% females (0.5 × 0.25). With random mating among hermaphrodites at the same genotypic proportions one would expect 6.25% females (0.5 × 0.5 × 0.25), or that self-fertilization results in a 6.25% difference in the proportion of hermaphrodite offspring. This is much smaller than the results shown here (Sh/Wh = 0.73, see Table 6). One explanation would be that nuclear restoration involves two or more nuclear restorer loci that interact epistatically.

It is also possible that differential mortality with regard to gender is related to inbreeding. This cannot be evaluated directly as gender cannot be determined until flowering. However, if there were differential mortality by gender at some point between seed and flowering, one might expect a negative correlation between survivorship within a sibship produced by a hermaphrodite mother and the proportion of the surviving individuals that are hermaphroditic. This was tested by calculating Kendall's nonparametric correlation between the proportion of seeds that survived to flowering and the proportion of the survivors that were hermaphroditic. No significant correlation was detected when hermaphrodite mothers from the S treatment were pooled across populations (τ = 0.079, P = 0.49, n = 39).

The discussion outlined above concerns the results of the crosses conducted with individuals from the same population. The experimental design also allowed a comparison of crosses within and among populations (the W vs. A contrasts). These crosses were motivated by the observation that the patchy distribution of S. vulgaris in Virginia results in a moderate amount of local population structure as measured by allozymes, chloroplast DNA and mitochondrial DNA (McCauley, 1998; Olson & McCauley, 2002). Thus, within population outcrosses might represent a type of bi-parental inbreeding when compared with among population crosses. If so, the prediction regarding offspring fitness would be that the products of among population crosses would be fitter than the products of within population crosses. This prediction was only partly supported, in that only the Population–Cross interaction was significant for female mothers when measures of fitness were considered (Tables 3 and 4). Inspection of Fig. 3 reveals that the interaction was significant because the offspring of females crossed within the HO population did poorly relative to the offspring of HO females crossed outside the population, or the offspring of TA females. Further inspection of Fig. 3 reveals that the offspring of hermaphrodites crossed within the HO population also did relatively poorly, although neither the Cross main effect on the interaction of Cross with Population were significant in this case. Note that the HO population was the smallest and most isolated of those studied (Table 1).

The expected consequences of within vs. among population outcrossing for offspring gender are less clear. It might seem that population structure would increase the likelihood that among population crosses produce hermaphrodites, as such crosses should increase heterozygosity. The observation that among population crossing increases offspring fitness, at least in some cases, would support this. However, the loci contributing to inbreeding depression need not be structured in the same way as the restorer loci. Further, with cytonuclear sex determination one must consider both the nuclear and cytoplasmic genomes. As mentioned above, it has been suggested that when different nuclear alleles are required to restore different CMS factors one might get cytonuclear coevolution on a very local spatial scale. In that case one would predict that among population crosses would result in an increase in the proportion of female offspring, because crosses between different coevolving systems might increase the likelihood of a mismatch between CMS factor and restorer (Gigord et al., 1998). Such a prediction might seem reasonable as there is evidence for spatial structuring of CMS factors in S. vulgaris (Taylor et al., 2001). The results presented here are in partial support of the first view, because crossing females with hermaphrodites from other populations increased the proportion of hermaphroditic, not female, offspring. It could be, however, that spatial variation in cytonuclear coevolution occurs on a spatial scale greater than the separation of populations used in this study or that the relatively small number of populations used in this study happened to contain the same CMS factor. It is not clear whether the two contrasting predictions are mutually exclusive, so perhaps the net effect of the two underlying mechanisms is often to cancel each other out with regard to offspring gender.

In conclusion, the consequences of self-fertilization by hermaphrodites must play an important role in both the evolution of the sex ratio in this gynodioecious system and in the evolution of inbreeding depression.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Growing conditions
  6. Crossing design
  7. Fitness measurements and offspring gender
  8. Data analysis
  9. Results
  10. Offspring fitness
  11. Offspring gender
  12. Discussion
  13. Acknowledgments
  14. References

We thank Chris Richards and Matt Olson for their advice and support throughout this project and Jonathan Ertelt for his expertise in the greenhouse. Mountain Lake Biological Station provided logistical support for the field portion of the study. Financial support was provided by National Science Foundation (USA) grants DEB-9610496 and DEB-0078531 awarded to DEM. The results reported here are in partial fulfilment of the requirements for a Master's of Science degree awarded by Vanderbilt University to SNE.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Growing conditions
  6. Crossing design
  7. Fitness measurements and offspring gender
  8. Data analysis
  9. Results
  10. Offspring fitness
  11. Offspring gender
  12. Discussion
  13. Acknowledgments
  14. References
  • Charlesworth, B. & Charlesworth, D. 1978. A model for the evolution of dioecy and gynodioecy. Am. Natur. 112: 975997.
  • Charlesworth, D. & Laporte, V. 1998. The male-sterility polymorphism of Silene vulgaris: analysis of genetic data from two populations and comparison with Thymus vulgaris. Genetics 150: 12671282.
  • Couvet, D., Atlan, A., Belhassen, E., Gliddon, C., Gouyon, P.H. & Kjellberg, F. 1990. Co-evolution between two symbionts: the case of cytoplasmic male-sterility in higher plants. Oxford Surv. Evol. Biol. 7: 225250.
  • Emery, S.N. 2001. Inbreeding Depression and its Consequences in Silene vulgaris. Masters Thesis. Vanderbilt University, TN, USA.
  • Frank, S.A. 1989. The evolutionary dynamics of cytoplasmic male sterility. Am. Natur. 133: 345376.
  • Gigord, L., Lavigne, C., Shykoff, J.A. & Atlan, A. 1998. No evidence for local adaptation between cytoplasmic male sterility and nuclear restorer genes in the gynodioecious species Thymus vulgaris L. Heredity 81: 156163.
  • Holsinger, K.E. 1988. The evolution of self-fertilization in plants: lessons from population genetics. Acta Oecol. 9: 95102.
  • Jolls, C.L. & Chenier, T.C. 1989. Gynodioecy in Silene vulgaris (Caryophyllaceae): progeny success, experimental design and maternal effects. Am. J. Bot. 76: 13601367.
  • Keller, L.F. & Waller, D.M. 2002. Inbreeding effects in wild populations. Trends Ecol. Evol. 17: 230241.
  • Lewis, D. 1941. Male sterility in natural populations of hermaphroditic plants. New Phytol. 40: 5663.
  • Lloyd, D.G. 1974. Theoretical sex-ratios of dioecious and gynodioecious angiosperms. Heredity 32: 1134.
  • Maki, M. 1993. Outcrossing and fecundity advantage of females in gynodioecious Chionographis japonica var. kurohimensis (Liliaceae). Am. J. Bot. 80: 629634.
  • McCauley, D.E. 1998. The genetic structure of a gynodioecious plant: nuclear and cytoplasmic genes. Evolution 52: 255260.
  • McCauley, D.E., Olson, M.S., Emery, S.E. & Taylor, D.R. 2000. Population structure influences sex ratio evolution in a gynodioecious plant. Am. Natur. 155: 815819.
  • Olson, M.S. & McCauley, D.E. 2002. Mitochondrial DNA diversity, population structure, and gender association in the gynodioecious plant Silene vulgaris. Evolution 56: 253262.
  • Pettersson, M.W. 1992. Advantages of being a specialist female in nodioecious Silene vulgaris S.L. (Caryophyllaceae). Am. J. Bot. 79: 13891395.
  • Richards, C.M. 2000. Inbreeding depression and genetic rescue in a plant metapopulation. Am. Natur. 155: 383394.
  • Sakai, A.K., Weller, S.G., Chen, M.-L., Chou, S.Y. & Tasanont, C. 1997. Evolution of gynodioecy and maintenance of females: the role of inbreeding depression, outcrossing rates, and resource allocation in Schiedea adamantis (Caryophyllaceae). Evolution 51: 724736.
  • Sall, J., Lehman, A. & Creighton, L. 2001. JMP Start Statistics, 2nd edn. Duxbury Press, Pacific Grove, CA, USA.
  • Sokal, R.R. & Rohlf, F.J. 1995. Biometry: the Principles and Practice of Statistics in Biological Research, 3rd edn. W.H. Freeman, New York, NY, USA.
  • Taylor, D.R., Olson, M.S. & McCauley, D.E. 2001. A quantitative genetic analysis of nuclear-cytoplasmic male sterility in structured populations of Silene vulgaris. Genetics 158: 833841.
  • Thompson, J.D. & Tarayre, M. 2000. Exploring the genetic basis and proximate causes of female fertility advantage in gynodioecious Thymus vulgaris. Evolution 54: 15101520.
  • Van Treuren, R., Bulsma, R., Ouborg, N.J. & Van Delden, W. 1993. The significance of genetic erosion in the process of extinction. IV. Inbreeding depression and heterosis effects caused by selfing and outcrossing in Scabiosa columbaria. Evolution 47: 16691680.