SEARCH

SEARCH BY CITATION

Keywords:

  • gynodioecy;
  • Lobelia siphilitica;
  • natural selection;
  • population size;
  • sex ratio

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Variation in population sex ratio can be influenced by natural selection on alternate sex phenotypes as well as nonselective mechanisms, such as genetic drift and founder effects. If natural selection contributes to variation in population sex ratio, then sex ratio should covary with resource availability or herbivory. With nonselective mechanisms, sex ratio should covary with population size. We estimated sex ratio, resource availability, herbivory and size of 53 populations of gynodioecious Lobelia siphilitica. Females were more common in populations with higher annual temperatures, lower soil moisture and lower predation on female fruits, consistent with sex-specific selection. Females were also more common in small populations, consistent with drift, inbreeding or founder effects. However, small populations occurred in areas with higher temperatures than large populations, suggesting that female frequencies in small populations could be caused by sex-specific selection. Both selective and nonselective mechanisms likely affect sex ratio variation in this gynodioecious species.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Population sex ratio influences reproductive success and thus the evolution of alternate sex phenotypes. Because all offspring have one mother and one father, any deviation from a 1:1 population sex ratio will cause frequency-dependent natural selection to favour individuals who produce more of the rarer sex (Fisher, 1930; Lloyd, 1976; Charnov, 1982). Population sex ratio can also influence patterns of natural selection on traits related to reproductive success (Carroll & Corneli, 1995; Ashman & Diefenderfer, 2001). For example, selection on traits that increase individual male mating success should be stronger in populations with a high frequency of males because of high intrasex competition. Finally, population sex ratio can directly influence the mating success of different sex morphs. In sexually polymorphic species, individuals have to mate with the opposite sex. Such enforced disassortative mating affects the genetic architecture of populations (Gouyon & Couvet, 1987; Cuevas et al., 2006), which may alter rates and patterns of evolution in response to natural selection (e.g. Holeski & Kelly, 2006).

Variation in sex ratio among populations is expected to reflect the relative fitness (survival and/or reproduction) of alternate sex phenotypes within populations. Sex-specific natural selection has been extensively studied in gynodioecious plant species, in which there are female and hermaphrodite individuals. Abiotic resource availability and the presence of natural enemies have been implicated as important determinants of the relative fitness of females in many gynodioecious plants (reviewed in Ashman, 2006). Fitness of females is frequently higher than that of hermaphrodites in hot, dry or nutrient-poor habitats. This may be because when resources are limited, hermaphrodites are unable to maintain both pollen and seed production, but females are able to reproduce better under stressful abiotic conditions. Alternatively, resource limitation may intensify inbreeding depression, which would decrease hermaphrodite maternal fitness relative to females. Relative fitness can also be influenced by natural enemies, such as herbivores, whose damage is sex biased (Ashman, 2002; Cornelissen & Stiling, 2005; Ashman, 2006). If the herbivore is a seed predator, then it can directly affect seed set of females and hermaphrodites (e.g. Marshall & Ganders, 2001). If differential natural selection on the sex phenotypes caused by abiotic resource availability or natural enemies is contributing to variation in sex ratio, then both the relative fitness and frequency of females should vary along ecological gradients (Ashman, 2006).

Although the relative fertility of the sexes is the primary determinant of population sex ratio in species with nuclear sex determination, the causes of sex ratio variation may be more complex in gynodioecious species with nucleo-cytoplasmic sex determination. In these species, sex determination involves epistatic interactions between cytoplasmic genes for male sterility (CMS) and nuclear genes that restore male fertility (Rf; Lewis, 1941; Kaul, 1988). Individuals who do not carry CMS genes develop as hermaphrodites. Individuals who carry CMS genes but do not carry an appropriate Rf allele develop as females. However, individuals who carry CMS genes can develop as hermaphrodites if they carry an appropriate Rf allele. In these systems, equilibrium sex ratios depend on the relative frequencies of CMS and Rf alleles and the diversity of CMS/Rf systems within demes. Allele frequencies are predicted to vary cyclically (Frank, 1989; Gouyon et al., 1991) and to be highly structured within and between populations, generating complex multilevel selection that can result in extreme sex ratio variation among demes or populations (Olson et al., 2005).

For nucleo-cytoplasmic gynodioecious species, the importance of sex-specific natural selection in determining population sex ratio in the wild is uncertain. Modelling suggests that interactions between Rf and CMS genes can cause cyclic variation in population sex ratio independent of natural selection caused by ecological factors, such as abiotic resource availability and natural enemies (Gouyon et al., 1991). If so, then populations may take a long time to reach an equilibrium sex ratio, and the relationship between relative fitness of the sexes and the population sex ratio at any given time may be weak (e.g. Barr, 2004). Empirical data on the relationship between ecological factors, seed set of females and hermaphrodites and population sex ratio are mixed. For example, seed set of Nemophila menziesii females relative to seed set of hermaphrodites decreases when plants are watered. However, population sex ratio and soil water availability are not correlated with each other, suggesting that the effect of soil moisture on relative fertility of N. menziesii does not translate into an effect on population sex ratio (Barr, 2004). By contrast, relative seed set of females is correlated with population sex ratio in Plantago maritima, suggesting that sex-specific natural selection may contribute to sex ratio variation in this species (Nilsson & Ågren, 2006). Because the relative fertility of the sexes does not always correlate with the population sex ratio in nucleo-cytoplasmic gynodioecious species, it is likely that both selective and nonselective mechanisms are influencing sex ratio.

Nonselective mechanisms such as genetic drift, inbreeding, gene flow or founder effects can also contribute to population sex ratio variation by altering frequencies of CMS/Rf alleles (e.g. Barr, 2004; Nilsson & Ågren, 2006). If hermaphrodites are heterozygous at dominant Rf loci, then increased inbreeding should result in hermaphrodites producing more homozygous female offspring (Emery & McCauley, 2002; Bailey & McCauley, 2005). This will increase female frequency regardless of whether inbreeding depression is exposed. Genetic drift can either increase (Byers et al., 2005) or decrease (Murayama et al., 2004) female frequency, depending on whether CMS or Rf alleles are lost to drift. Finally, founder effects can increase female frequency if the appropriate Rf alleles are absent from newly founded populations (e.g. Manicacci et al., 1996). Founder effects should increase female rather than hermaphrodite frequency for three reasons. First, because females often have higher seed set than hermaphrodites (Shykoff et al., 2003; Case & Ashman, 2005), population founders are statistically more likely to have female mothers. Second, although only self-compatible hermaphrodites can found populations on their own, if they are offspring of female mothers they are likely to produce females upon selfing (Dudle et al., 2001; Emery & McCauley, 2002; Bailey & McCauley, 2005). Third, small, newly established populations are unlikely to have a full complement of Rf alleles for CMS types present in the founders, even if there is more than one founder (Nilsson & Ågren, 2006).

If inbreeding, genetic drift and/or founder effects are contributing to variation in population sex ratio, then female frequency should be higher in small populations than in large populations (Nilsson & Ågren, 2006). For example, fewer mates may be available in small populations, resulting in increased inbreeding (Ellstrand & Elam, 1993) and higher female frequencies if Rf alleles are dominant (Emery & McCauley, 2002; Bailey & McCauley, 2005). Genetic drift should have a proportionately greater effect on allele frequencies in small than in large populations (Ellstrand & Elam, 1993). Assuming that only Rf alleles are lost, increased drift should result in females being more common in small populations (Byers et al., 2005). Founder effects should result in higher female frequencies in small populations if populations are small because they are newly founded (Manicacci et al., 1996). Although the relationship between population size and female frequency can be used to infer whether nonselective mechanisms are contributing to variation in population sex ratio, it has been established in few gynodioecious species (Nilsson & Ågren, 2006). Of those that have been studied, females are more common in small populations of some species (P. maritima, Nilsson & Ågren, 2006) but not others (Silene vulgaris, Olson & McCauley, 2002; Geranium sylvaticum, Asikainen & Mutikainen, 2003). However, if abiotic resource availability or natural enemies differ between large and small populations, then any variation in female frequency with population size could be caused by natural selection rather than inbreeding, drift or founder effects.

To determine the contribution of natural selection and nonselective mechanisms to variation in population sex ratio, we studied populations of gynodioecious Lobelia siphilitica L. (Lobeliaceae). Sex determination in L. siphilitica is nucleo-cytoplasmic (Dudle et al., 2001) and female frequency can vary from 0% to > 90% (Mutikainen & Delph, 1998; Dudle, 1999). Dudle (1999) surveyed 14 L. siphilitica populations and found that females were more common in the southern portion of the species range. This latitudinal cline suggests that population sex ratio varies along an ecological gradient in L. siphilitica, as would be expected if natural selection contributes to variation in population sex ratio. However, it is not known if abiotic resource availability and natural enemies also vary latitudinally. It is also not known whether female L. siphilitica are more common in small populations, as would be expected if inbreeding, drift and/or founder effects influence variation in sex ratio.

We used estimates of population size and sex ratio, as well as abiotic resource availability and natural enemies, to answer the following questions:

  • 1
    Are female L. siphilitica more common in the southern portion of the species range?
  • 2
    Are females more common in populations with lower abiotic resource availability or less female-biased herbivory?
  • 3
    Are females more common in small populations, and if so does abiotic resource availability or herbivory differ between large and small populations?

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Study species and sites

Lobelia siphilitica is a short-lived, herbaceous perennial wildflower that grows in wet meadows and woods throughout eastern North America (Johnston (1991a) and references therein). Its 3-cm-long blue flowers are primarily pollinated by Bombus spp. (Beaudoin Yetter, 1989). Plants in natural populations flower from early August until early October and fruits dehisce from mid-September to November (C.M. Caruso and A.L. case, personal observation). Hermaphroditic flowers are protandrous and pollen is shed from a tube formed by the fused anthers and filaments (Johnston, 1991a). Although L. siphilitica is self-compatible (Johnston, 1992), the complete separation between staminate and pistillate phases of flower development reduces the opportunity for autogamous self-fertilization in hermaphrodites (Johnston, 1991b). Lobelia siphilitica can also reproduce clonally via rosettes that overwinter and produce a flowering stalk the following summer (Beaudoin Yetter, 1989).

Data collection

We sampled 53 L. siphilitica populations from across a large portion of the species’ range (Appendix S1 in the Supplementary Material; Fig. 1). Populations were located based on unpublished theses (Beaudoin Yetter, 1989; Dudle, 1999), personal communications, personal observations, herbarium specimens and WWW searches. Most populations (38) were located at sites in which we could not find any other L. siphilitica. Given that our sampling was performed in fragmented agricultural landscapes, the probability of seed and pollen dispersal between these populations should be low. The remaining 15 populations were located at sites with multiple L. siphilitica populations. At these sites, we defined a population as a discrete patch of plants that was not visible from other patches and was separated by habitat in which we could not find any L. siphilitica. When in doubt as to whether patches of plants constituted a single population or not, we considered them to be a single population. During August and September 2005, we visited all populations and estimated female frequency. If there were fewer than 100 flowering plants in the population, we sexed all of the individuals. If there were more than 100 flowering plants in the population, we generally sexed the first 100 individuals that we encountered. Female frequency was calculated as the proportion of the sexed plants that were female. Because we did not count the total number of plants in all of the large populations, we treat population size as a categorical rather than a continuous variable. Specifically, we categorized populations with n < 100 as small and those with n > 100 as large.

image

Figure 1.  Location and sex ratio of the 53 L. siphilitica populations included in this study. The fraction of hermaphrodites is in grey and the fraction of females is in black. (a) Large (n > 100) populations. (b) Small (n < 100) populations.

Download figure to PowerPoint

We measured soil moisture, a common correlate of sex ratio variation (e.g. Ashman, 1999), in a subset of our L. siphilitica populations. Volumetric water content (VWC) of the soil was measured using a soil moisture meter (Hydrosense C620, Campbell Scientific, Logan, UT, USA) during flowering (2–28 August 2005) and fruiting (28 September and 6–8 October 2005). We measured VWC at the base of ∼30 haphazardly chosen plants. However, we measured fewer individuals in small populations. This protocol was used to estimate VWC of 31 populations during flowering and 15 populations during fruiting. The populations measured during fruiting were a subset of those measured during flowering. Although measuring VWC of a population at a single time point during flowering and/or fruiting will not capture spatial or temporal variation in VWC, we were constrained by the number of populations sampled and the large distances between populations. However, VWC during flowering and fruiting were strongly correlated in L. siphilitica (r13 = 0.906, P < 0.001), suggesting that relative differences in VWC between populations are consistent across time.

Because herbivory can also influence the relative fitness of the sexes in gynodioecious species (Ashman, 2006), we quantified the intensity of predispersal seed predation by the weevil Cleopmiarus hispidulus LeConte (Coleoptera: Curculionidae) in 25 of our L. siphilitica populations. Cleopmiarus hispidulus is native to eastern North America (Anderson, 1964) and also colonizes Lobelia inflata and L. cardinalis (Anderson, 1973). We have documented C. hispidulus attacking L. siphilitica in Illinois, Indiana, Iowa, Michigan, New York, North Carolina, Ohio, Ontario, Wisconsin and West Virginia (this study; Parachnowitsch, 2005; C. M. Caruso and A. L. Case, unpublished data). However, the proportion of L. siphilitica plants (r23 = 0.288, P = 0.163) and fruits (r23 = −0.049, P = 0.814) within a population that are attacked by C. hispidulus is not correlated with latitude, suggesting that weevil abundance does not vary geographically. In L. siphilitica, the female weevil chews a hole in the ovary wall and deposits a single egg. The ensuing larva feeds on ovules and developing seeds, pupates within the fruit and emerges in the fall after chewing an exit hole or upon fruit dehiscence. Larvae and pupae from L. siphilitica populations in Iowa, New York and Ontario are attacked by an unidentified parasitoid wasp. Adults feed on pollen and chew through the corolla tube to consume nectar (Parachnowitsch, 2005).

To quantify the intensity of seed predation by C. hispidulus, we collected five fruits from 15 haphazardly chosen females and 15 haphazardly chosen hermaphrodites per population; we collected fruits from fewer plants in small or hermaphrodite-dominated populations, or when a marked plant was destroyed prior to fruit set. Fruits were dissected to determine whether they had been damaged by C. hispidulus. Damage was indicated by the presence of C. hispidulus, frass, exit holes and/or the complete absence of seeds (as in Parachnowitsch, 2005). We used these data to estimate the proportion of hermaphrodite and female plants in each population that had at least one fruit attacked and the mean proportion of fruits attacked on affected females and hermaphrodites.

We recorded GPS locations for our populations (Appendix S1) and extracted climate data for these locations from the WorldClim database (described in Hijmans et al., 2005). This database has a spatial resolution of 1 km and is interpolated from weather station measurements of temperature and precipitation. Because females often have higher fitness than hermaphrodites when growing in hot, dry environments (Ashman, 2006), we extracted estimates of the annual mean temperature and the annual precipitation for each population.

Statistical analysis

We used analysis of covariance (ancova) to assess the effects of population size, latitude and longitude on the sex ratio of L. siphilitica populations. Population size was included as a fixed categorical variable, whereas latitude and longitude were included as covariates. The dependent variable was proportion females. For this and all other ancovas presented in the paper, we tested the assumption of homogeneity of residual variance using Levene's test. We tested the assumption of normality of residual variance using Lilliefor's test (Wilkinson, 1997). When these assumptions were not met, the dependent variable was arcsine-square root or log-transformed. To estimate the goodness of fit of the model, we present the adjusted coefficient of determination (r2) for this and all other ancovas. The adjusted coefficient of determination increases when a new independent variable improves the fit of the ancova model to the data (Zar, 1999).

We used ancova to test whether variation in temperature, precipitation and predation by C. hispidulus correlated with latitudinal variation in the proportion of females in L. siphilitica populations (Fig. 2). A correlation between phenotype and environment can suggest natural selection, although any correlational relationship should be experimentally verified (Endler, 1986). Population size was included as a fixed categorical variable and latitude and longitude were included as covariates. Because we measured the ecological variables on different subsets of our 53 populations, we could not include them all as covariates in a single ancova without greatly reducing sample size. However, two sets of ecological variables (annual mean temperature/annual precipitation and proportion hermaphrodite plants attacked by C. hispidulus/mean proportion hermaphrodite fruits attacked by C. hispidulus) were measured in the same populations. Consequently, we ran four ancova models. Two of these models included two ecological variables each (annual mean temperature/annual precipitation or proportion hermaphrodite plants attacked by C. hispidulus/mean proportion hermaphrodite fruits attacked by C. hispidulus). The other two models each included one ecological variable (proportion of female plants attacked or mean proportion of fruits attacked on females) as a covariate. If an ecological variable had a significant effect on proportion females, but latitude also remained significant, then we concluded that the ecological variable correlated with nonlatitudinal variation in the population sex ratio. If an ecological variable had a significant effect on sex ratio, but latitude was no longer significant, then we concluded that the ecological variable contributed to latitudinal variation in the population sex ratio.

image

Figure 2.  Plot of latitude vs. sex ratio for large (> 100 individuals, solid regression line) and small (< 100 individuals, dashed regression line) L. siphilitica populations (n = 53).

Download figure to PowerPoint

We used a combination of parametric and nonparametric analyses to test whether variation in VWC of the soil correlated with latitudinal variation in sex ratio of L. siphilitica populations. We used the same ancova model as above to analyse populations sampled during flowering, except that VWC, rather than climate or herbivory, was included as a covariate. The interpretation of this ancova model is the same as for the analyses of climate and herbivory variables. Because VWC during fruiting was heteroscedastic even after transformation, we used a nonparametric Spearman rank correlation to test for a relationship between this variable and proportion females. Consequently, we cannot determine whether VWC during fruiting is correlated with the latitudinal versus nonlatitudinal component of variation in sex ratio among L. siphilitica populations.

We used t-tests to determine whether the ecological variables that did have a significant effect on sex ratio (annual mean temperature, VWC during fruiting and mean proportion of female fruits attacked; Fig. 3) also differed between small and large populations. Population size class (large vs. small) was the independent variable and one of the ecological variables was the dependent variable. We tested the assumption of homogeneity of residual variance using Levene's test. If this assumption was violated, we used a separate-variance t-test. We tested the assumption of normality of residual variance using Lilliefor's test (Wilkinson, 1997).

image

Figure 3.  Relationship between three ecological variables and sex ratio of L. siphilitica populations. (a) Annual mean temperature, (b) volumetric water content (VWC) during fruiting and (c) mean proportion female fruits attacked by C. hispidulus. n = 12–53.

Download figure to PowerPoint

We ran an additional ancova to verify that the annual mean temperature (Table 2; Fig. 3a) correlated with latitudinal variation in population sex ratio (Fig. 2), as well as to verify that population size influenced sex ratio even when the annual mean temperature was included in the model (Tables 1–2). In this model, population size was categorical, the longitude and the annual mean temperature were covariates and proportion females was the dependent variable. We compared this model with the ancova with population size as a fixed categorical variable and latitude and longitude as covariates (Table 1), as well as with the model where both the annual mean temperature and the latitude were included as covariates (Table 2).

Table 2.   Effect of two climate, one soil moisture (VWC) and four herbivory variables on sex ratio of L. siphilitica populations.
Ecological variableSourceMSFP
  1. The effect of these ecological variables on the sex ratio was tested using ancova. Population size was the fixed categorical variable. Latitude, longitude and one or more ecological variables were covariates. d.f.num, d.f.denom = 1, 43 for climate variables; 1, 23 for VWC; 1, 4–14 for herbivory by Cleopmiarus hispidulus.

Model 1: Annual mean temperature and annual precipitation (r2 = 0.459)Population size1.27 × 10−94.37 × 10−70.999
Latitude0.0010.3590.552
Longitude0.0030.9580.333
Temperature0.0124.0140.050
Precipitation0.0010.1870.668
Latitude × Pop. size1.53 × 10−40.0530.820
Longitude × Pop. size0.0041.2710.266
Temp. × Pop. size0.0020.5270.472
Precip × Pop. size1.50 × 10−40.0510.822
Model 2: Soil VWC during flowering (r2 = 0.416)Population size0.1002.6470.117
Latitude0.2316.0920.021
Longitude0.1042.7350.112
Soil VWC0.0010.0220.884
Latitude × Pop. size0.3048.0210.009
Longitude × Pop. size0.1443.8070.063
VWC × Pop. size0.0240.6450.430
Model 3: Proportion hermaphrodite plants attacked and mean proportion hermaphrodite fruits attacked by Cleopmiarus hispidulus (r2 = 0.270)Population size0.1101.9070.189
Latitude0.1642.8350.114
Longitude0.0100.1720.685
Prop. plants attacked0.0190.3350.572
Prop. fruits attacked0.0240.4170.529
Latitude × Pop. size0.2133.6770.076
Longitude × Pop. size0.0010.0130.911
Plants attacked × Pop. size0.0100.1750.682
Fruits attacked × Pop. size0.0020.0360.852
Model 4: Proportion female plants attacked by C. hispidulus (r2 = 0.261)Population size0.0370.6250.459
Latitude0.1192.0340.204
Longitude0.0430.7280.426
Prop. attacked0.0220.3780.561
Latitude × Pop. size0.0370.6270.459
Longitude × Pop. size0.0020.0280.873
Attack × Pop. size0.0030.0590.816
Model 5: Mean proportion female fruits attacked by C. hispidulus (r2 = 0.805)Population size0.0032.9820.159
Latitude1.19 × 10−40.1150.752
Longitude0.0021.8720.243
Prop. attacked0.01010.0230.034
Latitude × Pop. size0.0043.9290.119
Longitude × Pop. size2.35 × 10−40.2270.859
Attack × Pop. size0.0087.5330.052
Table 1.   Effect of latitude, longitude and population size on sex ratio of L. siphilitica populations.
SourceMSFP
  1. The effect of these variables on the sex ratio was tested using ancova. Population size was a fixed categorical variable. Latitude and longitude were covariates. r2 = 0.417. d.f.num, d.f.denom = 1, 46.

Population size0.0195.9450.019
Latitude0.04815.213< 0.001
Longitude0.0092.7280.105
Latitude × Pop. size0.03310.3810.002
Longitude × Pop. size0.0082.4030.128

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Latitude, population size and latitude × population size all influenced the sex ratio of L. siphilitica populations (Table 1). Female frequency varied from 0% to 100% among L. siphilitica populations. Populations at lower latitudes had a significantly higher proportion of females, as well as higher variance in female frequency (Figs 1 and 2). In contrast, longitude did not influence the population sex ratio (Table 1; Fig. 1). The proportion of females in small populations [mean ± 1 SE, n = 0.19 ± 0.05 (32)] was 2.5 times higher than in large populations [0.07 ± 0.02 (21); Table 1]. In addition, the relationship between sex ratio and latitude differed for large versus small populations (population size × latitude interaction), but this was not the case for longitude (Table 1). Small populations at lower latitudes had a higher proportion of females (Figs 1b and 2), whereas the sex ratio of large populations varied little with latitude (Figs 1a and 2).

Annual mean temperature, but not annual precipitation, correlated with latitudinal variation in the population sex ratio. Populations with a higher proportion of females occurred in locations with higher annual mean temperature (Fig. 3a). In contrast, annual precipitation did not have a significant effect on population sex ratio (Table 2). Two analyses suggest that temperature is an ecological factor underlying latitudinal variation in sex ratio. First, in the ancova model with both annual mean temperature and latitude as independent variables, temperature had a significant effect on the population sex ratio, but latitude did not (Table 2). Second, annual mean temperature also had a significant effect on population sex ratio in the ancova model from which latitude was excluded, as did population size and population size × annual mean temperature (Table 3).

Table 3.   Effect of annual mean temperature, longitude and population size on sex ratio of L. siphilitica populations.
SourceMSFP
  1. The ancova model is the same as Table 1, except that annual mean temperature was substituted for latitude. r2 = 0.500. d.f.num, d.f.denom = 1, 46.

Population size0.1897.6640.008
Temperature0.58223.628< 0.001
Longitude0.0371.5180.224
Temp. × Pop. size0.26910.9200.002
Longitude × Pop. size0.0692.7830.102

Volumetric water content of the soil during fruiting, but not during flowering, had a significant effect on the sex ratio of L. siphilitica populations. Populations with a higher proportion of females occurred in areas with lower VWC during fruiting (Spearman's r13 = −0.727, P < 0.01; Fig. 3b). By contrast, VWC during flowering did not have a significant effect on the population sex ratio (Table 2).

One of four herbivory variables correlated with latitudinal variation in population sex ratio of L. siphilitica. Populations with a higher proportion of females had a smaller mean proportion of fruits attacked on female plants (Fig. 3c). By contrast, the proportion of hermaphrodite plants, female plants and hermaphrodite fruits attacked by the predispersal seed predator C. hispidulus did not influence the population sex ratio (Table 2). In the ancova model with the mean proportion of fruits attacked on female plants and latitude as independent variables, only herbivory had a significant effect on the population sex ratio (Table 2). This suggests that the intensity of herbivory on fruits of female plants is another ecological variable contributing to latitudinal variation in population sex ratio.

One of the three ecological variables that influenced the population sex ratio also differed significantly between small and large populations. Annual mean temperature was ∼0.7 °C higher in small than in large populations (t51 = 2.275, P = 0.027). VWC during fruiting (t9.428 = 1.132, P = 0.214) and the mean proportion of female fruits attacked (t10 = 0.262, P = 0.799) did not differ between small and large populations.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Our data suggest that both natural selection and nonselective mechanisms contribute to variation in sex ratio among L. siphilitica populations. Consistent with the hypothesis that differential natural selection on the sex phenotypes contributes to variation in population sex ratio (Ashman, 2006), female frequency of L. siphilitica varied with abiotic resource availability and the intensity of attack by natural enemies. Females were more common in southern L. siphilitica populations where the annual mean temperature was higher and there was less seed predation on female plants by C. hispidulus. Female L. siphilitica were also more common in populations with lower soil moisture. Consistent with the hypothesis that inbreeding, genetic drift and/or founder effects can contribute to variation in population sex ratio, females were 2.5 times more common in small than in large L. siphilitica populations.

Although higher female frequencies in small populations may be the result of nonselective mechanisms, we also found evidence that sex-specific natural selection contributes to differences in female frequency between large and small populations. Specifically, small L. siphilitica populations occur in areas where the annual mean temperature is ∼0.7 °C higher. This difference in annual mean temperature represents 13.4% of the range (6.66–11.88 °C) of annual mean temperatures observed across our 53 populations. Although we used a t-test to detect differences in ecological variables between population size classes, the difference in temperature between small and large populations remains significant in an ancova with latitude included as a covariate (F1,50 = 4.367, P = 0.042). This suggests that even at the same latitude, small L. siphilitica populations occur in slightly warmer areas than large populations. More importantly, the difference in annual mean temperature between large and small populations suggests that variation in female frequency between population size classes could be caused by selective rather than nonselective mechanisms. If so, then the fitness of females relative to hermaphrodites should be higher in small populations growing in warmer sites than large populations growing in cooler sites.

In addition to an increase in the mean frequency of female L. siphilitica at warmer southern latitudes, the variance in female frequency was also higher, resulting in a fan-shaped distribution of population sex ratios. A similar pattern was observed by Barr (2004), who found that both mean female frequency and the variance in female frequency were higher in N. menziesii populations growing in dry soils. She explained this pattern by suggesting that differences in selection along the soil moisture gradient determined maximum female frequency of N. menziesii populations, but that the sex ratio of individual populations could vary below that maximum. Our data suggest that higher variance in female frequency with decreasing latitude in L. siphilitica was caused by an increase in the number of small populations at southern latitudes. Variance in female frequency may be higher among small than large populations because stochastic mechanisms such as genetic drift and founder effects have a larger effect on sex ratio in small populations (Nilsson & Ågren, 2006). This increase in stochastic variation could explain the fan-shaped distribution of sex ratios that we observed.

Annual mean temperature correlated with latitudinal variation in female frequency. Female L. siphilitica were more common in populations growing in warmer environments, suggesting that at higher temperatures females have higher relative fitness than hermaphrodites. Although a correlation between temperature and female frequency has been detected in many gynodioecious species, the sign of the correlation varies. Females are more common in populations growing in warmer environments in some species (e.g. Daphne laureola (Alonso & Herrera, 2001), Wurmbea biglandulosa (Vaughton & Ramsey, 2004) and colder environments in other species (e.g. Eritrichum aretioides (Puterbaugh et al., 1997), G. sylvaticum (Asikainen & Mutikainen, 2003). Given that females are hypothesized to be more common in harsh abiotic environments (Ashman, 2006), our results could imply that warmer environments are more stressful for L. siphilitica. Consistent with this hypothesis, we found that small L. siphilitica populations were more common at low latitudes. However, we do not know whether temperature has direct effects on the relative fitness of females and hermaphrodites, or whether temperature is a proxy for another ecological factor that varies latitudinally. For example, pollinator discrimination and pollen limitation vary along an altitudinal temperature gradient in Ranunculus acris (Totland, 2001). In addition, we cannot rule out the possibility that females are more common in warmer environments because rates of population turnover are higher in the south (Nilsson & Ågren, 2006). Growing hand- and open-pollinated female and hermaphrodite L. siphilitica plants at high and low latitudes would be one way to discriminate between these hypotheses.

Soil VWC correlated with variation in population sex ratio during fruiting, but not during flowering. To determine why VWC and female frequency were significantly correlated only during fruiting, we did two additional analyses. First, we pruned back the VWC during flowering data set to only those populations for which VWC was also measured during fruiting. In this reduced data set, VWC during flowering and female frequency were negatively correlated (Spearman's r13 = −0.645, P = 0.010). This result suggests either that VWC has a significant effect on female frequency only in the portion of L. siphilitica’s range (Wisconsin, Illinois and Iowa) for which we measured VWC during both flowering and fruiting or that the effect of VWC during flowering on female frequency was obscured by temporal variation, as VWC was measured over a longer time span during flowering than fruiting. Second, we reanalysed the complete VWC during flowering data set using a Spearman correlation to determine whether the different effects on population sex ratio of VWC during fruiting and flowering could be an artifact of the different models (correlation vs. ancova) used to analyse these data sets. We found that this correlation between VWC during flowering and population sex ratio was not significant (Spearman's r29 = −0.237, P = 0.199), suggesting that the different effects on sex ratio of VWC during flowering and fruiting was not a statistical artefact. More generally, our results are consistent with other studies which have found higher female frequencies in drier environments (e.g. Ashman, 1999), but L. siphilitica seems to differ in that soil water availability explains less of the variation in female frequency than does temperature.

Our data also support the hypothesis that sex-specific herbivory contributes to variation in population sex ratio in sexually dimorphic plant species (Ashman, 2006). Specifically, a higher mean proportion of fruits on female plants were attacked by a specialized seed predator in populations with fewer females. This sex-specific herbivory should reduce relative female fitness in populations with few females, resulting in natural selection for hermaphrodites. Although two other studies have examined the relationship between sex-specific herbivory and the population sex ratio in gynodioecious species (Marshall & Ganders, 2001; Ashman et al., 2004), ours is the first to detect a significant correlation. However, two caveats to our result are that we were only able to sample fruits from females in 15 populations and that herbivory can be quite variable between years (e.g. Rudgers & Hoeksema, 2003). In addition to surveying weevil damage in more L. siphilitica populations across multiple years, herbivore exclusion could be used to test whether sex-specific weevil damage causes differential natural selection on the sex phenotypes.

We collected data in only 1 year (2005), but female frequency and size of previously studied L. siphilitica populations were quite consistent across years. Female frequency varied little between samples at Deer Grove (0% in 1998 and 2005), Harker's Run (77% in 1984, 60% in 2005), CERA (4.1% in 1999, 4% in 2005), Krumm (9.2% in 1999, 3% in 2005) and Carthage (27.1% in 2004, 38.5% in 2005; data from Beaudoin Yetter, 1989; Dudle, 1999, C. M. Caruso, unpublished data). Similarly, population size was relatively stable at Harker's Run (n = 13 in 1984, n = 30 in 2005), CERA (> 100 in 1999 and 2005), Krumm (> 100 in 1999 and 2005), Reichelt (< 100 in 2000 and 2005) and Carthage (< 100 in 1999 and 2005; data from Beaudoin Yetter, 1989 and C. M. Caruso, unpublished data). These data suggest that the relationship between population sex ratio, population size and ecological factors may be consistent across years in L. siphilitica.

In summary, both natural selection and nonselective mechanisms may contribute to variation in sex ratio among L. siphilitica populations. Female L. siphilitica frequency varied along temperature, herbivory and soil moisture gradients, as predicted if sex-specific natural selection was contributing to variation in sex ratio. Female L. siphilitica were more common in small populations, as predicted if inbreeding, genetic drift and founder effects were contributing to variation in sex ratio. However, small and large populations differed ecologically from each other, suggesting that differences in female frequency between large and small populations could also be the result of natural selection. More generally, differential natural selection on the sex phenotypes may be a more important cause of variation in population sex ratio of L. siphilitica than predicted by models of sex ratio dynamics. Our results therefore highlight the importance of testing the assumption that much of the variation in population sex ratios of nucleo-cytoplasmic gynodioecious species does not have an adaptive or ecological cause (Gouyon et al., 1991).

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

We thank R. Cassells, A. Frewin, A. Hansgate, A. Lad, P. Rose, J. Shipp and particularly C. Shantz for assistance in the field and laboratory. C. Barr, H. Maherali, C. Shantz, M. Sherrard and two anonymous reviewers provided valuable comments on an earlier version of this manuscript. J. Bell provided data from five Wisconsin L. siphilitica populations. Thanks to the many people who provided information on and access to L. siphilitica populations, including T. Brock, T. Funke, W. Glass, T. Hartley, M. Jones, C. Mabry, L. Mottl, J. Phillips, R. Phillips, G. Sullivan, T. Trauscht, B. Turner, E. Ulaszek, W. Vanderploeg, M. Wilson and J. Wittig. This research was supported by funding to CMC from the Natural Science and Engineering Research Council of Canada, the Canadian Foundation for Innovation, the Ontario Innovation Trust and the University of Guelph and to ALC from Kent State University.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  • Alonso, C. & Herrera, C.M. 2001. Neither vegetative nor reproductive advantages account for high frequency of male steriles in southern Spanish gynodioecious Daphne laureola (Thymelaeaceae). Am. J. Bot. 88: 10161024.
  • Anderson, D.M. 1964. A review of the specific names in North American Miarus (Coleoptera: Curculionidae). Coleopterists’ Bull. 18: 2124.
  • Anderson, D.M. 1973. Keys to larvae and pupae of the Gymnetrinae of America north of Mexico (Coleoptera: Curculionidae). Proc. Entomol. Soc. Wash. 75: 133140.
  • Ashman, T.-L. 1999. Determinants of sex allocation in a gynodioecious wild strawberry: implications for the evolution of dioecy and sexual dimorphism. J. Evol. Biol. 12: 648661.
  • Ashman, T.-L. 2002. The role of herbivores in the evolution of separate sexes from hermaphroditism. Ecology 83: 11751184.
  • Ashman, T.-L. 2006. The evolution of separate sexes: a focus on the ecological context. In: The Ecology and Evolution of Flowers (L. D.Harder & S. C. H.Barrett, eds), pp. 419465. Oxford University Press, Oxford.
  • Ashman, T.-L. & Diefenderfer, C. 2001. Sex ratio represents a unique context for selection on attractive traits: consequences for the evolution of sexual dimorphism. Am. Nat. 157: 334347.
  • Ashman, T.-L., Cole, D.H. & Bradburn, M. 2004. Sex-differential resistance and tolerance to herbivory in a gynodioecious wild strawberry. Ecology 85: 25502559.
  • Asikainen, E. & Mutikainen, P. 2003. Female frequency and relative fitness of females and hermaphrodites in gynodioecious Geranium sylvaticum (Geraniaceae). Am. J. Bot. 90: 226232.
  • Bailey, M.F. & McCauley, D.E. 2005. Offspring sex ratio under inbreeding and outbreeding in a gynodioecious plant. Evolution 59: 287295.
  • Barr, C.M. 2004. Soil moisture and sex ratio in a plant with nuclear-cytoplasmic sex inheritance. Proc. R. Soc. Lond. B Biol. Sci. 271: 19351939.
  • Beaudoin Yetter, R. 1989. The expression of male-sterility in Lobelia siphilitica L. (Campanulaceae): a life history approach. PhD Dissertation, Miami University, Oxford, OH.
  • Byers, D.L., Warsaw, A. & Meagher, T.R. 2005. Consequences of prairie fragmentation on the progeny sex ratio of a gynodioecious species, Lobelia spicata (Campanulaceae). Heredity 95: 6975.
  • Carroll, S.P. & Corneli, P.S. 1995. Divergence in male mating tactics between two populations of the soapberry bug. 2. Genetic change and the evolution of a plastic reaction norm in a variable social environment. Behav. Ecol. 6: 4656.
  • Case, A.L. & Ashman, T.-L. 2005. Sex-specific physiology and its implications for reproductive cost. In: The Allocation of Resources to Reproduction in Plants (E.Reekie & F.Bazzaz, eds), pp. 129155. Springer-Verlag , Berlin.
  • Charnov, E.L. 1982. The Theory of Sex Allocation. Princeton University Press, Princeton, NJ.
  • Cornelissen, T. & Stiling, P. 2005. Sex-biased herbivory: a meta-analysis of the effects of gender on plant-herbivore interactions. Oikos 111: 488500.
  • Cuevas, E., Arias, D.M., Dominguez, C.A., Castillo, R.A. & Molina-Freaner, F. 2006. The genetic structure of the gynodioecious Kallstroemia grandiflora (Zygophyllaceae): the role of male sterility and colonization history. Heredity 97: 269274.
  • Dudle, D.A. 1999. Maintenance and consequences of females in the gynodioecious plant, Lobelia siphilitica. PhD Dissertation, Indiana University, Bloomington, IN.
  • Dudle, D.A., Mutikainen, P. & Delph, L.F. 2001. Genetics of sex determination in the gynodioecious species Lobelia siphilitica: evidence from two populations. Heredity 86: 265276.
  • Ellstrand, N.C. & Elam, D.R. 1993. Population genetic consequences of small population size: implications for plant conservation. Annu. Rev. Ecol. Syst. 24: 217242.
  • Emery, S.N. & McCauley, D.E. 2002. Consequences of inbreeding for offspring fitness and gender in Silene vulgaris, a gynodioecious plant. J. Evol. Biol. 15: 10571066.
  • Endler, J.A. 1986. Natural Selection in the Wild. Princeton University Press, Princeton, NJ.
  • Fisher, R.A. 1930. The Genetical Theory of Natural Selection. Clarendon Press, Oxford.
  • Frank, S.A. 1989. The evolutionary dynamics of cytoplasmic male sterility. Am. Nat. 133: 345376.
  • Gouyon, P.H. & Couvet, D. 1987. A conflict between two sexes, females and hermaphrodites. In: The Evolution of Sex and its Consequences (S. C.Stearns, ed.), pp. 245261. Birkhaeuser, Basel.
  • Gouyon, P.H., Vichot, F. & Van Damme, J.M.M. 1991. Nuclear-cytoplasmic male sterility: single-point equilibria versus limit cycles. Am. Nat. 137: 498514.
  • Hijmans, R.J., Cameron, S.E., Parra, J.L., Jones, P.G. & Jarvis, A. 2005. Very high resolution interpolated climate surfaces for global land areas. Int. J. Climatol. 25: 19651978.
  • Holeski, L.M. & Kelly, J.K. 2006. Mating system and the evolution of quantitative traits: an experimental study of Mimulus guttatus. Evolution 60: 711723.
  • Johnston, M.O. 1991a. Natural selection on floral traits in two species of Lobelia with different pollinators. Evolution 45: 14681479.
  • Johnston, M.O. 1991b. Pollen limitation of female reproduction in Lobelia cardinalis and L. siphilitica. Ecology 72: 15001503.
  • Johnston, M.O. 1992. Effects of cross and self-fertilization on progeny fitness in Lobelia cardinalis and L. siphilitica. Evolution 46: 688702.
  • Kaul, M.L.H. 1988. Male Sterility in Higher Plants. Springer-Verlag, Berlin.
  • Lewis, D. 1941. Male sterility in natural populations of hermaphrodite plants. New Phytol. 50: 5663.
  • Lloyd, D.G. 1976. The transmission of genes via pollen and ovules in gynodioecious angiosperms. Theor. Popul. Biol. 9: 299315.
  • Manicacci, D., Couvet, D., Belhassen, E., Gouyon, P.H. & Atlan, A. 1996. Founder effects and sex ratio in the gynodioecious Thymus vulgaris L. Mol. Ecol. 5: 6372.
  • Marshall, M. & Ganders, F.R. 2001. Sex-biased seed predation and the maintenance of females in a gynodioecious plant. Am. J. Bot. 88: 14371443.
  • Murayama, K., Yahara, T. & Terachi, T. 2004. Variation of female frequency and cytoplasmic male-sterility gene frequency among natural gynodioecious populations of wild radish (Raphanus sativus L.). Mol. Ecol. 13: 24592464.
  • Mutikainen, P. & Delph, L.F. 1998. Inbreeding depression in gynodioecious Lobelia siphilitica: among-family differences override between-morph differences. Evolution 52: 15721582.
  • Nilsson, E. & Ågren, J. 2006. Population size, female fecundity, and sex ratio variation in gynodioecious Plantago maritima. J. Evol. Biol. 19: 825833.
  • Olson, M.S. & McCauley, D.E. 2002. Mitochondrial DNA diversity, population structure, and gender association in the gynodioecious plant Silene vulgaris. Evolution 56: 253262.
  • Olson, M.S., McCauley, D.E. & Taylor, D. 2005. Genetics and adaptation in structured populations: sex ratio evolution in Silene vulgaris. Genetica 123: 4962.
  • Parachnowitsch, A.L. 2005. Natural selection on floral traits of Lobelia siphilitica by pollinators and a specialist pre-dispersal seed predator. MSc Thesis, University of Guelph, Guelph, ON.
  • Puterbaugh, M.N., Wied, A. & Galen, C. 1997. The functional ecology of gynodioecy in Eritrichum aretoides (Boraginaceae), the alpine forget-me-not. Am. J. Bot. 84: 393400.
  • Rudgers, J.A. & Hoeksema, J.D. 2003. Inter-annual variation in above- and belowground herbivory on a native, annual legume. Plant Ecol. 169: 105120.
  • Shykoff, J.A., Kolokotronis, S.O., Collin, C.L. & Lopez-Villavicencio, M. 2003. Effects of male sterility on reproductive traits in gynodioecious plants: a meta-analysis. Oecologia 135: 19.
  • Totland, O. 2001. Environment-dependent pollen limitation and selection on floral traits in an alpine species. Ecology 82: 22332244.
  • Vaughton, G. & Ramsey, M. 2004. Dry environments promote the establishment of females in monomorphic populations of Wurmbea biglandulosa (Colchicaceae). Evol. Ecol. 18: 323341.
  • Wilkinson, L. 1997. SYSTAT 7.0: Statistics. SPSS, Chicago, IL.
  • Zar, J.H. 1999. Biostatistical Analysis. Prentice Hall, Upper Saddle River, NJ.