SOURCES OF GENETIC AND PHENOTYPIC VARIANCE IN FERTILIZATION RATES AND LARVAL TRAITS IN A SEA URCHIN

Sexual selection, in the form of mate choice and mating competition (Andersson 1994), and their postmating equivalents, cryptic choice, and sperm competition (Birkhead and Møller 1998), may facilitate the selection of genetically compatible or intrinsically high-quality genes (Neff and Pitcher 2005). In the former case, incompatibilities result in nonadditive genetic variance in fertilization or offspring fitness, which can be due to a diverse array of genetic mechanisms all involving interactions between paternal and maternal genetic elements (Zeh and Zeh 1996; Tregenza and Wedell 2000). Thus, according to the genetic compatibility hypothesis, offspring fitness is determined by specific interactions between maternal and paternal haplotypes. By contrast, selection for good genes results in additive genetic variance in fitness, so that a male with good genes will produce offspring with higher fitness irrespective of the female's genetic background (Andersson 1994; Møller and Alatalo 1999).

Identifying and disentangling genetic and environmental sources of variance in sexual selection remains a key challenge in many animal systems, especially those with internal fertilization where (nongenetic) maternal effects may amplify or mask subtle genetic processes (e.g., Kotiaho et al. 2003). Animals and plants with external fertilization and sexual reproduction offer excellent opportunities to address this problem by providing tractable systems for attributing phenotypic variance in fitness-related traits to genetic and environmental causes. In these systems, gametes from two or more members of each sex can be crossed in all combinations to generate multiple sibling relationships (full siblings, paternal, and maternal half siblings), without recourse to female multiple mating or crosses involving inbred lines (see Lynch and Walsh 1998). Such cross-classified designs therefore expand on alternative approaches (e.g., parent-offspring regression, nested sibling analyses) to detect genetic effects that contribute toward familial resemblance (Lynch and Walsh 1998). A second important advantage of using external fertilizers is that they offer more tractable systems for controlling nongenetic (environmental) sources of variance, thus providing more precise estimates of genetic variance components.

In this article we identify sources of genetic and phenotypic variance in fertilization rates and two early phases of offspring development in Heliocidaris erythrogramma, a common sea urchin inhabiting coastal waters in southern Australia (Keesing 2001), and a model system for developmental studies (Williams and Anderson 1975). We used the North Carolina II (NCII) block breeding design (Lynch and Walsh 1998) to partition genetic variance in these traits following replicated factorial crosses between two males and two females within each block. The NCII design has previously been applied to externally fertilizing animals and plants, where split-clutch and in vitro fertilization (IVF) techniques make it possible to partition gametes from both sexes and cross them in all combinations (Wedekind et al. 2001; Evans and Marshall 2005; Marshall and Evans 2005; Rudolfsen et al. 2005; Trippel et al. 2005; Pitcher and Neff 2006; Steven et al. 2007). Indeed, IVF techniques are well developed in H. erythrogramma and offer unparalleled experimental control over staged artificial fertilizations (Marshall et al. 2004; Evans and Marshall 2005).

We analyzed sources of variance due to male and female effects (and their interaction) using a split-clutch IVF assay (see Evans and Marshall 2005). In common with other broadcast spawning marine invertebrates, we anticipated strong maternal effects at fertilization due to variation among females in the size distribution of eggs (Marshall et al. 2004). Such effects strongly influence fertilization kinetics in broadcast spawning marine invertebrates because larger eggs are more likely to be contacted by sperm than smaller eggs (Levitan 1996; Styan 1998; Marshall et al. 2002). Our analysis of offspring performance focused on two phases of early larval development: (1) premetamorphic survival (hereafter “embryo viability”), defined as the proportion of initially fertilized eggs that successfully hatched from their fertilization membranes and progressed to the free-swimming larval stage, and (2) metamorphosis, the proportion of hatched larvae that metamorphosed by six days into the juvenile form in the absence of settlement cues. Our ability to assess fertilization rates directly and then measure offspring traits during two stages of development enabled us to assess the interrelationships between these potential fitness traits. In particular, we were interested in the relationship between fertilization rates and embryo viability, because the latter is often used to estimate fertilization success in species in which it is not possible to estimate fertilization rates directly (i.e., most internally fertilizing species). This distinction is especially important in studies that estimate paternity (as a proxy for relative fertilization success) from newly hatched or adult offspring genotypes. As emphasized elsewhere, paternity patterns may not reflect fertilization patterns at conception (Gilchrist and Partridge 1997; García-González and Simmons 2007), especially where embryo viability is influenced by differential mortality, maternal and paternal effects or other sources of environmental variance (Olsson et al. 1999; Simmons 2005; García-González and Simmons 2007).

Materials and Methods

Results

Embryo viability (the proportion of initially fertilized eggs that reached the free-swimming larval stage of development) ranged from 65% to 100% (mean ± SD = 76.5 ± 18.8%). From these successful hatchings, a mean of 29.7% (±20.1 SD; range: 1–73.5%) successfully metamorphosed into the juvenile form by the six days census point.

The results from the NCII analysis are summarized in Table 1. The analysis of fertilization rates revealed highly significant male and female variance components and no significant interaction between these main effects. By contrast, embryo viability was strongly influenced by the interaction between males and females, indicative of nonadditive genetic variance in this trait. Finally, our analysis of the number of juveniles that subsequently underwent metamorphosis revealed that the component of variance associated with sires was significant, accounting for 31% of the observed phenotypic variance for this trait (Table 1).

We also examined the interrelationships between fertilization rates, embryo viability, and metamorphosis. The distribution of correlation coefficients for all possible (512) combinations of independent data arising from the blocks is shown in Figure 1 (A–C) for the three relationships. The 95% confidence limits for these distributions inform us about the significance of the association between the pairs of variables (Fig. 1). These analyses revealed that the relationship between fertilization rates and embryo viability is not significant (confidence interval at α= 0.05 included zero; see Fig. 1A). By contrast, we found a positive and significant relationship between fertilization rates and metamorphosis (Fig. 1B), and a negative and significant relationship between embryo viability and metamorphosis (Fig. 1C). In the former case we found that all possible correlation coefficients always yielded r-values that were above zero (range r= 0.197/0.667; see Figure 1B), whereas in the latter all possible correlation coefficients yielded r-values that were below zero (range r=−0.83/−0.08; Fig. 1C).

Discussion

Our analysis focused on three potential fitness traits in the sea urchin H. erythrogramma: noncompetitive fertilization rates, pre-metamorphic larval survival (embryo viability), and metamorphosis. The relative influence of additive and nonadditive sources of variance depended on which of these traits was measured, mirroring recent findings for decorated field crickets (Ivy 2007) and earlier work on alpine whitefish (Wedekind et al. 2001). Taken together, these studies emphasize the importance of considering multiple components of offspring fitness when considering the genetic basis of mate choice and other forms of selection (see Hunt et al. 2004).

Our analysis of fertilization rates revealed highly significant male and female variance components and no significant interaction between these main effects. In the absence of extrinsic environmental effects (which were rigorously controlled during the assays), this finding implies that individuals of both sexes exhibit consistent variation in gamete traits. The difference between the variance components associated with females and males at fertilization provides an estimate of the variance due to maternal effects (Lynch and Walsh 1998), which in our analysis would have been substantial (see Table 1). This strong maternal effect at fertilization is consistent with a large body of work on other marine broadcast spawners (e.g., Levitan 1993; 2000; Marshall et al. 2000) and may be due to consistent differences in the quality, size, or maturity of eggs among females, all of which constitute important sources of variation at fertilization in H. erythrogramma (Marshall et al. 2004). The lack of interaction at fertilization reported here contrasts with the findings from a different (eastern Australian) population of H. erythrogramma, where in addition to significant main effects, interacting parental effects were also important (Evans and Marshall 2005). This difference may reflect the different population origins of study animals, different study environments, or differences in selection pressures that operate on gamete recognition traits in the two populations (see Zigler et al. 2005).

Although we did not specifically examine sperm performance traits in this study, our finding that variance in fertilization was partially attributable to general male effects is consistent with an increasing body of work showing that consistent variation among males in sperm traits (e.g., motility, swimming speeds, and viability) can influence fertilization (Birkhead et al. 1999; Gage et al. 2004; García-González and Simmons 2005b). Our ability to carefully control sperm densities (and sperm-to-egg ratios) during the IVF trials leads us to speculate that individual males differ consistently in sperm or ejaculate traits (other than sperm number) that influence fertilization (see review by Snook 2005), a possibility that we are currently pursuing using sperm viability assays and computer-assisted sperm analyses. Nevertheless, to rule out possible environmental (i.e., nongenetic) effects on sperm performance we advocate future quantitative genetic experiments that specifically focus on the genetic basis of traits known to regulate fertilization (see forthcoming reviews by Evans and Simmons 2007).

Our analysis of embryo viability yielded a highly significant male-by-female interaction, indicative of nonadditive genetic variance in this trait. This finding is consistent with the genetic compatibility hypothesis because it suggests that during this life-history phase sexual selection will target compatible male–female crosses rather than intrinsically high-quality mating partners (cf. García-González and Simmons 2005a). This result is consistent with findings from several insect studies in which it has been argued that genetic incompatibility can limit hatching rates and offspring viability (Tregenza and Wedell 1998; Newcomer et al. 1999; Engqvist 2006). However, much of the previous support from animals for the genetic compatibility hypothesis comes from studies that have failed to provide firm support for alternative (good genes) processes. Only a few animal studies have implicated genetic compatibility (i.e., nonadditive variance in fitness-related traits) within an experimental framework that employs statistical designs powerful enough to exclude the effects of good genes (Wedekind et al. 2001; Pitcher and Neff 2006; Ivy 2007). By contrast, the plant literature provides ample compelling evidence for incompatibility avoidance, particularly with respect to avoiding self-fertilization and promotion of outcrossing (reviews by Barrett 2003; Bernasconi et al. 2004). Our preliminary analysis of potential fitness traits in H. erythrogramma provides further support for the genetic compatibility hypothesis, at least with regard to embryo viability. The mechanisms underlying these nonadditive genetic effects are largely unknown, but may include incompatibilities due to selfish genetic elements, inbreeding and heterozygote deficiency (see reviews by Zeh and Zeh 1996; Tregenza and Wedell 2000).

Although variation in embryo viability was clearly influenced by nonadditive effects, our analysis of the number of juveniles that subsequently underwent metamorphosis revealed that the component of variance associated with sires was significant, accounting for 31% of the observed phenotypic variance for this trait (Table 1). Although it is not clear whether early (or indeed delayed) metamorphosis, as measured in our study, has fitness implications for juvenile urchins, there is anecdotal evidence from work on the settlement ecology of this species which suggests that spontaneous metamorphosis in the absence of settlement cues (originating from the coralline alga Amphiroa anceps) results in lower postmetamorphic survival (D. J. Marshall, unpubl. data). Indeed, several studies on other marine invertebrates show a strong effect of settlement choice on the subsequent performance of settlers (Elkin and Marshall 2007). Consequently, variation in settlement rates may provide an important target for natural selection in H. erythrogramma. More generally, the settlement “choices” that marine invertebrate larvae make are thought to have important implications for fitness, and delaying settlement in the absence of cues indicating suitable habitats may be an adaptation to avoid high rates of mortality (Pechenik 1990; Elkin and Marshall 2007).

Interestingly, embryo viability and metamorphosis (at six days) were negatively correlated (Fig. 1C), raising the possibility that larvae from crosses with high hatching rates were more likely to reject poor settlement cues than those from less viable crosses. Alternatively, and equally plausible, rates of larval metamorphosis may have been influenced by larval density, which would have varied according to differences in hatching rates among crosses. Unfortunately, our study cannot distinguish between these possibilities. Nevertheless, our work does underscore the need for studies specifically aimed at understanding the fitness implications of variation in spontaneous juvenile metamorphosis, as such information is crucial before concluding that additive genetic variance in this trait is due to paternal good genes.

We also uncovered a positive association between fertilization and metamorphosis rates (the proportion of surviving larvae that successfully metamorphosed into the juvenile form). This finding is potentially informative in the context of hypotheses that link male fertilizing success to the genetic benefits they bestow on their offspring (Yasui 1997) (reviewed by Evans and Simmons 2007). However, it is important to emphasize two important limitations of our study with respect to its relevance for such “good sperm” processes. First, the importance of metamorphosis (as measured in our study) in influencing offspring fitness is unknown (see above), and therefore it is unclear whether the selection of males imposing high (or low) rates of spontaneous metamorphosis on their offspring actually represents a benefit. Second, and more crucially with respect to good sperm processes, we estimated fertilization rates during noncompetitive IVF assays; presently we do not know whether the performance of sperm undergoing noncompetitive IVF trials predicts their success during sperm competition. Our ongoing research addresses both of these questions in H. erythrogramma in an attempt to specifically address the good sperm hypothesis.

Finally, we found no association between fertilization rates, which were influenced by intrinsic male and female effects, and embryo viability, which was strongly influenced by interacting parental effects. This finding has potentially important ramifications for studies that use paternity data to infer fertilization rates, particularly in internal fertilizing species in which paternity is typically estimated from newly hatched offspring rather than recently fertilized eggs. Although we stress again that our fertilization data come from noncompetitive mating trials, and that the link between noncompetitive and competitive fertilization success needs to be made, our data may inform recent debates on the importance of estimating fertilization at conception rather than from the proportion of viable offspring sired by focal males, because the latter may be influenced by a range of (differential) effects that affect embryo survival (e.g., see Gilchrist and Partridge 1997; Olsson et al. 1999; García-González and Simmons 2007).

Associate Editor: D. Ayre