Temporally balanced selection during development of larval Pacific oysters (Crassostrea gigas) inherently preserves genetic diversity within offspring

Abstract

Balancing selection is one of the mechanisms which has been proposed to explain the maintenance of genetic diversity in species across generations. For species with large populations and complex life histories, however, heterogeneous selection pressures may create a scenario in which the net effects of selection are balanced across developmental stages. With replicated cultures and a pooled sequencing approach, we show that genotype-dependent mortality in larvae of the Pacific oyster (Crassostrea gigas) is largely temporally dynamic and inconsistently in favour of a single genotype or allelic variant at each locus. Overall, the patterns of genetic change we observe to be taking place are more complex than what would be expected under classical examples of additive or dominant genetic interactions. They are also not easily explained by our current understanding of the effects of genetic load. Collectively, temporally heterogeneous selection pressures across different larval developmental stages may act to maintain genetic diversity, while also inherently sheltering genetic load within oyster populations.

Keywords: balancing selection; genetics; larval development; oysters; poolseq.

Figure 1.

(a) Temporal patterns of mortality and significant changes in mean allele frequency across larval development. The points and line represent mean cumulative mortality across the 22-day culture period, bars represent bi-directional and uni-directional changes in allele frequencies at each time interval (non-cumulative). (b) Five general stages of development (embryo, D-hinge, veliger, pediveliger and spat) are depicted (not to scale) on the bottom row, relative to the developmental time in (a). (Online version in colour.)

Figure 2.

Change in allele frequency corresponding with larval mortality. (a) Mean survival of larvae in replicate cultures from fertilization to settlement (day 22). (b) Change (Δ) in survival between sampling periods. (c) Clustered trajectories of minor allele frequencies (n = 473 in total). (d) Change (Δ) in MAF between sampling points. Line colours correspond to patterns of change identified by sequential pairwise parametric tests (G, gradual; UD, uni-directional; BD, bi-directional). Bold red lines are localized estimate (loess) trajectories for all SNPs within a cluster (pink ribbons ± s.d.). (Online version in colour.)

Figure 3.

Overall changes in MAF between days 2 and 22 post-fertilization. (a) Loci with overall distortions were determined significant (in red) by pairwise comparison between day 2 and 22 (n = 181). (b) Loci with temporally significant distortions (in red) at one or more time points in development (n = 473). Each point represents a single locus (SNP). Distance from the dashed centre line is the overall distortion of allele frequency between day 2 and 22. (Online version in colour.)

Figure 4.

Modelled temporal patterns of genotype-specific fitness (0–1) across larval development. Each coloured line represents mean estimates of fitness for each locus from n = 50 simulations. Black lines are pooled means within each genotype and cluster (from figure 2; grey ribbons = ± s.d.). (Online version in colour.)

Figure 5.

Simulated changes in genotypic proportions from fertilization to settlement. Changes in spat genotypes (circles) relative to HWE from the MAF of the spat pool (dashed lines). (Online version in colour.)