Fig 1.
Effects of inbreeding, drift, and selection in captive populations.
Mean multilocus heterozygosity (MLH) for six captive populations and the wild source population. Replicate populations are shown by solid (replicate 1) and dashed (replicate 2) lines. Estimates are shown for all genotyped SNPs (A), neutral SNPs (B) and nonneutral SNPs (C) as determined via simulations. Error bars represent SE calculated across individual estimates.
Fig 2.
Comparison of the number of nonneutral SNPs detected per population at generations 6, 12, and 19.
Replicate populations are shown by solid (replicate 1) and dashed (replicate 2) lines. Although no significant difference was detected between different breeding protocols (ANOVA: F = 4.8037, df = 2, p = 0.19), the number of nonneutral SNPs identified per generation was significantly higher in generation 19 compared to generation 6 (ANOVA: F = 1.911, df = 2; Tukey: g19-g6: diff = 67.5, adjusted p = 0.026).
Fig 3.
Within-individual estimates of multilocus heterozygosity (MLH) calculated using neutral as well as nonneutral SNPs.
Diagonal line illustrates location of perfect concordance between neutral SNPs and non-neutral SNPs; individuals with equal MLH in neutral and nonneutral SNPs would be plotted on the diagonal. Overall, nonneutral SNPs have less diversity (lower MLH) compared to estimates calculated from neutral SNPs.
Table 1.
Mean difference of within individual estimates of heterozygosity (MLH) between nonneutral and neutral SNPs.
Bolded values are significantly different from each other. Differences between nonneutral and neutral SNPs in MLH were significantly smaller in the minimizing mean kinship populations compared to the randomly mating population.
Table 2.
Spearman correlations (r) between genetic diversity (MLH) and trait measures.
The p-values, shown in parentheses, were calculated via 1000 permutations. We estimated correlations using three groups of SNPs: all SNPs, nonneutral SNPs, and neutral SNPs. Significant values, identified by p-values < 0.05, are shown in bold. Italicized traits indicate traits most tightly associated with fitness.
Table 3.
Spearman fitness correlations (r) between pedigree-based inbreeding and trait measures.
The p-value, shown in parentheses, was calculated via 1000 permutations. We estimated genome-wide diversity using the inbreeding coefficient (F) as calculated from the pedigree. Note that F is inversely related to heterozygosity. Significant values, identified by p-values < 0.05 are shown in bold. Italicized traits indicate traits most tightly associated with fitness.
Table 4.
Expectations regarding heterozygosity-fitness correlations (HFCs).
Under the general effect hypothesis, HFCs are due to genome-wide effects that should result in stronger correlations between genetic diversity at neutral loci compared to nonneutral loci. In contrast, under both the local and direct effect hypotheses, HFCs should result from selection on or near the molecular markers themselves, in which case there should be stronger correlations between fitness and genetic diversity at nonneutral loci than between fitness and genetic diversity at neutral loci.