3. Results
Figure 1a shows the bimodal size distribution of the 518 juveniles that could be caught from the wild. As expected, the random sample of 375 juveniles that was genotyped for parental assignment did not significantly differ in body lengths from the non-genotyped fish (Welch’s F = 1.7, p = 0.19). In total 301 (80.3%) of these 375 wild-caught juveniles could be assigned to 56 of the 60 experimental sib groups. Their average (± SD) body lengths and weights were 95.6 ± 10.6 mm and 11.3 ± 3.9 g, respectively, and were always below the 125 mm that the Bayesian mixture model had identified as upper size for 0+ fish (Fig. 1b). All but one of them could be genetically sexed. The overall male ratio was 48.3% and not significantly biased (χ2 = 0.33, p = 0.56).
The Bayesian model identified 57 of the remaining 74 genotyped fish as wild-born 0+ juveniles and the larger ones as 1+ or older (Fig. 1b). The wild-born 0+ were on average 17.6 mm smaller than the experimentally bred 0+ (Fig. 1b; F = 122.5, p < 0.001). Among the experimentally bred 0+, males were on average 2.6 mm larger than females (F = 4.4, p = 0.03; Fig. 1b). Among the wild-born 0+, males were on average 3.0 mm larger than females, which was in this smaller sample not statistically significant (F = 1.6, p = 0.21). However, the wild-born 0+ had a male-based sex ratio (63.2% males) that was not observed in the experimentally bred 0+ (χ2 = 4.3, p = 0.04; Fig. 1c).
We obtained 47 kinship coefficients (mean ± SD = -0.0003 ± 0.04). One full-sib family with a kinship coefficient of 0.226 was classified as outlier (because »3 SDs away from the mean, following the three-sigma rule) and excluded from all further analyses, leaving 251 genetically sexed juveniles of 46 full-sib families for the final analyses. The kinship coefficients of these sib groups could partly be predicted by the parental inbreeding coefficients: Higher parental inbreeding coefficients led to higher average kinship coefficients among their offspring (rs = 0.61, n = 20, p = 0.004; Supplementary Figure S1).
The body sizes of female juveniles declined with increased kinship coefficients, while the body sizes of male juveniles did not seem to be affected (Figure 2; Table 1; Supplementary Figure S2; Supplementary Table S1). Juvenile body size also varied among maternal but not paternal sib groups (Table 1).
The recapture rates per full-sib family, i.e., the mean number of recovered juveniles per number of released larvae, varied between 0.0 and 0.68 and were not correlated with the kinship coefficients (rs = 0.13, n= 48, p = 0.38) nor with the inbreeding coefficients of the dams (rs = 0.39, n= 11, p = 0.23) or the sires (rs = 0.05, n= 9, p = 0.90).
The number of recovered juveniles per experimental sib group that was represented in our sample varied from 1 to 17 (mean = 5.3, SD = 3.1), i.e., sex ratios per individual family could mostly not be determined due to low N. However, the number of recovered juveniles per dam varied from 5 and 52 (mean = 24.3, SD = 10.6). Family sex ratio differed among the maternal sib groups (χ 2 = 22.5, d.f. = 11, p = 0.02) but could not be predicted by maternal inbreeding coefficients (rs = -0.19, p = 0.57) nor the average kinship per maternal sib group (rs = 0.30, p = 0.37). The number of recovered juveniles per sire varied from 17 to 51 (mean = 29.2, SD = 10.6). Family sex ratio did not differ among the paternal sib groups (χ2 = 10.8, d.f. = 9, p = 0.29) and was not correlated with paternal inbreeding coefficients (rs = 0.65, p = 0.06) or average kinship per paternal sib group (rs = 0.37, p = 0.33).