Calibration of the set of loci
To test the accuracy of the motherhood assignment provided by the software COLONY v.2.0.6.6 (Jones and Wang, 2010), we calibrated the assignment probability based on known mother-offspring relationships. The probability of assignment of a mother was plotted against its true motherhood, based on the known mother-offspring relationships from the field, which involved 58 mothers and 275 hatchlings from 105 nests. COLONY uses a full likelihood model to assign sibship and parentage relationships using multilocus genotype data. Its algorithm allows for missing data and genotyping errors. Three long runs were performed with high likelihood precision, allowing polygamy for both sexes under an inbreeding model, and assuming an error rate of 0.01 for both allelic dropout and genotyping error. It is assumed that hatchlings sampled in 2010/11 did not reach sexual maturity by 2020/21 based on the mean age at maturity of 25 years (Chaloupka et al., 2004), therefore, we did not consider the occurrence of overlapping generations. All the hatchlings were considered as potential offspring and all the females were candidate mothers.
To determine the effect of genotyping errors on the accuracy of motherhood assignment, the above calibration was run with three different subsets of markers to determine which one would best perform in assigning true mother-offspring relationships: subset 1 comprised all 23 loci described in Dolfo et al. (2023); in subset 2 the loci which were deviant from Hardy-Weinberg equilibrium, as emphasized by the authors, were removed, leaving 14 loci (list in Supplementary Data 1); in subset 3 the loci which were given the highest error rate by COLONY were removed (CMY19, CMY22, and CMY35), leaving 20 loci. Based on this calibration (Supplementary Data 1) all 23 loci were retained for the analyses.