How many fathers? Study design implications when inferring multiple paternity in crocodilians.

Up to 10 males were reported to sire clutches of crocodilian eggs but review of the underlying study designs raised questions of potential upward bias of inferred sire numbers. To test this premise, different scenarios were explored using a published dataset of 16 known single-sire saltwater crocodile pairs and their offspring which were originally confirmed using a 11 loci microsatellite panel in CERVUS. Varying the number of microsatellites, omitting one or both parental genotypes and using different parentage analysis techniques revealed that total allele number, rather than number of loci, determined inferred sire accuracy in two opposing ways. Using the single-locus minimum method and GERUD, which both require prior knowledge of family groupings (i.e., nests), fewer alleles (and loci) accurately inferred only one father. In contrast, CERVUS and COLONY required all 11 loci (65 alleles) and both parental genotypes to (a) assign correct family groups and (b) infer the correct sire number, except in one family where two sires were equally assigned based on their number of homozygous loci. When less genotype information was provided, CERVUS and COLONY inferred up to six and seven sires, respectively. Given this data is from confirmed single-sire matings, and yet up to seven sires could be inferred, the significance of inappropriate study design is clearly demonstrated. Consideration should be carefully given to genotype data, particularly those collected specifically for population diversity studies, which are also used to infer multiple paternity because the underlying data collection assumptions are not equivalent between the two outcomes.