Relatedness and microgeographic genetic variation in Rhytidoponera mayri, an Australian arid-zone ant

Summary

We use a multiallelic regression and jack-knife technique to estimate relatedness and its confidence limits in a sample of 117 nests of the Australian arid zone ant Rhytidoponera mayri, using the genotypes at a prolifically-polymorphic amylase locus. Relatedness between workers from the same nest is low, whether calculated with respect to the complete sample of nests (b=0.121<0.158<0.195), or by restricting the analysis to those nests bordering a cell in a Gabriel-connectedness graph (b=0.101<0.126<0.151). Relatedness between workers of neighboring nests was determined for nearest-neighbors (b=0.021<0.054<0.087), and for nests connected in the Gabriel network (b=0.018<0.036<0.054). Relatedness is thus low but significant at both within- and between-nest levels, as is consistent with a life history involving multiple egg-layers and colony foundation via fission. Estimating relatedness for the different alleles separately yields some significant differences between alleles. We also tested for geographic substructuring using autocorrelation analysis of nine alleles separately and the factor scores for the first two principal components of all the allele frequencies: six of the eleven patterns tested differed significantly from randomness at the 95% level.