Overestimation of the strength of size-assortative pairing in taxa with cryptic diversity: a case of Simpson's paradox

Size-assortative pairing is one of the most common pairing patterns observed in nature and it probably occurs in many taxa with cryptic diversity. Observed patterns of size-assortative pairing in natural populations may thus be influenced by the co-occurrence of noninterbreeding cryptic groups of individuals living in sympatry. To quantify this potential bias, we sampled amphipods from the Gammarus pulex/Gammarus fossarum crustacean species complex in rivers containing two sympatric and morphologically cryptic groups, i.e. molecular operational taxonomic units (MOTUs). Within each river, MOTUs did not interbreed and differed in mean body size. We measured the strength of size-assortative pairing both within MOTUs and overall, combining both MOTUs for the analysis to test for potential effects of cryptic diversity on pairing patterns. Owing to body size differences between MOTUs, we expected size assortment to be stronger when combining MOTUs, which could represent a case of Simpson's paradox on measures of size-assortative pairing. In accordance with our predictions, in most rivers, combined-MOTU size-assortative pairing was stronger than within-MOTU size-assortative pairing. Combined-MOTU size-assortative pairing also increased with increasing body size difference between the two sympatric MOTUs. We discuss how such spurious correlations may lead to inferential fallacies when studying potential causes of pairing patterns and their consequences for sexual selection and phenotypic diversification. Previous results from studies measuring mating patterns in species in which cryptic diversity is likely to occur should be reappraised in the light of our findings.