Assortment and the analysis of natural selection on social traits

A central problem in evolutionary biology is to determine whether and how social interactions contribute to natural selection. This can be addressed with selection analyses that relate individual fitness to individual and social phenotypes. One such approach, known as social selection analysis, leads to the intuitive result that fitness effects from social partners will contribute to selection only if there is a correlation between the traits of individuals and their social partners (non-random phenotypic assortment). However, there have been inconsistencies in the application of this approach that center around the measurement of phenotypic assortment. Here, we use data analysis and simulations to resolve these inconsistencies, showing that: (i) not all measures of assortment are suitable for social selection analysis; and (ii) the interpretation of assortment, and how to detect non-random assortment, will depend on the scale at which it is measured. We discuss links between social selection and kin selection theory, and we provide a practical guide for the social selection approach. This article is protected by copyright. All rights reserved