Kin and group selection are both flawed but useful data analysis tools

Abstract

For understanding the evolution of social behavior in microbes, mathematical theory can aid empirical research but is often only used as a qualitative heuristic. How to properly formulate social evolution theory has also been contentious. Here we evaluate kin and multilevel selection theory as tools for analyzing microbial data. We reanalyze published datasets that share a common experimental design and evaluate these theories in terms of data visualization, statistical performance, biological interpretation, and quantitative comparison across systems. We find that the canonical formulations of both kin and multilevel selection are almost always poor analytical tools because they use statistical regressions that are poorly specified for the strong selection and nonadditive fitness effects common in microbial systems. Analyzing both individual and group fitness outcomes helps clarify the biology of selection. We also identify analytical practices in empirical research that suggest how theory might better handle the challenges of microbial data. A quantitative, data-driven approach thus shows how kin and multilevel selection theory both have substantial room for improvement as tools for understanding social evolution in all branches of life.