Evolutionary bet‐hedging reconsidered: What is the mean–variance trade‐off of fitness?

In traditional theories, bet‐hedging in evolutionary biology is defined as a trade‐off between the within‐generation arithmetic mean fitness (AMF) of a genotype and between‐generation variance (BGV) in AMF across generations. The rationale of this definition is that a bet‐hedger genotype suppresses the BGV to increase between‐generation geometric mean fitness (GMF; an index of long‐term sustainability), which in turn entails costs in terms of AMF. However, too strict interpretation of this definition causes confusion among empirical researchers. For example, in empirical studies comparing a putative bet‐hedger (e.g., producing a generalist phenotype or mixture of various phenotypes) and non‐bet‐hedger control (e.g., producing only a specialist phenotype), reviewers sometimes request that a necessary condition of bet‐hedging is that the bet‐hedger candidate shows a smaller arithmetic mean of AMFs obtained from multiple generations and larger GMF than the control. However, the cost of bet‐hedging is incurred at the potential genotypic level and thus the decrease of AMF mean is not necessarily observed at the phenotypic level (especially if bet‐hedger individuals have good conditions). Moreover, contrary to previous arguments, the “fine‐grained” environments would promote bet‐hedging because even monomorphic specialist genotypes increase GMF if their population size is sufficiently large. Computer simulations support these views. I try to shift the definition of bet‐hedging from the trade‐off‐based one to the GMF‐based one: bet‐hedging is any strategy to increase the between‐generational GMF to avoid extinction of its controlling genotype against unpredictable environmental fluctuation. Under this new light, bet‐hedging will be a universal law of biology.