Full text loading...
Abstract
In evolutionary ecology, risk-spreading (i.e. bet-hedging) is the idea that unpredictably variable environments favor genotypes with lower variance in fitness at the cost of lower arithmetic mean fitness. Variance in fitness can be reduced by physiology or behavior that spreads risk of encountering an unfavorable environment over time or space. Such risk-spreading can be achieved by a single phenotype that avoids risks (conservative risk-spreading) or by phenotypic variation expressed by a single genotype (diversified risk-spreading). Across these categories, three types of risk-spreading can be usefully distinguished: temporal, metapopulation, and within-generation. Theory suggests that temporal and metapopulation risk-spreading may work under a broad range of population sizes, but within-generation risk-spreading appears to work only when populations are small. Although genetic polymorphisms have sometimes been treated as risk-spreading, the underlying mechanisms are different, and they often require different conditions for their evolution and thus are better treated separately. I review the types of evidence that could be used to test for risk-spreading and discuss evidence for risk-spreading in facultative diapause, migration polyphenism, spatial distribution of oviposition, egg size, and other miscellaneous traits. Although risk-spreading theory is voluminous and well developed in some ways, rarely has it been used to generate detailed, testable hypotheses about the evolution of risk-spreading. Furthermore, although there is evidence for risk-spreading, particularly in facultative diapause, I have been unable to find any definitive tests with unequivocal results showing that risk-spreading has been a major factor in the evolution of insect behaviors or life histories. To advance our understanding of risk-spreading in the wild, we need (a) explicit empirical models that predict levels of diversifying risk-spreading for several insect populations in several environments that vary in uncertainty, and (b) tests of these models using measurements of phenotypes and their fitnesses over several generations in each environment.