Risk-sensitive fertility The variance compensation hypothesis

Two conditions are sufficient to indicate the need for risk-sensitive, adaptive analysis: (i) the outcomes of a behavior must be unpredictable to some degree and (ii) the relationship between outcomes and their value (in terms of fitness or utility) must be nonlinear. We argue that these conditions are common, and develop a general model for the analysis of risk-sensitive fertility behavior when long-term reproductive outcomes are unpredictable. We show that unpredictability is likely to have a patterned effect on fertility behavior, beyond adjustment for expected or average mortality, and we analyze the conditions under which that effect will be to promote or dampen fertility (positive or negative variance compensation, respectively). We then describe the implications of this variance compensation hypothesis (VCH) for the analysis of demographic transitions, agricultural intensification, fertility differences in natural-fertility populations, and clutch size. We conclude by lamenting that data are lacking to test the VCH, probably because it has not been appreciated that there can be patterned and effective behavioral adjustments to stochastic features of the environment