Environmental quality, developmental plasticity and the thrifty phenotype: a review of evolutionary models

The concept of the thrifty phenotype, first proposed by Hales and Barker, is now widely used in medical research, often in contrast to the thrifty genotype model, to interpret associations between early-life experience and adult health status. Several evolutionary models of the thrifty phenotype, which refers to developmental plasticity, have been presented. These include (A) the weather forecast model of Bateson, (B) the maternal fitness model of Wells, (C) the intergenerational phenotypic inertia model of Kuzawa, and (D) the predictive adaptive response model of Gluckman and Hanson. These models are compared and contrasted, in order to assess their relative utility for understanding human ontogenetic development. The most broadly applicable model is model A, which proposes that developing organisms respond to cues of environmental quality, and that mismatches between this forecast and subsequent reality generate significant adverse effects in adult phenotype. The remaining models all address in greater detail what kind of information is provided by such a forecast. Whereas both models B and C emphasise the adaptive benefits of exploiting information about the past, encapsulated in maternal phenotype, model D assumes that the fetus uses cues about the present external environment to predict its probable adult environment. I argue that for humans, with a disproportionately long period between the closing of sensitive windows of plasticity and the attainment of reproductive maturity, backward-looking models B and C represent a better approach, and indicate that the developing offspring aligns itself with stable cues of maternal phenotype so as to match its energy demand with maternal capacity to supply. In contrast, the predictive adaptive response model D over-estimates the capacity of the offspring to predict the future, and also fails to address the long-term parent-offspring dynamics of human development. Differences between models have implications for the design of public health interventions.