Update to Blangero's "Statistical Genetic Approaches to Human Adaptability" (1993): A Unified Theory of Genotype × Environment Interaction

My 1993 paper, "Statistical Genetic Approaches to Human Adaptability," originally was part of a special issue of Human Biology edited by Sarah Williams-Blangero and me (Williams-Blangero and Blangero 1993) that focused on the potential utility of genetic epidemiological models for investigations of anthropological genetic relevance. My contribution to the volume was stimulated by the many discussions that I had with one of my mentors, Cynthia Beall, while I was a graduate student at Case Western Reserve University during 1979-1986. I had always believed that the field of human adaptability research had not adequately utilized modern statistical models to provide rigorous evidence for the potential for a genetic basis of human physiological response to environmental challenges. Although many researchers in human biology and biological anthropology are interested in the evolution of human phenotypes in relation to environmental variation, the necessary first step of demonstrating a genetic basis for the underlying trait (and its response to environmental change) has been largely ignored. Such evidence for between-generation transmission is essential because evolution must act on existing genetic variation. After my graduate work, I moved into the substantially more lucrative field of complex disease genetics and began developing statistical genetic methods to locate and identify genes that influence complex phenotypes. As I became interested in the formal genetic analysis of complex traits, I realized that many of the techniques that I was developing and using would also be of value in studies of normal human variation.

One of my main goals for the 1993 paper was to try to attract human adaptability researchers to pursue large-scale family-based studies that would formally test for the presence of such genetic factors and supply accurate estimates of their overall importance in the determination of observed phenotypic variation. In later studies in collaboration with Professor Beall, we did use this family-based approach to detect genetic components in physiological phenotypes of great relevance for high-altitude adaptation (Beall et al. 1997a, 1997b). Similarly, the large extended pedigree studies that Dr. Williams-Blangero and I have long pursued in the Jirels of Nepal have also focused on identifying the genetic basis of human host response to parasitic infection (Williams-Blangero et al. 2002, 2008). These studies have all extensively used the theoretical results and the analytical techniques that I formulated in the 1993 paper.

My main regret regarding this 1993 paper was my choice of what I now believe to be an unfortunate title. Over the subsequent years, when I have assigned this paper as obligatory reading for my trainees, I have often been met with quizzical looks when they glance at the title. As I explain to them the focus of the paper, I have realized that a more accurate title, given the generic value of the results, would have been "Statistical Genetic Approaches to Genotype × Environment Interaction."

A genetic component for phenotypic response to the environment is, in fact, evidence for genotype × environment (G × E) interaction. Hence the field of human adaptability is implicitly focused on G × E interaction. The 1993 paper was my attempt at a unified statistical genetic theory on G × E interaction that would operationalize its formal testing. Looking back, the paper is remarkably complete with its coverage of both polygenic and locus-specific types of G × E interactions, and I still use these results in current studies of the genetic basis of complex phenotypes. The basic mathematical results still hold and clearly show how genetic variation in response occurs as a result of G × E interaction.

The 1993 paper reflects my long-held belief that family studies are optimal for the detection of G × E interaction. The section on cross-sectional type data (individuals measured in a single environment) shows that quantitative genetic assessment of overall G × E is feasible in families given adequate relationships. Relatives provide identical genotypes the opportunity to experience and respond to different environmental conditions. Using missing data theory, I showed how this incomplete data situation can recover the...