The multivariate evolution of female body shape in an artificial digital ecosystem

Human bodies exemplify complex phenotypes, likely to be subject to complex evolutionary forces. Despite the importance of body shape to health, social interactions and self-esteem, our understanding of body evolution and integration remains simplistically focused on simple ratios like waist–hip ratio (WHR), and body mass index (BMI), or manipulations of one or a few traits. Evolutionary selection analyses give a multivariate perspective, but highly correlated body measures create multicollinearity problems. Here we develop an original approach mimicking Darwinian selection to study how clonal lines of bodies, allowed to vary in 24 attributes via a mutation-like process, evolve in a digital ecosystem over 8 generations. Ten of 24 traits changed by more than one |S.D.| over seven generations of selection. Analyses of selection within generations, change in population mean, and change within clonal family lines all implicate slenderness, particularly narrow waists and long legs as the most important dimension of body attractiveness. WHR did not offer any improvement on waist girth as a predictor of attractiveness. Within the most successful clonal lineages, selection favored greater shapeliness, including larger busts, in addition to slenderness. Our results reveal the complex, multivariate nature of attractiveness, and that the success of simple ratios like WHR and BMI in previous studies is probably incidental to the importance of waist girth and general slenderness. Our results also suggest that the integration of the entire body phenotype is at least as important as any one trait, and that more than one way exists to make an attractive body.