THE ULTIMATE CAUSES OF PHENOTYPIC INTEGRATION: LOST IN TRANSLATION1

Abstract

Since Darwin's time we have come to recognize two major proximate causes of genetic correlation within populations (pleiotropy and linkage disequilibrium). Today's challenge, taken up by Phenotypic Integration, is to understand the ultimate causes of phenotypic and genetic correlation. A major advance on this front was made by Olson and Miller (1958) who proposed that phenotypic correlations between morphological structures may often reflect functional interactions. Olson and Miller' s insight reverberates today as the first clear proposal for an ultimate cause of correlation. Olson and Miller coined the term morphological integration to describe character correlations that are shaped by selection. The first genetic model of morphological integration was constructed by Lande (1980). Lande was unable to derive a general expression for how genetic correlations evolve from generation to generation. Nevertheless, he was able to characterize the additive genetic variance-covariance matrix (G-matrix) at equilibrium as it pulsates and wobbles under the opposing forces of mutation, recombination, and selection. Among other things, the resulting expression identifies correlational selection (selection that directly changes the covariance between two traits) as the major selective force shaping the evolution of genetic correlation and hence morphological integration. In an important early implementation of Lande's (1980) results, Cheverud (1982) showed how morphological integration could be pursued in a quantitative genetic framework. For many years, Cheverud and his colleagues seemed to be the only modern champions of morphological integration. Thus, Pigliucci and Preston's book, Phenotypic Integration, brings welcome attention to empirical laws of correlation that remain, in many ways, mysterious. Phenotypic Integration brings together contributions from empiricists and theoreticians, botanists and zoologists, and carries a strong multivariate theme throughout its 19 chapters. The term phenotypic integration is a welcome broadening of morphological integration to include behavior and other kinds of nonmorphological traits. Does this volume introduce a new theory of phenotypic integration? Not only is the answer No!, in several chapters this volume steps backward more than two decades, as the authors create new conceptual muddles and needlessly fret over issues that have long enjoyed