REGRESSION ANALYSIS OF NATURAL SELECTION: STATISTICAL INFERENCE AND BIOLOGICAL INTERPRETATION

摘要: Recent theoretical work in quantitative genetics has fueled interest measuring natural selection the wild. We discuss statistical and biological issues that may arise applications of Lande Arnold's (1983) multiple-regression approach to selection. review assumptions involved estimation hypothesis testing regression problems, we note difficulties frequently as a result violation these assumptions. In particular, multicollinearity (extreme intercorrelation characters) extrinsic, unmeasured factors affecting fitness seriously complicate inference regarding Further, assumption residuals are normally distributed vitiates tests significance. For this situation, suggest recently developed jackknife While permits direct assessment form suitable for predicting response, aim inferring causal relationships about effects phenotypic characters on is greatly facilitated by manipulative experiments. Finally, alternative definitions stabilizing disruptive