Nonrandom Variation in Within-Species Sample Size and Missing Data in Phylogenetic Comparative Studies

摘要

Inferences about evolution are often made from interspecific comparisons in which species-specific traits are compared on the basis of within-species sampling (Harvey and Pagel 1991). Interspecific comparisons have traditionally been performed with comparative tools that are designed to control for the phylogenetic relatedness of species, which is needed to fulfill the statistical criterion of independent data points (Felsenstein 1985). Such phylogenetic comparative methods have been extensively used to make inferences about correlated trait evolution (Martins and Hansen 1997; Pagel 1999; Freckleton et al. 2002; Felsenstein 2004). However, it has recently been recognized that these approaches do not automatically deal with other statistical assumptions that are also of concern in relation to phylogenetically structured interspecific data (Ives et al. 2007; Freckleton 2009; Garamszegi and M?ller 2010). These assumptions are associated with distributional issues, covariance structures, evolutionary processes, and within-species variance and data quality. The present cautionary note elaborates on assumptions about data quality.