PARENT SESSION
Oral Session #65: Evolutionary Ecology/ Population Genetics. Presiding: G. Gerrish.
Thursday, August 9, 2001. 8:00 AM to 12:00 PM. Hall of Ideas J.
A new statistic for measuring phylogenetic structure in comparative data.
Blomberg, Simon1, Ives, Tony 1, Fisher, Diana2, Garland, Theodore1, 1 2
ABSTRACT- Many ecologists wish to know how much trait variation among species is attributable to shared ancestry ("phylogenetic inertia") versus recent adaptation. Several statistics have been proposed to estimate and/or separate phylogenetic effects, but so far all methods have faced problems (e.g., the phylogenetic autocorrelation method). We propose a simple statistic that measures the degree of phylogenetic structure in a data set containing values of a continuous trait for a number of species. Given the trait values and the phylogeny, the statistic ranges from 0 (trait best fits a "star" phylogeny) to 1 (trait best fits the given phylogeny). It is simply the scaling value that best describes the variance-covariance structure of the data, under conditions that minimize the rate of evolution of a character across the tree (maximum likelihood is used for estimation). Significance tests can be based on permutations or Monte Carlo simulations. The statistic can be calculated under both phylogenetically independent contrasts and generalized least squares paradigms. We present several examples, including analyses of a large data set of ecological, behavioral, and life history traits for marsupials.
KEY WORDS: comparative, method, phylogenetic, statistics