Detecting Diversification Rate Variation in Supertrees

Abstract

Although they typically do not provide reliable information on divergence times, supertrees are nevertheless attractive candidates for the study of diversification rates: by combining a collection of less inclusive source trees, they promise to increase both the number and density of taxa included in the composite phylogeny. The relatively large size and possibly more dense taxonomic sampling of supertrees have the potential to increase the statistical power and decrease the bias, respectively, of methods for studying diversification rates that are robust to uncertainty regarding the timing of diversification events. These considerations motivate the development of atemporal methods that can take advantage of recent and anticipated advances in supertree estimation. Herein, we describe a set of whole-tree, topology- based methods intended to address two questions pertaining to the study of diversification rates. First, has a given (super)tree experienced significant variation in diversification rates among its branches? Second, if so, where have significant shifts in diversification rate occurred? We present results of simulation studies that characterize the statistical behavior of these methods, illustrating their increased power and decreased bias. We also applied the methods to a published supertree of primates, demonstrating their ability to contend with relatively large, incompletely resolved (super)trees. All the methods described in this chapter have been implemented in the freely available program, SYMMETREE