Abstract:
|
Distinguishing whether a group of subspecies forms one single species or separate species is a relevant problem for evolutionary and conservation biology. Here, we propose a Bayesian statistical method for species delimitation, for which we develop iBPP, an open source and publicly available software that biologists can use to estimate species limits from genetic and/or morphological data. Morphological data is commonly modeled with a multivariate normal distribution whose covariance matrix is determined by the species tree. For this model, we develop a new conjugate prior adapted to two levels of dependency: among measured variables and among individuals sampled from a set of putative species. This new conjugate prior transcends the biological context in which it is applied and can be utilized in other contexts with complex correlation structure. We illustrate the method through applications to Australian amphibolurine lizard species and describe biological issues addressed by our method, such as the use of multiple data types in a unified framework and how the method is based on an evolutionary model (unlike other approaches based on statistical clustering).
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.