Explorando patrones en rasgos macroecológicos utilizando regresión secuencial de autovectores filogenéticos.

Abstract

Bini, L.M., Villalobos, F., Diniz-Filho, J.A.F. 2014. Exploring patterns in macroecological traits using sequential phylogenetic eigenvector regression. Ecosistemas 23(1):21-26. Doi.: 10.7818/ECOS.2014.23-1.04 A number of methods have been proposed to estimate the level of phylogenetic signal (autocorrelation) in macroecological traits. These methods are useful to devise alternative ways to circumvent the problem of lack of independence among species and, recently, they have also proved valuable to infer how fast a trait has evolved in comparison with alternative evolutionary models, such as a Brownian motion or Ornstein-Uhlenbeck process. Recently, we developed a method called phylogenetic signal-representation (PSR) curve, an expansion of the phylogenetic eigenvector regression (PVR) proposed in 1998, which consists in estimating different coefficients of determination by regressing a trait of interest on the eigenvectors extracted from a phylogenetic distance matrix. The first model uses only the first of these eigenvectors as an explanatory variable; the second model uses both the first and the second and so on. After, the resultant coefficients of determination are plotted against the cumulative eigenvalues, and the shape of this curve is related to evolutionary models driving trait variation (i.e., a linear pattern is expected under Brownian evolution). Here, we used the PSR curve to study patterns of interspecific variation in Carnivora body size and geographical range size, and compared them with simulated curves under distinct evolutionary processes. Our results unequivocally support our expectations based on previous studies that body size has a strong phylogenetic signal, approximated by an OU pattern with low restraining force, whereas geographic range size is more labile and better fits the null expectations (i.e., absence of phylogenetic signal)