An analysis of covariance (ANCOVA) model with body mass as a covariate was used to assess the effect of mating system and female promiscuity on brain size and testes size in the sample after accounting for body mass
http://plosone.org/article/info:doi/10.1371/journal.pone.0000062#article1.body1.sec4.sec1.p1

Did the authors use a regular ANCOVA? We find out, below, that the variables analysed present a significant phylogenetic signal; as such, the data points cannot be considered independent, and a phylogenetically-weighted ANCOVA is strongly recomended. PDTREE can do this.