Phylogenetic paired t-test following Lindenfors et al. (2010).
phyl.pairedttest(tree, x1, x2=NULL, se1=NULL, se2=NULL, lambda=1.0, h0=0.0,
fixed=FALSE, ...)An object of class "phyl.pairedttest" with the following elements:
phylogenetic mean difference.
standard error of dbar.
estimated evolutionary variance (of the difference).
fitted (or fixed) value of λ.
log-likelihood of the fitted model.
t-value ((dbar-h0)/se where se is computed from the Hessian).
P-value.
the degrees of freedom.
the null hypothesis that was tested.
This function conducts a phylogenetic paired t-test, roughly following Lindenfors et al. (2010).
This is not a phylogenetic ANOVA, in which we want to compare the means of different sets of species on the tree. Instead, we are interested in the difference between two characters, or two measures of a character within a species, and we want to know if this difference is significantly different from zero controlling for the phylogenetic non-independence of species.
Likelihood optimization is performed using optim with method="L-BFGS-B" with box constraints on λ (0,1).
Lindenfors, P., L. J. Revell, and C. L. Nunn (2010) Sexual dimorphism in primate aerobic capacity: A phylogenetic test. J. Evol. Biol., 23, 1183-1194.
Revell, L. J. (2024) phytools 2.0: an updated R ecosystem for phylogenetic comparative methods (and other things). PeerJ, 12, e16505.