Assessing Phylogenetic Signal
Because of shared evolutionary history, closely related host species may
have similar effect sizes. We use Pagel’s lambda
(Pagel 1999) to
estimate phylogenetic signal in the distribution of effect sizes across
taxa. When phylogenetic non-independence was detected, we included a
phylogenetically structured random effect of species in our final
models. For both steps, we obtained a phylogeny of relevant prey/host
species from the Open Tree of Life using the ROTL package
(Hinchliff et
al. 2015; Michonneau et al. 2016); then we used the apepackage to prune the tree to our host species, to resolve polytomies,
and to generate branch lengths
(Paradis & Schliep
2018). For host species with multiple effect sizes, we calculated the
average effect size for each species weighted by the sample size of each
component study. We then used the pgls function of thecaper package to estimate Pagel’s lambda by maximum likelihood
for each of our datasets
(Orme et al.2018). We statistically tested the difference between this estimate and
two alternative possibilities: phylogenetic independence (lambda = 0)
and phylogenetic dependence as characterized by Brownian motion models
(lambda = 1). We failed to detect evidence of phylogenetic dependence in
any dataset using this method.
RESULTS