Phylogenetic signal

Phylogenetic signal was tested using two methods: the ’aotf’ function in Phylocom 4.2 (Webb et al. , 2011) for the quantitative traits (maximum diameter, maximum height and relative coverage), and calculating the traitNRI for the qualitative traits (Gastauer et al. , 2017), except for Selaginella convoluta , which was not considered in the tests of the phylogenetic signal so as not to create bias caused by imbalanced phylogenetic trees because of the large phylogenetic distance (566 myr) between this only pteridophyte species and all the others (see (Duchêne et al. , 2015; Holman, 2005). For the endozoochory trait, maximum diameter and maximum height only the species of the woody component were considered (n=87). For the relative coverage trait (RC) only the species of the non-woody component were considered.
The ’aotf’ module compares the variance rank of the observed mean variance of the trait of interest across all nodes from a null generated distribution of 9999 random repeats to trait values through phylogeny (Webb et al. , 2011).
In order to evaluate trait conservatism, the significance of the variance of the independent contrasts was analyzed. 9999 randomizations were made and the trait was considered conserved if the variation of contrasts observed was less than 2.5% of the randomizations. TraitNRI and traitNTI were used to assert that all species possessing a given qualitative trait are more related to each other than expected by chance. The traitNRI was calculated for the phylogenetic structure of the community formed by the species with each trait, in order to emulate the phylogenetic signal of these traits. One thousand randomizations were performed and evaluated through null model. If traitNRIs or traitNTIs were lower than – 1.96 or higher than 1.96 (p<0.05), they were considered significant (Gastauer et al. , 2017).