Statistical analysis
We assessed the differences in the functional traits between active and passive climbing species by using t-tests. We log10-transformed the leaf area and seed mass before the analysis to meet the requirements of normality checked by quantile plots. In addition, we quantified the phylogenetic signal that indicates to what extent phenotypic expression is explained by the lineage to which a species belongs. We quantified the phylogenetic signal using the Blomberg“s K (Blomberg, 2003). We first generated a phylogenetic tree including all liana species for which we have functional trait information using the V.PhyloMaker R package (Jin & Qian, 2019). Consequently, we pruned separated phylogenies, one for each of the two main climbing mechanisms, using the original tree with all species as a base phylogeny. Subsequently, we pruned a tree for each group-trait combination independently, removing taxa for which information trait information was not available. Values of K = 1 imply that a trait shows exactly the amount of phylogenetic signal expected under a null, stochastic model of character evolution (Brownian motion evolution) (Blomberg, 2003). If K does not differ from zero, then the trait has no phylogenetic signal. The statistical significance of K was assessed via permutation tests with 1000 randomizations. The significance of the phylogenetic signal was based on the variance of phylogenetically independent contrasts relative to tip shuffling randomization implemented by the phylosignal function of thepicante package in R (Freckleton et al., 2002). The P-values were determined by comparing the variance of the standardized independent contrasts for the tip values against the randomized data.
We used piecewise structural equation modeling to examine the hypothesized relationships between abundance and richness per climbing mechanism and the climate, soil and forest structure variables (Lefcheck, 2016). One of the main advantages of the structural equation models (SEM) is that it allows partitioning the total effect of the relationship between variables into direct and indirect effects (Grace, 2006). We evaluated the generality of the model across the tropics by controlling the effect of biogeography (continent). We built the theoretical SEM (Appendix S2, Fig. S1) based on hypothesized and established relationships between environmental variables and the abundance and species richness of lianas and plants in general (Table 1). We also evaluated if the ratio of active, relative to passive, climbing species richness differed among biogeographic realms using a generalized linear model. All analyses were performed in the R environment (R Core Team 2018).