Quantification of taxonomic, functional and structural diversity metrics
We computed plot-level taxonomic, functional and structural diversity metrics. Taxonomic diversity was quantified using species richness, Shannon Index, Pielou Evenness and Simpson index. Species richness was defined at plot level, as the number of distinct species enumerated inside each plot.
To quantify functional trait-based diversity, we computed multiple- and single-trait diversity metrics and functional dominance. Multiple-trait diversity metrics include functional richness (FRic), functional evenness (FEve), functional divergence (FDiv) and functional dispersion (FDis) (Mouchet, Villéger, Mason, & Mouillot, 2010; Villéger, Mason, & Mouillot, 2008). They were calculated at plot level, using the relative abundance of each species in each plot and the values of functional traits. We quantified the single trait functional divergence (FDvar), as the variance of each trait value weighted by the species’ relative density in each plot (Mason, Mouillot, Lee, & Wilson, 2005). For functional dominance, we calculated the plot-level community weight mean (CWM) for each functional trait based on the relative density of the species. We used CWM because it reflects functional shifts in mean trait values and species dominance in a community (S. Mensah, Veldtman, Assogbadjo, et al., 2016; Ricotta & Moretti, 2011). Both the multiple-trait functional diversity and dominance were calculated with “FD” package in R (Laliberté, Legendre, & Shipley, 2015). FDvar was calculated for each trait (i.e. wood density and plant maximum height) using the FDiversity software.
Structural diversity can be expressed with measures that relate to the horizontal and vertical extent as well as to the internal branching pattern of the trees. Thus, we focused on tree height and diameter differentiation across individual trees inside each plot, as well as on number of primary branches at crown base level, and assessed structural diversity as coefficient of variation (standard deviation to mean) of each of these three variables at plot level (S. Mensah, Pienaar, et al., 2018).