Edwin Lebrija-Trejos, Eduardo A. Pérez-García, Jorge A. Meave, Frans Bongers, and Lourens Poorter. 2010. Functional traits and environmental filtering drive community assembly in a species-rich tropical system. Ecology 91:386–398.

Appendix A. Comparison between PCA results using all measured traits and a subset excluding theoretically interdependent traits.

We evaluated whether the analysis based on all measured traits, including theoretically interdependent traits, largely affected the results by inflating the percentage of variance explained by the main axis of variation (PCA axis 1) and, hence, the conclusions on the relative importance of functional traits and environmental factors. To this end, we performed a PCA eliminating three leaf traits that in theory (but not necessarily in practice, Table A1) may be directly correlated to leaf size, which was as a key component of between fallow differences in community assembly according to the all-trait (complete) PCA.

The complete PCA and the partial PCA (i.e., excluding leaf compoundness, slenderness, and petiole length) were nearly the same (Fig. A1).  The first axis of the partial PCA explained a little less of total trait variation along successional communities (37 vs. 40%), but the percentage of total variation explained by the first two axes of both PCAs was exactly the same (61%). In agreement with this similarity, the loadings of the functional traits on the first and second axes of both PCAs were strongly correlated (R = 0.99, P < 0.001), indicating that the relative importance of the traits was unaffected. The correlation was equally strong for the scores of the fallows on the first two PCA axes, indicating in turn that the relationships between the functional trait composition of the communities was also unaffected. Accordingly, as in the complete PCA, air temperature remained strongly related to the first axis of the partial PCA (R = 0.68, vs. 0.72 of the complete PCA) and air temperature was also more strongly related to this axis than the percentage of open-sky photosynthetic photon flux density (PPFD; R = 0.50 for the partial vs. 0.52 of the complete PCA). All correlations were statistically significant at P < 0.05.

FIG. A1. PCA ordinations of regeneration communities of secondary tropical dry forest plots and a mature-forest plot at Nizanda, Oaxaca, Mexico. The PCAs are based on (a) the community-level abundance-weighted-means of all measured traits of the most successful species in the regeneration assembly, and (b) a subset of the traits excluding leaf compoundness, slenderness, and petiole length. The scaling of the ordinations reflects the relationships between the species traits (in bold) and their approximate loadings on the ordination axes: AD, animal dispersal; De, deciduousness; Ex, exudates; Hmax maximum height; LC, leaf compoundness; LD, leaf density; LDMC, leaf dry matter content; LMA, leaf mass per area; LPb, leaf pubescence; LPl, leaf pulvination; LS, leaf slenderness; LSz, leaf size; LT, leaf thickness; Nfix, Nitrogen fixation; PL, petiole length; SSh, seed shape; SSz, seed size; Sp, spininess; UD, unassisted dispersal; WD, wind dispersal. Different symbols indicate the plots’ time since abandonment (open circles: < 10 yr; gray circles: 10–30 yr; closed circles: > 30 yr). Arrows represent the direction of increase of the environmental variables (in italics), and their lengths indicate approximate correlation strength (R) between the environment and the axes of variation. PPFD = percentage of open-sky photosynthetic photon flux density reaching the forest understory (i.e., 50 cm above soil level); Tair = daily mean air temperature at the forest understory; SMP = soil matric potential (0–10 cm depth); and Tsoil = soil temperature (ca. 13 cm depth) relative to ambient air temperature.