Multivariate analyses of the effect of forest plot (primary and secondary) and tree size traits (DBH, tree height, trunk height, crown height, crown width, total leaves weight; log-transformed data) on ant species composition.

Canonical analysis (CA) and canonical correspondence analysis (CCA) was performed on ant presence-absence data (i.e. number of occupied trees by species) for i) all ant occurrences combined, ii) foraging ant occurrences and iii) arboreal-nesting occurrences. Results are shown for all canonical axes (i.e. all variables) combined and then for each variable tested separately in order of its significance in forward selection (only significant results up to fourth canonical axis shown).
aRank of canonical axis and significant variable in forward selection.
bCanonical eigenvalue for all axes (total value) and for each axis and corresponding variable separately.
cThe per cent variance in species data explained by respective CA and CCA canonical axes.
dSignificances of canonical axes assessed via Monte-Carlo permutation (F-ratio value, P < 0.05, 999 runs per analysis).
eThe per cent variance in species data explained by variables across all and by individual CA axes.
Multivariate analyses of the effect of forest plot (primary and secondary) and tree size traits (DBH, tree height, trunk height, crown height, crown width, total leaves weight; log-transformed data) on ant species composition.