Assessing the relationship between forest types and canopy tree beta diversity in Amazonia.

Planning of conservation priorities has often taken mapped forest types as surrogates for biological complementarity. In the Brazilian Amazon, these exercises have given equal weight to each forest type as if they were all equally distinct. Here, we examine floristic similarity between forest types to assess the reliability of vegetation maps as a surrogate for canopy tree-community composition. We analyzed floristic differences at the genus level between twelve Amazonian forest types using 1184 one-hectare inventories of large trees with three complementary approaches. First, we compared a map of floristic composition, from a uni-dimensional NMDS ordination of the inventories, with a map of coarser-level forest types commonly recognized as distinct by classification systems across Amazonia. Using Mantel and means-difference tests, we next examined the distance-decay of floristic similarity for all paired samples and for the pairs drawn from within and between twelve more finely divided forest types. Finally, we examined the degree of floristic separation of each pair of the twelve forest types using non-parametric analysis of variance. Maps of floristic composition and coarse-level forest types were highly congruent. At the finer level of classification, similarity was only slightly higher when pairs were drawn from the same versus from different forest types. This was true for all geographic distances. Nonetheless, eighty percent of the 66 paired combinations of forest types were significantly different in the unreduced genus-space and nearly half showed little or no overlap in a two-dimensional ordination. Three types were most distinct from all others: white sand, seasonally dry, and bamboo-dominated forests. Here, we show that forest types exhibit variable degrees of separation. For this reason, treating all fine-level forest types as equally distinct results in poor representation of canopy tree beta diversity. We recommend explicitly considering the degree of floristic separation between all forest types – as presented here for Amazonian flora – as a way to improve the use of this biodiversity surrogate.