Abstract
Elucidating how environmental factors drive plant species distributions and how they affect latitudinal diversity gradients, remain essential questions in ecology and biogeography. In this study we aimed: 1) to investigate the relationships between all three diversity attributes, i.e., taxonomic diversity (TD), functional diversity (FD), and phylogenetic diversity (PD); 2) to quantify the latitudinal variation in these diversity attributes in western Amazonian terra firme forests; and 3) to understand how climatic and edaphic drivers contribute to explaining diversity patterns. We inventoried ca. 15,000 individuals from ca. 1,250 species, and obtained functional trait records for ca. 5,000 woody plant individuals in 50 plots of 0.1 ha located in five terra firme forest sites spread over a latitudinal gradient of 1200 km covering ca. 10°C in latitude in western Amazonia. We calculated all three diversity attributes using Hill numbers: q = 0 (richness), q = 1 (richness weighted by relative abundance), and q = 2 (richness weighted by dominance). Generalized linear mixed models were constructed for each diversity attribute to test the effects of different uncorrelated environmental predictors comprising the temperature seasonality, annual precipitation, soil pH and soil bulk density, as well as accounting for the effect of spatial autocorrelation, i.e., plots aggregated within sites. We confirmed that TD (q = 0, q = 1, and q = 2), FD (q = 0, q = 1, and q = 2), and PD (q = 0) increased monotonically towards the Equator following the latitudinal diversity gradient. The importance of rare species could explain the lack of a pattern for PD (q = 1 and q = 2). Temperature seasonality, which was highly correlated with latitude, and annual precipitation were the main environmental drivers of variations in TD, FD, and PD. All three diversity attributes increased with lower temperature seasonality, higher annual precipitation, and lower soil pH. We confirmed the existence of latitudinal diversity gradients for TD, FD, and PD in hyperdiverse Amazonian terra firme forests. Our results agree well with the predictions of the environmental filtering principle and the favourability hypothesis, even acting in a 10°C latitudinal range within tropical climates.