Abstract: Functional diversity (FD) reflects within- and between-site variation of species traits (α- and β-FD, respectively). Understanding how much data types (occurrence-based vs. abundance-weighted) and spatial scales (sites vs. regions) change FD and ultimately interfere with the detection of underlying geoclimatic filters is still debated. To contribute to this debate, we explored the occurrence of 1690 species in 690 sites, abundances of 1198 species in 343 sites, and seven functional traits of the Atlantic Forest woody flora in South America. All FD indices were sensitive and dependent on the data type at both scales, with occurrence particularly increasing α richness and dispersion (occurrence > abundance in 80% of the sites) while abundance increased β total, β replacement, and α evenness (abundance > occurrence in 60% of the sites). Furthermore, detecting the effect of geoclimatic filters depended on the data type and was scale-dependent. At the site scale, precipitation seasonality and soil depth had weak effects on α- and β-FD (max. R² = 0.11). However, regional-scale patterns of α richness, dispersion, and evenness strongly mirrored the variation in precipitation seasonality, soil depth, forest stability over the last 120 kyr, and cation exchange capacity (correlations > 0.80), suggesting that geoclimatic filters manifest stronger effects at the regional scale. Also, the role of edaphic gradients expands the idea of biogeographical filters beyond climate. Our findings caution functional biogeographic studies to consider the effect of data type and spatial scale before designing and reaching ecological conclusions about the complex nature of FD.

Key words: Environmental filtering, Historical events, Human pressure, n-dimensional hypervolume, Biodiversity hotspot, Tropical forest