Fine-grained water availability drives divergent trait selection in Amazonian trees

Water availability is an important driver of plant functional biogeography. Most studies focus on the effects of precipitation, and neglect the contribution of groundwater as a source when the water table depth (WTD) is accessible to roots. Previous studies suggested that shallow water tables select for acquisitive traits. These studies have mostly contrasted shallow vs. deep water tables, without considering a more fine-grained perspective within shallow water tables or the temporal WTD behavior. Here we tested whether the degree of variation in WTD translates into divergent modes of trait selection. We expect constantly shallow WTD leading to the selection of acquisitive traits, whilst high fluctuation of WTD would lead to tree communities with more conservative traits. We used community and trait data (wood density and leaf traits) from 25 1-ha forest monitoring plots spread over 600 km in central Amazonia, covering a gradient of shallow to intermediate (0–8 m deep) WTD along the Purus-Madeira interfluve. Wood density was measured directly in trunk cores (498 trees) and leaf traits (Specific Leaf Area, Leaf Dry Mass Content, Leaf Thickness) of >6,000 individuals were estimated with FT-NIR (Fourier-Transformed Near-Infrared Spectroscopy) spectral models calibrated with cross-Amazonian data. We observed a turnover of families, genera, and species along the gradient of temporal WTD fluctuation range. This taxonomic turnover was accompanied by a change in wood traits, with higher wood density associated to higher WTD fluctuation and higher climatic water deficit. Leaf traits, however, varied in the opposite direction than initially hypothesized, i.e., trees had more acquisitive traits toward intermediate WTD with higher fluctuation. Based on those results, we propose that the effect of WTD selection should be conceptualized in a quadratic form, going from water excess in very shallow WTD (5 m, limiting condition, with conservative traits again).