The biogeography of embolism resistance across resource gradients in the Amazon

Abstract

AbstractAimUnderstanding tree mortality under climate change‐induced droughts requires knowledge of hydraulic trait distribution across environmental gradients. The spatial distribution of embolism resistance (i.e., P50), a key trait linked to hydraulic failure, is currently unknown across the Amazon. Here, we tested how precipitation, water table depth (WTD) and soil fertility interact as filters modulating the geographic hydraulic trait distribution.LocationFour sites in three key regions across the Amazon basin: Central, Western peripheral area and Brazilian Shield.Time PeriodPresent.Major Taxa StudiedTree species.MethodWe measured hydraulic vulnerability curves of 64 tree species (165 individuals) using an orthogonal design contrasting seasonal precipitation gradients, soil fertility and WTD. To estimate the geographical distribution of hydraulic traits for the basin scale, we use the coefficients of the best generalized linear model, projected on a 1 km grid in each environmental layer.ResultsWe show that WTD and soil fertility are the main drivers of embolism resistance, while precipitation has a secondary effect. Trees in shallow WTD and fertile soils had riskier hydraulic strategies. Our spatial projection identified the Amazonian southern band and valleys across the basin as the most vulnerable areas. High soil fertility in these regions increases the variability of hydraulic traits and thus uncertainty in predictions of drought responses.Main ConclusionAt the scale of the Amazon basin, we show that dominant tree species traits converge toward hydraulic resistance when resource (water and nutrients) levels are low and towards hydraulic vulnerability (but with higher trait variability) when resources increase. These results highlight the importance of WTD and soil fertility in structuring hydraulic traits and their effects on tree mortality under climate change‐induced droughts. Further empirical tests covering a wider spatial range and more tree species will help validate our biogeographical distribution model.