Abstract Question: Drought-induced tree mortality has been documented in forests worldwide but the mechanisms related to drought recovery are still poorly understood. To better predict forest trajectories under future climate scenarios, it is essential to disentangle physiological mechanisms underlying plant mortality caused by El Niño droughts. Here, we assessed how vegetation structure, vapor pressure deficit (VPD), and functional traits interact to mediate tree mortality after a severe drought in a tropical forest. Location: Mata das Flores State Park, an Atlantic Forest fragment in southeast Brazil. Methods: We established 20 permanent plots with contrasting vegetation structure and topography. In each plot, we measured tree abundance and diameter at breast height (DBH) of every woody plant (1–10 cm diameter) at the end of the drought, and two years after the break of drought, to calculate mortality rates during drought recovery. Hydraulic (e.g., maximum stomatal conductance to water vapor, stomatal density, etc.) and economic traits (specific leaf area, wood density, etc.) were measured on the 10 most abundant species. We also measured local air temperature and air humidity using HOBO dataloggers in each plot to calculate the VPD. Results: The studied Atlantic Forest understorey did not recover from the 2014–2016 drought, in terms of tree mortality. Lower VPD, driven by big trees in the valley, protected understorey plants with acquisitive economic attributes and conservative hydraulic attributes against mortality. On the other hand, higher VPD, driven by smaller trees and higher stem density on the ridge and slope, increased the mortality of understorey plants with acquisitive attributes. Conclusion: Ridges represent the most important fraction of the Atlantic Forest and our results suggest this type of forest is at high climate risk due to global change. Altogether, our results highlight that valleys are microclimate refuges for understorey plants and might help mitigate drought impacts in tropical forest under forecasted climate changes.
