Short-term effects of drought on tropical forest do not fully predict impacts of repeated or long-term drought: gas exchange versus growth

This is the final version. Available from the Royal Society via the DOI in this record.Are short-term responses by tropical rainforest to drought (e.g. during El Niño) sufficient to predict changes over the long-term, or from repeated drought? Using the world's only long-term (16-year) drought experiment in tropical forest we examine predictability from short-term measurements (1-2 years). Transpiration was maximized in droughted forest: it consumed all available throughfall throughout the 16 years of study. Leaf photosynthetic capacity [Formula: see text] was maintained, but only when averaged across tree size groups. Annual transpiration in droughted forest was less than in control, with initial reductions (at high biomass) imposed by foliar stomatal control. Tree mortality increased after year three, leading to an overall biomass loss of 40%; over the long-term, the main constraint on transpiration was thus imposed by the associated reduction in sapwood area. Altered tree mortality risk may prove predictable from soil and plant hydraulics, but additional monitoring is needed to test whether future biomass will stabilize or collapse. Allocation of assimilate differed over time: stem growth and reproductive output declined in the short-term, but following mortality-related changes in resource availability, both showed long-term resilience, with partial or full recovery. Understanding and simulation of these phenomena and related trade-offs in allocation will advance more effectively through greater use of optimization and probabilistic modelling approaches.This article is part of a discussion meeting issue 'The impact of the 2015/2016 El Niño on the terrestrial tropical carbon cycle: patterns, mechanisms and implications'.This work was supported by UK NERC grant NE/J011002/1 to P.M. and M.M., NERC independent fellowship grant NE/N014022/1 to L.R., ARC grants FT110100457 and DP170104091 to P.M., CNPQ grant 457914/2013-0/MCTI/CNPq/FNDCT/LBA/ESECAFLOR to A.L.d.C. It was previously supported by NERC NER/A/S/2002/00487, NERC GR3/11706, EU FP5-Carbonsink and EU FP7-Amazalert to P.M., and by a grant from the Gordon and Betty Moore Foundation