Coupled climate-carbon cycle models indicate that Amazon forests may be vulnerable to drought, with some predicting climate change-induced collapse of the Amazon forest and conversion to savanna, while others predict resilience. Much progress has been made in understanding tropical forest drought response, but integration of individual with ecosystem scale responses remains a challenge.
Here we analyze the response of an eastern Amazonian forest at ecosystem and individual scales to the El Nino-induced drought of late 2015/ early 2016. We use both eddy flux measurements of net ecosystem exchange of latent and sensible heat fluxes, and the ecohydrological responses, including leaf water potentials and stomatal conductances, of individual trees of species representing a large fraction of the site’s basal area. We use individual trees' pre-dawn water potential as a proxy for average soil water potential available to their roots to compare against declines in individuals' peak stomatal conductance, and change in midday minimum water potentials between 2012 (typical dry season) and 2015 (El Nino-induced drought) as a measure of drought stress. We use a simple weighted average approach to upscale from leaf-level water fluxes to ecosystem-level water fluxes.
Results/Conclusions
Whole-forest water (latent heat) fluxes normally reach their annual peak during the dry season, driven by the dry-season maximum in solar energy input. During the 2015-2016 El Nino dry period (August 2015 through February 2016), latent energy fluxes significantly declined relative to the average across all years, and sensible heat fluxes significantly increased, reflecting a strong shift in energy partitioning indicative of water limitation. Individuals, however, exhibited a variety of responses, consistent with a diversity of plant hydraulic strategies, with most species showing statistically significant declines in leaf water potential and stomatal conductance, but with a range across species that varied from no response to large magnitude declines. Leaf-level dominance-weighted fluxes were significantly correlated with tower fluxes.
Understanding and predicting whole-forest responses to strong drought is a critical priority for tropical forests, but may depend on accurate assessment of composition of species and their associated functions and hydraulic strategies.