Drought impacts on physiology and mortality in rain forest: Results from a large scale rainfall exclusion experiment in Amazonia and their wider applicability

Background/Question/Methods

Drought has become a focal issue for understanding Amazon forest-climate interactions, partly because of its strong effects on forest functioning, and partly because of recent and possible future increases in the frequency or intensity of drought in the region.  To investigate the impacts drought on the carbon and water cycles of rain forest, in 2001 we implemented a large-scale throughfall exclusion experiment at Caxiuanã National Forest Reserve, State of Pará, Brazil; the experiment is on-going. Here, we synthesise 8 years of data acquired from the experiment which has been supported by LBA in Brazil, the UK-NERC, and the EU-FP. We made physiological measurements of leaves and soil, and combined these with longer term measurements of growth, mortality, recruitment and litterfall. We closely integrated our measurements with modelling at fine and coarse scales (a multilayer soil and canopy physiology model; and subsequently dynamic global vegetation models), and compared the results more widely with global data on the responses to drought by rain forests, examining for differences or similarities across the tropics.

Results/Conclusions

Leaf and soil physiology responded rapidly to the experimental drought at our site, and we demonstrated strong constraints on photosynthesis and soil respiration over the short term. Leaf respiration increased under drought and this affected the change in NEP at short and multi-year timescales. Tree growth declined following reductions in soil moisture availability, but tree mortality was resistant to drought, only increasing substantially after more than 3 years, and also providing evidence for taxon-specific differences in vulnerability to drought. Our fine-scale modelling of GPP was successful and demonstrated a need for good estimates of below-ground resistance terms for moisture transport. A test of three DGVMs against our data underlined the need for improved representation of physiology and mortality in these models, including how this might relate to species-based differences in drought vulnerability. Comparing results with other studies, first within Amazonia and then globally, we suggest that whilst there is evidence that Amazonian rain forest trees show some similarity in vulnerability to drought, this does not always hold true when intercontinental comparisons are made.