• Amazonia;
  • drought;
  • hydraulic conductivity;
  • rain forest;
  • sap flow;
  • soil water;
  • SPA model;
  • stomatal conductance;
  • throughfall exclusion;
  • water relations


Warmer and drier climates over Eastern Amazonia have been predicted as a component of climate change during the next 50–100 years. It remains unclear what effect such changes will have on forest–atmosphere exchange of carbon dioxide (CO2) and water, but the cumulative effect is anticipated to produce climatic feedback at both regional and global scales. To allow more detailed study of forest responses to soil drying, a simulated soil drought or ‘throughfall exclusion’ (TFE) experiment was established at a rain forest site in Eastern Amazonia, Brazil, for which time-series sap flow and soil moisture data were obtained. The experiment excluded 50% of the throughfall from the soil. Sap flow data from the forest plot experiencing normal rainfall showed no limitation of transpiration throughout the two monitored dry seasons. Conversely, data from the TFE showed large dry season declines in transpiration, with tree water use restricted to 20% of that in the control plot at the peak of both dry seasons. The results were examined to evaluate the paradigm that the restriction on transpiration in the dry season was caused by limitation of soil-to-root water transport, driven by low soil water potential and high soil-to-root hydraulic resistance. This paradigm, embedded in the soil–plant–atmosphere (SPA) model and driven using on-site measurements, provided a good explanation (R2 > 0.69) of the magnitude and timing of changes in sap flow and soil moisture. This model-data correspondence represents a substantial improvement compared with other ecosystem models of drought stress tested in Amazonia. Inclusion of deeper rooting should lead to lower sensitivity to drought than the majority of existing models. Modelled annual GPP declined by 13–14% in response to the treatment, compared with estimated declines in transpiration of 30–40%.