In the Mediterranean basin, precipitation is expected to decline as a consequence of climate change. The response of a Quercus ilex forest in southern France to such a decline in water availability was studied using a 4-year throughfall exclusion experiment. Seasonal courses of sap flow and leaf water potential were obtained from 2004 to 2007 and used to characterize tree water relations in a control and a dry treatment. The experiment reduced the average precipitation input to the soil by 29%, and resulted in a 23% reduction in annual transpiration. Soil water potential was significantly lower in the dry treatment only during summer drought, but transpiration was reduced all year round even during well-watered periods. Despite a tight stomatal control over transpiration, whole-tree hydraulic conductance was found to be lower in the trees growing in the driest conditions. This reduction in water transport capacity was observed jointly with a reduction in leaf transpiring area. Canopy leaf area decreased by 18% in the dry treatment as a consequence of the throughfall exclusion, which was found to validate the ecohydrological equilibrium theory.

SummaryUnderstanding whether tree growth is limited by carbon gain (source limitation) or by the direct effect of environmental factors such as water deficit or temperature (sink limitation) is crucial for improving projections of the effects of climate change on forest productivity.We studied the relationships between tree basal area (BA) variations, eddy covariance carbon fluxes, predawn water potential (Ψ pd ) and temperature at different timescales using an 8-yr dataset and a rainfall exclusion experiment in a Quercus ilex Mediterranean coppice.At the daily timescale, during periods of low temperature (< 5°C) and high water deficit (< À1.1 MPa), gross primary productivity and net ecosystem productivity remained positive whereas the stem increment was nil. Thus, stem increment appeared limited by drought and temperature rather than by carbon input. Annual growth was accurately predicted by the duration of BA increment during spring (Dt t0-t1 ). The onset of growth (t 0 ) was related to winter temperatures and the summer interruption of growth (t 1 ) to a threshold Ψ pd value of À1.1 MPa.We suggest that using environmental drivers (i.e. drought and temperature) to predict stem growth phenology can contribute to an improvement in vegetation models and may change the current projections of Mediterranean forest productivity under climate change scenarios.