Résumé
In the Mediterranean basin, precipitation is expected to decline by the end of the 21st century as a consequence of climate change. A throughfall exclusion experiment has been established in a Quercus ilex coppice in southern France to anticipate the response of this ecosystem to a reduced precipitation amount by removing part of the throughfall. The experiment involved four 140 m(2) plots: a control plot, a throughfall exclusion plot, a thinned plot and a throughfall exclusion in a thinned plot. This experiment has been used during nearly two years to monitor throughfall amounts in the different plots. Additionally, stemflow has been recorded on 20 trees between 25 April 2006 and 15 May 2007. The control plot had a stem density of 6885 stems ha(-1) and a leaf area index of 3.1, rainfall partitioning into interception loss, throughfall and stemflow was, respectively, 30.9%, 56.6% and 12.5% of the total precipitation (1605 mm over the study period). Thinning reduced the stem basal area by 33% and the total interception loss by 34.6%. Given the rainfall partitioning, the throughfall exclusion was found to remove 19% of total precipitation, which was confirmed by soil water content measurements. The revised Gash analytical model predicted interception with an agreement of -6.2% of total precipitation. Results show a good transferability of the model to a plot with a different stem density, thus supporting the idea that model parameters and evaporation from a wet canopy scale Linearly with canopy cover. (C) 2008 Elsevier B.V. All rights reserved.