C. Piedallu, V. Chéret, J.P. Denux, V. Perez, J.S. Azcona, I. Seynave, J.C. Gégout, Science of The Total Environment, Volume 651, Part 2, 15 February 2019, Pages 2874-2885
Several studies use satellite-based normalized difference vegetation index (NDVI) to monitor the impact of climate change on vegetation covers. Good understanding of the drivers of NDVI patterns is hindered by the difficulties in disentangling the effects of environmental factors from anthropogenic changes, by the limited number of environmental predictors studied, and by the diversity of responses according to periods and land covers. This study aims to improve our understanding of the different environmental drivers of NDVI spatial variations for different stand type characteristics of mountain and Mediterranean biomes. Using NDVI values extracted from MODIS Terra time series, we calculated Spring Greenness (SG) and annual Relative Greenness (RGRE) to depict spring and summer vegetation activity, respectively, in a contrasted area of 10,255 km2 located in the south of France. We modeled SG and RGRE at different scales, using 20 environmental predictors characterizing available energy, water supply, and nutrient supply calculated for different periods of the year. In spring, high minimum temperatures, good nitrogen availability, and acidic or neutral pH turned out to be determining for greenness, particularly for stand types located in altitude. In summer, an important soil water reserve and low temperatures promoted vegetation dynamics, particularly for stands located in areas with a Mediterranean climate. Our results show that NDVI dynamics was not only driven by climatic variability, and should not be studied using only mean temperature and rainfall. They highlight that different environmental factors act complementarily, and that soil parameters characterizing water stress and soil nutrition should be taken into account. While the factors limiting NDVI values varied according to the season and the position of the stands along the ecological gradients, we identified a global temperature and water-stress threshold when considering the whole vegetation.