Refining plant traits in vegetation models using forest inventory and LAI measurements. An application to the modelling of Cedrus atlantica in the Rif Mountains with the CARAIB model

It appears today established that climate change will alter biodiversity, since the migration speed of many species, especially plants, are presumably too small to follow climate change. Mountain ecosystem floras of Mediterranean regions are particularly vulnerable to the climatic threat, because they combine high ecosystem diversity and large proportion of endemic species, with the risk of reaching the summits of the mountains which would limit their migration. Moreover, these environments are often strongly impacted by man. Being able to identify and predict the areas favourable to the species – microrefugia - becomes crucial in view of the fragmentation of the space devoted to their conservation. Dynamic vegetation models (DVMs) are well-designed tools for performing such projections, since they incorporate the physiological effects of CO2. However, they are usually run at the plant functional type level (PFT), whereas conservation studies require specific projections for each individual species. Thus, some efforts focus now on applying DVMs at species level, refining the definition of morphophysiological parameters from initial PFT traits to specific traits collected in the field or found in trait databases.
Here we simulated the modern distribution of Cedrus atlantica, an endangered species of the north Africa mountains with the CARAIB DVM (Dury et al., iForest - Biogeosciences and Forestry, 4:82-99, 2011), over the Rif Mountains. Model results in terms of biomass and NPP are evaluated against data coming from forest inventory and LAI measurements. Morphological traits of C. atlantica derived from plant material collected in situ (such as specific leaf area, C:N ratio of leaves, etc) are adapted in the model simulation. CARAIB is run at high resolution using either climatic inputs derived from the Climate Research Unit climate dataset combined with WorldClim climatology at 30 arc sec or the ouputs of a 5 km resolution simulation of the regional climate model MAR (Fettweis et al., The Cryosphere, 7 :469-489, 2013) over the focal area.