Sensitivity of forests in the European Alps to future climatic change

Model-based assessments of the impacts of climatic change on forests are confronted with 2 fundamental problems: first, there is a considerable uncertainty in the predictions of future climate; second, the forest models contain simplified parameterizations of ecological processes. In this paper, the sensitivity of forest models to different steady-state climate scenarios, to different process formulations and to different assumptions on the transient behaviour of climate is studied. The effects of 3 scenarios of climatic change and the behaviour of 5 forest gap models of the FORECE/FORCLIM family are compared at sites along an elevational gradient in the European Alps. A wide variety of species compositions may be obtained al a given site depending on the scenario of future climate. At some sites all future forests differ radically from today's forest, suggesting that these current forests are highly sensitive to climatic change. At some sites, the results of the 5 forest models differ strongly with respect to species composition and carbon storage when subjected to the same climate scenario, showing that the models need to be improved in order to arrive at reliable and robust parameterizations of abiotic and biotic influences in forest models. When comparing the effects of step, Linear, and sigmoid changes of the mean climatic parameters over 100 yr, it becomes evident that the type of change modelled is not crucial because the climatic change proceeds fast compared to the successional dynamics. It is concluded that simulations of the possible effects of climatic change on forests should be considered as sensitivity tests, not as predictions. Given the current uncertainties in atmospheric sciences (climate predictions) and in ecology (modelling of long-term forest dynamics), the most promising research strategy is to compare the effects of several climate scenarios and the projections of several forest models to arrive at slate-of-the-art ecological impact assessments.