Importance of climate uncertainty for projections of forest ecosystem services

Abstract

Mountain forests provide a wide range of ecosystem services (ES, e.g., timber production, protection from natural hazards, maintaining biodiversity) and are especially sensitive to climate change. Dynamic vegetation models are commonly used to project climate change impacts on forests, but the sensitivity of process-based forest landscape models (FLMs) to uncertainties in climate input data has received little attention, especially regarding the response of ES. Using a dry inner-Alpine valley in Switzerland as a case study, we tested the sensitivity of a process-based FLM to different baseline climate data, lapse rates, and future climate change derived from different climate model combination chains and downscaling methods. Under the current climate, different sources of baseline climate accounted for the majority of the variation at lower elevations, while differences in lapse rates caused large variability at higher elevations. Under climate change, differences between climate model chains were the greatest source of uncertainty. In general, the largest differences for species were found at their individual regional distribution limits, and the largest differences for ES were found at the highest elevations. Thus, our results suggest that the greatest uncertainty for simulating forest ES is due to differences between climate model chains, and we recommend using as many climate scenarios as possible when projecting future forest response to climate change. In addition, care should be taken when evaluating climate impacts at landscape locations that are known a priori to be sensitive to climate variation, such as high-elevation forests.