Explicit avalanche-forest feedback simulations improve the performance of a coupled avalanche-forest model

• Published in: Ecological complexity 2014-03, Vol.17, p.56-66
• Main Authors: Zurbriggen, N., Nabel, J.E.M.S., Teich, M., Bebi, P., Lischke, H.
• Format: Article
• Language: English
• Subjects: Avalanche-forest feedback; Explicit feedback simulation; Forest landscape model; Mountain forests; Sensitivity analysis; TreeMig-Aval
• ISSN: 1476-945X
• DOI: 10.1016/j.ecocom.2013.09.002

•TreeMig-Aval is a spatially explicit forest model coupled with an avalanche module.•The simulated avalanche-forest feedback depends strongly on environmental factors.•We evaluated sensitivity to slope steepness, additional disturbances and temperature.•Model complexity could not be reduced by omitting the feedback or forest dynamics.•Explicit simulation of avalanche-forest feedbacks is crucial for treeline simulation. Many temperate and boreal mountain landscapes are strongly affected by snow avalanches. Forests can reduce avalanche release probability, leading to a positive feedback between forests and avalanches. The effects of this feedback, especially when influenced by changing environmental conditions, make the projection of the future developments of mountain forests and avalanches challenging. In order to study this feedback under a wide range of environmental situations, we coupled a forest landscape model with a new probabilistic avalanche module. The coupled model TreeMig-Aval allows yearly spatially explicit simulations of climatically driven forest dynamics, with species-specific growth, mortality, and reproduction. Simulated spatially explicit avalanche release is driven by climate, topography, forest type and density. These factors, together with additional factors increasing tree mortality, influence the strength of the positive feedback between forests and avalanches. We investigated (a) the influences of the three environmental factors temperature, slope steepness, and additional mortality on the simulated dynamics of mountain forests and avalanches, (b) the plausibility of TreeMig-Aval, and (c) whether the complexity of TreeMig-Aval could be reduced. The sensitivity of avalanche release probability to environmental changes was thus compared between TreeMig-Aval and two simplified model versions. The three environmental drivers had strong and often nonlinear influences on the simulated forest and avalanche dynamics. The simulated avalanche release probability showed linear to sigmoidal decreases with temperature, a peak-shaped response to slope steepness, and steep sigmoidal increases with additional mortality. However, these response shapes of avalanche release probability to each environmental factor changed along the axes of the two other factors studied. These interactions suggest that future mountain forest simulation studies should explicitly account for the influence of environmental drivers on the avalanche-forest feedback. The s