Natural forest dynamics are driven by the ecosystem processes regeneration, growth and mortality. Within this project we address natural forest dynamics from two complementary viewpoints using remote sensing. We investigate i) mortality, in the sense of natural stand-replacing disturbances, across a network of protected forest landscapes covering the entire European Alps. Zooming in on one protected forest landscape and embracing all ecosystem processes we ii) investigate the spatiotemporal dynamics of the upper and lower subalpine ecotones in Berchtesgaden National Park since the 1950ies.

Disturbances are key drivers of forest ecosystem dynamics. Under natural conditions, abiotic disturbance agents like wind, fire and avalanches, as well as biotic agents like insects, are the main causes for stand-replacing disturbances in temperate mountain forests. However, for large parts of temperate forests worldwide, the patterns of natural disturbance regimes are unknown, due to direct and indirect human influence over centuries to millennia and/or a lack of (spatially explicit) data in areas less impacted by humans. We aim to close this gap for the European Alps by quantifying the natural forest disturbance regimes of Alpine mountain forests, focusing on the distribution of patch sizes. We make use of the Alparc network (link to: https://alparc.org/) to identify 12 strictly protected forest landscapes distributed across the entire European Alps, with good representation of the prevailing ecological and climatic gradients. We derive the size of disturbance patches using satellite-based disturbance maps . Focusing on the no-intervention zones of the protected areas allows us to exclude direct human impact on disturbance regimes, and thus to characterize the natural disturbance regime with distribution functions for the patch sizes. Disturbance regimes shape the physical template for future regeneration dynamics. Our results allow the mapping of return intervals of extreme events across the European Alps. Comparing natural and observed patch size distributions further allows for informing about naturalness of disturbance regimes outside protected areas and reveals chances and challenges for nature conservation.

Focussing on Berchtesgaden National Park, we investigate the spatiotemporal development of mountain forests ecotones in the national park since the 1950ies. In particular, we are interested in the spatiotemporal development of i) the subalpine-alpine ecotone (treeline) and ii) the montane-subalpine ecotone throughout the national park. We base our analysis on historic and recent aerial images covering the period from 1953 to 2020 in nine time steps. We use deep learning to create instance-segmented (classification and object detection) forest maps from the images. To this end, we manually interpret a set of training data, consisting of randomly distributed 0.5 ha segments through all nine time steps, for convolutional neural networks. Based on parameters capturing forest structure and composition we will delineate the two ecotones under study, track their development over time, and link changes to geophysical parameters.

Berchtesgaden National Park

Alparc

Dolomiti Bellunesi National Park

Dürrenstein Wilderness Area

Ecrins National Park

Gazälli Zegerberg Wilderness Area

Gesäuse National Park

Hohe Tauern National Park

Kalkalpen National Park

Mercantour National Park

Swiss National Park

Triglav National Park

Mont Ventoux Integral Biologic Reserve