Assessing bird functional diversity in the Soyang Lake Watershed with the aid of airborne LIDAR and the forest landscape model LandClim
TERRECO Cluster A-07
From 03/2013Principal Investigator: Björn Reineking, Chan Ryul Park, Dowon Lee
Staff: So-Yeon Bae
Abstract 2013: While ecosystem management in agricultural landscapes can be modified in a short time, potentially leading to rapid changes in landscape function, the results of forest management are evidenced only after decades. Specific tools are required to link forest processes such as succession, and management measures such as stand thinning, to services in biomass production, advantageous maintenance of water balances, the export or retention of materials and biodiversity. The forest landscape model LandClim has been specifically developed to assess the effects of climate change, disturbance, and management options on forest dynamics, and provides a framework for investigating many dimensions of the forested landscapes of Soyang Watershed (cf. the poster session).
In this study, LandClim is used with other methods to address general questions about habitat suitability for bird populations. Our long-term goal is to assess bird functional diversity within the Soyang Watershed by modeling the distribution of different bird functional groups. In a second step, we aim to project future changes in bird functional diversity as a consequence of endogenous succession, climate change, and forest management. We employ airborne Light Detection And Ranging (LiDAR) data to detect vertical and horizontal structure of forest stands with defined composition and LandClim to extrapolate this information to forested landscapes.
In a preliminary study, we evaluated the utility of LiDAR data to identify habitat for an endangered forest species, the hazel grouse (Bonasa bonasia), in the Bavarian Forest National Park in southeast Germany. We created a hazel grouse distribution model using boosted regression trees (BRTs), and then compared it to data from field inventories and aerial photo interpretations. We found that the cross-validated discrimination ability of the model using LiDAR data was slightly higher (as measured by AUC, 0.79) than obtained by using aerial photography (AUC 0.76) or field survey data (AUC 0.78). The predictive performance consistently increased with the adding of more predictors from different sources, indicating that the three data sources complemented each other. Each data source is likely to have an advantage in constituting one of three key aspects of the hazel grouse habitat, namely vertical stand structure, horizontal mixture of successional stages, and preferred deciduous trees as a better food resource. We concluded that public LiDAR data can be used to produce habitat suitability maps over large areas, and should be used more widely to detect valuable wildlife habitat. During the second phase of the TERRECO project, these methods will be extended from habitat analyses for a single species to evaluating habitat for the general bird functional diversity in the Soyang Watershed. Model development will be based on a future field survey. Future forest structures as simulated by LandClim will then be interpreted with the help of the newly developed statistical models of bird functional diversity.
Keywords: species distribution model, bird functional diversity, forest structure, forest landscape model
