Linking Complex Vegetation Patterns to Topographic Factors Derived from a High-Resolution Digital Terrain Model in a Mountain Mire of Hokkaido, Northern Japan
Аннотация
Peatlands are important ecosystems that store vast amounts of carbon, support biodiversity conservation, and mitigate flooding. Topography, including fine-scale variation, contributes to the formation of mire vegetation patterns by creating heterogeneous hydrological conditions. However, fine-scale drone-derived terrain data have not been widely utilized in the study of topography-vegetation relationships. In this study, we investigated how the vegetation distribution within Ukishima Mire, a mountain mire with complex fine-scale topographic patterns in Hokkaido, northern Japan, could be explained by topographic factors. The mire is located at an elevation of 870 m and covers an area of 0.15 km^2. A digital terrain model (DTM) was generated using drone photogrammetry. Eighteen topography-related variables were derived from the DTM. Across the mire, we recorded the locations of nine herbaceous vegetation types, which consisted of Sphagnum mosses, sedges, grasses, shrubs, and a disturbed type. We constructed a random forest model to classify vegetation types from topographic variables and assessed its predictive performance. The overall prediction accuracy was 67%, indicating that topographic variables captured a substantial component of vegetation patterns. The influential variables included proximity to open water surface, altitude, relative height to surroundings, terrain ruggedness, proximity to forest edge, slope, and topographic wetness index. We detected vegetation turnovers at multiple spatial scales along the gradients of topographic factors related to groundwater and surface water availability. These results demonstrate the potential of high-resolution DTM-derived variables for identifying multiscale topographic gradients associated with complex mire vegetation patterns.