Raster-based detection of vegetation patterns at landscape scale levels
Plant communities are more or less regularly combined in space. The main goal of the presented study is the detection of those spatial patterns in the most replicable way. The study took place in the Northern Alps near Salzburg. The whole area of approx. 8 km2 has been covered with a regular and equal-sized plot raster. The plots have been used for an inventory of plant communities. The statistical analysis of the relevés revealed regular combinations of plant communities (= vegetation complexes). Classification and ordination approaches have been applied to the data in order to get a better idea of the vegetation structure. Using a geographical information system (GIS), classes and ordination scores have been transferred into spatial patterns. Taking advantage of GIS techniques, there are also possibilities of causal analyses of the spatial structure of vegetation. The even-sized plots offer a scale-dependent cross-section of the vegetation structure. A major problem is the "modifiable areal unit problem": i.e., placement and dimension of the plots have an influence on the results of the analysis and may distort or hide patterns. Further problems are caused by the underlying classification of plant communities. If patterns of spatial plant community combinations are the object of research, raster-based detection provides methods that are relatively replicable. The schematic sample design is a convenient base for geo-statistical analyses, e.g. on biodiversity at various scale levels. The approach is useful if quantifiable and reliable information about the spatial vegetation structure is required, and if the mapping of plant community boundaries is too time-consuming or too little replicable. This is especially valid for monitoring purposes.