Spatial models as a tool to identify spatial patterns of surficial sediment composition and their contributing factors in the littoral zone of Lake Constance (Germany)

摘要

For a lake-wide investigation of littoral surface sediments, we collected approximately 800 samples from Lake Constance along 235 parallel transects of 1 km distance along the shore at water depths of 1, 2, 4 and 8 m. Sediment samples were analyzed for the presence of spatial patterns in mineralogical and granulometric composition and contents of carbon and sulphur, with the purpose of identifying general and lake specific processes that contribute to the observed patterns. In particular, we wanted to test if the general factors known for contributing to littoral sediment composition could be revealed by spatial modeling of a dataset based on a systematic sampling scheme. Explanatory variables for a regression model were derived from GIS-analyses of the dataset, and also from other available data on Lake Constance. Highest levels of variance explained by the model (30-40%) were reached for the parameters calcite, clay minerals, sulphur and inorganic carbon content. Regional patterns in sediment composition, described by the proportion of explained between-transect variance (VE_t) in the model, are explained by up to 70%. The observed east-west gradient of mineralogical sediment composition is ascribed to the respective regional patterns of allogenic and endogenic sources. Allogenic minerals such as clay, dolomite and quartz dominate the littoral surface sediments of the eastern part of the lake and the areas near river mouths. Endogenic minerals like calcite dominate the littoral sediments of the western parts of the lake, reflecting internal process of biogenic carbonate precipitation that corresponds to the distribution patterns of submersed macrophytes (particularly Charophytes). This study confirms the major factors contributing to littoral sediment composition found in previous studies. However, their explanatory power for spatial patterns of a specific lake will be weak if respective spatial patterns of influence are neglected. The present study provides a guide to future sampling schemes and corresponding spatial statistics for lake specific application of general models. It also provides a basis to support engineering decisions for specific lake development.