The relative contribution of sediment food ( e. g. organic matter, carbohydrates, proteins, C, N, polyunsaturated fatty acids) and environmental variables ( e. g. oxygen, pH, depth, sediment grain size, conductivity) in explaining the observed variation in benthic macroinvertebrates is investigated. Soft bottom sediments, water and benthic macroinvertebrates were sampled in several water systems across The Netherlands. The variance partitioning method is used to quantify the relative contributions of food and environmental variables in structuring the benthic macroinvertebrate community structure. It is assumed that detritivores show a significant relationship with sediment food variables and carnivores and herbivores do not. The results of the variance partitioning method with data sets containing only detritivores, herbivores or carnivores confirm this assumption. This indicates that the variance partitioning method is a useful tool for analyzing the impact of different groups of variables in complex situations. Approximately 45% of the total variation in the macroinvertebrate community structure could be explained by variables included in the analyses. The variance partitioning method shows that sediment food variables contributed significantly to the total variation in the macroinvertebrate dataset. The relative importance of food depends on the intensity of other environmental factors and is lower on broad spatial scales than on smaller scales. The results of the partitioning depend on the selected variables that are included in the analyses. The method becomes problematic in case variables from different groups of variables ( e. g. one food variable and one environmental variable) have a high inflation factor and thus are collinear. The choice of the variable that is left out impacts the variance allocated to the different groups of variables. The variance partitioning method was able to detect the spatial scale dependent contribution of food variables in structuring macroinvertebrate communities. This spatial scale dependency can also be caused by the size, the composition, and the heterogeneity of the dataset. Performing extra analyses in which specific samples are removed from the original dataset can give insight in under- or overestimation of the impact of certain factors and offers the possibility to test the robustness of the obtained results.
|Publication status||Published - 2004|
- multivariate analyses
- fish communities