The influence of different spatial-scale variables on caddisfly assemblages in Flemish lowland streams

Abstract

1. Patterns in species assemblages are the result of the combined influence of processes acting on different spatial scales. Various studies describe the distribution of macroinvertebrate communities and their relationship with environmental factors at different geographical scales, but only a few of these studies concentrate on Western European lowlands. 2. Using Flanders as representative for the densely populated Western-European lowlands, the specific aims of this study are: (i) to identify the different trichopteran species assemblages and to characterise them biologically using indicator species; (ii) to determine which environmental gradients most influence the observed species assemblages; and (iii) to analyse the relative importance of different spatial scale variables in constraining the Trichoptera distributions. 3. Assessment of the main environmental gradients suggested that the absence of Trichoptera from certain locations was mainly due to elevated nutrient concentrations and lower oxygen contents, confirming their sensitivity to anthropogenic disturbance. 4. Five Trichoptera species assemblages were distinguished based on Bray–Curtis dissimilarity coefficients. These assemblages did not differ significantly in species richness, but a shift in stream zonation preference was observed. In the ordination analysis 11 variables that were selected using a stepwise model building function manifested themselves as upstream–downstream and size-related gradients. The Trichoptera assemblages in lowland streams thus appear to follow a longitudinal succession pattern that corresponds with the species-specific preferences. 5. Partitioning the variance over the different spatial scales indicated that the reach-scale variables were far more important in explaining the variation in species composition. The study design, which limited the minimum–maximum range of catchment-scale characteristics, however, may have led to an overestimation of the impact of the local-scale variables.