From May to August 2004, the available food base and diet of two cyprinid species, eurytopic roach, represented by 75 small specimens, and rheophilic dace, represented by 37 small specimens, were investigated in seasonal patches of submersed aquatic macrophytes (SAM) in the impounded lowland Warta River, Poland. The aim of the study was to recognize spatial and temporal patterns in the feeding of both species in relation to available resources, and to evaluate their food niche overlap. To distinguish homogenous classes of fish alimentary tracts on the basis of their contents, a Kohonen artificial neural network (i.e., a self-organizing map, SOM) was used. Indicator food categories were identified using the IndVal index. Roach and dace partitioned the food niche, which was demonstrated in this study by 1) insignificant values of the Schoener's interspecific diet overlap index on particular sampling occasions, and two SOM sub-clusters (homogenous diet classes) with alimentary tracts almost exclusively of roach (axis of resources), 2) zero or low percentage of specimens of both fish species coming from the same sampling occasions and assigned to any of the remaining homogenous diet classes (axis of time), and 3) absence of roach in May and dace in August in the SAM patches, i.e., on almost half of the sampling occasions (axis of space). The diet overlap was highest when the SAM patches and food base were most developed, which is congruent with the niche overlap hypothesis saying that maximal tolerable niche overlap can be higher in less intensely competitive situations.We recommend the combined application of SOM and IndVal index, which have both previously been used in biocoenology, to the analyses of animal diets. They effectively allowed getting insight into the complex trophic relationships.
Read full abstract