The influence of riparian vegetation on macroinvertebrate community structure and functional organization in six new Guinea streams

Abstract

Information on the ecology of New Guinea streams is meagre, and data are needed on the trophic basis of aquatic production in rivers such as the Sepik in Papua New Guinea which have low fish yields. This study investigates the relationship between riparian shading (from savanna grassland to primary rainforest), algal and detrital food, and macroinvertebrate abundance and community structure in 6 Sepik River tributary streams. A particular aim was to elucidate macroinvertebrate community responses to changes in riparian conditions. All streams supported diverse benthic communities, but morphospecies richness (overall total 64) was less than in streams on the tropical Asian mainland; population densities of benthic invertebrates, by contrast, were similar to those recorded elsewhere. Low diversity could reflect limited taxonomic penetration, but may result from the absence of major groups (Plecoptera, Heptageniidae, Ephemerellidae, Psephenidae, Megaloptera, etc.) which occur on the Asian mainland. Population densities of all 19 of the most abundant macroinvertebrate taxa varied significantly among the 6 study streams, but community composition in each was broadly similar with dominance by Baetidae and (in order of decreasing importance), Leptophlebiidae, Orthocladiinae, Elmidae and Hydropsychidae. Principal components analysis (PCA) undertaken on counts of abundant macroinvertebrate taxa clearly separated samples taken in two streams from the rest. Both streams contained high detrital standing stocks and one was completely shaded by rainforest. Stepwise multiple-regression analysis indicated that population densities of the majority of abundant taxa (11 out of 19) across streams (10 samples per stream; n = 60) were influenced by algae and/or detritus, although standing stocks of these variables were not clearly related to riparian conditions. When regression analysis was repeated on mean counts of taxa per stream (dependent variables) versus features of each stream as a whole (thus n = 6), % shading and detritus were the independent variables yielding significant regression models most frequently, but pH, total-nitrogen loads and algae were also significant predictors of faunal abundance. Further regression analysis, undertaken separately on samples (n = 10) from each stream, confirmed the ability of algae and detritus to account for significant portions of the variance in macroinvertebrate abundance, but the significance of these variables varied among streams with the consequence that responses of individual taxa to algae or detritus was site-specific.