Dissertation presented in partial fulfilment of the requirements for the degree of Doctor of Science (PhD): Biology) December 2020
River ecosystems encompass river channels and its floodplains and form a diverse mosaic of habitats upon which countless species of animals and plants depend for survival. They provide a plethora of services for humans including a source of clean water for domestic and industrial uses, a source of food, a means of waste disposal, transportation, power production, and sites for the pursuit of leisure activities. Yet, they belong to the most threatened ecosystems on earth. Major threats to river ecosystems include habitat degradation, water pollution, flow modification, overexploitation, and invasion by exotic species. This is especially true for (sub) tropical developing countries where intensification of land-use for agriculture and poor disposal of untreated waste have markedly degraded rivers and associated floodplain ecosystems. Nevertheless, a proper understanding of ecosystem functioning and biological diversity is lacking. In this Ph.D., we contribute to bridging this knowledge gap. We investigate different factors that explain biodiversity and ecosystem quality in (afro) tropical river systems and associated temporary pool ecosystems in northeastern-Tanzania by using macroinvertebrates as biological indicators and collecting environmental and biological data at various spatial-temporal scales. Firstly, we assessed how seasonality (i.e., wet and dry seasons) influences macroinvertebrate community structure and water quality conditions (Chapters 1 and 2). An extensive repeated-sampling survey was conducted to measure water quality, macroinvertebrates, and other presumed important environmental variables in the two sub-catchments of the Upper Pangani River Basin (UPRB). We found evidence that water quality conditions and macroinvertebrate assemblages differ between seasons and that these differences are associated with high flow velocity, and runoff carrying sediment and nutrient loads from the catchment area to the river systems during the rainy season. Moreover, our results revealed that chlorophyll-a, oxygen and phosphorous (dry season), nitrogen and turbidity (wet season), and substrate composition and agricultural land-use (both seasons) are important determinants for the variation in macroinvertebrates assemblages between sites. We also attempted to identify indicator taxa linked to specific water quality conditions and found that families Hydropsychidae (Trichoptera), Potamonautidae (Decapoda), Baetidae (Ephemeroptera), and Heptageniidae (Ephemeroptera) showed to be indicator taxa of good water quality conditions, while Hirudinea (Annelida) and Chironomidae (Diptera) appeared to be indicator taxa of poor water quality conditions (Chapter 1). Secondly, we focused on the impact of land-use at different spatial scales on river quality (Chapters 1 and 2). To quantify this we used three different spatial methods of land-use estimation; (i) land-use of the entire watershed area above the monitoring site, (ii) a circular buffer around a monitoring site, and (iii) a circular buffer immediately upstream of a monitoring site, with circular buffers varying from 100m to 2km. The land-use percentage compositions in the dry and wet seasons were quantified using Landsat-8 satellite images with a maximum mapping resolution of 30m. We found that physico-chemical water quality and macroinvertebrate assemblages responded differently to land-uses at different scales in both dry and wet seasons. Nevertheless, the relationships were not always straightforward and clearly scale-dependent, suggesting that the spatial estimate used, and the spatial scale considered can strongly confound the conclusions (Chapter 2). Land-use of the entire watershed area upstream of the monitoring site better explained variation in physico-chemical water quality and macroinvertebrate indices whereas macroinvertebrate abundances showed strong links with more local land-use patterns within 100m and 2km radii. In Chapter 3, we added the main constraint that is not always included in studies of river systems i.e., connectivity and spatial autocorrelation among sites. For this, we use a spatially explicit analysis framework (spatial stream network (SSN) models) to test to what extent dendritic stream network structure affects spatial patterns of benthic macroinvertebrates and water chemistry at the catchment scale. We showed that spatial patterns and spatial autocorrelation exist in stream water chemistry and macroinvertebrate indices at both fine- and broad- spatial scale comprising both flow-connected and flow-unconnected spatial relationships. And that SSN models managed to make good predictions of water chemistry concentrations and macroinvertebrate indices at unsampled sites with estimates of uncertainty. The results highlight the value of SSN models and stress the need to specify spatial dependencies representing the dendritic network structure of river ecosystems. Finally, we assessed to what extent the seasonal connectivity of the river with temporary wetlands in the surrounding landscape is a crucial determinant of aquatic communities and environmental conditions in the floodplain wetlands. This was achieved by comparing environmental conditions and diversity and composition of macroinvertebrate communities from river connected pools with endorheic pools (Chapter 4). Macroinvertebrate communities from the two habitats were clearly differentiated and spatial taxon turnover was the main determinant of variation in community composition among pools. Hydrological connectivity facilitated the migration of fish to the river connected pools which structured the invertebrate community assemblages through selective predation, particularly of large prey such as large branchiopod crustaceans. Based on our dataset we identified indicator taxa for the different habitat types and found no specific fauna unique to river connected pools. Overall, our results suggest monitoring of river systems in wet and dry seasons given the fact that different selective filters limit invertebrate assemblages in both seasons. We recommend the creation of intact riparian buffer zones of at least 60 m from each side of the riverbank to help alleviate some of the observed negative effects of the land-use activities on the river systems. In addition, conservation and management schemes of temporary pools should focus on both river connected and endorheic pools to support high regional diversity. More importantly, SSN models should be used to support river basin management in the region in a rapid and cost-effective way.