Context  Land use and other human activities generally reduce downstream water quality, with consequences for ecosystem services and freshwater biodiversity. However, the relative effects of different types of land use and associated activities across large spatial scales can be difficult to understand, and thus to address, in heterogeneous landscapes.

Objectives  To assess how different types of landscapes affect the water quality and aquatic biodiversity downstream. To assess if coffee agroforestry landscapes are intermediate between forests and agriculture in terms of their effects on biodiversity in streams.

Methods  We measured water quality parameters and sampled stream macroinvertebrates in 46 catchments with perennial streams draining through mosaic landscapes managed by smallholder farmers in southwest Ethiopia. For each of the catchments, we estimated the proportion of each land use, including agriculture, coffee agroforestry and forests, estimated the settlement density and calculated an index representing the number of and distance to coffee washing stations.

Results  We found that water quality was better (e.g. high dissolved oxygen, low turbidity and low nutrient concentration) for streams draining through landscapes with high forest cover as compared to landscapes dominated by agroforestry or agriculture. Furthermore, we found elevated Escherichia coli (E. coli) concentrations in most streams, especially those draining agroforestry landscapes. The composition of macroinvertebrates varied much across streams with sensitive taxa dominating clean streams and vice versa. This pattern became even clearer when we analysed the direct effects of settlement density and coffee washing stations on the abundances of families with different sensitivity to poor water quality, but there were also inconsistent responses in some groups.

Conclusion  Our finding suggests that a single measurement of water chemistry alone inadequately captured land use effects on water quality, highlighting the importance of including biodiversity assessment. Since people frequently use these streams for various purposes and freshwater biodiversity is at stake, our findings highlight the urgent need of measures to reduce these negative effects. Such measures should focus on agriculture and agroforestry-dominated landscapes to primarily reduce the negative impacts of wastes from settlement areas and wet coffee washing stations.

Water quality and habitat conditions influence community composition of aquatic biota because different taxa vary in their tolerance to environmental stress. Adult dragonflies and damselflies are conspicuous insects that are associated with freshwater habitats in their larval stage and often also as adults. We sampled adult dragonflies and damselflies (collectively ‘odonates’ from the order Odonata) from 30 small streams flowing through heterogeneous landscapes in the Afromontane landscapes in southwestern Ethiopia and asked what environmental factors are driving the variation in species composition and to what extent odonates could serve as bioindicators of stream water quality. We found that the composition of odonate species differed between streams varying in shade from surrounding trees, presence of adjacent wetlands as well as with varying water quality, affected by upstream land-use. We found a number of species endemic to the Ethiopian highlands predominantly associated with streams draining more forested catchments. Different aspects of water quality could be differently well predicted by the species presence/absence. For example, the level of dissolved oxygen was better predicted than E. coli loads (cross-validated R2 = 0.51 vs 0.36, respectively). Combinations of different species explained the variation in different water quality variables. However, no single species was a positive or a negative indicator for all water quality parameters, even if the presence of Pinheyschna waterstoni and Platycypha caligata were consistently associated with good water quality whereas Pseudagrion kersteni and Pseudagrion spernatum were consistently associated with poor water quality conditions. We conclude that there is a large variation in dragonfly and damselfly species composition along streams and that the species of greatest conservation concern mostly are associated with streams draining less disturbed catchments. However, before using odonates as indicators of water quality, the target quality parameters need to be defined (e.g., low turbidity or high oxygen). Moreover, it is worth noting that in this stream-habitat context there are as many indicators of poor as of good water quality among the studied species. Given the conspicuous appearance of odonates and the ease and low cost of sampling, we encourage further development and tests of the value of using them as bioindicators.

Small streams provide key ecosystem services essential for the livelihoods of rural communities, but both drought and flooding could be harmful. To assess temporal variations in water flow along a land use gradient, we implemented a participatory stream monitoring approach in southwestern Ethiopia. We selected 16 catchments with different land use composition (e.g forest, agroforestry, annual crops) and trained local farmers living near the outlets to record water depth twice a week over more than two years covering three rainy seasons and two dry seasons. During this period, we also measured discharge five times at different seasons. Using the water depth records and discharge we calculated estimated runoff (mm/hour) and produced a time series for each stream. We modelled runoff as the function of land use, irrigation practice, precipitation and slope. We also interviewed 28 farmers living close to the outlets and assessed their perception of water availability and flood risks. There was strong seasonal variation in runoff related to the rainfall patters, with high flows in the midst of the rainy season July-August and low flows in the dry season January-February, exposing farmers to both flooding and water scarcity. The dry season flow was reduced in catchments with irrigation practices, and surprisingly also in catchments with high forest cover. No significant relationships were found between rainy season run-off and land use composition. We found that smallholder farmers experienced irrigation water shortages during the dry season and flooding during the rainy season, but these challenges were not significantly associated with the measured runoff metrics. Despite that this region is known for its abundance of surface water, our results suggest that upstream small-scale irrigation can be a threat to downstream farmers access to stream water. The participatory monitoring approach demonstrated the huge variability in flows. However, even more close monitoring of discharge patterns is needed to understand mechanism behind the variability among streams and to what extent land-use restoration could reduce drought or flood risks.

Water security represents a critical challenge worldwide especially in low-income countries due to poor water quality, insufficient water quantity, climate change and low investments in water supply infrastructure. However, water security challenges may vary across different catchments in the same landscape. We interviewed 83 smallholder farmers in southwestern Ethiopia to understand what water sources they used during different seasons and how that varied between catchments with different land use. We also investigated what water-related problems the farmers had and how they varied across catchments with different land use. The land use composition for each household’s upstream catchment was estimated using a land use map. Although smallholder farmers rely on multiple water sources, protected springs are the dominant source for drinking water. Farmers in catchments dominated by forests relied more on open sources such as unprotected springs than farmers in agroforestry or open agricultural landscapes. Farmers experienced a variety of water-related problems, such as water-related illnesses owing to poor-quality water sources, that fetching water required excessive time, and that women and children were sometimes exposed to violence related to water use conflicts. In coffee dominated landscapes, wastewater from wet coffee processing polluted the water downstream, used by smallholder farmers for washing clothes, hygiene and to water their cattle. Since there is a variation in water source types and associated problems between different types of landscapes, different interventions are needed to resolve problems at the local scale related to source-specific problems and at the catchment scale related to upstream catchment land use.