Abstract
Most ecosystems are exposed to multiple anthropogenic stressors that may interact additively or non-additively, where the combined stressor effects are larger or smaller than expected based on their individual effects, and therefore cause “ecological surprises”. Among the known threats for freshwater biodiversity and ecosystem processes, agriculture and water abstraction for irrigation are important stressors of stream ecosystems, especially in drier parts of the world. Due to global climate change, both the number of temporary rivers and the severity of flow intermittence are expected to increase, and there is an urgent need to improve our understanding of the consequences of water abstraction for stream ecosystem functioning. Agricultural development can impose a number of stressors in streams, including augmentation of nutrients in the water and fine sediment on the stream bed. Water abstraction, on the other hand, reduces discharge and current velocities downstream from water takes. These changes to the hydrological regime can be expected to further enhance the accumulation of fine sediment, and to change in-stream temperature regimes and the rate of processing of organic matter. The wide range of consequences makes it difficult to tease apart the individual and combined effects of agricultural development and water abstraction on stream ecosystems.
To study the effects of multiple stressors on stream ecosystem structure and function, I surveyed 43 stream sites in the Manuherikia River catchment in the Central Otago region of New Zealand. Strategic choice of study sites meant that agricultural intensity and water abstraction intensity were uncorrelated. I used general linear models and an information-theoretic model selection approach to identify the best models predicting the interactions among farming intensity, water abstraction and related physicochemical variables for a series of biological response variables, including fish presence and densities (Chapter 2), invertebrate taxa and 15 biological traits (Chapter 3), stream algal traits (Chapter 4), and δ15N isotope ratios of primary consumers as indicators of nitrogen dynamics (Chapter 5). I determined whether patterns in these response variables followed positive, negative or unimodal relationships along stressor gradients and whether interactions occurred among paired stressors. For the algal study, I also developed and tested a conceptual model and compiled a detailed database for ten biological traits of stream algae because to date no such comprehensive trait-based approach exists for stream algae.
Algal and invertebrate traits proved to be suitable tools for disentangling the mechanisms by which multiple stressors affect stream communities. Overall, when focusing on stressors operating at the landscape scale, I found that farming intensity had stronger effects on algae, invertebrate, fish and ecosystem processes than water abstraction. The latter seemed to affect invertebrate communities mainly through changes in organic matter dynamics and increased levels of deposited fine sediment. My results also suggest that farming intensity affected algal communities mainly through changes to in-stream nutrient and light regimes and invertebrate and fish communities through reduced habitat availability and altered food supply. Primary consumer stable isotope ratios responded to well-known stressors (% agriculture in the catchment and stream water nitrogen concentrations), and to the previously unstudied stressors of water abstraction and fine sediment deposition. An important result in the context of developing this bioindicator as a measure of nitrogen dynamics is that three common consumer taxa showed distinctively different response patterns to the studied stressors. This also implies that δ15N isotope ratios are not stressor-specific but can be affected by multiple environmental variables.
Overall, biological trait metrics and traditional metrics of community structure performed similarly well in detecting the effects of multiple stressors on stream invertebrate and algal communities along the catchment-scale and in-stream stressor gradients (with the exception that structural invertebrate metrics showed stronger relationships to nutrient concentrations). However, biological traits have the advantage of being applicable over much larger geographical scales than their taxonomic counterparts, thus allowing comparisons on a global scale.
While abstraction was generally less influential than agriculture, these landscape-scale stressors often interacted antagonistically and managers need to be aware of their cumulative effects. At the reach-scale, sediment and nitrogen enrichment were both influential stressors; therefore particular attention also needs to be paid to minimising overall catchment and stream bank erosion. However, deposited sediment often acted in concert with nitrogen concentrations on invertebrate communities and the combined response (antagonistic or synergistic) was not predictable on the basis of single-stressor patterns. Water quality in the Manuherikia main stem can be regarded as ‘good’, but there is pressure to expand the area of intensively managed pastures in New Zealand. However, my combined findings suggest that the proportion of highly-intensively managed pastures upstream from a given stream reach should not exceed 40% if managers wish to maintain trout populations and not exceed 7% to avoid marked changes in invertebrate communities. Similarly, the amount of stream flow reduction should not exceed 40% to provide for near-natural algal and invertebrate communities.