|dc.description.abstract||Agriculture is a major source of diffuse contamination of surface waters by nutrients, sediment and pesticides, and streams located in agricultural regions are particularly exposed to these kinds of stressors. I combined a field survey and an outdoors stream mesocosm experiment to investigate the impacts of agricultural intensification on stream health. In the field survey, my study design included five replicates of each farming practice (organic, integrated management and conventional farming), arranged in blocks, in agricultural land dedicated to the farming of sheep and beef cattle on pasture in southern New Zealand. This survey demonstrated that streams associated with conventional, organic and integrated farming practices contrasted in their physicochemistry (particularly inorganic fine sediment on the bed and glyphosate in the fine sediment) and in both the taxonomic organization and trait representation of stream invertebrate communities. Conventional sheep and beef farming resulted in poorer stream health in general. Consequently, my field survey provided data supporting the hypothesis of a causal relationship between multiple anthropogenic stressors from agricultural land use and stream condition.
To establish strong cause-effect relationships between the two agricultural stressors identified in my field survey and the resulting ecological consequences, I performed a mesocosm experiment. Here, I investigate whether (i) increased levels of sediment and glyphosate had individual and/or additive combined effects on benthic invertebrates, algae and ecosystem functioning, (ii) increased sediment reduced the bioavailability of glyphosate and thus its toxicity (antagonistic interaction), or (iii) sediment-adsorbed glyphosate prolonged the effects of exposure (synergistic interaction). To test these hypotheses, I randomly applied four levels each of fine sediment (0, 25, 75, 100% cover) and glyphosate (0, 50, 200, 370 μg l-1) to 96 outdoor circular stream mesocosms using a full factorial, repeated measures design. Sediment was added once at the start of the 28-day manipulative phase and glyphosate for 7 days one week later, after which the mesocosms were allowed to recover for two weeks. Biological response variables included densities and community composition of invertebrates and algae, invertebrate drift, adult insect emergence and, as a measure of ecosystem functioning, leaf decay rates. Responses to my experimental stressors were analysed immediately after glyphosate addition had ceased and one week and two weeks later.
Overall, my results showed that sediment and glyphosate addition affected the benthic communities both independently and in an interactive manner and that the combined effects may persist over a relatively long period. Sediment clearly had a much more pervasive influence on invertebrate and algal response variables at both population and community levels than the seven-day herbicide pulse. Propensities of larval stream invertebrates to drift and to emerge as adult insects responded differently to the two stressors compared to the traditionally used static invertebrate density variables. For the algal response variables, several significant overall interactions detected at the community level indicated that the two stressors were acting either synergistically (in their effect on algal community evenness and on high-profile and motile algal guilds) or antagonistically (effect on the low-profile guild). Similarly, leaf breakdown responded positively to both increasing sediment and high glyphosate levels, but the two stressors acted antagonistically in their effect on leaf toughness loss. These findings add to the growing evidence that resource managers need to be aware of complex multiple-stressor effects when trying to identify thresholds of harm for stream ecosystems.||