| Summary: | In England and Wales, eutrophication, physical habitat modification and water abstraction have been identified by the Environment Agency as the most significant stressors preventing freshwaters from reaching Good Ecological Status. A similar pattern has emerged across the European Union. These stressors often exist in combination; however understanding of their ecological impact has tended to be based on the spatial pattern of individual stressors. This thesis investigates relationships between multiple stressors and standard river macroinvcl1ebratc community metrics. A general framework using linear multilevel (or mixed effects) models is developed, allowing integrated analysis of selected historical monitoring datasets which have hierarchical and spatial-temporal structure. Multi-model inference is used to define multiple partial relationships, including interaction terms (describing synergistic I antagonistic effects), without over-fitting or neglecting tapering effects. Initially, partial relationships are demonstrated with seasonal antecedent high and low flows. A cross-level interaction term is used to deduce that compared to more natural channels, the macroinvertebrate community in more resectioned channels appears more sensitive to antecedent low flow. Furthermore, there is the suggestion that although macroinvertebrates in lowland rivers arc more sensitive to low flow and extent of channel resectioning separately. together these may have. a greater •effect in upland rivers. For a series of Chalk stream sites, supra-seasonal drought is shown to be associated with incremental ecological decline, and there is a particular partial relationship between the autumn community and summer flows two years prior to sampling. Trampling of the channel/banks by livestock (poaching) was also associated with lower metric scores. Spatial associations between stressors and reference-corrected WFD monitoring metrics are shown to exist: arable and improved grassland cover in the nearby landscape, phosphorus (SRP) concentration in the water and extent of channel/bank resectioning (negative). A positive relationship between the metrics and riparian woody vegetation exists both in space and through time. In general, partial relationships could be estimated despite the inevitable co-occurrence of these stressors. The negative effects of arable land cover may have reduced since 1990, and there was a suggestion of greater sensitivity to SRP in upland rivers. Although the results are statistical associations and would benefit from testing in a more experimental manner, they are relatively robust given certain caveats. They have implications for the management of stressors as part of Water Framework Directive (WFD) Programmes of Measures. Impacts of water abstraction must be considered in the light of channel modification and climate (drought status). Whilst it makes sense to focus on ameliorating eutrophication when it is dominant, addressing land management stressors, including riparian and channel habitat would still be needed to achieve WFD goals. The study demonstrates the wealth of knowlcdge that can be gained from existing monitoring datasets using analytical techniques which respect the structure of the data. However an overly risk-based monitoring approach may compromise our capability to extract this knowledge. This study provides a basis for future work, needed to address potential non-linear responses, interactions of habitat modification and supra-seasonal drought, response of individual taxa and generalisation to other biological quality elements.
|
|---|
