|dc.description.abstract||Modern agriculture is necessary to feed our growing global population, yet agriculture is also the most extensive cause of freshwater ecosystem degradation. Accordingly, determining how to maintain or improve freshwater values within intensively farmed riverscapes is a key environmental challenge facing society. Broadly, my PhD thesis aims to consolidate and create knowledge to make the outcomes of stream rehabilitation in agricultural landscapes more predictable. More specifically, my aims are to 1) develop practical habitat data collection and rehabilitation planning methods to improve stream fisheries, and 2) determine outcomes for fish species resulting from rehabilitation projects set within agricultural catchments.
To this end, in Chapter 2, I review the salmonid literature to identify key attributes that can limit stream brown trout populations. I also determine numerical thresholds for stream attributes that are likely to support various trout population levels—from non-existent to thriving. This literature review (presented in Appendix 1) demonstrates there are a wide range of attributes that should be considered in order to achieve community aspirations for productive stream fisheries. The literature review underpins a Bayesian Belief Network-based decision support model (BBN). This BBN directs users to assemble a parsimonious environmental data set to inform stream fishery management. It also integrates and interrogates these data to generate standardised and testable hypotheses about which environmental factors are likely to limit trout productivity. I tested the BBN on the Horokiri Stream, a data-rich catchment in Wellington, New Zealand. The BBN results suggest that the fishery was recruitment-limited in its reference state and limited by cover and low summer-flows in its degraded state. These model results were comparable with the conclusions of five experienced fishery biologists, following their detailed investigation into the factors that led to the loss of the Horokiri trout fishery between 1951 and 1990. Chapter 2 demonstrates the suitability of BBN modeling for conducting a limiting-factor analysis on stream fish.
Mechanically reshaping stream banks is a common practice to mitigate bank erosion in streams that have been channelised and lowered for agricultural land drainage. However, the response of fish populations to this practice has rarely been quantitatively evaluated. In Chapter 3, I assess the fish and habitat responses to a catchment-scale bank reshaping event in Waituna Creek (Southland, New Zealand), a low-gradient stream that drains an intensive agricultural landscape. Fish and instream habitat data were collected before and annually for three years after the reshaping event using a Before-After-Control-Impact study design. I hypothesised that large-bodied fish, such as eels and trout, would be negatively affected by the practice. After reshaping, deposited fine sediment levels increased in impact reaches and there was also a significant reduction in longfin eel biomass (by 49%). Three years after reshaping, fish community structure had largely returned to its pre-impact state in the reshaped areas. These results show that stream bank reshaping can have substantial effects on fish populations, even in streams which are subject to regular mechanical disturbance.
In Chapter 4, I use a space-for-time substitution design to investigate the response of instream habitat and fish populations to different riparian management practices throughout the Waikakahi Stream—a Dairy Best Practice Catchment in South Canterbury (New Zealand). I found a significant negative correlation between the upstream area of stock exclusion fencing and deposited instream fine sediment cover. Furthermore, I determined that this relationship emerges when ≥ 300 m lengths of upstream riparian area were included in the analysis—indicating the scale at which stock exclusion fencing results in a desirable instream habitat response. This result shows that if a catchment-scale approach is taken to improving riparian habitat, with broad landowner buy-in and determined community leadership, then some instream habitat improvement can be achieved within just a decade.
In my General Discussion (Chapter 5), I outline the wider implications of my research with specific reference to contemporary freshwater management in New Zealand. I also present a stream management ‘thought experiment’ and a new conceptual model that provide theoretical frameworks for considering how spatiotemporal scale might influence the outcomes of stream rehabilitation. Overall, my PhD research highlights the importance of considering a broad range of attributes when assessing the ecological effects of habitat degradation or rehabilitation on instream habitat and fish. In particular, physical instream habitat attributes such as structural fish cover can limit fish populations in agricultural streams, even if they have relatively degraded water quality. Successful stream fish population management in New Zealand will require incorporating physical/structural habitat attributes into the relevant management frameworks.||en_NZ