Relationships between multiple land-use pressures and individual and combined indicators of stream ecological integrity DOC ReseaRCh anD DevelOpment seRies 326

Joanne Clapcott, Roger Young, Eric Goodwin, John Leathwick, David Kelly
1 References 48 Appendix 1 Boosted regression tree model parameters 54 Appendix 2 Relative contributions of environmental predictor variables 56 Abstract The identification and prioritisation of rivers for their conservation management is a key issue facing the Department of Conservation, and effective tools for doing so are needed. This study investigated the relationships between indicators of river health and measures of human pressure, and evaluated the usefulness of land-use pressure
more » ... ements for predicting the ecological integrity (EI) of New Zealand streams. Fourteen metrics of stream health based on fish, invertebrate, water quality and ecosystem process data were used and tested against three land-use pressure gradients (native vegetation cover, impervious cover and predicted nitrogen loading). We used boosted regression trees (BRT) to quantify the shape of the response of each metric to the pressure gradients. The strongest metrics were oxidised nitrogen concentrations, δ 15 N of primary consumers and the macroinvertebrate community index, while the weakest overall indicators were gross primary productivity, fish richness and invertebrate richness. Monotonic decreases in response to native vegetation clearance and impervious cover were generally found, while distinct non-monotonic responses were observed for some indicators in relation to predicted nitrogen loading. Environmental variability, summarised by Freshwater Environments of New Zealand groups, significantly improved the explanatory power of individual BRT models. However, the direct use of environmental variables provided even better model performance. Output from the BRT analyses was used to predict observed and expected values for each metric for all stream reaches in New Zealand. Ten metrics calculated using the observed and expected data were then combined to produce a multi-metric index (MMI) of EI. Ecological response curves for the MMI suggested distinct thresholds for change in EI in response to land-use pressure-a potentially important implication for sustainable water resource planning and management.