6. Conclusions
This study has shown that distinct patterns in water quality and biological variables can be discriminated using REC's hierarchy of controlling factors. By delineating these patterns the REC provides a method for interpolating monitoring results for state of environment analysis and reporting and provides a spatial framework for managing water quality at a range of scales.
The study has highlighted the very heterogeneous nature of New Zealand's rivers and the difficulty this creates for effective environmental monitoring. Each of the four factors that define the top levels of the REC was implicated in differences in water quality and biological variables. This makes it difficult to attribute differences in state, or changes in time, to land use or management. To be certain that differences in state or trends are attributable to human activities, data from sites that are similar with respect to Climate, Source of Flow and Geology but which differ with respect to land use or management, must be compared. Monitoring networks must be either very targeted or must have a large number of sites that cover a wide range of types. Greater coverage of different types of rivers involves increased monitoring effort. Thought could be given to achieving greater spatial coverage by monitoring more integrative variables (e.g. biological variables) with decreased sampling frequency.
Consideration needs to be given to the effect of climatic variability on water quality, as it may confound attempts to determine the causes of observed water quality patterns. One approach for controlling climate effects is the use of paired sites; impact and control. The net difference between data at the sites could be used in trend analyses. However, this solution does not provide replicate sites that would allow confidence that the trends can be interpolated across similar REC sites. Monitoring networks that could allow pattern detection would involve increased replication, particularly of baseline sites. Both of these solutions could potentially require significant changes the structure of some Regional Council monitoring programs.
Further work should include more data from regional council monitoring networks to increase replication. This may enable statistical tests of significance of differences in state variables among REC classes. Furthermore, greater replication may provide increased certainty that spatial patterns in trends are associated with differences in land use.
