Water quality in low-elevation streams and rivers of New Zealand: recent state and trends in contrasting land-cover classes

River water quality in New Zealand is at great risk of impairment in low elevation catchments because of pervasive land-use changes, yet there has been no nationwide assessment of the state of these rivers. Data from the surface-water monitoring programmes of 15 regional councils and unitary authorities, and the National River Water Quality Network were used to assess the recent state (1998–2002) and trends (1996–2002) in water quality in low-elevation rivers across New Zealand. Assessments were made at the national level, and within four land-cover classes (native forest, plantation forest, pastoral, and urban). Finer-scaled assessments were made by subdividing the large number of pastoral sites into six climate classes, and seven stream orders. At the national level, median concentrations of the faecal indicator bacterium Escherichia coli, and dissolved inorganic nitrogen and dissolved reactive phosphorus exceeded guidelines recommended for the protection of aquatic ecosystems and human health. Water quality state varied widely within land-cover classes: E. coli and dissolved nitrogen and phosphorus concentrations in the pastoral and urban classes were 2–7 times higher than in the native and plantation forest classes, and median water clarity in the pastoral and urban classes was 40–70% lower than in the native and plantation forest classes. Water quality state in the pastoral class was not statistically different from that of the urban class, and water quality state in the plantation forest class was not statistically different from that of the native forest class. Significant trends in low-elevation rivers were limited to four parameters: flow (trending down in all instances), and temperature, clarity, and conductivity (trending up in all instances). The trends in flow, temperature, and clarity were apparent at the national scale, and within the pastoral class. The magnitudes of these trends were very low, corresponding to changes of £0.5%/ year in parameter medians.