Difficulties in using land use pressure and soil quality indicators to predict water quality

• Land pressures and soil quality can influence local water quality.
• Land use, catchment and climatic factors were strong predictors of 15 years of water quality.
• A snapshot of land value and stocking rate poorly predicted catchment water quality.
• Too few soil quality samples were available to predict catchment water quality well.
• Increasing the temporal frequency of pressures and spatial frequency of soil quality data may improve predictions.

Intensive agriculture can impair river water quality. Soil quality monitoring has been used to measure the effect of land use intensification on water quality at a point and field scales but not at the catchment scale. Other farm scale land use pressures, like stocking rate and the value of land, which relate to land use intensity are now publicly available, nationally. We therefore tested whether point scale soil quality measures, together with newly available farm scale land use pressures (land valuation and stocking rate) and existing catchment and climatic characteristics could help predict the behaviour of water quality data across 192 catchments in New Zealand.

We used a generalised additive model to make predictions of the change in nitrogen fractions (r² = 0.65–0.71), phosphorus fractions (r² = 0.51–0.70), clarity and turbidity (r² = 0.42–0.46), and E. coli (r² = 0.35) over 15 years. The state and trend of water quality was strongly related to a refined farm scale land use classification, and to catchment and climatic characteristics (e.g. slope, elevation, and rainfall). Relationships with point scale soil quality measures and the land use pressures were weak.

The weak relationship with land use pressures may be caused by using a single snapshot in time (2022), which cannot account for lag times in water quality response but leaves room for additional temporal data to improve predictive power. The weak relationship to soil quality measures was probably caused by limited data points (n = 667 sites) that were unrepresentative of land use, and areas of catchment subject to processes like runoff or leaching.

While national soil quality measures might be useful for evaluating environmental risk at the field or farm scale, without a large increase in sampling, they were not relevant at the catchment scale. Additional analyses should be performed to determine how many samples would be needed to detect a change using an environmentally focused soil test that can guide water quality management.