Pastoral agriculture is the dominant land use in New Zealand and intensification has led to increased nitrogen (N) losses to the wider environment. An indicator that could identify soils which are vulnerable to N loss would be useful for the development of management practices and regulations aimed at reducing unwanted N losses. The natural abundance of ¹⁵N relative to ¹⁴N (δ¹⁵N) in soils is one potential indicator, because most N cycle processes associated with N losses (e.g. nitrification, denitrification, and volatilisation) discriminate against ¹⁵N.

In this study we analysed archived surface soils from five long-term (17–50 year) superphosphate fertiliser or irrigation trials, a medium term (4 year) rate of nitrogen (N) trial, and we also sampled from three chronosequences where pine forest had been converted to pasture.

The two longest trials were at Winchmore on the Canterbury plains where different rates of irrigation and superphosphate had been applied for ~50 years. In both trials δ¹⁵N increased with time, and δ¹⁵N increased significantly more in treatments receiving higher rates of superphosphate or irrigation (increase of ~1‰ over 50 years). Higher rates of superphosphate and irrigation promoted greater pasture growth and therefore higher grazing intensities. Calculation of a simple N budget (based on estimates of N fixation from clover abundance and measured N immobilisation in soil) indicated that total N losses were much higher in the more intensive treatments. The other three long-term superphosphate trials were on North Island hill country, and unlike the flatland trials at Winchmore, no differences between treatments were observed. Slope and aspect had a significant influence on soil δ¹⁵N. Higher variability due to complex terrain and the shorter duration of the hill country trials may explain the lack of observed differences between treatments. In the N trial, δ¹⁵N increased significantly with time in treatments receiving >100 kg N ha^–1 y^–1 and the increase was greater at higher N application rates. Measured N leaching losses also increased with higher N inputs and there was a significant positive correlation between nitrate leaching and the change in soil δ¹⁵N. In the forest to pasture chronosequences there was no clear change in soil δ¹⁵N between forests and recently converted pastures (2–5 years old), but δ¹⁵N in long-term pastures (>40 years since pasture was established) was significantly higher than forest sites (~2‰ higher), likely reflecting cumulative fractionating N losses during pasture development.

These results have shown that soil δ¹⁵N can be related to past N losses and therefore could be a useful indicator of the vulnerability of soils to ongoing N losses.