Nitrogen mineralisation in sugarcane soils in Queensland, Australia: I. evaluation of soil tests for predicting nitrogen mineralisation

Abstract

We explored soil properties as indices of mineralisable nitrogen (N) in sugarcane soils and whether we could increase the accuracy of predicting N mineralisation during laboratory incubations. Utilising historical data in combination with samples collected during 2016, we: (i) measured mineralised N over the course of short-term (14 days) and long-term (301 days) laboratory incubations; (ii) compared models representing mineralisation; then (iii) related model parameters to measured soil properties. We found measures representing the labile organic N pool (Hydrolysable NaOH organic N; amino sugar Illinois soil N test) best related to short-term mineralised N (R2 of 0.50–0.57, P < 0.001), while measures of CO2 production (3, 7, 10 and 14 days) best related to longer-term mineralised N (R2 of 0.75–0.84, P < 0.001). Indices were brought together to model the active and slow pools of a two-pool mineralisation model in the statistical framework of a mixed-effects model. Of the models that relied on measurement of one soil property, cumulative CO2 production (7 days) performed the best when considering all soil types; in a cross-validation test, this model gave an external R2 of 0.77 for prediction of the 301-day mineralised N. Since the mixed-effects model accounts for the various sources of uncertainty, we suggest this approach as a framework for prediction of in-field available N, with further measurement of long-term mineralised N in other soils to strengthen predictive certainty of these soil indices.