Accurate prediction of soil nitrogen (N) mineralization in agricultural soils is of major concern because uncertainty in making fertilizer N recommendations can lead to economic losses and environmental pollution. This study examined the suitability of three temperature functions ( Q 10, Arrhenius, Logistic) as predictors of the temperature dependence of soil N mineralization rate, k, in soil using previously published data sets. Each function fits k/k 0, where k 0 is the reference mineralization rate, against soil temperature T, where k/ k 0 = 1 at the reference temperature, T 0. No single value of soil temperature was common to all data sets, and consequently a series of values of T 0 from 5 °C to 35 °C were tested. The influence of the temperature zone, land use and soil textural class of soils in the data set on the temperature response function was also tested. Despite the different mathematical forms of the functions evaluated, the fitted curves for each function were very similar and choice of temperature response function had a limited effect on prediction of soil N mineralization rate. An additional model, the Logistic Fixed M model, is proposed which fits the data sets as well as the previous models, but also takes into account the existence of optimal and maximal temperatures in a reasonable temperature range for biological organisms. In contrast, choice of T 0 had a much more pronounced impact on the k/ k 0 values, and thus on the predicted N mineralization rate, than choice of temperature model. A greater response of N mineralization rate (i.e. k/ k 0) to changes in temperature was observed in soils originating from colder climatic zones (mean annual temperature < 2 °C) compared with warmer climate zones (mean annual temperature > 6 °C). There was also a greater temperature response of soil N mineralization rate for agricultural compared with forested soils. Among agricultural soils, sand-loam soils had a greater temperature response compared with clay soils. Overall, selection of temperature response model did not appear to be critical to prediction of soil N mineralization rate, and consequently a form of the model which best represents the biological system is therefore preferable, whereas more attention should be given to the choice of the appropriate T 0 for field prediction of N mineralization.
Read full abstract