A Simple Dynamic Model of Soil Test Phosphorus Responses to Phosphorus Balances

Soil test P (STP) concentration indicates whether annual P applications can be expected to give yield increases and can also indicate an elevated risk of P mobilization and potential for P transfer to surface waters and groundwater from a particular field. Changes in STP with time thus project agronomic benefits and environmental risks of different P use strategies. To predict STP changes with time, we constructed a simple dynamic model for which the input variables are P balance and initial STP. The model parameters (soil type‐specific constants) were fitted using data originating from 44 P fertilizer experiments with different P rates. Model performance was evaluated using independent data sets that either had reasonably accurate input values (n= 103) or were obtained from farmers through interviews (n= 638). The simulations were in agreement with measured STP changes for both evaluation data sets when fittings were performed separately for four main soil types (clays, silts, coarse mineral soils, and organic soils). Statistical analysis confirmed that the model captured the trends in STP (NH4OAc test) with acceptable accuracy and precision, with r2of 0.83 and 0.66 for the data with more accurate input and for farmer interview data, respectively; the corresponding model efficiency statistics were 0.88 and 0.66. The model is not restricted to use with one soil test, as fittings for several different types of soil tests can be generated. In this study, we fitted the model for Olsen P data retrieved from the literature. Agronomic use of the model includes evaluation of P use strategies, e.g., when a certain STP level is targeted or when long‐term economy of P use is calculated. In an environmental context, the model can be used to predict STP changes with time under variable P balance regimes, which is essential for realistic assessment of changes in the potential for dissolved P losses. A simple model to project changes in STP was created.Model inputs are present STP and P annual balance.The model can be calibrated for several soil P tests.Tests with separate data showed the model's adequacy in predicting STP changes with time.