Simulating phosphorus responses in annual crops using APSIM: model evaluation on contrasting soil types

Crop simulation models have been used successfully to evaluate many systems and the impact of change on these systems, e.g. for climatic risk and the use of alternative management options, including the use of nitrogen fertilisers. However, for low input systems in tropical and subtropical regions where organic inputs rather than fertilisers are the predominant nutrient management option and other nutrients besides nitrogen (particular phosphorus) constrain crop growth, these models are not up to the task. This paper describes progress towards developing a capability to simulate response to phosphorus (P) within the APSIM (Agricultural Production Systems Simulator) framework. It reports the development of the P routines based on maize crops grown in semi-arid eastern Kenya, and validation in contrasting soils in western Kenya and South-western Colombia to demonstrate the robustness of the routines. The creation of this capability required: (1) a new module (APSIM SoilP) that simulates the dynamics of P in soil and is able to account for effectiveness of alternative fertiliser management (i.e. water-soluble versus rock phosphate sources, placement effects); (2) a link to the modules simulating the dynamics of carbon and nitrogen in soil organic matter, crop residues, etc., in order that the P present in such materials can be accounted for; and (3) modification to crop modules to represent the P uptake process, estimation of the P stress in the crop, and consequent restrictions to the plant growth processes of photosynthesis, leaf expansion, phenology and grain filling. Modelling results show that the P routines in APSIM can be specified to produce output that matches multi-season rotations of different crops, on a contrasting soil type to previous evaluations, with very few changes to the parameterization files. Model performance in predicting the growth of maize and bean crops grown in rotation on an Andisol with different sources and rates of P was good (75-87% of variance could be explained). This is the first published example of extending APSIM P routines to another crop (beans) from maize.