Asseng, S., CSIRO Plant Industry, Private Bag 5, PO, Wembley, WA 6913, Australia, Centre for Legumes In Mediterranean Agriculture, University of Western Australia, Nedlands, WA 6907, Australia
Bar-Tal, A., Centre for Legumes In Mediterranean Agriculture, University of Western Australia, Nedlands, WA 6907, Australia, Institute of Soils, Water and Environmental Sciences, Volcani Center, Agricultural Research Organisation, P.O. Box 6, Bet-Dagan 50250, Israel
Bowden, J.W., Centre for Legumes In Mediterranean Agriculture, University of Western Australia, Nedlands, WA 6907, Australia, Agriculture Western Australia, 3 Baron-Hay Court, South Perth, WA 6151, Australia
Keating, B.A., CSIRO Sustainable Ecosystems/APSRU, Long Pocket Laboratories, 120 Meiers Road, Indooroopilly, QLD 4068, Australia
Van Herwaarden, A., CSIRO Plant Industry, GPO Box 1600, Canberra, ACT, Australia
Palta, J.A., CSIRO Plant Industry, Private Bag 5, PO, Wembley, WA 6913, Australia
Huth, N.I., CSIRO Sustainable Ecosystems/APSRU, Long Pocket Laboratories, 120 Meiers Road, Indooroopilly, QLD 4068, Australia
Probert, M.E., CSIRO Sustainable Ecosystems/APSRU, Long Pocket Laboratories, 120 Meiers Road, Indooroopilly, QLD 4068, Australia

APSIM-Nwheat is a wheat crop system simulation model within the APSIM framework which consists of modules that incorporate aspects of soil water, nitrogen, residues, wheat (Triticum aestivum L.) crop development and growth, including grain protein content. APSIM-Nwheat has been validated for soil water, soil N, crop phenology, biomass production and yield. However, previous analyses have indicated that model performance was poor in terms of grain protein simulations particularly under very high or very low N input conditions together with terminal drought. The original routine for grain protein content simulated grain protein as a function of independent dry matter and N accumulation into the grain. To constrain grain protein content simulations under very high and very low N input conditions, without effecting other simulations, a simple but physiologically sound link was incorporated between dry matter and N translocation to the grain. An upper boundary of daily protein transfer to the grain was set to 4% N (22.8% grain protein), and a lower boundary was set at 1.23% N (7% grain protein). In addition, grain N was initialised with up to 3% N (17.1% grain protein) at the beginning of the grain filling phase, depending on the availability of tissue N. The new modified grain protein routine was tested across field data sets from a wide range of growing conditions and showed an improved performance with a RMSD, consistent across environments and soil types, at or below 0.35% N (2% grain protein). Independent tests, including data from different climates and a sensitivity analysis confirmed the improved simulation of grain protein under a wide range of conditions. The improved model was found to be reliable and robust enough to be used for specific simulation experiments to study grain protein interactions with management, soil types and environments in different climatic regions. © 2002 Elsevier Science B.V. All rights reserved.