Modelling and predicting durum wheat yield and quality in Mediterranean environments

Durum wheat (Triticum turgidum L. var durum) is mainly produced and consumed in the Mediterranean environments, where yield and grain protein content are usually constrained by environmental variables. A 1.4 million ha of durum wheat are currently grown in Italy with an average production of 4 million tons and an average yield of 3.8 t/ha. 73% of the production, which accounts for 65% of total production, is located in South Italy, while in the Northern regions yield is usually higher due to different pedology and climatic conditions. The cultivation of durum wheat in Italy generates a vast range of allied activities, “upstream” such as seed and technical supplies industries, and “downstream” such as storage centries, primary and secondary transformation industries. Due to the strategic importance of the durum wheat production, a tool able to accurately predict yield and grain protein content before harvesting and an assessment of the agronomic and environmental variables that most influence the quantitative and qualitative parameters in the Mediterranean environments, can be of great value in policy planning. Currently the national and international agricultural statistics services provide regular updates during the growing season of total acreage planted with a specific crop, as well as the expected yield levels. Traditionally, forecasts have been based on a combination of scouting reports as well as statistical techniques based on historical data. Based on the expected yield, the price of grain can vary significantly, with a high impact on both commodity prices and farmers incomes. Growing season forecasts of crop yields are therefore of considerable interest also to commodity market participants and price analysts. Crop simulation models can play a critical role in crop yield and quality forecasting applications: the relatively low cost and speed of assessment makes crop growth simulation models promising for areas where meteorological information is readily available. The overall aim of the present study was to test the predictive capability of Delphi system, based on the AFRCWHEAT2 model, for yield and quality forecast at local and regional scale in long-term analysis and to determine the general principles underlying how Mediterranean environments affects grain protein content (GPC). General findings allowed us to implement a new simple model with high predictive ability in terms of GPC, only based on gridded climate data, supporting strongly the key role played by weather pattern for a crop such as durum wheat under rainfed conditions.