Wheat protein prediction from climatic factors in southern NSW

Abstract

Association between regional wheat protein levels and numerous rainfall, temperature, and sowing time variables in southern New South Wales over the period 1948-49 to 1968-69 was assessed from simple linear correlation coefficients. Of all the simple variables, average maximum temperatures in October and November were most highly correlated (positively) with protein. On the basis of the preliminary assessment, a small number of combined rainfall and temperature variables were built up for screening through a series of multiple regression analyses. The best of the multiple regression models took in a further variable, time (in years), to take account of the factors not covered by the other variables. This model accounted for more than 85 per cent of regional protein variation over 21 years and included three highly significant variables (P<0.01) : time (in years), midseason (May to July) rainfall, and late (August to November) rainfall and one significant variable (P<0.05) : midseason rainfall squared. Protein level increased by approximately 0.1 per cent per annum over the study period once significant climatic effects were removed.