Short-term forecasting of wheat yield loss caused by the grain aphid (Sitobion avenae) in summer

A damage model was used to take into account the effect on winter wheat yield of the size, duration and timing of Sitobion avenae infestation in summer, and the influence of insecticide applications. It was validated as far as possible using available data, and appeared to be sufficiently accurate to be useful. The model was applied to a total of 32 winter wheat fields over the period 1975 to 1983 in eastern England and the Netherlands, to obtain estimates of yield losses associated with not applying insecticides at the end of ear emergence (G.S. 59), mid-anthesis (G.S. 65) or the end of anthesis (G.S. 69). These estimates of future yield losses were related in multiple regressions to the number of S. avenae per tiller at the time the insecticide was applied, and the rate of population increase over the previous few days. Basing the decision about whether to apply insecticide on forecasts from the multiple regression equations appeared to be slightly more profitable than prophylactic spraying when the chemical cost alone was considered, but substantially more profitable when the application cost and wheeling loss were also taken into account. The strategy minimising losses involved forecasts at all three crop growth stages. It was validated using a further 21 fields studied in 1984 and 1985, and gave better results than the EPIPRE system, which over-estimated the number of fields requiring spraying, and the economic threshold of George & Gair (1979), which under-estimated the number requiring spraying. However, the accuracy of the damage model needs to be checked against measured yield losses in order to establish the success of the multiple regression strategy conclusively.