Forecasting of periodic fluctuations in annual abundance of the bean aphid: the role of density dependence and weather

Abstract

Abstract Linear regression models have been constructed to describe variation in abundance of Aphis fabae, Myzus persicae and Acyrthosiphon pisum populations in Rostock, Germany. Egg densities, which have been used successfully in the past to predict A. fabae outbreaks, were not found to be a significant predictor of summer population size. Linear regression models that described existing data well were on the whole poor at predicting future patterns. Apparent delayed density dependence has been shown to be an artefact of the weather patterns that occurred between 1980 and 1992. It is hypothesized that much of the variation in A. fabae densities in the summer is due to variable rainfall and thus food quality for the aphid at budburst of spindle, the primary host. The importance of food quality in spring appears to be similar for M. persicae. It is not year clear what determines changes in the year‐to‐year abundance of A. pisum. The results are discussed with reference to the entire philosophical framework of multiple‐regression forecasting.