A new approach to the use of meteorological and suction trap data in predicting aphid problems1

Abstract

The network of suction traps operated by the Rothamsted Insect Survey has provided records of aphid activity since 1965. Previous attempts to use this information with meteorological data for predicting aphid problems have concentrated mainly on simple linear regression techniques. In this paper the value of a more complex multiple regression approach is investigated using additional data from an increased range of sites, species and years. Data from 1966 to 1988 are used to produce models to predict the date of the first record of each of 49 aphid species at each of six sites. The models used were particularly successful in predicting the date of the first record at a range of sites in 1989 of four species: Myzus persicae, Macrosiphum euphorbiae, Sitobion avenae and Phorodon humuli.