AN INVESTIGATION INTO THE USE OF MULTIPLE-SITE CHARACTERISTICS TO PREDICT THE PRESENCE AND INFESTATION LEVEL OF WIREWORMS (AGRIOTES SPP., COLEOPTERA, ELATERIDAE) IN INDIVIDUAL GRASS FIELDS

Sixty-two grass fields were sampled in England and Wales over a three year period to assess the level of wireworm population present. Data o n site-specific variables including soil physical characteristics, gra ss duration, grass genera diversity and other abiotic factors such as field aspect, altitude, and meteorological parameters were collected a t each site. Only grass duration and soil bulk density showed any asso ciation with wireworm infestation status when considered as single var iates. Data from a sub-set of 41 fields were used to develop a series of multi-variate discriminant rules to predict wireworm presence/absen ce in individual fields. These were validated using data from the rema ining 21 fields. In general, the rules tended to over-estimate the num ber of wireworm-infested fields by misclassifying uninfested fields as infested. Multivariate models to predict wireworm population levels i n infested fields were also developed using multiple and generalised l inear regression. The predictive accuracy of these was poor. Neither t he population prediction models nor the presence/absence rules account ed fully for the large inter-field variation in wireworm infestation s tatus.