AN EXAMINATION OF SPATIAL INPUT PARAMETERS IN ORDER TO IMPROVE CORN-EARWORM (LEPIDOPTERA, NOCTUIDAE) DAMAGE PREDICTIONS FOR A PHEROMONE TRAP CATCH REGRESSION-MODEL

A regression model relating cumulative pheromone trap catch and date o f corn silking to subsequent damage in sweet corn by Helicoverpa tea ( Boddie) was improved in one year by including additional spatial input parameters. Spatial inputs included information from timings and loca tions of corn plantings around the trap as well as from locations of w ind blocking features. Wind blocking features were: tree rows, wooded areas, large buildings close to the trap, and abrupt hillsides. Pherom one traps were monitored at 28 and 30 sites in 1990 and 1991. Corn dev elopment through the year and damage levels at harvest (percent infest ation) were also recorded for these locations. Maps of all corn planti ngs and wind blocking features within 2.5 kilometers of the trap were created, digitized, and entered into a Geographical Information System for each site. A stepwise regression analysis considering 18 spatial and two non-spatial variables resulted in a highly significant (P < 0. 001) regression model with four variables (trap catch, silking date, t he number of hectares of corn within 2.0 kilometers of the trap, and t he average distance to wind blocking features on the north side of the trap) explaining 82% of the variability for the 1990 data. In 1991, o f the 20 input parameters considered only one was significant, date of first silk. An extremely low corn earworm population was considered t o have caused this inability to find correlations for the other parame ters in 1991.