Jon Tollefson and Tim Nowatzki
Department of Entomology, lowa State University, Ames, lowa, USA 50011-3140

In order to prescribe corn rootworm controls, the manager must be able to predict damaging larval populations. Because the subterranean stages are difficult to sample, the most common practice used has been to determine beetle densities and to use these estimates to forecast larval populations. Monitoring beetle densities throughout the oviposition period requires numerous, repeated visits to fields, greatly increasing the costs of making management decisions. The sampling techniques that have been used to monitor beetle numbers will be reviewed and a model being developed to predict critical sampling periods will be presented.

Data for the model were collected over 4 years (1992, 1993, 1997, and 1998) in 41 continuous cornfields across lowa. Emergence in each field was monitored at least weekly for approximately 8-10 weeks, beginning in late June. In each field, 13 single-plant emergence cages were placed in a "U-shaped", stratified-random arrangement. Emergence cages were monitored regularly to determine the first date a beetle of either species or sex (biofix) was captured in each field. The biofix was defined as the date midway between cage placement and when the first beetle was captured in a cage. Beetles were collected at least weekly throughout the season after the biofix was determined. The sex and species of each beetle was recorded, as well as the total number of beetles that emerged on each date.

Beetle emergence over time for each field was expressed as a cumulative percentage starting at the biofix. Emergence above 95% was excluded from the data to linearize the distributions. Linear regression was used to explain the relationship between days after biofix and percentage emergence, by year. An F-test was used to compare the slopes and intercepts between years. The slopes, which measure the rate of emergence, were not significantly different (F = 1.52, df = 3,257; P > 0.05). However, the intercepts between years were significantly different (F= 48.2; df = 3,257-, P < 0.05). When data from all four years were combined, linear regression explained 68% of the variability between days after the biofix and percent emergence.

To validate the model, emergence data were collected from 21 continuous comfields across lowa in 1999 using the same methods described for the development of the model. Linear regression was used to explain the relationship between days after biofix and percentage emergence across all fields in 1999. The slope and intercept were compared to the model's slope and intercept. The slopes for the prediction model and the combined data from all fields in 1999 were significantly different (t = 4.21; df--405; P<0.05). However, the emergence model gave an acceptable prediction for 21 days after the biofix. Temperature data will be used to attempt to improve the model's forecast of late-season beetle numbers.

Data on sex ratios and female reproductive development was collected at each location. These data will be used to determine the critical times when sampling should occur to make management decisions. The model can then be used to forecast these dates and sampling will be conducted only on those dates. The forecasts will reduce the number of scouting trips to a field and, therefore, sampling costs.