The concept of bollworm, Heliothis zea (Boddie), management communities (BMCs) is the result of a three year study which began in 1972 in 50 square mile areas at three locations in Arkansas. Based on the findings of the three year study, and general knowledge of the seasonal history of the bollworm, a new series of studies began in 1976. These studies were initiated in order to determine if control measures for the bollworm could be managed on an area-wide basis rather than a field-by-field "as-needed" basis.

The data presented deal with the preliminary analysis of bollworm data collected during 1979 and 1980 in a BMC located in the Portland community. Four 14 rowfoot whole plant samples for bollworm were conducted each week in 34 - 60+ fields. The data collected each week were pooled and analyzed as a two stage sampling problem. All analyses were conducted on the actual and transformed, X+0.5, data. One objective of data analysis was to determine whether bollworm sample unit size could be reduced from 14 row-feet to 7 row feet during any part or all of the growing season. A second objective of the analysis was to determine the optimum allocation of resources for both sample units.

The comparison of the two sample units involved the ability to detect changes in population size, agreement between the two when converted to absolute or relative estimates, and the mean relative variation (RV).

Both sample units detected population changes, and the difference between the estimates by the two sample units were minimal in both years. RVs of the 7 row-foot sample unit were less than those of the 14 rowfoot unit in most cases. However, in 1979 the 7 rowfoot sample unit did not detect large larvae (>=1/2 inch) until the sixth week of sampling. Large larvae were detected by the 14 row-foot sample units during each week of sampling in 1979.

Optimum allocation of resources was determined using actual and transformed data for both sample units. The optimum sample size most often estimated for each sample unit was one. Differences in optimum sample size estimates for actual and transformed data were minimal, and the estimated optimum sample size exceeded the number of samples taken in each field only once.

The analysis of variance indicated less variation among fields than among samples within fields. Because the management decisions are based on the community mean, the within-field variation was considered acceptable. However, a reduction in sample size is not warranted because of the within-field variation.