|Giles, Kristopher - OKLAHOMA STATE UNIVERSITY|
|Royer, Tom - OKLAHOMA STATE UNIVERSITY|
Submitted to: Journal of Economic Entomology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: May 30, 2000
Publication Date: October 1, 2000
Interpretive Summary: Greenbug (Schizaphis graminum) infestations were sampled from hard red winter wheat fields in Oklahoma 115 times between 1997 and 1999. Statistical models for developing sequential sampling plans for the greenbug differed for fall and spring infestations. Sequential sampling stop lines were calculated for economic thresholds (ET) of 3 or 6 greenbugs sper tiller for fall infestations, and 6 or 9 greenbugs per tiller for spring infestations. It was found that statistical models for presence/absence sampling explained over 95% of the variance in the number of greenbugs per tiller in relation to the number of tillers with one or more greenbugs on them. Sequential sampling plans for fall and spring infestations were developed based on presence/absence models. The models will be useful to wheat growers and crop consultants because they will make it easier and less time consuming for them to determine whether greenbug infestations are high enough in a field that it is cost effective to spray the field with insecticide. The models are a significant improvement over the currently recommended method based on counting the number of greenbugs per foot of crop row, because they are more accurate and less time consuming.
Technical Abstract: The number of Schizaphis graminum was estimated on 115 occasions between 1997 and 1999 from hard red winter wheat fields in Oklahoma. A total of 32 and 83 fields were sampled during the fall and spring, respectively. Parameters of linear regressions relating the proportion of infested tillers (PT) and the mean number of greenbugs per tiller (m) differed significantly between fall and spring infestations. The PT to m linear model provided a good fit to data for fall and spring infestations at tally thresholds of 0, 1, 2, and 3. A regression model for fall with a tally threshold of 2 was the most appropriate for classifying S. graminum populations in the sense that it explained a greater amount of the variation in the PT to m relationship (97%) than models for other tally thresholds. A spring model with a tally threshold of 1 was the most appropriate for classifying populations during spring growth of winter wheat. Sequential sampling stop lines based on sequential probability rati tests were calculated for action thresholds (AT) of 3 or 6 greenbugs per tiller for fall infestations, and 6 or 9 greenbugs per tiller for spring infestations. Operating characteristic and average sample number functions were also generated. With the current parameters (P0 and P1 equal PAT+/- 16.7% of PAT, and alpha and beta set at 0.1), the average sample number required for decision making on greenbug populations varied 207 (fall, T = 2, AT = 3) to 69 (spring, T = 1, AT = 9).