|Ascough Ii, James|
Submitted to: International Weed Science Congress
Publication Type: Abstract Only
Publication Acceptance Date: 2/13/2008
Publication Date: 6/27/2008
Citation: Wiles, L., Fathelrahman, E.M., Wilkerson, G.G., Ascough II, J.C. 2008. Using Stochastic Effciency Analysis To Factor Distribution Of Weed Escapes Into Weed Management Decisions. International Weed Science Congress. Interpretive Summary:
Technical Abstract: Weeds in patches may be more easily managed than the same number of weeds spread throughout the field. We explored choosing weed management strategies based on both net return and the distribution of weed escapes within a field. Expected net returns with several different postemergence herbicides often are similar for a field; however, the distribution of escapes may vary. A grower might be willing to “pay” (forgo net return) for a distribution of weeds that is perceived to be easier to manage. Our goals were to examine variability of distributions of weed escapes among postemergence herbicides and investigate techniques to present information about those distributions so a grower may choose based on both net return and a preference for weed escapes in patches. Our approach is borrowed from methods used to evaluate risk in crop production. Risk is evaluated by ranking alternative management strategies based on how much income each produces relative to the variability in income. We explored the ranking of herbicide strategies based on net return and variability of weed counts for the population of escapes. Weeds typically grow in patches. Consequently, with good weed control, the variability of counts of escapes within a field provides information about the spatial distribution. We simulated weed counts of escapes using weed management decision models for corn, peanuts and soybean and weed counts collected prior to postemergence application in actual fields. We compared rankings of herbicide strategies with efficiency analysis based on net return and simple measures of the distribution of weed counts or competitive load. Our measures included variance, Green’s index, kurtosis and the ‘k’ parameter of the negative binomial distribution. We also adapted methods for rankings based on more detailed information about the distributions. These methods included maximizing utility and stochastic dominance with respect to a function. The strategies compared minimized weed escapes for a field or generated the top ten values of net return for a field. Several of the simple measures, except variance, produced similar rankings. The methods based on utility and stochastic dominance allowed us to more precisely describe preference for patchiness of escapes. A stop light chart of distributions of competitive load, expressed as percent yield loss, was a simple, visual method to present the tradeoff between net returns and the distributions of escapes.