Validation of Weed Competitive Indices for Predicting Peanut Yield Losses in Oklahoma

Abstract

Weed interference experiments have not been extensively conducted in Oklahoma peanut. Research was conducted in three environments to evaluate usefulness of single-weed density experiments with the use of several weeds to measure their relative competitive abilities with a crop. These data can be used to validate current competitive indices (CIs) used by a model to predict peanut yield loss due to weeds. This model is used by the Herbicide Application Decision Support System (HADSS) and Pesticide Economic and Environmental Tradeoffs (PEET), two decision-support systems (DSSs) available for Oklahoma peanut. Six weeds common in Oklahoma peanut were used: crownbeard, eclipta, ivyleaf morningglory, johnsongrass, Palmer amaranth, and prickly sida plus two others, barnyardgrass and common cocklebur, as benchmark species. Each weed was planted into peanut uniformly at eight weeds/10 m of row. Dry weed biomass accounted for 77 to 90% of variation in in-shell peanut yield loss; however, model parameters only allow for weed number. Yield losses from these experiments were compared to those predicted by the model to test original CI accuracy. Treatment means were compared to the prediction model with the use of protected LSD. Several significant differences were noted, and the CIs for those weed species were adjusted accordingly. Adjusting CIs improved actual yield data goodness of fit to model predictions specific to environment in question, but not necessarily in different environments. The CI changed at Ft. Cobb were eclipta from 1.8 to 4.5 and ivyleaf morningglory from 3.4 to 5.0. CI adjustments at Perkins were common cocklebur from 10.0 to 5.8 and johnsongrass from 3.0 to 4.6. Collecting data for several weed species at uniform density in a crop provides a more time-efficient method for obtaining accurate relative weed interference data; this method is useful in validating or establishing CI lists in areas and/or crops with limited data.