The effect of weed distribution on predictions of yield loss.
Abstract
A statistical model is presented which explores the effects of weed patchiness on crop yield. Input data are a non-linear regression of yield versus weed density, and the parameters of a frequency distribution. Simulations are presented for yield estimation assuming either a random or a negative binomial distribution. The assumption of a random distribution under-estimated yields in the presence of aggregation, and the discrepancy increased with the degree of aggregation. For Bromus sterilis in winter wheat, although estimates of yield may be in error at high densities, the error at densities where practical control decisions would be made are minimal. However, weed distribution may still contribute to decisions relating to weed control priorities.