The effect of weed distribution on predictions of yield loss.

A statistical model is presented which allows the exploration of the effects of weed patchiness on crop yield. Input data are a non-linear regression for 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 (...)