Yield effects of herbicides on competition between crop and weed communities

Abstract

A literature review showed that the relationship between crop and weed top dry matter yields at harvest time is often linear. It is suggested that by including monocultures of both crop and weeds and a series of partial weeding treatments in any herbicide screening experiment, the true effect of a herbicide can be determined with the separation of any effect due to interspecific competition. To do this, confidence limits are constructed for a linear regression fitted to the crop by weed yield coordinates for the non-herbicide treatments. Points produced by yield coordinates from herbicide-treated plots which lie outside these confidence limits indicate either depression or stimulation of both yield components. From the relationships between these points and the regression line, weed competition and herbicide phytotoxicity esects can be estimated separately. Evidence supporting this theory was obtained in a field experiment with a crop of Desmodium intortum and a natural weed community.