Figure 1.

Relationship between weed species richness and crop yield loss on the Broadbalk winter wheat experiment (begun in 1843). Weed species richness is assessed on herbicide‐free plots annually, and weed diversity varies in response to contrasting fertiliser treatments. Winter wheat yield loss from weed competition can be calculated as a percentage of the equivalent plots with the same fertiliser treatments but where weeds are controlled with herbicides. Data are presented for 19 years collected between 1991 and 2014 (the plots were fallowed in some years during this period) and sorted by weed species richness. For each level of weed species richness, the average yield loss is presented with error bars indicated the standard error of the mean (r ² = 0.59, P < 0.001). Plots with no nitrogen but with added phosphorus and potassium are excluded from the analysis as the abundance of leguminous weeds leads to facilitation and greater yield in the weedy plots compared to the weed‐free plots.