Modeling effects of herbicide drift on the competitive interactions between weeds.

Herbicides may drift onto road verges or natural areas adjacent to arable fields and affect nontarget plants. The effect of low doses of mecoprop-P on the competitive interactions and plant community dynamics was investigated in a model system using Capsella bursa-pastoris and Geranium dissectum as test plants. Dose-response experiments on single species showed that compared to G. dissectum, C. bursa-pastoris was more affected by mecoprop-P. Consequently, we expected that G. dissectum would outcompete C. bursa-pastoris when mecoprop-P was applied at a low dose in the competition experiment. Indeed, mecoprop-P had a significant effect on the interspecific competitive ability of both C. bursa-pastoris and G. dissectum. Our previous expectation, however, was not met: The interspecific competitive ability of both species increased significantly with the dose of the herbicide, and it was predicted that C. bursa-pastoris and G. dissectum are more likely to coexist in natural habitats with low concentrations of the herbicide compared to natural habitats with relatively high concentrations. The results from the dose-response experiments on the single species and the more laborious competition experiment approach, which is assumed to mimic the dynamics of plant communities more closely, show considerable discrepancies even though the experiments were performed at the same time and in the same greenhouse. This finding generally reduces the credibility of using single-species tests in ecological risk assessment of herbicide use.