RESUMO
BACKGROUND: In field experiments, assessment of herbicide selectivity and efficacy rarely takes advantage of dose-response regressions. The objective is to demonstrate that logarithmic sprayers, which automatically make a logarithmic dilution of a herbicide rate, can extract biologically relevant parameters describing the efficacy of herbicides in crops, and compare localities and time of assessment. RESULTS: In a conventional and an organic field, canola, white mustard, and no crop plots were sprayed with diflufenican and beflubutamid. A mixed effect log-logistic dose-response regression, with autoregressive correlation structure, estimated ED50 and ED90 for visual and Excess Green Index symptoms at various days after treatment (DAT). For visual assessment, ED50 differed within no crop between locations for beflubutamid at 12 DAT and 26 DAT. For diflufenican, the ED50 was different within crops at the two fields at 12 DAT, but not at 26 DAT. The Excess Green Indices at ED50 were not different among herbicides, locations, and corps; ED90 differed for white mustard and canola for beflubutamid but not for diflufenican. CONCLUSION: Suitable nonlinear regression models are now available for fitting dose-response data from a logarithmic sprayer in field experiments. The derived parameters (e.g. ED50 ) can compare selectivity and efficacy at numerous cropping systems. © 2018 Society of Chemical Industry.