Modelling evolution and management of glyphosate resistance in Amaranthus palmeri

A population-based model was developed to simulate the evolution of glyphosate resistance in populations of Amaranthus palmeri. Model parameters were derived from published and unpublished sources, and the model was implemented using previously established principles and methods. Sensitivity analyses indicated that the model was sensitive to variations in population size, mutation rate and seed bank dynamics. A distribution was assigned to these parameters and Monte Carlo type simulations were performed. Simulation results are therefore derived from a range of possible input parameters, enabling the risk of resistance evolution to be assessed when parameter values were unknown, uncertain or variable. In the 'worst-case' of five annual glyphosate applications in continuous glyphosate resistant cotton, evolution of glyphosate resistance was predicted in 39% of populations after 5 years and in c. 60% of populations after 10 years. These results are consistent with observations of the timescale for evolution of glyphosate resistance in A. palmeri in the field. The main drivers for glyphosate resistance evolution were selection pressure and population size, the greatest risks being associated with the largest A. palmeri populations. Risks of resistance were reduced when one of the five glyphosate applications was replaced by another mode of action with identical efficacy. However, not all glyphosate applications exerted the same selection pressure. Application of a soil residual herbicide at the time of crop sowing can provide control of A. palmeri well into the growing season and significantly reduced the rate and risk of glyphosate resistance evolution.