Weeds are highly diverse organisms capable of quickly adapting to a variety of environments. This is exemplified by evolved herbicide resistance in weeds due to over-reliance on herbicides with similar mode of action in modern agriculture. Computer-based population models are useful tools in representing and predicting weed population dynamics associated with herbicide resistance. Good knowledge of the agroecology and quality data are of vital importance to build such predictive models. However, empirical experiments that are designed specifically to provide data to model development rarely exist, and trade-offs between using existing and sometimes suboptimal data vs. generating new data are commonplace. Here we present the development of a population model of a key weed species, Amaranthus palmeri, in Argentina, for use in the evaluation of sustainability of weed control programs. Field studies characterising the weed emergence pattern and laboratory experiments revealing the genetic basis of resistance were conducted to inform the design and parameterisation of the corresponding population model.
Results/Conclusions
Variation in emergence time was impactful on the simulated weed density. An updated genetic assumption on glyphosate resistance, representing a weak target-site mutation, instead of gene amplification, decelerated the predicted evolution of resistance to both glyphosate and its mixture partner, fomesafen. The case study demonstrated the value of better data for better model predictions. Recommendations on choice of model type and validation methods will be made. Finally, we proposed a checklist for the development of robust herbicide resistance models, namely weed biology and ecology, genetics and fitness cost associated with the resistance trait, effectiveness of chemical and non-chemical weed control practices, standing genetic variation in weed populations and environmental variability.