Predictive models of weed population dynamics

• Published in: Weed research Vol. 49; no. 3; pp. 225 - 232
• Main Authors: FRECKLETON, R P, STEPHENS, P A
• Format: Journal Article
• Language: English
• Published: Oxford, UK Blackwell Publishing Ltd 01.06.2009; Wiley-Blackwell
• Subjects: agriculture; Biological and medical sciences; Chemical control; Fundamental and applied biological sciences. Psychology; herbicides; Parasitic plants. Weeds; Phytopathology. Animal pests. Plant and forest protection; predictive modelling; stochastic model; weed control; Weeds; Cultural practice; Weed control; Stochastic model; predictive modelling; Weed; Pesticides; Forecast model; Modeling; Herbicide; Weed science; Predictive value; Agriculture; Population dynamics; herbicides
• ISSN: 0043-1737; 1365-3180
• DOI: 10.1111/j.1365-3180.2009.00694.x

Summary Many of the challenges faced by weed ecologists can be met only by the capability to predict the responses of weed populations to changes in their environment or management. In spite of this, a review of papers published in Weed Research suggests that weed ecologists are remarkably reluctant to produce detailed, quantitative predictions. This may result from uncertainty in the accuracy of predictions and indeed, a variety of reasons have been put forward to suggest that the potential utility of weed models may be limited in this regard. In this study, we review the applications to which weed models have been put. Focusing on predictive population modelling, we highlight several limitations that can lead to failures of this approach and we discuss the likely prospects for weed population modelling. We make three points regarding the future of weed modelling. First, owing to prohibitive data requirements, the development of highly mechanistic models that attempt to make detailed predictions of weed population numbers is unlikely to be very successful. Second, data collection for developing weed models needs to be rethought. Weed models are most commonly compromised by a lack of spatial and temporal replication, preventing modellers from measuring parameter variability and error effectively and limiting assessments of model uncertainty. Finally, the utility of models needs to be better appreciated; models are key tools in making long range predictions of how management will affect weed populations, but, we estimate, they are used in only a small fraction of studies. Without the further development of models for weed population dynamics, our ability to predict long‐term dynamics will be restricted.