The use of the WISDEM model in the simulation of weed shifts in glyphosatetolerant cropping systems

Abstract

WISDEM (Weed Interference and Simple Demographic Model) is a decision-aid model developed initially by Wiles et al. (1996) and recently augmented in response to some major theoretical work by Canner, Wiles, and McMaster (2002). This model has been significantly upgraded by the present author and is presented as an ecologically-based approach to the simulation of weed shifts as well as a realistic decision-aid management tool for glyphosate-tolerant cropping systems, specifically for selecting optimal weed management products and optimal application timings. This work begins with a discussion of previous modeling efforts, competition studies, and the inherent weed science concepts that provide the foundation for such models. Such a discussion provides insight into the theoretical and historical development of WISDEM. On this theoretical level and from this historical legacy the model represents each crop rotation as a controlled sequence of scheduled plant disturbance events whose timing and intensity can be varied. Crop emergence is represented as a singular pulse event, while emergence for each weed species is depicted via an empirically derived Weibull curve. Population dynamics for each plant species are calculated via a set of proportional yield equations that estimate seed production per unit area for each plant species. These equations assess the competitive influence of each cohort of each plant species according to the species' relative competitive index, degree of competitive asymmetry, and emergence timing. In this way it has been postulated that much in the way of explicit multi-species plant competition can be represented through the use of empirical biological parameters that are relatively easy to obtain. It has also been postulated that additional simplification may be possible by integrating parameters associated with yield loss with parameters associated with weed seed production. In addition, in the context of the present work two algorithmic upgrades have been implemented, one which imposes a reduction in per-plant fecundity for late-emerging weed species and the other defining an effective weed density for crop yield reduction by averaging the weed densities for each species over the crop's competitive period. These two advancements, respectively referred to as the advent of "reproductive density" and "crop competitive density" are presented along with a re-parameterization of the curves that define the emergence timings of each weed species. Sensitivity analyses are performed along with validation studies that suggest a significant improvement in WISDEM's ability to represent realistic dynamics, and to, thus, serve as an effective ecological model and decision-aid tool. Lastly, two modeling experiments are performed on the field study from which validation comparisons have taken place. In each case various model runs are performed in an effort to find an option that is economically optimal from a long-term planning perspective. One experiment seeks to find a suitable bum-down herbicide replacement for a pre-plant tillage event, one which would effectively render the crop production scenario now as a "no-till" practice. The other experiment seeks to find optimal timings for the applications of Roundup Ultra™. In both instances the WISDEM model provides realistic recommendations that confirm its usability as a decision-aid tool. This work concludes with a description of the status of the ongoing project and suggestions for further efforts.