Assessing variability in fecundity of Amaranthus powellii using a simulation model

Summary Estimates of weed fecundity and its variability are critical for the development of population dynamic models and for evaluating the long‐term consequences of weed management practices. The purpose of this research was to estimate Amaranthus powellii fecundity across years, seasons and competitive environments using a mechanistic model. Existing models were modified to account for weed responses to shade, and to dynamically simulate seed production among subthreshold densities of A. powellii . The model was parameterized and tested using five sets of field data in which A. powellii was grown either alone or with transplanted broccoli. The model overestimated A. powellii height, but predicted both dry weight and fecundity well. Mean simulated fecundity for A. powellii ranged from 0 to 268 000 seeds per plant depending on year, crop maturity date, relative time of emergence and location. Year‐to‐year variability in simulated fecundity was large with coefficients of variation ranging from 77 to 128 under low and high competitive environments respectively. Our results suggest that estimates of weed fecundity based on 1 or 2 years of empirical data may result in significant errors in population dynamic models, and that the use of economic optimum thresholds based on weed density alone entails considerable risks.