Evaluation of Continuous Host-Parasitoid Models

Abstract

In this paper the performance of continuous host-parasitoid models were investigated. The parameter values for several well-known models: Lotka-volterra, Holling Tanner Type 2, Holling Tanner Type 3, Leslie, Bazykin, Beddington-DeAngelis, Yodzis and Rosenzwing-Macarthur models were estimated. The models were tested on 40 consecutive sets of time series data collected at 14 days interval for pest and parasitoid population obtained from a highland cabbage growing area in Eastern Kenya. Model parameters were estimated from the minimization of the loss functions between the theoretical and experimental time series datasets following the Nelder-Mead multidimensional method. Initial values of population size and parameters were randomly chosen. Durbin-Watson statistic was applied for comparison of model outputs and experimental population trajectories. Among the eight different hostparasitoid models, Holling Tanner model Type 3 presented relatively better approximations compared to the other models.