Effect of initial condition sensitivity and chaotic transients on predicting future outbreaks of gypsy moths

• Published in: Ecological modelling 2001, Vol.136 (1), p.49-66
• Main Author: Wilder, J.W.
• Format: Article
• Language: English
• Subjects: Animal and plant ecology; Animal, plant and microbial ecology; Animals; Biological and medical sciences; Chaos; chaos theory; Demecology; Europe; Fractals; Fundamental and applied biological sciences. Psychology; General aspects. Techniques; Gypsy moth; Lymantria dispar; Methods and techniques (sampling, tagging, trapping, modelling...); North America; Population dynamics; Protozoa. Invertebrata
• ISSN: 0304-3800; 1872-7026
• DOI: 10.1016/S0304-3800(00)00385-9

A previously validated, simple model of the population dynamics of the gypsy moth, Lymantria dispar (Lepidoptera: Lymantriidae), is used to show that initial condition sensitivity in the cases of (a) chaotic dynamics, and (b) fractal basin boundaries associated with non-chaotic attractors both severely limit the ability to predict future behaviour even for short time scales. Also demonstrated is the fact that short-term transients can cause even small discrepancies in initial population densities to lead to erroneous conclusions about future behaviour over periods of the order of a decade. This work addresses the quality of data needed to make predictions based on population data from a single year. It is demonstrated that, even when it is assumed that the model perfectly reflects the population dynamics and that the parameters are known exactly, extremely small errors in the specification of initial conditions are required to obtain reliable predictions. The model also shows that the observed dynamical difference between populations in North America (chaotic dynamics) and Europe (periodic dynamics) could be explained by the presence of chaotic transients.