Lack of predictability in model ecosystems based on coupled logistic equations

One of the major tasks for conservation and fishery biologists is the prediction of future population dynamics from relatively short time series of past data. Inspired by a recent paper of Alan Hastings and Kevin Higgins (Science, 263 (1994) 1133–1136) on long-term transients in realistic model ecosystems for spatially distributed populations, we explored the problem of predictability in analogous systems of coupled logistic equations. For such systems, with parameters in the ‘chaotic range’ as studied by A. Hastings and K. Higgins, we show that such predictions are at best extremely difficult. Consequences for conservation and fishery are explored.