Estimating characteristic length scales of dynamic biological systems : removing the need for long time series

The choice of appropriate spatial s.cales for observing, conserving and managing systems are pervading questions in applied ecological research. Determining the characteristic length scales (CLSs) of ecological systems is likely to provide valuable information towards answering these questions. The CLS is the scale at which the ratio of deterministic signal to noise in a system's dynamics is maximised i.e. the scale that captures the meaningful signal in the system's dynamics. Recent methods for identifying the CLS are attractive because they accommodate the complex non-linear behaviours that occur in ecological systems. However, these methods require long temporal data series and so are unrealistic for most natural systems. This thesis develops and examines two alternatives to using long time series data to estimate CLSs. The first is a short time series approach that requires data from only three or four consecutive landscapes. The second approach uses spatial data from a single point in time. The performance of these methods is compared with current techniques, using data from spatial competition systems. The model systems employed in this study are more complex than models examined by previous authors and provide a better indication of how CLSs might perform with real data sets. Results indicate that the short time series approach to estimating CLSs is more consistent in its interpretation than the long time series method, and has great potential for application to natural systems. A comparison of CLS results with more conventional analyses for identifying scales of spatial pattern (variograms and nested ANOV A) suggests that a combination of the two approaches may be most successful for defining characteristic scales in applied ecological contexts.