DETERMINING CHARACTERISTIC STRESSOR SCALES FOR ECOSYSTEM MONITORING AND ASSESSMENT

While most environmental scientists appreciate the importance of appro priate spatial and temporal scales for environmental assessment and mo nitoring, the issue of how to determine these scales has not (in our v iew) been addressed in sufficient depth. Here we define the characteri stic spatial scale for a stressor as the nearest distance from the sou rce at which stressor values cannot be statistically distinguished fro m those further away, while the characteristic time scale is the lengt h of time required to detect a statistically significant decline in av erage stressor values within the defined characteristic spatial scale in response to reduced stressor emission at the source. We show that f or a simple model of exponential distance-decay of average stressor va lues away from the source, the characteristic distance d for a stress or increases with (1) decreasing spatial signal strength, (2) decreasi ng spatial noise, (3) increasing sample size and (4) decreasing sampli ng resolution. By contrast, the characteristic time-scale tau increas es with (1) decreasing sample size and (2) decreasing sampling resolut ion. Moreover, unlike d, the relationship between tau*, signal streng th and spatial noise is not straightforward. Consequently, neither is the relationship between d and tau*; stressors with small characteris tic spatial scales need not also have short characteristic time scales . Thus, the accepted wisdom that spatial and temporal scales are posit ively correlated need not be true in practice primarily because the ab ility to defect statistically significant spatial or temporal change m ay vary considerably among different levels of biological organization . (C) 1997 Academic Press Limited.