Dos and don'ts of spatially explicit ecological risk assessments

Abstract

Location affects exposure and response to stressors at scales ranging from small sites to large regions. Tools such as geographic information systems (GIS) make it feasible to conduct spatially explicit ecological risk assessments (ERAs). However, no tools provide a panacea, and complex models based on sparse data can be inappropriate and misleading. Operations such as an interpolation within a GIS are models, and so they contain assumptions and uncertainties. Errors can propagate easily as numerous operations are performed, and the resulting uncertainties should be made explicit. Analysts should assure that space actually matters by using explanatory data prior to investigating spatial correlation. Exposure and risk estimates from experimental plots or monitoring stations may require scaling factors if they are to be used for larger watersheds or regional analyses. Maps are useful because they present complex spatial information in a manner that is easily interpreted. However, they must be prepared and interpreted with caution because they can suggest that there is a great deal more information than actually exists. In addition to presenting maps of predicted risks, maps of the spatial patterns in the uncertainties in these risk estimates should be included in risk assessments.