Bayesian Networks for the Assessment of the Effect of Urbanization on Stream Macroinvertebrates

It is generally acknowledged that macroinvertebrates are good indicators of water quality in streams, as a number of taxa are sensitive to pollution and integrate their response to pollution over time. This fact has led some states (e.g., Maine) to incorporate macroinvertebrates in their water quality standards. To quantify the relationship between macroinvertebrates and urban land use, this study used data from the US Geological Survey "Effects of Urbanization on Stream Ecosystems" (EUSE) program. Bayesian networks were developed to characterize the relationship between urban land use metrics and selected macroinvertebrate taxa. It is shown that impervious surfaces and percent urban land have a strong effect on sensitive species of macroinvertebrates, demonstrating the adverse effect of urbanization and thus the value of these macroinvertebrate taxa as indicators of pollution. The resultant models are probabilistic and graphical, making them easily interpretable techniques for urban planning to achieve downstream water quality goals and sustain aquatic ecosystems.