Development of a socio-ecological environmental justice model for watershed-based management

• The utility of integrating stream health and socio-economic metrics was studied.
• Limited correlation was observed between the social and ecological measures.
• Bivariate maps illustrated the spatial dependence of social and ecological variables.
• The census tracts with no streams corresponded to higher non-white populations.

SummaryThe dynamics and relationships between society and nature are complex and difficult to predict. Anthropogenic activities affect the ecological integrity of our natural resources, specifically our streams. Further, it is well-established that the costs of these activities are born unequally by different human communities. This study considered the utility of integrating stream health metrics, based on stream health indicators, with socio-economic measures of communities, to better characterize these effects. This study used a spatial multi-factor model and bivariate mapping to produce a novel assessment for watershed management, identification of vulnerable areas, and allocation of resources. The study area is the Saginaw River watershed located in Michigan. In-stream hydrological and water quality data were used to predict fish and macroinvertebrate measures of stream health. These measures include the Index of Biological Integrity (IBI), Hilsenhoff Biotic Index (HBI), Family IBI, and total number of Ephemeroptera, Plecoptera, and Trichoptera (EPT) taxa. Stream health indicators were then compared to spatially coincident socio-economic data, obtained from the United States Census Bureau (2010), including race, income, education, housing, and population size. Statistical analysis including spatial regression and cluster analysis were used to examine the correlation between vulnerable human populations and environmental conditions. Overall, limited correlation was observed between the socio-economic data and ecological measures of stream health, with the highest being a negative correlation of 0.18 between HBI and the social parameter household size. Clustering was observed in the datasets with urban areas representing a second order clustering effect over the watershed. Regions with the worst stream health and most vulnerable social populations were most commonly located nearby or down-stream to highly populated areas and agricultural lands.