Their first step was to divide the U.S. into twenty ecologically
homogeneous regions using methods based on firm scientific principles. They
divided the country into eight million patches, and for each patch, they collected
nine pieces of information about its ecology and climate. They then used a
supercomputer to cluster the patches into similar regions and picked a
representative site within each region. They then reanalyzed the data to make
sure that the twenty sites were as different from one another as possible and
represented the full spread of ecological conditions. This method worked well,
but it would be better to consider 100 different ecological properties rather than
just nine. New techniques will be needed to make that computationally practical.
Figure 10: The National Ecological Observatory Network had to choose twenty ecologically
representative sites across the United States to monitor to see what effects changes in land use,
climate, and invasive species might be having. They started by dividing the country into eight
million patches, which they classified by ecological type. This generated the map above. Credit:
William Hargrove, U.S. Forest Service.
Measuring progress toward sustainability might require us to understand
how to measure the health of specific “indicator species” that, like the canary in
the coal mine, indicate the overall health of an ecosystem. For example, lichens
respond to changes in forest structure (air quality, climate) and disappearance of
lichens may indicate environmental stress (high levels of sulfur dioxide, nitrogen
oxides, etc.). Algal species in aquatic systems may indicate organic pollution and
nutrient loading (e.g., nitrogen, phosphorus). Mussels are sensitive to siltation
and low dissolved oxygen in water. Efforts have been made to build
mathematical models that show how one can find “clusters” of unhealthy plants
such as lichens. These “dynamical spatio-temporal models” of the distribution of