The connectivity result files from Circuitscape represent the "adjusted cumulative current density" flowing across the landscape for each of several species, at a 90-meter resolution across the South Atlantic Landscape Conservation Cooperative region. Rasters are classified using quantiles with 20 categories (each 5% of region) to integer scores from 1-20. 1 = lowest 5% of the landscape, 20= top 5% of landscape Expert opinion was used to define a resistance surface for each of the target animals, with higher resistance representing map units expected to be more difficult and dangerous for species to move through. A set of nodes for each species, with node points indicating center locations for potential source populations, are also defined. Note actual species population data to define the nodes is not used, as that data was often unavailable, and the focus is on the potential spread of the species across the SALCC region and not limited to models to known populations. Therefore, node locations were determined using an innovative approach to search for local minima in the resistance surfaces, as such areas likely represent favorable habitat for each species in terms of the number of dispersing animals that might be produced. The Circuitscape program was then used to calculate expected flow of animals between each pair of nodes within a species-specific threshold of each other. Circuitscape works similarly to ordinary least cost path analysis, but instead of returning a single least cost path or corridor, it calculates the expected flow of the target species across all of the different pathways from one node to the other, treating the nodes as electrodes and the landscape as a circuitboard matrix with varying levels of resistance. Pathways expected to receive lots of dispersing animals are scored with high current density values, whereas cities, major roads, and out-of-the-way routes between the nodes tend to get low current density values. Once the pairwise runs were done, they were summed together to create a cumulative current map for each species. The current sums were calculated using an innovative weighted average approach, with pairwise current layers that included a larger node counting more than layers that only included smaller nodes. Circuitscape models were run on a study region that was defined by a 100km buffer of the SALCC region. This helped models avoid the edge effects that can be prevalent in connectivity studies of this variety. After results were obtained for the entire study region, they were clipped to the SALCC region. Results beyond the SALCC are available, but due to aforementioned edge effects, their use is not highly recommended. Known issues: The Circuitscape current density outputs are in relative terms and cannot be directly converted to numbers of animals crossing a given pixel of habitat per unit time. Additional empirical studies can hopefully provide calibration. Until then, users should be aware that Circuitscape may show connections between nodes that are too difficult for any single animal of the target species to actually complete, even after taking into account the maximum node-node distance thresholds employed. Also, broad areas of suitable habitat may have low current densities compared to narrow "chokepoints" where expected flow becomes more concentrated. Thus, the output maps should be interpreted cautiously. Important areas for connectivity could be broad "sheet flow" linkages, or tight restricted corridors, depending on the priorities of decision makers. Note the highest current density values tend to be found at the nodes themselves, which is an artifact of the way current flow is calculated. Post-processing the layers to remove this effect, before using the model results to set conservation priorities, is recommend as otherwise the node points will receive undue attention as connectivity routes (they represent more of the core habitats that need to be connected). Areas with little to no current flow can be interpreted as less important for connectivity, but only for the node pairs used. Circuitscape results are highly sensitive to the specific nodes used in the analysis. Users should examine node locations (available on request) before discounting a particular area of interest that appears to have low or zero current density. Consulting the Connectivity Analysis Toolkit results from the same project, which did not incorporate specific nodes and are therefore much less sensitive to the node selection process, may also be useful. Not all species included in the overall index are found across the entire SALCC. For example, the eastern diamondback rattlesnake is absent from most of the piedmont and foothills. As a result, coastal areas where all species were potentially present receive a higher potential index score for connectivity than the piedmont areas outside of the range of the diamondback rattlesnake and pine snake. If a smaller area is being considered (e.g. how to connect two specific core areas of habitat) then it would be more appropriate perhaps to use the single node pair output from Circuitscape corresponding to that area, instead of the regional cumulative map. The single pair results are available upon request. A box turtle model was attempted, but it proved to be computationally infeasible given the sheer number of nodes that would be involved across the study region. Complete details about the methodology of the project will be contained in the final report for the South Atlantic LCC, and in journal publications to follow. Users with urgent questions should send them to Ron Sutherland at ron@wildlandsnetwork.org.
