The starting and ending points of a corridor can greatly impact the location of the modeled corridor. A terminus is a part of a wildland block that forms one end of a modeled corridor. You can define it as a point (or pixel), a linear edge (e.g., the wildland boundary), or a patch (population patch or breeding patch). Often there is more than one potential terminus in each wildland block.
Some linkage designs use each wildland block in its entirety as a terminus for each single-species analysis. However, in our experience this procedure sometimes produces a corridor that connects to a part of the wildland block far from any potential habitat for the focal species. You can avoid this result by allowing only patches of potential breeding habitat within each wildland block serve as a potential terminus. Better yet, if you have a map or known breeding patches (Arizona has such data for pronghorn, for example), you should use each known breeding patch as a terminus.
Wildland blocks vary in how much known or potential breeding habitat occurs in the block. A flexible strategy is to select start/end points in the following order of priority:
If you have no pixels of breeding habitat in either wildland block, you probably should not build a corridor model for that species. You can modify the linkage design to serve this species later.
An important component of cost-distance is Euclidean distance. If your two wildland blocks nearly touch at one or more locations, least-cost modeling will tend to identify the narrowest gap as the best corridor, even if the modeled corridor is low in habitat value and does not effectively connect the most important habitat patches between wildland blocks. To reduce this problem, we often give the GIS model “room to run” by defining the terminuses well inside of the wildland blocks. The easiest way to do this is to draw roughly parallel lines a few kilometers apart, with most matrix land centered between the two lines, and then selecting population patches behind these lines. Even if this modification does not change the modeled corridor for the focal species, it will demonstrate that the modeled corridor is not merely an artifact of a point where two wildland blocks nearly touch.