Buffered or bundled, least-cost paths are not least-cost corridors: Computational experiments on path-based and wide-path-based models for conservation corridor design and effective distance estimation

摘要

Conservation of wildlife corridors has been an option to counter the problem of loss and fragmentation of wildlife habitat. A mathematical model often adopted in this option uses an algorithm for finding least-cost paths over a cost surface represented in raster format, i.e., as a grid of weighted cells. This model is easy to implement and computationally efficient, but not without drawbacks. The most notable one is that a path is a string of cells and may not be an appropriate representation of an actual wildlife corridor, which is often significantly wider than a single cell. Through computational experiments with artificial landscapes, this paper compares the existing model and an alternative model that 1) represents a corridor as a swath of cells, called a "wide path," swept by a preset form, 2) defines the cost of a wide path as the sum of the cost-weighted areas of all its cells, and 3) finds a least-cost wide path with a selected width between two terminuses over a raster cost surface. Results suggest that the wide-path-based model is an improvement over the path-based model in that both take the same raster cost surface as input but the former finds a better corridor in terms of form and cost without sacrificing the ease of implementation and computational efficiency that the latter has. This implies that the wide-path-based model will immediately benefit ecologists and planners who use geographic information systems for the design of wildlife corridors or the study of landscape connectivity.