An evaluation of site selection methods in the Assiniboine Delta region of Manitoba

Abstract

Resources that are devoted to the conservation of biodiversity are limited. Conservation planners are often faced with the challenge of preserving the maximum amount of biodiversity within a give minimum amount of land. Based on these circumstances, the extent to which protected areas are capable of representing elements of biodiversity will largely depend on the manner in which protected areas are selected and where they are located. This thesis focuses on the application of systematic selection methods in order to identify potential networks of protected areas that comprehensively represent enduring landscape features. Two conservation scenarios are explored: one takes into consideration the current level of conservation within the Assiniboine Delta region of Manitoba, and the other treats the study area as if no protected areas have been previously established within the region. A greedy heuristic algorithm and a simulated annealing algorithm were utilized to select candidate areas for conservation that would fulfill specific conservation goals. The two algorithms were free to select sites from crown lands or a combination of crown lands and privately owned natural lands. Variations in the magnitude of spatial clustering and conservation targets were applied to the two algorithms as they selected candidate areas for the two conservation scenarios. The simulated annealing algorithm outperformed the greedy heuristic algorithm with regards to generating a selection set that was spatially compact, and was much closer at meeting conservation goals in the most efficient manner. The results generated from this analysis contained some important implications for conservation planning in the Assiniboine Delta region. Most importantly, efficient ecosystem representation could not be achieved for this region. The reasons for this were that, first, in order to achieve full ecosystem representation, a much larger land base would be required than was available. Second, the presence of existing protected areas reduced the efficiency with which a network of protected areas could be established. The reason for this was that the manner in which the protected areas have been previously established resulted in the over representation of conservation features. Third, the manner in which the planning units were spatially distributed across the landscape did not adequateiy sample all of the enduring landscape features. If full ecosystem representation based on enduring landscape features is going to be accomplished, not only would conservation planners have to look outside crown lands to include privately owned lands, but they would also have to examine the options of restoration and rehabilitation.