Cultural Landscapes: Protection of Traditional Hunting and Trapping Areas in the Northwest Territories
GEOB 370 Final Project
University of British Columbia
Figure 2. Flowchart of Analytical Methods I.
Figure 6. Flowchart of Analytical Methods II.
Figure 4. Analytic Hierarchy Process: Hunting Layer
Figure 5. Analytic Hierarchy Process; Trapping Layer.
Analysis for the identification of areas where trails are likely to exist was divided into a two-part method. The first portion of analysis consisted of producing a map of likely hunting areas and a map of likely trapping areas through the conduction of two separate multicritera analysis. The second portion of analysis consisted of combining the weighted hunting and trapping maps to produce an overall map of likely trail use, and then extracting the most likely areas where trails would be found.High areas of likely trail use that fell within designated Sahtu land were identified and displayed differently from high areas of likely trail use that fell outside of Sahtu land boundaries. The final map displays a layout of the distribution of areas with high trail likelihood falling within and outside of designated Sahtu land, and can be viewed on the Protected Area Map Results page.
A flowchart of the steps taken in the first portion of analysis can be viewed in figure 1.
Methods
Figure 3. Portion of "Yamoria:The One Who Travels", depicting important sites within Dene oral tradition of the cultural hero Yamoria. Prince of Wales Northern Heritage Centre (2013).
The first step in analysis was to create a reference DEM as a base layer. DEMs of the region surrounding Tulit'a were collected for NTS sheets 096B, 096C, 096D, 096E, 096F, 096G, 095M, 095N, and 095O at the 1:250,000 scale. The DEMs were then mosaicked to create a reference DEM with the geocoordinate system NAD 1983.
Communities and highways were included in the maps to provide perspective for the areas later identified. The highway can also serve as a proxy for the waterways of the Deh Cho (Mackenzie River) and Sahtu De (Great Bear River). In the final Protected Area Map, zones of fire history were included as a proxy for areas of boreal forest regrowth, as a potential indicator of animal habitat.
Several input variables were selected for the creation of the hunting and trapping MCE maps. The criteria for each map favoured areas of low slope, and were of close proximity to Tulit'a, the habitat of the targeted animal species, and sites of importance in oral tradition.
To create the slope layer, a slope analysis was conducted using the reference DEM to identify areas of high and low slope. The slope layer was then normalized on a scale of 0 to 1 in order to be used in the multicriteria analysis.
To identify sites of importance in oral tradition, a pdf map of sites featuring the cultural hero Yamoria was georeferenced, and selected sites were digitized as a point feature layer. The portion of the pdf map used in analysis is shown in figure 2. The map was georeferenced using a first order polynomial regression with a resulting RMS Error of 0.0635157. Oral tradition points were then manually digitized and created as a separate layer.
To determine the proximity values for each of the feature variables, euclidean distance was calculated from each of the selected criteria (e.g.Tulit'a, Habitat, Oral Tradition Sites). Once the distance from each feature had been calculated, the euclidean distance layers were normalized by using a linear method of fuzzy relationship. An example of this method used for Tulit'a can be seen on the Euclidean Distance Map and Fuzzy Membership Map pages.
The normalized slope and feature proximity layers were then assigned weights using an analytical hierarchy process (AHP) for both hunting and trapping weighted surface maps. In each case, low slope and the primary targeted animal species were of greatest and equal importance. The reasoning for this ranking prioritizes areas that are easy to access by foot and that are nearby identified areas of species habitat. Proximity to the town of Tulit'a was of next greatest importance, as smaller distances to travel to the town as a destination would be more favourable. Secondary species were ranked equally in the trapping AHP, as they are available for harvest during a limited part of the seasonal trapping cycle. Oral tradition sites were given the lowest rank, as they would have had an important role in determining routes from an earlier period but would have less of an influence in a more modern context.
The results of the weights assigned to each criteria for the hunting and trapping AHP's are shown in figures 4. and 5., respectively.
The resulting weights designated by each AHP were assigned to each variable in a weighted multicriteria evaluation. This method produced two weighted surface maps displaying the likelihood of an area to contain hunting or trapping trails. The results can be viewed on the Hunting Area Map and Trapping Area Map pages, respectively.
The second portion of analysis consisted of methods to investigate the overall likelihood of trail existence from a combination of the hunting and trapping area data. Three surfaces were produced, displaying the overall distribution of likely trail use, the most likely areas of trail existence, and likely areas of trail existence that fell within designated areas of Sahtu land. A flowchart of the methods used in the second portion of analysis is shown in figure 6.
To produce a map of overall trail use likelihood, the weighted hunting and trapping surface maps had to be combined. Before doing this, however, each of the maps had to be normalized on a 0-10 scale. Once this step was completed, the normalized layers were combined using map algebra to produce a likelihood map of trail surfaces. The trail surface map was then reclassified to create 10 classes with grouped values between 2-20 of suitability in order to identify the areas of highest likelihood for trail use.
To identify the priority areas most likely to exhibit trail use, a mask of the higher ranked classes (2-9) above the mean weighted value was created and used to extract areas of highest likelihood of trail use. Figure 7. displays the mask used to extract areas of highest trail likelihood from the composite trail surface map.
Figure 7. Mask of highest ranked areas of likely trail use overlaid on the composite trail surface map.
A second mask was needed to designate areas of high trail likelihood that already fell within Sahtu land from those areas of high likelihood that fall outside of Sahtu boundaries. The Sahtu land layer was used as a mask to extract the targeted areas from the high likelihood trail use layer. The results were then compiled in a final map, which can be found on the Protected Area Map page. Finally, the total area of the high likelihood trail use was calculated in square kilometers using cell unit size, as well as the total area in square kilometers of high likelihood trail use falling within Sahtu boundaries. A comparison between the two totals are presented on the Results page.
A sensitivity test was conducted for each of the weighted maps produced. The methodology for this step first involved conducting a multicriteria analysis for both the hunting and trapping scenarios in which the criteria were assigned equal weights on a scale of 0-100. In the hunting scenario, Slope, Moose, Tulit'a, and OTSites were all given weights of 25; in the trapping scenario, Slope, Beaver, Tulit'a, Lynx, Marten, and OTSites were all given weights of 16.66. The results were two equally weighted maps, one for Hunting Areas, and another for Trapping Areas.
The next step was to normalize the data on a common scale. The weighted Hunting and Trapping layers were reclassified to contain classes, and then assigned new values for each class. The four highest classes (i.e. 6-10) were assigned new values of 1 and the other classes assigned values of 0. A similar procedure was carried out for the equally weighted layers, only the four highest classes were assigned values of 2.
Finally, the weighted layers and equal weight layers were combined in a Weighted Sum procedure in which each layer was assigned a weight of 1. Two composite maps resulted, one for the Hunting scenario and another for the trapping scenario. The results of the sensitivity case indicate that the methods used to establish the weighted maps were fairly robust. In each case, any land deemed as highly suitable by the weighted analysis was also selected by the equal weights analysis. There was little difference between the weighted and equal weights analysis of trapping areas, as the assesments are almost identical. The robusticity of the weighted hunting analysis was less accurate. Although every area identified by the weighted analysis was also identified by the equal weights analysis, there were some areas that the weighted MCE failed to recognize. The weighted hunting MCE becomes less accurate in areas that are farther away from Tulit'a, despite proximity to moose habitat and low slope (as can be seen in the final Hunting Area Map). The visual results of the sensitivity procedure can be found on the Sensitivity Map :Hunting Areas and Sensitivity Map: Trapping Areas pages.