John A. Nagy, Deborah L. Johnson, Nicholas C. Larter, Mitch W. Campbell, Andrew E. Derocher, Allicia Kelly, Mathieu Dumond, Danny Allaire, and Bruno Croft. 2011. Subpopulation structure of caribou (Rangifer tarandus L.) in arctic and sub-Arctic Canada. Ecological Applications 21:2334–2348.

Appendix E. Comparison of assignments by fuzzy (fuzzy c-means) and hierarchical (Ward’s method) clustering of individual migratory and tundra-wintering barren-ground, Dolphin and Union island, and boreal caribou to subpopulations in the Northwest Territories, Nunavut, and northern Alberta, Canada.

TABLE E1. Comparison of fuzzy (fuzzy c-means) and hierarchical (Ward’s method) clustering of migratory barren-ground caribou that used calving grounds A (Cape Bathurst), B (Bluenose-West), C (Bluenose-East), D (Bathurst), and E-1 and F (Beverly) in the Northwest Territories and Nunavut, Canada. The subpopulation class for each animal from fuzzy clustering was represented by the group with the largest membership coefficient (m = 2.0). Entries represent the number of animals; categories with zero entries are blank. Results are based on clustering of median monthly interval x, y coordinates or 24 variables.

TABLE E2. Comparison of fuzzy (fuzzy c-means) and hierarchical (Ward’s method) clustering of migratory barren-ground caribou that used calving grounds C (Bluenose-East), D (Bathurst), and E-1 and F (Beverly) and Dolphin and Union island caribou that calved on Victoria Island (VI) in Nunavut, Canada. The subpopulation class for each animal from fuzzy clustering was represented by the group with the largest membership coefficient (m = 2.0). Entries represent the number of animals; categories with zero entries are blank. Results are based on clustering of median monthly interval x, y coordinates or 24 variables.

TABLE E3. Comparison of fuzzy (fuzzy c-means) and hierarchical (Ward’s method) clustering of migratory barren-ground caribou that used calving grounds E-1 and F (Beverly) and G (Qamanirjuaq) and tundra-wintering caribou that used calving ground E including E-1, H, and I (Queen Maude Gulf, Lorillard, and Wager Bay) in Nunavut, Canada. The subpopulation class for each animal from fuzzy clustering was represented by the group with the largest membership coefficient (m = 2.0). Entries represent the number of animals; categories with zero entries are blank. Results are based on clustering of median 14-day interval x, y coordinates or 52 variables.

TABLE E4. Comparison of fuzzy (fuzzy c-means) and hierarchical (Ward’s method) clustering of migratory barren-ground caribou that used calving grounds E-1 and F (Beverly) and tundra-wintering caribou that used calving ground E including E-1 (Queen Maude Gulf) in Nunavut, Canada. The subpopulation class for each animal from fuzzy clustering was represented by the group with the largest membership coefficient (m = 2.0). Entries represent the number of animals; categories with zero entries are blank. Results are based on clustering of median 14-day interval x, y coordinates or 52 variables.

TABLE E5. Comparison of fuzzy (fuzzy c-means) and hierarchical (Ward’s method) clustering of tundra-wintering barren-ground caribou that used calving grounds E including E-1, H, I (Queen Maude Gulf, Lorillard, and Wager Bay) in Nunavut, Canada. The subpopulation class for each animal from fuzzy clustering was represented by the group with the largest membership coefficient (m = 2.0). Entries represent the number of animals; categories with zero entries are blank. Results are based on clustering of median 14-day interval x, y coordinates or 52 variables.

TABLE E6. Comparison of fuzzy (fuzzy c-means) and hierarchical (Ward’s method) clustering of boreal caribou in the Northwest Territories and northern Alberta, Canada. The subpopulation class for each animal from fuzzy clustering was represented by the group with the largest membership coefficient (m = 2.0). Entries represent the number of animals; categories with zero entries are blank. Results are based on clustering of median monthly interval x, y coordinates or 24 variables.