Which species have similar responses to urbanization across natural and modified sites in Edmonton?
Figure 11. Direct gradient analysis of land and vegetation cover variables with species frequencies for 89 camera trap sites in natural (n=69; green) and modified (n=20; beige) areas in Edmonton. Sites in the river valley are marked as circles and sites outside of the river valley are marked with triangles.
|
Table 3. R-squared and p-values for a direct gradient analysis using non-metric multidimensional scaling of landscape features and species frequencies in the city of Edmonton. Chipmunks, jackrabbits, porcupines, white-tailed deer, and moose were significantly associated with the ordination of land and vegetation cover variables. See appendix for species code interpretation.
|
To determine how the composition of mammal species communities was related to vegetative and urban features, I preformed a direct gradient analysis using non-metric multidimensional scaling (NMDS) and a Bray-Curtis distance measure (Figure 11). The analysis was preformed using the MetaMDS function in R studio. Modified sites were characterized by a higher density of roads, impervious surfaces, houses and grass cover (largely representing the abundance of mowed lawns in urban areas), while natural sites were characterized by a higher density of natural vegetation, shrub, forest, leaf litter, woody debris, canopy cover, and vegetation diversity. Vectors for predictor variables that were short and uninterpretable were removed from the ordination. Jackrabbits, red squirrels, and skunks were most detected at modified sites with a higher degree of urbanization, suggesting that these species are most characteristic of urban exploiters in Edmonton. Of these three species, the frequency of jackrabbits was significantly associated with the ordination of the landcover variables (p=0.001; Table 3). Jackrabbit populations are thought to be increasing in Edmonton's urban landscape, with the city being at the limits of their northern range (Visscher and Wood 2022). The results of my study suggest that the abundance of lawns in the city may be attracting jackrabbits to areas of higher urban density. Red squirrels and skunks were most frequently detected at sites outside of the river valley with a higher density of roads and impervious surfaces, suggesting that these species also exploit areas with a high degree of human disturbance. For red squirrels in particular, these results are supported by other studies which have shown that red squirrel abundance increases with human density in urban areas (Jokimäki et al. 2017).
The remaining species (moose, white-tailed deer, mule deer, coyote, snowshoe hare, red fox, flying squirrel, porcupine, beaver, and chipmunk) were most often detected at natural sites and sites that were in the river valley. Of these species, moose (p=0.004; Table3), white-tailed deer (p=0.006; Table3), and chipmunks (p=0.003; Table3) were significantly associated with the ordination of landcover variables. Ungulate species (moose, white-tailed deer, mule deer) showed similarities in their selection of landscape features. These associations were most pronounced for moose and white-tailed deer, who were most often detected at sites with higher vegetation diversity and shrub density. Moose in particular appeared to occur most frequently at natural sites outside of the river valley, suggesting they may not rely on the river valley heavily as a movement corridor. Additionally, the frequency of coyote, snowshoe hare, red fox, flying squirrel, beaver, and porcupine detections were closely associated with each other, although none of these relationships were significantly associated with the ordination of landcover variables (Table 3). Despite the lack of significant relationships, these species still appear to be associated with sites with a higher percentage of leaf litter, woody debris, and canopy cover. Chipmunks were also more frequently detected at natural sites, but they appear to neither be selecting for sites with a higher density of forest cover and vegetation or sites with a higher density of buildings, roads, and impervious surfaces. The lack of association with either extreme suggests that chipmunks may be acting as urban adapters in Edmonton and are able to exploit both natural and urbanized landscapes. It is important to note the possibility that camera traps at sites with high shrub cover and tall grass failed to adequately detect small bodied organisms due to their lower detectability in dense vegetation. Future studies which aim to assess how small mammals select landscape features using camera traps may also want to incorporate the use of track tubes or additional methods to increase the detectability of small mammals (O’Brien, Kinnaird, and Wibisono 2011).
The remaining species (moose, white-tailed deer, mule deer, coyote, snowshoe hare, red fox, flying squirrel, porcupine, beaver, and chipmunk) were most often detected at natural sites and sites that were in the river valley. Of these species, moose (p=0.004; Table3), white-tailed deer (p=0.006; Table3), and chipmunks (p=0.003; Table3) were significantly associated with the ordination of landcover variables. Ungulate species (moose, white-tailed deer, mule deer) showed similarities in their selection of landscape features. These associations were most pronounced for moose and white-tailed deer, who were most often detected at sites with higher vegetation diversity and shrub density. Moose in particular appeared to occur most frequently at natural sites outside of the river valley, suggesting they may not rely on the river valley heavily as a movement corridor. Additionally, the frequency of coyote, snowshoe hare, red fox, flying squirrel, beaver, and porcupine detections were closely associated with each other, although none of these relationships were significantly associated with the ordination of landcover variables (Table 3). Despite the lack of significant relationships, these species still appear to be associated with sites with a higher percentage of leaf litter, woody debris, and canopy cover. Chipmunks were also more frequently detected at natural sites, but they appear to neither be selecting for sites with a higher density of forest cover and vegetation or sites with a higher density of buildings, roads, and impervious surfaces. The lack of association with either extreme suggests that chipmunks may be acting as urban adapters in Edmonton and are able to exploit both natural and urbanized landscapes. It is important to note the possibility that camera traps at sites with high shrub cover and tall grass failed to adequately detect small bodied organisms due to their lower detectability in dense vegetation. Future studies which aim to assess how small mammals select landscape features using camera traps may also want to incorporate the use of track tubes or additional methods to increase the detectability of small mammals (O’Brien, Kinnaird, and Wibisono 2011).
Urban avoiders and urban adapters can better be distinguished when considering habitat selection in natural sites
Figure 12. Indirect gradient analysis of land and vegetation cover variables with species frequencies for 89 camera trap sites in natural areas in Edmonton. Sites are grouped based on their distance from the city centre (CC; <5km, 5-10km, 10-15km, >15km).
|
Table 4. R-squared and p-values for an indirect gradient analysis using non-metric multidimensional scaling of species frequencies and landscape features in the city of Edmonton. The density of natural forest (forest) natural shrub (d_shrub), natural vegetation (natveg) , modified vegetation (modveg), roads, and impervious surfaces (imperv_surfaces) was significantly associated with the species frequency ordination.
|
In order to determine how species selected sites in relations to land and vegetation cover variables within natural sites, I preformed an indirect gradient analysis using non-metric multidimensional scaling (NMDS) and a Bray-Curtis distance measure with the MetaMDS function in R studio (Figure 12). Sites were grouped by their distance to the city centre (<5km, 5-10km, 10-15km, >15km). The density of natural forest (p=0.001), natural shrub (p=0.019), natural vegetation (p=0.001) , modified vegetation (p=0.001), roads (p=0.001), and impervious surfaces (p=0.001) was significantly associated with the species frequency ordination (Table 4). Vectors for predictor variables that were short and uninterpretable were removed from the ordination. When looking at just natural sites, mule deer and white-tailed deer most often appeared at sites further from the city centre (>15km) which had a higher density of shrubs, natural forest, natural vegetation, and modified vegetation. Moose follow a similar pattern, while also being closely associated with shrubby sites. These results suggest that of the species detected, ungulates are least tolerant of urban landscapes and have habitat selection preferences most consistent with urban avoiders. On the other hand, even within natural areas, jackrabbits appear to be selecting sites with the highest degree of urbanization, suggesting that out of all species detected, jackrabbits may have the strongest ability to exploit urban landscape features like roads, impervious surfaces, and maintained grass (lawns). Beavers, porcupines, coyotes, snowshoe hares, flying squirrels and chipmunks were most often detected at sites at intermediate distances from the city centre, but this relationship was most apparent for coyotes. We do not find support for coyotes acting as urban exploiters in Edmonton. Instead, patterns of coyote habitat selection most closely resembles that of an urban adapter. Coyotes were not associated with the density of forest and vegetation cover or with the density of roads, impervious surfaces, and houses, suggesting that coyotes can persist in either landscape type.
If coyotes actively use both natural and urbanized spaces, the question then becomes, how can we promote coyote use of natural areas and discourage use of highly urbanized areas, to minimize human-coyote conflict? Reducing coyote attractants in urban centres is one possible solution for discouraging coyotes from select habitat in densely populated areas. It may be that coyotes are not actively selecting urban features such as buildings, roads, and houses, but they are being drawn to those locations due by the abundance of anthropogenic food they provide, like compost (Raymond and St. Clair 2023). Reducing anthropogenic food attractants may provide less incentive for coyotes to access highly modified landscapes. It is also possible that urban areas create good habitat for coyote prey, such as rodents, rabbits, and hares. Jackrabbits are a particularly common prey species for coyotes (Visscher and Wood 2022) and it is possible their high abundance in urban landscapes is attracting coyotes into urban spaces.
If coyotes actively use both natural and urbanized spaces, the question then becomes, how can we promote coyote use of natural areas and discourage use of highly urbanized areas, to minimize human-coyote conflict? Reducing coyote attractants in urban centres is one possible solution for discouraging coyotes from select habitat in densely populated areas. It may be that coyotes are not actively selecting urban features such as buildings, roads, and houses, but they are being drawn to those locations due by the abundance of anthropogenic food they provide, like compost (Raymond and St. Clair 2023). Reducing anthropogenic food attractants may provide less incentive for coyotes to access highly modified landscapes. It is also possible that urban areas create good habitat for coyote prey, such as rodents, rabbits, and hares. Jackrabbits are a particularly common prey species for coyotes (Visscher and Wood 2022) and it is possible their high abundance in urban landscapes is attracting coyotes into urban spaces.
Conclusions
By looking at the associations between species frequencies and landscape features, my study was able to describe the relative tolerance of urbanization for 13 mammal species in Edmonton within the urban avoider, adapter, and exploiter framework (Figure 13). Jackrabbits emerged as the species most exploitative of urban landscapes, while red squirrels and skunks also showed preference for modified areas with a higher anthropogenic disturbance. Our results suggest that populations of these species have the potential to increase in the City of Edmonton with increasing urban expansion. Ungulates were most characteristic of urban avoiders, preferring sites further from the city centre and sites with higher vegetation diversity and shrub cover. These results suggest that creating green spaces with a higher density of natural shrub may support ungulate populations along the perimeter of the city. The remaining species showed selection preferences consistent with urban adapters, appearing to choose sites at an intermediate distance from the city centre. Importantly, our findings do not show support for coyotes as urban exploiters, but instead show support for coyotes as urban adapters. These are promising results from a management standpoint because they suggest that coyotes likely can not depend on urban areas exclusively and still require adjacent natural landscape features. With this knowledge, managers can adopt strategies that will encourage coyotes' use of natural spaces by reducing coyote attractants in urban centres, such as compost and potentially jackrabbit populations, in order to reduce the likelihood of human-coyote conflict.
Figure 13. An illustration of the urban avoiders, adapters, and exploiters in the City of Edmonton according to the associations between species frequencies and landscape variables across the cities' urban gradient.