Using GIS Spatial Analysis and Statistical Models to Predict Habitat

Coyotes are common predators across North America, but little is known about their features, distribution, and response to the urban environment. Unlike their bigger wild dog counterparts, there have been few studies trying to associate them with the suburban and urban locales. According to Sears et al. (2003), in the Northern America, the coyote species expand to regions where the wolf population has declined such as the sections not occupied by the Rufus and the Lycaon species in Ontario. However, conversion of the natural habitat to cities, residential areas, agricultural lands, parks, and other human development and industrial structures has continuously reduced the number of natural habitats, a fragmentation that significantly impacts the coyote’s survival. They remain a shy class of the wild dogs that has not yet fully learned to interact with humans in the open like the gray foxes and opossums. The insufficient information about the animals is associated with their survival tactics and unavailability in the public spaces. They have been able to keep a low profile hence causing fewer conflicts with the humans, a characteristic that has made them the acceptable type of urban carnivores. Nevertheless, the current change in the urban setup and reduction of the natural habitats is forcing them to redefine their survival and hence distribution.
Modernization of the Calgary cities comes with adverse effects for the coyote. Although the increasing urban population and changing eating habits are providing a perfect setup for the coyote diets, there is a higher stake in their safety as they continue to breed.

According to Riley et al. (2003), animals associated with night activities like the coyotes are naïve to use public spaces in urban setups. They are thus forced to roam at night to avoid contact with people. However, the increasing rates of transformation to a day and night working economy within these urban and suburban perimeters continues to pose a significant challenge for the survival of the species. Human population does not entertain any carnivores and especially during the night. The mortality rate of the coyotes is thus bound to increase due to human-related causes such as poisoning, shooting, collision with the vehicles and direct hunting in the ideology that the animals dangerous to them, their pets, and children. However, their distribution and survival rest on the influence of the natural habitats, human population pervasiveness, the wildlife agencies intervention and the access to particular diets and prey abundance (Gehrt, 2007).
Apart from the encroachment of their natural habitats, many other factors determine the distribution of coyotes in North America and particularly in Calgary. According to Holroyd, (2008), habitat availability is measured by the amount of each vegetation type and its suitability to support the species. One of the factors evident from the study is the population of other wild dog family members especially the bigger ones. These pose a threat to the coyotes that are seemingly the smallest in the class. Apart from attacking and killing them, the larger carnivores are not opportunistic feeders and hence make savaging for coyotes a challenge. For some time, this has acted as the reason why the species had to move from one region to the other. Nearing the urban centers happen to be an easier solution due to the abundance of food, availability of shelter and eviction of the other carnivores that are presumably dangerous to humans and domestic animals lives. According to Lukasik and Alexander (2012), coyotes have of late subsidized their diet with the trash, domestic pets, and anthropogenic food items within the urban areas hence making them able to thrive in almost all cities within America and Canada in particular. Their research indicates that coyote is the dominant predator within Calgary whose introduction has influenced the community structure. Although the study was based in Calgary and in particular the region between the Bow and Elbow rivers in the southwest Alberta, the distribution of coyote extends to all other parts of the city ranging from Arbor Lake, Edworthy Park, Fish Creek Park, Nose Hill, North Glenmore Park to Stanley Park and Tom Campbell. Compared to the past few decades, the southwest of Calgary has experienced a population surge making it one of the largest metropolises in Canada. The residential development act as the leading cause of public space invasion thus defragmenting the coyote species habitations.
The survey data revealed that Coyotes were evident within many regions in Alberta. Most of this data was collected from observations and citizens’ reports from October 2008 to April 2013. The samples (886) were sufficient for predicting the entire distribution of the species. Nonetheless, some pseudo-absences had to be assumed and conflicting information discarded. The resultant 1685 absence location was a clear indication of the internal fragmentation of Coyote and their modern distribution. The species distribution was found to be closely related to the predictor variables such as the topography, vegetation, disturbances and distance from particular natural resources. Topographic aspects that came into play were the aspect of slope and terrain. The results of the correlation analysis from the habitat modeling proved that the degree of elevation, change of landscape, east-ness or north-ness, the ground cover, and distance to water sources do not necessarily impact the prevalence of the coyote species. Rather, the slope variations, distance to roads and parks, terrain ruggedness and the number of pets and children were then positive indicators of the presence of the animals. The factors differ with those of other species like raccoons whose existence is dictated by the forest cover and proximity to water sources (Baldwin et al. 2006). According to the analyzed data, it the likelihood of the species occurring in Calgary was at a rate of one coyote for every 30m x 30 m area. However, this was only true within the highly suitable environments where the rate could be zero for the other unsuitable habitats. The average probability was 0.5 coyote in the same coverage.
The analytic and graphical mapping of the habitat distribution of the coyote species showed a closer link between the models. The regions in the Northwest and South of Calgary shown in the Figures 3 (a-c) signify a larger correlation between the anthropogenic structures and the natural formations. They stand out as the sections with the highest population of the coyotes. Further, the spatial distribution of the habitat is highly fragmented. There is a close link between these results and the previous research conducted withing the same region. According to the study carried out by Lukasik and Alexander (2012), for coyote ‘scats’ over the period between August 2006 and September 2007, Coyotes were reported to follow along a linear feature and ‘scats’ were strategically deposited at famous locations such as along the borders and on trail junctions. A similar path as denoted in the generated maps which signified more scatter within adjacent northwest and southeastern regions. The prevalence of the coyote scats according to the study was higher in in Arbor Lake (AL), Fish Creek Park (FC) and North Glenmore Park (NGW). On a closer assessment, Al is in the Northwest of Calgary, FC to the southern end and NGW to the western end of Calgary. All the three falls under the same collection sites recorded in the scatter maps generated in this research. Riley at al. (2003) argues that natural areas are the largest component of the home range of coyotes. Despite them being omnivores, and while the developed and altered open areas offer more food sources for any optimistic feeder, nun-natural areas are still unsuitable for the species. The northern and southwestern regions of Calgary are the less developed hence maintain some of the natural aspects. It is one of the defining reasons for the nocturnal stagnation to the less developed areas by the species. Their low rates in the developed areas to the east of Calgary are a result of human-related activities such as anticoagulant poisoning. Their feeding habits and dependence on the rodents predisposes them to consume the poison meant for controlling the later. Most of these mortalities occurred within the residential settings, public parks, and golf courses.
Poessel, Gese and Young (2016) study of the coyote population in Southern California indicated a closer relationship between the species occurrence with the intensity and proximity of urbanization. Whenever a protective cover is available, Indiana coyotes were reported to occupy areas with dense housing adjacent to the large forested patches. In other sections of Michigan, Chicago, and Detroit, the species were reported to prefer areas with more ground cover and near water sources. Although coyote uses water for drinking, this research indicated that there was a stronger link between the animals and water sources or the riparian areas. The availability of the resource similar to food source dictates the animals’ distribution.
According to Hinton, Manen and Chamberlain (2015), “transient coyotes do not maintain territories and exhibit nomadic movements with no fidelity for any one area.” It means that their migration is determined by other factors and primarily the human activities and social changes. The more the humans encroach the territories occupied by the coyote, the further they have moved until recently when almost all the remaining coverage is insufficient or occupied by other carnivores and coyotes can only learn to live with humans. The research indicates that extensive migrations by transients follow decisions by individuals that impact the coyote ecology including foraging behavior, competition, and habitat selection. It, in turn, affects the population structure and processes over a wide geographical area. From the results of the research on coyote population in the North Carolina, only 30% of the species are transients. Therefore, it is usually challenging to break up the breeding pairs in a group making the transients the dispersing juveniles. Previous studies by Dodge and Kashian (2013) linked coyote to the access to tree cover and the availability of green spaces. On the other hand, Poessel, Gese and Young (2016) tied the conflicts between the species and humans on the developed land cover. Combining the two and the results of this research, it is evident that the habitat distribution of the coyotes is a factor of not only the food sources but also the ground cover, the human activities, ecological factors and presence of other carnivores within a region.
Model Comparison
Modeling techniques are important when predicting and estimating various aspects of research such as quantity, suitability of certain conditions and displaying locations. In the case of coyotes, they are significant when predicting the suitable habitat and areas currently used by the species. They thus serve as a means to explain the spread of the coyotes, their re-introduction, and future management. The presence-only data is the most modeled due to the availability of data from individuals and previous research databanks. However, this lacks an accurate absence data and contains an unknown sampling bias. According to Williams (2003), the quality and quantity of data occurrence is important and often a limiting factor. In habitat modeling, the accuracy is determined by the absence data. Besides having doubts on the plot precision, the data is primarily dependent on particular factors like the proximity to the cities, roads, and natural resources.
One method applied in the research was the statistical modeling. It involved collecting data from trusted sources such as the Calgary University databases and professors as well as accessing raw information by contacting respondents. Statistical analysis tools were then applied to analyze the data including the use of SPSS. It culminated into the discriminant function analysis prediction that examined the statistical significance of the predictor variables. Although the method depended on firsthand information, there are multiple pseudo-absences generated that could result in an exaggerated analysis of the habitat. The second prediction employed the logistic regression model that tested the significance level and estimated the coefficient for each predictor variable. For this type of modeling, the predictors were only significant at the values of p<0.05. The generalized additive model prediction employed a four-predictor variable whose significance rested on p-value<0.05 as well. However, the vegetation heterogeneity, distances, and terrain stand out as the primary stimulants of the habitat distribution. Although the models produced close results, the logistic regression was more efficient than the discriminant function model. The generalized additive model was the less efficient.
Landscape Analysis
From the map of the summary of road densities for the period between 2002 and 2005 in Calgary (figure 7), there has been an increased impact on the landscape in the eastern and southeastern parts of Calgary. With the buildings occupying an area of 31670m2 in 2015 from 26291m2 in 2002, the amount of the land cover is significantly reduced, a process that has caused wild animals to migrate. The modern landscape is entirely different from that of decades ago. Too much has changed in the region primarily from human development activities like setting up of new roads, cities and oil companies. According to the research results, trails and roads density has increased significantly from 12197 to 14348 square meters between 2002 and 2015 respectively representing a percentage of 2.54% decrease and hence a similar reduction in the forest cover. Further, the population of the forested cover such as the herbaceous and grasslands that acts as a source of food for the coyotes has decreased in size from 42651 to 34785 m square within the same period. All these culminate into the recent changes in the landscapes that are proving detrimental to the survival of many wild animals.
According to the study, the trend of increased human disturbances from heterogeneity, fragmentation and landscape changes is bound to break significantly within the next decade. The patch size has been reported to have enlarged, but the density lowered for the year 2002. The generated area-weighted mean shape indexes presently indicate an incredible landscape compared to the previous years, a distinct suggestion it is gradually but continuously transforming to uneven terrain unsuitable of the survival of coyotes. According to the information contained in table 10 regarding the landscape and land cover measures between 2002 and 2015 in Calgary, it is evident that the landscape has drastically changed over the past period. The area built-up has moved up from 30.98 to 41.0 percent which is an increase of 10.02 percent. In the end, the extent of the forest cover has reduced by 12.94 percent from 50.27 to 37.33. The two factors are inversely proportional in that when the built up area increases, the forest cover decreases but the later happens at a higher rate. A further expansion of the other development processes such as roads could be the reason for the massive reduction in the forest cover. The area-weighted mean shape index is an actual proof of the total reduction of the ground cover whose values indicated a reduction of 11.63m square mean shape. The areas hit by the landscape changes in the far north and south as well as the southeastern section in Calgary. According to the analytical results of the research conducted by Poessel, Gese and Young (2016), human-coyote conflicts depend on the type of landscape. Rural landscapes recorded the fewest cases of conflicts where the exurban had the highest. The relationship between the geographies also indicated the population size where the developed regions such as urban centers had the largest population but a proportional number of conflicts with the species. The urban sector was reported to be dense in the west, southwest, midwest, southeast and northeast at a reducing rate respectively. Figure 1 generated in this research verifies this where the primary causes of the drift in the population density are pointed out as the residential settings in the west and industrial growth in the southeastern region. In the end, the only parks and green spaces remain in the far south and northern boundaries. Nonetheless, these spaces are however getting depleted at a higher rate, and the coyote population is at risk.
Conclusion
The future existence of the coyote species rests on a range of factors from the exploitation by other predators, encroachment of their natural habitats, poor conservation mechanisms and human-coyote conflict that is rising each day with the increase in the number of animals within the urban settings. Despite having a broad niche tolerance and high reproduction rate, lack of natural habitats is becoming a challenge for the continuation of the species. The research has revealed that the public space has been made private through the erection of homesteads, industries, and agricultural plantations. The remaining spaces are slowly turning into parks, development areas for roads, airports and other national development as deemed by the government. Soon, Calgary will have transformed into an urban setting with the forest cover occupying less than 10 percent of the entire coverage. This presents a larger challenge for not only the coyote species but also the wild dog family.
Coyotes have gained significant interest in the wildlife management of late owing to their relatively negative character primarily brought about by human processes. Nevertheless, the management units have been challenging to define due to the increase in the rates of reproduction, immigration and reducing food resources. Habitat availability and their continuous distribution is the central factor in the conservation of the animals. This research has revealed that coyotes are not a dangerous species due to their shy characteristics. They try to avoid any overlap with humans and whenever people infringe their space, they retaliate by either migrating or adapting to the changes. The only risk they pose is to the pets and other smaller domestic animals. However, with the increasing changes in lifestyle, has been proven to change where the animals may come out openly and try to reclaim their habitats. It calls for better habitat modeling techniques, conservation procedures and mass education on the most efficient ways to deal with the species. The three modeling methods applied in this research (discriminant function analysis, generalized additive method, and logic regression model) indicates a close link between the population of coyotes and other ecological factors such as terrain, vegetation heterogeneity, urbanization, and food sources. All these were reported to play a central role in the existence of the animals and their acceptance in the urban settings within Calgary. While the statistical analysis may not have been very accurate, it pinpoints that more need to be done to conserve the natural environment if we were to manage the animals. As suggested by Williams (2003), probability detection through modeling provides a better estimate than making a total assumption. Based on the estimates, the government can initiate a more thorough research and devise long-term conventional methods for coyotes.
Notably, coyotes utilize natural areas than the developed regions. With every square meter change in development, the risk factor for the species is proportionally increased. These ranges from poisoning, collisions with vehicles and direct hunt down in case a coyote attacked oneself (Riley et al. 2003). The conflicts are more in the areas with exurban housing densities. The government must embrace the responsibility of safeguarding her animals through protecting and zoning certain areas to act as forest reserves if they were to end human-coyote discords and conserve the species. On the one hand, this will reduce the human animal (carnivore) interaction and hence reduce animal-based hazards while the wildlife organization will have distinct working boundaries. The citizens must also be educated to value the wildlife and promote their conservation by preserving green spaces, forests and being cautious with the carnivores on the road.
References
Baldwin, A. at al., (2006). Predicting Raccoon, Procyon lotor, occurrence through the use of microhabitat variables. Canadian Field-Naturalist 120(2): 225-231.
Dodge, W. B., & Kashian, D. M. (2013). Recent distribution of coyotes across an urban landscape in southeastern Michigan. Journal of Fish and Wildlife Management, 4(2), 377-385. DOI: 10.3996/062013-JFWM-040
Gehrt, S. (2007). ECOLOGY OF COYOTES IN URBAN LANDSCAPES (pp. 2-10). Dundee: Ohio State University. Retrieved from http://digitalcommons.unl.edu/icwdm_ wdmconfproc
Hinton, J., van Manen, F., & Chamberlain, M. (2015). Space Use and Habitat Selection by Resident and Transient Coyotes (Canis latrans). PLOS ONE, 10(7), e0132203. http://dx.doi.org/10.1371/journal.pone.0132203
Holroyd, P. (2008). Towards Acceptable Change: A Thresholds Approach to Manage Cumulative Effects of Land Use Change in the Southern Foothills of Alberta (Masters). UNIVERSITY OF CALGARY.
Lukasik, V. & Alexander, S. (2012). Spatial and Temporal Variation of Coyote (Canis latrans) Diet in Calgary, Alberta. Cities and The Environment, 4(1), 1-25. http://dx.doi.org/10. 15365/cate.4182011
Poessel, S., Gese, E., & Young, J. (2016). Environmental factors influencing the occurrence of coyotes and conflicts in urban areas. Landscape And Urban Planning, 157, 259-269. http://dx.doi.org/10.1016/j.landurbplan.2016.05.022
Riley, S., Sauvajot, R., Fuller, T., York, E., Kamradt, D., Bromley, C., & Wayne, R. (2003). Effects of Urbanization and Habitat Fragmentation on Bobcats and Coyotes in Southern California. Conservation Biology, 17(2), 566-576. http://dx.doi.org/10.1046/j.1523-1739.2003. 01458.x
Sears, Hilary J. et al., (2003). Landscape influence on Canis morphological and ecological variation in a Coyote-Wolf C. lupus × latrans hybrid zone, southeastern Ontario. Canadian Field-Naturalist 117(4): 591-600.
Williams, A. (2003). The Influence of Probability of Detection When Modeling Species Occurrence Using GIS and Survey Data (Doctorate). Virginia Polytechnic Institute and State University.