Data

DATA FRAME

A Microsoft Office Excel spreadsheet was used to store the data frame that was created using the process illustrated in Figure 5. The table consists of 90 rows, representing wildlife monitoring sites, and 80 columns containing information on the visitation rate per site of 25 mammal species and 44 bird species, as well as urban factors that characterize the monitoring sites and their surrounding areas.

Data Types

Predictor Variables → The 11 urban factors are defined as the numerical-continuous predictor variables for the multivariate statistical analyses.

Response Variables → The numerical-continuous response variables are defined as the rates of wildlife visitation rate [Wildlife Visits/Day] of 25 mammal species and 44 bird species estimated for each monitoring site.

Table 1. Data frame description

Column A: Categorical-Ordinal Variable : ID from 1 to 90 assigned to each monitoring site.
Columns B-Z: Numerical-Continuous Variables : Wildlife visitation rate [Visits/Day] of 25 mammal species estimated in each monitoring site from the data recorded from 2018 - 2021. The real values are not presented due to data management policies implemented by the City of Edmonton to safeguard biodiversity.
Columns AA-BR: Numerical-Continuous Variables : Wildlife visitation rate [Visits/Day] of 44 bird species estimated in each monitoring site from the data recorded from 2018 - 2021. The real values are not presented due to data management policies implemented by the City of Edmonton to safeguard biodiversity.
Columns BS-BV: Numerical-Continuous Variables : Distance, in metres, calculated from each monitoring site to the city centre (DCC), nearest road (NRo), nearest trail (NTr), and nearest water body (NWB).
Columns BW-BX: Numerical-Continuous Variables : Density, in km/km², of roads (RD) and trails (TD) calculated within a buffer area of 3.1 ha surrounding each monitoring site.
Columns BY-CC: Numerical-Continuous Variables : Calculated area ratio [%] of four land covers: water (WAT), developed area (DEV), modified area (MOD), naturally wooded (WOOD) and non-naturally wooded (NWOOD) areas within a buffer area of 3.1 ha surrounding each monitoring site.

DATA EXPLORATION

WILDLIFE VISITATION RATE

During the monitoring period from 2018 to 2021, the mammal species most frequently recorded were coyotes, white-tailed deer, snowshoe hares, red squirrels, and white-tailed jackrabbits, which will be referred to as the common species for further analysis. On the other hand, cougars, elks, Canadian lynxes, Richardson's ground squirrels, and American badgers were the least frequently detected species during the three-year monitoring period and will be identified as the rare species within the city (Figure 9).

Figure 9. Total visits of mammal species in Edmonton b

Dot plots depicting the total number of visits by mammal species as recorded on 90 monitoring trap cameras (Figure 1) between May 2018 and September 2021. The images in the dots represent the five most common (coyotes, white-tailed deer, snowshoe hares, red squirrels, and white-tailed jackrabbits) and rare (cougars, elks, Canadian lynxes, Richardson's ground squirrels, and American badgers) detected species. The x-axis shows the common names of the species, which are described as follows:

Mammals

American badger [1], Beaver [2], Canada lynx [3], Common muskrat [4], Cougar [5], Coyote [6], Elk [7], Fisher [8], Least chipmunk [9], Moose [10], Mouse sp. [11], Mule deer [12], Northern flying squirrel [13], Porcupine [14], Raccoon [15], Red fox [16], Red squirrel [17], Richardson’s ground squirrel [18], Small mammal sp. [19], Snowshoe hare [20], Squirrel sp. [21], Striped skunk [22], Weasel sp. [23], White-tailed Deer [24], White-tailed jackrabbit [25].

During the three-year monitoring period, the most frequently detected bird species were magpies, gulls, terns and allies, crows, mallards, and bohemian waxwings, which are classified in this study as the most common species. In contrast, Brewer's blackbirds, eastern kingbirds, ruffed grouse, short-eared owls, and white-breasted nuthatches were the least frequently detected bird species between 2018 and 2021 and will be referred to as the rare species for further analysis (Figure 10).

Figure 10. Total visits of bird species in Edmonton b

Dot plots depicting the total number of visits by bird species as recorded on 90 monitoring trap cameras (Figure 1) between May 2018 and September 2021. The images in the dots represent the five most common (magpies, gulls, terns and allies, crows, mallards, and bohemian waxwings) and rare (Brewer's blackbirds, eastern kingbirds, ruffed grouse, short-eared owls, and white-breasted nuthatches) detected species. The x-axis shows the common names of the species, which are described as follows:

Birds

American Coot [1], American Robin [2], Bird of Prey sp. [3], Blue jay [4], Blue-winged teal [5], Bohemian waxwing [6], Boreal chickadee [7], Brewer’s blackbird [8], Brown-headed cowbird [9], Canada goose [10], Chipping sparrow [11], Common merganser [12], Common raven [13], Common redpoll [14], Cooper’s hawk [15], Crow [16], Downy woodpecker [17], Duck sp. [18], Eastern kingbird [19], Gray partridge [20], Green-winged teal [21], Gray catbird [22], Gulls, terns and allies [23], Hairy woodpecker [24], House sparrow [25], Magpie [26], Mallard [27], Northern flicker [28], Ovenbird [29], Pileated woodpecker [30], Red-breasted nuthatch [31], Red-winged blackbird [32], Ring-necked pheasant [33], Rock pigeon [34], Ruddy duck [35], Ruffed grouse [36], Short-eared owl [37], Songbird sp. [38], Sparrow sp. [39], Starling [40], Warbler sp. [41], White-breasted Nuthatch [42], Wilson’s snipe [43], Woodpecker sp. [44].

When the wildlife visitation rate is aggregated spatially, the total number of mammals detected per monitoring day (Figure 11-Left) is concentrated mostly around the North Saskatchewan River Valley. However, the number of visits shows greater spatial variation, as indicated by the three classification ranges (yellow, orange, and red), which are widely distributed throughout the study area, with the lowest detections concentrated around Mill Creek. On the other hand, the distribution of the bird visitation rate detected per monitoring day (Figure 11 - Right) shows low spatial variability, with almost all monitoring sites recording the same value per day (≤ 0.19). The pattern of spatial variability mentioned here is confirmed by the distribution of data represented in the box and whisker plots (Figure 12).

Figure 11. Spatial distribution wildlife visitation rate

Spatial distribution of the 90 wildlife monitoring sites, illustrating the total number of wildlife visits detected, which has been normalized by the days each site was active [Wildlife Visits/Day]. The maps represent the rate recorded for mammals (Left) and birds (Right).

The boxplots (Figure 12) show the 25 mammal species (a) and 44 bird species (b) and their total visits per monitoring day. Generally, mammals exhibit more variability than birds in terms of visitation rate, which is consistent with what is observed in Figure 11. However, when this analysis is broken down by species, most species appear to be evenly distributed across the monitoring sites, with a few exceptions related to species identified as common and rare (Figures 9 and 10), indicated by larger boxes and visible outliers.

Figure 12. Visitation Rate distribution of all mammal and bird species

Box and whisker plots summarizing the distribution of the wildlife visitation rates of the 25 mammal species (a) and 44 bird species (b) recorded on 90 monitoring cameras (Figure 1) from May 2018 to Sept 2021.

Based on Figure 12, it can be inferred that the visitation rate data does not follow a normal distribution, likely due to zero-inflated datasets. Since these data are considered response variables, they are typically expected to meet the normality assumption. Although the multivariate statistical analyses employed in this study are robust enough to disregard the normality assumption, transforming the data to fit a normal distribution can improve interpretability by standardizing all values to a mean of 0 and standard deviation of 1, thereby giving equal weight to each species in the analysis. To achieve this, the data were transformed using a combination of inverse and square root transformations. This transformed data (Figure 13) is the best approximation of normality that can be achieved for these datasets, and will be used for both rotation and distance-based multivariate techniques to simplify the complexity of the data.

Figure 13. Distribution of the transformed Visitation Rate of all mammal and bird species

Box and whisker plots summarizing the distribution of the transformed wildlife visitation rates of the 25 mammal species (a) and 44 bird species (b) recorded on 90 monitoring cameras (Figure 1) from May 2018 to Sept 2021. The transformations applied were inverse and square root which standardize all values to a mean of 0 and standard deviation of 1.

URBAN FACTORS

Figure 14. Distance to urban factors

Box and whisker plots summarizing the data within Columns BS-BV in Table 1. This data represents the distance from wildlife monitoring sites to four spatial features within the city of Edmonton, city centre, roads, trails, and water bodies.

Distance

Three metrics for urban factors were assessed. Firstly, the distance from the monitoring sites to the city centre, roads, trails, and water bodies was estimated (Figure 14). The data showed a positive skew, and the distance to the nearest water body had the highest variability.

Density [km/km²]

Secondly, the surroundings of all monitoring sites were evaluated by creating a 3.1 ha polygon or a 100 m buffer. Within this buffer, it was examined the density of roads and trails, and found that trail density varied more than road density (Figure 15).

Figure 15. Density of urban factors

Box and whisker plots summarizing the data within Columns BW-BX in Table 1. This data represents the density of roads and trails around 3.1 ha surrounding the wildlife monitoring sites.

Figure 16. Proportion of urban factors

Box and whisker plots summarizing the data within Columns BY-CC in Table 1. This data represents the area proportion of land cover classes around 3.1 ha surrounding the wildlife monitoring sites.

Proportion [%]

Finally, the area proportion of all geographic factors was estimated. Forested areas were found to occupy most of the sites, followed by urbanized areas (Developed and Modified). This is due to the fact that the majority of sites are located in river valleys and ravines (Figure 16).

b The total abundance of each species has been computed for every monitoring site, but the data has not been included in Table 1. This is due to data management policies implemented by the City of Edmonton to safeguard biodiversity.