Native prey, not landscape change or novel prey, drive cougar (Puma concolor) distribution at a boreal forest range edge

Abstract

Many large carnivores, despite widespread habitat alteration, are rebounding in parts of their former ranges after decades of persecution and exploitation. Cougars (Puma concolor) are apex predator with their remaining northern core range constricted to mountain landscapes and areas of western North America; however, cougar populations have recently started rebounding in several locations across North America, including northward in boreal forest landscapes. A camera-trap survey of multiple landscapes across Alberta, Canada, delineated a range edge; within this region, we deployed an array of 47 camera traps in a random stratified design across a landscape spanning a gradient of anthropogenic development relative to the predicted expansion front. We completed multiple hypotheses in an information-theoretic framework to determine if cougar occurrence is best explained by natural land cover features, anthropogenic development features, or competitor and prey activity. We predicted that anthropogenic development features from resource extraction and invading white-tailed deer (Odocoileus virgianius) explain cougar distribution at this boreal range edge. Counter to our predictions, the relative activity of native prey, predominantly snowshoe hare (Lepus americanus), was the best predictor of cougar occurrence at this range edge. Small-bodied prey items are particularly important for female and sub-adult cougars and may support breeding individuals in the northeast boreal forest. Also, counter to our predictions, there was not a strong relationship detected between cougar occurrence and gray wolf (Canis lupus) activity at this range edge. However, further investigation is recommended as the possibility of cougar expansion into areas of the multi-prey boreal system, where wolves have recently been controlled, could have negative consequences for conservation goals in this region (e.g. the recovery of woodland caribou [Rangifer tarandus caribou]). Our study highlights the need to monitor contemporary distributions to inform conservation management objectives as large carnivores recover across North America.

Keywords: anthropogenic development; boreal; camera traps; carnivore recovery; predator–prey interactions; range expansion.

FIGURE 1

(a) Map of six camera trap arrays deployed across the oil sands region (OSR) in Alberta, Canada. (b) Distribution of camera traps across OSM Landscape Unit 2 (LU2), the study area. Each circle represents a camera location (n = 47) and increasing circle size corresponds to a higher number of cougar detections. (c) Example image of multiple cougars (n = 5) on a single camera.

FIGURE 2

Relative activity indices at each camera site (n = 47) for potential prey and competitor predator species, including gray wolf (Canis lupus), moose (Alces alces), snowshoe hare (Lepus americanus), native prey (moose and snowshoe hare combined), and white‐tailed deer (Odocoileus virgianius). Activity indices were calculated by summing all independent detections (defined by events occurring a minimum of 30 min apart) for each species and dividing by the total number of active camera days, to create a metric of species activity on the landscape.

FIGURE 3

Predicted relationship between the probability of cougar (Puma concolor) occurrence (±95% confidence intervals) and the best‐supported model at all spatial scales, the relative activity of native prey (summed relative activity of moose, Alces alces, and snowshoe hare, Lepus americanus).

FIGURE A1

Histograms of the percentage of deciduous forests at camera sites (n = 47) collected at (a) 500 m, (b) 2500 m, and (c) 5000 m.

FIGURE A2

Histograms of the percentage of shrublands at camera sites (n = 47) collected at (a) 500 m, (b) 2500 m, and (c) 5000 m.

FIGURE A3

Histograms of the percentage of water at camera sites (n = 47) collected at (a) 500 m, (b) 2500 m, and (c) 5000 m.

FIGURE A4

Histograms of the percentage of linear features (summed proportion of roads, seismic lines, railways, pipelines, trails, and transmission lines) at camera sites (n = 47) collected at (a) 500 m, (b) 2500 m, and (c) 5000 m.

FIGURE A5

Histograms of the percentage of block features (summed proportion of well‐sites, borrow pits, harvest areas, and residential areas) at camera sites (n = 47) collected at (a) 500 m, (b) 2500 m, and (c) 5000 m.

FIGURE A6

Pearsons correlation plots for unscaled continuous variables were used to build a candidate set of generalized linear models at each spatial scale (500, 2500, and 5000 m). Found a significant correlation (Pearson's r > .7) between linear features and shrubland at the 5000 m spatial scale (r = .82, p < .0001).

FIGURE A7

Camera operability across 2022 (designated by the horizontal black lines) for each camera (n = 47) in the OSM Landscape Unit 2 (LU2) study area in northeastern Alberta, Canada. Red dashes represent cougar detections at a camera.

FIGURE A8

Independent detections of cougars by month across all cameras (n = 47) in the OSM Landscape Unit 2 (LU2) study area in northeastern Alberta, Canada.