Are coyotes becoming more dangerous? This case study highlights a concerning growing trend across North American cities.

Habituation, Den Defence and Predatory Opportunism Appear to Have Contributed to Coyote Aggression in an Urban Park

Nathan C. Lewis, Colleen Cassady St. Clair, William L. Harrower, Sarah Benson-Amram
Animal Behaviour 238 (2026) 123617

Article Info

  • Received: 29 January 2025
  • Initial acceptance: 27 May 2025
  • Final acceptance: 23 February 2026
  • Available online: 6 July 2026
  • Keywords: carnivore attack, coyote attack, den defence, habituation, human-wildlife coexistence, human-wildlife conflict, human-wildlife interaction, urban carnivore, urban coyote, urban park

Abstract

Human-wildlife conflicts in urban settings are increasing and sometimes cause injury or mortality for people and wildlife. Conflicts may be exacerbated by large increases in outdoor human recreation, such as during the COVID-19 pandemic, when people were prevented from gathering indoors and made increasing use of parks. This appears to have occurred in Vancouver's Stanley Park, British Columbia, Canada, where an unprecedented 45 coyote, Canis latrans, attacks on humans occurred between December 2020 and August 2021. To understand how this situation arose and might predict similar future problems, we compared reported aggressive human-coyote encounters and coyote sightings in Stanley Park from 2020 to 2021. We sought to identify predictors of aggression among spatial and temporal variables associated with human activity and coyote behaviour. We also compared characteristics of human victims of attacks to those of park users in 2024.

Coyote aggression was higher:

  1. during pandemic restrictions on indoor gatherings (associated with higher human activity outdoors),
  2. on trails with greater human activity,
  3. during times of the day and week with relatively low human activity, and
  4. in areas closer to coyote dens and farther from potential sources of anthropogenic food, including garbage bins and picnic areas.

While proximity to anthropogenic food sources did not positively predict aggression, anecdotal reports described the behaviour of the aggressive coyotes as likely food-conditioned. Relative to observed park users, victims of coyote attacks were more often running and alone when attacked, but there was no effect of a person's age or gender, or of domestic dog presence. Our results suggest coyote aggression in Stanley Park likely resulted from a combination of habituation, den defence, opportunistic predatory behaviours, and possibly food conditioning. To avoid similar situations in the future, wildlife managers could encourage the public to recreate in groups, during peak-use hours and away from known den sites.

Introduction

Increasing urban populations and infrastructure cause reductions in natural land cover and intensifies human visits to urban parkland, both of which are associated with increasing reports of human-wildlife conflict in urban areas. Conflicts take many forms, including the spread of diseases, competition for resources and physical confrontations between wildlife and humans, pets or vehicles. Wildlife attacks on humans typically attract the most negative public attention and can be detrimental to conservation efforts and policies that promote coexistence with wildlife. Thus, understanding the behavioural mechanisms and contextual variables involved in aggressive encounters between people and wildlife could improve both public safety and wildlife conservation.

The likelihood and severity of wildlife conflict with humans depends partly on the individual animal's perception of and experience with humans. Animals that perceive humans as threats are more likely to avoid them, whereas animals with repeated benign experiences with humans are likely to habituate to them. In bears (Ursus spp.), habituated individuals make greater use of developed landscapes, which may increase the likelihood of accessing anthropogenic food resources and becoming food conditioned. In elk, Cervus canadensis, habituated animals are less likely to flee from an interaction with humans, in turn making conflict more likely. When carnivores live among people, conflict is more likely to escalate into attacks when it involves defence of dens, offspring or food resources. Some carnivore attacks may result from predatory impulses in response to humans that run away from or unintentionally run past a carnivore. All these sources of conflict (habituation, food conditioning, defence, predatory drive) appear to be occurring with increasing frequency in urban-dwelling coyotes, Canis latrans, which are involved in 31% of the carnivore attacks on humans in North America and are increasingly prevalent in urban areas.

Coyote attacks on people in urban areas are most often attributed to habituation and food conditioning, which stem from past benign encounters with humans, intentional feeding by humans and accessible attractants in urban areas. Urban coyotes also appear to be bolder than rural coyotes and this tendency increases in parks where they often encounter people. Defensive aggression may also contribute to some attacks on humans and domestic dogs, Canis lupus familiaris, because aggression towards humans and dogs tends to increase with proximity to dens. On the rare occasion that coyote attacks appear to be motivated by predation, they more often target people who are small (including children) and alone.

An opportunity to conduct a comprehensive analysis of coyote attacks at a single location within a short time frame was presented by a cluster of coyote attacks on humans in Vancouver, Canada's Stanley Park. Highly unusual coyote aggression occurred in this urban park between December 2020 and August 2021, during the global COVID-19 pandemic. In addition to 45 attacks by coyotes biting or scratching people during that period, there were 63 aggressive encounters without physical contact reported. Before 2020, only one other aggressive encounter and no bites had been reported in Stanley Park since 2011. Attacks ceased only after the park was closed on 1 September 2021 and 11 coyotes were removed. The 45 attacks in this 9-month period in Stanley Park are equivalent to more than 70% of the coyote attacks reported by the news media in Canada between 2000 and 2015.

Methods

Study Site

Stanley Park (49.3036°, -123.1442°) is a 4 km² urban park peninsula connected to central Vancouver that is ~90% surrounded by ocean in the Burrard Inlet and English Bay. The land cover includes mixed coniferous-deciduous rainforest (~75%), with more than 27 km of forested trails, large open areas (~10%), and two large bodies of water: Lost Lagoon and Beaver Lake.

Coyote Reports

We collected and compared three types of coyote reports from Stanley Park: attacks, noncontact aggression, and sightings.

  • Attack: a coyote initiates physical contact and causes bodily harm.
  • Noncontact aggressive encounter: a coyote aggressively advances or displays aggressive body language (e.g., lunging, snarling).
  • Sighting: a human observation of a coyote that does not engage with a human or pet.

Figures

Figure 1: Map of Reported Encounters in Stanley Park

View Figure 1 Data via Mendeley Data Repository

Figure 1 Description: A map showing all human-coyote interactions with location data in Stanley Park between 11 March 2020 and 31 August 2021. Points show attacks (38/45), noncontact aggressive encounters (63), and sightings (517) across land cover classes (Developed, Open, Natural, Water).

Results

Probability of Aggressive Encounters

In total, we reviewed 545 reports from between 11 March 2020 and 31 August 2021, including 45 attacks, 63 noncontact aggressive encounters and 437 coyote sightings.

Figure 2: Coefficient Estimates of Predictor Variables

View Figure 2 Data via Mendeley Data Repository

Figure 2 Description: Estimated coefficient and 95% confidence interval of each variable included in the top logistic regression models. Pandemic lockdown phase was the strongest predictor overall based on effect sizes.

Figure 3: Model Predictions for Aggression Probability

View Figure 3 Data via Mendeley Data Repository

Figure 3 Description: Model predictions showing that the probability of aggression increased with:

  • Pandemic lockdown phase (Phases 2 and 3)
  • Proximity to the nearest den
  • Proximity to the ocean
  • Times of day with lower human activity
  • Greater daily precipitation
  • Weekdays vs. Weekends

Figure 4: Density Distribution of Reports over Time

View Figure 4 Data via Mendeley Data Repository

Figure 4 Description: Density plot showing the dramatic surge in attacks and aggressive encounters beginning in late 2020 when indoor gatherings were prohibited.

Characteristics of Attack Victims

Figure 5: Observed vs. Expected Characteristics of Victims

View Figure 5 Data via Mendeley Data Repository

Figure 5 Description: Key victim demographics compared to baseline park users:

  • Gender/Age: No significant difference from expected park user rates.
  • Activity: Runners were overwhelmingly over-represented among victims (P < 0.001).
  • Group Size: Lone individuals were significantly more likely to be attacked (P < 0.001).
  • Dog Presence: No victims were accompanied by dogs.

Discussion

Relative to sightings without aggressive coyote behaviour, we found that aggressive encounters were more frequent during the pandemic-induced prohibition on indoor social gatherings. However, on a finer temporal scale, aggressive encounters occurred more often on days and at times of day when human activity was lower (e.g., night/crepuscular hours, weekdays, rainy days).

Key Takeaways:

  • Habituation & Human Activity: Outdoor activity spiked in Stanley Park during pandemic restrictions. Continuous habituation may have contributed to aggressive encounters becoming increasingly severe.
  • Den Defence: Aggressive encounters occurred closer to presumed den sites than did sightings.
  • Predatory Opportunism: 54% of victims were running at the time of attack, which is known to induce chases by coursing predators. Coyotes were also more likely to attack lone individuals.
  • Food Conditioning: Though aggressive encounters occurred farther from structured garbage bins/picnic areas, anecdotal evidence strongly suggests illegal feeding and unmanaged trash contributed significantly to bold behavior.

Key Implications for Management & Public Safety

  1. Anticipate Conflict: Park managers should anticipate increasing human-wildlife conflicts as human outdoor activity rises in urban natural areas.
  2. Den Buffers: Recreationists should be kept away from known den sites when possible, regardless of season.
  3. Safety Guidelines for the Public: Visitors recreating in urban parklands with known carnivore presence should be warned that they face an elevated risk of aggressive encounters if they are:
    • Running
    • Recreating alone
    • Recreating during hours of low human activity (dusk, dawn, night)

Open Copyright & Usage Notice

  • Source Publication: Animal Behaviour, Volume 238, 123617 (2026)
  • DOI: 10.1016/j.anbehav.2026.123617
  • Open Access License: Distributed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0).