A mule deer steps onto a trail in Yellowstone at 2 a.m., crosses quickly, and vanishes into timber. During the 2020 COVID shutdowns, that same crossing happened in broad daylight. The deer did not change because the trail moved. It changed because we came back.
That pattern, repeated across hundreds of animals and thousands of data points, is the central finding of a GPS-collar study published in Proceedings of the Royal Society B in 2025. Researchers tracked 229 individual mammals from 10 species across 14 National Park Service units, comparing movement data from 2019, a normal visitation year, with 2020, when pandemic closures emptied trails and campgrounds. The results were consistent: animals avoided roads, trails, and buildings during normal years, expanded their daytime activity when people disappeared, and contracted it again once visitors returned.
The study is the largest controlled test to date of how ordinary human presence, not hunting, not development, just people walking through a landscape, reshapes wildlife behavior inside the very places designated as refuges.
Avoidance shows up everywhere researchers look
The GPS-collar data did not land in isolation. A separate camera-trap study published in Scientific Reports in 2023 used motion-triggered cameras along trails and in nearby backcountry at a single national park. It found that many species shifted where they spent time and became markedly more nocturnal when hikers were present. That study reported that 16 out of 22 mammal species altered where and when they moved in response to non-motorized recreation alone. Animals that once traveled ridgelines and drainages during the day began using them at night or rerouting to less-trafficked corridors.
The clearest causal evidence came from an experiment with pumas. In a playback study published in Proceedings of the Royal Society B in 2017, researchers broadcast recordings of ordinary human conversation at puma feeding sites in the Santa Cruz Mountains of California. The cats fled more often, took longer to return, and cut their feeding time by more than half over the following 24 hours. No gunshots, no vehicles, no dogs. Just the sound of people talking was enough to make an apex predator abandon a kill.
These three lines of evidence, large-scale collar tracking, camera-trap occupancy modeling, and controlled sensory experiments, point in the same direction. Wildlife treats human presence as a predation-level threat. A widely cited 2015 analysis in Science by Chris Darimont and colleagues characterized humans as a “super predator” whose kill rates on adult prey far exceed those of wolves, bears, or mountain lions. That framework helps explain why the fear response is so strong and so general: animals that evolved under predation pressure do not distinguish between a hiker and a hunter. Both register as danger.
What the data cannot yet tell us
The GPS-collar study is broad, but it has gaps. The published analysis does not include per-park raw location files or exact closure timelines for each of the 14 units. That makes it hard to compare how quickly wildlife behavior rebounded after reopening, or whether parks that allowed partial access during 2020 saw different outcomes than those with total shutdowns.
The camera-trap work, meanwhile, quantified species-level changes in occupancy and activity timing, but full model outputs and activity tables have not been made widely available for independent reanalysis. That limits the ability of outside researchers to test alternative explanations or explore how responses vary by body size, diet, or social structure.
Perhaps the biggest open question is whether these behavioral shifts carry real fitness costs. Animals that go nocturnal to dodge hikers may face higher predation risk from native carnivores, or lose access to food sources available only during daylight. Conversely, some species might learn to exploit human-modified niches, like roadside forage or campground scraps, without obvious short-term penalties. Without multi-year demographic data linking movement records to survival, body condition, and reproductive success, researchers cannot say whether the observed avoidance is a temporary inconvenience or a chronic stressor with population-level consequences.
There is also almost no comparable data from non-park public lands. Forests managed by the Bureau of Land Management, state wildlife areas, and private conservation easements host different mixes of recreation and resource extraction: off-road vehicles, timber harvest, dispersed camping. Whether animals on those lands show the same intensity of avoidance as animals inside national parks remains untested. Extending the research beyond NPS boundaries would clarify whether park-specific management, or simply any sustained human presence, drives the response.
What this means for trails, closures, and park management
For the roughly 312 million recreation visits the National Park Service recorded in 2023, the practical implications are straightforward. Quiet, non-motorized recreation is not ecologically neutral. A person on a trail is not invisible to the elk bedded 200 yards uphill or the black bear foraging in the drainage below. Both are tracking that person’s movement, adjusting their own schedules, and paying an energy cost to do so.
That does not mean closing parks. It does mean that trail design, seasonal closures around sensitive habitats, and visitor education about staying on designated routes are not just courtesy measures. They are wildlife management tools backed by field data. Concentrating foot traffic on fewer corridors, for example, could leave larger blocks of undisturbed habitat available during critical feeding and breeding windows.
As of June 2026, more collar and camera datasets are being collected across NPS units, and researchers are beginning to link behavioral data to demographic outcomes. The core finding, however, is already settled: in the places we set aside for nature, our presence is not a backdrop. It is an ecological force, and the animals have been responding to it all along. We are only now catching up to what they already knew.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
