By pairing 4,581 animal GPS tracks with neighborhood level cellphone data, researchers show that human presence and landscape modification drive wildlife through separate channels, and that one size fits all conservation prescriptions miss half the picture.
For decades, wildlife managers have folded two questions into one: how much has the land been altered, and how much are people actually here? The answers looked identical in most datasets, so the policy advice sounded the same. Reduce human footprint, restore habitat, give animals space. A new study published in Science suggests that convergence was a measurement artifact, not a biological truth. Wildlife respond to pavement and to people as separate inputs, and they often push in opposite directions.
The research, led by ecologists at UC Santa Barbara's Bren School, the Smithsonian's National Zoo and Conservation Biology Institute, and Yale, drew on weekly GPS records from 4,581 individual mammals and birds across 37 species in the continental United States, comparing the same calendar weeks in 2019 and 2020, the pandemic "anthropause," when human movement dropped sharply. The methodological leap was data the team says had not been paired before: fine-scale animal GPS tracks layered against anonymized, neighborhood-level cellphone geolocation that separately tracked how many people were in a given place at a given time. The paper appeared in Science in 2026; the underlying journal text was not directly retrievable in the session that sourced this article, so the figures below trace to the institutional release and its republication.
The headline result is that 57% of species studied were affected by the combination of human presence and landscape modification, not by either one alone. Sixty-seven percent of mammal species and 68% of bird species showed shifts in occupied area or environmental niche size tied to human activity. The direction of the response split: 67% of mammal species and 41% of bird species reduced the areas they used when both human presence and habitat modification rose. That is the part that breaks the "less human pressure, more wildlife" rule of thumb. What matters is which kind of human pressure, in which species, in which kind of landscape.
The effects were strongest in less-developed settings, including national parks. Animals in wilder landscapes appear more sensitive to the moment-to-moment presence of people than animals in heavily urban ones, who have presumably already absorbed a steady human load. Co-lead author Ruth Y. Oliver, an assistant professor at the UCSB Bren School, framed the takeaway in the press materials: you cannot understand the full impact of people on wildlife without information on both landscape change and human presence.
The species-by-species response map is where the new framework earns its keep. Gray wolves expanded their ranges in response to human presence. White-tailed deer expanded their environmental niches as landscapes became more modified, then contracted those niches when human presence rose. Sandhill cranes did the reverse. A prescription written from a single habitat variable would push the wrong way for at least one of those species. The implication, drawn directly from the findings, is that effective conservation and wildlife management have to be species-specific and landscape-specific, not a single template applied across a biome.
The dataset itself is part of the COVID-19 Bio-Logging Initiative, a roughly 600-partner collaboration that has accumulated around a billion location records from about 13,000 animals, and whose earlier work documented mammal behavioral shifts and marine traffic changes during lockdowns. The new analysis sits inside that broader effort to use the pandemic as a natural experiment. Oliver's group has now moved on to the next question: whether the behavioral responses they documented raise or lower wildlife mortality. That is announced future research, not a finding, and it is the obvious watch item for anyone managing lands where these species live.
A few scope limits are worth holding onto. The data are U.S.-only, so the framework needs testing in other biomes and regulatory regimes before it travels into global policy. The sample skews toward larger, GPS-trackable mammals and birds, which means smaller herptiles, insects, and less-mobile species are largely absent from this picture. Cellphone geolocation is a useful but imperfect proxy for human presence. It misses who is outdoors, on foot, or off-network, and it is unevenly available across rural and urban America. The "first study to pair animal GPS with cellphone mobility data" claim comes from the press materials; it should be read as a methodological claim about this particular pairing, not a guarantee that no one has ever touched either dataset.
What changes for a wildlife manager reading this is the unit of analysis. The old question was how to reduce human footprint in a given habitat. The new question, with a second axis now empirically populated, is which species in this habitat is responding to which axis, and at what magnitude. That is a more demanding brief, but it is also one that finally matches the biology on the ground.