A recent study published in Science shows that in Yellowstone National Park in the United States, when wolves hunt large prey, the reason why crows are able to arrive at the carcass in a very short time is not to follow the wolves in the air for a long time, but to rely on the memory and navigation ability of "high-yield hunting areas" to fly to areas where wolves are more likely to hunt in advance. This international cooperation project, led by the Institute of Wildlife Ecology of the University of Veterinary Medicine in Vienna, the Max Planck Institute of Animal Behavior in Germany, and other institutions, uses long-term precise positioning and tracking to subvert people's long-standing intuitive understanding that "crows can eat as long as they follow the wolves closely."
For decades, observers have noticed that whenever wolves successfully hunt caribou, bison, or deer in Yellowstone, crows almost always appear quickly and gather nearby to grab the scraps of meat before the wolves have even finished eating. This almost "synchronous appearance" behavior was once simply explained as crows continue to track the movement of wolves over long distances, so they can locate the latest hunting scene as soon as possible. However, the research team found through systematic data analysis that this statement is not true. What crows display is a more complex cognitive strategy.
The research team conducted tracking work in Yellowstone Park for more than two years. During this period, about a quarter of the wolves in the park were wearing positioning collars, allowing scientists to track their movements and hunting locations in real time. At the same time, the researchers installed micro-GPS "backpacks" on 69 crows, a large-scale attempt with a rare sample size in this type of study. The process of capturing crows is quite difficult because the birds are extremely sensitive to their environment and highly wary of unfamiliar objects. Researchers had to disguise their capture devices as camp garbage or even fast food remnants to lower their vigilance.
During the winter—the season when crows and wolves interact most frequently—the researchers recorded the GPS locations of crows at 30-minute intervals and the wolf locations at hourly intervals, noting the times and locations of wolf kills of large prey such as caribou, bison, and deer. Unexpectedly, during two and a half years of monitoring, the research team found only one case that could be clearly defined as "a crow continued to follow a wolf for more than 1 km or more than 1 hour." This result is obviously inconsistent with the previous hypothesis that "crows rely on wolves to forage", and also made the research team puzzled as to why the crows could still "rush to the scene" so quickly.
More in-depth trajectory analysis gave a new explanation: the crows frequently traveled not to the current location of the wolf, but to specific areas where "wolves have historically been more likely to successfully hunt." Data show that wolf hunting events tend to cluster at the bottom of valleys with flatter terrain, and these areas have been proven to be "high-yield areas" with a higher probability of finding prey carcasses. The crows visited these areas significantly more frequently and visited less frequently sites where historically few hunting events had taken place, showing that they "learned and remembered" long-term differences in food abundance in different landscape areas.
Studies have recorded that some crows can fly up to 155 kilometers in a day, and their flight paths are often quite straight, pointing to specific areas where many hunting incidents have occurred before, rather than searching aimlessly along the routes of real-time movement of wolves. This shows that crows do not "track the present" myopically, but build a "food hotspot map" based on past experience, and then rely on excellent spatial memory and navigation capabilities to make search decisions on a broad landscape scale. As Matthias Loretto, the first author of the paper, points out, crows can fly for hours on end, flying directly to areas where "there may be carcasses" without having to follow the wolves all day to make a profit.
The team emphasizes that this does not mean that crows do not use wolf behavioral cues at short distances. When the ground is close enough, crows are likely to still pinpoint the location of the kill by observing the wolf's behavior or even listening to short-range signals such as wolf howls. However, on a larger scale, memory-driven navigation based on long-term experience dominates its foraging decisions, rather than simply "walking right next to the wolf." This combination strategy from large-scale memory selection to small-scale precise positioning based on instant cues reflects a highly flexible and complex foraging behavior pattern.
Dan Stahler, a long-term researcher on Yellowstone wolves and a senior biologist, pointed out that in field observations, people often see crows hovering directly over moving wolves, or following the pace of wolves at close range during hunting. This image has a profound impact on the impression of "crows clinging to wolves for food." However, no previous research has truly systematically quantified these activities from a "scavenger perspective", nor has crows been the real subject of behavioral analysis. This study puts crows at the center of the research and uses synchronized bird and predator trajectory data to overturn the simple rule of "follow closely" and reveal the more sophisticated cognitive mechanisms behind it.
Previous research has shown that crows are able to remember stable food sources such as landfills, and their memory and problem-solving skills are among the best among birds. This study further extends this understanding to natural scenarios where "the probability distribution is not fixed": even if the specific time and location of a single wolf killing event are unpredictable, crows can still extract from long-term accumulated experience "which areas are more likely to have food in the future" and plan large-scale movements accordingly. According to co-author John M. Marzluff, a professor at the University of Washington, this flexibility shows that crows do not rely on a specific wolf pack, but rely on their keen senses and past memories to choose between multiple foraging opportunities in a vast space.
The research team believes that this work not only changes people's traditional understanding of the crow-wolf interaction, but also proposes revisions to the entire theoretical framework of how scavengers find food. In the past, people often underestimated the ability of such animals to integrate spatial and temporal information in complex environments, simplifying their behavior to passively following predators or relying on short-range searches based on immediate scent. The "predictive foraging" strategy displayed by crows in Yellowstone Park reminds scientists that it is necessary to re-evaluate the cognitive upper limit of other scavenging species and even a wider range of wildlife.
The study titled "Ravens anticipate wolf kill sites across broad scales" was published in the journal Science on March 12, 2026, and was jointly completed by scientists from multiple institutions in Austria, Germany, and the United States. They pointed out that in a rapidly changing global ecological environment, understanding how animals cope with uncertainty in resource distribution through learning and memory is of great significance for predicting species adaptability and formulating conservation strategies. As researchers say, humans may have long underestimated the intelligence of this seemingly "scavenging" animal, and the tracks drawn by crows over Yellowstone are a powerful correction to this prejudice.