Urban birds often develop remarkable behavioral adaptations to their complex and challenging environment. Moving vehicles are an important feature of that environment, and birds learn not just to avoid the danger they represent, but also to exploit them in various ways. Crows (Corvus spp.) drop walnuts, small vertebrates, shelled mollusks and other prey/food items onto busy roads to have them killed and/or crushed by cars (Maple, 1974; Grobecker and Pietsch, 1978; Nihei, 1995; Nihei and Higuchi, 2001; Marzluff and Angell, 2007; Kim Anderson, pers. comm.), with various populations having invented/learned this technique at different times and some not at all (Cristol et al., 1997; Nihei and Higuchi, 2001). Carrion-eating birds routinely monitor or patrol busy roads to immediately snatch roadkill (Santos et al., 2011; Planillo et al., 2015; Harju et al., 2018). Passerines glean dead insects from cars (Jokimäki and Kaisanlahti-Jokimäki, 2023) and sometimes even nest in moving cars, trains and boats (Turner, 2015). There are observations of small birds using moving cars as mobile shelters from pursuing raptors (Daan Drukker, pers. comm.; Hadija Mustafa, pers. comm.).

Accipiter hawks are stealthy hunters that routinely and masterfully use various objects as cover to get closer to their prey before the final charge (Roth and Lima, 2003) and combine long waits in ambush with lightning-fast chases (Thiollay, 1994). To hunt this way, they must be able to plan ahead, to know and understand the prey’s behavior, particularly its movement patterns, and to be highly observant – in short, they need remarkable cognitive abilities. Their intelligence and flexibility make them highly successful hunters in both natural and human-modified environments: moving into cities can actually improve their hunting success (Malan and Robinson, 1999; Estes and Mannan, 2003; Rutz, 2006), so many Accipiter species are now nesting and/or wintering in cities around the world (Thiollay, 1994). Urban hawks develop novel ways of hunting not observed in their “rural” conspecifics. Rutz (2006), who studied Eurasian goshawks (A. gentlis) in Hamburg (Germany), listed three such techniques: “(i) Peregrine-like perched hunts from exposed vantage points like roof-tops and TV aerials; (ii) ‘backyard jumping’ and low quartering flights in residential areas where buildings and other objects are used for cover during surprise attacks; and (iii) foraging after sunset under artificial light conditions”. Other researchers reported urban Eurasian sparrowhawks (A. nisus) chasing prey into narrow streets with no escape routes (Dietrich, 1982) or into large glass windows (Theo Squires, pers. comm.); the latter tactic has also been reported for sharp-shinned hawks (A. striatus) (Klem, 1981). In the city of Kharkiv (Ukraine), Eurasian sparrowhawks have long been known to use moving cars and streetcars as cover to sneak up on their prey undetected (Krivitsky and Reutsky, 2003), and a Eurasian goshawk has recently been observed doing so as well (Dmitry Strelkov, pers. comm.).

Cooper’s hawk (A. cooperi) began to utilize cities as breeding and (more commonly) wintering habitat in the 1970s and quickly became the most successful colonizer of cities among Accipiter hawks (see review by Rosenfield et al., 2018). This medium-sized, hypercarnivorous raptor is an exception to the general trend of small, generalist raptors being more successful in cities (Cooper et al., 2022; Headland et al., 2023). However, the only novel hunting technique adapted specifically for urban environments reported for this species to date is hunting around artificial lights (Boal, 1997).

In 2021–2022, I repeatedly observed Cooper’s hawk(s) using cars backed up at a particular intersection as cover for approaching the prey undetected, and using the sound signals for visually impaired pedestrians (hereafter “sound signals”) as a clue that the red light phase of the streetlight will be longer and more cars will line up. This appears to be the most advanced case of raptors adapting to use traffic patterns reported to date.

In 12 hours of observation, 6 attempted attacks were observed. From the side of Side Street where I parked the car, I could see the hawk approaching the part of the street where the prey flock was located, but not the attack itself, so I do not know how many attacks were successful. Once the hawk was seen flying away back across Side Street with a house sparrow in its talons, and once it was observed eating a mourning dove on the ground near the site of attack. After each attack, the bird flock scattered and never returned on the same morning; more than one attack attempt per day has never been observed.

It took me the first two observations to determine the entire course of attacks; it was the same during the remaining four attacks. The hawk would first fly into the tree in front of house #11 ( Figure 1 ) and perch on a low branch near the trunk for up to one minute while the queue of cars was getting longer. The cars obstructed the hawk’s view of the prey and vice versa. As soon as the queue reached house #8, the hawk would take off, fly ~65 m (less than 1 m above ground) along the pedestrian sidewalk on the southern side of Side Street, make a 90° right turn under the tree in front of house #1, and cross Side Street between cars toward the prey flock feeding in front of house #2 ( Figure 1 ).

While observing the fourth attack, I noticed that the hawk appeared in the tree in front of house #11 when the sound signals at the streetlight were on. I realized that the sound accurately predicted that the car queue would be longer than usual. In two subsequently observed attacks, the hawk also appeared in the tree when the sound signals were on, but before the long car queue formed. That was not a random occurrence (arithmetic probability P= 0.0375³ ≈ 0.000053, as the signal was on for only ~3.75% of the time). In all three cases, the signal was no longer active by the time the hawk attacked. It is unknown if the first three observed attacks also started when the signals were on.

The hawk was never seen on weekends, even though I drove or walked through the intersection many times. In the winter of 2022–2023 I twice accidentally observed an adult Cooper’s hawk using the same method at the same location. It is very likely that it was the same individual as the immature hawk the year before. These two observations were not included in the analysis because total observation time was not recorded and the identity of the bird was not certain. In the summer of 2023, the sound signals at the streetlight stopped working for unknown reason, and the residents of house #2 stopped leaving leftovers. No hawks were ever observed at the intersection after that.

This is a case study, and as such it would be difficult or impossible to replicate. However, it does not particularly stand out among existing observations of complex hunting behavior of raptors: various species can be taught to selectively hunt certain prey species other than their own preferred ones (Bildstein, 2006); others hunt in cooperating and coordinating packs (Bednarz, 1988), use tools to procure food and nesting material (Stoyanova et al., 2010; Santiago, 2025), and use burning sticks to start wildfires, facilitating prey detection (Bonta et al., 2017).

The observations presented here could be explained as a result of simple Pavlovian training, but that would be an obvious oversimplification. To hunt as observed, the hawk had to develop a precise mental map of the street (as the target flock was invisible to the hawk until the last stage of attack), notice the connection between the sound signals and the length of the car queue, and figure out that only longer queues provided cover for the entire approach. The ability to plot the most efficient approach to a food item obscured by obstacles has been documented in corvids (i.e., Bugnyar et al., 2004) and some mammals (i.e., Dumas et al., 2006), but not in birds of prey. However, tests have shown that birds of prey can be as good at inventing and learning as parrots and corvids (Smith and Colbert-White, 2023).

The behavior described here is an impressive feat of intelligence, going a long way to explain the species’ ability to successfully colonize such unusual and dangerous environment as urban landscape. This intelligence was likely pre-existing rather than evolved in the novel environment. Red foxes (Vulpes vulpes) have been found to become bolder but not more intelligent in cities (Morton et al., 2023), even though they’ve been living there longer than Cooper’s hawks. Urban chimango caracaras (Milvago chimango) show faster learning and lower neophobia than rural ones, but their learning process is similar (Biondi et al., 2024). Moreover, most Cooper’s hawks wintering in cities in our area are visitors from non-urban areas (Boyle, 2011), so it is likely that the immature hawk I observed was a very recent immigrant to urban habitat. One can only imagine the level of knowledge and understanding of the environment possessed by hawks living in more natural landscapes.