How much do you think that WBCs are due to errors of judgement in flight, i.e. not recognizing a window as a window, and how much to not recognizing their own reflection as themselves, rather than another bird?

The reason I ask this question is that one side of our lab building is attractive as a site for nesting house martins, Delichon urbicum, who place their nests close under the hanging eaves but in almost all cases immediately above a window. Ours is not the only building in the village where they build nests over windows and they also do it on some domestic premises.

This placement of nest was something of a puzzle because we thought the birds would run a high risk of WBCs or could be disturbed in their flight paths by either the reflections or movements of people within.

However, one warm summer all the windows were opened and the birds seemed a bit agitated, possibly because the casement windows looked like obstructions, but another issue proved to be the problem when we found the birds were having difficulty judging their approach and sometimes came in through the open window. So our conclusion from this was that the birds benefit from the windows by being able to see their reflections on the approach and thereby judge their approach path in much the same way as a pilot would use lights and other landing aids to judge the approach to an airstrip.

I note that in your list of birds chimney swifts did not appear to be major casualties, so could it be that in that in the Hirundinidae and Apodidae, nearly all of whom nest in similar circumstances, the birds either instinctively or have learned to use reflections and shadows against walls or windows as flight aids whereas other birds cannot utilize these visual cues, and are possibly more likely to mistake them for competitor individuals of their own species?

waiting for moderation
1 Answer
Accepted answer

"Looking but not 'seeing' the way ahead" (Martin 2011)

During flight, the avian visual system is structurally and functionally primed for maximum resolution and perception in the lateral fields of view (Rogers 2008, Martin 2011). Visual lateralization allows birds in flight to detect conspecifics, predators, and foraging areas, which are features present from any direction around the bird except in the direction of flight (Rogers 2008, Martin 2011). Moreover, adjusting head position (yaw, pitch and roll) and rotation of the eyes enhances lateral vision when important environmental features are detected, but may also cause natural blind spots to project in the frontal space (Martin 2011). Birds may fail to predict windows in buildings emerging above vegetation where, presumably, birds evolved or learned to fly in open and uncluttered air space (Martin 2011). Thus, birds ‘look’ ahead during flight, but can be blind to or fail to ‘see’ windows. Even if obstacles were detected, avoiding a collision would be difficult because the rate of information gain lags far behind flight velocity (Martin 2011).

Below is a list of hypotheses for bird-window collisions in light of bird vision and behavior. Note that these hypotheses are not mutually exclusive and may interact depending upon the social and environmental context.

‘Defective/impaired vision’

Anatomical/physiologic abnormalities in vision system may limit an individual’s ability to perceive glass as a barrier.


Birds may be deceived and fly into windows during smoky conditions, inclement weather, or while in a drunken state after consuming fermented fruit. No study has assessed vulnerability of drunken birds to window strikes.


Collisions occur as territorial individuals engage in aggressive attacks on their own window reflections from short distances, i.e., 1-2 meters.

‘Tunnel Flight’

Flight through small, restricted passageways in heavy vegetated cover and toward lighted areas, i.e., a ‘tunnel’, causes birds to hit windows (Ross 1946, Snyder 1946). Tunnel flights may explain why thrushes, Accipiter hawks, waxwings, and grouse fly into windows and die at high rates (Klem 1989, Hager 2009).

‘Predator/Prey Chase’

Distracted flight where both prey and predator are fixed upon each other, and one or both individuals fail to perceive glass as a barrier, e.g., Accipiter hawks, shrikes (Klem 1989, Hager 2009), and gulls in Toronto (Ogden 1996) hunting birds near windows.

‘Social Chase’

Inter- and intraspecific chases by a territory holder on an intruder or by dominant individual on subordinate. This may occur during any season and explain simultaneous strikes of individuals of the same species in sapsuckers, Cooper’s Hawks, and Ruby-throated Hummingbirds (Hager 2009, S. B. Hager, unpubl. data). Related to this is the social behavior of flocking birds, such as waxwings.

‘Panic Flight’

Window collisions happen during high speed escape flights by flocking birds at feeder stations due to the sudden appearance of a predator or perceived predator (e.g., Blue Jay).

‘Non-social Flight’

Birds flying in non-social contexts fail to perceive reflective glass as a barrier, which results in window collisions (Dunbar 1949, Martin 2011).

waiting for moderation

Thanks for all the background information on bird-window collision hypotheses. Those possible scenarios all make sense in relation to direct head-on collisions and, of course for the species mentioned as examples, buildings are actually an obstruction to normal flight in much the same way as a stationary tree or a moving vehicle would be (although how Accipiter hawks flying through forests in pursuit of prey manage to avoid trees defeats me somewhat). However, martins are unusual in that unlike the other birds they do actually utilize buildings or cliffs for their own purposes and therefore court the risk of collision. Solid objects like walls and cliffs are understandably easier to avoid because you can see them distinctly from some distance but, although windows often appear black against the building, they are less easy to judge. Having just spent 5 minutes watching the martins I noted that they rarely, if ever, approach the nest directly and always fly in at a slight angle, possibly only 2 or 3 degrees off direct but this would allow them to see just enough laterally to fit with the visual lateralization concepts of (Rogers 2008, Martin 2011). It looks as though martins have adapted their behaviour to avoid those issues quite nicely.

waiting for moderation