It is estimated that several million to as many as one billion individuals of North America’s native bird species collide with buildings each year with the majority dying on impact. These collisions are caused mainly by the fact that most birds do not understand what glass is: they perceive reflections of trees, the sky and the general landscape on glass surfaces to be real scenes that they can fly into, or they fly into non-reflective glass because it may appear invisible to them. Ambient lighting at night and lights on inside of buildings at night exacerbate these problems. Collisions occur most often during migratory periods when millions of migrating birds pass through cities while traveling between their breeding and wintering grounds. Hundreds of North America’s bird species are affected including many that are uncommon and declining, and collisions are now believed to be one of the most important causes of decline in many of these species. Bird collisions have been monitored in Toronto, New York City, Chicago and other large cities for several decades and it is estimated that one million collisions occur each year in the city of Toronto alone.
In order to obtain baseline data about the nature and extent of the problem in Philadelphia Audubon Pennsylvania, the Philadelphia Zoo and the Academy of Natural Sciences of Philadelphia conducted a collaborative study of bird collisions in Philadelphia from 2008 through 2011. During the study a 3.5 square block section of center city Philadelphia was monitored for dead and injured birds 7 days a week throughout the spring and autumn migration periods. Detailed information was recorded about each bird found. From this data we estimate that about 1000 bird collisions probably occur within the study area each year.
It has always been assumed that bird collisions occur randomly and that no species is more likely to fly into a particular structure than another. But the data collected by this study (which contains more detailed information on each specimen than has ever been gathered by any other collision monitoring project) suggests that the distribution of specimens may not have been random. Spatial analysis would be needed to determine whether the distributions are random or not. If non-random distributions are found this would demonstrate for the first time that factors other than glass and lights may also be involved in causing bird collisions. Specifically, we suspect that birds may be interpreting buildings as areas of potential habitat and certain kinds of buildings or groupings of buildings may appear more attractive to some species than to others.
Throughout the 3-year study detailed information was recorded about each dead or injured bird found including the date, time, location (street address, side of nearest building and distance from nearest building), species name, specimen condition and other details. Dead or injured individuals of approximately 50 different species were found during the study.
One map layer would be produced for each of the species found showing where all dead and injured individuals of that species were discovered within the study area during the 3-year study. Spatial analysis would then be used to evaluate each distribution pattern for randomness. If non-random distributions are found one of the variables most likely to be associated with this is landscape structure which could be described through features like building height, building closeness, etc. Exploring which variables might be associated with non-random distributions is an area a student working on this project would be encouraged to pursue. Audubon Pennsylvania currently has limited aptitude to do this kind of work because we have only one (part-time) GIS technician who works almost exclusively on large-scale projects that are being coordinated with other parts of the national organization.
The maps will indicate whether the specimens of each species were randomly distributed within the study area or not. We have not yet determined the exact method that will be used to evaluate each distribution pattern for randomness. However if non-random distributions are found and these can be associated with specific factors like structural features of the landscape such a finding could enable recommendations to be created for improved design of buildings and cityscapes to reduce collision rates throughout the world. The hypothesis that bird collisions may not occur randomly is one that has never been investigated before. If non-random distributions are found such a finding would also merit publication so that other researchers could help us think creatively about this finding.
The Summer 2020 session has finished. Sign up for notifications about future opportunities.
