Drivers of fatal bird collisions in an urban center

Millions of nocturnally migrating birds die each year from collisions with built structures, especially brightly illuminated buildings and communication towers. Reducing this source of mortality requires knowledge of important behavioral, meteorological, and anthropogenic factors, yet we lack an understanding of the interacting roles of migration, artificial lighting, and weather conditions in causing fatal bird collisions. Using two decades of collision surveys and concurrent weather and migration measures, we model numbers of collisions occurring at a large urban building in Chicago. We find that the magnitude of nocturnal bird migration, building light output, and wind conditions are the most important predictors of fatal collisions. The greatest mortality occurred when the building was brightly lit during large nocturnal migration events and when winds concentrated birds along the Chicago lakeshore. We estimate that halving lighted window area decreases collision counts by 11× in spring and 6× in fall. Bird mortality could be reduced by ∼60% at this site by decreasing lighted window area to minimum levels historically recorded. Our study provides strong support for a relationship between nocturnal migration magnitude and urban bird mortality, mediated by light pollution and local atmospheric conditions. Although our research focuses on a single site, our findings have global implications for reducing or eliminating a critically important cause of bird mortality.

North America has lost nearly one-third of its birdlife in the last half-century, with migratory species experiencing particularly acute declines (1). Fatal collisions with built structures represent a major source of direct, human-caused bird mortality across North America, second only to predation by domestic cats (2). Estimates indicate that between 365 million and 988 million birds die annually in collisions with buildings in the United States, with another 16 million to 42 million annual deaths in Canada (2, 3). Birds may collide with glass windows because they reflect the surrounding environment or allow birds to perceive a seemingly open pathway to the interior of the building (4). For the billions of birds that migrate at night, outdoor lighting (e.g., streetlights and floodlights) and interior lighting from buildings may be disorienting and draw birds into built-up areas, at high risk to collide with infrastructure (5–8). Light pollution not only alters nocturnal migratory behavior on a large scale (5, 7), but is also an acute conservation concern. Nocturnal collisions with well-lit communication towers alone are estimated to kill appreciable percentages of the populations of sensitive species (9).Avian collisions with lighted structures have been documented in the scientific literature as early as the 19th century (10–12). In recent decades, this link between collisions and light pollution has been the subject of detailed investigation (8, 13–16). Observers of bird–building collisions and tower kills have long remarked on the apparent influence of meteorological factors such as cloud ceiling, fog, frontal passage, and abrupt changes in conditions, all of which have been associated with large mortality events (10, 13, 17–24). Steady-burning lights may be particularly hazardous (25). Due to high building density and intensity of artificial lighting, cities are of particular concern. Reports of mass collisions at lighted buildings in urban areas are frequent in both the popular and scientific press (13, 19–21, 26).Understanding, predicting, and preventing collision mortality are areas of active scientific inquiry and priorities for policymakers (1, 13). Collisions occur more frequently during migration seasons and impact numerous species of migratory birds (29), and recent work suggests that nocturnal migratory movements can be useful for predicting bird–window collisions (30). Lights-out programs, which encourage the public to extinguish outdoor lighting to protect migratory birds, are receiving increasing attention (13). The act of extinguishing lighting allows birds to immediately return to normal, safe behavior (7) and reduces mortality at lighted buildings (13). Presently, advisories are generally issued for a given time period (e.g., peak migration periods) or on specific nights when weather conditions are favorable for large migratory movements e.g., using migration forecasting, (31, 32)].Here, we integrate meteorological, migration-intensity, and window-radiance data to understand how these factors interact to cause bird collisions. We use a 21-y dataset of fatal collisions recorded at a single large building (McCormick Place Lakeside Center) in Chicago, IL (Fig. 1), to understand the behavioral, environmental, and anthropogenic drivers of these mortality events. Chicago poses the greatest potential risk from light pollution to migrating birds of all cities in the United States (33), and over 40,000 dead birds have been recovered from McCormick Place alone since 1978 (Figs. 2 and ​and3).3). Since 2000, we have recorded the number of birds and the lighting status of each window bay during dawn collision monitoring. Nocturnal lighting at McCormick Place correlates positively with bird collisions in many songbird species (34), but this association has not been quantified in the context of other important factors, including migration intensity and weather conditions. We estimate the effect of window lighting on collision counts and assess how the intensity of nocturnal bird migration mediates this relationship. We also test whether wind and weather conditions may magnify these associations. Finally, we investigate the spatiotemporal scales at which weather and migration data best explain collision mortality, identifying the times of night and areas of airspace associated with these events.Open in a separate windowFig. 1.Location of McCormick Place along the Chicago lakefront. The Lakeside Center building monitored in this study is highlighted in red in a three-dimensional rendering.Open in a separate windowFig. 2.Summary of collisions recorded at McCormick Place and regional bird migration between 2000 and 2020. (Upper) Individual years are drawn in different colors. Dates are given for mortality events totaling more than 50 birds. Pie charts show the family (fam.) composition of collected birds, with families representing less than 5% of total collisions merged into a single “other” category. (Lower) Summed annual migration passage at the KLOT radar in estimated number of individual birds (years colored). (Lower, Inset) Summed seasonal passage totals in estimated number of birds crossing a 75-km transect, with each point representing a year. Estimates are based on methods from ref. 35.Open in a separate windowFig. 3.Recorded collisions by year and window lighting. (A) Collisions recorded at McCormick Place between 1982 and 2020 for spring (light gray) and fall (dark gray) seasons. Horizontal lines with numeric labels show average seasonal collision totals before and after the window-lighting regime changed from fully lighted to partially lighted in 1999. The year 1997 is not shown because construction limited access to the site during that year. (B) Mean recorded daily collisions by window-lighting status from 2000 to 2020.