Bird-building collisions are a major source of wild bird mortality, with hundreds of millions of fatalities each year in the United States and Canada alone. Here, we use two decades of daily citizen-science monitoring to characterize day-to-day variation in building collisions and determine the factors that predict the highest risk times in two North American cities. We use these analyses to evaluate three potential causes of increased collision risk: heightened migration traffic during benign weather, increased navigational and flight errors during inclement weather, and errors in response to reflective glass. The seasonal phenology of collisions was consistent across sites and years, with daily collision rates approximately two-fold higher in autumn as compared to spring. During autumn migration, day-to-day variation in collision rates is best explained by weather conditions around dawn with four major drivers: no precipitation, high atmospheric pressure, cold temperatures, and clear skies. In spring, collision risk was best predicted by either overnight or morning weather, driven by a lack of precipitation and warm temperatures. Based on these results, we hypothesize that collisions are influenced by two main weather-driven mechanisms. First, benign weather causes an increase in migration traffic in both spring and autumn, creating greater opportunity for collisions to occur. Second, cold clear conditions cause an additional increase in collision risk for autumnal migrants. These conditions may be particularly hazardous in autumn because of the high abundance of naive and diurnal migrants at that time of year. Our analysis also establishes that a relatively small proportion of days (15%) are responsible for 50% of the total collision mortality within a season, highlighting the importance of targeting mitigation strategies to the most hazardous times.

Bird-building collisions are a major source of wild bird mortality, with hundreds of millions of fatalities each year in the United States and Canada alone. Here, we use two decades of daily citizen science monitoring to characterize day-to-day variation in building collisions and determine the factors that predict the highest risk times in two North American cities. We use these analyses to evaluate three potential causes of increased collision risk: heightened migration traffic during benign weather, increased navigational and flight errors during inclement weather, and increased errors in response to highly directional sunlight that enhances reflected images. The seasonal phenology of collisions was consistent across sites and years, with daily collision rates approximately twofold higher in autumn as compared to spring. During both migration seasons, collision risk was best predicted by the weather conditions at dawn. In spring, peak collision risk occurs on days with warm temperatures, south winds, and a lack of precipitation at dawn. In autumn, peak collision occurs on days with cool temperatures, north winds, high atmospheric pressure, a lack of precipitation, and clear conditions with high visibility. Based on these results, we hypothesize that collisions are influenced by two main weather-driven mechanisms. First, benign weather at dawn and winds that are favorable for migration cause an increase in migration traffic in both spring and autumn, creating greater opportunity for collisions to occur.Second, for autumnal migrants, cold clear conditions may cause an additional increase in collision risk. We propose that these conditions may be particularly hazardous in autumn because of the high abundance of naïve and diurnal migrants at that time of year. Our analysis also establishes that a relatively small proportion of days (15%) are responsible for 50% of the total collision mortality within a season, highlighting the importance of targeting mitigation strategies to the most hazardous times.