NEW PERSPECTIVES ON BIRD-WINDOW COLLISION: THE EFFECTS OF SPECIES TRAITS AND LOCAL ABUNDANCE ON COLLISION SUSCEPTIBILITY

Birds provide substantial ecosystem services such as pest control and act as indicators of environmental health. Diverse, widespread, and conspicuous, they have also helped advance our understanding of how ecological communities function. Unfortunately, human activities are now adversely affecting many bird species, driving populations downward. While major threats like habitat destruction have deservedly drawn the attention of conservationists and researchers, other human impacts have been largely overlooked. I focus this paper on one of these commonly neglected problems, window collision. Despite killing as many as one billion birds per year in the U.S., window collision has received little research until very recently and many questions about this issue remain unanswered. The research that does exist gives a complicated and often conflicting impression of window collisions. In particular, there is a poor understanding of which species are more likely to collide. Some studies have suggested year-round, resident birds are more vulnerable. Others have argued just the opposite, that window collisions disproportionately affect transient groups such as neotropical migrants. The purpose of this study was to create a better understanding of collision vulnerability by exploring the issue in the context of taxonomy, behavior, and local abundance. Discovering what drives these collisions could help further justify existing conservation concerns and inform future collision mitigation efforts. I used two approaches to analyze this issue. The first approach, referred to as the Triangle Study, involved tallying collisions at three study sites in the Triangle Region of North Carolina during fall 2015. At each of these three sites, standardized surveys were conducted at six buildings to determine what species were colliding and how often. I then reclassified the data based on taxonomic family and order, feeding guild, feeding location, migration and breeding status, and synanthropic status, the degree to which a species otherwise benefits from human development. Finally, I examined the collision frequencies among these classification groups for an indication of relative vulnerability. For my second analysis, the Duke Case Study, I examined whether indications of collision vulnerability persisted when local abundance was considered. The existing literature has given little attention to potential effects of local abundance when characterizing collision vulnerability. Consequently, impressions of vulnerable species may be biased towards those that are common around study sites. For the Duke Case Study, I looked at collision data collected at six buildings on Duke University’s Main Campus during spring and fall migration of 2015. To be able to compare these collision frequencies to local abundances, I conducted the first ever systematic survey of Duke University’s Main Campus bird community. I then simulated what collision rates might look life if collision were solely a product of local abundance and chance. I plotted the observed collision rates against the 95% confidence intervals of the simulated rates to determine if species were colliding below, within, or above the null range of collision rate. I also extended this analysis to the other taxonomic and functional classifications used in the Triangle Study. The Triangle Study revealed several characteristics that appeared to contribute to collision vulnerability. First, the majority of collision victims belonged to the order Passeriformes and the remaining observations were all near-passerines. At the family level, Parulidae, Turdidae, and Trochilidae accounted for a large proportion of the observed collisions. Ruby-throated Hummingbirds were by far the most common species, followed by Common yellowthroat, Northern Parula, and Red-eyed Vireo with nine collisions each. When considered under the functional classifications, the vast majority (80%) of collisions were insectivores. Among these insectivores, over a third foraged in the lower canopy and another 21% foraged on the ground. Additionally, birds that tended to benefit less overall from human develop appeared to collide more often. Well over half of the birds that collided were locally breeding species and the vast majority of were also migratory. The Duke Case Study provided a similar characterization of which groups were vulnerable, with a few notable exceptions. Only one of the top five colliders from the Triangle Study, the Red-eyed Vireo, recorded exceptionally high rates of collision in the Duke Case Study. However, both studies did indicate the Parulidae family was vulnerable, supporting previous research. The most notable disparity between the Triangle Study and the Duke Case Study involved breeding status. While the Triangle Study provided an impression that local breeders were susceptible to collision, the Duke Case Study demonstrated that these high rates were due to high local abundance and that locally breeding species were actually colliding less often than expected. Regarded independently of the Triangle Study, the Duke Case Study still offered several interesting results. Most significantly, it demonstrated the collision vulnerability is a real phenomenon and abundance and chance alone cannot account for collision frequencies. Additionally, the second analysis helped identify groups that were relatively invulnerable to collision, including the invasive House Finch, omnivores, and the ubiquitous Paridae family. The Duke Case Study also provided a preliminary indication that foraging height may be tied to collision vulnerability This paper has significant implications for how future research on bird collisions should be conducted. The Duke Case Study revealed that local abundance does often mask actual levels of vulnerability among species. Future studies must account for this potential effect when attempting to characterize collisions. Further, both analyses demonstrated that behavior appears to influence susceptibility to collision. More research is needed to better understand how activities such as territoriality and migration may be contributing to collisions