COS 56-9
Testing bird-friendly glass prototypes in an experimental flight tunnel

Tuesday, August 11, 2015: 4:20 PM
348, Baltimore Convention Center
Lucas W. DeGroote, Powdermill Nature Reserve, Carnegie Museum of Natural History, Pittsburgh, PA
Matt Webb, American Bird Conservancy, The Plains, VA
Christine Sheppard, American Bird Conservancy, The Plains, VA
Background/Question/Methods

Because birds can’t tell reflections from trees, plants, and sky, an estimated 599 million individuals die every year after colliding with windows.  Over 40 years of research coupled with grassroots advocacy efforts has resulted in the creation of a LEED certification credit which has greatly increased commercial interest in bird-friendly glass.  We tested glass in an experimental rotating flight tunnel stationed at Powdermill Avian Research Center in western Pennsylvania by artificially representing glass installed on a building using mirrors and a background.  Birds released from one end of the tunnel fly towards either a test pattern or clear (control) pane mounted on the opposite end of the tunnel before colliding with a mist net.  Flights were video recorded and coded as “colliding” with the test pattern (ultraviolet/non-ultraviolet), control pane, wall, ceiling, floor, or center.  We also coded observer confidence (1 to 3), cloud cover (1 to 5), speed (fast/slow), and flight type (direct/indirect).  Utilizing this data we investigated the potential biases and utility of tunnel testing.  

Results/Conclusions

We found a high degree of repeatability (98%) between observers coding “collisions” in the tunnel.  Of the 31 patterns tested with at least 60 usable flights, 22 panes were significantly avoided indicating that birds can see patterned glass in a tunnel setting. Avoided panes reduced collisions by 33% to 77% and probability of pattern detection was consistent among 15 species tested with more than 30 usable flights.  Using Akaike’s Information Criterion we determined that birds flying directly at a glass pane were more likely to avoid the test pane thus indicating that patterns can be detected more than 10 feet away.  In addition we determined that there was little association between flight speed, cloud cover, and pattern detection.  Our results are consistent with recent field experiment results and indicate that flight tunnels can be used to rapidly test numerous bird-friendly glass prototypes with a high degree of repeatability.  The ability to quickly identify effective bird collision reducing glass is crucial to addressing what has recently been identified as the second greatest direct threat to avian populations.