Communities measure bird-safe building progress by establishing a baseline collision rate, implementing mitigation, and then comparing post-mitigation collision counts using standardized survey protocols. Key metrics include carcass counts, collision evidence (feathers, smudges), and scavenger-adjusted mortality estimates. Reliable community monitoring requires consistent survey frequency, documented search areas, and correction for carcass removal by scavengers or humans.
Without consistent measurement, it is impossible to determine whether a bird-safe retrofit actually reduced collisions. Research on bird-building collisions emphasizes that collision rates are influenced by multiple interacting factors, including glass area, nearby vegetation, lighting, and migration intensity. For example, a study of a large urban building in Chicago found that halving the lighted window area decreased collision counts by 11-fold in spring and 6-fold in fall, with an estimated 60% reduction in mortality when lighted window area was minimized. This demonstrates that specific, measurable changes in building features can produce substantial, quantifiable reductions in bird deaths.
Effective community monitoring programs, such as those documented in the U.S. South, combine digital data logging with physical evidence collection. Volunteers typically track three types of data:
One of the most significant sources of error in collision monitoring is the removal of carcasses by scavengers (crows, raccoons, etc.) or human maintenance crews. A study of glass bus shelters in Stillwater, Oklahoma, addressed this by conducting experimental scavenger removal trials. Researchers left known carcasses at shelters to monitor their disappearance over time, using statistical models to estimate carcass persistence. By applying these persistence estimates to their raw counts, they derived a minimum adjusted mortality estimate. Communities can adopt similar methods by conducting small-scale scavenger trials or using published persistence rates for their region to adjust their raw counts.
Citizen science is essential for scaling collision monitoring to a community or city level, but data quality depends on rigorous training and standardization. Volunteers in the U.S. South, for instance, use mobile applications to log observations and collect specimens for university collections, creating a dual dataset of digital records and physical evidence. This approach supports advocacy by providing both quantitative trends and tangible examples of the problem. However, to ensure data reliability, programs should:
To rigorously measure progress, communities should design their monitoring as a Before-After-Control-Impact (BACI) study. This involves:
This design helps distinguish true mitigation effects from natural variation in bird activity. The table below illustrates how data points are organized to calculate the impact:
| Site Type | Before Mitigation | After Mitigation | Change (After - Before) |
|---|---|---|---|
| Impact (Treated) | Baseline Rate | Post-Mitigation Rate | Δ Impact |
| Control (Untreated) | Baseline Rate | Current Rate | Δ Control |
The net effect of the mitigation is estimated by comparing Δ Impact to Δ Control, isolating the reduction attributable to the intervention from background environmental variation.
Communities can begin measuring progress by implementing the following checklist for each monitored structure:
Next Step: Contact your local Audubon chapter or municipal planning office to request a bird-safe building audit or to join an existing monitoring network.
Community monitoring data is subject to several limitations. Scavenger removal rates can vary by season, location, and species, making persistence estimates approximate. Observer bias can occur if surveyors are not trained consistently or if search areas are not strictly defined. Additionally, collision rates can fluctuate due to weather, migration intensity, and local bird population changes, which is why control sites are essential. Communities should report their data with these uncertainties in mind, using phrases like "minimum estimated mortality" rather than absolute counts.
