|
Abstract:
|
Monitoring populations at the multi-national scale is essential for making informed conservation and management decisions for North American bat species. The North American Bat Monitoring Program (NABat) relies heavily on call files obtained from acoustic detectors. Bat call files are processed, run through species classification software, and sometimes through a manual vetting procedure before receiving a species identification. Uncertainty can be introduced in each of these steps, and classifications can suffer from imperfect detection or erroneous detection. These errors in detection have been shown to bias estimates of occupancy when ignored. Existing occupancy models can account for imperfect detection, but accounting for erroneous detection requires additional information about true occupancy status, or about the rate at which erroneous detections are made. In this talk, we provide a brief overview of existing methods for accounting for erroneous detections, discuss challenges in obtaining ground-truth data at the site-level and observation-level in the context of acoustic bat calls, and propose extensions to methodology that exploit information that is currently ignored.
|
