20th Annual Student Research Conference
Interdisciplinary
Quantitative Identification of Northeast Missouri Bats using Acoustic Data Alone
Joshua B. Kelly
Dr. Jason E. Miller and Dr. Scott Burt, Faculty Mentors
Over the years, conservation biologists have used various methods to identify bats to species using their echolocation calls. The Anabat bat detector and Analook software made it possible to visualize and quantitatively describe echolocation calls. Researchers assert that such quantitative descriptions can be used to identify the species of bat recorded with the instrument. Using linear discriminant function analysis (DFA) and classification trees (CT), classification rates in excess of ninety percent have been reported. Two years of acoustic data collected from Northeast Missouri bats suggests that the rates reported in the literature may be overestimates of true classification rates. Classification analysis like that reported in the literature was performed on the data. CT analysis uniformly yielded higher classification rates than DFA, but both methods gave poor results when V-fold validation was used. We explore reasons for this.
Keywords: statistics, mathematics, bats, echolocation, endangered species
Topic(s):Mathematical Biology
Presentation Type: Oral Paper
Session: 39-4
Location: VH 1432
Time: 2:15 pm