Britzke dissertation 2003 use of bat detectors

AN ABSTRACT OF A DISSERTATION USE OF ULTRASONIC DETECTORS FOR ACOUSTIC IDENTIFICATION AND STUDY OF BAT ECOLOGY IN THE EASTERN UNITED STATES Eric R. Britzke Doctor of Philosophy in Environmental Sciences Bats in the eastern United States use echolocation to locate prey and navigate in their surroundings. Recent advances in technology have enhanced use of ultrasonic detectors in field studies. Data recorded using ultrasonic detectors have been used to investigate a wide variety of questions involving ecology of bats.

Despite their abundant use, fundamental questions remain unanswered on appropriate uses and limitations of this technology. Using the Anabat II bat detector, acoustic identification of 12 bat species in the eastern United States was investigated using discriminant function analysis. Cross- validation yielded accuracy rates that ranged from 56.5% (Myotis grisescens), with 10 of 12 species having accuracy rates > 70%. To assess the impact of ambient light levels on bat activity, passive recording was conducted with light intensity meters at three fixed stations in Kentucky.

While temperature and time past sunset were significant factors in explaining variation in bat activity, ambient light level was not. Typically, bat habitat use studies assume equal bat activity throughout habitats being sampled. Using a 6-station grid, spatial variation of bat activity within two stands (mature forest and timber harvest area) was examined. Spatial variation in bat activity among stations was twice as high in the mature forest stand as in the timber harvest area, thereby suggesting that a different number of ultrasonic detectors is required to adequately sample bat activity in these habitats.

Acoustic identification cannot be performed on all recorded echolocation calls. Thus, an objective filter was constructed to assess if call sequences were identifiable. Effect of habitat type on proportion of recorded calls surviving the filter was determined. Habitats with greater structural complexity (e., mature forests) had lower proportions of call sequences that were identifiable.

Taken together, these studies assist in determination of appropriate uses of frequency division ultrasonic detectors for the study of bats. USE OF ULTRASONIC DETECTORS FOR ACOUSTIC IDENTIFICATION AND STUDY OF BAT ECOLOGY IN THE EASTERN UNITED STATES ___________________ A Dissertation Presented to the Faculty of the Graduate School Tennessee Technological University By Eric R. Britzke ___________________ In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSPHY Environmental Sciences ___________________ May 2003 CERTIFICATE OF APPROVAL OF DISSERTATION USE OF ULTRASONIC DETECTORS FOR ACOUSTIC IDENTIFICATION AND STUDY OF BAT ECOLOGY IN THE EASTERN UNITED STATES By Eric R. Britzke Graduate Advisory Committee: ________________________________ _________ Chairperson Date ________________________________ __________ Member Date ________________________________ __________ Member Date ________________________________ ___________ Member Date _______________________________ ____________ Member Date Approved for the Faculty: ___________________________ Associate Vice President for Research and Graduate Studies _______________________ ii Date ACKNOWLEDGEMENTS I thank members of my committee, Dr.

Roberts, and Dr. Wells, for assistance provided throughout this lengthy process from study design to completion of this manuscript. I also thank all additional people that assisted in editing associated manuscripts. Financial support for this project was provided by the Arkansas Game and Fish Commission, Bat Conservation International, Great Smoky Mountains National Park, Mammoth Cave National Park, National Forests in North Carolina, Ozark-St.

Francis National Forest, Southern Research Station of the U. Forest Service, Tennessee Wildlife Resources Agency, and the U. Fish and Wildlife Service. Additionally, the Center for the Management, Utilization, and Protection of Water Resources and Department of Biology at Tennessee Technological University provided funding and support for this project.

This project would not have been possible without the assistance of numerous people throughout the eastern United States. While I cannot name all of these volunteers, some people deserve special attention: Tom Biebighauser, Robert Currie, Alan Hicks, Susan Loeb, and John MacGregor. Chris Corben provided extensive assistance for his technical advice on the use of Analook software throughout this study. Finally, special thanks to Tammy Jones, Michael Schirmacher, and Andrew Scott; they managed to persevere despite difficult situations I thrust upon them.

iii TABLE OF CONTENTS Page LIST OF TABLES…………………………………………………………………… vi LIST OF FIGURES…………………………………………………………………. vii INTRODUCTION…………………………………………………………………… 1 Echolocation Properties…………………………………………………. 5 Uses of Ultrasonic Detectors……………………………………………. A Quantitative Method for Acoustic Identification of Bats in the Eastern United States……………………………………………………… 9 Introduction……………………………………………………………… 10 Methods………………………………………………………………….

Effect of Ambient Light Levels on Bat Activity. Spatial Variation in Bat Activity in Two Forest Stands. 36 iv Page PART 4. Effect of Habitat Type on Potential Identification of 42 Bat Echolocation Calls.

46 LITERATURE CITED……………………………………………………………… 52 v LIST OF TABLES Table Page 1.1 Number of recording locations and sample sizes of each bat species represented in the call library used to construct the DFA classification model……………………………………………………….2 Accuracy rates (%) of 12 species of bats for 3 iterations and average of the cross-validation procedure using DFA based on 10 echolocation call parameters.3 Average classification rates (%) from cross-validation testing of the DFA. N = number of test sequences for each bat species. Species codes represent the first two letters of the genus and the specific epithet. Actual species identifications are listed across the top, while predicted species identifications are along the side.

Unknown species identifications resulted from 2 species being identified by an equal number of calls within the sequence……………… 21 vi LIST OF FIGURES Figure Page 1.1 Locations from which bat echolocation calls were recorded with the Anabat II bat detector system…………………………………… 22 2.1 Mean number of echolocation files (±SE) recorded at 3 stations during 3 different moon phases. Number of files of bat activity did not differ among moon phases……….2 Relationships between rank of number of bat call sequences recorded and rank light intensity for 15-minute periods pooled for 12 nights of recording at each site…………………………………….1 Frequency distributions of coefficients of variation in bat activity among 6 stations in mature forest and shelterwood habitat stands……………………………………………………………… 39 3.2 Frequency distributions of similarities in bat activity among 6 stations sampled in mature forest and shelterwood stands……………….3 Relationships between distance between stations and similarities among those stations in the number of files recorded in mature forest and shelterwood stands…………………………………………….1 Relationship between number of bat call sequences that survived cleaning of filter 1 and number of bat call sequences surviving the cleaning of filter 2……………………………………………………….2 Proportion of bat call sequences surviving cleaning of filter 1 by habitat (F = 32. Habitats with the same letters are not statistically different in pairwise comparisons………………………. 51 vii INTRODUCTION Worldwide, many bat species have experienced severe population declines (Fujita and Tuttle 1991; Pierson 1998; Racey 1998; Richards and Hall 1998).

In the eastern United States, two monotypic species and two subspecies of another species are listed as federally endangered by the U. Fish and Wildlife Service (Harvey et al. Listing bats as threatened or endangered has prompted extensive research into ecology of bats (Racey and Entwistle 2003). Until the late 1990’s, most research focused on the use of capture and/or observational techniques.

However, relatively little is known about habitat use of many bats. Recently, ability to study bats has been improved by advances in technology (e., ultrasonic detectors and radio-telemetry). Initial ultrasonic detectors were expensive, susceptible to damage, and logistically difficult to use (Griffin 1958), but early efforts indicated that some species could be acoustically identified (Fenton and Bell 1981; Simmons et al. Recent advances in technology have resulted in widespread use of detectors for study of bat ecology (Betts 1998); however, several aspects of their use need refinement.

1 2 Echolocation Properties A single emission of sound is referred to as a call, and a series of calls is a call sequence (Fenton 1999). The sound produced with the lowest frequency is designated as the fundamental harmonic. As a by-product of sound production, other harmonics are produced at whole number multiples of the frequency of lower harmonics. For example, if sound is produced at 20 kHz, harmonics will also be produced at 40 kHz, 60 kHz, etc.

Further, as frequencies increase, harmonics are produced at increasingly lower amplitudes (energy). Because high frequency sound attenuates faster and is produced at lower amplitudes, higher harmonics travel shorter distances from the source than lower harmonics. Echolocation calls of bats consist of three phases: search, approach, and terminal (Griffin et al. Search phase calls are produced to locate prey, approach phase calls are produced to identify exact locations of prey, and terminal phase calls are produced just prior to capture.

Search phase calls are useful in the study of bat echolocation because they constitute a majority (ca. 90%) of calls produced by bats, exhibit consistency in structure throughout the call sequence, and may possess species-specific characteristics (Betts 1998; Fenton and Bell 1981; O’Farrell et al. Ultrasonic Detectors Orientation capabilities of bats were first studied by Italian scientist, Lazzaro Spallazani in the 1790’s (Griffin 1958). Spallazani observed that bats flew equally well 3 without vision, but he was unable to determine the method they used to orient to their surroundings.

However, he demonstrated that bats deprived of both sight and hearing could not successfully orient to their surroundings. Although ultrasound was suspected, high frequency sounds produced by bats were not detected until the late 1930’s (Griffin 1958). Before analysis of echolocation calls can be conducted, incoming signals must be converted to frequencies that are within the range of human hearing. Based on differences in methods to convert this incoming sound, three classes of ultrasonic detectors have been developed: heterodyne, frequency division, and time expansion.

Heterodyne detectors are narrowband instruments; i., they only detect ~ 10 kHz frequency range at any one time. A frequency is manually selected, and the ultrasonic detector converts all sounds within 5 kHz of the selected frequency into audible clicks. Heterodyne detectors can only be operated through active recording (researcher present), do not permit more advanced analysis of detected signals, and fail to detect bats outside the narrow band of frequencies being sampled (Fenton 1988). Consequently, heterodyne detectors are useful for a narrow range of study designs.

For example, heterodyne detectors can be tuned to a frequency of a target species to determine presence or activity of this species. Despite the limitations of this detector type, they are frequently used to study bat activity (Ahlen and Baagoe 1999; Crampton and Barclay 1998; Grindal and Brigham 1999; Limpens and Kapteyn 1991; Walsh and Harris 1996). Frequency division detectors are broadband (10-200 kHz) instruments that divide incoming frequencies by preset values. The Anabat II bat detector system (Titley 4 Electronics; www.au) is a widely used instrument of this type (Betts 1998).

Frequency division detectors can be used under active or passive (researcher absent) modes. Echolocation calls are recorded to tape recorders (Hayes and Hounihan 1994) or directly to laptop computers (O’Farrell 1998). Time expansion systems are also broadband ultrasonic detectors in which incoming signals are stored in a digital buffer. When the buffer is filled or when specified by the user, signals are downloaded at 1/10 of their normal speed to a tape recorder.

For example, incoming sounds that are 1.7 seconds in length will require 17 seconds to download, a period in which additional echolocation calls cannot be detected. Frequency division and time expansion systems retain different information from incoming sounds. The Anabat system only uses the harmonic with the most energy for recording (i., usually the fundamental harmonic), and information on harmonic structure and amplitude is lost. Time expansion systems retain complete incoming signals, providing for more thorough analysis.

This increased information retention comes at a large cost with file sizes being much larger for the time expansion system (> 1 MB each) than for the Anabat system (1-15 KB) each). Thus, time expansion systems require more powerful computers to record and analyze echolocation calls. The Anabat system measures lower maximum and higher minimum frequency values than time expansion systems (Fenton et al. However, these 1-2 kHz differences are less than variation from other sources (Brigham et al.

1989; Murray et al. 2001; Thomas et al. Additionally, simultaneous sampling indicated that Anabat systems detect fewer echolocation calls than time expansion detectors (i., microphones 5 on time expansion systems were more sensitive)(Fenton 2000; Fenton et al.