Surveying bat communities: a comparison between mist nets and the Anabat II bat detector system

Transcription

1 Acta Chiropterologica, 1(1): , 1999 PL ISSN Museum and Institute of Zoology PAS Surveying bat communities: a comparison between mist nets and the Anabat II bat detector system KEVIN L. MURRAY, ERIC R. BRITZKE, BRAD M. HADLEY, and LYNN W. ROBBINS I Depantnent of Biology, Southwest Missouri State University, Springfield, MO 65804, USA 'Corresponding author: LynnRobbins@mail.smsu.edu A survey of bat communities was conducted in Missouri during the summer of 1998 using both mist nets and Anabat. Mist nets provide valuable information about population demographics and species diversity within bat communities. However, they can yield biased samples of bat community activity and may cause undue stress to captured bats. The use of ultrasonic bat detectors like Anabat can ameliorate some of these problems. Bat detectors are relatively easy to set up, require no direct contact with bats and can sample a wider variety of habitats. To test the relative merits of these two methods, we sampled a variety of habitats including ponds, streams, and flyways using Anabat and mist nets in a paired design. This allowed for the simultaneous sampling of bat community activity necessary for direct comparison. Echolocation calls recorded by the Anabat were identified using a discriminate function analysis model based upon a library of known call sequences. Overall, values for species richness were significantly higher for Anabat IT than for mist nets. Species richness was consistently higher for Anabat for all habitats and seven individual species and one species group was detected more frequently with Anabat than with mist nets. Key words: Chiroptera, Anabal, mist nets, method comparison, acoustic identification, echolocation INrRODUCTlON Two commonly used methods of surveying bat communities are mist nets and ultrasonic bat detectors (Kuenzi and Morrison, 1998). Mist nets have several advantages such as the collection of demographic data (Erkert, 1982), but also have several drawbacks. Mist nets can be difficult and timeconsuming to erect in the field, and can yield biased samples of bat species assemblages (Kunz and Kurta, 1988; Findley, 1993). Certain habitats, like open fields, large bodies of water, or high in the canopy, cannot be sampled effectively using mist nets (Kunz and Brock, 1975). Also, some bat species may vary in their ability to avoid mist nets due to differences in activity patterns or ecomorphology (Kunz and Kurta, 1988). Perhaps the most serious drawback to the use of mist nets is the stress and potential harm to bats caused by capture and handling (Erkert, 1982; Kunzand Kurta, 1988; Findley, 1993). Ultrasonic detectors provide a means to survey bat communities while avoiding -several of these problems. Compared to mist nets, detectors are easily set up and can sample a wider variety of habitat types (O'Farrell, 1997). In addition, unlike mist nets, which require constant monitoring (Kunz and Kurta, 1988; Findley, 1993), ultrasonic detectors can be deployed and left

2 106 K. L. Murray, E. R. Britzke, B. M. Hadley, and L. W. Robbins to record automatically (Krusic and Neefus, 1996). This makes it possible to sample more sites with fewer people and in less time. Most importantly, a bat detector can sample bat species assemblages without any direct contact or disturbance to bats (Erkert, 1982; Kuenzi and Morrison, 1998). This can be a tremendous advantage, especially when dealing with endangered species. Ultrasonic detectors also have several potential biases (Findley, 1993; Barclay, 1999; O'Farrelletal., 1999b). Certain habitat types and conditions limit the use of bat detectors. Large bodies of water, sites with dense vegetation, and areas with high levels of insect noise are all difficult to sample effectively using detectors. Ultrasonic detectors set at ground level can fail to record bat species that fly high above the canopy (Fenton and Griffin, 1997; Barclay, 1999). Potentially, the most serious bias of ultrasonic detectors is the underrepresentation of species with low intensity echolocation calls (Thomas and LaVal, 1988; Findley, 1993; O'Farrell and Gannon, 1999). Though problems with ultrasonic detectors do exist, detectors are a comparatively easy, non-invasive method of surveying bat communities with fewer potential biases (Kuenzi and Morrison, 1998; O'Farrell et al., 1999b). Numerous studies have used either mist nets (Kunz, 1973; Gardner et al., 1989) or ultrasonic detectors (Hickey and Neilson, 1995; Krusic and Neefus, 1996; Lance et ai., 1996; O'Farrell e/ al., 1999b) to sample bat species assemblages. However, few have compared these two methods (Kunz and Brock, 1975; Rautenbach et al., 1996; Kuenzi and Morrison, 1998; O'Farrell and Gannon, 1999a). Moreover, previous comparison studies each used qualitative analysis for species identification and some failed to sample simultaneously with both detectors and mist nets. The goal of this study was to evaluate the relative effectiveness of mist nets and the Anabat II ultrasonic bat detector in surveying various habitats in Missouri. Sites were sampled simultaneously with both methods, and recorded echolocation calls were identified using quantitative analysis. MATERIALS AND METHODS Field Techniques Standard mist netting techniques (Gardner el ai., 1989) were used throughout the survey. Net height varied from 2.5 m to 9 m depending on habitat type and canopy cover, and net length varied from 6 m to 18 m depending on habitat type and width of flyway. A variety of habitat types were netted. including ponds. creeks. intermittent streams. water-filled road ruts. and closed canopy forest corridors (flyways). Sites were selected in areas where a high level of bat activity was expected and an effort was made to sample a wide variety of habilat types in the forests of Missouri. In general. nets were erected around sunset and taken down at midnight. Species were identified according to Schwartz and Schwartz (1981). Anabat systems were paired with nets at each site. Each system consisted of an Anabat Il bat detector. a Zero Crossings Analysis Interface Module (ZCAIM). and laptop computer (equipped with Anabat 5 and Analook software). Anabat Il bat detectors were positioned in the field to sample the same general area and habitat as the corresponding mist net. Detectors were never pointed toward nets and were always set up greater than 6 m away from nets. This enhanced the likelihood of recording natural echolocation calls from free-flying bats. rather than calls emitted by a bat avoiding or being captured by a mist net. Anabat systems were set to passively monitor sites using monitor mode. a feature of the Anabat 5 software and sampled the same time period as the corresponding mist net at each site. Sensitivity on the Anabat Il was adjusted from 3-7 depending on varying levels of insect noise at each site. The angle at which the Anabat Il was set varied from based on type of habitat sampled. The area sampled by Anabat units varied depending on the sensitivity setting and the positioning angle of the detector. As with mist nets. Anabat units were deployed to maximize the area sampled. Dala Analysis Recorded echolocation sequences were quantilatively identified (0 species and used to determine species richness (defined as number of species) at each site. Species identification was accomplished using a discriminate function analysis (DFA) model according

3 to methods found in Britzke el al. (1999). A library of known echolocation call sequences for each species was used to construct the model. All known call sequences were recorded from light-tagged bats producing search phase calls. Each call sequence used in construction of the model consisted of ten or more individual calls (pulses). The accuracy of the DFA model ranged from 71.4% (Myolis sodalis) to 100% (M. grisescens, M. seplentrionalis) depending on species. In all cases, documentation of the presence of a particular bat species was very conservative. Documentation at each site depended upon the identification of multiple recorded search phase call sequences for each species (i.e. several recorded Anabat files identified as a particular species were required before that species was determined to be present in an area). This decreased the possibility of misidentifying species using the DFA model. Due to insufficient known reference calls, Lasi"nls cinere"s was not included in the DFA model. In addition, when known L cinere"s calls were entered into the model, they were misidentified as Eplesic"s fusc"s. Therefore, these two species were grouped together for both mist nets and Anabat in order to maintain a direct comparison between the two surveying techniques. For each site, species richness values were deter mined for both mist nets and Anabat. Values for all twenty sites were then compared using a two-tailed Wilcoxon signed rank test with an IX of Species - Mist Net 5 c::=j Anabat '" = Methods for surveying bat communities 107 presence for both methods was examined on a site by site basis. Sites where only Anabat accounted for a species, sites where only mist nets accounted for a species, and sites where both accounted for a species were tallied and compared. RESULTS Twenty sites were sampled with both mist nets and Anabat, for a total of 70 hours. One hundred and five bats were captured with mist nets and a total of 1,964 files were recorded with Anabat during the study. Nine species of bats from five genera (Eptesieus fuseus, Lasiurus borealis, L. cillereus, Myotis septelltriollalis, M. lllcifugus, M. sodalis, M. grisescells, Nyctieeills humeralis, and Pipistrellus sllbflavus) were captured using mist nets. Anabat documented seven individual species (L. borealis, M. septelltriollalis, M. lllcifllgus, M. sodalis, M. grisescells, N. humeralis, and P. subflavus) and one species group (E. juseusll. cillereus). For individual sites within the study area, Anabat II consistently detected more species than mist nets (Fig. I). Overall, species li a S = u.- 4 n=20 n = 7.- u '" 3 Co r:n tl!i o Pond Flyway Stream Road Ruts Overall Habitat Type FIG. I. Average species richness determined by Anabat II and mist nets in various habitat types

4 108 K. L. Murray, E. R. Britzke, B. M. Hadley, and L. W. Robbins richness was significantly higher (Wilcoxon test, t, = 109.5, n = 20, P < 0.01) for Anabat than for mist nets. In addition, mean species richness values for Anabat were higher in all habitats examined (Fig. 1). At each site bat species were documented as present by either mist nets exclusively, Anabat exclusively, or by both methods (Table 1). For all twenty sites combined, various species were documented as present a total of 83 times. Anabat documented species as present 71 out of 83 times (85.5%) and mist nets documented species as present 42 out of 83 times (50.6%). Anabat was more effective at sampling all seven individual species and one species group (Table 1). Species commonly captured by mist nets (L. borealis, M. septentrionalis, and P. subflavus) as well as species rarely captured by mist nets (M. lucifugus and N. humeralis) were documented more often using Anabat. Anabat also detected endangered species more regularly, with M. grisescens being detected at six sites with Anabat, but only captured at three sites, and M. sodalis being captured at only one site, but detected at three. DISCUSSION There have been few studies comparing the effectiveness of ultrasonic detectors to mist nets. Fenton et al. (1987) used bat detectors to determine the distribution of Ellderma maclliatum in the western U.S. and southern Canada, and then compared their results to previous studies that used various capture techniques. Both methods revealed similar species distributions. Kunz and Brock (1975) found that the activity patterns of M. IllcifugllS determined by mist nets and ultrasonic detectors were equivalent. In a more complex system containing 22 insectivorous bat species, Rautenbach et al. (1996) noted no correlation between general bat activity as determined by an ultrasonic detector and bat captures as determined by mist nets. The fact that bat activity data were correlated with insect abundance while capture data were not indicates that ultrasonic detectors provided a better representation of bat activity than mist nets. Other studies involving the qualitative identification of several bat species have also compared detectors and mist nets. O'Farrell and Gannon (1999) reported that the number of species detected by Anabat was significantly higher than the number captured by mist nets and harp traps. In their study, Anabat identified 14 of 20 species sampled more often than capture methods. Kuenzi and Morrison (1998) found that average species richness values were similar for both methods (mist nets = 2.7, detector = 3.2), even TABLE I. Comparison of species specific capture success between mist nets and Anabat in 20 sampled sites Species Mist net only Anabatonly Both Lasiurus borealis 3 Pipistrellus subflavus 3 Eptesicus Juscus/Lasiurus cine reus 2 Nycticeius humeralis Myotis septentrionalis 2 M. grisescens M. lucijugus 0 M. sodalis 0 Total I

5 Methods for surveying bat communities 109 though the echolocation calls of three species of Myotis could not be differentiated by the qualitative methods used in the study. Species with similar echolocation calls are often grouped together (Hickey and Neilson, 1995; Krusic and Neefus, 1996; Hayes, 1997; Kuenzi and Morrison, 1998) leaving the true species composition in an area unresolved. It is possible, had these species been separately identified rather than grouped as a single species, that species richness values for the ultrasonic detector may have been considerably higher. Our study directly compared mist nets and detectors by sampling the same area and time period with both methods. Kuenzi and Morrison (1998) sampled on different nights, using detectors the first night and nets the second. Species composition and activity at a given site may vary considerably on a nightly basis (Hayes, 1997; O'Farrell and Gannon, 1999). Thus, the observed differences in species richness between mist nets and detectors in Kuenzi and Morrison (1998) may have simply been a result of this variation. O'Farrell and Gannon (1999) sampled intermittently with Anabat using active recording (following bats with the detector and saving call sequences manually). They stated that this technique was used to visually observe recorded call sequences and eliminate sequences of poor quality. However, this meant that portions of the night were not sampled acoustically because removal of bats from nets prevented constant monitoring of detector units. We chose to use passive recording methods (remote monitoring) rather than active recording for several reasons. Using our methods it was possible to remove low quality sequences in the analysis process, after they had been recorded. Also, visual cues (size of bat, flight style) that can be obtained by a researcher using active recording methods were not required for quantitative species identification. Most importantly, passive recording is a repeatable and less biased acoustic survey method that does not require constant monitoring. This makes it possible for one individual to set up acoustic monitoring systems that can record continuously at several sites in a single night. It is evident that the use of ultrasonic detectors is an invaluable tool in the study of echolocating bats. Researchers have used Anabat IT to determine species composition (Krusic and Neefus, 1996; Lance et al., 1996; O'Farrell et al., 1999a), examine activity and foraging behavior (Bradshaw 1996; Hayes and Adam, 1996; Hayes, 1997), and investigate possible management strategies (Crampton and Barclay, 1996; Hayes and Adam, 1996; Parker et ai., 1996). However, there are some concerns with current methods of acoustic monitoring, namely, that certain species of bats may be missed, misidentified, or underrepresented. These species include those that fly high above the canopy out of range of ground-based ultrasonic detectors (Fenton and Griffin, 1997). Also, some bat species use low intensity echolocation calls (e.g., Corynorhinus spp. and some Myotis spp.) and are not detected as readily as bats that use higher intensity echolocation calls (Fenton, 1982; Findley, 1993; O'Farrell and Gannon, 1999). O'Farrell and Gannon (1999) discovered that three species (Corynorhinus townsendii, Myotis auriculus, and M. evotis) with low intensity calls were captured by mist nets more often than they were detected by Anabat, and one species (M. thysanodes) that was detected as well by Anabat. However, they found no significant difference in the detection of low intensity species between Anabat and various capture techniques. Our survey dealt with only one species that uses a low intensity echolocation call, M. septentrionalis (Faure et al., 1993; Waters and Jones, 1995). Our data indicate that the echolocation calls of M. septentrionalis are detectable using the Anabat system and that

6 110 K. L. Murray, E. R. Britzke, B. M. Hadley, and L. w. Robbins this species is sampled more effectively by Anabat than mist nets in our study area. This result is somewhat surprising since M. seplentrionalis is considered difficult to detect acoustically (Hickey and Neilson, 1995). However, these findings can be explained by the fact that this species was relatively common in the study area (present' at 15 of 20 sites). Acoustic documentation of the species was verified by capture in mist nets at eight separate sites. In addition, average capture height of M. septenlrionalis in the study area was 1.5 m. Thus, individuals were flying close enough to ground-based detector units to be detected regularly. There is also evidence that M. seplenlrionalis lowers the intensity of its echolocation call when gleaning (Miller and Treat, 1993), meaning that in non-gleaning flight they may use a higher intensity call, which is easier to detect (Waters and Jones, 1995). It is difficult to say how our results can be applied to M. seplentrionalis in other geographic areas or to low intensity bats in general. There is no doubt that bats with higher intensity echolocation calls will be detected at greater distances and relatively more frequently than species with low intensity calls (Fenton, 1982; Findley, 1993; O'Farrell and Gannon, 1999). However, it is still unclear how this bias affects differential detection between mist nets and ultrasonic detectors. Another factor that hinders acoustic surveys of bat communities is the difficulty in obtaining known reference calls from species that are rarely captured. When dealing with these species, identification may not always be possible (Vaughan el ai., 1997). In our study, we had difficulty recording known calls from L. cinereus because only 4-5 individuals of this species were captured in two years of netting. With so few individuals we were unable to obtain sufficient known call sequences for this species. Thus, L. cinereus was not included in the DFA model, and, as a result, we were unable to document its presence quantitatively using Anabat. An additional problem is that the echolocation calls of species that are not included in the DFA model will be misidentified as species that are included (Vaughan et ai., 1997). In our case, echolocation calls of L. cinereus were misidentified as E.fuscus. The acoustic identification of these two species is relatively easy and has been accomplished both qualitatively (Fenton and Bell, 1981; Hickey and Neilson, 1995; O'Farrell et ai., 1999a) and quantitatively (Obrist, 1995). However, because L. cinereus was not included in the DFA model, it was not possible to reliably assign a species designation to the echolocation calls of these two species with the quantitative methods used in this study. As a result, calls from E. fuscus and L. cinereus were placed into a species group. Our results provide evidence of the utility of Anabat as a method of surveying bat communities and demonstrate its effectiveness when compared to mist nets. However, there are certainly limitations to its use. Like mist nets, Anabat has several potential biases such as a limited ability to detect bats that use low intensity echolocation calls. It is also clear that the identification of echolocation calls of species rarely captured may be problematic. While our results show that Anabat detects several bat species more effectively than mist nets, they also demonstrate that the combination of both survey techniques provides the most effective means of determining bat species composition in an area. ACKNOWLEDGMENTS Thank you \0 Sybil Amelon, Derek Bossi, Sarah Hartje, Scott Robenson, 10hn Ashc.aft, and all graduate and undergraduate students for their invaluable assistance in the field. Many thanks \0 SMSU support services for providing quality equipment for use in the study. Thank you to two anonymous reviewers for their helpful comments on an earlier version of this manuscript. Funding for the project was provided by the United States Forest Service, the Missouri Depart-

7 f Methods for surveying bat communities III I ment of Conservation, the Missouri National Guard and the SMSU Biology Department. ' LITERATURE CITED BARCLAY, R. M. R Bats are not birds: a cautionary note on using echolocation calls to identify. bats. Journal of Manunalogy, 80: BRADSHAW, P. A The physical nature of vertical forest habitat and its importance in shaping bat species assemblages. pp , in Bats and Forest Symposium (R. M. R. BARCLAY and R. M. BRIGHAM, eds.). British Columbia Ministry of Forests, Victoria, British Columbia, 292 pp. BRITZKE, E. R., D. W. BOSSI, B. M. HADLEY, and L. W. ROBBINS The acoustic identification of bals in Missouri. Bat Research News, 39: 158. CRAMPTON, L. H., and R. M. R. BARCLAY Habitat selection by bats in fragmenled and unfragmented aspen mixedwood stands of different ages. Pp , in Bats and Forest Symposium (R. M. R. BARCLAY and R. M. BRIGHAM, eds.). British Columbia Ministry of Forests, Vicloria, British Columbia, 292 pp. ERKERT, H. G Ecological aspecls of bat activity rhythms. pp , in Ecology of bats (T. H. KuNZ, ed.). Plenum Press, New York, 425 pp. FAURE, P. A., J. H. FULLARD, and J. W. DAWSON The gleaning attacks of the northern longeared bat, Myotis septentrionalis, are relatively inaudible to moths. Journal of Experimental Biology, 178: FENTON, M. B Echolocation, insect hearing, and feeding ecology of insectivorous bats. pp , in Ecology of bats (T. H. KUNZ, ed.). Plenum Press, New York, 425 pp. FENTON, M. B., and G. P. BELL Recognition of species of insectivorous bats by their echolocalion calls. Journal of Manunalogy, 62: FENTON, M. B., and D. R. GRIFFIN High altitude pursuit of insects by echolocating bats. Journal of Marnma1ogy, 78: FENTON, M. B., D. C. TENNANT, and J. WYSZECKI Using echolocation calls to measure the distribution of bats: the case of Eudenna macutatum. Journal of Marnma1ogy, 68: FINDLEY, J. S Bats: a community perspective. Cambridge University Press, Cambridge, UK, 167 pp. GARDNER, J. E., J. D. GARNER, and 1. E. HOFMANN A portable mist netting system for capturing bats with emphasis on Myotis sodalis (Indiana bat). Bat Research News, 30: HAyES, J. P Temporal variation in activity of bats and the design of echolocation-monitoring studies. Journal of Marnma1ogy, 78: HAYES, J. P., and M. D. ADAM The influence of logging riparian areas on habitat utilization by bats in western Oregon. Pp , in Bats and Forest Symposium (R. M. R. BARelA Y and R. M. BRIGHAM, eds.). British Columbia Ministry of Forests, Victoria. British Columbia, 292 pp. HICKEY, M. B. C., and A. L. NEILSON Relative activity and occurrence of bats in southwestern Ontario as determined by monitoring with bat detectors. Canadian Field-Naturalist, 109: KRuslC, R. A., and C. D. NEEFUS Habitat associations of bat species in the white mountain national forest. Pp , in Bats and Forest Symposium (R. M. R. BARCLAY and R. M. BRIGHAM, eds.). British Columbia Ministry of Forests, Victoria, British Columbia, 292 pp. KUENZI, A. J., and M. L. MORRISON Detection of bats by mist-nets and ultrasonic sensors. Wildlife Society BuUetin, 26: KUNZ, T. H Resource utilization: temporal and spatial components of bat activity in central Iowa. Journal of Manunalogy, 54: KUNZ, T. H., and C. E. BROCK A comparison of mist nets and ultrasonic detectors for monitoring flight activity of bats. Journal of Marnma1ogy, 56: KUNZ, T. H., and A. KURTA Capture methods and holding devices. Pp. 1-29, in Ecological and behavioral methods for the study of bats (T. H. KUNZ, ed.). Smithsonian Institution Press, Washington, D.C., 533 pp. LANCE, R. F., B. BOLLICH, C. L. CALLAHAN, and P. L. LEBERG Surveying forest-bat communities with Anabat detectors. pp , in Bats and Forest Symposium (R. M. R. BARCLAY and R. M. BRIGHAM, eds.). British Columbia Ministry of Forests, Victoria, British Columbia, 292 pp. Mn.LER, L. A. and A. E. TREAT Field recordings of echolocation and social signals from the gleaning bat Myoris septentrionalis. Bioacoustics, 5: O'FARRELL, M. J Use of echolocation calls for the identification of free-flying bats. Transactions of the Western Section of the Wildlife Society, 33: O'FARRELL, M. J., and W. L. GANNON A comparison of acoustic versus capture techniques for the inventory of bats. Journal of Marnma1ogy, 80: O'FARRELL, M. J., B. W. MiLLER, and W. L. GANNON. 1999a. Qualitative identification of free-flying bats using the Anabat detector. Journal of Marnma1ogy, 80:

8 Jl2 K. L. Murray, E. R. Britzke, B. M. Hadley, and L. W. Robbins O'FARRELL, M. J., C. CORBEN, W. L. GANNON, and B. W. MILLER. I 999b. Confronting the dogma: a reply. Journal of Mammalogy, 80: OBRIST, M. K Flexible bat echolocation: the influence of individual, habitat, and conspecifics on sonar signal design. Behavioral Ecology and Sociobiology, 36: PARKER, D. I., J. A. COOK, and S. W. LEWIS Effects of timber harvest on bat activity in southeastern Alaska's temperate rainforests. pp , in Bats and Forest Symposium (R. M. R. BARCLAY and R. M. BRIGHAM, eds.). British Columbia Ministry of Forests, Victoria, British Columbia, 292 pp. RAUTENBACH, I. L., M. B. FENTON, and M. J. WHIT ING Bats in riverine forests and woodlands: a latitudinal transect in southern Africa. Canadian Journal of Zoology, 74: SCHWARTZ, C. W., and E. R. SCHWARTZ The wild mammals of Missouri. University of Missouri Press and Missouri Department of Conservation, Columbia and London, 356 pp. THOMAS, D. W., and R. K. LA VAL Survey and census methods. Pp. 779, in Ecological and behavioral methods for the study of bats (T. H. KUNZ, ed.). Smithsonian Institution Press, Washington, D.C., 533 pp. VAUGHAN, N., G. JONES, and S. HARRIS Identification of British bat species by multivariate analysis of echolocation call parameters. Bioacoustics, 7: WATERS, D. A., and G. JONES Echolocation call structure and intensity in five species of insectivorous bats. Journal of Experimental Biology, 198: Received 21 February 1999, accepled 21 April 1999