Subcontract no. 97-162A

Between Idaho State University and Davidson College

DEVELOPMENT AND TESTING OF AN AUTOMATED

RECORDING SYSTEM FOR MONITORING ANURANS AND BIRDS

ON THE SAVANNAH RIVER SITE

Idaho State University, Pocatello, ID 83209

Davidson College, Davidson, NC 28036

INTRODUCTION

Recent worldwide declines in numbers of species make monitoring animal populations necessary as the first step in preserving the species that remain (Heyer et al., 1994). In the southeastern United States, declines in bird populations, especially neotropical migrants, are of special concern (Robbins et al., 1989; Terborgh, 1989). Likewise, apparent worldwide declines in many amphibian species make monitoring their populations especially important (Corn and Fogleman, 1984; Beiswenger, 1986; McAllister and Leonard, 1990; National Research Council Workshop on Declining Amphibians, 1990; Wake, 1991). The potential occurrence of sensitive species on the Savannah River Site in South Carolina requires the consideration of the impact of human activities on those species when making management decisions. Potentially occurring sensitive bird species include the Red-cockaded Woodpecker (Picoides albolarvatus), Swainson's Warbler (Limnothlypis swainsonii), Bachman's Sparrow (Aimophila aestivalis), and Painted Bunting (Passerina ciris). Sensitive anuran species that occur, or potentially occur, include the Carolina Gopher Frog (Rana capito), River Frog (Rana hecksheri), and the Pine Barrens Tree Frog (Hyla andersoni). Manual calling surveys have been a valuable technique traditionally utilized by researchers when monitoring both birds and anurans. However, temporal, interspecific, and environmentally induced variation in calling may lead to failure to detect some species (Peterson and Dorcas, 1992).

Automated recording systems can overcome many of the problems associated with manual calling surveys (Peterson and Dorcas, 1994). Advantages of automated recording systems include: (1) the ability to sample for extended periods of time, thus increasing the probability of detecting a given species; (2) decreased disturbance of calling anurans, thus decreasing the probability of missing easily disturbed species; (3) the ability to sample multiple sites simultaneously; (4) consistent sampling without the need for experienced field personnel; and (5) a permanent sampling record. If combined with environmental measurements, data from automated recording systems can also be used to optimize when, where, and under what conditions to conduct traditional calling surveys (Peterson and Dorcas, 1992).

We have developed an inexpensive automated recording system (ARS) which consists of a recycling timer, a cassette tape recorder, an omnidirectional microphone, a voice clock to audibly time stamp each sampling interval, and a weather resistant case. Several researchers from various universities, industries, and state and federal agencies have used these systems to monitor bird and anuran populations. Preliminary results of these studies have revealed that ARS can be a valuable technique in monitoring anurans. For example, using ARS, we were able to detect the presence of the gopher frog (Rana capito) at new locations on the Savannah River Site.

Although ARS has been shown to be an effective technique for monitoring anuran populations, detailed evaluation of ARS must be conducted to maximize its potential as a monitoring technique. This evaluation should include determination of the optimal sampling rates for ARS, determination of the spatial coverage of ARS, comparisons of results obtained from ARS with results from traditional techniques, and development and testing of the quantitative capabilities of ARS.

The purpose of this study was to conduct studies to properly evaluate the effectiveness of ARS as a monitoring technique. These tests were conducted in conjunction with studies already underway using ARS to monitor anuran populations at wetland locations on the Savannah River Site and elsewhere. Our evaluation of ARS includes:

1. Tests to determine optimal sampling rates and sampling times for detecting the presence of particular anuran species.

2. Tests to determine the spatial coverage of ARS.

3. Comparisons of species richness estimates determined by ARS with estimates using traditional nonauditory techniques (e.g.,drift fences) and traditional auditory techniques (i.e., point-count measurements).

4. Tests to determine the quantitative capabilities of ARS for selected species of anurans.

RESULTS

Optimal Sampling Times and Sampling Rates

Determination of the optimal times to conduct calling surveys will help to maximize the success of manual calling surveys, the standard technique of the North American Amphibian Monitoring Program. To determine the optimal sampling times for a variety of summer breeding species of anurans at an isolated wetland, we sampled at Flamingo Bay on the Savannah River Site (Bridges and Dorcas, in press). We chose Flamingo Bay because of the high diversity of anurans and because the species that occur there are typical in that they occur at many locations in the southeast. We sampled using ARS for 12 seconds every 30 minutes for 26 consecutive days during the months of June and July, 1997. Tapes were transcribed in the laboratory and the intensity of calling for each species was coded as follows: 0 = no vocalization recorded, 1 = one male heard vocalizing, 2 = multiple males vocalizing, but not overlapping into a full chorus and 3 = many males overlapping into a full chorus. Mean calling intensity was calculated for each species for each 30 minute sampling period.

Nine species of anurans were recorded calling at Flamingo Bay during the study including the southern cricket frog (Acris gryllus), eastern narrow-mouthed toad (Gastrophryne carolinensis), green treefrog (Hyla cinerea), pine woods treefrog (H. femoralis), barking treefrog (H. gratiosa), gray tree frog (H. chrysocelis), bullfrog (Rana catesbeiana), bronze frog (R. clamitans), and southern leopard frog (R. sphenocephala). Our sampling regime (12 sec every 30 min) resulted in a total of 1248 sampling intervals for each species.

Analysis revealed considerable interspecific variation in calling patterns over the sampling period. Some species were recorded on every day of the study, while others were not. For example, H. cinerea was recorded on all 26 days but a closely related species, H. gratiosa, was recorded on only18 days. Other species that were recorded every day of the entire study include, R. clamitans and A. gryllus.

Considerable interspecific variation was found in the daily temporal calling patterns of the species recorded in this study. Gastrorphyne carolinensis usually began calling at about 1500 hrs and gradually increased throughout the afternoon until dusk (Fig. 1A). Afternoon thunderstorms occurred frequently during the study and G. carolinensis frequently would often call most intensely just after these storms. The sharp drop in calling at about 2000 hrs may not actually be a decrease in calling activity, but more likely is a result if being "drowned out" by loudly calling hylids (e.g., H. gratiosa). Acris gryllus called during the daytime and at relatively high levels throughout every evening (Fig. 1B). Daytime calling by A. gryllus was typically lowest in the morning but gradually increased throughout the day and into the evening.

The four species of Hyla we recorded exhibited a relatively consistent daily calling pattern during which they would begin calling at sundown and then stop at about midnight. Hyla cinerea (Fig. 2A) exhibited a definite calling peak at between 2130 and 2230 hours. Hyla gratiosa (Fig. 2B) exhibited a sharp peak beginning shortly after sundown (2130 hrs). Hyla femoralis and H. chrysoscelis (Fig. 2C and 2D) showed peaks in calling shortly before midnight as well but did not occur in choruses as large as the other two Hyla.

All three species of Rana recorded during study showed peaks in calling activity between midnight and dawn. Rana sphenocephala typically did not start calling until after midnight and reached a peak of calling activity between 0230 and 0500 hrs (Fig. 3A). Although R. catesbeiana was recorded throughout the entire day, peak calling acitivity occurred between 0200 and 0600 hrs (Fig. 3B). Likewise, R. clamitans was recorded calling throughout the day, but peak calling occurred after midnight, between 0100 and 0530 hrs (Fig. 3C).

On average, 7.3 species of anurans were recorded per day. Our results indicate that if you went out on only one night during our study period you would miss, on average, 1.65 of the 9 species (18%) per night, if you sampled all night long. However, if you sampled only between 2100 and 2300 hrs (the time recommended by NAAMP) you would miss an average of 3.35 of the 9 (37%) species we recorded.

Spatial Coverage

To determine the spatial coverage of ARS, we repeatedly played the call of a Spring Peeper (Pseudacris crucifer) at varying distances (10 meter intervals up to 100 meters) from the system microphone. We use a clear call played at a volume that simulated the calling intensity of the test species (determined by a decibel meter). We conducted tests using four different microphones (microphones described below). To determine the effect of vegetative cover on call transmission, we conducted tests in a completely open habitat, an open canopy with shrubs, a closed canopy with open understory, and a closed canopy with thick understory. Tests were conducted at times when wind and other ambient noise was minimal. Tapes from the system were transcribed in the laboratory to determine how distance from the microphone affects call detection.

Results are provided in Figure 4 and indicate that the cardioid powered microphone (AudioTechnica - AT815A) provided the best distance coverage in all habitats, followed by the omnidirectional powered microphone (Radio Shack RS-9382), the cardiod nonpowered microphone (Electovoice - EV635A), and the omnidirectional nonpowered microphone (Radio Shack RS-3892). It was noteworthy that there was little difference in the performance of the two non-powered microphones, even thought the Electrovoice costs about 9 times as much ($89) as the Radio Shack model ($10). The effect of habitat on call transmission was similar to what we expected. Calls carried best in the open canopy and open understory, good for a short distance in the closed canopy with an open understory and open canopy with a closed understory, and poorly in the closed canopy with the closed understory.

Comparisons with Traditional Techniques

Comparison with Bird Point-Count - To compare results obtained by ARS to point-count surveys of bird communities, we set up ARS at an isolated wetland in Aiken, SC from 28-30 April 1997 and enlisted the expertise of Mark Komorowski, an expert in bird call identification, to assist with bird call verification. Mr. Komorowski visited the site for 2 hours on 28 April 1997 while the ARS was recording and manually recorded all birds he both observed and heard. Later, he listened to the ARS tape recording in the laboratory and identified the bird calls whenever possible. Results are listed in Table 1 and indicate that ARS can be a very successful technique for monitoring birds. ARS detected several nocturnal species that were not detected manually and likely detected several species that did not call during the manual survey due to disturbance by Mr. Komorowski.

Table 1. Comparison of ARS and a bird-point count at an isolated wetland in Aiken, SC. The number missed is the number of species not detected that were detected using the other technique.

Comparison to drift-fences - To compare the results of ARS with traditional nonauditory techniques, we compared the results of ARS set up at Rainbow Bay on the Savannah River Site for several weeks in April to the results obtained from a drift fence daily run during the same time. Comparisons indicate that during the sampling period, ARS was much more successful in monitoring all hylids and ranids, primarily because species of both of these families can either jump or climb over drift fences, but did not detect either spadefoot toads or southern toads, because they were not calling at that time. These results are an excellent demonstration of how ARS can complement existing techniques.

Quantitative Capabilities

We found it very difficult to quantify anuran population sizes using ARS. We did, however, find that ARS was useful in detecting relative variation in population sizes between two Carolina Bays, Flamingo Bay and Ellenton Bay (Mohr and Dorcas, 1999). We measured anuran calling activity using automated recording systems from 24 June to 18 July 1998 (Peterson and Dorcas, 1992 and 1994). Equipment consisted of a cardioid microphone (Model AT815A Audio Technica, Singapore) attached to a stereo analog tape recorder (Model TCD-5PROII, Sony Electronics, Park Ridge, NJ) controlled by a recycling timer (Model RS-1A12, SSAC, Baldwinsville, NY). This equipment, along with a voice clock (Model RS-63-915, Radio Shack, Ft. Worth, TX) to audibly time stamp the sampling period, was placed in a weather resistant toolbox. The microphone was shielded with plastic from 2-liter soda bottles and affixed to nearby trees approximately 2 m off the ground at the water's edge and facing the center of each bay. High bias, type II cassette tapes (C-100CDT2A, Sony Electronics, Park Ridge, NJ) were used for recording. The timers were set to record 12 seconds every 30 minutes, 24 hours per day for the entirety of the 25-day study.

Every three days the tapes were changed, returned to the lab, played on a stereo cassette deck (Technics Model RS-TR232; Matsushita Industrial Co, Ltd., Singapore), and calling analyzed. Anuran calls were identified to species, and intensity was quantified as follows: 1 = a single frog calling; 2 = a few frogs calling; 3 = loud chorus with many frogs. For all species other than Rana catesbeiana and R. clamitans, a "2" meant one could distinguish among individual calling males, and a "3" meant individuals could not be distinguished from the chorus. However, because of a low density of R. catesbeiana and R. clamitans at our sites, "2" meant two of that ranid species were calling and "3" meant that three or more were calling.

Five species of anurans were consistently heard throughout the study period. Two of these were treefrogs: Hyla cinerea (green treefrog) and H. gratiosa (barking treefrog), two ranids, R. catesbeiana (bullfrog) and R. clamitans (bronze frog), and Acris gryllus (southern cricket frog). Two other species of anurans, Gastrophryne carolinensis (narrowmouth toad) and H. femoralis (pine woods treefrog), were recorded only rarely during the study. It should be noted these two species were only heard after rain, and H. femoralis was only heard at the end of the study period after a two-day period of rain.

The treefrogs usually began calling shortly after sundown (2100 to 2200) at both sites (Fig. 5). Calling intensity of R. catesbeiana and R. clamitans reached its highest intensity in the early morning hours from about midnight to 0600 (Fig. 6), but for R. catesbeiana the increase was much greater at Flamingo Bay. Acris gryllus called consistently at all times during the 24-hour cycle but increased calling was observed at Ellenton Bay during 2230 to 0330 (Fig. 7. We also discovered considerable temporal differences in calling intensity over the 25-day study period between the two study sites. Both treefrogs and R. clamitans called at approximately the same intensity at both sites (Fig. 8). However, R. catesbeiana and A. gryllus (Fig. 9) had very different average intensities between the two sites.

The automated recording system allowed us to study anurans in their environment with minimal disturbance and to sample anuran calls systematically over a long period of time, allowing us to draw several conclusions regarding temporal variation in anuran calling behavior. Treefrogs have a high intensity temporal calling period from 2100 to midnight. Past midnight, calling by treefrogs significantly lessened in intensity. In contrast, calling intensity in R. catesbeiana (only at Flamingo Bay) and R. clamitans did not peak until well after midnight (Fig. 6). The smallest anuran of the study, A. gryllus, called 24-hours a day with only an increase in average intensity at night at Ellenton Bay.

Although there is a small increase in temporal calling intensity for the treefrogs and R. clamitans at Flamingo Bay (Figs. 5 and 6), this increase is not considered significant. This increase is attributed to the placement of the microphone near vegetation in which the anurans called from, thus giving a louder call than at Ellenton Bay. The difference in intensity for R. catesbeiana (Fig. 6) is a much greater increase than resulting from microphone placement alone and is therefore considered a significant difference in temporal calling intensity. As a result, comparisons of calling intensity at our two sites showed similar temporal results for all species. Thus, we conclude that temporal variation in calling behavior is not unique to individual sites of our study area.

Although the temporal calling patterns were similar between the two study areas, calling intensities of some species differed. Even though both treefrogs and R. clamitans had similar calling intensities at both sites studied, this was not the case with R. catesbeiana and A. gryllus. With R. catesbeiana and A. gryllus, calling intensity was never equal between the two sites. For R. catesbeiana, Flamingo Bay had the higher calling intensity, yet for A. gryllus, calling intensity was higher at Ellenton Bay. Assuming calling intensity is positively correlated population size the R. catesbeiana population is presumably larger at Flamingo Bay and the A. gryllus population is larger at Ellenton bay. A possible reason for differences in calling intensity, and thus population sizes, are differences in vegetative composition between the two bays. Flamingo Bay is a wetland with many large trees and limited amounts of understory vegetation and R. catesbeiana is known to prefer an aquatic, open habitat (Conant and Collins 1998). The opposite is true for the small anurans such as A. gryllus. The ideal habitat for this small frog is grasses and lily pads (Conant and Collins 1998) characteristic of Ellenton bay.

Manual calling surveys are a frequently used method of monitoring anurans. Data obtained using automated recording systems can assist in the optimization of anuran monitoring programs that use manual calling surveys (Peterson and Dorcas 1994). Automated recording systems provide the ability to collect data systematically over a long period of time with minimal disturbance to the anurans. Automated recording systems are particularly useful in detecting species that call late at night, which might be missed in manual calling surveys. In this study, if manual surveys were conducted during early evening, treefrogs would be detected but ranids might not be. However, if one waited until early morning to conduct calling surveys, R. catesbeiana and R. clamitans would be perceived as abundant, yet the treefrogs would have likely ceased calling. Thus, if a certain species is targeted during a manual survey, data obtained from automated recording systems can be used to give a good indication of when the anuran in question is most likely to be heard.

Another advantage to the automated recording system over manual monitoring programs is the ability to detect calls of infrequently calling species (Peterson and Dorcas, 1992). For example, H. femoralis and G. carolinensis were only heard a few times during the 25-day study period. In fact, H. femoralis was only heard the last two days of the study and only after a two-day period of rain. If a manual survey were conducted during our study period it would be highly unlikely these infrequent calling species would be heard and thus the manual survey data would not include these species even though they inhabit the area.

Beiswenger, R.E. 1986. An endangered species, the Wyoming Toad Bufo hemiophrys baxteri -the importance of an early warning system. Biological Conservation 37(1986):59-71.

Bridges, A. S. and M. E. Dorcas. in press. Temporal variation in anuran calling behavior: implications for surveys and monitoring programs. Copeia.

Corn, P.S. and J.C. Fogleman. 1984. Extinction of montane populations of the Northern Leopard Frog (Rana pipiens) in Colorado. Journal of Herpetology 18(2):147-152.

Heyer, W.R., M.A. Donnelly, R.W. McDiarmid, L.C. Hayek, and M.S. Foster (eds).1994. Measuring and monitoring biological diversity: Standard methods for amphibians. Smithsonian Institution Press, Washington, D.C.

Gibbons, J.W. and R.D. Semlitsch. 1981. Tesrrestrial drift fences with pitfall traps: An effective technique for quantitative sampling of animal populations. Brimleyana 7:1-16.

McAllister, K.R. and B. Leonard. 1990. 1989 Progress Report - Past distribution and current status of the spotted frog in western Washington. Washington Department of Wildlife, Wildlife Management, Nongame Program.

Mohr, J. R. and M. E. Dorcas. 1999. A comparison of anuran calling patterns at two Carolina Bays in South Carolina. Journal of the Elisha Mitchell Scientific Society 115:63-70.

Peterson, C.R. and M.E. Dorcas. 1992. The use of automated data acquisition techniques in monitoring amphibian and reptile populations. pp. 369-378. in Wildlife 2001: Populations. (D.R. McCullough and R.H. Barrett, eds.). Elsevier Applied Science, London.

Peterson, C.R. and M.E. Dorcas. 1994. Automated data acquisition. pp. 47-57. in Measuring and Monitoring Biological Diversity - Standard Methods for Amphibians. (W.R. Heyer, R.W. McDiarmid, M. Donnelly, and L. Hayek, eds.). Smithsonian Institution Press, Washington, D.C.

Terborgh, J. 1989. Where have all the birds gone? Essays on the biology and conservation of birds that migrate to the American tropics. Princeton University Press, Princeton, N.J.

Robbins, C.S., J.R. Sauer, R.S. Greenberg, and S. Droege. 1989. Population declines in North American birds that migrate to the neotropics. Proceedings of the National Academy of Sciences (USA) 86:7658-7662.

Wake, D.B. 1991. Declining amphibian populations. Science 253: 860.

Return to Funding Home Page