Patuxent Wildlife Research Center

NAAMP III Archive - new techniques
Home | Archive by Alphabetical Order | Archive by Category

The Use of PVC Pipes to Capture Hylid Treefrogs

Cynthia A. Moulton*, W. James Fleming, and Brenden R. Nerney

* Address as of January 1997
Department of Natural Sciences
Castleton State College
Castleton, Vermont 05735
email: brncam@worldweb.net

W. JAMES FLEMING
North Carolina Cooperative Fish and Wildlife Research Unit
North Carolina State University
Raleigh, North Carolina 27695
email: jim_fleming@nbs.gov
and
BRENDAN R. NERNEY
The Chelsea School
Silver Spring, Maryland 20910
email: brncam@worldweb.net

Introduction
Drift fences, pitfall traps, and spotlighting have long been used for amphibian censusing (Campbell and Christman 1982). However, these methods are ineffective for hylid treefrogs (Dodd 1991, Jones 1986, Greenberg et al. 1994). Many treefrogs can climb out of pitfall traps and over drift fences. They may escape visual detection from spotlighting because of their small size, cryptic coloration, and densely-vegetated habitat. Anecdotal accounts have reported sighting treefrogs in PVC pipes or bamboo stalks, although this was not necessarily the intended use of these objects. Because of these observations, we explored the use of PVC pipes for quantitative treefrog population assessments. Field studies in 1993 and 1994 addressed the following questions:

1.  If a number of pipes were placed in randomly selected wetland habitat sites, would treefrogs      use them for refuge during the day?

2.  Would it be possible to capture enough treefrogs in the PVC pipes to conduct mark-recapture      population analysis?

3.  Would different species, sizes and age classes of treefrogs use the pipes?

4.  What design configuration would maximize trapping success? These issues and the      applicability of using PVC pipes in future quantitative field studies are discussed.

Methods
Study Area.--The Pocosin Lakes National Wildlife Refuge (PLNWR) in eastern North Carolina covers 45,325 ha where ditches and canals intersperse between pocosin habitat of grasses (Juncus effusus, other Juncus spp. and Andropogon spp.), flowering plants (Saururus cernuus, Solanum carolinense, Boehmeria cylindrica, and Polygonum setaceum), and low shrubs (Persea palustris, Rhododendron viscosum, Rosa palustris, Ilex glabra, and Myrica cerifera). Some areas of the refuge support agricultural crops. Wheat, corn, and soybeans are alternately grown on leased sections within the refuge and much of the surrounding region consists of farming communities. Three species of hylid treefrogs (Hyla cinerea, H. squirella, and H. chrysoscelis), as well as many ranid and bufonid frogs annually breed along the ditches while the pine woods treefrog (H. femoralis) and little grass frog (Pseudacris ocularis) breed in small ephemeral and permanent ponds. During the summer months daily temperatures usually exceed 30°C with rainfall occurring in heavy sporadic episodes ( Table 1).

Methods.--Unless otherwise indicated, 1 m sections of opaque white PVC pipe (approximately 2 cm inside diameter) were inserted upright in the ground to capture treefrogs. Frogs are not restrained or restricted in these pipes; they may move freely in or out of the hollow opening at the top. During the trapping sessions, all pipes were checked daily between 0700 and 1700 h. When frogs were present, individuals were removed from the pipes, measured, and marked by toe-clipping for individual recognition. When extremely overcast days and/or thick vegetative cover impaired visibility inside the pipe, we passed a flashlight along the outside of the pipe to see if a frog had taken refuge inside.

In 1993, 10 sites 250 m in length were randomly chosen from 35.5 km of ditch bank habitat. Habitat characteristics were similar for all sites because of the uniformity of the land on the refuge. Paired PVC pipes were placed at 10-meter intervals along ditches at 1 and 2 m from the water's edge. At three of the 10 sites, 10 PVC pipes, each 1 m in length and 5-cm inside diameter, were also positioned. Trapping was conducted at each site from June through August in three 5-day and two 3-day trapping sessions (21 total trapping days per site). Additionally, we surveyed each site twice per trapping session for nighttime frog calling. Night surveys entailed visiting each site at least 30 minutes after sunset for a 10-minute period during which all calling frogs were noted. Each calling species was assigned a category of abundance as follows: (1) single; one individual calling, (2) few; at least two but no more than three frogs calling, (3) several; more than three individuals calling, with each call being distinguishable from the other, (4) abundant (chorus); so many frogs calling that individual calls could not be distinguished.

In 1994, nine sites were randomly chosen. At each site, we embedded a round, 884-liter (137 cm diameter) cattle trough, at least 10 cm into the ground. We set pipes in an 80-point grid formation (9 x 9 excluding the middle point) around each trough. Points were 1 m apart, with two pipes placed at each point. Along the ditch nearest to each trough (within 100 m), paired pipes were set at 1-m intervals for 25 m. Thus, each site consisted of 210 pipes and a cattle trough. We conducted 3-day trapping sessions every six days from June 7 through August 11. We ran four additional 2-day trapping sessions on Aug. 17-18, 27-28; September 9-10, and 24-25 (totaling 33 trapping days at each site). Early in June we heard H. femoralis chorusing from a small pond near one of the sites. We set 50 pipes 1 m apart along the edge of this pond, which was about a quarter of the entire perimeter. We checked these pipes for 10 two-day trapping sessions from June through September.

Statistical Methods -- We used the Chi-square test for heterogeneity to decide whether the 1994 data from sites and sessions could be pooled for mark-recapture analysis. To test if frogs preferred no vegetative cover, partial cover, or full vegetative cover around the pipes, we used the chi-square goodness of fit.

Results
In 1993, there were 62 individuals marked and released, with 71 recaptures. PVC pipe yielded three species of tree frogs; H. cinerea, H. femoralis, and H. squirella. All frogs were of adult size, and ranged from 21 to 51 mm, with a mean and standard deviation (s.d.) of 31.69 +/- 4.69 mm. Most captures occurred in the smaller diameter (2 cm) PVC, with only two of the 133 captures found in the larger size (5 cm) PVC. Individuals were usually found in the same pipe, or a nearby pipe, each time they were discovered.

We consistently found H. femoralis at two of the ten sites, with 41 individuals captured from the PVC pipes and 35 subsequent recaptures. Several H. femoralis were heard calling (and chorusing) from these two sites during the night surveys. We identified H. squirella during the night surveys at three sites. However, no more than two individuals at a site were ever noted during call surveys. From these three sites, 16 individual H. squirella were captured in the pipes, with 32 recaptures. On two sites, several H. cinerea were heard, but we caught only four individuals, with five recaptures. In every case and for all sites, if a treefrog species was noted in the night surveys, that species was also captured in the PVC pipes; conversely, where no frogs were heard, none were trapped ( Table 2).

In 1994, we captured 78 individual H. femoralis, with 210 recaptures. We also caught five frogs that had been marked in 1993. These captures occurred at two sites that were approximately 2 km from the frogs' original points of capture the previous year. Sizes of adult treefrogs ranged from 28 to 45 mm with a mean and s.d. of 35.21 +/-3.53 mm. In addition, we found newly emerged H. femoralis in PVC pipes a total of nine times; at the natural breeding pond and at two trough sites. Since we wanted to minimize handling of the newly-emerged frogs, we did not toe-clip any of them and measured only four whose lengths were all about 8 mm. Finally, we found as many as four individual P. ocularis (sizes 15-16 mm), with 12 total captures, at the natural pond site. The peak number of H. femoralis found at the trough sites occurred from late June to early August (Fig. 1). We trapped between 0 and 6 individuals per trough site per day, yet consistently trapped between 4 and 8 frogs at the natural breeding pond site. Most frogs were caught less than four times during the trapping period, and were seen in only a few sessions. But, 16 individuals (mostly male) were captured repeatedly and lingered at the troughs throughout much of the summer (Fig. 2&Fig. 3). On five occasions, we found two frogs in the same pipe. During the 1994 trapping season, 3 H. femoralis moved between sites. Two individuals moved to sites that were approximately 1700 m apart. One frog moved from the natural breeding pond site to a trough site, a distance of 2.5 km in 10 days (or less).

For 283 of the captures, we recorded whether the pipe was in vegetative cover, partial cover, or no cover. After determining available cover for the nine sites, we found no significant difference in capture rates between the three categories (P<0.05; X2=3.75). Captures occurred at pipes without vegetative cover 109 times, partially covered pipes 104 times, and 70 times in pipes covered by vegetation.

Discussion
Monitoring amphibian populations is accomplished in a variety of ways. Some techniques focus on trapping adults of a species by drift fences and pitfall, double-ended, and single-ended traps. Most amphibians and reptiles follow along the fence on the ground, and eventually enter or fall into traps. Adult treefrogs have eluded conventional traps because they can climb over drift fences and out of pitfall traps. Other means of capturing treefrogs for mark-recapture studies include catching them in nets at night in breeding areas. Netting adult frogs during breeding activity not only interferes with such activity and may injure frogs, but may not be possible in all areas. Other monitoring programs may include using call surveys, but there is little validation of calling activity as an index of population size. Monitoring larval populations via numerous types of netting strategies can be used but are time-consuming in both catching and identifying tadpoles and salamander larvae. With some species of treefrogs, such as the ones mentioned in this study, tadpoles are very small and quick, and are commonly found in darkly-stained water (high tannic acid content). Nets that would have mesh fine enough to capture the tadpoles would also be slow when moving through water, possibly too slow to capture tadpoles, thus biasing these efforts.

Using PVC pipes to trap adult and newly-emerged treefrogs for quantitative field studies appears to be a feasible way to monitor hylids and possibly other frogs. The Lincoln-Peterson index (for closed populations) and Jolly-Seber models (for open populations), with computer programs such as CAPTURE (Otis et al. 1978) and JOLLY (Pollock et al. 1990), may be useful for mark-recapture population analysis of PVC pipe trapping data. If one were conducting experimental field studies, parameters such as abundance, recruitment, birthrate, and/or survival could be estimated, along with their associated variances. In fact, because of the clear size differences in adults vs. newly-emerged frogs, recruitment and birth rates may be differentiated. It is not possible to do this, as easily, for other types of organisms.

In this study, we captured four species of frogs in the PVC pipes: H. cinerea, H. femoralis, and H. squirella along with the little grass frog, P. ocularis. We caught frogs of sizes ranging from 8 to 51 mm in the pipes. Also, we could distinguish two age classes of H. femoralis, that used the pipes, adults and newly-emerged frogs. In addition, five adults were caught both in 1993 and in 1994.

When comparing call survey results with actual PVC-trapped frogs a definite relationship exists. First, where we heard a species calling during night surveys, we caught that species in the pipes. When we heard one or two individual H. squirella and H. femoralis during call surveys, we very often trapped one or two individuals in the PVC pipes. When choruses of H. femoralis were noted during call surveys, we captured several individuals in the pipes. However, where choruses of H. cinerea were discovered, only a few individuals were ever captured in the PVC. Chorusing H. cinerea appeared to be more territorial and widespread than H. femoralis, so perhaps the site was too small to accommodate more trapping of H. cinerea .

Pipes placed at the natural breeding pond yielded higher numbers of treefrogs and in fewer trapping days, but this site was unreplicated. Also, we placed the pipes after hearing several nights of chorusing at this pond. Therefore, if one intended to survey areas specifically for treefrogs, placing pipes after frogs begin chorusing would likely maximize trapping success. Furthermore, of the nine trough sites we set up in 1994, the only site where frogs were neither captured nor heard, was located within 500 m of this natural breeding pond. Trapping success at sites >1500 m appeared unaffected by the presence of the natural breeding pond. Thus, when planning experiments using artificial breeding pools, one should avoid setting up sites too close to natural breeding areas.

In designing field studies on this topic, whether experimental or observational, the following criteria should be considered:

1.   Pipes should be arranged near breeding water, either natural or artificial (i.e. cattle troughs,       barrels).

2.   Distances between pipes at small ephemeral ponds or artificial pools should be small, but not       so close that the vegetation around pipes becomes severely altered over several trapping       sessions. Pipes may be placed further apart in large wetlands where breeding activity is more       widespread. We recommend a 1-5 m pipe interval to facilitate trapping and recapture success       in experimental studies and small sites, and 10 m (or more) for large sites or survey studies.

3.   The best size for a site will vary depending on the type of study, but, generally, larger sites       would be needed for surveys.

4.   Trapping sessions should be 2 or 3 days. Longer sessions increase site disruption, do not add       significant numbers of individuals to the session data, and are more likely to violate the       assumption of a "closed population" necessary for some types of analyses.

5.   As always, replication is mandatory. Many pipes in one area does not qualify as a replicated       study. Even if one is interested in the population size of a single park, county, etc. multiple sites       should be included in the study design.

Using PVC pipes to capture treefrogs eliminates trap mortality and allows movement for breeding and foraging. Also, this trapping method is inexpensive, easy to use, set up and store. Perhaps the most beneficial aspect of PVC pipes to capture hylids is its use for sampling in extremely dense vegetation near otherwise inaccessible breeding habitats. By using PVC pipes to trap treefrogs, quantitative field studies may serve to supplement or validate call survey results, to assess effects of external treatments (i.e., pesticide applications) on treefrog population dynamics, and to monitor population trends. More research is needed to determine the optimum pipe configuration to maximize trapping effectiveness, to assess its usefulness with other species of Hylids, and to address trapping success in different habitat types or climates.

Acknowledgments.- We thank Cheryl Kelley and Kristen Lehman for their assistance in trapping and David Kitts for sharing his general knowledge of habitat on PLNWR and help in site selection. Many thanks go to Herman Berkhoff, Joe Phelps, and Paris Trail for recounting their experiences of finding treefrogs in pipes and bamboo. Kenneth H. Pollock, George Barthalmus and Harold Heatwole provided useful guidance for the study design. Harold Heatwole and James D. Nichols kindly assisted in the review of this manuscript.

Literature Cited
Campbell, H. W. and S. P. Christman. 1982. Field techniques for herpetofaunal community analysis. In N. J. Scott (ed.), Herpetological Communities. U.S. Fish and Wildl. Serv., Wildl. Res. Rept. 13, pp. 193-200

Dodd, C. K. Jr. 1991. Drift fence-associated sampling bias of amphibians at a Florida sandhill temporary pond. J. Herpetol. 25:296-301.

Greenberg, C. H., D. G. Neary, and L. D. Harris. 1994. A comparison of herpetofaunal sampling effectiveness of pitfall, single-ended, and double-ended funnel traps used with drift fences. J. Herpetol. 28:319-324.

Jones, K. B. 1986. Amphibians and reptiles. In A. Y. Cooperrider, R. J. Boyd, and H. R. Stuart (eds.), Inventory and Monitoring of wildlife Habitat, pp. 267-290. U.S. Dept. Interior, BLM Service Center, Denver, Colorado.

Otis, D. L., K. P. Burnham, G. C. White, and D. R. Anderson. 1978. Statistical inferences from capture data on closed animal populations. Wildl. Monogr. 62.135 pp.

Pollock, K. H., J. D. Nichols, C. Brownie, AND J. D. Hines. 1990. Statistical inference for capture-recapture experiments. Wildl. Monogr. 107. 97 pp.

Author's Note:
This paper was part of C.A.M.'s Ph.D. disseration. A shorter version of this paper will be published in Herpetological Review, Vol. 27 (4).

U.S. Department of the Interior
U.S. Geological Survey
Patuxent Wildlife Research Center
Laurel, MD, USA 20708-4038
http://www.pwrc.usgs.gov/naamp3/naamp3.html
Contact: Sam Droege, email: Sam_Droege@usgs.gov
Last Modified: June 2002