Patuxent Wildlife Research Center

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

PVC Pipe Refugia: A Sampling Method for Studying Treefrogs

Elizabeth Domingue O'Neill and Robin G. Boughton

Soil and Water Science Dept.
University of Florida, Gainesville, FL

Dept. of Wildlife Ecology and Conservation
University of Florida, Gainesville, FL

[ Abstract ]


Of the popular sampling methods used to study herpetofauna, few if any are efficient in capturing treefrogs particularly during the nonbreeding season. However, Murphy (1993) designed a modified drift fence specifically for capturing treefrogs which he concluded was efficient for their capture. The use of poly-vinyl chloride (PVC) pipes is a much more efficient sampling method in regard to installation time and cost, maintenance, and ease in monitoring of treefrogs. The PVC pipe refugia sampling method is effective in capturing Hylid treefrogs and is capable of providing a variety of data on their biology and life history. An important limitation of this technique is that it does not lend itself to precise estimates of population size.

In 1958, Goin reported on a study conducted in her yard on Hylid treefrogs. Often artificial shelters provided by Goin, a 1-in diameter metal pipe had the highest total count of frogs recorded. With the natural refugia included, the pipe captures were second only to palmetto. Goin's results demonstrated that a pipe .refugia-type of 'trap' worked well for capturing Hylid treefrogs. Yet, according to the literature, this method has been little utilized by the scientific community until recently. Only a few studies have used PVC pipe refugia for the capture of treefrogs (Buchanan 1988, K. Greenberg pers. comm., Phelps 1993, Moulton et al. 1995, O'Neill 1995, Boughton and Staiger unpub. data). Several researchers have captured treefrogs using other types of artificial refugia such as bamboo (Stewart and Pough 1983) and wooden nest boxes (McComb and Noble 1981).

In each of these studies treefrogs used retreats heavily. No technique paper has been published describing this sampling method, however, one is in press (Moulton pers. comm., also see Moulton this conference). Data collected in the flatwoods and upland studies will be published at a later date. Suggestions, ideas, and data presented in this paper are supported by an ongoing study in a slash pine flatwoods ecosystem and by a second study conducted in an upland sandhill, both in north central Florida. This paper will discuss various aspects of the PVC pipe technique including its benefits, biases, areas requiring further study, and recommendations for study design. The use of PVC pipe refugia is a very attractive and feasible methodology for use in a regional monitoring program.

Flatwoods Site: In August of 1993, 120 PVC pipes were installed upright in the ground to be used in conjunction with drift fence/funnel trap arrays, concentrated at 9 small temporary ponds. Pipes of two diameters were utilized, 1.9 cm (3/4 in) and 3.8 cm (11/2 in). On the same site, a grid of 154 additional pipes was installed in January of 1996. A single 3.8-cm pipe was set every 15 meters throughout the pond and upland habitats. All pipes are approximately one meter tall and are inserted about 5-10 cm into the soil so as to be free standing and vertical. All animals captured are weighed, measured and individually marked with toeclips and (if large enough) passive integrated transponders (PIT tags). For a more detailed description of the study site and methods refer to O'Neill 1995.

Upland Site: At this site, 776 pipes were placed in trees at three separate locations; along the shore of a permanent lake, around a sinkhole pond, and in an upland pine-oak hammock. Variables of pipe design and placement were chosen to cover a spectrum of factors which treefrogs would likely consider when choosing a retreat. Pipes of two diameters (1.9 cm and 3.8 cm) were hung vertically in pine and oak trees at several heights. Caps and "T" fittings were installed on equal portions of these pipes. A portion of capped pipes were maintained with several inches of water inside. Finally, pipes were of two lengths, 30 cm and 60 cm. All treefrogs were weighed, measured and toe-clipped.

Benefits over other sampling methods

PVC pipe refugia are highly effective in capturing treefrogs. Although drift fence arrays are useful for capturing a variety of species, they are less effective in capturing treefrogs (Gibbons and Bennett 1974, Gibbons and Semlitsch 1981, Campbell and Christman 1982, Vogt and Hine 1982, Bury and Corn 1987, Dodd and Scott 1994). In the flatwoods study all amphibian captures in PVC pipe refugia have been of the Pinewoods treefrog, Hyla femoralis. Funnel traps used on the site also captured H. femoralis, though a drastic difference in capture effectiveness was observed between, thee two sampling methods. In the pine flatwoods study, 205 individuals and 276 ecaptures were recorded after 29,372 pipe-checks. Over the same time eriod only 59 individual and 7 recaptured H. femoralis were captured in funnel traps after 52,477 trap nights. In funnel traps, one new individual was captured for every 889 trap nights and one recapture was made for every 7497 trap nights. In. comparison, for PVC pipe refugia one new individual was captured for every 143 pipe-checks and one recapture was made for every 13 pipe-checks. At the upland site, 52,416 pipe- checks found 788 individuals and 2501 recaptures of four species, H. squirella, H. cinerea, H. femoralis, and H. gratiosa. One new individual was captured for every 66 pipe-checks and one recapture was made for every 21 pipe-checks.

High recapture rates of marked individuals and the subsequent recording of measurements at each capture provides information on sex ratio, growth rates, life span, age of sexual maturity, movement, breeding phenology and other life history aspects.

This method is inexpensive.

Installation and monitoring can be done quickly and easily and does not degrade the habitat as do some other techniques.

Little maintenance is required. Occasionally pipes need to be reinserted into the ground. Also, jumping spiders (Salticidae) and their webs regularly need to be removed from inside the pipes in the flatwoods and upland studies. Pipes have remained on the flatwoods site for 3 years (year-round) and no replacements have been necessary.

Because animals are not literally 'trapped', pipes can be checked as often, or infrequently, as a person's schedule allows.

Little or no mortality occurs with this method. The primary risk of mortality is in crushing a frog when replacing a pipe, thus care must be taken when returning pipes.

Because the technique requires less time and effort than other methods, it may make it more feasible for larger areas to be monitored over longer periods of time. Researchers can sample year-round across different habitats, providing valuable information not typically obtained. While most researchers concentrate their efforts at the breeding sites (ponds), PVC pipes can be installed across breeding and nonbreeding habitat. This will likely give researchers a fuller understanding of treefrog biology, including dispersal distances, daily activity patterns, and winter home range size.

PVC pipe refugia may be the best option in areas where typical drift fence arrays would be difficult to install (e.g., steep topography and/or dense vegetation). The placement of pipes in trees allows access to those treefrog species which remain in the shrub or tree canopies and infrequently, or never, venture onto the ground surface.

Pipes installed upright in the ground are highly predator-proof. Snakes have difficulty climbing the pipes, and the small diameter of the pipes prevents entry by most other predators. At the upland site, however, it is likely that mortality did occur in tree pipes via tree-climbing snakes. On the flatwoods site, there has been only one known occurrence of predation on frogs using pipes. A juvenile H. femoralis was eaten by a large jumping spider (Phidippus regius) while using a pipe.

Biases associated with PVC pipe refugia sampling method
Individuals are attracted to PVC pipe refugia and can be considered trap-happy. In the flatwoods study, 77% of all individuals captured are recaptured at least once. Over an 18-month period on the flatwoods site, individuals were captured an average of 12 times (maximum 80). In a ten-month period at the upland site, individuals were recaptured an average of 5 times at the sinkhole pond and 2.2 times at the lake and upland hammock locations.

Use of pipes by different treefrog species and by individuals in general can be affected by habitat type, pipe placement (ground vs. tree), structure of vegetation adjacent to pipes, pipe diameter, tree species, and environmental conditions such as season, temperature, and rainfall. For example, occupation of PVC pipe refugia by H. femoralis in the flatwoods showed a significant relationship with rainfall. H. femoralis occupied pipes on significantly more days with low rainfall (< 2.0 cm) than on days with significant rainfall (> 2.0 cm) (c2 = 8.155, df = 1, P = 0.0043). At the upland site, regression showed that adjacent vegetation at the tree base, at the midstory, and at the canopy were not strong predictors of capture rate. Tree DBH and species were strong predictors of capture rate. Phelps (1993) found no correlation between adjacent vegetative cover and capture rate in a South Carolina bottomland. Capture rate at the upland site was positively correlated with mean temperature for each capture day but was not correlated with rainfall or relative humidity.

Intra- and interspecific interactions may affect the sex, number of individuals, and species using pipes.

Considering these points it is unlikely that precise, statistically sound measures of population size can be determined with this method. However, this method does provide information on relative abundance, species occurrence, and trends in the population using the pipes. Trends seen in captures can serve as a representative sample of what is occurring within the overall local population.

Only select species will or can use PVC pipe refugia. While Anolis carolinensis individuals will use the pipes, this method is strongly biased toward use by treefrogs.

Recommendations for Study Design / Applications for regional use
Use a grid system coupled with mark/recapture methodology for population studies. Transects can be used for species occurrence studies. A grid of PVC pipe refugia can be used to sample across different habitats, such as uplands and wetlands (both often required by treefrog species). By using the same sampling method in both habitat types, trap bias will be eliminated (i.e., different capture rates would not be due to different trap types). Use of cover objects in grids or transects is considered to be a useful technique to evaluate changes in the activity and abundance of terrestrial salamanders as well as provide a systematic and repeatable way to survey for salamander populations over a period of years (DeGraaf and Yamasaki 1992, Fellers and Drost 1994, Brooks 1996). The same should be true for the use of PVC pipe refugia in monitoring treefrog populations, as treefrogs likely are attracted to and use PVC pipe refugia for the same or similar reasons as salamanders are attracted to and use cover boards.

Ground Pipes
Use 3.8 cm (1 1/2 in) diameter pipes approximately 1-m in length. In the flatwoods study, frogs are found in the large pipes (3.8 cm) 1.5 times more often than in the smaller pipes (1.9 cm). PVC pipe comes in 10-ft sections. These can be cut into 3 sections of approximately one meter each (or shorter if desired).

Install 100 or 150 pipes in a grid with one pipe every 15 to 25 meters apart. These numbers are based on data from the flatwoods study. Pipe occupancy for H. femoralis is variable and can range from 0% to 35%. With the use of 150 pipes, an average occupancy of approximately 10% (about 15 frogs) can be expected for each day pipes are checked. As the number of days sampled increases, the number of frogs captured will also increase. Distance moved by H. femoralis ranged from 4.5 - 26.5 m. The average distance detected was approximately 13 m. Therefore, placing pipes at least 15 m apart may provide a better chance for sampling different individuals rather than the same individual moving within its home range area. Because this data is based on one species in one habitat type, recommendations for the number of pipes required and their spacing should be considered only as a starting point for projects, until we can gather more information.

Tree Pipes
Pipes should be 3.8 cm in diameter, approximately 60 cm in length, and hung approximately 2 m high in hardwood trees. At the upland site, both length and diameter of pipes were highly significant predictors of capture rate. Long pipes (60 cm) were favored 5:1 over short pipes (30 cm) and pipes 3.8 cm in diameter were favored 3:1 over pipes 1.9 cm in diameter. Pipe location was also an important predictor of capture rate. Pipes in hardwoods were favored 2.5:1 over pipes in pines and pipes 2 m and higher were favored 3:1 over pipes hung at the base of trees. Pipes fitted with bottom caps and holding water were favored over other designs in the upland study. While only 20% of all pipes were of this pipe design, they represented over 55% of all captures.

A grid or transect may be set up using available trees. One pipe per tree is sufficient. For studies using ground pipes, tree pipes or a combination

Pipes could be checked at least once each month. Frequency of monitoring could be increased as desired, particularly in regions where species are active for only a portion of the year (treefrogs are active year-round in Florida).

Monitoring treefrogs with this method would likely be a good system for a volunteer effort. The low cost, ease of installation and monitoring, low maintenance, and flexibility in time investment should be very attractive for regional surveys. The amount of time invested will vary depending on volunteer and other effort, level of data collection, number of pipes installed, frequency of pipe checks, and seasonality of the local treefrog species' activity. Coordinating a mark/recapture effort is likely beyond the scope of most volunteer-based programs. Species occurrence data and total number of captures per trap day can be gathered without marking animals. This information may be useful for year-to-year comparisons. Involvement by youths in schools, 4-H Clubs, scouts, etc. would help expand a volunteer effort for regional monitoring especially in areas less likely to be sampled (in or near residential and on smaller tracts of land).

Aspects that require further study

Additional Notes
If a larger budget is available, researchers may want to consider marking individuals with PIT tags. The PVC pipe refugia/PIT tag system may serve as a very efficient method for sampling treefrog populations if researchers are interested in measuring relative abundance over the long term and their study does not require marked individuals to be measured repeatedly. PIT tags can be read directly through the pipe without disturbing the frog.

Why are treefrogs using PVC pipes? PVC pipe refugia placed upright in the ground retain moisture longer than the surrounding vegetation and they seem to provide a place out of the wind. Frogs change their position in the pipes (up and down) relative to the air temperature. Pipes in trees probably provide increased protection from predation, shade, and pipes capped with water inside provide moisture. The attraction to pipes may simply be due to the fact that the pipes mimic a natural shelter, providing protection from the weather (elements) and aiding the frogs in controlling water loss and body temperatures.

Frequent disturbance of frogs by toe-clipping, weighing, and measuring may lead to lower capture rates. When the loss of moisture, energetic expense, and the predation-like experience which occurs during data collection outweigh the benefits of the pipe refuge, treefrogs may evacuate. Recapture rates at the flatwoods site and at Phelps (1993) blackwater bottomland show that treefrogs have a very high tolerance for disturbance by researchers. However, at the upland sites recapture rates were far lower, and disturbance was evidently more significant. Therefore, it may be beneficial to avoid checking traps more than twice weekly.

Literature Cited
Brooks, R.T. 1996. Terrestrial salamander monitoring, in Protocols and Strategies for Monitoring North American Amphibians (unpublished). North American Monitoring Program (url: //amphib/naampsala.html)

Buchanan, B.W. 1988. Territoriality in the Squirrel treefrog, Hyla squirella: competition for diurnal retreat sites. Unpubl. M.S. Thesis, Univ. of SW Louisiana, Lafayette.

Bury, R.B. and P.S. Corn. 1987. Evaluation of pitfall trapping in northwestern forests:

Trap arrays with drift fences. Journal of Wildlife Management 51(1):112-119.

Campbell, H.W. and, S.P. Christman. 1982. Field techniques for herpetofaunal community analysis. pp. 193-200. In N.J. Scott (ed.), Herpetological Communities. U.S.D.I. Fish and Wildlife Service, Wildlife Research Report 13.

DeGraaf, R.M. and M. Yamasaki. 1992. A nondestructive technique to monitor the relative abundance of terrestrial salamanders. Wildlife Society Bulletin 20:260-264.

Dodd, C.K., Jr. and D.E. Scott. 1994. Drift fences encircling breeding sites. pp.125-130. In W.R. Heyer, M.A. Donnelly, R.W. McDiarmid, L.C. Hayek, and M.S. Foster (eds.) Measuring and Monitoring Biological Diversity: Standard Methods for Amphibians. Smithsonian Institution Press, Washington DC. 364pp.

Fellers, G.M. and C.A. Drost. Sampling with artificial cover. pp.146-150. In W.R. Heyer, M.A. Donnelly, R.W. McDiarmid, L.C. Hayek, and M.S. Foster (eds.) Measuring and Monitoring Biological Diversity: Standard Methods for Amphibians. Smithsonian Institution Press, Washington DC. 364pp.

Gibbons, J.W. and D.H. Bennett. 1974. Determination of anuran terrestrial ctivity atterns by a drift fence method. Copeia 1974(1):236-243.

Gibbons, J.W. and R.D. Semlitsch. 1981. Terrestrial drift fences with itfall traps: an ffective technique for quantitative sampling of animal populations. rimleyana 7:1-16.

Goin, O.B. 1958. A comparison of the nonbreeding habits of two treefrogs, Hyla squirella and Hyla cinerea. Quarterly Journal of the Florida Academy of Sciences 21(1):49-60.

McComb, W.C. and R.E. Noble. 1981. Herpetofaunal use of natural tree cavities and nest boxes. Wild. Soc. Bull. 9(4):261-267.

Moulton, C.A., B.R. Nerney, and W.J. Fleming. 1995. The use of PVC pipes to capture hylid treefrogs for quantitative field experiments. Abstract from The Wildlife Society Meeting, September 1995, Portland, Oregon.

Murphy, C.G. 1993. A modified drift fence for capturing treefrogs. Herpetological Review 24(4):143-145.

O'Neill, E.D. 1995. Amphibian and reptile communities of temporary ponds in a managed pine flatwoods. M.S. Thesis, University of Florida, Gainesville. 106pp.

Phelps, J.P. 1993. The effect of clearcutting on herpetofauna of a South Carolina blackwater bottomland. M.S. Thesis, North Carolina State University, Raleigh. 66pp.

Stewart, M.M. and F.H. Pough. 1983. Population density of tropical forest frogs: relation to retreat sites. Science 221:570-572.

Vogt, R.C. and R.L. Hine. 1982. Evaluation of techniques for assessment of amphibian and reptile populations in Wisconsin. pp.201-217. In N.J. Scott (ed.), Herpetological Communities. U.S.D.I. Fish and Wildlife Service, Wildlife Research Report 13.


U.S. Department of the Interior
U.S. Geological Survey
Patuxent Wildlife Research Center
Laurel, MD, USA 20708-4038
Contact: Sam Droege, email:
Last Modified: June 2002