Home Archived April 13, 2016
(i)
clear pixel
U. S. Geological Survey - click to go to the USGS homepage
clear pixel clear pixel

 

frog graphic  

Sampling Techniques and Protocols


Dip netting for salamander larvae in Abrams Creek. - click to enlarge

       In the section that follows, brief examples are listed of how certain techniques have been used to sample amphibians. As stated in Things to Consider During Planning, there may be vastly different amounts of time associated with using the different techniques, different reasons for choosing them, and different biases when interpreting the results. In every instance, researchers should quantify the amount of search time or sampling effort involved in the survey.

Active Sampling

       Time constrained __ In this technique, a predetermined amount of time is set for sampling the area or habitat. The presence of different species and the number of individuals (or even sex and life stagemales, females, juveniles) observed are recorded. Visual encounter protocols are

Figure 32. Turning logs in time constrained survey at Beech Flats. - click to enlarge
Figure 32. Turning logs in time constrained survey at Beech Flats.

followed; that is, animals are counted as they walk over the forest floor or stream bottom, hide in crevices or cling to cave walls, found by turning over surface debris (figs. 32, 33), heard calling, or captured in random dip (fig. 34) or sweep nets (fig. 30). The number of observers x total amount of time sampled is recorded. In terrestrial and aquatic situations, times may be set for 15 or 30 minutes, occasionally longer, depending on the number of observers and the amount or quality of habitat to be surveyed.

       Example. A sampling protocol is set whereby three researchers hike along Noland Divide Trail for 30 minutes, conduct a 30 minute time-constrained survey, hike another 30 minutes followed by another 30 minute sample, and so on throughout the day.

Figure 33. Terrestrial time-constrained survey in thickly vegetated habitat at Balsam Mountain. - click to enlarge
Figure 33. Terrestrial time-constrained survey in thickly vegetated habitat at Balsam Mountain.

Four to six sites per day can be sampled with this method, depending on trail conditions and terrain. The sampling effort would be 3 x 30 = 90 person-minutes at each site sampled. Sample data might be 3 adult D. imitator, 5 P. jordani (2 males, 3 juveniles), and 1 subadult E. wilderae at site 1, with similar data recorded at every sampling location.

     What this tells the observer. Time constrained surveys provide information on: (1) species presence (but not absence) at the time of sampling; (2) life history information, such as when eggs are deposited, larval presence, and activity patterns; and (3) habitat information. Sampling effort is easily quantified.

       Limitations. Detectability is influenced by all the factors listed in Things to Consider During Planning. Even if every attempt is made to standardize sampling (for example, by sampling at the same time of day and during the same time of year), environmental factors likely will be different and thus influence

Figure 34. Dip netting for salamander larvae in Abrams Creek. - click to enlarge
Figure 34. Dip netting for salamander larvae in Abrams Creek.

whether a species will be observed. Because environmental variables influence the number of animals observed, differences in counts over time may be more reflective of differences in environmental conditions during the sampling periods among years than changes in status. It is very difficult to determine any kind of trend based on periodic counts because it is unknown what the relationship is between the counts and actual abundance. In addition, there may be considerable variation in the ability of the field observers to locate and count animals; some observers may find animals easily, whereas others might have great difficulty finding amphibians. Observer bias, thus, could skew count data in a manner which has nothing to do with the actual abundance of the animals counted.

       Area constrained __ In this technique, a defined amount of habitat is selected for sampling. For example, researchers might choose to sample large, randomly selected plots (such as 30 x 40 m plots; fig. 35); they might survey smaller plots (for example, 10 x 10 m plots) during a hiking survey; or they might survey a pond, wetland, or cave entrance, regardless of how much time is required. Plots may be singular or in groups (fig. 36). As above, the presence of different species and the number of individuals (or even sex and life stage-males, females, juveniles) observed are recorded. Visual encounter protocols also are followed; that is, animals are counted as they walk over the forest floor or stream bottom, hide in crevices or cling to cave walls, found by turning over surface debris, heard calling, or captured in random dip or sweep nets. The number of observers x total amount of time sampled is recorded.

       Example. Two persons search Gourley Pond for 67 minutes. The sampling effort is 2 x 67 = 134 person-minutes. Sample data might be: larval A. opacum (> 50 observed), 14 egg masses of A. maculatum, larval R. sylvatica (hundreds of tadpoles), 4 P. crucifer heard calling.

       What this tells the observer. Area constrained surveys provide information on: (1) species presence (but not absence) at the time of sampling; (2) life history information, such as when eggs are deposited, larval presence, and activity patterns; (3) habitat information; and (4) in some cases, a very crude estimate of density (the amount of area / number of animals). Sampling effort is easily quantified.

       Limitations. Detectability is influenced by all the factors listed in Things to Consider During Planning. Even if every attempt again is made to standardize sampling, environmental factors likely will be different and thus influence whether a species is observed. Since environmental variables influence the number of animals observed, differences in counts over time may be more reflective of differences in environmental conditions during the sampling periods among years rather than changes in amphibian population status. As with timeconstrained sampling, it is very difficult to determine any kind of trend based on periodic counts because the relationship between counts and actual abundance is unknown.

Figure 35. Schematic of a 30 x 40-meter sampling plot. - click to enlarge

Figure 35. Schematic of a 30 x 40-meter sampling plot. The grid is marked off in 5-meter intervals. The outside of the grid is marked with blue survey flags, whereas the rows are marked with pink survey flags. A stream is included on the left margin of the plot, so that both stream and terrestrial salamanders may be surveyed. Automated data loggers (red dot, DL) can be installed to record air and water temperature and relative humidity. Researchers walk up the survey lines turning coarse woody debris, rocks, and leaf litter. In addition to information on the species, size, and age class of salamanders observed or captured, the distance from water also can be recorded. This gives an idea of the spatial distribution of species across the plot.

 

 

 

 

 

 

Figure 36. Diagram of the relationship of three 30 x 40-meter fixed sampling plots at a location. - click to enlarge

 

 

 

Figure 36. Diagram of the relationship of three 30 x 40-meter fixed sampling plots at a location. Plots need not be isolated. In this schematic, three plots are located along the course of a stream. Each plot is surveyed once per year during the summer, all in the same order (A in June; B in July; C in August), for the length of the study. A single data logger station is located at one of the plots.

 

 

 

 

 

       Transects __ Transect sampling can be conducted using simple visual encounter survey techniques, such as by walking a preselected line transect at night and counting all the salamanders seen, or it can be used in conjunction with passive sampling techniques, such as the placement of coverboards along a preselected survey line. When using transects, sampling locations are determined through a stratified random process. A survey line of a prescribed length is selected, and observers use the line as a base from which to make observations.

       Example 1. Researchers select 50 locations in the fir-spruce forest where transects of 100 m length will be established. During the day, a starting point for the transect is selected. The direction of the transect is then determined from a set of random numbers from 1 to 360 (based on the number of degrees in a circle). Using a compass and a 100-m survey tape, fluorescent tape is used to designate the survey line. After dark, two researchers walk along the transect line, 5 minutes apart, and count all the salamanders, categorized by species, observed in their flashlight beams. The distance from the starting point where the salamanders were observed also is recorded. Using two researchers allows for a measure of potential observer bias.

       Example 2. A three-party survey crew samples the Little River for Hellbenders. The total amount of the river to be sampled is marked off in 100-m sections on a map, and ten 100-m sections are selected for sampling based on a random numbers chart. At the river, a starting point and an end point are marked using red survey flagging. Wearing wet suits, two observers snorkel along parallel transects about 4 m from the shore and look for Hellbenders under rocks, ledges, and other underwater hiding places. Observations are relayed to the third researcher walking parallel to the shore.

Figure 37. Stream sampling at Balsam Mountain. - click to enlarge
Figure 37. Stream sampling at Balsam Mountain.

       Example 3. Researchers select 50 stream locations on the northern side of the Park for sampling; the locations are selected based on elevation and accessibility. At each location, the stream is marked off in 5-m transects for a total of 100 m of stream length. Using a random numbers chart, seven transects are selected for sampling. A two-person team turns over all the rocks and searches hiding places, beginning downstream and working upstream, capturing and measuring salamanders (fig. 37). They call out the data (species, sex, length, age-class) to a third researcher walking parallel to the stream who records the information (fig. 38).

       Example 4. Researchers select 50 locations in the fir-spruce forest where transects of 100 m in length will be established. A starting point for the transect is selected. The direction of the transect is then determined from a set of random numbers (from 1 to 360, based on the number of degrees in a circle). Using a compass and a 100-m survey tape, fluorescent tape is used to mark the survey line. At every 10-m increment, a series of eight coverboards are laid out in a grid parallel to the transect line (fig. 39). The coverboards are then monitored periodically for salamander presence (see Coverboards).

Figure 38. Checking identification and recording data during stream sampling at Balsam Mountain - click to enlarge
Figure 38. Checking identification and recording data during stream sampling at Balsam Mountain.

       What this tells the observer. Areaconstrained surveys provide information on: (1) species presence (but not absence) at the time of sampling; (2) life history information, such as when eggs are deposited, larval presence, size-class structure, and activity patterns; (3) habitat information; and (4) in some cases, a very crude estimate of density (for example, a minimum number of salamanders inhabiting the selected length of the stream surveyed). Sampling effort is easily quantified.

       Limitations. Detectability is influenced by all the factors listed in Things to Consider During Planning. Even if every attempt is made to standardize sampling (for example, by sampling at the same time of day and during the same time of year), environmental factors likely will be different and thus influence whether a species is observed. Because environmental variables influence the number of animals observed, differences in counts over time may be more reflective of differences in environmental conditions during the sampling periods among years than changes in amphibian population status. It is very difficult to determine any kind of trend based on periodic counts, because it is unknown what the relationship is between the counts and actual abundance. On the other hand, the lifehistory information obtained using transect surveys may be valuable for understanding the basic biology and demography of the species sampled.

salamander graphic       Sweep samples __ Sweeping a large, small-mesh dip net through the water column or in submerged leaf litter in ponds or larger wetlands allows observers to capture amphibian larvae and sometimes breeding adults. Sample locations may be completely randomized or some measure of design can be incorporated into sampling, such as by sampling areas along pond margins every 10 or 15 m, depending on the circumference of the area to be sampled. Species richness, the number of larvae in each sweep, and the total number of sweeps are recorded.

       Example. Two persons search the entire circumference around Gourley Pond by sweeping a dip net five times every 15 m. If the pond margin is 600 m, then 40 locations could be sampled and 200 sweeps could be made. The sampling effort is 200 sweeps. Sample data might be: 240 larval A. opacum; 6 egg masses of A. maculatum; and, 1,246 larval R. sylvatica. The amount of area sampled in relation to available habitat could be estimated visually.

Figure 39. Schematic of a combined transect/coverboard survey design. - click to enlarge
Figure 39. Schematic of a combined transect/coverboard survey design. A series of eight coverboards are located at the origin and thereafter at 10-meter intervals along a 50-meter transect. In this design, the boards are placed perpendicular to the transect. This survey design could be combined with a night survey, whereby a team of observers walks along the transect, spotlighting and counting salamanders. The boards would not be disturbed during such survey. Using dual observers at close intervals helps quantify observer bias. See text for layout details.

       What this tells the observer. Sweep surveys provide information on: (1) larval species presence at the time of sampling; (2) life history information, such as when eggs are deposited and tadpole developmental stage; (3) habitat information, such as microhabitat preferences and distribution of various larvae; and, (4) in some cases, an estimate of density (number of animals ) the amount of area sampled in reference to available habitat). Sampling effort is easily quantified.

       Limitations. Detectability may be influenced by many of the factors listed in Things to Consider During Planning. Even if every attempt is made to standardize sampling, environmental factors (for example, water availability and depth; water temperature) likely will be different among sampling occasions and thus influence whether a species is observed. Since environmental variables influence the number of animals observed, differences in counts over time may be only reflective of differences in environmental conditions during sampling periods. As with time-constrained sampling, it is very difficult to determine any kind of trend based on periodic counts, because the relationship between counts and actual abundance is unknown. Also, the number of larvae observed may not reflect the number of breeding adults, or tell anything about future reproductive success and the rate of successful metamorphosis. For example, the wetland could dry 10 days after a sampling visit, and all larvae could perish.

frog graphic       Call surveys __ All species of male frogs in Great Smoky Mountains National Park call to establish breeding territories and attract females. Species that may be quite difficult to find throughout most of the year can be readily heard at this time, their breeding sites identified, and relative abundances of adult calling males estimated. Call surveys are easy to conduct. A biologist simply periodically visits wetlands or drives park roads at night during the breeding season and records the locations of species heard calling. In very large choruses, it may be necessary to record abundance in terms of categories: 1 = 0 frogs calling; 2 = 1 individual calling; 3 = < 5 individuals calling; 4 = > 5 to 10 individuals calling; 5 = > 10 individuals calling.

       Areas appropriate for call surveys within the Great Smokies include the Cades Cove Loop Road and associated roads in Cades Cove, the road through Cataloochee Valley, Laurel Creek Road, Little River Road, lowland areas of Newfound Gap Road at Sugarlands and Smokemont, Big Cove Road, and the entry roads to Greenbrier, Cosby, and Deep Creek. Two methods may be used: (1) drive slowly and listen for frog choruses, or (2) conduct systematic searches using periodic stops with defined amounts of time for listening.

       Example. Starting at the entry gate to Cades Cove Loop Road, drive slowly and stop every 0.5 miles. At each stop, turn off the engine, and listen for 5 minutes. Record the species heard and the compass direction from which the call is heard; possible breeding sites can be identified during daylight hours as time permits.

       What this tells the observer. Call surveys provide information on: (1) adult male presence at the time of sampling; (2) the dates and environmental conditions when males call; (3) the location of breeding sites; and (4) an estimate of breeding male relative abundance can be attained through the use of the abundance categories. Sampling effort is easily quantified.

       Limitations. Detectability may be influenced by many of the factors listed in Things to Consider During Planning. Even if every attempt is made to standardize sampling, environmental factors (for example, weather, temperature, rainfall patterns) likely will be different among sampling occasions and thus influence whether a species is heard. Since environmental variables influence the number of animals calling, differences among abundance categories over time may be only reflective of differences in environmental conditions during sampling periods. Thus, call surveys must be conducted at multiple occasions during the potential breeding season. Further, call surveys tell nothing about the presence and number of females and nonbreeding males, or whether reproduction was successful. Call surveys are best implemented where researchers have access by road; isolated breeding sites could be overlooked, or ignored when access is difficult (such as along lower Hazel and Eagle Creeks). Since frogs often call diurnally or during different intervals of the night (several hours after dusk or before dawn), species could be missed or relative abundances underestimated. One way to circumvent this problem is to use automated data loggers to periodically sample frog calls throughout the day and night.

       Egg mass or nest counts __ A number of amphibians (Spotted Salamander, Wood Frog) deposit globular egg masses that are readily identified and can be counted. Other species (Marbled Salamander, Four-toed Salamander) deposit eggs in terrestrial habitat on dry pond bottoms or in the vegetation bordering ponds. As the pond fills, the eggs are inundated and hatching occurs (Marbled Salamander) or the eggs hatch and larvae wiggle through the vegetation to reach the pond (Four-toed Salamander). Counting egg masses or nests should give an indication of reproduction during the sampling period. This method has been used in the Great Smokies by James Petranka and Charles Smith; Crouch and Paton (2000) have suggested that the method is an effective way to gage trends in Wood Frog population size and reproduction.

       Example 1. Researchers visit Gum Swamp shortly after Wood Frogs have bred. Each separate egg mass can be identified and a flag placed next to it. Flags mark the distribution of the egg masses, are easily counted, can be left in place to follow reproductive parameters (for example, whether successful hatching takes place), and help to reduce observer bias (single observers can miss 10 percent or more of the egg masses (Crouch and Paton, 2000)). Because each female deposits one mass, the number of breeding females at a pond can be monitored through time.

       Example 2. The dry pond basin at Gum Swamp can be searched in October when female A. opacum have deposited their eggs and are sitting over them until the autumn rains arrive. By carefully turning logs, researchers can locate nests, place flags in the ground adjacent to them, and obtain an idea of the number of nests and their spatial distribution. Numbers of females and males can be counted (see Dodd, 2004, for sex determination criteria).

     What this tells the observer. Egg mass or nest surveys provide information on: (1) the number of females breeding successfully in a year; (2) the dates and environmental conditions when eggs are deposited; and (3) egg masses that can be followed through time to obtain an idea of the extent of successful reproduction. Crude estimates of the number of metamorphs produced can be obtained (number of egg masses x the percentage of masses with successful hatching x the mean number of eggs per mass). In the case of nests, the reproductive potential (number of nests x the mean number of eggs per nest) can be determined. Sampling effort is easily quantified as the amount of time spent searching an area.

       Limitations. Counting egg masses assumes that there is one female per egg mass. This assumption seems to hold true for those species depositing large, globular, jelly masses. However, this assumption will not be valid for all species depositing eggs in nests (for example, the Four-toed Salamander) because nests may include the eggs of more than one female. Be sure to check information on life history (Dodd, 2004). Counting egg masses generally does not give an indication of the number of males or nonbreeding females (but see Crouch and Paton, 2000). Unless the hatching success of egg masses is recorded, counting egg masses will not provide an estimate of the number of metamorphs produced during the breeding season. Care must be taken not to disturb brooding females because nest abandonment virtually ensures reproductive failure. Although some species are more tolerant of disturbance than others, a nest should not be disturbed repeatedly.

Easy Passive Sampling

Figure 40. Coverboards. - click to enlarge
Figure 40. Coverboards.

       Coverboards __ Herpetologists have a long history of turning over surface cover objects to look for terrestrial salamanders and reptiles. Coverboards are simply an extension of this search technique, albeit with a more formalized sampling design. Coverboards may be made of many types of materials (for example, wood, tarpaper shingles, plastic sheets), but the most common material is nonchemically treated plywood. The boards are cut into small sizes (for example, 20 x 25 cm; 35 x 35 cm; fig. 40) and placed in a grid of various design. Boards should not be too large, because the leaf litter underneath them becomes dry in the center and discourages salamander residency. Pressuretreated boards should never be used.

       In the Great Smoky Mountains, National Park Service personnel have used four boards placed within a few centimeters of one another at each sampling site along a long transect. Sampling sites might be located at 10-m intervals along the transect, such that a 50-m transect would have 24 coverboards placed along it (stations 0-5 x 4 boards/station). Coverboards must be placed in location for at least a month prior to beginning a survey to ensure they age properly and provide secure hiding places. Ideally, coverboards should be set out in the autumn of the preceding year prior to sampling. Some researchers scrape the ground underneath coverboards to ensure that the area underneath is not too large to discourage residence or will not increase air flow. Coverboards should be checked once every week or two; too much disturbance will inhibit salamander occupancy.

       Example. In a study of sampling techniques on the north side of Mt. LeConte, Hyde and Simons (2001) used two sizes of coverboards (three 13 x 26 cm; two 26 x 26 cm) placed at 10-m intervals along a 50-m transect (5 boards x 5 sampling stations = 25 boards/ transect). Using a stratified sampling design to locate transect sites, they sampled 101 locations and captured 1,224 salamanders over a 2-year period. Coverboards were only checked three times the first year, and four times the second year.

       What this tells the observer. Coverboard surveys provide information on: (1) species presence at the time of sampling; (2) life history information, such as data on size-class structure, reproduction, and activity patterns; and (3) habitat information. If used in conjunction with mark-recapture techniques, they also might be used to examine site fidelity, movement, and population size. Sampling effort is easily quantified (number of coverboards x number of days sampled).

       Limitations. Capture probability is influenced by all the factors listed in Things to Consider During Planning. Even if every attempt is made to standardize sampling, environmental factors likely will be different and thus influence whether a species is observed. Because environmental variables influence the number of animals observed, differences in counts over time may be more reflective of differences in environmental conditions during the sampling periods among years than changes in status. It is very difficult to determine any kind of trend based on periodic counts, because it is unknown what the relationship is between the counts and actual abundance. Hyde and Simons (2001) found that counts of terrestrial salamanders in the Great Smokies were highly variable and that sampling variability and detectability were not constant among species or even habitat type. Recapture rates of marked salamanders also are notoriously low, making estimates of population size unreliable. Finally, coverboards may provide artificially favorable cover, although preliminary evidence suggests this capture bias may not be as serious as previously believed. Some size classes of terrestrial salamanders are more likely to use coverboards than other sizes (for example, data from Virginia suggest that hatchlings and juveniles are found less often under coverboards than they are under natural cover objects). Coverboards are labor intensive to cut and haul to a sampling site. They are subject to vandalism, and bears and pigs will readily turn them over or move them around.

       PVC pipes __ A method that has proved successful in the southeastern United States for monitoring treefrog (Hyla) populations is to place polyvinyl chloride (PVC) pipes in the ground or to mount them on trees (Boughton and others, 2000; http://cars.er.usgs.gov/posters/Herpetology/Artificial_Refugia/artificial_refugia.html). The pipes are readily colonized by treefrogs, even during the nonbreeding season when the treefrogs are dispersed away from ponds.

Figure 41. PVC pipes on trees in Okefenokee National Wildlife Refuge. - click to enlarge
Figure 41. PVC pipes on trees in Okefenokee National Wildlife Refuge.

The placement of the pipes and their characteristics (diameter, structure, possibly color) are important. Frogs are captured most often in pipes of 3.8 to 5.0 cm (1.75-2 inch) in diameter located 2- to 4-m high, on a large trunked, deciduous, hardwood tree; they are captured much less frequently in pipes on tree trunks near the ground, in pipes of larger diameter, or in pipes located on pine trees (fig. 41). Pipes capped on the bottom to allow some standing water within the shaft and presumably to increase humidity also capture more frogs than pipes that are open on both ends. Free-standing pipes (91.4 cm; 36 inches) sunk directly in the ground near breeding ponds also are used by treefrogs.

       Example. A series of PVC pipes are to be placed around Gourley Pond to monitor the population of Cope’s Gray Treefrog (Hyla chrysoscelis). Twenty transects are established evenly spaced around the pond perimeter at its edge (fig. 42). Each transect consists of five pairs of pipes (N = 10/transect; total N = 200 pipes) spaced 10 m apart, and radiates outward perpendicular to the pond’s edge, similar to the spokes of a wheel. The first two pairs are inground pipes, whereas the last three pairs are nailed to hardwood trees (if possible) at a 2-m height. Each pair of pipes consists of one 3.8- and one 5.0-cm pipe. The pipes on trees are fitted with bottom caps, with a hole drilled 9 cm above the base to allow drainage. Pipes are painted camouflage green on the outside for concealment, and each pipe is marked with a distinct number. Pipes are checked once a week from March through September.

Figure 42. Schematic of a survey design using paired PVC pipes located at 10-meter intervals around a ponds perimeter. - click to enlarge
Figure 42. Schematic of a survey design using paired PVC pipes located at 10-meter intervals around a ponds perimeter. The first set of pipes is located at the ponds margin, and thereafter at 5- or 10-meter intervals perpendicular to the pond. The second set of pipes is located at the margin of the pond basin (dashed line). The first two sets of pipes are ground pipes (black dots), whereas the last three (gold dots) are located (preferably) on large-diameter deciduous hardwoods. Pipes are placed at a height of 2 meters on opposite sides of the trunk (red dots).

The number of frogs observed is recorded. Frogs could be marked via individual or cohort toe clips, or digitally photographed for identification. Recording the data separately for unmarked animals and recaptures is important, because results from other studies show that frogs take up residency within pipes.

       What this tells the observer. PVC pipe surveys provide information on: (1) species presence at the time of sampling; (2) life history information, such as when animals arrive at breeding ponds, how long they stay, sex ratios, size-class structure; (3) movement patterns while at the ponds; and (4) information on the direction and distance of dispersal. Sampling effort is easily quantified (number of pipes x the number of 24-hour periods sampled).

       Limitations. The only species that can be monitored in the Park using PVC pipes is Cope’s Gray Treefrog. Even then, sampling results for this species have revealed mixed results at other locations where pipes have been used. In some areas, Cope’s Gray Treefrogs will use pipes as retreats, whereas in other areas they seem to avoid PVC pipes. Whether they will use PVC pipes in the Great Smokies is unknown. If simple presence data are needed, call surveys would be more appropriate, although PVC sampling might prove valuable if more detailed life-history information is required. PVC pipes are likely to be stolen or vandalized. Bears, in particular, seem to be attracted to PVC and will often bite it or carry pieces around.

Figure 43. Leaf litterbag in Little River. - click to enlarge
Figure 43. Leaf litterbag in Little River.

       Larval litterbags __ One relatively new method for inventorying and sampling most stream-dwelling salamanders, especially larvae, involves the use of artificial refugia (leaf litterbags) placed in shallow streams (fig. 43). In 2000, Waldron and others (2003) tested the utility of using litterbags to sample salamanders in Great Smoky Mountains National Park. Three transects of six litterbags each (two large, two medium, and two small) were placed in five small, medium, and large streams. A total of 690 larval, juvenile, and adult stream-dwelling salamanders from 11 species were captured from June to November in the 90 litterbags. Sampling salamanders in small streams was most productive using large and medium-sized litterbags, although all bag sizes worked equally well in medium and large streams. The number of salamanders captured varied seasonally, with most captures occurring in June and July. The depth of bag submergence significantly influenced litterbag use by adult and larval salamanders, but had no effect on use by juvenile salamanders. The ease of deployment and nondestructive sampling methodology suggest that litterbags could be useful in determining salamander presence during large-scale inventory programs, especially when the time available for sampling a large number of individual sites is limited and when sampling for secretive or uncommon larvae, such as Pseudotriton or Gyrinophilus.

       Example. Litterbags of two sizes (70 x 70 and 90 x 90 cm) are constructed as outlined in Waldron and others (2003). In the field, three or four small rocks are placed in the netting to give the bag weight, then covered with leaves. Once filled with leaf litter, the corners of the netting are pulled together and tied with plastic cable ties to form a bag. Blue flagging is tied to the top of each bag so that researchers can easily locate bags in the field. Precautions are taken to prevent the loss of bags from fast-flowing water and flooding by placing one or two large rocks against or just downstream from each bag, and by tethering each bag to the nearest root, log, or large rock using monofilament fishing line.

       Streams are selected using a stratified sampling protocol for size, location, and ease of access (see Sampling Streams). All streams are < 50 cm in depth at the sampling site. Sampling sites are spaced so that a watershed can be sampled in 1 day, allowing all of the sites to be completely sampled in 1 week. One 50-m transect is set up in each stream study area. Eight bags, four of each size class, are placed 10 m apart along transects. The order of presentation of medium and large bags from 0 to 50-m is randomized along the transect. Litterbags are sampled biweekly from April through September. Prior to sampling each litterbag, the percentage of litterbag submergence under water is recorded. Bags are removed quickly from the stream and gently shaken over a white dishpan for approximately 15 seconds to remove salamanders (fig. 44). Adult, juvenile, and larval salamanders that fall into the dishpan are identified to species, measured for total length (TL, tip of snout to end of the tail) and snout-to-vent-length (SVL, tip of snout to the posterior end of the cloacal opening), and released. If field identification is not possible, individuals are taken to the laboratory for identification, and later released into their respective streams.

Figure 44. Checking leaf litterbag at Little River. - click to enlarge
Figure 44. Checking leaf litterbag at Little River.

       What this tells the observer. Leaf-litterbag surveys provide information on: (1) species presence (but not absence) at the time of sampling; (2) life-history information, such as larval size and activity patterns; and (3) habitat information. Sampling effort is easily quantified.

       Limitations. Although the technique may be effective for determining the presence of many stream-dwelling salamander larvae in Great Smoky Mountains National Park, the variation in the numbers of individuals captured and the inability to relate captures to overall abundance make trends impossible to monitor without considerable additional effort, such as by employing mark-recapture techniques on, often, very small larvae. Capture may be influenced by the factors listed in Things to Consider During Planning. Even if every attempt is made to standardize sampling (for example, by sampling at the same streams during the sible, individuals same time of year), environmental factors, as well as natural variation in reproductive output, likely will be different among years and locations and thus influence whether a species is captured. Since environmental and other variables influence the number of animals captured, differences in counts over time may not reflect changes in status. Additionally, it is difficult to determine whether the bags are selected by adult and large larval salamanders as places of retreat or for foraging, and to determine the amount of area actually being sampled using the method.

       Automated frog call data loggers __ Automated data loggers have been used successfully to determine the presence of calling frogs at breeding sites (fig. 45). They can be set to record at variable time intervals for various amounts of time throughout the entire day, or they can be programed to record only at certain times of a 24-hour period, such as from dusk to dawn. Frog calls are easily discerned by listening to the tapes, and it is sometimes possible to gain an index of calling intensity, provided large choruses are not involved.

Figure 45. Recording data in field as storm approaches at Cataloochee Divide. - click to enlarge
Figure 45. Recording data in field as storm approaches at Cataloochee Divide.

       Example. At a pond the size of Gum Swamp, three data loggers could be installed to monitor chorusing frogs: one on the east shore, one on the west shore, and one on either the north or south shore midway between the other two. The program could be set to record for 5 minutes every hour throughout the day, or for 5 minutes only from dusk to dawn (the starting and ending times would vary with season to account for day length). Both sides of the tape can be used, thus extending the amount of time between tape changes. Data loggers measuring water and air temperature, and barometric pressure, could be placed near the call logger to account for environmental influences on calling activity.

       What this tells the observer. Automated frog call data loggers provide information on: (1) species presence at the time of sampling (species likely to be overlooked during time-constraint sampling can be recorded with greater reliability); (2) life history and phenology information, such as when frogs call (especially if different species call at different times of the day), what environmental influences affect calling; and (3) a relative index of the number of males calling.

       Limitations. Although species can be easily identified, categorizing abundance may be very difficult in even moderately sized choruses because of call-overlapping interference. It is also often not possible to separate individual callers, allowing the possibility that a single calling male could be counted multiple times. Since environmental variables influence the number of animals calling, differences among abundance categories over time may be only reflective of differences in environmental conditions during sampling periods. Thus, call surveys using automated data loggers must be conducted at multiple occasions during the potential breeding season. Further, call surveys tell nothing about the presence and number of females and nonbreeding males, or whether reproduction was successful. Frog call surveys using automated data loggers are best implemented where researchers have limited access by road (such as along lower Hazel and Eagle Creeks) or when rare species are suspected.

       Whereas automated frog call data loggers are relatively easy to assemble (appendix IV), they are somewhat expensive (about $350 in 2002). Unfortunately, there are no computer programs currently available that can identify calls and categorize abundance by reading the tapes. Thus, researchers must listen to tapes and manually record the results, a time-consuming, tedious exercise. At the Florida Integrated Science Center, two observers independently listen to the tapes as a measure to reduce and quantify observer bias. Automated data loggers must be well hidden to reduce theft and vandalism, and this can limit their effectiveness. Curious bears have been known to investigate and attempt to dismember the data loggers.

Intensive Passive Sampling

frog graphic       Traps (aquatic or terrestrial): funnels, bottles, minnow, wire basket __ Various types of aquatic traps have been used to sample amphibian larvae; on occasion, some of these traps have been used to capture adults, such as the Common Mudpuppy, in fine wire-mesh basket traps. They are all based on the premise that an animal entering the trap will be unable to escape because it would be difficult to exit through the inward-directed funnel opening. However, few studies have examined this assumption, and unhindered movement into or out of a trap (termed trespass) undoubtedly occurs with varying degrees of frequency. Minnow traps come in wire-mesh, collapsible soft, and plastic variations. Wire-mesh minnow traps seem to capture the most larvae, whereas plastic-mesh traps seem to have the least capture success. A drawback to wire-mesh traps is that they cause injury to tadpoles, even when checked every day, because the animals tend to beat themselves against the metal mesh attempting to escape. Wire-basket traps are usually larger with larger mesh, and are more often used to sample fishes and turtles than amphibians. In Florida, a modified crayfish trap with a fine mesh plastic insert is used to capture aquatic salamanders (Amphiumas, Sirens) (http://cars.er.usgs.gov/posters/Herpetology/Sirens_and_Amphiuma/sirens_and_amphiuma.html). The trap has not been tested specifically to capture amphibians in more temperate habitats. Wire-mesh screen funnel traps have been used for both aquatic and terrestrial sampling. These traps are placed flush with a downed log, rock, or drift fence. As the animal enters the trap, it falls to the center and, presumably, cannot find its way back out of the trap. None of these traps are baited, although larvae may attract invertebrate and vertebrate (turtles, snakes) predators.

       Example. Researchers place 15 wiremesh minnow traps around the perimeter and in the center of Big Cove Beaver Pond. Traps are spaced at about 5 m apart, secured to a branch to prevent loss, and placed in such a manner that trapped air-breathing animals have access to surface air. Traps are checked daily, perhaps even once in the morning and once at night. The number of animals caught are recorded by species, size, and developmental stage, then released. Sampling should only require a few days at each location, although a location may be trapped more than once per season to capture both early and late breeders. Sampling effort is easily quantified (number of traps x number of days = number of trap days).

       What this tells the observer. Funnel traps are used to detect a species’ presence, and perhaps to obtain a crude abundance estimate (that is, very large numbers of larvae versus very few larvae). Counts have little meaning except in this context. Funnel trapping is often used during mark-recapture studies, especially if there are no known capture biases (that is, trap avoidance or trap happiness). Traps might be useful in sampling for rare species.

       Limitations. Some types of traps require assembly, whereas others can be purchased ready-to-use directly from a supplier. They are subject to vandalism by both wildlife (bears, pigs) and people; minnow traps, in particular, may be stolen. Trapped animals are vulnerable to drowning, predation, and injury, making daily checking, preferably in the early morning, absolutely essential to minimize mortality. Traps capture nontarget organisms, such as invertebrates and fish. Even if every attempt is made to standardize sampling (for example, by sampling at the same exact location and during the same time of year), environmental factors likely will be different and thus influence whether a species is captured. It is very difficult to determine any kind of population trend based on periodic counts since it is unknown what the relationship is between the counts and actual abundance. Captures also may be biased by trap avoidance or trap happiness (that is, returning to a trap again and again because of the availability of food or shelter). It may be necessary to conduct a pilot study prior to employing trapping methods to determine sampling biases.

       Drift fences __ Drift fences are the most labor intensive method for sampling amphibians. In brief, the idea is to intercept an animal during its daily wanderings, direct it along a fence constructed of metal (galvanized or aluminum) or cloth (highway department silt cloth; plastic sheeting) to where it either falls into a pitfall trap (a bucket or can sunk flush with the ground surface) or funnel trap (wire-mesh screening with inward-directed funnels; once the animal gets inside the funnel, it should be difficult for it to escape). Sometimes buckets and funnels are used simultaneously. There are a number of different array configurations, but they usually take some form of a Y or X shape; each arm is 7.5-10 m long. Drift fences also can be used to completely encircle breeding ponds. Each sampling unit may consist of three or four arrays randomly placed in an area. In a region the size of the Great Smokies, dozens of arrays would be necessary to sample the terrestrial amphibian communities. Arrays should be opened at least four times per year for a minimum of 2 weeks per sampling period; at high elevations, the winter sampling period could be skipped. There are several excellent descriptions of the technique and various configurations, and the reader is referred to chapters in Vogt and Hine (1982) and in Heyer and others (1994) for more information.

       Example. Researchers decide to use a Y-shaped drift fence configuration to sample lowland, terrestrial amphibians in the Cades Cove region. Twenty sampling locations are randomly selected, and three arrays are placed at each location approximately 50 m from one another. The fence must be trenched so that animals cannot walk underneath the fence, and so that erosion does not create areas for underfence trespass. Pitfalls may not be feasible because of the rocky soils, so two funnel traps are placed on each side of a fence arm (that is, 12 per array). Funnel traps may need to be shaded to prevent desiccation of trapped animals and are placed flush with the base of the fence. Traps must be checked daily to avoid animal desiccation and minimize predation. The number of captured individuals of each species for each funnel trap is recorded. Animals are released at least a few meters away in appropriate cover to minimize chances of recapture. Funnel traps are opened and checked four times per year for a period of 2 weeks per sampling occasion to ensure that different amphibian faunas are sampled (that is, those species which are active during the cool versus the warm times of the year).

salamander graphic       What this tells the observer. Drift fence surveys provide information on: (1) species presence (but not absence) at the time of sampling; (2) life history information, such as population size-class structure, reproduction, and activity patterns; and (3) when used with mark-recapture techniques (toe-clipping, elastomer marking, photographic identification), to obtain a measure of abundance. A drift fence-pitfall-funnel trapping regimen might be useful in capturing rare species or, when completely encircling a breeding site, in measuring reproductive effort and success. Sampling effort is easily quantified (number of buckets or funnels x the number of nights over which the sampling was conducted = number of bucket-or trap-nights).

       Limitations. Drift fences take a great deal of work to install and maintain, even without digging holes for pitfalls and carrying heavy metal flashing to a study site. They are subject to vandalism by both wildlife (bears, pigs) and people; drift fence materials may also be stolen. Animals are very vulnerable in pitfalls and traps, making daily checking, preferably in the early morning, absolutely essential to minimize animal desiccation and predation from reptiles and small and large mammals. Pitfalls also capture large numbers of shrews which either eat the other animals present or die from stress.

       As previously mentioned, the probability of catching an animal is influenced by all the factors listed in Things to Consider During Planning. Even if every attempt is made to standardize sampling (for example, by sampling during the same time of year), environmental factors likely will be different and thus influence whether a species is captured. Since environmental variables influence the number of animals that are active, differences in captures over time may be more reflective of differences in environmental conditions among the yearly sampling periods than changes in status. It is very difficult to determine any kind of trend based on periodic counts since it is unknown what the relationship is between the counts and actual abundance, unless mark-recapture techniques are employed.

       Even with mark-recapture techniques, only a very small portion of the population may be sampled (for example, terrestrial plethodontids may be territorial and thus unlikely to move about very much), so it may be difficult to extrapolate estimates of abundance in a wide area where animals are patchily distributed. Recapture rates are notoriously low in most mark-recapture studies of terrestrial salamanders, making estimates of variance quite high and unacceptable. Many amphibians may not walk along a fence (treefrogs might just climb it, hop over, or just pass it by), enter a funnel, or fall into a pitfall; some amphibians may be readily able to crawl out of a pitfall. Little is known about capture biases, but data from other studies indicate that the color (Crawford and Kurta, 2000) and size of the bucket may influence capture; that some individuals learn to avoid buckets; and, that other individuals may come to recognize buckets as a source of shelter or food. Therefore, capture probabilities are likely to vary considerably among species, even if the species is locally abundant.

Data Handling

salamander graphic

 

clear pixel
clear pixel clear pixel

U.S. Department of the Interior
U.S. Geological Survey

clear pixel clear pixel


Home