Author: Sam Droege, USGS Patuxent Wildlife Research Center, 12100 Beech Forest Rd., Laurel, MD 20708, frog@usgs.gov, 301-497-5840.

Species list
Description of technique
       Field time components
       Office time
       Equipment
Things that could bias your counts
       Call saturation
       Call interference
       Weather
       Time of year/time of day
       Habitat change
       Observer effects
Advantages and disadvantages
Approaches to analyzing your data
Existing protocols and programs using this technique
Estimates of variation of counts for this technique
Literature cited
Send a comment on this technique (this takes you to another page)
See existing comments (this takes you to another page)

Note: Some western species of rana (Leopard Frogs) call underwater and may rarely or never call above water. Even for those that call above water the calls tend to have low carrying capacity. Thus calling surveys are largely inappropriate for these species. Species whose names are in BOLD print are species whose calling, while usually vigorous, most often occurs only for a day or two after heavy rains (usually in mid- to late-summer in the southwest).

Bufo alvarius, Colorado River Toad; Bufo americanus, American Toad; Bufo baxteri, Wyoming Toad; Bufo boreas, Western Toad; Bufo californicus, Arroyo Toad; Bufo canorus, Yosemite Toad; Bufo cognatus, Great Plains Toad; Bufo debilis, Green Toad; Bufo exsul, Black Toad; Bufo fowleri, Fowler's Toad; Bufo hemiophrys, Canadian Toad; Bufo houstonensis, Houston Toad; Bufo marinus, Marine Toad; Bufo microscaphus, Arizona Toad; Bufo nelsoni, Amargosa Toad; Bufo punctatus, Red-spotted Toad; Bufo quercicus, Oak Toad; Bufo retiformis, Sonoran Green Toad; Bufo speciosus, Texas Toad; Bufo terrestris, Southern Toad; Bufo valliceps, Gulf Coast Toad; Bufo velatus, East Texas Toad; Bufo woodhousii, Woodhouse's Toad; Acris crepitans, Northern Cricket Frog; Acris gryllus, Southern Cricket Frog (Florida and Coastal); Hyla andersonii, Pine Barrens Treefrog; Hyla arenicolor, Canyon Treefrog; Hyla avivoca, Bird-voiced Treefrog; Hyla chrysoscelis, Cope's Gray Treefrog; Hyla cinerea, Green Treefrog; Hyla femoralis, Pine Woods Treefrog; Hyla gratiosa, Barking Treefrog; Hyla squirella, Squirrel Treefrog; Hyla versicolor, Gray Treefrog; Hyla wrightorum (eximia), Madrean Treefrog; Osteopilus septentrionalis, Cuban Treefrog; Pseudacris brachyphona, Mountain Chorus Frog; Pseudacris brimleyi, Brimley's Chorus Frog; Pseudacris cadaverina, California Treefrog; Pseudacris clarkii, Spotted Chorus Frog; Pseudacris crucifer, Spring Peeper; Pseudacris feriarum, Southeastern Chorus Frog; Pseudacris maculata, Boreal Chorus Frog; Pseudacris nigrita, Southern Chorus Frog; Pseudacris ocularis, Little Grass Frog; Pseudacris ornata, Ornate Chorus Frog; Pseudacris regilla, Pacific Treefrog; Pseudacris streckeri, Strecker's Chorus Frog; Pseudacris triseriata, Western Chorus Frog; Pternohyla fodiens, Lowland Burrowing Treefrog; Smilisca baudinii, Mexican Smilisca; Eleutherodactylus augusti, Barking Frog; Eleutherodactylus coqui, Coqui; Eleutherodactylus cystignathoides, Rio Grande Chirping Frog; Eleutherodactylus guttilatus, Spotted Chirping Frog; Eleutherodactylus marnockii, Cliff Chirping Frog; Eleutherodactylus martinicensis, Martinique Greenhouse Frog; Eleutherodactylus planirostris, Greenhouse Frog; Leptodactylus labialis, Mexican White-lipped Frog; Gastrophryne carolinensis, Eastern Narrow-mouthed Toad; Gastrophryne olivacea, Western Narrow-mouthed Toad; Hypopachus variolosus, Sheep Frog; Scaphiopus couchii, Couch's Spadefoot; Scaphiopus holbrookii, Eastern Spadefoot; Scaphiopus hurterii, Hurter's Spadefoot; Spea bombifrons, Plains Spadefoot; Spea hammondii, Western Spadefoot; Spea intermontana, Great Basin Spadefoot; Spea multiplicata, Mexican Spadefoot; Rana areolata, Crawfish Frog; Rana berlandieri, Rio Grande Leopard Frog; Rana blairi, Plains Leopard Frog; Rana boylii, Foothill Yellow-legged Frog; Rana capito, Gopher Frog; Rana cascadae, Cascades Frog; Rana catesbeiana, American Bullfrog; Rana chiricahuensis, Chiricahua Leopard Frog; Rana clamitans, Green Frog; Rana draytonii, California Red-legged Frog; Rana grylio, Pig Frog; Rana heckscheri, River Frog; Rana luteiventris, Columbia Spotted Frog; Rana muscosa, Mountain Yellow-legged Frog; Rana okaloosae, Florida Bog Frog; Rana onca, Relict Leopard Frog; Rana palustris, Pickerel Frog; Rana pipiens, Northern Leopard Frog; Rana pretiosa, Oregon Spotted Frog; Rana rugosa, Wrinkled Frog; Rana septentrionalis, Mink Frog; Rana sevosa, Dusky Gopher Frog; Rana sphenocephala, Southern Leopard Frog; Rana subaquavocalis, Ramsey Canyon Leopard Frog; Rana sylvatica, Wood Frog; Rana tarahumarae, Tarahumara Frog; Rana virgatipes, Carpenter Frog; Rana yavapaiensis, Lowland Leopard Frog; Rhinophrynus dorsalis, Burrowing Toad

Calling surveys of amphibians are designed to provide an index to changes in amphibian populations. The technique can be applied in monitoring sampling designs to provide estimates of abundance and change at scales ranging from continental down to single wetlands. A powerful attribute of a uniform, randomly allocated system of calling surveys is that they can be repartitioned in many meaningful ways. Estimates of trends from the same system of surveys can be developed for watershed, political, physiological, or survey-wide units of scale, dependent upon research and management requirements, in a way similar to the North American Breeding Bird Survey (Droege, 1990). By correlating observed trends and distribution patterns with ancillary variables, it is also possible to gain insight into the factors affecting population changes over time (e.g., declines may be associated with agricultural regions; increases with regions where wetlands have increased).

Most calling survey protocols in North America use a simple index to population size based on the one developed for the Wisconsin Frog and Toad Survey (for details see Mossman et al., 1998; Weir and Mossman, this volume).

The Wisconsin Index:
0 = no frogs, of a given species, can be heard calling;
1 = individual calls, not overlapping;
2 = calls are overlapping; but individuals are still distinguishable;
3 = numerous frogs can be heard; chorus is constant and overlapping.

Surveys are usually run three to four times per year in synchrony with the mating and peak calling seasons of the local species. Surveys are often put together as a series of stops (a "route") at wetlands along rural roads. Starting time is 30 - 60 minutes after sunset, and a route typically takes about two hours to complete. To reduce disturbance effects, an observer usually waits one minute at each wetland stop, then starts a three-five minute listening period. The observer records which species were heard along with a calling index value.

• Record data (3-5 minutes per stop)
• Weather and ancillary information (2-5 minutes per stop)
• Travel between stops
• Travel to start of route
• Return trip

• Data entry
• Editing and cross checks of entered data
• Analysis
• Report writing

• Hard copy of your protocol
• Data sheet for each visit (use a clipboard covered with plastic or photocopied on water-resistant paper)
• Pencil or indelible ink pen
• One flashlight per person
• Extra flashlight (just in case)
• Thermometer
• Watch
• Rain gear
• Tape recorder - in case you want to compare the sounds at your site with the audio recordings you have obtained

Things that could bias your counts

Calling saturation is a bias linked to changing detection rates with different densities of calling amphibians. As the number of individual frogs or toads calling increases at a site, the ability of the human observer to differentiate or count individuals rapidly declines (as anyone experiencing the painful impact of a large chorus of spring peepers can attest to). The calling index scale acknowledges this human limitation and uses only 3 intensity levels. Here the true population could range broadly, yet the index peaks at intensity level three and stays constant from then on. Thus, over time, the population could fluctuate widely, but the index has no ability to discern as long as most counts remain above the threshold where a level two changes to a level three. At lower population levels, the index tracks the population change, but only as far as a scale of 0 - 3 permits.

The reality of most calling surveys is that it is never the case that at a series of survey locations all points are saturated with detections of a species at the highest calling level (level 3). Rarely is it even the case that a species is even detected at all the points on a night's round of surveys. A conservative approach to the analysis of calling frog and toad data is therefore to collapse all the calling categories into a simple detected/not detected category. This does not eliminate bias in detections associated with population size (i.e., large populations are likely to be more detectable on an average night than small ones), but it makes the analysis more understandable. For example, it is easier to understand what changes in the frequency of points with green frogs (Rana clamitans) are over time than it is in the changes to a set of index values for green frogs taken at the same point. Frequency of points are clearly related to the number of wetlands occupied by green frogs, while changes in the index are harder to interpret. This is because it is not exactly clear how an index of 1, 2, or 3 relates to the real number of green frogs out there, nor what would be the best means of combining index scores across several wetlands.

The practical consequences of using a simple 1 or 0 index (and acknowledging that large populations are more detectable than small ones) will be an under-estimation of the rate of decline. That is, the slope will be correctly determined to be negative or flat, but it could be shallower than what is actually occurring. Thus a monitoring program based on calling surveys is likely (analyzed either with the index or as presence absence) to sound the alert to declines late. Consequently declines detected using calling surveys would be of greater concern as the real rate of decline would be under-reported.

A less clear problem is call obfuscation, occurring when one species' call interferes with the detection of another, quieter species. Bias could potentially be severe enough that a population increase in the quiet species could be misidentified as a decline, if the louder species (population also increasing) were to overwhelm the calls. It is unclear how many stops in a set of surveys would ever be affected by such a situation. An evaluation of regional chorus timing and loudness would be simple to conduct.

Most calling survey protocols minimize the biasing effects of weather on sampling by providing guidelines for appropriate sampling conditions, with prohibitions on sampling in nights with strong wind or heavy rainfall. Despite the guidelines however, weather induced variation exists, despite the guidelines, as droughts and wetter than normal years affect the number and rate of frogs calling. Consequently, short-term comparisons (2 - 5 years) of population counts are difficult to interpret, because weather effects on detectability and their impact on population fluctuations are intertwined. Over longer periods of time, the effects of weather diminish and simply become extra noise in the system, not a bias.

Another source of variation and potential bias in calling surveys is the onset of calling behavior. Advent of the calling season for a frog or toad begins earlier in warm years (often associated with large rainfalls), at times by several weeks (Mossman et al., 1998). To mitigate that effect, the sampling windows of regional programs are often based on temperature cues. In addition, monitoring programs in the U.S. and Canada have started to track the amphibian calling phenology. The U.S. program is called Frogwatch USA and is based on three successful programs in Canada. The goals of the Frogwatch USA program are: documenting the yearly phenological pattern of calls, determining amphibian population changes at individual wetland sites, and educating the public about amphibians.

Biasing factors can also creep in over extended periods of time if landscape, observers, or other initial conditions change and affect the detection of amphibians on calling surveys. Forests, ponds, fields, human development, and wetlands change with time in their dominance, succession, and interspersion. As these habitats change, so too does the mix of amphibian populations. Some species are favored and others not. These are the types of changes we would expect a monitoring program to track. However, another consequence of these changes may be shifts in the acoustical aspects of the landscape. Calling populations in some areas may become more detectable and others less as vegetation impedes or promotes sound (Varhegyi et al., 1998). Any large and uncorrected shift in the detectability of calls will affect (i.e., bias) the resulting calculated trends. It may be possible to correct such a bias by documenting habitat changes along routes or, more likely, using groups of years as co-variables (to diminish the effects of long-term habitat changes). Fortunately, such landscape changes occur over decades, providing some lead-time in determining and developing a correction, if warranted.

Observer effects include bias and errors caused by the survey participants themselves, factors that may be caused by hearing losses or identification errors. It could be argued that if you have many observers participating that relative hearing levels would be randomly distributed throughout that population yielding extra noise in the system but not bias. However, many researchers now statistically factor out the differences among observers in their analyses of trend (e.g., they control for the folks who are over-estimators and those who are underestimators) by making observer co-variables. This reduces the noise in the data caused by differences in how people count, but assumes that observer's hearing does not change over time (which it can and ultimately does). Thus a negative bias is built into the system (because observers hearing deteriorates as they age, possibly resulting in their recording relatively fewer frogs and toads over time). Hearing loss in observers can be dealt with in several ways. The best way would be to test observer's hearing and to impose strict guidelines for participation based on the results of those tests.

Most existing amphibian calling surveys use volunteer observers to collect survey data. The use of non-biologists to gather scientific data has been criticized. It should be noted that all observers, whether volunteers or not, have the potential for adding both bias and unwanted variance to counts of calling frogs. At issue here, however, is what additional bias or variance volunteers and laypeople bring to a system over and above the paid technician.

Based on a small sample survey of participants in the Wisconsin Frog and Toad survey, most observers reported that they had no or minimal prior experiences with amphibian identification (Mossman et al., 1998). While demographics in most other regions are unstudied, there is reason to suspect that calling surveys for amphibians will be more attractive to non-professional biologists than are bird surveys. Amphibian vocalizations are simpler and there are fewer species in any given region (roughly 7 - 30 for amphibians versus 100 - 225 for birds).

Three independent studies (Shirose et al. 1997; Hemeseth, 1998; and Kline, 1998) have investigated the ability of previously untrained lay-observers to learn and identify calling frogs. In Kline's (1998) study, students in grades 5 - 8 (note that at this point almost no students in this age group are participants in calling surveys in most states/provinces) listened to a training tape of calling amphibians and went out on one practice run prior to being tested. Groups of students were sent to a set of ponds during a 5-day interval and asked to record species and calling intensity level. The results demonstrated that variation among observers was less than the variation that occurred from day to day. The ability to detect and distinguish species was not reported. Hemeseth's (1998) study demonstrated that variance in species identification among observers was low, but that assignment of calling intensity showed high variance for some species. Shirose (1997) demonstrated similar results in Ontario, with no significant differences between raw recruits and professionals in species identification, but variation in assignment of calling intensity between both participant groups.

Volunteers can be expected to have longer tenure on routes than technicians, decreasing observer turnover and consequently improving the statistical adjustment of counts for observers (Sauer et al., 1994; Mossman et al., 1998). Note too that many technicians start on the job at fundamentally the same place that the volunteer does - that is, with no prior knowledge of the calls of amphibians. Finally, it is unlikely that most of the existing monitoring programs could afford to pay technicians to run surveys. With training, tests, and the vetting of the resulting data, the quality assurance of volunteer collected data can be safeguarded. An Internet-based amphibian call identification quiz is being developed by the USGS, which will help volunteer improve their identification skills and help us assess volunteer identification abilities.

Advantages:

• Scalable: The technique can be applied in monitoring sampling designs to provide estimates of abundance and change at scales ranging from continental down to single wetlands. A powerful attribute of a uniform, randomly allocated system of calling surveys is that they can be repartitioned in many meaningful ways. Estimates of trends from the same system of surveys can be developed for watershed, political, physiological, or survey-wide units of scale, dependent upon research and management requirements, in a way similar to the North American Breeding Bird Survey (Droege, 1990).
• Insight into reasons for population change: By correlating observed trends and distribution patterns with ancillary variables, it is also possible to gain insight into the factors affecting population changes over time (e.g., declines may be associated with agricultural regions; increases with regions where wetlands have increased).

Once you've used this technique to gather data, decide which approach to use to analyze your data.

• NAAMP (North American Amphibian Monitoring Program)
• Frogwatch USA
• Frogwatch Canada

Amphibian CV database

Droege, S. 1990. The North American Breeding Bird Survey. Pp. 1-4. In Sauer, J.R. and S. Droege (Eds.), Survey Designs and Statistical Methods for the Estimation of Avian Population Trends. U.S. Fish and Wildlife Service Biological Report 90, Washington, D.C.

Hemeseth, L.M. 1998. Iowa's Frog and Toad Survey, 1991-1994. Pp. 206-216. In Lannoo, M.J. (Ed.), Status and Conservation of Midwestern amphibians. University of Iowa Press, Iowa City, Iowa.

Kline, J. 1998. Monitoring amphibians in created and restored wetlands. Pp. 360-368. In Lannoo, M.J. (Ed.), Status and Conservation of Midwestern amphibians. University of Iowa Press, Iowa City, Iowa.

Mossman, M.J., L.M. Hartman, R. Hay, J.R. Sauer, J.R. and B.J. Dhuey. 1998. Monitoring long-term trends in Wisconsin frog and toad populations. Pp. 169-198. In Lannoo, M.J. (Ed.), Status and Conservation of Midwestern Amphibians. University of Iowa Press, Iowa City, Iowa.

Sauer, J.R., B.G. Peterjohn and W.A. Link. 1994. Observer differences in the North American Breeding Bird Survey. Auk 111:50-62.

Shirose, L.J., C.A. Bishop, D.M. Green, and others. 1997. Validation tests of an amphibian call count survey technique in Ontario, Canada. Herpetologica 53:312-320.

Varhegyi, G., S.M. Mavroidis, B.M. Walton, C.A. Conaway and R.A. Gibson. 1998. Amphibian surveys in the Cuyahoga Valley National Recreation Area. Pp. 137-154. In Lannoo, M.J. (Ed.), Status and Conservation of Midwestern Amphibians. University of Iowa Press, Iowa City, Iowa.