Ambystoma macrodactylum Baird, 1849
David S. Pilliod
Julie A. Fronzuto
1. Historical versus Current Distribution. Historical and current distributions
of long-toed salamanders (Ambystoma macrodactylum) are similar; there
is no evidence of a change in distribution. The type-locality for this common,
broadly distributed northwestern species is Astoria, Oregon (Baird, 1849), with the
syntype deposited at the U.S. National Museum 4042 (Ferguson, 1963). Populations
occur from the Alaskan Peninsula across British Columbia; south through Washington into
Oregon and the Sierra Nevada of California; and across the Rocky Mountains into eastern
Alberta, western Montana, and central Idaho. Isolated populations occur in central
California (Russell and Anderson, 1956) and in southeastern Oregon (Petranka,
1998). An erroneous record exists from Iowa (Ruthven, 1912). There are five
subspecies recognized, based primarily on morphologically distinguishable dorsal banding
patterns (Mittleman, 1948; Ferguson, 1961a; Crother et al., 2000). Generally, each
subspecies is distributed allotopically along north–south ranges (see Petranka,
1998) as follows:
· Western long-toed salamanders (A. m.
macrodactylum; Baird, 1849) range from Vancouver Island and southwestern British
Columbia through western Washington (west of the Cascade Mountains) and north of the
Calapooya Divide in western Oregon (Nussbaum et al., 1983).
· Eastern (also cited as central or Columbia) long-toed
salamanders (A. m. columbianum; Ferguson, 1961a) range from
southeastern Alaska through British Columbia, eastern Washington, Oregon (east of the
Willamette Valley), and into central Idaho (Nussbaum et al., 1983; Petranka, 1998).
· Northern (also erroneously cited as eastern or western)
long-toed salamanders (A. m. krausei, Peters, 1882) range from
eastern British Columbia and Idaho through western Alberta and Montana (Nussbaum et al.,
1983; Petranka, 1998; Walsh, 1998; Graham and Powell, 1999).
· Southern long-toed salamanders (A. m.
sigillatum; Ferguson, 1961a) range from southwestern Oregon (south of Calapooya
Divide) into the Sierra Nevada as far south as Carson Pass in California (R. Cutter,
· Santa Cruz long-toed salamanders (A. m.
croceum; Russell and Anderson, 1956) represent an isolated subspecies found at
11 locations in Santa Cruz and Monterey counties, California (U.S.F.W.S., 1999c).
Listed in 1967, this is the only Federally Endangered subspecies of A.
macrodactylum (Bury et al., 1980). Recent efforts to halt habitat threats
to this narrowly distributed population are being evaluated (Ruth, 1988; U.S.F.W.S.,
2. Historical versus Current Abundance. Little information exists regarding the
historical abundance of long-toed salamanders; it is therefore difficult to determine if
abundance has changed over time. In addition, determining the abundance of adults
in an area is usually difficult. Even when populations are large, long-toed
salamanders are rarely found outside of the breeding season. When encountered in
ponds during spring breeding aggregations, long-toed salamanders have been considered
locally abundant (Anderson, 1967a; Beneski et al., 1986; Powell et al., 1997; Fukumoto
and Herrero, 1998). Historical descriptions from Crater Lake, Oregon (Evermann,
1897; Bishop, 1943), Corvallis, Oregon (Storm and Pimentel, 1954), and Mt. Rainier,
Washington (Slater, 1936b; Stebbins, 1966), indicate long-toed salamanders were abundant
in those areas. In northwestern Idaho, over 2,030 adults were captured by a drift
fence encompassing a small (0.3 ha), fishless pond, providing habitat for an estimated
3,141 breeding adults (Beneski et al., 1986). Recent mark-recapture studies in
Alberta found similarly large population sizes (Powell et al., 1997; Fukumoto and
Herrero, 1998). However, the effective population sizes (estimated from allozyme
data) of six high elevation populations in Idaho and Montana were considerably smaller
(mean = 123; Funk et al., 1999). A five-year census of 11 high elevation basins in
central Idaho indicated that long-toed salamanders may have been extirpated or reduced to
low numbers in six of these basins, possibly due to the stocking of non-native trout into
deep, fishless breeding habitats (Pilliod and Peterson, 2001).
3. Life History Features.
Reproduction is aquatic.
i. Breeding migrations. In the Pacific Northwest, long-toed salamanders are the
earliest breeding amphibians (Leonard et al., 1993; Corkran and Thoms, 1996), often
migrating across snow and depositing eggs before complete ice-out. In the
Willamette Valley, Oregon, adults migrate to breeding ponds in late October to early
November (Stebbins, 1966; Nussbaum et al., 1983; Leonard et al., 1993), and as late as
June–July at higher elevations in the Cascades, Rockies, Sierra Nevada, and
Wallowas (Kezer and Farner, 1955; Stebbins, 1966; Howard and Wallace, 1985; Leonard et
al., 1993; Walls et al., 1993a; Pilliod, 2001; Thompson, 2001). Males are the first
to arrive at breeding ponds (Nussbaum et al., 1983; Beneski et al., 1986), probably to
court arriving females (Slater, 1936a; Knudsen, 1960; Anderson, 1961) and compete with
other males (Verrell and Pelton, 1996). While females, as a group, spend
approximately 3 wk at a breeding site (Beneski et al., 1986), depositing eggs over a
6–7 d period (Anderson, 1967a), individual females spend only 1–2 d (Verrell
and Pelton, 1996). Males generally leave the breeding site ~ 1 wk after the females
(Beneski et al., 1986), but may remain in the ponds for the entire breeding season (up to
2 mo or more; Anderson, 1968; Beneski et al., 1986). In ephemeral habitats,
breeding may only last 1–2 d (Walls et al., 1993a).
ii. Breeding habitat. Long-toed salamanders are opportunistic breeders, depositing
eggs in a variety of habitats, including seeps (Hamilton et al., 1998); along the
backwaters of slow-flowing streams (Beneski et al., 1986; Hamilton et al., 1998; Llewelyn
and Peterson, 1998), temporary pools at lower elevations (Leonard and Klaus, 1994); and
small–large permanent lakes and ponds at higher elevations (Anderson, 1967a; Howard
and Wallace, 1985; Leonard et al., 1993; Hamilton et al., 1998; Pilliod, 2001) and
higher latitudes (Green and Campbell, 1984). Eggs and larvae have also been found
in disturbed areas, such as newly formed (Hamilton et al., 1998), recently disturbed
(Corkran and Thoms, 1996), and human-influenced (Beneski et al., 1986; Llewelyn and
Peterson, 1998; Monello and Wright, 1999) pools down to the size of tire ruts (K.R.
McAllister, personal communication).
i. Egg deposition sites. Eggs are deposited in shallow water (< 0.5 m) with
silt-mud substrates, but also along rocky shorelines (Hamilton et al., 1998). Eggs
are attached to vegetation, floating or submerged woody debris (logs, branches), and
rocks, or placed unattached on the bottom in shallow (< 20 cm) water (Slater, 1936a,b;
Stebbins, 1954a; Nussbaum et al., 1983; Howard and Wallace, 1985; Corkran and Thoms,
ii. Clutch size. Clutch size is geographically variable, from 90–411 eggs
(Slater, 1936b; Gordon, 1939; Ferguson, 1961a; Anderson, 1967a; Howard and Wallace, 1985;
see also Petranka, 1998). Females deposit eggs over several hr, releasing
1–81 eggs in a cluster before moving to a new location in the pond (Petranka, 1998;
i. Length of larval stage. Length of larval period varies with elevation,
latitude, and pool permanence (Slater, 1936b; Kezer and Farner, 1955; Anderson, 1967a;
Howard and Wallace, 1985; Watson, 1997). Eggs hatch in 5–35 d, depending on
water temperature (Anderson, 1967a; Leonard et al., 1993). The larval period can be
as short as 50 d in some temporary ponds (Nussbaum et al., 1983) or last ≤ 3 yr in
permanent lakes at higher elevations (Bishop, 1943; Stebbins, 1966; Leonard et al., 1993;
ii. Larval requirements.
a. Food. Larvae are opportunistic carnivores and begin feeding shortly after
hatching (Petranka, 1998). Prey size generally increases with salamander body size
and includes crustaceans (amphipods, cladocerans, copepods), a variety of aquatic and
terrestrial insects (coleopterans, dipterans, ephemeropterans, plecopterans,
tricopterans), mollusks (gastropodans, pelecypodans), annelids (hirudineans,
oligochaetes), and ranid frog tadpoles (Anderson, 1968; Tyler et al., 1998a).
Larger larvae may cannibalize smaller larvae (Anderson, 1967a; Walls et al., 1993a),
possibly resulting in increased growth and size at metamorphosis of the cannibals (Wildy
et al., 1998, 1999).
b. Cover. Because of their diverse diets, feeding larvae are found in the open
water column and within cover. During the day, larvae may use cover to avoid
predation from vertebrate and invertebrate predators and are often found in or under
bottom detritus (rotting leaves, woody debris), submerged logs, rocks, and aquatic
vegetation (Anderson, 1967a; Green and Campbell, 1984; Liss et al., 1995; Corkran and
Thoms, 1996; Munger et al., 1997b; Hamilton et al., 1998; Petranka, 1998). In
fishless lakes, larvae may move more freely across open substrates or in the water column
(Liss et al., 1995; Tyler et al., 1998a). However, in fishless ponds in southeastern
Washington that contain a variety of other vertebrate and invertebrate predators, larvae
are seldom observed in open water and even less often captured in minnow traps (J.A.F.
and P. Verrell, unpublished data). At high elevations, second-year larvae may use
more open habitat compared to first-year and metamorphosing larvae that remain under
cover (Anderson, 1967a).
iii. Larval polymorphism. Morphologically distinct cannibalistic larvae have been
reported from at least one small, subalpine pond in Oregon (Walls et al., 1993a,b;
Petranka, 1998). These cannibalistic larvae have longer, wider heads and larger
vomerine teeth compared to conspecifics of the same size and from the same population
that were reared in the laboratory on a diet of live Tubifex (Walls et al.,
1993a). Diet is one of several intrinsic and extrinsic factors that influence the
expression of the cannibalistic larval morphology (Walls et al., 1993b).
iv. Features of metamorphosis. The timing of metamorphosis varies with
environmental conditions and is triggered by either intrinsic factors (possibly size of
the animal) or extrinsic factors such as temperature and pond drying (Anderson,
1967a). Size at metamorphosis is highly variable, ranging from 23–48 mm SVL
(40.5–90 mm TL; Carl, 1942; Howard and Wallace, 1985). Although size at
metamorphosis does not appear to be associated with elevation (Anderson, 1967a), larvae
that take 2–3 yr to transform are generally larger at metamorphosis (Kezer and
Farner, 1955; Howard and Wallace, 1985).
v. Post-metamorphic migrations. Long-toed salamanders exhibit a strong breeding
site fidelity and generally will only migrate within 100 m of breeding ponds (Anderson,
1967a; Sheppard, 1977; Powell et al., 1997). However, outside of the breeding
season, longer migrations may occur. For example, in Montana, adults have been
captured in pitfall traps at least 600 m from the nearest breeding site (J. Pierson,
unpublished data, as cited in Maxell, 2000). Terrestrial post-metamorphic
migrations are generally associated with rains, high soil moisture, and air temperatures
above 0 ˚C (Anderson, 1967a; Howard and Wallace, 1985; Beneski et al., 1986; Powell
et al., 1997; Fukumoto and Herrero, 1998). When moisture levels are sufficient,
metamorphic animals disperse away from breeding sites shortly after transforming
(Anderson, 1967a). Dispersal of newly metamorphosed animals may be spread out over
several months (May–August in southeastern Washington; J.A.F. and P. Verrell,
unpublished data) or occur as a mass migration when conditions permit (Anderson, 1967a;
vi. Neoteny. Although the prevalence of delayed metamorphosis in long-toed
salamanders indicates a potential for neoteny (Sprules, 1974a),
paedomorphosis—retention of larval characteristics in reproductively active
adults—has not been observed.
Habitat. As far as we know, juvenile habitats are similar to adults. Recently
transformed juveniles may remain close to the breeding pond (under cover objects) until
conditions for migration are favorable (Anderson, 1967a).
Habitat. Adult long-toed salamanders occur in a wide range of habitats ranging from
sea level to 3,000 m in California (Stebbins, 1966) to ≤ 2,030 m in Washington
(Leonard et al., 1993) to 2,470 m in Oregon (Howard and Wallace, 1981, 1985) and 2,725 m
in Idaho (Munger et al., 1997b; Pilliod, 2001). Suitable habitats include semiarid
grasslands and sagebrush steppes, alpine meadows, dry oak woodlands, humid coniferous
forests, rocky shorelines of subalpine lakes, beaver ponds (Reichel, 1996), and even
disturbed agricultural areas (Nussbaum et al., 1983; Monello and Wright, 1999), timber
harvest areas (Hamilton et al., 1998; Naughton et al., 2000), pastures (Leonard et al.,
1993), and residential green belts (Leonard et al., 1993). Adults are typically
subterranean outside of the breeding season, hiding under logs, bark, rocks, and within
rotten wood or rodent burrows, generally within 100 m of water (Gordon, 1939; Bishop,
1943; Stebbins, 1954a; Stebbins, 1966; Green and Campbell, 1984; Corkran and Thoms, 1996;
Powell et al., 1997). In a comprehensive search conducted in August, of all cover
objects within 10 m of the shoreline of three high elevation lakes in central Idaho, the
majority (82%) of adults were found under logs (5–50 cm diameter) within 5 m of
water (D.S.P. and M. Reed, unpublished data). Microhabitats are typically
associated with higher substrate moisture (Anderson, 1967a). When soil moisture
levels are low, juveniles and adults may aggregate (≤ 43 individuals have been
observed in close proximity) and entwine (Anderson, 1967a), a behavior that reduces water
loss (Alvarado, 1967). Adults can be found above ground at night or during rains
in the summer months not associated with the breeding season. Migrations do not
necessarily occur along stream corridors or within obvious habitat types (Beneski et al.,
F. Home Range
Size. Due to the limited vagility of long-toed salamanders, home range sizes are
relatively small. Sheppard (1977) monitored 25 salamanders with implanted
radioactive tags from July–October in Alberta and estimated (minimum-area convex
polygon) home ranges to be 115.6 m2 for females, 167.5 m2 for males, and 281.6
m2 for juveniles. The rugged topography in which long-toed salamanders are
often found may reduce movements between distant populations, potentially reducing gene
flow among populations (Howard and Wallace, 1981). However, in the Bitterroot
Mountains in Montana, allozyme data indicate populations within basins are panmictic,
with salamanders moving among breeding populations (Tallmon et al., 2000). These
allozyme data also suggest that salamanders move across mountain ridges more frequently
than they move across valley bottoms. In Idaho, a salamander was observed in June
crossing a snow-covered ridge that separated two cirque basins at an elevation of 2,600 m
(D.S.P., personal observations).
Although larvae are not known to be territorial, they are aggressive, possibly resulting
in the spacing of individuals (Anderson, 1967a). Both juveniles and adults can be
aggressive in competition over food (J.A.F., personal observations). Non-breeding
adults may be social, rather than territorial—conspecifics aggregate rather than
exclude each other Verrell and Davis (2003).
Aestivation/Avoiding Dessication. Aestivation has not been documented, but low
precipitation may inhibit migratory behavior and result in reduced surface activity
(Howard and Wallace, 1985).
Migrations. Long-toed salamanders migrate to breeding habitats in the spring and to
overwintering habitats in the fall. Individuals home to breeding and wintering
locations, but do not appear to follow population-level migratory routes (Beneski et al.,
(Hibernation). Although little is known about long-toed salamander hibernation,
adults probably hibernate terrestrially. Sheppard (1977) found three terrestrial
hibernacula aggregations located in gravel substrate at 45–70 cm below the ground
surface (frost line was estimated at 45 cm). When water temperatures drop and
surface ice forms, over-wintering larvae become less active and retreat under logs and
bottom debris (Anderson, 1967a). In shallow (< 1 m), high elevation ponds and
ephemeral pools, larvae may move into subsurface springs when pools freeze solid
(D.S.P., personal observations).
In lowland areas,
adults can remain active year-round (Stebbins, 1966). In southeastern Washington,
adults have been observed migrating to breeding ponds during the coldest winter months
and have been retrieved from submerged minnow traps in early January, below ≤ 15 cm of
ice, with air temperatures from 0 to -16 ˚C (J.A.F. and P. Verrell, unpublished
Associations/Exclusions. Long-toed salamanders occur in habitats used by other
amphibians, including blotched tiger salamanders (Ambystoma tigrinum
melanostictum), California slender salamanders (Batrachoseps
attenuatus), arboreal salamanders (Aneides lugubris),
rough-skinned newts (Taricha granulosa), Pacific treefrogs
(Hyla regilla,) mountain yellow-legged frogs (Rana
muscosa), Oregon spotted frogs (Rana pretiosa), and western
toads (Bufo boreas). Long-toed salamanders co-occur with Columbia
spotted frogs (Rana luteiventris) at > 80% of survey sites in Idaho
and Montana (Werner and Reichel, 1994; O'Siggins, 1995; Munger et al., 1997b; Llewelyn
and Peterson, 1998; Pilliod, 2001). This is probably due to a similarity in
habitat characteristics, but adult and larval long-toed salamanders may also utilize
frequently abundant Columbia spotted frog tadpoles as a food source (Munger et al.,
1997b). In Washington, long-toed salamanders and northwestern salamanders
(Ambystoma gracile) have generally allotopic distributions, possibly
resulting from competition and predation (Hoffman et al., 2003). Introduced trout
(Oncorhynchus sp.; Liss et al., 1995) and goldfish (Carassius
auratus; Monello and Wright, 2001) prey on long-toed salamander eggs and larvae,
substantially reducing their numbers, sometimes to the point of excluding them from
breeding sites (Liss et al., 1995; Munger et al., 1997b; Beck et al., 1998; Tyler et al.,
1998a; Yeo and Peterson, 1998; Funk and Dunlap, 1999; Hoffman and Pilliod, 1999; Monello
and Wright, 1999; Pilliod and Peterson, 2001). Artificial-pond experiments
indicate that trout reduce growth and survivorship of larvae, presumably due to both
indirect and direct effects of predation, such as limited foraging activity associated
with refuge use and increased predation (Tyler et al., 1998b).
L. Age/Size at
Reproductive Maturity. Sexual maturity is reached in 1–3 yr (at 50–55
mm SVL) after metamorphosis for both sexes (Anderson, 1967a; Nussbaum et al., 1983; Green
and Campbell, 1984; Howard and Wallace, 1985; Russell et al., 1996).
Skeletochronological techniques indicate that long-toed salamanders live ≤ 10 yr
(Russell et al., 1996).
Behavior. Larvae are carnivorous, opportunistic predators, consuming small
(zooplankton) to large (tadpoles) aquatic and terrestrial prey depending on size and
availability. Young larvae use a sit-and-wait technique, lunging at approaching
prey. Older larvae stalk or pursue prey (Anderson, 1968). Adults are also
carnivorous, preying on a variety of terrestrial organisms, such as annelids, mollusks,
and a variety of arthropods: arachnids, coleopterans, collembolans, dipterans, formicids,
lepidopterans, and orthopterans (Schonberger, 1944; Farner, 1947; Anderson, 1968).
Males feed at breeding sites, taking similar aquatic organisms as larvae (e.g., aquatic
dipterans; Anderson, 1968); females do not feed at breeding sites. This difference
may result from the longer time males spend at the breeding sites.
Invertebrate predators of larvae include predaceous diving beetles (Dytiscus
sp.; Marnell, 1997), odonate naiads, and belostomatids (J.A.F., personal
observations). Larvae and adults are also preyed upon by vertebrates, such as
salmonid fish (Liss et al., 1995), goldfish (Monello and Wright, 2001), northwestern
salamanders (Hoffman and Larson, 1999), blotched tiger salamanders (J.A.F., unpublished
data), non-native American bullfrogs (Rana catesbeiana; Nussbaum et al.,
1983; M.P. Hayes, personal communication),western terrestrial and common garter snakes
(Thamnophis elegans and T. sirtalis, respectively;
Ferguson, 1961a; Marnell, 1997; Pilliod, 2001), and belted kingfishers (Ceryle
alcyon; P. Murphy, personal communication).
Mechanisms. The highly secretive, subterranean life-style of long-toed salamanders
may be one of their most successful avoidance mechanisms (Ferguson, 1961a).
However, in the presence of predators, such as tiger salamanders, adult long-toed
salamanders may increase their activity and use cover objects less often (J.A.F. and P.
Verrell, unpublished data). The yellow dorsal stripe may serve as a warning to
predators. When attacks are simulated in a laboratory, long-toed salamanders
demonstrate a combination of behavioral and chemical defenses, including coiling, tail
undulations and lashing, and production of skin secretions (Anderson, 1963; Brodie, 1977;
Williams and Larsen, 1986). Adult long-toed salamanders can vocalize with squeaks
and clicks, possibly to startle predators once captured (Hossack, 2002). Adults
will avoid areas occupied by a damaged conspecific (Chivers and Kiesecker, 1996),
indicating the importance of chemical signals for avoiding predators. Larval
long-toed salamanders use chemical cues and learned recognition (non-lethal encounters)
to avoid inter- and intraspecific predators (Tyler et al., 1998b; Wildy et al.,
2001). The presence of methoxychlor increases vulnerability to predators (Verrell,
Little is known about the susceptibility of long-toed salamanders to diseases. Two
types of water molds, Saprolegnia ferax and Achlya
racemosa, have been found growing on long-toed salamander eggs in Montana and
Oregon (D.S.P. and others, unpublished data). Water molds, such as
Saprolegnia spp., have been reported to increase mortality of injured
ambystomatid salamanders (Walls and Jaeger, 1987) and may infect eggs that have been
stressed by environmental conditions such as low water levels, cold temperatures, or
increased UV-B radiation (Blaustein et al., 1994c; Kiesecker and Blaustein, 1995;
Kiesecker et al., 2001a). Investigations into the susceptibility of long-toed
salamanders to the recently discovered chytrid fungus are warranted.
Trematode parasites (Ribeiroia sp.) may be responsible for supernumerary limbs
and related deformities observed in > 1,600 long-toed salamanders collected from
Oregon in the late 1980s (Sessions and Ruth, 1990; Johnson et al., 1999; Sessions et
4. Conservation. Long-toed salamanders are widespread across their historical
range. Santa Cruz long-toed salamanders are Federally listed as Endangered and
occur in three population clusters (metapopulations) in coastal areas of Santa Cruz and
Monterey counties, California (U.S.F.W.S., 1999). The species is considered
“Secure” in Washington, Oregon, Idaho, Montana, and British Columbia
according to the National and State Heritage Status Ranks. Long-toed salamanders
are on the “Yellow B List” in Alberta, meaning that the species is not at
risk but vulnerable to limiting factors such as habitat alteration, destruction of
critical habitats, and non-native predatory fish (Graham and Powell, 1999). In
Alaska, long-toed salamanders are ranked as “Imperiled” by the State Heritage
system, but insufficient information exists to adequately assess the species’
may suffer local and possibly regional threats, but apparently these threats are not
resulting in widespread declines. In the vicinity of urban and agricultural areas,
lowland populations may be impacted by the loss of wetlands (Bury and Ruth, 1972; Ruth,
1974, 1988), road mortality during breeding migrations (Fukumoto and Herrero, 1998),
chemical contaminants (Ingermann et al., 1997; Fukumoto and Herrero, 1998; Nebeker et
al., 1998; Verrell, 2000), and predation from introduced goldfish (Monello and Wright,
2001). In relatively undisturbed mountain habitats, a number of possible threats
have been identified, including increased UV-B radiation (Blaustein et al., 1997; Belden
et al., 2000; Belden and Blaustein, 2002b), timber harvesting (McGraw, 1998; Naughton et
al., 2000), and introduced salmonid fish (Liss et al., 1995; Tyler et al., 1998a; Funk
and Dunlap, 1999; Hoffman and Pilliod, 1999; Pilliod and Peterson, 2001), but not
increased acid deposition or mobilization of aluminum (Bradford and Gordon, 1992;
Bradford et al., 1994c). Elasticity analyses of demographic models for long-toed
salamanders suggest that populations are more likely to decline when environmental
stressors result in higher mortality of post-metamorphic life stages as compared to
stressors that result in mortality of embryos or larvae (Vonesh and De la Cruz,
Acknowledgments. We thank several people for commenting on this account
and providing helpful natural history observations and access to unpublished documents,
including J. Howard, B. Leonard, K. McAllister, G. McLaughlin, C. Peterson, L. Powell,
and P. Verrell. Support for D.S.P. was provided by USGS Amphibian Research and
Monitoring Initiative during the final phase of this project.
Literature references for Amphibian Declines: The Conservation Status of United States Species, edited by Michael Lannoo, are here.
Feedback or comments about this page.
Citation: AmphibiaWeb. 2017. <http://amphibiaweb.org> University of California, Berkeley, CA, USA. Accessed 24 Apr 2017.
AmphibiaWeb's policy on data use.