Staying warm or moist? Operative temperature and thermal preferences of common frog (Rana temporaria), and effects on locomotion.

Abstract

Ambient temperature largely determines the body temperature of amphibians, and thus their hydration state and physiological performance. Microhabitat conditions chosen by terrestrial amphibians may represent a trade-off between high ambient temperatures, which maximize performance but cause high rates of water loss, and low temperatures, which, in turn, slow desiccation, but potentially hinder performance. We determined the operative temperature of common frogs (Rana temporaria) by placing 3% agar models in different microhabitats and measuring their temperature and water loss. Temperature measurements derived from the models accurately matched the body temperature of live frogs placed in the same microhabitat. Operative temperatures were lower than ambient temperatures on a warm day, probably because of evaporative water loss, but they were similar to or even slightly higher than ambient temperatures on a cool day, possibly because of warmth from the substrate. Frogs in the field selected moist and cool habitats, and their body temperatures ranged from 15 to 21 °C. In a temperature gradient in the laboratory, captive frogs chose significantly higher temperatures (19.4±1.7 °C) when the gradient floor was covered entirely with wet sand than when sand was wet in the cool end, but dry in the warm end (17.6±2.5 °C). The relevance of the preferred temperature was assessed through jumping performance experiments, using frogs with different body temperatures. Jump length was lower at low body temperature (6 °C) than at higher body temperatures, and peaked at 15 °C. Our results suggest that the frogs select favourable microhabitats of intermediate temperature, which could result in reduced water loss and peak physiological and behavioural performance.

Full text

17
Staying warm or moist? Operative temperature and
thermal preferences of common frogs (Rana temporaria),
and eects on locomotion
Angela Köhler1, Julita Sadowska2, Justyna Olszewska3, Paulina Trzeciak4,
Oded Berger-Tal5 & Christopher R. Tracy6
1
Department of Zoology and Entomology, University of Pretoria, South Africa
2
Department of Animal Ecology, University of Białystok, Poland
3
Department of Animal Toxicology, Institute of General and Molecular Biology, Nicolaus Copernicus University, Poland
4
Laboratory of Histology and Embryology of Vertebrates, Institute of Ecology and Environment Protection, Nicolaus Copernicus
University, Poland
5
Mitrani Department of Desert Ecology, Jacob Blaustein Institutes for Desert Research, Ben-Gurion University of the Negev, Israel
6
School of Environmental & Life Sciences, Charles Darwin University, Darwin, Australia
Ambient temperature largely determines the body temperature of amphibians, and thus their hydration state and physiological
performance. Microhabitat conditions chosen by terrestrial amphibians may represent a trade-off between high ambient
temperatures, which maximize performance but cause high rates of water loss, and low temperatures, which, in turn, slow
desiccation, but potentially hinder performance. We determined the operative temperature of common frogs (Rana temporaria)
by placing 3% agar models in different microhabitats and measuring their temperature and water loss. Temperature measurements
derived from the models accurately matched the body temperature of live frogs placed in the same microhabitat. Operative
temperatures were lower than ambient temperatures on a warm day, probably because of evaporative water loss, but they were
similar to or even slightly higher than ambient temperatures on a cool day, possibly because of warmth from the substrate. Frogs
in the eld selected moist and cool habitats, and their body temperatures ranged from 15 to 21 °C. In a temperature gradient
in the laboratory, captive frogs chose signicantly higher temperatures (19.4±1.7 °C) when the gradient oor was covered
entirely with wet sand than when sand was wet in the cool end, but dry in the warm end (17.6±2.5 °C). The relevance of the
preferred temperature was assessed through jumping performance experiments, using frogs with different body temperatures.
Jump length was lower at low body temperature (6 °C) than at higher body temperatures, and peaked at 15 °C. Our results
suggest that the frogs select favourable microhabitats of intermediate temperature, which could result in reduced water loss
and peak physiological and behavioural performance.
Key words: agar models, body temperature, jumping performance, microhabitat, skin temperature, thermal gradient
HERPETOLOGICAL JOURNAL 21: 17–26, 2011
Correspondence: Angela Köhler, Department of Zoology and Entomology, University of Pretoria, Pretoria 0002, South Africa.
E-mail: akoehler@zoology.up.ac.za
INTRODUCTION
Body temperature (Tb) of animals affects their
physiological and behavioural performance. Un-
like endotherms, ectothermic animals do not produce
metabolic heat to defend a constant Tb; instead, their Tb
is largely determined by the temperature of their envi-
ronment (Tracy, 1975; Navas et al., 2008). However,
ectotherms are known to thermoregulate behaviourally
by selecting favourable microhabitats (Lillywhite, 1970;
Christian & Weavers, 1996; Vences et al., 2002; Seebach-
er & Franklin, 2005). In doing so, they maintain a range of
Tb in which physiological and behavioural performance
are optimized (Walvoord, 2003; Seebacher & Franklin,
2005).
Warm microhabitats are favourable for ectotherms, as
locomotor performance is reduced at low ambient tem-
perature (Ta), and improves with increasing temperature
(Rome et al., 1992). Locomotion is crucial for survival
and tness as it facilitates escape from predators, forag-
ing and detection of mates. In amphibians, locomotor
performance generally increases with increasing Tb, until
it reaches a performance plateau and decreases rapidly at
very high temperatures (Wells, 2007). The optimal tem-
perature range varies between species, with some showing
wide performance plateaus over a 10–20 °C temperature
range, while others have narrower thermal optima (Tracy,
1979; Hirano & Rome, 1984; Knowles & Weigl, 1990;
Walvoord, 2003).
Although they are advantageous for locomotion, high
ambient temperatures result in higher evaporative water
loss (EWL) in terrestrial amphibians. The wet skin of
most amphibians offers essentially no barrier to water
loss through evaporation (Young et al., 2005). The selec-
tion of microhabitats has been found to be inuenced by
the hydration state of amphibians, thus indirectly affect-
ing thermoregulation (O’Connor & Tracy, 1992; Tracy
& Christian, 2005). Amphibians should prefer warm and
moist microhabitats to regulate their hydration state and
maximize physiological performance. However, a higher
Ta is often linked to dry conditions, so amphibians may
face a trade-off between choosing a lower Ta to avoid

18
desiccation and a higher Ta that maximizes locomotor
performance through increased Tb.
Tb and water loss of ectotherms in different micro-
habitats can be simulated by placing physical models
of similar size, shape and absorptivity into the natural
habitat of a particular species (e.g. Bartelt & Peterson,
2005; Tracy et al., 2007; for a review see Dzialowski,
2005). Such models produce a meaningful thermal index,
termed environmental or operative temperature (Te). Te is
dened as the Tb of an animal in thermal equilibrium with
its environment (Bakken et al., 1985; Dzialowski, 2005).
Models made out of 3% agar imitate evaporative prop-
erties of amphibian skin well (Spotila & Berman, 1976;
Navas & Araujo, 2000) and have been used in a variety of
studies of the thermal biology and water loss of amphib-
ians (Navas, 1996; Schwarzkopf & Alford, 1996; Navas
& Araujo, 2000).
The aim of this study was to explore the thermal ecol-
ogy of common frogs (Rana temporaria) by determining
Te under different conditions and comparing those tem-
peratures to thermal preferences in the eld and under
captive conditions. We also investigated the effect of Ta
on the physical performance of the frogs. We rst placed
different-sized physical models (simulating immature and
fully grown frogs) in various microhabitats and measured
their temperature and water loss to determine Te in R. tem-
poraria. We predicted that Te would be similar for frogs
of different sizes in the same habitat. The accuracy of our
temperature measurements was tested by placing models
and live specimens in the same microhabitat, predicting
that their Tb would be similar. Second, while we made
measurements of Te, we caught frogs in different micro-
habitats and determined Tb and environmental parameters.
Third, we investigated the thermal preferences of captive
frogs in a temperature gradient. We hypothesized that
the frogs would select moist microhabitats where their
Tb would not exceed 25 °C and water loss is minimized.
Fourth, to provide a performance context to the tempera-
ture selection data, i.e. to show the consequences of being
at different Ta, we investigated the effect of Tb on locomo-
tor activity of R. temporaria. We measured jump lengths
of frogs with different Tb and hypothesized that jumping
performance would increase with Tb and would be opti-
mal near the average Ta of their microhabitat.
MATERIALS AND METHODS
Study animals and validation of agar models
We caught 50 common frogs (Rana temporaria) of dif-
ferent body sizes in the Tuchola Forests region, Poland
(53°57'N, 17°48'E). Frogs were found in pine forests
(Pinus sylvestris), where the oor was patchily covered
mainly by sphagnum moss and blueberry bushes, and in
open grassland close to Lake Brzeźno and small streams
feeding the lake. They were weighed to ±0.01 g (Scout
Pro SP 402, Ohaus Corp., Pine Brook, NJ, USA), and
their snout–vent length (SVL), maximum head width and
maximum body width were recorded (Table 1). These
measurements were used to carve three different sizes of
physical models: the size of the smallest and largest live
specimen caught, and an average size deriving from the
mean of each measurement from all individuals. Models
were cut out from blocks of 3% agar, which simulates the
thermal and evaporative properties of live frogs (Navas
& Araujo, 2000).
To determine the accuracy of measurements using the
physical models, we compared Tb of the models to that of
live specimens. Medium-sized models and newly caught
frogs were placed in a plastic container (which provided
protection from wind) and placed in the sun, while a sec-
ond pair was placed in a mesh cage in the shade nearby.
The pairs remained in these microhabitats for 30 min, af-
ter which internal temperature was recorded by inserting
a type K thermocouple (associated with a digital thermo-
couple reader; ZyTemp TN40ALC, Radiant Innovation
Inc., HsinChu, Taiwan) into the cloaca of the frog and 3
mm under the dorsal surface of the model. We also meas-
ured the following environmental variables after the 30
min experimental period: Ta at 2 cm above the ground,
relative humidity (RH; Vaisala HMI41 humidity and tem-
perature indicator with a HMP44L probe, Vaisala Oyj,
Helsinki, Finland) and wind speed (Kestrel 4000 pocket
weather meter, Nielsen Kellerman, Boothwyn, Pennsyl-
vania, USA). The experiment was repeated four times
for each environment over the course of one day, using
different models and frogs. The agar models accurately
reected frog Tb, as the internal temperature of the physi-
cal models (Te sun: 20.0±1.0 °C; Te shade: 18.6±1.0 °C) and
of live specimens (Tb sun: 21.2±0.9 °C; Tb shade: 18.7±0.8
°C; all values are means ± SD) could not be distinguished
statistically (ANCOVA: F1,13=1.72, P=0.21) when placed
in the same environment. Tb of both models and frogs was
signicantly lower in the shade than in the sun (ANCO-
VA: F1,13=16.75, P<0.01).
To determine whether skin temperature (Tskin) could be
used as a surrogate for Tb in subsequent experiments, we
measured Tskin and cloacal Tb of the 50 frogs caught ini-
tially with an infrared thermometer and a thermocouple
(ZyTemp TN40ALC, Radiant Innovation Inc., HsinChu,
Taiwan). The frogs were held by their hind legs and placed
on the ground and Tskin and Tb were measured within a few
seconds to minimize the warming effect of the hand (Na-
vas & Araujo, 2000). Temperature measurements were
A. Köhler et al.
Table 1. Body measurements of 50 Rana temporaria
(mean ± SD and range) that were used to carve different
sized physical models. SVL = snout–vent length. Body
mass of the frogs is also presented.
SVL
(cm)
Head
width
(cm)
Body
width
(cm)
Body
mass (g)
Mean ± SD 4.15
±0.96
1.39
±0.32
1.87
±0.56
5.49
±3.90
Minimum 2.70 0.80 1.20 1.32
Maximum 7.00 2.40 3.30 18.08

19
made indoors, at a room temperature of 22 °C, after the
frogs had been habituated to this temperature for 30 min.
These conditions were not meant to represent any natural
situation, but were used to test whether Tskin and Tb were
equal. Tskin varied between consecutive measurements
with slight changes in the distance between thermom-
eter and frog skin (length of the infrared beam), and was
therefore not used as a surrogate for Tb. We report Tb for
subsequent experiments, as it was a more precise tem-
perature measurement.
Nineteen medium-sized frogs (SVL 4.14±0.79 cm and
body mass 7.39±2.31 g; mean ± SD) were kept in cap-
tivity for one week. Of these, 13 individuals were used
for the locomotion experiment and six for temperature
preference tests in a temperature gradient. We were not
able to distinguish between sexes, since most individu-
als were immature and paired vocal sacs and the nuptial
pads on the rst nger of males, used for gripping females
during mating, were not yet discernable. Frogs were kept
in plastic containers containing moist moss and live in-
sects caught at the eld site were provided daily after
experiments. All frogs were released at the site of capture
following completion of the study.
Measurements of Te
Te was estimated by placing medium-sized models in dif-
ferent microhabitats (3–4 replicates each) at the site of
initial capture. Four microhabitat categories were distin-
guished: warm & dry (sun, dry soil, low humidity), warm
& wet (sun, moist soil, high humidity), cool & dry (shade
or under plant cover, dry soil, low humidity) and cool &
wet (shade or under plant cover, moist soil, high humid-
ity). Two replicates (models of the same size) were placed
next to each other in a particular microhabitat for 30 min.
We weighed all models before and after this experimental
period to determine water loss. After the 30 min period, Tb
of the models was measured as described above. Simulta-
neously, we recorded Ta at 2 cm above the ground, wind
speed and RH. All RH values obtained in this study were
converted to vapour density using Smithsonian meteoro-
logical tables (List, 1966). We calculated vapour density
of the air (VDair) of each microhabitat from the recorded
RH and Ta, and vapour density at the frog surface (VDskin)
from the ambient RH and Tskin. Vapour density decit
(VDD) is the difference between VDair and VDskin.
To test for differences in Te between model sizes, we
placed the three freshly made models of different sizes in
an open spot (no plant cover, soil covered with dry moss)
and three others in a shaded spot (next to large bush, grass
cover, moist sandy soil), 3 m apart. Data were collected
for one sunny, warm day (day 1) and one overcast, cool
and windy day (day 2). On both days, Tb of the models
was recorded every 1 s, and averaged every 1 min, using
copper-constantan (Type-T) thermocouples interfaced to
a data logger (Campbell Scientic Inc. 21X Micrologger,
Logan, Utah, USA). We recorded Ta at 2 cm above the
ground in both environments using thermocouples, either
manually every 1 h or continuously by connecting them
to the data logger. RH and wind speed were recorded in
both spots three times daily. Models were weighed every
2 h and were replaced when they had lost more than 15%
of their initial mass.
Tb of wild frogs
To compare Te to temperatures selected by wild frogs,
we sampled an area consisting of different habitat types,
namely meadow (grass up to 80 cm high), pine forest
and articial garden (short grass and ower beds, next to
buildings), in close proximity to a lake (<15 m) and fur-
ther away from it (>15 m). All habitat types were sampled
with equal intensity three times a day (0900–1800), on the
same days the models were placed in the environment.
We caught 29 wild frogs of various sizes over the course
of the day and measured their Tb. The environmental con-
ditions (Ta at 2 cm above the ground, RH and wind speed)
were also recorded in each place a frog was captured.
Temperature preference
In addition to Tb selected in the eld, we measured ther-
mal preferences of six captive frogs in a thermal gradient
system, consisting of a long and narrow aluminium trough
(120 × 10 cm). One end of the trough was heated to 50 °C
by a FBH 604 Fisherbrand® thermostat and the second
end was cooled down to 0 °C by a FBH 635 Fisherbrand®
cryostat, resulting in a temperature gradient ranging from
1 to 45 °C. The temperature gradient was divided into
16 compartments of equal length. Temperature increases
were greater between the ve compartments at each end of
the gradient (2–4 °C difference between two neighbour-
ing compartments) than in the six middle compartments
(1.5–2 °C difference). Frog movements in the gradient
were not restricted. However, in each experimental series,
low cardboard barriers were placed at 12 °C and 40 °C in
the gradient to ensure that frogs were choosing very low
(<12 °C) or high (>40 °C) temperatures. A single frog was
placed in the middle of the gradient and was allowed to
habituate for 10 min. Thereafter, we observed its behav-
iour in the gradient for 1 h, during which the compartment
the frog stayed in was recorded every 60 s. Substrate tem-
peratures in each compartment were measured after the
1 h observation period using a thermocouple. The six cap-
tive frogs were tested in two experiments on consecutive
days. In the rst experiment, the oor of the gradient was
covered entirely with wet sand. This experiment allowed
us to assess general thermal preferences of the common
frog when there were no constraints on hydration. In the
second experiment, the gradient oor was covered with
wet sand at the cold end (0–12 °C) and dry sand at the
warm end (13–45 °C). These temperatures were chosen
based on preferences found in the rst series. The lowest
temperature preferred by frogs on wet sand was 15 °C
(see results below), but we put the wet sand at 3 °C cooler
than the lowest preferred temperature to achieve a clear
distinction between the regulation of Tb or hydration.
The effect of Tb on locomotion
As an indicator of the consequences of choosing different
Tb, we tested jumping performance at three temperatures.
We altered Tb of 13 captive frogs by placing them into
different Ta or water temperatures. Three temperature
categories were used: warm (sunny day, mean ± SD Ta
= 22.25±2.05 °C), medium (temperature of lake water:
Temperature regulation in common frogs

20
12.25±2.19 °C) and cold (water with melted ice: 2 °C).
The same individuals were exposed to each temperature
category on consecutive days, always at the same time of
day to minimize the effects of daily rhythms. They were
habituated to a certain temperature for 10–15 min, or until
Tb had adjusted to the environmental temperature (<5 min
in ice water). Cloacal Tb was measured and the frog was
immersed in very dilute water-soluble paint and placed in
a jumping arena. This arena consisted of a wooden base
(200 × 50 cm), on which paper was placed. A plastic mesh
was used to cover the sides and top so that the frog could
only jump in one direction. Frogs were stimulated to jump
by gently tapping their hind legs. A plastic box containing
moss was placed at the end of the jumping arena and the
frogs jumped towards this hiding place. Immediately after
the jumps, Tb was measured again, but generally did not
differ from the initial value. Environmental parameters
(Ta at 2 cm above the ground, RH and wind speed) inside
the jumping arena were also recorded at the start and dur-
ing the experiment.
The length of the individual jumps was later measured
using the paint imprints. A jump was measured from the
end of the hind legs of one imprint to the end of the hind
legs of the next imprint. Depending on the length of the
jumps and willingness to jump, we obtained four to 10
jumps per frog, and the mean of all jumps for each in-
dividual at a particular temperature was calculated. We
calculated the temperature coefcients (Q10) from the
relationship between mean jump length and Tb. We also
measured SVL of the frogs to determine whether the
maximum jump length of each individual was related to
frog size.
Statistical analysis
All data were tested for normality (Kolmogorov–Smirnov
and Shapiro–Wilk tests) and homogeneity of variance
(Levene’s test). Log, square-root or arcsine transfor-
mations were used when data were heteroscedastic. A
General Linear Model (ANCOVA) was used to identify
which environmental factors affect Tb of the wild frogs.
The categorical predictor was body size and continuous
predictors were Ta, wind speed, VDair and VDskin. Variables
were removed from the analysis until the best model t
was found; the adjusted R2 value was used as an indicator
for model t. For each experimental series in the tempera-
ture gradient, the proportion of time spent in cold (≤12
°C) and warm (>12 °C) areas were compared by paired
t-test. Temperatures selected by the frogs (i.e. the Ta at
which each frog spent most of the time) were compared
between the series using a paired t-test. The temperatures
frogs chose in each minute of observation in each experi-
mental series were exposed to a Levene’s test, assuming
that the frogs would show greater variance in their tem-
perature choices when they faced a trade-off (wet and
dry sand). For the locomotion experiment, Tb and mean
and maximum jump lengths were compared between the
three temperature categories using repeated-measures
ANOVA. To eliminate a possible effect of body size on
jumping performance, this analysis was performed on
SVL-corrected jumping data. Post-hoc comparisons were
conducted with Tukey’s Honestly Signicant Difference
test for equal sample sizes, followed by a Bonferroni
correction for multiple comparisons (Rice, 1989). Lin-
ear regression analysis (GLM) was used to test for a
relationship between SVL and maximum jump length at
intermediate temperature (where frogs jumped furthest).
Statistical analysis was performed using StatisticaTM 9.0;
the level of signicance was α≤0.05 for all tests. All data
are presented as means ± SD.
RESULTS
Measurements of Te
Te and water loss were highest in the warm & dry mi-
crohabitat, where Ta, wind speed and VDD were highest
(Table 2). Water loss was lowest in the cool & wet micro-
habitat, while Te was lowest in the cool & dry habitat (Table
2). Te derived from the three model sizes was similar. On
day 1, Ta reached a maximum of 25.4 °C in the shade and
29.2 °C in the sun (Fig. 1), with larger temperature uc-
tuations in the sunny microhabitat, as this spot was either
exposed to direct sunlight or occasionally shaded by small
A. Köhler et al.
Table 2. Ambient temperature (Ta), relative humidity (RH), vapour density (VDair) and wind speed of the four different
environmental categories (mean ± SD; n is sample size). Operative temperature (Te), vapour density at the frog
model surface (VDskin) and water loss deriving from the physical models placed in these microhabitats are also
given (mean ± SD; n is sample size). VDD is the vapour density decit (difference between VDskin and VDair). VD was
calculated from mean values of Ta and RH using Smithsonian meteorological tables (List, 1966).
Ambient conditions Model
Environmental
category n Ta (°C) RH (%)
VDair
(gm–3)
Wind
speed
(ms–1)n Te (°C)
VDskin
(gm–3)
VDD
(gm–3)
Water loss
(g per
30 min)
Warm & dry 320.53±2.38 34.73±8.24 6.20 0.20±0.35 6 20.45±2.42 17.76 11.56 0.39±0.03
Warm & wet 319.30±0.75 56.47±6.62 9.37 0.03±0.06 6 19.42±0.91 16.72 7.34 0.22±0.04
Cool & dry 316.17±0.75 44.97±6.56 6.20 0.07±0.12 6 16.07±0.77 13.69 7.49 0.13±0.06
Cool & wet 4 15.33±0.46 56.57±3.06 7.41 0±0 8 17.62±1.61 15.03 7.62 0.11±0.03

21
clouds. RH averaged 47.8±2.1% (VDair 9.3±1.9 gcm–3) in
the shade and 43.7±3.7% (VDair 9.6±1.9 gcm–3) in the
sun. There was no wind in the shade, while wind speed
was 0.1±0.1 ms–1 in the sun. In both microhabitats, Te was
generally below Ta, and slightly lower in medium-sized
models than in large and small ones. Day 2 was overcast
and windy, and Ta was similarly low in both microhabi-
tats (Fig. 1). Ta increased in the afternoon when the sun
appeared shortly, and the temperature was higher in the
shaded spot, protected from wind, than in the sunny, open
spot. RH averaged 81.0±1.8% (VDair 11.6±0.5 gcm–3)
in the shade and 72.5±3.2% (VDair 10.4±0.2 gcm–3) in
the sun. Wind speed was 0.1±0.1 ms–1 in the shade and
2.4±0.6 ms–1 in the open microhabitat. Te was remarkably
similar to Ta throughout the day in the shaded spot, while
it was occasionally slightly higher than Ta in the open
microhabitat, with the highest Te obtained from the large
models. Note that Ta was only recorded hourly in the open
microhabitat on day 2; it is therefore possible that brief
temperature peaks may have been missed.
Tb of wild frogs
The frogs that were captured over the course of the day
(on the same two days that the models were placed in the
two environments) were generally found in close prox-
imity to permanent bodies of water on moist or dry soil
that was covered by high grass or small bushes. No frogs
could be found more than 15 m away from water or in
open locations exposed to direct sunlight. Thus, frogs
were found in similar microclimates and had similar Tb
throughout the day. Tb of all 29 captured individuals aver-
aged 18.3±1.5 °C (range 15.4–21.2 °C) at a mean Ta of
Temperature regulation in common frogs
Fig. 1. Ambient temperatures (Ta) in one shaded microhabitat (next to large bush, grass cover, moist sandy soil –
shade, left) and one open microhabitat (no plant cover, soil covered with dry moss – sun, right), which were 3 m
apart. Operative temperature (Te) was derived from physical frog models of different sizes (small, medium, large,
N=1 each) placed in each environment. Te was recorded continuously over one sunny, warm day (day 1, above)
and one overcast, cool and windy day (day 2, below). Missing values are due to malfunctioning thermocouples or
interruptions in temperature recordings during replacements of models. Similar Te were obtained from the models
when placed in the same microhabitat. Te closely followed Ta on the cool day, but was generally lower than Ta on
the sunny day.

22
19.6±1.4 °C (range 18.0–22.8 °C). Wind speed was gen-
erally low, averaging 0.2±0.3 ms–1 (range: 0–0.9 ms–1),
and RH was 71.0±9.8% (range 47.7–82.9%). VDair aver-
aged 11.9±1.2 gcm–3, VDskin 15.3±2.1 gcm–3, and VDD
was 3.4±1.4 gcm–3. Field Tb were independent of body
size (ANCOVA: F2,23=2.04, P=0.15), but depended on
VDskin (F1,23=8.86, P<0.01) and wind speed (F1,23=4.83,
P=0.04). Field Tb was further affected by Ta, although this
was not statistically signicant (F1,23=3.59, P=0.07).
Temperature preference
Frogs placed in the gradient covered with wet sand spent
more time in the warm compartments (>12 °C) than in
cold ones (paired t-test: t5 =–6.19, P<0.01; Fig. 2). They
chose mean substrate temperatures of 19.4±1.7 °C. After
choosing a particular Ta, all frogs stayed at that tempera-
ture for the remainder of the hour of observation. Frogs
placed in the gradient with wet sand at the cold end and
dry sand at the warm end showed more variation in select-
ed temperatures (Levene’s test: P<0.001), moving more
often between the cold and the warm ends of the gradi-
ent. Thus, they divided their time approximately equally
between the cold end and the warm end of the gradient
(paired t-test: t5 = –1.43, P=0.21; Fig. 2). When station-
ary, frogs chose lower temperatures (17.6±2.5 °C) on wet/
dry sand than on wet sand only (paired t-test: t5 = –2.63,
P=0.04).
The effect of Tb on locomotion
Tb of the 13 frogs prior to jumping differed signicantly
between the three temperature categories (RM-ANOVA:
F2,24=336.56, P<0.001), with Tb being slightly higher than
the temperature recorded in the air or water that surround-
ed the frogs. The ambient conditions inside the jumping
arena were not controlled; Ta was similar (20.1–21.3 °C),
while RH increased (from 47.2% to 65.0%; VDair 8.6–
11.3 gcm–3) from the warm to the cold category. Wind
speed was very low for tests of all temperature categories
(<1.00 ms–1). The mean jump length of the frogs differed
signicantly between the three temperature categories
(RM-ANOVA: F2,24=19.45, P<0.001; Fig. 3), with jump
length being lower in the cold than at intermediate and
high temperatures (Tukey HSD test: P<0.01). Two of
the 13 frogs repeatedly did not jump at all at the low-
est Tb (6.3 °C), but were not excluded from the analysis
because the statistical results did not change when these
two individuals were omitted. Jump length was higher
in the medium category than in the warm category, but
this was only marginally signicant (P=0.05 after Bon-
ferroni correction). The temperature coefcients for the
jump lengths of R. temporaria were determined as: Q10
(6.3–14.6 °C) = 1.99; Q10 (14.6–22.8 °C) = 0.80; Q10 (6.3
–22.8 °C) = 1.27. Similarly to mean jump length, maxi-
mum jump length also differed between the temperatures
(RM-ANOVA: F2,24=6.15, P<0.01), being higher in the
medium and warm categories than in the cold (Tukey
HSD test: P<0.02). The maximum jump length at inter-
mediate temperatures was not signicantly related to SVL
(linear regression: F1,11=0.65, P=0.53, R2=0.04; Fig. 4).
A. Köhler et al.
Fig. 2. Proportion of time six Rana temporaria spent
in the cold (≤12 °C) and warm (>12 °C) end of a
thermal gradient (mean + SD). Frogs were tested in two
experimental series with different moisture content of
the substrate: 1) wet and 2) wet at the cold end and
dry at the warm end. When the entire gradient was wet,
frogs spent signicantly more time at the warm end,
while they divided their time more equally between
cold and warm end when the warm end of the thermal
gradient was dry. Statistical results derive from a paired
t-test (**P<0.01).
Fig. 3. Jump length of 13 Rana temporaria at different
body temperatures (mean + SD). Frogs jumped the
farthest at the intermediate Tb of 14.6 °C. Statistical
results derive from a Tukey HSD test that followed a
RM-ANOVA (*P≤0.05, **P≤0.01, ***P≤0.001).

23
DISCUSSION
The use of physical models to determine Te
In earlier studies, the rates of cutaneous water loss and
estimates of Te derived from agar models matched eld
Tb of amphibians (Spotila & Berman, 1976; Navas &
Araujo, 2000). It should be noted, however, that there are
differences in absorptivity between the transparent agar
models and the green/brownish colour of frogs. To test the
accuracy of our models and the possible effect of differ-
ences in absorptivity, we compared the Tb of agar models
and live frogs placed in the same microhabitat, and found
that they were very similar. These results suggest that
the thermal properties of agar models and R. temporaria
are comparable. Thus, agar models provide an accurate
measurement of Te and are useful in increasing our under-
standing of amphibian physiology and ecology. However,
to overcome the possible effect of differences in absorp-
tivity between agar models and frogs, future studies could
either colour the agar or use plaster models, which can
be coloured to match the absorptivity of amphibian skin
and provide similar results to agar models (Tracy et al.,
2007).
As expected, frog Te was highest in warm and dry en-
vironments and lowest in cool and dry environments. In
the cool environments, Te was lower in the dry than in the
wet microhabitat, possibly due to lower RH (and lower
vapour density decit between air and frog surface) and
higher wind speed in the dry microhabitat, resulting in
more heat loss. Water loss was higher in the warm than
in the cool environments, indicating that cool micro-
habitats should be more favourable for amphibians that
need to reduce their EWL. Water loss was higher in the
dry environments than in the wet ones, as water is lost
faster to dry soil and dry air than to moist soil and humid
air. Our models always lost mass, probably because of a
very small osmotic gradient between the agar model and
the substrate. Live frogs, on the other hand, can absorb
water from the surrounding environment when in water,
on moist soil, or during rain (Walker & Whitford, 1970;
Wells, 2007).
Te was generally lower than or equal to Ta. However,
when the sky was overcast and the weather cooler, the Te
obtained from some models was higher than Ta. The sub-
strate these models were placed on may have been slightly
warmer than the air, as Ta was higher before our experi-
ment and the soil may have stored heat and increased in
temperature. Te was similar in the different sizes of model,
with no particular size absorbing more heat than others.
Thermal preferences of R. temporaria
According to Te estimates and water loss of the agar
models, frogs would be expected to select cool and wet
habitats, protected from sun and wind by plant cover, as
their water loss is minimal in this microclimate. Wild
frogs that we captured over the course of the day were
generally found under plant cover, in cool and often moist
microhabitats, and never in the direct sun. Thus, frogs
maintained similar Tb throughout the day, and Tb was also
similar between different body sizes and light and dark
individuals. Coloration also had no effect on eld Tb in
an earlier study, where a high-altitude population of R.
temporaria also showed active thermoregulatory behav-
iour: Tb was higher than Ta during the day due to the frogs
basking in the sun, and at night, frogs retreated to a pond
where the water acted as a temperature buffer during the
nocturnal drop in Ta (Vences et al., 2002).
In addition to using direct eld observations of tem-
perature, an ectotherm’s preference for favourable
microhabitats can also be determined in the laboratory
by placing the animal in thermal gradients (for a review
see Brattstrom, 1979). In such a temperature gradient, our
frogs preferred intermediate temperatures (19.4 °C) on
wet substrates, probably due to the fact that they would
dehydrate quicker at higher temperatures and face hypo-
thermia at lower temperatures. When the frogs were faced
with a trade-off between heat and humidity, with only the
cold end of the gradient covered with wet substrate, they
chose even lower temperatures (17.6 °C) and spent more
time in the cold end of the gradient (Fig. 2), suggesting
that lowering Tb is less harmful than dehydration for these
frogs. Similarly, toads (Anaxyrus (Bufo) woodhousei) se-
lect lower temperatures on dry than on wet sand to avoid
desiccation (O’Connor & Tracy, 1992). Rapid heat loss at
the cool end and water loss at the hot and dry end caused
our frogs to move more often between the cool and warm
segments. The frogs may have had to leave the cool end
to raise their Tb, but then returned to the cool and humid
microclimate to avoid desiccation.
Behavioural hypothermia in ectotherms is dened in
the literature as the phenomenon of seeking lower tem-
peratures to reduce EWL or energy consumption (Tracy
et al., 1993), and has been observed in toads, Anaxyrus
(Bufo) americanus (Tracy et al., 1993) and lizards, Scelo-
porus undulatus (Crowley, 1987). The lower Tb reduces the
Temperature regulation in common frogs
Fig. 4. The maximum jump length of 13 Rana
temporaria at an intermediate temperature, in relation
to snout–vent length (SVL). Each frog was tested only
once at this temperature. Maximum jump length
was not signicantly related to body size (see text for
statistical results).

24
difference between water vapour density in the environ-
ment and at the skin, thus slowing the rate of EWL (Tracy
et al., 1993). This may be important for R. temporaria,
which tend to dehydrate quickly due to their relatively
small body size and high evaporation rate through the
skin (Lillywhite, 2006). In a habitat-like experimental
environment with different Ta (8–31 °C), Sinsch (1984)
found that R. temporaria exhibited behavioural cooling
in the heat (by staying in water or cool hiding places and
using evaporative cooling in dry microclimates), while
warming behaviour was demonstrated in the cold. At the
same time, these frogs changed their preferred time of
activity from the night at high temperatures to the day at
low temperatures, thus maintaining a fairly constant Tskin
at all Ta. The frogs preferred moderate Ta (10–20 °C), and
lengthy exposure to more than 30 °C led to 100% mortal-
ity (Sinsch, 1984). If shaded and moist areas provide the
optimal habitat for the regulation of body temperature, R.
temporaria may be affected by human-induced landscape
alteration and fragmentation. It remains to be investigated
to what extent the local distribution of these frogs is inu-
enced by a reduction in shadow refuges, such as caused
by deforestation and removal of natural vegetation lead-
ing to open canopy ponds.
The effect of Tb on locomotion
Tb had an effect on the jump performance of R. temporar-
ia. We measured average jump lengths of the frogs, as we
were interested in the average locomotor performance of
the frogs at a particular temperature. While the maximum
jump length indicates how well individuals are able to
escape from predators, average jump length may also in-
dicate how well the frogs are able to select microhabitats,
and nd food and mates. Frogs in our study had longer
jumps at higher Tb than at 6 °C, and jumped farthest at
15 °C (although only marginally significantly further
than at 23 °C). Navas et al. (1999) reported very simi-
lar jump lengths for R. temporaria at comparable body
temperatures, and reported an increasing jump length up
to 20 °C.
Jump length of common frogs was halved at low Tb
(≤6 °C) in both the present study and the Navas et al.
(1999) study, compared to 15 °C. In addition, two of our
frogs repeatedly did not jump at all at the lowest tem-
perature. This suggests that 6 °C is approaching the frog’s
critical thermal minimum, at least for movement on land.
Tattersall & Boutilier (1999) found that R. temporaria was
capable of swimming at temperatures as low as 1.5 °C, but
swimming speed and distance were signicantly reduced
compared to 7 °C. Critical thermal minima, dened as the
temperature at which an animal has lost the ability to es-
cape as temperatures fall to lethally low levels (Cowles &
Bogert, 1944), have been reported to be between 2 and 7
°C for ectotherms, including frogs (Christian et al., 1988),
lizards (Gvoždík & Castilla, 2001) and snakes (Doughty,
1994). However, lizards are rarely active at near-threshold
Tb (Huey & Stevenson, 1979) and frogs are also known to
hibernate in burrows or under water during winter when
Ta is low (Irwin et al., 1999; Roots, 2006). The ecological
relevance of locomotor performance at 6 °C is therefore
uncertain.
Frogs jumped the farthest at a Tb of 15 °C, suggesting
that a Tb around 15 °C maximizes locomotor performance,
although smaller temperature intervals should be tested to
clearly dene the performance plateau. This also indicates
that the temperatures chosen by the frogs in the thermal
gradient experiment are ecologically relevant, and opti-
mized (in a broad sense) their locomotor performance.
We did not measure jump lengths at higher Tb, as eld Tb
was not found to be higher than 22 °C in our study. Future
experiments should cover a wider temperature range to
determine the critical thermal minimum and maximum
of these frogs.
The temperature dependence of jump performance
has also been shown in other amphibians. Cricket frogs
(Acris crepitans), for instance, jumped equally well at
temperatures of 23 and 30 °C, but frogs produced shorter
jumps at 15 °C (Walvoord, 2003). Maximum jump dis-
tance of Lithobates (Rana) pipiens was lowest at 14 °C,
increased with increasing temperature, reached a maxi-
mum at 25 °C, and decreased thereafter (Hirano & Rome,
1984). The maximum jump distance of Cuban tree frogs
(Osteopilus septentrionales) increased over a range of Tb
from 11 to 30 °C (Peplowski & Marsh, 1997). Knowles
& Weigl (1990) tested ve frog species, whose maximum
jump length increased from 5 °C up to 20 or 30 °C. How-
ever, Rana clamitans and Lithobates (Rana) sylvaticus
showed narrow thermal optima, whereas the other three
species (Acris crepitans, Hyla femoralis and Pseudacris
triseriata) showed wider performance plateaux. In the
eld, amphibians select a Tb that allows optimal locomo-
tor performance, as shown in cricket frogs (Walvoord,
2003).
Conclusion
Tb affects the locomotor performance of R. temporaria,
and frogs showed maximal jumping distances at an inter-
mediate temperature (15 °C). Field Tb ranged from 15 to
21 °C, even on a very warm summer day, reaching a Ta of
30 °C in the sun. This suggests behavioural temperature
regulation. Frogs were not found in microhabitats with
Ta higher than 23 °C, indicating that they actively avoid
high temperature microhabitats in the environment. Agar
models demonstrated that water loss is higher at high
temperatures and in dry microhabitats, and the selection
of intermediate temperatures and moist environments
thus prevents extensive water loss. Amphibians gener-
ally prefer warm and moist environments, but may face a
trade-off as these are not always together in the wild. Our
results from the thermal gradient suggest that temperature
preferences of R. temporaria are affected by the moisture
content of the surrounding environment, and that frogs
alternate between cool moist and warm dry microhabi-
tats. In maintaining water balance and choosing optimal
temperatures, these frogs optimize their physiological and
behavioural performance.
ACKNOWLEDGEMENTS
This research was conducted as part of the Physiological
Ecology Workshop (course 0501-PEW-DX) at the Nico-
laus Copernicus University (NCU), Toruń, Poland. The
A. Köhler et al.

25
workshop was funded by NCU and the Blaustein Center
for Scientic Cooperation (BCSC) at the Jacob Blaustein
Institutes for Desert Research, Ben-Gurion University
of the Negev, Israel. We would like to thank Dr Michał
S. Wojciechowski and Prof. Berry Pinshow for organ-
izing the workshop, and the other course instructors, Dr
Małgorzata Jemow and Dr Andreas Mölich (in addition
to C.R.T. and O.B.-T.), for helpful comments during the
experiment. We are grateful to NCU for permission to
stay at their Popówka eld station. Thanks to Prof. B. Pin-
show for helpful comments on an earlier version of this
manuscript. Prof. Sue W. Nicolson and the National Re-
search Foundation, South Africa, are thanked for funding
A.K.’s journey to Poland. C.R.T. was supported by a grant
from the Australian Research Council (DP0879851). Our
experiments were approved by the NCU Committee for
Ethics in Animal Research and by the Nature Conserva-
tion Authorities.
REFERENCES
Bakken, G.S., Santee, W.R. & Erskine, D.J. (1985). Operative
and standard operative temperature: tools for thermal
energetics studies. American Zoologist 25, 933–943.
Bartelt, P.E. & Peterson, C.R. (2005). Physically modeling
operative temperatures and evaporation rates in amphibians.
Journal of Thermal Biology 30, 93–102.
Brattstrom, B.H. (1979). Amphibian temperature regulation
studies in the eld and in the laboratory. American Zoologist
19, 345–356.
Christian, K.A., Nunez, F., Clos, L. & Diaz, L. (1988). Thermal
relations of some tropical frogs along an altitudinal gradient.
Biotropica 20, 236–239.
Christian, K.A. & Weavers, B.W. (1996). Thermoregulation of
monitor lizards in Australia: an evaluation of methods in
thermal biology. Ecological Monographs 66, 139–157.
Cowles, R.B. & Bogert, C. (1944). A preliminary study of
the thermal requirements of desert reptiles. Bulletin of the
American Museum of Natural History 83, 265–296.
Crowley, S.R. (1987). The effect of desiccation upon the
preferred body temperature and activity level of the lizard
Sceloporus undulates. Copeia 1987, 25–32.
Doughty, P. (1994). Critical thermal minima of garter snakes
(Thamnophis) depend on species and body size. Copeia
1994, 537–540.
Dzialowsky, E.M. (2005). Use of operative temperature and
standard operative temperature models in thermal biology.
Journal of Thermal Biology 30, 317–334.
Gvoždík, L. & Castilla, A.M. (2001). A comparative study of
preferred body temperatures and critical thermal tolerance
limits among populations of Zootoca vivipara (Squamata:
Lacertidae) along an altitudinal gradient. Journal of
Herpetology 35, 486–492.
Hirano, M. & Rome, L.C. (1984). Jumping performance of frogs
(Rana pipiens) as a function of muscle temperature. Journal
of Experimental Biology 108, 429–439.
Huey, R.B. & Stevenson, R.D. (1979). Integrating thermal
physiology and ecology of ectotherms: a discussion of
approaches. American Zoologist 19, 357–366.
Irwin, J.T., Costanzo, J.P. & Lee Jr., R.E. (1999). Terrestrial
hibernation in the northern cricket frog, Acris crepitans.
Canadian Journal of Zoology 77, 1240–1246.
Knowles, T.W. & Weigl, P.D. (1990). Thermal dependence of
anuran burst locomotor performance. Copeia 1990, 796–
802.
Lillywhite, H.B. (1970). Behavioral thermoregulation in the
bullfrog, Rana catesbeiana. Copeia 1970, 158–168.
Lillywhite, H.B. (2006). Water relations of tetrapod integument.
Journal of Experimental Biology 209, 202–226.
List, R.J. (1966). Smithsonian Meteorological Tables. Publ.
4014. Washington, D.C.: Smithsonian Institution.
Navas, C.A. (1996). Implications of microhabitat selection and
patterns of activity on the thermal ecology of high elevation
neotropical anurans. Oecologia 108, 617–626.
Navas, C.A. & Araujo, C. (2000). The use of agar models to
study amphibian thermal ecology. Journal of Herpetology
34, 330–334.
Navas, C.A., Gomes, F.R. & Carvalho, J.E. (2008). Thermal
relationships and exercise physiology in anuran amphibians:
integration and evolutionary implications. Comparative
Biochemistry and Physiology A 151, 344–362.
Navas, C.A., James, R.S., Wakeling, J.M., Kemp, K.M. &
Johnston, I.A. (1999). An integrative study of the temperature
dependence of whole animal and muscle performance
during jumping and swimming in the frog Rana temporaria.
Journal of Comparative Physiology B 169, 588–596.
O’Connor, M.P. & Tracy, C.R. (1992). Thermoregulation by
juvenile toads of Bufo woodhousei in the eld and in the
laboratory. Copeia 1992, 865–876.
Peplowski, M.M. & Marsh, R.L. (1997). Work and power output
in the hindlimb muscles of Cuban tree frogs Osteopilus
septentrionalis during jumping. Journal of Experimental
Biology 200, 2861–2870.
Rice, W.R. (1989). Analyzing tables of statistical tests. Evolution
43, 223–225.
Rome, L.C., Stevens, D. & John-Alder, H.B. (1992). The
influence of temperature and thermal acclimation on
physiological function. In Environmental Physiology of
the Amphibians, 183–205. Feder, M.E. & Burggren, W.W.
(eds). Chicago: University of Chicago Press.
Roots, C. (2006). Hibernation. Westport, Connecticut:
Greenwood Press.
Schwarzkopf, L. & Alford, R.A. (1996). Desiccation and
shelter-site use in a tropical amphibian: comparing toads
with physical models. Functional Ecology 10, 193–200.
Seebacher, F. & Franklin, C.E. (2005). Physiological mechanisms
of thermoregulation in reptiles: a review. Journal of
Comparative Physiology B 175, 533–541.
Sinsch, U. (1984). Thermal inuences on the habitat preference
and the diurnal activity in three European Rana species.
Oecologia 64, 125–131.
Spotila, J.R. & Berman, E.N. (1976). Determination of skin
resistance and the role of the skin in controlling water loss
in amphibians and reptiles. Comparative Biochemistry and
Physiology A 55, 407–412.
Tattersall, G.J. & Boutilier, R.G. (1999). Does behavioural
hypothermia promote post-exercise recovery in cold-
submerged frogs? Journal of Experimental Biology 202,
609–622.
Tracy, C.R. (1975). Water and energy relations of terrestrial
amphibians: insights from mec hanistic modelling. In
Perspectives of Biophysical Ecology, 325–246. Gates, D.M.
Temperature regulation in common frogs

26
& Schmerl, R.B. (eds). New York: Springer.
Tracy, C.R. (1979). Further thoughts on anuran thermoregulation:
discussion. In The Behavioral Signicance of Color, 63–68.
Burtt Jr., E.H. (ed.). New York: Garland STPM Press.
Tracy, C.R., Betts, G., Tracy, C.R. & Christian, K.A. (2007).
Plaster models to measure operative temperature and
evaporative water loss of amphibians. Journal of Herpetology
41, 597–603.
Tracy, C.R. & Christian, K.A. (2005). Preferred temperature
correlates with evaporative water loss in hylid frogs from
northern Australia. Physiological and Biochemical Zoology
78, 839–846.
Tracy, C.R., Christian, K.A., O’Connor, M.P. & Tracy, C.R.
(1993). Behavioral thermoregulation by Bufo americanus:
the importance of the hydric environment. Herpetologica
49, 375–382.
Vences, M., Galán, P., Vieites, D.R., Puente, M., Oetter, K. &
Wanke, S. (2002). Field body temperatures and heating rates
in a montane frog population: the importance of black dorsal
pattern for thermoregulation. Annales Zoologici Fennici 39,
209–220.
Walker, R.F. & Whitford, W.G. (1970). Soil water absorption
capabilities in selected species of anurans. Herpetologica
26, 411–418.
Walvoord, M.E. (2003). Cricket frogs maintain body hydration
and temperature near levels allowing maximum jump
performance. Physiological and Biochemical Zoology 76,
825–835.
Wells, K.D. (2007). The Ecology and Behaviour of Amphibians.
Chicago: University of Chicago Press.
Young, J.E., Christian, K.A., Donnellan, S., Tracy, C.R. & Parry,
D. (2005). Comparative analysis of cutaneous evaporative
water loss in frogs demonstrates correlation with ecological
habitats. Physiological and Biochemical Zoology 78, 847–
856.
Accepted: 9 September 2010
A. Köhler et al.