Biological Journal of the Linnean Society, 2010, 101, 437–446. With 1 figure
Threatened tadpoles of Bokermannohyla alvarengai
(Anura: Hylidae) choose backgrounds that enhance
crypsis potential
PAULA C. ETEROVICK1*, FRANCISCO F. R. OLIVEIRA1 and GLENN J. TATTERSALL2
1
Programa de Pós-Graduação em Zoologia de Vertebrados, PUC Minas, Belo Horizonte, 30535-610,
Minas Gerais, Brazil
2
Department of Biological Sciences, Brock University, St Catharines, Ontario, Canada
Received 13 April 2010; revised 15 April 2010; accepted for publication 21 April 2010
bij_1501
437..446
Crypsis results from a complex interaction among prey coloration, background matching, behaviour and predator
visual perception. Tadpoles are known to have varied adaptations to escape predation, but the use of crypsis is little
explored, although it is likely for certain species. We investigated potential escape mechanisms related to active
escape (fleeing) and crypsis improvement in Bokermannohyla alvarengai tadpoles, proposing a new method to
measure cryptic potential. We studied the range of distances covered by threatened and fleeing tadpoles and the
proportion of tadpoles that seek shelter or remain exposed after fleeing. We hypothesized that tadpoles that remain
exposed may use alternative strategies to avoid detection, such as reaching deeper microhabitats or positioning
themselves on substrates that confer greater crypsis than the ones they were on before disturbance. A significantly
greater proportion of tadpoles remained exposed after disturbance and positioned themselves on backgrounds that
offered greater cryptic potential, but did not differ in depth. Tadpoles may respond to a trade-off between sheltering
and being cryptic. On the one hand, they may remain close to retreat sites or they may escape to microhabitats
that provide appropriate background matching as a means to achieve crypsis. On the other hand, the absence of
matching backgrounds in the tadpoles’ vicinities may induce them to seek shelter. © 2010 The Linnean Society
of London, Biological Journal of the Linnean Society, 2010, 101, 437–446.
ADDITIONAL KEYWORDS: background choice – background matching – escape strategy.
INTRODUCTION
Defensive coloration is widespread among animals
and may work to reduce predator detection risk (camouflage), warn predators of potential harm through
unpalatable or toxic chemicals in the prey’s body
(aposematism) or mislead predators into mistaking
the prey for something else (mimicry) (Endler, 1978;
Stevens, 2007; Stevens & Merilaita, 2009). Among
these, camouflage constitutes a complex subject and
may include crypsis (when prey detection is prevented, for instance, through background matching,
disruptive coloration or distractive markings), masquerade responses (where the prey resembles some
component of the background, such as a leaf), motion
*Corresponding author. E-mail: eterovick@yahoo.com
dazzle or motion camouflage (which impair estimates
of speed and trajectory or movement detection by the
predator) (Endler, 2006; Stevens & Merilaita, 2009).
A combination of background matching and disruptive patterns is expected to provide the best means
of avoiding detection, especially in prey that use
many different backgrounds (Endler, 2006). However,
according to Merilaita & Lind (2005), cryptic potential
may vary, depending on what part of the background
the prey matches, so that resembling a random part
of the background is not necessarily sufficient to
achieve crypsis.
Clearly, predator visual perception is crucial for
defensive colour mechanisms to be effective (Stevens,
2007). Visual orientation is usually important in vertebrate predators, so they are prone to being deceived
by and/or responding to defensive colorations. For
instance, an increase in edge profiles, resulting from
© 2010 The Linnean Society of London, Biological Journal of the Linnean Society, 2010, 101, 437–446
437
438
P. C. ETEROVICK ET AL.
disruptive coloration, may prevent frog detection by
the garter snake Thamnophis sirtalis (Osorio & Srinivasan, 1991).
In tadpoles, behavioural defences against predators include spatial avoidance, hiding and decreased
activity level (Teplitsky & Laurila, 2007). Morphological changes may be induced by predators and vary
depending on the type of predator, being more efficient
against the inducing type (Relyea, 2003; Benard,
2006). Alternatively, greater investment in growth (by
increased feeding activity) may lead to a size refuge
or early metamorphosis, decreasing predation risk
as well (Van Buskirk, 2000). Defence mechanisms
in tadpoles constitute an interesting study subject. However, crypsis in tadpoles is little explored.
Predator-induced colour changes in Hyla chrysoscelis
tadpoles have been interpreted as a potential disruptive coloration (McCollum & Leimberger, 1997), but
they probably have the same effect as increased tail
height (also an inducible defence in tadpoles) in misdirecting predator strikes to the tail and away from the
vital body area (Van Buskirk et al., 2003; Richardson,
2006).
Tadpoles in open Cerrado and montane meadow streams are preyed upon by aquatic insects, such
as Heteroptera (Belostomatidae) and Anisoptera
(Eterovick & Barata, 2006). Tadpoles of some species
seek shelter to decrease predation risk (Kopp,
Wachlevski & Eterovick, 2006), while others, however,
remain exposed most of the time, resting on the bottom
substrate (Eterovick & Sazima, 2004). Although
aquatic insects are known to use movement as a cue to
locate prey (e.g. Victor & Ugwoke, 1987), which
induces reduced movement responses in tadpoles
(Skelly & Werner, 1990), visual stimuli may also be
important (e.g. Victor & Ugwoke, 1987). In fact, tadpoles of Hyla chrysoscelis exhibit changes in colour,
shape and growth rate in response to the presence of
dragonfly nymphs (McCollum & Leimberger, 1997);
these induced morphological changes are efficient at
reducing predation (McCollum & Van Buskirk, 1996).
Besides aquatic predators, it is known that tadpoles may also be preyed upon by birds and snakes
in Cerrado/montane meadow habitats (Kopp &
Eterovick, 2006). Such vertebrate predators may use
visual cues to detect tadpoles, so defensive colorations might evolve as a mechanism to avoid detection
in palatable species. Both terrestrial reptiles and
birds retain the five ancestral sets of visual pigments,
including four cone classes with different spectral sensitivities, which allow for the efficient coding
of spectral information, meaning that colour vision
is likely to be important for object detection and
classification (Osorio & Vorobyev, 2005).
Bokermannohyla alvarengai (Bokermann, 1956) is
a large tree frog (adult males reach 110 mm snout-
vent length) restricted to the Espinhaço mountain
range, south-eastern Brazil (Bokermann, 1956;
Eterovick & Sazima, 2004). The adults are nocturnal
and they may rest on rocks exposed to direct sunlight
during the day (Tattersall, Eterovick & Andrade,
2006). Their coloration resembles lichens, conferring
effective crypsis through background matching
(Sazima & Bokermann, 1977; Eterovick & Sazima,
2004; Tattersall et al., 2006) that may help them
avoid detection by visually oriented predators. Bokermannohyla alvarengai breeds during the rainy season
in temporary streams where it lays aquatic egg
clutches of c. 1500 eggs. The tadpoles are diurnal and
dwell in backwaters, usually on submerged rocks.
Their development takes approximately 4 months.
They are relatively small, attaining a maximum size
of 53 mm total length (Sazima & Bokermann, 1977),
have no bright colours likely to indicate unpalatability and remain in the streams for several
months, exposed to diurnal, visually oriented predators. Combined, these traits ensure that they constitute a good model to study strategies used by tadpoles
that remain exposed in shallow montane meadow
temporary streams, where detection by terrestrial vertebrates, especially birds and snakes, is
likely.
In order to understand how B. alvarengai tadpoles
manage to survive a potentially high predation risk
from visually oriented predators, we investigated
potential escape mechanisms related to active escape
(fleeing) and camouflage. We first studied the range
of distances covered by fleeing tadpoles and the proportion of tadpoles that sought shelter or remained
exposed after fleeing. We hypothesized that tadpoles that remained exposed after fleeing would use
alternative strategies to avoid detection, such as
(1) reaching deeper microhabitats or (2) positioning
themselves on substrates that confer greater background matching than the ones they were on before
disturbance. We assumed that deeper microhabitats
might make tadpoles harder to detect for terrestrial
predators and also give them a little extra time to
escape once detected.
MATERIAL AND METHODS
STUDY SITE
The Rio Preto State Park (43°30′676″W; 18°00′799″S)
is located in the state of Minas Gerais, on the
southern–central portion of the Espinhaço mountain
range, in south-eastern Brazil. It encompasses vegetation types belonging to the Cerrado biome, including
montane meadows. The climate is tropical wet (sensu
Ab’Saber, 1977), with two well-defined seasons: a wet,
warm season from October to March and a dry, cold
© 2010 The Linnean Society of London, Biological Journal of the Linnean Society, 2010, 101, 437–446
CRYPSIS OF THREATENED TADPOLES
season from April to September. Annual rainfall in the
region averaged 1351.22 mm from 1995 to 2003. Air
temperatures and humidity measurements gathered
during a 1-year study at two rivers in the park were
within the ranges of 8.5–38.0 °C and 20.0–89.0%,
respectively (Oliveira & Eterovick, 2009). The present
study was carried out in seasonal streams formed
during the wet season, which run amidst rocky outcrops and bushy fields, at elevations of between 1300
and 1400 m. These temporary streams shelter no fish,
but there are birds (e.g. Pitangus sulphuratus, Tyrannidae) and snakes (e.g. Xenodon merremii, Liophis
poecilogyrus, L. miliaris, Helicops modestus, Dipsadidae, sensu Zaher et al., 2009) in the area that can
potentially prey on tadpoles (A. M. Souza, pers.
comm.). There are also reports in the literature of some
of these species feeding on tadpoles in other areas (e.g.
Pitangus sulphuratus; Kopp & Eterovick, 2006; Helicops modestus; Sawaya, Marques & Martins, 2008)
and one of us (F.F.R.O.) saw an individual of Liophis sp.
inside one of the sampled streams.
TADPOLE
SAMPLING
We sampled tadpoles of Bokermannohyla alvarengai in seven montane meadow streams with heterogeneous bottoms containing sand, rocks, some leaf
litter, submerged grasses and pebbles, all forming a
mosaic. The extent of sampling and sampled area
varied among streams, according to B. alvarengai
tadpole presence. The water was very clear and
the sampled sections exhibited low depth values
439
(1–20 cm), allowing us to detect all tadpoles present
through careful examination of the whole bottom.
Deeper sections occur sporadically among rock crevices and we may have missed some tadpoles that were
originally hidden. However, we searched such microhabitats with nets and found no tadpoles, indicating
that they were usually exposed before we disturbed
them (as described below). We collected data on
tadpole escape behaviour in one stream on December
2006 and in another six streams in November and
December 2008 (Table 1). Based on 32 measured
specimens, which we believe are representative of our
whole sample, tadpole total size varied from 43 to
52 mm and tadpole developmental stage varied from
27 to 41 (sensu Gosner, 1960). Tadpole behaviour may
change with developmental stage, although it was not
our aim to test such variation in the present study. For
this purpose, it would be appropriate to have a representative sample of tadpoles in stages distributed
through the whole developmental period, which was
not the case, as we found tadpoles that could not have
varied by more than 20% of total body length in the
field. The maximum size of tadpoles of B. alvarengai is
attained in stage 40 (53 mm total length, Sazima &
Bokermann, 1977).
We carefully photographed each tadpole located
before it made any attempt to escape and then one of
us moved a finger from above (P.C.E. in December 2006
and F.F.R.O. in November and December 2008) slowly
and continuously towards the tadpole, inside the
water, until a escape response was observed (it always
happened before the finger touched the tadpole). This
Table 1. Mean depths of microhabitats occupied by Bokermannohyla alvarengai tadpoles before and after disturbance,
depth change (depth after–depth before) and distance covered when disturbed, by sampling dates and streams
Streams
S1
S2
S3
S4
S5
S6
S7
Sampling periods
December 2006
November 2008
December 2008
Totals (overall means)
Sample size*:
nexp (nphoto),
nshelter
Coordinates
18°11′19″S,
18°11′36″S,
18°12′42″S,
18°11′35″S,
18°12′39″S,
18°11′17″S,
18°08′09″S,
43°20′14″W
43°20′08″W
43°20′07″W
43°20′08″W
43°20′07″W
43°20′13″W
43°20′43″W
Streams sampled
S7
S1–S6
S1–S5
Depth
before
(cm)
Depth
after
(cm)
Depth
change
(cm)
Distance
covered
(cm)
(33), 12
(8), 3
(6), 1
(1), 1
(9), 1
(5), 1
(7), 10
9.59 ± 4.28
8.00 ± 4.00
9.29 ± 3.30
11.25 ± 4.35
8.23 ± 5.22
4.00 ± 1.41
9.46 ± 3.59
10.00 ± 4.82
8.50 ± 4.43
8.43 ± 4.08
7.50 ± 3.32
8.53 ± 5.61
4.50 ± 1.64
9.60 ± 3.68
0.42 ± 5.19
0.50 ± 2.31
-0.86 ± 4.45
-3.75 ± 7.32
0.29 ± 6.26
0.50 ± 2.88
0.14 ± 3.25
9.51 ± 3.91
10.07 ± 6.01
10.86 ± 4.06
12.00 ± 5.48
10.06 ± 3.86
11.00 ± 2.00
25.37 ± 17.84
21 (7), 10
30 (7), 7
59 (55), 12
9.46 ± 3.59
4.37 ± 1.96
10.98 ± 3.55
9.60 ± 3.68
4.77 ± 2.32
10.97 ± 4.34
0.14 ± 3.25
0.40 ± 2.21
-0.02 ± 5.91
25.37 ± 17.84
9.20 ± 3.04
10.44 ± 4.67
110 (69), 29
8.86 ± 4.29
8.99 ± 4.62
0.13 ± 4.73
12.36 ± 9.58
41
14
7
4
17
6
21
*All values refer to tadpoles that remained exposed after attempting to escape (nexp) and are given as mean ± SD. The sample
sizes presented include tadpoles that did not seek shelter after disturbance and remained exposed (nexp); among these, the number
of tadpoles that could be successfully photographed both before and after disturbance (nphoto) and the number of tadpoles that
sheltered themselves after disturbance (nshelter).
© 2010 The Linnean Society of London, Biological Journal of the Linnean Society, 2010, 101, 437–446
440
P. C. ETEROVICK ET AL.
stimulus was used to elicit a escape response in a
controlled way that would allow us to make minimal
disturbance in the water surface and thus observe
where all tadpoles went after disturbed. Different
predators may have different approaches and elicit
different responses and whether escape speed or
covered distance vary depending on type of predator
is an interesting subject that deserves further study.
But our aim here was to study tadpole behaviour in
response to a standardized frightening stimulus. A
similar approach has already been used to study choice
of cryptic backgrounds by grasshoppers (Eterovick,
Figueira & Vasconcellos-Neto, 1997).
We recorded whether the tadpole was exposed or
sheltered after moving away from the disturbance and,
whenever it was exposed, we took another photograph.
We then measured with a metric tape and recorded
the distance covered by the tadpole from its initial
to its final position, the depth of both positions and
the distance from the tadpole’s initial position to the
closest structure that could potentially provide some
shelter (under which the tadpole could hide). We took
these measurements in situ, after we were finished
photographing the tadpoles. These structures included
rock crevices, submerged herbs or leaf litter (we considered whichever was the closest to the tadpole’s
original position). Movements were usually straight,
but in a single instance a tadpole made a curved
trajectory, which we considered in determining its
escape distance. We moved upstream while making
observations to avoid sampling the same tadpole more
than once.
In all sampling periods we conducted tadpole observations while natural light allowed good visualization
of tadpoles (between 11:00 and 17:00 h). We discarded
pictures that were blurry or had light reflections that
interfered with tadpole or background colour visualization. We took pictures of tadpoles with an automatic
Sony Cyber Shot DSC-P100 camera with Carl Zeiss
lenses, 3 ¥ optical zoom, 5.1 megapixels in December
2006 and with a Canon Powershot A630 (in automatic mode, with 2.8 aperture and 1/13–1/40 speed) in
November and December 2008. We positioned the
cameras almost perpendicular to the water surface,
but somewhat at an angle to minimize reflections while
still providing a dorsal view of the tadpole. We tested
for differences between sampling periods, which would
also account for differences between cameras when
conducting colour measurements and analyses (see
‘Colour measurement and analyses’). According to
Stevens et al. (2007), calibration for calculating reflection from digital photographs needs to be performed for
each session/light, because the light set-up may change
the ratio between different wavelengths (long, medium
or short waves). We did not perform such calibration
during our fieldwork because we were comparing pic-
tures of the same tadpole which were taken less than
30 s apart, so there was likely not enough time for light
conditions to change. Additionally, even if we considered that light conditions could still change, any
distortion caused in a particular wavelength would be
corrected for when we calculated the differences
between the tadpole and the background it was on,
because any detection-based distortion in the wavelengths would be the same for the tadpole and its
background. Furthermore, the tadpole was the same
individual in both pictures whose difference between
tadpole/background would be compared (see ‘Colour
measurement and analyses’). This way, the tadpole
itself worked as a calibration for background colour
when differences between tadpole and background
were calculated. Furthermore, any inter-individual
variation in the colour estimation of the images would
have decreased the chances of detecting significant
differences by increasing the error term in our statistical analysis.
TADPOLE
MICROHABITAT DEPTH AND
ESCAPE DISTANCE
We tested whether proportions of tadpoles escaping to
shelters and remaining exposed varied among different streams using a c2-test (7 ¥ 2 contingency table,
Ayres et al., 2005) to account for a possible effect of
local stream features or population on tadpole
response to disturbance. We then compared the mean
number of tadpoles sheltered and exposed using a
Mann–Whitney U-test to assess whether tadpoles
exhibited a preferential response to disturbance. We
compared the distance from a tadpole’s initial position
to the closest potential shelter (rock crevice, leaf litter
or aquatic vegetation) between tadpoles that used and
that did not use these shelters with a Mann–Whitney
U-test.
We compared data on depth (both before and after
disturbance), depth variation between microhabitats
occupied before and after disturbance and distance
covered by tadpoles during escape between sampling
dates and among streams using Kruskal–Wallis
tests. We wanted to know whether sampling date or
stream influenced the estimated parameters or not;
if not, we could consider dates and streams as replicates and pool all data for subsequent analyses. We
then compared depths of microhabitats used before
and after disturbance using a Wilcoxon test to test
the hypothesis that disturbed tadpoles would escape
towards deeper microhabitats. We used Shapiro–
Wilk tests to test for data normality and chose statistical tests accordingly. We conducted all statistical
tests using the software Systat (Systat Software,
2007).
© 2010 The Linnean Society of London, Biological Journal of the Linnean Society, 2010, 101, 437–446
CRYPSIS OF THREATENED TADPOLES
COLOUR
MEASUREMENT AND ANALYSES
To test whether tadpoles that remained exposed were
maximizing their cryptic potential through background matching, we compared their colour with background colour both in their initial position and in their
position after escaping. Background vs. prey texture
(colour patch size) variation is also important for
camouflage (e.g. Shohet et al., 2007). When patches of
the background that exhibit noticeable colour variation are smaller than the prey, it requires a sufficiently
accurate match to achieve good camouflage (Merilaita,
2003). But, in the present study, we did not make any
measurement of texture. Most substrates (rocks, sand)
441
were relatively uniform and similar in colour to the
tadpoles’ bodies. A few tadpoles positioned themselves
on mosaic substrates such as pebbles, but as the
texture of such substrates was not smaller than the
tadpole’s bodies, we considered that matching parts
of the substrate would still provide camouflage (for
examples of substrates, see Fig. 1).
The backgrounds used by these tadpoles are mottled,
with grain size similar to the colour grain size in
tadpoles’ bodies. The exceptions are rare pebbles/rocks,
which can be seen in Figure 1, that shows examples of
the different types of substrates observed. Transitions
between colours in the background or in the tadpoles
were rarely abrupt and, even when light and dark
Figure 1. Pictures of Bokermannohyla alvarengai tadpoles showing an example of tadpole position (A) before and (B) after
disturbance; layers created in Adobe Photoshop for colour analyses of (C) tadpole and (E) background; and examples of
tadpoles on (D) sandy and (F) rocky substrates. Notice the similarity between tadpole and background texture. The two light
dots on tadpole dorsum (arrows) may function as distractive markings against light backgrounds. Scale bar, 10 mm (A).
© 2010 The Linnean Society of London, Biological Journal of the Linnean Society, 2010, 101, 437–446
442
P. C. ETEROVICK ET AL.
tones were present, there were always intermediate
shades in the background mosaic, so we believe the
method we used was appropriate to show tadpole
cryptic potential on their background. Although we
are not specifically looking at disruptive patterns, a
tadpole sitting amongst alternating contrasting backgrounds would be difficult for any animal to see in the
type of habitat where they occur. Under such conditions, in order for a tadpole to be visible, the lightness
or amount of different wavelengths should be, in
general, lighter or darker in its background and our
method would detect this situation.
Colour analyses can be performed in several colour
spaces, which consist of Cartesian spaces where the
visual sensation of a colour can be uniquely defined
by a set of numbers representing chromatic features
(Drimbarean & Whelan, 2001). The ‘Lab’ colour space
was originally proposed by Hunter (1948) and modified by the International Commission on Illumination
(CIE, 1986) to what is currently called ‘CIE L*a*b*’.
The ‘L*’ component refers to a scale of lightness
(low values are dark, whereas high values are light),
whereas the ‘a*’ and ‘b*’ components represent colour
changes through the red–green and the yellow–blue
spectra, respectively.
For an appropriate colorimetric test, the lightness component must be analysed separately from
the a*b* component, as the lightness is critical for the
spatial formation of the image (McCormick-Goodhart
& Wilhelm, 2003). In computer image manipulation,
a* and b* can be adjusted in levels corresponding
to 6-unit increments resulting in changes noticeable to the human eye, without any interference in
L*. Their simultaneous adjustment, that is, a combined analysis of both a* and b*, can be defined
as Da*b* = √(Da*2 + Db*2) (McCormick-Goodhart &
Wilhelm, 2003).
We used Adobe Photoshop CS2, version 9.0.2
(Adobe Photoshop 1999–2005, Adobe Systems Incorporated) to quantify L*, a* and b* components in the
pictures of Bokermannohyla tadpoles before and after
disturbance. We assessed the components in the tadpoles by selecting them as a separate layer (Fig. 1). In
order to assess background components, we arbitrarily marked five elliptical shapes throughout the
background adjacent to the tadpole, merged them in
a separate layer and proceeded to a component measurement of the whole set of five elliptical shapes, to
obtain a general background measurement (Fig. 1).
We marked elliptical shapes close to tadpole body
shape and size and we tried to encompass all substrates surrounding the tadpoles in the available proportions. We avoided parts of the pictures that
eventually had light reflections on water surface or
the tadpole’s shadow. We used the means of L*, a*
and b* components to calculate |DL*| and Da*b* for
tadpole vs. background colour before and after disturbance, as follows:
ΔL* before = L* bb − L*tb ;
ΔL* after = L* ba − L*ta ;
Δa*b* before = [(a*bb − a*tb)2 + (b* bb − b*tb)2 ];
Δa*b*after = [(a* ba − a*ta)2 + (b* ba − b*ta)2 ];
where bb indicates background before tadpole disturbance, tb is tadpole before disturbance; ba is background after tadpole disturbance and ta is tadpole
after disturbance.
We performed paired t-tests to check whether tadpoles moved towards backgrounds with lightness (L*)
and colours (a*b*) more similar to their own.
We compared |DL*| and Da*b* among streams and
among sampling periods using Kruskal–Wallis tests.
Differences in L*, a* and b* among pictures are likely
to be caused by light differences throughout the day.
However, by always using the difference in brightness
and colour between tadpole and background, which
were in the same picture, we believe these variations were controlled. Besides, we considered that this
variation is part of the natural context of camouflage,
as predators may attack at different times and under
different light conditions.
RESULTS
When disturbed, tadpoles of B. alvarengai always
reacted by fleeing away from the stimulus, but they
usually covered a short distance (12.36 ± 9.58 cm;
Table 1). After this initial swimming burst, tadpoles
either retreated to shelters or remained exposed and
motionless. The proportion of tadpoles that sheltered
themselves and that remained exposed after disturbance did not change significantly among streams
(c2-test = 5.73, d.f. = 6, P = 0.455), so we could pool
samples from all streams to compare numbers of
tadpoles sheltering and not sheltering. A significantly
greater number of tadpoles remained exposed (110
out of 139) after attempting to escape from a potential threat (Mann–Whitney U-test statistic = 6.00,
P = 0.017). The tadpoles that moved to shelters when
disturbed (29 out of 139) were significantly closer to
potential shelters before disturbance than the ones
that remained exposed after disturbance (Mann–
Whitney U-test statistic = 874.50, P = 0.016). The shelters most frequently used were crevices among rocks
(23 tadpoles), followed by leaf litter (three tadpoles)
and aquatic vegetation (three tadpoles), which was in
accordance with a higher availability of rock crevices
as potential shelters (among 108 records of the type of
© 2010 The Linnean Society of London, Biological Journal of the Linnean Society, 2010, 101, 437–446
CRYPSIS OF THREATENED TADPOLES
shelter closest to the tadpole, 97 were rock crevices,
eight were leaf litter and four were aquatic vegetation).
The proportional use of the three types of shelter was
not significantly different from expected based on
shelter availability, indicating no preference by the
tadpoles (c2-test = 2.51, d.f. = 2, P = 0.284).
Depths of microhabitats occupied by tadpoles before
and after disturbance varied among sampling dates
(depth before: Kruskal–Wallis test statistic = 50.68,
P < 0.001; depth after: Kruskal–Wallis test statistic = 43.46, P < 0.001) but not among streams (depth
before: Kruskal–Wallis test statistic = 12.15, P =
0.059; depth after: Kruskal–Wallis test statistic =
10.43, P = 0.108). However, the depth difference
between microhabitats occupied before and after disturbance was not significant among sampling dates
(Kruskal–Wallis test statistic = 0.406, P = 0.816)
or streams (Kruskal–Wallis test statistic = 1.115,
P = 0.981), so we pooled data from all dates and
streams to test whether tadpoles move to deeper
microhabitats when disturbed. Distances covered by
tadpoles when disturbed varied among sampling dates
(Kruskal–Wallis test statistic = 17.36, P < 0.001) and
among streams (Kruskal–Wallis test statistic = 18.23,
P < 0.001). The distances covered were longer in the
stream sampled in December 2006 (S7, Table 1). We
conducted the same analysis, excluding data from this
stream (all from December 2006), and the distances
no longer varied among sampling periods (Mann–
Whitney U-test statistic = 744.50, P = 0.221) or
streams (Kruskal–Wallis test statistic = 2.69, P =
0.747).
When disturbed, tadpoles did not move to deeper
microhabitats (Z = 0.297, P = 0.766). However, they
moved to backgrounds significantly more similar to
them in lightness (mean |DL*|before = 32.42 and mean
|DL*|after = 25.37; t = 3.63, d.f. = 68, P < 0.001) and
in colour (mean Da*b* before = 8.50 and mean
Da*b*after = 7.39; t = 1.70, d.f. = 68, P = 0.046). For
colour analyses, we also pooled data from all sampling periods as there was no significant difference
among them (|DL*|before: Kruskal–Wallis test statistic = 3.04, P = 0.218; |DL*|after: Kruskal–Wallis test
statistic = 1.32, P = 0.518; Da*b*before: Kruskal–Wallis
test statistic = 5.41, P = 0.067; Da*b*after: Kruskal–
Wallis test statistic = 2.96, P = 0.228). There was
no significant difference among streams in |DL*|before
(Kruskal–Wallis test statistic = 6.10, P = 0.412),
|DL*|after (Kruskal–Wallis test statistic = 3.79, P =
0.705) or Da*b*after (Kruskal–Wallis test statistic =
9.90, P = 0.129). Da*b*before was significantly different
among streams (Kruskal–Wallis test statistic = 19.73,
P = 0.003), but we still pooled the data from all
streams to make subsequent tests more robust. We
were mostly interested in the tadpole’s escape behaviours, which were expressed after disturbance, when
443
there were no significant differences in the measured
parameters (|DL*|after and Da*b*after) among streams.
DISCUSSION
Tadpoles of B alvarengai attempted to escape the
threatening stimulus imposed on them by fleeing and
either sheltering themselves or remaining exposed
and motionless. The availability of suitable shelters
close to the tadpole may be important in determining the outcome of the escape strategy, as tadpoles
that chose to shelter were initially closer to potential
shelters than tadpoles that remained exposed. In
these circumstances, tadpoles seem to use any type of
shelter, showing no preferences, which is in accordance with an urgency to reach the closest available
shelter to escape predation. Tadpoles covered short
distances relative to their body size (mean distance
covered was approximately three times their body
size) when escaping, in what may be related to energetic or burst swimming constraints (Dayton et al.,
2005). It is also possible that fast tadpole movement
gives a signal to the predator and then this signal is
lost when the tadpole suddenly stops and remains
motionless upon a matching background. Prey movement is usually necessary for first detection, but
background matching allows the prey, once detected,
to ‘freeze’ and merge again into the background before
the predator can strike (Osorio & Srinivasan, 1991).
On the one hand, a significantly larger number of
tadpoles remained exposed after being disturbed
and they increased background matching, as demonstrated here, when they stopped moving, indicating that this hypothesis is likely. On the other hand,
tadpoles did not seek deeper microhabitats, maybe
because choosing the deepest microhabitats in the
depth gradient available in the shallow streams
where they live would not render this strategy
efficient to impair predator capture.
Studies conducted on tadpole vision of some species
show that they are myopic (Hoff et al., 1999). Their
vision seems to improve during development, representing a refractive index gradient, especially during
metamorphosis, probably attributable to their change
from aquatic to terrestrial habitat (Mathis, Schaeffel &
Howland, 1988). The range of developmental stages of
B. alvarengai we used here are not supposed to include
great changes in vision, as they were not very close to
metamorphosis, so they could all be considered myopic
(see Mathis et al., 1988). This could be another reason
for their short-distance movements; they may choose
microhabitats/shelters within their visual field. Otherwise, they may choose microhabitats based on cues
other than visual ones, which may instead coincide
with light/colour (see Hoff et al., 1999).
© 2010 The Linnean Society of London, Biological Journal of the Linnean Society, 2010, 101, 437–446
444
P. C. ETEROVICK ET AL.
It is also interesting to notice that features of the
tadpoles other than lightness and colour seemed to
contribute to crypsis. When on sandy backgrounds, the
lighter outline of the tadpole’s body seemed to help
hinder detection of tadpole’s body shape (obliterative
shading, sensu Stevens & Merilaita, 2009; fig. 1D),
while the two light spots on the dorsum (Fig. 1A)
seemed to work as ‘dazzle’ markings (distractive markings, sensu Stevens & Merilaita, 2009), which are
believed to act like distractors drawing the eyes of the
predator away from the body outline, preventing recognition (Stevens, 2007). Although such colour features are likely to be deceptive to visually oriented
predators, their actual effects on predator avoidance
remain untested for B. alvarengai tadpoles.
The presence of both physical and behavioural features that enhance crypsis adds evidence to its adaptive importance for B. alvarengai tadpole survivorship.
Animals that are exposed to different kinds of backgrounds may minimize predation risk by achieving a
compromise cryptic strategy that confers some advantage in more than one background (Houston, Stevens
& Cuthill, 2007). This may be the case for B. alvarengai
tadpoles, which exhibit a combination of colours, features and behaviours that potentially increase crypsis.
Besides, it has been demonstrated previously that
complex backgrounds facilitate selection for cryptic
mechanisms (Merilaita, 2003) and the structurally
complex backgrounds of montane meadow streams are
likely good for tadpole crypsis.
Differences in stream characteristics could result
in differences in tadpole escape opportunities and
should be accounted for. For instance, the differences
observed in distance covered by fleeing tadpoles in the
stream sampled in December 2006 (S7) in relation to
the remaining streams (S1–S6) could be because of
physical differences of the stream beds. Alternatively,
tadpoles may change their behaviour in response to
changes in light, temperature or depending on their
daily activity patterns. It is also possible that this
result was influenced by different observers disturbing tadpoles, although we made an effort to standardize all sampling procedures. Differences in tadpole
behaviour among streams could also be as a result of
genetic differences among populations; however, there
is no information available on the genetic structure of
B. alvarengai populations to test this hypothesis.
The differences between resting tadpoles (before
disturbance) and background were significantly higher
in some streams than in others, but, after disturbance,
escaping tadpoles always achieved the same result of
increasing similarity to background visual signals. It is
possible that the differences observed before disturbance are because of different substrate availability
among streams. If this is true, tadpoles were able to
increase their cryptic potential, even in streams that
offered more contrasting backgrounds, and to attain
the same level of background matching achieved by
tadpoles in streams with backgrounds more similar to
their own coloration. We did not evaluate to what
extent (if any) tadpoles can change their colours in
order to match background. However, the time interval
between the moments when we disturbed and photographed each tadpole after fleeing did not exceed 30 s,
minimizing the eventual influence of colour changes, if
they happen. It has been previously shown that adult
individuals can change colour, but the process is slow
(Tattersall et al., 2006). Thus, we interpret our results
as an effect of behavioural maximization of crypsis
through background selection for background matching. It is also possible that some tadpoles had been on
the backgrounds where we detected them for a shorter
time in some instances than in others, possibly having
less time to adjust their colour accordingly. Even if it is
the case, tadpoles were still able to choose a better
matching background in a short time interval when
disturbed, showing that, even if physiological colour
change happens and aids to crypsis, the behavioural
component is still important and is probably more
efficient to escape imminent predator attacks.
Tadpoles of B. alvarengai and other species are
known to select microhabitats in a non-random way in
montane meadow streams (Eterovick & Barros, 2003).
Although B. alvarengai has a moderately broad spatial
niche in the larval stage (Eterovick & Barros, 2003), it
is not known whether preferences and niche breadth
vary individually. In order to achieve better crypsis
and increase survival chances, B. alvarengai tadpoles
need to be able to select the most appropriate microhabitats that may be achieved through natural
selection or tadpoles’ perception of cryptic potential.
Tadpoles that succeed in choosing better matching
backgrounds will have greater chances of surviving,
whether they can actually see the substrate or use any
other cue that will produce the same result. As has
been shown for the salamander Plethodon cinereus,
behaviours that improve an individual’s ability to
escape predation may be selected for according to
individual phenotype (Venesky & Anthony, 2007).
Thus, an animal may be able to select backgrounds and
position itself in ways that decrease predation risk
through improved crypsis, as shown for grasshoppers
(Eterovick et al., 1997). As prey motion is crucial for
detection by predators, aquatic prey can also position
themselves in ways that minimize involuntary motion
caused by currents, as seen in cuttlefish (Sepia pharaonis and S. officinalis, Shohet et al., 2006). Tadpoles of
B. alvarengai may be under selection to have colour
types that provide crypsis through background matching in their natural habitat and behaviours that allow
them to choose the appropriate background to match
their colour while remaining motionless. This is an
© 2010 The Linnean Society of London, Biological Journal of the Linnean Society, 2010, 101, 437–446
CRYPSIS OF THREATENED TADPOLES
interesting subject that deserves further study, including studies focusing on specific predators and their
particular visual systems (see Stevens et al., 2007).
Colour differences between tadpole and background
after disturbance (Da*b* = 7.39) were still high enough
to be noticeable to the human eye (see McCormickGoodhart & Wilhelm, 2003). However, considering
that birds may have lower contrast sensitivity than
humans (Ghim & Hodos, 2006), the reduction in
lightness and colour contrasts relative to background
achieved by escaping tadpoles may be enough to
deceive predatory birds. It is important to consider
that birds probably have separate sets of receptors for
luminance and colour vision (Osorio & Vorobyev, 2005)
and B. alvarengai tadpoles managed to decrease their
contrast from the background in both these signals.
Birds and many other animals have a visible spectrum spanning 300- to 700-nm range wavelength,
whereas humans see within wavelengths of 400–
700 nm (Endler & Mielke, 2005). We have no information on whether B. alvarengai tadpoles and their
backgrounds reflect light in wavelengths between 300
and 400 nm in order to evaluate a potential effect of
these wavelengths on tadpole crypsis.
In this study we showed, for the first time, that
tadpoles can increase cryptic potential by selecting
microhabitats that better match their lightness and
colour. When disturbed, tadpoles of B. alvarengai may
seek shelters, provided they are available within a
short distance. Otherwise, they may remain motionless after a short escape response, on a matching
background. We believe therefore that a potential
trade-off in site selection exists for tadpoles, whereby
they select microhabitats that offer the appropriate
proximity to retreat sites; failing that potential, they
will escape to microhabitats that provide appropriate
background matching. Alternatively, the absence of
matching backgrounds in the tadpoles’ vicinities may
induce them to seek shelter. The use of defensive
colorations by tadpoles is still little explored and
constitutes a very interesting and complex subject.
Future studies should examine the real effects on
predation risk offered by different background types in
order to explain tadpole behavioural adaptations.
ACKNOWLEDGEMENTS
We are thankful to C. M. F. Morais and D. A. Coelho
for help in an initial phase of this work, during
the Herpetology field course of the Programa de Pós
Graduação em Zoologia de Vertebrados, PUC Minas,
and to A. M. Souza for providing useful information on
tadpole potential predators that occur in the Park.
Permits were provided by Ibama (12813-1), and IEF
(Instituto Estadual de Florestas; 085/06). Financial
support was provided by FIP (Fundo de Incentivo
445
à Pesquisa) PUC Minas, and Fapemig (Fundação
de Amparo à Pesquisa do Estado de Minas Gerais).
A Research Productivity grant (305889/2007-9) was
provided to P.C.E. by CNPq.
REFERENCES
Ab’Saber AN. 1977. Os domínios morfoclimáticos na América
do Sul. Primeira aproximação. Geomorfologica 52: 1–21.
Ayres M, Ayres Jr M, Ayres DL, Santos AS. 2005. Bioestat.
Versão 4.0. Belém, Pará, Brazil: Sociedade Civil Mamirauá,
MCT-CNPq.
Benard MF. 2006. Survival trade-offs between two predator
induced phenotypes in Pacific treefrogs (Pseudacris regilla).
Ecology 87: 340–346.
Bokermann WCA. 1956. Sobre uma nova especie de Hyla
do estado de Minas Gerais, Brasil (Amphibia, Salientia,
Hylidae). Papéis Avulsos do Departamento de Zoologia, São
Paulo XII: 357–362.
CIE (International Commission on Illumination). 1986.
Colorimetry. Publication 15.2. 2nd edn. Vienna, Austria:
CIE Central Bureau.
Dayton GH, Saenz D, Baum KA, Langerhans RB, DeWill
TJ. 2005. Body shape, burst speed and escape behavior of
larval anurans. Oikos 111: 582–591.
Drimbarean A, Whelan PF. 2001. Experiments in colour
texture analysis. Pattern Recognition Letters 22: 1161–1167.
Endler JA. 1978. A predator’s view of animal color patterns.
Evolutionary Biology 11: 319–364.
Endler JA. 2006. Disruptive and cryptic coloration.
Proceedings of the Royal Society B 273: 2425–2426.
Endler JA, Mielke Jr PW. 2005. Comparing entire color
patterns as birds see them. Biological Journal of the
Linnean Society 86: 405–431.
Eterovick PC, Barata IM. 2006. Distribution of tadpoles
within and among Brazilian streams: the influence of predators, habitat size and heterogeneity. Herpetologica 62: 365–
377.
Eterovick PC, Barros IS. 2003. Niche occupancy in southeastern Brazilian tadpole communities in montane-meadow
streams. Journal of Tropical Ecology 19: 439–448.
Eterovick PC, Figueira JEC, Vasconcellos-Neto J. 1997.
Cryptic coloration and choice of escape microhabitats by
grasshoppers (Orthoptera: Acrididae). Biological Journal of
the Linnean Society 61: 485–499.
Eterovick PC, Sazima I. 2004. Anfíbios da Serra do
Cipó: Amphibians from the Serra do Cipó. Belo Horizonte,
Brazil: Editora PUC Minas.
Ghim MM, Hodos W. 2006. Spatial contrast sensitivity of
birds. Journal of Comparative Physiology A 192: 523–534.
Gosner KL. 1960. A simplified table for staging anuran
embryos and larvae with notes on identification. Herpetologica 16: 183–190.
Hoff KvS, Blaustein AR, McDiarmid RW, Altig R.
1999. Behavior. Interactions and their consequences. In:
McDiarmid RW, Altig R, eds. Tadpoles. The Biology of
Anuran Larvae. Chicago and London: The University of
Chicago Press, 215–239.
© 2010 The Linnean Society of London, Biological Journal of the Linnean Society, 2010, 101, 437–446
446
P. C. ETEROVICK ET AL.
Houston AI, Stevens M, Cuthill IC. 2007. Animal camouflage: compromise or specialize in a 2 patch-type environment? Behavioral Ecology 18: 769–775.
Hunter RS. 1948. Accuracy, precision, and stability of new
photoelectric color-difference meter. Journal of the Optical
Society of America 38: 1094.
Kopp K, Eterovick PC. 2006. Factors influencing spatial
and temporal structure of frog assemblages at ponds in
southeastern Brazil. Journal of Natural History 40: 1813–
1830.
Kopp K, Wachlevski M, Eterovick PC. 2006. Environmental complexity reduces tadpole predation by water bugs.
Canadian Journal of Zoology 84: 136–140.
Mathis U, Schaeffel F, Howland HC. 1988. Visual
optics in toads (Bufo americanus). Journal of Comparative
Physiology A 163: 201–213.
McCollum SA, Leimberger JD. 1997. Predator-induced
morphological changes in an amphibian: predation by dragonflies affects tadpole shape and color. Oecologia 109: 615–
621.
McCollum SA, Van Buskirk J. 1996. Costs and benefits of
a predator-induced polyphenism in the gray treefrog Hyla
chrysoscelis. Evolution 50: 583–593.
McCormick-Goodhart M, Wilhelm H. 2003. A new test
method based on CIELAB colorimetry for evaluating the
permanence of pictorial images. Wilhelm Imaging Research,
Inc. Available at http://www.wilhelm-research.com/articles_
wir_additional.html.
Merilaita S. 2003. Visual background complexity facilitates
the evolution of camouflage. Evolution 57: 1248–1254.
Merilaita S, Lind J. 2005. Background-matching and disruptive coloration, and the evolution of cryptic coloration.
Proceedings of the Royal Society B 272: 665–670.
Oliveira FFR, Eterovick PC. 2009. The role of river longitudinal gradients, local and regional attributes in shaping
frog assemblages. Acta Oecologica 35: 727–738.
Osorio D, Srinivasan MV. 1991. Camouflage by edge
enhancement in animal coloration patterns and its implications for visual mechanisms. Proceedings of the Royal
Society B 244: 81–85.
Osorio D, Vorobyev M. 2005. Photoreceptor spectral sensitivities in terrestrial animals: adaptations for luminance
and colour vision. Proceedings of the Royal Society B 272:
1745–1752.
Relyea RA. 2003. How prey respond to combined predators:
a review and an empirical test. Ecology 84: 1827–1839.
Richardson JL. 2006. Novel features of an inducible defense
system in larval treefrogs (Hyla chrysoscelis). Ecology 87:
780–787.
Sawaya RJ, Marques OAV, Martins M. 2008. Composição
e história natural das serpentes de Cerrado de Itirapina,
São Paulo, sudeste do Brasil. Biota Neotropica 8: 127–
149.
Sazima I, Bokermann WCA. 1977. Anfíbios da Serra
do Cipó, Minas Gerais, Brasil. 3: Observações sobre a
biologia de Hyla alvarengai Bok. (Anura, Hylidae). Revista
Brasileira de Biologia 37: 413–417.
Shohet AJ, Baddeley RJ, Anderson JC, Kelman EJ,
Osorio D. 2006. Cuttlefish responses to visual orientation
of substrates, water flow and a model of motion camouflage.
Journal of Experimental Biology 209: 4717–4723.
Shohet A, Baddeley R, Anderson J, Osorio D. 2007.
Cuttlefish camouflage: a quantitative study of patterning.
Biological Journal of the Linnean Society 92: 335–345.
Skelly DK, Werner EE. 1990. Behavioral and life-historical
responses of larval American toads to an odonate predator.
Ecology 71: 2313–2322.
Stevens M. 2007. Predator perception and the interrelation
between different forms of protective coloration. Proceedings
of the Royal Society B 274: 1457–1464.
Stevens M, Merilaita S. 2009. Animal camouflage: current
issues and new perspectives. Philosophical Transactions of
the Royal Society B 364: 423–427.
Stevens M, Párraga CA, Cuthill IC, Partridge JC, Troscianko TS. 2007. Using digital photography to study
animal coloration. Biological Journal of the Linnean Society
90: 211–237.
Systat Software I. 2007. Systat 12 for Windows. San Jose:
Systat Software Inc.
Tattersall GJ, Eterovick PC, Andrade DV. 2006. Tribute
to R. G. Boutilier: Skin colour and body temperature
changes in basking Bokermannohyla alvarengai (Bokermann, 1956). Journal of Experimental Biology 209: 1185–
1196.
Teplitsky C, Laurila A. 2007. Flexible defense strategies:
competition modifies investment in behavioral vs. morphological defenses. Ecology 88: 1641–1646.
Van Buskirk J. 2000. The costs of an inducible defense in
anuran larvae. Ecology 81: 2813–2821.
Van Buskirk J, Anderwald P, Lüpold S, Reinhardt L,
Schuler H. 2003. The lure effect, tadpole tail shape, and
the target of dragonfly strikes. Journal of Herpetology 37:
420–424.
Venesky MD, Anthony CD. 2007. Antipredator adaptations
and predator avoidance by two color morphs of the eastern
red-backed salamander, Plethodon cinereus. Herpetologica
63: 450–458.
Victor R, Ugwoke LI. 1987. Preliminary studies on predation by Sphaerodema nepoides Fabricius (Heteroptera:
Belostomatidae). Hydrobiologia 154: 25–32.
Zaher H, Grazziotin FG, Cadle JE, Murphy RW, MouraLeite JC, Bonatto SL. 2009. Molecular phylogeny of
advanced snakes (Serpentes, Caenophidia) with an emphasis on South American Xenodontines: a revised classification
and descriptions of new taxa. Papéis Avulsos de Zoologia,
São Paulo 49: 115–153.
© 2010 The Linnean Society of London, Biological Journal of the Linnean Society, 2010, 101, 437–446