Content uploaded by Bibiana Rojas
Author content
All content in this area was uploaded by Bibiana Rojas on Mar 03, 2016
Content may be subject to copyright.
Content uploaded by John A Endler
Author content
All content in this area was uploaded by John A Endler on Jan 19, 2015
Content may be subject to copyright.
rsbl.royalsocietypublishing.org
Research
Cite this article: Rojas B, Devillechabrolle J,
Endler JA. 2014 Paradox lost: variable
colour-pattern geometry is associated with
differences in movement in aposematic frogs.
Biol. Lett. 10: 20140193.
http://dx.doi.org/10.1098/rsbl.2014.0193
Received: 3 March 2014
Accepted: 25 May 2014
Subject Areas:
behaviour, evolution, ecology
Keywords:
predator–prey interactions, warning signals,
polymorphism, visual illusions, poison frog
Author for correspondence:
Bibiana Rojas
e-mail: bibiana.rojas@jyu.fi
Electronic supplementary material is available
at http://dx.doi.org/10.1098/rsbl.2014.0193 or
via http://rsbl.royalsocietypublishing.org.
Animal behaviour
Paradox lost: variable colour-pattern
geometry is associated with differences
in movement in aposematic frogs
Bibiana Rojas
1,2
, Jennifer Devillechabrolle
3
and John A. Endler
1
1
Centre for Integrative Ecology, School of Life and Environmental Sciences, Deakin University, 75 Pigdons Road,
Geelong, Victoria 3216, Australia
2
Centre of Excellence in Biological Interactions, Department of Biology and Environmental Sciences,
University of Jyva
¨
skyla
¨
, PO Box 35, Jyva
¨
skyla
¨
40014, Finland
3
Association de Gestion des Espaces Prote
´
ge
´
s, Rue Cle
´
mencin Ne
´
ron, Regina 97390, French Guiana
Aposematic signal variation is a paradox: predators are better at learning and
retaining the association between conspicuousness and unprofitability when
signal variation is low. Movement patterns and variable colour patterns are
linked in non-aposematic species: striped patterns generate illusions of altered
speed and direction when moving linearly, affecting predators’ tracking abil-
ity; blotched patterns benefit instead from unpredictable pauses and random
movement. We tested whether the extensive colour-pattern variation in an
aposematic frog is linked to movement, and found that individuals moving
directionally and faster have more elongated patterns than individuals
moving randomly and slowly. This may help explain the paradox of poly-
morphic aposematism: variable warning signals may reduce protection, but
predator defence might still be effective if specific behaviours are tuned to
specific signals. The interacting effects of behavioural and morphological
traits may be a key to the evolution of warning signals.
1. Introduction
Some animals inform predators about their unprofitability with warning colour
patterns (aposematism [1]). Because this relies on predators learning and remem-
bering the relationship between colour patterns and unprofitability, selection
should favour colour patterns with little or no variation [2]; patterns with low
variation are easier to learn [2,3] and remember [4]. Paradoxically, within-
population variation exists in some aposematic species [5,6]. Despite studies
suggesting a natural–sexual selection interaction as a cause [5,7], the mechanisms
by which within-population variation is maintained, and its ecological and
behavioural correlates, are poorly understood. Attempts to explain the mainten-
ance of variable aposematic signals have primarily addressed among-population
variation [8,9] or involved laboratory studies with artificial prey or computer
games. Here, we address it using a wild natural population.
In non-aposematic snakes with variable coloration, certain patterns seem to
be more effective when accompanied by matched escape behaviours. For
example, individuals with striped patterns benefit from fleeing from predators,
because the pattern appears to remain stationary or move more slowly than the
escaping animal [10–13], whereas individuals with spotted patterns tend to
stay motionless and change direction during escape [10–13]. The use of compu-
ter games with human ‘predators’ has yielded similar results: moving targets
with stripes running along the movement axis (elongated) are missed more
often by observers than targets with other types of patterns [14], and targets
moving over longer segments are more difficult to catch [15]. However, these
studies do not consider aposematic species, which have different dynamics
than cryptic species [2]. At the species level, birds can discriminate palatable
&
2014 The Author(s) Published by the Royal Society. All rights reserved.
from unpalatable butterflies on the basis of their flight [16].
Palatable butterflies fly fast or erratically and have inconspic-
uous coloration, whereas unpalatable species fly slowly or
regularly and are conspicuously, warningly coloured [16].
That example, however, does not address variable apose-
matic coloration within a species. Here, we explore whether
an association between behaviour and colour pattern would
benefit aposematic species with variable aposematic signals.
We tested this hypothesis in a natural population of the
aposematic frog Dendrobates tinctorius, which has extensive
colour-pattern variation [6].
2. Material and methods
(a) Study site and species
Dendrobates tinctorius (Dendrobatidae) is a diurnal frog occurring
in primary forests in the Eastern Guiana Shield [17]. It exhibits
striking variation in colour patterns (figure 1) [6,17], has skin
alkaloid defences, and field experiments with Plasticine models
suggest birds as major predators [18]. This study was carried
out at Camp Parare
´
, Les Nouragues Reserve, French Guiana
(3859
0
N, 52835
0
W, 120 m.a.s.l.), February–July 2011.
(b) Trajectories
Twenty-five females and 14 males were followed for two con-
tinuous hours each. This duration was chosen based on prior
observations of individuals remaining motionless on the same
spot for almost 1 h. Preliminary observations allowed us to
determine the distance at which the frog would not show fleeing
behaviour: about 2.5 m (less when the observer moved very
slowly). This allowed us to observe the frogs without affecting
their behaviour and record their pattern of movement under
normal circumstances (i.e. not in response to a potential preda-
tor). We stuck a flag into the ground wherever the frog
remained still for at least 5 s. After the 2-h observation period,
we measured the distance and angle between pairs of flags
(segments) and estimated the total linear distance travelled
(between starting and endpoints), the average linear speed
(total linear distance/2 h) and the path length of each individual
by adding all segment lengths.
We calculated the mean angle and distance of displacement
for each segment and ran Rayleigh tests [19] on the segment vec-
tors for each frog in order to know whether their movement was
random or directional. With this information, we created a new
categorical variable, ‘directionality’. A frog was classified as
‘directional’ if the Rayleigh test was significant at the 5% level,
indicating a significant directional vector, and ‘random’ if not.
(c) Colour patterns
Each frog was photographed against graph paper for scale. Its
colour pattern was analysed with a method that uses grid trans-
ects across colour patterns and allows the estimation of geometric
parameters based on the distribution of the number of transi-
tions between adjacent colours (here yellow and black) [20].
The parameters used were pattern simplicity (mean distance
between colour transitions, longitudinally and transversally),
and pattern elongation (ratio of transition densities along and per-
pendicular to the body axis [20]). Pattern analyses were done
(a)
(c) pattern elongation
high
low
pattern simplicity no. interruptions
(b)
Figure 1. Colour patterns in D. tinctorius. A typical elongated pattern (a) and an interrupted one (b); and (c) examples of high and low values of the colour-pattern
parameters measured.
rsbl.royalsocietypublishing.org Biol. Lett. 10: 20140193
2
with MATLAB. We also recorded the number of interruptions of
yellow on each frog’s back (figure 1c).
(d) Statistics
Both coloration and movement variables were tested for normality
(Shapiro–Wilks tests with p . 0.05) and log-transformed when
this assumption was not met. Colour-pattern parameters and
movement variables (path length, mean segment length and
linear speed of individual trajectories) were compared between
the two directionality groups with MANOVA, or Mann–Whitney
tests when the variables violated normality after transformation.
Statistical analyses were performed with SPSS v. 19.0 for Mac,
and all tests were two-tailed.
3. Results
Thirty-six per cent of the frogs were classed as ‘directional’
and 64% ‘random’. We found no differences between the
number of females and males in each directionality group
(
x
2
¼ 0.986, d.f. ¼ 1, p . 0.05).
‘Directional’ frogs have trajectorieswithlongersegments and
higher average linear speeds than ‘random’ frogs (figure 2a,b;
MANOVA on directionality: Pillai’s trace ¼ 0.727, F
4,33
¼ 17.56,
p , 0.001; log path length F
1,38
¼ 4.80, p ¼ 0.035; log mean seg-
ment length F
1,38
¼ 25.33, p , 0.001; linear speed F
1,38
¼ 78.37,
p , 0.001). Note that average speeds underestimate the speed
over each segment; if we just use the time moving, the speeds
are 1.14 + 0.24 cm s
21
(for directional frogs) and 0.31+
0.07 cm s
21
. This still underestimates the true speed as it does
not account for the very brief pauses between the fast jerks.
Colour-pattern geometry was significantly different
between the two movement groups. ‘Random’ frogs have
more interruptions in their yellow patches (figures 1b and 2c;
Mann–Whitney U ¼ 88.0, p ¼ 0.006, n ¼ 39), whereas ‘direc-
tional’ frogs have more elongated patterns [20] (figures 1a
and 2d; F
1,39
¼ 4.88, p ¼ 0.034). Pattern elongation is signifi-
cantly negatively correlated with the number of interruptions
of yellow (Spearman’s
r
¼ 20.489, p ¼ 0.002, n ¼ 39), but
not correlated with pattern simplicity (Spearman’s
r
¼ 0.053,
p ¼ 0.750, n ¼ 39). Simplicity did not differ between the two
directionality groups (F
1,39
¼ 1.27, p ¼ 0.27).
4. Discussion
Polymorphic aposematic species could use polymorphic be-
havioural strategies to mitigate the disadvantages of variable
0.3
(a)(c)
(b)(d)
7
6
5
4
3
2
1
0
0.45
0.40
0.35
0.30
0.25
0.20
0.2
0.1
–0.1
log
10
(mean segment length)
log
10
(pattern elongation) no. interruptions in yellow
speed (m h
–1
)
–0.2
–0.3
–0.4
–0.5
14
12
10
8
6
4
2
0
random directional random directional
0
Figure 2. Differences in movement and colour-pattern geometry between frogs moving directionally and randomly. (a) Mean segment length; (b) linear speed;
(c) number of interruptions and (d) pattern elongation. Triangles enclose segments (‘notches’) which, if they do not overlap between directional and random plots,
indicate that the medians are different at the 5% level or better [21].
rsbl.royalsocietypublishing.org Biol. Lett. 10: 20140193
3
colour patterns, which could be particularly strong when
exposed to naive predators. Within a single population, frogs
with more elongated patterns move continuously in a given
direction rather than randomly. This pattern–movement com-
bination might create the illusion of a static pattern or a pattern
with a greatly reduced speed that affects predators’ abilities to
track the trajectory of moving individuals and predict their
attack angle [11,14]. This may be more pronounced when
movements occur at a higher speed [22,23] and over longer seg-
ments [15], as in these frogs. Frogs moving randomly, with
unpredictable changes of direction, have more interrupted pat-
terns and move at a lower average speed, over shorter
segments than directional frogs. Interrupted patterns may be
visually disruptive [11] or cryptic at a distance [24], and the
combination of disruptive patterns and slower movements,
or alternating movement and freezing, might be advantageous
for the avoidance of motion-oriented predators [12,13,25].
A possible explanation for the match between movement
behaviour and colour pattern could be correlational selection,
which favours specific combinations of traits expressed simul-
taneously in a given individual without necessarily altering the
distribution of each trait on its own [26,27]. The presence of
the ‘wrong’ combinations of colour pattern and behaviour of
every generation favours the evolution of genetic correlations
between their gene loci, because genetic correlations reduce
the frequency of the lower fitness combinations [26,28]. For
the association to strengthen over evolutionary time, one or
more factors causing genetic correlations would have to
occur so that the phenotypic correlations would not have to
be reformed from scratch every generation. Possible mechan-
isms include gametic phase or linkage disequilibrium,
pleiotropy among loci, physical linkage or epistasis from a
gene affecting both traits. Note that some of these genetic fac-
tors could occur in the absence of correlational selection.
Plasticity can also favour phenotypic correlations. If predator
misses were frequent enough, as is possible for aposematic
species, inefficient predators would favour developmental
plasticity or active learning to move in particular patterns. In
addition, positive assortative mating for colour patterns
would incidentally favour or reinforce the association.
Both directional movements paired with striped patterns
and random movements paired with broken patterns
may be equally good alternative strategies, allowing
variation in both. Moreover, correlational selection often
results from frequency-dependent interactions such as those
between predators and prey, or parasites and hosts [28];
frequency-dependence can easily result in stable polymorph-
isms [2] and stable quantitative variation [29]. Provided that
phenotypic correlations are consistent and based upon some
genetic correlation, this can favour variation in both traits.
How distinct and how variable each alternative phenotypic
cluster can be depends on the local combination of correla-
tional selection and genetics. The system may evolve to two
distinct phenotypic combinations if the selection gradients
are steep and genetic correlations very high, but if they
are low then there may be more variation and less distinct
alternatives, as we find in D. tinctorius.
One characteristic of a population at a polymorphic equi-
librium is equality of fitness among the forms [2]. However,
the mechanisms of fitness equality differ. The elongated pat-
tern with directional movement might make it difficult for a
predator to track the frog, hence increasing the probability
of a missed attack by attacking in the wrong place. The
broken pattern might be relatively more cryptic from a dis-
tance and forms a disruptive pattern up close, reducing the
number of attacks rather than redirecting them. The
elongated pattern shows off the aposematic pattern even
when it is moving, but the broken pattern may show off
the aposematic pattern better when it is not moving and
the predator is close enough to resolve the separate yellow
spots. Further experimental research is needed to support
either possibility.
The evolutionary factors contributing to the maintenance
of polymorphisms in aposematic species are more intricate
than originally thought; a match between behaviour and
colour patterns may be as important as predation and
sexual selection in preserving different aposematic signals
within populations. This study highlights the importance of
considering behaviour–pattern interactions in future
attempts to understand the paradox of intra-populational
warning signal diversity, and signal evolution in general.
The study complied with local environmental legislation.
Acknowledgements. We are grateful to Les Nouragues staff for logistic
support; and to Johanna Mappes, Janne Valkonen and two anon-
ymous referees for helpful comments on the manuscript.
Data accessibility. Data are available in the electronic supplementary
material.
Funding statement. We thank the Centre of Integrative Ecology (Deakin
University) and CNRS for financial support.
References
1. Poulton EB. 1890 The colours of animals: their
meaning and use. London, UK: Kegan Paul.
2. Endler JA, Mappes J. 2004 Predator mixes and the
conspicuousness of aposematic signals. Am. Nat.
163,532–547.(doi:10.1086/382662)
3. Ruxton GD, Sherratt TN, Speed MP. 2004 Avoiding
attack: the evolutionary ecology of crypsis, warning
signals and mimicry. Oxford, UK: Oxford University
Press.
4. Speed MP. 2000 Warning signals, receiver
psychology and predator memory. Anim. Behav. 60,
269– 278. (doi:10.1006/anbe.2000.1430)
5. Nokelainen O, Hegna RH, Reudler JH, Lindstedt C,
Mappes J. 2012 Trade-off between warning signal
efficacy and mating success in the wood tiger moth.
Proc. R. Soc. B 279,257–265.(doi:10.1098/rspb.
2011.0880)
6. Rojas B, Endler JA. 2013 Sexual dimorphism
and intra-populational colour pattern variation
in the aposematic frog Dendrobates tinctorius.
Evol. Ecol. 27,739–753.(doi:10.1007/s10682-013-
9640-4)
7. Maan ME, Cummings ME. 2009 Sexual dimorphism
and directional sexual selection on aposematic
signals in a poison frog. Proc. Natl Acad. Sci. 106,
19 072 – 19 077. (doi:10.1073/pnas.0903327106)
8. Rudh A, Breed MF, Qvarnstrom A. 2013 Does
aggression and explorative behaviour decrease
with lost warning coloration? Biol. J. Linn. Soc. 108,
116– 126. (doi:10.1111/j.1095-8312.2012.02006.x)
9. Willink B, Brenes-Mora E, Bolan
˜
os F, Pro
¨
hl H. 2013
Not everything is black and white: color and
behavioral variation reveal a continuum between
cryptic and aposematic strategies in a polymorphic
poison frog. Evolution 67,2783–2794.(doi:10.
1111/evo.12153)
rsbl.royalsocietypublishing.org Biol. Lett. 10: 20140193
4
10. Brodie ED. 1989 Genetic correlations between
morphology and antipreda tor beha v iour in na tur a l
popula tions of the garter snake Thamnophis ordinoides.
Natur e 342,542–543.(doi:10.1038/342542a0)
11. Jackson JF, Ingram W, Campbell HW. 1976 Dorsal
pigmentation pattern of snakes as an anti-predator
strategy: multivariate approach. Am. Nat. 110,
1029–1053. (doi:10.1086/283125)
12. Creer DA. 2005 Correlations between ontogenetic
change in color pattern and antipredator behavior
in the racer, Coluber constrictor. Ethology 111,
287– 300. (doi:10.1111/j.1439-0310.2004.01062.x)
13. Allen WL, Baddeley R, Scott-Samuel NE, Cuthill IC.
2013 The evolution and function of pattern diversity
in snakes. Behav. Ecol. 24,1237–1250.(doi:10.
1093/beheco/art058)
14. Stevens M, Searle WTL, Seymour JE, Marshall KLA,
Ruxton GD. 2011 Motion dazzle and camouflage as
distinct anti-predator defenses. BMC Biol. 9, 81.
(doi:10.1186/1741-7007-9-81)
15. Sherratt TN, Rashed A, Beatty CD. 2004 The
evolution of locomotory behavior in profitable and
unprofitable simulated prey. Oecologia 138, 143 –
150. (doi:10.1007/s00442-003-1411-4)
16. Chai P. 1986 Field observations and feeding
experiments on the responses of rufous-tailed
jacamars (Galbula ruficauda) to free-flying
butterflies in a tropical rainforest. Biol. J. Linn. Soc.
29,161–189.(doi:10.1111/j.1095-8312.1986.
tb01772.x)
17. Noonan BP, Gaucher P. 2006 Refugial isolation and
secondary contact in the dyeing poison frog
Dendrobates tinctorius. Mol. Ecol. 15, 4425– 4435.
(doi:10.1111/j.1365-294X.2006.03074.x)
18. Noonan BP, Comeault AA. 2009 The role of predator
selection on polymorphic aposematic poison frogs.
Biol. Lett. 5, 51– 54. (doi:10.1098/rsbl.2008.0586)
19. Zar JH. 1996 Biostatistical analysis, 3rd edn. Upper
Saddle River, NJ: Prentice-Hall.
20. Endler JA. 2012 A framework for analysing colour
pattern geometry: adjacent colours. Biol. J. Linn.
Soc. 107,233–253.(doi:10.1111/j.1095-8312.2012.
01937.x)
21. McGill R, Tukey JW, Larsen A. 1978 Variations of
boxplots. Am. Stat. 32, 12–16.
22. Stevens M, Yule DH, Ruxton GD. 2008 Dazzle
coloration and prey movement. Proc. R. Soc. B 275,
2639–2643. (doi:10.1098/rspb.2008.0877)
23. von Helversen B, Schooler LJ, Czienskowski U. 2013
Are stripes beneficial? Dazzle camouflage influences
perceived speed and hit rates. PLoS ONE 8, e61173.
(doi:10.1371/journal.pone.0061173)
24. Endler JA. 1978 A predator’s view of animal colour
patterns. Evol. Biol. 11, 319 – 364.
25. Hatle JD, Salazar BA, Whitman DW. 2002 Survival
advantage of sluggish individuals in aggregations of
aposematic prey, during encounters with ambush
predators. Evol. Ecol. 16,415–431.(doi:10.1023/
a:1020814110102)
26. Endler JA. 1986 Natural selection in the wild.
Princeton, NJ: Princeton University Press.
27. Brodie ED. 1992 Correlational selection for color
pattern and antipredator behavior in the garter
snake Thamnophis ordinoides. Evolution 46,
1284– 1298. (doi:10.2307/2409937)
28. Sinervo B, Svensson E. 2002 Correlational selection
and the evolution of genomic architecture. Heredity
89,329–338.(doi:10.1038/sj.hdy.6800148)
29. Mani GS, Clarke BC, Shelton PR. 1990 A model of
quantitative traits under frequency-dependent
balancing selection. Proc. R. Soc. Lond. B 240,
15– 28. (doi:10.1098/rspb.1990.0024)
rsbl.royalsocietypublishing.org Biol. Lett. 10: 20140193
5
