ArticlePDF Available

Abstract and Figures

Local warning colour polymorphism, frequently observed in aposematic organisms, is evolutionarily puzzling. This is because variation in aposematic signals is expected to be selected against due to predators' difficulties associating several signals with a given unprofitable prey. One possible explanation for the existence of such variation is predator generalization, which occurs when predators learn to avoid one form and consequently avoid other sufficiently similar forms, relaxing selection for monomorphic signals. We tested this hypothesis by exposing the three different colour morphs of the aposematic wood tiger moth, Arctia plantaginis, existing in Finland to local wild-caught predators (blue tits, Cyanistes caeruleus). We designed artificial moths that varied only in their hindwing coloration (white, yellow and red) keeping other traits (e.g. wing pattern and size) constant. Thus, if the birds transferred their aversion of one morph to the other two we could infer that their visual appearances are sufficiently similar for predator generalization to take place. We found that, surprisingly, birds showed no preference or aversion for any of the three morphs presented. During the avoidance learning trials, birds learned to avoid the red morph considerably faster than the white or yellow morphs, confirming previous findings on the efficacy of red as a warning signal that facilitates predator learning. Birds did not generalize their learned avoidance of one colour morph to the other two morphs, suggesting that they pay more attention to conspicuous wing coloration than other traits. Our results are in accordance with previous findings that coloration plays a key role during avoidance learning and generalization, which has important implications for the evolution of mimicry. We conclude that, in the case of wood tiger moths, predator generalization is unlikely to explain the unexpected coexistence of different morphs.
Content may be subject to copyright.
Colour alone matters: no predator generalization among morphs of an
aposematic moth
Katja R
onk
a
*
,
1
, Chiara De Pasqual
1
, Johanna Mappes, Swanne Gordon, Bibiana Rojas
Department of Biological and Environmental Science, University of Jyv
askyl
a, Centre of Excellence in Biological Interactions, Finland
article info
Article history:
Received 12 June 2017
Initial acceptance 24 July 2017
Final acceptance 19 October 2017
MS. number: 17-00475R
Keywords:
learning
polymorphism
predatoreprey interactions
predator generalization
warning signals
wood tiger moth
Local warning colour polymorphism, frequently observed in aposematic organisms, is evolutionarily
puzzling. This is because variation in aposematic signals is expected to be selected against due to
predators' difculties associating several signals with a given unprotable prey. One possible explanation
for the existence of such variation is predator generalization, which occurs when predators learn to avoid
one form and consequently avoid other sufciently similar forms, relaxing selection for monomorphic
signals. We tested this hypothesis by exposing the three different colour morphs of the aposematic wood
tiger moth, Arctia plantaginis, existing in Finland to local wild-caught predators (blue tits, Cyanistes
caeruleus). We designed articial moths that varied only in their hindwing coloration (white, yellow and
red) keeping other traits (e.g. wing pattern and size) constant. Thus, if the birds transferred their aversion
of one morph to the other two we could infer that their visual appearances are sufciently similar for
predator generalization to take place. We found that, surprisingly, birds showed no preference or
aversion for any of the three morphs presented. During the avoidance learning trials, birds learned to
avoid the red morph considerably faster than the white or yellow morphs, conrming previous ndings
on the efcacy of red as a warning signal that facilitates predator learning. Birds did not generalize their
learned avoidance of one colour morph to the other two morphs, suggesting that they pay more
attention to conspicuous wing coloration than other traits. Our results are in accordance with previous
ndings that coloration plays a key role during avoidance learning and generalization, which has
important implications for the evolution of mimicry. We conclude that, in the case of wood tiger moths,
predator generalization is unlikely to explain the unexpected coexistence of different morphs.
©2017 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
Aposematic organisms display warning signals that predators
learn to associate with their unprotability (Poulton, 1890). The
survival of such prey is thus highly dependent on a predator's
ability to learn, remember and generalize their learned avoidance
to other individuals sharing the same warning signal (reviewed in
Ruxton, Sherratt, &Speed, 2004). Signal sharing among aposematic
prey benets both the prey and their potential predators: (1) a
given individual has a lower risk of predation when more in-
dividuals share the same warning signal, and (2) predators benet
from not having to sample as many unprotable or toxic prey and
can more easily remember one and not multiple signals (ten Cate &
Rowe, 2007; Ghirlanda &Enquist, 2003; Guilford &Dawkins, 1991;
Müller, 1878; Rowland, Ihalainen, Lindstr
om, Mappes, &Speed,
2007). Therefore, local polymorphism in warning coloration is ex-
pected to be selected against (Chouteau, Arias, &Joron, 2016;
Endler, 1991; Joron &Mallet, 1998; Lindstr
om, Alatalo, Lyytinen,
&Mappes, 2001; Mallet &Barton, 1989; Mallet &Joron, 1999; but
see also Ihalainen, Lindstr
om, &Mappes, 2007 who found no evi-
dence for slower avoidance learning of single versus multiple
signals).
Despite the predicted disadvantages, warning signal poly-
morphisms are present in several aposematic taxa, such as frogs
(Am
ezquita, Castro, Arias, Gonz
alez, &Esquivel, 2013; Rojas &
Endler, 2013), ladybirds (O'Donald &Majerus, 1984; Pr
uchov
a,
Nedv
ed, Veselý, Ernestov
a, &Fuchs, 2014) and butteries (Jiggins
&McMillan, 1997). In fact, they seem to be more common than
expected considering that warning signals are predicted to be un-
der positive frequency-dependent selection (Müller, 1878; Ruxton
et al., 2004). One possible explanation for the co-occurrence of
several warning signal forms within the same population is pred-
ator generalization. This refers to a predator's ability to transfer its
learned avoidance of one signal to other signal(s) that share
*Correspondence: K. R
onk
a, Department of Biological and Environmental
Science, University of Jyv
askyl
a, Centre of Excellence in Biological Interactions, P.O.
Box 35, FI-40014, University of Jyv
askyl
a, Finland.
E-mail address: katja.h.ronka@jyu.(K. R
onk
a).
1
Shared rst authorship.
Contents lists available at ScienceDirect
Animal Behaviour
journal homepage: www.elsevier.com/locate/anbehav
https://doi.org/10.1016/j.anbehav.2017.11.015
0003-3472/©2017 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
Animal Behaviour 135 (2018) 153e163
common characteristics (Gamberale-Stille &Tullberg, 1999;
Guilford &Dawkins, 1991; Lindstr
om, Alatalo, Mappes, Riipi, &
Vertainen, 1999; Mappes &Alatalo, 1997). Generalization can be
symmetric, meaning that once one colour is learned it is equally
possible to transfer the learned aversion to other similar colours, or
asymmetric, implying that transferring a learned avoidance from
one colour to other(s) depends on the signal salience (Aronsson &
Gamberale-Stille, 2008; Exnerov
a et al., 2006; Gamberale &Tull-
berg, 1996; Gamberale-Stille &Tullberg, 1999; Ham, Ihalainen,
Lindstrom, &Mappes, 2006; Ruxton, Franks, Balogh, &Leimar,
2008; Waldron et al., 2017).
Predator learning involves different cognitive processes that
establish the association between warning coloration and unprof-
itability, and aid the memorization of this association once estab-
lished. This learning process may vary between predators even at
intraspecic levels (e.g. Adamov
a-Je
zov
a, Hospodkov
a, Fuchsov
a,
Stys, &Exnerov
a, 2016; Endler &Mappes, 2004; Exnerov
a et al.,
2015; Exnerov
a, Sv
adov
a, Fu
cíkov
a, Drent, &
Stys, 2010; Karlíkov
a,
Veselý, Ber
ankov
a, &Fuchs, 2016; Lindstr
om, Alatalo, &Mappes,
1999; Sherratt &Macdougall, 1995; Skelhorn, Halpin, &Rowe,
2016). Predators may also vary in their ability to cope with defen-
ded prey, due for example to dietary conservatism (Marples &Kelly,
1999; Mettke-Hofmann, Winkler, &Leisler, 2002; Turini, Veselý, &
Fuchs, 2016; Webster &Lefebvre, 2000). Therefore, investigating
how predators learn to associate the appearance of prey with the
noxious effects of their unprotability is crucial to understanding
how signal variation can be maintained within a population. During
the learning process predators acquire information about the
nutrient and toxin content of aposematic prey. Thus, individual
predators are expected to make different decisions on how to use
the information gathered from an encounter with aposematic prey
(Exnerov
a et al., 2003, 2007; Halpin, Skelhorn, &Rowe, 2014; Lynn,
2005; Skelhorn et al., 2016; Trimmer et al., 2011), and modify their
ingestion of toxic prey according to their toxic burden (Skelhorn &
Rowe, 2007).
Generalized avoidance should be broad and persist for a rela-
tively long time to offer protection to different warningly coloured
prey morphs. On the other hand, naïve predators can also avoid
warningly coloured prey due to innate wariness, neophobia or di-
etary conservatism (Exnerov
a et al., 2007; Lindstr
om et al., 1999;
Marples &Kelly, 1999; Marples &Mappes, 2011), which could be
further reinforced by the short-term effects of negative experience
with other aposematic prey. It has been suggested that multiple
modalities of warning signals can help predators discriminate be-
tween palatable and unpalatable prey (Kazemi, Gamberale-Stille, &
Leimar, 2015; Siddall &Marples, 2008). However, generalized
avoidance of aposematic prey can also be based on cues of different
sensory modalities, such as odour, sound, colour or pattern or
combinations of these. Depending on the cognitive processes of
predators, they could also associate their negative experience with
certain stimuli to any other stimuli encountered simultaneously
(Mackintosh, 1975; Pavlov, 1927). These results emphasize the
importance of studying how multiple cues and separate signal
components inuence a predator's decision to attack prey (Kikuchi,
Mappes, Sherratt, &Valkonen, 2016; Rowe &Halpin, 2013).
Here, we tested the hypothesis that the hindwing colour poly-
morphism of an aposematic moth is enabled by predator general-
ization, and investigated whether or not that generalization is
symmetric. We exposed paper models of the different hindwing
colour morphs of the wood tiger moth, Arctia plantaginis (formerly
known as Parasemia plantaginis,R
onk
a, Mappes, Kaila, &Wahlberg,
2016) to natural predators (blue tits, Cyanistes caeruleus), and
examined whether, once they learned to avoid one of the colour
morphs, they would generalize this aversion to the two unlearned
colour morphs, which would allow multiple morphs to coexist. A
lack of generalization among colour morphs would mean that
birds pay more attention to colours than to other cues of the moth
wings.
METHODS
The wood tiger moth is an aposematic diurnal moth with a
Holarctic distribution (Hegna, Galarza, &Mappes, 2015). They have
two different chemical defences, one of which is secreted from the
prothoracic glands. Although the chemical composition is not fully
known, these uids contain two types of methoxypyrazines, which
are produced de novo (Burdeld-Steel, Pakkanen, Rojas, Galarza, &
Mappes, 2016) and make them a deterrent to birds. Experiments
with bird predators suggest that the uids of yellow males have a
more repulsive odour (Rojas et al., 2017), while those of white
males taste worse (Rojas, Burdeld-Steel &Mappes, 2015). In-
dividuals vary in the degree of melanization and black patterning of
the wings, as well as in levels of chemical defence, but the most
striking feature of the wood tiger moth is its local hindwing colour
polymorphism (Hegna et al., 2015). In Europe, its forewings present
a black and white pattern in both males and females, whereas the
hindwing colour combined with black pattern differs between the
sexes (e.g. Galarza, Nokelainen, Ashra, Hegna, &Mappes, 2014;
Hegna &Mappes, 2014). The distinct white and yellow male
morphs are genetically determined by one autosomal locus and at
least three alleles, dominant white, recessive white and interme-
diate yellow (Galarza, Nokelainen, &Mappes, 2016), while female
hindwing coloration varies continuously from yellow to red
(Lindstedt, Schroderus, Lindstr
om, Mappes, &Mappes, 2016;Fig.1).
In Finland, for example, yellow and white males may occur within
one population (Nokelainen, Valkonen, Lindstedt, &Mappes, 2014)
whereas female hindwing coloration is mostly red (Hegna et al.,
2015).
To study the reaction of bird predators (see below for details on
procedure) to the different hindwing colour morphs, we used
articial moth models. The usage of articial prey allows for the
controlled manipulation of one or more warning signal compo-
nents at a time, while accounting for how predators (i.e. birds)
would see them (Endler &Mielke, 2005). In this way, other com-
ponents can be kept constant and independent of prey qualities,
such as the variation in the level of chemical defence or behaviour
(Karlíkov
a et al., 2016; Lindstr
om, Alatolo et al., 1999; Veselý &
Fuchs, 2009). Here, our articial moth models eliminated indi-
vidual variation in moth size, shape, degree of melanization, wing
pattern, wing posture, behaviour, smell or taste. Model wings were
constructed with the software GIMP (2.8.16; http://www.gimp.org/
) from pictures of a real male wood tiger moth specimen collected
in Finland. Pictures of one forewing and one hindwing of a typical
white moth were duplicated to obtain a symmetric pattern for the
whole model. The melanization pattern of the moths used was a
representative sample of a wing pattern in Finland (Fig. 1). To
control for the amount and shape of melanized (mainly black)
pattern of the wings, yellow and red models were created from the
same wing picture, changing the hue of the white parts of the
hindwing towards yellow or red. Finished models were printed
double sided (HP Color LaserJet CP2025) on waterproof (Rite In The
Rain, Tacoma, WA, U.S.A.) paper. To ensure that the model colours
resembled the real wood tiger moth morphs, colour reectance
was measured with an Ocean Optics Maya2000 Pro spectrometer
and average reectance curves from three spots in the model
hindwing coloration were compared to average reectance curves
of white, yellow and red moth hindwings (Fig. 1). Models were
then cut out from the paper and completed with a body made of
rolled pastry, composed of two parts of lard, six parts of coarse
wheat our and one part of water to make them edible. The total
K. R
onk
a et al. / Animal Behaviour 135 (2018) 153e163154
body weight was 0.04 ±0.005 g. Bodies were dyed on top with
black food colouring, to make models resemble the real moths as
accurately as possible. Finally, bodies were glued on the paper
models with nontoxic glue (UHU stick).
Bird Predators
Blue tits were chosen as predators for several reasons: (1) they
are visual foragers and their visual capabilities are well known
(Hart, Partridge, Cuthill, &Bennett, 2000; Hart &Vorobyev, 2005),
ensuring that they are able to distinguish all of the wood tiger
moth's colour morphs; (2) they have been used in several experi-
ments on coloration (e.g. Dimitrova &Merilaita, 2010; Exnerov
a
et al., 2007; Kikuchi et al., 2016) and wood tiger moths
(Nokelainen, Hegna, Reudler, Lindstedt, &Mappes, 2012), and also
with similar moth models (Rojas, Burdeld-Steel et al., 2015); (3)
tits are likely to be important natural predators of wood tiger moths
in Finland (Nokelainen et al., 2014); and (4) blue tits are common in
central Finland, and easy to capture and keep in captivity for a short
period of time.
The birds used for the experiment were caught from Konnevesi
Research Station and City of Jyv
askyl
a (central Finland), maintained
individually in plywood cages with a perch, water bowl and food ad
libitum, and kept on a 12:12 h light:dark cycle. Each bird was
weighed before and after the experiment, ringed, and its sex and
age were determined before being released to the same place of
capture. Birds were used with permission from the Central Finland
Centre for Economic Development, Transport and Environment and
licensed from the National Animal Experiment Board (ESAVI/9114/
04.10.07/2014) and the Central Finland Regional Environment
Centre (VARELY/294/2015). All experimental birds were used ac-
cording to the ASAB/ABS Guidelines for the treatment of animals in
behavioural research and teaching.
Experimental Procedures
The experiment consisted of three phases: a preference test, a
learning test and a generalization test (see details below). Each bird
was tested individually and only once for each part of the experi-
ment. The experiment was conducted between November 2015 and
March 2016 at Konnevesi Research Station, in central Finland and
lasted, on average, 3 days for each individual, depending on how
long the bird took to complete the different tests.
Trials took place in experimental custom-built plywood cages
(50 50 cm and 70 cm high) illuminated with a daylight lamp (Exo
Terra Repti Glo 10.0 UVB, http://exo-terra.com/). Each aviary had a
perch and a water bowl (access ad libitum). Birds were observed
through a small mesh-covered window situated on the front of the
cage, and lmed with a Canon Powershot S120 camera. The
experiment took place in a dark room to minimize observers dis-
turbing the birds.
Food and experimental models were offered on a green platform
through a moveable tray behind a visual barrier, allowing us to
estimate the exact time when the bird rst saw the model and thus
started the trial (see details in Nokelainen et al., 2012). A standard
green background was used, because wood tiger moths rest on
green leaves in nature (Hegna, Nokelainen, Hegna, &Mappes, 2013;
Nokelainen et al., 2012). All colours used in the moth models are
easily distinguished from the background by birds: Hegna et al.
(2013) reported just noticeable difference (JND) values in colour
contrast ranging from 8.6 to 11.6 for white and yellow articial
moth models and real wood tiger moths against the green back-
ground used also in this experiment, and Lindstedt et al. (2011)
calculated JND values above 27.27 for orange and red females on
natural green leaves of Alnus incana.
During pretraining, birds were allowed to habituate to the
experimental cages and learned to eat three sunower seeds from
100
(a) (b) (c)
80
60
40
20
0
100
80
60
40
20
0
100
80
60
40
20
300 400 500 600 700 800
Wavelen
g
th (nm)
Reflectance (%)
0
Figure 1. (a) Typical wood tiger moth colour morphs from central Finland (pictures: Kari Kulmala Coll.), (b) articial white, yellow and red moth wings used in the experiment and
(c) reectance curves of the white, yellow and red hindwings of real moths (darker colours) compared to reectance curves from white, yellow and red model hindwings (lighter
colours). Spectral measurements were taken from three wild-caught individuals of each colour from the spots marked with blue circles on the white moth. The same spots were
used to measure the model hindwing colours. Model wings were set in a more natural posture, less spread than the spread collection samples, but unfolded to show the hindwing
colour.
K. R
onk
a et al. / Animal Behaviour 135 (2018) 153e163 155
the green platform. To motivate the birds to attack the moth models
during the experiment (see below), they were food deprived for 2 h
before the preference test, 1 h before the learning test and 1 h
before the generalization test. After food deprivation, bird moti-
vation was tested with a sunower seed; if eaten, the bird was
considered ready to begin the test.
Phase 1: Preference test
A preference test was included in the experimental protocol for
two reasons. As we used wild-caught birds, we rst tested whether
they had any pre-existing biases towards white, yellow or red moth
morphs. Second, by offering palatable morphs several times to
birds we ensured that any potential unlearned or learned biases
disappeared, allowing us to test the effect of the coloration on
learning and generalization (Ghirlanda &Enquist, 2003).
All three morphs (white, yellow and red) were offered simulta-
neously on the green platform for 5 min, starting from when the
bird rst saw them.If the bird did not attack (i.e. grab or peck) any of
the edible model pastry bodies during the 5 min, the models were
taken away and presented again after a break. Once the rst attack
was made, the models were kept in the cage until the bird nished
eating all the pastry bodies. To ensure that all birds had an equally
rewarding experience with all the colours, we let the birds nish
eating the pastry bodies of all models in three consecutive trials
during the preference test. Between the trials, the presentation
(order) of the models on the platform was always changed (Fig. 2).
As birds were hesitant to attack the moth models for the rst
time (hesitation times varying from 17 s to 2 h consisting of 5 min
presentations), we did not use time to attack in analyses. Instead,
we recorded the order in which the models were attacked and
eaten during the three consecutive trials. We compared the order of
attacks between the rst and the last preference test to be sure that
all the birds got rid of any potential bias in preferences before the
learning phase. Preference test presentations were continued for a
maximum of 2 days. Eight of 53 birds did not attack or nish eating
the articial moth models offered during the preference test and
were, therefore, excluded from further tests.
Phase 2: Learning test
In the second phase of the experiment we tested whether blue
tits learn to avoid white, yellow and red models differently, and
established learned avoidance towards one of the colour morphs
before the following generalization test. Birds that completed the
preference test were divided into three groups for avoidance
learning: 15 birds wereoffered white models,15 yellow models and
16 red models as unpalatable. Groups of birds were selected as
similar as possible (i.e. similar sex, age and size distribution) and
birds from all groups were tested simultaneously. All models were
made unpalatable by replacing the water in pastry bodies with 15%
chloroquine diphosphate solution (Sigma Life Science, St Louis, MO,
U.S.A.). As the pastry bodies were coloured with black dye on top,
we also added 15% chloroquine diphosphate solution on top of the
bodies and let it dry before the following trials. Chloroquine solu-
tion was used because it is odourless (Hong, 1976) and thus all
qualities other than palatability (i.e. taste) of the prey items
remained the same throughout the experiment.
During the learning test, unpalatable models were presented
individually in consecutive trials alternating with sunower seeds
(Fig. 2). Sunower seeds were offered to monitor the birds' moti-
vation to forage and avoid unnecessary starvation. If the bird did
not attack the sunower seed, it got a 10 min break without food
and was then offered a sunower seed again. If the bird attacked
the sunower seed, the next unpalatable model was offered 2 min
after the bird nished eating. As long as the bird attacked the
models, trials were continued alternating with sunower seeds. If
the bird did not attack the unpalatable model, but ate the sunower
seed, it was considered to reject the model. After a bird did not
attack the moth model the second time in a row, a small live
mealworm (<20 mm Tenebrio molitor larva) was offered instead of
the sunower seed to test the bird's motivation to attack insect prey
and increase its motivation to forage. If the bird now attacked the
unpalatable model offered after the mealworm, trials were
continued again alternating with sunower seeds, but if it rejected
the unpalatable model, it got another mealworm (Fig. 2). We
considered the bird to have learned to avoid the unpalatable
models when it refused to attack three models in a row, but
consumed the sunower seeds and mealworms offered in between
and after the rejected models.
Presentation time was set to 5 min from when the bird rst saw
the model for the rst three trials to make sure that each bird had
the opportunity to attack and taste the model. To keep the overall
duration of the generalization experiment within the permitted 4-
day limit, a maximum of 30 presentations divided into 2 days was
set for the avoidance learning. Furthermore, we reduced the pre-
sentation time to 2 min from the fourth to the sixth trial, and to
1 min for the rest of the trials. Based on our observations during a
pilot experiment with six birds and xed durations of trials, birds
were unlikely to attack the model and did not consume it within
5 min if they did not attack within the rst minute. Sunower seeds
were usually attacked quickly, and hesitation time declined to a few
seconds as the trials proceeded, implying that 1 min was sufcient
to test the bird's willingness to attack the models. Two of the 46
birds did not stop attacking (white and red) models within 30
presentations and were therefore excluded from the following
generalization test.
Phase 3: Generalization test
In the third phase, we tested whether the 44 birds that had
learned one of the colour morphs as unpalatable would avoid
attacking the other two colour morphs. When birds had completed
the learning test, half of them had a break of at least 2 h with food
and 1 h of food deprivation before the last phase of the experiment,
1: Preference test 2: Learning test 3: Generalization test
1231 2 3 1 ... NN+1 N+2 N+3
Figure 2. Schematic illustration of the experimental design. Each green circle represents a platform presented to a bird in one trial. Moth models as presented to a bird that learned
to avoid the red morph are illustrated on the platforms and alternative food offered between the trials is shown above the platforms. For details of the experimental protocol see
Methods (Experimental procedures; phases 1e3).
K. R
onk
a et al. / Animal Behaviour 135 (2018) 153e163156
and half were tested the following day. The generalization test
started after the bird had consumed a sunower seed offered to test
its motivation to attack. Birds were tested for the generalization
with the colours that they did not learn as unpalatable: yellow and
red for those that learned white as unpalatable, white and red for
those that learned yellow, and white and yellow for those that
learned red (Fig. 2). The two colours tested were presented
simultaneously on the green platform in alternating positions for
three trials lasting 5 min each. This allowed us to test the repeat-
ability of bird behaviour. The trials were interspersed with sun-
ower seed presentations to make sure that birds were not
attacking the models due to lack of motivation. Moreover, offering
alternative food ensured that birds were not forced to eat the
models simply because of hunger. Models used for the general-
ization test were palatable.
Statistical Analysis
Phase 1: Preference test
The potential colour bias of blue tits was analysed separately for
the rst trial (Fig. 2) and all three trials pooled. Colour biases are
most likely to be detected reliably by checking the order of attacks
on the white, yellow and red models in the rst trial (N¼53), when
the birds rst encountered the models. The number of moths of
each colour taken rst, as well as left last, were compared by means
of a chi-square test. Additionally, all three trials were pooled in
another analysis to nd out whether the potential biases dis-
appeared as the birds learned to eat all the models. In the pooled
data, each colour was scored based on the order of choice by the
bird in each trial; the colour chosen rst was scored 1, that chosen
second was scored 2 and the colour chosen last was scored 3.
Thereby, the minimum score expected for a preferred colour was 3
(i.e. always chosen rst), and the maximum score expected for an
avoided colour was 9 (i.e. always chosen last). To study the popu-
lation level bias to all colours, the scores of each colour in each of
the three trials were summed and compared to an even distribution
by means of a chi-square test. The potential inuence of the rst
colour chosen on the subsequent choice was checked with a
binomial exact test.
Phase 2: Learning test
Potential differences in learning rate between the three colour
morphs were analysed using a mixed-effect Cox regression model,
using the coxmepackage (version 2.2e5; Therneau, 2015)in
RStudio (v. 0.99.902; RStudio, 2015). The response variable was the
probability that the presented model was attacked in each trial;
time was represented as number of trials. Model colour was added
as an explanatory factor and bird individual as a random effect.
Phase 3: Generalization test
If birds generalized their learned aversion of a given colour to
the two nonlearned colours, we would expect them to refrain from
attacking models offered during the generalization test but eat the
alternative prey offered between trials. Hence, the probability of
attack on palatable models is expected to be signicantly lower
than random (<0.5). If, in contrast, birds were unable to generalize
their learned avoidance, we would expect the attack probability to
be signicantly higher than 0.5. High attack probability is expected
(in the case of no generalization) since the birds had attacked and
eaten similar palatable models in the preference test and did attack
the models presented rst in the learning test within the 5 min
presentation. Thus, to test whether the birds generalized and the
attack probability on the models was lower (or higher) than
random, we built two generalized linear mixed models (GLMM 1
and 2) with a logit link and binomial distribution, including
whether the prey was attacked (1) or not (0) as the dependent
variable. Bird ID and bird ID nested within trial in GLMM 2 were
added as random factors using package lme4 (Bates, Maechler,
Bolker, &Walker, 2015) in R. GLMM 1 was used to test for gener-
alization in the rst trial only, and GLMM 2 in all three trials.
To test for asymmetric generalization, we divided the birds into
six treatment classes by the colours they learned (white, yellow and
red) and were offered (yellow and red, white and red, white and
yellow, respectively). This classication was then used as the
explanatory variable (colour combination) in two GLMM models
separately for the rst trial only (Table A1 in the Appendix) and all
three trials (Table A2 in the Appendix) of generalization (again with a
logit link and binomial distribution, including whether the prey was
attacked (1) or not (0) as the dependent variable, and bird ID nested
within trial and/or bird ID as random factors). A chi-square test was
used to check whether the birds attacked one colour morph rst
more frequently between the two colour morphs offered, both in the
rst trial and in the rst three trials pooled (Table A3 in the
Appendix). Birds tested the same or the following day after avoid-
ance learning were pooled in all analyses, as there were no differ-
ences in the number of attacks between birds tested the same or the
following day after avoidance learning in the rst trial (unpaired
two-sample Wilcoxon test: W¼276, N¼44, P¼0.21)or in the three
trials pooled (W¼262, N¼44, P¼0.59). We also checked whether
the rate of learning correlated with the number of attacks in the
generalization test with a Spearman correlation.
RESULTS
Preference Test
At the population level, birds did not show any preferences
(Table 1) or aversion (chi-square test:
c
2
2
¼2.577, P¼0.28) towards
any of the colours (white, yellow or red) during the rst trial. Birds
chose the second colour to attack with the same probability be-
tween the two colours left, irrespective of the rst colour chosen
(binomial exact test: P>0.05 for all comparisons).
At the individual level, 35 birds (85.4%, N¼41) chose at least
one colour in the same order for two different trials (for instance,
the same bird chose the yellow morph as last choice in two trials
out of three). Two birds showed a strong preference for one of the
colour morphs, choosing the same colour (yellow and red,
respectively) rst for all three trials. Three birds showed avoidance
for one colour morph (one for white, two for red), leaving the same
colour as last in all the trials. All other birds changed their order of
choice during the three trials, showing that they got rid of potential
biases towards the colours during training. When we tested the
overall scores for each colour morph during the three trials, birds
did not show differences between the colour morphs (chi-square
test:
c
2
2
¼0.789, P¼0.67).
Learning Test
Apart from two individuals, all birds (N¼44) learned to avoid
their moth model according to the criterion of no attack over three
Table 1
Distribution of colours chosen rst in the preference test trials (phase 1)
White rst Yellow rst Red rst Ndf
c
2
P
Trial 1 34 32 34 53 2 0.04 0.98
Trial 2 44 22% 34 41 2 2.98 0.23
Trial 3 29 27% 44 41 2 2.10 0.35
The percentages of birds that chose white, yellow or red models rst in the three
trials of the preference test, and the corresponding chi-square comparison for
preference for each trial, are shown.
K. R
onk
a et al. / Animal Behaviour 135 (2018) 153e163 157
subsequent trials. The number of trials needed to learn to avoid the
unpalatable model varied between 2 and 23 among the birds
(mean ¼7). The Cox regression model (Fig. 3) showed that birds
learned to avoid the red colour morph signicantly faster than the
yellow (Z¼2.17, P¼0.03), but showed no signicant differences
between the yellow and white morphs (Z¼0.87, P¼0.38).
Generalization Test
Overall, blue tits did not generalize their learned avoidance from
one colour morph to the other two, as the attack probabilities were
signicantly higher than 0.5 in the rst trial (GLMM 1: Z ¼4.33,
P<0.001; Fig. 4) and the three trials pooled (GLMM 2: Z¼6.42,
P<0.001). Only three of 44 individuals did not attack any of the
palatable models during the generalization test, showing general-
ized avoidance.
We did not nd clear evidence of asymmetric generalization.
The estimated attack probabilities did not differ signicantly be-
tween the combinations of colour learned and colour offered in
generalization trials (Tables A1 and A2 in the Appendix), and no
differences were found in which colour the birds attacked rst
during the rst trial of the generalization test (chi-square test:
P>0.05 in all cases; Table A3 in the Appendix). In the rst trial,
however, birds that learned yellow attacked fewer white models
compared to the other colour combinations, and the effect is near
the 0.05 signicance level (Table A1 in the Appendix). Also, when
the three trials were pooled, we found that birds that learned to
avoid the white morph attacked the red morph rst signicantly
more often than the yellow one (chi-square test:
c
2
1
¼5.9, P¼0.02).
The rate of learning did not correlate signicantly with the
proportion of attacked models in the rst generalization trial
(Spearman correlation r
S
¼0.12, N¼44, P¼0.45) or the total
number of attacks in the three generalization trials (Spearman
correlation: r
S
¼0.27, N¼44, P¼0.07), thus allowing us to
compare the effect of the colour learned on generalization despite
different learning rates of red versus the other colours.
DISCUSSION
No Generalization Based on Hindwing Colour
Generalized avoidance by local predators from one warning
signal to another has been proposed to contribute to the mainte-
nance of local warning signal polymorphism in aposematic species
(Am
ezquita et al., 2013; Exnerov
a et al., 2006; Gamberale &
Tullberg, 1996; Gamberale-Stille &Tullberg, 1999; Ham et al.,
2006; Hegna &Mappes, 2014; Rojas, Rautiala, &Mappes, 2014;
Ruxton et al., 2008; Waldron et al., 2017). Here we studied in more
detail how bird predators learn and generalize the warning colours
of a polymorphic (red, yellow, white) wood tiger moth population
using articial moth models. Attack rates during the generalization
test were in general very high. Indeed, the birds did not generalize
their learned avoidance among the wood tiger moth morphs, but
instead treated them as different prey types based on the differing
hindwing colour alone, as the morph models used did not differ in
size, shape, pattern, taste or smell.
The Importance of Colour
Our ndings are in line with previous experiments showing that
colour is of foremost importance in avian predator learning,
contributing especially to the discrimination between palatable
and unpalatable prey (Aronsson &Gamberale-Stille, 2008; Kazemi,
Gamberale-Stille, Tullberg, &Leimar, 2014). A large body of
research has demonstrated birds' ability to learn to avoid con-
spicuous, unpalatable prey (Aronsson &Gamberale-Stille, 2008;
Rowe, Lindstr
om, &Lyytinen, 2004; Sv
adov
a et al., 2009). This is
because conspicuous warning coloration enhances prey recogni-
tion (Guilford, 1986; Sherratt &Beatty, 2003), speed of avoidance
learning and memorability (e.g. Roper &Redston, 1987). Different
predators may use different components of the warning signal as a
primary cue depending on their sensory systems (Aronsson &
Gamberale-Stille, 2012; Endler, 1992; Guilford &Dawkins, 1991)
and disregard others. Studies done with birds have demonstrated
that colour seems to be a more important feature in warning signals
than size or pattern (Aronsson &Gamberale-Stille, 2008; Exnerov
a
et al., 2006; Sill
en-Tullberg, 1985; Terhune, 1977).
As predators can associate palatability or unpalatability with
several different kinds of prey traits, it is convenient to compare the
relative importance of those traits with how much they facilitate
associative learning. The expectation is that more salient signals are
learned faster (Kazemi et al., 2014). Our results indicate that red
was the most salient warning colour: birds learned to avoid the red
morph faster than the other morphs. This is in accordance with
Lindstedt et al. (2011), who found that the red female morph of the
0 5 10 15 20 25
Number of trials
0
0.2
0.4
0.6
0.8
1
Cumulative attack risk
Figure 3. Proportion of models attacked during the learning trials for each colour
model. The lines represent the cumulative attacks on unpalatable moth models of
white (black line), yellow (yellow line) or red (red line) hindwing colour.
0
0.25
0.50
0.75
1
White Yellow Red
Colour learned
Mean attack probability ± SE
Figure 4. Proportions (±SE) of models that were attacked in the rst trial of the
generalization test. Symbol styles refer to white (black circle), yellow (yellow square)
and red (red triangle) models offered in the test.
K. R
onk
a et al. / Animal Behaviour 135 (2018) 153e163158
wood tiger moth was better protected against bird predators,
suffering fewer attacks than its orange or yellow counterparts.
Indeed, red has been shown to be a very efcient warning signal
compared to other warning colours such as orange, yellow or white,
and other colours such as violet, blue, green and brown, at least for
some bird predators (Cibulkov
a, Veselý, &Fuchs, 2014; Exnerov
a
et al., 2006; Gamberale-Stille &Tullberg, 1999; Lindstedt et al.,
2011; Sv
adov
a et al., 2009).
In the present study prey items were made to resemble real
wood tiger moth morphs as closely as possible, keeping all traits
other than hindwing colour constant. This allowed us to compare
the effects of warning coloration of hindwings only. Changing the
warning colour hue altered not only the internal contrast on model
hindwings, but also the contrast between the model and the green
background. Although all colours in our experiment were clearly
conspicuous to the birds, red had the highest colour contrast
against the green background whereas white had the lowest. This
might explain why red seems to be the most salient signal.
Aronsson and Gamberale-Stille (2009) found similar results using
domestic chicks, Gallus gallus domesticus, which learned to avoid
red prey faster if presented on a contrasting background compared
to a background of similar hue. In another experiment, however,
red prey colour was found to inuence predator avoidance inde-
pendent of background colour (Sill
en-Tullberg, 1985). Thus, it
seems that both prey coloration per se and its contrast against the
background can contribute to predator avoidance, but it is still
relatively unclear which properties of prey coloration, chromatic or
achromatic, play the most important role. Previous work with
wood tiger moths has shown that the achromatic contrast against a
green background is highest for white morphs, which are the most
luminous of the three (Henze, Lind, Mappes, Rojas, &Kelber, 2017;
Lindstedt et al., 2011). Luminance has not been found to affect
predator responses towards the wood tiger moth, while the chro-
matic contrast in hue seems to be very important (Nokelainen et al.,
2012).
Generalization has been suggested to stabilize selection towards
aposematic signals via a peak shift phenomenon (Leimar, Enquist,
&Sillen-Tullberg, 1986; Lindstr
om, Alatalo, Mappes, Riipi et al.,
1999). The minimum (and maximum) responses of predators (i.e.
peaks of the generalization gradient) have been found to be dis-
placed from the negative (and positive) stimulus (Gamberale &
Tullberg, 1996; Hanson, 1959), such as yellow, towards a similar,
but more salient novel stimulus, such as red. Overall, we did not
nd strong evidence of asymmetric generalization, but there were
some trends between the colours tested. Birds that learned to avoid
red models attacked almost all the white and yellow models in the
generalization trials, whereas birds that learned the less salient
colours yellow and white generalized more, hinting at a tendency
to generalize from the less salient signals towards the more salient
signal.
Sv
adov
a et al. (2009) found asymmetric generalization using
great tits, Parus major, which did not generalize from red rebugs,
Pyrrhocoris apterus, to white or yellow mutants, but did generalize
from yellow mutants to red rebugs. Interestingly, in our experi-
ment, four blue tits that learned to avoid the yellow morph (N¼15)
refrained from attacking white models, while only two did so for
the red ones. Birds that learned to avoid the white morph attacked
both unlearned morphs equally, but chose red models rst more
often than yellow ones. This indicates that birds tended to gener-
alize more between the white and yellow than between white and
red. The yellow morph seems to benet least from the other col-
ours, since only between 7 and 34% of yellow models were left
unattacked (Table 2,Fig. 4).
Limitations of Testing Generalization in the Laboratory
Despite the majority of birds showing no generalization in our
experiment, the possibility remains that predators might gener-
alize among morphs of the wood tiger moth under different cir-
cumstances. Studying generalization in the wild is practically
impossible due to the rareness of predation events on aposematic
prey as well as difculties in observing the choices of individual
predators in natural conditions. A previous study aiming to explain
the variability in the warning signals of the harlequin poison frog,
Oophaga histrionica, showed that predators avoided attacking
aposematic frog models but not cryptic ones in areas where
aposematic frogs occur, exhibiting some generalization among
different frog colour morphs in the eld. However, the same study
found no generalization by naïve chicks tested in the laboratory
(Am
ezquita et al., 2013). This might imply that naïve and experi-
enced predators in the wild can use different generalization stra-
tegies (see also Ihalainen, Lindstr
om, Mappes, &Puolakkainen,
2008). Birds might also be prone to generalize more or less
widely under different circumstances (Aronsson &Gamberale-
Stille, 2012), for example under physiological stress during winter
months (Barnett, Bateson, &Rowe, 2007; Chatelain, Halpin, &
Rowe, 2013; Veselý, Ernestov
a, Nedv
ed, &Fuchs, 2017), limited
food availability (Ihalainen, Rowland, Speed, Ruxton, &Mappes,
2012; Lindstr
om, Alatalo, Lyytinen, &Mappes, 2004), limited time
to make decisions (Ings &Chittka, 2008), when the prey is
dangerously toxic (Lindstr
om, Alatalo, &Mappes, 1997; Sherratt,
2002), when the prey community is complex versus simple
(Ihalainen et al., 2007), or when the prey population has palatable
Batesian mimics in addition to the unprotable prey (Plowright &
Owen, 1980).
Avoidance learning has been suggested to happen in two steps:
rst, the birds learn simple rules based on certain cues, and once
the basic rules are formed, they then learn in more detail about
prey quality (Chittka &Osorio, 2007). Recent studies indicate that
birds are able to assess the nutritional benets of unprotable prey
and use this information in subsequent encounters (Halpin et al.,
2014). This ability could have affected not only bird learning
rates, but also their decision to attack in the generalizationphase of
our experiment. As our models' pastry bodies were of high nutri-
tional value and the birds were hungry, it is possible that the birds
were willing to take more risks and thus took more trials to learn to
avoid the models than it would take them to learn to avoid
defended prey in the wild. In addition, the 5 min presentations gave
the birds plenty of time to decide whether to attack or not, and to
make more sophisticated assessments of prey quality than might
be possible in the wild. Birds were given alternative food between
the presentations, but not enough for saturation, and would thus
have beneted energetically from discriminating between the un-
protable and protable models. Nevertheless, the costebenet
relationship was exactly the same for all morphs in our experiment
Table 2
Proportions (±SE) of tested colour morphs attacked in the generalization test trials
(phase 3) in relation to the colour learned
Colour learned Colour tested First trial Second trial Third trial
White Yellow 0.85±0.09 0.92±0.07 0.64±0.13
White Red 0.85±0.09 0.78±0.11 0.78±0.11
Yellow White 0.73±0.11 0.73±0.11 0.73±0.11
Yellow Red 0.86±0.09 0.73±0.11 0.60±0.13
Red White 0.93±0.06 0.86±0.09 0.73±0.11
Red Yellow 0.93±0.06 0.86±0.09 0.86±0.09
K. R
onk
a et al. / Animal Behaviour 135 (2018) 153e163 159
and, thus, we can safely compare the relative differences between
morphs in their salience.
The avoidance learning was based on counterconditioning,
where the colour signal was rst associated with a positive rein-
forcement (i.e. palatability) and then with a negative reinforcement
(i.e. unpalatability). Previous research has shown that in cases of
single counterconditioning the associations learned second are
forgotten at higher rates than those learned rst (Speed, 2000 and
references therein). Therefore, it is possible that the birds' experi-
ence and learned association with palatability in the preference test
exceeded the effect of generalized avoidance among the morphs for
most of the birds, which could partly explain the low level of
generalization observed. Offering the models as palatable at the
beginning of the experiment was necessary to get rid of any pre-
existing biases or neophobia prior to learning and testing gener-
alization effects of the birds; this was also necessary to motivate the
birds to attack and taste the unpalatable models during the rst
learning trials.
In the preference and generalization tests, simultaneous prey
choice was used to decrease the numbers of birds needed to
accomplish the experiment. Simultaneous prey choice is also a very
powerful set-up to detect any potential predator biases but, obvi-
ously, this approach has disadvantages too (Fig. 2). For example, it is
possible that long hesitation delays during the rst presentation of
the preference test were partly due to an aggregation effect, as
aggregations of conspicuous prey have been found to be aversive to
predators (Gamberale-Stille, 2000; Riipi, Alatalo, Lindstrom, &
Mappes, 2001). On the other hand, Nokelainen et al. (2012) pre-
sented wood tiger moths singly to birds, several of which also
hesitated for a long time before attacking them. Thus, it is difcult
to say how much the simultaneous presentation inuenced our
results, but during the ying season wood tiger moth morphs
typically aggregate at the same sites. Males of both morphs are
often found near calling females, and thus all morphs can be visible
and vulnerable to predators simultaneously.
Lastly, if the wood tiger moths are able to survive bird attacks,
the use of articial models does not necessarily give an accurate
estimate of selection. A considerable proportion of attacks in the
generalization test were just a single peck, leaving the models
uneaten, and thus not necessarily killed. The birds' willingness to
attack but reluctance to consume the models could stem from the
psychology of birds' decision making (Marples &Kelly, 1999).
Adamov
a-Je
zov
a et al. (2016) showed that for great tits and coal
tits, Periparus ater, neophobia (i.e. the avoidance of novel prey
affecting the decision to attack), but not dietary conservatism (i.e.
restriction of diet to certain prey types affecting the decision to
consume the prey), was deactivated during pretraining with a
palatable prey, but the initial hesitation of blue tits was not
affected by earlier experience. Blue tits have been found to show
higher general aversion even towards palatable prey than, for
example, great tits, probably because of higher dietary conserva-
tism (Prokopov
a, Veselý, Fuchs, &Zrzavý, 2010; Turini et al., 2016;
Veselý, Luhanov
a, Pr
a
skov
a, &Fuchs, 2013; Veselý, Vesel
a, Fuchs, &
Zrzavý, 2006). This indicates that predators' decisions of whether
to attack or not after avoidance learning might be species specic
and, thus, not generalizable from blue tits to other predators. In our
experiment, 45 of 53 blue tits overcame their initial avoidance
during the preference test and attacked and consumed the models
readily in the following avoidance trial. As those birds that did not
overcome their hesitation during the preference test were not
included in the following phases of the experiment, our results
describe the generalization tendency of the less hesitant in-
dividuals, which are more likely to attack aposematic prey in the
wild in the rst place. Many of these less hesitant individuals,
however, seemed to regain their dietary conservatism after they
had learned avoidance, as they no longer consumed the palatable
models attacked. In conclusion, whereas no generalization was
found regarding the attack probabilities, we did nd individual
variation in avoidance learning and dietary conservatism, which
could affect selection in the wild.
The Importance of Other Cues
Somewhat surprisingly, the blue tits had no initial biases to-
wards any of the hindwing colours. Earlier studies on the wood
tiger moth have found differential predation pressure in the eld
(Lindstedt et al., 2011; Nokelainen et al., 2012, 2014) and different
hesitation times by local predators (Lindstedt et al., 2011;
Nokelainen et al., 2012) towards the different colour morphs. As
the differences in hesitation times were found using living moths
(Nokelainen et al., 2012), it is possible that other cues, such as
odour, inuenced the results. In nature, the wood tiger moth relies
on multiple signal components (i.e. odour, taste) in addition to the
visual cues when exposed to potential predators (Rojas, Burdeld-
Steel &Mappes 2015). Its chemical defence contains pyrazines
(Burdeld-Steel et al., 2016; Rojas et al., 2017), a group of com-
pounds with a characteristic aversive smell, which is effective
against birds (Guilford, Nicol, Rothschild, &Moore, 1987; Rowe &
Guilford, 1996). In fact, pyrazine is known to trigger hidden aver-
sions to red and yellow colours (Rowe &Guilford, 1996) and con-
spicuous prey (Lindstr
om, Rowe, &Guilford, 2001), and enhance
both learning and memorability of yellow (Siddall &Marples, 2008)
or red coloured prey (Barnea, Gvaryahu, &Rothschild, 2004)at
least in domestic chicks (Siddall &Marples, 2008). Pyrazine odour
has been associated with Müllerian mimicry rings of insects and
suggested to function as a warning signal (Rothschild, 1961). On the
other hand, pyrazine odour has also been shown to assist in
discriminating prey and thus reduce avoidance generalization be-
tween differently coloured prey if the odour is present on only
some of them (Rowe &Guilford, 1996; Siddall &Marples, 2008).
The specic roles of different cues in predatoreprey interactions
are uncertain. It might be that odour is easy to associate with
palatability, but only functions close up, whereas conspicuous
colours aid in memorizing which prey to avoid even from a dis-
tance. Here we were interested in colour only. However, it has been
shown that when colour is kept constant, predators can discrimi-
nate prey based on pattern (e.g. Prokopov
a et al., 2010; Veselý et al.,
2013), and when both colour and pattern are equal, other visual
and/or chemical features of the prey are used for prey recognition
(Karlíkov
a et al., 2016).
Conclusions
Overall, the colour polymorphism of the wood tiger moth in
Finland seems unlikely to be maintained by generalized avoidance
based on its warning coloration only. However, predators were
hesitant to attack any of the aposematic morphs in the rst place,
and if they were to encounter them in the wild sharing other
warning cues such as pyrazine odour, general aversion seems likely
to occur. More knowledge on how predators acquire and use in-
formation on prey qualities in different contexts is needed to
conclude whether predator generalization contributes to the
maintenance of multiple aposematic morphs (see also Skelhorn
et al., 2016). The possibility that wild predators can generalize on
the basis of the pyrazine odour or the combination of colour and
odour requires further investigation. Alternative explanations for
K. R
onk
a et al. / Animal Behaviour 135 (2018) 153e163160
the occurrence of local warning signal polymorphism include
negative frequency-dependent natural selection, sexual selection,
frequency-dependent ight activity (Rojas, Gordon, &Mappes,
2015), signal efcacy trade-offs with other life history traits
(Hegna et al., 2013; Nokelainen et al., 2012), predator species-
specic mortality differences between morphs (Nokelainen et al.,
2014) or combinations of these mechanisms (Gordon, Kokko,
Rojas, Nokelainen, &Mappes, 2015). Colour polymorphism could
also be explained by multiple-model mimicry (Edmunds, 2000), if
the different morphs share warning colours with other defended
prey species, and predators generalize their avoidance from one
species to the other based on similar coloration. Thus, generaliza-
tion of learned avoidance remains as a possible contributor to the
maintenance of local polymorphism in wood tiger moth pop-
ulations. In conclusion, we argue based on our results that although
predator generalization could well contribute to the maintenance
of different aposematic morphs under certain circumstances, it is
unlikely to occur among distinct colour morphs of otherwise
similar prey of visually oriented avian predators.
Acknowledgments
We are indebted to Helin
a Nisu and Tuuli Salmi for help with the
birds at Konnevesi Research Station, and to Andr
es L
opez-Sepulcre,
Sebastiano De Bona and Janne Valkonen for their thoughtful advice
on statistical analyses. We also thank the Darwin groupat the
University of Jyv
askyl
a for discussions and two anonymous referees
and the editor for helpful comments on the manuscript. This study
was funded by the Centre of Excellence in Biological Interactions
(Academy of Finland, project no. 284666). C.D.P. was funded by the
Erasmus Exchange Programme.
References
Adamov
a-Je
zov
a, D., Hospodkov
a, E., Fuchsov
a, L.,
Stys, P., & Exnerov
a, A. (2016).
Through experience to boldness? Deactivation of neophobia towards novel and
aposematic prey in three European species of tits (Paridae). Behavioural Pro-
cesses, 131,24e31. https://doi.org/10.1016/j.beproc.2016.07.014.
Am
ezquita, A., Castro, L., Arias, M., Gonz
alez, M., & Esquivel, C. (2013). Field but not
lab paradigms support generalisation by predators of aposematic polymorphic
prey: The Oophaga histrionica complex. Evolutionary Ecology, 27(4), 769e782.
https://doi.org/10.1007/s10682-013-9635-1.
Aronsson, M., & Gamberale-Stille, G. (20 08). Domestic chicks primarily attend to
colour, not pattern, when learning an aposematic coloration. Animal Behaviour,
75,417e423. https://doi.org/10.1016/j.anbehav.2007.05.006. ISSN 0003-3472.
Aronsson, M., & Gamberale-Stille, G. (2009). Importance of internal pattern contrast
and contrast against the background in aposematic signals. Behavioral Ecology,
20(6), 1356e1362. https://doi.org/10.1093/beheco/arp141.
Aronsson, M., & Gamberale-Stille, G. (2012). Colour and pattern similarity in mim-
icry: Evidence for a hierarchical discriminative learning of different components.
Animal Behaviour, 84(4), 881e887. https://doi.org/10.1016/j.anbehav.2012.07.011.
Barnea, A., Gvaryahu, G., & Rothschild, M. (2004). The effect of the odour of pyrazine
and colours on recall of past events and learning in domestic chicks (Gallus
gallus domesticus). In H. F. Van Emden, & M. Rothschild (Eds.), Insect and bird
interactions (pp. 205e216). Andover, U.K.: Intercept Ltd.
Barnett, C. A., Bateson, M., & Rowe, C. (2007). State-dependent decision making:
Educated predators strategically trade off the costs and benets of consuming
aposematic prey. Behavioral Ecology, 18(4), 645e651. https://doi.org/10.1093/
beheco/arm027.
Bates, D., Maechler, M., Bolker, B., & Walker, S. (2015). Fitting linear mixed-effects
models using lme4. Journal of Statistical Software, 67,1e48. https://doi.org/
10.18637/jss.v067.i01.
Burdeld-Steel, E., Pakkanen, H., Rojas, B., Galarza, J. A., & Mappes, J. (2016). De novo
synthesis of chemical defences in an aposematic moth. Submitted manuscript.
ten Cate, C., & Rowe, C. (2007). Biases in signal evolution: Learning makes a dif-
ference. Trends in Ecology &Evolution, 22(7), 380e387.
Chatelain, M., Halpin, C. G., & Rowe, C. (2013). Ambient temperature inuences
birds' decisions to eat toxic prey. Animal Behaviour, 86(4), 733e740. https://
doi.org/10.1016/j.anbehav.2013.07.007.
Chittka, L., & Osorio, D. (2007). Cognitive dimensions of predator responses to
imperfect mimicry. PLoS Biology, 5(12), e339. https://doi.org/10.1371/journal.
pbio.0050339.
Chouteau, M., Arias, M., & Joron, M. (2016). Warning signals are under positive
frequency-dependent selection in nature. Proceedings of the National Academy of
Sciences of the United States of America, 113(8), 2164e2169.
Cibulkov
a, A., Veselý, P., & Fuchs, R. (2014). Importance of conspicuous colours in
warning signals: The great tit's (Parus major) point of view. Evolutionary Ecology,
28(3), 427e439. https://doi.org/10.1007/s10682-014-9690-2.
Dimitrova, M., & Merilaita, S. (2010). Prey concealment: Visual background
complexity and prey contrast distribution. Behavioral Ecology, 21(1), 176e181.
https://doi.org/10.1093/beheco/arp174.
Edmunds, M. (2000). Why are there good and poor mimics? Biological Journal of the
Linnean Society, 70(3), 459e466. https://doi.org/10.1111/j.1095-8312.200 0.
tb01234.x.
Endler, J. A. (1991). Interactions between predators and prey. In J. R. Krebs, &
N. B. Davies (Eds.), Behavioural ecology (3rd ed., pp. 169e203). London: Black-
well Scientic.
Endler, J. A. (1992). Signals, signal condition and the direction of evolution. Amer-
ican Naturalist, 139,S125eS153. https://doi.org/10.1086/285308.
Endler, J. A., & Mappes, J. (2004). Predator mixes and the conspicuousness of
aposematic signals. American Naturalist, 163(4), 532e547. https://doi.org/
10.1086/382662.
Endler, J. A., & Mielke, P. W., Jr. (2005). Comparing entire colour patterns as birds see
them. Biological Journal of the Linnean Society, 86(4), 405e431. https://doi.org/
10.1111/j.1095-8312.2005.00540.x.
Exnerov
a, A., Jezov
a, D., Stys, P., Doktorovov
a, L., Rojas, B., & Mappes, J. (2015).
Different reactions to aposematic prey in 2 geographically distant populations
of great tits. Behavioral Ecology, 26(5), 1361e1370. https://doi.org/10.1093/
beheco/arv086.
Exnerov
a, A., Landov
a, E.,
Stys, P., Fuchs, R., Prokopov
a, M., & Cehl
arikov
a, P. (2003).
Reactions of passerine birds to aposematic and non-aposematic rebugs (Pyr-
rhocoris apterus; Heteroptera). Biological Journal of the Linnean Society, 78(4),
517e525. https://doi.org/10.1046/j.0024-4066.20 02.00161.x.
Exnerov
a, A.,
Stys, P., Fu
cíkov
a, E., Vesel
a, S., Sv
adova
́
, K., Prokopovv
a, M., et al.
(2007). Avoidance of aposematic prey in European tits (Paridae): Learned or
innate? Behavioral Ecology, 18,148e156.
Exnerov
a, A. , Sv
adov
a, K. H., Fu
cíkov
a, E., Drent, P., &
Stys, P. (2010). Personality
matters: Individual variation in reactions of naive bird predators to aposematic
prey. Proceedings of the Royal Society B: Biological Sciences, 277(1682), 723e728.
https://doi.org/10.1098/rspb.2009.1673.
Exnerov
a, A. , Sv
adova
́
, K., Stys, P., Barcalov
a, S., Landov
a, E., Prokopovv
a, M., et al.
(2006). Importance of colour in the reaction of passerine predators to apose-
matic prey: Experiments with mutants of Pyrrhocoris apterus (Heteroptera).
Biological Journal of the Linnean Society, 88,143e153.
Galarza, J. A., Nokelainen, O., Ashra, R., Hegna, R. H., & Mappes, J. (2014). Temporal
relationship between genetic and warning signal variation in the aposematic
wood tiger moth (Parasemia plantaginis). Molecular Ecology, 23(20), 4939e4957.
https://doi.org/10.1111/mec.12913.
Galarza, J. A., Nokelainen, O., & Mappes, J. (2016). Genetic architecture of colour
polymorphism in the aposematic moth Arctia plantaginis. Manuscript in
preparation.
Gamberale, G., & Tullberg, B. S. (1996). Evidence for a peak-shift in predator
generalization among aposematic prey. Proceedings of the Royal Society B: Bio-
logical Sciences, 263(1375), 1329e1334.
Gamberale-Stille, G. (2000). Decision time and prey gregariousness inuence attack
probability in naive and experienced predators. Animal Behaviour, 60,95e99.
https://doi.org/10.1006/anbe.2000.1435.
Gamberale-Stille, G., & Tullberg, B. S. (1999). Experienced chicks show biased
avoidance of stronger signals: An experiment with natural colour variation in
live aposematic prey. Evolutionary Ecology, 13(6), 579e589. https://doi.org/
10.1023/A:1006741626575.
Ghirlanda, S., & Enquist, M. (2003). A century of generalization. Animal Behaviour,
66,15e36.
Gordon, S. P., Kokko, H., Rojas, B., Nokelainen, O., & Mappes, J. (2015). Colour
polymorphism torn apart by opposing positive frequency-dependent selection,
yet maintained in space. Journal of Animal Ecology, 84(6), 1555e1564. https://
doi.org/10.1111/1365-2656.12416.
Guilford, T. (1986). How do warning colourswork? Conspicuousness may reduce
recognition errors in experienced predators. Animal Behaviour, 34, 286e288.
https://doi.org/10.1016/0003-3472(86)90034-5.
Guilford, T., & Dawkins, M. S. (1991). Receiver psychology and the evolution of
animal signals. Animal Behaviour, 42,1e14.
Guilford, T., Nicol, C., Rothschild, M., & Moore, B. P. (1987). The biological roles of
pyrazines: Evidence for a warning odour function. Biological Journal of the Linnean
Society, 31(2),113e128. https://doi.org/10.1111/j.1095-8312.1987.tb01984.x.
Halpin, C. G., Skelhorn, J., & Rowe, C. (2014). Increased predation of nutrient-
enriched aposematic prey. Proceedings of the Royal Society B: Biological Sci-
ences, 281(1781). https://doi.org/10.1098/rspb.2013.3255.
Ham, A. D., Ihalainen, E., Lindstrom, L., & Mappes, J. (20 06). Does colour matter? The
importance of colour in avoidance learning, memorability and generalisation.
Behavioral Ecology and Sociobiology, 60(4), 482e491. https://doi.org/10.1007/
s00265-006-0190-4.
Hanson, H. M. (1959). Effects of discrimination training on stimulus generalization.
Journal of Experimental Psychology, 58(5), 321.
K. R
onk
a et al. / Animal Behaviour 135 (2018) 153e163 161
Hart, N. S., Partridge, J. C., Cuthill, I. C., & Bennett, A. T. D. (2000). Visual pigments, oil
droplets, ocular media and cone photoreceptor distribution in two species of
passerine bird: The blue tit (Parus caeruleus L.) and the blackbird (Turdus merula
L.). Journal of Comparative Physiology A, 186(4), 375e387. https://doi.org/
10.1007/s003590050437.
Hart, N. S., & Vorobyev, M. (2005). Modelling oil droplet absorption spectra and
spectral sensitivities of bird cone photoreceptors. Journal of Comparative Phys-
iology A, 191(4), 381e392. https://doi.org/10.1007/s00359-004-0595-3.
Hegna,R. H., Galarza, J. A.,& Mappes, J. (2015). Globalphylogeography and geographical
variation in warning coloration of the wood tiger moth (Parasemia plantaginis).
Journal of Biogeography, 42, 1469e1481. https://doi.org/10.1111/jbi.12513.
Hegna, R. H., & Mappes, J. (2014). Inuences of geographic differentiation in the
forewing warning signal of the wood tiger moth in Alaska. Evolutionary Ecology,
28(6), 1003e1017. https://doi.org/10.1007/s10682-014-9734-7.
Hegna, R. H., Nokelainen, O., Hegna, J. R., & Mappes, J. (2013). To quiver or to shiver:
Increased melanization benets thermoregulation, but reduces warning signal
efcacy in the wood tiger moth. Proceedings of the Royal Society B: Biological
Sciences, 280(1755). https://doi.org/10.1098/rspb.2012.2812, 20122812.
Henze, M., Lind, O., Mappes, J., Rojas, B., & Kelber, A. (2017). An aposematic colour-
polymorphic moth seen through the eyes of conspecics and predators eSensi-
tivity and colour discrimination in Arctia plantaginis. Submitted manuscript.
Hong, D. D. (1976). Chloroquine phosphate. Analytical Proles of Drug Substances, 5,
61e85. https://doi.org/10.1016/S0099-5428(08)60315-9.
Ihalainen, E., Lindstr
om, L., & Mappes, J. (2007). Investigating Müllerian mimicry:
Predator learning and variation in prey defences. Journal of Evolutionary Biology,
20(2), 780e791. https://doi.org/10.1111/j.1420-9101.2006.01234.x.
Ihalainen, E., Lindstr
om, L., Mappes, J., & Puolakkainen, S. (2008). Can experienced
birds select for Müllerian mimicry? Behavioral Ecology, 19, 362e368. https://
doi.org/10.1093/beheco/arm151.
Ihalainen, E., Rowland, H. M., Speed, M. P., Ruxton, G. D., & Mappes, J. (2012). Prey
community structure affects how predators select for Müllerian mimicry. Pro-
ceedings of the Royal Society B: Biological Sciences, 279(1736), 2099. https://
doi.org/10.1098/rspb.2011.2360.
Ings, T. C., & Chittka, L. (2008). Speed-accuracy tradeoffs and false alarms in bee
responses to cryptic predators. Current Biology, 18(19), 1520e1524. https://
doi.org/10.1016/j.cub.2008.07.074.
Jiggins, C. D., & McMillan, W. O. (1997). The genetic basis of an adaptive radiation:
Warning colour in two Heliconius species. Proceedings of the Royal Society B:
Biological Sciences, 264(1385), 1167. https://doi.org/10.1098/rspb.1997.0161.
Joron, M., & Mallet, J. (1998). Diversity in mimicry: Paradox or paradigm? Trends in
Ecology &Evolution, 13,461e463.
Karlíkov
a, Z., Veselý, P., Ber
ankov
a, J., & Fuchs, R. (2016). Low ability of great tits to
discriminate similarly inconspicuous edible and inedible prey. Ethology, 122(2),
180 e191. https://doi.org/10.1111/eth.12454.
Kazemi, B., Gamberale-Stille, G., & Leimar, O. (2015). Multi-trait mimicry and the
relative salience of individual traits. Proceedings of the Royal Society B: Biological
Sciences, 282(1818), 2015e2127. https://doi.org/10.1098/rspb.2015.2127.
Kazemi, B., Gamberale-Stille, G., Tullberg, B. S., & Leimar, O. (2014). Stimulus
salience as an explanation for imperfect mimicry. Current Biology, 24(9),
965e969. https://doi.org/10.1016/j.cub.2014.02.061.
Kikuchi, D. W., Mappes, J., Sherratt, T. N., & Valkonen, J. K. (2016). Selectionfor multi-
component mimicry: Equal feature salience and variation in preferred traits.
Behavioral Ecology, 27(5), 1515e15 21. https://doi.org/10.1093/beheco/arw072.
Leimar, O., Enquist, M., & Sillen-Tullberg, B. (1986). Evolutionary stability of
aposematic coloration and prey unprotability: A theoretical analysis. American
Naturalist, 128(4), 469e490. https://doi.org/10.1086/284581.
Lindstedt, C., Eager, H., Ihalainen, E., Kahilainen, A., Stevens, M., & Mappes, J. (2011).
Direction and strength of selection by predators for the color of the aposematic
wood tiger moth. Behavioral Ecology, 22(3), 580e587. https://doi.org/10.1093/
beheco/arr017.
Lindstedt, C., Schroderus, E., Lindstr
om, L., Mappes, T., & Mappes, J. (2016). Evolu-
tionary constraints of warning signals: A genetic trade-off between the efcacy
of larval and adult warning coloration can maintain variation in signal
expression. Evolution, 70(11), 2562e2572. https://doi.org/10.1111/evo.13066.
Lindstr
om, L., Alatalo, R. V., Lyytinen, A., & Mappes, J. (2001). Strong antiapostatic
selection against novel rare aposematic prey. Proceedings of the National
Academy of Sciences of the United States of America, 98(16), 9181e9184. https://
doi.org/10.1073/pnas.161071598.
Lindstr
om, L., Alatalo, R. V., Lyytinen, A., & Mappes, J. (20 04). The effect of alter-
native prey on the dynamics of Batesian and Müllerian mimicries. Evolution, 58,
1294e1302. https://doi.org/10.1554/03-271.
Lindstr
om, L., Alatalo, R. V., & Mappes, J. (1997). Imperfect Batesian mimicrydthe
effects of the frequency and the distastefulness of the model. Proceedings of the
Royal Society B: Biological Sciences, 264(1379), 149. https://doi.org/10.1098/
rspb.1997.0022.
Lindstr
om, L., Alatalo, R. V., & Mappes, J. (1999). Reactions of hand-reared and wild-
caught predators toward warningly colored, gregarious, and conspicuous prey.
Behavioral Ecology, 10(3), 317e322. https://doi.org/10.1093/beheco/10.3.317.
Lindstr
om, L., Alatalo, R. V., Mappes, J., Riipi, M., & Vertainen, L. (1999). Can
aposematic signals evolve by gradual change? Nature, 397(6716), 249e251.
https://doi.org/10.1038/16692.
Lindstr
om, L., Rowe, C., & Guilford, T. (2001). Pyrazine odour makes visually con-
spicuous prey aversive. Proceedings of the Royal Society B: Biological Sciences,
268(1463), 159e162 .
Lynn, S. K. (2005). Learning to avoid aposematic prey. Animal Behaviour, 70,
1221e1226. https://doi.org/10.1016/j.anbehav.2005.03.010. ISSN 00 03e3472.
Mackintosh, N. J. (1975). A theory of attention: Variations in the associability of
stimuli with reinforcement. Psychological Review, 82(4), 276e298.
Mallet, J., & Barton, N. H. (1989). Strong natural selection in a warning-color hybrid
zone. Evolution, 43(2), 421e431.
Mallet, J., & Joron, M. (1999). Evolution of diversity in warning color and mimicry:
Polymorphisms, shifting balance, and speciation. Annual Review of Ecology and
Systematics, 30,201
e233. https://doi.org/10.1146/annurev.ecolsys.30.1.201.
Mappes, J., & Alatalo, R. V. (1997). Batesian mimicry and signal accuracy. Evolution,
51,20148e22051. https://doi.org/10.1111/j.1558-5646.1997.tb05129.x.
Marples, N. M., & Kelly, D. J. (1999). Neophobia and dietary conservatism: Two
distinct processes? Evolutionary Ecology, 13(7e8), 641e653. https://doi.org/
10.1023/a:1011077731153.
Marples, N. M., & Mappes, J. (2011). Can the dietary conservatism of predators
compensate for positive frequency dependent selection against rare, conspic-
uous prey? Evolutionary Ecology, 25(4), 737e749 .
Mettke-Hofmann, C., Winkler, H., & Leisler, B. (2002). The significance of ecological
factors for exploration and neophobia in parrots. Ethology, 108(3), 249e272.
https://doi.org/10.1046/j.1439-0310.2002.00773.x.
Müller, F. (1878). Ueber die Vortheile der Mimicry bei Schmetterlingen. Zoologischer
Anzeiger, 1,54e55.
Nokelainen, O., Hegna, R. H., Reudler, J. H., Lindstedt, C., & Mappes, J. (2012). Trade-
off between warning signal efcacy and mating success in the wood tiger moth.
Proceedings of the Royal Society B: Biological Sciences, 279(1727), 257e265.
https://doi.org/10.1098/rspb.2011.0880.
Nokelainen, O., Valkonen, J., Lindstedt, C., & Mappes, J. (2014). Changes in predator
community structure shifts the efcacy of two warning signals in Arctiid moths.
Journal of Animal Ecology, 83, 598e605. https://doi.org/10.1111/1365-
2656.12169.
O'Donald, P., & Majerus, M. E. N. (1984). Polymorphism of melanic ladybirds
maintained by frequency-dependent sexual selection. Biological Journal of the
Linnean Society, 23(2e3), 101e111. https://doi.org/10.1111/j.1095-8312.1984.
tb00131.x.
Pavlov, P. I. (1927). Conditioned reexes: An investigation of the physiological ac-
tivity of the cerebral cortex. Annals of Neurosciences, 17(3), 136.
Plowright, R. C., & Owen, R. E. (1980). The evolutionary signicance of bumble bee
color patterns: A mimetic interpretation. Evolution, 34(4), 622e637. https://
doi.org/10.2307/2408017.
Poulton, E. B. (1890). The colours of animals: Their meaning and use (Vol. 26). London,
U.K.: Kegan Paul, Trench, Trubner.
Prokopov
a, M., Veselý, P., Fuchs, R., & Zrzavý, J. (2010). The role of size and colour
pattern in protection of developmental stages of the red rebug (Pyrrhocoris
apterus) against avian predators. Biological Journal of the Linnean Society, 100(4),
890e898. https://doi.org/10.1111/j.1095-8312.2010.01463.x.
Pr
uchov
a, A., Nedv
ed, O., Veselý, P., Ernestov
a, B., & Fuchs, R. (2014). Visual warning
signals of the ladybird Harmonia axyridis: The avian predators' point of view.
Entomologia Experimentalis et Applicata, 151(2), 128e134. https://doi.org/
10.1111/eth.12454.
Riipi, M., Alatalo, R. V., Lindstrom, L., & Mappes, J. (2001). Multiple benets of
gregariousness cover detectability costs in aposematic aggregations. Nature,
413(6855), 512e514. https://doi.org/10.1038/35097061.
Rojas, B., Burdeld-Steel, E., & Mappes, J. (2015). Primary vs. secondary defences in an
aposematic moth: trade-off or additive effect?. Submitted manuscript.
Rojas, B., Burdeld-Steel, E., Pakkanen, H., Suisto, K., Maczka, M., Schulz, S., et al.
(2017). How to ght multiple enemies: Target-specic chemical defences in an
aposematic moth. Proceedings of the Royal Society B: Biological Sciences, 284.
https://doi.org/10.1098/rspb.2017.1424, 20171424.
Rojas, B., & Endler, J. A. (2013). Sexual dimorphism and intra-populational colour
pattern variation in the aposematic frog Dendrobates tinctorius.Evolutionary
Ecology, 27(4), 739e753. https://doi.org/10.1007/s10682-013-9640-4.
Rojas, B., Gordon, S. P., & Mappes, J. (2015). Frequency-dependent ight activity in
the colour polymorphic wood tiger moth. Current Zoology, 61,765e772.
Rojas, B., Rautiala, P., & Mappes, J. (2014). Differential detectability of polymorphic
warning signals under varying light environments. Behavioural Processes, 109,
164e172. https://doi.org/10.1016/j.beproc.2014.08.014.
R
onk
a, K., Mappes, J., Kaila, L., & Wahlberg, N. (2016). Putting Parasemia in its
phylogenetic place: A molecular analysis of the subtribe Arctiina (Lepidoptera).
Systematic Entomology.https://doi.org/10.1111/syen.12194. n/aen/a.
Roper, T. J., & Redston, S. (1987). Conspicuousness of distasteful prey affects the
strength and durability of one-trial avoidance learning. Animal Behaviour, 35(3),
739e747. https://doi.org/10.1016/S0003-3472(87)80110-0.
Rothschild, M. (1961). Defensive odours and Müllerian mimicry among insects.
Transactions of the Royal Entomological Society of London, 113(5), 101e123.
https://doi.org/10.1111/j.1365-2311.1961.tb00803.x.
Rowe, C., & Guilford, T. (1996). Hidden colour aversions in domestic clicks triggered
by pyrazine odours of insect warning displays. Nature, 383(6600), 520e522.
https://doi.org/10.1038/383520a0.
K. R
onk
a et al. / Animal Behaviour 135 (2018) 153e163162
Rowe, C., & Halpin, C. (2013). Why are warning displays multimodal? Behavioral
Ecology and Sociobiology, 67(9), 1425e1439. https://doi.org/10.1007/s00265-
013-1515-8.
Rowe, C., Lindstr
om, L., & Lyytinen, A. (2004). The importance of pattern similarity
between Mullerian mimics in predator avoidance learning. Proceedings of the
Royal Society B: Biological Sciences, 271(1537), 407e413. https://doi.org/10.1098/
rspb.2003.2615.
Rowland, H. M., Ihalainen, E., Lindstr
om, L., Mappes, J., & Speed, M. P. (2007). Co-
mimics have a mutualistic relationship despite unequal defences. Nature, 448,
64e67.
RStudio. (2015). RStudio: Integrated development environment for R (Version
0.99.441). Available from: http://www.rstudio.org/.
Ruxton, G. D., Franks, D. W., Balogh, A. C. V., & Leimar, O. (2008). Evolutionary
implications of the form of predator generalization for aposematic signals and
mimicry in prey. Evolution, 62(11), 2913e2921. https://doi.org/10.1111/j.1558-
5646.2008.00485.x.
Ruxton, G. D., Sherratt, T. N., & Speed, M. P. (2004). Avoiding attack: The evolutionary
ecology of crypsis, warning signals and mimicry. Oxford, U.K.: Oxford University
Press.
Sherratt, T. N. (2002). The evolution of imperfect mimicry. Behavioral Ecology,13(6),
821e826. https://doi.org/10.1093/beheco/13.6.821.
Sherratt, T. N., & Beatty, C. D. (2003). The evolution of warning signals as reliable
indicators of prey defense. American Naturalist, 162(4), 377e389.
Sherratt, T. N., & Macdougall, A. D. (1995). Some population consequences of vari-
ation in preference among individual predators. Biological Journal of the Linnean
Society, 55(2), 93e107. https://doi.org/10.1111/j.1095-8312.1995.tb01053.x.
Siddall, E. C., & Marples, N. M. (2008). Better to be bimodal: The interaction of color
and odor on learning and memory. Behavioral Ecology, 19(2), 425e432. https://
doi.org/10.1093/beheco/arm155.
Sill
en-Tullberg, B. (1985). The signicance of coloration per se, independent of
background, for predator avoidance of aposematic prey. Animal Behaviour,
33(4), 1382e1384. https://doi.org/10.1016/S0003-3472(85)80211-6.
Skelhorn, J., Halpin, C. G., & Rowe, C. (2016). Learning about aposematic prey.
Behavioral Ecology, 27(4), 955e964. https://doi.org/10.1093/beheco/arw009.
Skelhorn, J., & Rowe, C. (2007). Predators' toxin burdens inuence their strategic
decisions to eat toxic prey. Current Biology, 17(17), 1479e1483. https://doi.org/
10.1016/j.cub.2007.07.064.
Speed, M. P. (2000). Warning signals, receiver psychology and predator memory.
Animal Behaviour, 60, 269e278. https://doi.org/10.1006/anbe.2000.1430.
Sv
adov
a, K., Exnerov
a, A., Stys, P., Landov
a, E., Valenta, J., Fucíkov
a, A., et al. (2009).
Role of different colours of aposematic insects in learning, memory and
generalization of naive bird predators. Animal Behaviour, 77(2), 327e336.
https://doi.org/10.1016/j.anbehav.2008.09.034.
Terhune, E. C. (1977). Components of a visual stimulus used by scrub jays to
discriminate a Batesian model. American Naturalist, 111(979), 435e451. https://
doi.org/10.1086/283178.
Therneau, T. M. (2015). coxme: Mixed effects cox models (Version 2.2-5).https://
CRAN.R-project.org/package¼coxme.
Trimmer, P. C., Houston, A. I., Marshall, J. A. R., Mendl, M. T., Paul, E. S., &
McNamara, J. M. (2011). Decision-making under uncertainty: Biases and
Bayesians. Animal Cognition, 14(4), 465e476. https://doi.org/10.1007/s10071-
011-0387-4.
Turini, A., Veselý, P., & Fuchs, R. (2016). Five species of passerine bird differ in their
ability to detect Batesian mimics. Biological Journal of the Linnean Society, 117(4),
832e841. https://doi.org/10.1111/bij.12708.
Veselý, P., Ernestov
a, B., Nedv
ed, O., & Fuchs, R. (2017). Do predator energy demands
or previous exposure inuence protection by aposematic coloration of prey?
Current Zoology, 63(3), 259e267. https://doi.org/10.1093/cz/zow057.
Veselý, P., & Fuchs, R. (2009). Newly emerged Batesian mimicry protects only un-
familiar prey. Evolutionary Ecology, 23(6), 919. https://doi.org/10.1007/s10682-
008-9281-1.
Veselý, P., Luhanov
a, D., Pr
a
skov
a, M., & Fuchs, R. (2013). Generalization of mimics
imperfect in colour patterns: The point of view of wild avian predators.
Ethology, 119(2), 138e145. https://doi.org/10.1111/eth.12045.
Veselý, P., Vesel
a, S., Fuchs, R., & Zrzavý, J. (2006). Are gregarious red-black
shieldbugs, Graphosoma lineatum (Hemiptera: Pentatomidae), really apose-
matic? An experimental approach. Evolutionary Ecology Research, 8(5),
881e890.
Waldron, S. J., Endler, J. A., Valkonen, J. K., Honma, A., Dobler, S., & Mappes, J. (2017).
Experimental evidence suggests that specular reectance and glossy appear-
ance help amplify warning signals. Scientic Reports, 7,257.https://doi.org/
10.1038/s41598-017-00217-5.
Webster, S. J., & Lefebvre, L. (2000). Neophobia by the Lesser-Antillean Bullnch, a
foraging generalist, and the Bananaquit, a nectar specialist. Wilson Bulletin,
112 (3), 424e427. https://doi.org/10.1676/0043-5643(2000)112[0424:NBTLAB]
2.0.CO;2.
Appendix
Table A1
Test for asymmetric generalization in the rst generalization trial
Model df LRT Pr(Chi) Model AIC
(Intercept)þcolour combination 5 10.70 0.058 37.4
(Intercept) 38.1
Random effects Variance SD
Bird ID 4808 69.34
Fixed effects Estimate SE ZP
(Intercept: colour combination: yw) 12.48 3.32 3.75 <0.001
Colour combination: yr 14.85 5.22 2.85 0.0044
Colour combination: rw 2.27 8.04 0.28 0.78
Colour combination: ry 2.27 8.19 0.28 0.78
Colour combination: wr 1.38 6.16 0.22 0.82
Colour combination: wy 1.38 6.15 0.22 0.82
LRT: likelihood ratio test; y: yellow; w: white; r: red. Model selection was based on
model t, i.e. the model chosen was the one with the lowest Akaike information
criterion (AIC) value. The signicance level of
c
2
(Chi) indicates a change from the
model with colour combination as an explanatory variable to the model below, with
intercept only. Estimates of the best-tting model (in italics) are shown below. Of
the colour combinations, the combination of yellow learned and white offered had
least attacks, and was thus set to the intercept.
Table A2
Test for asymmetric generalization in all three generalization trials
Model df LRT Pr(Chi) Model AIC
(Intercept)þcolour combination 5 4.32 0.50 161.9
(Intercept) 156.2
Random effects Variance SD
Trial: Bird ID 167.82 12.96
Bird ID 58.01 7.62
Fixed effects Estimate SE ZP
(Intercept) 10.05 1.57 6.42 <0.001
LRT: likelihood ratio test. Model selection was based on model t, i.e. the model
chosen was the one with the lowest Akaike information criterion (AIC) value. The
signicance level of
c
2
(Chi) indicates a change from the model with colour com-
bination as an explanatory variable to the model below, with intercept only. Esti-
mates of the best-tting model (in italics) are shown below.
Table A3
Comparisons of colours attacked rst in the generalization trials
Colour
learned
Colour
tested
Attacked rst df
c
2
P
First trial White Yellow 6/14 1 0 1
Red 6/14
Yellow White 6/15 1 0 1
Red 7/15
Red White 8/15 1 0.13 0.71
Yellow 6/15
Three trials
pooled
White Yellow 12/42 1 5.89 0.02
Red 24/42
Yellow White 20/45 1 0.75 0.39
Red 15/45
Red White 22/45 1 0.41 0.52
Yellow 18/45
Chi-square comparisons of how many times each tested colour morph was attacked
rst in the generalization trials, by the colour morph learned. Models that were not
attacked or attacked second are included in the total number of models offered. Bold
indicates signicant difference.
K. R
onk
a et al. / Animal Behaviour 135 (2018) 153e163 163
... We focused on males, which are polymorphic regarding their hind-wing coloration: their hind wings may be yellow (chroma-rich) or white (luminance-rich). We chose Blue tits (Cyanistes caeruleus) as predators because their visual physiology is well-understood (Hart et al. 2000), they occur in the same areas as wood tiger moths, and are known to separate wood tiger moths' white and yellow color morphs (Rönkä et al. 2018b). It has also been shown that bird communities that inhabit different light environments select for different wood tiger moth warning signals (yellow and white) in nature (Nokelainen et al. 2014). ...
... In low light intensities luminance contrast should be more detectable (and salient) than chromatic contrast (Arenas et al. 2014;Cronin et al. 2014;Kelber 2019). Thus, provided that birds have completed avoidance learning on conspicuous, defended prey (Rönkä et al. 2018b), we expected shady conditions to hamper yellow warning signal efficacy (i.e., to get more attacks); in higher-light intensities, on the other hand, chromatic warning signals should work better (i.e., and to get less attacks). Finally, we expected the opposite pattern (i.e., yellow signals being attacked more often in shady conditions, and the opposite in high light intensity) if attacks are more dependent on sheer prey detectability. ...
... Cummings 2012; Arenas et al. 2014;Rojas et al. 2018). Instead, in variable light environments, signals being categorized as unprofitable may be enough either due to generalization for the common signal or simply by possessing conspicuous coloration (Ham et al. 2006;Rönkä et al. 2018b). ...
Article
Full-text available
A big question in behavioral ecology is what drives diversity of color signals. One possible explanation is that environmental conditions, such as light environment, may alter visual signaling of prey, which could affect predator decision-making. Here, we tested the context-dependent predator selection on prey coloration. In the first experiment, we tested detectability of artificial visual stimuli to blue tits (Cyanistes caeruleus) by manipulating stimulus luminance and chromatic context of the background. We expected the presence of the chromatic context to facilitate faster target detection. As expected, blue tits found targets on chromatic yellow background faster than on achromatic grey background whereas in the latter, targets were found with smaller contrast differences to the background. In the second experiment, we tested the effect of two light environments on the survival of aposematic, color polymorphic wood tiger moth (Arctia plantaginis). As luminance contrast should be more detectable than chromatic contrast in low light intensities, we expected birds, if they find the moths aversive, to avoid the white morph which is more conspicuous than the yellow morph in low light (and vice versa in bright light). Alternatively, birds may attack first moths that are more detectable. We found birds to attack yellow moths first in low light conditions, whereas white moths were attacked first more frequently in bright light conditions. Our results show that light environments affect predator foraging decisions, which may facilitate context-dependent selection on visual signals and diversity of prey phenotypes in the wild.
... In sites (Moyobamba, Ahuashiyacu) in which the morphs present are less colourful, non-local morphs were attacked at a much higher rate both when similarly colourful and when more colourful, showing again the absence of peak-shift in a different set of morphs. These observations are consistent with the lack of generalization in predators of a polymorphic moth (Rönkä et al. 2018). Kin selection is ruled out as adult butterflies disperse. ...
Article
Full-text available
Butterflies are frequently conspicuous. The function of this conspicuousness is understudied and may vary between species or family. Aposematism has frequently been proposed, as well as sexual signalling; even allowing for other functions, gaps seem to remain. Butterflies are also frequently social and many species will aggregate in large clusters. Here I propose that striking colourations may have evolved in some species to improve visibility to conspecifics, even if it also increases visibility to predators and even if colouration provides no other benefit, merely because the improved visibility increases the probability of being part of a cluster which provides protection against predation through dilution. As well as showing the potential existence of a new mechanism which can lead to bright colour occurring in flying insects, the proposed mechanism may provide an explanation to the mimicry rings of Heliconius butterflies that is superior to Müllerian mimicry. Several features of Heliconius rings seem to contradict the logic and premises of Müllerian mimicry, but would be predicted by, or consistent with, the hypothesis that the marks on Heliconius wings are signals towards co-roosters.
... Moreover, recent studies in other species have shown that UV may facilitate separation of incipient species as recently demonstrated in Colias butterflies(Ficarrotta et al., 2022) and that the differences in UV reflection may arise from novel duplication of the gene producing sex-specific differences in reflectance as in Zerene cesonia butterfly(Rodriguez-Caro et al., 2021). Previous experiments have shown that birds learn to avoid red wood tiger moths more effectively than yellow or white ones(Ham et al., 2006;Lindstedt et al., 2011;Rönkä et al., 2018), but the selection F I G U R E 5 Representation of the wood tiger moth genotypes to illustrate how they may appear to ecologically relevant receivers. These false image examples show genotypes of both sexes organized in vertical columns and human, avian and moth false colour images in horizontal groups. ...
Article
Full-text available
The definition of colour polymorphism is intuitive: genetic variants express discretely coloured phenotypes. This classification is, however, elusive as humans form subjective categories or ignore differences that cannot be seen by human eyes. We demonstrate an example of a 'cryptic morph' in a polymorphic wood tiger moth (Arctia plantaginis), a phenomenon that may be common among well-studied species. We used pedigree data from nearly 20,000 individuals to infer the inheritance of hindwing colouration. The evidence supports a single Mendelian locus with two alleles in males: WW and Wy produce the white and yy the yellow hindwing colour. The inheritance could not be resolved in females as their hindwing colour varies continuously with no clear link with male genotypes. Next, we investigated if the male genotype can be predicted from their phenotype by machine learning algorithms and by human observers. Linear discriminant analysis grouped male genotypes with 97% accuracy, whereas humans could only group the yy genotype. Using vision modelling, we also tested whether the genotypes have differential discriminability to humans, moth conspecifics and their bird predators. The human perception was poor separating the genotypes, but avian and moth vision models with ultraviolet sensitivity could separate white WW and Wy males. We emphasize the importance of objective methodology when studying colour polymorphism. Our findings indicate that by-eye categorization methods may be problematic, because humans fail to see differences that can be visible for relevant receivers. Ultimately, receivers equipped with different perception than ours may impose selection to morphs hidden from human sight.
... Moreover, recent studies in other species have shown that UV may facilitate separation of incipient species as recently demonstrated in Colias butterflies(Ficarrotta et al., 2022) and that the differences in UV reflection may arise from novel duplication of the gene producing sex-specific differences in reflectance as in Zerene cesonia butterfly(Rodriguez-Caro et al., 2021). Previous experiments have shown that birds learn to avoid red wood tiger moths more effectively than yellow or white ones(Ham et al., 2006;Lindstedt et al., 2011;Rönkä et al., 2018), but the selection F I G U R E 5 Representation of the wood tiger moth genotypes to illustrate how they may appear to ecologically relevant receivers. These false image examples show genotypes of both sexes organized in vertical columns and human, avian and moth false colour images in horizontal groups. ...
Article
Full-text available
Phenotypic variation is suggested to facilitate the persistence of environmentally growing pathogens under environmental change. Here we hypothesized that the intensive farming environment induces higher phenotypic variation in microbial pathogens than natural environment, because of high stochasticity for growth and stronger survival selection compared to the natural environment. We tested the hypothesis with an opportunistic fish pathogen Flavobacterium columnare isolated either from fish farms or from natural waters. We measured growth parameters of two morphotypes from all isolates in different resource concentrations and two temperatures relevant for the occurrence of disease epidemics at farms and tested their virulence using a zebrafish (Danio rerio) infection model. According to our hypothesis, isolates originating from the fish farms had higher phenotypic variation in growth between the morphotypes than the isolates from natural waters. The difference was more pronounced in higher resource concentrations and the higher temperature, suggesting that phenotypic variation is driven by the exploitation of increased outside‐host resources at farms. Phenotypic variation of virulence was not observed based on isolate origin but only based on morphotype. However, when in contact with the larger fish, the less virulent morphotype of some of the isolates also had high virulence. As the less virulent morphotype also had higher growth rate in outside‐host resources, the results suggest that both morphotypes can contribute to F. columnare epidemics at fish farms, especially with current prospects of warming temperatures. Our results suggest that higher phenotypic variation per se does not lead to higher virulence, but that environmental conditions at fish farms could select isolates with high phenotypic variation in bacterial population and hence affect evolution in F. columnare at fish farms. Our results highlight the multifaceted effects of human‐induced environmental alterations in shaping epidemiology and evolution in microbial pathogens.
... First, under some circumstances, novel color alleles may evolve neutrally or under weak selection from predators. For example, when predators focus only on a specific element or combination of elements in an aposematic signal, variation in other signal elements may have little to no impact on predation rates (Winters et al. 2017;Rönkä et al. 2018a). There can also be variation in predator cognition and foraging behavior due to neophobia or dietary conservatism, which can result in weaker selection against novel signal forms by predators (e.g., Thomas et al. 2004;Aubier and Sherratt 2015), especially if the prey is aggregated (Mappes and Alatalo 1997;Mappes et al. 1999). ...
Article
Full-text available
Our understanding of how novel warning color traits evolve in natural populations is largely based on studies of reproductive stages and organisms with endogenously produced pigmentation. In these systems, genetic drift is often required for novel alleles to overcome strong purifying selection stemming from frequency‐dependent predation and positive assortative mating. Here, we integrate data from field surveys, predation experiments, population genomics, and phenotypic correlations to explain the origin and maintenance of geographic variation in a diet‐based larval pigmentation trait in the redheaded pine sawfly (Neodiprion lecontei), a pine‐feeding hymenopteran. Although our experiments confirm that N. lecontei larvae are indeed aposematic—and therefore likely to experience frequency‐dependent predation—our genomic data do not support a historical demographic scenario that would have facilitated the spread of an initially deleterious allele via drift. Additionally, significantly elevated differentiation at a known color locus suggests that geographic variation in larval color is currently maintained by selection. Together, these data suggest that the novel white morph likely spread via selection. However, white body color does not enhance aposematic displays, nor is it correlated with enhanced chemical defense or immune function. Instead, the derived white‐bodied morph is disproportionately abundant on a pine species with a reduced carotenoid content relative to other pine hosts, suggesting that bottom‐up selection via host plants may have driven divergence among populations. Overall, our results suggest that life stage and pigment source can have a substantial impact the evolution of novel warning signals, highlighting the need to investigate diverse aposematic taxa to develop a comprehensive understanding of color variation in nature. This article is protected by copyright. All rights reserved
... Generalization is thought to be a key mechanism sustaining Batesian mimicry, because when mimics resemble models well enough, receivers are expected to generalize their learned responses to models, to mimics (Ham et al., 2006;Ruxton et al., 2008;Speed and Ruxton, 2010;Rönkä et al., 2018). Signal detection theory predicts that receivers perceive signals as more similar when the signal distribution is broader, which can be a result of phenotypic variation among models and mimics (Figure 1; Lynn et al., 2005). ...
Article
Full-text available
Mutualisms involve cooperation, but also frequently involve conflict. Plant-pollinator mutualisms are no exception. To facilitate animal pollination, flowering plants often offer pollen (their male gametes) as a food reward. Since plants benefit by maximizing pollen export to conspecific flowers, we might expect plants to cheat on pollen rewards. In intersexual floral mimicry, rewarding pollen-bearing male flowers (models) are mimicked by rewardless female flowers (mimics) on the same plant. Pollinators should therefore learn to avoid the unrewarding mimics. Plants might impede such learning by producing phenotypically variable flowers that cause bees to generalize among models and mimics during learning. In this laboratory study, we used partially artificial flowers (artificial petals, live reproductive parts) modeled after Begonia odorata to test whether variation in the size of rewarding male flowers (models) and unrewarding female flowers (mimics) affected how quickly bees learned both to recognize models and to reject mimics. Live unrewarding female flowers have 33% longer petals and have 31% greater surface area than live rewarding male flowers, which bees should easily discriminate. Yet while bees rapidly learned to reduce foraging effort on mimics, learning was not significantly affected by the degree to which flower size varied. Additionally, we found scant evidence that this was a result of bees altering response speed to maintain decision accuracy. Our study failed to provide evidence that flower size variation in intersexual floral mimicry systems exploits pollinator cognition, though we cannot rule out that other floral traits that are variable may be important. Furthermore, we propose that contrary to expectation, phenotypic variability in a Batesian mimicry system may not necessarily have significant effects on whether receivers effectively learn to discriminate models and mimics.
... The color polymorphism is under selection by bird predators in the wild (Rönkä et al., 2020). In predation experiments, birds respond differently toward the hindwing morphs, avoiding either yellow (Nokelainen et al., 2012(Nokelainen et al., , 2014 or white , but see Rönkä et al. (2018). Rojas et al. (2019) speculate that the variable response by predators could be due to differences in cues between the moths and their model stimuli, differences in light environment between experiments (Nokelainen in prep b), or the presence or absence of methoxypyrazine odor. ...
Article
Full-text available
Aposematic organisms warn predators of their unprofitability using a combination of defenses, including visual warning signals, startling sounds, noxious odors, or aversive tastes. Using multiple lines of defense can help prey avoid predators by stimulating multiple senses and/or by acting at different stages of predation. We tested the efficacy of three lines of defense (color, smell, taste) during the predation sequence of aposematic wood tiger moths ( Arctia plantaginis ) using blue tit ( Cyanistes caeruleus ) predators. Moths with two hindwing phenotypes (genotypes: WW/Wy = white, yy = yellow) were manipulated to have defense fluid with aversive smell (methoxypyrazines), body tissues with aversive taste (pyrrolizidine alkaloids) or both. In early predation stages, moth color and smell had additive effects on bird approach latency and dropping the prey, with the strongest effect for moths of the white morph with defense fluids. Pyrrolizidine alkaloid sequestration was detrimental in early attack stages, suggesting a trade-off between pyrrolizidine alkaloid sequestration and investment in other defenses. In addition, pyrrolizidine alkaloid taste alone did not deter bird predators. Birds could only effectively discriminate toxic moths from non-toxic moths when neck fluids containing methoxypyrazines were present, at which point they abandoned attack at the consumption stage. As a result, moths of the white morph with an aversive methoxypyrazine smell and moths in the treatment with both chemical defenses had the greatest chance of survival. We suggest that methoxypyrazines act as context setting signals for warning colors and as attention alerting or “go-slow” signals for distasteful toxins, thereby mediating the relationship between warning signal and toxicity. Furthermore, we found that moths that were heterozygous for hindwing coloration had more effective defense fluids compared to other genotypes in terms of delaying approach and reducing the latency to drop the moth, suggesting a genetic link between coloration and defense that could help to explain the color polymorphism. Conclusively, these results indicate that color, smell, and taste constitute a multimodal warning signal that impedes predator attack and improves prey survival. This work highlights the importance of understanding the separate roles of color, smell and taste through the predation sequence and also within-species variation in chemical defenses.
... However, yellow models were attacked by juvenile birds at significantly lower rates compared to red models. Some studies showed that, among aposematic colours, yellow was more effective for avoidance learning than red (Lawrence and Noonan 2018), while in other studies, birds learned to avoid the red moth models considerably faster than the yellow models (Rönkä et al. 2018). We suggest that in our study, this difference between attacks on yellow and red models may be explained by the innate nature of aversive responses to yellow prey, as indicated by some experiments (Lindström et al. 1999;Hauglund et al. 2006). ...
Article
Full-text available
The direction and strength of selection for prey colouration by predators vary in space and time and depend on the composition of the predator community. We tested the hypothesis that bird selection pressure on prey colouration changes through the season due to changes in the proportion of naïve juvenile individuals in the bird community, because naïve and educated birds differ in their responses to prey colours. Bird predation on caterpillar-shaped plasticine models in two boreal forest sites increased sevenfold from early summer to mid-summer, and the time of this increase coincides with the fledging of juvenile birds. In early summer, cryptic (black and green) models were attacked at fivefold higher rates compared with conspicuous (red and yellow) models. By contrast, starting from fledging time, cryptic and conspicuous models were attacked at similar rates, hinting at a lower selectivity by naïve juvenile birds compared with educated adult birds. Cryptic models exposed in a group together with conspicuous models were attacked by birds at a threefold lower rate than cryptic models exposed singly, thus supporting the aposematic commensalism hypothesis. However, this effect was not observed in mid- and late summer, presumably due to the lack of avoidance of conspicuous prey by the juvenile birds. We conclude that selection pressure on prey colouration weakens considerably when naïve birds dominate in the community, because the survival advantages of aposematic colouration are temporarily lost for both the conspicuous and their neighbouring cryptic prey.
Article
The study of both animal colouration and patterning across the natural world has been imperative for understanding some of the key principles of biology throughout the past century, particularly with respect to evolution and genetics. Generally, colour and pattern have each been described qualitatively, often being binned into discrete groups relying on human perception of colour or pattern, rather than being considered in a biologically relevant context. ‘Binning’ traits into discrete groups has the consequence that variation within discrete morphs is often overlooked. Terrestrial gastropods, such as the selected study genera Cepaea and Auriculella, provide an ideal system for the study of polymorphisms and colour variation due to the extreme variety of morphs present across the taxa, as well as the nature of shell growth providing a complete ontogeny of an individual. The aims of this thesis are threefold; firstly, I aimed to understand finescale variation within and between established banding pattern morphs in Cepaea, to allow inferences to be made regarding the genetic mechanisms responsible for this variation. The implementation of two quantitative methods for measuring variation in band position and width in individual shells found that individual band absence has a minor but significant effect on the position of the remaining bands, implying that the locus controlling band presence/absence acts mainly after the position of bands is determined. I establish a method which is useful for comparative studies of quantitative banding variation in snail shells, and for extraction of growth parameters and morphometrics, highlighting the importance and usefulness of gastropod shells in the understanding of how variation is established and maintained in a population. Secondly, I aimed to understand the shell colour variation present in both Cepaea nemoralis and Cepaea hortensis by using spectrophotometry and psychophysical modelling in tandem. It was revealed that colour variation in Cepaea hortensis is continuous, with no detectable effects of geographic location with the exception of an association of the paleness of yellow shells with latitude. Differences between the colour of Cepaea hortensis and Cepaea nemoralis, both in terms of exact shade and overall colour were revealed; Cepaea hortensis are generally paler, and less pink-toned, but slightly more brown-toned. Precise shade variation of yellow individuals from genetically diverse lineages of Cepaea nemoralis were also detected. The results presented have significance in furthering the understanding of the precise nature of the colour polymorphism displayed in Cepaea spp., and the nature of the selection which acts upon it, as well as highlighting the importance of considering colour as a continuous trait, rather than binning it into discrete groups. Thirdly, I aimed to investigate colour variation across a number of scales in the Hawaiian land snail genus Auriculella, to allow inferences about the genetic architecture responsible for the variation, and to highlight the usefulness of museum collections of gastropod shells in understanding variation in extinct or endangered species. I demonstrated that there are differences in colour within single shells of Auriculella, similar to variation displayed by other Pacific Island snails. I described significant variation between isolated populations of the same species, and determined that there is no difference in colour variation between shells on the islands of Maui and Oahu. Finally, I demonstrated that there is no difference in colour between shell chiralities, suggesting that interchiral mating is not uncommon, and that the loci responsible for colour variation and chirality are not closely linked. By describing the variation present in Auriculella in a biologically relevant context, inference regarding genetic mechanisms of variation becomes possible in a taxa of conservation concern. By achieving these aims, and synthesising conclusions drawn from their achievement, I have highlighted the importance of accurately defining phenotypes for the purposes of evolutionary ecology and genetics. Defining phenotypes and investigating variation present within morphs has allowed inferences to be made regarding the underpinning genetic mechanisms which control variation in two gastropod genera, although the principles are applicable to other taxonomic groups. Finally, and more broadly, I have demonstrated the usefulness of gastropods as study systems, particularly where large collections of shells are available.
Preprint
Chemical defences often vary within and between populations both in quantity and quality, which is puzzling if prey survival is dependent on the strength of the defence. We investigated the within- and between-population variability in chemical defence of the wood tiger moth (Arctia plantaginis). The major components of its defences, SBMP (2secbutyl3methoxypyrazine) and IBMP (2isobutyl3methoxypyrazine) are volatiles that deter bird attacks. We expected the variation to reflect populations predation pressures and early-life conditions. To understand the role of the methoxypyrazines, we experimentally manipulated synthetic SBMP and IBMP and tested the birds reactions. We found a considerable variation in methoxypyrazine amounts and composition, both from wild-caught and laboratory-raised male moths. In agreement with the cost of defence hypothesis, the moths raised in the laboratory had a higher amount of pyrazines. We found that SBMP is more effective at higher concentrations and that IBMP is more effective only in combination with SBMP and at lower concentrations. Our results fit findings from the wild: the amount of SBMP was higher in the populations with higher predation pressure. Altogether, this suggests that, regarding pyrazine concentration, more is not always better, and highlights the importance of testing the efficacy of chemical defence and its components with relevant predators, rather than relying only on results from chemical analyses
Article
Full-text available
Many animals protect themselves from predation with chemicals, both self-made or sequestered from their diet. The potential drivers of the diversity of these chemicals have been long studied, but our knowledge of these chemicals and their acquisition mode is heavily based on specialist herbivores that sequester their defenses. The wood tiger moth (Arctia plantaginis, Linnaeus, 1758) is a well-studied aposematic species, but the nature of its chemical defenses has not been fully described. Here, we report the presence of two methoxypyrazines, 2-sec-butyl-3-methoxypyrazine and 2-isobutyl-3-methoxypyrazine, in the moths' defensive secretions. By raising larvae on an artificial diet, we confirm, for the first time, that their defensive compounds are produced de novo rather than sequestered from their diet. Pyrazines are known for their defensive function in invertebrates due to their distinctive odor, inducing aversion and facilitating predator learning. While their synthesis has been suspected, it has never previously been experimentally confirmed. Our results highlight the importance of considering de novo synthesis, in addition to sequestration, when studying the defensive capabilities of insects and other invertebrates.
Article
Full-text available
Although predation is commonly thought to exert the strongest selective pressure on colouration in aposematic species, sexual selection may also influence colouration. Specifically, polymorphism in aposematic species cannot be explained by natural selection alone. 2.Males of the aposematic wood tiger moth (Arctia plantaginis) are polymorphic for hindwing colouration throughout most of their range. In Scandinavia, they display either white or yellow hindwings. Female hindwing colouration varies continuously from bright orange to red. Redder females and yellow males suffer least from bird predation. 3.White males often have higher mating success than yellow males. Therefore, we ask whether females can discriminate the two male morphs by colour. Males approach females by following pheromone plumes from a distance, but search visually at short range. This raises the questions whether males discriminate female colouration and, in turn, whether female colouration is also sexually selected. 4.Using electroretinograms, we found significantly larger retinal responses in male than female A. plantaginis, but similar spectral sensitivities in both sexes, with peaks in the UV (349 nm), blue (457 nm), and green (521 nm) wavelength range. 5.According to colour vision models, conspecifics can discriminate white and yellow males as separate morphs, but not orange and red females. For moths and birds (Cyanistes caeruleus), white males are more conspicuous against green and brown backgrounds, mostly due to UV reflectivity, and red females are slightly more conspicuous than orange females. 6.The costly red colouration among females is likely selected by predator pressure, not by conspecifics, whereas male colour polymorphism is probably maintained, at least partly, by a the opposing forces of predation pressure favouring yellow males, and female preference for white males. Whether or not the preference for white males is based on visual cues requires further testing. 7.The evolution of polymorphic aposematic animals can be better understood when the visual system of the species and their predators is taken into consideration. This article is protected by copyright. All rights reserved.
Article
Full-text available
Animals have evolved different defensive strategies to survive predation, among which chemical defences are particularly widespread and diverse. Here we investigate the function of chemical defence diversity, hypothesizing that such diversity has evolved as a response to multiple enemies. The aposematic wood tiger moth (Arctia plantaginis) displays conspicuous hindwing coloration and secretes distinct defensive fluids from its thoracic glands and abdomen.We presented the two defensive fluids from laboratoryreared moths to two biologically relevant predators, birds and ants, and measured their reaction in controlled bioassays (no information on colour was provided). We found that defensive fluids are target-specific: thoracic fluids, and particularly 2-sec-butyl-3-methoxypyrazine, which they contain, deterred birds, but caused no aversive response in ants. By contrast, abdominal fluids were particularly deterrent to ants, while birds did not find them repellent. Our study, to our knowledge, is the first to show evidence of a single species producing separate chemical defences targeted to different predator types, highlighting the importance of taking into account complex predator communities in studies on the evolution of prey defence diversity.