Behav Ecol Sociobiol DOI 10.1007/s00265-013-1515-8

REVIEW

Why are warning displays multimodal? Candy Rowe & Christina Halpin

Received: 29 August 2012 / Revised: 28 January 2013 / Accepted: 1 March 2013 # Springer-Verlag Berlin Heidelberg 2013

Abstract Multimodal defensive displays are commonplace, with prey combining conspicuous coloration, sounds, odours and other chemical emissions to deter predators. These components can signal to predators in multiple signal modalities to warn them that prey are defended. The aim of our review is to examine the form and function of multimodal warning displays. Data collected from the literature on multimodal insect warning displays show the degree of complexity and diversity that needs to be explained, and we identify patterns in the data that may be worthy of more rigorous investigation. We also provide a theoretical framework for the study of multimodal warning displays, and evaluate the evidence for different functional hypotheses that can explain their widespread evolution. Our review highlights that whilst multimodal warning displays are well documented, particularly in insects, we lack a good understanding of their function in natural predator–prey systems. Keywords Aposematism . Mimicry . Multimodal warning display . Multicomponent signal . Receiver psychology . Predator . Prey

Introduction It is nearly 150 years since Darwin’s curiosity was aroused by the occurrence of conspicuous coloration in insect larvae, the Communicated by E. A. Hebets This manuscript is part of the special issue Multimodal Communication—Guest Editors: James P. Higham and Eileen A. Hebets C. Rowe (*) : C. Halpin Centre for Behaviour and Evolution, Institute of Neuroscience, Newcastle University, Newcastle upon Tyne, NE2 4HH, UK e-mail: [email protected]

evolution of which he could not attribute to sexual selection. In a letter to Wallace in 1867, Darwin asked: “My difficulty is, why are caterpillars sometimes so beautifully and artistically coloured?” (Darwin 1887). Wallace replied that the larvae were brightly coloured to warn predators of chemical defences and to deter them from initiating an attack (Wallace 1867). This defensive strategy was later called ‘aposematism’ (Poulton 1890). Aposematism is a defence strategy found in many taxa (see Cott (1940) and Edmunds (1974) for excellent reviews). Aposematic prey typically employ yellow, orange and red in their signals, often combining these warning colours with black to produce a high-contrast pattern (for example, the yellow-and-black pattern of the common wasp, Vespa vulgaris, and the red-and-black pattern of the seven-spot ladybird, Coccinella septempunctata). Whilst being conspicuous can increase the risk of being detected by some visually hunting predators (e.g. Gittleman and Harvey 1980; Järvi et al. 1981; Alatalo and Mappes 1996), it can also be beneficial by enhancing predator avoidance. Predators can show unlearned avoidance behaviour towards prey with colours and patterns typically associated with aposematism (e.g. Smith 1975; Schuler and Hesse 1985; Roper and Cook 1989), and can also learn to avoid toxic prey much more quickly if they have a conspicuous signal compared to a cryptic one (Gittleman and Harvey 1980; Roper and Redston 1987). The benefits to possessing conspicuous warning coloration have been well studied, and there is abundant evidence that warning signals can enhance prey survival (see Ruxton et al. (2004) for further in-depth review and discussion). However, aposematic prey are not only brightly coloured, but they often also produce sounds, odours and distasteful secretions (or any combination of these) upon attack (e.g. Carpenter and Ford 1933; Cott 1940; Dunning 1968; Alexander 1964; Rothschild 1965; Claridge 1974; Edmunds 1974; Rothschild et al. 1984; Ruxton et al. 2004). We will

Behav Ecol Sociobiol

refer to these complex displays as ‘multimodal defensive displays’ since whilst the additional components have clearly been selected to increase the survival chances of the prey, we don’t know whether they are performing a signalling role or are part of the defence itself. We restrict our definition of a ‘multimodal warning display’ to those cases where multiple components are known to be or could be playing a communicatory function. Because of this distinction between multimodal defensive displays and multimodal warning displays, we will first give a brief overview of the nature of multimodal defensive displays, and evaluate what role different components might play, particularly whether components may act as a deterrent rather than a signal. Our analysis is focussed on insects simply because there is a wealth of data from this particular taxon; however, we anticipate that our discussion can be applied to defensive displays more generally. The main part of our review will subsequently focus on evaluating the functional hypotheses that could explain the evolution of multimodal warning displays. Our aim is to provide a framework of testable hypotheses that can help advance research in this area, and to identify key areas for future research.

Multimodal defensive displays in insects: a brief review Since the nineteenth century, multimodal defensive displays have been well documented in insects (see Table 1). That is not to say that they do not occur in other taxa: there are good examples in other arthropods, e.g. the display of the harvestman, Holpobunus mexicanus, that combines aposematic coloration with stridulation and a secretion (Pomini et al. 2010) as well as in vertebrates, e.g. the striped skunk, Mephitis mephitis, combines a high-contrast striped pattern with teethchattering and a production of a foul-smelling spray (Cott 1940). However, by far the majority of observations and investigations have been made on insects, providing us with an excellent opportunity to get an overview of the variety of forms that multimodal defensive displays can take. We have collected together examples of multimodal defensive displays of insects that we have opportunistically collected from reviews, notes and empirical studies that were readily available to us (see Table 1). Sources included our previous work on the subject (Rowe 1998; Rowe and Guilford 2001), papers that we already possessed, and citation searches on some of the older literature. To be included in the table, a species needed to have a visual warning signal and at least one additional non-visual component in their display, or have at least two non-visual components if they were not obviously warningly coloured. We categorised the different components on the basis of which sensory modality they would primarily stimulate in predators. We focused on collecting examples from chemically defended prey;

however, during this search we did find some examples of multimodal defensive displays in species that are not known to contain toxins. We have also included these in the table since they are pertinent to our discussions about the role of mimicry in the evolution of multimodal warning displays. We have recorded the description of each component where possible (an ‘X’ in the table means that the presence of a component was reported but the exact form of that component was not), and have verified prey coloration against colour photographs published on the internet where possible. Our aim has not been to provide a comprehensive survey of the multimodal defensive displays found in insects, but to give an overview of the sensory modalities of predators that can be stimulated by the displays of different insect species. Therefore, the number of examples in each taxon is not necessarily a reflection of the frequency of multimodal displays in that group, and certainly any gaps in the data should not be interpreted as meaning that multimodal displays are not found in a particular taxon. For example, the high number of examples from Lepidoptera included in the table (47/99) is likely to reflect the enthusiasm of early naturalists to collect and study butterflies and moths, and should not be taken as evidence for a higher occurrence of multimodal displays in this particular taxon. Whilst we are cautious in suggesting that these data are an accurate representation of multimodal defensive displays in nature, we think that they can provide insight into the nature of these displays, and generate questions for future research. Patterns of complexity and diversity There are three take-home messages from the data presented in Table 1. First, it is clear that insect defensive displays can be incredibly complex and contain up to four components that can be detected by different sensory systems of predators (vision, audition, olfaction, gustation/nociception). One of the most complex displays is that of the lubber grasshopper (Romalea microptera), that has spectacular yellow and black coloration, but also produces hissing as well as a chemical secretion, containing phenols and quinones, and a repugnant odour upon attack (Roth and Eisner 1962; Rothschild and Haskell 1966; Hatle et al. 2002; Jones et al. 1989). This level of multimodal complexity is not restricted to adult insects: the larvae of Saturnia pyri are green and blue with black hairs, and not only produce a secretion that is foul smelling (to humans) upon attack but also an audible chirping sound (Bura et al. 2009). However, having four sensory components in a display is seemingly quite rare (only 8 % of the documented cases). This leads to a second point of note: that as the number of different components in a defensive display increases, the

Behav Ecol Sociobiol Table 1 Examples of multimodal defensive displays in insects showing how components can stimulate different sensory modalities in predators Species

Vision

Audition

Olfaction

Gustation/ nociception

Anisomorpha buprestoides Red/black

Penetrating stench

Fine mist

Cercopis vulnerata

Red/black

Pyrazine

Dictyophorus laticinata

Red/black

Hissing

Foul smelling

Fluid

Dictyophorus productus

Red/white/black

Hissing

Foul smelling

Fluid

Eurycnema goliatha

Red wing flash

Swishing

Mantis religosaa

Green wing flash

Hissing

Phymateus leprosus

Bright green/black

Rattle

Phymateus morbillosus Phymateus puniceus

Reference

Orthoptera Eisner 2003 Moore et al. 1990 Carpenter 1938 Carpenter 1938 Bedford and Chinnick 1966 Hill 2007 Foul smelling

Fluid

Carpenter 1938

Red/yellow/black

Evil smelling

Fluid

Carpenter 1938

Red wing flash

Evil smelling

Poekilocerus bufonius

Yellow/black

Pyrazine

Romalea microptera

Yellow/black

Hiss

Tropidoderus childreniia

Red wing flash

Swishing

Vestria ‘Crayola katydid’b

Carpenter 1938 Moore et al. 1990

Repugnant odour

Secretion

Roth and Eisner 1962; Rothschild and Haskell 1966; Hatle et al. 2002

Green/blue/yellow/ black

Pyrazine

Secretion

Nickle et al. 1996

Brachystethus cribrum

Red/black

Powerful odour

Cercopis vulnerata

Red/black

X

Coranus subapterus

Brown/grey/red

Graphosoma lineatum

Red/black

X

Nematopus indus

Red/black

Powerful odour

Pyrrhocoris apterus

Red/black

Rhodinus prolixus

Brown/grey

Stridulation

Pungent odour

Schilman et al. 2001

Triatoma infestans

Brown/grey

Stridulation

Pungent odour

Schilman et al. 2001

Bedford and Chinnick 1966

Hemiptera Cott 1940 Rothschild 1961

Stridulation

Haskell 1961

X

Spray

Johansen et al. unpubl. Cott 1940

Secretion

Exnerová et al. 2003

Coleoptera Adalia bipunctata

Red/black

Adelium pustulosum

Black

Bracon coccineus

Red/black

Pyrazine Stridulation

Reflex bleeding Moore et al. 1990; de Jong et al. 1991 Secretion

Eisner et al. 1974

Coccinella septempunctata Red/black

Smells like Coccinellids Pyrazine

Reflex bleeding Moore et al. 1990; Marples et al. 1994

Coccinella 11-punctata

Red/black

Pyrazine

Reflex bleeding Rothschild 1961

Coccinella transversalis

Red/black

Pyrazine

Elaphrus riparius

Brown

Stridulation

Rothschild 1961

Moore et al. 1990 Secretion

Schilman et al. 2001

Epilachna curcurbitae

Red/black

Pyrazine

Epilachna 26-punctata

Red/black

Pyrazine

Moore et al. 1990 Moore et al. 1990

Eumorphus tetraspilotus

Red/black

Pyrazine

Moore et al. 1990

Pyrazine

Moore et al. 1990

Harmonia conformis

Orange/black

Lilioceris lilii

Red/black

Micraspis frenata

Orange/black

Omophron labiatus

Brown/green

Pseudolycus haemopterus

Orange/black

Pyrazine

Pyrochroa serraticornis

Red/black

Pyrazine

Rothschild 1961

Rhagonycha fulva

Orange/brown

Pyrazine

Moore et al. 1990

Scaphinotus andrewsi

Black

Stridulation

Secretion/spray

Scaphinotus viduus

Black

Stridulation

Secretion/spray

Zonitis lutea

Orange/brown

Stridulation

Haskell 1961 Pyrazine

Stridulation

Moore et al. 1990 Secretion

Pyrazine

Masters 1979 Moore et al. 1990

Wheeler et al. 1970 Wheeler et al. 1970 Moore et al. 1990

Behav Ecol Sociobiol Table 1 (continued) Species

Vision

Audition

Olfaction

Gustation/ nociception

Reference

Eristalis arbustoruma

Yellow/black

Buzzing

Rashed et al. 2009

Eristalis flavipesa

Pale yellow/black

Buzzing

Rashed et al. 2009

Eristalis transversa

Yellow/black

Buzzing

Rashed et al. 2009

Helophilus fasciatusa

Yellow/black

Buzzing

Rashed et al. 2009

Spilomyia longicornisa

Yellow/black

Buzzing

Rashed et al. 2009

Spilomyia sayia

Yellow/black

Buzzing

Rashed et al. 2009

Acherontia atropus

Yellow/black

Clicking/hissing

Actias luna

Green/red spots

Clicking

Amorpha juglandisa

Whistling

Antheraea polyphemus

Thrashing brown display Green/white/red

Callosamia promethean

Green/red/yellow

Dryas iulia

Red/white/black

Diptera

a

Lepidoptera Larvae Cott 1940; Brown et al. 2007 Regurgitant

Brown et al. 2007 Bura et al. 2011

Clicking

Regurgitant Regurgitant Pyrazine

Brown et al. 2007 Brown et al. 2007 Moore et al. 1990

Papilio rumanzovia

Green/white

Pyrazine

Moore et al. 1990

Heliconius chartiona

White/black

Pyrazine

Moore et al. 1990

Heliconius melpomene

White/black

Manduca sexta

Green/white

Pyrazine Clicking

Saturnia pyri

Green/blue

Chirping

Smerinthus jamaicensis

Green/white/red

X

Foul smelling

Moore et al. 1990 Secretion

Bura et al. 2012

Secretion

Bura et al. 2009

Regurgitation

Brown et al. 2007

Yellow fluid

Rothschild 1961

Yellow fluid

Rothschild 1961

Yellow fluid

Rothschild 1961

Secretion

Bisset et al. 1960; Moore et al. 1990; Fenton and Roeder 1974

Adults Acraea horta

Black/orange/clear

Alaena amazoula

Black/yellow

Amauris echeria

Black/yellow

Arctia caja

Yellow/red/black

Arctia villica

Cream/black/orange

Pyrazine

Battus polydamus

Yellow/black

Pyrazine

Cissura plumbea

Back/red

X

Composia fidelissima

Black/blue/white/red

Chirping

Cyncia tenera

White/orange

Clicking

Danaus plexippus

Orange/black

Pyrazine

Moore et al. 1990

Delphyre rubricincta

Grey/orange

Foul odour

Blest 1964

Diospage chrysobasis

Red/black

Dyschema tiresias

Yellow/black

Groaning/chirping

Erasmia sanguiflua

Blue/white/black

Squeaking

X

White/red

Rothschild 1961 Moore et al. 1990 Blest 1964 Froth

Rothschild and Haskell 1966 Barber and Conner 2007

Froth

Euplagia quadripunctaria Yellow/red/black Eupseudosoma aberrans

Smells like Coccinellids Smells like Coccinellids Smells like Coccinellids Pyrazine

Acrid smell

Blest 1964

Fluid

Carpenter 1938

Froth

Rothschild and Haskell 1966

Pyrazine

Moore et al. 1990

X

Blest 1964

Heliconius chartiona

Pale yellow/black

Pyrazine

Moore et al. 1990

Heliconius hydarus

Red/black

Acrid

Cott 1940

Pyrazine

Moore et al. 1990

Heliconius melpomene

Red/yellow/black

Inachis ioa

Coloured wing flash

Hissing

Haskell 1956

Pachydota affinis

Back/yellow

X

Blest 1964

Pachydota punctata

Grey/yellow

X

Panaxia dominula

Cream/black/orange/red

Pyrazine

Papilio machaon

Yellow/black

Decaying pineapple

Blest 1964 Secretion

Rothschild 1961 Cott 1940

Behav Ecol Sociobiol Table 1 (continued) Species

Vision

Audition

Olfaction

Gustation/ nociception Fluid

Phragmatobia fuliginosa

Red/black

Pyrazine

Rhyparia purpurata

Yellow/red/black

Pyrazine

Rhodogastria bubo

White

Rhodogastria lupia

White/orange

Robinsonia sanea

White/black/orange

Spilosoma lubricipeda

White/black/orange

Thyanoprymna superba

Black/red

Tyria jacobaeae

Red/black

Utethesia bella

Red/white/Black

Sizzling

Reference

Rothschild 1961 Rothschild 1961

Acrid smell

Froth

Carpenter 1938

Acrid smell

Froth

Carpenter 1938

X

Blest 1964 Pyrazine

Rothschild 1961

X

Blest 1964 Pyrazine

Moore et al. 1990 X

Zerynthia polyxena

Yellow/black

Pyrazine

Zygaena lonicerae

Red/black

Pyrazine

Rothschild and Haskell 1966 Moore et al. 1990

Fluid

Rothschild 1961; Moore et al. 1990

Hymenoptera Bombus impatiens

Pale yellow/black

Buzzing

Rashed et al. 2009

Bombus terrestris

Yellow/white/black

Hissing, buzzing

Kirchner and Rochard 1999, Rowe and Guilford 2001

Apis florea

Pale yellow/black

Buzzing

Sen-Sarma et al. 2002

Apis mellifera

Pale yellow/black

Buzzing

Rashed et al. 2009

Croesus septentrionalis larvae

Yellow/green/black

Unpleasant

Secretion

Cott 1940

X component present but full description not available a

Species known or thought to be palatable (note that these species tend to be cryptically coloured and use flash displays)

b

Species not yet given formal classification

number of species showing that level of complexity decreases (there are 61 displays with two components, 20 with three components, and 8 with four components; see Table 2). Interestingly, this pattern tends to be repeated in each of the individual taxa (see Table 2): there are only two cases of the number of species increasing with increasing number of components (from three to four components in Orthoptera and from two to three components in Hemiptera). This suggests that this could be a genuine pattern worthy of more rigorous investigation. Finally, the displays show an incredible diversity of form. Not only do defensive displays have different numbers of components, but species appear to use different combinations of those components, even within the same taxon. For Table 2 The number of species (taken from Table 1) with multimodal defensive displays containing two, three or four different components

example, amongst the beetles in Table 1, some species combine warning coloration with sounds or odour alone (e.g. Lilioceris lilii and Epilachna curcurbitae), whilst others also produce secretions or sprays in combination with an odour (e.g. Adalia bipunctata). In addition, some defended beetles appear not to be warningly coloured at all but have multiple components in different sensory modalities (e.g. Scaphinotus andrewsi and Scaphinotus viduus). Although outside the scope of this current review, it may be possible to collect more data and use phylogenetic techniques to explore this variation to understand the selection pressures acting on multimodal defensive displays. These observations raise a number of questions. For example: Can the number or the diversity of predators that

Insect order

Two components

Three components

Four components

Orthoptera

6

3

4

Hemiptera

3

5

0

Coleoptera

12

8

0

Diptera

6

0

0

Lepidoptera (larvae)

7

4

1

Lepidoptera (adults)

23

9

3

Hymenoptera

4

1

0

Total

61

20

8

Behav Ecol Sociobiol

a species has drive defensive displays to be increasingly complex? Why are some toxic prey not typically aposematically coloured and yet have multiple components in nonvisual sensory modalities? Why do palatable prey have multimodal displays? But perhaps one of the most important, and key to this review, is: do non-visual components form part of a multimodal signal to predators, or are they simply additional lines of defence? This is the question that we address in the following section. Defence or communication? To be confident that we are studying multimodal warning signals, we need to rule out the possibility that non-visual components in an insect’s display are not a ‘secondary defence’. Secondary defences are those traits that enhance the chances of an individual surviving capture by a predator (Edmunds 1974) and can be considered to make an encounter unprofitable or aversive (Rowe and Guilford 2001; Ruxton et al. 2004). Under this definition, we have identified three ways that non-visual components can act as secondary defences: (1) by startling an attacking predator (‘startle hypothesis’); (2) by being inherently noxious (‘noxious hypothesis’); and (3) by confusing the sensory system of a predator to prevent a successful attack (‘sensory confusion hypothesis’). If we want to ensure that we are studying a multimodal warning signal, we need to make sure that a component is not fulfilling one of these alternative defensive functions. Startle hypothesis A startle display is produced suddenly and without warning at close proximity to the predator, and produces a period of hesitation in a predator that increases a prey’s chance of escape (Edmunds 1974). Although the best known work on startle displays has focused on visual startle (e.g. Schlenoff 1985; Ingalls 1993), non-visual components also appear to function in this way. For example, the whistles of the larvae of North American walnut sphinx (Amorpha juglandis) cause birds to hesitate or jump back from the sound source (Bura et al. 2011). Alternatively, non-visual components could startle predators that are hunting using sensory modalities other than vision. For example, peacock butterflies (Inachis io) produce sounds that startle mice hunting at night (Olofsson et al. 2011). In the case of the peacock butterfly, the sounds that are produced when it flicks it wings have no effect on a visually hunting predator that is startled instead by the visual component of the display (Vallin et al. 2005). Although these examples provide evidence for an acoustic startle response in predators, examples are currently restricted to species that are palatable and not known to contain toxins (Bura et al. 2012; Olofsson et al.

2011). Therefore, whilst it is possible for non-visual components to startle predators, there is no evidence for this in species that are aposematic. Noxious hypothesis In contrast, there is plenty of evidence that non-visual components of multimodal defensive displays can be aversive, particularly chemical components. Chemicals can be intrinsically noxious, for example, they can be bittertasting, irritating, or cause pain or sickness to predators (see Pasteels et al. (1983) and Eisner (2003) for good reviews). Predators can learn to avoid prey that they find distasteful or cause discomfort (e.g. Dean 1980; Brower and Fink 1985). Whilst we know that the chemicals produced by aposematic insects can be a deterrent, we should be cautious in readily ascribing this function to all chemical components. For example, pyrazine odour is a common component of multimodal warning displays (Moore et al. 1990), and yet it does not appear to be aversive to domestic chicks (Gallus gallus domesticus) (Guilford et al. 1987; Rowe and Guilford 1996) or passerines in the lab or in the field (Kelly and Marples 2004; Siddall and Marples 2011a). In addition, secretions are not necessarily aversive to a degree that prevents ingestion (Eisner and Grant 1980; Moore et al. 1990). For example, whilst ants (Formica spp.) may take time to clean their antennae following contact with the regurgitant of the common silkmoth caterpillar (Antheraea polyphemus), it does not deter them from returning to the nest with prey coated in the regurgitant (Brown et al. 2007). In addition, whilst bitter-tasting chemicals can be unpleasant to taste, there is evidence that distastefulness is not sufficient to prevent a successful attack from a predator and could instead be used as a signal (Marples et al. 1994; Skelhorn and Rowe 2009, 2010). Therefore, whilst some chemicals can indeed be intrinsically aversive, we should not automatically view all chemicals as being secondary defences, and explore their potential for being signals of toxicity (Eisner and Grant 1980; Rothschild et al. 1984; Ruxton and Kennedy 2006; Skelhorn and Rowe 2010). Sensory confusion hypothesis Finally, it is possible that non-visual components can affect the sensory systems of attacking predators to reduce the chances of a successful attack, although this appears to be rare. The tiger moth (Betholdia trigona) is the only species known to defend itself in this way, by using clicks to jam bat sonar (Corcoran et al. 2011). This highly specialised form of clicking appears to have evolved from clicks that were originally used as warning signals to bats, although a clear evolutionary

Behav Ecol Sociobiol

pathway has yet to be established (Corcoran et al. 2011; Conner and Corcoran 2012). Given that this is not even a widespread form of defence in arctiid moths that are predated by bats, it is unlikely to be a widespread form of secondary defence in aposematic prey more generally. As these examples show, non-visual components of defensive displays can be aversive to predators; equally, there is evidence that sounds, odours and distastefulness can act as signals rather than be a deterrent. Indeed, there is good evidence that non-visual components can play a signalling role in a wide variety of aposematic species (e.g. arctiid moths (Bates and Fenton 1990); ladybirds (Marples et al. 1994) and bees (Kirchner and Rochard 1999); see Rowe and Guilford (2001) for further examples), and the presence of similar non-visual components in species that share the same warning signals (i.e. they are visual mimics) strongly suggests that predators are using non-visual components as signals of defence (Moore et al. 1990; Rashed et al. 2009). However, the functional hypotheses that we discuss can only be applied to those species where we will now know that different components fulfil communicatory roles, and establishing how aversive non-visual components are to predators will be fundamental to this process.

Multimodal warning displays: a framework of functional hypotheses There is significant diversity and complexity in the defensive displays of chemically defended insects (Table 1). Over the last 20 years, there has been an increasing interest in the occurrence of multimodal warning signals and what selection pressures have led to their widespread evolution (Rowe and Guilford 2001). In particular, there is a clear need to explain why aposematically coloured prey also have additional components in their displays that can act as signals to predators. If aposematic coloration is such an effective signal, as is commonly believed, why do so many prey species invest in additional signalling components in their warning displays? Although many hypotheses have been put forward to explain their occurrence, there has been no critical evaluation of the evidence to support or refute these hypotheses. In the remainder of this review, we will discuss the supporting evidence for the different hypotheses and suggest ways in which they could be further tested. We aim to provide the first comprehensive theoretical framework for the study of multimodal warning signals that can help to identify areas for future research. In order to understand the evolution of multimodal warning displays, it is important to have a framework of testable hypotheses. Such a framework already exists for explaining the function of complex signals (Hebets and Papaj 2005), and here we will adapt it for the study of multimodal

warning displays. Because we are focusing on a specific signalling system, not all the hypotheses can be readily applied to our system and we have selected only those that directly apply in the context of aposematism (see Table 3). Following Hebets and Papaj (2005), we will separate our hypotheses into ‘content-based’ hypotheses (i.e. those based on ‘strategic design’; Guilford and Dawkins 1991) and ‘efficacy-based’ hypotheses (i.e. those based on ‘tactical design’; Guilford and Dawkins 1991). Unlike Hebets and Papaj (2005), however, we make no distinction between ‘efficacy-based’ hypotheses and ‘inter-signal interaction’ hypotheses, and consider both of these to be properties of signal reception and processing by receivers. Content-based hypotheses There are two ways in which having multiple signal components in a display can increase the information value of the signal. First, each component can inform the receiver about different qualities of the signaler (Møller and Pomiankowski 1993), and can be considered more broadly to include ‘any instance involving the transfer of more than one type of information’ (Hebets and Papaj 2005). These are known as ‘multiple messages’ (Møller and Pomiankowski 1993). The second way is by each component being a redundant or ‘backup’ signal (Møller and Pomiankowski 1993; Johnstone 1996), where multiple signals of the same quality will allow for better assessment of the overall quality of interest to the receiver. Both of these hypotheses can be applied to complex warning displays (see Table 3). Multiple messages Multiple messages can either provide information about multiple qualities of a signaler or about different aspects of the same quality (Hebets and Papaj 2005). In the context of aposematism, the quality of interest to a predator is how toxic a prey is, and signal components could provide different information about the toxins that prey contain. Chemical defences vary within and across species (e.g. Brower and Calvert 1985; Cohen 1985; de Jong et al. 1991), and reliable signals of toxin content could be selected for if prey with a high investment of toxins benefit from these signals (Guilford 1994). This could occur by predators selecting less toxic individuals in order to keep their toxin burden low (Sherratt et al. 2004; Skelhorn and Rowe 2007; Barnett et al. 2012; Halpin et al. 2012). Given the interest in multiple messages in the context of multicomponent sexually selected traits (Møller and Pomiankowski 1993; Candolin 2003), it may seem surprising that there has been no application of this idea in a multimodal warning display. This is most likely because research in aposematism tends to focus on the efficacy of

Behav Ecol Sociobiol Table 3 Functional hypotheses to explain the evolution of multimodal warning displays Hypothesis

Description

Content-based hypotheses Multiple messages

Each component informs the predator about a different quality of the prey

Redundant signals

Multiple components allow a more accurate assessment of the same quality of the prey by a predator

Efficacy-based hypotheses Efficacy back-up

Multiple components ensure detection by a predator in a variable environment

Perceptual variability

Signal components in different sensory modalities are most effective against different types of predator

Statistical summation

Multiple components decrease the reaction time to a signal by a predator

Attention-alerting

One component improves the degree to which a predator can discriminate along a different sensory dimension

Context-setting

One component provides the context in which another component is assessed by a predator

Differentiation

Multiple components enable aposematic prey to be more distinct from palatable mimics

Potentiation

One component enhances the aversion learning of another component of a warning display by a predator

Reduced generalisation

Predators show less generalisation around the visual signal when a non-visual component is present

Social learning

Non-visual components alert multiple predators to learn to avoid the prey

the signal, and so we know very little about whether components of multimodal warning displays could be informing predators about the toxin content of an individual. However, there have been a couple of recent studies that suggest that aspects of the visual colour pattern can be indicative of the amount of toxin that an individual prey contains (Bezzerides et al. 2007; Blount et al. 2012), and we know that avian predators in both the laboratory and in the wild can select differentially defended prey based upon how bitter they taste through taste-rejection behaviour (Brower and Calvert 1985; Skelhorn and Rowe 2006; Halpin et al. 2008a, b; Skelhorn and Rowe 2010). Therefore, it is possible that signal components could be informing predators of a prey’s toxicity and that each component signals a different aspect of the chemical defence. In order to test this idea, we need to know how different display components correlate with different toxins within a single species, and identify two separate display components that correlate with different aspects of the defence. In addition, it is also important to show that a predator is attentive to those signals and selects prey according to its current toxin burden. Currently, this is a tall order: the defensive chemistry of prey can be complex (Pasteels et al. 1983; Ritland 1994; Eisner 2003), and we are only starting to explore the possibility of display components being reliable signals of toxicity for predators (Bezzerides et al. 2007; Skelhorn and Rowe 2010; Blount et al. 2012). However, it is possible that the signals and defence chemistry of ladybirds could be a tangible system in which to investigate this since the visual and the chemical signals could both be reliable signals of toxicity (de Jong et al. 1991; Marples et al. 1994; Blount et al. 2012). It remains to be seen whether multiple messages can explain the evolution of signal complexity in warning displays.

Redundant signals The assumption underlying the redundant signals hypothesis is that components contain a degree of error in the quality that they are signalling. In the case of aposematism, this would mean that the expression of a signal does not perfectly correlate with underlying toxicity. As discussed in the previous section, there is the possibility that some colour signals may correlate with toxicity, but like any signal, this is unlikely to be perfect. In addition, warning signals are likely to have an additional source of unreliability through palatable Batesian mimics copying the colours (Dittrich et al. 1993), sounds (Rashed et al. 2009), and odours (Moore et al. 1990) of aposematic prey. Therefore, a predator cannot be fully sure of the toxicity associated with a given signal. At best, signals are likely to provide information on the probability that an individual is defended rather than allow a fully accurate assessment. To provide evidence in support of this hypothesis, we would need to show that multiple components co-vary with prey toxicity, and that predators benefit from an improved assessment of prey toxicity using multiple components (Johnstone 1996; Hebets and Papaj 2005). If we learn more about how signal components vary with prey toxicity within a species, it is possible to test whether or not the components signal the same or different qualities about prey toxicity. We would also need to show that predators acquire more accurate information about prey toxicity from multiple components compared to when just a single component is available to them. For example, perhaps predators are better able to regulate their toxin burden when multimodal signals are available (Skelhorn and Rowe 2007). We would predict that birds would be able to more accurately respond to changes in their toxin burden when they have two signal components

Behav Ecol Sociobiol

compared to one. By combining carefully controlled laboratory experiments with knowledge about how signals vary with toxicity, it may be possible to test this hypothesis. Efficacy-based hypotheses Efficacy-based hypotheses are concerned with the efficient transfer of the signal content from the signaler to the receiver (Guilford and Dawkins 1991). Factors that affect signal efficacy are the environment in which the signal is produced and the sensory and cognitive processes of the receiver. In this section, we will review various efficacy-based hypotheses (see Table 3) that could account for the evolution of aposematic signals. These hypotheses tend to lend themselves more easily to being empirically tested (see also Hebets and Papaj 2005), and consequently there appears to be some good evidence in support of some of these ideas. Efficacy back-up The efficacy back-up hypothesis is based on the idea that multiple components evolve in response to variability in the environment (Candolin 2003; Hebets and Papaj 2005). It predicts each component should provide the same information (i.e. they should co-vary), but each will act as a back-up to the other in the presence of environmental noise (Hebets and Papaj 2005). In order to test this idea, it is important to show that each component can act as a warning signal to predators, and that the probability of responding is greater when the two components are produced together compared to separately. To our knowledge, this hypothesis has not been tested in the context of aposematism. This is perhaps surprising given that we know that environmental noise can affect the transmission properties of different types of signal (Endler 1993; Bruum and Slabbekoorn 2005), and in particular because the conspicuousness and effectiveness of an aposematic prey’s visual signal will change according to the background it is on when found by a predator (e.g. Giittleman and Harvey 1980). Testing whether or not multiple warning display components can be efficacy back-up signals could be established in the laboratory using model empirical systems. For example, it may be possible to change the visual and olfactory environments of artificial prey using a ‘novel world’ set-up (e.g. Alatalo et al. 1996; Lindström et al. 1999) in order to test whether these signals can act as a back-up to one another. However, this would not tell us about whether or not these signals act as back-up signals in the wild. This could perhaps be investigated using comparative methods to test the prediction that aposematically coloured prey would be more likely to have additional nonvisual components in more visually complex environments. Alternatively, perhaps it is possible to take laboratory

experiments out into the field and manipulate the noise in a predator’s environment and measure the effect that has on the predation of either naturally occurring or artificial signalling prey (e.g. pastry baits (Speed et al. 2000) or artificial moths (Cuthill et al. 2005)). This hypothesis could certainly explain the evolution of multiple signal components, and there are opportunities for it to be tested. Perceptual variability The perceptual abilities of different species of predators are highly variable, for example, some diurnal predators possess excellent colour vision (e.g. birds; Bennett and Cuthill 1994), whilst others are unable to see colour and detect prey using luminance contrast instead (e.g. praying mantids; Prudic et al. 2007). As a consequence, prey must adapt their warning signals to predators’ sensory systems, and may need to produce signal components in different sensory modalities to be most effective against different types of predator. Although Hebets and Papaj (2005) conceived perceptual variability to be at the level of receivers within a species, there is no reason to restrict it to this particular case. Perceptual abilities are likely to vary hugely between different types of predator: whilst warning coloration is effective during the day to visually hunting predators, it is likely to be less effective against nocturnal predators that rely on different sensory modalities to detect and capture prey. Therefore, visual and non-visual components of warning displays could have evolved to be effective warning signals to predators hunting during the day and at night (Rothschild 1965; Pearson 1989). There is certainly good evidence from arctiid moths to support the idea that different display components have been selected to target diurnal and nocturnal predators. Species of tiger moths vary in the degree to which they are conspicuous and warningly coloured (some are cryptic whilst other have high-contrast red-and-black and yellowand-black patterns) as well as in whether or not they produce clicks when detected or attacked by a predator (Ratcliffe and Nydam 2008). This interspecific variation can be explained by the probability with which each species encounters diurnal and nocturnal predators. Species that are visually conspicuous are more active during the day (e.g. Apantesis nais and Spilosoma congrum) indicating that their coloration has been selected by visually hunting predators, such as birds. In contrast, ultrasonic clicks are produced predominantly by species that emerge later in the year and coincide with the peak of bat activity (e.g. Virbia aurantaca and Hypoprepia fucosa). For tiger moths at least, components in different sensory modalities appear to be independently targeting predators that are encountered at night and during the day (Ratcliffe and Nydam 2008). Whilst comparative studies certainly do show how selection acts on warning displays,

Behav Ecol Sociobiol

they do not demonstrate the use of different signals by a single prey species against different predators. This hypothesis still awaits conclusive evidence. Increased detection Detection is most likely to be enhanced by the use of multiple components through reducing a predator’s reaction time to a warning signal (Rowe 1999). Specifically, a predator’s reaction time to respond to a multimodal warning display is predicted to be more rapid compared to any single component encountered in isolation. In the psychological literature, this is referred to as ‘statistical summation’ and occurs only when the components in a display are produced together (Raab 1962). It is not clear how common simultaneous production of components is in nature. For example, aposematically coloured prey have usually already been detected before they produce odours and sounds in response to closer inspection or handling (e.g. Rothschild 1961; Haskell 1956; Marples et al. 1994). However, some components can be produced together, for example sounds and odours (e.g. Carpenter 1938; Rothschild and Haskell 1966; Eisner et al. 1974), or a chemical secretion accompanied by an additional visual display (e.g. Carpenter 1938; Blest 1964; Boppré 1986; Laurent et al. 2005). Clearly it is in the prey’s interest that the predator responds to its signal and that the signal stops an attack: the quicker this occurs, the less damage is likely to be done to the prey (Barnett 2007). To test this particular hypothesis, it would be necessary to manipulate the ability of prey to produce different components and measure predators’ reaction times with high-speed cameras. We would predict that not only would reaction times be faster to multimodal displays compared to single components but that survival of the prey would also be higher. As yet, there has been no specific investigation of predators’ reaction times to multimodal warning displays, and the question of whether complex displays could have evolved to reduce reaction times in predators remains an open one. Improved discrimination Discrimination occurs when animals need to respond differently to different stimuli, either due to innate predispositions or through learning (Rowe 1999). Discriminating among prey according to how profitable they are is clearly important to predators. We have identified three hypotheses that could explain how multimodal warning signals have evolved to enhance discrimination between aposematic prey and more profitable prey in the environment: (1) the ‘attention-alerting’ hypothesis; (2) the ‘context-setting’ hypothesis; and, (3) the ‘differentiation' hypothesis.

The attention-altering hypothesis states that ‘one signal can increase the degree to which a receiver focuses attention on another sensory field, and by doing so, improve discrimination within that field’ (Hebets and Papaj 2005). This idea is not new in the context of warning signals (Guilford 1994; Guilford and Rowe 2000). Guilford (1994) first suggested the idea that visual warning signals might be 'go-slow' signals that cause a predator to sample prey more cautiously in order to better assess their palatability through odours and distasteful chemicals. Testing this hypothesis would require investigating whether predators discriminate better on the basis of taste when prey are conspicuous compared to cryptic, but as yet no critical test exists. The second idea for how discrimination can be improved by multimodality is that one component acts as a context for another (Hebets and Papaj 2005). The context-setting hypothesis predicts that predators will change their responses to a signal component in the presence of another component. Studies of unlearned biases in avian predators provide support for this hypothesis. For example, the presence of pyrazine odour increases the reluctance of birds to attack and eat prey that are typically warningly coloured, i.e. those prey that are red or yellow (Rowe and Guilford 1996; Jetz et al. 2001; Kelly and Marples 2004), novel (Rowe and Guilford 1999; Jetz et al. 2001), or conspicuous (Lindström et al. 2001). In addition, buzzing sounds can enhance chicks’ biases against yellow and novel-coloured crumbs (Rowe and Guilford 2001; but see Hauglund et al. (2006) and Siddall and Marples (2011b)), and the taste of quinine (a bitter-tasting alkaloid) enhances chicks' biases against yellow and red food (Rowe and Skelhorn 2005). Taken together, these studies show that the presence of a non-visual component (an odour, sound or taste) could provide a contextual cue that makes birds reduce their attacks on prey that are typically warningly coloured. If this is the case, we would predict that the contextual components would only be used by predators if they provided them with information about how toxic prey were likely to be. There is some evidence to support this: birds are more likely to avoid warningly coloured prey when the non-visual contextual component is associated with a natural defence in the wild (Marples and Roper 1996; Roper and Marples 1997; Rowe and Guilford 2001; Skelhorn et al. 2008; but see Jetz et al. (2001) and Hauglund et al. (2006)). This suggests that predators may have co-evolved adaptive responses to reduce the chances of ingesting toxic prey. The final way in which discrimination could be improved is through additional components making aposematic prey more distinct from their palatable mimics. We refer to this as the differentiation hypothesis. In Batesian mimicry, a palatable mimic is typically thought to benefit from sharing the colour pattern of a defended species, but the mimicry comes at a cost to the defended model species since predators will

Behav Ecol Sociobiol

view the pattern as more palatable (Bates 1862; Gilbert 2005). Consequently, whilst mimics should be selected to be indistinguishable from their models to avoid predation, models should be selected to look dissimilar to their mimics to enable predators to discriminate between them. Having additional components would be one way to achieve this. If this has played an important role in the evolution of complex warning displays, we would predict that models will have more non-visual components than their visual mimics, and that increasing the number of components in different sensory modalities would make the signal more costly or impossible for mimics to copy (which has also been suggested to explain the evolution of slow flight among aposematically coloured prey (Srygley, 2004)). Currently, there are not enough data to test these predictions. However, the fact that visual mimics can have non-visual components that are similar to those of their models (e.g. Gaul 1952; Rothschild 1961; Moore et al. 1990; Rashed et al. 2009) argues against this hypothesis since models may not be able to escape their mimics just by having multimodal displays. For example, those hoverflies that are visually similar to bees and wasps also buzz upon attack like their hymenopteran models. However, these buzzes do not appear to have been selected to sound very similar to those of their specific model species (Rashed et al. 2009). This could mean that syrphids find it difficult to mimic the hymenopteran buzzes, although alternatively it could be that predators do not discriminate finely between different types of buzzing. There are no data to test the prediction that models are more likely to have a greater number of components in their warning displays than their mimics do. A comparative approach would be required to test whether this is indeed the case. Increased learning and memory Claridge (1974) first suggested for sounds, and Rothschild et al. (1984) for odours, that non-visual components might act to enhance the learned association between the coloration and the unpalatability (Rothschild et al. 1984). This would occur not because the sounds and odours were aversive, but because they cause the visual signal to be more salient to a predator (Rothschild and Moore 1987). We refer to this as the ‘potentiation’ hypothesis because this is the term used in the psychological literature to refer to this phenomenon (e.g. Mackintosh 1974). The potentiation hypothesis predicts that the aversion learning of a visual signal should be stronger or faster in the presence of a non-visual component, and that the memory for that warning signal should be more robust. Learning to avoid a visual cue associated with unpalatability is certainly enhanced by the presence of non-visual components. Pyrazine odour, a common feature of multimodal warning

displays (Rothschild et al. 1984; Moore et al. 1990), increases the ability of Lister hooded rats (Rattus norvegicus) to associate a visual environmental cue (cage colour) with unpalatability (Kaye et al. 1989), and chicks to associate a colour signal with distastefulness (Siddall and Marples 2008). Birds also show enhanced colour avoidance when both acoustic (Rowe 2002; but see Hauglund et al. (2006) and Siddall and Marples (2011b) that have failed to replicate this result) and gustatory cues (Franchina et al. 1997). Although there is good evidence that one sensory component can potentiate the learning of another, these experiments have been conducted in the laboratory, and there is no evidence that they enhance colour aversion learning in the wild. Indeed, the only experiment that has attempted to do this did not find a potentiating effect of pyrazine on wild birds’ avoidance of yellow prey (Siddall and Marples 2011a). Therefore, it is not clear whether such a cognitive mechanism can explain the widespread evolution of multimodal warning signals. If the strength of the association between the colour signal and the defence is increased by the presence of a non-visual component, this could mean that predators are less likely to generalise around the visual signal (Ghirlanda and Enquist 2003). This effect could be advantageous for species that are trying to escape the resemblance of a visual Batesian mimic (Rowe and Guilford 2000). It would be possible to test if predators do show restricted generalisation around a colour signal in the presence of a non-visual component in laboratory experiments, where generalisation could be readily measured. Measuring learning and generalisation in free-living predators would be more challenging. However, we would also predict that aposematic prey with mimics would be more likely to have additional components. Whilst this is also a prediction of the differentiation hypothesis, it could be discriminated from this hypothesis by showing that mimics appear more similar to their models when the models have multiple components in their display. Again, comparative approaches would be useful for this type of analysis. An extension of this idea that has not been explored is that a non-visual component could enhance aversion through social learning in group-living predators. We call this a ‘social learning’ hypothesis. Although individuals can learn to avoid prey that conspecifics find aversive using visual cues (Mason and Reidinger 1982; Fryday and Greig-Smith 1994), these are not always available in complex environments (Skelhorn 2011). The production of a sound or an odour by a defended prey could have better transmission properties, perhaps alerting predators nearby to pay more attention to the behaviour of an attacker and the appearance of the rejected prey. In this way, sounds and odours could simultaneously alert multiple predators to learn to avoid the prey. The likelihood that this is a

Behav Ecol Sociobiol

significant selection pressure will depend on the social structure of the predators, but it could certainly enhance the learned avoidance of prey in group-living predators.

Skelhorn, Martin Stevens and Jeri Wright for helpful discussions (that they may or may not remember) on various aspects of the manuscript. The review was supported by a BBSRC and NERC co-funded project grant (BB/G00188X/1).

Summary References It is clear that there are differing degrees of support for each of the functional hypotheses that we have presented here: some are empirically supported whilst others are as yet untested. However, much of the supporting data come from laboratory experiments which one could argue does not provide conclusive evidence to support any one of these hypotheses. Whilst the laboratory certainly has its advantages of being able to control both the presentation of components as well as the experience of the predators, it is also disadvantageous because the stimuli used may not be ecologically valid. Therefore, although laboratory studies can establish how multimodal displays can function, we still do not know if selection pressures measured in the laboratory can and do operate in the wild. For this, we argue that we need data from more studies of the prey themselves. Table 1 suggests that there may be enough data to conduct some comparative analyses to help identify the selection pressures acting on multimodality in insect warning signals. In addition, it is perhaps time to turn our attention towards trying to understand the selection pressures acting on a complex warning display of a single species. To our knowledge, we still do not know this for any species leaving clear room for future research.

Conclusion Our review has revealed the abundance and diversity of multimodal warning displays found in insects. There is a huge amount of variation in both the number of components that prey have in their displays and the form that those components can take. In the same way that warning coloration can vary among species, so can the type of sounds that prey produce, and the properties of their chemical secretions. It has been surprising to us to discover that we know relatively little about the evolution of these displays. We have reviewed 11 functional hypotheses that could explain the evolution of multimodal warning displays, but many of them lack any critical testing. We hope that our review encourages researchers to attempt to identify the selection pressures acting on multimodal warning displays, particularly in natural predator–prey systems. Acknowledgments We would like to James Higham and Eileen Hebets for inviting us to submit a review to this special issue, and to Eileen and two anonymous reviewers for their enormously stimulating and helpful comments on the manuscript. We would also like to thank Melissa Bateson, Ben Brilot, Sue Healy, Domhnall Jennings, John

Alatalo RV, Mappes J (1996) Tracking the evolution of warning signals. Nature 382:708–710 Alexander RD (1964) Acoustic communication in arthropods. Ann Rev Ent 12:495–526 Barber JR, Conner WE (2007) Acoustic mimicry in a predator–prey interaction. Proc Nat Acad Sci 104:9331–9334 Barnett CA (2007) The effects of energetic and physiological state on the foraging decisions of birds. Unpublished PhD thesis, Newcastle University Barnett CA, Skelhorn J, Bateson M, Rowe C (2012) Educated predators make strategic decisions to eat defended prey according to their toxin content. Behav Ecol 23:418–424 Bates HW (1862) Contributions to an insect fauna of the Amazon valley (Lepidoptera: Heliconidae). Trans Linn Soc Lond 23:495–566 Bates DL, Fenton MB (1990) Aposematism or startle? Predators learn their responses to the defenses of prey. Can J Zool 68:49–52 Bedford GO, Chinnick LJ (1966) Conspicuous displays in two species of Australian stick insects. Anim Behav 14:518–521 Bennett AD, Cuthill IC (1994) Ultraviolet vision in birds: what is its function? Vis Res 34:1471–1478 Bezzerides AL, McGraw KJ, Parker RS, Husseini J (2007) Elytra color as a signal of chemical defense in the Asian ladybird beetle Harmonia axyridis. Behav Ecol Sociobiol 61:1401–1408 Bisset GW, Frazer JFD, Rothschild M, Schachter M (1960) A pharmacologically active choline ester and other substances in the garden tiger moth Arctia caja (L.). Proc R Soc Lond B 152:255–262 Blest AD (1964) Protective display and sound production in some New World arctiid and ctenuchid moths. Zoologica 49:161–181 Blount JD, Rowland HM, Drijfhout FP, Endler JA, Inger R, Sloggett JJ, Hurst GDD, Hodgson DJ, Speed MP (2012) How the ladybird got its spots: effects of resource limitation on the honesty of aposematic signals. Funct Ecol 26:334–342 Boppré M (1986) Insect pharmacophagously utilizing defensive plant chemicals (pyrrolizidine alkaloids). Naturwissenschaften 73:17– 26 Brower LP, Calvert WH (1985) Foraging dynamics of bird predators on overwintering monarch butterflies in Mexico. Evolution 39:852–868 Brower LP, Fink LS (1985) A natural toxic defense system: cardenolides in butterfly versus birds. Ann N Y Acad Sci 443:171–188 Brown SG, Boettner GH, Yack JE (2007) Clicking caterpillars: acoustic aposematism in Antheraea polyphemus and other Bombycoidea. J Exp Biol 210:993–1005 Bruum H, Slabbekoorn H (2005) Acoustic communication in noise. Adv Study Behav 35:151–209 Bura VL, Fleming AJ, Yack JE (2009) What's the buzz? Ultrasonic and sonic warning signals in caterpillars of the great peacock moth (Saturnia pyri). Naturwiss 96:713–718 Bura VL, Rohwer VG, Martin PR, Yack JE (2011) Whistling in caterpillars (Amorpha juglandis, Bombycoidea): sound-producing mechanism and function. J Exp Biol 214:30–37 Bura VL, Hnain AK, Hick JN, Yack JE (2012) Defensive sound production in the tobacco hornworm, Manduca sexta (Bombycoidea: Sphingidae). J Insect Behav 25:114–126

Behav Ecol Sociobiol Candolin U (2003) The use of multiple cues in mate choice. Biol Rev 78:559–571 Carpenter GDH (1938) Audible emission of defensive froth by insects. Proc Zool Soc A 242–251 Carpenter GDH, Ford EB (1933) Mimicry. Methuen, London Claridge MF (1974) Stridulation and defensive behaviour in the ground beetle, Cychrus caraboides (L.). J Entomol A 49:7–15 Cohen JA (1985) Differences and similarities in cardenolide contents of Queen and Monarch butterflies in Florida and their ecological and evolutionary implications. J Chem Ecol 11:85–103 Conner WE, Corcoran AJ (2012) Sound strategies: The 65-millionyear-old battle between bats and insects. Annu Rev Entomol 57:21–39 Corcoran AJ, Barber JR, Hristov NI, Conner WE (2011) How do tiger moths jam bat sonar? J Exp Biol 214:2416–2425 Cott HB (1940) Adaptive coloration in animals. Methuen, London Cuthill IC, Stevens M, Sheppard J, Maddocks T, Parraga CA, Troscianko TS (2005) Disruptive coloration and background pattern matching. Nature 434:72–74 Darwin C (1887) The life and letters of Charles Darwin: including an autobiographical chapter, edited by his son Francis Darwin. Murray, London de Jong PW, Holloway GJ, Brakefield PM, Vos H (1991) Chemical defense in the ladybird beetles (Coccinellidae). II. Amount of reflex fluid, the alkaloid adaline and individual variation in defense in 2spot ladybirds (Adalia bipunctata). Chemoecology 2:15–19 Dean J (1980) Encounters between bomardier beetles and two species of toads (Bufo americanus, Bufo marinus): speed pf prey capture does not determine success. J Comp Physiol 135:41–50 Dittrich W, Gilbert F, Green P, McGregor P, Grewcock D (1993) Imperfect mimicry: a pigeon’s perspective. Proc R Soc Lond B 251:195–200 Dunning DC (1968) Warning sounds of moths. Z Tierpsychol 25:129–138 Edmunds M (1974) Defence in animals. Longman, Harlow Eisner T (2003) For the love of insects. Harvard University Press, Cambridge Eisner T, Grant RP (1980) Toxicity, odor aversion and “olfactory aposematism”. Science 213:416 Eisner T, Aneshansley D, Eisner M, Rutowski R, Chong B, Meinivald J (1974) Chemical defense and sound production in Australian tenebrionid bettles (Adelium spp.). Psyche 81:189–208 Endler JA (1993) The color of light in forests and its implications. Ecol Monogr 63:1–27 Exnerová A, Landová E, Stys P, Fuchs R, Prokopová M, Cehláriková P (2003) Reactions of passerine birds to aposematic and nonaposematic firebugs (Pyrrhocoris apterus; Heteroptera). Biol J Lonn Soc 78:517–525 Fenton MB, Roeder KD (1974) The microtymbals of some Arctiidae. J Lepidopt Soc 28:205–211 Franchina JJ, Moon C, Peters S (1997) Effects of toxin magnitude on taste aversion and taste-potentiated aversion to visual cues in chicks (Gallus domesticus). Physiol Behav 62:605–609 Fryday SL, Greig-Smith PW (1994) The effects of social learning on the food choice of the house sparrow (Passer domesticus). Behaviour 128:281–300 Gaul AT (1952) Audio mimicry: an adjunct to colour mimicry. Psyche 59:82–83 Ghirlanda S, Enquist M (2003) A century of generalization. Anim Behav 66:15–36 Gilbert FS (2005) The evolution of imperfect mimicry. In: Fellowes MDE, Holloway GJ, Rolff J (eds) Insect evolutionary ecology. CABI, Wallingford, pp 231–288 Gittleman JL, Harvey PH (1980) Why are distasteful prey not cryptic? Nature 286:149–150 Guilford T (1994) Go-slow signaling and the problem of automimicry. J Theor Biol 170:311–316

Guilford T, Dawkins MS (1991) Receiver psychology and the evolution of animal signals. Anim Behav 42:1–14 Guilford T, Rowe C (2000) Aposematism: to be red or dead. Trends Ecol Evol 15:261–262 Guilford T, Nicol C, Rothschild M, Moore BP (1987) The biological roles of pyrazines: evidence for a warning odour function. Biol J Linn Soc 31:113–128 Halpin CG, Skelhorn J, Rowe C (2008a) Being conspicuous and defended: selective benefits for the individual. Behav Ecol 19:1012–1017 Halpin CG, Skelhorn J, Rowe C (2008b) Naïve predators and selection for rare conspicuous defended prey: the initial evolution of aposematism revisited. Anim Behav 75:771–781 Halpin CG, Skelhorn J, Rowe C (2012) The relationship between sympatric defended species depends upon predators’ discriminatory behaviour. PLoS One 7:e44895 Haskell PT (1956) tape recording of sounds emitted by the peacock butterfly. Proc R Ent Soc C 21:20–22 Haskell PT (1961) Insect sounds. HF&G Witherby, London Hatle JD, Salazar BA, Whitman DW (2002) Survival advantage of sluggish individuals in aggregations of aposematic prey during encounters with ambush predators. Evol Ecol 16:415–431 Hauglund K, Snorre HB, Lampe H (2006) Responses of domestic chicks Gallus gallus domesticus to multimodal aposematic signals. Behav Ecol 17:392–398 Hebets EA, Papaj DR (2005) Complex signal function: developing a framework of testable hypotheses. Behav Ecol Sociobiol 57:197– 214 Hill SA (2007) Sound generation in Mantis religiosa (Mantodea: Mantidae): stridulatory structures and acoustic signal. J Orthopt Res 16:35–49 Ingalls V (1993) Startle and habituation responses of blue jays (Cyanocitta cristata) in a laboratory simulation of anti-predator defenses of Catocala moths (Lepidoptera: Noctuidae). Behaviour 126:77–96 Järrvi T, Sillen-Tullberg D, Wiklund C (1981) The cost of being aposematic: an experimental study of predation on larvae of Papilio machaon by the great tit Parus major. Oikos 36:267–272 Jetz W, Rowe C, Guilford T (2001) Non-warning odors trigger innate color aversions—as long as they are novel. Behav Ecol 12:134–139 Johnstone RA (1996) Multiple displays in animal communication: “back-up signals” and “multiple messages”. Phil Trans R Soc Lond B 352:329–338 Jones CG, Whitman DW, Compton SJ, Silk PJ, Blum MS (1989) Reduction in diet breadth results in sequestration of plant chemicals and increases efficacy of chemical defense in a generalist grasshopper. J Chem Ecol 15:1811–1822 Kaye H, Mackintosh NJ, Rothschild M, Moore BP (1989) Odour of pyrazine potentiates an association between environmental cues and unpalatable taste. Anim Behav 37:563–568 Kelly DJ, Marples NM (2004) The effects of novel odour and colour cues on food acceptance by the zebra finch (Taeniopygia guttata). Anim Behav 68:1049–1054 Kirchner WH, Röschard J (1999) Hissing in bumblebees: an interspecific defence signal. Insect Soc 46:239–243 Laurent P, Braekman JC, Daloze D (2005) Insect chemical defence. Top Curr Chem 240:167–229 Lindström L, Alatalo RV, Mappes J, Riipi M, Vertainen L (1999) Can aposematic signals evolve by gradual change? Nature 397:249–251 Lindström L, Rowe C, Guilford T (2001) Pyrazine odour makes visually conspicuous prey aversive. Proc R Soc Lond B 268:1–4 Mackintosh NJ (1974) The psychology of animal learning. Academic, London Marples NM, Roper TJ (1996) Effects of colour and smell on the response of naive chicks towards food and water. Anim Behav 51:1417–1424

Behav Ecol Sociobiol Marples NM, van Veelen W, Brakefield PM (1994) The relative importance of colour, taste and smell in the protection of an aposematic insect Coccinella septempunctata. Anim Behav 48:967– 974 Mason JR, Reidinger RF (1982) Observational learning of food aversions in red-winged blackbirds (Agelaius phoeniceus). Auk 99:548–554 Masters WM (1979) Insect disturbance stridulation: its defensive role. Behav Ecol Sociobiol 5:187–200 Møller AP, Pomiankowski A (1993) Why have birds got multiple sexual ornaments? Behav Ecol Sociobiol 32:167–176 Moore BP, Brown MV, Rothschild M (1990) Methylalkylpyrazines in aposematic insects, their hostplants and their mimics. Chemoecology 1:43–51 Nickle DA, Castner JL, Smedley SR, Attygalle AB, Meinwald J, Eisner T (1996) Pyrazine emission by a tropical katydid: an example of chemical aposematism? (Orthoptera Tettigoniidae: Copiphorinae: Vestra Stål). J Orthopt Res 5:221–223 Olofsson M, Vallin A, Jakobsson S, Wiklund C (2011) Winter predation on two species of hibernating butterflies: monitoring rodent attacks with infrared cameras. Anim Behav 81:529–534 Pasteels JM, Grégoire J-C, Rowell-Rahier M (1983) The chemical ecology of defense in arthropods. Ann Rev Ent 28:263–289 Pearson DL (1989) What is the adaptive significance of multicomponent defensive repertoires? Oikos 54:251–253 Pomini AM, Machado G, Pinto-da-Rocha R, Macías-Ordóñez R, Marsaioli AJ (2010) Lines of defense in the harvestman Hoplobunus mexicanus (Arachnid: Opiliones): aposematism, stridulation, thanatosis, and irritant chemicals. Biochem Syst Ecol 38:300–308 Poulton EB (1890) The colours of animals: their meaning and use especially considered in the case of insects. Kegan Paul, Trench, Trübner and Co, London Prudic KL, Skemp AK, Papaj DR (2007) Aposematic coloration, luminance contrast and the benefits of conspicuousness. Behav Ecol 18:41–46 Raab DH (1962) Statistical facilitation of simple reaction times. Trans NY Acad Sci 24:574–590 Rashed A, Khan MI, Dawson JW, Yack JE, Sherratt TN (2009) Do hoverflies (Diptera: Syrphidae) sound like the Hymenoptera they morphologically resemble? Behav Ecol 20:396–402 Ratcliffe JM, Nydam ML (2008) Multimodal warning signals for a multiple predator world. Nature 455:96–99 Ritland DB (1994) Variation in palatability of queen butterflies (Danaus gilippus) and implications regarding mimicry. Ecology 75:732–746 Roper TJ, Cook SE (1989) Responses of chicks to brightly coloured insect prey. Behaviour 110:276–293 Roper TJ, Marples NM (1997) Odour and colour as cues for tasteavoidance learning in domestic chicks. Anim Behav 53:1241–1250 Roper TJ, Redston S (1987) Conspicuousness of distasteful prey affects the strength and durability of one-trial aversive learning. Anim Behav 35:739–747 Roth LM, Eisner T (1962) Chemical defenses of arthropods. Annu Rev Entomol 7:107–136 Rothschild M (1961) Defensive odours and Mullerian mimicry among insects. Trans R Ent Soc Lond 113:101–121 Rothschild M (1965) Proc R Ent Soc Lond C 30:3 Rothschild M, Haskell PT (1966) Stridulation of the garden tiger moth Arctia caja L. audible to the human ear. Proc R Ent Soc Lond A 41:167–170 Rothschild M, Moore BP (1987) Pyrazines as alerting signals in toxic plants and insects. In: Labeyrie V, Fabres G, Fachaise D (eds) Insect–plants. W. Junk, Dordrecht Rothschild M, Moore BP, Brown WV (1984) Pyrazines as warning odour components in the monarch butterfly, Danaus plexippus,

and in moths of the genera Zygaena and Amata (Lepidoptera). Biol J Linn Soc 23:372–380 Rowe C (1998) Multicomponent signals. Unpubl. DPhil thesis, University of Oxford Rowe C (1999) Receiver psychology and the evolution of multicomponent signals. Anim Behav 58:921–931 Rowe C (2002) Sound improves visual discrimination learning in avian predators. Proc R Soc Lond B 269:1353–1357 Rowe C, Guilford T (1996) Hidden colour aversions in domestic chicks triggered by pyrazine odours of insect warning displays. Nature 383:520–522 Rowe C, Guilford T (1999) Novelty effects in a multimodal warning signal. Anim Behav 57:341–346 Rowe C, Guilford T (2000) Aposematism: to be red or dead. Trends Ecol Evol 15:261–262 Rowe C, Guilford T (2001) The evolution of multimodal warning displays. Evol Ecol 13:655–671 Rowe C, Skelhorn J (2005) Colour biases are a question of taste. Anim Behav 69:587–594 Ruxton GD, Kennedy MW (2006) Peppers and poisons: the evolutionary ecology of bad taste. J Anim Ecol 75:1224–1226 Ruxton GD, Sherratt TN, Speed MP (2004) Avoiding attack: the evolutionary ecology of crypsis, warning signals and mimicry. OUP, Oxford Schilman PE, Lazzari CR, Manrique G (2001) Comparison of disturbance stridulations in five species of triatominae bugs. Acta Trop 79:171–178 Schlenoff DH (1985) The startle responses of blue jays (Cyanocitta cristata) to Catocala (Lepidoptera: Noctuidae) prey models. Anim Behav 33:1057–1067 Schuler W, Hesse E (1985) On the function of warning coloration: a black and yellow pattern inhibits prey-attack by naive domestic chicks. Behav Ecol Sociobiol 16:249–255 Sen-Sarma M, Fuchs S, Werber C, Tautz J (2002) Worker piping triggers hissing for coordinated colony defence in the dwarf honeybee Apis florae. Zoology 105:215–223 Sherratt TN, Speed MP, Ruxton GD (2004) Natural selection on unpalatable species imposed by state-dependent foraging behaviour. J Theor Biol 228:217–226 Siddall EC, Marples NM (2008) Better to be bimodal: the interaction of color and odor on learning and memory. Behav Ecol 19:425– 432 Siddall EC, Marples NM (2011a) The effect of pyrazine odor on avoidance learning and memory in wild robins Erithacus rubecula. Curr Zool 57:208–214 Siddall EC, Marples NM (2011b) Hear no evil: the effect of auditory warning signals on avian innate avoidance, learned avoidance and memory. Curr Zool 57:197–207 Skelhorn J (2011) Colour biases are a question of conspecifics’ taste. Anim Behav 81:825–829 Skelhorn J, Rowe C (2006) Avian predators taste-reject aposematic prey on their level of chemical defence. Biol Lett 2:348–350 Skelhorn J, Rowe C (2007) Predators’ toxin burdens influence their strategic decisions to eat toxic prey. Curr Biol 17:1479– 1483 Skelhorn J, Rowe C (2009) Distastefulness as an antipredator defence strategy. Anim Behav 78:761–766 Skelhorn J, Rowe C (2010) Birds learn to use distastefulness as a signal of toxicity. Proc R Soc Lond B 277:1729–1734 Skelhorn J, Griksaitis D, Rowe C (2008) Colour biases are more than a question of taste. Anim Behav 75:827–835 Smith SM (1975) Innate recognition of coral snake pattern by a possible avian predator. Science 187:759–760 Speed MP, Alderson NJ, Hardman C, Ruxton GD (2000) Testing Müllerian mimicry: an experiment with wild birds. Proc R Soc Lond B 267:725–731

Behav Ecol Sociobiol Srygley RB (2004) The aerodynamic costs of warning signals in palatable mimetic butterflies and their distasteful models. Proc R Soc Lond B 271:589–594 Vallin A, Jakobsson S, Lind J, Wiklund C (2005) Prey survival by predator intimidation: an experimental study of peacock butterfly defense against blue tits. Proc R Soc Lond B 272:1203–1207

Wallace AR (1867) Journal of Proceeding of the Entomological Society of London meeting March 4th, 5:lxxx-lxxxi, In Transactions of the Entomological Society of London 3rd series, vol III, London 1864–1869 Wheeler JW, Chung RH, Oh SK, Benfield EF, Neef SE (1970) Defensive secretions of cychrine beetles (Coleoptera: Carabidae). Ann Ent Soc Am 63:469–471