VANE-WRIGHT, R.I. & P.R. ACKERY, editors. 1984.
The Biology of Butterflies.
Symposium of the Royal Entomological Society of London, No. 11
Academic Press. London.
xxiv + 429 pp.
20. Visual Communication and Sexual Selection
Among Butterflies
tRobert E. Silberglied
Smithsonian Tropical Research Institute, Balboa, Republic of Panama
The wings of butterflies are adorned with a wealth Butterfly wing colours and patterns serve several
of pattern and a display of colour unrivalled in the important functions. Unlike some colours associated
living world. Some butterflies have uniformly- with metabolically important compounds (e.g.
coloured, unpatterned wings ofwhite; others reflect haemoglobin, chlorophyll, cytochrome), butterfly
yellow, orange, red, green, blue, violet, brown or wing pigments and structural colours must serve
black. The colours are produced in various ways (Fox biophysical functions because they are located outside
& Vevers 1960), including selective absorption by the metabolic pool, in the dead cellular skeletons of
pigments (Ford 1945), tyndall scattering (Huxley the wing scales. Certain colours absorb or reflect
1976) and thin-film interference (Simon 1971). In radiation effectively (Watt 1968). A butterfly may reg-
some species, the wings bear simple patterns, such ulate its body temperature by positioning itself and
as contrasting colour along the veins. In others its wings with respect to the sun (Vielmetter 1954,
multicoloured patterns ofanastomosing lines, ripple Clench 1966, Watt 1968, Findlay et al. 1983), but
patterns or eyespots are found (Nijhout 1980a). only the basal regions of the wings play an important
These complex patterns are genetically determined role in thermoregulation (Wasserthal 1975, Douglas
and developmentally controlled (Nijhout 1978, 1979). We must look elsewhere if we are to under-
1980b, 1981). The coloration and pattern of a stand the function of the coloration of the remainder.
butterfly's dorsal and ventral wing-surfaces may be The colour patterns ofbutterflies have usually been
the same or different. The two sexes may be identical, discussed in the context of one or both of two
slightly different, or in some cases so distinctive that contrasting, communicative functions: protection
they were originally described in different genera. from predators (Poulton 1908, Cott 1940), and social
Wing colour may be seasonally, geographical or signals used during courtship (Silberglied 1977).
locally variable or polymorphic. Furthermore, closely Darwin (1874) devoted a chapter to the Lepidoptera
related species may appear extremely different, while in The descem of Man and seleetion in relation 10 sex.
unrelated species can display nearly identical colour He provided numerous examples and observations
patterns (Vane-Wright 1971, 1979a,b). to support his view that, 'although many serious
Attempts to understand this panoply of spectral objections may be urged, it seems probable that most
diversity have resulted in major contributions to our of the brilliantly coloured species ofLepidoptera owe
knowledge of the evolutionary process. The study their colours to [inter-] sexual selection, excepting
ofbutterfly colour and pattern has played an impor- in certain cases ... in which conspicuous colours
tant role in the development of the theories of both have been gained through mimicry as a protection.'
natural (Darwin 1859) and sexual selection (Darwin Sexual selection on pattern and colour may oppose
1874, Hingston 1933), lJlimicry (Bates 1862, Miiller other selective forces, such as predation (e.g.
1878c, 1879, Carpenter & Ford 1933, Wickler 1968) Carpenter & Ford 1933, Stride 1956, 1958b, Brower
and genetic polymorphism (Sheppard 1961a, Robin- 1963, Turner 1977a, 1978, Vane-Wright 1971, 1975,
son 1971). In return, information from ecological, 1976, 1979a, Ch.23). A fuller understanding of the
behavioural, genetic, developmental and evolutionary nature and intensity of sexual selection for colour
studies has contributed greatly to our understanding and pattern in butterflies is therefore ofgreat impor-
the significance of butterfly colour and pattern. tance in evaluating other evolutionary processes.
tBob Silberglied died in a tragic accident on 13 January 1982, a few days after submitting his fmal draft of this paper.
208 Robert E. Silberglied
Darwin's view that intersexual selection is of
primary importance in the evolution ofbrilliant male
butterfly coloration and sexual dimorphism has
persisted with little change for more than a century.
During this time a host of physiological and
behavioural studies, both observational and experi-
mental, have greatly augmented our knowledge of
butterfly vision and its role in courtship. This paper
explores the extent to which his view, which has been
incorporated into numerous reviews, texts and theore-
tical discussions, is corroborated by our present
knowledge ofvisual communication among butterflies.
Darwin's Views on Sexual Selection
and Butterfly Coloration
Darwin (1874) was particularly impressed with sexual
dimorphism as exhibited by many butterfly species,
and by the disturbation of this phenomenon within
genera. He was intrigued with the problem of why
the sexes differ in some species, but not in close
relatives. Were the colours and patterns attributable
to direct effects of the environment in which the
butterflies lived, were they largely protective, were
they employed as attractive devices by the two sexes,
or did they perhaps serve some unknown purpose?
He pointed out that when the sexes differed, it
generally affected the colour and pattern of the dorsal
wing surfaces, those usually concealed when the
butterfly is at rest. The male usually has the more
beautiful, brilliant or striking dorsal coloration.
According to Darwin, the female generally exhibits
'ancestral' features, from which the male pattern
departs to greater or lesser degree. Thus females of
closely related species tend to resemble one another
more than do the males.
Darwin recognized that 'whenever colour has been
modified for some special purpose, this has been, as
far as we can judge, either for direct or indirect
protection, or as an attraction between the sexes.' The
similarity of the ventral wing surfaces in both sexes
reflects the fact that they generally rest with the wings
raised, with this surface exposed to view. The ventral
surface is usually mottled or patterned in such a way
as to resemble the background against which the
butterfly rests, thus making it more difficult to detect.
That most nocturnal Lepidoptera were drab in
coloration also impressed Darwin, as well as the fact
that diurnal moths often have brilliant coloration.
However, unlike butterflies, diurnal moths rarely
exhibit strong sexual dimorphism. (The major
exceptions involve sex-limited mimicry, e.g. Callo-
samia promethea, Pericopis spp.)
Darwin recognized that butterfly courtships are
often prolonged. He knew that males are aggressive
and competitive but more or less indiscriminate in
their choice of mates, and that a female could refuse
a courting male. The location of the colours, and the
behaviour of the males, indicated to him an active
function in display toward the opposite sex (Darwin
1880). Because butterflies exhibit colour preferences
while feeding, and can be attracted to coloured
decoys, he reasoned that they perceive and 'admire'
bright colours. He believed that females were more
excited by, and preferred, the more brilliant, beautiful
males. By choosing mates on the basis of colour,
females would select for males that departed from
the duller ancestral pattern, resulting in the evolution
of brilliant male coloration and striking cases of
sexual dimorphism.
Though Darwin clearly perceived bright male
coloration as being due to female sexual selection,
he also discussed some evidence contrary to his view.
He mentioned cases of matings to 'battered, faded
or dingy males' (see also Ch.21) but dismissed these
as being a result of protandry.
Wallace (1889) believed that the sexual dimorphism
exhibited by many butterflies was often due to the
acquisition ofprotective coloration by females, rather
than to the development ofbrilliant colours in males
by intersexual selection. However, the diversity of
brilliant patterns of the males, and the similarity of
closely-related females, argue strongly against this
view, suggesting that protective coloration represents
an ancestral character state that existed before the
evolution of brilliant dorsal wing displays. Wallace
felt that 'the varieties ofcolour and marking, forming
the most obvious distinction between allied species,
... have ... in all probability been acquired in the
process ofdifferentiation for the purpose ofchecking
the intercrossing ofclosely allied forms.' (It is difficult
to understand how Wallace reconciled this "need for
recognition" with his denial offemale choice.) Both
Darwin and Wallace argued that bright colours were
not due to any special influence oftropical conditions.
In those few instances in which the females have
the more brilliant patterns, Darwin suggested that
the males prefer to mate with the more 'beautiful' fe-
males. When the sexes are alike, either the male has
retained theancestral colourpattern, or the malecolours
have been 'transferred' to the female (see Ch.23).
Darwin also recognized that mimicry affords a
special explanation for the colours of many butter-
flies. When mimicry is limited to onesex, it is expressed
in the female: the non-mimetic males generally retain
the colour pattern typical of their phyletic group,
from which the females depart. Darwin accepted
Thomas Belt's explanation for this phenomenon,
'that the females had a choice of mates and preferred
those that retained the primordial appearance of the
group' (Belt 1874, see also Turner 1978).
Summarizing, Darwin (1874) said
it is impossible to admit that the brilliant colours of
butterflies, and of some few moths, have commonly
been acquired for the sake of protection. We have seen
that their colours and elegant patterns are arranged and
20. Visual Communication and Sexual Selection 209
exhibited as if for display. Hence I am led to believe
that the females prefer or are most excited by the more
brilliant males; for on any other supposition the males
would, as far as we can see, be ornamented to no
purpose. With butterflies we have the best evidence,
as the males sometimes take pains to display their
beautiful colours; and we cannot believe that they
would act thus, unless the display was of use to them
in their courtship. Judging from what we know of the
perceptive powers and affections of various insects,
there is no antecedent improbability in sexual selection
having come largely into play; but we have as yet no
direct evidence on this head, and some facts are opposed
to the belief. Nevertheless, when we see many males
pursuing the same female, we can hardly believe that
the pairing is left to blind chance-that the female
exerts no choice, and is not affected by the gorgeous
colours ... with which the male is adorned' (my
emphasis).
The Evidence
Several kinds of direct evidence bear on Darwin's
conjecture. For instance, we need to ask of butterfly
vision: Over what spectral range do they see?
Can they discriminate colours from one another?
Have they a well-developed ability to distinguish
among patterns? What other factors (e.g. depth
perception, motion, polarization) are important in
their vision?
Which characteristics of butterfly colour patterns
might be used for visual communication? How are
the mechanisms of colour production related to the
qualities of visual stimuli? Which features are
constant, and which variable, within and between
species? Do male and female butterflies behave
during courtship in a manner consistent with the use
of optical signals? Do males rely on visual stimuli
to choose among females? Do females refuse, and so
choose among, males? If so, do they respond to the
males' colours?
Butterfly Vision
When physiological methods uncover color receptors
in the compound eye, their biological significance in
a color vision system is usually assumed, even in the
absence of behavioral data
(Goldsmith & Bernard 1974)
Spectral sensitivity
The spectral range visib. Ie to butterflies extends from,
the ultraviolet (Lutz 1924) through the red
(Eltringham 19 I9, Schlieper 1928), fully encom-
passing the visible spectrum of humans as well as
that of other insects. It is the broadest visible
spectrum known in the animal kingdom (Silberglied
1979).
Colour vision
It has long been supposed that butterflies are sensitive
to colour (Eltringham 1919). Flower constancy
exhibited while feeding, and responses of males to
dead specimens and coloured dummies, strongly
suggested that they could distinguish hues. Simple
choice experiments (e.g. Lutz 1924; see McIndoo
1929) also appeared to reveal colour preferences, but
none of the early experiments controlled for
differences in stimulus intensity (brightness). It
remained for Ilse (1928; Kuhn & Ilse 1925) to
demonstrate that true colour vision (hue discrimi-
nation) exists in butterflies. Additional behavioural
studies (e.g. Ilse 1937, Tinbergen et al. 1942,
Tinbergen 1958, Crane 1955, Ilse & Vaidya 1956,
Magnus 1958a, Swihart 1969, 1970, 1972b, C. A.
Swihart 1971, Swihart & Swihart 1970) reveal colour
preferences that may change with age and reproduc-
tive state, context, or learning.
Neurophysiological studies (it is extremely
frustrating to find that several neurophysiologists
rarely mention the sexes of the individual animals
they so painstakingly study) have confirmed and
extended these behavioural findings. Electroretino-
grams, single-cell recordings, optomotor responses
and other data reveal broad sensitivity from less than
300nm in the UV through 700nm in the red, and
more than one type of colour receptor in both
butterflies and skippers (e.g. Swihart 1963, 1964,
1967a, 1970, Post & Goldsmith 1969, Schumperli
1975, Bernard 1979). Swihart (I969) has modelled
colour vision as a dichromatic system in a skipper,
and 'at least a trichromatic system' in a Papilio, a
Heliconius and a Morpho (Swihart 1970, I 972a, b),
but since his experiments (and those of Schumperli
1975) do not include the UV portion of the spectrum,
such conclusions and models must be considered
premature and incomplete. It is probable that more
than three colour receptors are present in some
butterflies, as in some moths (e.g. four in Spodoptera
exempta; Langer et al. 1979). The complexities of
receptor structure (Bernard & Miller 1970), function
(Swihart 1973) and integration (Swihart 1965, 1968,
1970, 1972b) are such that a full understanding of
the neurophysiological basis of butterfly colour vision
is still a long way off. (See also Goldsmith & Bernard
1974, Horridge 1975, Mazokhin-Porshnyakov 1969,
Menzel 1975.)
Building upon the foundation ofCrane's work with
Heliconius, Swihart (I963, 1964, 1965) studied the
integration of visual information in the central
nervous system of several species (Swihart 1967a).
He had found that 'the behavioral sensitivity of
[H. erato] seems ... to be related not to any
modification of the [visual] receptors, but rather to
the development of [neural] pathways which
"selected" the output of those receptors which
210 Robert E. Silberglied
transduced information with special biological
significance.' His findings generated the hypothesis
that 'there was a selective advantage in developing
neural mechanisms which demonstrate dispropor-
tionate sensitivity to the basic wing coloration,
presumably because of the role played by such colors
in releasing courtship behavior.' Studies on six
species resulted in the discovery ofa match between
the spectral reflectance curve of the butterflies' wings,
and the spectral efficiency curve for summated
responses of high-order neurons in the medulla
interna (Swihart 1967a). For example, the red and
black Heliconius erato was found to have a peak
response in the red, the green Philaethria dido and
Siproeta stelenes a peak response in the green, and
the iridescent blue Morpho peleides to have a maximal
response to blue. He concluded that 'butterflies
possess a neural mechanism which "selects" the
output from various receptors in such a manner as
to make the visual system respond maximally to
stimulation with colors approximating the wing
pigmentation.' It thus appears that colour preferences
are effected by means of stimulus filtering in the
butterfly brain. At the receptor level, contrast may
be enhanced for biologically meaningful colours by
selective back-reflection of narrow spectral regions
from laminar tapetal interference filters (Bernard &
Miller 1970, Bernhard et al. 1970, Miller & Bernard
1968, Ribi 1980).
Spatial resolution
The spatial resolution ofa butterfly's eyes is related
to the angular separation between ommatidia.
However, the correspondence between interom-
matidial angles, and neurophysiological and
behavioural measurements of resolution, is limited
by complexities of ommatidial optics and receptor
physiology (Goldsmith & Bernard 1974, Palka &
Pinter 1975, Wehner 1975). In general, butterflies
have a resolution ofseveral degrees ofarc, hundreds
of times coarser than humans (O.5min ofarc). From
a behavioural standpoint, the amount of information
obtained per solid angle of view is ofgreater interest
than the simple angular resolution. Because the
former is a squared function of the latter, the spatial
information-gathering ability ofbutterfly vision may
be tens of thousands of times lower than that of
humans and other vertebrates. It is unlikely that
pattern details of small objects, such as other
butterflies, are resolved until they are extremely close
(see Eltringham 1919, Mazokhin-Porshnyakov 1969,
Yagi & Koyama 1,963). Little is known about central
processing of spatial information in butterflies
(Swihart & Schumperli 1974, Schiimperli 1975,
Schumperli & Swihart 1978), but visual field
organization and hierarchy generally resembling that
of vertebrates has been found in some insects
(Schumperli 1975, Wehner 1975).
The angle of view of butterfly eyes is very large,
accounting in part for the difficulty of approaching
alert species from any direction. One might think
that the dorsal colours of a butterfly would not be
visible to another that was below or in front of it
(Wallace 1889). But the wide angle of view, the
lowering of the wings to a nearly vertical position
on the downbeat (Silberglied & Taylor 1978), and
the forward projection of certain structural colours
(Darwin 1880) make it likely that one butterfly can
see the dorsal surface ofanother from most positions
during flight. While there is little direct experimental
evidence for depth perception in butterflies, a
neurophysiological basis has been found in numerous
protocerebral cells containing input from both eyes
(Schiirnperli 1975).
The meagre behavioural data on pattern and spatial
discrimination indicate that ovipositing female
butterflies are capable of visually discriminating
between leaves ofdifferent shapes (Rauscher 1978).
They possibly detect eggs visually (Gilbert 1975;
Ch.3) as well as chemically (Rothschild &
Schoonhoven 1977, Ch.6). However, male
Hipparchia butterflies do not discriminate among
dummy butterflies ofdifferent shapes (Tinbergen et
al. 1942, Tinbergen 1958), and pattern details play
little role in species and sexual discrimination
(Tinbergen et al. 1942, Tinbergen 1958, Crane 1955,
Stride 1957, Magnus 1958b, Obara 1970, Rutowski
I977a) although there do appear to be some
exceptions; Papilio xuthus (Hidaka & Yamashita
1975, 1976) and Limenitis camilla (Lederer 1960).
In general, it does not appear that shape information
is particularly important in the context of butterfly
communication.
Temporal resolution and motion detection
The temporal resolution of the eye is generally
measured in terms of the frequency at which
successively presented images fuse. Argy>mis paphia
has been shown to have flicker fusion at about
150 images per second (A. Muller in Magnus
1958a,b), compared with about 40 per second in
humans. It has been suggested that, for a moving
insect or subject in view, the information gained from
more frequent sampling of the visual environment
may compensate to some extent for low spatial
resolution.
A butterfly can detect a moving object if it differs
in colour from the surrounding field, an ability that
is greatly enhanced if a difference in brightness is
also present (Kaiser 1975).
Polarization
Arthropod photoreceptors exhibit a general
sensitivity to polarization, used at the behavioural
level for orientation (von Frisch 1968). The iridescent
reflection from the wings ofsome butterflies should
20. Visual Communication and Sexual Selection 211
be strongly polarized at certain angles. To my
knowledge, this phenomenon has not actually been
demonstrated, and its possible role in visual
communication is unknown.
Non-imaging photoreceptors
Such receptors might also play a role in
communication. Adult butterflies lack ocelli, but
extra-ocular photoreceptors have recently been
reported on the genitalia ofboth sexes in a small but
diverse array of species (Arikawa et al. 1980). This
'anatomical provision for hindsight' (C. M. Williams)
may serve some function during copulation.
Butterfly Colour Patterns
The colours and patterns of butterfly wings have
been responsible for their great popularity among
scientists, collectors and the general public. Attention
has been given to the nature and synthesis of
pigments, the mode of production of structural
colours, the development and genetics ofwing colour
and pattern, and the role of visual signals in
communication with vertebrate predators. However,
the signal properties ofbutterfly colours and patterns,
as related to communication, are as yet poorly
understood. It is not intended to provide a
comprehensive review of this formidable subject,
which has been reviewed in general terms by
Hailman (1977) and Hamilton (1973). Rather, a few
features of butterfly coloration especially pettinent
to the topic of this paper will be discussed. Readers
are referred to the voluminous literature on butterfly
coloration for reviews of other aspects (e.g. Crane
1954, Graham 1950, Nijhout 1978, 1981,
Papageorgis 1975, Poulton 1908, Robinson 1971,
Schwanwitsch 1924, Silberglied 1977, 1979, Turner
1977a, 1978, Vane-Wright 1975, 1976, 1979a).
Ultraviolet patterns
The human observer describes butterfly colours in
terms of our system of colour nomenclature, which
is inadequate for the task because butterfly vision
includes an ultraviolet (UV) component. With some
exceptions, diurnal terrestrial vertebrates do not see
UV, so this spectral region may provide a 'private
channel' for communication among insects
(Silberglied 1979). A butterfly's wings may reflect
little or much UV; as with 'visible' colours the
reflection may be produced by scattering in the
absence of UV-absorbing pigments, or by physical
means. A great variety, of UV patterns exists, even
in 'visibly' concolorous groups such as pierids and
satyrines, and striking differences in UV patterns
between otherwise similarly coloured species occur
in many genera (e.g. Colias, Silberglied & Taylor
1973, 1978; G01zepte1yx, Mazokhin-Porshnyakov
1957, Nekrutenko 1968; Phoebis, Allyn & Downey
1977; Pieris, Bowden 1977; Prepona, Descimon et
al. 1973-74) (cfWynne-Edwards 1962: 34). Sexual
dimorphism is more pronounced in the UV than in
the 'visible' spectrum; local polymorphism, geo-
graphical variation and other colour phenomena also
occur (Silberglied 1979). Since insect compound eyes
are maximally sensitive in the UV, as much attention
should be paid to UV patterns as potential sources
of behaviourally meaningful signals, as to 'visible'
reflection patterns. Furthermore, since vegetation
generally absorbs UV, reflection in this region serves
to maximize colour contrast in terms of insect vision.
Iridescence
Iridescence is usually defined as 'exhibiting a
rainbow-like display of colours'. Most cases of
iridescence in butterflies are attributable to reflection
from multiple thin-film interference filters present
in the outer layer of wing scales (Mason 1926-27,
Anderson & Richards 1942, Simon 1971). In most
such cases, including spectacular examples such as
M01pho and many lycaenids, the range of reflected
colour lies in the blue-to-violet end of the 'visible'
spectrum; further investigation usually reveals
intense UV reflection as well (Ghiradella et al. 1972,
Ghiradella 1974). Due to the mechanism of colour
production, such colours change in hue and intensity
as the angle of the wing changes during flight (Crane
1954). At any patticular angle, the reflection is of
very high spectral purity and thus unlike other
reflected light in the terrestrial environment. To an
observer watching through a narrow spectral window
(e.g. a television camera with a narrow-band filter;
Eisner et al. 1969) or with species whose reflection
barely enters the 'visible' spectrum (e.g. Apatura
spp.), these butterflies are conspicuous at con-
siderable distances as their wing reflection flashes
'on' or 'off'. The reflection may also be partially
polarized at certain angles.
Such iridescent patterns are widespread among
butterflies, especially in the three largest families,
Lycaenidae, Nymphalidae and Pieridae. In pierids
and some other groups, these spectacular patterns
of iridescence are nearly always totally invisible to
the human eye, as the reflection does not pass beyond
400nm into the violet. Iridescence, whether 'visible'
or UV, is largely restricted to the dorsal wing
surfaces, and is far more strongly developed in males.
Light colours
A large number of species are light in colour; nearly
all of the Pieridae, for example. White and other light
colours also provide contrast with vegetation, and
the suggestion has been made at various times that
white may be aposematic (Wallace 1889, Kettlewell
1965, T. Eisner, pers. comm., C. L. Remington,
pers. comm.). In light coloured butterflies, the ventral
wing surfaces are usually darker and bear cryptic
212 Robel'! E. Silberglied
patterns and colours. Where sexual dimorphism
occurs among light-coloured species, the males
usually have a uniform field of colour, while that of
females is infuscated or obscured by other markings.
Orange and red
Orange and red patterns must be considered to have
potential signal value, since butterflies, unlike other
insects, can see these colours. Red has been
demonstrated to have a communication function in
He/iconius (Crane 1955).
Sexual dimorphism
When the sexes differ in coloration, females usually
have a protective pattern, either more cryptic or more
mimetic than males. The male pattern is often visible
from a greater distance than that of the female, even
in cases where the latter is mimetic ofan aposematic
unpalatable model. Female coloration also appears
to be more variable, as has been shown quantitatively
in moths (Fisher & Ford 1928).
Variation and polymorphism
In species with local polymorphism, either both sexes
are polymorphic, or only the females. Male-limited
polymorphism is virtually unknown in butterflies
(Brower 1963, Turner 1978). Some species are
polymorphic on a seasonal basis. The males of
polymorphic species cannot use a single, narrow,
visual searching-image for females. When both sexes
are polymorphic, females likewise cannot use a single
colour pattern to recognize conspecific males. Where
geographic or seasonal variation or polymorphism
occurs, there must be some compensation in the use
of visual signals, by corresponding variation in the
signal-receivers' behaviour (Burns 1967), a broad
range of, or multiple, acceptable signals, constancy
of the signal component (Rutowski 1981), or less
reliance on visual signals for communication
(Richards 1927, Brower 1963).
Mimicry
Mimetic species cannot use visual signals exclusively
for communication, or confusion with similarly-
coloured species would result. In sex-limited
mimicry, the male cannot use visual signals alone
for the recognition ofconspecific females, but females
could use them to recognize males. Brower (1963;
see also Ch.25), in a brief but seminal discussion of
sex-limited mimicry, suggested that in Mullerian
mimicry where both sexes are mimetic, 'scent plays
a more important role in courtship than sight.' It is,
alternatively possible that mimetic species use
distinctive patterns of UV reflection for sex- and
species-recogition (Remington 1973, Silberglied
1979).
A wide range of problems involving the use of
visual signals by variable, polymorphic or mimetic
species awaits investigation. For example, what visual
signals are employed by male Papi/io dardanus, and
do these signals vary geographically with the female
morphs? If red coloration is critically important in
He/icollius erato, what signals are used by the yellow
and blue Colombian H. e. chestertonii, the only
subspecies that lacks red markings (see Turner
1975a)? How do similarly coloured sympatric
congeners (e.g. Adelpha spp.) tell one another apart?
Butterfly Courtship
The courtship of butterflies is a complex ritual,
involving the exchange ofvisual, chemical and tactile
stimuli. While it differs in detail from one species
to another, a general sequence of behaviour is
followed by most species (Scott 1973d, Hidaka 1973,
Silberglied 1977).
Courtship is usually initiated by the male. Males
locate females by using one or both of two behaviour
patterns: 'waiting' for females in locations where they
are likely to be encountered, or 'seeking' females by
active, persistent search of the habitat (Magnus 1963;
these behaviours have also been named 'perching'
and 'patrolling' by Scott 1973d, 1974b). Both types
of behaviour may be found within a species (Dennis
1982), or in an individual at different times (Davies
1978).
Actual courtship begins in response to the sight
of the female. It is easy to study the initial approach
behaviour of males, by using 'dummies' of various
sizes, shapes and colours (e.g. Crane 1955, Magnus
1958b, Petersen et al. 1952, Petersen & Tenow 1954,
Silberglied & Taylor 1978, Stride 1957, 1958a,
Tinbergen et al. 1942, Tinbergen 1958). The visual
stimuli required to attract a male are surprisingly
unspecific in many species. Movement, colour and
size are usually important, while shape and detail
of pattern are not (but cf. Lederer 1960, Hidaka &
Yamashita 1975, 1976). Males of some species fly
after anything that might be another butterfly,
including falling leaves, birds, or a moving butterfly
net. Further male behaviour appears to depend on
the response of the chased object-other males may
be chased away, birds avoided, and females courted.
In Cohos, only a stationary visual stimulus is
needed to release the entire sequence of male
courtship behaviour, so that males may actually
copulate with paper dummies (Silberglied & Taylor
1978). In other species (ArgYll/lis; Magnus 1950,
1958b), the male requires motion and olfactory cues
before continuing his courtship.
The later stages of courtship depend on the
receptivity of the female. In those species carefully
studied to date, olfactory signals released by the male
are usually required for successful mating (e.g.
Brower et al. 1965, Lundgren & Bergstrom 1975,
Magnus 1950, Pliske & Eisner 1969, Rutowski
20. Visual Communication and Sexual Selection 213
1977b, 1978a, 1980a, Stride 1958a, Tinbergen ec at.
1942, Tinbergen 1958; cf. Pliske 1975b; for reviews
see Scott 1973d, Silberglied 1977 and Ch.25). The
more elaborate behaviour associated with courtship
appears at this stage, when the male disseminates his
odour(s) in the vicinity of, or directly upon, the
female's antennae. Depending on the species, the
male may dust her antennae during flight or while
she is resting (Danaus gi/ippus; Brower ec at. 1965),
rub her antennae with one or both wings (Tinbergen
eC at. 1942, Temple 1953), or perform other
specialized, species-specific behaviours (Crane 1955,
1957).
Most females encountered by males have already
mated, and in most cases mated females respond
negatively towards males. This negative response is
characterized by 'rejection postures', 'ascending
flights', and other behaviours which vary in detail
from one species to another (Scott 1973d, Stride
1958b; see also Rutowski 1978a,b, Edmunds 1969),
Chovet 1983). Males usually persist for a while in
spite of the rejection posture, and a female that
initially rejects a male may later accept him (Petersen
& Tenow 1954).
In the final stage of courtship, the male positions
himself alongside the female and probes laterally and
ventrally along her wings with his abdominal tip, in
an attempt to locate and couple with her genitalia. In
most species, there is no visible indication that the
female has accepted the male, other than absence of
rejection posturing. In a few cases the female may
be seen to adjust her position and lower her abdomen
(Rutowski 1977b).
In some species, a receptive female may initiate
courtship with a 'solicitation' flight towards a male
(Crane 1955, Stride 1956, 1958b; Ch.21, Lederer
1960, Rutowski 1980b, 1981). Courtship may also
be said to be initiated by the female in those few
species where a sex-attractant pheromone is used (see
also Ch.25). Such cases sometimes involve mating
by the pharate or teneral female. Facultative mating
with teneral females is common in many butterflies;
in these circumstances the female may be unable to
reject a male and prevent insemination (Taylor 1972).
In most species, the female becomes unreceptive
after mating, and rejects subsequent matings for some
time. Because the number of matings by a female
may easily be determined by counting the number
ofspermatophores in the bursa copulatrix, substantial
data exists on female mating frequency (Burns 1968;
Ch.22). In general, females mate at most only a few
times; large numbers of matings are very unusual
(Ehrlich & Ehrlich J978). Thus most mating
attempts by males are not successful. Unsuccessful
courtships may result from general female
unreceptivity, female coyness or female preference
(Ch.21).
While mating behaviour is usually initiated by
males in response to visual stimuli, subsequent
courtship behaviour is more complex. The male may
require olfactory stimulation or he will not persist,
and the female generally requires male olfactory
stimuli before accepting (Ch.25). Tactile stimuli may
be exchanged, especially in the later stages just prior
to and during copulation. The role of visual
stimulation during the later phases of courtship is
poorly understood. In particular, the question of
whether the female 'chooses' the male (or the male
'chooses' the female) on the basis of visual stimuli
is not easily answered (see also Ch.21).
Visual Signals affecting Male Behaviour
The responses of male butterflies to visual stimuli
have been studied in virtually every experimental
investigation of butterfly courtship behaviour.
Generally, dead specimens or 'dummies' have been
used as stimuli, and the number of approaches by
males counted as a measure of attractiveness. Such
experiments are simple to perform, and the results
have generaIIy been consistent from one such study
to another. One of their drawbacks is that little
information is gained about the response of males
to colour and pattern in later stages of courtship.
Another problem with 'approach tests' is that the
males' behaviour may easily be misinterpreted: males
may be attracted to stimuli for reasons other than
courtship (e.g. feeding, aggression, roosting,
puddling, etc.). Some of these difficulties are
circumvented by studying the behaviour of males
toward living females in flight cages, or on tethers
(e.g. Crane 1955, Brower 1959, Rutowski 1981,
Silberglied & Taylor 1978).
Another method used to assess the importance of
visual stimuli to males, compares mating frequencies
of female colour phenotypes in polymorphic species
relative to the frequencies of these morphs in the
population (e.g. Smith 1973, Burns 1966, Pliske
1972; Chs 21, 22). Such data are obtained from
mating pairs collected under field conditions, or from
spermatophore counts in wild-eaught females of the
various morphs. While of great value in documenting
assortative mating or sexual selection, this work does
not necessarily identify the colour patterns as the
signals used by male butterflies in mate choice. The
correspondence between colour phenotypes and
mating success may also be due to differences
between female morphs in vigour, behaviour
(coyness), pheromone production, longevity, or other
properties correlated with or genetically linked to
pattern loci.
In general, males are attracted to visual stimuli
coloured most like the conspecific female (but see
also Ch.23). The detailed pattern of the butterfly
need not be duplicated in the more attractive
dummies; only colour and UV reflection need
214 Roberl E. Silberglied
resemble the female (e.g. Magnus 1963, Stride 1956,
Obara 1970, Obara & Hidaka 1968, Silberglied &
Taylor 1978). Often the ventral wing colour, exposed
by the female when at rest, is most attractive. Both
saturation and intensity of the colour make important
contributions to a dummy's attractiveness. It has
sometimes been possible to make artificial dummies
that were more attractive than female specimens
(Magnus 1958b, Silberglied & Taylor 1978).
In sexually dimorphic species, the colour of the
male may be delen'em to other courting males, rather
than neutral or attractive (but see also Ch.23). This
was first discovered by Stride (1956, 1958b), who
added the white of male Hypolimuas misippus wings
to otherwise attractive female dummies. Males
initially attracted to some of these dummies departed
immediately, rather than persist in courtship. It is
significant that the most pronounced deterrent effect
was produced when he used white pieces of males'
wings (rather than white areas from another species
or an artificial material), which are intensely UV
reflecting and iridescent (see also Ch.21 concerning
this work). Rutowski (1977a) found that the UV flash
of sitting, fluttering male Eurema lisa inhibited
further courtship attempts by other males; courting
males did not distinguish males lacking UV from
females. In Colias, Silberglied & Taylor (1978) found
that simply adding UV-reflection to an attractive
dummy produced a 30- to 50-fold decrease in its
attractiveness. They also found that mating males
of C. eurylheme and C. philodice expose their
hindwings when approached by another male; in C.
eurYlheme this action produces a sudden flash of
intense UV-reflection that is surely repellent to the
intruder.
In some species, male coloration elicits agonistic
behaviour in other males. The deterrent or
intimidating effect, of male colour on the behaviour
ofother males, is probably 'turned off' (or interpreted
differently) during puddling behaviour, a
predominately male activity (Arms el al. 1974).
Where studied, motion has been shown to
significantly increase the attractiveness of a dummy
(e.g. Tinbergen el af. 1942, Crane 1955, Magnus
1958a,b). In Magnus' classic experiments with
A 'gyuuis paphia, motion of the dummy was necessary
to hold the attention of the male. He found that more
and more effective super-normal stimuli could be
created by increasing the degree of flicker in the
dummy's apparent motion, until the flicker-fusion-
frequency of the butterfly's eye was reached.
Size is another important stimulus. A graded
response to size 1)as usually been found, with low
responses to dummies of the appropriate colour
smaller than the female, good responses to dummies
of the same size, and super-normal responses to giant
dummies (Tinbergen el al. 1942, Tinbergen 1958,
Silberglied & Taylor 1978). Males do not appear to
discriminate against dummies shaped differently
from the female, unless the differences assume
extreme proportions (Tinbergen el al. 1942).
The details of wing-pattern usually do not
contribute to the attractiveness of a dummy, except
where the pattern affects the saturation or intensity
of the entire wing. Thus the minor sexual differences
of pattern found in genera like Colias (light spots in
the dark borders offemales), Pieris, etc. probably do
not matter to males, at least in the initial attraction
(Rutowski 1981).
Ethological studies of species with female
polymorphism have sometimes demonstrated male
preference for dummies resembling one of the
morphs relative to others. Yellow females of Pieris
bryoniae are less attractive than white ones (Petersen
1952). Magnus (1958b) found that the dark female
form 'valesina' was less attractive to male Algymzis
paphia than the normal phenotype. Extrapolating
from the result that white is inhibitory to male
Hypolimllas misippus, Stride (1956) tried to account
for the rarity of its 'alcippoides' form (but cf.
Edmunds 1969b, Smith 19760, Ch.21). Burns (1966,
1967) used spermatophore counts from field-collected
females to argue that male preferences maintain the
non-mimetic female morph in Papilio glauCIIs (but
cf. Prout 1967, Pliske 1972, Levin 1973, Barrett
1976; Ch.22). MaleAllartiafalima are more attracted
to dummies with white bands than to those with
yellow bands; in this species the colours are not under
genetic control, but change gradually from yellow
to white with age (Emmel 1972, 1973a, cf. Taylor
1973b, Young & Stein 1976, Silberglied el af. 1979).
The data and interpretation ofseveral of these studies
have been subject to dispute. I will discuss (below)
only the relevance of dummy experiments to male
mating behaviour under natural conditions.
Visual Signals affecting Female Behaviour
Literally not one particle ofevidence [exists J that the
female is influenced by colour.
(Wallace 1877)
There is no proof to date for the assumption that any
visual selection in favor ofnormally colored males takes
place by butterfly females.
(Magnus 1963)
In contrast with the numerous studies of the role
of visual stimuli in male courtship behaviour, few
attempts have been made to obtain similar
information about females. There are many reasons
for this. Male butterflies are aggressive in initiating
courtship attempts, but females usually are initially
passive or preoccupied with other activities. Any
study of female courtship behaviour requires sexually
active males. Second, 'determined' approaches can
be used to bioassay attractive visual stimuli for males,
but such tests can rarely be performed with females.
20. Visual Communication and Sexual Selection 215
The coy responses offemales are such that only actual
copulation (or in rare cases where it is recognizable,
solicitation flights or acceptance behaviour such as
exposing the abdominal tip; Rutowski 1977b, 1981)
can be used to compare the relative acceptability of
various male phenotypes. Because females become
unreceptive after mating, each can be used only once,
so many virgin females are required. Third, many
species have polymorphic females, and the
experimenter may use the various female forms, dead
specimens or even paper dummies as stimuli for
courting males. But since males are rarely
polymorphic, and because active males are required
to test female responses, the investigator must modify
living males with various experimental treatments
to produce different colour or pattern phenotypes.
Fourth, most experiments involving choice by
females, berween males from different groups, must
be conducted in cages to keep the butterflies from
dispersing, but many species will not behave
normally in such an environment. Only those
individuals that do not spend all their time trying
to escape can be used, so large numbers must be
reared to produce a cage-adapted population. Fifth,
since mating behaviour changes with age, all
experimental groups ofeach sex must be 'balanced'-
they must have the same age distribution. Sixth, I
have found that in nearly every experiment of this
kind, untreated males mated more frequently than
did treated control males of the same colour. This
is probably due to minor injury of the latter during
handling and treatment (see also ChA). A comparison
between experimentally colour-modified males, and
normally coloured, but untreated control males, is
generally meaningless. Both untreated and treated
control groups, as well as one or more experimental
groups, should be included in the design of such
experiments. For these and other reasons, few have
attempted to perform them.
Crane (1955) and Rutowski (1981) used 'angling'
techniques to assay for female responses to tethered
males and artificial dummies. This method can
succeed only if there is a visible bioassay for female
acceptance behaviour.
Field data, on assortative mating in polymorphic
species, are of limited value in determining the
importance of male colour to females. Non-random
mating and/or sexual selection are not necessarily due
to visual discrimination and choice, even when there
are coloration differences between phenotypes. While
male phenotypes may differ in their relative
representation in mated pairs, such a result may arise
because of differences 'between morphs in thermal
properties, vigour, courtship activity, competitive-
ness, pheromone production, longevity, etc. The
rarity of male polymorphism is a serious practical
limitation to the use of this technique, but it has been
employed by Smith (1975c, 1980, 1981, Ch.21) in
Danaus chrysippus. The method may also be used
in species like Chlosyne lacinia, in which several
colour forms occur sympatrically in both sexes. Field
mating data may suggest that visual discrimination
is occurring, but only experimental studies can test
for the operation of visual signals.
Given these limitations, it is not surprising that
data on female colour preferences are few, and needs
to be augmented by additional studies of other
species. Anyone with large cultures of living
butterflies has the opportunity to test the ideas
presented below. Experimental studies of the role
of colour in mate selection in 'visibly' dimorphic
species are particularly desirable.
Crane (1955) was the first to attempt the
experimental study of the role of colour in female
acceptance behaviour. She painted both sexes of
Heliconius erato various colours, and observed the
behaviour of other butterflies towards them. She
claimed that 'the farther the altered color of the
forewing from the normal scarlet in the spectrum,
the less notice is taken of the butterfly, either by the
opposite sex or as a general subject for social chases.'
But she also remarked that 'positive courtship
responses, sometimes including copulation, were
obtained at least once for each color change effected
in each sex', and that 'females were in general less
influenced [in behaviour1by color change than were
males.' Her discussion of this experiment includes
no indication ofsample size for each colour beyond
'at least one butterfly of each sex', and she did not
include treated control individuals. She describes
difficulties with variables of weather and the
physiological state of the butterflies, and the reader
is presented with a generalized summary of the
results rather than quantitative data. It is not clear
that social interactions were fully separable from
sexual behaviour. Finally, most ofher discussion of
the role of colour deals with the responses of males.
Another series ofexperiments performed by Crane
(1955) used cloth dummies angled from bamboo
wands. The single female tested responded 'positively'
to red and orange, but minimally or not at all to other
colours. The 'female's responses were ... gauged
from the strength of her courting behavior, shown
by the degree of abdomen elevation, extrusion of
yellow gland, apposition offorewings and fluttering
ofhindwings.' But this is a description offemale re-
jection behaviour (L. Gilbert pers. comm., J. Mallet
pers. comm.). One may not conclude that female
H. erato discriminate between males on the basis of
colour.
Stride (1958b) studied the responses of female
Hypolimnas misippus to two kinds ofmodified males.
Two 'black' males had their white spots blackened
chemically, a treatment that did not affect the
UV-blue iridescence. Three 'colourless' males had
most of their wing scales removed. Males were tested
216 Robert E. Silberglied
individually in a large flight cage containing six virgin
females in each test. None of the 'colourless' males
mated successfully (18 courtships), while both of the
blackened males were successful (three courtships).
All four untreated control males mated (four
courtships). Unfortunately, Stride did not include
a treated control group, so there is no way to evaluate
the traumatic effect of scale removal. He did not
control for odour (Brower 1963), so the results may
have been affected by disruption of a chemical
communication system involving wing-born phero-
mones (cf. Silberglied & Taylor 1978, Rutowski
1980a, Grula et al. 1980). His sample size is too small
for statistical analysis. Nevertheless, the all-black
males mated successfully, so one can probably agree
with Stride's conclusion that 'the white spots on the
wings of the males played little part in the courtship
of the butterfly.' There was no female selection
against males lacking the most conspicuous 'visible'
pattern element.
Rutowski (1981) used his 'angling' technique to
study solicitation flights of female Pieris protodice.
By removing UV-absorbing pterins from the wings
of males, he produced dummies that resembled
females. He found that females use the UV-
absorption of males as a sexual recognition signal.
No colour modifications were performed in the
'visible' spectrum.
In Dallalls chrysipplls, a species in which both sexes
are polymorphic, males heterozygous at a locus
affecting colour pattern have a significantly higher
mating success than either homozygote (Smith 1981,
Ch.21). However, the minor colour differences
between the heterozygote and one ofthe homozygotes
are probably too small, in my opinion, to be acted
upon by visual selection. Smith partly attributes the
difference in mating success to inter-male
competition, mediated through the interaction of
climatic conditions and the thermoregulatory
properties of these colour patterns.
Silberglied & Taylor (1978) reported an extensive
series of experiments on two species of Colias. Their
goal was to identifY the basis ofconspecific assortative
mating by females. The experiments, performed in
flight cages, involved large numbers of receptive
females of both species, both untreated and treated
control males as well as experimental groups, and
balance for age distribution.
Colias emytheme has orange, UV-reflecting males,
while male C. philodlce are yellow and UV-absorbing.
Males of the yellow species were coloured orange by
means offelt-tipped pens, and the males of the orange
species yellow, by transplanting the discal area of the
wings, of a yellow-winged, UV-reflecting strain of
hybrid origin, onto the corresponding area of
C. emytheme. Females were still fully discriminatory:
they mated conspecifically, and they accepted the
peculiarly-coloured conspecific males as frequently
as they did the normally-coloured treated control
males of their own species. In another experiment,
males C. eurytheme (Colour plate 4A) and C. philodlce
(Colour plate 4B) dyed orange, red, green, blue and
black mated as frequently and as conspecifically as
did treated control individuals!
The only visual component, the modification of
which had any effect on female mating behaviour,
was the ultraviolet reflection of C. ellrytheme.
Regardless of their 'visible' colour, males of this
species whose UV reflection had been destroyed,
suffered a significant drop in the number of
successful conspecific matings. Yet they were not
accepted by C. philodice females, even though the
males of C. philodice are UV-absorbing, and even
though C. philodice females accept their own males
in any colour. In short, the females of C. ellrytheme
responded only to the UV component of wing
coloration, and the females of C. philodice acted as
though they were totally blind to the male colours
or UV reflection!
The olfactory basis ofmate selection has since been
identified in these Colias species (Rutowski 1980a,
Grula et al. 1980). So far as female discrimination
is concerned, olfactory cues are far more important
to the females of these species than are visual signals
(Silberglied & Taylor 1978). It is significant that the
females of these species are polymorphic for colour,
while the males are not. This colour distribution
agrees with the idea of stronger female than male
sexual selection stabilizing coloration of the opposite
sex. But the experimental results directly conflict this
suggestion.
I have since carried out one additional experiment
on the role ofcolour in female choice, this time with
the brilliant red Neotropical species Allartia amathea
(Colour plate 4C). It was performed in collaboration
with Annette Aiello at the Smithsonian Tropical
Research Institute, and is published here for the first
time. The experiment meets all of the criteria
discussed at the beginning of this section.
Freshly-eclosed but hardened A. amathea were
used. Virgin females were mixed with equal numbers
ofmales, divided into three groups: untreated control
(red), treated control (red), and experimental (black).
Treatment involved painting the dorsal wing-surfaces
with clear (Ad Marker (R), Jacksonville, Florida:
Warm Gray I) or black (same brand, Super Black)
felt-tip pens. The butterflies were released into a
flight cage and subsequent matings recorded. Because
there were three groups for each sex, there were nine
mating combinations possible. The results are
presented in Table 20.1.
Only 21 matings were obtained, so the number in
each cell is very small. For both logical and statistical
reasons (the total number of matings in each cell is
the result of two phenomena: male attractionl
courtship of females and female acceptancelrejection
20. Visual Communication and Sexual Selection 217
Table 20.1. A mating experiment on sexual selection for colour in the brilliant red butterfly Anartia amathea
(Nymphalidae)
Females
Red
(untreated control)
Red
(treated control)
Black
(experimental) TOTAL
Males:
Red
(untreated control) 5 3 9
Red
(treated control) 3 2 0 5
Black
(experimental) 4 3 0 7
TOTAL 12 8 21
The experiment was performed in a 3m x 3m x 2.1m screened cage on Barro Colorado Island, Panama, with stocks collected
in the Darien region of eastern Panama, in 1976.
behaviour, and the small sample size in each cell),
the results in each of the nine cells cannot be
meaningfully compared with one another. However,
if we compare the number of matings by group
within each sex, a distinct picture emerges.
Notice first the number of matings by females in
each of the three groups. Black (experimental) females
did not mate as frequently as did either of the red
females (one-tailed binomial test of black vs. red
treated control females, P ~ 0.02; Siegel 1956). It
is highly unlikely that the acceptance/rejection
behaviour of females is changed by painting them
a different colour. The only logical conclusion is that
the males spent little time courting them compared
with the red females. This is entirely consistent with
what we know about the role of colour in male
behaviour.
On the other hand, notice that those females that
mated did not discriminate between the red and black
males (one-tailed binomial test ofblack vs. red treated
control males, P ~ 0.44) (or if they did, they might
have preferred the black!). Thus females did not
choose on the basis of the red colour - the most
conspicuous feature of the wings and the only colour
present. A devil's advocate might argue that they
prefer black, but this is hardly what we are concerned
with when we speak of the brilliant colours that have
been or could be so nicely explained by female sexual
selection. This brilliant red butterfly sees red
(Bernard 1979), and the males respond to red.
Because the male has the more brilliant red
coloration, it is only logical that it be the result of
female selection. Yet it is clear even from this minor
experiment that such i,s not the case.
Brower et al. (1971) have argued convincingly that
A. amatlzea is an 'incipient mimic' of Heliconius
erato/melpomone. This contention is supported by
their experimental data which show that predators,
having learned to discriminate H. erato, will also
sight-reject A. amathea. Thus the red coloration
might be 'explained' on that basis. But the male of
A. amathea has a more brilliant red and a darker
black, and would probably be a better mimic. Unless
the red of the male serves some other role, we have
a case in which the male is the better mimic than
the female, which would be a unique situation so far
as I am aware (Carpenter & Ford 1933).
Discussion
Is it possible, or wise, to attempt to make gener-
alizations about the use of visual signals for
communication among the 10 000-15 000 species of
butterflies? I believe it is, because they all have to
deal with the same problem: recognition of and
communication with conspecific individuals. Their
vision is constrained by the same solar spectrum.
Their use of colour is limited by biosynthetic
versatility, as well as by the activities of visually-
oriented predators. Diverse solutions to the problem
are found, including the discarding of visual
communication for many functions. But a general
picture of the use of visual signals by butterflies has
emerged from the confusing wealth ofspecies, colour
patterns and behaviour.
Before proceeding, it is well to emphasize that
exceptions will be found to such generalizations, and
that the exceptional cases, when studied in depth,
often serve to confirm, refute or qualify the
hypotheses generated. Since so few intensive studies
of butterfly communication have been carried
out, I cannot hope to convince - only to raise
doubt where it should exist, about the classical
interpretation of the role ofbutterfly colour patterns
in communication.
The two functions-protection from visually-
oriented predators, and social signals used during
218 Robert E. Silberglied
courtship-may at first seem to be mutually
exclusive, because the former often requires patterns
that are difficult to detect (crypsis) or are confusing
(mimicry), while the latter demands a high signal-
to-noise ratio and uniqueness with respect to species
and sex. But it should also be remembered that these
functions involve different visual systems: the
vertebrate lens-eye and the lepidopteran compound
eye. It is possible that some panerns serve both
functions by taking advantage of the differences
between these visual systems (Silberglied 1979).
Vision and Pattern
Butterfly vision is characterized by excellent colour
discrimination across an extremely broad visible
spectrum, low spatial resolution and high sensitivity
to motion or flicker. Colour contrast is increased in
particular spectral regions by physical devices
(e.g. tapetal reflection) and neurophysiological
mechanisms (stimulus filtering). The ability of the
insect's eye to detect a moving object is strongly
enhanced by brightness- as well as colour-contrast.
At the behavioural level, butterflies are capable of
discriminating quite subtle differences of hue,
intensity and saturation (e.g. the difference between
greenish-yellow and yellowish-green on the ventral
hindwings of male and female Colias; Silberglied &
Taylor 1978), and oflearning (C. A. Swihart 1971,
Swihart & Swihart 1970).
Hence the extremely brilliant patterns that
characterize the dorsal surfaces of so many male
butterflies probably are not required for
communication by any intrinsic limitations of the
visual system. For iridescent colours, the change of
hue with angle of the wing, narrow spectral
reflectance at any given angle, abrupt intensity
change with angle at any given wavelength, and
strong ultraviolet component (and possible
polarization) that contrasts strongly with vegetation,
are features that would serve well as signals for long-
range communication. Morpho can easily be seen from
low-flying aircraft; the gleam of their wings, as they
sail above the forest canopy, makes them appear as
giant, blue, flashing beacons. Describing M. rhetenor,
H. W. Bates (1864) reported that 'when it comes
sailing along, it occasionally flaps its wings, and then
the blue surface flashes in the sunlight, so that it is
visible a quarter of a mile off.' Where iridescence
is not involved, the flicker of the alternating dorsal
and ventral surfaces exposed by beating the wing,
often enhanced by colours that contrast well with
the green of vegbation, might perform the same
function. The patterns of females, being more
variable and less conspicuous, ought not to be as
apparent over the same visual distances as are those
of the males.
The mode of colour production in butterflies
allows for an overlay of transparent scales producing
structural colour over a base of pigmented scales. As
a result, these two types of coloration may evolve
and be expressed independently of one another. We
see a striking example of this in males of Hypolimnas
misippus and related species, where a patch of intense
ultraviolet reflection overlays both white and black.
In terms of insect vision, this produces a white bull's-
eye, surrounded by a ring of pure ultraviolet in a
field of black. While the pigments have not been
studied, only melanin or some other dark pigment
would be required to achieve this effect, because
white is produced by diffuse scattering, and the UV
reflection by a transparent interference filter overlay.
Thus, a remarkable range ofcoloration, with diverse
optical properties, is made possible by the
combination of pigmentary and structural colours.
A glance through a collection reveals the extent to
which these possibilities have been realized.
Thus we find two solutions to the problem of
providing increased or decreased brightness- and
colour-eontrast. Reflectance over much of the
spectrum may be increased (or decreased) (the pieridl
satyrine solution). Alternatively, a particular colour
may be increased (or decreased) in intensity beyond
anything found in the habitat, by optical interference
(the iridescence solution). The latter allows for the
combination of intense reflection (or absorption) with
any underlying visible coloration, andlor restriction
of iridescence to the ultraviolet. The use of dark
borders around a brilliant iridescence (as in most
Morpho, some Hypolimnas, many Colias, etc.)
provides a high-eontrast edge to the intense colour
field borne on the wings.
A visual signal may be effective in any spectral
region that coincides with the spectral sensitivity
range of the receiver. In butterflies we should expect
to find important visual signals scattered from the
UV to the red. Feeding and mate-finding behaviour
are mediated in part by visual signals throughout the
spectrum, and the bunerfly's responses to differences
of hue, saturation and intensity are most evident in
these contexts. However, a surprising number of
important signals involving sexual communication
are effected through bright coloration at all
wavelengths (white), UV-reflectance and iridescence.
Many ofthese appear to be all-or-none signals, such
as 'I am a male' or 'go away'. They differ from the
full-spectrum group in that the signal's function is
ofgreat value to the sender as well as to the receiver.
Iridescent colours and/o' ordinary high-intensity
reflection, as well as large size, are characteristic of
butterflies of open spaces, where long-range
signalling may be important. The most spectacular
examples of such colours are found in wide-ranging
pierids and nymphalids that frequent the upper
canopy, and species that use treefall gaps, shafts of
sunlight and other openings in the forest. Such colour
20. Visual Communication and Sexual Selection 219
patterns are relatively less well developed in the
satyrines, ithomiines and other groups that frequent
deep forest or dense vegetation. While there are
numerous exceptions, I believe most lepidopterists
would agree with these generalities (cf. Hingston
1933, Papageorgis 1975).
Courtship and Intersexual Selection
Male and female behaviour during courtship agrees
well with observations on other organisms, and with
recent sexual selection theory (see Blum & Blum
1979, Maynard Smith 1978, O'Donald 1980, Parker
1974, Rutowski 1982, Thornhill 1976, 1979, 1980,
Trivers 1972, West-Eberhard 1979, Williams 1975).
Males are aggressive toward rivals, and persistent in
locating and courting females, while females are coy
and effective at rejecting males. Female mate-
rejection behaviour has been described in numerous
species (Scott 1973d). A male butterfly usually cannot
rape an unreceptive, rejecting female unless she is
teneral (Taylor 1972).
Female rejection behaviour is elicited by courting
males in a variety of situations. Recently fertilized
females are generally unreceptive. A receptive female
may respond with rejection behaviour in the initial
stages of courtship, or she may respond negatively
if the courting male lacks appropriate signals (e.g.
another species with different pheromones). While
UV visual signals are employed in a few cases,
pheromones are almost always involved; a male that
has been experimentally deprived of the necessary
pheromones is usually rejected (e.g. Tinbergen et al.
1942, Tinbergen 1958, Pliske & Eisner 1969; Ch.25).
Male butterflies depart immediately after
copulation: there is no behavioural post-mating
investment by males on behalf of their offspring.
Males do not control access to resources. Besides
gametes, the male's only contribution toward his
offspring is the spermatophore and its contents,
which in one species have been shown to be
metabolized by the female and incorporated into the
contents ofeggs (Boggs & Gilbert 1979; but see also
Chs 3, 6).
If a female were somehow to assess male quality
during courtship, in terms of potential paternal
investment, she would have to do so on the basis of
the spermatophore he would produce if she accepted
him. Older males that have previously mated produce
smaller spermatophores, and also place the female
at risk to predation due to prolonged copulation times
(Rutowski 1979, L. E. Gilbert, cited in Thornhill
1976, Silberglied, unpublished). But because of her
low visual resolution, it is unlikely that a female can
discriminate between young and old males on the
basis of colour pattern or wing wear. Most of a
female's efforts in a generally short life are spent
locating suitable hostplants for oviposition (Watt
1968; Ch.6). A female is unlikely to waste valuable
time selecting among males on the basis of colour
or pattern, features that are likely to be poor
predictors of age and male quality (cf. Thornhill
1980). A female that mated with a male who
produced a small spermatophore (or who was judged
unsatisfactory on any other grounds) could always
remate sooner than she otherwise would. (Because
strong sperm precedence exists in butterflies, a male
should produce large spermatophores to prevent
this - Ehrlich & Ehrlich 1978; see also Rutowski
1978a, 1978b.) Another male adaptation that may
perform a similar function is the antiaphrodisiac
pheromone (Gilbert 1976). The female should
minimize the time to copulate with a healthy male,
so as to get on with oviposition-and the identifi-
cation and health of a male may be better determined
by chemical signals than by visual ones (Rutowski
1978b, 1981; Ch.21). Chemical signals may be more
reliable at close range than visual ones, due to
variation in lighting conditions and in the relative
positions of sender and receiver. Initial rejection
behaviour by receptive females may also be inter-
preted as a means ofassessing male vigour or potency
(Rutowski 1979).
Butterfly Coloration and Male Behaviour
Male butterflies definitely use colour as a signal to
recognize females. This has been demonstrated in
nearly every experimental study. But generally males
do not encounter more than one female at a time,
and they respond to anything that mIght be a female
by approaching it. In natural conditions, males must
decide whether to chase sequentially presented
stimuli that might be conspecific females. A wild
male rarely chooses between different stimuli
presented simultaneously, as is usually the case in
'dummy' experiments. Males make inspection flights
in response to a much broader range ofvisual stimuli
than one might predict on the basis of results of
approach tests.
Unlike females, males may mate many times;
variance in the number of matings by males is
probably quite high. Because receptive females are
scarce and competition among males is high, one
should not expect to find strong male discrimination
against conspecific receptive females. In fact, males
often persistently court, and sometimes mate with,
females of other species and even on rare occasions
with other males (Tilden 1980, Silberglied, un-
published).
It is unusual to find unmated older females (Ehrlich
& Ehrlich 1978), even in species having female
polymorphism with 'unattractive' forms such as
'valesina' of A Igynnis paphia (Magnus 1958b) and
'turnus' of Papilio glaucus (Burns 1966, 1967; cf.
Prout 1967, Pliske 1972, Levin 1973; Ch.22). Such
220 Robert E. Silbe1glied
females mayor may not attract as many males, over
as great a distance or as often, as 'normal' phenotypes,
but it is unlikely that the availability ofwilling males
is ever limiting, or that copulation by such females
is significantly delayed. Females may also compensate
for lower attractiveness to males by performing
'solicitation' flights.
For these reasons I do not believe that the colour
preferences shown by males, under experimental
conditions, necessarily translate into significant sexual
selection on female colour in the field. There exists
a graded response to sign stimuli, but it is in the
male's best interest to mate at every opportunity, and
to drive away or avoid competing males. A male's
colour preferences serve him well in locating
conspecific females, in shunning or thwarting rivals,
and in wasting less time with members of other
species. But because opportunities for choice between
receptive females rarely occur, male colour
preferences probably do not serve to stabilize female
pattern to any great extent (Magnus 1963, Pliske
1972). The high variability offemale coloration, and
female polymorphism, are indirect reflections of this
(cf. Richards 1927).
Butterfly Coloration and Female Behaviour
It is not surprising that we know so little of female
preferences in butterflies. This siruation is a reflection
of the difficulties involved with such experiments,
including the few female behaviours that can be used
to measure the relative acceptability of males, the
requirements that receptive (i.e. virgin) females be
used and that a separate female be used for each data
point (i.e. mating), and the fact that many, if not
most, species do not mate readily in cages. There
has also been a widespread assumption of colour
discrimination by females, hence little interest in such
experiments. Yet all experiments of this type point
to the fact that colour is used less by females than
by males, as a basis for discrimination, that females
use colour little, if at all, and that when females use
visual cues, these are primarily in the UV (Silberglied
& Taylor 1978, Rutowski 1981).
However, I do not deny that females choose. They
exhibit effective rejection behaviour, especially after
mating and during other periods of unreceptivity.
The evidence simply does not support choice on the
basis of 'visible' colour characters. Olfactory and
ultraviolet signals are used. Perhaps if we were as
receptive and attentive to olfactory stimuli as we are
to bright colours and complex patterns, we would
appreciate the fragrant world of male butterflies as
a sensory nirvana surpassing even their kaleidoscopic
adornments.
A hypothesis cannot be proven; one can only
attempt to refute it by means of experiment and
observation. Darwin's hypothesis is that female
choice is based on male colour. Experiments have
been designed to try to disprove this hypothesis.
Females have been offered choices among males
bearing different colours. They exhibited little
preference, if any, and at most far less preference
for colour of their mates than do males. Hence the
hypothesis is falsified, in spite of its logic, in spite
of its apparent success at accounting for many colour
phenomena in butterflies, and in spite of its
attractiveness to evolutionary theorists.
The argument might be made that sexual selection
is operating, and that females are choosing males on
the basis of colours not changed in the experiments
(e.g. the black of Colias spp. and Anartia amathea).
However, the question is, to what extent are the
spectacular patterns and colours of butterflies, as we
see and interpret them, a product of sexual selection
by females? Ultraviolet signals and dull 'background'
colours are not at issue here; the former were
unknown to Darwin, and the latter have never been
suggested as products of intersexual selection.
Male Intrasexual Selection-
an Alternative Hypothesis
The season of love is that of battle.
(Darwin 1874)
If female colour preferences are unimportant
selective forces on male coloration, we are left with
the problem of accounting for the peculiar sexual
distribution of colour and pattern in butterflies. The
problem is now more acute, for the old argument
is now reversed: if colour is more important to the
male than to the female, we should expect females
to be less variable in colour than males.
I would like to suggest an alternative hypothesis,
one that may explain the distribution of many
butterfly colour patterns, and yet be in accord with
the experimental and observational data on butterfly
courtship. I believe that inn·asexual communication
between males (see also Ch.23), rather than inter-
sexual or interspecific communication, is the major
selective agent ,·esponsible for brilliant male coloration,
low male colour variability, lack of male-limited
polymorphism, and absence of male sex-limited
mImIcry.
Darwin (1874) recognized conflict between males
as the other important component ofsexual selection,
but his discussion of characters produced by intra-
sexual selection was concerned primarily with
physical structures, such as the horns of beetles,
narwhals and moose. He was aware of aerial battles
between males in some butterfly species, but did not
link such observations with sexual dimorphism for
colour or pattern. Darwin appreciated that inter-male
displays figure in the evolution of bird coloration,
but he did not consider brilliant butterfly wing
colours as important signals used for communication
20. Visual Communication and Sexual Selection 221
between males. He instead attributed them almost
entirely to intersexual selection by females.
The alternative hypothesis, of male intrasexual
selection, as an important factor in the evolution of
butterfly sexual dimorphism, is not new. Wallace
(1877,1889), Hingston (1933) and Huxley (1938a)
refuted intersexual selection in general, in favour of
intrasexual selection among males. In particular,
Hingston (1933) discussed the anomalous distribution
of colour and pattern in butterflies. He presented a
wide variety ofobservational evidence in support of
his view that the brilliant dorsal colours ofmany male
butterflies function in fighting. To Hingston, male
butterflies engage in 'psychological warfare', 'a battle
of bravado, gesticulation and threat'; 'their colours
are . . . their weapons'.
Other recent authors have mentioned male
interactions as a possible factor in the evolution of
sexual dimorphism, but usually without further
elaboration or supporting evidence (e.g. Turner
1978). Vane-Wright (1980), reviewing possible
combinations of sexual interactions between
butterflies, has also suggested that colour patterns
might play a role as male-male signals (see also
Ch.23). The hypotheses presented by Wallace,
Hingston and Huxley are in accord with many of
the observable facts ofbutterfly colour distribution.
Inter- and intrasexual selection are not mutually
exclusive agents. In many cases, they serve equally
well to 'explain' the same observations. Until
recently, it has not been possible to weigh the relative
importance of intersexual versus intrasexual selection,
due to lack of direct evidence. Now, having
demonstrated experimentally that female selection
on male colour is weak at best, we must evaluate the
evidence regarding male intrasexual selection on
colour and pattern. This evidence comes from diverse
sources, including field observations and
experiments, as well as from the physical nature of
colour and its distribution among butterfly species.
The male is the active, mate-locating sex in
butterfly courtship. If brilliant colour had evolved
as a signal facilitating mate location, we would expect
to find it better developed in the female. This is not
the case. Colours detectable at a great distance are
not necessary for female choice, because butterfly
courtship usually takes place in a space ofless than
one cubic metre.
Fighting between males has been reported in
numerous species (Baker I 972a, 1978, Davies 1978,
L. E. Gilbert in Maynard Smith & Parker 1976, Joy
1902, Richards 1927, Hingston 1933). The behaviour
can be observed at food,sources as well as in other,
but not all, contexts (cf. puddling, migration; Shapiro
1970). In a few stout-bodied species, the wings are
modified as weapons (e.g. Charaxes spp.; Owen
197Ia), and physical damage may be inflicted during
flights (Darwin 1874). But direct physical contact
should be avoided, for such action may result in a
pyrrhic victory, in which the winner may be as badly
damaged as the vanquished. Selection for effective
alternatives, such as ritualized threat displays, would
be expected. The signals need not be visual-loud
snapping noises accompany inter-male fights in
Hamadryas (Darwin 1839, Swihart 1967b).
Males of many butterfly species are 'territorial'
(Baker 1972a, 1978; see also Silberglied 1977). Such
behaviour is best developed in species exhibiting
'waiting' mate-location behaviour (cf. Scott 1973d).
Territorial defence in butterflies probably serves to
prevent access by rival males to sites at which
receptive females fly. (Among birds, food resources,
perches, and nests are also defended.) For this reason,
butterfly 'territories' are not necessarily fixed in
location from day to day, or even from one time of
day to another (Baker 1972a, Owen 1971a, Davies
1978). These differences between butterfly and bird
territoriality have served to complicate discussions
with semantic problems and confound researchers
looking for strict analogies (e.g. Ross 1966, Scott
1973d). In species that 'hilltop' males defend
prominences to which receptive females fly in search
of mates (Shields 1968, Scott 1974b). In territorial
or aggressive species, brilliant male colour may serve
to intimidate rivals for prime locations. Males that
are victorious in such encounters would receive a
disproportionate share of matings, resulting in
selection for stronger male signals.
Male butterflies of many species make inspection
flights whenever anything that might be a female
appears (e.g. Tinbergen er al. 1942, Lederer 1960,
Stride 1957). As a result, frequent male-male
interactions occur, especially in species exhibiting
'seeking' mate-location behaviour (though also in
'waiting' species). Mating success in such species
should be related to the efficiency with which an area
is searched for females. It would be of advantage to
a male to identify his sex to other 'seeking' males
from a great distance; thus reducing time wasted in
fruitless homocourtships. It would be of advantage
to the signal receiver for the same reason, and also
because his time may be better spent searching areas
not recently explored by another cruising male (cf.
Rutowski 1981). While such behaviour may result
in greater dispersion ofmales throughout the habitat,
its explanation is based on individual advantage
rather than group selection (Wynne-Edwards 1962).
Experimental evidence also supports the hypothesis
that males respond to one another's colours. Colour
patterns are more important to males than to females.
Not only are males attracted to visual stimuli
resembling females in approach tests, they are repelled
by visual stimuli resembling other males (Obara
1970, Stride 1958a (but see also Ch.21), Rutowski
1977a, 1981, Silberglied & Taylor 1978). White, UV
reflectance and iridescent colours seem to be most
222 Robert E. Sitberglied
important in this regard. Some species have special
displays in which the male's dorsal colours are
exhibited while sitting (e.g. 'flutter response'; Obara
& Hidaka 1964, Rutowski 1977b, 1978a; see also
David & Gardiner 1961). Another striking example
is illustrated by the stereotyped rejection behaviour
of mating male Colias eurylheme, in which the
hindwings are flashed at approaching intruders
(Silberglied & Taylor 1978). However, to my
knowledge no one has performed an experiment to
determine if colour may affect the outcome of inter-
male combats between butterflies.
Males of the Grayling butterfly studied by
Tinbergen and his colleagues (Tinbergen el at. 1942,
Tinbergen 1958) are unusual in discriminating
colours while feeding but not in relation to social
interactions. This lack ofcolour use is consistent with
the lack of intense colour and absence ofstrong sexual
dimorphism in this species.
Like birdsongs, brilliant butterfly colours may
serve as agonistic devices, used for threat, intimi-
dation and rejection. The attractive, alluring,
seductive characteristics ofmale butterflies, required
for successful courtship are their odours. Male
pheromone glands are widespread in nocturnal as
well as diurnal Lepidoptera, and their pheromones,
unlike female sex-'attractants', are chemically diverse
(Ford 1945, Silberglied 1977; see also Ch.25).
Intersexual selection has probably played the
dominant role in the evolution of these scent organs
and their secretions-not in the evolution ofbrilliant
male colours.
Darwin (1874) noted the close analogy between the
secondary sexual characters of birds and insects,
especially in terms of beautiful colours and the
distribution and nature of sexual dimorphism.
'Whatever explanation applies to the one class
probably applies to the other; and this explanation
... is sexual selection.' However, Darwin treated
male coloration in birds largely as adornment evolved
under intense intersexual selection. This view has
not been supported by recent studies. According to
Rohwer el at. (1980), 'bright plumages are evolved
strictly for aggressive signalling. We know ofno good
support for the alternative hypothesis that bright
coloration serves in female attraction or as an isolating
mechanism.' In contrast, Baker & Parker (1979)
concluded that 'bird coloration has evolved almost
entirely in response to predation-based selective
pressures. Although plumage and coloration are
involved in species and sex recognition systems, they
have not evolved in response to sexual selection
pressures.' Thus, while ornithologists do not fully
agree on the selective agents responsible for bird
coloration, intersexual selection no longer has their
enthusiastic support (but cf. Burley 1981 and Ch.21;
see also Hingston 1933). Butterflies also differ from
birds in the clearer physical partitioning of
communicative and protective functions onto the
dorsal and ventral wing surfaces, exposed,
respectively, during display or at rest (Darwin
1874).
Baker & Parker (1979) also support the hypothesis
that 'bright colours may commonly be favoured when
an individual is anyhow obvious (e.g., through
activity), and where it represents an "unprofitable"
prey for a predator', as was suggested for insects by
Jones (1932; see also Young 1971, Gibson 1980).
Flash coloration-the sudden appearance of colour
during flight and its sudden disappearance at rest-
may startle or confuse predators. These and other
'predation hypotheses' do not conflict with, and are
not mutually exclusive of, the suggested role of
intrasexual selection as the major selective factor
producing brilliant male coloration. They do not
explain why brilliant coloration should be so
advantageous for the male, but less so for the female.
These hypotheses should be supported to the extent
that they are in agreement with experimental data,
but are unlikely to account for the phenomena of
male brilliance, low variation, etc. Colour may also
play other roles (e.g. interspecific signalling in social
aggregations, such as puddling) that do not conflict
with these conclusions.
Another hypothesis that has been proposed to
account for brilliant male coloration is the so-called
'handicap principle' (Zahavi 1975). This suggestion,
that females prefer to mate with males that have
survived i1l spile of the 'handicap' of brilliant
coloration, has serious theoretical flaws (Maynard
Smith 1976, 1978, O'Donald 1980; cf. West-
Eberhard 1979, Thornhill 1980). But even if
that was not the case, Zahavi's hypothesis would
predict female choice based on male colour. This
prediction is not supported by the experimental
evidence.
The concept of intersexual selection on colour
pattern has figured prominently in discussions of
butterfly systematics, genetics, behaviour, mimicry
and evolution (e.g. Carpenter & Ford 1933; Wickler
1968, Vane-Wright 1971, Scott 1973d, Silberglied
1977, Turner 1977a, 1978). Some theories developed
using this concept do not depend on male-female
communication as a selective force, and may easily
be modified by the partial or complete substitution
of intrasexual for intersexual selection. Others will
require major revision. Male intrasexual selection
cannot account for all interesting colour phenomena
in butterflies, many of which are better understood
in the contexts of thermoregulation, camouflage,
mimicry, and in some cases (especially UV),
intersexual selection. These agents are not mutually
exclusive, but act in concert, sometimes one or
another playing a more important role, producing
infinite variations, concerti and symphonies on the
themes of colour, pattern and vision.
20. Visual Communication and Sexual Selection 223
Summary
Butterflies have a visual spectral sensitivity extending
from the ultraviolet to the red, the widest known
among animals. They have excellent colour vision,
and exhibit both innate and learned colour
preferences. Butterfly wings bear diverse colour
patterns, and colour is used by butterflies during
feeding, oviposition and sexual behaviour.
Nevertheless, there is little evidence to support
Darwin's argument that intersexual selection, in the
form ofmale and female mating preferences, acts as
a potent force in the evolution of 'visible' butterfly
coloration.
For males of many species studied, the colour of
the female provides an important visual stimulus for
mate-location and recognition. However, male
behaviour is probably not an important selective
factor in determining female colour, and males do
not often discriminate between potential mates on
its basis. This is so because male colour preferences
are relatively broad, females are usually encountered
and courted one at a time, the number of receptive
females encountered usually limits male reproductive
success, and males can mate many times. Female
variation and polymorphism, sex-limited mimicry
and low pattern diversity among females of closely
related species provide indirect evidence that males'
preferences probably do not act as important
stabilizing selective forces on female colour and
pattern.
Female butterflies exhibit well-developed rejection
behaviour. The bases for discrimination by females
have been little studied, and there are very few data
on colour preferences. Available data reveal that
female colour preferences are weak or absent, or are
at most even less precise than those of males. The
only visual signals on the males' wings that have been
shown to affect female behaviour lie in the ultraviolet.
As determinants of female acceptance behaviour,
olfactory stimuli appear to be far more important
than visuaI.
The brilliant colour patterns of male butterflies
have signal properties that would serve well for long-
range communication. It is suggested that male-male
interactions may be the major selective agent in
the evolution of bright colours and iridescence in
males, low variability of male coloration, lack of
male sex-limited mimicry, and several other general
colour phenomena in butterflies. Recognition of
other males, and advertisement of his own sex,
are advantageous for a male in the contexts
ofagonistic, territorial and mate-location behaviour.
This suggestion is consistent with conclusions of
recent reviews of the evolution and function of bird
coloration.
Acknowledgements
I would like to thank A. Aiello, S. Foster, I. Rubinoff, D. Roubik, N. G. Smith, M. ].
Wesr-Eberhard, H. Wolda, and other colleagues at the Smirhsonian Tropical Research
Institute for stimulating discussion of some of the ideas presented and for helpful comments
on an earlier draft. Other helpful information, suggestions or access to unpublished material
were provided by M. Boppr., B. H6lldobler, K. Mikkola, R. Rutowski, D. Schneider,
D. A. S. Smith, and R. I. Vane-Wright. A special debr is owed to T. Eisner, who first
kindled my interest in problems ofcoloration) and with whom several afthe ideas presented
were discussed on various occasions. The colour photographs for Plate 4 were prepared
with technical aid ofN. Smythe and K. Dressler. The Smithsonian Institution, Scholarly
Srudies Program and Fluid Research Fund are gratefully acknowledged for financial support
of some of the experiments discussed.