Oceanography and Marine Biology: An Annual Review (Volume 43) - Chapter 9 pot

455

Oceanography and Marine Biology: An Annual Review,

2005,

43

, 455-482
© R. N. Gibson, R. J. A. Atkinson, and J. D. M. Gordon, Editors
Taylor & Francis

ECOLOGY AND EVOLUTION OF MIMICRY
IN CORAL REEF FISHES

EVEN MOLAND,

*

JANELLE V. EAGLE & GEOFFREY P. JONES

School of Marine Biology and Aquaculture, James Cook University,
Townsville, 4811 Queensland, Australia

*E-mail:
Address correspondence to: Even Moland, Carsten Ankersgt. 11, N-1524 Moss, Norway

Abstract

This review examines the literature on mimicry in coral reef fishes and evaluates the

prevalence of mimicry in different taxa, its ecological consequences and postulated modes of
evolution. Mimicry appears to be a widespread and common phenomenon in coral reef fishes, with
approximately 60 reported cases. Although many are largely anecdotal accounts based on colour
resemblance, recent quantitative comparisons and experimental manipulations have confirmed that
many do represent mimic-model relationships. The distribution of mimics and models among reef
fish families appears largely serendipitous. Mimics are most common in the families Blenniidae,
Serranidae and Apogonidae and models in the families Pomacentridae, Blenniidae and Labridae.
Mimics and model species usually represent less than 10% of species within families, although
imperfect forms of mimicry are likely to have been underestimated. Mimicry appears to be partic-
ularly important during juvenile stages, with 28% of mimic species losing their mimic colouration
when they outgrow their models. All cases of mimicry support predictions that mimics are rare
relative to their models. Furthermore, the abundance of mimics in different areas may increase in
proportion to model abundance. The spatial distribution of mimics appears to be limited by that of
their model species, although some change models in different habitats or in different parts of their
range. Many mimics live in close association with their models, and both foraging advantages and
predator avoidance have been experimentally demonstrated. Aggressive mimicry appears to be the
most prevalent type of mimicry overall in coral reef fishes, constituting 48% of all cases reported
to date, followed by Batesian (40%) and social mimicry (12%). Müllerian mimicry seems to be
rare, although it may contribute to the mimetic complexes involving members of the blenniid tribe
Nemophini. However, these traditional classifications are too simplistic for reef fishes because both
foraging advantages and predator avoidance can apply in a single mimetic relationship, and their
relative importance has not been evaluated. Preliminary data suggest a high degree of phenotypic
plasticity in mimetic colouration and little genetic differentiation among different mimics of the
same species. Overall, the review highlights the many significant steps that need to be taken towards
a more complete understanding of the ecological and evolutionary significance of mimicry in coral
reef fishes.

Introduction

The phenomenon of mimicry, where one species evolves to closely resemble another, has arisen

many times throughout the plant and animal kingdoms (Wickler 1965, Turner 1977, Gilbert 1983,
Malcolm 1990, Mallet & Joron 1999). The evolution of mimicry occurs in response to selective

3597_book.fm Page 455 Friday, May 20, 2005 6:26 PM
© 2005 by R.N Gibson, R.J.A. Atkinson and J.D.M. Gordon

EVEN MOLAND, JANELLE V. EAGLE & GEOFFREY P. JONES

456
pressures favouring individuals that are able to disguise their identity by masquerading as another
species. The study of mimicry began in the mid-1800s, with most of the theory developed from
field studies on terrestrial animal groups, most notably butterflies. Despite this long history of
interest, the prevalence of mimicry, its ecological significance and mechanisms of evolution have
yet to be evaluated for many animal groups.
The proposed fitness advantage of mimicry depends upon who is being deceived. Four different
types of mimicry have been traditionally recognised. In

Batesian

mimicry, harmless and palatable
species closely resemble unpalatable or venomous species usually avoided by predators (Bates
1862). In

Müllerian

mimicry, two unpalatable species share similar colour patterns and thus
reinforce their predator deterrence (Müller 1879). In

aggressive

mimicry, a predatory species
resembles a harmless or beneficial species and thus achieves increased opportunities for foraging
by deceiving prey (Wickler 1965, 1968, Malcolm 1990).

Social

mimicry, proposed for birds by
Moynihan (1968), refers to examples of mimics that associate with similarly coloured individuals
in order to escape the attention of predators. Despite these relatively well-defined categories, the
underlying bases of many mimic-model relationships are poorly understood.
Mimic-model systems are thought to share a number of defining ecological characteristics that
should apply if mimicry is to provide a fitness advantage. Mimic species must be rare compared
with their model species because the deception will not work if the predators or prey that are being
deceived encounter too many mimics and learn from these experiences (Bates 1862). Mimics should
also occupy the same habitat (Randall & Randall 1960) and geographic range as the model (Turner
1977, Thresher 1978) so that signal receivers are able to recognise them. Across this range, variation
in the mimic should match any geographic variation in the model (Turner 1977, Thresher 1978)
or change to other model species with complementary ranges (Gilbert 1983, Mallet & Joron 1999).
Mimetic species should also alter their behaviour from that characteristic of their taxonomic group
to enhance resemblance to their models (Snyder 1999). These basic ecological patterns are untested
for most presumed cases of mimicry. If such relationships hold, the distribution and abundance of
mimics will be closely tied to that of their models. An understanding of the ecological relationships
between mimics and models is critical to assessing its ecological significance and underlying
evolutionary mechanisms.
The potential significance of mimicry in coral reef fishes was first recognized by Randall &
Randall (1960) who drew attention to many of the examples that are well known today. Many
additional cases have been reported over the last four decades, suggesting that mimicry may be a
relatively widespread phenomenon on coral reefs (Russell et al. 1976, Siegel & Adamson 1983,
Kuiter 1995, Snyder 1999, Snyder et al. 2001). However, despite the increasing records, there have
been relatively few studies specifically addressing the causes and consequences of mimicry in reef

fishes. Much of the evidence for mimicry in the early literature on coral reef fishes was anecdotal,
based on colour resemblance and observer intuition. This is gradually being superseded by more
rigorous observational studies addressing the criteria necessary to establish mimicry (e.g., Snyder
1999, Eagle & Jones 2004) and experimental studies establishing cause-effect links between mimic
and model species (e.g., Springer & Smith-Vaniz 1972, Caley & Schluter 2003, Munday et al.
2003, Moland & Jones 2004). Although there have been a number of reviews of different kinds of
mimicry in fishes that include coral reef species (Randall & Kuiter 1989, Randall & McCosker
1993, Sazima 2002a,b) there have been no recent reviews evaluating the prevalence of and evidence
for different types of mimicry in coral reef fishes. Early reviews of the ecology of reef fishes argued
that mimicry could have important implications (Ehrlich 1975, Sale 1980) but in recent texts this
topic has received virtually no attention (Sale 1991, 2002).
The prevalence of mimicry within a community can be difficult to assess. Descriptions of
mimicry usually focus on the more spectacular and specialized mimics that are striking to the
observer, which may lead to an underestimate of its importance (Russell et al. 1976). Three clear

3597_book.fm Page 456 Friday, May 20, 2005 6:26 PM
© 2005 by R.N Gibson, R.J.A. Atkinson and J.D.M. Gordon

MIMICRY IN CORAL REEF FISHES

457
examples are the similarity of the blenny

Aspidontus taeniatus

to the cleaning labrid

Labroides
dimidiatus

(Eibl-Eibesfeldt 1959, Randall & Randall 1960) (see Figure 1A,B, in the colour insert
following page 278), the similarity of the monocanthid

Paraluteres prionurus

to the unpalatable
tetraodontid

Canthigaster valentini

(Tyler 1966, Caley & Schluter 2003) (Figure 1C,D, see colour
insert) and the similarity of the harmless blenny

Petroscirtes breviceps

to the blenny

Meiacanthus
grammistes

which possesses a venomous bite (Smith-Vaniz et al. 2001) (Figure 1E, see colour
insert). Few would dispute that these are mimic-model pairs. The problem begins when the colour
similarities are not so striking to the observer. Theory predicts that both good and poor mimics can
evolve in different situations (Lindström et al. 1997, Edmunds 2000, Sherratt 2002); however,
distinguishing poor mimics from fortuitous resemblance can be a challenge. Recent experimental
studies (Munday et al. 2003) suggest that there can be a strong association between similarly
coloured reef fishes that would not be considered obvious cases of mimicry. In addition, what the
human eye records as identical may be very different from what the fishes actually perceive
(Marshall 2000).
Coral reef fishes may provide a challenge to the assumption that mimicry can be neatly classified

according to one of the four established evolutionary mechanisms. For example, cleaner wrasse
mimics may be unique among all known cases of mimicry. They are primarily thought to be
aggressive mimics, increasing their feeding opportunities when, masquerading as cleaner wrasses,
they remove scales or chunks of flesh rather than parasites from the bodies of their ‘customers’
(Randall & Randall 1960). However, they may also benefit from a special type of anti-predation
mechanism. Cleaner wrasses are afforded an ‘amnesty’ from predators because of the parasite
removal service they provide. Cleaner wrasse mimics may also benefit from this amnesty, thus
gaining the dual advantages of both an increase in feeding and a decrease in predation (Kuwamura
1983).
The overall aims of this review are to examine the current literature on the prevalence of
mimicry in coral reef fishes and to evaluate its ecological and evolutionary significance. In particular,
the degree to which reef fishes conform to a body of theory largely developed to explain mimicry
in terrestrial animals is evaluated and the evidence for different types of mimicry in coral reef fishes
critically assessed. Specific questions that are addressed include: How prevalent is mimicry in coral
reef fishes? Which families of coral reef fishes are involved in mimetic relationships, what is the
prevalence of mimicry within these families, and within species, what life history stages are
involved? What is the relationship between the distribution and abundance of mimics and models?
Do the geographic ranges of mimics and models coincide, are mimics rare relative to their models
and do mimics and models influence each other’s abundance? What is the evidence for the four
widely recognised types of mimicry (Batesian, Müllerian, aggressive and social) in reef fishes? Do
reef fishes conform to this classification, which type of mimicry is most important, or is the evidence
inadequate to reach this conclusion? Finally, which research directions must be taken to complete
our understanding of the ecology and evolution of mimicry in reef fish communities?

Prevalence among reef fishes

Taxonomic distribution

Approximately 60 species in 16 families have been reported to mimic other coral reef fish species
(Figure 2A). Three speciose families (Blenniidae, Serranidae and Apogonidae) are disproportionately

represented, accounting for about two-thirds of the reported cases. Nine families are represented
by only a single mimetic species. The proportion of species in each family that are mimics is
usually quite low (<10%). The exception appears to be the Nemipteridae, where 27% of species
are mimetic at some stage in their life history. Mimicry is least prevalent in the two of the most

3597_book.fm Page 457 Friday, May 20, 2005 6:26 PM
© 2005 by R.N Gibson, R.J.A. Atkinson and J.D.M. Gordon

EVEN MOLAND, JANELLE V. EAGLE & GEOFFREY P. JONES

458
speciose of coral reef fish families, the Pomacentridae and Labridae, with one and three mimics
representing 0.3% and 0.8% of their total species diversity, respectively. Many reef fish families
that are characteristic of coral reefs (e.g., Chaetodontidae, Pomacanthidae and Scaridae) are notably
absent from the list of mimics. The taxonomic distribution of mimicry appears largely serendipitous
and it has evolved independently on numerous occasions.
The models for mimetic species are also found in a wide range of reef fish families (Figure 2B).
However, the taxonomic distribution of mimics and models is quite different, with approximately

Figure 2

(A) Prevalence of mimicry in coral reef fish families involved in mimetic relationships, (B) preva-
lence of species functioning as models in the same families. Percentages represent proportion of mimics and/or
models within each family. Number of mimics and models are drawn from the literature. Percentages were
calculated using information on species diversity derived from Bellwood (1997) and Munday & Jones (1998).
0.6% 5%
3%
- - 4% 2%
6%
6%

10% 4%
3%
1%
1%
3.5%
Number of models
Tetraodontidae
Tripterygiidae
Pomacanthidae
Chaetodontidae
Scorpaenidae
Atherinidae
Muraenidae
Monacanthidae
Chaenopsidae
Pom
ace
ntridae
Mullidae
Carang
idae
Malacanthidae
Plesiopidae
Pseudochromidae
Acanthuridae
Gobiidae
Labridae
Lutjanidae
Nemipteridae
Apogonidae

Serranidae
Blenniidae
22
20
18
16
14
12
10
8
6
4
2
0
B
A
4%10%10% - 4% 0.3% 10% 4%
1% 0.1% 2%
4%
27%
3%
7%
6%
Number of mimetic species
Tetraodon
tidae
Tripterygiidae
Pomacanth
idae
Chaetodontidae

Scorpaenidae
A
therinidae
Muraenidae
Monacanthidae
Chaenopsidae
Pomacentr
idae
Mullidae
Car
angidae
Malacanthidae
Plesiopidae
Pseudochromidae
Acanthuridae
Gobii
dae
Labridae
Lutjanidae
Nemipteridae
Apogonidae
Serranidae
Blenniidae
20
18
16
14
12
10
8

6
4
2
0

3597_book.fm Page 458 Friday, May 20, 2005 6:26 PM
© 2005 by R.N Gibson, R.J.A. Atkinson and J.D.M. Gordon

MIMICRY IN CORAL REEF FISHES

459
three-quarters of model species found in the families Pomacentridae, Blenniidae and Labridae. The
proportion of species in these families that serve as models is also low (<10%) and for nine families
there is only one species that is a model. Only the families Blenniidae and Serranidae include
relatively high numbers of both mimics and models.
Around three-quarters of mimetic relationships involve one mimic and one model species.
However, there are cases where a model has more than one mimic (Figure 3A) and cases where a
mimic has more than one model across its geographic range (Figure 3B). There are several cases
of mimetic species having four or more models. For example, juvenile

Lutjanus bohar

mimic four
different

Chromis

species (Russell et al. 1976, Moyer 1977) and the two labrids

Oxycheilinus

diagrammus

and

Epibulus insidiator

(Ormond 1980). Some model species appear to be sought
after, such as the cleaner wrasse

Labroides dimidiatus,

which is mimicked by three blennies

,
Aspidontus taeniatus

,

A. filamentosus

and

Plagiotremus rhinorhynchos

(Randall & Randall 1960,
Wickler 1961, Springer & Smith-Vaniz 1972, Kuwamura 1983). Other cases of several mimics to
a single model include the mimetic complex (‘mimicry rings’) centred on each of the fangblennies

Meiacanthus nigrolineatus

,

M. oualanensis

and

M. vittatus

(Smith-Vaniz et al.



2001).

Body size and ontogenetic patterns

Mimicry appears to be particularly important in small coral reef fishes and for larger fishes, mimicry
is confined to the juvenile stage. Of the 60 mimetic species reported to date, 17 (28%) are mimetic

Figure 3

(A)



Number of mimic species per model and (B)



number of model species per mimic. Percentages

represent the prevalence of the different relationships.
0
5
10
15
20
25
30
35
40
45
50
123 4
123 4
Number of mimic species per model
Frequency
75%
18%
7%
0
5
10
15
20
25
30
35
40
45
Number of model species per mimic

Frequency
72%
9%
10%
9%

A
B

3597_book.fm Page 459 Friday, May 20, 2005 6:26 PM
© 2005 by R.N Gibson, R.J.A. Atkinson and J.D.M. Gordon

EVEN MOLAND, JANELLE V. EAGLE & GEOFFREY P. JONES

460
only as juveniles (Figure 4). These species appear to lose mimetic colouration when they grow to
a larger size than their model species (Sazima 2002a, Eagle & Jones 2004, Moland & Jones 2004).
In the family Serranidae, the seven species mimetic throughout ontogeny are the small hamlets,

Hypoplectrus

spp. (Randall & Randall 1960, Thresher 1978), which grow to a maximum size of
13 cm (Lieske & Myers 1996). In contrast, the seven species that are mimetic only as juveniles
are all epinepheline groupers (Russell et al. 1976, Randall & Kuiter 1989, Kuiter 1995, Snyder
1999, Snyder et al. 2001) that reach a much larger body size ranging from 52–110 cm depending
on the species (Lieske & Myers 1996).
The blenniid

Meiacanthus nigrolineatus

is interesting because it starts life first as a mimic,
then itself becomes a model. It is a social mimic of several

Cheilodipterus

spp. (Apogonidae) as
a juvenile (Dafini & Diamant 1984). As an adult, it serves as a model in Batesian and aggressive
mimicry for the blenniids

Ecsenius bicolor

and

Plagiotremus townsendi,

respectively (Springer &
Smith-Vaniz 1972, Smith-Vaniz et al. 2001).

Ecological patterns and consequences

Rarity and local abundance

There is surprisingly little published data on the relative abundance of mimic and model species
(Figure 5). Generally, reef fishes conform to the prediction that mimetic species must be rare relative
to their model species (Bates 1862). However, while mimic species are usually less abundant than
their models, the percentage of mean abundance per site ranges from 1–78% of the abundance of
their model species (mean 29.3%

±

7.3%).
Recent quantitative studies suggest that the abundance of mimics is influenced by the abundance
of their models. Eagle & Jones (2004) showed that the densities of

Acanthurus pyroferus

, a mimic
during its juvenile stage, are promoted by the abundance of one of its models,

Centropyge vroliki

.
The abundance of mimics (juveniles) increased in proportion to increased abundances of the model
at spatial scales both within and among reefs. Similarly, Moland & Jones (2004) found a positive
relationship between the mimic

Plagiotremus rhinorhynchos

and its model

Labroides dimidiatus

.

Figure 4

Frequency of mimicry in juveniles vs. mimicry throughout life history for all coral reef fish families
with mimetic species.
0
2

4
6
8
10
12
14
16
18
20
Blenniidae
Serranidae
Apogonidae
Nem
ipteridae
Lutjanidae
Labridae
Gobiidae
Acanth
urid
ae
Pseudoch
rom
idae
Plesiopidae
M
alacanthidae
C
arangidae
Mullidae
Pom

acentridae
Chaenop
sidae
Monacanthidae
Frequency
Throughout ontogeny
Juvenile stage only

3597_book.fm Page 460 Friday, May 20, 2005 6:26 PM
© 2005 by R.N Gibson, R.J.A. Atkinson and J.D.M. Gordon

MIMICRY IN CORAL REEF FISHES

461
If this proves to be a trend for a number of mimetic relationships, especially those involving juvenile
mimics, then mimicry could be of vital importance for survival of some species through the more
vulnerable ontogenetic stages.

Geographic range and local distribution

Reef fishes also conform to the prediction that mimetic species should occupy the same habitat
and geographic range as their model species, or change to other model species in other habitats or

Figure 5

Relative abundance of coral reef fish species in proposed model-mimic relationships. White bars =
model species. Black bars = mimic species. N = number of sites. Percentages represent average mimic to
model ratio per site. References at top of each panel refer to the study in which each example of relative
abundance between model and proposed mimic was reported.
23%

N=10
P. r h i norhynchosL. dimidiatus
5
4
3
2
1
0
Moland & Jones 2004
N=10
5%
A. pyroferusC. vroliki
14
12
10
8
6
4
2
0
Eagle & Jones 2004
43%
N=6
S. guamensisFowleria sp.
8
7
6
5
4
3

2
1
0
Seigel & Adamson 1983
41%
N=6
A. taeniatusL. dimidiatus
6
5
4
3
2
1
0
Kuwamura 1983
1%
N=2
L. boharC. flavomaculata
700
600
500
400
300
200
100
0
Moyer 1977
35%
N=5
A. taeniatusL. dimidiatus

10
8
6
4
2
0
Losey 1974
N= 7
36%

78%
P. townsendiE. gravieriM. nigrolineatus
25
20
15
10
5
0
Springer & Smith-Vaniz 1972
4%
27%
N=4
P. l. laudandusE. bicolorM. atrodorsalis
35
30
25
20
15
10
5

0
Losey 1972
Model and mimic species
Average abundance per site

3597_book.fm Page 461 Friday, May 20, 2005 6:26 PM
© 2005 by R.N Gibson, R.J.A. Atkinson and J.D.M. Gordon

EVEN MOLAND, JANELLE V. EAGLE & GEOFFREY P. JONES

462
parts of their range (Thresher 1978, Gilbert 1983). Spatial concordance between mimics and their
models appears to apply across a range of spatial scales including: (1) regional geographic ranges;
(2) local ranges within regions and (3) depth and microhabitat ranges within locations. A good
example of multiscale spatial relations can be seen in the mimic surgeonfishes. In the Indian Ocean

Acanthurus tristis

juveniles are mimics of the pygmy angelfish

Centropyge eibli

(Kuiter & Debelius
1994). In the Indo-Pacific, juveniles of the sister species

Acanthurus pyroferus

mimic three different

Centropyge

species,

C. vroliki

,

C. heraldi

and

C. flavissima

(Randall & Randall 1960, Myers 1989,
Kuiter 1996)



(Figure 6, see colour insert). In general

C. vroliki

is found in the central Indo-Pacific,

C. flavissima

is most abundant on coral atolls of the far Eastern Pacific and

C. heraldi

is in between.
There is, however, a very large area where the ranges of all three

Centropyge

model species for

Acanthurus pyroferus

overlap (Allen et al.1998). It is unclear what determines the model species
that

A. pyroferus

mimics in the area of overlap but it may be separation among the species at local
scales within this region. For example, whereas all three model species are reported from the Great
Barrier Reef,

Centropyge vroliki

is widespread but most abundant in inshore areas (Eagle, unpub-
lished data),

C. flavissima

is commonly seen in northern offshore areas and

C. heraldi

is uncommon,

occurring only occasionally on outer reefs and in the Coral Sea (Australian Institute of Marine
Science Monitoring Team, unpublished data).
The adopted model species may also reflect the depth distributions of potential model species.
There are many sympatric

Centropyge

species found across the geographic range of

Acanthurus
pyroferus

that are not mimicked. The four model

Centropyge

species are closely related sister taxa,
known to hybridise where their boundaries overlap (Allen et al. 1998, Debelius et al. 2003).
Furthermore, they are all shallow water species, found at their highest abundance closest to the
reef crest relative to their congeners (Allen et al. 1998, Eagle et al. 2001). Eagle & Jones (2004)
propose that this case of mimicry has evolved in part so that

Acanthurus pyroferus

juveniles can
deceive aggressive territorial herbivorous fishes and gain access to algal food resources on the reef
crest. If this is the case, mimicking

Centropyge

species that inhabit deeper microhabitats on the
reef would be of no benefit.
Mimics appear to be able to expand their distributions by changing to different model species
in different habitats. For example, juvenile

Anyperodon leucogrammicus

are aggressive mimics of
congeners

Halichoeres biocellatus

,

H. purpurascens

and

H. melanurus

(Russell et al. 1976,
Randall & Kuiter 1989). These three wrasses have disjunctive distributions along both depth and
exposure gradients (Jones et al. unpublished data). Hence, by mimicking all three species,

Anyperodon
leucogrammicus

may increase the habitat and area that it is able to utilise. Similarly,

Pseudochromis

fuscus

is an aggressive mimic of

Pomacentrus adelus, P. amboinensis

,

P. c hrysurus

and

P. moluccensis,

four damselfishes that are found at different depths and in different microhabitats
but in the same reef areas (Munday et al. 2003).
Other mimics with more restricted geographic ranges, but which overlap broadly in depth and
microhabitat, mimic a range of species from different families. For example, the hamlets,

Hypoplec-
trus

spp.

,

are aggressive mimics found only in the Caribbean that mimic species of angelfishes,
wrasse and damselfishes. The subset of microhabitats used by each species closely mirrors that of
their respective models (Thresher 1978).
Overall, the nature of the spatial relationship between mimics and models is likely to depend

on many factors including the type of mimicry and the geographic range size and habitat specificity
of both mimic and model species.

Dispersion and behaviour

Recent quantitative studies have found a close



spatial association between mimics and their models.
Eagle & Jones (2004) found that juvenile

Acanthurus pyroferus

were found in local scale areas of

3597_book.fm Page 462 Friday, May 20, 2005 6:26 PM
© 2005 by R.N Gibson, R.J.A. Atkinson and J.D.M. Gordon

MIMICRY IN CORAL REEF FISHES

463
reef (30 m

2

) inhabited by one or more fish of the model species

Centropyge vroliki,

more often
than would be expected if mimics were distributed at random. Within this scale, mimic

Acanthurus
pyroferus

were never observed >2 m away from an individual of the model species

Centropyge
vroliki

, and 10% of the time that they were observed they were actually located at a distance of
less than 10 cm from a model fish. Similarly, Moland & Jones (2004) found that

Plagiotremus
rhinorhynchos

mimics of the cleaner wrasse

Labroides dimidiatus

were often found less than 1 m
from a model fish. Many published anecdotes of mimetic relationships include observations of
behavioural associations between mimics and models (e.g., Barlow 1975, Caley & Schluter 2003)
and it is likely that more in-depth studies of other model-mimic pairs may uncover similarly close
patterns of spatial dispersion. This pattern suggests that just being located in a region or locality
where the model is found may not be enough to deceive most species, and that constant reinforce-
ment of the mimicry through interactions of the signal receivers with models is required.
For mimicry to work, it may not even be enough to look like and be found in close proximity
to models, mimics may also need to alter their behaviour (Snyder 1999). It has been observed in

at least two cases of mimicry in coral reef fishes that mimics alter their swimming behaviour to
enhance their resemblance to their model. The fins and swimming motion used for propulsion are
features that make it easy to distinguish different families of coral reef fishes underwater. Eagle &
Jones (2004) have observed that when foraging as a pair with a model (

Centropyge vroliki

), mimic
surgeonfish juveniles (

Acanthurus pyroferus

) do not swim with the pectoral fin propulsion charac-
teristic of other surgeonfishes, but instead adopt the ‘wiggling’ swimming motion of angelfishes.
Similarly, instead of using caudal propulsion, the juveniles of the grouper

Plectropomus laevis

have
been observed to fold their caudal fin, erect the front section of their dorsal spines and use their
pectoral fins in a sculling motion for propulsion in the style of the distinct posture and swimming
motion of the pufferfish

Canthigaster valentini

(A.M. Ayling, personal communication). Further
observations of the swimming motion and behavioural characteristics of other mimics may bring
to light similar modifications for enhancing model resemblance.

Types of mimicry and the evidence

There has been considerable variation in the types of evidence used to establish mimicry and
distinguish among the different types. Most early published studies provide only anecdotal obser-
vations of colour similarities and speculation about the basis of the mimicry. Only a few studies
quantify the ecological relationships between pairs of species (Springer & Smith-Vaniz 1972,
Kuwamura 1983, Côté & Cheney 2004, Eagle & Jones 2004) and even fewer provide experimental
evidence of the ecological advantages resulting from the mimetic relationship (Caley & Schluter
2003, Munday et al. 2003, Moland & Jones 2004). No studies provide all the evidence required to
establish mimicry and at present there is no clear guide to distinguish among the different types.
Here a set of criteria is established that can be used as ‘minimum evidence’ as an initial step to
identify the potential fitness advantages of mimicry and distinguish among the four recognised
types in each case observed (Table 1).

Batesian mimicry

In Batesian mimicry, harmless and palatable species closely resemble unpalatable or venomous
species usually avoided by predators (Bates 1862). Batesian mimicry has been reported for 25
species from 10 families of coral reef fishes. It is most prevalent in the families Apogonidae and
Blenniidae, with five and six species, respectively (Table 2).
Batesian mimicry has been recognized on the basis of six criteria (Rettenmeyer 1970



in
McCosker 1977): “(i) a species, the model, is undesirable to predators; (ii) a second species, the

3597_book.fm Page 463 Friday, May 20, 2005 6:26 PM
© 2005 by R.N Gibson, R.J.A. Atkinson and J.D.M. Gordon

EVEN MOLAND, JANELLE V. EAGLE & GEOFFREY P. JONES

464
mimic, is desirable to predators but has evolved from its ancestral appearance until it resembles
the model so closely that potential predators are deceived and leave it alone; (iii) the mimics are
less abundant than the models; (iv) the mimics are found at the same place and time as the models;
(v) both model and mimic are conspicuous or readily seen by potential predators and (vi) the
predators learn or associate undesirability with the appearance of the model”. However, although
this list is comprehensive, some of the criteria are common to other forms of mimicry (Table 1).
Batesian mimicry could be postulated on the basis of just three critical pieces of evidence,
although very few studies meet all these criteria (Table 2): (1) overall resemblance between model
and mimic in shape and colour, and known non-palatability of the model, (2) known palatability
in the mimic and (3) predator avoidance in the mimic (Table 2). The first requirement establishes
visual similarity and that the model has some feature that makes it less susceptible to predation.
The second requirement is crucial in ruling out the possibility of dealing with Müllerian mimicry,
in which both mimic and model are unpalatable or possess means of predator deterrence. The third
requirement is important in order to establish that the visual similarity has a realised adaptive
advantage. As shown in Table 2 there are only five of 25 coral reef fish species that fulfil these
criteria.
Although mentioned by Whitley (1935), Clark & Gohar (1953



in Randall & Randall 1960)
and others, the first proposed case of Batesian mimicry among coral reef fishes was the filefish

Paraluteres prionurus

(Monacanthidae), which is strikingly similar to the pufferfish

Canthigaster

valentini

(Tetraodontidae) (Tyler 1966). Model and mimic are virtually indistinguishable when
observed in the field. The evidence presented in the original article is anecdotal and does not contain
records of experiments testing the palatability of the mimic or to what degree the model is
unpalatable. It is now known, however, that

C. valentini

is likely to be avoided by predators
throughout its life history because the eggs and larvae are unpalatable (Gladstone 1987) and have
toxin in their skin and inner organs (Allen et al. 1975).
Caley & Schluter (2003) described the adaptive advantage gained by a mimic species as a
‘protective umbrella’ that may vary in size and shape according to the extent the mimic is protected
by its resemblance to the unpalatable model. They provide an excellent demonstration of a study
designed to measure the ‘protective umbrella’ that the pufferfish

C. valentini

provides for its mimic

Paraluteres prionurus

. By exposing predators to plastic models with increasingly divergent colour

Table 1

Multiple criteria to distinguish among the four recognised types of

mimicry when reporting cases of mimicry in coral reef fishes

Criteria Batesian Müllerian Aggressive Social

Record of overall resemblance Yes Yes Yes Yes
Information on diet No No Yes Yes
Model unpalatable Yes Yes No No
Mimic unpalatable No Yes No No
Mimic palatable Yes No No No
Record of predator avoidance in mimic Yes Yes No Yes
Record of schooling with model No No Yes/No* Yes
Record of taking prey disguised as model No No Yes No
Record of foraging on same food source as model No No No Yes
Record of foraging on other food source than model No No Yes No

* Referring to different trends suggested for species exhibiting aggressive mimicry; Yes = species that
feed on prey smaller than themselves tend to mimic and join species harmless to their prospective prey;
No = species that feed on prey larger than themselves tend to mimic mostly beneficial species (cleaners)
(modified from Sazima 2002a).

3597_book.fm Page 464 Friday, May 20, 2005 6:26 PM
© 2005 by R.N Gibson, R.J.A. Atkinson and J.D.M. Gordon

MIMICRY IN CORAL REEF FISHES

465

Table 2

Coral reef fish species reported as Batesian mimics and their putative models, with

references and the kind of evidence presented for each species
Mimic Model Model family Reference Evidence
Centrogenysidae
Centrogenys vaigiensis Parascorpaena picta Scorpaenidae Whitley 1935 A
Serranidae
Plectropomus laevis* Canthigaster valentini Tetraodontidae Randall & Hoese 1986 AB
Plesiopidae
Calloplesiops altivelis Gymnothorax meleagris Muraenidae McCosker 1977 AB
Apogonidae
Cheilodipterus
nigrotaeniatus
Meiacanthus grammistes Blenniidae Smith-Vaniz 1987 AB
Cheilodipterus
parazonatus
Meiacanthus vittatus Blenniidae Smith-Vaniz et al. 2001 AB
Cheilodipterus zonatus Meiacanthus geminatus Blenniidae Gon 1993 AB
Fowleria abocellata Scorpaenodes guamensis Scorpaenidae Goren & Karplus 1983 AB
Fowleria sp. Scorpaenodes guamensis Scorpaenidae Seigel & Adamson 1983 AB
Haemulidae
Pomadasys ramosus* Oligoplites palometa* Carangidae Sazima 2002b AB
Nemipteridae
Pentapodus trivittatus Meiacanthus crinitus Blenniidae Smith-Vaniz et al. 2001 AB
Scolopsis bilineatus* Meiacanthus lineatus,
M. oualanensis,
M. smithi
Blenniidae
Blenniidae
Blenniidae
Springer & Smith-Vaniz 1972
Russell et al. 1976

Smith-Vaniz et al. 2001
ABC
Scolopsis
margaritifer*
Meiacanthus vittatus Blenniidae Russell et al. 1976, Smith-
Vaniz et al. 2001
AB
Blenniidae
Ecsenius bicolor Meiacanthus atrodorsalis Blenniidae Springer & Smith-Vaniz 1972,
Losey 1972
ABC
Ecsenius gravieri Meiacanthus
nigrolineatus
Blenniidae Springer & Smith-Vaniz 1972 ABC
Petroscirtes brevipes Meiacanthus grammistes,
M. kamoharai,
M. vittatus
Blenniidae
Blenniidae
Blenniidae
Smith-Vaniz 1987
Smith-Vaniz et al. 2001
A
Petroscirtes fallax Meiacanthus lineatus Blenniidae Springer & Smith-Vaniz 1972,
Russell et al. 1976, Smith-
Vaniz et al. 2001
ABC
Plagiotremus
laudandus flavus
Meiacanthus oualanensis Blenniidae Springer & Smith-Vaniz 1972,

Losey 1972, Smith-Vaniz
et al. 2001
AC†
Plagiotremus
laudandus laudandus
Meiacanthus atrodorsalis Blenniidae Springer & Smith-Vaniz 1972,
Smith-Vaniz et al. 2001
AC†
Plagiotremus phenax Meiacanthus smithi Blenniidae Springer & Smith-Vaniz 1972,
Smith-Vaniz et al. 2001
AC†
Gobiidae
Amblygobius linki Meiacanthus anema Blenniidae Springer & Smith-Vaniz 1972 A
Valencienna
helsdingenii
Meiacanthus anema Blenniidae Springer & Smith-Vaniz 1972 A
Monacanthidae
Paraluteres arqat Canthigaster margaritata Tetraodontidae Randall & Randall 1960 AB
Paraluteres prionurus Canthigaster valentini Tetraodontidae Tyler 1966, Caley & Schluter
2003
ABC
3597_book.fm Page 465 Friday, May 20, 2005 6:26 PM
© 2005 by R.N Gibson, R.J.A. Atkinson and J.D.M. Gordon
EVEN MOLAND, JANELLE V. EAGLE & GEOFFREY P. JONES
466
patterns from the model Canthigaster valentini, they showed a relatively broad region of protection
surrounding the colour pattern of the model species. Predators did not approach plastic models
even roughly similar in colouration to C. valentini. Their findings indicate that evolution of Batesian
mimicry may begin with only moderate resemblance of a mimic to a model species and that close
resemblance evolves gradually thereafter.

The broad region of protection associated with a particular colour pattern may explain less than
perfect Batesian mimics. For example, small juveniles (<12 cm total length) of Plectropomus laevis
(Serranidae) also resemble Canthigaster valentini, but to a much lesser degree (Randall & Hoese
1986). Future studies focusing on the experimental approach used by Caley & Schluter (2003) for
other proposed cases of Batesian mimicry are needed to attain a greater understanding of how
Batesian mimicry evolves in coral reef fish species.
Of particular interest for studies of Batesian mimicry is the evolution of prey avoidance.
Predators must instantly recognize a particular morphology or behaviour as belonging to highly
toxic or unpalatable prey species (Springer & Smith-Vaniz 1972, Smith 1975, Greene & McDiarmid
1981). This avoidance is likely to be genetically ‘hard-wired’ because there is little scope for
learning from mistakes. Failure to recognise unpalatable species will be fatal if the species is highly
toxic. The same argument may apply to the unusual case of Batesian-type mimicry where cleaner
wrasses, such as the species Labroides dimidiatus, are afforded a special predator ‘amnesty’ due
to the parasite removal service they provide (Kuwamura 1983). Various species of blennies that
mimic this species may also benefit from this predator ‘amnesty’. Presumably a predator would
not ‘learn’ to spare L. dimidiatus from consumption, there being no repercussions from eating it,
but this species is somehow innately recognised as ‘not for consumption’. Whether or not predators
have innate avoidance of certain species based on their colour patterns would help to determine
why some species are mimicked and others are not.
Müllerian mimicry
In Müllerian mimicry, two unpalatable species share similar colour patterns and thus reinforce their
predator deterrence (Müller 1879). Although Bates (1862) commented on the occurrence of two
common unpalatable species with similar phenotypes, it was Müller (1879) who proposed the
Table 2 (continued) Coral reef fish species reported as Batesian mimics and their putative
models, with references and the kind of evidence presented for each species
Mimic Model Model family Reference Evidence
Acanthuridae
Acanthurus pyroferus* Centropyge flavissima,
C. vroliki,
C. heraldi

Pomacanthidae
Pomacanthidae
Pomacanthidae
Randall & Randall 1960
Randall et al. 1997
A
Acanthurus tristis* Centropyge eibli Pomacanthidae Randall & Randall 1960,
Kuiter & Debelius 2001
Notes: A = overall resemblance plus information on means of predator avoidance in model; B = known palatability in
mimic; C = record of predator avoidance in mimic. Sequence of mimic families follows Randall et al. (1997); genera and
species in alphabetical order. Species denoted by * are only mimetic or act as models in the juvenile stages. Format of
table modified from Sazima (2002a).
†
Referring to evidence reported in Smith-Vaniz et al. (2001) suggesting Batesian mimicry with elements of i) Müllerian
mimicry, because some predatory fishes ingest and then reject the mimic (both dead and alive individuals) and ii) aggressive
mimicry, because the Plagiotremus spp. feeds on mucus and epidermis (including scales) of other fishes.
3597_book.fm Page 466 Friday, May 20, 2005 6:26 PM
© 2005 by R.N Gibson, R.J.A. Atkinson and J.D.M. Gordon
MIMICRY IN CORAL REEF FISHES
467
benefits of mimicry for pairs of unpalatable species. If a constant number of unpalatable individuals
must be sacrificed per unit time to teach local predators to avoid a given colour pattern, the fraction
suffering mortality in each species will be reduced if they share a colour pattern. Müllerian mimicry
is a mutualism once attained, but can be sustained as an asymmetric mutualism with unequal
benefits (Mallet & Joron 1999). However, the rare species will gain far more from a mimetic
relationship than will the more common one (Müller 1879).
Müllerian mimicry has not been proposed as the single basis for any mimetic relationship
among coral reef fishes. However, Springer & Smith-Vaniz (1972) and Smith-Vaniz et al. (2001)
propose examples of mixed-type mimicry that include Müllerian mimicry in several interspecific
mimetic complexes. Mimetic complexes are also known as ‘mimicry rings’ and involve more than

two species. (e.g., Losey 1972, Springer & Smith-Vaniz 1972, Russell et al. 1976, Smith-Vaniz
et al. 2001). Many of the known cases of mimetic complexes in coral reef fishes involve one or
more species of the family Blenniidae, in particular from the tribe Nemophini, the ‘sabertoothed’
blennies or ‘fangblennies’. This essentially Indo-Pacific tribe of blennies derives its common name
from an impressive pair of dentary canines, which are used in territorial threat displays or for
defensive purposes. One of the five nemophinine genera, Meiacanthus, possesses a toxic buccal
gland positioned near the base of a pair of deeply grooved canines. Field and laboratory experiments
and observations have revealed that members of the genus Meiacanthus can use their fangs to inject
a noxious substance that causes some predatory fishes to avoid them as potential prey (Losey 1972,
Springer & Smith-Vaniz 1972).
On the basis of this predator deterrence the members of the Meiacanthus genus are reported
as models in several cases of Batesian mimicry (Table 2). However, mimics of Meiacanthus include
members of another genus of the Nemophini, Plagiotremus, which possess fangs but no toxic buccal
glands. In light of the definition of unpalatability “any trait that acts on predators as punishment,
and that causes learning leading to a reduction in attacks” (Mallet & Joron 1999), the fangs of
Plagiotremus may be sufficient to provide some degree of predator protection. Furthermore, because
predatory fishes are known to ingest and then reject both live and dead individuals of Plagiotremus
townsendi, Smith-Vaniz et al. (2001) proposed an element of Müllerian mimicry in the complexes
involving members of both genera. Plagiotremus species feed on mucus, epidermis and the scales
of other fishes, in contrast to Meiacanthus species, which feed on benthic and planktonic crusta-
ceans. Hence there may also be elements of aggressive mimicry in this relationship (Losey 1972,
Springer & Smith-Vaniz 1972). A future study extensively testing the unpalatability of the Pla-
giotremus species involved in mimetic relationships with Meiacanthus species may add strength
to the proposal that Müllerian mimicry is a partial explanation of colour similarities in these
complexes.
Most species of coral reef fishes have some form of predator evasion tactic and it can be argued
that some cases of presumed Batesian or other mimicry could also have elements of Müllerian
mimicry. In the case of the palatable filefish Paraluteres prionurus and toxic pufferfish Canthigaster
valentini, Paraluteres prionurus has some degree of predator protection through a set of retrorse
spines on the caudal peduncle and a semi-erectile first dorsal spine (Tyler 1966, Randall et al.

1997). Similarly, when speculating about the adaptive advantage gained by the mimic surgeonfish
Acanthurus pyroferus and its model pygmy angelfishes from the genus Centropyge, Randall &
Randall (1960) cited the diagnostic spines of each family, caudal blades of the surgeonfish (Acan-
thuridae) and cheek spines of the angelfish (Pomacanthidae), as possible evidence of a Müllerian
relationship. However, given the high rates of mortality reported for a species of Centropyge
(Aldenhoven 1986) and the evidence in support of other types of mimicry for this example, it is
unlikely that Müllerian mimicry is a significant factor in this relationship. Experimental tests of
prey selection by predators comparing preferences for mimics, models and other species are required
to detect Müllerian mimicry in cases such as these.
3597_book.fm Page 467 Friday, May 20, 2005 6:26 PM
© 2005 by R.N Gibson, R.J.A. Atkinson and J.D.M. Gordon
EVEN MOLAND, JANELLE V. EAGLE & GEOFFREY P. JONES
468
Aggressive mimicry
In aggressive mimicry, a predatory species resembles a harmless or beneficial species and conse-
quently achieves increased opportunities for foraging by deceiving prey (Wickler 1965, 1968,
Malcolm 1990). Wickler (1965) cites Poulton (1890) as the first to draw the distinction between
defensive, or protective, mimicry by which the adaptive advantage decreases predation, and aggres-
sive mimicry, which increases feeding opportunities. Aggressive mimicry is reported for 32 species
from 10 families of coral reef fishes, but is most prevalent in the families Serranidae and Blenniidae,
with 13 and 8 species, respectively (Table 3). Wickler (1960, 1961, 1963, 1965, 1968) carried out
pioneer work on aggressive mimicry in coral reef fishes in his studies of the relationship between
the cleaning labrid Labroides dimidiatus and its aggressive mimic, the blenniid Aspidontus taenia-
tus. Larger fishes, expecting only to have their parasites removed, allow A. taeniatus to approach
them and it uses this opportunity to bite off pieces of fins and epidermis (e.g., Eibl-Eibesfeldt 1959,
Randall & Randall 1960, Wickler 1961, Kuwamura 1981, 1983).
However, even seemingly unequivocal cases of aggressive mimicry such as the cleaner wrasse
example are not straightforward. Since some authors had observed A. taeniatus also to feed on
demersal fish eggs (Randall & Randall 1960, Losey 1974, Smith-Vaniz 1976), Kuwamura (1983)
examined the gut contents of 11 specimens of A. taeniatus. According to this study, fish fins were

rarely found in the gut contents and, in fact, the mimic blenny fed mostly on demersal fish eggs
as well as polychaete tentacles. Furthermore, during behavioural observations A. taeniatus rarely
bit pieces from the fins of host fishes even when they were posing for cleaning (Kuwamura 1983).
Kuwamura (1981, 1983) suggests that the low rate of exploitation of posing fish by the mimic is
a strategy to prevent fish visiting cleaning stations from learning its disguise. It is possible that the
main benefit of this mimicry is immunity from predation but mimics may rely on aggressive mimicry
when other kinds of prey are rare (Losey 1978, Kuwamura 1983).
It is also possible that this relationship between A. taeniatus and Labroides dimidiatus is a form
of Batesian-type mimicry, due to amnesty of models rather than non-palatability. During observa-
tions of the same two species in Kimbe Bay, Papua New Guinea, carried out at the same time as
another study (Moland & Jones 2004) we never observed any life history stage of Aspidontus
taeniatus strike fish. However, small groups of two to three individuals were observed to feed on
demersal eggs in nests protected by various egg-caring pomacentrids on several occasions. The
fierce aggression elicited from egg-caring individuals was ignored by the blennies. Being a scav-
enger of demersal fish eggs involves much travelling around in search for food and immunity from
predation in a Batesian-type relationship may facilitate this type of lifestyle. However, aggressive
mimicry could have an important role in the juvenile stage when the blenny does not yet have the
body size required to withstand attacks form egg-caring fishes protecting their nests. In light of
these findings, it seems inappropriate to consider A. taeniatus only as an aggressive mimic.
The above example illustrates why it is necessary to impose strict criteria to distinguish among
types of mimicry in order to assess cases of mimicry. Sazima (2002a) reviewed the literature on
aggressive mimicry in fishes and sorted the evidence into two categories: (1) evidence of overall
resemblance plus information on diet and social behaviour and (2) a record of prey taken by the
mimic under the supposed disguise. Category 2 evidence is vitally important in ruling out the
possibility of an element of social mimicry.
These two categories are used in this review to assess the evidence given in cases of aggressive
mimicry reported for coral reef fishes (Table 3). Of the 32 species reported as aggressive mimics,
less than half (13) include reports of category 2 evidence.
Cases of aggressive mimicry found in coral reef fishes can also be categorised into three trends
described by Sazima (2002a) for fish species exhibiting aggressive mimicry in freshwater cichlids:

(1) species that feed on prey smaller than themselves tend to mimic and join species harmless to
3597_book.fm Page 468 Friday, May 20, 2005 6:26 PM
© 2005 by R.N Gibson, R.J.A. Atkinson and J.D.M. Gordon
MIMICRY IN CORAL REEF FISHES
469
Table 3 Coral reef fish species reported as aggressive mimics and their putative models, with
references and the kind of evidence presented for each species
Mimic Model Model family Reference Evidence
Serranidae
Aethaloperca rogaa* Centropyge spp.
Pomacentrus spp.
Stegastes spp.
Pomacanthidae
Pomacentridae
Pomacentridae
Snyder et al. 2001 AB
Anyperodon leucogrammicus* Halichoeres biocellatus
H. purpurascens (?=H.
leucurus)
H. melanurus
Labridae
Labridae
Labridae
Russell et al. 1976
Randall & Kuiter 1989
AB
Hypoplectrus aberrans Stegastes variabilis Pomacentridae Thresher 1978 A
Hypoplectrus chlorurus Microspathodon chrysurus

Pomacentridae Randall & Randall 1960,

Thresher 1978
A
Hypoplectrus gemma Chromis cyanea Pomacentridae Randall & Randall 1960,
Thresher 1978
A
Hypoplectrus guttavarius Holacanthus tricolor Pomacanthidae Thresher 1978 A
Hypoplectrus nigricans Stegastes adustus Pomacentridae Thresher 1978 A
Hypoplectrus sp. (tan hamlet) Stegastes planifrons Pomacentridae Randall & Randall 1960,
Thresher 1978
A
Hypoplectrus sp.
(blue-black hamlet)
Bodianus rufus Labridae Thresher 1978 A
Mycteroperca acutiorostris* Halichoeres poeyi Labridae Sazima 2002a A
Mycteroperca interstitialis* Halichoeres maculipinna Labridae Sazima 2002a A
Mycteroperca tigris* Thalassoma bifasciatum Labridae DeLoach 1999, Snyder
1999
AB
Plectropomus oligacanthus* Oxychelinus celebicus Labridae Kuiter 1995 A
Pseudochromidae
Pseudochromis fuscus Pomacentrus adelus
P. amboinensis
P. chrysurus
P. moluccensis
Pomacentridae
Pomacentridae
Pomacentridae
Pomacentridae
Munday et al. 2003 AB
Apogonidae

Cheilodipterus zonatus Meiacanthus sp. Blenniidae Russell et al. 1976 A
Carangidae
Oligoplites saurus* Atherinella brasiliensis Atherinidae Sazima & Uieda 1980 AB
Lutjanidae
Lutjanus bohar* Chromis ternatensis
C. flavomaculata,
C. miyakensis
C. weberi
Pomacentridae
Pomacentridae
Pomacentridae
Pomacentridae
Russell et al. 1976,
Moyer 1977
A
Ocyurus chrysurus Mulloidichthys martinicus Mullidae Sikkel & Hardison 1992 AB
Nemipteridae
Scolopsis margaritifer* Meiacanthus sp. Blenniidae Russell et al. 1976 A
Labridae
Oxycheilinus diagrammus Parupeneus macronema
Amblyglyphidodon
leucogaster
Plectroglyphidodon
lacrymatus
Stegastes nigricans
Mullidae
Pomacentridae
Pomacentridae
Pomacentridae
Pomacentridae

Ormond 1980 AB
3597_book.fm Page 469 Friday, May 20, 2005 6:26 PM
© 2005 by R.N Gibson, R.J.A. Atkinson and J.D.M. Gordon
EVEN MOLAND, JANELLE V. EAGLE & GEOFFREY P. JONES
470
Table 3 (continued) Coral reef fish species reported as aggressive mimics and their putative
models, with references and the kind of evidence presented for each species
Mimic Model Model family Reference Evidence
Epibulus insidiator Chaetodon larvatus
Pomacentrus sulfureus
Ctenochaetus striatus
Zebrasoma veliferum
Chaetodontidae
Pomacentridae
Acanthuridae
Acanthuridae
Ormond 1980 AB
Blenniidae
Aspidontus taeniatus Labroides dimidiatus Labridae Randall & Randall 1960,
Kuwamura 1983
AB
Aspidontus filamentosus Labroides dimidiatus Labridae Springer & Smith-Vaniz
1972
AB
Plagiotremus azaleus Thalassoma lucasanum Labridae Hobson 1969 A
Plagiotremus laudandus flavus Meiacanthus oualanensis Blenniidae Springer & Smith-Vaniz
1972, Smith-Vaniz
et al. 2001
A
†

Plagiotremus laudandus
laudandus
Meiacanthus atrodorsalis
Ecsenius bicolor
Blenniidae
Blenniidae
Losey 1972
Springer & Smith-Vaniz
1972, Smith-Vaniz
et al. 2001
A
†
Plagiotremus rhinorhynchos Labroides dimidiatus
Anthias mortoni
Pseudanthias
squamipinnis
Labridae
Serranidae
Serranidae
Wickler 1961
Russell et al. 1976,
Kuwamura 1981, 1983,
Randall et al. 1997,
Moland & Jones 2004
Côté & Cheney 2004
AB
Plagiotremus tapeinosoma Thalassoma
amblycephalum
Trachinops taeniatus,
Forsterygion sp.

Labridae
Plesiopidae
Trypterygiidae
Russell et al. 1976 AB
Plagiotremus townsendi Meiacanthus nigrolineatus
Ecsenius gravieri
Blenniidae
Blenniidae
Springer & Smith-Vaniz
1972
Smith-Vaniz et al. 2001
AB
†
Chaenopsidae
Hemiemblemaria simulus Thalassoma bifasciatum Labridae Randall & Randall 1960,
DeLoach 1999
A
Acanthuridae
Acanthurus pyroferus* Centropyge vroliki
C. flavissima
C. heraldi
Pomacanthidae
Pomacanthidae
Pomacanthidae
Eagle & Jones 2004 A
Acanthurus tristis* Centropyge eibli Pomacanthidae Eagle & Jones 2004 A
Notes: A = overall resemblance plus information on diet and social behaviour; B = record of taking prey under the supposed
disguise. Sequence of mimic families follows Randall et al. (1997); genera and species in alphabetical order. Species denoted
by * are only mimetic in the juvenile stages. Format of table adapted from Sazima (2002a).
† Referring to evidence reported in Smith-Vaniz et al. (2001) suggesting aggressive mimicry with elements of i) Batesian

mimicry, as the Meiacanthus spp. possesses predator avoidance through venomous fangs, and ii) Müllerian mimicry, as
some predatory fishes ingest and then reject the mimic Plagiotremus spp. (both dead and alive individuals).
3597_book.fm Page 470 Friday, May 20, 2005 6:26 PM
© 2005 by R.N Gibson, R.J.A. Atkinson and J.D.M. Gordon
MIMICRY IN CORAL REEF FISHES
471
their prospective prey, (2) species that feed on prey larger than themselves tend to mimic mostly
beneficial species (cleaners) or, less frequently, join species harmless to their prospective prey,
(3) species that feed on prey about their own size tend to mimic their prospective prey species.
Species typically fitting trend 1 are the hamlets (Hypoplectrus spp.; Serranidae), which feed on
small fishes and crustaceans while their models, wrasses, damselfishes and an angelfish are plank-
tivores or omnivores (Lieske & Myers 1996). Trend 2 is appropriate for the blennies reported as
aggressive mimics. The two Aspidontus species and juvenile Plagiotremus rhinorhynchos are
mimics of the cleaner wrasse Labroides dimidiatus, while other blenniids mimic species harmless
to their prey. Currently there are no definitive examples among coral reef fishes of trend 3, although
the aggressive mimic Pseudochromis fuscus mimics adult damselfish in order to fool the juveniles
of these damselfish species that it preys upon. While the mimic and adult models of these species
are approximately the same size as required for trend 3, the actual prey (juveniles) are much smaller.
This may also be the case for larger specimens of particular mimic hamlets but has not been
explored in detail (Fischer 1980).
Social mimicry
‘Social’ mimicry refers to examples of mimics that associate with similarly coloured individuals
in order to escape the attention of predators (Moynihan 1968, Dafini & Diamant 1984, Randall &
McCosker 1993, Smith-Vaniz et al. 2001). The term ‘social mimicry’ was proposed to describe the
conspicuous colouration and signal patterns adapted to promote formation and maintain cohesion
of mixed flocks of birds (Moynihan 1968). Moynihan (1968) correctly suggested that the study of
fishes might demonstrate the same phenomenon. Dafini & Diamant (1984) first observed social
mimicry in fishes and referred to the phenomenon as ‘school-oriented’, a term that was later
identified by Randall & Kuiter (1989) as congruous with the social mimicry defined by Moynihan
(1968).

Social mimicry is reported for eight species from six families of coral reef fishes and is most
prevalent in the Blenniidae (three species) (Table 4). When reporting cases of social mimicry, the
evidence should ideally include records of the following: (1) overall resemblance plus information
on interspecific schooling behaviour and (2) as above plus record of foraging on the same food
source as the putative model species. Type 2 is important in order to rule out a possible element
of aggressive mimicry. Aggressive mimicry can be argued if the mimic forages at a higher trophic
level than its model. There are three reported cases of social mimicry for which type 2 evidence
has not been recorded (Table 4).
In fishes, social mimicry is defined as where two or more species with the same colour pattern
and similar shape school together for mutual protection from predators (Randall & McCosker
1993). Half of the eight reported cases of social mimicry involves a species mimetic only during
its juvenile stage, which is when they are likely to be most susceptible to predation (Jones &
McCormick 2002). The occurrences and advantages of interspecific group foraging are well known
in the literature (Fishelson 1977, Ormond 1980, Foster 1985, Strand 1988) and include increased
freedom to forage due to the sharing of predator vigilance when in a group (e.g., Sakai & Kohda
1995). There is no real need for individuals in groups to look the same, although some individuals
are known to select groups of similarly coloured individuals (Crook 1999). For rare species, close
resemblance to more common species may allow them to ‘hide’ amongst other fishes and avoid
standing out to predators as obviously different. If this is the case, mimicry may well be a process
that helps to maintain diversity in coral reef fish assemblages (Gilbert 1983).
As described for other types of mimicry, there are examples of mimetic relationships in which
some element of social mimicry is apparent even though it may not be the main adaptive advantage.
3597_book.fm Page 471 Friday, May 20, 2005 6:26 PM
© 2005 by R.N Gibson, R.J.A. Atkinson and J.D.M. Gordon
EVEN MOLAND, JANELLE V. EAGLE & GEOFFREY P. JONES
472
For example, the Batesian mimic filefish Paraluteres prionurus shows remarkable similarity to the
pufferfish Canthigaster valentini in habitat affinity and behaviour. Caley & Schluter (2003) observed
individuals of Paraluteres prionurus shelter among groups of the model species Canthigaster
valentini when approached by a diver. This close behavioural association between mimic and model

is also observed between juvenile mimic surgeonfish Acanthurus pyroferus and one of its mimics
Centropyge vroliki (Eagle & Jones 2004). For typical Batesian and Müllerian mimicry it should be
sufficient just for the mimic to exist in the vicinity of the model and where predators would avoid
mimics after learning from encounters with the model regardless of whether they occurred together.
Whereas there are likely to be some cases of pure social mimicry in coral reef fishes, if close spatial
association is an ecological requirement for mimicry, it is likely that it will be difficult to distinguish
social mimicry from close spatial association in other types.
Can the classical framework be applied?
Taking the published evidence at face value, aggressive mimicry is the most prevalent type of
mimicry in coral reef fishes, constituting 48% of all cases reported to date, followed by Batesian
(40%) and social mimicry (12%). Examples of all three types appear to be present in the families
Blenniidae and Apogonidae (Tables 2–4). However, as discussed above, relatively few studies meet
Table 4 Coral reef fish species reported as social mimics and their putative models, with references
and the kind of evidence presented for each species
Mimic Model Model family Reference Evidence
Apogonidae
Apogon compressus* Cheilodipterus macrodon* Apogonidae Kuiter 1990, 1992, Smith-
Vaniz et al. 2001
A
Malacanthidae
Hoplolatilus starcki Pseudanthias pascalus
P. tuka
Serranidae
Serranidae
Randall & McCosker 1993 AB
Lutjanidae
Apsilus dentatus* Chromis cyanea Pomacentridae Bunkley-Williams &
Williams 2000
AB
Mullidae

Mulloidichthys mimicus Lutjanus kasmira Lutjanidae Randall & Guézé 1980,
Randall & McCosker 1993
A
Pomacentridae
Lepidozygus tapeinosoma Luzonichthys whitleyi
Pseudanthias bartlettorum,
P. dispar
Serranidae
Serranidae
Serranidae
Randall & McCosker 1993 AB
Blenniidae
Ecsenius midas Luzonichthys whitleyi
Pseudanthias bartlettorum,
P. dispar
P. squamipinnis
Serranidae
Serranidae
Serranidae
Serranidae
Randall & McCosker 1993
Starck 1969
AB
Meiacanthus nigrolineatus* Cheilodipterus spp.* Apogonidae Dafini & Diamant 1984 AB
Meiacanthus urostigma* Cheilodipterus
quinquelineatus*
Apogonidae Smith-Vaniz et al. 2001 A
Notes: A = overall resemblance plus information on interspecific schooling behaviour; B = record of foraging on the same
food source as model. Sequence of mimic families follows Randall et al. (1997); genera and species in alphabetical order.
Species denoted by * are only mimetic or act as models in the juvenile stages. Format of table modified from Sazima (2002a).

3597_book.fm Page 472 Friday, May 20, 2005 6:26 PM
© 2005 by R.N Gibson, R.J.A. Atkinson and J.D.M. Gordon
MIMICRY IN CORAL REEF FISHES
473
all the criteria necessary to classify cases into the four types of mimicry. Studies that not only
provide observations and quantify ecological relationships but also experimentally distinguish
among the different hypotheses (Munday et al. 2003, Moland & Jones 2004) provide the best
evidence. However, there are clearly serious limitations with the approach of classifying all cases
of mimicry in reef fishes into traditional categories.
It is now commonly recognised that mimicry among coral reef fishes may be complex and
involve more than one selective mechanism, including both foraging advantages and predator
avoidance (Smith-Vaniz et al. 2001, Caley & Schluter 2003, Eagle & Jones 2004). Many cases of
mimicry are difficult to classify because they contain elements of two or more recognized forms.
Establishing one selective advantage does not necessarily eliminate another. In mimetic complexes
involving more than two species, the different types of mimicry are not mutually exclusive and
elements of Batesian, Müllerian, aggressive and social mimicry are all potentially present (e.g.,
Losey 1972, Springer & Smith-Vaniz 1972, Russell et al. 1976, Smith-Vaniz et al. 2001).
Distinguishing between the relative benefits of decreased predation pressure and increased
foraging opportunity is difficult, yet crucial to understanding the evolution of mimicry. One par-
ticularly difficult case to categorise is the blenny mimicry of cleaner wrasses. There are three
potential underlying bases of the mimicry of the cleaner wrasse Labroides dimidiatus by the blenny
Aspidontus taeniatus: (1) an ‘amnesty’ from predation, (2) aggressive mimicry in terms of the
opportunity to bite food from fish posing to be cleaned and (3) aggressive mimicry in terms of
gaining access to demersal egg masses defended by territorial damselfishes. Adaptive advantages
(1) and (3) are slightly unusual cases within their categories, where the anti-predation mechanism
is not ‘unpalatability’ but ‘amnesty’, and the aggressive mimicry does not deceive the prey, but
rather the species defending the prey.
The other case of mimicry where multiple advantages apply is the mimic surgeonfish, where
there are also benefits from reduced predation pressure and increased access to food. The traditional
view of the relationship between juvenile Acanthurus pyroferus and A. tristis and their Centropyge

models is that the adaptive advantage is decreased predation pressure due to the notion that the
‘secretive’ models are difficult to catch (Randall et al. 1990, Kuiter & Debelius 2001). Centropyge
spp. spend a large proportion of their time sheltering in reef crevices. When venturing out to feed
they closely hug the contours of reef topography and are very quick to hide in the presence of
perceived danger. Kuiter & Debelius (2001) hypothesise that the mimicry is linked to the relative
sizes at which the mimic and model species settle. Juvenile Acanthurus pyroferus settle out of the
plankton at a much larger size than Centropyge spp. Hence, by mimicking juvenile angelfishes,
surgeonfishes may benefit from their greater experience in predator avoidance. If this is the basis
for this case of mimicry, it is another unusual case of Batesian mimicry where the model species
is not ‘unpalatable’; instead it is avoided due to a low ‘catch-per-unit-effort’ experienced by
predators.
While an anti-predation adaptive advantage cannot be entirely ruled out, Eagle & Jones (2004)
provide quantitative results that indicate mimic Acanthurus pyroferus gain greater foraging time as
well as greater access to algal food resources because they resemble and associate with their models.
Like the cleaner wrasse mimics, mimic surgeonfishes feed on a demersal food source also guarded
by aggressive territorial damselfishes. However, damselfishes attacked models and mimics foraging
in algal territories approximately half the number of times they attacked other surgeonfishes,
including adult A. pyroferus. This may be because the diet of A. pyroferus was most similar to that
of damselfishes, whereas angelfishes consumed different types of algae from those preferred by
the damselfishes (Eagle & Jones 2004).
The highly aggressive and territorial damselfishes (Pomacentridae) may be particularly influ-
ential on the behavioural adaptations of coral reef fishes in evolutionary terms. This family contains
most of the species that lay and guard demersal eggs, as well as the species that ‘farm’ algae. In
3597_book.fm Page 473 Friday, May 20, 2005 6:26 PM
© 2005 by R.N Gibson, R.J.A. Atkinson and J.D.M. Gordon
EVEN MOLAND, JANELLE V. EAGLE & GEOFFREY P. JONES
474
both the mimic cleaners and mimic surgeonfish cases discussed here, mimicry may have evolved
to reduce the risk of being attacked by damselfishes guarding a food source, as well as to reduce
the risk of predation to allow individuals more time to search for these food sources. Future

experiments including (1) prey choice experiments by predators of models, mimics and other
species, (2) removal of the model in an area, followed by monitoring of the behaviour, growth and
fate of the mimics and (3) ‘bottle experiments’ to test aggressive displays from territorial fishes
may yield new answers and perspectives on the importance of each of these likely benefits. The
division of mimicry into its classical types is useful in determining whether there is enough evidence
to suggest that the resemblance of two species is in fact mimicry. However, these two examples
highlight the problem that types are not mutually exclusive and closer inspection usually reveals
greater complexity in mimic-model relationships.
Is mimicry ecologically important in coral reef
fish assemblages?
Mimicry is thought to be an important ecological process potentially maintaining the abundance
of rare species and so promoting species diversity (Gilbert 1983) but this hypothesis has proved
difficult to evaluate. It is notable that mimicry appears to be most prevalent in the high-diversity
tropical communities composed of butterflies, birds and freshwater and coral reef fishes (Trewavas
1947, Cody 1969, Sazima 1977, Brosset 1997, Hori & Watanabe 2000). These are all highly
colourful species assemblages found in habitats where colour appears to be an important signalling
mechanism (Marshall 2000). However, whether mimicry is a cause or an effect of high diversity
is difficult to assess. Mimicry may evolve only in communities that already have high species
diversity, where there would be an increased probability that two species might fortuitously resemble
one another in the first place (Domeier 1994).
Based on the number of cases reported, mimicry may appear unimportant as an ecological and
evolutionary force in coral reef fishes. The 60 mimetic species reported to date represent only
~1.5% of an estimated fish diversity of about 3750 species inhabiting coral reefs globally (Lieske &
Myers 1996). In a brief review of the status of mimicry in the context of coral reef fish ecology,
Sale (1980) concluded that there was no evidence indicating that mimicry or other behavioural
interactions determine the species composition of reef fish assemblages. Nevertheless, the experi-
mental evidence to date suggests that mimicry is crucial to explaining the distribution and abundance
of the species that have been examined, allowing them to persist as rare species and increase in
abundance in places where models are abundant.
The conclusion that mimicry is not ecologically important to reef fish communities as a whole

is probably premature. The most important question, from an ecological perspective, is how many
cases of mimicry remain to be discovered. While most of the cases of ‘perfect’ mimicry have
probably been identified, it is likely that most of the cases of ‘imperfect’ mimicry have not. The
study by Munday et al. (2003), which shows that different colour morphs of Pseudochromis fuscus
associate with a different range of similarly coloured damselfishes, highlights the potential advan-
tages of being a poor mimic that can pass for a range of different species (Edmunds 2000, Sherratt
2002). The process of identifying associations between reef fish species only superficially similar
in body colouration is in the early stages.
In coral reef fishes, mimicry may be particularly important to the survival of smaller species
or juveniles of larger species. The prevalence of mimicry in smaller individuals may be part of a
suite of other anti-predatory strategies associated with body size. Small individuals are more likely
to be eaten by predators and, where confined to restricted habitats for shelter, are more likely to
compete among themselves for resources (Munday & Jones 1998, Jones & McCormick 2002). A
3597_book.fm Page 474 Friday, May 20, 2005 6:26 PM
© 2005 by R.N Gibson, R.J.A. Atkinson and J.D.M. Gordon
MIMICRY IN CORAL REEF FISHES
475
large number of coral reef fishes exhibit a juvenile colouration that is distinct from adults but the
degree to which these colours function in mimicry, crypsis, predator warning or intra-specific
signalling remains to be fully evaluated.
Is mimicry under genetic or phenotypic control?
The extent to which mimic colour patterns are under either genetic or phenotypic control, or both,
is mostly unresolved. Many polychromatic species appear to exhibit mimic forms but the underlying
genetic basis of differences in colour is unclear. The presence of distinct colour morphs in mimetic
and other coral reef fishes has resulted in confusion about species boundaries. There are cases
where (1) recognised species are genetically indistinct but interbreed where their distributions
overlap, forming hybrids with gradients in colour pattern between the two species (McMillan et al.
1999), (2) a widely distributed species with a uniform colour pattern actually comprises genetically
distinct populations in different geographic areas (Bernardi et al. 2001), (3) colour morphs display
assortative mating, yet have no genetic differentiation and are considered the same species (Doherty

et al. 1995, Planes & Doherty 1997a,b) and (4) cases where species display assortative mating,
have no genetic differentiation and are considered different species, which is the case for the mimetic
hamlets Hypoplectrus spp. (Thresher 1978).
Thresher (1978) proposed that mimicry might be the driving force behind speciation in the
hamlets, because mimics are required to be rare compared with models. Once a mimic population
reaches a proportion of the size of a model population, speciation may be favoured if the adaptive
advantage is lost through learning behaviour of the signal receiver. However, whether or not colour
morphs represent different species within the genus Hypoplectrus has been debated for over 20 yr
(Thresher 1978, Fischer 1980, Graves & Rosenblatt 1980, Domeier 1994, Ramon et al. 2003,
McCartney et al. 2003, García-Machado et al. 2004). Over 10 morphospecies of Hypoplectrus are
distinguishable only by colour pattern, despite detailed studies of biological and ecological attributes
(Fischer 1980, Domeier 1994) and genetic composition (Graves & Rosenblatt 1980, Ramon et al.
2003, McCartney et al. 2003, García-Machado et al. 2004). Most authors speculate that colour
morphs originated in geographic isolation but there is current gene flow, or that this is a case of
speciation in progress. In support of prior speciation are patterns of abundance that indicate potential
historic centres of origin for morphospecies (Domeier 1994) and behavioural preferences for mating
with like-coloured individuals, even though hybridisation does occur where species are rare (Fischer
1980, Domeier 1994). Whereas three studies of genetic structure found no genetic differentiation
among morphospecies (Graves & Rosenblatt 1980, Ramon et al. 2003, Garcia-Machado et al 2004),
McCartney et al. (2003) claim to have found some non-random separation between hamlets from
different geographical regions, and among morphospecies at some locations.
Many fishes are capable of dramatic colour changes in response to changes in their social and
physical environment (Crook 1997, Arigoni et al. 2002). The colours displayed by mimics may be
equally plastic and dependent upon local conditions. For example, when model Labroides dimid-
iatus were removed from the vicinity of juvenile Plagiotremus rhinorhynchos mimics, the mimics
rapidly lost their model coloration (Moland & Jones 2004). The two colour morphs of the dottyback
Pseudochromis fuscus, both of which appear to be mimics, have recently shown to lack genetic
differentiation and individuals are capable of switching between colour morphs (Messmer et al. in
press). Similarly, we have observed that Acanthurus pyroferus individuals, when they are at the
size that they begin to lose the mimic coloration of model Centropyge vroliki, ‘flash’ the adult

coloration when in the presence of conspecific adults, demonstrating a physiological colour change.
Mimic Acanthurus pyroferus, which may resemble one of three different Centropyge species, may
have flexibility in colour differentiation, depending on the species that is most abundant when they
3597_book.fm Page 475 Friday, May 20, 2005 6:26 PM
© 2005 by R.N Gibson, R.J.A. Atkinson and J.D.M. Gordon
EVEN MOLAND, JANELLE V. EAGLE & GEOFFREY P. JONES
476
recruit to the reef. A. pyroferus also appears to have an apparently non-mimetic form that has
coloration somewhat intermediate between the yellow Centropyge flavissima and C. heraldi, and
the brown C. vroliki (Kuiter & Debelius, 2001). This colouration may be adopted if there are no
model species in close proximity.
Mimicry and colour: what do the fishes see?
To date, our perception of the resemblance of mimics and models has been based on the human
perception of colour. The prevalence of mimicry in the highly coloured fishes on coral reefs may
be associated with the water clarity and the high light intensity in the shallow water habitat. But
what do the fishes themselves perceive? The answer to this question may totally change current
views as to the prevalence and importance of mimicry. It is known that visual perception in fishes
is relatively good, even from an early stage (Fernald 1993, Shand 1997, Hawryshyn 1998) but also
that reef fishes do not see the same colour spectrum that humans see and individual species see
differently from one another (Marshall 2000). Therefore, it is as yet unclear to what extent mimics
are able to deceive signal receivers.
Knowledge of the visual acuity of different reef fish groups may enable us to more precisely
identify mimic-model pairs and establish the signal receivers in mimetic relationships. To date,
combinations of the 21 major categories of colours have been characterised for only about 200
coral reef fish species, the transmission of ocular media is known for only 185 species and spectral
sensitivities for only 20 species (Marshall 2000). Some species, such as the planktivorous damselfish
Chromis atripectoralis, have ocular media that transmit light down to shorter wavelengths of 300 nm
including ultraviolet (UV). Others, such as the angelfish Pygoplites diacanthus, have ocular media
that pass only wavelengths above 400 nm (Marshall 2000). Interestingly, the colours on the body
pattern of P. diacanthus include four that have significant UV components with peak reflectance

in wavelengths between 300–400 nm but these cannot be seen by conspecifics. These colours are
presumably part of a signalling repertoire to other species (Marshall 2000). Nevertheless, because
species-specific recognition systems in fishes are well developed (Thresher 1974, 1979, Barry &
Hawryshyn 1999), individuals masquerading as members of a model species are unlikely to deceive
the models themselves. Therefore, other species, such as predators or competitors, might also not
be easily fooled. Because gobies and damselfishes can distinguish between mimic hamlets and
models (Thresher 1976, 1979), it has been proposed that mimicry in this situation is directed at
deceiving crustacean prey (Thresher 1978). However, fishes have been found in the gut contents
of larger individuals of some hamlet morphospecies (Fischer 1980). Further information about the
colours that are transmitted, as well as what can be seen by mimics, models and intended signal
receivers, would provide useful information to help determine the adaptive advantage of many
cases of mimicry.
It is not clear whether or not certain colours are more often involved in mimetic relationships
than others. A number of coral reef fishes with the same genetic species identity are characterised by
dimorphism involving ‘yellow’ and ‘dark’ variants, including the butterflyfish Forciper longirostris,
the goatfish Parupeneus cyclostomus, the wrasse Epibulus insidiator, the trumpetfish Aulostomus
chinensis and the dottyback Pseudochromis fuscus (Randall et al. 1997). Whereas brown is a good
camouflage colour on the reef, yellow is thought to be a good signal colour because it is transmitted
well in oceanic water (Marshall 2000). Whether there is a common underlying mechanism for the
evolution of yellow coloration among these different taxa is unknown but it does occur in many
examples of mimicry (Munday et al. 2003). For example, the vivid colours of some of the Hypoplec-
trus species include yellow and are an unusual feature among serranids that, as reef hunters,
generally have disruptive coloration.
3597_book.fm Page 476 Friday, May 20, 2005 6:26 PM
© 2005 by R.N Gibson, R.J.A. Atkinson and J.D.M. Gordon
MIMICRY IN CORAL REEF FISHES
477
Many juvenile surgeonfish species exhibit yellow morphs that have been implicated in mimicry.
The yellow form of juvenile Acanthurus pyroferus, a mimic of the yellow pygmy angelfish Cen-
tropyge flavissima, is the most well-known example. However, a number of other surgeonfish species

often have conspicuous yellow juveniles, including Acanthurus olivaceus, A. tennentii, Ctenocha-
etus flavicauda, C. strigosus and Zebrasoma scopas (Kuiter & Debelius 2001). Guiasu & Winter-
bottom (1998) observed that the yellow juveniles of Ctenochaetus strigosus were subject to less
aggression from territorial damselfishes than other surgeonfish juveniles. They hypothesised that
this may be due to its similarity to Centropyge flavissima (one of the Acanthurus pyroferus model
species), which did not seem to be attacked. Some Ctenochaetus strigosus juveniles, besides being
all yellow, actually have the blue eye and operculum markings that both Centropyge flavissima and
Acanthurus pyroferus mimics of this species have (Myers 1999). The mimetic forms of A. pyroferus
may have developed from initially plain yellow juveniles, which based on its appearance in other
taxa, may be an ‘easy’ morph to produce. If a plain yellow A. pyroferus morph had survival
advantages in areas where the two yellow species of model angelfishes (Centropyge flavissima and
C. heraldi) were abundant, then further mimic morphs of C. vroliki and C. eibli may have simply
coevolved with this sister group of Centropyge. Knowledge about the history of speciation in models
and mimics would provide further clues about the evolution of mimicry in coral reef fishes.
Conclusions
This review indicates that studies of mimicry in coral reef fishes have moved into a new phase.
Anecdotal accounts of mimicry are being superseded by detailed studies of the ecological and
behavioural relationships between mimic and model pairs, and critical tests of hypotheses concern-
ing the underlying selective advantages of mimicry. Whereas most cases have still not been fully
validated, evidence to date suggests that individual cases of mimicry in coral reef fishes may be
complex and are not easily classified according to a single adaptive mechanism. Many instances
in which mimics enjoy both foraging advantages and reduced likelihood of predation have been
identified but the relative importance of these is unknown.
Despite the recent advances, there are many other questions that remain unanswered and others
that are only beginning to be asked. The impacts of model species on the distribution and abundance
of mimics are largely unknown and the costs to the model species are only beginning to be evaluated
(Côté & Cheney 2004). A complete understanding of the ecological and evolutionary significance
of mimicry in reef fishes must await a full appraisal of the prevalence of mimicry in different
taxonomic groups. Whereas the ‘good’ mimics have been identified, there are likely to be many
examples of ‘poor’ mimics that have not come under scrutiny and it is not known how close colour

resemblance needs to be to qualify as mimicry. This appraisal must take into account a greater
understanding of the role of vision and colour in reef fishes because what appears to be a mimic
to the human eye may not be, and there may be other mimics that we cannot perceive. Our
understanding of the phenotypic and genetic basis of mimicry is just beginning, so far suggesting
minimal genetic differentiation between different mimics of the same species and a high degree of
phenotypic plasticity in mimetic colouration. These many gaps in our knowledge highlight the
continued need for multidisciplinary and multifactorial approaches to our understanding of the
mimicry phenomenon.
Acknowledgements
The structure of our review was built upon the strong foundations laid down by the pioneers of
fish mimicry Wolfgang Wickler, John E. Randall and William F. Smith-Vaniz. Even Moland
3597_book.fm Page 477 Friday, May 20, 2005 6:26 PM
© 2005 by R.N Gibson, R.J.A. Atkinson and J.D.M. Gordon
EVEN MOLAND, JANELLE V. EAGLE & GEOFFREY P. JONES
478
expresses appreciation to his early teachers, his father Anders Moland and Dr. Magne Saetersdal
who sparked a fascination for the mimicry phenomenon. Discussions with Julian Caley, Philip
Munday and Mark McCormick have contributed to the ideas expressed in this review. We are
indebted to Rudie Kuiter, whose eye for mimicry can be seen in the photographic images he kindly
allowed us to use. Our work has been supported by an Australian Research Council grant to G.P.J.
References
Aldenhoven, J.M. 1986. Local variation in mortality rates and life-expectancy estimates of the coral-reef fish
Centropyge bicolor (Pisces: Pomacanthidae). Marine Biology 92, 237–244.
Allen, G.R., Russell, B.C., Carlson, B.A. & Starck W.A. 1975. Mimicry in marine fishes. Tropical Fish
Hobbyist 24, 47–56.
Allen, G.R., Steene, R. & Allen, M. 1998. A Guide to Angelfishes and Butterflyfishes. Perth: Odyssey
Publishing/Tropical Reef Research.
Arigoni, S., Francour, P., Harmelin-Vivien, M.L. & Zaninetti, L. 2002. Adaptive colouration of Mediterranean
labrid fishes to the new habitat provided by the introduced tropical alga Caulerpa taxifolia. Journal
of Fish Biology 60, 1486–1497.

Barlow, G.W. 1975. On the sociobiology of some hermaphroditic serranid fishes, the hamlets, in Puerto Rico.
Marine Biology 33, 295–300.
Barry, K.L. & Hawryshyn, C.W. 1999. Spectral sensitivity of the Hawaiian saddle wrasse, Thalassoma
duperrey, and implications for visually mediated behaviour on coral reefs. Environmental Biology of
Fishes 56, 429–442.
Bates, H.W. 1862. Contributions to an insect fauna of the Amazon Valley. Lepidoptera: Heliconidae. Trans-
actions of the Linnean Society of London 23, 495–566.
Bellwood, D.R. 1997. Reef fish biogeography: habitat associations, fossils and phylogenies. In Proceedings
of the 8th International Coral Reef Symposium, Panama, 24 to 29 June 1996, Volume 1, H.A. Lessios &
I.G. Macintyre (eds). Smithsonian Tropical Research Institute: Panama, 379–384.
Bernardi, G., Holbrook, S.J. & Schmitt, R.J. 2001. Gene flow at three spatial scales in a coral reef fish, the
three-spot dascyllus, Dascyllus trimaculatus. Marine Biology 138, 457–465.
Brosset, A. 1997. Aggressive mimicry by the characid fish Erythrinus erythrinus. Ethology 103, 926–934.
Bunkley-Williams, L. & Williams, E.H.J. 2000. Juvenile black snapper, Apsilus dentatus (Lutjanidae), mimic
blue chromis, Chromis cyanea (Pomacentridae). Copeia 2000, 579–581.
Caley, M.J. & Schluter, D. 2003. Predators favour mimicry in a tropical reef fish. Proceedings of the Royal
Society of London Series B 270, 667–672.
Cody, M.L. 1969. Convergent characteristics in sympatric species: a possible relation to interspecific compe-
tition and aggression. The Condor 71, 222–239.
Côté, I.M. & Cheney, K.L. 2004. Distance-dependent costs and benefits of aggressive mimicry in a cleaning
symbiosis. Proceedings of the Royal Society of London Series B 271, 2627–2630.
Crook, A.C. 1997. Colour patterns in a coral reef fish. Is background complexity important? Journal of
Experimental Marine Biology and Ecology 217, 237–252.
Crook, A.C. 1999. Quantitative evidence for assortative schooling in a coral reef fish. Marine Ecology Progress
Series 176, 17–23.
Dafini, J. & Diamant, A. 1984. School-oriented mimicry, a new type of mimicry in fishes. Marine Ecology
Progress Series 20, 45–50.
Debelius, H., Tanaka, H. & Kuiter, R.H. 2003. Angelfishes: A Comprehensive Guide to Pomacanthidae.
Chorleywood, U.K.: TMC Publishing.
DeLoach, N. 1999. Reef Fish Behaviour: Florida, Caribbean, Bahamas. Jacksonville: New World Publishers.

Doherty, P.J., Planes, S. & Mather, P. 1995. Gene flow and larval duration in seven species of fish from the
Great Barrier Reef. Ecology 76, 2373–2391.
Domeier, M.L. 1994. Speciation in the serranid fish Hypoplectrus. Bulletin of Marine Science 54, 103–141.
Eagle, J.V. & Jones, G.P. 2004. Mimicry in coral reef fishes: ecological and behavioural responses of a mimic
to its model. Journal of Zoology 264, 33–43.
3597_book.fm Page 478 Friday, May 20, 2005 6:26 PM
© 2005 by R.N Gibson, R.J.A. Atkinson and J.D.M. Gordon
MIMICRY IN CORAL REEF FISHES
479
Eagle, J.V., Jones, G.P. & McCormick, M.I. 2001. A multi-scale study of the relationships between habitat
use and the distribution and abundance patterns of three coral reef angelfishes (Pomacanthidae).
Marine Ecology Progress Series 214, 253–265.
Edmunds, M. 2000. Why are there good and poor mimics? Biological Journal of the Linnean Society 70, 459–466.
Ehrlich, P.R. 1975. Population biology of coral reef fishes. Annual Review of Ecology and Systematics 6,
211–247.
Eibl-Eibesfeldt, I. 1959. Der Fish Aspidontus taeniatus als Nachamer der Putzers Labroides dimidiatus.
Zeitschrift für Tierpsychologie 16, 19–25.
Fernald, R.D. 1993. Vision. In The Physiology of Fishes, D.H. Evans (ed.). Boca Raton, Florida: CRC Press,
161–189.
Fischer, E.A. 1980. The relationship between mating system and simultaneous hermaphroditism in the coral
reef fish, Hypoplectrus nigricans (Serranidae). Animal Behaviour 28, 620–633.
Fishelson, L. 1977. Sociobiology of feeding behaviour of coral fish along the coral reef of the Gulf of Elat
(=Gulf of Aqaba), Red Sea. Israel Journal of Zoology 26, 114–134.
Foster, S.A. 1985. Group foraging by a coral reef fish: a mechanism for gaining access to defended resources.
Animal Behaviour 33, 782–792.
García-Machado, E., Chevalier Monteagudo, P.P. & Solignac, M. 2004. Lack of mtDNA differentiation among
hamlets (Hypoplectrus, Serranidae). Marine Biology 144, 147–152.
Gilbert, L.E. 1983. Coevolution and mimicry. In Coevolution, D.J. Futuyama & M. Slatkin (eds). Sunderland,
Massachusetts: Sinauer Associates, Inc., 263–281.
Gladstone, W. 1987. The eggs and larvae of the sharpnose pufferfish Canthigaster valentini (Pisces: Tetra-

odontidae) are unpalatable to other reef fishes. Copeia 1987, 227–230.
Gon, O. 1993. Revision of the cardinalfish genus Cheilodipterus (Perciformes: Apogonidae), with description
of five new species. Indo-Pacific Fishes 22, 1–59.
Goren, M. & Karplus, I. 1983. Preliminary observations on the scorpionfish Scorpaenodes guamensis and its
possible mimic the cardinalfish Fowleria abocellata. Developments in Ecology and Environmental
Quality 2, 327–336.
Graves, J.E. & Rosenblatt, R.H. 1980. Genetic relationships of the color morphs of the serranid fish Hypoplec-
trus unicolor. Evolution 34, 240–245.
Greene, H.W. & McDiarmid, R.W. 1981. Coral snake mimicry: does it exist? Science 213, 1207–1212.
Guiasu, R.C. & Winterbottom, R. 1998. Yellow juvenile color pattern, diet switching and the phylogeny of
the surgeonfish genus Zebrasoma (Percomorpha, Acanthuridae). Bulletin of Marine Science 63,
277–294.
Hawryshyn, C.W. 1998. Vision. In The Physiology of Fishes, D.H. Evans (ed.). Boca Raton, Florida: CRC
Press, 345–374.
Hobson, E.S. 1969. Possible advantages to the blenny Runula azeala in aggregating with the wrasse Thalas-
soma lucasanum in the tropical eastern Pacific. Copeia 1969, 191–193.
Hori, M. & Watanabe, K. 2000. Aggressive mimicry in the intra-populational color variation of the Tanganyikan
scale-eater Perissodus microlepis (Cichlidae). Environmental Biology of Fishes 59, 111–115.
Jones, G.P. & McCormick, M.I. 2002. Numerical and energetic processes in the ecology of coral reef fishes.
In Coral Reef Fishes: Dynamics and Diversity in a Complex Ecosystem, P.F. Sale (ed.). San Diego:
Academic Press, 221–339.
Kuiter, R.H. 1990. Nature’s copies. Sportdiving 23, 114–116.
Kuiter, R.H. 1992. Tropical Reef-fishes of the Western Pacific Indonesia and Adjacent Waters. Jakarta: Penerbit
PT, Gramedia Pustaka Utama.
Kuiter, R.H 1995. The juvenile vermicular cod Plectropomus oligacanthus, a mimic of the slender maori
wrasse Cheilinus celebicus. Revue Francaise d’Aquariologie 21, 77–78.
Kuiter, R.H. 1996. Guide to Sea Fishes of Australia. Sydney: New Holland.
Kuiter, R.H. & Debelius, H. 1994. South-East Asia: Tropical Fish Guide. Frankfurt: IKAN–Unterwasserarchiv.
Kuiter, R.H. & Debelius, H. 2001. Surgeonfishes, Rabbitfishes and their Relatives. Chorleywood, U.K.: TMC
Publishing.

Kuwamura, T. 1981. Mimicry of the cleaner wrasse Labroides dimidiatus by the blennies Aspidontus taeniatus
and Plagiotremus rhinorhynchos. Nanki Seibutu 23, 61–70.
3597_book.fm Page 479 Friday, May 20, 2005 6:26 PM
© 2005 by R.N Gibson, R.J.A. Atkinson and J.D.M. Gordon