Content uploaded by Will Sowersby
Author content
All content in this area was uploaded by Will Sowersby on May 02, 2023
Content may be subject to copyright.
Vol.40
No.2
April
2023
168 S. Sogawa et al.
ZOOLOGICAL SCIENCE 40: 168–174 (2023) © 2023 Zoological Society of Japan
* Corresponding author. E-mail: kohda.tanganyika@gmail.com
doi:10. 2108/zs2200 88
INTRODUCTION
A common feature of stable animal societies organized
by social relationships such as dominance hierarchies, ter-
ritoriality, and sexual pair-bond is the development of social
signals facilitating the recognition of individuals (Bshary et
al., 2002; Tibbetts and Dale, 2007; Leopold and Rhodes,
2010; Bshary, 2011; Bshary and Brown, 2014). Animals living
in these societies are under strong selective pressure to use
sensory information to recognize other individuals during
social interactions (Tibbettes and Dale, 2007; Tibbetts et al.,
2008). For example, visual cues are commonly exploited for
individual recognition in mammals and birds (e.g., Rosenfeld
and van Hoesen, 1979; Whiteeld, 1986; Brown and Dooling,
1992; Kendrick et al., 1995; Peirce et al., 2000; Gothard et
al., 2009; Kano and Tomonaga, 2009). Similar to other taxa,
it appears that many species of social sh are capable of
visually recognizing other individuals (Bshary et al., 2002;
Siebeck et al., 2010; Bshary, 2011; Bshary and Brown, 2014;
Male Guppies Recognize Familiar Conspecic Males by
Their Face
Shumpei Sogawa1,2 , Rio Fukushima1, Will Sowersby1,2, Satoshi Awata1,2 ,
Kento Kawasaka1,2, and Masanori Kohda1,2*
1Laboratory of Animal Sociology, Department of Biology and Geosciences, Graduate School of Science,
Osaka City University, Osaka, Japan
2Laboratory of Animal Sociology, Department of Biology, Graduate School of Science,
Osaka Metropolitan University, Osaka, Japan
Individual recognition is a necessary cognitive ability for the maintenance of stable social relation-
ships. Recent studies have shown that like primates, some sh species can distinguish familiar
sh from unfamiliar strangers via face -recognition. However, the taxa of the studied sh species
are restricted (within Perciformes) and the visual signal used for the recognition of sh remains
unclear. Here, we investigated the visual signal for individual-recognition in males of a sexually
dichromatic guppy (Poecilia reticulata, Cyprinodontiformes). Using guppy males, we examined the
hypothesis that sh distinguish between familiar individuals and unknown strangers by their faces
rather than by body coloration. We randomly presented focal sh with four types of composite
photo-models: familiar (familiar-face and familiar-body = F/F), stranger (stranger-face and
stranger-body = S/S), familiar face combined with stranger body (F/S) and stranger face combined
with familiar body (S/F). Focal males infrequently attacked familiar-face models but frequently
attacked stranger-face models, regardless of body types. These behavioral reactions indicate that
guppy males discriminate between familiar and stranger males by their face, not body coloration
with wide variation. Importantly, male faces contain clear individual-variation in white/metallic col-
ored patches on the operculum visible for humans. Considering the photo-model, our results sug-
gest that these patches might be an important visual stimulus for face-recognition in guppy males,
like some cichlids. Comparative examination among males of different guppy variants, including
wild type phenotype, suggests that the face color-patch is stable regardless of variation in body
color, with a different genetic mechanism potentially underlying face and body colors.
Key words: body coloration, face recognition, facial coloration, guppy, individual recognition
Kohda et al., 2015), in addition to using other sensory cues
such as acoustic and olfactory cues (e.g., Myrberg and
Riggio, 1985; Hawkins, 1986; Bshary, 2011; Bshary and
Brown, 2014).
The face appears to be a particularly important feature
for individual recognition in many social animals, including
primates, non-primate mammals, and some birds (Kendrick,
1994; Parr et al., 2000; Peirce et al., 2000; Racca et al.,
2010; Coulon et al., 2011). In shes, individual face-
recognition was rst suggested in territorial damselsh,
which seem to use facial color variations on the operculum
to recognize individuals (Siebeck et al., 2010). More recently,
face-recognition by exploiting small individual differences in
the shape and size of color patches near the eyes on the
operculum has been observed in the territorial cichlid
Neolamprologus pulcher (Kohda et al., 2015). Other cichlid
species that live in stable social groups also appear to have
individual variations in the face (Julidochromis transcriptus:
Awata et al., 2005; Hotta et al., 2017) and a discus sh with
a strong pair bond is capable of visually recognizing indi-
vidual faces (Symphysodon aequifasciatus, Satoh et al.,
2016). Individual face recognition is also reported in a
169
Face recognition in sh
harem-polygynous wrasse (Kohda et al., 2023). As such,
these examples of face recognition in sh species are docu-
mented from a variety of stable social systems. However,
these are restricted to shes of the order Perciformes, and
studies on social sh other than this order are required to
examine the face-recognition hypothesis that the face will be
pertinent for individual recognition in various sh taxa.
Another current problem in the studies of sh face rec-
ognition is that many studied sh species exhibit color varia-
tion on the operculum but exibit no or less color variations on
body or ns (Kohda et al., 2015, 2023; Satoh et al., 2016;
Hotta et al., 2019). It remains unknown whether sh rely only
on individual differences in the face for individual recognition
or also exploit variation in color and patterns on the body
and/or ns. Thus, studies of sh face recognition should be
conducted using sh that exibit a wide individual variation in
the body and/or ns as well as or rather than in the face.
To test the predictions from the face-recognition hypoth-
esis, we chose the ornamental guppies (Poecilia reticulata)
variant “Neon tuxedo” as our study species. Guppy is a sh
belonging to the order Cyprinodontiformes and is a popular
ornamental sh species. This sh exhibits large variations in
color across variants due to selective breeding (Fig. 1).
Guppies are a model organism in behavioral ecology and
cognitive studies (Magurran, 2005) and are thought to be
able to distinguish between individuals (Croft et al., 2003,
2004). We hypothesize that individual differences in face
color patterns and the ability to exploit these differences for
individual recognition will be highly conserved in guppies
despite selective breeding for various color variants on the
body and ns. In our preliminary observations, we found that
in males of the Neon tuxedo guppy, there are possible
individual differences in color patches on the operculum
(F i g. 1).
Therefore, in Experiment 1, we tested the dear enemy
effect to establish whether male guppy aggression is
reduced when they recognize familiar individuals. The dear
enemy effect is an ethological phenomenon in which once
territorial neighbors establish the border and become famil-
iar, these neighbors become tolerant of each other (e.g.,
Temeles, 1994). When territory owners have this relation-
ship, they still exihibit strong aggression to stranger individu-
als, and we can judge from their responses whether they
distinguish familiar and unfamiliar sh. In Experiment 2, we
presented males with sh models, including composite
models of familiar individuals and unknown strangers, to
establish whether guppies recognize individuals due to indi-
vidual differences in the face or body coloration. Our study
will make clear whether sh outside of Perciformes recog-
nize the faces of other individuals or rely on other features of
body and ns with great individual variation.
MATERIALS AND METHODS
Study sh and maintenance
We obtained 50 adult male Neon tuxedo guppies
from a commercial breeder (Kamihata Fish Ind. Ltd.,
Japan). The breeder kept sh in a large tank with many
individuals (Kamihata Fish Ind. Ltd., personal commu-
nication). The males were approximately 30 mm in
standard length (SL) and were housed under labora-
tory conditions in four stock tanks (10–14 males per
tank; 30 × 60 × 30 cm3) at 23–27°C, under 12:12 h
light-dark cycle at Osaka City University, Japan. Gup-
pies were housed under these conditions for > 30
days and were visually isolated from other tanks in the
laborator y. In the stock tanks we often observed the
sh interacting and forming shoals. We considered
sh from the same stock tank as familiar individuals
and sh from other tanks as unknown strangers in the
following experiments.
Experiment 1: Dear enemy relationship with famil-
iar neighbor
The dear enemy relationship describes a phe-
nomena where aggression between individuals is
reduced once territorial boundaries are set (Sogawa et
al., 2016). To examine whether these guppies form
dear enemy relationships, we placed one focal sh
(mean SL ± s.d. = 28.09 ± 0.95 mm; n = 14) and a
similar sized ( < 2 mm in size difference) stimulus
(opponent) sh, from a different stock tank, in separate
adjacent experimental tanks (15 × 15 × 15 cm 3; water
depth 12 cm) separated by a 1 cm gap (Fig. 2A). In
each experimental tank we placed a small stone (3 ×
4 × 4 cm3) in the center to facilitate territorial behav-
iors. The two sh were kept visually isolated from each
other by a white PVC partition between the two exper-
imental tanks. The rear side of the experimental tanks
was covered with black paper so that we could record
clear movies of sh behaviors. In all 14 trials the tank
in which the focal guppy was placed was randomly
Male
Female
Fig . 1. A whole body and ve male and ve female guppy (Poecilia reticulata)
faces of the “Neon tuxedo” variant. Photographs were taken under identical light
conditions.
170 S. Sogawa et al.
chosen.
We allowed the focal sh to habituate and form territories
in the experimental tanks for 3 days. We then removed the
partition and video recorded social interactions using a Sony
HDR-CX485 for 6 minutes every morning for 7 days. During
that time the white partitions were not reset until after the end
of the experimental period. We predicted that aggression
between the focal sh and the stimulus neighbor would
decrease with time (Kohda et al., 2015; Sogawa et al., 2016;
Saeki et al., 2018), demonstrating a dear enemy ef fect (Fisher,
1954). Attacks and aggressive displays with spreading ns
towards opponents were recorded as aggressive behaviors.
Experiment 2: Presentation of stimulus models
We created familiar and stranger photo-models using the
familiar neighbor males (F) from the dear enemy experiment
(n = 14 males) and stranger males (S) from different stock
tanks. We took photographs (Nikon D610) of both sides of
neighbor males while they were under anesthesia prior to
Experiment 1. Similarly, we took photographs of both sides of
unknown stranger males (< 2 mm in size difference; Fig. 3).
Photographs were taken under white uorescent lamps
(FL-202, LPL Co. Ltd.) that were attached to both sides of a
copy stand (CS-A4, LPL Co. Ltd.) and included a color
checker (Spyder CHECKR 24, Datacolor Inc.). The color bal-
ance of each photograph was assessed and corrected using
Adobe Photoshop 2021. A UV-cut protection lter was
attached to the camera lens, so UV-light from sh would not
be reected on an image of a photo-model.
We then created four different types of digital models
(Kohda et al., 2015; Satoh et al., 2016; Hotta et al., 2017):
familiar neighbor face and body (F/F), stranger face and body
(S/S), familiar neighbor face and stranger body (F/S), and
stranger face and familiar neighbor body (S/F) (Fig. 3). We
accounted for any differentiation in coloration between the
face and body areas using GIMP 2.10.28 (The Gimp Team,
https://www.gimp.org/). The four different model types were
then printed (EPSON EP-30VA, 200 dpi) in color on glossy
photographic paper, cut to match the contours of the model
sh, and laminated.
We predicted that if male guppies distinguish between
familiar neighbors and unknown strangers, the focal sh
would attack F/F models less frequently than the S/S model.
Furtherm ore, we predicted that if g uppies use face- recognition
to distinguish between neighbors and strangers,
the focal sh would attack S/S and composite S/F
models (i.e., stranger face) at a greater frequency
than F/F and F/S models (i.e., familiar face). In
contrast, if males use variations in body color to
identify individuals, focal sh would attack S/F
and F/S at an intermediate rate between F/F and
S/S models.
To test these predictions, we presented the
four different types of models to the focal males.
We rst reset the white partition bet ween the two
tanks after the lights went out on the evening of
day 7 of Experiment 1 and removed the familiar
individual. We randomly attached one of the four
models to the center of a transparent acrylic plate
(15 × 15 cm2), which was itself placed 3 cm away
from the closest side of the tank facing toward the
focal sh tank (Fig. 2B). On the next morning, the
white partition was removed again, and the behav-
ior of the focal sh was video-recorded for 6 min
starting from the rst reaction of the focal sh
toward the model. Afterwards, the partition was
placed back between the two tanks; the transpar-
Focal fish
Focal fish Opponent
1 cm
(A)
Focal fish
Photo model
3 cm
(B)
Fig. 2. Schematic illustrations of experiment tanks. (A) Experiment 1. A
focal sh and an opponent are put in two tanks with a 1 cm gap in which a
white sheet is put. A small stone on the bottom. (B) Experiment 2. A front
view of the two experiment tanks during the presentation of photograph
models. A photograph model attached on a transparent board is presented
with a 3 cm distance from the glass wall facing the focal sh tank (See
MATERIALS AND METHODS for procedure of these experiments).
Fig. 3. Four types of male guppy composite models used in Experiment 2. F: familiar
and S: stranger. F/F = familiar face and familiar body; S/S = stranger face and stranger
body; F/S = familiar face and stranger body, and S/F = stranger face and familiar body.
171
Face recognition in sh
ent acrylic plate with the model was removed; and the original famil-
iar neighbor in Experiment 1 was returned to the tank. The par tition
was once again removed so the focal sh and the familiar neighbor
were visible to each other and the dear enemy relationship was
maintained. This procedure was repeated every 2 days until the
focal sh was nished being exposed to all of the four model types.
Statistical analyses
To avoid any human disturbance inuencing our results we did
not quantif y focal sh behavior in the rst minute of each 6 minute
video recording in both Experiment 1 and 2. We categorized guppy
aggressive behavior towards the stimulus neighbor sh with refer-
ence to Baerends et al. (1955). In Experiment 1, to test the dear
enemy effect (i.e., that aggression declined over time), changes in
the total time that the focal sh spent performing aggressive behav-
iors towards the opponent (sec per 5 min) from day 1 to day 7 was
assessed by a Friedman test. As a post hoc, multiple comparison
tests between the days, Exact Wilcoxon signed-rank tests with a
sequential Bonferroni adjustment was performed. In Experiment 2,
to test whether focal sh behaved differently towards the face and
body of familiar neighbors and unknown strangers, the total time
that the focal sh spent performing aggressive behaviors towards
the photo model was compared among four models (F/F, S/S, F/S,
S/F), using a Friedman test, followed by Exact Wilcoxon signed-
rank tests with a sequential Bonferroni adjustment. All statistical
analyses were performed in R version 4.2.1 (R Core Team, 2022).
All tests were two sided, and the signicance levels were 0.05.
Ethical note
No animals were sacriced during our experiments. Guppies
were fed a commercial ake food (Tetramin; Spectrum Brands
Japan Inc.) once a day and housed in appropriate conditions. Sick
or injured sh in the stock tanks were removed to a separate tank
and treated with medication where possible. All experiments were
conducted in compliance with the Regulations on Animal Experi-
ments of Osaka City University and the Japan Ethological Society.
RESULTS
In Experiment 1, all focal sh exhibited strong aggres-
sion toward their neighbors on day 1, but their aggression
decreased drastically from day 2 and remained constant
until day 7 (Fig. 4). Consequently, the time focal sh spent
acting aggressively toward their neighbors on day 1 (mean ±
s.e. = 225.79 ± 19.8 sec) was signicantly longer than that
on the other days (Day 2–7: all < 35 sec; Friedman test:
χ
2 =
37. 5 8 , n = 14 males, P < 0.0001), and no signicant differ-
ences were found in the time spent acting aggressively
among day 2 through day 7 (Fig. 4).
In Experiment 2, there was a signicant difference in the
time spent in aggression by the focal sh toward four types
of models, F/F (familiar neighbor face and body used for
Experiment 1), S/S (stranger face and body), FS, and SF
(Friedman test:
χ
2 = 16.20, n = 14 males, P = 0.001). The
focal sh acted more aggressively toward the stimulus mod-
els when the face was stranger (i.e., S/S and S/F) than when
the face was familiar neighbor (F/F and F/S; Fig. 5). There
were no signicant differences in time spent in aggression
between the S/S and S/F models or between the F/F and
F/S models. Note that the focal sh behaved aggressively at
a similar level toward the F/F model in Experiment 2 and the
familiar neighbor individual on day 7 in Experiment 1 (Exact
Wilcoxon signed-rank test, V = 51.00, P = 0.95; Figs. 4 and
5). In contrast, the focal sh spent signicantly more time
acting aggressively toward the S/S model than it did toward
the familiar neighbor individual on day 7 (Exact Wilcoxon
signed-rank test, V = 0.00, n = 14 males, P = 0.0001).
DISCUSSION
Our results suggest that male guppies (Neon tuxedo)
can recognize the faces of conspecics. By using a combi-
nation of live sh, stimulus models and composite models
we demonstrate that the face is used to visually recognize
other individuals rather than the body. Below we outline why
this ability is important for social shes, and suggest the sig-
nal likely used for individual recognition in many sh spe-
cies. Our study demonstrates that social shes, including
guppy, are capable of complex social cognitive behaviors,
and are likely under similar selective pressures to visually
recognize individual conspecics.
Fig. 4. Changes in the time focal male guppies spent acting
aggressively toward neighbor males over a period of 7 days.
Means ± s.e. (sec/5 min) are shown (n = 14). Different alphabets
denote statistically signicant differences (P < 0.05) by Exact
Wilcoxon signed-rank tests with sequential Bonferroni correction
method.
0
50
100
150
200
250
1 2 3 4 5 6 7
Days
a
bb
bb
bb
Time in aggressive behavior
of focal fish (sec / 5 min)
Fig. 5. The time of aggressive reactions of male guppies (sec/5
min) against four types of photo models (F/F: familiar face and
familiar body; F/S: familiar face and strange body; S/S: stranger
face and stranger body; S/F: stranger face and familiar body) (n =
14). The box plot shows the median (thick line within the box), 25th
and 75th percentiles (box), ranges (whiskers), and outliers (white
circles). Different alphabets denote statistically signicant differ-
ences (P < 0.05) by Exact Wilcoxon signed-rank tests with sequen-
tial Bonferroni correction method.
0
50
100
150
200
250
300
a
F/F F/S S/S S/F
a
b
b
Time in aggressive behavior
against models (sec / 5 min)
Four types of models
172 S. Sogawa et al.
When exposed to a stimulus male for 7 days, focal-male
aggressive behavior decreased signicantly after the rst
day and remained low for the remainder of the experimental
period. This pattern of reduced aggression toward a known
neighbor rival is referred to as the dear enemy effect, and
this will be the rst documented nding of it in a species
other than cichlid species (Leiser and Itzkowits, 1999;
Frostman and Sherman, 2004; Kohda et al., 2015; Sogawa
et al., 2016; Saeki et al., 2018; Hotta et al., 2019). The dear
enemy effect is predicted to increase individual tness by
reducing the costs of defending a territory (Fisher, 1954). A
key element of the dear enemy effect is therefore the ability
to recognize conspecics individually. After the 7 day period,
we exposed focal sh to the photo-models of the same
familiar neighbour and an unknown stranger. The results
demonstrated that decline in focal male aggression was
directly related to their ability to recognize the particular indi-
vidual and not because they had become
habituated to having an adjacent occu-
pied tank. Furthermore, although male
Neon tuxedo guppies have individual
variations in bright body and n colors,
our familiar/stranger body face compos-
ite models clearly demonstrated that
focal sh recognized individuals by their
face, rather than their body. Our results
are the rst documented face-recogni-
tion in sh outside the order Perciformes
(Seibeck et al., 2010; Kohda et al., 2015,
2023; Satoh et al., 2016; Hotta et al.,
2017, 2019), and support the hypothesis
that individual face-recognition in sh
may be widespread in a variety of sh
(probably including Beloniformes: Wang
and Takeuchi, 2017).
The face appears to be particularly
important for the visual recognition of
conspecics in guppy males. Kohda et
al. (2015) found in the cichlid N. pulcher
that this sh recognizes individuals via
individual-specic color patches near
the eye on the operculum via an experi-
ment with replacement of only their color
patches between photographs of familiar
and unfamiliar conspecic males. That
study clearly showed that the color patch
is the most important visual signal for
individual recognition among various
face features, including eyes, head
shape, face coloration other than the
color patches, and so on. Interestingly,
other cichlids, damselshes, a wrasse,
and a medaka that have demonstrated
the ability to visually distinguish between
conspecic individuals via face recogni-
tion also have color patches with indi-
vidual variation similarly located near the
eye on the operculum (Siebeck et al.,
2010; Satoh et al., 2016; Hotta et al.,
2017; Wang and Takeuchi, 2017; Kohda
et al., 2023). Male guppies, including
the Neon tuxedo used in this study, have comparable whit-
ish/metallic color patches with individual variation on their
operculum, indicating that they may use a similar trait for
identifying conspecics. In contrast to males, female gup-
pies are typically less colorful on the body and ns, yet they
are suggested still likely to have the ability to distinguish
between familiar individuals and unknown strangers in their
social networks (Griffiths and Magrran, 1997; Swaney et al.,
2001; Croft et al., 2004). Females of the Neon tuxedo guppy
also have observable white/metallic color patches on the
operculum (Fig. 1). We therefore suggest that variation in
color patterns or patches on the operculum are likely
exploited as a visual signal for individual recognition in gup-
pies and across social shes.
The damselshes, cichlids, medaka, and guppy are
known to have UV vision, which should be taken into account
in considering visual cues for face recognition (Benett et al.,
Female
Blue Glass Cobra Yellow Tuxedo
Wild type
Wild type
Blue Glass guppy
Male
Female
Fig. 6. A wild type male (right top) and 11 male variations and a wild type female (right
bottom) and two female variations of guppy, Poecilia reticulata. English names of the
variations are shown. Note that male body colorations are largely different between varia-
tions, but the face-color on operculum is observed in all variations and wild types. The
operculum colors are also observed in females. Photographs are through the courtesy of
Kamihata Fish Ind. Ltd. Japan.
173
Face recognition in sh
1996; Smith et al., 2002; White et al., 2003; Siebeck et al.,
2010, Wang and Takeuchi, 2017). The photo-models in the
current study and the previous studies were made of photo-
graphs taken by cameras with a UV protection lter lens
(Kohda et al., 2015; Kawasaka et al., 2019). Thus, the focal
sh see photo-models of no UV-light, likely as we see, yet
they clearly distinguish the models of familiar sh from unfa-
miliar one quickly and accurately (Kohda et al., 2015). As
guppy males also distinguish the face of a UV-light-free
model accurately, we can say that the prominent feature in
the face of a guppy will be the color patch on the operculum
like the one in N. pulcher as we see it, and the role of UV-
light, if any, will be small (Fig. 1). The color patches near the
eyes on the operculum visible to humans will be the main
signals in the visual recognition in these shes.
Across taxa, the eyes seem to play an important role in
individual face-recognition (e.g., sh: Karplus and Algom,
1981; Karplus et al., 1982, 2006; Karina et al., 2012; pri-
mates: Nahm et al., 1997; humans: Kano and Tomonaga,
2010; Gothard et al., 2009). For instance, animals often look
rst at an opponent’s eyes before assessing the rest of the
face (e.g., humans and chimpanzees: Kano and Tomonaga,
2009, 2010; rhesus monkey: Gothard et al., 2009; and pos-
sibly also sh: Hotta et al., 2019). It is therefore reasonable
to assume that selection may favor individual identifying
traits that are located close to the eyes to facilitate rapid sig-
nal transmission. Indeed, experiments have shown that N.
pulcher can accurately distinguish between familiar and
unfamiliar faces within 0.5 seconds (Kohda et al., 2015). The
locations of color patches in Neon tuxedo guppy are both on
the operculum close to the eye and are supercially similar
to the locations of color patches on the operculum of social
cichlids and damselshes and medaka, which are also
capable of visually recognizing individual faces (Siebeck et
al., 2010; Kohda et al., 2015; Hotta et al., 2017; Wang and
Takeuchi, 2017). We speculate that if color patches were
located elsewhere on the body they would not be as effec-
tive at signaling individual identity and that having individual
identifying traits close to the eye is important across social
shes.
Previous studies have found that similar to birds and
mammals, shes are affected by the face inversion effect,
where faces that are presented to focal individuals upside-
down are difficult to recognize (sh: Wang and Takeuchi,
2017; Kawasaka et al., 2019; mammals: Kendrick, 1994;
Kendrick et al., 1995; birds: Whiteeld, 1986; Brown and
Dooling, 1992). These experiments have provided additional
evidence that animals, including social sh, typically regard
the face as a whole rather than as individual components.
Based on the ndings in the guppy in the present study, we
hypothesize that in sh, color patches and patterns near the
eye on the operculum are also integrated into the whole face
by observers (sensu Kawasaka et al., 2019) and directly
increase inter-variation in the face allowing for quick and
accurate recognition. Importantly, sh will have a mental
image of the familiar’s face, including the inter-variations of
the color patch (Kohda et al., 2023). Another study of ours on
the guppy Neon tuxedo is likely to provide positive results
regarding their inversion effect, which will be published else-
where in the near future.
In this study we used a guppy variant, the Neon tuxedo,
which has undergone selective breeding by hobbyists and in
the aquarium trade. However, by looking at images of wild
type and 11 other varieties of guppy, it is clearly apparent
that the metallic colored patch on the operculum is con-
served across guppy variants (see Fig. 6). Moreover, we
speculate that within guppy variants, the patch on the oper-
culum will vary between individuals, as in Neon tuxedo
(Kohda and Fukushima, personal observations). Male guppy
body coloration has received widespread attention in evolu-
tionary research, but considerably less attention has been
given to face color (e.g., Houde, 2019). Interestingly, while
the characteristic orange and black wild-type male body
phenotype has resulted in several brightly colored male
guppy variants via selective breeding, the metallic color
patch on the face has remained conserved. Female body
color on the other hand has remained comparatively dull
across guppy variants but the color patch on the operculum
is still visible (see Fig. 6). These anecdotal observations
suggest that the color patch on the operculum is a relatively
stable trait and is likely used for individual recognition rather
than as a sexual signal. Furthermore, given the retention of
the face color patch across variants despite dramatic
changes in body coloration, we propose the hypothesis that
body and face coloration will have a different underlying
genetic inheritance.
ACKNOWLEDGMENTS
We thank the members of the Laborator y of Animal Sociology,
Osaka City University for their fruitful discussions. The present
study was nancially supported by KAKENHI (Nos. 17K18712,
19H03306, and 20K20630 to MK) and the Osaka City University
Strategic R esearch Grant 2018 and 2019 for Top Priority Researc hes
to MK and SA.
COMPETING INTERESTS
The authors have declared that no competing interests exist.
AUTHOR CONTRIBUTIONS
SS, RF, and MK substantially contributed to the study concep-
tualization. KK contributed to the methodology and the resources.
SS and RF contributed to the investigation and the data curation.
SA contributed to formal analysis and data visualization. SS and
MK wrote the original manuscript with support from RF and WS. All
authors contributed to review and editing.
REFERENCES
Awata S, Munehara H, Kohda M (2005) Social system and repro-
duction of helpers in a cooperatively breeding cichlid sh
(Julidochromis ornatus) in Lake Tanganyika: eld observations
and parentage analyses. Behav Ecol Sociobiol 58: 506–516
Baerends GP, Brouwer R, Waterbolk HT (1955) Ethological studies
on Lebistes reticulatus (Peters) 1. An analysis of the male
courtship pattern. Behaviour 8: 249–334
Bennett ATD, Cuthill IC, Partridge JC, Maier EJ (1996) Ultraviolet
vision and mate choice in zebra nches. Nature 380: 433 –435
Brown SD, Dooling RJ (1992) Perception of conspecic faces by
budgerigars (Melopsittacus undulatus): I. Natural faces. J
Comp Psychol 106: 203–216
Bshary R (2011) Machiavellian intelligence in shes. In “Fish Cogni-
tion and Behavior” Ed by C Brown, K Laland, J Krause, Wiley-
Blackwell, Hoboken, NJ, pp 277–297
Bshary R, Brown C (2014) Fish cognition. Current Biology 24:
R947–R950
Bshary R, Wickler W, Fricke H (2002) Fish cognition: a primate’s
174 S. Sogawa et al.
eye view. Animal Cognition 5: 1–13
Coulon M, Baudoin C, Heyman Y, Deputte BL (2011) Cattle dis-
criminate between familiar and unfamiliar conspecics by
using only head visual cues. Anim Cog 14: 279–290
Croft DP, Albanese B, Arrowsmith BJ, Botham M, Webster M,
Krause J (2003) Sex-biased movement in the guppy (Poecilia
reticulata). Oecologia 137: 62–68
Croft DP, Krause J, James R (2004) Social networks in the guppy
(Poecilia reticulata). Proc R Soc B 271: S516–S519
Fisher J (1954) Evolution and bird sociality. In “Evolution as a Pro-
cess” Ed by J Huxley, AC Hardy, EB Ford, Allen and Unwin,
London, pp 71–83
Frostman P, Sherman PT (2004) Behvioral response to familiar and
unfamiliar neighbors in a territorial cichlid sh. Ichthyol Res 51:
283–285
Gothard KM, Brooks KN, Peterson MA (2009) Multiple perceptual
strategies used by macaque monkeys for face recognition.
Anim Cog 12: 155–167
Griffiths NW, Magurran AE (1997) Schooling preferences for famil-
iar sh vary with group size in a wild guppy population. Proc R
Soc B 264: 547–551
Hawkins AD (1986) Underwater sound and sh behaviour. In “The
Behaviour of Teleost Fishes” Ed by TJ Pitcher, Springer,
Boston, MA, pp 114–151
Hotta T, Satoh S, Kosaka N, Kohda M (2017) Face recognition in the
Tanganyikan cichlid Julidochromis transcriptus. Anim Behav
127: 1–5
Hotta T, Kawasaka K, Shatoh S, Kohda M (2019) Fish focus primar-
ily on the faces of other sh. Sci Rep 9: 8377
Houde AE (2019) Sex, Color, and Mate Choice in Guppies. Prince-
ton University Press, Princeton
Kano F, Tomonaga M (2009) How chimpanzees look at pictures: a
comparative eye-tracking study. Proc R Soc B 276: 1949–1955
Kano F, Tomonaga M (2010) Face scanning in chimpanzees and
humans: continuity and discontinuity. Anim Behav 79: 227–235
Karina KA, Giljov AN, Malashichev YB (2012) Eye as a key element
of conspecic image eliciting lateralized response in sh. Anim
Cogn 16: 287–300
Karplus I, Algom D (1981) Visual cues for predator face recognition
by reef shes. Z Tierpsychol 55: 343 –364
Karplus I, Goren M, Algom D (1982) A preliminary experimental
analysis of predator face recognition by Chromis caeruleus
(Pisces, Pomacentridae). Z Tierpsychol 58: 53 –65
Karplus I, Katzenstein R, Goren M (2006) Predator recognition and
social facilitation of predator avoidance in coral reef sh Das-
cyllus marginatus juveniles. Mar Ecol Prog Ser 319: 215 –223
Kawasaka K, Hotta T, Kohda M (2019) Does a cichlid sh process
face holistically? Evidence of the face inversion effect. Anim
Cogn 22: 153 –162
Kendrick KM (1994) Neurobiological correlates of visual and olfac-
tory recognition in sheep. Behav Process 33: 89–112
Kendrick KM, Atkins K, Hinton MR, Broad KD, Fabre-Nys C,
Keverne B (1995) Facial and vocal discrimination in sheep.
Anim B eh av 49: 1665 –1676
Kohda M, Jordan LA, Hotta T, Kosaka N, Karino K, Tanaka H, et al.
(2015) Facial recognition in a group-living cichlid sh. PLOS
ONE 10: e0142552
Kohda M, Bshary R, Kubo N, Awata S, Sowersby W, Kawasaka K,
et al. (2023) Cleaner sh recognize self in a mirror via self-face
recognition like humans. Proc Natl Acad Sci USA 120:
e2208420120
Leiser J, Itskowitz M (1999) The benets of dear enemy recognition
in three-contender convict cichlid (Cichlasoma nigrofasciatus)
contests. Behaviour 136: 983 –1003
Leopold DA, Rhodes G (2010) A comparative view of face percep-
tion. J Comp Psychol 124: 233–251
Magurran AE (2005) Evolutionary Ecology: The Trinidadian Guppy.
Oxford University Press, Oxford
Myrberg A, Riggio RJ (1985) Acoustically mediated individual rec-
ognition by a coral reef sh (Pomacentrus partitus). Anim
Be h av 3 3: 411– 416
Nahm FKD, Perret A, Amaral DG, Albright TD (1997) How do mon-
keys look at faces? J Cog Neurosci 9: 611–623
Parr LA, Winslow JT, Hopkins WD, de Waal FBM (2000) Recogniz-
ing facial cues: Individual discrimination by chimpanzees (Pan
troglodytes) and rhesus monkeys (Macaca mulatto). J Comp
Psychol 114: 47–60
Peirce JW, Leigh AE, Kendrick KM (2000) Congurational coding,
familiarity and the right hemisphere advantage for face recog-
nition in sheep. Neuropsychologia 38: 475–483
Racca A, Amadei E, Ligout S, Guo K, Meints K, Mills D (2010) Dis-
crimination of human and dog faces and inversion responses
in domestic dogs (Canis familiaris). Anim Cog 13: 525–533
Rosenfeld SA, van Hoesen GW (1979) Face recognition in the rhe-
sus monkey. Neuropsychologia 17: 503–509
Saeki T, Sogawa S, Hotta T, Kohda M (2018) Territorial sh distin-
guish familiar neighbours individually. Behaviour 155: 279–293
Satoh S, Tanaka H, Kohda M (2016) Facial recognition in a discus
sh (Cichlidae): Experimental approach using digital models.
PLOS ONE 11: e0154543
Siebeck UE, Parker AN, Sprenger D, Mathger LM, Wallis G (2010)
A species of reef sh that uses ultra-violet patterns for cover t
face recognition. Current Biol 20: 407–410
Smith EJ, Partridge JC, Parsons KN, White EM, Cuthill IC, Bennett
ATD, Church SC (2002) Ultraviolet vision and mate chpice in
the guppy (Poecilia reticulata). Behav Ecol 13: 11–19
Sogawa S, Ota K, Kohda M (2016) A dear enemy relationship in a
territorial cichlid: evidence for the threat-level hypothesis.
Behaviour 153: 387–400
Swaney W, Kendal J, Capon H, Brown C, Laland KN (2001) Famil-
iarity facilitates social learning of foraging behaviour in the
guppy. Anim Behav 62: 591–598
Temeles EJ (1994) The role of neighbours in territorial systems:
when are they ‘dear enemy’? Anim Behav 47: 339– 350
Tibbetts EA, Dale J (2007) Individual recognition: it is good to be
different. Trends Ecol Evol 22: 529–537
Tibbetts EA, Sheehan MJ, Dale J (2008) A testable denition of
individual recognition. Trends Ecol Evol 23: 356
Wang MY, Takeuchi H (2017) Individual recognition and the ‘face
inversion effect’in medaka sh (Oryzias latipes). eLife 6:
e24728
White EM, Julian C, Partridge JCP, Church SC (2003) Ultraviolet
dermal reexion and mate choice in the guppy, Poecilia reticu-
lata. Anim Behav 65: 693–700
Whiteeld DP (1986) Plumage variability and territoriality in breed-
ing turnstone, Arenaria interpres: status signaling or individual
recognition? Anim Behav 34: 1471–1482
(Received October 11, 2022 / Accepted February 20, 2023 /
Published online April 1, 2023)
