Full text of DOI:10.1163/156853900502286

CHEMICAL TRANSFER OF WARNING INFORMATION IN
NON-INJURED FISH
by
1,2)
LUCIANA C. JORDÃO and GILSON L. VOLPATO
(Laboratory of Animal Physiology and Behaviour, Dept. de Fisiologia, IB, CAUNESP,
UNESP, 18618-000, CP 510, Botucatu — SP, Brazil)
(Acc. 7-II-2000)
Summary
In this study we describe pacus,
Piaractus mesopotamicus
, detecting the presence of a
predator by conspeciŽc alerting pheromone. Moreover, we investigate whether this chemical
information indicates the presence of a speciŽ c predator, or whether it indicates general
Hoplias
disturbance. We exposed groups of pacus to the view of a predator Ž sh (trahira,
malabaricus
), a non-predator Ž sh (piracanjuba,
Brycon orbignyanus
) or an aquarium without
any Ž sh (control), and then we transferred their water to isolated conspeciŽcs. We set up
six trials of each condition in which we analysed the dispersion and the distance from the
visual stimulus in water-donor Ž sh and the distance from the chemical stimulus in water-
receiver pacus. This study showed that pacus visually identiŽ ed the presence of another
Ž sh and recognised it as predator or non-predator. This is interpreted as an innate response.
Such heterospeciŽc detection affects the behaviour of pacus, which release chemicals that
induce conspeciŽ cs to adopt a similar behaviouralresponse. At least two chemicals might be
involved, one of them possibly an alerting pheromone.
Keywords
: chemical communication, disturbance pheromone, alerting pheromone, predator,
Piaractus mesopotamicus
Ž sh behaviour, pacu,
.
1) Correspondiningauthor; e-mail address: volpato@ibb.unesp.br
2)
°
¢
This study is part of L.C. Jordão s MSc that was founded by FAPESP (n 96/09884-1). The
authors thank very much A.C.B. Tardivo for technical assistance; Dr. N.R. Liley for revising
the last manuscript and providingvery helpfulsuggestions;Dr. L.M. Dill and Dr. D.P. Chivers
for providing helpful reprints; and Fish-Braz®R for donating the Ž sh.
c
®
Koninklijke Brill NV, Leiden, 2000
Behaviour 137, 681-690

682
JORDÃO & VOLPATO
Introduction
In predator-prey interactions, early detection of a predator may be crucial
(Lima & Dill, 1990). In shoaling Ž sh species, transfer of such information
among conspeciŽ cs is achieved by different sensory modalities according
to their habits and environment. Some species, e.g. minnows Phoxinus
phoxinus (Magurran & Highan, 1988) and fathead minnows Pimephales
promelas (Mathis et al., 1996), give a fright reaction that warns conspeciŽ cs
after detecting a predator. The fright reaction is clearly composed by visual
displays, and these cues are widely used in Ž sh communication (Guthrie &
Muntz, 1993). However, considering the poor quality of aquatic images and
the low transparency in turbid rivers, chemical cues are also expected to be
used for transfer of information.
In Ž sh, chemically mediated information among conspeciŽ cs is well re-
ported. A range of studies has shown that many Ž sh species release chemi-
cal alarm cues when injured (e.g. predator attack). Such alarm substance is
stored in club cells located in the skin of most of Ostariophysan Ž shes, and it
is released when these cells are burst by mechanical damage (for review see
Pfeiffer, 1977; Smith, 1992; Chivers & Smith, 1998).
Although alarm substance has been largely investigated in Ž sh, chemical
communication in dangerous situations between non-injured conspeciŽ cs
has scarcely been studied. The chemical released by stressed (but not injured)
animals that is recognised by their conspeciŽ cs has been referred to as a
¢
‘disturbance pheromone , and is found in many aquatic organisms (Hazlett,
1989; Wisenden et al., 1995; Mathis & Lancaster, 1998; Kiesecker et al.,
1999).
As far as we know, Wisenden et al. (1995) are the only investigators to
experimentally describe such a disturbance pheromone in Ž sh. They found
that individual Iowa darters, visually stimulated by an artiŽ cial model of
a predator, release chemicals that induce alert posture in conspeciŽ cs. But
¢
the term ‘disturbance pheromone is not speciŽ c to this type of disturbance.
¢
Thus, we adopt in this study the term ‘alerting pheromone , as suggested by
N.R. Liley (pers. comm.) to indicate the vigilance response to the cue.
The present study is the second to show such transfer of alerting informa-
tion by chemical cues in Ž sh. This was achieved in the Brazilian Characidae
pacu, Piaractus mesopotamicus, using a natural predator as visual stimuli
to induce release of chemical signals. Moreover, we investigated whether in

CHEMICAL WARNING INFORMATION IN FISH
683
such chemical cues pacus discriminate the heterospeciŽ c as a predator or
not. Pacu is a schooling Ž sh living in turbid rivers where vision is limited,
and thus elaborate chemical communication may be expected.
Methods
Our investigative strategy consisted of exposing groups of pacus (4 Ž sh per aquarium)
Hoplias malabaricus
) and transferring their water to
to the view of a predator (trahira,
isolatedconspeciŽcs. In a second conditionwe used a piracanjuba,
Brycon orbignyanus
(non-
predator), instead of the trahira to test whether the effect of exposure to any Ž sh produced
alerting pheromone. Either trahiraor piracanjubawere always turned to face the donor pacus.
In a third condition, we used only the view of an aquarium with water but without any other
Ž sh. We set up six trials of each condition in which we analysed the behaviour of both donor
and receiver pacus.
Fishes and holding conditions
Piaractus mesopotamicus
We used 180-day old juvenile pacus,
(Holmberg, 1887), grown in
a hatchery and which had never experienced another Ž sh species. These Ž sh were transferred
to the laboratory and housed in a 500-l tank (2 Ž sh per l over this period). The trahira was
captured one week beforethe experimentand kept alone in a 500-l tank. The piracanjubawas
held in a 2000 l tank with three other tropical species for about two months(1 Ž sh per 100 l).
The lengths of the trahira and piracanjubawere about 15 cm each, and they were about three
times larger than the pacus.
All the tanks were in a closed room with 12L:12D cycle and water temperature about
°
25 C. Commercial Ž sh food was provided once per day in excess, but feeding was detected
only in pacus and piracanjuba. The Ž sh were food-deprived at least 24 h before the
experiment. Food leftoversand solid Ž sh excreta were removed at least once per week.
Experimental apparatus
´
´
donor receiver
or
and the conditioned water was
We used glass aquaria(28.0 10.0 18.0 cm; 4.8 l) as either
containers.
receiver donor
Each
aquarium was connected to only one
always transferredtoward the receiver.Water was transferredthrough a pipe by gravity(80 ml
per min) and reached the receiver aquarium at an outlet in a corner situated 1.75 cm from the
bottom and 1.75 cm from the wall. The donor containerwas not reŽ lled duringthe experiment
to avoid the dilution of the released chemical substances. In pretrial studies, we tested the
system by adding methylene blue in the donor containers: after 20 min of water transference,
the transmitancein the receiver aquaria was equivalent to 88% of the donor one.
An aquarium with heterospeciŽc Ž sh (trahira or piracanjuba)or water only was set next to
each donor aquarium and an opaque partition avoided visual contact between them before the
tests. The receiver aquaria had no visual contact with either donor or heterospeciŽc aquaria.
We used a dark curtain surrounding the whole aquarium system to prevent disturbanceby
external stimuli. Manipulationof the water  ow was made outside this area to avoid handling
interference.

684
JORDÃO & VOLPATO
Behaviour recording and analysis
Pacus were introduced into experimental aquaria approximately16 h before the beginning of
the experiment. Four pacus were placed in each donor and only one in each receiver aquaria.
Although we used a shoal species in an isolated condition(a stressful element), this situation
¢
has been recognisedto induce the Ž sh to give a ‘pure responseto chemical stimuli(Lawrence
& Smith, 1989).
Two video cameras were set up for simultaneousrecordingof Ž sh behaviour in both donor
´
and receiver aquaria. We drew a 28.0
frontal side of these aquaria to analyse Ž sh position.
17.5-cm grid divided into 3.5-cm squares on the
The pacus were videotaped for 25 min. At the 5th min the partition between the donor and
the heterospeciŽc Ž sh (or water) aquarium was lifted and transfer of water started.
¢
For videotape analyses, every 30 s we registered the position of the Ž sh s eye in the grid.
These data were then plotted on an X-Y axis. Mathematical analysis considered a set of 10
points per Ž sh each 5 min. Thus, for the grouped (4 Ž shes) and the isolated pacus, 200 and
50 points were collected in 25 min, respectively.
The mean of the Ž sh position on X axis and the mean on Y axis were the barycentric
coordinates(calculatedto each 5-min periodalong 25 min). From these data, threeparameters
were evaluated:
Dispersion
(1)
is the mean of the distances between each position and the respective
barycenter in both receiver and donor aquaria(Thines & Wandenbussche, 1966). The
lower these values, the lower the dispersion(greater the cohesion) of the group.
Distancefrom the visual stimulus
(2)
is the mean distanceof the donor Ž sh from aquarium
with the visual stimulus (trahira, piracanjuba or water). Trahira and piracanjuba were
too long to swim freely in the aquaria.Thus, theystayedmotionlessand the distanceof
the donor pacus to the aquarium with the heterospeciŽc Ž sh indicated approximately
the distance between these Ž sh.
Distance from the water source
(3)
is the mean distanceof the receiver Ž sh from the water
source(end of the pipe).
In the Ž rst 5-min periodof observation(beforemanipulations),the investigatedparameters
±
N
= 6) are expressed
were similar between the conditions, as follow. Mean values ( SE,
Dispersion of the donor Ž sh
±
±
in cm.
: predator = 5.31 1.14; non-predator = 5.96 0.75;
±
F
p
Dispersion of the receiver Ž sh
and control = 6.12 0.94 ( = 0.20; = 0.82).
: predator =
±
±
±
F
p
3.12 1.15; non-predator = 2.66 1.54; and control = 4.20 1.38 ( = 0.33; = 0.72).
Distance from the visual stimulus
±
±
: predator = 10.51 3.24; non-predator = 15.00 1.74;
Distance from the water source
= 7.24 0.92; non-predator = 14.19 4.05; and control = 12.71 3.68 ( =
±
F
p
and control = 10.22 2.16 ( = 1.18; = 0.33).
: predator
±
±
±
F
1.30;
p
= 0.30). Therefore, we analysed the effect of either visual stimulus or water transfer by
subtracting the results of this Ž rst period from the respective values in the subsequent ones
(post-pre differences). Thus, negative values indicate that pacus decreased the dispersion or
¢
the distance from the stimulus. The medians of these values were compared by Friedman s
¢
two-way analysisof variance followed by Dunnett s multiple comparison test(Lehner, 1996).

CHEMICAL WARNING INFORMATION IN FISH
685
Results
No signiŽ cant effect was detected between the conditions before test stimuli
were imposed. However, signiŽ cant changes occurred after presentation of
the stimuli in all the parameters, except for the dispersion of the donor Ž sh
(Fig. 1).
The visual stimulus of either heterospeciŽ c species or only water did not
x2
signiŽ cantly change the dispersion of the donor Ž sh (Friedman,
= 4.75,
p = 0.93), but the donors were closer to the non-predator and farther away
x2
¢
= 17.33, p = 0.0017, Dunnett s multiple
from the predator (Friedmann,
comparison test) (Fig. 1).
Fig. 1. Donor Ž sh. Behavioural patterns elicited by visual stimuli in juvenile pacus
as measured by change in dispersion (A) and change in distance from cue source (B).
Behavioural changes were calculated by subtracting pre-stimulus (Ž rst 5 min) from post-
±
stimulus scores. Data are expressed as mean values ( SE) obtained from 6 trials. Different
p
letters indicate statistically signiŽ cant differences at
< 0.05 (Friedmann, corrected for
Dunnett multiple comparisons of the medians).

686
JORDÃO & VOLPATO
Fig. 2. Receiver Ž sh. Behavioural patterns elicited by chemical stimuli in juvenile pacus
as measured by change in dispersion (A) and change in distance from cue source (B).
Behavioural changes were calculated by subtracting pre-stimulus (Ž rst 5 min) from post-
±
stimulus scores. Data are expressed as mean values ( SE) obtained from 6 trials. Different
p
letters indicate statistically signiŽ cant differences at
< 0.05 (Friedmann, corrected for
Dunnett multiple comparisons of the medians).
After water transfer, the receiver Ž sh changed their dispersion (Fried-
x2
¢
mann,
= 9.00, p = 0.011, Dunnett s multiple comparison test) and
x2
distance from the water source (Friedmann,
= 11.64, p = 0.003, Dun-
¢
nett s multiple comparison test) (Fig. 2). In the latter case, predator and non-
predator stimuli elicited different patterns of response from the receivers.
¢
Furthermore, the proŽ le of the receiver s distance from the chemical source
¢
stimulus is very similar to that of the donor Ž sh s response to the visual stim-
ulus.

CHEMICAL WARNING INFORMATION IN FISH
687
Discussion
This study showed that predator-naïve juvenile pacu, Piaractus mesopota-
micus, can visually identify and distinguish other Ž sh species as predatory
or non-predatory. This is interpreted as an innate response. Furthermore,
such heterospeciŽ c detection affects the behaviour of pacus, which release
chemicals that induce conspeciŽ cs to adopt a similar behavioural response.
At least two chemicals might be involved, one of them possibly an alerting
pheromone.
Visual recognition of predators
The behavioural response of the donor Ž sh to the view of a heterospeciŽ c
was signiŽ cantly different from that adopted after the view of an aquarium
without Ž sh. Moreover, these responses were species dependent: pacu ap-
proached the piracanjuba but retreated from trahira. Trahira is a predator
Ž sh (Nelson, 1994; Lowe-McConnell, 1995) while piracanjuba is a herbiv-
orous one (Lowe-McConnell, 1995). Therefore naïve P. mesopotamicus dis-
tinguished by visual cues a predator from a non-predator Ž sh.
Many studies have already reported the importance of vision as a mecha-
nism for predator recognition in Ž shes (e.g. Mathis et al., 1993). Such Ž sh
abilities to recognise predators may be genetic and modulated by early ex-
perience (Magurran, 1990; Mathis et al., 1993; Chivers & Smith, 1998).
In our study, because we used naïve pacus, the different responses of the
donor pacus to the predator or to the non-predator strongly suggests that
the discrimination have a genetic basis. Another consideration is that these
heterospeciŽ cs are sympatric with pacu (Lowe-McConnell, 1995) and thus
recognition of the trahira is an important component in the anti-predatory
mechanisms evolved in this species. In fact, according to Malyushina et al.
(1991) predator cues evokes innate responses in conspeciŽ cs and these re-
sponses are more pronounced when predator-prey have coevolved and coex-
ist in a same environment (see this point in Kats & Dill, 1998).
Chemical transfer of warning information among non-injuried conspeciŽcs
Magurran & Higham (1988) showed that minnows (Phoxinus phoxinus),
which could see threatened conspeciŽ cs (but not the predator), modiŽ ed
their own behaviour. But while the visual transfer of information among

688
JORDÃO & VOLPATO
individuals of a shoal may be important, vision may also be limited in the
aquatic environment (Mathis et al., 1993), and thus other sensory abilities
may be involved. The second important conclusion of the present study is
that chemicals released by the Ž sh may mediate such transfer of information.
When pacus were faced with heterospeciŽ cs (it did not matter if it was
a predator or not), they released chemical cues that increased dispersion in
conspeciŽ cs. In such case, however, the change in dispersion may be a mere
consequence of the movements of the pacus approaching to (non-predator
condition) or retreating from (predator condition) the chemical cue. Thus,
the main biological response to chemicals released by conspeciŽ cs is not
related to the dispersion of the Ž sh, but to the direction it is moving to.
Chemical communication is widely used by Ž sh in different contexts, in-
cluding the alarm situation (for review see Liley, 1982; Smith, 1992; Chivers
& Smith, 1998). However, most of these studies report chemicals released
from damaged skin (club cells) (see reviews by Pfeiffer, 1977; Chivers &
Smith, 1998). Literature is very scarce concerning chemical communication
by non-injured Ž sh. Some studies have suggested that stressed Ž sh may re-
lease different chemical stimuli (Todd et al., 1967; Malyushina et al., 1991;
Lebedeva et al., 1994), but only Wisenden et al. (1995) have linked both pre-
dation threats to releasing of chemicals and chemical detection to changes in
vigilance of conspeciŽ c Ž sh. In this context, the present study is the second to
clearly show behavioural modulation by chemicals of conspeciŽ cs induced
by predator presence.
Piracanjuba attracted pacus directly by visual cues and these grouped
pacus attracted conspeciŽ cs by chemical means. The sympatric occurrence
of these species may suggest common habitat preferences and thus staying
closer to each other may be of biological signiŽ cance.
Two chemically-mediated responses were detected in this study (alert
and attraction) and the participation of at least two chemicals is strongly
expected. Nitrogenous waste product, possibly ammonium, has been pro-
posed as component of the disturbance pheromone in crayŽ sh (Hazlett, 1989,
1990), frogs(Kiesecker et al., 1999) and Iowa darters (Wisenden et al., 1995)
and might be involved in the alerting responses described in this study. How-
ever, the two different responses reported here may indicate that at least an-
other chemical may also be involved.

CHEMICAL WARNING INFORMATION IN FISH
689
References
Chivers, D.P. & Smith, J.R.F. (1998). Chemical alarm signalling in aquatic predator-prey
systems: a review and prospectus. — Écoscience 5, p. 338-352.
Guthrie. D.M. & Muntz, W.T.A. (1993). Role of vision in Ž sh behaviour. — In: Behaviour of
teleost Ž shes(T.J. Pitcher, ed.). Chapman & Hall, London, p. 363-439.
Hazlett, B.A. (1989). Additional sources of disturbance pheromone affecting the crayŽ sh in
Orconectes virilis
. — J. Chem. Ecol. 15, p. 381-385.
— — (1990). Source and nature of disturbance-chemical system in crayŽ sh. — J. Chem.
Ecol. 16, p. 2263-2275.
Kats, L.B. & Dill, L.M. (1998). The scent of death: chemosensory assessment of predation
risk by prey animals. — Écoscience 5, p. 361-394.
Kiesecker, J.M., Chivers, D.P., Marco, A., Quilchanos, C., Anderson, M.T. & Blaustein, A.R.
Rana aurora
(1999). IdentiŽ cation of a disturbance signal in larval red-legged frogs,
— Anim. Behav. 57, p. 1295-1300.
.
Lawrence, B.J. & Smith, J.R.F. (1989). Behavioral response of solitary fathead minnows,
Pimephales promelas
, to alarm substance. — J. Chem. Ecol. 15, p. 209-219.
Lebedeva, N.Y., Vosilene, M.Z.Y. & Golovkina, T.V. (1994). Aspects of stress in rainbow
Salmo gairdneri
trout,
. — J. Ichthyol. 23, p. 66-74.
Lehner, P.N. (1996). Handbook of ethological methods. — Cambridge University Press,
Cambridge.
Liley, N.R. (1982). Chemical communication in Ž sh. — Can. J. Fish Aquat. Sci. 39, p. 22-35.
Lima, S.L. & Dill, L.M. (1990). Behavioral decisions made under the risk of predation: a
review and prospectus. — Can. J. Zool. 68, p. 610-640.
Lowe-McConnel,R.H. (1995). Ecologicalstudiesin tropicalŽ sh communities.— Cambridge
University Press, Cambridge.
Magurran, A.E. (1990). The inheritanceand developmentof minnow anti-predatorbehaviour.
— Anim. Behav. 39, p. 834-842.
— — & Higham, A. (1988). Information transfer across Ž sh shoals under predator threat. —
Ethology 78, p. 153-158.
Malyushina,G.A., Kasumyan, A.O. & Marusov, E.A. (1991). Ecological aspectsof chemical
signals in Ž sh. — Scripta Technica, Inc. 31, p. 1-7.
Mathis, A., Chivers, D.P. & Smith, R.J.F. (1993). Population differences in response of
Pimephalespromelas
fathead minnows(
— Ethology 93, p. 31-40.
) to visual and chemical stimuli from predators.
— —, — — & — — (1996). Cultural transmission of predator recognition in Ž shes:
intraspeciŽc and interspeciŽc learning. — Anim. Behav. 51, p. 185-201.
— — & Lancaster, D. (1998). Response of terrestrial salamanders to chemical stimuli from
distressed conspeciŽcs. — Amphibia-Reptilia19, p. 330-335.
Nelson, J.S. (1994). Fishes of the world. — John Wiley & Sons, Inc., New York.
Pfeiffer, W. (1977). The distributionof fright reaction and alarm substance cells in Ž shes. —
Copeia 4, p. 653-665.
Smith, J.R.F. (1992). Alarm signals in Ž shes. — Rev. Fish Biol. Fish. 2, p. 33-63.
Thines, G. & Vandenbussche, E. (1966). The effects of alarm substance on the schooling
Rasbora heteromorpha
behaviour of
Duncker in day and night conditions. — Anim.
Behav. 14, p. 296-302.

690
JORDÃO & VOLPATO
Todd, J.H., Atema, J. & Bardach, J.E. (1967). Chemical communication in social behavior of
Ictalarus natalis
a Ž sh, the yellow bullhead(
). — Science 158, p. 672-673.
Wisenden, B.D., Chivers, D.P. & Smith, R.J.F. (1995). Early warning in the predation
Etheostoma exile
sequence: a disturbance pheromone in Iowa darters (
Ecol. 21, p. 1469-1480.
). — J. Chem.