Colorado State University
Fort Collins, Colorado 80523
Karl von Frisch (1941) first demonstrated that when the European minnow Phoxinus
phoxinus is eaten by a predator, damage to the skin releases an alarm substance
("Schreckstoff", or scary stuff) that elicits a fright reaction in conspecifics.
Fish dart randomly when they first detect the substance, then form a tight school
and retreat from the source.
At first, von Frisch (1941) speculated that this system would be common among
schooling fishes, where the defensive behavior would be most effective. Further
study, however, revealed that the presence of alarm pheromone is taxonomically
distributed. All fish known to have alarm substance were of the order Ostariophysi
(Pfeiffer 1963) until Smith (1979) demonstrated a similar system in two species of
darters, Etheostoma exile and E. nigrum (Perciformes: Percidae). The latter
system seems to be widespread within the Percidae, and has recently been
demonstrated for the gobies Brachygobius sabanus and Asteropteryx semipunctatus
(Smith et al. 1991, Smith & Lawrence 1992), also Perciformes. Finally, chemical
alarm signals have been demonstrated for a sculpin (Scorpaeniformes) (Smith 1992).
Since then, much experimental work has been done on the nature of the response of
Ostariophysan and Percid fish to the alarm substance. Unfortunately, very little
is known of the chemistry of any fish alarm pheromone, and pure pheromone has
never been isolated for detailed chemical analysis. It is known, however, that the
alarm pheromone of a species can affect even members of another superorder, if the
receiving species also has an alarm pheromone system (Mathis & Smith 1993a).
Interspecific and intergeneric responses are common and often as sensitive as
intraspecific responses (Smith 1982). Thus, alarm pheromones or the mechanism to
detect them may be very similar among species.
The possession of an alarm system presents an evolutionary dilemma. The sender is
presumed not to receive any direct benefit since the alarm substance does not deter
predators and fish which are damaged in an encounter with a predator presumably
rarely survive. Threatened but uninjured fish do not release alarm pheromone (von
Frisch 1941, Smith 1979); release of the pheromone is dependent upon mechanical
damage to the skin. This leaves some form of kin selection as the most likely
genetic benefit, by warning related individuals of the presence of a predator.
The Ostariophysan and Percid Alarm Pheromone Systems
In the Ostariophysi, specialized club cells (alarm substance cells, or ASCs) on the
epidermis outside of the protection of scales were identified by Pfeiffer (1960).
These cells have been well characterized by cytomorphology and staining properties.
All species studied with an alarm reaction possess ASCs, and they are unique to the
Ostariophysi. The cells are disrupted by even minor mechanical damage and release
their contents into the environment. ASCs have no other known function.
In the Percidae, injury to the skin also releases an alarm pheromone which affects
other fish in the area. Club cells similar to ASCs have been identified from the
epidermis. They are distinct from ASCs and are not likely to be homologous,
although they are apparently functionally analogous in the alarm system.
In the Ostariophysi alarm pheromones are widespread, and show little correlation
with life history traits. Fish which school, are solitary, bottom-dwelling, or
even predacious have a chemically-based alarm system. This system is innate and
will develop by a certain age regardless of experience. Where different species
begin to differ significantly is in how and when they respond to the alarm
Pfeiffer (1963) looked at the fright reaction of five species of Cyprinidae and two
species of Catostomidae. All of these species school when young. They all develop
ASCs at least ten days before they begin to respond to alarm pheromone. For
example, in the zebrafish (Brachydanio rerio) alarm pheromone is present in the
skin by age 20 days, but they did not begin to respond to pheromone until at least
32 days old, reared at 26øC. Raised at lower temperatures so that development
occurred more slowly, the reaction to pheromone was delayed. Therefore it seems
that the ability to respond begins only after the fish have reached a certain stage
Additionally, the timing of sensitivity to alarm pheromone may vary among
populations of fish. In a study by Waldman (1982), naive zebrafish fry did not
respond to pheromone until 48-52 days after hatching, at least 17 days after they
begin to produce their own alarm pheromone. This study was also conducted at
26øC, although it is not known if diet or population density had an effect. In the
young schooling fish of all seven species, the reaction to pheromone consisted of
tighter aggregation and a net downward movement. In addition, serial 10-fold
dilutions of conspecific skin extract were effective to a 1/1000 dilution.
Upon maturity, reactions to alarm pheromone began to differ. Schooling adults
still retained the same type of reaction, although at very low dosages freezing
behavior, where fish remain motionless for several minutes or more, was also
observed. The longnose dace (Rhinichthys cataractae) is a solitary bottom-
dweller as an adult. When alone, these fish tend to exhibit a freeze behavior or
rapid diving into cover when they sense alarm pheromone. The adult northern
squawfish (Ptychocheilus oregonense) is predaceous as an adult. Young predators
showed a fright reaction after eating another squawfish, and then refused to eat
for several days in some cases. After repeated predation on other squawfish,
however, the fright reaction disappeared. Experienced predators are still
sensitive to alarm pheromone from other Cyprinids. Thus, while an alarm system is
not present only in schooling Ostariophysi, the reaction to alarm pheromone can
change throughout the life of an individual, and among species, depending on the
life history of the species.
Smith (1979) showed a fright reaction to conspecific skin extract by the Iowa
darter (Etheostoma exile) and johnny darter (Etheostoma nigrum), of the family
Percidae. This was the first record of alarm pheromones outside the Ostariophysi.
Reaction to conspecific skin extract consisted of an initial freeze and alertness
followed by reduced movement which may last 30 minutes or more. Darters are
bottom-dwelling fish, often with cryptic coloration, so such a response may reduce
their apparency to predators. In some cases, a rapid darting occurred followed by
freezing. Fish experiencing a higher concentration of the pheromone may dart to
distance themselves from the predator before reducing movement, but this is
The results to this point made it obvious that alarm pheromones in fish had more
complex behavioral effects than assumed at first. The context of the situation in
which the pheromone is perceived could elicit different responses. The plasticity
of the behavioral response to alarm pheromone raised an additional important
question. Could the alarm pheromones of different species be similar enough so
that they could be perceived and exploited by allospecifics? The recipient could
then make an appropriate behavioral response. Clearly, more detailed studies were
Smith (1982) used the darter system to study intergeneric responses to alarm
pheromones. He used three species from each of the three genera of Percids: the
blackbanded darter, Percina nigrofasciata, the naked sand darter, Ammocrypta beani,
and the gulf darter, Ethiostoma swaini. The results seemed consistent with both a
taxonomic explanation and a habitat explanation. P. nigrofasciata and A. beani
responded more strongly to each other's extracts than to extract from E. swaini.
Etheostoma swaini responded more strongly to conspecific extract than to extract
from the other two species. This is consistent with the accepted phylogeny of the
Percidae, which considers Percina and Ammocrypta more closely related to each other
than either is to Etheostoma. Ammosrypta beani showed no significant response to
E. swaini extract but the response to P. nigrofasciata extract was statistically
significant. In addition to this phylogenetic correlation, P. nigrofasciata can be
found in the habitat of E. swaini, with lots of aquatic vegetation cover, and in
the open sandy habitat of A. beani. There may be a selective advantage for
species to retain sensitivity to the alarm pheromone of other species with which it
A recent report (Mathis & Smith 1993a) reinforce the idea that different species
sharing the same habitat, and especially the same predation pressure, would have a
selective advantage if they could perceive allospecific alarm pheromones. This
article demonstrated that the brook stickleback, Culaea inconstans, responds to
injured conspecifics by increased shoaling, an apparent predator defense. This was
the first evidence of an alarm pheromone in the order Gasterosteiformes.
Experiments showed that C. inconstans exhibited the same fright reaction when
exposed to chemical stimuli from fathead minnows (Pimephales promelas).
Conversely, fathead minnows were not sensitive to chemical stimuli from C.
inconstans. The author concludes that since the two species are sympatric in areas
of their range, that C. inconstans derives a benefit from responding to the alarm
pheromone of P. promelas.
Lawrence & Smith (1989) completed a detailed study of the behavioral response of P.
promelas to conspecific alarm pheromone. Movements of solitary fish were recorded
using a tracking system interfaced with a computer. The tracking system consisted
of a grid of light beams which were periodically scanned by the computer;
interception of the light beams by the fish were converted into different behavior
patterns. The limits of this assay were such that only one fish could be tested at
a time (P. promelas is a schooling fish). However it has been documented that
schooling fish will respond to the visual cue of a fish reacting to alarm pheromone
and react similarly (Verheijen 1956). Therefore behavioral responses observed in
this assay are pure responses to skin extract.
Four types of behavior were observed: dashing, slowing and/or freezing (it was not
clear whether these were different responses or levels of intensity of the same
response), exploring, and no response. Four ten fold dilutions of skin extract
were used (0.1, 0.01, 0.001, 0.0001) and all were effective, indicating that the
active space of alarm pheromone can exceed 58,000 L/cm2 skin. The distribution of
the four responses at each dilution provided insight into the flexibility of the
behavioral response to alarm pheromone. At the two highest concentrations, the no
response was rare or not at all observed. Slowing/freezing and dashing were the
most common types of behavior, and exploring was never observed. The dashing is a
random, rapid movement, assumed to be a predator escape mechanism; this behavior is
very common in fish. It can also be elicited by mechanical and visual stimuli. At
high alarm pheromone concentrations, a feeding predator is likely to be nearby, so
this behavior is appropriate. Dashing was always followed by reduced activity,
which would make the fish less conspicuous to the predator.
At the lower concentrations, dashing was much less frequent. Perhaps the fish did
not perceive an immediate danger. Reduced activity was the most common response,
which almost always lasted through the eight minute observation period. Exploring
behavior also appeared and was about half as common as reduced activity. This
response resembles food search behavior (Lemly & Smith 1985); the skin extract may
contain substances that elicit food searching and override the fright response at
such low concentrations. Exploring is also consistent with "predator inspection
behavior" (sensu Magurran 1986) seen in the European minnow Phoxinus phoxinus.
These fish will inspect an area to gain information on predators in the area. The
low concentrations of alarm pheromone would alert prey to a predator in the area.
The alarm pheromone also can chemically label predators which feed on fish with
alarm pheromone. (Mathis & Smith 1993b) demonstrated that water from a tank
occupied by a northern pike (Esox lucius) which was fed on Pimephales promelas (our
friend the fathead minnow) elicited a fright response in naive P. promelas. Such
a response was not observed upon exposure to chemical stimuli from a pike that had
eaten green swordtails (Xiphophorus helleri) with no known alarm pheromone. It was
assumed that the pike excretes the pheromone into the environment. Thus prey would
be alerted to the presence of a predator well after a predation event has occurred.
In a field experiment, bait traps with similar pike/alarm pheromone stimuli were
conspicuously avoided by larger (presumable older and more experienced) Cyprinids.
In addition it was demonstrated in the laboratory that after experience with
pike/alarm substance cues, the same fish gave a fright response to pike stimuli
alone (Mathis & Smith 1993c). This suggested that learning was going on: the fish
were able recognize a predator based on a previous association of the predator
chemical signature with alarm pheromone. Chemical Composition of the Ostariophysan
Pfeiffer & Lemke (1973) showed that the compound isoxanthopterin elicited a fright
reaction in the giant danio (Danio malabaricus); the response was not identical to
genuine alarm pheromone. Pfeiffer (1978) also showed several heterocyclic
compounds to be effective in eliciting a fright response. These results were
qualitative, and no idea of the quantitative response of fish to these compounds
compared to genuine alarm substance was shown.
A quantitative study was done using the black tetra, Gymnocorymbus ternetzi
(Pfeiffer et al. 1985). A videorecorder measured the change in the angle of
inclination of the tetra toward a light source (apparently a response to make this
reflective fish less conspicuous). This provided a quantitative measure of the
alarm response. Hypoxanthine-3(N)-oxide was found to be as effective as skin
extract in producing the behavioral response. It was concluded that this compound
is the active component or most important active component in the alarm pheromone.
This same compound has been found in Phoxinus phoxinus skin extract. Evolution of
Chemical Alarm Signals in Fish
In the absence of group or species selection (a theory abandoned by most
evolutionary biologists), what benefit could the sender gain by warning
conspecifics? Such signals are generally thought to have evolved by direct
benefits to the emitter by behavior modifications induced in receivers (Weldon
1983). The fright reaction carries no benefit to the individual being eaten. The
response to certain metabolites released during predation would be selected for in
the receiver if the behavior produced served as a defense against predation. This
seems to be the case in fish alarm pheromones. The receiver is the obvious
beneficiary, and it is not hard to envision such traits being selected for.
However, this does not explain the inception of the alarm system in the sender.
The fact that allospecifics benefit is less of a dilemma. Once the system
appeared, another species which could exploit that system for its own use would
gain the same evolutionary benefits as the sender. But how did this system arise
in the first place? Its presence in so many species suggests a significant
selective advantage. In the Ostariophysi, the ASCs may constitute over 30% of the
epidermis and have no known function other than alarm signalling (Smith 1986). It
should be noted that all of the evolutionary explanations for the presence of ASCs
have centered around the benefit being defense against predators by behavior
changes in the prey.
Direct benefits to the sender are a possibility. It has been suggested that a
predator which has reduced hunting success may concentrate on easier prey or leave
the area entirely (eg., Vinyard 1980). Such mechanisms require that the individual
survives the encounter. Smith & Lemly (1985) reported less than 16% of fathead
minnows survived encounters with a piscivorous bird and reproduced. It would seem
unlikely that this would be enough to drive the production of ASCs into a
population, although data are sorely lacking.
ASCs are present in some species that do not show a fright reaction. These are all
highly specialized Ostariophysi: three species of blind cave fishes, piranhas, and
silver dollars (Smith 1986). In these specialized species, the need for a fright
reaction might be unnecessary because of other defensive capabilities, yet they
still retain the ASCs. They also produce alarm pheromone, since their skin extract
causes fright reaction in other species. The ASCs may be vestigial or may serve
Other specialized Ostariophysans lack ASCs. These all have very effective anti-
predator defenses: extremely cryptic coloration and sand burying ability, armored
scales, or electric defense. This may explain why ASCs were no longer necessary,
but other extremely cryptic, armored, and strongly electric species still have
retained ASCs. The distribution of loss of ASCs or fright reaction will need
further study to shed light on the evolutionary development of ASCs.
The other major line of reasoning for the evolution of ASCs is that of kin
selection. If the alarm pheromone warns related individuals of a predator, this
could provide the necessary genetic advantage. Since most Ostariophysi and Percids
school as juveniles, this is a feasible explanation. Since the release of the
pheromone requires attack by a predator, injury or death would occur in any case
and is not a cost of the alarm system. Unfortunately, very little data exists on
the relatedness of individuals within a school. There is only weak evidence of
genetic structuring in a Cyprinid (Ferguson & Noakes 1981).
Relatedness of juveniles in a school is the most likely explanation, since in only
a very few species of Ostariophysi is there any parental care at all. Therefore,
parents would not provide a benefit for their young. Also, by the time fry become
sensitive to alarm pheromone, juvenile schools have already formed. The effect of
kinship on the evolution of alarm systems in fish remains unclear. Summary
Alarm pheromones in fish have been described from several orders but details are
only known for the Ostariophysi and the Percidae. In the Ostariophysi, specialized
epidermal cells (ASCs) contain the alarm pheromone; these cells have no other
known function. Chemicals released by mechanical damage to the skin elicit a
fright response in conspecifics, and in many cases, other species as well. Active
space of the pheromone is on the order of 104 L/cm2 skin. Interspecific responses
seem to be affected by phylogenetic relatedness of the species as well as the
selective advantage of responding to pheromone of species with which predators are
shared. Behavior induced by the fright reaction is flexible and can change from
species to species (according to habitat and life history traits), during an
individual's lifetime, and according to the concentration of pheromone, in order to
produce an appropriate response. Predators excrete active alarm pheromone after
eating fish with the pheromone, alerting sensitive fish to their presence. Fish
can learn to associate predator chemical signatures with pheromone, so that with
experience the predator stimulus alone can produce appropriate anti-predation
behavior. Hypoxanthine- 3(N)-oxide has been shown to be an important component of
alarm pheromone, but detailed chemical analysis of pure pheromone has not been
done. The evolutionary development of the alarm system in fish is poorly
understood, and is another important avenue of research in understanding this
Ferguson, MM & DLG Noakes. 1981. Social grouping and genetic variation in common
shiners, Notropis cornutus (Pisces, Cyprinidae). Env Biol Fish 6:357-360.
Frisch, K von. 1941. Uber einen Schreckstoff der Fischhaut und seine biologische
Bedeutung. Z vergl Physiol 29:46-145.
Lawrence, BJ & RJF Smith. 1989. Behavioral response of solitary fathead minnows,
Pimephales promelas, to alarm substance. J Chem Ecol 15:209-219.
Lemly, AD & RJF Smith. 1985. Effects of acute exposure to acidified water on the
behavioral response of fathead minnows, Pimephales promelas, to chemical feeding
stimuli. Aquat Toxicol 6:25-36.
Magurran, AE. 1986. Predator inspection behavior in minnow shoals: Differences
between populations and individuals. Behav Ecol Sociol 19:267-273.
Mathis, A & RJF Smith. 1993a. Intraspecific and cross- superorder responses to
chemical alarm signals by brook stickleback. Ecology 74:2395-2404.
Mathis, A & RJF Smith. 1993b. Chemical labeling of northern pike (Esox lucius) by
the alarm pheromone of fathead minnows (Pimephales promelas). J Chem Ecol 19:1967-
Mathis, A & RJF Smith. 1993c. Fathead minnows, Pimephales promelas, learn to
recognize northern pike, Esox lucius, as predators on the basis of chemical stimuli
from minnows in the pike's diet. Anim Behav 46:645-656.
Pfeiffer, W. 1960. Uber die Verbreitung der Schreckreaktion bei Fischen.
Pfeiffer, W. 1963. The fright reaction of North American fish. Can J Zool 41:69-
Pfeiffer, W. 1977. The distribution of fright reaction and alarm substance cells
in fishes. Copeia 1977:653-665.
Pfeiffer, W. 1978. Heterocyclic compounds as releasers of the fright reaction in
the giant danio Danio malabaricus (Jerdon) (Cyprinidae, Ostariophysi, Pisces). J
Chem Ecol 1978:665-673.
Pfeiffer, W & J Lemke. 1973. Untersuchungen zur Bolierung und Identifizierung des
Schreckstoffes aus der Haut der Elritze, Phoxinus phoxinus (L.) (Cyprinidae,
Ostariophysi, Pisces). J Comp Physiol 82:407-410.
Pfeiffer, W, G Riegelbauer, G Meier & B Scheibler. 1985. Effect of hypoxanthine-
3(N)-oxide and hypoxanthine-1(N)-oxide on central nervous excitation of the black
tetra Gymnocorymbus ternetzi (Characidae, Osaryiophysi, Pisces) indicated by dorsal
light response. J Chem Ecol 11:507-523.
Smith, RJF. 1992. Alarm signals in fishes. Rev Fish Biol Fish 2:33-63.
Smith, RJF. 1986. The evolution of chemical alarm signals in fishes. In: Chemical
Signals in Vertebrates, Vol 4. Duvall, D, D Muller-Schwarze & RM Silverstein [eds],
Plenum Press NY
Smith, RJF. 1982. Reaction of Percina nigrofasciata, Ammocrypta beani, and
Etheostoma swaini (Percidae, Pisces) to conspecific and intergeneric skin extracts.
Can J Zool 60:1067-1072.
Smith, RJF. 1979. Alarm reaction of Iowa and johnny darters (Etheostoma, Percidae,
Pisces) to chemicals from injured conspecifics. Can J Zool 57:1278-1282.
Smith, RJF & BJ Lawrence. 1992. The response of a bumblebee goby, Brachygobius
sabanus, to chemical stimuli from injured conspecifics. Env Biol Fish 34:103-108.
Smith, RJF & AD Lemly. 1985. Survival of fathead minnows after injury by predators
and its' possible role in the evolution of alarm signals. Env Biol Fish
Smith, RJF, BJ Lawrence & MJ Smith. 1991. Cross-reaction to skin extract between
gobies, Asteropteryx semipunctatus and Brachygobius sabanus. J Chem Ecol 17:2253-
Verheijen, FJ. 1956. Transmission of a flight reaction amongst a school of fish
and the underlying sensory mechanisms. Experentia 12:202-204.
Vinyard, GL. 1980. Differential prey vulnerability and predator selectivity:
effects of evasive prey on bluegill (Lepomis macrochirus) and pumpkinseed (L.
gibbosus) predation. Can J Fish Aquat Sci 37:2294-2299.
Waldman, B. 1982. Quantitative and developmental analyses of the alarm reaction in
the zebra danio, Brachydanio rerio. Copeia 1982:1-9.
Weldon, PJ. 1983. The evolution of alarm pheromones. In: Chemical Signals in
Vertebrates, Vol. 3. Muller-Schwarze, D & RM Silverstein [eds], PlenumPress, NY.