J. Fish B i d . (1977) 11, 363-376

A review of chemical communication in freshwater fish D. J. SOLOMON

Salmon arid Freshwaler Fisheries Laboratory, Ministry of Agriculture, Fisheries and Food, I0 Whitehall Place, London SWIA 2HH

(Accepted 2Y August, 1976)

Since communication between individuals of a species of fish by chemical agents (phero- mones) was first demonstrated in 1932, such a process has been suggested in many aspects of fish behavidur and development. This review describes observations on such mechanisms in shoaling behaviour and beneficial conditioning of water, homing of migratory fish, com- munication of alarm, ' crowding factor ' (which adversefy affects growth, survival and fecundity in dense populations), pair formation and spawning, and a range of other social interactions. Some of the chemicals involved have been isolated and identified, but most are indicated by behavioural observations. Pheromones are of great significance in fish behaviour and ecology, and are likely to be an important factor in culture operations.

I. INTRODUCTION The earliest demonstration of chemical communication between freshwater fish was reported by Wrede (1932), who showed that members of a shoal of minnows Phoxinus phoxinus were capable of detecting the species-specific emanations from individuals of the shoal. Such a mechanism had been considered a possibility earlier but was usually rejected as unlikely in the absence of evidence for similar mechanisms else- where, e.g., Chidester, 1924; see Section 111. Von Frisch (1938) described a reaction of minnows to the smell of physically damaged individuals of the same species, and initiated and inspired a succession of studies that has since dominated the field of fish chemical communication research. Karlson & Luscher (1959) proposed the term ' pheromones ' for such agents of communication, defining them as ' substances which are secreted to the outside by an individual and received by a second individual of the same species, in which they release a specific reaction, for example a definite behaviour or a developmental process '. Most pheromone systems studied have been in insects, but they have also been implicated in various aspects of fish behaviour, including shoaling, territory formation, pairing and reproduction, homing, and crowding- influenced depression of growth, reproduction and viability.

Bardach & Todd (1970) and Pfeiffer (1974) have partially reviewed the literature; this review attempts to cover the whole field of chemical communication in freshwater fish rather than, as the previous reviews, dealing mainly with one aspect in detail in order to indicate principles. Where a considerable literature exists on one aspect not all studies can be given consideration here, and key papers and reviews will be referred to: thus no attempt is made to make the reference list exhaustive.

As more behavioural processes are being shown to be influenced by pheromones there has been a move to reconsider various processes with the possibility of such a mechanism in mind; several contributions to our knowledge have been made in this way. Therefore the intention of this paper is to provide information and to stimulate

363

364 D. J . S O L O M O N

consideration of various aspects of fish behaviour and development that might involve pheromones. For the purposes of this review the term communication is used in a wide sense to cover the passage of any information between individuals.

11. SHOALING BEHAVIQUK AND CONDITIONING OF WATER Several authors have analysed the various behavioural factors involved in shoaling,

e.g., Wrede, 1932; Keenleyside, 1955; and Hemmings, 1966. Wrede (1932) observed that the mucus of minnows (Phoxinus laevis L.=Phoxinus phoxinus (L.) ) attracted other members of the same species. Blindlng, or removing the forebrain, did not stop the reaction or prevent learning. It was suggested that vision kept a shoal together during daylight, and the attractive odour at night. In his study on shoaling in roach [RutiZus rutilus (L.)], Hemmings (1966) discussed the theory that inter-individual distance in shoals is brought about by an equilibrium between attractive and repulsive forces. Vision was clearly the most important sense in daylight at least, and olfaction appeared to play a role in preventing break up of the shoal at night. The repulsive element appeared to operate through the lateral-line pressure sense. Roach were strongly attracted to the odour of other roach in a ' choice chamber ' experiment.

Keenleyside (1955) made similar observations of schooling in rudd [Scardinills erythrophthalmus (L.)] and three-spined stickleback [Gasterosteus aculeatus (L.)]. Blinded rudd failed to maintain a proper shoal, but did remain in the area of the odour of other rudd. This would allow the shoal to remain aggregated during dark- ness.

Hoglund & fistrand (1973) studied the odour preferences of young char [Salvelinus alpinus (L.)] in a flowing water ' Y ' choice chamber. Intact fish showed a preference for the water flow containing the odour of other char to that containing no such odour. The preference disappeared when the olfactory epithelium was destroyed.

As distinct from shoaling behaviour, many species of fish appear capable of con- ditioning the surrounding water to produce an effect on other individuals. The effect can be harmful or inhibitory (crowding factor, dealt with in Section V) or advan- tageous. Allee et uZ. (1940) described experiments in which goldfish [Carassius aurafus (L.)] grew better in water which had previously held other goldfish, than in ' fresh ' water. The main factor involved appeared to be the calorific contribution of small suspended particles of regurgitated food, but a chemical secretion was also involved. This was indicated by the beneficial value of water which had contained only starved conditioning fish, by the stimulating properties of filtered conditioned water, and the beneficial effect of protein extracts from the mucus of other goldfish. A 2.5 p.p.m. aqueous solution of an extract promoted a significantly increased growth rate, an observation which indicates that it was not the direct food value of the secretion which was responsible.

McCaulcy (1968) found that some factor in water which had previously held other individuals conferred an advantage on rainbow trout (Salnzo gairdneri Richardson) experiencing thermal stress. Conditioned water was prepared by maintaining the experimental temperature until heat death of the first batch of fish occurred. Time taken for heat narcosis to set in was significantly longer in a second batch of fish in the conditioned water than for a control group in ' fresh ' water.

Miles (1968) found that the presence of adult eels (Ai?guiZla rostrata Le Sueur) in a stream rendered it more attractive to migrating elvers, whereas the presence of large

C H E M I C A L C O M M U N I CAT I 0 N I N F R E S H W A T E R YI S €1 365

numbers of other elvers reduced the attraction. The benefit of this mechanism in distributing elvers to areas which are suitable for survival and growth of eels is obvious. Only in this last case has any attempt been made to identify the chcmicals responsible. Miles (loc. cir.) found the agent to be biodegradeable, unaffected by autoclaving, non-volatile, and to have dissolved and suspended fractions.

111. HOMING OF MIGRATORY FISH Pheromones appear to play a role in homing behaviour of migratory fish, in what

can be considered a specialised type of aggregating behaviour. Chidester (1924) first recorded the suggestion that Atlantic salmon (Salnzo salur L.) responded to emana- tions of young fish in the river, but concluded that it was ' rather unlikely, although it is an interesting conjecture '.

White (1934a) presented the earliest experimental evidence for such a mechanism in a study on the two branches, East and West, of thc Apple River in Nova Scotia, which have a common estuary. Salmon were observed to ascend only the West branch, even though there was no barrier on the East where a mill dam had been removed some ten years earlier. In 1932,25 000 fry of unrecorded origin were planted in the East branch, and later that year a considerable run of adults was seen in the stream. The followiiig spring there were an estimated 350000 newly hatched fry present indicating successful spawning (White, 19346). It seems that the presence of young fish in the stream had rendered it attractive to adult fish, but without knowledge of the origin of the planted fry no further inference can be drawn from these results. It was suggested that the milt produced by precocious male parr was the source of the attractant but there was little evidence to support this.

Surprisingly, little further attention was paid to the possible role of pheromones i n homing until Nordeng (1971) suggested that a race-specific emanation from young niigratory char [SalveEiirtts alpinus (L.) J attracted adults of that race. Artificially fertilised eggs from char in the Salangen river in Northern Norway were flown to a hatchery several hundred km south, and reared for four years, when they were ready to migrate. They were flown back to their area of origin, and 143 released in their river of origin, and 174 in the L~ltsebotn estuary, about 10 km distant. Of the first group 27 were recaptured, all but one in the Salangen river. Of the second group, 31 were recaptured, 21 in the Salangen and 10 in the Laksebotn. These results are interpreted as deinonstrating that the fish were attracted to their ancestral river, which in the case of the L~ksebotn release they had never previously experienced, by a race-specific emanation from their relatives upstream. This would not of necessity be an inherent recognition and attraction, as the fish had of course been reared together and could have learned to recognize their race odour. Further evidence was given by the behaviour of a group of 200 tagged Salangen char released in the river near the south Norway hatchery where they were being reared. After two weeks many of these fish were observed to have gathered beneath the floor of the hatchery by the outlets from the tanks still containing 2000 of the original hatch. The local population of non-migratory char were not observed to behave in this manner.

Further evidence for the involvement of pheromones in the homing of Atlantic salmon was presented by Solomon (1973). Observations on the mixing of stocks of salmon between the estuaries of neighbouring salmon rivers were contrasted with the rarity of the presence of ' wandering ' fish in the estuaries of non-salmon bearing rivers.

366 D. J. SOLOMON

Further evidence was provided by the commercial catches on the River Parret in Somerset, England. A small net fishery, representing a fairly consistent effort, has existed for a considerable time on the estuary. Although it is not a salmon river, it is assumed that a very small proportion of the stocks returning to surrounding major salmon rivers frequented the estuary. Following the stocking of a tributary of the Parret with salmon eggs from a variety of distant sources, the estuary catch was considerably increased. It was shown that the increase in the catch was mainly of fish not derived from the stocked eggs; it appeared that a greater number of fish from neighbouring streams was entering the estuary. This situation was interpreted as indicating the existence of a general attraction for adult salmon to young fish in the river. It appeared that fish had to penetrate a considerable way into the estuary in order to determine whether it was their home river. There is little to indicate whether the factor finally identifying the river is a river-race specific pheromone which is only distinguishable from that of nearby races in greater concentrations, or whether a further mechanism is involved.

Dsving et al. (1973) examined the mechanism of race discrimination in the char populations studied by Nordeng (Ioc. cit.) by recording the electrical activity of the olfactory bulb and telencephalon of fish experiencing the odours of various races. They found the smell of conspecifics caused a definite response, but were unable to demonstrate the fish’s ability to discriminate between races. Later experiments (Dsving et al., 1974), involving recordings from single cells in the olfactory bulb indicated an ability to distinguish between fish from different areas. Mucus taken from fish appeared to provide the information for this discrimination.

IV. ALARM SUBSTANCE A very considerable literature has accumulated on chemical communication of

alarm since the mechanism was first described by Von Frisch in 1938. Pfeiffcr (1974) has provided an extensive review, so only a selection of the more important observa- tions, and results published since that review, will be considered in detail here.

DESCRIPTION AND OCCURRENCE Von Frisch (1938) observed that a damaged minnow conveyed alarm to other

members of the shoal nearby, and his experiments indicated that this was mediated by a water-borne chemical from the damaged skin of the fish. He then demonstrated the existence of such a mechanism in many species of fish, mainly cyprinids. Other authors have demonstrated the alarm substance in many species, e.g. Schutz (1956) in 19 species of cyprinids, 1 cobitid, and 2 characids; Reed (1969), in 3 cyprinids; Reed et al. (1973), in 2 cyprinids; and many others. Pfeiffer (1974) reviev,s the occurrence throughout the Pisces andconcludes that it is restricted to theostariophysi and Gonor- rhynchiformes and thusmissing from many groupsof shoaling fish, e.g. Clupeidae and Percidae. From various observations, including species which possess the ability to react to but not to produce alarm substance, it was concluded that some groups, mainly those with a specialised mode of life, e.g. blind cave fish and piranha, had secondarily Iost the alarm substance mechanism. I t has often been suggested, and finally demonstrated (Reutter & Pfeiffer, 1973), that the alarm substance is secreted by the club cells, secretory cells in the skin without ducts to the surface. The ability to produce and respond to alarm substance is not present on hatching. For example the

CHEMICAL COMMUNICATION I N F R E S H W A T E R FISH 367

reaction appears in minnows at about 50 days, the substance being present in the skin for a week or so before this. The shoaling habit develops at about 25 days (Pfeiffer, 1974).

THE ALARM REACTION Von Frisch (1938, 1931) classified the alarm reaction of minnows into six categories

according to intensity, from apparent sudden shock, dashing to a hiding place swimming rapidly round the tank, followed by a long-term avoidance of the locality of the experience, to an apparent slight unease on the part of the fish. He also found some individuals were unaffected on some occasions. Many other species show a similar range of reactions while others demonstrate some specialisation consistent with their way of life and habitat. The tench Tirzca tirzca (L.), which inhabits ponds and slonr-flowing rivers, adopts an attitude of about 60" to the bed with its head down and swims excitedly, stirring up the mud and debris to hide itself (Pfeiffer, 1974). The cyprinid Hybopsis aestiilalis, a somewhat sedentary bottom dweller, remains motion- less (Reed, 1969).

BEHAVIOURAL ASPECTS It has been suggested that the alarm substance mechanism serves to prevent young

fish being eaten by adults of the same species, or by other predators possessing the alarm substance reaction. (Von Frisch, 1941). Schutz (1956) considered the observa- tion that young fish can produce alarm substance for some time before they themselves respond to it supported this hypothesis. However, there is evidence that such a mechanism does not operate. Verheijen & Reuter (1969) studied the process of intra- and-inter-specific predation, including the effect of any alarm substance on the predator, and on ' witnesses' to the event. They observed that large roach and minnows consumed small roach without any apparent alarm reaction. Blinded minnows reacted to water taken from a tank in which a pike Esox Iucius L. had consumed a small roach, whereas no reaction was seen to water in which a minnow had consumed a small roach. It was suggested that two mechanisms could beinvolved; first, a lower sensitivity of larger fish, and second, minimal damage being done to the prey during consumption by a cyprinid. Any damage done by the pharyngeal teeth would release the minimum of alarm substance to the outside, in contrast to the method of feeding of the pike, which does considerable surface damage to prey with its mandibular teeth. It is therefore suggested that the alarm reaction is not a major factor in preventing intra-specific predation.

A chemically elicited alarm reaction can be passed to other individuals visually. Schutz (1956) observed the reaction could be passed between adjacent aquaria of minnows in this way. This is a very meaningful behavioural feature, as it suggests that only one member of a shoal needs to detect the pheromone to induce a reaction in all individuals. The communication of a visual signal is likely to be far faster than a chemical one.

A further aspect of chemical communication was reported by Goz (1941). He found the majority of a population of blinded minnows being studied showed an alarm reaction, either flight or ' freezing ', to the smell of pike. A proportion of the minnows which did not show such a reaction did so after they had been present when a conspeci- fic was devoured. It does appear that minnows learn to recognise pike scent as dan- gerous by association of minnow alarm substance and pike odour.

368 D. J. SOLOMON

The reproduction of some cyprinids involves physical abrasion by contact with the spawning substrate. Smith (1973) studied the apparent suppression of alarm substance production by male fathead minnows Pimephales promelas Rafinesque, during spawning when damage to the skin occurs. He Sound treatment with testosterone markedly reduced the number of club-cells in the skin. The mechanism is apparently appropriate for several reasons. It allows year-round maintenance of alarm substance potential by females and juveniles. The loss of potential by the males is not signifi- cantly detrimental as during the spawning period the males become solitary and territorial, reducing the value of the alarm substance mechanism.

CHEMISTRY Unfortunately relatively little progress has been made in the identification of alarm

pheromones in fish. Tt has been observed that the alarm substances of cyprinids are effective in producing a reaction in a range of related species, but that the intraspecific reaction is strongest (Schutz, 1956). This suggests that the substances are chemically similar, but with a large variety of subtle specific variation. Huttcl (1941) suggested that minnow alarm substance was purine-like or pterin-like (double ringed compounds usually associated with pigments), was non-volatile and extremely water soluble. Schutz (1956) tested a suggestion that ichthyopterin was the agent, but concluded that while there were chemical similarities, it did not induce a fright reaction. Reed et a!. (1973) found that the fright reaction was induced in some cyprinids by histamine, though it was not as strong as the reaction to the specific agent. They suggest that a ringed or double ringed compound might be involved and that the phenomenon of reduced but positive interspecific effect suggests the substance could be a polypeptide- like molecule, with a high potential for specific variability in the order of sub-units. Pfeiffer & Lemke (1973) concluded the alarm substance was probably a pterin, and were able to demonstrate an alarm reaction to isoxanthopterin.

In common with most pheromone systems, communication can be effected by very low concentrations. Schutz (1956) found a standard extract of minnow skin made by placing a lacerated piece of skin weighing 200 mg into 200 ml of water could be detected by minnows at a dilution of 1 in 50 000. Pfeiffer (1973) reports an experiment in which Danio malabaricus (Jerdon) reacted when a 5 ml sample containing 0.25 pg of extracted alarm substance was poured into a 20 litre aquarium.

V. CROWDING FACTOR The complex interactions of individual fish under crowded conditions, resulting in

overall decrease in growth rate (stunting) or differential growth rates, and adversely affectcd performance, have been recognised for a considerable time. However, the possibility that chemical communication might play a major role was largely over- looked until Rose (1959a, b) conducted experiments on several species of fish under densely crowded conditions, following earlier studies on frog tadpoles. In the first experiments (Rose 19594 observations were made on breeding white cloud mountain fish Tanichthys albonubes Lin. A pair in a 15 1 aquarium would produce up to 200 fertile eggs, but the tank appeared capable of supporting only about 20 1-cm fish. Shortly after feeding commenced, size differences became apparent. The larger fish continued to grow, but the smaller ones stopped eating and died in spite of an abun- dance of food. Tf removed to another tank, the smaller fish grew well. The situation

CHEMICAL C O M M U N I C A T I O N JN F R E S H WATER FISH 369

with Burbus tetrazona (Bleeker) was similar; only 15 I-cm fish from 200 fertile eggs could be supported. However, when half the water in the tank was replaced four times daily, up to 174 could be reared to that size. This demonstrates that the effect was not due to physical bullying or dominance. Rose (1959b) studied the phenomenon of apparently reduced praduction of young by the live-bearing guppy Lebistcs rcticuh- tus (Peters) under crowded conditions which had previously been ascribed to cannibal- istic infanticide. Different densities (14, 10 and 2 per 12 1 tank) were established, and all young produced were removed immediately. Results indicated that the number of young per female was inversely proportional to the number of females, i.c., similar numbers were produced in each tank. Presence of Tanichthys albonubes did not affect the situalion-indeed, they appeared to enhance the survival of the young. Fecundity was increased when a considerable proportion of the water was replaced each day.

Rose & Rose (1965) reviewed the field of control of growth and reproduction i n freshwater organisms by specific products. They suggested that the results of Brown (1946) indicated such a mechanism. In her experiments two groups of similar sized sibling brown trout Salmo trutta L. were reared in aquaria and fed ad libitum. A considerable size range soon appeared in each tank, with the smallest individuals dying. After a few months the fish were reallocated into two new groups, large and smaller individuals. The growth of the smallest larger-fish group was slowed down, and that of the largest smaller-fish group was accelerated, to such an extent that after seven months the weight ranges of the two populations were the same. It was assumed at the time that the explanation lay in some psychological dominance of the large over the small.

Some of the findings of Rose & Rose (1965) on frog tadpoles Ram spp. suggest features of the mode of action of chemically mediated inhibition. Large and small tadpoles produce a siiiiilar amount of crowding factor per unit of body weight, but the small individuals are more sensitive to its effect. In a crowded group of individuals of a range of sizes the growth of the larger ones will be slowed, while that of the smaller ones is stopped completely. A single individual, or a group, would never produce enough crowding factor to stop all growth, indicating the existence of a regulating negative feedback. A rapidly growing tadpole produces a greater amount of inhibiting factor than a slow growing one of the same weight.

Yu & Perlmutter (1970) examined growth inhibiting factors in the zebra fish Bruchydanio rerio (Hamilton-Buchanan). Initial experiments involved the collection of crowding factor with a charcoal filter, followed by elution with chloroform, ethanol, acetone, ether, petroleum and water. The residues were collected after evaporation of the solvents, redissolved in water, and administered in a bio-assay tcst usingzebra fish embryos. Only the chloroform fraction had any effect, retarding devel- opment and depressing heart rate. In a second series of experiments on three stages of the fish, crowding factor was removed with a methyl choloroform filter in experi- mental tanks, control tanks having no such filter. In post larval and adult fish indivi- dual growth rate appeared to be unaffected by crowding factor, but mortality was increased. The growth of juvenile fish was retarded, but it recovered when a methyl chloroform filter was used on the retarded control group tank. The chloroform extract did not have any effect on embryos of the blue gourami, Trichogaster tricliop- terus (Pallas), or Rivulus harti (Boulenger).

Yu & Perlmutter quote unpublished findings of Greene who extracted a species- specific spawning inhibition factor from goldfish water with chloroform. Swingle

370 D. J. SOLOMON

(1 953) discovered several species of fish produced such a spawning inhibition factor. If water in which a dense population had been held was presented to uncrowded fish about to spawn, it prevented spawning, whereas if crowded fish were given adequate fresh water, spawning took place.

Francis et al. (1974) attempted to identify crowding factor, using two bioassay techniques, from the effects on growth and heartbeat rate of goldfish. Reduced growth-rate of fry indicated the presence of a retarding factor. The heartbeat assay was used in two ways-visual counts of heartbeat in newly hatched fry under a micro- scope, and ECG recording from implanted electrodes in juveniles. Heart rate in fry dropped from 200h25 to 125115 in the presence of crowding factor, and in juveniles, from about 140 to 70. The factor was extracted from the water with 1,1,1 trichlore- thane and carbon tetrachloride, or concentrated in an activated charcoal filter and eluted with chloroform. Fractionation was carried out through a silicic acid column. Some results of this study have been presented by Pfuderer et al. (1974). The effective crowding factor was eluted from the silicic acid (or Sephadex LH-20) columns with the diglyceride fraction. A different and rather variable effect was caused by the sterol fraction. Further analysis was complicated by the presence in the fish tissues and water of foreign phthalate esters presumed to be derived from commercial plasticizers, though the peak activity associated with the natural diglycerides had a UV spectrum distinguishable in detail from that of the foreign phthalate esters. I t was concluded the crowding factor may have been a chemical similar to these esters, or a complex between a natural phthalate and other lipids. In all these studies the fish have been held in crowded conditions in containers of limited size. In nature, levels of crowding below those causing retarded growth and reproduction, may serve to dis- perse fish and prevent undue competition. Miles (1968) observed the presence of large numbers of elvers in a stream rendered it less attractive to other migrating elvers. It has been suggested that a reaction of young Atlantic salmon to skin extract of other individuals, involving a rolling movement from side to side and disturbed frequency of respiratory movements of the gill covers, may be a territorial defence reaction to the proximity of a potential rival (Marusov, 1975). The suggestion that ' crowding factors ' may operate a dispersal process is not in conflict with the observations on the attraction of ' conditioned ' water in some species; as in shoaling discussed in Section 11, optimum dispersion may be achieved by a dynamic equilibrium between attractive and repulsive influences.

When chemical emanations produce direct physiological effects, as may be the case in some of the experiments outlined and in beneficial conditioning of water, rather than initiating a behavioural response via a sensory receptor, there is some doubt whether they fall strictly within the definition of pheromones proposed by Karlson & Luscher (1959). Where a crowding factor works under natural conditions to effect dispersal at lower concentrations than cause the direct physiological effect, it would appear to fit the definition. It is considered that such a distinction is not im- portant for the purposes of this review, and that it is useful to consider together a wider range of chemical communication.

VI. PAIR FORMATION AND SPAWNING Although many pheromone influenced reproductive mechanisms have been observed

in insects, until recently there was surprisingly little information of the role of chemical

CHEMICAL COMMUNICATION I N FRESHWATER FISH 37 1

communication in the spawning of fish. Bardach & Todd (1970, citing Roule, 1931) record that the reproductive activity of male shad A h a sp. may be initiated by secre- tions from the female, and Belding (1934) mentions the possibility of a similar mecha- nism in Atlantic salmon. Very few detailed observations had been published, until Timms & Kleerekoper (1972) studied the attraction of ripe inales of the channel catfish Ictaluruspunctatus to emanations from ripe females. The males located the source by a tropotactic response, turning continuously towards a greater concentration. Thus a point source of the pheromone was circled in one direction, keeping the greater concentration on one side. The authors commented on the lack of fast olfactory adaptation to the pheromone, which is a normal feature of fish olfaction. They suggested that either such adaptation was lacking in the species, or that the fluctuating stimulus experienced during the locating procedure prevented adaptation. It could further be suggested that in the case of such a specialized system as pheromone detection and response, the normal olfactory reactions, which are obviously advanta- geous in most situations, may be overridden or do not operate. A change in the electrical activity of the olfactory bulbs, both evoked and spontaneous has been reported in male goldfish treated with sex hormones (Oshima & Gorbman, 1968). Timms and Kleerekoper (loc. cir.) also reported that fishermen in the Mississippi basin use ripe female catfish as a lure in traps to catch large numbers of males. Similar observations on sex attractants have been made by Newcombe & Hartinan (1973) on rainbow trout SaZmo gairdmri. In an experimental Y maze, both sexes were strongly attracted to water taken from downstream of spawning fish, were weakly attracted to water associated with a ripe female, and females were weakly attracted to ripe males. These authors quote Gandolfi (1949) as postulating a sex attractant in the guppy. The author has received reports of male sea trout (migratory Salmo trutta L.) being attrac- ted to the outflow from a pen containing ripe females. Partridge et al. (1976) found that ripe male goldfish were able to discriminate between ovulated and unovulated females by a pheromone apparently contained in the ovarian fluid, which is released shortly after ovulation. They found that the olfactory acuity of males to other odours (brine shrimp extract) was also increased with the increased differential behaviour towards ovulated and unovulated females. This indicates a hormonally mediated modification of the olfactory system coinciding with its pheromone detecting function, similar to that observed by Oshima & Gorbman (loc. cit.).

Apart from these studies and observations there is little information on sex phero- mones in fieshwater fish. In view of the major involvement of pheromones in the sexual activity of other animal groups, and the wide application of chemical communi- cation in other aspects of the behaviour of freshwater fish, it is suggested that such mechanisms may play a major role in co-ordination of maturation and onset of spawning in shoal breeding fish, and in pair formation in pair-breeding fish.

VII. OTHER SOCIAL INTERACTIONS Although in most species there is no parental care of young, or social hierarchy

other than shoaling, in those fish which do exhibit such behaviour, chemical informa- tion can play an important part. Many members of the family Cichlidae have complex parent-young interactions, involving retrieval of straying young by mouth. Noble & Curtis (1939) concluded that parents of three species could recognise their own young beyond about three days old, if the ' foreign ' young were at least two days different

D. I. SOLOMON 372

in age from their own, or if they had had previous brooding experience. Myrberg (1966) further studied these relationships, and found the powers of discrimination were improved when the parents were able to learn the chemical characteristics of their brood in isolation, without other broods in the same aquarium. Chemore- ception appears to be the dominant sense in identifying the young during the first few days, vision becoming more important thereafter.

Brooding behaviour in Hemichroniis bimaculatus Gill (one of the chichlids studied by Noble & Curtis), appears to be initiated by chemical emanations from the young. Kuhme (1963) was able to extend the period of exhibited brooding behaviour by three weeks by presenting to a female water from a tank containing fry.

A detailed series of studies of the role of olfaction in the social hierarchy of the yellow bullhead catfish Zctalurus natalis has been reviewed by Bardach & Todd ( I 970). Only a summary of the work and conclusions will be given here. Individuals could be rapidly trained to discriminate between the odour of two other individuals by a reward/punishment conditioning technique. Thc mucus from the ‘ donor ’ fish did not provide as reliable a signal as the whole fish, indicating the involvement of some other secretion.

Behavioural interactions between individuals involved attack of strangers, accep- tance of acquaintances, and ritual display between previously acquainted individuals where dominance was disputed. Water li-om a tank holding a previously established dominant initiated submissive behaviour in the submissive individual, and water from the submissives tank caused the dominant to attack the apparent source. The fish were basically territorial, but there were soine co-operative aspects of their behavour. When a stranger was introduced into a community containing established dominant and submissive individuals, the submissives fled to the shelter occupied by the domi- nant. They were evicted only when the danger had passed. Chemical recognition of dominance of an individual disappeared after it had been defeated in an encounter, even in fish which has not witnessed the event; that is, tile chemical signals were altered. Bullheads in which the sense of smell was destroyed took a much longer time to establish a hierarchy, with persistant challenging of dominance. I t thus appears that a range of chemical cues is involved in this complex social process.

In a study to identify the source of chemicai information not present in the mucus of catfish ictalurus nebulosus Le Seur, Richards (1974) found that cues for individual recognition were provided by urine and extracts from the urophysis, an organ in the caudal neurosecretory tract. While caudal neurosecretions had not been demonstrated to be voided to the outside of the animal, their involvement in control of urinary bladder function suggested a possible process. I t was therefore postulated that a function of the caudal neurosecretory system could be pheromone production.

VIII. GENERAL CONSIDERATIONS From the diversity of sites of production and chemistry of substances involved in

chemical communication it would appear that many of the systems have developed independently to fulfil a range of similar requirements. In a medium such as water, which is often turbid and carrying a high level of background noise, it is understand- able that such systems have developed.

Major limitations of chemical communication in water are the low rate of travel of the signal, and in many cases the non-directional nature of the information. For some

C H E M I C A L C O M M U N I C A T I O N I N F R E S H W A T E R FISH 373

mechanisms, such as beneficial conditioning and crowding factor, these limitations are unimportant, but for the communication of discrete signals they severely restrict the performance of the system. It is therefore understandable that such a system is often utilised in conjunction with another, for example, with the fright reaction, the signal being passed on visually after the initial chemical reception.

Concentration gradients may be detectable by fish, giving a directional aspect to the signal, or in the case of aggregating pheromones, a kinesis may be involved. In flowing water, the presence of a pheromone eliciting a rheotactic response can operate a very effective homing or pairing mechanism with a very reliable directional element inherent in the signal.

The selective advantages of most of the systems described are obvious, and their evolution understandable. The evolution of the alarm substance/reaction mechanism is more difficult to explain, as in most cases the possession of the alarm substance cells, as opposed to the ability to react to the alarm substance, does not confer any advantage to its possessor. This is especially true if Verheijen & Reuter (1969) are correct in suggesting that the mechanism plays little or no part in reducing intraspecific predation. Bardach & Todd (1970) suggest the alarm substance evolved as an aggre- gating pheromone, effective at low concentrations as it diffused slowly from the club cells to the skin surface. The opposite reaction to very much larger doses, they suggest, is a secondary development. Some insect pheromones work in this manner, causing aggregation at low concentrations, and a dispersion at higher levels. As little work has been done on the chemistry of the aggregating pheromones, there is little evidence for or against the suggestion that these two functions are affected by the same or related chemicals.

A second mechanism that could have affected the spread of the alarm reactions throughout a species is kin selection. If shoals of sibling individuals are maintained, a new trait possessed by a proportion of a brood shoal could be selected for by increased viability of the shoal due to apparent altruistic behaviour of a few members.

The suggestion by Bardach & Todd (1970) that one pheromone can have two different effects a2 different concentrations leads to the consideration of the relation- ship between crowding factor and aggregating pheromones. A simple mechanism of optimising inter-individual distance can be postulated if aggregation and dispersion were affected by the same secretion at different concentrations.

Several behavioural aspects of pheromone action suggest they represent a communi- cation system of much greater sophistication than appears at first sight. The observa- tion by Goz (1941) (see Section IV) that minnows learned to associate the scent of pike with danger after a single experience of an association of pike odour and minnow alarm substance indicates an important application of the mechanism. Similarly, Von Frisch (1941) observed that minnows avoided a feeding tube for some days after an individual had become trapped and damaged by it. The equilibrium of beneficial conditioning, shoaling pheromones and crowding factor probably represents a non- wasteful density-dependent feedback mechanism in optimising population density.

IX. SIGNIFICANCE IN FISHERIES MANAGEMENT AND CULTIVATION It is obvious that a group of mechanisms which play an important role in the lives

of fishes under natural conditions would have implications also for fisheries manage- ment, exploitation and fish cultivation. In the past little attention has been paid to this

374 I). J. SOLOMON

aspect. The manipulation of stocks of migratory or shoaling species is likely to be influenced by pheromones, Re-stocking with eggs or young of a distant race of fish which home by race-specific pheromones may be useless, or even positively damaging to the homing of native individuals. On the other hand, shoahg, homing or reproductive pheromones could be exploited for guiding or trapping fish, as described by Timms & Kleerekoper (1972) who recorded the use of caged female catfish to capture large numbers of males by fishermen in the Mississippi basin. Poor produc- tivity as a result of chemically depressed growth in stunted populations is unlikely to be improved by increasing stock density, which has in the past often been a quickly- adopted panacea for poor angling. Indeed, reduction in stock may well be a better course of action. An understanding of the mechanisms involved could lead to a rational optimising of stock, according to the varied interests of fisheries users.

In the field of fish culture, where fish are often reared in densities never approached under natural conditions, chemical interactions are even more likely to be a significant factor. While build-up of crowding factor is often prevented in intensive units by high flow-through rates of water, in more extensive culture it may be a limiting factor for growth and reproduction. If it were, chemical removal or selective breeding of strains which produced less of the secretion could reduce the influence. Even where such effects were avoided by high water throughput, there remains the possibility of an influence on the downstream natural populations, or other fish farms.

On the credit side, there may be great potential for exploitation of pheromones in fish culture, for example in beneficial conditioning, ' sorting ' of mixed species populations by using aggregating pheromones, and co-ordination of maturation of brood-stocks.

References Belding, D. L. (1934). The spawning habits of the Atlantic salmon. Trans. Am. Fish. SOC.

64,211-218. Bardach, J. E. & Todd, J. H. (1970). Chcrnical communication in fish. In Communication by

Chemical Signals, (J. W. Johnston, D. G. Moulton & A. Turk eds). New York: Appleton-Century-Croft s.

Allee, W. C., Finkel, A. J. & Hoskins, W. H. (1940). The growth of goldfish in homotypically conditioned water: a population study in mass physiology. J . exp. Zool. 84,417-443.

Brown, M. E. (1946). The growth of brown trout (Salino trutta L) I. Factors influencing the growth of trout fry. J . exp. Biol. 22,118-144.

Chidester, F. E. (1924). A critical examination of the evidence for physical and chemical influences on fish migration. J. exp. Biol. 2,79-118.

Dsving, K. B., Enger, P. S. & Nordeng, H. (1973). Electrophysiological studies on the olfactory sense in char (Sabno alpinus L.). Comp. Biochem. Physiol. 45A, 21-24.

Dslving, K. B., Nordeng, H. & Oakley, B. (1974). Single unit discrimination of fish odours released by char. (Salmo alpinus L.) populations. Comp. Biochem. Physiol. 47A,

Francis, A. A., Smith, F. & Pfuderer, P. (1974). A heart-rate bioassay for crowding factors

Gandolfi, G. (1969). A chemical sex attractant in the guppy, Poecilia reticulata (Pisces,

Goz, H. (1941). Uber den Art-und Individualgeruch bei Fischen. 2. vergl. Physiol. 29,145. Hemmings, C. C. (1966). Olfaction and vision in Fish schooling. J . exp. Biol. 45,449-464. Hoglund, L. B. & Astrand, M. (1973). Preferences among juvenile char (Salvelinusalpinus L.)

to intraspecific odours and water currents studied with the fluvarium technique. Inst. Freshwat. Res. Drottningholm 53,21-30.

1051-1063.

in goldfish, Progve Fish Cult. 36(4), 196-200.

Poeciludae). MonitZool. Ital. 3,89-98.

CHEMICAL COMMUNICATION I N FRESHWATER FISH 375

Huttel, R. (1931). Die chemische Untersuchung des Schrcckstoffes aus Elritzenhaut. Natur-

Karlson, P. & Luscher, M. (1959). ' Pheromones ': a new term for a class of biologically

Keenlcyside, M. H. A. (1955). Some aspects of the schooling behaviour of fish. Behaviour 8,

Kuhme, W. (1963). Chemischc ausgeloste Brutpflege und Schwarmreaktionen bei Hemichro-

Marusov, Y . E. (1975). Reactions of young sdhonids to certain natural chemical stimuli.

McCauley, R. W. (1968). Suggested physiological interaction among rainbow trout finger- lings undergoing thermal strcss. J . Fish. Res. Bd Can. 25,1983-1986.

Miles, S. G. (1968). Rheotaxis of elvers of the American eel (Anguilla rostrata), in the labora- tory to water from different streams in Nova Scotia. J . Fish. Res. Bd Can. 25, 1591- 1602.

Myrberg, A. A. (1966). Parental recognition of young in cichlid fishes. Anim. behav. 14,

Newcombe, C. & Hartman, G. (1973). Some chemical signals in the spawning behaviour of rainbow trout (Salmo gairdneri). J . Fish Res. Bd Can. 30(7), 995-997.

Noble, G. K. & Curtis, B. (1939). The social behaviour of the jewel Fish Xemichrornis binmculatus Gill. Am. Mus. not. Hist. B., 76,146.

Nordeng, H. (1971). Is the local orientation of anadromous fishes determined by phero- mones? Nature233,411413.

Oshima, K. & Gorbman, A. (1968). Modification by sex hormones of the spontaneous and evoked bulbar electrical activity in goldfish : dose-response relationships. J . Endocrinol. 40,409420.

Partridge, B. L., Liiey, N. R. & Stacey, N. E. (1976). The role of pheromones in the sexual behaviour of the goldfish. Anim. Behav. 24,291-299.

Pfeiffer, W. (1974). Pheromones in Fish and amphibia. In Pheromones, (M. C. Birch, ed.), 495 pp. Amsterdam &London: North Holland Publishing Co.

Pfeiffer. W. & Lemke, J. (1973). Untersuchungen zur Isoliarung und Identifizierung des Schrcckstoffes aus der Haut der Elritze, Phoxinus phoxinus (L.). J. comp. Physiol. 82,

Pfuderer, P., Wiiliams, P. &Francis, A. A. (1974). Partial purification of the crowding factor from Carassius auratus and Cyprinus carpio. J . exp. Zool. 187(3), 375-382.

Reed, J. (1969). Alarm substances and fright reaction in some fishes from the southeastern United States. Trans. Am. Fish. Soc. 98,664-668.

Reed, J. R., Wieland, W. & Kimbrough, T. D. (1973). A study of the biochemistry of alarm substances in fish. Proc. 26th Ann. Con$ SE Assoc. of Game and Fish Commissioners

Reutter, K. & Pfeiffer, W. (1973). Fluoreszenzmikroskopischer Nachweis des Schreckstoffes in den Schreckstoffzellen der Elritze, Phoxinusphoxirzus (L.) (Cyprinidae, Ostariophysi, Pisces). J. comp.Physiol.82,411-418.

Richards, I. S. ( I 974). Caudal neurosecretory system: possible role in pheromone production. J. exp. Zool. 187,405-408.

Rose, S. M. (1959~). Failure of survival of slowly growing members of a population. Science 129,1026.

Rose, S. M. (19596). Population control in guppies. Am. Mid. Naf. 62,474-481. Rose, S. M. & Rose, F. C. (1965). The control of growth and reproduction in Freshwater

Roule, L. (1931). Lespoissons et le inonde vicantdeseaux. Paris: Delagrave. Schutz, F. (1956). Vergleichende Untersuchungen uber die Schreckreaktion bei Fishen und

Smith, R. J. F. (1973). Testosterone eliminates alarm substance in male fathead minnows.

Solomon, D. J. (1973). Evidence for pheromone-influenced homing by migrating Atlantic

wissetzschaften 29,3 3 3-3 34.

active substances. Nuture 183,55-56.

183-248.

mis hiinadatus (Pisces). Z . Tierpsychol. 20,688-704.

J . Zchthyolog~ 15,341-343.

565-571.

407-410.

Oct. 22-25 1972,608-610.

organisms by specific products. Mitt. itzt. Vereh. theor. angew. Limnol. 13,21-35.

deren Verbreitung. 2. vergl. Physiol. 38,861 35.

Can. J . Zool. 51,875-876.

salmon Sulmo salar (L.). Nature 244,231-232.

376 D. J. SOLOMON

Swingle, H. S. (1953). A repressive factor controlling reproduction of fishes. Proc. Eighth Paeif. Sci. Congr. IIIA, 865-870.

Timms, A. M. & Kleerekoper, H. (1972). The locomotor responses of male Tctalurus punc- tatus, the channel catfish, to a pheromone released by the ripe female of the species. Trans. Am. Fish. SOC. 101,302-310.

Verheijen, F. J. & Reuter, J. H. (1969). The effect of alarm substance on predation among cyprinids. Anirn. Behuv. 17,551-554.

Von Frisch, K. (1938). Zur Physiologie des Fisch-Schwarmes. Naturvissenschaften, 601-606. Von Frisch, K. (1941). Uber einen schreckstoft der Fischhaut und seine biologische Bedeu-

White, H. C. (1934~). Some facts and theories concerning the Atlantic salmon. Trans. Am.

White, H. C. (19343). A spawning migration of salmon in the E. Apple river. Rep. Biol.

Wrede, W. L. (1932). Versuche uher den Artduft der Elritzen. Z . vergl. Physiol. 17, 510-519. Yu, M.-L., & Perlmutter, A. (1970). Growth inhibiting factors in the zebrafish (Bruchydanio

tung. 2. vergleich. Physiol. 29,46-145.

Fish. SOC. 64,360-362.

Board Canada 1933,41.

rerio) and the blue gourami (Trichogaster frichopterrrs). Growth 34,153-175