Studies On the Shoaling Behaviour of Fish: I. Responses of Two Species to Changes of Illumination and to Olfactory Stimuli

One of the most striking characteristics of many species of fish is the fact that they live in shoals or schools, and since the fisheries of the world depend in large measure upon this habit it is one of great importance to mankind. Yet it is a subject that has been surprisingly neglected by students of animal behaviour, with the result that our understanding of the mechanism of shoaling has lagged behind the accumulation of empirical knowledge obtained by the application of echo-sounding and other new techniques in fishery research. These, in addition to providing us with much new information on the movements of fish shoals in the sea, may suggest problems that are suitable for experimental study.

Considerable advances have also been made in recent years in our knowledge of the sensory capabilities of fish and of the physiology of their nervous systems ; and this knowledge too suggests new approaches to the problems of shoaling behaviour.

Most studies on shoaling have been concerned with a single aspect of the subject, such as the sensory basis of the mutual attraction of fish to one another, and there has been as yet no attempt to describe the whole behaviour pattern of a pelagic species in terms of the various external and internal stimuli that are of significance to the fish. As a contribution to this end the following paper describes observations on the responses of shoals of two species of fish found in Caribbean inshore waters to changes of light intensity and to olfactory stimuli associated with food.

The fish used for the studies here described were the silversides, Hepsitia stipes (Müller & Troschel), and the tomtate, Bathystoma rimador (Jordan & Swain), which are common in shallow water around Jamaica and the tropical Western Atlantic. Both species live in shoals but have somewhat different habits in other respects. Silversides fulfil the criteria of an ‘ideal’ shoaling species as defined by Parr (1927); that is to say they are always found in shoals made up of individuals of nearly uniform size ; the two sexes are indistinguishable by any external character ; if the shoal is made to disperse it, nevertheless, reforms again at once, and the fish feed exclusively on plankton. They show no sign of any territorial behaviour and most of the time swim in midwater. I have never seen individuals of this species pick up food from the bottom. In many respects therefore the behaviour of silver-sides resembles that of the pelagic Clupeidae.

The tomtate is mainly a plankton feeder but will also pick up food from the bottom and will attack anything that is not too large for it. Shoals in the sea tend to be associated with features in the submarine landscape, such as the piles of wharves and coral formations, and when alarmed the whole shoal retreats into the most sheltered corner of its territory. The sexes are similar in external appearance but the individuals in a shoal vary in size more than do the silversides. It may be considered representative of a demersal type of shoaling species.

The silversides used in these experiments were 5−7 cm. in length, the tomtates 7−11 cm. Immediately after capture the fish were placed in the experimental aquaria and left there undisturbed for a few days. Silversides formed a shoal within a few minutes and appeared to have accustomed themselves to life in the aquarium within 24 hr. Tomtates usually remained in a corner on the bottom for 2 or 3 days and did not form a shoal which swam freely around the aquarium until 5−7 days after capture. They were fed two or three times weekly on small pieces of beef liver or fresh clam meat and occasionally on live plankton. Small shoals of silversides could be kept for several weeks without special feeding, the fish presumably obtaining sufficient food to maintain themselves from the water supply to the aquarium.

The shoals were discarded and replaced by fresh ones after each series of experiments, which usually lasted from 2 to 4 weeks. This was done to minimize conditioning to their food and other possible changes in behaviour resulting from long periods of confinement, and to permit experiments to be repeated with freshly caught fish.

The experimental aquaria and the arrangement of their water supply are shown diagrammatically in Figs. 1 and 2. Two glass-sided aquaria, A and A′ 150 × 38 cm. and 38 cm. deep, were mounted side by side and supplied continuously with fresh sea water from a common reservoir, R. A master tap, T_1, controlled the overall rate of flow and the delivery rate to each aquarium could be adjusted by the taps T₂ and $T2'$⁠. The system was balanced so that both aquaria received the same amount of water at all times, although the total rate delivered to each aquarium varied between 1·7 and 2·4 l./min. according to the head of water in the supply reservoir. The water was discharged into the aquaria close to one end by a number of small holes bored through 15 mm. internal diameter polythene tubing, the end of which was plugged, and the outlet was by an overflow pipe at the other end. With this system the water drifted slowly through the aquaria from the inlet to the outlet and the mechanical effect of small changes in the rate of supply was negligible. Glass taps and T pieces and polythene tubing were used throughout the system.

For testing the reactions of fish to substances in solution a 1l. aspirator bottle, B and B′, was connected with the delivery tube to each aquarium and mounted just above the water level so as to drain into the aquaria when the control taps, T₃ and $T3'$⁠, were opened. Narrow bore taps were used here and the system was adjusted to deliver 100 ml. from the bottles into the aquaria in about 60 sec. The rate of delivery from the bottles varied a little with the head of water in the common supply reservoir, but never exceeded an increase of 8 % in the total rate of supply to the aquaria. Control tests carried out with sea water in the bottles showed that the behaviour of the fish was not affected by an increase of this amount in the rate of flow. To check that no back flow took place from either bottle into the aquarium on the opposite side of the common supply line the system was tested with solutions of potassium permanganate and indian ink, which showed that all the liquid from each bottle was carried with the supply current into the aquarium with which it was directly connected. The time interval between opening tap T₃ and the appearance of coloured liquid at the inlet to the aquarium was 15−18 sec. and there was good mixing with the sea-water supply during passage down the common part of the tube.

Solutions for testing were made up in sea water from the same source as supplied to the tanks and care was taken to see that it was delivered at the same temperature, which varied from 25 to 27·5°C.

It is not possible with this system to estimate the concentrations of substances in the aquaria, where they were diluted rapidly in a much greater volume of water, but results can be expressed in terms of the total amount of substance delivered and, knowing the rate of supply from the bottles and reservoir, the concentration at the inlets to the aquaria can be estimated.

A removable screen, S₁, was placed between the two aquaria so that the shoals were invisible to each other and each aquarium received illumination only from above and along one glass side. Another screen, S₂, which projected about 50 cm. above and beyond the aquaria, was mounted across the inlet end. The control taps and injection bottles of the supply system were attached to the outside of this screen, which was also provided with a peephole above each aquarium. The experimenter could manipulate the taps and observe the reactions of the fish from behind this screen without being seen by them.

Levels of illumination were measured with an Eel photometer, the photocell of which was mounted pointing upwards just above the surface of the water. During the daytime light from the windows of the laboratory entered through the exposed glass face so that light intensities were more or less uniform throughout the aquaria.

Swimming speeds were measured from 16 mm. ciné films of the fish taken through the open glass side of an aquarium against a 10 cm. square grid marked on the background screen, S₁. Illumination was by 48 in. fluorescent strip lights mounted a few inches above the surface of the water and augmented by two 40 W. Photofloods. Film was shot at 16 frames per second and swimming speeds calculated by counting the number of frames exposed while a fish travelled in a horizontal direction across one or more successive squares. Measurements were made only of fish which appeared to travel parallel with the grid and at right angles to the optical axis of the camera, i.e. in which there was no apparent change of size or of focus while crossing a square. The camera was mounted approximately 150 cm. from the middle of the long side of the aquarium and filming was limited to the four grid squares immediately facing that point, which gave a maximum angle of about 6·5° on either side of the optical axis. Estimates of swimming speeds have not been corrected for the degree of parallax or for the distance between the fish and the background grid, since the errors involved are of minor order.

Silversides and tomtates showed similar diurnal changes in their patterns of shoaling behaviour. During the day, when the illumination at the water surface ranged from 2 f.c. to more than 50 f.c., the fish kept in tight shoals for most of the time near the bottom of the aquaria. They remained more or less stationary for long periods and swam relatively slowly unless alarmed or otherwise stimulated. Average swimming speeds recorded at illumination of 6−13 f.c. were 7·9 cm./sec. for silversides and 12·8 cm./sec. for tomtates.

Movements of a shoal were usually initiated by a single fish, though not always the same one, and the rest of the shoal followed. If, as frequently happened, the others failed to follow it the leading fish soon stopped, turned through 180° and rejoined the shoal. A single fish rarely travelled a distance greater than one and a half times its own length from the main body of the shoal before stopping or turning back.

When the shoal moved it frequently happened that one or more fish remained behind for a second or two. These then swam faster to catch up with the main body and the greater the distance by which they were separated the faster they swam. Fish turning to rejoin or hurrying to catch up with the shoal swam faster than those departing from it.

Swimming speeds of fish within the main body of the shoal were closely adjusted-to each other and in general were slower than those of single fish that had become separated from it. Fish did not keep any definite station within a shoal but continually shifted their positions relative to one another, sometimes lagging a little behind or accelerating and slowing up again.

Deviations from the shoal of single fish in lateral or vertical directions were followed by similar corrective movements. Fish that deviated to one side usually made a complete turn through 360° to rejoin the shoal from the rear.

Some recorded swimming speeds illustrating these remarks are given in Table 1.

The conclusion drawn from these observations is that during the day the shoals are kept in being by a balance between the tendency of single fish to depart and go their own way and the mutual attraction they have towards others of their own kind. As Spooner (1931) found with Morone labrax the coherence of a shoal is the result of continuous adjustments of the speed and direction of swimming.

At night the shoals were dispersed ; the fish swam singly in all directions and at all levels in the aquaria, with no apparent co-ordination of direction or speed. Groups of two or three fish frequently formed for a second or two but dispersed again. Swimming speeds, though not measurable, were markedly faster than by day and certain actions seen rarely or not at all during the day were common at night.

Silversides frequently rose to the surface to investigate or to seize small floating objects, while tomtates repeatedly picked up pieces of matter from the bottom.

To summarize, the fish were more active and did not swim in shoals at night and their behaviour was clearly directed towards the investigation and sampling of potential food.

Shoals were observed continuously from the onset of twilight to see if there was a sudden switch from the day to the night pattern of behaviour at a certain level of illumination. An electric pocket torch fitted with a deep red filter was used as an aid to observation when it became too dark to see the movements of the fish. This had no apparent effect upon their behaviour.

The behaviour of both species followed a similar course as the light intensity decreased and may be best illustrated by the following typical series of observations of a shoal of eleven silversides.

The change in behaviour of tomtate shoals, with the onset of night, followed a similar course and need not be described separately, except to record that they rose to the surface less frequently than silversides but investigated the bottom and sides of the tank very actively, picking up and sampling small objects whether edible or not.

These observations show clearly that the change-over from day to night patterns of activity is a progressive one and there is no sudden switch from one pattern to the other at any level of illumination. The changes in the order in which they become apparent may be summarized as follows :

1. An increase of swimming speed.

2. An increase in the distance that individual fish move away from the rest of the shoal before turning to rejoin it. This is what distinguishes a loose from a tight shoal.

3. Dispersal of the shoal with the fish swimming in all directions; at first only momentary but progressively for longer and longer periods until dispersal is general.

4. Individual fish rise to the surface (silversides) or explore the bottom of the tank (tomtates).

The responses of fish to substances in solution were studied by introducing small amounts of test substances dissolved in sea water into one aquarium by the injection apparatus already described, while an equal quantity of sea water was injected into the other aquarium. Tests were replicated, the shoal in each tank being used in turn as a control for the other. The behaviour of the fish was watched for at least 2 min. before and 5 min. after each injection. If a response was obtained, it usually occurred within a minute of starting the injection. An interval of an hour or more was allowed between successive tests and substances from which doubtful responses were obtained were retested on another occasion. During the whole series of experiments lasting 8 months each preparation and substance was tested several times with different shoals of both species.

The rests were carried out during the day or with the aquaria illuminated after dark and only with inactive or slowly swimming shoals.

The response to an exciting olfactory stimulus resembled in some respects the normal night-time behaviour pattern and consisted of: (1) an increase of swimming speed; (2) dispersal of the shoal; (3) swimming to the vicinity of the inlet and returning thereto repeatedly; (4) in some cases rising to the surface, picking up objects from the bottom and attacking any small object seen in midwater.

It was possible with tomtates, but not with silversides, to distinguish clearly between two types of response to an olfactory stimulus, namely, the detection of a substance in the water and the activation of the fish to exploratory feeding behaviour. A substance was said to be perceived when (1) there was an initial increase in swimming speed which was not maintained for more than 2 or 3 min., and (2) the fish swam to the vicinity of the inlet and sometimes sampled the water issuing from it.

Fish were said to show exploratory feeding behaviour when in addition to the above (1) fast swimming was maintained for 5 min. or longer, and (2) they repeatedly investigated, picked up and sometimes swallowed pieces of filter-paper and other objects from the bottom.

Measurements of swimming speeds of unactivated and activated fish are given in Table 2. Average speeds of fish activated by an olfactory stimulus were approximately double those of inactive ones, while peak speeds were nearly three times as great. The greatest swimming speeds were recorded when fish were in the act of feeding.

Swimming speeds remained high for some time after olfactory stimulation and then declined slowly to about the initial level, which was usually attained in from 15 to 40 min. (Table 2). Shoals reformed during this period.

Shoals of silversides and tomtates responded with intense and sustained exploratory feeding activity to dilute sea-water extracts of a number of natural food substances. Extracts were prepared by chopping with scissors or grinding with sand 1 or 2 g. of tissue and allowing it to stand in a small quantity of sea water for a few minutes with occasional stirring. The suspension was then filtered or centrifuged and the filtrate made up to a convenient volume. Extracts of fresh plankton were prepared by filtering, scraping the plankton from the filter-paper, crushing lightly, mixing with more sea water and refiltering. The effectiveness of preparations was assayed roughly by serial dilutions of the initial extract made up to 100 ml. in sea water. All extracts injected into the aquaria were extremely dilute and without colour or odour detectable by the experimenter.

In a preliminary series of experiments it was found that exploratory feeding activity was evoked by almost any ‘meaty’ extract. A list of substances tested and an indication of their effectiveness is given in Table 3. The most intense reactions were obtained in response to extracts of freshly caught plankton, pieces of the adductor and foot muscles of the lamellibranch Area zebra and the body fluids of the echinoids Diadema antillarum and Lytechinus variegatus. Some idea of the sensitivity of the fish to these substances is given by the fact that preparations containing the equivalent of 20 mg. of Area muscle evoked a clear feeding response. Assuming the active substance to constitute 0·1 % of the wet weight of muscle, this would represent a concentration of the order of 1 × 10⁻⁹ in the water delivered to the aquarium.

Exploratory feeding activity was also obtained in response to filtered sea water which had contained live plankton for a few minutes. Freshly caught plankton consisting mainly of unidentified copepods was brought to the laboratory without delay. Living animals were decanted into a large filter funnel and filtered through Whatman No. 1 filter-paper with continuous addition of fresh sea water. Portions of the animal-free filtrate were then injected into the aquaria.

There was some indication that the effectiveness of preparations was increased by longer contact with the plankton. The shortest period from the collection of plankton to injection of filtrate into an aquarium was 20 min., and 500 ml. of this preparation were required to evoke a response. When, however, the plankton was in contact with the water for 30–60 min., 100 and 200 ml. portions of filtrate were found to be effective.

To test the possibility that the fish were responding to substances originating from dead plankton organisms killed during the manipulations involved in collection and filtration, the injection apparatus was modified so that a flow of water from the supply reservoir passed through the bottles and thence to the inlet tubes leading to the aquaria as shown in Fig. 3. Live plankton were pipetted into the bottles and were prevented from passing through to the aquaria by a piece of gauze stretched over the outlet tube. After the plankton had been introduced, the bottles were first flushed with several litres of water which were discarded in order to get rid of substances released from dead animals which might have been introduced by the pipetting. Bottles which appeared to contain any dead plankton were not used for experiments. After being flushed out the bottles were connected with the inlet tubes of the aquaria.

Fish responded with exploratory feeding behaviour in most experiments when water from the bottles containing live plankton was first injected into the aquaria, but did not long continue to do so if the flow was maintained. Repeated responses were obtained, however, if the flow through the bottles was interrupted for a few minutes and then restarted.

Although no attempt was made to express the effect of plankton extracts or live plankton filtrates quantitatively, these experiments show that exploratory feeding behaviour may be stimulated by substances released into the sea by these organisms and suggest that they are released continuously.

The reaction of fish to a visual and an olfactory stimulus presented simultaneously was investigated in the following way. A square-sided Perspex column, 45 cm. long and 7 · 5 cm. side, was placed in each aquarium as shown in Fig. 4. The columns were fitted with two short side-tubes near the base and the top which were connected to the sea-water supply so that water flowed through them and was carried away to a drain without entering the aquaria. A piece of gauze was fitted across the outlets to retain zooplankton. The columns were left in the aquaria containing shoals of six silversides for 2 days before experiments were carried out, so that the fish might become habituated to their presence. At the end of this period they appeared to take no notice of the columns.

Live plankton was introduced into both columns and the reactions of the fish observed. Initially the shoals in both aquaria showed great interest in the columns, swam repeatedly around them and returned to them again and again. Frequent attacks on the plankton were observed in which the fish collided with the intervening Perspex or turned aside at the last moment. After a few minutes the attacks became infrequent and the shoals resumed their previous pattern of behaviour, swimming slowly around the aquaria and taking little interest in the plankton, though this was still visible to them in the columns.

An olfactory stimulus consisting of 100 ml. of a filtered extract of plankton was introduced from an injection bottle into one of the aquaria in the usual way about 2 hr. after live plankton was placed in the column, and the attacks made by fish on the column were counted. The fish reacted with intense exploratory feeding activity and during the next 3–5 min. made many attacks on the column. Thereafter, the frequency of attacks declined to approximately the same level as before while the fish resumed their normal unactivated behaviour. A second and a third injection of plankton extract were followed by renewed exploratory feeding behaviour and many attacks on the column. The shoal in the other aquarium reacted in a similar manner to injections of plankton extract. The results of both experiments are shown graphically in Fig. 5.

Fish activated in the same way by extracts of Area muscle also made numerous attacks on plankton in the Perspex columns.

These experiments show that when excited by an appropriate olfactory stimulus silversides will make visual attacks on objects to which they have previously been negatively conditioned.

Extracts of Area muscle were prepared in different ways in an attempt to identify the substance or group of substances that evoked exploratory feeding behaviour.

1. 2 · 1 g. of fresh muscle were chopped, ground with sand and boiled for 30 min. in sea water. The mixture was cooled, filtered and the filtrate made up to 250 ml. in sea water. Aliquots of this preparation made up to 100 ml. were injected into the aquaria. Extracts prepared in this way evoked feeding responses in dilutions containing the equivalent of 40 mg. of muscle ; they were therefore to all intents and purposes as effective as cold water extracts. Extracts prepared from second and third boilings of the muscle were effective also, but those from the fourth boiling evoked no response.

2. 4 · 6 g. of fresh muscle were chopped, ground and dialysed through a cellophane membrane for 15 hr. The dialysate, which had a faintly cloudy appearance, was made up to 250 ml. and serial dilutions of this preparation tested in the usual way. Feeding reactions were obtained from dilutions containing 5 ml. (equivalent to 2 %) of the stock solution of dialysate. The non-dialysable fraction was diluted, filtered, made up to volume and tested in the same way. This too evoked feeding responses in dilutions equivalent to 2 % of the stock solution. Another portion of muscle was dialysed for 48 hr. and the dialysate changed three times during this period. The non-dialysed residue still evoked feeding responses.

3. 2 · 8 g. of chopped muscle were extracted successively with a mixture of ether and chloroform followed, by acetone. The extracts were combined and evaporated to a sticky residue which was mixed with sea water. This preparation did not evoke a feeding response. The residue was mixed with sea water, boiled for a few minutes, cooled, filtered, made up to volume and tested in the usual way. Feeding responses were obtained from dilutions equivalent to 10 % of the stock solution.

The conclusion drawn from this series of experiments was that the constituent of Area muscle that stimulates exploratory feeding behaviour is not a protein or a lipoid. It is water soluble, heat stable at 100 ° C. and has a molecular weight not greater than about 5000.

The effect of a number of pure substances in sea-water solution was tested in the hope that one or more might evoke exploratory feeding behaviour similar to that obtained with extracts of muscle or plankton. They were selected on the assumption that a water-soluble constituent of muscle or an excretory product was most likely to yield a positive result. A list of the substances tested and the results obtained are summarized in Table 4. All substances were injected into the aquaria in 100 ml. of sea water and the behaviour of the shoals observed in the usual way. In the case of substances which evoked a response the smallest quantity that was effective is given in the table and for those to which no response was observed the largest amount tested is given.

The results summarized in Table 4 show that a number of substances were detected by the fish, which responded by an increase in swimming speed and congregation around the inlet, but only four substances evoked reactions that approximated to exploratory feeding behaviour as described in a previous section, and in no case was the response so prolonged as with extracts of muscle or plankton. Of these four substances glutamic acid and creatinine evoked feeding behaviour in tomtates only, ammonia in silversides only, while lactic acid was effective with both species.

To test the possibility that some of these substances might represent to the fish only one component of an exciting olfactory stimulus the effect of some arbitrary mixtures was tried. Mixtures containing one or more of the four substances that evoked exploratory feeding in pure solution were usually effective, but not more so than the single substances, and the responses were still of considerably shorter duration than those obtained with extracts of muscle, plankton or other foods.

The conclusion drawn from these experiments is that fish can detect and respond with a change in behaviour to a variety of pure organic substances in trace amounts and a few of these substances may excite a type of exploratory feeding behaviour of short duration.

Tomtates were blinded under 1 % urethane anaesthesia by making a slit in the side of the orbit, removing the lens and expressing most of the aqueous humour so as to collapse the eyeball. The operation could be performed in less than a minute and the fish made a quick recovery. For several days they were darker in colour than unoperated fish and no longer swam with the shoal but spent most of the time swimming slowly on the bottom or resting in a corner of the aquarium. Silversides did not survive this operation.

Ten fish in all were operated on. Three days after operation four of them showed motor responses to sudden changes of illumination and were destroyed. The remaining six showed no sensitivity to light and were used for subsequent tests.

Single blinded fish were placed in separate aquaria, left for 1 or 2 days and then tested for their response to an extract of Area muscle in the usual way. In every case a characteristic exploratory feeding reaction was observed, which differed from that of normal fish only in the fact that blinded ones remained in contact with or very close to the bottom of the aquarium, swimming in tight circles below the inlet, where they sampled and picked up small objects. When so activated they soon found their way to the vicinity of the inlet from any part of the aquarium.

Their ability to find food was tested by introducing small pieces of Area muscle into the tank, depositing them carefully on the bottom with the least possible disturbance some distance from the fish. Eight such tests were carried out on two blinded fish. In each case the latter became active after a short interval and started to search for the food. The time interval between introducing the muscle and the start of active swimming varied from 7 to 62 sec. Once activated the fish swam along the bottom of the tank and located the general vicinity of the food within a few seconds. They then found it by a series of rapid changes of direction, turning back whenever their course took them further away until eventually they touched the morsel with their snout, when they at once seized and swallowed it.

Blinded tomtates were able also to locate pieces of food suspended in midwater by cotton thread, but usually took longer to do so owing to their tendency to remain in close contact with the bottom.

Silversides gave alarm reactions in response to extracts of fresh skin of one of their own species. A fish was taken from another aquarium, killed and a few milligrams of the skin and scales removed by scraping with a scalpel. The scrapings were shaken with a small quantity of sea water for a few minutes, filtered and made up to 100 ml., which was then injected in the usual way into an aquarium containing a shoal of nine fish.

No striking change in behaviour was seen when the experiment was carried out during the day, when the fish were already in a tight shoal. A sharp increase in the frequency of turning was first noticed 90 sec. after the injection was started. The shoal remained very tight close to the bottom and moved to a position near the outlet end of the aquarium, where it remained more or less stationary for the next 15 min.

A more dramatic result was observed when the experiment was performed at night, the fish being then more active and the shoal dispersed. Two fish first became ‘alarmed’ 33 sec. after the injection was started and their alarm was communicated to the rest within a few seconds. The shoal reformed and kept in a tight formation close to the bottom for the next 8 min., remaining for most of this time in the middle section of the aquarium. Redispersal began when one or two fish started to swim away from the shoal for a few seconds, though sometimes the rest followed. After about 20 min. the shoal dispersed once more.

Similar experiments with tomtates gave inconclusive results. Sudden changes in the behaviour of the shoal were sometimes observed shortly after injection of the skin extract but there was no unequivocal alarm reaction.

It is generally accepted that shoaling fish use their eyes to keep station during the daytime and that blinded fish cannot keep their place in a shoal with unblinded ones, though they may attempt to do so. These facts have tended to divert attention from the part played by other sense organs in the life of pelagic species inhabiting the illuminated waters near the surface of the oceans. It seems to be widely assumed that since these species have for the most part well-developed eyes they search for and capture their food by sight. Consideration of the optical limitations of the fish eye, and the medium in which it operates, suggests that their vision is less satisfactory for finding food than is commonly supposed. The teleost eye is markedly myopic, according to Walls (1942), by as much as 15 diopters, and when in the rest position the lens is accommodated for near vision. It is adapted for seeing objects most clearly at distances of the order of 10 cm. The distance at which any object of a certain size can be seen depends, other things being equal, on the resolving power or visual acuity of the eye. This is best in good illumination and is diminished, though often with a gain in absolute sensitivity, in poor light. There are no reliable data on the visual acuity of fish, but if we assume a value similar to those found for man—it is unlikely to be better—an object 0 · 09 mm. long should be visible at 30 cm. and one 0 · 3 mm. at a distance of 1 m. in clear water and good illumination. Bainbridge (1952) found that with practice he could see single Calanus finmarchicus, an organism 3 or 4 mm. in length, at a distance of 120–150 cm. under optimum conditions and smaller but opaque species at about 30 cm. It seems unlikely that fish can do much better than this.

Observations made in the course of the experiments described in this paper support the view that fish feeding on plankton use their eyes to detect small organisms in their immediate vicinity only. Silversides were seen frequently to make direct attacks upon small copepods from distances up to about 40 cm., but it could not be shown that they saw their prey at greater distances. On the contrary, they often failed for a considerable period to notice plankton introduced near one fend of the aquarium while they were swimming at the far end between 100 and 150 cm. from it.

There is reason, therefore, to doubt whether even the well-developed eyes possessed by most pelagic shoaling fish are of much use for detecting food organisms at distances greater than a metre or so. We have now to consider the evidence in favour of long-distance perception by means of other senses. Tactile and acoustic mechanisms have been relatively little studied, but there is now a considerable body of information on the ability of fish to detect and respond by a change in behaviour to substances in solution. Many fish are known to be extremely sensitive to olfactory and gustatory stimuli, to traces of organic substances and dissolved gases as well as to small differences in the temperature, pH and salinity of water masses. Hasler & Wisby (1951) found that fish may learn to discriminate between and remember odours associated with a particular river, and there is good evidence that certain organic substances may be detected in extremely dilute concentrations. Schutz (1956), for instance, estimated that the ‘alarm’ substance liberated from the skin of wounded minnows could be detected by other minnows at a dilution of 2 × 10 ^{− 11}, and many substances have been shown to bring about changes of behaviour at concentrations in the range 1 × 10 ^{− 8} to 1 × 10^{− 10}. As mentioned above, the substance in Area muscle that excited Hepsitia and Bathystoma was effective in concentrations of this order at the point of delivery to the aquaria.

There is reason to believe also that olfactory communication plays some part in maintaining the coherence of shoals, since Keenleyside (1955) found that blinded rudd (Scardinius erythrophthalmus L.) could detect and preferred water that had contained other rudd to water which had not, but showed no such preference after their olfactory epithelium was destroyed. A recent summary of work in this field has been given by Hasler (1957).

Almost all previous studies of the olfactory sense have been made on freshwater or on non-shoaling marine species, but Tester and his colleagues (1955) found in the flesh and body fluids of tuna and other fish a substance that stimulated feeding behaviour in the predatory pelagic species Euthynnus ajfinis and Neothunnus macropterus. They were able to show that the exciting substance was water-soluble and dialysable, at least in part, that it contained phosphorus, the amide link and the benzene ring but not sulphur. They were not able to stimulate feeding behaviour with any of a large number of pure chemicals, though the fish appeared to sense the presence of some of them. Their findings, which were unknown to me at the time, are in general similar to mine with Hepsitia and Bathystoma and suggest that these pelagic species may hunt their food by following olfactory clues. All we can claim to have demonstrated, however, is that these species can detect very small amounts of certain water-soluble non-protein substances present in a variety of animal tissues and may respond to them with feeding behaviour. These or similar substances are liberated into the water by live zooplankton.

The reaction of Hepsitia and Bathystoma to an exciting olfactory stimulus has both a kinetic and an orienting component. The former is manifested by an increased rate of swimming as soon as the presence of a substance is detected, the latter by movement towards the source of the stimulus and repeated return thereto The orienting mechanism was demonstrated most clearly by the manner in which blinded Bathystoma located pieces of food introduced into the aquarium some distance from them. The initial swimming response sometimes took the fish towards the food and sometimes further away from it. In the latter case it always turned after travelling a few centimetres and swam in the right general direction. Further turns were made whenever the position of the food was overshot by a few centimetres and the morsel was eventually located by a series of such turns of progressively increasing accuracy.

These experiments have some bearing on the movements of fish shoals in the sea. There is now ample evidence, recently summarized by Hardy (1956) and Lucas (1956), that shoals of plankton-feeding fish aggregate around areas where their principal food is abundant and tend to be absent from areas where food is scarce. The association between North Sea herring (Clupea harengus) and Cal anus fin-marchicus has been more intensively studied than any other, but there is evidence of a similar relation between mackerel (Scomber scomber) and Calanus in the western English Channel and Celtic Sea. It has been shown too that whalebone whales can somehow find areas of high density of the zooplankton on which they feed. It seems reasonable to suppose that shoals can be guided to their food from long distances by olfactory perception and this may well be the most important single factor determining their movements, compared with which differences in temperature, salinity and other characteristics that distinguish different water masses may be of minor significance. We may, in general, expect fish to swim up concentration gradients of substances excreted by their food organisms, and shoals of the same species may find each other by this means. We may also expect them to swim away from alarming or noxious olfactory stimuli and thus avoid unsuitable areas such as those of dense phytoplankton production, though there appears to be still some doubt whether in fact they do so.

In addition to horizontal migrations some pelagic species make regular diurnal vertical movements similar to those of the zooplankton on which they feed. For instance, Balls (1951) and Richardson (1952) have shown by echo-sounding that shoals of herring (Clupea harengus) and sprat (C. sprattus) rise towards the sea surface at night and descend again to deeper water at daybreak. Richardson believes this behaviour to be a direct response to the change of light intensity, since the vertical movements were observed in non-feeding as well as in feeding fish, but there is no reason to conclude that this behaviour pattern is innate in fish as it seems to be in the case of the invertebrate organisms of the plankton. It seems equally likely that the fish learn to associate the regular diurnal changes in light intensity with vertical movements of their food and, having become so conditioned, continue to respond to the light stimulus even when not feeding. Another possibility is that non-feeding as well as feeding fish may continue to follow the concentration gradient of substances liberated by plankton in their diurnal vertical migrations.

These speculations have now led some way beyond the facts ; we clearly need to know more about the nature of all kinds of olfactory stimuli that are of significance to fish, and the way in which associations between different sources of information are formed in the course of their lives. From this type of study we may hope eventually to arrive at an understanding of the behaviour of fish shoals that will enable their behaviour in the sea to be predicted with greater certainty than is at present possible.