OBSERVATIONS ON THE PHYSIOLOGY OF THE SWIM BLADDER IN CYPRINOID BY H. M. EVANS AND G. C. C. DAMANT.

(Received 2.6th February 1928.)

(With Five Text-figures.)

THE swim bladder of fishesi s primarily a hydrostatic and with rare exceptions contains or tends to contain the exact quantity of gas which is necessary to make the specific gravity of the whole equal to that of the water in which it is swimming, so that it can rest in mid-water tending neither to rise nor sink: this normal con- dition is called neutral . Since gas is compressible and water is not, any increase of external or atmospheric by acting through the non-rigid body walls will reduce the volume of gas in the swim bladder and cause the fish to sink in the water (condition of negative buoyancy). This condition can also be produced by aspirating some of the gas from the swim bladder or by attaching a small weight to the fish. When in a state of negative buoyancy produced by any of these treatments (provided that the interference has not been excessive), the fish will compensate (i.e. restore its ) by introducing additional gas into its swim bladder. In fish with closed swim bladders () it has long been known that this additional gas is mainly and is secreted into the swim bladder by organs known as red bodies or gas glands. Such organs are absent in many fishes whose swim bladders are furnished with ducts communicating with the exterior, and experiments which have been published on the method by which such fish compen- sate are inconclusive. Working with such Cyprinoids as , Roach and Goldfish, which have a long pneumatic duct leading from the oesophagus to the posterior sac of the swim bladder, we find that when a condition of negative buoyancy is induced artificially they combat it by rising to the surface, taking air into the mouth and passing it thence to the swim bladder. If they are prevented from doing this an alternative remains, for we find that in spite of the absence of red bodies they can slowly secrete a gas rich in O2 into their swim bladders. The following experiments demonstrate these facts. When fish are observed in glass sided aquaria it is possible to judge with great exactness whether they are in neutral buoyancy and if not how far compensation has proceeded, provided that incautious movements or changes of illumination are not allowed to scare them into rapid excited swimming. Two small goldfish were placed in a corked glass bottle containing water (Fig. 1). The cork was pierced by a tube at which the experimenter sucked with his Physiology of the Swim Bladder in Cyprinoid Fishes 43 mouth till the reduction of atmospheric, pressure (by about 25 cm. Hg) was seen to cause the escape of a bubble or two of air from the swim bladder via the pneumatic duct and mouth. They were then quickly transferred to a glass aquarium (Fig. 2) where one was imprisoned under a bell-jar completely filled with water and so cut off from access to the surface while the other was left free to promenade and "take the air." The fish were much agitated and showed marked negative buoyancy, dropping "like stones" to the bottom at each temporary cessation of active swimming. After five minutes both were trying to reach the surface; the prisoner

J) 1

Fig. 1. could only knock its head against the glass dome, but the free fish came to the sur- face and took gulps of air then dropping back to the bottom where it spat out a few bubbles, repeating the performance at short intervals. Some of the air remained inside for the negative buoyancy of this fish steadily diminished and had disappeared in ii hours. Meanwhile the prisoner retained its negative buoyancy, though after 24 hours it was thought to be somewhat less heavy than at the start; after 48 hours compensation had progressed further but the negative buoyancy was still so great that the fish rested on the bottom and fell there (though less heavily than at first) after each swimming effort had ceased. Air was now blown up under the bell-jar so as to form a large bubble at the top. 44 H. M. EVANS arid G. C, C. DAMANT The prisoner immediately swam up, gulped air in the same way as the free fish had done and, in less than two hours it had fully compensated and was resting com- fortably in mid-water (neutral buoyancy). In another type of experiment negative buoyancy was produced by puncturing the swim bladder and removing a certain quantity of gas. Roach were placed in a large tank in which there was a constant flow of water. A vertical wire netting par- tition divided the tank into two equal compartments one of which was provided with a wire netting cover resting about one inch below the surface of the water. The fish confined in this half are called non-access fish because they cannot reach the surface or gulp air while the fish in the other half are called free-access fish.

Fig. 2.

The puncture was made in all the experiments at a point 2 scales above the at a distance of 7 lateral line organs posterior to the head. This corresponds with the centre of the anterior sac. They were punctured with the needle, of a hypodermic syringe and approximately 2 c.c. of gas removed from the swim bladder of each. Eight of them were then placed in the free-access and seven in the non-access com- partment of the tank. All showed marked negative buoyancy. After two hours three of the free-access fish were partially compensated and were observed to be swimming on the surface at frequent intervals with \ inch of the head exposed; they gulped air and spat out froth. After seven hours all the free-access fish were poised freely in the tank and swam without effort in a condition of neutral buoyancy while the non-access fish were observed to be swimming up to the wire netting and then coming down tail firsts still showing marked negative buoyancy and resting hori- zontally on the bottom or at an acute angle with their tails touching it. After 24 hours three non-access and three free-access fish were killed and the Physiology of the Swim Bladder in Cyprinoid Fishes 45 gas from their swim bladders collected over water and carefully measured when it was found that each of the free-access fish contained more gas per gramme of body weight than did any of the non-access fish, the average difference being between 3 and 4 cc. in the case of 100 gm. fish. The subsequent history of the remaining fish was as follows. After 29 hours the four non-access fish were still in negative buoyancy and trying to reach the surface; at one moment all four were swimming upwards in an almost vertical attitude. The fish with free access naturally remained in neutral buoyancy. After 48 hours two of the four non-access fish appeared to be nearly, if not quite, compensated; all four could move more freely and no longer swam obliquely with the head pointing upwards. After 72 hours two were in neutral buoyancy but two still tended to rest on the bottom. After 112 hours all four had neutral buoyancy. Summing up, we see that of four fish without access to the surface two required between two and three days to compensate and two required more than four days, while all eight of those with access to the surface compensated within seven hours. Further experiments showed that the period of compensation in non-access fish is variable and may be much longer; we have cases in which neutral buoyancy was not attained till the 16th, 28th and even the 48th day. The following experiment was devised to study more closely the feeble and variable power of compensation which the foregoing had shown us was possessed by Cyprinoids prevented from gulping air. In it negative buoyancy was produced by a more natural method than aspiration and means were provided for measuring the progress of compensation with some accuracy. Three goldfish were confined in a large aspirating bottle supplied at its base with running water from a tap. After passing through the bottle the water escaped through a discharge pipe which could be raised or lowered so as to vary the hydrostatic pressure within the bottle between nine feet of water as a maximum and six inches which was the depth of the con- tainer. With no access to air {i.e. when the aspirating bottle was completely filled with water (Fig. 3, A) the fish under of a few feet of water showed negative buoyancy and uneasiness with the usual signs of wanting to gulp air. After half an hour the discharge tube was lowered and the pressure reduced to normal, whereupon two of the fish started to float upwards with positive buoyancy showing that they had already partly compensated towards the increased pressure. By similar experiments it is found that compensation goes on slowly and in about six hours fish can compensate to five feet of water pressure, but in two experiments where the fish were left under a head of eight feet of water for 24 and 48 hours respectively they remained in negative buoyancy, though by lowering the pressure it was found that they had compensated to five feet but apparently could carry the process no further in the time. (It will be understood that the fish are kept under the high pressure continuously except for the intervals of about five minutes' duration when, by lowering the discharge tube till the fishshowe d signs of positive buoyancy, the degree of compensation was gauged.) On varying the experiment so that a large bubble of air was trapped over the running water in the aspirating bottle (Fig. 3, j?) 46 H. M. EVANS and G. C. C. DAMANT the fish were seen to make use of it. Almost directly the pressure was raised they came to the surface, gulped air in the usual way and compensated to eight feet pressure in the space of five minutes or so.

Discharge which can be raised or lowered

9 Feet of rubber 9feet of rubber tube tube t

Fig. 3-

It might be suggested that fish which compensate by swallowing air store it in the stomach or intestines and not in the swim bladder, but careful examinations made to decide this point have always revealed a total absence of gas in the digestive tract. For instance, a Roach weighing one pound was punctured in the anterior sac of the swim bladder and between 3 and 4 c.c. of gas aspirated. The fish was Physiology of the Swim Bladder in Cyprinoid Fishes 47 observed to regain neutral buoyancy by gulping air in the course of three quarters of an hour. It was then killed and dissected under water. On opening the ab- dominal cavity two or three tiny bubbles of gas escaped from the peritoneal cavity; no doubt this gas had escaped there from the punctured swim bladder. The in- testines and stomach were then removed without interfering with the pneumatic duct. No gas escaped as they were removed and no gas was apparent as they were divided with scissors throughout their length.

COMPOSITION OF THE SWIM BLADDER GAS IN AND OTHER . We have analysed the gases from-normal Roach and Bream; the body of the fish was completely immersed under water while the sample was collected through an aspirating needle over mercury. The analysis was done in a Haldane portable apparatus. About half the samples were examined for combustible gases but as no trace of such was ever found the residual gas is assumed to be in all cases.

Table I. Analyses of gases from the swim bladders of Cyprinoid fish.

V^g V /O/ CO2(%) /io-5 4-2

8-3 2'5 6-8 4*4 Roach , io-8 4*3 3'3 2"3 3'9 2"O 3"O 2*3 \ 3*5 2" I Bream

Average 6-o- 2*8 N2 (by diff.) 91*2%. The Bream was caught on a line and its gases collected in a floating laboratory within a few minutes of capture. The gases from Roach were removed a few days or hours after capture, the fish having been meanwhile confined in a large tank with constant supply of fresh water. F. G. Hall (1924) to whom we are indebted for the most recent work on swim bladder gases, gives the results of analyses of 12 samples from (American) Carp. The average was O2 5*7 per cent. CO2 3*7 per cent., which is strikingly similar to our own determination. His series included three samples with the O2 between 7 and 8 per cent., one over 12 per cent, and the remainder below 5 per cent, which corresponds with the range of variation found by us. In the American Perch, which has a closed swim bladder and cannot swallow air, he found the gas to contain

O2 19-9 per cent. CO2 0*63 per cent, (average of 28 samples). H. M. EVANS and G. C, C. DAMANT He then proceeded to investigate the effect of removal of gas by taking a second sample from each fish some hours after the first and found that in Perch the oxygen percentage had risen to nearly double its original volume while in Carp there was little change. He remarks "The Carp has an open duct leading from the swim bladder while the Perch has a closed swim bladder. This is believed to be the ex- planation of the difference in response of the two types to the withdrawal of air from the swim bladder" and with this comment leaves the matter. With Roach, allowed free access to the surface after being aspirated, we get results closely resembling those of Hall with Carp, but if the fish are prevented from gulping air the difference is striking. Approximately equal quantities of gas were removed from the swim bladders of seven Roach by aspiration with a hypodermic syringe. The fish were then placed in the experimental tank described on p. 44, four of them being cut off from the surface by wire netting and the other three, as controls, being able to reach it and gulp air. When the process of compensation was seen to be complete, which in the case of the non-access fish was after some days, they were killed and the gases drawn off and analysed. Table II. Comparison of gases found in the swim bladders of Roach after compensating with and without free access to the surface.

Non-access fish O2% co2 23-8 2-3 24-8 3-2 26-8 3-8 27-5 4-4 Average 25-7 3-4 Free-access fish 7-8 46 7-6 2-3 2-S Average 8-9 3-1 These results show that Roach with negative buoyancy requiring gas to inflate their swim bladders are able, if prevented from reaching the surface and gulping it, to secrete oxygen to about the same extent as do Perch which are Physoclisti with prominent red glands. In Roach no gland or modified epithelium has been de- scribed on the inner coat of the swim bladder. The process is however very slow as compared with fish possessing gas glands, for instance, Woodland (1911) found that Pollack of about the same weight as our Roach could secrete 5 c.c. of gas in less than 12 hours. Physiology of the Swim Bladder in Cyprinoid Fishes 49 THE PRESSURE OF GASES IN THE SWIM BLADDER OF CYPRINIDAE. If a Cyprinoid such as a Minnow or Roach which has been kept for weeks or perhaps its whole life in water no more than six inches deep be killed and opened, it will be noticed that the swim bladder is tense and feels quite hard, because the gases contained in it are at a pressure considerably above that of the atmosphere plus the six inches hydrostatic pressure which is the maximum external pressure of the fish's recent environment. That such a pressure exists during life becomes evident when gas is drawn off from the swim bladder as in the experiments described on p. 44. In none of the fish was it necessary to suck with the syringe, for as soon as its needle entered the bladder the piston was driven to the top by the pressure of the contained gases. We have found no trace of this phenomenon in non-cyprinoid fish such as Perch whose (closed) swim bladders in the same circumstances are quite soft to the touch; the contained gases being at atmospheric pressure. The excess pressures in the swim bladders of Roach have been measured by introducing a hollow needle connected with rubber tubing (filled with liquid paraffin) to a "Tyco" pressure gauge. The fish were rapidly killed, generally by destroying the brain, and the needle thrust into the swim bladder, either more or less blindly through the flank of the entire fish or carefully after opening the body and exposing the swim bladder. In the first case time is saved, but readings of

Table III. Gas pressures in swim bladders of Normal Roach.

Pressure in mm. of Hg Method of killing Body opened or not 1 Anterior sac Posterior sac.

Pithed Yes 70 5> >> 67 90 € >> j> 80 >> >> 42 120 80 >> 5> 80 70 >> >> 70 >> 3> 50 90 >> »> 60 40 »> >> 40 80 >> »> 70 >> 72 40 >J >> 50

J> J> 54 54 20 20 »> J> >> >> 38 38 40 M >> 44 >> No 84 j> 83" *9 58 yy 58 Etherised >> 53 64 »> H.C.N. 62 67 Yes 64 . IS Average 59 62

BJEB-Vli 50 H. M. EVANS and G. C. C. DAMANT pressure may be too low through gas escaping from the bladder into the peritoneal cavity at the side of the needle puncture. In the second case leakage is prevented by doing the operation under water and carefully inserting the needle through a thick muscular part of the bladder wall; on the other hand the effect of the necessary manipulation on the contractile bladder walls may cause some alteration of pressure. Table III gives the results of a number of experiments, the method by which the fish was killed is indicated and also whether the body was opened before measuring the pressure or not. In most cases the pressure in the two sacs into which the bladder is divided was taken separately. Generally they were much the same within the limits of possible error but in some cases there was a difference showing that com- munication between the two sacs had been cut off by closure of the sphincter as previously described by Evans. We find then that when Roach are kept in water less than a foot deep the average pressure in their swim bladders will be 60 mm. Hg or 2§ ft. of water pressure. A reduction of atmospheric pressure by about 60 mm. Hg which is equivalent to raising the bladder pressure by a further 60 mm. Hg will generally cause the escape of a bubble of air from the bladder via the pneumatic duct and mouth, so that speaking figuratively the safety valve in the average Roach is loaded to about 120 mm. Hg. It would be interesting to know whether when such fish are living at a greater depth, say four feet of water, they still maintain a bladder pressure in excess of that due to hydrostatic head, but we have no satisfactory means of de- termining the point. Undoubtedly the pressure is partly due to tone in the muscles of the bladder wall which have been described by Evans, but it is difficult to decide in what way it is advantageous to the fish. Apparently it is incompatible with the existence of such structures as the red glands of the Physoclisti, for with a bladder pressure in excess of the aortic blood pressure it is difficult to see how blood could reach them. The Roach's bladder wall is so impermeable to gases that when dissected out and left on the laboratory table in a dish with a little saline solution to keep it moist, the pressure within does not appreciably fall till putre- faction sets in after two or three days. Ability to compress the contents of the bladder as required would obviously enable a fish to counteract the effect of changes of hydrostatic pressure on its buoyancy within a limited range: thus a Roach living in a pond three feet deep might keep its swim bladder contents at a constant pressure of 67 mm. Hg (and therefore at a constant volume), maintaining this by muscular exertion when at the surface and allowing hydrostatic pressure to take over part of the work at greater depths till at the bottom the muscles could be fully relaxed. The result would be that the fish's buoyancy would be the same at all depths in that pond. We do not believe this to be the right explanation. Goldfish with fairly high bladder pressure when confined in the apparatus shown in Fig. 3 and suddenly exposed to an increase of pressure of no more than 18 inches of water show signs of negative buoyancy and discomfort which they try to remedy by swallowing air in a way that certainly does not suggest that the muscular mechanism indicated above is available. Again if one attaches small pieces of cork or lead to fish so as to Physiology of the Swim Bladder in Cyprinoid Fishes 51 upset their neutral buoyancy, they will compensate within limits by altering the volume of gas in their swim bladders, and measurements of the pressure should show if the bladder muscles were used to assist the process. The following experi- ment was done to investigate the point. A batch of Roach of average weight 100 gm. were divided into two groups of six one lot being kept as controls, while to each fish of the other a small cork float was attached. After five hours all the fish were killed and the bladder pressures determined. The pressures in the corked fish, being much below the normal, demonstrate clearly that compensation was effected by expulsion or absorption of gas and not by compression of the original quantity of gas into a smaller volume. Finally the gases were squeezed out of the bladders of all the fish and measured, when it was found that the corked fish only had 6-5 c.c. per 100 gm. of body weight and the controls 9-1 c.c. per 100 gm. of body weight. We obtained a similar decisive answer from fish to which small lead weights had been attached. In these the pressure rose much above normal. Table IV. Pressure of gases in swim bladders of " Corked" fish and Controls. " Corked " fish Controls (mm. Hg) (mm. Hg) 25 70 33 80 22 44 20 90 25 63 40 50 Average 27 66 A chance observation is worthy of record as illustrating how the buoyancy can be affected by the bladder pressure. A number of Roach that had been kept in a tank with about one foot depth of water for nearly three months became infected by fungus disease and died: soon after death some were seen to be floating on the surface while others lay on the bottom. The average pressure in seven fish found floating was 30 mm. Hg while the average pressure in three which were lying dead at the bottom of the tank was 90 mm. Hg. Apparently the fish floated or sank according to whether their bladder muscles contracted or relaxed about the time of death. Since the bladder pressure does not seem in life to be a means of regulating buoyancy we can only indicate one other way in which it might be useful. Since as far as we know the high pressures described above are confined to the Cyprinidae which are one of the group of fishes possessing the Weberian mechanism, it is natural to suppose that the two are connected, and we suggest provisionally that the proper working of the Weberian ossicles demands a certain tension in the bladder walls which is normally maintained by the reaction of the contained gases against the compressive force applied by the bladder muscles.

4-2 52 H. M. EVANS and G. C. C. DAMANT

ANATOMY OF THE PARTS CONCERNED IN INFLATION OF THE SWIM BLADDER. Evans (1925) has described and figured the muscular bands and sphincters and nerve supply of the swim bladder in Cyprinidae, including the sphincter at the exit of the pneumatic duct from the posterior air sac and the nerve plexus at its junction with the oesophagus. One of the objects of this paper is to describe the bulbous enlargement of the pneumatic duct which occurs at this point and appears to con- stitute a definite organ concerned with the inflation of the swim bladder. A similar structure occurs in the Gymnotidae and John Hunter (1861), after describing the two separate "air bags" of Gymnotus electricus, mentions that their ducts unite to form a common passage which becomes cellular and then opens into the oeso- phagus. Sections of this cellular portion of the common duct show a structure similar to that found in the bulb of the pneumatic duct in Cyprinidae. We propose to call this organ the pneumatic bulb. Guyenot (1909), describing it, writes "the pneumatic duct for 1 to 1-5 cm. presents a fleshy enlargement with longitudinal and circular striated muscle continuous with that of the oesophagus.... In the lumen of this enlargement are a number of blind diverticula separated into compartments and lined by villi like those of the oesophagus." He then proceeds to state that "the small diameter of the pneumatic duct and the abundance of mucous secretion contributes, with the muscular sphincter, to form an apparatus not only opposing the penetration of external air but also, to a certain extent, the exit of air from the swim bladder, and that the muscles of the pneumatic sphincter and particularly the striated muscles have a tonic contraction which prevents egress of air and makes the penetration of air impossible at least in Cyprinidae." He finds that "the minimum pressure required to produce exit of air is 65 cm. of water, and for the entry of air 195 cm. of water." We believe that the foregoing experiments prove that the entry of air is not only possible but is the most rapid and usual method adopted by these fish to inflate the swim bladder. The anatomical structure of the pneumatic bulb is not only consistent with this function but is presumably evolved for the special purpose of enabling swallowed air to be pumped into the swim bladder. The organ has been examined by one of us (H.M.E.) in Roach, Tench, Bream, Minnow and Carp and a similar type of structure is present in all of these. The specimen now described was taken from a Roach weighing six ounces. The pneu- matic duct in its passage forward from the posterior sac of the swim bladder be- comes gradually thicker at a point about 4-8 mm. before its junction with the wall of the oesophagus. Fig. 4 is a diagrammatic longitudinal section of these parts and will enable the reader to follow the structure of the duct as it passes through the wall of the oesophagus. Starting at the narrow entrance of the duct into the oesophagus, it is seen to dilate and form a considerable cavity occupying three quarters of the thickness of lie oesophageal wall. Next follows a narrower portion into which several diverticula pen: these diverticula run parallel to the long axis of the duct. The total length of Physiology of the Swim Bladder in Cyprinoid Fishes 53 the duct lying within the wall is 47 mm. so that the total length of the pneumatic bulb is approximately 1 cm. About 0-3 mm. before joining the gullet and for a distance of about 07 mm. within the oesophageal wall, the duct breaks up into a spongy portion the total length of which is therefore about i-o mm. The opening into the gullet is surrounded by a sphincter muscle and the duct throughout its course is surrounded by a thick investment of striated muscle con- tinuous with the striated muscle of the oesophageal wall. The mucous membrane of the oesophagus is continued throughout the anterior portion of the duct for about 4-0mm. and presents numerous "taste buds" surrounding the orifice, but the last 0-7 to O'S mm. of the duct within the oesophageal wall presents an entirely different structure as described above, it has here a spongy structure and is now lined with columnar epithelium. T IS

LUMEN OF (ESOPHAGUS

Fig. 4. Longitudinal section of wall of gullet through oesophageal bulb (diagrammatic). Fig. 5 is a detailed drawing under a quarter-inch objective of the pneumatic bulb of the Tench before its junction with the oesophagus. The lumen of the main duct is shown and one large diverticulum surrounded by circular muscle fibres, external to which are longitudinal muscle bundles. Apparently in the pneumatic bulb we have an air pump with its entrance guarded by "taste buds" and a sphincter,and its main cavity lined with mucous epithelium and provided with powerful striated muscular walls. Between this part and the duct proper is interposed a spongy labyrinth which evidently might act as a filter. It is interesting to compare the situation of the taste buds in the above fishes with that in mammals. In man they are found at the base of the tongue on the sides of the papillae vallatae and also on the fungiform papillae. According to Schafer (1894) tney are specially numerous over a small area just in front of the anterior pillar of the fauces, also on the anterior surface of the soft palate and the posterior surface of the epiglottis. In the process of swallowing the post-nasal space and the entrance into the 54 H. M. EVANS and G. C. C. DAMANT larynx have to be cut off so that food can pass only into the oesophagus. The taste buds appear in just those positions where one would expect organs to be posted as sentinels to protect the air cavities. In Cyprinoid fish we find taste buds at the orifice of the pneumatic bulb, and conclude that they are there as part of the mechanism for preventing food from entering it.

We desire to express our gratitude to Prof. A. E. Boycott for allowing us to work in his laboratory at University College Hospital and for much valuable help and many suggestions.

Fig. 5. Pneumatic bulb of Tench before its junction with oesophagus, showing duct and one dir- verticulum and circular and longitudinal muscle bundles. J in. obj.

SUMMARY. 1. Experimental proof is given that Cyprinoid fish can inflate their swim bladders either by swallowing air or, far more slowly, by secreting a gas rich in oxygen, although they have no gas glands in the ordinary sense of the term, 2. Attention is directed to the high gas pressure which such fish maintain in their bladder and its probable use is discussed. 3. The of the pneumatic bulb at the oesophageal end of the pneumatic duct in Cyprinoid fishes is described. Physiology of the Swin Bladder in Cyprinoid Fishes 55

REFERENCES.

EVANS, H. M. (1925). Proc. Roy. Soc. B, 97, 545-576. GUYENOT, E\ (1909). Bull. Sci. Nat. de France et Belgique, 43, 203—296. HUNTER, J. (1861). Essays and Observations, ed. by R. OWEN, p. 419. HALL, F. G. (1924). Biol. Bull. Woods Hole Mass. XLVII, pp. 79-115. SCHAFER, E. A. (1894). Quain's Anatomy, vol. in, pt. 3. Tenth ed. p. 149. WOODLAND, W. N. F. (191 I). Anatamischer Anzeiger, 40, 225-242.