The Reactions of Fish to Water of Low Oxygen Concentration

In polluted rivers fish may encounter a great variety of toxic substances, and with the development of new industries the fist of chemicals occurring as pollutants is a lengthening one. But one of the chief hazards is a lack of dissolved oxygen brought about by the decomposition of sewage or organic effluents from slaughter-houses, dairies, tanneries and paper mills. Freezing may produce a similar condition (Thompson, 1925) by cutting the water off from contact with the atmosphere. The literature on the effects of oxygen deficiency on fishes is very extensive. Gardner (1926) has a long bibliography of the older literature, and more recent work is reviewed by van Dam (1938) and Black (1951). In general previous research has consisted mainly of studies of the degree to which the oxygen consumption of fish is affected by the oxygen concentration of the water and the temperature, the comparative tolerance of different species, the effects of low oxygen concentrations on the rate and amplitude of the respiratory movements and the way the animal’s power to extract oxygen from the medium is affected by the concentration of carbon dioxide. Doudoroff & Katz (1950) have recently reviewed this last aspect of the problem, and important relevant papers are those of Hart (1945), Black (1942), Fry (1947) and Fry, Black & Black (1947). There appears to have been comparatively little study of the degree to which fish can detect and avoid water of abnormally low oxygen concentration. Shelford & Allee (1913) applied their ‘gradient tank’ method of experiment to the problem and studied the behaviour of about sixteen American species of fish in gradients of oxygen, carbon dioxide, nitrogen and ammonia. They concluded that there appears to be little or no response to a nitrogen gradient, and that the concentration of this gas is of no importance provided that it is not present in such excess as to cause ‘gas-bubble’ disease; that fish react negatively to carbon dioxide and to a deficiency of oxygen, indicating their dislike by moving away, rising to the surface, gulping and ‘coughing’; that fish do not react negatively to ammonia and sometimes enter concentrations which cause death. Apart from indicating that fish show a general tendency to avoid water of low oxygen concentration, the experiments of Shelford & Allee do not throw much light on the problem; they do not, for instance, indicate how far the reaction is decided by the actual concentrations of oxygen encountered or by the temperature of the water, and some of their results are rather vague; thus in one series of experiments (Shelford & Allee, 1913, p. 236, Table 9), in which eighteen fish were used, the fish spent 56% of their time in the high-oxygen part of the tank and 44% of their time in the low-oxygen part, not a difference of great significance.

The present paper records the results of about 100 experiments to test the reactions of small minnows, sticklebacks and trout to water of low oxygen concentration, using the apparatus described in earlier papers by the writer (1947, 1948). In this no gradient is produced, but two contrasted supplies of water, in and out of which the fish can swim freely, are very sharply differentiated. The general method of experiment is fully described in these papers; details relevant to the present investigation follow in the next section.

The apparatus used is essentially similar to that described by the writer (1948) in a paper dealing with the reactions of fish to sulphide solutions, lead nitrate and ammonia. The same arrangement has recently been used by Hodgson (1951) to study the reactions of aquatic beetles to salts and alcohols. Modifications of the original apparatus to adapt it to present requirements are shown in Fig. 1. Water of low oxygen concentration is generally prepared by boiling normal water, by subjecting it to low pressures or by passing through it a stream of nitrogen; elaborate apparatus for the preparation of large quantities of oxygen-free water has been described by Shelford (1918), and Fry (1951) has devised a fractionating apparatus to provide water of various dissolved oxygen tensions. With the apparatus used by the writer large volumes are not required, and boiling, aided by the filter pump, was found quite satisfactory. The filter pump is preferable to a vacuum pump of the ‘Speedivac’ oil-immersed type, as it is unaffected by the large quantities of water vapour drawn off. A (Fig. 1) is the reaction tube which measures internally 3-4 × 50 cm.; aspirator B supplies water fully saturated or slightly supersaturated with air. Aspirator C, of thick Pyrex glass, is detached from the apparatus and its outlet closed with a screw-clip. About 9 1. of warm or hot water is poured in, the bung is replaced, tap D is closed and E is connected to the filter pump. The pump is set working and the aspirator shaken vigorously to give a swirling motion to the water; if this is not done it boils at the top only and very little oxygen is drawn off.

The temperature of the water placed in C and the time the pump is operated depend on the oxygen concentration required. To produce water of very low oxygen tension it is necessary to bring it to boiling-point before pouring it into the aspirator and then boil by means of the filter pump for 15 min. or more. Boiling for a short period removes far less of the dissolved oxygen than is commonly supposed. If water 50 % saturated is required it need not be warmed at all and a few minutes’ evacuation and shaking will suffice. After a certain amount of practice the temperature and evacuation time necessary to produce the required depression of the oxygen concentration can be estimated with fair accuracy. When the pump has been running for a sufficient time the tap E is shut and the water allowed to cool to the correct temperature with the aspirator still exhausted of air. Tube F permits determination of the temperature. When this is correct and equal to that of B pure liquid paraffin is admitted via the funnel and stopcock D in sufficient quantity to cover the surface of the water; air is then allowed to enter and a Winkler determination done on a sample taken by attaching a delivery tube to the rubber connexion G and very rapidly running about 200 ml. through a 50 ml. sampling bottle. When dealing with water of very low oxygen concentration a small sampling bottle of the type described by Powers (1918) was found useful.

The use of liquid paraffin is not always necessary. When the oxygen content of the water in the supply aspirator is not below 40 % saturation and when not more than about two-thirds of the supply is discharged during the experiment the paraffin can be omitted without serious error; this is an advantage, for cleaning the aspirator after the use of paraffin is a tedious performance. In this case, of course, the aspirator is kept exhausted of air until the experiment is begun.

When both water supplies are ready a fish is placed in the reaction tube. This is filled with water from B, and a little water is run in from C also to clear out all bubbles which are finally expelled via H. No bubble traps can be used for obvious reasons. The fully oxygenated supply is set running through both halves of the reaction tube ; when the fish has become accustomed to its surroundings its movements are recorded for 2-3 min. and then the flow in one half of the tube is replaced by water from aspirator C. If the fish performs exploratory movements and visits both ends of the reaction tube at the commencement of the experiment it is preferable to start running the poorly oxygenated water through the half of the tube from which the fish is absent ; if the fish is sluggish and remains in one half of the tube it is obviously necessary to admit the poorly oxygenated water on this side. When a reversal of flow is carried out normal water is passed through both halves of the tube for at least a minute before the supply from aspirator C is set running in the new direction. The Aberystwyth tap water is a very soft and pure water of pH normally about 6-8 and does not contain any appreciable quantity of carbon dioxide. The method used for lowering the oxygen content of the water results in the removal of some of the dissolved nitrogen. In these experiments it is assumed that fishes do not react to differences in nitrogen concentration ; ideally, of course, the nitrogen should be restored so that the only difference between the two supplies of water presented to the fish is the difference in oxygen concentration.

The first experiments were carried out with sticklebacks (Gasterosteus aculeatus L.) at 13° C. Four typical records are given in Fig. 2, each for a single 28-32 mm. fish. The reaction to oxygen concentrations below 3 mg./l.—at this temperature about 28 % saturation—conforms to a fairly definite pattern. The fish has no immediate appreciation of the low oxygen concentration and generally swims across the boundary zone without any hesitation; occasionally it will cough, i.e. make a sudden, very jerky respiratory movement, but swim on. In a series of fifteen experiments at 13° C. with oxygen concentrations up to 3 mg./l. the fish crossed from the fully oxygenated water into the poorly oxygenated supply without hesitation about 92 times. The record for o-8 mg. in Fig. 2, for instance, shows five such traverses between 8 and 10 min. If the fish remains in the low-oxygen zone it develops a dyspnoea, the amplitude and frequency of the respiratory movements increasing; at the same time random movement appears to be incited and the fish may struggle violently. Sooner or later this random swimming takes the animal across the boundary into the well-oxygenated water, when the stimulus to swimming disappears. Usually this happens with the greatest abruptness, so that the fish only swims about 10 cm. into the well-oxygenated water and then becomes almost motionless as its respiratory movements rapidly return to normal. Thus it appears to select the high-oxygen zone, but the basis of this selection is simply the sudden removal of the stimulus to swimming. If the poorly oxygenated supply is run through both halves of the reaction tube a stickleback will keep in continual motion up and down the tube, becoming more and more agitated ; when well-aerated water is admitted the fish stops moving immediately on coming in contact with this and generally remains still until it has recovered.

Other workers have noted that a lack of oxygen stimulates swimming in fish; see, for instance, the review by Black (1951, pp. 98-9). The reaction has not been explained; it is known that the mackerel does not perform any breathing movements but keeps up a current of water through the gill chambers by swimming along with its mouth open (Scheer, 1948, p. 399), and movement through the water probably aids respiration in most fishes. Thus more rapid movement may be a natural response to oxygen deficiency.

The records in Fig. 2 are largely self-explanatory, but a description of the 0.26 mg. experiment will help to make things clear. The low-oxygen water was first admitted at 30 min. immediately after the fish had moved over to the left. Between 4 and 5 min. the right half of the tube was visited three times; the first two visits produced no great discomfort, but on the third occasion a violent dyspnoea developed with furious swimming. At 5 min. the stickleback got across the boundary and recovered. The trial at 8 min. produced no definite result, as the fish crossed the tube with a natural movement before any respiratory distress developed. At iz| min, the 0-26 mg. water was again admitted on the right; the fish remained quite unperturbed for about 50 sec., struggled for about 40 sec., reached the well-oxygenated water, stopped and recovered, its opercular movement rate returning to normal in a little over a minute.

When the oxygen concentration tested is 3-5 mg./l. the results are more variable. Respiratory distress, agitated swimming, apparent recognition and selection of the well-aerated water, rest and recovery generally form the behaviour pattern, the dyspnoea taking a little longer to develop. Occasionally, however, the fish does not leave the poorly oxygenated water, it does not struggle wry much, and by increasing the amplitude and frequency of its breathing movements seems to succeed in obtaining enough oxygen for its needs. The fourth record in Fig. 2 is for water containing 6·6 mg./l. ; in this case the stickleback seemed to have a preference for the left half of the tube, and though it made several visits to the fully aerated water it persisted in returning. In this experiment it proved possible to make some counts of the respiratory movements. These had a frequency of 96 per min. when the experiment began and had speeded up to 164 per min. when the experiment was ended at 18 min. It was then decided to make some similar observations at low oxygen concentrations; the majority of these attempts were failures, but eventually three experiments succeeded, the movements of the fish being copied together with a fair number of reasonably accurate counts of the opercular movement rate. These three records are set out in Fig. 3 and help to confirm the general interpretation of the reaction given above. In the third record in Fig. 3 the fish revisited the low-oxygen zone at 8-9 min. before it had recovered completely.

All the successful experiments carried out at 13° C. with Gasterosteus are summarized in Fig. 4. Here the horizontal scale gives the oxygen concentration of the ‘low’ zone; in every case fully oxygenated water is presented as the alternative. The ordinates (‘reaction times’) are the times the fish took to leave the poorly oxygenated water. In some cases they swam into this of their own volition, in others it was admitted into the half of the reaction tube in which they had taken up a position. The reaction times are very approximate, and in some cases it was obvious that the stickleback took an inordinate time to find its way to the boundary zone. One conclusion can be drawn from Fig. 4: it is evident that as the oxygen concentration tested is decreased there is no dramatic decrease in the time taken by the fish to ‘select’ the well-aerated water. It should be noted that the concentration range 4·0-0·26 mg./l. represents in terms of percentage saturation a drop from 38 to 2·5%.

At low temperatures the stickleback exhibits a generally similar reaction to water of low oxygen concentration, but the whole performance is so slowed down that it can be easily watched. Several experiments were carried out at 3° C. with fish that were gradually cooled down to this temperature from 13° C. over a period of 9-10 hr. Three records are given in Fig. 5, and a description of the first experiment, with water containing 0·3 mg./L, will be sufficient. Before the low-oxygen water is admitted the fish rests motionless on the bottom of the reaction tube, breathes slowly, regularly and almost imperceptibly and appears to be asleep. On admitting the 0-3 mg. water it does not move but soon the respiration rate begins to rise. At 5 min. it is 87 per min. and the opercular movements are appreciably deeper; at 6 min. they have become heaving and of great amplitude; suddenly the fish begins to swim, it crosses the boundary zone, almost immediately rests again upon the bottom of the reaction tube and once more appears to be asleep, its respiration returning to the previous frequency and amplitude. Reversal of the flows at 9-10 min. repeats the performance. In the second record in Fig. 5 it will be noted that over 9 min. go by before the boundary zone is crossed, and in the 3·5 mg. experiment the fish was motionless with no sign of respiratory distress at the end of the experiment.

At 20° C., on the other hand, when a stickleback comes into contact with water of very low oxygen concentration, respiratory difficulties seem to develop almost immediately; so rapidly that the entry of the fish is checked and it moves away, remaining in the well-aerated water (see the record for 1·0mg./l. O₂ in Fig. 6). Thus on encountering the poorly oxygenated water the fish behaves as if it were violently irritated and retreats immediately, just as minnows appear to be violently irritated by and at once avoid a 0·03% solution of paracresol (Jones, 1951, pp. 265-6). At higher oxygen concentrations (3·5 mg., Fig. 6) the reaction is a little slower; if the fish swims at its normal speed the boundary zone is crossed, dyspnoea develops in 5-10 sec., and after some erratic movement the fish retreats. At 3·5 mg. the reaction appears to be a much speeded-up version of that seen at lower temperatures ; whether the animal’s behaviour at 20° C. and very low oxygen concentrations is a further speeding-up of the same sequence of events is not clear. It is impossible to avoid the conclusion, however, that the essential feature of the reaction is the rate of development of dyspnoea ; it may be that at any particular low oxygen concentration there is a critical temperature at which respiratory distress is developed so quickly that a fish swimming at the normal rate is turned back at once. The reaction fails occasionally when the fish crosses the boundary zone with a wild rush (see Fig. 6, 2·0 mg., 5·6 min.).

It will be noted in the 1·0 mg. experiment recorded in Fig. 6 that the responses at 4-6, 8-9 and 12-14 min. are much more prompt than those displayed on the admission of the poorly oxygenated water at 3, 7 and 10½ min., but when it is remembered that the poorly oxygenated water takes some 30 sec. to replace the previous supply the explanation for this is obvious.

With minnows twenty-four experiments were carried out at temperatures from 3 to 24° C. and oxygen concentrations from 0·27 to 60 mg./l.; four representative results are given in Fig. 7. The behaviour of the fish is essentially similar to that observed in the case of Gasterosteos but is less dependent on the temperature, so that the reaction at low temperatures is less delayed. In the 3° C. 1·3 mg. experiment the first reaction was rather indefinite, but after 7 min. the fish responded comparatively promptly; it would swim across the boundary zone, develop gulping respiratory movements of great amplitude in 20 sec. or less and return to the well-oxygenated water. At the same temperature 3 · 0 mg./l. of oxygen produced a fairly good response in about 4 min., but water containing 4 · 6 mg./l. produced neither dyspnoea nor movement in 15 min. In the 2 · 8 mg. 12 · 5 ° C. experiment the fish swam about fairly quickly, would penetrate some distance into the low-oxygen zone, hesitate 3-5 sec., begin gulping wildly and retreat. At 0 · 53 mg. and 18 ° C. the minnow was very active, would rush into the poorly oxygenated water, become very distressed almost immediately and rush out; at 13 min. the fish swam deep into the right half of the tube and began violent struggles which did not get it to the boundary zone until about 25 sec. had elapsed, by which time it had lost its sense of balance. Swimming on one side it reached the well-oxygenated water and recovered in about 2 min.

The 24 ° C. experiment was one carried out rather later in the spring (June) when the minnows had become more acclimatized to high temperatures. In this trial the fish did not swim so wildly and generally would not penetrate more than 2-3 cm. into the low-oxygen zone, developing respiratory difficulties almost immediately. At this temperature lack of oxygen very quickly produces in minnows an extremely high opercular movement rate exceeding 270 per min., and an avoiding reaction is promptly developed to water containing up to 4 · 0 mg./l.

The minnows seemed to be much more adept than the sticklebacks at finding their way about in the reaction tube, and when stimulated to swim by respiratory difficulties would very often move at once in the right direction. This was particularly evident in the case of the fish used in the 1 · 3 mg. experiment recorded in Fig. 7. Not all the fish behaved in this way; some, when stimulated by oxygen lack, began what appeared to be a random movement.

A number of trials were made with 30-32 mm. trout fry, and four typical results are given in Fig. 8. At 3 ° C. the response takes some time to develop; at first the fish breathes very quietly and regularly about 64 times per min., and resting on the bottom of the tube appears to be asleep. When the water of low oxygen concentration is run in the respiratory movements gradually quicken and attain a surprising amplitude; the fish, still resting on the bottom, raises itself on its pectoral fins with its body arched backwards in a curious posture resembling opisthotonos. Eventually it begins to swim in an uneasy, aimless fashion, and on crossing into the well-oxygenated water immediately ‘falls asleep’ again. At 13 ° C. the response is much more prompt; if the fish swim across the boundary zone they develop a marked dyspnoea in 5-7 sec. and return to the well-oxygenated water.

Some of the experiments at 13 ° C. were not very satisfactory, for the trout, which were obtained from a hatchery, appeared to be very tame and tended to remain motionless for long periods, making few voluntary excursions to the poorly oxygenated water. At 20 ° C. the fish were more active, and the response to a concentration of 0 · 56 mg./l. was so rapid that the behaviour of the fish appeared to be an immediate recognition and rejection of the poorly oxygenated water, for on swimming into it the trout would stagger, gulp violently and retreat in less than 5 sec. A 3 · 2 mg. experiment at this temperature also resulted in a very good avoiding reaction.

In a general discussion of their results Shelford & Allee (1913, p. 261) note ‘that fishes are able to react to stimuli by simple turnings back, and that as a rule, they remain longer in water which does not clearly influence the details of their activities’. In the course of subsequent work to test the reactions of fishes to various polluting chemicals Shelford (1917) obtained results suggesting that fish react ‘positively’ to a great number of polluting chemicals, swimming into the solution as if attracted by it, remaining in it and even avoiding pure water. On the basis of these experiments the assumption appears to have been made that while fish have no instinctive, protective reaction to poisonous, unfamiliar substances, they do have an instinctive capacity for actively avoiding water of abnormally low oxygen concentration, a condition they may be accustomed to meeting in their normal environment. Thus Ellis (1937, p. 372), after an extensive survey of fresh waters in America involving 5809 oxygen determinations at 982 stations, concluded that ‘these data collected from localities where the fish had had opportunity to choose for themselves point very strongly to 5 p.p.m. as the lower limit of dissolved oxygen, if the complex is to maintain a desirable fish faunae under natural river conditions’. Eriksen & Townsend (1940, p. 84) in a study of pollution by waste sulphite liquors refer to Ellis’s work stating that it supplies evidence ‘that salmon and trout—“representatives of the fine fish group”—either select water having dissolved oxygen content of 4 p.p.m. or more, or such fish find it impossible to establish important populations in waters having less than 4 p.p.m.’ Thompson (1925, p. 429), discussing the reactions of fish to oxygen deficiencies in the Illinois river, attributes to fishes an ability to avoid concentrations that would be fatal to them. The writer’s experiments described in this paper appear to indicate that fish do not have any instinctive ability to recognize immediately and avoid water of abnormally low oxygen content, that they swim into it with little or no hesitation when the oxygen concentration and temperature do not provoke immediate and acute respiratory distress, and if they eventually avoid it the basis of the reaction would appear to be the active, random swimming and struggling that appears to be incited by dyspnoea. A full discussion of the application of the results to problems of river pollution would be beyond the scope of this paper, but it may be pointed out that in further investigations the temperature factor must be regarded as one of the greatest importance. That the oxygen demands of fishes vary greatly with the temperature is well known; van Dam (1938, p. 71), for instance, found that a rise of about 8 ° C. caused the oxygen intake of a rainbow trout to increase from 42 · 5 to 98 · 0 c.c./hr., equivalent to a Q₁₀ of about 2 · 7. More significant than the increase in oxygen demand is the increase in the rate at which dyspnoea develops ; this appears to be the essential factor deciding whether fish will enter water of a different oxygen concentration. When we consider, for example, that at 3 ° C. water containing about 0 · 5 mg. O₂/l. may be tolerated by Gasterosteus for over 5 min., whereas the same concentration at 20 ° C. brings on a severe dyspnoea in less than 5 sec., the importance of the temperature factor is clear. Under natural conditions, however, we have to take into consideration the degree to which polluted and unpolluted waters may mix to form gradients, the presence in polluted water of specific toxic substances, such as the sulphides produced by the decomposition of sewage (Longwell & Pentelow, 1935), the concentration of carbon dioxide and the degree to which a more rapid appreciation of unfavourable conditions at high temperatures is counterbalanced by greater oxygen demands.