A Further Study of the Reactions of Fish to Toxic Solutions

In a previous publication (Jones, 1947 b) the writer has reviewed the work of Shelford & Allee (1913), Wells (1913, 1915a, b) and Shelford (1917, 1918), who studied the reactions of fish to toxic solutions with the ‘gradient tank’. A modified type of apparatus was described in which the fish are confined in a horizontal glass tube, half of which may be filled with flowing tap water, the other half with flowing solution; water and solution are so sharply differentiated that in their movements up and down the tube the fish are presented with a definite concentration step, not a gradient. The reactions of the fish can be checked by reversing the directions of flow. Under these conditions Pygosteus pungitius reacts negatively to solutions of ethyl alcohol, chloroform, formalin, mercuric chloride and zinc sulphate, but appears to have no power of recognizing a copper sulphate solution. Into this the fish swims, to be stupefied and eventually killed.

The present paper describes further experiments on the same lines. When the work was resumed it was the writer’s intention to continue using Pygosteus as test animal, but no further supply of this material could be obtained. Fortunately an adequate supply of Gasterosteos aculeatus was available and work was therefore carried on with this species. Some results have also been obtained with minnows (Phoxinus phoxinus L.).

The most important paper dealing with the toxicity of sulphide solutions to fish is that of Longwell & Pentelow (1935), who have shown that soluble sulphides may be formed by the anaerobic decomposition of sewage. The toxicity of sodium sulphide solutions to trout was tested and it was found that the rapidity of their effect varies with the pH of the solution; from 6·8 to 6·0 the toxicity increases slightly, from 6·8 to 7·8 it varies little, from 6·8 to 9·0 it decreases considerably. Presumably this is because in neutral or acid solution a greater proportion of the toxic substance is present in the form of unionized molecules of hydrogen sulphide. It appears to be generally recognized that the unionized molecule of hydrogen sulphide penetrates living tissue more readily than sulphide ions (see, for instance, Jacques, 1936). Stroede (1933) also deals with the mode of formation of hydrogen sulphide in natural waters and states that concentrations down to 1 mg./l. are harmful to trout; 8·12 mg./l. harmful to carp and tench. Some estimations of the toxicity of this substance to goldfish were made by Ellis (1937, p. 416). The toxic action of hydrogen sulphide resembles that of hydrogen cyanide; some observations on its effect on respiration in Gasterosteus have been made by the writer (1947a). Shelford (1917, chart 3) records two experiments to test the reactions of fish (rock bass and minnow) to sulphide solutions, but both are inconclusive.

Experiments with sulphide solutions present some difficulty. Solutions of H₂S of predetermined concentration are not easy to prepare in quantity; concentrated solutions of sodium sulphide are relatively stable, but are highly alkaline and when diluted to the degree necessary are very unstable as rapid oxidation to water and sulphur takes place. This decomposition appears to be accelerated when the solution is brought to pH 7 by the addition of hydrochloric or sulphuric acid, For satisfactory results it is necessary to employ an apparatus in which a dilute, neutral solution is automatically made up as it runs into the apparatus in which the fish are placed. The arrangement adopted in the present investigation is shown in Fig. 1.

Aspirator A, with inflow and outflow, provides a constant head supply of tap water. Aberystwyth tap water is fully aerated, very soft, of pH 6-8. On opening clip 4 water is delivered to the left side of the experiment tube, opening clip 6 delivers water to the right. Aspirator B contains water of the same temperature as that circulating in A; this runs into the funnel at a constant rate of 200 ml./min. Its rate of flow can be regulated exactly by adjustment of the height of the air intake. The small aspirator C contains a sodium sulphide solution of a concentration one hundred times that required for experiment. This drips into the funnel at 2 ml./min. ; here again the rate of flow is regulated by adjustment of the air intake tube, and a short length of fine capillary tubing makes a satisfactory delivery tube for the dripper. In the funnel water and sulphide solution mix; sufficient sulphuric acid is added to the water in B for the mixture to have a pH of 6-8. This is most easily accomplished by starting with neutral tap water in B, withdrawing a 500 ml. sample of solution through D, titrating this with N/10 sulphuric acid and bromothymol-blue, and thus estimating the amount of acid required to be added to the supply in B when this is refilled, but a certain amount of trial and error is generally necessary before the mixture is completely satisfactory. Finally, the concentration of the mixture is checked by withdrawing another 500 ml. sample and titrating this against standard iodine and starch. The diluted, neutralized solution appears to decompose so rapidly that the concentrations measured in this way are generally slightly below values calculated from the strength of the solution in the dripper and the degree of dilution. The values given in the results which follow are based on the iodine titrations.

When not required the solution can be run to waste through D. When solution is to run into the experiment tube clip 2 is closed, opening clip 3 delivers it to the left side, clip 5 to the right. The rate of discharge from the experiment tube is regulated by screw-clip 11 so that the solution level in the mixing funnel is maintained at approximately the same height as the water supply level in A.

The part of the apparatus shown on the right resembles that used in the earlier investigation. Clips 9 and 10 are on delivery tubes which permit the expulsion of air when filling the apparatus and can also be used for drawing off samples of solution for checks of pH or concentration. The whole solution and water supply arrangement was fixed up on the experiment bench some distance from the tube in which the fish are placed, so that they are not disturbed by the operator’s movements. The coil E is included to make the delivery tube to the left side equal in length to that leading to the right. The general method of experimentation and recording results followed closely that adopted in the earlier investigation. Before beginning an experiment the tap water is run for 1-2 hr. to ensure stability of temperature. The temperatures at which the experiments were conducted are recorded in the legends to the figures. It was found that provided this did not exceed 17° C. the movements of the fish are comparatively slow and deliberate. At 18-20° C., however, they tend to swim wildly up and down the tube and the results are very unsatisfactory.

A representative series of the results obtained with sodium sulphide is set out in Fig. 2. At 0·0007 N the display much distress immediately the solution is admitted and swim actively until all have gathered in the water zone; thereafter they avoid the solution in most definite fashion. At the lower concentrations it will be noted that the fish take a progressively longer time to react, and at 0·00008 N a definite preference for the water zone is not established until the experiment has been in progress for over 47 min. The time taken to produce this effect, that is the time for which the solution must be admitted before all of five fish definitely select the water zone and will not voluntarily enter the solution, may be termed the ‘reaction time’ ; in Fig. 3 the reaction times observed in all the sulphide experiments are plotted.

Below 0-0005 N the relation between reaction time and concentration is expressed fairly accurately by the equation log₁₀t = 2– 4000c, where t is the reaction time in minutes, c the concentration in normality. At higher concentrations theoretical values of t given by the equation are shorter than those observed, but in many of these experiments it was obvious that the ‘reaction time’ observed was really the time the fish took to find their way out of the solution zone, the unfavourable change in their environment being perceived in a much shorter time. Survival times for Gasterosteus are also given in Fig. 3, i.e. the times the fish survives when continually immersed in neutralized Na₂S solutions. It will be noted that the reaction time is substantially shorter than the survival time over the whole concentration range tested.

It is well known that the human olfactory organs can detect hydrogen sulphide at great dilution. A I/100,000 mixture is most unpleasant and Moncrieff (1944, p. 177) states that a 10^-7 mixture of air and H₂S is detectable. No information appears to be available as to the capability of the human organs of smell and taste for detecting the gas in solution.

Inspection of the result for o-ooo 15 N (Fig. 2) shows that whereas the initial reaction time is 19-21 min. the response on reversing the water-solution flow at 25 min. is much more rapid (about 4 min.), and the third response at 30 min. more prompt still. A similar result is generally observed at any other concentration at which the primary response takes some measurable period of time, as the other recordings in Fig. 2 indicate. This is not due to the fish becoming more familiar with the vessel in which they are confined, for keeping them in the apparatus for 7-8 hr. before the experiment is begun, during which period they explore their prison most thoroughly, results in no shortening of the primary reaction time. It is possible that in their many journeys up and down the tube the fish gradually learn to distinguish between water and solution, for when the reaction time is several minutes some fish may cross the water-solution junction thirty or forty times; But it seems more probable that the improved response displayed on reversing the flow results from a much more rapid development of the toxic symptoms which excite the response. Thus at the end of 25 min. (in the 0-00015 N experiment) the fish have all spent some considerable time in the solution, have developed some degree of toxaemia, and spent an insufficient time in the water zone for recovery. When the solution is now run in on the side on which they have congregated the toxic symptoms necessary to induce the negative response are developed much more rapidly, ‘with an accompanying improvement in the speed of avoiding action. Before this hypothesis can be established a technique must be devised whereby the progress of the negative reaction is observed, together with measurement of the rate at which the sulphide solution depresses the oxygen utilization of the fish.

Using Gasterosteus as test animal the effect of lead nitrate solutions was studied over the concentration range O-OOOOI-O-I N. The apparatus lay-out for these experiments resembles that adopted for sodium sulphide, but as lead nitrate is reasonably stable, even at great dilution, the dripper arrangement can be dispensed with and aspirator B filled with solution of the required concentration before beginning the experiment. With the air-intake tube in position this runs into the funnel at 200 ml./min. ; this was found a better arrangement than connecting B directly with the supply tubes leading to the bubble traps, for the level of solution in the funnel provides a convenient check on the rate at which water and solution flow through the observation vessel.

Four typical results are given in Fig. 4. At O-IN the experiment was not satisfactory, owing to stratification of the heavy solution, but it was evident that the fish reacted positively. At 0·04N a better separation of water and solution was achieved ; at this concentration the fish react negatively at first (note their temporary retreat at 6-7 min.), then they persist in swimming into the solution, and sooner or later are seen to be deliberately avoiding the water. When the directions of flow are reversed they select the lead nitrate zone almost immediately. Shelford (1917) noted many instances of fish reacting positively to toxic substances ; this he attributed to the fish failing to recognize the solution, swimming into it and becoming ‘intoxicated’.

Some efforts were made to elucidate this positive reaction, but without complete success. The pH of 0·04N lead nitrate is 5·4; this is not an attraction for Gasterosteus which is fairly definitely negative to water of this degree of acidity (see Fig. 5ft). Another explanation that might be suggested is that the fishes’ chemical sense is dulled or destroyed by exposure to concentrated lead nitrate, for it is well known that washing out the mouth with a strong astringent temporarily impairs the sense of taste very considerably. Fig. 5 a records one of the experiments carried out to test this theory. Five fish were placed in the observation vessel and their reaction to 0·0007N lead nitrate tested in the usual way ; the result, a prompt negative response. Then after 5 min. of double water flow 0·04 N lead nitrate was run in at both ends for 15 min. This was washed out with water and then 0·0007N lead nitrate was run in on one side. Though the fish showed obvious distress after their 15 min. exposure to the more concentrated solution, they reacted to the dilute solution negatively, and quite as promptly as before.

Next the reaction of Gasterosteus to calcium nitrate was examined. To dilute solutions (e.g. 0·001N) the fish appear quite indifferent, but to 0·04N Ca(NO₃)₂ a positive reaction is displayed. A similar result is observed with 0-03-0-05 N sodium nitrate or 0·03-0·07N sodium chloride. These results suggested that the osmotic pressure of the 0·04N lead nitrate might be its attractive feature and the reaction of Gasterosteus to glucose solutions of similar osmotic pressure was tested. Here, however, the reaction of the fish was negative ; they preferred the water and if they ventured into the glucose appeared irritated, swimming in an erratic and nervous maimer. It has to be admitted that at the present stage a satisfactory explanation of the positive reaction of the stickleback to concentrated lead nitrate is not forthcoming; in a subsequent section of this paper it will be shown that Gasterosteus is positive to dilute ammonia solutions.

At 0·01 N the reaction is vague, the fish keep wandering restlessly to and fro, and on further dilution (0·004N) a negative reaction is displayed. At 0·0004N the response is similar, and equally prompt; on running in the solution the fish are greatly agitated and quickly choose the water zone. Here they remain and should one venture to the centre of the experiment tube it pauses, gobbles, and retreats with obvious distaste. At 0·00002N (approximately 2 mg. Pb/1, survival time about 16 hr.) a delayed, but nevertheless definite negative reaction is displayed. The movements of the fish are leisurely and in their visits to the solution zone they display little irritation. On reversing the water-solution flow (at 25 min.) they again have some difficulty in selecting the water zone and some fish swim up and down the tube ten or more times before making up their minds. One experiment was run at 0·00001 N ; in this a vague negative reaction developed in about 30 min.

In Fig. 6 the reaction time-concentration curve is plotted together with some reaction times observed with zinc sulphate and with lead nitrate when the minnow (Phoxinus phoxinus) is used as test animal. Five fish were used at each concentration. The minnow is much more sensitive than the stickleback; 0·0007 N lead nitrate is avoided immediately, the fish retreating from the solution as it enters, and even a 0·000004N solution (0·4 mg./l.) excites a definite negative response in 8 min. Single fish experiments in which all the movements of the fish were recorded were also made with Phoxinus-, three results which illustrate very well the reactions of the minnow to dilute lead nitrate are given in Fig. 7; note the almost immediate rejection of the solution at 10^-4N, the several trials and eventual rejection of the solution at 10^-5N.

It is possible that the minnow can detect, but is not irritated by, lead nitrate solutions of greater dilution than 0·000004 N, but even if this represents the threshold it is remarkably low. Moncrieff (1944, p. III) supplies some information on the sensitivity of the human sense of taste to heavy metal salts, stating that silver nitrate is just detectable in 0·003 % solution. This would be about 0·00018N, or 19 mg./Ag/l.

It is well known that the hydrogen-ion concentration of the water is one of the most important factors determining the type of fauna found in any aquatic environment and observations in field and laboratory on the pH ranges different animals can withstand are in plentiful supply. Using the goldfish as test animal Ellis (1937, p. 409) has studied the lethality of eleven acids, and (1937, p. 379) reviews the results of a number of workers which seem to show that fresh-water fish have a considerable tolerance for variations in hydrogen-ion concentration. On the other hand, little study appears to have been made of the motor reactions of fish to water of abnormal acidity or alkalinity. Wells (1915 a) made some attempt at the problem, but his results are somewhat vague; using the gradient tank and three different species of fish he decided that fish are negative to neutrality in favour of slight acidity. According to Shelford (1929, p. 480, footnote) Wells took neutrality as the ‘turning point’ of phenolphthalein, presumably about 9·0.

The reaction of Gasterosteo to acid and alkaline water has been studied over the pH range 3-2-12-0. In every experiment the fish were offered acid or alkaline water as an alternative to the Aberystwyth tap water (pH 6-8), in which they were kept until used for experiment, and which has a pH approximately that of the river from which they were obtained. The scheme of apparatus resembles that of Fig. 1. Aspirator A supplies water of pH 6-8; aspirator B is replaced by a duplicate of A, providing a second constant head supply from the same pipe. This runs at 200 ml./min. into the funnel to dilute, about one hundred times, a hydrochloric acid or sodium hydroxide solution dripping steadily from C. Samples of the mixture are withdrawn at D, their pH determined, and the concentration of the supply in C and its rate of flow adjusted until the mixture has the pH required. This is finally checked by the examination of samples drawn at 9 or 10.

Colorimetric estimation was found quite satisfactory for the acid range; for the alkaline range phenol-thymol-phthalein is satisfactory up to pH 10·4. Above this value no completely reliable indicator could be found, though alizarin S, methyl blue, nile blue and brilliant cresyl blue were tried, and pH determinations over the range 10·4-12·0 were therefore checked with a Cambridge hydrogen-ion apparatus.

One of the experiments with acid water is that recorded in Fig. 56, already referred to. All the other hydrogen ion experiments with Gasterosteus are summarized in Fig. 8. To water of acidity greater than pH 5-4 the stickleback is definitely negative, avoiding it with the usual symptoms of discomfort. The writer (1939) has shown that the hydrogen ion threshold is about pH 5-0, which the stickleback survives for 9 days. Using nitric acid Carpenter (1927) obtained a similar result with Phoxinus.

At 5-8 a very vague negative reaction is displayed and from 6·0 to 7·0 the fish appear indifferent. Water of pH 7·0 to 11·0 similarly excites no definite response; if any reaction at all develops in the course of 30 min. it is an indefinite preference for the alkaline zone. At 11·4 a fairly definite negative reaction is developed, and still more alkaline water is very quickly avoided with widely distended mouth and other symptoms of distress, the alkali portion of the reaction time curve dropping much more steeply than the acid.

The toxicity of ammonia solutions to fish has been studied by a number of investigators interested in pollution problems, including Ellis (1937), Shelford (1917), and Mason-Jones (1930). Grindley (1946) has measured the toxicity of ammonium chloride to rainbow trout, his experiments seem to place the threshold at 314 mg. NH₄C1/1. (about 0·006N). Shelford concluded that fish are unable to recognize even highly toxic solutions of the gas, but the two experiments he records (1917, chart 2) are not very satisfactory for the fish were driven to the ammonia side of the gradient tank, nor does he state the actual concentration tested.

Four of the writer’s results with Gasterosteus are given in Fig. 9. The apparatus arrangement used for lead nitrate was found satisfactory; the ammonia solutions were prepared by diluting a normal solution of ammonia standardized with N/IO HC1, the concentrations of the solutions being checked by titrating 500 ml. samples at the end of the experiments, and the pH values given are those of samples drawn from the experiment tube.

To 0.04N ammonia the stickleback is promptly negative, a negative reaction is also developed towards a O-OIN solution, but somewhat tardily, though this concentration is highly toxic (survival time 4$12$ min.). Two fish persisted in swimming into the solution with the result that they died suddenly with convulsive movements and widely distended mouths. With further dilution the negative response disappears and-is replaced by a most definite positive reaction; in the 0·001 N experiment note how the fish avoid the water and gather in the solution zone within 2 min. of the flow reversal at 12 min. At 0·000IN the reaction is still positive.

In dilute ammonia solutions we apparently have a clear case of positive reaction to a lethal substance. As the osmotic pressure of the solution differs little from that of the tap water, and as the pH of the solution does not appear attractive, we can only conclude like Shelford that the toxic substance ‘intoxicates’ the fish in some way.

Furthermore, it will be noted that the general result with ammonia solutions is the converse of that observed with lead nitrate, to which Gasterosteus is positive at high concentrations, negative at great dilution. Hence, when the reaction of fish to any toxic substance is studied it is important to cover a wide concentration range, otherwise any conclusions based on the results may be inaccurate.

Reviewing the whole of the writer’s observations on the behaviour of fish to toxic solutions the following general observations may be made. The reaction observed may be of three types: (1) Positive, the fish being attracted by the toxic substance itself (ammonia), possibly by some other property of the solution, such as its osmotic pressure (concentrated lead nitrate). (2) A complete failure to distinguish between the solution and water, with a pseudo-positive reaction when the substance in question is highly toxic (copper sulphate). (3) A negative reaction, the promptness of which depends on the concentration of the solution, and which may be maintained down to the threshold of toxicity (sulphide). Fish may be positive to a substance in high concentration, negative to the same substance in dilute solution, and vice versa.

Even when substances whose physiological effects are similar are considered the fishes’ capability of detecting and rejecting them is not related to their degree of toxicity. The four heavy metal salts tested, in decreasing toxicity form the order mercuric chloride, copper sulphate, lead nitrate, zinc sulphate. According to the stickleback’s ability to detect and avoid their presence the order is lead, mercury, zinc, copper.

Lastly it is evident that the sensitivity of one fish to a particular substance may differ considerably from that of another, for the minnow is some ten times as sensitive to lead nitrate as the stickleback.