The Relation Between Growth and Food Consumption in the Brown Trout (Salmo Trutta)

The experiments described in this paper were designed to compare the relations between food and growth in a fresh-water fish with those of the plaice, and also as a step towards the understanding of the part played by a carnivorous fish in the ecology of a stream.

The experiments began in October 1933 and were continued, with some intermissions, until November 1935. Subsequently a series of experiments on the loss of weight of starved fish were carried out in 1936.

Brown trout (Salmo trutta L.), obtained from a commercial hatchery and kept in stock ponds at the Alresford station until required, were used for these experiments and proved to be very suitable. They withstood confinement well (two fish, for example, lived in the experimental tanks from October 1933 until the summer of 1935), were easily handled and their principal food in the Itchen, Gammarus pulex, was easily obtainable in sufficient quantity.

The experiment was begun with twelve fish which were approximately eight months old, having been hatched in the early spring of 1933. They were small for their age, their weights varying from 1·1 to 3·0 g. Each fish was accommodated in a wooden tank 56 cm. long, 26 cm. wide and 27 cm. deep (internal measurements) fitted with a gauze fid. For convenience the tanks were arranged in a double row supplied the tank immediately below, the two top tanks were fed by a bifurcated pipe from the main laboratory supply, of river Itchen water. Although this water was untreated it is unlikely that it carried any appreciable food, for animal plankton (trout appear to be entirely carnivorous) is very scanty in the Itchen, and the fish in the top tanks, which would have had the best chance of getting any food from this source, showed no better growth than the others. Any larger organisms, which, incidentally, were never observed in the water, would be excluded by the gauze lids which also prevented the access of any air-borne food in the form of insects.

To prevent any loss of food the outlets from all tanks were covered with bronze gauze of 1 mm. mesh. The tanks were numbered 1–6 on one side and 7–12 on the other, and the fish were identified by the number of the tank they occupied. The maximum and minimum water temperature was recorded each day in tanks 1 and 12.

With very few exceptions the fish were weighed weekly. They were dried gently with a towel to remove as much surface water as possible and then placed in a tube of water of known weight and weighed on an air-damped balance. The weights of small fish were recorded to the nearest milligram, for, although it was originally assumed that an accuracy of 0·1 g. would be sufficient, it was found that duplicate weighings of the same fish gave values which did not generally differ by more than a centigram. When the fish became so big that they could no longer be weighed on a sensitive balance, they were dried, placed in a weighed towel and weighed on a rough balance to the nearest gram.

The fish were fed on live G. pulex (the fresh-water shrimp), obtained freshly for each meal from the river Itchen or channels connected with it. They were usually fed six times a week, no food being given on the day they were weighed, but when it was necessary to miss a day they were given a double ration on the day before. They were, however, always given a normal one-day ration on the day preceding that on which they were weighed, to eliminate, as far as possible, errors due to varying amounts of food in the gut. The shrimps were dried on blotting paper and then weighed in a tube of water. So long as the amounts used were small they were counted as well as weighed, in order to obtain figures which might later be compared with data on the population of Gammarus in a river. The tanks were examined every day, and any dead Gammarus were removed, dried on blotting paper and weighed. Live Gammarus were not removed until the day on which the fish were weighed. On this day all the tanks were thoroughly cleaned, the water siphoned out and all Gammarus, alive or dead, were removed, dried on blotting paper and weighed. By this method the amount of food eaten per week was estimated.

In order to check this estimation, at one period during the experiment one tank was left for 3 weeks without a fish in it ; a weighed quantity of Gammarus was put in each day, dead ones picked out and weighed daily, and the surviving Gammarus recovered and weighed at the end of each week. In each of the 3 weeks the weight of Gammarus (living and dead) recovered slightly exceeded that put in (Table I).

The differences are perhaps small enough to be due to experimental error; if they have any significance they represent the amount of growth made by Gammarus during the experiment. Clearly there was no loss of food during this time, and in the feeding experiments the difference between the weights of food put in and recovered can be taken to be the amount eaten by the fish.

Six fish, those in the odd-numbered tanks, were given an excess of food in order to obtain data on the relation between the amount of food eaten and the growth made in fully fed fish. The rations of the other six were carefully regulated in an attempt to keep the fish at constant weight from week to week, in order to estimate the amount of food required for the bare metabolic needs of the animal, allowing no excess to be used for growth. It will be shown that it is impossible to foretell exactly how much food will be required for this purpose, but a fair measure of success was achieved.

The fish were first placed in the tanks and fed on Gammarus on 12 October 1933, and, after some preliminary trials, routine weighings were begun on 2 November. They were continued without intermission until 31 August 1934. Trout 10 died on 24 November and no. 2 on 24 December. They were replaced, no. 10 by a fish of approximately the same size (no. 10 A), and no. 2 by one a year older and weighing 12 g., but this fish was lost after 1 week and another fish (2B) weighing 14 g. was substituted. Owing to tank 7 developing leaks large enough to allow Gammarus to escape, the data provided by fish 7 were unreliable until the leaks were finally stopped on 12 April 1934. On 14 June 1934 all the fish except nos. 1, 2B and 3 were killed accidentally, and until the end of August the experiment was continued with the three survivors only.

The experiment was then discontinued for a period of 7 weeks. During this time, as at all other periods when the work was discontinued, the fish were retained in the tanks, fed on Gammarus and the temperatures were recorded daily but neither the fish nor food were weighed.

On 18 October the experiment was restarted. Fishes 1 and 3 were given an excess of food and no. 2B a maintenance ration as before, and six more fish were placed in tanks 7–12. These fish will be referred to as nos. 7 A, 8A, 9A, 10B, 11A and 12 A. They were approximately 8 months old and their weights ranged from 1to 9 g. The rationing of these fish was different from that of the corresponding numbers in the original experiment; nos. 7 A and 10B received maintenance rations, nos. 8 A and 12 A were given an excess of food, and nos. 9 A and 11A were given about 1·7 and 0·75 g. of food per week respectively. These amounts were roughly twice the maintenance requirements of fishes of their size, and it was hoped to see in the early weeks of the experiment whether fish on an intermediate ration used their food more efficiently for purposes of growth than did those receiving an excess of food (cf. Dawes, 1930–1), but insufficient data on this point were obtained. As the fish grew this constant amount of food became a smaller proportion of the body weight until eventually it was not sufficient to allow any further growth, and it was hoped by keeping the amount of food constant to obtain variations in the weight of fish corresponding to variations in the amount of food required for maintenance. Fish 10 B died after 2 days and was replaced by another somewhat larger trout (10 C).

The experiment was suspended from 21 December to 10 January 1935, when it was continued with the same fishes and the same rationing system until 18 April. On 7 March it was impracticable to continue to weigh fishes 1 and 3 (which now weighed over 50 g.) on an accurate balance, and thenceforward they were weighed to the nearest gram as described on p. 447.

On 28 March, tank 12 was found to be empty, the fish having escaped. The weighings were again suspended from 18 April to 9 May. Then they were recommenced and continued without intermission until 8 August. On 6 June fish 1 jumped from the trough in which it was confined whilst the tank was being cleaned and died. When the experiment began in November 1933 this fish weighed 3 g. and at the time of its death it weighed 91 g. On the same day (6 June) a trout fry about 3 months old was weighed and placed in tank 12. This fish (12B) was fed on young Gammarus of such a size that it could easily swallow them until it grew sufficiently to take shrimps of the ordinary size. On 24 June fish 3 was found dead, presumably owing to the effect of the high water temperature on a comparatively large fish in a small tank. At the beginning of the experiment this fish weighed 3 g., and at the time of its death 99 g.

On 8 August the experiment was again suspended until 19 September. From then until 7 November, when the experiment was concluded, no more interruptions occurred. Fish 2B was found dead on 17 August, but there were no further losses.

During 1936 eleven experiments were carried out to determine the loss of weight of fish kept without food for a week. The fish were taken from a stock pond, dried on a towel, weighed and kept, each usually in a separate tank, for a week without food and then reweighed. At first a number of fish were weighed daily, but this involved too much handling and all these fish died within the week.

During the final experiment in November the fish, after being weighed at the end of a week’s starvation, were replaced in the tanks and starved for another week. They all survived, but the loss of weight in the second week was considerably less than in the first. The implication of this will be discussed later.

To avoid any more handling of the fish than was necessary no measurements of length were made, and the only data obtained on the rate of growth were the figures for increase of weight. Similar series of determinations of the rate of growth of trout do not appear to be available elsewhere, and they are worth a brief discussion purely as measurements of growth.

For the study of the growth of fish which were at all times supplied with an excess of food, and where therefore the amount of food available was not a factor limiting the rate of growth, the data given in Table II are available.

The data are clearly insufficient to allow any statistical investigation of growth in brown trout, yet, as Minot (1891) has pointed out, the regular measurements of individuals over a considerable period of time often reveal features of growth which are masked or misinterpreted when the figures are based on the average of a large number of determinations.

There are considerable variations in the rate of growth expressed as increase in body weight per unit time. The fish which grew most rapidly was 8A, which increased its original weight by nearly 80 g. in 385 days, whilst no. 11, which showed the poorest growth, only increased by 5·7 g. in 224 days. Since, as will shortly be shown, this rate of growth is continually changing, determination of the average amount of growth per day from these figures gives an entirely misleading comparison between one fish and another and cannot therefore be used.

Surber (1935) found that American brook and rainbow trout fed solely on Gammarus fasciatus gave the growth in weight as shown in Table III.

It will be seen that the fish used by Surber grew considerably faster than the brown trout. There are several possible explanations of this fact. First, the American species may be faster growing than the British one. Secondly, in Surber’s experiment the water temperature throughout the experiment was constant at 54° F. In the experiments described in this paper the temperature varied considerably, and it will be shown later that the increase in weight in unit time was greater when the water temperature was between 50 and 60° F. Thirdly, the rainbow and brook trout were considerably larger when measurements began than were the brown trout, and when we come to discuss the weekly increments in weight it will be seen that they increase, within limits, with the size of the fish.

It may here be noted that the brown trout used in these experiments appeared to grow as quickly as fish of the same age in the Itchen at Alresford. Wild fish in this area are generally from 6 to 8 in. long when they are 2 years old, and in this experiment fishes 1 and 3 were both 8 in. long when they died in June 1935, at approximately $214$ years old, and fish 8 A was 8 in. long in November 1935 when it was only $134$ years old.

The complete records, showing the increase of weight from week to week have been filed and are available to anyone interested. Many are illustrated graphically in Fig. 2. The curves are not smooth, and, since the weighings were not begun until the early phases of growth were past and were discontinued before the fish became mature, they are by no means complete, but in form they resemble those for increase in weight in man, guinea-pig, mouse and carp given by Ostwald (1908) and for the plaice by Dawes (1930–1).

They show that, except at certain periods, increasing size was accompanied by increasing increments in weight, so that when the figures are plotted a curve concave to the ordinate is produced. It may be assumed that, had the measurements continued, at some point this relation would no longer hold, increasing size being accompanied by diminishing increments and the curve would become convex, thus producing the S-shaped curve typical of growth. There is no indication that this point of inflexion was reached in these experiments.

All the fish used grew very slowly from November 1933 to about the middle of February 1934, and from then onwards the rate of growth increased, though it was more marked in some fish, e.g. no. 5, than in others, e.g. no. 11, until 14 Jume, when four of the six fish were killed. The rate of growth of fishes 1 and 3 appeared to diminish somewhat in July but increased again in August and continued to be rapid until the beginning of November when it declined, and indeed on 30 November both fishes weighed less than they did the week before. After this, however, rapid growth recommenced, to be interrupted again in the middle of January, an interruption that was more marked and persistent in no. 3 than in no. 1. Thereafter the growth rate of both fish continued high until both fish died in June. The growth of fish 8 A was more regular. Growth was rather slow when the fish was first brought into the experiment in October 1934, but it increased fairly steadily up to the middle of July. Then growth practically ceased for a fortnight and was slow until the end of August. Then the rate increased and remained high until the experiment was discontinued on 7 November 1935. It is noteworthy that all three fish showed a period of less rapid growth at midsummer as well as during the winter.

The variations in growth rate are, however, very much more easily realized and measured if successive increments of weight are plotted against time and a curve drawn through the points so obtained. This is the acceleration curve. It rises with an increasing growth rate, is parallel to the axis when the rate is constant, and falls when the growth rate diminishes. The weekly increments have been plotted and curves drawn for fishes 1 and 8 A in Fig. 3. The other fish show similar features and have been omitted in order to avoid undue complication of the graph.

The figure shows that the actual growth rate constantly varied, periods of rapid growth alternating at frequent intervals with periods of slow growth, and that it was only occasionally that the rate was the same even for a fortnight. A similar phenomenon has repeatedly been observed when a growing mammal is weighed at regular intervals (Davenport, 1908, footnote on p. 288) but it does not appear in the data for the growth of plaice given by Dawes (1930–1), either because in this animal the growth rate is more regular or because the fish were not weighed frequently enough to record it. The variations of the growth rate deduced from the study of Fig. 2 represent therefore changes in the average rate over a period of some weeks, for it is only when the differences are considerable and continued for several successive weeks that, owing to the scale of the diagram, the slope of the curve is appreciably affected.

It has already been explained that a number of fish were given a restricted amount of food so adjusted as to maintain their weight approximately constant. The growth of these fish from week to week cannot therefore profitably be discussed except as part of the study of their food requirements. At intervals, however, the experiment was suspended and during these periods the fish were given an unweighed and presumably ample ration and were therefore free to grow. Their growth rate at these times is worth a brief examination.

Fish 2B was given unweighed quantities of Gammarus from 31 August 1934 to 18 October 1934. During this period its weight increased from 15·8 to 27·6 g., an increment of 11·8 g. During the same period fish 1 showed an increase from 24·6 to 32·7 g., i.e. 8·1 g., and fish 3 from 27·9 to 35·6 g., i.e. 7·7 g. The experiment was again suspended from 20 December 1934 to 10 January 1935. During this period:

During the same period the fish which had at all times received ample food showed the following increases :

The experiment was suspended again from 18 April to 9 May. The fish usually kept on restricted diets showed the following increases :

whilst the fully fed fish showed the following growth :

During the period between 8 August and 19 September the following increases were obtained :

During the same period the other fish grew as follows :

These data show that whenever its food supply was uncontrolled fish 2 B increased its weight more over the same period than did nos. 1 and 3. There is here some indication that a fish, whose growth has been retarded by lack of food, can, when food becomes abundant, grow more rapidly than those whose previous growth has been unrestricted. Fishes 7 A and 10 C, when their food supply was unregulated, grew at approximately the same rate as no. 8 A and do not appear therefore to support this view, but these fish were never kept on maintenance rations for more than 14 weeks without a break, whereas no. 2B, from the time it came into the experiment, was rationed for 34 weeks continuously. As during the intermissions the fish were free to grow it is clear that nos. 7 A and 10 C were never so far removed from the normal weight for their age as was no. 2B, and we should not expect therefore such striking effects on their growth rate.

Minot (1891) found that if any of his guinea-pigs showed for a period a growth rate lower than normal, e.g, on account of illness, the loss was very soon made up and, as he says, “each individual appears to be striving to reach a particular size”. It seems to be a reasonable hypothesis that any young animal has the power to compensate for a period of reduced growth.

The average weekly water temperature has been plotted in Fig. 3, and it is quite clear by inspection of this curve and the weekly growth increments of the fish plotted on the same graph that there is no simple and direct relation between the amount of growth made in any one week and the temperature of the water during that week. This does not mean, of course, that differences in temperature have no effect on the growth rate of trout; the experiments were not designed to study the effect of temperature, and there were too many other uncontrolled variables to enable it to be clearly shown in the primary data. If, however, the experiment is divided into periods during which the temperature remained approximately constant, the average temperature for these periods calculated, and the growth data from all fully fed fish used, then some idea of the optimum temperature conditions for the growth of brown trout can be obtained. Since these periods were not all of the same length, and the fish differed in size, it was necessary to reduce the growth data to some common unit. The method adopted was to calculate the variation in weight (increase or decrease) in milligrams per week per gram of the weight of the fish at the beginning of each period (known for convenience as the starting weight). This method of presenting the data is similar to that used by Minot (1891), and it has been criticized by D’Arcy Thompson (1917) on the grounds of unnecessary complexity and by Gray (1929) because it is based on the assumption that the amount of growing tissue in an organism is directly proportional to the whole weight of the organism. As, however, no better way of comparing the growth of one fish over a given period with that of another of a different size over a different period has been found, this method has necessarily been used, though its limitations have been fully realized.

In Table IV are shown, then, the average temperatures of the various periods into which the experiment was divided, the weight of the fish at the beginning of each period, the average increase or decrease in weight per week expressed as milligrams per gram of the starting weight, and the average amount of food consumed per week by fully fed fish also expressed as milligrams per gram of the mean body weight. Discussion of this last column will be deferred for the present.

From these data a dot diagram (Fig. 4) has been prepared to show the relation between the increase in weight per week and the average temperature. It is immediately apparent that no curve can be drawn to express this relation, for there were large variations in the amount of growth at all temperatures, but it does appear that on the whole growth increased with increasing temperature from 38 to 50° F., reached a maximum between 50 and 60° F. and declined when the temperature exceeded 60° F. This is more clearly shown if the data are arranged in a correlation table in which the temperatures have been grouped at 5° intervals and the increases (or decreases) of weight at 20 mg. intervals and the number of times each growth class was associated with each temperature class recorded (Table V).

It appears then that increasing temperature up to about 60° F. was accompanied by an increase in the growth rate of these fish, but that above 6o° the growth rate tended to diminish. It has previously been noted that in both years of the experiment there was a period in the summer when the growth of fully fed fish was slow, and it seems likely that when the water temperature exceeds 6o° F. the optimum conditions for trout growth have been passed. From the known facts of the geographical range of the brown trout (Tate Regan, 1911) we should expect that high water temperatures would not be the most favourable for the growth of these fish, for they are natives of northern lands and inhabit the colder parts of streams, the head-water or spring-fed portions. In the southern half of their range they do not occur in the lowland rivers which become warm in summer, and although they have been introduced in the tropics it is only where, as in Kashmir or Kenya, the effects of latitude are modified by the considerable altitude of the streams that they have been successful.

A fish, like any other organism, needs food to provide the energy for its vital processes such as respiration, digestion and excretion, and for its various activities involving movement of the body. If it obtains more food than is necessary for these metabolic needs the surplus is available for increasing the body substance, that is, growth. Before, therefore, the relation between food and growth can be discussed, it is necessary to determine what proportion of the food is required for the maintenance of the ordinary activities of the trout. It should be noted that as this study was primarily ecological no attempt has been made to assess the energy value of Gammarus as food nor to discuss its composition in physiological terms. It must also be understood that the relation between the amount of food required to maintain the fish at constant weight and the basal metabolism of the fish cannot be determined from these data, for, within the limits of their tanks, the fish were free to move as they liked and clearly their activity is a factor of unknown magnitudeaffecting the data. Yet, since all the fish, both those on maintenance rations and those fully fed, were kept under uniform conditions it is reasonable to suppose that the figures obtained from the different fish are approximately comparable.

To determine the “maintenance” requirements a number of fish were given restricted rations which were adjusted from week to week in an attempt to maintain the weight of the fish constant. The full weekly data have been filed for reference ; they show that the experiment met with a very fair degree of success, for, although in only one instance was the weight of the fish the same in two successive weeks, the variations were very small and the amount of food given in any week was a reasonable approximation to the maintenance requirements of the fish.

It was at once apparent that, as would be expected, the maintenance requirements depend on the size of the fish and they vary seasonally, being high in the summer and low in the winter. In order to eliminate, therefore, the factor due to the varying sizes of fish used in the experiment and to investigate the seasonal changes in food requirements, in Table VI the data have been grouped and recalculated. The experiment has been divided into periods of differing lengths, but of approximately constant temperature, the mean temperature and the mean body weight of each fish have been calculated and the amount of food eaten has been expressed as milligrams per week per gram of the mean body weight and the variations in weight as milligrams per week per gram of starting weight, for each period.

Since in only a few cases was the weight of a fish the same at the end as at the beginning of a period, it is impossible in most cases to state the exact maintenance requirements of the fish but, so far as can be judged from the amount of food eaten and the gain or loss in weight, the requirements of each fish in any one period were roughly the same and, indeed, show much less variation than might have been expected. In particular it may be noted that the body weight of the fish did not appear to influence its relative maintenance requirements; for example, fish 2B when it weighed about 14 g. gave much the same values as no. 4 which only weighed i g., and later on when no. 2B weighed about 26, 31 and 35 g. it gave figures similar to those of no. 7 A, which during these same periods weighed only 9, 11 and 13 g. respectively. Dawes (1930–1) found that there was a definite diminution in the maintenance requirements of the plaice with increasing body weight, but no evidence of a similar phenomenon was shown in these experiments with trout.

It seems then that, as the variations in weight of the fish from week to week were so small, if the average food consumption per fish expressed in milligrams per week per gram of mean body weight be calculated for each period a measure of the maintenance requirements is obtained, which, whilst not quite accurate, is a good approximation to the true value and which can be applied, without serious error, to all the fish used in the experiment. The values so obtained are given in Table VII. It will be seen that the amount required varied from 51 to 270 mg./g. of body weight, but extreme values were not numerous and in most cases the figure was between 70 and 102 mg. The median is at 83·5 and the semi-interquartile range is ± 16·0.

Dawes (1930–1) found that the maintenance requirements in plaice varied seasonally and also according to the body weight of the fish. The values determined by him vary between 28 and 315 mg. per week per g. of mean body weight but most of them lie between 70 and 140 mg. That plaice fed on mussels and trout fed on Gammarus should give approximately the same values for maintenance requirements is curious and can, at present, only be regarded as a coincidence.

The variations in the maintenance requirements of trout are obvious in Table VI and have been illustrated graphically in Fig. 5. During the first 32 weeks of the experiment, when six fish were under observation, the points lie approximately on a reasonably smooth curve which follows closely the curve of temperature variations. Later, when the experiment was continued with one, two or three fish the relation between food requirements and temperature was not nearly so regular or well marked. It is to be expected that in a cold-blooded animal the amount of food required would be greatly influenced by the external temperature, and the results of the earlier months of the experiment were in accord with this expectation. Whether the discrepancies in the subsequent data were due to some unrecorded factor, or whether they were merely a result of the less satisfactory experimental conditions, the small number of fish and the periods when observations were suspended, is not known. Dawes (1930–1) found that the maintenance requirements of plaice were influenced by the temperature of the water and that during the winter “a plaice requires only a fraction of the daily ration which is required to maintain the weight constant during the summer months”.

During 1936, as has already been recorded, experiments were carried out to determine the loss in weight of trout kept without food for a week, in the hope that the data so obtained could be correlated with the amount of food required for maintenance. This hope was not fulfilled but the experiments are worth a brief discussion.

The methods used have already been described on page 450 and the data obtained have been summarized in Table VIII. In this table the loss of weight has been expressed in mg./g. of starting weight, and the observations have been grouped according to the temperature, and to the size of the fish. The average values obtained have been given for 2, 10 and 40 g. intervals, and the number of fish in each group, is shown by the index figure. The full data on which this table is based have been filed and are available to anyone interested.

These figures show only that, as would be expected, the loss in weight is considerably greater at temperatures between 55 and 60° F. than it is below 50°. There is not, however, any simple relation between the temperature and the weight lost, for, although there are sequences in the averages, they are not progressive from a low loss at a low temperature to a high one at a high temperature. The proportionate loss in weight does not appear to bear any relation to the total weight of the fish, a result which accords with the finding that the relative maintenance requirements of trout do not appear to vary with the body weight, but it is doubtful whether this evidence is of any value, for the figures for loss of weight in Table VIII bear no apparent relation to the amounts required for maintenance as shown in Table VII.

It is probable that these unsatisfactory results were due to deficiencies in the experimental procedure. The fish were taken direct from the stock pond, and no information on how recently they had fed was obtained. Some, therefore, probably had food in their stomachs, whilst others were already empty. The digestion of an unknown weight of food must clearly produce a diminution in the weight of the fish which is quite unrelated to the normal wasting due to starvation. Another probable source of error was shown in the first series of experiments. It has previously been recorded that an attempt was made to weigh some of the fish every day, but that all those subjected to this procedure died at or before the end of the week. Table IX shows that in every case the weight of the fish increased before it died although it received no food. The most reasonable explanation of this is that when the fish became weakened by starvation and excessive handling the normal mechanism controlling the water intake into and output from the body broke down, with the result that an excess of water entered and the weight therefore increased. It is not unreasonable to suppose that a similar variation in water intake may have occurred in some, at any rate, of the starved fish which were not weighed every day. In this connexion it may be noted that the fish which were starved for a fortnight (see p. 450) without exception showed a greater loss of weight in the first week than in the second, and it is not certain whether this was due to less loss of body substance in the second week, or to greater water absorption during this time, and the data will not therefore be further discussed.

From 18 October 1934 until the experiment ended on 7 November 1935, fishes 9A and nA were given respectively approximately 1·7 and 0·75 g. of Gammarus per week, except at periods when weighings were suspended, when they, like all the other fish, received an unknown, but probably ample, meal of Gammarus every day. At the beginning 1·7 and 0·75 g. were, for these fish, rations considerably in excess of their maintenance requirements, for they represented 238 and 269 mg./g. of body weight respectively, while the maintenance requirements (see Table VII) were at this time about 100 mg./g. of body weight. As can be seen from Table X (the full data have been filed for reference) the fish accordingly grew steadily. Obviously as the fish grew the proportion of the weight of food to body weight declined and eventually growth stopped. In both fishes this point was reached in early summer, and as at this time, owing to the high water temperature, the maintenance requirements continued to increase, the fish began to lose weight, for the amount of food they received was below their maintenance ration. Then in the autumn, when the temperature- and the amount of food needed to keep the body weight constant fell, these rations again became sufficient to allow growth to recommence. These facts are summarized in Table XI below, which is abstracted from the data in Tables VII and X.

The weight variations of these two fish therefore agree reasonably well with expectations based on the amount of food required to keep fish at a constant weight.

The growth of those fish which at all times received an excess of food has already been described, and it only remains to discuss the amount of food eaten and the relation of food eaten to growth made.

The data are available in full for reference and have been summarized and reduced to a common form in Table IV. Those for fishes 1 and 8 A are illustrated in Fig. 3.

In the last two months of the first year (fishes 1, 3, 5, 9 and 11) the amount of food eaten per week in the winter was small, less than 1 g., and in all fish except no. i was lower in December than in November. From the beginning of January onwards the amount increased week by week with an almost unbroken regularity until by the beginning of June the fish were eating between 20 and 30 times as much as they were during the first week of December. All the fish save nos. 1 and 3 were then lost, but the data for the two survivors were continuous until the end of August. The amount of food eaten by fish 1 continued to increase until early in July when it reached 12 g. per week, but then it showed a marked decrease and remained approximately constant for the rest of the summer. The amount eaten by fish 3 was at its maximum (13·5 g.) in the middle of June. It then fell markedly, but there occurred a small increase in July, another fall early in August, and another increase towards the end of that month.

In October 1934 the experiment was restarted using fishes 1 and 3 of the original batch and 8A and 12A, newcomers to the tanks. During the first week of the experiment the food requirements of all these fish were fairly high, but in all of them the amount eaten was considerably lower in November and increased again in December. It fell off again from the middle to the end of January 1935, an increase occurred in February, but there was another period of low consumption in March, and then the increase was continued somewhat irregularly through April, May and early June. Fishes 1 and 3 died in June, and the marked decrease in the amount.eaten by no. 3 in the last fortnight of its life is probably of little significance from the point of view of normal feeding. Fish 8 A continued to eat large quantities (up to 30 g.) each week until early in August, when suddenly its requirements fell to less than half. In the autumn this fish again fed heartily, though during October its appetite was not quite so good as in either September or November.

From 6 June onwards tank 12 was occupied by a fish only a few months old. This young fish ate more food in proportion to its body weight than did the others, and in the week ending 11 July ate more than its own body weight, i.e. 1·8 g., a relative consumption which was never approached by any of the others.

It is clear from an inspection of Fig. 3 that there is no simple relation between the amount of food eaten and the amount of growth made in the same week, for although there are many cases where a large increase in weight accompanied a large food intake, there are also many where comparatively little growth was made although a large amount of food was eaten. It appears then that the efficiency of digestion varies greatly from week to week.

If, however, the figures be grouped into longer periods according to the temperature, as has been done in Table IV, then a more simple picture of the relation between food and growth is obtained, for the short period fluctuations are smoothed out. In Fig. 6 the data given in Table IV are illustrated graphically. Points have been plotted to show the amount of growth made (in mg./g. of starting weight per week) according to the amount of food eaten (mg./g. of mean body weight) over periods of approximately constant temperature. Dots represent observations made at temperatures below 50° F., circles those between 50 and 60° F., and crosses those at a temperature of over 60° F.

It will be seen that the dots are grouped reasonably evenly about the straight line A–B, indicating that at temperatures below 50° F., on the average, the growth made is roughly proportional to the amount of food eaten, and that $212$ units increase in the food eaten produced on the average one unit of extra growth. It will also be observed that the point at which A–B cuts the abscissa and which therefore indicates the amount of food required to keep the body weight constant, is 85 mg. of food per gram of body weight per week, a value which is in fair agreement with those given in Table VII. The circles, indicating the relation between food intake and growth at temperatures between 50 and 60° F., are very scattered and no curve can be drawn to which they approximate. It is obvious, however, that within this range it requires more food to produce a unit increase of body weight than at the lower temperature, but since generally speaking the fish ate more than when the water was colder, the amount of growth made was high. Some part of the lower efficiency of food utilization at this temperature is clearly due to the increased maintenance requirements of the fish, but from the irregularity of the points other factors, on which no information was obtained, are also involved. When the temperature rose above 60° F. there was a further marked decline in the efficiency of food conversion, and since this was accompanied by no increase, and often indeed a decrease, in appetite, the growth rate fell markedly. Here again, the points are irregularly distributed and no further deductions can be made from them. It is clear, however, that the check in the growth rate of all the fully fed fish used in the experiment during the summer (see p. 453) is due partly, if not entirely, to the low efficiency of food conversion, coupled in some cases with a decline in appetite. It may be noted in passing that the crosses in the right-hand top comer of the graph, which are exceptions to the statements made above, were given by fish 12B, and it has previously been stated that at this time this fish was growing very fast and eating proportionately far more food than any of the others.

It is perhaps worth while to examine the question of the efficiency of food conversion in somewhat more detail. An index of efficiency can be obtained simply by dividing the amount of food eaten in any period by the amount of growth made, and it then represents the amount of food required to produce a unit increase in the weight of the fish over that period. This may be called the “crude efficiency”. Since, however, a proportion of the food eaten is used for maintenance, and is not available for fresh growth, this figure is only valuable when a fish is growing reasonably fast, for when it is eating only a little more than its maintenance ration, and therefore growing slowly, very high crude efficiency figures are obtained. A better idea of the use made of the food is obtained if we subtract from the total amount of food eaten the amount estimated to be required for maintenance and then divide the amount available for growth by the amount of growth made. This figure will be called the “net efficiency”. It is true that the maintenance requirements for all individual fish at all times are not known, but we can take those given in Table VII as reasonable approximations and so obtain figures comparable with those given by Dawes (1930–1) for plaice feeding on mussels.

Since, as has already been explained, there were from week to week wide fluctuations in the relation between food eaten and growth made, in Table XII the grouped data from Table IV have been used instead of the detailed figures for food eaten and growth made per week.

Even with this grouping there is a good deal of variation, but of the 78 values of “net efficiency” calculated, 6 lie between 1 and less than 2, 22 between 2 and less than 3, 22 between 3 and less than 4, 17 between 4 and less than 5, and only 11 at 5·0 and over. The median is at 3·35 and the semi-interquartile range ±0·9. There is no regularity about the variations, and although of the 11 values greater than 5, 8 were recorded when the temperature was 50° F. or more, there is no correspondence between the more normal values and the temperature. The figures do show, however, that for brown trout, Gammarus is a highly nutritious food, for it may take less than 2 g. of food over the maintenance requirements to produce 1 g. increase in weight, and it very seldom takes more than 4 g. Dawes (1930-1), feeding plaice on mussels, found a similar inconstancy in “net efficiency” from one period to another to that here recorded, but his figures show that mussel is a less efficient food for plaice than Gammarus is for trout, for the highest “net efficiency” he records is 2–4, and the figures are more generally in the region of 5.

Surber (1935), in his experiments with brook and rainbow trout feeding on G. fasciatus, did not determine the maintenance requirements of his fish and gives therefore figures of “crude efficiency” only. He found that for brook trout it required 6·05 g. (wet weight) of food to produce 1 g. increase of weight and for rainbow trout 6·63 g. These figures show good agreement with many of the values for the “crude efficiency” of brown trout feeding on G.pulex given in Table XII.

During the course of this experiment, whenever the quantity of Gammarus was reasonably small, the amount given to the fish was counted as well as weighed. The method of weighing has already been described, and although it was not exact, as they were weighed alive with an indeterminate amount of water adhering to them, it is possible to make an estimate of the average weight and so express the amount of food eaten in numbers as well as weight.

In all 31,763 Gammarus were counted and weighed and the mean weight was 0·026 g. As the shrimps were generally weighed several at a time, and only rarely individually, it is impossible to give the actual size range of those used, for it happened that all those that were weighed one at a time were all very near the mean weight. The smallest average weight for any group was 0·008 g. and the largest 0·067 g

It is probable that the average weight of 0·026 g. is lower than the actual average weight of the Gammarus population of the River Itchen at Alresford, for a deliberate attempt was made to feed small fish on small Gammarus, and as it was generally for the small fish that the rations were counted as well as weighed (for when the fish became larger and could take fully grown shrimps their food requirements were too high for numbers to be counted) it is certain that an undue proportion of small Gammarus were included in the 31,763.

Since, however, it is reasonable to suppose that the smaller and younger fish prey principally on the smaller shrimps, for a full-grown Gammarus is rather a large mouthful for a fish less than 10 gr. in weight, this average of 0·026 g. can be used without serious error to calculate the numbers of shrimps eaten by the fish to produce a given increase in weight.

In Table XIII are shown, for various periods, for which the data are available the increase of weight of the fish, the total amount of food eaten, the estimated number of Gammarus this quantity represents and the average number of shrimps required to produce 1 g. increase in weight of the fish.

It will be seen from Table XIII that, although the numbers vary considerably, in order to increase its weight by 1 g. a brown trout, generally speaking, ate between 200 and 300 Gammarus. Surber (1935) estimated the average live weight of each individual G. fasciatus to be 0·017 g., and the average number required to produce 1 g. increase of weight to be 356 for brook trout and 390 for rainbow trout.

These fish were, of course, restricted to a diet of Gammarus, and as, under wild conditions, other foods are available, a fish in the river will not eat so many shrimps for the same increase in weight. Yet in the Itchen at Alresford, Gammarus is the dominant invertebrate, and as it has been repeatedly shown (Neill, 1938; Slack, 1934; Pentelow, 1932) that trout feed on the most easily obtainable food, there is little doubt that taking all seasons óf the year Gammarus will comprise at least half the food eaten. Therefore, as a general statement, it may be said that for every gram increase of weight of the trout in the Itchen at Alresford at least 100 Gammarus are consumed. These figures may, in the future, be useful in assessing the stock of trout that a given length of stream can profitably hold.

1. The growth of brown trout (Salmo trutta), fed on Gammarus pulex, in their first and second years has been studied.

2. The growth in weight varies considerably from week to week but, generally speaking, it increases with increasing size of the fish. It is assumed that in these experiments the second point of inflection of the normal S-shaped growth curve was not reached because the fish were too young.

3. In all the fishes studied there was a period of slow growth during the winter and during the summer. Growth is at its maximum at temperatures between 50 and 60· F.

4. By careful adjustment of the rations it was possible to keep the body weight of the fish approximately constant from week to week. The amount of food required for this purpose varied from 51 to 270 mg./g. of body weight per week, but was mainly between 70 and 102 mg. and was apparently affected by the water temperature, being higher when the water was warmer.

5. Starved fish lost more weight at higher temperatures than at lower, but the loss of weight could not be related to the amount of food required to maintain the body weight constant at a given temperature.

6. The appetite of fully fed fish increases as the temperature rises to 60° F. but generally decfines at temperatures higher than this. Between 40 and 50° F. the amount of growth made is roughly directly proportional to the amount of food eaten, but above 50° no such simple relation exists.

7. G. pulex is a very efficient food for trout; generally speaking about 5 g. of this food produce 1 g. increase in weight. If from this amount the quantity required to maintain the body weight constant is subtracted, it is found that 1 g. increase in weight is produced by about 3 g. of food available for growth.

8. The average weight of the Gammarus used as food in this experiment was 0·026 g., and it is estimated that for every gram increase of weight each fish consumed between 200 and 300 Gammarus.

The author wishes to express his very sincere thanks to Dr E. S. Russell, Director of Fishery Investigations, Ministry of Agriculture and Fisheries, for his constant interest and advice in this work, to Mr T. Edser and the staff of the Statistical Branch of the Ministry for invaluable help in the treatment of the data, to the Department for permission to publish these results, and to his colleagues Dr J. Grindley and Mr H. Stokes for their assistance in the preparation of the diagrams.