The Growth of Brown Trout (Salmo Trutta Linn.): II. The Growth of Two-Year-Old Trout at a Constant Temperature of 11·5°C.

The investigation now to be described was an attempt to study the growth of individual trout when the environmental factors were as completely controlled as possible. Observations on the early growth of trout fry under such conditions have already been reported (Brown, 1946), but these are very delicate and are difficult to identify as individuals in the early stages, so 2-year-old trout were used for a more detailed investigation, especially of the relation between food and growth in different environments. The present paper is an account of observations on trout living at a constant temperature of 11·5° C., and experiments concerned with the effect of temperature on growth will be reported later.

The following factors were controlled or recorded throughout: temperature, amount and intensity of illumination, rate of water flow, chemical composition and aeration of the water, quality and quantity of food, volume of living space and number of individuals in each group. The following were investigated : the growth of individuals in groups, the effect of crowding on growth, changes in growth rate with age, the relation between food and growth and the effect of duration of light on growth. The number of individuals studied was limited by the space and time available, but the observations are published since some of the results were unexpected.

The brown trout used in these experiments were obtained from the Midland Fishery, Nailsworth, Gloucestershire. They had been bred at the hatchery, and each consignment was all of the same age and of similar sizes, so that they had grown at the same average rates during the 18 months they had lived in the hatchery.

The fish were kept in tanks in a constant-temperature room which was set at 11·5° C., this temperature being chosen because it lies within the optimum range postulated by Pentelow (1939) and Wingfield (1940). The tanks were of glass, with angle-iron edges and slate floors, and they were arranged in two tiers, one at bench level and the other on the floor. The water supply came from a constant-level tank on the roof of the room, so that it was under constant pressure. On entering the room it flowed through a coil of metal tubing in the air and then through another coil immersed in standing water. This reduced its temperature to that of the room. The temperature was recorded by a thermograph in one of the tanks, and it seldom varied more than 0·2° from 11·5°C.

A T-junction distributed the water to each tier of tanks, and glass siphons between the tanks and an overflow device from the last one in each series maintained a constant level. The rate of flow was adjusted so that the water never became foul in any of the tanks, and it was then maintained constant. The siphons between the tanks were enclosed in cylinders of fine wire gauze so that particles of food should not be swept from one into the next, and certain tanks were divided by closely fitting glass plates into two spaces of known volume.

Cambridge tap water, an analysis of which was given by Brown ( 1946), was used throughout. All the tanks were supplied with compressed air, which bubbled through continuously and kept the water stirred. It was adjusted so that the water movement was not sufficiently violent to disturb the fish, while keeping the whole volume well aerated. The room was blacked out and the tanks were lit by 40 W. electric bulbs suspended 36 cm. above the water and arranged so that there was one for each 20 sq.cm, of water surface. The lamps were controlled by a clockwork time switch, and most of the tanks received 12 hr. of light and 12 hr. of darkness each day. One tank was surrounded with cardboard shields and a curtain of black material and its lamps were controlled by a second time switch, so that the fish in it could be allowed different amounts of light of the same intensity as that in the rest of the room.

The tanks did not all contain the same volume of water, and it was arranged that tanks of equal volumes contained different numbers of fish while tanks of different volumes contained equal numbers of fish, thus comparing the effects of the total amount of living space and of the amount of living space per individual on the growth of trout. Two of the tanks received water direct from the supply, while the rest received water siphoned over from another tank. The inhabitants of tanks of the same volume, but receiving fresh and conditioned water, were exchanged at intervals, so that any effect of ‘conditioning’ might be detected, though no effect was observed. All the tanks were covered with glass plates to prevent the fish from leaping out and the water from being contaminated with dust.

The fish were fed every day except Sundays and were given coarsely minced meat, from which the fat and gristle had been removed, mixed with about one-half its volume of minced liver. From a dish of this mixture, rations were weighed out for each tank and a sample was set aside for dry-weight determination. In order that the water should not become foul, the amount added to each tank was adjusted so that nearly all of it was eaten by the fish. Three times a week, as much as possible of the excess food and faeces in each tank was siphoned out, filtered with a Büchner funnel, and dried in an air oven at 80° C. for at least 20 hr. The dry weight of the excess food and faeces, subtracted from the dry weight of food added to the tank, gave the amount which had been absorbed by the fish. This amount is a maximum estimate, since some was probably lost by solution in the water and some may have been swept away through the gauze gratings. The food consumption has been expressed as milligrams of ‘standard meat’ per gram of fish per week, the percentage dry weight of the standard meat being 22·5.

The fish were generally weighed and measured once a fortnight, and they were starved on the previous day. Each was held with a damp cloth on a measuring board, and the length between the tip of the nose and the fork of the tail was read to the nearest millimetre. Each fish was then transferred to a waxed wooden box containing some damp cottonwool. One box measured 20 × 5 × 5 cm. and another 30 × 5 × 5 cm., and each had a waxed wooden lid which could be held firmly in position with a rubber band. The box containing the fish was weighed to the nearest tenth of a gram, and the fish was then placed in water with ice floating in it. The box was weighed again and the weight of the fish was obtained by subtraction. The process of weighing and measuring occupied about 3 min. for each individual, and trout did not appear to suffer from the treatment if they were returned to cool, well-oxygenated water.

The individual trout differed among themselves in the shape of such parts as the operculum and in the number and arrangement of spots on the body. It was therefore possible to recognize individuals by their appearance and to keep individual records of their length and weight. Fish were kept for at least 6 weeks under the experimental conditions before observations on their growth were begun.

Most of the trout remained healthy, though a few suffered from disease. When a fish died, it was dissected to determine its sex and the state of its gonads. It was then either dried in the air oven at 80° C. in order to determine its dry weight, or was preserved in 4 % formalin solution. Some individuals were examined thoroughly for parasites, from which they were remarkably free. Samples of scales were taken from all the dead fish, and have been preserved for future examination.

The individuals in each experimental group belonged to the same consignment from the hatchery and were of uniform size when the observations began. They had been living in the same conditions at the hatchery and were the same age, so that they had grown at the same average rates before the experiments began. Soon afterwards, however, the dispersion of individual weights within each group increased, the large fish becoming much larger than the smaller ones.

In Fig. 1 the fish in each group are arranged in order of decreasing weight, and their positions in this order are shown at intervals of 3 months. The largest fish generally remain at the top of these orders and the smallest ones at the bottom, but there are interchanges of position among the intermediate fish.

In Fig. 2 individual specific growth rates have been plotted against individual positions in the size orders. Fish with high specific growth rates occur in positions 1–2, 4–5–6 and 9–10–11 in the orders of decreasing weight, while those in positions 3–4 and 7–8 grew at low rates. These maxima and minima are variable, but can be seen in the records for all groups of six or more fish. In these groups the fish at the head of the order is generally growing at a maximum rate and that at the bottom at a minimum rate, so that they remain in these positions. The fish in positions 5 and 10 approximately have high rates and should grow larger than those in positions 4 and 9. In the next 3-monthly period, however, the latter will have the higher specific rates and should regain their original positions in the order. Deviation from this pattern of specific growth-rate distribution among the members of a group was not uncommon.

The tank space in the constant-temperature room was divided up so that fish might be grown with the following amounts of living space per individual: 3, 5, 12, 23, 35 and 50 1. The number of fish which could be grown with the larger amounts of living space was necessarily small, and it was difficult to maintain these experiments because the trout frequently developed ‘tail-rot’ and died, so the conclusions about growth under these conditions are very tentative. The more crowded trout seemed to be much more resistant to disease, but had a high mortality on the few occasions when the compressed air supply failed, while the fish with intermediate amounts of living space survived. All the tanks received the same water supply, the same amount of aeration and illumination, and were at the same temperature. All were supplied with more food than the fish would eat. They thus differed only in amount of living space and number of individuals.

In all the experiments there were considerable differences between individual specific growth rates, and these were associated with individual position in the size orders. When the averages for each tank are compared over 3-monthly periods, there appears to be an optimum amount of living space per fish for rapid growth, and the average specific rates decrease for volumes greater and less than this amount (Fig. 3). The standard deviations from the means are large compared with the mean values and the numbers of fish are not large, so the differences between the means are not statistically significant. Comparison of individual specific growth rates, however, shows that the values for fish in the most crowded tank were consistently lower than those for fish in the same position in the size order, but with more living space, while the few fish with large amounts of living space generally grew at lower rates than the corresponding more crowded fish (Fig. 2).

This effect of living space was not merely a correlation between the number of fish in a group and the average specific growth rate, since equal numbers of fish were kept with 5 and 23 1. per fish and approximately equal numbers with 3 and 12 1. each. There was also no obvious correlation between the total volumes of the tanks and the growth of the fish, since the two most crowded tanks each held 40 1., those with 12 and 35 1. per fish held 120 1., and the other two, with 23 and 50 1. per fish, held 160 and 150 1. respectively. Thus the effect depended on the degree of crowding (i.e. the amount of space per individual).

There were considerable differences between the specific growth rates of individuals in the same tank during each fortnight, but the average monthly specific rates of all the fish show an annual cycle of variation. For each tank, the standard deviation from the mean rate remained fairly constant throughout the year, and the growth rates of all the individuals varied in the same way.

The mean specific growth rates calculated from the weight and length data for the ten fish with a living space of 12 1. per fish are shown in Fig. 4. For weight the specific growth rate fell to a low value in the autumn of 1942. The fish became 2 years old in December 1943, and the growth rate rose to a maximum in February 1943 and remained high till the late summer. It then fell markedly, and the behaviour of the fish changed at the same time. Instead of remaining inactive except when being fed, they became very restless and chased each other round the tank. In September, several of the larger fish died, apparently because of asphyxia, and on dissection they were found to be nearly sexually mature. The survivors were combined with the survivors in an adjacent tank. These were the same age and had had a similar growth history. Their growth rate continued to be low, and they were stripped of ova and milt in November 1943. They became 3 years old in December 1943, and the growth rate rose again in January 1944.

For length the specific growth rate followed the same cycle. There was a minimum in the autumn of 1942, a high rate from January to March 1943, and the rate decreased steadily to a minimum in the autumn of 1943 and rose again in January 1944.

This growth-rate cycle, with a low specific rate in the autumn, a maximum in early spring, a high summer rate and a decrease in the next autumn, accompanied by maturation of the gonads, occurred in all the groups of trout at 11 ·5° C., and the average rates for all the groups varied in the same way at the same times of year. During the whole of this period, however, there was no variation in any physicochemical factor and the fish were allowed to eat as much as they would. This cycle must therefore reflect physiological changes in the individual trout. The onset of sexual maturity, which was accompanied by depression of the specific growth rate, occurred at the same time as it would have occurred had the fish been living in a hatchery. There was no correlation between the size of the fish and maturation of the gonads, since all became mature at the same time although they differed considerably in size.

Thus, 2-year-old trout, living in constant conditions of temperature, illumination, food supply, rate of water flow, and aeration and chemical composition of the water, had an annual growth-rate cycle, with slow growth in the autumn and rapid growth in early spring, decreasing during the summer to a ‘check’ in the autumn. They became sexually mature when they were just 3 years old.

In all the tanks at 11 ·5° C. most of the fish showed an alternation of periods of rapid and slow growth in weight and similar, though less well-marked, fluctuations in the rate of growth in length. For weight a period of rapid specific growth rate was followed by one of slow growth, and this was followed by another period of rapid growth, each period being about 2 weeks long. The closer the specific growth rate was to the average value (about 3 –3 ·5 % weight per week), the less marked was the fluctuation. The specific rate of growth in length for a given period generally showed a direct correlation with the specific rate of growth in weight during the preceding fortnight. Thus, fluctuations in the rate of growth in length alternated with those in the rate of growth in weight.

A picture of the growth of an individual trout may be obtained by combining these correlation curves. Table 1 shows growth-rate values which have been calculated for fish weighing 50 g. with different condition factors. The relation between the specific rates for growth in weight and in length is such that for fish with condition factors below about 1 ·8 the factor increases after a fortnight of growth, while there is a decrease in condition factor for fish with an original factor above 1 ·08. The greater the divergence of the condition factor from 1 ·08, the greater is the change in value towards this figure. The fluctuations in the specific rates of growth in weight and in length are thus more marked, the more the condition factor differs from 1 ·08.

Katabolic changes are always occurring in living cells, so that a certain amount of food must be utilized to replace the energy dissipated and the protoplasm broken down. Starved animals lose Weight, and an animal must eat a definite amount of food, the ‘maintenance requirement’ or ‘maintenance ration’, to avoid loss of weight. Since it is necessary to know the maintenance requirement for a given set of environmental conditions before the efficiency of growth can be calculated, and since the maintenance requirement may change as an animal grows, experiments were designed to determine the relationship between it and the body weight of trout.

A group of ten fish were kept in an ‘unconditioned’ tank in the constant-temperature room, and were weighed once a week. The change in their average weight was expressed as a proportion of their original weight (gain or loss as milligrams per gram of fish). The weight of meat absorbed was also measured and expressed as milligrams per gram of fish, and the two quantities were plotted against each other. At the ‘maintenance level’ there was a direct relationship between the amount of food absorbed and the change in weight, so the total amount of food added to the tank was reduced until at least three consecutive values showed this relationship. The value for amount of food absorbed for which no change in weight occurred was then taken as the maintenance requirement.

As the rations were reduced towards the maintenance level, there was an increase in the efficiency of growth, but when the maintenance level had been reached, there was no further change in efficiency. Only the last three or four in each series of values have therefore been used to calculate the requirement. In one experiment, with fish of average weight 111 ·7 g., rations were kept at the maintenance level for eight consecutive weeks, and there was no change in efficiency. Fig. 10 A is a typical series of values giving a maintenance requirement.

After determination of a maintenance requirement, the fish were fed at a higher level until the average weight had increased to a new value. The amount of food was then reduced until another maintenance requirement could be calculated. At the maintenance level, all the fish in the tank gained or lost weight simultaneously, so it is assumed that each received a fair share of the food. During the intervals with higher rations some grew faster than the others, but all increased in weight.

The maintenance requirement decreased, relatively, as the fish increased in weight, reaching a constant level for about 100 g. average weight (Fig. 6). An unexpectedly low value was obtained in November to December 1943, with fish which were just 3 years old and must have been nearly sexually mature. After 1 month of increased food, two of them were stripped of ova (about 15 % of the body weight) and three of milt (about 5 % of the body weight). During two subsequent maintenance periods their requirements were at the higher level which had been expected from the earlier values. It seems likely that these fish had well-developed gonads, which had grown in size during the preceding growth period, and these were partially resorbed when the food rations were reduced.

The total amount of food absorbed by the fish in each tank was measured, and the values have been divided by the total weight of fish in the tank so that they are expressed as milligrams of meat per gram of fish per week. The amount absorbed by ten fish in 120 l. of water is shown in Fig. 7. The amount varied from week to week but decreased as the fish grew older and larger, and was lowest, on the average, when the fish were aged 150 to 160 weeks, and had fully developed gonads.

Maintenance requirements decreased, relatively, as the fish grew larger, and the total amount of food absorbed must have been used for maintenance as well as for growth. The amount available for growth may be found by subtracting the maintenance requirement for fish of that average weight from the total amount absorbed during each week. This quantity varied from week to week (Fig. 7), but fluctuated round a constant level for fish aged between 95 and 140 weeks.There was no apparent correlation between it and the average weight of the fish.

The amount of food eaten in excess of the maintenance requirements has been correlated with the condition factor of the average fish. The relation (Fig. 8) is similar to that between the specific rate of growth in weight and the condition factor, but with the maximum at 1 ·12 instead of 1 ·08. Unfortunately, there are no values for fish with an average condition factor less than 0 ·95.

The change in average weight (expressed as milligrams per gram of fish per week) was roughly proportional to the amount of food eaten in excess of the maintenance requirements (Fig. 7), but that this relationship is not direct can be shown by calculating the ‘efficiency of utilization’ of the food. This is defined here as the increase in average weight (as milligrams per gram fish per week) per unit amount of food eaten in excess of the maintenance requirement (also as milligrams per gram fish per week). It may also be called’efficiency of conversion of food into fish ‘or’ efficiency of growth ‘, and it was not constant from week to week.

There is a strong positive correlation (r = 0 ·87) between the efficiency and the amount of increase in average weight (Table 2). Larger increases in weight were more efficient than smaller increases, so that the former required relatively less food. Since rapid growth in weight was associated with high efficiency, the curve correlating efficiency with the condition factor resembles that correlating the specific rate of growth in weight with the condition factor. There is a maximum at about K= 1 ·12 (Fig. 8), and again there are no values for fish with an average condition factor less than 0 ·95.

These observations are based on the feeding and growth of the average fish, and they support the conclusions derived from growth records for individual trout, that short-period fluctuations in the rates of growth in weight and in length are related to the condition factor. The specific rate of growth in length varies directly with this factor. The amount of food eaten in excess of the maintenance requirements and the efficiency with which this food is used for growth are both high for a condition factor of about 1 ·10, and growth in weight is most rapid for fish with approximately this condition factor. With higher and lower condition factors, the amount of food eaten and the efficiency are lower, and so is the specific rate of growth in weight. Unfortunately, there are no figures for the food eaten and efficiency of fish with an average condition factor less than 0 ·95, but, by analogy with the specific rate of growth in weight, they should be high if the condition factor is less than 0 ·90.

Assuming that maintenance requirements vary with the weight of the fish and are independent of the amount of living space, the food eaten in excess of these requirements has been calculated for each tank in the constant-temperature room (Table 3). For each group of fish, the values were independent of the average weights and fluctuated about a mean. There were no marked differences in the amount of food eaten by fish with amounts of living space between 5 and 23 l. per fish ; those with 3 l. per fish ate less food and showed less variation in the amounts eaten from week to week; while those with 50 l. per fish showed marked fluctuations in the amount eaten, which in some weeks was little more than their maintenance requirements. The average efficiency of utilization of the food was low in the more crowded tanks compared with fish with greater amounts of living space (Table 3).

The degree of crowding thus has little, if any, effect on the feeding of fish with moderate amounts of living space. Very crowded fish show less variation in the amount of food eaten, eat less food and use it less efficiently, while those with large amounts of living space show large fluctuations in the amount eaten and in efficiency. These observations on feeding thus support the conclusion that over- and undercrowding both lead to reduction in the average specific growth rate, so that there is an optimum population density for rapid growth.

In order to investigate more thoroughly the relation between the amount of food eaten and the change in body weight, seven fish in 40 l. of water were given amounts of food which were more than their maintenance requirements but less than the amounts which they would have eaten if not restricted. It was intended that their rations should be kept at a constant level above the maintenance requirement, but fluctuations in the dry weight of the food made this impossible. The fish were weighed and measured once a week, and they all survived until the end of the experiment. They all gained and lost weight in the same way, so it is assumed that they each obtained a fair share of the food. Their average condition factor varied between 0 ·95 and 1 ·00.

Fig. 9 shows the amount of food that was added to the tank and the amount that was actually absorbed by the fish (both expressed as milligrams per gram fish per week). The latter quantity generally varied directly as the amount added, but in some weeks, e.g. when aged 115, 122, 127 –130, 136 and 139 weeks, the fish did not take full advantage of their food supply.

When the maintenance requirements are subtracted from the amount of food eaten, to give the amount available for growth, this quantity shows considerable variation from week to week, but it was never as high as the average amount available to fish with unrestricted rations.

The weekly change in average weight is also shown in Fig. 9; it was very variable, a period of rapid increase being followed by a period in which there was no gain and sometimes even loss of weight. The peaks in the growth curve generally coincided with peaks in the curve for food eaten in excess of the maintenance requirement, while weeks in which little food was eaten generally followed periods of rapid increase in weight. The efficiency of utilization of the food (Fig. 9) was generally at a maximum value after a period of poor growth and low food intake, and these maxima preceded maxima for change in weight and food eaten in excess of the maintenance requirements.

Table 2 shows that, for changes in weight greater than 20 mg. per g. fish per week, the efficiency of growth was markedly higher than that of fish of the same age with unrestricted rations.

Thus 2-year-old trout with restricted rations showed cycles of growth and rest. A period of little growth in weight and small amount of food eaten was followed by one of increased food intake associated with high efficiency. The latter fell as the food intake increased, and the change in weight rose to a maximum with the amount of food eaten. The peak of rapid growth, which lasted for 1 or 2 weeks, was followed by decrease in food intake and consequent poor growth. The cycle was then repeated, taking between 6 and 8 weeks. Growth in length occurred alternately with growth in weight, as with an unrestricted food supply.

When growth in weight occurred, the efficiency was greater, so that the fish with a restricted food supply attained a size which was not very much smaller than that of fish of the same age with unrestricted rations. The former grew spasmodically, possibly by exaggeration of the short-period fluctuations in growth rate, but their rapid growth periods compensated for the intervening periods of little or no growth.

As the amount of food added to the tank was reduced during each determination of a maintenance requirement, the fish first lost weight, then became ‘adapted’ to the smaller amount of food and gained weight. If the amount of food was reduced further, they again lost weight for a week, then became adapted once more and gained weight. When the maintenance level was reached, however, the change in weight was directly proportional to the amount of food absorbed, and this relationship remained constant.

This ‘adaptation’ to reduced levels of feeding was observed during every determination of maintenance requirements, and a typical record is given in Fig 10 A.

Fig. 10 B shows the effect of increasing the rations of fish which had lived for 6 weeks at the maintenance level. At first they were very efficient, but they soon became adapted to the higher level of feeding, and a slight reduction in the amount of food then led to loss of weight, although the amount eaten was greater than the maintenance requirement. After another week at this lower level, however, they became adapted to it, and an increase in the amount of food again resulted in rapid increase in weight. During the following week, with the same amount of food, there was less than half as much increase in weight, the efficiency having again decreased.

Trout therefore use their food more efficiently when the amount is restricted, and they become adapted to different levels of feeding in 1 or 2 weeks. At the maintenance level, growth in weight is directly proportional to the amount of food absorbed, but at higher levels there are cycles of alternating rapid and slow growth.

One 120 l. tank in the constant-temperature room was enclosed with cardboard screens and a black curtain so that it was shielded from the lights in the rest of the room. The two electric bulbs above the tank were connected to a separate time switch, and the six trout in the tank were grown for 5 months with 18 hr. of light per diem ; the amount of light was reduced at the rate of 1 hr. per diem per week, and they were then grown for 2 months with 6 hr. of light per diem. They were allowed more food than they would eat, so that their growth should be directly comparable with that of fish of the same age in the other tanks.

The mean specific growth rates, calculated from the weight data, for fish of the same age with 6 and 12 and with 12 and 18 hr. of light per diem, are given in Table 4. There was no difference between the specific growth rates of fish with 12 and 18 hr. per diem, except that there was slightly less individual variation with the greater amount. Fish with only 6 hr. of light per diem, however, grew significantly better than those of the same age with 12 hr. (t = 3·1, P = 0·01).

These rather limited observations suggest that when trout are given as much food as they will eat once a day, growth is more rapid when the total duration of illumination is less than 12 hr. per diem, but increase in the duration above this amount does not result in further depression of the growth rate. No experiments were performed with lights of different intensities.

In the first paper of this series (Brown, 1946) it was shown that the factor which had the greatest influence on the early growth of individual trout fry in aquaria was the size of a fish relative to that of the others in the same group. The larger individuals grew faster, and if they were removed the smaller individuals showed increase in their specific growth rates. Among groups of 2-year-old trout, also, the relative size of an individual had an effect on its specific rate of growth. In groups of fry, an individual’s growth rate was proportional to its position in the hierarchy of decreasing weight, but for the older fish the position was more complex. The largest and smallest individuals in each group generally retained these positions in the size order throughout the experiments, but there were exchanges in position among the intermediate fish, occurring about every 3 months. The individuals appeared to form ‘subgroups’ of three, four or five fish of similar weights, and within each subgroup the largest fish grew fastest and the smallest most slowly. After about 3 months the largest individual of one subgroup would be larger than the smaller ones of the next larger subgroup, and there was a rearrangement of hierarchies. There were, however, many exceptions to this pattern of growth rate distribution.

The older fish, like the fry, were allowed to eat as much as they would, and the smaller individuals appeared to eat as readily as the larger ones. Thus, competition for a limited food supply cannot be invoked to explain the size hierarchy effect. The older trout, like the fry, soon developed the habit of resting in definite positions in their tanks, and if new fish were added to the tank and came to rest in one of these positions, the ‘owner’ would chase it away. Certain individuals were more aggressive and would chase the others round their tanks. Thus, within each group there were individual differences in behaviour, and at least some fish appeared to react to others as towards recognizable individuals. Every trout seemed to acquire certain territorial rights, and it is not inconceivable that individuals of approximately the same size formed subgroups within the larger group. Dominance-sub-ordination relations within the subgroups may then have determined the growth rates of the individuals. Such an effect of social dominance on growth rate was reported by Noble (1939), who also found that stable hierarchies could last for 2 or 3 weeks among Xiphophorus, but in these cases all the fish in the tank were involved in a single group.

Anglers often find that certain stations in rivers and lakes are occupied by large trout. If one of these is caught its position is taken, sometimes within an hour, by another fish, generally slightly smaller. There is thus evidence that there is a ‘size hierarchy’ in natural conditions, and that larger trout occupy more favourable parts of rivers and lakes.

The degree of crowding affects the level of growth rate within the group. Overcrowded fish eat less, use the food less efficiently and grow more slowly than fish with more living space. Those with a large amount of individual living space, however, are very susceptible to disease and grow and feed erratically and not so well, on the average, as those with intermediate amounts. The fish were exchanged at intervals between tanks with ‘conditioned’ and ‘unconditioned’ water, and there appeared to be no correlation between the type of water and the growth rate. The existence of an optimum degree of crowding for rapid growth therefore cannot be explained by the theory of chemical conditioning of the medium put forward by Allee (1939). There is also no evidence that the actual volume of the tanks limited the growth of the fish, since some of the larger fish grew in the smaller tanks, and vice versa.

Welty (1934) demonstrated that individuals of the genera Notropis, Macropodus, Brachydanio and Carassius ate more food when grouped than when isolated, the stimulus for extra feeding being the sight of other individuals feeding. Among the 2-year-old trout, those with the greatest amount of living space always rested far apart in the tank, and there was no evidence of mutual relations. The most crowded fish were continually disturbing each other and had a generally lower level of appetite and efficiency. Guhl & Allee (1944) found that a flock of hens in which there was continuous upheaval of the social structure ate less and did not grow so well as one in which there was a stable hierarchy, and it is possible that the subgroups were less stable in the more crowded tanks of trout than in those with intermediate amounts of living space per fish. In the latter, appetite and efficiency were high; there was probably little mechanical disturbance; there may have been mutual visual stimulation ; and the hierarchies were probably stable.

As trout fry grow older, their specific growth rates decrease (Brown, 1946). The average growth rates of the 2-year-old trout were markedly lower than those of the fry, in accordance with the principle propounded by Minot (1890) that the acceleration of specific growth rate is negative. It might be expected that the decrease in rate would be gradual and continuous under constant environmental conditions. All the trout at 11 ·5° C., however, had an annual cycle of growth with an autumn check when they became 2 years old, rapid growth in the next spring, slower summer growth and another autumn check which coincided with maturation of the gonads as they became 3 years old. This cycle was presumably the result of an annual physiological cycle of changes in the internal environment, possibly the variation of secretion of an endocrine organ such as the pituitary gland. Hasler, Meyer & Field (1939) showed that extract of carp pituitary can induce premature spawning in rainbow trout, and pituitary extracts affect growth and reproduction in other vertebrates.

Since this cycle occurred in the absence of variation of any environmental factor it cannot depend on the existence of environmental ‘time-markers’. These have been shown to be important in the maturation of the gonads of Salvelinus fontinalis (Hoover & Hubbard, 1937) and Phoxinus laevis (Bullough, 1939, 1940). The brown trout became sexually mature at the same age and at the same season as those of the same stock under natural conditions. Maturation of the gonads is generally followed by decrease in the growth rate (Fage & Veillet, 1938) and is therefore an important event in the growth history of fish. There was no apparent difference between the growth rates of 2-year-old male and female trout.

Within this annual cycle of growth, individuals showed short-period fluctuations in specific rate taking between 4 and 6 weeks and correlated with the condition factor. Trout in ‘good’ condition are heavier relative to their length than those in ‘poor’ condition, and probably contain more food reserves. Growth in length, which involves regional differentiation, occurs at a rate directly proportional to the condition factor and thus to the amount of reserve food. Fish in very good condition eat little food, while those in very poor condition probably eat plenty, i.e. appetite is low for fish with food reserves and high for those with none. The amount of food eaten is also high for trout in medium condition and low for those in poor condition. The efficiency of utilization of the food and the specific rate of growth in weight vary with the condition factor in the same way as the amount of food eaten. The balance between condition factor, growth in weight and growth in length is such that there are alternating periods of slow and rapid growth; rapid growth in length alternates with rapid growth in weight, and the condition factor tends towards a value of approximately 1 ·08.

When the level of feeding is reduced and the fish are allowed amounts between the maintenance requirement and the amount they would eat with unrestricted rations, the short-period fluctuations in specific rate of growth are exaggerated. Since the efficiency of utilization of the food is higher, on the average, these fish reach a size not very much smaller than that of trout of the same age with unrestricted rations. These observations agree with those of McCay & Tunison (1935, 1937) on Salvelinus fontinalis, for which the efficiency was inversely proportional to the level of feeding. When this level was altered, the brown trout became ‘adapted’ to the new level in 1 or 2 weeks.This was observed also by McCay, Dilley & Crowell (1929) using Salvelinus and by Moore (1941) using Apomotis, Helioperca and Perca flavescens.

At the maintenance level, the relation between the amount of food absorbed and the change in weight remained constant. The maintenance requirement decreased relatively with increase in weight of the fish, reaching a constant level at about 100 g. body weight. Pentelow (1939) estimated the maintenance requirements and efficiency of utilization of trout of average weight at about 20 g., but since most of his determinations were at temperatures between 45 and 50° F., his values are not comparable with those given in this paper. He found no correlation between maintenance requirements and body weight.

Surber (1935) estimated that individual Salmo gairdnerii of average weight about 50 g. required 6 ·63 g. of Gammarus at 12° C. for an increase in weight of 1 ·0 g. This amount of food included the maintenance requirement, which was not estimated, and the value calculated for Salmo trutta under the same conditions is 6 ·3 g. of meat for an increase in weight of 1 ·0g., if allowance is made for the rate of growth. Dawes (1930) investigated the maintenance and growth requirements of plaice fed with Mytilus, and his values are of the same order as those for Salmo trutta. He found also that the plaice were more efficient at lower levels of feeding.

Wingfield (1940) remarked that in temperate zones the annual variation in temperature is accompanied by annual variation in the amount and intensity of light, but there has been little attempt to separate the effect of these two factors on growth. The rather limited experimental evidence reported in this paper showed that the growth of 2-year-old trout was more rapid with less than 12 hr. of the standard illumination per diem. No experiments were performed with different intensities of light, and it is possible that there is an optimum amount (duration × intensity) for growth. Light might affect the growth rate through the pituitary gland, which is known to mediate melanophore responses in salmonid fry (Neill, 1940).

Under natural conditions the duration of light may affect growth directly, as in these experiments, but trout feed by sight, so it will also affect the amount of time during which the fish may capture food. If the food supply requires much search, the duration of light might be a factor reducing the growth rate by limiting the amount of food which could be eaten each day. Theoretically, therefore, there should be an optimum length of day, depending on the ‘availability’ of food organisms, such that the duration of light is sufficiently long for feeding but short enough for growth not to be inhibited.

1. Two-year-old trout were grown in environments where the following factors were controlled: temperature, amount and intensity of illumination, rate of flow, composition and aeration of the water, quality and quantity of food and amount of living space.

2. The specific growth rate of an individual depended on its size relative to that of the others in the group. It is suggested that subgroups of four or five individuals existed within the size hierarchy and were reorganized at intervals of about 3 months.

3. There was an optimum degree of crowding for rapid growth, and overcrowding led to lower appetite and efficiency of utilization of food, while undercrowded trout ate and grew erratically.

4. In spite of constant environmental conditions, all the fish had an annual growth-rate cycle, with an autumn check, a spring maximum, rapid summer growth and another autumn check, which coincided with maturation of the gonads when they became 3 years old.

5. Individual specific growth rates fluctuated over periods of 4 –6 weeks, and rapid growth in length alternated with rapid growth in weight. The specific rate of growth in length was directly proportional to the condition factor. The amount of food eaten, the efficiency of utilization of food and the specific rate of growth in weight varied with the condition factor and were maximal for a factor of about 1 ·10.

6. The growth-rate fluctuations were exaggerated and the efficiency was greater when the food supply was restricted. At the maintenance level the change in weight was directly proportional to the amount of food eaten. The maintenance requirement decreased, relatively, with increase in body weight.

7. The mean specific growth rate was higher with less than 12 hr. per diem of the standard illumination.