Physiology & Behavior, Vol. 16, pp. 613-615. Pergamon Press and Brain Research Publ., 1976. Printed in the U.S.A.
Sleep in a Schooling Fish, Tilapia mossambica C. M. SHAPIRO AND H. R. HEPBURN
Department o f Physiology, University o f the lqitwatersrand Johannesburg, Republic o f South Africa (Received 28 July 1975) SHAPIRO, C. M. AND H. R. HEPBURN. Sleep in a schooling fish, Tilapia mossambica. PHYSIOL. BEHAV. 16 (5) 613-615, 1976. - There are only a few infra-mammalian studies on sleep. For this reason, behavioural sleep in the fish T. mossambica was investigated and the increased response threshold to electrical and feeding stimuli measured. The results of these studies afford little evidence for REM-like sleep; however, the possibility of this kind of sleep is not precluded. Behavioural sleep is established for this species of fish. Sleep
Fish
Phylogeny
Behavioural criteria
ALTHOUGH the major behavioural and electrophysiological correlates of mammalian sleep have been well established [6,9], there is an extreme paucity of both behavioural and neurophysiological observations for the lower vertebrates [2]. This is especially so with regard to fish. Theories that relate the significance of sleep to phylogeny might well benefit from additional studies in several of the more primitive vertebrates; but it must be borne in mind that the application of electrophysiological criteria based on mammalian sleep may be totally inappropriate when applied to more primitive vertebrates. Therefore, for the present report the common-sense criteria of Flanigan etal. [2] for describing sleep-wakefulness cycles are used. These include: stereotypic posture, behavioural quiescence, elevated behavioural thresholds and very rapid state reversibility following relatively intense stimulations. METHOD
Animals Preliminary observations in other experiments have established that the fish, T. mossambica (Perciformes: Cichlidae) are very sensitive to movements such as by persons walking near the tanks [3]. For this reason, the experimental tanks of 50 1 capacity were made opaque on three sides and the front fitted with a sheet of solar-shield one-way glass. This glass has the advantages of allowing direct observations on the fish when the room lighting is dim with respect to tank illumination and also preventing the fish from detecting outside movements. The tanks were covered with a close fitting roof with light bulbs placed at each end. A small opening at one end of the tank provided entry for an air supply, water-filtering and food. Each of two tanks contained 15 fish and they were fed daily on hand-rolled bread pellets. This method of feeding was found to be satisfactory in other experiments [3]. Feeding experiments were conducted only after the fish had been trained to accept the pellets and had been adapted to a 15 hr light/9 hr dark photoperiod. Except where 613
otherwise noted, the fish were fed immediately with the onset of the light period and this at a rate of 1 pellet/6 sec for 1 min. The number of pellets out of each series of l 0 pellets reaching the bottom of the tank uneaten was noted. A minute later, another series of l0 pellets were given to the fish. Once 5 or more pellets reached the bottom of the tank uneaten, it was assumed that the fish were replete and feeding was stopped. The fish were only distrubed twice a week for tank cleaning.
Procedu re To observe the fish under laboratory night conditions the tank lights were connected to a variac so that the lights could be suitably dimmed at the end of the artificial light periods. The tanks were kept in a photographic darkroom to obviate undersirable light leaks. Lines were painted on the back wall of the tanks thus dividing them into quadrants. This allowed a crude quantitative measure of fish activity as the displacement of any one fish from any quadrant to another per unit time. A simple stimulator connected to 2 wire leads with metal plates at their ends and placed at diagonally opposite ends of the tank was used to measure the effects of electrical stimuli during different activity periods. Preliminary observations established that the first discernible response to electrical stimuli involved gulping, an alternating opening and closing of the mouth. This response was used to measure response threshold, which was taken as the first stimulus level at which any of the fish displayed gulping on gradually increasing stimulation. In the latency of feeding test as in the gulping response to electrical shock, the first pellet eaten by any fish in the tank is taken as the latency value. Finally, the activity rhythms of the fish were classified on the basis of several hundred hours of direct observation. It should be clearly noted that in no case were individual animals studied. Only populations of fish are considered because this approach is obviously the more natural way in view of the fact that this species is day-active and forms schools in its normal environment.
614
SHAPIRO AND HEPBURN RESULTS
Under conditions of a 15 hr light/9 hr dark photoperiod regime, the gross behaviour of T. m o s s a m b i c a can be clearly divided as follows. During the light phase 2 behavioural patterns were observed: a very active pattern characterized by much swimming (more than 10 fish displacements/min) referred to as State A; a second more passive phase (State B) of 3 or fewer displacements/min with less than 3 fish resting on the bottom of the tank at any time. State A fish exhibit higher respiratory activity, judged as gill movement, than do State B fish. These two phases alternate throughout the light period with an average of a 20 min active State A behaviour followed by 10 min of the passive State B behaviour. However, both of these states are quite labile and are readily altered by environmental stimuli such as feeding. At the onset of the dark period there is a marked change in the behaviour of the fish compared with that of the light period. For an initial period of 90 + 30 min there is a general reduction in swimming and a gradual downward vertical movement of the fish. This is followed by an extended period of some 6.5 to 7.5 hr during which at least 12 of the 15 fish in each tank were resting on the bottom of the tank (State C). The respiratory rate of State C fish is, however, no different from that of State B fish and no eye movement of State C fish could be observed (as apposed to States A and B). Towards the end of the dark period there is much irregular behaviour: several fish actively swim about, others remain resting on the bottom of the tank and yet others remained stationary in the middle of the tank. With the onset of light, the fish that had been actively swimming around immediately returned to the bottom of the tank and exhibited a peculiarly distinct behaviour for about 5 min as follows. The fish were initially still (about 30 sec) and then began to swim backwards in small circles for about 90 sec. After this there is a 1 - 3 min rest period during which the colour pattern of the fish changes from one of broad vertical stripes through a bleached stage and finally to a pattern of only one or two horizontal stripes. The fish then exhibit darting behaviour for a minute or two after which there is intense general swimming activity (more than 45 displacements per min) and strong schooling behaviour for 5 - 1 0 min. Finally, the fish enter the normal light phase behaviour of alternating A and B states. When the photoperiod regime was switched to 12 hr light and dark respectively, the only alteration in the above described behaviour was that the normally encountered sequence associated with the end of the 9 hr dark period was extended by an additional 3 hr under the new darkness regime. In another series of 8 experiments during which the fish were kept in total darkness for 48 hr, the behaviour of the fish at the end of this period was the same as that described for the end of the 9 hr period in the 9/15 hr photoperiod regime. Having observed the basic activity rhythms of the fish under three different photoperiod conditions, a number of experiments were superimposed on these activity rhythms in order to determine which, if any, of the observed patterns meet the sleep criteria as formulated by Flanigan e t a l . [2]. The effect of continuous illumination was investigated in eight experiments in which the fish were exposed to 96 hr of uninterrupted light. The fish were regularly monitored throughout this period and only States A and B activity were observed. On switching off the light, a shorter time was required for the fish to display
night resting behaviour (30 + 15 min). The effects of arbitrarily shortening the light period after a normal course of 9 hr darkness was investigated in another series of experiments in which it was found that fish which had been subjected to light periods of less than 6 hr behave, in the ensuing dark period, in a manner indistinguishable from that encountered under normal light conditions, i.e. the normal alternation of States A and B occur. Only after 6 hr of light did some of the fish display the normal sequence of behaviour associated with the 9/15 hr photoperiod regime. In another series of experiments electrical stimuli were given to fish in States A, B or C. Thirty such experiments were performed for each activity state for two tanks of fish. The minimum stimulus required to elicit the gulping response, based on 60 measurements for each state, is shown in Table 1 from which it can be seen that stimulatory thresholds for the A and B fish were not significantly different; however, C is significantly different from both A and B. TABLE I ELECTRICAL STIMULUS-RESPONSETHRESHOLDS FOR T. MOSSAMB1CA IN THREE DIFFERENTACTIVITYSTATES
Threshold, in V S.D. p
A
States B
C
5.1 1.2
4.8 l.l
9.1 1.6
Avs. B Avs. C
n.s.
0.05
Bvs. C
0.05
In order to further distinguish the three major states of behaviour, a biologically more natural measure of response threshold, that of feeding, was assayed. States A and B fish did not differ with respect to the time delay before the eating of the first bread pellet and averaged 5 + 2 sec, while State C fish averaged 70 + 30 sec. As a control to test whether or not this long C response was specifically related to light deprivation effects, other fish which had been exposed to light for only 2 hr after their normal 9 hr dark period were resubjected to a 2 hr dark period at the end of which they were given bread pellets and the mean reaction time in these fish was 8 + 3 sec. It should be noted that the 2 hr period of darkness preceeding the feeding stimulus is coincident with the longest period, 90 + 30 minutes, required for the fish to display darkness resting behaviour after a normal period of light. All of the above results and observations pertain only to adult fish. Early experiments based on the use of young fish produced quite erratic results: for example, none of the immatures exhibited backward circular swimming nor the changes in colour pattern. It is also of interest to note that the young fish were less active than the adults during light periods, but became more active on the introduction of adults into their tanks. DISCUSSION Several findings of interest emerge from the above observations on the photoperiod related activity rhythms of the fish, T. m o s s a m b i c a , that may readily be interpreted in terms of the behavioural sleep criteria of Flanigan e t al. [2]. In terms of stereotypical posture, for example, the fish were observed to rest at the bottom of the tank only during
SLEEP IN A SCHOOLING FISH
615
the dark period and the only other nonswimming behaviour is that exhibited in State B wherein the fish simply remain bouyant and afloat usually near the middle of the tank. The approximately 7 hr period of quiescence observed during darkness included greatly reduced responses to feeding and electrical stimuli (State C) thus fulfilling both the quiescence and raised response threshold criteria. The labile nature of this state is evidenced by the effects of both electrical and light stimuli which converts C behaviour to that of A regardless of the length of the dark period used. Although the above observations collectively satisfy the sleep criteria set out in the introduction, the lack of rapid eye movement (REM) during activity State C would imply that there is no REM sleep in this species of fish; and, this is consistent with other observations on the tench [5]. However, other studies on a number of coral reef fish species [8] report the occurrence of REM sleep. Moreover, Marshall [4] has reported that there are day-active as well as night-active species of fish; but, interestingly enough, in each case there is stereotypical and sepcies-specific resting posture and that there is reasonable qualitative evidence of a raised response threshold for several of the coral reef species during normal nocturnal resting periods. Although it has previously been shown that the full complement of those anatomical centres responsible for mammalian sleep is not present in those species of fish
which have thus far been examined [ 1], it is a fact that several species of fish exhibit behavioural sleep [4, 5, 8] in addition to the present evidence in the case of T. mossambica. The possibility that other and more primitive centres subserving a sleep function in fish might be considered, and, by the same token, it is certainly conceivable that the neurophysiological basis of lower vertebrate sleep is both qualitatively and quantitatively different from that of higher vertebrates. However, even in the very few recent investigations dealing with reptiles [2,7] there is lack of agreement on the occurrence of quiet and active sleep. Similarly, in the present report immature fish were rejected during the early stages of preliminary observations on the grounds that their behaviour was erratic with respect to the normal schooling behaviour (which, of course, makes description less difficult). This would imply that even the use of simple models in studies of sleep ontogeny are fraught with the problems of complex interactions in gregarious or social animals, and that considerably more comparative data on both active and quiet states will be required in order to develop a predictive theory of sleep. ACKNOWLEDGEMENT Wearegrate~ltoMr. D. Veenhof~rtechnicalassistance.
REFERENCES 1. Broughton, R. Phylogenetic evolution of sleep systems. In: The Sleeping Brain, edited by M. H. Chase. Los Angeles: Brain Research Institute, 1972. 2. Flanigan, W. F., R. H. Wilcox and A. Rechtschaffen. The EEG and behavioural continuum of the corcodilian, Caiman sclerops. Electroenceph. clin. Neurophysiol. 34: 521-538, 1973. 3. Hepburn, H. R., N. J. Berman, H. D. Jacobson and P. Fatti. Trends in arthropod defensive secretions, an aquatic predator assay. Oecologia 12: 373-382, 1973. 4. Marshall, N. B. Sleep in fishes. Proc. R. Soc. Med. 65: 177, 1972. 5. Peyrethon, J. and D. Dusan-Peyrethon. Etude polygraphique du cycle veiUe-sommeild'un Teleosteen (Tinca tincaJ. C r. hebd. S~anc. Acad. Sci., Paris 161: 2533-2537, 1967.
6. Snyder, F. Toward an evolutionary theory of dreaming. Am. J. Psyehiat. 123: 121-136, 1966. 7. Susic, V. Electrographic and behavioural correlations of the rest-activity cycle in the sea turtle, Caretta caretta L. (Chelonia). J. exp. mar. lh'oL Ecol. 10: 81-87, 1972. 8. Tauber, E. S. and E. D. Weitzman. Eye movements during behavioural inactivity in certain Bermuda reef fish. Communs Behav. Biol. 3: 131-135, 1969. 9. Williams, R. L., M. W. Agnew and W. B. Webb. Sleep patterns in young adults: an EEG study. Electroenceph. clin. Neurophysiol. 17: 376-381, 1964.
Sleep in a Schooling Fish, Tilapia mossambica C. M. SHAPIRO AND H. R. HEPBURN
Department o f Physiology, University o f the lqitwatersrand Johannesburg, Republic o f South Africa (Received 28 July 1975) SHAPIRO, C. M. AND H. R. HEPBURN. Sleep in a schooling fish, Tilapia mossambica. PHYSIOL. BEHAV. 16 (5) 613-615, 1976. - There are only a few infra-mammalian studies on sleep. For this reason, behavioural sleep in the fish T. mossambica was investigated and the increased response threshold to electrical and feeding stimuli measured. The results of these studies afford little evidence for REM-like sleep; however, the possibility of this kind of sleep is not precluded. Behavioural sleep is established for this species of fish. Sleep
Fish
Phylogeny
Behavioural criteria
ALTHOUGH the major behavioural and electrophysiological correlates of mammalian sleep have been well established [6,9], there is an extreme paucity of both behavioural and neurophysiological observations for the lower vertebrates [2]. This is especially so with regard to fish. Theories that relate the significance of sleep to phylogeny might well benefit from additional studies in several of the more primitive vertebrates; but it must be borne in mind that the application of electrophysiological criteria based on mammalian sleep may be totally inappropriate when applied to more primitive vertebrates. Therefore, for the present report the common-sense criteria of Flanigan etal. [2] for describing sleep-wakefulness cycles are used. These include: stereotypic posture, behavioural quiescence, elevated behavioural thresholds and very rapid state reversibility following relatively intense stimulations. METHOD
Animals Preliminary observations in other experiments have established that the fish, T. mossambica (Perciformes: Cichlidae) are very sensitive to movements such as by persons walking near the tanks [3]. For this reason, the experimental tanks of 50 1 capacity were made opaque on three sides and the front fitted with a sheet of solar-shield one-way glass. This glass has the advantages of allowing direct observations on the fish when the room lighting is dim with respect to tank illumination and also preventing the fish from detecting outside movements. The tanks were covered with a close fitting roof with light bulbs placed at each end. A small opening at one end of the tank provided entry for an air supply, water-filtering and food. Each of two tanks contained 15 fish and they were fed daily on hand-rolled bread pellets. This method of feeding was found to be satisfactory in other experiments [3]. Feeding experiments were conducted only after the fish had been trained to accept the pellets and had been adapted to a 15 hr light/9 hr dark photoperiod. Except where 613
otherwise noted, the fish were fed immediately with the onset of the light period and this at a rate of 1 pellet/6 sec for 1 min. The number of pellets out of each series of l 0 pellets reaching the bottom of the tank uneaten was noted. A minute later, another series of l0 pellets were given to the fish. Once 5 or more pellets reached the bottom of the tank uneaten, it was assumed that the fish were replete and feeding was stopped. The fish were only distrubed twice a week for tank cleaning.
Procedu re To observe the fish under laboratory night conditions the tank lights were connected to a variac so that the lights could be suitably dimmed at the end of the artificial light periods. The tanks were kept in a photographic darkroom to obviate undersirable light leaks. Lines were painted on the back wall of the tanks thus dividing them into quadrants. This allowed a crude quantitative measure of fish activity as the displacement of any one fish from any quadrant to another per unit time. A simple stimulator connected to 2 wire leads with metal plates at their ends and placed at diagonally opposite ends of the tank was used to measure the effects of electrical stimuli during different activity periods. Preliminary observations established that the first discernible response to electrical stimuli involved gulping, an alternating opening and closing of the mouth. This response was used to measure response threshold, which was taken as the first stimulus level at which any of the fish displayed gulping on gradually increasing stimulation. In the latency of feeding test as in the gulping response to electrical shock, the first pellet eaten by any fish in the tank is taken as the latency value. Finally, the activity rhythms of the fish were classified on the basis of several hundred hours of direct observation. It should be clearly noted that in no case were individual animals studied. Only populations of fish are considered because this approach is obviously the more natural way in view of the fact that this species is day-active and forms schools in its normal environment.
614
SHAPIRO AND HEPBURN RESULTS
Under conditions of a 15 hr light/9 hr dark photoperiod regime, the gross behaviour of T. m o s s a m b i c a can be clearly divided as follows. During the light phase 2 behavioural patterns were observed: a very active pattern characterized by much swimming (more than 10 fish displacements/min) referred to as State A; a second more passive phase (State B) of 3 or fewer displacements/min with less than 3 fish resting on the bottom of the tank at any time. State A fish exhibit higher respiratory activity, judged as gill movement, than do State B fish. These two phases alternate throughout the light period with an average of a 20 min active State A behaviour followed by 10 min of the passive State B behaviour. However, both of these states are quite labile and are readily altered by environmental stimuli such as feeding. At the onset of the dark period there is a marked change in the behaviour of the fish compared with that of the light period. For an initial period of 90 + 30 min there is a general reduction in swimming and a gradual downward vertical movement of the fish. This is followed by an extended period of some 6.5 to 7.5 hr during which at least 12 of the 15 fish in each tank were resting on the bottom of the tank (State C). The respiratory rate of State C fish is, however, no different from that of State B fish and no eye movement of State C fish could be observed (as apposed to States A and B). Towards the end of the dark period there is much irregular behaviour: several fish actively swim about, others remain resting on the bottom of the tank and yet others remained stationary in the middle of the tank. With the onset of light, the fish that had been actively swimming around immediately returned to the bottom of the tank and exhibited a peculiarly distinct behaviour for about 5 min as follows. The fish were initially still (about 30 sec) and then began to swim backwards in small circles for about 90 sec. After this there is a 1 - 3 min rest period during which the colour pattern of the fish changes from one of broad vertical stripes through a bleached stage and finally to a pattern of only one or two horizontal stripes. The fish then exhibit darting behaviour for a minute or two after which there is intense general swimming activity (more than 45 displacements per min) and strong schooling behaviour for 5 - 1 0 min. Finally, the fish enter the normal light phase behaviour of alternating A and B states. When the photoperiod regime was switched to 12 hr light and dark respectively, the only alteration in the above described behaviour was that the normally encountered sequence associated with the end of the 9 hr dark period was extended by an additional 3 hr under the new darkness regime. In another series of 8 experiments during which the fish were kept in total darkness for 48 hr, the behaviour of the fish at the end of this period was the same as that described for the end of the 9 hr period in the 9/15 hr photoperiod regime. Having observed the basic activity rhythms of the fish under three different photoperiod conditions, a number of experiments were superimposed on these activity rhythms in order to determine which, if any, of the observed patterns meet the sleep criteria as formulated by Flanigan e t a l . [2]. The effect of continuous illumination was investigated in eight experiments in which the fish were exposed to 96 hr of uninterrupted light. The fish were regularly monitored throughout this period and only States A and B activity were observed. On switching off the light, a shorter time was required for the fish to display
night resting behaviour (30 + 15 min). The effects of arbitrarily shortening the light period after a normal course of 9 hr darkness was investigated in another series of experiments in which it was found that fish which had been subjected to light periods of less than 6 hr behave, in the ensuing dark period, in a manner indistinguishable from that encountered under normal light conditions, i.e. the normal alternation of States A and B occur. Only after 6 hr of light did some of the fish display the normal sequence of behaviour associated with the 9/15 hr photoperiod regime. In another series of experiments electrical stimuli were given to fish in States A, B or C. Thirty such experiments were performed for each activity state for two tanks of fish. The minimum stimulus required to elicit the gulping response, based on 60 measurements for each state, is shown in Table 1 from which it can be seen that stimulatory thresholds for the A and B fish were not significantly different; however, C is significantly different from both A and B. TABLE I ELECTRICAL STIMULUS-RESPONSETHRESHOLDS FOR T. MOSSAMB1CA IN THREE DIFFERENTACTIVITYSTATES
Threshold, in V S.D. p
A
States B
C
5.1 1.2
4.8 l.l
9.1 1.6
Avs. B Avs. C
n.s.
0.05
Bvs. C
0.05
In order to further distinguish the three major states of behaviour, a biologically more natural measure of response threshold, that of feeding, was assayed. States A and B fish did not differ with respect to the time delay before the eating of the first bread pellet and averaged 5 + 2 sec, while State C fish averaged 70 + 30 sec. As a control to test whether or not this long C response was specifically related to light deprivation effects, other fish which had been exposed to light for only 2 hr after their normal 9 hr dark period were resubjected to a 2 hr dark period at the end of which they were given bread pellets and the mean reaction time in these fish was 8 + 3 sec. It should be noted that the 2 hr period of darkness preceeding the feeding stimulus is coincident with the longest period, 90 + 30 minutes, required for the fish to display darkness resting behaviour after a normal period of light. All of the above results and observations pertain only to adult fish. Early experiments based on the use of young fish produced quite erratic results: for example, none of the immatures exhibited backward circular swimming nor the changes in colour pattern. It is also of interest to note that the young fish were less active than the adults during light periods, but became more active on the introduction of adults into their tanks. DISCUSSION Several findings of interest emerge from the above observations on the photoperiod related activity rhythms of the fish, T. m o s s a m b i c a , that may readily be interpreted in terms of the behavioural sleep criteria of Flanigan e t al. [2]. In terms of stereotypical posture, for example, the fish were observed to rest at the bottom of the tank only during
SLEEP IN A SCHOOLING FISH
615
the dark period and the only other nonswimming behaviour is that exhibited in State B wherein the fish simply remain bouyant and afloat usually near the middle of the tank. The approximately 7 hr period of quiescence observed during darkness included greatly reduced responses to feeding and electrical stimuli (State C) thus fulfilling both the quiescence and raised response threshold criteria. The labile nature of this state is evidenced by the effects of both electrical and light stimuli which converts C behaviour to that of A regardless of the length of the dark period used. Although the above observations collectively satisfy the sleep criteria set out in the introduction, the lack of rapid eye movement (REM) during activity State C would imply that there is no REM sleep in this species of fish; and, this is consistent with other observations on the tench [5]. However, other studies on a number of coral reef fish species [8] report the occurrence of REM sleep. Moreover, Marshall [4] has reported that there are day-active as well as night-active species of fish; but, interestingly enough, in each case there is stereotypical and sepcies-specific resting posture and that there is reasonable qualitative evidence of a raised response threshold for several of the coral reef species during normal nocturnal resting periods. Although it has previously been shown that the full complement of those anatomical centres responsible for mammalian sleep is not present in those species of fish
which have thus far been examined [ 1], it is a fact that several species of fish exhibit behavioural sleep [4, 5, 8] in addition to the present evidence in the case of T. mossambica. The possibility that other and more primitive centres subserving a sleep function in fish might be considered, and, by the same token, it is certainly conceivable that the neurophysiological basis of lower vertebrate sleep is both qualitatively and quantitatively different from that of higher vertebrates. However, even in the very few recent investigations dealing with reptiles [2,7] there is lack of agreement on the occurrence of quiet and active sleep. Similarly, in the present report immature fish were rejected during the early stages of preliminary observations on the grounds that their behaviour was erratic with respect to the normal schooling behaviour (which, of course, makes description less difficult). This would imply that even the use of simple models in studies of sleep ontogeny are fraught with the problems of complex interactions in gregarious or social animals, and that considerably more comparative data on both active and quiet states will be required in order to develop a predictive theory of sleep. ACKNOWLEDGEMENT Wearegrate~ltoMr. D. Veenhof~rtechnicalassistance.
REFERENCES 1. Broughton, R. Phylogenetic evolution of sleep systems. In: The Sleeping Brain, edited by M. H. Chase. Los Angeles: Brain Research Institute, 1972. 2. Flanigan, W. F., R. H. Wilcox and A. Rechtschaffen. The EEG and behavioural continuum of the corcodilian, Caiman sclerops. Electroenceph. clin. Neurophysiol. 34: 521-538, 1973. 3. Hepburn, H. R., N. J. Berman, H. D. Jacobson and P. Fatti. Trends in arthropod defensive secretions, an aquatic predator assay. Oecologia 12: 373-382, 1973. 4. Marshall, N. B. Sleep in fishes. Proc. R. Soc. Med. 65: 177, 1972. 5. Peyrethon, J. and D. Dusan-Peyrethon. Etude polygraphique du cycle veiUe-sommeild'un Teleosteen (Tinca tincaJ. C r. hebd. S~anc. Acad. Sci., Paris 161: 2533-2537, 1967.
6. Snyder, F. Toward an evolutionary theory of dreaming. Am. J. Psyehiat. 123: 121-136, 1966. 7. Susic, V. Electrographic and behavioural correlations of the rest-activity cycle in the sea turtle, Caretta caretta L. (Chelonia). J. exp. mar. lh'oL Ecol. 10: 81-87, 1972. 8. Tauber, E. S. and E. D. Weitzman. Eye movements during behavioural inactivity in certain Bermuda reef fish. Communs Behav. Biol. 3: 131-135, 1969. 9. Williams, R. L., M. W. Agnew and W. B. Webb. Sleep patterns in young adults: an EEG study. Electroenceph. clin. Neurophysiol. 17: 376-381, 1964.
