SleepBrazilian in electricJournal fishof Medical and Biological Research (1999) 32: 1223-1228 1223 ISSN 0100-879X Short Communication

Electrolocation-communication discharges of the Gymnotus carapo L. (Gymnotidae: Gymnotiformes) during behavioral sleep

R.M. Stopa and Departamento de Ciências Biológicas, Faculdade de Ciências, K. Hoshino Universidade Estadual Paulista, Bauru, SP, Brasil

Abstract

Correspondence Technical problems have hampered the study of sleep in . The Key words R.M. Stopa electrical discharges of Gymnotus carapo L. (Gymnotidae: Gymnoti- · Sleep · Departamento de Ciências Biológicas formes) were monitored to evaluate their ease and reliability as Fish Faculdade de Ciências, UNESP · parameters to study sleep. The discharges were detected by electrodes Electric discharges Av. Eng. Luiz Edmundo C. Coube, s/n · Cannibalism immersed in a glass aquarium and were recorded on a conventional 17033-360 Bauru, SP · Antipredation mechanism Brasil polygraph. G. carapo showed conspicuous signs of behavioral sleep. During these periods, opercular beat rates were counted, electric Publication supported by FAPESP. discharges recorded, and the “sharp discharge increase” (SDI) of the orienting reflex was investigated. All 20 animals monitored main- tained electrical discharges during behavioral sleep. The discharge frequencies during sleep (50.3 ± 10.4 Hz) were not significantly Received September 24, 1998 Accepted July 26, 1999 different from those observed when the fish was awake and inactive (57.2 ± 12.1 Hz) (Wilcoxon matched-pairs signed-ranks test, P>0.05). However, the SDI, which was prevalent in the awake fish, was not observed during periods of behavioral sleep. Additional observations showed that the species had cannibalistic habits. When presented with electrical discharges from a conspecific, the sleeping fish showed an initial decrease or pause in discharge frequency, while the awake fish did not have this response. We conclude that the electrical discharges of G. carapo were not conspicuous indicators of behavioral sleep. Discharges may have been maintained during sleep for sensory pur- poses, i.e., conspecific detection and avoidance of cannibalistic at- tacks.

The heuristic value of comparative stud- ria, and the universal occurrence of sleep in ies for sleep research is widely recognized, the group is still questioned (2,3). but many animal groups have received little Gymnotus carapo L. (Gymnotidae: Gym- attention. Fish, in particular, have been ne- notiformes), popularly called “tuvira”, is a glected because of the technical difficulties low voltage electric pulse triggering fish, and in obtaining and interpreting conventional may be a potentially valuable species for electrographic recordings (1). As pointed out sleep research. It uses electric organ dis- by others (1-6), existing studies of sleep in charges (EOD) to interact with its environ- fish are based primarily on behavioral crite- ment, e.g., to communicate with conspecif-

Braz J Med Biol Res 32(10) 1999 1224 R.M. Stopa and K. Hoshino

ics and to scan its habitat (7-9). Since re- vibration of the aquarium, etc). Another 20 duced interaction with the environment is a fish were observed to determine the occur- marked characteristic of sleep, the electrical rence of behavioral sleep. For this purpose, discharges of G. carapo would be expected each animal was left undisturbed (except for to change during a sleep-activity cycle. To the placement of worms in the tank) in the investigate this hypothesis, in the present experimental aquarium for 7 days. Several study we determined the occurrence of be- times a day, the fish were observed for signs havioral sleep in G. carapo and then ob- of behavioral sleep, i.e., prolonged immobil- served if the state was correlated with changes ity in a particular posture, reduced fin move- in EOD. The electrical discharges of this ment, and reduction of the frequency of cy- species oscillate between 40 and 100 Hz and clic opercular movements. EOD recordings can be measured continuously and easily by were carried out when signs of behavioral a noninvasive technique. sleep were apparent and during the orienting G. carapo individuals of both sexes from reflex test, characterized in the awake ani- the Paraná River basin were used during the mal by a sharp discharge increase (SDI) in study. Size ranged from 50 to 400 mm stan- response to a plastic rod immersed and stirred dard length. Ten groups of 15 animals were in the vicinity of the fish. Near the end of the kept in aerated water tanks (1,000-l capac- behavioral sleep recording periods, the ity) under ambient temperature and light aquarium wall was knocked upon to induce conditions. The fish were fed earthworms locomotion by the fish and to confirm the daily in amounts corresponding to approxi- reversibility of the behavioral state. mately 10 g/animal. They were allowed to Recordings for the first 10 animals at 23- acclimate for 7 days and then one animal 25oC showed that inactive, awake fish emit- was singled out and transferred to the exper- ted very regular pulse discharge frequencies imental chamber, an aerated glass aquarium ranging from 42 to 68 Hz (mean ± SD of 53.5 (250 x 500 x 250 mm) surrounded by a visual ± 12.1 Hz) (Figure 1). Discharge amplitude isolation fence and containing a thermom- changed with animal position in relation to eter. The bottom of the aquarium was filled the electrodes and the maximum voltage with a 10-mm deep layer of washed sand. A observed was 1 mV. The frequency of pulses shelter made with a piece of PVC tubing (24 increased during swimming, reaching 75.5 ± to 40 mm in diameter, 50 to 280 mm in 16.0 Hz (mean ± SD of 15 randomly selected length) was positioned in the sand so that the episodes). No relationship was observed be- animal could be observed through a sight tween animal size or previous swimming hole from outside the visual isolation fence. and the frequency of discharge. The mean Electrical cables with 20-mm non-insulated and standard deviation of the SDI frequency tips were immersed and fixed on the oppo- was 71.4 ± 12.8 Hz. SDI was elicited by site longitudinal walls of the aquarium to rapidly moving shadows and unexpected light transmit and record electric organ discharges changes, observer approximation, knocks on via the electromyography channel of an EM the aquarium wall, or immersion and slow polygraph (Electronics for Medicine, Hous- movement of a plastic rod. EOD frequency ton, TX, USA). decreased and increased directly with tem- Preliminary observations were made on perature, but ceased outside the 9-32oC range. 10 animals to evaluate recording quality, Near the extremes of this range, the fish lost baseline discharge frequencies, the effect of their natural posture and showed EOD arrest temperature, and the response to presenta- for 3-10 s in response to painful pinching. tions of some environmental stimuli (light There were clear signs of behavioral sleep flash, moving shadows, sounds, mechanical for G. carapo. Sleep was characterized by a

Braz J Med Biol Res 32(10) 1999 Sleep in 1225 motionless posture with the tail down (touch- ful stimulation (pinching). G. carapo emit- ing the sand) and the head up (30 degrees ted electrical discharges throughout the above with respect to the bottom of the aquarium), behavioral sequence, and then stopped them reduction in amplitude and frequency of oper- 5-20 min after quiescence had begun. Be- cular movements, and ventral fin immobil- cause EOD was maintained after behavioral ity. Plastic rod immersion into the aquarium responses had ceased, it is likely to be con- during these periods did not evoke an SDI, trolled independently. EOD was recovered indicating a reduction in the vigilance level. after 5-10 min in the original aquarium, con- However, the more extreme stimulus of a comitantly with postural and swimming nor- knock on the aquarium wall woke the fish up malization. as indicated by generalized body movements Interactions with conspecifics were in- and undulations of the ventral fins. EOD vestigated in order to determine the func- recordings unexpectedly showed that elec- tional role played by the sleep EOD of G. trical discharges were maintained during carapo. Small size class G. carapo (50-90 behavioral sleep periods. Discharge frequen- mm) were introduced into an aquarium with cies (50.3 ± 10.4 Hz) were not significantly a larger resident animal, which was awake different from those recorded during inac- and had been starved for two days (the resi- tive awake periods (57.2 ± 12.1 Hz) (Wil- dent fish was at least 30 mm longer than the coxon matched-pairs signed-ranks test, intruder fish). A brief period of attentive P>0.05). Amplitude remained constant dur- behavior of the resident animal was observed, ing sleep and waking periods if the fish did followed by a rapid exit from the PVC shel- not change its position in the aquarium. ter, pursuit and capture of the intruder fish, To determine if electrical discharges oc- and immediate cannibalism (within 2 to 5 curred during sleep via mechanisms inde- min). Alternatively, in two cases (15.4%), pendent of those that control other aspects of the smaller fish was killed within the first 5 sleep, behavior and EOD were monitored in min, but ingestion was delayed for up to 5 h. anesthetized fish. Thirteen fish were anes- The ingestion latencies were shown to in- thetized for the experiment individually by crease up to 24 h if a second or more intrud- immersion in an aquarium containing 200 ers were introduced (16 observations). mg/l of compound MS-222. Later, they were These observations suggest the possibil- returned to their original aquarium and moni- ity that G. carapo uses its EOD during sleep tored to assess the reversibility of anesthetic- to detect conspecifics and to avoid cannibal- induced sleep compared with natural sleep. istic attacks. This hypothesis was investi- The sequence of behaviors observed was gated by transmitting the EOD of a stimulus agitated normal swimming, uncoordinated fish (SF) to a sleeping fish whose EOD was swimming with postural loss (initiated after monitored (response fish; RF). SF discharges 10-15 min of latency), and behavioral quies- were transmitted to the RF aquarium by cence with loss of motor responses to pain- flipping a switch interposed in the cables Figure 1 - Electric organ dis- charge of Gymnotus carapo, used for social communication and electrolocation. A segment of a recording made at rest and after glass rod presentation (ar- row) in the animal’s vicinity is illustrated. Time calibration: 1 s. e, Stimulus presentation. e

Braz J Med Biol Res 32(10) 1999 1226 R.M. Stopa and K. Hoshino

connecting the aquaria. Stimulation lasted frequent than the other responses for awake about 30-60 s and, in order to compare the fish (chi-square = 62.6, d.f. = 2, P<0.05). response, it was repeated after at least 3 min None of the SF-RF size combinations were if the RF showed signs of being awake. The associated with a particular response during electrical discharges of 10 fish classified as the initial phase of the EOD recordings. small (100 to 200 mm standard length), 10 as Following the first 3 s, the focused intermediate (210 to 300 mm), and 10 as showed a variety of EOD patterns. Some large (310 to 400 mm) were monitored indi- were silent for only a brief period and then vidually on different days during EOD trans- resumed their EOD, while others remained missions of small, intermediate, and large silent during and for 30 s or more after the stimulus fishes. A total of 90 recordings end of SF discharge transmission. Some ap- were made when animals were asleep and 90 peared to be competing with the SF trans- when they were awake. mission. These fish emitted alternate dis- Independent of SF or RF size, 82 (91.1%) charge bursts or simultaneously superim- of the tests conducted on sleeping animals posed their discharges on those of the SF. showed a discharge frequency reduction or These observations were not the focus of silence during the first 3 s of SF discharge this study, but it was interesting to note that presentation (Figure 2A). This response was six of the focused fishes attacked the tips of significantly more common than either dis- the transmission cables. charge frequency increases or superpositions The data obtained indicate that G. carapo (SF and RF simultaneously emitting electri- does sleep. The species showed a particular cal discharges), which occurred with the sleep posture, relative immobility, increased other 8 fish (8.9%) (chi-square = 134.8, d.f. sensory thresholds, and reversibility of the = 2, P<0.05). In contrast, 72.2% of the awake state, which are the four widely accepted focused fishes displayed discharge frequency behavioral criteria for sleep (10). Conse- increases or superpositions in response to quently, this can be added to the SF transmissions (Figure 2B). Another the category of other species in which be- 18.9% of the awake fish showed discharge havioral sleep has already been demonstrated frequency decreases, and 8.9% displayed no (1-6,10). observable change. A chi-square test showed The sustained EOD observed during be- that increases and superpositions were more havioral sleep in G. carapo may suggest that

Figure 2 - Initial response of Gymnotus carapo to conspecific F discharges presented during sleep (A) and wakefulness (B). Segments of polygraphic record- ings for the same animal are pre- sented. The stimulus started at the arrows and lasted 30 s. Note A the brief interruption of dis- charge in A, when discharges of the stimulus fish were recorded (low-voltage pulses), and subse- F quent overlapping discharges (probably competition). Such dis- charge arrest was not observed when the animal was awake. Time calibration: 1 s. F, Onset of discharge presentation. B

Braz J Med Biol Res 32(10) 1999 Sleep in electric fish 1227 the periods considered as sleep, according to ity for the purpose of concealment by preda- behavioral criteria, are not true sleep. In- tors or prey is a well-known adaptive mech- stead, the periods may correspond to the anism in aquatic animals and has been dem- acrophases of rest periods during the circa- onstrated mainly in the cardiorespiratory ac- dian rest-activity cycle. This seems unlikely, tivity of decapod crustaceans (13,14). In fish, however, because electrical discharges con- however, concealment via cardiorespiratory tinued even after the fish lost their natural arrest is not a viable strategy, because sus- posture during pharmacologically induced pension of heart and opercular beats can last sleep. In other words, the mechanisms con- only a few seconds (15). It is instead an trolling EOD may be largely independent of anticipatory protective reflex before the ex- those controlling behavioral sleep. The re- plosive muscular activity of a flight response duction or suppression of electrical dis- (15). In contrast, the arrest observed in the charges in the RF, observed when the EOD electrical discharges of G. carapo was lengthy of an SF were presented during behavioral for some fish, suggesting its use for conceal- sleep, seems to be strong evidence for the ment. Considering the cannibalistic habits of occurrence of significant changes in central the species, lengthy EOD suspension may be nervous system activity. A similar reduction considered as part of an antipredation mech- or silence of EOD in G. carapo, elicited by anism. In this scenario, the brief initial “si- sensory stimulation, was also reported dur- lence” period, observed primarily in sleep- ing light anesthesia (a state in which there ing fishes, is elicited as part of the startle are changes in central nervous activity) response and then used to analyze exog- (Correa SM and Hoffman A, personal com- enous electrical discharges without interfer- munication). ence from their own EOD. The information Reduction in cerebral serotonin level (11) obtained by the fish during the “silence” and sleep potentiation by vasoactive intesti- period may determine its decision to main- nal peptide (2) corroborate the assumption tain EOD suppression or resume its electri- that behavioral sleep in fish is determined by cal discharges. Important points for future a specific mechanism that occurs in homeo- investigation include whether heart beats therms. In addition, maintenance of non- are concomitantly arrested during the peri- vegetative activities during sleep, like swim- ods of EOD suspension, and whether G. ming in some fish species and aquatic mam- carapo may detect cardiac electrical activity mals, is well known. Maintaining a sensory of conspecifics. channel active during sleep may also be Assuming that the brief “silences” were adaptive. For example, the lowering of the part of the startle-orienting reflex, it may be auditory stimulus threshold to awake a rat if possible to explain why a significantly greater the sound is conditioned to a signaling elec- number of sleeping G. carapo displayed a tric shock (12). Maintenance of EOD during reduction or silence of discharge frequen- sleep may have adaptive significance for G. cies in response to EOD transmissions from carapo as well, by allowing them to detect conspecifics. Cardiac arrest induced by star- conspecific cannibals. Contrary to what was tling stimuli has been observed in dominant expected, the use of EOD to assess the oc- (Oreochromis niloticus, Cichlidae: currence of sleep was not possible with Perciformes), which, because of their domi- the methods used in this study. However, nance, are able to rest. Submissive tilapias, the existence of small, undetectable, differ- which remain startled almost all the time, do ences in EOD during sleep should not be not show the cardiac arrest response (16). ruled out. This suggests that defensive concealment Temporary suppression of visceral activ- may be more effective for resting or sleeping

Braz J Med Biol Res 32(10) 1999 1228 R.M. Stopa and K. Hoshino

fishes. For G. carapo, the immediate cessa- on the misinterpretation of observations about tion of electrical discharges avoids or retards species which swim during sleep for passive its detection, granting sufficient time for iden- (ram) ventilation of the gills or move their tification of exogenous stimuli and initiation fins to make postural corrections. The uni- of other defensive responses. versal occurrence of sleep in fishes seems Although EOD were not useful for de- likely since it has been reported in inverte- tecting sleep, the present study demonstrated brates as well (17,18). Seep compensation that behavioral sleep occurs in G. carapo. after a period of deprivation (“rebound This finding adds to the diversity of taxo- sleep”) (19,20) indicates the importance of nomic fish groups in which sleep has been this physiological state among numerous and described. Reports of fish species which do distantly related animal taxa. not sleep seem to be based, at least partially,

References

1. Peyrethon J & Dusan-Peyrethon D (1967). telencéfalo em Gymnotus carapo (Teleos- lobster Homarus americanus. Canadian Étude polygraphique du cicle veille- tei: Gmonotoidei). Avaliação de sua Journal of Zoology, 50: 165-170. sommeil d’un téleosten (Tinca tinca). participação nos processos de alerta e 14. Cuadras J (1981). Behavioral determi- Comptes Rendus des Séances de la habituação. Master’s thesis, Universidade nants of severe cardiac inhibition. Physi- Société de Biologie, 161: 2533-2537. de São Paulo, Ribeirão Preto, SP. ological Psychology, 9: 384-392. 2. Campbell SS & Tobler I (1984). Animal 8. Lissmann H (1958). On the function and 15. Oliveira PC & Hoshino K (1989). Heart sleep: a review of sleep duration across evolution of electric organs in fish. Jour- rate in fishes during interspecific interac- phylogeny. Neuroscience and Biobehav- nal of Experimental Biology, 35: 156-191. tions. Boletim de Fisiologia Animal, 13: ioral Reviews, 8: 269-300. 9. Hopkins CD (1988). Neuroethology of 49-54. 3. Lenke R (1988). Hormonal control of electric communication. Annual Review 16. Stopa RM (1992). Paradas cardíacas da sleep-appetitive behaviour and diurnal ac- of Neuroscience, 11: 497-535. reação de sobressalto em tilápias (Oreo- tivity rhythms of cleaner Labroides 10. Hartse KM (1994). Sleep in insects and chromis niloticus) dominantes e submis- dimidiatus (Labridae, Teleostei). Behav- nonmammalian vertebrates. In: Kryger sas. Master’s thesis, Universidade ioural Brain Research, 27: 73-85. MH, Roth T & Dement WC (Editors), Prin- Estadual Paulista, Botucatu, SP. 4. Karmanova IG & Lazarev SG (1979). ciples and Practice of Sleep Medicine. WB 17. Arechiga MD (1994). Do crustaceans Stages of sleep evolution (facts and hy- Saunders, Philadelphia. sleep? Newsletter of the World Federa- pothesis). Waking and Sleeping, 3: 506- 11. Fingerman SW (1976). Circadian rhythms tion of Sleep Research Societies, 4: 7-9. 510. of brain 5-hydroxy-tryptamine and swim- 18. Tobler I (1994). Evidence for sleep in the 5. Shapiro CM, Woolf CJ & Borsook D ming activity in the Fundulus scorpion and cockroach. Newsletter of the (1981). Sleep ontogeny in fish. In: Laming grandis. Comparative Biochemistry and World Federation of Sleep Research Soci- PR (Editor), Brain Mechanisms of Behav- Physiology, 54C: 49-53. eties, 4: 9-11. ior in Lower Vertebrates. Cambridge Uni- 12. Trigona MC, Ciancia F & Bloch V (1968). 19. Tobler I & Bórbely AA (1985). Effect of versity Press, Cambridge. La profondeur du sommeil paradoxal chez rest deprivation on motor activity of fish. 6. Melesio-Nolasco S & Ayala-Guerrero F le chat: seuils de stimulation réticulaire et Journal of Comparative Physiology, 157: (1983). Filogenía del sueño en peces y détection des stimuli sonores significants 817-822. anfibios. Boletín de Estudios Medicos y ou non. Journal de Physiologie, 60: 559- 20. Inoue S (1994). Behavioral versus telen- Biologicos, 32 (Suppl 1): 101-107. 560. cephalic sleep. Newsletter of the World 7. Correa SSAL (1995). Conexões intrínsecas 13. McMahon BR & Wilkens JL (1972). Si- Federation of Sleep Research Societies, e extrínsecas da região dorsocentral do multaneous apnea and bradycardia in the 4: 11-12.

Braz J Med Biol Res 32(10) 1999