- Email: [email protected]
Demand-feeding rhythm in rainbow trout and European catfish
Physiology & Behavior 73 (2001) 625 – 633
Demand-feeding rhythm in rainbow trout and European catfish Synchronisation by photoperiod and food availability Vale´rie Bollieta,*, Ana Arandab, Thierry Boujarda a
Laboratoire de Nutrition des Poissons, Unite´ Mixte INRA/IFREMER, Station d’Hydrobiologie, INRA, BP 3, Saint Pe´e sur Nivelle 64310, France b Department of Physiology and Pharmacology, Faculty of Biology, University of Murcia, Campus Espinardo, Murcia 30100, Spain Received 7 December 2000; received in revised form 1 February 2001; accepted 11 April 2001
Abstract The effect of light – dark (LD) cycle and food availability was tested on the demand-feeding rhythm of single and groups of rainbow trout and European catfish. Under LD and free food access, most trout and catfish displayed, respectively, a diurnal and a nocturnal pattern of demand-feeding activity, whereas a few fish or groups of fish switched from diurnalism to nocturnalism or vice versa. In both species held under constant lighting conditions and a restricted feeding (RF) cycle (RF 20:4), the demand-feeding rhythm rapidly synchronised to food availability. The demand-feeding rhythm was under endogenous control and, in rainbow trout, periodogram analysis suggested the existence of two oscillators, one synchronised by photoperiod (LEO) and the other by food (FEO). When submitted to both LD and RF cycles, LD was, at least in the rainbow trout, the dominant zeitgeber synchronising the demand-feeding rhythm. In catfish, food availability rapidly synchronised demand-feeding rhythm. Finally, in both species, the synchronisation of single fish to LD or feed availability appeared slower than that of groups of fish, supporting the idea that social organisation affects the circadian activity in fish. D 2001 Elsevier Science Inc. All rights reserved. Keywords: Feeding behaviour; Demand-feeding; Circadian rhythm; Catfish; Rainbow trout; Photoperiod; Food availability
1. Introduction In fish, like in most vertebrate species, behavioural and physiological processes are rhythmic and the phasing of the rhythm is generally species-dependent. There is an increasing body of evidence that a number of fish species can change their circadian rhythm of behavioural or feeding activity on a seasonal basis [9– 12,16,20,21,27]. Furthermore, some species, such as the sea bass or the goldfish, can spontaneously switch from one type of phasing to another without any apparent changes in their keeping conditions [23 – 25]. Such variability raises the question of the regulation of the circadian system in fish species. In vertebrates, photoperiod is considered as the most important synchroniser of biological rhythm, but periodic food access is also known to synchronise many rhythms [5,19]. Sanchez-Vazquez et al. * Corresponding author. Tel.: +33-5-59-51-59-96; fax: +33-5-59-5451-52. E-mail addresses: [email protected] (V. Bolliet), boujard@st-pee. infra.fr (T. Boujard).
[24] reported in the sea bass that periodic food access could synchronise the demand-feeding rhythm. These authors also suggested that food availability might be one of the factors inducing a shift in phasing and that, as reported in mammals, two different oscillators, one synchronised by photoperiod and the other by food, might entrain the rhythmic pattern of feeding activity in this species [19,24]. However, sea bass is known for its highly flexible circadian system and whether the synchronisation of trophic activity by feeding as well as its endogenous control is similar in more rigid nocturnal or diurnal species remains obscure. In order to answer this question, we investigated the regulation by light –dark (LD) and restricted feeding (RF) cycles of the rhythmic pattern of trophic activity in two teleost species chosen for their contrasted characteristics in feeding behaviour, the rainbow trout Oncorhynchus mykiss, a diurnal species, and the European catfish Silurus glanis, a nocturnal species. The assumption was that LD and RF cycles may regulate in a different manner the rhythmic pattern of trophic activity in nocturnal and in diurnal species. Such study, dealing with biological rhythm in fish, might be performed either on single fish or on groups of fish. But
0031-9384/01/$ – see front matter D 2001 Elsevier Science Inc. All rights reserved. PII: S 0 0 3 1 - 9 3 8 4 ( 0 1 ) 0 0 5 0 5 - 4
626
V. Bolliet et al. / Physiology & Behavior 73 (2001) 625–633
because there is direct evidence that fish circadian activity depends on group size and social organisation [2,13 – 15,20], we ran this experiment on both single fish and groups of fish. The main objective of our study was to determine in both species: (i) the daily pattern of feeding behaviour; (ii) the respective role of photoperiod and food availability in the synchronisation of the demand-feeding rhythm; (iii) the endogenous control of the rhythm; and (iv) the effect of group formation on the synchronisation of the rhythm.
2. Materials and methods 2.1. Animals and housing 2.1.1. Rainbow trout Immature rainbow trout obtained from the experimental fish farm of Donzacq (Landes, France) were kept for 3 weeks in 300-l flow-through tanks supplied with river water prior to experiment. Following this acclimation period, fish of approximately 50 g were randomly selected in order to compose eight groups of 10 trout each. In addition, eight trout were kept single. Groups and single fish were held from January 1998 to April 1998 (93 days) in 16 100-l flowthrough tanks. From the beginning to the end of the experiment, water temperature ranged between 8C and 13C. During free-running conditions, daily fluctuations ranged between 0.2C and 0.7C. 2.1.2. European catfish European catfish of approximately 50 g each were obtained from the same experimental farm. They were randomly divided among eight groups with five catfish in each and eight catfish were kept single. Groups and single fish were held in 16 100-l tanks supplied with re-circulated water maintained at 23 ± 0.2C. The experiment was carried out between the end of June 1998 and the beginning of October 1998 (115 days). 2.2. Experimental conditions In both experiments, the artificial LD cycle (12L/12D, light onset at 0700 h) was programmed by means of an electronic timer. Light was provided by 40-W incandescent bulbs situated 1 m above the water surface. Dim achromatic red light (PF 712B; Philips, France) was used during all periods of darkness. Fish were fed by means of electronic self-feeders (Imetronic Sarl, France), which delivered approximately 0.15 g to individual fish and 0.30 g to groups of fish, each time a fish activated the rod located 1 cm above the water surface. These devices were connected to a computer, as previously described [3]. Each tank was equipped with a sediment trap. The efficiency in collecting waste feed was near 100% for
all tanks. Pellets found in the trap at the outlet of each tank were counted daily. Feed used was a commercial diet (Biomar) containing 42% of protein and 22% of lipids. 2.3. Experimental protocol Experimental conditions are summarised in Table 1. In order to establish the daily pattern of feeding activity, fish were submitted over a 4-week period to a 12L:12D cycle and self-fed ad libitum. Then, to investigate the synchronising effect of food availability on the demand-feeding rhythm, fish were submitted to a time-restricted feeding cycle. Every day, food was available for 4 h, either in the middle of the subjective night or in the middle of the subjective day (RF 20:4). In order to suppress the synchronising effect of photoperiod, fish were held in constant lighting conditions: rainbow trout, considered as a diurnal eater, were maintained under constant light; and catfish, classified as a nocturnal eater, were held in complete darkness. In a third phase, the internal origin of the daily feeding rhythm was tested. Over a period of 2 weeks, rainbow trout were kept under free-running conditions and allowed to eat ad libitum again. Because free-running rhythm in mammals is better expressed under complete food restriction than under free food access [19], catfish were completely fooddeprived for 1 week (free food access being tested at the end of the experiment; see below). Then for both species, in order to determine the respective role of photoperiod and food availability on the synchronisation of the demand-feeding rhythm, fish were submitted to a 12L:12D cycle and allowed to eat for 4 h either in the middle of their natural phase of feeding activity (day for the trout, night for the catfish) or in the middle of their inactive phase. On day 73, catfish were submitted again to complete darkness and periodic food access. Then, to elucidate whether free-running rhythm was, as in mammals, better expressed under complete food restriction than under ad libitum condition, fish were given free food access. At the end of the experiment, catfish were finally held under an LD cycle and given free access to food. Table 1 Experimental conditions Rainbow trout
European catfish
Experimental conditions
Days
Experimental conditions
Days
LD + FF LL + RF 20:4 LL + FF LD + RF 20:4
1 – 29 30 – 64 65 – 81 81 – 93
LD + FF DD + RF 20:4 DD + no food LD + RF 20:4 DD + RF 20:4 DD + FF LD + RF 20:4 LD + FF
1 – 28 29 – 56 57 – 63 64 – 73 74 – 87 88 – 94 95 – 102 103 – 115
LD = light – dark cycle; LL = continuous light; DD = continuous darkness; FF = free food access; RF 20:4 = time-restricted food access.
V. Bolliet et al. / Physiology & Behavior 73 (2001) 625–633
2.4. Data analysis Because of the 1-h recording interval, oscillation patterns of food demands were often irregular and did not allow us to detect significant free-running period. Thus, under freerunning conditions, data for each species were smoothed using a moving average method for a window of five points prior to converting them to a contingency periodogram [17]. This periodogram analysis uses a contingency table containing the states of the variables at different periods. The values in the table thus represent the frequencies fij of the r states of the variable (row i) observed at various times (columns j) of a period T. A different table is constructed for each period considered in the periodogram (e.g., T = 20 h or T = 24 h). The significance of a given period length T is obtained from a c2 table for a critical level a and degrees of freedom r = (r 1)(T 1). Period lengths of the rhythm were determined for data series representing four to six 24-h cycles.
627
It is noteworthy that the use of moving averages can slightly lengthen the period of a series. However, the aim of the experiment was to detect significant rhythms of feeding activity and not to determine with accuracy period length of the free-running rhythms.
3. Results Rainbow trout and European catfish learned rapidly how to operate the feeders (Fig. 1). In both species, no longer than 1 week was necessary to reach a stable feeding level. Food waste, initially observed in all tanks, decreased rapidly. After the acclimation period, uneaten pellet was observed only in tanks where single or groups of fish showed arrhythmic or untypical patterns of feeding activity (see below). There was no mortality during the experimental period.
Fig. 1. Actograms of demand-feeding records of one group of trout (A), one single trout (B), one group of catfish (C), one single catfish (D). Periods of food access during restrictive feeding are delimited by a vertical box. Horizontal bars on top of the figure represent the LD cycle; dark bars represent the dark or subjective dark phase; empty bars represent the light or subjective light phase. Data have been double-plotted for convenient visualisation.
628
V. Bolliet et al. / Physiology & Behavior 73 (2001) 625–633
3.1. Diel pattern of food demands under LD conditions Under LD cycle and free food access (days 1 –29), all groups of trout immediately displayed a very consistent diurnal feeding behaviour (Fig. 1A). In contrast, seven of eight single fish were initially arrhythmic. After 1 week of LD cycle, most of them demanded for food during the light phase (daily diurnal demands: 80– 100%), but two single fish remained arrhythmic and another one displayed a clear nocturnal feeding behaviour until day 29. All singles and five of eight groups of European catfish exposed to LD cycle and self-fed ad libitum (days 1 – 28) initially displayed an arrhythmic pattern of demand-feeding activity (Fig. 1C,D). After 4 weeks of experiment, six of eight groups and seven of eight single fish were feeding mostly during the night. The daily percentage of nocturnal demands in groups and individuals ranged between 80%
and 90% and between 60% and 90%, respectively. The others group or single fish fed mostly during the light phase. From days 93 to 115, groups and single catfish submitted again to the same experimental conditions displayed immediately a clear nocturnal pattern of feeding activity (Fig. 1C,D). Daily percentage of nocturnal demands averaged 80 – 100% in groups and 60– 70% in single fish. 3.2. Time-restricted feeding cycle When submitted to a time-restricted feeding cycle (RF 20:4) and constant light, groups of trout rapidly demanded food during the hours when food was available (Fig. 2A). In contrast, at least 1 week was necessary for groups of catfish to synchronise their demand-feeding rhythm to food availability (Fig. 2C). In both species, synchronisation of the demand-feeding rhythm to food availability was slower and
Fig. 2. Evolution of the daily percentage of rewarded demands in group and single rainbow trout (A, B) and European catfish (C, D) held under RF (20:4) cycle and constant lighting conditions (LL for trout, DD for catfish). Groups and single fish were allowed to feed for 4 h in the middle of the subjective night (full circle) or in the middle of the subjective day (open circle). The initial raw data series was smoothed using a moving-average method with a window of five points. Values are the mean ± S.D. of four replicates.
V. Bolliet et al. / Physiology & Behavior 73 (2001) 625–633
weaker in single fish than in groups (Fig. 2B,D). After 4 weeks of food restriction, while the daily percentage of diurnal demands was similar in single and groups of rainbow trout, the percentage of nocturnal demands made by single catfish remained lower than that of groups. No significant differences were observed between fish allowed to eat during the subjective day or during the subjective night. At the beginning of the restricted period, a few single trout and groups of trout eating in the middle of the subjective night also continued to demand for food during the subjective day as they did before the restriction. The periodogram analysis revealed two period lengths, one of 24 h and the other of 26 h. After 1 week of food restriction, there were no more diurnal demands and only one period
629
(24 h) was revealed by the periodogram analysis. This phenomenon was not observed in catfish. In trout, the daily profile of demands for food over a period of 5 days in the middle of the restriction period is shown in Fig. 3. Regardless of the period of food availability, most demands occurred during the feeding window. In both single and groups of fish, a burst of advanced activity was observed 2– 3 h prior to food access. Similar results were obtained in catfish. Both single and groups of catfish exposed to similar conditions in the second part of the experiment (days 74– 87) synchronised rapidly to food availability. Daily percentage of nocturnal demands averaged 80– 100% in groups and 60 – 70% in single fish.
Fig. 3. Mean ± S.D. diel demand feeding profile over 5 days (days 43 – 47) of groups of (A, C) and single (B, D) rainbow trout maintained under constant light and submitted to RF (20:4) cycle. Period of food access is represented by vertical lines. (A, B) Group and single trout were allowed to eat for 4 h in the middle of the subjective day. (C, D) Group and single trout were allowed to eat for 4 h in the middle of the subjective night. Hatched boxes at the bottom of the figure represent subjective night.
630
V. Bolliet et al. / Physiology & Behavior 73 (2001) 625–633
3.3. Free-running period Under constant light and ad libitum conditions, a demand-feeding rhythm persisted at least for 1 week in 11 of 16 single trout or groups of trout. No clear differences could be observed between single trout and groups of fish. A higher number of fish maintained a rhythm when they were previously allowed to eat during the subjective day rather than during the subjective night (seven and four, respectively). Length periods also presented a higher variability when fish were previously synchronised to the light phase than to the dark phase (23 – 30 and 24– 25 h, respectively). Under constant darkness and free food access (days 88 –94), all groups and single catfish exhibited self-sustained rhythm of demand-feeding activity over a period of 4 – 6 days. The period length ranged from 21 to 26 h regardless of the treatment. In contrast, only three groups and six single fish exhibited a free-running feeding rhythm when submitted to complete restriction (days 57– 63, period length: 20– 26 h). 3.4. LD and RF In trout subjected to LD and RF cycles (days 81 –93), the demand-feeding rhythm was preferentially synchronised to LD cycle (Figs. 1A,B and 4A). When food availability coincided with the natural feeding phase of the fish (food available during the day; Fig. 4A, open square and open circle), demands for food occurred during the feeding window (high percentage of diurnal and rewarded demands). In contrast, when food availability was not in phase with the natural feeding period of the fish (food available during the night; Fig. 4A, full square and full circle), most fish continued to demand for food during the
light phase despite the restriction (high percentage of diurnal demands, low percentage of rewarded demands). When submitted to similar conditions (LD + RF, days 64 – 73), all catfish allowed to eat in the middle of the night (full square, full circle) showed a nocturnal pattern of feeding behaviour and synchronised to food availability (Figs. 1C,D and 4B). Most catfish allowed to eat only during the day (open square, open circle) reversed their natural demand-feeding pattern and became diurnal (high percentage of rewarded demands, low percentage of nocturnal demands). At the end of the experiment (days 95 –102), catfish submitted to similar conditions of LD and RF cycles showed a different feeding pattern (Fig. 1C,D). In contrast to that observed from days 64 to 73, most fish allowed to eat during the day synchronised initially to LD and demanded for food during the night despite the restriction (Fig. 1C). However, they rapidly reversed their feeding pattern and only one group of fish remained nocturnal at the end of the phase. When food restriction was lifted (LD + FF, days 103 – 115), all fish presented a nocturnal pattern of feed demands (Fig. 1C,D).
4. Discussion 4.1. Rhythmic pattern of demand-feeding activity under LD Most trout and catfish submitted to LD cycle and self-fed ad libitum displayed a consistent diurnal and nocturnal feeding behaviour, respectively. These results are in good agreement with previous observations performed in the same species [2– 4,8,22]. However, after 4 weeks of experiment, one single trout presented a nocturnal pattern of
Fig. 4. Percentage of daily rewarded demands (demands that occurred during the feeding window) vs. the percentage of demands that occurred during the light phase (A: trout) or the dark phase (B: catfish). In both cases, fish were submitted to LD cycle and allowed to eat for 4 h either in the middle of the light phase or the dark phase. Open square: isolated fish allowed to eat during the light phase; open circle: groups of fish allowed to eat during the light phase; full square: isolated fish allowed to eat during the night phase; full circle: groups of fish allowed to eat during the dark phase.
V. Bolliet et al. / Physiology & Behavior 73 (2001) 625–633
trophic activity, whereas two single catfish and one group of catfish were strictly diurnal. Such interindividual differences were previously described within a population of rainbow trout by Alana¨ra¨ and Bra¨nna¨s [1]. These results highlight the interindividual variability in feeding behaviour in fish and may suggest that feeding activity in trout and catfish is not as rigidly confined to the light or the dark phase as it is generally admitted. However, in the present study, feeding activity of nocturnal single trout and diurnal single catfish was associated with an important amount of uneaten pellet, whereas almost no feed waste was noticed in fish showing a typical feeding behaviour. A similar result was previously reported in the turbot [6]. In light of these observations, it could be argued that the daily profile of food demands of nocturnal trout and diurnal catfish may be an artifact. Thus, according to current knowledge, concurrent dualism, at least in rainbow trout, is far from being clearly demonstrated. In any case, dualism observed in both catfish and trout appears different from that observed in sea bass or goldfish [23,24,26]. Indeed, in these last species, single fish held in similar controlled conditions can display different feeding behaviour, but also switch spontaneously from one type of phasing to another. Such spontaneous phase shift was not observed in the present study. 4.2. Synchronisation of the demand-feeding rhythm to food availability In fish, the synchronising effect of feeding has been mostly studied on locomotor activity (for review, see Refs. [26,28]). The present results provide direct evidence that food availability is also a good and rapid synchroniser of fish feeding rhythm. This is in agreement with previous findings obtained in the sea bass [24] and suggests that in nature, fluctuations of daily food availability throughout the year might be one of the factors inducing seasonal changes in feeding behaviour.
631
consistent with previous results obtained in trout by Sanchez-Vazquez and Tabata [22]. In contrast to mammals, such differences are frequently observed in fish. Their circadian system is thought to be composed of weak oscillators coupled together and according to SanchezVazquez and Tabata [22], fluctuations in period length may reflect different assemblies of the circadian system during the free-running period. As previously demonstrated in rodents [19], the circadian system of rainbow trout may contain two independent oscillators, one entrained by food (FEO) and the other by light (LEO). Although not yet located in fish, a LEO has been previously evidenced in trout [8,18,22], but nothing is known about the FEO in this species. In the present study, trout held under LL and food restriction exhibited two demand-feeding rhythms with different period, one corresponding to the feeding schedule and the other free-running. These data strongly support the existence of a FEO in addition to the LEO in this species, as also suggested in the see bass and the goldfish [24,26]. Under free-running conditions, it is generally assumed in mammals that animals have a longer persistence of food anticipatory activity when they are food-deprived than when they have free access to food (for review, see Ref. [19]). This contrasts with the present results in European catfish, showing a demand-feeding rhythm better sustained under ad libitum conditions than under continuously food-deprived condition. However, fish given free food access were studied at the beginning of the experiment when they were possibly not fully adapted to their new holding conditions. Indeed, the fact that most catfish held under LD conditions and free food access displayed an arrhythmic pattern of feeding behaviour at the beginning of the experiment (days 1 –28) but not at the end (days 103– 115) suggests that a problem of acclimation might have initially affected their circadian organisation. 4.4. Interactions of food availability and photoperiod on the synchronisation of the demand-feeding rhythm
4.3. Endogenous control of the demand-feeding rhythm Regardless of their period of feeding under restriction (middle of the subjective day or middle of the subjective night), most rainbow trout and catfish maintained their rhythmic pattern of feeding activity when held under freerunning conditions. This result suggests that the feeding rhythm observed under RF conditions was not only a passive response to food restriction, but that food availability directly affected the circadian organisation of the fish and entrained endogenous oscillators generating the demand-feeding rhythm. This contrasts with previous findings in sea bass showing a rapid damping of the oscillations under free-running conditions and supports the idea of a weak coupling of the oscillators in this last species [24]. The variability in period length that appeared both in trout and catfish during the free-running conditions is
When rainbow trout were submitted to both LD and RF cycles, LD was, at least in our experimental conditions, the dominant zeitgeber for the coupled oscillator complex. When the feeding window corresponded to the active phase of the fish (day), both LD cycle and food availability interacted to synchronise the demand-feeding rhythm, and demand for food was mainly concentrated during the hours of its availability. The respective importance of photoperiod and feed in the synchronisation of feeding activity in catfish is far less evident. Food availability appeared to be the dominant factor synchronising the rhythm. This is in good agreement with previous finding obtained in the same species [2] and also in the sea bass [24]. Such results might suggest that feed availability is a more potent zeitgeber in European catfish and sea bass than in rainbow trout. However, it
632
V. Bolliet et al. / Physiology & Behavior 73 (2001) 625–633
should be noted that European catfish submitted to new LD and RF cycles at the end of the experiment (days 95 – 102), entrained initially to photoperiod. Dominance of feed as synchroniser occurred only after a few days. This slower synchronisation to food might be attributed to the important phase shift of the demand-feeding rhythm observed under free-running conditions (days 88– 94) prior to LD and RF. Indeed, the phase shift of the rhythm was more important under free food access than under complete restriction (days 57 –63), suggesting that more time might be necessary in the first case to synchronise the cycle to food again. The different strength of the synchronizer in both species might be related to their feeding behaviour. Indeed, the fact that rainbow trout is a visual eater [3] and that catfish detect food with their barbels and do not depend on the lighting conditions to feed might explain why photoperiod is the dominant synchronizer in the trout while food availability appears to be a more potent zeitgeber in the catfish. Further experiments in other diurnal or nocturnal species would be required to elucidate this point.
European catfish. Whether this observation is related to the nocturnal or the diurnal behaviour of the species remains to be elucidated. The present results emphasize the complexity of the circadian system in fish and the need for further investigations in this field. From a practical point of view, a good understanding of the regulation of fish feeding rhythms would be of prime interest for fish farmers to establish optimal feeding schedule and reduce feed waste and water pollution.
4.5. Effect of groups formation on the synchronisation of feeding rhythm
[1] Alana¨ra¨ A, Bra¨nna¨s E. Diurnal and nocturnal feeding activity in Arctic char (Salvelinus alpinus) and rainbow trout (Oncorhynchus mykiss). Can J Fish Aquat Sci 1997;54:2894 – 900. [2] Boujard T. Diel rhythms of feeding activity in the European catfish Silurus glanis. Physiol Behav 1995;58:641 – 5. [3] Boujard T, Leatherland JF. Demand-feeding behaviour and diel pattern of feeding activity in Oncorhynchus mykiss held under different photoperiod regimes. J Fish Biol 1992;40:535 – 44. [4] Boujard T, Luquet P. Rythmes alimentaires et alimentation chez les siluriformes. Aquat Living Resour 1996;9:113 – 20 (hors se´rie). [5] Boulos Z, Terman M. Food availability and daily biological rhythms. Neurosci Biobehav Rev 1980;4:119 – 31. [6] Burel C, Robin J, Boujard T. Can turbot, Psetta maxima, be fed with self-feeders? Aquat Living Resour 1997;10:381 – 4. [7] Clayton DA. Social facilitated behavior. Q Rev Biol 1978;53:373 – 92. [8] Cuenca EM, de la Higuera M. Evidence for an endogenous circadian rhythm of feeding in the trout (Oncorhynchus mykiss). Biol Rhythms Res 1994;25:336 – 7. [9] Eriksson L. Nocturnalism versus diurnalism; dualism within fish individuals. In: Thorpe JE, editor. Rhythmic activity of fishes. New York: Academic Press, 1978. pp. 69 – 80. [10] Eriksson L, Van Veen T. Circadian rhythm in the brown bullhead, Ictalurus nebulosus. Evidence for an endogenous rhythm in feeding, locomotor and reaction time behavior. Can J Zool 1980;58:1899 – 907. [11] Fraser NH, Heggenes J, Metcalfe NB, Thorpe JE. Low summer temperatures cause juvenile Atlantic salmon to become nocturnal. Can J Zool 1995;73:446 – 51. [12] Heggenes J, Krog OM, Lindas OR, Dokk JG, Bremmes T. Homeostatic behavioural responses in a changing environment: brown trout (Salmo trutta) become nocturnal during winter. J Anim Ecol 1993;62:295 – 308. [13] Kavaliers M. Social groupings and circadian activity of the killifish, Fundulus heteroclitus. Biol Bull 1980a;158:69 – 76. [14] Kavaliers M. Circadian activity of the white sucker, Catostomus commersoni: comparison of individual and shoaling fish. Can J Zool 1980b;58:1399 – 403. [15] Kavaliers M. Period lengthening and disruption of socially facilitated circadian activity rhythms of goldfish by lithium. Physiol Behav 1981; 27:625 – 8. [16] Landless PJ. Demand-feeding behavior of rainbow trout. Aquaculture 1976;7:11 – 25. [17] Legendre P, Fre´chette M, Legendre P. The contingency periodogram:
In both rainbow trout and catfish, synchronisation to photoperiod and food availability was slower in single fish than in groups of fish. These results are consistent with previous findings in European catfish demonstrating that fish held in groups of three or more individuals displayed a typical nocturnal feeding rhythm while isolated fish tended to be arrhythmic [2]. Such phenomenon might be due to interactions between individuals and formation of shoal. It is generally assumed that group formation enables a social facilitation of behaviour [7,13 –15] and the present results support the idea that isolation might affect the circadian activity in fish.
5. Conclusion The present study provides the first evidence for the existence of an FEO in addition to an LEO in rainbow trout. Most rainbow trout and European catfish displayed a diurnal and nocturnal pattern of feeding behaviour, respectively. Some fish were able to switch from one phasing to another, supporting the idea that flexibility is widespread in fish. However, according to current knowledge, the strength of the circadian system appears species-dependent and fish might be classified as mainly diurnal, mainly nocturnal, or dual-phasing species. Feed and photoperiod were rapid synchronizers of the demand-feeding rhythm in trout and catfish. When tested simultaneously, both factors interacted to control the rhythm. Photoperiod appeared to be the dominant synchroniser in rainbow trout while food availability, in our experimental conditions, was a more potent zeitgeber in
Acknowledgments The present research was supported by a National Sciences and Engineering Research Council of Canada grant to V. Bolliet.
References
V. Bolliet et al. / Physiology & Behavior 73 (2001) 625–633
[18]
[19]
[20]
[21]
[22] [23]
a method identifying rhythm in series of non-metric ecological data. J Ecol 1981;69:965 – 79. Martinez-Bebia M, Boujard T, Madrid JA. Comportamiento alimentario de truchas (Oncorhynchus mykiss) expuestas a ciclos de luz/ oscuridad de periodicidad ultradiana. In: Castello F, Calderer A, editors. Actas V Congreso Naciolal Acuicultura. Barcelona: Publicacions de la Universitat de Barcelona, 1995. pp. 546 – 51. Mistlberger RE. Circadian food-anticipatory activity: formal models and physiological mechanisms. Neurosci Biobehav Rev 1994; 18: 171 – 95. Muller K. The flexibility of the circadian system of fish at different latitudes. In: Thorpe JE, editor. Rhythmic activity of fishes. New York: Academic Press, 1978. pp. 91 – 104. Randolph KN, Clemens HP. Some factors influencing the feeding behavior of channel catfish in culture ponds. Trans Am Fish Soc 1976; 6:718 – 24. Sanchez-Vazquez FJ, Tabata M. Circadian rhythms of demand-feeding and locomotor activity in rainbow trout. J Fish Biol 1998;52:255 – 67. Sanchez-Vazquez FJ, Madrid JA, Zamora S. Circadian rhythms of feeding activity in sea bass, Dicentrarchus labrax L: dual phasing
[24]
[25]
[26]
[27]
[28]
633
capacity of diel demand-feeding pattern. J Biol Rhythms 1995a; 10:256 – 66. Sanchez-Vazquez FJ, Zamora S, Madrid JA. Light – dark and food restriction cycles in sea bass: effect of conflicting zeitgebers on demand-feeding rhythms. Physiol Behav 1995b;58:705 – 14. Sanchez-Vazquez FJ, Madrid JA, Zamora S, Iigo M, Tabata M. Demand-feeding and locomotor circadian rhythms in the goldfish, Carassius auratus: dual and independent phasing. Physiol Behav 1996; 60:665 – 74. Sanchez-Vazquez FJ, Madrid JA, Zamora S, Tabata M. Feeding entrainment of locomotor activity rhythms in the goldfish is mediated by a feeding-entrainable circadian oscillator. J Comp Physiol 1997; 181:121 – 32. Sanchez-Vazquez FJ, Azzaydi M, Martinez FJ, Zamora S, Madrid JA. Annual rhythms of demand-feeding activity in sea bass: evidence of a seasonal phase inversion of the diel feeding pattern. Chronobiol Int 1998;15:607 – 22. Spieler RE. Feeding-entrained circadian rhythms in fishes. In: Ali MA, editor. Rhythms in fishes. NATO ASI, Ser A: Life Sci, vol. 236. New York: Plenum, 1992. pp. 137 – 48.
Demand-feeding rhythm in rainbow trout and European catfish Synchronisation by photoperiod and food availability Vale´rie Bollieta,*, Ana Arandab, Thierry Boujarda a
Laboratoire de Nutrition des Poissons, Unite´ Mixte INRA/IFREMER, Station d’Hydrobiologie, INRA, BP 3, Saint Pe´e sur Nivelle 64310, France b Department of Physiology and Pharmacology, Faculty of Biology, University of Murcia, Campus Espinardo, Murcia 30100, Spain Received 7 December 2000; received in revised form 1 February 2001; accepted 11 April 2001
Abstract The effect of light – dark (LD) cycle and food availability was tested on the demand-feeding rhythm of single and groups of rainbow trout and European catfish. Under LD and free food access, most trout and catfish displayed, respectively, a diurnal and a nocturnal pattern of demand-feeding activity, whereas a few fish or groups of fish switched from diurnalism to nocturnalism or vice versa. In both species held under constant lighting conditions and a restricted feeding (RF) cycle (RF 20:4), the demand-feeding rhythm rapidly synchronised to food availability. The demand-feeding rhythm was under endogenous control and, in rainbow trout, periodogram analysis suggested the existence of two oscillators, one synchronised by photoperiod (LEO) and the other by food (FEO). When submitted to both LD and RF cycles, LD was, at least in the rainbow trout, the dominant zeitgeber synchronising the demand-feeding rhythm. In catfish, food availability rapidly synchronised demand-feeding rhythm. Finally, in both species, the synchronisation of single fish to LD or feed availability appeared slower than that of groups of fish, supporting the idea that social organisation affects the circadian activity in fish. D 2001 Elsevier Science Inc. All rights reserved. Keywords: Feeding behaviour; Demand-feeding; Circadian rhythm; Catfish; Rainbow trout; Photoperiod; Food availability
1. Introduction In fish, like in most vertebrate species, behavioural and physiological processes are rhythmic and the phasing of the rhythm is generally species-dependent. There is an increasing body of evidence that a number of fish species can change their circadian rhythm of behavioural or feeding activity on a seasonal basis [9– 12,16,20,21,27]. Furthermore, some species, such as the sea bass or the goldfish, can spontaneously switch from one type of phasing to another without any apparent changes in their keeping conditions [23 – 25]. Such variability raises the question of the regulation of the circadian system in fish species. In vertebrates, photoperiod is considered as the most important synchroniser of biological rhythm, but periodic food access is also known to synchronise many rhythms [5,19]. Sanchez-Vazquez et al. * Corresponding author. Tel.: +33-5-59-51-59-96; fax: +33-5-59-5451-52. E-mail addresses: [email protected] (V. Bolliet), boujard@st-pee. infra.fr (T. Boujard).
[24] reported in the sea bass that periodic food access could synchronise the demand-feeding rhythm. These authors also suggested that food availability might be one of the factors inducing a shift in phasing and that, as reported in mammals, two different oscillators, one synchronised by photoperiod and the other by food, might entrain the rhythmic pattern of feeding activity in this species [19,24]. However, sea bass is known for its highly flexible circadian system and whether the synchronisation of trophic activity by feeding as well as its endogenous control is similar in more rigid nocturnal or diurnal species remains obscure. In order to answer this question, we investigated the regulation by light –dark (LD) and restricted feeding (RF) cycles of the rhythmic pattern of trophic activity in two teleost species chosen for their contrasted characteristics in feeding behaviour, the rainbow trout Oncorhynchus mykiss, a diurnal species, and the European catfish Silurus glanis, a nocturnal species. The assumption was that LD and RF cycles may regulate in a different manner the rhythmic pattern of trophic activity in nocturnal and in diurnal species. Such study, dealing with biological rhythm in fish, might be performed either on single fish or on groups of fish. But
0031-9384/01/$ – see front matter D 2001 Elsevier Science Inc. All rights reserved. PII: S 0 0 3 1 - 9 3 8 4 ( 0 1 ) 0 0 5 0 5 - 4
626
V. Bolliet et al. / Physiology & Behavior 73 (2001) 625–633
because there is direct evidence that fish circadian activity depends on group size and social organisation [2,13 – 15,20], we ran this experiment on both single fish and groups of fish. The main objective of our study was to determine in both species: (i) the daily pattern of feeding behaviour; (ii) the respective role of photoperiod and food availability in the synchronisation of the demand-feeding rhythm; (iii) the endogenous control of the rhythm; and (iv) the effect of group formation on the synchronisation of the rhythm.
2. Materials and methods 2.1. Animals and housing 2.1.1. Rainbow trout Immature rainbow trout obtained from the experimental fish farm of Donzacq (Landes, France) were kept for 3 weeks in 300-l flow-through tanks supplied with river water prior to experiment. Following this acclimation period, fish of approximately 50 g were randomly selected in order to compose eight groups of 10 trout each. In addition, eight trout were kept single. Groups and single fish were held from January 1998 to April 1998 (93 days) in 16 100-l flowthrough tanks. From the beginning to the end of the experiment, water temperature ranged between 8C and 13C. During free-running conditions, daily fluctuations ranged between 0.2C and 0.7C. 2.1.2. European catfish European catfish of approximately 50 g each were obtained from the same experimental farm. They were randomly divided among eight groups with five catfish in each and eight catfish were kept single. Groups and single fish were held in 16 100-l tanks supplied with re-circulated water maintained at 23 ± 0.2C. The experiment was carried out between the end of June 1998 and the beginning of October 1998 (115 days). 2.2. Experimental conditions In both experiments, the artificial LD cycle (12L/12D, light onset at 0700 h) was programmed by means of an electronic timer. Light was provided by 40-W incandescent bulbs situated 1 m above the water surface. Dim achromatic red light (PF 712B; Philips, France) was used during all periods of darkness. Fish were fed by means of electronic self-feeders (Imetronic Sarl, France), which delivered approximately 0.15 g to individual fish and 0.30 g to groups of fish, each time a fish activated the rod located 1 cm above the water surface. These devices were connected to a computer, as previously described [3]. Each tank was equipped with a sediment trap. The efficiency in collecting waste feed was near 100% for
all tanks. Pellets found in the trap at the outlet of each tank were counted daily. Feed used was a commercial diet (Biomar) containing 42% of protein and 22% of lipids. 2.3. Experimental protocol Experimental conditions are summarised in Table 1. In order to establish the daily pattern of feeding activity, fish were submitted over a 4-week period to a 12L:12D cycle and self-fed ad libitum. Then, to investigate the synchronising effect of food availability on the demand-feeding rhythm, fish were submitted to a time-restricted feeding cycle. Every day, food was available for 4 h, either in the middle of the subjective night or in the middle of the subjective day (RF 20:4). In order to suppress the synchronising effect of photoperiod, fish were held in constant lighting conditions: rainbow trout, considered as a diurnal eater, were maintained under constant light; and catfish, classified as a nocturnal eater, were held in complete darkness. In a third phase, the internal origin of the daily feeding rhythm was tested. Over a period of 2 weeks, rainbow trout were kept under free-running conditions and allowed to eat ad libitum again. Because free-running rhythm in mammals is better expressed under complete food restriction than under free food access [19], catfish were completely fooddeprived for 1 week (free food access being tested at the end of the experiment; see below). Then for both species, in order to determine the respective role of photoperiod and food availability on the synchronisation of the demand-feeding rhythm, fish were submitted to a 12L:12D cycle and allowed to eat for 4 h either in the middle of their natural phase of feeding activity (day for the trout, night for the catfish) or in the middle of their inactive phase. On day 73, catfish were submitted again to complete darkness and periodic food access. Then, to elucidate whether free-running rhythm was, as in mammals, better expressed under complete food restriction than under ad libitum condition, fish were given free food access. At the end of the experiment, catfish were finally held under an LD cycle and given free access to food. Table 1 Experimental conditions Rainbow trout
European catfish
Experimental conditions
Days
Experimental conditions
Days
LD + FF LL + RF 20:4 LL + FF LD + RF 20:4
1 – 29 30 – 64 65 – 81 81 – 93
LD + FF DD + RF 20:4 DD + no food LD + RF 20:4 DD + RF 20:4 DD + FF LD + RF 20:4 LD + FF
1 – 28 29 – 56 57 – 63 64 – 73 74 – 87 88 – 94 95 – 102 103 – 115
LD = light – dark cycle; LL = continuous light; DD = continuous darkness; FF = free food access; RF 20:4 = time-restricted food access.
V. Bolliet et al. / Physiology & Behavior 73 (2001) 625–633
2.4. Data analysis Because of the 1-h recording interval, oscillation patterns of food demands were often irregular and did not allow us to detect significant free-running period. Thus, under freerunning conditions, data for each species were smoothed using a moving average method for a window of five points prior to converting them to a contingency periodogram [17]. This periodogram analysis uses a contingency table containing the states of the variables at different periods. The values in the table thus represent the frequencies fij of the r states of the variable (row i) observed at various times (columns j) of a period T. A different table is constructed for each period considered in the periodogram (e.g., T = 20 h or T = 24 h). The significance of a given period length T is obtained from a c2 table for a critical level a and degrees of freedom r = (r 1)(T 1). Period lengths of the rhythm were determined for data series representing four to six 24-h cycles.
627
It is noteworthy that the use of moving averages can slightly lengthen the period of a series. However, the aim of the experiment was to detect significant rhythms of feeding activity and not to determine with accuracy period length of the free-running rhythms.
3. Results Rainbow trout and European catfish learned rapidly how to operate the feeders (Fig. 1). In both species, no longer than 1 week was necessary to reach a stable feeding level. Food waste, initially observed in all tanks, decreased rapidly. After the acclimation period, uneaten pellet was observed only in tanks where single or groups of fish showed arrhythmic or untypical patterns of feeding activity (see below). There was no mortality during the experimental period.
Fig. 1. Actograms of demand-feeding records of one group of trout (A), one single trout (B), one group of catfish (C), one single catfish (D). Periods of food access during restrictive feeding are delimited by a vertical box. Horizontal bars on top of the figure represent the LD cycle; dark bars represent the dark or subjective dark phase; empty bars represent the light or subjective light phase. Data have been double-plotted for convenient visualisation.
628
V. Bolliet et al. / Physiology & Behavior 73 (2001) 625–633
3.1. Diel pattern of food demands under LD conditions Under LD cycle and free food access (days 1 –29), all groups of trout immediately displayed a very consistent diurnal feeding behaviour (Fig. 1A). In contrast, seven of eight single fish were initially arrhythmic. After 1 week of LD cycle, most of them demanded for food during the light phase (daily diurnal demands: 80– 100%), but two single fish remained arrhythmic and another one displayed a clear nocturnal feeding behaviour until day 29. All singles and five of eight groups of European catfish exposed to LD cycle and self-fed ad libitum (days 1 – 28) initially displayed an arrhythmic pattern of demand-feeding activity (Fig. 1C,D). After 4 weeks of experiment, six of eight groups and seven of eight single fish were feeding mostly during the night. The daily percentage of nocturnal demands in groups and individuals ranged between 80%
and 90% and between 60% and 90%, respectively. The others group or single fish fed mostly during the light phase. From days 93 to 115, groups and single catfish submitted again to the same experimental conditions displayed immediately a clear nocturnal pattern of feeding activity (Fig. 1C,D). Daily percentage of nocturnal demands averaged 80 – 100% in groups and 60– 70% in single fish. 3.2. Time-restricted feeding cycle When submitted to a time-restricted feeding cycle (RF 20:4) and constant light, groups of trout rapidly demanded food during the hours when food was available (Fig. 2A). In contrast, at least 1 week was necessary for groups of catfish to synchronise their demand-feeding rhythm to food availability (Fig. 2C). In both species, synchronisation of the demand-feeding rhythm to food availability was slower and
Fig. 2. Evolution of the daily percentage of rewarded demands in group and single rainbow trout (A, B) and European catfish (C, D) held under RF (20:4) cycle and constant lighting conditions (LL for trout, DD for catfish). Groups and single fish were allowed to feed for 4 h in the middle of the subjective night (full circle) or in the middle of the subjective day (open circle). The initial raw data series was smoothed using a moving-average method with a window of five points. Values are the mean ± S.D. of four replicates.
V. Bolliet et al. / Physiology & Behavior 73 (2001) 625–633
weaker in single fish than in groups (Fig. 2B,D). After 4 weeks of food restriction, while the daily percentage of diurnal demands was similar in single and groups of rainbow trout, the percentage of nocturnal demands made by single catfish remained lower than that of groups. No significant differences were observed between fish allowed to eat during the subjective day or during the subjective night. At the beginning of the restricted period, a few single trout and groups of trout eating in the middle of the subjective night also continued to demand for food during the subjective day as they did before the restriction. The periodogram analysis revealed two period lengths, one of 24 h and the other of 26 h. After 1 week of food restriction, there were no more diurnal demands and only one period
629
(24 h) was revealed by the periodogram analysis. This phenomenon was not observed in catfish. In trout, the daily profile of demands for food over a period of 5 days in the middle of the restriction period is shown in Fig. 3. Regardless of the period of food availability, most demands occurred during the feeding window. In both single and groups of fish, a burst of advanced activity was observed 2– 3 h prior to food access. Similar results were obtained in catfish. Both single and groups of catfish exposed to similar conditions in the second part of the experiment (days 74– 87) synchronised rapidly to food availability. Daily percentage of nocturnal demands averaged 80– 100% in groups and 60 – 70% in single fish.
Fig. 3. Mean ± S.D. diel demand feeding profile over 5 days (days 43 – 47) of groups of (A, C) and single (B, D) rainbow trout maintained under constant light and submitted to RF (20:4) cycle. Period of food access is represented by vertical lines. (A, B) Group and single trout were allowed to eat for 4 h in the middle of the subjective day. (C, D) Group and single trout were allowed to eat for 4 h in the middle of the subjective night. Hatched boxes at the bottom of the figure represent subjective night.
630
V. Bolliet et al. / Physiology & Behavior 73 (2001) 625–633
3.3. Free-running period Under constant light and ad libitum conditions, a demand-feeding rhythm persisted at least for 1 week in 11 of 16 single trout or groups of trout. No clear differences could be observed between single trout and groups of fish. A higher number of fish maintained a rhythm when they were previously allowed to eat during the subjective day rather than during the subjective night (seven and four, respectively). Length periods also presented a higher variability when fish were previously synchronised to the light phase than to the dark phase (23 – 30 and 24– 25 h, respectively). Under constant darkness and free food access (days 88 –94), all groups and single catfish exhibited self-sustained rhythm of demand-feeding activity over a period of 4 – 6 days. The period length ranged from 21 to 26 h regardless of the treatment. In contrast, only three groups and six single fish exhibited a free-running feeding rhythm when submitted to complete restriction (days 57– 63, period length: 20– 26 h). 3.4. LD and RF In trout subjected to LD and RF cycles (days 81 –93), the demand-feeding rhythm was preferentially synchronised to LD cycle (Figs. 1A,B and 4A). When food availability coincided with the natural feeding phase of the fish (food available during the day; Fig. 4A, open square and open circle), demands for food occurred during the feeding window (high percentage of diurnal and rewarded demands). In contrast, when food availability was not in phase with the natural feeding period of the fish (food available during the night; Fig. 4A, full square and full circle), most fish continued to demand for food during the
light phase despite the restriction (high percentage of diurnal demands, low percentage of rewarded demands). When submitted to similar conditions (LD + RF, days 64 – 73), all catfish allowed to eat in the middle of the night (full square, full circle) showed a nocturnal pattern of feeding behaviour and synchronised to food availability (Figs. 1C,D and 4B). Most catfish allowed to eat only during the day (open square, open circle) reversed their natural demand-feeding pattern and became diurnal (high percentage of rewarded demands, low percentage of nocturnal demands). At the end of the experiment (days 95 –102), catfish submitted to similar conditions of LD and RF cycles showed a different feeding pattern (Fig. 1C,D). In contrast to that observed from days 64 to 73, most fish allowed to eat during the day synchronised initially to LD and demanded for food during the night despite the restriction (Fig. 1C). However, they rapidly reversed their feeding pattern and only one group of fish remained nocturnal at the end of the phase. When food restriction was lifted (LD + FF, days 103 – 115), all fish presented a nocturnal pattern of feed demands (Fig. 1C,D).
4. Discussion 4.1. Rhythmic pattern of demand-feeding activity under LD Most trout and catfish submitted to LD cycle and self-fed ad libitum displayed a consistent diurnal and nocturnal feeding behaviour, respectively. These results are in good agreement with previous observations performed in the same species [2– 4,8,22]. However, after 4 weeks of experiment, one single trout presented a nocturnal pattern of
Fig. 4. Percentage of daily rewarded demands (demands that occurred during the feeding window) vs. the percentage of demands that occurred during the light phase (A: trout) or the dark phase (B: catfish). In both cases, fish were submitted to LD cycle and allowed to eat for 4 h either in the middle of the light phase or the dark phase. Open square: isolated fish allowed to eat during the light phase; open circle: groups of fish allowed to eat during the light phase; full square: isolated fish allowed to eat during the night phase; full circle: groups of fish allowed to eat during the dark phase.
V. Bolliet et al. / Physiology & Behavior 73 (2001) 625–633
trophic activity, whereas two single catfish and one group of catfish were strictly diurnal. Such interindividual differences were previously described within a population of rainbow trout by Alana¨ra¨ and Bra¨nna¨s [1]. These results highlight the interindividual variability in feeding behaviour in fish and may suggest that feeding activity in trout and catfish is not as rigidly confined to the light or the dark phase as it is generally admitted. However, in the present study, feeding activity of nocturnal single trout and diurnal single catfish was associated with an important amount of uneaten pellet, whereas almost no feed waste was noticed in fish showing a typical feeding behaviour. A similar result was previously reported in the turbot [6]. In light of these observations, it could be argued that the daily profile of food demands of nocturnal trout and diurnal catfish may be an artifact. Thus, according to current knowledge, concurrent dualism, at least in rainbow trout, is far from being clearly demonstrated. In any case, dualism observed in both catfish and trout appears different from that observed in sea bass or goldfish [23,24,26]. Indeed, in these last species, single fish held in similar controlled conditions can display different feeding behaviour, but also switch spontaneously from one type of phasing to another. Such spontaneous phase shift was not observed in the present study. 4.2. Synchronisation of the demand-feeding rhythm to food availability In fish, the synchronising effect of feeding has been mostly studied on locomotor activity (for review, see Refs. [26,28]). The present results provide direct evidence that food availability is also a good and rapid synchroniser of fish feeding rhythm. This is in agreement with previous findings obtained in the sea bass [24] and suggests that in nature, fluctuations of daily food availability throughout the year might be one of the factors inducing seasonal changes in feeding behaviour.
631
consistent with previous results obtained in trout by Sanchez-Vazquez and Tabata [22]. In contrast to mammals, such differences are frequently observed in fish. Their circadian system is thought to be composed of weak oscillators coupled together and according to SanchezVazquez and Tabata [22], fluctuations in period length may reflect different assemblies of the circadian system during the free-running period. As previously demonstrated in rodents [19], the circadian system of rainbow trout may contain two independent oscillators, one entrained by food (FEO) and the other by light (LEO). Although not yet located in fish, a LEO has been previously evidenced in trout [8,18,22], but nothing is known about the FEO in this species. In the present study, trout held under LL and food restriction exhibited two demand-feeding rhythms with different period, one corresponding to the feeding schedule and the other free-running. These data strongly support the existence of a FEO in addition to the LEO in this species, as also suggested in the see bass and the goldfish [24,26]. Under free-running conditions, it is generally assumed in mammals that animals have a longer persistence of food anticipatory activity when they are food-deprived than when they have free access to food (for review, see Ref. [19]). This contrasts with the present results in European catfish, showing a demand-feeding rhythm better sustained under ad libitum conditions than under continuously food-deprived condition. However, fish given free food access were studied at the beginning of the experiment when they were possibly not fully adapted to their new holding conditions. Indeed, the fact that most catfish held under LD conditions and free food access displayed an arrhythmic pattern of feeding behaviour at the beginning of the experiment (days 1 –28) but not at the end (days 103– 115) suggests that a problem of acclimation might have initially affected their circadian organisation. 4.4. Interactions of food availability and photoperiod on the synchronisation of the demand-feeding rhythm
4.3. Endogenous control of the demand-feeding rhythm Regardless of their period of feeding under restriction (middle of the subjective day or middle of the subjective night), most rainbow trout and catfish maintained their rhythmic pattern of feeding activity when held under freerunning conditions. This result suggests that the feeding rhythm observed under RF conditions was not only a passive response to food restriction, but that food availability directly affected the circadian organisation of the fish and entrained endogenous oscillators generating the demand-feeding rhythm. This contrasts with previous findings in sea bass showing a rapid damping of the oscillations under free-running conditions and supports the idea of a weak coupling of the oscillators in this last species [24]. The variability in period length that appeared both in trout and catfish during the free-running conditions is
When rainbow trout were submitted to both LD and RF cycles, LD was, at least in our experimental conditions, the dominant zeitgeber for the coupled oscillator complex. When the feeding window corresponded to the active phase of the fish (day), both LD cycle and food availability interacted to synchronise the demand-feeding rhythm, and demand for food was mainly concentrated during the hours of its availability. The respective importance of photoperiod and feed in the synchronisation of feeding activity in catfish is far less evident. Food availability appeared to be the dominant factor synchronising the rhythm. This is in good agreement with previous finding obtained in the same species [2] and also in the sea bass [24]. Such results might suggest that feed availability is a more potent zeitgeber in European catfish and sea bass than in rainbow trout. However, it
632
V. Bolliet et al. / Physiology & Behavior 73 (2001) 625–633
should be noted that European catfish submitted to new LD and RF cycles at the end of the experiment (days 95 – 102), entrained initially to photoperiod. Dominance of feed as synchroniser occurred only after a few days. This slower synchronisation to food might be attributed to the important phase shift of the demand-feeding rhythm observed under free-running conditions (days 88– 94) prior to LD and RF. Indeed, the phase shift of the rhythm was more important under free food access than under complete restriction (days 57 –63), suggesting that more time might be necessary in the first case to synchronise the cycle to food again. The different strength of the synchronizer in both species might be related to their feeding behaviour. Indeed, the fact that rainbow trout is a visual eater [3] and that catfish detect food with their barbels and do not depend on the lighting conditions to feed might explain why photoperiod is the dominant synchronizer in the trout while food availability appears to be a more potent zeitgeber in the catfish. Further experiments in other diurnal or nocturnal species would be required to elucidate this point.
European catfish. Whether this observation is related to the nocturnal or the diurnal behaviour of the species remains to be elucidated. The present results emphasize the complexity of the circadian system in fish and the need for further investigations in this field. From a practical point of view, a good understanding of the regulation of fish feeding rhythms would be of prime interest for fish farmers to establish optimal feeding schedule and reduce feed waste and water pollution.
4.5. Effect of groups formation on the synchronisation of feeding rhythm
[1] Alana¨ra¨ A, Bra¨nna¨s E. Diurnal and nocturnal feeding activity in Arctic char (Salvelinus alpinus) and rainbow trout (Oncorhynchus mykiss). Can J Fish Aquat Sci 1997;54:2894 – 900. [2] Boujard T. Diel rhythms of feeding activity in the European catfish Silurus glanis. Physiol Behav 1995;58:641 – 5. [3] Boujard T, Leatherland JF. Demand-feeding behaviour and diel pattern of feeding activity in Oncorhynchus mykiss held under different photoperiod regimes. J Fish Biol 1992;40:535 – 44. [4] Boujard T, Luquet P. Rythmes alimentaires et alimentation chez les siluriformes. Aquat Living Resour 1996;9:113 – 20 (hors se´rie). [5] Boulos Z, Terman M. Food availability and daily biological rhythms. Neurosci Biobehav Rev 1980;4:119 – 31. [6] Burel C, Robin J, Boujard T. Can turbot, Psetta maxima, be fed with self-feeders? Aquat Living Resour 1997;10:381 – 4. [7] Clayton DA. Social facilitated behavior. Q Rev Biol 1978;53:373 – 92. [8] Cuenca EM, de la Higuera M. Evidence for an endogenous circadian rhythm of feeding in the trout (Oncorhynchus mykiss). Biol Rhythms Res 1994;25:336 – 7. [9] Eriksson L. Nocturnalism versus diurnalism; dualism within fish individuals. In: Thorpe JE, editor. Rhythmic activity of fishes. New York: Academic Press, 1978. pp. 69 – 80. [10] Eriksson L, Van Veen T. Circadian rhythm in the brown bullhead, Ictalurus nebulosus. Evidence for an endogenous rhythm in feeding, locomotor and reaction time behavior. Can J Zool 1980;58:1899 – 907. [11] Fraser NH, Heggenes J, Metcalfe NB, Thorpe JE. Low summer temperatures cause juvenile Atlantic salmon to become nocturnal. Can J Zool 1995;73:446 – 51. [12] Heggenes J, Krog OM, Lindas OR, Dokk JG, Bremmes T. Homeostatic behavioural responses in a changing environment: brown trout (Salmo trutta) become nocturnal during winter. J Anim Ecol 1993;62:295 – 308. [13] Kavaliers M. Social groupings and circadian activity of the killifish, Fundulus heteroclitus. Biol Bull 1980a;158:69 – 76. [14] Kavaliers M. Circadian activity of the white sucker, Catostomus commersoni: comparison of individual and shoaling fish. Can J Zool 1980b;58:1399 – 403. [15] Kavaliers M. Period lengthening and disruption of socially facilitated circadian activity rhythms of goldfish by lithium. Physiol Behav 1981; 27:625 – 8. [16] Landless PJ. Demand-feeding behavior of rainbow trout. Aquaculture 1976;7:11 – 25. [17] Legendre P, Fre´chette M, Legendre P. The contingency periodogram:
In both rainbow trout and catfish, synchronisation to photoperiod and food availability was slower in single fish than in groups of fish. These results are consistent with previous findings in European catfish demonstrating that fish held in groups of three or more individuals displayed a typical nocturnal feeding rhythm while isolated fish tended to be arrhythmic [2]. Such phenomenon might be due to interactions between individuals and formation of shoal. It is generally assumed that group formation enables a social facilitation of behaviour [7,13 –15] and the present results support the idea that isolation might affect the circadian activity in fish.
5. Conclusion The present study provides the first evidence for the existence of an FEO in addition to an LEO in rainbow trout. Most rainbow trout and European catfish displayed a diurnal and nocturnal pattern of feeding behaviour, respectively. Some fish were able to switch from one phasing to another, supporting the idea that flexibility is widespread in fish. However, according to current knowledge, the strength of the circadian system appears species-dependent and fish might be classified as mainly diurnal, mainly nocturnal, or dual-phasing species. Feed and photoperiod were rapid synchronizers of the demand-feeding rhythm in trout and catfish. When tested simultaneously, both factors interacted to control the rhythm. Photoperiod appeared to be the dominant synchroniser in rainbow trout while food availability, in our experimental conditions, was a more potent zeitgeber in
Acknowledgments The present research was supported by a National Sciences and Engineering Research Council of Canada grant to V. Bolliet.
References
V. Bolliet et al. / Physiology & Behavior 73 (2001) 625–633
[18]
[19]
[20]
[21]
[22] [23]
a method identifying rhythm in series of non-metric ecological data. J Ecol 1981;69:965 – 79. Martinez-Bebia M, Boujard T, Madrid JA. Comportamiento alimentario de truchas (Oncorhynchus mykiss) expuestas a ciclos de luz/ oscuridad de periodicidad ultradiana. In: Castello F, Calderer A, editors. Actas V Congreso Naciolal Acuicultura. Barcelona: Publicacions de la Universitat de Barcelona, 1995. pp. 546 – 51. Mistlberger RE. Circadian food-anticipatory activity: formal models and physiological mechanisms. Neurosci Biobehav Rev 1994; 18: 171 – 95. Muller K. The flexibility of the circadian system of fish at different latitudes. In: Thorpe JE, editor. Rhythmic activity of fishes. New York: Academic Press, 1978. pp. 91 – 104. Randolph KN, Clemens HP. Some factors influencing the feeding behavior of channel catfish in culture ponds. Trans Am Fish Soc 1976; 6:718 – 24. Sanchez-Vazquez FJ, Tabata M. Circadian rhythms of demand-feeding and locomotor activity in rainbow trout. J Fish Biol 1998;52:255 – 67. Sanchez-Vazquez FJ, Madrid JA, Zamora S. Circadian rhythms of feeding activity in sea bass, Dicentrarchus labrax L: dual phasing
[24]
[25]
[26]
[27]
[28]
633
capacity of diel demand-feeding pattern. J Biol Rhythms 1995a; 10:256 – 66. Sanchez-Vazquez FJ, Zamora S, Madrid JA. Light – dark and food restriction cycles in sea bass: effect of conflicting zeitgebers on demand-feeding rhythms. Physiol Behav 1995b;58:705 – 14. Sanchez-Vazquez FJ, Madrid JA, Zamora S, Iigo M, Tabata M. Demand-feeding and locomotor circadian rhythms in the goldfish, Carassius auratus: dual and independent phasing. Physiol Behav 1996; 60:665 – 74. Sanchez-Vazquez FJ, Madrid JA, Zamora S, Tabata M. Feeding entrainment of locomotor activity rhythms in the goldfish is mediated by a feeding-entrainable circadian oscillator. J Comp Physiol 1997; 181:121 – 32. Sanchez-Vazquez FJ, Azzaydi M, Martinez FJ, Zamora S, Madrid JA. Annual rhythms of demand-feeding activity in sea bass: evidence of a seasonal phase inversion of the diel feeding pattern. Chronobiol Int 1998;15:607 – 22. Spieler RE. Feeding-entrained circadian rhythms in fishes. In: Ali MA, editor. Rhythms in fishes. NATO ASI, Ser A: Life Sci, vol. 236. New York: Plenum, 1992. pp. 137 – 48.