Circadian locomotor activity and its entrainment by food in the crayfish Procambarus clarki

It is commonly accepted that overt biological rhythms result from the interplay of endogenous timing mechanisms and a complex host of environmental entraining agents (Pittendrigh, 1981). The characterization of entraining influences and the nature of their interactions with the intrinsic pacemaking systems are critical to the understanding of biological rhythms.

There is ample evidence that food can entrain circadian rhythms (Boulos and Terman, 1980), and its interactions with endogenous rhythms and with light:dark cycles have been described in mammals (Stephan, 1986; Coleman and Francis, 1991; Hau and Gwinner, 1992; Jilge, 1992). In invertebrates, the interaction of food with the time sense of bees has been known for a long time (see Kolterman, 1971), although the information on food as an entraining agent is rather limited.

Although the circadian rhythm of locomotor activity in crayfish has been known for a long time (see Aréchiga and Naylor, 1976), only light:dark cycles have been explored as entraining agents and little information is available about the fundamental properties of the rhythm under free-running conditions (Page and Larimer, 1972). The starting point of these experiments is a series of observations showing an increase in the levels of locomotor activity in crayfish following the supply of food. We have explored the possibility of entraining the locomotor activity rhythm by restricting food delivery. To set a baseline, some experiments were conducted to explore the properties of the rhythm under the same programmes of exposure to light and darkness used when testing for food entrainment.

Experiments were conducted in adult specimens of freshwater crayfish Procambarus clarki of either sex and in intermoult at the time of the experiment. One week before recordings started, crayfish were placed in individual two-compartment recording chambers under 12 h:12 h light:dark cycles of illumination, with 2800 lx as background intensity. Meat pellets were supplied once per week at the light:dark transition time and, once recordings started, animals were no longer fed unless feeding was part of the experiment.

Recordings of locomotor activity were made by using dual-channel recording chambers. These chambers were made of black Perspex, using a modification of the design described by Atkinson and Naylor (1973) to study the activity of Nephrops norvegicus and adapted to crayfish size. One of the compartments was a tunnel simulating a burrow (see Fig. 1), to provide a resting place for the animals (Matsura and Hamano, 1984). The other compartment was a wide chamber. Light was provided by a bulb placed 40 cm above the wide chamber and controlled by a programmable timer. The water level in the chambers was kept at 2.5 cm and water was exchanged via two outlets placed at each extreme of the chamber. Temperature in the chamber was kept at 20±1 °C and monitored continuously.

Movements were detected with an array of five diode–photocell couplers in each of the two compartments. The emission wavelength of diodes was 900 nm, which is far from the range of sensitivity of crayfish visual pigments (Shaw and Stowe, 1982). Signals from each compartment were counted separately. A detailed description of the chambers and of the electronic circuit for activity detection has been given elsewhere (Fernández de Miguel et al. 1989).

The usual format for the analysis of locomotor activity was to express the number of movements of each separate channel in 10 min bins. Locomotor activity levels were different from one day to the next and from one animal to the other. Moreover, under constant conditions it was common to observe that the periodicity of the free-running rhythm changed after 5–7 days. For these reasons, the activity of every crayfish had to be analysed cycle by cycle.

The analysis of rhythmicity was based on the presence of periodic outbursts or ‘peaks’ of activity. Two criteria were used to define activity peaks. (a) The levels of locomotor activity had to be larger than the mean ± standard deviation (S.D.) of basal activity (mostly small movements, such as those of the antennae, or slight displacements within the same compartment). This type of activity appeared like noise in the recordings. (b) Activity peaks appeared over successive cycles with periodicities between 22.0 and 27.0 h. Under LD conditions, out-of-burrow activity was observed at the same time as activity increases inside the burrow. This was also true in the food-related activity peaks that appeared during illumination. For this reason, out-of-burrow activity was used as a good marker of rhythmicity.

To determine periodicities, average values ± S.D. of activity per hour were measured and activity peaks appearing over five or more subsequent cycles were adjusted by linear regression. Within 5–7 cycles, high linear correlations (r>0.95) were found. Correlation coefficients decreased as the number of cycles of free-running rhythms was increased. This was more obvious in food-related rhythms. For this reason, periodicities were determined on the basis of only the first 5–7 cycles.

Shape estimates of locomotor activity rhythms were made by plotting average values ± S.E.M. of activity per hour over several cycles against the period of the rhythm (see Enright, 1981). For the estimate of activity (α) and rest (ρ) quotients, durations of activity peaks on each cycle were added and divided by the duration of ρ, defined as the difference between the length of the period and α.

Locomotor activity was continuously monitored in nine crayfish under a 12 h:12 h light:dark photoperiod. It was a consistent feature that, during illumination, activity was very low and restricted to the burrow (Fig. 1). After the transition from light to darkness, a period of locomotion was initiated. Most of the activity displayed by the animals was observed during darkness, as previously reported (Page and Larimer, 1972). In most of the animals (eight out of nine), locomotor activity consisted of two peaks, one at the beginning of darkness and the second one delayed by several hours relative to the first one (3.44±0.52 h for a total population of 15 cycles). The first activity peak started shortly after the onset of darkness. This peak lasted several hours during the first few cycles (4.41±0.35 h for the whole population) and then gradually decreased (1.12±0.05 h after 10 days). In contrast, the duration of the second peak increased on successive days from 1.1±0.24 h on the first 2 days, to 4.33±1.33 h on the fourth day and to 6.16±0.94 h on day 10 of recording. However, the phase relationship was maintained during this recording time (Fig. 1).

Out-of-burrow activity was present in both peaks and, in many cases, it was the main component of the total activity period (Fig. 1). Moreover, whereas in the open compartment there was virtually no activity during illumination, within the burrow the difference in the amount of activity in darkness and in light was not as clear-cut (Fig. 1). At the onset of light, an activity peak was observed inside of the burrow. It is known that this is a reflex response to the light onset since it disappears under DD and LL (Page and Larimer, 1972). Fig. 1 also shows an estimate of the total activity obtained from seven cycles of five animals. After a variable lapse of 8–20 cycles, the rhythm faded out and activity became irregular, with a reduction of activity in the dark phase and an increase in the light phase.

To test whether the locomotor activity rhythm observed under LD had an endogenous origin, locomotor activity was studied under DD and LL, after crayfish had been entrained to 12 h:12 h light:dark cycles. Changing the programme from LD to DD in five animals caused a free-running rhythm of locomotor activity to appear. Here, activity occurred mostly in the open compartment and this is therefore the only one shown in Fig. 2A. Changes of compartment occurred during activity peaks and therefore could be used again as a good marker of phase. The first activity peak in the dark phase ran freely for 8–20 days, with a period shorter than 24 h (τ=22.3±0.84 h for the total population). A second peak was only apparent in three crayfish, in which it preserved its phase relationship with the first peak (τ=22.5±0.6 h). The total duration of α under DD was 13.5±1.1 h, giving an α/ρ quotient of 1.21. As mentioned above, immediately after the start of DD, the peak corresponding to the time of light onset disappeared.

In a second series of experiments, the LD program was changed to LL with an intensity of 2800 lx. Only in five out of nine animals was a free-running rhythm detected during 5–9 cycles. After this, the rhythm died out and activity appeared at random. In contrast with observations made in DD, the total amount of locomotor activity in LL was greatly diminished during the first few cycles, and the periodicity of the discernible onset of activity was delayed (τ=24.8±0.27 h for five animals). In no case was out-of-burrow activity detected, even during activity peaks. For this reason, Fig. 2B shows only activity recorded within the burrow. The total duration of α for the free-running rhythm in LL was 6.12±0.2 h, giving an α/ρ quotient of 0.34.

The possibility of food influencing an activity rhythm was tested in 35 animals, exploring different combinations of light and darkness: (a) food delivered in LD; (b) food delivered in DD; and (c) food delivered in LL.

In crayfish that had been entrained to LD conditions and deprived of food for 1 week, the delivery of a single food pellet to the open compartment evoked an attraction response. Within a few minutes after the food delivery, crayfish emerged from the burrow, took the pellet, walked back to the burrow and started eating. Remaining portions of the pellet were dropped outside the burrow. After some trials, it was found that if the pellet weight was about 10 % of the body weight of the animal, it was totally eaten in less than 1 h. Consequently, this size of pellet was chosen for the rest of the experiments. The completion of the sequence of movements leading to the location, catching and carrying of food took several minutes. It was also observed that, even after the animal had eaten the pellet, it kept walking in and out of the burrow for several hours.

It is noteworthy that crayfish responses to food supply were unpredictable and capricious. Sometimes animals that had not eaten for a week or more did not respond to food delivery. In contrast, others that had eaten the previous day continued displaying the whole feeding behaviour described above. Only data from animals that were responsive to food delivery have been included in the present report.

The effect of food delivery was first tested at the beginning of the dark phase of the activity rhythm. When a single food pellet was administered in the open space simultaneously with the offset of light, excitation and all the feeding-related events mentioned above were triggered. As can be seen in Fig. 3, the burst of activity in the open space was markedly increased for several hours. In none of the animals were noticeable changes observed in activity during the next cycles (not shown). In contrast, when feeding was repeated for consecutive days, reductions in the level and duration of the locomotor activity peak were noted in five out of six animals. Here, the light-evoked burst of activity in both compartments became quite prominent. Fig. 3 shows responses obtained in one of these tests. The absence of food on subsequent days restored the original levels of locomotor activity and the LD-entrained rhythm persisted again for several cycles (only one is shown in Fig. 3).

In 20 animals deprived of food for 1 week, a single pellet of meat was supplied during the illumination period of the light:dark cycle. In these experiments, food induced the usual enhancement of locomotor activity during the ensuing hours and the whole pattern of feeding behaviour was reproduced. In nine cases, anticipatory reactions were observed in the next cycles. Interestingly, in 12 animals a peak of activity continued to appear during the ensuing days, at approximately the time of food administration. In five animals in which food was supplied 6 h after the onset of light (Fig. 4), a rhythm was followed for at least 5 days with a periodicity of 24.10±0.37 h. After 5–7 days, the period increased to 24.87±0.42 h. As observed in the light-entrained rhythm, ‘food-induced’ activity peaks were accompanied by out-of-burrow activity in most of the cases. The amplitude and the duration of these peaks were far smaller than the corresponding values for the darkness activity peaks described above, and they were therefore difficult to distinguish in the noisy recordings. For this reason, the presence of out-of-burrow activity was again taken as a good phase marker when analysing periodicities. There was also a peak of activity during the dark phase of the cycle. Over successive days, this peak appeared in phase with the food-related peak instead of remaining phase-shifted to the beginning of darkness. In contrast, the reflex peak at the onset of light remained unchanged. Fig. 5 compares shape estimates of locomotor activity rhythms of a crayfish under LD conditions (Fig. 5A–C) with that of crayfish fed 6 h after the onset of light (Fig. 5D–F). The various peaks of activity can be clearly distinguished in the out-of-burrow activity panels.

A single food delivery was also tested at different times of the dark phase in five crayfish. Although food induced an increase in the levels of locomotor activity during the first night, no noticeable changes in the phase of the rhythm or the activity levels were observed in subsequent cycles.

From observations under LD cycles, it seemed feasible to evoke an activity peak in phase with the time of food delivery, but the presence of a free-running rhythm of locomotor activity, as we had seen in our control experiments, could obscure the analysis. A change was introduced in the experimental protocol and the rhythm was allowed to run freely until it damped out. As described above, this usually took 10–15 days. Once the rhythm became undetectable, food was supplied. Shortly after this, there was an increase in activity lasting several hours. On the following days, the onset of activity appeared in phase with the food-induced outburst of activity. Fig. 6 shows the recording from an arrhythmic animal after 15 days under DD (the last 24 h of recording before food delivery are shown as day I). The period of this rhythm was longer than 24 h (26.2±0.12 h, for a population of seven cycles in each of eight animals). A second peak was detected in five animals with a similar period to that of the first peak. The food-induced rhythm ran for 7–14 cycles. In these cases, no dependence on the time of delivery was noted. Fig. 6 also shows a form estimate of this rhythm, considering the average periodicity for seven cycles for eight animals. In this case, time zero was taken to be the beginning of the food-induced activity peak at every subsequent cycle. Therefore, time zero also reflects the period estimated for this rhythm (26.2 h).

A group of nine animals was subjected to a programme of continuous illumination at 2800 lx, while activity was monitored. When the free-running rhythm could no longer be detected, a single meat pellet placed in the open space produced a peak of locomotor activity with a variable delay (from about 30 min to several hours). As had occurred in the light-entrained rhythm, out-of-burrow activity was totally inhibited and the animals, therefore, did not eat the pellet. Surprisingly, during the following days, a rhythm was detected in five of the animals, with a period shorter than 24 h (22.5±0.46 h). In two of the animals, a single small activity peak was detected after the food delivery, which lasted for as long as 10 days (Fig. 7). A form estimate of this rhythm is also shown in Fig. 7. Again the period (22.5 h) was considered to be time zero. Average values per hour of locomotor activity over 5–7 days have been pooled for all five animals.

The features of the crayfish activity rhythm that we observed under LD conditions are, in general, similar to those described by Page and Larimer (1972). The presence of two activity peaks during the dark phase of the cycle has not been previously reported in crustaceans, although it resembles observations made in other species (see Aschoff, 1966). Both activity peaks were also observed in constant darkness with a similar period; therefore, it is likely that they are driven by the same circadian pacemaker.

The weakness of the circadian rhythm has also been noticed by other authors, especially under constant illumination, where no free-running rhythms had been detected (Page and Larimer, 1972). Differences between previous observations and the data reported here under LL conditions could be due to the design of our recording chambers, which allow burrowing behaviour (Atkinson and Naylor, 1973; Matsura and Hamano, 1984). Since the initial burst of activity at the onset of illumination disappears under DD or LL, we agree with Page and Larimer (1972) that this activity is a reflex response to light, unrelated to the circadian rhythm of activity.

Values of circadian periodicity and α/ρ quotients for the free-running rhythm under DD and LL conform to the empirical Aschoff’s rule (see Hoffman, 1965), according to which, for nocturnally active species such as crayfish, the periodicity would be expected to shorten and the α/ρ quotient to increase under DD, and the opposite to happen under LL. It is interesting to point out that the values in our experiments are similar to those reported for circadian rhythms in the visual system of the crayfish (Aréchiga et al. 1973) and for other crustacean rhythms (see De Coursey, 1983).

Evidence presented here strongly suggests that food may act as an entraining agent. The presence of a locomotor activity peak triggered by a single food delivery and reappearing over the next 2 weeks suggests that food delivery is controlling the phase of an activity rhythm. One possibility is that food induces advances or delays of phase in an existing rhythm. To confirm this it would be necessary to determine the phase–response curves for both rhythms. An attractive speculation based on experiments presented here is that food supply after deprivation entrains a circadian activity rhythm and induces locomotion, whereas satiety inhibits locomotion and feeding behaviour. This would explain the desynchronization of the LD-entrained rhythm in animals without a food supply, the appearance of activity peaks days after delivery, and the reduction in the levels of locomotor activity after several subsequent days of reinforcement. This dual effect of food supply, together with the irregularities in crayfish feeding behaviour stated above, make it difficult to evaluate whether food induces stable phase relationships and the control of period length during locomotor activity. However, the appearance of a rhythm after food delivery in the absence of other environmental cues and its persistence with its own free-running period allow us to propose that food is an entraining agent. It must be emphasized that food has a strong influence on the locomotor activity rhythm of crayfish, since a single supply may exert a long-lasting effect. This, by itself, could be a criterion for proposing food as an entraining agent of crayfish locomotor activity.

The observation that the food-related rhythm exhibits a period-dependence during continuous darkness or light, a response opposite to that found for the LD-entrained rhythm, is not readily explainable. One possibility is that, as has been suggested for other species, the oscillator driving the food-related rhythm is different from that driving the LD-entrained rhythm (Stephan, 1984; Weaver and Reppert, 1989). Interactions between these (dual or multiple) systems would also explain the coexistence of the food-related and darkness peaks of activity, which remain phase-shifted under LD conditions, and the influence of light and darkness on the food-related rhythm. The possibility of a dual-clock system in crustaceans has been proposed on several grounds and for several rhythms, since the earlier postulation by Brown and Webb (1949) for the fiddler crab Uca pugilator, and includes more recent reports in other species (Palmer and Williams, 1986; Reid and Naylor, 1990).

The free-running rhythm under LL in the presence of food but in the absense of eating suggests that the initial stimulus for food entrainment is chemical. Unpublished observations have shown that a food homogenate is able to reproduce this effect (F. Fernández de Miguel and H. Aréchiga). Several chemical substances present in meat extracts have been shown to induce strong attractant motor reactions in various crustacean species (Carr and Derby, 1986), and specific chemosensitive neurones have been identified (Ache, 1982). It remains for further studies to determine the precise nature of the entraining agent(s) and the possible neural pathways underlying this complex behavioural integration.