The Locomotor Activity Rhythm of Talitrus Saltator (Montagu) (Crustacea, Amphipoda)

Previous studies on the sun-orientation of Mediterranean Talitrus saltator (Montagu) have revealed the presence of an internal timing mechanism which operates at solar frequency (Papi & Pardi, 1953, 1959; Pardi, 1960). Transfer of animals to a longitude different from that where they originated showed that the angle between the sun and the direction of escape at a given time is fixed for a given population (Papi, 1955). The timing basis of the navigational mechanisms in Talitrus and in Talorchestia deshayesi (Audouin) can be re-set by artificially changing the light/dark cycles. Thus animals subjected to a light/dark cycle delayed by 6 h show an angle of escape which is at approximately 90 ° to that of controls (Pardi & Grassi, 1955). These authors also showed in Talitrus that raising the temperature from 35 to 37 °C slightly accelerated the timing process which controls the angle of orientation, and similar effects are reported in Orchestia platensis Kröyer, even in regular light/dark conditions (Jankowsky, 1969). There is therefore a great deal of evidence to suggest that talitrids which navigate using a sun-compass mechanism have an endogeneous time-sense.

In British localities there is so far little evidence of a time-sense in Talitrus since its navigational ability seems more largely based on visual orientation to the silhouettes of sand-dunes (Williamson, 1951). It therefore seemed worth while to investigate locomotor rhythmicity in this species and to assess the persistence of any endogeneous component of the rhythm.

Talitrus saltator is a semi-terrestrial amphipod inhabiting non-permanent burrows in sand somewhat above high-water mark, so that burrows are found lower down the shore during neap tides than on springs. The animals emerge at dusk and forage between the level of the burrows and the water’s edge. When seaweed, upon which they largely feed, has been washed up in sufficient quantity, they do not move so far, most remaining around the strand-line. Geppetti & Tongiori (1967) also report variability in nocturnal migrations in Mediterranean Talitrus, depending upon availability of food at the water’s edge. The same authors report that high temperature and low humidity may inhibit migration. Present observations suggest that surface activity is also reduced during rain. Around dawn, animals return to the burrow region, perhaps partly guided by solar navigation, but mainly by visual stimuli from silhouettes of sand-dunes (Williamson, 1951).

On the coast of South Wales animals overwinter in sand well above high water, at depths from 30 to 100 cm. They appear on the shore during nights about the end of April and remain active at night until about late October. Pallault (1954) considers a temperature of 10 °C to be critical for the seasonal activity of Normandy populations, and present observations suggest that this is also true for South Wales populations. Animals maintained throughout the winter in the laboratory at room temperature and under natural light regimes displayed normal activity patterns. Normal behaviour was also shown by animals collected from the field during the winter and brought into the laboratory.

Experimental animals were collected by night at Oxwich Bay, near Swansea. They were located with a torch and picked up by hand. They survived well in the laboratory in glass tanks containing moist sand and under normal light conditions. They were fed occasionally with seaweed and pieces of apple. Large males, easily distinguished by long, orange first antennae, were used in all experiments unless otherwise stated.

The actographs were rectangular boxes 75× 150×250 mm constructed of 3·2 mm perspex and painted black on the outer surface. Each box contained two wells 75× 75 mm filled with moist sand to a depth of 75 mm and into which the experimental animals were able to burrow. The two wells were separated by a perspex platform at the same level as the sand surface. A dim red light beam passed across the platform and was focused sharply by a lens (focal length 10 cm) on to a phototransistor (OCP 71) taped to the outside of the box. The perspex platform was necessary to prevent animals throwing sand against the photo-transistor while burrowing. The light source was a 6·5 V, 0·3 A bulb, painted black except for a small slit and fitted with a red filter. The intensity of the light was controlled by a variable resistor of 100 Ω and a very dull light was found to be adequate.

The photo-transistor was connected to a circuit slightly modified from that of Southward & Crisp (1965), so that when an animal crossed the platform and interrupted the light beam, the resulting change in resistance of the photo-transistor was recorded as a mark upon the moving chart-paper of a ‘Rustrak’ event recorder. Power for the circuit was provided by a mains-operated transformer producing a rectified current at 12 V. Greatest sensitivity of the photo-transistor was found to occur with minimum power supply, achieved by use of a variable resistor of 1 kΩ.

Each actograph had a well-fitting clear perspex lid, and the experiments were carried out in a constant temperature room, at 15 °C unless otherwise stated, usually with a dim red background illumination of immeasurably low intensity.

Preliminary experiments showed that the most consistent results were obtained using groups of four animals in each actograph. Activity is expressed as number of events for each hour, usually in histogram form. Also, on account of the clear-cut and unimodal nature of the activity rhythm, it is possible to calculate the time of the mid-point of an activity period, and this is given, when appropriate, to the nearest 0·01 h. All times are expressed as G.M.T.

Many experiments were conducted in dim red light, and the animals showed a clearcut unimodal circadian rhythm which persisted for as long as the experiments were run (up to 46 days). In the longer experiments there was some diminution of activity after the first 2 weeks or so. In most cases the interval between successive activity peaks was slightly more than 24 h, and there was no sign of a twice daily or tidal component.

Fig. 1 shows the activity of a group of four animals over 46 days. For the first 21 days the mean period is 24·49 h (S-D-± 0·49). After about this time there is a clear increase in mean period (25·49 h, S.D. ±2·10) and the rhythm is not so precise in appearance or regular in period.

The percentage total hourly activity for each hour of the first activity period of 20 groups of four animals in constant dim red light is given in Fig. 2. Maximum activity occurs between midnight and 01.00 h. These experiments were carried out between late May and late November, and no seasonal differences in activity patterns were noted.

To test the effects of continuous white light three groups of four freshly collected animals were placed in actographs at 15 °C and 0·9 m vertically below a 100 W light bulb (about 200 lux). Activity was recorded for 5 days and then for a further 3 days in dim red background illumination. A control actograph was maintained in continuous darkness throughout by means of light-proof covers. The results are given in Fig. 3, in which the control animals show a normal activity rhythm, with a period of slightly more than 24 h (Fig. 3,a). The activity of those under constant illumination, however, is almost entirely suppressed in two cases. In the third actograph (Fig. 3 d) duration of activity is restricted and its onset is delayed by an hour or two, but a circadian rhythm is still apparent. In the 3 days under dim red illumination following continuous light the period and approximate phase of the rhythm are seen to have been maintained, despite the previous suppression of activity by light. The inactivity of animals under continuous light is probaby accounted for by there being a threshold value of light intensity, above which the animals remain in their burrows. Moreover, since there was a little surface activity during the light-on phase of these experiments, it is possible that the threshold may be in the region of the light intensity used, i.e. 200 lux.

In another experiment run simultaneously, modified actographs were used in which the perspex platform extended throughout the box and there was no sand available for burrowing. A few small pieces of moist cotton wool were introduced to maintain humidity and to provide a substratum on which the animals could rest when inactive. The results are given in Fig. 4. The control group of animals in continuous darkness show a normal circadian activity rhythm (Fig. 4 a), but the rhythm of the three groups of animals under continuous illumination is somewhat different. With these groups the onset of activity is at first somewhat delayed, but thereafter the activity rhythm has an average period which is clearly less than 24 h. The mean periods for two sets of experimental animals in Fig. 4(b, c) are 22·2 h and 23·08 h, while that for the controls (Fig. 4 a) is 24·36 h. In the subsequent continuous dim red light the advanced timing of the rhythm is maintained relative to the control animals.

The animals in these modified actographs under continuous illumination were unable to avoid the light by burrowing, and this apparently resulted in the different period of the rhythm. The advance of the timing of the rhythm does not appear to be complete with one group of animals (Fig. 4d), perhaps due to a slower response by one or two of the four animals tested. The bimodal pattern persists in subsequent constant dim red light.

To test for the effectiveness of light in synchronizing the rhythm four freshly collected animals were placed in an actograph with sand 0·9 m below a 100 W lamp (about 200 lux) and subjected to a regime of 12 h light and 12 h darkness with the onset of darkness beginning at 12 noon. The mid-darkness point was thus 6 h in advance of normal, and the onset of darkness was about 8 h in advance of normal. Another actograph was maintained under the same light regime, but 1·2 m horizontally from the first one, at about 40 lux. A control actograph was kept beneath lightproof covers. After 3 days of experimental light/dark, the actographs were recorded in dim red light for 3 days, and the mean hourly activity over 24 h during this time is shown in Fig. 5 a. The timing of the rhythm of the animals exposed to the new light regime using 200 lux is largely rephased to the ‘expected’ night, but that of the animals at 40 lux is virtually unchanged and close to that of the controls.

In another experiment run simultaneously, modified actographs were used. These provided no sand in which the animals could burrow but small pieces of moist cotton wool were introduced to maintain humidity. Mean activity over 24 h for the 3 days of constant dim red background light, following the experimental light/dark treatment as described above, is shown in Fig. 5(b). Animals in the actographs at 200 lux and 40 lux both re-synchronized their activity rhythms to match the new light regime. Both these experiments were repeated twice, with similar results. Clearly light is a major entraining factor for Talitrus. The failure of the animals to re-synchronize when in sand under intermittent dim light at 40 lux is no doubt accounted for by their being in burrows. They were probably unable to detect the light-intensity difference during the change from light to darkness. All these experiments suggest that under normal conditions Talitrus relies on a light/dark cycle with a light-phase threshold of between 40 and 200 lux.

In initial experiments to test the effects of temperature on the rhythm three actographs were maintained in three separate constant-temperature rooms at 15, 20 and 25 °C, and in continuous dim red light. The activity of four animals in each actograph was recorded simultaneously for 15 days. A further actograph was maintained under natural light conditions at temperatures which fluctuated between 17 and 22 °C. At the time of collection of the animals in the field the air temperature was 10-5 °C, and they were introduced into the actographs within 3 h of this time.

The results, expressed as time of mid-activity points on successive days, are given in Fig. 6. The animals at fluctuating temperatures and in normal light/dark conditions are seen to maintain the timing of their rhythm, with mid-activity around midnight. The animals at 15 °C likewise maintain the timing of their rhythm, showing, in this case, no appreciable drift, probably because they were subjected to a 4·5 °C rise in temperature on being brought in from the field. Both groups of animals at the higher temperatures have an activity mid-point which is advanced to well before midnight during the first few days of the experiment, but after 10 days there is very little difference between them and the other groups of animals. On day 8 the 15 °C constanttemperature room suffered a mechanical failure, and between 03.00 h and 15.00 h the temperature rose to 20 °C. An advance of about 3 h occurred in the succeeding midactivity point, but this again was compensated for within two days.

In another experiment at 15 °C in continuous dim red light, a group of animals was transferred to a constant-temperature room at 22·5 °C from 13.00–19.00 h on day 4 and then returned to 15 °C. An advance of about 3 h occurred in the succeeding mid-activity point (Fig. 7), but after 4 days that of both control and experimental groups of animals occurred at about the same time. These results seem to suggest that an increase in temperature results in a temporary shortening of the period of the rhythm, when an apparently temperature-sensitive part of the timing mechanism runs fast for a few days. Compensation occurs possibly by a more basic temperature-independent timing mechanism, and this seems to take somewhat longer if the animals are maintained at the increased temperature.

To assess the role of temperature as a synchronizer of the rhythm the effects of regular temperature fluctuations in constant dim red light were tested by subjecting two actographs, each with four animals to alternating temperatures at 15 and 25 °C at 12 h intervals, so that one actograph was out of phase with the other. This involved carrying the actographs a few yards between two constant-temperature rooms. A control actograph was maintained at 15 °C and was disturbed in the same way as the experimental actographs at 12 h intervals. The changes took place at 08.00 and 20.00 h for 4 days. The animals exposed to the higher temperature during the day (Fig. 8a) do not show an advance in the timing of their rhythm to the same extent as has been reported above. However, the timing of the rhythm remains nearly constant for the 4 days, unlike that of the control animals whose rhythm drifted by about 4 h (Fig. 8 c). There is no difference between the control animals and those exposed to a reversed temperature cycle, with higher temperatures during the expected night. It seems that a temperature cycle with higher temperatures during the day may help to maintain the timing of the locomotor rhythm, but a temperature cycle which is the reverse of this, even with a difference of 10 °C, does not cause re-synchronization of the rhythm. Longer-term experiments may prove otherwise.

Single animals in the actograph displayed locomotor rhythms whose period and duration often differed markedly from one another. Activity was reduced, and sometimes even absent, for the entire duration of the experiment. In all these cases the animals were healthy at the end of the experiment. On the other hand, groups of four animals gave consistent results. The mid-activity points of 15 single animals and 8 groups of four animals in experiments of 10 days’ duration in constant dim red light are given in Fig. 9. In each case the mid-activity points follow a pattern to which a linear regression line can be fitted to confirm a constant rate of drift of activity peaks. The two regression coefficients are significantly different (P < 0·001) with the single animals losing time (0·87 h/day) twice as fast as the groups of four animals (0·39 h/ day). It seems likely that the timing mechanisms of single animals have periods which differ between individuals, and that the animals tend to influence one another so that as a group they phase their activity peaks to the timing of those animals showing least drift.

Groups of 8, 16 and 32 animals in the usual actographs all show much less drift of activity peaks than single animals. The results of an experiment with a group of 4 and a group of 32 animals is shown in Fig. 10. The phasing of the two activity patterns are very similar, each with a period of only slightly more than 24 h. A further point emerges from these results in that the amount of activity recorded by the 32-animal group in each spell of activity is much less than eight times that of the 4-animal group, so it is reasonable to suggest that a high density of animals will to some extent inhibit the activity of individuals. In most cases the time at which activity ceases is somewhat later in the 32-animal group, and this is probably because it takes individuals longer to find an unoccupied burrow or find space to make a new one.

In view of the differences in period recorded between the activity rhythms of single animals and groups of animals, and the inference that individuals are able to influence the time-keeping abilities of one another, attempts were made to re-synchronize the locomotor rhythm of Talitrus by contact with other animals possessing rhythms that were out of phase. The fact that chilling can apparently retard or even stop the timing mechanism of Talitrus (Bregazzi, 1972) makes it reasonably easy to obtain two groups of animals, each possessing a rhythm which is out of phase with the other. In preliminary experiments, differences in phase between ‘chilled’ and ‘normal’ animals were maintained in separate actographs for up to 19 days.

The term ‘mutual entrainment’ has been used previously to denote the entrainment of more than one supposed internal timing mechanism within a single animal by mutual interaction (Harker, 1964). In the present account the term is used to denote the entrainment of individual animal’s rhythms by contact with other rhythmically active individuals. A number of different experiments failed to reveal an unmistakable case of mutual entrainment. In the first experiment three animals, entrained to natural light conditions, were introduced into an actograph together with three animals that had been chilled so that their locomotor activity pattern was 12 h out of phase with the first group. Two such actographs, together with controls of normal and chilled animals, were monitored in dim red background illumination for 15 days (Fig. 11). At first there are two distinct peaks per day, each following closely the phase of the control animals as indicated by the subjective shading on the histograms. Thereafter, activity declines and in one case is completely absent for a few days. It is then resumed at the previous level, but with each peak in an intermediate position between the two controls and difficult to relate by shading. It seems that two groups of animals that are sufficiently out of phase with one another can maintain the asynchrony of their locomotor rhythms, even in the same actograph. While one group was active, the other was buried in the sand, and the two groups need not necessarily have come into contact at all. It remains to investigate the possibility that the two groups of animals interacted to result in an overall reduction of activity, while the subsequent reappearance of a phased-shifted rhythm also requires investigation.

Since in the experiments so far described the animals could avoid each other in sand, additional experiments were carried out using 12 cm glass dishes fitted with perspex lids and containing only about 3 mm of damp sand, into which animals could partly burrow when inactive, but which was not deep enough for them to disappear entirely beneath the surface. Normal and chilled animals in the same dish were identified by being of different sex, or by the tip of one first antennal flagellum being removed. Animals so treated were allowed 3 or 4 days to recover from wounding, and behaved normally. The animals were tested for 3 days in actographs following some days in the experimental dishes. All experiments were conducted under dim red background illumination.

The number of experiments carried out, together with their duration, the numbers of animals used and the differences in phase are given in Table 1. In no case did any animal obviously alter its phase to that of the group in contact. A typical result, in which groups of ‘normal ‘males were in contact for 4 days with chilled males 8 h out of phase, is given in Fig. 12.

An additional point of interest has emerged from these experiments. It was found that the rate at which the locomotor activity peaks of animals in the dishes with little sand drifted across natural time varied considerably and was usually at a greater rate than in those animals kept in sand into which complete burrowing was possible. The difference in period between different groups of animals was not great enough to affect the interpretation of the results if the time spent in the dishes was not more than a few days, but it imposed a serious limitation upon this method for longer-term investigations of mutual entrainment. In one experiment two control dishes, both containing groups of animals with naturally synchronized rhythms to begin with, were 8 h out of phase with each other after 17 days.

The locomotor activity rhythm of Talitrus under conditions of constant temperature and dim red background illumination has a regular circadian period of somewhat more than 24 h. It persists for several weeks, which indicates that it has a very significant endogeneous component. The shortening of the period of the rhythm under constant illumination (Fig. 4) is contrary to Aschoff’s rule (Pittendrigh, 1960), which states, in part, that the rhythms of dark-active species have a shorter period in constant darkness than in constant light and that the period of the rhythms in constant light increase with increasing intensity of illumination. Another probable reversal of Aschoff’s rule has been reported for Aedes aegypti (L.) (Taylor & Jones, 1969).

Present experiments show that light-intensity cycles are likely to be the chief, and perhaps the only, entraining agent for Talitrus under natural conditions. A change in time of activity to coincide with the dark period, under artificial light conditions 6 h out of phase with normal daylight, has also been reported by Pardi & Grassi (1955).

Ercolini (1960) describes the locomotor activity of Mediterranean Talitrus as nocturnal under normal light/dark conditions and, in some populations, as having possible bimodal characteristics. No such bimodal pattern occurred in the locomotor rhythm in the present study under constant conditions (Fig. 2), and if it does exist in the field, it may be due to activity representing the search for a suitable burrowing substrate as dawn approaches. The difference between Mediterranean and South Wales populations may also be related to the different ecological situations in these regions. In South Wales, Talitrus is a largely intertidal scavenger; in the Mediterranean there is a restricted intertidal region and animals move inland to feed, returning to burrow in sand near the water’s edge at dawn. Furthermore, due to differences in latitude, the summer-active South Wales populations experience a shorter night than that experienced by the Mediterranean populations, and this may have restricted the development of a bimodal activity pattern. Geppetti & Tongiori (1967) suggest that sunrise causes the seaward migration in Mediterranean Talitrus, the animals having moved inland to forage at the time of the previous sunset.

Although some organisms do possess rhythms which can be entrained by regular temperature fluctuations (Pittendrigh, 1954; Roberts, 1959; Williams & Naylor, 1969), it is not surprising that experimental temperature cycles failed to change significantly the timing of the locomotor rhythm of Talitrus. This is because the animals lie buried in sand during the day, where the effect of daily temperature fluctuations is likely to be small and often irregular. It follows that adaptation to temperature cycles alone could hardly have arisen. Pardi & Grassi (1955) report that the re-setting of the navigational timing mechanism of Talitrus by artificial light/dark cycles proceeds in constant temperature just as readily as when there is an imposed cycle of high temperature during the light period and low temperature during the dark period.

Near or complete independence of rhythms from the effects of different ambient temperatures has been reported for many organisms (e.g. Brown & Webb, 1948; Pittendrigh, 1954; Bünning, 1958; Naylor, 1963). For Talitrus, it seems that an increase in temperature causes a temporary advance of the timing mechanism controlling locomotor activity. An increase in temperature some time before the expected time of activity results in a delay in the timing of the locomotor rhythm and a lowering of the temperature advances of the timing of the rhythm in both Carcinus (Naylor, 1963) and in Blattella germanica (L.) (Dreisig & Nielsen, 1971) but not in Talitrus.

Compensation for the effects of increase in temperature in Talitrus occurs spontaneously after a few days (Figs. 6, 7), which suggests a dual control mechanism, partly temperature-sensitive and partly temperature-independent. A similar coupled dual mechanism has been advanced by Pittendrigh & Bruce (1959) for the control of the eclosion rhythm in Drosophila. Two levels of temperature sensitivity have also been suggested for Carcinus by Naylor (1963) in view of the different effects of cooling within the normal temperature range, which increases locomotor activity, and chilling to temperatures below that normally experienced by the animal, which suppresses activity and inhibits the timing mechanism which controls locomotor rhythmicity.

In the field, Talitrus is protected to a large extent from fluctuations in air temperature by being several centimetres below the sand surface during daytime when fluctuations are more likely to occur. The small or gradual temperature changes that the animal might encounter can probably be accommodated by the endogenous timing mechanism which is itself, in any case re-synchronized every day by light changes.

No clear demonstration of mutual entrainment emerged from the laboratory experiments, but this does not necessarily invalidate the inference from the investigations with single animals and groups of animals that it does occur. The experiments employed phase differences of between 7 h and 12 h in order that any mutual effect would be clear. In view of the results, it is likely that mutual entrainment is only possible between animals which possess relatively small phase differences, such as are found between single animals after free-running for 24 h. This makes unequivocal demonstration of mutual entrainment all the more difficult.

The clear-cut and persistent nature of the locomotor timing mechanism of Talitrus points to a strong adaptive significance, although Ruppell (1967) does not come to this conclusion for the locomotor rhythm of Orchestia platensis.Aschoff (1964) considers that a physiological oscillatory mechanism confers an advantage in its own right, because the periodic increases in amplitude allow, at certain times, higher levels of energy expenditure and hence, perhaps, of efficiency (alternating with low-level quiescent periods) than could be achieved by a non-oscillating system for the same overall expenditure of energy. Be this as it may, in view of the fact that the vast majority of organisms live in a cyclically fluctuating environment of one sort or another, there is an obvious advantage in terms of energy conservation in having the physiological oscillatory system correctly synchronized with the environment. In the case of Talitrus the best time for feeding and mating contacts is clearly the night-, time, when humidity is high, rate of evaporation low and bird predators are absent.

Although the animals are sometimes disturbed by tidal action, especially during spring tides and storms, there has been no advantage in developing a timing mechanism of tidal frequency, for Talitrus is a supra-littoral species, opportunist in feeding habits and not necessarily dependent upon excursions into the inter-tidal zone for food or shelter. Furthermore, regular synchronization of a timing component of tidal frequency would clearly be difficult to achieve. Wildish (1970) found that Orchestia gammarella, which is found beneath wrack and stones around high-water mark, possesses an endogenous locomotor rhythm of circadian frequency only, but O. mediterrinea, which lives somewhat lower down the shore, has endogenous rhythms of both circadian and tidal frequency.

The endogenous timing mechanism is of further importance to a burrowing organism such as Talitrus for it provides a means whereby the fluctuations of the environment can be predicted and so prepared for, without constant reference to them. Although Talitrus burrows may remain intact during daytime and allow information of light intensity to reach the animal, more often than not the burrows collapse near the entrance, particularly when the surface sand dries out. Furthermore, the manner of burrowing, namely scooping the sand with the gnathopods and first two pairs of pereieopods, passing it between the last three pairs of pereieopods and then flicking it backwards with the pleon, is such that the entrance is often blocked before the animal has finished its burrowing activity. On account of the endogenous ‘clock’, excursions to the surface need only be made as the time for locomotor activity approaches, and so the animal does not require information about light intensity during the day, and also, should the burrow remain open, the animal is not ‘misled’ by spurious light fluctuations.

Another adaptive advantage of the endogenous timing mechanism of Talitrus is seen in the much-studied ability to navigate by reference to the sun. It is not possible at present to state whether this mechanism is the same as that which controls the timing of the locomotory rhythm. However, they do have certain features in common. Both are re-synchronized by light/dark changes and both appear to ‘run fast’ initially when subjected to a temperature rise of more than a few degrees. Rates of advance which are similar under free-running conditions have been demonstrated in both the activity and navigational timing mechanisms of starlings by Hoffmann (1960). The lack of ‘valuable results’ from navigational experiments by Pardi & Grassi (1955) after Talitrus was chilled may be due to the same sort of differential effect of chilling upon the navigational timing mechanism as occurs with the locomotor activity rhythm (Bregazzi, 1972).

1. Talitrus saltator possesses a clear-cut locomotor activity rhythm, largely under the control of a persistent endogenous timing mechanism of circadian frequency. In dim red illumination the period of the rhythm is somewhat greater than 24 h; in continuous white light it is less than 24 h, provided that the animals are unable to avoid the light by burrowing. The rhythm is synchronized by light/dark fluctuations, and the animal is active during the dark period.

2. The timing mechanism is temporarily advanced by increases in temperature, but otherwise it possesses a large measure of temperature independence within the normal environmental range.

3. There is some evidence that, in groups, the animals influence one another and so reduce the rate of drift of the activity rhythm in constant conditions.

4. The adaptive significance of the endogenous timing mechanism of Talitrus is discussed.

We are grateful to Mr R. G. James for helpful discussion, Professor E. W. Knight-Jones for the provision of laboratory facilities and the Science Research Council for financial support.