After some years of relative inactivity the pace of circadian rhythm research seems to be gathering again. The renewed energy, however, is focussing on molecular solutions and identification of the genes involved with the clock mechanism. Although I applaud such efforts, what is being written seems too often to suggest some miscomprehension of what the circadian system, or systems, actually involve. Certainly there is a lack of reference to actual rhythmic activity patterns observed in the field, the laboratory or both. What follows therefore is an attempt to redress the situation. The reader will find some repetition from the earlier parts of the website but this is done to highlight my arguments and to enable this section to "stand-alone".

Orthodox circadian theories

The literature on circadian rhythms contains much, rather dogmatic, theory which has been derived either from animals showing unimodal diurnal or nocturnal activity (often with activity throughout the light or dark), or else from observations of once-in-a-lifetime events (such as eclosion) which require conclusions to be drawn from populations. For instance, a prime author, C.S. Pittendrigh, subscribed to the view that a 'day-active program appropriately tracked dawn while that of night-active species tracked sunset' (Pittendrigh, 1993 et ante). The timing of activity under field conditions has been attributed to 'gating' (Beck, 1980), and bimodal field activity of day-active insects has been described as 'restricted to the hours near dawn and dusk when the saturation deficit is lower and the hazard of water loss thereby lessened' (Pittendrigh, 1993). Most explanations of entrainment and the mechanisms underlying the observed patterns of daily activity stem from experiments that appear to show a single change in light intensity, most commonly light-off, as being responsible for phase-setting. For insects perhaps the most widely cited theoretical approach is that of Peterson (1980a, b, c), which is based on observations of a mosquito, Cx. p. quinquefasciatus. It is common also to find the terms 'noise' and 'transients' used to justify minimising, or ignoring, the causation of activity at unexpected times (for example, Pittendrigh in a seminar discussion response to Lankinken and Lumme, 1984).

In contrast, the stimulus for the research described in this review was the discovery that both light-on and light-off have phase-setting effects on the circadian rhythm of flight activity of Ae. aegypti (Taylor and Jones, 1969). Just after the Ae. aegypti findings had been submitted for publication, White (1968) published his observation that the morning peak of hatching of the eggs of a psyllid, Cardiaspina densitexta, was controlled by two 'clocks'. One clock apparently was set by light-off 12-14h earlier and the other by light-on to entrain the hatching some 45min to 2h later. These two papers were the first such reports, at least for insects, and both stemmed from observations made using a range of artificial LD regimes. Within the natural range, southern Australia, of C. densitexta, the length of darkness varies from just under 10 h to a little over 14h in midwinter. White concluded that the dual timing ensured that hatching took place only when other conditions of temperature and humidity were favourable for the survival of larvae. The suggestion for Ae. aegypti was that the co-entrainment fitted the range of summer daylengths, from 12-16h daylight, found within the known geographical range (0-40° latitude) of an otherwise pan-global species (Taylor and Jones, 1969).

The significance of 'moving' peaks

Saunders (1982) reviewed many aspects of 'insect clocks' and wrote (pp 116-117) -

'Satisfactory evidence for the existence for separate dawn and dusk oscillators is provided by either (1) stably different phase relationships between the activity peaks, established during entrainment to cycles containing different durations of light, and retained during subsequent DD free-run, or (2) the “splitting of the activity into two or more components which, at least, for a while free-run with different periods. In the most convincing cases the two components span the entire 360° of mutual phase relationship, and may even cross and recross'.

This website shows that such a study had been made (Taylor, 1969) although it is only now that ready availability of quick and relatively easy computer analysis has enabled the necessary depth of analysis.

Mosquito studies

Published studies of mosquito species in a range of LD regimes are few. Nayar and Sauermann (1971) exposed Aedes taeniorhynchus to a range of LD regimes and found it to be a nocturnal species, with a clear bimodal E and M pattern in LD 12:12, which persisted in DD after LD 12:12. In regimes with D > 12h the M peak settled around 12-15h after dusk. In L > 12h activity compressed into D, with light inhibiting activity, except in LD 23:1 when there was activity starting some 12h after light-on. They did not comment on the drifting of peaks away from the bimodal E and M. The East Asian species, Cx. p. pallens, also shows this group pattern in LD 16:8 and LD 12:12 (Chiba, 1964; Chiba et al. 1982). Chiba and Tomioka (1992) examined their findings on Cx. p. molestus in the light of Aschoff's (1981) consideration of how insects respond to a change in LD ratio. That consideration gave three ways where phase angle difference stays in parallel to light-off, light-on, or the middle of light-time. It seems, however, to depend on only one zeitgeber operating in each of the first two ways, and to require that both steps in light intensity have equal weight in the third. Even so, it was an advance from earlier thinking (Tyschen, 1978, derived from Aschoff, 1965) where it was thought that, under normal photoperiods, rhythms with peaks in the light period fit straight lines of slope intermediate between the slope of the dark/light transition and the vertical; and, rhythms with peaks in the dark period fit straight lines of slope intermediate between the vertical and the light/dark transition.

The studies presented on this website, thus, provide a more extensive range of experimental results than any other findings on mosquitoes. One of the features of mosquito rhythms that has permitted the observation of what appear to be multiple zeitgebers is that none of the species studied shows purely unimodal activity. Most have readily discernible and clear peaks of activity, which surely are the product of the internal rhythms, or rhythms. Flight activity of individuals has proved to be easy to monitor and the pattern can be observed for several successive 24h-cycles. Here, in the scrutiny of rhythm patterns, it is recognised that adverse illumination (i.e. D for L-active and L for D-active species) tends to suppress activity, and, so, upsurges in activity may be observed only as low percentages of the overall amount in any 24h period. Accordingly, all upsurges in activity are assumed to have significance. From that, although they are not included on the already complex photoperiodograms, it is easy to spot trend-lines linking the obvious peaks in favourable light conditions and the smaller upsurges in adverse light conditions.

With regard to 'noise', all the photoperiodograms are derived from recordings of the activity patterns of individual mosquitoes and those patterns were examined. In total, across the 14 species, plus 4 strains of Ae. aegypti, there are recordings from 86 LD regimes and over 1400 mosquito-days. The diverse origins of the studied insects - including long-standing laboratory colonies, quite new laboratory colonies, larvae reared from nature, and wild-trapped adults - strengthens the core of evidence, rather than weakening it. Further strength comes from the variety of ways in which the insects were exposed to the recording LD regimes - with some experiments using insects reared in the final LD regimes, and others using insects recorded after shifting light-on to achieve a new LD regime. A final useful feature of mosquitoes is that the activity of many species has been observed in the field, and, so, laboratory results can be verified as representing valid patterns rather than simply spurious end-products of the experimental conditions.

It is important to emphasise that for all the mosquito species studied there is evidence of more than one zeitgeber operating to affect the activity pattern. At the least, in assymetric LD regimes, most show shifting of peaks which is consistent with separate entrainment by light-on and light-off, supporting the long-standing report by Taylor and Jones (1969). Particularly noticeable is the splitting of E' in the day-active and early crepuscular species, and of M and E in late-crepuscular species. This evidence is made more compelling by the fact that the splitting can be observed as activity in the light condition (i.e. absence or presence of light) which in median LD regimes appears to be inhibitory.

Insects other than mosquitoes

Although the phenomenon of peaks occurring away from the 'normal' point, or points (i.e. the pattern in regimes of around LD 12:12), has been found for a number of other insect taxa, the significance of such observations has tended to pass without comment. Tyschen (1978) examined the circadian rhythm of mating in the Queensland fruit fly, Dacus tryoni, and reviewed a number of other reports of circadian rhythms underlying behaviour. His summary followed earlier thinking, in accepting that entrainment of rhythms in the light period follows the dark-light transition and rhythms in the dark period follow the light-dark transition. Almost all the examples he examined were from studies with LD regimes within the range L 6-18h, and none was of the evening (E'-E) pattern. The D. tryoni results showed very tight E'-E mating responsiveness in median LD regimes, but the peak became much more diffuse as the regimes become more extreme. In LD 22:2 the peak was markedly forward, and in LD 4:20 the activity continued well into the dark.

Brady and Crump (1978) compared field and laboratory observations of circadian activity rhythms in a tsetse fly, Glossina morsitans. This species shows a V-shaped pattern with all activity in L and almost all concentrated in M and E', especially the first hour after light-on. When flies in an LD 12:12 regime experienced a change to LD 18:6 the pattern retained its V-shape. When light-on was advanced 6h the V shifted forward over two 24h-cycles. When light-off was delayed 6h the V remained unaltered, with activity dropping after the E' in the 12th hour of light. They concluded that sunrise was the primary zeitgeber, and that this was not surprising for a species originating from Zimbabwe (17°S), where natural LD variations were only from LD 11:13 to LD 13:11.

Smith (1985) reported that the peak of eclosion of the blowfly, Lucilia cuprina, occurred close to light-on in all photoperiods, but in long L the mean eclosion was after dawn, whereas in short L it was before dawn. He did not comment on the timing mechanism.

Moore and Rankin (1993) studied the honeybee, Apis mellifera ligustica, (from Austin, Texas, maximum LD 14:10). They used LD regimes from LD 4:20 to LD 22:2, and found a totally day-active pattern with a broad, almost flat, peak, somewhat skewed towards afternoon activity (see Fig. 53). They denoted onset of activity and noted this appeared to follow light-off by some 10h. Their interpretation was that the main peak was entrained by the previous light-off but their view did not embrace the evidence from long L regimes. In particular, they offered no explanation as to the strong evidence from LD 22:2, which clearly showed an early peak which appeared to be entrained to follow light-on by some 10-14h.

A specially interesting report is that of Tomioka, Uwozumi and Matsumoto (1997) who investigated the effect of various LD cycles given during development on 'wild-type Canton-S' and the 'per^S and per^L1 mutants' of Drosophila melanogaster. The analysis and discussion centred on the activity patterns recorded in DD following the rearing regimes but the actograms do show the patterns in several days prior to the change to DD. Although they make no comment on it, a primary difference is that Canton-S and per^S are light-active whereas per^L1 is night-active. Their analysis was based solely on actograms (essentially assemblages of the actual recording charts) and they did not assess the variation in activity level on, say, an hourly basis. From the actograms, however, it is clear that in LD12:12 and LD 16:8 (i.e. the natural summer daylength) Canton-S and per^S show a bimodal M-E'. Although the authors noted this bimodalism they ignored it, chosing to base their analysis on straight lines 'eye-fitted' to 'activity onset' and 'activity offset' of what was, thus, taken as a unimodal activity pattern. From simple visualisation, in the extreme LD regimes the peak activity clearly resembles that shown, for instance, by Ae. aegypti. In LD 20:4, the E' in per^S follows light-on by some 10-12h, with a small burst of activity in the 'normal' E' position; whereas E' in Canton-S follows light-on by 14-18h. The night-active per^L1 shows a primarily N peak of activity, with a short-duration E in LD 8;16 and LD12:12. The activity in the more extreme LD regimes appears to show poor entrainment with the N appearing to drift later in the dark periods with each successive 24h LD period.

Conclusions

Skeleton photoperiods and phase response curves

Many published studies of activity have been limited to a single regime (such as LD 12:12 for species of tropical origin, or LD 16:8 for species of temperate origin). The mosquito photoperiodograms show clearly how such studies will conceal more than they reveal, and that use of a range of LD regimes (within the frame work of the natural earthly 24h-cycle) is essential to investigations of entrainment. Similarly, orthodox entrainment theory (most notably stemming from the work of Pittendrigh, 1981 and 1993) has its origins in the extensive use of 'skeleton' photoperiods and relies heavily on the 'phase response curve'. The concept involves the demarcation of a single 'phase reference point' and then plotting the timing of that point in relation to what is defined as being a single zeitgeber. The evidence, from mosquitoes and other insects, that many activity patterns are bimodal, with the timing of activity probably determined by both light-on and light-off, and, further, that peaks can be split in more extreme LD regimes, surely is likely to impair or even invalidate the concept. At the simplest level, a light pulse given during an otherwise continuous period of darkness will have both an upward and a downward change in light intensity. White (personal communication, 1969) also felt that skeleton photoperiods could well be 'independent dark periods with light periods between them'. The widely cited work of Peterson (1980a, b, c), with its development of the 'limit cycle' model, also relied on a single phase reference point. My results (click here, and they are results which match most field reports for the species, suggest that Peterson studied either an unusual strain of Cx. p. quinquefasciatus or, perhaps, the pre-mating stage of the adult females. Moreover, the species exhibits bimodalism.

Clock genes and 'mutants'

Current literature reveals a wide and ever-expanding core of biochemical and molecular studies aimed at elucidating the true nature of the 'clock'. It is worrying, however, that some of the reports appear to show a loss of sight of the whole organism. For instance, the 'clock' is always referred to in the singular, and often there seems no obvious link between the published activity patterns and the mechanism illustrated (although that mechanism may well form only part of the overall system). Although the insect species is not always clearly stated (for example Goldbeter, 1995, and Zeng et al., 1996, wrote simply of "the Drosophila period protein") a fruit fly, Drosophila melanogaster, has become the prime insect for such studies and much emphasis is placed on what are known as the 'wild-type Canton-S' and the 'per^S and per^L1 mutants' (Tomioka, Uwozumi and Matsumoto, 1997).

As noted earlier, a primary difference between these D. melanogaster 'mutants' is that Canton-S and per^S are light-active whereas per^L1 is night-active. The evidence from the literature, however, is that the difference is ignored. Given that flight is a necessary precursor to most other behaviour, including mating, such differences in activity would predispose the forms to reproductive isolation and speciation. To me, as an entomologist with specialist interests in behaviour and population studies, that alone suggests that the 'mutants' of D. melanogaster may be separate members of a species complex. On this web site, the photoperiodograms of the Cx. pipiens complex show markedly different patterns of flight activity - i.e. the M-E' pattern in day-active Cx. p. molestus, the E-M pattern in late-crepuscular Cx. p. pipiens, the N-M' pattern in night-active Cx. quinquefasciatus and the apparently unimodal N pattern of the closely-related night-active Cx. torrentium. Known also is the bimodal E -M pattern of Cx. p. pallens (Kasai & Chiba, 1987). Another example is the Anopheles punctulatus group which now is known to be a species complex with at least seven sibling members, including An. farauti. These are regarded as inseparable on reliable morphological grounds but are distinguishable by allozyme electrophoresis (Foley et al. 1994).

Indeed, I find the whole story of the D. melanogaster 'mutants' rather worrying. Their origin can be found in the report by Konopka and Benzer (1971). This reveals that D. melanogaster was chosen because it 'offers certain advantages', with readily available genetic methodology and much already 'known about the rhythm of eclosion'. Moreover, there was a precept that - 'Since the expression of a rhythm requires an integrated system, mutation of the genes responsible for development and functions of the system could lead to abnormal rhythms'. The starting point was the 'C-S' (Canton-Special) wild strain. After 'mutagenesis' by ethyl methyl sulfonate, 'treated males' from the strain were mated to virgin females. The progeny were reared in LD 12:12 and stocks of 'rhythm mutants' were separated out as those which emerged 'abnormally', by comparison with 'normal' female stocks where at least twice as many emerged during the light as during the dark. Graphs of emergence and 'modulo-plotted' actograms of locomotor activity were published. Even allowing for the slight difficulty in visualising activity patterns form the modulo plots, the day-active pattern of the wild and short-period flies and the dark-active pattern of long-period flies is obvious. This, however, was not remarked on by Konopka and Benzer. The separation of the 'mutants' on the grounds of abnormal emergence also is worrying, specially as the 'normal' females showed a mixed pattern of emergence, in itself an unusual phenomenon given the precision timing of most behavioural events.

Konopka and Benzer's key hypothesis, accepted by the many who have successively used the 'mutants', was that the 'alteration of a single gene' had led to 'missense mutations producing alterations in quality or quantity of a gene product involved in the clock mechanism'. Without doubting the genetic stability of the various D. melanogaster forms and other more recently identified 'mutants', the mosquito evidence clearly suggests that the concept of 'missense mutations' may be a misunderstanding of wholly sensible and viable interspecific variations. I have to suggest, therefore, that clarification of the purity of the original colony of the now very widely used 'wild type Canton-S' is very desirable, if not essential.

END