Although D. pseudoobscura was found to be day-neutral and without a diapause in its life cycle, Pittendrigh soon concluded that the time measurement inherent in photoperiodic induction could be explained by comparison with an insect’s circadian responses (see, for example, Pittendrigh and Minis 1964; Pittendrigh 1966). The following sections describe subsequent tests of that notion in S. argyrostoma.

‘Complete’ PPRCs for insect diapause induction have been described on numerous occasions (Lees 1955; Saunders 2001; Denlinger 2022); these cover the full range of photophases, natural and unnatural, together with continuous darkness (DD) and continuous light (LL). Responses to light–dark cycles outside the natural range for the latitude of origin are included, since any clock model proposed must also explain such data. Ultra-short photoperiods, for example, may be defined as those not occurring naturally at the insect’s latitude or during the winter when the insect is already in diapause and therefore unresponsive. The proportion entering diapause in these ultra-short photoperiods is variable and generally reduced in comparison with that in ‘strong’ natural short days.

The reason for this reduction in diapause in ultra-short photoperiods in the flesh fly was investigated by establishing PRCs for pulse durations ranging from 1 to 20 h (Fig. 2) (Saunders 1978) on first instar larvae, a stage when photoperiodic sensitivity was most marked. Pulse lengths of 1 to 3 h were found to give rise to low-amplitude type 1 responses, whereas those of longer pulses produced responses approaching or reaching type 0. Light pulses of 12 h or more reset the oscillation to near constant phase equivalent to the onset of the subjective night (CT 12 h) (Saunders 1978) as in D. pseudoobscura and C. vicina.

Phase response curves (PRCs) for the adult emergence (pupal eclosion) rhythm of the flesh-fly Sarcophaga argyrostoma exposed as larvae to light pulses of 1–20 h. Short pulses (1–3 h) give rise to Type 1 PRCs in which the resulting curve is roughly parallel to the beginning of the pulse. Longer and ‘stronger’ pulses give rise to Type 0 PRCs with resulting curves roughly parallel to the end of the pulse. From Saunders (1981b)

In the eclosion rhythm of D. pseudoobscura (Pittendrigh and Bruce 1957)—and probably also in S. argyrostoma—eclosion behaviour is regulated by a two-tier system comprising a light-sensitive pacemaker that, in turn, controls a more downstream and temperature-sensitive slave. In D. pseudoobscura, pacemaker phase shifts are achieved almost immediately in response to a light pulse (Chandrasekaran 1967) but observable transients occur, both to the pacemaker and to the slave oscillator ‘catching up’ with the former. In doing so, type 1 responses induce only small phase shifts and pass through a larger number of transient cycles before achieving steady state, whereas stronger pulses giving rise to type 0 responses induce larger phase shifts and more rapid entrainment. It is probable that diapause induction is regulated in the same way.

An experiment was therefore conducted (Saunders 1982, 2022) in which cultures of larvae were exposed to trains of light pulses (of 1–21 h in duration) with the light pulses starting already close to entrainment at CT 12 (dubbed in phase), or a full half-cycle away at CT 24/0 (dubbed out of phase) (Fig. 3). The number of transient cycles before steady state was estimated using a computer program based on the light pulse PRCs. Results showed that cultures of larvae initially out of phase and passing through a greater number of transients than cultures starting in-phase, resulted in a lower incidence of diapause. On the other hand, cultures starting in-phase produced levels of diapause equivalent to ‘strong’ short days of 8 h or more, suggesting that steady-state entrainment of the presumed photoperiodic oscillator facilitates full diapause induction.

Sarcophaga argyrostoma. Photoperiodic response curves (PPRCs) for light pulses of 1–21 h in duration with the first pulse in the train starting either at CT 12 (close to final steady-state entrainment) or a full half-cycle away at CT 24/0 (far from entrainment). With shorter or ‘weaker’ light pulses (less than 8 h), pulses starting at CT 12 h induce levels of pupal diapause comparable to those engendered by longer, stronger pulses, whereas those starting at CT 24/0 h produce much lower diapause incidence. This suggests that diapause incidence depends on the rate of entrainment and, therefore, the number of transient cycles experienced before final steady state. Redrawn from Saunders (1982)

In a second experiment, light–dark cycles each containing 2, 4, 8, or 12 h of light, but with the irradiance of the photophase at either 240 μW cm−2 or much increased to about 16,000 μW cm−2 (Saunders 1982, 2022). Figure 4 shows that when the strength of the light pulse was increased, diapause incidence was also increased, particularly for the shorter photoperiods. This increase was not caused by an unavoidable rise in temperature at the increased irradiance, because higher ambient temperature in DD reduces diapause (Saunders 1971). Reciprocity between pulse duration and photophase intensity probably means that increased illuminance strengthened the ultra-short photoperiods from type 1 PRC towards that of type 0, thereby increasing diapause incidence by reducing the number of transients.

Sarcophaga argyrostoma. Incidence of pupal diapause produced by cultures of larvae exposed to ultra-short light pulses of 2, 4, and 8 h at light intensities of 240 µW cm−2 (closed circles) or to much higher irradiance of approximately 16,000 µW cm−2 (open circles). Increased irradiance induces a greater incidence of pupal diapause, probably because PRCs are raised from Type 1 to Type 0 and induce fewer transient cycles. Redrawn from Saunders (1982)

In S. argyrostoma, the phase shifting effects of 12 h light pulses were followed throughout larval development (Saunders 1979b) to delineate the photoperiodic sensitive period. Sensitivity began in the intra-uterine embryo then, in first instar larvae gave a type 0 PRC which subsequently declined through a series of type 1 responses until no response was recorded in third instar, post-feeding or ‘wandering’ larvae prior to their burrowing into the soil to pupate. Initially, this reduction in phase shifts during larval development was attributed to oscillator dampening. However, since the photoperiodic oscillator must continue through the pupal instar to time adult emergence, it seems more likely that the attenuation results from the photoreceptors becoming ‘detached’ from the oscillation. The effect, however, is the same: larval sensitivity to photoperiod steadily declines during larval development. In other species of flesh fly, this aspect shows variation. In North American S. crassipalpis (Denlinger 1971), for instance, sensitivity to photoperiod is largely restricted to the intra-uterine embryos, whereas in Japanese S. similis, it extends into the wandering larva itself (Goto and Numata 2009). Rapid attenuation of the photoperiodic response (likened to ‘oscillator dampening’) is also seen in some aphids—Megoura viciae, for example (Lees 1973)—and in some drosophilids at very high latitudes (Vaze and Helfrich-Fȍrster 2016; Kauranen et al. 2019; Tyukmaeva et al. 2020) in which the photoperiodic clock presents characteristics resembling an ‘hourglass’.

Pittendrigh and Minis (1964) showed that the eclosion rhythm of D. pseudoobscura could synchronise to a regime consisting of two short (15 min) pulses of light per cycle which they called a symmetrical ‘skeleton’ photoperiod (PPs) to distinguish it from a ‘complete’ photoperiod (PPc): this result attracted attention to the importance of the ‘on’ and ‘off’ transitions of the photophase as important signals in the entrainment phenomenon. Skeletons of short photophases up to about 11 h (PPs 11) were found to simulate almost exactly their complete counterparts (PPc 11), with the two 15 min pulses causing phase shifts in steady state that ‘balanced’ each other by the first causing phase delays (−Δφ) and the second, phase advances (+ Δφ) of the same magnitude but of opposite sign (Pittendrigh 1965). Attempts to entrain the eclosion rhythm to skeletons longer than about 14 h, however, were more complicated. A skeleton of PPs 14, for example, was open to two ‘interpretations’, one of about 14 h and the other of about 10 h. Between these conditions, the oscillator underwent a ‘phase jump’ and came to accept the shorter of the two steady states as ‘day’. Longer initial skeletons similarly jumped to the shorter alternative.

Similar experiments were conducted using S. argyrostoma with respect to the production of pupal diapause (Saunders 1975a) (Fig. 5). Symmetrical skeletons (PPs) of 4–10 h formed from two 1-h light pulses per cycle led to a high incidence of diapause similar to that of their complete (PPc) counterparts. Attempts to entrain to skeletons of 16 h or more, however, led to a phase jump to the shorter alternative, also inducing a high diapause incidence. Skeletons of 11, 12, and 13 h, close to half the circadian period, however, produced less diapause (27–51%) than their corresponding complete photoperiods. The behaviour of diapause induction in S. argyrostoma, was clearly regulated by the entrainment of an oscillatory system similar to that in D. pseudoobscura.

Sarcophaga argyrostoma. a Diapause incidence in cultures of larvae exposed to complete (PPc) photoperiods (open circles) or to symmetrical ‘skeleton’ (PPs) photoperiods formed from two 1-h pulses of light (closed circles). Complete photoperiods produce a response curve with a typical critical daylength. ‘Skeletons’ of PPs4 to PPs10 also produce a high incidence of pupal diapause similar to that produced by their complete counterparts. Attempts to entrain to skeletons longer than PPs14, however, lead to a phase jump to the shorter interpretation. This also induces a high incidence of diapause. Skeletons with an interval close to half the circadian period (11–13 h), produce less diapause. b Experimental design of the ‘skeleton photoperiods’. Redrawn from Saunders (1975a)

In D. pseudoobscura, symmetrical skeletons initially close to half the circadian period were found to present an additional property: the final steady state depended on two initial variables, (1) the phase illuminated by the first pulse in the train, and (2) the value of the first interval between the two pulses (Pittendrigh 1966). This could result in two alternative steady states called the ‘bistability phenomenon’, predicted by computer simulation and considered to be important evidence that pupal eclosion was indeed an oscillatory phenomenon.

Similar results were obtained for diapause induction in S. argyrostoma (Saunders 1975a) in which larval cultures were exposed to two skeleton regimes formed from 1-h light pulses, either LD 1:9:1 (PPs 11) or LD 1:13:1 (PPs15) each with the first light pulse in the train commencing at all circadian phases. These two skeletons were chosen, because the former, if taken as ‘day’, should produce a short-day, high incidence of diapause equivalent to that induced by an 11 h complete photophase, whereas the latter should act as a long day, producing a low incidence equivalent to that in a 15 h photophase. The results of this experiment (Saunders 1975a) produced two curves of diapause incidence, mirror images of high and low diapause, showing that the bistability phenomenon applied to diapause induction in a fashion equivalent to those of circadian entrainment in D. pseudoobscura described above. Similar ‘positive’ bistability results have since been observed in C. vicina (Vaz Nunes et al. 1990) and in the cabbage moth Mamestra brassicae (Kimura and Masaki 1993) and constitute strong evidence for the circadian basis of photoperiodic time measurement and diapause regulation.

Interrupting the scotophase with additional short light pulses (‘night interruption’ experiments) were originally performed to investigate possible light-sensitivity phases during the dark period. Early results, often for the additional light pulse placed centrally in the night, seemed to indicate that such treatments were sometimes ineffective (Lees 1955). However, Adkisson (1964, 1966), working with the pink bollworm moth Pectinophora gossypiella and using systematic scanning of the scotophase by 1 or 2 h light pulses, showed that long-day effects (diapause avoidance) were observed at two phases, one early (point A) in the night and the other (point B) later (Fig. 6). Pittendrigh (1966) observed the similarity between these results and his own observations in D. pseudoobscura—which he called ‘asymmetrical skeletons’. Using PRCs, he showed that, in conjunction with the main photophase, the early supplementary pulse acted as a ‘new dusk’ and the later pulse as a ‘new dawn’ and constituted strong evidence for the circadian basis of insect photoperiodism. Such bimodal responses to night interruption experiments have since proved almost universal in the insects, including the flesh fly S. argyrostoma (Saunders 2002, 2020).

Sarcophaga argyrostoma. Diapause responses of larvae exposed to asymmetrical ‘skeletons’ (or night interruption experiments) using basic photoperiods of LD 7:17 to LD 14:10. In the three shorter photoperiods (panels a–c), scanning 1-h light pulses produced two phases of reduced diapause incidence (A and B), whereas with LD 14:10 diapause was reduced to near zero with all scanning pulses (panel d). In panels a–c, point B occurred about 9 h after the end of the main light component. This result suggests that point B is the diapause-regulating photoinducible phase φi (see text for details). Solid and dotted lines show data from replicate experiments. Redrawn from Saunders (1975a)

In S. argyrostoma, two further experiments were conducted using night interruption techniques but in non-24 h light–dark cycles (Saunders 1979a) inspired by earlier experiments with the green vetch aphid by Lees (1970). In the first test, flesh-fly larvae were exposed to a series of asymmetrical skeletons in which the supplementary pulse was placed early in the night (at point A) and the terminal hours of darkness systematically increased from 7 to 13 h. In the second test, the supplementary light pulse was placed later in the night at point B (9 h after light-off) and was then followed by 12 h of darkness, a period greater than the critical nightlength for diapause induction. Results of the first test showed that the diapause-averting effects of light falling on point A could be reversed by a terminal dark period longer than the critical night length, whereas results of the second test showed that the ‘long day’ diapause-averting effect of illuminating point B could not be reversed by the terminal diapause inducing long night. This latter result suggested that point B marked the position of the photoperiodic photoinducible phase. This evidence proved to be of crucial importance in understanding how the circadian system regulates diapause induction.