This review examines some of the models to account for time measurement in insect photoperiodism. It considers the supporting evidence for these models and the attempts to discriminate among them. Although hourglass timers may exist, it is suggested that most photoperiodic mechanisms, including many hourglass-like timers, are circadian-based, making Bünning's original hypothesis, that the circadian system somehow provides the essential “clockwork” for photoperiodic timing, the most persuasive unifying principle. The apparent diversity among modern species in their modes of time measurement is probably the result of differences between the underlying circadian systems that were adopted for seasonal night length measurement as the insects, or groups of insects, moved northwards into areas with a pronounced winter season. Photoperiodic time measurement, therefore, exhibits both unity (in their common circadian basis) and diversity in detail. Attention to this diversity may provide invaluable insights into the problem of photoperiodic time measurement at comparative, and molecular, levels.
As their distributions extended northwards, insects encountered increasingly severe and longer winters, coupled to shorter favorable summers or “growing seasons”. The evolutionary response to these adverse periods of winter cold was dormancy, specifically diapause. In multivoltine species with a facultative winter diapause, further growth and reproduction is suspended, most frequently as day length falls below (or night length increases above) a well-defined critical value (Saunders 2002). Such photoperiodically regulated diapause is now known in over 500 insect species (Nishizuka et al. 1998), occurring in species-specific diapause stages such as the egg (embryo) nymph, larva, pupa or adult.
Induction of facultative diapause comprises a linked series of events including: (i) an input pathway involving photoreception; (ii) the “measurement” of day length (or more usually, night length); (iii) the simultaneous accumulation of inductive night lengths by a “counter” mechanism; and (iv) endocrine effectors to regulate the diapause or nondiapause developmental pathways (Saunders 2002). The role of the neuroendocrine system in the inception, maintenance and termination of diapause is now well documented (Denlinger 1985; Saunders 2000), but the earlier or “upstream” events remain obscure. In particular, the identity of the photoreceptors is not fully resolved, the nature of the night length timer and the counter mechanism still conjectural and, in addition to light, the input pathway also responds to daily temperature cycles (thermoperiods) and to low temperature itself.
Until recently, the central photoperiodic mechanism (the “clock–counter” system) responsible for night length measurement and the accumulation of inductive cycles was investigated merely by noting the diapause responses of groups of insects exposed to a range of experimental inputs of different light and temperature cycles. Formal experiments of this type were frequently designed to test for the possible involvement of the circadian system in time measurement. More than 50 years of such research has generated a vast amount of data describing the basic properties of the insect photoperiodic system, together with a range of theoretical models to explain the phenomenon of time measurement (Vaz Nunes & Saunders 1999; Saunders 2002). To date no consensus of opinion has been forthcoming and a wide variety of possible mechanisms has become apparent. This review considers these data and models in relation to the apparent diversity seen in the insect photoperiodic response; in doing so it revisits an older publication (Saunders 1978a) addressing the same issue.
BRIEF HISTORY OF PHOTOPERIODIC CLOCK MODELS
The earliest attempts to model time measurement in photoperiodism envisaged a biochemical mechanism set in motion at dawn (or more frequently, at dusk) and measuring the duration of the light (or the dark) component of the daily cycle. The night length timer was likened to an “hourglass” in which dawn, if it came early, signified a short (summer) night length resulting in uninterrupted development, but if it was delayed, signified a long (autumnal) night length leading to developmental arrest (diapause) or to an autumnal morph. This idea was expressed in its most detailed form by A. D. (Tony) Lees of Cambridge University and Imperial College, London. In his monumental work on seasonal morph determination in the green vetch aphid Megoura viciae Buckton, Lees regarded the night length hourglass timer as a series of linked biochemical reactions occurring during the scotophase (Lees 1973, 1986). This model was essentially a linear or non-circadian type of night length timer that did not reset itself automatically in extended periods of darkness: the hourglass timer needed to be “turned over” by an intervening light phase before it could measure night length again. This concept of an hourglass-like photoperiodic mechanism lives on today, particularly for some Lepidoptera (see below).
Circadian photoperiodic clock: Bünning's hypothesis
An alternative type of model was proposed by the German botanist Erwin Bünning (1936) of Tübingen University who suggested that night length measurement in plant photoperiodism was one of the functions of the circadian system now known to provide temporal organization in almost all eukaryotes (and some prokaryotes). Bünning suggested that a circadian rhythm was entrained by the light–dark cycle in such a way that a particular light-sensitive phase occurred during the night. Under the long nights of autumn this phase remained in the dark, but under the short nights of summer it was illuminated, giving rise to an appropriate summer response. In this model, therefore, light had a dual role: entrainment and photoinduction.
Early tests of Bünning's hypothesis, initially with plants and later with animals, used light–dark cycles longer than 24 h duration, designed to examine possible “free-running” circadian rhythmicity in the photoperiodic response. These experiments were of two main types. In the first, often referred to as the Bünsow protocol (Bünsow 1953), organisms were exposed to 48- or 72-h cycles containing a main photophase of, say, 8 to 12 h with the extended scotophase systematically interrupted by a short 1 or 2 h “scanning” pulse. In the second, usually called the Nanda–Hamner (NH) protocol (Nanda & Hamner 1958), organisms were exposed to exotic cycles each containing a “short” photophase (e.g. 8 to 12 h) coupled, in different experimental subsets, to an increasing number of dark hours to give overall cycle lengths ranging from 18 to 72 h or more in length. Both types of experiment may show peaks and troughs of the photoperiodic phenomenon under study, for example, flowering maximum, at intervals of about 24 h in the extended night, clearly indicating a circadian involvement in the response. Insect examples of such experiments will be discussed below.
External coincidence model
Night-interruption experiments, in which the long night of an inductive light–dark cycle (e.g. 12 h light : 12 h dark; LD 12:12) is systematically interrupted by a 1-h scanning pulse, frequently reveal two positions in the night at which the supplementary light pulse produces a short night or nondiapause response (see Saunders 2002). These are referred to as point A early in the night and point B late in the night. In a typical case such as the flesh fly Sarcophaga argyrostoma (Robineau-Desvoidy) (Saunders 1975), where the critical night length is about 9.5 h, point A, early in the night, occurs about 9.5 h before the dawn transition, whereas point B occurs about 9.5 h after dusk. In the 1960s, Colin Pittendrigh of Princeton University and then Stanford University interpreted such results in terms of the entrainment of the circadian system to complex two-component light–dark cycles, and suggested a version of Bünning's general hypothesis that was more appropriate for the insects. This model he called external coincidence (Pittendrigh & Minis 1964; Pittendrigh 1966, 1972).
This model incorporated a more modern appreciation of the phenomenon of circadian entrainment based on the behavior of the adult eclosion rhythm in Drosophila pseudoobscura (Frolova & Astaurov) (Pittendrigh & Minis 1964; Pittendrigh 1966). The external coincidence model retained Bünning's proposition that light had a dual role, entrainment and photoinduction. It differed from Bünning's original in that it recognized that the insect circadian system was reset to a near constant phase (about circadian time, or CT 12) at the end of a long light phase (Pittendrigh 1966). This property meant that any further lengthening of the light component of the cycle involved the “tracking back” of the dawn transition of the cycle to illuminate phases in the late subjective night. In insects, therefore, the photosensitive (or “photoinducible”) phase was thought to lie in the latter part of the night (at point B) rather than early in the night as envisaged by Bünning for plants. Experimental evidence supporting these aspects of the model will be presented in subsequent sections.
In its original form (Pittendrigh's 1966) external coincidence effectively modeled the abrupt switch at the critical day or night length, but failed to explain the frequently observed drop in diapause incidence under ultra-short day lengths and continuous darkness, and the occasional rise in diapause sometimes seen under ultra-long days (Pittendrigh 1972).
To address these and other problems, Lewis and Saunders (1987) suggested a modified version of external coincidence based on a dampening circadian oscillator, as first proposed by Bünning (1969, 1973). This model is reviewed here only in outline; more detailed information is given in the original papers (see Lewis & Saunders 1987; Saunders & Lewis 1987a,b). In these papers, a computer-based feedback control systems model was constructed, creating an oscillation in which the synthesis and loss of a hypothetical oscillating chemical (c) occurred in an autoregulatory feedback loop in relation to a hypothetical threshold concentration. Important parameters of this model included: (i) the time delay (TD) between the synthesis and the loss of c, giving rise to simulated “circadian” period; (ii) the synthesis rate (SR) of c regulating the dampening coefficient of the oscillation; and (iii) a simulation of the effect of light. A high value of TD simulated a long-period oscillation and a low value of TD a short period. A high synthesis rate (SR) provided a self-sustained oscillation whereas a low value gave a rapidly dampening oscillator. Values of SR were selected to give moderately dampened oscillations in continuous darkness. Simulated light lowered the concentration of “c” in an intensity- and duration-related fashion, whereas removal of the light gave rise to a rapid increase in c to boost the oscillation to higher amplitude. Using these parameters the oscillation in continuous darkness or under very short photophases dampened below the arbitrary threshold concentration, whereas longer photophases maintained amplitude. Amplitude was also maintained at above threshold values when the oscillation became entrained to the light–dark cycle, but fell below threshold when entrainment failed.
The essential features of external coincidence were modeled by identifying the phase at which c rose above the arbitrary threshold concentration as the photoinducible phase. The model then accumulated such phases, either illuminated or not illuminated, over consecutive cycles through the temperature-dependent “sensitive period” to produce a hypothetical “induction sum” (INDSUM). A high accumulated value of INDSUM at the end of the sensitive period simulated a high final incidence of diapause, whereas a low value of INDSUM simulated a low incidence. The damped oscillator version of external coincidence effectively described the formal properties of the photoperiodic clock, not only in the flesh fly S. argyrostoma, for which it was originally written, but for a wide range of other species (Saunders & Lewis 1987a,b).
The concept of a dampening oscillator may appear to be at odds with generally accepted features of circadian rhythms, namely that they are both persistent and fully self sustained. However, diapause incidence in insects is a population response expressed as a percentage entering diapause in the group of insects exposed to a particular set of environmental conditions. Since individual insects in the population may vary (e.g. show different free-running periods) the apparent dampening may be attributable to an increasing asynchrony among the members of the population as they free-run in darkness. If this also occurred within the photoperiodic mechanism, the developing asynchrony among the population would collectively resemble a dampening oscillator. It is this possibility that makes the concept of a dampening oscillator so useful in modeling the photoperiodic mechanism.
Internal coincidence: dawn and dusk oscillators
In addition to external coincidence, Pittendrigh (1972) suggested two other possibilities, namely internal coincidence and the more general principle of circadian resonance (see below); these concepts arose from an increasing understanding during the 1960s that the circadian system in complex, multicellular organisms comprised a number of cellular oscillators.
Like Bünning's original hypothesis, external coincidence comprised a single circadian oscillation and two functions for light, entrainment and photoinduction. Pittendrigh's internal coincidence model, on the other hand, derived from an earlier idea developed by Russian entomologists (Tyshchenko 1966; Danilevskii et al. 1970), was thought to involve at least two circadian oscillators (or groups of oscillators), one phase set by dawn (morning) and the other by dusk (evening), with light having a single role: that of entrainment. As day length (or night length) changed with the seasons, the mutual (internal) phase relationship between the two oscillators, or groups of oscillators, would change, thus regulating possible up- or down-regulation of genes involved in the diapause or nondiapause developmental pathways. Recent work on Drosophila melanogaster Meigen by Grima et al. (2004) and Stoleru et al. (2004) has revealed the presence of two groups of neurons, the morning or M-cells and the evening or E-cells; these cells are responsible for anticipatory locomotory activity preceding lights-on and lights-off, respectively. Although the circadian output these cells regulate is behavioral rather than dormancy-related, their activity demonstrates a possible seasonal aspect to the circadian system in D. melanogaster that might support such a model.
Circadian resonance: the multioscillator circadian system
Pittendrigh (1972) suggested a third way in which circadian rhythmicity could be involved in photoperiodic time measurement. Like internal coincidence, the “resonance effect” recognized the multioscillator construction of the circadian system in complex (multicellular) organisms and quoted several early examples of physiological performance (e.g. longevity or growth rate) being apparently impaired when the circadian system was driven by light cycles outside the normal ranges of entrainment. With particular reference to such procedures as Nanda–Hamner (NH) experiments, he stressed that “. . . one cannot simply treat such protocols as the search (with light) for the recurrence of a photoinducible phase of a circadian rhythm: the light used in the experiments drives, or entrains, the circadian system and we cannot set aside the complication that the performance of the system will be a function of its proximity to resonance, no matter how the photoperiodic time measurement is made.”
Although it does not concern diapause induction per se, an example of this may be afforded by the rhythm of adult eclosion in the flesh fly S. argyrostoma which shows highly rhythmic eclosion patterns under light–dark cycles whose period (T h) resonates with that of the endogenous circadian period (τ) or multiples of it (i.e. where T = 24, 48 or 72 h), but highly arrhythmic patterns when driven by cycles where T is far from τ (i.e. T = 36 or 60 h) (Saunders 1978b). If a similar phenomenon occurs in photoperiodic time measurement, the mutual phase relationships of constituent oscillators and their proximity to resonance may play a major role in the success with which a physiological function such as night length measurement is discharged. One consequence of this could be that the circadian system is not necessarily involved in night length measurement itself, but that the performance of the system is a function of its proximity to resonance.
This possibility was developed by Vaz Nunes and Veerman (1982) in their analysis of diapause induction in the red spider mite Tetranychus urticae Koch. They proposed that night length measurement in this mite was a function of a noncircadian hourglass, but the positive result from NH experiments, with its distinct peaks and troughs of diapause incidence at circadian (about 20 h) intervals, was the result of “downstream” action of the circadian system on the photoperiodic counter, with high diapause occurring when T was close to τ (i.e. when the driving light cycle resonated with the endogenous circadian system) but low diapause incidence when T was far from τ. They called this novel proposition the “Hourglass clock – Oscillator counter” model. This model, however, has been critically re-evaluated (Saunders 2010a), suggesting that the Tetranychus data could be adequately explained by the damped oscillator version of external coincidence outlined above. A separate effect of circadian resonance on the counter was also deemed unlikely because the two processes, the clock (night length timer) and the counter, are almost certainly just aspects of the same event, occurring hand-in-hand during the sensitive period.
A form of the resonance principle may also operate in T 24 h light-dark cycles, thus being of potential importance in the discrimination between long and short nights. In the years leading up to his death in 1996, Colin Pittendrigh (pers. comm., year unknown) explored this possibility with a model incorporating multiple cellular oscillators with different circadian parameters, particularly their endogenous circadian periodicities and responses to light pulses. Although never formally published, such modeling suggested that constituent oscillators adopted different internal phase relationships as the photophase shortened or lengthened.
EXPERIMENTAL EVIDENCE FOR THE VARIOUS CLOCK MODELS IN INSECT PHOTOPERIODISM
Circadian vs hourglass-like responses
Nanda–Hamner (NH) experiments with insects have produced both positive and negative responses (Vaz Nunes & Saunders 1999). Positive responses are those that reveal alternating peaks and troughs of diapause and nondiapause development as the period (T) of the light–dark cycle is extended. In the flesh fly Sarcophaga argyrostoma, peaks of high diapause occur at intervals slightly greater than 24 h (Saunders 1973a), whereas in the red spider mite Tetranychus urticae high diapause peaks are about 20 h apart (Veerman & Vaz Nunes 1980). These peaks of high diapause incidence thus occur with circadian frequency, and are interpreted as evidence that the circadian system is involved in photoperiodic time measurement. Other evidence that circadian rhythmicity is involved in the photoperiodic phenomenon including responses to complex two-component (“skeleton”) light cycles and “bistability” responses are not described here, but are reviewed elsewhere (Saunders 2010b).
Negative responses to NH photocycles, on the other hand, lack such peaks and troughs and are usually interpreted as revealing an hourglass type of night length timer. In the green vetch aphid Megoura viciae maintained at 15°C, for example, incidence of the short-night or summer morph (virginopara producer) occurs until the night length exceeds the critical value of about 9.5 h whereupon an abrupt switch to the long-night or autumnal morph (ovipara producer) occurs (Lees 1965, 1973, 1986). Once the switch has taken place, ovipara production remains consistently high with no peaks and troughs as observed in Sarcophaga or Tetranychus.
Although NH experiments with S. argyrostoma at 20°C revealed peaks of high diapause at roughly 24-h intervals, reduction of the temperature to 16 or 14°C provided an hourglass-like response (Saunders 1973a, 1982), rather similar to that observed for M. viciae at 15°C (Lees 1973). This prompted Lees to examine the responses of Megoura to NH experiments at higher temperatures. Even at 18 and 20°C, however, no regular peaks and troughs were observed (Lees 1986), bolstering the conclusion that the night length timer in this species was indeed an hourglass. Nevertheless, NH experiments conducted with the related Aphis fabae Scopoli at 20°C produced a “positive” NH response (Hardie 1987), and even the quintessential hourglass of M. viciae showed evidence for the accumulation of successive long-night cycles in periods of extended darkness (Vaz Nunes & Hardie 1993), strongly suggesting a circadian component in Megoura's photoperiodic mechanism. Hourglass-like photoperiodic responses such as these have been attributed to night length timers incorporating heavily damped circadian oscillators (Saunders & Lewis 1987b; Saunders 2010a,b).
Both positive and negative responses to NH experiments have been recorded in the Lepidoptera (for reviews, see Vaz Nunes & Saunders 1999; Saunders 2002). For example, in the large white butterfly Pieris brassicae (L.) maintained at 20°C in NH cycles containing 8 or 10 h photophases, three peaks of high diapause incidence were recorded at cycle lengths of about 24, 44 and 66 h, clearly indicating a circadian involvement in the photoperiodic response (Claret et al. 1981). In a Dutch strain of the same species, however, alternate peaks and troughs of diapause were less evident (Veerman et al. 1988). At 19°C, using a constant photophase of 8 h, diapause incidence was high once the critical night length was exceeded, but longer scotophase durations showed a steady, non-rhythmic decline in cycles up to about 60 h. At higher temperature (22.5°C) a marked peak in diapause incidence occurred at about T 24 h (LD 8:16) and a second about 16 h later at T 40 h (LD 8:32). Although (Veerman et al. 1988) conceded that the circadian system was “. . . somehow involved in the photoperiodic induction of diapause . . .” their overall conclusion was that an hourglass timer was responsible for night length measurement.
In the Asiatic rice borer Chilo suppressalis (Walker), maintained in NH photocycles at 28°C with a constant photophase of 12 h, successive peaks of high diapause incidence at circadian intervals were also absent (Chen et al. 2011). However, as with the Dutch strain of P. brassicae (Veerman et al. 1988), a pronounced high diapause peak was observed in cycle lengths close to 24 h. As Pittendrigh (1972) observed, this is itself strong evidence that the circadian system is involved in photoperiodic timing. Both this result and that for the Dutch strain of P. brassicae may be explained in terms of the damped oscillator version of “external coincidence” as outlined above.
Tests for external coincidence
The crux of external coincidence is the dual action of light; i.e. entrainment of the photoperiodic oscillator and photoinduction of the nondiapause–diapause developmental pathways when light coincides or fails to coincide with a specific photoinducible phase (ϕi). Pittendrigh's interpretation of the two points of low diapause incidence in night interruption experiments was that point B, late in the night, represented the position of ϕi, whereas point A merely represented a phase where the light pulse elicited a phase delay of the oscillator so that ϕi was delayed into the “next” photophase (Pittendrigh & Minis 1964; Pittendrigh 1966).
Although originally designed to investigate sequential changes in the perceived Megoura hourglass, Lees (1970, 1971) performed an experiment that substantiated Pittendrigh's interpretation. This experiment, which exposed the aphids to complex cycles comprising a main photophase and supplementary pulses during the night, was conducted in two halves. In the first, he placed the 1-h supplementary light pulse 1.5 h into the dark period in a position where it coincided with point A, and then followed this pulse with a final dark period that increased from 9 to 10 h before the onset of the next “main” photophase. In the second experiment, the 1-h supplementary pulse was placed at hourly intervals (1–10 h) into the night and, in all cases, was followed by a final dark period exceeding the critical night length (9.5 h). These complex light regimes were then repeated throughout the insect's sensitive period and the resulting progeny assessed as either ovipara (long night) or virginopara (short night) producers.
In the first of these experiments, virginopara production occurred when the final dark period was less that 9.5 h, but ovipara production when it exceeded this value. In the second experiment, virginopara production was low until the scanning pulse fell about 6 h into the night where point B would occur; it then rose to 100% even though the terminal dark period exceeded the critical night length in every case. These results demonstrated that the short night effects (virginopara production) induced by light falling on point A were reversible by a subsequent long night, whereas those induced by light falling on point B were not. A similar experiment was subsequently conducted with diapause induction in the flesh fly Sarcophaga argyrostoma (Saunders 1979) with comparable effects. These results with both Megoura and Sarcophaga were clearly compatible with external coincidence indicating a photoinducible phase lying late in the night at point B. In the flesh fly, since the circadian oscillation was set to a phase equivalent to circadian time (CT) 12 at the end of the photophase (Saunders 1973a, 1976), and the critical night length was about 9.5 h, the photoinducible phase was calculated to occur at about CT 21.5 h.
In S. argyrostoma, the position of the photoinducible phase late in the night (at point B or CT 21.5 h) was further indicated in an experiment in which only that phase was illuminated (Saunders 1979). In this experiment, larvae were exposed to a so-called “T-experiment” comprising a single 1-h light pulse in cycles ranging from T 21.5 (LD 1:20.5 h) to T 30.5 (LD 1:29.5 h), a series covering the primary range of entrainment of the circadian oscillator, as judged from the 1-h phase response curve for the adult eclosion rhythm (see below). Only in a cycle of LD 1:20.5 in which the light pulse came on at CT 20 and finished at CT 23.5, and therefore illuminated the putative photoinducible phase (at CT 21.5), were long-day or low diapause effects observed. In all other regimes the photoinducible phase was calculated to fall in the dark, and diapause incidence was high. Since the postulated photoperiodic oscillator was probed by a single short pulse of light in the absence of any other photoperiodic influence, this result offered compelling evidence for the reality of the photoinducible phase at CT 21.5.
Lees (1981) subsequently examined the spectral sensitivity of points A and B in M. viciae to further characterize their differences. Test virginoparae were exposed to a 1-h pulse of monochromatic light placed either 1.5 h after dusk (at point A) or to a 30 min pulse 7.5 h after dusk (at point B) in an otherwise ovipara-producing cycle (LD 13.5:10.5 h). The intensity of the light at each wavelength was adjusted in each experiment until a 50% response (50% virginopara production) was obtained. The results for the early night interruption (at point A) showed that maximum sensitivity was in the blue (450–470 nm) but for the late night interruption (at point B) maximum effectiveness extended from the blue some way into the red. Very similar results were obtained for the flesh fly Sarcophaga similis (Maede) by Goto & Numata (2009), indicating that different types of photoreceptor are probably involved at points A and B. These authors concluded that cryptochrome may be the photoreceptor acting at point A, promoting phase delays of the photoinducible phase into the next photophase, whereas a red-sensitive opsin may be the photoreceptor molecule that directly produces diapause-averting effects at point B.
Internal coincidence and the resonance principle
In insects, fewer data support the concept of internal coincidence. However, diapause induction data in formal photoperiodic experiments with the parasitic wasp Nasonia vitripennis (Walker) (Saunders 1974) are consistent with this model. When adult wasps were exposed to NH cycles comprising 4 to 28 h of light combined with variable periods of darkness to give overall light–dark cycles up to 72 h in duration, peaks and troughs of larval diapause incidence were observed, indicating a circadian involvement in photoperiodic time measurement. Moreover, in a three-dimensional “circadian topography”, as advocated by Pittendrigh (1972), the “ascending slopes” of the high diapause peaks were found to be parallel to light-off, whereas the “descending slopes” were parallel to light-on (Saunders 1974). These results, therefore, suggested the involvement of “dawn” and “dusk” oscillators in the Nasonia photoperiodic clock, an interpretation which received additional support from the use of daily temperature cycles (thermoperiods) in the complete absence of light (see below).
Tests to discriminate external from internal coincidence
The essential difference between internal and external coincidence is that light in the former merely acts as an entraining agent for the dawn and dusk oscillators whereas in the latter it plays two roles, entrainment and photoinduction, probably using two different photopigments. Since temperature cycles and light cycles are both effective entraining agents, the use of daily thermoperiods in the complete absence of light offered a test to discriminate between the two models.
Daily thermoperiods in otherwise complete darkness were applied to the parasitic wasp N. vitripennis and to the flesh fly S. argyrostoma, insects that present properties of the internal and external coincidence types of clock, respectively. In N. vitripennis, daily square wave temperature cycles of 23°C (thermophase) and 13°C (cryophase) were found to induce larval diapause when the thermophase was less than about 12 h per day, but to induce nondiapause development when the thermophase was greater than 14 h per day; between the two was a well-marked critical thermoperiod (Saunders 1973b). Since temperature cycles are known to be effective circadian entraining agents and light was entirely excluded from this experiment, this result suggests internal coincidence. When a similar experiment was conducted with S. argyrostoma, however, the result showed that the incidence of pupal diapause in daily thermoperiods was very close to that in a constant temperature when the former were expressed as the arithmetic mean temperature of the cycle (Saunders 1984); the linear response with no “critical thermoperiod” suggested that there was no obvious thermoperiodic effect. Since diapause–nondiapause regulation relied on light, external coincidence was strongly indicated. Subject to certain caveats outlined below, these experiments suggested that the wasp and the fly measured night length by different mechanisms that resembled internal and external coincidence, respectively.
In the Lepidoptera, thermoperiodic induction of diapause has been demonstrated in several species, most notably in the cabbage butterfly, P. brassicae (Dumortier & Brunnarius 1977) and the southwestern corn borer, Diatraea grandiosella (Dyar) (Chippendale et al. 1976). These observations may indicate that the photoperiodic mechanism in these species, like that in Nasonia, is of the internal coincidence type. More significantly in this respect is probably the observation by Masaki and Kikakawa (1981) that larval diapause in the meal moth Plodia interpunctella (Hübner) may be induced by daily thermoperiods under continuous illumination. This seems to rule out any form of external coincidence since all circadian phases are illuminated, unless the “photo”-inducible phase is both light- and temperature-sensitive (see below).
In a second test of internal coincidence using N. vitripennis (Saunders 1978a), newly emerged female wasps were exposed to five consecutive cycles of long-night (LD 12:12) or short-night (LD 16:8) cycles before being transferred to continuous darkness (DD). Control groups were maintained throughout in LD 12:12, LD 16:8 or DD. Five cycles of LD 12 : 12 were previously known to be insufficient to program the wasps for the production of diapausing larvae (Saunders 1966), but prediction from the internal coincidence model was that these five initial short- or long-night cycles would entrain the separate dawn and dusk oscillators to different phase relationships, and that these phase relationships would then persist (free-run) in the ensuing darkness to induce nondiapausing or diapausing progeny, respectively. Results showed that wasps experiencing five long nights and then darkness switched one by one to the production of diapausing larvae in a manner similar to those experiencing long nights throughout their adult life. The DD control showed no similar upward trend and never exceeded 40% diapause. Wasps exposed to five short-night cycles before DD, on the other hand, continued to produce nondiapausing progeny (like the LD 16:8 control) for a further 12 days after transfer to darkness. It was concluded that separate dawn and dusk oscillators were entrained by the light-on and light-off signals of the five initial cycles and that their mutual phase relationships persisted during the following period in darkness: the results were therefore consistent with the internal coincidence model.
UNITY AND DIVERSITY IN PHOTOPERIODIC RESPONSE
Insect photoperiodic responses comprise a linked sequence of events from photoreception to the endocrine effectors regulating the seasonally appropriate developmental outcomes. Without photoreceptors the insects would not be able to sense the seasonal changes in photoperiod; without the hormonal effectors the alternate developmental pathways could not be regulated. Between photoreception and the endocrine responses are events concerned with the measurement and summation of successive night lengths during the sensitive period. This review focuses on these central events in the photoperiodic phenomenon.
The four main photoperiodic models considered here, external coincidence, internal coincidence, circadian resonance and hourglass-like responses, have been presented as explanations for these central events by various authors working on particular species. Although there is strong experimental evidence that the circadian system is involved in the first three, such evidence may be less persuasive in the fourth. However, many hourglass-like responses resemble heavily damped circadian oscillators (Bünning 1969; Saunders & Lewis 1987a,b) and even in Megoura viciae, a species presenting the strongest evidence for hourglass timing, the similarities to external coincidence are striking (Lees 1973). Therefore, although the exact role of circadian rhythmicity in insect photoperiodism remains unsettled, Bünning's hypothesis that photoperiodic time measurement is a function of the circadian system (Bünning 1936) remains the most persuasive unifying principle.
The photoperiodic models considered in this review underline differences between the various species of insect investigated, but they may not be mutually exclusive. For example, the possible relationship between hourglass-like timers and dampening circadian oscillators has already been mentioned. Night length timers of the external and internal coincidence types may also contain a resonance element since a number of separate cellular oscillators may be involved and accurate time measurement may rely, at least in part, on the degree of harmony between them.
External coincidence and internal coincidence appear to be distinctly different models, but there are a number of caveats affecting this conclusion. For example, tests to differentiate between them, as described in this review, have only been applied to a limited number of species. The use of thermoperiod to replace photoperiod as a test for internal coincidence has been conducted mainly with Lepidoptera (Chippendale et al. 1976; Dumortier & Brunnarius 1977) and Hymenoptera (Saunders 1973b) but rarely with higher Diptera (e.g. flesh flies), species that provide the strongest evidence for external coincidence. In Sarcophaga argyrostoma, moreover, daily thermoperiods were applied to larvae developing in complete darkness (Saunders 1984) but in this species the young larvae (the most photoperiodically sensitive stages) were deep in their food mass, a site where the effects of thermoperiod might have been significantly attenuated. A more meaningful experiment of this type could be performed with a species such as S. similis in which the larval sensitive period extends into the mature “wandering” larva (Goto & Numata 2009) during which stage thermoperiods might be more effective since the larvae have now left their food mass.
A further complication in the use of thermoperiod is seen in the work of Van Houten et al. 1987. These authors reported that thermoperiod affected diapause induction in the mite Amblyseius potentillae through the agency of vitamin A (as in the photoreceptor), suggesting that both photo- and thermoreception involved a rhodopsin-based system. Recent work on D. melanogaster (Shen et al. 2011) has also shown that rhodopsin is involved in larval temperature sensitivity. The “photoinducible phase”, diagnostic of external coincidence, may therefore be both light- and temperature sensitive, thereby reducing the utility of thermoperiods in distinguishing between the internal and external alternatives. In addition, the Lees experiment (Lees 1970, 1971; see above) has not been applied to those species such as N. vitripennis whose clocks most closely resemble the internal type. Clearly, many more experiments need to be conducted on a comparative basis to determine the extent of these apparent differences. Nevertheless, at the present time, the various models underline an apparent diversity in the photoperiodic response.
It seems probable that all insect photoperiodic clocks possess a common root in circadian rhythmicity but, at the same time, show diversity in detail. Hourglass-like timers not derived from the circadian system may well exist, but the evidence for them remains slim. Those cases that have been proposed, particularly those suggesting day-length timers (Bradshaw et al. 2003), are sufficiently unusual to warrant careful investigation involving a full range of experiments such as night-interruption, Nanda–Hamner (NH), Bünsow, bistability, thermoperiod and T-experiments (Saunders 2010b). All these experiments are designed to test for canonical circadian properties in the photoperiodic response, such as free-running rhythmicity with a near-24-h periodicity and entrainment by light and temperature cycles. In particular, NH experiments using a range of photophases are required to determine whether time measurement begins at light-on or light-off and, therefore, whether the photoperiodic mechanism measures day length or night length (Saunders & Bertossa 2011).
Modern photoperiodic responses probably evolved on numerous occasions as insects, or groups of insects, moved into higher latitudes where they experienced seasons inimical to growth and development. In doing so, the pre-existing circadian system provided the night-length measurement inherent in the response. Based on this unifying principle was diversity in detail arising from differences in the molecular mechanisms that had evolved in the different insect groups. For example, in many Lepidoptera (Zhu et al. 2005), Hymenoptera, mosquitoes and beetles, mammalian-type cryptochrome-2 (cry2) acts as a transcriptional repressor of the CLOCK/CYCLE dimer in the circadian autoregulatory feedback loop (Yuan et al. 2007), whereas in D. melanogaster and possibly other higher Diptera, cry1 encodes a blue light photoreceptor used in entrainment. The form of the clock involving cry2 is regarded as ancestral (Sandrelli et al. 2008; Merlin & Reppert 2010; Tomioka & Matsumoto 2010) whereas that in the fruit fly is thought to be derived. In the Drosophila clock, cry2 has been lost, whereas in the honey bee (Rubin et al. 2006) and N. vitripennis it is cry1 that has been lost. If these differences, and others, also occur in the photoperiodic mechanism, substantial differences in the way clock genes may be involved in photoperiodism might also be expected (Saunders & Bertossa 2011). Further investigations will be needed to see whether such differences underlie the diversity suggested by the formal experiments discussed in this review.
In conclusion, insect photoperiodic mechanisms find unity in their common circadian origin, but diversity resulting from their separate and parallel evolution between different groups as they extended into areas where seasonal diapause offered marked selective advantages. Consideration of this diversity may provide valuable insights into the comparative, and molecular, aspects of time measurement in insect photoperiodism.