Insect photoperiodism: effects of temperature on the induction of insect diapause and diverse roles for the circadian system in the photoperiodic response



This review considers the effects of temperature on insect diapause induction and the photoperiodic response, and includes constant temperature, temperature cycles, pulses and steps in daily light–dark cycles, constant darkness and in constant light, all with reference to various circadian-based “clock” models. Although it is a comparative survey, it concentrates on two species, the flesh fly Sarcophaga argyrostoma and its pupal parasite Nasonia vitripennis, which possess radically different photoperiodic mechanisms, although both are based upon the circadian system. Particular attention is given to the effects of daily thermoperiod in darkness and to low and high temperature pulses in conjunction with a daily light–dark cycle, treatments that suggest that S. argyrostoma “measures” night length with a “clock” of the external coincidence type. However, N. vitripennis responds to seasonal changes in photoperiod with an internal coincidence device involving both “dawn” and “dusk” oscillators. Other species may show properties of both external and internal coincidence. Although the precepts of external coincidence have been well formulated and supported experimentally, those for internal coincidence remain obscure.


At higher latitudes, seasonal changes in photoperiod and temperature (and interactions between them) are the major environmental signals regulating diapause onset in insects. Thus, lengthening nights (or shortening days), together with falling autumnal temperatures, lead to an over-wintering diapause. On the other hand, the shorter nights (or longer days) of summer combined with elevated temperatures lead to continuous or nondiapause development (Saunders 2002).

Photoperiodic responses are generally thought to involve a sequence of events from: (i) photoreception; through (ii) measurement of night length (or perhaps, in a few cases, day length); to (iii) the accumulation of inductive photoperiods by a “counter” mechanism (Saunders 1981); and (iv) the downstream regulation of events leading to release or retention of neurohormones regulating diapausing or nondiapausing development.

Although photoperiod is the major environmental factor involved in this seasonality, temperature may affect the induction of diapause at a number of levels including photoreception, time measurement, accumulation of inductive photoperiods and neurohormone regulation. This review surveys some of these temperature effects in relation to various clock models, which are outlined below. Although the review is a comparative survey of species with an over-wintering diapause, particular attention is given to two, the flesh fly Sarcophaga argyrostoma (Robineau-Desvoidy) and the parasitic wasp Nasonia vitripennis (Walker), which appear to have substantially different photoperiodic clock mechanisms. The review includes tests for photoperiodic clock models based on various temperature treatments. It does not include the effects of temperature on diapause development and termination.

Photoperiodic response

Insects living at higher latitudes frequently show a “long-day” photoperiodic response in which activity, development and reproduction occur during the short nights of summer but an over-wintering diapause occurs when autumnal night lengths exceed a critical value (Saunders 2002). The blow fly Calliphora vicina Robineau-Desvoidy presents a life cycle of this type (Saunders 1987) (Fig. 1). In this species a maternally operating photoperiod regulates larval diapause provided that the temperature of the larval environment falls below about 15°C. This type of summer-active or “long-day” response is dominated by the ecologically important critical night length (or day length) separating the short summer nights, inducing continuous nondiapause development from the longer autumnal nights leading to diapause. In C. vicina the maternal critical night length is at about 9.5 h (or critical day length of about 14.5 h.) per 24 h. The photoperiodic response curve of such a long-day species also frequently shows a decrease in diapause incidence under very long nights and in continuous darkness. Models to account for this type of photoperiodic response, and others, are outlined in the next section.

Figure 1.

Photoperiodic regulation of larval diapause in the blow fly Calliphora vicina (Saunders 1987). Adult flies were exposed to a range of photoperiods at 20°C and the resulting larvae were maintained in darkness at <15°C. The photoperiodic response curve shows a “long-day” response with a well-marked critical day length (CDL) at about 14.5 h per day (or critical night length of about 9.5 h) and a drop in diapause incidence under very short days and continuous darkness. Models for the photoperiodic response need to account for both of these characteristics.

Circadian and hourglass-like models for photoperiodic time measurement in insects: the art of the possible

As with all theoretical constructs, models for the photoperiodic mechanism describe possible components of the system rather than concrete elements. Therefore clock models merely explore ways that photoperiodic time measurement may occur, although the best of them are soundly based on experimental data and serve to predict the outcome of future experiments. Historically, some of the most useful and predictive models are those that propose that core events in photoperiodic time measurement are functions of the circadian system.

Bünning's hypothesis and coincidence models

The hypothesis that the circadian system and its “clock” genes provide the time measurement for the photoperiodic response remains somewhat controversial, although increasing evidence (e.g. Koštál 2011; Saunders & Bertossa 2011; Goto 2013) supports this proposition. Both the measurement of successive photoperiods and their accumulation by the counter mechanism are thought to be functions of the circadian system (Saunders 2012) although non-circadian or hourglass-like timers remain a possibility in some species.

The theoretical association between circadian rhythmicity and photoperiodism began more than 70 years ago when the German plant physiologist Erwin Bünning (1936) proposed a simple model now known as Bünning's general hypothesis (Pittendrigh 1972). Experimental evidence in favor of some sort of circadian involvement (in terms of entrainment of circadian oscillations by light and temperature cycles) has been reviewed in a number of publications (Pittendrigh & Minis 1964; Pittendrigh 1966, 1972; Saunders 2002, 2010, 2011, 2012). This will not be re-examined in detail in the current review.

One of the most frequently used procedures to investigate possible circadian involvement in photoperiodism is the so-called Nanda–Hamner or “resonance” protocol (Nanda & Hamner 1958). In this type of experiment, organisms are exposed to a range of light–dark (L–D) cycles each containing a “short-day” photophase (say L = 8 or 12 h) combined, in different experimental sub-sets, with variable periods of darkness (D hours) to give overall cycles (L + D = T hours) covering several multiples of the circadian period. In a number of species in the Orthoptera, Coleoptera, Lepidoptera, Hymenoptera, Diptera and Acarina, Nanda–Hamner (NH) experiments have revealed periodic peaks and troughs of diapause incidence at about 24 h intervals as T is extended. Such periodicity is usually attributed to a role for the circadian system in photoperiodic regulation (Saunders 2002) (Fig. 2).

Figure 2.

Results of Nanda–Hamner experiments in which a “short-day” photophase (L = 8–12 h) is coupled, in different experimental sub-sets, with dark periods (D) of increasing duration to provide light–dark cycles (L + D = T h) up to 84 h or more. In many species, such as the blow fly Calliphora vicina (Vaz Nunes et al. 1990) and the flesh fly Sarcophaga argyrostoma (Saunders 1973a) diapause incidence (the long-night response) shows alternate peaks and troughs with a periodicity of about 24 h, indicating a circadian involvement in the photoperiodic response. In others, such as the aphid Megoura viciae (Lees 1986), ovipara production (the long-night response) is low until T exceeds the critical night length, after which it remains consistently high, indicating an hourglass-like response.

Several models for circadian involvement in insect photoperiodism have arisen from such studies (Vaz Nunes & Saunders 1999; Saunders 2002, 2011). Two of the most influential of these are “internal coincidence” and “external coincidence” (Pittendrigh 1972; see also Saunders 2011). Internal coincidence suggests that the photoperiodic mechanism is based on (at least) two circadian oscillators; seasonal changes are sensed by changing “internal” phase relationships between them as days (or nights) lengthen or shorten. This kind of model, often invoking separate “dawn” and “dusk” oscillators, is exemplified by the parasitic wasp Nasonia vitripennis. The model suggests that light has a single role, that of entrainment (Saunders 1974). In mammals, similar two-oscillator models based on separate “morning” and “evening” oscillators have also been proposed (Daan et al. 2001; Oster et al. 2002).

External coincidence, on the other hand, envisages a single oscillator entrained and phase-set by the light cycle in such a way that, after a photophase of about 12 h or more, the oscillator is reset to a constant phase equivalent to the start of the subjective night, defined as Circadian time, CT 12 (Pittendrigh 1966; Saunders 2012, 2013) whereupon it is in a position to “measure” the duration of the following dark hours. A particular light-sensitive (or “photo-inducible”) phase (φi) then occurs at the end of the critical night length (Fig. 3). Under the increasingly long nights of autumn the photoinducible phase begins to fall in the dark of each cycle, and accumulation of such events by a “counter” mechanism during the insect's photoperiodic sensitive period (Saunders 1981) leads to the diapause state. On the other hand, under the short nights of summer, the photoinducible phase is illuminated by dawn light of each cycle leading to continuous or nondiapause development. Unlike internal coincidence this model proposes that light has two roles: (i) entrainment of the circadian oscillator; and (ii) illumination (or non-illumination) of the photoinducible phase. External coincidence seems to be the most appropriate model to explain photoperiodic induction of pupal diapause in flesh flies, Sarcophaga spp. (Saunders 2002, 2011).

Figure 3.

External coincidence model for photoperiodic timing (after Pittendrigh 1972). The circadian oscillation, here shown as a phase response curve, is reset to a constant phase (Circadian time, CT 12) at the end of the photophase (vertical arrows). (A) Long night cycles; the photoinducible phase (φi) falls in the dark (solid circles), leading to the autumnal pathway (diapause induction). (B) Short night cycles; φi is illuminated by the dawn transition of the daily photophase (open circles) to give the summer developmental pathway (nondiapause development). Open horizontal arrows show the “movement” of the dawn transition of the daily photophase in relation to the oscillation as nights lengthen in the autumn (A) or shorten in the summer (B).

Tests to distinguish internal from external coincidence have been described in a number of earlier publications (Saunders 2002, 2011). Other tests, particularly those employing non-photic zeitgeber such as thermoperiod and temperature pulses, are discussed in this review. In addition, recent considerations of the role of light in photoperiodism, particularly spectral sensitivity studies (Saunders 2012) and genome sequencing, have underlined differences between these two types of photoperiodic mechanism.

Damped oscillator version of external coincidence

Lewis and Saunders (1987) proposed an extension of the external coincidence model, specifically for the photoperiodic induction of pupal diapause in S. argyrostoma, but also to address many of the known properties of the photoperiodic mechanism in a range of other species. This model proposes that the circadian oscillator involved in external coincidence may dampen in extended periods of darkness, the degree of dampening varying between species and environmental conditions. This dampening oscillator is derived from a feedback control systems approach to circadian rhythmicity (Gander & Lewis 1979; Christensen et al. 1984; Lewis 2002) which, by incorporating the accumulation of successive photoperiods by a “counter” mechanism (Saunders 1981), converted the essentially qualitative model of external coincidence into a quantitative one. Many of the theoretical components of this modified external coincidence model also invite comparison with concrete molecular components of the circadian system uncovered in more recent investigations (e.g. Hall 2003; Saunders et al. 2004).

The damped oscillator model follows the synthesis and loss of a hypothetical substance (called chemical “c” in the model), which acts as a component of the negative feedback loop generating the circadian cycle. The degree of dampening of this oscillator is determined by the rate of synthesis (SR) of c, perhaps representing the rate of assembly of the PERIOD–TIMELESS protein (PER–TIM) dimer in a Drosophila-type oscillation. In computer simulations a high value of SR gives a self-sustained oscillation, whereas a low value of SR leads to rapid dampening. Circadian period (τ) is determined by the time delay (TD) between the synthesis of c and its degradation, where a high value of TD gives a longer value of τ than a lower value of TD. The effect of simulated “temperature” on the oscillation in darkness is to give a more heavily dampened oscillation at lower values. Last, simulated “light” destroys c in proportion to its intensity, but when light is removed a high synthesis rate gives a rapid increase in c, thereby boosting the rhythm.

In the damped circadian oscillator model, diapause depends on the accumulated concentration of a “diapause titer” (or INDSUM) present in each individual at the end of the photoperiodic “sensitive period”. The diapause titer is generated in each cycle when the amplitude of the oscillation remains above a threshold but occurs in the dark. If the oscillation declines below threshold the diapause titer is not generated and the total production of INDSUM and consequently percentage diapause is curtailed.

In the model, it is difficult to assign the photoinducible phase to a particular point on the oscillation. However, if INDSUM represents a light-sensitive product that is synthesized and accumulated during the above-threshold portion of the curve, the titer of INDSUM will become maximal toward the end of this period. Although this aspect of the model invites a somewhat circular argument, calculations based on the end of the above-threshold portion representing the photoinducible phase provide good simulations of the external coincidence model for the photoperiodic mechanism in S. argyrostoma.

Light falling on the photoinducible phase therefore affects the level of INDSUM by promoting its destruction and causing a reduction in the final incidence of diapause. In common with most long-day insects, diapause incidence in S. argyrostoma is also higher at lower temperature, whether in darkness or in long nights. For this reason the model also assumes that INDSUM production is inversely related to temperature. The model therefore resembles a circadian-based device of the external coincidence type but incorporates most of the known properties of the system gained from real experiments. Computer simulations using this model and systematic variation of its various components (e.g. synthesis rate, time delay, “temperature” and “light intensity”) provide close parallels to observed photoperiodic responses in a wide range of insects (Lewis & Saunders 1987; Saunders & Lewis 1987a,b).

Effects of temperature on photoperiodic mechanism

The following sections describe some of the effects of temperature on diapause induction and the photoperiodic mechanism. These include constant temperature, temperature cycles, pulses or steps, all of them in conjunction with concurrent light-dark cycles (LD) or in conditions of continuous darkness (DD) or of light (LL). Results of these experiments are discussed in terms of some of the photoperiodic clock models described above.

Particular attention is given here to “long-day” species such as the ovoviviparous flesh fly S. argyrostoma and its pupal parasite, the wasp N. vitripennis. In the fly, photoperiodic sensitivity begins in the intra-uterine embryos (Denlinger 1971) and may continue through larval development up to puparium formation (Saunders 1971). Long autumnal nights experienced during this sensitive period induce diapause in the pupae, whereas short summer nights lead to uninterrupted or nondiapause development to the adult flies (Saunders 1971). In the wasp, however, photoperiodic sensitivity is strictly maternal with diapause occurring in the larval progeny just before their pupation. Photoperiod has no diapause-regulating role in the larvae themselves (Saunders 1966). In a group of wasps maintained under diapause-inducing long nights, individual females switch abruptly from the production of nondiapausing to diapausing larvae, the switch within the population being completed by about 8–10 days after adult eclosion. Under short nights, however, the switch to the production of diapausing progeny, if it occurs at all, is delayed until near the end of adult life (Saunders 1966).

Effects of constant temperature, with photoperiod and in darkness

In most insects with an over-wintering diapause, low temperature acts in conjunction with long nights to induce diapause while higher temperature acts with short nights to reduce it (e.g. Danilevskii 1965; Saunders 2002). Figure 4 shows such a relationship for the flesh fly S. argyrostoma (Saunders 1971). Conversely, in short-day insects (e.g. Kogure 1933; Masaki 1980), not considered further in this review, high temperatures augment summer diapause under short night photoperiods but low temperatures during autumnal long nights reduce it.

Figure 4.

Effect of constant temperature (15, 20 and 25°C) on the photoperiodic induction of pupal diapause in the flesh fly Sarcophaga argyrostoma, showing a higher incidence of diapause at lower temperature. Data from Saunders (1971).

In long-day insects an inverse relationship between temperature and diapause incidence may also be seen in the results of Nanda–Hamner (NH) experiments. For example, in S. argyrostoma raised as larvae at 16, 18 or 20°C in a series of light–dark regimes consisting of 12 h of light and increasing periods of darkness to give overall light cycles (T) from 12 to 84 h (Saunders 1973a), pupal diapause incidence at 16°C approached 100% under all T cycles greater than about 24 h to give a response that resembled an essentially “negative” NH profile similar to that in the aphid Megoura viciae Buckton (Lees 1965, 1986) (Fig. 5), whereas at higher temperatures diapause incidence is reduced. The appearance of negative or apparently hourglass-like NH responses at lower temperature has also been recorded in a number of other species including Drosophila auraria Peng (Pittendrigh 1981), D. triauraria Bock & Wheeler (Yoshida & Kimura 1993), the blow fly C. vicina (Vaz Nunes et al. 1990), the cabbage white butterfly Pieris brassicae (L.) (Veerman et al. 1988) and the predacious mite Amblyseius potentillae (Garman) (Van Houten & Veenendaal 1990), indicating its generality.

Figure 5.

Effect of constant temperature (16, 18 and 20°C) on diapause induction in Sarcophaga argyrostoma exposed to Nanda–Hamner photocycles, showing an overall reduction in the heights of the diapause maxima as the temperature rises. At the lowest temperature tested (16°C) the response resembles that of an “hourglass”. Data from Saunders (1973a).

At higher temperature a clear NH rhythm becomes evident in S. argyrostoma (Fig. 5) but with diapause incidence reduced at all T values. At the intermediate temperature tested (18°C) the magnitude of the high diapause peaks declined with an increase in T, indicating that the photoperiodic oscillator slowly dampens in extended periods of darkness. At 20°C, the highest temperature shown in Figure 5, the incidence of diapause was low throughout but the three peaks (close to T 27, 51 and 75 h) were of roughly the same magnitude, suggesting that the oscillator was more self-sustained at this temperature.

The results described in this section suggest that constant temperature has at least two effects on diapause incidence. First, low temperature has a general diapause-inducing effect, which may be seen in continuous darkness, under long nights and in Nanda–Hamner experiments. Second, temperature is thought to affect the dampening rate of the photoperiodic oscillator, as revealed in NH experiments, the oscillator being more self-sustained under higher temperature despite the lowered overall incidence of diapause. Both of these characteristics are in accord with the damped oscillator model for the photoperiodic mechanism in S. argyrostoma (Lewis & Saunders 1987).

Effects of concurrent daily light and temperature cycles

In their natural environment, insects experience concurrent daily cycles of light (photoperiod) and temperature (thermoperiod) with nights colder than the days. In early work by Goryshin (1955) with the knot grass moth Acronycta rumicis (L.), larvae were exposed to daily thermoperiods (17–30°C) in conjunction with various light–dark cycles. The resulting pupae were then assessed for diapause. Results showed that the incidence of diapause was greater when days were at 30°C and the nights at 17°C, than when the nights were warmer than the days. It was concluded that the photoperiodic effect was more dependent on the temperature of the dark period than the light.

In a series of papers Beck (1982, 1984, 1985, 1987) examined the effect of day and night temperatures on diapause induction in the European corn borer Ostrinia nubilalis and also concluded that the temperature of the cool dark period of the daily cycle was “more important” for diapause regulation than the warm light phase. Similar results were obtained by Van Houten et al. (1987) for the predacious mite Amblyseius potentillae.

Effects of daily thermoperiod in the absence of a light cycle

A daily temperature cycle (or thermoperiod) may also have diapause regulating effects in the absence of a light cycle, most frequently in continuous darkness (DD) but occasionally also under continuous light (LL). For example, working with larval diapause induction in the European corn borer Ostrinia nubilalis (Hübner), Beck (1962) exposed larvae, in DD, to a daily temperature cycle comprising 11 h at 31°C, 11 h at 10°C and the remaining two 1-h periods in the cooling and warming phases. Under this regime, almost all larvae entered diapause, whereas control larvae maintained at constant temperatures of 31°, 26° and 21°C rarely became dormant.

Similar effects of thermoperiod in DD were recorded by Menaker and Gross (1965) for the pink bollworm moth Pectinophora gossypiella (Saunders) and by Goryshin and Kozlova (1967) for the Lepidoptera P. brassicae, A. rumicis and Spilosoma menthastri (L.).

A clear-cut dependence of diapause induction on thermoperiod in continuous darkness was later demonstrated for N. vitripennis (Saunders 1973b). In this experiment, groups of wasps were raised from the egg stage in complete darkness (DD) until the adult instar, thus avoiding the use of red light known to be effective in photoperiodic photoreception (Saunders 1975, 2012). Adult wasps were then maintained, also in DD, in a range of square-wave temperature cycles between a higher temperature (thermophase, T = 23°C) and a lower temperature (cryophase, C = 13°C), ranging from TC 6 : 18 h to TC 18 : 6 h. Control groups of wasps were maintained at constant temperatures of 13 and 18°C (the arithmetic mean of 13 and 23°C), again in DD. The proportions of female wasps producing diapausing or nondiapausing progeny were then assessed by maintaining the eggs deposited during a 2-day “test period”, between the 15th and 17th day of adult life, after a further 15 days in darkness. Results (Fig. 6) showed that almost all of the progeny produced by wasps maintained in “short-day” thermoperiods (TC 6 : 18 h to TC 10 : 14 h) became diapausing larvae, but all of those produced by wasps in “long-day” thermoperiods (TC 14 : 10 h to TC 18 : 6 h) developed without arrest. An abrupt “critical thermophase” occurred at about TC 13 : 11 h. The proportions of females producing diapause progeny at constant temperatures of 13 and 18°C (the mean of 13 and 23°C) were about 44% and 2%, respectively, illustrating the inverse relationship between temperature and diapause incidence outlined above. In addition, diapause incidence under TC 12 : 12 h was about 70%, considerably higher than the 2% at a constant temperature of 18°C, clearly demonstrating the inductive effect of thermoperiod over a constant temperature with the same mean value.

Figure 6.

Induction of diapause in the parasitic wasp Nasonia vitripennis and in the flesh fly Sarcophaga argyrostoma exposed to daily thermoperiods in complete darkness. In N. vitripennis, a daily thermoperiod (23°C : 13°C) produces a response curve resembling that for light (Saunders 1974). On the other hand, in S. argyrostoma, exposure to a thermoperiod (25°C : 15°C), produces a straight line response with no critical thermophase (Saunders 1984). The difference between the responses of these two species to thermoperiod suggests that N. vitripennis possesses a photoperiodic clock of the internal coincidence type, but that in S. argyrostoma is of external coincidence.

The experiment using a daily thermoperiod in darkness was conceived as a test for internal coincidence in N. vitripennis and considered to be consistent with that model since light was not required for the response, thermoperiod most probably acting as zeitgeber for circadian entrainment as a substitute for photoperiod. Similar responses were later recorded for larval diapause induction in the south-western corn borer Diatraea grandiosella Dyer (Chippendale et al. 1976), for pupal diapause induction in P. brassicae (Dumortier & Brunnarius 1977) and for the induction of diapause in the predacious mite A. potentillae (Van Houten et al. 1987). Working with O. nubilalis maintained in darkness, Beck (1982) showed that diapause induction required a cold phase in excess of about 9.5 h with a temperature between about 10 and 17°C. With the Indian meal moth Plodia interpunctella (Hübner), however, Masaki and Kikukawa (1981) showed that daily thermoperiod could regulate diapause induction in both DD and in LL. These observations are all consistent with some form of internal coincidence.

Experiments using daily thermoperiods with S. argyrostoma produced quite different results. Larvae, newly deposited by female flies maintained in DD, were established in cultures subjected to daily temperature cycles (T = 25°C; C = 15°C) ranging from TC 4 : 20 h to TC 20 : 4 h, all in continuous darkness. Pupae were later assessed for diapause or nondiapause development. The results showed a linear relationship between thermoperiod and diapause incidence with no evidence of a “critical thermophase” as in Nasonia (Fig. 6). Plotting diapause incidence as a function of the arithmetic mean temperature of the thermoperiodic cycle (see Saunders 1984) suggested that the observed increase in diapause at shorter thermoperiods was merely an effect of the overall lower temperature. These results with S. argyrostoma are consistent with external coincidence, rather than with internal coincidence as in the wasp.

Photoperiod with concurrent low or high temperature pulses

Early experiments using low temperature pulses (LTPs) and high temperature pulses (HTPs) in conjunction with a daily photoperiod were often designed to evaluate effects of temperature extremes on day- or night-length measurement. For example, working with the knot grass moth A. rumicis, Danilevskii (1965) reported work using 3-h periods at 5°C occurring at different times of either the photophase or scotophase in cycles of LD 17 : 7 or LD 14 : 10. Chilling at the beginning or at the end (but not the middle) of the 17-h photophase converted the response from that of a long day (0% diapause) to that of a short day (100% diapause). It was concluded that chilling interfered with photoreception thereby converting the 7-h scotophase into the equivalent of a diapause-inductive 10-h night.

In a later study using Megoura viciae, Lees (1986) exposed aphids, at 15°C, to a 4-h LTP at the beginning of an inductive long night. Night length measurement was accomplished perfectly when the cold pulse occurred at temperatures down to 6°C but temperatures lower than this apparently caused time measurement to “slow down”. Time measurement still occurred at −3°C but only at one-quarter of its rate. Chilling at temperatures down to −3°C during the photophase was thought to interfere with photoreception, the chilled portion being “read” as darkness.

The simplest explanation for these results is that low temperature interferes with either (or both) photoreception or the process of time measurement. In terms of the nature of the photoperiodic mechanism itself, however, it has to be recognized that low and high temperature pulses will also cause phase shifts of the oscillator(s) involved in the time measurement process. Experiments considering such phase changes, and tests to discriminate between external and internal coincidence, were therefore conducted with N. vitripennis and its host S. argyrostoma in cultures exposed to pulses of low and high temperature at different phases of the light–dark cycle. The results of these experiments are dealt with in the next two sections.

Low temperature pulses in S. argyrostoma: external coincidence

Larval cultures of S. argyrostoma were subjected to daily low temperature pulses (LTPs at 5°C) at sequential times during the night of an otherwise diapause inductive light cycle of LD 14 : 10 (Saunders 1984). Figure 7 shows that when the LTP fell in the first half of the night it caused an increase in the incidence of pupal diapause, but LTPs falling in the second half of the night caused a low incidence of diapause. These results were interpreted in terms of the entrainment of the circadian oscillator involved in diapause regulation exposed to the combined phase-resetting effects of the daily light and temperature pulses. This interpretation, strongly suggesting external coincidence, is as follows.

Figure 7.

(A) Induction of pupal diapause in the flesh fly Sarcophaga argyrostoma maintained under LD 14 : 10 at 17°C with low temperature pulses (3-h LTPs at 5°C) systematically scanning the 10-h “night”, showing an increase in the incidence of diapause when the LTP falls in the early part of the night, but reduced diapause late in the night. Horizontal line (control) shows the incidence of diapause in LD 14 : 10 without LTP. Data from Saunders (1984). (B) Interpretation of this result as a function of the circadian system in photoperiodic time measurement (schematic). The phase of the oscillation is reset in each cycle to CT 12 (the beginning of the subjective night) with the photoinducible phase (closed circle, φi) occurring 9.5 h (the critical night length) later, just before dawn. The curve is a schematic representation of the phase response curve for LTPs in D. pseudoobscura with phase advances early in the subjective night and delays in the late subjective night (data from Zimmerman et al. 1968; Chandrashekaran 1974). When the LTP falls early in the night (at a), φi is advanced to earlier times in the night (left pointing arrow), enhancing the diapause inductive effect; when the LTP falls late in the night (at b), φi is delayed from the dark into the light (right pointing arrow) and diapause incidence is greatly reduced. The results shown in (A) are therefore consistent with external coincidence.

Regular daily cycles of light intensity and temperature are the main entraining agents (or zeitgeber) for circadian rhythms and also act as signals for the seasonal photoperiodic clock. When the daily photophase is of sufficient duration and intensity, circadian oscillations are frequently set, at light-off or “dusk”, to a narrow range of phases close to the beginning of the subjective night (defined as Circadian time, CT 12) (Pittendrigh 1966). In the external coincidence model the photoinducible phase is then timed to occur 9.5 h later at the end of the critical night length (see Fig. 3 for details). In laboratory experiments, short supplementary light pulses falling in the early subjective night (CT 12 to 18) cause phase delays, but pulses falling in the late subjective night (CT 18 to 24) cause phase advances. The degree of dampening of the oscillator is determined by the rate of synthesis (SR) of c, perhaps representing the rate of assembly of the PERIOD-TIMELESS protein dimer (Pittendrigh 1966; Winfree 1970).

Perturbations caused by temperature pulses would also yield PRCs but these are less frequent in the literature and are not known for the Sarcophaga case. In the absence of an LTP phase response curve for S. argyrostoma, a PRC for LTPs based on the eclosion rhythm of Drosophila pseudoobscura Frolova & Astaurov (Zimmerman et al. 1968; Chandrashekaran 1974) is used. A schematic representation of such a curve (Fig. 7B) showing phase advances in the early subjective night (CT 12 to 18) and phase delays in the late subjective night (CT 18 to 24), is assumed to be a qualitative representation of the unknown PRC for 3-h LTPs in S. argyrostoma. Using this PRC, an LTP commencing early in the subjective night would be expected to phase advance the photoinducible phase to earlier times of the night and thereby enhance the light cycle's diapause-inducing effect; however, an LTP commencing in the late subjective night would be expected to phase delay the photoinducible phase into the next photophase, thereby reducing diapause incidence. The results of the chilling experiment with S. argyrostoma, therefore, are fully consistent with the external coincidence model.

Low and high temperature pulses in N. vitripennis: internal coincidence

Reversals of photoperiodic effect by chilling were also recorded in N. vitripennis (Saunders 1967, 1969). When wasps were chilled at 2°C for 4 h at the beginning or the end of the night in LD 14 : 10, the response was converted from that of a long night (high diapause incidence) to that of a short night (low diapause incidence). Conversely, chilling in the light period of this cycle merely strengthened the short day response. In the converse experiment, at LD 16 : 8, chilling in the dark had no effect but chilling in the light reversed the response from that of a long day (low diapause incidence) to that of a short day (high diapause incidence).

These results, unlike those for S. argyrostoma reviewed above, indicated that both day and night lengths are “measured” by the photoperiodic mechanism in N. vitripennis. Together with the results of Nanda–Hamner experiments (Saunders 1974, 2012) and of thermoperiod in darkness (Fig. 6; Saunders 1973b) these results appear to be consistent with an internal coincidence model for the photoperiodic clock in N. vitripennis with separate “dawn” and “dusk” oscillators.

In a later investigation, adult females of N. vitripennis in an otherwise diapause inductive light cycle of LD 14:  10 and a “background” temperature of 18°C were systematically exposed to 3-h LTPs (at 2°C) or HTPs (at 35°C) across both the light and the dark components of the cycle (Saunders 2002). Results showed that LTPs or HTPs timed to occur during the 14-h light component of this cycle had little effect on reversal of diapause incidence, thus confirming earlier observations (Fig. 8; Saunders 1969). However, both LTPs and HTPs scanning the 10-h night produced a low incidence of diapause (a delayed switch) when the temperature pulses (either high or low) fell in both the early and the late portions of the night. Temperature pulses falling in the middle of the night close to “subjective midnight” had less effect. Comparable results with both LTPs and HTPs were later obtained with Drosophila triauraria (Yoshida & Kimura 1999).

Figure 8.

Effect of low temperature pulses (LTPs, 3 h at 2°C) and high temperature pulses (HTPs, 3 h at 35°C) systematically scanning a light cycle (LD 14 : 10, 18°C) on the photoperiodic responses of the parasitic wasp Nasonia vitripennis. A rapid switch to the production of diapausing progeny (∼10 days) represents a long night effect (high diapause incidence), a delayed switch (20+ days) a short night effect (low diapause incidence). In contrast to Sarcophaga argyrostoma (Figure 7), results show a bimodal response during the 10-h “night” with diapause reversal (delayed switches) occurring both early and late in the night, for both LTPs and HTPs. The two horizontal lines close to 10 days represent control switching points for LD 14 : 10 at 18°C without temperature pulses.

Therefore, unlike the results for LTPs in S. argyrostoma, which showed a single lowered incidence of diapause or short night response in the second half of the night (Fig. 7), results for LTPs (and HTPs) in N. vitripennis produced a markedly bimodal response with short night effects both early and late in the 10-h dark phase. This bimodal response does not lend itself to an interpretation in terms of a simple external coincidence model, but may be explained by a two-oscillator internal coincidence mechanism with separate “dawn” and “dusk” oscillators.

Chilling and heating probably have two effects upon the photoperiodic mechanism: (i) they might adversely affect either photoreception or the process of time measurement; and (ii) they probably cause phase shifts of the constituent oscillators involved in the photoperiodic response. Such temperature-pulse phase shifts have not been described in N. vitripennis. They are indicated, however, in the results of experiments in which wasps were exposed to a 4-h LTP (2°C) at the beginning of the photophase of a diapause inductive light cycle (LD 14 : 10) and the 10-h night systematically scanned by a 1-h light pulse to reveal points of long-day sensitivity (diapause reversal) (Saunders 1969). A low-temperature pulse in this position caused a delay in the peak of diapause reversal, suggesting that at least part of the photoperiodic mechanism (the “dawn” oscillator?) was phase-set by light-on (Fig. 9). The results shown in Figure 9, however, provide no data for possible effects of the low-temperature pulse on the putative “dusk” oscillator and further experiments are needed. The results, however, are at least partially consistent with the concept of internal coincidence. This result is therefore consistent with the concept of internal coincidence.

Figure 9.

Effect of a low-temperature pulse (LTP 4 h at 2°C) at the beginning of the photophase of an otherwise diapause-inductive light–dark cycle (LD 14 : 10) on diapause reversals in Nasonia vitripennis revealed by systematic scanning of the 10-h “night” by 1-h supplementary light pulses. Solid circles, scanning the night without the 4-h LTP showing two points of low diapause incidence (or delayed switch to the production of diapausing progeny) one early in the night (at A) and one later in the night (at B); open circles, scanning the night in cycles with the 4-h LTP, showing that peak A is delayed indicating that at least one component of the photoperiodic mechanism is timed from “dawn”. This result is considered to be consistent with internal coincidence. Data from Saunders (1969).

In the interpretations given above, a fundamental difference is indicated between external coincidence involving a photoinducible phase (in S. argyrostoma) and internal coincidence (in N. vitripennis), which lacks such a phase and relies on changing phase relationships between constituent oscillators within the circadian system.

Temperature effects on the photoperiodic counter

Temperature also affects the summation of inductive light cycles by the counter mechanism through an interaction between the photoperiodically sensitive period and the number of light–dark cycles needed to induce diapause. For example, in N. vitripennis the sensitive period (SP) begins soon after adult eclosion (or emergence from the host puparium) and since the wasps’ larval progeny are unresponsive to photoperiod (Saunders 1966), SP effectively ends when the eggs are laid. Under long night (diapause inductive) conditions wasps initially produce eggs that develop into non-diapausing larvae, but switch one-by-one to the production of progeny that become dormant, the mean number of long night cycles required to effect this switch constituting the “required day number” (RDN) (Saunders 1966).

The effects of temperature on the interaction between the SP and the RDN in N. vitripennis are shown in Figure 10A. In this experiment, groups of wasps were exposed to diapause inductive cycles of LD 12 : 12 in constant temperatures of 15, 20, 25 and 30°C (Saunders 1966). Results showed that the mean life span of the wasps was temperature dependent, ranging from about 11.5 days at 30°C to about 33 days at 15°C. The pattern and duration of daily progeny production, equivalent to the maternal sensitive period, SP, was also temperature dependent with a peak production on about the fourth day at 30°C to about the 14th day or later at 20 and 15°C. However, the age the switch to the production of diapausing progeny (the RDN), showed a marked degree of temperature compensation, only changing from about 7 days at 30°C to 8.4 days at 15°C. Consequently, interaction between the temperature-sensitive SP and the temperature-compensated RDN meant that, at 30°C, only 27.4% of the offspring entered diapause, but at 25, 20 and 15°C, diapause incidence was, respectively 61.0%, 71.2% and 90.8%.

Figure 10.

Effect of temperature on the interaction between the length of the sensitive period (SP) and the “required day number” (RDN). (A) Adults of Nasonia vitripennis maintained under diapause inductive long nights (LD 12 : 12) at four temperatures (30, 25, 20 and 15°C). The solid line shows the rate of egg production, with the shaded portion of each polygon the proportion of the larvae subsequently entering diapause. The dotted line shows the survival rate for the wasps in each group. Temperature has a marked effect on survival and oviposition rate (equivalent to the maternal sensitive period, SP, in this species), but it has a negligible effect on the number of long-night cycles needed to effect the switch to diapausing progeny production (the “required day number”, RDN), which is therefore markedly temperature compensated. Data from Saunders (1966). (B) Comparable data for Sarcophaga argyrostoma. Larval cultures were maintained under diapause inducing long nights (LD 10 : 14) at six temperatures (26, 24, 22, 20, 18 and 16°C). Pupae were collected daily as they formed and incubated at 20°C in the dark; they were then later assessed for diapause or nondiapause development. Diapause incidence increased with an increasing number of long nights experienced during the larval sensitive period (SP). The number of long nights required to raise diapause incidence to 50% (the required day number, RDN) was, however, like that in N. vitripennis, temperature compensated. Data from Saunders (1971).

A similar interaction between a temperature-sensitive SP and a temperature-compensated RDN is seen for pupal diapause induction in the flesh fly S. argyrostoma (Saunders 1971), embryonic diapause in the mosquito Aëdes atropalpus (Coquillett) (Beach 1978), larval diapause in the blowfly C. vicina (Saunders 1987) and for ovarian diapause in Drosophila montana Stone, Griffen & Patter (Salminen & Hoikkala 2013). In the flesh fly, for example (Fig. 10B), the SP starts in the intra-uterine embryos and continues through larval development, ceasing just before puparium formation. In this experiment, larval cultures were raised under diapause inductive long-night cycles in a range of temperatures from 16 to 26°C; the pupae were then collected daily and later assessed for their diapause or nondiapause status (Saunders 1971). When diapause incidence was plotted against the number of long night cycles experienced, a “family” of curves was obtained showing that the RDN, the number of cycles inducing about 50% diapause, was about 12 to 13 with little effect of temperature. At moderate temperatures of 24, 22 and 20°C, the first larvae to pupate became nondiapause pupae, but those that pupated after the RDN was accumulated became dormant. However, at the highest temperature tested (26°C), all larvae pupated before this point and none of them entered diapause, but at the lower temperatures (18 and 16°C) nearly all did so.

Temperature effects on stored INDSUM and neurohormone release and retention

Larvae of S. argyrostoma raised under both short and long nights show a major peak of hemolymph ecdysteroids associated with pupariation and pupal development (Richard & Saunders 1987). About 72 h after pupariation, short-night (nondiapause developing) pupae then show a sustained high production of ecdysteroids driving adult differentiation, but those reared under long nights (diapause destined individuals) show a long period of extremely low ecdysteroid titers as they enter the diapause state. It may be assumed that S. argyrostoma pupae initiate diapause sometime during this 72-h period, after pupal differentiation but before the cerebral neurosecretory cells are due to release prothoracicotropic hormone (PTTH).

Working with S. argyrostoma reared under diapause-inductive long nights, Gibbs (1975) showed that a single temperature step-up after pupariation lead to a reduction of pupal diapause but a temperature step-down at this point increased it. The degree of these effects was found to be a function of the magnitude of the temperature step. This result suggested that a stepwise change in temperature after pupariation altered the subsequent incidence of diapause by affecting the “diapause titer” or INDSUM accumulated during the sensitive period, although temperature effects on events further downstream can not at present be excluded.


Pervasive effects of temperature on the photoperiodic mechanism and the nature of the photoperiodic clock

In northern temperate species with an over-wintering diapause, temperature affects all levels of the photoperiodic response from photoreception to hormonal events triggering the diapause syndrome. Temperature signals include, inter alia, constant temperature, temperature cycles or thermoperiods, low and high temperature pulses, and temperature steps, up or down, all of them in conjunction with a daily photoperiod or in continuous darkness or continuous light. The effects of these temperature treatments on the photoperiodic induction of diapause have been considered in this review. Of particular importance, because they contribute to our understanding of the mechanism of the photoperiodic response itself, are daily thermoperiods in darkness and temperature pulses in conjunction with a light cycle.

Exposure of N. vitripennis and S. argyrostoma to daily thermoperiods in complete darkness (Saunders 1973b, 1984) suggested that light was essential for the photoperiodic response in Sarcophaga but not in Nasonia (Fig. 6) in accordance with the external coincidence model for the former and internal coincidence for the latter (Pittendrigh 1972). This conclusion was underlined in experiments using temperature pulses systematically interrupting the night of diapause inductive light cycles (Figs 7, 8), which again suggested external and internal coincidence for Sarcophaga and Nasonia, respectively. The external coincidence model for S. argyrostoma (Fig. 3) has been tested on a number of occasions (Saunders 2011) and its derivative, the damped circadian oscillator model of the clock-counter mechanism (Lewis & Saunders 1987; Saunders & Lewis 1987a,b) successfully accounts for the major features of photoperiodic time measurement in this species.

The internal coincidence model has been less well formulated than external coincidence and less completely characterized experimentally. Its origins lie in the numerous observations that the insect circadian system is of a multi-oscillator construction (Saunders 2002) with constituent oscillators varying both in period (τ) and their sensitivity to light, temperature and other entraining agents. In N. vitripennis, available evidence from Nanda–Hamner experiments, thermoperiod and low and high temperature pulses strongly indicate internal coincidence with separate, but interacting, “dawn” and “dusk” oscillators. Further elucidation of the photoperiodic mechanism in this species, however, must include more fundamental molecular and physiological investigations. These approaches are dealt with in the next section.

Evolution of insect photoperiodic clock mechanisms

Experiments involving daily thermoperiods in DD (Saunders 1973b, 1984) and low and high temperature pulses in light–dark cycles (Saunders 1984, 2002; this review) have indicated that the flesh fly S. argyrostoma and its parasitic wasp N. vitripennis measure seasonal changes in photoperiod with substantially different mechanisms: external coincidence in the former, but internal coincidence in the latter. In S. argyrostoma the external coincidence model has been extensively tested; the photoinducible phase (φi) has been located in the latter half of the night (at CT 21.5) by night interruptions and the so-called Lees experiment (see Lees 1973; Saunders 1979, 2012). The photoinducible phase has also been isolated from the effects of the “main” photophase in T experiments (Saunders 1979). On the other hand, in N. vitripennis, photoperiodic regulation of the diapause/nondiapause switch in development has been shown to operate in the absence of light, and temperature- and light-pulse experiments have indicated that: (i) more than one oscillation is involved; and (ii) a specific photoinducible phase is not necessary.

Further differences between external and internal coincidence, as shown in Sarcophaga spp. and N. vitripennis, have been revealed in spectral sensitivity and genomic studies. In spectral sensitivity studies using the flesh fly S. similis, Goto and Numata (2009) showed maximum sensitivity to UV and blue light (395–470 nm) when light fell early in the night, but a broader peak from 395 nm up through the blue and green (583 nm) into the red end of the spectrum for light falling in the second half of the night. This indicated that two photoreceptors were involved, one absorbing maximally in the blue and used for entrainment and the other, late in the night, absorbing at longer wavelengths and probably associated with diapause regulation at the photoinducible phase. In contrast, spectral sensitivity studies with N. vitripennis (Saunders 1975) showed that light falling both early and late in the night engendered identical action spectra with a broad peak at a maximum at 554–586 nm extending into much longer wavelengths up to 617 nm or more. This suggested that, although several photoreceptive molecules were involved, light had only a single role: that of entrainment. These data therefore supported the conclusion that Sarcophaga spp. and N. vitripennis measure photoperiod by substantially different mechanisms, seen here as external and internal coincidence, respectively.

The genome of a Sarcophaga sp. has yet to be sequenced but it may resemble that of Drosophila melanogaster Meigen, which includes timeless1 (tim1) and cryptochrome1 (cry1), but lacks cryptochrome2 (cry2) (Zhan et al. 2011). In Drosophila, the protein CRY1 is a dedicated blue-light photoreceptor responsible for entrainment (Emery et al. 1998; Stanewsky et al. 1998). In N. vitripennis, however, the genome lacks cry1 (and tim1) but contains the “mammalian-type” cry2 that encodes CRY2 acting not as a photoreceptor but as a transcriptional repressor in the auto-regulatory feedback loop itself (Schurko et al. 2010). In a number of other insects (Lepidoptera, mosquitoes and an aphid, for example) cry1, cry2 and tim1 are all represented (Zhan et al. 2011). Following earlier authors (Yuan et al. 2007; Sandrelli et al. 2008; Merlin & Reppert 2010), the form of the photoperiodic clock in these last species may be regarded as “ancestral”, whereas those in Sarcophaga (probably lacking cry2) and Nasonia (lacking tim1 and cry1) are probably “derived”.

Photoperiodically regulated diapause may occur at any stage of metamorphosis (embryo, larva, pupa or adult) but usually at a species-specific stage of development (Tauber et al. 1986). On the whole, there appears to be little phylogenetic relationship concerning the stage at which diapause may occur. Among the Culicidae, for example, different species may enter diapause as embryos, larvae or adults. These observations suggest that photoperiodic regulation and the type of diapause have evolved on numerous occasions as insects extended their distributions into higher latitudes or into areas with climates inimical to development and reproduction. However, many Sarcophaga spp. over-winter in a pupal diapause, most northern Drosophila spp. undergo an adult reproductive diapause and many Aëdes mosquitoes become dormant as embryos, suggesting some phylogenetic relationship in photoperiodic responses occurring between closely related species.

Given the increasing evidence that photoperiodic time measurement (even in some of those species presenting hourglass-like properties) is a function of the circadian system (Saunders 2002, 2010; Goto 2013) it is probable that insects have repeatedly “adopted” the pre-existing circadian system as the “clockwork” for photoperiodic time measurement. In larvae of Sarcophaga spp. (external coincidence) for example, the blue-light photoreceptor responsible for entrainment may be CRY1 and the longer wavelength photoreceptor(s) functioning at the photoinducible phase might be opsin-based (and probably brain-centered). In adults of N. vitripennis (internal coincidence), on the other hand (where CRY1 is absent) the photoperiodic photoreceptors in both the early and late subjective night might be opsin based and perhaps located in the compound eye. Recent data for the cricket Modicogryllus siamensis (Tamaki et al. 2013) also locate opsin-based photoreceptors in the compound eyes and gene silencing techniques (RNAi) for the genes encoding UV-, blue- and long-wavelength opsins result in a partial (although substantial) alteration of the crickets’ short-night responses (rapid nymphal development) to long-night responses (protracted nymphal development). Whether opsin-based photoreceptors are operating at the photoinducible phase in this species or merely for oscillator entrainment, however, is not clear.

The photoperiodic mechanisms in N. vitripennis and S. argyrostoma are clearly quite distinct and probably also derived from a more ancestral type of photoperiodic mechanism containing cry1 and cry2 as in various Lepidoptera, Homoptera, Heteroptera and Diptera–Nematocera, cited above. If that is the case these “ancestral” photoperiodic clocks might show characteristics of both external and internal coincidence. The latter, for instance, is indicated by the ability of several species of Lepidoptera to show diapause regulation by daily thermoperiod in the complete absence of light (Chippendale et al. 1976; Dumortier & Brunnarius 1977). Future progress in determining the nature of the photoperiodic mechanism in insects should involve a broad comparative approach including genomic and molecular studies on the one hand and “formal” and physiological studies on the other.