Periodic heartbeat reversals cause cardiogenic inspiration and expiration with coupled spiracle leakage in resting blowflies, Calliphora vicina

Insects breathe by a branching air tube system called the tracheae, formed by lateral segmental invaginations of the exoskeleton. The blind tracheolar ends invade the tissues, supplying them directly with oxygen. In many flight-adapted adult insects, the tracheae are partly enlarged to form air sacs, which are suspended between the exoskeleton and the organs and occupy a vast part of the body cavity. The inflow and outflow of respiratory air is regulated by valves behind the filter-protected atrium of the lateral openings, called the spiracles. The gas exchange in many resting insects is characterized by cyclic CO₂ release. These widely distributed cyclic mechanisms have been studied for many years (Schneiderman, 1960; Lighton, 1996; Marais et al., 2005; Chown et al., 2006). The majority of recent investigations focussed on the impact of physical ecological factors, such as temperature, humidity and gas concentrations, on the diverse gas exchange patterns to evaluate their functional significance and ecological or evolutionary advantages.

The aim of this study was to analyse the assumed functional interplay of the periodic heartbeat reversals with the respiratory cycles and the involved structures such as the heart and spiracular valves. Their action in quiescent Diptera has scarcely been visualized or recorded, because direct observation of most valves is difficult. They are hidden behind dense filter structures composed of ramified trichomes. Therefore the influence of the valves on gas exchange was deduced from the CO₂ output measured by flow-through experiments (Kestler, 1985; Lighton, 1988; Hetz et al., 1994), partly with cannulation to specified spiracles or compartments of the body (Wasserthal, 2001; Duncan and Byrne, 2002), by thermographic-anemometric (Wasserthal, 1981; Slama, 1988) and volumetric-manometric measurements of expired or consumed air (Jõgar et al., 2007; Karise et al., 2010; Jõgar et al., 2011). A more immediate influence of the spiracles on the gas exchange was revealed by intra-tracheal pressure recordings (Brockway and Schneiderman, 1967; Kestler, 1985; Hetz et al., 1994) and oxygen uptake (Hetz et al., 1994; Wobschall and Hetz, 2004; Matthews and White, 2011). The gas exchange cycles are often characterised by a sequence of constriction, fluttering and opening of the spiracles (CFO) without or with coordinated bouts of ventilation movements (CFV) (Kestler, 1985), and are called discontinuous or in the absence of constriction cyclic gas exchange (DGC or CGE) (Lighton, 1996). It has been shown in Calliphora and Drosophila that periodic heartbeat reversal causes changes in haemocoelic and tracheal pressure and volume, alternating in the anterior body and in the abdomen (Wasserthal et al., 2006; Wasserthal, 2007; Wasserthal, 2012) because the tracheal systems and haemocoels of both compartments are functionally separate (Faucheux, 1973; Wasserthal, 1999). Therefore, it was hypothesized that the haemolymph shift ventilates the tracheal system by alternating between the anterior and posterior body. Under resting conditions, the tracheal pressure differs from the atmosphere because the spiracles impede immediate gas exchange and pressure equalization with the atmosphere by the more-or-less closed valves.

The action of the thoracic spiracular valves was analysed by pressure measurements directly at the intact spiracles. It was of interest to see whether the valves open and close synchronously or in a different manner. Simultaneous measurements at two spiracles was necessary to determine whether observation of one spiracle is representative for the other spiracles of the thorax. In order to avoid

a blockage of the gas exchange by the tight sensor attachment directly at the spiracles, most pressure measurements and all oxygen measurements were performed by intubating one of the dorsal air sacs, which proved to be a mildly invasive procedure. For analysis of the influence of the spiracular valves on the uptake of oxygen and the release of the CO₂, simultaneous video recordings and reflectance measurements of the thoracic spiracular valves were made. The influence of valve action on the emission of CO₂ and water loss was recorded by flow-through measurements concurrently with the air sac pressure registration. This study provides the first documentation of how tracheal pressure cycles, resulting from periodic heartbeat reversals, contribute to respiratory gas exchange.

Callipora vicina has two pairs of thoracic spiracles, the mesothoracic Sp1 and the metathoracic Sp2. Both spiracles can be closed by valves, which consist of opposing anterior and posterior membranous lips (Fig. 1A–C). The lips are stretched and kept in tension by the underlying haemolymph. The free edges bordering the aperture are stabilized by sclerotized elastic bars, which are ventrally connected to the V-shaped muscle. In Sp2, the dorsal part of the valves is soft and flexible (Fig. 1C). The seven pairs of abdominal spiracles are very small and have a simple circular opening with loosely arranged bristles in the atrium (Fig. 1A). Their inner valves were not visible under our experimental conditions and were not analysed further. The surface of the thoracic spiracles is equipped with peritrema filter plates of ramified bristles, which protect the atrium and delicate valves behind. The filter bristles of Sp1 form a dense stable roof (Fig. 2A). Sp2 has an anterior fixed plate (Fig. 2G) and a posterior pin-jointed plate, which can be opened passively by a strong expiratory air stream (Fig. 2H). The filter plates were removed for observation of the valves. The thoracic valves are said to be closed by a muscle and to be opened by the elasticity of the sclerotized rims of the valve lips (Krancher, 1881; Hassan, 1944; Faucheux, 1973). However, our own observations and recordings suggest that these valve muscles are openers (L.T.W. and A. S. Fröhlich, unpublished).

During and immediately after CO₂ anaesthesia, the valves of Sp1 and Sp2 were widely open (Fig. 2F,N). Visual observations suggested that after recovery in quiescent blowflies, the valves were mostly closed (Fig. 2B–D,I–K). The valves opened during and especially for a while after activity such as locomotion, grooming and feeding. When ceasing activity, the constricted valves seemed to be closed. In the hours after feeding, quiescent flies opened the valves at intervals (mean 21.6±9.8 min; N=9, n=8 per fly) without visible exercise. The valves of Sp1 and Sp2 on both sides opened and closed generally synchronously, showing the same tracheal pressure in front of the spiracular openings (Fig. 3A,B). However, the degree of opening and correspondingly the pressure amplitude at the tubed spiracle could differ (Fig. 3C). The opening slit of the valves varied continuously between 0 and 8% of the maximal possible open area. With widely opened valves (above 40%, Fig. 2F,N), the tracheal pressure was equilibrated with the atmosphere, and no pressure pulses could be recorded (Fig. 4). This happened especially after intensified activity.

The measurement of the tracheal pressure in front of the spiracles is problematic because the closing of the valves interrupts the connection of the sensor with the tracheal lumen. Moreover, the partial opening of the valves can result in local pressure differences at the individual spiracle (Fig. 3C). The sensor blocks the connection of the spiracle opening with the ambient air and may thus artificially affect the pressure of the entire thoracic system. To circumvent the possible disturbance of the spiracles, measurements of tracheal pressure were performed using intubations of the dorsal air sacs to obtain integrated values of the pressure and partial pressure of oxygen (P_O2) from the thoracic or abdominal tracheal system. Basically, the pressure curves show the same pattern whether recorded at the spiracles or at the dorsal air sacs.

In addition to the active movements of the spiracular valves, a minute leak at the dorsal soft part of the rims of the valve lips of Sp2 could be detected (Fig. 5). This leak is difficult to see and was hitherto overlooked. During volume and pressure decrease in the thoracic haemocoel by backward heartbeat, the leak was passively widened (Fig. 5C–E). This explains why ambient air could be sucked into the thoracic air sacs and the P_O2 increased even though the proper valve remained constricted (Fig. 5). The minimal extension of the leak during forward pulse period could be near 0% and expand to only 1% during backward beating (N=8; Fig. 5A,B). The widening of the leak can start from ~1% and extend to 5% (N=5; Fig. 5B,C) or from ~8.5 to 13.7% (N=3; Fig. 5D,E, supplementary material Movie 1). The regular periodicity of the leakage was also measured by the changes of reflected light from the valve opening and the resulting exposure of the dark inner tracheal background (N=3; Fig. 6). In the video recordings, it was seen that the valve lips vibrate in the frequency of the backward pulses, while they continuously increased the leak. The haemolymph on the rear of the valve lips is sucked in by the conical heart chamber connected by the lateral venous channels with the meta-thoracic spiracular region, as in Drosophila (Wasserthal, 2007) (L.T.W., unpublished data).

Oxygen fluctuations in the scutellar air sacs were measured in intubated flies with simultaneous registration of intra-tracheal pressure or heartbeat. As postulated, the compensatory pressure decline in the air sacs of the anterior body by the negative pulses during backward beating was correlated with an oxygen increase in the anterior body (Figs 7, 8). At an ambient temperature (T_a) of 21°C, the P_O2 ranged between a lower mean of 17.5±1.1 kPa and an upper mean of 18.9±1.1 kPa. (N=17, evaluated sequences: 1907; time: 20 h). At low metabolism in hibernating flies at a T_a between 3 and 19°C, intratracheal pressure cycles continued with a coincident O₂ increase during backward pulse periods leading to a mean P_O2 of between 16.9±4.05 and 17.9±3.2 kPa (N=7, n=523; Table 1, supplementary material Table S1). The single lowest P_O2 of 4.5 kPa was measured at an intermediate T_a of 10°C. No general reduction of P_O2 in hibernating flies at low ambient temperature was recorded (supplementary material Table S1). The O₂ rise (ΔP_O2 per peak) during each negative pulse period ranged between 0.1 and 2.5 kPa with a mean of 0.5±0.2 kPa at 21°C. During sustained quiescence with Sp2 leaking, the P_O2 remained constant at a high level (18.57±1.09, N=19), and no decrease could be seen towards the end of a several minutes-long period of resting heartbeat cycles. The mean P_O2 was also high when the O₂ peaks were weak and disappeared in the noise (mean 18.4±1.3 kPa, N=14, 20–30 sequences per fly). In a few cases, when the backward pulses were exceptionally omitted and pressure pulses remained positive for 4.8±0.28 min (N=3; Fig. 8B), the mean P_O2 decreased by about 3±0.5 kPa in the thorax. After the reappearance of the rhythm with backward pulse periods and closed but leaking spiracles, the P_O2 returned to the original, higher values (Fig. 8B at 18:20 h).

Measurements in the abdominal air sacs showed the reciprocal correlation of O₂ increase during forward pulse periods with a rise in ΔP_O2 of 0.96±0.45 Pa (n=46 sequences), which produced a mean P_O2 level of 18.4±0.7 (N=3 females; evaluated sequences: 328 and time: 3.3 h; Fig. 9).

The CO₂ output of the entire insect was measured by flow-through respirometry combined with recording of the intra-tracheal pressure of the scutellar air sac (N=14, evaluated sequences: 2287 and time: 13 h). The CO₂ micro-bursts ranged between a minimum of 4.4 nmol s⁻¹ g⁻¹ and a maximum of 28.8 nmol s⁻¹ g⁻¹ (Table 2, supplementary material Table S2). The pressure increase in the scutellar air sacs corresponding to periodic haemolymph accumulation in the thorax was in most cases correlated with a CO₂ micro-burst (Fig. 10A). The mean burst amplitude was 6.4±1.5 nmol s⁻¹ g⁻¹. Although the O₂ rise in the anterior body occurred during backward pulse periods and in the abdomen during forward pulse periods, the moment of maximal CO₂ emission could change for some time within measurements in the same fly, possibly depending on the phases of digestion of crop contents (Fig. 10B). The coincidence of the CO₂ maxima with one of the pulse directions was not as reliable as the O₂ increase in the thorax and abdomen. As the CO₂ recordings by flow-through respirometry comprise the CO₂ emission of the entire body, it can only be deduced from the pressure conditions whether the emission came from the anterior or from the posterior body or from both parts. In three of 14 flies, distinct CO₂ bursts occurred during forward pulse periods and backward pulse periods (Fig. 10C, supplementary material Table S2). In these cases, the one burst during forward pulse periods was attributed to expiration of the anterior body, and the other one during backward pulse periods to expiration of the abdomen. In some sequences, both bursts were equally strong and fused. Although being generally cyclical, the CO₂ emission never decreased to 0.

In addition to the cardiogenic CO₂ micro-bursts in the leaky phase, during active opening of all thoracic spiracles, the residual CO₂ was released as a macro-burst with a high mean amplitude of 273.5±151.4 nmol s⁻¹ g⁻¹ (N=9, 7–8 per fly at 22°C; Fig. 10D). The inter-burst phase between the CO₂ macro-bursts lasted 21.55±9.82 min. The amplitude of the macro-bursts was up to 683.8 nmol s⁻¹ g⁻¹ in well-fed quiescent individuals. The resting periods were often interrupted by phases of running on the Styrofoam ball, grooming or regurgitating and re-imbibing the crop contents, accompanied by irregular CO₂ release omitting the cyclical macro-bursts. All flies exhibited long phases of intermittent activity without the cyclical CO₂ macro-bursts. As an unexpected result, the P_O2 in the scutellar air sac dropped by ΔP_O2 of 2.75±1.16 kPa (N=11) when the spiracles fully opened (Figs 11, 12). This is the contrary to what one would have expected (see Discussion).

The simultaneous intratracheal P_O2 measurements and pressure measurements confirm the hypothesis that the cardiogenic negative pressure periods cause active inspiration. Because the negative intra-tracheal pressure arises in the anterior body during backward (retrograde) periods and in the abdomen during forward (anterograde) pulse periods (Wasserthal, 2012), it is evident that the heartbeat reversal is the causal force of this mechanism. The periodic P_O2 increase cannot be explained by a mere physical effect due to pressure changes. Under negative pressure in a closed system, the P_O2 should decrease and not increase. Moreover, the P_O2 reduction during pressure decrease is so small that it does not significantly counteract the observed P_O2 rise. The determined pressure-dependent physical P_O2 changes between 0.02 and 2 Pa have an effect of only 0.004 and 0.4%, respectively, with regard to the higher mean ΔP_O2 rise of 0.5±0.2 kPa.

It is a remarkable result that the P_O2 peaks occur and the relatively high mean P_O2 (17.5–18.9 kPa) remains constant, although the thoracic spiracles are constricted in the neuromuscular sense. However, a critical inspection of the spiracular valves revealed a very small leak of Sp2 widening in the course of the backward pulse periods of the heart. It is this leakage, increased by the negative pressure, that allows the respiratory inflow and rise of the tracheal P_O2. At the first spiracle, no comparable leak could be detected. The possibility cannot, however, be fully excluded that the ventral slit of the Sp1 valve is slightly opened below 1%. The periodic P_O2 rise during the negative pressure periods (forward pulses) in the abdomen is assumed to function in a similar way to spiracle leakage.

Passive suction ventilation (PSV) is the result of the discontinuous gas exchange cycle (DGC) based on the constricted phase, flutter phase and open phase (CFO) of the spiracles described in lepidopteran pupae (Schneiderman, 1960; Levy and Schneiderman, 1958; Brockway and Schneiderman, 1967). During the C-phase, the P_O2 decreases to a low value of 3–7 kPa because of O₂ consumption, and the tracheal pressure becomes sub-atmospheric. When in the following F-phase the spiracles open briefly, the tracheal pressure increases to nearly atmospheric values and fresh air is sucked in, preventing the P_O2 from sinking below the above fairly constant low value. This gas exchange by simultaneous diffusion and inward convection of O₂ is based on a great difference in partial pressure and hydrostatic pressure (Miller, 1974). It retains CO₂ and H₂O while allowing maximal O₂ uptake in a N₂ equilibrium of outward diffusion and inward convection (Kestler, 1985).

In the cardiogenic gas exchange of C. vicina, air is also sucked in by the negative intratracheal pressure, but it is not supported by a high diffusive gradient of P_O2 between the tracheal lumen and the atmosphere as in the PSV of the pupae. The suction mechanism in C. vicina is based on the negative pressure in the haemocoel, which expands the tracheal lumen. The inspiration by tracheal volume increase enables the airflow to reach the terminal tracheoles of the blindly ending tracheae or air sacs. By contrast, in the classical PSV in lepidopteran pupae, the suction force arises by the higher molar oxygen uptake than molar CO₂ release due to buffering of metabolic CO₂ in the tissues and haemolymph and delayed transition into the tracheae in the O-phase (Levy and Schneiderman, 1958). The resulting negative intratracheal pressure in the C-phase is partly responded to by the reduction of the tracheal volume, putting the compliant tracheae under tension and shortening the abdomen. The opening of the spiracles in the F- and O-phases leads to a relaxation of the tracheae and abdomen resulting in volume and length increases, respectively, with the consequence of inspiration. Both suction mechanisms avoid stagnant air and the dependence on diffusion alone.

The cardiogenic form of active suction ventilation is performed by negative pressure periods alternating in the anterior body and abdomen and passively adapting spiracle leakage. This stereotypical rhythm fulfils a similar job as the flutter phases of the CFO-type in other insects, by enabling a longer lasting convective and diffusive O₂ uptake, whereas CO₂ and H₂O have to diffuse against the inflowing air. In the flutter phases, an active role of the spiracles in regulation of the inspiratory airflow is assumed, and the term ‘flutter’ describes the sudden neuromuscular movements of the spiracular valves (Miller, 1981). By contrast, the proper valve mechanism in the leaky phase in C. vicina is not fluttering as the valve remains constricted, and the Sp2 leak widens gradually and vibrates with heart pulses in the course of each backward pulse period of the heart and narrows in the course of the forward pulse period. A similar spiracular behaviour can be deduced in the abdomen from the O₂ peak during the negative pressure period in the abdominal air sac. A fluctuating leakage of the abdominal spiracles could not be detected here because of their microscopic dimension.

A difference between the CFO-type and the cardiogenic gas exchange is the convective release of CO₂ in that compartment, which, because of the positive pressure periods, receives an increasing haemolymph volume, which is compensated for by the decreasing tracheal volume, and leads to a local and temporal separation of the interburst O₂ uptake and CO₂ micro-burst in the anterior body (Fig. 10A) and in the abdomen (Fig. 10B) or alternating in both (Fig. 10C). Independent of the cardiogenic gas exchange in C. vicina, macro-bursts of CO₂ emissions occur in longer intervals. They are reminiscent of the cyclical gas exchange (flutter–burst-type) in other insects with a long inter-burst phase and a short macro-burst phase. The inter-burst phase corresponds to the leaky phase in C. vicina with a variable number of heartbeat sequences. A correlation between inter-burst duration and number of heartbeat sequences, corresponding to the cardiogenic micro-bursts, could not be found. With a sequence duration of 27.5–39 s (data from Tables 1, 2) between 32 and 46 heartbeat sequences could occur during one inter-burst phase with a mean duration of 21 min.

The P_O2 drop in the scutellar air sac during the phases of full spiracle opening differs from published results in other insects with cyclical constriction, flutter and burst (CFB) phases of the spiracles, such as in lepidopteran pupae, beetles or cockroaches: the tracheal O₂ rises when the CO₂ is released as a macro-burst in the open phase (Punt et al., 1957; Levy and Schneiderman, 1966; Lighton, 1988; Hetz et al., 1994; Matthews and White, 2011). O₂ uptake occurs simultaneously with CO₂ release and additional small amounts during the flutter phase. In C. vicina the O₂ uptake seems to be fully disconnected from the CO₂ macro-burst.

The probable reason for the P_O2 drop during opening of the spiracles is the accumulation of CO₂ in the tracheae during the transition from the dissolved phase in the tissues and haemolymph into the gaseous phase just before and during release through the spiracles. Intra-tracheal CO₂ measurements inside the flies are not available, but it is known that the haemolymph pH becomes less acidic during the open phase in butterfly pupae (Hetz and Wasserthal, 1993) and in cockroaches (Matthews and White, 2011). The alkalosis in the haemolymph is an indication that the relative share of CO₂ increases the tracheal P_CO2 and the total tracheal pressure, and leads to an enhanced outflow of O₂ into ambient air, which lowers the concentration of the O₂. Measurements of O₂ consumption in lepidopteran pupae showed a similar O₂ drop ascribed to CO₂ bursts (Jõgar et al., 2011). The clear separation of the CO₂ burst during the open phase and the O₂ uptake during the leaky phase suggests that the active spiracle opening serves for CO₂ release. This separation confirms results in Hyalophora pupae (Levy and Schneiderman, 1958; Buck, 1962) and in the grasshopper Taeniopoda eques. In both insects the spiracle opening is triggered by a certain threshold of P_CO2 (Levy and Schneiderman, 1966; Harrison et al., 1995).

The persistent, relatively high P_O2 of 17.5–18.9 kPa in the phases with constricted, but leaking spiracles in quiescent C. vicina is contrary to the hypothesis that the constriction phase prevents toxic damage by O₂ radicals under low oxygen demand, which has been argued in connection with lepidopteran pupae (Hetz and Bradley, 2005). In adult flies, the maintenance of a high P_O2 with leaking spiracles favours the classical hypothesis of water retention, which has been suggested for saturniid moth pupae (Buck, 1962; Miller, 1974), and for Blattodea (Kestler, 1985; Schimpf et al., 2009), in which, under normoxic conditions, the P_O2 lies above 15 kPa (Matthews and White, 2011). As in any active suction ventilation during inspiration over a short valve distance, water vapour cannot diffuse outwards against the inflowing air stream (Kestler, 1985). But in C. vicina, water can be lost during positive pressure periods, which cause several CO₂ micro-bursts during the interburst (Fig. 10A–C). The difference between the cardiogenic micro-bursts and the macro-bursts is a convective gas exchange versus a diffusive gas exchange of the macro-burst. As water molecules diffuse quicker than CO₂ molecules, a diffusive gas exchange of CO₂ should lead to 59.6% higher water loss than a convective gas exchange of the same amount (Kestler, 1983; Kestler, 1985). This water retention can be deduced from the data in Fig. 12. A first comparison of the ratio of H₂O evaporation per 1 nmol of CO₂ emission during the interburst of 12 cardiogenic micro-bursts, which amounts to 0.33 nmol g⁻¹ with that of one macro-burst, which amounts to 0.79 nmol g⁻¹ H₂O per 1 nmol g⁻¹ of CO₂, results in a 58.2% higher diffusive water loss in the macro-burst. This preliminary evaluation confirms the above hypothesis that the cardiogenic mechanism is advantageous to withhold water as the number and duration of diffusive macro-bursts is reduced in favour of prolonged interbursts with convective CO₂ micro-bursts.

There are few known examples of cyclic gas exchange in other flies, and no comparison exists with the periodic heartbeat reversals. In Drosophila melanogaster, CO₂ bursts occur in populations selected for desiccation resistance with a burst cycle frequency of 1.2 per minute (Williams et al., 1997). The CO₂ release during the ‘closed’ phase was discussed with regard to a possible leakage of the spiracles. In mosquitoes a similar burst cycle frequency of 1.1 per minute was recorded at a flow rate of 1000 ml min⁻¹ at 20°C (Gray and Bradley, 2006). In tsetse flies, CO₂ macro-bursts have been recorded with a cycle frequency of 0.054–0.080 Hz, which is equal to 3.2–4.8 cycles min⁻¹ (Terblanche and Chown, 2010). All these burst frequencies are high when compared with C. vicina. Fewer heartbeat sequences or only a single one might occur during one of these relatively short inter-burst phases. Drosophila with a heartbeat sequence length of 19–25 s (Dulcis and Levine, 2005; Wasserthal, 2007) and mosquitoes with a sequence length of ~20 s (Glenn et al., 2010) would perform two to three cardiogenic cycles per one inter-burst. In contrast to these fast frequent macro-burst cycles, Drosophila mimica releases CO₂ bursts with pronounced flutter phases of the classical CFO cycle alternating with long-lasting leaky phases of ~30 min (Lehmann and Schützner, 2010). It is probable that these leaky phases are accompanied by numerous heartbeat reversals and reflect the cardiogenic ventilatory mechanism.

Periodic heartbeat reversal is a widespread phenomenon in insects (Gerould, 1929; Jones, 1977; Wasserthal, 1996), and it is predicted that it contributes to gas exchange in all cases in which the haemolymph is periodically shifted between the anterior body and abdomen. In contrast to the flies described here, in adult giant silk moths, the CO₂ emission cycle is strongly coupled with the heartbeat sequence (Wasserthal, 1996) with periodic expansion and contraction of abdominal movements and superimposed bouts of peristaltic contractions at the end of each backward pulse period, which are coordinated with metachronic closing of the abdominal spiracles (Wasserthal, 1981).

Tracheal ventilation in quiescent Calliphora turned out to be mostly a side effect of periodic heartbeat reversal causing the changes in tracheal volume and appropriate spiracular valve leaks. Inspirations are shown to be only cardiogenic, due to backward beating through increasingly leaking valves in the thorax and during forward heartbeat in the abdomen. The expirations follow mostly during forward heart pulses. Single abdominal pumping strokes coincide with the onset of each forward pulse period of the heart during the interburst (Wasserthal, 2012). They have no measurable effect upon the gas exchange, whereas in other insects abdominal ventilatory movements are typically volleys during the macro-bursts. In C. vicina the residual CO₂, which is retained in the tissues and haemolymph during the long interburst, is released as a macro-burst during open phases. Thus, at rest, the respiratory gas exchange in the flies is water saving and energetically economic, because no special ventilation movements and spiracle muscle activity, as in the flutter and open phase of other insects, are required. This provides reserves for the high metabolic demands during flight, which in Drosophila is matched by continuous active opening of the spiracles (Lehmann, 2001). Whether in C. vicina during flight, periodic haemolymph shifts by heartbeat reversals continue and whether they contribute to the gas exchange will be addressed in a future study.

In Lepidoptera and scarabaeid beetles, the elastic tracheae enable a tidal haemolymph flow in the wings and elytra, respectively, functioning as counterforce to the periodic heartbeat reversals and intermittent and coordinated action of the accessory pulsatile organs (Wasserthal, 1982; Wasserthal, 1996). In flies the role of the accessory pulsatile organs in the cyclic haemolymph supply and the functional and structural adaptations of the tracheal system in the gas exchange still need to be analysed.

Blowflies (Calliphora vicina Robineau-Desvoidy 1830) from the field and their offspring were used for experiments. They were treated and fed as described in a previously published paper (Wasserthal, 2012). The flies were anesthetized with CO₂ gas only for fixation and surgical treatment, but not during measurements. The mass of a well-fed adult was in the range 53–120 mg (mean ± s.d., 84.5±16.4 mg). It diminished by 8–20 mg when fasting for 24 h at a temperature of 20–23°C and relative humidity of 60–82%.

In order to see whether the thoracic spiracles open or close simultaneously, the pressure at two spiracles (Sp1 and Sp2) in 12 flies was measured, either on the same side or on opposite sides of the same segment. Plastic tubes were glued with Pattex (Henkel, Düsseldorf, Germany) and sealed in front of the peritrema with Fixogum rubber cement (Marabu, Tamm, Germany), leaving the atrium and valves intact. Thereby these prae-spiracular measurements recorded the atrial pressure conditions, which were expected to be identical with the intratracheal pressure when the valves were open.

For measurement of the tracheal pressure inside the body, the dorsal cuticle with the underlying air sacs was perforated and connected with a bi-tubated plastic cone using Pattex, which allowed the fly to be handled and clamped in the apparatus while being simultaneously connected to the pressure sensor (Sensym SCXL 004 DN, Sensortechniques, Puchheim, Germany) and to the O₂ optode (Fig. 1). The positive or negative pressure pulses at the spiracle and in the scutellar and the abdominal air sacs reflect the activity and direction of the heart pulses and were used as the reference for the periodic heartbeat reversal and the resulting haemolymph shift between thorax and abdomen. This method of pressure measurements has been described in detail and compared with electrophysiological records in a previously published paper (Wasserthal, 2012). The dead space of the sensor of 25 μl and the 48 mm long tube connection to the spiracles or air sacs resulted in a 50% attenuation of the pressure signal. This was considered in the scaling of the curves. The response time was 7–10 ms and the time constant was 30 ms. In some flies the heartbeat was measured by extracellular electrical resistance myographs, as described previously (Wasserthal, 2012).

In addition to the dorsal pressure sensors, the flies were equipped with a fibre-optic optode (Microx TX3 AOT, PreSens, 93053 Regensburg, Germany). The tapered tip (diameter 50 μm) of this fibre was oriented directly above the perforation or inside the scutellar or abdominal air sacs, arranged beside the air pressure tube in the bi-tubated adapter cone (Fig. 1D). The measurements were run under controlled temperature, between 20 and 23°C, and with hibernating flies in an outside Faraday cage at ambient temperature between 2 and 19°C (Table 1, supplementary material Table S1). The sampling rate of the optode was 1 Hz. Response time was 40 ms and the time constant (interval from 17.4 to 20 kPa) in the experimental setup was 1.5 s. Calibrations in the O₂-free and ambient atmosphere were repeated before and after each experiment. The stability of the optodes allowed continuous use over several weeks without significant reduction in sensitivity and only slow, gradual loss in response time.

The influence of pressure changes on the P_O2 was checked by simulation experiments. In the first series, a microlitre syringe instead of the fly was combined with the pressure sensor and the optode. Doubling the syringe volume (45 μl + dead space volume of 10 μl) by 50 μl at T_a 22°C in the closed system led to a P_O2 decrease from 19.8 to 9 kPa, a value that would also be expected theoretically. This shows that the measured increase in P_O2 in the fly experiments during pressure decrease can only be explained by an inflow of ambient air into the open tracheal system.

In a second simulation series, the influence of pressure changes on the O₂ optode signal was tested in a closed 20 ml chamber. In the pressure range of 0.01–10 Pa, the corresponding physical P_O2 values were between 0.02 and 2 Pa, respectively. This was considered in the rise and drops of the O₂ measurements with the flies (see Discussion).

For visualization of the valve action, the plates of ramified bristles of the peritrema, which hide the valve lips, were removed with surgical micro-scissors, paying special attention not to injure the membranous attachments of the valve lips to avoid bleeding. The movements of the thoracic spiracular valves were observed with a Macroscope (Leica M420: 35–70-fold magnification, Leica AG, CH-9435 Heerbruck) and recorded using the video and single-frame mode of a Eos D60 SLR camera (Canon, Otha-ku, Tokyo, Japan) or reflex measurements using a Nikon F2 camera (Nikon, Chiyoda-ku, Tokyo, Japan) with an integrated Si-photocell of 2.8×3.1 mm on the interchangeable SLR screen (Fig. 1B,C) with an external connection to the amplifying DC interface. The spiracles were illuminated with a light ring of white light-emitting diodes arranged around the front lens of the Macroscope. The surfaces of the valve lips are white with brownish, sclerotized areas. They have a higher reflectance than the shadowy tracheal lumen, which becomes exposed if the valves are open. On the basis of the video frames and photographs, the area of the spiracular valve opening was traced and calculated as the percentage of the maximal possible open area, using custom-made software.

The CO₂ measurements were performed in a specimen chamber with controlled constant airflow (1000 ml min⁻¹) and adjustable pressure at controlled temperatures between 20 and 30°C. The volume of the specimen chamber was 20 ml and as small as possible (Fig. 1E) for recording at short response time for gas flow in the chamber. Response time was 1.4±0.2 s. The fly was glued at the mesoscutellum to a cannula, which at the same time served to fix the fly in position to the upper gum plug and connect the perforated air sac to the pressure sensor. A lateral port in the chamber allowed feeding and manipulation of a Styrofoam ball provided for foot contact in order to quieten the fly. The airflow stimulated prolonged running and grooming activities. Orientation of the head against the air stream appeased the flies noticeably. The chamber was connected directly to a CO₂/H₂O infrared gas analyser (LI-7000, LI-COR, Lincoln, NE, USA). Before entering the specimen chamber, the air passed a CO₂- and water-absorbing scrubber containing pellets of NaOH and the reference chamber. The pressure of the chamber was adjusted to a value between 10 and 50 Pa above ambient. The baseline of the CO₂/H₂O analyser and system was checked for drift after each experiment without a fly. The CO₂ output was calibrated between experimental runs in the absence of flies using a 50 ml syringe metering pump (Glenco, Houston, TX, USA), simulating the release of a constant volume of pure CO₂ gas in steps at different flow rates (0.1–10 μl s⁻¹).

Data were continuously recorded on an Apple Powermac or Powerbook using a custom-made amplifier and a Powerlab 8-channel AD-Interface with software (Chart 5.54: CB Sciences, Milford, MA, USA). The sampling rate was 200 Hz.

I thank Thomas Messingschlager and Alfred Schmiedl for technical assistance and Prof. Manfred Frasch, Department of Developmental Biology, for laboratory use. I am also indebted to the reviewers for constructive comments and discussion.