Flight-motor-driven respiratory airflow increases tracheal oxygen to nearly atmospheric level in blowflies (Calliphora vicina)

Flies are among the best fliers in the animal kingdom. They have indirect muscle fibres attached to the inner surface of the mesothoracic exoskeleton, which, by deformation, move the wings up and down. The high wingbeat frequency is maintained by stretch activation, based on the vibrational resonance of the thorax without synchronous nervous stimuli (Pringle, 1974). Among flies, Calliphoridae are powerful fliers. Vigorous flies can easily be obtained from the field and raised. Their size allows for micro-sensor access to the inner organs, such as the heart, accessory pulsatile organs and tracheal system, in contrast to the preferred fly model organism, Drosophila. In Calliphoridae, the structural basis and mechanics of the flight apparatus, its control and energetics have been investigated in detail (Miyan and Ewing, 1985; Nachtigall, 1985; Ennos, 1987; Wisser, 1988; Nalbach, 1989; Dickinson and Tu, 1997; Walker et al., 2014). There is a disproportion of thorough research on the flight mechanism in flies in contrast to the aspect of gas exchange during flight.

Flight muscles consume more oxygen than do any other type of tissue. This applies especially to small insects, which expend relatively more energy compared with larger ones (Schmidt-Nielsen, 1997). An increase of blowfly metabolism during flight occurs by a factor of 100 (Davis and Fraenkel, 1940) and can be maintained for more than 30 min. This surpasses values measured in Drosophila, with an increase of CO₂ output that is 7.4 times higher during flight than at rest (Dickinson and Lighton, 1995) or of metabolic rate with respect to muscle mass by a factor of 8.5 during flight (Lehmann and Dickinson, 1997). Although it is generally accepted that flight is the most energy-absorbing activity, the mechanism of respiratory gas exchange in flying flies is not understood. Only a few papers address the question of how the high oxygen requirement during flight is achieved. In Drosophila, diffusion alone and adjustment of the spiracle opening have been suggested to provide adequate supply, sporadically supported by a hydraulic proboscis reflex (Lehmann and Heymann, 2005). The possible role of autoventilation by the flight apparatus in flies has been discussed and dismissed (Weis-Fogh, 1964). This scepticism was substantiated by the small contraction of the flight muscles of 1–2%, as well as the corresponding slight deformation of the thorax and tracheae (Boettiger, 1960). However, it was hypothesised that in contrast to a diffusive gas exchange, a convective gas exchange would provide a selective advantage, especially in small and flying insects, by reducing the respiratory water loss (Kestler, 1983,, 1985).

Flies are sophisticated in most structural and functional regards, constituting the basis for their high flight performance. It was assumed that an ingenious mode of ventilatory gas exchange may be uncovered when using new and gentle analysing methods. Considering the indefatigable flight of blowflies, their oxygen supply should not run short during sustained flight. The present study tested the hypothesis that the small thoracic deformations by the asynchronous flight muscles in Calliphora vicina Robineau-Desvoidy 1830 also induce a ventilatory gas exchange, questioning the suggestion that diffusion alone provides sufficient gas exchange. A priority objective was the measurement of oxygen content in the tracheal air sacs using O₂-optodes concurrently with tracheal pressure and monitoring the wingbeat by photo-diodes. For analysing gas exchange, a technique similar to that successfully applied in the hawk moth Manduca sexta (Wasserthal, 2001) was adopted for this study to measure the tracheal pressure in front of the spiracles (pre-atrial) and in the dorsal air sacs, as well as the CO₂ release from specified spiracles using a split flow-through chamber. For comparison, the CO₂ and H₂O emissions of the entire fly were measured using flow-through respirometry with concurrent measurement of pressure or P_O2 in the scutellar air sacs. In addition, I analysed whether the periodic heartbeat reversals, which are basically involved in gas exchange at rest (Wasserthal, 1999,, 2007,, 2012,, 2014), are continued during flight and whether they contribute significantly to gas exchange.

The tethered flies showed different types of flight behaviour. The younger flies (1–4 weeks old) showed mainly intermittent volleys of several short flight phases, each lasting only few seconds or a fraction of a second with few wingbeats. Because of the high wingbeat frequency (range 132–159 Hz, mean 145±9.4 Hz, at 24°C, N=8), these short flights would be sufficient for flies to escape or change location. The volleys lasted up to 10 min. Long, continuous flight was performed more frequently by the older (1–6 months old) flies. The mean duration of the evaluated continuous flight was 117±79 s (N=7).

The percentage of the aperture surface in relation to the maximum spiracle opening was calculated with custom-made software. The maximum possible spiracle opening was 0.234 mm² in the mesothoracic spiracles (Sp1) and 0.38 mm² in the metathoracic spiracles (Sp2). However, these wide apertures were rarely displayed and only for short moments. The thoracic spiracles often opened widest shortly at the initiation of a flight phase or after flight stop, with the consequence that the maximum P_O2 was attained at the end of flight or after flight had stopped. A typical opening during flight start was 0.036 mm² in Sp1 (Fig. 1A) and 0.14 mm² in Sp2 (Fig. 1B), representing a 3.9 times smaller aperture in the Sp1. The negative intratracheal pressure during flight was maintained with only slightly opened spiracles.

Tracheal pressure has a crucial influence on respiratory gas exchange. It reveals whether the gas exchange is performed by diffusion alone or through the support of convection. The pre-atrial pressure was measured at Sp1 and at Sp2 to evaluate a possible pressure gradient between anterior and posterior thoracic spiracles and to see whether heartbeat reversals have some influence on gas exchange during flight, just as documented during rest (Wasserthal, 2014). The pre-atrial pressure at the Sp1 and Sp2 was measured in different combinations: at a single spiracle, simultaneously at Sp1 and ipsilateral Sp2, or simultaneously at ipsilateral and contralateral spiracles (N=6). At all thoracic spiracles, the downstroke of the wings was correlated with a pressure decrease, while the upstroke was correlated with a pressure increase. During flight, the relative mean pressure at Sp1 on both sides was sub-atmospheric (−19±12 Pa, N=3) (Fig. 1A, Fig. 2). At Sp2, the mean pressure was over-atmospheric (5±3 Pa, N=3, Fig. 1B, Fig. 2). Positive pressure at Sp2 arose only if the flight was regular and stable. Irregular flight resulted in variable pressure above or at atmospheric levels (Fig. 1B). Unilateral wingbeat resulted in a one-sided pressure effect. If the left wing beat alone, the pressure decreased only at the ipsilateral spiracle and attained almost the same values as when the two wings were beating together (Fig. 3). There seems to be a side-bound pressure effect in the thorax and little pressure equalisation between the left and right tracheal system in the thorax.

In the paired scutellar air sacs, side-specific pressure measurements were not analysed. The septum between neighbouring scutellar air sacs was always opened by the surgical intervention and the paired air sacs were connected in unity with the sensors. In the scutellar air sac, the mean tracheal pressure was always sub-atmospheric during flight: the relative air sac pressure decreased from the resting atmospheric values with open spiracles (=0 Pa) by −37.5±29.7 Pa (N=17, Fig. 4, Table 1). In this relative mean pressure decrease, the fluctuations of the positive and negative pressure periods in the tracheal system caused by heartbeat reversals, ranging between a mean of +8.8±13 and −8.1±17 Pa, are not relevant, as the spiracles during flight were more open than at rest, when most of the time they were only leaking (Wasserthal, 2014).

Resting flies maintained a high mean intratracheal P_O2 of 18.6±1.36 in the thorax, ventilated by cardiogenic haemolymph pressure changes (Wasserthal, 2014). It was unknown whether this oxygen concentration could be maintained or exceeded under high oxygen consumption during flight. P_O2 was measured in the scutellar or abdominal air sacs using combined or separate pressure and O₂ sensors (Wasserthal, 2014). During short and steady flight, P_O2 increased in the scutellar air sacs. Before or at the initiation of flight, P_O2 decreased transiently by 1–2 kPa below the resting level (t-test, P<0.005, N=12, Fig. 4, Table 1). It then rapidly increased to 1–2 kPa above resting level, becoming almost atmospheric, with a mean P_O2 of 19.5±1.27 kPa (t-test, P<0.02, N=17).

In the abdomen, intratracheal P_O2 revealed an opposite effect. In the large abdominal air sacs during flight, P_O2 decreased from 17±0.94 kPa to 14±1.58 kPa (t-test, P<0.002, N=6, Fig. 5). It is assumed that CO₂ from the thoracic haemolymph was transferred into the air sacs before being released via the anterior three pairs of abdominal spiracles, which are connected to the air sacs. A similar P_O2 drop in the thoracic air sacs was previously found to be correlated with transient intratracheal CO₂ accumulation during spiracle opening in resting C. vicina (Wasserthal, 2014).

The flight phases caused corresponding CO₂ and H₂O emissions. CO₂ emissions attained a mean value of 385±184 nmol s⁻¹ g⁻¹ and H₂O emissions attained a mean value of 216±110 nmol s⁻¹ g⁻¹ (Fig. 6A, Table 2). The flight phases were often initiated by an increase of CO₂ emissions without any visible activity (Fig. 6B, asterisk). These CO₂ peaks probably represent so-called macro-bursts, which are an effect of the spiracle opening during discontinuous gas exchange in resting insects (Wasserthal, 2014). The CO₂ emissions caused by intermittent flight phases replaced the CO₂ macro-bursts.

CO₂ and H₂O release in the 20 ml flow-through chamber experiments with a flow rate of 1000 ml min⁻¹ started with a response time of 1.4±0.2 s after onset of flight. While H₂O and CO₂ emissions increased during flight, tracheal P_O2 decreased at the beginning of flight (Fig. 7). In the course of the flight phases, tracheal P_O2 increased and attained its maximum at the end of the flight phase, sometimes even shortly after flight stopped. The preceding O₂ decrease is explained with a premature opening of the spiracles before flight onset. A periodic P_O2 drop in the scutellar air sac has already been described and interpreted as a consequence of the CO₂ transition from the haemolymph into the tracheal system before its release during the opening of the spiracles (Wasserthal, 2014).

The difference between the CO₂ emission levels during rest and tethered flight was not as high as one would have expected according to published data regarding the gain in metabolic rate of 148-fold from rest to full flight in hawk moths (Bartholomew and Casey, 1978). The CO₂ emissions during pauses between the flight phases were already rather high and the emissions during flight were only 3.3 times higher. After deep rest, the mean CO₂ emissions during flight were 33 times higher, or 81.4-fold when compared with the maximum values of CO₂ emissions at the beginning of the flight phases. The average values corresponded to a 15-fold increase of metabolic rate during flight in bumblebees (Ellington et al., 1990).

Water loss rose from 32.2±12 nmol s⁻¹ g⁻¹ at deep rest to 216±110 nmol s⁻¹ g⁻¹ during steady flight, which is a 6.7-fold increase. This rise is markedly below the 33-fold increase of the CO₂ output, from 11.7±9.4 nmol s⁻¹ g⁻¹ at rest to 385±184 nmol s⁻¹ g⁻¹ during flight (Table 2).

As the pre-atrial pressure showed a decrement between Sp1 and Sp2, it was of interest to see whether inspiration and expiration were different at these spiracles. To examine the emissions through selected spiracles, the CO₂ emissions were measured using tubed flies in a split flow-through chamber, which allowed pairs of Sp1 or Sp2 to be connected separately with an interposed empty chamber to the CO₂-measuring device (infrared gas analyser, IRGA) (Figs 8, 9, 11). When the insect compartment was directly connected to the IRGA and the Sp2 tubes were deviated via the empty chamber into the scrubber, the emissions of the Sp1 and the abdominal spiracles were analysed together (Fig. 8A). The resulting CO₂ output was 24±10 nmol s⁻¹ g⁻¹ and followed on each flight period with a delay of 2–3 s (N=4). During the crosscheck experiment, the Sp2 was connected to the IRGA via the empty compartment, and 366±100 nmol s⁻¹ g⁻¹ of CO₂ was emitted (N=4, Fig. 8B).

To see whether and how much CO₂ is released by Sp1 alone, the Sp1 tubes were connected to the IRGA via the empty chamber (Fig. 9A). In this arrangement, a possible effect of all other spiracles was excluded. No CO₂ was emitted through Sp1 (N=3). In the crosscheck, Sp2 and all abdominal spiracles released CO₂ into the insect chamber without Sp1. The CO₂ emissions of Sp2 and of all abdominal spiracles were directed to the IRGA (Fig. 9B). The CO₂ level attained a similar value of 342±98 nmol s⁻¹ g⁻¹ (N=3), similar to the measurements of Sp2 output alone (Fig. 8B). The small fraction of CO₂ emissions in Fig. 9A must come from the abdominal spiracles because the contribution of Sp1 is zero (Fig. 8A). Because the small abdominal share of CO₂ emissions is not contained in the output of Sp2 alone (Fig. 8B), the bulk of CO₂ leaves from Sp2.

It was of interest to see whether the respiratory airflow was directed from Sp1 to Sp2 by an intratracheal valve mechanism, or whether it could be inverted by adjusting a pressure decrement between the compartments of the split chamber. While in the above experiments, the pressure in the two compartments of the split chamber was kept identical, in a second series, the pressure in the insect compartment was increased over that in the empty compartment with the discharging tubes of Sp1. A small pressure difference of +5 Pa in the insect compartment led to only a little, dampened increase of Sp1 emissions of below 24 nmol s⁻¹ g⁻¹ (N=3), as seen when comparing Fig. 9C with A. An increase of the pressure excess of +18 Pa in the insect compartment reversed the CO₂ emissions from the Sp2 to the anterior Sp1 (Fig. 9D). The amount (290–390 nmol s⁻¹ g⁻¹, N=3) of the reversed CO₂ emissions was comparable with the unchanged airflow, but in contrast to the normal time response of 1–3 s, the artificially inverted airflow was delayed by 5–6 s after the initiation of flight. This experiment supports the existence of a tracheal valve, which prevents a reverse airflow, when the counter-pressure is moderate (Fig. 9C). This postulated valve seems to be overcharged by higher counter-pressure (Fig. 9D).

Periodic heartbeat reversals have a significant influence on respiratory gas exchange in resting blowflies (Wasserthal, 2014). During flight, the thermistor measurements with heat-marking of the thoracic haemolymph showed the typical periodic temperature increase above the abdominal heart characteristic for the backward pulses of the heart. During flight, no parallel conspicuous pressure effects were detected at Sp2 (Fig. 10). Owing to the increased thorax temperature of 27–31°C during flight, the heartbeat sequences appeared in a shorter succession (7±2 cycles min⁻¹, N=7), in contrast to the resting flies (1.53±0.61 cycles min⁻¹ at 21°C, N=17; Wasserthal, 2014). The influence of the more frequent backward pulse periods supports a more efficient removal of the CO₂-charged haemolymph from the thorax into the abdomen, where it contacts the air sacs beside the Sp2 and the large abdominal air sacs. The rising CO₂ levels of the abdominal air sacs during flight also confirm this (Fig. 5).

Autoventilation in flying insects has been documented in larger insect species, such as locusts (Weis-Fogh, 1967; Miller, 1981). A unidirectional gas transport has been deduced from the segmental succession of spiracular openings. In large beetles, a passive inflow into the exposed anterior spiracles during incoming flow was called draught ventilation (Miller, 1981). In the hawk moth M. sexta, a unidirectional airflow with CO₂ emissions through the metathoracic spiracles represents a similar mechanism to that in the blowfly and lends itself to comparison. In the moth, the flight apparatus generates the suction force for the inspiration flow at the anterior spiracles as a result of inspiration prevention through the posterior thoracic spiracles (Wasserthal, 2001). During the downstroke, the volume of the thoracic air sacs increases, while the posterior thoracic spiracles are automatically enclosed in a pleural fold between the mesothorax and metathorax. During the upstroke, the air sac volume decreases and the moth expires through the exposed and open posterior spiracles. The mechanical coordination of spiracle opening and closing with the wingbeat cycle is possible because the wingbeat frequency of the moths is only 27±3 Hz and the longitudinal deformation of the thorax by the synchronous muscles is more enhanced.

In Calliphora, with a higher wingbeat frequency of 145±9.4 Hz performed by the asynchronous muscles, the spiracles act with a wingbeat-independent, less-frequent closing pattern. The metathorax with the posterior spiracles does not approach the mesothorax during downstroke and leaves the spiracle exposed. The pressure gradient between Sp1 and Sp2 of 20–34 Pa demonstrates that, during flight, a unidirectional airflow is produced, entering through the anterior spiracles (Sp1) and leaving through the posterior spiracles (Sp2).

How is the negative pressure at Sp1 and in the air sacs generated? The permanent sub-atmospheric pressure in the scutellar air sac during flight is based on the sub-atmospheric pressure in the haemocoel (Wasserthal, 2012). During flight, the increased consumption of O₂ might further contribute to the reduction of tracheal pressure as, because of the haemolymph buffering capacity, not all CO₂ is released immediately into the tracheal system to compensate for the loss of O₂ (Chown et al., 2006). In addition, the strict correlation of the pressure decrease with each downstroke of the wings suggests a mechanical effect, which is produced by the volume changes of the deforming thorax. The resulting pressure difference causes a unidirectional airflow by autoventilation. This conclusion is substantiated by the split-chamber flow-through experiments, which show the bulk of CO₂ release through the Sp2 and, to a smaller extent, from the abdominal spiracles (Figs 8, 9).

Diffusion alone in small insects was regarded as sufficient for respiratory gas exchange during flight (Weis-Fogh, 1964; Miller, 1974; Lehmann and Heymann, 2005). The possibility of gas exchange by ventilation due to thoracic distortions in small insects has been disclaimed because the deformations of the thorax by the indirect flight muscles are very small. The contraction of the indirect flight muscles in Sarcophaga blowflies is only 1–2% of its resting length (Boettiger, 1960) or 2–5% of the strain amplitude in Drosophila virilis (Chan and Dickinson, 1996). Moreover, in Drosophila, it was argued that in addition to the small thoracic distortion in these small tracheae, autoventilation and tracheal ventilation due to the Bernoulli effect are greatly attenuated by the low Reynolds number for tracheal airflow (Lehmann, 2001); Heymann and Lehmann (2006) state: ‘Although tethered flying Drosophila sporadically employ ventilation to manipulate tracheal gas flow, diffusion is still assumed to be the main type of respiration in fruit flies and diffusive theory may be applied’. These authors concluded that their ‘present results provide direct evidence for the general assumption in respiratory research that the tracheal development of a simple diffusion-based system matches the respiratory needs at maximum metabolic activity of the animal’ (Heymann and Lehmann, 2006).

However, it has been shown that distortion of the thorax in Calliphora results in upward and outward movement during the downstroke: ‘The dorsolongitudinal muscles provide increased scutal arching, localised at the scuto-scutellar border and outward splaying of the tergal fissure’ (Ennos, 1987). In video sequences of C. vicina, the scutal arching during downstroke is 60–96 µm of a 5 mm high thorax, confirming the small, approximately 1–2%, vertical distortion (L.T.W., unpublished).

In C. vicina fixed dorsally at the meso-scutum, the scutellum moves downwards during downstroke and the antero-ventral thorax prolongates by 3.3–4.7%. This distortion is sufficient to enlarge the thoracic volume during the downstroke, increasing the volume of the scutal air sac and the tracheae between the longitudinal and vertical muscle bundles (L.T.W., unpublished). When the air sacs expand, they produce a negative pressure pulse. The high-frequency summation of the pulses (145±9.4 Hz) leads to the mean sub-atmospheric pressure at the Sp1 and in the mesothoracic tracheal system. The suction at the Sp1 is the precondition for the uniform inflow of fresh air and the rise from a mean resting P_O2 of 18.6 to 19.5 kPa oxygen supply during flight in the scutellar air sac. This relatively high O₂ concentration corresponds to the high P_O2 in flying hawk moths (Komai, 1998) and bumblebees (Komai, 2001). However, the question remains as to why the mean tracheal pressure continues to be sub-atmospheric and no CO₂ is released through Sp1 during the upstroke. The existence of a back-pressure valve behind Sp1 is therefore postulated. The experiments with the artificially imposed higher pressure on the insect compartment of the split flow-through chamber with the intention to reverse the respiratory air stream from Sp2 to Sp1 are in favour of the occurrence of a valve that is, however, only effective as long as the counter-pressure is moderate and not higher than 5 Pa; it did not resist higher, probably un-physiological pressure.

A comparison of the outer morphology of Sp1 and Sp2 reflects the different functions (tasks) of these spiracles. Sp1 has a 3.9-fold narrower aperture than Sp2 when, for example, Sp1 is 15% open and Sp2 is 37% open (Fig. 1). Sp1 has a fixed roof of filter bristles, whereas the hinged posterior filter plate of Sp2, which is pressed passively outwards by a strong expiratory airflow, supports the expiratory function of Sp2. The negative intratracheal pressure in the mesothorax during flight is maintained by only a slight opening of the spiracular valves. It is suggested that a wide opening of the valves causes a pressure equalisation with the atmosphere, resulting in a breath. The convective oxygen gain would, however, diminish when the spiracles remain wide open for a longer time, preventing the re-establishment of a negative tracheal pressure.

Proboscis movements in Drosophila during flight have been reported to represent a sporadic type of tracheal ventilation (Lehmann and Heymann, 2005). The flight-associated proboscis movements may be a taxon-specific trait. In C. vicina, proboscis pulsations were not observed during flight. The ventilation effect by extension and retraction of the proboscis has, however, been recorded in resting C. vicina and the hoverfly Eristalis tenax (L.T.W., unpublished). Here, tracheal pressure pulses were correlated with the dabbing of water or honey solution. During extension, the haemocoel of the proboscis is filled with haemolymph and its volume is compensated for by the expanding cephalic air sacs, causing inhalation. At rest and only rarely during flight, the flies regurgitated and slowly resucked their crop contents without dabbing and pressure pulses.

Bouts of droplet extrusion were also performed by resting calliphorid flies under high ambient temperatures in the field (L.T.W., unpublished). Proboscis extension and fluid exposition were performed under high ambient temperature (above 28°C) and interpreted as a mechanism for evaporative cooling in the hawk moth Pholus achemon (Adams and Heath, 1964) and in honeybees (Heinrich, 1979,, 1980). Calliphora vicina often discharged fluid from the crop during and after preparation for the experiment. This is interpreted as a stress reaction, possibly relieving the body weight for escape flight. Before attributing the proboscis extrusion to ventilation in flying Drosophila (Lehmann and Heymann, 2005), the authors stated that the proboscis extensions were often correlated with water emission spikes during the initial stage of flight, soon after the fly had been placed into the respiratory chamber: ‘Water spikes only occur during flight and are often correlated with the extension of the fly's proboscis’ (Lehmann et al. 2000).

It is questionable whether proboscis extension is a typical ventilation procedure during flight in flies. It is probable that it exposes the humidified tongue for evaporative cooling or when suffering from stress. The CO₂ emissions during proboscis retraction would then be a side effect as is the case during dabbing at rest (L.T.W., unpublished).

Flying blowflies generate a unidirectional ventilatory airflow. This was concluded from the pressure difference between Sp1 and Sp2 and was confirmed by the selective CO₂ release through the posterior spiracles (Sp2 and abdominal spiracles). The gas exchange is performed by the flight apparatus, deforming the thoracic air sacs. This autoventilation gas exchange – despite the small deformations of the thorax by only 1–2% – raises the oxygen concentration within the thoracic air sacs to nearly ambient levels during full flight. These findings contradict the opinion that respiratory gas exchange during flight in small insects relies on diffusion alone. Instead, they confirm the hypothesis – based on a theoretical model – ‘that replacement of diffusive gas exchange by convective gas exchange should be favoured in evolution of small animals, whenever it is possible’ to reduce respiratory water loss (Kestler, 1985). Experiments with pressure reversal in the tracheal system suggest the existence of an intratracheal valve mechanism. An X-ray tomographic analysis of the tracheal system needs to be performed to answer this question (L.T.W. and P. Cloetens, in preparation).

Blowflies (C. vicina) were captured from the field. They were examined immediately, or the F1 offspring of vigorous females were bred and the adults were used after the fourth day at the earliest point after eclosion. Some flies were kept alive in a flight cage for up to 6 months, hibernating in a cool winter garden. For presentation of the results, the data for specimens with different ages (mostly between 5 days and 2 months) were pooled. There was no obvious fundamental difference in the behaviour and the physiological reactions in flies of different ages. However, the older flies were much more appetent to fly longer continuously. The use of the wild C. vicina and F1 generation guaranteed powerful fliers. The procedure to engage wild specimens and only the F1 generation avoided inbreeding and was a consequence of my experience that flies from commercial or homogeneous laboratory stocks were short-lived and incapable of flying persistently. The mean mass and s.d. of the flies was 84±18 mg. Their mass changed by 18% during the experimental runs, depending on the fluctuating intestine content. Several times a day, the flies were given a droplet of honey–water solution on the Styrofoam running ball. Flies were anesthetised with CO₂ gas only for fixation and surgical treatment.

The spiracular valves were recorded by video sequences during rest and flight, as described in Wasserthal (2014). The imported videos were then analysed using Autodesk Maya 3 (111 McInnis Parkway, San Rafael, CA, USA). The contours of the valve lips around the aperture were traced in an upper layer, and the positional changes were depicted along the time axis with the ‘Hypergraph’ tool (Autodesk).

For measurement of the pre-atrial pressure, the mesothoracic spiracles (Sp1) and metathoracic spiracles (Sp2) were connected to polyethylene tubes (outer diameter 1.2 mm, inner diameter 0.8 mm, Fig. 11A–C). The plastic tubes or stubs were glued with Pattex (Henkel, Düsseldorf, Germany) and sealed in front of the spiracular peritrema with Fixogum rubber cement (Marabu, Tamm, Germany), without injuring the atrium and valves. These pre-atrial measurements recorded the atrial pressure conditions, which, in the resting flies, essentially corresponded to the intratracheal pressure in the scutellar air sacs when the spiracular valves were open (Wasserthal, 2014). The pressure measurements at the spiracles were not unproblematic because the tight connection of the spiracles with the sensor blocked the gas exchange with the ambient air. The measuring device was a closed system with a limited gas volume of 60 µl on each side. If both spiracles of the same thoracic segment were blocked, dyspnoea occurred. This became visible in the form of short and intermittent flight phases or irregular flight. The air volume had to be replenished by repeated opening of the connecting tubes. If only one spiracle of the same segment was connected to the pressure sensor, gas exchange was less restricted and flight was more sustained. The restricted gas exchange on one side was probably not fully compensated for via the open contralateral spiracle. This is suggested by the fact that the contralateral wing movement did not have the same pressure-decreasing effect at the Sp1 as the ipsilateral wing movements (Fig. 3).

For O₂ and pressure measurements, the flies were fixed at the mesoscutellum or abdominal tergite 2. For CO₂ and H₂O flow-through experiments, the flies were glued to the mesonotum. In all experiments, a combination of Pattex with a layer of Fixogum was used. The Fixogum functioned like an elastic cushion, which was necessary to avoid impeding the thoracic deformations caused by the indirect flight muscles. Otherwise, the flies refused to fly. There were no visual movement patterns used for flight activation. Flight started spontaneously or as a reflex action when the support, a Styrofoam running ball, was withdrawn from the legs.

For measurement of the tracheal pressure inside the body, the dorsal cuticle with the underlying air sacs was perforated and connected to a bi-tubed plastic cone using a mixture of Pattex and Fixogum, which allowed the fly to be handled and tethered in the setup, while being simultaneously connected to the pressure sensor (Sensym SCXL 004 DN, Sensor-techniques, Puchheim, Germany) and to the O₂ optode. The dead space of the pressure 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 and tables. The response time was 7–10 ms and the time constant was 30 ms.

Heartbeat was measured by thermistors utilising the thermal effect of the metabolic heat of the flight muscles or by artificially warming the haemolymph of the thorax and anterior abdomen by a laser beam, as previously described (N=7, F1 offspring, 6–57 days old; Wasserthal, 2012).

Measurement of the P_O2 was performed using fibre-optic optodes (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, and arranged beside the air pressure tube in the bi-tubed adapter cone (Wasserthal, 2014). The measurements were run under controlled ambient temperature, between 20 and 24°C. 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 for continuous use over several weeks without significant reduction in sensitivity and only a slow, gradual loss in response time. For further details, see Wasserthal (2014).

CO₂ and H₂O emissions of flies without spiracle tubes were measured in a 20 ml volume flow-through chamber. The flies were glued by the meso-notum to a bracket. The chamber was connected directly to a CO₂/H₂O IRGA (LI-7000, LI-COR, Lincoln, NE, USA). Before entering the reference chamber and the specimen chamber, the air passed a CO₂- and water-absorbing scrubber containing pellets of NaOH. The airflow was a constant 1000 ml min⁻¹ at controlled temperatures of between 20 and 24°C and controlled pressure. This was slightly over-atmospheric (ΔP=10–100 Pa) to avoid unwanted gas inflow by tiny leaks. In a number of experiments, the flies were fixed at the scutellum by a cannula passing through the upper plug and connecting the scutellar air sac with the O₂ optode (N=3) or pressure sensor (N=6), as described earlier (Wasserthal, 2014; Fig. 7).

For measurement of the CO₂ release from selected spiracles, the insects were transferred into the 18 ml insect compartment of a split flow-through chamber as described previously (fig. 1 in Wasserthal, 2001). The 4 mm long tube stubs of the spiracles were extended by 24–32 mm long polyethylene tubes connected via sleeves and canalised into the 18 ml empty compartment of the split chamber (Figs 8, 9, 11). The respiratory air stream of both Sp1 and both Sp2 could thus be separately directed to the gas analyser, while the insect compartment was connected to the scrubber, or vice versa. By exemption, the effects of Sp1, Sp2 and abdominal spiracles could be derived. The pressure inside the compartments of the split chamber was measured and adjusted to identical values. For checking the possible occurrence of an internal tracheal valve, an artificial pressure decrement between the compartments was produced in order to test the possibility of a flow reversal. In the split-chamber experiments, the CO₂-free air was humidified before passing through the chamber compartments.

The wingbeat was recorded by measuring the changes in brightness caused by the upstroke and downstroke, interrupting a constant light beam projected on a Si photo-diode. The wingbeat frequency was determined by optical freezing of the wing position by stroboscopic flashes or by recording the pressure pulses with a high sampling rate of 1–40 kHz.

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.