Respiration in the Desert Locust: II. The Control of the Spiracles

The observations of a number of authors have shown that the spiracles of various insects are capable of graded opening. In the absence of ventilation this is termed diffusion control. Hazelhoff (1927) demonstrated diffusion control in Periplaneta: in 1 % carbon dioxide the spiracles are slightly open, 2 % causes further opening and in 3 % they are wide open. It has been observed in the flea (Wigglesworth, 1935), in houseflies (Case, 1957) and in the tsetse fly (Bursell, 1957), and its occurrence has been inferred in the pupae of Hyalophora (Buck, 1958), whose spiracles are very slightly open during the ‘interburst’.

Graded opening has not previously been demonstrated in an insect whose spiracles are normally synchronized with abdominal ventilation, and in this paper it will be shown that the spiracles of the locust can individually modify their behaviour within the overall pattern of synchronization, and suggestions will be made as to how this is controlled.

The first four pairs of spiracles of the locust open during the inspiratory phase of abdominal ventilation and the remainder during expiration (McArthur, 1929). That these synchronized movements in fact produce an anterior to posterior flow of air through the insect has been demonstrated by Fraenkel (1932). More recent measurements of the volume of air pumped through the insect in flight (Weis-Fogh, 1960) have shown that hyperventilation does not greatly increase the flow, so that probably a modification of spiracular behaviour takes place.

Adult Schistocerca gregaria were obtained and kept as described elsewhere (Miller, 1960a). The first three pairs of spiracles have been studied in detail, and observations made on the remainder suggest that they are similar to the third.

To record the behaviour of several spiracles simultaneously, very small mirrors were attached to the spiracle valves with a resin and wax mixture. The mirrors were made from silvered fragments of drawn out coverslips and each weighed approximately 0·05 mg.—about the same weight as one valve of spiracle 2. A weight twenty times heavier did not hamper the movements of the valves. A beam of light from a Baker microscope lamp was reflected by the mirrors through concentrating lenses on to a moving roll of photographic recording paper (Kodak R.P. 20), in a simple electrically driven camera. Movements of the spot on the film which resulted from ventilation were clearly distinguishable from those caused by opening and shutting of the valves. Where only the extreme positions of valve movement were required, the beam was focused onto a ground-glass screen and the positions marked by pencil. A slight but negligible error arose from not using a cylindrical screen.

Before a mirror can be attached to spiracle 1, part of the overlying pronotum must be cut away and the prothorax rigidly waxed to the mesothorax. Spiracles 3 and 10 are sunk in pits, and the mirror must be supported on a very small glass rod waxed to the valve. With this technique it was possible to record from three spiracles simultaneously.

Nerve impulses in the spiracular nerves were recorded with two hooked platinum and 10% iridium wire electrodes, 0·03 mm. in diameter, and insulated down to the hooks with ‘Araldite’. The nerve was lifted into a drop of paraffin oil contained in a polythene ring (diameter 1 mm.), which was glued to one electrode. Frequently this failed to contain the oil, but it was still possible to record impulses when more was added. Provided that oil did not spread into the cut ends of tracheae, the nerve remained alive for at least an hour. The impulses were amplified, displayed and photographed as described elsewhere (Miller, 1960a). Impulses were recorded from the spiracle nerves close to the ganglia and close to the spiracles but no differences were detected.

Spiracle movements and other events were recorded on the oscilloscope by the ‘buzzer’ method (Miller, 1960a).

Nerves were stimulated with two coupled, single channel, square-wave stimulators, using pulses of 1 msec, duration at various frequencies. Most stimulation was carried out under oil.

Sections of the spiracles and their muscles were cut after fixation in Camoy or Baker’s formaldehyde-calcium and double embedding by Peterfi’s Celloidin paraffin technique (Pantin, 1948). The course of the spiracle nerves was traced in fresh material after supravital staining with methylene blue.

Gassing techniques are described elsewhere (Miller, 1960a).

Nerves are named according to Ewer’s (1953) scheme for Acanthacris refucornis. The muscle numbering is that used by Snodgrass (1929, 1935). The spiracles are numbered 1–10. It should be remembered that spiracle 3, situated on the first abdominal segment, supplies tracheae principally to the thorax, while spiracle 4 does so exclusively.

Mirror recordings of the activity of spiracles 1–4 and 10, and direct observations on the remainder, have confirmed the conclusions of earlier authors concerning their synchronization with ventilation. Spiracle 10 has been reported as inspiratory at rest but expiratory in flight (Hamilton, 1937). All the present records have shown that spiracle 10 opens towards the end of expiration, although when ventilation is weak it may in addition open during inspiration.

At rest, spiracles 3 and 5–9 remain closed, so that air probably enters by spiracles 1, 2 and 4, and leaves by spiracle 10. During greater activity, and normally in mature females, spiracles 5–9 are brought into use, the more posterior first. In some locusts spiracle 3 may open only during and for a short period after flight.

The durations of the open and closed phases of the spiracles are very variable (Hoyle, 1959), but in a resting locust ventilating at 30/min., the open phase of the inspiratory spiracles seldom lasts more than 20%, and that of the expiratory spiracles 5–10% of the whole ventilatory cycle (Fig. 1). Opening of the inspiratory spiracles always follows that of the expiratory spiracles immediately : there is then a pause, which includes the compression phase (McCutcheon, 1940), before the expiratory spiracles re-open. When ventilation is accelerated, the duration of the closed phase is reduced while that of the open phase stays approximately the same, and the compression phase disappears.

In an atmosphere of 5 % carbon dioxide, the inspiratory and expiratory spiracles may each be open for as much as 80–90% of the whole cycle: the considerable overlap allows air to be inspired and expired to some extent through all spiracles. Indirect measurements of the intratracheal pressure of various insects (Watts, 1951) show an increase with moderate ventilation from 1 to 7–10 mm. Hg, whereas in hyperventilation induced by carbon dioxide the pressure falls back to 1–2 mm. These figures are probably explained by changes in spiracular behaviour, similar to those described here.

Adequate general descriptions of the spiracles of Acrididae have been given by Snodgrass (1929, 1935), Jannone (1940), Karandikar (1939) and Albrecht (1953, 1956). Only a brief description and a number of details not mentioned by these authors will be given here.

Spiracle 1 is situated on the soft membrane between the pro- and mesothorax, underneath the posterior expansion of the pronotum, which is kept clear of the spiracle by a small knob. Unlike the other spiracles the atrium leads into two separate tracheal trunks: the dorsal one supplies the head and prothorax and the smaller ventral one the first pair of legs and the flight muscles. The openings of these trunks into the atrium will be termed the dorsal and ventral orifices. The spiracle is contained on a small peritreme to which the anterior grooved valve is fixed and the posterior valve hinged. The ventral part of the posterior valve comprises well sclerotized cuticle and is saucer-shaped; the dorsal part is of soft cuticle with the typical ‘thorns ‘of intersegmental membrane (Fig. 2). A sclerotized rod runs down the anterior margin of the valve; it passes between the orifices and then inwards and round the ventral orifice. Half way round it bears a large process which curves anteriorly and upwards. The closer muscle (79), arising from an apodeme on the peritreme, runs dorsally to this process. One end of the opener (80) shares the insertion on the peritreme with the closer, and the other is attached to the posterior margin of the hinged valve near its ventral end.

The closer muscle is short (0·5 × 0·3 mm.) with outer fibres 20–30μ in diameter and an inter-Z distance of 3−5μ The central fibres are more slender, 10−15μ in diameter with an inter-Z distance of 5−8μ. The opener is longer and thinner (0·15 × 0·7 mm.) with fibres 10−20μ in diameter and an inter-Z distance of 3−5μ

The spiracle is innervated by the median nerve of the prothoracic ganglion (Fig. 3). The course of the nerve is very similar to that of Acanthacris refucornis (Ewer, 1954a). Near the spiracle it divides and two axons enter each muscle.

As in the cockroach (Case, 1957) and the dragonfly nymph (Zawarzin, 1924), the motor axons themselves divide at the branching of the median nerve into two transverse nerves. This has been shown in two ways ; by simultaneously recording from the two transverse nerves close to the spiracles, when identical patterns of impulses are obtained from each (Fig. 4), and secondly, after cutting the median nerve close to the spiracle, by electrically stimulating one transverse nerve when identical movements are seen in both spiracles. Similar tests have shown this to be true for spiracles 2 and 3 as well.

Mirror recordings were made of the open and closed positions of the valve every 30 sec. in various concentrations of carbon dioxide in air. Fig. 5 shows the average positions from thirty locusts, the vertical bars representing extreme values. In air and when the locust is at rest the valve opens 20–30 % with inspiration; it opens wider in increasing carbon dioxide concentrations and maximally in 3–4%. In more than 10% it fails to close fully.

After section of the closer the valve opens 20–30%. Section of both muscles shows that hinge elasticity is responsible for this amount of opening. Wider opening must therefore depend on differential contractions of the opener. By cutting the surrounding cuticle, inverting the spiracle and removing the tracheae, it is possible to watch the activity of both muscles. Together with observations on the intact spiracle this has provided the following account.

In air and at rest the opener makes no contraction, spiracle opening depending on relaxation of the closer and hinge elasticity alone. The action is unaltered after destruction of the opener. In 1 % carbon dioxide the valve opens approximately 50% with inspiration and the opener makes rhythmical contractions, which start during and finish after those of the closer (Fig. 6A). The instants of opening and closing of the valve are still determined by the closer relaxations and contractions.

The slower contraction of the opener draws the posterior margin of the moving valve inwards and thereby tenses a cuticular spring. The more the valve is distorted, the more energy is stored by the spring and the wider the valve opens when the closer relaxes. Before the closer contracts again, the opener relaxes and the valve closes from half to.about a quarter open. In 2% carbon dioxide the phasing changes (Fig. 6B); the opener now contracts while the closer is relaxed so that the spiracle opens wider towards the end of inspiration, from 25 % to about 50–70 % open. A further change takes place in 3–4% carbon dioxide, when both relaxation and contraction of the opener take place while the spiracle is closed and the closer contracted (Fig. 6C). Throughout inspiration the opener is therefore contracted and the spiracle 70% or more open. Finally, in more than 4–5% carbon dioxide the relaxations of the opener get smaller and disappear, so that the opener remains continuously contracted and the closer shuts the valve against the fully tensed spring (Fig. 6C).

Since both muscles make contractions while the spiracle is closed, it is not surprising that both have been termed closers by some authors (Lee, 1925), and some other Orthoptera have been described with two closers in spiracle 1 (Maki, 1938).

A volley of large impulses (up to 80/sec.), recorded from the transverse nerve, corresponds to the contraction of the closer, while a volley of smaller impulses (smaller by a factor of five or six and usually at frequencies lower than 20/sec.) corresponds to the contraction of the opener (Fig. 7 G). In air when the opener stays relaxed no small impulses appear, and in 5% carbon dioxide when the opener remains contracted the small impulses are continual. No change in the pattern of impulses has been detected after squashing the nerve peripherally, but after its section near the ganglion they cease. There is no evidence for sensory nerves in the transverse nerve, which are associated with the spiracle. Sensory impulses have been recorded from nerve IAα, which probably innervates cuticular sense organs near the spiracle, but they remain unaltered in carbon dioxide.

Inspection of the records reveals that the large (closer) impulses often occur in pairs and are of two sizes; moreover, one usually occurs at a frequency slightly different from that of the other. This means that they come in and out of phase, that is, they beat (Fig. 7 A, B). The same is usually apparent in the small (opener) impulses. When they coincide, the impulses are superimposed and an extra-large impulse is recorded. Similar patterns have been observed in recordings from spiracles 2 and 3.

Hoyle (1959) describes spiracle 2 of the locust as being innervated by two axons, a ‘slow’ and a ‘fast’. They often fire together but sometimes the slow discharge is absent, while the fast is always there. He has observed pairs of impulses, but states that both members occur in the same axon. However, in spiracle 1, since one member of the pair is of a slightly but constantly different size from the other, and since the second sometimes follows the first by 1 msec, or less (presumably within the absolute refractory period of the first) and may even be superimposed on it, it seems more likely that each axon provides one member of the pair, and that both axons are therefore involved in the normal operation of the spiracle. Hoyle draws attention to paired impulses in records from the cockroach (Case, 1957); in these, too, extra large impulses occur infrequently.

Measurements of the speed of conduction in the four axons of the spiracle 1 transverse nerve have been made by recording under oil from two pairs of electrodes separated by 3 mm. (Fig. 7H). Impulses in the two axons to the closer travel at approximately 1·09 and 0·97 m./sec. ; in the two axons to the opener they travel at approximately 0·43 and 0·36 m./sec. These speeds are slow compared with 6–7 m./sec. in the giant fibres and 2–3 m./sec. in the cereal nerves of the cockroach (Roeder, 1948).

Stimulation of the transverse nerve of spiracle 1 with single shocks, after destruction of the opener, produces twitches in the closer, and these summate to give a smooth tetanus at frequencies greater than 15/sec. The opener, however, does not respond to single shocks; at 3/sec. a very slow and weak contraction occurs; at 6/sec it contracts more strongly still taking 2–3 sec. to do so, and at higher frequencies the contractions become stronger and faster. The artificial and natural frequencies of impulses produce about the same speed and amount of contraction in each muscle, and they show that the closer is a fast (phasic) muscle and the opener a slow and at times a tonic muscle.

The foregoing experiments have shown that carbon dioxide affects the activity of the opener. To determine whether this is initiated centrally or peripherally, mirrors were waxed to the valves of both spiracles i and a very gentle stream of carbon dioxide directed into one. After a delay of several seconds increased opening occurred in both spiracles simultaneously. By comparison with results from spiracle 2, and since each spiracle receives the same motor input, the reaction probably takes place through the ganglion, rather than by means of a local system. The possibility remains that an extra-ganglionic system may control both spiracles simultaneously. However, after section of the median nerve close to the ganglion the limited amount of spontaneous activity which then appears is not related in the two spiracles (see p. 256).

Modifications in the patterns of nerve impulses in the transverse nerve, which took place as a result of carbon dioxide treatment, were unaltered after crushing the nerve peripherally and after completely denervating the prothoracic ganglion, except for the median nerve and the anterior and posterior connectives. Consequently, carbon dioxide reception must take place within the ganglion or in another segment.

After section of the nerve cord between pro- and mesothoracic ganglia (or between meso- and metathoracic ganglia) all trace of synchronized action disappears from spiracle 1. The valve usually remains 10 or 20% open with continual fluttering movements due to the closer. Prodding the insect, or struggling, causes immediate full closing, and afterwards the valve opens more widely for a few seconds. In 1–2 % carbon dioxide it opens wide and in higher concentrations all fluttering ceases: opening results from a maintained contraction of the opener and relaxation of the closer.

Records from the transverse nerve show a constant stream of pairs of large closer impulses (7–12 pairs/sec.)—too slow apparently to maintain a tetanus in the closer (Fig. 8). The slight difference in the frequencies in the axons is more constant than in the intact insect, and the number of extra-large impulses (beats) gives an indication of the frequency difference in the firing of the two axons. With struggling the frequency in both axons increases and the same phasing is maintained. In 1 % carbon dioxide a few small (opener) impulses appear but the large impulses are unchanged (Fig. 8B); in 2%, the small impulses increase in frequency and the large decrease; in 3–4% the large disappear altogether (Fig. 8D).

Since section of the nerve cord between the pro- and mesothoracic ganglia abolishes the rhythmic opening and closing movements in spiracle 1, but not its sensitivity to carbon dioxide, the latter must depend on the head or prothoracic ganglion.

Head-perfusion experiments were undertaken to demonstrate a carbon-dioxide receptor which controlled the activity of the opener. (The technique is described elsewhere, Miller, 1960a.) 1–2% carbon dioxide was Injected into the mandibular air-sac and it produced almost immediate contraction of the opener. After section of the nerve cord in the neck the injections had no effect, although 4–5 % carbon dioxide directed at the ganglion then produced a weak contraction.

It seems that the head contains sensitive carbon-dioxide receptors whose stimulation increases abdominal ventilation, induces neck and prothoracic ventilation and causes contraction of the opener of spiracle 1. It is possible that one receptor controls all the reactions. In addition the prothoracic ganglion has less sensitive receptors affecting abdominal ventilation and the spiracle 1 opener. Stimulation of the head receptors produces neck and prothoracic ventilation only when the metathoracic ganglion is intact: for opener contractions, however, no more than the prothoracic ganglion need be intact. Since the dorsal trunk leads directly from spiracle 1 to the head, it is not surprising that the head should have an overriding influence on the activity of the spiracle 1 opener. It has no comparable influence over other spiracles.

The opener is inserted on the posterior margin of the moving valve behind the ventral orifice, and the closer in front on the upturned process (Fig. 2). When both muscles contract strongly and simultaneously, the long narrow orifice is completely constricted by the scissor-like action of the rod bearing the process and the posterior margin. In more than 4–5 % carbon dioxide, when the opener remains contracted, the orifice opens slightly only during inspiration as the closer relaxes. On the other hand, when the opener remains relaxed the orifice is open all the time. The significance of this surprising consequence of opener contractions (a misnomer in the context) is discussed elsewhere (Miller, 1960b).

A detailed account of the structure of spiracle 2, together with a description of the histology of the closer muscle and its innervation, has been given by Hoyle (1959). Only a small number of additional points will be mentioned here.

Spiracle 2 possesses a single closer muscle (111) and the valves are opened by the elasticity of the cuticular surround. Hoyle points out the occurrence of two types of muscle fibre in the closer: a central core of thin fibres with an inter-Z distance of 6–7μ, and surrounding thicker fibres with an inter-Z distance of 3–6μ. Inner fibres with an inter-Z distance as great as 10μ have been observed during the present investigation. The thin fibres are more numerous than in the closer of spiracle 1, and the inter-Z distance of many is greater. Hoyle suggests that they might be normal fibres whose development has been arrested, but neither Tiegs (1955), Wigglesworth (1956) nor Smith (1958) mentions a stage in fibre development when the striations appear more widely separated than in the mature muscle.

Secretory tissue occurs in the thickness of the anterior valve, and yellow or colourless droplets have been observed hanging inside the valve after rough handling of the insect. Spiracle 2 of Romalea microptera froths a black noxious liquid when the insect is irritated, and the secretions in Schistocerca may be similar to this.

After section of the closer muscle the valves open and are separated maximally by 0·1−0·2 mm. Observations made during flight, however, show the valves then to be separated by 0·4−0·5 mm. This wide-opening is achieved by a cuticular device.

The suture between the meso- and metathoracic lateral wall, dorsal to spiracle 2, comprises on the anterior margin of the metepistemum a well-formed apodeme to which the metacoxal muscle (125) is attached. The apodeme is joined to the thickened posterior margin of the mesepimeron by a bridge of cuticle with specific staining properties (Fig. 9 B). The bridge, lying under the mesepimeron and barely visible from the exterior, extends about three-quarters of the way from the spiracle to the wings. It is continuous with the cuticle surrounding the spiracle valves but, while the latter acts like a metal spring, the bridge is rubber-like and has many of the properties of the elastic ligaments of the wings (Weis-Fogh, 1958). The rubbery quality is not lost after fixation in alcohol or after drying and re-wetting. In sections fixed in Camoy or Baker’s fixatives and stained with Hansen’s haematoxylin or chlorazol black, the bridge stains more intensely than normal cuticle : with Mallory’s triple stain it is red.

The bridge allows the mesothorax to be pulled very slightly anteriorly and up-wards from the metathorax ; on release it snaps back into position. By pulling both ends of a thin horizontal section of the lateral wall under a microscope, the action of the elastic part can be watched. A thin superficial layer of sclerotized cuticle prevents the movement from exceeding about 0·05 mm.

In the nymphs of Schistocerca this movement pulls the base of the spiracle valves apart and tends to close the spiracle. In the adult, however, a small and sometimes darkened ligament runs in the cuticle from a knob on the dorsal end of the anterior valve to the apodeme of the metepistemum (Fig. 9,A). It is attached very close but just medial to the vertical axis of rotation of the valve, so that a small anterior movement of the mesothorax pivots the valve wide open on the ligament, and this in turn cants open the posterior valve (Fig. 9C). After section of the ligament no wide-opening occurs, although the normal movements are unaffected.

Wide-opening has been observed momentarily during struggling and always in flight. It can be artificially induced by gentle pressure on the cuticle just anterior to the spiracle. After section of both pleural apophyses through their respective coxal cavities, it no longer occurs in flight: section of the metathoracic apophysis alone is nearly as effective. Electrical stimulation at 100/sec. of nerve IV Bbi, which innervates the metathoracic stemopleural muscle (115), causes a slow contraction of that muscle but wide-opening does not occur. Locusts were flown after section of this nerve and of the homologous nerve in the mesothorax, but wide-opening was unaffected. Apparently the rigidity provided by the stemo-pleural apophysis is essential for wide-opening, but it is not dependent on contraction of the stemo-pleural muscles. Section of the coxal muscles which are inserted on the pleural and intersegmental sutures does not prevent wide-opening in flight, although these muscles are probably responsible for its occurrence during struggling. The basalar and subalar muscles in the mesothorax (98, 99) and in the metathorax (128, 129) assist in the downstroke of the wings (Pringle, 1957), and after their unilateral section, the wing movements of that side are much reduced and wide-opening almost disappears. Comparison with the intact side, where wide-opening continues, suggests that contraction of the basalars and subalars causes it. Electrical stimulation of the mesothoracic basalar and subalar muscles is more effective in producing wide-opening than stimulation of the metathoracic muscles. Weis-Fogh (unpublished) has suggested the presence of tonic components in these muscles, which may be responsible for wide-opening.

The spiracle is innervated by the median nerve of the mesothoracic ganglion (Fig. 3). The course of the nerve is similar to that of Acanthacris refucornis (Ewer, 1954a). Near the salivary duct the transverse nerve is joined by a branch from nerve 1 of the metathoracic ganglion. Before and after crossing the duct it supplies a number of branches (meso IF) to the vestige of the nymphal spino-pleural muscle. Close to the spiracle muscle the transverse nerve divides. One branch, meso IG (meso IE of Ewer), supplies the vestige of the mesothoracic pleuro-subalar muscle: as it passes dorsal to the spiracle, very slender branches leave it and appear to end on the surface of air-sacs. At the bases of these branches there are swellings in the nerve, which contain large cells. The other branch, meso IE, sends two axons into the spiracle muscle.

Mirror tests on the open and closed positions of the valves have been made in various carbon-dioxide concentrations (Fig. 10). The valves never open more than 30–40% with inspiration; that is, there is no wide-opening in carbon dioxide. In more than 7–8% carbon dioxide they usually fail to close completely, but this value is variable and in some locusts the valves may make only the weakest closing movements in 5 %, whereas in others more or less full closing continues in 10–12% carbon dioxide. A comparable variation in the sensitivity of the dragonfly spiracles to carbon dioxide has been shown to depend on the water balance of the insect (Miller, unpublished), and the same may be true for the locust. To conclude, spiracle 2 in the non-flying locust shows much less of a graded response than spiracle 1.

Unilateral gassing with 10 % carbon dioxide has demonstrated that one spiracle may remain open while its partner continues to make full closing movements: the volleys of impulses remain unchanged in both transverse nerves. The direct action of carbon dioxide on the neuromuscular junction, reducing the electrical responses and the tension developed, has recently been demonstrated (Hoyle, 1960), and is no doubt responsible for the independent action of one spiracle.

The volleys of impulses giving rise to spiracle closing nearly always comprise pairs of impulses of slightly different sizes. For the reasons already given for spiracle 1, both axons appear to be involved in the normal operation of the closer (Fig. 11).

After section of the nerve cord between the meso- and metathoracic ganglia, the behaviour of spiracle 2 is similar to that of spiracle 1. The volleys of impulses in the transverse nerve are replaced by continual pairs of impulses, usually between 6 and 12 pairs/sec. (Fig. 11D). In some locusts the spiracle remains closed, but in many it is 20% open and continually fluttering: with struggling it closes fully. The sensitivity to carbon dioxide is much increased and it will open slightly in 1–2 % and completely in 2–3%, when all fluttering ceases. As in the intact insect, the impulses in the transverse nerve remain unaffected during opening. The sensitivity of the response remains undiminished for several weeks.

After the spiracle has been uncoupled from ventilation, the differing frequencies in the two axons become more regular. By watching the valves and listening through earphones to the impulses recorded close to the spiracle, the apparent drop in frequency as the impulses nearly and then completely coincide can be seen to correspond to a larger twitch of the valve. Periods of coincidence and larger twitches occur about once a second, while there may be 8–10 small twitches per second. Hoyle (1959) describes a rhythm persisting in spiracle 2 after section between meso- and metathoracic ganglia, often seen as a regular rise and fall in the frequency of impulses. The asynchronous firing of the two axons gives the impression of a rise and fall of frequency. This rhythm is clearly of a different nature from that which occurs in the intact insect involving a rhythmic cessation of all impulses. Hoyle states that a pair of impulses separated by 3–4 msec, gives rise to a considerably greater tension than would be evoked by either alone; his account refers to impulses in the same fast axon, whereas the pairs under consideration here, which give rise to larger twitches, occur in different axons.

It is not surprising that the apparent sensitivity of the peripheral carbon-dioxide reaction should be greater after the spiracle is uncoupled from ventilation, since the frequency of motor impulses reaching the spiracle is then less. In an intact but immobilized locust the spiracle may fail to close fully in 7% carbon dioxide, whereas after release full closing may occur in higher concentrations. Variation of the impulse frequency provides a method of altering the sensitivity of the spiracle reaction to carbon dioxide. To what extent it is used, perhaps in relation to the water balance, is unknown.

To conclude, gradations of the open position of spiracle 2 are affected primarily through the wide-opening mechanism and appear only in flight, although a reduced amount of opening (10–20%) is infrequently seen in resting locusts. Gradations of the closed position do not normally take place in less than 5–7% carbon dioxide, and then depend on the local action of carbon dioxide on the neuromuscular junction (Hoyle, 1960).

Spiracle 3, lying just anterior to the auditory capsule and belonging to the first abdominal segment, supplies large tracheae to the thoracic air-sacs and flight muscles, and also to the gut. The inner end of the atrium is closed by the hinged valve which is produced into a manubrium for the attachment of the dorsal closer and ventral opener muscles (Fig. 12). The short closer, 148 (0·32 × 0·096 mm.), is inserted on a rigid apodeme immediately above the spiracle, and the long narrow opener, 147 (2·6 × 0·064 t0 0·032 mm.), is inserted on the soft ventral interseg-mental membrane, beside the tensor of the tympanum (146). For over half its length the opener runs parallel to muscle 146 and is bound to it by connective tissue. The opener comprises fibres normal in appearance with an inter-Z distance of 3–5μ. The structure of the closer is similar to that in spiracle 1. Section of both muscles shows the hinge to possess limited elasticity sufficient to open the spiracle valve 20%.

The innervation is again similar to that of Acanthacris ruficornis (Ewer, 1953, Fig. 3). The anterior median nerve divides into transverse nerves shortly after leaving dorsally from the metathoracic ganglion. Each transverse nerve (meta 1C) sends a branch (meta 1F) to muscle 117 and one to nerve ABD₁. It then crosses the ventral end of coxal muscles 119 and 120, sending several branches into the former. Shortly before reaching the spiracle a further branch leaves to supply the vestige of the nymphal longitudinal oblique intersegmental muscle of the metathorax (140). The transverse nerve joins the opener half way along its length and after sending several twigs into that muscle, runs parallel and into the closer.

Three patterns of behaviour are commonly seen in spiracle 3 of the non-flying locust : in the first it remains closed, in the second it opens slightly during inspiration and in the third it opens fully.

The first pattern usually appears when the locust is at rest or moderately active and ventilation is not greatly stimulated. At such a time records from the transverse nerve show a constant stream of impulses (Fig. 13 A). The second occurs when ventilation is rapid (c. 60/min.) but shallow, and the valve opens about 10% or less in phase with the other inspiratory spiracles : sometimes it opens twice per cycle. Opening results from partial relaxation of the closer and from hinge elasticity; the action is not disturbed by section of the opener. Records from the transverse nerve show alternate volleys of impulses and silent periods (Fig. 13B).

With slow deep ventilation (30/min.), the third pattern appears when the spiracle opens 50–100% during inspiration. The amount of opening is dependent on differential contractions of the opener, although unlike spiracle 1 there is no storing of energy in an elastic system and the contractions of opener and closer alternate regularly. This is reflected in the patterns of nerve impulses from the transverse nerve (Fig. 13C–E), which show a regular alternation of large (closer) and small (opener) impulses—smaller by a factor of 4–5. The small impulses never overlap the large and occur at comparable frequencies (60–70/sec.) when the spiracle opens 100%, and at lower frequencies (10–20/sec.) when the spiracle opens less.

Mirror tests on the open and closed positions of the valve in different carbon dioxide concentrations show that the behaviour of the spiracle is comparable to that of spiracle 1 (Fig. 14). The amount of spiracular opening is dependent on whether the second or third pattern appears and much individual variation is again seen. In concentrations greater than 10–12% the spiracle fails to close fully, and contractions continue only in the opener, so that the spiracle moves between 20 and 100% open.

Unlike those of spiracle 1, both muscles respond with twitches to single shocks, and tetanus is produced at frequencies of over 15/sec. in both. In the intact insect contractions of the opener are slower than those of the closer.

With mirrors fixed to both spiracles and a gentle stream of carbon dioxide aimed into one, it can be seen that modifications of behaviour always affect both spiracles simultaneously. After the destruction of all lateral nerves and the anterior or posterior connectives of the metathoracic ganglion, and after squashing the transverse nerve peripherally, comparable changes in the patterns of impulses with carbon dioxide can still be recorded. It can be concluded that the modifications of spiracle behaviour are regulated entirely from the metathoracic ganglion and do not depend on any external sensory input. Since the ventilatory rhythm originates from the metathoracic ganglion (Miller, 1960a) no method of uncoupling this spiracle from ventilation has been discovered.

As in spiracles 1 and 2, the volleys of impulses to the opener and closer each comprise impulses of two different sizes (Fig. 13) and for the reasons already discussed it is probable that the operation of each muscle is controlled by impulses in two axons.

The continued activity of the denervated spiracle as an independent effector has been described by many authors (Wigglesworth, 1935, 1941; Beckel & Schneiderman, 1956; Case, 1957).

After section of the transverse nerve of spiracle 1 or 3, the opener and closer muscles show no activity. One or two hours later the closer contracts and the spiracle is closed. Subsequently, it relaxes in more than 5% carbon dioxide, but the opener makes no contractions.

The left transverse nerve of sixty locusts was cut close to the spiracle, but medial to its junction with meso IG. In the majority the closer showed no more activity for 1 or 2 days. Later it contracted and the spiracle closed. The locusts were kept in a cage under normal conditions (see Part I) and were inspected three times a day. At each inspection the locusts were pinned on one side to a cork and observed for 5 or 10 min. During closing, five stages can be recognized spread over about 24 hr. :

• Valves are open and motionless, but close momentarily after one is pulled out and released.

• Small spontaneous movements appear.

• Valves are half closed and make continual small movements.

• Valves are almost fully closed but still moving.

• Valves are fully closed.

Most spiracles close after 2 or 3 days—some take as long as 5.

In a further batch of twelve locusts the transverse nerve was cut close to the ganglion, but the subsequent behaviour was unchanged.

Fig. 15 shows the times of closing: eight spiracles closed immediately after nerve section. When the locusts were kept for 3 hr. in the dark or were pinned down and kept at 4° C., then in about 70% spiracle closing followed nerve section immediately. If the locusts were kept immobilized the spiracles remained closed, but shortly after their release they opened. This suggests that the denervated spiracle is normally kept open for the first day or two after nerve section by the level of metabolic end-products in the insect, and that only after 3 hr. immobilization do these fall sufficiently low for the spiracle to close. Herber & Slifer (1928) concluded that the ventilation frequency of Melanoplus femur-rubrum did not reach a true ‘resting rate’ for at least 1 hr. after the insect was pinned down and then only in the complete absence of struggling. It would follow from the hypothesis that spontaneous closing is normally inhibited in the intact locust by the level of metabolic end-products, so that the spiracle is able to open when the motor impulses cease; but this seems improbable and there is no evidence to suggest that the spontaneous closing mechanism is operative in the innervated muscle.

The subsequent closing of spiracle 2 after 2 or 3 days may result from a decrease in sensitivity to metabolic products or an increase in the excitability of the muscle. That it was not a result of water loss through the open spiracle was shown by keeping locusts in moist and in dry air following nerve section, and observing that spiracle closing occurred after approximately the same interval in each.

The closed denervated spiracle is very sensitive to carbon dioxide, opening fully in 1–2 % and reclosing shortly after return to air. It opens a few seconds after the start of struggling or flight. It does not react to oxygen lack, and remains closed in 100% nitrogen until nearly all movement of the locust has ceased. The movements are always slow and smooth—very different from the fluttering seen in the spiracle uncoupled from ventilation. This behaviour is quite unlike that of the denervated cockroach spiracle 2 (Case, 1957) which closes within 24 hr. of denervation and then will not open in less than 15% carbon dioxide. Closing is followed after 4–6 days by ‘fasciculation’, which Case associates with the degeneration of the peripheral nerve stump.

In the locust the sensitivity remains for 2–3 months; occasionally, when the sensitivity was much reduced, the tracheae supplying the closer muscle were found to be partially filled with liquid. Hoyle (1960) comments on the need for the air passages into the muscle to be unblocked for carbon dioxide to affect the muscle, and this observation suggests that the same mechanism is responsible for muscle relaxation in the denervated spiracle. Nerve regeneration, indicated by the resumption of synchronized movements, occurred in two locusts only (3%). Case (1957) reported regeneration in 36% of operated cockroaches.

It seemed that the independent behaviour of the spiracles could be explained if motor nerve cells were present in the muscle. To investigate this possibility, 6μ. sections of the muscle were cut and stained for cholinesterases with myristoyl choline and fight green (Denz, 1953 ; Bowden & Lowy, 1955). Dark areas appeared in the sections but they gave no clear indication of the presence of nerve cells.

The effect of nicotine on insect nervous tissue is well known (Roeder, 1939). After initial excitation, it causes the complete block of ganglia. To test for the presence of nerve cells in the closer, the spiracles were cut out and floated on a solution of 0·001 % nicotine sulphate in Ringer. They closed immediately and remained closed for several hours, but would open in 5% carbon dioxide. In stronger nicotine solutions (0·01 %) they would not open in less than 20% carbon dioxide. Nicotine excites the closer to contract and its subsequent failure to destroy the autonomous action suggests that contraction of the closer is not dependent on continual excitation from a motor neurone in the muscle, but is a property of the muscle fibres themselves.

Hoyle (1959) pointed out that the thinner muscle fibres with a long inter-Z distance might be expected to perform slow tonic contractions. An attempt was made to destroy most of the large outer fibres with a small scalpel and a tungsten hook. In other spiracles the thin inner fibres, together with the large fibres of one side, were destroyed. The operations were performed on twenty denervated spiracles, and after observation the muscles were fixed in Camoy and stained with acid fuchsin, so that an estimate of the remaining fibres could be made. Provided that the thin central fibres were not damaged, the spiracle remained closed and continued to react to carbon dioxide. If, however, they were destroyed, but more than half of the large fibres were left intact, the spiracle opened and showed no further activity.

Although not conclusive, the results suggest that the maintained contraction of the denervated spiracle and the subsequent relaxations in low carbon-dioxide concentrations depend on the small central fibres. The observation that the closers of spiracles 1 and 3, possess a few thin fibres and make maintained contractions accords with this suggestion.

The operation of the first three pairs of spiracles can be compared when their valves are represented by levers moved by springs and muscles (Fig. 16). This emphasizes the similarity of the closing and differences in the opening mechanisms, the latter being effected in spiracle 1 by hinge elasticity and a muscle in series with a spring; in spiracle 2 by a spring with variable tension (the piston), and in spiracle 3 and the remaining spiracles by a muscle which probably includes a variable elastic element and at times works like a rubber band (Miller, 1960b).

It is noteworthy that the median nerves in each segment, in addition to innervating the spiracles, supply branches to some of the nymphal muscles which degenerate in the adult (Ewer, 1954b). Thomas (1954) suggested that these muscles may make nymphal respiratory movements or possibly that they help to rupture the cuticle at ecdysis. Ewer (1954b) has proposed that they may be important in preserving the shape of the pterothorax immediately after moulting.

Recordings made from meso IG of immature adults, which supplies the pleuro-subalar muscle, have shown bursts of motor impulses which occur during abdominal expiration. The frequency of impulses increases when ventilation is more vigorous. This observation supports both Thomas’s first and Ewer’s suggestions; however, the situation of the muscles makes it unlikely that they could contribute a significant ventilating movement, whereas their contraction during expiration when the soft pterothorax tends to be blown up, may be important in preserving its shape.

It has been shown that the openers of spiracles 1 and 3 are controlled from their respective ganglia and that their contractions in response to carbon dioxide do not depend on receptors outside the central nervous system. In another paper (Miller, 1960a) the possible direct action of carbon dioxide on the membrane of post-synaptic fibres was discussed in relation to the motor neurones of the ventilatory muscles. It seems that a similar mechanism could give rise to the increased activity of the opener motor neurones in response to carbon dioxide, and thereby contribute further to the economy of interconnexions in the central nervous system, as was suggested elsewhere (Miller, 1960a).

Spiracle 2 is an ‘odd-man-out’ and it is difficult to explain why this spiracle alone in the locust should have one muscle and a peripheral control mechanism. Observations on the one-muscle spiracles of Diptera and Odonata have shown that they have a similar peripheral control, and this may be a property of all such spiracles. No insect is known to have an opener muscle in spiracle 2 (Maki, 1938), and openers are not found in spiracle 1 outside the Orthoptera. It may be that the peripheral mechanism has some particular significance for the locomotory segments, although what part it plays in the adult locust remains uncertain. Alternatively, the peripheral mechanism may represent a more primitive system which is overridden in the locust in the interests of synchronization with ventilation. There is plenty of evidence (to be discussed in a later paper), however, that in Odonata and Diptera it takes part in the normal control of the spiracle. In the first instar locust spiracle 2 is rarely synchronized with ventilation; more often it appears to act independently and most probably under the influence of the peripheral mechanism.

Both the ventilatory and the spiracular rhythms are autonomous although modified by sensory input (Hoyle, 1959). The presence of carbon dioxide in the ganglion may be necessary for their initiation, but each cycle is not dependent on the build-up and release of the gas. A comparison may be made with the cockroach where ventilation and synchronized movements appear only in more than 10% carbon dioxide (Hazelhoff, 1927); below this the spiracles are regulated perhaps by a peripheral mechanism, similar to that of the locust. The locust, possessing the same mechanisms, is much more strongly biased in favour of ventilation synchronization.

Attention has been drawn to the firing of two motor axons at slightly different frequencies during the normal operation of the spiracle. This has suggested an hypothetical mechanism to account for the rhythmical disappearance of motor impulses in the nerves to the spiracle muscles during ventilation, as follows. Two intemuncial neurones each firing spontaneously and continually, but at slightly different frequencies, relay onto the motor neurones of the spiracle nerves. One intemuncial is inhibitory and the other excitatory. The excitatory impulse is inhibited at the synapse if, say, the inhibitory impulse precedes it by less than 200 msec. For peripheral α-inhibition in Crustacea the inhibitory impulse must precede the excitatory by only a few msec, to be effective (Katz, 1949); however, at least in vertebrates, a central inhibitory volley can still be effective after 100 msec. (Sherrington, 1947). If, for example, the excitatory intemuncial in the locust fires at 10/sec, and the inhibitory at 11 /sec., two excitatory impulses will be inhibited at the synapse every second. If two controlling cells are now postulated, one influencing the frequency in both intemuncials by the same amount, and the other altering the frequency in one only, both changes in frequency and variations in the duration of periods of firing and of silence can be achieved (Fig. 17). To obtain some combinations it may be necessary to postulate a second inhibitory intemuncial.

By some such plan two continually firing cells can produce a much slower rhythm in a post-synaptic fibre without any sensory feed-back. Such a system could be responsible for the rhythmic firing of the motor cells of the ventilatory muscles as well as those of the spiracle muscles.

The same principle is used in beat-frequency electronic oscillators, where two oscillators produce a third frequency much lower than that of either alone.

After section of the nerve cord between the meso- and metathoracic ganglia, the rhythm disappears from spiracles 1 and 2, and is replaced by continual firing in the motor nerves. This can be explained if the inhibitory intemuncial is in the metathoracic and the excitatory in the mesothoracic ganglion. During strong flight, motor impulses in the nerve to spiracle 2 cease entirely (Miller 1960b); this is explicable if at such times the excitatory and inhibitory intemuncials fire in phase.

It should be emphasized that this scheme is entirely hypothetical, and recalled that the patterns of impulses in the transverse nerves involve probably two motor fibres, so that the scheme must perhaps be duplicated.

I would like to thank Prof. V. B. Wigglesworth for encouragement and supervision of this work. I am most grateful to Prof. T. Weis-Fogh for much useful discussion and help. My thanks are due to Dr F. S. J. Hollick for the supply of small mirrors, to Dr G. Hoyle and Dr D. S. Smith for permission to quote their unpublished results, and to the Agricultural Research Council for financial support.