J. Exp. Biol. (1964), 41, 331-343 3 3 I

With 12 text-figures

Printed in Great Britain

FACTORS ALTERING SPIRACLE CONTROL INADULT DRAGONFLIES: WATER BALANCE

BY P. L. MILLER*

Department of Zoology, Makerere College, Kampala, Uganda

{Received 5 October 1963)

INTRODUCTION

The valves of all the spiracles of adult dragonfiies are controlled by single closermuscles. These muscles respond to carbon dioxide directly (Miller, 1962) probablythrough the same mechanism as that which occurs in spiracle 2 of the locust (Hoyle,i960). The threshold of the response to carbon dioxide is determined in part at leastby the frequency of motor impulses in the nerve to the closer muscle. Hypoxia,flight and abdominal ventilation have been shown to alter this frequency in spiracle 2 ofdragonfiies in different ways (Miller, 1962), and the water balance is here shown to beanother factor which may affect it and hence exert an influence over spiracle control.In dragonfiies which are partially desiccated spiracle control is tight and the thresholdof the carbon dioxide response is high, while in well-hydrated insects the threshold islow.

Mellanby (1934) and Ramsay (1935) have clearly demonstrated the importance ofthe spiracle valves in reducing water loss from an insect. Changes in spiracularactivity following alteration of the water content, however, have only recently beenrecognized (Bursell, 1957), although the likelihood of such changes has been envisagedfor many years (see Wigglesworth, 1953, p. 433).

Some initial observations were made in the field in Uganda on about 60 specimensof Brachythemis leucosticta (Burmeister) and Trithemis aimulata (Palisot de Beauvois)(Libellulidae). They indicated that spiracle control varies very much in differentindividuals caught on the same occasion and examined immediately after capture.For example spiracle 2 might close instantly after a burst of struggling or it mightcontinue to make movements synchronized with abdominal ventilation for a further2-3 min. Such variation may depend on many factors including the age, condition, sexand temperature of the dragonfly and on the interval since the last meal, but the experi-ments reported here indicate that water balance may also be a contributory factor.

MATERIAL

Most of the experimental work was carried out in Uganda on Ictinogomphus feroxRambur (Gomphidae). Methods for obtaining teneral and mature adults have alreadybeen described (Miller, 1962). Other species used are mentioned in the appropriatesections.

• Present address: Department of Zoology, Oxford, England.

332 P. L. MILLER

METHODS

The means of registering spiracle valve movements and of recording nerve impulseshave already been described (Miller, 1962). In all experiments the central recordingsite was chosen and paraffin oil was not employed. The experiments were carried outat 23-250 C. and at 60-70% R.H.

The frequency of nerve impulses was determined from counts made on films of theoscilloscope trace. The values given represent the sum of the frequencies in the twomotor axons to the spiracle, which were seen to fire at more or less the same frequencyunder all experimental conditions. They are derived from the mean of 10-20 countseach of 200 msec, duration.

The physiological saline used contained 153-1 mM./l. sodium, 2-6 mM./l. potassium,i-8 mM./l. calcium and 24 mM./l. glucose, pH 6-8-7-0.

Other techniques are described under the appropriate experiments.

RESULTS

Experiment 1. The effect of changed water balance on the control of spiracle 2

This experiment was carried out in order to determine whether dragonfiies whichhave been made to drink copiously exercise a different degree of control over theirspiracles compared with those which have been partially desiccated. In addition toIctinogomphus ferox, several small libellulid species were used: Brachythemis leucosticta(Burmeister), B. lacustris Kirby, Trithemis amtulata (Palisot de Beauvois) and Orthetrumiulia Kirby in Uganda, and Sympetrum striolatum Charp. in England. Mature insectswere chosen 1-2 days after their capture, by which time most of their gut contents hadbeen voided. The tests were each made on batches of 4-7 insects and in all over100 have been examined.

Dragonfiies were made to drink by allowing them to chew at the end of a pipettefrom which distilled water was expressed. By adding 0-5% methylene blue to thewater and subsequently dissecting a few specimens the occurrence of drinking wasverified by the presence of the dye in the gut. The amount of water imbibed wasdetermined by weighing the insect on a torsion balance after blotting off superficialdrops. As much as a 25 % increase in weight was brought about in this way.

Dehydration was achieved by keeping insects in dry air over phosphorus pentoxide.The rate of desiccation was accelerated in some individuals by lifting a flap of cuticlefrom the pterothorax which was later waxed back in position. As an alternative methodsome dragonfiies were placed in a stream of 1 o % carbon dioxide in which their spiraclesremained open. The progress of desiccation was followed by periodic weighings on atorsion balance, and in some insects up to 10% of the original body weight was lostafter several hours' treatment. Dragonflies which defaecated during the experimentor which were unduly active were rejected. Many batches of insects were alternatelydesiccated and hydrated several times. They survived this treatment well and changesin weight brought about by these means were maintained for several hours.

At intervals during treatment the valve activity and hence the control exercised byspiracle 2 was checked in 2-5 % oxygen, 5 % carbon dioxide in air, and after a 10 sec.flight. This flight was performed with the insect glued to a small glass rod in front of a

Spiracle control in adult dragonflies: water balance 333

wind tunnel: it was initiated by pinching the abdomen and terminated by allowing thelegs to grasp a small ball of paper. The rod was weighed with the insect and its weightwas subsequently subtracted.

In Fig. 1 the percentage change in weight of I. ferox is plotted against the amount ofopening of spiracle 2 in 3 % oxygen. After a gain in weight of 10%, brought about bydrinking, spiracle 2 opens fully in this gas mixture, whereas after a weight loss of 3-6 %,accompanying desiccation, it remains more or less closed in the same gas.

100-1

80-

60-

C 40-

O00c.c

o

20-

- 8 - 6 - 4—r~ 0 +2

Change In weight (%)

— 1 —+ 4

— I —+ 6 ~7iT

Fig. 1. Ictmogompkus ferox. Changes in weight brought about by desiccation or hydrationplotted against the percentage opening of spiracle 2 in 3 % oxygen in nitrogen.

In Fig. 2 the loss in weight with desiccation of initially hydrated Sympetrumstriolatum is plotted against the interval between the end of a 10 sec. flight and the fullclosing of spiracle 2. After a weight loss of 6-13% the spiracle closes with no delayafter a short flight.- Prompt closing also follows considerably longer flights in desiccatedinsects.

Fig. 3 shows the increase in length of the interval between the end of a 10 sec. flightand the full closing of spiracle 2 as the dragonflies are progressively hydrated, followingearlier desiccation. Fig. 4 shows the rise in the threshold of the response to 2 % carbondioxide following prolonged treatment with 10% carbon dioxide. After these lasttests saline was injected into the thorax and there was an immediate lowering of thethreshold to a value similar to that in hydrated insects. This suggests that prolongedtreatment with 10% carbon dioxide does not itself bring about fatigue of the responsebut has its effect through the water balance.

The results show that variations in the amount of spiracle control exercisedcorrespond to the state of hydration of the insect. The experiments discussed belowwere designed to throw light on the mechanism involved.

Experiment 2. Measurements of the frequency of nerve impulses in the spiracle nervefollowing desiccation or hydration

At least two mechanisms could account for the results reported in Expt. 1: the waterbalance might affect the spiracular response to carbon dioxide directly, or the action

334 P. L. MILLER

M-i

12-

60 90Time (sec)

Fig. 2. Sympetrum striolatum. The percentage loss in weight of four dragonflies, initially wellhydrated, plotted against the interval between the end of a io sec. flight and full closing ofspiracle 2.

30-i

60

Time (sec.)

Fig. 3. Four Brackythemis lacustris (C © • O) and three B. leucosticta ( x A A). The gain inweight, after drinking, plotted against the interval between the end of a 10 sec. flight and the fullclosing of spiracle 3.

Spiracle control in adult dragonflies: water balance 335

might take place through the central nervous system. If the latter holds, then changesin spiracular sensitivity might be reflected in changes in the frequency of motorimpulses to the closen Since the threshold to hypoxia as well as to carbon dioxidevaries with water balance, and since hypoxia has no direct action on the spiracle muscle(Miller, 1962), it seemed likely that the central nervous system was involved. Thefollowing experiments show this to be so, and evidence discussed later suggests that infact both central and peripheral mechanisms play a part in regulating spiracle move-ments with respect to the water balance.

100-

40-

S.O

20- |Oo

§

60 120 180

Time In 10% CC^ (mln.)

Fig. 4. Sympetrum striolatum. The percentage opening of spiracle 2 in 2 % carbon dioxideplotted against the time spent in a continuous stream of 10 % carbon dioxide. The experimentstarted with twelve dragonflies; after 90 min., one was dead; after 180 min. a further four weredead.

Table 1. Correlation of the frequency of motor impulses in the nerve to spiracle 2with the behaviour of the valve in 2 % oxygen after dehydration or desiccation

Species

Macromia reginac

Ictinogomphui ferox

Regime

DryWet

DryWetDryWet

Weight change(mg.)

-7o (-11%)+ 40 ( + 6%)

-60 (-7%)+ 75 ( + 8%)-90 (-9%)+ 35 ( + 4%)

Impulses/sec.10895

85435411

Valve activity

0-10 % open95-100% open

o-5 % open100% open

0-25 % open100% open

Ictinogomphus ferox and a large corduliid dragonfly, Macromia reginae Le Roi, weresubjected to dry and wet regimes as described in Expt. 1. The response of spiracle 2was tested in 2% oxygen; the spiracle nerve was then dissected out as quickly aspossible and a recording was taken. The results indicate that an increased frequencyof motor impulses accompanies desiccation while a decrease follows hydration; someexamples are shown in Table 1. This technique; however, is not wholly satisfactorysince time is not allowed for the insect to recover from the operation, and this mayaccount for the great variation in the results. In consequence an attempt was made toreproduce the conditions of hydration and desiccation by perfusion of the operateddragonfly with saline solutions of different concentrations.

22 Exp. Biol. 41, 2

336 P. L. MILLER

Experiment 3. The effect of varying the strength of the perfusing salineson the frequency of impulses to spiracle 2

Salines of the following concentrations were made up: 0-5,0-75, i-o, 1-5, 2-0, 2-5 and3-0 x normal. The appropriate solution was then perfused through the inverted insectby running it into the base of a prothoracic leg from a small pipette and allowing it toleave from an incision at the posterior end of the abdomen. In some experiments thesolution was added through an opening in the fronts but the site of perfusion did notaffect the results. Between 1 and 2 ml. saline was added before recordings were made,and more was pipetted in between readings. Although nearly all the haemolymph wasreplaced by saline, insects survived well for many hours. As an indicator of 'goodcondition' the ability to perform flight movements on pinching the abdomen was used.After more than about 2 hr. perfusion with 2*0 x normal saline, the condition began todeteriorate, but this was fully reversible on return to saline of normal strength. In3-0 x normal the dragonfly was in poor condition after 15-30 min., but again the effectwas reversible.

120-1

Saline

x 2 0

x1-0

xO-5

20

Time (sec)

Fig. 5. Ictinogotnpkus ferox. Histogram of the frequency of motor impulses in the two axonsto spiracle 2 in successive 200 msec, periods, with the insect perfused with double strength,normal strength and half strength saline. The cyclical rise and fall in frequency results fromthe activity of abdominal ventilation centres and both crests and troughs are affected by theperfusing saline.

Recordings from the spiracle nerve were taken at intervals during perfusion untilno further change was detected. The response was often slow to appear and might notbe complete in less than 30 min. Faster responses were obtained with strongersolutions. In 0-5 x normal saline the discharge was sometimes completely extinguishedafter long periods of perfusion. The insect showed no ill effects and spiracle closurecould still be evoked by stroking the valve hairs. This showed that the neuromuscularmechanism had not been blocked by the dilute solution, although spontaneous firingof the motor neurones had ceased. On return to normal saline firing was soon resumed.This has been repeated several times on the same preparation.

Spiracle control in aduh dragonflies: water balance 337

In Fig. 5 the frequencies in 0-5, i-o and 2-0 x normal saline are shown from the samedragonfly as consecutive readings every 200 msec. It can be seen that the high-frequency bursts, coinciding with expiration (Miller, 1962), as well as the troughs oflower frequency are affected by the perfusing saline. In Figs. 6 and 7 the effect ofswitching between 0-5 and 2-0 x normal salines is shown in .teneral and mature insects.In these preparations the connectives between the thorax and abdomen were cut sothat there was no interference from the abdominal ventilation centres and steadyreadings were obtained. After perfusion with a different saline, teneral insectsnormally achieve a new value more rapidly than mature specimens, and the changes intenerals are usually greater than in matures.

SalinexO-5 x20 xO-5 x2-0 x20 xOS x2O

40 60Time (mln.)

40 60Time (mln.)

Fig. 6. Ictinogomphus ferox. A, Teneral individual 7 days after emergence. The frequency ofmotor impulses to. spiracle 2 is plotted against time and the effect of switching between halfand double strength saline is shown. B, Teneral insect 1 day after emergence, as in A.

In Fig. 8 the results from many tests are summarized. In 0-5 x normal saline themean frequency from 34 tests is 31/sec. (s.E. + 2-10); in normal saline it is 43-5/sec.(S.E. ± 2-4) from 33 tests; and in 2-0xnormal it is 56-5/sec. (s.B. ± 1-95) from 31tests.

Experiment 4. The effect of varying the osmotic pressure of the solutionindependently of the ionic concentration

For this experiment 0-5 x normal saline was made isosmotic with i-o and 2-0 xnormal by the addition of 31 and 92 g./l. glucose respectively. In Fig. 9 the resultsof perfusing /, ferox with 2-0 x normal saline followed by 0-5 x normal and 0-5 xnormal + glucose are shown. The low frequency achieved in 0-5 x normal saline

338 P. L. MILLER

is maintained when this is changed to a solution containing excess glucose, but is re-placed by a high frequency when 2-0 x normal saline is again perfused through theinsect. In Fig. 10 comparable results are shown in which the order of perfusions ischanged; this experiment was performed on Phyllogomphus orientate, a gomphid ofabout the same size as / . ferox. A similar result has been obtained from six otherI. ferox. It seems likely, therefore, that the concentration and perhaps the ratios ofspecific ions in the haemolymph have an important regulatory function in controllingthe frequency of motor impulses to the spiracles, while the osmotic pressure plays nopart.

80-

x2070-

60-

iS.50-

5-

I 40-V

f 30-E

20-

10-

40 60 80 100 0-5 10

Time (mln.) Saline strength (log scale)

Fig. 7 Fig. 8

Fig. 7. Ictinogompkus ferox. Mature insects treated with salines of various strengths as in Fig. 6.

Fig. 8. Ictinogomphus ferox. The frequency of motor impulses in both axons to spiracle 2 inteneral and mature insects plotted against the strength of the perfusing saline. Vertical linesshow maximum and minimum values; horizontal bars denote standard errors.

2-0

Experiment 5. Localization of the site of the response to water balance

The mesothoracic ganglion of /. ferox is partially fused to the metathoracic ganglion.For the following experiments, therefore, Acanthagyna villosa (Gruenberg) (Aeshnidae)was used since in this species the ganglia are well separated. After cutting all thelateral nerves and the posterior and anterior connectives, the mesothoracic ganglionstill responds to'changes in the concentration of the perfusing saline (Fig. 11). Thispreparation is successful only when the ventral tracheal trunks supplying the gangliaare undamaged and the abdomen continues to ventilate them. Thus complete de-afferentation does not abolish the response and it may therefore depend on intra-

Spiracle control in adult dragonflies: water balance 339

ganglionic receptors or alternatively it may take place through a direct action oninterneurones or on motor cells. The possibility of additional receptors outside thecentral nervous system is not excluded but their presence seems unlikely.

xO-5 x20

10-

20 AOTime (mln.)

Fig> 9 Ictinogomphiu ferox. The frequency of motor impulses to spiracle 2 in a teneral insectplotted against time, with double strength, half strength, half strength + glucose (isosmoticwith double strength) (o'5 G), and double strength saline successively perfusing through thethorax and abdomen.

The removal of parts of the central nervous system may have certain long-lastingeffects on the discharge in the nerve to spiracle 2. These are summarized in Fig. 12,in which results from I. ferox and A. villosa are shown. The values given are all frominsects perfused with small amounts of normal saline. When the cord is intact thenerve to spiracle 2 of A. villosa passes impulses at a frequency two or three timeshigher than that in / . ferox. After decapitation this falls to less than half the originalvalue, the frequency now becoming similar to that in the nerve to spiracle 1, as alreadyreported (Miller, 1962). When the supra-oesophageal ganglion alone is cauterizedthere is no change and the sub-oesophageal ganglion seems to be responsible for themaintenance of a high frequency in the spiracle 2 nerve in this species. The possiblesignificance of this in dragonflies with a crepuscular flight period will be discussedelsewhere.

Decapitation oil. ferox produces no change in the frequency of impulses to spiracle 2(Fig. 12). Removal of the prothoracic ganglion in either species may cause a temporarydrop, but there is then usually a return to the value held before the operation. Aftercutting the central nervous system posterior to the mesothoracic ganglion, the

34° P. L. MILLER

discharge is unaffected except in so far as the removal of abdominal ventilatory centresmay abolish synchronizing bursts (Miller, 1962).

To summarize, in / . ferox removal of parts of the central nervous system anteriorto the mesothoracic ganglion has no appreciable effect on the discharge, whereas inA. villosa removal of the sub-oesophageal ganglion depresses the frequency to less thanhalf the value in the intact insect.

8O-1 Saline

x0-5G x20

2L 5 0 -

10-1

Time (mln.)

Fig. 10. Phyllogompkus orientalis. Details as in Fig. 9, with the order of perfusing solutions changed

DISCUSSION

Lengthy discussion of the results seems premature at present when the experimentalwork is incomplete. Experiments are now in progress on British species to determinewhich components of the physiological saline are important in setting the frequencyof the endogenous motor output. It is also hoped to compare by analyses of thehaemolymph of hydrated and desiccated insects the natural changes in ionic concentra-tions with those used in perfusing solutions.

The latency of the reaction is probably accounted for by diffusion through theouter layers of the ganglion. The time course described here is of the same order asTreherne (19626) found when he measured the rate of spike height reduction in thecord of the cockroach bathed in a solution containing 70 mM. K+/1. Although theneural lamella does not present a barrier to the diffusing ions a Donnan equilibriumis maintained across the sheath (Treherne, 1962 a). Hence, while the composition ofthe fluid in contact with the nerve cells may be very different from that in the haemo-lymph, changes in the latter will probably alter the position of the equilibrium and soaffect the environment of the neurones.

Spiracle control in adult dragonflies: water balance 341

Changes in the concentration of potassium and calcium ions are known to affectspontaneous activity in the central nervous system of the cockroach, calcium lacklowering the level and potassium excess (more than 20 mM./l.) raising it, this latterbeing antagonized by calcium (Roeder, 1948). Raised potassium level might actdirectly on the spiracle motor cells slightly depolarizing them and thereby increasingtheir rate of firing. That changes in the ionic composition of the haemolymph occurnaturally in some insects is clear from the work of Hoyle (1954, 1956; see also Shaw &Stobbart, 1963). If potassium concentrations are found to be significant for regulatingthe frequency in the spiracle nerves, then the high potassium level which Hoyle (1956)has found in moulting locusts may not only reduce movement in the insect by de-polarizing muscles, but also reduce water loss by increasing spiracle control at a timewhen feeding temporarily ceases.

x 1 0

40 50Tl/ne (min.)

Fig, 11. Acanthagyna villosa. The frequency of motor impulses in the transverse nerve fromthe mesothoracic ganglion with all lateral nerves and the posterior and anterior connectives cut.The insect is perfused successively with the following salines: i-o, i's, 2-o, i-oxnormalstrength.

However, while it is likely that changes in the frequency of motor impulses to thespiracle determine the threshold of the spiracular response to carbon dioxide, themechanism of the response to hypoxia is less clear. This depends on a ganglionicrather than a peripheral mechanism and will be the subject of discussion in a laterpaper.

The mechanism in the dragonfly is clearly different from that in the slug, where areduction of the osmotic pressure following dilution of the bathing medium causes an

342 P. L. MILLER

increase in the spontaneous activity of the central nervous system. As in the dragonfly,this result also can be interpreted functionally (Hughes & Kerkut, 1956; Kerkut &Taylor, 1956).

160-i

140-

120-

100-

g- 80-

«H

40-

20-

A

20 40 60 80

Time (mln.)

100 120

IOU-

140-

120-

100-

80-

60-

-

40-

-

20-

VO£

co-«,PS&Ouun>

O

1a

o

o

oo

•

• • •

•

0o •

• • o

1 1 1 1 1 1 120 40 60

Time (min.)

80

Fig, 12. Ictinogomphu ferox and Acanthagyna villosa. A, Frequency of motor impulses tospiracle 2 before and after decapitation. B, The same before and after section of the nerve cordbetween the prothoracic and mesothoracic ganglia. • , Acanthagyna villosa; A, A. villosa withsupra-oesophageal ganglion cauterized; O, Ictinogompkus ferox; A, / . ferox with supra-oesophageal ganglion cauterized.

Water balance may affect the spiracle closer muscle directly in addition to its actionthrough the nervous system. After denervation of the spiracle close to its musclethe valve sometimes shows autonomous activity by closing and responding to lowcarbon dioxide tensions in a way familiar in other insects (Beckel& Schneiderman, 1957;Hoyle, 1961). When removed from the insect and floated on saline it opens, but whenfloated on saline of 1-5 or 2-0 x normal strength it closes and still shows sensitivity tocarbon dioxide. These results are comparable with those of Hoyle (1961) from spiracle 2of the locust, where he showed that the haemolymph potassium produced contractureof the denervated spiracle muscle when its concentration was more than 30 mM./l.Although this mechanism has not been demonstrated to act in the intact preparation,indirect evidence suggests it is operative there too, and it was at first taken to be theonly one involved in the spiracular response to water balance (Miller, 1961). However,it now seems likely that there are two separate mechanisms: the first depends on theresponse of the ganglion to the concentration of one or more unidentified ions and thesecond on the action of high potassium levels on the closer muscle.

Spiracle control in adult dragonflies: water balance 343

SUMMARY

1. Dragonflies caught in the wild display a marked variation in the degree of controlexercised over their spiracles.

2. In the laboratory desiccation produces tighter control and hydration loosercontrol of spiracle 2: that is, in a partially desiccated insect the thresholds of thespiracular responses to carbon dioxide and to oxygen lack are raised.

3. In desiccated insects the frequency of motor impulses to the spiracles ishigher than in hydrated individuals. These effects can be reproduced by perfusionwith physiological salines of various strengths.

4. The reaction does not depend on the osmotic pressure of the solution but on theconcentration of one or more of its constituents.

5. The isolated mesothoracic ganglion is able to mediate this reaction.

I am most grateful to Mr Jonathan Kingdon for assistance in the capture of dragon-flies. My thanks are due to the Tropical Medicine Research Board for the award of agrant for the purchase of apparatus.

REFERENCES

BECKBL, W. E. & SCHNEIDERMAN, H. A. (1957). Insect spiracle as an independent effector. Science,136, 352-53-

BURSELL, E. (1957). Spiracular control of water loss in the tsetse fly. Proc. R. Ent. Soc. Lond. A, 33,21-9.

HOYLE, G. (1954). Changes in the blood potassium concentration of the migratory locust (Locustarmgratoria imgratorioides R. and F.) during food deprivation and the effect on neuromuscular activity.J. Exp. Biol. 31, 260-70.

HOYLE, G. (1956). Sodium and potassium changes ocurring in the haemolymph of insects at the timeof moulting and their physiological consequences. Nature, Lond., 178, 1236—7.

HOYLE, G. (i960). The action of carbon dioxide gas on an insect spiracular muscle. J. Ins. Pkytiol. 4,

HOYLE, G. (1961). Functional contracture in a spiracular muscle, jf. Int. Pkytiol. 7, 305-14.HUGHES, G. M. & KERKUT, G. A. (1956). Electrical activity in a slug ganglion in relation to the con-

centration of Locke solution. J. Exp. Biol. 33, 282-94.KBRKUT, G. A. & TAYLOR, B. J. R. (1956). The sensitivity of the pedal ganglion of the slug to osmotic

pressure changes. J. Exp. Biol. 33, 493-501.MELLANBY, K. (1934). The site of water loss from insects. Proc. Roy. Soc. B, 116, 139-49.MILLER P. L. (1961). Spiracle control in dragonflies. Nature, Lond., 191, 623.MILLER, P. L. (1062). Spiracle control in adult dragonflies (Odonata). J. Exp. Biol. 39, 513-35.RAMSAY, J. A. (1935). The evaporation of water from the cockroach. J. Exp. Biol. 12, 373-83.ROEDER, K. D. (1948). The effect of potassium and calcium on the nervous system of the cockroach,

Periplaneta americana. J. Cell. Comp. Physiol. 31, 327-38.SHAW, J. & STOBBART, R. H. (1963). Osmotic and ionic regulation in insects. In Advances in Insect

Physiology, pp. 315-99. Ed. J. W. L. Beament, J. E. Treherne and V. B. Wigglesworth. AcademicPress.

TREHERNE, J. E. (1962a). Distribution of water and inorganic ions in the central nervous system of aninsect (Periplaneta americana). Nature, Lond., 193, 750-2.

TREHERKE, J. E. (19626). Some effects of the ionic composition of the extracellular fluid on the electricalactivity of the cockroach abdominal nerve cord. J. Exp. Biol. 39, 631—41.

WiGGiBSWORTH, V. B. (1953). The Principles of Insect Physiology, 5th ed. London: Methuen.