Unit Activity in the Abdominal Nerve Cord of a Dragonfly Nymph

During recent years, several investigations have analysed the organization of invertebrate central nervous systems in terms of the activity of single units. Use has been made of micro-electrodes to record selectively from such units in the crayfish (Kennedy & Preston, 1960) and in Aplysia (Hughes & Tauc, 1962), similar techniques have not been used in insects, except in special cases such as the corpora pedunculata (Maynard, 1956), optic lobes (Burtt & Catton, 1960) and motor neurones of the sound-producing organs of cicadas (Hagiwara & Watanabe, 1956). The technique of recording from small bundles of axons split off from the inter-ganglionic connectives has not been extensively applied to insects and therefore seemed a profitable one to explore. In some ways this method has proved to have advantages over the micro-electrode technique for a general analysis of neuronal connexions especially in the crayfish, at first in the circumoesophageal commissures (Wiersma, Ripley & Christensen, 1955; Wiersma, 1958) and later in the connectives of the abdominal nerve cord (Hughes & Wiersma, 1960; Wiersma & Hughes, 1961). In applying it to insects, however, several difficulties are encountered which were not so serious in the crayfish, notably due to the smaller size but also due to differences in their respiratory mechanisms. Many investigators of the insect central nervous system have pointed out the importance of maintaining its tracheal supply if normal activity is to persist. This supply is profuse in most insects and any attempt to dissect the connective is bound to rupture some of these structures. In addition, insects are unable to renew the oxygen contained in the main tracheal trunks when immersed in physiological saline. The dragonfly nymph possesses advantages over many insects with respect to this problem in that more normal activity can persist because gaseous exchange continues at its gill. Further advantages of this insect are that work has already been done on the histology (Zawarzin, 1924) and physiology of the central nervous system (Hughes, 1953 ; Fielden, 1960) and of the abdominal proprioceptors (Finlayson & Lowenstein, 1958).

The present paper describes different patterns of activity which have been encountered in a study of single units in the abdominal connectives of the dragonfly nymph. It has already become clear that many of the interneurones have properties similar to those found in the central nervous system not only of other invertebrates but also of vertebrates.

With few exceptions last instar nymphs of Anax imperator were used throughout this study. The nymph was pinned ventral side uppermost in a dissecting dish containing physiological saline (7·5 g. NaCl; 0·1 g. KC1; 0·2 g. CaCl2; 0·2 g. NaHCO3; distilled H20,1000 ml.). This solution was found to give best survival of the preparations. Most preparations were made between the fifth and sixth ganglia but in other cases the connectives in front of the last ganglion were prepared. A small portion of the sternum was removed and the ventral nerve cord exposed under a dissecting microscope. The connectives were separated from one another, and for the finer dissection electrolytically-tapered tungsten needles were used. These were made by pulling out tungsten wire in an oxygen/coal-gas flame and the sharpening was completed by passing 6 v. a.c. between the needle and a large copper electrode immersed in strong (20%) potassium hydroxide solution. Dissection of the connective was not easy because of its small size, both in external diameter and in the diameter of the individual nerve fibres. Thus although the total cross-section of a connective contains approximately 1200 fibres in both the crayfish and the dragonfly nymph, the size range in the latter is only up to 17 μ (Hughes, 1953), whereas it was only fibres above this size that Wiersma & Hughes (1961) considered they were able to establish with any certainty. De-sheathing of the connective made later stages of the preparation more readily manageable, but it was not always possible. The connective was split repeatedly into smaller bundles, and the direction of the split was indicated on a diagrammatic cross-section. As yet, however, precision of localization has not been great and no conclusions have been based upon it.

A platinum-recording electrode was hooked beneath a small bundle of fibres which was raised above the surface of the bathing saline containing the indifferent electrode. Activity of individual units could be readily distinguished when the bundles were sufficiently small. Recording was usually carried out with the bundle retaining both its anterior and posterior connexions, but later one of these was cut and recording made from either the anterior or posterior lead. Stimulation of the animal was carried out mechanically by means of fine paint brushes for stimulating hairs or by means of needles for applying more localized stimulation to spines, etc. Proprioceptive stimulation by compression or by telescoping different parts of the abdomen was not easy because of the smallness of the preparation and the difficulty of maintaining constant recording conditions while performing such stimulation. The whole question of the localization of the area of stimulation was much more complex than in the crayfish partly because of the size of the animal but also because of the difference in sensory innervation. The impulses were displayed on a cathode-ray oscilloscope having been amplified by a Tektronix pre-amplifier and were heard simultaneously through a loudspeaker unit. The viability of these preparations at room temperature was fairly good and most retained activity for at least 3 hr. The results were obtained from about forty preparations.

The types of unit activity recorded in a connective may be classified in several different ways. The axons present are mainly either interneurones, or primary sensory fibres which enter through one of the segmental nerves and pass up or down the cord before ending in the neuropile. A further possibility is that some are motor neurone collaterals ascending or descending in the connective before leaving in a segmental nerve from a ganglion other than that containing the cell body. This type, however, is not well known in insects and the other two types can be divided ing to the length of the intracentral pathway. Some of the interneurones described by Zawarzin (1924) pass from the brain to the last abdominal ganglion, whereas others have much shorter pathways. Another feature which may be used in their classification is the particular modality of stimulation to which they respond. However, the most obvious feature of a fibre which one observes on first recording from it is whether it has a resting discharge or not. These discharges may be present for a variety of reasons. It is well established that many sense organs have resting discharges, and evidently some of the activity recorded in the connective was in fibres from such endings. Other units showing a resting discharge were interneurones receiving tonic input from similar sources. Both sorts of unit will be affected by peripheral stimulation, but some interneurones remained apparently unaffected by the types of stimulation used here. Such fibres are here designated ‘spontaneously active’, the assumption being made that their resting discharge would persist even when the central nervous system is completely isolated from the animal which has been shown to be true in some cases.

Units showing this type of activity, completely unaffected by any external stimulation, vary a great deal in the particular form of their resting discharge. This was independent of the age or condition of the preparation and ranged from continuous regular spikes to random firing and intermittent repetitive bursts (Fig. 1). All types of spontaneous activity were found propagating in both ascending and descending directions and this activity was often shown to persist when a single ganglion was completely isolated from the rest of the nervous system. In some cases the intervals between spikes were relatively constant and the discharge persisted for a very long time, usually at quite low frequencies (2–20 per sec.). Many spontaneously active units, however, were characterized by irregularities in their discharge pattern which Kolmodin (1957) and Hunt & Kuno (1959) noted as a common feature of spinal interneurones. This irregularity can be seen in Fig. 1 a and in Fig. 2 where the interval between successive impulses is plotted and shows a wide range though the overall frequency remains fairly constant. In this, and many other cases, a fairly normal distribution curve could be obtained about the mean frequency, though in other cases the discharge appeared more random. This unit was quite unaffected by mechanical stimulation which had a very marked effect on another unit in the same bundle.

Units discharging intermittently (Fig. 1,c, d) or tending to produce bursts of impulses of higher frequency were very common. These frequency-modulated bursts were not related to any overt breathing or heart rhythms and again appeared to be endogenous to the central nervous system. The frequency of the bursts covered a fairly wide range (10-30 per min.). In some instances the pattern of each burst was remarkably constant and often showed a high initial frequency which decayed exponentially during the burst (Fig. 3). The presence of a repeating pattern in the discharge of a single neurone was sometimes found to be true of several neurones which fired more or less in step with one another and gave rise to a more complex overall pattern of discharge (Fig. 1c). Such activity strongly suggests the presence of some interneuronal connexion between the units or possibly an even more direct one.

Many fibres which showed a resting discharge were affected by stimulation of the animal. These are discussed in the next section along with those fibres which did not have such discharges but only became active when the preparation was stimulated. Differences between these two types are presumably due to the type of sensory fibre or, if an interneurone, partly to the sensory input to which it is subjected and partly to its properties of adaptation.

Mechanical stimulation, whether of proprioceptive or tactile endings, often proved difficult to localize because of the small size of the animal. In addition, stimulation was sometimes provided by movements of the animal itself, especially those due to respiration which in many cases persisted a long time after the preparation was made. Units excited by these movements were more common in recordings made between the last two ganglia, suggesting that not many ascending respiratory pathways are present above the level of these ganglia. Some units showed a definite phase relationship to the sequence of movements of the abdomen, paraprocts and anal sphincter which result in respiratory movements and/or jet propulsion. No doubt some of these discharges are in the so-called ‘giant fibres’ in which impulses have been recorded in the intact connectives (Hughes, 1953; Fielden, 1960). In some preparations it was difficult to distinguish between effects produced directly as a result of mechanical stimulation and those which arose secondarily because of reflex movements produced by stimulation of some part of the nymph. These difficulties were more apparent in fresh preparations which tended to become more quiescent as time elapsed.

The units which were direct responses to touch of hairs or spines or to proprioceptive stimulation could be divided into primary sensory fibres and interneurones. In the crayfish the distinction between these two types of response is fairly clear. The frequency of discharge is higher from primary sensory fibres which are affected by stimulation of more localized areas. In insects this distinction is not so easy because axons derived from many sensory endings, sometimes from quite widespread areas, join together to form a single afferent fibre which enters through one of the segmental nerves and many pass up or down the cord. Nevertheless, in many cases responses from a very localized region of the animal were recorded which showed a fairly regular high-frequency discharge and these were clearly in primary sensory fibres. The size of the spikes recorded tended to be smaller and more rapidly adapting in most cases and they could be evoked even by stimulation of a single spine or group of hairs. Responses to compression of a segment in either the dorso-ventral or the longitudinal direction produced clear proprioceptive responses presumably from end organs of the types described by Finlayson & Lowenstein (1958). These were generally fairly slowly adapting responses in comparison with those from the spines. Units with and without a resting discharge responded to proprioceptive stimulation and occasionally burst activity could be attributed to them (Fig. 4). Whether or not these responses were in primary fibres has not been clearly established, although this is definitely the case in the crayfish. In the dragonfly nymph but not in the crayfish, however, some responses to abdominal proprioceptive stimulation were obtained on the side opposite to the connective from which recordings were being made and across several segments. This suggests the presence of an interneurone but it is not impossible that some primary fibres pass right across the ganglion and ascend on the contralateral side, for this has been noted for tactile receptors also. In general primary sensory responses were obtained only from segments innervated by the two ganglia adjoining the dissected bundle, but instances have been found which indicate that some primary sensory fibres traverse at least one ganglion. As in the crayfish, the size of the spikes recorded from proprioceptors tended to be much larger than those recorded from primary sensory hair fibres. Spike sizes also varied considerably among units identified as interneurones. Responses from hair fibres of this type were smaller than the rapidly adapting responses obtained from the spines on the posterior lateral edge of the last three abdominal segments (Fig. 5).

Many units identified as interneurones had a resting discharge but this was by no means universal. These interneurones showed excitation or inhibition on stimulation and were often characterized by a post-excitatory depression of their resting discharge (Fig. 6). Response of these and silent units was, in general, not so regular or high in frequency as that obtained from many sensory fibres and appeared more slowly adapting. Occasionally, however, interneurones responded with only a few impulses to tactile stimuli; the large units excited by stimulation of the lateral spines on the last few segments (Fig. 5) were characteristically of this type. Besides these low, often irregular frequencies, interneurones were characterized by the presence of more ‘central’ properties. They tended to be more labile and were sometimes inhibited by one type of stimulation but excited by another. These and cases of inhibition of one unit with concurrent excitation of a second provide examples of the complexity of the integration that can occur at the interneuronal level (Fig. 7). Some interneurones showed a trigger type of response in which a single stimulus, so long as it reached a threshold level, fired off a train of impulses in a given unit. The pattern of this discharge remained fairly constant following stimulation of many different parts of the body and for varying suprathreshold intensities. Other types of after-discharge were also found many times and these possibly suggest the presence of second-order interneurones.

The great majority of interneurones investigated responded to a single modality only, although some units responding both to proprioceptive stimulation and movement of tactile hairs have been identified. Responses to vibration have also been observed. One of these fibres was also sensitive to touching of the thoracic legs and its response was of the trigger type. This unit responded to taps on the bench but was not responsive to airborne sounds (Fig. 8). It was noted that with frequent application of the stimulus there was a decline in the response. When the taps were made very frequently (to per min.) the unit, following an initial excitatory phase, became quite inactive for a certain period and then its activity returned again for a brief time; and this pattern was repeated so long as the stimulation was continued. This ‘adaptation’ of the central interneurone is reminiscent of the adaptation reported by Pumphrey & Rawdon-Smith (1937) for the giant fibres ascending from the last abdominal ganglion of the cockroach. Such a response clearly involves a complex situation which requires more careful control of stimulating conditions before its precise mechanism can be established.

Interneuronal responses have been observed from varying areas of the abdomen and anal paraprocts and some were found from the thorax, legs and head. It is therefore possible to recognize units conducting spikes in ascending and descending directions with respect to the lead. This assisted in the identification of interneurones on the basis of their peripheral sensory representation. Fig. 9 is a ventral view of the abdomen showing areas which, when stimulated, excited interneurones that have been found several times. It can be seen that the majority of units found so far are ipsilateral, i.e. the sensory field to which they respond occurs on the same side of the body as the connective in which they run. A relatively large proportion (20%), however, responded to contralateral stimulation. In ipsi-, contra-, and bi-lateral categories clear interneuronal responses have been obtained from receptive areas ranging from only a single abdominal segment to several segments. Unisegmental responses were localized to varying degrees. Many ipsilateral hair fibres responded only to dorsal or ventral stimulation and others only to stimulation of the lateral spines or of the hairs near the developing spiracles. Several of these localized responses were also found following contralateral or bilateral stimulation. An example of a bilateral fibre responding to rather discrete areas of one segment is shown in Fig. 5, together with the larger unit excited by stimulation of the ipsilateral spine. Ventral hair fibres appear to be more abundant than in the crayfish where the ventral surface is sparsely represented but this may be the result of differences in dissecting technique Many examples were found of interneurones excited in turn by stroking the lateral edge of consecutive segments on the same side of the animal. Occasionally such unisegmental units could be identified in the same dissected bundle following stimulation of three of four contiguous segments, confirming the impression that in some cases similar fibres run together in the cord as in the crayfish. Unlike the crayfish, however, the dragonfly nymph shows no clear evidence of a difference between the external segmentation of the body and the segmentation of its nerve supply.

The complexity of interneuronal responses was specially well illustrated in those occurring in multisegmental fibres. A wide variety of response patterns has been recorded and a considerable range of receptive areas found, ranging from two segments to the whole of one side of the animal. Multisegmental responses almost invariably came from homologous areas of the segments. Some areas were represented by several fibres and it is clear that stimulation of any part of the abdomen excites a very large number of interneurones which transmit information of varying degrees of specificity.

An understanding of central integration requires a knowledge of the morphological relationships of the neurones. Among insects most is known about the dragonfly nymph as a result of the methylene blue studies of Zawarzin (1924), but even for this insect the relationships are difficult to establish because of the complexity of the neuropile. From histological work it is possible to show the paths of sensory neurones and to some extent of motor neurones, but the connexions of interneurones are the most difficult to decide Physiological methods are not only essential for the study of functional properties of central neurones, but are also of great value in unravelling structural features of the central nervous system. For sensory fibres the techniques used here have confirmed the histological demonstration of tracts which enter one ganglion and pass up or down the cord, usually on the same side, but sometimes after having crossed to the opposite side. Evidence for the latter type of arrangement has been found for proprioceptive and tactile fibres and contrasts with the crayfish cord where primary sensory fibres appear to be homolateral. Hughes & Wiersma (1960) have drawn attention to the value of ganglionated nerve cords as material for the study of the integrative properties of interneurones, because it is often possible to record between two different sites where they can be affected synaptically. Such interneurones which integrate inputs over several segments of the body were designated ‘type C’ and this type of connexion has been shown many times for internen-rones of the dragonfly nymph. As was pointed out for the crayfish, these interneurones make possible a considerable amount of convergence of different sensory inputs. This will result in summation or occlusion depending on the relative frequencies of the spikes initiated in the different ganglia. Inevitably the information conveyed to any higher centre by such interneurones is less than that carried by the equivalent number of ‘private lines’. Nevertheless, an interneurone of this type, if it has efferent connexions with motor neurones concerned in startle reactions, will fulfil its function more economically and perhaps more efficiently, especially if it is fired at a lower threshold. This sort of economy of neurones in arthropods was stressed by Wiersma (1952) with reference to the lateral giant fibres of the crayfish which have efferent connexions in the segmental ganglia. Roeder (1959) has emphasized the decrease in quality of information transmitted by the giant fibres of the cockroach which has been sacrificed in favour of greater speed of conduction resulting from their increased diameter. There is no evidence that the fibres with type C connexions observed in the dragonfly nymph are those of largest diameter, and it is evident that the sacrifice in information is here not necessarily compensated by an increase in speed of conduction. The presence of synaptic regions in different ganglia makes it possible that these interneurones may integrate purely local inputs provided spikes are not initiated which spread to other ganglia. Mechanisms whereby this might occur have been described for Aplysia neurones (Tauc & Hughes, 1961, 1962).

Quantitative data about the actual numbers of arthropod central neurones is not very great. Zawarzin (1924) estimated that in Aeschna nymphs each abdominal ganglion contains 90 neurones and each thoracic ganglion 150. These estimates were based upon methylene blue preparations of whole ganglia in which it is doubtful if all the cells would have stained. Counts made from silver-stained serial sections of Anax (Fielden, unpublished) have given much higher figures of 600 cells for a typical abdominal ganglion and twice this number for the last ganglion. This number compares well with that of Wiersma (1957) who counted about 500 cells in each abdominal ganglion of the crayfish. It is noteworthy that the number of fibres in a connective is approximately 1200 for each animal. Relative to the vertebrate, then, there is no doubt that fewer neurones are present in the central nervous system. To the economy effected by the small number of motor neurones and the ‘push-button’ action of giant fibres, may be added the possibility of interneurones with synaptic contacts in many ganglia having areas functioning independently.

The occurrence of multisegmental fibres is not the only similarity between the organization of central interneurones in the dragonfly and in the crayfish. Similar patterns of central representation have been found and the proportion of contralateral fibres is approximately the same (20%). As yet, however, no asymmetric fibres excited by stimulation of unequal areas on the ipsilateral and contralateral sides have been observed in the dragonfly. In both animals a given sensory area is represented centrally by many interneurones. Conversely interneurones vary in the extent of the peripheral areas which they represent and from this convergence there arises the possibility that the site of stimulation may affect the nature of the response in a given interneurone. Many of the interneurones have a resting discharge which shows irregularity in the pulse interval and this appears to be a common feature of interneurones in the cat, crayfish, Aplysia and the dragonfly nymph. The presence of multiple representation may be necessary for the recognition of significant signals in such ‘noisy lines’.

Interneurones with resting discharges that are unaffected by any peripheral stimulus have been described here as spontaneously active fibres. Activity of this type in whole animal preparations has also been observed in the crayfish circumoesophageal commissure (Wiersma, 1958) and in the abdominal ganglion cells of Aplysia. As was pointed out for Aplysia (Hughes & Tauc, 1962) it is difficult to be certain that such neurones are completely unaffected, for it may be that they only respond to more complex patterns of stimulation than are normally employed. Even so they must maintain a more or less constant level of output at their efferent connexions which may be important in the functioning of other neurones. As yet nothing is known about efferent connexions of spontaneously active interneurones or of any of the other interneurones studied. These may be directly on to motor neurones or the next stage in the integrative mechanism. It would certainly be of interest to know if the multisegmental fibres act as relaying centres for motor neurones as well as receiving centres from several segments and possibly electrical stimulation of central bundles will clarify this point.

One of us (A.F.) would like to thank the Medical Research Council for their financial support during the course of this work.