1. The roots of the segmental nerves in nymphs of Anax imperator originate from separate dorsal and ventral tracts in the ganglionic neuropile.

  2. Axons forming the dorsal part of the nerve root can sometimes be traced to ganglion cells and tend to be large and thick-walled compared with the ventral axons which are smaller and thin-walled.

  3. In the roots of the fifth nerves of the last ganglion the two parts can be separated by dissection. Recording from each part under various conditions of stimulation shows that sensory activity occurs predominantly in the ventral part of the nerve root whilst motor spikes are recorded almost entirely from the dorsal part.

  4. It is concluded that there is a functional localization of motor and sensory fibres in the root of an insect nerve comparable to that in the dorsal and ventral roots of vertebrate nerves.

The roots of insect peripheral nerves have been described by some authors as originating from two sites within the segmental ganglia. Furthermore, the nerves show a certain distinction into dorsal and ventral portions, the former tending to be composed of large, thick-walled axons, the latter, smaller thin-walled axons. It has been suggested, on this structural evidence, that the dorsal part of the nerve root is composed largely of motor fibres whilst the ventral portion is formed from sensory fibres (Hilton, 1911; Zawarzin, 1924; Wigglesworth, 1959). Some circumstantial evidence from behavioural experiments has also been interpreted as supporting this suggestion but there has not been any direct physiological demonstration of this functional localization. Histological work on the abdominal ganglia of nymphs of Anax imperator has shown that the nerves have two origins in the dorsal and ventral neuropile and has indicated that it might be possible to separate these two parts by dissection and record the electrical activity of each. Zawarzin (1924) described these two origins of motor and sensory fibres in Aeschna nymphs and went so far as to draw an analogy with the clearly distinguishable roots of vertebrate spinal nerves.

These observations, combined with the good viability of dissected fibres of the dragonfly nervous system (Fielden & Hughes, 1962), suggested an investigation of the function of the two portions of the nerve root. The present paper describes the origin of a typical nerve as shown by histological methods and an examination of the functional localization in the two dissected parts using various combinations of stimulating and recording conditions. The physiological results support the conclusions drawn from the structural differences demonstrating that the two parts of the nerve are concerned with separate functions.

Ventral dissections were made of last instar nymphs of Anax imperator immersed in physiological saline. For the histological investigation pieces of nerve cord were pinned directly on small wax blocks and immediately transferred to fixative. Preparations were fixed in aqueous and alcoholic Bouin, Carnoy or mercuric formol for silver staining and in buffered 1 % osmium tetroxide for the ethyl gallate method (Wigglesworth, 1957). Satisfactory results were obtained with modifications of Holmes’s (1943) and Peters’s (1955) silver techniques which were used, in conjunction with the osmium-ethyl gallate method, for a general investigation of structure. Fibre analyses in the segmental nerves were made from material fixed in saturated picric acid plus o·2% osmium tetroxide and subsequently stained with iron haematoxylin. The structure of the nerve roots was reconstructed from serial drawings and photomicrographs of sections of the ganglia cut in transverse, vertical, and horizontal planes.

The most convenient nerves for the purpose of dissection are the fourth and fifth pair of the last abdominal ganglion. These extend posteriorly from the ganglion instead of laterally and hence their structure facilitates dissection and recording. Fine dissection of these nerves was carried out using electrolytically tapered tungsten needles (Fielden & Hughes, 1962). The nerves were de-sheathed and dissected horizontally over small pieces of exposed film in situ. Because of the small size of both the nerve and its component axons (less than 12μ) the dissection was always approached from a point as near to the ganglion as possible. Electrical activity was recorded with a small platinum electrode hooked beneath the dissected part of the nerve which was raised above the surface of the saline containing the indifferent electrode. Stimulating electrodes were made of silver/silver chloride through which square wave pulses were passed, ranging from o· 1 to 0· 5 msec, duration. Impulses were amplified by a Tektronix pre-amplifier and displayed on a cathode-ray oscilloscope. They were heard simultaneously through a loud-speaker unit.

Structure and origins of the segmental nerves

The pathways of fibres from the segmental nerves were mainly studied from sections of the last abdominal ganglion of the Anax nymph. The innervation of the terminal abdominal segments from the five paired nerves of this ganglion has been described previously, when it was shown that all the nerves are composed of sensory and motor fibres except the third (N3) which is purely motor in function (Fielden, 1960). Three pairs extend laterally from the ganglion whilst nerves 4 and 5 run posteriorly and are easier to dissect into dorsal and ventral portions. The origins of these nerves are therefore described more fully since they were used for the physiological work. Silver and osmium-ethyl gallate staining methods do not enable single fibres to be traced as easily as in the methylene-blue studies of Zawarzin (1924) or the Golgi preparations of Guthrie (1961). Hence the present description is largely concerned with tracts of fibres, both to and from the peripheral nerves, which could be followed in the different planes of serial sections.

A typical abdominal ganglion consists of a dense central neuropile surrounded peripherally on lateral and ventral sides by closely packed ganglion cells and invested by a tough fibrous connective tissue sheath (Text-fig. 1 a). The ganglion cells tend to be grouped between the origins of the paired nerves and are of two main types: large (35–45 μ, presumably motor neurones) and small (10–15 μ, association neurones). A few large ganglion cells also lie mid-dorsally above the neuropile in the last ganglion. The axons from the larger cells initially have a diameter of 2–4 μ, but narrow down to about 1·5 μ on entering the neuropile where they often form distinct vertical tracts. The axons enlarge again as they form motor fibres in the nerve roots, a feature also noted by Wigglesworth (1959) in his work on the central ganglia of Rhodnius. In addition to those from the ganglion cells, tracts can be traced running in various directions through the central neuropile. These have been investigated in some detail and different regions of the neuropile have been identified but this paper is restricted to the major tracts of the nerve roots.

Text-fig. 1.

Diagrammatic representations of sections through the last abdominal ganglion of Anax nymph, (a) Transverse section through the origin of a segmental nerve showing the position of the major ganglion cells and fibre tracts, (b) Longitudinal section showing the main ganglion cell groups and dorsal and ventral tracts to and from the segmental nerves on the left. ----, Dorsal tracts; ——, ventral tracts; N1-N5, segmental nerves; conn., connective.

Text-fig. 1.

Diagrammatic representations of sections through the last abdominal ganglion of Anax nymph, (a) Transverse section through the origin of a segmental nerve showing the position of the major ganglion cells and fibre tracts, (b) Longitudinal section showing the main ganglion cell groups and dorsal and ventral tracts to and from the segmental nerves on the left. ----, Dorsal tracts; ——, ventral tracts; N1-N5, segmental nerves; conn., connective.

Sections of a nerve as it approaches the ganglion show its characteristic division into dorsal and ventral parts (Pl. 1 a). The larger, thick-walled axons lie predominantly in the dorsal half while the ventral half is composed of smaller thin-walled axons. These groups of fibres have different origins within the neuropile and are comparable to those which in other insects are considered to be motor and sensory in function (Hilton, 1911; Zawarzin, 1924; Wigglesworth, 1959; Guthrie, 1961). The axons occupying the ventral part of the nerves are difficult to trace after they enter the ventral neuropile. Some fibres run in the direction of the connectives and into the mid-part of the ganglion on both ipsilateral and contralateral sides (Text-fig. 1a, b). Others can be traced from the ventral tracts to the more posterior part of the neuropile. Single sensory neurones with branches running in comparable directions have been described by Zawarzin (1924) in the ganglia of Aeschna. The dorsal, motor, fibres enter the nerve from the dorsal part of the neuropile and can occasionally be traced to the larger ganglion cells via distinct vertical tracts. They also give off numerous branches into the neuropile. The paths of some of these dorsal fibres could be traced from fibre tracts in the connectives and in particular from the more posterior part of the last ganglion.

The fourth and fifth nerves of the last ganglion show the differences between the dorsal and ventral parts of a nerve root particularly well. Pl. 1 a is a photomicrograph of a section through these nerves as they approach the last ganglion between the groups of ganglion cells. Both nerve roots show the different types of axon in the two parts. These axons were counted, in more peripheral parts of the nerves, from highly enlarged photomicrographs of osmium-fixed material stained with iron haematoxylin. N4 is the smaller of the two paraproct nerves and it comprises four large axons (10 – 14 μ), fourteen to fifteen smaller axons (5 – 10 μ) and approximately 180 axons below 5 μ diameter, as seen under the light microscope. The larger axons lie dorsally and have thicker walls which often appear double due to the enveloping mesaxon (Pl. 1b). The fourth nerve arises more anteriorly in the ganglion than does the fifth. It first becomes distinguishable as a separate fibre bundle which crosses the mid-part of the ganglion diagonally and then turns to run in a posterior direction parallel with N5 (Text-fig. 1 b). This ventral tract is the most prominent part of this nerve. The latter enters the ganglion on the dorsal surface and then turns ventrally inwards to join the ventral fibres of N5 described above. These combined tracts form a distinct longitudinal bundle in transverse sections. The fibres branch in both medial and lateral neuropile and it was impossible to tell from the methods employed in the present study whether any of them pass through to the connectives without synapsing. Only degeneration experiments comparable to those of Hess (1958) on the cockroach will show this conclusively. The dorsal fibres of N4 are less distinct and they become identifiable only after the ventral tract has bent medially into the ganglion. Motor fibres from ganglion cells from groups between N 2 and N4, N4 and N 5, and the two N5’s can be traced to this root.

The fifth nerve is the largest nerve of the last ganglion and is a mixed nerve supplying part of segment 9, segment 10, the anal appendages and the respiratory chamber. It is composed of two large axons (10 – 14 μ), twelve to fourteen smaller axons (5-10 p) and approximately 400 axons between 1 and 5 μ in diameter. The small fibres are largely ventral, while the larger thick-walled fibres are dorsal. The dorsal bundle shows some distinction into thin-walled and thick-walled motor axons, as described by Wiggles worth (1959) in Rhodnius, since the larger axons lie more laterally and tend to lose their thick sheaths as they enter the ganglion (Pl. 1 a). The dorsal fibres appear to be derived from cell groups between the segmental nerves, and both ipsilateral and contralateral processes have been traced. No very prominent tracts can be delimited dorsally but fibres have been traced from the mid-anterior region of the neuropile to the fifth roots. It has been mentioned that the ventral fibres of N5 form a prominent longitudinal tract running through the ganglion. This tract has many branches in the neuropile, particularly in areas believed to be largely composed of association fibres. None of the larger fibres, either of the connectives or paraproct nerves, could be traced through the ganglion, and these observations are supported by the experimental work where it was shown that at least the larger fibres synapse in the last ganglion (Fielden, 1960).

To summarize, at its origin each nerve shows a distinct division into dorsal and ventral parts largely composed of axons of different sizes and structures. These axons can be traced to separate areas of the neuropile and sometimes those of the dorsal tracts can be traced to ganglion cells. The large size of the fifth root and its clear division into two parts enables the function of these parts to be investigated. The methods by which this was achieved are described below.

Localization of function in the root of the fifth nerve

The fifth nerve has a diameter in the order of 300– 400 μ. as it leaves the ganglion. A fine tungsten needle was inserted horizontally through the sheath and the nerve was split longitudinally. This was done most easily close to the ganglion and normally in this position the split divided the nerve into two more or less equal parts. Some preparations were sectioned and stained with iron haematoxylin following dissection and recording. These sections showed that in only relatively few cases did the split separate the two bundles but in most experiments it did make an approximate division between the two types of fibre described histologically. The fibre types obviously mix as the nerve becomes more peripheral, and it is only at its origin that the dissection tended to split the nerve between the bundles. It was occasionally possible to record from the two portions of the nerve simultaneously but this proved difficult owing to the small size of the preparation. For this reason records from each half were usually taken consecutively, using the same source of stimulation. Various stimulating inputs were used in order to deduce the function of the two parts and to eliminate the effects of antidromic stimulation which might confuse the results. The methods of stimulation used and the results obtained are described below and illustrated in Fig. 2. The ganglion was de-afferented except for the nerves under stimulation.

(1) Mechanical stimulation of the tactile hairs bordering the anal appendages

The paraprocts and epiproct are edged by tactile hairs whose afferent fibres are contained in N5. Movement of these hairs by a fine needle or brush elicits the characteristic evasion response of the dragonfly nymph and it has been shown that the afferent fibres can excite efferents of the same nerve root (Fielden, 1960). Hence, if the deductions from the histological work are correct, stimulation of the paraproct hairs should show that small sensory impulses occur largely in the ventral portion of the nerve and larger motor spikes in the dorsal portion. This was found to be the case and the origins of these spikes were checked by cutting the portions of the nerve proximally and distally to the electrodes respectively (Pl. 2d, e). Sensory discharges were only recorded in the distal ventral part and efferent spikes only in the proximal dorsal part. The latter were observed only if the rest of the nerve was kept intact.

(2) Electrical stimulation of the whole nerve

The reflex stimulation of efferent fibres by afferents of the same root has also been shown to occur when the nerve is stimulated electrically. In the present experiments when stimulating electrodes were placed on the fifth nerve peripheral to the dissected part a compound action potential occurred in the ventral portion of the nerve and a burst of large spikes in the dorsal portion. On cutting the two halves of the nerve consecutively between stimulating and recording electrodes the burst of spikes was shown to be due to synaptic excitation of motor axons and the compound spike to the direct excitation of afferent fibres (Text-fig. 2 ; Pl. 2f, g). An attempt was made to stimulate the proximal dissected ends of the fifth nerve and to observe any muscular movements or record any muscle action potentials which followed. The results were somewhat inconclusive owing to the placing of stimulating electrodes in such a small preparation. However, in several experiments in which the dorsal stump was stimulated contractions occurred in the sphincter muscles and muscle potentials could also be recorded. Comparable muscular activity could not be seen on applying the electrodes to the ventral stump when the dorsal part was cut, and hence these observations support the results obtained by direct recording from the dissected nerve.

Text-fig. 2.

Diagram of the last abdominal ganglion and a dissected fifth nerve to show the stimulating and recording positions used in localizing function in the nerve root. The ganglion is shown from the ventral aspect. The conditions used are numbered 1–4 and described in the text S, stimulating electrodes; R, recording electrodes.

Text-fig. 2.

Diagram of the last abdominal ganglion and a dissected fifth nerve to show the stimulating and recording positions used in localizing function in the nerve root. The ganglion is shown from the ventral aspect. The conditions used are numbered 1–4 and described in the text S, stimulating electrodes; R, recording electrodes.

(3) Electrical stimulation of the contralateral connective

This connective was used to avoid antidromic stimulation which might possibly occur if there are any ‘through’ fibres to the ipsilateral connective. Normally no response was seen in the ventral part of the dissected fifth root on stimulating this connective but a discharge of large spikes occurred in the dorsal part. The latter persisted in the cut proximal end of this dorsal portion and were therefore motor in origin (Pl. 2 a, b).

(4) Electrical stimulation of the contralateral fifth nerve

Stimulation of this nerve considerably reduced the possibility of antidromic stimulation of fibres in the dissected fifth nerve and again produced a large response in the dorsal portion and only an occasional small response in the ventral part. Stimulation of ipsilateral and contralateral N1, N2 and N4 also produced a similar result.

These results represent the conclusions based on a large number of experiments as it was rarely possible to dissect the nerve into entirely separate motor and sensory parts and hence a certain amount of overlap occurred. However, it appears that the structural differences in the origins of the parts of the nerve result in the physiological localization of predominantly sensory fibres in the ventral part and predominantly motor fibres in the dorsal part of the root. Some of the experiments described above were performed on dissected preparations of the fourth nerve roots and the same general results were obtained. Since this nerve becomes differentiated more anteriorly in the ganglion it does not provide as good a preparation as the fifth nerve.

The functional localization of fibres described in the present results receives support from histological work on other insects, utilizing methods which depict both single cells and fibre tracts. The methylene-blue and Golgi studies of Hilton (1911) on the Corydalis larva, of Zawarzin (1924) on Aeschna and of Guthrie (1961) on Gerris are useful in showing the paths of single motor and sensory neurones in the dorsal and ventral parts of the ganglion and the relationships of these with each other and with association neurones. Further evidence for a comparable localization of sensory and motor tracts is found in the work of Power (1948), who mentions their different origins in Drosophila nerves, and in the description of Periplaneta ganglia given by Pipa, Cook & Richards (1959). Wigglesworth’s (1959) work on the localization of fibres in the dorsal and ventral parts of the leg nerves of Rhodnius has already been mentioned. He has counted the small axons in the ventral half and found that approximately the same number of sensilla exist on the segments of the mesothoracic leg adding support to the view that they are sensory neurones.

Some observations of insect behaviour can also be interpreted as supporting functional localization in a nerve root. Binet (1894) found that in beetles with immovable wings the alar nerve is considerably reduced and only the ventral, presumably sensory, root exists. He also found that in Dytiscus pressure on the dorsal part of the ganglion caused motor paralysis without anaesthesia, while the same treatment of the ventral part caused anaesthesia without loss of movement. The current evidence therefore seems to refute the claim of Roeder, Tozian & Weiant (1960) that there is no clear differentiation of insect nerves into dorsal and ventral parts as there is in the vertebrate. Although the present results provide no evidence to the contrary they do not necessarily mean that the two parts of the root are composed exclusively of motor and sensory axons. The emphasis is on a predominant localization of fibre types in the nerve origin which become mixed more peripherally.

It is of interest to note that Zawarzin’s (1924) comparison between the dorsal and ventral roots of the dragonfly and of the vertebrate is supported by the physiological work on the Anax ganglion. In both, a segmental reflex can be demonstrated which is dependent on separate fibre tracts leading to and from the central nervous system.

However, it is significant that the functional specialization of dorsal and ventral roots in the insect is directly opposite to the well-known dorsal (sensory) and ventral (motor) roots in the vertebrate. In Anax, reflex activity in the fifth nerve arises from stimulation of afferent fibres passing into the ganglion by a ventral root and exciting efferent neurones in the dorsal root. This reflex mediates the escape response and the importance of its local co-ordination has been stressed previously, but the separate roots were then unknown (Fielden, 1960). The possibility of separating the two roots therefore facilitates an investigation of the relationship between sensory and motor neurones in an insect.

I am grateful to Dr G. M. Hughes for his constructive criticism during the course of the work and for reading the manuscript. I would also like to thank the Medical Research Council for their financial support.

Binet
,
A.
(
1894
).
Contribution à l’étude du système nerveux sous-intestinal des insectes
.
J. Anat. Physiol
.
30
.
449
58O
.
Fielden
,
A.
(
1960
).
Transmission through the last abdominal ganglion of the dragonfly nymph, Anax imperatar
.
J. Exp. Biol
.
37
,
832
44
.
Fielden
,
A.
&
Hughes
,
G. M.
(
1962
).
Unit activity in the abdominal nerve cord of a dragonfly nymph
.
J. Exp. Biol
.
39
,
31
44
.
Guthrie
,
D. M.
(
1961
).
The anatomy of the nervous system of the genus Gerris (Hemiptera-Heteroptera)
.
Phil. Trans. B
,
244
,
65
102
.
Hess
,
A.
(
1958
).
Experimental anatomical studies of pathways in the severed nerve cord of the cockroach
.
J. Morph
.
103
,
479
92
.
Hilton
,
W. A.
(
1911
).
Some remarks on the motor and sensory tracts of insects
.
J. Comp. Neurol
.
21
,
383
91
.
Holmes
,
W.
(
1943
).
Silver staining of nerve axons in paraffin sections
.
Anat. Rec
.
86
,
157
.
Peters
,
A.
(
1955
).
Experiments on the mechanism of silver staining
.
Quart. J. Micr. Sci
.
96
,
84
115
.
Pipa
,
R. L.
,
Cook
,
E. F.
&
Richards
,
A. G.
(
1959
).
Studies on the hexapod nervous system. II. The histology of the thoracic ganglia of the adult cockroach, Periplaneta americana (L.)
.
J. Comp. Neurol
.
113
,
401
23
.
Power
,
M. E.
(
1948
).
The thoracico-abdominal nervous system of an adult insect, Drosophila melanogaster
.
J. Comp. Neurol
.
88
,
347
91
.
Roeder
,
K. D.
,
Tozian
,
L.
&
Wbiant
,
E. A.
(
1960
).
Endogenous nerve activity and behaviour in the mantis and cockroach
.
J. Ins. Physiol
.
4
,
45
62
.
Wigglesworth
,
V. B.
(
1957
).
The use of osmium in the fixation and staining of tissue
.
Proc. Roy. Soc. B
,
147
,
185
99
.
Wigglesworth
,
V. B.
(
1959
).
The histology of the nervous system of Rhodnnu proHxux (Hemiptera). II. The centr l ganglia
.
Quart. J. Micr. Sci
.
100
,
299
313
.
Zawarztn
,
A.
(
1924
).
Zur Morphologie der Nervenzentren. Das Bauchmark der Insekten. EinBeitrag zur vergleichenden Histologie (Histologische Studien Uber Insekten VI)
.
Z. wiss. Zool
.
122
,
323
424
.

Plate 1

(a) Transverse section through the posterior part of the last abdominal ganglion of an Anax nymph showing the dorsal and ventral parts of the paraproct nerves N4 and N5. Both nerve roots show the dorsal thick-walled ‘motor’ axons and ventral thin-walled ‘sensory’ axons surrounded by ganglion cells. The occurrence of the two types of motor axon is seen in N5. Osmium-ethyl gallate method, 4 μ section.

(b) Transverse section of the fourth nerve of the last ganglion showing the large thick-walled dorsal axons, many with an enveloping mesaxon. Schwann cells can be seen in the nerve sheath. Osmium fixed, iron haematoxylin preparation, 6 μ section.

(a) Transverse section through the posterior part of the last abdominal ganglion of an Anax nymph showing the dorsal and ventral parts of the paraproct nerves N4 and N5. Both nerve roots show the dorsal thick-walled ‘motor’ axons and ventral thin-walled ‘sensory’ axons surrounded by ganglion cells. The occurrence of the two types of motor axon is seen in N5. Osmium-ethyl gallate method, 4 μ section.

(b) Transverse section of the fourth nerve of the last ganglion showing the large thick-walled dorsal axons, many with an enveloping mesaxon. Schwann cells can be seen in the nerve sheath. Osmium fixed, iron haematoxylin preparation, 6 μ section.

Plate 2

Localization of function in the dorsal and ventral roots of N5 by dissection of the nerve and recording from the two halves under various conditions.

(a) Recording from the ventral half, electrical stimulation of the contralateral connective (position 3, Text-fig. 2). Only a few small spikes are seen.

(b) Recording from the dorsal half, electrical stimulation of the contralateral connective in the same preparation. A burst of large spikes is seen.

(c) Recording from the whole nerve in the same preparation as a check on recording conditions.

(d) Recording from the ventral half in a different preparation, mechanical stimulation of the hairs edging the paraproct (position 1, Text-fig. 2). Small sensory discharges can be seen.

(e) Recording from the dorsal half of the nerve in the same preparation. The nerve is cut distally. Shows larger motor spikes on stimulation of the tactile hairs.

(f) Recording from the ventral half in another preparation. Electrical stimulation of the whole N5 distally (position 2, Text-fig. 2). A compound afferent spike can be seen following the stimulus artifact.

(g) Same preparation as above, recording from the dorsal half following distal stimulation of N5. Motor discharges can be seen.

(a) Recording from the ventral half, electrical stimulation of the contralateral connective (position 3, Text-fig. 2). Only a few small spikes are seen.

(b) Recording from the dorsal half, electrical stimulation of the contralateral connective in the same preparation. A burst of large spikes is seen.

(c) Recording from the whole nerve in the same preparation as a check on recording conditions.

(d) Recording from the ventral half in a different preparation, mechanical stimulation of the hairs edging the paraproct (position 1, Text-fig. 2). Small sensory discharges can be seen.

(e) Recording from the dorsal half of the nerve in the same preparation. The nerve is cut distally. Shows larger motor spikes on stimulation of the tactile hairs.

(f) Recording from the ventral half in another preparation. Electrical stimulation of the whole N5 distally (position 2, Text-fig. 2). A compound afferent spike can be seen following the stimulus artifact.

(g) Same preparation as above, recording from the dorsal half following distal stimulation of N5. Motor discharges can be seen.