The Neural Basis of Escape Swimming Behaviour in the Squat Lobster Galathea Strigosa: I. Absence of Cord Giant Axons and Anatomy of Motor Neurones Involved in Swimming

Closely related species may display patterns of behaviour which are basically similar, but differ in detail between species either qualitatively or quantitatively (Tinbergen, 1958, 1960; Berg 1974; Blest, 1960). Such patterns, which have diverged from a common ancestor through different selection pressures in their respective habitats (Baerends, 1958; Wickler, 1961), are termed homologous. In multisegmented animals, basically similar structures may be differentiated and specialized in successive segments, and the types of behaviour in which these structures are used can also differ either qualitatively or quantitatively. Interspecific and intersegmental differences in homologous patterns of behaviour are reflected in the structure and function of neurones and neural circuits. A determination of the extent and nature of neuronal homology and variation through the comparative study of related segments and species should facilitate an understanding of the processes which underly the development and evolution of neural circuits and of behaviour.

Many decapod crustaceans escape from threats by means of a rapid abdominal flexion, or tailflip. In crayfish, this behaviour is mediated by two pairs of interneurones with giant axons (for recent review see Wine & Krasne, 1982). These make direct electrical connections with a pair of specialized giant motor neurones (the Motor Giants, MoGs) in the connectives between each abdominal ganglion, and indirect weaker connections with a pool of fast flexor motor neurones in each ganglion. During escape the anterior and posterior segments behave differently, depending on the direction of the initiating stimulus. Mittenthal & Wine (1973, 1978) have found that the connections between the giant axons and MoGs are present in some ganglia, but absent in others. In the lobster, escape behaviour does not differ qualitatively from the crayfish, and the giant interneurones and fast flexor motor neurones show apparent strict homology with those of the crayfish (Otzuka, Kravitz & Potter, 1967; Mittenthal & Wine, 1978). In hermit crabs (Crustacea, Anomura) escape behaviour consists of a withdrawal reflex during which they rapidly retract into their shells (Wiersma, 1961 ; Chapple, 1966). The behaviour is mediated by only one pair of giant axons, thought to be homologous with the crayfish MG axons (Umbach & Lang, 1981) ; the LG axons appear to be absent in this species. The giant motor neurones (GMs) of the hermit crab are functional homologues of the crayfish MoGs, but their anatomy is significantly different since they synapse with the giant axons close to the ganglionic neuropile. In contrast, the homologous connection in crayfish is located in the connectives near to the exit of the third root (Furshpan & Potter, 1959).

The squat lobster, Galathea strigosa (Crustacea, Anomura), differs markedly from both crayfish and hermit crabs. When threatened or startled, it escapes with a series of extension-flexion movements of the abdomen which resemble crayfish tailflips. In contrast to crayfish, however, the abdomen is held flexed beneath the cephalothorax in the stationary animal (Sillar & Heitler, 1982). Thus escape must begin with extension and not flexion. In this paper we examine the anatomy of neurones underlying the escape. The behaviour is not mediated by giant interneurones since these are absent in the abdominal nerve cord. The fast flexor and extensor motor neurones appear homologous with related species. However the neurone homologous with the crayfish MoG is an unspecialized fast flexor. The results, which suggest that escape behaviour is homologous with non-giant swimming in crayfish (Schrameck, 1970), are discussed in the context of the evolution of giant-fibre mediated escape circuits. In the two papers which follow (Sillar & Heitler 1985a,b) we have analysed the motor programme for the behaviour, the role of sensory feedback and the mechanisms for burst production in flexor and extensor motor neurones during swimming.

Squat lobsters, Galathea strigosa, were collected from lobster creels in St Andrews Bay and maintained in circulating sea water aquaria at temperatures ranging from 3 to 15 °C. Mature, healthy adults of either sex, measuring 6—9 cm from tip of rostrum to caudal edge of telson, were used in all experiments.

Relevant parts of the abdominal nervous system were dissected, fixed for 3 h in 3 % phosphate buffered gluteraldehyde and post-fixed in 1 % osmium tetroxide for 30min. l-μm sections were stained with Toluidine Blue. The peripheral nervous system of the abdomen was stained with 1 % Methylene Blue (BDH 154490). To examine the central anatomy of extensor and flexor motor neurones, respectively, cobalt chloride (Pitman, Tweedie & Cohen, 1972) was introduced into the cut ends of axons in abdominal 2nd and 3rd roots. On occasions where staining was light, silver intensification was used to enhance cobalt-sulphide stains, using the method of Davis (1982). The following results are based on over 100 such preparations.

Conventional methods of extracellular and intracellular recording and display were used throughout. Fast flexor muscle fibres were penetrated using glass microelectrodes filled with 3 mmol 1-1 potassium acetate and having resistances of 20-40 MΩ. The somata of motor neurones were penetrated in an isolated nerve cord preparation. The nerve cord was pinned ventral side up in a Sylgard-lined Petri-dish and immersed in chilled oxygenated saline (Mulloney & Selverston, 1974, with Tris substituted by TES and buffered to pH7·2-7·4). The ganglion under study was mechanically desheathed and motor neurone somata were located visually on the basis of soma size and position. Somata were penetrated with glass microelectrodes filled with 5% Lucifer Yellow (Stewart, 1978) in l moll-1 LiCl (40-60MΩ). Motor neurones were filled using 0·5-s, 10-nA pulses of negative current every second for 20—60 min.

The abdominal nervous system in Galathea consists of five free abdominal ganglia (G1-G5) linked by paired connectives (Fig. 1C). In crayfish there are six abdominal ganglia. In Galathea the first true abdominal ganglion (TAG) has become fused to the posterior end of the thoracic ganglionic mass (TGM). Therefore reference to G1 in Galathea corresponds to G2 in crayfish, and so on. Three paired roots arise from each ganglion. As in crayfish and hermit crabs, abdominal third roots (r3) appear to be purely motor and exclusively innervate the flexor muscles. The point of exit of r3 from left and right connectives is asymmetrical; r3 on one side leaves the connective up to 1 mm more rostrally than the other. In addition the third roots of G3 leave the connectives from its dorsal aspect while other abdominal roots arise more laterally (Fig. 1C). These features of abdominal asymmetry and torsion may reflect the close taxonomic relationship between the Galatheid and Pagurid anomurans. However the abdomen of Galathea appears to be externally symmetrical (Fig. 1A), and no asymmetries have been observed in either the motor neurone pools or the abdominal musculature.

A striking feature of the abdominal connectives in Galathea is an absence of large diameter axons. Cross sections cut at all levels of the abdomen reveal no axon profiles of similar relative diameter to the giant fibres of crayfish or hermit crabs (Fig. 2). In the dorsal region of the connectives just posterior to each ganglion is a group of seven or eight axons which are relatively large in diameter (approx. 20-30 μm) compared with other profiles. In serial sections of the nerve cord these axons leave the connective via r3 and are absent in sections cut more posteriorly. Thus, these are the axons of fast flexor motor neurones (FFs). With the exception of these, the dorsal region of the connectives contains a homogeneous population of axon profiles which rarely exceed the diameters of the FFs. In addition, FF axons all have similar diameters, both in the connectives and in r3. This feature also contrasts with crayfish and hermit crabs, in which the axon of the giant motor neurone is significantly larger than other FFs. These results suggest that Galathea lacks both giant interneurones and the giant motor neurone although it is likely that homologues of the crayfish neurones are present in reduced form.

1 mm after leaving the connective, r3 bifurcates into a small superficial branch (r3s) which extends laterally to innervate a thin sheet of ventral tonic flexor muscles, and a larger main branch (r3m) which ascends dorsally and rostrally to innervate the fast flexor muscles. The central anatomy of fast motor neurones has been visualized by selective cobalt backfilling of r3m. Backfilling r3m G2 stains two soma clusters in G2 and one in G3 (Fig. 3). In describing these clusters the terminology of Mittenthal & Wine (1978) for crayfish FFs will be used for ease of comparison. Anterior and contralateral to the filled root is a medial cluster of cells (flexor medial contralateral cluster: FMC) which normally contains four somata. The two largest (approx. 90-110 μm diameter) are homologous with the crayfish Fl and MoG neurones (see below). A second group of neurones have somata located ipsilateral to the filled root in G2 (flexor posterior ipsilateral cluster: FPI). This cluster also contains four somata in G2 with diameters in the range 70—90 μm. Often the FPI neurones form two pairs and the neurites of each pair follow similar paths into the neuropile. A third group (flexor anterior contralateral cluster: FAC), comprising two cells, has somata in the anterior portion of G3 contralateral to the axons.

In terms of the number, size and distribution of somata, FFs in G2 show almost strict homology with crayfish FFs in G3. An exception to this is the occurrence of three FAC FFs in crayfish compared with only two in the homologous cluster in Galathea. A further difference is that no branches have been observed arising from any FF axons in the connectives (e.g. Fig. 3A). This contrasts with the crayfish in which the MoG neurone sprouts featherlike processes from its axon near r3, and provides further evidence that Galathea may lack a specialized giant motorneurone.

The three FF soma clusters have been visualized in G1 to G5 by backfilling r3m of these ganglia (Fig. 4). Attempts to fill r3m to the fused thoracic-abdominal ganglion (TAG) have failed to stain FMC and FPI clusters successfully, probably due to the large distances involved (up to 20mm in large specimens). However two FAC neurones stain in G1 and eight large axons ascend the connectives towards TAG, suggesting strict homology with G2.

The FAC cluster contains two somata in G1 to G4, but no FAC neurones are present in G5. Each FAC neurone has almost identical anatomy. The axons and neurites of each pair follow similar paths and the two somata are always closely apposed in the anterior contralateral portion of the ganglion.

The FMC cluster contains four somata in G1 to G3, three in G4 and one in G5. Thus G1, G2 and G3 show strict serial homology in the number, size and distribution of FMC somata. Of the three somata in G4, one is large (approx. 100 μm diameter) and extremely lateral in its location, and is therefore likely to be a homologue of the crayfish FI neurone (see below). The remaining two are smaller and anterior in location. Therefore the missing neurone in this cluster may be the homologue of the crayfish MoG neurone. The single FMC neurone in G5 is large (approx. 100 μm diameter) and located near the midline of the ganglion (Fig. 4Av). It is difficult to establish homology with anterior ganglia on the basis of soma position. However the soma can be traced to an axon which bifurcates before exiting via r3. This may indicate that the neurone is an FI homologue (see next section).

The FPI cluster contains four somata in G1 to G4 and shows strict serial homology in these ganglia. G5 deviates from strict homology by having six FPI somata. The similarities with crayfish are striking (Fig. 4B). In both species there is a small anterior-posterior reduction in the total number of FFs. In crayfish this has been correlated with a reduction in the flexor muscle mass in caudal segments.

In each of G1, G2 and G3 the FMC cluster is strictly homologous with the crayfish in number, size and distribution of somata. The largest and most lateral of these is homologous with the crayfish FI neurone, identified by Wine & Mistick (1977). Cobalt backfilling either r3m or r3s stains the FI soma and an axon which projects down the remaining unstained branch of r3. Thus this neurone is the only member of the FF pool whose axon branches at the first bifurcation of r3. Its unique peripheral branching pattern has allowed the following unambiguous identification of the neurone’s function and the establishment of homology with the crayfish FL

Physiological identification of the FI branch point has been obtained as follows (Fig. 5). Extracellular stimulation of r3s or r3m G2 results in a short latency spike recorded in the remaining r3 branch (Fig. 5Bi, ii). This potential is elicited at a set threshold of stimulation and further increases in stimulus intensity (up to 100 V) do not alter its amplitude or waveform. Subsequent intracellular penetration of ventral FF muscle fibres in the region depicted in Fig. 5A, give records showing an inhibitory junctional potential (IJP, Fig. 5Biii) which is tightly coupled to the extracellular spike in threshold and latency. The FI IJP is not always visible in intracellular recordings, presumably because its reversal potential lies close to the resting potential of FF muscle fibres (approx. —70mV). It was found difficult to alter the membrane potential of FF muscle fibres by injection of up to 10 nA current, probably because of their large volume. On several occasions, however, sufficient depolarization of the muscle membrane was achieved to enhance the amplitude of the FI IJP (Fig. 5Biii). It is not clear whether this was possible because of lower electrode resistance or because penetrations were made closer to FI endplates. On a number of other occasions FI IJPs were hyperpolarizing at resting potential. This may have been due to artificial membrane depolarization resulting from the penetration process.

The central branching of FI has been revealed by injection of Lucifer Yellow into its soma in each of Gl, G2 and G3 (Fig. 6). The FI soma was located visually on the basis of its size and position. The dendritic field and major dendritic branches of FI are similar in each ganglion. In addition, the central branching of FI closely resembles that of the crayfish homologue (see Fig. 7 in Mittenthal & Wine, 1978), and no major differences exist between the two species. The FI soma is always large (approx. 110μm diameter) and located at the extreme lateral edge of the ganglion, near the origin of r 1. The FI soma often appears ovoid in shape and on some occasions projects partially into the base of the 1st root. This may be due to tissue distortion during processing.

The more medial of the two large FMC somata in Gl, G2 and G3 is homologous with the crayfish MoG neurone in terms of its size and position. In contrast to the crayfish MoG, however, this neurone has a complex dendritic tree in the neuropile (Fig. 7A-C) and in this respect appears to be a typical FF. The crayfish MoG has no central dendrites but sprouts featherlike processes in the connectives in the region of the giant fibres (Mittenthal & Wine, 1978). No processes have been observed arising from the MoGH axon in the connectives of Galathea. The crayfish MoG branches extensively in the periphery and bifurcates at the first branch point of the third root (Furshpan & Potter, 1959). The MoGH axon in Galathea does not branch down r3s but its axon often branches at each bifurcation of r3m suggesting a wide distribution in the periphery.

The crayfish MoG is specialized with respect to the central giants, and receives excitatory input only from them (Wine & Krasne, 1982). In contrast, intrasomatic recordings from the MoGH in Galathea reveal a low level of spontaneous synaptic input from a number of unidentified neurones. Single electrical shocks to the TAG-G1 connective elicit a complex compound EPSP in the MoGH soma in G2 (Fig. 8A) in addition to suprathreshold activation of other r3m units. We were unable to observe soma spikes in response to cord stimulation. However, injection of 10 nA depolarizing current into the soma resulted in a train of spikes which correlated 1 : 1 with spikes recorded extracellularly in r3m (Fig. 8B).

As in crayfish and lobsters, the axons of efferent and afferent neurones involved in abdominal extension enter the CNS via the second root (r2) of abdominal ganglia. Cobalt backfills of r2 G2 stain up to 12 somata in G2 which are distributed among three clusters. Identification of these extensor efferents is based on homologies with neurones in crayfish (Treistman & Render, 1975; Wine & Hagiwara, 1977) and lobster (Otsuka et al. 1967) abdominal ganglia.

As shown in Fig. 9, two clusters of somata are located ipsilateral and one contralateral to the filled root. The somata of four or five fast extensor motor neurones (FEs) form a tight cluster at the ipsilateral edge of the ganglion between the bases of rl and r2. The diameters of FE somata show considerable variability with a mean of 64 μm. A second cluster lies anterior to the FEs, approximately at the origin of r 1, and comprises the somata of four smaller, slow extensor motor neurones (approx. 20-30 μm diameter). In most preparations the neurites of SEs enter the neuropile along a tight tract, distinct from the neurite path of FEs.

In both crayfish (Wine & Hagiwara, 1977) and lobsters (Otsuka et al. 1967) the soma of the peripheral inhibitor to the fast extensor muscles (El) is large (approx. 110μm diameter in crayfish) and located contralateral to the filled root. Located contralateral to r2 in Galathea are two somata (Fig. 9). Of these the largest has a soma diameter of approximately 95 μm and is therefore labelled the phasic extensor inhibitor (El). The remaining cell in this cluster has a much smaller soma (approx. 30 μm diameter). In view of its close proximity to El and its similarity to a small contralateral soma in crayfish this cell is labelled the tonic extensor inhibitor (el). In crayfish ganglia (Wine & Hagiwara, 1977) a third smaller contralateral soma (approx. 15 μm diameter) of unknown function stains in about 25 % of preparations. In contrast only two contralateral somata are observed in r2 G2 fills in Galathea.

The only consistent differences between these results and those of Wine & Hagiwara (1977) for crayfish are the absence of a third contralateral soma and a small reduction in the total number of ipsilateral cells. There are four or five SEs and five or six FEs in crayfish G2. In all essential respects the two pools are homologous in soma size and distribution.

Homologous extensor efferents have been stained in each abdominal ganglion except the 5th (last ganglion, not studied) and the fused last thoracic - first abdominal ganglion (Fig. 10). Gl, G2 and G3 show strict serial homology in number size and distribution of somata (Fig. 10A-C). In G4, however, there are only three SEs, three FEs and the two contralateral inhibitors (Fig. 10D). This caudal decline in the number of efferent somata is consistent with the result of Wine & Hagiwara (1977) for the penultimate abdominal ganglion in crayfish. TAG also departs slightly from strict serial homology in that only seven ipsilateral and two contralateral somata are stained. Of the ipsilateral cells there are four FEs and three SEs.

The major dendritic domain of the extensor efferents forms a well defined U-shaped region of dorsal neuropile, (Fig. 11), with the densest accumulation of processes lying on the ipsilateral side of the ganglion. Some of the contralateral neuropilar arborizations can be attributed to EI When El is stained selectively by injection of Lucifer Yellow into its soma it is seen to have two major dendritic loci which project anteriorly from the neurite, in line with each contralateral hemiganglion (Fig. 11A). In favourable cobalt fills a tight tract of several large neurites courses horizontally across the ganglion, suggesting that other extensors arborize contralaterally. This has been confirmed by intrasomatic injection of Lucifer Yellow into a number of FEs (Fig. 11B-D). They have complex bilateral branching patterns. The major dendritic branching occurs ipsilateral to the soma but in each fill a large neurite courses laterally across the ganglion to an area of often dense arborization. This feature of FE anatomy contrasts directly with homologous neurones in crayfish (Wine & Hagiwara, 1977), where only a few branches of the extensor excitors may cross the midline of the ganglion.

Bullock & Horridge (1965) state that ‘Crustacea with a rapid tailflip can be expected to have giant fibres running the length of the cord’. Indeed, among the Crustacea giant fibre systems for escape appear to be widespread, especially amongst those with elongated, flexible abdomens. There are two pairs in Homarus (Allen, 1894), Cambarus and Palaemonetes (Johnson, 1924, 1926) and Leander (Holmes, 1942). The stomatopod, Squilla has one pair only (Bullock & Horridge, 1965), as do the Anomurans, Callianassa (Turner, 1950) and Pagurus (Umbach & Lang, 1981). The squat lobster, Galathea strigosa (Anomura) has a tonically flexed abdomen at rest (Fig. 1A) which can be repeatedly extended and flexed during escape but it does not possess giant fibres.

In crayfish the giant fibre circuit is now well understood and involves three key neural components: the giant fibres themselves; the MoGs and the Segmental Giants (SGs; Roberts et al. 1982). Each of these is thought to be derived from primitive elements in an ancestral non-giant swimming circuit (Wine & Krasne, 1982). Thus it is thought that the giant fibres evolved from smaller premotor interneurones, the MoGs from smaller unspecialized FFs and the SGs from conventional limb motor neurones. The swimming circuit, still present in crayfish, is under the control of a central pattern generator (CPG) which drives extension first with flexion following at short and near constant latency (Reichert & Wine, 1982). Thus the species from which crayfish evolved must have lacked both giant axons in the cord and a pool of MoGs in abdominal ganglia. Moreover escape must have involved backwards swimming by repeated extension-flexion cycles of the abdomen which began with extension. On the basis of the present data, we hypothesize that Galathea has features of such an ancestor. If this is so then, in theory at least, each of the three key elements of the crayfish giant fibre circuit will be present in unspecialized or reduced form.

Since the premotor interneurones of the crayfish swimming circuit are numerous, small and largely unidentified (Wine & Krasne, 1982), it may be impossible to identify primitive homologues of the giants in Galathea on an anatomical basis. The complete absence of giant fibres in the connectives provides evidence only that escape in Galathea involves non-giant circuits. Similarly the SGs probably evolved from conventional limb motor neurones. The remaining 50 or so swimmeret motor neurones in each hemiganglion are very similar anatomically and hence the search for a primitive SG homologue in Galathea is also likely to be unsuccessful.

In contrast, the crayfish MoG was recruited from a relatively small population of large, uniquely identifiable FFs in each ganglion. The MoG differs from other FFs in its soma size and position, location of dendrites and extent of peripheral distribution. Therefore, the primitive homologue of the MoG is identifiable in a non-giant circuit (MoGH).

In terms of the number, size and distribution of FF somata, the homologies that exist between crayfish and Galathea are striking (Fig. 4B). In both species anterior ganglia show strict serial and close interspecific homology. In posterior ganglia a decline in the number of FFs occurs and this is correlated in crayfish with an observable reduction in the volume of muscle to be innervated in these segments (Mittenthal & Wine, 1978). The most convincing evidence for homology among FFs in the two species is the FI. This neurone is identifiable on the basis of soma size and position, central branching and peripheral function in Galathea and crayfish (Wine & Mistick, 1977). These data imply that the two species derive from a common ancestor and that the FI neurone has changed little since they diverged.

In contrast to FI, the MoGH differs considerably from the crayfish MoG (despite the fact that the two neurones are homologous in soma size and position), having a complex dendritic tree in the neuropile and no axonal branches in the connectives. The major selection pressure on the evolution of the crayfish MoG must have been to ensure a rapid and complete tail flexion in response to a single impulse in the giant fibres. It would seem logical, then, for crayfish to have recruited the largest FF from the available pool (perhaps with the widest peripheral distribution) in order to minimize the amount of modification required in its structure. MoGH in Galathea happens to be the largest flexor excitor motor neurone, but other important factors such as axon course and established functional connections with particular premotor elements may well have been important. The possibility that another different FF has assumed the role of MoG in Galathea can be ruled out since no axonal specializations have been seen in any FF from cobalt backfills. The FF pools in Galathea consist of a homogeneous population of neurones, each with similar anatomy.

The only notable difference that we have observed among FEs in the two species is the presence of strongly bilateral dendritic domains among the fast extensors in Galathea. The homologous neurones in crayfish have mainly ipsilateral dendrites with few processes ever crossing the midline of the ganglion (Wine & Hagiwara, 1977).

In crayfish, the presence of ipsilateral FE dendrites has been associated with the retention of a system suitable for independent lateral activity whereas the flexors, with bilateral dendrites (Wine & Hagiwara, 1977), appear specialized to operate bilaterally. But clearly, if Galathea does represent the primitive crayfish condition as we have suggested, then the demonstration of bilateral dendrites in this species suggests that crayfish FEs may have secondarily lost their contralateral processes. Crayfish may have evolved lateralized FE dendrites as a concomitant of the giant-fibre system. Postgiant extension in crayfish differs from non-giant extension in being a chain reflex induced by the initial flexion and which is dependent upon patterned sensory feedback (Reichert, Wine & Hagiwara, 1981). Perhaps it is crucial that post-giant extension is lateralized with respect to this afferent input which may be non-uniform in a changing environment.

Thus several features of escape in Galathea resemble the predicted ancestral crayfish form. (1) There are no giant axons in the connectives. (2) The FF and FE pools are homologous with crayfish. (3) The MoGH is an unspecialized FF. (4) Escape involves repeated extension-flexion cycles of the abdomen which begin with extension.

Recent evidence on the escape withdrawal reflex of the hermit crab shows that there is a single pair of giant axons in the connectives which may be homologues of the crayfish medial giant axons. Each central giant axon makes direct electrical connections with a pool of specialized giant motor neurones in the abdomen. These GMs, homologues of the crayfish MoGs, possess most of the anatomical specializations of the crayfish neurone. However, the axonal dendrites are located at the extreme caudal edge of the neuropile and not in the mid-connectives (Umbach & Lang, 1981). Thus the hermit crab lies somewhere between Galathea and crayfish in terms of the anatomy of its escape circuitry. The homologies that exist between the FF pools in these three species imply that they all derive from a common ancestor, and that the differences in their escape circuits have resulted from different selection pressures in their respective habitats. The giant fibres of the crayfish are a late development (Wine & Krasne, 1982) and thus unless Galathea has secondarily lost the giant fibre-MoG system it must represent an ancestral form in terms of the neural control of abdominal flexion. This means that either the escape system of crayfish and hermit crabs evolved independently but convergently after the divergence of the Macrura and Anomura, or perhaps hermit crabs are more closely related to crayfish than they are to Galathea. In either case the non-giant, backward swimming of Galathea may represent the primitive form of escape behaviour from which the others evolved.

We wish to thank Mrs C. Lamb for typing the manuscript. This work was supported by an SERC Studentship to KTS and in part by an SERC grant to WJH. We thank W. Stewart for providing a sample of Lucifer Yellow.