Naturally occurring abnormalities (Bruchdreifachbildungen) in decapod crustacean appendages are described. They are similar to the range of structures experimentally produced by cutting notches in the sides of insect legs (Bohn, 1965). It is argued that they result from failure of wounds to heal. Regeneration from a free surface along the proximodistal axis is always in a distal direction. Surfaces regenerating circumferentially can regenerate in either direction around the circumference. Regeneration will proceed until the two surfaces of the wound meet. Then, where the two surfaces on either side are non-congruent, intervening tissues will be intercalated. This explanation accounts for the range of structures observed after notching experiments (Bohn, 1965) and seen in crustacean Bruchdreifachbildungen. The explanation says that regeneration will occur wherever wounds fail to heal. This avoids the difficulties of the complete circle rule (French, Bryant & Bryant, 1976) and explains why appendages with mirror-image symmetry are often capable of regeneration.

From time to time, fishermen capture lobsters and crabs with curious, mirror-image symmetrical lateral outgrowths from their chelae and occasionally from other appendages. Such abnormalities occur in a variety of animals, including crustaceans, insects, amphibians, birds and mammals (Przibram, 1921). They have been called’ Bruchdreifachbildungen ‘because they are thought to result from damage to the original structure and they produce a triplication distal to the site of injury (Przibram, 1921). In this paper we describe a collection of abnormal chelae which has been assembled over a period of at least 43 years by the Marine Laboratory, Aberdeen, and provide a possible explanation for the phenomenon. The deformities resemble the lateral outgrowths some-times found in cockroach limbs after deep notches have been cut into their sides during post-embryonic growth (Bohn, 1965). They take various forms from a small bump to pairs of extra segments. They often exhibit the phenomenon of distal expansion, which has been noted previously in experimentally formed double-posterior amphibian limbs (Slack, 1977, 1980a) and which is also seen in some of the lateral regenerates of insect legs (Bohn, 1965). The fact that crustacean limbs regenerate so readily after autotomy (see Bliss, 1960; Paul, 1914) and that they can respond locally to produce abnormal lateral supernumeraries suggests that the crustacean limb might offer a valuable new system for studying pattern formation. Indeed preliminary reports suggest that a gradient system, similar to that found in insects, may determine proximodistal organisation of the crayfish leg (Mittenthal, 1978).

We have now examined 11 decapod crustacean limbs showing various abnormalities. They include specimens from the lobster Homarus gammarus (L.), the edible crab Cancer pagurus L. and the Norway lobster Nephrops norvegiens (L.). Together with some of the specimens originally described by Bateson (1894) and Przibram (1921) there is now a sufficient range of examples to permit a useful discussion of the cause of the phenomenon. The fact that Bruchdreifachbildungen occur in such a wide variety of animals suggests that our findings may be of some general significance.

All the outgrowths described in this paper are on the terminal or the penultimate segments of the chela (the dactylopodite and the propodite). However, although this is the most common site for the abnormality, similar outgrowths can occur on antennae (Przibram, 1921, specimen nos 13 a, 13 b), and walking legs (Bateson, 1894, specimen no. 808). They can be derived from the carpopodite (Przibram, 1921, specimen no. 16), the meropodite (Bateson, 1894, specimen no. 826; Przibram, 1921, specimen no. 37), the basipodite (Bateson, 1894, specimen no. 808) and the coxopodite (Przibram, 1921, specimen no. 23). There is a natural dimorphism of chelae in decapods (Emmel, 1908) in which one chela is adapted for ‘cutting’ and the other for ‘crushing’. We found examples in both cutters and crushers. We adopted the following convention to identify the origin of the lateral supernumeraries. The chela is flattened laterally and we call the surface which faces the contralateral chela the internal face. The outward-facing opposite side is called the external face. The propodite extends ventrally as the index (or propopodite extension) to make the lower jaw of the claw structure. The dactylopodite forms the moveable dorsal element of the claw. Details of the various limbs and their abnormalities including relevant examples from Bateson (1894) and Przibram (1921) are tabulated (Table 1) and representative examples are illustrated diagrammatically (Fig. 1). Where appropriate, we have amplified this description with photographs.

Table 1
graphic
graphic
Fig. 1

Drawings to show the eleven abnormal claws. For further details see Table 1. Bar = 5·0 cm.

Fig. 1

Drawings to show the eleven abnormal claws. For further details see Table 1. Bar = 5·0 cm.

Although not all lateral outgrowths have the same structures, there are several features that most have in common:

  1. All but four of the outgrowths are symmetrical for at least part of their length (Fig. 1). Often the symmetry takes the form of a pair of mirror-image structures which may be fused (e.g. nos 1 – 4) or separate (no. 7) at the base. Exceptions to this are: no. 9, where the outgrowth is neither symmetrical nor divided and is in the form of a bump; no. 11, which is a curly spike lacking clear cuticular markers and thus of uncertain symmetry; no. 10, which has two lateral outgrowths, one narrow and featureless and one divided at the end to form a small pair of mirror image dactylopodite ends.

  2. Where features can be identified on the outgrowths they are always ones that lie distal to the level of the limb from which the outgrowth originates. This can include joints and more distal limb segments (nos. 2, 12, 15, Table 1, Fig. 1).

  3. Where an outgrowth is proximally fused, the base of the structure only includes features from the same side of the circumference as that from which it originates. This is particularly clear in no. 2, where the knobs around the propodite/dactylopodite joint of the outgrowth are typical of the dorsal part of the limb (Fig. 2). In other cases, where there are no clear markers at the site of origin, structures characteristic of other parts of the circumference are never represented at the base of the outgrowth. For instance, teeth, which are found on the side of the limb furthest from the outgrowth (e.g. nos. 3 and 4), are not present in the basal region of the structure. More distal regions of symmetrical outgrowths include features from further round the circumference so that at the bifurcation point where the outgrowth divides into two, two complete circumferences occur (Figs. 1, 3). We refer to this situation as distal expansion after Slack (1980a).
    Fig. 2

    External (left) and internal (right) views of specimen no. 2 showing the origin of the lateral outgrowth. Light coloured featureless cuticle at its base probably indicates the extent of the original damage. The mid-lateral ridges and adjacent cuticle dorsal to them are undisturbed. The knobs at the dactylopodite/ propodite joint are typical of the dorsal part of the limb.

    Fig. 2

    External (left) and internal (right) views of specimen no. 2 showing the origin of the lateral outgrowth. Light coloured featureless cuticle at its base probably indicates the extent of the original damage. The mid-lateral ridges and adjacent cuticle dorsal to them are undisturbed. The knobs at the dactylopodite/ propodite joint are typical of the dorsal part of the limb.

    Fig. 3

    Ventral view of the dactylopodite of specimen no. 5. It shows several features typical of Bruchdreifachbildungen. The more proximal of the lateral outgrowths (p) is longer than the other (d). Not all circumferential levels are represented at the base of the structure. Thus, teeth appear only just below the bifurcation level. On the distal side of the outgrowth there is a pattern discontinuity in the form of a bump (arrow), but on the proximal side the outgrowth emerges smoothly from the dactylopodite. Bar = 1·0 cm.

    Fig. 3

    Ventral view of the dactylopodite of specimen no. 5. It shows several features typical of Bruchdreifachbildungen. The more proximal of the lateral outgrowths (p) is longer than the other (d). Not all circumferential levels are represented at the base of the structure. Thus, teeth appear only just below the bifurcation level. On the distal side of the outgrowth there is a pattern discontinuity in the form of a bump (arrow), but on the proximal side the outgrowth emerges smoothly from the dactylopodite. Bar = 1·0 cm.

  4. Outgrowths can occur on many points of the chelae, including the carpo-podite (no. 12), the propodite (e.g. no. 2) and the dactylopodite (e.g. no. 4). In some instances the outgrowth arises from the joint at the base of the dactylopodite (nos. 6, 7). From these observations and the previous data showing that similar supernumeraries have been recorded on walking legs and virtually all limb segments (we have been unable to find an example of an outgrowth on the ischiopodite) it appears that they can arise from any point on the proximo-distal axis of a limb including the intersegmental membranes of joints.

  5. Outgrowths can occur on the internal (nos. 1, 5), external (no. 10), dorsal (nos. 2, 4) and ventral (nos. 3, 8) faces of the limb.

  6. The bifurcation point can occur at different levels along the proximodistal axis of the outgrowth. It may be at the base with little (no. 14) or no (no. 7) fusion, or it may be in the next segment up the series. Thus in one case (no. 13) the origin of the structure is the meropodite, the carpopodite is fused and the two elements are separate only at the level of the carpopodite/ propodite joint. Small outgrowths consist of bumps (no. 9) or spikes (no. 11) and are undivided.

  7. The distal side of the base of the outgrowth often has a pattern discontinuity where it meets the main part of the limb. If there is a clear pattern at this point, it is symmetrical about this discontinuity. For instance, there is a ‘V’ pattern in the row of teeth in no. 9 (Fig. 4). More often it is simply a bump or groove in the cuticle (nos. 4, 5, Figs. 1, 3).
    Fig. 4

    Ventral view of the distal side of the bump of specimen no. 9. It shows a distortion of the tooth row associated with the polarity reversal on the distal side of the outgrowth, b, bump; d, dactylopodite. Bar = 5·0 mm.

    Fig. 4

    Ventral view of the distal side of the bump of specimen no. 9. It shows a distortion of the tooth row associated with the polarity reversal on the distal side of the outgrowth, b, bump; d, dactylopodite. Bar = 5·0 mm.

  8. The proximal side of the outgrowth is often longer than the distal side and this results in the outgrowth pointing distally (no. 3).

  9. Two specimens with lateral outgrowths moulted while in captivity. The new cuticle of no. 10 is a little damaged so that the tip of the internal lateral supernumerary is broken and lost. This was the most interesting feature of the limb because it was formed into a pair of mirror-image dactylopodite tips (Fig. 1). Nevertheless, comparison of the structure before and after ecdysis suggests that other aspects of the pattern are the same from moult to moult. This is confirmed by no. 9 which is still alive and has moulted in captivity. Here the abnormality is in the form of a bump on the ventral edge of the dactylopodite (Figs. 1, 5). The shape of the structure is almost the same in the cast and in the living specimens, and the tooth pattern remains unchanged (Figs. 4, 5). From the fact that decapods can regenerate a complete limb within a moult cycle (Bliss, 1960) it seems likely that the lateral supernumeraries are formed within a single moult cycle. Our observations suggest that thereafter the major pattern elements remain stable.
    Fig. 5, 6

    Outgrowth on specimen no. 9 at two moult stages. The same pattern of teeth (arrows) is recognizable in the cast (Fig. 5) and in the living specimen (Fig. 6). The general appearance of the outgrowth is virtually unchanged from moult to moult. Bars = 5·0 mm.

    Fig. 5, 6

    Outgrowth on specimen no. 9 at two moult stages. The same pattern of teeth (arrows) is recognizable in the cast (Fig. 5) and in the living specimen (Fig. 6). The general appearance of the outgrowth is virtually unchanged from moult to moult. Bars = 5·0 mm.

As the claws were all recovered from wild specimens the cause of the deformities remains uncertain. They are unlikely to be due to a genetic abnormality in the pattern forming mechanism, because that would probably result in much more widespread deformities. Damage during embryogenesis or damage to a limb bud reforming after a limb had been autotomized would also tend to produce more widespread effects. Much more likely is damage to the limb during fighting, possibly while the limb is still soft after moulting, followed by regeneration. Shortly after moulting the cuticle of H. gammarus is very brittle. Squeezing a limb with forceps at this stage causes splits in the exoskeleton and the damaged region can be pulled away easily (Truby, unpublished observations). When two decapods fight they often grasp each other by the chelae. It is easy to imagine how such behaviour could produce localized damage to the claws or their intersegmental membranes. Support for this hypothesis comes from the observation that the abnormalities have been found found mainly on the chelae and that most cases involve outgrowths on the claws or preceding segment. It is, of course, possible that such abnormalities occur with similar frequency on more proximal segments but that major deformities there increase the likelihood of autotomy.

The idea that damage causes the outgrowths is supported by the views of other workers. Huxley (1884) reported lateral outgrowths on crayfish limbs and suggested that they are due to regeneration following limb damage in newly moulted animals. Bodenstein (1953) concluded that naturally occurring triple leg structures in insects are caused by a distal portion of the leg being partially broken off followed by distal regeneration from both the proximal and distal wound surfaces. Similarly, Przibram (1921) concluded that Bruch-dreifachbildungen, similar to the ones we have described, are caused by regeneration from both the proximal and distal sides of a wound in the side of the limb. There is now direct evidence from experiments with cockroaches to support the general principle that lateral regenerates are caused by wounding (Bohn, 1965). Bohn cut V-shaped notches in the ventral and dorsal faces of tibiae of Leucophaea maderae (Fabr.). Although the wound normally healed perfectly or left only a small bump, some of the ventral notches gave a range of Bruchdreifachbildungen of the same type that we have described for crustacea. These may include raised bumps at the site of the injury, mirrorimage lateral outgrowths of tibial tissue, similar structures but also including the tibia/tarsal joint and some tarsal structures terminally. In some cases regeneration resulted in the formation of all proximodistal levels including the limb tip. In such structures there is a range of types from those where there is a bifurcation at the tip to form two separate sets of tarsal claws to those where the bifurcation point is at the base of the structure before the first joint. The dorsal notches never produced more than short outgrowths of tibial tissue. This difference between the effects of ventral and dorsal wounding may not reflect a physiological difference between cells at the two sites for the following reasons. First, there are reported cases of naturally occurring dorsal outgrowths from insect legs showing regeneration as far as the limb tip (Przibram, 1921, specimen nos. 173 a, b). Second, in the crustaceans, which are probably built according to similar rules, complete outgrowths can occur at any point on the circumference (see above). It will be shown that our explanation for the phenomenon depends on failure of wounds to heal. Differences in the curvature of the limb surface may well affect this process. We think that differences in regeneration behaviour at different points on the insect leg may be due to some mechanical factor. It is important to establish this point because, according to a polar coordinate model (French et al. 1976), one would expect similar types of cellular behaviour at all points on the circumference. If, however, cells are specified with reference to a Cartesian system of coordinates, one might expect cells on opposite sides of the limb to exhibit different types of behaviour. For instance, they may be able to regenerate only in one direction along a particular axis (see for instance Slack, 1980a, b).

Another special case of ‘wounding’ giving rise to pattern duplications or triplications is known in Drosophila imaginal discs. Here, temperature-sensitive cell-lethal mutants can be used to produce localized cell death in the discs (Girton & Russell, 1980). Following such damage to the leg discs, structures remarkably similar to those described by Bohn (1965) for the cockroach have been produced in Drosophila (see Girton & Bryant, 1980).

Regarding the general problem of the mechanism generating Bruchdreifach-bildungen, there has been no really satisfactory explanation. Why are the results so variable and why does a notch cause the phenomenon infrequently? There have been numerous studies on the insect leg and from, them certain consistent facts emerge. First, when tissues from different proximodistal levels are recombined, tissues normally separating those levels are regenerated by intercalation. Cells forming the regenerate are derived from both proximal and distal faces (Bohn, 1976). A similar pattern of intercalary regeneration follows when tissues from different points around the circumference are confronted (French, 1978). Intercalation in that instance is by the shortest possible route (French, 1978) so that a slightly damaged limb repairs itself rather than regenerates a mirror-image copy. These findings have led to the formulation of the ‘clockface’ or polar coordinate model for explaining distal regeneration (French et al. 1976). Complete distal regeneration occurs when a complete circumference is exposed or can be formed by intercalary regeneration from sections of the circumference. According to the formal model (French et al. 1976) distal transformation occurs only when these conditions are met. However, distal regeneration can occur in other circumstances. In insects, experimentally created mirror-image symmetrical lateral limb outgrowths are capable of some regeneration (French, 1976 a). In amphibians, mirror-image symmetrical limbs show varying abilities to regenerate. In the axolotl, double-posterior limbs give complete distal regeneration when amputated (Slack & Savage, 1978). In the newt Notophthalmus viridescens, amputated double-half limbs show partial regeneration. Here the distal regeneration is elicited by confrontation of non-congruent circumferential values (Bryant & Baca, 1978). In this case only a few rounds of intercalation are necessary to resolve all pattern discontinuities. However, Bryant & Baca (1978) still maintain that a complete circumference is necessary for total distal regeneration. In insects, after telescoping experiments where congruent proximodistal tibial levels are combined, lateral regenerates can form if there is partial failure of the host/graft junction to heal (French, 1976a). Our interpretation of all these phenomena is that regeneration follows wherever wounds fail to heal. Where a complete circle forms, healing is effectively prevented by intercalation across the clockface and the production of an unresolvable point at the limb tip (French, 1976a). Healing is known to be inhibited when cells from different positions of an axis are confronted in insect grafting experiments (Niibler-Jung, 1977). This can lead to rounding up or even rejection of the graft. According to our hypothesis, amputated mirror-image symmetrical limbs regenerate because mechanical factors prevent the exposed parts of the circumference from coming together. To explain Bruchdreifachbildungen it is necessary to assume that regeneration occurs when wounds fail to heal and that, where complete circles are formed, distal regeneration will continue until terminal structures have formed. The clockface model by itself fails to explain the range of structures formed after notching (Bohn, 1965). It predicts only two outcomes. When the notch is shallow the limb should repair itself, when it is deep two completely separate laterals should be produced with no fusion at the base. In the cases we have described, there is only one instance (no. 7) where the two supernumeraries are separate. In the others, the bifurcation point is a considerable distance from the base. This was also true of Bohn’s (1965) examples in the cockroach. In addition, examination of some of our specimens shows that the original damage was localized to much less than half of the circumference. A clear case is no. 2 (no. 1 is also derived from a small wound). This shows a symmetrical propodite/dactylopodite joint and a mirror-image pair of dactylo-podites arising from the dorsal side of the propodite. The base of the outgrowth occupies no more than 20% of the circumference of the main limb axis. In addition, the raised mid-lateral ridges which occur on the internal and external faces are undisturbed (Fig. 2). The base of the regenerate, which shows mirror-image symmetry, consists of tissues normally occurring in the dorsal third of the limb. This argues that notches extending much less than half-way round the circumference can produce lateral regenerates. According to the clockface model, damage to a small region of the circumference could cause laterals to form if positional values are not evenly distributed around the circumference. Local damage at a site where the values are clustered could expose more than half the circumferential levels. However, in that case the supernumeraries should be separate at the base (see below). In all the cases but one (no. 7), that we have described, the laterals are fused at the base. Finally, the clockface model does not explain how extra circumferential positional values are intercalated to cause the common phenomenon of distal expansion.

Conditions necessary for the production of Bruchdreifachbildungen

Except in the case of the production of two completely separate mirror-image laterals it is necessary to assume that the wound remains open after notching. Although the notch may be sealed with a clot of haemolymph, the edges of the wound may remain apart for a considerable period of time after injury. Where physical contact of the two sides of confronted tissues fails to occur, each side may regenerate independently of the other. When proximal and distal levels of a cockroach limb are telescoped together, both faces usually contribute to the intercalary regenerate, with the distal face forming more proximal levels and the proximal face forming more distal levels (Bohn, 1976; French, 1976a). However, if the two cut surfaces fail to establish cellular contact, both surfaces behave like distally amputated limbs and, even with congruent grafts, the proximal face completes the limb and the distal face regenerates a mirror-image (along the proximodistal axis) of the original graft (Bohn, 1965; French, 1976a). Thus, although distal regions can give rise to more proximal ones, when the surface is free, a cut surface always regenerates more distal parts of the proximodistal axis. There is no evidence concerning the direction of regeneration around the circumference when a particular circumferential level is exposed and prevented from joining the other side of the wound. However, we believe that it can regenerate and that it can regenerate in either direction around the circumference. Only by making these assumptions can all Bruchdreifachbildungen be explained. According to the shape of the notch and position of cells on the cut edge, the exposed faces may behave like distally amputated surfaces or exposed parts of the circumference. If the cells along the cut surface have serial circumferential values then regeneration will produce more distal structures. If they have serial proximodistal values then regeneration will result in new circumferential values being formed (Fig. 7). In a situation where the surface is at an angle to both axes, they will both have serial values. Therefore regeneration will proceed along both axes. In the following account, for simplicity, we have considered only regeneration along one of the two axes. Thus for a steep-sided notch we have considered regeneration along the proximodistal axis and for notches with a shallower angle we have considered regeneration around the circumference. Later we will describe what happens when regeneration along both axes is considered at the same time.

Fig. 7

Regions of the limb cylinder can be arbitrarily defined in terms of circumferential and proximodistal coordinates (position values) (A) (see French et al. 1976). Different shaped notches (B, C) can expose edges in which either one or both sets of position values are arranged serially. A transverse cut exposes serial circumferential values only (D). Such a surface will always regenerate distally irrespective of stump polarity (D, E). A cut parallel to the proximodistal axis (F) exposes serial proximodistal values. It is proposed that regeneration of circumferential values may be in either direction around the circumference (G, H).

Fig. 7

Regions of the limb cylinder can be arbitrarily defined in terms of circumferential and proximodistal coordinates (position values) (A) (see French et al. 1976). Different shaped notches (B, C) can expose edges in which either one or both sets of position values are arranged serially. A transverse cut exposes serial circumferential values only (D). Such a surface will always regenerate distally irrespective of stump polarity (D, E). A cut parallel to the proximodistal axis (F) exposes serial proximodistal values. It is proposed that regeneration of circumferential values may be in either direction around the circumference (G, H).

The production of bumps or spikes undivided at the tip

Spikes or bumps are likely to arise where the two sides of the notch are at an acute angle to one another so that each surface behaves like a distally amputated limb. In this case we will consider regeneration along the proximodistal axis only. The proximal and distal faces of a notch may fail to fuse after notching because of the mechanical factors involved. For instance the remaining intact side of the limb will sometimes hold the two edges apart. Following the rule that a free cut end always regenerates distally (see above), both proximal and distal sides of the wound will regenerate in a distal direction. Since the two faces of the notch confront each other the regenerating faces will eventually meet (Figs 7 & 8). At the base of the notch the two faces will meet almost immediately. Consequently there will be no great discrepancy of proximodistal values at the fusion point. Further out along the sides of the notch, regeneration will proceed a considerable way until the two sides meet. This results in the confrontation of tissues from significantly different proximodistal levels. Intercalation of intervening values will follow to produce a stable bump (Fig. 8) or spike (specimen no. 11) which would persist unchanged from moult to moult as observed in specimen no. 9 (Figs. 5, 6). Clearly the nature and size of the bump would depend on the geometry of the original notch. Observable features of our specimens are consistent with this explanation. Distal to the base of the outgrowth there is often a pattern discontinuity in the form of a groove (no. 4) or a bump (Fig. 3) while on the proximal side the junction between the main limb and the sidegrowth is not visible. This is consistent with distal regeneration from the two sides of the notch which produces a polarity reversal on the distal side but not on the proximal side (Fig. 4). We have noted that the lateral outgrowths often point distally (specimen no. 11 and many of the mirror-image divided structures) (Fig. 1). This follows from our explanation because the proximal side of the wound has to form more distal levels before reaching the tip than the distal side.

Fig. 8

A bump may be explained with reference to the proximodistal axis only. Proximodistal coordinates showing the position of the notch (A). The cut surfaces at the top of the notch are held apart (B). Distal regeneration occurs from both free cut surfaces (C). Regenerated tissues meet and fuse (D). When non-congruent values i and q meet, intercalation of intermediate values occurs to produce a bump (E). Note that the values m to q are represented three times, the centre set being a mirror-image of each of the outer sets.

Fig. 8

A bump may be explained with reference to the proximodistal axis only. Proximodistal coordinates showing the position of the notch (A). The cut surfaces at the top of the notch are held apart (B). Distal regeneration occurs from both free cut surfaces (C). Regenerated tissues meet and fuse (D). When non-congruent values i and q meet, intercalation of intermediate values occurs to produce a bump (E). Note that the values m to q are represented three times, the centre set being a mirror-image of each of the outer sets.

A feature of the theory is that it does not necessarily involve consideration of the circumferential positional values. Assuming that each side of the notch regenerates independently until fusion, we expect the newly regenerated cells to derive their circumferential values from the free cut surface just as they do when a distal amputation occurs. When the two sides of the wound meet during regeneration the circumferential values should be approximately in register.

The production of two complete lateral regenerates without fusion

The most complete mirror-image laterals, unjoined at the base and consisting of all levels distal to the wound site, require a different explanation. It is that the notch penetrated more than halfway through the appendage. In this case we do not believe that opposite sides of the wound are necessarily held apart. Fusion of the two lateral faces, where they are closest together at the base of the notch, results in the intercalation of intervening circumferential values by the shortest possible route across the middle of the damaged aiea. This creates two sets of circumferential values on the side of the limb (Fig. 9). The wound will effectively remain unhealed at the centre of each set because of the unresolvable points there. Distal regeneration will follow to completion.

Fig. 9

Production of two sets of circumferential values in the side of the limb after a deep notch. The diagram shows the outline of the notch as seen from above with intercalation across the middle and free regeneration (arrows) to produce two sets of circumferential values. This will produce two complete lateral regenerates without fusion. Shaded area = open wound.

Fig. 9

Production of two sets of circumferential values in the side of the limb after a deep notch. The diagram shows the outline of the notch as seen from above with intercalation across the middle and free regeneration (arrows) to produce two sets of circumferential values. This will produce two complete lateral regenerates without fusion. Shaded area = open wound.

Bruchdreifachbildungen fused at the base and showing distal expansion

Wherever the two lateral supernumeraries are completely separate at the base, the inference must be that the notch extended more than halfway across the circumference. If the notch extends less than halfway and the damage is followed by the two lateral edges of the wound coming together, then the shortest intercalation rule (French et al. 1976) requires the missing circumferential values to be intercalated. The result is local repair without lateral supernumeraries. This is probably the normal outcome of local damage to a restricted part of the circumference. Nevertheless in some of our specimens the original damage was highly localized on one side of the limb and yet laterals were produced. In these cases the Bruchdreifachbildungen are fused at the base and distal expansion results in complete separation of the two laterals at a more distal level. In this case we believe that the original notch was shallow and/or had sections of the lateral wound surface approximately parallel to the proximodistal axis of the limb. It is assumed that the wound remains unhealed and that regeneration begins at the lateral edges. We also assume that such an exposed edge may regenerate in either direction around the circumference. Each edge will regenerate without reference to the other until the two sides finally meet (Fig. 10). After the lateral edges meet, we predict intercalation by the shortest route around the circumference if there is a confrontation of non-congruent values. This could have three possible outcomes. First, the two edges can regenerate towards each other. In this case there will be repair of the wound without lateral regenerates because, when the two sides meet, we expect near or complete congruences of values. Second, the two edges can regenerate values in the reverse direction (Fig. 10). If such regeneration proceeds sufficiently far, values more than halfway round the circumference from the centre of the original wound can be formed at the exposed edges. When they meet, intercalation at the interface will occur until two complete sets of circumferential values have been created. Following our previous reasoning, the inherent instability of this situation allows distal regeneration to completion. Note that, at the base of such a structure, the two laterals will each consist of less than half of the circumferential values and they will be joined together in mirror image symmetry. The distal expansion can be explained by the well established intercalation rule and does not require special new rules (see for example Slack, 1980a, b). The third possibility is that on one side of the wound, regeneration proceeds in one direction and that on the other side, it is in the reverse direction. This would not result in lateral regenerates because, assuming equal rates of regeneration from each edge, the relative difference in circumferential values will be approximately the same as the difference between the original exposed surfaces. However, intercalation after fusion will result in a bump at the site of the notch.

Fig. 10

A notch seen from above cutting the proximodistal axis at a shallow angle. Free regeneration (solid arrows) of values in the reverse direction around the circumference proceeds to a point where, after fusion, intercalation (dotted arrows) produces two sets of circumferential values. The resulting structure will be fused at the base and will bifurcate at a more distal level. Dashed lines in D join points having the same circumferential values. (E) Plan view of the resulting structure as it would appear from the side.

Fig. 10

A notch seen from above cutting the proximodistal axis at a shallow angle. Free regeneration (solid arrows) of values in the reverse direction around the circumference proceeds to a point where, after fusion, intercalation (dotted arrows) produces two sets of circumferential values. The resulting structure will be fused at the base and will bifurcate at a more distal level. Dashed lines in D join points having the same circumferential values. (E) Plan view of the resulting structure as it would appear from the side.

So far we have considered regeneration along each of the two axes separately and we have chosen ideal examples where the exposed surface is nearly parallel to one of the axes and cuts the other at an acute angle. However, in most notches the wound will be more or less U shaped when viewed from the side. So, in some places along the wound margin a given axis will be cut at a shallow angle and in others it will be cut at an acute one. For this reason the resulting outgrowth will have features of bumps and of laterals with distal expansion. Thus, an outgrowth may have a polarity reversal (with respect to the proximodistal axis) at the distal side of the lateral’s base, but also show distal expansion (Fig. 3).

In this discussion we have attempted both to explain the phenomenon of Bruchdreifachbildungen and to provide an explanation for distal regeneration which does not depend upon the complete circle rule (French et al. 1976) alone. This is because the complete circle rule demonstrably fails in the case of the distal regeneration of mirror-image symmetrical limbs (French, 1976; Slack & Savage, 1978) and because by itself the complete circle rule cannot explain the range of structures found in the Bruchdreifachbildungen we have described for crustaceans and Bohn (1965) has observed in insects. After due consideration, we concluded that the main cause of regeneration in limbs is the failure of wounds to heal. Distal regeneration following the formation of complete circles is just a special case of wounds failing to heal. It may be brought about because of the unresolvable point at the centre of a complete circumference or because cells on opposite sides of the circumference have sufficiently different surface properties that healing is inhibited. Our hypothesis could explain why amputated mirror-image symmetrical limbs are capable of distal regeneration. All that is required is regeneration of all values distal to the cut before the cells at the circumference come together at the tip. This is quite possible; most fields are small at the time of determination (Wolpert, 1969; Crick, 1970) and subsequent cell divisions merely increase the size of the organ. In addition, just the geometry of an amputated cylinder is not conducive to healing at the tip. An important point deriving from our argument is that it is immaterial which circumferential values are represented in the wounded surface. For this reason we can explain the regeneration of anomalous mirror-image symmetrical lateral supernumeraries in amphibians after 180° rotations of the limb (Maden, 1980). In that case we propose that, after the operation, the wound fails to heal over part of the circumference due to some obstruction (possibly a blood clot) or perhaps to a misalignment of host and graft tissues. The partial set of values intercalated around the wound will have mirror-image symmetry and distal regeneration will result in these values being perpetuated until either the most terminal levels are formed or the wound finally heals. In conclusion, we can summarize our explanation as follows. A cut free surface will sequentially regenerate values along the axis orthogonal to the cut. Any exposed part of the circumference may begin to regenerate so long as the surface remains free. In the case of the proximodistal axis regeneration from a free surface will always proceed in a distal direction while in the circumferential axis it can proceed in either direction around the circumference. Clearly our ideas are speculative and they are based on a rationalisation of phenomena exhibited by a set of naturally occurring monstrosities. Nevertheless, they provide a possible way out of the complete circle impasse and they suggest the need for further experiments and accurate observation of cell behaviour at wound surfaces.

We thank Angela Chorley for drawing the chelae and preparing Fig. 1. We are also grateful to Dr P. J. Hogarth for helping with our literature search. P.R.T. was supported by an M.R.C. studentship.

Bateson
,
W.
(
1894
).
Materials for the Study of Variation
.
London, New York
:
MacMillan
.
Bliss
,
D. E.
(
1960
).
Autotomy and regeneration
.
In Physiology of Crustacea
, vol.
1
(ed.
T. H.
Waterman
), pp.
561
589
.
New York, London
:
Academic Press
.
Bodenstein
,
D.
(
1953
).
Regeneration
.
In Insect Physiology
(ed.
K. D.
Roeder
), pp.
886
878
.
New York
:
Wiley
.
Bohn
,
H.
(
1965
).
Analyse der Regenerationsfâhigkeit der Insekten-extremitat durch Amputations-und Transplantationsversuche an Larven der afrikanischen Schabe Leuco-phaea maderae Fabr. (Blattaria). 1. Regenerationspotenzen
.
Wilhelm Roux Arch. Entw-Mech. Org
.
156
,
49
74
.
Bohn
,
H.
(
1976
).
Regeneration of proximal tissue from a more distal amputation level in the insect leg (Blaberus craniifer, Blattaria)
.
Devi Biol
.
53
,
285
293
.
Bryant
,
S. V.
&
Baca
,
B. B
, (
1978
).
Regenerative ability of double-half and half upper arms in the newt, Notophthalmus viridescens
.
J. exp. Zool
.
204
,
307
324
.
Crick
,
F. H. C.
(
1970
).
Diffusion in embryogenesis
.
Nature, Lond
.
225
,
420
422
.
Emmel
,
V. E.
(
1908
).
The experimental control of asymmetry at different stages in the development of the lobster
.
J. exp. Zool
.
189
,
421
427
.
French
,
V.
(
1976a
).
Leg regeneration in the cockroach, Blatella germánica. I. Regeneration from a congruent tibial graft/host junction
.
Wilhelm Roux’ Arch, devl Biol
.
179
,
57
76
.
French
,
V.
(
1976b
).
Leg regeneration in the cockroach, Blatella germánica. IL Regeneration from a non-congruent tibial host/graft junction
.
J. Embryol. exp. Morph
.
35
,
267
301
.
French
,
V.
(
1978
).
Intercalary regeneration around the circumference of the cockroach leg
.
J. Embryol. exp. Morph
.
47
,
53
84
.
French
,
V.
,
Bryant
,
P.
&
Bryant
,
S.
(
1976
).
Pattern regulation in epimorphic fields
.
Science, N.Y
.
193
,
969
981
.
Girton
,
J. R.
&
Bryant
,
P. J.
(
1980
).
The use of cell lethal mutations in the study of Drosophila development
.
Devl Biol
.
77
,
233
243
.
Girton
,
J. R.
&
Russell
,
M. A.
(
1980
).
A clonal analysis of pattern duplication in a temperature-sensitive cell-lethal mutant of Drosophila melanogaster
.
Devl Biol
.
77
,
1
21
.
Huxley
,
T. H.
(
1884
).
The Crayfish
.
London
:
Kegan Paul, Trench
.
Maden
,
M.
(
1980
).
Structure of supernumerary limbs
.
Nature, Lond
.
286
,
803
805
.
Mittenthal
,
J. E.
(
1978
).
Intercalary regeneration in crayfish legs
.
Am. Zool
.
18
,
610
.
Nübler-Jung
,
K.
(
1977
).
Pattern stability in the insect segment. 1. Pattern reconstitution by intercalary regeneration and cell sorting in Dysdercus intermedins Dist
.
Wilhelm Roux Arch, devl Biol
.
183
,
17
40
.
Paul
,
J. H.
(
1914
).
Regeneration of the legs of decapod Crustacea from the preformed breaking plane
.
Proc. R. Soc. Edinb
.
35
,
78
91
.
Przibram
,
H.
(
1921
).
Die Bruchdreifachbildung im Tierreiche
.
Wilhelm Roux Arch. Entw-Mech. Org
.
48
,
205
444
.
Slack
,
J. M. W.
(
1977
).
Control of anteroposterior pattern in the axolotl forelimb by a smoothly graded signal
.
J. Embryol. exp. Morph
.
39
,
169
182
.
Slack
,
J. M. W.
(
1980a
).
Morphogenetic properties of skin in axolotl limbs
.
J. Embryol. exp. Morph
.
58
,
265
288
.
Slack
,
J. M. W.
(
1980b
).
A serial threshold theory of regeneration
.
J. theor. Biol
.
82
,
105
140
.
Slack
,
J. M. W.
&
Savage
,
S.
(
1978
).
Regeneration of mirror symmetrical limbs in the axolotl
.
Cell
14
,
1
8
.
Wolpert
,
L.
(
1969
).
Positional information and the spatial pattern of cellular differentiation
.
J. theor. Biol
.
25
,
1
47
.