Neosorny in fleas, and the sessile life-style

J. Zool., Lond. ( 1992) 226, 6 1 3-629 Neosorny in fleas, and the sessile life-style MIRIAM ROTHSCHILD Ashton Wold, Peterborough, UK (Accepted 22 May 1991) (With 14 plates in the text) Fleas arc basically free-living parasites. and the sessile life-style is an evolutionary after-thought. The specializations described are somewhat similar in the various neosomes, only they have been developed within restraints imposed by the family strait-jacket and the contrasting characters and habits of different hosts. Contents Page Introduction................................ 613 Material and methods............................ 614 Classification of the neosomic fleas...................... 615 Discussion.................................. 626 Conclusions................................ 628 Relercnces.................................. 628 Introduction Neosomy was defined by Audy er al. (1972) as the formation of new external morphological structure and taxonomically significant enlargement, resulting at least in part from the secretion of new cuticle during a single active stadium of an invertebrate in a group that normally changes in external form only through a moult. These authors give examples from fleas, gravid queen ants, certain flies, mites, ticks and copepods. Jordan (1 962) had previously described the process in Tunga (Siphonaptera) as teleomorphosis of form affecting function. He cited as further examples of post metamorphic growth the intersegmental membranes of queen termites, repletes of honey-ants and the oil beetle (Meo2). I believe neosomy is a more convenient term for describing this phenomenon, although, as usual, Jordan had priority. Fleas are, broadly speaking, free mobile parasites, but about 90 species, chiefly among the Pulicoidea, have adopted a sessile or a semi-sessile life-style and the female (very occasionally the male also) remains, after eclosion, more or less permanently attached to the integument of the host; in the genus Tunga and Neotunga the female burrows beneath the skin and is endoparasitic. Neosomy associated with the sessile life-style has been developed independently in several unrelated super-families of fleas. Like mimicry in the Lepidoptera it is an evolutionary afterthought. One would therefore expect to find striking analogous specializations resulting from the sedentary life-style, modified by the type of host involved, but developed within the restraints imposed by ancient family characters. Such restraints are well illustrated in the mimetic 613 0 1992 The Zoological Society of London

614 M. ROTHSCHILD PLATE I. EchidnopliuRu mymiecoheii. A male stick-tight flea with a typical Pulicoid body-build. Note the contracted thorax, powerful vertical notal ridge (N) and angled frons. The pleural arch (P) is ventrad in position. Lepidoptera. Thus the speedy aposematic hawk-moths (Sphingidae) mimic fast-flying models such as bumblebees or bluebell flies, while the white butterflies (Pieridae) and their Mullerian mimics flutter about slowly and conspicuously in the sunshine. In the past, attention has centred almost exclusively on the external anatomy of neosomic fleas (Ioff, 1950; Hopkins& Rothschild, 1953; Jordan, 1962; Smit, 1962), but it has now been possible to examine the internal organs of species from two unrelated families. Materials and methods The specimens, originally preserved in alcohol in the Rothschild collection, are now desiccated and thus unsuitable for histological study. Fortunately, Professor S. G. Medvedev was able to send me 2 neosomic female specimens of Vermipsyllu ulukurt preserved in alcohol. Although this is not an ideal fixative for the production of serial sections (furthermore the fleas had been kept in this preservative for 40 years), using the technique previously described (Rothschild, Schlein & Ito, 1986), satisfactory sections were obtained and the principal internal organs could be examined. Specimens of Tungu monositus received from Professor F. J. Radovsky, fixed in Bouin Brazil, proved ideal material for serial sectioning. Six thousand specimens of the semi-sessile rabbit flea (Spilopsyllus cuniculi) which were scctioned by the same method were available for comparison.

NEOSOMY IN FLEAS 615 PLATE 11. Nosr)~s,.llu.s./usciu,us (male). A characteristic Ceratophylloid body-build with extended thorax and abdomen, a slender slanting notal ridge (N) and rounded frons. The pleural arch (P) is notal in position. Gravid female ants (Myrrnicu ruhra) were provided by Jeremy Thomas and the queen termite (Mucrorermes suhliyulinu.~) by the Department of Entomology (British Museum of Natural History). Sections of thc former specimens were stained with a triple stain (Heidenhains Azan) and the latter with haematoxylin, counter stained with eosin. and a comparison made of their intersegmental membranes. Photographs werc taken by the author with a Plan 2.5/0.08 (whole fleas), or a Plan 25/0.45 or a Neofluar 40/0.75 with a KPL 8 x eyepiece. Enlargements were varied during processing. The film was a colour reversal Kodak Ektachrome 50 Tungsten. Classification of the neosomic fleas Smit (1982) split Jordan s three super-families (see Traub, Rothschild & Haddow, 1983) of the Siphonaptera into five super-families. Neosomic species are found in the Pulicoidea, Vermipsylloidea and Malacopsylloidea. Body-build und the position of the pleural urch The strait-jacket of family characters is most evident in the contrast between body-build of the Pulicoid semi-sessile and sessile fleas and those of the Vermipsylloidea. This fundamental difference was stressed before Smit (1982) split the Order into five super-families and the Vermipsyllidae were at that time still included in the Ceratophylloidea (Hopkins & Rothschild, 1953; Rothschild. 1964).

616 M. ROTHSCHILD PLATE 111. Lacinia of female V. ulukurt. showing slight serrations The Pulicoids are more closely associated with the bodies of their hosts and are found in their fur or feathers, less frequently free in their nests or lairs. They have a more compact build, due to the closer approximation of the segments, which results in a 'dumpy' body with a shorter abdomen and contracted thorax, and legs adapted for the initial leap from the substratum to the passing animal but not for perambulation away from the host (Plate I). Ceratophylloids (sensu la&) have more elongated, less compact bodies (Plate 11), with a supple thorax and rangy legs adapted for walking or crawling on uneven surfaces. In the Pulicoid fleas the pleural arch is situated nearer the lateral centre of the thorax, or may even be ventrad in position. This entails a short stout pleural rod and a correspondingly elongated vertical metanotal ridge (Plate I). In the majority of other fleas, the pleural arch is situated near to the dorsum (Plate 11) and consequently the pleural rod is long and straight, whereas the metanotal ridge is short and oblique, sometimes almost horizontal. In the neosomic females these family tendencies are clearly delineated. Thus the bodies of the embedded Tunga are roughly spherical, whereas those of Vermipsylla and the related Dorcadia are elongated and maggot-like. The abdomen of Neotunga is more worm-like, although the head and thorax are definitely Pulicoid. Head, mouthparts and antennae The semi-sessile and sessile Pulicoid fleas are almost all characterized by an angled frons (Plate I) which allows a close application of the head to the feeding site and point of attachment. The angled frons is lacking in a few Hectopsylla and in the Vermipsylloidea. In the Malacopsylloidae only one of the two known species has an angled frons (see figs 2 and 15 in Smit, 1987). The Pulicoid sessile and semi-sessile fleas are anchored in position by their laciniae which are broad, heavily-serrated blades (Plate IV). There is a tendency throughout the super-family to find

NEOSOMY IN FLEAS 617 PLATE IV. Lacinia of Cediopsyllo wpolim. showing strong serrations characteristic of sedentary Pulicoid fleas. slightly serrated laciniae even in 'free' mobile species like the cat flea (CtenocephulidesSelis). In addition, the epipharynx of Neotungu is also serrated. Ceratophylloid fleas in general have narrow laciniae lacking coarse serrations. In Verrnipsyllu the mouthparts are unusually elongated and only lightly serrated (Plate 111). In the related genus Dorcadiu only the tips of the laciniae are toothed. The Malacopsyllids both have feebly serrated laciniae but their modified legs and highly developed and enlarged hook-like tarsal setae suggest they anchor to their hosts by means of their feet rather than their mouthparts (see figs 3-7 in Smit, 1987). Several authors have suggested that the unusually long laciniae of the Vermipsyllids (see fig. 197 in Hopkins & Rothschild, 1956) are necessary to support the weight of the sessile neosomic female body, but this feature could also be attributed to the need to penetrate the thick pelt and integument of mountain sheep which are their principal hosts. Male fleas which fertilize the female while she is attached to the host, and hence immobilized, lack suckers on the clasping surface of their antennae (Rothschild & Hinton, 1968). Such organs are absent, for example, in Tungu, Echidnophagu, the semi-sessile rabbit fleas, and Hectopsyllu (Pulicoidea). However, they are well developed in male Vermipsyllu alakurt (Vermipsylloidea), Mulucopsyllu grossiiientris and Phthiropsyllu ugenoris (Malacopsylloidea), and one can assume that copulation may also occur before the female fixes to the host, and still requires restraint in order that insemination can be satisfactorily effected. The prouentriculous and salivary glands A baffling specialization of a number of the sessile species is the reduction in number (often associated with thickening) of the proventricular spines. Tunga monositus and an unnamed sibling

618 M. ROTHSCHILD PLATE V. Proventriculus of female V. ulukuri with numerous fine spines (longitudinal section). PLATE V1. Proventriculus of female E. gullinuceu with reduced number of thickened spines (longitudinal section).

NEOSOMY IN FLEAS 619 species (collected by R. Traub) are the only known fleas altogether lacking such spines, and they are greatly reduced in other members of the genus. The same applies in a lesser degree to the rabbit fleas Spilopsyllus and Cediopsyllu, and the stick-tight fleas Echidnophuga (Plate VI). In Chuetopsylla some species have reduced spines and in others they are numerous and small. Yet in the female neosome of V. alukurt (Plate V), Neotunga, Malacopsylla and Phthiropsylla, the proventricular spines are numerous, sometimes unusually so, closely set and slender. An equally curious contrast can be found in the salivary glands, for two extremes are met with in the sessile Pulicoid fleas. In T. monositus these organs are the largest known, with huge polyploid cells, ramifying through the body, while those of Echidnophuga gallinacea are the smallest yet recorded, with only four small polyploid cells apiece. There are 14-24 large polyploid cells in the female rabbit fleas salivary glands (see figs 43-47 in Rothschild et a/., 1986). Those of V. alakurt are small, compact, pear-shaped organs with very long narrow ducts resembling Ceratophyllid glands. Exceptionally, they stain a uniform blue colour (with Heidenhain s Azan triple stain) apparently lacking tissues and secretions which, in other fleas, stain bright red (see figs 40-42 in Rothschild et ul., 1986). Rectul umpullue A morphological modification which seems to be connected with the sessile life-style is a reduction in the rectal ampullae (Rothschild, 1976, 1988). In Tungu and Echidnophugu* they are reduced from the normal six to two in number, and in Spilopsjdlus and related genera and the Vermipsyllids they are smaller than is normally the case and appear structurally degenerate. The oiliduct und nerve cord In the semi-sessile rabbit fleas, the gravid female leaves the host to lay in the nest but Vermipsyllu, Dorcudiu, Echidnophuga and Tungu oviposit while still attached to, or free on the body of, the host. The oviduct is then modified to expel the eggs forcibly. In the case of T. monositus. the end portion of the oviduct and vagina are lined with cilia-like setae** (see fig. 76 in Rothschild et ul., 1986) which sweep them outwards with the assistance of powerful muscles. In Vermipsjdlu ulukurt the muscles of the oviduct contract sharply so that the oviduct closes behind each egg, catapulting it forwards as it passes to the exterior (see also Jordan, 1962: 362-365). Closely connected with the spherical body-build of the neosomic female Tunga is a reduction in length of the ventral nerve cord. This is achieved by the shortening of the connectives and the rounded contours of the ganglia (Plate VIII). The reverse situation pertains in the maggot-shaped, elongated abdomen of the female Vermipsyllids in which the connectives of the ventral nerve cord are greatly extended and the ganglia are lozenge-shaped, long and narrow, not circular in outline (Plate VTI). The intersegmental membranes The exoskeleton of the adult flea, including the head and mouthparts, does not grow after eclosion and is not involved in the enormous expansion of the intersegmental membranes. In both * F. G. A. M. Smit inlbrms nic that three African species have the normal six rectal ampullae. ** Incorrectly described as spines in Rothschild CI d.. 1986.

620 M. ROTHSCHILD PLATE VI1. Elongated lozenge-shaped ganglia and elongated connective (C) of ventral nerve cord of female V. ulukurt. PLATE VIII. Rounded thoracic ganglia and shortened connective (C) of female T. munosirus. the neosomic females of Tunga, Neotunga and Vermipsylfa the terga and sterna remain like small straps of tanned cuticle (very heavily tanned in the latter genus) atop and beneath the grotesquely enlarged abdomens (see fig. 196 in Hopkins & Rothschild, 1956). Unfortunately, freshly hatched specimens were only available for Tunga monositus and the semi-sessile rabbit flea Spilopsyllus cuniculi. The intersegmental membranes in the newly emerged female T. monositus are deeply infolded,

NEOSOMY IN FLEAS reminiscent of a gathered or ruched piece of material, tucked in tightly beneath the terga and sterna. In both Spilop.sy1lu.s and Echidnophugu, semi-sessile species with large sclerites unusually tough and hard, the swelling of the gravid female abdomen is also achieved by stretching of infolded intersegmental membranes. It is interesting that in the rabbit flea the expansion of the abdomen is controlled by the sex hormones of the host for it occurs only on a pregnant doe host or new-born young h&re ripening eggs are present to exert any internal pressure (Rothschild & Ford, 1973, PI. la). In T. monositus, once the female is embedded beneath the skin of the host, the body wall expands very rapidly and all traces of folding of the intersegmental membranes vanish, but again, egg production occurs later and takes no part in expanding the neosome. By the third day after penetration, polyploid cells are conspicuous below the arthrodial membrane, with the thinnest of untanned cuticular covering. Tanning is confined to the portions of the adult flea's body, head and legs, etc. and the area surrounding the caudal disc. The polyploid cells beneath the region of the caudal disc are gigantic (Plate IX) and pendulous, but are elongated and flattened (Plate X) in those portions whcre the intersegmental membrane forming the body wall is stretched. The anterior region of the neosome is expanded into 4-8 bulbous lobes between which the head is protectively retracted. In this region beneath the cuticle the polyploid cells are again enormous. In Vermipsyllu ulukurt the intersegmental membranes are greatly expanded but the body is not distorted and assumes a fat maggot-like appearance. Unlike the membranes of Tungu, which are smooth and lack setae, those of Verruiipsyllu and Dorcudia (and to a lesser degree Chuetopsylla) are lightly tanned and are covered with short, well-spaced, symmetrically arranged spine-like setae (Plates XI & XII), except immediately beneath the dorsal tergites where they are lacking (illustrated in Hopkins & Rothschild, 1956, figs 201-203 but not described). Polyploidy is entirely 62 I PL.ATI IX. Gigantic polyploid cclls secreting the body wall in thc region ol' the caudal disc of the embedded gravid Icniale T. ~IOU~J.S~IU.Y.

622 M. ROTHSCHILD PLATE X. Single polyploid cell beneath the intersegmental membrane of the body wall of an embedded gravid female T. ntonosirus. PLATE XI. Intersegmental membrane of female V. uhkuri (transverse section)

NEOSOMY IN FLEAS 623 PLATE XII. Outer surface of spined intersegmental membrane of female V. cilcikurr. PLATE XIII. Intersegmental membrane of the gravid queen ant Myrmica ruhra. Note the absence of polyploid cells and spicules.

624 M. ROTHSCHILD PLATE XIV. Intersegmental membrane of the gravid queen termite Mucrotermes suhhvulinus. Note the 'wavy' structure reminiscent of the intersegmental membrane of Y. alakurt (Plate XI) but lacking spicules. lacking and the walls of the expanded membranes appear wavy in serial section (Plate XI). Similarly, the intersegmental membranes of Neotunga, Hectopsyllu and Malacopsylla lack spinelike setae. Polyploidy is also lacking in the cells which secrete the smooth expanded intersegmental membranes of the gravid female ant (Myrmica rubra) and the gravid female termite (Macrotermes suhhyulinus) (Plates XI11 & XIV). The wavy structure of certain portions of the latter are surprisingly similar to those of Vermipsylla alakurt (Plate XI). The absence of spines is to be expected in the queens which inhabit a protected environment. Resilin und the jumping mechanism Since free-living fleas-except for a small number of so-called nest fleas-progress by running and vigorous jumping, one would expect the sessile and neosomic species to have lost some of the structures associated with this form of progression (Rothschild & Schlein, 1975). But the species concerned must still achieve, after emergence and before settling, the all important leap from the substratum to the passing host. The large size and mobility of the ungulate hosts with which Vermipsylla are associated must account for the big cone-shaped cap of resilin within the pleural arch cavity (fig. 15A in Rothschild & Schlein, 1975), the long vertical pleural rod and powerful outer coxal ridge. The pangolin flea, Neotunga euloidea also possesses a large cap of resilin, but the pleural rod, which is very short and thick and the pleural arch are ventrad in position. In Tunga which have small mammals and Echidnophaga which sometimes, in addition, have domestic hens as hosts (Hopkins & Rothschild, 1953), the resilin pad is reduced and flattened and the pleural rod

NEOSOMY IN FLEAS 625 is bent, thickened and shortened. Nevertheless, they are both capable jumpers before settling. In the Malacopsyllids which, again, are associated with large mobile hosts, the resilin pad is very large and the pleural rod greatly thickened but markedly bent. After fixation the legs of Tunga tend to be lost. The life-cycles The life-cycles of only approximately half a dozen sessile or semi-sessile fleas are known in detail and of these five are unusual and highly specialized. The two rabbit fleas Spilopsyllus cuniculi and Cediopsylla simplex depend for reproduction on the influence of the sex hormones and pheromones of their hosts (Rothschild & Ford, 1973*). Both male and female undergo sexual development, maturation, copulation, fertilization, oviposition and regression while feedingfixed for long periods-on the pregnant female rabbit and subsequently on her new-born young. No development can take place on the buck or non-pregnant doe. Eggs are laid in the nest, away from the colony, and the larvae consume debris and blood-soaked faecal pellets of the adult fleas. After oviposition the spent females and some of the male fleas return to the doe to await the next litter; if at any stage a pregnancy is interrupted the fleas ovaries regress and they remain feeding on the rabbits ears. The neosomes of Tunga monositus, a truly sessile species in the female sex, penetrates below the integument of the dorsal surface of the ear pinna of Peromyscus maniculatus and related species (Lavoipierre, Radovsky & Budwiser, 1978; see fig. 80** in Rothschild et al., 1986). The site of attachment is intensely specific as the neosome can only survive in this circumscribed area (Lavoipierre et al., 1978). For the first 12-14 days the female feeds on fluid and cellular exudate. Copulation occurs between the 14th and 19th day after eclosion; before this the female will not accept the male. Egg production follows within a day or two of insemination and continues for about six weeks. Simultaneously, the diet of the neosome switches to whole blood drawn from a vessel, and later to blood-pool feeding. The female Echidnophaga gallinacea once settled on the wattles of the host remains in situ until death (Suter, 1964). Copulation and oviposition takes place on the host. The female gorges? with blood from the moment of attachment onwards, while whole blood platelets are only found in the rabbit fleas intestine soon after attachment to the host s ears and again when the gravid female contains well-developed eggs and defecation greatly accelerates. The female T. monositus lays on an average 500 large eggs, which contain all the nourishment required for the development of the adult, since the mouthparts of the larva are atrophied and it never feeds. Moults are reduced to two. Although small salivary glands are present in the male it also never feeds; active sperm are already present in the pharate adult (see fig. 233 in Rothschild et a/., 1986). The neosome of T. monositus may live for several months, while the female Echidnophaga survives barely three weeks after attachment. The female rabbit flea, settled on the host s ears can live for over a year. At eclosion the female T. monositus is barely 1 mm in length but during its endoparasitic life, by * It is not possible on the information available to state categorically that the Mexican rabbit flea Cediopsyllu repolita is also hormone-bound but morphological and circumstantial evidence indicates that this is so (Rothschild, pers. obs.). ** The caption of this figure should read partly digested cellular exudate, not blood. t A curious observation was made when the author was rearing this species. The freshly emerged fleas fed and defecated much faster when fed on her arm than on that of her 8-year-old son!

626 M. ROTHSCHILD virtue of the expansion and growth of the intersegmental membranes, it increases 1000 times in volume. The body is then roughlycircular but there are four (or more) anterior projections or lobes between which the head is withdrawn (Lavoipierre et al., 1978). Only the caudal disc is in contact with the outside world, through which defecation, oviposition, respiration and copulation are effected. The legs are usually lost during the neosomes endoparasitic life. The life-cycle of the neosomic Vermipsyllids present a sharp contrast to that of the Pulicoid species. The hosts of these fleas which are adapted to high mountain pastures are principally ungulates but V. ulukurt also infests foxes, dogs and wolves, and if hungry will eagerly attack man (Ioff, 1950). The related genus Chuetopsyllu parasitizes carnivores. The adult fleas, both male and female, spend only the winter months, from October to April, on the body of their hosts. Development takes place on the ground among vegetation and debris but after hatching in the autumn the fleas-both sexes-spend a considerable time in open country pastures hopping about on the ground, actively searching for the passing hosts (Ioff, 1950). The female once attached to the integument by the elongated mouthparts grows rapidly from 1-2 mm to about 6-8 mm in length, by expansion of the intersegmental membranes. She begins egg-laying immediately, ovipositing near the surface of the pelt. A feature of this neosome is the long spells of gorging with blood which continue throughout adult life. Copulation takes place on the body of the host but often when the female is unattached and moving about in the wool or hair. In the genus Dorcadia the size of the fully expanded female neosome is considerably larger, attaining a length of 10-1 6 mm by 5-7 mm in width. Jumping about on the body of the host is well nigh impossible and this species then progresses through the wool in a worm-like manner. Waves of muscular contraction pass backwards along the body. Both Vermipsyllu and Dorcadiu lay enormous numbers of eggs which become entangled in the wool of their sheep hosts and are rubbed off on the ground. The eggs of Dorcadia are unique among fleas, for they are black not white. Ioff surmises that the black cuticle and setae of these fleas and the black eggs of Dorcadia are a protection from the effects of the radiant energy experienced at high altitudes. All sessile and semi-sessile fleas share at least one behavioural trait in common; they must select a feeding site on the host s body protected from teeth or beak. Rabbit fleas which fix on the host s ears congregate in an area just out of reach of the rabbit when it pulls down an ear over its face to groom it. The dorsal surface of the mouse pinna is a relatively safe area for T. monositus and the hen s wattles for E. gallinaceus. Vermipsyllu alakurt favours the fat tail of sheep and Dorcadiu the sides of the neck. Discussion The most striking fact about the neosomes of the different species involved in this life-style is that they have retained their essential and contrasting family body-build (Plates I & TI), incorporating the various convergent specializations associated with a sessile life-style. Chief of these is the expansion and growth of the intersegmental membranes linked with the simultaneous production of huge numbers of eggs. There is little doubt that these species of fleas which parasitize large mobile hosts such as deer, sheep, horses, armadillos, pangolins, wart hogs and large carnivores (particularly those which have scaly integuments, or lack nests or lairs) must evolve modifications which ensure that they remain on the bodies of their hosts once they have reached them. This requirement is one of the driving forces for the evolution of the sessile and neosomic life-styles. The size of the hosts is also associated with the large store of resilin in the dome-shaped

NEOSOMY IN FLEAS 621 pleural arch of all these fleas, since the initial jump from the substratum to the passing animal is of primary importance. The Tungidae, however, are principally parasites of small mammals (Hopkins & Rothschild, 1956)*. They are Pulicoid fleas and there are, throughout this superfamily, modifications in response to a close association with the host. Tungids develop on the ground in sandy soil frequented by small mammals which they invade via their feet or muzzle. Their pleural arch is flattened and reduced in bulk. They and Neotungu are the only endoparasitic fleas, and the neosome develops when buried beneath the skin of the host. The need to produce large numbers of eggs also seems to be linked to the females immobility. There are certain characteristics which are undoubtedly evolved in response to the sedentary life-style. In the sessile fleas the laciniae are serrated since they form organs of permanent or intermittent attachment. In the former case, they are then broad blades and inflexible, but in female fleas which spend relatively brief periods attached to the integument, like V. ulukurt or the Malacopsyllids, the laciniae are relatively narrow, sometimes flexible and only slightly toothed. These fleas often lack the angled frons which allows close and prolonged application of the parasite to the host s integument. Possibly the family strait-jacket plays a part in these variations, since the heavily serrated blade-like laciniae and angled frons are essentially characteristic of the Pulicoidea and the slender, flexible type and rounded frons are Ceratophylloid. Smit (1982) draws attention to the similarity between the adaptations of the Malacopsyllids parasitizing armadillos and the Vermipsyllids with ungulate hosts (which he places in different super-families). He also notes that the aberrant family Ancistropsyllidae (placed in another superfamily) with only three known species, resembles Vermipsyllu in some respects, and these fleas also parasitize deer. Little is known about its life-cycle but it is believed (Robert Traub, pers. comm.) that Ancistropsyllu attaches itself to the hosts by its enormously developed hooked ocular setae (see fig. 826 in Hopkins & Rothschild, 1971) favouring protected sites such as the skin folds of the groin. The lack of all but a few minute suckers on the male antennae indicates that the females are probably fixed during copulation. The very large spiracular fossa of tergum VIII and the associated tracheae may be an adaptation to a semi-sessile life since they are also found in female Tungu, Neotunga and some Echidnophugu. One species of Chimueropsyllu (placed by Smit ( 1982) in the super-family Hystrichopsylloidea) displays at least four adaptations associated with a sessile life-style: (1) strongly serrated laciniae; (2) a sharply angled frons; (3) male antennae lacking suckers; (4) a forward swing of the mouthparts (Rothschild & Hinton, 1968). It is enigmatical but suggestive, that ante-sensilial bristles are absent in Tungu, Hectopsyllu, Neotungu and Vermipsyllids and vestigial in Ancistropsyllu. Similarly, an anal stylet is lacking in the female Tungu, Hectopsyllu, Ancistropsylla and Vermipsyllids, but present in Neotungu. The most interesting aspect of the development of the intersegmental membranes of the neosome is the fact that those of all five species of Tungu so far examined are secreted by giant polyploid cells. As we have seen, there is no evidence of these cells in the membranes of Vermipsylla, nor in those of either the queen ant or the queen termite (Plate XI11 & XIV). Unfortunately, the female Neotunga is not available for sectioning; it would be interesting to determine whether polyploidy was involved in the secretion of the membranes in this flea, which is placed in the same super-family. Only approximately 30 species of fleas pertaining to ten families (Rothschild, 1976) have been * Tungrr penerrom is also a parasite of man, and is of medical importance in tropical Africa and Central and South America.

628 M. ROTHSCHILD cut into serial section by us but all those have polyploid cells in the salivary glands. However, there is a marked difference in the size of these cells in different families of fleas, for example in Ceratophyllids and Vermipsyllids they are insignificant in size and few in number. Whereas in Pulicoids and Hystrichopsyllids they are usually large and sometimes numerous. The malpighian tubules of Hystrichopsylla consist of huge polyploid cells (see figs 99, 100 in Rothschild et al., 1986). However, the development of polyploidy reaches a maximum in Tunga although here also there is considerable variation within the genus. This is well illustrated by the lack of polyploid cells lining the rectal ampulla of T. penetrans and their presence in T. monositus, where each of those giant cells about equals the size of the rectal pads! (See figs 106 & 107 in Rothschild et al., 1986). In fact, polyploidy attains fantastic proportions in this species. The cuticle of the intersegmental membranes of Vermipsylla, as we have seen, is covered in small spicules, whereas that of the neosome of Tunga is smooth and spineless. This difference is to be expected since Tunga is protected by the host s integument while those of Vermipsylla are exposed to the buffetings of life on an active host. It is rather surprising that neither Malacopsyllids nor Neorunga appear to have developed intersegmental spicules. Conclusions The classification of fleas is bedevilled by the exceptions to every devised rule and the neosomic species which constitute an evolutionary after-thought are equally unruly. It is possible, however, to separate ancient family traits from: (1) the influence of unusual hosts; and (2) the specializations imposed by a sessile life-style. The secretion of the expanded intersegmental membranes by polyploid cells in Tunga may be unique among neosomes. My very best thanks are due to S. G. Medvedev, L. A. Mound, F. J. Radovsky, F. G. A. M. Smit, J. Thomas and R. Traub for material, translations of Russian literature and many helpful discussions and suggestions. I am especially grateful to F. G. A. M. Smit and R. Traub for reading my manuscript. REFERENCES Audy, J. R., Radovsky. F. J. & Vercammen-Grandjean. P. H. (1972). Neosomy: Radical intrastadial metamorphosis associated with arthropod symbiosis. J. med. En!. 9 487-494. Hopkins, G. H. E. & Rothschild, M. (1953). An illustra~edcutalogue of the Rothschild collection offleas (Siphonuprera) in the British Museum (Natural History) 1. London: Trustees of the Br. Mus. (Nat. Hist.). Hopkins, G. H. E. & Rothschild, M. (1956). An illustrated caralogue ofihe Rothschildcollection ofjieas (Siphonaptera) in the Brirish Museum (Natural History) 11. London: Trustees of the Br. Mus. (Nat. Hist.). Hopkins, G. H. E. & Rothschild, M. (1971). An illusiratedcaialogue of the Rorhschildcollection @peas (Siphonapteru) in the British Museum (Natural History) V. London: Trustees of the Br. Mus. (Nat. Hist.). IoK, 1. G. (1950). The Alakurt. Mater, K. Poznan. Fauny i Flory SSSR. Ectoparasites (N.S.) (Zool.) 15(30): 4-29. Jordan, K. (1962). Notes on Tunga caecigena (Siphonaptera: Tungidae). Bull. Br. Mus. nai. Hist. (Em.) 12: 353-364. Lavoipierre, M. M. J.. Radovsky, F. J. & Budwiser, P. D. (1978). The feeding method of Tunga monositus and host response. J. med. Ent. 15: 187-217. Rothschild, M. (1964). Remarks on the life-cycle of fleas (Siphonaptera). Proc. inr. Congr. Parasitol. I: 29. Rothschild, M. (1976). Notes on Fleas (Part 11): The internal organs: Can they throw any light on relationships within the order? Proc. Br. En!. nut. Hist. Sue. 9: 97-1 10. Rothschild, M. (1988). Giant polyploid cells in Tunga monositus (Siphonaptera: Tungidae). Biosystematics d.f haematophagous insects. Systs Ass. (Spec. Vol). 37: 313-323. Rothschild, M. & Ford, R. (1973). Factors influencing the breeding of the rabbit flea (Spilopsyllus cuniculi): A spring-time accelerator and a kairomone in nestling rabbit urine with notes on Cediopsylla simplex, another hormone bound species. J. Zool., Lond. 170 87-137.

NEOSOMY IN FLEAS 629 Rothschild,M.&Hinton. H. E.(1968). Holdingorgansontheantennaeofmalefleas. Proc. R.ent. Soc. Lond. (A143 105-107. Rothschild, M. & Schlein, J. (1975). The jumping mechanism of Xenopsyllu cheopis. I. Exoskeletal structures and musculature. Phil. Trans. R. SOC. (B) 271: 457-490. Rothschild, M., Schlein. J. & Ito, S. (1986). A rolour atlas of insect tissues via theflea. London: Wolfe. Smit, F. G. A. M. (1962). Neotuugueuloideu gen.n., sp.n. (Siphonaptera: Pulicidae). Bull. Br. Mus. nat. Hist. (Ent.) 12365-378. Smit, F. G. A. M. (1982). Classification of the Siphonaptera. In Svnopsis andclassijcation of living organisms 2 557-563. Parker. S. P. (Ed.). New York: McGraw-Hill Book Co. Smit. F. G. A. M. (1987). An illustrated catalogue qf the Rothschild collection of,peus (Malacopsylloidea) in the British Museum (Nururul History) VII. London: Trustees of the Br. Mus. (Nat. Hist.). Suter, P. R. (1964). Biologie von Echidnophagagallinucea (Westw.) und Vergleich mit andern Verhaltenstypen bei Flohen. Aria trop. 21: 193-238. Traub, R., Rothschild. M. & Haddow, J. (1983). The Rothschildcollection offleas: the Ceratophyllidae: key to rhe genera and host relationships. London: Academic Press Inc.