Notes on the genus Tunga (Siphonaptera: Tungidae) II – neosomes, morphology, classification, and other taxonomic notes

This review focuses on the neosomes, morphology, and taxonomy of adult species of the genus Tunga, complementing the previously published data on the phylogeny, ecology, and pathogenic role. Neosomes are structures formed after penetration of adult females into the skin of hosts resulting in significant enlargement, being the most characteristic and most frequently observed form in hosts. Neosomes can be differentiated by shape, measurements, and sites of attachment to principal hosts. The taxonomic value and morphometric data of the most widely used characteristics to separate species – such as frontal curvature, head chaetotaxy, preoral internal sclerotization, ventral and dorsal genal lobes, eyes, maxillary palps, fusion of pronotum and mesonotum, metacoxae, metatarsi chaetotaxy, spermatheca (females), manubrium, basimere, telomere, and phallosome (males) – are comparatively analyzed. The sexes, individual variations, undescribed species, higher taxa, as well as a proposal for division of the genus into two subgenera (Tunga and Brevidigita) are presented (as previously given by Wang). A key for females, males, and gravid females (neosomes) also is included for identifying the 13 known species. Data on host specificity and geographical distribution may also support the identification of Tunga species because some sand fleas and their hosts may have co-evolved.

Among the Siphonaptera, Tunga is the most specialized genus because the adult females penetrate into the skin of their hosts. Similar to other fleas, both males and females are blood-feeding [18,53], however, the larvae and adult male of T. monositus do not feed [26].
Recently, a review of the genus Tunga concerning taxonomy, phylogeny, ecology and pathogenic role was presented by Beaucournu et al. [5], although the review does not include T. hexalobulata, which was described subsequently. Similarly, De Avelar et al. [13], when describing T. bossii for the first time, presented a widely used dichotomous key for identifying the known species and their neosomes and excluded T. bonnetti and T. hexalobulata, which were discovered later.
The present study complements this review with regard to the neosomes, morphology, classification, and other taxonomic notes. The taxonomic value and morphometric data of several characteristics are comparatively analyzed. Although the morphological aspects are often not known for many taxa, here we present a new key for identifying the 13 known Tunga species, including a more classical vision and emphasizing neosomal characteristics.
Morphological studies of the genus Tunga have concentrated on the description of the neosomes because they are the most characteristic and most frequently observed form in hosts. Females and males measure approximately 1 mm, but after penetration, the gravid females increase considerably in size, reaching approximately 10 mm (Fig. 1), and expose, on the surface of hosts, only the vital respiratory, anal, and vaginal apertures, which are contained in a caudal disk or conical prominence. The caudal disk exhibits some differences among species: (i) flattened, as in T. penetrans (Fig. 2), T. monositus, and T. bonneti; (ii) conical, as in T. caecata, T. travassosi, T. trimamillata, and T. hexalobulata; or (iii) cylindrical, as in T. terasma, T. bondari, T. caecigena, and T. callida. Otherwise, the caudal disk can be (i) as wide as long, or almost as wide as long, as in T. caecata, T. travassosi, and T. callida; (ii) wider than long, as in T. penetrans, T. monositus, T. trimamillata, T. bossii, and T. hexalobulata; or (iii) longer than wide, as in T. terasma, T. bondari, and T. caecigena. Neosomes can live more than three months attaching to different sites of their respective hosts [14,26]. Consequently, neosomes are important for identifying the species in the genus Tunga.
A review solely on the neosomes of tungid fleas that infest wild and domestic animals and concentrated mainly on hosts, infestation, sites of attachment, and impact on the hosts was recently presented by Linardi and Avelar [32]. Table 1 shows the known Tunga species with their respective geographic distributions, sexes, stages, neosomes, and sites of attachment to principal hosts. Excepting neosomes, in which the size is indicated in millimeters, all measurements included in other tables are in micrometers.

Head (Fig. 3)
The cephalic capsule includes the majority of the characteristics used for identifying these species. Members of the genera Tunga, Hectopsylla (with the exception of the female of Hectopsylla coniger), Echidnophaga, Phacopsylla, and Neotunga euloidea are always characterized by an angular profile and a well-pronounced frontal tubercle, as in T. caecigena (Fig. 3C), T. travassosi (Fig. 3D), T. bondari (Fig. 3E), T. trimamillata (Fig. 3J), T. bossii (Fig. 3K), and T. hexalobulata (Fig. 3M). The front curvature varies little but is gently convex or sharper as in T. caecigena (Fig. 3C) or almost straight as in T. libis (Fig. 3H). Generally, the ventral profile of the cephalic capsule shows a genal lobe, which is often more pronounced in females; it seems absent in bossii, but is very clear in caecata, caecigena, travassosi, bondari, terasma, and monositus. Preoral internal sclerotization presents as the posterior arm longer than the anterior arm (caecata, monositus, trimamillata, and hexalobulata) or with the posterior arm subequal to or shorter than the anterior arm (caecigena, penetrans, bondari, bossii), or much shorter than the anterior arm (callida). A dorsal genal lobe either covers part of the antennal segment III (bondari, travassosi, caecigena, callida) or not, as in other species. The number of bristles on the postantennal region, posterior region to the frontal tubercle, antennal segment II, and base of the maxilla may be a differential characteristic. Inside the cephalic capsule, the eyes are the most characteristic structures for species identification. They may be absent in caecigena and callida or present in other species, or they may be pigmented (penetrans, bondari, terasma, libis, trimamillata, bonneti, and hexalobulata) or without black pigment (caecata, travassosi, monositus, and bossii). When present, the eyes can be small (caecata, travassosi, and monositus), or large, in the case of other species. In some species such as penetrans, bondari, trimamillata, the eyes also have a recess into their internal margins. The greatest eye diameters and the numeric variations of bristles on some structures of Tunga species are presented in Table 2.
Maxillary palps ( Fig. 4) with segments vary in size and chaetotaxy; in caecigena, callida, and bonneti, segment IV is longer than segment I, though the reverse is true in T. trimamillata and T. hexalobulata. In other species, the palps are approximately the same length. In T. bossii, segment I was described as very small and segments III and IV as being incompletely divided; however, a new observation showed that, in fact, what was thought to be ''segment I'' was a strengthened area at the base of the palp and an incomplete division occurred between segments II and III and between segments III and IV.
Morphometric data regarding the maxillary palps, including new measurements for T. bossii, laciniae and preoral internal sclerotization are indicated in Table 3.

Thorax
As Smit [46] noted, the fusion of the pronotum and mesonotum can be dorsally complete, a characteristic of the penetrans group or incomplete, as in the caecata group. Currently, the penetrans group includes T. penetrans, T. travassosi, T. bondari, T. terasma, T. trimamillata, and   T. hexalobulata [11,13]. In the caecata group, the following species are included: T. caecata, T. caecigena, T. callida, T. libis, T. monositus, T. bossii, and T. bonneti. The chaetotaxy is generally sparse, consisting of 1-7 bristles on the prothorax. Legs III show the classical increase in length, compared to the first two pairs, typical of Siphonaptera. However, these legs are slender, and members of the Tunga genus are poor jumpers despite the existence of a pleural arch. At this leg, Beaucournu et al. [5] noted an external guard tooth, whose size is variable depending on the species and ranges from completely absent to normally developed. The coxae are always preserved regardless of the autotomy of the legs. The metacoxae (Fig. 5) project  downward at an anterodistal angle, like a wide tooth, and exhibit a variable number of bristles both on the external and internal surfaces. Species such as penetrans, trimamillata, and hexalobulata present a greater number of bristles. In the species of the caecata group, the metacoxae are slightly wider at the base than at the apex (Figs. 5E, H, I), though in the penetrans group, excepting T. terasma (Fig. 5F), the basal part is nearly two times wider than the apical part (Figs. 5A, B, H, J, K). According to Hopkins and Rothschild [20], T. penetrans is characterized by a slower and less complete deterioration of the legs than the other species known at that time. The chaetotaxy of tibiae, tarsi, and especially of the distal segment of leg III are discriminant characteristics. According to Smit [46], the chaetotaxy of the distal tarsal in the species belonging to the penetrans group is strongly reduced, with only two pairs of hair-like lateral plantar bristles and no patch of minute plantar setae segment bristles (Fig. 6A). In contrast, the chaetotaxy is only slightly reduced in the species of the caecata group, which exhibit three or four pairs of stiff subspiniform lateral plantar bristles and a patch of minute plantar bristles (Figs. 6B-F). Unfortunately, these articles are often missing even when studying a female that is recently embedded.

Abdomen
Externally, the most striking distinguishing feature between the penetrans and caecata groups is the variation in the size of the spiracles of the hypertrophied females. In the penetrans group, the spiracles of terga II-IV have disappeared, while in the caecata group, they are smaller than the others, though present [46]. The spermatheca presents a certain enigma. In the female neonate or a female not yet embedded, this structure is invisible upon first examination, as far as we know for the fleas that have a body or bulga, is most often sclerotized, and is a distal appendage, tail, or hilla that acts as a pump for sperm. In fleas that are not attached, a thorough examination will reveal, however, a transparent area called cribiform area, which is riddled with multiple orifices, at the base of the bulga. As in other genera, it is connected to the vagina through a duct, the ductus spermathecae, a visible though poorly defined structure. The spermatheca itself is diaphanous, and more or less conical in some species such as penetrans or trimamillata. In contrast, this structure is heavily sclerotized in neosomatic and fertilized females and has a consistent shape within a given species. Both Karsten [25] and Bonnet [8]  described the spermatheca perfectly in the penetrans group, although Bonnet was not able to identify its purpose. Bonnet also noted the presence of an occasional double spermatheca in T. penetrans, which was omitted by Beaucournu et al. [4] and is, thus far, unique to this family. Figure 7 shows the spermathecae of 11 species. T. travassosi, T. bondari, T. terasma, T. callida, T. monositus, and T. bossii exhibit bulgae that are rounded and spherical, while in T. penetrans, T. caecata, T. libis, T. monositus, T. trimamillata, and T. hexalobulata they are elongated and ellipsoidal. Spermatecae with short hillae are seen in penetrans, trimamillata, and hexalobulata, while long hillae are found in caecata, travassosi, bondari, terasma, callida, libis, and bossii. On the other hand, the width of the hilla may be thin (penetrans, terasma, monositus, trimamillata, and hexalobulata) or thick (bondari, callida, and libis). The measurements of the bulga and hilla of the spermathecae are presented in Table 4.
In males, segment IX (clasper) and the phallosome are the most striking features for differentiating species. Similar to other fleas, segment IX is divided into two processes, the basimere (also often called the immovable or fixed process, or even p 1 ) and the telomere (also called the movable process, movable finger, or even p 2 ) which are connected to another structure, the manubrium. Figure 8 shows the shape and chaetotaxy of segment IX for the eight species whose males are known. The ratios between the width of the basimere and telomere in its median portion and the length of basimere/manubrium, and the shape of the basal and dorsal part of the manubrium are observed in Table 4. The phallosome consists of proximal and distal arms articulated in the middle, just before the ductus ejaculatorius and is almost as long as the penis-plate.
The size ratio and the angle between the two arms are differential features for the species, as seen in Figure 9. In T. penetrans and T. monositus, the two arms are angled at approximately 90°, whereas in T. callida and T. caecigena, they are connected in a nearly straight line. Table 4 exhibits some morphologic features and morphometric data of the modified segments of the male and the spermathecae of the females.

Taxonomy
The principal synonymies, sexes, individual variations, undescribed species, subgenera, higher taxa and a key for adult species and neosomes are included and discussed below.

Synonymies
It is interesting to note that the older the description of the species, the greater the number of existing synonymies. Thus, T. penetrans presents seven major synonyms followed by T. caecata with two of them. All the synonymies for the genus were already cited by Beaucournu et al. [5].

Sexes
As shown in Table 1, the species caecata, travassosi, bondari, bossii, and hexalobulata are known only through their hypertrophied females. T. penetrans, T. callida, T. monositus, T. trimamillata, and T. bonneti are species in which the holotype and the allotype were described simultaneously. In caecigena and libis, the allotype males were described 37 (1958) and 6 (1968) years, respectively, after the holotype females. The males of T. terasma were described incorrectly by Fonseca [17] as males of T. travassosi. Interestingly, the holotype female of T. terasma was described the following year by Jordan. It is possible that this is the only case in the entire order of Siphonaptera in which the allotype was known before the holotype.

Individual variations
Regardless of sexual dimorphism (length, size of spiracles, chaetotaxy, maxillary palps), most of the individual variations are found in the modified segments, especially among the males. Hopkins and Rothschild [20] illustrate variations in the manubrium of T. penetrans and T. terasma. The illustrations of the male of T. caecigena presented by Chen and Ku [9] also show variations in the shape of the manubrium, basimere, and telomere. The figures of Wang [51] stress variations in the form of denticles on the distal portion of the basimere of T. caecigena. Similarly, when describing T. monositus, Barnes and Radovsky [2] illustrated and drew attention to the fact that the manubrium presents a highly variable shape, even on two sides of the same individual, with the proximal portion ranging from broad and blunt to slender, curved, and acuminate.

Undescribed species
1. As reported in Linardi and Guimarães [34], hypertrophied females of the caecata group were observed by Linardi and Botelho [31] to parasitize Oryzomys nigripes (currently Oligoryzomys nigripes) and Nectomys squamipes from Caratinga, Minas Gerais State, Brazil, with the neosomes located near the base of the dorsal surface of the ears.   Linardi) reported that this species is now being described and that it forms a discoid neosome in the carapace of Zaedyus pichiy perforating the osteoderms.

Subgenera
When Smit [46] divided the genus Tunga into two groups of species, he considered morphological characteristics such as the dorsal fusion of the pronotum and the mesonotum, the chaetotaxy of the fifth tarsal segment and the presence or the size of spiracular fossae on terga II-IV of the females, in addition to the parasitized host groups. At that time, only eight species were known. Four were included in the penetrans group (T. penetrans, T. travassosi, T. bondari, and T. terasma) and four others in the caecata group (T. caecata, T. caecigena, T. callida, and T. libis). Later, another proposal by Wang [51], based solely on geographical distribution, included the two known Chinese species, caecigena and callida, in a distinct subgenus (Brevidigita). When presenting the supraspecific classification for the genus Tunga, Lewis [28] accepted Wang's proposal, but seems to have taken into consideration only those Tunga species that parasitize commensal rats for inclusion in Brevidigita because only T. caecata was added, though two other species, T. monositus and T. libis, were already known and were improperly left in the subgenus Tunga. Currently, with 13 known species, we consider the genus divided into two subgenera, Tunga and Brevidigita, though this division may be debatable:

Higher taxa
The vast majority of the Siphonaptera show a certain consistency for certain wildcard characteristics; for example, the profile of the cephalic capsule, development of the eye (unless a host to a different ecology), the proportions of the various articles of the palp, or the form of the spermatheca. However, in the genus Tunga, these rules are not respected. As has been done in the genus Pulex (Pulex, Juxtapulex. . .), further divisions may be required when the puzzle of taxa known as Tunga is complemented by a comparative study of all males.
Beaucournu (in Aberlenc) [  For this author, the Tungoidea comprises only the genus Tunga. Hectopsylla remains in the Pulicoidea. Linardi asserts (as seen in Linardi and Guimarães [34]) that Hectopsylla must be retained in Tungidae, as evidenced by DNA analysis [52].

Key for adults and neosomes
The first dichotomous key for this genus was that of Hopkins and Rothschild [20] and did not include six species that were subsequently described (callida, libis, monositus, trimamillata, bossii, bonneti, and hexalobulata). While it is no longer usable, it proposed to differentiate Tungidae, but using characteristics that have proved worthless, for example, the indentation of the eye. This same key, excluding the two Chinese Tunga, was later reproduced by Johnson [21] to identify South American sand fleas. Barnes and Radovsky [2], when describing T. monositus, presented a key exclusively for the caecata group species known at that time.
Recently, De Avelar et al. [13], when describing Tunga bossii, proposed a new key, including all known taxa (both bonneti and hexalobulata are absent since their descriptions appeared after bossii) in which the appearance of the neosome is widely used. Although often the morphological aspects of many taxa that we would like to use in classification are not known (the autotomy of legs is mainly in females), we attempt to provide a more classical vision of differentiation, with emphasis on independent characteristics of neosomes. Thus, this new key also includes the means to differentiate known male characteristics. 2. Eye without black pigment (Fig. 3D). Base of maxilla with only a micro-bristle. Maxillary palp as in Figure 4D. Metacoxa as in Figure 5D. Spermatheca as in Figure  3. Eye small, its greatest diameter only slightly exceeding half the distance from eye to dorsal margin of head (Fig. 3E). Dorsal genal lobe covering part of antennal segment III. Segment IVof maxillary palp longer than segment I (Fig. 4E). Metacoxa as in Figure 5E. Spermatheca with bulga wider than long (Fig. 7D). 4. Frontal tubercle slightly pronounced (Fig. 3F). Antennal segment II with only two bristles. Maxillary palp with few bristles (Fig. 4F). Metacoxa with proximal portion as wide as distal portion (Fig. 5F). Manubrium with ventral margin concave and dorsal margin convex (Fig. 8C). Spermatheca with globular bulga two times longer than hilla (Fig. 7D). Hypertrophied female not globular and containing four prominent lobes, measuring (mm) 10 (length), 9 (width), and 13 (height) and with cylindrical caudal disk, longer than wide .. 5. Head with pronounced ventral genal lobe (Fig. 3A). Anterior base of maxilla with three thick bristles. Segment II of maxillary palp longer than segment I (Fig. 4A). Anteromedial surface of posterior tibia without bristles. Manubrium with proximal portion wide (Fig. 8A). Phallosome as Figure 9A. Bulga of spermatheca approximately 3.5 times longer than hilla (Fig. 7A). 6. Head with the postantennal region exhibiting 17-19 bristles (Fig. 3I). Antennal segment II with four bristles. Segment IV of maxillary palp almost as long as segment II (Fig. 4I). Metacoxa without a projection at the margin of the proximal portion (Fig. 5H). Spermatheca as in Figure 7I, with curved hila and bulga nine times wider than hilla. Anteromedial surface of posterior tibia with row of 9-12 bristles. Manubrium with proximal portion tapering and facing up (Fig. 8G). Phallosome as Figure 9E. Neosome with three lobes located anteriorly, measuring 8. Occipital region with 8-10 bristles (Fig. 3C). Preoral internal sclerotization with anterior arm extending to half the distance between frontal tubercle and base of maxillary palp. Lacinia as long as maxillary palp. Manubrium almost as long as basimere (Fig. 8B). Phallosome as in Figure 9B. Elliptical neosome, with dorsal and ventral portions of similar dilatation, measuring (mm) 7-10 (length), 5 (width), and 6 (height) and caudal disk longer than wide ........T. (B.) caecigena Jordan and Rothschild. Occipital region with 3-5 bristles (Fig. 3G). Preoral internal sclerotization with anterior arm extending near the base of maxillary palp. Lacinia longer than maxillary palp. Manubrium slightly longer than basimere (Fig. 8D). Phallosome as in Figure 9C. Spherical neosome, with the dorsal portion more swelled than the ventral portion, measuring (mm) 4 10. Fifth metatarsomere with numerous minute plantar bristles (Fig. 6D). Segment II of the maxillary palp longer than segment III. Telomere as wide as basimere (Fig. 8E). Vertically elliptical neosome, higher than long and without lobes. Fifth metatarsomere of the metatarsus with few minute plantar bristles (Fig. 6F). Segment II of the maxillary palp as long as segment III (Fig. 4K). Telomere wider than basimere (Fig. 8H). Horizontally elliptical neosome, longer than high, rugby-ball shaped and without lobes. 11. Posterodorsal lobe of proepimeron large, strongly projecting. Base of maxilla with no bristles. Metacoxa with basal portion much wider than apical and with anterior basal corners dilated (Fig. 5G). Spermatheca with bulga four times longer than hilla (Fig. 7H). Males with segment IX as in Figure 8F and phallosome as in Figure 9D. Bell 12. Preoral internal sclerotization with posterior arm more than eight times the length of anterior arm (Fig. 3B). Frontal tubercle slightly pronounced. Eye with the greatest diameter less than half the length of the antennal segment II. Base of maxilla with one bristle. Segment IV of the maxillary palp less than the length of segments II + III (Fig. 4B). Metacoxa as in Figure 5B. Spermatheca with bulga longer than wide and hilla two times longer than wide (Fig. 7B). Neosome, measuring (mm) 7 (length), 6 (width), and 6 (height) and conical caudal disk, almost as wide as long .  (Enderlein). Preoral internal sclerotization with posterior arm less than four times the length of anterior arm (Fig. 3K). Conspicuous frontal tubercle. Eye with the greatest diameter almost as long as antennal segment II. Base of maxilla has two bristles. Segments II, III, and IV of the maxillary palp undivided, but apparently IV greater than II + III (Fig. 4J). Metacoxa as in Figure 5I. Spermatheca with bulga wider than long and hilla about four times longer than wide (Fig. 7J). Neosome, measuring (mm) 9 (

Final remarks
The geographical center of the distribution of the genus Tunga is concentrated in South America, including 77% of the species that most likely originated there. Because approximately 23% of the species were described in the last 2 years from Ecuador, Brazil, and Chile, the biomes and regions contained in these countries offer great opportunities for new findings. There are at least three or four undescribed species, with one of them awaiting description and another two or three remaining confused with species that are already known.
Thirty-nine percent of the species are known only by the characteristics of the embedded females (neosomes). Only two out of the 13 described species of sand fleas have known larvae, although several larvae of the 1st instar of an undescribed species were obtained in the laboratory by Linardi and Botelho [34] from hypertrophied females infesting Brazilian wild rodents. Consequently, investigation of the alternate sexes and immature forms provides promising lines of research in the respective regions of their occurrences.
Because larval development occurs in fine-grained soils and the sand fleas are univoltine, the best results can be obtained in the dry-cool season.
Given the epidemiological and economic importance of T. penetrans and T. trimamillata, which infest both domestic animals and humans, the correct identification of species is indispensable. T. penetrans has been found parasitizing at least 28 genera of hosts [10], although some occurrences are incorrect records [32,33]. Morphological variations must be used carefully for taxonomic purposes. Molecular biology should also be used for such purposes. As showed by De Avelar and Linardi [12] the Multiple Displacement Amplification technique (MDA) may be a valuable tool for molecular studies involving samples of sand fleas that are preserved in scientific collections.
Rodents and edentates are the main hosts of Siphonaptera, housing approximately 85% of the known species.
Data on host specificity and geographical distribution may support the identification of Tunga species because some sand fleas and their hosts may have co-evolved. According to Traub [48], most fleas have clearly evolved with their hosts and primitive hosts tend to have primitive fleas [47]. Mammals such as Edentata might have been the primitive hosts of tungids because they are devoid of incisor teeth and nails to remove the neosomes attached on toes and on the ventral abdominal, regions regularly in contact with the soil and of great difficulty for the hosts to dislodge the parasites by grooming or eating. On the other hand, based on the molecular phylogeny, Whiting et al. [52] placed Tunga at the base of flea phylogeny and its association with basal mammal hosts suggests that the origin and diversification of Siphonaptera coincided with basal mammal diversification. Sloths (Pilosa) and armadillos (Cingulata) belong to an ancient stock of mammalians and constitute the majority of the natural mammalian hosts of the genus Tunga [52].