Head morphology of Tricholepidion gertschi indicates monophyletic Zygentoma

The relic silverfish Tricholepidion gertschi is the sole extant representative of the family Lepidotrichidae. Its phylogenetic position is of special interest, since it may provide crucial insights into the early phenotypic evolution of the dicondylian insects. However, the phylogenetic position of T. gertschi is unclear. Originally, it was classified among silverfish (Zygentoma), but various alternative relationships within Zygentoma as well as a sistergroup relationship to all remaining Zygentoma + Pterygota are discussed, the latter implying a paraphyly of Zygentoma with respect to Pterygota. Since characters of the head anatomy play a major role in this discussion, we here present the so far most detailed description of the head of T. gertschi based on anatomical studies by synchrotron micro-computer tomography and scanning electron microscopy. A strong focus is put on the documentation of mouthparts and the anatomy of the endoskeleton as well as the muscle equipment. In contrast to former studies we could confirm the presence of a Musculus hypopharyngomandibularis (0md4). The ligamentous connection between the mandibles composed of Musculus tentoriomandibularis inferior (0md6) is also in contact with the anterior tentorium. Phylogenetic analysis of cephalic data results in monophyletic Zygentoma including T. gertschi. Zygentoma are supported by the presence of a set of labial muscles originating at the postocciput, presence of an additional intralabral muscle, and four labial palpomeres. Character systems like the genitalic system, the mating behaviour, the segmentation of the tarsi, the overall body form, and the presence of ocelli which were proposed in other studies as potentially useful for phylogenetic reconstruction are evaluated.


Introduction
The relic silverfish Trichlepidion gertschi occurs only in the coastal region of northern California. The species is characterized by a number of peculiarities with respect to all other extant silverfish species (= Euzygentoma) such as the presence of ocelli (in addition to compound eyes), 5-segmented tarsi [1], styli and coxal vesicles on almost all pregenital segments of the abdomen, and a ligamentous head endoskeleton. Its phylogenetic position with Zygentoma or Dicondylia is unclear (for an overview see [1,2]). Molecular [3][4][5][6][7] and morphological studies [8][9][10][11][12][13] disagree whether T. gertschi is the sister species to all other dicondylians (Zygentoma + Pterygota), sister species to all other Zygentoma, or a subgroup within Zygentoma ( Figure 1). However, the phylogenetic position of T. gertschi is crucial to understand the evolution of several morphological characters in the stem lineage of Dicondylia, e.g. presence of a proventriculus [10], sperm configuration [20,21], the general organisation of the head [11,14] and thorax [22] including muscle equipment, and the composition of ovaries [10].
Available cephalic data did not help thus far to resolve the phylogenetic relationships of T. gertschi to remaining Zygentoma [11,14]. This contribution aims to augment the analyses of the head character system by providing a detailed description of the head morphology of T. gertschi with a strong focus on the documentation of mouthparts, endoskeleton and muscle equipment. We show that characters of the cephalic morphology indeed provide information on a sistergroup relationship of T. gertschi and the remaining studied Zygentoma.
SR-μCT was done using the recommendations of Betz et al. [24] and the respective beamline staff. Prior to scanning, the sample was critical point dried (CPD) (Model E4850, BioRad), mounted on specimen holders, and put into the scan chamber several hours in advance to allow for temperature acclimatisation which avoids movement artefacts. Scanning was performed at the Paul-Scherrer Institute (PSI, Villigen, Switzerland) with a stable energy beam of 8.5 keV in attenuation mode, 10× magnification, 500 ms exposure time, and 1601 projections within 180°(see [25] for the beamline configuration) and at the Deutsches Elektronen Synchrotron (DESY, Hamburg, Germany) at the beamline DORIS III/BW2 and beamline PETRA III/IBL P05 with a stable energy of 8 keV with high density resolution [26] in attenuation contrast mode. The field of view, exposure time, and number of projections was adjusted for each specimen.
Segmentation of structures and rendering of the resulting 3D model was performed with the software packages Reconstruct [27] and Blender (www.blender.org). Both software packages are distributed under the general public license (GPL). Tables and figures were edited with GIMP ver. 2.8, Inkscape ver. 0.48 and Scribus ver. 1.4.1 (all GPL). A 3D model of the head of T. gertschi is available (additional file 4) which facilitates identification of internal structures. Please download the software Blender to view the model. Additionally, transverse sections of the head are provided as a film sequence (in AVI format; see additional file 5) For scanning electron microscopy (SEM) the specimen was transferred in a series of steps into 100% ethanol, critical point dried (Model E4850, BioRad), and sputter coated (Model Anatech Hummer VII). Microscopy was performed on a Hitachi S-2460 N scanning electron microscope using a rotatable sample holder [28]. The terminology of skeletal elements follows Seifert [29], the muscular one Wipfler et al. [30].

Phylogenetic analyses
Parsimony analyses of 139 cephalic characters (see Appendix) and Bremer support (BR) calculations were carried out with TNT [31] using 1,000 heuristic searches starting with random addition of taxa (Wagner trees; Tree-Bisection-Reconnection (TBR) branch swapping, with 100 trees saved per replicate). All characters were equally weighted and unordered. The archaeognathan Machilis germanica was selected as the outgroup. Only unambiguous changes were mapped on the optimal trees. Character numbers and states are given in brackets using the following syntax: (character number : character state). The character matrix is derived from Blanke et al. [32] [11,14] and characters of the whole body with a focus on attachment structures [13], b) Lepidotrichidae as the sistergroup to all remaining Dicondylia was hypothesized by Kristensen [15] and Stys & Zrzavý [16], c) T. gertschi as the sistergroup to the remaining Zygentoma, with unclear resolution of †L. pilifera, was hypothesized by Koch [17] and Engel [18], d) Lepidotrichidae as closely related to Nicoletiidae (+ Ateluridae) within Zygentoma was hypothesized by Wygodzinsky [10] and is supported by sperm characters according to Dallai et al. [19]. and is based on the matrix of Wipfler et al. [30]. Please refer to the electronic supplement (additional file 6) for a complete tree showing all taxa.

External head capsule
The orthognathous head (Figure 2a) bears numerous sensilla. Trichoid sensilla (= setae) are up to 100 μm long, directed anteriorly and occur with a density of 10-12 sensilla per 100 μm 2 . Among them, numerous (>500/100 μm 2 ) small tubercles on the exoskeleton, possibly sensorial in function, cover the entire head (Figure 2e). Both sensillum types also occur on certain regions of the mouthparts (details see below). The compound eyes, con-taining~40 ommatidia, are positioned immediately behind the antennae and dorsal to the externally visible posterior mandibular articulation (Figure 3a, 4b). The three ocelli are barely visible in SEM specimens while they are of whitish colour in living specimens. The middle ocellus lies centrally directly above the epistomal ridge, the lateral ocelli behind the antennal bases (Figures 3a, 5a). Their lenses do not protrude from the head.
In anterior view the undivided clypeus is formed like a dorsoventrally elongated hexagon (Figure 5a). The welldeveloped epistomal ridge forms its medio-dorsal and the antennal bases its dorso-lateral delimitation. The  depression so that a ginglymus is formed. The straight occipital ridge (in dorsal view; Figure 3a) clearly separates the frons from the rectangular occiput. The width mentioned head regions here considered as occiput and postocciput see the discussion.

click 3D
clypeus is covered by the same two sensillum types like the rest of the head capsule except for its ventral part near the clypeo-labral ridge where it is smooth and entirely devoid of setae ( Figure 5d). The epistomal ridge is located at half height of the antennal bases. An interantennal ridge is absent. The frons is roundish in lateral view (Figure 4b). It continues ventrally into the genal area which also harbours the posterior mandible articulation (Figure 4b). In dorsal view the frons appears like a semi-circle, its corners almost meeting the eyes anteriorly (Figure 3a). The frontal sutures and the coronal suture form an inverted Y when seen from dorsofrontal. The frontal sutures are barely visible, the coronal suture continues from the frontal sutures posteriorly until the postocciput (Figure 3a).
The head-wise part of the posterior mandibular articulation comprises a depression and a pyramidal condylus with its tip oriented ventrally ( Figure 4a). The condylus formed at the posterior end of the mandible lies in the depression so that a ginglymus is formed. The straight occipital ridge (in dorsal view; Figure 3a) clearly separates the frons from the rectangular occiput. The width of the occiput (in lateral view) corresponds to the width of the compound eyes. The occiput is posteriorly delimited by a strong postoccipital ridge. The postoccipital ridge is equally well developed and serves as an attachment point for several thoracic muscles. The postocciput is one third the length of the occiput (in dorsal view; Figure 3a) and wider than the rest of the head capsule. In the lateral region the postocciput bears three very long (~150 μm) trichoid sensilla which are laterally oriented. For a different interpretation of the above mentioned head regions here considered as occiput and postocciput see the discussion.

Cephalic endoskeleton
The cuticular endoskeleton (Figures 3b, 4c, 6a) is composed of two main elements: a posterior tentorium and an anterior tentorium. The anterior tentorium is composed of paired anterior and dorsal tentorial arms and an anterior tentorial plate. The posterior tentorium is connected to the anterior tentorial plate by small muscles (0te5 + 6). The anterior tentorial pits are externally not visible. They are located ventral to the voluminous antennal bases in a cavity delimited by the mandibles, the clypeus and the antennae (Figure 5b). The massive, anterior tentorial arms emerge from the anterior tentorial pits and coalesce into the anterior tentorial plate at the level of the transverse mandibular tendon. At the point of fusion the dorsal tentorial arms emerge. They are not in contact with the head capsule but suspended to it by two muscle bundles (0te2 and 0te4). Posterior  to the dorsal arms the anterior plate narrows and widens again at level of the posterior mandibular articulation, just where the posterior tentorium begins.

Mandibles
The mandibles (Figure 7) are formed like an elongated bowl in dorsal view with an oval dorso-mesally oriented opening to which several muscles attach (Figures 5f, 7d, 8). The mandibles are overall rigidly sclerotised with the greatest wall thickness at the gnathal region (incisivi and mola) and the posterior and anterior mandibular articulation regions (Figures 7c + e). Anteriorly, the mandibles form a sharp median edge with a dorsal mola and three ventral incisivi (Figures 7c + e). The gnathal edges are almost symmetrical on both mandibles.
The posteriormost part of the mandibles bears the posterior mandibular articulation which is continuous with the overall form of the mandible (Figure 7a + b), so that no distinct knob is formed. The loose anterior mandibular articulation complex is situated at height of the dorsal part of the mola, a short distance distal to it. It is composed of two parts: the headwise part is a "caliper-like" structure formed by parts of the anterior tentorial arms and the clypeus (Figure 5f + g). The mandibular part of the articulation consists of a thickened mandibular dorsal margin (the mandibular ridge), a depression ventral of this ridge, and a lateral edge delimiting the depression laterally (Figure 5c).
The caliper is "wrapped" around the mandibular edge, its clypeal part forms a process which touches the mandibular depression and fits into it. The posterior part of the caliper (a ventral apodeme of the anterior tentorial arm) reaches into the lumen of the mandible directly posterad the mandibular ridge (Figure 5g). The two processes (clypeal and tentorial) of the caliper thus prevent an antero-posterad movement of the mandible.
The mola is almost formed like a right-angled triangle in lateral view (Figure 7e

Maxillae
The body of the maxillae (from cardo to galea) is three times longer than wide in overall shape (Figure 9a). The cardo is approximately triangular and contains a medially oriented lever (Figure 9a) serving as attachment for the M. craniocardinalis (0mx1). The cardo bears some setae and is moveably connected to the stipes by a very narrow articular membrane lying at the base of the cardo-stipital ridge.
The stipes is composed of a narrow basistipes and a much larger mediostipes bearing the palpus, galea and lacinia (Figure 9f). The whole stipes is devoid of sensilla except for the part directly posterior to the palpal base (the area which is externally visible in lateral view; see Figure 4). The base of the maxillary palpus is surrounded by protrusions of the stipes forming a ring around the palpal foramen (palpifer).
The palpus is five-segmented and densely covered with trichoid sensilla. The first segment is one third as long as the second one. The third one is slightly longer than the second one, the fourth and fifth are as long as the second one. Each segment is slightly thinner than the preceding one. The fifth segment bears distally six special sensilla formed by a basal cylindrical segment densely covered with microtrichia and four to six tubular extensions on the tip of the cylindrical base (Figures 9f1 + f2). These sensilla are arranged in a pentagon with the 6th sensillum at its center.

click 3D
The galea is sickle-shaped (Figure 9a + f), distally covered with setae and bears apically two sensorial projections with several minute cones at their apices, each equipped with a terminal pore (Figures 9c1 + 2). The lacinia is also sickle-shaped (Figure 9a

Labium
The labium (Figure 2a) is divided into postmentum, prementum, glossae, paraglossae and palpus. All externally visible parts are covered with trichoid sensilla in the same density as the rest of the head capsule. The postmentum is an almost rectangular plate. The prementum is also rectangular in ventral view and bears a deep median premental cleft. Glossae and paraglossae are short, the former finger-like and a bit longer than the broadened paraglossae. The palpi are four-segmented; a short basal segment is followed by two elongate segments and a widened apical segment. The anterior (or ventral) surface of the apical segment shows a median cleft and is densely covered with trichoid sensilla. Two additional sensilla types are present in the distal region. Towards the apical margin there are three brush-like clusters composed of thin tubular extensions with flattened tips Figure 8 Frontal virtual SR-microCT sections through the head of Tricholepidion gertschi with a) the 0md4 and 0md8 muscles coloured on the right side of the animal to show the presence of the 0md4 and the insertion point of the 0md8, and with b) the ligamentous connection of 0md6 between both mandibles in yellow color and the muscular connection to the anterior tentorium in red colour. Abbreviations: ata, anterior tentorial arm; br, brain; fg, foregut; md, mandible. For muscle references see main text. ; c2) detail of the apical area of the galea (meso-anterad view); d) detail of the apical and subapical area of the lacinia; e) the muscle equipment in anterior view; f) anterior view of the outer anatomy; f1) detail of the apical area of the maxillary palpus; f2) detail of a sensillum at the apex of the maxillary palpus. Abbreviations: bst, basistipes; ca, cardo; cal, lever of the cardo; csen, conical sensillum; gal, galea; gc1 + 2, first and second apical cones of galea; gden, galeal denticles; lac, lacinia; lam, lamellae; mp, maxillar palpus; mst, mediostipes; mxinc, maxillary incisivi; plam, pectinate lamellae; st, stipes. Images not to scale to each other, scale bar for a) and f) 100 μm; for c) + d) + f1) 10 μm; for f2) 1 μm. Please click on the figure to activate the 3D content. For muscle references see main text. ; c2) detail of the apical area of the galea (meso-anterad view); d) detail of the apical and subapical area of the lacinia; e) the muscle equipment in anterior view; f) anterior view of the outer anatomy; f1) detail of the apical area of the maxillary palpus; f2) detail of a sensillum at the apex of the maxillary palpus. Abbreviations: bst, basistipes; ca, cardo; cal, lever of the cardo; csen, conical sensillum; gal, galea; gc1 + 2, first and second apical cones of galea; gden, galeal denticles; lac, lacinia; lam, lamellae; mp, maxillar palpus; mst, mediostipes; mxinc, maxillary incisivi; plam, pectinate lamellae; st, stipes. Images not to scale to each other, scale bar for a) and f) 100 μm; for c) + d) + f1) 10 μm; for f2) 1 μm. Please click on the figure to activate the 3D content. For muscle references see main text.

click 3D
( Figures 2b + d). Mesally three wart-like groups of sensilla occur that closely resemble those on the apical segment of the maxillary palpus; each sensillum is composed of a basal cylindrical segment densely covered by microtrichia and a single distal tubular extension (Figures 2b + c).

Hypopharynx, epipharynx, and salivarium
The hypopharynx has a wide lumen and is strengthened by a suspensorium made of several sclerites (Figure 6b) to which muscle bundles attach (see below). Two transverse sclerotized ribbons are present in the anterior part of the hypopharynx and cross the anterior surface. The proximal transverse sclerite (pts; Figure 6b) originates from the oral arm (oa), the distal transverse sclerite (dts, Figure 6b) originates from the distal part of the suspensorium below the loral arm (loa). The epipharynx (= inner side of the clypeolabrum) is a concave structure and bears two fields of hairs on the inner side. The mandibles fit into the concave space of the epipharynx.
When closed, the right mandible is positioned a short distance in front and more ventrally of the left one. The joint, unpaired efferent duct of the salivary glands and labial nephridia is bowl-shaped ( Figure 6b)

Foregut
The foregut has a wide lumen and is not distinctly subdivided into pharynx and oesophagus. Various muscles hold the foregut into position (0ci1, 0bu1-3, 0bu5 + 6,

Discussion
In this study we characterize all muscles and endoskeletal features of the head of T. gertschi. The description of the outer anatomy largely conforms with the one of Wygodzinski [10]. In contrast to Wygodzinski [10] we interpret the general head organisation as orthognathous (only the labial palpi point to the rear (Figures 2a & 4b)), although we believe that this incongruence is due to the different use of the terms "orthognathous" and "hypognathous". In the Anglo-Saxon language the terms "hypognathous" and "orthognathous" are often used synonymously [34]. The interpretation of the region dorsal of the eyes as occiput (Figures 3a, 4b) -and consequently the designation of the postoccipital area -is unclear. Matsuda [35] and Snodgrass [36] proposed a tripartite gnathocephalon where the postocciput is regarded as the labial segment and the occiput as the maxillary segment. They are supposed to be separated by the occipital ridge and the postoccipital ridge. Accordingly, the frons harbours the eyes. However, we consider cephalic ridges and sulci as lines of mechanical strengthening (ridges) or weakening (sulci) which either deflect mechanical strain or serve as predetermined breaking points during ecdysis. Thus, they are not associated with any head segments [37][38][39][40] or head regions such as occiput, postocciput, or frons. For the time being we adhere to the latest accounts on this problem [11,14], which favour the interpretation of the ridge posterior of the eyes as the postoccipital ridge.
In general, the cephalic morphology of T. gertschi is characterized by the presence of several potential autapomorphies: absence of 0lb2 and 0hy10, apex of the labial palpi with two different types of sensilla, and a clypeus and labrum each with two distinguishable subareas. Due to the absence of an intraclypeal ridge we refrain from interpreting the clypeus as distinctly separated into ante-and postclypeus despite the differing surface structure (Figure 5d).

Monophyletic Dicondylia and Pterygota are corroborated
Generally, the monophyly of Dicondylia and Pterygota is well supported by molecular and morphological data [2,41] even though some authors doubted this view [42,43]. Our phylogenetic analysis including T. gertschi as well as L. saccharina, T. domestica and A. formicaria corroborated the monophyly of Dicondylia. Potential autapomorphies are the presence of a coronal suture, cuticular dorsal tentorial arms, presence of an additional anterior mandibular joint (for which the group is named), presence of M. labroepipharyngealis (0lb5), M. verticopharyngealis (0ph1), M. tentoriopharyngealis (0ph2) and the five-segmented maxillary palpus.
The monophyly of winged insects is strongly supported (BR 9; Figure 10). Unambiguous autapomorphies of Pterygota are the divided clypeus (15:1), the origin of the antennal muscle 0an2 at the dorsal tentorial arms (32:2; although character states are shifting among Neoptera), the fusion of the pre-and posttentoria (47:1) and the loss of several tentorial muscles (56-59:1), as well as the absence of a circumesophageal vessel ring (35:1) and the loss of labial musculature (0la7; 115:1 & 0la9; 118:1). The loss of hypopharyngeal muscles 0hy6 (131:1) and 0hy11 (133:1) may represent further autapomorphies of Pterygota, but the ancestral states of these characters remain ambiguous due to the lack of a more distantly related outgroup.

Head data supports monophyletic Zygentoma
The position of T. gertschi was discussed controversially: It was considered as sistergroup to Euzygentoma [17,18], as sistergroup to Euzygentoma + Pterygota [15,16], or as sistergroup to Nicoletiidae and Ateluridae [10,19] within Euzygentoma. Analysis of cephalic data did not support a hypothesis about the phylogenetic position of T. gertschi, since the M. hypopharyngo-mandibularis (0md4) was considered absent [11,14] and due to the existence of a ligamentous connection of M. tentoriomandibularis lateralis inferior (0md6) between the mandibles [11,44,45]. In our specimens a M. hypopharyngo-mandibularis (0md4) is clearly visible, origin, insertion, and course of this muscle are in line with its organisation in other dicondylian taxa. Thus, only the ligamentous connection of muscle M. tentoriomandibularis lateralis inferior (0md6) between the mandibles [11,44,45], which is shared with Archaeognatha, is left as an argument against the monophyly of Zygentoma. It has to be emphasized that parts of this muscle are also in contact with the anterior tentorium despite the presence of a ligamentous connection. Archaeognatha possess only a ligamentous connection and the remaining insects (excl. Tricholepidion) exhibit an origin of this muscle exclusively on the tentorium. Our phylogenetic analysis corroborates monophyletic Zygentoma despite the inclusion of the mandible ligament in the character matrix (character 69) with high support (BR 12; Figure 10). Synapomorphies of Euzygentoma and T. gertschi revealed in our study concern the composition of the labial musculature (107:0; 108:0; 109:0; 110:0). Euzygentoma and T. gertschi possess a remarkable set of extrinsic labial muscles originating from the postoccipital region and extending dorso-ventrally through the whole head into the labium. Archaeognatha and Pterygota clearly do not possess this set of labial muscles [32,44,46]. The situation in Protura, Collembola, and Diplura is unclear due to homologisation problems of the labial structures and the corresponding muscles. Both, T. gertschi and Euzygentoma, possess four labial palpomeres (103:3). Also, the intralabral muscle equipment is characterized by an additional muscle, the M. epistoepipharyngalis (0lb3; 22:0). In the present analysis the low number of ommatidia (less than 80; 2:1) is a homoplastic character since Grylloblattodea also show reduced eyes [30] which however may be due to convergence. Wing-like tentorial processes reaching into the lumen of the mandible are also present in Odonata (49:1; [32]). A M. labrolabralis (0lb6; 24:0) and the cylinder shaped posterior mandibular joint (68:0) are characters also present in Ephemeroptera [14], the number of lacinial incisivi (83:0) is shared with the grylloblattodean Galloisiana yuasai [30]. Except for the (homoplastic) loss of the mandible ligament (see above), the loss of ocelli (but see below), and the loss of M. verticopharyngealis (0ph1) we found no cephalic apomorphies characterizing Euzygentoma. Evidence from cephalic characters therefore suggests monophyletic Zygentoma, with T. gertschi as sister to Euzygentoma.

Other character systems
With few exceptions most molecular studies advocate monophyletic Zygentoma. Based on secondary structure alignment of the 18S unit Kjer [5] recovered Tricholepidion as sistergroup to Odonata with low support while Giribet [47], using a more inclusive molecular dataset and a compilation of published morphological data, proposed Tricholepidion as sister taxon to Euzygentoma + Pterygota with low support. All other molecular works supported monophyletic Zygentoma [3,4,6,7].
The monophyly of Zygentoma was further argued by the occurrence of sperm conjugation (or sperm pairing) in T. gertschi and Lepismatidae [48]. Zrzavý [1] considered this a weak argument, since the mode of sperm aggregation seems to be quite diverse among Zygentoma (reviewed by Dallai et al. [19]). In a detailed study on the sperm ultrastructure and sperm pairing mode of T. gertschi Dallai et al. [20,21] discovered that the spermatozoa morphology in T. gertschi resembles the ancestral state of insects with a 9 + 9 + 2 axonemal pattern and accessory tubules with 16 protofilaments. The sperm pairing in T. gertschi is a true fusion between two spermatozoa along the entire sperm head region, which is different from sperm aggregations in other Zygentoma [19]. Characters related to sperm conjugation accordingly provide no convincing evidence in favour or against zygentoman paraphyly. The same concerns the number of ovarioles that seems to be plesiomorphic in T. gertschi (seven, as in Archaeognatha [10]). The ovariole number is reduced in euzygentomans (five in Lepismatidae; three in Nicoletiidae; [49]) and highly variable in Pterygota [49].
Other possible synapomorphies of Euzygentoma and T. gertschi include similarities in the mating behaviour of T. gertschi and Lepismatidae [50], a unique type of sensillum on the terminal filament of males of T. gertschi and some Nicoletiidae [10], and the widened apical segment of the labial palpus. Although the similarity of the main elements of mating is remarkable, the mating behaviour among Zygentoma is variable and especially difficult to compare between Archaeognatha and Zygentoma: Archaeognatha show at least three distinctly different modes of sperm transfer, another two are hypothesized for Petrobiellus and Mesomachillis species [46]. The lack of a robust phylogeny of Archaeognatha presently impedes a polarization of the modes of sperm transfer in this taxon. The scattered occurrence of special sensilla on the terminal filaments of males of some Nicoletiidae (including some Ateluridae) likewise demands a robust hypothesis on the interrelationships among blind silverfish to clarify the phylogenetic significance of the presence of these sensilla in T. gertschi. The widening of the apical labial palpus segment is paralleled in males of several Machilidae and Meinertellidae (e.g. Silvestrichilis, Trigoniophthalmus, Promesomachilis, nearly all Meinertellidae [46]) and, hence, seems to be homoplastic.
Also, we interpret the five-segmented tarsi [10] not as an autapomorphy of T. gertschi, since these occur as well in several pterygote taxa and the phylogenetic significance is therefore unclear. Zrzavý [1] considered five-segmented tarsi as the plesiomorphic condition for Pterygota. Taking into account the tarsus configuration in T. gertschi, it is conceivable that five-segmented tarsi already existed in the stem lineage of Dicondylia. Engel [18] even considered the five-segmented tarsus as the ancestral state of insects, with the number of tarsomeres being reduced not only in Euzygentoma, but also in Archaeognatha. However, it has to be emphasized that the number of tarsomeres is variable within Archaeognatha (2-3) and Zygentoma (2-4 in Euzygentoma; e.g. [46]).
Engel [18] proposed the dorsoventral flattening of the body as a synapomorphy uniting all Zygentoma. He classified T. gertschi in its own family Tricholepidiidae because the extinct type species of the Lepidotrichidae, Lepidotrix pilifera Silvestri, 1912, might be more closely related to the Euzygentoma than to T. gertschi due to the apparent absence of ocelli [17]. Engel [18] proposed the loss of ocelli as a synapomorphy uniting a clade "Neozygentoma" (= L. pilifera + Euzygentoma). Although T. gertschi clearly possesses three ocelli, we consider this a weak argument for this classification, since loss of ocelli occurred several times within Dicondylia, e.g. in Xenonomia (= Notoptera + Mantophasmatodea), Phasmatodea [51] and Zoraptera [52]. However, a more thorough re-examination of L. pilifera is mandatory. As for the head (judged from Silvestri's illustration of L. pilifera, his Figure V1 [53]) we particularly consider the corresponding, unique expression of the occiput as potentially synapomophic. excellent support at the synchrotron facilities. SR-μCT scanning was done under proposals no. I-20080169, I-20100029, I-20120065 (DESY), and 20110069 (PSI) which is gratefully acknowledged. We would also like to thank two anonymous reviewers for their helpful comments. The Alexander Koenig Gesellschaft (AKG) provided financial support for the field trip. The members of the ZFMK provided support in lab and field.