The thorax of the cave cricket Troglophilus neglectus: anatomical adaptations in an ancient wingless insect lineage (Orthoptera: Rhaphidophoridae)

Background Secondary winglessness is a common phenomenon found among neopteran insects. With an estimated age of at least 140 million years, the cave crickets (Rhaphidophoridae) form the oldest exclusively wingless lineage within the long-horned grasshoppers (Ensifera). With respect to their morphology, cave crickets are generally considered to represent a `primitive’ group of Ensifera, for which no apomorphic character has been reported so far. Results We present the first detailed investigation and description of the thoracic skeletal and muscular anatomy of the East Mediterranean cave cricket Troglophilus neglectus (Ensifera: Rhaphidophoridae). T. neglectus possesses sternopleural muscles that are not yet reported from other neopteran insects. Cave crickets in general exhibit some unique features with respect to their thoracic skeletal anatomy: an externally reduced prospinasternum, a narrow median sclerite situated between the meso- and metathorax, a star-shaped prospina, and a triramous metafurca. The thoracic muscle equipment of T. neglectus compared to that of the bush cricket Conocephalus maculatus (Ensifera: Tettigoniidae) and the house cricket Acheta domesticus (Ensifera: Gryllidae) reveals a number of potentially synapomorphic characters between these lineages. Conclusions Based on the observed morphology we favor a closer relationship of Rhaphidophoridae to Tettigoniidae rather than to Gryllidae. In addition, the comparison of the thoracic morphology of T. neglectus to that of other wingless Polyneoptera allows reliable conclusions about anatomical adaptations correlated with secondary winglessness. The anatomy in apterous Ensifera, viz. the reduction of discrete direct and indirect flight muscles as well as the strengthening of specific leg muscles, largely resembles the condition found in wingless stick insects (Euphasmatodea), but is strikingly different from that of other related wingless insects, e.g. heel walkers (Mantophasmatodea), ice crawlers (Grylloblattodea), and certain grasshoppers (Caelifera). The composition of direct flight muscles largely follows similar patterns in winged respectively wingless species within major polyneopteran lineages, but it is highly heterogeneous between those lineages. Electronic supplementary material The online version of this article (doi:10.1186/s12862-016-0612-5) contains supplementary material, which is available to authorized users.


Background
The evolution of wings is considered to be a key innovation responsible for the unrivaled evolutionary success of insects, improving dispersal capability, predator avoidance, as well as the access to scattered food sources and mating partners [1]. Beyond flight, wings can provide additional advantages, contributing to thermoregulation, defensive behavior and acoustic communication [2][3][4]. Yet, wing loss is a common phenomenon among pterygotes [1]. In Ensifera (long-horned grasshoppers), one of the most species-rich lineages among the Polyneoptera, wings are often reduced to tiny remnants whose only purpose appears to be the production of sound [5,6]. Orthoptera in general have long been of interest to scientists studying intra-specific acoustic communication and hearing systems. Crickets (Gryllidae) and bush-crickets or katydids (Tettigoniidae) in particular are well known for their elaborate acoustic signaling via tegminal stridulation that is associated with mating and territorial behavior [4]. In the last century, numerous biologists dedicated their research to bioacoustics and countless studies have been conducted illuminating the neuroanatomical [7,8], behavioral [9] and evolutionary [10,11] background of ensiferan bioacoustics.
Some ensiferan taxa have completely reduced their wings, nevertheless. To understand the evolution of bioacoustics within the Ensifera special attention was paid to these wingless and deaf taxa, such as the Rhaphidophoridae, commonly known as camel and cave crickets. The neuroanatomy of their chordotonal organs [10] as well as their vibratory communication through low frequencies [12] is assumed to reflect the ancestral condition of bioacoustics within the Ensifera. Also in regard of their overall morphology, cave crickets are considered a´primitivel ineage among Ensifera preserving several characters in their plesiomorphic state, e.g. the morphology of the ovipositor, the absence of tarsal pulvilli and the absence of posterofurcal connectives in the thorax [13]. With about 550 described species, these insects form an ecologically specialized group mainly adapted to cave life [5]. Rhaphidophoridae has a disjunct geographical distribution restricted to the temperate areas of the Northern and Southern hemispheres as reflected by their phylogeny [14]. Rhaphidophoridae comprises two major groups: Rhaphidophorinae, distributed in Eurasia and North America, and Macropthinae that is restricted to South Africa, South America and New Zealand [15,16]. Although the monophyly of Rhaphidophoridae is well supported in molecular analyses [17][18][19][20], cladistic analyses of morphological characters indeed could not identify any supporting apomorphy for this clade yet [21,22]. The species Troglophilus neglectus investigated in this study appears to branch off from a basal node, forming the sister taxon to the remaining Rhaphidophoridae [19]. In this respect, T. neglectus likely retains characters from the last common ancestor of Rhaphidophoridae and can be considered representative for this taxon in general.
Numerous hennigian (mental) and cladistic studies of Ensifera including Rhaphidophoridae have led to competing hypotheses with respect to the relative positions of the two most species-rich groups within the Ensifera, the true crickets (Gryllidae) and the bush-crickets (Tettigoniidae) (Additional file 1). Traditionally, ensiferan taxonomy is based on the morphology of wings and wing venation in particular. Interestingly, the phylogenetic hypotheses based on this specific character complex differ remarkably. Following the classification scheme of Handlirsch [23], Zeuner [24] proposed a closer relationship of crickets ('Grylloidea' therein) and bush-crickets ('Tettigoniidae' therein) and considered both taxa as having evolved from different fossil representatives of the Prophalangopsidae. He considered the tegminal stridulation and its specific wing morphology as an apomorphic character in the last common ancestor of crickets and bush-crickets. On the other hand, Karny [25,26] and Sharov [27] shared the opinion that the true crickets and relatives (mole crickets, Gryllotalpidae, and antloving crickets, Myrmecophilinae) originated from the gryllacridids (including Rhaphidophoridae), whereas the bush-crickets (Tettigoniidae) were assumed to form an independent lineage within the Ensifera. However, the majority of hennigian and cladistic morphological studies [13,21,22,28] as well as phylogenetic analyses based on molecular data [19,[29][30][31][32][33] propose a division of the Ensifera in two major groups: the "grylloid" clade, including true crickets (Gryllidae), mole crickets (Gryllotalpidae) and antloving crickets (Myrmecophilinae), and a "tettigonioid" clade, comprising the bush-crickets (Tettigoniidae), cave crickets (Rhaphidophoridae), wetas (Anostostomatidae), Jerusalem crickets (Stenopelmatidae) and raspy crickets (Gryllacrididae). Dune crickets (Schizodactylidae) are assigned to either of these two clades according to different authors [21,22].
While studies solely based on molecular data may provide a robust phylogenetic framework for any given organismic group, comparative morphological research is essential for interpreting evolutionary scenarios [34] and tracing functional transformations and adaptations [35]. In particular, the morphology of insect thoraces has repeatedly played a substantial role in understanding the systematics and evolution of certain insect groups [36][37][38][39]. In Ensifera this character complex is hitherto insufficiently studied, with publications that either give only a scarce description of the thoracic skeleton and/or merely include a part of the thoracic musculature. Very few detailed investigations of ensiferan thoraces provide characterizations of skeletal structures in addition to a complete description of the muscular equipment. These studies only consider representatives of the most species-rich ensiferan lineages: Voss [40][41][42][43] gives an exceedingly detailed description of the thorax of the house cricket Acheta domesticus (Gryllidae), whereas Maki [44] provides the only existing description of the thoracic musculature of a bush-cricket, Conocephalus maculatus (Tettigoniidae). Studies focusing on the thoracic morphology of Rhaphidophoridae are scarce. Carpentier [45] gives a brief description of the thoracic skeleton of the greenhouse stone cricket Diestrammena asynamora (Rhaphidophorinae) in addition to a study of its pleural musculature [46]. Furthermore, Richards [47] presents a fragmentary description of the thoracic morphology of Macropathus filifer, a rhaphidophoridean species belonging to the southern group Macropathinae.
Here we present a detailed description of the skeletal structures and the muscular equipment of the thorax of the East Mediterranean cave cricket Troglophilus neglectus (Rhaphidophorinae). The thoracic morphology of T. neglectus is compared to the conditions found in other representatives of Orthoptera in order to detect possible apomorphic traits of Rhaphidophoridae. Furthermore, the investigated character complex is evaluated in the context of its phylogenetic information content, and potential synapomorphies of the competing phylogenetic hypotheses of ensiferan relationships are discussed. Moreover, the general nomenclature recently proposed for thoracic musculature of Neoptera [36] is critically reviewed in light of our results. It is evident that within the Neoptera wings were lost several times independently in evolution and this was a step-like process with numerous morphological transformations in each lineage. Therefore, our observations are compared to the thoracic morphology of other wingless polyneopteran representatives, such as Zoraptera [36], Mantophasmatodea [48] or Phasmatodea [49] in order to compile common adaptations of the thoracic skeletal and muscular system related to secondary winglessness. Based on our novel anatomical data we will provide a detailed description of the consequences of wing loss on the functional anatomy of insect thoraces and thoroughly address the question whether these transformations follow a similar pattern.

Specimens
The specimens investigated in this study were collected in Brje pri Komnu, Slovenia, in July 2008 and identified as Troglophilus (Paratroglophilus) neglectus Krauss, 1879 [50]. All specimens were preserved in 70% ethanol. For the sake of consistency in subsequent comparative studies, all investigated specimens are female adults. In total, four individuals were investigated using the following different methods.

High-resolution photography
Three specimens were used to investigate and illustrate the thoracic skeleton. One complete and undamaged specimen was dehydrated in a graded ethanol series and critical-point dried (Balzer CPD 030) to visualize the outer lateral and dorsal view. Another specimen was sagitally cut and macerated in 5% KOH (1 h in a heating cabinet with 60°C) and likewise dried at critical point. Criticalpoint drying was applied to improve the contrast of the thoracic sclerites against the membranous areas and to visualize the sclerites in more detail. One specimen was fixed in a ventrally overstretched position to expose the neck region and subsequently dried using the HMDS (Hexamethyldisilazane, Carl Roth GmbH & Co KG, item number 3840.2) procedure [35]. Photographs of the HMDS-dried specimen were taken using a digital camera (OLYMPUS Pen E-P2) mounted on a stereomicroscope ZEISS Stemi SV11. The critical-point dried specimens were photographed with a CANON EOS 550D equipped with a macro lens (100 mm) and a ring flash (METZ 15 MS-1). The overall sharp images are composed of image stacks edited in Helicon Focus® (Helicon Soft) and Adobe Photoshop® CS3.

Synchrotron radiation micro computer tomography (SRμCT) and 3D-reconstruction
In order to investigate the thoracic musculature, one specimen was dehydrated in a graded ethanol series, criticalpoint dried (Balzer CPD 030) and mounted on a specimen holder (aluminium stub). The scan was performed at the synchrotron radiation facility BESSY II (Berlin, Germany). The three-dimensional model of the thorax was created using AMIRA®5.4.3 and Autodesk Maya® 2013. Rendered images were edited using Adobe Illustrator® CS3.

Skeleton
The thorax of T. neglectus comprises approximately two thirds of the total body length and is strongly curved downwards with the dorsal side nearly two times longer than the ventral side. The sclerites are colored light brown, speckled with dark reddish brown. All thoracic terga are ventrally elongated and saddle-shaped, masking great parts of the thoracic pleura in a lateral view (Fig. 1a). Wings and wing base sclerites are lacking. The phragmata are weakly developed and function as attachment points for the poorly developed dorsal longitudinal muscles. Ventrally, the anterior parts of the sterna, the membranous areas between these sclerites, and the inner surfaces of the coxae are covered by numerous setae (Fig. 1e).

Mesothorax
The meso-and metathorax are almost identical in size. Like the pronotum nt1, also the pterothoracic nota nt2/ nt3 show no external or internal sculpturing and are ventrally elongated covering the most part of the pterothoracic pleura (Fig. 1a, d). The mesopleura has a triangular form tapering at the dorsal side. The mesepisternum est2 is much broader than the epimeron em2 (Fig. 2). The mesepisternum is folded inwards at the anterior edge projecting into a median direction in an obtuse angle. This inwardly folded part of the episternum is referred to as anterior margin amest2 (Fig. 2a, e) and serves as an attachment area for several muscles (m38, m39). The anterior edge of the mesepisternum, connecting the episternum with its anterior margin, is forming a strongly sclerotized ridge (marked by white asterisks in Fig. 2a). The anterior margin of the mesepisternum extends medially onto the level of the trochantinocoxal joint. A massive and long pleural arm pla2 protrudes from the straight mesopleural ridge plr2 (Fig. 2d, e). A sclerotized bridge between the pleura and the sternum is absent in the mesothorax. The mesosternum st2 has a trapezoid shape, the longer edge orientated towards the head. The margins of the mesosternum are relatively indistinct because it is not delimited by strongly marked ridges as is the prosternum. The furcal pit fup2 and the spinal pit spp2 are located along a longitudinal groove at the posterior margin of the mesosternum st2 (Fig. 1e). The mesothoracic furca fu2 has a long lateral process lfup and a short posterolateral process plfup (Fig. 2d). The form of the mesothoracic spina sp2 is reminiscent of a butterfly with expanded wings consisting of paired dorsolateral and ventrolateral processes and an unpaired posterodorsal one (Figs. 2d, e; 4b). The mesospina is situated slightly posterior from and between the laterally exposed furcae. A distinct and isolated spinasternum is absent. Directly posterior to the mesospinal pit spp2, the sterna of the meso-and metathorax are flexibly connected by a lathy median sclerite ms (Mediansklerit in [13]), Fig. 1e). The slender and feather-shaped mesothoracic trochantin ti2 articulates anteroventrally with the coxa cx2.
Thoracic musculature of T. neglectus and its homologization with that of other Neoptera The thoracic muscles of T. neglectus are illustrated in Figs. 3 and 4. The detailed description of these muscles is provided in Table 1 containing origin, insertion and specific characteristics. In addition, Table 1 provides a hypothesis for the homology of the muscles of T. neglectus with the muscles generally reported from neopteran insects according to the nomenclature of Friedrich & Beutel [36]. In general, a thoracic muscle is treated as an individual unit when both origin and insertion and, in addition, the function of this specific muscle are different from other thoracic muscles found in the thorax. Muscles that possess several bundles are characterized through differently originating muscle parts running together in one tendon at a common insertion point (e.g. m16). On the other hand, muscles can run parallel but their origin and insertion is clearly separated nontheless having the same function. These muscles are treated as derivatives of a single muscle (e.g. m44, m45). The nomenclature of neopteran thoracic muscles presented by Friedrich & Beutel [36] provides a solid basis for homologizing thoracic muscles across insect groups. In some cases, however, the homologization of the thoracic muscles of Troglophilus with the muscles of the "generalized neopteran thorax" [36] proves to be difficult, because muscles are solely defined by their origin and insertion points. While we were able to largely homologize the thoracic muscles unambiguously, we will discuss some problematic cases in the following: The M. pronoto-trochantinalis anterior (Idvm13) and M. pronoto-trochantinalis posterior (Idvm14) both share the same insertion point on the trochantin and have only a slightly different origins on the pronotum: Idvm13 originates from the anterior region of the pronotum, whereas Idvm14 arises from the central region of the pronotum [36]. In Troglophilus, the muscle m8 originates at the dorsolateral area of the pronotum slightly above the cryptopleura, inserting at the trochantin via a long and thin tendon. As m8 is the only muscle originating from the dorsal area of the pronotum it is questionable whether m8 is homologous to Idvm13 or Idvm14. Therefore, further criteria for homologization are necessary. A similar muscle with a long thin tendon is also present in other ensiferans [13]. According to Ander [13], the point of origin of this pronotal muscle has shifted from an anterior laterodorsal area above the cryptopleura to the lateral or central area of the pronotum behind the cryptopleura. Thus, the muscle m8 of Troglophilus is most likely homologous to Idvm13 according to the nomenclature of Friedrich & Beutel [36].
The M. profurca-phragmalis (Idvm10) is a common feature among major polyneopteran taxa [36,48]. This muscle usually connects the profurca with the prophragma. However, in some orthopteran species, like in the grasshopper Dissosteira carolina (muscle 59) [56] or the stick grasshopper Cephalocoema albrechti (muscle 59) [57], Idvm10 has an insertion point shifted to the anterior part of the mesopleura. In Troglophilus, both conditions are present at the same time (m7 and m12). The muscle m7 is undoubtedly homologous to Idvm10 as it arises on the dorsal face of the profurca and inserts at the ventrolateral part of the prophragma. The second muscle (m12) takes a more horizontal course and arises from the ventral surface of the profurca inserting ventrally at the anterior margin of the mesepisternum. Because of their diverging courses and their differing origins on the profurca, the muscles m7 and m12 are most likely two separate muscles and not portions of a single muscle. Therefore, we conclude that muscle m12 of Troglophilus is homologous to M. profurca-intersegmentalis posterior (Ispm5) [36]. This assumption is also supported by the presence of serially homologues of m12 in the mesoand metathorax of Troglophilus (m36 and m59). Furthermore, a simultaneous presence of Idvm10 and Ispm5 is only known from Phasmatodea (Megacrania tsudai, Carausius morosus) and Embioptera (Oligotoma saundersii) [36]. In contrast to the morphology of Troglophilus, the muscle Ispm5 is attached to the peritreme in Megacrania [53] and Oligotoma [44], but to the intersegmental fold in Carausius [52]. These different attachment points cause uncertainties in regard to the homology of the muscle m12. Therefore, a question mark is added here (see Table 1).
In the generalized neopteran thorax, three pterothoracic dorsoventral muscles are attached to the posterior coxal rim [36]: M. noto-coxalis anterior (II/III dvm4), M. noto-coxalis posterior (II/IIIdvm5) and M. coxasubalaris (II/IIIdvm6). In winged Neoptera, the muscles II/IIIdvm4 and II/IIIdvm5 originate at the central region of the nota, while II/IIIdvm6 inserts at the subalare. According to literature data [48,49], the insertion point of II/IIIdvm6 is translocated to the lateral region of the nota in wingless Neoptera. This interpretation is consistent with the assumed tergal origin of the subalare, as proposed before [44,58,59]. In winged orthopterans, all three dorsoventral muscles are also well developed with the muscle II/IIIdvm6 inserting at the subalare. In contrast, the same muscle inserts at the epimeral face of the pleura in wingless Orthoptera: in the cave crickets Troglophilus neglectus (m32 and m55; present study) and Diestrammena asynamora (cx-em 2 ) Table 1 List of thoracic muscles of the cave cricket Troglophilus neglectus, specifying origin and insertion of each muscle including noteworthy characteristics and corresponding figure in the article. Furthermore, homologization (Hom*) according to the nomenclature after [36] [46], in the New Zealand tree weta Hemideina femorata (Ab4) [60], in the apterous proscopiids Cephalocoema albrechti (90a and 120) [57], in morabine grasshoppers (99 and 129) [61], in wingless females of Pamphagidae, Lamarckiana sp. (depressor extensor muscle) [62], and also in micropterous species of Acrididae, e.g. Barytettix psolus (99 and 129) [63]. These findings are more consistent with the assumption of a pleural origin of the subalar sclerite, as suggested by other authors [40,51,[64][65][66]. It is noteworthy that the hypothesis of a pleural origin of the basalar and subalar plates is exclusively based on developmental studies on orthopterans. With reference to Snodgrass [51], the aforementioned plates of nymphal Ensifera (Gryllus) and Caelifera (Melanoplus) are not yet differentiated from the pleura, and the M. coxa-subalaris (3E' and 3E") arises from the upper edge of the pterothoracic epimeron. Voss [41][42][43] who compared the thoracic musculature of different developmental stages of the house cricket Acheta domesticus also observed the epimeral insertion of the M. coxa-subalaris in the first instar (II and IIIpm6 in [41]; II and IIIldmv2 in [42,43]), in which the basalar and subalar plates (Pleuralgelenkplatten) are not yet present. Muscle m37 of T. neglectus is not described in Orthoptera or other insect taxa [59]. Due to its sternal origin at the anterior face of the mesofurca and its pleural insertion at the posterior edge of the cryptopleura, this muscle should be assigned to the sternopleural muscles [36]. Compared with the generalized neopteran thorax, muscle m37 is likely homologous to M. mesofurcaintersegmentalis anterior (IIspm7) with an insertion point shifted from the intersegmental membrane/ intersegmental sclerite to the posterior edge of the propleura. A muscle connecting the intersegmental sclerite between the pro-and the mesothorax with the mesothoracic furca is present in Corydalus (Megaloptera) [59]. In Mantodea, a muscle that arises on the prosternum near the prothoracic spina inserting at the metafurca, is apparently homologous to muscle IIspm7 [36,59]. The specific traits of m37 in Troglophilus cannot be compared with the conditions reported from the aforementioned insect taxa. For this reason, we cannot homologize this muscle with any muscle listed by Friedrich & Beutel (see Table 1).

Phylogenetically informative characters
The thoracic muscles found in Troglophilus are compared to that of a cricket, Acheta domesticus [40][41][42][43], and a bush-cricket, Conocephalus maculatus [44], in order to find similarities and differences between the major ensiferan groups represented by these species. Two fully winged locusts, the African Migratory Locust Locusta migratoria migratorioides [44] and European Migratory Locust Locusta migratoria migratoria [67], and a brachypterous representative, Atractomorpha sinensis [44], of the Caelifera, the sister group of Ensifera [68,69], are also considered for comparison to delineate apomorphic and plesiomorphic traits. Moreover, further taxa of Polyneoptera, either having fully developed wings or being apterous, are also studied to draw reliable conclusions about the importance and effect of winglessness on the thoracic muscular system. The phylogenetically informative characters, which have a different manifestation in the Caelifera, are compiled in Fig. 5. A table providing the complete data set of the thoracic muscles of the aforementioned representatives is available as an additional data file (Additional file 2).

Characters unique for cave crickets
Rhaphidophorids are generally considered as the morphologically most homogenous taxon within the Ensifera [13,26]. Interestingly, rhaphidophorids are the only ensiferan subgroup for which no apomorphic character was reported in the cladistic analysis of Desutter-Grandcolas [21]. However, the thoracic muscular system of T. neglectus differs in significant points from that of other ensiferans, providing a number of potential autapomorphies (see Fig. 6). In general, the enlarged number of sternopleural muscles is a novelty for Troglophilus. In particular, the presence of m36 (IIspm6) and m37 (IIspm?) is Table 1 List of thoracic muscles of the cave cricket Troglophilus neglectus, specifying origin and insertion of each muscle including noteworthy characteristics and corresponding figure in the article. Furthermore, homologization (Hom*) according to the nomenclature after [36] [41,43], are framed by a rectangle. Species marked with an asterisk (*) bear different names in the respective cited publication (modified after [54] and [55]) Fig. 6 Unique muscular characters of Troglophilus neglectus as compared to other polyneopteran representatives. Potential positive apomorphies are indicated in light grey. Direct flight muscles, as indicated by Voss [41,43], are framed by a rectangle. Species marked with an asterisk (*) bear different names in the respective cited publication (modified after [53] and [54]) unique within Orthoptera. Troglophilus is characterized by a largely reduced set of direct and indirect flight muscles. Both orthopteran representatives of the species-rich crickets (Gryllidae) and bush-crickets (Tettigoniidae) that we used for comparison are fully winged. In contrast, cave crickets completely lack wings. Thus, it is difficult to decide whether a flight muscle absent in Troglophilus is only a result of winglessness or represents an apomorphic character of Rhaphidophoridae. Since the ratio of flightless species to volant ones among orthopterans ranges between 30 and 60 % [1], the small taxon sampling of our study is insufficient to address this question.
It is particularly noteworthy that in Troglophilus the well developed musculature is important for operating the legs. These muscles are attached to the coxal rim or the trochanter and enable diverse movements of the legs. These muscles are either strongly developed, like Mm. noto-trochanteralis (m11, m33, m56), or their number is increased, like in the pro-and mesothoracic sternocoxal muscles scm1 (m23-25, m44-45). This strengthening of the sternocoxal muscles through multiplication is also reported from the wingless New Zealand tree weta Hemideina thoracica [60]. M. coxo-subalaris (II/IIIdvm6), which has an additional function as a flight muscle in winged insects [70], exclusively acts as leg retractor in Troglophilus. Additionally, Troglophilus has several sternopleural muscles that have not been described for other orthopterans. These include the serially homologous muscles m12 (Ispm5?), m36 (IIspm6) and m59 (IIIspm5) as well as the not homologized m37 (IIspm?). The connection of sternal and pleural elements by these muscles might lead to an enhanced movability of the thoracic segments (against each other), since there are no rigid connections of e.g. the pterothoracic sterna as in grasshoppers [13,71]. Together with the strong leg musculature, the sternopleural musculature probably facilitates the scrambling movement of Troglophilus on cave walls and an increased jumping capability.
As suggested by authors of similar morphological studies [13,72], the morphology of the thoracic sternum and associated sclerites in particular differs in decisive points between major ensiferan lineages. Including data on the thoracic skeletal anatomy of Diestrammena asynamora (Rhaphidophorinae) [45,46] and Macropathus filifer (Macropathinae) [47] this specific character complex indeed provides some apomorphic traits for the Rhaphidophoridae. Prothoracic spinasternum and prospina. The characteristics of the prothoracic spinasternum and its internal protrusion, the prospina, have a unique appearance in rhaphidophorids. The prospinasternum of cave crickets is completely reduced externally (see Fig. 1e and [13]). Its presence is only noticeable by the existence of the prospina located in the membranous fold between the proand the mesosternum. In other ensiferan taxa, the prospinasternum is either exposed in the sternal intersegmental fold as a fully developed sclerite or merged with the posterior part of the prosternum or the anterior part of the mesosternum [13,71,72]. Also the starshaped prospina, consisting of paired anterolateral and posterolateral processes and an unpaired anterior process, is a unique feature of rhaphidophorids. It has also been described in Diestrammena asynamora [45] and Macropathus filifer [47], two other representatives of cave crickets. In tettigoniids the prospina is triangular or t-shaped [72], when present. Voss [40] describes the prospina of Acheta domesticus as an irregular foursided plate. The prospina of the mole cricket Gryllotalpa vulgaris is a long blade-like structure [73].
Median sclerite between meso-and metasternum. A narrow median sclerite, situated in a longitudinal arrangement between the sterna of the meso-and metathorax, is a typical feature of all rhaphidophorids [13]. This sclerite is frequently present in other ensiferan taxa, but the specific condition is different. In tettigoniids it can be rectangular or trapezoid, mostly spanning the whole width of the metasternum [72]. A triangular or semicircular sclerite is embedded at the anterior part of the metasternum in Anostostomatidae [13,60], whereas in schizodactylids it is narrow and rectangular, inflexibly connecting meso-and metasternum ( [71], unpublished observations for Comicus FL). Since the anatomical situation in rhaphidophorids is similar to that found in Grylloblatta, Ander [13] assumes that this sclerite is at least the posterior part of the mesothoracic spinasternum, since the mesospina is situated at the posterior end of the mesosternum right between the furcal apophyses. In contrast, Matsuda [59] and Naskrecki [72] refer to this sclerite as metathoracic presternum. As another alternative, Matsuda [59] characterizes the sclerite in question as the secondarily detached anterior part of the metathoracic basisternum. Due to these uncertainties, we simply refer to the sclerite as median sclerite ms following Ander [13].
Metafurca. The shape and specific structure of the metathoracic furca is another peculiarity of the thoracic skeleton of cave crickets. Rhaphidophorids possess a triramous furca with continuously tapered processes: an anterior, a lateral and a posterolateral one (see Fig. 2 and [45,47]). Most other ensiferans have a biramous metafurca bearing a lateral and a posterior process [40,72]. Like rhaphidophorids, the metafurca of Anostostomatidae has three processes, but the lateral one differs in shape from that of Rhaphidophoridae. In Anostostomatidae it is a flat, blade-like structure, termed apophysis wing, which directly projects beneath the pleural arm [60].

Phylogenetic implications
The scarce information available for ensiferan thorax morphology is not yet sufficient for a cladistic analysis.
However, the thoracic characters found in Troglophilus neglectus, Acheta domesticus (Gryllidae) and Conocephalus maculatus (Tettigoniidae) in comparison to other polyneopteran representatives (see Additional file 2) shows potential synapomorphies for certain subgroups within the Ensifera. As summarized in Fig. 7, the most parsimonious hypothesis of the phylogenetic position of cave crickets within the Ensifera supports a closer relationship to bushcrickets (Tettigoniidae) than to true crickets (Gryllidae). Hence, the hypothesis of ensiferan relationships favoured by the majority of authors (see Additional file 1) is also supported by thoracic muscle characters. Interestingly, all of the potential synapomorphies of Rhaphidophoridae and Tettigoniidae are negative character traits, i.e. reductions. This implies that the number of thoracic muscles decreases in a specific lineage among Ensifera, viz. Rhaphidophoridae + Tettigoniidae.
The thorax of Troglophilus neglectus and the evolution of secondary winglessness in general The consequence of wing reduction and flight loss largely affects thorax morphology in insects, both cuticular structures and the muscular system, which includes secondarily undifferentiated terga, less extensive phragmata and reduced or poorly developed dorsal longitudinal muscles (II/IIIdlm1, II/IIIdlm2), as well as the absence of wing base sclerites and associated wing-steering muscles [36,60]. These distinctive traits are also found in the thorax of Troglophilus. In contrast to other wingless taxa like Grylloblatta [75] and the wingless morph of Zorotypus [36], the pleural arms in the pterothorax of Troglophilus are still well pronounced. Additionally, well developed pleural arms seem to be a common feature of Orthoptera, regardless the wing status, either fully winged [40,56], micropterous [63] or wingless [46,57]. In Mantophasmatodea, the well-developed pleural arms are explained by the climbing lifestyle among shrubs [48].
M. pleura-sternalis (II/IIIspm1), which is attached dorsally on the basalare and ventrally on the lateral part of the sternum, is thought to act as an extensor and flexor of the wing, and therefore is considered to be a direct flight muscle [56]. With the exception of Grylloblattodea and Mantophasmatodea, the general trend among wingless insects is the reduction of this muscle [48]. This trend is also observed within Orthoptera. In Caelifera, M. pleurasternalis is present in the meso-and metathorax of winged locusts [44,56], whereas it is absent in the micropterous Mexican grasshopper Barytettix psolus [63], and also reduced in wingless Proscopiidae [57] and morabine b a c Fig. 7 Informative characters of a comparative morphological study of the thoracic muscular system of representatives of Ensifera. The characters are mapped on the three competing hypotheses of the relationship between crickets (Gryllidae), bush-crickets (Tettigoniidae) and cave crickets (Rhaphidophoridae). Based on homologization in Table 1 (compiled in Additional file 2). R! indicates a reduced character in the respective taxa grasshoppers [61]. The assumption that M. pleura-sternalis is at least present in the mesothorax of Ensifera is based on the description of a single cricket species [41][42][43]. After investigation of several additional ensiferan species, we can now reliably conclude that muscle IIspm1 is only present in Grylloidea, e.g. Acheta domesticus (IIpm14) [41] and Gryllus campestris (ls-es 1 ) [46], and in the mole cricket Gryllotalpa gryllotalpa (LS-EP 2 ) [76]. The muscle is lacking in the meso-and the metathorax of the cave cricket Troglophilus, the schizodactylid Comicus calcaris (unpublished observations FL) and the winged bush-cricket Conocephalus maculatus [44]. This reduction of muscle spm1 in the pterothorax, especially in Tettigoniidae, might be a phylogenetically informative character, which needs to be tested in a future cladistic analysis based on an enlarged taxon sampling.
In the pterothorax of Troglophilus, dorsal longitudinal (II/IIIdlm2), dorsoventral (II/IIIdvm1) and tergopleural muscles (tpm) are absent, muscles that are indirectly or directly involved in flying [36,48]. Most notably, the number of wing-steering tergopleural muscles is reduced, as has also been reported from other wingless taxa, e.g. Phasmatodea [49,52] or Orthoptera [57,60]. The only tergopleural muscle retained in both pterothoracic segments of Troglophilus is M. epimero-subalaris (II/IIItpm10). In winged species, this muscle connects the dorsal part of the epimeron with the subalar sclerite [36]. As in Troglophilus, the insertion point of tpm10 is translocated to the notum in wingless species of Phasmatodea [49] or Mantophasmatodea [48].
In Euphasmatodea (the majority of extant stick insects) on the other hand, thoracic morphology of wingless species largely resembles conditions found in Ensifera. Klug [49] observed a significantly reduced set of tergopleural muscles in wingless stick insects, only consisting of muscles II/IIItpm10 and II/IIItpm13 (tpm13 is a unique muscle of Phasmatodea). These partly comparable patterns imply that the mechanism and morphology of secondary winglessness may follow similar routes in closely related taxa. In contrast, in Embioptera (webspinners), the assumed sister taxon of Phasmatodea [69], the set of tergopleural muscles (II/IIItpm1, II/IIItpm5, II/IIItpm6, II/ IIItpm7, II/IIItpm10; homologized in [48]) does not differ between winged males and wingless females of the same species [78,79].
Another pattern providing support for the assumption of similar evolutionary trajectories in closely related taxa can be observed in the entirely wingless Xenonomia [80] comprising heelwalkers (Mantophasmatodea) and ice crawlers (Grylloblattodea). Here, the set of tergopleural muscles is different from that of wingless representatives of Orthoptera, Phasmatodea or Embioptera. Grylloblatta campodeiformis (Grylloblattodea) is characterized by a set of IItpm1/5 and IIItpm1/5 [75] (homologized in [36]). Based on the description of Klug [49], Austrophasma caledonensis (Mantophasmatodea) exhibits the same set of tergopleural muscles in the pterothorax, IItpm1/5 and IIItpm1/5. According to the reinvestigation of the same species [48] a considerably higher number of tergopleural muscles is reported: IItpm1/2/3/4/5/?10 and IIItpm1/2/3/ 4/5/?10. These studies used different µCT data sets for analysis. Depending on the quality of the data sets, it is possible that some muscles were initially overlooked, e.g. tpm10 characterized as a flat muscle closely fitting the skeletal elements. Nevertheless, muscle tpm1 in Klug [49] and the four muscles tpm1/2/3/4 described for Austrophasma by Wipfler et al. [48] are located in the same small area between the anterior part of the tergum and the dorsal part of the pleural ridge. A further explanation of these striking differences might lie in the different life stages or sexes investigated in both studies. Klug [49] examined a nymphal stage of unknown sex of Austrophasma caledonensis, whereas in the study of Wipfler et al. [48] no explicit information about the developmental stage or the sex of the investigated specimens is provided. However, studies about the postembryonic development of the flight musculature of hemimetabolous insects show that these muscles are less developed in early nymphal stages, significantly increasing in size during their ontogenesis [81][82][83][84].
Other studies comparing the thoracic musculature report a differing number of muscles in nymphs and adults of the same species [41,42,85]. In consequence, the presence of tpm1 and tpm5 in the meso-and metathorax of Grylloblattodea and Mantophasmatodea might still be considered a synapomorphic character of both taxa.
Principally, the flight ability and performance of insects also depend on the total mass of flight muscles present, and not only on the concrete set of direct and indirect flight muscles [84]. Nonetheless, the concrete set of tergopleural muscles differs between major insect groups [36]. Regarding the Orthoptera, their flight ability and performance become of secondary importance, since many species primarily move by jumping. In these cases, wings are mainly used to control the direction and trajectory during the jumping process [5,86]. For instance, the house cricket Acheta domesticus [41], with a set of IItpm1/2/5/9/10 and IIItpm1/2/5/9/10, and the tettigoniid Conocephalus (Anisoptera) maculatus [44], with a reduced set of IItpm2/5/9 and IIItpm2/9/10, exhibit similar flight capability [44,86]. On the other hand, the absence of specific tergopleural muscles as in the brachypterous gaudy grasshopper Atractomorpha sinensis [44] having only a single duplicated tergopleural muscle in the meso-and metathorax (II/IIItpm9) causes a low vagility [87]. In contrast, Sipyloidea sipylus, a winged stick insect, only has the ability to control its speed and trajectory during free fall with a set of six different metathoracic tergopleural muscles in the flight apparatus (tpm1/3/4/6/9/10) [49,88]. In conclusion, there appears to be no correlation between an increased number of pterothoracic tergopleural muscles and an enhanced flight capability. However, an extremely reduced set of tergopleural muscles does consequently lead to the inability to fly.
Anatomical structures that are no longer used will be reduced in the course of evolution, and the degree of reduction can be an indicator of the time elapsed [89]. Nevertheless, conservative anatomical elements can be retained although associated traits of the periphery are lost [90]. As we have outlined, the loss of wings in insect groups like Orthoptera, Xenonomia [48] or Phasmatodea [49] has been followed by a number of anatomical adaptations of skeletal and muscular elements in the thorax. The insect lineages compared above exhibit significantly different evolutionary histories in regard of the time span since wing loss, affecting the degree of reduction or anatomical adaptations towards flightlessness. The radiation of Rhaphidophoridae began at least 140 million years ago [16,19]. Thus, the Rhaphidophoridae may represent the oldest exclusively wingless lineage within Ensifera [19], and wing loss occurred most probably in the last common ancestor (autapomorphy) of all Rhaphidophoridae. The likewise wingless Xenonomia, heelwalkers (Mantophasmatodea) and ice crawlers (Grylloblattodea), are roughly the same age as the Rhaphidophoridae [69]. We have demonstrated that the thoracic musculature differs significantly in both lineages. In comparison, the wingless representatives of Euphasmatodea are significantly younger. The diversification of their major extant lineages took place during a period of about 20 million years, and presumably started after the Cretaceous-Tertiary boundarỹ 66 million years ago [91,92]. The thoracic musculature of wingless Ensifera, Rhaphidophoridae in particular, is most similar to the conditions found in the much younger wingless representatives of Euphasmatodea than in the equally old Xenonomia, refuting any dependency between level of reduction and evolutionary time. This might be explained by the degree of correlation of the structures in question to other, still adaptive features [89].

Conclusions
Secondary winglessness, a widespread phenomenon among pterygote insects, largely affects the thoracic anatomy including skeletal structures and the muscular system. By comparing the thoracic morphology of various wingless representatives of Polyneoptera, we demonstrate that anatomical adaptations towards flightlessness, especially regarding the flight musculature, are highly homogenous within major lineages, viz. Ensifera, Caelifera, Xenonomia, or Euphasmatodea. However, in most cases these specific adaptations are strikingly different between the aforementioned taxa indicating a markedly dissimilar role of these muscles after wing loss.
The thoracic morphology of Ensifera is a highly structured character complex whose investigation is a worthwhile endeavor, leading to a deeper understanding of functional adaptations during the evolution of Ensifera in general. We have shown that the thoracic morphology can be a valuable source for characterizing individual ensiferan taxa, providing a number of potential apomorphies for cave crickets (Rhaphidophoridae). Based on our comparison with other ensiferans, we can provide arguments for a closer relationship of Rhaphidophoridae to Tettigoniidae, rather than to Gryllidae. These findings are consistent with previous assumptions [19,21,22].