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 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 pro- and 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 star-shaped 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 four-sided 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 bush-crickets (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.
On the other hand, the alternative hypotheses also gain support by few characters of the thoracic musculature (Fig. 7). Gryllidae and Rhaphidophoridae share the presence of Ivlm6. However, this ventral longitudinal muscle frequently occurs within the Polyneoptera: in Austrophasma caledonensis (m26) [48], Periplaneta americana (101) [74], Grylloblatta campodeiformis (81) [75], Oligotoma saundersii (35) [44], and Zorotypus hubbardi (Ivlm6) [36]. Considering the thoracic muscular system, the presence of muscle Iscm6 and IIspm3 are the unique common characters of Gryllidae and Tettigoniidae. Nevertheless, Iscm6 is also present in the outgroup representatives Atractomorpha sinensis (29) [44] and Austrophasma caledonensis (m34) [48]. Muscle Iscm6 connects the profurca with the trochanter of the foreleg. In Troglophilus, the profurca is relatively short and does not extend beyond the opening of the coxa. This specific morphology would not allow lscm6 to reach the trochanter, which, from a functional point of view, could explain its secondary absence in Troglophilus. Although lacking in the representatives of the Caelifera, muscle IIspm3 appears to represent a common character of other polyneopteran taxa since it is present e.g. in Blattodea, Periplaneta americana (149) [74], Phasmatodea, Carausius morosus (IIildvm) [52] and Megacrania tsudai (148) [53], Mantophasmatodea, Austrophasma caledonensis (m51) [48], and Zoraptera, Zorotypus hubbardi (IIspm3) [36].
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. pleura-sternalis 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 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 [41–43]. 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-es1) [46], and in the mole cricket Gryllotalpa gryllotalpa (LS-EP2) [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].
Regarding the two major lineages of Orthoptera, Caelifera (grasshoppers) and Ensifera (katydids and crickets), muscle tpm10 is only known to exist in the meso- and metathorax of ensiferan taxa [41, 44, 76]. Only Maki [44] described a muscle tpm10 in the mesothorax of the African Migratory Locust Locusta migratoria migratorioides (see Additional file 2), but neither Albrecht [67] observed this muscle in the European Migratory Locust Locusta migratoria migratoria, nor did Snodgrass [56] in his study about the thoracic morphology of the Carolina Grasshopper Dissosteira carolina. In general, the number of tergopleural muscles that have been described for Locusta (II/IIItpm1, II/IIItpm2, II/IIItpm5, II/IIItpm9 and IItpm10) is exceptionally large [44]. Somewhat surprisingly, only M. epimero-axillaris tertius (II/IIItpm9) is known in Locusta migratoria migratoria (85 and 114) [67], Dissosteira carolina (85 and 114) [56], the wingless morabine grasshoppers (tergopleural muscle) [61], and even in the brachypterous Atractomorpha sinensis (37/38 and 62/63) [44]. In wingless Caelifera, like Lentula callani [77] and Cephalocoema albrechti [57], even this muscle is reduced and not a single tergopleural muscle has ever been reported. In summary, the distinctive set of tergopleural muscles differs significantly between Caelifera and Ensifera and the role of these muscles after wing loss is markedly dissimilar.
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 [81–84]. 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 boundary ~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].