HOMOLOGY OF PARTICULAR MUSCULAR STRUCTURES IN TURTLES

Muscular Units Unique in Particular Species

Some muscular units are only known in singular species (indicated by two asterisks each in Appendix 1, Figure 5, Figure 6, Figure 7, Figure 8, Figure 9, Figure 10, Figure 11, Figure 12, Figure 13, Figure 14, Figure 15: No. 6, 16, 18, 30, 36, 39, 44, 48, 50, 62, 72) or are found in only a few often closely related species (indicated by one asterisk: No. 10, 20, 32, 33, 61, 83). These findings may be correlated to autapomorphic or plesiomorphic conditions seen in the respective taxa. Moreover, the breadth of the studied taxa and the differing accuracy of the authors, the different extent and the quality of manual dissections, and other techniques used such as serial sectioning (e.g., No. 6, 48), led to the identification of possibly unique structures. In addition, structures reduced during development may be observed as rudiments in subadult specimens (m. levator bulbi, No. 16). Guigova et al. (2009) discovered a contralateral variability in the number of eye muscles (retractor bulbi) in giraffes related to a unilateral combat behaviour in males. Intraspecific and contralateral variability in the number (No. 36) and extent (No. 66, Walter 1887) of cranial musculature also can be found in turtle species. This pattern could possibly be associated first to the phylogenetic identity (Pleurodira vs. Cryptodira) and second to the individual retracting behaviour of the head/neck region. However, if a muscular unit is only described for one species this does not mean that the muscle is absent in all other turtles. A detailed survey comparing several specimens of many species and focussing on all muscular units of the head is still lacking and would provide a more comprehensive understanding of cranial musculature in turtles. A first study tending toward this approach was performed by Jones et al. (unpublished work) on most of the feeding-related muscles of two marine turtle species. Appendix 1 and Figure 5, Figure 6, Figure 7, Figure 8, Figure 9, Figure 10, Figure 11, Figure 12, Figure 13, Figure 14, Figure 15  only represent the current state of knowledge. Once a broader set of taxa and specimens has been examined the scheme presented here can be expanded upon.

External Eye Muscles (No. 1-4, 9-10, 34-38) (nn. III, IV, VI)

In the optic chamber of Tetrapoda four mm. recti (No. 2-4, 37), two mm. obliqui (No. 1, 9), and one m. retractor bulbi (No. 38) are present. The latter autapomorphically exists in this group and represents a separation of the m. rectus posterior of non-tetrapod vertebrates (Nishi 1938b). Within Sauropsida, additional muscles were described to attach the eye bulbus. Together with m. retractor bulbi (No. 38), they form an enormous diversity among sauropsids, and their homology is not completely understood (Nishi 1938b; Underwood 1970). Those differences may be explained by the extent of the membrana nictitans and an increase of its mobility. This "third eyelid" is a medial skin duplication of the lower eyelid and may have a cleaning and protecting function. It can be greatly reduced in some turtle taxa such as Carettochelys insculpta (Underwood 1970, p. 67).

The m. quadratus (No. 35) originates at the medial surface of the eye bulbus, runs caudally to insert into the posterior angle of the eyelids. Alternatively, it may develop two tendons, with the dorsal tendon inserts posteriorly into the upper lid, while the ventral tendon inserts posteriorly into the lower eyelid (Nishi 1938b). Anterior to m. quadratus (No. 35), m. pyramidalis (No. 34) originates at the medial surface of the eye bulbus. Via one tendon it runs to the membrana nictitans (Pelodiscus sinensis), in most species also via a second tendon to the lower eyelid. The muscle only occurs within Archosauria and Testudines and forms a potential synapomorphy of both taxa (Thomson 1932; Schumacher 1972; Rieppel 2004; Eger 2006). Edgeworth (1935) found the pyramidalis muscle in birds to be ontogenetically derived from the abducens primodium, and he identified an n. abducens (VI) innervation. For Emydura subglobosa, I also found an n. abducens (VI) innervation of m. pyramidalis (No. 34), which supports the homology of this muscular structure between birds and turtles. In Chelonia mydas (Edgeworth 1935), Clemmys japanicus (Nishi 1938b), Pelodiscus sinensis (Ogushi 1913b), or Dermochelys coriacea (Schumacher 1972), several modifications of m. pyramidalis (No. 34) are known. To clarify homologies among species, ontogenetic studies are necessary, and several new taxa must be studied because this structure seems to have an enormous diversity.

One may assume two new eye muscles in the ground pattern of Sauropsida (reptiles and birds). One membrana nictitans related muscular structure (~pyramidalis) and one muscle related to the angle of the upper and the lower eyelid (~quadratus). In P. sinensis, both muscles are still separated from each other. In E. subglobosa and other turtles (Eger 2006), only an m. pyramidalis (No. 34) is present inserting to the membrana nictitans, an m. quadratus (No. 35) is absent, but a second tendon of m. pyramidalis also inserts to the posterior part of the lower eyelid. Developmental studies should focus on the early development of this muscle in turtles such as E. subglobosa, which may be (at least partly) a fusion of the two above mentioned muscles in sauropsids. However, m. quadratus (No. 35) may be reduced in most turtles, and m. pyramidalis (No. 34) may have gained a new insertion to the lower eyelid. Though both muscles are innervated by n. abducens (VI), the detailed branching pattern of the nerve may give some information to the identity of the muscle. In C. japanicus the m. quadratus (No. 35), inserting to the upper lid, has one additional insertion to the tendon of membrana nictitans. M. pyramidalis (No. 34) is not present in this species as a separated muscle, but it may be homologous to the superficial part of m. retractor bulbi (No. 38) in C. japanicus, which inserts to the membrana nictitans (Nishi 1938b) – an assumption supported by my own observation of E. subglobosa. This situation could indicate that the pyramidalis muscle (No. 34) phylogenetically/ontogenetically incorporates material from the quadratus (No. 35) as well as from the retractor bulbi (No. 38). With this hypothesis in mind, an insertion of m. pyramidalis (No. 35) to the lower eyelid in most turtles may also be explained.

In Squamata, the bursalis muscle (Underwood 1970) is clearly derived or still connected to the Mutterboden muscle, m. retractor bulbi (Nishi 1938b). One may assume m. bursalis of squamates to be homologous to the m. quadratus of birds and turtles (No. 35), because the m. quadratus (No. 35) is partly connected to retractor bulbi (No. 38) in C. japonicus and D. coriacea. In both species, it partly originates from the interorbital septum, a feature it shares with m. rectus posterior (No. 37), which is known to be the Mutterboden for retractor bulbi in tetrapods (Nishi 1938b). While in amphibians and most squamates the tendon to the membrana nictitans originates from the interorbital septum and is indirectly moved by eye activity, Varanus salvator shows an integration of this tendon to the bursalis muscles (Nishi 1938b). This character is an additional indication that the homology of bursalis is at least partly the Mutterboden of m. pyramidalis (No. 34) in turtles and archosaurs.

As a case of intraspecific variability, Ogushi (1913b) found one unique eye muscle in two adult specimens of P. sinensis inserting to the ventral face of the upper eyelid – m. quadratus superior (No. 36). Although discussed as a potential separation of m. quadratus principalis (No. 35) by Ogushi (1913b) himself, the muscle also may have been derived from rectus anterior (No. 2), which is partly connected to the posterior forth of lower eyelid in D. coriacea. The latter eyelid reaches relatively far into the dorsal region of the head due to the vertical orientation of the eyelids in this species; hence, spatial similarilarity to the insertion site found in P. sinensis exists. The innervation of m. quadratus superior (No. 36) is not clear but Ogushi (1913b) argued for an n. abducens (VI) innervation of the muscle and refutes an n. oculomotorius (n. III) innervation as gained by m. rectus anterior (No. 2).

Ogushi (1913b) described one additional external eye muscle (No. 10) in older specimens of the trionychid P. sinensis that may originate phylogenetically from m. obliquus superior (principalis) (No. 9). In my own observations on Emydura subglobosa, I found such a structure in one subadult specimen. In contrast to the finding in P. sinensis, the structure was only present as a portion in E. subglobosa and consequently should be called m. obliquus superior Pars accessorius (No. 10) in that specimen (Figure 28.1) versus m. obliquus superior accessorius (No. 10) in the respective specimens of P. sinensis.

Smooth Muscles of the Eye and the Nose (No. 5-8, 11-15) and
M. Levator Bulbi (No. 16) (nn. III, V)

In contrast to former researchers, Lakjer (1926, p. 26) distinguished two muscles ventral to the eye bulbus. On the one hand, he defined the striped fibred m. levator bulbi (No. 16) as a homologue of Constrictor I dorsalis. On the other hand, he defined an m. depressor palpebrae inferioris (No. 14-15) composed of two muscle portions in turtles, which would have been consequently misinterpreted in the literature. He assumed that the "new" – smooth muscle fibred – m. depressor palpebrae inferioris "potentially" is "a particular transformation of connective tissue fibres of the periorbita." Different authors (Olsson et al. 2001; Ericsson et al. 2004) have shown that the contribution of cNCC to the patterning of the muscles of the mandibular and hyoid arch is mainly restricted to the formation of connective tissue material. Nevertheless, it has also been shown cardiac NCC form smooth muscles in the aortic vessels within visceral arches (see Hall 1999, p. 98); moreover, they contribute to all blood vessels of the face and forebrain with muscular material (Etchevers et al. 2001). In the eye region, Yamashita and Sohal (1987) have shown that the dorsal parts of iris muscles in birds also originate from cNCC. Bruner (1901) studied the smooth muscles in the facial region of Lissamphibia and refered to connective tissue and surrounding mesenchym to form nasal and eyelid muscles. Based on those considerations and the possibility of a smooth muscle origin out of NCC, I follow the opinion of Lakjer (1926) in discussing the m. depressor palpebrae inferioris (No. 14-15) to be of mesoderm-independent, cNCC-derived origin. M. levator bulbi (No. 16), in contrast, originates from n. trigeminus (V) innervated (jaw) muscle Anlagen. Pars transversalis of m. depressor palpebrae inferioris (No. 14) seems to be a novel (cNCC-derived) structure within turtles. It functionally replaces the dorsal head of m. levator bulbi (No. 16) and gains innervation from a branch of n. trigeminus (V₂) that innervates muscular structures ventral to the eye (Lubosch 1938b) – but never parts of m. levator bulbi (No. 16) (Lakjer 1926).

Except for birds, the Pars equatorialis of m. depressor palpebrae inferioris (No. 15) is known in all reptiles studied so far (Lakjer 1926). In addition to this muscle portion, those groups show well-developed – non-smoothed – derivates of Constrictor I dorsalis. The coexistence of both structures and their differing histology clearly speak for non-related muscles.

The smooth muscle fibres of m. depressor palpebrae inferioris Pars equatorialis (No. 15) and the striped fibres of m. levator bulbi (No. 16, dorsal head) have both an anterior-posterior orientation. Additionally, the fact that in Dermochelys coriacea a dorsal head of levator bulbi still exists and an equatorialis portion of m. depressor palpebrae inferioris (No. 15) is missing (Lakjer 1926) speaks for the hypothesis of a functional but no material replacement.

The Anlage of m. levator bulbi (No. 16) is visible in turtle embryos (Edgeworth 1907) and shows a different degree of postembryonic regeneration (Versluys 1912, p. 599; Fuchs 1915; Lakjer 1926). It may be remained as connective tissue (Nick 1912, p. 125) or be completely reduced (Schumacher 1973) in adults. A well-developed m. levator bulbi (No. 16), originating from the skull roof and reaching ventrally to the eye with two heads, was observed in a subadult D. coriacea (Lakjer 1926). In adult D. coriacea specimens, m. levator bulbi (No. 16) would be only visible as a thin fascia (Edgeworth 1935: cited after Poglayen-Neuwall 1953-54). Jones et al. (unpublished work: pers. obs. IW) possibly discovered an m. levator bulbi originating from the skull in a subadult Caretta caretta. Also in an adult non marine-turtle the muscle was found, in the trionychid Apalone ferox (Lubosch 1933), having only one insertion, which would be homologous to the ventral head found in D. coriacea. In most turtle species, such as Terrapene carolina, m. levator bulbi (No. 16) is completely reduced in adults – possibly related to the reduced mobility of the palatal region in turtles – and both portions of m. depressor palpebrae inferioris (No. 14-15) are well developed (Lakjer 1926).

Referring to Ogushi's (1913b) work, Edgeworth (1935) considered the "m. depressor palpebrae inferioris" (No. 14-15) and the "m. arrector rostri" (No. 13) as a single structure to be synonymised with his "Smooth (unstriped) ocular muscles of Chelonia." One should take his note with caution, because Ogushi (1913b) neither described the nose muscles having smooth muscle fibres, nor did he describe any innervation pattern. In contrast, he cautions against a homologisation to smooth muscles of other tetrapods as described by Bruner (1901).

Winokur (1982) studied the nasal muscles of turtles more in detail and found in sum three smooth muscle masses. Surprisingly not only in the proboscis of Trionychia, but also in the "flat" noses of some pleurodires the author found smooth nasal muscles (chelids: Chelodina longicollis, Elseya latisternum; pelomedusid: Podocnemis unifilis). The m. nasalis Pars arrector rostri (No.13) was not found in all pleurodires and only in Podocnemis unifilis a Pars internarialis (No. 12) is present. I define the three nasal muscles as portions s. s., because Winokur (1982) highlights the connectivity of the portions by fibres crossing in the anterior region of the muscular units. In species, where parasaggital cartilages are present (not Chelodina longicollis), a complete separation of the Pars arrector rostri (No. 13) and the circumnarial fibres (No. 11) occurs. One may hypothesise the Pars circumnarialis (No. 11) to be the phylogenetic Mutterboden of the nasal muscular units. However, in that case a sistergroup relationship of Trionychia (Carettochelys insculpta and trionychids, sensu Gaffney and Meylan 1988) to all remaining Cryptodira must be assumed. In the latter, consequently nasal muscles would have been reduced. Nasal muscles seem to be related to nasal closure in species living in the water. Winokur (1982) noted that marine turtles, such as Caretta caretta, have erectile non-muscular tissue in the nasal region, which enables a nasal closure.

N. Trigeminus (V) Innervated Jaw Musculature (No. 17-30)

The mm. adductor mandibulae externus (No. 17-21), internus (No. 23-28), et posterior (No. 29, 30) are the most discussed muscles in chelonian literature, first because of the prominence and diversity of those muscles in turtles. Second, the unclear position of turtles within amniotes is particularly correlated to the anapsid skull condition, and several authors assumed especially the m. adductor mandibulae externus to have an important influence to the architecture of temporal bones in evolution (e.g., Fuchs 1909; Gregory and Adams 1915; Zdansky 1923-25; Kilias 1957; Frazetta 1968; Rieppel 2008) and development (Rieppel 1990). Recently, Tvaroková (2006) was able to show a prior influence of jaw muscle development to the formation of the emarginations of the skull, a process occurring relatively late in embryogenesis. However, jaw muscle development seems to be independent of the formation of temporal openings.

The differentiation of constrictor I lateralis musculature into three muscles represents a plesiomorphic condition that can be recognised in non-tetrapods, amphibians, and amniotes (e.g., Lubosch 1938b, 1938c; Diogo et al. 2008a, 2008b).

The Portions of M. Adductor Mandibulae Externus (No. 17-21)

A partition of m. adductor mandibulae externus (No. 17, 19, 21) into three major portions, as proposed by Lakjer (1926) (Figure 2), is in some cases not clearly identifiable. Also in Emydura subglobosa, I was not able, to clearly distinguish between those in superficial view. When compared to other sauropsid taxa (Lakjer 1926; Lubosch 1933, 1938b), the interrelationship and proportions of these three portions are completely rearranged in turtles. This peculiar anatomy of m. adductor mandibulae externus architecture of turtles lead Iordansky (1987, 1996) to the confidence that the turtle's m. adductor mandibulae externus should be categorised in a completely different way than in other sauropsid groups. He proposed a postorbital, superior, and inferior part of this muscle (Figure 4). Concerning the above-defined nomenclature of muscular units (Appendix 1), the proposal of Iordansky (1987, 1996) represents a more functional categorisation rather than an evolutionary identity of these muscular structures. Nevertheless, thanks to the thorough considerations of this author a fundamental re-evaluation of jaw musculature is inspired.

My impression from the here presented study is that jaw musculature – at least m. adductor mandibulae externus – in turtles possibly experienced a completely different evolution than in Sauria. When compared to Lissamphibia, neither the saurian- nor the turtle-like m. adductor mandibulae externus condition has a comparable shape. Including mammalian jaw muscle architecture, one has to define four morphotypes of constrictor primus lateralis homologues in tetrapods. The herein proposed scheme of plastic jaw muscle behaviour (Figure 16) may be most convenient to interpret the evolutionary history of this structure among land living vertebrates. Comparative ontogenetic studies are needed to postulate homologies herein. Possibly those studies may result in the sobering corollary, that only m. adductor mandibulae externus as a whole structure is homologous among the tetrapod taxa, and the plastic paraxial mesoderm reorganises separately within the different clades: Lissamphibia, Mammalia, Testudines, and Sauria. When referring to the peculiar Crocodylian jaw muscle arrangement (Poglayen-Neuwall 1953b; Iordansky 1964; Tvaroková 2006; Holliday and Witmer 2007) for example also complete reorganisations of the head mesoderm are imaginable within those taxa.

As the homology of jaw muscle portions among Testudines is demonstrated in the presented study (Appendix 1), I discuss particular structures in the following.

Except for Dermochelys coriacea (Burne 1905; Lakjer 1926; Schumacher 1972) and Chelus fimbriatus (Poglayen-Neuwall 1966; in which it can be completely reduced: compare Lemel et al. 2010 and Appendix 1) m. adductor mandibulae externus Pars profundus (No. 19) is the most prominent muscle portion of the jaw apparatus in turtles. It is the dorsomedial most muscular unit originating mainly from the parietal and the supraoccipital crest and mainly inserts to the coronar aponeurosis (Appendix 1).

In trionychids and kinosternids the coronar aponeurosis develops multiple tendinuous differentiation (Appendix 5) resulting in a variety of muscle fibre courses and muscle heads (e.g., Dalrymple 1975). Lakjer (1926) defined five different partitions in Amyda cartilaginea that are also visible in related species such as Pelodiscus sinensis (Poglayen-Neuwall 1953a; Iordansky 1987) and in Lissemys punctata (Schumacher 1954-55a, 1954-55b). Except for his "portion D", Lakjer (1926) described a major region of the "portions" to originate from the parietal/supraoccipital crest, partly by the supraoccipital aponeurosis (Appendix 5, Figure 16, Lakjer's: "Sehnenband c"). His "portion D" is described to span between two branches of the coronar aponeurosis and hence should be described as a particular portion s. s. in the presented study. However, in Lakjer's figures (Lakjer 1926, fig. 152-153) a clearly unseparated origin of "D" (at least to "C") is visible. Finally, I defined the m. adductor mandibulae externus profundus (No. 19) to be only one muscle portion s. s. in trionychids. One could define Lakjer's (1926) "portions" as muscle heads because they all show different insertions to parts of the coronar aponeurosis, but a shared origin. Further investigations on the detailed anatomy of fibre courses and eventually separated origins may verify or change the nomenclature of those structures proposed herein.

An additional portion - m. adductor mandibulae externus Pars profundus atypica (No. 20) - is only known for Mauremys caspica and Cuora amboinensis (Poglayen-Neuwall 1953a). When compared to other sauropsid groups, this muscle cannot be homologised to any other known muscular structure. Both species having a Pars profundus atypica (No. 20) are characterised by several highly derived characters within cryptodire turtles. Within those, they are not closely related to each other. Consequently, one has to discuss the Pars profundus atypica (No. 20) as a novel structure that independently evolved within those species. Poglayen-Neuwall (1953a) did not refer to the innervation pattern of that muscular unit, however due to its insertion to the coronar aponeurosis one can assume, that it ontogenetically developed from the adductor externus Anlage and phylogenetically it is a separation of Pars profundus principalis (No. 23).

The m. adductor mandibulae externus Pars medialis portions (No. 17-18) originate at the quadrate in turtles and insert laterally to the lower jaw and/or the coronar aponeurosis (Appendix 1). M. adductor mandibulae externus Pars superficialis (No. 21) generally originates broadly medially on the lateral wall of the adductor chamber. In Chelydra serpentina, Rieppel (1990) clearly defined two strong muscle heads in the origin of Pars superficialis (No. 21), the postorbital and the squamosal related head. The former generally occurs in taxa with a strong zygomatic arch or in taxa with complete dermal armour in the postorbital/temporal region (e.g., Chelonioidea, Platysternidae). In the species studied herein, Emydura subglobosa, having only a posterior bony bridge, such a muscle head is not present. However, some fibres are known to attach the skin spanning above the adductor chamber in several turtles species (also E. subglobosa). To which extent those can be homologised to that muscle head should be investigated in the future.

In several species, the Pars medialis (No. 17) as here defined (Appendix 1) is not clearly distinguishable from the Pars superficialis (No. 21), which resulted in conflicting interpretations of the anatomy observed. Poglayen-Neuwall (1953a) for example declared the fused Pars medialis/Pars superficialis (No. 17/21) s. s. to be his "P. media" (compare to Figure 2, Figure 3, Figure 4, Appendix 1) and only the postorbital region related muscle head of Pars superficialis (No. 21) to be his "P. superficialis" (e.g., in Mesoclemmys nasuta; similarly to Poglayen-Neuwall 1966: Chelus fimbriatus).

As discussed above, the Pars profundus (No. 19) of kinosternids and trionychids is separated into several muscle heads in the origin site. In Kinosternon scorpioides Poglayen-Neuwall (1953a) defined one of the muscle heads of Pars profundus (No. 19) to be his "P. media" due to the fusion of No. 17/21. In Pelodiscus sinensis, Poglayen-Neuwall (1953a) named m. zygomaticomandibularis (No. 22) as his "P. superficialis" and Pars superficialis (No. 21) s. s. as his "P. media". All other authors who studied P. sinensis specimens have been able, to clearly distinguish the Pars medialis (No. 17) from the other portions. The proposal of Poglayen-Neuwall (1953a) to declare m. zygomaticomandibularis (No. 22) to be his "P. superficialis" may have lead him – if no case of variation – to overlook the separation of a Pars medialis (No. 17) that usually originates from the quadrate in P. sinensis separately.

In Amyda cartilaginea, Lakjer (1926) separated the "medialis region" of m. adductor mandibulae externus into three parts. Following the here presented definition of portions and the criteria of homology as summarised in Appendix 1, I define Lakjer's (1926) "dorsalis-portion" ("II dors") as the m. adductor mandibulae externus superficialis (No. 21): It has its own origin on the squamosal, its own course, and a unique insertion on the lateral face of the coronal aponeurosis. As mentioned above, a comparative arrangement of muscle bundles is also visible or described in other trionychids, P. sinensis, or L. punctata, and in kinosternids (Poglayen-Neuwall 1953a; Schumacher 1954-55a, 1954-1955b; Iordansky 1987).

All these considerations on Pars medialis (No. 17) and Pars superficialis (No. 21) highlight first the plasticity of cranial musculature not only between major taxonomic groups, and second the necessity to follow a clear nomenclature of muscular structures (Appendix 1).

Due to the particular posterior extension of the lower jaw in trionychids a clear Pars medialis principalis (No. 17) is visible; moreover, in Amyda cartilaginea (Lakjer 1926) it separates another portion, Pars medialis inferior (No. 18), to also enable an insertion to the posterior most extension of the lower jaw.

The m. levator anguli oris of lepidosaurs (e.g., Abdala and Moro 2003) stands in close relationship to the Pars superficialis of this group and may be convergent to the rictal plate attaching fibres of the superficialis portion (No. 21) that occurs in some turtles (Werneburg 2010), given that both are homologous among sauropsids.

M. Zygomaticomandibularis (No. 22)

The m. zygomaticomandibularis (No. 22) is a particular muscular unit only existing in Carrettochelydae (pers. communication by Shigeru Kuratani) and Trionychidae. It has a convergent topology to the masseter muscle in mammals. However, it gains a different innervation and has to be declared as a "real adductor mandibulae muscle" (quote from Ogushi 1913b). Hence, most authors separate this zygomandibular structure as a real muscle s. s. apart from the three main portions (No. 17, 19, 21) of the adductor mandibulae externus complex (e.g., Schumacher 1973; Dalrymple 1975, 1977; Iordansky 1987). In contrast, Lakjer (1926) and Poglayen-Neuwall (1953a) described it as their "superficialis" (No. 21) in A. cartilaginea and P. sinensis based on its spatial relationships.

As mentioned above, Pars superficialis (No. 21) in several turtles exposes a postorbital region related muscle head that – in the origin face – is clearly separated from the squamosal-related muscle head of this portion. Together, both heads inseparabably fuse and form one muscle belly, which inserts as one structure to the lateral face of the coronar aponeurosis as well as to the lateral face of the lower jaw. As such, Pars superficialis (No. 21) of the species showing this anatomy is partially comparable to m. zygomaticomandibularis (No. 22), which also originates around the postorbital region and inserts laterally to the lower jaw.

Jones et al. (unpublished work) have found a particular lateral head of the Pars superficialis (No. 21) in Caretta caretta and Lepidochelys kempii that is almost completely separated from the rest of the muscle portion. Next to the insertion to the anterolateral face of the coronar aponeurosis and the coronar process, Pars superficialis (No. 21) broadly inserts laterally to the lower jaw, and reaches ventrad almost to the ventrolateral edge of the lower jaw in these species. For C. caretta, such a separated head was not described by any other author before, and the insertion to the lower jaw does not show such extent. Intraspecific variability and/or a different focus of the authors may be the reasons for this difference. Also for other marine turtles, Chelonia mydas and Eretmochelys imbricata, such a comparable anatomy was never described or depicted. In Dermochelys coriacea (Schumacher 1972), the Pars superficialis (No. 21) does not insert to the lateral face of the lower jaw at all. – Marine turtles show almost complete dermal armour of the temporal region (Kilias 1957); Trionychia show a strong zygomatic arch. The lateral head of Pars superficialis (No. 21) in Chelonioidiae and the m. zygomaticomandibularis (No. 22) of Trionychia originate from homologous bones and, to a different extent, insert to the lateral face of the lower jaw. Based on this comparable anatomy, I hypothesise the lateral head or at least lateral parts of the Pars superficialis (No. 21) of Chelonioidae – depending on the underlying phylogenetic topology (Werneburg 2010, figure 7.11) – to represent either a homologous or a convergent structure to the m. zygomandibularis (No. 22) in Trionychia.

Developmental observations as well as more detailed comparative anatomical studies – particularly on the innervation pattern of the lateral region of the m. adductor mandibulae externus complex – will help to test the here proposed hypothesis.

The Portions of M. Adductor Mandibulae Internus (No. 23-28)

The m. adductor mandibulae internus (No. 23-28) exposes a more comparable pattern among turtles (and tetrapods) than the external adductor muscular structures (No. 17-22) do. However, the identity of the 'pseudotemporalis', 'pterygoideus' and 'intramandibular' portions and their interrelationship to m. adductor mandibulae posterior (No. 29-30) were repeatedly discussed in the literature (Rieppel 1990, see below).

In several turtle species, a remarkable integration of m. adductor mandibulae posterior (No. 29-30) and m. adductor mandibulae internus Pars pseudotemporalis principalis (No. 23) may occur. This led Schumacher (1953-54, 1954-55a) and his student Hacker (1954) to name both units as "Pars caudalis" (No. 29) et "Pars rostralis" (No. 23) of the posterior adductor (Figure 2, Figure 3). After observing a broader range of taxa, Schumacher (1954-55b) separated Pars pseudotemporalis principalis (No. 23) as "m. adductor mandibulae anterior." Later, Schumacher (1972, 1973) went back to the nomenclature and partitions of jaw musculature as proposed by Lakjer (1926) (Figure 2). Here I also follow the approach of the latter author. I separate Pars pseudotemporalis principalis (No. 23) as a portion of m. adductor mandibulae internus (Appendix 1). As reported by Poglayen-Neuwall (1953a, 1953-54, 1966), the innervation pattern of Pars pseudotemporalis (No. 23-24) is completely different to that of m. adductor mandibulae posterior (No. 29-30). In addition, the developmental and evolutionary identities of both structures differ as discussed below.

The Pars pseudotemporalis principalis (No. 23) has an abnormal horizontal orientation in Chelus fimbriatus due to its flat skull. In regard to the following considerations it is worth mentioning that Poglayen-Neuwall (1966) homologised this deep pseudotemporalis structure ("profundus") of C. fimbriatus to that of Trachemys scripta (Poglayen-Neuwall 1953a), although it has a different origin – I follow his proposal to homologise this structure in C. fimbriatus and other turtle species (Appendix 1).

In several species, an m. adductor mandibulae internus Pars pseudotemporalis superficialis (No. 24) may be developed. In contrast to the present study, Poglayen-Neuwall (1966) does not homologise the Pars pseudotemporalis superficialis (No. 24) of T. scripta and of C. fimbriatus because in those species that portion inserts differently. While in T. scripta the portion inserts to the Zwischensehne (Appendix 5, Figure 16), which is connected to m. intramandibularis (No. 25), the Pars superficialis (No. 24) of C. fimbriatus inserts to the subarticular aponeurosis. I do not follow the nomenclature of the author, because – as shown by himself – the innervation of Pars pseudotemporalis superficialis (No. 24) shows the same pattern in both species. As will be shown for the Partes pterygoidei (No. 26-28) below, the insertions of m. adductor mandibulae internus portions (23-28) can easily separate from or refuse to the subarticular aponeurosis and e.g., forming a pterygoideus- or a posterior aponeurosis (Appendix 5, Figure 16)). The identity of the m. intramandibularis (No. 25) was previously discussed by Rieppel (1990) as well as by Iordansky (2008), and will also be discussed herein. The criterion of spatial orientation has to be preferred when discussing the identity of the Pars pseudotemporalis superficialis (No. 24). However, it is to be mentioned that a pseudotemporalis superficialis portion (No. 24) only occurs in a few turtle groups (Werneburg 2010: characters 110-113) that are not assumed to be closely related. Pars pseudotemporalis superficialis (No. 24) may have the same developmental origin in those taxa; however, it has to be declared as a convergent structure in the adult specimens of those groups.

The Pars pseudotemporalis (No. 23-24) structures in T. scripta provide a case study of the presented definition of muscular units (Poglayen-Neuwall 1953a; Iordansky 1987). Whereas T. scripta shows a common inseparable origin of Pars pseudotemporalis principalis (No. 23) and Pars pseudotemporalis superficialis (No. 24) – to be named as a portion No. 23/24 – C. fimbriatus shows two clearly separated portions s. s.

The m. adductor mandibulae internus Partes pterygoidei (No. 26-28) originate in the pterygoid region and insert directly, via an own (pterygoid) tendon and/or – together with the remaining m. adductor mandibulae internus portions – via the subarticular aponeurosis to the lower jaw (Appendix 1, Appendix 5). Schumacher (1973) mentioned a non-homologous arrangement of two Partes pterygoidei in Pleurodira and Cryptodira, a third portion would only occur in Podocnemis (Figure 4). After revising the original references and based on the above-defined terminology, I was able to define three homologous portions s. s. (Appendix 1) and the Pars pterygoideus posterior (No. 27) is present in several species. Partes pterygoidei dorsalis (No. 26) et ventralis (No. 28) are clearly separated in their attachment sides and by their fibre courses (Lakjer 1926: Amyda cartilaginea); however, in superficial view – also depending on the extend of the attachment of Pars pterygoideus ventralis (No. 28) – both may look like continuous structures (Lakjer 1926: Eretmochelys imbricata). Schumacher (1972) described two clearly separated portions in an adult specimen of Dermochelys coriacea, Lakjer (1926) did not find those in a young specimen. In the studied specimens of Emydura subglobosa, I recognised crucial heterotopic rearrangements in the Partes pterygoidei (No. 26-28). Intraspecific variability as well as ontogenetic differentiation must be taken into account when discussing those structures within a phylogenetic framework.

As mentioned above, m. adductor mandibulae internus Pars intramandibularis (No. 25) independently occurs in a few turtle species. Intramandibular, n. trigeminus (V) innervated muscles – attaching to the medial face of the lower jaw, mostly to cartilago meckeli – occur several times independently in vertebrate evolution (Albrecht 1876; Lubosch 1914; Rieppel 1990; Hertwig 2008; Iordansky 2008; Holliday and Witmer 2009; Werneburg 2009b). Iordansky (2008) functionally correlated their occurrence with strong dermal armour of the lower jaw in several reptiles. In Mammalia and Lissamphibia, either a less armoured lower jaw or a modified arrangement of jaw bones is visible, hence an intramandibular muscle would be absent in those groups. According to Rieppel (1990), Iordansky (1990) defined two non-homologous kinds of intramandibular muscles in sauropsids that evolved from different parts of the ancestral jaw adductor musculature. The "crocodiloidan (respectively crocodilian)" type is an intramandibular muscle associated to the m. adductor mandibulae posterior (crocodiles, some birds). In contrast, the "lacertil[i]oidan" (several lepidosaurs) muscle type is associated to the m. adductor mandibulae internus muscular units. In Testudines, an intramandibular muscle (No. 25) associated to the n. adductor mandibulae internus portions was discovered in the marine turtle C. caretta (several authors, see Appendix 1), in the emydids Chrysemys picta, Graptemys pseudogeographica, T. scripta (Poglayen-Neuwall 1953a), Trachemys terrapen (Iordansky 1996), as well as in the chelydrids Chelydra serpentina (Rieppel 1990: embryologically), Macrochelys temminckii, as well as in the taxon Platysternon megacephalum (Schumacher 1953-54, 1954-55b). The marine turtle Dermochelys coriacea does not have an intramandibular muscle although through of oversight differently mentioned by Iordansky (2008) (compare to Burne 1905; Poglayen-Neuwall 1953a, 1953-54; Schumacher 1972, 1973). Although turtles obviously show the "lacertil(i)oidan" type, Iordansky (2008) mentioned an intermedial condition of the intramandibular musculature in turtles. In Emys orbicularis, he refers to an intramandibular aponeurosis (Figure 16, Appendix 5), which does not bear any muscular material and is connected to the posterior aponeurosis, that serves as insertion tissue of the m. adductor mandibulae posterior (No. 29-30) muscle. This muscle is completely separated from other muscular units in that species – a hint for the m. adductor mandibular posterior (No. 29-30) related origin of m. intramandibularis (No. 25), at least in emydids. Moreover, Rieppel (1990) has shown the intramandibular muscle in lacertilians and turtles to develop differently. In C. serpentina, a close relationship of intermandibular (No. 31-33) and intramandibular (No. 25) muscle Anlagen exists in early development. Later on, the intermandibularis (No. 31-33) Anlage separates, and a clear relation of m. intramandibularis (No. 25) Anlage to the m. adductor mandibulae internus (No. 23-28) Anlage is recognisable. Rieppel (1990) correlated this close development with similar innervation patterns of mm. intra-/intermandibularis (No. 25, 30-31) by posterior branches of n. alveolaris trigemini (V) (see also Poglayen-Neuwall 1953b).

Developmental studies in Emys orbicularis may show, if there are any muscular Anlagen within the intramandibular aponeurosis (Figure 16) that are reduced during embryogenesis or if the intramandibular aponeurosis is only a separation of the posterior aponeurosis allowing the m. adductor mandibulae posterior a broader area of activity. While the condition of m. adductor mandibulae posterior (No. 29-30) muscle indicates a closer relationship of turtles to archosaurs, the developmental program of intramandibularis formation out of m. adductor mandibulae internus Anlagen (Rieppel 1990) – with a potential muscle formation in adults – suggested a closer relationship of turtles to lepidosaurs. As mentioned for m. adductor mandibulae externus (No. 17-21/22) musculature, the homology assumption of Lakjer (1926) and his followers is not entirely convincing. Particularly in crocodiles, the n. trigeminus (V) innervated jaw muscle portions more or less form a continuous muscle mass (e.g., Poglayen-Neuwall 1953b; Iordansky 1964; Schumacher 1973) – hence the topographical identity of m. intramandibularis may not necessarily reflect a different phylogenetic origin (Iordansky 1996: 'they are perhaps homologous'; Holliday and Witmer 2007).

In this context, the term 'm. intramandibularis' may be misleading, because it would imply a homology to m. intramandibularis (Aω) of teleost fishes (Werneburg and Hertwig 2009). In that clade m. intramandibularis (Aω) evolved from the m. adductor mandibulae externus/posterior complex (A2 incl. A2-PVM sensu Diogo et al. 2008a), whereas the intramandibularis portion of turtles possibly separated from m. adductor mandibulae internus (A3 sensu Diogo et al. 2008a) (Schumacher 1953-54, 1954, see above).

M. Adductor Mandibulae Posterior (No. 29-30)

I named the muscle originating at the ear capsule and inserting to the posteromedial aspect of the lower jaw as m. adductor mandibulae posterior (No. 29) (Appendix 1). In several turtles, this posterior part of adductor musculature inserts to the subarticular aponeurosis (e.g., Chelonia mydas, Chelus fimbriatus, Dogania subplana, Podocnemis). Consequently (s. s.) one should call it a Pars posterior of m. adductor mandibulae internus in those cases. In other cases (e.g., Chrysemys, Geochelone, Pelodiscus), the posterior muscle (No. 29) has a completely separated identity, exposing an own tendon (Figure 16.2: apo. pos) or/and exposing an own direct insertion to the lower jaw. In addition, intraspecific variability may occur such as in Emys orbicularis showing a separated or an integrated behaviour of those muscles (Hoffmann 1890; Poglayen-Neuwall 1953a). For Testudo horsfieldi, a contralateral variability of the course of V₂ was documented, which results in different conditions in the partition of the mandibular muscles (Iordansky 1987, 1990; see also Haas 2001).

Luther (1914) described a continuously increasing separation of externus, internus, and posterior adductor parts in the evolution of tetrapods – divers transitions are visible in several turtle species (Pelodiscus sinensis, Terrapene carolina, Testudo graeca). While Lepidosauria more likely show an integration of the posterior and the external muscle masses (Rieppel 1987) and their innervation patterns, Archosauria, some lepidosaurs (some lizards and chamaeleons), and turtles show an integration of the posterior and the interior muscle masses and their innervation patterns (Poglayen-Neuwall 1953a). In several turtle species (Chelus fimbriatus, Hydromedusa tectifera: Poglayen-Neuwall 1966), an integrated relationship of the posterior part to the internus part is recognisable. Within a phylogenetic framework, one may argue this condition to be a unifying character of a group consisting of Archosauria + Testudines. However, the particular patterns of innervation still have to be studied in several groups, and the ground pattern of lepidosaurian and turtle innervation has to be defined.

Following Diogo et al. (2008a), the mm. adductor mandibulae externus et posterior plesiomorphically form one single muscle within Osteichthyes, the A2. This close relationship is still recognisable in lepidosaurs and speaks for a derived condition uniting Archosauria and Testudines. Contrary to the hypothesis presented in Figure 16 – saying the posterior aponeurosis separates from the subarticular aponeurosis – Rieppel (1990) discussed the potential developmental origin of the pterygoideus/pseudotemporalis tendon from the posterior tendon. Moreover, Kesteven (1942-45) mentions Pars pterygoideus posterior (No. 27) of the internal adductor to possibly evolve from m. adductor mandibulae posterior (No. 29).

The close developmental relationship of m. adductor mandibulae internus Pars pseudotemporalis (No. 23) to the anterior head of m. adductor mandibulae posterior (No. 29) in Chelydra serpentina highlights the integration of those structures (see above; Rieppel 1990). However, Rieppel (1990) also mentioned that Lakjer's (1926) topological homologisation of this muscle head as a part of m. adductor mandibulae posterior would be correct – he adequately warns not to mix developmental plasticity and homology criteria of adult topology. The developmental variability of the structure was not studied by Rieppel (1990). It may explain the peculiar topographical condition of jaw musculature that the author observed (see Werneburg 2009b).

In a few species, an m. adductor mandibulae posterior Pars rostralis (No. 30) was found. Only more integrative studies can find a solution for the question, if this portion corresponds to the anterior head in the origin of m. adductor mandibulae posterior principalis (No. 29), which is found in several species (see Werneburg 2010).

Assuming the fluid pattern formation model as introduced above and reminding the examples presented in this chapter, I hypothetically assume the posterior Anlage (No. 29) (= m. adductor mandibulae Anlage) to represent the phylogenetic Mutterboden of both, m. adductor mandibulae externus (No. 17-22) et internus (No. 23-28) respectively (Figure 16; longest arrows). As a whole, the "pulsating" structure overridingly points either to the internal or to the external area in development. As evolution acts on organisms in the whole life span (Maier 1999), the differentiation of the three major jaw adductor muscles may be determined in this early time of development, running different paths of development and finally forming non-homologous muscle portions in adults.

Mm. Intermandibularis et. Submentalis (No. 31-33)

The identity of the intermandibular structures in tetrapods were continuously discussed in the literature (Appendix 1), partly reviewed by Rieppel (1990). In sum, the anterior part of the intermandibular muscular structures is innervated by n. trigeminus (V) and is called m. intermandibularis (No. 31-32), an internal m. submentalis (No. 33) can occur (Appendix 1). The posterior part is innervated by n. facialis (VII). It is called m. constrictor colli Pars intermandibularis (No. 42) in this study (see below for details). Both can be very continuous – forming one muscle s. s. – or a large gap (posterior trigonum) may be present separating m. intermandibularis (No. 31-32) as a whole muscle (compare to Werneburg 2010: characters 178-182). As mentioned above, the most separated condition of muscular structures is defined in Appendix 1. However, in a comparative description, the integration of No. 31/42 should be reflected in the name of respective structures.

Sondhi (1958) depicts one remarkable muscle portion in Asperideretes leithii (Trionychia). This m. intermandibularis Pars profundus (No. 32) lies dorsally to m. intermandibularis Pars principalis (No. 31). Pars profundus and Pars principialis are innervated by two branches splitting from one branch that originates from n. alveolaris (V).

N. Facialis (VII) Innervated Musculature (No. 39-46)

Ruge (1896) developed a detailed system for the evolutionary differentiation of n. facialis (VII) innervated musculature in vertebrates. In its plesiomorphic condition (Gnathostomata), the constrictor of the second pharyngeal arch (Constrictor secundus: C₂) has a dorsoventral (dv) orientation of muscle fibres (C₂dv). In evolution, this structure serves as Mutterboden for several n. facialis innervated muscles (Lubosch 1933). Anterodorsally it separates a mandibular part (C₂md). In turtles, three muscles corresponing to this C₂md are known: m. depressor mandibulae (No. 45), m. dilatator tubae (No. 46), and the m. cervicomandibularis (No. 39). The latter was only described for Apalone ferox [Lubosch 1933: note, although redrawn by Schumacher (1973) himself, he did state the muscle to be missing in turtles]. Lubosch (1933) referred to Gräper (1932) and Ogushi (1913b, by mistake he wrote "Osawa"), who would also have found this muscle in A. ferox and Pelodiscus sinensis. However, Ogushi (1913b) only referred to the "cervico-hyo-capitis" muscle complex that he mentioned to be a fusion of different homologous muscular units of other species (see below). This muscle does not attach to the lower jaw as the m. cervicomandibularis (No. 39) does (Lubosch 1933) and gains a different innervation. In addition, Gräper (1932) only mentioned an m. cervicocapitis [partly homologous to Ogushi's (1913b) muscle], that among others, attaches the otic region in A. ferox and not to the lower jaw as described by Lubosch (1933). In P. sinensis, the muscle is innervated by n. hypoglossus (XII). Developmental studies may observe if the n. facialis (VII) innervated m. cervicomandibularis (No. 39) of Apalone ferox (of Lubosch 1933) is incorporated into more medial neck muscles during ontogeny. In that case the muscle would have changed its innervation as well as its insertion pattern – or if m. cervicomandibularis (No. 39) develops the other way around.

M. dilatator tubae (No. 46), which attaches to the eustachian tube, is strongly associated to m. depressor mandibulae (No. 45) by connective tissue, hence Ogushi (1913b) decided to name it the internal depressor mandibulae muscle (Appendix 1). However, no overlapping fibres were ever described to declare both as portions of one muscle. Commonly described to be innervated by n. facialis (VII) (e.g., Ogushi 1913b; Schumacher 1973), McDowell (1963) second-guessed his former observations of 1961 and assumed an n. glossopharyngeus (n. IX) innervation of m. dilatator tubae (No. 46), a finding that would relate the muscle to cornu branchial-I (3rd pharyngeal arch) related musculature. The author focused on cranial arteries mainly and comparative muscle descriptions are absent in his studies that would more strongly justify his findings. In fact, he mainly based his observations on Podocnemis expansa, which has a unique orientation of cornu branchial-I distally ending between m. dilatator tubae (No. 46) and m. depressor mandibulae (No. 45). In this context the n. glossopharyngeus (n. IX) innervated m. branchiomandibularis visceralis (No. 47) is to be mentioned (see also below), which itself originates on the distal tip of cornu branchial-I. McDowell (1963) – who did not mention this commonly present muscle – could have intermixed the muscle masses and consequently their innervation pattern in the lower otic region. Alternatively, the strong integration of cornu branchial-I with the m. branchiomandibularis visceralis (No. 47) bearing n. glossopharyngeus (n. IX) in Podocnemis expansa may have resulted in a partial shift of some branches of the IXth cranial nerve. Serial section series would be able to detect the actual innervation pattern in this species. In Caretta caretta, Jones et al. (unpublished work) found a quiet short m. branchiomandibularis visceralis (No. 47) that runs parallel to the ventral face of m. dilatator tubae (No. 46). Both muscles attach the lower jaw in that species; however, no intercrossing fibres or nerves were detected.

Jones et al. (unpublished work) found the m. depressor mandibulae (No. 45) to include a complex tendinuous framework in Caretta caretta that almost separates it in two singular muscle portions. However, along the whole length of the muscle the two parts share a continuous muscle fibre distribution on the surface as well as in the insertion/origin areas. That defines both parts to form one muscle s. s. and not to form muscle portions. As a case of intraspecific variability, one may expect specimens, in which both parts are separated from each other and form two muscle portions. In that case a new muscle portion has to be added to the list (Appendix 1, Figure 6, Figure 7, Figure 8, Figure 9, Figure 10, Figure 11, Figure 12, Figure 13, Figure 14, and Figure 15,) as No. 89 after the last number in the list or – to be preferred – as a reference to the donor m. depressor mandibular (No. 45) as No. 45-1 (note: not as 45a, which would indicate a muscle head following the definition of this paper).

Anteroventrad the Mutterboden muscle C₂dv separates an intermandibular part in turtles, herein defined as m. constrictor colli Pars intermandibularis (No. 42). As discussed above, the dual ancestry of intermandibular muscles has been demonstrated by several authors (e.g., Poglayen-Neuwall 1953a), whereas the anterior, n. trigeminus (V) innervated part is Constrictor primus ventralis derived (m. intermandibularis, No. 31), the posterior part (No. 42) is n. facialis (VII) innervated. In my opinion, Lubosch (1933) misidentified all intermandibular muscle(s) to be of Constrictor secundus origin due to his focus on facialis innervated musculature. In Chelodina longicollis, the m. constrictor colli Pars intermandibularis (No. 42) may extend its general origin from the posteromedial aspect of the lower jaw to cornu branchial-I (Kesteven 1942-45).

C₂dv itself can be separated in two portions in turtles, the m. constrictor colli Pars oralis (No. 43) et Pars aboralis (No. 40). Whereas the former is attached to the cranium, the latter originates in a median raphe on the top of the neck/dorsal tip of the anterior cervical vertebrae. Pars aboralis (No. 43) may extend caudally and, fused with the posterior neck constrictor – m. sphincter corticis (Ogushi 1913b: his No. 53; synonym: m. sphincter colli posterior sensu Fürbringer 1874 and Lubosch 1933) – it may cover the whole neck. Due to the diversity of m. constrictor colli (Lubosch 1933) and different degrees of fusion with other neck muscles in turtles a high confusion exists about the identity and synonymy of its muscular units (e.g., Schumacher 1973, p.163-166).

Mediad the C₂dv may separate two originally hyoid apparatus associated muscular parts in vertebrates, a dorsal C₂hd and a ventral C₂hv. The former only occurs in a very few cases within tetrapods and Lubosch (1933) mentions an m. stylohyoideus in birds and mammals and the m. strapedius in crocodiles to eventually be derivates of this dorsal hyoid constrictor (C₂hd). The n. XII innervated m. collosquamosus (No. 57) or the n. IX innervated m. branchiomandibularis visceralis (No. 47) – who may topographically correspond to the C₂hd – would have shifted their origins or insertions and gained a different innervation if they were descendents of C₂hd. However, based on current knowledge, it is most parsimonious to simply define the C₂hd to be absent in turtles. Ontogenetic studies as well as detailed anatomical observations in adults may detect the destiny of this structure.

The ventral C₂hv may be found in turtles. It is represented by a muscle portion of m. constrictor colli for which I introduce the name m. constrictor colli Pars spinalis (No. 41). It originates from the lateral aspect of the anterior cervical vertebrae, is present in several species, and seems to replace the m. constrictor colli Pars aboralis (No. 40) in most cases. Whereas the former originates laterally from the cervical vertebrae and lies ventrally to several epaxial muscles (such as No. 80, 81, 82), the latter originates dorsally to the anterior cervical vertebrae in a medial raphe or dorsally on the neural arches and lies superficially to all other muscles of the neck. Both portions (No. 40, 41) insert ventrally in a median raphe with the contralateral muscle. Surprisingly No. 40 (C₂dv Partim) and No. 41 (C₂hv) never definitely occur at the same time. This could indicate a direct homology of both portions – in that case a radical shift of the origin site would have happened quite often in turtle evolution, perhaps enabled by extreme fluid developmental process of the constrictor colli Anlage (see discussion on Pars superficialis, No. 44, below). Nevertheless, after re-checking the very comprehensive study of Lubosch (1933) – who compared dissections of a broad range of sauropsid species and brought them in context with a large literature review on vertebrate taxa (e.g., Ruge 1896) – I follow his global interpretation where he discussed the condition seen in Apalone ferox (Lubosch 1933, figure 17). Here he found constrictor colli Pars oralis (No. 40) and Pars aboralis (No. 40), both of C₂dv origin, to be fused. Next to them, C₂hv (Pars spinalis, No. 41) is present, which valuates it to be an additional portion of m. constrictor colli. Again, only developmental studies can finally evaluate this opinion. – As trionychids are for Cryptodira, Chelodina longicollis, is characterised by several derived features within Pleurodira (as known so far). This may be possibly related to the high mobility and extention of necks in both taxa. Pars spinalis (No. 41) is commonly described to gain n. VII innervation (Gräper 1933). However, as an intraspecific variation of C. longicollis Kesteven (1942-45) described the portion to be innervated by spinal nerves nn. S1 et S2. One could argue the m. sphincter corticis of Ogushi (1913b: his No. 53) – the posterior constrictor of the neck, that is often indistinguishable fused with the n. facialis i(VII) nnervated m. constrictor colli (No. 40-44) – to have shifted rostrad and to suppress the facialis constrictor. Kesteven (1942-45) continued his description and first described three distinct heads in the origin of the Pars spinalis (No. 41), a character of this portion also found in other turtle species (e.g., Dermochelys coriacea, Schumacher 1972). Second, Kesteven (1942-45) described a different innervation pattern of the more posterior neck constrictor (m. sphincter corticis) by the ventral branch of spinal nerve n. S2. Ogushi (1913b) particularly mentioned a ventral spinal nerve innervation of m. sphincter corticis in turtles – but in Pelodiscus sinensis, it would come from spinal nerve n. S7. Hence, it is perhaps more reasonable, that Pars spinalis (No. 41) may have extended caudally as a whole in C. longicollis, suppressed m. sphincter corticis, and gained a new innervation. However, based on that differing innervation patterns, one cannot finally state the homology of this muscular unit in C. longicollis. I synonymised the muscle portion described by Kesteven (1942-45) to Pars spinalis (No. 41) due to similar origin and insertion when compared to other turtle species.

As demonstrated in the trionychid Aspideretes leithii (Sondhi 1958), the Mutterboden muscle C₂dv may also separate a lateral muscular unit, m. constrictor colli Pars superficialis (No. 44): a potential of this muscular Mutterboden that was never described in literature before. As its potential phylogenetical donor muscle (C₂dv), it gains innervation by n. VII (Sondhi 1958), inserts to the median raphe at the bottom of the neck, and is situated superficially to all other neck muscles. Surprisingly, the course of muscle fibres (= horizontally) is rotated by 90° to the fibre course of the remaining (common) constrictor colli portions (= vertically), that resembles an enormous positional change in evolution. Ontogenetically non-terminated myoblasts (No. 44 Anlage) that separate from the myoblast of the donor muscle Anlage (C₂dv) may have "simply" changed the program of their alignment, and the accumulated fibres may have consequently changed their orientation in the adult. An alternative, less parsimonious scenario is imaginable: It would be worth checking, if the constrictor colli Pars superficialis (No. 44) ontogenetically originates from the n. XII innervated coracohyoideus (No. 58) muscle Anlage. It also lies ventrally in the neck, has an anterior-posterior orientation of muscle fibres, and serves as a Mutterboden for several other muscles ventrally in the neck (see below). The ontogenetic muscle precursors of m. constrictor colli Pars superficialis (No. 44) may have been shifted from the coracohyoideus Anlage to a superficial position and may have been separated from its donor after the dorsoventrad expansion of constrictor colli Anlage (C₂dv). Here it may have gained the innervation of n. VII by positional reasons. In that case, the muscular portion should be named "coracohyoideus Pars superficialis."

N. Glossopharyngeus (IX) Innervated Musculature

Amniotes show a reduced number of splanchnocranial elements, and only two muscles are known in turtles that are exclusively innervated by n. glossopharyngeus (n. IX), which plesiomorphically is associated to the third pharyngeal arch. The partly very strong m. branchiomandibularis visceralis (No. 47) arises from the distal end of cornu branchial-I and it inserts medially to the posterior region of the lower jaw. Gräper (1932, 1933) considered a breathing correlated function while the lower jaw is closed. However, in several species, the cornu branchial-I is distally deeply imbedded between the neck muscles, and it is connected by connective tissue (Gräper 1932) or ligaments (Jones unpublished work: pers. obs. IW) to the occipital/squamosal region. A feeding mechanism does not include only snapping in turtles but there is some evidence for an anteroposterior movement (M. Jones, personal commun., 2009). The influence of this muscle should be tested in biomechanical analyses. The developmental origin of the muscle (No. 47) is not entirely clear. Instead of arising from the posterior most n. glossopharyngeus (IX) innervated part of the paraxial mesoderm, it may also develop from n. hypoglossus (XII) innervated myotom. A very strong connection/integration of m. branchiomandibularis visceralis (No. 47) to m. hyoglossus (No. 67) (Walter 1887) or also to m. genioglossus (No. 63) (von Bayern 1884, p. 63) was reported for some species (e.g., Emys orbicularis).

The second n. glossopharyngeus (IX) innervated muscle was only described for Pelodiscus sinensis based on serial section evaluations (Ogushi 1913b): m. tensor s. dilatator vaginae venae nasoophtalmicae (No. 48). The author argues that the muscle would pull the respective venae to the skull to avoid a constriction of the vessel, whilst the head/neck is retracted in the cryptodire style. Comparative studies should show if the muscle (No. 48) autapomorphically occurs in P. sinensis or soft-shelled turtles, or if the muscle is unique to all hidden-necked turtles. In the pleurodire turtle Emydura subglobosa, I was not able to identify such a muscle in serial sections, which would support Ogushi's (1913b) functional assumption for now.

Nn. Vagus et Accessorius (X, XI) Innervated Musculature

The laryngeal muscles (No. 49, 50) show a relatively conserved count and shape among Testudines. Particularly the complete separation of cartilago cricoidea or its integration to cartilago thyreoidea (together as cartilago cricothyroidea: Schumacher 1972, 1973) has an influence to the insertion site of m. dilatator laryngis (No. 51). Only in Caretta caretta an additional muscle portion was found (No. 50) (Alessandrini 1834b), that apparently separated medially from m. constrictor laryngis principalis (No. 49).

In constrast, m. plastrocapitis (No. 52) develops a high degree of variability among turtle taxa and the muscle gained particular interest by some classic authors due to its phylogenetic relationship to sternocleidomastoideus muscle of Mammalia (e.g., Meckel 1828; Cuvier 1835; Fürbringer 1874). Nevertheless, it was often not recognised, because a confusing diversity of plastrocapitis (No. 52) evolved within turtle. This fact is realisable by the differing innervation pattern of the muscle among species. While it is innervated by n. accessorius (XI) in most species (sensu Ogushi 1913b; Shiino 1913), it may also be supported by branches of n. vagus (X) (Scanlon 1982: Chrysemys picta), and/or by the rami ventralis of the 3^rd to the 4^th spinal nerve (Fürbringer 1874; Ogushi 1913b: Pelodiscus sinensis, Schumacher 1972: Dermochelys coriacea). Finally, also the n. hypoglossus (XII) was described to contribute to the innervation of m. plastrosquamosus (No. 52) (Gräper 1932; Kesteven 1942-45).

Schumacher (1972) argued for the loss of m. plastrocapitis (No. 52) in Chelodina longicollis. He apparently based his knowledge on Gräper's study (1932); however, that author also discussed the muscle to be possibly integrated to other strong developed muscles (Gräper 1932, p. 185, his No. 4) of the neck region. In C. longicollis he discovered (page 186, his No. 10 N.N.) a new muscular structure spanning between the distal third of cornu branchial-I and the mastoid area (squamosal/quadrate). I call it the m. squamosobranchial (No. 53). Kesteven (1942-45, p. 264) found a comparable muscle to be present in the same species (his "cornu-hyoideus-capitis"). He also described an m. plastrocapitis (No. 52, his "sterno-thyroid"), which apparently shifted its origin from the plastron to the ventral area of the neck – possibly due to the exceptionally elongated neck of C. longicollis. In addition, the insertion seems to be replaced from the skull ventrad to the dorsal area of the hyoid apparatus – fibre attachments at the distal area of cornu brachiale-I and a few fibres at cornu branchial-II are reported. Although drawn in connection with plastrocapitis (No. 52), Ashley (1962) described his "sternomastoideus" to only connect one cornu branchial (by a tendon) with the base of the skull. Kesteven's (1942-45) and Ashley's (1962) observations show a strong relationship between both muscles (No. 52, 53). Schumacher (1972) and many others reported a tendinuous connection of m. plastrocapitis to cornu branchial-I in addition to its insertion to the mastoid area. One may finally argue for a phylogenetic origin of m. squamosobranchiale (No. 53) from the anterior part of m. plastrocapitis (No. 52).

Recently, Jones et al. (unpublished work) could show a so far unknown insertion of a thin m. plastrocapitis (No. 52) to the lateral face of the atlas in C. caretta. In this species m. squamosobranchial (No. 53) is well developed. In Lepidochelys kempii, the authors did not find an m. squamosobranchial (No. 53); however, a thin tendon connecting cornu branchial-I and the squamosal may be homologous to the structure. Jones et al. (unpublished work) also did not find a separated m. plastrocapitis (No. 52) in L. kempii, but in this species, the muscle might be laterally fused with the superficially situated m. constrictor colli (No. 41-43). With this finding, one may perhaps argue for an origin of m. plastrocapitis (No. 52) from n. facialis (VII) innervated muscle Anlagen in turtles – that would strongly contradict its homology to the sternocleidomastoideus muscle of Mammalia. But Gräper (1933), who missed m. plastrocapitis (No. 52) in Chelonia mydas as a separated structure argues for an integration of the muscle into the n. hypoglossus (XII) innervated m. coracohyoideus (No. 58-60), which itself would be very complex in this species. In contrast, Hoffmann (1890) found the muscle to be completely separated in the same species – a case of intraspecific variability.

In sum, m. plastrocapitis (No. 52) is highly variable at all. It is reduced in size, replaced to other insertion sites, or integrated to other muscular structures of the neck. For this retractor muscle functional constrains – reduced retractibility (marine turtles), high mobility of long necks (C. longicollis), or also the distinguishing retraction mode of Cryptodira and Pleurodira – should be taken in account. Up to this date, m. plastrocapitis (No. 52) and m. squamosobranchiale (No. 53) are only described in detail for a few species, and one needs to check the correlations proposed above. Gräper (1932, p. 191) correlates the absence of n. accessorius (XI) with the very thin and reduced shape of m. plastrocapitis (No. 52) in trionychids (his Apalone ferox and Ogushi's 1913b: P. sinensis). Consequently, he later argues (Gräper 1933, p. 277) for a replacement of m. plastrocapitis (No. 32) by the colloplastralis muscle of Ogushi (1913b: his No. 32), spanning between the plastron and the neck. However, in P. sinensis both muscles (Ogushi 1913b: his No. 32, 33), m. colloplastralis and m. plastrocapitis (No. 52), are present as complete muscles (par definition: Appendix 1) at the same time, and one should not declare a replacement of both, neither functionally nor concerning homology.

N. Hypoglossus (XII) Innervated Musculature

Sewertzoff (1929) discussed the developmental origin of several muscles in the hypoglossal region; Gräper (1932, 1933) very comprehensively demonstrated the immense interspecific diversity of muscular integration in tongue related muscles. Especially m. genioglossus (No. 63), which resembles the main developmental Mutterboden in this area, is partly connected to other muscular units (No. 64, 65, 69, 70, 72, 73, 74) in several species. As mentioned above, I named muscular units based on the most separated state of a muscular structure known from literature. However, particularly in the tongue region one could easily argue that a structure is only a muscle head of m. genioglossus (No. 63) s. s. when observing a few species. However, problems of homology will arise when making comparisons to species where this muscle "head" is separated as a muscle or a muscle portion.

One example for the consideration of synonymy, as it was accomplished for each cranial muscular unit separately (Appendix 1), should be demonstrated herein: One muscular structure (No. 65) was described by George and Shah (1955a) as belonging to m. genioglossus (No. 63) in Lissemys punctata, because as this muscle it originates from the symphysis of the dentaries. However, such as m. geniohyoideus principalis (No. 64), this muscular structure (No. 65) inserts to the cornu branchial-I. After rechecking the drawings, I realised – contra the description of the authors – the structure mainly originates from the medial aspect of the dentary (as No. 64), and only one part of the muscle has an anterior origin at the symphysis of the dentaries. In this example, intraspecific (or ontogenetic) variability finally shed light on the identity of the structure. In the Lissemys punctata specimen studied by Gnanamuthu (1937), m. geniohyoideus Pars lateralis (No. 65) is still connected to its Mutterboden m. geniohyoideus principalis (No. 64), while George and Shah (1955a) dissected a specimen of the same species in which those muscular units are completely separated. George and Shah (1955a) did not describe intercrossing fibres between their "Genio-glossus externus et internus" (No. 63 and 65), which would instead define them as portions of the same muscle. In their specimen, two separated muscles were described. George and Shah (1955a) and Schumacher (1973) mentioned the m. genioglossus (No. 63) to originate from the symphysis as well as from the medial side of the anterior third of the dentaries. Hence, it would be a correct synonymisation to call the "genioglossus externus" of George and Shah (1955a) the only homologous structure of m. genioglossus (No. 63) in the present study (No. 63). George and Shah (1955a) may have confused both muscles (No. 63 and 65) due to their strikingly similar orientation in the tongue and potentially similar functions – protraction and elevation of hyoid. Finally, I follow Gnanamuthu (1937) in naming the muscular structure mentioned as m. geniohyoideus Pars lateralis (No. 65), and hence to be in closer relationship to m. geniohyoideus Pars principalis (No. 64). Moreover, Kesteven (1942-45) described a structure in Chelodina longicollis, which should be named as m. geniohyoideus Pars lateralis (No. 65) par definition (Appendix 1). Apparently, the Pars principalis (No. 64) is lost in this species, probably due to changing anatomical conditions correlated to feeding behaviour. The "flexibility" of m. geniohyoideus (No. 64) can also be recognised by its innervation; while generally innervated by n. hypoglossus (XII) (Schumacher 1972), Shiino (1913) reported an additional support by n. glossopharyngeus (IX) in Trachemys decussata.

M. coracohyoideus (No. 58,) spanning between the shoulder girdle and the hyoid apparatus, possibly resembles the ontogenetic and phylogenetic Mutterboden for several muscular units at the base of the neck (No. 59, 60, 61, 62, 66, 71). This is indicated by fibres that still intercross in some species (e.g., No. 60, No. 71; Gräper 1932: Chelydra serpentina and Chelodina longicollis) or a very similar complete course of those muscular units, which are unique in particular species (e.g., 61, 62 or 59: Gräper 1932). The potential origin of m. constrictor colli Pars superficialis (No. 44) out of m. coracohyoideus (No. 58) was already mentioned above. Myotomic material of (perhaps interspecifically) different myotom origin may be the ontogenetic precursor of m. coracohyoideus (No. 58). Myotomic material (perhaps interspecifically) originating from different myotomes may be the ontogenetic precursor of m. coracohyoideus (No. 58). This conclusion is supported by the muscle's innervation pattern. In addition to n. hypoglossus (XII) (Scanlon 1982), the muscle is supported by the rami ventralis of spinal nerves 2-3 in its anterior and of spinal nerves 4-5 in its posterior region (Gräper 1932; Schumacher 1972). In addition, the high anteroposterior extend of the muscle speaks for this multimyogen assumption, a condition comparable to the longest muscles of turtles, the m. retrahens capiti collique (including No. 88), which is also innervated by spinal nerves of the whole vertebral column (e.g., Ogushi 1913b; Scanlon 1982; Callister and Peterson 1992). The extension of only one myotom of the posterior head region (n. XII innervated) over the whole neck or body in early development is hardly conceivable.

The fluid character of cranial musculature as demonstrated for jaw muscles above is also exemplified in the "cornu hyoideus muscle," unique in Aspideretes leithii (Sondhi 1958). The structure originates from cornu branchial-II and runs to cornu branchial-I and corpus hyoidei. By the definition of origin and insertion (Appendix 1), m. coracohyoideus Pars interbranchialis (No. 60) and m. branchiohyoideus (No. 55) must have been fused in this species while loosing their origin/insertion on cornu branchial-I. In comparative studies, "cornu hyoideus" should be labelled as No. 55/60 following the above-mentioned proposal.

M. collosquamosus (No. 57) and m. atlantoexoccipitis (No. 54), two muscles arising from the cervical vertebrae, were described by Ogushi (1913b) as being innervated by n. hypoglossus (XII). However, an additional n. accessorius (XI) support was reported for the former, and a singular spinal nerve (n. S2) support was described for the latter in Chryssemys picta (Scanlon 1982). Ogushi (1913b) mentioned an autapomorphic loss of n. accessorius (XI) for Pelodiscus sinensis; hence, the double innervation of m. collosquamosus (No. 57) seems to be the more common situation. Myogenic material of n. accessorius (XI) innervated paraxial mesoderm as well as n. vagus (X) innervated myotomes is reduced to a large extent in tetrapods due to the loss of gill structures (Maier et al. 2004). For collosquamosus (No. 57), an integration of n. XI innervated muscular material to n. hypoglossus (XII) innervated material is still possible. However, n. accessorius (XI) may have changed the muscle to innervate in most species or it simply was reduced.

The innervation pattern of m. atlantoexoccipitis (No. 54) was only described in the above-mentioned studies (n. XII or n.S2), and no consensus existed regarding evolutionary identity of this muscle. M. atlantoexoccipitis (No. 54) is innervated by ramus epistropheo-squamosi of n. hypoglossus (XII) in Pelodiscus sinensis (Ogushi 1913b). It originates from the ventrolateral aspect of cervical vertebrae-1 and inserts to the exoccipital. For Chrysemys picta, Scanlon (1982) described a muscular structure, the "m. collo-capitis longus," which is innervated by n. S2. It originates with a thin tendon from the ventrolateral surface of m. collo-squamosus (No. 57) and it inserts to the exoccipital. Due to its tendinuous connection to this muscle (No. 57), the structure (No. 54) would be a portion of collosquamosus (No. 57) by definition, but as mentioned above, the most common condition of the muscle known in turtle literature should be mentioned when naming a structure. Scanlon (1982) homologised this structure with the cranium-associated part of m. longus colli (No. 86/87) of other authors, which is situated ventrally in the neck. However, he additionally described an m. longus colli Pars capitis (No. 87) to be present, which would be separated in two "portions" – his "longus colli of CV-2 / -3". He did not synonymise these two "portions" with muscles described by other authors. I homologise Scanlon's (1982) "m. collo-capitis longus" to m. atlantoexoccipitis (No. 54) due to the synopsis of following aspects: 1. a close relationship of atlantoexoccipitis (No. 54) and m. collosquamosus (No. 57) by a shared n. hypoglossum (XII) innervation in Pelodiscus sinensis (Ogushi 1913b). 2. the connection of Scanlon's (1982) "m. collo-capitis longus" to m. collosquamosus (No. 57). 3. a similar course, as well as 4. the same insertion of both structures to the exoccipital. Consequently, m. atlantoexoccipitis (No. 54) in C. picta must have changed both its origin site and its innervation pattern, possibly subsequent to a slightly changed spatial orientation.

Neck Musculature (n. Cd/v)

Hypaxial and epaxial muscle layers – that are still distinct in fish – show a complicated morphology in tetrapods due to adaptations to land locomotion and holding the head in the air (Steiner 1977; Liem et al. 2001; Maier et al. 2004). Especially in turtles and birds the musculature of the very mobile neck underwent a large number of modifications (Nishi 1916, 1938a; Gasc 1981) resulting in several fusions and partitions. The homologisation of muscle regions seems to be possible only based on innervation patterns. In addition, innervation patterns may change during evolution (Haas 2001, 2003a, 2003b), especially in such a variable region as the neck. Hence, the most valuable source of homologisation would be comparative developmental studies of the neck region, but so far, those are missing for turtles. While discussing n. facialis (VII) innervated neck constrictors (No. 40-44), Lubosch (1933, p. 589) introduced the term myobiotaxis that adumbrates the ability of muscles to partly fuse, differentiate, re-fuse, etc. Keeping this in mind, the homologisations of the cranium-associated neck muscular units presented here (No. 54, 57, 75-88) are of a very premature nature in terms of research. Considering neck musculature, the lotus approach as presented for the adductor mandibulae above (Figure 16) reaches a new level of relevance.

In this regard, one example was extensively discussed by Ogushi (1913b) in which he defines a peculiar "m. cervico-hyo-capitis" (his No. 35; in this study at least parts of No. 71, 75, 78, 79: see below) in the trionychid turtle Pelodiscus sinensis. He described this muscle to originate from cervical vertebrae 3-6. The muscle belly forms four muscle heads (Ogushi 1913b: "caudae") anteriorly, attaching to the squamosal, the occipital region, the cornu branchial-II and finally the ventral aspect of corpus hyoidei. Ogushi (1913b) argued that this muscle does not find any corresponding homologue as a whole structure in P. sinensis, but most muscle fibres may be homologous to m. cervicocapitis (No. 79). Ventrally, the apparently epaxial muscle of Ogushi (1913b) could have been fused with hypaxonic musculature forming the anterior part of the muscle head which inserts to corpus hyoidei ("venter anterior" =anterior to the inscription of "cauda hyoidea" sensu Ogushi 1913b: No. 35a). However, I assume this muscle head to originate phylogenetically from the dorsal aspect of m. levator pharyngis (No. 71). The muscle head originates from cornu branchial-I and is innervated by ramus levator pharyngis branches of the n. hypoglossus (XII). The persistence of the innervation in addition to a somehow "major" shift of the muscle mass additionally supports this assumption. For another trionychid turtle, Apalone ferox, Gräper (1932, p. 190) described the m. levator pharyngis (No. 71) as still connected to m. coracohyoideus Pars principalis (No. 58), which highlights the "flexibility" of this structure among genera.

Ogushi (1913b) also argued that the m. atlantoopisthoticus (No. 78, his No. 47) and m. atlantoepistropheooccipitis (No. 75, his No. 46) are derived from the anterior heads of "m. cervico-hyo-capitis" namely from capita squamosi et occipitalis. Both muscles (No. 75, 78) exist next to the proposed donor muscle and are not connected to it, they may at most be partly homologous to those muscle heads. Hence, I list them as unrelated muscular units (no portions of one muscle s. s.) in Figure 6, Figure 7, Figure 8, Figure 9, Figure 10, Figure 11, Figure 12, Figure 13, Figure 14, and Figure 15. Without the integrated approach of the lotus system at this point, a general question of homology would arise again. Could one homologise a muscle A in one species with a muscle A in an other species where "lotus spoken" unrelated "muscle drops" could have originated from B? Again, the lotus system does not list clear tables of homology, as it is simply a list of muscular units of a fluid material.

M. cervicocapitis (No. 79), the major part of Ogushi's (1913b) "m. cervico-hyo-capitis" is well defined as a separated muscle by several authors and is innervated by the dorsal rami of the first four (Scanlon 1982) to five (Vallois 1922) spinal nerves.

Whether m. collosquamosus (No. 57) phylogenetically separates from the "cauda squamosi" part of Ogushi's (1913b: No. 35) "m. cervico-hyo-capitis" (~ m. cervicocapitis, No. 79) can only be tested ontogenetically. Both muscles (No. 57 and ~ No. 79) originate from different cervical vertebrae and gain a different innervation, arranging the former to hypoglossus (n. XII) and the latter to epaxial muscular regions.

For neck muscles, a large confusion exists about homology, nomenclature, and evolutionary identity (see synonyms in Appendix 1). Some further problematic structures are exemplarily mentioned next.

The identity of m. atlantoopisthoticus (No. 78) was discussed in literature. It is innervated by R. m. epistropheo-squamosi of n. XII in P. sinensis (Ogushi 1913b, p. 360) but spinal nerve n. S1 innervated in Chrysemys picta (Scanlon 1982). Ogushi (1913b) assumed the muscle – together with the S1d-innervated m. atlantoepistropheooccipitis (No. 75) – to originate phylogenetically from the anterior part of the S2d- to S5d-innervated m. cervicocapitis (~ No. 79, see above). In contrast, Scanlon (1982) assumed it being the spinal nerve-1 innervated atlas-part of the serially arranged intracervical m. interspinalis cervicis. Only developmental studies can solve this problem. However, both hypotheses assume an epaxial muscular origin of m. atlantoopisthoticus (No.78), which I also do. The shape of the posterior skull region, as well as the related retracting mechanism (Dalrymple 1979) differentiates trionychids such as P. sinensis (Ogushi 1913a; Dalrymple 1977) from the emydid C. picta (Jamninsky 2007) and all other turtle taxa. This may have resulted in a rostrad shift of m. atlantoopisthoticus (No. 78) insertion, correlated to a shift of innervation from spinal nerve-1 to n. hypoglossus (XII) in P. sinensis. The same scenario is also imaginable for the above-mentioned n. XII innervated m. atlantoexoccipitis (No. 54) from a spinal nerve innervated region, which would also give some further evidence for the presented identity of "m. collo-capitis longus."

The phylogenetic origin of m. atlantoepistropheooccipitis Pars profundus (No. 76) is not entirely clear. Topographically, it seems to be a separation of m. atlantoepistropheooccipitis Pars principalis (No. 75); and in Chelonoidis denticulata, it is still connected to No. 75 (Wiedemann 1803). However, George and Shah (1955a, figure 7) have drawn it as a cranium-attached continuation of the intercervical m. longissimus cervicis complex. As the authors did not draw any fibres crossing between those muscles, I correlated m. atlantoepistropheooccipitis Pars profundus (No. 76) to m. atlantoepistropheooccipitis Pars principalis (No. 75), and as the latter muscle (after Ogushi 1913b) I hypothesise it to be innervated by the dorsal branch of n. S1.

Vallois (1922) distinguished two muscles spanning between the anterior part of the carapace and the cervico-cranial region, one lateral, and one medial one. Each of them has one capitis- (No. 81, 82 and related portions) and one cervicis-portion. The observation of Vallois (1922) is supported by descriptions of Burne (1905) and Herrel et al. (2008). The diversity of m. carapacocervicocapitis lateralis Pars capitis (No. 81) was discussed by Vallois (1922), who distinguished two major types. It inserts to the basicranial region in Pleurodira (Shah 1963), whereas in marine turtles it attaches the posterodorsal edge of the skull (e.g., Rathke 1848; Hoffmann 1890; Vallois 1922; Schumacher 1972; Jones et al. unpublished work). In all remaining cryptodires, the muscle portion (No. 81) is missing (Meckel 1828; Rathke 1848; Hoffmann 1890; Ogushi 1913b; Shah 1963; Scanlon 1982). The shared condition with pleurodires may be evidence for a basal position of marine turtles within cryptodires; however, a functional explanation for the different insertions is difficult. A ventral insertion in Pleurodira may be explained by the laterad retracting mechanism of this group. Cryptodires retract their neck differently to pleurodires and may have lost the insertion of m. carapacocervicocapitis lateralis to the head (No. 81). Marine turtles lack the ability to retract their neck, and the presence of m. carapacocervicocapitis lateralis Pars capitis (No. 81) may represent a plesiomorphic relict of Testudines-ancestors. As with marine turtles, the stem taxa of Testudines apparently did not retract their neck, indicated by bony spines in the neck region (Proganochelys quenstedtii: Gaffney 1990), an absence of a carapace, or an absence of a domed rib cage (Odontochelys semitestacea: Li et al. 2008). Assuming a basal position of marine turtles (Werneburg and Sánchez-Villagra 2009) one may declare the non-retracting morphotype of No. 81 to plesiomorphically perform a stabilisation function, whereas in Pleurodira No. 81 – having a derived condition with a shifted insertion – acts as a lateral retractor. In cryptodires, able to retract their neck (non-marine turtle cryptodires), No. 81 may have been reduced because in this taxon the longest muscle of turtles – m. retrahens capiti collique (No. 88 with three related portions, Scanlon 1982) – acts as the main retractor of the head/neck (Callister et al. 1992).

With a very similar shape, m. carapacocervicocapitis medialis Pars capitis (No. 82) occurs dorsally in the neck of Cryptodira (excl. trionychids: e.g., Rathke 1848; Ogushi 1913b), but it is missing in Pleurodira (e.g., Gräper 1932: Chelodina longicollis). Due to the multidirectional flexibility of the neck in trionychids and the lateral flexure behaviour of pleurodires, No. 82 may have been reduced in those taxa.

M. carapacocervicocapitis lateralis Pars capitis (No. 81) was described for Chelodina longicollis as a muscular structure with two broad heads in the origin (Shah 1963). One may assue a fusion of No. 81 and No. 82 in this species or a phylogenetic / ontogenetic donor-derivate relationship of both muscular units at this point. In actual fact, the mm. testocervicocapitis lateralis et medialis may have been derived from one donor muscle (m. longissimus, Gasc 1981; Tsuihiji 2005, 2007); however, the comparative approach of this study speaks for a derived two-headed origin of No. 81 in C. longicollis and not for a fusion of No. 81 and No. 82. The integration of No. 81 to the cervicalis-portion of m. carapacocervicocapitis lateralis was not described or depicted by Shah (1963); further observations will help to understand the anatomy in this species.

The spinal nerves nn. S1v-S7v innervated (Ogushi 1913b) m. longus colli is a serially arranged muscle ventral to the cervical vertebrae. It forms two portions attaching via one tendon to the basioccipital region of the cranium. M. longus colli Pars capitis-II/III (No. 87) originates with two heads/tendons ventrally from the 2^nd and the 3^rd cervical vertebrae, m. longus colli Pars capitis-I (No. 86) originates ventrally from the 1^st cervical vertebrae. The whole m. longus colli was described as the ventralmost muscle in the neck and not to be included into the n. XII innervated m. coracohyoideus complex (No. 58-60) (Shah 1963). Most authors recognised intercrossing fibres between the serially arranged portions; however, Bojanus (1819-21), Meckel (1828), or Hoffmann (1890) named No. 87 as a separated muscle (Appendix 1). Due to the inconsequence of muscle nomenclature and the diverse anatomy of the anteriormost muscular structures of the neck, a strange confusion is recognisable in literature (Appendix 1). For example, m. collocapitis brevis (No. 84) was assumed being part or a homologue to the anteriormost portion of m. longus colli (No. 87) or vice versa (Scanlon 1982). In addition to a separated m. collocapitis brevis (No. 84), Hoffmann (1890) still recognised three independent origins of the two cranium-attaching portions of m. longus colli (No. 86, 87), which converge in one tendon. Hence, I elaborated separated muscular units at this point. After accurate observations, Jones et al. (unpublished work) did not find No. 84 in Caretta caretta and Lepidochelys kempii (Chelonioidea). Where the muscle occurs, one may infer m. collocapitis brevis (No. 84) to originate phylogenetically from anterior portions of m. longus colli complex. Scanlon (1982) also described a close relation of No. 84 to both longus colli capitis-I (No. 86) and -II/III (No. 87) in insertion site.