Morphology of the jaw, suspensorial, and opercle musculature of Beloniformes and related species (Teleostei: Acanthopterygii), with a special reference to the m. adductor mandibulae complex

The taxon Beloniformes represents a heterogeneous group of teleost fishes that show an extraordinary diversity of jaw morphology. I present new anatomical descriptions of the jaw musculature in six selected beloniforms and four closely related species. A reduction of the external jaw adductor (A1) and a changed morphology of the intramandibular musculature were found in many Beloniformes. This might be correlated with the progressively reduced mobility of the upper and lower jaw bones. The needlefishes and sauries, which are characterised by extremely elongated and stiffened jaws, show several derived characters, which in combination enable the capture of fish at high velocity. The ricefishes are characterised by several derived and many plesiomorphic characters that make broad scale comparisons difficult. Soft tissue characters are highly diverse among hemiramphids and flying fishes reflecting the uncertainty about their phylogenetic position and interrelationship. The morphological findings presented herein may help to interpret future phylogenetic analyses using cranial musculature in Beloniformes.


INTRODUCTION
The m. adductor mandibulae complex belongs to one of the most intensively studied soft tissues in vertebrates. It primarily moves the skeletal elements associated to the mandibular arch and is the main head and the most powerful feeding musculature. The m. adductor mandibulae complex is highly adapted to different feeding strategies among vertebrate clades and, as such, experienced a large amount of diversification. Its anatomy is informative for different phylogenetic levels and a mutual evolution with jaw and skull anatomy can be observed (e.g., Gosline, 1986;Diogo, 2008;Diogo & Abdala, 2010;Datovo & Vari, 2013).
The complex jaw musculature of Beloniformes has only been studied in very few species so far, and most published descriptions of beloniform species are superficial and insufficiently illustrated, making broad scale phylogenetic comparisons impossible. That makes broad phylogenetic comparisons impossible. The aim of the present study was to illustrate and describe the morphological diversity of cranial musculature of six selected species of Beloniformes in great detail and to compare it to external jaw anatomy. By using manual dissections and histological slide sections I aim to provide a comprehensive anatomical basis for future researchers studying more species in a phylogenetic context.
In the present, purely anatomical study, the great diversity within beloniform subgroups or within non-beloniform groups could not be studied by maintaining the provided extent and detail of illustrations and descriptions. However, I present some considerations about the potential phylogenetic relevance of some characters that have to be tested in future studies. Therefore, four selected near related acanthopterygian species, which may serve as outgroup in future phylogenetic studies, are described. In addition to two atherinomorph species, I included the percomorph Perca fluviatilis, which was recently used to define the ancestral pattern of atherinomorph jaw musculature (Hertwig, 2008), and the mugilomorph Rhinomugil corsula, which is possibly closer related to Atherinomorpha (Stiassny, 1990;Setiamarga et al., 2008;Near et al., 2013). A preliminary character mapping is presented.

Anatomical observations
Standard procedures for histology and manual dissection are those used by Werneburg (2007) and Werneburg & Hertwig (2009).
For dissection, two or more specimens per species were used. In the first step of dissection (summarised in Fig. 2) the lateral view of the skinned head including all muscles in their unaltered place, including the jaw adductor musculature, opercle-, and suspensoric-related musculature, was documented. In the second step, the external section of m. adductor mandibulae (A1) was mostly removed and the course of the internal section of m. adductor mandibulae (A2/3) was depicted. Further steps of dissection    Arrangement of the species studied herein and those from the literature (*) used for the reconstruction of character evolution (character mapping); following Lovejoy, Iranpour & Collette (2004;compare to Fig. 1C). Outlines indicate the species, which were manually dissected herein; not to scale . did allow inspection of the symplectic in lateral view with the A2/3 completely or partly removed. Finally, the medial view of the jaw apparatus was documented with a focus on the musculature medial of the lower jaw, namely the intramandibular section of m. adductor mandibulae (Aω), the anterior part of m. protractor hyoidei, and m. intermandibularis.
Serial sections were prepared for all species (slice thickness = 12 µm), except for Pe. fluviatilis and B. belone due to the size of these species. The positions of the sections are indicated in the dissection figures (Figs. 6,8,10,12,14,16 and 19). For S. saurus, a juvenile specimen was used for histological sectioning (Fig. 20), whereas for manual dissections and character coding (as for all species), adult specimens were used (Fig. 20).

Character evolution
Using PAUP* (Swofford, 2003), a character mapping was performed. Therefore, the topology of Lovejoy, Iranpour & Collette (2004) was used as template to arrange the phylogeny of the beloniform species studied herein and of three additional hemiramphid species ( Fig. 4; cf. Fig. 1C). For the interrelationship of major acanthopterygian groups, the present study follows the findings of Stiassny (1990), Setiamarga et al. (2008), and Near et al. (2013). Therein, Percomorpha form the sister taxon to Ovalentaria. Consequently a polarisation of characters is given. The topology for the character mapping was drawn using the move branch function in Mesquite 2.01 (Maddison & Maddison, 2011).

Characters and character mapping
In total, 37 soft tissue characters are described and discussed below. The character matrix can be found in Table 1. The results of the character mapping are listed in Table 2. Therein, the consensus of Acctran and Deltran optimizations are documented. Due to the particular focus on the morphological descriptions and illustration of this study, the taxonomic sampling is limited. Also the available data from the literature record is limited. As such, I avoid discussing the character changes in detail. They should serve as summary of character distribution of the species studied herein. The phylogenetic relevance of the characters should be subject of evaluation and discussion in future, more quantitative analyses of the cranial musculature of Beloniformes. Those studies may also consider more closely related species for the comparison with Atherinomorpha.

External section of the m. adductor mandibulae complex (A1)
The m. adductor mandibulae is differentiated into different muscle sections in teleost fishes, representing a complex of individual muscles, each having a separated origin, course, and insertion (Diogo, 2008;Diogo & Abdala, 2010). The external section of m. adductor mandibulae complex, A1, is the lateral-most jaw muscle. If present, it originates posteriorly on the suspensorium and/or on the preopercle, it runs rostrad, and has a tendinous insertion to the upper or lower jaw (i.e., Allis, 1897).
General appearance. An A1 is present in Perca fluviatilis ( Figs  General appearance 0 0 1 1 0 1 0 0 1 0 0 1 0 Orientation 2 3 X X X 2 X 2 X 1 0 X 2 External section of m. adductor mandibulae (A1) Insertion 4 3 X X X 1 X 2 X 0 1 X 1 Origin 1 X 1 0 0 0 0 1 0 0 1 1 1 Lateral head 0 X 0 1 0 0 0 3 2 2 0 0 3 Medial head 2 X 4 4 X X X 3 2 0 1 4 3 Intermedial head 1 X 1 X X X X 2 X X 0 1 2         In O. latipes, Hertwig (2008) and Werneburg & Hertwig (2009) described a lateral muscle of the adductor complex with an insertion to the lower jaw. It could be interpreted in two different ways: First, it could represent A1, the possession of which is plesiomorphic; A1 is present in all non-beloniform fishes studied and in O. latipes, it autapomorphically would have shifted its insertion to the lower jaw. Second, A1 could be reduced in O. latipes (Hertwig, 2005). In that case, one additional step of transformation would be needed, as the internal section of m. adductor mandibulae (A2/3) would be modified secondarily. Hertwig (2005) followed the principle of parsimony and opted for the first explanation. Werneburg (2007) interpreted an insertion of A1 to the maxilla and homologised the muscle to the A1 of the outgroup representatives. After reanalysing, this finding was revised and A1 actually inserts on the posterior edge of the dentary at two-thirds of its height below the coronoid process of this bone and has contact via connective tissue to the lig. maxillo-mandibulare in this species (Werneburg & Hertwig, 2009). Previously, the latter connection was misinterpreted as an upper jaw insertion (Werneburg, 2007). Wu & Shen (2004) mentioned a small ventrolateral portion of A1, their A1-VL, in two flying fish species. As Hertwig (2005: 39) already pointed out, the homologisations of those authors remain unclear. Moreover, the illustration of that portion is lacking. It appears that Wu & Shen (2004) may have confused this portion with the lateral subdivision of A2/3. Hertwig (2008, 149) wrote: 'In an extensive comparative study of the m. adductor mandibulae in teleostean fishes, [the authors], however, did not mention a subdivision of A2/3 either in the Mugilomorpha or in the Atherinomorpha, but this is probably down to their limited taxon sample, which comprised only three species of the latter.' If Wu & Shen (2004) actually identified the remainder of A1 as their A1-VL (supported by the fact that an insertion of A1-VL to the maxilla is present), a high interspecific variability may be hypothesised for the flying fishes. Starks (1916) dissected a belonid species, Tylosurus acus, in which he described an A1-muscle. Following the present homologisation, however, that muscle clearly represents the lateral head of the muscle A2/3, which has a similar anatomy as found in B. belone (see also below) and S. saurus .
For the ground pattern of Atherinomorpha, Hertwig (2005) proposed that the external (A1) and internal (A2/3) sections are situated next to each other in a horizontal plane. As an outgroup of Atherinomorpha, the author used Pe. fluviatilis, in which the A2/3-portions are situated above each other in a horizontal plane ( Figs. 2A and 5). In the present study, R. corsula was dissected as an additional, potential outgroup species, which is closely related to Atherinomorpha. Similar to Atherinomorpha (sensu Hertwig, 2008), the A1 of that species also has to be interpreted to be lateral to the A2/3 in a horizontal plane. As such, that character has to be withdrawn as an autapomorphy of Atherinomorpha. More detailed observation among Percomorpha could identify the orientation of A1 to A2/3 in Pe. fluviatilis (Figs. 2A and 5) as autapomorphy of Percomorpha or only of that species. In the latter case, the 'A1 in horizontal plane to A2/3' would need to be interpreted as plesiomorphic among Acanthopterygii. Observations among Mugilomorpha could identify the orientation of A1-A2/3 as a homoplastic character of R. corsula and Atherinomorpha. If all members of Mugilomorpha had an A1 lateral to A2/3, and when following the phylogenetic hypothesis of Stiassny (1990), that spatial orientation would need to be interpreted as a synapomorphy of Mugilomorpha + Atherinomorpha.
Insertion The insertion of A1 to the jaws is different in all species studied. A definition of homology (e.g., A1 inserts laterally to the maxilla) was not made, because the differences of A1 were too large. Hertwig (2008) observed several atherinomorph species and defined the insertion of A1 at the lateral face of the maxilla to be present in Pe. fluviatilis and "Aplocheilidae". In contrast to Pe. fluviatilis (Figs. 2A and 5), however, the A1 inserts on the other end of the maxilla in Ap. lineatus (Figs. 2D, 10 and 11). The latter species has an additional tendon to the medial face of the lacrimal, a character which was found by Hertwig (2008) to be present in the ground pattern of Atherinomorpha (compare to Alexander, 1967;Parenti, 1993;Stiassny, 1990). For Cyprinodontiformes (incl. Aplocheilus), Hertwig (2005) was not able to define an unambiguous constellation of the insertion of A1. However, he argued that the insertion of A1 shifted based on the rotation of the maxilla in this taxon. As such, the insertion of A1 to the lateral face of the maxilla could be interpreted as being plesiomorphic among Atherinomorpha.

Internal section of the m. adductor mandibulae complex (A2/3)
The A2/3 usually originates with two or three muscle heads on the suspensoric and on the preopercle and inserts as a consistent muscle to the lower jaw. Muscle heads are defined as partial differentiations of a muscle. They have separated origins or insertions (Werneburg, 2007;Werneburg, 2011). Muscle heads gain a descriptive nomenclature herein; their position of origin (or insertion) and the spatial orientation were considered. This nomenclature differs from Winterbottom (1974), because that one is not applicable for muscle heads herein.
A2/3 can have an intramandibular portion. A muscle portion is defined as having a separate origin, course, and insertion, but as having some intertwining fibres or a shared tendon with another muscle portion of the same ontogenetic and/or phylogenetic origin (Werneburg, 2007;Werneburg, 2011).
Similar to the present study, Hertwig (2005) and Hertwig (2008) found the origin of the medial head of A2/3 to be highly variable. In addition to an adult specimen of S. saurus, a juvenile was studied (Figs. 9E-9H). In this specimen, a different orientation of the A2/3-heads was found (Werneburg, 2007). One could hypothesise that the medial head of A2/3 in the juvenile shifts its origin to a dorsal position and the intermedial head of A2/3 could shift its origin to a more ventral position (two transformation steps). Alternatively, the origin of the medial A2/3-head of the juvenile could shift ventrolaterally to the intermedial head of A2/3 and would be homologous to the intermedial head of A2/3 in the adult. Hence, the intermedial head of A2/3 in the juvenile (then the medial head of the adult) would keep its origin at the sphenotic (one transformation step). Those scenarios are very speculative because they are derived from only one observation. No final answer can be presented, because the variability of that character within S. saurus cannot be estimated. The species D. pussila, B. belone, and S. saurus show a very drastic ontogenetic elongation of the lower jaw (Hemiramphidae) or of both jaws (Belonidae, Scomberesocidae) (Boughton, Collette & McCune, 1991;Lovejoy, 2000;Lovejoy, Iranpour & Collette, 2004). It would be valuable to study if, correlated to the elongation of jaws, changes in the anatomy of the jaw musculature occur (origin, volume, course, insertion). Comparative ontogenetic and electromyographic studies (Focant, Jacob & Huriaux, 1981;Osse, 1969) could help to interpret the specific case mentioned herein. Ontogenetic changes in the anatomy of the jaw musculature were already observed by Hertwig (2005) in representatives of Goodeidae (Cyprinodontiformes: Crenichthys). Nanichthys (Scomberesocidae) is often not accepted as a 'genus' in a taxonomic sense and is often referred to as a dwarf morphotype of Scomberesox (Collette, 2004;Collette et al., 1984). However, if the juvenile specimen of S. saurus studied herein would actually represent a member of a valid genus Nanichthys, the arrangement of the A2/3-musculature may serve as a criterion to distinguish both species taxonomically.
Intermedial head. The intermedial head of A2/3 is situated between the lateral and the medial head. It originates only on the horizontal aspect of the preopercular in R. corsula (Figs. 6 and 7) [state 0]. It takes its origin from the horizontal aspect of the preopercle and at the processus caudalis quadrati in Ap. lineatus (Figs. 10 and 11), B. belone (Fig. 18), and S. saurus (Figs. 19 and 20) [state 1] and originates only on the processus caudalis quadrati in O. latipes (Figs. 12A-12C and 13) and X. oophorus (Fig. 3) [state 2]. An intermedial head is not present in Pe. fluviatilis, At. boyeri, Pa. brachypterus, and D. pussila. Muscle portions. Unlike in all other species [state 0], A2/3 is laterally separated into two portions (by definition; see above and Werneburg, 2011) in At. boyeri (Figs. 8,9 and 12D) [state 1]. The muscle portions of A2/3 have separated origins lateral at the posterior part of the suspensoric as well as separated insertions medial to the lower jaw. The medial portion of A2/3 is differentiated into two heads at its origin. The lateral portion of its A2/3 is not separated into heads. Among the species studied herein, and indeed, considering data from Hertwig (2008) regarding several other atherinid species, this condition has to be declared autapomorphic for At. boyeri (Atheriniformes).
Orientation of muscle heads. The spatial orientations of the medial and the lateral head of A2/3 are different among species. In Pe. fluviatilis ( Figs. 2A and 5), Ap. lineatus (Figs. 10 and 11), D. pussila (Figs. 2G, 16 and 17), X. oophorus (Fig. 3), B. belone (Fig. 18), and S. saurus (Figs. 19 and 20), the medial head of A2/3 is situated dorsally to the lateral head or is at least clearly visible in lateral view [state 0]. The medial head of A2/3 is situated ventrally to the lateral head in At. boyeri (Figs. 8,9 and 12D)  ]. In the former species, it also inserts on the coronomeckelian bone [state 1], which is only found in these two species. It represents a bone, which is posterodorsally fused with the border of processus primordialis anguloarticularis. Both bones are separated from each other by a clear suture (Werneburg, 2007). Intramandibular portion. An intramandibular portion of A2/3 is lacking in all Beloniformes [state 0]. It is present in R. corsula (Figs. 6 and 7) and has a narrow insertion on the medial face of processus coronoideus dentalis [state 1]. In Ap. lineatus (Figs. 10 and 11), it has broad insertions to the processus coronoideus dentalis, to cartilago Meckeli, and to the anguloarticular [state 2]. It inserts medially to the dentary in At. boyeri (Figs. 8,9 and 12D) [state 3] and has a narrow insertion medially to the anguloarticular in Pe. fluviatilis (Fig. 5) [state 4].
The configuration of the intramandibular portion of A2/3 is different among non-beloniforms species studied here. As the criterion of homology, the intramandibular portion is defined to originate from an A2/3-associated aponeurosis or tendon herein. Hertwig (2008), who observed few species of Beloniformes (O. latipes and some hemiramphids), argued for an autapomorphic reduction of an intramandibular portion of A2/3 within Beloniformes, which I can confirm herein.
Intramandibular muscles possibly act in positioning the jaw (Karrer, 1967: "Stellbewegung"). Hertwig (2005) and Hertwig (2008) mentioned the reduction of intramandibular muscles and found a correlation between the loss of those muscles and a reduced mobility of particular bone elements. For Empetrichthys latos (Cyprinodontiformes), he noticed an ontogenetic reduction of intramandibular muscles. The movement of upper jaw bones in Beloniformes may be coupled to the movement of the lower jaw (see above) and hence they may underlie large mechanical stresses in fish hunting species. To withstand those forces, the bones of the lower jaw may have a higher degree of fusion resulting in the tendency to reduce intramandibular musculature.
Like Hertwig (2008), I defined an intramandibular portion of A2/3 as present in Pe. fluviatilis. However, the configuration of the intramandibular musculature of Pe. fluviatilis could be interpreted differently. In the present study, two intramandibular muscles were differentiated. First, an intramandibular portion of A2/3 is described as originating from the tendon of A2/3 by only a few muscle fibres. It narrowly inserts on the medial face of the anguloarticulare. Second, an intramandibular m. adductor mandibulae (Aω) is described, which is tendinously originating from the preopercular and the quadrate. That muscle has a flat insertion medially to the dentary, to cartilago Meckeli, and to the anguloarticular.
In contrast, Osse (1969) only described one intramandibular muscle for Pe. fluviatilis. That muscle, "Aω" in Osse (1969), has one origin at the tendon of A2/3. This "Aω" also has a narrow attachment to the anguloarticular, one tendinous attachment to the prearticular/quadrate and one flat insertion to the medial face of the lower jaw. Osse (1969) combined the Aω and the intramandibular portion of A2/3 of the present study as his "Aω." Therefore, he did not differentiate the course of muscle fibres and other associated structures. The fibres of the intramandibular portion of A2/3 of the present study run anteroventrad. The fibres of the Aω were found to originate as a double fibred muscle from the tendon originating from the prearticular/quadrate. However, some fibres also originate from the tendon of A2/3, which is only partly fused with the tendon of Aω. While both tendons fuse, the course of the Aω-tendon is still separable (Fig. 5D). The fusion of the tendons and the origin of some Aω-fibres at the A2/3-tendon may have persuaded Osse (1969) to define only one intramandibular muscle.
One additional interpretation of intramandibular muscle configuration is possible. If a tendinous insertion of A2/3 to the tendon of Aω is hypothesised, the origin of some Aω-fibres may have been shifted to the tendon of A2/3. In that case, no intramandibular portion of A2/3 would exist in Pe. fluviatilis. If this configuration is a plesiomorphic condition of Acanthopterygii, the character should also be interpreted as a reversal within Beloniformes. In contrast, if one hypothesises the intramandibular portion of A2/3 to be independently reduced in Pe. fluviatilis, the character should be considered as homoplastic in Pe. fluviatilis (Percomorpha) and Beloniformes. To clarify that controversy, additional species of Percomorpha and Acanthopterygii need to be observed in great detail, but this was outside the scope of the present study.

Intramandibular section of the m. adductor mandibulae complex (Aω)
The intramandibular section of the m. adductor mandibulae complex (Aω) connects the suspensoric with the medial face of the lower jaw.
Origin. It originates with a tendon anteriorly at the medial face of the symplectic in Pa. brachypterus (Figs. 14 and 15) and D. pussila (Figs. 16 and 17) (Figs. 6 and 7), B. belone (Fig. 18), and S. saurus (Figs. 19 and 20), Aω originates broadly on the medial face of the quadrate and a part of the muscle can have a tendinous origin [state 2]. It attaches with a tendon anteroventrally to the medial face of the quadrate in At. boyeri (Figs. 8,9 and 12D) and Ap. lineatus (Figs. 10 and 11) [state 3]; and in Pe. fluviatilis (Fig. 5), it originates with a tendon anteriorly at the medial face of the horizontal aspect of the preopercular and to a small amount medially at the middle area of processus caudalis quadrati [state 4]. The Aω is absent in X. oophorus [state 5].
Hertwig (2005) defined as a common character of hemiramphids: The origin of the flat tendon of Aω is situated at a part of the symplectic, which points rostrad. He studied species of Hyporhamphus, Nomorhamphus, and Hemiramphodon. Due to the diverging observation in D. pussila herein (Figs. 16 and 17), this character on the origin of Aω cannot be confirmed to be diagnostic for all hemiramphids. However, as that character was also found in Pa. brachypterus (Figs. 14 and 15), a potential synapomorphic character of (Exocoetidae + Hemiramphidae) is identified and a possible monophyly of Hemiramphidae could be indicated (Rosen, 1964;Rosen & Parenti, 1981;Collette et al., 1984). This would contradict the works of Lovejoy, Iranpour & Collette (2004) and Aschliman, Tibbetts & Collette (2005), who found "Hemiramphidae" paraphyletic. In the work of Lovejoy, Iranpour & Collette (2004), the Zenarchopteridae (among others Dermogenys, Hemiramphodon, Nomorhamphus) oppose the paraphyletic "Belonidae" (incl. Scomberesocidae) and Hyporhamphus belongs to a group, which opposes (Zenarchopteridae + "Belonidae"). Several species of "Hemiramphidae" that are closely related to Exocoetidae in the work of Lovejoy, Iranpour & Collette (2004), as well as several other species of the remaining groups of Beloniformes need to be observed to gain a better understanding on how that character is distributed. The absence of Aω was documented for some atherinomorph species by Hertwig (2008) and the reduction must have occurred several times independently.
Shape. In R. corsula (Figs. 6 and 7), Aω is separated into two heads at the level of the quadrate. The lateral head inserts broadly to the medial face of the dentary and cartilago Meckeli. The medial head of Aω inserts ventrally to the medial face of the dentary and anteriorly to the medial face of the anguloarticular [state 0]. The Aω represents a double-feathered muscle in Pe. fluviatilis (Fig. 5), At. boyeri (Figs. 8,9 and 12D), D. pussila (Figs. 16 and 17), B. belone (Fig. 18) Insertion. On the medial face of the lower jaw, the Aω (when not differentiated into heads) inserts broadly to the dentary, cartilago Meckeli and/or to the anguloarticular in Pe. fluviatilis (Fig. 5), At. boyeri (Figs. 8,9 and 12D),Pa. brachypterus (Figs. 14 and 15),and D. pussila (Figs. 16 and 17) [state 0]. It inserts broadly to the dentary, to the anguloarticular, and to the cartilago Meckeli, whereby a ventral part in feathered muscles inserts far anteriorly to the medial face of the dentary in B. belone (Fig. 18)  Hertwig (2005) and Hertwig (2008) has shown that the configuration of Aω is highly variable among Cyprinodontiformes. In comparison, this can also be concluded for the species observed herein.
In each species studied, several specimens were observed and a tendency of a rounder cross-section of the muscle was found in At. boyeri (Figs. 8,9 and 12D). In addition, the assignment to big-bellied or elongated oval has to be understood as a tendency in the variability of the specimens observed.
Course. During its course from origin to insertion, the thickness of m. adductor arcus palatini hardly changes in R. corsula (Figs. 2B, 6 and 7), O. latipes (Figs. 2E, 12A-12C and 13), and X. oophorus (Fig. 3) (Figs. 16 and 17), B. belone (Fig. 18) An autapomorphy in the ground pattern of Atherinomorpha may be the presence of a lig. palato-maxilla. The absence of the ligament in R. corsula (Figs. 6 and 7) and a different attachment of the ligament in Pe. fluviatilis makes it impossible to reconstruct the ground pattern.
Lig. parasphenoido-suspensorium. This ligament is present in Pe. fluviatilis (Fig. 5), At. boyeri (Figs. 8, 9 and 12D For Pe. fluviatilis, Osse (1969) described two ligaments (his No. XVII and XVIII) that originate from the parasphenoid and insert to the dorsal edge of the suspensoric. This differentiation of the ligament could not be identified in the manual dissections performed for the present study.

CONCLUSIONS
In the present study, the variety of jaw, suspensoric, and opercle muscles was described for several acanthopterygian fishes with a focus on Beloniformes. The diversity of jaw muscles within Beloniformes corresponds to the external differences in their jaw morphology. As such, long beaked forms and species with protractible mouths show remarkable differences in their jaw musculature that may be correlated to stiffening or high mobility of the jaws.
The A1 lowers the upper jaw in most fishes. As an autapomorphy of Beloniformes, Mickoleit (2004) mentioned the reduced mobility of bones related to the upper jaw. Hertwig (2005) hypothesised that the reduced mobility of those bones might be correlated with the reduction of A1 within Beloniformes or the displacement of the A1-insertion apart from the upper jaw. In the present study, such a replacement of A1 was discovered in O. latipes ( Fig. 2E; see also Werneburg & Hertwig, 2009). This species can still move its upper jaw during feeding (I Werneburg, pers. obs., 2006), which questions the possibility of a functional correlation of the character pair mentioned by Hertwig (2005) and Hertwig (2008), namely 'A1 no longer attached to upper jaw' and 'non-moveable upper jaw bones' .
Moreover, in the flying fish Pa. brachypterus, which has no A1 (Fig. 2F), a protrusible jaw was discovered herein. Therefore, the upper jaw bones are moveable against each other (Figs. 14 and 15).
The hemiramphid Dermogenys pusilla, which hunts at the surface of the water (Meisner, 2001), is able to easily move its short upper jaw, although the species has no A1 ( Fig. 2G). Hence, coupled by ligament attachments, the lifting of the upper jaw appears to be indirectly performed by lowering the lower jaw. A deep coupling of those structures can be hypothesised for most other A1-lacking Beloniformes. In addition, the mobility of the protrusible upper jaw of Pa. brachypterus suggests a strong ligament-bone interaction (Figs. 14 and 15).
Among hemiramphids, whose phylogenetic relationship is debated, A1 can be absent (this study: Dermogenys pussila; Hertwig, 2008: Hyporhamphus unifasciatus) or can be present (Hertwig, 2008: Nomorhamphus sp., Hemiramphodon phaiosoma;Rosen, 1964: Arrhamphus brevis). Also Exocoetidae seem to have members with an A1 (Wu & Shen, 2004: Cypselurus cyanopterus, Parexocoetus mento; but see comments in the Results section) and members without an A1 (this study: Pa. brachypterus). The phylogenetic significance of those conditions can first be adequately estimated when more species are observed and more clarity exists about phylogenetic interrelationship. But this requires further detailed and comprehensive observations.
At least for B. belone (Fig. 2H) and S. saurus (Figs. 19 and 20), one may hypothesise that the loss of the A1 could be related to a strong fixation of the upper jaw to the cranium, realised by lig. premaxillo-frontale. Whether the upper jaw of both species is still moveable in vivo is not known so far, but is not expected.
As seen in hemiramphids, an elongated lower jaw not necessarily involves the reduction of A1. Xenopoecilus oophoris, an adrianichthyid with duckbill-like jaws, also has an A1 (Fig. 3), which is attached to the upper jaw. This indicates that also an elongated upper jaw, which possibly was present in the ground pattern of Beloniformes already (Parenti, 1987), not necessarily implies the loss of A1. Only the derived condition of two species, B. belone and S. saurus, which possess a stiffened upper jaw, may be clearly correlated to the loss of A1. As such, it can be expected that another belonid, Potamorrhamphis eigenmannii (Miranda Ribeiro, 1915), which has a moveable upper jaw in vivo (I Werneburg, pers. obs., 2006), could have an A1, but this hypothesis needs further observation. The present study shows that the loss of A1 must not be interpreted only in correlation to elongated jaws. Other biomechanical requirements must be considered.
The studied selection of non-beloniform species must be handled with care when choosing them as potential outgroup species (as example see Hertwig, 2008). Compared to the insufficient documentation of the cranial musculature of most acathopterygian groups, the species dissected herein appear to show several derived characters. E.g., Rh. corsula has three main components of A2/3. Most mugiliform taxa, however, are reported to have a different arrangement of that muscle (Gosline, 1993: Agonostomus;Van Dobben, 1935: Mugil;Wu & Shen, 2004: Chelon, Crenimugil;Starks, 1916: Mugil;Eaton, 1935: Mugil). As the authors of these studies did not observe histological sections, these findings could represent artefacts caused by the lower resolution of manual dissection.
As representative of the potential sister group to all remaining Beloniformes, the adrianichthyids Oryzias latipes and Xenopoecilus oophorus were studied herein. Hertwig (2005), Hertwig (2008) and Werneburg & Hertwig (2009) already diagnosed several derived characters for O. latipes that could be affirmed herein and together with X. oophorus, it shares several derived characters. Due to the distinctive morphology of Adrianichthyidae, problems could arise when reconstructing the jaw muscle configuration in the ground pattern of Beloniformes. In addition to several derived characters, the taxon seems to display several plesiomorphic characters shared with Cyprinodontiformes. This finding persuaded Rosen (1964) and Li (2001) to postulate a sister group relationship of Adrianichthyidae + Cyprinodontiformes, named as Cyprinodontoidei (Fig. 1A). The present study highlights which characters are most variable among near related species and may assist taxon and character selection in future phylogenetic studies.
The differing external jaw morphology of diverse beloniform fishes is nicely reflected in the anatomy of their jaw musculature. Apparent changes concern the absence or presence of the A1 and arrangements of the intramandibular musculature. Both muscles are coupled to the upper or lower jaw, which are connected by ligaments themselves. The strong attachment of the upper jaw to the neurocranium, as visible in needlefishes and sauries, involves complex rearrangements of the soft tissue of the jaw apparatus.