Evolution of jaw depression mechanics in aquatic vertebrates: insights from Chondrichthyes

The widely accepted phylogenetic position of Chondrichthyes as the sister group to all other living gnathostomes makes biomechanical analyses of this group of special significance for estimates of skull function in early jawed vertebrates. We review key findings of recent experimental research on the feeding mechanisms of living elasmobranchs with respect to our understanding of jaw depression mechanisms in gnathostome vertebrates. We introduce the possibility that the ancestral jaw depression mechanism in gnathostomes was mediated by the coracomandibularis muscle and that for hyoid depression by the coracohyoideus muscle, as in modern Chondrichthyes and possibly placoderms. This mechanism of jaw depression appears to have been replaced by the sternohyoideus (homologous to the coracohyoideus) coupling in Osteichthyes following the split of this lineage from Chondrichthyes. Concurrent with the replacement of the branchiomandibularis (homologous to the coracomandibularis) coupling by the sternohyoideus coupling as the dominant mechanism of jaw depression in Osteichthyes was the fusion and shift in attachment of the interhyoideus and intermandibularis muscles (producing the protractor hyoideus muscle, mistakenly refereed to as the geniohyoideus), which resulted in a more diversified role of the sternohyoideus coupling in Osteichthyes. The coracohyoideus coupling appears to have been already present in vertebrates where it functioned in hyoid depression, as in modern Chondrichthyes, before it acquired the additional role of jaw depression in Osteichthyes.


INTRODUCTION
The vertebrate jaw has proven to be a model subject of research on the evolution of functional systems. This complex system of muscles and skeletal elements that serves as a crucial interface between vertebrates and their environment has provided the material for several major concepts and repeating themes in the evolution of organismal design (Gans, 1961;Shaeffer & Rosen, 1961;Osse, 1969;Crompton & Parker, 1978;Barel, 1983;Lauder & Liem, 1983Bramble & Wake, 1985;Frazzetta, 1986;Aerts et al., 1987;Aerts, 1991). However, our understanding of the evolution of vertebrate jaws has been limited because functional morphological research on the sister group to all other living gnathostomes, the Chondrichthyes, has lagged behind work on other major clades. Because of the phylogenetic position of Chondrichthyes, establishing details of jaw mechanics in this group is vital to our understanding of the feeding mechanism in the earliest jawed chordates. In this paper we discuss some implications of recent research on the functional morphology of feeding in elasmobranchs (Haller, 1926;Moss, 1972;Motta, Hueter & Tricas, 1991;Frazzetta, 1994;Motta & Wilga, 1995;Motta et al., 1997;Wilga, 1997;Wilga & Motta, 1998a,b, 2000 for our understanding of gnathostome jaw evolution. We focus on the history of mouth opening mechanisms and particularly the systems of muscles and skeletal elements that are involved in depression of the lower jaw. One of the general conclusions from comparative studies of aquatic feeding mechanisms in bony fishes and salamanders is that the major biomechanical couplings involved in mouth opening have been largely retained throughout the radiation of these vertebrate groups (Reilly & Lauder, 1990;Lauder & Shaffer, 1993). The primary mechanism mediating lower jaw depression in all groups of living fishes and salamanders that have been studied to date, with the exception of Chondrichthyes, is a linkage involving a ligamentous connection between the hyoid bar and mandible that transmits posterior rotation to the mandible (Lauder, 1980a;Lauder & Shaffer, 1985;Bemis & Lauder, 1986;Bemis, 1987). Recent research on chondrichthyans has revealed that this linkage is not a viable mechanism of jaw depression in this group (Ribbink, 1971;Moss, 1972;Motta et al., 1991Motta et al., , 1997Motta & Wilga, 1995;Wilga, 1997;Wilga & Motta, 1998a,b, 2000. Our primary purpose in this paper is to review this observation and discuss its implications for our understanding of early vertebrate feeding systems in the light of current estimates of vertebrate interrelationships. We introduce and explore the possibility that the ancestral jaw  Shirai, 1996) showing taxa investigated functionally. 1, Negaprion brevirostris, Sphyrna tiburo; 2, Squalus acanthias; 3, Rhinobatos lentiginosus. depression mechanism in gnathostomes was mediated by the coracomandibularis muscle and that of hyoid depression by the coracohyoideus as seen in modern Chondrichthyes. METHODS The morphology and function of the muscles involved in the coracomandibularis and coracohyoideus couplings in lower vertebrates was compiled from the literature. Muscle synonymies are from Edgeworth (1935), Winterbottom (1974), and Miyake, McEachran & Hall (1992). We reserve the chondrichthyan terms 'coracomandibularis' and 'coracohyoideus' couplings as general term for those couplings commonly referred to as the 'geniohyoideus' and 'rectus cervicis'. In this paper, we use the term 'hyoid' to refer to the ventral elements of the hyoid arch, which usually consist of the ceratohyal and the basihyal in most groups, but includes the hyomandibula in chondrichthyans. Schematic diagrams of musculoskeletal couplings in elasmobranchs were produced from anatomical dissections of at least five freshfrozen specimens each of four species of elasmobranchs: the lemon shark Negaprion brevirostris (63-229 cm TL); the bonnethead shark Sphyrna tiburo (56-82 cm TL), the spiny dogfish Squalus acanthias (46-65 cm TL), and the Atlantic guitarfish Rhinobatos lentiginosus (52-63 cm TL) (Motta & Wilga, 1995;Motta et al., 1997;Wilga, 1997;Wilga & Motta, 1998a,b, in review). These species represent each of the three major elasmobranch radiations (Fig. 1).
Computer axial tomography scans and photographs of intact heads approximating the location of the jaws in the resting position ( jaws closed) and with the jaws and hyoid maximally depressed were used to measure the angle of the hyoid relative to the lower jaw (for more detail see Motta & Wilga, 1995). Hyoid rotation in degrees was calculated as the difference between the resting position and its position at peak lower jaw depression in one fresh-frozen specimen each of N. brevirostris (102 cm TL), Squalus acanthias (70 cm TL), and Sphyrna tiburo (84 cm TL). This angle was measured at the intersection of lines drawn along the dorsal edge of the teeth on the lower jaw and the dorsal edge of the ceratohyal.
Motor activity and kinematics of head movements during feeding in N. brevirostris (5 individuals, 36 capture trials), Sphyrna tiburo (3 individuals, 12 capture trials), Squalus acanthias (8 individuals, 44 capture trials), and R. lentiginosus (5 individuals, 42 capture trials) were studied using simultaneous electromyography and high-speed video (for more detail see: Motta et al., 1997;Wilga, 1997;Wilga & Motta, 1998a, b, in review). Lateral and ventral video recordings were made during feeding experiments using a NAC 200 high-speed video camera at 200 fields per second. Bipolar electrodes were implanted in select cranial muscles using 26 gauge hypodermic needles. Fish were anaesthetized for surgery using 0.065 g/l of tricaine methanesulfonate (MS 222) and maintained on this solution during surgery. After surgery, fish were returned to the experimental tank to recover. Feeding trials were begun after normal swimming behavior was observed for at least one hour postrecovery and continued until the fish was satiated. Prey items found naturally in the diet were offered as follows: Atlantic thread herring (Opisthonema oglinum) and crevalle jack (Caranx hippo) for N. brevirostris; Pacific herring (Clupea pallasii) for Squalus acanthias; speckled crab (Arenaeus cribrarius), pink shrimp (Penaeus duorarum) and Atlantic thread herring (Opisthonema oglium) for Sphyrna tiburo; pink shrimp (Panaeus duorarum) for R. lentiginosus. Electrode wires were attached to differential amplifiers set at a gain of 1000, bandpass 100-3000 Hz with a 60 Hz notch filter. Signals were simultaneously monitored on a four-channel oscilloscope and an eight-channel thermal array recorder and recorded on a pulse code modulator that multiplexed the signals to a videocassette recorder. The EMG and video recordings were synchronized using a unit that directed a preprogrammed repeating pulse simultaneously to one channel of the tape recorder and to LED strobes that were recorded by the video camera. Chart recordings of electromyographic data from each muscle were analysed by measuring burst duration and burst onset relative to the start of lower jaw movement as determined by the pattern of synchronization pulses on the video images and EMG tracings. At the termination of each experiment, sharks were euthanized by MS-222 overdose according to Institutional Animal Care and Use Committee guidelines of the University of Washington, Friday Harbor Laboratories, the University of South Florida, and Mote Marine Laboratory. Positions of the electrodes were verified by dissection and body length measured. The following muscles were recorded from: epaxialis, coracomandibularis, coracohyoideus (not implanted in Sphyrna tiburo), coracoarcualis (N. brevirostris and Squalus acanthias only), quadratomandibularis, depressor mandibularis (R. lentiginosus only), coracohyomandibularis (R. lentiginosus only) and depressor hyomandibularis (R. lentiginosus only). The time of the following kinematic events was determined from the video recording: onset of lower jaw depression, onset of hyoid depression, peak hyoid depression, onset of lower jaw elevation, onset of upper jaw protrusion, onset of upper jaw retraction and complete jaw closure.  Figure 2. Jaw depression couplings in a teleost (after Lauder, 1985 Our comparisons of motor patterns among species are based on major qualitative differences concerning the synchrony of motor activity with specific kinematic events. In the jaw depression analysis, lower jaw depression and hyoid depression partially overlap in time, however the onset and completion of these kinematic events consistently occurred independently and allow differences in the motor pattern and the corresponding kinematic event to be detected. Muscle morphology, function and motor pattern were mapped by hand onto a cladogram of gnathostomes. The resulting phylogenetic distribution of these traits was used as a basis for inferring evolutionary sequences of change in these components of the feeding mechanism.
posteroventrally. This causes posterodorsal rotation of the proximal end of the hyoid, which is transmitted to the posterior end of the mandible through the MH ligament. As a result, the anterior tip of the mandible rotates posteroventrally around the quadratomandibular joint to depress the lower jaw. Contraction of the hypaxialis muscles may fix or retract the pectoral girdle allowing the CH-coupling to work more effectively (Lauder, 1985).
The CH-coupling in sharks consists of the hypaxialis-pectoral girdlecoracoarcualis-coracohyoideus-hyoid linkage (see Fig. 3) (Motta et al., 1997;Wilga, 1997;Wilga & Motta, 1998a, 2000. The origin of the coracohyoideus is from the coracoarcualis and the insertion is onto the hyoid, while the insertion of the coracoarcualis is onto the coracohyoideus and the origin from the pectoral girdle, thus they are linked functionally and morphologically as a coracohyoideus complex. Functional analyses have revealed that in sharks, contraction of the coracohyoideus complex pulls the anterior-medial end of the hyoid posteroventrally, as in teleosts. However, unlike the teleost hyoid, the proximal end of the shark hyoid does not move dorsally with contraction of the rectus cervicis, rather, it rotates anteriorly and ventrally (Fig. 4). This anteroventral rotation of the proximal hyoid is seen in the CAT scans, radiographs, and photographs of the head skeleton of sharks with the mouth closed and with the hyoid maximally depressed (see Fig. 4). Such images reveal that the distal hyoid is rotated posteroventrally around the mandibularceratohyal articulation through an angle of 55°in N. brevirostris, 60°in Squalus acanthias, and 70°in Sphyrna tiburo (Motta & Wilga, 1995;Wilga & Motta, 1998a, 2000. The hyoid appears to rotate around the articulation to the lower jaw as well as, and independently of, the hyomandibular-ceratohyal articulation, thus the MH ligament does not appear to transmit rotation to the mandible. The coracomandibularis coupling (CM-coupling) in teleosts, coupling 1 in Figure  2, consists of the hyoid-protractor hyoideus-mandible linkage. Contraction of the protractor hyoideus may either elevate or protract the hyoid compressing the buccal cavity or depress the lower jaw if the hyoid is fixed by coupling 2. The CM-coupling in elasmobranchs is composed of the hypaxialis-pectoral girdlecoracomandibularis-mandible linkage (coupling 1 in Fig. 3). As the pectoral girdle in sharks is fixed in position by the hypaxialis, contraction of the coracomandibularis pulls the anterior tip of the mandible posteroventrally. The coracomandibularis in elasmobranchs is not linked to the hyoid as it is in Osteichthyes and its line of action is always below the jaw joint and thus it always acts to depress the lower jaw.
The above interpretations of function in the CM-coupling and the CH-coupling in sharks are supported by electromyographic data from four species of elasmobranchs (see Figs 5 and 6). Motor activity in the coracomandibularis always precedes that of the coracohyoideus and coracoarcualis and always begins prior to the onset of lower jaw depression. Motor activity in the coracohyoideus and coracoarcualis may not begin until after the onset of lower jaw depression and may not continue throughout lower jaw depression. However, motor activity in these muscles begins just before the start of hyoid depression and ends prior to peak hyoid depression. In N. brevirostris, peak hyoid depression often occurs after activity in the coracohyoideus and coracoarcualis muscles have stopped. The extended duration of hyoid depression is probably due to water influx pushing against the hyoid when the mouth is opened and the shark is swimming. Note that peak hyoid depression occurs during elevation of the lower jaw.

DISCUSSION
The mechanisms of jaw action during feeding have now been studied in representatives of most of the major living non-amniotic, aquatic vertebrate lineages.  Wilga & Motta, 1998a,b). Black boxes indicate the motor pattern with left and right error bars showing one standard error of the burst onset and duration times respectively. Dark grey box indicates a second burst of activity. White circles indicate time of peak hyoid depression. The first grey region indicates the expansive phase of mouth opening from the start to maximum lower jaw depression. The middle grey region indicates the compressive phase of mouth closing from maximum lower jaw depression and the start of upper jaw protrusion to complete jaw closure and maximum upper jaw protrusion. The last grey region indicates the recovery phase from jaw closure and maximum upper jaw protrusion to complete retraction of the upper jaw. Muscle name abbreviations: EP, epaxialis; DM, depressor mandibularis; DH, depressor hyomandibularis; CM, coracomandibularis; QM, quadratomandibularis; CH1, coracohyoideus; CH2, coracoarcualis; CH3, novel division coracohyomandibularis. Note the lack of activity in CH1 in R. lentiginosus even though the muscle was correctly implanted and was active during other behaviours.

Anatomical features of jaw depression couplings
The lower jaw depression couplings in aquatic gnathostomes involve the coracomandibularis and coracohyoideus muscles and their derivatives. The muscles that are believed to be homologous to the coracomandibularis and the coracohyoideus are known by a variety of names in the different vertebrate clades (Tables 1, 2). In keeping with our hypotheses that chondrichthyans possess the ancestral mechanisms and to simplify discussion throughout this paper, when we refer to couplings that involve these homologous muscles, we will refer to them using the terms  (1985), compresses buccal cavity Shaffer & Lauder (1985) Muscle synonymies from Edgeworth (1935), Winterbottom (1974), and Miyake et al. (1992). * Based on anatomical studies. * * Note that the muscle commonly called the geniohyoideus in teleosts is actually the protractor hyoideus, which arises from fusion of the intermandibularis posterior and the interhyoideus muscles and therefore is not homologous to the geniohyoideus of other lower vertebrate groups; however, we use it in this study because it is analogous to the geniohyoideus (Edgeworth, 1935;Winterbottom, 1974;Miyake et al., 1992). coracomandibularis or coracohyoideus coupling as the general term and follow with the specific muscle by group in parenthesis when it differs from chondrichthyans.
The protractor hyoideus muscle in teleosts is commonly, albeit mistakenly, referred to as the geniohyoideus muscle, which is involved in the coracomandibularis coupling. According to Edgeworth (1935) and Winterbottom (1974), the protractor hyoideus is composed of a fusion of the intermandibularis posterior and the interhyoideus muscles which resulted in the protractor hyoideus which spans the hyoid and mandible. The intermandibularis spans the mandible while the closely apposed interhyoideus spans the hyoid in other fishes. Furthermore, they concluded that any muscle that is homologous to the geniohyoideus (coracomandibularis coupling) in other lower vertebrates has been lost in teleosts, as well as gars. However, the protractor hyoideus muscle is functionally analogous to the coracomandibularis coupling of other vertebrates and so we use it in our discussion to show the phylogenetically broad roles of these couplings in jaw mechanics.

Function of the coracohyoideus coupling
Anatomical and experimental evidence on Chondrichthyes supports the role of the CH-coupling in mediating hyoid depression in aquatic lower vertebrates. In Osteichthyes, the CH-coupling also functions to depress the lower jaw but this function has not been found in chondrichthyans (Fig. 7). The CH-coupling depresses the hyoid in both Chondrichthyes and Osteichthyes (sternohyoideus), which expands the buccal cavity and aids in suction feeding and directing food posteriorly. In sharks, as in other extant aquatic gnathostomes, the proximal end of the hyoid is connected to the proximal end of the lower jaw by several ligaments, but this linkage does not appear to effect jaw depression via hyoid retraction (Gadow, 1888;Daniel, 1915;Allis, 1923;Nobiling, 1977;Motta & Wilga, 1995, 1999. Manual manipulation of the hyoid does not depress the mandible beyond the effect of pushing on the ventral floor of the buccal cavity and the underlying coracomandibularis muscle. In video images of feeding sharks, a distinct ventral bulging of the hyoid is observed to travel posteroventrally from behind the mandibular symphysis shortly after the start of lower jaw depression (Wu, 1994;Motta et al., 1997;Wilga, 1997;Ferry-Graham, 1998;Wilga & Motta, 1998a, 2000. As the hyoid is depressed, it rotates around the hyomandibular-ceratohyal and mandibular-hyomandibular articulations in N. brevirostris, Squalus acanthias, and Sphyrna tiburo through an angle between the mandible and the hyoid, from 0°to 55-70°at peak hyoid depression. Thus, the hyoid rotates independently of the lower jaw and the MHL ligaments do not appear to transmit hyoid rotation to the mandible.
When the mouth is opened in sharks, the proximal end of the hyoid does not move posterodorsally as in teleosts. Instead this element is rotated anteroventrally, displacing the entire jaw apparatus anteroventrally relative to the chondrocranium. Since posterodorsal elevation of the proximal end of the hyoid does not occur it cannot function to depress the lower jaw in a manner similar to that in bony fishes. Furthermore, peak hyoid depression occurs during elevation of the lower jaw in sharks, well after peak lower jaw depression has taken place.
Electromyographic data support this interpretation as activity in the coracohyoideus and coracoarcualis muscles do not begin until 25-43 ms after the onset of coracomandibularis activity and may not begin until well after the onset of lower jaw depression in Squalus acanthias and N. brevirostris (Wilga & Motta, 1998a;Motta et al., 1997). Thus, the CH-coupling is unlikely to mediate lower jaw depression in elasmobranchs by transmitting movements of the hyoid to the mandible through the MH ligaments.
The CH-coupling does not mediate lower jaw depression in batoids and does not appear to do so in holocephalans, although experimental data on the latter group is lacking. These taxa present unique cases due to the distinctive morphology of their jaw suspension systems. The hyoid arch has separated in batoids with the ventral portion of the hyoid arch (ceratohyal) associated with the first branchial arch and the dorsal portion (hyomandibula) connected to and supporting the jaws (Gregory, 1904;Maisey, 1980). The coracohyoideus inserts onto the ventral portion of the hyoid arch, which is not connected to the jaws and therefore precludes it from mediating lower jaw depression (Wilga & Motta, 1998b). In support of this, the coracohyoideus (see CH1 in Fig. 6) is not active during lower jaw depression in R. lentiginosus (Wilga & Motta, 1998b). However, batoids have a novel division of the coracohyoideus muscle, the coracohyomandibularis that arises from the embryonic coracohyoideus along with the coracoarcualis and inserts onto the hyomandibula (Marion, 1905;Miyake et al., 1992). The coracohyomandibularis depresses the hyomandibula and in doing so also depresses the jaw apparatus, but is not active until the latter half of lower jaw depression when the mouth has already been partially opened (Wilga & Motta, 1998b). Experimental evidence suggests that the role of the coracohyomandibularis muscle is to expand the orobranchial region for the production of suction (Wilga & Motta, 1998b).
Holocephalans are the only living gnathostomes to possess a morphologically complete hyoid arch that is not involved in suspending the jaws from the cranium and is free from the cranium (Gregory, 1904;Maisey, 1980;Didier, 1995). The coracohyoideus in holocephalans also appears to function in depressing the hyoid (Ribbink, 1971;Didier, pers. comm.). Although a MHL ligament is present that may assist lower jaw depression in holocephalans, this mechanism must work in conjunction with the coracomandibularis in those holocephalan groups in which the coracohyoideus originates from the coracomandibularis (Didier, 1995).
Although the visceral skeleton in placoderms is poorly known, the presence of hypobranchial muscles that depress the hyoid arch has been inferred in detailed morphological analyses of other researchers (Miles & Westoll, 1968;Denison, 1978).
Thus, it appears that the CH-coupling in Chondrichthyes and possibly placoderms serves to expand the orobranchial cavity by depressing the hyoid. It is unclear whether the CH-coupling in placoderms included an MH ligament and if such a linkage was involved in jaw depression.
Experimental analyses of the coracohyoideus coupling in bony fishes (sternohyoideus) and aquatic salamanders (rectus cervicis) indicates that it functions to depress the hyoid (see Fig. 2) (Lauder, 1985;Lauder & Shaffer, 1993). During hyoid depression the proximal end of the hyoid is rotated posterodorsally, pulling the MH ligament which then pulls the proximal end of the lower jaw posterodorsally resulting in depression of the lower jaw (Liem, 1980;Lauder, 1979Lauder, , 1980aLauder, , 1985Shaffer & Lauder, 1985a;Bemis & Lauder, 1986). Although activity in all of the head muscles overlap broadly during feeding, activity in the sternohyoideus (coracohyoideus homologue) in bony fishes and rectus cervicis (coracohyoideus homologue) muscles in salamanders coincides with the mouth opening phase and thus the CH-coupling is the primary mechanism of lower jaw depression in bony fishes and salamanders (Osse, 1969;Liem, 1980;Lauder, 1979Lauder, , 1980aBarel, 1983;Shaffer & Lauder, 1985b;Lauder & Shaffer, 1985;Bemis & Lauder, 1986;Aerts et al., 1987;Aerts, 1991). The hyoid arch is poorly known in acanthodians. However, a small accessory element between the hyomandibula and ceratohyal of Acanthodes is presumed to be an interhyal. The presence of this element indicates that the hyoid arch may have played a role in the jaw mechanism in Acanthodes similar to that in extant bony fishes (Miles, 1964(Miles, , 1968, which suggests that it possessed a CH-coupling similar to that of extant bony fishes. Secondary mechanisms for jaw depression that operate independently of the CHcoupling exist in various groups. Teleosts possess an opercular coupling (see Fig. 2) (Osse, 1969;Anker, 1974;Barel et al., 1977), and batoids, lungfish and salamanders have independently evolved a depressor mandibularis coupling (Lauder & Shaffer, 1985;Shaffer & Lauder, 1985a,b;Bemis & Lauder, 1986).

Function of the coracomandibularis coupling
Functional analyses of feeding in Squalus acanthias, N. brevirostris, Sphyrna tiburo and R. lentiginosus indicate that the CM-coupling mediates depression of the lower jaw in this phylogenetically broad sample of elasmobranchs (Wilga & Motta, 1998a,b, in review;Motta et al., 1997). Ribbink (1971) proposed that the coracomandibularis muscle is responsible for depressing the lower jaw in holocephalans. As illustrated by Didier (1995) and Ribbink (1971), the origin of the coracomandibularis muscle is ventral to the jaw joint in holocephalans, resulting in a line of action that would depress the lower jaw. Good evidence based on detailed morphological studies of placoderms show that a CM-coupling connected the pectoral girdle to the lower jaw, with a line of action ventral to the jaw joint (Heintz, 1932;Stensio, 1959;Miles, 1969;Miles & Westoll, 1968). As a result of these studies, the CM-coupling in placoderms has been hypothesized to depress the mandible (Heintz, 1932;Stensio, 1959;Miles, 1969;Miles & Westoll, 1968).
The role of the CM-coupling during feeding is variable in bony fishes and aquatic salamanders. The branchiomandibularis (coracomandibularis coupling) is active during lower jaw depression in Amia in capture events and during lower jaw elevation in manipulation events in Polypterus and Amia (Lauder, 1980a). Some insight into this apparent dual role is provided by Elshoud-Oldenhave and Osse (1976) and Lauder (1979Lauder ( , 1981, who noted that the function of the coracomandibularis coupling changes depending on its line of action. When the line of action is dorsal to the jaw joint it elevates the lower jaw and when it is ventral to the jaw joint it depresses the lower jaw. In an anatomical study of the head of Acipenser, Stengel (1962) proposed that the branchiomandibularis acts to depress the lower jaw. In teleosts, instead of a branchiomandibularis, the protractor hyoideus (analogous to the coracomandibularis coupling) protracts the hyoid and compresses the buccal cavity (Osse, 1969;Lauder, 1979;Liem, 1980;Wainwright & Lauder, 1986;Wainwright et al., 1989). Perhaps, as the sternohyoideus (coracohyoideus coupling) took an increasing role in depressing the lower jaw, the branchiomandibularis (coracomandibularis coupling) atrophied until it was lost. Meanwhile, the protractor hyoideus evolved to function primarily in protracting the hyoid and compressing the buccal cavity.
In an anatomical study of Latimeria, Lauder (1980b) proposed that the coracomandibularis acts to elevate the lower jaw or compress the buccal cavity rather than to depress the lower jaw as reported by Millot and Anthony (1958), Thomson (1967Thomson ( , 1970, and Trewavas (1959). In Ambystoma, the geniohyoideus (coracomandibularis coupling) is active during lower jaw depression and buccal compression after the jaws have closed (Shaffer & Lauder, 1985a,b;Lauder & Shaffer, 1985). The geniothoracis (coracomandibularis coupling) is active throughout the feeding event from lower jaw depression through lower jaw elevation in Lepidosiren (Bemis & Lauder, 1986). Distinct muscle attachment sites have led researchers to the hypothesis that general 'fish type' mandibular and hyoid muscles can be restored in Acanthodes (Miles, 1968). If so, then this suggests that a protractor hyoideus (coracomandibularis coupling) that functioned similarly to that in Osteichthyes may have been present in Acanthodes. Thus, the role of the coracomandibularis coupling in basal bony fishes and acanthodians appears to be both jaw depression and buccal compression and that in teleosts is primarily in hyoid retraction and buccal compression.

Evolution of jaw depression couplings
We define the CH-coupling as a coracohyoideus muscle, or its derivative, originating on the pectoral girdle and attaching on the hyoid with a ligamentous connection between the hyoid and the mandible. This anatomical configuration exists in all living gnathostome groups, apparently existed in acanthodians (Denison, 1978), and may have existed in placoderms (Miles, 1968(Miles, , 1969. We therefore hypothesize that the anatomical CH-coupling was present in the common ancestor of living gnathostomes (see Fig. 7, Bar #1). However, there is no clear indicaion of the functioning of the CH-coupling in lower jaw depression in placoderms. A CH-coupling that functions in jaw depression is not known from Chondrichthyes. However, the CH-coupling is the primary mechanism of jaw depression in actinopterygians, lungfishes, and aquatic salamanders. It has been inferred to function in Latimeria and acanthodians (Miles, 1964(Miles, , 1968Lauder, 1980a,b). Therefore, a CH-coupling that functions to depress the jaw appears to be primitive for the Teleostomi, as Lauder (1980a, b) suggested (see Fig. 7, Teleostomi=Acanthodii and Osteichthyes). This interpretation implies that the anatomical CH-coupling existed in gnathostomes before a role in jaw depression evolved.
To summarize our interpretation of the evolution of gnathostome jaw mechanics, we recognize two states of the CM-coupling in gnathostomes (see Fig. 7). A CMcoupling between the pectoral girdle and mandible that functions to depress the lower jaw is present in elasmobranchs plus holocephalans and is therefore inferred to be primitive for Chondrichthyes. Existing interpretations indicate that a coracomandibularis-like muscle in placoderms connected the pectoral girdle and mandible (Heintz, 1932) and we infer by similarity to living groups that this muscle may have been a jaw depressor in placoderms. If our functional interpretations of the fossil taxa are correct, and in light of the phylogenetic hypothesis presented in Figure  7 we note that this condition is estimated to have been present in the common ancestor of placoderms and living gnathostomes (Bar #1 Fig. 7). This conclusion must be tempered by the lack of unequivocal evidence for the mechanism of jaw depression in placoderms. In actinopterygians, coelacanths, lungfish, amphibians, and probably acanthodians the CM-coupling attaches onto the hyoid or branchial arch and is primarily involved with protraction of the hyoid and compression of the buccal cavity, but may also contribute to depression of the lower jaw in some groups. Determination of coracomandibularis muscle function in Acipenser, Polyodon, and Latimeria awaits experimental analyses. Thus, because this character state is shared by the major lineages of the Osteichthyes we hypothesize that it is primitive for the clade (see Bar #2, Fig. 7). These observations imply a transformation series of the jaw depression mechanism in gnathostomes that begins with jaw depression by the CM-coupling, as seen in living chondrichthyans. This mechanism was then replaced by the CH-coupling in Osteichthyes, in conjunction with a shift in attachment of the analogous CM-coupling which resulted in a more diverse role of the CHcoupling.
Future research should focus on two areas. First, a refined anatomical understanding of fossil lineages, such as placoderms, that pre-date the presumed split between Chondrichthyes and Osteichthyes is needed. Of particular interest will be estimates of the origin of the mandibulohyoid ligament, and attempts to test the hypotheses of coracomandibularis function in jaw depression. Because Chondrichthyes are the only living vertebrate lineages in which jaw depression is effected by the CM-coupling further functional analyses of holocephalans and basal elasmobranch taxa will be important to our interpretation of the fossil forms.