Three-Dimensional Kinematics of the Jaws and Hyoid During Feeding in White-Spotted Bamboo Sharks

Elasmobranchs (sharks, skates and rays) use a distinct set of cartilaginous endochondral elements to capture and process prey using suction or bite mechanisms. In contrast, teleost use a complex linkage of dermal and endochondral bones to capture prey whether by suction or biting. This difference in complexity has, in part, enabled teleosts to radiate into a variety of different feeding styles where the jaw and hyoid arches are linked and function as a unit. However, elasmobranchs have decoupled depression of the lower jaw from depression of the hyoid arch and expand the buccal cavity using a different mechanism than teleosts. White-spotted bamboo sharks, Chiloscyllium plagiosum, are trophic generalists that use suction and crushing behaviors to take a wide variety of prey items. This study reconstructs the threedimensional kinematics of the mandibular and hyoid arches during suction food capture in three bamboo sharks using X-ray Reconstruction of Moving Morphology (XROMM). In this method, two biplanar high-speed x-ray video cameras are used to record skeletal motions during feeding strikes, after which tracking data is applied to animate CT scans, from which kinematic data are compared to hypotheses regarding the feeding mechanism. This research reveals that the mandibular and hyoid arches of bamboo sharks move in previously unsuspected ways when feeding. The hyomandibular cartilage swings backward and downward up to 58° while the ceratohyal simultaneously swings backward and downward up to 66° while suction feeding. At the same time, both cartilages undergo long-axis rotation up to 50°.: the dorsal surface of the hyomandibula rotates rostrally while the dorsal surface of the ceratohyal rotates caudally. The lower jaw supinates as the mouth opens and pronates past resting position as the jaws close. The upper jaw mirrors this motion, pronating as the mouth opens and supinating as the mouth closes. The extra degrees of rotation in the hyoid arch elements may explain the remarkable variation in feeding performance of bamboo sharks. This variation in motion of hyoid motions is achieved without limiting expansion of the oral cavity, which could comprise suction generation. Long-axis rotation of the jaws may function to enhance grasping of prey by exposing more tooth rows to the prey when feeding, while otherwise maintaining a more hydrodynamic profile. The additional motions and mechanisms of the jaws and hyoid during feeding may provide bamboo sharks with an adaptation to take to a wide variety of prey items despite specializations for suction feeding.

x

LIST OF FIGURES
Chapter 1 FIGURE PAGE Figure 1. Proposed motions of the hyomandibula (dark grey) and ceratohyal (light grey) of a white-spotted bamboo shark during a feeding event , and   medially, protruding the distal ends of the jaws rostrally as proposed in .

Summary Statement
A precise three-dimensional reconstruction of motions of the hyoid arch in a suction feeding shark.

ABSTRACT
Elasmobranchs feed in a variety of ways. Much of their success comes from deriving different feeding mechanisms from the same basic components of the hyoid and mandibular arches. The mechanism is then dependent on movements of skeletal elements during feeding. For suction-feeding elasmobranchs (sharks, skates and rays), ventral expansion of the hyoid cavity through depression of the hyoid arch generates most of the suction force used to capture prey. Three-dimensional movements of the hyoid arch of white-spotted bamboo sharks, Chiloscyllium plagiosum Bennet 1830, are quantified using X-ray Reconstruction Of Moving Morphology (XROMM). Three axes of rotation and were reconstructed for the hyomandibula and ceratohyal, along with translation of the distal end of each element. Long-axis rotation in elements of the hyoid arch of an elasmobranch is observed for the first time. As the ceratohyal and hyomandibula depressed, the hyomandibula also rotated rostrally from its proximal end and the ceratohyal rotated caudally. In addition, the hyomandibula rotated around its long-axis as it depressed. The dorsal margin of the ceratohyal rotates rostrally then caudally as the ceratohyal depresses. Long-axis rotation of the hyomandibula is predominantly in the opposite direction of the ceratohyal. Rostral rotation of the hyomandibula may contribute to rostral jaw protrusion in elasmobranchs and long-axis rotation of the hyomandibula and ceratohyal may increase rotational motions, specifically depression, of the hyoid arch by permitting reorientation of the elements during hyoid depression, but had little effect on protrusion/retraction of the hyoid arch.
Bamboo sharks are able to vary the timing and extent of protrusion and retraction relative to depression of the hyoid. This can allow bamboo sharks to adjust their feeding motions according to prey size or environment, enabling them to feed on variety of prey items, despite a highly specialized suction mechanism.

INTRODUCTION
Elasmobranchs (sharks, skates and rays) are capable of feeding using a variety of strategies. The success of a feeding strategy is directly dependent on the mechanism used to obtain prey. Numerous mechanisms have been proposed for feeding, based on morphological and ecological aspects of different shark species; however, kinematic tests of feeding mechanisms remain few, and are typically limited to motions along a single plane (e.g. Wilga and Sanford 2008). The feeding mechanism of elasmobranchs consists of a mandibular arch (the jaws) and a hyoid arch (a derived gill arch that supports the jaws; Mallatt 1996). The orientation and motions of the mandibular and hyoid elements determines the feeding mechanics of different shark species (Motta et al. 1997;Motta and Wilga 1999;. These elements are loosely connected to the cranium and highly kinetic. As such, motions of the skeletal elements of the head can be complex and three-dimensional. The skeletal motion of suction-feeding sharks is of particular interest because the mechanism appears multiple times among elasmobranchs and directly converges on the condition in teleost fishes Wainright et al. 2015) and the sequence of motions are highly stereotyped as a result of strong physical constraints .
Suction-feeding sharks, such as the white-spotted bamboo shark, are capable of equal or even lower oral pressures than teleost fishes, but powered primarily through expansion of their hyoid region (as compared to expansion of hyoid and mandibular elements in teleost fishes, Day et al. 2015;Lauder 1980;Wainright et al. 2007;Wainright et al. 2015) and despite using fewer skeletal elements (Nauwelearts et al.2008;Wilga and Sanford 2008;). At the same time bamboo sharks, are also capable of feeding on a variety of prey Lowry and Motta 2007 For suction-feeding sharks the motions of the hyoid arch generate the forces that draw water and prey into the mouth Wilga and Sanford 2008). The mandibular and hyoid arches are able to move independently, connected only at the hyoidiomandibular joint through a series of ligaments.
Transverse motions of the hyoid elements are responsible for expanding the buccal cavity during suction-feeding Wilga and Sanford 2008); however, other mechanisms may be involved in feeding beyond suction. In bamboo sharks the buccal cavity compresses laterally by adduction of laterally directed hyomandibulae (  . Protrusion of the hyomandibula may then protrude the lower jaws, a possible mechanism for moving the jaws closer to prey during feeding . Transverse motions of the hyoid elements relate to expansion of the buccal cavity and have been well studied because of this; however, the rostro-caudal motions, particularly at the hyomandibuloceratohyal joint ( Fig. 1A) are linked to protrusion of the jaws . Jaw protrusion follows prey capture in Orectolobiformes and does not appear to relate to suction in elasmobranchs . As prior studies on bamboo sharks have focused on the effects of hyoid mechanics on suction, these rostrocaudal motions have gone largely unexamined.
Based on manipulation of deceased specimens the hyomandibuloceratohyal joint of Chiloscyllium plagiosum may move rostrally simultaneous to ceratohyal depression ( Fig. 1B; Ramsay 2012) as in other orectolobiforms ), contributing to rostral jaw protrusion and jaw depression . In Orectolobus rostral protrusion of the proximal end of the ceratohyal depresses and protrudes the lower jaw rostrally via an articulation with the mandibular knob . A similar mechanism could be present in Chiloscyllium; however, the mandibular knob is smaller than in Orectolobus and the hyomandibula, rather than the ceratohyal, articulates directly with the Meckel's cartilage . In the morphologically similar nurse shark, ceratohyal depression follows jaw depression ) and any connected hyomandibula protrusion would not contribute to leverage of the lower jaw during depression. Instead hyomandibula motion would contribute rostral protrusion of the jaws.
The way that longitudinal and transverse motions of the hyoid arch interact during feeding has not been measured for any elasmobranch. Motion of hyoid elements through longitudinal and transverse planes is complex, both as a result of operating in three-dimensions, but also as a result of articulations at multiple joints.
Complex multi-planar motions can result in long-axis rotation of skeletal elements, an additional degree of motion that has not been described or hypothesized for hyoid elements (Kambic et al. 2014). Long-axis rotation of cranial elements has not been studied for aquatic vertebrates; however, the limbs of walking birds exhibit long-axis rotation during complex multiplanar motions (Kambic et al. 2014). Long-axis rotation can increase range of motion of skeletal elements, as well as permit complex maneuvering behavior (Kambic et al. 2014). Depression of the ceratohyal not only reorients it relative to the cranium, but creates a complex series of changes in the orientation of the basihyal and hyomandibula (Fig. 1). Long-axis rotation in the hyomandibula and ceratohyal may act to maintain pitch of the basihyal, the insertion for muscles that depress the hyoid arch (Motta and Wilga 1999; relative to the cranium, or increase the possible range of motion during ceratohyal depression.
Measuring three-dimensional motions of the hyoid arch is difficult. Thick muscle and skin of the head prevents the use of video .
Sonomicrometry can measure internal motions with high precision ; however, measurements beyond a single plane require implantation of numerous crystals and complex calculations to infer direction simply from changes in distance. The low density of cartilage relative to water has also inhibited radiocinematographic analysis of hyoid elements ). X-ray Reconstruction of Moving Morphology (XROMM, ) tracks skeletal elements, here using dense markers in the hyoid and jaw elements, and biplanar high speed x-ray video to record and reconstruct motions in three dimensions Gidmark et al. 2012).
Bamboo sharks move the hyoid elements transverse to the cranium to generate suction; however, bamboo sharks feed on a broad variety of prey and show a high degree of variation in suction performance. Additional mechanisms, specifically protrusion of the hyomandibuloceratohyal joint for prey capture may account for variation in diet and performance. Three-dimensional kinematics of the hyomandibula and ceratohyal of a suction-feeding elasmobranch are recorded in-vivo for the first time, including the magnitude and timing of rotation and translation. XROMM was used to reconstruct three-dimensional motions of the hyoid elements. Kinematic data was used to test for protrusion of the hyomandibula and long-axis rotation of the hyomandibula and ceratohyal during suction. If protrusion of the hyomandibula corresponds with prey capture then it is predicted to occur during or prior to peak gape. Long-axis rotation is predicted to result from protrusion/retraction and depression of the hyomandibula and ceratohyal and greater long-axis rotation is predicted to correspond to increased range of motion, including greater magnitude protraction/retraction and depression. In addition variation in hyoid motions is compared across multiple individuals, variation is predicted to be high, corresponding to previous reports of variation in suction performance in bamboo sharks .
The sharks were anaesthetized with 0.033 gL -1 of tricaine methanesulfonate  in a recirculating fish anaesthesia system . All experimental procedures were approved by Institutional Animal Care and Use Committees of Brown University and the University of Rhode Island. All sharks recovered to feed normally following implantation. Prior to feeding trials food was withheld from individuals for one week.

CT Scans
Each shark was CT scanned to determine marker positions and create models of the cranium and each element in the mandibular and hyoid arches (the basihyal was The tool shelf is available at xromm.org. Centroids of markers are exported from Maya as CT coordinates (CTex tool). For the distal end of the ceratohyal, rostral and ventral translation was measured.

Video data collection
Medial translation of the distal end of the ceratohyal was also measured, but only varied from zero in some trials, and medial translations were small because the distal end of the ceratohyal is directly limited by articulation with the basihyal.

Statistical Analysis
To test for intraspecific differences in rotations for both the hyomandibula and the ceratohyal, single factor ANOVAs were performed using the General Linear Model in SAS 9.4 software (SAS Institute, Cary NC). The ANOVAs compared timing of the peak for rotational axes of each element, and comparing magnitude of the peak rotation for axes of each element. Individuals were the factor and time and magnitude were the measured variables for each axis of rotation of the ceratohyal and the hyomandibula. Specifically based on observed direction of peak motion, the significant differences between individuals were tested for time and magnitude of: negative peak in long-axis rotation, protrusion (positive) peak rotation, and peak depression (negative) of the hyomandibula and ceratohyal, as well as negative longaxis rotation of the ceratohyal (Table 1). ANOVAs were also used to assess peak rostral and ventral translations of the distal ends of the hyomandibula and ceratohyal and medial translation of the hyomandibula. For trials with significant (p<0.05) individual differences a Tukey adjusted multiple comparisons test was used to determine which individuals varied significantly (Table 2) for homogeneity of variance and normality of the distribution respectively, for each ANOVA.

Hyoid motions during feeding
Bamboo sharks captured food using inertial suction in all trials and transported the food to the back of the mouth in a single motion in eleven of the twelve trials. For a single trial prey was sucked to the mouth, grasped and then transported in a second motion. Very little cranial elevation was observed during feeding strikes, but the head was positioned so that the food was directly in front of the mouth.
Feeding began with the mouth opening by depression of the Meckel's cartilage ( Fig. 4 between -100 and 0 ms). As the Meckel's cartilage depressed and gape increased the ceratohyal began to retract and depress, and the hyomandibula began to rotate such that its distal end depressed ventrally and protruded rostrally (most clear in

Rotations from joint coordinate systems (JCSs)
The hyomandibula and ceratohyal elements are nearly cylindrical making it difficult to visualize long-axis rotation in the XROMM animations. Data from the JCSs, however, show substantial long-axis rotation in both elements (Fig. 5). The hyomandibula and ceratohyal rotated in opposite directions, with the dorsal margin of the hyomandibula moving rostrally and the dorsal margin of the ceratohyal moving caudally. The magnitude of long-axis rotation of the hyomandibula is greater than protrusion and depression, with means (±standard error of mean, s.e.m.) of -34.4±3.2, 29.8±1.9 and -28.2±3.6 degrees respectively (Table 1). Initially the dorsal margin of the ceratohyal rotated slightly rostrally, but then made a greater caudal rotation (Fig.   5B). Rostral long-axis rotation of the ceratohyal was small, -12.22±2.0 deg, and caudal long-axis rotation was greater, 21.2±5.0 deg, but smaller than retraction and depression at -43.4±2.0 and -47.0±4.4 deg, respectively.
The mouth began to open substantially before the hyoid elements began to depress, and the majority of ceratohyal and hyomandibula motion occurred after peak gape (Fig. 5). All three rotations of both the ceratohyal and hyomandibula peaked at about 50 ms after peak gape, but with a wide range of variation (Table 1).

Translations of points in an anatomical coordinate system (ACS)
The  Table 2). At the same time depression moved the distal end of the hyomandibula -6.2±0.7 mm ventrally and 6.8±0.6 mm medially, constricting the buccal cavity as the distal end of the ceratohyal translated -12.8±0.7 mm ventrally to expand the buccal cavity (Table 2). Ceratohyal rotation about its Y-axis moved the distal end caudally by 13.3±1.1 mm. In the trial shown there is also a lateral translation of the distal end of the ceratohyal (Fig. 6B) that is not seen in all trials.

Variation among individuals
Individual variation was common for motions of the hyoid arch. Significant levels of individual variation can be seen as p-values in Tables 1 (rotations) and 2 (translations) and how individuals vary in timing and magnitude of motions is shown in Table 3 (Table 3). Magnitudes of ceratohyal long-axis rotation also differed significantly, as positive long-axis rotation was greater in Bamboo 2 than Bamboo 3 and negative long-axis rotation was higher magnitude in Bamboo 3 than either of the other individuals. Bamboo 1 retracted and caudally translated the ceratohyal earler than other individuals, and non-long-axis rotations were typically of lower magnitude in Bamboo 1 and Bamboo 2 than rotations of Bamboo 3 (Table 3).

Regression
Least squares linear regression analysis (

DISCUSSION
The

Protrusion
The hyomandibula of C. plagiosum protrudes as the ceratohyal depresses and retracts , as in other orectolobiform shark species. Protrusion of the hyomandibula pushes the mandibular knob and the jaw joint rostrally (Fig. 1B); however, protrusion of the hyomandibula follows lower jaw depression, and therefore does not contribute to jaw opening mechanics during feeding. As in other elasmobranchs the timing of protrusion is also too late to enhance suction forces Wilga and Sanford 2008) through rostral protrusion of the jaws (Holzman et al. 2007).
Protrusion of the hyomandibula coincides with the retraction and depression of the ceratohyal, and may be an important component of the buccal expansion mechanism of C. plagiosum. By protruding the hyomandibuloceratohyal joint rostrally the horizontal rotational axis (Z) of the ceratohyal is shifted so that as the ceratohyal retracts the distal end of the ceratohyal is being constantly depressed. If there was no protrusion then, as the ceratohyal retracted, it would achieve peak depression ventral to the hyomandibuloceratohyal joint, elevating or remaining level as the ceratohyal continued to retract. Instead retraction and depression peak at nearly the same time in all trials. Simultaneous peak in retraction and depression allows bamboo sharks to increase buccal volume throughout the contraction of the hyoid depressors connecting the ceratohyal and basihyal to the pectoral girdle. While hyoid and buccal pressure peaks at or before peak gape , water continues to flow into the buccal cavity until the jaws close. Continuous expansion of the buccal cavity can prevent backflow and any subsequent prey escape until the jaws are closed (Muller and Osse 1984).

Long-axis rotation
Long-axis rotation of the hyoid elements had not been recorded for any other vertebrate, but based on this study may be a key component of the rotational motions of the hyoid arch during expansion and compression of the buccal cavity of elasmobranchs. The degree to which dorsal margin of the hyomandibula rotates rostrally is directly related to the degree of depression of the hyomandibula (slope= 1.08, Table 4), so that long-axis rotation is an integral part of the depression mechanism and therefore adduction of the hyomandibula. A similar relationship is seen with the depression and dorsal to caudal long-axis rotation of the ceratohyal, though less extreme (Table 4), and more varied across all trials and the proximal articulation of the ceratohyal is less constrained by hyomandibula than the proximal end of the hyomandibula is by the cranium. At the joint with the chondrocranium the hyomandibula has a pair of articulations (Motta and Wilga 1999); it may be that the hyomandibula must rotate around its long-axis to maintain contact with both theses articulations during depression.
Positive Long-axis rotation of the ceratohyal, where the dorsal margin rotates caudally (Fig. 7), has a strong relationship with depression of the ceratohyal (Table 4).
While the ceratohyal will still depress in the absence of positive long-axis rotation, depression is stronger when long-axis rotation is positive, and greater dorsal-to-caudal rotation is associated with greater depression. Feeding attempts with only negative (dorsal margin rotating rostrally) long-axis rotation showed lower depression, but still sufficient for a successful strike. It is worth noting that ventral translation did not vary significantly among individuals including trials those trials without positive long-axis rotation, suggesting that during those food captures there was an alternative successful mechanism for lowering the distal end of the ceratohyal beyond just ceratohyal depression while rotating the dorsal margin caudally.
Long-axis rotation may reorient the distal end of the ceratohyal relative to the basihyal. As the basihyal descends and moves caudally it may reorient so that the rostral end depresses relative to the caudal end, or it may stay level relative to the Depression of hyoid elements may be closely linked to long-axis rotation in hyomandibula and the ceratohyal, but the amount of protrusion/retraction appears to vary independent of depression or long-axis rotation ( Table 4). The rostro-caudal motions of the joints of the hyoid arch are timed closely with the depression of the hyoid elements, but the magnitudes of the motions appear to be independent of depression or long-axis rotation, such that how much the hyomandibula depresses does not affect the amount that it protrudes rostrally (Table 4). Decoupling of the magnitudes of protrusion and depression in the expansion mechanism may account for the high variability in suction pressures that characterizes bamboo sharks as a functional generalist . Minor changes in head shape are estimated to produce noticeable changes in pressures , so increased protrusion/retraction of hyoid elements may lengthen the buccal cavity and alter pressure profiles during strikes, increasing or decreasing pressure, or shifting the timing of peak pressure according to possible prey items. Variation in the kinematic variables driving buccal expansion and suction mechanics in bamboo sharks may be one of the main characteristics that allow to adjustments their highly specialized feeding strategy ) for a wide variety of prey . Further research focusing on the suction performance of elasmobranchs should incorporate three dimensional kinematic reconstructions of the hyoid elements with changes in oral volume and in-vivo measurements of pressure  Long-axis rotation may also have broader implications for estimates of loading and force distribution across and between elements during suction capture events.
Long-axis rotation may transmit compressive or tensile forces from the elements to the ligamentous connections that support and connect the feeding apparatus of bamboo sharks and other suction feeding shark species Motta and Wilga 1999), reducing the load during suction events. A redistribution of forces could have particular significance for the hyomandibula as it is exceptionally stiff compared to other shark species and is expected to be under exceptional load (Balaban et al. 2014).
Individuals varied considerably in amount of rotation of the hyomandibula and ceratohyal. Variation in motions of these elements may also explain the variation in pressure from other bamboo shark feeding studies . This variation was less apparent in translation of the distal ends of the same elements, so it is possible that differences in motions between individuals are adjustable depending on position or type of prey, but the overall suction mechanics constrain the changes in dimensions of the oral cavity to ensure sufficient flow for capturing and keeping prey.

CONCLUSION
Bamboo sharks are dietary generalists, despite capturing prey almost entirely through suction. The general diet of bamboo sharks has been attributed to their ability to vary suction performance. Suction is generated by movement of the hyoid arch. This study describes additional degrees of motion beyond those previously attributed to suction.
These motions vary considerably in magnitude and may explain the high variation in suction performance. Figure 1: Proposed motions of the hyoid arch, including the hyomandibula (dark grey) and ceratohyal (light grey) of a white-spotted bamboo shark during a feeding event , and . A) Cartilaginous elements of the head, in left lateral view. B) Cartilaginous elements at maximum gape and hyoid depression. The distal end of the ceratohyal moves ventrally and caudally (thick black arrows 1) as the proximal end moves rostrally and medially (black arrow 2). C) Anterior view of the cranium and hyoid elements in resting position and D) depressed position, with the arch expanding ventrally (black arrows 1) and constricting medially (black arrows 2

Author Contributions
Bradley Scott is responsible for the introduction and discussion of the study and analysis of the trials, with input and commentary from the other authors. Implantation, anaesthesia, feeding, and contributions to the written components of the study came from Dr. Wilga. Dr. Brainerd ran the X-ray equipment, and contributed the experimental protocol used to record, track and analyze motions of bamboo sharks, as well as the written contributions to description of methods and results of the study.

Summary Statement
A precise three-dimensional reconstruction of jaw motions in a suction feeding shark.

Abstract:
The white-spotted bamboo shark is a generalist suction-feeding elasmobranch, capable of generating exceptional subambient pressures within its mouth. The main mechanism used to generate this drop in pressure is an expansion of the hyoid arch; Medial translation of the jaw joint occurs following maximum gape and after prey capture, and too late to enhance the protrusion of the upper jaw. Pronation of the lower jaw is likely too low in magnitude to change buccal volume or contribute to suction.
Instead pronation may be a result of extreme stress from activation of the hypertrophied jaw adductors of bamboo sharks as the jaws are closed.

Introduction:
Aquatic vertebrates use a numerous feeding strategies to obtain prey. Methods range from ram, filter and biting to suction feeding (Motta and Wilga 2001). The mechanisms underlying different feeding strategies vary based on the underlying morphology and the range of motion. The presence of feeding mechanisms can be tested by comparison with kinematic data for various taxa. In elasmobranchs (sharks, skates, and rays) the feeding strategies, and underlying mechanisms, are varied, but share constraints on the number of elements in the feeding apparatus (Motta and Wilga 2001;. Much of the flexibility in feeding strategy may come from a decoupling of the motions of the mandibular and hyoid arches (Motta and Wilga 1999;. The jaws open and close the mouth, biting or processing prey, while the hyoid either protrudes the jaws or generates negative pressure inside the mouth (Motta et al. 1997;Wilga and Sanford 2008).
Suction feeding elasmobranchs use powerful hyoid motions to expand the oral cavity ventrally and generate powerful suction forces in front of the mouth. While motions of the hyoid arch have been studied due its importance for generating suction forces Wilga and Sanford 2008;; Chapter 1), threedimensional motions of the mandibular arch in suction feeding sharks are less wellunderstood . White-spotted bamboo sharks, Chiloscyllium plagiosum, are specialized for suction capture, yet have a generalized diet . Additional degrees of motion of the jaws may allow a broader range of mechanisms for different prey. While timing and magnitude of protrusion and depression of the jaws has been measured in other studies (e.g. Wilga and Sanford 2008), medial translation of the jaw joint, a possible contributor to jaw protrusion, and long-axis rotation, a possible mechanism for grasping prey, have not been examined. Here X-ray Reconstruction of Moving Morphology (XROMM) is used to measure additional degrees of motion including medial translation of the jaw joint ) and longaxis rotation of the upper (Frazzetta and Prange 1987) and lower jaws .
The mandibular arch of elasmobranchs is suspended from the cranium as a palatoquadrate and Meckel's cartilage, supported by the hyomandibula and connected to the hyoid arch at the hyomandibular joint ( Fig. 1A; Motta and Wilga 1999;. In elasmobranchs suction force is primarily generated by ventral expansion of the hyoid arch Wilga and Sanford 2008;). This contrasts with teleosts, where the hyoid elements are closely connected to the upper jaws , and protrusion of the upper jaws contributes considerably to suction force at the mouth (Holzman et al. 2007). Among elasmobranchs, feeding initiates when the lower jaw depresses, then the jaws protruded to close rapidly (Fig. 1B, C; ), but jaw protrusion contributes little to suction forces (Motta et al. 1997;. Instead protrusion of the jaws reduces the time required to close the mouth following prey capture ).
The white-spotted bamboo shark is a suction-feeding generalist ) that eats a wide variety of prey items such as small fish and crustaceans . Capable of strong suction forces for capturing prey, but also highly variable in the timing and magnitude of pressures and kinematics during feeding Wilga and Sanford 2008;; Chapter 1), Bamboo sharks may use this flexibility in feeding to adjust for different prey or feeding strategies, such as using substrate to enhance suction forces (Nauwelerts et al. 2007) and adjusting the orientation of their teeth during prey capture or processing . To accommodate different prey items, the threedimensional motions of the upper and lower jaws of bamboo sharks may be more complex than depression and protrusion. Erection of the teeth during prey capture is hypothesized to activate through tension of the dental ligament . The dental ligament can be tensed by supination of the lower jaw . As the ventral margin of the jaws rotate medially, the labial margin rotates laterally, here this is termed supination, after the motion of turning the hands palm up or forward. The opposite rotation around the same axis is pronation. Lateral rotation of the labial margin of the lower jaw pulls the dental ligament by increasing the labial area of the lower jaw, resulting in tension . Longaxis rotation of mandibular elements has not been reported for any elasmobranch species, but bite-feeding sharks may utilize a similar mechanism for the upper jaw during protrusion (Frazzetta and Prange 1987). Long-axis rotation can have considerable effects on maneuverability and orientation of elements in actions as disparate as terrestrial locomotion (Kambic et al. 2014) and the hyoid elements of bamboo sharks (Chapter 1).
Other orectolobiform shark species medially constrict their jaw joints during protrusion, a mechanism proposed to enhance the magnitude of jaw protrusion during prey capture . Bamboo sharks may use a similar mechanism during feeding . The jaws are connected at the symphysis (distal end) by flexible cartilage rather than fused or interdigitated . This allows the jaws to rotate. As the jaw joint moves medially, the angle between the jaws is reduced and the jaws are thought to protrude rostrally as well as ventrally (Fig. 1D, E ;. If the jaws are protruded as the jaw joints move medially, then the distal end of the upper jaw should move further forward than the jaw joint at the same time.
The purpose of this study is to describe the three-dimensional motions of the upper and lower jaws during food capture in the White-spotted bamboo shark. In particular to record the long-axis rotation of the jaws and the medial translation of the jaw joint. As predicted, the upper and lower jaws compress medially at the jaw joint, and exhibit long-axis rotation during feeding. As the lower jaw depresses the ventral margin rotates medially, supinating. As the jaws close and protrude the ventral margin of the lower jaw rotates laterally, pronating. Medial translation occurs as the jaws are protruded  following peak gape, but does not appear to contribute to additional rostral protrusion of the jaws (contrary to .

Methods and Materials: Implantation
Four conical tungsten carbide markers (Kambic et al. 2014) were implanted in the chondrocranium, and three markers in each of palatoquadrate, Meckel's Cartilage, hyomandibula, ceratohyal from the left side (Fig. 2) in three individual white-spotted bamboo sharks (Chiloscyllium plagiosum Bennett) of similar size, with body lengths of 78.6 cm 79.2 cm 85 cm, and cranium lengths of 6.47 cm, 6.35 cm, 6.58 cm at the University of Rhode Island Graduate School of Oceanography. General anaesthesia was applied using 0.033 gL -1 of MS-222 Tricaine methanesulfonate in a recirculating fish anaesthesia system  for implantation and CT scan procedures. All sharks recovered and fed normally following implantation. Individuals were fed in a separate transparent tank to acclimate them for experimental trials. Food was withheld for one week prior to trials to encourage feeding.

CT Scan
Individuals were scanned following the first day of trials using a FIDEX CT scanner (Animage). Scans were taken at 448 X 448 pixels (416 X 416 pixels for Bamboo 3) resolution and a slice thickness of 0.185 mm. Images were compiled and reconstructed in Osirix (Pixmeo, Geneva, Switzerland). Three-dimensional surface meshes of the cranium, Meckel's, palatoquadrate, hyomandibula, and ceratohyal cartilages and as well as the markers were exported as .obj files to Maya (2015; Autodesk). The centroid for the vertices of each marker was calculated using the vertAvG tool in the XROMM_tools shelf for Maya (2015; Autodesk) available at xromm.org. Centroids of markers were exported from Maya as CT coordinates using the CTex tool.

XROMM
Feeding trials were recorded at the Keck XROMM Laboratory, Brown University.
Each individual was recorded feeding on a piece of squid or herring for each trial.
Feeding events were recorded on a pair of Phantom v.10 high speed video cameras.
Each camera had a fluoroscopic image intensifier and a corresponding X-ray emitter (Imaging Systems and Service; . Food items were placed in front of the shark and recording began when subject moved towards the food; it ended when the subject was finished processing or after ten seconds, whichever occurred first. The motion of implanted markers were recorded at 320 Hz or 330 Hz, with X-ray energies of 110-120kV and 100mA. CT coordinates of each marker were imported into XMALab (open-sourced software authored by Ben Knörlein; bitbucket.org/xromm/xmalab). Marker positions in both videos from each trial were tracked using XMALab and the three-dimensional translations of markers in each element were used to calculate translations and rotations of rigid bodies . Rigid body transformations were imported into Maya and applied to 3D meshes of each element imported from the CT scans.

Joint Coordinate System Analysis
The rotations of each element in each trial were measured in Maya using a Joint Coordinate System (JCS, Brainerd 2015 Kambic et al. 2014; jointAxes tool, XROMM_tool_shelf for Maya, xromm.org). Two sets of three (x, y, z) axes were aligned to the elements. The X-axis was aligned with the long-axis of the element pointing distally, the Z-axis was aligned horizontal to the cranium and pointed medially (Fig. 3). The Y-axis is aligned perpendicular to the other two axes, pointing approximately dorsally (Fig. 3). Here axes were centered at the quadratomandibular joint. One set of axes tracked the motion of the cranium, the other tracked the motion of the distal element, the Meckel's cartilage (Fig. 3A) or the palatoquadrate (Fig. 3B), and the relative difference in motion between the sets of axes was recorded. The calculated difference in motion between the two axes gave the motion of lower jaw or upper jaw relative to the cranium (Fig. 3). Timing and magnitude of the peak rotation along each of the three axes was recorded. Timings were recorded relative to peak gape. Magnitudes were recorded as relative to resting position. Resting position was recorded at a time prior feeding, where the ceratohyal was neither elevating nor depressing, or the initial frame of the trial if feeding began shortly after the start of filming. A total of 12 trials were analyzed for three individuals; four trials per individual.

Anatomical Coordinate System Analysis
Translation data were obtained using an Anatomical Coordinate System (ACS, Gidmark et al. 2012) in Maya, applied to the same trials as the JCS. A single set of axes was aligned with the cranium. As in Chapter 1 the Y-axis pointed rostral, the Z-axis pointed dorsally, and the X-axis pointed medially (relative to the left side cranial elements) at right angles to each other. Locators were placed on the jaw joint and caudal angle of the Meckel's cartilage; and the distal end of the palatoquadrate. On the palatoquadrate rostral and ventral translation was measured for the distal locator. Medial, rostral and ventral translation were recorded for the locator on the jaw joint of the Meckel's cartilage ( Table 2). The medial translation of the caudal angle of the Meckel's cartilage was also measured ( Table 2). All translation was relative to the cranium.

Feeding Sequence
Food was captured by inertial suction and transported to the oral cavity in a single motion for all trials analyzed except one. For the single exception the food item was drawn to the edge of the mouth, grasped, and then transported in a later motion.
While the magnitude of hyoid motions of that trial were much lower than other trials (Chapter 1) no obvious impact to the mandibular arch motions was observed. Cranial elevation was present prior to feeding in some trials to position the head in front of the food; however, little or no cranial elevation was observed during feeding strikes.
Opening of the mouth appears to begin earlier in these trials compared to other studies of kinematics in bamboo sharks Wilga et al. 2008). This is due to a difference in definition. Initiation of the expansive phase typically begins when gape is 10% of maximum ; however, some motions, such as supination of the lower jaw, begin as the jaws start to open and prior to 10% of maximum gape, so here the initial jaw depression is defined as the last jaw closure before maximum gape.
Feeding was initiated with depression of the lower jaw and opening of the gape (Fig. 4,. The lower jaw supinated as it depressed (most easily seen in Fig.   4B, C at -18 ms). The ventral margin of the Meckel's cartilage moved medially. Near peak gape the jaw joint began to protrude, moving rostrally, as well as ventrally and medially ( Fig. 4A at -18 ms). The upper jaw also began to pronate, the ventral margin rotating laterally (Fig. 4 0 ms). Protrusion of the jaw joint coincided with the whbeginning of depression and retraction of the ceratohyal and adduction and protrusion of the hyomandibula (Fig. 4A, -18 -0 ms, and Chapter 1 fig. 4). Peak depression of the Meckel's cartilage did not always coincide with peak gape, as depression of the palatoquadrate began at varied times among trials. At or slightly before peak gape pronation of the palatoquadrate peaked and the distal end of the palatoquadrate began to translate rostrally and ventrally as it the upper jaw depressed (Fig. 4, 0 ms). As the jaws closed the distal end of the hyomandibula fully protruded rostrad, depressed and adducted, while the ceratohyal fully retracted and depressed

Rotations from joint coordinate systems (JCSs)
Long-axis rotation of the upper and lower jaws can be seen as slight medial and lateral displacements of the ventral margins relative to the jaw joint in ventral and anterior views (Fig. 4B, C), but the motions are best quantified using JCSs (Fig. 5).
Supination of the lower jaw is smaller than pronation, with means (±s.e.m) of -8.0±1.1 deg and 11.2±1.5 deg respectively ( Fig. 5A and Table 1). Pronation of the upper jaw (12.3±0.9 deg) is slightly greater than pronation of the lower jaw (Table 1). The depression of the lower jaw is the largest rotation of the jaw joints at -22.7±0.6 deg., but depression of the upper jaw is also quite high, -15.7±1.8 deg (Fig. 5, Table 1).
Motions of the mandibular arch are much more sequential compared to the hyoid arch, where majority of the motion happens at the same time (Chapter 1). Peak supination of the lower jaw at -19±7 ms is followed by peak pronation of the upper jaw 0±1.6 ms, then depression of the lower jaw (6±2 ms), depression of the palatoquadrate (55±5 ms) and finally pronation of the lower jaw at 87±6 ms (Fig. 5, Table 1).

Translations of points in an anatomical coordinate system (ACS)
Rotations of the mandibular arch elements resulted in translation of the jaw joint, the caudal angle of the lower jaw, and the distal end of the upper jaw ( Fig. 6 and Table 1 generally quite low (1±0.7 mm) and the timing is considerably earlier than the medial translation of the jaw joint, 7±0.2 ms after peak gape compared with 54±4 ms after peak gape.

Individual Variation
Timing and magnitude of showed considerably variation among the individuals, particularly the medial translation of the jaw joint and long-axis rotation of the upper and lower jaws ( Table 1) Individual Bamboo 1, achieved peak supination earlier than the other two individuals, and peak pronation later than Bamboo 2 (Table 4). Bamboo 1 also had later medial translation of the lower jaw joint and depression of the upper jaw compared to Bamboo 2, but not significantly later than Bamboo 3, who was typically in between the other individuals in terms of timing (Table 4) (Table 4).
The relationship between medial translation of the jaw joint and rostral protrusion was also tested. If the medial translation of the jaw joint increased rostral protrusion of the jaws, then differences in rostral translation between the distal end of the upper jaw and the jaw joint should have occurred at the same time as medial translation. Peak medial translation occurred as the jaws closed, while the peak difference in rostral protrusion occurred at maximum gape (p-value<0.001).

DISCUSSION
The elements of the mandibular arch have highly complex three-dimensional motions during feeding. In addition to opening and closing the jaws, and ventral depression of the upper jaw, white-spotted bamboo sharks, Chiloscyllium plagiosum, use long-axis rotation of the jaws to assist in prey capture. Long-axis rotation of the upper and lower jaws times consistently with prey capture. Medial translation of the jaw joint is present; however, it occurs after prey capture and does not appear to contribute to rostral protrusion of the jaws. Despite specializing in suction feeding, bamboo sharks appear to have additional mechanisms for capturing prey increasing the range of prey types that they can consume.
Supination peaks prior to peak gape, consistent with the timing predicted for aiding prey capture by tensing the dental ligament and erecting the teeth . Following peak gape the lower jaw pronates, swinging the caudal angle of the meckel's cartilage medial to lateral relative to the jaw joint. While pronation of the lower jaw could improve suction performance by expanding the buccal cavity, as the caudal angle moves laterally, pronation and lateral translation of the caudal angle have extensive variation in timing as well as magnitude (Fig. 6A, Tables 1 and 2). In addition the caudal angle moves laterally during supination, while during pronation the caudal angle moves medially (Fig. 6A), the opposite motion that would be expected if pronation expanded the buccal cavity. Medial translation of the caudal angle coincides with medial translation of the jaw joint, but is of lower magnitude. As the jaw joint moves medially pronation of the lower jaw may act to reduce the compression of buccal cavity during feeding, rather than increase total volume. Instead of increasing volume pronation of the Meckel's cartilage may be a byproduct of the force from the hypertrophied jaw adductors of C. plagiosum (Ramsay and Wilga 2012;). The muscles insert along the caudal and ventral margins of the Meckel's cartilage and activate during compression to close the mouth (Motta and Wilga 1999;. With sufficient force they may pull the ventral margin laterally relative to the labial margin of the lower jaw. The pronation of the upper jaw at peak gape may have a similar function to lower jaw supination. As the upper jaw pronates the labial margins move apart from each other, potentially tensing the dental ligament and erecting upper teeth as well (Frazzetta and Prange 1987). This would occur shortly after supination in the lower jaw, as the palatoquadrate depressed, where it could come into contact with larger prey items. Other sharks may use a similar mechanism for erecting teeth despite broad ecological differences. Pronation during jaw protrusion has been hypothesized for bite-feeding Carcharinid shark species for the same role (Frazzetta and Prange 1987) despite distinctly different ecologies compared with to bamboo sharks . While the timing of long-axis rotation of the upper and lower jaws is consistent with grasping prey , erection of teeth during supination was not examined, and food was grasped during prey capture in only one trial, while longaxis rotation was found in all trials. As with direction and sequence of kinematics for other cranial elements during feeding Wilga and Sanford 2008), the sequence of long-axis rotation may be highly consistent during feeding events, but the duration and magnitude may be highly variable even within the same species ; Chapter 1) Assessing significance of supination and long-axis rotation of elements of the mandibular arch for the feeding mechanics and ecomorphology of suction feeding shark species requires further comparison with other elasmobranchs. Morphological correlates of supination and pronation may include a loosely connected symphysis as well as a distinct angel between the distal ends of the jaws that allows the ventral margins of the lower jaw to rotate medially without abutting  Food items were kept small during trials to maximize number of feeding events before satiation, but if long-axis rotation improves prey capture by increasing the ability of teeth to grasp , then larger food items may induce a greater, or more stereotypical response. Younger bamboo sharks also use less suction and more locomotion when feeding (Lowry and . As bamboo sharks grow in size suction becomes the dominant component of feeding (Lowry and . Grasping prey may be a more significant component of the feeding strategy of young bamboo sharks until suction performance alone is sufficient to capture prey items, and as such long-axis rotation of the jaws may have greater significance for success of juvenile bamboo sharks than for adults. For this study body size varied little among the individuals, allowing differences in feeding kinematics to be associated for factors other than ontogeny, but, if long axis rotation, particularly supination, is of greater importance earlier in ontogeny then magnitude of long-axis rotation would be expected to scale negatively with body size. Pronation of the lower jaw, if related to suction performance may increase in magnitude later in ontogeny, as bamboo sharks transition to a more suction dominated feeding strategy (Lowry and . Magnitude of pronation of the lower jaw may be small, but it starts at approximately the same time as peak buccal pressure, and peak rate of increase in hyoid area . Estimation of change in buccal volume of C.
plagiosum during feeding could allow a more accurate assessment of the impact of pronation on changes in overall buccal volume and suction forces.
Medial translation of the jaw joint compresses with the hyoid cavity , during and following peak gape, when food is typically already captured  and peak buccal pressure has passed . Medial translation of the jaw joint also comes after the peak increase in rostral translation of the distal end of the upper jaw relative to the jaw joint (Fig. 6C). The distal end of the upper jaw protrudes rostrally more than the jaw joint at peak gape in all but one trial, but the medial translation of the jaw joint does not peak until well after peak gape, delayed by approximately 50 ms ( Table 2).
Because of this difference in timing it is unlikely that medial translation of the jaw joint protrudes the jaws rostrally. Rather, given the similarity in timing to the pronation of the lower jaw, medial translation of the jaw joint may be the result of pronation, through medial rotation of the labial margin of the lower jaw as the jaws close. The presence of medial translation of the jaw joint in other suction feeding orectolobiform sharks  indicates that the long-axis rotations of the lower jaw observed here may be widespread within the Orectolobiformes, a group that includes numerous suction feeding specialists and that feeds on a variety of prey in complex environments . Rostral translation of the jaws also occurs too late to contribute to grasping of prey or suction performance. Depression of the upper jaw reduces the time required to close the mouth, closing gape before the lower jaw is fully elevated (Fig. 5A, the blue depression trace is still negative at end gape, i.e. not returned to resting position) as in other elasmobranchs . Rostral protrusion of the jaws may also extend the buccal cavity to prevent escape of prey once captured.
Variation in mandibular arch motions was greater between Bamboo 1 and 2 than 3, strikingly different from the motions of the hyoid arch, where differences in long-axis rotation of the ceratohyal resulted in greatest variation being between Bamboo 3 and the other two individuals. Differing trends in variation between hyoid and mandibular arches may result from their independent depression mechanisms. The independence of depression of the mandibular and hyoid arches of bamboo sharks allows for greater variety in the timing and magnitude of strike motions, while a consistent sequence of motions is retained . This may allow individuals to adjust for various environmental effects including substrate (Nauwelearts et al. 2007) and a wide variety of prey items ) that they may encounter.

CONCLUSIONS
This study shows that the elements of the mandibular arch undergo complex threedimensional motions during feeding, including long-axis rotation of the upper and lower jaws. In addition to varying suction performance, bamboo sharks may use longaxis rotation of the upper and lower jaws as additional mechanisms for obtaining a variety of prey. By rotating the jaws around their long-axes the teeth may be erected to aid in grasping of soft bodied prey. Alternatively, long-axis rotation may improve suction performance by making the inside of the mouth more conical. Medial motion of the jaw joint does not contribute to prey capture and may be the result of long-axis rotations of the jaws during feeding.

Author Contributions
Bradley Scott is responsible for the introduction and discussion of the study and analysis of the trials, with input and commentary from the other authors. Implantation, anaesthesia, feeding, and contributions to the written components of the study came from Dr. Wilga. Dr. Brainerd ran the X-ray equipment, and contributed the experimental protocol used to record, track and analyze motions of bamboo sharks, as well as the written contributions to description of methods and results of the study.

Funding
Funding was provided by the National Science Foundation, NSF grants IOS-1354189 to Cheryl Wilga and DBI-1262156 to Elizabeth Brainerd, as well as Graduate Teaching Assistantships from the University of Rhode Island.