Deploy the proboscis!: Functional morphology and kinematics of a novel form of extreme jaw protrusion in the hingemouth, Phractolaemus ansorgii (Gonorynchiformes)

Abstract

Premaxillary protrusion and the performance advantages it confers are implicated in the success of diverse lineages of teleost fishes, such as Cypriniformes and Acanthomorpha. Although premaxillary protrusion has evolved independently at least five times within bony fishes, much of the functional work investigating this kinesis relates to mechanisms found only in these two clades. Few studies have characterized feeding mechanisms in less‐diverse premaxilla‐protruding lineages and fewer yet have investigated the distinctive anatomy underlying jaw kinesis in these lineages. Here, we integrated dissection, clearing and staining, histology, micro‐CT, and high‐speed videography to investigate an isolated and independent origin of jaw protrusion in the hingemouth, Phractolaemus ansorgii, which employs a complex arrangement of bones, musculature, and connective tissues to feed on benthic detritus via a deployable proboscis. Our goals were to provide an integrative account of the underlying architecture of P. ansorgii's feeding apparatus and to assess the functional consequences of this drastic deviation from the more typical teleost condition. Phractolaemus ansorgii's cranial anatomy is distinct from all other fishes in that its adducted lower jaw is caudally oriented, and it possesses a mouth at the terminal end of an elongated, tube‐like proboscis that is unique in its lack of skeletal support from the oral jaws. Instead, its mouth is supported primarily by hyaline‐cell cartilage and other rigid connective tissues, and features highly flexible lips that are covered in rows of keratinous unculi. Concomitant changes to the adductor musculature likely allow for the flexibility to protrude the mouth dorsally and ventrally as observed during different feeding behaviors, while the intrinsic compliance of the lips allows for more effective scraping of irregular surfaces. From our feeding videos, we find that P. ansorgii is capable of modulating the distance of protrusion, with maximum anterior protrusion exceeding 30% of head length. This represents a previously undescribed example of extreme jaw protrusion on par with many acanthomorph species. Protrusion is much slower in P. ansorgii—reaching an average speed of 2.74 cm/s—compared to acanthomorphs feeding on elusive prey or even benthivorous cypriniforms. However, this reorganization of cranial anatomy may reflect a greater need for dexterity to forage more precisely in multiple directions and on a wide variety of surface textures. Although this highly modified mechanism may have limited versatility over evolutionary timescales, it has persisted in solitude within Gonorynchiformes, representing a novel functional solution for benthic feeding in tropical West African rivers.

We describe a novel form of jaw protrusion in Phractolaemus ansorgii, a fish that feeds with a deployable proboscis‐like snout that can be protruded to a distance greater than 30% of its head length. This proboscis is unique from the feeding apparatuses of other fishes; it features a dentary that is rotated nearly 180 degrees in the head and an oral aperture that is supported by flexible hyaline cartilages rather than bone, which may be advantageous for feeding along irregular benthic substrate.

1. INTRODUCTION

The ability to protrude the upper jaw, or premaxilla, toward prey has evolved independently in at least five lineages of ray‐finned fishes, with each origin requiring an amalgam of modifications to the ancestral actinopterygian skull (Wainwright et al., 2015). The ancestral condition featured few mobile elements, tight ligamentous connections, and a small premaxilla firmly attached to the neurocranium (Schaefer & Rosen, 1961; Westneat, 2004). Increased premaxillary mobility has been implicated in the success of some of the most diverse groups of teleosts including cypriniform and acanthomorph fishes that together represent one‐third of all vertebrate species (Figure 1a) (Bellwood et al., 2015; Motta, 1984; Nelson et al., 2016). The evolution of premaxillary protrusion has long fascinated functional morphologists as it represents multiple mechanical solutions to the same ecological challenge—that of feeding in a viscous, fluid medium (Lauder, 1982; Westneat, 2004). Numerous functional studies have demonstrated a suite of ways in which premaxillary protrusion contributes to resource acquisition in midwater and benthic suction feeders as well as in more specialized foraging behaviors such as “picking” that is observed in Cyprinodontiformes (Ferry‐Graham et al., 2007; Staab, Ferry, & Hernandez, 2012; Staab, Holzman, et al., 2012; Wainwright & Day, 2007). Most teleosts employ some combination of ram and suction feeding, both of which are augmented by anterior jaw protrusion. For example, premaxillary protrusion doubles ram distance and increases ram velocity in some predatory cichlids (Wainwright et al., 2001; Waltzek & Wainwright, 2003), and increases the total hydrodynamic force exerted on the prey in sunfish (Holzman et al., 2008).

FIGURE 1.

(a) Five independent origins of jaw protrusion within bony fishes, demarcated by stars. Note that stars do not indicate an ancestral state, only that jaw protrusion has evolved at least once within this group. Red stars on Cypriniformes and Acanthomorpha indicate that jaw protrusion is coincident with large species radiations. This phylogenetic tree is modified from Hughes et al. (2018). To the right, micro‐CT scans of skulls of representative gonorynchiform (b) and cypriniform (c) species, emphasizing bones of the oral jaws and position of the jaw joint. Dent, dentary; mx, maxilla; pmx, premaxilla.

Jaw protrusion has long been hypothesized to enhance the effectiveness of prey capture and may have evolved and persisted in multiple lineages because it improves trophic performance (Wainwright et al., 2015). Most of our understanding of how fishes effect premaxillary protrusion and what performance advantages it confers come from studies of select cypriniform and acanthomorph fishes, two incredibly speciose and ecologically diverse clades that have evolved divergent mechanisms to protrude the upper jaw. In both groups, jaw protrusion represents a trophic novelty involving a rearrangement of musculoskeletal and ligamentous architecture that increases the mechanical complexity of the feeding apparatus and imparts enhanced feeding performance and greater kinematic flexibility. These rearrangements, as with many morphological novelties, are posited to have permitted new functions or the expansion of existing functional repertoires throughout the evolution of these groups (Galis, 2001; Hernandez & Staab, 2015).

In contrast to Cypriniformes and Acanthomorpha, other origins of premaxillary protrusion are not correlated with high lineage diversification or ecological and morphological diversity (Wainwright et al., 2015). The independent origin of premaxillary protrusion in Characiformes is represented by a single genus with five species (Bivibranchia) (Gery, 1962). Another independent origin in Gonorynchiformes is represented by the single species Phractolaemus ansorgii—a freshwater fish that is endemic to West Africa and feeds on detritus along the benthos with a deployable, proboscis‐like snout (Figure 1) (Gery, 1962; Thys van den Audenaerde, 1961; Vari, 1985). There is a dearth of functional studies that investigate protrusion mechanisms beyond Cypriniformes and Acanthomorpha. Moreover, these studies have focused primarily on musculoskeletal architecture, overlooking the potentially important roles that other connective tissues play in the feeding apparatus. To better understand the morphological limitations imparted by the teleost skull and the demands of feeding in fluid, it is necessary to investigate these neglected structures and independent origins of protrusion. The isolated occurrence and lack of accompanying diversity of these origins belie their significance as novel solutions to a ubiquitous ecological challenge.

Gonorynchiformes is a species‐depauperate order of ostariophysan fishes that exhibits a disproportionately disparate array of craniofacial morphologies given its small size (37 species). Among the most enigmatic of its members is the hingemouth, Phractolaemus ansorgii, the only gonorynchiform capable of feeding via premaxillary protrusion. The cranial skeleton and oral jaws of P. ansorgii are quite distinct; thin and rodlike, the maxillae are curved posteriorly and attached to the small, triangular premaxillae (Thys van den Audenaerde, 1961). Even more unusual, however, is that the mandible is flipped in the skull, rotated nearly 180 degrees from the typical, rostrally oriented teleost position such that it is oriented caudally and its articulation with the quadrate is the anteriormost point of the head (Thys van den Audenaerde, 1961; Figure 1b,c—note relative position of jaw joint). Although the shape and orientation of the oral jaws are unique among teleosts, the complexities of P. ansorgii's feeding apparatus are not fully captured by this bizarre skeletal anatomy alone, which represents only one set of many modifications that have evolved to facilitate this novel mode of benthic feeding.

While the jaw protrusion mechanism employed by P. ansorgii is present in only a single extant species, investigating how a fish can feed with such a complex rearrangement of cranial structures provides an opportunity to investigate both the morphological and functional capacity of the teleost skull. Compared to research on cypriniform and acanthomorph fishes, there have been few anatomical studies of P. ansorgii and even fewer that present a robust functional hypothesis for how premaxillary protrusion is actuated (Diogo, 2010; Howes, 1985; Thys van den Audenaerde, 1961). Furthermore, previous hypotheses were developed in the absence of in vivo observation and were unable to account for many of the nuances and complexities of the feeding behavior that can be informed by kinematic data. This study is the first to provide three‐dimensional visualization of P. ansorgii's cranial anatomy as well as functional data from recorded feeding trials, both of which greatly supplement our ability to understand and appreciate this unusual reorganization of the cranial anatomy. In this study, we characterize the mechanism of jaw protrusion employed by P. ansorgii in three ways: (1) we describe a novel arrangement of the skin and connective tissues forming the deployable proboscis and the highly modified oral aperture, hereafter called the mouth, at its terminal end; (2) we use high‐speed video to quantify the kinematics of this novel feeding apparatus; and (3) we propose functional hypotheses for jaw protrusion and retraction based on musculoskeletal and ligamentous anatomy.

2. MATERIALS AND METHODS

2.1. Gross and functional anatomy

A total of 24 specimens (25–127 mm SL) were acquired through the aquarium trade or borrowed from the American Museum of Natural History (AMNH 33471, 71872, 217860, 240179, 246013, 257,165, and 274,663) and used to characterize musculoskeletal, ligamentous, and connective‐tissue anatomy associated with the feeding apparatus. Seven fresh and formalin‐fixed specimens were dissected to determine muscular and ligamentous anatomy. Three additional specimens were cleared and stained using the protocol presented in Dingerkus and Uhler (1977) with modifications by Potthoff (1984). Cleared and stained specimens were used to assess the shape and presence of bones and cartilaginous elements of the proboscis. For histological examination, whole heads of seven specimens were infiltrated with JB‐4 following an embedding media protocol from Electron Microscopy Science, sectioned at 2 μm, and stained with Lee's methylene blue‐basic fuchsin stain. Internal cranial anatomy of seven specimens was visualized using microcomputed tomography (micro‐CT) scans made at varying resolutions (10.3–49.6 μm) using a GE phoenix v|tome|x s240 dual‐tube 240/180 kV system (General Electric, Fairfield). Soft tissue of one of the seven specimens was contrast enhanced by staining for 36 h in 5% Lugol's iodine solution. Six specimens were fixed with their proboscises retracted, while one (AMNH 217860) was fixed with its proboscis partially protruded. CT scans were segmented in Materialise Mimics (version 20.0) and used to visualize and reconstruct the three‐dimensional musculoskeletal anatomy of the oral jaws.

Morphological analyses included characterization of the following structures: premaxilla, maxilla, dentary, anguloarticular, retroarticular, quadrate, and the muscles of the adductor mandibulae complex. We also describe several other connective tissues and ligamentous structures associated with the underlying architecture of the proboscis and movement of the oral jaws.

2.2. Animals and filming procedure

Live Phractolaemus ansorgii were acquired through the aquarium trade and housed communally with conspecifics in a 29‐gallon tank. Three individuals (33–59 mm SL) were trained to feed under experimental conditions in a separate 8‐gallon filming tank that had been fitted with a false bottom, two opaque acrylic dividers, and a rock to serve as refuge. During feeding trials and training, the tank was filled approximately one‐third of the way full to reduce the feeding arena to a 16 × 24.9 × 10 cm chamber. In addition to ambient lighting in the room, the tank was placed under five 60 W filming lights (Portable Luminaire Model LD1003, 120Vac, 60 Hz) to adequately illuminate the structures of interest. Prior to feeding trials, individuals were fasted for 48–72 h to avoid the effects of satiation (Sass & Motta, 2002). Individuals were then transferred to the feeding aquarium 30 min prior to the start of the feeding trial to allow for adequate acclimation to the low‐water and high‐light conditions. This time was determined from pilot trials, in which fish resumed normal behaviors within 30 min of being transferred to the tank. Fish were fed a liquid slurry of crushed algae pellets, spirulina powder, and frozen brine shrimp, which was delivered to the bottom of the tank with a pipette. Feeding trials were recorded at 1000 frames per second with two synchronized high‐speed cameras (Phantom VEO 340S, Vision Research Inc.) that simultaneously filmed dorsal and lateral views through Canon EF 100 mm f/2.8 L Macro IS USM lenses. All experimental and animal care procedures were approved by the Institutional Animal Care and Use Committee at The George Washington University (protocol #A035).

2.3. Kinematic analysis

A total of 26 feeding trials were recorded and used for either characterization of feeding behavior or for measuring kinematic variables related to the movement of cranial bones during feeding (n_Pa1 = 3, n_Pa2 = 16, n_Pa3 = 7). Detailed descriptions of feeding in P. ansorgii have yet to be published; thus, we define several pertinent terms here to characterize the observed behaviors. Frames for most videos were cropped to include only a single strike. We define a strike to be when the proboscis is deployed from a fully retracted state to a protruded state. Other videos were cropped to include multiple probes, which we define to be when the proboscis was not fully retracted following the initial strike but rather partially retracted and protruded again. Videos were imported into Fiji (version 2.15.0; Schindelin et al., 2012) to track the following landmarks of interest over the course of each feeding sequence: (1) anterior tip of the upper lip, (2) center point of the ethmoid, (3) posteriormost point of the opercle, and (4) position of expelled food particles (Figure S1). These four landmarks were used to generate 2D coordinates to calculate the following displacement and timing variables: (1) maximum upper jaw displacement (measured as the straight‐line distance between the tip of the ethmoid and the tip of the upper lip), (2) time to maximum upper jaw displacement, (3) protrusion speed, and (4) probes per second.

3. RESULTS

3.1. Musculoskeletal and ligamentous anatomy of the oral jaws

The oral jaws, by definition, constitute the premaxillae, maxillae, dentaries, and angulars. At rest, these bones lay tucked into a depression of the skull bordered posteriorly by the ethmoid, ventrally by the quadrate, and laterally and anteriorly by a pseudochin (Figures 2 and 3a). This pseudochin extends anterodorsally from the distal ends of the preopercles and is composed of fat and other loose connective tissues; this structure is hereafter referred to as the chin. The small, bilaterally paired premaxillae are triangular with a rounded medial edge, roughly tear‐drop shaped. They are very thin along their dorsoventral axis and extend laterally to about half the width of the maxillae. They are medially united by a thick ligamentous tissue rich in hyaline‐cell cartilage (Figure 3b asterisk). The premaxillae are inflexibly linked to the maxillae by ligamentous connective tissue. They lie posterodorsally to the tips of the maxillae such that these maxillary tips form the anteriormost point of the upper jaw (Figure 2b red asterisk) rather than the premaxillae, the condition most often seen in other teleosts. The maxillae are curved and sickle shaped with their convex aspect facing posteriorly toward the neurocranium (Figure 2a). They bow medially along their vertical axes and together create a wide, semicircular shape that forms the lateral border of a rounded buccal cavity (Figure 2c). The distal, dorsalmost aspect of the maxilla, ventral to the premaxilla, is quite flat and thin terminating in a rounded tip. The ventral aspects of the maxillae are far more broad and complex, bearing an anterior process for adductor muscle attachment that is distinctly cupular and concave (Figure 4a arrow). Further posterior, along the medial face of this ventral portion, a thin spine‐like projection extends dorsally (Figure 4c arrow). This maxillary spine is connected to the distal end of the lower jaw by connective tissue that wraps around the lateral face of the dentary.

FIGURE 2.

Micro‐CT scans of the skull of Phractolaemus ansorgii (AMNH 240179) in (a) lateral view with some skeletal elements (interopercles, orbitals) omitted to show anatomical structures of interest. (b) Dorsal and (c) frontal views cropped and enlarged to the same scale to emphasize oral jaws. Distal tip of maxilla marked by red asterisk.

FIGURE 3.

Dorsal view of the proboscis of Phractolaemus ansorgii, highlighting positions of the upper jaw bones, rictal cartilages, and cartilaginous ligament at rest and during protrusion. (a) Still from feeding video of a live fish at rest showing folded up proboscis and dorsally oriented mouth. (b) Illustration based on cleared and stained specimens showing position of bones and ligaments beneath the skin. Premaxillae united on their medial aspect by ligament rich in hyaline‐cell cartilage (asterisk). Cartilaginous ligament is folded beneath the skin of the proboscis (cl). (c) Still from feeding video of a live fish protruding its jaws demonstrating proboscis at maximum protrusion. (d) Illustration based on manipulated cleared and stained specimens showing position of bones and the attachments of the cartilaginous ligament when the proboscis is protruded. cl, cartilaginous ligament; lf, labial flap; mx, maxilla; pal, autopalatine; pch, pseudochin; pmx, premaxilla; rc, rictal cartilage; ul, upper lip; vom, vomer.

FIGURE 4.

Micro‐CT scans of the right half of the oral jaw elements of Phractolaemus ansorgii (AMNH 240179) with close‐up in (a) lateral, (b) medial, and (c) frontal (turned slightly laterally to show morphology of medial spine) views. (a) Cupular anteroventral process of the maxilla marked by arrow. (c) Arrowhead pointing to finger‐like process on medial face of the dentary, arrow pointing to medial spine on the maxilla, and bracket emphasizing the empty space between maxilla and dentary where the rictal cartilages sit. Oral jaws overlaid on video stills of a live individual with proboscis retracted (d) and protruded (e). Note that the micro‐CT images are from a museum specimen and not from the filmed specimen. Position of oral jaws was approximated by the position of the premaxilla (covered) and maxilla (asterisk), which is visible through the skin of the proboscis.

The lower jaw (dentary, retroarticular, and anguloarticular) is rotated ~180 degrees such that the mandible articulates with the quadrate at the anteriormost point of the head (Figures 2a and 4a). The anguloarticular is triangular with a ventral V‐shaped process that forms the jaw joint, abutting the quadrate. The retroarticular is small, rectangular, and sits anterior to the anguloarticular (Figure 4). As previously described in Howes (1985), it is connected to the interopercle via the retroarticular–interopercular ligament (Figure 5a). Unlike any fish previously described, the dentary extends posteriorly from this anteriorly displaced jaw joint, with its distal tip lying ventral to the ventral aspect of the maxilla. The large gap between the medial face of the maxilla and the lateral face of the dentary (Figure 4c, bracket) contains two round, discrete pieces of hyaline cartilage (rictal cartilages sensu Howes, 1985) embedded within hyaline‐cell cartilage as well as two divisions of the adductor mandibulae (Figure 5b). At rest, the distalmost tip of the dentary—from its articulation with the quadrate—lies posterior to the proximal tip of the jaw. Together, the dentaries form a broad V‐shape with their distal tips pointing laterally (Figure 2b,c). The dentaries seemingly maintain this V‐shape even while protruded, as exemplified by one scanned specimen, which was fixed with its proboscis partially protruded. There is no ossified mental symphysis. Proximally, the dentary bears a finger‐like process on its medial face (Figure 4b,c arrowhead). These processes point posteriorly and are connected to each other by a broad mass of hyaline‐cell cartilage that lies dorsal to the intermandibularis muscle. At rest, the dentaries do not lie parallel to the quadrate but are instead very slightly tilted anterodorsally (Figure 2a). The quadrate is a rectangular bone tapering anteriorly and composed of cartilage posteriorly (Figures 2a and 4a). Importantly, each quadrate bears a hook‐like process (called the quadrate spine, sensu Howes, 1985) at its anterior extremity near the jaw joint. This quadrate spine extends laterally and curves dorsally, forming a groove or tunnel through which adductor musculature passes (Figures 4c and 5a).

FIGURE 5.

Cranial musculature involved in movement of the oral jaws, featuring muscles of the adductor mandibulae complex and levator operculi. (a) The A1 division of the adductor mandibulae narrows into a tendon, passes through the tunnel formed by the quadrate spine, and inserts on the anteroventral aspect of the maxilla. The opercular series is connected to the mandible by an interopercular–mandibular ligament between the elongate interopercle and the retroarticular. (b) The A1o and A1i divisions of the adductor mandibulae attach to the maxilla and the rictal cartilage between the maxilla and dentary, respectively. (c) The A2 division of the adductor mandibulae has two divisions, A2α and A2β. A2α branches into two divisions, the dorsal of which invests the skin of the proboscis and the ventral of which wraps around the chin. The proboscis and mouth are outlined in light grey. Ch, chin; dent, dentary; iop, interopercle; ll, lower lip; LO, levator operculi; lri, retroarticular–interopercular ligament; max, maxilla; pmx, premaxilla; prob, proboscis; op, opercle; ul, upper lip.

Our examination of jaw musculature is in partial disagreement with what has been previously reported by Howes (1985) and Diogo (2010). Diogo's (2010) description of adductor musculature does not differ from that of Howes (1985), thus we refer only to Howes here. There are two prominent divisions of the adductor mandibulae complex, A1 and A2 (Figure 5a–c). As originally described by Howes (1985), the ventral division, A1, originates on the ventral aspect of the preopercle and runs anteriorly along the lateral face of the quadrate. As it approaches the quadrate spine, it narrows into a tendon that passes through a tunnel bound medially by the quadrate, laterally by the quadrate spine, and dorsally by a ligamentous connection between these two structures. However, we did not find that this tendon joins an aponeurosis from which the divisions A1i and A1o extend; instead, the tendon of A1 is separate from and unassociated with the muscles A1i and A1o. The tendon of A1 bends dorsally sharply as it exits the quadrate tunnel and continues dorsally to attach to the cupular anteroventral process of the maxilla.

The subdivisions of A1, A1i and A1o, are discrete bundles of muscle, thickest and broadest at their origin, and tapering dorsally adjacent to their insertion points (Figure 5b). The more superficial bundle, A1i, originates on the lateral face of the quadrate near the quadrate spine and wraps medially around A1o close to its dorsal insertion. A1i does not fully envelope A1o, remaining medial to cover and insert on the posterior aspect of the rictal cartilage that lies between the maxilla and the dentary. A1o seemingly originates within A1i, covering the connective tissue surrounding the distal aspect of the dentary as A1o continues dorsally to attach along the posterior aspect of the maxilla and above the broad ventral base. These observations differ from Howes (1985), in which A1o is illustrated as being continuous with the main body of A1 and described as having bundles that insert on the distal cavity of the maxilla and the connective tissue between the maxilla and premaxilla; the rest of muscle is described as covering “the rictal cartilages that lie ventrally and between the dentaries.” We did not observe A1o extending to the distal end of the maxilla or covering any structure ventral to the dentaries. Furthermore, we find the illustrations of these muscles as “a thin sheet of muscle fibers,” misleading; A1o and A1i are depicted as bifurcating from one another, instead of wrapped around each other, with A1o covering the anterior face of the maxilla and inserting onto the cupular anteroventral process of the maxilla and A1i attaching along the posterodorsal extreme of the maxilla.

Unlike Howes (1985), we find that the dorsal and ventral subdivisions of A2 are readily distinguishable from one another as discrete muscle bundles originating on the preopercle, dorsal to A1 (Figure 5c). The dorsal and more superficial subdivision of A2, A2α (sensu Winterbottom, 1974), continues anteriorly along the preopercle and runs deep to the lateral ethmoid. On the anterior side of the lateral ethmoid, A2α bifurcates into tendinous dorsal and ventral branches. The dorsal branch terminates in many small threadlike tendons that invest the skin of the proboscis. The ventral branch of A2α continues anteriorly as a tendon, running through the flexible connective tissue of the chin. This tendinous ventral branch of A2α runs all the way around the chin and is continuous with the A2α muscle that originates on the other side of the head. These findings are novel to our study, whereas Howes (1985) describes the dorsal subdivision of A2 as inserting on the anterior face of the antorbital and the connective tissue covering the rictal cartilages. Consistent with Howes, the ventral subdivision of A2, A2β, narrows into a tendon that attaches to the coronomeckelian bone on the anteromedial aspect of the dentary (Figure 5c). However, we did not observe the two tendinous divisions of A2β Howes described.

We confirm the presence of a cartilaginous ligament (pseudocartilaginous ligament sensu Thys van den Audenaerde, 1961; X‐shaped ligament sensu Howes, 1985) that is divided into three bilaterally paired branches (Figure 3c,d). The posteriormost pair of branches are attached to the horns of the vomer, the middle pair extend laterally to the autopalatines (ectopterygoid sensu Thys van den Audenaerde, 1961), and the anteriormost pair are found between the muscles A1i and A1o inserting on the rictal cartilages between the maxillae and the dentaries (Figure 3c,d). This ligament folds up on itself at rest and stretches taut during maximum protrusion (Figure 3c,d). It is rich in hyaline‐cell cartilage along its length and contains a longitudinal series of three thin rods of hyaline cartilage in each of its anteriormost branches (Figure 3d, 6c).

FIGURE 6.

Histological overview of the mouth in Phractolaemus ansorgii. (a) Image of a live fish with dashed lines depicting the region examined (sagittal section). (b) Upper lip supported by a thin band of dense connective tissue (ct) and lined with keratinous unculi (unc). Sensory cells (sc) present on the ventral aspect of the upper lip. (c) At rest, the folded skin of the proboscis (asterisk) lies over the maxilla (mx) and premaxilla (pmx). Ventral to the maxilla is the cartilaginous ligament rich in hyaline‐cell cartilage (hcc) with a lump of hyaline cartilage embedded within (hc). (d) Pseudochin (Pch) composed of loose connective tissue with goblet and sensory cells (gc and sc, respectively) studded along the outer epithelium. Lower lip (Ll) composed of hyaline‐cell cartilage (hcc). Folded skin of the proboscis marked by asterisk. (e) Tip of lower lip featuring tall uncular row.

3.2. Anatomy of the mouth and proboscis

Here, we use the term “mouth” to refer to the oral aperture at the end of the proboscis. Importantly, the position of the jaws in relation to the mouth is modified such that neither the premaxillae nor dentaries border the mouth opening (Figures 3d and 6b,d), as seen in all other fishes previously described. Instead of skeletal elements, the mouth is bordered by an upper and lower set of pseudolips composed of cartilages and other connective tissues, hereafter referred to as the lips. The upper lip forms the dorsal border of the mouth opening (Figure 6b). Here, the maxillae and premaxillae sit posteriorly to a flat rod of dense connective tissue that provides rigid support to the distal extreme of the upper lip; at rest, there is a gap between the distal end of this connective tissue and the skin of the upper lip (Figure 6b). Although the greatly reduced premaxillae abut this connective tissue, they do not extend to the margins of the mouth opening, nor do the maxillae (Figures 3d,e and 6a–c). Sensory cells are embedded along the ventral surface of the upper lip, but we do not find sensory cells on any surface of the lower lip (Figure 6b,d). The lower lip has no bony support from the lower jaw. In fact, the dentaries do not form any part of the mouth opening (Figure 6d), and we note again that the distal end of the dentary, which in most teleosts forms the ventral border of the oral aperture, is posterior to the mouth opening and connected by ligaments to the ventral aspect of the maxilla (Figure 4d,e). All structural support of the lower lip is instead provided by large amounts of hyaline‐cell cartilage (Figure 6d).

The upper and lower lips are lined with transverse rows of keratinous unculi (chitinous denticles sensu Thys van den Audenaerdae, 1961), which are unicellular horny projections that superficially resemble curved and conical teeth (Figure 6b,d,e) (Roberts, 1982). These unculi cover the dorsal and ventral surfaces of the lips. In the upper lip, they are largest and most conical at the distal extreme, appearing thinner and shorter along the proximal aspect. This pattern is seen again in the lower lip but with marked differences. The lower lip possesses only one distinctly conical row at its distal tip, this row being the widest and tallest while the remaining uncular rows are much shorter and flatter on the anterior and posterior surfaces.

At rest, the mouth is dorsally oriented. The lips do not touch, and the mouth remains slightly open at all times (Figures 3a,b and 6a). The mouth sits at the terminal end of the proboscis, which is elongate and tubular when protruded but is folded into a pocket‐like depression of the head that is bordered posteriorly by the concave ethmoid and anteriorly by the chin when at rest (Figures 3a and 6a). The chin is flexible, folding forward when the proboscis is deployed, and bears a thick epithelial layer studded by numerous goblet cells and sensory cells (Figure 6d). At rest, the lower lip is almost entirely concealed behind the chin. The proboscis itself is formed by a continuous sheet of skin that extends anteriorly from the ethmoid to the upper lip along its dorsal aspect and from the chin to the lower lip along its ventral aspect (Figure 6). At rest, the skin of the proboscis is folded and distinctly crimped (Figure 6c,d asterisks). Dorsally, the proboscis encases the maxillae, the premaxillae, and the dense connective tissue that extends anteriorly beyond these bones to form the lips (Figures 3a–d and 6a–c). Structural support of the proboscis is provided primarily by the maxillae which reinforce its lateral borders, allowing the proboscis to maintain a rounded, tubular shape when deployed. At the opening of the proboscis, the skin folds around the anterior rictal cartilages that are embedded in the lateral borders of the mouth opening. The upper and lower lips are united at their lateral margins by this intricate folding, which forms a thick, circular, and darkly pigmented flap of skin and cartilage on either side of the mouth (Figure 7). These flaps, hereafter called the labial flaps, slot under the upper lip and extend ventrally to cover the lower lip, concealing the lower lips' lateral extremes and the entirety of the labial commissure (Figure 7a,c).

FIGURE 7.

Histological overview of the lips and mouth folds in Phractolaemus ansorgii. (a) Close up of the mouth as a reference for the histological section with dashed lines depicting the region examined. At rest, the proboscis is tucked into the head, and the mouth is slightly open. (b) Sagittal section through the lateral border of the mouth showing the labial flap (lf) that is continuous with the upper lip. The lower lip (ll) extends dorsally as it approaches the labial commissure. (c) Close‐up of the open mouth to show the right labial flap and the proximal end of the lower lip that lies medial to it.

3.3. Behavioral observations

In our trials, a feeding event began upon the initiation of a strike. Strikes appear unidirectional and limited to the sagittal plane with little lateral excursion (no more than 7°) occurring at the distal tip of the proboscis that encompasses the maxillae, premaxillae, and lips (Figure S2). Directionality of protrusion depended on the position of the food. In these trials, food was administered to the bottom of the tank, so most protrusion was anteroventral to capture food particles along the substrate (Figure 8a). We did, however, capture one incidence of dorsal protrusion where a fish fed on particles of food suspended above them in the water. Dorsal protrusion of this nature was also observed outside of recorded trials as individuals would often lie underneath rocks and refugia to feed on food particles clinging to the surfaces above them.

FIGURE 8.

Representative frames from a feeding sequence of Phractolaemus ansorgii. (a) In a typical strike, the proboscis is protruded anteroventrally. (b) Ejection of a large particle of food (black arrow) involves dorsal rotation of the mouth.

Following strikes, the proboscis was not always immediately retracted but instead protruded again to probe (supplementary Video [Link], [Link], [Link]). Strikes were defined by deployment of the proboscis from a fully retracted to a protruded state and distinct from probes, which were partial protrusions, often iterative, from a non‐fully retracted state following the initial strike. During probing events, individuals would protrude their proboscis to varying distances while using their mouths to scrape at and suction food items along the substrate. Probing occurred following a strike over 70% of the time and involved some degree of protrusion that did not always extend to the maximum distance observed for that individual (see kinematic analysis below). The number of probes per feeding event was highly variable, ranging between 1 and 19 probes. Because the length of each recording was limited by our high frame rate, not all feeding events were captured to completion.

Fish were also capable of expelling unwanted food particles away from their faces (Figure 8b; Supplementary Video S4 and S5). When this occurred, the proboscis was retracted (likely by retraction of the maxillae past their starting position) directing the mouth opening dorsally. When the proboscis was protruded or the mouth was directed upward to expel food, the lateral borders of the mouth were stretched taut such that the lower lip was fully visible, having been pulled out from beneath the labial flap, and the lips formed a continuous, circular opening (Figure 8b; 0–68 ms). All feeding events (strikes, probes, and expulsions) involved a similar stretching of the lips and labial flaps to fully open the mouth (Figure 8a; 65–100 ms). Successful strikes and probes involved contraction of the mouth opening such that the lips were drawn together as the mouth was lifted off the substrate. Mouth opening often appeared coincident with protrusion but was also observed when the proboscis was fully retracted, such as during food expulsion. Mouth closure and retraction of the proboscis were not necessarily coincident, and full mouth closure was rarely observed.

3.4. Kinematic analysis

We quantified the maximum distance that the proboscis was protruded, protrusion speed, and speed of expelled food particles (Table S1). Across our three individuals, we observed a maximum protrusion distance of 33% of head length, with a minimum protrusion distance of 15% (average 27.26% ± 3.25 SD). The distance to which the proboscis was protruded varied between trials and individuals, with certain individuals exhibiting a wider range of protrusion distances than others (Pa1 average 26.53% ± 3.96, Pa2 average 28.36 ± 2.23, and Pa3 average 26.38 ± 3.79). There was no noticeable difference between protrusion distances in trials that featured a single strike as opposed to strikes followed by probing. Time to peak protrusion averaged 117 ± 53 ms with a range 52–241 ms. Average speed of protrusion was 2.74 ± 1.26 cm/s with a range 1.29–5.42 cm/s. Expulsion of food particles was observed in nine feeding events with an average speed of 9.60 ± 2.80 cm/s and a range 5.77–13.56 cm/s.

4. DISCUSSION

We describe the novel anatomy and kinematics of jaw protrusion in Phractolaemus ansorgii and examine the feeding mechanics of their highly complex cranial anatomy. Modifications in feeding architecture are reflected as changes in shape and spatial orientation of the oral jaws, a reorganization of the adductor musculature, and the formation of a newly constructed mouth supported not by skeletal elements but rather by connective tissue. With greatly reduced upper jaws and a mandible rotated nearly 180 degrees within the skull, the distal tips of the oral jaws do not support the oral aperture. As such, in order to feed benthically this species has evolved a protrusile proboscis with an unusual mouth opening at its end that is composed entirely of skin, hyaline cartilages, and other connective tissues that effectively take on the role of the oral jaws. With high‐speed videography, we visualized the subtle movements and feeding behaviors enabled by this flexible mouth, which facilitates the scraping of detritus and algae while also permitting the suction feeding of suspended particles. Furthermore, we measured several kinematic variables that allowed us to better understand the functional consequences of this unique anatomy during benthic feeding. We find that P. ansorgii is capable of protruding its jaws to a distance that exceeds 30% of its head length, representing a previously undescribed form of extreme jaw protrusion in fishes.

In accordance with published hypotheses and electromyographic studies on the role of various cranial muscles in fish feeding, we suggest that the anterior movements of the upper and lower jaws are actuated by two separate mechanisms involving levator operculi and adductor mandibulae. Anteroventral rotation of the lower jaw about the jaw joint may be achieved by contraction of the levator operculi, a mechanism that has been described as a means of mandibular depression in cichlids and other acanthomorph fishes (Liem, 1970, 1978, 1980; Westneat, 1990). In P. ansorgii, the mandible is attached to the opercular series by the retroarticular–interopercular ligament (Howes, 1985); thus, we posit that when the levator operculi contracts, the bones of the opercular series are pulled posteriorly, and the dentary, rather than being depressed, pivots anterior on the jaw joint, pushing the maxilla anteriorly and thus deploying the proboscis. Protrusion of the upper jaw may be achieved by contraction of A1, which we describe here as a fixed pulley system with its tendon running along the quadrate and through the quadrate tunnel before turning sharply dorsally to attach to the maxilla. Protrusion is then achieved by two muscular actions that may act synchronously or asynchronously to modulate the degree of anterior protrusion or to accomplish different feeding behaviors. This is further supported by electromyography (EMG) studies investigating motor activity for both biting and suction‐feeding fishes in which similar muscle firing patterns have been reported with the levator operculi firing prior to the adductor mandibulae during prey capture, although the amount of concurrent muscle activity is highly variable (Alfaro et al., 2001).

Retraction of the proboscis may be accomplished by contraction of both branches of A2, A2β and A2α, the former of which pulls the dentary posteriorly and the latter of which pulls the chin and proboscis posteriorly. Given the maxilla is united to the distal end of the dentary by a ligamentous connection, contraction of A2 would pull the maxilla posteriorly as well. We hypothesize that the scraping and pinching movements observed during probes may be possible due to contraction of A1o and A1i which would serve to pull the maxilla ventrally toward the quadrate. These movements appear to be independent of protrusion, as they occur at intermediate protrusion distances, and individuals have been observed opening and closing their mouths while the proboscis is fully retracted. Furthermore, this anatomy seems to permit quick expulsion of unwanted food items. As P. ansorgii possesses an epibranchial organ, a muscular pocket‐like structure in the dorsoposterior pharynx that functions to aggregate food, the ability to expel food quickly during feeding may be advantageous for sorting particles that may be non‐nutritive or too large for proper aggregation. A highly complex repertoire of mouth movements permitted by coordination of the adductor mandibulae muscles is also found among cypriniforms. With EMG, Ballintijn et al. (1972) demonstrated not only jaw closure but also downward extensions of the lips and a capacity to eject food particles via contraction of different branches of adductor mandibulae in common carp. Although we believe our revisions to the existing muscular descriptions have strengthened our understanding of P. ansorgii's cranial anatomy, allowing us to propose a newly informed functional hypothesis for protrusion, how exactly these muscles work to move the oral jaws cannot be confirmed without EMG data.

Although previous anatomical studies of P. ansorgii have suggested how musculoskeletal elements of the oral jaws and neurocranium may coordinate premaxillary protrusion, surprisingly little attention has been given to the composition of the mouth itself. Among the most striking architectural modifications is that, given the orientation and diminutive nature of the oral jaws, the mouth is entirely supported by cartilages and other connective tissues in lieu of bone. Indeed, there is a large amount of hyaline‐cell cartilage in the head: surrounding the bony elements of the jaws, in the ligaments, serving as symphyseal tissue, and forming the rigid borders of the mouth where there are otherwise no bones at all. Hyaline‐cell cartilage is a highly flexible tissue found in the lips, rostral folds, ligaments, and articular surfaces of other benthic‐feeding fishes (Benjamin, 1989). In the cypriniform Gyrinocheilus aymonieri, hyaline‐cell cartilage has been shown to support the oral sucker and lips, allowing for large degrees of deformation during adherence and benthic feeding (Benjamin, 1986). In many other bottom‐dwelling ostariophysans, it forms periosteal cushions on bone, which have been posited to act as resilient buffers against stress during burrowing and probing (Benjamin, 1989, 1990). We suggest that the large quantity of hyaline‐cell cartilage in the oromandibular region of P. ansorgii serves similar functions to what Benjamin (1989) has described in other ostariophysans, namely that of providing rigid yet yielding structure to the mouth. During probing events, the flexibility of these connective tissues and their loose attachment to the oral jaws may facilitate an increased range of motion and rapid flexion of the lips, lending an appearance of fine motor control that is characteristic of mammalian lips or an elephant's trunk despite P. ansorgii's mouth bearing no intrinsic muscle (Supplementary Video S3). Such structural flexibility would be particularly advantageous for feeding on irregular surfaces such as wood, rock and gravel; in conjunction with the keratinized epithelium that overlies these connective tissues, it is one of several features of P. ansorgii's trophic anatomy that may aid in dislodging invertebrates, algae, detritus, and other flocculent organic matter from the substrate.

Indeed, another striking feature of P. ansorgii's feeding apparatus is their unculiferous mouth (Figure 6a,b,e). Unculi are horny projections present in at least 16 ostariophysan families, often on the adhesive pads or the lips of edentulous bottom‐dwelling species where they have been suggested to provide mechanical protection of the skin, aid in adhesion, and facilitate the rasping of algae, insect larvae, and other food items from the substrate (Dana Ono, 1980; Pinky et al., 2004; Roberts, 1982). In P. ansorgii, keratinous unculi line the top and bottom lips, and this keratinization of the epithelium may provide a greater capacity for friction between the mouth and the substrate when feeding. In the upper lip of P. ansorgii, we found a noticeable gap, or bursa, between the unculiferous epithelium and the distal end of the dense connective tissue supporting it (Figure 6b). Benjamin (1989) noted that the lips of many algae‐eating cypriniformes bear a keratinized epithelium separated from the underlying masses of hyaline‐cell cartilage by bursae, allowing a degree of movement between the epithelial layer and the supportive connective tissue beneath. In P. ansorgii's upper lip, it is possible that during protrusion, anterior movement of the maxilla pushes the supportive connective tissue against the epithelium, pulling it taut and creating a stiffer structure to support the overlying unculi as the fish probes. During retraction of the maxilla, this connective tissue may then be pulled back, allowing the skin of the upper lip to slacken. The resulting conveyer‐like movement of the overlying unculi could then aid in dislodging food particles deposited along the substrate. P. ansorgii's unculi are small (≥20 μm), too small to visualize in our videos with any clarity, preventing us from confirming this hypothesis. Nevertheless, we suggest that changes in the shape and position of the oral jaws have necessitated a novel arrangement of both pliant connective tissue and stiff keratinous unculi, structures that likely serve complementary functions during benthic feeding (Benjamin, 1989).

Despite the novel tissue composition and structure of the mouth, the performance of P. ansorgii's feeding apparatus is comparable to other fishes. The proboscis is highly protrusile, reaching a maximum protrusion distance of 33% of its head length, a distance equivalent to that of the giant pikehead, Luciocephalus pulcher, which employs both lower jaw depression and cranial elevation to achieve premaxillary protrusion (Lauder & Liem, 1981). Although extreme jaw protrusion has been posited as an adaptation for capturing elusive prey, the diet of P. ansorgii is composed primarily of detritus and algae which they suction off the substrate (Odum & Anuta, 2001). These food items are largely immobile organic debris and microorganisms resting on or attached to solid substrates, thus it is not entirely surprising that P. ansorgii exhibits a slow average protrusion speed when compared to acanthomorph and cypriniform species that feed on faster‐moving midwater prey (Table 1) (Day et al., 2005; Gibb & Ferry‐Graham, 2005; Lauder, 1980; Staab, Ferry, & Hernandez, 2012; Staab, Holzman, et al., 2012). A previous study in Cyprinodontiformes found that jaw morphologies that permit enhanced dexterity during protrusion may incur functional trade‐offs in other aspects of feeding performance, namely strike speed (Ferry‐Graham et al., 2007). Although the kinematic mechanism effecting protrusion in P. ansorgii may be constrained in terms of speed, the requisite reorganization of the feeding apparatus may have made the proboscis better suited for increased maneuverability and kinematic flexibility (e.g., probing the benthos at varying distances, rotating the mouth dorsally to expel particles away from the face, and feeding at highly variable protrusion velocities). Furthermore, the maneuverability of the proboscis and the mouth is especially relevant for feeding on heterogeneous substrates, wherein the deformations permitted by fleshy lips rich in hyaline‐cell cartilage can allow the mouth to form a better seal on irregular substrates during scraping. Even with such drastic modifications to the underlying architecture enabling protrusion, the mechanism employed by P. ansorgii approaches and in some ways exceeds the functional capacity of the feeding mechanisms seen in more species‐rich and morphologically diverse groups of fishes.

TABLE 1.

Comparison of average protrusion speed.

Order	Family	Species	Average speed	Reference
Cypriniformes	Catostomidae	Catostomus insignis	5.73 cm/s	Staab, Ferry, & Hernandez. (2012)
Cypriniformes	Cyprinidae	Carassius auratus	15.6 cm/s	Staab, Ferry, & Hernandez (2012)
Cypriniformes	Cyprinidae	Danio rerio	4.71 cm/s	Calculated from Gibb & Ferry‐Graham (2005), table 2
Cypriniformes	Cyprinidae	Danio rerio	4.36 cm/s	Staab, Ferry, & Hernandez (2012)
Cypriniformes	Cyprinidae	Devario aequipinnatus	3.97 cm/s	Staab, Ferry, & Hernandez (2012)
Cypriniformes	Leuciscidae	Gila robusta	10.7 cm/s	Staab, Ferry, & Hernandez (2012)
Cyprinodontiformes	Poeciliidae	Poecilia sphenop	2.40 cm/s	Calculated from Gibb & Ferry‐Graham (2005), table 2
Gasterosteiformes	Syngnathidae	Syngnathus leptorhynchus	5.93 cm/s	Calculated from Gibb & Ferry‐Graham (2005), table 2
Gonorynchiformes	Kneriidae	Phractolaemus ansorgii	2.74 cm/s	This study
Perciformes	Centrarchidae	Lepomis macrochirus	11.02 cm/s	Calculated from Gibb & Ferry‐Graham (2005), table 2
Perciformes	Chaetodontidae	Chaetodon xanthurus	13.89 cm/s	Calculated from Gibb & Ferry‐Graham (2005), table 2
Perciformes	Chaetodontidae	Heniochus acuminatus	10.40 cm/s	Calculated from Gibb & Ferry‐Graham (2005), table 2
Perciformes	Labridae	Choerodon anchorago	2.03 cm/s	Calculated from Gibb & Ferry‐Graham (2005), table 2
Perciformes	Labridae	Coris gaimard	4.23 cm/s	Calculated from Gibb & Ferry‐Graham (2005), table 2
Perciformes	Labridae	Hologymnosus doliatus	3.67 cm/s	Calculated from Gibb & Ferry‐Graham (2005), table 2
Perciformes	Labridae	Novaculichthys taeniourus	4.97 cm/s	Calculated from Gibb & Ferry‐Graham (2005), table 2
Perciformes	Labridae	Oxycheilinus digrammus	6.11 cm/s	Calculated from Gibb & Ferry‐Graham (2005), table 2
Perciformes	Osphronemidae	Betta splendens	6.76 cm/s	Calculated from Gibb & Ferry‐Graham (2005), table 2
Pleuroncetiformes	Paralichtyidae	Xystreurys liolepis	17.36 cm/s	Calculated from Gibb & Ferry‐Graham (2005), table 2
Pleuroncetiformes	Pleuronectidae	Pleuronichthys verticalis	23.88 cm/s	Calculated from Gibb & Ferry‐Graham (2005), table 2

It is tempting then to speculate on why P. ansorgii occupies such an isolated position within the phylogeny when it bears many of the hallmarks of a successful benthic feeder. We need to look no further than the rest of Ostariophysi to find any number of bizarre mouths and feeding mechanisms associated with sucking and scraping at algae and detritus. Yet benthic‐feeding ostariophysans like anostomids, catostomids, prochilodontids, and loricariids belong to groups that are more speciose, some by several orders of magnitude, than all of Gonorynchiformes (Nelson et al., 2016). Wherein lies the fundamental difference? Such speculation lies beyond the scope of this paper, but it is worthy to note that in contrast to many other benthic and substrate‐feeding fishes, P. ansorgii's jaws are relatively inflexible. Loricariids, for example, have left–right decoupled dentaries that are specialized for maintaining constant contact with irregular substrates (Lujan and Armbruster, 2012; Schaefer & Lauder, 1986), and various marine algivores such as acanthurids, kyphosids, labrids, and pomacanthids possess novel intramandibular joints similarly conducive to scraping (Allen et al., 1998; Konow & Bellwood, 2005; Moran & Ferry, 2014; Price et al., 2010; Purcell & Bellwood, 1993). P. ansorgii's oral jaws, however, lack such intrinsic osteological flexibility. It is instead the connective tissues of the proboscis that likely accommodate feeding on irregular surfaces and that are responsible for the repertoire of subtle mouth movements observed during probing. Furthermore, the proboscis possesses some functional versatility in that it is protruded dorsally during surface breathing (pers. obvs.) and expels unwanted particles. The need to deploy the proboscis ventrally while also performing any of these additional functions may have constrained the orientation, shape, and mobility of the oral jaws, rendering the exploration of any other ecomorphological variation biomechanically impossible.

The capacity for jaw protrusion has been hypothesized to increase cranial complexity, allowing for and benefitting from the decoupling of previously associated structures, which can facilitate the evolution of multiple morphological solutions to achieve the same functional goal, namely that of acquiring food (Hernandez & Staab, 2015; Lauder, 1982; Staab, Ferry, & Hernandez, 2012; Staab, Holzman, et al., 2012; Wainwright et al., 2015). While jaw protrusion has been implicated in the success of lineages like Cypriniformes and Acanthomorpha, it is not true of all fishes capable of feeding with even extreme forms of cranial kinesis. Asymmetry in species richness exists broadly across teleosts but is especially prominent among the groups capable of jaw protrusion. In the case of P. ansorgii, we might consider that the reorganization of not only the musculoskeletal architecture but also much of the connective tissues of the head imposes a large constraint on this lineage's capacity to adapt to new or changing environmental conditions, effectively serving as an evolutionary dead end (Kelley & Farrell, 1998; Stebbins, 1957; Stireman, 2005). Ultimately, selection acts upon phenotypic traits that affect feeding performance, and as demonstrated here, P. ansorgii feeds benthically in a way that is novel among bony fishes. Phractolaemus ansorgii represents a unique opportunity to study the functional consequences of morphological novelty, its effect on performance, and the limits of the teleost skull, and we propose that while architecturally complex, this mechanism may have limited versatility over evolutionary timescales.

Supporting information

Figure S1:

Figure S2:

Table S1:

Video S1:

Video S2:

Video S3:

Video S4:

Video S5:

Evans, A.J. , Naylor, E.R. , Lujan, N.K. , Kawano, S.M. & Hernandez, L. (2024) Deploy the proboscis!: Functional morphology and kinematics of a novel form of extreme jaw protrusion in the hingemouth, Phractolaemus ansorgii (Gonorynchiformes). Journal of Anatomy, 244, 929–942. Available from: 10.1111/joa.14020

DATA AVAILABILITY STATEMENT

Data that supports the findings of this study are available in the supplementary material of this article. Additional videos or ct‐scans are available from the corresponding author upon reasonable request.