JOURNAL OF MORPHOLOGY 187:81-108 (1986)

Morphology and Function of the Feeding Apparatus of the Lungfish, Lepidosiren paradoxa (Dipnoi)

W. E. BEMIS AND GEORGE V. LAUDER Department of Anatomy, University of Chicago, Chicago, Illinois

ABSTRACT The feeding mechanism of the South American lungfish, Lep idosiren paradoxa retains many primitive teleostome characteristics. In partic- ular, the process of initial prey capture shares four salient functional features with other primitive vertebrates: 1) prey capture by suction feeding, 2) cranial elevation at the cranio-vertebral joint during the mouth opening phase of the strike, 3) the hyoid apparatus plays a major role in mediating expansion of the oral cavity and is one biomechanical pathway involved in depressing the mandible, and 4)peak hyoid excursion occurs after maximum gape is achieved. Lepidosiren also possesses four key morphological and functional specializa- tions of the feeding mechanism: 1) tooth plates, 2) an enlarged cranial rib serving as a site for the origin of muscles depressing the hyoid apparatus, 3) a depressor mandibulae muscle, apparently not homologous to that of amphibi- ans, and 4) a complex sequence of manipulation and chewing of prey in the oral cavity prior to swallowing. The depressor madibulae is always active during mouth opening, in contrast to some previous suggestions. Chewing cycles include alternating adduction and transport phases. Between each adduction, food may be transported in or out of the buccal cavity to position it between the tooth plates. The depressor mandibulae muscle is active in a double-burst pattern during chewing, with the larger second burst serving to open the mouth during prey transport. Swallowing is characterized by prolonged activity in the hyoid constrictor musculature and the geniotho- racicus. Lepidosiren uses hydraulic transport achieved by movements of the hyoid apparatus to position prey within the oral cavity. This function is analagous to that of the tongue in many tetrapods. A key event in the evolution of land verte- more distant outgroup (Schultze and Trueb, brates from fishes was the transition from '81),lungfshes are the closest living out- aquatic suction feeding to terrestrial feeding group of tetrapods, Latimeria probably being with the tongue and jaws. This transition more distantly related to tetrapods than pre- involved numerous major changes in the viously supposed (Miles, '77; Wiley, '79; Ro- feeding mechanism, such as the evolution of sen et al., '81). As such, dipnoans are in a novel mechanisms for acquiring food, han- unique position to offer insight into the evo- dling it in the oral cavity, and swallowing lution of tetrapod feeding systems. At the the processed food (Bramble and Wake, '85). same time, members of the radiation of lung- Although much has been written about the fishes, distinct from its first appearance in fish--amphibiantransition, little information the lower Devonian, have diverged exten- exists on the feeding mechanics of the Dipnoi sively from the common primitive body plan or lungfshes, one of the most important liv- that lungfishes and tetrapods shared. The ing groups relevant to the origin of terres- extant lungfishes are in some respects poor trial feeding (Thomson, '69). The phylogen- models of phylogenetically primitive lung- etic position of lungfishes makes this group critical to our understanding of vertebrate Dr. Bemis' present address is Department of Zoology, Univer- evolution. Whether lungfishes are regarded sity of Massachusetts, Amherst, MA 01003-0027. as the sister group of tetrapods (Rosen et al., Dr. Lauder's present address is Department of Developmental '81) or whether they are considered to be a and Cell Biology, University of California, Irvine CA 92717.

0 1986 ALAN R. LISS, INC. 82 W.E. BEMIS AND G.V. LAUDER fishes. We know this, because their excellent mammalian mastication. A final goal of this fossil record documents extensive morpholog- paper will thus be to discuss some of these ical change during their evolution (Bemis, convergent features, particularly the lingual '84b). For example, early in their history, transport system typically used by tetrapods lungfishes exhibited much evolutionary div- and the hydraulic transport system used by ersification and specialization in the feeding lungfkhes. apparatus, and two distinct groups-one with specialized tooth plates and the other show- MATERIALS AND METHODS ing periodic growth and shedding of denti- Animals and diet cles-can be recognized (Campbell and Bar- Nine individual Lepidosiren paradoxa ob- wick, '83). tained from commercial dealers were avail- In this paper, we provide the first cinema- able for this study. The five specimens used tographic and electromyographicstudy of the for experiments ranged in size from 15 cm feeding system of any dipnoan. The subject total length (TL) to 20 cm TL. Individuals of our study is the South American lungfish, were housed in 20-liter aquaria held at ap- Lepidosiren paradoxa. Lepidosiren and the proximately 26 degrees C. Animals were closely related African lungfishes, Protopte trained to feed readily on pieces of earth- rus spp., are members of a clade that is dis- worm (Lumbricus).A special training period tinct in many respects from the Australian to accustom the animals to feeding under the lungfish, Neoceratodus. Thomson ('69) pro- bright movie lights began one month prior to vided a general discussion of dipnoan food the start of filming. Individuals were starved and feeding systems, and concluded that the for up to two weeks prior to either filming manner of feeding in all three genera is sim- sessions or electromyographic studies. Large ilar. In the only detailed examination of the African lungfish, Protopterus aethwpicus, are feeding mechanics of living lungfishes, Per- known to feed extensively on molluscs, al- kins ('72) based conclusions about muscle though smaller individuals consume a diver- function solely on dissections. Most research sity of both hard and soft bodied inverte- on the functional morphology of lungfishes brates (Corbet, '61). Sparse ecological data has focused on respiration and circulation, available for Lepidosiren is inconclusive con- such as the electromyography of respiration cerning dietary specialization, though it in Protopterus (McMahon, '691, cineradiogra- seems likely that they are opportunistic om- phy of air breathing in Lepidosiren (Bishop nivores. Carter and Beadle ('30) maintained and Foxon, '68), and functional anatomy and young specimens on a diet of earthworms, physiology of lungfish hearts and circulatory prepared (presumably shell-less) snails, and systems (e.g., Johansen and Hanson, '68). plant material. They considered plant mate- The major goal of our study was to describe rial to be especially important, and stated and analyze the feeding of Lepidosiren and that only a few potential prey items-worms, use this information to identify anatomical insects, and insect larvae-were available in and functional features of the feeding system the water where the fish were taken. On the that are primitive to dipnoans. These primi- other hand, Lankester (1896),citing the field tive features are then compared to those observations of Bohls, stated that Lepidosi- previously identified in the feeding systems ren feeds chiefly on the hard-shelled snail of other lower vertebrates (e.g., ray-finned Ampullaria, and that plant material may be fishes, coelacanths, and salamanders). Com- ingested by accident. parison allows us to recognize which features of the vertebrate feeding mechanism are A natorny phylogenetically common to all of these lin- Gross morphological features of the feeding eages, which features are common to tetra- apparatus (osteology, arthrology, and myol- pods and lungfishes, and which are special- ogy) were examined by dissection using a izations limited to the Dipnoi. A second goal Zeiss IVb dissecting microscope. A camera was to provide a detailed functional analysis lucida attachment was used to prepare the of the tooth plates, which are the major spe- figures from dissected specimens and cleared cialization of the feeding mechanism of ex- and stained specimens. Three individuals tant lungfkhes. Unlike the vast majority of were cleared and stained (for both cartilage fishes, lungfishes process food items using a and bone) following procedures modified from marginally placed dental apparatus, with Wassersug ('76) and Dingerkus and Uhler movements that are broadly convergent to ('77). In addition, prepared skeletons were FEEDING MECHANICS OF LUNGFISH 83 used to examine osteological details. Both both the lower jaw and the hyoid apparatus. dry and cleared and stained material was It was measured as the perpendicular dis- also available for Protopterus. tance from a line drawn between the hyoid The serially sectioned head of a single in- and the lower lip to the most indented point dividual prepared for an earlier study (Be- on the lower jaw between the two end points. mis, ’84a) was available for analysis. Briefly, Together, these two measurements provide a this specimen was fixed in 5% neutral buf€- qualitative estimate of both the length and ered formalin, embedded in low viscosity ni- strength of the adduction phase of chewing. trocellulose (type RS 1/2, Hercules Inc., To aid in interpreting electromyographic Wilmington, Delaware), and serially cross recordings, simultaneous video recordings sectioned at 30 pm. Most sections were were made of feeding behavior. A Panasonic stained with Ehrlich’s hematoxylin or Ver- video cassette recorder, two Panasonic video hoeff s elastin, counterstained with picro- cameras, and a special effects generator per- ponceau or PAS, following methods outlined mitted the recording of the animal’s behavior in Humason (’72). These sections were stud- and the simultaneous electromyographic ac- ied for details about the tooth plates, cranial tivity as displayed on the chart recorder. The bones, and muscles. time between frames on the videotape was .0333 sec, more than adequate for the analy- Cinematography and video sis of prolonged chewing sequences. Overall, Kodak 4X Reversal Film was used in a ten simultaneous video and electromyogra- Photosonics 16-1PL camera to record all phic sequences were analyzed, each includ- phases of feeding. Filming rate varied from ing sustained chewing bouts; because each 200 frames per second (for suction feeding) to feeding sequence involves one prey-capture 100 frames per second (for extended slow suction and a small number of constrictions, chewing sequences). Three 600 Watt Smith- we recorded fewer successful simultaneous Victor filming lights provided illumination. recordings of these events. We filmed more than 40 suction feeding se- quences. Nineteen of these sequences met E lectromyography the minimum requirements for further kine- Eledromyograms were obtained using matic analysis, in that the view was lateral standard procedures (see Lauder, ’83a,b). throughout the sequence and the film was Briefly, the fish was anaesthetized with tri- judged to be up to speed on the basis of tim- caine methane sulfonate, aild fine wire steel ing light marks on the edge of the film. Ten alloy bipolar electrodes were implanted in as chewing sequences were filmed, and the two many as eight cranial muscles. We based our longest usable sequences, each representing protocol for electrode placement on detailed an entire “bout” of activity (see below), were study and measurement of dissected speci- analyzed in detail. mens. Owing to the small number of living Kinematic patterns were analyzed by pro- specimens available, we did not kill individ- jecting every third frame of a sequence onto uals to check our electrode placements fol- a Houston Instruments Hi-Pad digitizer and lowing successful recordings. Consistency of measuring a series of variables. Measuring electromyographic patterns from implanta- every third frame provided more than ade- tion to implantation was accepted as evi- quate resolution of the extreme excursions, dence of consistent electrode placement. based on the frame-by-frameanalysis of sev- Simultaneous signals from up to six mus- eral test sequences. Variables measured in cles were amplified with Grass P511J pream- both the suction feeding and chewing se- plifiers (set for a 30 to 3000 Hz bandpass) and quences were: hyoid depression, gape dis- recorded on tape with a Bell and Howell tance, distance from the prey to the plane of 4020A FM tape recorder. Electromyograms the gape, and cranial elevation. Two addi- were played back at one-eighth speed on a tional variables were measured in the chew- Gould 260 chart recorder, so that the effective ing sequences: adductor bulge and “crush.” frequency response of the myograms dis- Adductor bulge is the perpendicular distance played on the chart recorder was nearly 1,000 that the adductor mandibulae posterior mus- Hz. cle bulges above the surface of the head. The Electromyographic patterns obtained dur- “crush” measurement is highly correlated ing feeding were analyzed in several ways. with adductor bulge, but differs in that it First, the overall pattern of muscle activity incorporates information on the positions of was summarized using the method outlined 84 W.E. BEMIS AND G.V. LAUDER in Lauder (‘83a). For each of the three behav- differences, differences due to electrode im- iorally discrete events discussed below, a plants and experimental days) follows proce- summary diagram was constructed (e.g., dures presented in Shaffer and Lauder (‘85a), Figs. 10, 13). This was done by choosing a where this approach is discussed further. reference muscle for each behavior, and us- ing the onset of activity in this muscle as the RESULTS point from which the onset and offset times Anatomy for activity in other muscles were measured. Osteology and tooth plates Our method for determining onset and offset Figure 1 illustrates the cranial bones and times is somewhat subjective, particularly in cartilages of Lepidosiren. In comparison to the case of muscles such as the adductor the primitive condition for osteichthyans and mandibulae, in which a train of isolated amphibians, the skulls of living dipnoans spikes typically precedes the major continu- show a reduction of many bony elements ous burst of activity (e.g., Figs. 12, 14). In (Bridge, 1898; Miles, ’77). For example, the such cases, we accepted the first spike of an maxillary-premaxillary arcade, which is only evenly spaced train of spikes as the onset of debatably present in any dipnoan (Rosen et a muscle’s activity period. A number of elec- al., ’81; Campbell and Banvick, ’83), is ab- tromyograms was measured and, for each sent in all three living genera. Such extreme muscle, the mean onset and offset times, the reduction and modification make it difficult standard errors of these means, and the per- to identify homologies of some cranial bones. centage of experiments that the muscle was Our terminology follows Bemis (‘84a). active were calculated. This information was The largest and most important dermal combined into a summary diagram showing elements from the standpoint of the feeding the overall pattern of activity for the be- mechanism are the pterygoid and prearticu- havior. lar Fig. 1: PT, P). These bones support the We also analyzed the amplitude and spike massive, three-bladed tooth plates character- patterns of electromyograms by playing the istic of lepidosirenids. The right and left signals back from the tape recorder into a 12 prearticular bones are immovably attached bit DAS analog to digital converter. The tape to each other at the broad mandibular sym- recorder speed was slowed down to give an physis; a similar symphysis binds the right effective sampling frequency of 8,000 Hz. The and left pterygoids to each other. Because of digitized myograms were stored on an IBM the absence of the maxillary arcade and den- XT computer where the signals were pro- tary, the prearticular and pterygoid tooth cessed using the spike counting algorithms plates topologically make up the margins of developed by Beach et al. (‘82) and Gorniak the jaws, even though they are located well and Gans (‘80). Both digitized myograms and behind the lips and the oral opening (Fig. 1). the derived spike and amplitude analyses During feeding, the anatomical limitations were plotted using an HP 7470A plotter of the jaw joint and symphysis restrict the (1000 points per inch resolution; Fig. 14). motion of the lower jaw to an up and down To be sure that the sample rate of the ana- hinge-like motion, which in turn, causes the log to digital conversion was adequate to rep- tooth plates to occlude precisely (Bemis, ’84a). resent the information in the electromyo- Anteriorly, a small pair of so-called vomerine grams, we sampled one channel of data at a teeth are embedded in cartilage beneath the rate of 31.4 KHz. A Fast Fourier Transform dermal ethmoid (Fig. 1). The articulation of (FFT, calculated using the Cooley-Tukey al- the dermal ethmoid with the supraorbital is gorithm) of these signals showed that vir- not kinetic as believed by Berman (‘79). The tually no power is present above 1000 Hz, fossa for the large adductor mandibulae mus- and thus our sample rate of 8,000 Hz more cles are bounded above by the supraorbital, than satisfies the Nyquist sample criterion. on its median surf ce by the frontal bone and braincase, and be Bow by the squamosal. Be- Statistical analyses neath the squamosal are the two small ele- Statistical procedures used in this study ments of the oDercular series. a-eatlv reduced (analysis of variance, both nested and one- from their coAdition in fossiydipnoans (Fig. way, regression, paired t-tests) generally fol- 1: OP. SOP’). low Sokal and Rohlf (‘81). One of the statisti- The’ only ‘element of the visceral skeleton cal analyses in this paper, the nested analysis exhibiting any ossification is the ceratohyal of variance (used to assess interindividual (Fig. 1:CH). The ceratohyals are suspended FEEDING MECHANICS OF LUNGFISH 85

Fig. 1. Lateral and ventral views of the cranial skel- and cranial rib (CR). Other labeled elements are: AN, eton of Lepidoscren paradoxa Note the vomerine tooth angular; DE, dermal ethmoid; EO, exoccipital; FR, fron- (V), and the pterygoid (PT)and prearticular (P) bones, tal; MC, Meckel’s cartilage; N, notochord OP, opercu- which support the tooth plates. Also, note the relation- lum; S, supraorbital; SOP, suboperculum; SQ,squamosal. ship of the ceratohyal (CH) to the pectoral girdle (PG) by hyoid suspensory ligaments from the ven- corners of the skull. Anterior-posterior mo- tral surface of the quadrate cartilage and tion of the ribs is allowed by the well-devel- squamosal bone. A second ligament, the oped synovial articulations with the skull. mandibulohyoid ligament, ties each cerato- The pectoral girdle is embedded in the mus- hyal to the lower jaw. Cranial ribs (Fig. 1:CR) culature halfway between the cranial ribs extend ventro-posteriorly from the posterior and the ceratohyal. 86 W.E. BEMIS AND G.V. LAUDER

Fig. 2. Lateral and ventral views of the superficial electromyographic purposes); IH, interhyoideus; IM, in- cranial musculature of Lepidosiren paradona. AMa and termandibularis; RAO, retractor anguli oris superfici- AMP, adductor mandibulae anterior and posterior; DM, alis; RCa and RCp, rectus cervicis (anterior and posterior depressor mandibulae; EP, epaxialis; GTa and GO,gen- subdivisions are for electromyographic purposes). iothoracicus (anterior and posterior subdivisions are for

jaw opening; 3) jaw and hyoid compressing or Myology constricting; 4) hyoid depressing; and 5) head The major cranial muscles of Lepidosiren raising. are illustrated in Figure 2. Many authors There are four muscles in the first group have contributed to understanding the cra- (jaw closing muscles), and all are innervated nial musculature of dipnoans, including by the trigeminal nerve: the adductor man- Owen(1839)' Bridge (1898), Luther ('13)'Edge- dibulae anterior and posterior, and the re- worth ('26, ,351, Fox ('651, and McMahon ('69). tractor anguli oris superficialis and pro- However, not all authors agree on the details fundus (Fig. 2). The major jaw closing mus- of cranial myology in the group (e.g., Wiley, cles are the pinnately fibered anterior and '79; Jollie, '82). For the present purpose of posterior divisions of the adductor mandibu- clarifying function, it is simplest to divide lae. Although these two portions are distinct, the musculature into five functional cate- they arise developmentally from a common gories based on origins and insertions, and primordium (Edgeworth, '35)' and insert on then describe the individual muscles of each the coronoid process of the lower jaw via a group. These groups are: 1) jaw closing; 2) common tendon. Both portions of the adduc- FEEDING MECHANICS OF LUNGFISH 87 tor mandibulae were studied electromyo- the interhyoideus forms the muscle sur- graphically. rounding the opercular opening. The inter- The remaining two muscles of the first mandibularis and interhyoideus muscles group, the lip retractors, originate from the function as constrictors, acting to raise the fascia covering the posterior portion of the hyoid apparatus or stiffen the opercular flap. adductor mandibulae as well as the squamo- Both were studied electromyographically. sal bone. They insert on the connective tissue In the fourth group is the major hyoid de- of the upper lip. The superficial lip retractor pressor, the rectus cervicis (= sternohyoi- was studied electromyographically. deus). At its posterior end, the large paired The three muscles in the second group are rectus cervicis muscles are continuous with all potential jaw-opening muscles: the geni- the ventral trunk musculature: some fibers othoracicus, rectus cervicis, and depressor originate from the cranial ribs, others at the mandibulae. The geniothoracicus and rectus pectoral girdle. The fibers of the recti cervicis cervicis are hypobranchial muscles, while the run antero-medially to insert on the cerato- depressor mandibulae is derived from the hyals. The primary action of these muscles is hyoid arch. The geniothoracicus originates to depress the hyoid; however, hyoid depres- on the fascia of the ventral trunk muscles sion is coupled to lower jaw depression by the and its straight, parallel fibers continue an- mandibulohyoid ligament. Anterior and pos- teriorly to the area beneath the hyoid (Fig. terior subdivisions of this muscle were rec- 2). The muscle becomes a thin tendinous ognized for electromyographic study. Anter- sheet by the point of its insertion near the ior electrodes were placed near the muscle’s symphysis. For the purposes of electromyog- insertion on the ceratohyal. Posterior elec- raphy, we distinguished an anterior portion trodes were located in the main body of the at the hyoid and a posterior portion in the muscle, ventral to the base of the pectoral belly of the muscle. The large recti cervicis fin. muscles are described under group 4 below. The fifth group, the head-elevating mus- Although hyoid depression is the main func- cles, consists of the epaxialis. The epaxial tion of these muscles, they are potential jaw muscles continue from the trunk up over the depressors due to the ligamentous connection posterior end of the adductor musculature, between the hyoid and mandible. where they are attached both to the skull The muscle we refer to here as the depres- and to the fascia overlying the adductor mus- sor mandibulae has been known by a variety cles. Epaxial muscles were studied by plac- of names, most commonly, retractor mandi- ing electrodes about 0.5 cm posterior to the bulae (e.g., McMahon, ’69; Edgeworth, ’35). skull. Thomson, (’69, pg. 98) refers to a postero- dorsal depressor mandibulae muscle of dip- Histology noans, presumably the same muscle we des- Parker (1892) is an excellent source on cra- ignate here. Our use of the name depressor nial histology of Protopterus and Lepidosiren. mandibulae is intended to reflect the correct In cross section, the oral cavity is shaped like function of the muscle. The muscle originates an inverted “U” (Fig. 3A,B). The palatal mu- from the surfaces of the opercular and subop- cosa is underlain by the parasphenoid bone, ercular bones and inserts via a tendon onto ethmoid cartilage, and dense connective tis- the posterior corner of the lower jaw. Due to sue, and the sides of the cavity are supported the large diameter of the jaw joint, the mus- by the ventral wings of the pterygoid bones. cle actually inserts quite far from the center Anteriorly, the mucosal surface of the of rotation of the joint. Thus, mechanically, tongue is supported by a stiff, chondroid con- the depressor mandibulae is well suited to nective tissue “pad” (Fig. 3A,B). The tongue depress the lower jaw. Functionally, the jaw pad continues posteriorly, along the lateral joint cannot allow retraction. Electrodes were margins of the tongue, where it supports a placed in the center of the muscle. prominent ridge on each side. Posteriorly, The two muscles of group three, the inter- the mucosal surface of the tongue is sup- mandibularis and interhyoideus, are derived ported by a thin layer of connective tissue from the mandibular and hyoid arch respec- lying over the large recti cervicis muscles. tively. The muscles originate as thin sheets There is no separate intrinsic musculature. from the ventral borders of the lower jaw and The outer fibrous layer of the tongue pad is suboperculum and meet in a tendinous sheet continuous with the periosteum of the cera- at the ventral midline. The posterior part of tohyal bones. Thus, the tongue cannot move Fig. 3. A) Mid-sagittal view of Lepidosiren head, head of Lepidosiren, showing the shape of the oral cavity, showing oral cavity, large rectus cervicis muscle (RC) ceratohyal bones (CH), lateral wings of the tongue pad and tongue pad (PD) forming the body of the tongue, (PD), loose connective tissue beneath hyoid and tongue position of prearticular tooth plates VP), and margins of GCT), and the large adductor mandibulae posterior mus- lips extending anteriorly. B) Cross section through the cle (AMP). FEEDING MECHANICS OF LUNGFISH 89 independently of the hyoid apparatus. When snout or within several millimeters of it (Fig. the hyoid is in a raised position, the tongue 4: frame 1). Strikes are characterized by a nearly fills the oral cavity, so that there is relatively short duration (between 50 and 200 little dead volume prior to mouth opening. ms), during which the prey is engulfed and Strong ventral movements of the hyoid ap- carried into the buccal cavity with the flow paratus and its attached tongue occur during of water created by expansion of the mouth feeding and respiration. The sublingual re- cavity (Fig. 4). There is usually a pause of gion is filled with a very loose connective from one to several seconds following the tissue, providing an easily deformed cushion strike before chewing begins. This is shown into which the hyoid apparatus can be moved in Figure 5 by the absence of any electromy- (Fig. 3). Scalation of the skin in this region is ographic activity immediately following the suppressed, presumably facilitating disten- strike. Chewing of the prey between the tooth sion. plates occurs in several stages that we refer Anteriorly, the mouth is constricted to an to as chewing bouts, with successive bouts approximately circular opening. The lips ex- separated by pauses (Fig. 5). From one to five tend out, away from the underlying bones. or more chewing bouts may occur during a They are supported by various cartilaginous feeding (mean = 2.5, based on 16 feedings). derivatives of the nasal capsule and Meckel's Each chewing bout consists of a series of cartilage as well as abundant chondroid tis- chewing or adduction cycles (mean number sue. The margin of the upper lip curls lat- of cycles per bout = 7.2, based on 40 bouts). erally around the margin of the lower lip. Each chewing cycle in turn consists of a pe- The upper lip is supported by a rod of chon- riod of jaw adduction in which the tooth droid tissue that contains elastic fibers (Bert- plates are adducted and the prey crushed (= mar, '66, believed this rod to be homologous adduction phase), and a period during which with an upper labial cartilage). The lower lip the prey is transported within the oral cavity does not contain a comparable support bar, and repositioned between the tooth plates (= though its medial edge is well defined, with transport phase). Alternation of these two a slight curl that mates with the upper lip. phases produces the key characteristic of the The tooth plates are surrounded by fleshy chewing bouts, which is the movement of pads of tissue so that only the apical ridges prey into and out of the oral cavity, with of the tooth plates appear to project from the crushing of the prey occurring between each epithelium (Parker, 1892). Like the tongue transport phase. Figure 6 illustrates a chew- pad, these tooth plate pads are supported by ing sequence in which the prey is being a core of chondroid connective tissue. The transported into the mouth cavity (frames a, pads are easily deformed or moved when b, and c), crushed (frames d to k), and then pressed upon. Thus, food items being chewed again transported into the mouth (frames 1 presumably contact more than just the ex- to 0)before being crushed again. posed ridges of the tooth plates. After the final chewing bout has ended, the prey is located within the mouth cavity and Overview of events during feeding swallowing occurs. Swallowing may be initi- The process of prey capture and swallowing ated by rapid hyoid depression that creates a in lungfishes may be divided into four com- strong flow of water through the oral cavity ponents: 1)approach to the prey; 2) the strike; and moves the prey posteriorly to the pha- 3) chewing; and 4) swallowing. The approach ryngeal region. Subsequent contraction of the to prey located on the bottom often occurs hyoid musculature constricts the posterior with the head depressed ventrally and the part of the buccal cavity and forces the food anterior body arched. Prolonged searching into the esophagus. We term this behavior a may occur before prey are located, and dur- constriction or buccal compression. ing this time the head is usually maintained at an angle to the substrate. The strike: initial suction After a food item is located, the entire feed- Photographs of a complete strike are given ing sequence may last from 10 seconds to in Figure 4. Considerable variation was several minutes, with the average in this found both within and between individuals study being 36 seconds. Figure 5 summarizes in the time course of prey capture by suction the events and terminology used here to de- feeding. Gape profiles, for example, can be scribe feeding. The strike is usually initiated either unimodal or have multiple peaks and only when the prey is either touching the can vary significantly in duration (Fig. 7A). 90 W.E. BEMIS AND G.V. LAUDER

Fig. 4. Eight frames from a film taken at 200 frames/ bottom of the tank. Peak hyoid depression occurs in the second to show the pattern of head movement that oc- last frame. curs during a strike at a piece of worm located on the

The dominant kinematic pattern for suction gin until the mouth nears maximum gape, feeding is shown in Figure 7B. The prey and peak hyoid excursion occurs after peak moves toward the jaws as the mouth opens gape (Fig. 7B). The cranial angle trace is and usually enters the mouth at or near peak similar to that of the gape, achieving its peak gape. Initial depression of the hyoid occurs value at about the same time as peak gape. at the start of mouth opening, but a slow Variation in the four kinematic measure- hyoid elevation typically follows. The major ments as well as the total gape cycle time in ventral movement of the hyoid does not be- nineteen feedings by four individuals is FEEDING MECHANICS OF LUNGFISH 91

U I I I! 1- STRIKE CHEWING BOUT CHEWING BOUT CONSTRICT & SWALLOW

A 1

ADDUCTION CYCLE

DM Y 71 1-sec

IH r I' *+++

Fig. 5. Five electromyograms recorded simultane- Two chewing bouts then occur, with about a one second ously to show the behavioral events in a feeding se- pause between bouts. Within a chewing bout are several quence and the terminology used in this paper. adduction cycles (= chewing cycle), each consisting of an Calibration bars on left side of each trace = 200 micro- adduction phase and a transport phase. At the end of the volts. Initial prey capture occurs at the strike, and is last bout, two constrictions (= buccal compressions) occur. followed by a one second pause with no muscle activity. shown in Figure 8, and the results of pair- more rapid, a decrease in time between the wise analyses of variance (ANOVA) among onset of the depressor mandibulae and the these variables is given in Table 1. In all onset of the adductor was seen (Fig. 9B). feedings, the hyoid reaches its maximum Other muscles, such as the interhyoideus and ventral excursion significantly after peak the geniothoracicus become active also. The gape (Table 1). Cranial angle, however, can fastest strikes (gape cycle times of 40 ms) reach its peak either before or after gape display considerable overlap between mouth (Fig. B), and the times to maximum gape and opening and closing muscles (Fig. 9C): the cranial angle are not significantly different jaw adductors and depressors become active (Table 1). within 20 ms of each other. The electromyographic patterns during A summary pattern for N = 34 strikes is suction feeding differed greatly with the shown in Figure 10. The depressor mandibu- speed of the strike (Fig. 9). In slow strikes, lae was active in all strikes as was the rectus only the depressor mandibulae and adductor cervicis posterior (first burst). In at least two- mandibulae muscles showed activity, and thirds of all feedings analyzed, the anterior there was a 100 to 200 ms delay between the rectus cervicis also was active with the onset offset of depressor activity and the onset of of depressor activity. The epaxial and inter- the adductor muscles (Fig. 9A). In the slowest hyoideus muscles become active, on the av- strikes (gape cycles of a second or more), no erage, during the first half of the activity activity was recorded from the adductor period of the depressor mandibulae. The ac- mandibulae muscles. When the strike was tivity period of the interhyoideus muscle cor- 92 W.E. BEMIS AND G.V. LAUDER

k 2cm 3

b n Ix C m=

i anx

e

Fig. 6. Tracings from a film taken at 100 frames per second (every fifth frame traced) of Lepidosiren chewing on an earthworm. Arrow in frame e points to the characteristic dimple that forms under the mandible during strong crushing movements. The perpendicular depth of -this dimple is the “crush” measurement described in the text. See text for further discussion.

TABLE 1. Pairwise analyses of variance on five kinematic variables from the initial strike by Lepidosiren* Time to peak Prey Time to Time to peak Gape cycle cranial angle distance peak gape hyoid depression time (1) (2) (3) (4) (5) (1) - 0.11 2.54 22.95’ 25.81‘ (2) 1, 17 - 1.24 28.30’ 37.611 (3) 1, 18 1, 17 - 43.29l 63.511 (4) 1, 18 1, 17 1, 18 - 8.79’ (5) 1, 18 1, 17 1, 18 1, 18 - *All variables measured in milliseconds. Values above the diagonal are the F statistic for the ANOVA, while values below the diagonal are the degrees of freedom used in the test. ‘Significant at the P < .01 level. A VARIATION IN GAPE PROFILES FOR LEPIDOSIREN 1

h E "

LLW 0

h E "

WL .27 0 :::: :::: .12 0 .2 .4 .6 .B 1

h E " W .::;I < .::;I .25 .1 0 .2 .4 .6 .8 1

h f " W <

B GAPE, PREY DISTANCE, HYOID DEPRESSION. AN0 CRANIAL ANGLE IN LEPIDOSIREN 3

,. .65k I E " W c

,. f " +

+8 -:::I 0 -:::I >W .ls L 0 0 .15 .3 .45 .6 ,. E " K W0 E > 1

0 B 0 < z 2 n I: I: 0 .15 TIME .3(SECONOS) .45 .6

Fig. 7. AIVariation in the profile of gape distance dur- with maximum gape. The four kinematic measurements ing four feedings by one individual. B)Four kinematic are: Gape, the distance between the upper and lower measurements plotted against time to show the kine- jaw; Prey Dist, the distance from the prey to the plane matic pattern during capture of a worm. Note that peak of the gape; Hyoid Depr, the distance of hyoid depression; excursion of the hyoid is reached after peak gape, and Cran Ang, the angle of the head with respect to the body. that maximum cranial elevation is nearly coincident 94 W.E. BEMIS AND G.V. LAUDER

LEPIDOSIREN 1 (N = 6) LEPIOOSIREN 2 (N = 4)

ANGLE ANGLE I +-

PREY PREY

GAPE I GAPE HYOID +- HYOID -- CYCLE CYCLE

0 .2 .4 .6 .8 1 1.2 d ' .> .i' .6 ' .a i 1.2 TIME (SECONDS) TI ME (SECONDS)

LEPIOOSIREN 3 (N = 3) LEPIOOSIREN 4 (N = 6)

ANGLE ANGLE I

PREY PREY GAPE t GAPE HYOID t HYOID --I CYCLE CYCLE +- I

L. 1 Ail- A.2L- 0 .2 .4 .6 .8 1 1.2 7 I ME (SECONDS)

Fig. 8. Mean (vertical bar) and ranges (horizontal 1. The five variables are: Angle, time from the start of lines) for five kinematic variables in four individual mouth opening to maximum angle of the head on the lungfishes measured during the initial strike. Each panel vertebral column; Prey, time from the start of mouth represents a different individual. The four individuals opening until the prey has completely entered the mouth; differ in the relative time of maximum gape, prey en- Gape, time from the start of mouth opening to peak trance, and cranial angle, but show the same pattern of gape; Hyoid, time from the start of mouth opening to hyoid and gape timing. Pooled data for all four individ- peak hyoid depression; Cycle, total duration of the gape uals shown are analyzed in a pairwise fashion in Table cycle, from start of mouth opening to closing. relates with the slow and slight hyoid and b, is followed by depression of the hyoid elevation seen in Figure 7B. Two muscles are that moves the worm further into the mouth. active during the mouth-closing phase of the By frame d adduction of the jaws has started. gape cycle: the adductor mandibulae and the The strongest adduction occurred in frames rectus cervicis. The muscle with the most g through j, which show the posterior part of variable pattern of activity during the strike the adductor mandibulae muscle bulging was the geniothoracicus posterior (Fig. 10: above the posterior surface of the head. GQ). No well defined period of activity could Two entire bouts of chewing by two differ- be identified, and this muscle was active in ent individuals are shown in Figure 11. Fig- less than 33%of all experiments. ure 11A includes a prey capture, followed almost immediately by a chewing bout con- Chewing and constriction sisting of seven chewing cycles. Figure 11B Figure 6 illustrates the pattern of events shows a longer chewing bout consisting of in a single chewing cycle. This sequence is nine chewing cycles, followed by two con- 750 milliseconds long and the frames are 50 strictions and swallowing. milliseconds apart. The prey transport phase The gape profiles shown in Figure 11 are occurs first, and mouth opening, in frames a the best guide to interpreting these se- A. SLOW STRIKE 6. MEDIUM STRIKE C. FAST STRIKE

AMP

DM I I. I. IH 4 RCa 100 msec IM 100 maw - I - GTa GTP I RAO 2 1 I

GTP

Fig. 9. Differences in EMG patterns for different strike speeds. A) Slow strike; B) moderately rapid strike; C) rapid strike. In B, chewing on the prey begins just after the offset of activity in the first adductor mandibulae posterior muscle burst. EMG calibration bar = 200 microvolts. 96 W.E. BEMIS AND G.V. LAUDER

I

DM I AMP I -lB%z+

EP -I ‘*

In ‘-I

RCo

REP -

GTP I+ - I I 0 200 400 600 eon I000 TIME (MILLISECONDS) Fig. 10. Summary block diagram illustrating the pat- strikes. The ends of the bars indicate the mean onset tern of muscle activity during suction feeding. In this and offset times, while the thin lines extending from the and subsequent block diagrams, black bars represent ends indicate the standard error of that mean. The over- activity that occurred in more than 66% of all experi- lapping error bars for the geniothoracicus posterior mus- ments, shaded bars represent activity in more than 33% cle indicate that a pattern we believed to consist of two but less than 66% of all experiments, and white bars bursts is in fact not distinguishable from one highly indicate that a muscle was active in less than 33% of all variable burst of activity. See text for further discussion. experiments. This figure summarizes data from 34 quences. For each chewing cycle, the mouth the hyoid movement is studied. During the is opened and closed rapidly. A large gape first half of the sequence shown in Figure distance after mouth closing means that the 11B, each time the mouth is opened, the tooth plates are in contact with the food. hyoid is raised, expelling the prey slightly Thus, in the gape profile in Figure 11B, the out of the mouth. During the second half, lungfish made two chewing cycles before the this pattern reverses, and the worm is prey was transported to a position between brought back into the mouth. the tooth plates. In the next seven chewing The approximate strength of an addudion cycles (C3-C9),the prey was crushed between is reflected by the tracings for the crush vari- the tooth plates. In Figure llA, there are able. The first two chewing cycles in Figure fewer %uccessful” chews, that is, in some 11B show weak adductions, but the subse- cases, the food was not between the tooth quent seven adductions are larger. This is plates during peak adduction. comparable to patterns of electromyographic Figure 11 also documents the tendency of activity in the adductor muscles, in which Lepidosiren to transport prey items part way out of the mouth during a chewing bout. This is shown by the traces of prey distance. In Fig. 11. Representative kinematic patterns for ex- tended chewing bouts in two individuals. The measure- Figure 11B, the worm is expelled partially in ment labelled “crush” provides an indication of the time the fourth chewing cycle. It is transported the prey is being crushed between the tooth plates; fur- out a few more millimeters before each of the ther discussion is given in the text. A) A successful strike next two adduction phases. Then, the pattern is followed by seven chews; C1C7 indicate the adduction phases of the chewing cycles. B) This record is the final reverses, and the worm is transported back chewing bout of a longer feeding sequence. Nine adduc- into the mouth for the next two crushes. A tion phases (ClC9) are indicated, during most of which, similar pattern of partially expelling the prey the worm was successfully located between the tooth item during transport phases is shown in plates. The “stepped” appearance of the prey distance measurement clearly shows the consequences of the in- Figure 11A. tra-oral prey transport system, serving to move the prey The underlying basis for the patterns of in and out of the mouth during a chewing bout and prey transport is apparent if the record for expose it to the full action of the tooth plates. FEEDING MECHANICS OF LUNGFISH 97

A PREY CAPTURE AND CHEWING BY LEPIDOSIREN 1 c A E " wa L? I- Lr4W , -, "" 0 .5 1 1.5 2 2. 5 3 3.5 4

h E 0 w 1.2-

b- L3 .8-

.4- w> (L:[L 0- 1 -,

h E .45[ I w

K w 0 0 H 0 > 0 .5 1 1.5 2 2.5 3 3.5 4

h t I E " 1 3 KU

6 CHEWING AN0 SWALLOWING BY LEPIOOSIREN 2

L

A 0 .4- v c3 W CoNSTRIC a CONSTRIC 2

.11 I 0 1.5 3 4.5 6 7. 5

A E " + -ul 0 > n!d a

n E " na w

0 u 0 I

A E " 1 3 L1: U 98 W.E. BEMIS AND G.V. LAUDER

c

Q 0 [1: FEEDING MECHANICS OF LUNGFISH 99

I-ADDUCTIDN PHASE- -TRANSPORT PHASE AMP 7+ START OF NEXT CHEWING CYCLE AMo - RAO - 1 EP I OM I -- I IH -- I IM - I *-- RCo 1- I RCP - 1 GTO - I GTP - - I +zZBZ-+ZzmBm& I 0 200 400 600 800 1000 TIME (MILLISECONDS)

Fig. 13. Summary diagram of eledromyographic ac- food between the tooth plates, while the transport phase tivity during a chewing cycle. Conventions as in Figure reflects intra-oral movement of food. The three adductor 10. This figure summarizes data from 89 chewing cycles. muscles (AMP, AMa, RAO) are active only during the The adduction phase and the transport phase of chewing adduction phase, while many muscles, including in par- cycles are definable both kinematically and electromy- ticular the depressor mandibulae (DM),have a double- ographically. The adduction phase indicates crushing of burst pattern. the amplitude of addudor bursts in a given during a chewing bout. The depressor man- chewing bout starts out low, and builds up to dibulae shows two periods of activity in each a much larger level as the chewing bout pro- chewing cycle (Fig. 13: DM). A synchronous gresses (e.g., Fig. 5). burst of activity in the depressor accompan- The kinematics of constriction movements ies the adductor burst. The second, and usu- are shown at the end of the sequence in Fig- ally larger, period of activity in the depressor ure 11B. During constrictions, the mouth is mandibulae occurs between successive ad- kept tightly closed, the prey is located within ductor bursts. This activity indicates mouth the buccal cavity, and the hyoid is moved opening associated with the transport phase. sharply down and then slowly elevated. In The summary block diagram (Fig. 13) for this sequence, there are two constrictions. eledromyographic activity at all 11 record- During each, the amount of hyoid depression ing sites shows an obvious break between is steadily reduced. After these two constric- adduction and transport phases. During the tions, swallowing occurred. adduction phase, the three adductor muscles Figure 12 shows an electromyogram for a (AMP, AMa, RAO) are active. During trans- chewing bout consisting of eight chewing port, these muscles are silent. Many muscles cycles. The start of this bout is indicated by exhibit two bursts of activity comparable to slight activity in the depressor mandibulae the synchronous and asynchronous bursts of muscle. This activity indicates mouth open- the depressor mandibulae. However, it is ing and the beginning of a transport phase. usually the second burst that is the stronger After two such transport phases, during of the two. The most consistently active mus- which the food is positioned between the cle during transport is the depressor mandi- tooth plates, there is a small burst in the bulae, because mouth opening occurs in every adductor muscles. The subsequent, much transport phase. Several muscles act directly larger adductor bursts are correlated with or indirectly on the hyoid, tending to depress crushing of the food. These larger adductor or raise it. Those such as the recti cervicis bursts occurred regularly, with a period of tend to depress the hyoid, bring the food item about one second. There is some tendency farther into the mouth, and make the trans- (described below) for this rhythm to speed up port appear kinematically similar to initial 100 W.E. BEMIS AND G.V. LAUDER

UNSCALED EMG - ADDUCTOR MANDIBULAE POSTERIOR (AMP)

c ' II I

20

--*,,- ..?.., . * - = 5) 12 NUMBER SPIKE AMPLITUDE AMP (N

4- - MEAN SPIKE NUMBER * SPIKE AMPLITUDE 3- - DEPRESSOR MANDIBULAE (N = 5) 2-

1-

D. L A-

prey capture. The interhyoideus in particu- three small trains of spikes precede the start lar, but also the intermandibularis, tend to of major activity. Once the major period of elevate the hyoid region, forcing the prey activity does start, a degree of rhythmicity is item forward to position it between the tooth retained, as is particularly obvious from the plates. transformed EMG. The average time be- Figure 14 presents an analysis of the activ- tween these pulses of activity seen in the ities of the adductor and depressor muscles transformed EMGs is 19 milliseconds (N = averaged over five successive chewing cycles. 65; SD = 8.92), indicating an average pulse A sample electromyogram for the adductor frequency of 53 Hz. This pulse frequency is mandibulae posterior muscle from one of the considerably lower than the mean (N = 5 five cycles is presented at the top of the fig- bursts) power peak of 297 Hz determined by ure. A transformed version of this same myo- FFT analysis of the entire EMG. Note also gram, in which the number of spikes in a that the myogram builds up to a peak in given bin width has been multiplied times activity, and then declines before the end of the mean amplitude of the spikes in that bin the period of contraction. The trace below the (Beach et al., '82), is presented in the second transformed myogram is the average of five trace. Like most of the adductor bursts, activ- successive transformed adductor bursts. This ity in this muscle appears rhythmic. Thus, trace shows that there is a central peak in FEEDING MECHANICS OF LUNGFISH 101

TABLE 2. Three level nested ANOVA on the duration of electrical activity in the adductor mandibulae Dosterior muscle* Source of Degrees of Sum of Mean variation freedom squares square F Significance Among individuals 4 330,023 82,506 1.6 NS Among implants within indiv. 3 209,349 69,783 3.5 NS Among days within implants 7 136,871 19,553 6.0 P < ,001 Among bursts within days 74 240,800 3,254 Total 88 917.045

*Variance component (%I: Among individuals = 17.5; among implants = 30.8; among days = 24.1; and among bursts = 27.6. activity for all five of these myograms. Fi- ber within the bout and the length of time nally, the bottom trace in Figure 14 shows that the muscle is inactive. This is investi- the corresponding pattern of activity in the gated further in Figure ED, which shows depressor mandibulae for these five chewing the burst and interburst periods plotted cycles. There is no evidence of crosstalk in against cycle number within each of the six raw myograms of these two muscles, and we chewing bouts. The regression line through regularly recorded strong alternating activ- the points indicating the burst period is not ity in both muscles (e.g., Fig. 5B). Note that significantly different from 0 (r = -.26; t = the synchronous depressor bursts are not 1.59, NS). However, the slope of the line symmetrically located with respect to the ad- through the interburst periods is signifi- ductor bursts. Note also that the second pe- cantly different from 0 and is negative (r = riod of activity in the depressor mandibulae -.71; t = 5.83, p < .001). Thus, during an is much larger than the synchronous burst. average chewing bout, the later chewing This is interpreted as the major burst of ac- cycles tend to be more rapid, due to the tivity responsible for mouth opening. There shorter interburst period. is a one to one correlation between the tim- We performed three-level nested analyses ing of the rhythmic pulses of activity in the of variance on the periods of activity for sev- two muscles (r = .99; N = 37). eral muscles to determine the extent of vari- Activity in the adductor mandibulae mus- ation and to evaluate the repeatability of cle shows no overall pattern during a com- muscle activity patterns. In Table 2, we re- plete feeding sequence, but there is a port the results for the adductor mandibulae tendency for later bursts within a bout to posterior, the key muscle used in chewing: occur more frequently (Fig. 15). These data the analyses of other muscles (e.g., depressor were measured from electromyographic rec- mandibulae) show comparable results. Activ- ords of a long chewing sequence, consisting ity periods for 89 bursts of activity in the of a total of 46 chewing cycles in six separate adductor mandibulae posterior were mea- chewing bouts. The burst period (length of sured. These bursts were grouped by experi- time that the adductor mandibulae was ac- mental day within electrode implant and by tive) and interburst period (length of time implant within individual lungfkhes. Varia- between successive bursts of adductor man- tion within days accounted for nearly 28% of dibulae activity, unless this was the end of a the total variance. Variation among days ac- chewing bout, in which case no measurement counted for about one quarter of the total is reported) were measured in these 46 chew- variance and was significant. Although the ing cycles. The two period measurements are greatest percentage of variation (30.8%) was plotted against chewing cycle number (Fig. present among implants, this variation is not 15A,B), and against each other (Fig. 15C). significant, and the relatively low variation These graphs suggest that there is neither a (17.5%) among individuals was not signifi- consistent change in the pattern of activity cant. This analysis suggests that 1) there within a total feeding event (Fig. 15A,B), nor was a good deal of repeatability from individ- is there an obvious relationship between the ual to individual, and 2) that our electrode length of time that the adductor muscle is placements recorded consistent patterns of active and the length of time that it is inac- activity. The only significant variation is at tive (Fig. 15C). However, within each of the the level of days, which presumably reflects six chewing bouts, there is a significant re- small differences in the details of a particu- lationship between the chewing cycle num- lar feeding event, such as the prey size. 102 W.E.BEMIS AND G.V. LAUDER

PERIOOS OF ADOUCTOR ACTIVITY AN0 INACTIVITY DURING A FEEDING EVENT .35, I

x BOUT 2 1.25- BOUT 1 BOUT 4 ' BOUT 5 * * B~~~ 6 x f ** XI) .85YfX* * ** **

.45 , , *,I, , ,*, I I I I I , I > ,

r = .06

*I

ff *x

.05 .45 . 65 .85 1.05 1.25 1.45 INTERBURST TIME. SECONDS

PERIODS OF AOOUCTOR ACTIVITY AND INACTIVITY WITHIN BOUTS (N = 6) 1. 6-

1.4 - *

*+t INTERBURST IME, SECONOS 1.2- r = -.71 * 1- *\: *

.8-

.6- * *

.4- BURST TIME, SECONOS r = -.26 e e c e c c .2- m e e e e m e m f 0 e 0

01 0 1 2 3 4 5 6 CHEWING CYCLE NUMBER WITHIN A BOUT Fig. 15. Periods of adductor activity and inactivity. A) interburst duration, showing no relationship; D) burst Burst duration (time in seconds of adductor activity) in and interburst durations within single chewing bouts, the adductor mandibulae posterior muscle, plotted showing a significant relationship between the length of against chewing cycle number within an entire feeding the interburst and the number of the chewing cycle. sequence consisting of 46 chews; B) interburst durations Thus, later chews within a chewing bout tend to occur (time in seconds between successive adductor bursts) for closer together in time. these same 46 chews; C) burst duration as a function of FEEDING MECHANICS OF LUNGFISH 103

GTo -7- I GTP I I I 0 200 400 600 BOO 1000 TIME (MILLISECONOS>

Fig. 16. Summary diagram of the pattern of muscle constrictions, although at different amplitudes. Note the activity during constriction (buccal compression) follow- overlapping but not synchronous bursts in the adductor ing chewing bouts. Conventions for this figure follow and depressor mandibulae, and the extended activity in those of Figure 10. This figure summarizes data from 45 the interhyoideus (IH),rectus cervicis (RCp), and geni- constrictions. All muscles recorded are active during othoracicus (GTp).

Figure 16 diagrams patterns of electrical more constriction events. When chewing is activity during constriction. The most impor- complete, a final series of constrictions forces tant muscle during constriction movements the prey item into the esophagous, and swal- is the interhyoideus and it is used as the lowing ensues. reference muscle for this diagram. These The mechanics of each of these phases are eight muscles all showed significantly over- summarized in Figure 17. The white arrows lapping activity during constriction. Three of represent the movements seen externally. the muscles (the depressor mandibulae, in- The black arrows represent the approximate termandibularis, and geniothoracicus poste- lines of action of the muscles involved, as rior, Fig. 16: DM, IM, GTp), are active well as an estimate of the relative amplitude significantly later than the other muscles ex- of electromyographic activity. During suc- amined. The last half of depressor mandibu- tions, the lower jaw is opened by jaw-depress- lae activity does not overlap activity in the ing muscles, the depressor mandibulae and adductor mandibulae. rectus cervicis, and the hyoid is depressed by the large recti cervicis. During the adduction phase of chewing, the jaw-closing muscles DISCUSSION are strongly active, while most other compo- Behavioral patterns and functional nents are weakly active. During the trans- morphology port phase, the adductor muscles are inactive. We distinguish three basic behavioral pat- The lower jaw is abducted by the depressor terns in the feeding system of Lepidosiren. mandibulae, while the hyoid region is moved These are: 1) the strike or prey capture by up or down through the action of the recti suction; 2) chewing cycles, each consisting of cervicis or interhyoideus muscles. During an adduction phase and a transport phase; constriction (buccal compression), the mus- and 3) constriction or buccal compression. A cles elevating the hyoid region, principally given feeding event includes an initial strike, the interhyoideus muscle, are strongly ac- followed by a variable number of chewing tive. Other muscles, notably the adductor cycles arranged into chewing bouts. Usually, mandibulae, are active to prevent food from each chewing bout is terminated by one or escaping through the front of the mouth. 104 W.E. BEMIS AND G.V. LAUDER

A. SUCTION C. CHEWING - TRANSPORT

6. CHEWING - ADDUCTION D. CONSTRICTION

Fig. 17. Summary diagram showing the dominant features of the kinematic and electromyo- graphic patterns during four stages of feeding. White arrows indicate head movements visible externally, while black arrows indicate EMG activity. Width of the black arrows approximates the relative amplitude of EMG activity.

An especially noteworthy feature of muscle skinks feeding on snails. They suggest that activity during chewing is the pattern of because such a pattern produces graded force spike activity within adductor muscle bursts. development at the tooth row, it may reduce During the adduction phase of chewing, the shock impacts between opposing teeth. Irish adductor mandibulae muscles consistently ('83) reported that pulsed adductor activity show rhythmic pulses of electrical activity so occurred less frequently in the fish, Colos- that each overall burst is composed of shorter, soma, when fed on soft rather than hard food relatively high amplitude and high fre- items. Lungfishes are widely regarded as du- quency activity followed by lower level adiv- rophagic, yet the earthworm prey used in ity (Fig. 14A,B). The result is a "pulsed" this study scarcely qualify as hard. Never- pattern with an average frequency of 53 Hz, theless, we did find pulsed activity during which stands out sharply in contrast to the feeding on earthworms. It will be interesting average peak power frequency of the myo- to study the pulsed adductor burst patterns grams, 297 Hz. This pulsed pattern is suffi- for lungfish feeding on hard prey items. ciently consistent that averaging several There are few relevant data on the physio- transformed bursts does not decrease its clar- logical properties of fish muscles (e.g., Bone, ity (Fig. 140 During these adductor muscle '78; Johnston, '831, and the comparative data bursts, the antagonistic depressor mandibu- that do exist suggest that there can be impor- lae muscle is also active. It also exhibits a tant differences among taxa. Nevertheless, rhythmic pattern of pulses, with a one to one based on the data of Johnston ('80) €or skate correspondence to the adductor pulses. and cod fin muscle fibers, our observed pulse A pulsed pattern of EMG activity during frequency of 53 Hz indicates that the muscle crushing may reflect biomechanical or phys- probably is being stimulated at a rate suffi- iological properties advantageous in crush- cient to produce at least unfused tetanus (see ing (Irish, '83). Gans et al. ('85) documented also Gans and de Vree, '84). By stimulating a pulsed pattern of adductor activity in the muscle at about 50 Hz, the tension devel- FEEDING MECHANICS OF LUNGFISH 105 oped probably rises and falls, as in unfused A final important aspect of the functional tetanus. These and other ideas need further morphology of feeding in lungfishes is the study before the functional significance of mechanism of chewing and the process of the pulsed EMG pattern in lungfish adductor food transport within the oral cavity. During muscles is fully understood. a chewing bout, the prey is alternately A further noteworthy feature of the chew- crushed and transported, in a fashion remi- ing phase is the consistent presence of elec- niscent of generalized mammalian mastica- trical activity in antagonistic muscles. The tory systems (see examples in Gans et al., best example of this is the antagonistic activ- ’78; Hiiemae, ’78). Unlike mammals, how- ity in the depressor and adductor mandibu- ever, the mechanism of prey movement is lae muscles when prey are crushed (Fig. 14D). hydraulic, with water movement and its di- The first, or synchronous, burst of depressor rection of flow determined by the hyoid ap- mandibulae activity is a regular feature of paratus. As the mouth is opened during the our data, although it is usually of lower am- transport phase of a chew, the hyoid can be plitude than the second burst. Antagonistic elevated to push water and the prey ante- muscle activity in vigorous movements is riorly. At other times in a chewing bout, the common in musculoskeletal systems (Bas- hyoid acts in a similar manner to its role in majian, ’74), and perhaps reflects the stabili- suction feeding and draws water and the prey zation of joint articulations or increased posteriorly. Thus, it is the control of water control due to modulation of activity in movement by the hyoid apparatus that al- antagonists. lows positioning of prey between the tooth The nested ANOVA performed on the burst plates and provides control of prey position duration in the adductor mandibulae pro- in a manner analagous to the tetrapod vides considerable insight into the experi- tongue. mental procedure and into the variability Our results on the functional morphology present in the feeding mechanism. By parti- of mouth opening conflict with some of the tioning the variation in one variable (such as limited literature available on the feeding burst duration of a muscle) into three levels, systems of living lungfishes, particularly in variation among individuals, among elec- the case of the jaw opening system. Thomson trode implants on the same individual on (’69) realized the correct function of the de- different days, and among recording sessions pressor mandibulae muscle, but in the ab- on different days with the same electrode sence of functional data, did not elaborate on implants, it is possible to assess the effect of this point. Edgeworth (’35) considered that repeated experiments on the same individu- jaw opening in lungfishes was the result of als as well as to determine how much indi- geniothoracicus activity. This idea was rei- viduals differ from each other. Our results terated by Perkins (’72), who stated (pg. 69): (Table 2) show that although there is varia- “The m. geniothoracicus lowers the jaw dur- tion in our data owing to different electrode ing feeding; no other muscle appears to be implantations on the same individuals (many active at this time.” Perkins went on to sug- weeks apart), this variation is not statisti- gest that the depressor muscle is not active cally significant. There is also no evidence during feeding. However, neither Edgeworth from this analysis that the individuals we nor Perkins used electromyography to deter- studied differed significantly from each other. mine muscle activity patterns, and our re- The analysis does show, however, that there sults suggest that they misinterpreted func- is significant variation among days given tion on the basis of static morphology. It is that the same set of electrodes was used. This more difficult to resolve a difference between is consistent with the results of ShafTer and our study and McMahon’s (’69) electromy- Lauder (’85a) who found a significant varia- ographic study of respiration in the Afri- tion among days in many of the EMG vari- can lungfish, Protopterus aethiopicus. Mc- ables they used in a study of suction feeding Mahon found that the depressor mandibulae in aquatic ambystomatid salamanders. We muscle was only active during peak move- recommend this type of analysis in studies of ments of the lower jaw. Thus, he stated (’69, vertebrate functional morphology because of pg. 417): “Activity in the retractor (= depres- its heuristic value and the increased under- sor) mandibulae was always associated with standing of the sources of variation in the maximal depression of the lower jaw, as seen experimental data (also see ShafTer and Lau- in the ‘yawning movements’ discussed be- der, ’85b). low.” This conflicts with our finding that the 106 W.E. BEMIS AND G.V. LAUDER depressor was active every time the lower intact after 400 million years of vertebrate jaw was opened. evolution and extensive divergence in feed- ing form, function, and habit. Evolution of the feeding mechanism in One aspect of the feeding system of lepido- lower vertebrates sirenid lungfishes appears to be a specializa- A key component of our understanding of tion at least functionally similar to a feature vertebrate evolution is the analysis of both found in the feeding mechanism of salaman- morphological and functional patterns in the ders. Lepidosiren possesses a well-developed feeding mechanism. The definition of such muscle (Fig. 2) that is consistently active dur- patterns forms a necessary basis for discus- ing mouth opening. In recognition of this sions of the rate of evolution in the skull, the function, we term the muscle depressor man- origin of terrestrial feeding, and scenarios dibulae, in accordance with Thomson ('69) about morphological changes and the envi- but different from a widely used name, re- ronment. The recognition of such morpholog- tractor mandibulae (e.g., McMahon, '69; Fox, ical and functional patterns is difficult '65; Edgeworth, '35). Aquatic salamanders because general patterns become clear only possess a depressor mandibulae muscle in a after extensive comparative analyses of the topographically comparable position that is relevant clades. This study contributes to our also used in mouth opening (Shaffer and Lau- understanding of the evolution of the verte- der, '85a). Although the depressor mandibu- brate feeding mechanism by providing com- lae muscles of lungfkhes and salamanders parative data on a critical clade, the are both derived from the hyoid arch, they lungfkhes, and allowing the definition of are not regarded as homologues by embryol- primitive characteristics of the tetrapod feed- ogists (e.g., Fox, '63, '65; Edgeworth, '35). ing mechanism. The present data must be The basic developmental difference between considered in the light of other recent work lungfishes and salamanders is that the me- on patterns of evolution in the feeding mech- dial portion of the constrictor hyoideus sheet anism of aquatic vertebrates (salamanders, gives rise to the depressor muscle of lung- Shaffer and Lauder, '85a, '85b; coelacanths, fishes (Edgeworth, '35: pg. 97), while the de- Lauder, '80a; ray-finned fishes, Lauder, 'gob, pressor muscle of salamanders is derived '82, '85). from the dorsal (= levator hyoideus) portion The comparative data cited above on the of the constrictor hyoideus sheet (Edgeworth, major lower vertebrate clades indicate une- '35, pg. 102). quivocally that at least four aspects of the The depressor mandibulae muscle is simi- feeding mechanism are primitive for teleos- lar in size and position in Lepidosiren and tome fishes (reviewed in Lauder, '80): 1) ini- Protopterus. In the Australian lungfish, Nee tial prey capture occurs by suction feeding; ceratodus forsteri however, the muscle is ap- 2) the epaxial muscles produce cranial ele- parently much less well developed, origi- vation during mouth opening; 3) the hyoid nating from the ceratohyal and passing to its apparatus plays a major role in mediating insertion on the lower jaw deep to the oper- expansion of the mouth cavity and is one cular elements, instead of originating from biomechanical system involved in depressing their outer surface (Fox, '65; Perkins, '72). the mandible; and 4) peak hyoid excursion The size and position of the muscle in Neocer- occurs after maximum gape has been atodus is probably closer to the primitive con- achieved. This study confirms that all of dition for lungfkhes, because there has been these features also characterize Lepidosiren, a general tendency during dipnoan evolution strongly suggesting that the four primitive to reduce both the number and size of the characteristics of the teleostome feeding opercular and subopercular elements (e.g., mechanism are also primitive for tetrapods. Uranolophus-Scaumenacia-Neoceratodus- The fundamental biomechanical systems in- Lepidosiren), and there is no indication of a volved in initial prey capture exhibit re- muscle attachment on the outer face of the markable conservatism. For example, ab- opercular and subopercular in the descrip- duction of the mandible during suction feed- tions of phylogenetically primitive dipnoans ing in primitive ray-finned fishes, coela- (e.g., Uranolophus, Denison, '68; Griphogna- canths, lungfishes, and aquatic salamanders thus, Miles, '77). The enlarged depressor occurs at least in part by posteroventral mandibulae of Lepidosiren and Protopterus movement of the hyoid and the transmission is, therefore, a specialized condition within of this movement to the retroarticular pro- the Dipnoi, and is yet another synapomorphy cess of the mandible by the mandibulohyoid indicative of the close relationship between ligament. This mechanical system remains these two genera. Even though the muscle is FEEDING MECHANICS OF LUNGFISH 107 unknown in outgroup osteichthyans, the em- mouth opening is best interpreted as having bryological considerations discussed above been achieved by a combination of cranial render it unlikely that the depressor man- elevation caused by the epaxial muscles, and dibulae is a synapomorphy of lungfishes and mandibular depression mediated by postero- tetrapods. The convergent development of ventral excursion of the hyoid. The mandi- depressor mandibulae muscles in lungfishes bulohyoid ligament would transmit hyoid as well as several lineages of tetrapods is movement to the mandible as in other lower striking evidence of the functional impor- vertebrates. No evidence available from fos- tance of specialized jaw opening systems. sils contradicts these inferences, but neither The salient specialization in the feeding does fossil evidence provide much critical in- mechanism of Lepidosiren does not lie in ini- formation on the presence of fleshy lips that tial prey capture, but rather in the extensive occlude the margins of the gape (as in Protop intra-oral processing of food after it has been terus and Lepidosiren), or on the possible captured. Movement of the prey in and out of presence of a depressor mandibulae muscle. the mouth accompanied by strong adduction Further progress in interpreting the func- of the jaws to crush prey between the mas- tional characteristics of extinct lower verte- sive tooth plates is an aspect of the feeding brates awaits a more precise understanding mechanism not found in most lower verte- of the interrelationships between form and brates. Associated with the tooth plates and function in living clades. crushing behavior is the hydraulic transport of food that allows positioning of food within ACKNOWLEDGMENTS the oral cavity. In many ways, this hydraulic food transport system in lungfishes performs This study was supported by grants from NIH 1F32 DE05352 to W.E.B., and NSF DEB analagous functions to those of the tongue in 81-15048 and the Whitehall Foundation to tetrapods (e.g., Hiiemae, et al., ’78; Bramble and Wake, ’85; Gorniak, et al., ’82; Smith, G.V.L. G.V.L. also received support from the ’84). Water movement mediated by the hyoid Louis Block Fund (University of Chicago). apparatus positions prey between teeth, aids We thank Steve Barghusen, Cathy Smither, in swallowing (constriction or buccal com- and Peter Wainwright for assistance in prep- pression in Lepidosiren), and can act to re- aration of the paper, and for computer pro- move unwanted particles from the oral cav- gramming the data analysis. Peter Wain- ity. Although other fishes have been wright, Fran Irish, Carl Gans, and Marvalee documented to use water flow through the Wake provided helpful comments on the oral cavity to position food (see, for example, manuscript. Lauder, ’81), ray-finned fishes have not yet been shown to do the extensive processing of LITERATURE CITED food that Lepidosiren does with the oral teeth. Basmajian, J.V. (1974) Muscles Alive, Their Functions Hiiemae et al. (’78: 206) stated that Trans- Revealed Through Electromyography, Ed. 3. Balti- more: Williams and Wilkins. port is the fundamental mechanism of feed- Beach, J., G.C. Gorniak, and C. Gans (1982) A method ing and the hyoid apparatus is a fundamental for quantifying electromyograms. J. Biomech. 15:611- part of that mechanism,” a conclusion as ap- 617. propriate for Lepidosiren as it was for the Bemis, W.E. (1984a) Morphology and growth of lepidosi- renid lungfish tooth plates (Pisces: Dipnoi). J. Morphol. mammalian feeding systems they discussed, 179:73-93. despite the major anatomical differences be- Bemis, W.E. (198413) Paedomorphosis and the evolution tween tongue and hydraulic transport sys- of the Dipnoi. Paleobiology 10:293-307. tems. Beman, D.S. (1979) Gnathorhizu bothrotreta (Ostei- chthyes: Dipnoi) from the Lower Permian Abo Forma- This paper and the comparative data now tion of New Mexico. Ann. Carnegie Mus. 48:211-230. available on other lower vertebrate clades Bertmar, G. (1966) The development of the skeleton, set the stage for a new examination of the blood-vessels and nerves in the dipnoan snout, with a feeding mechanism in “rhipidistian” fishes. discussion on the homology of the dipnoan posterior nostrils. Acta Zool. Stockh. 47:82-150. The presence of several features of the initial Bishop, I.R. and G.E.H. Foxon (1968) The mechanism of prey capture mechanism in all living lower breathing in the South American lungfish, Lepidosiren vertebrate clades studied to date indicates parudoxa; a radiological study. J. Zool. Lond. 154:263- that Eusthenopteron, Rhizodus, Glyptolepis, 271. Bone, Q. (1978)Locomotor muscle. In W.S. Hoar and D.J. and their relatives (whatever their phyloge- Randall (eds): Fish Physiology, Vol. 7. New York: Aca- netic position) shared several key character- demic Press, pp. 361-424. istics of the lower vertebrate feeding mech- Bramble, D. and D.B. Wake (1985) Feeding mechanisms anism. Although there is no evidence for the of lower tetrapods. In M. Hildebrand, D. Bramble, K. presence of a depressor mandibulae muscle, Liem, and D. Wake (eds): Functional Vertebrate Mor- 108 W.E. BEMIS AND G.V. LAUDER

phology. Cambridge: Harvard University Press, pp. New York Praeger, pp. 36-67. 230-261. Lankester, E.R. (1896) On the Lepidosiren of Paraguay, Bridge, T.W. (1898) On the morphology of the skull in and on the external characters of Lepidosirerz and Prc- the Paraguayan Lepidosiren and in other dipnoids. topterus. Trans. Zool. Soc. Lond. 14:12-24. Trans. Zool. Soc. Lond. 14:325-376. Lauder, G.V. (1980a) The role of the hyoid apparatus in Campbell, K.S.W. and R.E. Barwick. (1983) Early evolu- the feeding mechanism of the coelacanth Latimeria tion of dipnoan dentitions and a new genus Speonesy chalumnae. Copeia 198O:l-9. drwn Mem. Ass. Australas. Palaeontols. l:17-49. Lauder, G.V. (198Ob) Evolution of the feeding mecha- Carter, G.S. and L.C. Beadle (1930) Notes on the habits nism in primitive actinopterygian fishes: A functional and development of Lepidosiren paradoxa J. Linn. Soc. anatomical analysis of Polypterus, Lepisosteus, and Lond., Zool. 373197-203. Amia J. Morphol. 163:283-317. Corbet, P.S. (1961) The food of non-cichlid fishes in the Lauder, G.V. (1981) Intraspecific functional repertoires Lake Victoria Basin, with remarks on their evolution in the feeding mechanism of the characoid fishes Lebi- and adaptations to lacustrine conditions. Proc. 2001. asina, Hoplias, and Chalceus. Copeia 1981:154-168. Soc. Lond. 136:l-101. Lauder, G.V. (1982) Patterns of evolution in the feeding Denison, R.H. (1968) Early Devonian lungfishes from mechanism of actinopterygian fishes. Am. Zool. Wyoming, Utah, and Idaho. Fieldiana, Geol. 27:353- 22,275-285. 413. Lauder, G.V. (1983a)Functional design and evolution of Dingerkus, G. and L.D. Uhler (1977) Enzyme clearing of the pharyngeal jaw apparatus in euteleostean fishes. alcian blue stained whole small vertebrates for dem- Zool. J. Linn. Soc. Lond. 77:l-38. onstration of cartilage. Stain Technol. 52229-232. Lauder, G.V. (198313) Functional and morphological bases Edgeworth, F.H. (1926) On the development of the cra- of trophic specialization in sunfishes (Teleostei, Cen- nial muscles in Protopterus and Lepidosiren Trans. trarchidae). J. Morphol. 178:l-21. Roy. SOC.Edinburgh. 54:719-734. Lauder, G.V. (1985) Functional morphology of the feed- Edgeworth, F.H. (1935) The Cranial Muscles of Verte- ing mechanism in lower vertebrates. In H.-R. Duncker brates. Cambridge: Cambridge Univ. Press. and G. Fleischer (eds): Functional Morphology of Ver- Fox, H. (1963) The hyoid of Neoceratodus and a consid- tebrates. New York Springer-Verlag, (in press). eration of its homology in urodele amphibia. Proc. Zool. Luther, A. (1913) Uber die von N. Trigeminus versorgte Soc. Lond. 142:803-810. Muskulatur der Ganoiden und Dipneusten. Acta SOC. Fox, H. (1965) Early development of the head and phar- Sci. Fennicae 41:3-72. YM of Neoceratodus with a consideration of its phylo- McMahon, B.R. (1969) A functional analysis of the geny. J. 2001. Lond. 146:470-554. aquatic and aerial respiratory movements of an Afri- Gans, C., F. de Vree, and G.C. Gorniak (1978) Analysis can lungfish, Protopterus aethiopicus, with reference of mammalian masticatory mechanisms: Progress and to the evolution of the lung ventilation mechanism in problems. Zbl. Vet. Med. C. Anat. Histol. Embryol. vertebrates. J. Exp. Biol. 51:407-430. 7:226-244. Miles, R. (1977) Dipnoan (lungfish) skulls and the rela- Gans, C. and F. de Vree (1984)Temporal summation in tionships of the group: A study based on new species the adductor musculature of the lizard Trachydosau- from the Devonian of Australia. Zool. J. Linn. Soc. rus: Treppe and tetanus in the crushing of snails. Am. Lond. 61:l-328. Zool. 24t108A (abstract). Owen, R. (1839) Description of the Lepidosiren annec Gans, C., F. de Vree, and D. Carrier (1985)Usage pattern tens. Trans. Linn. Soc. 18:327-361. of the complex masticatory muscles in the shingleback Parker, W.K. (1892) On the anatomy and physiology of lizard, Trachydosaurus rugosus: A model for muscle Protopterus anneetens. Trans. Roy. Irish Acad. 30:109- placement. Am. J. Anat. 173:219-240. 230. Gorniak, G.C., and C. Gans (1980)Quantitative assay of Perkins, P.L. (1972) Mandibular mechanics and feeding electromyograms during mastication in domestic cats groups in the Dipnoi. Unpublished doctoral disserta- (Felis catus). J. Morphol. 163:253-281. tion, Yale University, New Haven, Connecticut. Gorniak, G.C., H.I. Rosenberg, and C. Gans (1982) Mas- Rosen, D., P.L. Forey, B.G. Gardiner, and C. Patterson tication in the tuatara, Sphenodon punctatus (Reptilia: (1981) Lungfkhes, tetrapods, paleontology, and ple- Rhyncocephalia): Structure and activity of the motor siomorphy. Bull. Am. Mus. Nat. Hist. 167:159-276. system. J. Morphol. 171:321-353. Schultze, H.-P., and L. Trueb (1981) Basic Structure and Hiiemae, K.M. (1978)Mammalian mastication: A review Evolution of Vertebrates, Vols. I, 11. E. Jarvik. 1980. of the activity of the jaw muscles and the movements London: Academic Press. (review). J. Vert. Paleontol. they produce in chewing. In P.M. Butler and K.A. 1:389-397. Joysey (eds): Studies on the Development and Function Shaffer, H.B., and G.V. Lauder (1985a) Aquatic prey of Teeth. New York Academic Press, pp. 359-398. capture in ambystomatid salamanders: patterns of Hiiemae, K.M., A.J. Thexton, and A.W. Crompton (1978) variation in muscle activity. J. Morphol. 283:273-284. Intra-oral food transport: The fundamental mechanism Shaffer, H.B., and G.V. Lauder (1985b) Patterns of vari- of feeding. In D. Carlson and J. McNamara, eds. Mus- ation in aquatic ambystomatid salamanders: kinemat- cle Adaptation in the Craniofacial Region. Ann Arbor: ics of the feeding mechanism. Evolution. 39.33-92. Univ. of Michigan Press, pp. 181-208. Smith, K.K. (1984) The use of the tongue and hyoid Humason, G.L. (1972)Animal Tissue Techniques, Ed. 3. apparatus during feeding in lizards (Ctenosaura simi- San Francisco: W.H. Freeman. lis and Tupinambis nigropunctatus). J. Zool. Lond. Irish, F. (19831 The dynamics of seed-eating in characi- 202:115-143. form fish, Colossoma. Am. Zool. 23:1027 (abstract). Sokal, R.R., and F.J. Rohlf (1981) Biometry. San Fran- Jollie, M. (1982)Ventral branchial musculature and syn- cisco: W.H. Freeman. apomorphies questioned. Zool. J. Linn. SOC. Lond. Thomson, K.S. (1969) The biology of the lobe-finned 74:35-47. fishes. Biol. Rev. Camb. Phil. SOC.44:91-154. Johansen, K. and D. Hanson (1968) Functional anatomy Wassersug, R.J. (1976)A procedure for differentia! stain- of the hearts of lungfishes and amphibians. Am. Zool. ing of cartilage and bone in whole formalin fixed ver- 8: 191-2 10. tebrates. Stain Technol. 51:131-134. Johnston, LA. (1980) Contractile properties of fish fast Wiley, E.O. (1979) Ventral gill arch muscles and the muscle fibers. Marine Biol. Letters. 2:323-328. interrelationships of gnathostomes, with a new classi- Johnston, I.A. (1983)Dynamic properties of fish muscle. fication of the Vertebrata. Zool. J. Linn. SOC.Lond. In P.W. Webb and D. Weihs (eds): Fish Biomechanics. 67:149-179.