Mechanical Properties of the Dorsal Fin Muscle ofSeahorse (Hippocampus) and Pipefish (Syngnathus)

MIRIAM A. ASHLEY-ROSSn

Department of Biology, Wake Forest University, Winston-Salem, North Carolina27109

ABSTRACT The dorsal and pectoral fins are the primary locomotor organs in seahorses(Hippocampus) and pipefish (Syngnathus). The small dorsal fins beat at high oscillatory frequenciesagainst the viscous medium of water. Both species are able to oscillate their fins at frequencies likelyexceeding the point of flicker fusion for their predators, thus enhancing their ability to remaincryptic. High-speed video demonstrated that seahorse dorsal fins beat at 30–42Hz, while pipefishdorsal fins oscillate at 13–26Hz. In both species, the movement of the fin is a sinusoidal wave thattravels down the fin from anterior to posterior. Mechanical properties of seahorse and pipefish dorsalfin muscles were tested in vitro by the work loop method. Maximum isometric stress was 176.1 kN/m2 in seahorse and 111.5 kN/m2 in pipefish. Work and power output were examined at a series offrequencies encompassing the range observed in vivo, and at a number of strains (percent lengthchange during a contractile cycle) within each frequency. At a given strain, work per cycle declinedwith increasing frequency, while power output rose to a maximum at an intermediate frequency andthen declined. Frequency and strain interacted in a complex fashion; optimal strain was inverselyrelated to cycle frequency over most of the frequency range tested. Seahorse dorsal fin muscle wasable to generate positive work at higher cycling frequencies than pipefish. Both species producedpositive work at higher frequencies than have been reported for axial and fin muscles from other fish.J. Exp. Zool. 293:561–577, 2002. r 2002 Wiley-Liss, Inc.

Locomotion via movements of the median finshas arisen independently multiple times in theevolution of fishes. The methods of median finpropulsion are as varied as the taxa that employthem (reviewed in Lindsey, ’78). Locomotion bydorsal fin undulation is classified as amiiform(exemplified by Amia calva, but also seen in manymormyrids, Gymnarchus, and syngnathids, in-cluding Syngnathus and Hippocampus). Gymnoti-form locomotion is characterized by undulations ofthe anal fin only (seen in gymnotids and notopter-ids). Undulation of both dorsal and anal fins isreferred to as balistiform (seen in triggerfish andtheir relatives, some cichlids, centriscids, andflatfish). Finally, tetraodontiform locomotion isdefined as oscillations of short-based dorsal andanal fins (seen in pufferfish and ostraciids).Median fin propulsion is associated with relativelyslow, but precise, locomotion in complex habitatssuch as coral reefs and obstacle-strewn stream orsea beds (Lindsey, ’78; Webb, ’82; Videler, ’93).Most fish that use median fin propulsion generatelow-frequency waves (E2Hz) of high amplitude inthe fins, allowing them to swim with highhydrodynamic efficiency (defined as useful power

output divided by power input to move the fin;Blake, ’80). The family Syngnathidae, comprisingthe seahorses and pipefish, is an exception:Members of this group undulate their dorsal finsat very high frequencies (440Hz in seahorses;Breder and Edgerton, ’42; Blake, ’76) with lowamplitude, leading to reduced efficiency and veryslow swimming speeds (Blake, ’80).

Because the swimming speeds of seahorses andpipefish are slow, one may expect that themusculature powering dorsal fin movementswould be relatively weak. However, the dorsal finmuscle is required to contract against substantialresistance as it beats the fin back and forththrough the viscous medium of water. Further-more, the muscle must contract at a highfrequency while performing positive work againstthe environment. Indeed, the oscillation is so rapid

Grant sponsor: Grass Foundation (Grass Fellowship in Neurophy-siology at Woods Hole); Grant sponsor: National Science Foundation;Grant number: IBN-9813730.

nCorrespondence to: Miriam A. Ashley-Ross, Department ofBiology, Box 7325, Wake Forest University, Winston-Salem, NC27109. E-mail: rossma@wfu.edu

Received 16 November 2001; Accepted 22 May 2002Published online in Wiley InterScience (www.interscience.wiley.

com). DOI: 10.1002/jez.10183

r 2002 WILEY-LISS, INC.

JOURNAL OF EXPERIMENTAL ZOOLOGY 293:561–577 (2002)

Fig. 1. Cleared and stained specimens ofH.hudsonius (A) and S. fuscus (B) showing theskeletal supports of the dorsal fin. Surfacescales of S. fuscus over the proximal radialshave not been removed; proximal radials arevisible through the translucent scales. Boneappears red, cartilage appears blue. The dorsalfin consists of a series of 16–20 bony raysconnected by a flexible membrane. The fin raybases are supported by a series of midlineskeletal elements radiating dorsally from thevertebral column; in order, these are theproximal radials, middle radials, and distalradials, followed by the fin ray base itself. Theproximal radials are bony, while the middleand distal radials are cartilaginous. The mid-dle radials are fused together into a rigidbeam-like structure in the seahorse; in pipe-fish, the middle radials do not form a singlecontinuous structure, but individual middleradials are bound together by connectivetissue. Scale bars in both panels are 0.5 cm.

Fig. 2. Anatomy of the dorsal fin muscle in the seahorse,Hippocampus. The box over the seahorse on the right side ofthe figure indicates the area that is enlarged on the left side ofthe figure. Muscle is colored yellow. The dorsal fin muscleconsists of a layer of inclinator muscle slips and a layer ofdepressor muscle slips. The inclinator muscles arise from thelateral processes of the vertebrae and insert via a narrowtendon onto the fin ray base at approximately the midpoint of

the distal radial upon which the fin ray base rides. Thedepressor muscles resemble the inclinator muscles in appear-ance; they are located deep to the inclinators, arise from thevertebrae, and narrow to long tendons that insert on theposterolateral fin ray base. For clarity, only the inclinatormuscles are shown. Anatomy of the dorsal fin and itsassociated muscle is similar in the pipefish. Scale bar, 1 cm.

that it exceeds the flicker fusion frequency of thehuman eye, and likely those of predators as well(Blake, ’76). The dorsal fin is thus renderedeffectively invisible as it propels the animal.In vertebrates, high frequency muscles are

associated with at least three activities: soundproduction, eye movements, and locomotion.Sound producing and ocular positioning musclestend to have the faster contraction times. In thesound producing group are the bat cricothyroidmuscle, the midshipman sonic muscle, the toadfishsonic muscle (all with in vivo contraction frequen-cies Z100Hz; Fawcett and Revel, ’61; Revel, ’62;Fine, ’89; Bass and Marchaterre, ’89; Bass, ’90;Fine et al., ’90; Rome et al., ’96, ’99, respectively),and rattlesnake shaker muscle (contraction fre-quency E50Hz; Clark and Schultz, ’80; Schultz etal., ’80). These muscles are all thought to contractnearly isometrically and to do little mechanicalwork in vivo. Likewise, muscles responsible for eyemovements have very fast contraction times(o10ms), but produce little force, and hencesmall amounts of mechanical work (Cooper andEccles, ’30; Close and Luff, ’74).In contrast, fast locomotory muscles, such as

hummingbird pectoral muscle (30–78Hz; Green-ewalt, ’75) and sea horse dorsal fin muscle(430Hz; Breder and Edgerton, ’42), tend to havelower frequencies of contraction. However, thesemuscles must produce mechanical work and powerif they are to propel their bearer about. Humming-bird pectoral muscles are responsible for produ-cing sufficient power to allow sustained hovering,one of the most energetically expensive locomotoractivities (Suarez et al., ’91; Withers, ’92). Assum-ing a Q10 value of 2, when brought to the sametemperature the contraction frequency of thedorsal fin musculature of the sea horse would besimilar to that of hummingbird pectoral muscle.Seahorse locomotion has been previously stu-

died from the standpoint of anatomy of thelocomotor apparatus (Consi et al., 2001), kine-matics using high-speed video (Breder and Edge-rton, ’42; Blake, ’76), and hydrodynamic theory(Blake, ’76). Consi et al. (2001) described in detailthe anatomy of the dorsal fin of seahorses,including the joints between skeletal elements ofthe fin (Fig. 1). Transverse movements of the finray are allowed at the joint between the middleand distal radials (with the fin ray base ridingpassively on the distal radial), while longitudinalmovements of the fin ray (elevation and depres-sion) occur at the joint between the fin ray baseand the distal radial (Consi et al., 2001). Two

muscles move each individual fin ray on either sideof the animal: an inclinator muscle and adepressor muscle (collectively referred to as thedorsal fin muscle; Fig. 2). The inclinator musclesof seahorses and pipefish are likely derived fromthe dorsal fin elevator muscles of more typical fish(Winterbottom, ’74; Geerlink and Videler, ’74;Consi et al., 2001).

Properties of the dorsal fin muscle of seahorseshave been investigated by examination of sub-cellular anatomy and isometric properties. Pub-lished ultrastructural studies of sea horse dorsalfin muscle have demonstrated polyneuronal in-nervation similar to other fish muscles (Smith etal., ’88), large numbers of mitochondria (makingup to 10% of a muscle cell’s volume; Smith et al.,’88), and an extensive sarcoplasmic reticulum(occupying approximately 28% of the fiber volume;Bergman, ’64b; Smith et al., ’88) similar to that ofother fast muscles. The sarcoplasmic reticulumshows a highly ordered distribution within thecell, with triads situated at the Z lines (Smith etal., ’88). Dorsal fin muscle also shows an excep-tionally high ratio of sarcoplasm volume tomyofibril volume (Bergman, ’64b). Investigationof the contractile capabilities of the dorsal finmuscle of Hippocampus hudsonius has thus farbeen limited to measurements of twitch contrac-tion time and fusion frequency (Bergman, ’64a).No studies have yet addressed the mechanicalproperties of the dorsal fin muscle under condi-tions resembling those in vivo. Therefore, the goalof this study was to characterize the work andpower output capabilities of the dorsal fin muscleof the seahorse, Hippocampus, and its closerelative, the pipefish, Syngnathus, as a first stepin understanding the changes in muscle composi-tion that must have occurred in the evolutionof these specialized, high-frequency dorsal finmuscles.

MATERIALS AND METHODS

Animals

Northern pipefish, Syngnathus fuscus (12.6–18.8 cm standard length), were obtained from theMarine Resources department of the MarineBiological Laboratory at Woods Hole, MA. Theywere maintained in 80 l tanks with flow-throughseawater and fed adult brine shrimp several timesper day.

Two species of seahorses, Hippocampus hudso-nius and H. zosterae, were obtained through theaquarium trade, primarily from Florida-based

WORK AND POWER OF DORSAL FIN MUSCLE 563

vendors. Specimens ofH. hudsonius ranged in sizefrom 14.9–16.5 cm standard length which, due tothe unusual body posture of seahorses, wascalculated as the sum of two measurements: thedistance from the tip of the snout to the posteriorside of the base of the cornet, and the distancefrom the posterior side of the base of the cornet(see Fig. 2) to the tip of the uncurled tail. Dwarfseahorses, H. zosterae, were 2.4–3.3 cm standardlength. Seahorses were kept either in 80 l tankswith flow-through seawater (all H. zosterae, someH. hudsonius) or in 40 l aquaria filled withartificial seawater (Instant Ocean; some H. hud-sonius). Seahorses were fed a mixture of adultbrine shrimp and ghost shrimp (H. hudsoniusonly) several times per day. All fish, both sea-horses and pipefish, were used for experimentswithin one week of arrival in the laboratory.

High-speed video of fin movements

Fin movements of pipefish and seahorses werevideotaped at 500 frames/sec using a Kodak Ekta-pro EM1000 high-speed video system equippedwith a zoom lens. As none of the species involvednaturally maintains steady swimming, sequenceswere recorded as the fish moved freely in aquaria.Only periods of steady fin movement weresampled, and 3–13 fin beat cycles were recordedfor each individual. Two H. hudsonius, two H.zosterae, and four S. fuscus were videotaped. Finbeat frequencies were calculated from the videos.

In vitro measures of work and poweroutput

Only S. fuscus and H. hudsonius were used forin vitro work, as H. zosterae dorsal fin musclesproved inconveniently small. Fish were firstanesthetized by immersion in a solution of tricainemethane sulfonate (MS-222; 1.5 g/l) prepared inseawater or artificial seawater until the rightingreflex disappeared. Fish were then decapitatedand the rest of the body was cut away from thesection containing the dorsal fin. The abdominalwall (skin and musculature) was cut away, and theinternal organs in this region were removed,leaving the dorsal portion of the body containingthe dorsal fin, vertebral column, epaxial muscu-lature, and the bony plates. This remainingsection of the body was then placed in a dissectingdish under Ringer’s solution. The Ringer’s solu-tion contained, in mM: NaCl 145, KCl 4, CaCl2 2.4,MgSO4 1.7, MgCl2 1.5, Na2HPO4 2.5, KH2PO4 0.4,D-glucose 10, HEPES 10. pH was adjusted to 7.6 at

201C with KOH (Knapp and Dowling, ’87). Therest of the preparation was carried out under adissecting microscope. The section was pinned tothe bottom of the dissecting dish so that the dorsalfin pointed up, and incisions were made with ascalpel on either side of the dorsal fin through thebony plates, taking care not to cut into theunderlying muscle too deeply. Cuts were madewith scissors through the bony plates on the sidesof the section, just dorsal to the lateral processes ofthe vertebrae. Then the bony ‘‘exoskeleton’’ wasdeflected laterally with forceps, and the myotomalmuscle was cut away from the connective tissueoverlying the dorsal fin muscle, eventually freeingthe dorsolateral portion of the body section to beentirely discarded. The remaining part of thepreparation was composed of the dorsal fin,vertebral column, and the dorsal fin muscle onboth sides of the vertebral column. The sheet ofconnective tissue that overlies the dorsal finmuscle was carefully removed, finally exposingthe dorsal fin muscle, which is composed of a seriesof parallel slips of inclinators and depressorsconnected by tendons to the bases of the dorsalfin rays (Fig. 2). Next, the dorsal fin muscle on oneside of the body was completely removed, exposingthe proximal radials that support the dorsal fin.Very fine-tipped spring scissors were used tocarefully cut through the proximal radials, thusfreeing the ends of the muscle slips to moverelative to one another without the bony strutbetween them (which normally turns the linearcontraction of the dorsal fin muscle into lateralmovement of the dorsal fin rays; see Fig. 3A). Themuscle slips were examined for any damage, and aseries of 4–6 intact slips (composed of bothinclinators and depressors) were selected, alwaysnear the middle of the dorsal fin. This series ofslips, together with the associated fin rays andsection of vertebral column, was cut away fromthe surrounding tissues and transferred into aPlexiglas experimental chamber containing 40mlof Ringer’s solution. Chamber temperature wasmaintained at 201C by a water bath circulatingfluid through a channel surrounding the experi-mental chamber. The Ringer’s solution wasaerated throughout the experiment. The lateralprocesses of the vertebrae on the side opposite thedorsal fin muscle were fixed in a clamp at thebottom of the chamber, and the fin rays weregathered together and tied with silk suture ontothe arm of a servomotor ergometer (CambridgeTechnology, Cambridge, MA and Aurora Scienti-fic, Ontario, Canada; model 300B) that also

M.A. ASHLEY-ROSS564

functioned as a force transducer. One advantage ofthe arrangement of the dorsal fin structures isthat the middle radials, which articulate with thedistal radials and fin ray bases, are fused or boundtogether to form a rigid beam, which permits thecontribution of multiple muscle slips to be mea-sured without introducing potential complicationsof the outside slips contracting at an angle to theinner slips (see Fig. 3B).The muscle was stimulated using a Grass S88 or

S8800 stimulator connected to silver plate electro-des that were placed in the bath on either side ofthe muscle. Force and position, both sensed by theergometer, were recorded using a custom-writtencomputer program that also triggered the stimu-lator. Muscle length, stimulus strength, andstimulus frequency were adjusted to produce

maximal isometric tetanic force. Typical valuesfor stimulation frequency necessary to producefused tetani were 100–120Hz, at a voltage in-tensity of 80 V through the bath electrodes.Maximal twitch and tetanic force were recorded,and the muscle length (Lo) was measured througha dissecting microscope. Mechanical work andpower output was measured using the work loopmethod (Josephson, ’85), in which the muscle wassubjected to symmetrical sinusoidal lengthchanges while being stimulated phasically duringthe length change cycle. Plotting muscle forceversus length over a cycle of oscillation results in aloop-shaped trace. Because work is the product offorce and distance, the area enclosed by the loop isthe work performed by the muscle during thatcycle. Mechanical work and power output (workper cycle times oscillation frequency) were mea-sured at frequencies that encompassed the rangeof fin oscillation observed in the high-speed videorecordings. For pipefish, the oscillation frequen-cies used were 5, 10, 15, 20, 25, and 30Hz. Forseahorses, frequencies were 20, 30, 40, and 50Hz.Several different strains (peak-to-peak musclelength change as a percentage of Lo) were used.Pipefish were tested at strains of 2.5%, 5%, 10%,15%, and 20% of Lo. Preliminary experiments onseahorse muscle demonstrated rapid deteriorationof the muscle at 20% strain, so a more restrictedrange was used: 2.5%, 4%, 5%, 6%, 8%, 10%, 12%,and 15% of Lo. The phase and duration of thestimulus train were optimized to produce maximalwork per cycle at each frequencyþstrain combina-tion. The order in which frequencies were testedwas random; however, the order in which strainswere tested within each frequency was alwaysfrom smallest to largest. This was done due toconcern that the longest strains might causedamage to the muscle. The computer programgenerated a series of four consecutive lengthchange cycles, and work per cycle was measuredfor cycle 3. Trials were separated by two minutes.Three measurements were made at each frequen-cyþstrain combination for each individual. Theexperiment was ended when isometric tension fellbelow 90% of the initial value. The Ringer’ssolution was changed and isometric tension waschecked periodically during the experiment. Pre-parations were extremely stable, often lasting for10–12 hours. Values for a particular frequencyþstrain combination represent the mean of datafrom 3–6 individuals. Temperature effects onmuscle mechanics were examined in pipefish bytesting four individuals at both 101C and 201C in

Fig. 3. A: Schematic of how the dorsal fin inclinatormuscles work. The figure depicts a cross-section of thevertebral column. Fin muscles on either side of the vertebralcolumn alternately contract, pulling the fin ray first to oneside, then the other. A traveling wave passes down the fin dueto a phase lag in the fin ray movements from anterior toposterior. B: Schematic of how the fused middle radialsfunction as a rigid beam to maintain separation between thefin rays and fin ray muscles in the experimental setup.

WORK AND POWER OF DORSAL FIN MUSCLE 565

the following frequencyþstrain combinations:5Hz and 10Hz; within each frequency, 2.5%, 5%,and 10% strain.Following each experiment, the muscle was

dissected away from the surrounding tissues andweighed to the nearest 0.1mg. Cross-sectionalarea was calculated by converting mass to volume(assuming muscle density of 1.05 kg/m3) and thendividing by Lo.For each muscle, Lo, cross-sectional area, max-

imum isometric twitch and tetanic forces, timefrom stimulus to peak force (both twitch andtetanus), maximum isometric twitch and tetanicstress, mean work per cycle, and mean poweroutput were determined. Data for the musclesfrom the two species were analyzed for statisticalsignificance in StatView for Macintosh (version5.0; SAS Institute, Cary, NC) using MANOVA forthe isometric variables, and three-way ANOVAsfor work and power output that consideredspecies, oscillation frequency, and strain as themain effects. The three-way ANOVAs were con-ducted only on the frequencies and strains forwhich data was available for both seahorses andpipefish: 20 and 30Hz, and 2.5%, 5%, 10%, and15% strain. Temperature effects were analyzed bycomputing Q10 values and comparing work percycle using a single factor ANOVA with tempera-ture as the main effect. Differences were consid-ered significant at a¼0.05.

RESULTS

High-speed video

Representative fin beat sequences from S.fuscus and H. hudsonius are shown in Fig. 4.These sequences, as well as additional movies ofoscillating dorsal fins (including one of H. zoster-ae), are posted as QuickTime movies on the WorldWide Web at http://www.wfu.edu/~rossma/finmo-vies.html. For pipefish, the oscillation frequenciesobserved in swimming animals were 13–26Hz(mean 20.7Hz). Multiple waves are present in thefin simultaneously (the wavelength is shorter thanthe fin; Fig. 4A). For H. hudsonius, fin beatfrequencies during swimming were 30–42Hz(mean 37.6Hz). As in pipefish, the propulsivewavelength is shorter than the fin (Fig. 4B). Finbeat frequencies were higher in the dwarf sea-horse, H. zosterae, ranging from 42–54Hz (mean47.4Hz). Unlike the larger species examined, inthe dwarf seahorses, only a single propulsive wavewas visible in the fin at any given instant; in fact,the wavelength appeared to be substantially long-

er than the length of the fin (not shown, but seeQuickTime movie). In all fin beat sequencesobserved for all species, oscillation of the fin eitherbegan along the entire length of the fin at once, orthe oscillation began at the anterior end andprogressed posteriorly until the entire fin wasinvolved. Oscillation was never observed to beginin the posterior regions of the fin and spread tomore anterior sites. Each individual fin ray beatsside to side in a sinusoidal pattern (Fig. 4C); aphase lag between the rays such that each rayleads the one posterior to it generates thetraveling wave pattern.

Isometric and anatomical measures

The series of slips of dorsal fin muscle removedfrom H. hudsonius and S. fuscus were similar incross-sectional area, but differed in their restinglength (Lo) and isometric force production(Table 1). MANOVA performed on the anatomicaland isometric properties of the muscles from thetwo species demonstrated a significant multivari-ate difference between them (Wilk’s l¼0.104,F¼17.284, P¼0.0001). All variables, except cross-sectional area (which was heavily influenced bythe experimenter according to how many muscleslips were removed), were significantly different inpost-hoc tests (Table 1). Resting length, twitchand tetanic stress were all higher in H. hudsoniusthan S. fuscus, while the rise times to maximumtwitch and tetanic force were faster in H.hudsonius (Table 1). The faster rise times arecorrelated with the faster fin beat frequenciesobserved in seahorses (see above).

Representative twitch and tetanic contractionsfor seahorses and pipefish are shown in Fig. 5. Inboth species, the twitch record looks similar tothat seen from other muscles. However, thetetanic traces always showed an unusual patternin which the force would rise to an initial peak andthen decline rapidly before stopping at a plateau offorce (that was always higher than the twitchforce) that was maintained throughout the stimu-lus period (Fig. 5). The decline of force from thetetanus plateau followed a slower timecourse thanthat seen in a twitch contraction (Fig. 5).

In vitro muscle mechanics

The cycling frequencies selected for each speciescompletely overlapped the range of fin beatfrequencies observed in swimming animals (seeabove). Only two frequencies (20Hz and 30Hz)were examined in both seahorses and pipefish, and

M.A. ASHLEY-ROSS566

Fig.4.

Rep

resentativefinbea

tsequen

cesrecord

edat500frames/sec.In

dividualframes

havebeencropped

andcompositedusingCanvas7(D

eneb

aSystem

s,In

c.,

Miami,FL).A:Pipefi

sh,S.fuscus;anterioristoward

thetopoftheframes.B:Sea

horse,

H.hudsonius;anterioristoward

thetop.In

both

(A)and(B

),thesolidhorizontal

lines

indicate

theanteriored

geofthedorsalfin,whilethedash

edlines

connectthecrestofasingle

propulsivewaveasitmoves

posteriorlyalongthefin.Asingle

finbea

tcycleissh

ownin

each

sequen

ce.C:Sch

ematicindicatingthepatternofside-to-sidemovem

entofasingle

finray.Numberswitharrowsindicate

thepositionofthefinray

attimes

duringasingle

finbea

tcycleandcorrespondto

thenumbered

arrowsin

thepanelsof(A

).Sim

ilarly,letterswitharrowscorrespondto

theframes

of(B

).

WORK AND POWER OF DORSAL FIN MUSCLE 567

work and power output at these frequencies werecompared statistically. Representative work loopsfrom several frequencies at 5% strain in bothspecies are shown in Fig. 6. Individual work loopswere selected because they were close to the meanwork (in mJ) recorded for that frequencyþstraincombination. At the same strain, Hippocampusdorsal fin muscle is able to generate positive workat higher frequencies than that of Syngnathus.

Seahorse dorsal fin muscle performs positive work(albeit a small amount) at frequencies as high as50Hz, while pipefish dorsal fin muscle is no longerable to contract fast enough to keep up with theservomotor arm, and thus performs negative work(it is being lengthened by the servomotor duringthe time of maximum activation; Fig. 6). In bothspecies, the phase of stimulation for maximal workproduction decreased (became earlier) as cyclingfrequency increased. Likewise, the duty factor(proportion of the cycle occupied by the stimulus)for maximal work production decreased asfrequency increased (Fig. 6).

Mean values for work per cycle and poweroutput of the dorsal fin muscle of Hippocampusand Syngnathus at different frequencies andstrains are shown in Figs. 7 and 8, respectively.In general, work per cycle at a given straindeclined with increasing frequency in both species(Fig. 7). At higher frequencies, the absolute timeduring which force can be developed and producepositive work is reduced, leading to a lower network during a single contraction cycle. The effectof strain on work performed was more complicatedand dependent upon the cycling frequency. At lowfrequencies, work per cycle increased as strainincreased from the lowest value used (2.5% Lo) upto some optimum value, which was typically 6% inHippocampus and 10% in Syngnathus (Figs. 7, 9).Increasing strain beyond this value resulted in adecrease in work per cycle and, in fact, would oftenresult in negative work (Fig. 7). As the frequencyof oscillation was increased, the value of musclestrain that resulted in maximum work decreasedby a small increment in the seahorse, but verynoticeably in the pipefish (Figs. 7, 9). Seahorsemuscle was able to generate positive work athigher cycling frequencies than pipefish muscle(Fig. 7). Several trials of pipefish dorsal fin muscleat 40 Hz (not shown) resulted in negative work at

TABLE1. Anatomical and isometric parameters for dorsal ¢nmuscle ofH. hudsonius and S. fuscus

Hippocampus Syngnathus P-value

Cross-sectional area (mm2) 2.89 (0.82) 2.64 (1.33) nsLo (mm) 5.40 (1.20) 2.30 (0.44) o0.0001Max twitch stress (kN/m2) 124 (53.3) 50 (23.9) 0.0008Time to max twitch force (ms)2 19 (2.0) 22 (4.1) 0.0206Max tetanic stress (kN/m2) 176 (62.1) 112 (48.7) 0.0252Time to max tetanic force (ms)3 25 (5.2) 36 (5.9) 0.0021

1Values are given as mean (SD). For all Hippocampus values, n=6; for Syngnathus values, n=13, except where otherwise noted.2n=113n=10

Fig. 5. Representative isometric twitch and tetanus fromseahorse (A) and pipefish (B). Upper traces in each panel areforce, lower traces are the stimulus monitor. Black tracesdepict tetanus, grey traces depict twitch contractions andstimulus.

M.A. ASHLEY-ROSS568

all strains. Pipefish muscle was able to generatepositive work at longer strains than seahorsemuscle: At 15% strain, seahorse muscle producednegative work at all frequencies tested, whereaspipefish muscle produced positive work at all butthe two highest frequencies (Fig. 7).The graphs of average power output at the

different frequency þ strain combinations havethe same general shape as those for work per

cycle, with an important difference. Power outputdoes not monotonically decrease with increasingcycle frequency. In fact, power production ismaximal at an intermediate frequency (30Hz inseahorse, 20Hz in pipefish; Figs. 8,10). This is dueto power output being simply the product of cyclefrequency and work per cycle: work per cycledeclines, but for a time the increasing frequencyoutweighs this, and power output rises. However,at some point, the work per cycle declines to thepoint that even very high frequencies cannotcounteract this effect, and power output thendeclines as well (Fig. 10).

The dorsal fin muscles of Hippocampus andSyngnathus were statistically compared by athree-way ANOVA at the two frequencies (20Hzand 30Hz) and four strains (2.5%, 5%, 10%, and15%) for which data was recorded in both species.Interestingly, there was not a significant effect ofspecies on either work per cycle (Table 2) or poweroutput (Table 3). However, both work and poweroutput showed significant effects of frequency,strain, and the interactions of speciesnstrain andfrequencynstrain (all with Po0.0001; Tables 2,3).These results indicate that even within therestricted range used in this analysis, both workand power output differ significantly at differentstrains and frequencies. The significant speciesn-

strain interaction term indicates that the seahorseand pipefish are responding differently to thestrain levels tested, which is evident from Figs. 9and 10. Finally, the significant frequencynstraininteraction indicates that the relationship be-tween cycling frequency and strain level iscomplex, as discussed above. It must be pointedout that the lack of a significant speciesdifference in these ANOVAs is somewhat mislead-ing, as the frequencies that overlapped werevery limited. Pipefish muscle was not able togenerate any positive work at high fre-quencies, whereas seahorse muscle still performedpositive work at oscillation frequencies as high as50Hz.

Temperature effects on muscle mechanics

Temperature exerts a significant effect on theability of pipefish dorsal fin muscle to performwork. Maximum isometric stress did not differsignificantly with temperature. Figure 11 sum-marizes the effect of temperature on work percycle and power output. At each oscillationfrequencyþstrain combination, work per cycleand power output at 201C is significantly higher

Fig. 6. Representative work loops at different cyclingfrequencies from the seahorse (left column) and pipefish(right column). Frequency and work recorded for the cycleshown is indicated in the upper left of each panel. All workloops were measured at 5% total strain. All loops fromHippocampus show positive work (loops are traversed coun-ter-clockwise; the muscle is generating force while it isshortening in length), whereas the loops for the two highestfrequencies in Syngnathus show negative work (loops aretraversed clockwise; the muscle is generating force while it isbeing forced to lengthen). Thick regions of the loops indicatethe period of stimulation.

WORK AND POWER OF DORSAL FIN MUSCLE 569

(as determined by single-factor ANOVA) than at101C (Table 4). Q10 values are dependent upon thefrequencyþstrain combination being considered,ranging from just over two at 5Hz/2.5% strain tojust under four at 5% strain (Table 4). Due to thedifferent responses to temperature of the muscleat 10% strain (increase in work and power at 201C,but decrease at 101C), Q10 values were notcalculated for this strain.

DISCUSSION

Comparison of seahorse to pipefish dorsalfin muscle

While both Hippocampus and Syngnathus usehigh-frequency oscillations of the dorsal fin astheir primary means of movement, the dorsal finmuscles from the two species showed significant

Fig. 7. Average values of workper cycle at different frequencies andstrains in Hippocampus (A) andSyngnathus (B). Error bars are stan-dard errors; some of the error barsare completely hidden by their re-spective symbols. Different frequen-cies are represented by differentcolored symbols and lines. Filledbrown square, 5Hz; open blue trian-gle, 10Hz; open teal diamond, 15Hz;filled black circle, 20Hz; filled greyinverted triangle, 25Hz; open redsquare, 30Hz; filled purple diamond,40Hz; open green circle, 50Hz.

M.A. ASHLEY-ROSS570

differences in several areas. Seahorse dorsal finmuscle bundles are significantly longer than thoseof pipefish (Table 1), while pipefish have longerdorsal fins with higher numbers of fin rays (andhence muscle slips; Fig. 1). Seahorses havenoticeably deeper bodies than pipefish, with alarge ventral ‘‘belly,’’ as well as an elaboration ofscales into spines all over the body. It may be thatthe greater length of the dorsal fin muscle slips inseahorses is simply due to the greater distance

between the vertebral column and the dorsalsurface, rather than having specific functionalsignificance. Maximum twitch and tetanic stressare significantly higher in Hippocampus dorsal finmuscle than in Syngnathus; mean twitch stress isnearly 2.5 times higher, while mean tetanic stressshows less of a difference, being 1.6 times higherin the seahorse. Twitch-tetanus ratio was 0.69 inseahorse and 0.45 in pipefish. Time to maximumforce was also significantly shorter in Hippocam-

Fig. 8. Average values of poweroutput at different frequencies andstrains in Hippocampus (A) andSyngnathus (B). Error bars are stan-dard errors; some of the error barsare completely hidden by their re-spective symbols. Different frequen-cies are represented by differentcolored symbols and lines. Filledbrown square, 5Hz; open blue trian-gle, 10Hz; open teal diamond, 15Hz;filled black circle, 20Hz; filled greyinverted triangle, 25Hz; open redsquare, 30Hz; filled purple diamond,40Hz; open green circle, 50Hz.

WORK AND POWER OF DORSAL FIN MUSCLE 571

pus than in Syngnathus for both twitch andtetanus (Table 1), which is likely correlated tothe higher in vivo contraction frequencies in theseahorse.At the same cycling frequencies, seahorse and

pipefish dorsal fin muscle generated similaramounts of work and power output (nonsignificantby ANOVA; Tables 2,3). However, work and power

differed significantly with frequency and strain(Tables 2,3). Additionally, Hippocampus andSyngnathus responded differently to strain (sig-nificant speciesnstrain interaction, Tables 2,3).Pipefish produced maximum work per cycle atlonger strains than seahorse over much of thefrequency range tested (10% versus 6%; Figs. 7,9).Because the Lo of pipefish dorsal fin muscle isconsiderably shorter than that of the seahorse,when the percent strains are converted to absoluteshortening distances, the optimal distances aresimilar, on the order of 0.2–0.3mm in both species.This may be of functional significance, in that themagnitude of lateral deflection of the fin rays isdependent on the absolute amount of shorteningin the muscle. Similar absolute muscle shorteningwould thus be predicted to produce similarangular deflection of the fin rays. Though therewas not a significant difference attributable tospecies in the ANOVA, it must be pointed out thatthere is nonetheless a real difference in theperformance capabilities of seahorse dorsal finmuscle relative to pipefish, as Hippocampus wasable to produce positive work and power at highcycling frequencies (430Hz) where Syngnathusdorsal fin muscle only produced negative work.Negative work values indicate that the servomotoris doing work on the muscle, and hence that themuscle is unable to contract fast enough to keepup with the servomotor.

Fig. 9. Strain at which both work and power productionare maximal for Syngnathus (filled circles) and Hippocampus(open squares) across a range of frequencies.

Fig. 10. Mean maximum work (A) and power output (B) atdifferent frequencies for Syngnathus (filled black circles) andHippocampus (open grey squares), irrespective of strain. Errorbars are standard errors of the mean; some of the error barsare completely hidden by their respective symbols.

TABLE 2. ANOVAresults for work per cycle (inJ/kg) of the dorsal¢nmuscles ofH. hudsonius and S. fuscus at two frequencies and

four strains

E¡ect F-value P-value

Species (df=1) 0.697 0.405Frequency (df=1) 69.481 o0.0001Strain (df=3) 55.101 o0.0001SpeciesnStrain (df=3) 20.553 o0.0001FrequencynStrain (df=3) 7.860 o0.0001

TABLE 3. ANOVAresults for power output (inW/kg) of the dorsal¢nmuscles ofH. hudsonius and S. fuscus at two frequencies and

four strains

E¡ect F-value P-value

Species (df=1) 1.259 0.263Frequency (df=1) 55.003 o0.0001Strain (df=3) 67.655 o0.0001SpeciesnStrain (df=3) 23.137 o0.0001FrequencynStrain (df=3) 15.739 o0.0001

M.A. ASHLEY-ROSS572

Comparisons to other studies

An unusual feature of the tetanic contractionfor both Hippocampus and Syngnathus is the‘‘sag’’ from the initial peak of force to the lowerplateau that is maintained for the duration ofstimulation (Fig. 5). This was observed in everypreparation tested for seahorses and pipefish, butnot for salamander muscle tested in the sameexperimental setup (Ashley-Ross and Barker,unpublished data), so it does not appear to be anartifact of the apparatus used. According to Berg-man (’64a), the tetanic force in dorsal fin muscle ofH. hudsonius does not show a similar decline,

though the experimental preparation used in thatstudy was quite different from that used here.Bergman mounted the dorsal fin horizontally andtied his force transducer to the distal end of the finrays (which acted as a long lever arm). Since thefin rays will bend along their long axis, it is likelythat Bergman’s (’64a) arrangement acted todampen an initial peak of force before the plateau.In the preparation used here, the tissues in serieswith the force transducer are the fin rays, tendonsof insertion, and the dorsal fin muscles them-selves. It is at present unclear what is responsiblefor the sag; it seems unlikely to be due to ‘‘give’’ inone of these tissues, as the sag was visible in all ofthe tetanic contractions, not just the first (onewould expect the tissue to ‘‘give’’ once and then bestable). It may represent very fast relaxation ofsome muscle fibers; however, this remains to betested experimentally. Functionally, the sag wouldappear to be of little consequence for the seahorseor pipefish. Patterns of dorsal fin muscle activa-tion are unknown, but it is unlikely that longtetanic contractions occur during normal in vivooperation. At contraction frequencies resemblingthose in vivo, the entire strain cycle wouldhave taken place before the plateau region of thetrace had been reached. Thus, the force developedby the muscle is likely to resemble the initialpeak of force, rather than the maintained plateauforce.

TABLE 4. E¡ects of temperature onmuscle mechanics in S. fuscus1

101C 201C Q10 F-value P-value

Isometric tetanic stress (N/m2) 74.29 (41.75) 111.5 (48.72) n/a 1.888 0.18Work (J/kg) 133.253 o0.00015Hz2.5% 0.074 (0.050) 0.16 (0.024) 2.185% 0.10 (0.095) 0.27 (0.044) 2.5410% 0.041 (0.17) 0.37 (0.11) n/a10Hz2.5% 0.042 (0.034) 0.14 (0.043) 3.285% 0.061 (0.047) 0.24 (0.069) 3.9110% �0.047 (0.11) 0.32 (0.097) n/aPower (W/kg) 117.975 o0.00015Hz2.5% 0.37 (0.25) 0.80 (0.12) 2.185% 0.52 (0.48) 1.33 (0.22) 2.5410% 0.20 (0.83) 1.85 (0.55) n/a10Hz2.5% 0.42 (0.34) 1.39 (0.43) 3.285% 0.61 (0.47) 2.38 (0.69) 3.9110% �0.47 (1.13) 3.23 (0.97) n/a

1Work and power were measured at two cycling frequencies (5 and 10Hz) and three strains (2.5, 5, and 10%Lo).Values for parameters are mean (SD).

Fig. 11. Temperature effects on work (A, B) and powerproduction (C, D) in Syngnathus at 2.5%, 5%, and 10% Lo

strain. Filled circles, 101C; open squares, 201C. (A, C) 5Hz. (B,D) 10Hz. Error bars are standard deviations.

WORK AND POWER OF DORSAL FIN MUSCLE 573

Time to peak twitch tension in Hippocampusmeasured in this study (19ms; Table 1) is close tothe value given by Bergman (’64a), who reported amean of 13ms to reach peak tension. The slowertime in the present study is likely due tomeasurements being made at 201C in this studyversus 251C in Bergman’s (’64a) work. Valuesreported by Bergman (’64a) for twitch tensioncannot be compared to those given here, asinformation regarding muscle dimensions wasnot reported, and the ‘‘lever arm’’ arrangementused is not comparable to the direct measure-ments used here. The seahorse and pipefish dorsalfin muscles show a slower twitch rise time thanindividual fast fibers from Pacific sand dab dorsalfin muscle (Gilly and Aladjem, ’87), which attainpeak tension in approximately 10ms. However,the fast fibers examined by Gilly and Aladjem (’87)were determined by histochemical staining to befast glycolytic (white) fibers with very littleoxidative capacity (Franzini-Armstrong et al.,’87). This finding contrasts with seahorse dorsalfin muscle, which has been shown to stain ashighly oxidative (Hale, ’96).The maximum tetanic stress in Hippocampus

and Syngnathus dorsal fin muscle is smaller thanthat measured in locomotor muscle from othervertebrate species, which usually average 200–

300 kN/m2 (Table 5, Josephson, ’93). Myofibrilsoccupy only 61% of the fiber volume in seahorsedorsal fin muscle (Smith et al., ’88); whencorrected for myofibrillar cross-sectional area,maximum tetanic stress is approximately 290kN/m2 in Hippocampus. Ultrastructure of pipefishdorsal fin muscle has not been examined, but if themyofibrils comprise the same proportion of cellvolume as in the seahorse, then the maximumtetanic stress in Syngnathus would be approxi-mately 180 kN/m2. The estimates for maximumtetanic stress when corrected for myofibrillarvolume are thus similar to those reported forother locomotor muscles.

The time to peak twitch force for seahorse andpipefish dorsal fin muscle is shorter than otherlocomotor muscles measured at the same tem-perature (Table 5). A faster rate of rise in force isexpected due to the high in vivo cycling frequen-cies of dorsal fin muscles in the two species.However, neither pipefish nor seahorse dorsal finmuscle can match the rapidity of the twitchkinetics of the sound-producing swimbladdermuscle of toadfish, with a time to peak twitchforce of E6ms (Table 5; Rome et al., ’96). Theswimbladder muscle contracts and relaxes extre-mely quickly, primarily due to an unusually highcrossbridge dissociation rate (Rome et al., ’99) that

TABLE 5. Comparative values of isometric parameters, work, and power output from representative vertebrate skeletal muscles

MuscleTemperature

(1C)

Maxisometric

stress (kN/m2)

Time topeak twitch

(ms)

Max workper cycle(J/kg)

Max poweroutput(W/kg) Reference

Hippocampus dorsal ¢n 20 176 19 0.36 [20Hz] 7.71 [30Hz] This studySyngnathus dorsal ¢n 20 112 22 0.42 [10Hz] 6.48 [20Hz] This studyDog¢sh white myotome 12 2412 120 [5Hz] (Curtin andWoledge,’96)Stenotomus chrysops redmyotome

20 1833 10.9 [5Hz] 27.9 [5Hz] (Rome and Swank,’92)

Thunnus albacares slow¢bers

20 110 6.5 [4Hz] (Altringham and Block,’97)

Sarda chiliensisslow ¢bers

20 100 11 [5Hz] (Altringham and Block,’97)

Dipsosaurus dorsalisilio¢bularis

22 1881 39 7.6 [5.9Hz] 42.6 [5.9Hz] (Swoap et al.,’93)

42 11 7.8 [20.1Hz] 153.7 [20.1Hz]Toad¢sh white myotome 15 228 79 (Rome et al.,’99)Toad¢sh swimbladdermuscle

15 56 6 (251C) (Rome et al.,’99)

Frog sartorius 20 350 15 [4Hz] 60 [4Hz] (Stevens,’96)Mouse soleus 35 224 22 4.9 [5Hz] 34.0 [5Hz] (James et al.,’95)Mouse extensor digitorumlongus

35 233 10 11.8 [10Hz] 107.2 [10Hz] (James et al.,’95)

1Value taken from (Marsh and Bennett,’85), measured at 251C.2Value taken from (Curtin andWoledge,’88).3Value taken from (Rome et al.,’92).

M.A. ASHLEY-ROSS574

has the side effect of limiting the muscle to lowforce production (Table 5) due to low numbers ofattached crossbridges. For twitch contraction timeand isometric stress, the dorsal fin muscles ofHippocampus and Syngnathus are intermediatebetween toadfish swimbladder muscle and otherlocomotor muscles (Table 5).Comparison of values for work and power

output of dorsal fin muscle from Hippocampusand Syngnathus to those of other muscles isdifficult due to the paucity of data from compar-able cycling frequencies. Nonetheless, some con-clusions are possible. Maximum work per cycle inboth species is an order of magnitude lower thanthe values reported for other vertebrate species.At similar frequencies but higher temperatures,lizard iliofibularis and mouse extensor digitorumlongus (EDL) muscles produce over 20� the workper cycle of the dorsal fin muscles studied here(Table 5). Using the Q10 value for 10Hz, 5% strainfor Syngnathus, raising its dorsal fin muscle to thesame temperature as the mouse EDL would resultin net work per cycle of 3.24 J/kg, still less thanone-third the value of mouse muscle.At comparable temperatures, frequencies giving

optimal work and power production in othervertebrate locomotor muscles are considerablylower than the optimal frequencies for seahorseand pipefish dorsal fin muscles. Due to the higherpreferred cycling frequencies, maximum poweroutput values for Hippocampus and Syngnathusare closer to those reported for slow fibers in otherfish, but much lower than the values for white(anaerobic) myotomal muscle (Table 5). Seahorsedorsal fin muscle is highly aerobic (Bergman, ’64b;Smith et al., ’88) and is classified as oxidativemuscle by histochemical staining (Hale, ’96).However, it operates at higher frequenciesthan other highly oxidative muscles. In fact,seahorse and pipefish dorsal fin muscles areable to generate positive work at cycling fre-quencies higher than those that have beenrecorded for myotomal muscle (red or white) fromother fish.Why are values for work per cycle so low in

Hippocampus and Syngnathus? The high cyclingfrequencies likely explain much of this result. Thetime taken up by the cycle is so short that themuscle would have to generate extraordinarilyhigh forces to produce large values of work percycle. Even at 20Hz (the lowest frequency testedfor the seahorse), the entire lengthen-shorten-relengthen cycle takes place so rapidly (entirecycle in 50ms) that the maximum isometric force

does not have time to develop before the musclebegins shortening at 25% of the way through thecycle (compare time to maximum twitch force inTable 1). Force falls rapidly during shortening andis very low during the re-stretch to Lo (Fig. 6). Ifcontraction and relaxation of the dorsal fin musclefollowed the time course of an isometric twitch,then the muscle would still be generating force atthe end of the cycle (about half way throughrelaxation; see Fig. 5). However, it is well-knownthat muscle force and shortening velocity varyinversely (reviewed in Josephson, ’93). The morerapidly a muscle shortens, the less force it is ableto sustain. At the frequencyþstrain combinationsused here, shortening velocities range from 0.125muscle lengths/sec (pipefish at 5Hz, 2.5% strain)to 7.5 lengths/sec (seahorse at 50Hz, 15% strain).At the higher frequencies and larger strains,shortening velocity is likewise high, and theforce-velocity relationship results in the forcedeveloped during shortening being low. Thus,low values for work per cycle are an expectedconsequence of the high muscle shortening velo-cities resulting from the high cycling frequencies.Indeed, at the higher frequencies tested forseahorse (40Hz and 50Hz), strain above 10%resulted in negative work, indicating that themuscle was unable to keep pace with the servo-motor. However, work per cycle was also low atlow cycling frequencies, where the shorteningvelocity of the muscle would be expected to be ina range of the force-velocity curve where forcewould still be high. Small amounts of work at lowcycling frequencies, and the very low force duringre-stretch at all cycling frequencies, may indicatethat shortening (work-dependent) deactivation(Josephson and Stokes, ’99) is occurring in thedorsal fin muscle. Work-dependent deactivation isan important property of muscles that performcyclic contractions, particularly at high frequency,as it allows the muscle to be re-lengthened byantagonist muscles without generating negativework that would impede movement (Josephson,’99; Josephson and Stokes, ’99). While shorteningdeactivation is important in allowing the dorsal finmuscles to function at high frequencies, it reducesthe force maintained during the cycle, and hencethe work performed.

Relatively low power output by the muscles,combined with inefficient transfer of momentumto the surrounding water (Blake, ’80), restrictsseahorses and pipefish to slow swimming speeds.However, both species have little need to moverapidly, relying on crypsis to avoid predation. The

WORK AND POWER OF DORSAL FIN MUSCLE 575

dorsal fin muscles are well-suited to this task: theybeat the fin back and forth so rapidly that itsmovement is above the frequency for flicker fusionin humans, and likely for their natural predatorsas well (Blake, ’76). The fin is thus renderedinvisible, and the swimming animal appears to benothing more than a drifting bit of eelgrass, kelp,or coral.

ACKNOWLEDGMENTS

I thank Robert Josephson for teaching me howto measure muscle work, for many valuablediscussions on muscle, and for critical commentson the manuscript. William Conner, HermanEure, and Wayne Silver also provided criticalcomments on the manuscript.

LITERATURE CITED

Altringham JD, Block BA. 1997. Why do tuna maintainelevated slow muscle temperatures? Power output of muscleisolated from endothermic and ectothermic fish. J Exp Biol200:2617–2627.

Bass AH. 1990. Sounds from the intertidal zoneF vocalizingfish. Bioscience 40:249–258.

Bass AH, Marchaterre MA. 1989. Sound-generating (sonic)motor system in a teleost fish (Porichthys notatus)F sexualpolymorphism in the ultrastructure of myofibrils. J CompNeurol 286:141–153.

Bergman RA. 1964a. Mechanical properties of the dorsal finmusculature of the marine teleost,Hippocampus hudsonius.Bull Johns Hopkins Hosp 114:344–353.

Bergman RA. 1964b. The structure of the dorsal finmusculature of the marine teleosts, Hippocampus hudso-nius andH. zosterae. Bull Johns Hopkins Hosp 114:325–343.

Blake RW. 1976. On seahorse locomotion. J Mar Biol AssocUK 56:939–949.

Blake RW. 1980. Undulatory median fin propulsion of twoteleosts with different modes of life. Can J Zool 58:2116–2119.

Breder CM, Edgerton HE. 1942. An analysis of the locomotionof the seahorse, Hippocampus, by means of high speedcinematography. Ann N Y Acad Sci 43:145–172.

Clark AW, Schultz E. 1980. Rattlesnake shaker muscle. II.Fine structure. Tissue Cell 12:335–351.

Close RL, Luff AR. 1974. Dynamic properties of inferior rectusmuscle of the rat. J Physiol Lond 236:259–270.

Consi TR, Seifert PA, Triantafyllou MS, Edelman ER. 2001.The dorsal fin engine of the seahorse (Hippocampus sp.). JMorphol 248:80–97.

Cooper S, Eccles J. 1930. The isometric responses ofmammalian muscles. J Physiol Lond 69:377–385.

Curtin N, Woledge R. 1996. Power at the expense of efficiencyin contraction of white muscle fibres from dogfish Scyliorhi-nus canicula. J Exp Biol 199:593–601.

Curtin NA, Woledge RC. 1988. Power output and force-velocity relationship of live fibers from white myotomalmuscle of the dogfish, Scyliorhinus canicula. J Exp Biol140:187–197.

Fawcett DW, Revel JP. 1961. The sarcoplasmic reticulum of afast-acting fish muscle. J Biophys Biochem Cytol Suppl10:89.

Fine ML. 1989. Embryonic, larval and adult development ofthe sonic neuromuscular system in the oyster toadfish.Brain Behav Evol 34:13–24.

Fine ML, Burns NM, Harris TM. 1990. Ontogeny and sexualdimorphism of sonic muscle in the oyster toadfish. Can JZool 68:1374–1381.

Franzini-Armstrong C, Gilly WF, Aladjem E, Appelt D. 1987.Golgi stain identifies three types of fibres in fish muscle. JMuscle Res Cell Motil 8:418–427.

Geerlink PJ, Videler JJ. 1974. Joints and muscles of the dorsalfin of Tilapia nilotica L. (Fam. Cichlidae). Neth J Zool24:279–290.

Gilly WF, Aladjem E. 1987. Physiological properties of threemuscle fibre types controlling dorsal fin movements in aflatfish, Citharichthys sordidus. J Muscle Res Cell Motil8:407–417.

Greenewalt CH. 1975. The flight of birds: the significantdimensions, their departure from the requirements fordimensional similarity, and the effect on flight aerodynamicsof that departure. Trans Am Phil Soc 65:1–67.

Hale ME. 1996. Functional morphology of ventral tail bendingand prehensile abilities of the seahorse, Hippocampus kuda.J Morphol 227:51–65.

James RS, Altringham JD, Goldspink DF. 1995. The mechan-ical properties of fast and slow skeletal muscles of the mousein relation to their locomotory function. J Exp Biol 198:491–502.

Josephson RK. 1985. Mechanical power output from striatedmuscle during cyclic contraction. J Exp Biol 114:493–512.

Josephson RK. 1993. Contraction dynamics and power outputof skeletal muscle. Annu Rev Physiol 55:527–546.

Josephson RK. 1999. Dissecting muscle power output. J ExpBiol 202 Pt 23:3369–3375.

Josephson RK, Stokes DR. 1999. Work-dependent deactivationof a crustacean muscle. J Exp Biol 202:2551–2565.

Knapp AG, Dowling JE. 1987. Dopamine enhances excitatoryamino acid-gated conductances in cultured retinal horizon-tal cells. Nature 325:437–439.

Lindsey CC. 1978. Form, function, and locomotory habits infish. In: Hoar WS, Randall DJ, editors. Fish physiology. NewYork: Academic Press. p 1–100.

Marsh RL, Bennett AF. 1985. Thermal dependence of isotoniccontractile properties of the skeletal muscle and sprintperformance of the lizard Dipsosaurus dorsalis. J CompPhysiol [B] 155:541–555.

Revel JP. 1962. The sarcoplasmic reticulum of the batcricothyroid muscle. J Cell Biol 12:571.

Rome LC, Swank D. 1992. The influence of temperature onpower output of scup red muscle during cyclical lengthchanges. J Exp Biol 171:261–281.

Rome LC, Sosnicki A, Choi IH. 1992. The influence oftemperature on muscle function in the fast swimming scup.II. The mechanics of red muscle. J Exp Biol 163:281–295.

Rome LC, Syme DA, Hollingworth S, Lindstedt SL, BaylorSM. 1996. The whistle and the rattle: the design ofsound producing muscles. Proc Natl Acad Sci U S A93:8095–8100.

Rome LC, Cook C, Syme DA, Connaughton MA, Ashley-RossM, Klimov A, Tikunov B, Goldman YE. 1999. Tradingforce for speed: why superfast crossbridge kinetics leads tosuperlow forces. Proc Natl Acad Sci U S A 96:5826–5831.

M.A. ASHLEY-ROSS576

Schultz E, Clark AW, Suzuki A, Cassens RG. 1980. Rat-tlesnake shaker muscle. I. A light microscopic and histo-chemical study. Tissue Cell 12:323–334.

Smith DS, Morales M, Del Castillo J. 1988. Dorsal fin andmyotome muscle of the sea horse, Hippocampus. PuertoRico Health Sci J 7:183–184.

Stevens ED. 1996. The pattern of stimulation influences theamount of oscillatory work done by frog muscle. J PhysiolCambridge 494:279–285.

Suarez RK, Lighton JRB, Brown GS, Mathieu-Costello O.1991. Mitochondrial respiration in hummingbird flightmuscles. Proc Natl Acad Sci U S A 88:4870–4873.

Swoap SJ, Johnson TP, Josephson RK, Bennett AF. 1993.Temperature, muscle power output and limitations on burstlocomotor performance of the lizard Dipsosaurus dorsalis. JExp Biol 174:185–197.

Videler J. 1993. Fish swimming. London: Chapman and ;Hall.Webb PW. 1982. Locomotor patterns in the evolution ofactinopterygian fishes. Amer Zool 22:329–342.

Winterbottom R. 1974. A descriptive synonymy of the striatedmuscles of the teleostei. Proc Acad Nat Sci Philadelphia125:225–317.

Withers PC. 1992. Comparative animal physiology. FortWorth, TX: Saunders College Publishing.

WORK AND POWER OF DORSAL FIN MUSCLE 577