J Comp Physiol A (2009) 195:843–852

DOI 10.1007/s00359-009-0463-2

ORIGINAL PAPER

Prey location, biomechanical constraints, and motor program choice during prey capture in the tomato frog, Dyscophus guineti

Jenna A. Monroy · Kiisa C. Nishikawa

Received: 3 May 2009 / Revised: 15 July 2009 / Accepted: 15 July 2009 / Published online: 6 August 2009© Springer-Verlag 2009

Abstract This study investigated how visual informationabout prey location and biomechanical constraints of thefeeding apparatus inXuence the feeding behavior of thetomato frog, Dyscophus guineti. When feeding on prey atsmall azimuths (less than § 40°), frogs aimed their headstoward the prey but did not aim their tongues relative totheir heads. Frogs projected their tongues rapidly by trans-ferring momentum from the lower jaw to the tongue. Stor-age and recovery of elastic energy by the mouth openingmuscles ampliWed the velocities of mouth opening andtongue projection. This behavior can only occur when thelower jaw and tongue are aligned (i.e., within the range ofmotion of the neck). When feeding on prey at large azi-muths (greater than § 40°), frogs aimed both the head andtongue toward the prey and used a muscular hydrostaticmechanism to project the tongue. Hydrostatic elongationallows for frogs to capture prey at greater azimuthal loca-tions. Because the tongue moves independently of thelower jaw, frogs can no longer take advantage of momen-tum transfer to amplify the speed of tongue projection. Tofeed on prey at diVerent azimuthal locations, tomato frogsswitch between alternative strategies to circumvent thesebiomechanical constraints.

Keywords Feeding · Behavioral choice · Biomechanical tradeoVs · Frog · Amphibian

Introduction

Many vertebrates feed on a wide variety of prey that diVerin their physical and behavioral characteristics (Schaerlae-ken et al. 2007).Thus, animals use diVerent kinematicstrategies for capturing diVerent prey (e.g., sharks, Ferry-Graham 1998; Wshes, Liem 1978; Wainwright and Lauder1986; van Wassenberg et al. 2006; salamanders, Deban1997; frogs, Anderson 1993; Valdez and Nishikawa 1997;and lizards, Lappin and German 2005). In many animals,visual information about prey size, shape, or location is par-ticularly important for modulating prey capture behavior(parrotWsh, Rice and Westneat 2005; salamanders, Deban1997, frogs, Anderson and Nishikawa 1996; and lizardsSchaerlaeken et al. 2007). For example, the leopard frog,Rana pipiens will apprehend prey with either its tongue orits jaws depending on the size of the prey (Anderson andNishikawa 1996). Visual information about the prey iscombined with other information about the animal to selectamong alternative kinematic strategies that are appropriatefor diVerent conditions.

Like other vertebrates, frogs exhibit a complex reper-toire of feeding behaviors from which they choose the mostappropriate strategy for catching prey of varying sizes,shapes, and locations. Due to biomechanical constraints, nosingle kinematic strategy can be used to feed successfullyon all prey types (van Wassenberg et al. 2006). As men-tioned above, the leopard frog, R. pipiens, uses either itstongue or its jaws to apprehend diVerent sized prey. Asprey increase in size, they reach a mass at which wet adhe-sion to the tongue can no longer be used for prey prehen-sion. At this size, the frogs switch from tongue prehensionto jaw prehension (Anderson and Nishikawa 1996). Toadsuse inertial elongation to project their tongues rapidlybut this strategy is limited by the range of head turning

J. A. Monroy (&) · K. C. NishikawaDepartment of Biological Sciences, Northern Arizona University, Bldg 21, FlagstaV, AZ 86011-5640, USAe-mail: jenna.monroy@nau.edu

123

844 J Comp Physiol A (2009) 195:843–852

(Nishikawa and Gans 1996; Nishikawa 2000). Momentumproduced by rapid mouth opening is transferred from thelower jaw to the tongue, resulting in rapid tongue elonga-tion (Mallett et al. 2001; Lappin et al. 2006). Movements ofthe lower jaw and tongue must be aligned and preciselycoordinated to maximize momentum transfer. Thus, whenprey are located beyond the range of head turning, toadsturn their whole bodies so that they can use inertial elonga-tion to catch elusive prey. These examples demonstratehow individual frogs adjust their behavior when confrontedwith a biomechanical constraint.

Among frogs, only microhylids and hemisotids haveintrinsic tongue muscles that enable frogs to project andaim their tongues hydrostatically (Ritter and Nishikawa1995; Nishikawa et al. 1999; Monroy 2000; Meyers et al.2004). Hydrostatic elongation allows for prey capture atlarge azimuthal angles around the body but since the tonguemoves independently of the head, the frogs cannot uselower jaw momentum for rapid tongue projection as intoads. Thus, hydrostatic elongation is a successful strategyfor catching slow prey at large azimuths relative to thebody. A tradeoV between the speed of tongue projectionand accurate tongue placement appears to constrain frogs tousing one strategy or another. Because of these biomechanicalconstraints, frogs choose alternative motor strategies.

The most successful strategy for capturing a given typeof prey will depend on biomechanical constraints thatemerge from the interaction between the predator’s feedingapparatus and the physical characteristics of the prey. Thegoal of the present study was to investigate the biomechan-ical constraints that inXuence the selection of alternativestrategies in frogs to capture prey at diVerent azimuthallocations.

The frog, Dyscophus guineti is appropriate for this studybecause it appears to use both inertial elongation andhydrostatic elongation to project its tongue during prey cap-ture. Like other microhylid frogs, Dyscophus possessesintrinsic tongue muscles that enable frogs to aim theirtongues > 90° around their heads. However, tongue projec-tion in Dyscophus is about ten times faster (> 2 m/s)(Monroy 2000) than in other microhylid frogs (» 0.2 m/s)and is comparable to tongue projection speeds of frogs thatuse inertial elongation (e.g., Rana) (Nishikawa 2000). Wehypothesized that Dyscophus uses diVerent feeding strate-gies in response to changes in prey location. SpeciWcally,we hypothesized that Dyscophus uses inertial elongation tofeed on prey within the range of head turning and hydro-static elongation when the prey are located beyondthe range of head turning. When the tongue is aligned withthe head, momentum is transferred from the lower jaw to thetongue and the tongue can be projected rapidly. However,when prey are located beyond the range of head turning, thetongue moves independently of the lower jaw and momentum

can no longer be used to propel the tongue from the mouth.The speciWc objectives of this study were: (1) to determinethe mechanism for rapid tongue projection in Dyscophus,speciWcally, whether Dyscophus uses inertial elongation;(2) to determine whether prey location is used to choosebetween alternative motor programs; and (3) to determinehow biomechanical constraints inXuence motor programchoice.

Materials and methods

Specimens

Ten adult tomato frogs, D. guineti, were used in this study.Frogs with snout-vent lengths ranging from 6.2 to 8.2 cmwere obtained from a commercial animal supplier (GladesHerp Farm, Bushnell, FL, USA). To document changes infeeding behavior with respect to prey location, seven indi-viduals were imaged at 250 Hz while feeding on cricketsplaced at diVerent azimuthal locations around the frog. Toobserve changes in motor output of the jaw muscles, elec-trodes were implanted into the mm. depressor mandibulaeand the mm. levator mandibulae posterior longus, on theright and left sides of three additional frogs. These frogswere also imaged at 250 Hz while feeding on prey at diVer-ent azimuthal locations.

Kinematic analysis of feeding behavior

Feeding sequences were recorded at 250 frames s¡1 in lat-eral and ventral views using two synchronized RedLake™high-speed digital imaging systems. The cameras werepositioned lateral to the animal, one on top of the other, anda 45° mirror underneath a glass platform was used to recordthe ventral view in one of the cameras. Synchronized stro-boscopic LED illumination was used to eliminate the needfor high Watt light bulbs and to enhance picture quality.Frogs were placed on the platform and allowed to feedunrestrained on half-inch crickets at a distance of approxi-mately 2 cm. Forceps were used to restrain the crickets andensure that the head of the cricket faced the frog. A back-ground of 0.5 cm2 was used for scaling and aspect ratio cal-culation. At least six feeding sequences were recorded fromeach individual. All the sequences collected were consi-dered successful hits in which the tongue was placed on thecricket but frogs were not always allowed to eat the cricket.During three feeding sequences, crickets were placeddirectly ahead of the frog (less than § 20°) and during threefeeding sequences, crickets were placed at varying anglesto the right or left of the frog (greater than § 20°). Attemptswere made to Wnd the greatest angle at which a frog wouldfeed on a cricket without moving its body. Prey angles were

123

J Comp Physiol A (2009) 195:843–852 845

measured from still images after data collection. Feedingstoward the right and left sides of the head were not found todiVer in any of the measured kinematic variables. Thus,these trials were combined into a single group.

Feeding sequences were digitized using Didge 2.2motion analysis software (A. Cullum, Creighton Univer-sity). For each lateral frame, the digitized points (X, Y)included the upper jaw tip, jaw joint, lower jaw tip, base oftongue (i.e., the point of attachment between the tongue andthe lower jaw), tongue tip, and prey. For each ventralframe, seven points on the right and left sides of the headwere digitized to determine the midlines of the head andtongue: jaw joint, base of tongue, tongue tip, and prey. TheX coordinate in the ventral view was used to produce three-dimensional coordinates for the lower jaw, base of thetongue, tongue tip, and prey. In this study, we focused onlyon tongue projection. Thus, trials were analyzed from theonset of forward head movement to the completion of for-ward head movement, which did not include tongue retrac-tion or mouth closing.

For each frame, gape distance (i.e., distance between theupper and lower jaw tips) and the gape angle (i.e., anglesubtended by the upper and lower jaw tips with the jawjoint at the vertex) and tongue distance (i.e., distancebetween the lower jaw and tongue tip) was computed. Jawand tongue distances were smoothed using a quintic splinefunction (QuickSAND; Walker 1998) and instantaneousvelocities were estimated as the Wrst derivative of thesmoothed position data.

For each feeding sequence, the time of onset, time ofcompletion, and duration were calculated for head, jaw, andtongue movements. Maximum head angle, maximumtongue angle, and prey angle were computed using ventralimages. Maximum head angle, deWned as the angle towhich the head turned toward the prey during feeding, wascalculated as the angle between the midline of the head inthe frame before Wrst movement, and the same line in theframe when head movement ceased. The midline of thehead was deWned as the line that bisects the jaw joints andpasses through the upper jaw tip. Maximum tongue angle,deWned as the angle of the tongue with respect to the headat maximum projection, was calculated as the angle formedby the midline of the head and the midline of the tongue.The midline of the tongue was deWned as the line thatbisects the right and left base of tongue and the tongue tips.Prey angle was deWned as the angle formed by the midlineof the head and the line between the upper jaw tip and theprey. The part of the cricket closest to the frog was used tocalculate prey angle. Most often this was the head of thecricket. Relative maximum tongue reach was computed asthe distance between the maximally protracted tongue tipand the lower jaw tip, divided by the jaw length. Jaw lengthwas used as an estimate of resting tongue length.

To examine how feeding behavior changed with preyangle, a two-way ANOVA was performed on 13 kinematicvariables using SPSS 11.0. The main eVects were individ-ual (random) and feeding strategy (head-only orhead + tongue). Feeding strategies were deWned as either(1) head-only aiming, when a frog turned its head towardthe prey but did not aim its tongue relative to its head (lessthan or equal to § 10°) or (2) head + tongue aiming, whena frog aimed both its head and tongue (greater than § 10°)toward the prey. The feeding strategies were coded as a cat-egorical variable because there was a discrete separationbetween the two behaviors at a prey angle » 40°.

A step-wise discriminant function analysis was used todetermine which kinematic variables best discriminatebetween head-only and head + tongue aiming. Variableswith F values > 3.84 were entered into the analysis. The Ftest of Wilks’ lambda determines whether the discriminantmodel as a whole is signiWcant. Wilks’ lambda values rangefrom 0 to 1 with 0 indicating that group means diVer and 1indicating that group means are the same. A random sampleof the original data set (70%, 57/82 trials) was used to cre-ate the discriminant function and was also cross-validatedto identify any outliers that were not correctly classiWed bythe discriminant function. The remaining trials 30% (25/82trials) were used to test the validity of the classiWcation.

Electromyography

Electromyograms (EMGs) were recorded from the mouthopening muscle, mm. depressor mandibulae, and the largestof the six jaw adductor muscles, the mm. levator mandibu-lae posterior longus, on the right and left sides of threefrogs. EMGs were synchronized with high-speed imagingusing a single TTL trigger signal from the imaging systemto the LabView computer that recorded EMG data. Bipolarhook electrodes were constructed from 0.05 mm diameterformvar-coated stainless steel wire with a bared tip lengthof 0.5 mm separated by 1 mm. Frogs were anesthetized in asolution of 0.1 g tricaine methanesulfonate (MS-222) perliter of water buVered to a pH of 7.0 with sodium bicarbon-ate. Electrodes were implanted percutaneously into themuscles of anesthetized subjects using 23 gauge hypoder-mic needles. Electrodes were threaded under the skin to theback of the frog and secured to the skin using 4.0 silksuture.

Analog signals were ampliWed 1,000 times (A-M Sys-tems) and Wltered using a 60-Hz notch Wlter and a band-pass Wlter set at 10–5,000 Hz. Signals were digitized at4,000 samples s¡1 (Peak Performance Technologies, Inc.,Centennial, CO, USA) and Wltered digitally using a secondband-pass Wlter set at 200–1,000 Hz. EMG signals weresynchronized with digital images to determine the onset ofjaw muscle activity relative to the onset of mouth opening.

123

846 J Comp Physiol A (2009) 195:843–852

For each trial, the duration, peak amplitude, and rectiWedintegrated area prior to mouth opening were measured fromdepressor mandibulae EMG signals. A two-way ANOVAwith direction (head + tongue right, head + tongue left,head-only) and individual (random) as the main eVects, wasused to compare EMG variables during head-only andhead + tongue aiming. The relationships between depressormandibulae activity, prey angle, lower jaw displacement,and lower jaw velocity were examined using Pearsonproduct–moment correlation coeYcients, P = 0.05.

Fig. 1 Relationship between prey angle, head angle, and tongue angleduring feeding in Dyscophus guineti. a Head angle versus prey angle.b Tongue angle versus prey angle. The solid line represents a one-to-one relationship with prey angle. At prey angles less than § 40°, thehead turns toward the prey (r = 0.97, slope = 0.77, 95%CI 0–32,P = 0.0001) and the tongue is not aimed (headonly aiming). At prey an-gles greater than § 40°, both the head (r = 0.6, slope = 0.41, 95%CI13.7–53, P = 0.0001) and tongue (r = 0.73, slope = 0.59, 95%CI 17.6–70, P = 0.0001) are aimed toward the prey (head + tongue aiming).The vertical line at 40°, separates head-only and head + tongue aiming.n = 10 individuals, 88 trials

Fig. 2 Head aiming (HA, closed triangles) and tongue aiming (TA,open diamonds) strategies at large prey angles (greater than 70°). Thethree points below 70° were not included in this Wgure to focus only onhead + tongue aiming trials. a In three individuals, tongue angle(r = 0.96, P = 0.0001) increased whereas head angle (r = 0.01P = 0.78) was independent of prey angle (n = 12 trials). b In Wve indi-viduals, both head (r = 0.82, P = 0.0001) and tongue (r = 0.87,P = 0.0001) angle increased with prey angle (n = 30 trials). c In twoindividuals, head angle (r = 0.85, P = 0.0001) increased but tongue(r = 0.69, P = 0.27) angle was independent of prey angle (n = 16 trials)

123

J Comp Physiol A (2009) 195:843–852 847

Results

Feeding strategies

Tomato frogs used two feeding strategies to capture prey atdiVerent locations around their heads. When feeding onprey within § 40° on either side, the frogs turn their headstoward the prey but do not aim their tongues relative totheir heads even though they possess the ability to do so(Fig. 1a, b). We termed this behavior head-only aiming.When feeding on a prey item at an azimuthal angle greaterthan § 40°, frogs must aim both their heads and tongues tosuccessfully capture prey. We termed this behaviorhead + tongue aiming.

During head-only aiming, frogs turned their heads 0–34°toward the prey and projected their tongues in the sagittalplane (Fig. 1a, b). The greater the prey angle, the more thefrogs turned their heads toward the prey (r = 0.97,P = 0.0001). Maximum head turning angles recorded dur-ing head-only aiming were close to but still less than maxi-mum head angles (average maximum = 49.6°) recorded foreach individual. There was a small but signiWcant relation-ship between prey angle and tongue angle (r = 0.5,P = 0.003). However, in all cases, the tongue angle

was < 10°. Therefore, head-only aiming was the only strat-egy frogs used within the range of head turning.

During head + tongue aiming, the head turned 20–60° andthe tongue was aimed 18–68° relative to the head (Fig. 1a, b).Both head angle (= 0.60, P < 0.0001) and tongue angle(r = 0.73, P = 0.0001) increased with prey angle. However,the extent to which the head and tongue were aimed variedamong individuals and from trial to trial. In three individuals(Fig. 2a), head angle varied little across prey angle (r = 0.47,P = 0.78), whereas tongue angle increased with prey angle(r = 0.96, P = 0.0001). In Wve individuals (Fig. 2b), bothhead and tongue angle increased with prey angle (tongue—r = 0.87, P = 0.0001; head—r = 0.82, P = 0.0001), and intwo individuals (Fig. 2c), tongue angle varied little with preyangle (r = 0.69, P = 0.27) and head angle increased with preyangle (r = 0.85, P = 0.0001). These results demonstrate thatall possible combinations of head + tongue aiming may bepresent among individuals within a given population.

Kinematics of head-only aiming versus head + tongue aiming

In all head-only aiming trials, movements of the lower jawand tongue were precisely timed such that the onset of

Fig. 3 Representative image se-quences of head-only and head + tongue aiming. Note that the lateral and ventral views were recorded simultaneously using two synchronized cam-eras. a Headonly aiming is char-acterized by rapid movements of the head, lower jaw and tongue (prey angle » 0°). b During head + tongue aiming, lower jaw and tongue movements are no longer tightly coupled and are signiWcantly slower than during head-only aiming (prey angle » 90°). Time 0 ms indi-cates the onset of mouth opening

123

848 J Comp Physiol A (2009) 195:843–852

tongue projection always occurred within 12 ms of theonset of mouth opening (Fig. 3; Table 1). When prey werelocated at azimuths § 0–10°, frogs lunged toward the preyand continued to lunge forward while the tongue was pro-jected. When prey were located at azimuths § 10–40°, thehead turned toward the prey and stopped turning just as themouth began to open (Fig. 4a; Table 1). Thus, during head-only aiming, head turning occurred before simultaneousmovements of the mouth and tongue.

When prey were located at azimuths beyond § 40°, headturning, mouth opening, and tongue projection overlappedin time, but were less tightly coupled (Fig. 4b; Table 1).The head turned toward the prey throughout the duration ofmouth opening. The onset of tongue projection (averageonset = 13.7 ms) occurred later relative to mouth openingand was also more variable (F ratio = 22.6, P < 0.0001).Thus, during head + tongue aiming, head, lower jaw, andtongue movements were not as stereotyped as during head-only aiming.

The velocities of mouth opening and tongue projectionwere twice as fast during head-only aiming compared tohead + tongue aiming (Fig. 5; Table 1). During head-onlyaiming, maximum jaw and tongue velocities ranging from86 to 229 cm s¡1 occurred at the onset of movement andslowed as the tongue was projected (Fig. 5). Duringhead + tongue aiming, the mouth opened at an averagemaximum velocity of » 70 cm s¡1 and the tongue was pro-jected at an average maximum velocity of » 60 cm s¡1.

Because movements of the lower jaw and tongue were notas tightly coupled, maximum tongue velocity occurred sev-eral milliseconds after the onset of mouth opening andmaximum lower jaw velocity.

We used a step-wise discriminant function analysis toidentify those kinematic variables that best separate head-only from head + tongue aiming. Maximum jaw velocity,onset of head movement, and the onset of tongue projec-tion were the most important kinematic variables in dis-criminating among feeding types. Head-only andhead + tongue aiming were classiWed correctly in 54 of 56trials (96.4%) (Wilk’s Lambda = 0.321, P < 0.0001). Tri-als that were not included in creating the discriminantfunction were classiWed correctly in 23 of 26 trials(88.5%). Thus, movements of the head, jaws, and tonguewere distinctly diVerent between head-only andhead + tongue aiming.

Jaw muscle activity during head-only versus head + tongue aiming

In general, the depressor mandibulae muscles were active25–141 ms before the onset of mouth opening, and reachedtheir peak just before or at the onset of mouth opening(Fig. 6). However, the relative timing and magnitude ofactivation diVered between head-only and head + tongueaiming (Table 2). Prior to mouth opening, the depressormandibulae were activated longer when feeding on prey at

Table 1 Mean kinematic vari-ables (§ SEM) for head-only and head + tongue aiming and results of a two-way ANOVA with feeding strategy (Wxed) and individual (random) as the main eVects. F- and P-values are for the eVect of feeding strategy. Head-only (no head turning) n = 18; Head-only (w/head turn-ing) n = 9 trials; Head + tongue n = 61 trials. Timing variables are relative to the onset of mouth opening. NS P > 0.05; * P < 0.05; ** P < 0.01

Variable Head-only Head + tongue F P

Mean Mean

Head movements

Start of lunge (ms) ¡61.4 § 10.2

End of lunge (ms) 48.1 § 5.6

Lunge duration (ms) 109.8 § 12.4

Start of head turn (ms) ¡38.8 § 12.7 ¡38.8 § 4.3 0.0 0.99

End of head turn (ms) 3.1 § 3.9* 65.8 § 4.8 24.0 0.01

Head turn duration (ms) 32.0 § 15.1* 104.2 § 9.6 9.3 0.04

Jaw movements

Completion of lower jaw movement (ms) 11.9 § 0.8** 31.8 § 4.6 16.3 <0.0001

Max jaw velocity (cm/s) 132.6 § 5.4** 69.7 § 5.8 105.8 <0.0001

Time of max velocity (ms) 1.7 § 0.9 3.3 § 1.2 2.3 0.19

Tongue movements

Start of projection (ms) 5.2 § 0.5** 13.6 § 2.6 6.9 <0.0001

End of projection (ms) 21.8 § 1.0** 47.1 § 5.0 23.2 <0.0001

Projection duration (ms) 16.6 § 0.9** 33.5 § 2.7 54.4 <0.0001

Tongue distance (cm) 1.1 § 0.1 1.0 § 0.1 3.0 0.12

Max tongue velocity (cm/s) 118.7 § 6.5** 60.8 § 3.6 68.1 <0.0001

Time of max velocity (ms) 6.4 § 1.2** 26.6 § 4.7 19.5 <0.0001

123

J Comp Physiol A (2009) 195:843–852 849

azimuths less than § 40° (head-only aiming) than whenfrogs fed on prey at azimuths greater than § 40°(head + tongue aiming). The average onset of depressormandibulae activity was 100 ms prior to mouth openingduring head-only aiming and 55 ms prior to mouth openingduring head + tongue aiming. The relative integrated areaof depressor mandibulae activity preceding mouth openingshowed a similar association with feeding strategy. Therewas a larger burst of activity prior to mouth opening whenfeeding on prey within § 40° than when frogs fed on preyat angles greater than § 40°. Within a strategy, there wasno eVect of prey angle on either the relative timing ormagnitude of depressor mandibulae activity. Therefore,head-only aiming feedings were associated with greateractivation of the depressor mandibulae prior to mouthopening than head + tongue aiming.

The extent to which the depressor mandibulae muscleswere activated prior to mouth opening varied among indi-viduals. However all individuals showed a similar associa-tion with lower jaw displacement and jaw velocity (Fig. 7).The onset and integrated area of EMG activity recordedfrom the depressor mandibulae increased with larger lowerjaw displacements and velocities, which correlated withhead-only aiming. Thus, greater activation of the depressormandibulae resulted in larger, faster movements of thelower jaw that occurred during head-only aiming.

Fig. 4 Relative timing of head, lower jaw, and tongue movements.Each box represents the average § SEM onset and completion of headturning and tongue projection with respect to average § 95%CI gapeduring mouth opening. Time 0 indicates the onset of mouth opening. aDuring head-only aiming (n = 9 trials), a frog Wrst turns its head towardthe prey after which the mouth opens and the tongue is projected. bDuring head + tongue aiming (n = 61 trials), the mouth opens and thetongue is projected while the frog turns its head toward the prey

Fig. 5 Kinematic plots of average jaw and tongue movements duringhead-only aiming (n = 27 trials) (open triangles) and head + tongueaiming (n = 61 trials) (closed circles). a Average gape distance 95%CIversus time. b Average gape velocity § 95%CI versus time. c Averagetongue distance § 95%CI versus time. d Average tongue velocity §95%CI versus time. During head-only aiming, the mouth opened andthe tongue was projected farther and faster than during head + tongueaiming

123

850 J Comp Physiol A (2009) 195:843–852

Discussion

Biomechanics of head-only aiming

Head-only and head + tongue aiming diVer biomechani-cally. Several lines of evidence support the hypothesis thatDyscophus uses inertial elongation of the tongue duringhead-only aiming. The biomechanics of inertial elongationhas been investigated in toads, Bufo marinus (Mallett et al.2001; Lappin et al. 2006). These studies demonstrate thatprior to mouth opening, the mouth opening muscles, thedepressor mandibulae, are preloaded during which elasticenergy is stored. Elastic recoil of the mouth opening mus-cles powers rapid mouth opening and momentum transferfrom the lower jaw to the tongue powers rapid tongue pro-jection. To maximize momentum transfer, lower jaw andtongue movements are precisely coordinated and thetongue moves in the same direction as the lower jaw.

Head-only aiming in Dyscophus shares several impor-tant features with inertial elongation in toads. Kinematicanalyses show that during head-only aiming in Dyscophus,the mouth opens rapidly, the onset of tongue projection istightly coupled with the onset of mouth opening and thetongue moves in the same direction as the lower jaw. Inaddition, EMG analyses show that preloading of the mouthopening muscles results in faster movements of the lower

jaw and tongue, suggesting that elastic energy is stored andrecovered during mouth opening. Thus, our results demon-strate that Dyscophus uses inertial elongation to project thetongue during head-only aiming.

Biomechanics of head + tongue aiming

In contrast to head-only aiming, Dyscophus uses a muscularhydrostatic mechanism for tongue projection duringhead + tongue aiming. In microhylids, including Dysco-phus, the tongue is a muscular hydrostat composed ofintrinsic tongue muscles that can produce complex move-ments such as elongation and bending (Monroy 2000;Meyers et al. 2004). Unilateral contraction of the tongueprotractor muscle, the genioglossus, causes the tongue toelongate and bend toward the side opposite the activatedmuscle (Monroy 2000; Meyers et al. 2004). The slowertongue and jaw movements associated with hydrostaticelongation and head + tongue aiming may allow for onlinecorrection and accurate placement of the tongue on the prey(Ritter and Nishikawa 1995; Meyers et al. 2004).

Several lines of evidence suggest that lower jaw momen-tum is not used to elongate the tongue during head + tongueaiming. First, the tongue moves out-of-plane with the lowerjaw. Momentum of the lower jaw can only be transferred inthe same direction as jaw movement, and will oppose the

Fig. 6 Mean rectiWed activity of the left depressor mandibulae (DM) and left levator mandibu-lae posterior longus (LM) with respect to the average gape proWle § 95% CI during a head-only aiming (n = 3 trials) and b head + tongue aiming (n = 3 tri-als) from a single individual. Time 0 ms indicates the onset of mouth opening. Depressor man-dibulae activity was greater prior to mouth opening during head-only aiming compared to head + tongue aiming

Table 2 Mean (§ SEM) EMG variables for the depressor mandibulae prior to mouth opening during head-only and head + tongue aiming andresults of a two-way ANOVA with feeding strategy (Wxed) and individual (random) as the main eVects

Head-only n = 10 trials; head + tongue n = 21 trials. F- and P-values are for the eVect of feeding strategy

NS P > 0.05; * P < 0.05; ** P < 0.0001

Variable Head-only Head + tongue F P

Mean Mean

Duration (ms) 99.4 § 10.5** 54.4 § 9.1 34.0 0.0001

Relative integrated area (Vs/Vs) 0.9 § 0.08** 0.4 § 0.04 26.7 0.0001

Relative maximum amplitude (mV/mV) 0.8 § 0.08 0.6 § 0.06 10.9 0.06

Relative time of peak amplitude (ms) 16.0 § 0.01 7.2 § 0.01 0.8 0.38

123

J Comp Physiol A (2009) 195:843–852 851

out-of-plane forces produced by the intrinsic tongue mus-cles particularly at large prey angles. Second, the relativetiming of lower jaw and tongue movements are not tightlycoupled. During head + tongue aiming, the onset of tongueprojection occurs when the lower jaw is decelerating, whichdecreases the momentum transferred to the tongue. Third,mouth opening velocity is signiWcantly slower, due to theabsence of preloading associated with storage and recoveryof elastic energy in the mouth opening muscles.

Biomechanical constraints of frog feeding

Head-only and head + tongue aiming confer diVerentadvantages and constraints. During head-only aiming,momentum produced by rapid mouth opening ampliWes thespeed of tongue projection but head-only aiming can onlybe used for prey locations that fall within the range ofmotion of the neck (» §40°). During head + tongue aim-ing, Dyscophus can capture prey at greater azimuthal anglesdue to hydrostatic elongation of the tongue, but the frogscan no longer take advantage of momentum transfer toamplify the speed of tongue projection. Due to these bio-mechanical trade-oVs, there appears to be no single prey-capture strategy that works reasonably well across the fullrange of prey locations.

The switch between head-only and head + tongue aim-ing appears to occur at prey azimuths near the limit ofmotion of the neck, between 40° and 50°. It is unknownwhether Dyscophus switches between head-only andhead + tongue aiming or exhibits an intermediate behaviorat the transition between the two behaviors. We havelimited data at these azimuths. Previous work has shownthat R. pipiens switches from tongue to jaw prehension inresponse to an increase in prey size (Anderson and Nishikawa1993). At the transition between the two behaviors,frogs switch randomly from one behavior to the other.Other animals have been shown to combine features ofeach behavior into a single behavioral blend at the transi-tion (Mortin et al. 1985). More feedings at and around 40°are needed to determine what occurs at the transitionbetween head-only and head + tongue aiming in Dysco-phus.

In theory, frogs could use one or more of several strate-gies for capturing prey at small azimuths, frogs could cap-ture prey: (1) by turning the head toward the prey and notaiming the tongue, (2) by turning only the head partiallyand aiming the tongue partially, or (3) by leaving the headstationary and aiming only the tongue. However, Dysco-phus exhibited only one of these three strategies. We neverobserved tongue-only aiming, although in principle thefrogs could have aimed the tongue alone to capture prey at

Fig. 7 Relationship between relative depressor mandibulae inte-grated area and a prey angle (r = 0.76, P = 0.0001), b lower jaw dis-placement (r = 0.72, P = 0.0001), and c lower jaw velocity (r = 0.71,P = 0.0001). The diVerent symbols represent three individuals, n = 31trials. Relative integrated area varied negatively with prey angle andpositively with lower jaw displacement and velocity, suggesting thatgreater activation of the depressor mandibulae results in larger, fastermovements of the lower jaw

123

852 J Comp Physiol A (2009) 195:843–852

all prey angles. In addition, frogs never aimed their tongueswhen feeding on prey at angles less than § 40°. Instead,frogs used only one strategy, head-only aiming, whichmaximizes the speed of tongue projection and may improveprey capture success.

For prey at large azimuths, frogs could capture prey: (1)by turning the head maximally and aiming the tongue therest of the way, (2) by turning the head partially and aimingthe tongue partially, or (3) by aiming the tongue maximallyand turning the head the rest of the way. All three strategieswere used by at least one individual in this study. At largeazimuths, some frogs aimed only their tongues, othersaimed only their heads, while others aimed both their headsand tongues. No single individual appeared to change itsfeeding strategy across trials. These results demonstratethat all possible strategies are present in a given populationof frogs and suggest that the frogs are not constrained bio-mechanically to choose a particular strategy for feeding onprey at large azimuths.

Acknowledgments This research was supported by grants IBN-0215438, IBN-0240349, IOS-0623791, and IIS-0827688 from theNational Science Foundation. Helpful comments on earlier versions wereprovided by Jean Block, Carrie Carreno, Leslie Gilmore, Marci Hol-lenshead, A. Kristopher Lappin, Stan Lindstedt, Laura Krebs, MaureenMaloney, Erika Nowak, David Pierotti, and Eric Zepnewski. We alsothank two anonymous reviewers for their insightful comments on pre-vious versions of this manuscript. The experiments in this study wereconducted in accordance with NIH publication No. 85-23, “Principlesof laboratory animal care” and with the approval of the InstitutionalAnimal Care and Use Committee at Northern Arizona University.

References

Anderson CW (1993) The modulation of feeding behavior in responseto prey type in the frog Rana pipiens. J Exp Biol 179:1–12

Anderson CW, Nishikawa KC (1993) A prey-type dependent hypo-glossal feedback system in the frog Rana pipiens. Brain BehavEvol 42:189–196

Anderson CW, Nishikawa KC (1996) The roles of visual and proprio-ceptive information during motor program choice in frogs.J Comp Physiol A 179:753–762

Deban SM (1997) Modulation of prey-capture behavior in the pleth-odontid salamander Ensatina echscholtzii. J Exp Biol 200:1951–1964

Ferry-Graham L (1998) EVects of prey size and mobility on prey cap-ture kinematics in leopard sharks Triakis semifasciata. J Exp Biol201:2433–2444

Lappin AK, German M (2005) Feeding behavior modulation in theleopard lizard (Gambelia wislizenii): eVects of noxious versusinnocuous prey. Zoology 108:287–295

Lappin AK, Monroy JA, Pilarski JQ, Zepnewski ED, Pierotti DJ, Nis-hikawa KC (2006) Storage and recovery of elastic potential ener-gy powers ballistic prey capture in toads. J Exp Biol 2009:2535–2553

Liem K (1978) Modulatory multiplicity in the functional repertoire ofthe feeding mechanism in cichlid Wshes. I. Piscivores. J Morphol158:323–360

Mallett ES, Yamaguchi GT, Birch JM, Nishikawa KC (2001) Feedingmotor patterns in anurans: insights from biomechanical modeling.Am Zool 41:1364–1374

Meyers JJ, O’Reilly JC, Monroy JA, Nishikawa KC (2004) Mecha-nism of tongue protrusion in microhylid frogs. J Exp Biol 207:21–31

Monroy JA (2000) Aiming during prey capture in microhylid frogs.Am Zool 40:6A

Mortin LI, Keifer J, Stein PSG (1985) Three forms of the scratch reXexin the spinal turtle: movement analyses. J Neurophysiol 53:1501–1516

Nishikawa KC (2000) Feeding in frogs. In: Schwenk K (ed) Feeding,form, function and evolution in tetrapod vertebrates. Academic,San Diego, pp 117–144

Nishikawa KC, Gans C (1996) Mechanisms of prey capture and narialclosure in the marine toad, Bufo marinus. J Exp Biol 199:2511–2529

Nishikawa KC, Kier WM, Smith KK (1999) Morphology and mechan-ics of tongue movement in the African pig-nosed frog Hemisusmarmoratum: a muscular hydrostatic model. J Exp Biol 202:771–780

Rice AN, Westneat MW (2005) Coordination of feeding, locomotor,and visual systems in parrotWshes (Teleostei: Labridae). J ExpBiol 208:3503–3518

Ritter DA, Nishikawa KC (1995) The kinematics and mechanism ofprey capture in the African pig-nosed frog (Hemisus marmora-tum): the description of a radically divergent anuran tongue. J ExpBiol 198:2025–2040

Schaerlaeken V, Meyers JJ, Herrel A (2007) Modulation of prey cap-ture kinematics and the role of lingual sensory feedback in the liz-ard Pogona vitticeps. Zoology 110:127–138

Valdez CM, Nishikawa KC (1997) Sensory modulation and behavioralchoice during feeding in the Australian frog, Cyclorana novae-hollandiae. J Comp Physiol A 180:187–202

van Wassenberg S, Herrel A, Adriaens D, Aerts P (2006) Modulationand variability of prey capture kinematics in clariid catWshes.J Exp Zool 305:559–569

Wainwright P, Lauder G (1986) Feeding biology of sunWshes: patternof variation in the feeding mechanism. Zool J Linn Soc 88:217–228

Walker JA (1998) Estimating velocities and accelerations of animallocomotion: a simulation experiment comparing numerical diVer-entiation algorithms. J Exp Biol 201:981–985

123