Elastic recoil action amplifies jaw closing speed in an aquatic feeding salamander

Abstract

Tendon springs often influence locomotion by amplifying the speed and power of limb joint rotation. However, less is known about elastic recoil action in feeding systems, particularly for small aquatic animals. Here, we ask if elastic recoil amplifies the speed of gape closing during aquatic food processing in the Axolotl (Ambystoma mexicanum). We measure activation of the adductor mandibulae externus via electromyography and strain of the jaw adductor muscle–tendon unit (MTU), and gape kinematics via fluoromicrometry. The muscle is pre-activated coincident with gape opening, which causes MTU stretch. Activation lasts significantly shorter for fish than cricket processing, and muscle shortening during MTU lengthening yields significantly greater elastic strain for cricket processing. The speed of MTU shortening, which dictates the speed of gape closing is 2.5–4.4 times greater than the speed of the initial shortening of the muscle fascicles for fish and cricket gape cycles, respectively. These data demonstrate a clear role for elastic recoil, which may be unexpected for a MTU in a feeding system of a small, aquatic animal. Amplification of jaw-closing speed resulting from elastic recoil likely confers ecological advantages in reducing prey escape risks during food processing in a dense and viscous fluid environment.

1. Introduction

The elastic action of biological springs is common across vertebrate locomotor systems, where stretch and recoil of compliant tissues including tendons and fascia play diverse roles [1]. Biological springs are known to amplify the power of movement [2,3], as well as the rate at which kinetic energy is released from muscle contraction to power joint movement [4,5]. Elastic recoil action may also reduce the need for muscles to do mechanical work by recycling gravitational-potential energy from joint movements [5–7], thereby improving movement economy. The latter can be accomplished as elastic stretch and recoil relegates the production of muscular force and work to recovery phases within the motion cycle where stress is low, e.g. the upstroke phase of a wing flap or the swing phase of a limb stride [5,6]. Our fairly detailed understanding of biological spring performance is to an overwhelming extent sourced from terrestrial limb systems. By contrast, we have a relatively limited knowledge about biological springs in feeding and fluid-based motion systems.

The occurrence and utility of springs in feeding systems could potentially be as diverse as in limbs. There are many prominent connective tissue sheets associated with the jaw musculature of most tetrapods, and especially basal aquatic-feeding anamniotes [8,9]. Nevertheless, we have fewer clear examples of elastic recoil action for feeding, despite the tendency of jaw systems to operate rapidly and powerfully [10–14]. Elastic action has the ability to amplify movement speed with respect to the initial shortening speed of activated muscle fascicles [1,4] and may, therefore, potentially help minimize jaw gape duration, which in turn minimizes prey escape risks.

Several studies have questioned the utility of elastic recoil action in small animals and fluid-based movement [15–17], whereas others have discovered evidence of elastic action in flight [5,7,18] and swimming [19,20]. Here, we probe the idea that elastic recoil might be important during aquatic food processing, where high viscosity and drag [20] limit the speed at which elements such as jaws or fins can move [21]. Food might escape the jaws of a predator if gape is prolonged following a strike, but the escape risk remains during food processing. Thus, while elastic recoil has been shown to confer benefits during the fast, gape-expansive behaviours involved in food capture [10–14,22,23], we suspect that elastic recoil may also be beneficial during the slower, gape-compressing movements involved in food processing. As the food is processed, interactions of the jaw and the adductor musculature with food substrates that have different mechanical properties may influence the generation of reaction forces and the potential for elastic recoil action.

There is indirect evidence of elastic energy storage in the jaw system during food capture from earlier work on the tiger salamander (Ambystoma tigrinum); Reilly & Lauder [24] used electromyography (EMG) to demonstrate that the adductor mandibulae internus (AMi) was activated approximately 20–30 ms prior to the onset of jaw-closing during the strike. Without direct measurements of muscle length, it was not possible to determine the length-change trajectory of muscle contraction and demonstrate elastic recoil action.

Here, we use fluoromicrometry to measure the strain of the adductor mandibulae externus (AMe) and its entire muscle–tendon unit (MTU) [4,5] in the Axolotl (Ambystoma mexicanum). We test the hypothesis that, following food capture, the muscle fascicles undergo an active stretch-shortening cycle as the gape is opened and closed during food processing. We predict a period of muscle fascicle shortening as the MTU is stretched during gape expansion, resulting in a significant stretch of the tendinous structures associated with ambystomatid jaw muscles [8,9]. Since collagenous structures are capable of recoiling much faster than the shortening speed of contracting muscles [2], we predict the speed of jaw closing (as driven by MTU shortening) to be amplified in the presence of elastic recoil, as compared to a hypothetical stiff jaw adductor muscle system [5].

By using two food types with differing mechanical properties (hardness, toughness) and escape abilities, we test the hypothesis that the contribution of elastic recoil to jaw-closing speed may be modulated according to the compressive force requirements and risk of prey escape associated with food processing. Because the magnitude of compliant tissue strain, and therefore the effects of elastic recoil on jaw closure speed, could potentially be modulated by variation in the duration and intensity of AMe activation during MTU lengthening, we predict that food types of different mechanical properties and escape abilities will elicit different amounts of speed amplification during MTU recoil in AMe.

2. Material and methods

(a). Animals

Six adult female Axolotls of similar size (TL 24.22 ± 2.01 cm; SVL 12.72 ± 1.21 cm; body mass 71.20 ± 18.73 g) (mean ± s.d.) were purchased from the Ambystoma Genetic Stock Center. They were housed, trained to feed on food held in forceps, and implanted at UMass Lowell and then transported to Harvard University's Concord Field Station for experiments. Animals were anaesthetized in a 2.5% aqueous solution of benzocaine (Orajel) so that radio-opaque markers could be implanted. Procedures for marker implantation are described in detail elsewhere [4,5], but briefly: marker implants of relevance to this study included two 0.5 mm tantalum markers at the ends of the AMe muscle belly to measure changes in fascicle length (L_fasc) (figure 1). We also implanted two 0.5 mm markers implanted into the bone at the insertion and origin points of the AMe on the mandible and squamosal bones, respectively, to measure changes in the length of the whole MTU (L_MTU). Subjects resumed normal feeding behaviours within hours after surgery but were given 6 days of recovery to allow markers to lodge in place before data collection. In a second anaesthesia procedure, right before the experiment, we implanted bipolar-offset fine-wire EMG electrodes into several muscles, including the AMe (mid-belly) to measure muscle activity during feeding.

Figure 1.

3D reconstruction of the Axolotl skull, mandible and the adductor mandibulae externus (AMe) muscle–tendon unit (MTU). Example locations of implanted radio-opaque markers are shown as spheres, along with measurements of fascicle length (L_fasc, red) and MTU length (L_MTU, blue bracket). Subtraction of L_fasc from L_MTU allows for estimation of the length of the tendon structures that connect AMe to the mandible (ventrally) and the squamosal–neurocranial junction (dorsally). (Online version in colour.)

(b). Data collection

Animals were habituated for several days to a custom-built tank (910 × 70 × 150 mm; 7 mm acrylic wall thickness), which could be positioned within the visualization volume of two near-orthogonally arranged high-speed video cameras (Photron PCI-1024) mounted onto c-arm fluoroscopes (GE-OEC 9600). Biplanar x-ray videos of food processing kinematics were recorded, so that gape, MTU and muscle length-change trajectories could be measured for each cycle [4,5]. Minimally filtered EMG data were acquired at 4000 Hz via an amplifier (Grass P511) through an A/D converter (Powerlab 8/30 ADinstruments) to software (ADinstruments LabChart). We also recorded a square-wave pulse from the camera software so that EMG data could be synchronized with the digitized kinematics data for analyses.

We sought to compare data from food processing gape cycles on two distinct food types, cricket and fish, which differ in their escape ability and materials properties. Both common house crickets (Acheta domesticus) and rosy minnows (Pimephales sp.) were presented to all six animals, but data for both food types could only be collected from four (sample is detailed in electronic supplementary material, table S1). We carefully size-matched crickets and minnows and size selected food to equal one mouth width of the subject. After feeding trials, subjects were euthanized in an overdose of benzocaine (4%) and fixed in 10% phosphate-buffered formalin.

(i). Measurement conditioning and data analyses

The phase-locked video pairs were undistorted and calibrated in XMAlab 1.5.0 [25]. Three-dimensional (3D) movement data of the tantalum markers were then tracked frame-by-frame, extracted from XMAlab, imported to Igor Pro 7.0 (Wavemetrics) and used to calculate inter-marker distances for muscle fascicle beads (L_fasc).

In order to confirm maker placement and perform segment corrections that took into account any muscle fascicle distance not contained between our intramuscular markers (figure 1), micro-CT scans were taken of all cadavers after soft-tissue contrast-enhancement in a 4% alcoholic phosphomolybdenum-acid solution for 5 days [26]. We acquired 0.036 mm image-slices using a 90 kV x-ray tube voltage and 110 µA continuous power with a 0.25 mm brass filter (Bruker Skyscan 1173). X-ray projections were then reconstructed in NRecon Reconstruction Software (Micro Photonics, Allentown, PA) with an automatic beam hardening correction factor. Muscle, bone and marker elements where then segmented (The Horos Project) and imported to Maya (Autodesk) [27] where the locator tool from the XROMM tool-shelf was used to measure MTU length as well as inter-marker distances.

For analyses, gape cycles were selected following the food-capture strike, with a cycle being defined as the period where gape excursion transitioned from inactive closed, via peak expansion, and back to inactive closed [28]. To calculate tendon length (L_ten), we subtracted instantaneous L_fasc from L_MTU. We differentiated true L_fasc and L_MTU to obtain mean speed measurements. Both these measurements were normalized to the segment-corrected measurement of resting muscle fascicle length (L_rf), as measured from videos of the quiescent live animal [5]. Lengthening speeds are reported as negative values. To calculate strain, we used resting length measurements for the muscle and MTU (ε = ΔL/L_r). We calculated speed amplification by dividing average V_MTU during jaw closing, i.e. the additive speed of elastic recoil and any residual muscle fascicle shortening, by the average speed of muscle shortening (V_fasc) during gape opening, i.e. the speed with which muscle contraction stored energy into elastic structures.

(ii). Data representation and statistical analyses

Summary data on duration of fascicle shortening accompanied by MTU lengthening, data on tendon strain and speed amplification due to elastic recoil are shown in box and whisker plots giving the median, quartile and range of data (figure 4). In the box plot for EMG duty factor (figure 4a), the box bounds give mean onset timing (top), and mean offset timing (bottom), with whiskers indicating timing variability (s.e.m.) across individuals and gape cycles for a given food type. We used onset of gape closing as the reference time for calculations of EMG duty factor. To determine if kinematics, muscle–tendon function, and EMG varied with respect to food type we used General Linear Models in Systat (v.11). These analyses factored the effect of the individual (random factor) as well food type (fixed effect).

Figure 4.

Mechanical determinants of speed amplification in the Axolotl AMe musculature. (a) Duty factor of AMe activation. (b) Interval where fascicle shortening and MTU lengthening co-occurred. (c) Peak tendon strain (ε_tendon = ε_MTU − ε_fasc) per gape cycle. (d) Speed amplification conferred by MTU recoil during jaw closing. (e) Shortening speed for AMe fascicles (grey) and the whole MTU (black). (a) Shows activation onset (upper box bound) and offset (lower box bound) as grand mean ± s.e.m. (whiskers) for all cycles, with respect to the onset of gape closing (dotted line). (b–e) Show median (line), quartiles (box-bounds) and data range (whiskers) with dots representing individual processing cycles.

3. Results

During a food processing cycle of an Axolotl, the duration of gape closing is significantly faster on crickets (GLM; F_1,147 = 8.359; p = 0.004) than fish, and there is a significant effect of individual behaviour (F_1,147 = 2.566; p = 0.030). However, other gape kinematics variables are not statistically significantly different with respect to food type, but there are significant individual effects, including for total gape cycle duration (F_1,147 = 0.984; p = 0.323; effect of individual F_1,147 = 7.828; p < 0.001), active gape duration (F_1,147 = 2.298; p = 0.132; individual F_1,147 = 7.068; p < 0.001) and gape opening duration (F_1,147 = 0.177; p = 0.675; individual F_1,147 = 9.496; p < 0.001).

The AMe muscle is pre-activated relative to the onset of gape closing, by 0.014 ± 0.008 s for cricket and 0.016 ± 0.006 s (mean ± s.d.) for fish with no statistically significant difference between food types (F_1,71 = 1.74; p = 0.19). However, the duration of muscle activity during cricket processing (34.84 ± 23.84%) is significantly longer compared to during fish processing (21.07 ± 28.81%) (F_1,71 = 14.96; p < 0.0001) (figure 4a).

The fascicles of AMe begin shortening early in the gape cycle, well before the onset of MTU lengthening that by necessity accompanies gape opening (figure 2a). We found a statistically significant effect of food type on the duration of this muscle pre-shortening (F_1,45 = 10.92; p = 0.002) (figure 4b), which lasts longer on crickets (0.097 ± 0.064 s) (mean ± s.d.) than on fish (0.056 ± 0.056 s). The tendinous structures within AMe operates at greater strains during cricket processing (4.62 ± 2.94%) as compared to fish processing (2.51 ± 1.72%) (F_1,36 = 19.63; p = 0.0001) (figure 2, figure 4c).

Figure 2.

Three gape cycles from one individual illustrating the time-varying strain of the AMe muscle (red, pale), MTU (black, dark) and tendon (blue) during processing of cricket (b) and fish (c). Speed of muscle fascicles (red) and MTU (black) is shown in the bottom row. Grey bars indicate period of gape closing. For cricket processing, note the periods of AMe fascicle shortening associated with MTU lengthening, after which recoil of the MTU drives gape closing, while the AMe fascicles remain isometric, the hall mark pattern of elastic recoil action. Stretch-and-recoil dynamics are less clear for fish processing, where AMe fascicle shortening appears in closer coupling with MTU shortening and gape closing. Note that the AMe tendon and MTU operate at greater strains, and thus force for cricket than fish processing (see text). Shortening speeds (positive V) are lower for AMe fascicles than the MTU, as expected for elastic recoil-driven speed amplification. (Online version in colour.)

During gape opening, which is the phase where elastic energy can be stored by the AMe tendon, the speed of muscle shortening is borderline faster on cricket (0.95 ± 0.62 L_r s⁻¹) than fish (0.78 ± 0.53 L_r s⁻¹) (F_1,53 = 4.04; p = 0.05) (figure 3, figure 4d). The resulting speed of MTU recoil is significantly greater on cricket (3.01 ± 1.45 L_r s⁻¹) than fish (2.05 ± 0.95 L_r s⁻¹) (F_1,52 = 6.91; p = 0.011) (figure 4e). The degree of speed amplification conferred by elastic recoil differs significantly across food types; series elastic action amplifies the speed of MTU shortening and jaw occlusion by a factor of 4.36 ± 2.52 on cricket, but significantly less on fish (F_1,54 = 8.60; p = 0.005), by only a factor of 2.51 ± 1.72 (figure 4d).

Figure 3.

Plot of mean AMe fascicle shortening speed (V_fasc) against mean MTU shortening speed (V_MTU) for each gape cycle. Colours separate subjects, circles indicate cricket prey and squares indicate fish prey, respectively. Data distribution above the line of unity (V_fasc = V_MTU; long dash) indicates that MTU shortening is faster than fascicle shortening. (Online version in colour.)

4. Discussion

Our data show that food processing in the Axolotl (A. mexicanum) involves elastic recoil action to varying degrees between two food types with distinct materials and behavioural properties. The MTU of AMe is stretched during jaw opening, while the fascicles of the muscle shorten. Jaw closing then results from MTU recoil, as the AMe fascicles either continue to shorten or remain isometric. The decoupling of AMe shortening and gape closing is seen more consistently during cricket than fish processing (figure 2). This pattern demonstrates stretch and recoil of compliant tissues within the adductor MTU, resulting in a stretch-shortening cycle that amplifies gape closing speed, from 2.5-fold on average during fish processing to 4.4-fold on average during cricket processing.

The finding that speed amplification via elastic recoil is food-type dependent is unexpected given previous studies of feeding in ambystomatids, which concluded that ambystomatid feeding is stereotyped with respect to food type [29]. However, most performance variables in our study suggest that food processing in the Axolotl relies on cues that signal differences in prey mechanical or behavioural properties. During cricket processing, factors such as the prolonged period of muscle fascicle shortening during MTU lengthening, the increased tendon operating length and strain, and the increased duration of muscle activity may all contribute to the observed speed amplification, as compared to fish processing. Specifically, the increase in tendon operating length-range and strain, despite the relatively consistent length-change trajectory for the muscle fascicles across food types (figure 3), suggest that cricket processing involves greater production of compressive force by the AMe. Common to processing on both food types is a pattern of peak tendon stretch coinciding with the onset of jaw closing. This strain pattern indicates that the application of force to food compression occurs during jaw closure, as previously observed in other salamander species [30,31]. The speed of jaw closing is also amplified to a greater extent during cricket than fish processing, suggesting there is likely an increase in the mechanical power (P = F • V) involved with processing of crickets.

The pattern of increased gape closing and MTU shortening speed and likely also force and power for cricket processing matches expectations of a behavioural response to the hard cuticle found in most insects [32]. However, some data suggest that the common house cricket (A. domesticus) used in this study has a relatively soft cuticle [33], and that fish likely was the harder food treatment in our study [34–36]. The use of greater compressive force, speed and power for processing cricket compared to fish suggest that Axolotls might have an ingrained neurobehavioural response to food types, an interpretation that is consistent with prior results [29] indicating a stereotyped prey-type response during food acquisition, which, possibly, is triggered prior to prey capture through visual and chemoreceptive cues [37,38]. It is also possible that the faster MTU recoil speed during cricket processing results from a more readily compressible soft body than the harder minnow body.

Analyses of gape kinematics during Axolotl food processing show that the only variable with a food-type effect is the duration of gape closing, which is significantly faster during cricket than fish processing. This is also the gape cycle phase that is at least partially driven by elastic recoil action. Our performance data for gape closing speed contrast with earlier conclusions of ambystomatids having a strike behaviour that is invariant across food types [29,37]. However, consistent with these earlier studies all gape kinematics variables measured showed clear evidence of individual variability. Whereas our findings support the notion that cricket as a food type elicits unique neurobehavioural responses from the Axolotl, data from other ambystomatids and salamander species will be needed to fully understand these results.

Despite the impressively rapid and powerful feeding behaviours that can be observed across the animal kingdom, elastic recoil action, specifically in fluid environments, has rarely been demonstrated for feeding systems. When described, these instances are exclusively ballistic prey acquisition mechanisms with great speed and power demands, such as amphibian tongue projection [10,11,13,14,22] and teleost suction feeding [12,23]. In these systems, a biological spring is stretched slowly by muscle during a prolonged preparatory phase, then allowed to recoil. Elastic recoil can occur much faster than the peak shortening speed of muscle (approx. 16 resting lengths per second [39,40]). However, elastic recoil action in slower, cyclical movements, such as those involved in food processing might confer equally important ecological benefits. For instance, elastic recoil may benefit by minimizing risks of prey escape and enhance processing performance, but such mechanisms remain understudied.

The degree of speed amplification we demonstrate here is well within the speed capabilities of vertebrate skeletal muscle [39,40]. We posit that delivering the force available for food processing at a greater speed is likely functionally significant to facilitate rapid gape closure in a dense and viscous fluid environment. The evolutionary relevance of this speed amplification might be an abbreviation of the time the mouth is kept open, and thus a reduction of the risks of prey escape in a dense and viscous fluid environment where gravity offers little aid to food handling. A shorter duration of the gape cycle is likely important for the neotenic Axolotl, which does not naturally undergo metamorphosis and feeds exclusively underwater. Going forward, we anticipate studies of elastic structures in muscles of the jaw and tongue of Axolotls and other salamanders to present a fertile system for broadening our understanding of elastic mechanisms in aquatic, as well as terrestrialized craniofacial systems.

Ethics

All procedures and experiments adhered closely with approved IACUC protocols from UMass Lowell and Harvard University (FAS).

Data accessibility

Data generated for this study are available under the permanent ID BROWN47 on the XROMM portal (http://xmaportal.org/webportal/).

Authors' contributions

N.K. conceptualized the project, N.K. and M.R. collected in vivo data, J.S. and M.R. analysed data and J.S. generated XROMM models. All authors participated in writing, editing and approving the manuscript for submission.

Competing interests

We declare we have no competing interests.

Funding

Funded by UML start-up (N.K.) and two UML Honors college awards (M.R., J.S.).