JOURNAL OF MORPHOLOGY 271:1527–1536 (2010)

Motor Patterns of Distal Hind Limb Muscles in Walking Turtles: Implications for Models of Limb Bone Loading Heiko L. Schoenfuss,1 John D. Roos,1 Angela R. V. Rivera,2 and Richard W. Blob2* 1 2

Aquatic Toxicology Laboratory, St. Cloud State University, St. Cloud, Minnesota Department of Biological Sciences, Clemson University, Clemson, South Carolina

ABSTRACT Previous studies of limb bone loading in walking turtles indicate that the ground reaction force exerts a flexor moment at the ankle during stance, requiring extensor muscle activity to maintain joint equilibrium. Of four proposed ankle extensors in turtles, two (gastrocnemius medialis, pronator profundus) originate on the tibia and fibula, respectively, while the other two (flexor digitorum longus, gastrocnemius lateralis) originate from the distal femur, crossing the flexor aspect of the knee and potentially eliciting compensatory forces from antagonist knee extensor muscles that could contribute to femoral stress. Published bone stress models assume all four proposed ankle extensors are active during stance in turtles. However, if only the ankle extensors that cross the knee were active then femoral stresses might be higher than predicted by published models, whereas if only extensors that do not cross the knee were active then femoral stresses might be lower than predicted. We analyzed synchronized footfall and electromyographic activity patterns in slider turtles (Trachemys scripta) and found that all four proposed ankle extensors were active during at least part of stance phase in most individuals, corroborating bone stress models. However, activation patterns were complex, with multiple bursts in many ankle extensors that frequently persisted into swing phase. In addition, two hypothesized ankle flexors (tibialis anterior, extensor digitorum communis) were frequently active during stance. This might increase the joint moment that ankle extensors must counter, elevating the forces they transfer across the knee joint and, thereby, raising femoral stress. Recognition of these activity patterns may help reconcile differences between evaluations of loads on turtle limbs based on force platform versus in vivo strain studies. Moreover, while some variation in motor patterns for the distal hind limbs of turtles may reflect functional compartmentalization of muscles, it may also indicate flexibility in the control of their limb movements. J. Morphol. 271:1527–1536, 2010. Ó 2010 Wiley-Liss, Inc.

pherson, 1991; Roy et al., 1991; Biewener and Gillis, 1999; Gillis and Blob, 2001; Kargo and Rome, 2002; Reilly and Blob, 2003). A variety of factors may influence which muscles among a group of potential synergists are actually active during the production of a particular motor behavior (Zernicke and Smith, 1996; Biewener and Gillis, 1999; Gillis and Blob, 2001). For example, if muscles across a joint show differences in fiber type proportions, primarily slow oxidative muscles might be activated for slower motions, whereas primarily fast oxidative-glycolytic or fast glycolytic muscles might be activated during faster motions. Alternatively, if two muscles crossing the same joint exhibit different lever arm ratios, muscles showing high mechanical advantage might be activated for behaviors requiring high force output, whereas muscles showing high velocity advantage might be activated for behaviors requiring rapid movement. Thus, redundancy among limb muscles can facilitate both breadth of function and refinement of performance in the execution of motor behaviors. Although of great significance to vertebrate limb performance, redundancy in limb muscle systems can complicate the evaluation of some aspects of limb function. This is particularly true for studies of skeletal loading. The limb bones of terrestrial vertebrates must bear loads that result from their role in supporting the body against gravity, as well as from forces exerted by the actions of limb muscles during stance and movement (Alexander, 1974; Biewener, 1983, 1989, 1990). One common way in which the loads borne by limb bones are evaluated is from quasi-static equilibrium models derived from simultaneous measurement of ground

KEY WORDS: locomotion; biomechanics; electromyography; turtle; bone loading; muscle; modeling

Contract grant sponsor: NSF; Contract grant number: IOS0517340.

INTRODUCTION In many limbed vertebrates, the muscles that control limb movements are highly redundant systems, with multiple potential synergists and antagonists situated in positions that can generate or restrict movement at a particular joint (MacÓ 2010 WILEY-LISS, INC.

*Correspondence to: Richard W. Blob, 132 Long Hall, Department of Biological Sciences, Clemson University, Clemson, SC 29634. E-mail: [email protected] Received 28 June 2010; revised 13 August 2010; Accepted 22 August 2010 Published online 21 October 2010 in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/jmor.10901

1528

H. L. SCHOENFUSS ET AL.

Fig. 1. Schematic diagram of muscular forces and GRFs acting across the ankle and knee joints of the turtle hind limb, modified from Butcher and Blob (2008), with portions of the shell and pelvis (black shading) illustrated for reference. To satisfy the condition of static rotational equilibrium assumed in force platform-based models for calculating limb bone stress (Eq. 1), the moment exerted at the ankle by ankle extensor muscles must equal the flexor moment imposed at the ankle by the GRF, which is calculated as the vector product of the GRF and its moment arm at the ankle (RGRF). The moment exerted by ankle extensor muscles was initially modeled (Butcher and Blob, 2008) as the vector product of the summed force of uniarticular (pronator profundus, gastrocnemius medialis) and biarticular (gastrocnemius lateralis, flexor digitorum longus) ankle extensors, and their mean moment arm at the ankle joint (rm). Because the biarticular ankle extensors cross the knee joint, any flexor moment they exert must be countered by knee extensor muscles (femorotibialis, iliotibialis), the contraction of which will place stress on the femur (grey shading). An additional condition of the model is that ankle flexor muscles (tibialis anterior, extensor digitorum communis) are inactive during stance. Stance phase activity by these muscles would add to the flexor moment that ankle extensors would have to counter, and potentially transfer a larger moment across the knee, requiring greater knee extensor force to maintain equilibrium and potentially elevating femoral stress.

reaction forces (GRFs) (via force platform recordings) and limb position (via film or video of limb kinematics) (Biewener and Full, 1992). Assuming a state of quasi-steady equilibrium, these data can be used to calculate estimates of the forces exerted by muscle groups (and transferred to the skeleton) by the equation: Fm ¼ ðRGRF 3GRFÞ=rm ;

ð1Þ

where Fm is the force exerted by the muscle, GRF is the ground reaction force, RGRF is the moment arm of the GRF about the specified joint, and rm is the weighted mean moment arm of the muscles active to counter the GRF at that joint, with weighting determined by the relative force generating capacities of the muscles as reflected by their physiological cross-sectional areas (Fig. 1). The contributions of individual muscles to the total Fm Journal of Morphology

exerted by a group can then be estimated based on their relative physiological cross-sectional areas, with calculations for a limb typically beginning distally and proceeding to more proximal joints (Alexander, 1974; Biewener and Full, 1992). However, only muscles that are actually active during a behavior should be considered as contributing to this Fm (Blob and Biewener, 2001; Butcher and Blob, 2008). Knowledge of which synergists are active or inactive at a joint generally would not affect estimates of the total muscle force exerted across the joint in question (Biewener and Full, 1992). However, if any muscles under consideration were biarticular, knowledge that they were inactive would affect estimates of the muscle forces transferred up the limb to more proximal joints and limb bones (Blob and Biewener, 2001; Butcher and Blob, 2008). To proceed with calculations of muscle force from equilibrium models, a further initial assumption is typically made that at the distal joint where the analysis begins, only muscles antagonistic to the GRF are active (Alexander, 1974; Biewener, 1983, 1989; Biewener and Full, 1992; Blob and Biewener, 2001; Butcher and Blob, 2008). This assumption is necessary to resolve the muscular contributions to limb bone loads, to prevent unknown muscle forces from occurring on both sides of the equilibrium equation (Eq. 1; Biewener, 1983; Blob and Biewener, 2001; Butcher and Blob, 2008). However, if this assumption is incorrect for a given system and muscles on the same side of a joint as the GRF were active, then muscles antagonistic to the GRF would have to exert higher forces to maintain equilibrium at the joint. Such additional muscular forces could be transferred to the limb bones and elevate actual limb bone loads higher than estimates based on equilibrium models. Recent studies of bone loading in the hind limbs of turtles (Butcher and Blob, 2008; Butcher et al., 2008) provide an opportunity to evaluate the potential impact of muscular activity assumptions on models of skeletal loading based on force platform analyses. Limb bone loading in turtles was examined as part of a broader effort to evaluate the diversity and evolution of limb bone loading mechanics and safety factors in tetrapods, particularly whether ectothermic species that use sprawling limb posture might typically exhibit higher limb bone safety factors than endothermic species using more upright limb posture (Blob and Biewener, 1999, 2001; Butcher and Blob, 2008; Butcher et al., 2008). These studies found that, although all estimates of limb bone safety factors (in bending) for turtle femora were higher than those found in endothermic, upright species, calculations based on force platform analyses (Butcher and Blob, 2008) produced femoral safety factor estimates considerably higher (13.9) than calculations based on direct, in vivo measurements of skeletal

HIND LIMB EMGS IN TURTLE WALKING

loads using strain gauges (6.9: Butcher et al., 2008). Safety factors as high as those determined from force platform analyses would indicate that turtles diverge substantially even from other reptiles (with limb bone safety factors typically <10: Blob and Biewener, 2001), suggesting even greater evolutionary diversity in tetrapod bone loading patterns. However, the differences in results between experimental techniques might also indicate that force platform analyses underestimate femoral loading magnitudes in turtles, potentially through the assumptions of muscle activity employed in force-based loading models. Force platform data indicate that in turtles, as is typical in tetrapods, the GRF exerts a moment that would tend to dorsiflex the ankle during stance (Butcher and Blob, 2008). As a result of this dorsiflexing GRF moment at the ankle (hereafter referred to as a ‘‘flexor moment’’), ankle extensor muscles must be active to counter the GRF. Of four potential ankle extensors in turtles (Walker, 1973), two (gastrocnemius medialis and pronator profundus) originate on the tibia and fibula, respectively, while the other two (flexor digitorum longus, gastrocnemius lateralis) originate from the distal femur, crossing the flexor side of the knee and potentially eliciting antagonist forces from muscles on the extensor surface of the femur that could contribute to femoral stress (Fig. 2). Our published bone stress models assumed that all four ankle extensors are active during stance (Butcher and Blob, 2008). However, if only the ankle extensors that cross the knee were active during stance then femoral stresses might be higher than predicted, because the entire force exerted to counter the GRF at the ankle would be transferred across the knee, after which it would have to be countered by knee extensor muscles that span the femur and subject that bone to stress (Fig. 1). In contrast, if only the ankle extensors that did not cross the knee were active during stance, then stresses might be lower than predicted by our model because none of the force exerted to counter the GRF at the ankle would be transferred across more proximal joints (Fig. 1). Our published model of femoral loading in turtles based on force platform data also assumed inactivity of muscles on the flexor surface of the shank during stance (Butcher and Blob, 2008). However, this assumption has not been tested, and if such activity were present, it could require ankle extensors to compensate and exert greater forces that could cascade up the limb and also elevate stresses even on proximal bones such as the femur (Fig. 1). To assess the roles of the shank muscles in production of limb bone stresses in turtles, we collected recordings of electromyographic activity (EMG) during treadmill walking from the four putative ankle extensor muscles of red-eared slider turtles, as well as two muscles on the flexor sur-

1529

face of the shank (tibialis anterior and extensor digitorum communis) (Fig. 2). These data allowed us to evaluate patterns of coactivation for potentially synergistic extensor muscles across the ankle joint in walking turtles, evaluate the functional roles of muscles on the flexor surface of the turtle shank during the swing and stance phases of walking, and test current assumptions about shank muscle activity that have been incorporated into models for calculating femoral stresses in turtles, potentially providing insight into the differences in results from force-platform- versus straingauge-based analyses of limb bone loads.

MATERIALS AND METHODS Experimental Animals Concurrent kinematic and EMG data were collected from eight male slider turtles (Trachemys scripta) during treadmill locomotion to assess the timing and contribution of ankle extensor and flexor muscles to the locomotor cycle. Sliders are a generalized species from the emydid lineage that spends considerable time traveling over land at slow to moderate speeds, particularly between bodies of water (Gibbons, 1970; Gibbons et al., 1990; Bodie and Semlitsch, 2000). Turtles (287 6 31 g, mean 6 s.e.m.) were purchased from a commercial vendor (Concordia Turtle Farm, Wildsville, LA) and housed in groups in 150-gallon stock tanks equipped with pond filters and dry basking platforms. Tanks were located in a temperature-controlled greenhouse, providing ambient light regimes. Turtles were fed daily a diet of commercially available pellets (Reptomin). All animal care and experimental procedures were conducted in accordance with Clemson University IACUC guidelines (protocol 50110).

Electromyography Before data collection, turtles were evaluated for their willingness to walk on a treadmill (Jog A Dog model DC5, Ottawa Lake, MI) with proficient animals selected for experiments. Electrodes were implanted in the morning of the recording day, following modifications of established protocols (Loeb and Gans, 1986; Gillis and Blob, 2001; Blob et al., 2008, Rivera and Blob, 2010) that were applied to minimize stress to the animals, facilitate rapid recovery from implant procedures, and improve signal quality. Local anesthetic (0.025 ml Lidocaine1) was injected into the belly of the shank muscle mass, allowing percutaneous implantation of bipolar fine-wire electrodes (0.002 mm diameter; insulated stainless steel; 1 mm barbs; California Fine Wire, Grover Beach, CA) using hypodermic needles. External landmarks for implants were determined through dissection of additional turtle specimens before experiments. Once an electrode needle was inserted to the desired depth in a muscle the needle was carefully retracted, leaving only the electrode wire behind. This procedure was repeated three times, from proximal to distal, for each of the four putative extensor muscles. By targeting three implants for each of our focus muscles, we included sufficient redundancy to accomplish good recording coverage for most turtles and target muscles, and also obtained incidental sampling of two additional muscles on the anterior (dorsiflexor) surface of the shank (Table 1). For each animal, all 12 needles were pulled back along the electrode wire and attached to the highest (dorsal) portion of the carapace with adhesive tape. The wire bundle was then secured to the skin near the insertion by a suture. The total weight of this structure was less than 4 g (
1% animal weight) and did not produce any discernable effects on the behavior of the turtles. The free ends of each electrode were soldered into a

Journal of Morphology

1530

H. L. SCHOENFUSS ET AL.

Fig. 2. Hindlimb musculature of Trachemys scripta, modified from Walker (1973), highlighting focal muscles from which EMG data were collected. (A) Superficial extensors of the ankle and flexors of the knee, left side, posterior view. Shading highlights gastrocnemius lateralis (red), gastrocnemius medialis (green), and flexor digitorum longus (blue). (B) Deep extensors and flexors of the ankle, left side, deep posterior view. Pronator profundus highlighted in yellow. (C) Extensor and flexor muscles of the ankle, left side, ventral view. Color scheme as above with tibialis anterior (purple) and extensor digitorum communis (dark blue) added. See Walker (1973) for key to muscle name abbreviations. microconnector that was plugged into a shielded cable to conduct signals to EMG amplifiers (Grass 15LT system, West Warwick, RI) for amplification (10,0003) and filtering (60 Hz notch filter, 30 Hz–6 kHz bandpass). These procedures allowed turtles to recover rapidly after implantation and dramatically shortened the length of unshielded wire required, improving signal clarity.

Journal of Morphology

Approximately 4 hours after completion of implants, experimental turtles were placed on the treadmill for recording sessions. Belt speed was set at a minimum to stimulate consistent walking steps (<0.5 m/s). Analog EMG signals were converted to digital and collected at 5000 Hz using custom LabVIEW (v.6.1; National Instruments, Austin, TX) routines. Simultaneous with EMG recordings, turtles were filmed in lateral view

HIND LIMB EMGS IN TURTLE WALKING

1531

TABLE 1. Recording sample sizes from shank muscles of individual turtles examined in this study Anatomical ankle extensorsa Biarticular

Anatomical ankle flexorsa

Uniarticular

Turtle

Weight (g)

Pronator profundus

Gastrocnemius medialis

Gastrocnemius lateralis

Flexor digitorum longus

Tibialis anterior

Extensor digitorum communis

TS03 TS20 TS26 TS34 TS35 TS62 TS70 TS75

357 226 191 189 295 435 269 225

20 n/m n/m n/m n/m n/m 13 19

n/m 20 20 20 n/m 14 n/m 20

20 20 20 19 n/m 16 n/m n/m

n/m 20 20 20 20 16 13 20

n/m n/m n/m n/m 20 n/m n/m n/m

n/m n/m n/m 20 20 n/m n/m 20

The number of locomotor cycles for which EMGs were analyzed is indicated for each muscle and each turtle. n/m, not measured. a Designations of muscles as anatomical extensors or flexors based on descriptions of Walker (1973). (100 Hz) with a digital high-speed video camera (Phantom V4.1, Vision Research; Wayne, NJ) to allow evaluation of the timing of footfalls and ankle movements relative to muscle bursts. White correction fluid was used to mark the knee, ankle, and tip of digit 5 to enhance the visibility of these landmarks on videos, and EMG data were synchronized with kinematic videos by triggering a signal generator that simultaneously produced a light pulse visible in the video and a square wave in the EMG. Between 9 and 22 multicycle locomotor bouts were collected for each turtle (Table 1), with turtles allowed to rest for 5–15 minutes between bouts. Following data collection, turtles were euthanized (Euthasol, Delmarva Laboratories, Midlothian, VA, 200 mg/kg intraperitoneal injection) and electrode positions were verified by dissection. Before analysis, recordings were reviewed to select locomotor cycles for analysis based on the clarity of both kinematic and EMG data, avoiding the first and last steps in bouts. Custom LabVIEW (v. 6.1, National Instruments) software was used to determine the onset, offset, and duration of each activity burst from every muscle from which successful recordings were obtained. For individuals in which multiple recordings were obtained from a single muscle, the recording with the highest signal-to-noise ratio and clearest definition of onset and offset was analyzed. Joint positions during stance phase were digitized for every other video frame from 157 kinematic videos distributed across all eight individuals, using a modification of the public domain NIH Image program for Macintosh, developed at the US National Institutes of Health (the modification, QuickImage, was developed by J. Walker and is available at http:// www.usm.maine.edu/~walker/software.html). Limb position coordinate data were then processed using custom Matlab (v. 5.0.0, The MathWorks, Natick, MA) routines to calculate ankle kinematics and footfall timing. Calculated values for kinematic variables from each limb cycle were fit to a quintic spline (Walker, 1998), smoothing the data and allowing all of the trials to be normalized to the same duration for comparison (Blob et al., 2008).

RESULTS Although EMG recordings were not obtained from all muscles in all eight turtles, data were collected from between three and seven turtles for all four proposed ankle extensor muscles as well as the extensor digitorum communis, a putative ankle flexor. Additionally, data from the putative ankle flexor tibialis anterior were collected from one turtle (Table 1). To characterize muscle contraction

patterns, we analyzed between 13 and 20 locomotor cycles spread across a minimum of three locomotor bouts for each turtle, allowing us to assess pattern variation within and across individuals. Kinematics of the Distal Hindlimb in Slider Turtles The transition from swing phase, when the foot was off the ground, to stance phase, when the foot was placed on the ground and supported the body, occurred approximately one quarter (28% 6 0.5%) through the locomotor cycle (Fig. 3). Stance phase begins with the knee anterior to the ankle and the ankle highly flexed (
308), but starting by the middle of stance the ankle begins to extend rapidly to
1408 by the time the toe is lifted from the ground. Motor Patterns of Putative Ankle Extensor Muscles in Slider Turtles Pronator profundus. Activation patterns for the pronator profundus were assessed in three turtles (TS03; TS70; TS75). This deep, fan-shaped muscle originates broadly on the fibula and inserts on the first and second tarsals and the base of the first metatarsal (Walker, 1973; Fig. 2B). Contraction patterns for this muscle were consistent among all three turtles from which recordings were obtained, with one long burst beginning late in swing phase and continuing for most of stance phase in almost every (52/53) cycle evaluated (Fig. 3C; Table 2). The onset of this burst thus was close in time to the touchdown of the foot, and its offset occurred during the rapid ankle extension of the latter part of stance (Fig. 3B). Occasionally (
1532

H. L. SCHOENFUSS ET AL.

Fig. 3. Summary of distal hindlimb motion and control for slider turtles during walking, assessed through synchronized kinematic and EMG data. (A) Representative images of a turtle walking on the treadmill at the instants of (from left to right) beginning of locomotor cycle, toe touchdown, mid stance, and just before toe lift-off. White ellipses highlight the ankle joint, and time after the start of the cycle is indicated for each frame. Original images were cropped and adjusted for brightness and contrast. (B) Changes in the angle of the ankle through the course of stance phase (beginning at an average of 28% cycle duration across all turtles); inset figure illustrates the angle measured. Points and error bars represent means (6 s.e.m.) of values calculated from 157 steps across all eight turtles examined. Arrows indicate the timing and ankle angle of corresponding video images from (A). (C) Plots of EMG burst timings obtained from turtles walking on a treadmill, with X-axes of (B) and (C) drawn to the same scale. Bars indicate time during which a muscle was active in the individual turtles indicated parenthetically to the left of the graph, with whiskers indicating s.e.m. of onset or offset times (see Table 2 for corresponding values). Discrete bursts for a single muscle are indicated with different colors (black5first burst, gray5second burst, white5third burst) and, when necessary, offset vertically for viewing clarity. Data from turtles exhibiting similar patterns are plotted in a single row. PP5 Pronator profundus; GM5 Gastrocnemius medialis; GL5 Gastrocnemius lateralis; FDL5 Flexor digitorum longus; TA5 Tibialis anterior; EDC5 Extensor digitorum communis.

only the ankle joint, originating on the tibia and inserting via the plantar aponeurosis on the bases of the first four digits and the fifth metatarsal (Walker, 1973; Fig. 2A, C). Across the five turtles in which we evaluated the activity of this muscle, we observed three characteristic patterns (Table 2, Journal of Morphology

Fig. 3C). In two turtles (TS20 and TS62) a long burst (
50% cycle duration) was present for most of stance phase in all evaluated cycles, with a short additional burst occasionally present (12/33 steps) near the swing-stance transition. In two other turtles (TS26 and TS34), a much shorter

HIND LIMB EMGS IN TURTLE WALKING

1533

TABLE 2. Burst onset and offset times for sampled hind limb muscles in slider turtles Muscle Pronator profundus Gastrocnemius medialis

Individuals recorded

Burst number

TS03, TS70, TS75

1 2 3 1 2 1 2 3 1 2 1 1 2 3 1 2 1 1 2 1 2 1 2 1 1

TS20, TS62 TS26, TS34 TS75

Gastrocnemius lateralis

TS3, TS20, TS26 TS62 TS34

Flexor digitorum longus

TS20, TS34, TS70, TS75 TS35, TS62 TS26

Tibialis anterior Extensor digitorum communis

TS35 TS34 TS35, TS75

Onset time (6s.e.m.)a 0 16.1% 6 79.0% 6 20.6% 6 43.5% 6 0 26.9% 6 82.4% 6 0 92.4% 6 24.9% 6 2.8% 6 23.5% 6 73.3% 6 0 88.0% 6 21.9% 6 18.1% 6 78.6% 6 8.5% 6 33.7% 6 1.6% 6 54.4% 6 8.2% 6 17.6% 6

0.9 3.1 3.4 2.0 2.0 2.5 1.2 0.6 0.8 5.7 8.1 4.7 0.5 0.6 2.9 1.0 2.2 0.3 4.1 0.6 0.8

Offset time (6s.e.m.)a 11.3% 6 87.1% 6 97.4% 6 30.3% 6 94.7% 6 16.1% 6 37.2% 6 100 28.1% 6 100 86.5% 6 23.6% 6 67.9% 6 98.0% 6 31.7% 6 94.1% 6 91.4% 6 79.3% 6 94.1% 6 27.0% 6 93.1% 6 20.1% 6 65.5% 6 21.0% 6 94.6% 6

1.0 1.9 0.6 2.4 0.6 0.7 2.1 1.0 1.2 5.9 5.4 2.0 2.8 3.9 0.8 2.5 0.7 2.5 1.2 0.9 4.1 1.4 0.7

Frequencyb 7/53 52/53 16/53 12/33 33/33 39/39 37/39 35/39 20/20 18/20 58/58 12/17 17/17 4/17 19/19 17/19 70/70 35/35 18/35 20/20 20/20 20/20 17/20 20/20 37/37

Data for each muscle are reported according to the pooling of individuals with similar patterns illustrated in Figure 2, with burst numbers referring to the order illustrated. a Onset and offset times are reported as percentages of total cycle duration. Onsets averaging under 1% or greater than 99% are assigned values of 0 and 100%, respectively. b Frequency refers to the number of analyzed cycles in which a burst was identified, out of the total number of cycles examined for the muscle in the designated individuals.

burst (lasting roughly 10% of the limb cycle) was present early in stance, with additional periods of activity early in swing and late in stance, reflecting a continuous burst spanning the transition from stance to swing. Finally, one turtle (TS75) showed a third pattern, with activity during all of swing phase that persisted into the beginning of stance, and usually (18/20 steps) a late stance burst reflecting continuous activity across the stance-swing transition. Despite these variations, in all five animals the gastrocnemius medialis was active for substantial portions (or nearly all) of stance phase. Gastrocnemius lateralis. The gastrocnemius lateralis originates from the fibular epicondyle of the femur and crosses the knee and ankle joints before inserting in common with the gastrocnemius medialis via the plantar aponeurosis (Walker, 1973; Fig. 2A, C). Activation patterns for this muscle were also diverse across the individuals we tested (Table 2, Fig. 3C). Three turtles (TS03, TS20, and TS26) exhibited a single long burst for gastrocnemius lateralis lasting for nearly all of stance phase. A fourth turtle (TS62) exhibited a somewhat shorter burst during the beginning of stance phase, but also frequently (12/17 steps) exhibited an additional discrete burst during swing phase and sometimes also a third burst (
a fifth turtle (TS34) showed swing phase activity that persisted into the beginning of stance phase, as well as a second, shorter burst during late stance phase. Nonetheless, as for the gastrocnemius medialis, in all five animals the gastrocnemius lateralis was active for substantial portions (or nearly all) of stance phase. Flexor digitorum longus. The largest of the putative ankle extensors, flexor digitorum longus originates from the fibular epicondyle of the femur and the shaft of the fibula and, like gastrocnemius lateralis, crosses the knee and ankle joints, inserting via tendons on the plantar surface of the terminal phalanges (Walker, 1973; Fig. 2A, C). Three distinct patterns of activity for the flexor digitorum longus were observed across seven animals, though the differences among these patterns are less substantial than the variations observed across the two heads of gastrocnemius (Table 2, Fig. 3C). Four turtles (TS20, TS34, TS70, TS75) showed a single burst lasting nearly the entire duration of stance phase. Two additional turtles (TS35, TS62) also showed a long burst spanning most of stance phase, but also showed a second, short burst later in stance phase during over half of the limb cycles we evaluated. Finally, one turtle (TS26) showed a long, stance phase burst similar to the pattern described for the first four turtles evaluated for this muscle, but also showed an Journal of Morphology

1534

H. L. SCHOENFUSS ET AL.

early burst that began in the middle of swing phase and persisted through the swing-stance transition. Despite such differences, flexor digitorum longus was active for most of stance phase across all seven turtles we examined. Motor Patterns of Putative Ankle Flexor Muscles in Slider Turtles Tibialis anterior. This muscle originates on the anterior margin of the proximal tibia and inserts on the first metatarsal (Walker, 1973; Fig. 2C). The single turtle (TS35) in which we observed contraction patterns for this muscle consistently showed a burst lasting for most of swing phase, as well as a second, short burst halfway through the stance phase (Table 2; Fig. 3C). Extensor digitorum communis. This muscle originates from the dorsal surface of the distal femur, and spreads widely to insert on the terminal phalanx of digit 1 as well as the first four metatarsals, crossing the knee and ankle joints (Walker, 1973; Fig. 2C). Although one turtle (TS34) consistently showed a single burst of activity during swing phase, two other turtles (TS35, TS75) showed much longer, single bursts of activity that began during late swing phase but continued through most of stance phase (Fig. 3C). DISCUSSION Our EMG data from the distal hind limb muscles of slider turtles allow anatomically based hypotheses of the functional roles of these muscles (Walker, 1973; Butcher and Blob, 2008) to be assessed. This refined understanding of activation patterns for these muscles provides insight into how the limb skeleton of turtles is loaded during terrestrial locomotion, and indicates substantial variation in the motor patterns of some muscles with potential functional significance (e.g., German et al., 2008). Functional Roles of Turtle Shank Muscles and Their Implications for Models of Femoral Loading All four proposed ankle extensor muscles, whether uni- or biarticular, were active during most of the stance phase of the walking locomotor cycle across most of the turtles we examined (Fig. 3) as predicted by their anatomical positions (Walker, 1973; Butcher and Blob, 2008). The substantial ankle extension that occurs during stance phase (Fig. 3) suggests that the activity of these muscles does, as predicted, help to counter the flexor moment imposed at the ankle by the GRF. However, the roles of all four of these muscles appear to be more complicated than simply facilitating ankle extension. All four ankle extensors showed EMG bursts in at least some individuals that occurred primarily during swing phase (Fig. Journal of Morphology

3), though such bursts were less common for pronator profundus than they were for the other three muscles. Such activity could help to modulate or control the flexion of the ankle as the hind limb is swung into position for replacement of the foot on the ground. Ankle flexion rarely exceeds 308 at foot down (Fig. 3B), and contraction of ankle extensors may help to set this limit. During both stance and swing, however, the largely coinciding activity of all four ankle extensors in most individual turtles makes further distinction of functional roles among these muscles difficult to discern from available kinematic data. The activity patterns of putative ankle flexor muscles in turtles were also more complicated than initial predictions based on their anatomical position (Walker, 1973). Although both tibialis anterior and extensor digitorum communis showed EMG bursts during the ankle flexion that occurs during swing phase, both also showed major periods of activity during stance phase, while the ankle was extending (Fig. 3). Such activity could serve to stabilize the ankle joint while the foot is on the ground and limit hyperextension during stance. Because of its distal insertion on the digits, activity of extensor digitorum communis might also help to maintain engagement of the claws on the substrate, or prevent inversion of the foot. The complexities of the activity patterns for the six distal hind limb muscles we evaluated have substantial implications for our understanding of skeletal loading in turtles during locomotion. Our previous studies found that safety factors (in bending) calculated for turtle femora were much higher when based on force platform analyses (Butcher and Blob, 2008) than they were when based on in vivo measurements of femoral strains (Butcher et al., 2008). This difference suggested that force platform analyses might underestimate loading magnitudes in turtles, potentially through incorrect assumptions about the activity of muscles that might contribute to femoral stress. Electromyography data from this study allowed us to test two specific assumptions. First, the force platform model (Butcher and Blob, 2008) assumed that all four putative ankle extensors were active during stance, but if only the ankle extensors that cross the knee were, in fact, active during stance (i.e., gastrocnemius lateralis and flexor digitorum longus), then actual femoral stresses might be higher than originally calculated. This is because if only biarticular ankle extensors were active, the entire force exerted to counter the flexor moment of the GRF at the ankle would be transferred across the knee, in response to which knee extensor muscles that span the femur and subject it to stress would have to exert elevated forces. However, our EMG data showed that all four putative ankle extensors were active during stance, corroborating this aspect of our

HIND LIMB EMGS IN TURTLE WALKING

model of muscle activity (Butcher and Blob, 2008) and ruling out this assumption as a potential contributor to differences in force platform and strain-based estimates of femoral safety factors in turtles. Second, to prevent unknown muscle forces from occurring on both sides of equilibrium equations and to resolve the muscular contributions to limb bone loads (Biewener, 1983; Blob and Biewener, 2001; Butcher and Blob, 2008), our published model of femoral loading in turtles also assumed inactivity of muscles on the flexor surface of the shank during stance (Butcher and Blob, 2008). However, stance phase activity by ankle flexors could require compensation by ankle extensors, causing them to exert additional force that could be transferred across more proximal joints and elevate stresses on the femur (Blob and Biewener, 2001; Reilly and Blob, 2003). Our EMG data showed that both of the ankle flexors we examined, tibialis anterior and extensor digitorum communis, were frequently active during much of stance. Ankle extensor muscles thus likely must exert forces that are higher than those needed to counter the GRF alone, potentially elevating the forces that they transfer across the knee joint and that, ultimately, are countered by additional force from knee extensor muscles that would increase stresses on the femur above our previous estimates (Butcher and Blob, 2008). Quantifying the magnitude of this effect on bone stress estimates would require direct measurements of the forces that tibialis anterior and extensor digitorum communis contribute to the flexor moment of the ankle, data that are currently unavailable. Nonetheless, it seems likely that coactivity by antagonist ankle flexors may contribute to differences between the evaluations of safety factors in turtle limbs based on force platform versus in vivo strain studies. Although evaluations of the mechanics underlying femoral loading patterns in turtles that were determined from force platform data (Butcher and Blob, 2008) are not called into question by the EMG data reported here, the magnitudes of bending stresses previously reported may be underestimated, which could elevate safety factor estimates derived from this method of analysis. Diversity and Consistency of Distal Hind Limb Motor Patterns in Turtles We observed considerable diversity in the motor patterns of several muscles across the individuals we examined. Although all of the putative ankle extensors showed major stance phase bursts across sampled individuals, there were some differences in both the timing and number of activity bursts for each of these muscles across many of the turtles in our study. To some degree, such variability among individuals may reflect differences in the

1535

precise placement of electrodes within specific muscles. Functional heterogeneity among different portions of single muscles has been documented within limb muscles (including medial gastrocnemius) of several vertebrates (Ahn et al. 2003; Higham et al., 2008). Thus, individuals that exhibited differences in the contraction patterns of a muscle (particularly those showing dramatic differences, such as the two divergent patterns among individuals for extensor digitorum longus) may, in fact, have had functionally divergent portions of the muscles sampled. Although some variation in motor patterns across individuals was present in all the muscles we were able to sample from multiple turtles, patterns for flexor digitorum longus appeared much more consistent (i.e., stereotyped: Wainwright et al., 2008) across the seven individuals we sampled than patterns for the other muscles, which were generally evaluated from fewer individuals. Although, as noted above, some of the variation we observed in motor patterns for several muscles (particularly across individuals) may reflect differences in electrode placement, it is also likely that much step-to-step variation reflects adjustments to minor differences in foot placement or other sensory input (Zernicke and Smith, 1996). Although variation in motor output for a specific behavior is distinct from the flexibility of motor output across behaviors, the two are often correlated (Wainwright et al., 2008). With such limited variation in flexor digitorum longus activity, is it possible that turtles might accommodate changes in locomotor conditions through adjustments of motor output for muscles primarily affecting more proximal joints (potentially as far as the ankle), rather than muscles such as flexor digitorum longus that insert more distally and also control the digits? Such a pattern would seem counterintuitive given the proximity of the digits to the substrate that might be the primary source of sensory stimuli, and kinetic data from other taxa (e.g., running birds: Daley et al., 2007) indicating a proximal-todistal gradient of joint neuromechanical control, with more distal muscles typically showing greater sensitivity to changes in locomotor conditions. Such a proximal-to-distal gradient may be present to some extent in turtles, as data from two species indicate little variation in motor output for a uniarticular hip flexor (puboischiofemoralis internus) across locomotor behaviors, in contrast to variation seen in muscles controlling the knee (Blob et al., 2008) and the variation seen in several muscles affecting the ankle joint in this study (Fig. 3). Extending predictions from birds to turtles may be complicated, however, for muscles affecting the digits, particularly given differences in muscle architecture between these taxa (e.g., the lack of long tendons in turtle muscles: Walker, 1973). Nonetheless, the variation we did observe in the Journal of Morphology

1536

H. L. SCHOENFUSS ET AL.

motor patterns of distal hind limb muscles in turtles during walking suggests that these muscles could hold promise for future studies of motor flexibility across behaviors (e.g., Gillis and Blob, 2001; Blob et al., 2008; Rivera and Blob, 2010). ACKNOWLEDGMENTS The authors thank Tim Higham for comments during early analyses of this study, and Gabriel Rivera, Mike Butcher, Takashi Maie, Megan Wilson, and Nora Espinoza for their feedback and contributions during experiments. They also thank Warren Walker for permission to use modified versions of his published anatomical illustrations in our Figure 2 and two anonymous reviewers for their suggestions. LITERATURE CITED Ahn AN, Monti RJ, Biewener AA. 2003. In vivo and in vitro heterogeneity of segment length changes in the semimembranosus muscle of the toad. J Physiol 549:877–888. Alexander RM. 1974. The mechanics of a dog jumping, Canis familiaris. J Zool Lond 173:549–573. Biewener AA. 1983. Locomotory stresses in the limb bones of two small mammals: The ground squirrel and chipmunk. J Exp Biol 103:131–154. Biewener AA. 1989. Scaling body support in mammals: Limb posture and muscle mechanics. Science 245:45–48. Biewener AA. 1990. Biomechanics of mammalian terrestrial locomotion. Science 250:1097–1103. Biewener AA, Full RJ. 1992. Force platform and kinematic analysis. In: Biewener AA, editor. Biomechanics Structures and Systems: A Practical Approach. New York: Oxford University Press. pp45–73. Biewener AA, Gillis GB. 1999. Dynamics of muscle function during locomotion: Accommodating variable conditions. J Exp Biol 202:3387–3396. Blob RW, Biewener AA. 1999. In vivo locomotor strain in the hindlimb bones of Alligator mississippiensis and Iguana iguana: Implications for the evolution of limb bone safety factor and non-sprawling limb posture. J Exp Biol 202: 1023–1046. Blob RW, Biewener AA. 2001. Mechanics of limb bone loading during terrestrial locomotion in the green iguana (Iguana iguana) and American alligator (Alligator mississippiensis). J Exp Biol 204:1099–1122. Blob RW, Rivera ARV, Westneat MW. 2008. Hindlimb function in turtle locomotion: Limb movements and muscular activation across taxa, environment, and ontogeny. In: Wyneken J, Godfrey MH, Bels V, editors. Biology of Turtles. Boca Raton: CRC Press. pp 139–162. Bodie JR, Semlitsch RD. 2000. Spatial and temporal use of floodplain habitats by lentic and lotic species of aquatic turtles. Oecologia 122:138–146.

Journal of Morphology

Butcher MT, Blob RW. 2008. Mechanics of limb bone loading during terrestrial locomotion in river cooter turtles (Pseudemys concinna). J Exp Biol 211:1187–1202. Butcher MT, Espinoza NR, Cirilo SR, Blob RW. 2008. In vivo strains in the femur of river cooter turtles (Pseudemys concinna) during terrestrial locomotion: Tests of force platform models of loading mechanics. J Exp Biol 211:2397–2407. Daley MA, Felix G, Biewener AA. 2007. Running stability is enhanced by a proximo-distal gradient in joint neuromechanical control. J Exp Biol 210:383–394. German RZ, Crompton AW, Thexton AJ. 2008. Variation in EMG activity: A hierarchical approach. Int Comp Biol 48: 283–293. Gibbons JW. 1970. Terrestrial activity and the population dynamics of aquatic turtles. Am Mid Nat 83:404–414. Gibbons JW, Greene JL, Congdon JD. 1990. Temporal and spatial movement patterns of sliders and other turtles. In: Gibbons JW, editor. Life History and Ecology of the Slider Turtle. Washington, DC: Smithsonian Institution Press. pp 201–215. Gillis GB, Blob RW. 2001. How muscles accommodate movement in different physical environments: Aquatic vs. terrestrial locomotion in vertebrates. Comp Biochem Physiol A 131:61–75. Higham TE, Biewener AA, Wakeling JM. 2008. Functional diversification within and between muscle synergists during locomotion. Biol Lett 4:41–44. Kargo WJ, Rome LC. 2002. Functional morphology of proximal hindlimb muscles in the frog Rana pipiens. J Exp Biol 205:1987–2004. Loeb GE, Gans C. 1986. Electromyography for experimentalists. Chicago: University of Chicago Press. Macpherson JM. 1991. How flexible are muscle synergies? In: Humphrey DR, Freund HJ, editors. Motor control: Concepts and issues. Chichester, UK: Wiley. pp 33–47. Reilly SM, Blob RW. 2003. Motor control of locomotor hindlimb posture in the American alligator (Alligator mississippiensis). J Exp Biol 206:4327–4340. Rivera ARV, Blob RW. 2010. Forelimb kinematics and motor patterns of the slider turtle (Trachemys scripta) during swimming and walking: Shared and novel strategies for meeting locomotor demands of water and land. J Exp Biol doi:10.1242/ jeb.047167. Roy RR, Hutchinson DL, Pierotti DJ, Hodgson JA, Edgerton VR. 1991. EMG patterns of rat ankle extensors and flexors during treadmill locomotion and swimming. J Appl Physiol 70: 2522–2529. Wainwright PC, Mehta RS, Higham TE. 2008. Stereotypy, flexibility, and coordination: Key concepts in behavioral functional morphology. J Exp Biol 211:3523–3528. Walker JA. 1998. Estimating velocities and accelerations of animal locomotion: A simulation experiment comparing numerical differentiation algorithms. J Exp Biol 201:981–995. Walker WF Jr. 1973. Locomotor apparatus of Testudines. In: Gans C, Parsons TS, editors. Biology of the Reptilia, Vol. 4. Morphology D. London: Academic Press. pp 1–100. Zernicke RF, Smith JL. 1996. Biomechanical insights into neural control of movement. In: Rowling LB, Shepherd JT, editors. Handbook of Physiology Section 12: Exercise: Regulation and Integration of Multiple Systems. New York: American Physiological Society. pp 293–330.