Here, we used an obstacle treadmill experiment to investigate the neuromuscular control of locomotion in uneven terrain. We measured in vivo function of two distal muscles of the guinea fowl, lateral gastrocnemius (LG) and digital flexor-IV (DF), during level running, and two uneven terrains, with 5 and 7 cm obstacles. Uneven terrain required one step onto an obstacle every four to five strides. We compared both perturbed and unperturbed strides in uneven terrain to level terrain. When the bird stepped onto an obstacle, the leg became crouched, both muscles acted at longer lengths and produced greater work, and body height increased. Muscle activation increased on obstacle strides in the LG, but not the DF, suggesting a greater reflex contribution to LG. In unperturbed strides in uneven terrain, swing pre-activation of DF increased by 5 per cent compared with level terrain, suggesting feed-forward tuning of leg impedance. Across conditions, the neuromechanical factors in work output differed between the two muscles, probably due to differences in muscle–tendon architecture. LG work depended primarily on fascicle length, whereas DF work depended on both length and velocity during loading. These distal muscles appear to play a critical role in stability by rapidly sensing and responding to altered leg–ground interaction.
The muscles of animal legs must function to allow versatile and stable movement through a wide range of terrain conditions. Yet, little is known about the interplay of mechanics, muscle dynamics and neural control in the context of unsteady tasks, such as manoeuvring and stabilization . Animal movement requires a complex integration of central, peripheral and physical control mechanisms [2–4]. Owing to the complexity of the neuromuscular system, multiple possible combinations of muscle activity and sensory feedback can achieve any given target task [5–7]. However, different solutions for achieving a motor task may result in different characteristics such as stability, required total muscle activity, and fatigue rate (e.g. [7,8]). Animals probably select among motor control strategies in a context-dependent manner based on numerous criteria .
The aim of this study is to investigate the neuromuscular control strategies used by animals to maintain stability in uneven terrain. We are interested in both the immediate response to a perturbation, involving stride-to-stride adjustments, as well as shifts in the motor control strategy depending on terrain environment. In terrain known to be uneven or unpredictable, animals might select a different neuromuscular control strategy than in uniform terrain. For example, when cats are exposed to repeated perturbations to the paw during swing, they exhibit long-term adjustments, ‘high-stepping’ to avoid stumbling . Such a strategy has clear advantages in uneven terrain, but might increase energy cost. We expect animals to adjust motor control on both short and longer time scales to achieve robustly stable and economic movement through varying environmental conditions.
An important focus of our investigation is to explore the relationship between muscle–tendon architecture and neuromuscular control. Most in vivo studies of muscle function during locomotion have focused on steady locomotion over level ground or a constant slope. To understand the morphological and biomechanical factors that influence neural control, it is critical to investigate muscle dynamics over a broader range of animal behaviour. Muscles that play similar roles in steady movement might play distinct roles during non-steady tasks. Here, we examine this issue by measuring the in vivo force–length–activation dynamics of two distal hindlimb extensors in the guinea fowl: the lateral gastrocnemius (LG) and the digital flexor-IV (DF). Both these muscles have pinnate, short-fibred architecture with a relatively long free tendon. The LG and DF function similarly during steady level locomotion to provide economic weight support during stance and facilitate tendon elastic energy cycling [11–13].
Nonetheless, there are important architectural differences between the LG and DF. The LG acts across the knee and ankle, and has a relatively stiff tendon, whereas the DF crosses the knee, ankle and all distal joints, and has an exceptionally long tendon. The ratio of tendon length to muscle fascicle length is 10.8 for the DF, whereas it is 5.7 for the LG . The extreme architecture of DF may enhance force control over position control and facilitate economy, but limit its ability to actively control joint position and do external work against the environment [14,15]. In vivo studies suggest that the DF contributes more than LG to elastic energy cycling in steady running, but does not contribute to work for incline running . By contrast, the LG contributes moderately to elastic energy cycling, and also contributes to work on an incline [11–13].
Although muscle architecture influences the context in which a muscle does work, its mass determines total work capacity. Short-fibred, pinnate distal muscles do have a substantial mass and work capacity in most animals (e.g. [16,17]). The gastrocnemius and digital flexors of the guinea fowl make up approximately 30 per cent of hindlimb muscle mass . We do not yet understand the circumstances in which the distal muscles produce and absorb substantial energy, because muscle force–length dynamics have been measured for relatively few motor tasks . Distal muscles may contribute to work during non-steady tasks such as acceleration, jumping, manoeuvring and stabilization. Recent work suggests that distal muscles may absorb or produce energy to help stabilize the leg in the face of sudden terrain perturbations [18,19].
On the basis of these observations, we hypothesized that the specialized architecture of distal leg muscles reflects a proximo-distal gradient in joint neuromechanical function [18,19]. We suggested that the muscles at proximal joints control leg cycling and contribute the majority of work for tasks such as steady incline running and jumping, whereas distal muscles rapidly adjust force and work during non-steady tasks such as manoeuvring, maintaining stability and avoiding injury in uneven terrain. A recent study of in vivo dynamics of the LG during running over an unexpected drop in terrain drop supports this idea . The LG exhibits rapid changes in work output depending on the posture of the leg when it contacts the ground. This results in a context-dependent stabilizing response to unexpected terrain perturbations.
Here, we investigate neuromuscular control of the LG and DF during locomotion over uneven terrain with obstacles repeating every four to five strides. The bird must take a single step onto each obstacle before returning to the original ground level. To evaluate changes in neuromuscular control among terrain conditions, we ask the following four questions:
— How do muscle force–length–activation dynamics during perturbed strides (on obstacle) compare with level terrain?
— How do unperturbed strides in uneven terrain (between obstacles) compare with level terrain?
— Do the LG and DF respond similarly in work output in uneven terrain?
— Do similar neuromechanical factors underlie changes in work of both muscles?
Both unperturbed and perturbed strides in uneven terrain may differ from those in uniform terrain, owing to context-dependent adjustment of motor control. Additionally, the architectural differences between LG and DF may lead to differing responses to perturbations. We expect the DF to be especially sensitive to terrain variation due to its extreme architecture and action at more distal joints. This may result in larger intrinsic mechanical effects on contractile performance, when compared with the LG.
(a) Animals and training
We obtained six adult guinea fowl (Numida meleagris), 1.77 ± 0.26 kg body mass (mean ± s.e.m., n = 6) from a local breeder near Bedford, MA, USA. Animals were trained to run on a level motorized treadmill (Woodway, Waukesha, WI, USA) at speeds of 1.7–2.0 m s−1. Training sessions were 20–30 min in duration, with breaks for 1–3 min as needed. Animals were trained 3–4 days per week for three weeks. Experiments were undertaken at the Concord Field Station of Harvard University. All procedures were approved by the Harvard Institutional Animal Care and Use Committee.
(b) Surgical procedures
We followed similar surgical procedures as described previously . Transducers were implanted into the lateral head of the gastrocnemius (LG) and the digital flexor to the lateral toe (DF). The birds were anaesthetized using isoflurane delivered through a mask. The surgical field was plucked of feathers and sterilized with antiseptic solution (Prepodyne, West Argo, Kansas City, MO, USA). Transducers were connected to a micro-connector placed on the bird's back (GM-6, Microtech Inc, Boothwyn, PA, USA). We passed the transducer leads subcutaneously from a 1–2 cm incision over the synsacrum to a second 4–5 cm incision over the lateral right shank. E-type tendon buckle force transducers were implanted on the Achilles tendon and DF tendon. Sonomicrometry crystals (0.7 mm for DF, 1.0 mm for LG; Sonometrics Inc., London, Canada) were implanted along the fascicle axis in the middle one-third of each muscle. Crystals were placed in small openings created using fine forceps, approximately 3–4 mm deep and 10 mm apart. We verified signal quality using an oscilloscope, and then secured them by closing the overlying connective tissue with 5-0 silk suture.
Next to each sonomicrometry crystal pair, we implanted fine-wire, twisted, silver bipolar electromyography (EMG) hook electrodes with 0.1 mm diameter, 0.5–1.0 mm bared tips, 5–8 mm spacing (California Fine Wire, Inc., Grover Beach, USA). EMG electrodes were placed using a 23 gauge hypodermic needle and secured to the muscle's fascia using 5-0 silk suture. Skin incisions were closed using 3-0 silk. The birds could walk within 2 h and ran the following day without lameness. The birds were given analgesia every 12 h and antibiotics every 24 h. Experimental recordings took place over the next 1–3 days. After the experiments, the guinea fowl were killed by an intravenous injection of sodium pentobarbital (100 mg kg−1) under deep isoflurane anaesthesia (4%, mask delivery).
Post-mortem, we dissected each muscle free from the surrounding tissues to make morphological measurements and confirm placement of transducers. Crystal alignment relative to the fascicle axis (α) was within ±3°. Finally, the tendon force buckles were calibrated in situ .
(c) Transducer recording
The micro-connector on the bird's back was connected, via a lightweight shielded cable (Cooner Wire, Chatsworth, NJ, USA), to a sonomicrometry amplifier (120.2, Triton Technology Inc., San Diego, CA, USA), a strain gauge bridge amplifier (2120, Vishay Micromeasurements, Raleigh, NC, USA), and EMG amplifiers (P-511, Grass, West Warwick, RI, USA). EMG signals were amplified 1000× and filtered (10–10 kHz bandpass) before digital sampling. Signals were sampled by an A/D converter (Axon Instruments, Union City, CA, USA) at 5 kHz.
Digital high-speed video was recorded in lateral view at 250 Hz (PhotronFastcam-X 1280 PCI; Photron USA Inc., San Diego, CA, USA). Kinematic points were marked on the synsacrum, hip, middle toe and lateral toe, and tracked using custom software in Matlab (v. 7, Mathworks, Inc.; Natick, MA, USA). We tracked the following kinematic events (i) midswing (MS), the time at which the swing-leg toe crossed the midline of the stance leg; (ii) toe down (TD); and (iii) toe off (TO). Successive MS events were used to cut the muscle data into strides. For a subset of data, we manually digitized the marker positions at the kinematic events. We calculated effective leg length (Lleg), leg angle (LA), hip height (H) (figure 1a), stride period and stance period.
(e) Experimental protocol
The guinea fowl ran at 1.7 m s−1 in three conditions: (i) level terrain, (ii) terrain with repeated 5 cm high obstacles, and (iii) terrain with repeated 7 cm high obstacles. In uneven terrain, the bird encountered an obstacle every four to five strides. We constructed the obstacles using Styrofoam covered with cardboard and black neoprene, to create a light, stiff surface that matched the treadmill belt. The treadmill belt was composed of rubber-coated steel slats (0.56 × 0.07 m) with clearance around the entire surface for obstacles. The effective running surface measured 0.56 × 1.73 m. We matched the length and width of individual obstacles to the belt slats, and attached them with industrial strength Velcro. Seven obstacles were placed on sequential slats to form a total obstacle surface that was 0.49 m long, approximately one stride length.
(f) Data processing
We filtered the EMG signals using a sixth order, zero-lag high-pass Butterworth filter (70 Hz cutoff), then calculated the intensity over time using wavelet analysis as described previously . Within individuals, EMG intensity was normalized based on the mean total intensity per stride in level trials. We calculated fractional fascicle length from the sonomicrometry data, following methods described by others , using length during quiet standing as the reference length (Lo). Fascicle length was differentiated to obtain fascicle velocity (V, in lengths per second, L s−1). To calculate muscle power, velocity (converted to ms−1) was multiplied by tendon force (in Newtons). Muscle power was integrated over time for each stride to calculate work (in Joules, and divided by muscle mass to obtain mass-specific values, J kg−1). We then measured variables at a number of time points to evaluate muscle force, length and activation dynamics, see the glossary for a description.
For each trial, we analysed a 10–12 s sequence in which the bird maintained a constant speed, resulting in approximately 30 strides per trial. Stride cycles were put into the following categories: control (C) for level running, and in uneven terrain the stride prior to (s−1), the stride on (s0), the first stride following (s+1), and the second to third strides following (s+2) an obstacle. We grouped strides two to three together because the number of strides between obstacles varied.
All statistics were calculated using the statistics toolbox in Matlab (Mathworks, Inc.; Natick, MA, USA). To test for differences in muscle and kinematic variables, we used mixed-model ANOVA with stride category as a fixed factor and individual as a random factor. We used post hoc t-tests with Bonferroni correction to compare pairs of stride categories.
We used multiple linear regression analysis to evaluate the contributions of fascicle length, phase and activation factors on total force impulse (Jtot) and net work produced (Wnet) by each muscle. We included , , , , Phase, Esw and Etot as factors in the model (see glossary), with backward stepping to minimize multi-colinearity and find the minimal model that best fit the data.
(a) Changes in kinematics and leg posture during perturbed strides in uneven terrain
When guinea fowl negotiated an obstacle, the leg contacted the obstacle during late swing, initiating an early transition to stance compared with level terrain (figures 1 and 2). The birds exhibited a more ‘crouched’ leg posture on the obstacle, with a shallower leg angle and shorter hip height (figure 1b and table 1). Hip height increased during stance by 3.2 cm on both 5 and 7 cm obstacles, suggesting net positive work on the body (figure 1b and table 1).
(b) Muscle force–length–activation dynamics during perturbed strides in uneven terrain
Significant changes occurred in muscle force–length dynamics during obstacle strides in relation to altered leg loading and posture. As the foot contacted the obstacle, fascicle length of both LG and DF immediately diverged from level means (figures 2 and 3), shifting towards longer lengths. Force development began early in both muscles (figure 3) and the rise time to peak force (Trise) took longer (table 2). LG peak force (Fpk) increased by 32 and 39 per cent in 5 and 7 cm terrain, respectively; by contrast, DF Fpk decreased by 20 and 26 per cent (table 2). EMG intensity (Etot) of LG increased by over 80 per cent in obstacle strides in 5 cm and 7 cm terrain; by contrast, there was no significant change in Etot of the DF (figure 5 and table 2).
The average shifts in length and velocity on obstacle strides were similar for the two muscles (‘s0’, figure 4). Fascicle length remained longer throughout force development on obstacles (figure 3). Velocity shifted towards increased stretch during the initial loading phase from TD to T50 (), and towards greater shortening (or less stretch) from T50 to Fpk () (‘s0’, figure 4).
Overall, force–length dynamics in obstacle strides led to significantly greater work output compared with level control (figure 5 and table 2). LG work (Wnet) on ‘s0’ increased by 5.26 and 6.05 J kg−1 for 5 and 7 cm terrain, respectively. The DF also showed increased Wnet, averaging 9.16 and 13.79 J kg−1 in 5 and 7 cm terrain; however, the difference was significant only for 7 cm terrain, owing to high variation in DF work output (figure 5 and table 2).
(c) Muscle force–length–activation dynamics during ‘unperturbed’ strides in uneven terrain
The largest shifts in force, length and activation occurred in obstacle strides (s0); however, there were also a number of significant shifts across non-obstacle strides. In the stride immediately following the obstacle (s+1), force–length dynamics were similar to level terrain (table 2). A greater number of differences occurred in strides immediately preceding the obstacle (s−1) and during unperturbed strides between obstacles (s+2). Across non-obstacle strides, swing phase activation (Esw) increased by about 5 per cent, although this difference was significant only for DF (table 2). The DF operated at 5 per cent shorter lengths in 5 cm terrain (table 2). LG velocity shifted towards stretch, by about 1.5 L s−1 from TD to T50, and 0.7 L s−1 from T50 to Fpk (figure 4). LG Wnet decreased by 1 J kg−1 in 5 cm terrain (figure 5). The total force impulse in unperturbed strides in uneven terrain did not differ significantly from level control (table 2).
(d) Contractile properties underlying the changes in muscle work in uneven terrain
The force–length dynamics of the LG and DF varied substantially in uneven terrain, depending on how the leg interacted with the ground. Muscle length and force development rapidly deviated from level values as soon as the foot contacted the obstacle (figures 2 and 3). The DF exhibited more time-varying force–length dynamics within a stride, greater stride-to-stride variation (figure 3) and a greater range of mass-specific work output (figure 6), compared with the LG.
Multiple regression analysis suggests that a number of neuromechanical factors contributed to variation in LG and DF mechanical output. The largest factor in the regression model for total force impulse (Jtot) was length at 50 per cent peak force (), for both LG and DF. explained 60 per cent of the variation in Jtot for the LG, and 30 per cent for the DF (table 3). For the LG, and Etot also contributed to the variation in Jtot, explaining 7 and 9 per cent of the variance, respectively. For the DF, these factors did not contribute to the model for Jtot. Phase (between force and length) and swing Esw were significant factors for the DF, but explained only 1 and 5 per cent of the variance, respectively. The results suggest that 67 per cent of LG variance and 31 per cent of DF variance in Jtot could be explained by intrinsic mechanical factors relating to length and velocity. However, the total variance explained (r2) by the model for Jtot was lower for the DF (r2 = 0.76 for LG and 0.36 for DF), suggesting that a linear model may not be adequate for this muscle.
The neuromechanical factors in Wnet differed more substantially between muscles. was the largest single factor in LG Wnet, explaining 64 per cent of variance, whereas velocity at peak force () was the largest factor in DF Wnet, explaining 31 per cent of variance. Additional factors in LG work were , Esw and Etot, explaining 4, 1 and 7 per cent of the variance, respectively. For DF work, and Phase (between force and length) were additional significant factors, each explaining 16 per cent of the variance. Activation factors (Esw, Etot) contributed significantly to variation in Wnet of the LG, but not the DF.
(e) Relationship between leg posture and force–length dynamics across terrain conditions
The fascicle length of both muscles inversely correlated with leg posture (figure 7a). More crouched postures at TD led to greater extremes in fascicle length during force development. There was also an inverse relationship between leg posture and the change in hip height during stance (ΔH; figure 7b). Changes in hip height provide an estimate of the potential energy change of the body, so the positive ΔH associated with crouched leg posture suggests net positive work on the body. The significant positive correlation between and Wnet, observed for both muscles (table 3 and figure 6), suggests that both the LG and DF contribute to an increase in potential energy of the body when the leg contacts the ground in a crouched posture.
Recent studies investigating leg and muscle mechanics in response to an unexpected terrain drop suggest that distal leg muscles play an important role in stability and injury prevention during running over uneven terrain [18,19]. The response depends on the interplay of leg posture, leg loading and neuromuscular control . However, the previous studies focused on a single unexpected drop perturbation that may not reflect natural terrain conditions. Here, we investigated context-dependent changes in neuromuscular control in uneven terrain. We measured in vivo muscle function of two distal muscles, LG and DF, during running over terrain with obstacles every four to five strides. This terrain was not particularly challenging— the birds maintained average position on the treadmill belt and recovered from each obstacle within one stride following the perturbation.
(a) How do neuromuscular dynamics during perturbed strides (on obstacle) compare with level terrain?
The bird's dynamic response to the perturbed obstacle stride (s0) in uneven terrain is similar to that observed during an unexpected drop perturbation . Leg posture was more crouched and stance duration longer in obstacle strides, but stride duration remained similar (table 1). Muscle length was similar to level running until the toe contacted the obstacle (figure 3). The kinematic differences may have resulted from the intrinsic mechanical interaction of the swing leg with the obstacle, resulting in an early transition from swing to stance. Drop perturbations result in similar but inverse dynamics—delayed ground contact leads to a more extended leg posture and shorter stance duration .
In both obstacle and drop perturbation conditions, there is a strong relationship between the work produced by LG and leg posture at the time of ground contact (figure 8) . The drop perturbation experiment elicited greater extremes in LG work and leg posture. This is probably owing to the larger size of the perturbation used: 8.5 cm , compared with 5 and 7 cm in the current study. However, additional factors may contribute, such as variation in feed-forward control and reflex modulation across conditions, because LG work in the unexpected drop diverges more from the fit to both datasets (figure 8). Nonetheless, the trends are remarkably similar. Like the LG, the DF muscle exhibits rapid changes in force–length dynamics and work output during obstacle perturbed strides. The average change in work of the DF was higher than that of the LG (figure 5), but DF showed greater stride-to-stride variability.
Neural factors appear to play a greater role in the response of the LG during perturbed obstacle strides, when compared with the DF. Total EMG intensity of the LG increased by over 80 per cent on the obstacle stride (s0), but DF activity was not significantly different from level running. In the unexpected drop experiment, there was no significant change in LG EMG activity within the perturbed stride . This suggests either that neuromuscular control of LG is adjusted in anticipation of the obstacle in uneven terrain, or that the reflex response to a sudden increase in load differs from the response to a sudden decrease in load. The transmission delay for the stretch reflex of the gastrocnemius is approximately 6 ms, which is less than 5 per cent of stance duration , so monosynaptic (Ia) stretch reflexes could explain the increase in LG muscle activity. Additionally, heterogenic length feedback between agonist muscles acts in approximately the same time scale . Feedback from the DF as the toes contact the obstacle could excite LG activity through heterogenic pathways. Head pitch is another factor that could facilitate increased LG muscle activity in the obstacle stride. In cats, pitching the head downward elicits increased muscle activity analogous to uphill walking . If guinea fowl look downward in anticipation of obstacles, this could elicit increased LG activity similar to that observed on steady inclines [11,25].
(b) How do neuromuscular dynamics during unperturbed strides (between obstacles) compare with level terrain?
The kinematics in unperturbed strides remained similar to level terrain, suggesting that the overall target movement pattern is not adjusted in uneven terrain. There were no significant differences in kinematic variables in the unperturbed strides (table 1). Interestingly, the final leg angle before toe off (LATO) did not change significantly in any stride category (table 1), suggesting consistent control of the stance–swing transition. Cat studies suggest that the stance–swing and swing–stance transitions are controlled through a combination of two sensory signals: loading of ankle extensors, and position feedback from hip extensors [26,27]. The consistency of LATO across conditions may result from physical unloading of the leg beyond a particular angle, providing robust control of stance–swing transition across varied terrain conditions.
Although the overall movement pattern was not adjusted in the unperturbed strides in uneven terrain, muscle dynamics suggest context-dependent tuning of neuromuscular control. Swing phase activation increased by 5 per cent across unperturbed strides, which may facilitate a rapid response when the leg contacts an obstacle. DF operated at 5 per cent shorter length in uneven terrain, and LG shifted towards stretch during force development (table 2). These findings suggest that animals might tune force–length dynamics through small shifts in the timing of muscle activation. This is consistent with other recent studies that suggest context-dependent tuning of control, even when similar overall gait and speed are maintained. For example, humans and other animals adjust leg compliance through posture and co-activation , foot placement to avoid obstacles , and swing-leg kinematics to avoid stumbling . Yet, we still have a limited understanding of how animals select among control strategies while moving through complex terrain. This is because most of our knowledge of locomotion is based on controlled laboratory conditions, usually on steady, level locomotion on treadmills or uniform runways, or in ‘reduced’ preparations where much of the central nervous control has been removed or minimized [1,30]. Further studies are required that include both naturalistic terrains and neuromechanical simulations. Nonetheless, the results here suggest that animals use similar target movements across terrain conditions, with feed-forward adjustments and reflex feedback acting to tune leg impedance around the same global strategy.
(c) Neuromechanical factors contributing to the changes in muscle dynamics of lateral gastrocnemius and digital flexor-IV
The LG and DF differ in muscle–tendon morphology: although both have pinnate, short-fibred architecture, the DF has an exceptionally long tendon that crosses all distal joints. We expected the DF to be more sensitive to terrain perturbations due to its extreme architecture. Although the average fascicle length and velocity changes were similar for the two muscles (figure 4), the DF produced a greater range of work over the conditions measured (figure 6). This is consistent with results on steady level and incline running , suggesting that the stretch-shorten cycle of the DF results in greater sensitivity in work output to variation in length, velocity and phase. The DF has a weaker relationship between fascicle length and work than the LG, but greater sensitivity to velocity (table 3). The LG acts as a length-dependent actuator, generating positive work when the leg begins stance in a more crouched posture. By contrast, the DF acts as both a length- and velocity-dependent actuator. Work done by the DF depends on both leg posture and how rapidly the leg is loaded.
Although there are a number of adjustments to neural control in uneven terrain, it appears that intrinsic mechanical factors play a large part in the response of both LG and DF. The delay between activation and force development is approximately 30 ms in level running , suggesting that intrinsic mechanical factors lead to the initial rapid increase in force upon obstacle contact. Length and velocity were the largest factors in total force impulse and work of both muscles (table 3). Furthermore, DF exhibited an increase in mean work on obstacle steps, despite similar total EMG intensity (figure 5). Changes in fascicle length can immediately alter muscle contraction because of force–length and force–velocity properties of muscle [31,32]. History-dependent factors, such as stretch force enhancement, might also contribute . The small shifts in swing pre-activation may tune these intrinsic mechanical effects by adjusting muscle length and velocity during the loading phase of stance.
(d) A tradeoff in limb design for economy versus stability?
We propose that the diversity in distal leg muscle architecture among animals reflects tradeoffs among economy versus stability, injury avoidance and agility. Specialist hoppers and runners, such as wallabies and horses, have distal architecture that facilitates economy through short, pinnate fibre arrangement and long tendons [34,35]. However, wallaby distal muscles do not increase work output on an incline , and horse digital flexors may be limited to a high-frequency damping function . Furthermore, horse distal tendons are prone to injury and heat damage [38,39]. In contrast, the ankle extensors of generalists such as ducks, guinea fowl and turkeys, store relatively less elastic energy, but probably have higher safety factors and can control work against the environment for other tasks, such as swimming (ducks) and incline running (guinea fowl, turkeys) [11–13,40]. This and other recent studies suggest that distal extensor muscles also play a critical role in stability by rapidly adjusting work output in response to terrain perturbations [18,19,21]. Consequently, distal muscles might have greater mass and work capacity in animals that regularly negotiate uneven terrain. The guinea fowl is a capable runner, but also relatively small. Small animals may live in inherently ‘rough’ environments, frequently encountering large terrain changes relative to leg length. Thus, guinea fowl distal hindlimb morphology may reflect optimization towards robust stability in uneven terrain.
- lateral gastrocnemius muscle
- digital flexor muscle head to the lateral toe (digital flexor-IV)
- toe down (begin stance)
- toe off (end stance)
- midswing point in leg kinematics
- hip height (usually relative to hip height at TD in level terrain HTD,C)
- Effective leg length
- Leg angle
- peak muscle–tendon force
- time period from TD to Fpk
- time period of force decay from Fpk to the initial value
- time point of 50% peak force
- fascicle fractional length (L) and velocity (V) at T50
- L and V at Fpk
- mean V from TD to T50
- mean V from T50 to Fpk
- phase between Fpk and maximum fascicle length (phase = (TFpk–TLpk)/Tstride)
- force impulse during swing
- force impulse over the stride
- work done during swing
- net work over the stride
- swing EMG intensity
- total EMG intensity.
We thank colleagues at the Concord Field Station of Harvard University for assistance and feedback, including Craig McGowan, Jim Usherwood, Polly McGuigan, Russ Main, Ed Yoo, Chris Richards and Pedro Ramirez. Supported by an HHMI Predoctoral Fellowship and an NSF Bioinformatics Postdoctoral Fellowship to M.A.D. (DBI-0630664), and an NIH grant (AR047679) to A.A.B.
One contribution of 15 to a Theme Issue ‘Integration of muscle function for producing and controlling movement’.
- This journal is © 2011 The Royal Society