Plasticity of the gastrocnemius elastic system in response to decreased work and power demand during growth

Taking advantage of storage and release of elastic strain energy can enhance performance that would otherwise be limited by the force–velocity constraints of muscle. The temporal decoupling of energy production from energy delivery permitted by elastic energy storage allows muscles and tendons to produce force effectively over a wider range of shortening or lengthening speeds. Muscles may generate forces during slow or isometric contractions while elastic recoil augments the rate of energy delivery or absorption during rapid movements (Patek et al., 2011; Roberts and Azizi, 2011). By making use of energy storage in the tendon ‘spring’, a muscle–tendon unit (MTU) can produce force more economically or with greater power than a muscle alone (Roberts and Azizi, 2011). Yet, several studies have identified important differences among spring–muscle combinations. Wilson, Lichwark and colleagues (Lichtwark and Wilson, 2008; Wilson et al., 2000) showed that the efficiency of an MTU during cyclic loading depends on the tuning of relative muscle and spring properties. For instance, muscle efficiency varies with both fascicle length and tendon stiffness, with the specific optimal efficiency values depending on gait conditions (Lichtwark and Wilson, 2008). Several researchers (Astley and Roberts, 2012, 2014; Galantis and Woledge, 2003; Ilton et al., 2018; Richards and Sawicki, 2012) have shown that the opposing inertial or drag forces acting on a motor–spring system also influence whether springs enhance performance. Adding further complexity, several studies (Robertson and Sawicki, 2014; Robertson et al., 2018; Sawicki et al., 2015) have found that the timing of neural activation of muscle must be tightly controlled to take advantage of in-series springs. Together, this body of work suggests that understanding the conditions in which spring systems enhance performance may require expanding our focus from the MTU to that of the broader ‘elastic system’ that includes the muscle (motor), the spring, the resistive forces and the neural control of the system. The optimal performance of an elastic system may require tuning of both morphology and neural control. This approach recognizes the integrated nature of the neuro-musculoskeletal system (Nishikawa et al., 2007).

The sensitivity of elastic system efficiency to the tuning of its components complicates inferences for how elastic systems systematically adapt to functional demand during maturation. For instance, do growing individuals who regularly perform functions that utilize elastic strain energy develop elastic systems with greater energy storage capacity? This is still unknown because most studies of MTU plasticity have focused on how individual components of elastic systems [neural control (Gillis and Biewener, 2001; Kao et al., 2010; Lutz and Rome, 1996), muscle (Du and Standen, 2017; Estes et al., 2017; Minamoto et al., 2015; Lieber et al., 2017) and tendon (Bohm et al., 2014; Bohm et al., 2019; Docking and Cook, 2019; Eliasson et al., 2007; Katugam et al., 2020; Kubo et al., 2007)] vary with task or training, and how those individual changes influence function of an MTU (Albracht and Arampatzis, 2006; Mayfield et al., 2016; Robertson and Sawicki, 2014; Rosario et al., 2016; Sawicki et al., 2015). Yet, the integrated nature of the elastic system suggests that functional consequences of plasticity are difficult to predict by analyzing elements in isolation (Albracht and Arampatzis, 2006; Ettema, 1996; Zajac, 1992). Therefore, the complex nature of the neuromuscular adaptation of elastic systems may require analysis at the system level rather than at the level of individual components.

Here, we present a study of the morphological plasticity of an elastic system. Specifically, we ask whether individuals that jump during maturation (an activity requiring elastic energy storage and return; Henry et al., 2005) develop elastic systems that are more capable of storing elastic strain energy at maturity than those of individuals restricted from jumping. We focus on the elastic system most involved in storage and release of elastic energy during jumping (Alexander, 1968; Arellano et al., 2019; Biewener et al., 2004; Farris et al., 2016; Walmsley et al., 1978), the gastrocnemius elastic system. We test this by altering the rearing conditions of two groups of guinea fowl (Numida meleagris) across the entire growth period, allowing one group to engage in jump-to-perch behavior and preventing all jumping in the other group. Guinea fowl are a particularly good species with which to study these questions because they generate peak powers three times greater than possible with muscle alone during jumping, producing forces over six times their body weight (Katugam et al., 2020), which suggests that energy storage and release are particularly important to their jump performance (Roberts and Hoppeler, 2016). We previously reported that restricted birds in another portion of this study showed detriments in jump performance at adulthood (Cox et al., 2020). In this paper, we aim to link the morphological and functional consequences of our intervention.

In analyzing the morphology data, we took both an individual-component and a systems-level approach to evaluate the plasticity of an elastic system during growth. At the component level, we probed whether our treatment resulted in systematic morphological differences in individual components of the gastrocnemius elastic system between groups. We sought to determine whether components of this elastic system plastically adapt to variations in functional demand during growth. At the systems level, we asked how plastic changes at the component level interact to influence the capacity for elastic energy storage. To do this, we developed subject-specific musculoskeletal models that incorporated experimentally measured morphological properties of each bird's elastic system. With each subject-specific model, we simulated a fully activated muscle contraction under various postures and quantified the resulting tendon strain energy stored. The purpose of this systems-level analysis was to evaluate the integrated effects of morphological variation on the limits of energy storage capacity of each bird.

The component- and systems-level analyses serve as a case study for understanding how a particular elastic system changes with functional demand. We also took advantage of the variation within and across groups to ask broader questions about the relationship between form and function in elastic systems. Specifically, we asked which combinations of naturally occurring morphological variation most influence the ability of an elastic system to store energy. Lastly, because the elastic system requires tuning of both morphology and neural control and energy storage capacity is only one piece of the puzzle, we tested whether jump performance is constrained by the limits of energy storage capacity. Specifically, we asked whether birds with lower energy storage capacity produce less work and power during jumping.

We hypothesized that components of an elastic system plastically adapt to variations in functional demand during maturation, resulting in greater energy storage capacity in birds that jump during growth. We predicted that energy storage capacity would increase linearly with muscle force-generating capacity and inversely with tendon stiffness (Biewener and Baudinette, 1995; Biewener and Roberts, 2000; Kubo et al., 1999). Finally, we predicted that differences in jump performance would positively correlate with an animal's ability to store elastic strain energy in the tendon of the gastrocnemius elastic system.

To study these questions, 1-day-old guinea fowl [Numida meleagris (Linnaeus 1758)] keets were obtained from a regional breeder (Guinea Farm, New Vienna, IA, USA). After a 2-week brooding period, the keets were pen reared through skeletal maturity (>6 months) in one of two conditions, as we previously described in detail (Cox et al., 2020). A control group (C; n=8) was housed in a large, circular pen (3.14 m²) that allowed ample room for locomotion and objects for jumping and perching. The restricted treatment group (R; n=7) was raised in a smaller pen (1 m² at maturity) with low mesh ceilings that prevented jumping. Food and water were available ad libitum (food intake did not differ between groups). Lights were programmed to be on a 12 h:12 h light:dark cycle. This protocol resulted in no changes in time spent walking or standing between groups, but drastically altered the average number of jumps per day. The control group jumped twice their body height, on average, 194 times a day, whereas the restricted birds were restricted from jumping entirely (Cox et al., 2019a). The experimental protocol was approved by the Institutional Animal Care and Use Committee at The Pennsylvania State University (ref. 46435)

As described previously (Cox et al., 2020), at skeletal maturity (between 29 and 31 weeks old) jump performance was measured by placing each bird in turn on 6×6 inch (15.24×15.24 cm) force plates (AMTI HE6x6; Watertown, MA, USA) enclosed in a tapered box and encouraging the birds to jump. Jump power was calculated from the instantaneous net vertical ground reaction and the vertical center of mass velocity. The horizontal component of GRF was ignored because the experimental setup constrained jumps to be nearly vertical. Velocity was obtained by integrating the center of mass acceleration, which was in turn found from the net ground reaction force and the body mass. We calculated jump work by integrating the instantaneous power with respect to time over the course of the jump. At the end of functional data collection, birds were euthanized (intravenous pentobarbital, >160 mg kg⁻¹).

The pelvic limb was separated from the upper body and the left and right legs were then split by sectioning the pelvis at the midline while avoiding muscle attachments. Right limbs were placed into neutral buffered formalin for fixation (10%) for at least 2 weeks, while left legs were fresh-frozen and kept at −20°C. Right limbs were positioned with joint angles approximating those at mid-swing during running [hip: 30 deg; knee: 80 deg; ankle: 125 deg; within ±2 deg (Rubenson and Marsh, 2009)]. Joint angles were confirmed for the fixed limbs using photographs made with a digital camera (Canon EOS550D; Surrey, UK) and analyzed with ImageJ (National Institutes of Health, Bethesda, MD, USA).

We made measurements of the lateral and medial heads of the gastrocnemius muscle (LG and MG, respectively), the muscle group of the MTU thought primarily responsible for storage and release of elastic strain energy during running and jumping (Arellano et al., 2019; Henry et al., 2005). The third (intermedia) head of the gastrocnemius only comprises ∼10% of the total mass of the gastrocnemius muscles in this species (Rubenson et al., 2006) and thus was not included in the analysis. MG and LG were dissected from the fresh-frozen left limbs and weighed to the nearest 0.1 mg. The LG and MG were then dissected from the fixed limbs for fascicle length, pennation angle and sarcomere analysis. The LG was first split longitudinally through the mid-belly to view fascicle arrangement. Photographs of whole MG and split LG made with a digital camera (Canon EOS550D with Canon EFS 18-55 and 10× lens) were imported into ImageJ for measurement of the pennation angle between muscle fascicles and their insertions on the aponeurosis (Salzano et al., 2018).

Owing to the expected within-muscle heterogeneity of strain (Ahn et al., 2003; Azizi and Deslauriers, 2014), each muscle was divided into sections for analysis. The MG was split into anterior and posterior fascicles (Carr et al., 2011) and then further split into proximal and distal fascicles, resulting in four sections. The LG was split into proximal, middle and distal sections, each spanning one-third the length of the muscle belly. Average pennation angle was found for each section by taking the mean of three angle measurements. Sarcomere lengths for each section were found using the laser diffraction techniques described in Salzano et al. (2018). A minimum of three sarcomere length measurements were taken from each muscle fascicle bundle and these measurements were averaged to obtain the mean measured sarcomere length.

Optimal fascicle length, L_O, was calculated by multiplying the length of the fascicle by the ratio of optimal sarcomere length of guinea fowl muscle (2.36 μm; Carr et al., 2011) to the mean measured sarcomere length.

We specifically calculated isometric force along the muscle fiber for input into the musculoskeletal model rather than including the influence of pennation angle because pennation angle is a separate input into the OpenSim Millard muscle model (see below for model description), which accounts for the change in pennation angle with muscle length (Millard et al., 2013).

The gastrocnemius moment arm at the ankle was experimentally measured using the tendon travel method (Landsmeer, 1961; Spoor and van Leeuwen, 1992) as described by Salzano (2020). In short, the Achilles tendon (which attaches only to the LG, MG and IG in guinea fowl) was attached to a linear transducer (model P510-2-S11-N0S-10C, UniMeasure, Corvallis, OR, USA) to measure excursion and kept at a constant 10 N tension to prevent changes in tendon strain (Fig. 1). Retroreflective markers were placed on dissected limbs to track the relative movement of the tibia and tarsometatarsus in 3D across a range of joint angles using a four-camera motion analysis system (300 Hz; Kestrel, Motion Analysis Corporation, Santa Rosa, CA, USA), and automatically synchronized to the linear transducer data within the motion analysis software (Cortex, Motion Analysis Corporation). Joint centers and a mean helical axis were calculated from motion data for each trial and used to calculate flexion angle at the ankle at each time point (Lewis et al., 2006). A cubic spline was fit to the tendon excursion versus flexion angle points using least-squares approximation, and tendon excursion was differentiated with respect to angle to estimate moment arm across the measured range of motion (30–90 deg). Average values are reported in Table 1.

We quantified the tendon force–length properties with material analysis as described in Katugam et al. (2020). In short, tendons were detached from the gastrocnemius muscles but left attached at their insertion points on the tarsometatarsus bone. Both the bone and the tendon's proximal end were connected to a material testing machine (858 Mini Bionix II; MTS Systems, Eden Prairie, MN, USA). Samples were mounted vertically using custom clamps on the tendon aponeurosis and the tarsometatarsus and attached to a 50-pound load cell (MTS Systems). The upper clamp gripped the entire aponeurosis of each sample, leaving only the free tendon exposed to loading. The tendon force–length properties were quantified by loading the tendon cyclically (20 cycles) to 4% strain. The tendon force–length curves were calculated by averaging the data from last 5 cycles of the loading protocol. Tendon force–strain curves were calculated by normalizing displacement by the length of the tendon (T_L) measured to the nearest 0.1 mm with calipers while under zero force in the material testing setup. Average values for tendon stiffness given in Table 1 were calculated from the slope of the tendon force–length curve across the last 50 points measured in the last 5 cycles of trials, at strain between 3 and 5%.

The tendon slack lengths for the LG and MG were estimated from experimental measures as described in the Supplemental Materials & Methods. Because model based estimates of muscle fiber length in a given posture are particularly sensitive to the tendon slack length (Ackland et al., 2012; De Groote et al., 2010; Scovil and Ronsky, 2006) and our calculations involved several simplifying assumptions, we further refined our experimental estimates of tendon slack length by fine-tuning the tendon slack length parameter in the OpenSim model (see Supplemental Materials & Methods for experimental tendon slack length measurement, and see below and Supplemental Materials & Methods for model development). After experimental moment arms and tendon and muscle properties were added to subject-specific models, each model was posed in the individual's fixed posture. The model's tendon slack length was adjusted iteratively in the model until the LG and MG normalized fiber lengths were within 1% of the experimentally measured values. These final values are listed in Table 2.

Simulations of 100% activation of the LG and MG were performed across the range of experimentally measured joint angles for the ankle and knee joint (Fig. 2A,B). From these simulations, we extracted the maximum elastic energy storage across all postures for each bird and the ankle and knee angles at which the maximum was achieved, and recorded the values in a pre-jump posture (Fig. 2C) (Henry et al., 2005). Tendon elastic energy stored in the pre-jump posture has been found to be a requirement for the very high power generated in guinea fowl jumping (Cox et al., 2020; Henry et al., 2005). Additionally, we recorded the normalized fiber length for each muscle at this posture at zero activation.

To determine whether components of the gastrocnemius elastic system change systematically in response to changes in demand, we evaluated the influence of treatment group (restricted versus control) on each element of morphology measured. This was accomplished using t-tests if the homogeneity of variance assumption test was passed, and using a Kruskal–Wallis test by ranks when this criterion was not met. The threshold for statistical significance was set at 0.005 after a Bonferroni correction for multiple comparisons. Likewise, the relationship between treatment group and elastic energy storage capacity was evaluated with a t-test after data passed tests for normality and homogeneity of variance, as described above for evaluation of differences between groups of individual elastic system components.

We used stepwise comparison of Akaike information criterion (AIC) values (stepAIC R Mass package; Venables and Ripley, 2002) to determine the parameters and coefficients of the full model that best predicted elastic energy storage potential across natural variation of joint postures in preparation for jumps. The full statistical model evaluated included stored strain energy (PE) as a dependent factor and, as potential independent variables, tendon stiffness (tendonK), the summed maximum isometric force capacity of LG and MG along the tendon (sumFMax), the average LG and MG optimal fascicle length (avOFL) and starting muscle length at zero activation of the LG and MG in the pre-jump posture (avLenA0c). We included possible interaction terms between muscle force capacity, tendon stiffness and muscle start length (sumMaxF×tendonK×avLenA0c) and between optimal fascicle length and tendon stiffness (avOFL×tendonK) following recommendations by Zajac (1989) of functional equivalent muscle tendon joint properties. We did not include muscle moment arm or tendon slack length in the statistical model because they both contributed to the starting muscle length at any given joint posture.

To quantify the relative explanatory power of morphological variation of any individual element to predict stored energy to a systems-level approach, we compared individual parameter models with the best multi-parameter model found by stepwise comparisons described above. The AIC value of the best model was compared with AIC values of models with individual predictors and their relative explanatory power was computed (Wagenmakers and Farrell, 2004).

The relationship between Achilles tendon elastic energy storage capacity and experimentally measured muscle-mass-normalized peak power output and jump work were both tested with a linear model with elastic energy storage as the dependent variable and peak power or jump work as the independent variable.

We found no statistically significant differences in any individual morphological property between birds that jumped during growth and those that did not (all P≥0.1; Tables 1, 2).

We found no significant differences in the capacity to store energy in the strain of elastic elements between birds restricted and unrestricted from high power activities during maturation, despite small differences in tendon strain that did not reach significance (Table 3, Fig. 3). This held true both at the peak crouched posture before jump initiation (P=0.43; Fig. 2C) and at the posture that optimized elastic energy storage (P=0.44; Fig. 2D). It should be noted that the optimum posture for elastic energy storage was at the most extended knee angle and the most flexed ankle angle tested, and was an ∼85 deg more extended ankle angle than birds used in preparation for a jump. While this more extended posture lengthened the gastrocnemius and increased the energy stored in the Achilles, it shortened the operating length of the knee extensor muscles, reducing their force-generating capacity. In the pre-jump posture, the shorter gastrocnemius length decreased elastic storage capacity by 12% for control birds and 10% for restricted birds (Table 3) in comparison to the optimal posture for energy storage.

We found that the amount of energy stored in strain of the tendon was best explained by variation in the average of the passive LG and MG muscle length at activation onset, avLenA0 (Table 4). Fig. 4A illustrates that tendon strain energy increases at longer starting lengths both between individuals and, even more strikingly, within individuals, across postures. Muscle force capacity along the tendon was the next most explanatory variable and, like normalized muscle length, shows a positive relationship with energy storage (Fig. 4B). In contrast, longer muscle optimal fascicle lengths reduced energy storage (P=0.03; Table 4, Fig. 4D) when evaluated as an individual predictor, but was not a significant factor as a predictor in the full multi-parameter model. Opposite to our predictions, tendon stiffness did not significantly correlate with elastic strain energy when evaluated as an individual predictor (P>0.1; Table 4, Fig. 4C), but did improve the explanatory power of a full model. The stepwise AIC comparison of full and reduced models found the sum of LG and MG maximum force capacity along the tendon (sumFMax), average optimal fascicle length (avOFL), muscle length at activation onset (avLenA0), tendon stiffness and the interaction of tendon stiffness and muscle force capacity as the independent predictors that best correlated with stored elastic strain energy. The relative explanatory power of each predictor followed similar patterns as seen in the individual analyses, with muscle start length and force capacity showing the greatest predictive power. This full model had an R² of 0.93 and was over 280,000 times more likely to explain the variation in strain energy than any model with only one explanatory variable.

We found little to no correlation between energy storage capacity predicted by simulations and experimentally collected jump metrics of either muscle-mass-specific work or power (Fig. 5). A linear model showed no significant relationship between either peak power or jump power per kilogram of muscle mass capacity (t=−0.36, P=0.72, R²_adj=−0.07) or jump work (t=−0.05, P=0.96, R²_adj=−0.08) and strain energy in the pre-jump posture. The negative adjusted R² values for both tests show that the variation in jump work or peak power explains only a negligible amount of variation in elastic energy storage potential. The scatterplot of standardized predicted values versus standardized residuals for both variables showed that the data met the assumptions of homogeneity of variance and linearity and the residuals were approximately normally distributed.

We found the gastrocnemius elastic system of the guinea fowl robust to variations in locomotor conditions during growth. Neither properties of individual components nor energy storage capacity varied between birds that did and did not jump throughout maturation. Nor did we find any correlation between energy storage capacity and jump performance. Variation in muscle operating length across individuals predicted energy storage capacity better than any fixed morphological property, and a systems approach incorporating multiple components was substantially able to predict energy storage capacity better than variation along any individual element.

Contrary to our predictions, we saw, in general, no systematic changes in the gastrocnemius elastic system in response to decreased demand for high power and work activities during maturation. Surprisingly, birds that were restricted from jumping throughout their entire growth period (Table 1) developed elastic systems that were largely indistinguishable from the control group that jumped, on average, almost 200 times a day.

Two factors may account for why we saw no systematic changes in the elastic system whereas observations of morphological plasticity in response to changes in functional demand abound (Baldwin and Haddad, 2002; Blazevich et al., 2003; Fitts and Widrick, 1996; Mcdonagh and Davies, 1984; Wisdom et al., 2015), even in guinea fowl in particular (Buchanan and Marsh, 2001; Salzano et al., 2018). First, several studies suggest that plasticity may vary by life stage, with lower or inconsistent plasticity in growing animals (Aucouturier et al., 2008; Johnston, 2006; Legerlotz et al., 2016). Thus, the inconsistency between the lack of plasticity in our study and the morphological variation found by others suggests that guinea fowl may exhibit lower plasticity during maturation than in adulthood. Fast-growing species, such as guinea fowl, might outpace environmental fluctuations with rapid growth and not invest in developmental plasticity (Dewitt et al., 1998; Snell-Rood, 2012). A slow-growing species (humans, for example) might have a selective advantage with greater developmental plasticity. Thus, although our treatment may have been powerful enough to induce morphological changes in adults, rapidly growing guinea fowl may be more robust to environmental perturbations.

Second, our results could be consistent with the results of other studies if the plastic response to decreased demand is not inferable from changes in response to increases in demand. For example, it may be that the increase in muscle mass that occurs in response to a certain increase in functional demand is greater than the decrease in muscle mass that occurs in response to the equivalent decrease in demand. Many studies find clear evidence of morphological plasticity, but this was in response to increased mechanical load (Atherton and Smith, 2012; Buchanan and Marsh, 2001; Kubo et al., 2007; Mcdonagh and Davies, 1984; Mersmann et al., 2017; Salzano et al., 2018) and extreme disuse (Bajotto and Shimomura, 2007; Campbell et al., 2013; Clark, 2009). Our intervention, however, eliminated jumping while maintaining consistent low intensity exercise (i.e. walking). Thus, we did not induce chronic offloading, as had been the goal in several previous disuse studies. Our results suggest that there may not be a linear dose–response relationship between changes in functional demand and morphological variation. Instead, as recently suggested (Katugam et al., 2020), there may be a range of variation in demand that is not extreme enough to induce physiological or morphological modifications above those under developmental control. If this region of stasis encompasses a wider range of disuse, it could explain both why offloading studies often require extreme disuse, such as bedrest or limb immobilization (Bajotto and Shimomura, 2007; Bebout et al., 1993; Clark, 2009), to induce change, and why we found no systematic morphological changes here. Eliminating jumping may not be an extreme enough disuse signal to induce musculoskeletal plasticity.

Thus, although we found no systematic morphological variation when restricting high power activities during maturation, this does not necessarily imply that the morphology of elastic systems does not plastically adapt to variations in functional demand. But it does suggest that there are conditions in which elastic systems may be insensitive to functional variation.

Despite this lack of consistent morphological variation between our treatment groups, restricted birds showed performance deficits. This suggests either that small morphological changes in individual elastic elements compound to alter elastic system function, that variations are significant in other MTUs that we did not quantify, or that behavioral or neural variation account for the difference in jumping performance. Our systems-level analysis aimed to specifically address the question of whether morphological variation compound within the elastic system to enable unrestricted birds to store more energy in their Achilles tendon in preparation for a jump. Again, contrary to our predictions, simulations in our subject-specific models resulted in no differences between groups in their maximum ability to store elastic energy. Taken together, the findings of minimal changes to individual MTU components and no effect on the overall elastic energy storage could indicate that morphology necessary to enable jumping is highly conserved. This could happen if rapid movements are very critical to fitness, as may be the case for prey animals for whom evasion is critical.

The first two analyses focus on plasticity of elastic systems and quantified the influence of rearing conditions on the morphology of individual components and how that variation influenced energy storage capacity across treatment groups. Our last two analyses utilize the variation within and across our treatment groups to further probe the relationship between form and function in elastic systems.

Evaluating the best predictors of energy storage, we found that muscle properties far outweighed the influence of tendon stiffness. Surprisingly, maximum isometric muscle force, while correlating with energy storage, was not the most important factor. Instead, normalized muscle length at the start of contraction was the best individual predictor of energy storage, with the longest normalized muscle lengths enabling greatest elastic storage [in agreement with results from Azizi and Roberts (2010) and Rosario et al. (2016)]. This may be because muscles that start contracting on the descending or plateau region of the force–length curve increase force capacity as they shorten, resulting in a greater equilibrium force, while muscles starting on the ascending limb of the force–length curve lose force capacity as they shorten against a tendon during an isometric contraction (Cox et al., 2019b). Further, we were particularly surprised to find that tendon stiffness alone had little to no predictive power of energy storage. Together, these data suggest that, between individuals or across an individual's lifetime, the large variation in force capacity owing to force–length or force–velocity effects may overshadow the influence of variation in tendon properties in determining tendon strain energy. This conclusion is consistent with studies in humans that found no correlation between tendon stiffness and vertical jump height (Kubo et al., 1999). Yet, this idea runs contrary to the focus on spring properties (Albracht and Arampatzis, 2006; Biewener and Roberts, 2000; Bohm et al., 2019; Fletcher and MacIntosh, 2018; Khayyeri et al., 2017; Proske and Morgan, 1987; Waugh et al., 2012, 2014) or relative spring and maximal muscle properties (Lichtwark and Wilson, 2008; Mersmann et al., 2017) in many studies that try to connect form and function in elastic systems. Our results suggest that instead, between or within individuals, elastic energy storage capacity may be more sensitive to variations that alter muscle operating lengths (tendon slack length, optimal fascicle lengths, joint postures) or muscle cross-sectional area than to changes in the tendons themselves.

Further, our results also highlight the importance of analyzing the components of an elastic system in concert rather than trying to infer performance from variation in one component. Our full model that included both muscle (maximum muscle force and starting length) and tendon properties explained changes in energy storage capacity over 280,000 times better than variation in any individual property, even when penalizing models for complexity. This, again, emphasizes the limitations of reductionist approaches to understanding how musculoskeletal morphological variation influences the energy storage capacity of an elastic system.

Contrary to our expectations, individuals that developed elastic systems capable of storing greater energy in their tendons did not take advantage of that ability to produce more powerful jumps. This suggests that morphology may play a smaller role than neural control in determining contribution of elastic energy storage in jumping. Musculoskeletal morphological variation may not be the main factor limiting jump performance.

The interaction between tendon and muscle force–length curves may, in part, provide a mechanistic explanation for this disconnect between morphology and performance. Here, we include a conceptual diagram to illustrate how variations in morphology (here maximum muscle force capacity) may be less influential than properties that can be adjusted in real time (here muscle start length) (Fig. 6). If we plot the muscle and tendon force–length curves on the same figure, it is possible to visualize how they might interact. If the muscle operates on the ascending or plateau region of the force length curve, where passive forces are minimal, and we assume that there is no slack in the tendon, a tendon strain of zero will coincide with the muscle length at the start of a contraction. Any tendon length change, then, is equal and opposite to the change in muscle length. During a fixed-end contraction, the maximal tendon strain occurs when the tendon force equals the total muscle force of the three heads of the gastrocnemius (open circles, Fig. 6). The tendon strain at equilibrium, then, is dramatically influenced by the length of the muscle when it begins to contract (Fig. 6A), reaching higher values for contractions starting at longer muscle lengths. As our results suggest (and as can be visualized by comparing the differences in the areas of the grey shaded regions between Fig. 6A and B), these dynamics can be larger than the influence of naturally occurring variations in maximal force capacity. Although we constrained this thought experiment to muscles operating on the ascending or plateau region of the force–length curve because those were the operating lengths that correspond to the range of joint angles observed in this species and are most common in vertebrates (Burkholder et al., 2001), it is interesting to note that the above argument may not hold true for muscles starting at longer lengths. Although starting at longer lengths further increases energy storage for a given muscle (both muscle force capacity and tendon resistance increasing with muscle shortening), increasing force capacity for muscles starting on the descending limb also drastically increases energy storage capacity. Thus, although changes in operating lengths many not always be more influential on energy storage than changes in force capacity, muscle start length can dramatically alter energy storage potential of an elastic system in all conditions. And although muscle operating lengths are constrained by morphology (OFL, pennation angle and moment arms), they are also easily varied with joint angle. Large variations in maximum muscle force could be compensated for by small variations in posture largely under neural control. Thus, as our results suggest, dynamic factors such as muscle operating length may influence energy storage more than temporarily fixed musculoskeletal features such as maximum muscle force capacity.

This suggests that there may be a large range of morphological variation that can be compensated for with neural plasticity and brings up questions of how the two interact. Does variation along a particular morphological axis correlate with systematic changes in neural control? If so, how do individuals search though the neural possibility space? What are the limits of neural compensation? Do we see greater morphological plasticity of the components of elastic systems in conditions that push the limits of neural plasticity?

Although musculoskeletal morphology may set the bounds of possible energy storage, individuals may not operate at their limits of elastic potential. This suggests either a significant behavioral component (restricted birds simply may have not tried as hard to jump) or that there may be benefits to real-time tunability in elastic systems. The jump of a guinea fowl is powered by both tendon recoil and simultaneous muscle work (Henry et al., 2005), as is common in many larger animals (Alexander, 1974, 1995; Kubo et al., 1999; Roberts and Marsh, 2003). The muscles that load the tendon pre-jump also contract during take-off to contribute power to the jump. In these hybrid systems, trade-offs between maximizing tendon strain energy and muscle power may explain our finding that the pre-jump posture of guinea fowl did not optimize energy storage in the tendon. Adjusting muscle lengths to maximize tendon strain may hamper muscle fiber work during take-off. Further, in a complex system such as this, with dozens of individual MTUs spanning multiple joints and working in concert with direct drive muscles with little tendon, the difference between a great jumper and a good jumper might depend less on the maximal storage capacity of any one MTU (i.e. its musculoskeletal morphology) and more on fine adjustments of neural control to harmonize the output of the collective system (Adkins et al., 2006; Enoka, 1997; Häkkinen et al., 2000).

Together, our experimental and modeling analyses suggest that performance advantage of the control birds, which practiced jumping throughout maturation, may lie less in the body's modification of individual elastic elements, and more in the fine tuning of neural circuits to coordinate muscle activation timing to take better advantage of what they each possess. Although restricting normal locomotor behavior during growth (i.e. eliminating practice) likely leads to deficits in neural control, neural plasticity is potentially a rapidly reversible pathway to adapt an elastic system to functional variation. Given the potential short timescale of neural plasticity (Adkins et al., 2006; Yoxon and Welsh, 2019), greater sensitivity of neural locomotor/movement stimuli could allow the individual to adjust the dynamics of an elastic system during growth without making potentially irreversible changes to morphology that could be detrimental in subsequent stages of growth or in adulthood if environmental conditions or functional demand rapidly change. Thus, one could interpret the results of our study as suggesting that practice during growth may indeed be more related to forming the neural framework for jumping than for forming the musculoskeletal framework. This also suggests the specific hypothesis that individuals restricted from an activity during growth may be capable of reversing the resulting neural deficits with practice later in life.

Several modeling simplifications could have influenced our results. For instance, the gastrocnemius elastic system is not the only one that could contribute to jump power. Although the Achilles is the largest tendon involved, many other digital tendons spanning both the ankle and tarsometatarsus joint have the potential to contribute to jump power but were not included in our analysis. Furthermore, changes in the characteristics of muscles spanning proximal joints may also have contributed to the differences in jump power, but these muscles were not modeled. Additionally, potentially important dynamic effects were ignored. For simplicity, we simulated the amount of energy stored in the Achilles tendon during fixed-end contractions, where the joint posture was constant as the muscle and tendon dynamically responded to increasing muscle activation. Activating muscles while altering joint posture would alter these dynamics, perhaps amplifying the influence of individual differences in input or output lever lengths or force–velocity effects not apparent from group averages. Likewise, we did not measure potential variation in muscle-specific tension, which could alter our estimates of maximal muscle force capacity. Nor did we measure and include individual variation in muscle or aponeurosis passive elastic properties that could significantly alter energy storage (Arellano et al., 2019; Huijing and Ettema, 1988; Lemos et al., 2008). Although this is a common approach in musculoskeletal modeling (Millard et al., 2013; Seth et al., 2018; Zajac, 1989), variation in the aponeurosis and free tendon stiffness (Ettema and Huijing, 1989; Finni, 2006) have the potential to introduce errors (Epstein et al., 2006; Lieber et al., 2017).

Another possible limitation was the modeling choice to focus on the potential for an individual to store energy in the strain of their tendon. How that energy is released and how that energy release interacts with synchronous muscle activation could also influence jump performance (Ilton et al., 2018). Because jumps are likely powered both by tendon recoil and muscle work (Azizi and Roberts, 2010; Henry et al., 2005), there may be a trade-off between the work the muscle puts into tendon strain and that which is left available to power the jump during tendon recoil (Sutton et al., 2019). Future work could involve simulation of jumps in these subject-specific models to assess the contributions of these dynamic factors. Thus, although we found no consistent change in components of the gastrocnemius elastic system owing to decreased demand for high power activities during growth, more complex models may provide insight into the ways in which morphological variation constrains performance.

We found that decreasing the demand for high power and work during growth can influence adult performance but does not necessarily lead to morphological plasticity. We found no difference in energy storage capacity between groups that did and did not jump throughout maturation or any correlation with experimentally measured jump performance. We conclude that the gastrocnemius elastic system in the guinea fowl displays little to no morphological plastic response to decreased demand during growth and that neural control of elastic systems may constrain performance more than morphology.

We thank Dr Kirsty McDonald and Justin Csaszar for help collecting this data. We also thank the staff at the Central Biological Laboratory at The Pennsylvania State University for their care of our animals during the study and Dennis Ripka for support building housing.