The role of hind limb tendons in gibbon locomotion: springs or strings?

Tendons link muscle to bone and have an important function in force transfer from contracting muscle fibres to bone – or, in the case of bi-articular muscles, force transmission from one joint to another – resulting in joint motion. In addition, it has been shown that tendons also have an important role in reducing the cost of locomotion by: (1) reducing the amount and rate of shortening of muscle fibres, allowing near-isometric force production and contraction in their optimal fibre length range (Biewener and Roberts, 2000; Lichtwark and Wilson, 2005a; Lichtwark et al., 2007; Lichtwark and Barclay, 2010); and (2) acting as elastic springs, storing and releasing elastic strain energy during locomotion (Alexander, 1984; Alexander, 2002). Such an energy-saving function has been demonstrated for different tendons in a wide range of species, such as the hind limb tendons in hopping kangaroos and wallabies (Dawson and Taylor, 1973; Alexander and Vernon, 1975; Biewener and Baudinette, 1995; Ker et al., 2000) and the Achilles tendon in humans (Lichtwark and Wilson, 2005b; Lichtwark et al., 2007).

Not all tendons are, however, suitable to act as ‘elastic springs’. The amount of strain energy that can be stored in a tendon depends on the elongation of the tendon. In turn, this depends both on the tendon force and the dimensions and material properties of the tendon, in particular its volume and compliance or stiffness. To enable energy-saving in locomotion via stretch and recoil of tendons, tendons should have a compliance that allows a certain degree of elongation when subjected to forces acting during normal ambulation, ca. 4% strain in human walking (Ishikawa et al., 2005; Lichtwark and Wilson, 2007; Lichtwark et al., 2007) and 3–6% strain in horse walking (Dimery et al., 1986). If tendons are too stiff relative to the occurring forces, the capacity to store energy as elastic energy in the tendon will be limited, whereas tendons that are too compliant might hamper motor and positional control and increase the risk of injury.

While tendon stiffness is dependent on tendon dimensions, the Young's modulus, i.e. the ratio of stress over strain, is not affected by specific dimensions of the tendon (Peltonen et al., 2010; Peltonen et al., 2012). The Young's modulus has been documented for various mammalian tendons and amounts to roughly 1.2 GPa (Pollock and Shadwick, 1994), though reported values range from 0.4 to 1.7 GPa (Ker et al., 1988; Zajac, 1989; Pollock and Shadwick, 1994). Whether this wide range of reported values is due to differences between species, tendons (i.e. functional demand) or measurement techniques remains unclear (Matson et al., 2012).

Another important property of a tendon is its hysteresis, i.e. the amount of energy lost as heat upon recoil. The lower the hysteresis, the higher the efficiency of the tendon. Hysteresis rates vary widely, with values ranging from 3 to 38% in mammalian tendons (Ker, 1981; Pollock and Shadwick, 1994; Maganaris and Paul, 2000) and reported values of 19–26% for the human Achilles tendon (Maganaris and Paul, 2002; Lichtwark and Wilson, 2005).

In addition, a characteristic that is often disregarded – because it is difficult to measure – is the resonant frequency of the tendon, i.e. the frequency at which a maximum amplitude is achieved with a minimum input force. While resonant frequencies are to be avoided in man-made constructions, achieving resonance is actually desirable in animal locomotion as it minimizes locomotor cost (Farley et al., 1993; Ahlborn et al., 2006).

Because the tendon is one part of the functional muscle-tendon unit (MTU), tendon properties should always be viewed in relation to the properties of the muscle(s) to which the tendon is attached, namely its force-generating capacity and the muscle fibre length. The stress imparted to the tendon is proportional to the ratio of the physiological cross-sectional area of the muscle (PCSA) – which is a measure of the force-generating capacity of the muscle – to the cross-sectional area of the tendon (TCSA). The higher this ratio, the more stress will be exerted on the tendon, resulting in a higher tendon strain and energy absorption capacity. Thus a configuration where a long, thin tendon is connected to a large, pennate muscle will favour enhanced elastic energy savings because of the high force-generation potential of the relatively short muscle fibres (Biewener, 1998a) compared with a configuration where this tendon is connected to a slender, parallel-fibred muscle. The moment arm of the muscle (group) also plays a crucial role and, more specifically, the ratio of muscle moment arm to ground force moment arm [the so-called ‘effective moment arm’ (Biewener, 2005)]. Recent studies on human running biomechanics (Raichlen et al., 2011; Scholz et al., 2008) have demonstrated that the muscle moment arm has an even stronger influence on the amount of energy stored in the Achilles tendon than its mechanical properties, with shorter muscle moment arms leading to higher tendon stresses and higher amounts of energy stored.

Finally, alongside suitable tendon properties relative to the force-generating capacity of the muscle, appropriate kinematics and locomotor dynamics are required to allow elastic energy storage and release in the tendons. Negative joint power should precede positive joint power to facilitate conversion and storage of potential energy in elastic strain energy in stretched tendons and subsequent tendon recoil, allowing release of the elastic energy (Rubenson et al., 2011). This prerequisite is, for example, fulfilled in the calf muscles during human running, allowing storage and release of elastic energy in the Achilles tendon.

Among hominoids (Superfamily Hominoidea), the Achilles tendon is considered a hallmark feature of modern humans related to the evolution of habitual bipedalism. Its dimensions and mechanical properties make it particularly suitable to act as an elastic spring during human walking and, in particular, running (Bramble and Lieberman, 2004). Apart from humans, the lesser apes or gibbons (Family Hylobatidae) are the only extant hominoids to possess a relatively long Achilles tendon with a substantial external portion. In contrast, the (non-human) great apes only possess a short Achilles tendon, which is located inside the triceps surae muscle (Payne et al., 2006). This observation led us previously to speculate that the gibbon Achilles tendon might act as an elastic energy store during hind-limb-powered locomotion (Vereecke and Aerts, 2008).

Gibbons are highly arboreal primates that rely predominantly on suspensory locomotion, in particular brachiation, to travel across the forest canopy, yet hind-limb-powered locomotion – such as leaping, climbing and bipedalism – is also regularly used. Gibbons will leap to cross large gaps in the forest canopy, while bipedal bouts on top of branches or vines are used as a run-up for leaps or between periods of brachiation. Given that hind-limb-powered locomotion (i.e. leaping, climbing, bipedalism) constitutes an important part of the gibbon's locomotor repertoire [up to 40% of travel time (Fleagle, 1976)], it seems reasonable to assume that anatomical adaptations in the hind limb that might reduce the locomotor cost of leaping, climbing and bipedalism will be favoured, as long as this does not negatively affect the efficiency and/or performance of brachiation. This lets us suggest that a relatively long Achilles tendon could have evolved in the hylobatid lineage, independent from the evolution of the Achilles tendon in hominins, as an adaptation to increase overall locomotor efficiency. Long tendons in the hind limb allow the muscle mass of the hind limb to be located more proximally, minimizing distal inertia and, as such, the efficiency of brachiation, while they could simultaneously contribute to reducing the locomotor cost of leaping and bipedalism by acting as elastic springs. Without a direct measure of the energetics of gibbon locomotion it remains difficult to assess this proposition; however, we can evaluate the theoretical possibility for elastic energy storage and release in the hind limb tendons of gibbons by investigating their anatomical and mechanical properties. To evaluate the hypothesis that the relatively long tendons in the gibbon hind limb can act as elastic springs during hind-limb-powered locomotion, we have harvested samples of the Achilles and patellar tendon, both prominent hind limb tendons, in a series of fresh-frozen gibbon cadavers, and have determined their material properties.

Tendon samples were taken from the left or right hind limb of fresh-frozen gibbon cadavers (patellar tendon: N=8, Achilles tendon: N=14). Cadavers were obtained through collaboration with the National Museums of Scotland (Edinburgh, Scotland) and the Royal Zoological Society of Antwerp (Belgium). All animals included in this study were housed in zoos and died under natural circumstances. None of the specimens demonstrated musculoskeletal pathologies (Table 1).

The cadavers were thawed at room temperature and the triceps surae and quadriceps femoris were freed from surrounding tissue. The quadriceps femoris were cut at the site of origin (iliac spine and femur shaft) and insertion (tibial tuberosity), while the triceps surae were either cut at both origin (femoral condyles) and insertion (tuber calcanei) (set B) or only at origin and kept attached to the calcaneus (set A). The different preparation of sets A and B was due to a different clamping technique used for these two sets of Achilles tendon, but did not influence the anatomical measurements that were taken.

Following the procedure described in a previous anatomical study (Vereecke et al., 2005), a series of measurements was taken of each muscle head using a digital scale (Sartorius, Bradford, MA, USA; precision 0.1 g), a digital camera (Lumix DMX, Panasonic) and digital callipers (Mitutoyo, Tokyo, Japan; precision 0.01 mm). Measurements include: muscle belly mass (m; g), muscle fascicle length (FL; mm), pennation angle (PA; deg), muscle-tendon unit length (MTUL; mm), and external and internal tendon length (ETL, ITL; mm). ETL was defined as the length of the outer tendon portion (i.e. from the most distal muscle fibres to the distal end of the tendon), while ITL refers to the inner aponeurotic portion of the tendon (i.e. from the proximal end of the tendon to the most distal muscle fibres). Tendon length (TL) is defined as the total length of the tendon (i.e. TL=ETL+ITL).

For the triceps surae, the PCSA was calculated as the sum of the PCSA of the lateral and medial gastrocnemius and soleus muscles. For the quadriceps femoris, the PCSA was calculated as the sum of the PCSA of the rectus femoris and the three vasti (lateral, medial and intermedius). PCSA is used to estimate maximum force-producing capacity (Ward and Lieber, 2005). To allow comparison with humans, PCSA is scaled to body mass to the two-thirds power [i.e. PCSA/(body mass)^2/3].

The proximal and distal ends of the tendon samples were dried over a distance of 10 mm, while the rest of tendon was kept hydrated (using damp tissue and PVC film). The tendon samples were clamped at the proximal and distal dry parts using flat metal, self-tightening plates. In a subset of the Achilles tendon samples in which the distal part was kept attached to the calcaneus (set A), samples were fixed by clamping the proximal dry end (self-tightening clamp) and the calcaneus (custom-made clamp). In each case, the tendon section length corresponds to the actual part being tested and was measured as the distance between the inner borders of both clamps after the tendon sample had been mounted in a materials tester.

The patellar tendons were tested with the patella still imbedded in the tendon. In this way, the full length of the patellar tendon could be tested, a situation that is most representative of the condition in vivo. This means, however, that the actual measured tendon includes both the so-called quadriceps tendon (proximal to the patella) and the patellar ligament (between the patella and the tibial tuberosity). Thus, in this paper, the ‘patellar tendon’ is defined as the tendon running from the quadriceps femoris to its insertion on the tibial tuberosity. The choice to test both the proximal and distal part of the tendon, with imbedded patella, was based on the observation that both parts are continuous ontogenetically, a situation that is retained in adult morphology by collagen fibres running over the patellar bone (Mérida-Velasco et al., 1997; Franchi et al., 2009; Toumi et al., 2012). Testing the whole complex together was considered the best approximation of the situation in vivo. It should be noted that in most studies on the functional anatomy of humans, the ‘patellar tendon’ (sometimes called the patellar ligament) is most commonly defined as the tendon portion between the patella and the tibial tuberosity. This should be kept in mind when comparing values obtained in this study with other published values.

The material properties of the tendon samples were obtained using an Instron materials testing machine (Instron Corp., Canton, MA, USA), after calibrating the load cell to 0 N with slack tendon. A load-controlled cyclic sine wave test (1 kHz sampling rate, recording every cycle) with a starting load of 5 N was used for all tendon samples. For each tendon sample, two to three cyclic tests consisting of 40 cycles were performed, increasing the maximal load in each subsequent test. These cyclic tests were followed by a single ramp test to obtain data under maximal tendon stress. Test protocol conditions were based on previously published studies by Ker and colleagues (Ker, 1981; Ker et al., 2000) investigating material properties of tendons in vitro and were adapted to the specificities of the current samples.

Because of the limited sample size of our original data set – related to the limited and unpredictable availability of cadaveric material of gibbons – we chose to increase our data set by adding more samples in the following year. This resulted in a larger data set, yet one comprising two sets of samples tested with slightly different testing conditions. Set A included only Achilles tendon samples, while set B included both Achilles and patellar tendon samples. Tests for set A were carried out using an Instron model 8500 at 4 Hz and 36°C (University of Leeds, UK). Set B was tested using an Instron model E3000 at 2 Hz and 20°C (The University of Liverpool, UK). The frequencies correspond to the in vivo loading frequencies observed during slow to fast bipedalism (contact times range from 0.2 to 0.6 s).

Data were analyzed using GNU Octave 3.2.4 software. For each tendon sample, load versus displacement curves were plotted for the cyclic test with the highest maximal load. As some conditioning occurs during the first loading cycles [as is the case in all tendons (Pollock and Shadwick, 1994; Maganaris, 2003)], cycle 20 was selected for calculation of the hysteresis to ensure proper preconditioning of the tendon sample. A second-order polynomial was plotted to the load–displacement data during loading and unloading, and integrated for the maximal recorded load of the tendon sample in that test. The hysteresis, H (%), was calculated as the difference between the integral of the loading and unloading curve, divided by the integral of the loading curve. The hysteresis represents the amount of energy lost as heat during unloading as a percentage of total energy stored in the tendon during loading.

The tendon's material properties (stiffness, stress and strain) were calculated using the load and displacement data recorded during the single ramp test.

Stiffness, S (N mm⁻¹), was determined as the slope of the linear regression of the load–displacement data during loading above 50% maximum volumetric contraction (MVC, which was estimated by multiplying the PCSA with a maximal isometric muscle stress of 0.30 N mm⁻²) (Erskine et al., 2009). Coefficients of determination (r²) of the included data were between 0.9950 and 0.9998, supporting a good linearity of the load–displacement relationship (above 50% MVC). Mean stiffness and standard deviation (s.d.) of the samples were calculated for the Achilles and patellar tendon.

Stress, ε (MPa), and strain, σ (%), were calculated by dividing load by TCSA and tendon elongation by tendon sample length, respectively. In a similar fashion as determining stiffness, the Young's modulus, E (MPa), was calculated as the slope of the linear regression of the stress–strain data in the linear portion of the loading curve (above 5 MPa). The r² of the regression equations was between 0.9924 and 0.9997. For each tendon, the highest calculated modulus was reported. In addition, the stress values of each tendon were calculated at 2, 3, 4 and 5% strain and plotted as box plots (means ± s.d.) to illustrate the inter-individual variability.

Results are presented as mean ± s.d. All statistical analyses were performed in GNU Octave or Excel 2007.

Available kinematic and kinetic data from previous studies of bipedal locomotion (Vereecke and Aerts, 2008; Vereecke et al., 2006) and leaping (Channon et al., 2010a; Channon et al., 2011; Channon et al., 2012) in gibbons were used to investigate the role of the Achilles and patellar tendon during locomotion.

The triceps surae has a comparable architecture in gibbons and humans, consisting of a pennate medial and lateral gastrocnemius, originating from the medial and lateral femoral condyles, and a short-fibred, pennate soleus, inserting via a shared Achilles tendon to the calcaneal tuberosity. The volumetric division in soleus, medial and lateral gastrocnemius is, however, different in gibbons and humans. In gibbons, either the lateral gastrocnemius (36–46% of triceps PCSA) or soleus (25–40% of triceps PCSA) make up the largest part of the triceps surae, in both volume and PCSA (Channon et al., 2009; Vereecke et al., 2005), while in humans the soleus typically constitutes ~60% of the PCSA of the triceps surae (Morse et al., 2005). The Achilles tendon is well developed in gibbons, amounting for 40% of the triceps surae MTUL (i.e. ETL/MTUL; 74% if both external tendon and inner aponeurotic portions are considered). This relatively long Achilles of gibbons is remarkable, given its absence or small size in other non-human apes (Payne et al., 2006). Among the hominoids, a long Achilles tendon is also observed in humans, with an external tendon length amounting to 64–95% of the MTUL of the triceps surae (Table 2).

The gibbon quadriceps femoris consists of the large muscle bellies of the vastus lateralis and intermedius, and the more slender vastus medialis and rectus femoris, a similar organization as is found in humans (O'Brien et al., 2010; Blazevich et al., 2006). The muscle fibres of these four heads insert into a common quadriceps tendon, which is attached to the patella and continues as the ‘patellar tendon’ from the apex of the patella to the tibial tuberosity. The length of the patellar tendon amounts to 12% of the quadriceps MTUL in gibbons (i.e. ETL/MTUL; or 64% if external tendon and inner aponeurotic portion are considered; Table 2). In humans, the length of the patellar tendon (ETL) amounts to 19–22% of the quadriceps MTUL and is thus relatively longer than in gibbons (note that reported values of gibbon and human patellar tendon include length of the patella and of the quadriceps tendon; Table 2).

The triceps surae are relatively more slender in gibbons compared with humans (scaled PCSA of 278 versus 1098; Table 2, Fig. 1), which is expected given the predominantly suspensory lifestyle of gibbons. The same can be said for the quadriceps, the PCSA of which is also relatively smaller in gibbons compared with humans (scaled PCSA of 548 versus 884–1259; Table 2), yet the quadriceps are twice as large as the triceps surae in gibbons, while these muscle groups have almost equal PCSAs in humans. This might point to a task division between both muscle groups in gibbons, with the quadriceps being more important for production and/or absorption of work in leaping and bipedalism and the triceps surae being more reliant on elastic storage and recoil.

Comparative data on the dimensions of the triceps surae and quadriceps femoris are provided in Table 2 and Figs 1 and 2.

The mean stiffness of the gibbon's Achilles tendon was 99.6±42.7 N mm⁻¹, and that of the patellar tendon was 299.1±95.6 N mm⁻¹ (Fig. 3). The Young's modulus was 701.3±286.5 MPa for the Achilles tendon and 300.1±130.0 MPa for the patellar tendon (Fig. 3). Hysteresis was lower for the Achilles tendon, 13.5±3.4%, than for the patellar tendon, 25.5±4.2%.

The force–elongation curves of the 14 Achilles tendon samples and the eight patellar tendon samples during loading and unloading are shown in Fig. 4. The figures clearly illustrate the visco-elastic nature of tendon, with a nonlinear behaviour under low loads (‘toe region’), and a linear behaviour at higher load values (typically above 50% MVC). Stiffness and Young's modulus values are obtained in the linear portion of the force–elongation curves. To illustrate the tendon properties under lower load conditions, which might very well prevail during habitual locomotion, we have also calculated stress values (and s.d.) at 2, 3, 4 and 5% strain for both tendons (Fig. 5).

The energy absorption capacity of the Achilles tendon at 30 MPa, the average stress-in-life of the tendon, is 0.799 J using Eqn 3 (Biewener and Baudinette, 1995). Using the same formula and an average stress-in-life of 17 MPa, the estimated energy absorption capacity obtained for the patellar tendon amounts to 1.560 J.

Comparison of the gibbon Achilles and patellar tendon demonstrates some substantial differences in dimensions and material properties. The gibbon Achilles tendon has a relatively low stiffness and is long and thin compared with the dimensions of the triceps surae (high ETL/FL and PCSA/TCSA ratios; Table 2). This indicates that it will be subjected to relatively high levels of tendon stress and strain during locomotion, facilitating storage of elastic strain energy. The hysteresis of the Achilles tendon is reasonably low, indicating that upon recoil, 86.5% of the stored energy will be available to contribute to the positive mechanical work that has to be delivered to raise and accelerate the body centre of mass (COM) and the limbs. The patellar tendon, in contrast, has less favourable material properties to facilitate elastic energy storage and release. Its hysteresis is rather high, averaging 25.5%, meaning that a quarter of the stored strain energy will be lost as heat during recoil. Despite the quadriceps being bulky, powerful muscles, the patellar tendon is a low-stressed tendon with a high safety factor compared with the Achilles tendon because of its relatively large diameter and, hence, low PCSA/TCSA ratio (i.e. 56 versus 100 for the Achilles tendon; Table 2). However, the Young's modulus of the patellar tendon is also significantly lower than that of the Achilles tendon (E=0.3 GPa versus 0.7 GPa; P=0.0014, t=3.719, d.f.=20); this means that even under low stresses, the patellar tendon will be subjected to substantial tendon strain, which might facilitate elastic energy storage in locomotion.

The visco-elastic nature of tendons allows stretch and recoil during locomotion, increasing locomotor efficiency, yet this elasticity or compliance also affects positional control. An interesting parameter hereof is the ratio of external tendon length to muscle fascicle length (ETL/FL). The higher this ratio, the more influence the tendon properties have on the behaviour of the MTU, while a low ratio indicates a small effect of tendon extension on MTUL, suggesting optimization for precise positional control. The ratio for the gibbon Achilles tendon is six times higher than for the patellar tendon (2.4 versus 0.4; Table 2), suggesting a spring-like function for the Achilles tendon while the patellar tendon might be more suitable for rapid force transfer and positional control, despite its low Young's modulus. This observation also corresponds to the relatively high measured stiffness for the patellar tendon (300 N mm⁻¹ versus 100 N mm⁻¹ for the Achilles tendon).

Recent work by Lichtwark and Barclay (Lichtwark and Barclay, 2010) has shown that tendons with a low stiffness improve both power output and efficiency of muscles. In tests using artificial mice tendons, they showed that muscles connected in series to a relatively compliant tendon had a higher mechanical efficiency than those connected to a stiff tendon. Translated to gibbon tendons, this would mean that the gibbon Achilles tendon, with a relative compliance (i.e. RC=1/S×MVC/FL) of 10%, would achieve a muscle efficiency of ~32%, while the patellar tendon, with an RC of 3%, would have an efficiency of ~26% [by comparison, the RC of ‘stiff’ rat Achilles tendon is 5%, while the ‘compliant’ wallaby Achilles tendon has a RC of 35% (Litchwark and Barclay, 2010; Lichtwark and Wilson, 2005a)]. Compliant tendons are particularly important to improve the power output of short-fibred muscles, which have a restricted shortening range, such as the triceps surae in gibbons and humans.

Finally, stiffness values of tendons have also been correlated to tendon strength: more compliant tendons (lower stiffness) are weaker, stiffer tendons are stronger (Matson et al., 2012). This would suggest that the gibbon Achilles tendon is more susceptible to injury than the patellar tendon.

In conclusion, the anatomical dimensions and mechanical properties of the gibbon Achilles and patellar tendon suggest that these hind limb tendons fulfil a different role in locomotion. The Achilles tendon seems suitable for elastic energy storage and release, because of its low hysteresis and relatively high compliance and because the tendon is connected in series to the short-fibred, pennate triceps surae (leading to a high ETL/FL ratio of the MTU). Such muscle architecture combined with the material properties of the Achilles tendon likely results in a higher muscle efficiency compared with that achieved by the patellar tendon, which is stiffer, has a higher hysteresis and a lower ETL/FL.

No direct measures of tendon force, or stress, are available for gibbons, yet the stress-in-life value (i.e. PCSA/TCSA×0.3 MPa) can be used as a rough estimate of tendon stress occurring in locomotion, with high stress values pointing to a spring-like function and a higher fatigue resistance (Ker et al., 2000). For the Achilles tendon we obtained a stress-in-life value of 30 MPa, while the estimated stress in the patellar tendon was 17 MPa, suggesting that the Achilles tendon might be more suitable for elastic energy storage (and recoil), as well as being more fatigue resistant. Though these values are based on theoretical calculations, rather than on actual measurements, they roughly correspond to the stress levels obtained in our experiments at 4% Achilles tendon strain (27±11 MPa) and 5% patellar tendon strain (14±7 MPa; Fig. 5); strain values that are comparable to those observed in human walking and jumping [4–6% (Kurokawa et al., 2001; Kurokawa et al., 2003; Ishikawa et al., 2005; Lichtwark and Wilson, 2007; Lichtwark et al., 2007)].

During normal ambulation, stresses in mammalian tendons have been reported to range from 5 to 25 MPa (Biewener and Roberts, 2000; Svensson et al., 2012), while increasing in fast locomotion and strenuous activities. The tendon stresses of galloping horses amount to up to 40–50 MPa (Biewener, 1998b); in humans, stresses of 40 MPa have been reported in the patellar tendon during jumping (Svensson et al., 2012) and up to 111 MPa in the Achilles tendon during running (Ker et al., 1988).

Estimates of tendon stresses are useful, as they allow us to calculate the energy absorption capacity of the tendons during locomotion. Calculations using stress-in-life values of 30 MPa for the Achilles tendon and 17 MPa for the patellar tendon yielded energy absorption capacities of 0.799 and 1.560 J, respectively. When scaled to an average body mass of 6 kg, this gives us 0.13 and 0.26 J kg⁻¹, respectively. Thus, with its relatively low modulus, the patellar tendon achieves a higher energy absorption capacity than the Achilles tendon, at lower stress values. In these calculations, tendon volume is calculated as the product of TCSA and total tendon length, assuming a constant diameter of the tendon over its full length. The TCSA has been reported to vary across the length of feline (and human) tendons (Cui et al., 2009), yet a constant TCSA is not an unreasonable assumption to be made in the estimation of energy absorption capacity. The gibbon patellar tendon has a rather constant diameter (with exclusion of the patella) and although the Achilles tendon tapers distally (in the mediolateral direction), it also becomes thicker (in the anteroposterior direction), leading to a relatively constant cross-sectional area. It should also be noted that tendon length – which is used in our calculations of energy absorption capacity – includes both the outer tendon (ETL) and the inner aponeurotic part (ITL), because previous studies have shown that stress is dissipated over both parts (Muramatsu et al., 2001; Arampatzis et al., 2007). However, most human studies have only included the outer portion of the tendon in their measurements.

Previous studies on gibbon bipedalism (e.g. Vereecke et al., 2006) have shown that the amount of external work performed during a bipedal bout averages approximately 3.5 J or 0.72 J kg⁻¹ m⁻¹. This means that, in theory, the Achilles tendon could contribute 18% and the patellar tendon 36% of the energy requirements of bipedalism via elastic recoil, provided that tendon stress levels of 30 and 17 MPa, respectively, are reached, and that all requirements for elastic storage and recoil are fulfilled. During leaping, the external work amounts to 3.83 J kg⁻¹ m⁻¹ in an average orthograde single-footed leap (Channon et al., 2010a), leading to theoretical energy contributions of 1–7%. Yet, these theoretical calculations do not take into account that tendon stresses might be higher in leaping than in bipedalism. The obtained values should therefore be considered as rough estimates of the contribution of both tendons to the locomotor cost of gibbons via an elastic energy-saving mechanism.

Despite their predominantly suspensory lifestyle, gibbons also regularly engage in hind-limb-powered locomotion, namely bipedalism, climbing and leaping [up to 40% of their travel time (Fleagle, 1976)]. The kinematics and dynamics of bipedalism and leaping of gibbons have been the subject of previous studies, where we showed that gibbons are proficient leapers as well as adept in bipedalism even though showing no clear morphological adaptations to these locomotor modes (Vereecke et al., 2006; Channon et al., 2010a; Channon et al., 2011). We know that tendons can enhance muscle efficiency by allowing near-isometric contraction of the muscles to which they are connected, yet we specifically wondered whether the gibbon hind limb tendons could also contribute to the power-generating capacities of muscles via elastic energy storage and recovery.

The material properties of the Achilles tendon seem well suited to fulfil this task, and the dynamics of bipedalism also facilitate storage and recoil of elastic strain energy in the Achilles tendon. Although gibbons rarely run bipedally, i.e. have a flight phase, their bipedal locomotion can be described dynamically by a spring-mass mechanism, with in-phase fluctuation of potential and kinetic energy and lowest position of the COM during midstance (Fig. 6B, top), as is the case for human running (Vereecke et al., 2006). The downward motion of the body COM coincides with negative ankle joint power (Vereecke and Aerts, 2008) and knee flexion and ankle dorsiflexion, stretching the patellar and Achilles tendon, allowing conversion of potential energy to elastic strain energy (Fig. 6B, middle). In the final part of stance, knee extension is followed by ankle plantarflexion, unloading the tendons as the ground reaction force drops, and facilitating recoil and release of energy. The released energy can contribute to forward acceleration and propulsion of the COM, as it coincides with an upwards motion of the COM and positive COM and ankle joint work (Fig. 6B, bottom). This suggests that the Achilles tendon can have a similar role in gibbon bipedalism as in human running, acting as an elastic spring to increase the locomotor efficiency.

The role of the patellar tendon in gibbon bipedalism is more difficult to evaluate, as we have no accurate calculations of knee joint power. We can, however, infer its potential role in gibbon bipedalism by looking at the joint angles and by considering its function in human bipedalism. The knee joint kinematics show a similar double-humped profile in gibbon and human bipedalism, with knee flexion followed by extension in stance and swing (Fig. 6B, middle). These knee joint kinematics facilitate stretch and subsequent recoil of the patellar tendon, allowing elastic energy storage and release when coinciding with negative joint work (energy absorption) and positive joint work (energy generation), respectively.

In human locomotion, the knee displays a mainly negative joint work and power profile, which points to a primary function as a damper (energy absorption) and in adjusting limb stiffness (Farris and Sawicki, 2012). Energy recovery via recoil of the patellar tendon does not lead to a substantial reduction in energy requirements because of the high efficiency of the mostly eccentric contraction of the quadriceps in human bipedalism. In gibbon bipedalism, the knee remains more flexed throughout stance (Fig. 6B, middle) compared with in human bipedalism, requiring strong eccentric contraction of the quadriceps to counteract gravity and the activity of the hamstrings. As observed in humans, elastic energy storage could occur in the patellar tendon during gibbon bipedalism, yet tendon recoil probably contributes little to the energy requirements.

During leaping, the knee and ankle joint kinematics again facilitate stretch and recoil of the hind limb tendons (Fig. 6A, middle). Unlike bipedalism, leaping is an acyclic motion with a continuous upward motion of the COM (Fig. 6A, top) and hence predominantly positive COM work. The joints contributing to this mechanical energy production most likely show negative and positive power peaks – allowing storage and recoil of elastic energy – yet this cannot be unequivocally demonstrated because of the lack of separate work and power profiles of the different joints. We presume that loading of the hind limb tendons with energy release upon recoil will occur in leaping, in particular in leaps with an apparent countermovement, contributing to mechanical power generation and enhancing muscle performance. Again, the compliant Achilles tendon is the more likely candidate for elastic energy storage and release, while the quadriceps MTU seems more suitable for power production (considering the large volume of the quadriceps and short muscle moment arm) and transmission of energy from the large hip extensors (Channon et al., 2009; Channon et al., 2010b).

The values we obtained for the Young's modulus of the gibbon hind limb tendons – 0.3 GPa for the patellar tendon and 0.7 GPa for the Achilles tendon – are at the lower end of reported moduli of mammalian tendons [range 0.4–1.7 GPa (Ker et al., 1988; Zajac, 1989; Pollock and Shadwick, 1994)], but do fall within the range reported for human Achilles [0.14–0.90 GPa (Magnusson et al., 2003; DeFrate et al., 2006; Waugh et al., 2012)] and patellar tendons [0.12–0.85 GPa (Hashemi et al., 2005); 0.30–0.85 GPa (Svensson et al., 2012)]. However, using ultrasound techniques, much higher moduli were obtained in humans: 2 GPa and higher for the Achilles tendon (Maganaris et al., 2008; Peltonen et al., 2010) and 1–1.5 GPa for the patellar tendon (Magnusson et al., 2008; Hansen et al., 2006). This wide range of reported values illustrates the difficulties in obtaining accurate material properties from tendons, and the differences between in vitro and in vivo results suggests that the test protocol affects the obtained results (Hansen et al., 2006; Hashemi et al., 2005; Maganaris et al., 2008). One of the difficulties in in vivo studies seems to be the accurate determination of tendon dimensions (Magnusson et al., 2008; Svensson et al., 2012). In addition, in vitro studies work with post-mortem material and often test only part of the free tendon, whereas the aponeurosis is also accounted for in ultrasound studies and vascularisation is intact, which might affect material properties (Benjamin et al., 2008). This makes it difficult to compare material properties obtained from different studies; however, data obtained using the same methodologies (e.g. same research groups, equal force range) are largely comparable, as well as data of different tendons obtained in the same study, e.g. tendon properties in the feline (Cui et al., 2009) and turkey hind limb (Matson et al., 2012). These latter studies indicate that there might be differences between tendons of different muscles, suggesting a structural adaptation of tendon in relation to its function. This challenges earlier findings of similar tendon properties across different muscles and species (Bennett et al., 1986; Maganaris and Paul, 2002; Pollock and Shadwick, 1994). Our results on the gibbon Achilles and patellar tendon support the idea of adaptation of tendon structure to a differing functional demand, yet structural differences should be further documented by a histological comparison between the tendons of different muscles.

There are some limitations of this study that should be kept in mind when evaluating the data. This study presents unique data and includes as many different individuals as possible; however, the total sample size remains limited. This is an inevitable issue when studying endangered taxa, such as the hylobatids (Cites I list). To maximize sample size, experiments were performed in two separate sets (A and B) in consecutive years, as access to specimens was dependent on death by natural causes. All included specimens were captive animals and activity levels of the animals might have been reduced compared with wild gibbons. The specimens were all adult animals, yet covered a wide age range (9–41 years). These factors might affect tendon properties, though the effect of age and exercise has not been demonstrated unequivocally (Hashemi et al., 2005; Benjamin et al., 2008; Couppé et al., 2009). Secondly, care was taken to have proper clamping of the tendon samples, using serrated clamps gripping on dried-out ends, yet some clamping effects cannot be ruled out. During our experiments, the tendon samples were loaded until rupture, after the cyclic tests, and the site of rupture was visually inspected. Samples showing slipping during the single-ramp tests or that ruptured near the clamp (3 out of 25) were discarded from the analysis. Finally, the patellar tendon samples included the patella, which certainly affects measured stiffness and makes comparison with other studies, which have only investigated the patellar ligament, difficult. Testing the whole patellar tendon complex was, however, considered the best approximation of the situation in vivo.

This study suggests that, theoretically, the gibbon Achilles tendon may contribute to reducing the locomotor cost of bipedalism and leaping by acting as an elastic spring. The patellar tendon seems a less likely candidate for storage and release of elastic strain energy during hind-limb-powered locomotion but may be more suitable for rapid force transfer and positional control, despite its low Young modulus. Full body dynamics of gibbon bipedalism and leaping should be addressed in future studies to confirm or refute the energy-saving role of the hind limb tendons in gibbon locomotion.

We acknowledge the National Museums Scotland (NMS) and Royal Zoological Society Antwerp (RZSA) for providing access to a unique and valuable collection of gibbon specimens. We also thank the student (Amal Alami) and supporting staff from the Small Veterinary Hospital of the University of Liverpool who helped with data collection, and the two anonymous reviewers for their constructive comments on the original manuscript. Special thanks go to Dr Robert Ker (University of Leeds), who gave valuable advice on tendon testing and data interpretation and gave us access to his testing facilities.

 
• COM

  centre of mass

 
• E

  Young's modulus

 
• ETL

  external tendon length

 
• FL

  muscle fascicle length

 
• H

  hysteresis

 
• ITL

  internal tendon length

 
• m

  muscle belly mass

 
• MA

  moment arm

 
• MTU

  muscle-tendon unit

 
• MTUL

  muscle-tendon unit length

 
• MVC

  maximum volumetric contraction

 
• PA

  pennation angle

 
• PCSA

  physiological cross-sectional area

 
• RC

  relative compliance

 
• S

  stiffness

 
• TCSA

  tendon cross-sectional area

 
• TL

  tendon length

 
• u

  energy absorption capacity

 
• V_t

  volume of tendon

 
• ε

  stress

 
• εl

  stress-in-life

 
• ρ

  density of tendon tissue

 
• σ

  strain