Functional Analysis of the Gibbon Foot During Terrestrial Bipedal Walking: Plantar Pressure Distributions and Three-Dimensional Ground Reaction Forces

Transcription

1 AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 128: (2005) Functional Analysis of the Gibbon Foot During Terrestrial Bipedal Walking: Plantar Pressure Distributions and Three-Dimensional Ground Reaction Forces Evie Vereecke, 1,2 * Kristiaan D Août, 1,2 Linda Van Elsacker, 2 Dirk De Clercq, 3 and Peter Aerts 1 1 Department of Biology, University of Antwerp, B-2610 Wilrijk, Belgium 2 Center for Research and Conservation, B-2018 Antwerp, Belgium 3 Movement and Sport Sciences, University of Ghent, B-9000 Ghent, Belgium KEY WORDS Hylobates lar; bipedalism; primate locomotion; biomechanics; kinetics ABSTRACT This paper gives a detailed analysis of bipedal walking in the white-handed gibbon, based on collected pressure and force data. These data were obtained from four gibbons in the Wild Animal Park, Planckendael, Belgium, by using a walkway with integrated force plate and pressure mat. This is the first study that collects and describes dynamic plantar pressure data of bipedally walking gibbons, and combines these with force plate data. The combination of these data with previously described roll-off patterns of gibbons, based on general observations, video images, force plates, and EMG data, gives us a detailed description of foot function during gibbon bipedalism. In addition, we compare the observed characteristics of hylobatid bipedalism with the main characteristics of bonobo and human bipedalism. We found that gibbons are midfoot/ heel plantigrade, and lack the typical heel-strike of other hominoids. The hallux is widely abducted and touches down at the onset of the stance phase, which results in an L-shaped course of the center of pressure. The vertical force curve is trapezoid to triangular in shape, with high peak values compared to humans. The braking component is shorter than the accelerating component, and shortens further at higher walking velocities. Speed has a significant influence on the forefoot peak pressures and on most of the defined gait parameters (e.g., vertical force peak), and it alters the foot contact pattern as well. The investigation of existing form-function relationships in nonhuman primates is essential for the interpretation of fossil remains, and might help us understand the evolution of habitual bipedal walking in hominids. Am J Phys Anthropol 128: , ' 2005 Wiley-Liss, Inc. The acquisition of habitual bipedalism was a key event in human evolution (Benton, 1997), but to date it remains unresolved as to how and when this particular gait type evolved in our early ancestors. Studying the mechanics of occasional bipedal walking in living apes can help reveal the acquisition of bipedal walking in (pre)hominids. In combination with the study of their functional morphology, this leads to the establishment of plausible form-function relationships of extant primates. This information is essential for the interpretation of (extinct) primate and prehominid fossil bones, and for the reconstruction of their locomotor behavior from these bony remains (Fleagle, 1999; Aerts et al., 2000; Schmitt, 2003b). So far, bipedal walking has been studied in several primate species: gibbon, chimpanzee, bonobo, baboon, spider monkey, and Japanese monkey, among others (e.g., Jenkins, 1972; Kimura et al., 1977; Stern and Susman, 1981; Yamazaki and Ishida, 1984; Ishida et al., 1985; Shapiro and Jungers, 1988; Aerts et al., 2000; D Août et al., 2002; Hirasaki and Kumakura, 2003; Vereecke et al., 2003; Wunderlich, 2003), and it was found that primates mainly adopt a compliant walking gait, in contrast with the relatively stiff-legged bipedalism of humans (Alexander, 1977; Schmitt, 1999; Schmitt, 2003). Moreover, bipedal walking appears to differ markedly between different primate species. This is related to differences in habitual posture and habitual locomotion of primates and to different degrees of arboreal or terrestrial adaptation, as reflected in their functional morphology (Kimura et al., 1977). However, despite the infrequent use of bipedalism and the lack of specific bipedal adaptations in their morphology, many primates seem to be reasonably good bipeds. With this study, we extend the prevailing information on bipedal locomotion of extant primates by offering a detailed analysis of the gibbon foot during terrestrial bipedalism. Gibbons were chosen as study species because they frequently engage in bipedal walking, and their bipedalism is particularly interesting for comparison with human bipedalism, since gibbons and humans belong to the same family of Hominoidea. Gibbons are widely known as the acrobatic apes that travel through the canopy by hand-over-hand swinging from branch to branch. And logically, this remarkable Grant sponsor: Fund for Scientific Research, Flanders, Belgium; Grant numbers: G *Correspondence to: Evie Vereecke, Laboratory for Functional Morphology, Department of Biology, University of Antwerp, Universiteitsplein 1, B-2610 Wilrijk (Antwerp), Belgium. Evie.Vereecke@ua.ac.be Received 4 November 2003; accepted 9 June DOI /ajpa Published online 28 April 2005 in Wiley InterScience ( # 2005 WILEY-LISS, INC.

2 660 E. VEREECKE ET AL. TABLE 1. Year of birth, sex, body weight, and number of analyzed sequences of different individuals Individual Year of Body birth Sex 1 weight (kg) Number of sequences Ge 1980 M Na 1983 F Ya 1997 M Be 2000? Total M, male; F, female;?, unknown. 2 This is sum of body weight of adult mother and her 4-monthold infant (accounting for approximately 10% of mother s body weight). form of progression has been the subject of numerous studies (e.g., Andrews and Groves, 1976; Jungers and Stern, 1984; Preuschoft and Demes, 1984; Yamazaki, 1990; Bertram and Chang, 1996, 2001; Chang et al., 1997, 2000; Bertram et al., 1999). Gibbons are, however, not only excellent brachiators, but are also capable of a wide variety of other locomotor activities, such as leaping, diving, climbing, galloping, and quadrupedal, tripedal, and bipedal walking (Keith, 1923; Avis, 1962; Carpenter, 1964; Ellefson, 1967; Tuttle, 1972; Baldwin and Teleki, 1976; Fleagle, 1976, 1999; Gittins, 1983). Ten to twelve percent of their locomotor repertoire consists of bipedal walking (Carpenter, 1964; Tuttle, 1972; Andrew and Groves, 1976; Fleagle, 1976, 1999; Cannon and Leighton, 1994), and gibbons are thus the most bipedal of the nonhuman primates (Tuttle, 1972; Ishida et al., 1984; Yamazaki, 1990; Schmid and Piaget, 1996). Since gibbons move and feed mainly in the middle and upper levels of the canopy and virtually never descend to the ground (Carpenter, 1964; Ellefson, 1967; Tuttle, 1972; Fleagle, 1999), their bipedalism is predominantly arboreal, but they also appear to be surprisingly adept terrestrial bipeds. Of course, gibbons are, above all, exceptional brachiators, but their bipedalism is remarkable because it is a kind of by-product of their brachiator morphology. Gibbons have a primitive anatomy and specialized limb proportions with long, powerful forelimbs and relatively long hindlimbs (Andrew and Groves, 1976; Fleagle, 1999). This is an adaptation for arm-swinging, but when traveling over the ground, these long forelimbs might incite upright, and hence bipedal, walking (personal observations). Most preceding research on gibbon bipedalism focuses on general kinematics and forces (Prost, 1967; Kimura et al., 1977, 1983; Okada et al., 1983; Ishida et al., 1984; Yamazaki and Ishida, 1984; Kimura, 1985) and on the electromyography of the hindlimb and back muscles (Ishida et al., 1978, 1985; Tuttle et al., 1978; Stern and Susman, 1981; Okada and Kondo, 1982; Shapiro et al., 1987; Shapiro and Jungers, 1988). Together with some ethological studies (e.g., Tuttle, 1972; Baldwin and Teleki, 1976; Fleagle, 1976), these investigations provide valuable information about the general mechanics of hylobatid bipedalism. It was found that gibbon bipedalism has characteristics differing from that of man, such as the slightly inclined trunk, the flexed hips and knees, the single-peaked vertical force curve (due to the downward movement of the body s center of gravity during midstance), the simultaneous activity of the hindlimb muscles, the large interstep variability, and the absence of an initial heel-strike (Schmid and Piaget, 1994). However, Fig. 1. Diagram of indoor gibbon setup in Wild Animal Park, Planckendael, Belgium. gibbons do make heel contact with the substrate, but after or simultaneous with midfoot contact, and they are classified as midfoot/heel plantigrade (Schmitt and Larson, 1995). Aside from this information, detailed data of gibbon foot function are lacking, and discussion continues about the amount of resemblance between hylobatid bipedalism and human bipedalism, compared to the bipedalism of other nonhuman primates. These observations are also mostly based on walking trials of one or two trained (H. agilis) subjects and incorporate only a limited number of sequences over a restricted range of velocities. In order to enhance our understanding of gibbon bipedalism, we collected additional force data on four nontrained lar gibbons, and more importantly, we also collected the first dynamic plantar pressure profiles of bipedal walking gibbons. These give us detailed information about pressure distribution during stance phase, and in combination with the force data, allow us to focus on dynamic foot function during the bipedal walking of gibbons. In addition, we compare the observed characteristics of gibbon bipedalism with the bipedal walking of bonobos (Pan paniscus) and humans. Bonobos are an interesting species for comparison, in view of their prehominid-like morphology and their close phylogenetic relationship with humans (McHenry and Corruccini, 1981; Aerts et al., 2000; D Août et al., 2002). Moreover, force and pressure data of bipedal walking bonobos are available from a comparable study on bonobo bipedalism (Vereecke et al., 2003). MATERIALS AND METHODS For this study, we collected data from four whitehanded gibbons (Hylobates lar) in the Wild Animal Park, Planckendael, Belgium, during February April 2003 (16 recording days, N ¼ 182; Table 1). Our setup was placed in the inside play hall of the gibbon house and consisted of a wooden platform of m with a built-in pressure mat ( m; Rsscan International) and a force plate ( m; AMTI) covered with a thin (nonskid) rubber sheet (Fig. 1). The pressure mat was connected with an Rsscan three-dimensional (3D) box and a Macintosh PowerBook G3 with Footscan mst software to record dynamic plantar pressure profiles at high temporal and spatial resolution (250 Hz sampling and 8,192 pressure sensors of cm 2 ). The summed pressures

3 GIBBON FOOT FUNCTION DURING BIPEDALISM 661 Fig. 2. Sign convention of 3D ground reaction forces on AMTI force plate. Direction of walking is from IN to OUT. are calibrated online with the ground reaction forces to allow an accurate pressure sampling. The force plate was connected to an amplifier, the 3D box, and the Macintosh computer to obtain 3D ground reaction forces during foot contact, i.e., transverse force (Fx), longitudinal force (Fy), and vertical force (Fz) (Fig. 2). A cage of latticework 1.20 m high and with an entrance of m was placed over the platform to oblige the gibbons to walk over the whole length of the platform. The cage (and platform) was aligned with a standard capturing cage with similar dimensions leading to the outer door. Each time a gibbon wanted to go outside or inside, it had to walk over the pressure mat and force plate. Four cameras (50 Hz) were positioned in a 3D arrangement (not aligned) and were synchronized together and with the pressure mat using four LEDs, which were connected with the outgoing sync of the 3D box. The video recordings were collected for future kinematical analysis of hylobatid bipedalism, but were also used in this study to check visually if the selected pressure profiles corresponded with steady bipedal sequences (see below). Before recording took place, all systems were calibrated. Since we wanted to collect spontaneous bipedal bouts, no direct interaction with the animals was attempted, and only strictly bipedal bouts (without hand assistance) were analyzed. Sequences with apparent accelerations, decelerations, jumps, gallops, and other aberrant gaits were excluded. Finally, we retained 182 bipedal sequences with contact times ranging from sec. The distribution of the plantar pressure profiles was analyzed with Footscan mst software (Rsscan International) and Microsoft Excel On each plantar pressure profile, 11 squares of 1.5 cm 2 (2 2 sensors) were placed, covering nearly the whole foot contact area (Fig. 3). These squares were defined on the basis of gibbon foot morphology and correspond with the foot regions that were defined for the analysis of human plantar pressure profiles (Wearing et al., 2001). The foot regions were finally combined into six regions to analyze pressure distribution under the foot: 1) the heel, near the calcaneus; 2) the lateral midfoot, near the metatarsal V base; 3) the medial midfoot, near the navicular bone; 4) the metatarsals, near the metatarso-phalangeal joints II V; 5) the lateral toes II V, near the distal phalanges; and 6) the hallux, i.e., the first metatarso-phalangeal joint and the first digit (Ha ¼ M1 þ T1). For every foot region, pressure was plotted as function of time, which was normalized against stance phase duration (total stance phase duration ¼ 100%). Averages and confidence intervals (CI ¼ X 1.96*SE) of these pressure-time plots were calculated for each individual over a limited range of velocities (contact time range, sec), all Fig. 3. Definition of foot regions (indicated by squares) on a Footscan plantar pressure profile (showing maximal pressures during stance phase). Note absence of heel loading. Fig. 4. Degree of hallucal abduction (HA): angle between first ray and third ray of foot. with qualitatively similar gait characteristics as observed on the corresponding video images. In addition, the onset and duration ( SE) of these regional plantar pressures were calculated for the entire test group. We had no information about foot contact area, so we chose to divide the pressure data by body weight to allow comparison of different-sized subjects. Maximal plantar pressures were calculated for each foot region and for each individual separately, to illustrate the actual magnitude and amount of inter- and intraindividual variation of regional peak pressures. Furthermore, the course of the center of pressure (COP), the degree of hallucal abduction, defined by the angle between the first and the third toe, and the foot progression angle, the angle formed by the functional foot axis (i.e., the third ray of the foot) to the direction of progression, were evaluated on the pressure profiles (Fig. 4). Together with the distribution and relative timing of pressures across the foot, these parameters give us useful information regarding foot posture and foot function during stance phase. The 3D ground reaction forces were also recorded with Footscan software and were analyzed in Microsoft Excel Plots were made of Fx, Fy, and Fz as a function of the stance phase duration. Calculation of average force-time plots for the total test group (four individuals; contact time range, sec) was obtained in a way similar to that described above for the pressure data. Forces were divided by body weight for standardization. To quantify and evaluate foot function during

4 662 E. VEREECKE ET AL. HA 1 (8) TABLE 2. Average (X) and standard error (SE) of all gait parameters during hylobatid bipedalism CT (sec) DF SD (sec) DB (sec) Fz max (BW) Fz max (%) Fy min (BW) Fy max (BW) Fy brake (%) X SE N HA, degree of hallucal abduction; CT, contact time; DF, duty factor; SD, stride duration; DB, initial or terminal double support; Fz max, value of Fz maximum (times body weight, BW) and timing of maximum (in % of stance phase duration); Fy max and Fy min, maximum and minimum of Fy, (times body weight, BW); Fy brake, duration of braking period (in % of stance phase duration); N, number of sequences. hylobatid bipedalism, we calculated some additional force parameters: the value and timing of the Fz maximum, the value of the Fy maximum and Fy minimum, and the duration of the braking (negative Fy) and propelling (positive Fy) period. Furthermore, some spatiotemporal parameters could be obtained from the force plate data, namely, the total stance phase duration, the initial and/or terminal double-support periods (the sum of which is the double-support phase), the stride duration, and the duty factor (i.e., stance phase duration/gait cycle). These parameters allow comparison with the gait characteristics of humans and other nonhuman primates. The Footscan device allows recording of multiple steps and automatically assigns the vertical force component to separate feet, based on the recorded pressures. Obviously, this is not possible for the longitudinal and transverse force components. Therefore, we selected only braking and propelling forces that occurred during single foot recordings (i.e., the first and/or last footstep of a walking trial; see also Fig. 10) to calculate a mean braking and propelling force component. The transverse force component was investigated qualitatively, but was not included in the detailed analysis. Since our devices not only record plantar pressures and 3D forces but also give stance phase durations, we can investigate the effect of velocity on different gait parameters. Stance phase duration (i.e., contact time) is strongly correlated with walking velocity, and contact time might thus be used as a reliable measure of velocity (Kimura et al., 1983; Vilensky and Gehlsen, 1984; Demes et al., 1994; Vereecke et al., in press). However, we found a significant effect of individual on contact time, and therefore we account for these interindividual differences when evaluating the speed effect. We do so by using an ANCOVA (general linear model procedure in SPSS 11.5 for Windows), in which we define individual as factor and contact time as covariate. Additionally, we tested the Pearson product moment correlation coefficient between contact time and certain gait parameters for which interaction effects (ind*ct) were absent, over a wide range of walking speeds (contact times of sec). In addition, we divided walking sequences of the subadult male into two speed categories: slow-moderate walking (contact times between sec) and fast walking (contact times between sec), to look for differences in gait characteristics between slow and fast walking within one individual (Student s t-test). As our sample had a rather heterogeneous composition, we also calculated the repeatability or intraclass correlation coefficient of the different parameters to describe the proportion of inter- rather than intraindividual variation. The repeatability (r) is based on variance components derived from a one-way ANOVA (for more information on this topic, see Lessells and Boag, 1987). If r is high (near 1), than the variation within and among individuals is significantly different, and the variation can be ascribed to individual differences. If r is low (near 0), then there is no difference between the variation within and among individuals, and there are no significant individual differences. Other statistical tests (e.g., Pearson product moment correlations, Student s t-tests) were also executed using SPSS 11.5 for Windows. The data are normally distributed (Kolmogorov-Smirnov test), so parametric tests are used. For all variables, mean values x standard error (SE) are given, unless stated otherwise (Table 2). RESULTS Plantar pressure distribution and roll-off pattern The foot progression angle is variable during bipedal walking, as is the degree of hallucal abduction. Most often the hallux is strongly abducted (x = , N = 68), but there are significant differences between individuals, and values range from (Table 4). When we account for these interindividual differences, we do not find a significant correlation between degree of hallucal abduction and stance phase duration (Pearson product moment correlation, P > 0.05; Table 4). The foot is plantar-flexed at the end of swing phase, and the plantar side is placed flat upon the substrate during stance phase, although inversion with curling of pedal digits II V occurs as well (see Tuttle, 1972). In contrast with other hominoids, there is no clear heel-strike during hylobatid walking. However, the heel might be loaded after contact of the hallux and/or metatarsals during midstance, which is the case in 80% of the bipedal sequences of the adult male. The medial and lateral parts of the midfoot are loaded simultaneously, with usually a higher loading of the lateral part. The metatarsal heads are loaded throughout stance phase, but peak pressures occur at around 70% of stance phase. This peak pressure is followed by a slight loading of the middle toes (digit II IV) at the end of stance phase. There is a low and more or less constant loading of the hallux throughout stance phase, with the highest pressure occurring at around 70% of stance phase (Fig. 5). The relative timing of different foot regions with their onset and duration, expressed as percentages of total stance phase, is given in Figure 6. The course of the center of pressure (COP) is typically L-shaped, running backward from the tip of the hallux to the central part of the midfoot, and then running forward to the middle toe. This is caused by a footfall pattern in which initial foot contact is made by the hallux and/or metatarsal heads, followed by loading of the mid-

5 GIBBON FOOT FUNCTION DURING BIPEDALISM 663 Fig. 5. Pressure-time plots during hylobatid bipedalism. For each foot region and each individual, average pressure (N/cm 2 kg) is depicted in function of percent of stance phase duration (%). A: Ge: N ¼ 35. B: Ya: N ¼ 49. C: Na: N ¼ 43. D: Be: N ¼ 23. see below), in foot progression angle, and in degree of hallucal abduction. Although the pressure distribution and footfall pattern show a lot of inter- and intraindividual variation (Tables 3 and 4; see also below), the general pattern during hylobatid bipedalism can be summarized as follows. After initial contact with the hallux and/or metatarsals, the weight is shifted from midfoot to the metatarsal heads to the middle toes. The hallux is loaded throughout stance phase, and heel loading is variable. 3D forces and spatiotemporal parameters Fig. 6. Relative timing of different foot regions expressed as percent of total stance phase duration. For each region, average onset and duration ( standard error) of loading are given (N ¼ 150, 4 different subjects). foot during midstance and a weight shift toward the metatarsal heads and toes during push-off (Fig. 7). This is the general pattern, but alternatively the course of the COP can also follow a straight line, running from the midfoot to the toes, or it can be loop-formed (Fig. 8). These different COP patterns may be caused by differences in walking speed (loop-formed at higher speeds; The shape of the vertical force curve is typically trapezoid to triangular during bipedal locomotion of gibbons. The vertical force peak is reached in the first half of stance phase and approximates 1.4 times the body weight during steady walking (Table 2). The sagittal force is logically divided into a (negative) braking period and a (positive) propelling period (Fig. 9). The timing of the shift from braking to propelling is variable, but occurs usually somewhat after the vertical force has reached its maximum (first half of stance phase), and so the propelling period is longer than or as long as the braking period. The maximum and minimum values of sagittal force are around 0.2 times the body weight (BW), but the (positive) propelling peak is often higher than the (negative) braking peak, especially during fast locomotion (see below). A lot of variation was observed in

6 664 E. VEREECKE ET AL. Fig. 7. Sequential pressure profiles of a (right) foot roll-over during bipedal walking of a gibbon. From left to right: initial contact of hallux (touchdown), contact of metatarsal heads, midstance with midfoot contact, and push-off (toe-off) through metatarsal heads and toes. Statistical analysis of all the data (ANCOVA) revealed a significant interaction effect of individual and contact time on some variables (duty factor, stride duration, value of vertical force maximum, value of sagittal force minimum and braking period duration, and metatarsal peak pressure; Table 4), which means that the relationship with contact time differs between individuals. When an interaction effect is absent, we can investigate the main effect of individual, and this yields significant values for degree of hallucal abduction, timing of vertical force maximum, and peak pressures under the lateral and medial midfoot, toes, and hallux (Table 4). Obviously, the interindividual variation in bipedal walking characteristics of gibbons is important. To quantify the amount of variation among and within individuals, we calculated the repeatability (Lessells and Boag, 1987). For all gait parameters, except for pressure under the hallux, it was found that interindividual variation is larger than intraindividual variation, so a main part of the apparent variation can be ascribed to individual differences. The specific values for each gait parameter are given in Table 4. Effect of velocity Fig. 8. Examples of three different types of COP courses. A: Typically L-shaped. B: V-shaped, with more adducted hallux. C: Loop-formed, with positive foot angle (occurring at higher velocities). Each plantar pressure profile is of a left foot; line denotes course of COP. the transverse component, with an average magnitude around 0.1 BW. But since we were unable to divide the overall transverse force component into separate footsteps (during multifoot contact), we cannot make a detailed analysis of the transverse force pattern during hylobatid locomotion. The spatiotemporal parameters are given in Table 2. From these data, we can conclude that gibbon bipedalism is a rapid mode of locomotion with short cycle duration, stance phase duration, relative stance phase duration (i.e., duty factor), and double-support phase (Fig. 10). The presented data are mean values of the four subjects and of all analyzed sequences, but we must mentioned that there are substantial inter- and intraindividual differences (see below). Inter- and intraindividual variation The statistical analysis shows a significant effect of walking velocity (or contact time) on several gait parameters, and we also observed speed-related changes in the foot roll-over pattern. The hallux does not make initial contact at higher velocities. This initial contact is often made with the lateral part of the foot, and there is a latero-medial roll-off through the metatarsal heads and toes. As a result, the course of the COP is no longer L-shaped, but more loop-formed (Fig. 8C). This is associated with a large (positive) foot progression angle, and at maximal speeds, we also observed sliding of the foot at the onset of stance phase. When interaction effects (ind * CT) are absent, we can investigate the main effect of contact time on the parameters. This yields significant values for the double stance phase period, for the timing of the vertical force maximum, and for the value of the sagittal force maximum (Table 4). For variables with a significant interaction effect, we calculated separate regressions between contact time and the variable for each individual, and found very mixed results. Duty factor, stride duration, braking period duration, value of the vertical force maximum, value of the sagittal force minimum, and peak pressures under the metatarsal, toe, and hallux regions showed a significant correlation with contact time, but not consistently for every individual (Table 4). To clarify these results, we divided the sequences of one individual into two speed categories (contact times, sec vs sec) and compared the gait characteristics of both categories (Fig. 11). Almost all gait parameters of the slow-moderate walking category differed significantly from the fast walking category (Student s t- test). The peak pressures under the metatarsal, toe, and hallux regions increased at higher walking velocities (P < 0.001), but the peak pressures under the lateral and medial midfoot remained unaffected (P > 0.05). The value of the vertical force peak increased with increasing velocity (and decreasing contact time; P < 0.001), and the instant of the vertical force peak was delayed (6% delay in fast vs. slow walking; P < 0.05), which was reflected in a more triangular-shaped force curve at higher velocities. The value of the propelling force peak increased with increasing velocity, and the duration of the propelling period was lengthened (9% longer in fast vs. slow walking; P < 0.001), but the value of the braking force trough remained unchanged. When walking faster, as reflected in shorter stance phase durations, the cycle duration and the double-support phase shortened accordingly, and the duty factor (or relative stance period) decreased (P < 0.05).

7 GIBBON FOOT FUNCTION DURING BIPEDALISM 665 TABLE 3. Average of maximal pressures SE (N/cm 2 ) under six different foot regions for each individual separately 1 Heel Lateral midfoot Medial midfoot Metatarsals Toes Hallux Ge Ya Na Be Total N ¼ 147. See also Figure 5. TABLE 4. Results of ANCOVA of individual (ind) and contact time (interaction effects and main effects) together with repeatability (r), calculated from mean square within (MSW) and mean square among (MSA) individuals 1 Ind * CT Ind CT MSW MSA r 2 HA (8) , CT (sec) 8, , DF * (2) SD (sec) * (4) , DB (sec) 2, , Fz max (BW) 1/(4) Fz max (%) Fy min (BW) (2) Fy max (BW) * Fy brake (%) (1) * Heel Lateral midfoot Medial midfoot Metatarsals ** (1) Toes (1) Hallux (1) Abbreviations as in Table 2., not significant; or 1/positive/negative correlation with CT within one individual (number of individuals for which correlation is significant). * P < ** P < P < DISCUSSION The foot contact pattern found in gibbons differs from that in bonobos and humans on two major points. First of all, there is the absence of a heel-strike, i.e., there is no contact of the heel at the onset of stance phase. When observing bipedally walking gibbons, Morton (1924, 1935) noticed that gibbons do not heel-strike, and this was confirmed repeatedly by other researchers (Ishida et al., 1984, 1985; Gebo, 1992; Tuttle et al., 1992, 1993). Moreover, Morton (1924, 1935) and Gebo (1992, 1993) pointed to the heel-elevated (or semiplantigrade) foot posture of gibbons, in which the heel never rests upon the ground during locomotion. This was, however, not confirmed by our study. Gibbons do differ from other hominoids (i.e., the great apes and humans) by the absence of an initial heel-strike, but they also differ from other semiplantigrade primates (e.g., baboons and macaques) because the heel does contact the substrate during midstance. Heel contact occurs simultaneously with contact of the midfoot region (Schmitt and Larson, 1995; personal observations) in all subjects, but true loading of the heel region was only observed in the heaviest subject (i.e., the adult male), with a similar loading as for the midfoot (personal observations). Possibly, high body mass causes the rear foot to sag, which increases the loading area under the foot and therefore decreases the average plantar pressures. Unfortunately, we are unable to test this hypothesis, because we cannot perform an accurate normalization of the pressure data. Thus we might only suggest that heel loading functions as a safety mechanism to avoid excessive loading of the foot, and potential injuries, during bipedal walking of heavier gibbons. Secondly, during midstance (and also during bipedal standing), the gibbon foot is positioned on the substrate with full contact of the foot sole and with slightly bent digits. There is only occasional curling of the pedal digits II V (Tuttle, 1972; personal observations), comparable to the inverted foot posture observed in bonobos (Vereecke et al., 2003). Moreover, Hirasaki and Kumakura (2003) investigated the foot kinematics of a lar gibbon, and found that gibbons have a relatively plantar flexed tarsometatarsal joint during bipedal walking. These observations led us to conclude that gibbons are plantigrade, like other hominoids, but due to the absence of an initial heel-strike, they should be defined as midfoot/heel plantigrade (Schmitt and Larson, 1995). Heel-strike is considered to be a synapomorphous character of great apes and humans that must have evolved in the hominoid line after the Hylobatidae split off. Another point that deserves attention is the initial contact of the hallux. This remarkable foot contact pattern has been observed frequently in our gibbon group, and occurs also occasionally in bonobos (in 5% of cases; personal observations), but we did not find any references to this pattern in the literature. The gibbon foot is held in a resting posture during swing phase, and is plantar-flexed at touchdown. The slightly pronated foot posture and relatively long hallux (Schultz, 1973), as an adaptation for arboreal life, might cause the initial touchdown of the hallux, followed by contact of the other foot regions. Another possibility, suggested by Crompton (personal communication), is that the initial contact of the hallux serves to scan the substrate and attune the foot posture to the structure of the surface. However, in view of the relatively high walking velocity, we doubt if

8 666 E. VEREECKE ET AL. Fig. 9. General plot of vertical and sagittal force components during gibbon bipedalism. For each component, average forcetime plot and corresponding CI are shown (N ¼ 182; 4 individuals). Fig. 10. Vertical (30 Hz filtered), sagittal, and transverse (both 15 Hz filtered) forces during three consecutive footsteps of adult female gibbon during steady bipedal walking. Solid bars represent stance times of left and right foot. there is time for really scanning the substrate. Nonetheless, it is possible that contact of the hallux with the substrate gives some sensory feedback, which helps to control foot posture and also increases the sense of balance. A further remarkable characteristic of the hallux is the high degree of abduction. Generally, the hallux is more abducted in gibbons than in bonobos, but variability is high. We observed a hallucal abduction ranging from , with marked differences between left and right foot and between different individuals. The reason for this variation is unclear, but could be related to the foot progression angle and walking direction. When walking bipedally, gibbons often walk obliquely, thereby using a kind of cross step and placing their feet parallel to each other, resulting in a positive (toe-out) and negative (toe-in) foot progression angle with, respectively, a more and less abducted hallux. An influence of walking velocity was also expected, but the correlation of hallucal abduction and walking velocity was not significant. In view of the relatively low and constant pressure under the hallux, stabilization or propulsion does not seem to be a primary function of the hallux during bipedal walking, but favors a more passive function of the hallux. The hallux might be acting as a tripod in the gibbon foot to increase the base of support during bipedal walking. This is unlike the situation in humans (and to a lesser extent in bonobos), where the (adducted) hallux has an important contribution in the generation of propulsive forces during push-off (Vereecke et al., 2003). Stance phase duration, cycle duration, and double-support phase are shorter during gibbon bipedalism than during bonobo and human bipedalism (Kimura et al., 1977, 1983; Vereecke et al., 2003). These findings correspond with observations of other researchers who pointed to the very short cycle duration of bipedally walking gibbons (Kimura et al., 1977, 1979), with short support phases (Okada et al., 1983; Shapiro and Jungers, 1988) compared to other apes. However, the relative stance phase duration (or duty factor) is similar in gibbons, bonobos, and humans (D ¼ ; Kimura et al., 1977, 1983; Vereecke et al., 2003). We observed large interindividual variation in spatiotemporal, but also in pressure- and force-related parameters. This might be related to differences in body proportions, age, and walking velocity. Statistical tests did not separate juvenile from (sub)adult gibbons (Student s Fig. 11. Force-time plot of vertical and sagittal force components during fast (CT ¼ sec, N ¼ 28) vs. slow to moderate bipedal walking (CT ¼ sec, N ¼ 32) for subadult male gibbon. t-test, P > 0.05), but it is not unlikely that there may exist more subtle, ontogenetic changes in gait characteristics. With the current data set, we were not able to perform a detailed analysis of age-related changes in bipedal walking, but this will be the subject of a future paper, which will entail a longitudinal study of the juvenile gibbon. Another source of variation might be the fact that the adult female is carrying an infant, but we did not observe apparent differences in gait between the adult female and the adult male that might be related to infant-carrying. But again, a more detailed investigation of the effect of carrying is needed to draw some conclusions. Finally, we included walking trials in this study, which covered a wide range of speeds and contained moderate acceleration and deceleration. Most probably, then, walking velocity is also an important cause of the inter- (and intra-) individual variation. During fast walking, the double-support phase is minimized and occasionally true running sequences (D < 0.5) are observed, but never over several consecutive strides. The average walking velocity of gibbons is high, but there seems to be an effect of age, with adult animals walking rather slowly, and juvenile and subadult animals walking faster. However, all animals are capable of walking at a

9 GIBBON FOOT FUNCTION DURING BIPEDALISM 667 wide range of velocities, and due to the adoption of a compliant gait, they can keep on walking at speeds where humans would already be running. Slow walking is a steady, controlled gait, whereas fast walking resembles human forefoot running, although aerial phases are rare. Thus, we expect that walking velocity has a substantial influence on at least some gait characteristics. This velocity effect was tested, and we found indeed that walking velocity has a significant effect on forefoot peak pressures and on several spatiotemporal and force-related parameters, although interindividual differences obscure these effects somewhat. If we look at the results of the subadult male (fast vs. slow walking), we find that all spatiotemporal and force-related parameters are affected by speed, except for the value of the braking force minimum. This is not surprising, since it was found that force patterns are dependent on velocity in human walking (Keller et al., 1994; Kimura et al., 1977), but thus far no study has found a consistent velocity effect for gibbon walking (Prost, 1976; Kimura et al., 1977, 1979). Kimura et al. (1983) only found a significant correlation between walking velocity and vertical force peak in gibbons, whereas Ishida et al. (1978, 1984) did not find a correlation between velocity and vertical force peak, but only with the propelling peak. We did use stance phase duration as a measure of velocity, but since these are strongly correlated (Kimura et al., 1983), this is probably not the reason for the different results. Other sources that may have resulted in the absence of an important velocity effect in previous studies are the lower number of analyzed sequences, the smaller range of walking velocities, the use of a different study species (H. agilis), and the fact that only one (trained) subject was included in the study. In addition, since we observed an increase of the propelling force peak and a lengthening of the propelling period at higher speeds in our study, we assume that the fastwalking sequences often entail (moderate) acceleration. It is thus not improbable that part of the observed speed effect might be an accelerating effect. Particularly for the characteristics of the braking and propelling force, this might generate results that are not in correspondence with results of standardized treadmill studies (see below). The kinetic parameters are largely in correspondence with the results of previous investigations of foot force during gibbon bipedalism (Ishida et al., 1976; Kimura et al., 1977, 1983; Okada et al., 1983). The vertical force curve is triangular-shaped, due to the lower position of the body s center of gravity during midstance, and this entails a relatively higher vertical force peak (1.4 BW) compared to bonobos, chimpanzees, and humans (1.0 BW; Kimura et al., 1977; Kimura, 1985; Winter, 1991). However, at lower walking velocity, the force curve becomes more trapezoidal and resembles the force curve of bonobos and chimpanzees. The typical double-humped pattern of humans was never observed, which confirms the absence of an inverted pendulum gait (Alexander, 1976) in hylobatid bipedalism. The sagittal force curve has a multipeaked braking component with high variability and a less variable single-peaked propelling component. The magnitudes of the braking and propelling peaks are similar to those in humans (0.2 BW; Winter, 1991), but larger than in chimpanzees (0.1 BW; Kimura et al., 1977). The relative braking period is shorter than the propelling period in gibbons (44% vs. 56%), as in other nonhuman primates (Kimura et al., 1977; Okada et al., 1983), but the braking peak (0.19 BW) is also lower than the propelling peak (0.26 BW). This is not in accordance with previous results (Kimura et al., 1977), and is somewhat strange because the impulses of both braking and propelling components should be more or less equal during steady walking. However, these are mean values from different steps and different animals, so the imbalance between braking and propelling impulse could be caused by high interstep and/or interindividual variability, or might be due to acceleration. Since we were unable to standardize our walking trials, it is not unlikely that unsteady trials were included in the analysis, which might be reflected in the braking and propelling impulses. Therefore, the possible confounding effect of acceleration must be kept in mind when looking at the results of sagittal force characteristics. When we look at the pressure distribution, we see that peak pressure under the metatarsal heads coincides with the propelling peak, whereas the braking peak concurs with the peak pressure under the midfoot (and heel) region. This indicates that the midfoot (and heel) region decelerates at touchdown, and that the metatarsal region generates a propulsive force at push-off. We also noticed a positive correlation between walking speed and both the forefoot peak pressures and the propelling force peak (see above), whereas there is no correlation with midfoot peak pressures and the braking force peak. So, in accordance with our visual observations, gibbons become forefoot runners at higher speeds, comparable to some human runners (20%; Kerr et al., 1983). But again, in view of the apparent correlation with the propelling force (and the absent correlation with the braking force), this is more likely an effect of acceleration than of velocity. We did not analyze the transverse force component, but when we looked at the force plate output, we found that both feet where either pushing toward each other or away from each other, and not consistently pushing laterally, as mentioned by others (Okada et al., 1983; Ishida et al., 1984). CONCLUSIONS Hylobatid bipedalism is a distinct walking gait, characterized by midfoot/heel plantigrady, a short braking period, a high vertical force peak, initial contact with the hallux, an L-shaped course of the COP, and a short cycle duration. But it also shares some characteristics with human bipedalism, e.g., the short double-support periods and speed-related changes, and with bonobo bipedalism, e.g., the trapezoid to triangular force curve and the marked hallucal abduction. The apparent differences in bipedal walking gait between the three hominoids support the view of an independent evolution of bipedalism in the three genera. Furthermore, we suggest that gibbons are not extreme specialists, but highly adaptable primates who are able to progress in diverse habitats in numerous ways. It is obvious that the functional morphology of gibbons is primarily adapted to brachiation, but it is interesting to note that these adaptations do not restrict gibbons to brachiation, but enable them also to perform a wide variety of other locomotor activities, including bipedalism. This has interesting connotations for the interpretation of primate and/or prehominid fossil remains, since this suggests that marked adaptations to one gait type do not exclude other, less frequently adopted gaits, and that alternative gait types can exist

10 668 E. VEREECKE ET AL. without the presence of marked morphological adaptations. Understanding the relationship between form and function and studying the bipedalism of extant nonhuman primates are essential for the interpretation of (pre)hominid fossil bones, and might help us gain insights into the evolution of habitual bipedal walking in hominids. ACKNOWLEDGMENTS This study was supported by a grant to E.V. and research project funding to P.A. (G ) from the Fund for Scientific Research, Flanders (Belgium). We thank the Flemish Government for structural support to the Center for Research and Conservation of the Royal Zoological Society of Antwerp. We are also grateful to the editor-in-chief and two anonymous reviewers for useful comments on the manuscript. Finally, we thank the very cooperative staff of the Wild Animal Park (Planckendael, Belgium). LITERATURE CITED Aerts P, Van Damme R, Van Elsacker L, Duchene V Spatio-temporal gait characteristics of hind limb cycles during voluntary bipedal and quadrupedal walking in bonobos (Pan paniscus). Am J Phys Anthropol 111: Alexander RM Mechanics of bipedal locomotion. In: Spencer-Davies P, editor. Perspectives in experimental biology, volume 1. Oxford: Pergamon. p Alexander RM Mechanics and scaling of terrestrial locomotion. In: Pedley TJ, editor. Scale effects in animal locomotion. London: Academic Press. p Andrew P, Groves CE Gibbon and brachiation. In: Rumbaugh DM, editor. Gibbon and siamang: a series of volumes on the lesser apes, volume 4: suspensory behavior, locomotion, and other behaviors of captive gibbons: cognition. Basel: Karger. p Avis V Brachiation. The crucial issue for man s ancestry. Southwest J Anthropol 18: Baldwin LA, Teleki G Patterns of gibbon behavior on Hall s Island, Bermuda: a preliminary ethogram for Hylobates lar. In: Rumbaugh DM, editor. Gibbon and siamang: a series of volumes on the lesser apes, volume 4: suspensory behavior, locomotion, and other behaviors of captive gibbons: cognition. Basel: Karger. p Benton MJ Vertebrate palaeontology. 2nd ed. London: Chapman and Hall. Bertram JEA, Chang YH Gait change in gibbon brachiation. Am Zool 36:20. Bertram JEA, Chang YH Mechanical energy oscillations of two brachiation gaits: measurement and simulation. Am J Phys Anthropol 115: Bertram JEA, Ruina A, Cannon CE, Hui Chang Y, Coleman MJ A point-mass model of gibbon locomotion. J Exp Biol 202: Cannon CH, Leighton M Comparative locomotor ecology of gibbons and macaques: selection of canopy elements for crossing gaps. Am J Phys Anthropol 93: Carpenter CR A field study in Siam of the behavior and social relations of the gibbon (Hylobates lar). In: Carpenter CR, editor. Naturalistic behavior of nonhuman primates. University Park: Pennsylvania State University Press. p Chang YH, Bertram JEA, Ruina A A dynamic force and moment analysis system for brachiation. J Exp Biol 200: Chang YH, Bertram JEA, Lee DV External forces and torques generated by the brachiating white-handed gibbon (Hylobates lar). Am J Phys Anthropol 113: D Août K, Aerts P, De Clercq D, De Meester K, Van Elsacker L Segment and joint angles of hind limb during bipedal and quadrupedal walking of the bonobo (Pan paniscus). Am J Phys Anthropol 119: Ellefson JO A natural history of gibbons in the Malay Peninsula. Unpublished Ph.D. thesis, University of California at Berkeley. Demes B, Larson SG, Stern JT, Jungers WL, Biknevicius AR, Schmitt D The kinetics of primate quadrupedalism: hindlimb drive reconsidered. J Hum Evol 26: Fleagle JG Locomotion and posture of the Malayan siamang and implications for hominid evolution. Folia Primatol (Basel) 26: Fleagle JG Apes and human. In: Fleagle JG, editor. Primate adaptation and evolution. New York: Academic Press. p Gebo DL Plantigrady and foot adaptation in African apes: implications for hominid origins. Am J Phys Anthropol 88: Gebo DL On plantigrady and quadrupedalism reply. Am J Phys Anthropol 91: Gittins PS Use of forest canopy by the agile gibbon. Folia Primatol (Basel) 40: Hirasaki E, Kumakura H Foot kinematics of Hylobates lar, Ateles geoffroyi, and Macaca fuscata during locomotion on arboreal and terrestrial substrates. Am J Phys Anthropol [Suppl] 36: Ishida H, Kimura T, Okada M, Yamazaki N Kinesiological aspects of bipedal walking in gibbons. In: Preuschoft H, Chivers DJ, Brockelman WY, Creel N, editors. The lesser apes: evolutionary and behavioral biology. Edinburgh: Edinburgh University Press. p Ishida H, Okada M, Tuttle RH, Kimura T Activities of hindlimb muscles in bipedal gibbons. In: Chivers DJ, Joysey KA, editors. Recent advances in primatology, volume 3: evolution. London: Academic Press. p Ishida H, Kumakura H, Kondo S Primate bipedalism and quadrupedalism: comparative electromyography. In: Kondo S, Ishida H, Kimura T, Okada M, Yamazaki N, Prost JH, editors. Primate morphophysiology, locomotor analyses and human bipedalism. Tokyo: University of Tokyo Press. p Jenkins FAJ Chimpanzee bipedalism: cineradiographic analysis and implications for the evolution of gait. Science 178: Jungers WL, Stern JT Kinesiological aspects of brachiation in lar gibbons. In: Preuschoft H, Chivers DJ, Brockelman WY, Creel N, editors. The lesser apes: evolutionary and behavioral biology. Edinburgh: Edinburgh University Press. p Keith A Man s posture: its evolution and disorders. Br Med J 1: Keller TS, Weisberger AM, Ray JL, Hasan SS, Shiavi RG, Spengler DM Relationship between vertical ground reaction force and speed during walking, slow jogging and running. Clin Biomech 11: Kerr BA, Beauchamp L, Fisher V, Neil R Footstrike patterns in distance running. In: Kerr BA, editor. Biomechanical aspects of sport shoes and playing surfaces: Proceedings of the International Symposium on Biomechanical Aspects of Sport Shoes and Playing Surfaces. Calgary: Calgary University Press. p 135. Kimura T Bipedal and quadrupedal walking of primates: comparative dynamics. In: Kondo S, Ishida H, Kimura T, Okada M, Yamazaki N Prost JH, editors. Primate morphophysiology, locomotor analyses and human bipedalism. Tokyo: University of Tokyo Press. p Kimura T, Okada M, Ishida H Dynamics of primate bipedalism as viewed from the force of the foot. Primates 18: Kimura T, Okada M, Ishida H Kinesiological characteristics of primate walking: its significance in human walking. In: Morbeck ME, Preuschoft H, Gomberg N, editors. Environment, behavior and morphology: dynamic interactions in primates. New York: Gustav Fischer. p Kimura T, Okada M, Yamazaki N, Ishida H Speed of the bipedal gaits of man and nonhuman primates. Ann Sci Nat Zool 5: Lessells CM, Boag PT Unrepeatable repeatabilities: a common mistake. Auk 104: