AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 124:248 –256 (2004) Energetic Costs of Bipedal and Quadrupedal Walking in Japanese Macaques M. Nakatsukasa,1* N. Ogihara,1 Y. Hamada,2 Y. Goto,1 M. Yamada,1 T. Hirakawa,1 and E. Hirasaki3 1 Laboratory of Physical Anthropology, Kyoto University, Kyoto 606-8502, Japan Primate Research Institute, Kyoto University, Aichi 484-8506, Japan 3 Department of Biological Anthropology, Osaka University, Osaka 565-0871, Japan 2 KEY WORDS locomotor energetics; bipedalism; Macaca fuscata; respiratory physiology ABSTRACT We investigated the energetic costs of quadrupedal and bipedal walking in two Japanese macaques. The subjects were engaged in traditional bipedal performance for years, and are extremely adept bipeds. The experiment was conducted in an airtight chamber with a gas analyzer. The subjects walked quadrupedally and bipedally at ﬁxed velocities (⬍5 km/hr) on a treadmill in the chamber for 2.5– 6 min. We estimated energy consumption from carbon dioxide (CO2) production. While walking bipedally, energetic expenditure increased by 30% relative to quadrupedalism in one subject, and by 20% in another younger subject. Energetic costs increased linearly with velocity in quadrupedalism and bipedalism, with bipedal/quadrupedal ratios remaining almost constant. Our experiments were relatively short in duration, and thus the observed locomotor costs may include presteady-state high values. However, there was no difference in experimental duration between bipedal and quadrupe- dal trials. Thus, the issue of steady state cannot cancel the difference in energetic costs. Furthermore, we observed that switching of locomotor mode (quadrupedalism to bipedalism) during a session resulted in a signiﬁcant increase of CO2 production. Taylor and Rowntree ( Science 179:186 –187) noted that the energetic costs for bipedal and quadrupedal walking were the same in chimpanzees and capuchin monkeys. Although the reason for this inconsistency is not clear, species-speciﬁc differences should be considered regarding bipedal locomotor energetics among nonhuman primates. Extra costs for bipedalism may not be great in these macaques. Indeed, it is known that suspensory locomotion in Ateles consumes 1.3–1.4 times as much energy relative to quadrupedal progression. This excess ratio surpasses the bipedal/quadrupedal energetic ratios in these macaques. Am J Phys Anthropol 124:248 –256, 2004. © 2004 Wiley-Liss, Inc. There is no living nonhuman primate in which bipedalism comprises a major positional component. This poses an inherent problem in the study of human bipedal adaptations through cross-species comparative methods. Living humans are highly specialized bipeds. Although anatomical and physiological studies on living humans documented sufﬁcient conditions for habitual bipedalism, it is difﬁcult to discuss how capable those animals which exhibit a less specialized condition are for bipedal behaviors. Specialized nonhuman bipeds, however, can be obtained under experimental conditions. Bipedal monkey attractions have been developed widely in Asian countries since ancient times. Japanese monkey performance, for example, has a history of more than 1,000 years, and is acknowledged as one of the most popular traditional entertainments in the country. Japanese macaques (Macaca fuscata) that engage in this traditional performance are trained to stand and walk bipedally for about 1 hr daily (Hayama et al., 1992; Iwamoto, 1985; Nakatsukasa and Hayama, 2003). These macaques even develop a human-like lumbar spinal curvature (Hayama et al., 1992; Preuschoft et al., 1988). According to a computer simulations, these trained monkeys are estimated to save one third of the energy expenditure for bipedal walking of ordinary experimental monkeys, owing to a more erect posture of the trunk (Ishida, 1991). These macaques are thus excellent “experimental” bipeds, and intriguing subjects for investigating the potential for bipedalism in nonhuman primates. We report here on their locomotor energetics during bipedal and quadrupedal walking. Whereas locomotor energetics is an important aspect of foraging behavior in animals (Steudel, 2000), experimental data have rarely been accumulated since Taylor et al. (1982). In particular, experimental studies detailing the energy consumption for bipedal locomotion in nonhuman primates are very few. Taylor and Rowntree (1973) found © 2004 WILEY-LISS, INC. Grant sponsor: JSPS; Grant number: Grant-in-Aid 12440245. *Correspondence to: Masato Nakatsukasa, Laboratory of Physical Anthropology, Graduate School of Science, Kyoto University, Sakyo, Kyoto 606-8502, Japan. E-mail: firstname.lastname@example.org Received 20 April 2002; accepted 22 May 2003. DOI 10.1002/ajpa.10352 Published online 3 November 2003 in Wiley InterScience (www. interscience.wiley.com). LOCOMOTOR ENERGETICS IN JAPANESE MACAQUES 249 for bipedal performance since 2 years of age. Experiments were conducted from March 2001–June 2002. Experimental procedures Fig. 1. Body weight of subjects during experimental period. No experiment was done on Subject 2 in November 2001. that the energetic costs of bipedal and quadrupedal running were the same in chimpanzees and capuchins. Since then, no comparable study has been done. Thus, this is the third study to compare locomotor energetics in bipedal and quadrupedal locomotion in nonhuman primates. MATERIALS AND METHODS Subjects The subjects were two male Japanese macaques, housed at the Suo Monkey Performance Association (Kumamoto Prefecture, Japan). Among several trained macaques, these two were best accustomed to walking on a treadmill. This is an important condition, because the subject is required to walk continuously for 2– 6 min on a treadmill with stable gait during the experiment (see below). There is no particular strain or lineage of macaque for Japanese monkey performance. In bipedal training done at the Suo Monkey Performance Association, no restricting implements are used. The initial step of training involves a bipedal stand exclusively, and not a walk (Hayama et al., 1992). The trainer pays particular attention to the degree to which the subject extends the hip, knee joints, and lumbar spine. Each training session continues for 15–20 min and is repeated several times a day. After a stable upright posture has been acquired, usually after a week to a month, a long-distance walk can be accomplished with no difﬁculty. The Suo Monkey Performance Association is operated under municipal regulations regarding animal welfare and treatment. Experiments were conducted with the collaboration of the Suo Monkey Performance Association, following the guidelines for animal experimentation of Kyoto University. Subject 1 was 9 years, 11 months in age at the commencement of the experiments and had a body weight of 11.6 –12.5 kg (Fig. 1). Subject 2 was 3 years, 11 months in age and had a body weight of 4.7– 6.2 kg (Fig. 1). Both subjects had been trained Since we could not accustom the subjects to a respirator gas mask, experiments were conducted with subjects being placed in an airtight chamber made of clear acrylic panels and an aluminum frame (1.9 ⫻ 1.9 ⫻ 1.9 m), with the seams being coated with silicone resin (Fig. 2). The chamber was ﬁxed on an acrylic ﬂoor sheet. The air-tightness of the chamber was tested by releasing CO2 (carbon dioxide) into the chamber (ca. 2,000 ppm) and leaving it overnight. A treadmill was carried into the chamber during experiments. The subjects do not walk without the accompaniment of a trainer. Therefore, a trainer remained within the chamber during the course of the experiments while breathing through a breathing tube extending from outside the chamber. This condition necessitated a large chamber size and was less conducive to the accurate determination of gas concentrations in the chamber. An infrared gas analyzer (Model CGT-7000, Shimadzu Corp.) was positioned in the chamber, and CO2 concentrations were measured. Metabolic cost is usually evaluated as a consumption of oxygen per time and body mass (O2 ml/sec/kg). Thus, the evaluation of respiratory quotients (RQ ⫽ exhaled CO2/O2 uptake) was necessary in order to determine true metabolic cost. We were obliged to measure the CO2 concentration because of the accuracy of the gas analyzer. Reﬂecting the gas composition of the air (O2 21 percent volume (vol%) vs. CO2 0.05 vol%), the accuracy of standard O2 sensors is much lower (minimum scale, 10⫺2 vol%) compared to that of standard CO2 sensors (10⫺4 vol%). This level of accuracy was not proper for recording changes in O2 concentrations in our experimental chamber. However, we tried to estimate RQ values in a later series of experiment (see below). Ten electronic fans were set to enhance the diffusion of expired gas. Eight fans were set in corners of the chamber, and two were in front of and behind the subject (Fig. 2). The air was sampled from three positions on the wall and ceiling (arrows in Fig. 2). Sampling tubes were set to avoid collecting expired breath of the subject directly. Air mixing was checked preliminarily with a human being: a human came into the chamber and stood near the position of the trainer (Fig. 2) for more than 15 min. Soon after the chamber was closed, the concentration of CO2 changed unstably. The inner air, however, then diffused evenly, and a constant increase in CO2 concentration (resulting from breathing) was observed after 20 sec. We conducted this test several times. A similar pattern was also observed in experiments on macaques. Consequently, we began to record CO2 concentrations at least 30 sec after the chamber was closed. The concentration of CO2 was recorded 250 M. NAKATSUKASA ET AL. Fig. 2. Experimental chamber (left) and scene of experiment with Subject 1 (right). Air was sampled at three positions on wall and ceiling (arrows), avoiding collection of expired breath directly. Fig. 3. Change of CO2 concentration (open circles; ppm) and O2 concentration (solid circles; vol%) against elapsed time (sec) during an experimental session on Subject 1 (bipedal walk at 2.5 km/hr). CO2 and O2 concentrations (y) are linearly correlated with elapsed time (x): y ⫽ 0.97x ⫹ 523.2 (r2 ⫽ 0.999) for CO2, and y ⫽ ⫺1.4 ⫻ 10⫺4x ⫹ 21.07 (r2 ⫽ 0.998) for O2. Along regression line of O2 concentration, RQ value between adjacent two plots is indicated. for at least 90 sec at intervals of 10 sec. Figures 3 and 4 show plots of CO2 concentration against elapsed time. Experimental duration was variable (90 –360 sec). Although there are some differences, mean values are about 2 min in both bipedal and quadrupedal sessions. Thus, the subjects walked about 2.5 min in an average session (30 sec before measuring, and about 2 min for measuring). Experiments were curtailed if the subject showed signs of excessive tiring or irritation. When a subject ceased stable walking during an experimental session, the data collected prior to cessation were used or discarded completely. For this reason, the duration of most experimental sessions was not very long. Each subject walked quadrupedally and bipedally at a certain velocity (1.0/1.5– 4.5 km/hr) on the treadmill within the chamber. Before the ﬁrst experimental session, a test walk (warm-up) was conducted for more than 5 min, and consecutive sessions were separated by a resting interval of several minutes. Walking velocity and mode of locomotion (quadrupedal or bipedal) were changed at random. Between sessions, the chamber was kept open. Air refreshing was smoothly done, as enhanced by the electronic fans. The least-squares linear regression (LSR) was calculated between CO2 concentration and elapsed time (Fig. 4). The regression was highly signiﬁcant in all sessions (r2 ⬎ 0.98). Taylor et al. (1982) noted that when an animal begins to tire, its energetic cost increases. Thus, a linear regression may not be ﬁtted under conditions of extreme fatigue. However, this was not the case in our experiments under medium endurance. The slope of the LSR was standardized using the subject’s body weight, and was subsequently employed as a measure of energy consumption (␦CO2 ppm/sec/kg). The consumption of O2 can be calculated as (␦CO2 ppm/sec/kg ⫻ chamber capacity)/RQ. The chamber capacity was 6,859 ⫻ 103 ml minus the total volume of the experimental apparatus, trainer, and subject. We ignored differences in body size of trainers and of subjects when the LSR slope divided by the subject body mass (␦CO2 ppm/sec/kg) was used as a convenient measure of energy consumption, since these size differences were negligible relative to chamber size. The pressure within the LOCOMOTOR ENERGETICS IN JAPANESE MACAQUES 251 Fig. 4. Increase of CO2 concentration (ppm) against elapsed time in (a) Subject 1 and (b) Subject 2. CO2 concentration was measured each 10 sec from onset of session, and least-squares regression was calculated. Open circles and dotted lines represent bipedal walk; solid circles and solid lines represent quadrupedal walk. chamber throughout each experimental session was assumed to be constant. RQ Our primary objective was a comparison of locomotor costs of bipedal and quadrupedal locomotion. Thus, converting CO2 production to O2 consumption was not essential for this study. The most important point here was to prove that RQ did not ﬂuctuate signiﬁcantly through all experimental sessions in each subject. If this assumption were to hold, the ratio of CO2 production rates (bipedalism to quadru- pedalism) would be almost equivalent to the relative metabolic cost of bipedalism in each subject. Some evidence support this assumption. The RQ is theoretically 1.0 under the aerobic combustion of glucose, and decreases if lipid or protein is included in the respiratory substrate. Additionally, part of the produced CO2 is stored in the muscle and blood, resulting in a transient lowering of the RQ. Figure 1 shows the body weights of subjects from March 2001–June 2002. Although body weight is basically stable in each subject, Subject 1 lost weight in June 2001. However, the body weight loss did not 252 M. NAKATSUKASA ET AL. TABLE 1. Respiratory quotients measured during bipedal walking1 RQ relative to each decrease of 10⫺2 O2 vol% Subject 1 2.5 km/hr 2.5 km/hr 3.5 km/hr 3.5 km/hr 3.5 km/hr 3.5 km/hr Average Subject 2 2.5 km/hr 2.5 km/hr 2.5 km/hr 3.5 km/hr 3.5 km/hr 3.5 km/hr Average 4th2 Sessionaveraged RQ 1st 2nd 3rd 0.73 0.68 0.98 0.70 0.61 0.61 0.72 0.79 0.71 0.66 0.74 0.70 0.52 0.70 0.63 0.60 0.82 0.80 0.69 0.71 0.71 0.82 0.75 0.66 0.57 0.703 0.63 0.53 0.73 0.58 0.59 0.70 0.63 0.57 0.76 0.79 0.57 0.69 0.74 0.69 0.66 0.51 0.63 0.54 0.61 0.80 0.63 0.62 0.60 0.72 0.56 0.63 0.75 0.653 0.83 0.55 1 RQs were calculated as increase of CO2 concentration while O2 concentration decreases by 10⫺2 vol%. 2 Duration of session: approximately 4 –5 min in Subject 1, and 4 – 6 min in Subject 2. 3 Mean of session-averaged RQs. Fig. 5. Mass-speciﬁc CO2 production during 3.5-km/hr bipedal and quadrupedal walk in Subject 1 (␦CO2ppm/sec/kg ⫻ 10⫺2). Solid diamonds, July 2001; open diamonds, June 2002. Although Subject 1 lost weight in July 2001, no effect was observed. affect mass-speciﬁc CO2 production by locomotor exercise (Fig. 5). Thus, it is unlikely that body fat or muscles were particularly consumed as the respiratory substrate during this period. Since these subjects did not experience marked fat gain and loss, it is possible to disregard body fat as a major respiratory substrate. These macaques are fed combinations of carbohydrate-rich food items (potato, beans, maize, barely, and rice) with supplementary fruits and vegetables. Although no menu was recorded during the experiments, it is reasonable to predict that animals fed such balanced diets would not have extreme RQ values deviating from the normal range (probably within 0.8 – 0.9). During the latter series of experiments, we introduced an O2 sensor and measured both O2 and CO2 concentrations simultaneously in relatively long sessions (Table 1). Calculated RQs were quite variable, not only between sessions but also within session. However, they showed correlation with neither walking velocity nor duration. For two reasons, we interpret the variability of RQ values as being caused by the relatively low accuracy (or delay of response) of the O2 sensor rather than the variability of true RQ itself; the variability of RQ is minor. Firstly, CO2 production rates (B/Q ratios as well) converged in a narrow range in each experimental condition (Tables 2 and 3). If RQ is really as variable as it appeared in Table 1 (e.g., 0.52– 0.98 in Subject 1 walking at a velocity of 3.5 km/hr), such constant results are unexpected. Secondly, CO2 production rates increase linearly as walking velocity increases in each subject (Fig. 6). This regularity cannot be expected if RQ values are highly variable. Figure 3 shows O2 consumption during a long session. O2 concentration seems to linearly decrease through the session. However, calculated RQ values vary from 0.6 – 0.83, even in this single session. The average RQ was 0.7 in Subject 1 and 0.65 in Subject 2 (Table 1). These values are extremely low, particularly in Subject 2. It is impossible to give a clear idea about the reasons for this. Possible causes are a respiratory substrate involving more lipids and CO2 storing in the body. However, the ﬁrst explanation is not sufﬁcient for such extremely low RQ values. Even if lipid was exclusively used as the respiratory substrate, it should not be less than 0.7. In addition, it is unlikely that lipid is preferentially broken down as the respiratory substrate during relatively short-term exercise. Also, we do not think it is probable based on the diet of the subjects. The second explanation is also unclear. If this is the case, RQ values will rise with long exercise duration. Such a tendency was not observed (Table 1). Since these low values are not explainable, we did not use these RQ values to obtain O2 consumption in our subjects. RESULTS Table 2 and Figure 6a summarize the results of experiments on Subject 1. Thirty-nine and 43 sessions were done for bipedal and quadrupedal walking, respectively. Subject 1 walked with velocities of 1.5– 4.5 km/hr. In both quadrupedal and bipedal walking, %␦CO2 is correlated linearly with walking velocities. The LSR formula between velocities and mean %␦CO2 is: y ⫽ 1.961x ⫹ 2.156 (r2 ⫽ 0.957) in bipedal walking, and y ⫽ 1.504x ⫹ 1.582 (r2 ⫽ 0.927) 253 LOCOMOTOR ENERGETICS IN JAPANESE MACAQUES TABLE 2. CO2 concentration increase rates (ppm/sec/kg ⫻ 10⫺2) in bipedal and quadrupedal walking by subject 1 Walking velocity Bipedalism Mean SD n Range Mean duration of session (sec) Quadrupedalism Mean SD n Range Mean duration of session (sec) B/Q ratio1 1 1.5 km/hr 2 km/hr 2.5 km/hr 3 km/hr 3.5 km/hr 4 km/hr 4.5 km/hr 5.37 0.09 5 5.26–5.46 120 6.48 0.54 6 5.72–6.94 103 6.55 0.56 5 5.99–7.37 126 7.60 0.70 9 6.91–8.82 113 9.06 0.13 5 8.91–9.16 154 9.62 0.71 6 8.85–10.72 111 11.59 0.59 3 10.91–11.97 116 4.05 0.54 5 3.59–4.93 120 4.87 0.62 6 3.91–5.69 127 5.10 0.36 4 4.60–5.45 120 5.67 0.55 10 4.97–6.92 110 6.93 0.76 5 5.76–7.85 148 6.98 0.73 10 5.88–7.72 100 9.05 0.56 3 8.41–9.46 127 1.33 1.33 1.29 1.34 1.31 1.38 1.28 Bipedal/quadrupedal ratio. TABLE 3. CO2 concentration increase rates (ppm/sec/kg ⫻ 10⫺2) in bipedal and quadrupedal walking by subject 2 Walking velocity Bipedalism Mean SD n Range Mean duration of session (sec) Quadrupedalism Mean SD n Range Mean duration of session (sec) B/Q ratio1 1 1 km/hr 1.5 km/hr 2 km/hr 2.5 km/hr 3 km/hr 3.5 km/hr 4 km/hr 4.5 km/hr 7.15 0.89 4 5.84–7.78 98 6.08 0.22 4 5.84–6.37 123 8.65 0.77 4 7.90–9.72 100 7.94 0.42 6 7.31–8.37 148 9.66 0.72 9 8.51–10.59 123 10.61 0.80 6 9.66–11.95 132 11.56 0.59 5 10.89–12.36 102 12.04 1.31 2 11.11–12.96 120 5.98 0.11 3 5.85–6.08 120 6.28 0.50 5 5.73–6.87 102 6.45 0.72 4 6.07–7.59 140 7.75 0.53 6 6.77–8.22 100 8.50 1.08 4 7.80–10.10 147 9.61 0.25 3 9.43–9.78 110 1.02 1.38 1.23 1.25 1.25 1.20 10.01 1.24 2 9.19–10.88 140 1.20 Bipedal/quadrupedal ratio. in quadrupedal walking. Slopes are not signiﬁcantly different between these LSR trajectories (P ⫽ 0.852). The bipedal/quadrupedal ratio of %␦CO2 is rather constant, ranging from 1.28 –1.38, with an average of 1.32. Table 3 and Figure 6b summarize the results of experiments on Subject 2. Forty and 27 sessions were conducted for bipedal and quadrupedal walking, respectively. Subject 2 walked with velocities of 1.0 – 4.5 km/hr. No data could be gathered on quadrupedal walking at a velocity of 1.0 km/hr. Although %␦CO2 is generally correlated with velocities for both quadrupedal and bipedal walking, the variation is large for bipedal walking at low velocities. This result probably reﬂects that these velocities (ⱕ1.5 km/hr) were too low for this subject to accomplish stable walking (Hirasaki, unpublished data). At a velocity of 2.5 km/hr, %␦CO2 drops markedly in both bipedal and quadrupedal walking. Gaits at walking velocities faster than 2.0 km/hr were fairly stable (unpublished data); the reason for this is unclear. At higher velocities, %␦CO2 increases linearly, and bipedal/quadrupedal ratios of %␦CO2 are fairly constant, at around 1.2. LSR lines are calculated as: y ⫽ 1.65x ⫹ 4.68 (r2 ⫽ 0.901) for bipedal walking, and y ⫽ 1.486x ⫹ 3.34 (r2 ⫽ 0.960) for quadrupedal walking. There is no signiﬁcant difference of slopes between these LSR trajectories (P ⫽ 0.1079). Although the trials were much less abundant, we conducted a further experiment. Subject 1 walked quadrupedally at a certain velocity for about 2 min, and then switched locomotor mode to bipedalism, and continued to walk at the same velocity. Figure 7 indicates a change of CO2 concentration through the session. The walking velocities were 1.5, 3.0, and 4.0 km/hr. After the locomotor mode was switched to bipedalism, CO2 production signiﬁcantly increased in all trials (shown in the slope of the LSR line). The B/Q ratio in each velocity is 1.15 (1.5 km/hr), 1.19 (3.0 km), and 1.17 (4.0 km). Although all of these values are rather lower than the B/Q ratio calculated from averaged values in multiple sessions (Table 2), this result clearly proves that a bipedal walk costs signiﬁcantly more than a quadrupedal walk. DISCUSSION The experimental conditions in this study were different from those in previous studies. For example, subjects in Taylor et al. (1982) were intended to run with greater endurance (15–30 min) than would normally be required in nature. Running velocities in their experiments reached as high as 11 km/hr. 254 M. NAKATSUKASA ET AL. Fig. 7. CO2 concentration (ppm) increase in quadrupedal and bipedal walking by Subject 1 (walking velocities: 1.5, 3.0, and 4.0 km/hr). Locomotor mode was switched from quadrupedalism to bipedalism at middle of session (indicated by arrow). After switching to bipedalism, CO2 production signiﬁcantly increases in all trials (shown in slope of LSR line). LSR slopes and 95% conﬁdent limits are shown. Fig. 6. Standardized CO2 concentration increase rates (ppm/ sec/kg ⫻ 10⫺2) in bipedal and quadrupedal walking with different velocities (km/hr) in Subject 1 (a) and Subject 2 (b). Mean ⫾ 1 standard deviation. Diagonal lines are LSR trajectories: y ⫽ 1.961x ⫹ 2.156 (r2 ⫽ 0.957) in bipedal walking, and y ⫽ 1.504x ⫹ 1.582 (r2 ⫽ 0.927) in quadrupedal walking in Subject 1; y ⫽ 1.65x ⫹ 4.68 (r2 ⫽ 0.901) for bipedal walking and y ⫽ 1.486x ⫹ 3.34 (r2 ⫽ 0.960) for quadrupedal walking in Subject 2. Their experimental conditions reﬂected the different purpose of their study. They aimed to compare locomotor energetic physiology in diverse warm-blooded animals. On the other hand, we focused on a more restricted comparison: the difference in energetic cost for bipedalism and quadrupedalism in the same individual. Because our experiments were carried out with relatively short durations, critics may question the rigorousness of our estimates of energetic costs. It is generally thought that at least 4 min is necessary for an animal to reach a steady state from onset of exercise. If we measured presteady-state energetic cost, observed values would be higher than those at a steady state (Taylor et al., 1982). In several sessions with a relatively long duration (e.g., Fig. 3), CO2 production rates, however, appear almost constant through the session (ⱖ5 min), and there is no tendency for them to decrease in the later part of a session. This probably indicates that the exercise in our subjects was mostly supplied by the aerobic system, and that contributions from the anaerobic system, if any, were negligible. The relatively short experimental duration suggests that subjects did not yet reach a steady state. However, we have measured energetic costs for bipedalism and quadrupedalism under the same conditions (including costs for start-up). Thus, the comparisons should be relevant in evaluating the relative energetic costs of bipedal walkng in each subject. Since Taylor et al. (1982), comparative locomotor energetics in nonhuman primates have rarely been investigated. In particular, studies detailing meta- LOCOMOTOR ENERGETICS IN JAPANESE MACAQUES bolic costs associated with nonhuman primate bipedal walking are limited. Unfortunately, the methodological problems in this study, and particularly the uncertainty of true RQ, pose difﬁculties in comparisons with previous studies. However, if we credit the assumption that ﬂuctuation of RQ is negligible, some comparison is possible. It is known that massspeciﬁc O2 consumption is linearly correlated with walking/running velocities (Taylor et al., 1982). This is also the case in our study. Taylor et al. (1982) published energetic costs of quadrupedal walking in an 8.5-kg hamadryas baboon and a 5.1-kg stumpedtailed macaque. If a seemingly appropriate RQ is given (e.g., 0.8 – 0.9), the observed mass-speciﬁc CO2 production in our subjects yields similar values of mass-speciﬁc O2 consumption as in these equivalent-sized monkeys.1 This may give credibility to our estimates of locomotor energetics. Despite several methodological problems in our experiments, Figure 7 indicates higher energetic costs in bipedal walking, without a doubt. The only remaining uncertainty is the degree of difference. The B/Q ratios in the single-session experiments are lower than those calculated from averaged data in multiple sessions (Tables 2 and 3). Since the number of trials is only three, it is not clear if this difference is biologically meaningful or produced by chance, and (if the former is the case) what causes this difference. More trials are needed, and similar experiments with the reverse sequence (bipedalism to quadrupedalism) are also necessary. Taylor and Rowntree (1973) noted that the energetic costs for bipedal and quadrupedal walking were the same for chimpanzees and capuchins at velocities of 1.5–5 km/hr. In our results, however, the energetic cost for bipedalism was higher by 20 –30% than that of quadrupedalism in both subjects, regardless of walking velocity (1.5– 4.5 km/ hr). Despite the chance that we included data at presteady state, this does not explain the contradicting results, because the experimental conditions (e.g., walking duration) were the same in bipedal and quadrupedal walking. One exception was Subject 2, at a velocity of 1.5 km/hr (Fig. 4b), which is too slow for comfortable voluntary walking in Japanese macaques. Why have these two studies produced different results? Since the energetic cost for quadrupedalism in chimpanzees is very high compared to the general mammalian standard (⫹36%; Taylor et al., 1982), the “relatively low” energetic cost of bipedalism might not be surprising for chimpanzees. However, the same explanation is not possible for capuchins, which are more efﬁcient quadrupeds (only 10% extra 1 Mass-speciﬁc CO2 production rates can be converted to massspeciﬁc O2 consumption (mlO2/sec/kg) as follows: (CO2 values given in Tables 2 and 3) ⫻ 0.0676/RQ. For example, mass-speciﬁc O2 consumption in Subject 2 walking at a velocity of 3.5 km/hr (⫽ 0.97 m/sec) is 0.68 (when RQ is 0.85). This value almost coincides with data of a 5.1-kg stumped-tailed macaque in Figure 1C in Taylor et al. (1982). 255 cost; Taylor et al., 1982). There are a few possible explanations. One is that the relatively cheap cost of bipedalism in capuchins can be accounted for by their smaller body size (on average, 3.34 kg), as the excess energetic cost in bipedalism was lower in the younger Subject 2 than in Subject 1 (Tables 2 and 3). However, Subject 2 was 4.2 kg at the beginning of the experiment, and the size difference for the capuchins was less than 1 kg. Thus, this explanation is weak. Another explanation might be that Japanese macaques are particularly inefﬁcient bipeds. Biomechanical analysis by Yamazaki (1985) revealed that ordinary experimental Japanese macaques have relatively low potential abilities for bipedal walking when compared with gibbons, chimpanzees, and spider monkeys. Undoubtedly, these trained macaques are extremely adept walkers compared with most other nonhuman primates. However, genetically determined anatomical features might entail more extra energetic costs than in capuchins. This seems likely, because high muscular activities would be required in macaques to maintain hindlimb joints in ﬂexed positions and to control excursions of the center of gravity (Hirasaki et al., 2002), even if the forelimbs are almost free from body support and driving. In capuchins, the tails might have partly contributed to balancing the trunk. It would be necessary to examine the kinematics and/or kinetics of capuchin walking to discuss this possibility. Whatever the reason may be, it must be remembered that comparative studies of this kind are extremely few. Caution should thus be exercised in generalizing from the arguments of Taylor and Rowntree (1973) to other primates. However, is the energetic cost of bipedal walk really great for these trained macaques? When Ateles travel by suspensory locomotion, energy consumption is 30 – 40% greater than in quadrupedalism (at 2 km/hr; Parsons and Taylor, 1977). This excess ratio is even greater than the B/Q ratio in trained macaques. Bipedal walking cost at 2 km/hr was as much as quadrupedal walking cost at 3.0 km/hr in Subject 1, and bipedal walking at 2.5 km/hr cost as much as quadrupedal walking at 3.6 km/hr (Fig. 3). In Subject 2, the bipedal walking cost at 2 km/hr was equivalent to the quadrupedal walking cost at 3.2 km/hr, and bipedal walking at 2.5 km/hr cost as much as quadrupedal walking at 3.8 km/hr. Extra locomotor costs at this amount might be compensated by beneﬁts obtained through bipedalism. We observed two skeletons of trained monkeys which died of acute disease (13 and 9 years in age). Apart from a thickened cortex and somewhat enlarged articular surfaces (Nakatsukasa et al., 1995; Nakatsukasa and Hayama, 2003), these skeletons did not exhibit severe degenerative articular changes or vertebral body deformations. This observation suggests that stresses to the skeletal system introduced by bipedal behaviors with medium en- 256 M. NAKATSUKASA ET AL. durance were at a tolerable level throughout their lives.2 The locomotor kinematics of Subjects 1 and 2 are currently being analyzed and will be published elsewhere. Hirasaki et al. (2002) preliminarily revealed several kinematic features of bipedal walking which may improve locomotor efﬁciency in these trained macaques. These trained monkeys exhibit less frequent but longer strides than in untrained macaques. The hip-joint trajectory follows an upward convex curve during the support phase, resembling the inverted pendulum movement in human bipedalism. Angular movements of the head and trunk are reduced compared to those of ordinary macaques. These characteristics are common among trained macaques, despite individual variations. Thus, it is intriguing to evaluate how much the bipedal energetic cost is reduced in these macaques relative to the cost in ordinary macaques. Unfortunately, we have not yet succeeded in measuring the locomotor energetic cost of ordinary macaques in the same experimental setup, because those animals could not walk with constant gaits for as long as 2 min. Heart pulse rates, however, are a good measure of energy expenditure (Rose and Gamble, 1994), and can be more easily monitored. We are attempting to compare locomotor energetic by doing so. We have observed that bipedal gait becomes more stable through training in trained macaques. Longitudinal studies as well as studies of individual variation of locomotor energetics and kinematics/kinetics will reveal how much energy is saved in trained macaques in relation to different gait and postural patterns, and may shed some light on the adaptive processes of bipedalism in the earliest hominids. ACKNOWLEDGMENTS We are very grateful to the trainers and macaques of the Suo Monkey Performance Association for their 2 We observed a severe pathological deformation of the hip joint in a wild subadult Japanese macaque which congenitally lacked both forearms and had necessarily adopted bipedalism for traveling (Nakatsukasa, personal observations). This case suggests that a complete transition to bipedalism introduces too much stress to the skeletal system of the Japanese macaque. The kinematics of bipedal walking in this macaque were more ordinarily monkey-like than in a trained macaques (Ogihara, unpublished ﬁndings). enormous patience in carrying out the repetitive experiments required of this investigation. We thank K. Steudel-Numbers and other reviewers for thoughtful comments on our manuscript, and M.D. Rose for English-language editing of the manuscript. Many thanks go to H. Sato, J. Domoto, E. Ishizaki, A. Hidaka, and Shimadzu Medical Systems Corp. for kind advice and support in setting up the experimental system. LITERATURE CITED Hayama S, Nakatsukasa M, Kunimatsu Y. 1992. Monkey performance: the development of bipedalism in trained Japanese monkeys. Acta Anat Nippon 67:169 –185. Hirasaki E, Ogihara N, Hamada Y, Nakatsukasa M. 2002. Kinematics of bipedal locomotion in bipedally-trained Japanese macaques (monkey performance monkeys). Am J Phys Anthropol [Suppl] 34:85 [abstract]. Ishida H. 1991. A strategy for long distance walking in the earliest hominids: effect of posture on energy expenditure during bipedal walking. In: Coppens Y, Senut B, editors. Origine(s) de la bipédie chez les Hominidés. Paris: CNRS. p 9 –15. Iwamoto M. 1985. Bipedal of Japanese monkeys and carrying models of hominization. In: Kondo S, editor. Primate morphophysiology, locomotor analyses and human bipedalism. Tokyo: University of Tokyo Press. p 251–260. Nakatsukasa M, Hayama S. 2003. Skeletal response to bipedalism in macaques: with emphasis on cortical bone distribution of the femur. Cour Forschungsinst Senckenberg. 243:35– 45. Nakatsukasa M, Hayama S, Preuschoft H. 1995. Postcranial skeleton of a macaque trained for bipedal standing and walking and implications for functional adaptation. Folia Primatol (Basel) 64:1–29. Parsons PE, Taylor CR. 1977. Energetics of brachiation versus walking: a comparison of a suspended and an inverted pendulum mechanism. Physiol Zool 50:182–188. Preuschoft H, Hayama S, Günther MM. 1988. Curvature of the lumbar spine as a consequence of mechanical necessities in Japanese macaques trained for bipedalism. Folia Primatol (Basel) 50:42–58. Rose J, Gamble JG. 1994. Human walking. Baltimore: Williams and Wilkins. Steudel K. 2000. The physiology and energetics of movement: effects on individuals and groups. In: Boinski S, Garber PA, editors. On the move: how and why animals travel in groups. Chicago: University of Chicago Press. p 9 –23. Taylor CR, Rowntree VJ. 1973. Running on two or on four legs: which consumes more energy? Science 179:186 –187. Taylor CR, Heglund NC, Maloiy GMO. 1982. Energetics and mechanics of terrestrial locomotion. J Exp Biol 97:1–21. Yamazaki N. 1985. Primate bipedal walking: computer simulation. In: Kondo S, editor. Primate morphophysiology, locomotor analyses and human bipedalism. Tokyo: University of Tokyo Press. p 105–130.