Relationship between efficiency and pedal rate in cycling: significanceof internal power and muscle fiber type composition

E. A. Hansen1, G. Sjøgaard2,3

1Department of Physical Performance, The Norwegian School of Sport Sciences, Oslo, Norway, 2Department of Physiology, NationalInstitute of Occupational Health, Copenhagen, Denmark, 3Institute of Sports Science and Clinical Biomechanics, University ofSouthern Denmark, Odense, DenmarkCorresponding author: Ernst Albin Hansen, Department of Physical Performance, The Norwegian School of Sport Sciences,Postbox 4014, Ulleval Stadion, 0806 Oslo, Norway. Tel: 147 23 26 23 26, Fax: 147 22 23 42 20, E-mail: Ernst.albin.hansen@nih.no

Accepted for publication 31 May 2006

Cycling was performed to test the following two hypotheses:(1) muscular efficiency is unrelated to pedal rate (61, 88, and115 r.p.m.) for a group of subjects with a wide range of slowtwitch (ST) fibers in spite of decreasing whole-body effi-ciency and (2) muscular efficiency correlates positively with%STmuscle fibers, and this correlation is more pronouncedat low pedal rates than at high pedal rates.

Whole-body gross efficiency decreased from 20–22% at61 r.p.m. to 15–18% at 115 r.p.m.Mean muscular efficiencyfor all subjects (n5 16) was � 26%, with delta efficiencybeing constant and muscular efficiency (taking internalpower into account) slightly increasing with pedal rate.

Muscular efficiency correlated positively (R2 5 0.25) with% ST fibers (21–97% ST in m. vastus lateralis) at115 r.p.m. while not at 61 and 88 r.p.m.

In conclusion, the decrease in whole-body gross efficiencywith increasing pedal rate was not explained by a decrease inmuscular efficiency, and supported increase in internalpower to account for the increase in metabolic power withincreasing pedal rate. Furthermore, differences betweenindividuals in muscle fiber type composition affected mus-cular efficiency, which correlated positively with % STmuscle fibers during fast pedalling.

Contraction velocity, or rate, and muscle fiber typecomposition are variables known to affect the rate ofenergy turnover in single skeletal muscle fibers,including those from humans (Bottinelli & Reggiani,2000). Accordingly, both these variables may influ-ence the efficiency of whole muscles contractingin vivo. Cycling has been used to investigate muscularefficiency during human voluntary muscle activitybecause this type of exercise allows measurement ofexternal power (EP) on the bicycle as well as meta-bolic power (MP), the latter being determined fromthe rate of oxygen uptake at submaximal exerciselevels. EP encompasses the power generated by legmuscles to maintain or increase cycling speed, e.g., toovercome aerodynamic drag, rolling resistance, fric-tion resistance, and, if necessary, ascents and accel-erations. The influence of contraction rate and fibertype composition, as well as their combined effect,can be elucidated by studying subjects with diversemuscle fiber type compositions performing cyclingat various pedal rates.Regarding the effect of contraction rate on effi-

ciency, it has been shown that during cycling at aconstant EP, oxygen uptake (Coast & Welch, 1985;Marsh & Martin, 1993; Takaishi et al., 1996) and its

derived MP (Gaesser & Brooks, 1975; Kang et al.,2004) increase with increasing pedal rate for pedalrates above 50–80 r.p.m., depending on intensity(Foss & Hallen, 2004; Nielsen et al., 2004). Thisincreased MP, per definition, implies a decrease inwhole-body gross efficiency. However, it does notimply a corresponding decrease in muscular effi-ciency with increasing pedal rate, and the literaturein fact reveals contradictory findings. Some previousstudies have shown that muscular efficiency, esti-mated as delta efficiency (a given change in EPdivided by the corresponding change in MP), in-creased with increasing pedal rate (Asmussen, 1952;Boning et al., 1984; Sidossis et al., 1992; Chavarren &Calbet, 1999; Martin et al., 2002). However, anotherstudy found that delta efficiency decreased withincreasing pedal rate (Gaesser & Brooks, 1975), whileyet another study presented data fitting a U-shapedcurve (Francescato et al., 1995). In addition, muscu-lar efficiency across a wide range of contraction ratesduring knee-extension exercise was unaffected bycontraction rate (Sjøgaard et al., 2002). The lattermay occur for groups of subjects that are hetero-geneous with regard to muscle fiber type compositionbecause there may be reciprocal changes in the

Scand J Med Sci Sports 2007: 17: 408–414 Copyright & 2006 The Authors

Journal compilation & 2006 Blackwell MunksgaardPrinted in Singapore .All rights reservedDOI: 10.1111/j.1600-0838.2006.00580.x

408

efficiency or different changes in contribution topower of the two fiber types. Various methods andprocedures for estimation of muscular efficiency mayalso contribute to the inconsistent findings. In theknee-extension study, muscular efficiency was esti-mated as (EP1IP) MP� 1, where IP is the internalpower that represents the energy changes of movingbody segments. The rationale for including IP is thatduring knee extension or cycling, the muscles mustovercome both EP and IP, the sum of which is thetotal power (TP). IP has been estimated for a numberof exercise forms such as walking and running(Winter, 1979; Willems et al., 1995) as well as forsimple knee extension (Sjøgaard et al., 2002). Re-cently a quantification and evaluation of IP duringcycling was performed applying various modelsbased on kinematics (Hansen et al., 2004). Thisevaluation supported that IP during cycling couldbe estimated using the same model as that appliedfor knee extension exercise (Winter, 1979; Sjøgaardet al., 2002), although an even better estimationcould be attained by applying a modified version ofanother established model for walking and running(Willems et al., 1995). Muscular efficiency duringcycling, taking into account IP from evaluated mod-els based on kinematics, has not been reportedpreviously, and may shed new light on the contro-versies regarding the relationship between pedal rateand muscular efficiency.The effect of human muscle fiber type composition

on muscular efficiency also needs further clarifica-tion, particularly when complicated by variations incontraction rate. An elegant single fiber studyshowed peak efficiencies to be similar between slowtwitch (ST) and fast twitch (FT) fibers, but foundthat peak efficiency occurred at a higher contractionvelocity in FT compared with ST muscle fibers (Heet al., 2000). Still, caution must be exercised whengeneralizing this knowledge to voluntary muscleactivity, particularly because temperature was quitelow (12 1C), stimulation was maximal, and musclesamples from only two subjects were included. Thereis disagreement among studies that have investigatedthe influence of muscle fiber type composition onmuscular efficiency during cycling. Two studies re-ported a positive correlation between % ST musclefibers and efficiency at 80 r.p.m. (Coyle et al., 1992;Mogensen et al., 2006) while another study reportedno significant correlation at 70 r.p.m. (Mallory et al.,2002). Interestingly, one study reported that subjectswith high % ST muscle fibers executed work with alower delta efficiency at 100 r.p.m. than at 60 r.p.m.,while subjects with low % ST muscle fibers had ageneral lower delta efficiency and showed no suchrelation (Suzuki, 1979). Thus, the relationship be-tween muscular efficiency and fiber type compositionmay be affected by pedal rate. Experimental in vivo

data on this issue are largely missing and the presentstudy contributes to this discussion with such newdata. The hypotheses of the present study are basedon detailed theoretical considerations regarding theinteraction between muscle fiber type compositionand pedal rate to modulate muscular efficiency (Sar-geant & Beelen, 1993).The aim of the present study was to test the follow-

ing hypotheses for the voluntary muscle activityof cycling: (1) muscular efficiency, taking IP intoaccount, is unrelated to pedal rate for a group ofsubjects with a wide range of muscle fiber type com-position and (2) muscular efficiency correlates posi-tively with % ST muscle fibers, and this correlation ismore pronounced at low than at high pedal rates.

MethodsSubjects

Sixteen males (25.8 � 0.87 years, 177 � 1.27 cm, and73.7 � 2.05 kg) volunteered for the study following written-informed consent. The subjects performed cycling for sportand/or recreation and were all highly accustomed to cycling.The study was approved by the local ethics committee. Thesubjects had a needle biopsy taken from m. vastus lateralis fordetermination of % type I myosin heavy chain (MHC) iso-form (% MHC I), which in this study is taken to represent %ST muscle fibers. A previous analysis showed a correlationcoefficient of 0.96 between % MHC I and % Type I (histo-chemistry analysis; Hansen et al., 2002). All details regardingthe biopsy procedure and the subsequent sodium dodecylsulfate polyacrylamide gel electrophoresis (SDS-PAGE) ana-lysis have been reported previously (Hansen et al., 2002).Mean % MHC I for all subjects was 62.6 � 4.50%.

Exercise sessions

Each subject conducted three sessions on separate days, over aperiod of 5 weeks. The first session (I) conducted wasergometer cycling for determination of maximal oxygen up-take and identification of low and moderate submaximal EPcorresponding to 40% and 70% of the EP at which maximaloxygen uptake was attained. The second session (II) was apilot study for practising non-supported cycling on a tread-mill. The third session (III) was treadmill cycling at low andmoderate EP at a range of pedal rates. The sessions have beendescribed in detail previously (Hansen et al., 2002) but, insummary, session I was performed at 80 r.p.m. on a Monarkcycle ergometer (Monark AB, Varberg, Sweden) mountedwith an SRM crank dynamometer (Schoberer Rad Messtech-nik, Julich, Germany), and sessions II and III were performedon a 16-speed racing bicycle placed on a motorized WoodwayELG 70 treadmill (Woodway GmbH, Weil am Rhein, Ger-many). Session III consisted of a 5min warm-up at 100W EPfollowed by 10 cycling bouts, each lasting 5min and all beingseparated by 5min of rest. Six of the 10 bouts were performedat low and moderate EP with preset pedal rates of 61, 88, and115 r.p.m., in randomized order. The other bouts were per-formed with freely chosen pedal rate; data from those boutshave been reported separately (Hansen et al., 2002; Nielsen etal., 2004). EP, pedal rate, rate of oxygen uptake, and respira-tory exchange ratio were determined during the last 2min ofeach bout. EP and pedal rate were measured with the SRMcrank dynamometer while rate of oxygen uptake and respira-tory exchange ratio were determined using the Douglas bag

Muscular efficiency during cycling

409

technique as previously described (Franch et al., 1998). Thetwo EP levels were set sufficiently low to ensure predominantlyaerobic energy turnover in all cycling bouts. MP was thencalculated from the rate of oxygen uptake and respiratoryexchange ratio according to Coyle et al. (1992). MP wascalculated from the gross rate of oxygen uptake while netMP was estimated by subtracting a basal metabolic rateof 83W, corresponding to 250mLO2/min, from the MP.This basal metabolic rate in general agrees with textbookvalues and corresponds to the group mean of the individualestimates (ranging from 0.230 to 0.291mLO2/min) determi-ned by an equation taking into account body mass and height(Mifflin et al., 1990).

Kinematics

In order to quantify IP according to a model based onkinematics, round reflective markers were attached to thesubjects’ skin, clothes, or shoe at anatomical landmarksprimarily on the right side of the body, as previously describedin detail (Hansen et al., 2004). Two video cameras werealigned to record in the frontal and sagittal planes. Betweenthese two, a third camera was aligned so that each markercould be recorded by at least two cameras. The cameras weregenlocked and recorded, at 50Hz, the motion of the markersfor 30 s during the last 2min of each cycling bout. Five pedalrevolutions were digitized using Peak Motus 2000 (PeakPerformance Technologies Inc., Englewood, CO, USA). Thecoordinates were low-pass filtered with either second-, third-,or fourth-order Butterworth and either 2, 3, or 4Hz cutofffrequency. From the feet, shanks, thighs, hands, lower arms,upper arms, as well as the head, and trunk segments, a modelwas constructed as a linked system with 13 segments. Mass,center of mass, as well as gyration radius for each segmentwere taken from the literature (Winter, 1990, pp. 56–57). A 3Dposition file with the x, y, and z coordinates was providedas input to the software program ERGILA developed inMATLAB (The MathWorks Inc., Natic, MA, USA).

A physiological evaluation of a number of models based onkinematics suggests that IP during cycling is most accuratelyestimated as IPWillems-COM (Hansen et al., 2004). This IPestimate was calculated from an established model as thekinetic energy changes of the body segments, including energytransfer only between segments of the same limb (Willems etal., 1995), and being modified by disregarding the powerassociated with the potential and kinetic energy changes ofthe center of mass of the whole body (Hansen et al., 2004). Asreported in the evaluation paper, IP increased with an in-creased pedal rate. Accordingly, IPWillems-COM was 14 � 0.60,42 � 1.60, and 91 � 3.95W at low EP and 14 � 0.53,44 � 1.90, and 99 � 4.30W at moderate EP, at 61, 88, and115 r.p.m., respectively (Hansen et al., 2004). Significantdifferences between the two EP levels were found forIPWillems-COM at 88 and 115 r.p.m. All of the above IP valueswere used for calculation of muscular efficiency in the presentstudy.

Efficiency calculations

Muscular efficiency was calculated as (EP1IP)/(net MP).Delta efficiency was calculated as the difference in EP betweenbouts at low and moderate submaximal EP divided by thecorresponding difference in MP. Of note is that delta effi-ciency, due to the way in which it is calculated, does not allowcalculation of separate efficiencies for the two EP levels.Whole-body gross efficiency was calculated as EP/MP.

Statistics

All data are presented as mean � SE, unless otherwiseindicated. Paired t-test, ANOVA (repeated measures andfactorial), and Fisher’s protected least significant differencetest (post hoc analysis following significant ANOVA) werecalculated in Statview 5.0 (SAS Institute Inc., Cary, NC,USA). Regressions were calculated in Excel 2000 (MicrosoftCorporation, Redmond, WA, USA). Po0.05 was consideredstatistically significant.

ResultsOxygen uptake and EP

Maximal oxygen uptake was 4.52 � 0.131L/min,corresponding to 61.6 � 2.04mL/min/kg. The calcu-lated target values for low- and moderate EP were151 � 4.5 and 264 � 7.9W, respectively, with similaroverall mean values attained during the tests:152 � 2.7 and 266 � 4.8W, respectively, with nosignificant differences between the pedal rates of 61,88, and 115 r.p.m. Rate of oxygen uptake increasedwith increasing pedal rate at both EP levels (Po0.05;Table 1). When the physiological responses of rateof oxygen uptake and MP were plotted vs TP(5EP1IPWillems-COM), these variables were closelyrelated, independent of the combination of EP andpedal rate as demonstrated by R240.99. This rela-tionship was significant for each of the two linearregression lines (Fig. 1). The reciprocal of the slope(1/3.7653) in Fig. 1(b) was 26.6%, representing‘‘overall delta efficiency.’’

Efficiency vs pedal rate

Muscular efficiency calculated with IPWillems-COM

was not affected by EP level (Table 2). This allowedus to calculate mean muscular efficiency across thetwo EP levels to reduce data to only one value persubject for each pedal rate, which gave the mostrobust data for the subsequent part of the analysis.This showed that the pedal rate had a slight – yetsignificant – effect on muscular efficiency. A post hoctest showed that muscular efficiency was lower at61 r.p.m. (24.9 � 0.35%) than at 88 (26.7 � 0.33%)and 115 r.p.m. (26.8 � 0.56%) (corresponding to anapproximately 0.1� 0.2% point increase per 1 r.p.m.

Table 1. Oxygen uptake (l/min) during the different cycling bouts

61 r.p.m. 88 r.p.m. 115 r.p.m.

Low EP 2.26 � 0.052 2.37 � 0.046 2.94 � 0.076Moderate EP 3.50 � 0.098 3.61 � 0.082 4.08 � 0.093

Data are mean � SE.

Significant effect of pedal rate at both low and moderate EP.

Cycling bouts were performed at 61, 88, and 115 r.p.m. at low and

moderate external power (EP) corresponding to 40% and 70% of the EP

at which maximal oxygen uptake was attained.

Hansen & Sjøgaard

410

pedal rate, Po0.05). Mean muscular efficiency, forall cycling bouts, was 26.1 � 0.26%, and thus similarto mean delta efficiency (25.9 � 0.42%). Delta effi-ciency was unaffected by pedal rate.Gross efficiency values were lower than the values

of muscular efficiency and delta efficiency. Further,this whole-body efficiency measurement was higherat moderate compared with low EP and decreasedwith pedal rate (Po0.05; Table 2).

Efficiency vs muscle fiber type composition

Muscular efficiency calculated with IPWillems-COM

(mean of the two EP levels) correlated significantlywith % MHC I during cycling at 115 r.p.m. (Fig. 2).R2-values and slopes were lower and not significantfor the two lower pedal rates (61 r.p.m.: y5

0.0422x122.214, R2 5 0.1733; 88 r.p.m.: y5 0.0305x124.771, R2 5 0.099), but still indicated a weak trendof positive relationship between % MHC I andmuscular efficiency.

Discussion

A main finding of the present study was that mus-cular efficiency, in terms of delta efficiency, wasindependent of pedal rate across a wide range ofpedal rates typically used in cycling, when consider-ing a mixed group of subjects with a wide range ofmuscle fiber type compositions. This finding supportsour first hypothesis. Still, interestingly, muscularefficiency increased slightly with increasing pedalrate. The present study also showed that human

0

1

2

3

4

5

0 100 200 300 400Total power (W)

0 100 200 300 400Total power (W)

Rat

e of

oxy

gen

upta

ke (

l min

−1)

y = 3.7653x + 99.874 R2 = 0.995

0

400

800

1200

1600

Met

abol

ic p

ower

(W

)

low EP

moderate EP

low EP

moderate EP

y = 0.0107x + 0.3415 R2 = 0.992

(a)

(b)

Fig. 1. (a) Rate of oxygen uptake and (b) metabolic power(MP) as a function of total power (TP). Data are given asmean � SE. Symbols may exceed SE bars. TP is calculatedas the sum of external power (EP) and internal power(IPWillems-COM). Cycling bouts were performed at 61, 88,and 115 r.p.m. at low andmoderate EP corresponding to 40%and 70% of the EP at which maximal oxygen uptake wasattained. The data points on the ordinate axes represent thebasal metabolic rate (250mL O2/min corresponding to 83W)and are included in the computed regression lines. Po0.05.

Table 2. Various calculations of efficiency (%) during the different

cycling bouts

61 r.p.m. 88 r.p.m. 115 r.p.m.

Delta efficiency 25.0 � 0.83 26.5 � 0.59 26.3 � 0.67Muscular efficiencyw

Low EP 24.8 � 0.50 26.6 � 0.49 26.6 � 1.00Moderate EP 24.9 � 0.48 26.8 � 0.42 26.9 � 0.52

Whole body gross efficiency*,w

Low EP 20.2 � 0.44 18.5 � 0.38 15.2 � 0.55Moderate EP 21.9 � 0.45 21.4 � 0.34 18.3 � 0.43

Muscular efficiency was calculated with IPWillems-COM as estimate of

internal power (IP).

Data are mean � SE.*Significant effect of EP level.wSignificant effect of pedal rate.

EP, external power.

y = 0.0823x + 21.613 R2 = 0.249

20

22

24

26

28

30

32

34

36

0 20 40 60 80 100% MHC I

Mus

cula

r ef

ficie

ncy

(%)

Fig. 2. Relationship between % myosin heavy chain (MHC)I and muscular efficiency at 115 r.p.m. calculated withIPWillems-COM as an estimate of internal power (IP) andaveraged between low and moderate external power (EP)corresponding to 40% and 70% of the EP at which maximaloxygen uptake was attained. Po0.05.

Muscular efficiency during cycling

411

muscles with high % ST muscle fibers performedsubmaximal cycling exercise more efficiently duringfast pedalling (115 r.p.m.) than those with low % STmuscle fibers, supporting our second hypothesis.However, our expectation of this being more pro-nounced at low pedal rates than at high pedal rateswas not confirmed. On the contrary, % ST musclefibers and muscular efficiency were unrelated at 61and 88 r.p.m. Finally, the present results revealedthat the well-established decrease in whole-bodygross efficiency with increasing pedal rate was notdue to a decrease in the efficiency of the contractingmuscles, which, on the contrary, may have actuallyincreased. Rather, the decrease in gross efficiencywith increasing pedal rate was due to a correspond-ing increase in IP. Therefore, based on whole-bodygross efficiency, conclusions cannot directly be drawnregarding muscular efficiency.

Significance of pedal rate and muscle fiber typecomposition

The efficiency of the working muscles increasedslightly (muscular efficiency) or remained constant(delta efficiency) with increasing pedal rate. Oneinterpretation of this may be that the contractionrates used in the present study refer to the ascendinglimb or the flat part of the inverted U-shapedrelationship between efficiency and contraction velo-city of muscle fibers (Sidossis et al., 1992; Martin etal., 2002). However, it should also be considered thatthe relationship between efficiency of the workingmuscles and pedal rate reflects the summed effect ofcontraction velocity and rate, muscle fiber typecomposition, and muscle fiber type recruitment pat-tern. The latter may, in turn, be affected by pedalforce production, which is a hyperbolic function ofpedal rate. Accordingly, it is possible that cycling atsubmaximal EP at around 60 r.p.m. results in amixed muscle fiber type recruitment pattern of bothactive ST and FT muscle fibers. Further, it might bethat an increase in pedal rate, which decreases pedalforce (Lollgen et al., 1980; Boning et al., 1984;Takaishi et al., 1998; Hansen et al., 2002) andincreases TP via an increase in IP, for some subjectsresults in more ST and less FT muscle fiber recruit-ment. Indeed, a previous report found that FTmuscle fibers were less glycogen depleted after cyclingat 100 compared with 50 r.p.m. at 80% of maximaloxygen uptake (Ahlquist et al., 1992). For subjectswith high % ST fibers, such a shift in fiber recruit-ment toward the more efficient ST muscle fiber type,in addition to the theoretically increased efficiency ofeach muscle fiber with increasing pedal rate, wouldexplain an increase in muscular efficiency with in-creasing pedal rate. Subjects with low % ST fibershad probably recruited all their ST and some FT

muscle fibers at 60 r.p.m. When they increased thepedal rate, the decrease in force might have acted toreduce FT fiber recruitment while the increase in TPmight have acted to increase the total number ofactive muscle fibers. Such an increase in muscle fiberactivation would necessarily involve recruitment ofFT muscle fibers if all ST muscle fibers were alreadyrecruited at the lower pedal rate. The sum effectcould be a relatively larger involvement of FT musclefibers at high pedal rates compared with low pedalrates. In addition, efficiency of the working musclescould be lower or unchanged since the additionalrecruitment of the less efficient FT muscle fiber typemay then counteract the theoretical increase in effi-ciency for each muscle fiber with increased contrac-tion velocity. Based on these considerations, it ispossible that previous studies showing increasedmuscular efficiency with increased pedal/contractionrate (Asmussen, 1952; Boning et al., 1984; Sidossiset al., 1992; Chavarren & Calbet, 1999; Martin et al.,2002) tested subjects (often endurance-trained roadcyclists) who were homogeneous with regard tomuscle fiber type composition (Sjøgaard, 1984).This is not the case, however, in studies (Fergusonet al., 2002; Sjøgaard et al., 2002) where the subjectswere not specifically trained (or naturally selected viatheir sport) and would supposedly be more hetero-geneous with regard to muscle fiber type composi-tion, as in the present study. Future studies applyingimproved techniques for determination of musclefiber type recruitment pattern may investigate thesepropositions.Muscular efficiency correlated positively with

% ST muscle fibers at 115 r.p.m. Other studies havealso shown human muscles with high rather than low% ST muscle fibers to execute work more efficientlyduring cycling (Coyle et al., 1992; Mogensen et al.,2006). Thus, even that some variation exists forrepeated measures of muscle fiber type composition(Elder et al., 1982), and that the vastus lateralismuscle is only one, although a major, power gen-erator during pedalling (Neptune et al., 2000), it maybe possible to observe a positive correlation between% ST muscle fibers in m. vastus lateralis and effi-ciency during cycling. Therefore, it is likely thatmuscle fiber type composition, via muscular effi-ciency, accounts for some of the intersubject varia-tion in whole-body efficiencies observed in thepresent study, as well as for some of the differencesbetween published values. Our expectation of a morepronounced relationship between muscular efficiencyand % ST muscle fibers at low pedal rates comparedwith at high pedal rates was contrasted ratherthan satisfied. Perhaps the suggestions made aboveregarding fiber type recruitment pattern may alsoexplain this; e.g. a shift toward more active STmuscle fibers for subjects with high % ST fibers at

Hansen & Sjøgaard

412

increasing pedal rate in contrast to additional re-cruitment of FT fibers for subjects with low % STfibers,in other words a clearer expression in terms offiber recruitment and muscular efficiency of inter-subject differences in muscle fiber type compositionwith increased pedal rate.

Muscular efficiency estimates

Muscular efficiency, which takes into account bothbasal metabolic rate and the estimated power re-quired to move the body segments, is relevant forinvestigating the efficiency of the working muscle, asis delta efficiency. When comparing these two esti-mates of efficiency of the working muscles, it shouldbe noted that muscular efficiency is a particularlyattractive study parameter as long as the IP estimatesare valid. The reason is that it can be determinedfrom a single exercise bout while determination ofdelta efficiency requires at least two exercise bouts atdifferent EP levels. Further, delta efficiency cannotaccount for potential changes in IP with, e.g., EP.It follows then that muscular efficiency, as opposedto delta efficiency, can reveal possible differencesbetween two different specific exercise intensities.The mean muscular efficiency across all cycling

bouts was 26%, which was concordant with the otherindependent muscular efficiency estimate, delta effi-ciency, also being 26% across pedal rates. Deltaefficiency has previously been suggested as the mostvalid in vivomeasure of muscular efficiency (Whipp &Wasserman, 1969; Poole et al., 1992). The magni-tudes of the muscular efficiency estimates in thepresent study were well within the theoretical max-imum of 30%, which was determined based on aknowledge of phosphorylative coupling efficiencyand contraction coupling efficiency (Whipp & Was-serman, 1969; Stainbsy et al., 1980).As anaerobic metabolism was not accounted for,

an overestimation of muscular efficiency may haveoccurred. Correcting muscular efficiency for the sameanaerobic contribution for all subjects would prob-ably cause the level to decrease slightly. Still, it islikely that subjects with high % ST muscle fibers hada lower anaerobic contribution at the same relativeexercise intensity than subjects with lower % STmuscle fibers. If this is so, then the difference inmuscular efficiency between subjects would havebeen even larger than that shown in Fig. 2. In short,including the anaerobic contribution would lower theoverall level of efficiency to some extent. However,our findings of a positive correlation between mus-cular efficiency and % ST muscle fibers would belargely unaffected or possibly strengthened by acorrection for anaerobic contribution.

In summary, when considering subjects with awide range of muscle fiber type composition as awhole, the efficiency of the working muscles duringsubmaximal cycling was about 26% and remainedconstant (delta efficiency) or increased only slightly(muscular efficiency) with increasing pedal rate.Whole-body gross efficiency decreased with increas-ing pedal rate, which was not explained by a decreasein muscular efficiency but supports the likelihoodthat IP accounts for the increase in MP with increas-ing pedal rate. Differences between individuals inmuscle fiber type composition affected muscularefficiency, which correlated positively with % STmuscle fibers at 115 r.p.m. while not at 61 and88 r.p.m.

Perspectives

Various measures of efficiency, as addressed in thepresent study, are complementary in our understand-ing of optimizing human muscle function. Thus,muscular efficiency, taking IP into account, may bea useful tool for adding to our knowledge of effi-ciency of the working muscles during voluntarymuscle activity such as cycling. From a whole-bodyperspective, though, it may be MP that matters mostduring prolonged activity where breakdown of thelimited energy stores ought to be minimized. Thus,the finding of a decreasing whole-body gross effi-ciency throughout the range of pedal rates in thepresent study contradicts the common voluntarymotor behavior of pedalling fast (465 r.p.m.) duringcycling at low to moderate exercise intensity levels(for references see Nielsen et al., 2004), despite steadyor increasing muscular efficiency. Such motor beha-viour may then be chosen because maximum powercan be performed at higher pedal rates (Kohler &Boutellier, 2005), and implies a given EP to representa lower load relative to maximum power at higherpedal rates. Whole-body gross efficiency, and thusMP, is particularly interesting when energy supply isa limitation, as in endurance performance (Nielsenet al., 2004; Foss & Hallen, 2005), and is thereforerelevant from a practical point of view (Moseley &Jeukendrup, 2001).

Key words: internal work, pedalling frequency, 3Dkinematics.

Acknowledgements

This study was supported by The Danish Sports ResearchCouncil (Grant 980501-14) and a grant given to authorG. Sjøgaard by The Danish Elite Sport Institution TeamDanmark.

Muscular efficiency during cycling

413

References

Ahlquist LE, Bassett Jr DR, Sufit R,Nagle FJ, Thomas DP. The effect ofpedaling frequency on glycogendepletion rates in type I and IIquadriceps muscle fibers duringsubmaximal cycling exercise. EurJ Appl Physiol 1992: 65: 360–364.

Asmussen E. Positive and negativemuscular work. Acta Physiol Scand1952: 28: 363–382.

Boning D, Gonen Y, Maassen N.Relationship between work load, pedalfrequency, and physical fitness. IntJ Sports Med 1984: 5: 92–97.

Bottinelli R, Reggiani C. Human skeletalmuscle fibers: molecular and functionaldiversity. Prog Biophys Mol Biol 2000:73: 195–262.

Chavarren J, Calbet JAL. Cyclingefficiency and pedalling frequency inroad cyclists. Eur J Appl Physiol 1999:80: 555–563.

Coast JR, Welch HG. Linear increase inoptimal pedal rate with increasedpower output in cycle ergometry. EurJ Appl Physiol 1985: 53: 339–342.

Coyle EF, Sidossis LS, Horowitz JF,Beltz JD. Cycling efficiency is related tothe percentage of type I musclefibers. Med Sci Sports Exerc 1992: 24:782–788.

Elder GC, Bradbury K, Roberts R.Variability of fiber type distributionswithin human muscles. J Appl Physiol1982: 53: 1473–1480.

Ferguson RA, Ball D, Sargeant AJ. Effectof muscle temperature on rate ofoxygen uptake during exercise inhumans at different contractionfrequencies. J Exp Biol 2002: 205:981–987.

Foss Ø, Hallen J. The most economicalcadence increases with increasingworkload. Eur J Appl Physiol 2004: 92:443–451.

Foss Ø, Hallen J. Cadence andperformance in elite cyclists. Eur JAppl Physiol 2005: 93: 453–462.

Francescato MP, Girardis M,diPrampero PE. Oxygen cost ofinternal work during cycling. Eur JAppl Physiol 1995: 72: 51–57.

Franch J, Madsen K, Djurhuus MS,Pedersen PK. Improved runningeconomy following intensified trainingcorrelates with reduced ventilatorydemands. Med Sci Sports Exerc 1998:30: 1250–1256.

Gaesser GA, Brooks GA. Muscularefficiency during steady-rate exercise:effects of speed and work rate. J ApplPhysiol 1975: 38: 1132–1139.

Hansen EA, Andersen JL, Nielsen JS,Sjøgaard G. Muscle fiber type,efficiency, and mechanical optimaaffect freely chosen pedal rate during

cycling. Acta Physiol Scand 2002: 176:185–194.

Hansen EA, Jørgensen LV, Sjøgaard G.A physiological counterpoint tomechanistic estimates of ‘‘internalpower’’ during cycling at differentpedal rates. Eur J Appl Physiol 2004:91: 435–442.

He ZH, Bottinelli R, Pellegrino MA,Ferenczi MA, Reggiani C. ATPconsumption and efficiency of humansingle muscle fibers with differentmyosin isoform composition. BiophysJ 2000: 79: 945–961.

Kang J, Hoffman JR, Wendell M, WalkerH, Hebert M. Effect of contractionfrequency on energy expenditure andsubstrate utilisation during upper andlower body exercise. Br J Sports Med2004: 38: 31–35.

Kohler G, Boutellier U. The generalizedforce-velocity relationship explainswhy the preferred pedaling rate ofcyclists exceeds the most efficientone. Eur J Appl Physiol 2005: 94:188–195.

Lollgen H, Graham T, Sjogaard G.Muscle metabolites, force, andperceived exertion bicycling at varyingpedal rates. Med Sci Sports Exerc 1980:12: 345–351.

Mallory LA, Scheuermann BW, HoeltingBD, Weiss ML, McAllister RM,Barstow TJ. Influence of peak VO2 andmuscle fiber type on the efficiency ofmoderate exercise. Med Sci SportsExerc 2002: 34: 1279–1287.

Marsh AP, Martin PE. The associationbetween cycling experience andpreferred and most economicalcadences. Med Sci Sports Exerc 1993:25: 1269–1274.

Martin R, Hautier C, Bedu M. Effect ofage and pedalling rate on cyclingefficiency and internal power inhumans. Eur J Appl Physiol 2002: 86:245–250.

Mifflin MD, St.Jeor ST, Hill LA, ScottBJ, Dougherty SA, Koh YO. A newpredictive equation for resting energyexpenditure in healthy individuals. AmJ Clin Nutr 1990: 51: 241–247.

Mogensen M, Bagger M, Pedersen PK,Fernstrom M, Sahlin K. Cyclingefficiency in humans is related to lowUCP3 content and to type I fibers butnot to mitochondrial efficiency. JPhysiol 2006: 571: 669–681.

Moseley L, Jeukendrup AE. Thereliability of cycling efficiency. Med SciSports Exerc 2001: 33: 621–627.

Neptune RR, Kautz SA, Zajac FE.Muscle contributions to specificbiomechanical functions do not changein forward versus backward pedaling. JBiomech 2000: 33: 155–164.

Nielsen JS, Hansen EA, Sjøgaard G.Pedalling rate affects enduranceperformance during high-intensitycycling. Eur J Appl Physiol 2004: 92:114–120.

Poole DC, Gaesser GA, Hogan MC,Knight DR, Wagner PD. Pulmonaryand leg VO2 during submaximalexercise: implications for muscularefficiency. J Appl Physiol 1992: 72:805–810.

Sargeant AJ, Beelen A. Human musclefatigue in dynamic exercise. In:Sargeant A.J., Kernell D., eds.Neuromuscular fatigue. Amsterdam.New York: North-Holland, 1993:81–92.

Sidossis LS, Horowitz JF, Coyle EF.Load and velocity of contractioninfluence gross and delta mechanicalefficiency. Int J Sports Med 1992: 13:407–411.

Sjøgaard G. Muscle morphology andmetabolic potential in elite road cyclistsduring a season. Int J Sports Med 1984:5: 250–254.

Sjøgaard G, Hansen EA, Osada T. Bloodflow and oxygen uptake increase withtotal power during five different knee-extension contraction rates. J ApplPhysiol 2002: 93: 1676–1684.

Stainbsy WN, Gladden LB, Barclay JK,Wilson BA. Exercise efficiency: validityof base-line subtractions. J ApplPhysiol 1980: 48: 518–522.

Suzuki Y. Mechanical efficiency of fast-and slow-twitch muscle fibers in manduring cycling. J Appl Physiol 1979: 47:263–267.

Takaishi T, Yamamoto T, Ono T, Ito T,Moritani T. Neuromuscular,metabolic, and kinetic adaptations forskilled pedaling performance incyclists. Med Sci Sports Exerc 1998: 30:442–449.

Takaishi T, Yasuda Y, Ono T, MoritaniT. Optimal pedaling rate estimatedfrom neuromuscular fatigue forcyclists. Med Sci Sports Exerc 1996: 28:1492–1497.

Whipp BJ, Wasserman K. Efficiency ofmuscular work. J Appl Physiol 1969:26: 644–648.

Willems PA, Cavagna GA, Heglund NC.External, internal and total work inhuman locomotion. J Exp Biol 1995:198: 379–393.

Winter DA. A new definition ofmechanical work done in humanmovement. J Appl Physiol 1979: 46:79–83.

Winter DA. Biomechanics and motorcontrol of human movement. NewYork: John Wiley & Sons Inc, 1990.

Hansen & Sjøgaard

414