This article was downloaded by: [Biblioteca. Universidad de Extremadura]On: 05 March 2015, At: 04:07Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,37-41 Mortimer Street, London W1T 3JH, UK

Journal of Sports SciencesPublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/rjsp20

Science and cycling: current knowledge and futuredirections for researchGreg Atkinson a , Richard Davison b , Asker Jeukendrup c & Louis Passfield da School of Sport and Exercise Sciences , Loughborough University , UKb International Sports Consultancy , Brisbane, QLD, Australiac School of Sport and Exercise Sciences , The University of Birmingham , Birmingham, UKd School of Applied Sciences , The University of Glamorgan , Pontypridd, UKPublished online: 13 Jun 2008.

To cite this article: Greg Atkinson , Richard Davison , Asker Jeukendrup & Louis Passfield (2003) Science and cycling: currentknowledge and future directions for research, Journal of Sports Sciences, 21:9, 767-787, DOI: 10.1080/0264041031000102097

To link to this article: http://dx.doi.org/10.1080/0264041031000102097

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the Content) containedin the publications on our platform. However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of theContent. Any opinions and views expressed in this publication are the opinions and views of the authors, andare not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon andshould be independently verified with primary sources of information. Taylor and Francis shall not be liable forany losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoeveror howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use ofthe Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematicreproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Journal of Sports Sciences, 2003, 2 1, 767-787 iTay lor&Francis healthsciences Science and cycling: current knowledge and future directions for research

GREG AT KIN SON^' * RICHARD DAVISON,~ ASKER JEUKENDRUP~ and LOUIS PASSFIELD" '~choo l of Sport and Exercise Sciences, Loughborough University, UK, 2~nternational Sports Consultancy, Brisbane, QLD, Australia, 3~chool of Sport and Exercise Sciences, The University of Birmingham, Birmingham, U K and 4~chool of Applied Sciences, The University of Glamorgan, Pontypridd, U K

Accepted 10 May 2003

In h s holistic review of cycling science, the objectives are: (1) to identify the various human and environmental factors that influence cycling power output and velocity; (2) to discuss, with the aid of a schematic model, the often complex interrelationships between these factors; and (3) to suggest future directions for research to help clarify how cycling performance can be optimized, given different race disciplines, environments and riders. Most successful cyclists, irrespective of the race discipline, have a high maximal aerobic power output measured from an incremental test, and an ability to work at relatively high power outputs for long periods. The relationship between these characteristics and inherent physiological factors such as muscle capilliarization and muscle fibre type is complicated by inter-individual differences in selecting cadence for different race conditions. More research is needed on high-class professional riders, since they probably represent the pinnacle of natural selection for, and physiological adaptation to, endurance exercise. Recent advances in mathematical modelling and bicycle-mounted strain gauges, which can measure power directly in races, are starting to help unravel the interrelationships between the various resistive forces on the bicycle (e.g. air and rolling resistance, gravity). Interventions on rider position to optimize aerodynamics should also consider the impact on power output of the rider. All-terrain bicycle (ATB) racing is a neglected discipline in terms of the characterization of power outputs in race conditions and the modelling of the effects of the different design of bicycle frame and components on the magnitude of resistive forces. A direct application of mathematical models of cycling velocity has been in identifymg optimal pacing strategies for different race conditions. Such data should, nevertheless, be considered alongside physiological optimization of power output in a race. An even distribution of power output is both physiologically and biophysically optimal for longer ( > 4 km) time-trials held in conditions of unvarying wind and gradient. For shorter races (e.g. a 1 km time-trial), an 'all out' effort from the start is advised to 'save' time during the initial phase that contributes most to total race time and to optimize the contribution of kinetic energy to race velocity. From a biophysical standpoint, the optimum pacing strategy for road time-trials may involve increasing power in headwinds and uphill sections and decreasing power in tailwinds and when travelling downhill. More research, using models and direct power measurement, is needed to elucidate fully how much such a pacing strategy might save time in a real race and how much a variable power output can be tolerated by a rider. The cyclist's diet is a multifactorial issue in itself and many researchers have tried to examine aspects of cycling nutrition (e.g. timing, amount, composition) in isolation. Only recently have researchers attempted to analyse interrelationships between dietary factors (e.g. the link between pre-race and in-race dietary effects on performance). The thermal environment is a mediating factor in choice of diet, since there may be competing interests of replacing lost fluid and depleted glycogen during and after a race. Given the prevalence of stage racing in professional cycling, more research into the influence of nutrition on repeated bouts of exercise performance and training is required.

Keywords: aerodynamics, efficiency, ergometry, nutrition, pacing.

Introduction

Cycle ergometry is a common form of exercise for general investigations in the sport and exercise sciences.

* Address all correspondence to Greg Atlunson, School of Sport and Exercise Sciences, Loughborough University, Loughborough The of the upper body LEI 1 3TU, UK. e-mail: g.atkinson2@lboro.ac.uk and arms during cycling makes it easier for researchers

Journal of Sports Sciences ISSN 0264-0414 print/ISSN 1466-447X online 0 2003 Taylor & Francis Ltd DOI: lO.lO8O/O264O4lO3 1000 102097

Dow

nloa

ded

by [B

ibliot

eca.

Unive

rsida

d de E

xtrem

adur

a], [$

{indiv

idualU

ser.di

splay

Name

}] at

04:07

05 M

arch 2

015

Atkinson et al.

to make invasive measurements, for example using a catheter (MacLaren et al., 1999), while the cyclist can exercise for long periods of time. In addition, cycling is generally less associated with injuries than running. Therefore, training studies may be easier to implement and be less affected by participant drop-out. Competi- tive cyclists benefit from such research in that the results are generally relevant to their chosen mode of exercise.

Nevertheless, for research results to be directly relevant to the competitive cyclist, research needs to integrate the influences of several factors, which are summarized in the model presented in Fig. 1. First, the researcher should consider the physiological character- istics of the athlete that are specific to the event (e.g. muscle fibre composition, maximal oxygen uptake). These inherent characteristics influence the power that can be generated in a cycling race. Besides inherent physiological ability, training, nutritional and pacing strategies have also a strong influence on overall race power output (Fig. 1). The position of the rider on the bicycle, predominantly governed by the vertical and horizontal displacement of the saddle, can also influ- ence power output.

So far, the multiple factors that influence cycling power have been mentioned, but there are also several factors that influence the relationship between power and cycling velocity (Fig. 1). The rider who, through superior ability, training and nutrition, has the potential to generate the most power during a race, might not

necessarily win the race. Cycling is a high-velocity sport, and races can be held in all terrains and environments. Therefore, physical resistive forces present in the race environment (e.g. hills or winds) influence greatly the relationship between cycling power and overall race velocity. The presence of different resistive forces in a race can influence the choice of bicycle design or its components. Figure 1 also shows that rider position and pacing strategy do not just affect physiological power output. Rider position can also affect the extent to which air resistance influences the power-velocity relationship. Pacing strategy can also influence the magnitude of the effect of wind and gradient on the power-velocity relationship.

As a specific example of the diverse nature of competitive cycling, during the 1996 racing season, Chris Boardman competed in the Tour de France, a 3 week race with stages over various distances (10 to 250 km) and terrains, followed soon after by participa- tion in the 4 km individual pursuit in the World Championships. Such severe intensity and duration of activity in the months before a 4 km pursuit would seem to go against current thinking on training periodization for optimal performance. Nevertheless, Chris Boardman won the World Championship 4 km event. Boardman competed in the World Championships in a new aerodynamic position that would have lowered air resistance, but may have had a paradoxical influence on the rider position for optimal power output (Gnehrn et al., 1997). Such empirical observations serve to illustrate

Overall race velocity

A

I ~ a c i n s ]

Fig. 1. Diagram showing the various factors that influence cycling power output and speed, in line with the topics covered in the present review.

Race-specific retarding

forces ~ i k e d e s i g n l

.t Rider position

Race-specific training

.+

Cycling power t

b4 Race-specific

nutritional strategies strategies

Race-specific inherent physiological ability

Dow

nloa

ded

by [B

ibliot

eca.

Unive

rsida

d de E

xtrem

adur

a], [$

{indiv

idualU

ser.di

splay

Name

}] at

04:07

05 M

arch 2

015

Science and cycling

the importance of considering the many, and sometimes competing, influences on cycling performance.

Several excellent recent reviews on cycling science have considered the differing nature of cycling events (e.g. Burke, 2000; Neuman, 2000). In these reviews, the event-specificity of training strategies was empha- sized. Hence, training factors are not considered in the present review. It is only relatively recently that researchers have started to consider fully all the factors presented in Fig. 1, and the interrelationships between these factors, that characterize competitive cycling. For example, ergometer tests that mirror the intermittent nature of cycling road races and that simulate the variable resistive forces a rider might experience during a time-trial were formulated only in the last 10 years (Palmer et al., 1996b; Atkinson and Brunskill, 2000).

The aim of this paper is to provide an overview of the recent research on cycling that has attempted to 'close the loop' between the laboratory and the race environ- ment. A 'holistic' approach is adopted in an attempt to discuss current and future research pertaining to each factor bearing in mind interrelationships between other factors. The paper is structured with Fig. 1 in mind. Each of the factors that influence cycling velocity will be discussed, except, as mentioned above, for training strategies. Each of these discussions will end with suggestions for relevant research in the future.

Physiological factors

Although we have stressed the diverse nature of cycling events, cycling is predominantly an endurance sport. Even cycling performance lasting 5 min has been shown to be highly correlated with maximal aerobic power

(Passfield and Doust, 2000). Exceptions would be the 200 m and 1 km sprint events. Performance optimiza- tion for the latter event is mentioned in a later section of the review. In this section on physiological factors, we will deal mainly with the physiological aspects of aerobic metabolism that affect cycling performance. For a more detailed review of the integration of factors that affect general endurance performance, see Coyle (1995) and Hawley and Stepto (2001).

Most of the new information we review comes fi-om studies involving European professional cyclists. Although such information is of great interest, caution should be applied in extrapolating this information to all competitive standards in cycling. The typical professional cyclist who is based in Europe would cover 25,000 to 35,000 km in a year, with in excess of 90 days of competition and participation in at least one of the major 3 week tours (Giro dYItaliaJ Tour de France, Vuelta a Espaiia). In contrast, the amateur cyclist would normally cover no more than 25,000 km a year, have less than 50 competitive days and not normally race on more than two consecutive days. Figure 2 illustrates the relative time spent at different power outputs in typical professional and amateur road races lasting 6 h and 3.5 h, respectively. Clearly, there is a difference in relative power profiles and, therefore, the likely physiological demands and determinants of success would also be different.

Despite possible differences between competitive standards, the key physiological factors that are related to success in cycling are maximal aerobic power ( ~ o , , ~ , muscle fibre type, cycling efficiency and lactate threshold. The research in which these factors have been suggested to be important will now be discussed in detail.

20 60 100 140 180 220 260 300 340 380 420 460 500 540 580 620 660 700 740 780

Power (W)

Fig. 2. Distribution of power output during two individual road race performances, one top 6 h professional road race (m) and one top 3.5 h amateur road race (0). Unpublished data obtained by one of the authors (R.D.).

Dow

nloa

ded

by [B

ibliot

eca.

Unive

rsida

d de E

xtrem

adur

a], [$

{indiv

idualU

ser.di

splay

Name

}] at

04:07

05 M

arch 2

015

Atkinson et al.

Maximal aerobic power

As is the consistent a large

case for other endurance-based athletes, a observation is that world-class cyclists have V 2 An absolute range of 5.0-

6.4 1. min-' or relative range of 70-85 ml . kg-' . min-' is typical for professional cyclists (Coyle et al., 199 1; Lucia et al., 1998, 2001a; Mujika and Padilla, 2001). The largest relative values ( > 80 ml - kg-' - min-l) tend to be observed for specialist climbers who have a relatively low body mass (

Science and cycling

race cycling performance. Theoretically, there would be benefits of having a fast response in oxygen consumption to exercise, as this would reduce the contribution from anaerobic pathways to supply the energy for rapid changes in exercise intensity. One of the problems with current research into v02 kinetics is the tendency to use non-trained cyclists and 'square wave' exercise starting from rest (e.g. Ozyener et al., 2002), which does not replicate the circumstances in road cycling, where intermittent periods of high- intensity activity are superimposed on long periods of low- to moderate-intensity work. There is currently a poor understanding of what happens to v02 kinetics after several hours cycling during which exercise intensity is varied.

One of the important functional abilities of the cyclist that Coyle (1995) suggested is related strongly to endurance performance is gross mechanical efficiency [(work donelenergy consumed) x 1001. Gross mechan- ical efficiency during cycling has been reported to be between 18 and 26% in well-trained male competitive cyclists (vo~,, = 69 ml - kgp1 . min-'; Coyle et al., 1992), between 23.6 and 25.6% in amateur compe- titive cyclists (vo~,~, = 4.9 l . min-l; Lucia et al., 2002a) and between 24.4 and 28.8% in professional cyclists (v02,, = 5.0 1 . minpl; Lucia et al., 2002a). There is no difference in gross mechanical efficiency between amateur and professional cyclists at low workloads (Lucia et al., 2002a). In previous studies, both gross mechanical efficiency and delta efficiency [(Awork done1Aenergy expenditure) x 1001 were sug- gested to be related to the proportion of type I fibres in muscle (Coyle et al., 1992). More recently, Marsh et al. (2000) have shown that there are no differences in delta efficiency for different cadences ranging from 50 to 100 rev.minP1 for runners, less trained non- cyclists or trained cyclists. However, the authors commented that, although not statistically significant, the trained cyclists exhibited a delta efficiency that was slightly higher (- 1-2%) across the range of cadences. The highest power output used for these comparisons was 200 W, which would be considered a relatively low exercise intensity for a professional cyclist. It is possible that greater differences would be noted for intensities closer to competitive intensities. Horowitz et al. (1994) showed that, when a group of competitive cyclists was divided according to the proportion of type I fibres (high=73%, nor- mal=48%), the subgroup that was high in the proportion of type I fibres produced a significantly higher power for a given v02 during a 1 h test, indicating a higher efficiency.

The results of a recent study, which involved a ramp protocol, suggested that professional riders have a higher gross mechanical efficiency at intensities above the lactate threshold than highly trained amateur riders, with the difference increasing with greater exercise intensities (Lucia et al., 2002a). Nevertheless, the validity of the unusually high (28.8%) mechanical efficiency reported by Lucia et al. (2002a) and in a later study by Lucia et al. (2002~) has been questioned recently (Jeukendrup and Martin, 2003). As an indication of the possible benefit of an increased efficiency, Jeukendrup and Martin (2001) calculated that increasing gross efficiency by 1% for a 70 kg cyclist who can maintain a power output of 400 W for 1 h would result in a 48 s improvement for a time-trial over 40 km.

One aspect of efficiency that remains relatively under- researched is the change in efficiency as exercise progresses. The results of one study demonstrated that the reductions in efficiency resulting from - 60 min of constant-intensity exercise (60% VO~,,) are related to reductions in power output during a high-intensity bout (5 min) performed before and after the constant- intensity exercise (Passfield and Doust, 2000). There- fore, cycling performance could be determined by both high gross efficiency and changes in that efficiency as a result of continuing exercise. It is currently unclear whether professional cyclists exhibit an even more attenuated reduction in cycling efficiency as exercise progresses than trained cyclists, and also to what extent the changes in efficiency are affected by variable- intensity exercise rather than constant-intensity exer- cise.

Muscle capillariza tion The importance of muscle capillarization for endur- ance performance has been highlighted in several reviews (Coyle, 1995; Hawley and Stepto, 2001). Nevertheless, specific studies on muscle capilliariza- tion and cycling performance are very limited, and to date there have been no investigations of professional cyclists. Coyle et al. (1988) reported a strong correla- tion (r=0.74) between capillary density and time to fatigue at - 88% v02, in a group of well-trained cyclists. In a later study, Coyle et al. (1991) compared a group of elite national cyclists with good regional riders and showed that the elite riders had a 23% higher muscle capillary density. Capillary density was significantly correlated with the average absolute work rate for a 1 h performance. The importance of capillary density is clearly illustrated by the fact that, in both the above studies, capillary density was a key variable in regression equations to predict perfor- mance.

Dow

nloa

ded

by [B

ibliot

eca.

Unive

rsida

d de E

xtrem

adur

a], [$

{indiv

idualU

ser.di

splay

Name

}] at

04:07

05 M

arch 2

015

Atkinson et al.

Fibre type From the information available, it would appear that, to be a good endurance cyclist, it is necessary to have or attain a very high proportion of type I muscle fibres (Coyle et al., 1991; Bishop et al., 2000; Lucia et al., 2002b) or, conversely, a low proportion of type IIx fibres (Lucia et al., 2002b). Further evidence of the importance of type I fibres, and the optimization of their use, are the cadences used typically by professional cyclists. These relatively high ( > 90 rev min-') ca- dences do not tend to minimize oxygen uptake but rather minimize force per pedal stroke (Lucia et al., 2001b). This lowering of force per pedal stroke by higher cadences would seem to be a mechanism whereby the recruitment of type I1 fibres is minimized (Ahlquist et al., 1992).

There is good evidence to suggest that the preferred cadence of trained cyclists tends to rise with increasing workload (Marsh and Martin, 1997). For example, when high workloads (- 400 W or - 90% ~ 0 ~ ~ ~ ) are required for a considerable period in time-trials, professional cyclists adopt high cadences (-93 rev. min-') for optimal performance (Lacia et al., 200 1 b) . When environmental conditions are very constant and even higher power outputs ( > 600 W) are required - for example, in a team pursuit on a track - the preferred cadences are extremely high (1 33-1 39 rev.minP') (Broker et al., 1999).

Given the above information, it is puzzling that the cadences used by cyclists during uphill cycling are considerably lower (- 60-80 rev. min-') (Davison et al., 2000; Lucia et al., 2001b). Presumably the power outputs required are also high in uphill cycling. Closer examination of the top performers in mountain stages suggests that they adopt relatively higher cadences rather than a higher gear ratio (Lucia et al., 2001b). This observation has also recently been emphasized by a remarkable performance by one rider (Lance Arm- strong) in 2001 on the Alpe dYHuez climb, when to achieve the average speed of 22 km - h-' (estimated to require a power output of - 500 W), he maintained an average of 100 revmmin-' for the 40 min duration of the climb (Lucia, 2001b). Nevertheless, it is worth noting that the second rider to the top of Alpe dYHuez in 2001 (Jan Ulrich) used a much higher gear and had a correspondingly lower cadence ( - 70 rev min -'), similar to that measured for other professional riders while climbing ( - 7 1 rev min -I; Lucia et al., 200 1 b), but still fast enough to beat all but one rider. If higher cadences do improve cycling performance, possible mechanisms could include reducing the force per pedal stroke to either minimize recruitment of type I1 muscle fibres and optimize the use of the more efficient fatigue- resistant type I fibres, or to minimize the disruption of

blood flow to the active muscle mass (Takaishi et al., 2002). It has also been shown that higher pedal cadences result in a more effective skeletal-muscle pump, which increases muscle blood flow and venous return (Gotshall et al., 1996). Further work is required to identify the reasons for inter-individual differences in cadence selection, and whether training a cyclist to use different cadences actually improves performance.

The physiological consequence of a large number of highly trained type I fibres is the ability to sustain high power outputs for long periods as well as a high gross mechanical efficiency. Complementing these character- istics would be a high level of muscle capilliarization to supply enough oxygen and substrate to sustain the large energy outputs. As much of the more recent data on cycling performance have been obtained from profes- sional riders, it is difficult to unravel the extent to which adaptation to this physiological load contributes to the attributes measured in these athletes.

It should be noted that a high proportion of type I muscle fibres may be associated with overall success in the major tours and time-trial events, but that several riders can have more individual stage wins than the overall winner of the tours. The winner is deemed the fastest overall rider, not the most consistent finisher (although there is also usually a 'winner's jersey' for this rider), who is conventionally the rider with the best sprinting ability. The objective of the road sprinter is to complete most of a race with the least possible energy cost and then, usually with the aid of team-mates, produce a very high intensity burst of speed at the finish of the race. To enable this very high-intensity finish ( > 1500 W for the last 200 m), it could be that the fibre type distribution in these riders is different from that in riders who finish high up the overall classification. Nevertheless, there has been no investigation to characterize fully these road sprinters in particular. Studies that have involved explosive strength training (Bastiaans et al., 200 1) or interval training (Stepto et al., 1999) to enhance the general performance of trained cyclists have failed to produce consistently favourable results. It is also inter- esting to note that none of the recent overall winners of any of the major tours would be considered to be a top sprint finisher. To confuse the appraisal of the possible fibre typelanaerobic contribution to cycling performance even further, it is common for the top contenders in a major stage race to win or at least be very highly placed in the prologue time-trial, which is generally less than 10 lun, and would undoubtedly require some ( < 10%) anaerobic contribution for a top performance.

Ventilatory demand Several recent studies have shown that professional cyclists can maintain very high power outputs and

Dow

nloa

ded

by [B

ibliot

eca.

Unive

rsida

d de E

xtrem

adur

a], [$

{indiv

idualU

ser.di

splay

Name

}] at

04:07

05 M

arch 2

015

Science a n d cycling

percent VO~,,, particularly during time-trials (Lucia et al., 2001a). Obviously, to maintain such high workloads, there is a considerable ventilatory demand. The average ventilation for trained cyclists during a 40 km indoor time-trial has been measured at 1 1 1 + 13 1 . min-' (Smith et al., 200 l), and it would be logical to conclude that for shorter-duration ( < 40 km but > 5 km), higher-intensity races, the ventilatory demand would be even higher. Romer et al. (2002) showed recently that after a time-trial performance, the inspiratory muscles exhibit consider- able fatigue with a trend for the fatigue to be greater for shorter-duration, higher-intensity performances (20 vs 40 km). These researchers also reported that 6 weeks of specific inspiratory muscle training could significantly reduce this fatigue, resulting in improved performances in both 20 and 40 km time-trials.

Measurements of inspiratory muscle performance have not been reported for professional cyclists. There- fore, to date, it is unclear whether these athletes have more developed inspiratory muscles as a result of their large training load or whether they, too, could benefit as a result of specific inspiratory muscle training. How- ever, in one study it was suggested that professional cyclists exhibit different breathing patterns during incremental exercise than trained elite amateur cyclists (Lucia et al., 1999a). There was no difference in the maximal ventilation between the two groups, but the amateur riders had significantly higher minute ventila- tion (v& at 300,350 and 400 W and a higher breathing frequency at most of the moderate-to-maximal inten- sities. In contrast, the professional riders appeared to adopt prolonged expiratory times and there was no evidence of a tachypnoeic shift at high exercise intensities with the increase in ventilation being due to both an increase in tidal volume and in breathing frequency. Nevertheless, we acknowledge that some of the above differences could be explained by the different relative exercise intensities, particularly in relation to the respiratory compensation for metabolic acidosis.

Future directions From a sport scientist's point of view, professional cyclists represent an opportunity to investigate the adaptation that occurs in response to a massive overload of endurance exercise. This overload is probably more than for any other athlete. Unfortunately, until the recent work of Lucia, Padilla and co-workers, elite road cycling was a relatively under-researched area and many questions remain unanswered. Even this more recent work has focused on the physiological responses to exercise intensities that have been merely estimated from heart rate. The availability of power-measuring

devices on the bicyle provides the opportunity for more studies of the physiological responses to competitive cycling, while knowing the exact power output in races.

Professional cyclists understandably have been very reluctant to donate muscle tissue for research purposes. Therefore, virtually nothing is known about the morphological and biochemical characteristics of their muscles. There has also been little investigation into any correlation between changes in performance or mechanical efficiency within a yearly training cycle and changes in cardiac, skeletal muscle and enzymatic responses. Sports scientists have been investigating the physiological responses to different pedal cadences for some time, yet we still do not fully understand the interaction of cadence, workload, efficiency and hill climbing. The interaction of V O ~ kinetics and the variable exercise intensity experienced during road race cycling is another area requiring investigation.

Physical factors

In the previous section, we outlined the physiological characteristics that influence power production during a race. This power production is required to overcome a complex interaction of resistive forces presented by the competition environment. Consequently, race power may not necessarily relate to cycling velocity in a simple, direct way. Depending upon the nature of the event, these resistive forces are likely to include air resistance, gravity, rolling resistance, inertia and frictional losses from the chain and bearings. The major resistive force at normal racing speeds on flat terrain comes from air resistance, with rolling resistance and frictional losses likely to account for less than 10% of total energy expenditure (Kyle, 1996). The inertia of the rider and cycle become significant during acceleration and deceleration, while the effect of gravity becomes a major factor in hilly terrain.

Air resistance

At cycling speeds above 15 km . hp' on level ground, air resistance becomes the major resistive force (Kyle, 1986). This aerodynamic drag becomes significant because the air resistance increases as an approximate squared function of cycling velocity. Power output should, in theory, be expected to increase as a cubic function of cycling speed, because the total cycling power output is the product of air resistance and velocity. Nevertheless, Bassett et al. (1999) observed that the average exponent for power output versus cycling speed is 2.6 for indoor track cycling. The reason that the exponential is less than 3 is probably due to the combination of all resistive forces, which act with linear,

Dow

nloa

ded

by [B

ibliot

eca.

Unive

rsida

d de E

xtrem

adur

a], [$

{indiv

idualU

ser.di

splay

Name

}] at

04:07

05 M

arch 2

015

Atkinson et al.

quadratic and cubic functions. Accordingly, a useful model of cycling should incorporate these different terms and not use a simple power function to model the power-speed relationship.

The aerodynamic drag of a cyclist is commonly represented by his or her drag coefficient (Cd). This rather complex coefficient may be broadly summarized as an integration of pressure drag, related to a cyclist's shape and frontal area, and friction drag, affected by the surface smoothness of the cyclist and his or her bicycle. At typical racing speeds, pressure drag plays a much more significant role in determining a cyclist's drag coefficient (Kyle, 1996). It is the predominance of pressure drag that enables riders to benefit fiom slipstreaming (drafting) behind one another. By slip- streaming, the drafting cyclists benefit from riding into a lower pressure vortex created by the leading cyclist. Following cyclists may therefore experience a 30% reduction in the required power output compared with the lead rider (Broker et al., 1999). Broker et al. (1999) found that there is a further 6% saving in power output by riding in third compared to second place in a group for the same reasons. By alternating the additional work associated with leading, riders in a team or group are consequently capable of sustaining speeds significantly greater than that of a lone rider. From the data of Broker et al. (1 999), it may be calculated that four riders alternating the lead can increase the average power generated in the lead position by around 25% com- pared with that of a single rider. This magnitude of increase in power will not be fully reflected in race times, however, because of the disproportionately higher power output required for the faster speeds.

In recent years, competitive cyclists have adopted increasingly aerodynamic positions to help minimize their drag and improve performance. Bassett et al. (1999) have calculated that 60% of the increased distance covered in the hour record for track cycling over the previous 30 years is due to improved aerodynamics rather than fitness. This importance of aerodynamics is exemplified by an analysis of Chris Boardman's best hour record of 56.375 km set in 1996. Bassett et al. (1999) estimated that Boardman averaged 20 W less than Tony Rominger, the previous record holder. Nevertheless, Boardman was 6 kg heavier than Rominger and rode over 1 km further. Such a finding is likely to be largely accounted for by the extremely aerodynamic 'Superman' position that Boardman adopted, as he did for the Olympics, for his successful record ride. This position was subsequently banned by cycling's world governing body, effectively increasing the average power output required for any future successful record attempt by an estimated 5%. Gnehm et al. (1997) found that riding 'on the drops' of the handlebars was associated with a higher metabolic cost

of riding than when in a less streamlined position ('on the hoods' or 'on the tops' of the handlebars). Never- theless, these authors maintained that the aerodynamic advantages associated with more streamlined positions outweigh any detrimental effects on the ability to maintain power output while adopting them.

It should be remembered that aerodynamic drag is a function of relative air movement, not simply horizontal (ground) velocity. Accordingly, power output in cycling is determined both by horizontal velocity and by air speed. For example, cycling at 30 km-hp l into a headwind of 10 km . hpl requires an equivalent power output to riding at 40 km . h-' through still air. Figure 3 presents the effects of wind velocity (headwinds and tailwinds) and gradient (uphill and downhill) on cycling speed. This relationship is very nearly linear (Martin et al., 1998). Changes in air density will also influence aerodynamic drag. The major determinants of air density are air temperature, barometric pressure and, to a lesser extent, humidity. Changes in these variables will have a proportionate effect on air density and, consequently, the power required to maintain a given speed. In this manner, a combination of an increase in temperature of 5C and decrease in barometric pressure of 15 mmHg will reduce air density and aerodynamic drag by approximately 4%.

Rolling resistance Rolling resistance contributes significantly to the total energy cost of cycling at slow speeds ( < 15 km - h-I). Under common road- or track-racing conditions, roll- ing resistance is thought to be almost constant and independent of cycling speed (Kyle, 1996). Accord- ingly, at higher speeds typical of competitive road or track racing, rolling resistance tends not to be practi- cally significant as other factors become much more important. This fact can be seen when models of cycling velocity are examined. Although the rolling resistance term in these models has been related to velocity, the contribution to overall power output is small (Martin et al., 1998; Bassett et al., 1999). The major variables influencing rolling resistance are riding surface, tyre diameter, construction and pressure, and the total load on the tyre (Kyle, 1996). Of these variables, tyre pressure is perhaps the one that riders have most influence over. Surprisingly, Ryschon and Stray-Gundersen (1993) found no discernible effect upon the oxygen cost of cycling over a range of tyre pressures typically used by cyclists (6-10 bar). Never- theless, there is some scope for changes in rolling resistance to influence performance at high velocities, as it is generally estimated to account for - 10% of the total power output when cycling at 40 km . hpl (Kyle, 1986).

Dow

nloa

ded

by [B

ibliot

eca.

Unive

rsida

d de E

xtrem

adur

a], [$

{indiv

idualU

ser.di

splay

Name

}] at

04:07

05 M

arch 2

015

Science a n d cycling

o-' , 1 -10 -5 0 5 10

Wind velocity (m-s-1) I I I

- 3 0 3 Road gradient (%)

Fig. 3. The influence of wind speed and direction (dashed line) and road gradient (dotted line) on cycling velocity for a hypothetical cyclist with a constant power output of 255 W, a drag area of 0.264 m2 and a mass of 7 1.9 kg. The influence of these two resistive factors on cycling speed is very nearly linear (12 > 0.99) over the range of wind speeds and gradients presented and for the hypothetical data that were entered into the mathematical model. A negative wind speed corresponds to a tailwind; a negative gradient corresponds to a downhill. Figure drawn from data collected under the conditions reported by Martin et al. (1998).

Frictional losses

Frictional losses during cycling are thought to be relatively small, but variable (Kyle, 1988). Whitt and Wilson (1974) suggested that a new clean chain may incur frictional losses of only 1.5%. Kyle and Caiozzo (1 986) examined chain drive losses by comparing power input with output at a pedal rate of 72 rev min-'. These authors found that the percentage of energy lost increased with power output from 1.9% at 100 W to 3.9% at 300 W. In contrast, Martin et al. (1998) mathematically modelled chain drive losses at a fixed 2.4%. Martin and colleagues based their model on finding a consistent difference of this magnitude between two ergometers, one of which measured power output before the chain drive and the other after it. This method may not be valid, however, as Jones and Passfield (1998) suggested that both of the above ergometers already include chain drive losses in the calculated power output. Frictional losses in bearings (e.g. in the wheel hub) are likely to be trivial. For example, Martin et al. (1998) suggested that wheel bearing losses amount to about 1.4 W (0.6%) when cycling at 30 k m . h-l.

Resistance due to gravity

During uphill cycling, speeds frequently slow to such an extent that aerodynamic drag is no longer important. Instead, the most significant resistive force is caused by

lifting the rider's own mass and that of the bicycle up the slope against gravity. The influence of road gradient on cycling speed is, like the influence of wind, almost linear (Martin et al., 1998; Fig. 3). The work done in overcoming gravity when cycling uphill may be calcu- lated simply as the change in potential energy. Thus the work done is governed by the mass of rider and cycle, gravity and height gained. Cycling power output may be calculated simply by considering the rate of this change in potential energy. Under windless conditions, these principles may be used to calculate cycling power output while cycling up a gradient, where gradient and cycling speed are known (e.g. on an inclined treadmill) if rolling resistance and fictional losses are assumed to be negligible.

From the above it can be seen that the mass of rider and cycle directly influence power output when climb- ing. Consequently, riders specializing in hill or moun- tain climbing, and their bicycles (within regulations), should be as light as possible to maximize their power output to weight ratio. A lower inertia is also a function of less mass. A low inertia will aid performance in racing conditions that feature repeated accelerations, such as criterium racing, or Madison and points races on the track. Kyle (1996) pointed out that in such conditions the mass of rotating components should be as little as is practical, since the inertial effects are more pro- nounced. For time-trials over flat terrain, reducing the mass of rider or cycle is unlikely to yield significant

Dow

nloa

ded

by [B

ibliot

eca.

Unive

rsida

d de E

xtrem

adur

a], [$

{indiv

idualU

ser.di

splay

Name

}] at

04:07

05 M

arch 2

015

Atkinson et al.

improvements in performance. When cycling downhill, heavier riders have a distinct advantage over lighter ones, as Swain (1997) observed. This is because smaller riders tend to have a greater air resistance relative to their body mass.

Future directions A variety of methods have been used to estimate the energy requirements of road and track cycling under different resistive forces. These methods include the measurement of human energy expenditure from oxygen consumption (Swain et al., 1987; McCole et al., 1990), doubly labelled water techniques (Wester- terp et al., 1986), estimates of energy expenditure from heart rates (Palmer et al., 1994; Padilla et al. 2000), deceleration tests (de Groot et al., 1995) and the towing of cyclists (di Prampero et al., 1979; Capelli et al., 1993). Using data from such studies, Olds et al. (1993, 1995) derived mathematical models of road and track cycling. These approaches have provided useful insight into the demands of cycling. However, several of the methods used to identify the magnitude of the various terms in mathematical models may now have been superseded by the development of a cycle-mounted crank dynamometer (SRM, Julich, Germany). This crank dynamometer enables the direct measurement of power output in both laboratory and field conditions, and has proved to be reasonably accurate and reliable (Jones and Passfield, 1998).

Consequently, recent studies have been able to describe cycling power output in a range of cycling events from measures taken in the field (Broker et al., 1999; Craig and Norton, 200 1; Jeukendrup et al., 200 1) and from mathematical models (Martin et al., 1998; Bassett et al., 1999). These models should allow scientists and coaches to evaluate accurately the theoretical performance implications of changes in environmental, physiological, nutritional and technological factors. For example, as described in the next section on 'Pacing strategy', such models may be useful for exploring the influence of pacing strategy on performance under various hypothetical contexts (Swain, 1997).

The design of bicycle used in the studies that have generated mathematical models has improved over the years in terms of its relevance to competition. For example, Martin et al. (1998) used a racing bicycle with the rider in a time-trial position. Nevertheless, it can be claimed that all the data derived from these models are relevant only to the road and track disciplines of cycling. Although exercise intensity during mountain biking has been estimated from heart rates during races (Impellizzeri et al., 2002), few data appear to be available on the characteristics of off-road cycling, an issue that future research may wish to address.

Pacing strategy

Although pacing strategy is recognized as important by coaches and athletes, empirical studies on the optimum distribution of power within a race are relatively rare. Robinson et al. (1958) and Foster et al. (1 993) reported that an even distribution of work is optimum for 1.2 km running and 2 krn cycling time-trials, respectively. Close scrutiny of more recent research work on optimum pacing strategy, in conjunction with an appreciation of how the external factors outlined in the above section on 'Physical factors' influence cycling speed, suggests that a more complicated strategy than an even or 'flat-line' pacing strategy would be best in some races.

It is difficult to arrive at simple conclusions for pacing strategy in 'massed start' road races, due to the complexity of the event. Consequently, the current focus is on time-trial events. Elite road racers generally try to avoid varying expenditure of energy until the h a 1 50 km of a race, a tactic that has empirical support (Palmer et al., l996b). Nevertheless, many professional races have been won by riders working hard early in the race to establish an unassailable lead. Moreover, tactics in road racing depend on a myriad of variable factors, including environmental conditions, the amount of teamwork that is evident, and the number and gradient of hills in the race (Lucia et al., 1999b). It would be extremely difficult to model fully the influence of all these factors, but not impossible given enough cases for adequate statistical power in a regression model.

Track time-trials The relationships between physiological effort, power and cycling speed are relatively strong in track events, since there are few changes in the physical character- istics of the racing environment (e.g. changes in barometric pressure, temperature, wind direction). Consequently, as in running, split-times within track races may be used as relatively accurate indicators of pacing strategy. The longest Olympic time-trial event on the track is 4 km. However, world records are recognized for individual events of a longer duration (e.g. the 'hour record'). The results of Foster et al. (1993) suggest that an even pacing strategy of very similar split-times is optimal for events over 4 km in duration, this advice being formulated from an optimal physiological response to exercise.

Shorter time-trials over 1 and 4 km are complicated by the fact that the time required to bring the bicycle up to speed is a relatively larger proportion of total race time, and also by the relative contribution of kinetic energy at the end of a race. For example, elite cyclists in the 4 km pursuit complete the first 250 m of the race

Dow

nloa

ded

by [B

ibliot

eca.

Unive

rsida

d de E

xtrem

adur

a], [$

{indiv

idualU

ser.di

splay

Name

}] at

04:07

05 M

arch 2

015

Science and cycling

in a time that is about 3-4 s slower than the other 250 m sections (Wilberg and Pratt, 1988). Van Ingen Schenau et al. (1994) and de Koning et al. (1999) simulated 1 and 4 km track races, incorporating the characteristics of the standing start, and found that the best pacing strategy is an 'all-out' effort at the start, even if such a tactic leads to a drop in power towards the finish of the race. Van Ingen Schenau et al. (1994) maintained that it would be important for a 1 krn sprint cyclist not to 'hold back' on the 'all-out' strategy with the remainder of the race in mind, since a considerable advantage can be obtained by inputting more kinetic energy at earlier stages of the race. In effect, the rider would be releasing more energy during the first part of the race and this energy would lead to faster times even if power outputs drop considerably in the later stages of the sprint. Van Ingen Schenau et al. (1994) cited the research work of Hirvonen et al. (1987) to support their hypothesis that maximal rates of energy liberation are important to sprinting. Hirvonen et al. (1987) found that elite sprint runners have a larger rate of breakdown of creatine phosphate in the first seconds of a race. Recently, Bishop et al. (2002) found that the improve- ment in performance associated with an 'all-out' start in sprint kayaking may be mediated by faster V O ~ kinetics. More cycling-specific research work is re- quired to elucidate the exact balance between biophy- sical optimization of energy and physiological production of that energy. Craig and Norton (2001) suggested, as we have done in the section on 'Physical factors', that recent advances in technology for measuring power output under competitive conditions could be used to monitor the characteristics of such

track races and the physiological acceptability of high starting power outputs.

Road time-trials The cycling time-trial competitions that are typically held on open roads are much longer (10-60 km in elite stage races) than those held on the track. Therefore, the contribution of the standing start to total time is much less important in road time-trials. Moreover, an attempt to maximize kinetic energy input with an 'all-out' starting effort would lead to later reductions in power that could be disastrous to overall race time. Never- theless, there is evidence that it is very difficult for a rider not to start a time-trial of any distance with a high power output. In Fig. 4, we present the power profile for an Olympic 4 km pursuit cyclist who completed a simulated 16.1 km time-trial on a cycle ergometer (Computrainer Pro, Racer Mate Inc., Seattle, WA). In agreement with Nikolopoulos et al. (2001), it can be seen that the initial power was much higher than for the rest of the race. This profile is characteristic of most cyclists (e.g. the sample studied using the same cycle ergometer as above also presented in Fig. 4) and is not necessarily accompanied by an immediate increase in perceived exertion compared to a less powerful start (Firth, 1998; Atkinson and Brunskill, 2000). If cyclists are trained to blunt the starting effort, using feedback on their power, total race times are usually improved (Firth, 1998; Atkinson and Brunskill, 2000), adding weight to the argument that the optimal pacing strategy for road time-trials is one that varies little from the average power over the whole race. If a device for

Olympic-level cyclist

Distance (km)

Fig. 4. Power profile for a 4 km Olympic pursuit rider during a 16.1 km simulated time-trial. Average power over the simulated race was 327 W for the rider (unpublished data obtained by one of the authors, G.A.). Mean power profile of a sample (n = 7) of cyclists of varying ability also presented (Atkinson and Brunskill, 2000). All data were collected using an ergometer (Computrainer Pro, Racer Mate Inc., Seattle, WA).

Dow

nloa

ded

by [B

ibliot

eca.

Unive

rsida

d de E

xtrem

adur

a], [$

{indiv

idualU

ser.di

splay

Name

}] at

04:07

05 M

arch 2

015

Atkinson et al.

measuring power on the bike is not available, good coaching advice might be for the rider to choose a gear that is slightly lower than would normally be adopted at the start to encourage a less powerful first few minutes of a long time-trial. The monitoring of heart rate or ratings of perceived exertion by a rider appears to be unreliable for detecting differences in starting power (Atkinson and Brunskill, 2000).

The information above might suggest that holding power relatively constant over the duration of a race is the best strategy for all road time-trials. However, the identification of an optimum pacing strategy is compli- cated by variable external conditions within the race, such as gradient and wind direction. It is extremely unlikely that any road time-trial is held on a completely flat course with no wind - the conditions simulated in most laboratory time-trials (Hickey et al., 1992; Palmer et al., 1996a). A typical cyclist who generates a constant powerof255 Wwouldtravelat6 m-s- ' i na 10 m-s - ' headwind compared with 18 m - s-' in a 10 m . s-' tailwind. The same cyclist's speed would be 5 m - s-' while travelling up a hill with a 6% gradient compared with 18 m - s-' while travelling down the same hill (Fig. 3; Martin et al., 1998). The positive effects of tailwinds and downhills on cycling velocity do not completely cancel out the deleterious effect of headwinds and uphills on cycling velocity. Therefore, circuit or 'out and home' races in hilly terrain and windy conditions will always be slower than races in flat, calm environ- ments (Kyle, 1988; White, 1994; Martin et al., 1998). Just like the standing start in track racing, relatively slow sections of a race (headwinds, uphill) make up the greatest proportion of total race time. Accordingly, a successful strategy may be to focus upon improving cycling velocity in these slow sections.

In a simulation, Swain (1997) used the equation of motion of a cyclist provided by Di Prampero et al. (1979) to predict that, when a race is held on a hilly course or on a course with equal sections of headwind and tailwind, faster times would be recorded if the cyclist increases power on the uphill or headwind sections and compensates with reduced power on the downhill or tailwind sections (total or 'net' work done remaining the same). For example, Swain (1997) predicted that a rider who varies power output by 10% in variable conditions of a strong (24 km.h-') wind could be more than a minute faster over 40 km than a rider who chooses a consistent pacing strategy at the same average power output. So far, there has been no comprehensive attempt to replicate the research work of Swain (1997) with the mathematical model supplied by Martin et al. (1998). Using data supplied for the cyclist in Appendix I of Martin et al. (1998), we simulated the effects of varying power in parallel with variations in wind and gradient over a 50 km 'race'. We

found that if the rider produces a power output of 400 W over the 50 km race with no hills or wind, the time to completion would be 64 min 12 s. If the same rider produces the same power but against a headwind of 10 m - s-' for the first 25 km and with a tailwind of the same speed for the remaining 25 km, then the rider's time would be 76 min 10 s. Nevertheless, if the rider is able to increase power output by 20 W (5%) into the headwind section and decrease power in the tailwind section so that the average power over the whole race is still 400 W, then the 50 km time would be 75 min 7 s, a saving of over 1 min compared with a constant power strategy.

Atkinson and Brunskill (2000) confirmed such a 'time-saving' hypothesis when they examined the influ- ence of a constant versus variable pacing strategy during 16.1 km time-trials performed on an ergometer (Com- putrainer Pro, Racer Mate Inc., Seattle, WA) that could simulate changes in wind direction within the race. Race time was slightly better and overall perceived exertion was significantly lower when power output varied in parallel with the resistive force due to the simulated wind compared with a more constant pacing strategy.

Swain (1997) predicted that, in windy or hilly conditions, the more a rider can vary power, the more time is saved during a race. Such a prediction would depend on the distance raced in each headwind section and, hence, the length of race and nature of course. In the section on 'Physiological factors', we pointed out that elite cyclists can ride at about 90% of maximal oxygen consumption for 1 h, and so it would be impossible to vary oxygen consumption by more than 15%. Liedl et al. (1999) recently estimated acceptable power variation in a simulated laboratory time-trial. The mean power (corresponding to 78% V O ~ , ~ fkom an initial 1 h time-trial was used to examine the physiological and subjective strain of two pacing strategies: (1) constant pace at the mean work rate and (2) variable pace which consisted of alternate 5 min periods of riding 5% above and 5% below mean work rate. No differences in physiological strain were found between the two pacing strategies. Atkinson and Brunskill (2000) also found that riders were able to vary power by 5% during a 16.1 km simulated time- trial. As for the examination of pacing strategy in sprint events, it is still unclear what the exact trade-off is between physiological optimization of power produc- tion and biophysical optimization of cycling velocity.

Future directions Since the relative magnitude of starting power can be particularly important for time-trials, whether on the road or track, future research should attempt to identify the most appropriate way of monitoring starting power.

Dow

nloa

ded

by [B

ibliot

eca.

Unive

rsida

d de E

xtrem

adur

a], [$

{indiv

idualU

ser.di

splay

Name

}] at

04:07

05 M

arch 2

015

Science and cycling

Ideally, the rider should be trained in identifyng different starting powers and the influence of these powers on total race time should be examined. More information is also needed on which physiological or subjective measures are sensitive to subtle changes in pace.

To date, the suggestions of Swain (1997) have not been confirmed using the more recent and possibly more accurate mathematical models (e.g. Martin et al., 1998). Such work should simulate various race condi- tions of hills and wind that may be encountered by time-triallists. Future research should address whether training athletes to optimize the distribution of power in line with variations in the race environment makes a difference to overall race time. An important compo- nent of such investigations should be the identification of how much power can be varied without increasing overall physiological or subjective strain compared with a constant power strategy.

Nutritional strategies

The power outputs typically observed during amateur and professional cycling competitions (see Fig. 2) are associated with a very high energy expenditure. It can be estimated that a cyclist, with typical efficiency and aerodynamics, expends about 83 kJ . minpl when riding at 40 lun - h-' (Jeukendrup et al., 200 1). Professional cyclists can expend up to 120 kJ.min-' for longer periods of time, with most of this energy derived from the breakdown of muscle glycogen. During the Tour de France, energy expenditures may amount to 36 MJ - dayp1 (Saris et al., 1989). Riders in the Tour de France run a great risk of becoming depleted of muscle glycogen if energy costs are not repaid with adequate nutrition (Jeukendrup et al., 2001). A significant amount of heat can be produced as a by-product of the muscular work involved in cycling, particularly in hot ambient temperatures. Sweating is an effective mechanism to remove some of this heat from the body, albeit at the risk of hampering performance through dehydration, depending on the adequacy of fluid intake. Nutritional strategies designed to ameliorate the above problems are complicated but can be described in terms of time before, during and after a race.

The week before the race The classical studies of Bergstrom et al. (1967) and Hulnnan (1 967) demonstrated that endurance perfor- mance is increased when a high carbohydrate diet is adopted compared with a normal mixed diet in the week before a competition. There appears to be a linear relationship between the muscle glycogen concentra-

tion and the time to fatigue at a constant power. Although the original 'glycogen-loading' protocol in- cluded a so-called 'depletion phase', more recent studies indicate that this phase is not necessary in trained athletes (Sherman et al., 1981). In these early studies, time to exhaustion was investigated rather than performance in a time-trial. The relevance of time to exhaustion measurements for performance in real races has been questioned - that is, there are no races in which cyclists have to ride for as long as possible (Jeukendrup et al., 1996). Therefore, more studies are required that assess externally valid performance vari- ables rather than general endurance capacity.

The effects of elevated muscle glycogen on time-trial performance (i.e. to cover a certain distance as fast as possible) have been investigated by many research groups and, unlike the effects of carbohydrate loading on endurance capacity (time to exhaustion), the effects on time-trial performance appear to be less clear-cut. Researchers have generally found small effects of carbohydrate loading on time-trial performance lasting 1 h or less. In an excellent and complete review, Hawley et al. (1997) concluded that time-trial perfor- mance can be increased by 2-3% by carbohydrate loading. It is important to note, however, that in most studies the glycogen concentrations in the control condition were rather low and it is likely that no differences in time-trial performance can be observed when carbohydrate ingestion is 7 g - kgp1 - day-' or more. Interestingly, Coyle et al. (2001) recently demonstrated that cyclists who exercise daily (2 h at 65% ~ 0 ~ ~ ~ ) for 7 days were able to maintain or even increase their glycogen stores throughout the week. When they received a diet containing 11 g - kgh1 - dayp1 compared with a diet containing 7 g - kgp1 - dayp1, their muscle glycogen concentrations were elevated. In fact, the muscle glycogen concentrations in that study are among the highest ever reported (Coyle et al., 2001). Despite the higher glycogen stores on the high- carbohydrate diet, performance (as measured by means of a time-trial after 2 h of steady-state cycling) was not affected. Therefore, it was recently concluded that the intake of 7-10 g - kgp1 - day-' of carbohydrate should be sufficient to replenish the carbohydrate stores completely and to allow optimal performance (Jentjens and Jeukendrup, 2003). For a more detailed discussion of the role of muscle glycogen for exercise performance, the reader is referred to the recent reviews of Hawley et al. (1997) and Jentjens and Jeukendrup (2003).

Three to four hours before the race The ingestion of a carbohydrate-rich meal containing 140-330 g of carbohydrate, 3-4 h before exercise, has been shown to increase muscle glycogen (Coyle et al.,

Dow

nloa

ded

by [B

ibliot

eca.

Unive

rsida

d de E

xtrem

adur

a], [$

{indiv

idualU

ser.di

splay

Name

}] at

04:07

05 M

arch 2

015

Atkinson et al.

1985) and enhance exercise performance. After an overnight fast, liver glycogen can be substantially reduced (Nilsson and Hultman, 1973), which may result in difficulties in maintaining blood glucose concentrations during the subsequent exercise bout. The ingestion of carbohydrate may replenish liver glycogen reserves and contribute, together with any ongoing absorption of the ingested carbohydrate, to the maintenance of blood glucose.

Carbohydrate intake in the few hours before exercise has profound effects on metabolism. Generally, carbo- hydrate use is stimulated and fatty acid mobilization and oxidation are decreased. These metabolic changes can persist for up to 6 h after carbohydrate ingestion (Montain et al., 199 1). Nevertheless, the changes do not appear to be detrimental to exercise performance, because an increased carbohydrate availability compen- sates for the greater carbohydrate utilization.

One hour before the race Ingesting carbohydrate at rest results in a large increase in plasma glucose and insulin. However, when exercise is initiated 30-60 min after the ingestion of carbohy- drate, and even if carbohydrate is continued to be ingested, there is a rapid fall in plasma glucose as a consequence of the combined stimulatory effects of high insulin concentrations and muscle contraction on muscle glucose uptake.

Since the metabolic effects of pre-exercise carbohy- drate ingestion are a consequence of high glucose and insulin concentrations, there has been an interest in strategies that minimize the changes in plasma glucose and insulin before exercise. These strategies have included ingesting carbohydrate types (e.g. hctose) other than glucose, varying the carbohydrate load and ingestion schedule, adding fat, or including warm-up exercise in the pre-exercise period. In general, while these various interventions do modify the metabolic response to exercise, there appears to be no advantage for exercise performance in blunting the pre-exercise glucose and insulin concentrations (Hargreaves et al., 1987; Okano et al., 1988; Calles-Escandon et al., 199 1; Goodpaster et al., 1996). Nevertheless, there may be individuals who suffer more than others fkom the consequences of low blood glucose (hypoglycaemia) . When carbohydrate is ingested during prolonged exercise, the type of pre-exercise carbohydrate feeding has no effect on metabolism and performance (Burke et al., 1998).

It could be argued that if pre-exercise carbohydrate ingestion is the only mechanism by which a cyclist can increase carbohydrate availability during exercise, the cyclist would be well advised to ingest as much carbohydrate as possible, without undue gastrointest-

inal distress. This tactic would compensate for the reduced fat use and provide a pool of glucose that becomes available for use during the later stages of exercise. There appears to be little evidence to support the practice of avoiding carbohydrate ingestion in the hour before exercise. Individual procedures must be determined on the basis of individual experience with various pre-exercise carbohydrate ingestion protocols.

In addition to carbohydrate depletion, dehydration is a major contributor to fatigue in hot environments, with losses as little as the equivalent of 1-2% of body mass (600-1500 ml) associated with impaired exercise per- formance (Walsh et al., 1994). Thus, while preparing for one-day races, cyclists should ingest sufficient quantities of fluid to ensure that they are well hydrated before exercise. This is best monitored by body mass measurements. Ingesting large amounts of water may result in hyperhydration before exercise, but it is also likely to increase diuresis (fluid excretion) and gastro- intestinal distress (Maughan and Nadel, 2000). There has been some interest in adding glycerol to hydration beverages to enhance fluid retention and maintain a relative hyperhydration before and during exercise. The studies reported in the literature have produced conflicting results (Latzka et al., 1998; Hitchins et al., 1999), and there is the possibility that glycerol can promote intracellular dehydration with potential nega- tive consequences such as headaches. It is perhaps premature at this stage to recommend glycerol to achieve hyperhydration.

During the race

During prolonged cycling, energy has to be provided in the form of carbohydrate and fluid losses need to be replenished to prevent dehydration. Early research demonstrated that cyclists could ride 20-25% longer in rides to exhaustion at 65-70% VO~, (Coyle et al., 1986). More recent work with externally valid protocols has shown that performance in a simulated 40 km time- trial improved by more than a minute with carbohydrate ingestion compared with a water placebo (Jeukendrup et al., 1997). Consequently, it is generally believed that carbohydrate ingestion during exercise longer than about 45 min can improve exercise performance. During exercise of 90 rnin or longer, any improvement in performance due to dietary intervention is likely to be the result of a better maintenance of blood glucose concentration and high carbohydrate utilization rates. During high-intensity exercise, such as a 40 km time- trial, the mechanisms are less clear, as only small amounts of the ingested carbohydrate will become available to the working muscle. It has been suggested that the feeding during exercise may have central effects (Jeukendrup et al., 1997).

Dow

nloa

ded

by [B

ibliot

eca.

Unive

rsida

d de E

xtrem

adur

a], [$

{indiv

idualU

ser.di

splay

Name

}] at

04:07

05 M

arch 2

015

Science and cycling

The carbohydrate ingested during exercise will reduce the breakdown of liver glycogen (Jeukendrup et al., 1999a,b). It is generally believed that a larger contribution of ingested carbohydrate to energy ex- penditure is beneficial, partly because it will reduce the breakdown of endogenous carbohydrate sources (i.e. liver glycogen). It has been demonstrated that certain types of carbohydrates are oxidized more readily than others (Jeukendrup and Jentjens, 2000). Glucose, maltodextrins, sucrose, maltose and soluble starches can be oxidized at relatively high rates, whereas fructose and galactose are oxidized at lower rates.

The more carbohydrate is ingested, the more will be oxidized. However, when carbohydrate is ingested at a rate of about 1 .O-1.2 g min-l, additional carbohydrate intake will not result in increased utilization (Jeukendrup and Jentjens, 2000). Even the ingestion of very large amounts of carbohydrate will not increase the use over about 1 g - min-l.

Most cyclists do not drink enough fluid in races. Measurements during various professional cycling races (Tour de Andalusia, Mediterranean Tour, Tour of Switzerland, Tour de France) have shown that riders lose 2.1-4.5 kg of body mass after a typical stage. Some of this loss in body mass (100-300 g) may be carbohydrate and fat used during exercise, but most will be fluid loss. In some phases of the race, it is very difficult to drink simply because the hands are needed on the handlebars or there is very little time to drink. This is especially so in steep uphill sections, steep downhill sections and when riders are attacking. During off-road cycling, drinking may even be more difficult in the technical sections. Another problem is the fact that drink bottles are not always readily available and a maximum of about 1 litre can be carried on the bicycle.

After the race After exercise, and in particular stage races, it is important to replenish the fluid losses and to restore muscle glycogen. Depending on the environmental temperature, the replenishment of fluid may be more important, less important or equally important than the restoration of muscle glycogen. In most cases, riders would aim to achieve both these goals at the same time.

The effectiveness of post-exercise rehydration is mainly determined by the volume and composition of the fluid consumed. Water is not the ideal post-exercise rehydration beverage when rapid and complete restora- tion of body fluid balance is necessary (Maughan and Nadel, 2000). The ingestion of water alone in the post- exercise period results in a rapid fall in plasma sodium concentration and a drop in plasma osmolality.

These changes have the effect of reducing the stimulation to drink (thirst) and increasing the urine

output, both of which will impair the rehydration process. Plasma volume is more rapidly and completely restored in the post-exercise period if some sodium chloride (77 mmol-1-') is added to the water con- sumed (Nose et al., 1998). This concentration is similar to the upper limit of the sodium concentration found in sweat, but is considerably higher than the sodium concentration of many commercially available sports drinks, which usually contain 1 0-25 mmol - 1 -'.

Complete rehydration after exercise can only be achieved if the sodium lost in sweat is replaced as well as the water. T o achieve euhydration after exercise, the sodium intake must be greater than the sodium loss. Ingesting a beverage containing sodium not only promotes rapid fluid absorption in the small intestine, but also allows the plasma sodium concentration to remain elevated during the rehydration period and helps to maintain thirst while delaying stimulation of urine production. The inclusion of potassium in the beverage consumed after exercise is thought to enhance the replacement of intracellular water and thus promote rehydration, but currently there is little experimental evidence to support this. The rehydration drink should also contain carbohydrate (glucose or glucose poly- mers) because the presence of glucose will also stimulate fluid absorption in the gut and improve beverage taste. After exercise, the uptake of glucose into the muscle for glycogen resynthesis should also promote intracellular rehydration.

Until recently, it was generally recommended that athletes should consume a volume of fluid equivalent to their sweat loss incurred during exercise to rehydrate adequately in the post-exercise recovery period. It was recommended that athletes consume about 1 litre of fluid for every kilogram lost during an exercise session. It is now clear that this amount of fluid is insufficient, because it does not take into account the urine losses that occur after beverage consumption over a period of hours. Existing data indicate that the ingestion of 150% or more of body mass loss (i.e. 1.5 litres of fluid consumed during recovery for every kilogram of weight lost during exercise) may be required to achieve normal hydration within 6 h after exercise (Maughan and Shirreffs, 1996).

Cyclists should be encouraged to consume solid food as well as fluid between exercise bouts, unless food intake is likely to result in gastrointestinal disturbances. In one s t ~ d y j the same fluid volume consumed as a meal plus water combination compared with a sports drink resulted in a smaller volume of urine being produced and hence greater fluid retention (Maughan and Shirreffs, 1996). The greater efficacy of the meal plus water treaunent in restoring whole-body fluid balance was probably a consequence of its greater total sodium and potassium content. In exercise conditions

Dow

nloa

ded

by [B

ibliot

eca.

Unive

rsida

d de E

xtrem

adur

a], [$

{indiv

idualU

ser.di

splay

Name

}] at

04:07

05 M

arch 2

015

Atkinson et al.

where sweat losses are large, the total amount of sodium and chloride lost will be high. For example, the loss of 10 litres of sweat, with a sodium concentration of 50 mmol- 1-', amounts to a loss of about 29 g of sodium chloride. Clearly, food intake can be important in restoring these salt losses, since as mentioned above most commercially available sports drinks contain only about 10-25 mmol - 1 -'.

Figure 5 illustrates maximal muscle glycogen synth- esis rates reported in response to different rates of carbohydrate ingestion in the early hours post-exercise. It can be seen that the maximal rate of muscle glycogen resynthesis is reached at a carbohydrate intake of between 1.2 and 1.4 g - min-' (75-90 g of carbohydrate per hour). A carbohydrate intake of more than 90 g per hour provides no additional benefits in terms of muscle glycogen storage and may only increase the risk of gastrointestinal discomfort. The timing of carbohydrate intake is also important when a fast recovery is required. Complete glycogen repletion can be accomplished within 24 h provided the amount and timing of carbohydrate is right (Jentjens and Jeukendrup, 2003). The most rapid replenishment of glycogen stores occurs in the first 60-90 min after exercise. In one study, it was found that the rate of glycogen storage was almost twice as fast if carbohydrate supplements were provided immediately after exercise rather than several hours after (Ivy et al., 1988).

The highest rates of muscle glycogen synthesis have been reported in studies in which carbohydrates were

provided at regular intervals (every 15-30 min) (Jent- jens and Jeukendrup, 2003). This is most probably due to the maintenance of high blood glucose and insulin concentrations for a longer period. Thus, eating small meals frequently appears to have an extra benefit over eating fewer large meals. Additionally, eating small meals is likely to reduce the risk of gastrointestinal discomfort.

Most forms of carbohydrates will provide similar amounts of glucose for glycogen storage. When fructose is ingested, the rates of glycogen synthesis are relatively low because fructose has to be converted to glucose in the liver before it can be used for glycogen synthesis in the muscle. Most types of carbohydrate, however, will result in similar muscle glycogen resynthesis rates (glucose, sucrose, glucose polymers).

There appears to be no difference in muscle glycogen synthesis when either liquid or solid forms of carbohydrate are consumed in the first few hours after exercise (Keizer et al., 1987). However, in the study of Keizer et al., the total amount of carbohy- drate ingested was small and glycogen resynthesis rates were suboptimal. Liquid forms of carbohydrate or carbohydrate foods with a high fluid content are often recommended to athletes because they are easy to digest and are less filling and, therefore, do not tend to affect one's normal appetite Liquid carbohydrate supplements also source of fluid that may be beneficial rehydration.

50 1 - - - - - Maximal syntb~sis.rate. . . . . . . . . . . . . . . -a

Carbohydrate intake (g . m i d )

adversely. provide a for rapid

Fig. 5. Maximal muscle glycogen synthesis rates after the ingestion of different amounts of carbohydrate in the first few hours after exercise.

Dow

nloa

ded

by [B

ibliot

eca.

Unive

rsida

d de E

xtrem

adur

a], [$

{indiv

idualU

ser.di

splay

Name

}] at

04:07

05 M

arch 2

015

Science and cycling

Insulin promotes muscle glycogen synthesis after exercise and any means by which insulin concentrations can be elevated could help the resynthesis of glycogen after exercise. Some amino acids and proteins will cause insulin release from the pancreas and several studies have shown that co-ingestion of protein and/or amino acids with carbohydrate almost doubled the insulin response (Zawadzki et al., 1992; Van Loon et al., 2000; Jentjens et al., 2001). Muscle glycogen synthesis was also increased in these studies by 40-1 00% (Zawadzki et al., 1992; Van Loon et al., 2000). However, a recent study (Jentjens et al., 2001) has shown that when the total carbohydrate intake is very high (1.2 g - kgp1 body mass per hour), the presence of a protein-amino acid mixture does not further increase the rate of muscle glycogen synthesis despite a much higher insulin response. Therefore, there is no need to ingest proteins or amino acids with carbohydrate to maximize glycogen synthesis when the total carbohydrate intake is suffi- cient.

Future directions Given the prevalence of stage races in cycling and the general frequency of racing described in the section on 'Physiological factors', future researchers on nutrition could emphasize the influence of diet on recovery from exercise. New strategies to improve glycogen resynth- esis and to reduce protein breakdown after repeated bouts of exercise should be explored.

Although protein intake may not always have an effect on muscle glycogen synthesis, there is some evidence that the intake of essential amino acids in the hours after exercise may increase protein synthesis (Rasmussen et al., 2000). Increased protein synthesis after exercise could help the repair of muscle damage and synthesis of various enzymes and mitochondria. To date, research work in the latter area has been limited, because the techniques to study protein metabolism are complicated and very expensive. In addition, refinements should be made through the results of empirical research to optimize the fluid and carbohydrate delivery from sports drinks, especially under different conditions of heat strain on thermo- regulation.

Synopsis

In this review, we have attempted to adopt a 'holistic' approach to analysing cycling performance. The factors that influence cycling power and velocity have been considered with an appreciation of both the nature of the different disciplines of cycling and the interrelation- ships between the different mediating factors.

The characteristics of cyclists that are common across most disciplines of cycling include a high maximal aerobic power output and an ability to work at relatively high power outputs for extended periods of time. The relationship between these characteristics and inherent physiological factors such as muscle capilliarization and muscle fibre type is complicated by inter-individual differences in the selection of cadence in different race conditions. The physiological de- mands of many cycling disciplines have yet to be fully investigated.

Recent advances in the mathematical modelling of resistive forces on the bicycle are starting to help unravel the interrelationships between these resistive forces. Data from these models may be supplemented with information derived from the use of bicycle- mounted strain gauges that can measure power directly in races. All-terrain bicycle racing is a neglected discipline in terms of characterizing power outputs in race conditions and the modelling of the effects of the quite different bicycle components and associated resistive forces.

Pacing strategy during cycling depends on the relative magnitude of certain resistive forces during a race. A constant power output distribution, with a 'blunted' starting power, is best for longer ( > 4 lun) time-trials held in conditions of unvarying wind and gradient. For shorter track racing, a more powerful start is advised to 'save' time during the phase that contributes most to total race time and to optimize the kinetic energy contribution to total race time. The optimum pacing strategy for road time-mals may need to take into account varying conditions of hills and wind, but more research is needed to elucidate fully how much a variable power output can be tolerated by a rider and how much time such a pacing strategy might save in a real race or one simulated with the latest mathematical models that are available.

Diet is a multifactorial issue and many researchers have tried to examine aspects of cycling nutrition (e.g. timing, amount, composition) in isolation. Only re- cently have researchers attempted to analyse interrela- tionships between dietary factors (e.g. the link between pre-race and in-race dietary effects on performance). The thermal environment is a mediating factor in choice of diet, since there may be competing interests of replacing lost fluid and depleted glycogen during and after a race.

Future directions for research should centre on closing more 'loops' between the different factors that influence cycling performance. Researchers working in cycling science should have a good appreciation of the limitations of their work in terms of which riders, which race environment and which discipline of competitive cycling their results are relevant to.

Dow

nloa

ded

by [B

ibliot

eca.

Unive

rsida

d de E

xtrem

adur

a], [$

{indiv

idualU

ser.di

splay

Name

}] at

04:07

05 M

arch 2

015

Atkinson et al.

References

Ahlquist, L.E., Bassett, D.R., Sufit, R., Nagle, F.J. and Thomas, D.P. (1992). The effect of pedaling frequency on glycogen depletion rates in type I and type I1 quadriceps muscle fibers during submaximal cycling exercise. Euro- pean Journal of Applied Physiology, 65, 360-364.

Atkinson, G. and Brunskill, A. (2000). Pacing strategies during a cycling time trial with simulated headwinds and tailwinds. Ergonomics, 43, 1449-1460.

Balmer, J., Davison, R.C. and Bird, S.R. (2000). Peak power predicts performance power during an outdoor 16.1-km cycling time trial. Medicine and Science in Sports and Exercise, 32, 1485-1490.

Bassett, J.R., Kyle, C.R., Passfield, L., Broker, J.P. and Burke, E.R. (1999). Comparing cycling world hour records, 1967-1 996: modelling with empirical data. Medicine and Science in Sports and Exercise, 31, 1665-1676.

Bastiaans, J.J., van Diemen, A.B., Veneberg, T. and Jeukendrup, A.E. (2001). The effects of replacing a portion of endurance training by explosive strength training on performance in trained cyclists. European Journal of Applied Physiology, 86, 79-84.

Bergstrom, J., Hermansen, L., Hultman, E. and Saltin, B. (1967). Diet, muscle glycogen and physical performance. Acta Physwlogica Scandinavica, 71, 140-1 50.

Billat, V.L. (1996). Use of blood lactate measurements for predmion of exercise performance and for control of training. Sports Medicine, 22, 1 57-1 75.

Bishop, D., Jenkins, D.G., McEniery, M. and Carey, M.F. (2000). Relationship between plasma lactate parameters and muscle characteristics in female cyclists. Medicine and Science in Sports and Exercise, 32, 1088-1093.

Bishop, D., Bonetti, D. and Dawson, B. (2002). The influence of pacing strategy on V O ~ and supramaximal kayak performance. Medicine and Science in Sports and Exercise, 34, 1041-1047.

Broker, J.P., Kyle, C.R. and Burke, E.R. (1999). Racing cyclist power requirements in the 4000-m individual and team pursuits. Medicine and Science in Sports and Exercise, 31, 1677-1685.

Burke, E.R. (2000). Physiology of cycling. In Exercise and Sport Science (edited by W.E. Garrett and D.T. Kirken- dall), pp. 759-770. Philadelphia, PA: Lippincott Williams & Wilkins.

Burke, L.M., Claassen, A., Hawley, J.A. and Noakes, T.D. (1 998). Carbohydrate intake during prolonged cycling minimizes effect of glycemic index of preexercise meal. Journal of Applied Physiology, 85, 2220-2226.

Calles-Escandon, J., Devlin, J.T., Whitcomb, W. and Horton, E.S. (1991). Pre-exercise feeding does not affect endurance cycle exercise but attenuates post-exercise starvation-like response. Medicine and Science in Sports and Exercise, 23, 8 18-824.

Capelli, C., Rosa, G., Butti, F., Ferretti, G., Veicsteinas, A. and di Prampero, P.E. (1993). Energy cost and efficiency of riding aerodynamic bicycles. European Journal of Applied Physiology and Occupational Physiology, 67, 1 44- 1 49.

Coyle, E.F. (1995). Integration of the physiological factors determining endurance performance ability. In Perspectives in Exercise Science and Sports Medicine (edited by C.V. Gisolfi and D.R. Lamb), pp. 25-63. Indianapolis, IN: Benchmark Press.

Coyle, E.F., Coggan, A.R., Hemmert, M.K., Lowe, R.C. and Walters, T.J. (1985). Substrate usage during pro- longed exercise following a pre