Bosquet1, 3
A. Duchene1
V̇O2 Kinetics during Supramaximal Exercise:
G. Dupont1, 2
Relationship with Oxygen Deficit and 800-m L. Leger 3
Running Performance H. Carter 4
Training & Testing
Abstract plitude of the V̇O2 response (60.0 ± 5.0 and 53.5 ± 5.7 ml · min–1 ·
kg–1, respectively) and V800 m (6.27 ± 2.1 and 5.45 ± 0.38 m · s–1, re-
The aim of this study was to compare V̇O2 kinetics of highly- ver- spectively). O2def (24.6 ± 2.7 and 27.7 ± 7.8 ml · kg–1, respectively)
sus recreationally-trained subjects during a constant velocity and the gain of the V̇O2 response (193 ± 14 and 194 ± 13 ml · kg–1 ·
test of supramaximal intensity. Eighteen trained male subjects m–1, respectively) were similar between groups. τ was associated
were recruited to one of two groups: highly trained (HT, n = 8, with O2def (r = 0.90, p < 0.05), but not with V800 m (r = 0.30,
V̇O2max = 70.1 ± 6.5 ml · min–1 · kg–1) and recreationally trained (RT, p > 0.05). It was concluded that HT subjects exhibited faster on-
n = 10, V̇O2max = 63.2 ± 6.4 ml · min–1 · kg–1). All subjects performed kinetics and higher amplitude than their RT counterparts. The
an incremental test to exhaustion for the determination of higher amplitude was not thought to reflect any difference in
V̇O2max and peak treadmill velocity (PTV), two constant velocity underlying physiological mechanisms. The faster τ, whose exact
tests at 110% of PTV to determine V̇O2 kinetics and oxygen deficit mechanisms remain to be elucidated, may have practical impli-
(O2def), and a 800-m time trial to determine running perfor- cations for coaches.
mance (mean velocity over the distance, V800 m). We found signif-
icant differences between HT and RT for the on-transient of the Key words
V̇O2 response (τ, 24.7 ± 3.3 and 30.9 ± 7.0 s, respectively), the am- Middle distance running · oxygen deficit · fatigue · training
518
Introduction rest to the intensity required by this kind of event (110 to 120%
of V̇O2max) [7, 45, 46] are probably a discriminating parameter of
The physiological determinants of events lasting ∼ 2 minutes, performance. It is generally assumed that V̇O2 rises exponen-
like the 800 m in running, are particular in that they require a tially to match the energy production with the energy required
significant contribution of both anaerobic (34 to 40 %) and aero- for the exercise. In the case of supramaximal intensity exercise,
bic (60 to 66 %) pathways to cover the total energy demand where the energy demand exceeds V̇O2max, V̇O2 will tend toward
[11,16, 45], and that anaerobic capacity is nearly completely [11] its maximum, and eventually reach it if there is sufficient time
or completely [32] utilized at the end of the race. for the primary component of the V̇O2 response to develop com-
pletely [49]. In this theoretical frame, a faster V̇O2 kinetic re-
Since performance in this time domain relies on the interplay be- sponse at the same relative work rate will reduce the magnitude
tween aerobic and anaerobic pathways, the V̇O2 kinetics from of the oxygen deficit (O2def) [17,19], which is an indirect method
Affiliation
1
Faculty of Sport Sciences, University of Lille 2, Ronchin, France
2
Faculty of Sport Sciences, University of Artois, Liévin, France
3
Department of Kinesiology, University of Montreal, Montreal, Canada
4
Chelsea School Research Centre, University of Brighton, Eastbourne, United Kingdom
Correspondence
Laurent Bosquet · Department of Kinesiology · University of Montreal · CP 6128, succ. centre ville ·
Montreal · Canada H3C 3J7 · Phone: + 5143 43 89 49 · Fax: + 5143 43 2181 ·
E-mail: laurent.bosquet@umontreal.ca
Bibliography
Int J Sports Med 2007; 28: 518 – 524 © Georg Thieme Verlag KG · Stuttgart · New York ·
DOI 10.1055/s-2006-955896 · Published online March 15, 2007 ·
ISSN 0172-4622
of measuring anaerobic metabolism in the muscle [21]. It is ex-
pected that this faster adjustment of aerobic pathway will spare Table 1 Main characteristics of highly trained (HT) and recreational-
the fall in intramuscular phosphocreatine, which may attenuate ly trained (RT) groups. Values are mean ± SD
the accumulation of hydrogen ions and allow for a greater avail-
ability of stored energy sources at the end of the race [26]. HT (n = 8) RT (n = 10)
It is clear from numerous studies that enhanced physical fitness Age (years) 22.9 ± 4.1 23.7 ± 2.8
results in significant alterations of the V̇O2 kinetics in the heavy Height (cm) 180 ± 6 179 ± 6
intensity domain, including a faster on-transient (the speed of Weight (kg) 68.9 ± 6.8 75.2 ± 8.9
the rise in oxygen uptake from exercise onset until the “steady V̇O2max (ml · min–1 · kg–1 ) 70.1 ± 6.5 63.2 ± 6.4a
state”) and a smaller amplitude of the V̇O2 slow component (a
a
significantly different from HT (p < 0.05)
delayed and elevated phase of the oxygen uptake response to
heavy intensity exercise) [4, 29]. Surprisingly, very few studies
have reported values for the V̇O2 kinetics in the supramaximal
Bosquet L et al. V̇O2 Kinetics and Performance … Int J Sports Med 2007; 28: 518 – 524
the subject’s feet astride the moving belt and hands holding the where V̇O2(t) is the value for V̇O2 (ml) at time = t, V̇O2(baseline) is the
handrail. Time was measured to the nearest second from the mo- mean V̇O2 of the 2 min preceding the beginning of the test,
ment the subject released the handrail (usually less than 3 sec- V̇O2(amplitude) is the difference between the asymptotic value of
onds) until he grasped it again to signal exhaustion. Strong verbal the exponential term and V̇O2(baseline), and τ is the time constant
encouragement was given throughout each test. V̇O2 and related (s). O2def (ml · kg–1) was the product of τ (converted in min) and
gas exchange measures were determined breath-by-breath using V̇O2max [48]. O2def was also computed by taking V̇O2 at τ instead
an automated cardiopulmonary exercise system (Moxus, AEI of V̇O2max, in order to have an estimate of O2def at the beginning
Technologies, Naperville, IL, USA). The metabolic cart was cali- of exercise. The “gain” of the response (Gamplitude = V̇O2(amplitude)/∆
brated according to the same procedure as the incremental test. running speed, expressed relative to body mass) was also calcu-
lated.
Time trial: an 800-m race was performed on a synthetic indoor
track. Subjects were matched according to their personal best or Statistical analysis
their PTV when no performance was available, and ran in homo- Standard statistical methods were used for the calculation of
geneous groups of two or three. Subjects were encouraged to means and standard deviations. Normal Gaussian distribution
Training & Testing
maintain an even pace throughout the run in order to produce of the data was verified by the Shapiro-Wilk test, and homosce-
the best performance. Performance was measured to the nearest dasticity by a modified Levene test. A logarithmic transformation
0.1 second. was performed before the analysis when data exhibited hetero-
scedasticity. A Student’s t-test was used to evaluate the differ-
Data analysis ences between HT and RT subjects for relevant parameters. The
Determination of maximal oxygen uptake (V̇O2max), GET, and magnitude of the difference was assessed by the effect size (ES).
PTV: mean values of V̇O2, ventilatory equivalent of O2 (V̇E/V̇O2) Since there was no control group per se, pooled standard devia-
and ventilatory equivalent of CO2 (V̇E/V̇O2) were displayed every tion was used to compute this statistic. The scale proposed by
30 seconds during the test. As suggested by Duncan et al. [12], Cohen [5] was used for interpretation. The magnitude of the dif-
the primary criterion for the attainment of V̇O2max was a plateau ference was considered either small (0.2), moderate (0.5), or
in V̇O2 despite an increase in running velocity. In the absence of a large (0.8). Pearson product-moment correlation was used to
plateau, secondary criteria included a respiratory exchange ratio evaluate the association between relevant parameters. Statistical
of 1.15 or greater, and a post-exercise [La–]b of 8 mmol · l–1 or significance was set at p = 0.05 level for all analysis. All calcula-
greater [12]. GET was defined as the velocity at which there was tions were made with Statistica 6.0 (Statsoft, Tulsa, OK, USA).
a nonlinear increase in V̇E/V̇O2 without a concomitant nonlinear
increase in V̇E/V̇O2 [47]. PTV was computed from the duration
maintained during the last stage of the incremental test with Results
the following equation [30]:
520 The results obtained during the incremental test and the 800-m
PTV = v + (a · n/b) (Eq. 1) time trial are reported in Table 2. We found a large difference be-
tween HT and RT for PTV, V̇O2max, and performance in the 800 m.
where v is the velocity (m · s–1) of the last completed stage, a is The impact of these differences in aerobic fitness and perfor-
the increment of the test (m · s–1), n is the duration (s) maintained mance capacity on V̇O2 kinetics during an exercise bout in the
after the first 60 s of the stage and b is the duration (s) of the supramaximal intensity domain is presented in Table 3. There
stage. was no difference in baseline V̇O2 according to the level of exper-
tise. We found faster on-kinetics (τ) and higher amplitude for HT,
Modelling of the V̇O2 kinetic response and determination of oxy- which resulted in a higher end-exercise V̇O2 (baseline + ampli-
gen deficit (O2def): raw breath-by-breath V̇O2 data measured tude). This latter difference seemed to be primarily due to the dif-
during the two constant velocity tests were edited so that noisy ference in running velocity and not underlying physiological
breaths caused by swallowing, coughing or sighing, which were mechanisms, since Gamplitude was similar between groups (Fig. 1),
not reflective of the underlying kinetics, could be removed. A val- whether it was computed from end-exercise V̇O2 (193 ± 14 and
ue was considered noisy if it was greater than 4 standard devia- 194 ± 13 ml · kg–1 · m–1 for HT and RT, respectively) or from V̇O2(am-
tions from the local mean [36]. The signals from both tests were –1
plitude) (169 ± 8 and 170 ± 10 ml · kg · m
–1
for HT and RT, respec-
then resampled to second-by-second and ensemble-averaged to tively). End-exercise V̇O2 was also independent of τ (r = – 0.32,
produce an average response. The duration of this response was p > 0.05), but was associated with V̇O2max (r = 0.94, p < 0.05). It is
that of the shortest time to exhaustion (207 ± 34 and 173 ± 42 s noteworthy that there was no difference between HT and RT
for RT and HT, respectively, p > 0.05). The initial cardiodynamic when end-exercise V̇O2 was expressed in % V̇O2max, but also that
component (i.e., the first 20 s after the onset of exercise) was ex- the capacity to reach a high % V̇O2max during an exercise of su-
cluded. The iterative nonlinear least square modelling procedure pramaximal intensity seems to be independent from both τ
within Statistica 6.0 (Statsoft, Tulsa, OK, USA) was then used to fit (r = – 0.21, p > 0.05) and V̇O2max (r = 0.20, p > 0.05). There was no
this new regular signal from the 2 min preceding the beginning difference in O2def between groups, whether its maximal value
of the test until the endpoint of the shorter of the two tests, with (28.77 ± 3.86 vs. 32.71 ± 8.93 ml · kg–1), or when it was computed
the following monoexponential model, which has been shown to at the beginning of exercise by taking V̇O2 at τ instead of V̇O2(am-
be relevant for such exercise intensities [10]: –1
plitude) (15.49 ± 1.79 vs. 17.46 ± 4.92 ml · kg ). The correlation ma-
trix between performance in the 800-m time trial and V̇O2 ki-
V̇O2(t) = V̇O2(baseline) + V̇O2(amplitude) – (1-e(-t/τ)) (Eq. 2) netics parameters, O2def, PTV and V̇O2max is presented in Table 4.
Bosquet L et al. V̇O2 Kinetics and Performance … Int J Sports Med 2007; 28: 518 – 524
Table 2 Results from the incremental test and the time trial for the highly trained (HT) and recreationally trained (RT) groups. Values are mean
± standard deviation
Incremental test
PTV (m · s–1 ) 5.38 ± 0.25 4.75 ± 0.33 p < 0.001 2.12
V̇O2max (ml · min–1 · kg–1 ) 70.1 ± 6.5 63.2 ± 6.4 p < 0.05 1.06
[La–]b, peak (mmol · l–1 ) 8.91 ± 1.80 9.82 ± 1.75 n.s. 0.51
800 m-time trial
Performance (s) 127.4 ± 4.5 147.4 ± 10.3 p < 0.001 2.59
–1
V800 m (m · s ) 6.27 ± 0.21 5.45 ± 0.38 p < 0.001 2.59
PTV: peak treadmill velocity; [La ]b, peak: peak blood lactate concentration; V800 m: mean velocity over the 800-m time trial; n.s.: nonsignificant (p > 0.05); : 7.79 m · s–1 by
– a
Table 3 V̇O2 kinetics for highly trained (HT) and recreationally trained (RT) groups. Values are mean ± standard deviation (95% confidence
interval)
HT RT P (effect size)
Baseline (ml · kg–1 · min–1 ) 8.6 ± 2.4 (6.9 to 10.2) 7.3 ± 0.6 (6.9 to 7.7) n.s. (0.77)
Amplitude (ml · kg–1 · min–1 ) 60.0 ± 5.0 (56.5 to 63.5) 53.5 ± 5.7 (49.9 to 57.0) p < 0.05 (1.21)
End-exercise V̇O2 (ml · kg–1 · min–1 ) 68.6 ± 6.1 (64.3 to 72.8) 60.7 ± 5.5 (57.3 to 64.1) p < 0.05 (1.35)
End-exercise V̇O2 (%V̇O2max) 97.9 ± 3.2 (95.7 to 100.2) 96.2 ± 4.3 (93.6 to 98.9) n.s. (0.44)
τ (s) 24.7 ± 3.3 (22.4 to 27.0) 30.9 ± 7.0 (26.6 to 35.2) p < 0.05 (1.09)
521
Table 4 Correlation matrix between performance in the 800-m
time trial (s) and relevant parameters
The aim of this study was to compare the V̇O2 kinetics of highly- faster on-kinetics and a higher amplitude than their recreation-
versus recreationally-trained subjects during a constant velocity ally trained counterparts.
test in the supramaximal intensity domain, in relationship with
running performance in the 800 m. The salient fact of our study Contrary to our hypothesis, this difference in the V̇O2 response to
was that the difference in performance between highly- and rec- supramaximal exercise did not result in a significant difference
reationally-trained runners was associated with significant alter- in O2def between groups at the beginning of exercise. Such a dis-
ation of the V̇O2 on-kinetics. Highly trained runners exhibited crepancy with expected results is surprising, since both param-
Bosquet L et al. V̇O2 Kinetics and Performance … Int J Sports Med 2007; 28: 518 – 524
eters should follow each other. It should be noted, however, that action potential at high frequency [42], is more pronounced in
τ and O2def were clearly associated in our study (r = 0.90, type II muscle fibers [27]. Moreover, the decline in motoneurons
p < 0.001). Although interpretation of this correlation should be firing rates that characterize central fatigue has also been shown
made with caution (these measures are not independent), they to be higher in fast motor units [15]. Other mechanisms like a re-
suggest, however, that subjects with shorter τ exhibit lower duction in calcium release from the sarcoplasmic reticulum [13],
O2def. Consequently, the lack of difference between groups was a calcium desensitization at the neuromuscular junction [43] or
probably a statistical matter. In fact, statistical power of τ was at the troponin C level [18], or a failure of the action potential to
∼ 0.70 for the sample size used at the 0.05 alpha level, while it propagate along each branch of a motor axon [44] are some
was only ∼ 0.20 for O2def. It would have been necessary to raise mechanisms among a pleiad of others, that may also contribute
the number of subjects to ∼ 40 to reach a statistical power com- to fatigue or exhaustion in intense, short, exhaustive exercise.
parable to τ at the 0.05 alpha level, and eventually make the dif- Although it is only speculation, it is likely that such mechanisms
ference between groups significant. In parallel to this statistical were largely involved in our study, and contributed substantially
consideration, it should be kept in mind that the determination to the cascade of events leading to exhaustion before attainment
of O2def is associated with many shortcomings that clearly ques- of V̇O2max.
Training & Testing
tion its validity [1, 2]. These apply to the method of Whipp [48],
but also to other methods available in the literature [23, 32]. In PTV was the better predictor of running performance in the
addition to the undeniable lack of statistical power for this de- 800 m among the parameters we measured (r = – 0.88, p < 0.05).
pendant variable, the determination of O2def per se probably Such a close association has already been reported in the litera-
contributed to the fact that the alteration of τ was not accompa- ture for running events ranging from 800 m to the half-marathon
nied by a concomitant alteration of O2def. [8, 31, 34, 37]. The fact that it was better correlated with perfor-
mance than V̇O2max alone was expected, since PTV is determined
Performance level was associated with a higher amplitude of the by both V̇O2max and the energy cost of running [8]. The absence of
V̇O2 response during the constant velocity test for HT, which re- a relationship between O2def and performance in events of sim-
sulted in a higher end-exercise V̇O2. This difference seemed to be ilar duration than the 800 m has already been reported in the lit-
clearly associated with a higher V̇O2max in HT (r = 0.94, p < 0.05). erature [6, 35, 41]. Gastin [16] computed from the results of over
However, when end-exercise V̇O2 was expressed as a percentage 30 studies that anaerobic contribution accounted for 30 to 40 %
of V̇O2max, it was not different between groups, representing of the variance in events lasting 120 seconds, which corresponds
∼ 97 %. Together with similar Gamplitude between groups, the result to a maximal correlation of ∼ 0.60. As already underlined by Craig
of this normalization suggests that the difference of end-exercise and Morgan [6], it means that even under perfect conditions,
V̇O2, when expressed in absolute units, does not reflect any dif- O2def would not be expected to be a strong predictor of perfor-
ference in underlying physiological mechanisms. The non-at- mance. The same applied to τ. The fact that it was not signifi-
tainment of V̇O2max, despite reaching exhaustion, has consis- cantly correlated with performance reflected partly a relatively
522 tently been reported in exercise of such duration and intensity. poor statistical power (1-β = 0.24, n = 84 to obtain 0.80), but also
Hill et al. [22], as well as Spencer et al. [45] and Draper and a lower quantitative influence of this variable than other deter-
Wood, [9,10] found a ∼ 90 % V̇O2max at the end of supramaximal minants like V̇O2max. This does not mean, however, that it was
exercise lasting ∼ 120 seconds. This is clearly lower than our re- not a determinant of running performance in the 800 m, since it
sults. It is, however, worth noting that the duration of the shorter was different between HT and RT. We did not perform a multiple
trial, which was used to produce the average V̇O2 response, was stepwise regression because of the small number of subjects and
substantially longer in our study (192 ± 40 seconds). The primary the redundancy of some parameters (amplitude, end-exercise
component of the V̇O2 response had more time to develop, thus V̇O2 and V̇O2max for example).
reaching a higher % V̇O2max.
Some practical applications can be derived from the results of
The fact that V̇O2max was not reached at the end of exercise is in- this study. As expected, HT subjects exhibited faster on-kinetics
teresting considering discussions about underlying mechanisms than their RT counterparts during an exhaustive bout of supra-
causing fatigue, other than O2 delivery or utilization to/by the maximal intensity. It was not the scope of this study to under-
muscle. According to the size principle of Henneman [20], any in- stand the mechanisms underlying such a difference. Although
crease in exercise intensity will ensue with a higher recruitment this information has not been collected in our study, the training
of type II muscle fibers. It is therefore expected that more fati- background was likely to play a very important role, but also the
gable and also less efficient type II [24] muscle fibers will con- physiological adaptations that may accompany it, such as an in-
tribute substantially to the total power output during an exercise crease in muscle mitochondrial content [3] or a modification of
of supramaximal intensity, but with a greater relative contribu- the muscular typology [28]. In this respect, the optimal experi-
tion of anaerobiosis than type I muscle fibers. Such an allegation mental approach, although very difficult to achieve, would have
is coherent with the lower Gamplitude that has already been report- been to compare two groups of similar training background, but
ed in the severe and supramaximal intensity domains [4, 40]. different performance level. Whatever the reason for this differ-
Saltin et al. [39] pointed out that excitation-contraction failure ence, the main advantage was probably a decrease of anaerobic
coupled with reduced nervous drive were probably the main contribution at the beginning of exercise, although we were not
sources of fatigue and exhaustion during an intense, short, ex- able to provide indisputable statistical evidence in the context of
haustive exercise. In fact, it is well established that potassium ac- this study. The results of Millet et al. [33] suggest that increasing
cumulation in the interstitial space, which may result in an in- intensity of supramaximal exercises during an interval training
ability of the sarcolemma and the T-tubule system to propagate session induce a faster adjustment of the V̇O2 response primary
Bosquet L et al. V̇O2 Kinetics and Performance … Int J Sports Med 2007; 28: 518 – 524
component. Now, it remains to determine if the repetition of 9
Draper SB, Wood DM. The oxygen uptake response of sprint- vs. en-
such sessions would involve a chronic improvement of τ. If such durance-trained runners to severe intensity running. J Sci Med Sport
2005; 8: 233 – 243
a fact was confirmed, it would provide short distance/duration 10
Draper SB, Wood DM. The V̇O2 response for an exhaustive treadmill
coaches with a relevant method to improve this specific determi- run at 800-m pace: a breath-by-breath analysis. Eur J Appl Physiol
nant of performance. In other respects, this study, as well as 2005; 93: 381 – 389
11
others [9,10, 22, 45], showed that V̇O2max was not reached at the Duffield R, Dawson B, Goodman C. Energy system contribution to
400-metre and 800-metre track running. J Sports Sci 2005; 23:
end of a supramaximal running exercise. As already discussed, it
299 – 307
suggests that neuromuscular fatigue is probably of more concern 12
Duncan GE, Howley ET, Johnson BN. Applicability of V̇O2max criteria:
than the metabolic one to explain the reasons of exhaustion in discontinuous versus continuous protocols. Med Sci Sports Exerc
this kind of exercise [39]. Although anaerobic contribution is 1997; 29: 273 – 278
13
Favero TG. Sarcoplasmic reticulum Ca(2+) release and muscle fatigue.
quantitatively minor in the total energy supply of events lasting
J Appl Physiol 1999; 87: 471 – 483
∼ 2 minutes [11,16, 45], anaerobic training involving strength and 14
Elrick R, Gamboa JD, Martin DE, Mora AH, Patterson M, Piqueras MP,
speed training, because it has a great impact on neuromuscular Schmidt P, Vittori C. Speed in the 800 metre. New Studies in Athletics
function [38], is qualitatively more important than a simple bio- 1996; 11: 7 – 22
Bosquet L et al. V̇O2 Kinetics and Performance … Int J Sports Med 2007; 28: 518 – 524
35 42
Olesen HL, Raabo E, Bangsbo J, Secher NH. Maximal oxygen deficit of Sejersted OM, Sjogaard G. Dynamics and consequences of potassium
sprint and middle distance runners. Eur J Appl Physiol Occup Physiol shifts in skeletal muscle and heart during exercise. Physiol Rev 2000;
1994; 69: 140 – 146 80: 1411 – 1481
36 43
Ozyener F, Rossiter HB, Ward SA, Whipp BJ. Influence of exercise in- Sieck GC, Prakash YS. Fatigue at the neuromuscular junction. Branch
tensity on the on- and off-transient kinetics of pulmonary oxygen up- point vs. presynaptic vs. postsynaptic mechanisms. Adv Exp Med Biol
take in humans. J Physiol 2001; 533: 891 – 902 1995; 384: 83 – 100
37 44
Padilla S, Bourdin M, Barthelemy JC, Lacour JR. Physiological corre- Smith DO. Morphological aspects of the safety factor for action poten-
lates of middle-distance running performance. A comparative study tial propagation at axon branch points in the crayfish. J Physiol 1980;
between men and women. Eur J Appl Physiol Occup Physiol 1992; 301: 261 – 269
45
65: 561 – 566 Spencer MR, Gastin PB. Energy system contribution during 200- to
38
Ross A, Leveritt M. Long-term metabolic and skeletal muscle adapta- 1500-m running in highly trained athletes. Med Sci Sports Exerc
tions to short-sprint training: implications for sprint training and ta- 2001; 33: 157 – 162
46
pering. Sports Med 2001; 31: 1063 – 1082 Thomas C, Hanon C, Perrey S, Le Chevalier JM, Couturier A, Vande-
39
Saltin B, Bangsbo J, Graham T, Johansen L. Metabolism and perfor- walle H. Oxygen uptake response to an 800-m running race. Int J
mance in exhaustive intense exercise; different effects of muscle gly- Sports Med 2005; 26: 268 – 273
47
cogen availability, previous exercise and muscle acidity. In: Marcon- Wasserman K, Whipp BJ, Koyl SN, Beaver WL. Anaerobic threshold
net P, Komi P, Saltin B (eds). Muscle Fatigue Mechanisms in Exercise and respiratory gas exchange during exercise. J Appl Physiol 1973;
Training & Testing
524
Bosquet L et al. V̇O2 Kinetics and Performance … Int J Sports Med 2007; 28: 518 – 524