We recruited eight male participants (mean ± SD: age 24 ± 5 yr, body mass 76 ± 8 kg, height 1.81 ± 0.54 m, V˙O2peak 45.6 ± 6.5 mL·kg⁻¹·min⁻¹) for this study. All participants were screened for known heart or respiratory disease by completing a health and physical activity readiness questionnaire. The study received institutional ethics approval in accordance with the Declaration of Helsinki. All participants provided written informed consent before participating.

Each participant completed three visits. During the first visit, participants first completed an incremental exercise test to determine LT and V˙O2peak. A smooth ramped-incremental test protocol (Lode Excalibur Sport, Groningen, Netherlands) was used, which started at 100 W and increasing by an equivalent of 20 W per minute until the participant could no longer maintain a cadence of at least 60 rev. min⁻¹ at the specified power output, despite strong verbal encouragement. Pulmonary gas exchange and ventilation was measured breath-by-breath. Participants wore a flexible facemask that fully covered mouth and nose and breathed through a mouthpiece and impeller turbine assembly. Before each test, the gas analyzer (MetaMax 3B, Cortex Biophysik, Leipzig, Germany) was calibrated for gas and volume measurements, according to the manufacturer's guidelines. The gas analyser was calibrated with gases of known concentrations over the span expected during exercise, and the turbine volume transducer was calibrated using a 3-L syringe over a range of different flow rates (Hans Rudolph, MO). V˙O2peak was recorded as the highest exercise test mean oxygen consumption over a 30-s period. The LT was estimated as the V˙O2 equivalent to: (1) a non-linear increase in VCO₂ output, compared to V˙O2; (2) an increase in the ventilatory equivalent for oxygen (V_E/ V˙O2), without an increase in V_E/VCO₂; (3) an increase in end-tidal partial pressure of O₂ (P_ETO₂) without a fall in the end-tidal partial pressure of CO₂ (P_ETCO₂) (Beaver et al., 1986).

After a short rest, participants were familiarized with the maximal voluntary cycling power test (MVCP) used during the subsequent two visits. Participants were required to sprint maximally for 6 s on an SRM ergometer (SRM, Julich, Germany) set in “isokinetic mode” with a fixed pedaling cadence of 90 rev·min⁻¹. This was repeated a further two times for the purposes of familiarization.

Participants completed two further experimental visits (Figure 1). Each visit commenced with a baseline finger prick blood lactate sample (Biosen C-Line, EKF Diagnostics, London, UK) and 10-min warm up at 100 W at 90 rev·min⁻¹ following which participants performed a MVCP test. MVCP at an isokinetic pedaling cadence of 90 rev·min⁻¹ was taken as the highest power output recorded during the 6-s sprint. Power data was recorded at the crank of the ergometer once every revolution throughout the MVCP test. Prior to each test, the SRM ergometer was calibrated according to the manufacturer recommendations. Next, participants were asked to either perform the 100 intermittent drop-jumps protocol (pre-fatigue condition) or rest for 33 min (control condition) according to a random and counterbalanced order. In the pre-fatigue condition, participants dropped 100 times from a 40-cm high platform down to 90° knee angle before jumping upward as high as possible. Between each jump there was a 20-s rest period. Marcora et al. (2008) observed no increase in capillary blood lactate concentration after this fatiguing exercise protocol, which requires only moderate cardiovascular strain (58% of maximum heart rate) for 33 min. The control condition consisted of resting comfortably for 33 min. Two minutes after completing the allocated experimental manipulation, another finger prick blood lactate sample was taken before a second MVCP test was performed. Two minutes following the MVCP, participants then completed a bout of 6 min high-intensity cycling exercise to assess the V˙O2 kinetic response. This exercise protocol consisted of 2 min rest, sitting on the ergometer (Lode Excalibur Sport, Groningen, Netherlands), 3 min of “unloaded” cycling at 90 rev·min⁻¹, and finally a transition to a power output calculated to require 50% delta (Δ50%). The power output at Δ50% was calculated as the difference between estimated LT and V˙O2peak using the formula: Δ50% = LT + 0.5 × (V˙O2peak – LT). Following the bout of high-intensity cycling exercise, another finger prick blood lactate sample was taken before a third MVCP test was performed. Participants then rested for 60 min before repeating the exercise protocol described above to allow replication of the V˙O2 kinetic response. Due to the long lasting nature of fatigue induced by the intermittent jumping protocol (Twist and Eston, 2005) it was not deemed necessary for participants to complete a further bout of fatiguing exercise prior to the 2nd period cycling at Δ50%. Before and after this second bout of high-intensity cycling exercise, a finger prick blood lactate sample was taken before participants performed a MVCP test.

Heart rate was measured throughout all exercise tests using a monitor (Polar Electro Oy, Kempele, Finland) with a 1 Hz sampling frequency, and then averaged over each minute of the 6 min test. Rating of perceived exertion (RPE) was taken each minute using the Borg RPE scale and memory anchoring procedures (Borg, 1998). In both studies, all visits were completed within 21 days of the first visit, and at least 7 days apart and at the same time of day. Participants were instructed to arrive at the laboratory well hydrated and having abstained from intense exercise, alcohol and caffeinated beverages within the 24 h preceding each visit.

The breath-by-breath V˙O2 from each test were initially examined to exclude errant breaths caused by coughing, swallowing, sighing, etc., and those values lying more than four standard deviations from the local mean were removed. The breath-by-breath data were subsequently linearly interpolated to provide second-by-second values and, for each individual, the two repetitions of the Δ50% exercise were time-aligned to the start of exercise and ensemble averaged. The first 20 s of data after the onset of exercise were deleted to ensure the exclusion of the cardiopulmonary phase (Whipp and Rossiter, 2005) and a non-linear least-square algorithm was used to fit the data thereafter. In this equation V˙O2 baseline indicates the baseline value obtained during the 3 min of “unloaded” cycling at 90 rev·min⁻¹, calculated as the mean V˙O2 measured over the final 60 s unloaded cycling. A bi-exponential model was used to characterize the response kinetics in its constituent fundamental and slow components. Responses from the two transitions in each condition were ensemble averaged to improve signal-to-noise ratio and fitted according to the following function: V˙O2(t)=V˙O2b+Ap(1-e-(t-TDp/τp))+Asc(1-e-(t-TDsc/τsc)) where V˙O2 (t) represents the absolute V˙O2 at a given time t; V˙O2b represents the mean V˙O2 in the baseline period; A_p, TD_p, and τ_p represent the amplitude, time delay, and time constant, respectively, describing the fundamental or phase II increase in V˙O2 above baseline; and A_sc, TD_sc, and τ_sc represent the amplitude of, time delay before the onset of, and time constant describing the development of, the V˙O2 slow component, respectively. Data fitting were carried out by using a commercial available software (GraphPad Prism 4, GraphPad Software).

The statistical software package SPSS was used for all statistical analysis (Version 14.0, SPSS, Chicago, Illinois). Prior to analysis all data was checked for normality of distribution using a Shapiro-Wilk test. The effects of condition (pre-fatigue vs. control) on parameters associated with the V˙O2 kinetic response was assessed using paired t-tests. Two-way repeated measures ANOVAs (condition × time) were used to analyse the MVCP, blood lactate concentration. Heart rate and RPE data were analyzed by two-way repeated measures MANOVA (condition × bout × time). For all repeated measures analysis, if a significant interaction was revealed, the main effect of condition was not considered, and tests of simple main effects of condition were conducted as follow-up using the Least significant difference post-hoc test. Statistical significance was accepted at p < 0.05. Results are reported as mean ± SD unless otherwise stated.

There was a significant condition × time interaction on MVCP (P = 0.03; Figure 2A). Tests of simple main effects of condition at each time point revealed no significant difference in MVCP at baseline. However, after experimental manipulation, MVCP was significantly lower in the pre-fatigue condition compared to the control condition at all time points (P = 0.02). Tests of simple main effects of time within each condition revealed that 6 min of cycling exercise at Δ50% did not induce a significant reduction in MVCP in either the pre-fatigue (P = 0.73) or the control (P = 0.54) condition.

As shown in Table 1, V˙O2b was not significantly different between conditions (P = 0.78). Figures 3A–C demonstrates a typical individual response in the control and pre fatigued conditions during high-intensity cycling exercise. Locomotor muscle fatigue had no effect on either the fundamental phase responses or the slow component during high-intensity cycling exercise. Specifically, neither the kinetic response (τ_p, P = 0.72) nor the amplitude (A_p, P = 0.75) of the fundamental phase were significantly different between conditions. The V˙O2sc and the trajectory (rate of the development of the V˙O2sc) were also unaffected by locomotor muscle fatigue (P = 0.50 and P = 0.55, respectively). Figures 3D–F shows that the pattern of the V˙O2 kinetic response during high-intensity cycling exercise was similar between bouts 1 and 2 in the pre-fatigue condition.

There was no significant condition x time interaction (P = 0.48) and no significant main effect of condition (P = 0.75) on blood lactate concentration. There was, however, a significant main effect of time (P ≤ 0.01) with blood lactate concentration significantly increasing after each bout of high-intensity cycling exercise (P ≤ 0.01; Figure 2B).

There was no significant no significant condition x bout x time interaction (P = 0.89), condition x time interaction (p = 0.09), or bout x time interaction (P = 0.44). There was also no significant main effect of condition (P = 0.54), or exercise bout (P = 0.29) on heart rate. There was, however, a significant main effect of time (P ≤ 0.01) with heart rate progressively increasing during high-intensity cycling exercise. Heart rate responses are therefore presented as the grand average of both bouts in Figure 4A.

There was no significant no significant condition × bout × time interaction (P = 0.17) on RPE. The significant main effect of bout (P = 0.03) as well as significant condition x bout (P = 0.03) and bout × time (P = 0.02) interactions indicate that bout 2 was perceived as requiring more effort than bout 1. For sake of simplicity, RPE data are presented as the grand average of both bouts in Figure 4B. RPE increased over time in both conditions (condition × time interaction, P = 0.13; main effect of time, P ≤ 0.01) but was significantly higher in the pre-fatigue condition compared to the control condition (main effect of condition, P ≤ 0.01).

As in previous studies (Skurvydas et al., 2002; Nielsen et al., 2005; Marcora et al., 2008) the 100 intermittent drop jumps protocol induced a significant reduction in locomotor muscle function with no significant increase in blood lactate concentration (Figure 2). The significant and prolonged reduction in MVCP (Figure 2A) would have necessitated additional muscle recruitment to meet the demands of the subsequent high-intensity cycling exercise bouts. The higher RPE measured in the pre-fatigue condition (Figure 4B) supports indirectly the assumption that a greater level of central motor command and muscle recruitment was required to produce the same power output as in the control condition (de Morree et al., 2012). Furthermore, blood lactate accumulation during the two bouts of high-intensity cycling exercise (a marker of metabolic stress in locomotor muscles) was not different between conditions (Figure 2B). Therefore, with reference to the work of Skurvydas et al. (2000), our data suggests that we were able to experimentally isolate the effects of additional recruitment to compensate for locomotor muscle fatigue from the effects of mechanisms like muscle metabolite accumulation, limited O₂ or substrate availability, increased glycolysis and pH disturbance on V˙O2 kinetics during high-intensity cycling exercise.

There is a considerable body of literature suggesting an association between the V˙O2sc and the progressive recruitment of type II muscle fibers during aerobic exercise above LT (Saunders et al., 2000; Burnley et al., 2002; Krustrup et al., 2004a,b; Bernasconi et al., 2006; Endo et al., 2007). However, the link between the V˙O2sc and the progressive recruitment of Type II muscle fibers is not supported unanimously (Lucia et al., 2000; Scheuermann et al., 2001; Tordi et al., 2003; Zoladz et al., 2008). One of the difficulties is the poor validity of surface EMG to infer the level of muscle activation and the type of motor unit recruited, especially during dynamic contractions (Farina et al., 2004). Therefore, a more direct testing of the relationship between locomotor muscle fatigue and the V˙O2sc during aerobic exercise was required. The first to do so were Cannon et al. (2011) who found a significant correlation between the magnitude of locomotor muscle fatigue (measured by a reduction in MVCP) and the magnitude of the V˙O2sc measured during cycling at different exercise intensities. This finding lends some support to the “muscle fatigue and recruitment” hypothesis for the V˙O2sc observed during high-intensity aerobic exercise. However, in the same study, Cannon et al. (2011) found that locomotor muscle fatigue develops during the first 3 min of cycling exercise at Δ20 and Δ60%, with no further fatigue developing over the remaining 5 min of high-intensity cycling exercise even though a V˙O2sc is still evident. These data demonstrated for the first time dissociation between locomotor muscle fatigue and the development of the V˙O2sc during high-intensity cycling exercise. Therefore, the significant correlation between locomotor muscle fatigue and the V˙O2sc reported by Cannon et al. (2011) is most likely a spurious one, with both variables related to exercise intensity.

The V˙O2 kinetics data from the current experimental study provide experimental evidence that locomotor muscle fatigue (defined as an exercise-induced reduction in maximal voluntary force or power produced with the locomotor muscles) is not causally associated with the development of the V˙O2sc during high-intensity aerobic exercise. Indeed, none of parameters of the V˙O2 kinetics were affected by locomotor muscle fatigue (see Table 1, Figure 3). Specifically, the magnitude of V˙O2_sc during high-intensity cycling exercise was normal and not significantly different between the pre-fatigue and the control condition. We also found that, despite causing a significant V˙O2sc, cycling for 6 min at Δ50% did not induce any significant reduction in MVCP in either the pre-fatigue or control condition. Clearly, as noted by Cannon et al. (2011), locomotor muscle fatigue might not necessarily be required for the development of the V˙O2sc during high-intensity aerobic exercise.

As the V˙O2sc of the metabolic process have been suggested to arise from within the locomotor muscles (Jones et al., 2011), it is interesting to consider that pre-fatiguing the same muscles had no effect on altering the V˙O2 kinetic response to exercise above the LT; i.e., a reduced efficiency (or an increased O₂ cost) of muscle contractions still occurred even though there were different levels of muscle fatigue between conditions at the start of exercise. The degree of blood lactate accumulation observed in our study clearly demonstrates that cycling exercise for 6 min at Δ50% was metabolically stressful for our participants. Thus, instead of greater recruitment of less efficient Type II fibers because of locomotor muscle fatigue, the explanation for the V˙O2sc observed during exercise above the LT may be challenges to cellular homeostasis sufficient to impair the efficiency of the working muscle fibers. Indeed, using a 3 min all-out high intensity cycling test, Vanhatalo et al. (2011) demonstrated that progressive fiber recruitment was not requisite for the development of the V˙O2sc. Instead their results suggest that the V˙O2sc was likely generated by a higher muscle fiber oxygen cost per unit of external work done. Moreover, Ribeiro et al. (1986) found participants decreased mechanical power output in order to maintain a constant pulmonary V˙O2 during 40 min bouts at fixed percentages of V˙O2max (55–75%). Interestingly, the magnitude of the reduction in power output was linearly related to the relative metabolic power. However, in the current study, our data suggests that a reduction in efficiency and a V˙O2sc is still evident within pre fatigued muscles. Therefore, we argue that progressive muscle fatigue per se, may not be associated with muscle inefficiency and the development of the V˙O2sc.

One possible mechanism for the V˙O2sc could be a reduction in efficiency of energy transfer within the mitochondria, specifically in the ratio between ADP phosphorylation and oxygen consumption (P/O ratio). Cannon et al. (2014) have shown that during high intensity exercise eliciting the V˙O2sc, the tight coupling between ATP production V˙O2 seen during moderate intensity exercise is no longer evident. Indeed, changes in pH and high rates of glycolytic flux resulting from high intensity exercise have been suggested to reduce the P/O ratio (Özyener et al., 2001), suggesting that muscle inefficiency may result from both impaired ATP production and turnover. Another potential mechanism leading to increased mitochondrial uncoupling, reduced P/O ratio, and reduced efficiency is back leak of protons across the inner membrane without driving ATP-synthase (Rolfe and Brand, 1996). Alternatively, the V˙O2sc could also be caused by some mitochondrial ATP generation being used to reduce ROS generation within the cell (Brand, 2000). A high proton motive force that drives efficient ATP synthesis is associated with an additional ROS production. Proton leak across the mitochondrial membrane without driving ATP production may therefore assist in limiting the oxidative damage associated with high levels of ROS generated during the high intensity or prolonged aerobic exercise (Sahlin et al., 2010). High intensity aerobic exercise could also result in a reduction Gibbs free energy and necessitate an increased rate of ATP production to maintain steady-state SERCA ATPase function, and sarcoplasmic reticulum calcium flux. In turn, this could increase rate of ATP production and would thus reduce the efficiency of the muscular work during exercise (Grassi et al., 2015). Thus, there are potential mechanisms other than additional muscle recruitment to compensate for locomotor muscle fatigue which could explain the presence of the V˙O2sc. However, the evidence for the possible mechanisms outlined above is sparse, and readers are referred to the review by Grassi et al. (2015) for a more in-depth discussion on this topic.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.