Oxygen uptake (VO₂) kinetics has been analyzed through mathematical modeling of the constant-load exercise onset and offset VO₂ response. This response profile appears to be of an exponential nature, which could indicate first or second order kinetics operations (DiMenna and Jones, 2009). This analysis has shown that VO₂ exponentially increases at the onset of moderate exercise with constant power output (on-fast component), reaches a steady state, and rapidly decreases at the offset of moderate exercise (off- fast component) (Kilding et al., 2006; Ozyener et al., 2001; Paterson and Whipp, 1991; Scheuermann et al., 2001). First-order kinetics mandates on/off symmetry, which means that the change in VO₂ occurring when the contractile activity is ceased must be a mirror image of that which occurred when it was commenced (Rossiter et al., 2005). In the heavy intensity exercise, i.e., at intensities greater than the anaerobic threshold but below the maximal VO₂, an delayed increase (on-slow component) after the on-fast component is presented (Barstow and Molé, 1991; Barstow et al., 1996; Ozyener et al., 2001; Paterson and Whipp, 1991; Scheuermann et al., 2001), but at the offset only an off-fast component is developed (Ozyener et al., 2001; Scheuermann et al., 2001). At the severe exercise intensity, which is significantly above the anaerobic threshold, and neither VO₂ nor blood lactate levels can be stabilized (Poole et al., 1988), the on-transient VO₂ kinetics is reverted to a single-exponential profile (Ozyener et al. , 2001), while the off-transient kinetics is retained for a two-component form (Dupond et al., 2010; Ozyener et al., 2001). At the highest intensity - extreme exercise leading to exhaustion before maximal oxygen uptake is attained (DiMenna and Jones, 2009; Hill et al., 2002) - , the VO₂ on-kinetics response is characterized by the development of an evident fast component, being the slow component phenomenon not developed (Burnley and Jones, 2007; Figueiredo et al., 2011; Whipp, 1994). This area of intensity was recently described (Hill et al., 2002), and, to the best of our knowledge, the VO₂ off- kinetic profile has never been studied at this particular intensity.

VO₂ assessment has been carried out mainly in well controlled environments, particularly in exercise laboratories, and the number of studies conducted in field is very scarce (Billat et al., 2002; Fernandes et al., 2008). In fact, the VO₂ off-transient kinetics is documented in constant-load exercise performed from the moderate to severe intensities. Nevertheless, studies that aim to model the VO₂ recovery kinetics at extreme intensity exercise were not yet conducted in swimming. In this sense, the purpose of this study is to characterize the VO₂ off-transient kinetics, examining also the on/off symmetry, during a 200-m front crawl maximal effort performed at extreme intensity. It was hypothesized that an on/off symmetry of the VO₂ kinetics response would be preserved, although the post-exercise VO₂ did not match the O₂ deficit.

Participants
Eight highly trained male swimmers volunteered to participate in the study. The participants provided informed written consent before data collection, which was approved by the local ethics committee and was performed according to the declaration of Helsinki. Their mean performance for long course 200-m freestyle was 109.3 ± 2.0 s, corresponding to 90.3 ± 3.2% of the 2009 world record for this event. This sample included a finalist and five participants at the European Championships. Individual and mean (± SD) values for subjects' main physical and performance characteristics were: age (21.8 ± 2.4 years), height (184.5 ± 6.2 cm), body mass (76.1 ± 6.5 kg), fat mass (10.4 ± 1.7%) and lean body mass (62.4 ± 4.4%).

Data collection
In an indoor 25-m swimming pool, with a water temperature of 27ºC, each swimmer performed a 200-m front crawl effort at maximal speed. In water starts and open turns, without underwater gliding, were used. Each swimmer performed a 200-m front crawl maximal effort, according to his best individual 200-m performance and his own experiences, and was encouraged to swim at his best effort; therefore, no visual or acoustic pacing was implemented. VO₂ kinetics was measured using a telemetric portable gas analyzer (K4b², Cosmed, Italy), which was connected to the swimmer by a low hydrodynamic resistance respiratory snorkel and valve system (Keskinen et al., 2003; Rodriguez et al., 2008), and was calibrated before and after each test. Respiratory variables were continuously monitored after the 200-m effort until baseline VO₂ values were obtained (after approximately 12 min of recovery the assessment was ended). Swimmers were advised to use continuous rhythmical breathing during swimming, turning and in the recovery period. Expired gas concentrations were measured breath-by-breath and averaged every 5 s for a better temporal resolution (Sousa et al., 2010) in order to reduce inter breath fluctuations ("noise"). Peak oxygen uptake (VO_2peak) was considered as the highest value of this sampling interval.

Data analysis
The following equation was used to fit VO₂ kinetics on the on-transient period:

(1)

where t is the time, V_b is the oxygen uptake at the start of the exercise (mL·kg^-1·min^-1), A_on is the amplitude of the fast component (mL·kg^-1·min^-1), TD_on is the time for the onset of the fast component (s) and t_on stands for the time constant of the fast component, i.e., the time to reach 63% of the plateau of this phase during which physiological adaptations adjust to meet the increased metabolic demand. The cardiodynamic phase was not taken into consideration due to its amplitude insignificant value. The inexistence of a slow component was confirmed by the rigid intervals method, particularly by the difference between the last VO₂ measurement of the exercise and the value measured in the final 5 s of the 200-m event (adapted from Fernandes et al., 2003; Koppo and Bouckaert, 2002).

For the off-transient period, the individual responses were fitted by using both a single (equation 2) and a double exponential (equation 3) regression models for the entire recovery period, in which the exponential term started at the beginning of the off-transient period modeling (TD_1off in the equations):

(2)

(3)

where t is the time, A_off represents the amplitude for the exponential term and the t_off and TD_off are the associated time constant and time delay. A nonlinear least squares method was implemented in MatLab for the adjustment of these functions to VO₂ data.

After a visual exploratory inspection of all VO₂ curves, and for the sake of numerical stability, it was verified that, due to the extreme exercise intensity in which the 200-m held, all swimmers started the recovery period immediately after the 200-m effort. In this sense and assuming that TD_1off=0, the off-transient period was modeled according to the restructure equations:

(4)

(5)

The F-test (0. 28) showed the homogeneity of both models variances, confirmed also by the equality of their mean values (p=0.98), and therefore, the off-transient response was well described by a single exponential function. In fact, this characterization was not improved by using the double exponential model. In this sense, the on- and off-transient periods are symmetrical in shape (mirror image) once they were adequately fitted by single-exponential functions. An example of the oxygen (O₂) uptake on and off kinetics curve is shown in Figure 1.

The mean (± SD) values for swimming speed (200_speed), VO_2peak, A_on, TD_on, t_on and A_off and t_off for the 200-m front crawl effort and recovery period are presented in Table 1.

Significant differences were obtained between the on- and off- VO₂ kinetic parameters (all for p<0.01), and its amplitude was higher in the recovery period. Complementarily to the above referred data, direct relationships were observed between t_off and 200_speed (r = 0.77, p = 0.02), t_off and VO_2peak (r = 0.76, p = 0.03) and t_off and A_on (r = 0.72, p = 0.04) (see Figure 2). No significant correlations were found between VO_2peak and the other VO₂ off-transient parameters (A_off, r = 0.35, for p > 0.05). The absences of significant relationships were also observed between t_on and t_off (r = 0. 19) and A_on and A_off (r = 0.5), all with p > 0.05.

The aim of this study was to characterize the VO₂ off-transient kinetics and to examine the on/off symmetry during a self-imposed 200-m swimming at race pace. We tested the hypothesis that the VO₂ kinetics response will manifest a symmetric on/off response, even if the post-exercise VO₂ does not match the O₂ deficit.

An understanding of the VO₂ kinetics is considered an important parameter to improve sports training methodology and increase performance in sport (Billat et al., 2001). Furthermore, it was recently suggested that the determinants of exercise tolerance and the limitations to sports performance can be better understood through an appreciation of the physiological significance of the fast and slow components of the dynamic VO₂ response to exercise (Burnley and Jones, 2007). For a long time, studies regarding O₂ uptake assessment in swimming were conducted with either Douglas bags (di Prampero et al., 1974; Lavoie and Montpetit, 1986) or mixing chamber gas analyzers (Dal Monte et al., 1994; Demarie et al., 2001). It was only recently that the development of a swimming snorkel suitable for breath-by-breath analysis (Keskinen et al., 2003; Rodríguez et al., 2008) allowed assessing VO₂ dynamics in swimming pool conditions through direct oxymetry (Fernandes et al., 2003; Rodriguez et al., 2003). Nevertheless, in the O₂ uptake kinetics related literature, studies that aimed to characterize it in human non-constant load extreme intensity exercises are very scarce. Moreover, among these studies, only Rodriguez and Mader, 2003, Rodriguez et al., 2003, and Silva et al., 2006 implemented a swimming effort at intensities similar to our protocol.

Considering the total sample, VO_2peak ranged from 60.2 to 81.8 ml·kg^-1·min^-1, which is in accordance with recently reported data obtained in trained male competitive swimmers performing during swimming in pool conditions (Fernandes et al., 2008; Figueiredo et al., 2011; Reis et al., 2010; Rodríguez and Mader, 2003; Rodríguez et al., 2003; Silva et al., 2006).

Symmetry between the on- and off-transient phases: Since symmetry is an essential quality of VO₂ kinetic dynamics viewed as a first-order reaction kinetics (Rossiter et al., 2005), it was a focus of interest in the present study. The on/off symmetry of the fast components has been observed for the moderate intensity exercise domain performed in cycle ergometer (Paterson and Whipp, 1991; Ozyener et al., 2001; Scheuermann et al., 2001) and treadmill running (Kilding et al., 2006). For the heavy intensity exercise, an asymmetry in the VO₂ dynamics has been reported, describing an on-fast component and an off-fast and off-slow components at cycle ergometer (Ozyener et al., 2001) and knee extensor exercise (Rossiter et al., 2002). This asymmetry was also reported for severe exercise intensity, namely in indoor running (Dupond et al., 2010) and cycle ergometer (Ozyener et al., 2001). In contrast, in the present study the on- and off-transient phases were symmetrical, once they were adequately fitted by a single exponential function, compared to the double exponential one, and no slow component for the VO₂ response was developed (see Figure 1). Nonetheless the above referred studies, the symmetry observed in the present study can be explained by the implementation of a non-constant load, and to the greater exercise intensity. As expected, we observed only an on-fast component, since the non-constant load at freely-chosen maximal race pace induced an exponential rise in VO₂ kinetics that unable the development of a VO₂ slow component; this fact was previously mentioned but only for ergometer exercise (Burnley and Jones, 2007; Whipp, 1994).

On/off kinetic parameters: Although an on/off symmetry in the VO₂ kinetic response was observed in this extreme intensity exercise lasting 2.7 min on average, differences between the VO₂ on- and off-transient kinetic parameters were observed. In fact, greater A_off and t_off values are reported. This last parameter is a major focus of interest in the VO₂ kinetic related literature, once it is a determinant factor in VO₂ dynamics. A longer t_off value, as observed in this study, concur with previous data obtained in the heavy exercise domain (Cleuziou et al., 2004; Yano et al., 2007); however, other studies reported the opposite behavior for the same exercise intensity (Engelen et al., 1996; Ozyener et al., 2001; Scheuermann et al., 2001), as well as for the moderate domain (Patterson and Whipp, 1991). At the severe exercise intensity, Billat et al., 2002 and Ozyener et al., 2001 reported no differences in regarding on and off fast periods. In addition, the obtained t_off mean value was greater than the results reported in the literature for both moderate (Cleuziou et al., 2004; Kilding et al., 2006; Rossiter et al., 2002; Takayoshi et al., 2003), heavy (Rossiter et al., 2002) and severe intensities (Perrey et al., 2002).

However, as suggest, when we compared our data with studies using a double exponential fitting approach, t_off was shorter comparing to t_off of the slow component during heavy (Cleuziou et al., 2004) and severe intensity exercise (Dupond et al., 2010). As previously stated, the present study reported a symmetry on the on/off VO₂ kinetic response; however differences between the on- and off- VO₂ kinetic related parameters were found.

In fact, VO₂ kinetics is influenced by endurance training, being reported a faster VO₂ on-kinetics in trained subjects involved both in cross-sectional and longitudinal studies (Casaburi et al., 1987; Koppo et al., 2004; Murias et al., 2010; Phillips et al., 1995). Indeed, training seems to change the muscle fiber-type characteristics, mitochondrial density, oxidative enzyme activity, oxygen availability, capillary density and muscle perfusion (Koppo et al., 2004), existing evident differences between trained and untrained subjects. Although this study did not have the intention to investigate this phenomenon, the mean swimming speed was very high since the onset of the effort, which may induced a faster increase in ATP requirements, and a fast lactate accumulation, once a pattern of type I/II muscle fiber contribution seems to be established without delay (Cunningham et al., 2000). These facts (and being the off-set fast component explained by the restore of O₂ in blood and in muscle, a significant lactate removal, and by the resynthesis of ATP and PCr) may induce discernible slower responses during the recovery period. Hence, the oxygen debt must be larger than the oxygen deficit, i.e., the post-exercise VO₂ quantitatively did not match the O₂ deficit (Yano et al., 2007). In fact, since different pacing strategies were adopted during the maximal 200-m, different VO₂ on kinetics may occurred, which influenced the VO₂ response in the recovery period. This is a limitation of the current study comparing to constant pace researches.

Regarding the VO₂ amplitude, the greater observed A_off mean value (comparing to A_on) is not in accordance with the results reported for moderate and heavy intensities (Cleuziou et al., 2004), and for the severe intensity exercise (Perrey et al., 2002), that showed no significant differences between the A_on and A_off mean values. In our study, the greater values of A_off may be a result of the extreme exercise intensity in which our study was conducted, different modeling procedures that were used, as also mode of exercise performed. At this exercise intensity, in which highest work rates are observed, the VO₂ mean value is high even until the end of the effort. Once the A_off represents the difference between the VO₂ at the end of the exercise and the steady state VO₂, the greater A_off mean value seems justified.

Once the TD_off was assumed to be zero, in result of the sudden and instantaneous diminishing of VO₂, comparisons with previously reported data obtained for the moderate (Cleuziou et al., 2004) and heavy intensities domains (Billat et al., 2002) are difficult to establish. However, Takayoshi et al. (2003) reported low TDoff mean values (1, 2 s) for the moderate exercise intensity domain. Moreover, and contrasting the results of the present study, Perrey et al., 2002 found no differences between the TD_on and TD_off mean values at severe intensity.

Relationship between VO₂ kinetics on/off-transient phases and performance: The observed direct relationship between t_off and 200_speed evidences that the swimmers who performed a faster 200- m, needed more time to attained a VO₂ steady state in the off-transient phase; in addition, these swimmers presented greater VO_2peak and A_on mean values. These facts seem to evidence one more time that the very high swimming speed just after the beginning of the effort led to greater VO_2peak and A_on mean values, increasing both the need for a higher energy supply and the accumulation of fatigue-related metabolites, slowing the recovery phase. Indeed, the 200-m performance is strongly related to the t_off, which seems to be also a good predictor of VO_2peak and A_on. However, these data should be seen with precaution, once other factors might explain the performance variability in this specific distance.

This study was supported by grant: PTDC/DES/101224/2008 (FCOMP-01-0124-FEDER-009577).

• The VO₂ slow component was not observed in the recovery period of swimming extreme efforts;
• The on and off transient periods were better fitted by a single exponential function, and so, these effort and recovery periods of swimming extreme efforts are symmetrical;
• The rate of VO₂ decline during the recovery period may be due to not only the magnitude of oxygen debt but also the VO₂peak obtained during the effort period.