In order to determine (i) the modifications of AC_FS during maturation stages, and (ii) whether AC_ALT and AC_FS both determined in a single maximal swimming effort were similar to AOD, the present study was divided into two parts. Figure 1 presents the experimental design of the present study. In Part I, swimmers were subjected one maximum front crawl (without snorkel) 400 m effort to determine AC_FS; and, on the other day, body composition was analyzed by the Dual-energy X-ray absorptiometry (DEXA, General Electric Medical Systems, Fairfield, USA) explained elsewhere (Campos et al., 2012). All tests were performed in a 25-m swimming pool with water temperature of 25 ± 2°C and were preceded by a warm-up of ~1,000 m freestyle swimming of low to moderate intensity determined subjectively by the swimmers. Additionally, swimmers were instructed not to engage in strenuous activity the day before exercise tests and to maintain a consistent routine of training, sleeping, and diet throughout the study.

In Part II, swimmers were subjected to three experimental sessions, interspersed by at least 24 h. On the first visit, subjects performed four submaximal efforts aiming to establish VO₂-speed relationship. On the second day, the subjects were subjected to a submaximal exercise, and a maximal front crawl 400 m effort with snorkel. No warm-up was performed before the tests and the subjects started each trial when their VO₂ values exhibited two consecutive values within 2 mL·kg⁻¹·min⁻¹ of that recorded before the first submaximal exercise (observed on the first day; Reis et al., 2010b). This first maximal front crawl 400 m effort (second day trial) was used to evaluate AOD and AC_ALT (Figure 1) and the swimmers used snorkel during the effort. After at least 48 h, the swimmers were subjected to another 400 m maximal effort without snorkel (AC_FS).

Expired gases were collected breath-by-breath using either a gas analyzer (Quark PFT, Cosmed^®, Rome, Italy) in Part I, and a portable gas analyzer (K4b², Cosmed^®, Rome, Italy) connected to an Aquatrainer snorkel (Cosmed^®, Rome, Italy) in Part II. The gas analyzers were calibrated immediately before and verified after each test using a certified gravimetrically determined gas mixture, while the ventilometer was calibrated preexercise and verified postexercise using a 3-L syringe, in accordance with the manufacturer's instructions. Following the removal of outliers, breath-by-breath data were interpolated to give 1s values (OriginPro 8.0, OriginLab Corporation, Microcal, Massachusetts, USA) to enhance response characteristics of excess postoxygen consumption (EPOC) (Zagatto et al., 2011). Before the maximal 400 m and after 3, 5, and 7 min of recovery, blood samples were collected to determine [La⁻] using a blood lactate analyzer YSI-2300 (Yellow Springs Instruments^®, OH, USA).

Peak oxygen consumption (VO_2Peak) was estimated by the backward extrapolation technique, after a maximum front crawl effort of 400 m freestyle, that is, without snorkel. For this, the subjects were instructed to immediately breathe on a face mask (Hans Rudolph, Kansas City, MO, USA) connected to a breath-by-breath gas analyzer system. The equipment was calibrated immediately before the test according to the instruction of the manufacturer. The VO_2Peak was obtained using a 30 s backward extrapolation technique (Campos et al., 2017b; Monteiro et al., 2020); for this, VO₂ values were transformed in logVO₂, and plotted against time. Through a linear regression the y-intercept was considered as VO_2Peak.

Free-swimming anaerobic contribution was determined by the sum of Ana_ALA and Ana_LAC (Bertuzzi et al., 2010; Zagatto et al., 2011; Kalva-Filho et al., 2015). Swimmers were instructed to immediately breathe on a face mask (Hans Rudolph, Kanss City, MO, USA) connected to a breath-by-breath gas analyzer system (Quark PFT, Cosmed^®, Rome, Italy) for 5 min (Campos et al., 2017a). The AC_FS was calculated in Excel (Microsoft Corporation, Redmond, Washington, USA) and Origin (OriginPro 8.0, OriginLab Corporation, Microcal, Massachusetts, USA). Ana_ALA was assumed as the fast component of EPOC. For this EPOC, breath-by-breath measurements obtained during 5 min of recovery were adjusted as a function of time using a bi-exponential model (Equation 1) (Ozyener et al., 2001). The product between amplitude (A₁) and the fast component time constant (f₁) was assumed as Ana_ALA (Equation 2) (Knuttgen, 1970; Bertuzzi et al., 2010). Ana_LAC was obtained by net lactate accumulation (i.e., difference between [La-] peak and baseline values; Δ[La-]), considering a metabolic equivalent of 3 mL·O2·kg-1 for each unit of lactate elevated with maximal effort (di Prampero and Ferretti, 1999). Thus, AC_FS was assumed as the sum of Ana_ALA and Ana_LAC (Equation 3). AC_FS values were presented as absolute (L), and relative to body mass (mL·kg⁻¹).

where in Equation 1, VO_2(t) is the oxygen uptake at time t in recovery time, VO_2BASE was the oxygen uptake of at baseline measured before swimming, A is the amplitude, δ is the time delay, ƭ₁ is the time constant (tau) and ₁ and ₂ denote the fast and slow components, respectively. In Equation 2, Ana_ALA is the alactic anaerobic contribution and in Equation 3 AC_FS is the alternative method to determine anaerobic contribution in a single effort without snorkel and Ana_LAC is the lactic contribution. Data of one subject are presented in Figure 2.

The AC_ALT was determined as presented for AC_FS. The main differences between AC_ALT and AC_FS are due to the fact that at AC_FS the swimmers perform the effort without the snorkel and the fast component of values of EPOC, used to estimate the alactic anaerobic contribution, was obtained immediately after swimming (≈2 seg), while the swimmers swam with snorkel for AC_ALT.

Data normality was tested and confirmed by Shapiro–Wilk's test, which permitted the use of parametric tests. Data are presented as mean ± standard deviation (SD). Significance level was set at 5%. The minimal sample size to provide a statistical power of 80% was estimated using G^*Power software, version 3.1.9.4 (Franz Faul, Christian-Albrechts-Universität Kiel, Kiel, Germany).

The subject's characteristics are presented on Table 1.

Figure 4 presents the anaerobic contribution (i.e., AC_FS) of early-pubertal, middle- pubertal, and pubertal groups determined after the 400 m effort. Absolute Ana_ALA only tended to be different among groups [early-pubertal: 1.42 ± 0.84 L; middle-pubertal: 1.47 ± 0.69 L; pubertal: 2.11 ± 0.66 L; F = 2.86; p = 0.07; Power = 0.52; ηp2 = 0.15; 90% CI (0; 0.30)], without differences in relative Ana_ALA [early-pubertal: 30.27 ± 20.70 mL·kg⁻¹; middle-pubertal: 24.28 ± 10.13 mL·kg⁻¹; and pubertal: 31.63 ± 10.82 mL·kg⁻¹; F = 0.93; p = 0.40; Power = 0.19; ηp2 = 0.05; 90% CI (0; 0.17)]. Pubertal group presented greater absolute Ana_LAC than the other groups [early-pubertal: 0.64 ± 0.44 L; middle-pubertal: 1.01 ± 0.51 L; pubertal: 1.75 ± 0.83 L; F = 8.72; p = 0.001; Power = 0.95; ηp2 = 0.36; 90% CI (0.11; 0.49)], while no differences were found between early-pubertal and middle-pubertal.

Pubertal showed greater relative Ana_LAC than early-pubertal [early-pubertal: 12.77 ± 8.42 mL·kg⁻¹; middle-pubertal: 16.60 ± 7.24 mL·kg⁻¹; and pubertal: 25.44 ± 11.01 mL·kg⁻¹; F = 5.49; p < 0.01; Power = 0.81; ηp2 = 0.26; 90% CI (0.04; 0.41)]. AC_FS were greater in pubertal group than the other groups [early-pubertal: 2.10 ± 0.90 L; middle-pubertal: 2.48 ± 1.12 L; pubertal: 3.87 ± 1.12 L; F = 7.79; p = 0.002; Power = 0.93; ηp2 = 0.33; 90% CI (0.09; 0.47)], and no differences were found between early-pubertal and middle-pubertal (Figure 4). No differences were found for relative AC_FS between groups [early-pubertal: 44.82 ± 19.75 mL · kg⁻¹; middle-pubertal: 40.88 ± 15.55 mL · kg⁻¹; and pubertal: 57.08 ± 16.49 mL · kg⁻¹; F = 2.70; p = 0.08; Power = 0.49; ηp2 = 0.14; 90% CI (0; 0.29)].

Speed ranged between 64.42 ± 0.93 and 80.30 ± 6.85% of 400 m performance in submaximal efforts. The mean time for 400 m was 330.59 ± 13.20 s (mean speed = 1.20 ± 0.04 m·s⁻¹) and VO_2Peak was 3.07 L·min⁻¹. The VO₂-speed relationship presented values of angular, linear, and determination coefficients of 4.00 ± 1.22 (L·min⁻¹)·(m·s⁻¹)⁻¹, 1.82 ± 1.06 L·min⁻¹, and 0.94 ± 0.02, respectively. Figure 3 demonstrates the pacing used by swimmers during the maximal 400 m effort. Table 2 summarizes all parameters related to AOD, AC_ALT, and AC_FS.

No differences were found between absolute AOD (3.2 ± 1.3 LO₂) and AC_ALT (3.2 ± 1.5 LO₂), and AC_FS (4.0 ± 0.9 LO₂) determined in the 400 m maximal effort [F = 3.69; p = 0.06; Power = 0.54; ηp2 = 0.42; 90% CI (0; 0.60)]. The relative AOD (51.8 ± 12.2 mL · kg⁻¹), AC_ALT (50.5 ± 14.3 mL · kg⁻¹), and AC_FS (65.2 ± 8.8 mL · kg⁻¹) values presented main effect [F = 4.49; p = 0.04; Power = 0.62; ηp2 = 0.47; 90% CI (0.01; 0.64)]; however, post-hoc analysis did not indicate any differences among values (Figure 5).

The agreement analysis between methods are shown in Figure 6. The mean error between AOD and AC_ALT was 0.04 L, and between AOD and AC_FS was −0.74 L. However, the limits of agreement of AOD and AC_ALT were 0.96 and 0.87 L for upper and lower limits of agreement, while between AOD and AC_FS were 0.77 L for upper limit and 2.26 L for lower limit (four out of six presented greater AC_FS than AOD). AOD was very strongly correlated with AC_ALT (r = 0.95; p = 0.002), and strongly correlated with AC_FS (r = 0.82; p = 0.04).

In accordance with our hypothesis, AC_FS was sensitive to maturation stages in swimmers, with the pubertal group presenting significantly higher absolute AC_FS than middle-pubertal and early-pubertal groups. The pubertal and middle-pubertal groups presented greater muscle mass than early-pubertal; however, the difference between middle-pubertal and pubertal was of ≈7 kg on average, which can have practical influence on performance, besides the absence of statistical differences. Thus, expressing AC_FS values relative to total body mass and muscle mass is extremely important when comparing the anaerobic indices of swimmers of different biological ages.

These results agree with the findings of Kaczor et al. (2005), which have demonstrated that the quantity and activity of glycolytic enzymes are greater in more mature subjects. The study of Lätt et al. (2009) has also confirmed that net lactate accumulation was significantly greater when swimmers were on Tanner stages 3 and 4 than on stage 2, while no differences were found between stage 3 and 4; however, the authors did not take into account the alactic metabolism. When considering Ana_LAC and Ana_ALA, the latter only tended to be greater (p = 0.07) in pubertal than in the other groups. Thus, for swimmers, Ana_LAC is the main variable differing between maturation stages. Therefore, the difference in absolute AC_FS may be related to Ana_LAC since no differences were found in Ana_ALA between maturation stages. Furthermore, no differences were detected in relative AC_FS between maturation stages, indicating a possible influence of muscle mass on AC_FS.

Due to its importance in swimming context, a feasible tool to evaluate anaerobic contribution would be important, and AC_FS is practical because it enable swimmers to swim freely; however, it was important to compare it with currently used anaerobic contribution determination methods (i.e., AC_ALT and AOD).

The measurement of energy cost in swimming has received great attention on swimming, since it is important for performance (Zamparo et al., 2000). When calculating the netmetabolic power expenditure, both aerobic and anaerobic contribution must be accounted (Barbosa et al., 2006; Figueiredo et al., 2011). Faina et al. (1997) observed that the time to exhaustion at maximal aerobic speed is closely associated with anaerobic contribution in swimming, highlighting the importance of anaerobic metabolism for maximal efforts. To overcome AOD problems of excessive testing, an alternative method of AOD determination has been proposed using net lactate accumulation and off-transient oxygen consumption (Bertuzzi et al., 2010). As the oxygen consumption can be measured after swimming (Kalva-Filho et al., 2015; Campos et al., 2017a), AC_FS would be an even more interesting and applicable tool to evaluate the anaerobic contribution of swimmers without interfering on technique and speed.

The values of AOD observed in the present study were similar to those observed in exhaustive efforts (Ogita et al., 1996), but greater than other investigations that used fixed distance maximal efforts (Reis et al., 2010a,b). Ogita et al. (2003) investigated the possible influence of exercise duration on AOD values obtained in a swimming flume. Those authors observed that anaerobic contribution was similar when exhaustion occurred between one (≈2.8 L) and 5 min (≈2.9 L), with maximal values attained in 2–3 min (≈3.2 L). Thus, maximal AOD values (i.e., anaerobic capacity) can be obtained in a 200 m effort (2–3 min to exhaustion), with no significant differences in relation to a 400 m maximal effort (4–5 min to exhaustion) (Ogita et al., 2003). However, Reis et al. (2010b) observed lower values of AOD in a 400 m than in a 200 or 100 m maximal effort performed in front crawl (≈11.9 mL·kg⁻¹, ≈17.5 mL·kg⁻¹, and ≈21.0 mL·kg⁻¹, respectively). These results were confirmed in breaststroke for 200 and 100 m (≈23.1 mL·kg⁻¹ and 22.2 mL·kg⁻¹, respectively) (Reis et al., 2010b).

It has been suggested that combining sub and supraanaerobic threshold intensities (i.e., 30–90% of VO_2Max) affects the precision and validity of the AOD model (Buck and McNaughton, 1999). We did not analyze the anaerobic threshold of swimmers but ensured intensities greater than this physiological index by using the 400 m mean speed as well as a submaximal intensity (i.e., 95% of VO_2PEAK; unpublished data). Thus, although linear regression is the major concern for AOD calculation, this method is still considered the most acceptable for anaerobic evaluation (Noordhof et al., 2010; Reis et al., 2010b). Different from the present study, the AOD calculation performed in those above-mentioned studies used the effort mean speed to estimate demand, respecting the pace strategy of each swimmer. Thus, we calculated the estimated demand for each 25 m during the maximal effort (Figure 3), increasing the precision of these measurements. This approach together with the five points in the VO₂-speed relationship, indicate that AOD values were determined in a robust way during the present study, allowing its use to validate AC_ALT and AC_FS.

This is the first study to compare conventional AOD with AC_ALT in a maximal swimming effort in swimmers. Bertuzzi et al. (2010) compared a conventional and alternative method, in cicloergometer, to determine anaerobic contribution during an exhaustive cycling effort. Those authors observed similar values, positive significant correlation (r = 0.78) and a mean error very close to zero, which agrees with the present findings. Therefore, the difficulties implemented by the need for submaximal exercises to estimate VO₂-speed relationship are overcome in the alternative method. Finally, determination of AC_ALT allows the calculation of Ana_LAT and Ana_ALA separately, enabling the investigation of different training models on these two metabolisms.

Even though AC_ALT decreases the number of evaluations and allows the evaluation of Ana_LAT and Ana_ALA, it was still calculated with swimmers using snorkel during swimming. Besides changes in mechanics during swimming, the apparatus reduces the speed of the swimmers (330.5 ± 13.2 s vs. 303.6 ± 10.8 s), which might limit anaerobic contribution. Another important limitation refers to the impossibility of swimmers performing the turns and the underwater dolphin kick, a technique that has been commonly observed in swimming events. The use of snorkel also limits the use of “filipina” during breaststroke swimming, in addition to being uncomfortable for swimmers, limiting its use in practical settings.

We have shown no differences between AC_FS with AC_ALT and AOD; however, a tendency was detected in absolute values and an effect was found for relative anaerobic contribution (without detection in posthoc analysis). This might have occurred due to the reduced sample size. It is important to note that the limits of agreement between AOD and AC_FS highlighted a lower limit of 2.26 L. Four out of six presented significantly greater AC_FS than AOD (mean difference of 1.24 L). Thus, even though no statistical differences were observed, free swimming anaerobic contribution evaluation (AC_FS) might be recommended because it allows the athletes to perform in greater intensity, which is especially important since swimmers did not reach exhaustion during swimming.

The limitations of the present study were that athletes (both men and women) were evaluated in Part I which might have influenced the comparison between maturation stages, and the small sample size in Part II. It would be desirable to confirm these results with a larger sample size. Finally, for the Ana_ALA determination, 5 min of recovery was used. Bertuzzi et al. (2016) have observed that a minimum of 6 min is required for Ana_ALA evaluation; however, 5 min of recovery have been used in other studies (Kalva-Filho et al., 2015; Campos et al., 2017a; Andrade et al., 2021), and the fast component happens in the 1st min of recovery. Moreover, studies could also use bi-exponential decay equation as proposed by Scheuermann et al. (2011)—since it does not assume that athletes will reach baseline values at the end of recovery—and compare Ana_ALA using both Scheuermann et al. (2011) and Ozyener et al. (2001) equations.