The experiments were conducted on two sessions carried out in a controlled laboratory setting on 2 days apart at the same time of day. Participants were instructed not to undertake any strenuous activity during the 24 h preceding each session. The first session was dedicated to gathering participants’ physical characteristics (anthropometric measurements) and V̇O_2max assessment. Then, during the second session, the rowers covered a race distance of 2,000 m for U18, 1,500 m for U16 and 1,000 m for U14, on a rowing ergometer (Model D, Concept2, Morrisville, VT, United States A) as fast as possible, according to the recommendations of the French rowing federation. All the rowers were already fully familiarized with the equipment. The computer of the ergometer continuously delivered the power output (in W). The investigators set the resistance factor between 100 and 130 according to age and the expertise level of young rowers. The same resistance factor was used for the two exercise sessions. Verbal encouragement was systematically provided by the investigators during each exercise session.

Thirty-one male competitive rowers aged from 12 to 17 years volunteered to participate in the present study. They were classified into three age categories 1) under 18 years (U18: 16.0–17.9 years, n = 9), 2) under 16 years (U16: 14.0–15.9 years, n = 10) and 3) under 14 years (U14: 12.0–13.9 years, n = 12). For the whole population, all tests were carried out between October and March, i.e., during the winter training period, before the start of the competitive season in May. All participants trained 3 to 6 times per week (i.e., 2–4 “on-water” training sessions and 1–2 physical training sessions) in the year preceding the experiments and have already participated in regional or national competitions. None of the participants had a family history of cardiovascular disease or was using any medication. The present study was approved by an institutional ethics review board (Comité d'Éthique pour la Recherche en Sciences et Techniques des Activités Physiques et Sportives—CERSTAPS, n° 2017-29-11-20) and conformed to the standards of use of human participants in research as outlined in the Sixth Declaration of Helsinki. The participants were informed of the experimental procedures and gave their written consent before any testing was conducted. Also, written informed consent was obtained from the parents or legal guardians of the participants.

Age, height and BM are detailed in Table 1. Maximal heart rate (HR_max) was similar among the three age groups. Maximal oxygen uptake (V̇O_2max) was not significantly different between U18 and U16 (Table 1), but 27% lower in U14 than U16 (g = 1.80, p < 0.001). The power output corresponding to V̇O_2max (Pa_max) was significantly lower in U16 than U18 (− 28%, g = 1.07, p < 0.05) and in U14 than U16 (−28%, g = 1.49, p < 0.01).

Oxygen uptake at rest (V̇O_2rest) was 0.58 ± 0.08 (0.53–0.64) L∙min⁻¹ for U18, 0.56 ± 0.07 (0.51–0.61) L∙min⁻¹ for U16 and 0.47 ± 0.07 (0.42–0.51) L∙min⁻¹ for U14. The mean time (T_perf) to cover 2,000 m was 7 min 02 ± 18 s (6 min 48 s–7 min 16 s), 5 min 28 ± 18 s (5 min 15 s–5 min 40 s) for 1,500 m, and 4 min 06 ± 25 s (3 min 50 s–4 min 21 s) for 1,000 m. As indicated in Table 2, the mean power output sustained during the rowing ergometer exercise (PO_perf) was not significantly different between U18 and U16 but significantly lower in U14 than U16 (−27%, g = 1.40, p < 0.01). Relative to Pa_max, %PO_perf was not significantly different among the three age categories. The mean oxygen consumption sustained during the race was significantly higher in U18 than U16 (+13%, g = 1.11, p < 0.01) and in U16 than U14 (+26%, g = 1.56, p < 0.001). However, relative to V̇O_2max, %V̇O_2perf was not significantly different among the three age categories. [La]_wp was similar among the three age categories. While no significant difference was observed for [La]_max and ∆[La] between U18 and U16, [La]_max and ∆[La] were significantly lower in U14 than U16 (−15%, g = 1.42, p < 0.01 and −12%, g = 1.62, p < 0.01, respectively).

The total work produced during the rowing ergometer test (W_Tot) was significantly lower in U16 than U18 (−30%, g = 3.69, p < 0.001) and in U14 than U16 (−46%, g = 4.49, p < 0.001).

The amounts of energy released from each metabolic pathway, expressed in absolute value, are reported in Table 3. E_Ae and E_An were significantly lower in U16 than U18 (−34%, g = 5.18, p < 0.001 and -20%, g = 1.39, p < 0.01, respectively) and in U14 than U16 (−47%, g = 4.95, p < 0.001 and −28%, g = 1.77, p < 0.001, respectively). E_AnLa and E_AnAl were also significantly lower in U16 than U18 (−22%, g = 1.37, p < 0.05 and −10%, g = 1.17, p < 0.05, respectively) and in U14 than U16 (−31%, g = 1.81, p < 0.001 and -11%, g = 1.03, p < 0.05, respectively). As a result, E_Tot was significantly lower in U16 than U18 (−33%, g = 4.56, p < 0.001) and in U14 than U16 (−44%, g = 4.37, p < 0.001). E_Ae, E_An and E_Tot are illustrated by race distance (or age category) in Figure 1A.

BM was positively and significantly correlated to E_Tot (r ² = 0.67, p < 0.001) and W_Tot (r ² = 0.66, p < 0.001).

Allometric scaling exponents obtained from Eq. 2 are detailed in Table 4.

When scaled for BM ^b1 , E_Ae, E_An and E_Tot were not significantly different between 2,000 m for U18, 1,500 m for U16, and 1,000 m for U14 (p > 0.05) (Figure 1B). Similarly, no significant difference was observed for W_Tot scaled for BM ^b1 between the three race distances.

W_Tot was significantly correlated to E_Tot when expressed in absolute value (Figure 2A) or scaled for BM (Figure 2B).

As illustrated by Figure 3 , %E_Ae was lower in U16 than U18 (-3.4% g = 1.59, p < 0.01) and lower in U14 than U16 (−4.1% g = 1.72, p < 0.001), while %E_An was higher in the same proportions in U16 than U18 and in U14 than U16. %E_AnLa and %E_AnAl were also significantly higher in U16 than U18 (+1.8%, g = −1.30, p < 0.01 and +0.5%, g = −3.31, p < 0.001, respectively) and in U14 than U16 (+3.1%, g = −1.30, p < 0.01 and +1.1%, g = −4.95, p < 0.001, respectively).

Rowing performance parameters have been mostly investigated in adult rowers, particularly in high-level athletes (Ingham et al., 2002; Bourdin et al., 2004; Mikulić, 2011; Bourdin et al., 2017). Little information is available about specific rowing performance in competitive 12- to 17-year-old rowers. To the best of our knowledge, only three studies reported data that can be explicitly compared to our outcomes (Russell et al., 1998; Mikulić and Ružić, 2008; Giroux et al., 2017). For instance, in U18, rowing performance was 4.8% lower in our rowers compared to those evaluated by Russell et al. (1998) on 2,000 m (7 min 02 ± 18 s vs. 6 min 43 ± 16 s, respectively). In U16, our data are comparable to those reported by Giroux et al. (2017) on 1,500 m (5 min 28 ± 18 s vs. 5 min 26 ± 20 s, respectively). In U14, rowing performance was 2.8% higher in our rowers compared to those evaluated by Mikulić and Ružić (2008) on 1,000 m (4 min 06 ± 24 s vs. 4 min 13 ± 24 s, respectively). In addition, V̇O_2max measured in our rowers (Table 1) were comparable to those reported by Russell et al. (1998) and Mikulić and Ružić (2008) in U18 and U14 (4.6 ± 0.4 and 2.8 ± 0.5 L min⁻¹, respectively). Taken together, these data suggest that rowing performance and physical fitness were comparable among our French competitive rowers and U18 Australian and U14 Croatian national-level rowers.

Despite their young age, the rowers of the present study were able to sustain exercise intensity between 95 and 98% of their V̇O_2max for ∼7 min over the 2,000 m in U18 and for ∼4 min over the 1,000 m in U14 (Table 2). This finding is consistent with the data reported by Gillies and Bell (2000), showing a mean exercise duration of 7 min 06 s sustained at 94% of V̇O_2max over the 2,000 m rowing ergometer performance in adult rowers. Similarly, mean mechanical power outputs (PO_perf) sustained over the 2,000 in U18, 1,500 in U16 and 1,000 m in U14 were found to be 109, 114 and 117% of Pa_max, respectively. These remarkable results are closely comparable to those reported in high-level rowers performing an exhaustive 2,000 m exercise on a rowing ergometer (113% of Pa_max) (Bourdin et al., 2004). Thus, our data show for the first time that young competitive rowers were able to maintain supra-maximal intensities (i.e., higher than Pa_max) during prolonged exercise (from ∼4 to 7 min). This finding may be associated with their capacity to highly stimulate glycolytic metabolism, as evidenced by the high values of [La]_max reported at the end of the 2,000, 1,500 and 1,000 m exhaustive exercises (Table 2).

Our results showed that the absolute amount of energy released from aerobic (E_Ae) and anaerobic (E_An) energy pathways decreased with the reduction of rowing competition distances (Table 3). Comparatively, using a similar estimation method, de Campos Mello et al., 2009 reported values of E_Ae, E_An and E_Tot in adult rowers comparable to those obtained in our U18 rowers (563 vs. 560 kJ for E_Ae, 106 kJ vs. 84 kJ for E_An and 674 vs. 645 kJ for E_Tot).

From the method of accumulated oxygen deficit, Diry et al. (2020) showed that the quantity of anaerobic energy released during a 60 s ‘all-out’ exercise in young competitive rowers was positively influenced by their body dimensions. In the present study, an original allometric approach was used to remove the dimensional effect of BM on the absolute quantity of aerobic and anaerobic energy supplied during the 2,000-, 1,500-, and 1,000-m rowing exercises. The current results show that the BM-specific allometric coefficients (i.e., b ₁ in Eq. 1 and Table 4) are lower than 1 (0.76 for E_Ae and 0.84 for E_Tot), suggesting that E_Ae and E_Tot would increase in lower proportions than BM. Conversely, the BM-specific allometric coefficient was 1.20 for E_An (Table 4) suggesting that E_An would increase in higher proportion than BM. Interestingly, the BM-specific allometric coefficient associated to E_An (i.e., 1.20) is comparable to one previously reported by Maciejewski et al. (2016b) regarding power output during a modified rowing Wingate test in U16 rowers (i.e., 1.24).

In the present study, the total work (W_Tot) produced over 2,000, 1,500, and 1,000 m was found to be closely correlated to the total energy (E_Tot) expended over the corresponding distances. Our results show that W_Tot accounted for 98% in the variation of E_Tot (Figure 2A). This significant result clearly illustrates that, despite the assumptions made to estimate the relative aerobic and anaerobic contibutions (see Methods section), a very close relationship persists between the mechanical (W_Tot) and physiological (E_Tot) parameters. In addition, when calculating the rowing gross efficiency (in %), as E_Tot divided by W_Tot, there were no significant differences among categories: 19.5, 20.5 and 19.7% for U18, U16 and U14, respectively. Interestingly, these results are in the same range as those of highly trained adult rowers (i.e., 18.5%) (Bourdin et al., 2004), and show that young competitive rowers are able to reach the same rowing gross efficiency as their adult elite counterparts.

However, further analysis shows that BM would explain 66 and 67% of the variations in W_Tot and E_Tot, respectively. The allometric method used to remove the effects of BM on W_Tot and E_Tot shows that these two parameters have similar BM-specific allometric coefficients (Table 4), and they are no longer correlated when scaled for their respective allometric coefficients (r ² = 0.09) (Figure 2B), suggesting that BM would have a very strong influence (91%) on this relationship. Therefore, our outcomes clearly confirm 1) the interest and accuracy of the method used in the present study for estimating the energy amounts derived from aerobic and anaerobic pathways and 2) the need to use allometric modeling to appreciate the influence of BM on W_Tot and E_Tot.

The outcomes of the present study confirm that anaerobic metabolism is a non-negligible energy source during rowing competitions, whatever the considered competition distances (Figure 3). Although a statistical difference exists, the estimated relative anaerobic contributions over the three distances (13.0% for 2,000 m, 15.3% for 1,500 m and 19.4% for 1,000 m) were found to be relatively comparable from a practical perspective. Because these values are estimated and not directly measured, we cannot exclude approximations inherent to our calculation methods (vide supra Method section). However, it is worth noting that our estimations are similar to those previously reported in the literature for adult rowers covering the distance of 2,000 m. For example, using a similar method to ours, de Campos Mello et al., 2009 reported that the anaerobic pathways provided about 16% of the total energy expended over a 2,000 m rowing exercise in national-level adult rowers. Similarly, over the Olympic distance, Secher et al. (Secher et al., 1982; Secher, 1983) estimated from the oxygen debt method that about 14% of the energy expended was of anaerobic origin in elite rowers. These results were confirmed by Pripstein et al. (Pripstein et al., 1999), using the accumulated oxygen deficit method in university adult rowers (12%). The similarities between our results and those of previous studies (Secher et al., 1982; Secher, 1983; Pripstein et al., 1999; de Campos Mello et al., 2009) are consistent despite the 1) differences in rowing exercise duration between child, adolescent and adult rowers and, 2) heterogeneous muscle mass between populations, which is known to influence anaerobic energy supply (vide supra) (Bangsbo et al., 1993; Gastin, 2001; Diry et al., 2020).