Twenty-five physically active participants (12 women) were assessed. Anthropometric and pulmonary functional characteristics of the subjects are shown in Table 1. The participants reported no history of respiratory, cardiovascular, metabolic, musculoskeletal, neoplastic diseases or acute infections for at least 2 weeks before testing. They did not take anti-inflammatory medications, illicit drugs, antioxidants, or any other dietary supplements. All participants were thoroughly informed (in verbal and written forms) of the study procedures, and all of them signed informed consent forms. This study followed the Declaration of Helsinki (Harriss et al., 2017) and was approved by the ethics committee of the Faculty of Medicine of Pontificia Universidad Católica of Chile (project no. 19042213).

The participants were evaluated at the Laboratory of Exercise Physiology of Pontificia Universidad Católica of Chile during two sessions, separated by an interval of 24 h interval. All procedures were performed under laboratory environmental conditions (temperature, 22 ± 2°C; relative humidity, 40 ± 2%) and within a similar time frame (9:00 a.m.–2:00 p.m.). Participants were asked to avoid physical activities for 24 h before the measurements and to avoid alcohol, caffeine, and other stimulants and food for at least 3 h prior to the evaluations.

Each participant had an initial record of 90 s on the cycle ergometer, followed by 60 s of a baseline resting phase, during which data acquisition began was synchronized. For data analysis, data were compared across percentages of the test duration (0–100% of task), with the time of V.O_2–peak set as 100%. To complement the analysis, a triphasic model of exercise intensity was used, where ventilatory thresholds (VT1 and VT2) were calculated by two expert researchers using a visual method (Guazzi et al., 2012). An opinion of a third expert was considered in cases of discrepancy during the analysis (Vainshelboim et al., 2017). The values used for ventilatory variables [V.E, RR, and tidal volume (Vt)] and SmO₂ were obtained from the average of the last 30 s during each phase. At the end of the protocol, the exhaustion of the participants was evaluated using the Borg Modified Rating Perceived Scale (RPE) (see Supplementary Material I).

To compare differences of muscle oxygen levels between men and women, data are reported as: (i) maximum changes or differences (Δ), (e.g., ΔSmO₂-m.intercostales = SmO₂-m.intercostales rest—SmO₂-m.intercostales peak phase) and (ii) percentage of change, dividing the SmO₂ value of each phase by the SmO₂ of rest phase (%-change), (e.g., SmO₂-m.intercostales (%-change) = (SmO₂-m.intercostales phase⋅SmO₂-m.intercostales rest phase^–1). In addition, ΔSmO₂-m.intercostales was normalized by ΔV.E (SmO₂-m.intercostales ⋅ΔV.E^–1) and the ΔSmO₂-m.vastus lateralis was normalized to peak workload-to-weight (watts⋅kg^–1) (SmO₂-m.vastus lateralis⋅ PtW^–1).

Normality of the data was evaluated using the Shapiro-Wilk test. The descriptive characteristics of the participants were compared using the Student’s t-test. The differences between sexes with regard to SmO₂-m.intercostales, SmO₂-m.vastus lateralis, SmO₂ m.intercostales—m.vastus lateralis ratio, and ventilatory variables (V.E, RR, and Vt), which are expressed according to the % task, were analyzed using the two-way ANOVA test, reporting differences between conditions when the interaction of the factors (sex × time) was significant (p < 0.05). Subsequent multiple comparisons were analyzed used the Sidak post-hoc test. Comparisons of ΔSmO₂-m.intercostales⋅ΔV.E^–1 and ΔSmO₂-m.vastus lateralis⋅PtW^–1 were performed using the Student’s t-test. The correlation between ΔSmO₂-m.vastus lateralis (%) and ventilatory variables were assessed using Pearson’s correlation coefficient. Statistical significance was set at p < 0.05. The statistical analysis was performed using the GraphPad Prism (version 8.0; San Diego, California, United States).

Compared to women, men had higher absolute (3470 ± 436 ml⋅min^–1 vs. 2156 ± 189 ml⋅min^–1, respectively, p < 0.001) and relative (51.0 ± 5.3 ml⋅kg⋅min^–1 vs. 39.3 ± 3.0 ml⋅kg⋅min^–1, respectively, p < 0.001) V.O_2–peak values. Regarding the workload (watts), men demonstrated greater values in absolute terms (256 ± 23 vs. 164 ± 17, p < 0.001) and when normalized to body weight (PtW) (3.8 ± 0.4 vs. 3.0 ± 0.3, p < 0.001). Regarding the ventilatory variables, men reached greater V.E_max expressed in absolute terms (L⋅min^–1) (142.9 ± 21.6 vs. 98.9 ± 14.9, p < 0.001) and in %-change (15.6 ± 4.3 vs. 9.9 ± 2.2, p = 0.037) (see Supplementary Material I). The interaction of factors (sex × time) was different in the V.E (L⋅min^–1) (p < 0.001) and V.E (%-change) (p < 0.001). When expressing this variable regarding the % task in absolute terms (see Figure 3A) and %-change, men reached a greater value from 60 to 100% of the protocol (p < 0.001) (see Figure 3B). Concerning the RR_max expressed in absolute terms (bpm), there were no differences between sexes (p = 0.457). The interaction of factors (sex × time) was not different in the RR (breaths⋅min^–1) (p = 0.857), but RR (%-change) was different (p < 0.001). The level of RR (%-change) in the peak phase was greater in men than in women (4.1 ± 1.2 vs. 2.9 ± 0.4, p < 0.001). When expressing this variable in relation to the % task, men reached a greater value from 90 to 100% of the protocol (p < 0.001) (see Figure 3F). The level of Vt_max expressed in absolute terms (L) was greater in men than in women (2.9 ± 0.3 vs. 2.0 ± 0.3, respectively, p < 0.010). The interaction of factors (sex × time) was different in the Vt (L) (p < 0.001), but not in the Vt (%-change) (p = 0.070). When expressing absolute Vt_max in relation to the % task, it was greater in men (between 30 and 100% during the protocol, p < 0.001) (see Figure 3B). Regarding the Vt (%-change), the value was greater in men (between 80 and 90% of the test, p < 0.027) (see Figure 3E).

The variables mentioned above that consider the triphasic model of exercise intensity are presented in Supplementary Material I.

Women show a greater decrease of SmO₂-m.intercostales (%-change) from 40 to 100% of the maximal exercise test, possibly attributed to the great relative recruitment of the respiratory musculature, required to reach the V.E level to meet the exercise metabolic demands (Contreras-Briceño et al., 2021). This is consistent with previous findings that reported that the contractile activity of accessory respiratory muscles (m.sternocleidomastoideus and m.scalenus) was greater in women during a bicycle endurance test at a workload of 85%. This is attributed to their thoracic and pulmonary characteristics and dependence on thoracic and accessory respiratory muscles during intense exercise has been postulated to be an adaptation that diminishes the risk of diaphragmatic fatigue (Mitchell et al., 2018). Women have increased resistive respiratory (WOB_R) and elastic (WOB_e) work, lower forced vital capacity, and high limited expiratory flow (Sheel and Guenette, 2008; Sheel and Romer, 2012). A similar study reported that this greater level of WOB increases the respiratory oxygen consumption in women during maximal exercise (13.8 vs. 9.4% in men) (Lomauro and Aliverti, 2018). Our findings are aligned with these data, as we found a greater ΔSmO₂-m.intercostales in women than in men (39% ± 9% vs. 27% ± 19%, respectively) (see Supplementary Material I).

In both groups, the participants who achieved greater V.E values induced greater deoxygenation in m.intercostales, which is consistent with the finding of a previous study (Contreras-Briceño et al., 2019). Interestingly, even though women achieved a lower ΔV.E than men (88.7 ± 14.5 vs. 133.1 ± 21.8 L⋅min^–1, respectively), they induced a higher level of ΔSmO₂-m.intercostales normalized by ΔV.E; this is consistent with a previous study that reported women reaching a higher level of V.O_2–RM in spite of only achieving a V.E that was 25% of men’s values (Lomauro and Aliverti, 2018). Another related observation was that participants who demonstrated a greater decrease of ΔSmO₂-m.intercostales also showed a larger change in RR, for which may be an effect of ventilatory strategy during exercise.

Considering the triphasic model of exercise intensity proposed by Skinner et al. (Skinner and McLellan, 1980), the SmO₂-m.intercostales showed early changes that were lower in women at VT1 (0.8 ± 0.1 vs. 1.0 ± 0.1 in men) and VT2 (0.6 ± 0.4 vs. 0.8 ± 0.1 in men). Similarly, in women, exercise intensities on VT1 lead to a greater level of oxygen consumption by the respiratory muscles (V.O_2–RM), a phenomenon recognized as a limiting factor in physical performance progression in women (Guenette and Sheel, 2007). However, there are no differences between sexes at the peak exercise phase, which could be explained by the exponential increase of V.E at the expense of RR (Wanke et al., 1991).

Our results show that men have greater decreases in SmO₂-m.vastus lateralis from 60 to 100% of the workload, possibly because of a greater contraction speed and cross-sectional area in locomotor muscles, leading to a sustained high intensity of exercise requiring a higher level of oxygen consumption level in peripheral muscles (Hakkinen, 1993; Fulco et al., 1999). In addition, as expected, men reached a higher absolute peripheral workload (256 vs. 164 watts in women) and PtW (3.8 vs. 3.0 watts⋅kg^–1 in women), which is in agreement with previous reports, wherein it was found that men perform more peripheral muscle work that is supported by a higher neuromuscular activity and IIx fiber recruitment (Miller et al., 1993). Thus, our study found that greater decreased ΔSmO₂-m.vastus lateralis normalized by relative peripheral workload was higher in men (see Figure 4B). Further, women have a larger fiber I ratio and a higher capillary density in peripheral muscles, which provides better oxygen supply allowing them to sustain muscle work with a higher oxygen consumption capacity. Such related muscle mass and fiber type characteristics are worthy of investigation in future studies (Simoneau and Bouchard, 1989; Glenmark et al., 2004).

Another aspect to discuss is the length of time of exercise protocol used. The literature suggests that peak oxygen consumption tests should not exceed 10 min of duration to limit muscle fatigue as a cause of exercise cessation (Buchfuhrer et al., 1983). However, this finding has been questioned by Midgley et al. (2008) who showed that longer test generates higher V.O_2–peak values, suggesting that the cycloergometer tests should last between 7 and 26 min (Midgley et al., 2008). In this sense, incremental exercises by 20 watts⋅min^–1 had been associated with progressive deoxygenation in the m.vastus lateralis, m.rectus femoris, and m.vastus medialis, similarly to our findings (Chin et al., 2011). Consequently, utilization of this time-stage protocol can induce higher power and total work than ramp tests (Zuniga et al., 2012). Based on these previous findings, we used an exercise protocol with a longer 2-min incremental stage, without discarding that the fiber recruitments associated with exercising could affect the results obtained; this is a relevant aspect to elucidate in future research.

Synchronous evaluation of both muscle groups determined the SmO₂ ratio by calculating the preponderance of the muscle group with a higher oxygen demand during the study protocol. In this regard, we found that women had fewer changes than men (absolute, between 80 and 100%; relative, from 70 to 100%) (see Figures 3C,D). These findings could be explained by the increase in the cost of breathing, which leads to a higher blood flow to respiratory muscles, probably restricting oxygen supply to the locomotor muscles (metabolic reflex) (Boushel, 2010). Moreover, women reach a higher level of V.O_2–RM because of the increase in V.E, which limits oxygen supply to the locomotor muscles (Dominelli et al., 2015, 2017), also consistent with the study by Guenette et al. (2007). Nonetheless, other authors have reported a homogeneous distribution of blood flow in women, resulting from a lower sympathetic stimulation and vascular resistance in lower extremities, which does not negatively impact exercise progression (Smith et al., 2016). These conflicting data require further examination that would include evaluation methods to clarify of the effect of metabolic reflex on sports performance in women.

By device limitations, it was not possible to completely assess the respiratory and locomotor muscles in a single test. The lack of adipose thickness measures by skinfold and/or ultrasonography reports limits verification that the muscle tissue was within the measuring range of the MOXY device (12.5 mm) (McManus et al., 2018). However, our participants were at the lower end of the normal BMI range. In addition, cyclic hormonal variations should be considered in women participants, given that edema, dehydration, and altered thermoregulation are factors affecting physical performance by modifying the ventilatory center response (Kilbride et al., 2003); this is an aspect that requires further study. Future studies could consider the stage of the menstrual cycle; however, menstrual cycle is not consistently found to influence exercise performance (McNulty et al., 2020). Recently, it was reported that 36% of high-performance female athletes stated that their menstrual cycle impacted negatively on their performance for at least some or most of the time (Heather et al., 2021). In contrast, it did not affect the production of muscle strength and power (McNulty et al., 2020; Dasa et al., 2021). Furthermore, we recommend to future research could consider the confounding influence of physical fitness levels of participants.