Improving cardiorespiratory fitness (CRF) concomitant with metabolic function is an important component of swimming training (Jorgić et al., 2011) and serves as a key predictor of swimming performance and fitness (Lahart and Metsios, 2018). Several effective and time-efficient exercise strategies have been identified that improve CRF and induce positive metabolic changes (Latt et al., 2009). The search for more effective training interventions for both competitive swimmers and recreational swimmers has expanded to include several external aids that can enhance CRF variables including oxygen uptake (VO₂), respiratory minute ventilation (VE), tidal volume (VT), respiratory frequency (RF), heart rate (HR), and associated metabolic measures such as lipid oxidation (Heinicke et al., 2005; Koppers et al., 2006; Millet et al., 2010).

One method that has shown promise is inducing a condition known as hypercapnia or the abnormal elevation of the partial pressure of carbon dioxide in arterial blood (pCO₂) above 45 mmHg (Kato et al., 2005; Schneider et al., 2013). A low-cost strategy that can safely induce hypercapnia is by increasing the volume of external respiratory dead space such as by breathing through a tube (Hebisz et al., 2013; Khayat et al., 2003; Zatoń et al., 2010). According to numerous studies (Cathcart et al., 2005; Toklu et al., 2003), the added respiratory dead space (ARDS) supplemented by tube breathing increases the amount of CO₂ that is inspired in subsequent breaths proportional to the added volume. The retention of CO₂ induces a condition known as acute respiratory acidosis (Jones, 2008; Kraut and Madias, 2014). In response to increased CO₂ levels, chemoreceptors provide respiratory feedback (Kumar and Bin-Jaliach, 2007) increasing VE, VT, and RF (Mercier et al., 1992) to levels observed at higher training intensities (Zatoń and Smołka, 2011). Several studies suggested that regular training with induced hypercapnia can condition muscle cells to utilize lipids at a greater rate than other substrates and enhance the reduction of body fat content (Hollidge-Horvat et al., 1999; McLellan, 1991; Østergaard et al., 2012).

Original research on exercise with ARDS dates back to the 1970s in which participants breathed through a 1200 ml tube during 6 min of ergometer cycling corresponding to 30%, 50%, and 70% of predicted maximal oxygen uptake (VO₂max) (Kelman and Watson, 1973). This protocol resulted in an increase in RF and VE with the latter attributed to increased VT. In the previous study Zatoń et al. (2010) have found that a 5-week endurance cycling intervention with 1000-ml ARDS improved performance (total work) in the incremental exercise test with simultaneous decreases in hematocrit, hemoglobin, and red blood cell concentrations. In another study by the same research group, the respiratory response to cycle ergometry at 100 W with increasing ARDS volume (up to 1600 ml) was investigated (Zatoń and Smołka, 2011). In this study, VE increased proportional to the volume of ARDS and both pCO₂ and blood pH decreased. A study on incorporating 1000 ml ARDS in a 10-week training macrocycle in trained cyclists showed increases in VO₂max, VCO₂max, and VEmax during exercise testing although a decrease was observed in mechanical efficiency (Hebisz et al., 2013). Another investigation on cycling training with 1200 ml ARDS at submaximal intensities found no reduction in RER and VCO₂ but an increase in endurance capacity (Smołka et al., 2014). A recent study involving triathlon athletes included ARDS in an 8-week interval training intervention and found greater increases in VCO₂max, VE, maximal power, and post-exercise lactate concentrations when compared with a non-ARDS cohort (Michalik et al., 2018).

In the swimming domain, Adam et al. (2015) introduced ARDS in an 8-week training intervention targeting glycolytic power in a sample of competitive swimmers. Two sessions were held per week in which repeated 50-m swims were performed at maximal intensity until volitional exhaustion. This protocol enhanced 100-m freestyle swim speed as well as VO₂max and VE ascertained in an incremental exercise test. Another study examined the effects of a 5-week swimming intervention (two session per week) with ARDS in competitive swimmers and observed reduced HR and a lowered physiological cost of swimming (Zatoń and Ziebura, 2012). Additionally, Zoretić et al. (2014) investigated an 8-week hypercapnic-hypoxic training program in elite swimmers. At the end of the program they observed a 5.35% increase in hemoglobin concentration and a 10.79% increase inVO₂max. Karaula et al. (2016) examined the effects of 8 weeks of hypercapnic–hypoxic training in elite swimmers and reported improved swimming efficacy and increased maximal inspiratory and expiratory muscle strength.

The aforementioned studies (Zatoń and Ziebura, 2012, Adam et al., 2015) involving elite swimmers suggests that swimming with ARDS may enhance swimming performance and CRF and maintain CRF among athletes during the off-season mesocycle when exercise intensity is reduced. However, less is known about the effects of ARDS in recreational swimmers as previous studies recruited elite or trained cohorts. Furthermore, as a reduction in fat tissue is a common goal of recreational swimmers, there is a lack of data on changes in substrate metabolism after several weeks of swimming with ARDS. Hence, investigating the effects of swimming with ARDS on CRF and lipid metabolism in recreational swimmers can provide new insights on the use of this exercise enhancement. Therefore, the aim of this study was to assess the circulatory, respiratory, and metabolic effects of a moderate-intensity swimming intervention with ARDS in recreational swimmers. Moderate intensity exercise was selected as the literature recommends prescribing this intensity level to promote lipid utilization and cardiovascular adaptations (Alkahtani, 2014). It was hypothesized that ARDS would show large improvements in CRF with beneficial metabolic changes towards increased exercise-induced lipid oxidation.

Twenty-two healthy male and female physical education students were recruited. All swam regularly for recreational purposes averaging 2 km twice a week. Subjects were assigned to a control (C) and experimental (E) group based on VO₂max to ascertain similar CRF between the two groups. VO₂max was obtained for each participant in an incremental exercise test (detailed below). The VO₂max scores were ranked from highest to lowest and participants were then alternately assigned to group E (n = 11, age 24.27 ± 2.69 years, body height 1.73 ± 0.09 m, body mass 70.03 ± 13.14 kg, VO₂max 45.55 ± 7.47 mL·kg^-1·min^-1) and group C (n = 11; age 24.00 ± 3.35 years, body height 1.68 ± 0.03 m, body mass 72.32 ± 10.13 kg, VO₂max 47.09 ± 8.85 mL·kg^-1·min^-1). The groups were compared at this stage with the non-parametric Wilcoxon test (α = 0.05) and no between-group differences were observed for age (p = 0.72), body height (p = 0.50), body mass (p = 0.65), and VO₂max (p = 0.65). This generated comparable intervention groups with an objective baseline measurement of CRF.

All participants provided their written consent to participate in the study and were informed they could withdraw at any time. They were instructed to maintain their normal lifestyle and diet and refrain from any additional exercise external to the training prescribed in the study. The design was approved by the local ethics committee (No. 14/2017) and all procedures adhered to the prerogatives set out in the Declaration of Helsinki.

An incremental exercise test (IET) on a cycle ergometer was administered 3 days before and after the intervention to assess changes in CRF pre- and post-intervention. The IET was performed in controlled conditions in a climate-controlled (24ºC and 50% relative humidity) exercise laboratory (PN-EN ISO 9001:2009 certified). Testing was executed at the same time of day and participants were instructed to refrain from caffeine 24 hours prior to testing (Atkinson and Reilly, 1996). The cycle ergometer (Excalibur Sport, Lode BV, Netherlands) was adjusted to each participant and calibrated before each trial. Starting workload was 0 W and linearly increased by ~0.27 W s^-1 (50 W·3 min^-1) until volitional exhaustion. Gas exchange was measured breath-by-breath using a metabolic cart (Quark b2, Cosmed, Italy). The device was calibrated with a reference gas mixture of CO₂ (5%), O₂ (16%), and N₂ (79%). Respiratory function was measured 2 min prior to test start and continued 5 min after test conclusion with data averaged over 30-s intervals. HR was also continually measured with a non-invasive heart rate monitor (S810, Polar Electro, Finland). Primary outcome measures of VO₂ [mL·kg^-1·min^-1], VCO₂ [ml·min^-1], VE [l·min^-1], RER [VCO₂·VO₂^-1], and HR [beat·min^-1] were determined at four workloads (50, 100, 150, 200 W) and at maximal power (Max). VO₂max was defined as the highest 30-s average at in which relative VO₂ values plateaued (<1.35 mL·kg^–1·min^–1) despite an increase in workload or two of the following criteria: (a) RER >1.10, (b) attainment of HRmax (within 10 bpm of age-predicted maximum [220–age]), (c) voluntary exhaustion.

Twelve swimming sessions were administered over a period of 6 weeks with at least 72 h of rest provided between each session. All sessions were held in a short course 25-m swimming pool in homogenous conditions (water temperature 27°C, air temperature 28°C at a relative humidity of 60%, pool illuminated at 600 lumens). The participants swam the front crawl for 50 min at a moderate intensity corresponding to 60% IET-obtained VO₂max. This intensity was selected as it was below the lactate threshold and suitable for prolonged exercise in untrained subjects and most optimal for enhancing lipid metabolism (Aellen et al. 1993). Intensity was translated into a target HR for each individual and monitored by the subject with a waterproof heart rate monitor (RS400, Polar Electro, Finland) during each open turn (lap completion).

Group C breathed using normal front crawl breathing technique. Group E swam with a custom face mask integrated with a polypropylene frontal snorkel and 2.5-cm diameter ribbed tubing to provide 1000 ml of physiologic dead space (Figure 1). Dead space volume was identical for each participant and measured by filling the snorkel with water and then transferring the volume to a graduated cylinder as per Smolka et al. (2014). This volume was added with the volume of the custom face mask to ascertain a total volume of 1000 ml. The snorkel was sufficiently rigid to maintain a constant volume when swimming.

Previous reliability and validity testing for this apparatus showed that a 1000 ml volume can be sufficient to provoke hypercapnia (partial pressure of pCO₂ in arterial blood >45 mmHg) and induce respiratory acidosis in which pCO₂ increased with a concomitant decrease in blood pH without causing hypoxia (Zatoń et al., 2010; Zatoń and Smołka, 2011; Hebisz et al., 2013; Smołka et al., 2014). This apparatus had also been previously verified for use during swimming (Zatoń and Ziebura, 2012; Adam et al., 2015). When swimming with the snorkel, the subjects wore a nose clip to eliminate nose breathing. Breathing frequency and the ratio of inhalations to exhalations were not controlled.

Blood was drawn from the fingertip before and 3 min after completing the first (pre-intervention) and last (post-intervention) swim session to quantify the change in plasma glycerol concentration (GLY, micromol·l^-1). GLY was collected at this time point as the swimming sessions were sufficient in duration to fully promote lipolysis (Stallknecht et al. 1995). Quantitative enzymatic determination of GLY was performed using free glycerol reagent (Sigma-Aldrich, USA). The difference (∆) before and after the first and last swim session was compared to serve as an indirect measure of change in pre- and post-intervention lipid metabolism.

Body mass index (BMI; kg·m^-2), fat mass (FM; kg), and fat-free mass (FFM; kg) were determined by near-infrared interactance (6100/XL, Futrex, Great Britain) at the middle of the biceps brachii muscle of the dominant extremity. Measures were collected pre- and post-intervention as an indirect measure of changes in lipid metabolism. The measurements were taken by a laboratory member with the device calibrated before each trial (Fukuda et al., 2017).

Data are given as means ± standard deviations (SD) and the difference (∆) between pre- and post-intervention values. The distribution of the data set was screened for normality using the Shapiro–Wilk test and the homogeneity of variances checked with Levene’s test. Cardiorespiratory outcomes (VO₂, VCO₂, VE, RER, HR) for each workload (50, 100, 150, 200 W and Max) were compared using repeated measures multivariate analysis of variance (MANOVA) for dependent samples and Tukey’s Honest Significant Difference (HSD) test for pairwise post-hoc comparisons.

For anthropometric and metabolic outcomes (BMI, FM, FFM, ∆GLY), a normal distribution was not confirmed and the non-parametric Wilcoxon signed-rank test was used. Effect sizes were calculated using Cohen’s d and interpreted as small (d=0.20 to 0.49), moderate (d=0.50 to 0.79), and large (d>0.80). All calculations were performed with the Statistica 13.1 software package (StatSoft, USA) and followed accepted methodology (Thoma et al., 2015). Significance was set at an alpha level of 0.05 for all statistical procedures with p values provided for all results. The sample size was estimated using a stand-alone power analysis program for statistical tests (G*Power 3.1.9.2, Kiel University, Germany) (Faul et al., 2007) with a small effect size of f² = 0.29. Assuming an alpha error of 0.05 and power of 0.80, the required total sample size was estimated to be 26 subjects. However, due to the length and commitment of the intervention, we were able to include only 22 individuals in the final analysis.

Pre- and post-intervention cardiorespiratory and metabolic outcomes are presented in Table 1. No differences between groups were observed at post-intervention for VO₂ (p = 0.75), VCO₂ (p = 0.85), VE (p = 0.93), RER (p = 0.98), HR (p = 0.44), BMI (p = 0.69), FM (p = 0.71), FFM (p = 0.43), and ∆GLY (p = 0.21) (Table 1). Post-hoc analysis revealed a within-group difference only in group E for RER at maximum workload (p = 0.04, d = 1.30).

Several studies have investigated the longitudinal effects of incorporating ARDS as an external aid to further enhance exercise-induced cardiovascular adaptations and lipid utilization in a variety of exercise modalities and intensities (Hebisz et al., 2013; Smołka et al., 2014). However, no studies to date have investigated the effects of an extended duration intervention in recreationally trained swimmers. The present experiment does not corroborate earlier findings, nor does it confirm the positive effects of ARDS-induced hypercapnia on respiratory, cardiovascular, and metabolic function (Toklu et al., 2003; Cathcart et al., 2005; Kato et al., 2005; Kumar and Bin-Jaliach, 2007; Kraut and Madias, 2014; Zoretić et al., 2014; Karaula et al., 2016). Post-intervention VO₂, VCO₂, VE, HR in the IET did not change in any of the selected workloads and only RER increased at maximal workload. No changes were observed in BMI, FM, FFM, and ∆GLY.

Previous research on hypercapnic exercise reported a decrease in RER, which had been interpreted as a substrate shift towards increased lipid utilization (Kato et al., 2005; Østergdard et al., 2012), which was confirmed via tissue analysis (Hollidge-Horvat et al., 1999). Hence, our original assumption that we would observe enhanced lipid oxidation with reduced body fat content when compared with identical albeit non-hypercapnic swim training. Instead, we found an isolated increase in post-intervention RER that was observed only during incremental exercise at maximal workload and not steady-state exercise as in the aforementioned studies. Previous studies have found VCO₂ to increase during incremental exercise as a result of hyperventilation and that the increased buffering of blood lactic acid may cause RER to no longer reflect substrate usage but higher ventilation rates and blood lactate levels (Kenney et at., 2012).

No change was observed in ∆GLY in either the control or experimental group, confirming a similar level of hydrolyzed triglyceride with no enhanced reduction in adipose tissue. This was also confirmed by similar pre- and post-intervention BMI, FM, and FMM in the experimental group. This lack of change in ∆GLY could be explained by the discrepancy between the exercise modality (swimming with a large upper extremity component) and the testing modality (ergometer cycling with a dominant lower extremity component). Another consideration is the fact that hypercapnia has been found to promote adipogenesis (Kikuchi et al., 2017), possibly negating the fat-reducing effects of moderate-intensity exercise as adopted in the present study. A lack of significant differences can also be explained by the differences in GLY uptake by skeletal muscle during exercise (van Hall, 2002). Considering the above, it is also probable that the lack of difference in post-intervention cardiorespiratory (VO₂, CO₂, VE, RER and HR) and metabolic (BMI, FM, FFM, and GLY) measures can be explained by an inadequate exercise stimulus and not ARDS. The sample population was composed of young recreational swimmers with above average CRF (Kaminsky et al., 2015). The adopted training volume may have been insufficient to induce a quantifiable response in CRF or promote a substrate shift towards increased lipid utilization despite the use of an exercise intensity (60% VO₂max) that had been previously used to induce hypercapnia (Graham et al., 1982; Smołka et al, 2014).

Regarding the lack of differences in the present study, several limitations should be considered. First, the literature strongly supports sports-specific fitness testing (Fernandes and Vilas-Boas, 2012). A swimming ergometer or swimming flume would have better replicated the swimming task and enhanced the validity of CRF testing (Demarie et al., 2001). However, there is evidence that there are no significant differences in peak VO₂ between simulated swimming on an incline bench, tethered swimming, and bicycle ergometry, in which both land- and water-based testing are a reliable and valid method of assessing VO₂max in swimmers (Kimura et al. 1990). Nonetheless, exercise testing similar to the exercise modality would have enhanced our findings.

A second issue was that pulmonary function was not assessed. Ventilation may have been modulated with the incorporation of ARDS and influenced the results. Future studies should consider the inclusion of spirometry testing in order to better quantify and interpret the effects of ARDS. Third, the diet of the participants was not strictly controlled nor was energy expenditure estimated. Fourth, previous studies on ARDS in swimmers applied an 8-week training program (Adam et al., 2015). However, we employed a 6-week intervention to determine if a shorter duration protocol could induce similar effects. It is possible that 8 weeks is the minimum timeframe required to fully promote the beneficial adaptations of ARDS.

Finally, a relatively small, homogeneous sample of health young adults was used. Future research should involve an untrained sedentary or overweight/obese population to further investigate training with ARD including modulated exercise frequency, exercise intensity, and tube volume. However, if future works consider modifying the exercise protocol, it needs to be mentioned that several participants in the experimental group reported headaches during training. This may be due to the dilation of cranial blood vessels and may have limited exercise performance. Furthermore, increasing the volume of the snorkel/tubing may induce hyperventilation, reducing pCO₂ and therefore minimizing the post-training effects of respiratory acidosis on RER and VCO₂ (Hollidge-Horvat et al., 1999). This could negate the promoted enhancement of cardiorespiratory and metabolic adaptations by ARDS. As many questions pertaining to ARDS remain unanswered, further research appears necessary. The present study may serve as a baseline and provide a theoretical and methodological grounding for future research to provide more objective data on the effects of ARDS on cardiorespiratory and metabolic variables in recreational swimmers.

A 6-week swimming intervention with ARDS performed at a moderate intensity corresponding to 60% VO₂max did not improve CRF (VO₂, VCO₂, VE, HR) measured in an IET. While post-intervention RER measured at maximum workload increased, this does not suggest a substrate shift towards increased lipid utilization during exercise. No difference was also observed in ∆GLY, suggesting a similar level of hydrolyzed triglyceride in the last swim session when compared with the first that was confirmed by a lack of change in adipose tissue (FM and BMI). Despite various limitations, the present study can serve as a guide for future research on the implications of ARDS-enhanced exercise.