Twenty-four well-trained male kayakers (mean age=27 ± 4 years; height = 180 ± 2 cm; mass = 83 ± 5 kg; body fat = 9.4 ± 1.3%; years of experience = 11 ± 3 years) provided their written informed consent and volunteered to participate. All participants were members of the Iran national kayak sprint team and 18 were medalists of the Asian championships. Following screening for the presence of any unknown disease or conditions putting them at risk of adverse response to high-intensity exercise, participants were randomly assigned to paddling based HIIT (PHIIT), resistance-type HIIT (RHIIT), or a control group (CON). All procedures were in accordance with ethical principles of the Declaration of Helsinki and approved by the institutional review board of the University of Guilan and ethical committee of Sport Sciences Research Institute of Iran (approval ID: IR.SSRC.REC.1400.019).

In order to become oriented with all devices, testing procedures, and training protocols, all participants performed some familiarization visits to the laboratory at least 3 days before baseline measurement. Pre-testing of cardiorespiratory fitness and anaerobic power, along with cardiac structure and hemodynamic parameters as well as blood and biochemical parameters was conducted before the beginning of pre-season training, with post-testing held immediately after completion of the exercise programs. Prior to the beginning of training period, participants completed 3 sessions of a 30-min paddling time trial on a kayak ergometer (Dansprint, Hvidovre, Denmark) at a self-selected pace and an incremental paddling test to volitional fatigue. Participants also performed two upper body 30-s Wingate tests on a separate day to become familiar with these performance tests.

In the pre- and post-training, participants completed an incremental exercise test to determine peak oxygen uptake (V.O_2peak) and related physiological variables. Time for which vV.O_2peak can be maintained (Tmax), upper-body Wingate test, one repetition maximum (1RM) in one-armed cable row, and 500 and 1,000-m on-water paddling performances were also evaluated on separate days. Cardiac structure and hemodynamics were evaluated before and after training. Also, body composition was determined using bioelectrical impedance analysis (Inbody 520, South Korea). All the aforementioned tests were completed in the morning, with 24 h of recovery separating each test. Figure 1 shows a schematic of the sequence of methods and order of the tests used in the present study.

Using a breath-by-breath gas collection system (Cosmed K4B2, Rome, Italy), participants performed an incremental paddling test on a kayak ergometer to determine V.O_2peak, vV.O_2peak, maximal ventilation (v._E@V.O_2peak), respiratory frequency (R_f@V.O_2peak), and tidal volume (V._T@V.O_2peak). The test began at 6 km⋅h^–1 and workload increased 1 km⋅h^–1 every 1 min until volitional exhaustion (Sheykhlouvand et al., 2016a,b, 2018a,b). The drag of the ergometer was adjusted according to the kayakers’ body mass as recommended by the producer to simulate on-water drag during paddling. When three or more of the following criteria were met, participants were considered to have reached their V.O_2peak: (I) the V.O₂ ceased to increase linearly despite the increase in workload and approached a plateau or decreased slightly with the last two values within ± 2 ml^–1⋅kg^–1⋅min^–1; (II) a respiratory exchange ratio reached to 1.1 (Sheykhlouvand et al., 2016a,b, 2018a,b). The vV.O_2peak (or Vmax) was defined as the minimal speed at which the athlete was paddling when V.O_2peak occurred.

The hemodynamic function was evaluated using a transthoracic electrical impedance cardiograph device (PhysioFlow, Manatec, France). This method has been validated and described by previous researches and is known as a reliable method at rest and exercise up to V.O_2max (Charloux et al., 2000; Richard et al., 2001). Two electrodes were placed on the neck, two at the xiphoid sternum, and one on each side of the chest as recommended by the manufacturer. Once a 20-s calibration was completed, hemodynamic values were recorded. At termination of incremental exercise, maximal values for HR (HR_max), stroke volume (SV_max), cardiac output (Q._max), end-diastolic volume (EDV), end-systolic volume (ESV), and ejection fraction (EF) were obtained.

Using a mechanically braked arm ergometer (891E; Monark, Vansbro, Sweden), participants completed a 30-s all-out effort to determine peak power output (PPO) and average power output. Participants were instructed to crank against the internal resistance of the ergometer as fast as possible and a load equivalent to 0.075 kg⋅kg^–1 body mass (Forbes et al., 2014) was applied instantaneously (within 3 s). Participants were verbally encouraged to crank as fast as possible throughout the test. PPO and average power output were calculated using the devise software.

After a warm-up comprising 5-min of paddling at 50% vV.O_2peak, 5 min stretching, and another 5 min of paddling at 60% vV.O_2peak, the paddling speed was increased to vV.O_2peak and participants exercised on kayak ergometer until exhaustion. The test was terminated volitionally by the subject if the desired speed could not be maintained and the time maintained at vV.O_2peak (Tmax) was recorded. Participants were verbally encouraged to paddle as long as possible. In the post-training, this test was repeated following identical procedures.

1RM testing started with a warm-up consisting of 5 min of paddling on an ergometer (Dansprint, Hvidovre, Copenhagen, Denmark) at 6 km⋅h^–1 followed by upper-body joint mobilization exercises. Following a 2-min recovery, participants performed one set of 6 repetitions with a load equal to 60% estimated 1RM followed by one set of 2–3 repetitions at 80% estimated 1RM (Seated pulley cable row machine, Technogym, Italy). Thereafter, to control the work rate accurately and simplify the matching equation that will be explained in section “Exercise Training Protocols,” the movement distance of the one-armed cable row was set at 1 meter (distance between the beginning and the end of motion) and participants performed 3–5 one-repetition sets only with one arm individually and with a 4-min recovery between sets to determine 1RM. The heaviest load that each subject could properly lift in a 1-m motion distance was considered to represent his 1RM (Ualí et al., 2012).

Using a Vivid E95 Ultra edition machine (GE Healthcare, Chicago, Illinois, United States) and according to the recent guidelines (Lang et al., 2015), transthoracic echocardiography was performed at rest and in a left lateral decubitus position. HR was continuously measured by a single-lead electrocardiogram. In the two-dimensional view, structural parameters were recorded with linear internal measurements of the left ventricle (LV) acquired in the parasternal long-axis view. Stroke volume, interventricular septal wall thickness (IVSWT), left ventricle mass (LVM), and left ventricle end-systolic and end-diastolic diameters (LVESd and LVEDd) were recorded. All echocardiographic studies were reviewed by the same cardiologist blinded to group allocation.

Participants were sampled in the morning after an overnight fast exceeding 8 h. A 10-ml venous blood sample was collected in the pre- and post-training by venipuncture from an antecubital vein in the morning after an overnight fast exceeding 8 h. Seven milliliters of blood was immediately spun at 3,000 rpm for 15 min at 4°C and separated and stored at −80°C for measuring total testosterone and cortisol. Serum concentrations of total testosterone [Cayman Chemical, Ann Arbor, Michigan, United States; intra-assay coefficient of variation (CV) = 4.4%] and cortisol (Cayman Chemical, Ann Arbor, Michigan, United States; intra-assay CV = 6.7%) were determined by ELISA kits. Also, using an automated cell counter (Abacus C; Diatron, Budapest, Hungary), the remaining 3-ml blood sample was measured to record complete blood count. Also, plasma volume change was calculated using following equation (Rotstein et al., 1982):

Hb_pre = hemoglobin concentration before the exercise,

Hb_post = hemoglobin concentration after the exercise,

Hct_pre = hematocrit value before the exercise (%),

Hct_post = hematocrit value after the exercise (%).

In the pre- and post-training, participants completed 500- and 1,000-m on-water paddling tests over three consecutive days. The day first was dedicated to the 500-m test; participants then rest during the second day, and they completed the 1,000-m test on the third day. Prior to testing, they completed a standardized warm-up according to Borges et al. (2015). Each participant performed two trials of 500-m test and two trials of 1,000-m test interspersed with 1 h of passive recovery. Time was recorded using two synchronized stopwatches (Interval 2,000 XC Track and Field watch, Nielsen-Kellerman, Delaware, Pennsylvania, United States) and the best times were used for analysis. The tests were performed on flat water with an average tail wind of ∼3.2 m⋅s^–1 at an ambient temperature of ∼23°C. The condition was almost the same during pre- and post-training.

Approximately 48 h after the last baseline measurement, participants underwent 8-weeks of kayak ergometer training, on-water paddling program, or one-armed cable row. Participants undergoing HIIT (PHIIT and RHIIT) performed three HIIT sessions and three traditional on-water paddling sessions each week. In PHIIT, the subjects performed six intervals at 100% vV.O_2peak with training volume varying each week (60, 70, 75, 75, 75, 75, 70, and 60)%Tmax from first to eighth week, respectively, using a 1:1 work to recovery ratio. Traditional on-water paddling sessions consisted of 60 min of paddling at 70–80% HRmax (55–75% V.O_2max; Haff and Triplett, 2016).

The participants performing RHIIT completed one-armed cable row training matched with PHIIT with respect to total work and training duration. As the training mode was one-armed, the hands were alternated during efforts. For matching the total work, work rate at 100% vV.O_2peak [Watts (W)] was recorded from the kayak ergometer. Each W is equal to 1 Joule⋅s^–1 where each Joule is a result of force [Newton (N)] multiplied by distance [meter (M)] leaving:

Considering time commitment of %Tmax (sec) leads to the following equation:

By multiplying work (Newton-force Meter) by 0.10197, we can convert it to Kilogram-force Meter and the equation would be:

Subsequently, 1RM of one-armed cable row in a 1-m motion distance was evaluated and target 1RM [%1RM (kg)] was identified. The PHIIT regimen requires a high-volume training with respect to the time (%Tmax), the RHIIT and PHIIT were performed in the strength endurance phase of the kayakers’ yearly training program where the intensity of repetitions is low to moderate (50–75% 1RM) and the volume is high (Haff and Triplett, 2016). Then, the force (kg) that must be carried in 1-m distance divided by target 1RM [50% 1RM (kg)] and the number of repetitions in one-armed cable row in 1-m distance was specified. On this occasion, as the distance of motion in one-armed cable row is 1 m, the value of force (kg) and work [force (kg) multiplied by distance (1)] would be the same and the number of repetitions will be as follow:

Considering that each athlete had his own 1RM and Tmax, the number of repetitions was specialized and varied among the participants.

With such a matching method, the total work performed during PHIIT and RHIIT would be the same. The key point is the difference between imposed force by each stroke (during paddling HIIT) and each repetition (during resistance HIIT) and the total number of strokes or reps during the working time (Tmax). The difference is that the number of strokes during paddling HIIT is more than the number of repetitions during RHIIT but the imposed force in each repetition in RHIIT is greater than that of each paddling stroke in PHIIT.

The participants in CON performed six sessions of on-water kayak paddling per week including 60 min of traditional endurance paddling at 70–80% HR_max. Also, all three groups performed 1 d/wk of Fartlek training [45 min of long slow distance run (LSD)] and 2 sessions per week of weight training consisting of 3 sets of 8–12 repetitions at 70% 1RM including bench pull, bench press, seated row, bicep curl, military press, pulley pushdowns and trunk rotation) and push-ups, sit-ups, and pull-ups.

To avoid potential confounding of the results mediated by taking supplements and stimulants, participants were directed to continue the same habitual nutrition intake during the experiment. Consuming the same diet 48 h prior to and post-training assessment was encouraged. In addition, subjects were asked to refrain from participating in vigorous activity and to avoid the consumption of caffeinated food and beverages in the 24-h period prior to testing.

Sample size for three groups (N = 8) was calculated using G*Power software (Faul et al., 2007) and Statistical analyses were performed using SPSS, version 25.0 (Statistical Package for Social Science, Chicago, IL). Results were expressed as mean ± SD. The Shapiro-Wilk test was used to test the normality and Levene’s test was used to assess homogeneity of variances. The data were analyzed using a two-factor mixed analysis of variance (ANOVA) with the between factor “group” (PHIIT, RHIIT, and CON) and repeated factor “time” (pre-training, post-training). Significant interactions or main effects were subsequently analyzed using a Tukey’s honestly significant difference post-hoc test. One-way ANOVA was used to analyze difference between changes in plasma volume in different groups. Pearson product–moment correlations were used to examine relationships between variables. Effect size was calculated using Cohen’s d (d). Alpha level was set at 0.05.

No pre-training difference was observed between groups for these physiological parameters. After the 8-week training program, a significant time-regimen interaction (p = 0.04) was found in V.O_2peak. As shown in Figure 2A, the change in V.O_2peak (ml⋅kg^–1⋅min^–1) in response to PHIIT was significantly greater compared to CON (p = 0.03, d = 2.02). V.O_2peak was significantly increased in PHIIT (Post: 54.35 ± 3.65 vs. Pre: 48.39 ± 3.91 ml⋅kg^–1⋅min^–1,%Δ = 12.3, p = 0.0006, d = 1.6), RHIIT (Post: 52.31 ± 4.97 vs. Pre: 47.59 ± 4.48 ml⋅kg^–1⋅min^–1, %Δ = 9.1, p = 0.002, d = 1.0), and CON (Post: 48.31 ± 2.13 vs. Pre: 45.95 ± 2.73 ml⋅kg^–1⋅min^–1, %Δ = 5.1, p = 0.003, d = 0.9) compared with pre-training.

After the 8-week training period, vV.O_2peak significantly increased in PHIIT (Post: 18.2 ± 0.5 vs. Pre: 17.2 ± 0.9 km⋅h^–1, %Δ = 5.7, p = 0.001, d = 1.2), RHIIT (Post: 17.8 ± 0.5 vs. Pre: 16.9 ± 0.6 km⋅h^–1, %Δ = 5.3, p = 0.0004, d = 1.5), and CON (Post: 17.7 ± 0.4 vs. Pre: 16.8 ± 0.4 km⋅h^–1, %Δ = 5.5, p = 0.006, d = 2.5) compared with pre-training, but there was no time-regimen interaction (p = 0.91) (Figure 2B).

Maximal V._E significantly increased from pre- to post-training in PHIIT and RHIIT (p = 0.003, d = 1.4 and 1.2, respectively) but not in CON and there was no between-group difference for this variable (p = 0.65). In addition, a significant increase occurred in R_f@V.O_2peak from pre- to post-training in PHIIT and RHIIT (p = 0.001 and 0.004, d = 2.0 and 0.8, respectively) but not CON. Also, the change in R_f@V.O_2peak in response to RHIIT was significantly greater compared to the change in CON (p = 0.03, d = 1.2). There was no change in V._T@V.O_2peak across time (p > 0.05).

There were no between-group differences for PPO or average PO (p = 0.64 and 0.25, respectively). PPO significantly increased after training in PHIIT, RHIIT, and CON (p = 0.0003, 0.003, and 0.002, d = 1.1, 1.0, and 0.3, respectively). Average PO was significantly increased in response to PHIIT (p = 0.001, d = 0.7) and RHIIT (p = 0.05, d = 0.5) compared with pre-training but not CON (Table 1).

Maximum strength expressed as 1RM in one-armed cable row significantly improved over time for right and left hand in response to RHIIT (p = 00008, 0.00001; d = 1.7, 1.6, respectively). No training-induced increase in this variable was observed in PHIIT and CON with no between-group difference for the magnitude of changes pre- to post-training.

There was no pre-training difference (p > 0.05) observed between groups for any biochemical variables. Table 2 presents the resting hormone concentrations and hematological changes in response to the 8-week training period. Compared to pre-training, a significant increase was observed in total testosterone concentration in PHIIT and RHIIT (p = 0.02 and 0.03, d = 1.0 and 0.7, respectively). No significant differences were observed among groups (p = 0.35). There was a training-induced increase in Testosterone/Cortisol ratio (T/C ratio) in PHIIT and RHIIT (p = 0.05, d = 0.8) but not in CON (p > 0.05). Also, there was no between-group difference for T/C ratio (p = 0.76). Results showed no change in cortisol, hemoglobin (Hb), red blood cell (RBC), or hematocrit (Hct) in response to training (p > 0.05). Plasma volume changes over time were 4.7, 1.1, and –2.1% in PHIIT, RHIIT, and CON groups, respectively (Figure 3). No between-group difference was found for change in plasma volume (p = 0.13).

No pre-training difference was observed between groups for cardiac morphology or hemodynamics. As indicated in Table 3, SV_max significantly increased after training in response to PHIIT (p = 0.006, d = 1.0) and RHIIT (p = 0.003, d = 1.1) but there was no change in CON (p = 0.23). Q._max showed a significant main effect for time in PHIIT (p = 0.006, d = 1.2), RHIIT (p = 0.001, d = 1.7), and CON (p = 0.05, d = 0.5), but there was no time-regimen interaction for Q._max (p = 0.41). EDV and EF significantly increased from pre- to post-training in PHIIT (p = 0.004 and 0.05, d = 1.2 and 0.9, respectively) and RHIIT (p = 0.002 and 0.03, d = 1.3 and 0.5, respectively) groups, but not in CON. Data showed no change in ESV or HR_max across time in groups.

Resting values of SV and LVESd were significantly increased across time in PHIIT (p = 0.01 and 0.01, d = 0.7 and 0.5, respectively), RHIIT (p = 0.01 and 0.01, d = 0.9 and 0.7, respectively), and CON (p = 0.005 and 0.01, d = 0.6 and 0.5, respectively). No significant difference was observed for resting values of IVSWT, LVM, and left LVEDd compared with pre-training (p > 0.05). No between group difference was observed in the change in cardiac dimensions over time (Table 3).

No pre-training difference (p > 0.05) was observed between groups for 500- and 1,000-m paddling performance. Figures 4A,B show changes in 500- and 1,000-m kayak sprint performance from pre- to post-training. A significant time-regimen interaction (p = 0.05) was found in 500-m paddling performance as the change in 500-m paddling performance in response to PHIIT and RHIIT was significantly greater compared to the change in CON (p = 0.02, d = 0.6; and p = 0.05, d = 0.4, respectively). In response to training, 500-m paddling time significantly decreased in PHIIT (Post: 109.1 ± 5.1 vs. Pre: 112.5 ± 4.1 s, Δ≈ –3.4 s, p = 0.0008, d = 0.7), RHIIT (Post: 110.3 ± 4.9 vs. Pre: 114.0 ± 3.9 s, Δ≈ –3.7 s, p = 0.0006, d = 0.8), and CON (Post: 114.7 ± 2.1 vs. Pre: 116.6 ± 2.3 s, Δ≈ –1.9 s, p = 0.02, d = 0.8).

As shown in Figure 4B, a significant time-regimen interaction (p = 0.04) was found in 1,000-m paddling performance. The change in 1,000-m paddling performance in response to PHIIT was significantly greater compared to the change in CON (p = 0.05, d = 0.9). The 1,000-m paddling time was significantly decreased in PHIIT (Post: 227.1 ± 8.6 vs. Pre: 231.0 ± 7.4 s, Δ≈ –3.9 s, p = 0.0001, d = 0.5) and RHIIT (Post: 231.6 ± 7.8 vs. Pre: 233.7 ± 6.9 s, Δ≈ –2.1 s, p = 0.0004, d = 0.6) but not in CON (Post: 235.1 ± 2.2 vs. Pre: 236.0 ± 3.9 s, Δ≈ –0.9 s, p = 0.17) compared with pre-training.

Also, 500- and 1,000-m paddling performances were negatively correlated to V.O_2peak in pre- and post-training (r = –0.92, p = 0.00001; r = –0.91, p = 0.000001, respectively) and post-training (r = –0.91, p = 0.00001; r = –0.81, p = 0.000001, respectively).