Contribution of different strength determinants on distinct phases of Olympic rowing performance in adolescent athletes

ABSTRACT

Aerobic metabolism dominates Olympic rowing, but research on the relative contribution of strength and power demands is limited. This study aimed to identify the contribution of different strength determinants for distinct phases of rowing ergometer performance. The cross-sectional analysis comprised of 14 rowing athletes (4 female, 10 male, age: 18.8 ± 3.0y, 16.9 ± 2.2y). Measurements included anthropometrics, maximal strength of leg press, trunk extension and flexion, mid-thigh pull (MTP) and handgrip strength, VO₂max, and a 2000 m time trial, where peak forces at the start, middle and end phase were assessed. Additionally, rate of force development (RFD) was assessed during the isometric leg press and MTP with intervals of 150, 350 ms and 150, 300 ms, respectively. Stepwise regression models for ergometer performance showed that the start phase was mainly explained by maximal trunk extension and RFD 300 ms of MTP (R²= 0.91, p < 0.001) and the middle section by VO₂max, maximal leg press strength and sitting height (R²= 0.84, p < 0.001). For the end phase, a best fit was observed for trunk flexion, RFD 350 ms of leg press, body height and sex (R²= 0.97 p < 0.001), whereas absolute VO₂max, trunk flexion and sex explained variance over the entire 2000 m time trial (R²= 0.98, p < 0.001). It appears that for the high acceleration in the start phase, force transmission through maximum strength for trunk extension is essential, while fast power production along the kinetic chain is also relevant. Additionally, the results support that maximal force complements the reliance on VO₂max. Further intervention studies are needed to refine training recommendations.

Highlights

• This study highlights the importance of the neuromuscular system for rowing performance, which has to be considered in addition to the well-known factors VO₂max, anthropometry and sex.

• For overall 2000 m rowing performance, maximal leg strength appears to complement the reliance on VO₂max for maintenance of force production.

• For efficient force transmission in the start phase, maximal strength of trunk extension and a fast power production along the kinetic chain of legs, trunk and arms are essential, while towards the end of the race isometric trunk flexion and rate of force development of leg press appear to be important.

KEYWORDS:

• strength
• performance
• training
• competition
• endurance

Introduction

Rowing performance is characterised by a well-developed anaerobic and aerobic capacity but also muscle strength (Maestu et al., 2005; Ingham et al., 2002). Race times over the Olympic distance of 2000 m range from 5.5 to 7 min in elite and adolescent rowers, highlighting the complexity of metabolic and neuromuscular demands (Maestu et al., 2005; Steinacker, 1993; Russell et al., 1998). The energy distribution of the race has been estimated as 67% aerobic and 33% anaerobic, of which 21% are lactic and 12% alactic (Roth, 1983). Due to a relatively low duty cycle (around 32–38 strokes/min), contraction velocities of muscles are rather slow, and oarsmen are characterised by a high proportion of slow-twitch fibers of 70%–85% (Steinacker, 1993). With such a fiber type distribution, it is reasonable that strong correlations between maximum aerobic oxygen consumption (VO₂max) and rowing performance have been shown (Ingham et al., 2002; Cosgrove et al., 1999; Cerasola et al., 2020; Yoshiga & Higuchi, 2003; Nevill et al., 2011). As typically 210–230 powerful rowing strokes are performed per race, rowing is considered as power-endurance sport, where leg, trunk and arm muscle sequentially contribute to force propulsion and transmission (Maestu et al., 2005; Steinacker, 1993; Nugent et al., 2020).

Compared to the metabolic requirements of rowing performance in elite and adolescent athletes, considerably less is known about the role of distinct neuromuscular determinants. While it was shown that the linear oar velocity is dependent on the angular velocities of the lower leg, trunk, upper leg as well as lower and upper arm (Lamb, 1989), some studies showed a strong association between rowing ergometer performance and strength measurements, such as a multiple repetition maximum or a maximal repetition test (e.g. in leg press and bench pull) (Ingham et al., 2002; Cerasola et al., 2020; Nevill et al., 2011; Akça, 2014; Riechman et al., 2001; Huang et al., 2007; Jürimäe et al., 2010; Lawton et al., 2013). Furthermore, the role of various physiological, morphological, and anthropometric variables in predicting rowing ergometer performance was assessed using regression models, where the results suggest a combination of aerobic and anaerobic capacities, as well as large body dimensions and muscle volume of the vastus lateralis muscle to contribute significantly (Ingham et al., 2002; Cerasola et al., 2020; Nevill et al., 2011; Akça, 2014; Riechman et al., 2001; Huang et al., 2007; Lawton et al., 2013; van der Zwaard et al., 2018).

The importance of neuromuscular determinants is also indirectly underlined by interventional studies, comparing the effects of different strength training regimens on rowing performance. A recent study indicated beneficial effects of heavy-load over light-load resistance training to enhance rowing performance in successful collegial oarsmen (Gallagher et al., 2010), that is likely induced by improved maximal strength and consequently a lower relative strength requirement per stroke (Støren et al., 2008). However, in most intervention studies, strength training was performed alongside endurance training without a control group, leaving it unknown as to whether observed performance improvements are directly related to changes in neuromuscular performance (Maestu et al., 2005; Lawton et al., 2011). In this context, it was shown that the inclusion of heavy-load resistance training into the general and competitive preparation period of female rowing athletes led to increased rowing performance without muscle hypertrophy or significant changes in muscle architecture (van der Zwaard et al., 2021). However, as no control group was included in this study and an in-depth neuromuscular analysis was not performed the role of strength training for the observed performance improvements remains unknown.

In order to provide more precise training recommendations and overcome the gap between strength goals in adolescent and elite athletes to ensure a consistent physical development (McNeely et al., 2005), it is necessary to assess the relative contribution of distinct neuromuscular determinants. In fact, rowing performance is characterised by a unique pacing strategy, which typically involves an initial spurt, followed by a gradual decrease until the third quarter and finally an increase in the average speed until the end of the 2000 m race (Secher, 1993; Garland, 2005; Hagerman, 1984). It is reasonable to assume that high levels of maximal strength and rate of force development (RFD) are necessary to accelerate the boat against water resistance during the start phase, and high force production needs to be maintained throughout the race, despite the metabolic acidosis occurring over the Olympic race distance (Nugent et al., 2020; Treff et al., 2021). However, due to a lack of scientific information concerning the contribution of distinct strength determinants (i.e. RFD versus isometric and isokinetic maximal strength) to different phases of Olympic rowing performance, this remains speculative.

To further expand on previous research providing first hints for associations of rowing performance and neuromuscular factors, we aimed to examine the relative contributions of maximal isometric and isokinetic strength as well as different intervals of RFD to rowing ergometer performance in adolescent athletes. Moreover, we aimed to assess whether the identified strength determinants differentially affect the start, middle and end phases of a simulated 2000 m race.

Methods

Subjects

Fourteen athletes of the local rowing club volunteered to participate in this study [4 females: age 18.8 ± 3.0 years, height 168.1 ± 4.5 cm, body mass 65.6 ± 5.5 kg; and 10 males: 16.9 ± 2.2 years, height 181.4 ± 4.8 cm, body mass 76.7 ± 9.2 kg (mean ± SD). According to Mirwald et al. (2002), the age at peak height velocity was calculated as 14.2 ± 1.7 years for males and 12.7 ± 0.6 years for females, respectively. The athletes were classified as “Highly Trained/National Level” (Tier 3) (McKay et al., 2021). The participants were not divided in light- or heavyweight categories, as the Swiss rowing association does not distinguish between them up to the U23 category. Additionally, the athletes are trained in both scull and sweep rowing and have not been committed to one technique. The participants had at least three years of rowing and one year of strength training experience and were unafflicted by acute or chronic diseases or injuries. All procedures were carried out with the written informed consent of the participants. For athletes under the age of 18, written informed consent was obtained from both the adolescents and parents. The study was conducted according to the declaration of Helsinki and approved by the ethics committee of the German sport university.

Experimental protocol

To investigate different strength determinants in young rowers, we conducted a battery of strength tests, as well as a maximum ergometer test over 2000 m. The measurements took place in the rowers’ preparation phase, where the polarised endurance training with 90% high-volume, low-intensity and 10% low-volume, high-intensity training (9.0 ± 0.3 h per week) was accompanied by two sessions of strength-endurance training per week. The measurements were divided into a first day with anthropometric testing, the strength battery, and 2000 m time trial; and a second day with a ramp-test to determine VO₂max. The athletes were instructed to standardise the nutritional intake with the last high-carbohydrate meal taken 2 h prior to the tests. Furthermore, the participants were advised to avoid intensive exercise before 48 h of the test days, as well as physical training sessions 24 h before the test. To familiarise the participants with the measurements, an identical test was carried out in advance.

Strength determinants

First, anthropometric data such as body and sitting height, body mass and composition were measured. Strength abilities were determined after a ten-minute warm-up at self-paced, moderate intensity (151 ± 38 W) on the rowing ergometer (Concept2 Deutschland GmbH, Hamburg, Germany). Maximal strength was measured in the order of bilateral, isokinetic and isometric leg press, trunk extension and flexion, followed by isometric mid-thigh pull and grip force. RFDs were assessed from the force curves for isometric leg press and mid-thigh pull.

Body composition was estimated with an Inbody 720 (Biospace Co., Ltd., Seoul, Korea). The leg press, trunk extension and flexion were measured on IsoMed 2000 and IsoMed 2000 Back Module, respectively (D. & R. Ferstl GmbH, Hemau, Germany). These instruments provide good to excellent reliability in measuring leg and trunk strength (Roth et al., 2017; Dirnberger et al., 2013). The mid-thigh pull and hand grip force are measured with the force plate of Leonardo Mechanograph® (Novotec Medical GmbH, Pforzheim, Germany), which were reported to be highly reliable for assessing athletes (Rhodes et al., 2022).

All isometric measures were repeated three times with 30 s recovery between subsequent trials. According to Maffiuletti et al. (2016), the athletes were instructed to “push as fast and hard as possible” in order to focus on rapid increase of force development while also maintaining maximal force during a 3–4 s plateau. The leg press was performed with a knee angle of 110° and the backrest was set at an inclination of 70°, while a hip angle of 85° was used for the trunk extension and flexion. The mid-thigh pull was performed at knee and hip angles of 150° and grip force was measured with the arm extended vertically.

The peak force of the best trial was used for further evaluation. In addition to the standard interval of 150 ms for RFD in isometric leg press and mid-thigh pull, intervals of 350 respectively 300 ms were analysed, as force production is expected to be slower due to the slow duty cycle nature of rowing (Maffiuletti et al., 2016; Schneider, 1980).

Two trials with five repetitions each were executed for isokinetic measures, while a rest period of 60 s was prescribed. The velocity was set at 260 mm/s for leg press, and 60°/s for trunk extension and flexion. Only the highest value of peak torque of trunk strength and peak force of the concentric phase in the leg press were considered for further analysis.

2000 m time trial

The rowing ergometer is a suitable tool for assessing performance and training progress; it is reliable for well-trained rowers and in good agreement with on-water performance (typical error 2%, 95%CI 1.3–3.1%) (Schabort et al., 1999). Prior to the time trial over 2000 m, the athletes warmed up for five minutes at self-paced moderate intensity. The drag factor was set to 135. A calibrated piezoelectric force transducer (Type 9311B, Kistler Instrumente AG, Winterthur, Switzerland) was installed in the pull chain of the rowing ergometer to record and analyse the produced force curves (proEMG 2.0, prophysics AG, Kloten, Switzerland) throughout the entire bout. Due to the pacing nature of rowing, the focus was on the three characteristic phases, namely start, middle and end, and their definition was determined as follows: An analysis of the actual start sprint revealed the peak forces of the first five strokes differ from the second five strokes (MD: −0.137 kN, p < 0.001), so the start phase was refined as the peak force of the first five strokes in the following analyses. The peak force of ten strokes at half time of the trial was defined as middle phase, as it represents the 500–1500 m section of the whole race (MD: 0.010 kN, p = 0.236). The end phase was defined as the peak forces of the last 10 strokes, representative of the last 500 m (MD: 0.032 kN, p = 0.134), but also taking into account the gradual increase in force as well as individual differences towards the end. To ensure the athletes were working to their maximum capacity, they were asked for their subjective ratings of exertion before and after the measurements using the Borg scale (6–20 RPE) (Borg, 1998). Verbal encouragement was provided throughout the trial.

VO₂max

Following a 48-hour window for recovery, VO₂max was determined using a ramp protocol on the rowing ergometer with an increment of 20 watts per minute. The initial load was set to reach the maximum exhaustion between 8 and 12 min and calculated for each athlete by the expected power as the sum of the mean power output during 2000 m rowing and 20 watts. If more than five strokes did not meet the required wattage, two warnings were issued before the test was ended. Respiration was measured continuously breath-by-breath with a reliable gas analyser (Metamax® 3B, Cortex Biophysik GmbH, Leipzig, Germany) (Macfarlane & Wong, 2012). Athletes were encouraged to perform at their best to achieve maximal exhaustion. Additionally, the secondary exhaustion criteria maximal heartrate (HR_max≥ 95% of age-predicted heart rate ( = 220 – age)), respiratory exchange ratio (RER_max≥ 1.10) and reported perception of exhaustion by Borg (RPE_max≥ 19) (Borg, 1998) were considered (Wagner et al., 2020). If the criteria were reached, VO₂max was defined as the maximum of the 30 s moving average oxygen consumption.

Statistics

Descriptive statistic methods were used to present mean and standard deviation (mean ± SD), and 95% confidence interval (CI). The interclass correlation coefficients between the familiarization and main measurement days are given as mean value for each outcome and classified according to Koo & Li, 2016. The normal distribution of numeric variables was assessed using the Shapiro–Wilk test, and in case of significance, further verification was performed using Q-Q-plots. All of the variables were deemed normally distributed. Homoscedasticity was assessed visually with Tukey-Anscombe plots. Based on Pearson’s correlation coefficient r presented in Table 1, all correlated variables were included in four stepwise multiple regression models for the peak forces of the start, middle and end phases as well as the full 2000 m time trial, while sex was inserted as standard. The Akaike information criterion (AIC) was considered for the selection of variables; the root mean squared error (RMSE) returned the accuracy of the models. To avoid multicollinearity, the variance inflation factor was kept < 10. For generalizability, the models were tested using leave-one-out cross-validation. The level for statistical significance was set at α ≤ 0.05. Analyses were made using Jamovi (version 1.8.1.0); graphs are drawn with R and RStudio (version 4.2, package: graphics).

Table 1. Descriptive statistic presented as mean ± SD, 95% confidence interval (CI) and Pearson correlation (r) between anthropometric characteristics, VO₂max, maximal strength variables, RFD’s and ergometer performance (2000 m, start, middle and end phase). Intraclass correlation coefficients are given as mean of each outcome. The 2000 m ergometer performance was measured as time [s], while peak forces of the first 5, middle 10 and final 10 strokes were defined as start, middle and end phase. (N = 14; 10 male, 4 female).

	mean ± SD	CI	r (2000 m)	r (start)	r (middle)	r (end)	ICC
Age [y]	17.4 ± 2.6	(15.9; 19.0)	−.130	.111	.319	−.010
Body height [cm]	177.6 ± 7.6	(173; 182)	−.867**	.836**	.750**	.885**	1.00
Sitting height [cm]	92.3 ± 4.3	(90; 95)	−.837**	.699**	.804**	.716**	.98
Body mass [kg]	73.6 ± 9.7	(67.7; 79.4)	−.711**	.810**	.586	.847**	.99
Fat mass [kg]	11.9 ± 4.6	(9.2; 14.7)	.370	−.108	−.393	−.092	.97
Skeletal muscle mass [kg]	34.8 ± 5.8	(31.3; 38.3)	−.898**	.866**	.785**	.914**	.98
Relative VO2max [ml·min^−1·kg^−1]	57.9 ± 7.1	(53.7; 62.2)	−.654*	.379	.645*	.358	.96
Absolute VO2max [L·min^−1]	4.3 ± 0.8	(3.8; 4.7)	−.972**	.860**	.883**	.856**	.89
Maximal isometric leg press [kN]	4.2 ± 1.3	(3.4; 5.0)	−.847**	.671**	.865**	.894**	.61
Maximal isokinetic leg press [kN]	3.7 ± 8.9	(3.2; 4.3)	−.887**	.817**	.808**	.908**	.95
Maximal isometric trunk flexion [Nm]	113.7 ± 29.5	(1.4; 1.7)	−.912**	.851**	.856**	.939**	.92
Maximal isometric trunk extension [Nm]	258.9 ± 79.2	(212; 306)	−.936**	.918**	.869**	.898**	.91
Maximal isokinetic trunk flexion [Nm]	122.4 ± 28.3	(106; 139)	−.925**	.863**	.848**	.939**	.98
Maximal isokinetic trunk extension [Nm]	268 ± 79.7	(220; 316)	−.914**	.890**	.844**	.880**	.93
Maximal isometric mid-thigh pull [kN]	2.1 ± 0.4	(1.9; 2.4)	−.908**	.835**	.841**	.876**	.81
Handgrip force [kN]	0.45 ± 0.09	(0.39; 0.50)	−.765**	.706**	.618*	.745**	.95
RFD 150ms isometric leg press [N·150ms^−1]	0.99 ± 0.52	(0.68; 1.30)	−.648*	.662*	.677**	.843**	.61
RFD 350ms isometric leg press [N·350ms^−1]	0.75 ± 0.28	(0.58; 0.92)	−.787**	.749**	.804**	.891**	.68
RFD 150 ms isometric mid-thigh pull [N·150ms^−1]	1.14 ± 0.24	(0.99; 1.29)	−.413	.344	.330	.302	.69
RFD 300 ms isometric mid-thigh pull [N·300ms^−1]	1.71 ± 0.27	(1.54; 1.88)	−.812**	.903**	.769**	.894**	.78

*p < 0.05, **p < 0.01; RFD, rate of force development; ICC, intraclass correlation coefficient.

Results

Figure 1 shows the averaged force curves over the rowing strokes for the start, middle and end phase, respectively. The start phase was characterised by a high force in the initial 25% of the rowing stroke with male and female athletes generating 0.611 ± 0.2 kN and 0.472 ± 0.1 kN, respectively. The maximum force was achieved at 46% of the cycle for males (0.951 ± 0.2 kN), and at 43% for females (0.758 ± 0.1 kN). In the middle phase of the 2000 m time trial, both male and female athletes generated the highest force at 47% of the rowing stroke, with values of 0.846 ± 0.1 kN, and 0.684 ± 0.02 kN, respectively. The stroke characteristics of the end phase of the simulated rowing race showed a force maximum of 0.865 ± 0.2 kN at 44% of the rowing stroke for male athletes, and 0.578 ± 0.7 kN at 41% for female athletes, respectively.

Figure 1. Averaged force curves (solid line) [kN] and standard deviations (dotted lines) for the first 5 strokes in the start phase (A), 10 strokes in the middle (B) and end phase (C) as percentages of the rowing stroke cycle separated for female (N = 4) and male athletes (N = 10).

Descriptive statistics for anthropometric characteristics, maximal strength metrics, RFDs and VO₂max as well as the Pearson correlation coefficients for association between these variables and the 2000 m time (434 ± 31 s), the peak force of the start (0.938 ± 0.2 kN), middle (0.762 ± 0.2 kN) and end phase (0.835 ± 0.2 kN) are presented in Table 1. Significant correlations were found between maximal leg press, trunk extension and trunk flexion strength, as well as mid-thigh pull, handgrip force and the rowing performance over the 2000 m time trial (r = −0.765 to −0.936, p < 0.01) as well as the peak force over the different phases of the race (start, middle and end phase) (r = 0.671–0.918, p < 0.05) (see Table 1.). Furthermore, the rowing performance (2000 m time trial, start, middle and end phase) was significantly correlated with RFD of leg press over 150 and 350 ms (r = −0.648–0.891, p < 0.05) (Table 1). The RFD of the mid-thigh pull was only correlated with rowing performance measures over 300 ms (r = 0.769–0.903, p < 0.05) but not 150 ms (r = 0.302 to −0.413, p > 0.05) (Table 1).

The best single predictor of each multiple regression model is displayed in Figure 2. For the full Olympic rowing distance (2000 m), our regression model revealed that, 97.5% of variance was explained by absolute VO₂max, maximal isokinetic trunk flexion and sex (Table 2). During the starting phase, 91.7% of the variance was explained by the maximal isometric trunk extension, RFD 300 ms of mid-thigh pull and sex (Table 2). For the middle phase, 82.7% of the variance was explained by the absolute VO₂max, maximal isometric leg press, sitting height and sex (Table 2). Finally, 96.3% of the variance in the end phase were explained by maximal isometric trunk flexion, RFD 350 ms of leg press, body height and sex (Table 2).

Figure 2. Greatest predictors for the rowing performance (2000 m, start phase, middle phase, end phase). Correlations between start phase [kN] and isometric trunk extension [Nm] (r = 0.918; p < 0.001) (A), middle phase [kN] and VO₂max [L·min⁻¹] (r = 0.883; p < 0.001) (B), end phase [kN] and isometric trunk flexion [Nm] (r = 0.898; p < 0.001) (C) and 2000 m time trial [s] and VO₂max [L·min⁻¹] (r = 0.972; p < 0.001) (D).

Table 2. The regression model equation, estimate, root mean squared error (RMSE), significance level (p), standardized estimate (SE) and adjusted R² for the start, middle, end phase and the total 2000 m time trial.

Phase	Equation	Estimate	RMSE	p	SE	R^2
Start	Intercept	.262		.052
	Isometric trunk extension	.002	.050	.005	.686	.905
	RFD of mid-thigh pull 300 ms	.138	.046	.267	.217	.912
	Sex	.054	.043	.237	.301	.917
Middle	Intercept	−.971		.128
	VO2max	.043	0.076	.518	.203	.760
	Isometric leg press strength	.060	0.065	.043	.495	.806
	Sitting height	.014	0.056	.101	.361	.844
	Sex	−.010	0.056	.858	−.064	.827
End	Intercept	−.046		.935
	Isometric trunk flexion strength	.003	.064	.016	.420	.871
	RFD leg press 350 ms	.264	0.048	.005	.399	.919
	Body height	.002	.037	.631	.076	.946
	Sex	.096	.029	.064	.493	.963
2000 m	Intercept	586.2		<.001
	VO2max	−25.2	7.13	<.001	−.621	.941
	Isokinetic trunk flexion strength	−.3	5.5	.013	−.281	.962
	Sex	−10.4	4.3	.031	−.327	.975

RFD, rate of force development.

Discussion

The aim of this study was to investigate the maximal strength determinants exploited in the start, middle, and end phases of Olympic rowing in adolescent athletes. In addition to VO₂max, anthropometry and sex, our results suggest the relevance of different strength determinants for Olympic rowing: All maximum strength variables and most RFDs are strongly correlated to rowing ergometer performance and its distinct phases. Further, trunk extension strength and RFD contribute to the start phase, and leg strength supplements VO₂max in the middle phase.

Regarding the maximal strength variables, isometric trunk extension showed strong correlations for all phases of rowing ergometer performance and explained most of the variance in the start phase. This is reasonable from a biomechanical perspective, as it is crucial for the rowing stroke to apply the high forces produced in the legs to the oar via the complete muscle chain. As the acceleration of the boat at the start requires high levels of strength, our results suggest that maximal back strength is important for the force transmission in this phase (Baudouin & Hawkins, 2002). Contrarily, Huang et al. (2007) found that trunk extension strength did not correlate with rowing ergometer performance. However, this might be explained by their measurement protocol, which included the maximal number of repetitions, representing a strength endurance domain, whereas in our study maximum strength was measured (Huang et al., 2007).

Besides the trunk extension, the leg press strength belonged to the strongest correlates for the middle phase, which represents a constant, sustained, high force production between the start and end phases. The additional 4.6% of variance in the middle phase model explained by maximal isometric leg press highlight the relevance of maximal force production of the legs to maintain the high power over the course. Our results agree with previous studies, which found leg press (1-RM and 5-RM) to correlate well with rowing ergometer performance (Akça, 2014; Huang et al., 2007; Jürimäe et al., 2010; Lawton et al., 2013; Lawton et al., 2011). As 75%–80% of the power of a rowing stroke comes from the legs, maximum leg strength is important for the sustained middle section of the race (Cosgrove et al., 1999). Underlining these findings, another study demonstrated that the volume of the vastus lateralis muscle largely explains the variance in rowing ergometer performance in male and female athletes. While these findings were controlled for sex and body size differences it can be assumed that the ergometer performance can be enhanced by maximising muscle volume and muscle strength of the vastus lateralis muscle (van der Zwaard et al., 2021).

In contrast to trunk extension and leg strength, the presence of isokinetic trunk flexion in the model of the 2000 m time trial and the end phase may seem inappropriate at first sight, as trunk flexion does not contribute to the primary force production of a rowing stroke. Nevertheless, trunk flexors decelerate the trunk extension in the finish phase and return the trunk into the starting position during the recovery phase of the rowing stroke (Nugent et al., 2020). Further explanations for the significant role of trunk flexion are linked to breathing patterns, as trunk muscles are auxiliary for breathing. In rowing, respiratory muscles are reported to face a dual demand: On one hand, breathing muscles effect ventilatory control in order to meet the high demands of oxygen during rowing; On the other hand, they assist the propulsive force production as rowers hold their breath to stabilise the core at the initial catch phase (Treff et al., 2021). As the breathing rhythm is coupled to the rowing stroke by a ratio of 2:1 (breath:stroke), only little time is available for inspiration during the drive phase, which is even more challenging to meet the respiratory requirements (Secher, 1993; Treff et al., 2021; Mahler et al., 1991). Thus, breathing control through adequate levels of trunk muscles might be important for rowing performance, especially towards the end of the race. This can be linked to Bucher et al. (2018), who examined the change in double pooling performance after an exercise-induced trunk fatigue in well trained cross-country skiers. Their poorer performance was accompanied by lower ventilation and lower VO₂peak, which might be due to negative influence on respiratory muscle function, technique and posture, which all affect conditions for breathing (Bucher et al., 2018). However, there is no data on this topic in rowing yet.

In addition to the maximal strength variables, further variance in the start and end phases was explained by the RFD of 300 ms in mid-thigh pull and 350 ms in leg press, which is underlined by the force curves showing differences in timing and amplitude of applied force between different phases in the race (Figure 1). As visualised, the force curves show a strong initial increase, which differs between the phases of the race and is highest in the start phase. From a biomechanical perspective, the most efficient, force-generating technique is characterised by a powerful first part of the rowing stroke (Schwanitz, 1991). Thus, it was shown that independent of stroke frequency, the time to apply maximal force on the oar seems to be approximately 0.3–0.4 s (Schneider, 1980). Accordingly, it could be assumed, that especially in the starting phases as well as in the spurt to the finish line a high force generation in a short interval is crucial to the stroke characteristics associated with good performance. Since the RFDs of 150 ms were not or weaker correlated to these phases, the optimal interval to raise RFDs in rowing seems to be longer, representing the nature of slow power production in rowing (Steinacker, 1993).

While the role of RFD for rowing has not been investigated yet, its relevance was shown in other endurance sports, such as cycling and running. In these sports, a higher RFD correlated with a better movement economy and therefore had a positive impact on overall endurance performance (Støren et al., 2008; Rønnestad & Mujika, 2014). However, due to the short race duration in rowing, the reliance on rowing economy plays a minor role (Cosgrove et al., 1999). Beyond movement economy, other potential mechanisms suggested could still be relevant for rowers as the change in RFD exceeded the higher 1-RM after heavy resistance training in runners (Støren et al., 2008). A higher RFD indicates a reduced time to reach a given force, which in turn implies longer recovery periods. While blood flow is restricted during muscle contraction, the increased relaxation time might lead to increased circulatory flow, which suggests a better supply of O₂ and substrates to potentially improve peak performance (Støren et al., 2008; Rønnestad & Mujika, 2014). Although this potential mechanism has not been studied thoroughly, improved blood flow could also be beneficial for rowing performance.

In terms of training considerations, one study addressed plyometric training, where enhanced 500 m performances but no changes in rowing economy compared to the control group were found (Egan-Shuttler et al., 2017). So, improved short-time performance through neuromuscular training is supported by our findings, as the neuromuscular variables maximal strength and RFD determined the start phase. Because the start is of great importance for the psychological and tactical advantage due to the backward nature in rowing, this phase should be taken into account in further training intervention studies in which the recommendations for heavy and explosive resistance training are refined (Gallagher et al., 2010; Garland, 2005; Egan-Shuttler et al., 2017).

In addition to the strength determinants, VO₂max, anthropometry and sex were prominent in the models for rowing ergometer performance. Indeed, the models for the 2000 m time trial as well as the middle section are mainly explained through absolute VO₂max. This agrees with earlier research, where absolute but not relative VO₂max showed to have strong performance-predictive value due to the high contribution of aerobic metabolism in Olympic rowing (Ingham et al., 2002; Steinacker, 1993; Cosgrove et al., 1999; Cerasola et al., 2020; Yoshiga & Higuchi, 2003; Nevill et al., 2011; Secher, 1993; Treff et al., 2021). Considering the anthropometric data, a higher body mass, especially higher skeletal muscle mass as well as body and sitting height seem to be beneficial for rowing, while the height variables were even included in the models of middle and end phase. This clearly agrees with the literature, as it is well documented that successful rowers are taller and heavier, since the body weight is carried by the boat on the water (Cosgrove et al., 1999; Cerasola et al., 2020; Yoshiga & Higuchi, 2003; Akça, 2014; Secher, 1993; Hagerman, 1984). Additionally, large body dimensions provide the biomechanical advantage through a longer rowing stroke, which is associated with a higher level of rowing performance (Cosgrove et al., 1999; Cerasola et al., 2020; Akça, 2014).

Additional variance for the start and end phase as well as the total distance of 2000 m time trial, which could not be explained by the previous variables, is related to sex. Yoshiga and Higuchi (2003) examined sex differences in rowing ergometer performance with respect to various characteristics. After normalising for body size, about 10% sex differences remained. However, when matched to fat-free mass and VO₂max, about 4% persisted. In addition to differences in body size, a lower haemoglobin concentration could account for lower aerobic capacity, which may reflect that testosterone stimulates haemoglobin production (Yoshiga & Higuchi, 2003). Therefore, it stands to reason that sex adds variance to the models of the start, end phase and 2000 m time trial.

Limitations

The main limitation of this study is the heterogenous and small sample including both male and female athletes, with different levels of performance, age and body weight. Even though the biological maturity was calculated as peak height velocity, six of the athletes were already over 18 years old, whereby the formula becomes inaccurate. However, due to the cross-sectional nature of this study, we deem the effect of biological age to be minimal. In a similar manner, menses were not assessed in female participants. Nevertheless, it is questionable whether this would affect the findings of this study, as current evidence shows no systematic influence of the menstrual cycle on acute strength and endurance performance (Taipale-Mikkonen et al., 2021; Colenso-Semple et al., 2023). As the results of this study are cross-sectional, the question of compliance over time and the influence of training on strength determinants raises, which need to be investigated in further research. When interpreting our findings, one should bear in mind that isometric and isokinetic strength measurement methods are to some extent unnatural movements that depend on the device setting and joint angles, respectively.

Conclusion

In this study, rowing ergometer performance was analysed in terms of the different phases and the strength determinants that explain variability in these phases was assessed. Our results confirm that specific strength determinants do explain the variability in rowing ergometer performance of adolescence athletes particularly in the start phase. In addition to VO₂max, anthropometry and sex, which were already well documented in previous research (Ingham et al., 2002; Cosgrove et al., 1999; Yoshiga & Higuchi, 2003; Akça, 2014; Hagerman, 1984), we suggest a fast power production and transmission via trunk extension musculature to the oar seems to be essential in the start sprint, whereas for the sustained middle section force production from the legs supplements VO₂max as a limiting parameter. While this is important knowledge that helps to further characterise Olympic rowing performance that might provide more precise training recommendations for consistent performance development, it remains to be assessed whether chronic strength training targeting these specific strength determinants will lead to improved rowing performance.

Conflict of interests

We reported no conflicts of interest.

Acknowledgement

We would like to thank the rowing athletes and coaches for their engagement.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.