Effects of strength training on physical fitness and sport-specific performance in recreational, sub-elite, and elite rowers: A systematic review with meta-analysis

ABSTRACT

The purpose of this systematic review with meta-analysis was to examine the effects of strength training (ST) on selected components of physical fitness (e.g., lower/upper limb maximal strength, muscular endurance, jump performance, cardiorespiratory endurance) and sport-specific performance in rowers. Only studies with an active control group were included if they examined the effects of ST on at least one proxy of physical fitness and/or sport-specific performance in rowers. Weighted and averaged standardized mean differences (SMD) were calculated using random-effects models. Subgroup analyses were computed to identify effects of ST type or expertise level on sport-specific performance. Our analyses revealed significant small effects of ST on lower limb maximal strength (SMD = 0.42, p = 0.05) and on sport-specific performance (SMD = 0.32, p = 0.05). Non-significant effects were found for upper limb maximal strength, upper/lower limb muscular endurance, jump performance, and cardiorespiratory endurance. Subgroup analyses for ST type and expertise level showed non-significant differences between the respective subgroups of rowers (p ≥ 0.32). Our systematic review with meta-analysis indicated that ST is an effective means for improving lower limb maximal strength and sport-specific performance in rowers. However, ST-induced effects are neither modulated by ST type nor rowers’ expertise level.

Abbreviations

CON: control group; ICC: intraclass correlation coefficient; CRE: cardiorespiratory endurance; F: female; IG: intervention group; INT: intervention group; M: male; Sets: number of sets per exercise; SMD: standardized mean differences; SMD_wm: weighted mean SMD; ST: strength training; RCT: randomized controlled trial; Reps: repetitions; RM: repetition maximum; TF: training frequency (times per week); TI: training intensity (eg., % of 1 repetition maximum); TP: training periods (weeks)

KEYWORDS:

• Resistance training
• plyometric training
• on-water performance
• race time
• oarsmen
• athletic performance

Introduction

Competitive rowing is characterized by repetitive cyclic muscle actions and high demands on several components of physical fitness such as cardiorespiratory endurance [CRE] and muscular fitness (Akça, 2014; Bourgois et al., 2014; Mäestu et al., 2005). In this context, “muscular fitness” is used as an umbrella term for “maximum strength,” “local muscular endurance,” and “muscular power” (Granacher et al., 2016). For instance, Izquierdo-Gabarren et al. (2010) examined the effects of training status on measures of physical fitness in sub-elite male rowing athletes aged 21–30 years. These authors reported significantly higher maximum strength (i.e., bench-press 1-repetition maximum [RM]) and lower 2,000 m race times in elite compared with recreational athletes. Further, Olympic compared with high-school and club rowers (aged 23–35) exhibited higher maximum strength (normalized to body mass) in squat, deadlift, and bench pull exercises (McNeely et al., 2005). Moreover, Lawton et al. (2013b) reported that lower limb maximum strength (i.e., leg-press 5-RM) and upper limb muscular endurance (i.e., seated arm pull 60-RM) were the best predictors for 2,000 m rowing ergometer performance (R² = 59%) in elite male rowers aged 24 years.

To enhance muscular fitness and sport-specific performance, strength training (ST) appears to be an adequate and effective means in trained and untrained individuals (Lesinski et al., 2016; Rhea et al., 2003). For instance, the prospective cohort study of DuManoir et al. (2007) examined the effects of ST in addition to regular rowing training (i.e., CRE training) on maximum strength (e.g., 1-RM leg press) and sport-specific performance (e.g., 2,000 m rowing performance) in male novice rowers aged 31 years. The authors demonstrated that ten weeks of additional ST resulted in significant increases in 1-RM leg press and improved 2,000 m rowing ergometer time (DuManoir et al., 2007). Further, in addition to regular rowing practice, plyometric training showed significant improvements in 500 m time trial performance compared with aerobic cycling in male high-school rowers aged 16 years (Egan-Shuttler et al., 2017). Of note, the effects of ST on CRE in athletes in general and rowers, in particular, are crucial for performance during competition. For rowing, this is of particular interest because training-induced adaptations of strength training in addition to regular CRE-based rowing training (i.e., concurrent training) can be mitigated compared with single-mode ST in trained individuals (Coffey & Hawley, 2017; García-pallarés & Izquierdo, 2011). Previously, Lawton et al. (2011) summarized the literature on the effects of ST when integrated into regular rowing training on measures of physical fitness and sport-specific performance (Lawton et al., 2011). In their narrative review, Lawton et al. (2011) reported positive effects of heavy-resistance strength training and strength endurance training on sport-specific performance without reporting any statistically significant proof. However, the review of Lawton et al. (2011) is methodologically limited in terms of evidence level (i.e., narrative review) and the applied selection criteria (i.e., the inclusion of uncontrolled trials). To the best of our knowledge, no systematic review and/or meta-analysis have been conducted that compared the effects of ST vs. active controls in rowing athletes.

Thus, the purpose of this systematic review with meta-analysis was to examine the effects of ST vs. active controls on components of physical fitness (e.g., muscular fitness, CRE) and sport-specific performance (e.g., 2,000 m rowing ergometer performance, 20 min all-out) in rowing athletes (e.g., recreational, sub-elite, and elite athletes). With reference to the relevant literature (García-pallarés & Izquierdo, 2011; Lawton et al., 2011), we hypothesized that ST induces significant gains in measures of physical fitness (e.g., increase in 1-RM) and sport-specific performance (e.g., improved 2,000 m rowing performance) in rowers.

Methods

The present systematic review and meta-analysis was conducted in accordance with the recommendations of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA).

Systematic literature search

A computerized systematic literature search was performed in the databases PubMed, SPORTDiscus, and Web of Science. The following Boolean search syntax was applied using the operators “AND”, “OR”, “NOT”: (“elite athlete*” OR “sub-elite athlete*” OR “recreational athlete*” OR athlete* OR “youth athletes” OR “young athletes”) AND (rowing OR rower OR row* OR oarsmen) AND (“strength training” OR “resistance training” OR “weight training” OR “power training” OR “plyometric training” OR “complex training” OR “compound training” OR “weight-bearing exercise”) AND (“maximum strength” OR “jump performance” OR “cardiorespiratory endurance” OR “muscular endurance” OR “rowing performance”) NOT (patient OR patients). The vocabulary was controlled using the Medical Subject Headings (MeSH) in PubMed. The search was limited to full-text availability, publication dates (01/01/1970 to 06/11/2019), and language (English). Further, relevant review articles and the reference lists of the included studies were screened for titles to identify additional suitable studies for inclusion in this review article.

Selection criteria

Studies were included in the present systematic review and meta-analysis if they adhered to the PICOS approach. This standardized question approach addresses five review components: patient population or disease (P), interventions or exposure of interest (I), comparators (C), main outcome or endpoint of interest (O), and study design (S) (Liberati et al., 2009).

Based on the defined inclusion and exclusion criteria (Table 1), two independent reviewers (DT, OP) screened potentially relevant papers by analysing titles, abstracts, and full texts of the respective articles to decide upon their eligibility to be included in this paper. In case DT and OP did not reach an agreement concerning inclusion/exclusion of an article, an additional author (UG) was contacted.

Table 1. Selection criteria.

Category	Inclusion criteria	Exclusion criteria
Population	Healthy male and female rowers	Participants from other sports or without an athletic background (i.e., organized athletic training)
Intervention type	Heavy-resistance strength training Strength endurance training Power training	Less than 6 strength training sessions Strength training with inspiratory/respiratory muscle training Effects of nutritional supplements
Comparator	Active control group	No control or passive control group
Outcome	At least one measure of rowing performance, muscle strength, local muscular endurance, jump performance, and CRE	No reported means and standard deviations
Study design	Controlled study	Case study, observational study

CRE cardiorespiratory endurance

Coding of studies

Each study was coded for the following variables: number of participants, sex, age, and expertise level. The expertise level of the participants was classified as elite (i.e., international top-level athletes), sub-elite (i.e., national top-level athletes), and recreational athletes (i.e., competitive athletes; Prieske et al., 2016). Our quantitative analyses focused on measures of maximal strength (upper/lower limbs), local muscular endurance (upper/lower limbs), jump performance, CRE, and sport-specific performance (e.g., 2,000 m rowing ergometer performance). If studies examined more than one variable within these outcome categories, only one representative outcome variable was included in the analysis. The variables were ranked regarding their priority for each outcome including the variable with the highest ranking in the analysis (see Table 2). Strength training programmes were coded for the following training modalities: training type (i.e., heavy-resistance strength training), frequency (i.e., number of sessions per week), and volume (i.e., number of sets per exercise, number of repetitions per set).

Table 2. Study coding.

Outcome parameters	Ranking
Performance in maximal strength	Maximal isometric strength 1-RM 5-RM
Performance in muscular endurance	70% of the 1-RM The time needed for 60 repetitions
Jump performance	Countermovement jump Vertical jump Broad jump
Cardiorespiratory endurance performance	V̇O2max Watts at 4 mmol/l Repetitive Wingate test Shuttle run test
Sport-specific performance (rowing ergometer performance)	2,000 m 500 m 20 min all out test Distance rowed in 90 s

1-RM = one repetition maximum; V̇O_2max maximal oxygen uptake

In terms of training type, ST was classified as heavy-resistance strength training, strength endurance training, and power training. Heavy-resistance strength/power training and strength endurance training are each located at the opposite extremes of the ST continuum (Beattie et al., 2014; Knuttgen, 2007). If data were missing in the included studies, authors were contacted to kindly provide additional information (Lawton et al., 2013b, 2012). In two studies, missing pre- and post-test data were obtained from the published charts (Buckley et al., 2015; Izquierdo-Gabarren, Expósito et al., 2010).

Assessment of risk of bias

The Physiotherapy Evidence Database (PEDro) scale was used to provide general information on the methodological quality and to evaluate the risk of bias in the selected studies. The PEDro scale rates internal study validity and the presence of statistical replicable information using a 0- (high risk of bias) to 10-point (low risk of bias) scale with ≥6 points representing the cut-off for a low risk of bias (Maher et al., 2003).

Statistical analyses

To determine the effects of ST on measures of physical fitness and sport-specific performance in recreational, sub-elite, and elite rowers, between-subject standardized mean differences (SMD) were calculated for pre-test and post-test values of each study using the following formula: $SM{D}_{\hspace{0.17em}}=\frac{m_1-m_2}{s_{pooled}}$, with m₁ standing for the mean pre/post-test value of the intervention group, m₂ for the mean pre/post-test value of the control group, and s_pooled for the pooled standard deviation. In accordance with Hedges and Olkin (1985), pre/post-test SMDs were corrected for the respective sample size using the factor 1-(3/(4 N-9)). Additionally, adjusted SMD values were calculated as the difference between pre-test SMD to post-test SMD (Durlak, 2009). Finally, a random effects model was applied to weight each included study according to the magnitude of the respective standard error and to calculate the weighted mean adjusted SMD (SMD_wm). At least two intervention groups had to be included to calculate SMD_wm values for each proxy of physical fitness (Higgins & Green, 2011).

The meta-analysis was conducted using Review Manager 5.3 (Nordic Cochrane Centre, The Cochrane Collaboration, Copenhagen, Denmark). Subgroup analyses were computed for expertise level (i.e., recreational, sub-elite and elite rowers) and training type (i.e., strength/power training vs. strength endurance training). In order to improve readability, we consistently reported positive SMD values if superiority of ST compared with an active control were indicated. With reference to Cohen (1988), SMD_wm values were classified as small (SMD_wm < 0.50), medium (0.50 ≤ SMD_wm < 0.80), and large (SMD_wm ≥ 0.80). Further, between-study heterogeneity was assessed using I² and Chi² statistics. With respect to I2, heterogeneity was interpreted as low (I² < 25%), moderate (25% ≤ I² < 50%), high (50% ≤ I² < 75%) or considerably high (I² ≥ 75%) (Deeks et al., 2019; Higgins et al., 2003). The level of statistical significance was set at p ≤ 0.05.

Results

Study characteristics

Table 4 illustrates the risk of bias using the PEDro scale. Overall, the median score of the included studies was 4 with a range of 3 to 5. This score is indicative of low methodological quality.

Table 3. Included studies examining the effects of resistance training in athletes.

	No. of subjects							Interventions
Study	All	M	F	INT	CON	Age [years]	Expertise level	TP	TF	TI	Exercises	Sets	Reps	Rest
Buckley et al. (2015)	28	0	28	14	14	25 ± 5	Recreational	6	3	NA	3	6	all out in 60 s	3 min
Ebben et al. (2004) (IG, p. 1)	26	0	26	13	13	20 ± 1	Recreational and	8	3	80 to 100% 80 to 100%	6	3	05 to 12	NA
Ebben et al. (2004) (IG, p. 2)							Sub-Elite	8	3		6	2	15 to 32	NA
Egan-Shuttler et al. (2017)	16	16	0	8	8	16 ± 1	Recreational	4	3	100%	6 to 7	3 to 5	5 to 20	NA
Gallagher et al. (2010) (IG, p. 1)	18	18	0	6	5	20 ± 1	Sub-elite	8	2	NA	7	2	15 to 30	2 min
Gallagher et al. (2010) (IG, p. 2)				7	5			8	2	NA	7	3 to 5	1 to 5	2 min
Izquierdo-Gabarren et al. (2010) (IG 1)	43	43	0	14	8	26 ± 4	Sub-elite	8	2	75 to 92%	4	3 to 4	4 to 10	NA
Izquierdo-Gabarren et al. (2010) (IG 2)				15	8			8	2	75 to 92%	4	3 to 4	2 to 5
Izquierdo-Gabarren et al. (2010) (IG 3)				6	8			8	2	75 to 92%	2	3 to 4	2 to 5
Kramer et al. (1993)	24	0	24	13	11	19 ± 1	Recreational	9	3	100%	4	1 to 5	5 to 20	NA
Lawton et al. (2012)	22	12	10	12	10	24 ± 4	Elite	8	2	NA	6	3 to 4	6 to 15	NA
								8	2	NA	8	3 to 4	25 to 50	NA
Lawton et al. (2013a)	10	0	10	10	10	23 ± 4	Elite	14	1	90% of 1-RM	4	3 to 4	6 to 15	NA
									1	NA	7	3	15 to 30	20 s
Tse et al. (2005)	45	45	0	25	20	21 ± 1	Recreational	8	2	NA	NA	NA	NA	NA

1 RM one repetition maximum, IG intervention group, F female, M male, Reps repetitions, Sets number of sets per exercise, TF training frequency (times per week), TI training intensity (e.g., % of 1 repetition maximum); TP training periods (weeks).

Table 4. Physiotherapy Evidence Database (PEDro) score of the included studies.

Study	Eligibility criteria	Randomized allocation	Blinded allocation	Group homogeneity	Blinded subjects	Blinded therapists	Blinded assessor	Drop out <15%	Intention to treat analysis	Between-group comparison	Point estimates and variability	PEDro score
Buckley et al. (2015)	0	1	1	0	0	0	0	1	0	1	1	5
Ebben et al. (2004)	0	1	0	1	0	0	0	1	0	1	1	5
Egan-Shuttler et al. (2017)	0	0	0	1	0	0	0	1	0	1	1	4
Gallagher et al. (2010)	0	1	0	0	0	0	0	1	0	1	1	4
Izquierdo-Gabarren et al. (2010)	0	1	0	1	0	0	0	1	0	1	1	5
Kramer et al. (1993)	0	1	0	0	0	0	0	1	0	1	1	4
Lawton et al. (2012)	1	0	0	1	0	0	0	1	0	1	0	4
Lawton et al. (2013a)	1	1	0	1	0	0	0	1	0	1	0	5
Tse et al. (2005)	0	0	0	1	0	0	0	0	0	1	1	3

1 indicates a “yes” score; 0 indicates a “no” score.

In terms of potentially relevant journal articles, Figure 1 represents a flow chart that highlights the different phases of the systematic literature search. A total of 249 potentially relevant studies was identified in the electronic databases PubMed, SPORTDiscus, and Web of Science. The application of the predefined inclusion/exclusion criteria, reduced the number of studies to nine which were used for quantitative analyses. A total of 232 (males = 134/females = 98) athletes participated in these studies and 143 received heavy resistance strength training and/or strength endurance training in 5 studies and power training in 2 studies in overall 12 intervention groups.

Figure 1. Flow chart illustrating the different phases of the systematic literature search and study selection.

The sample size of the intervention groups ranged from six to 25 participants with an average age ranging from 16 to 25 years (Table 3). Eight out of nine studies recruited adult athletes, whereas only one study (Egan-Shuttler et al., 2017) included young athletes.

In terms of expertise level, two studies investigated elite athletes (Lawton et al., 2012, 2013b), three studies included sub-elite athletes (Ebben et al., 2004; Gallagher et al., 2010; Izquierdo-Gabarren, González De Txabarri Expósito et al., 2010), and five studies examined recreational athletes (Buckley et al., 2015; Ebben et al., 2004; Egan-Shuttler et al., 2017; Kramer et al., 1993; Tse et al., 2005).

Further, ST interventions lasted between four and fourteen weeks with training frequencies ranging from two to four sessions per week. The training volumes comprised two to six sets per exercise and one to fifty repetitions per set. Training protocols of the active control groups were heterogeneous and varied largely across the nine studies. It ranged from hardly any information on the training programme (Tse et al., 2005) to a detailed characterization of the training protocol (Lawton et al., 2013b).

The active control group did not undergo any ST but continued their regular training, similar to that performed by the intervention groups (Gallagher et al., 2010; Izquierdo-Gabarren, González De Txabarri Expósito et al., 2010; Kramer et al., 1993; Tse et al., 2005). For that reason, the training volume (i.e., total weekly minutes of exercise) of both groups are different in these four studies.

To match training volumes, the active control group was assigned to an additional training (e.g., endurance rowing, cycling), for the equivalent ST time of the intervention group in five studies (Buckley et al., 2015; Ebben et al., 2004; Egan-Shuttler et al., 2017; Lawton et al., 2012, 2013a). One out of five studies included another ST type (maximal strength vs. muscular endurance strength training) for the active control group to match the training volume of both groups (Ebben et al., 2004). In two of the five studies, the active controls and intervention groups performed the same ST in their regular training (Kramer et al., 1993; Tse et al., 2005). Six studies with ten ST interventions were included for subgroup-analyses to examine the effects of ST on sport-specific performance with heavy resistance strength/power training (five studies, eight interventions) and strength endurance training (two studies, two interventions). Furthermore, the same studies were included to examine the effects of ST for sport-specific performance in sub-elite (three studies, six interventions) or recreational (four studies, four interventions) athletes. The two studies with elite athletes had no sport-specific performance outcome (Lawton et al., 2012, 2013a).

Effects of strength training on physical fitness in recreational, sub-elite or elite rowers

Four studies were eligible for inclusion in our systematic review and meta-analysis that determined the effects of ST on measures of upper limbs maximal strength (e.g., 1-RM bench pull and isometric pull) compared with no or single mode regular rowing training (Izquierdo-Gabarren, González De Txabarri Expósito et al., 2010; Kramer et al., 1993; Lawton et al., 2012, 2013b). The analysis revealed a non-significant, small effect (SMD_wm = 0.3, I² = 0%, Chi² = 1.74, df = 5, p = 0.10) in favour of ST, irrespective of the expertise level (Figure 2). Further, four studies examined the effects of ST on measures of lower limbs maximal strength (e.g., 1-RM leg press) compared with no or single-mode regular rowing training (Buckley et al., 2015; Kramer et al., 1993; Lawton et al., 2012, 2013b). Results showed significant, small effects (SMD_wm = 0.42, I² = 0%, Chi² = 2.36, df = 3, p = 0.05) in favour of ST in recreational and elite rowers (Figure 3).

Figure 2. Effects of strength training compared with an active control (CON) on measures of upper limb maximal strength (e.g., 1-RM bench pull) in recreational, sub-elite, and elite rowers. CI = confidence interval, df = degrees of freedom, IG = intervention, IV = inverse variance, Random = random effects model, SE = standard error, SMD = standardized mean difference.

Figure 3. Effects of strength training compared with an active control group (CON) on measures of lower limb maximal strength (e.g., 1-RM leg press) in recreational and elite rowers. CI = confidence interval, df = degrees of freedom, IV = inverse variance, Random = random effects model, SE = standard error, SMD = standardized mean difference.

Two studies investigated the effects of ST on measures of upper limbs muscular endurance (e.g., 70% 1-RM bench pull and 60-RM seated arm pull) compared with no or single mode regular rowing training (Kramer et al., 1993; Lawton et al., 2012). Findings showed non-significant, small effects (SMD_wm = 0.49, I² = 0%, Chi² = 0.05, df = 1, p = 0.10) in favour of ST in recreational and elite rowers. Moreover, two studies investigated the effects of ST on measures of lower limbs muscular endurance (e.g., 70% 1-RM leg press and squats) compared with no or single-mode regular rowing training (Buckley et al., 2015; Kramer et al., 1993). The analysis revealed a non-significant, medium effect (SMD_wm = 0.78, I² = 80%, Chi² = 4.94, df = 1, p = 0.24) in favour of ST in recreational rowers.

In terms of jump performance (e.g., countermovement jump), only three studies with active controls analysed the effects of ST compared with no or single-mode regular rowing training (Buckley et al., 2015; Kramer et al., 1993; Tse et al., 2005). Results showed non-significant, small effects (SMD_wm = 0.06, I² = 0%, Chi² = 1.7, df = 2, p = 0.78) in favour of ST in recreational rowers (Figure 4).

Four studies determined the effects of ST on CRE performance (e.g., V̇O_2max) compared with no or regular rowing training only (Buckley et al., 2015; Izquierdo-Gabarren, González De Txabarri Expósito et al., 2010; Lawton et al., 2012; Tse et al., 2005). A non-significant and small effect was observed in favour of the control group in recreational, sub-elite, and elite rowers (SMD_wm = −0.09, I² = 12%, Chi² = 5.65, df = 5, p = 0.64) (Figure 5).

Figure 4. Effects of strength training compared with an active control group (CON) on measures of jump performance (e.g., countermovement jump) in recreational rowers. CI = confidence interval, df = degrees of freedom, IV = inverse variance, Random = random effects model, SE = standard error, SMD = standardized mean difference.

Figure 5. Effects of strength training compared with an active control group (CON) on measures of cardiorespiratory endurance (e.g., V̇O₂max) in recreational, sub-elite and elite rowers. CI = confidence interval, df = degrees of freedom, IG = intervention, IV = inverse variance, Random = random effects model, SE = standard error, SMD = standardized mean difference.

Effects of strength training on sport-specific performance in recreational, sub-elite or elite rowers

Six studies using twelve interventions determined the effects of ST on measures of sport-specific performance (e.g., 2,000 m rowing ergometer time) compared with no or single-mode regular rowing training (Ebben et al., 2004; Egan-Shuttler et al., 2017; Gallagher et al., 2010; Izquierdo-Gabarren, Expósito et al., 2010; Izquierdo-Gabarren, González De Txabarri Expósito et al., 2010; Kramer et al., 1993; Tse et al., 2005).

Figure 6. Effects of strength training compared with an active control group (CON) on measures of sport-specific performance (e.g., 2.000 m rowing ergometer time) in recreational and sub-elite rowers. CI = confidence interval, df = degrees of freedom, IG = Intervention, INT = intervention group, IV = inverse variance, Random = random effects model, SE = standard error, SMD = standardized mean difference.

The analysis revealed a significant, small sized effect (SMD_wm = 0.32, I² = 0%, Chi² = 5.1, df = 9, p = 0.05) in favour of ST (Figure 6). Subgroup analyses indicated non-significant (I² = 0%, Chi² = 0.14, df = 1, p = 0.71) differences between the effects of heavy resistance strength/power versus strength endurance training on sport-specific performance in recreational and sub-elite rowers.

Furthermore, subgroup analyses for expertise level regarding the effects of ST on sport-specific performance indicated non-significant differences between recreational and sub-elite rowers (I² = 0%, Chi² = 0.98, df = 1, p = 0.32).

Discussion

The aim of this systematic review with meta-analysis was to examine the effects of ST on components of physical fitness (e.g., muscular fitness, CRE) and sport-specific performance (e.g., 2,000 m rowing ergometer performance) in recreational, sub-elite or elite rowers. The main conclusions were that (i) ST produced significant small-sized effects on lower limbs maximal strength in recreational and elite rowers and small-sized effects on sport-specific performance in recreational and sub-elite rowers, and (ii) training type and expertise level did not influence the effects of ST on sport-specific performance.

Effects of strength training on physical fitness in recreational, sub-elite or elite rowers

Our findings revealed significant, small-sized effects of ST on lower limb maximum strength. Similarly, previous reviews and meta-analytical studies reported significant enhancements in maximum strength following ST in trained and untrained individuals of various ages (Lesinski et al., 2016; Rhea et al., 2003). For instance, Rhea et al. (2003) analysed 140 studies with a total of 1,433 effect sizes to identify a dose-response relationship for strength improvement following ST. The authors showed that ST with moderate intensity (i.e., 60% 1-RM, effect size [ES] = 2.8), four sets per exercise and training session (ES = 2.3), and three sessions per week (ES = 1.9) induced the largest strength gains in untrained individuals whereas ST with higher intensities (i.e., 80% 1-RM, ES = 1.8), four sets per exercise and session (ES = 1.2), and two sessions per week (ES = 1.4) were most efficient in trained individuals (Rhea et al., 2003). In this regard, our findings of significant effects of ST on lower limb maximum strength in rowers partly confirmed the results of Rhea et al. (2003). For instance, Lawton et al. (2012) investigated the effects of eight weeks of combined heavy resistance strength and strength endurance training (three to four sets per exercise and session, two sessions per week) in elite rowers (mean age: 24 years). Authors reported that combined maximum strength and muscular endurance training significantly improved lower limb maximum strength (e.g., leg press +2.4%), upper limb maximum strength (e.g., isometric pull +12.4%), and lower limb muscular endurance (e.g., leg press 30-RM +4.0%) in comparison with rowing only in elite rowers. Interestingly, adaptations to ST in addition to regular CRE-based training (concurrent training) can be mitigated compared with single-mode ST only in trained individuals (Coffey & Hawley, 2017; García-pallarés & Izquierdo, 2011). Given that the included studies conducted ST in addition to the regular CRE-based rowing training, future studies may need to examine whether a well-periodized program with ST only is more effective to improve maximum strength measures and/or sport-specific performance in rowers compared with concurrent training programs.

Interestingly, our analyses did not confirm the hypothesis that ST will improve upper limb maximal strength, muscular endurance, jump performance, and CRE. It can be speculated that methodological reasons such as the inappropriate training modalities implemented in ST interventions of the included studies are responsible for the lack of significant ST-induced fitness gains in rowers. Only one study (Izquierdo-Gabarren, González De Txabarri Expósito et al., 2010) adhered to the recommendations of Rhea et al. (2003) in regards to training intensity (75 to 92%), training volume (3 to 4 sets), and training frequencies (two times per week) for trained athletes. All the other studies used inappropriate training modalities. For instance, Lawton et al. (2012) implemented ST four times (instead of two times) per week without specifying the intensity for the maximum strength and strength endurance training. Accordingly included studies failed to specify respective parameters (Lawton et al., 2012, 2013a). It can be concluded that ST interventions should adhere more closely to the suggested dose-response relationships for strength development in trained and untrained individuals (Rhea et al., 2003).

Another possible explanation for the lack of performance gains in upper limb maximum strength, muscular endurance, jump performance, and CRE observed in this study may be the mixed training types (e.g., machine-based, free weight training) within one session. For instance, Lawton et al. (2013a) investigated the changes in lower-body strength development after two 14-week phases of intensive resisted on-water rowing, either incorporating weight training or rowing alone in elite rowers aged 23 years. The ST intervention included machine-based (e.g., leg-press), free weight (e.g., deadlifts) and plyometric training (e.g., hurdle jumping) exercises. Of note, each of these ST modalities has specific benefits and limitations. For instance, Lesinski et al. (2016) found that ST programs using free weights were most effective to enhance maximum strength in young athletes. Accordingly, to evaluating the effects of ST on different measures of physical fitness, it would be better to focus on one training type (e.g., free weight training) depending on the ST goal (e.g., enhance maximum strength).

Between two and four studies were included with regards to the different physical fitness outcomes. For that reason, more research is needed to make a concise statement on the different measures of physical fitness. In addition, the use of different measures of physical fitness across studies resulted in diminished comparability of the included studies.

For the largest ST-induced performance gains (e.g., maximal strength) it was postulated that ST intervention periods should last ten weeks and more (Lesinski et al., 2016; Lawton et al., 2011). However, only the cross-over study of Lawton et al. (2013a) compared ST programs lasting ten weeks with programs that lasted five to nine weeks(Buckley et al., 2015; Ebben et al., 2004; Egan-Shuttler et al., 2017; Gallagher et al., 2010; Izquierdo-Gabarren, González De Txabarri Expósito et al., 2010; Kramer et al., 1993; Lawton et al., 2012, 2013a; Tse et al., 2005). Thus, future studies with longer ST durations are required to clarify any beneficial effects of longer intervention periods on rowers’ physical fitness.

In order to evaluate the effects of ST on measures of physical fitness in recreational, sub-elite or elite rowers, future research should consider the following study characteristics: (i) controlled studies with one (active) control (i.e., rowing only) and one ST group. Both experimental groups should apply similar training volumes. (ii) training modalities should adhere to the recommendations intensities (i.e., 80% 1-RM), four sets per exercise and session, and two sessions per week in trained individuals as provided by Rhea et al. (2003). (iii) researchers should focus on only one training goal (e.g., improvement of maximal strength) and training type (e.g., power training) for the ST interventions. (iv) future research should implement the same tests in their studies for the different outcome parameters such as maximal strength (e.g., 1-RM), local muscular endurance (e.g., repetitions at 70% of the 1-RM), jump performance (e.g., countermovement jump height), and CRE (e.g., V̇O_2max).

Effects of ST on sport-specific performance in recreational, sub-elite or elite rowers

This meta-analysis revealed significant small-sized effects of ST on sport-specific performance in recreational and sub-elite rowers compared with active controls (Ebben et al., 2004; Egan-Shuttler et al., 2017; Gallagher et al., 2010; Izquierdo-Gabarren, González De Txabarri Expósito et al., 2010; Kramer et al., 1993; Tse et al., 2005). This is in line with the literature as it has previously been shown that ST is effective to improve rowing performance (Akça, 2014; DuManoir et al., 2007; Gee et al., 2011; Huang et al., 2007; McNeely et al., 2005). For instance, Akça (2014) and Huang et al. (2007) showed significant relationships between lower limbs maximal strength (1-RM leg press r = −0.536, p ≤ 0.05, r = −0.755, p < 0.01) and sport-specific performance (2,000 m rowing performance). DuManoir et al. (2007) showed a significant (p < 0.05) improvement of a 10 week heavy resistance strength training on physical fitness (lower limb maximal strength) and transfer effects on sport-specific performance (2,000 m rowing performance) in recreational rowers. Similarly, our systematic review with meta-analysis showed concomitant gains in lower limb maximal strength and rowing performance. From a functional perspective, it can be speculated that training-induced gains in leg muscle strength may partly be transferred to rowing performance. In fact, eight weeks of lower limb strength training improved maximal lower limb strength and movement velocity during a cycle ergometer test in adolescent soccer players compared with an active control (Hammami et al., 2018). In rowers, lower limb movement velocity essentially contributes to total oars velocity during on-water rowing particularly during the first half of the drive phase (i.e., 48.2–90.1%) (Lamb, 1989). Thus, training-induced gains in lower limb maximal strength may induce improvements in rowing performance through an increase of lower limb movement velocity during rowing.

Effects of ST on sport-specific performance are not influenced by training type or expertise level. This can most likely be explained by the heterogeneous training types and their corresponding effects that were aggregated in this study. For instance, Lesinski et al. (2016) showed that free weight ST produced the largest effects on maximal strength in young athletes, while complex training had the largest effect on sport-specific performance. Ebben et al. (2004) investigated the effects of heavy resistance strength vs. strength endurance training on sport-specific performance in recreational and sub-elite rowers with a mean age of 20 years. Free weight (e.g., walking lunge), machine-based (e.g., bench press), and complex training (e.g., deadlift) were included in the training protocol. In our systematic review with meta-analysis, no study was included that applied complex training exercises only. Three studies were included with at least one complex training exercise (e.g., deadlift (Ebben et al., 2004; Gallagher et al., 2010; Izquierdo-Gabarren, González De Txabarri Expósito et al., 2010) and three studies without complex training exercises (Egan-Shuttler et al., 2017; Kramer et al., 1993; Tse et al., 2005). In our analyses, Tse et al. (2005) examined the effects of core endurance ST in addition to regular rowing training on jump performance (e.g., vertical jump), CRE (e.g., shuttle run), and rowing performance (e.g., 2,000 m) in recreational rowers. No significant differences were found for any of the functional performance tests. However, the meta-analysis of Prieske et al. (2016) revealed that trunk muscle strength plays only a minor role in physical fitness and sport-specific performance.

Limitations

This study has a few limitations that warrant discussion. First, the methodological quality of the included studies was relatively low. In fact, none of the included studies reached the pre-determined PEDro cut-off score of ≥6 points. Low methodological quality may have been caused by a lack of randomization with regards to the allocation of individuals to experimental groups. Accordingly, future high-quality interventions studies (e.g., randomized group allocation) are needed that examine the effects of strength training on physical fitness and rowing performance.

Second, the included studies comprised participants of different expertise levels (recreational sub-elite and elite athletes). Different ST types were analysed (i.e., plyometric training (Egan-Shuttler et al., 2017; Kramer et al., 1993), core stability training (Tse et al., 2005), or ST with high load and low repetitions/low load and high repetitions (Buckley et al., 2015; Ebben et al., 2004; Gallagher et al., 2010; Izquierdo-Gabarren, González De Txabarri Expósito et al., 2010; Lawton et al., 2012, 2013b). Heterogenous testing methods were applied (e.g., CRE testing using incremental treadmill tests, 10 m shuttle run tests, rowing power tests at blood lactate of 4 mmol/L). Future studies should apply similar tests to allow comparison between studies, particularly with regards to sport-specific testing.

Conclusions

The present systematic review with meta-analysis showed that ST produced small-sized effects on lower limb maximal strength in recreational and elite rowers and small-sized effects on sport-specific performance in recreational and sub-elite rowers when compared with active controls. Due to the rather low methodological quality of the included studies, further high-quality studies (i.e., RCTs or CTs) are needed with rowing athletes. More specifically, future studies should specify all relevant ST training modalities (e.g., training volume, intensity etc) when describing the training protocols. The largest ST-induced performance gains (e.g., maximal strength) were noted for trials that lasted 10 weeks and longer (Lesinski et al., 2016; Lawton et al., 2011). However, this is a preliminary finding that needs to be verified in future studies. With reference to the findings of this study, coaches are advised to regularly implement ST in the training regime of rowers to enhance lower limbs maximal strength. More specifically, ST should address lower limbs muscles to improve rowing performance. These findings are irrespective of training type (heavy resistance strength/power training vs. strength endurance training) and expertise level (recreational vs. sub-elite).

Authors’ Contributions

DT, OP, HC, UG analyzed and interpreted the data and were a major contributor in writing the manuscript. All authors read and approved the final manuscript.

Availability of Data and Material

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Disclosure Statement

Dirk Thiele, Olaf Prieske, Helmi Chaabene, and Urs Granacher declare that there is no conflict of interest.

Additional information

Funding

This study is part of the research project “Resistance Training in Youth Athletes” that was funded by the German Federal Institute of Sport Science [ZMVI1-081901 14-18].