We systematically searched SCOPUS, PubMed, Web of Science, and SPORTDiscus (via EBSCOhost) electronic databases from inception until 30th May 2022. A systematic investigation of the topic was carried out utilizing the Boolean operations AND and OR. For keyword selection and search strategy development, the authors sought advice from experienced librarians. The keywords are as follows: (“plyometric training” OR “plyometric exercise*” OR “plyometric drill*” OR “plyometr*” OR “ballistic six” OR “ballistic training” OR “explosive” OR “force-velocity” OR “stretch-shortening cycle” OR “stretch-shortening exercise” OR “complex training” OR “jump training”) AND (“tennis” OR “tennis player*” OR “tennis athlete*”). The main databases search string is provided in Supplementary Appendix S1. Furthermore, to find additional literature that might not have shown up in the search results using the four databases, a search was also carried out on Google Scholar and based on the reference lists of selected papers, previously relevant reviews and meta-analyses (Singla et al., 2018; Sole et al., 2021; Colomar et al., 2022; Xiao et al., 2022).

Firstly, two authors (ND, DH) uploaded the collected literature information to the Endnote X9 reference management software based on title, type, authors, and year of publication. After that, all duplicates were removed (ND). Two independent authors (ND, KS) screened each study’s title and abstract for potentially relevant full-text literature. After that, the same authors compared the entire study text to the inclusion and exclusion criteria and chose publications that fulfilled the requirements. Two independent authors (ND, KS) worked out their disagreements through discussion; if they disagreed, a third author (BA) was consulted until they reached an agreement. Figure 1 illustrates the details of the selection procedure.

To evaluate whether studies could be included in the review, we employed the PICOS approach (Table 1). The selection criteria that we considered for our systematic review are stated below:

1) Healthy tennis players, with no limitations on their gender, age or level.

The following exclusion criteria were used in our systematic review:

1) Unhealthy tennis players or those with injuries.

Two reviewers (ND, DH) retrieved data from each study using a Microsoft Excel spreadsheet (Microsoft Corporation, Redmond, WA, United States), and a third reviewer (KS) validated the data. The data considered were: 1) Name of the first author and year of publication; 2) Subject characteristics: sample size, gender, age (years), and tennis experience; and 3) Characteristics of the PT intervention, which are training protocol, training frequency (days/week), duration (weeks), intensity level (e.g., maximal), rest time between sessions (hours), rest time between sets (s), rest time between repetitions (s), type of progressive PT overload (e.g., volume-based; technique-based), training period (e.g., in-season), PT replaced (if applicable) a component of the regular tennis practice.

According to the guideline on the webpage for Cochrane Training, two reviewers (ND, KS) independently assessed the risk of bias of each of the identified RCTs using the updated Cochrane risk of bias assessment for randomized trials (RoB-2) (Higgins et al., 2022). Risk Of Bias In Non-randomized Research of Interventions (ROBINS-I) was used to evaluate the risk of bias in non-randomized controlled trials (Higgins et al., 2022). GRADE was used to analyze and summarize the confidence of evidence following the GRADE handbook’s principles (Schünemann et al., 2020). First, we categorized all relevant trials based on the outcomes they reported. Then, to determine the confidence of evidence, the following six factors were considered: research design, study limitations, inconsistency, indirectness, imprecision, and publication bias. Outcomes were assessed separately for RCTs and non-RCTs. Two individuals authored this work (ND, KS).

If three or more relatively homogeneous studies supplied explicit pre-and post-test data for the control and experimental groups using the same parameters, these papers were combined for meta-analysis (Ramirez-Campillo et al., 2021b; Ramachandran et al., 2021). In contrast, a narrative synthesis of the findings is conducted (Elgueta-Cancino et al., 2022). On the basis of pre- and post-intervention performance means and standard deviations, between-group effect sizes (ES; Hedge’s g) were computed (SD). The data were standardized using the post-score SD. The inverse-variance random-effects model was utilized for meta-analyses because it assigns a proportional weight to trials depending on the magnitude of their individual standard errors (Deeks et al., 2008) and aids analysis while accounting for heterogeneity across studies (Kontopantelis et al., 2013). If the needed information was not available in the original article or additional materials, the authors were contacted. If contact was not returned or data was unavailable, the study was excluded from the meta-analysis. The ES values were presented with 95% confidence intervals (95% CIs). The ES magnitudes were interpreted using the following scale: < 0.2, trivial; 0.2–0.6, small; > 0.6–1.2, moderate; >1.2–2.0, large; >2.0–4.0, very large; >4.0, extremely large (Hopkins et al., 2009). In some investigations including multiple PT groups, the control group was proportionally divided to permit comparisons across all participants (Higgins et al., 2008). To assess heterogeneity, the I² statistic was employed. Low, moderate, and high degrees of heterogeneity, were determined to be 25%, 25–75%, and >75%, respectively (Higgins et al., 2003). The risk of publication bias across studies was assessed using the extended Egger’s test (Egger et al., 1997). Furthermore, a sensitivity analysis was performed when Egger’s test was significant (p < 0.05). An analysis of all available data was carried out using Comprehensive Meta-Analysis software (version three; Biostat, Englewood, NJ, United States).

The database search produced 361 articles, with an additional five articles discovered via Google Scholar and reference lists. In total, 224 research articles remained after duplicates were removed, and the remaining articles were eliminated based on title and abstract screening. An evaluation of the remaining 111 studies was conducted independently by two researchers. The systematic review included twelve studies after the final screening process (Table 2), and eight articles were included in the meta-analysis (Supplementary Appendix S2). The detailed selection procedure for the studies is shown in Figure 1.

RoB-2 was applied to five RCTs, while ROBINS-I was applied to seven non-RCTs. Ten trials were found to have an overall moderate risk of bias or presented some concerns, only two papers showed a lower risk of bias (see Figure 2). Figure 2A depicts the findings of the RoB-2 evaluations. Only one research described the method for generating randomization sequences utilizing stratified block randomization and specifically documented allocation concealment (Behringer et al., 2013), whereas four articles did not completely describe the randomized procedure (Salonikidis and Zafeiridis, 2008; Gelen et al., 2012; Rathore, 2016; Ziagkas et al., 2019). Only two studies had preregistered protocols (Salonikidis and Zafeiridis, 2008; Behringer et al., 2013), and three of the studies therefore had some concerns regarding bias in the selection of the reported results (Gelen et al., 2012; Rathore, 2016; Ziagkas et al., 2019). Figure 2B depicts a graphical representation of the outcomes of ROBINS-I evaluations. One of the non-RCTs discovered some concerns due to missing data, owing to an approximate 15% dropout rate (Fernandez-Fernandez et al., 2016), two studies had moderate risk of bias in domain of deviations from intended intervention (Ölçücü, 2013; Lakshmikanth et al., 2018), and three studies had moderate risk regarding bias in the selection of the reported results (Ölçücü et al., 2013; Fernandez-Fernandez et al., 2018; Hotwani et al., 2021).

According to the GRADE assessment (Supplementary Appendix S3), in terms of RCTs, the certainty of evidence is considered very lower to moderate. In terms of non-RCTs, the certainty of evidence is considered very lower to low.

Table 2 summarizes the study and intervention characteristics. Moreover, the training protocol of each study can be found in Supplementary Appendix S4. The twelve included papers were carried out between 2008 and 2021, with 443 tennis players. Across the studies, 1) Gender: nine studies focused on men (Salonikidis and Zafeiridis, 2008; Behringer et al., 2013; Fernandez-Fernandez et al., 2015; Fernandez-Fernandez et al., 2016; Fernandez-Fernandez et al., 2018; Mohanta et al., 2019), no studies focused on female tennis players alone, two studies reported mixed gender (Lakshmikanth et al., 2018; Hotwani et al., 2021), and one study did not specify gender (Gelen et al., 2012); 2) Age: all studies recorded the participants’ ages, and an overview of age reports from twelve research revealed that the participants’ ages ranged from 12.5 to 25 years; 3) Tennis experience: eight studies reported on the training experience of the tennis players (Salonikidis and Zafeiridis, 2008; Gelen et al., 2012; Behringer et al., 2013; Fernandez-Fernandez et al., 2015; Fernandez-Fernandez et al., 2016; Fernandez-Fernandez et al.2018; Mohanta et al., 2019), ranging from 12 to 96 months, four studies did not report on training experience (Ölçücü et al., 2013; Rathore, 2016; Lakshmikanth et al., 2018; Hotwani et al., 2021); 4) Intervention: regarding the training regimen in this review, four studies conducted PT for the upper and lower extremities (Behringer et al., 2013; Fernandez-Fernandez et al., 2016; Mohanta et al., 2019; Ziagkas et al., 2019), and one study used only upper limb PT (Gelen et al., 2012), six studies used lower-extremity PT (Salonikidis and Zafeiridis, 2008; Fernandez-Fernandez et al., 2015; Rathore, 2016; Fernandez-Fernandez et al., 2018; Lakshmikanth et al., 2018; Hotwani et al., 2021), but three studies did not detailed description the training protocol (Ölçücü, 2013; Rathore, 2016; Ziagkas et al., 2019). In addition, three studies combined PT with other types of exercise training (i.e., sprint training, acceleration/deceleration/change of direction drills) (Fernandez-Fernandez et al., 2015; Fernandez-Fernandez et al., 2018; Hotwani et al., 2021). The frequency of training was one to three times per week, and two studies did not report the frequency (Gelen et al., 2012; Hotwani et al., 2021). The duration of the intervention was primarily between 20 and 60 min, but one study did not report the period (Salonikidis and Zafeiridis, 2008). The interventions ranged from three to 9 weeks, and most of the interventions were conducted for 8 weeks (n = 8), and only one study did not report the length (Gelen et al., 2012).

The meta-analysis was conducted on certain trials that measured maximal serve velocity (n = 3), sprint speed (n = 3), lower extremity power (n = 3) and strength (n = 3), agility (n = 4). Other outcomes were insufficient and produced limited data for pooling, consequently, they were not considered for meta-analysis. The data used for meta-analyses and the forest plots are presented in Supplementary Appendix S3, S5 , respectively.

The PT showed a significant increase (p < 0.001) of maximal serve velocity with moderate effect (ES = 0.75; 95% CI = 0.38–1.12; Egger’s test p = 1.00; n = 115) and low heterogeneity (I² = 0.0%) (Supplementary Appendix S2; Supplementary Figure S1).

The PT showed a significant increase (p = 0.046) of sprint speed with moderate effect (ES = 0.43; 95% CI = 0.01–0.85; Egger’s test p = 1.00; n = 115) and low heterogeneity (I² = 18.7%) (Supplementary Appendix S2; Supplementary Figure S2).

The PT showed a significant increase (p = 0.022) of lower extremity power with small effect (ES = 0.50; 95% CI = 0.07–0.93; Egger’s test p = 0.734; n = 115) and lower heterogeneity (I² = 19.9%) (Supplementary Appendix S2; Supplementary Figure S3).

The PT showed a non-significant change (p = 0.115) in lower extremity muscle strength with small effect (ES = 0.30; 95% CI =-0.07–0.68; Egger’s test p = 0.471; n = 108) and lower heterogeneity (I² = 0.0%) (Supplementary Appendix S2; Supplementary Figure S4).

The PT showed a significant increase (p < 0.001) of agility with moderate effect (ES = 0.88; 95% CI = 0.53–1.23; Egger’s test p = 0.060; n = 105) and lower heterogeneity (I² = 6.42%) (Supplementary Appendix S2; Supplementary Figure S5).

Five studies contained in this review evaluated maximal serve velocity. Among these studies, ball speed measurement by radar gun (Gelen et al., 2012; Behringer et al., 2013; Ölçücü et al., 2013; Fernandez-Fernandez et al., 2016; Fernandez-Fernandez et al., 2018). Two RCTs measured this skill aspect. Gelen et al. (2012) compared EG3 and CG employing high-volume upper body PT vs. normal tennis practice. They discovered a significant (p = 0.001) acute effect in maximal serve speed. Behringer et al. (2013) conducted an 8-week PT for the upper and lower bodies and increased the maximum serve velocity in a statistically significant (p < 0.05) way. Three non-RCTs studied maximal serve velocity. Fernandez-Fernandez et al. (2016) found a significant increase (p < 0.05) in maximal serve velocity after an 8-week upper and lower body PT program. Meanwhile, in another study, the comparison of PT versus a traditional training regimen indicated positive effects on maximal serve velocity (p < 0.01) (Ölçücü et al., 2013). In addition, the report by Fernandez-Fernandez et al. (2018) demonstrates that the EG1 significantly improved the maximal serve velocity. In contrast, the EG2 did not observe any changes.

One RCT (Behringer et al., 2013) and one non-RCT (Fernandez-Fernandez et al., 2016) measured serve accuracy. The accuracy of the service is scored by calculating the position of the ball landing in the specified target area. The RCT reported neither positive nor adverse impacts have been observed on service accuracy scores after an 8-week PT program (Behringer et al., 2013). Interestingly, Fernandez-Fernandez et al. (2016) reported a significant change (p < 0.01) in serve accuracy after a similar 8-week PT program.

Five of the studies (one RCT and four non-RCTs) assessed sprint speed. The sprint tests used in these investigations comprised the linear sprint test of 4 m (Salonikidis and Zafeiridis, 2008), 5 m (Fernandez-Fernandez et al., 2016), 10 m (Fernandez-Fernandez et al., 2015; Fernandez-Fernandez et al., 2016; Fernandez-Fernandez et al., 2018), 12 m (Salonikidis and Zafeiridis, 2008), 20 m (Fernandez-Fernandez et al., 2015; Fernandez-Fernandez et al., 2016; Fernandez-Fernandez et al., 2018), 30 m (Fernandez-Fernandez et al., 2015), 50 m (Mohanta et al., 2019); lateral sprint test of 4 m (Salonikidis and Zafeiridis, 2008); and change of direction sprint test of 12 m with turn (Salonikidis and Zafeiridis, 2008), repeated sprint ability (Fernandez-Fernandez et al., 2015).

The RCT reveals that it significantly improves (p < 0.05) 12 m sprint without turn performance in the combined group (PT + regular tennis drills) and PT alone group. Meanwhile, the performance in the 12 m sprint with turn was improved only in the EG2 and failed to reach significant differences in the EG1 (Salonikidis and Zafeiridis. 2008). Three 8-week non-RCTs revealed improvements (p < 0.05) in the linear sprint test (Fernandez-Fernandez et al., 2016; Mohanta et al., 2019) and repeated sprint ability test (Fernandez-Fernandez et al., 2015). Of note, Fernandez-Fernandez et al. (2015) reported that the 10 m test was significantly improved (p < 0.05), in contrast to the 20 m test and 30 m test. Fernandez-Fernandez et al. (2018) reported that the EG1 had a positive effect on pre-to post-test measures of the sprint, while the EG2 training method had negative effects on 5 m, or trivial effects on 10 m and 20 m.

Two non-RCTs measured upper extremity power via medicine ball throw tests (Fernandez-Fernandez et al., 2016; Fernandez-Fernandez et al., 2018). Fernandez-Fernandez et al. (2016) compared EG and CG applying an 8-week PT program vs. normal tennis practice. They discovered that MBT performance had improved significantly (p < 0.05). Fernandez-Fernandez et al., 2018 observed a considerable increase in MBT following the EG1 training method, but no improvement in EG2.

One RCT (Salonikidis and Zafeiridis, 2008) and four non-RCTs (Fernandez-Fernandez et al., 2015; Fernandez-Fernandez et al., 2016; Fernandez-Fernandez et al., 2018; Mohanta et al., 2019) measured lower extremity power. This outcome test consisted of CMJ test (Fernandez-Fernandez et al., 2015; Fernandez-Fernandez et al., 2016; Fernandez-Fernandez et al., 2018), drop jump test (Salonikidis and Zafeiridis, 2008), SLJ test (Fernandez-Fernandez et al., 2016), VJ test (Mohanta et al., 2019). Salonikidis and Zafeiridis (2008) found that EG2 activities had a significant gain (p < 0.05) in DJ height when compared to EG1. Fernandez-Fernandez et al. (2016) demonstrate that EG vs. CG significantly enhanced (p < 0.01) CMJ and SLJ performance. Fernandez-Fernandez et al. (2015) compared EG and CG using an 8-week combination of explosive strength and repeated sprint training. The results indicate that CMJ achieved a positive effect. Fernandez-Fernandez et al. (2018) reported data on EG1 vs. EG2, and it is noted that the EG1 training approach markedly improved (p < 0.05) the CMJ performance, in contrast to the EG2. Furthermore, Mohanta et al. (2019) showed that 8 weeks of PT could significantly improve VJ height (p = 0.027).

One RCT (Behringer et al., 2013) and two non-RCTs (Ölçücü et al., 2013; Mohanta et al., 2019) measured upper extremity strength. The tests applied at this point involved the 1RM and 10RM chest press test (Behringer et al., 2013; Mohanta et al., 2019), and the torque of the shoulder joint test (Ölçücü et al., 2013). The RCT (8-week PT program) conducted by Behringer et al. (2013) observed a greater increase (p < 0.05) in upper extremity strength (10RM chest press test) when compared to the control group. Two similar 8-week PT programs (non-RCT design) also found a statistically significant difference (p < 0.05) in upper limb strength (1RM chest press and torque of the shoulder joint test) measurement between the experimental group’s pre-and post-test (Ölçücü et al., 2013; Mohanta et al., 2019).

The lower extremity strength test consists of a 10RM leg press test (Behringer et al., 2013), a maximum isometric force test (leg) (Salonikidis and Zafeiridis, 2008), and a torque of the knee joint test (Ölçücü et al., 2013). Two RCTs indicated that tennis players who trained with 8 weeks of PT had considerably larger increases in lower extremity strength (p < 0.05) than the control group (Salonikidis and Zafeiridis, 2008; Behringer et al., 2013). One non-RCT also achieved a positive effect in the strength test (Ölçücü et al., 2013) after 8-week PT.

In the twelve studies that were reviewed, agility was reported in seven. The factors valued and assessment instruments used included the 5-0-5 test (Fernandez-Fernandez et al., 2016; Fernandez-Fernandez et al., 2018), T-test (Mohanta et al., 2019; Hotwani et al., 2021), Illinois test (Rathore, 2016; Lakshmikanth et al., 2018; Hotwani et al., 2021), tennis-specific agility test (Lakshmikanth et al., 2018), hexagon test and spider test (Ziagkas et al., 2019). Two RCTs (Rathore, 2016; Ziagkas et al., 2019) show that EG vs. CG substantially increased agility performance (hexagon test, spider test, Illinois test; p < 0.0001). Among non-RCTs, a statistically significant difference between the pre-and post-test in several agility tests (i.e., 5-0-5 test, T-test, Illinois test, tennis-specific agility test) was revealed by the EG of studies (Fernandez-Fernandez et al., 2016; Lakshmikanth et al., 2018; Mohanta et al., 2019). Moreover, Hotwani et al., 2021 revealed that 3 weeks of combining plyometric with sprint training sessions resulted in an extremely significant (p < 0.0001) gain in agility (Illinois test and T-test). However, Fernandez-Fernandez et al. (2018) discovered positive effects from pre-test to post-test measures on agility (5-0-5 test) in the EG1. On the contrary, the agility test for the EG2 showed negative impacts.

One RCT (Salonikidis and Zafeiridis, 2008) and one non-RCT (Fernandez-Fernandez et al., 2015) evaluated reaction time and aerobic endurance performance, respectively.

The reaction time was measured based on the load cells. Salonikidis and Zafeiridis 2008 reported outcomes on EG1 vs. EG2; both PT regimens induce favorable changes to reaction time performance (p < 0.05). Another study using the 30–15 interval fitness test evaluated aerobic endurance performance, and the study reported no significant improvement in this aspect of factor (Fernandez-Fernandez et al., 2015).

It is worthy to note that one out of the twelve studies reported that eight players in the control group and one in the PT group were dropped from the final collection of data because of acute injuries (i.e., ankle sprain) (Fernandez-Fernandez et al., 2016). Apart from that, no other studies in this review reported fatigue, soreness, injury, pain, damage, or adverse effects associated with PT intervention.

The tennis serve is undoubtedly one of the most challenging tennis shots to master, but it can significantly contribute to a point win or advantage (Guillot et al., 2013). Based on our meta-analysis, PT interventions were found to have a moderate effect (ES = 0.75) on maximal serve velocity in tennis players. The serve is a complicated stroke that involves a sequence of forces (legs, trunk, and arm/racquet) moving from proximal to distal (Elliott, 2006). During a serve, it is crucial to transfer power from the legs, trunk, and arm to the ball as quickly as possible to maximize the velocity (van den Tillaar. 2004). According to Ferrauti and Bastiaens, (2007), service velocity results from an efficient force transfer along a convoluted kinetic chain dependent on intermuscular coordination and muscle strength. Simultaneously, PT helps to increase muscular strength and intermuscular coordination, so that force transfer during service can be improved (Fernandez-Fernandez et al., 2016). Previously, resistance training has been recommended to increase ball velocity in tennis (Kraemer et al., 2000; Kraemer et al., 2003); yet, a study in the current review specifically compared the effects of PT and resistance training on this factor and found that serve speed improvement was significantly higher in the PT group than in the resistance training group (Behringer et al., 2013). In addition, Fernandez-Fernandez et al., 2013 suggested that upper-extremity plyometric exercises can be frequently used by athletes in the pursuit of more powerful functional performance. In the present review, of the five studies measuring serve speed, four designed upper-body plyometrics in the training protocols (Supplementary Appendix S4), therefore, the increase in maximal serve velocity was expected.

In tennis, the main determinants of shot quality include not only ball velocity but also ball placement accuracy (Kovacs and Ellenbecker, 2011). Fernandez-Fernandez et al. (2016) speculated that the improved kinetic chain as a result of the power gains from the PT would contribute to the stability of tennis players from a technical perspective, leading to a higher accuracy test score. However, another study found that similar plyometric exercises did not affect serve accuracy (Behringer et al., 2013). In the literature, service accuracy was also unaffected by other forms of strength training (Ferrauti and Bastiaens, 2007; Fernandez-Fernandez et al., 2013). Terraza-Rebollo and Baiget. (2020) explained in their study that it was likely because subjects were instructed to hit the ball at maximum speed, so it was difficult to take into account the accuracy. However, there is some controversy because it has also been observed a positive correlation between velocity and accuracy in tennis (Cauraugh et al., 1990).

Therefore, PT is an effective method that may have greater chances of improvement in tennis players’ maximal serve velocity. Concerning serve accuracy, it was impossible to reach a firm conclusion, because only two studies examined the effect of PT on this point. Furthermore, more research is required to explore the relationship between training-induced increases in serve velocity and accuracy.

Better sprint performance will enable tennis players to get to the ball faster and will give them more time to prepare for the shot (Kramer et al., 2021). The meta-analysis, summarizing the effects of PT on the sprint speed performance of tennis players, showed a small but significant effect across studies (ES = 0.43). Increases in sprint performance following PT may be due to increased neuromuscular activation of the exercised muscles (Hakkinen and Komi, 1985). Moreover, these adaptations induced by lower body plyometric exercises (e.g., sprinting, hopping, jumping), such as improving muscle-tendon stiffness and increasing neural drive to agonist muscles (Markovic and Mikulic, 2010) may improve SSC efficacy. As a result of gains in SSC efficacy in lower limb musculature, stronger force generation likely occurs in the concentric action phase after a rapid eccentric muscle movement (Markovic and Mikulic., 2010; Radnor et al., 2018), which is a key requirement for enhancing tennis players’ sprint performance. However, one study found that PT had no effect on a 12 m sprint with a turn test (Salonikidis and Zafeiridis, 2008). This observation may have resulted from the activity being conducted only on one leg (Salonikidis and Zafeiridis, 2008). Furthermore, Fernandez-Fernandez et al. (2015) concluded that a PT program did not result in significant gains in the 20 m and 30 m sprint measurements. This discovery might be connected to the athletes’ stride frequency and the coordination of the lower limb muscles (Young et al., 2001). In fact, only a few studies in the literature did not report the positive effects of PT on sprint performance (Oxfeldt et al., 2019). For example, Hammami et al. (2019) investigated the impact of a night-week PT program on young female handball players. There were no changes reported for the 5 m and 10 m sprint times, which could be explained by the fact that initial acceleration (over 5 and 10 m) has proven to be more difficult to improve than maximal velocity, most likely due to the smaller margin for improvement and the different forces involved. Interestingly, Hammami et al. (2020) found inconsistent results on young female handball players, that there were significant increases in speed over distances of 5–30 m after 10 weeks of similar plyometric exercises. Differences between PT programs (e.g., frequency, duration) may help to explain the different magnitudes of physical fitness changes among studies (Ramirez-Campillo et al., 2020b). However, the paucity of available data limits our attempts to explore the effects of these factors on training effects. Therefore, although we have found evidence to support the use of PT as an effective training modality to improve sprint performance, more high-quality research on sprint speed performance is still needed.

Power development is paramount, irrespective of the sport and proportion of each energy system engaged, since certain critical actions are executed as quickly and forcefully as possible (Elliott et al., 2007). On the one hand, upper extremity power was studied in two studies included in this review, and this variable was examined using medicine ball throws. These outcomes showed positive effects (Fernandez-Fernandez et al., 2016; Fernandez-Fernandez et al., 2018). Ulbricht and others discovered that the upper extremity power test (MBT) was an important predictor of tennis performance (Ulbricht et al., 2016). These two studies in our review incorporating PT of the upper limb involving multiple medicine ball exercises could be an important factor in improving upper extremity power. Furthermore, PT works by utilizing the natural elastic components of muscles and tendons, and stretch reflexes, to increase the power of subsequent movements (Trajkovic et al., 2016).

On the other hand, a meta-analysis of lower extremity power (CMJ) was undertaken, and the PT indicated a significant increase with a small effect (ES = 0.50). Such effects were noted for participants with a wide range of sports backgrounds. A meta-analysis on individual sports athletes was done by Sole et al. (2021), and the results showed similar improvements in the CMJ performance of ES = 0.49. In short, factors including improved motor unit recruitment, increased intermuscular coordination, increased neural drive to the muscles of the agonist, and improved SSC utilization are responsible for jump performance improvements (Markovic and Mikulic, 2010; Taube et al., 2012). Of note, a study conducted by Fernandez-Fernandez et al. (2018) showed that performing NMT before regular tennis practice produced a substantially significant positive effect on CMJ. Conversely, when NMT was conducted after regular tennis drills, there was no improvement in CMJ assessed. The previous tennis regular practice that resulted in acute fatigue may have contributed to the lack of a marked increase in CMJ (Garcia-Pallares and Izquierdo, 2011). From this, the optimal design and implementation of training strategies are significant for tennis players. This should be taken into account in future research.

Overall, our findings imply that PT positively affects tennis players’ lower extremity power, and future research should consider further research on upper extremity power.

For successful athletic performance, strength is essential (Slimani et al., 2016). Moreover, it has been reported that a tennis player’s upper and lower body strength can be very helpful in preventing injuries (Ellenbecker and Roetert, 2004). Of the four studies included in this review assessing strength, three reported upper body strength. These studies showed significant improvement in upper extremity strength tests (Ölçücü et al., 2013, Behringer et al., 2013; Mohanta et al., 2019).

Meanwhile, it has been found in the present meta-analysis that PT led to small gains in lower extremity muscle strength (ES = 0.30), but this was not statistically significant. In contrast, de Villarreal et al. (2010) observed that PT has a significant, moderate effect on lower limb muscle strength (ES = 0.97). Theoretically, PT increases cross-bridge mechanics, activates motor units, enhances neural efficiency, and provides passive tension to the muscle-tendon complex, contributing to strength performance (Malisoux, 2005; Ramírez-Campillo et al., 2015). Moreover, improvements in muscle strength after PT may also be associated with muscle hypertrophy (Grgic et al., 2020).

Additionally, the narrative synthesis in our review showed a positive direction of evidence for muscle strength, but the certainty of the evidence was very low to low. As a result, these findings are unclear, though this could be due to the relatively few studies conducted in this field. Likewise, although our meta-analysis results do not show a significant effect from PT on lower extremity muscle strength, numerous previous studies have concluded that PT is an efficient training technique to enhance strength performance in other ball players, such as basketball players (Ramirez-Campillo et al., 2021a), handball players (Hammami et al., 2020). Therefore, to confirm the effects of PT on tennis players, further high-quality studies simultaneously evaluating the influence of upper and lower body muscle strength are needed to draw more solid conclusions.

The basic requirement for a tennis player is the ability to quickly switch between multidirectional movements (e.g., vertical, lateral, forward, backward) (Rathore, 2016). In the present review, one of the most studied physical attributes is agility (seven papers). The findings in the current meta-analysis show that PT has a significant, moderate effect (ES = 0.88) on agility. Similar improvement in agility performance was observed by Asadi et al. (2016) and Thapa et al. (2021), supporting the findings of our study. The present research evaluated agility performance using a variety of tests (5-0–5 test, T-test, Illinois test, Heroxge test, tennis-specific agility test). Of note, Asadi et al. (2016) suggested that PT could improve change-of-direction ability (ES = 0.26–2.8, small-to-large), depending on the types of plyometric exercises and change-of-direction tests. The current results are consonant with these findings, with extensively positive results in all of the agility tests after PT.

Additionally, Fernandez-Fernandez et al. (2018) found positive effects from pre-test to post-test analyzes on agility in the EG1. In contrast, the EG2 observed negative effects on agility. As aforementioned, the fatigue of regular tennis training may have contributed to the decline in agility. In the literature, it has been well identified that PT is a time-efficient, effective and simple method for improving agility among athletes (Asadi, 2013; Arazi et al., 2014; Ramírez-Campillo et al., 2014). Indeed, PT helps reduce ground contact by boosting muscular force output and movement efficiency, which is related to agility improvement (Markovic and Mikulic, 2010; Asadi et al., 2016). Moreover, this type of training approach could have increased eccentric strength in the legs, allowing athletes better to switch between deceleration and acceleration movements (Brughelli et al., 2008). Meanwhile, plyometric exercises which included powerful multidirectional movements helped improve the ability to change directions rapidly (Söhnlein et al., 2014). Therefore, the findings suggest that PT might assist tennis players to enhance their agility performance.

The tennis player’s ability to react quickly in response to their competitor’s movements is crucial during a game (Reid et al., 2013). Reaction time performance has only been investigated in one study, and this study reported significant improvement in the reaction time test (Salonikidis and Zafeiridis, 2008). Similar results were found in a study done by Turgut et al. (2019) which showed a significant improvement in reaction time as a result of PT. However, discussion on the mechanisms related to improved reaction time in athletes after PT remains unclear, with extensive empirical research required to elucidate such mechanisms.

Likewise, aerobic endurance is a major component of physical fitness (Wezenberg et al., 2013), and can be considered one of the benefits of PT. However, only one paper examined aerobic endurance in the current review, and this study found no significant improvement in the aerobic endurance test (30–15 interval fitness test) (Fernandez-Fernandez et al., 2016). This may be due to repetitive short-duration exercises which induced changes in glycolytic enzymes, muscle buffering, and ion regulation and led to improved anaerobic capacity (Dawson, 2012) rather than aerobic performance.

Based on two articles, we cannot get more information from these two variables, thus, in the future, more studies evaluating the effects of PT on reaction time and aerobic endurance are required to reach more reliable conclusions.

This systematic review can be regarded as having some limitations, which should be taken into account. The first limitation is that only 12 publications were analyzed, and there were insufficient RCTs (n = 5) included in this study. Secondly, there is no study specifically analyzing female tennis players in this review, which limits our comprehension of the overall effectiveness of PT in tennis players. This presents a crucial gap in the study, and should be addressed in further studies. Thirdly, three studies were excluded from the current meta-analysis because they lacked control group data or did not offer enough. In addition, several outcomes (such as serve accuracy, upper extremity power and strength) were excluded from our meta-analysis, since there was an insufficient number of studies involved. Fourthly, several studies did not give a comprehensive description of the training program; for example, three of the included publications did not include details regarding the training protocol. Fifthly, the majority of the studies included in this review had small sample sizes (8–30 people per EG), and only one research disclosed the sample size calculation technique (Mohanta et al., 2019), meaning statistically significant results may not reflect the actual effect. As demonstrated by GRADE, the certainty of the evidence ranged from very low to moderate for most of the study outcomes, lowering confidence in the reported estimates.