Introduction

“You need to be constantly playing well throughout five hours if you want to win a match like this. I guess there is an endurance part.” (Novak Djokovic about his win in 2019 over Roger Federer in the longest Wimbledon final in history (Amako, 2019)). This quote reflects the importance of maintaining high performance in tennis up to the end of every match. Even more, it reflects the importance of maintaining high performance under physically demanding situations, in which decrements of performance might be expected. Tennis performance is directly related to a player’s technical skills. They are crucial for discriminating tennis players of various performance level (Baiget, Fernandez-Fernandez, Iglesias, Vallejo, & Rodriguez, 2014; Baiget, Iglesias, & Rodriguez, 2016; Del Villar, Garcia Gonzalez, Iglesias, Perla Moreno, & Cervello, 2007; Kolman, Huijgen, Kramer, Elferink-Gemser, & Visscher, 2017; Lyons, Al-Nakeeb, Hankey, & Nevill, 2013; Vergauwen, Spaepen, Lefevre, & Hespel, 1998; Vergauwen, Madou, & Behets, 2004). Though the importance of technical skills is undisputed, it seems especially important to persevere technical performance throughout the match. Also, under the prolonged physical demands towards the end of matches players need to retain their technical skills at a high level (Kovacs, 2004; Kovacs, 2007). Measuring technical skills under increasing physical intensity is difficult and no tests so far captured all relevant components.

Tennis is a sport with an intermittent profile (Kovacs, 2006). The aerobic and anaerobic energy demands are high throughout these intermittent profiled matches (Bangsbo, 1994; Kovacs, 2007). During long and high-intensity rallies, heart rates of 70-80% of the maximum heart rate and peak heart rates of 100% of the maximum heart rate are elicited (Baiget et al., 2014). The aerobic and anaerobic capacity of a tennis player may therefore highly influence individual performance and match outcome. Consequently, the monitoring of technical and physical skills under increasing physical intensity in a tennis-specific setting is valuable for determining a tennis player’s performance level.

Technical skills may be evaluated based on stroke analysis (which captures the mechanical aspects of strokes), as well as based on stroke outcomes (such as stroke accuracy and stroke velocity). Further on, we continue using the term technical skills to describe the stroke outcomes. Technical tennis skills that are most often studied are stroke velocity and stroke accuracy. Stroke velocity (Kolman et al., 2017; Landlinger, Stöggl, Lindinger, Wagner, & Müller, 2012; Vergauwen et al., 1998; Vergauwen et al., 2004) and stroke accuracy (Baiget et al., 2014; Baiget et al., 2016; Del Villar et al., 2007; Kolman et al., 2017; Lyons et al., 2013; Vergauwen et al., 1998; Vergauwen et al., 2004) discriminate between players of different performance levels with better players reaching higher stroke velocities and being more accurate. Not surprisingly, a combination of stroke velocity and stroke accuracy, the VA-index also distinguishes tennis players of different performance levels (Kolman et al., 2017; Vergauwen et al., 1998). Moreover, higher ranked tennis players make fewer stroke errors than their lower ranked counterparts (Kolman et al., 2017; Vergauwen et al, 1998). Therefore, the proposed set of technical skills should contain stroke velocity, stroke accuracy, the VA-index and percentage of errors (PE).

The development of a player’s tennis performance is influenced by individual, environmental and task constraints (Newell, 1986). Over time these constraints may change and therefore influence tennis performance. Especially within youth athletes, in which individual constraints change rapidly due to changes in biological development and training (Malina, Rogol, Cumming, Coelho e Silva, & Figueiredo, 2015), improvements but also stagnation in tennis performance are quite common. For instance, physical fitness has a uniform rapid development for male youth tennis players, whereas the development flattens out at the age of 14-15 years old for female youth tennis players (Faff, Ladyga, & Starczewska-Czapowska, 2000; Kramer, Huijgen, Elferink-Gemser, & Visscher, 2016). Although studies measuring technical skills over age are scarce, it has been established that forehand stroke velocity and backhand stroke velocity are positively correlated with age for male youth tennis players (González-González, Rodríguez-Rosell, & Clavero-Martin, 2018). Still, the lack of information regarding the development of the remaining technical skills over age shows the need for new studies. Therefore, capturing the differences of physical and technical skills between male and female tennis players over age will be an important topic within this study.

The interaction between a tennis player and its opponent makes individual tennis performance difficult to determine in matches, as the opponent influences a player’s performance as well. Therefore, to capture a player’s performance at a certain moment or to monitor progress over time, fixed field test protocols are valuable to measure tennis skills (Kolman et al., 2017; Landlinger et al., 2012; Lyons et al., 2013; Vergauwen et al., 1998; Vergauwen et al., 2004). Fixed field test protocols are essential to compare a player to another player or with performance bench-marks for performance discrimination purposes. Monitoring technical skills under increasing physical intensity is important for determining a tennis player’s performance level in tennis. Test protocols including a tennis-specific execution of strokes in incremental stages to induce higher intensities and fatigue have been developed, such as the modified specific endurance field test and the TEST-protocol (Baiget et al., 2014; Brechbuhl, Girard, Millet, & Schmitt, 2016; Smekal et al., 2000). Yet, this specific endurance field test protocol lacks rest intervals given the intermittent character of tennis (Baiget et al., 2014) and the technical performance of strokes is no outcome measure of interest in the TEST protocol (Brechbuhl et al., 2016). Additionally, in these test protocols oncoming ball direction was alternately between the forehand and backhand side of the participant, leading up to a back-and-forth running protocol between the sidelines of the court. Such a back-and-forth running protocol is neglecting the importance of tennis-specific footwork between the strokes. Randomization of oncoming ball direction could solve this problem.

In order to successfully monitor technical tennis skills under increasing physical intensity in a maximized tennis-specific setting, a suitable field test protocol is required. Therefore, the TEST-protocol and specific endurance field test protocol (Baiget et al., 2014; Brechbuhl et al., 2016) will form the basis, and will be modified with respect to abovementioned required adjustments. The current study introduces the new Physical Technical Tennis-specific Field test (PTTF-test) protocol and aims to determine to what extent the PTTF-test is able to capture fluctuations in technical skill under increasing physical intensity in adolescent talented tennis players. The study aims to capture decreasing technical performance under increasing physical intensity. In addition, we will compare performance in the PTTF-test between age categories and between boys and girls. We hypothesize that the PTTF-test succeeds in capturing decreasing technical performance under increasing physical intensity for adolescent talented tennis players.

Materials and methodology

Participants

The minimum age for participation was 12 years old. Participants were excluded in case of a health condition or injury that would make them unable to perform a physical test, or if they were not allowed by the trainer or medical staff. A total of 21 male tennis players (age 15.5 ± 2.5 yrs) and 19 female tennis players (age 15.2 ± 2.2 yrs) from the Netherlands participated in this study. They were considered elite players, as they were the nationally highest ranked performers within their age categories: U14, U16 and 16+ (see also Table 1).

Before the measurements, participants and their parents were fully instructed on the objective, design, method and risks of the test by an information letter. Both the participants and parents (if the participant was under the age of 16) provided written informed consent prior to inclusion within the study. The local ethical committee of the psychology department of the University of Groningen approved the study (PSY-1819-S-0262) that was performed consonantly to the ethical standards derived from the Declaration of Helsinki.

PTTF-test set-up

The measurements for this study took place on a hardcourt indoor tennis court. This court has a Playsight Smartcourt for video-review and analytics and as such is equipped with 10 on-court cameras. The system allows for the valid registration of stroke velocity and ball placement and the registration of session video material. To measure stroke accuracy, target areas were presented on-court to which participants were instructed to direct their strokes. Two target area’s (squared plastic of 100.6 x 68.7 cm) were placed near the intersections of the baseline with both singles sidelines, precisely 100.6 cm from the baseline and 68.7 cm from the sidelines, as presented in Figure 1. Set-up of the target area’s was based on the D4T test (Kolman et al., 2017), but exact dimensions were slightly adjusted to the pre-specified dimensions of the line system that is available in the Playsight Smartcourt software. During the PTTF-test protocol the participants were fed with regular tennis balls (Dunlop Fort Max TP) by a manually programmed ball machine (Promatch SmartShot Xtra). The ball machine was located in the middle of the court between the singles sidelines and 100.6 cm behind the baseline, as shown in Figure 1. Participants used their own tennis racket during the test protocol and wore a pulse monitoring belt (Polar Team2 Pro) for heart rate registration.

PTTF-test protocol

Prior to the PTTF-test protocol, the participants performed a 5-min specific warming-up session, including running, arm rotations, lunges, squat jumps, sprints and 30 practice strokes fed by the ball machine. This warming-up session was created by and performed under the supervision of physical staff members. In the PTTF-test protocol the participants were fed with oncoming balls by the ball machine and instructed to hit both forehand and backhand strokes crosscourt towards the presented targets in an incremental physical intensity level design. Participants were instructed to hit balls with “the best possible velocity-accuracy ratio”, whereby slice strokes were not permitted. Physical intensity levels consist of 60 s activity, which are alternated with 30 s rest, as adopted from the TEST protocol (Brechbuhl et al., 2016). This rest period closely approaches the 25 s intervals between points in a tennis match. Oncoming ball frequency at the first intensity level is 12 strokes.min-1. Physical intensity increases for every new level of 60 s. Oncoming ball frequency increases with 2 strokes.min-1 up to 18 strokes.min-1 at level 4. From level 5 and on, oncoming ball frequency increases with 1 stroke.min-1 per level. Oncoming ball direction, velocity and height are randomised within the intensity levels for the participants to evoke tennis-specific footwork between the strokes. Oncoming ball velocity was based on the D4T test and further adjusted during trials to induce more variability (Kolman et al., 2017). Eventually, oncoming ball velocity in the PTTF-test is randomised within the range of 60 and 80 kph .The randomised pattern varies per intensity level as well, to prevent for habituation of the pattern by the participant. The same standardized pattern of oncoming balls is presented to all par-ticipants to secure a standardized test protocol. The PTTF-test protocol ends with a participant’s voluntary exhaustion or when the participant is not able to hit two subsequent oncoming balls with an adequate technique as determined by experienced coaches (hit the ball with the racket snares instead of the racket frame and hit the ball in a controlled manner in the direction of the net) or when the participant fails in striking three oncoming balls in total within a given intensity level.

Outcome variables PTTF-test

The determination of physical responses by Polar heart rate measurements has been commonly used throughout activity (Baiget et al., 2014; Baiget, Fernandez-Fernandez, Iglesias, Vallejo, & Rodriguez, 2015). This method for the determination of physical responses is a valuable tool in order to secure the ideal rest interval of tennis through the absence of any additional measurements during the rest intervals. The Polar pulse monitoring belts allow for the registration of participants’ heart rate during the PTTF-test as a measure for physical intensity. The continuous registered heart rate data are used to calculate mean heart rate for every intensity level that the participant performed. PTTFlevel is the outcome measure of step-out moment from the test protocol.

For technical performance measurements, Playsight Smartcourt allows for the registration of stroke velocity and ball placement for every performed stroke within the PTTF-test. For stroke accuracy, strokes are awarded with a score of 1, 3, 6 or 9 points based on ball placement location as presented in Figure 1. Balls that landed outside of these point awarding zones, but inside the singles tennis field are awarded with a score of 0 points. Finally, strokes that landed outside the singles tennis field or strokes that ended up in the net were given a score of -1 points. With all data it was possible to calculate the outcome variables stroke velocity, stroke accuracy, the VA-index, PE for every intensity level that was performed by the participant. The VA-index is a validated combination score of stroke accuracy and stroke velocity in which stroke velocity is squared and therefore more rewarded due to the harder increment on higher scores (Kolman et al., 2017; Vergauwen et al., 1998). The VA-index is calculated by the following formula whereby stroke velocity is expressed in kph:

PE is expressed by the number of errors as a percentage of the total strokes within a level. Strokes in the net, off-court or sliced strokes are counted as errors.

Figure 1. PTTF-test protocol set-up including the dimensions and location of the targets and the point awarding for stroke accuracy based on ball bounce location.

Statistical Analysis

Outcome variables are compared between situations with low physical intensity (LOW-Phys.Int.) and with higher physical intensity (HIGH-Phys.Int.). The ‘BASELINE’-situation (LOW-Phys.Int.) is represented by the mean value of the outcome variables from intensity level 2 and intensity level 3 for every participant. Intensity level 1 is not included within the ‘BASELINE’-situation, as it is a way for the participants to conform with the test protocol. The ‘FINAL’-situation (HIGH-Phys.Int.) is participant-dependant and represented by the mean value of the outcome variables from the final two intensity levels that the participant performed during the PTTF-test.

The statistical analysis was executed with the software IBM SPSS statistics 23.0. The accuracy scores inter-rater reliability was determined with Cohen’s kappa and by the examining of four full test protocols by two observers. Cohen’s kappa for inter-rater reliability showed an excellent consensus between two observers for accuracy score determination based on ball bounce location, κ = .85 (p<.001).

At first, Analysis 1 was performed to compare the mean heart rates, PTTFlevel and the technical skills in both intensity situations between age categories and genders. A MANOVA analysis with age category and gender as grouping factors and with heart rate, stroke velocity, stroke accuracy, VA-index and percentage errors of both intensity situations and PTTFlevel all as dependant variables was performed as a basic comparison between groups. Analysis 2 assessed the increase in physical intensity between these situations by mean heart rate applying a repeated measures ANOVA with grouping factors age category and gender, and with the ‘BASELINE’-situation and the ‘FINAL’-situation as repeated measures. Analysis 3, a repeated measures MANOVA with grouping factors age category and gender, and with the ‘BASELINE’-situation and the ‘FINAL’-situation as repeated measures, was used to assess the effect of physical intensity on the set of outcome variables mean stroke velocity, mean stroke accuracy, the VA-index, and percentage errors. Follow-up univariate analyses are performed to check for the effect of physical intensity on the outcome variables separately. The combination of analysis 2 and analysis 3 will be used to determine the criterion validity of the PTTF-test for capturing decreasing technical performance under increasing physical intensity.

Checks for the normality of distribution, independence of measures, the homogeneity of covariance matrices and sphericity are performed. Sphericity is checked by Mauchly’s test. In the case that sphericity assumption is violated, a Greenhouse-Geisser correction is used to adjust. The Bonferroni correction is performed for all post-hoc testing. Effect sizes in the data analysis are classified as large (η2>0.14), medium (0.06<η2<0.14) or small (0.01<η2<0.06) (Cohen, 1988). For all data analysis the level of significance is set at an alpha of 5%.

Results

Participants characteristics, and the PTTFlevel (‘FINAL’-situation) for each age category and divided by gender, are presented within

Table 1. Participants characteristics (mean ± SD) and their final reached intensity level (PTTFlevel) per age category group, divided by males (♂, n=21) in 1a. and females (♀, n=19) in 1b.

Analysis 1: comparison of the PTTF outcome variables in BASELINE and FINAL intensity situations between age categories and gender

The mean heart rates, mean stroke velocity, mean stroke accuracy, VA-index and PE in the ‘BASELINE’ and ‘FINAL’ situation and PTTFlevel are presented in Table 2, divided by age category and gender. The MANOVA found a multivariate Age category effect (F(22,48) =3.328, p=.000; Wilks’ Lambda = 0.157, partial η2 = 0.604). Univariate testing showed that the Age category effect was apparent for PTTFlevel (F(2,34) = 4.341, p=.021; partial η2 = 0.203), mean heart rate in the ‘BASELINE’-situation (F(2,34) =5.468, p=.009; partial η2 = 0.243), mean stroke velocity in the ‘BASELINE’-situation (F(2,34) =8.139, p=.001; partial η2 = 0.324), VA-index in the ‘BASELINE’-situation (F(2,34) =10.700, p=.000; partial η2 = 0.386), mean heart rate in the ‘FINAL’-situation (F(2,34) =6.958, p=.003; partial η2 = 0.290), mean stroke velocity in the ‘FINAL’-situation (F(2,34) =7.311, p=.002; partial η2 = 0.301) and PE in the ‘FINAL’-situation (F(2,34) =7.538, p=.002; partial η2 = 0.307). The results derived from the subsequent Bonferroni post-hoc testing described the differences between the groups as illustrated in Table 2. No Gender effect or Age category x Gender interaction effect was revealed.

Analysis 2: The increase of physical intensity in the PTTF-test

The Repeated Measures ANOVA results regarding heart rate are presented in Table 3. Mean heart rate was significantly higher in the ‘FINAL’-situation (190.5 ± 8.1) than in the ‘BASELINE’-situation (174.1 ± 11.5); F(1,34) = 140.269, p=.000; Wilks’ Lambda = 0.195, partial η2 = 0.805. Furthermore, no significant interaction effects were found (Table 3). The mean heart rates in both intensity situations are illustrated in Table 2 for all age categories and divided by gender.

Analysis 3: The effect of physical intensity on technical performance

The repeated measures MANOVA found a signi-ficant effect of Intensity situation (F(4,31) = 75.931, p=.000; Wilks’ Lambda = 0.093, partial η2 = 0.907) and Intensity situation x Age category interaction effect (F(8,62) = 2.441, p=.023; Wilks’ Lambda = 0.578, partial η2 = 0.240).

Further univariate analyses are presented in Table 4 and illustrate that mean stroke velocity was significantly lower in the ‘FINAL’-situation (94.3 ± 7.9) than in the ‘BASELINE’-situation (101.6 ± 8.1); F(1,34) = 100.310, p=.000; partial η2 = 0.747. Secondly, mean stroke accuracy was significantly lower in the ‘FINAL’-situation (1.6 ± 0.6) than in the ‘BASELINE’-situation (2,6 ± 0,5); F(1,34) = 144.388, p=.000; partial η2 = 0.809. Furthermore, VA-index was significantly lower in the ‘FINAL’-situation (16.1 ± 6.4) than in the ‘BASELINE’-situation (30.5 ± 7.4); F(1,34) = 278.325, p=.000; partial η2 = 0.891. Finally, PE was significantly higher in the ‘FINAL’-situation (28.9 ± 9.4) than in the ‘BASELINE’-situation (19.2 ± 9.3); F(1,34) = 32.353, p=.000; partial η2 = 0.488.

Table 2. Outcome variables for both intensity situations of the PTTF-test divided by age category and gender

n/a. non applicable variable in the ‘BASELINE’-situation;

In comparison with the Total U14 group: *. effect significant at the α=0.050 level; **. effect significant at the α=0.010 level;

***. effect significant at the α=0.001 level. In comparison with the Total U16 group: #. effect significant at the α=0.050 level.

Table 3. Repeated Measures ANOVA Test of Within-Subjects Effects for Heart Rate in the PTTF-test.

*. Intensity Situation; **. mean square; ***. degrees of freedom; #. effect significant at the α=0.001 level.

With regard to the Intensity situation x Age category interaction, a significant effect was found on VA-index; (F(2,34) = 6.432, p=.004; partial η2 = 0.274). However, the effect was not apparent for the remaining technical outcome variables. Table 2 shows the mean stroke velocity, mean stroke accuracy, VA-index and PE in both the intensity situations for all age categories and per gender.