Pelvis-Trunk coordination strategies differ a cross preparatory court movement distances during the tennis forehand

Participants

Eighteen male college athletes (age: 23.6 ± 2.2 years, body mass: 75.1 ± 6.5 kg, height: 175.2 ± 6.8 cm, and experience: 10.9 ± 3.1 years) from a local college tennis team participated in this study. The athletes had an International Tennis Number (ITN) ≥ 4, meeting the ITN standards, and trained for 20 h per week. To meet the inclusion criteria, each participant completed at least six years of structured tennis training and competition. Additionally, all participants were required to be in good health, with no current or chronic injuries in the past six months. To ensure adequate statistical power of 80%, the required sample size was calculated using G*Power software (version 3.1.9.7). The repeated-measures analysis indicated that at least 18 participants were needed, with parameters set at α = 0.05, β = 0.8, and r = 0.70. This study adhered to the principles outlined in the Declaration of Helsinki. The study protocol was explained to all participants, and written informed consent was obtained. The protocol was approved by the Jeonbuk National University Institutional Review Board and conducted in accordance with relevant guidelines and regulations (JBNU2022-04-008-002). The recruitment period spanned from February 6, 2023 to May 12, 2023.

Protocol

To minimize the effects of learning and fatigue on the formal experiments, all participants visited the laboratory twice. During their first visit, demographic variables were recorded, and a maximum movement distance test was conducted. Participants first completed a specific warm-up program, which included five minutes of low-intensity sideways and acceleration runs, followed by at least ten minutes of groundstrokes at submaximal speed. After the warm-up, the maximum movement distance test was performed. In this test, participants self-selected a distance and executed forehand strokes using crossover step footwork, directing the ball toward the center of a circle with a 1-meter diameter positioned above the ball machine. All participants followed a standardized down-the-line stroke protocol (Giles & Reid, 2021), and adopted an open stance for varying movement distances. Consistent with previous studies, a ball machine (The Tennis Ball Machine PRO, Seoul, South Korea) delivered new balls (Wilson TRAINER) with a standard rotation and speed (25.3 ± 0.4 m/s) to the same stroke position through machine speed model setting for each participant (Landlinger et al., 2010a; Landlinger et al., 2010b). To replicate the environment of a real competition, the circle was placed approximately 6 m from the participant and 1.3 m above the ground, reflecting the attacking shots scenarios and the net height. The distance was gradually increased in half-length increments from the ground to the left or right greater trochanter of the femur for each participant until they could no longer effectively execute the forehand stroke technique or maintain accuracy. The maximum movement distance was recorded, and the mean of five trials was used to determine three distances: (a) 100%, (b) 75%, and (c) 50% of the maximum movement distance, reflecting the different preparatory court movement task constraints (Fig. 1A). In this study, the average maximum movement distance was 3.47 ± 0.21 m.

Formal experiments were conducted after a minimum of 48 h. Participants were instructed to avoid caffeinated beverages for at least eight hours prior to the trial. All participants were tested at the same time of day and were required to wear their own tennis shoes and use their own rackets during the test to minimize the influence of external factors on stroke performance. A single researcher applied 57 reflective markers (Li et al., 2025) (14 mm; Biomech Marker Set, OptiTrack, Corvallis, OR, USA) to specific anatomical landmarks on the participants’ bodies (Fig. 1B). A 15-segment model was created, including the feet, shanks, thighs, pelvis, trunk, head, bilateral upper arms, forearms, and hands. Additionally, four markers were placed on the tennis racket, positioned at the 3 o’clock and 9 o’clock positions on the racket head and shaft to capture its trajectory (Landlinger et al., 2010a). The same warm-up program from the first laboratory visit was performed to ensure all participants had a similar stroke feeling as in a match scenario and adapted to the markers attached to their bodies. To minimize learning effects, the ball was placed in the machine at random intervals unknown to the participants. They were instructed to stroke the ball with maximum effort, aiming for accuracy by hitting the circle with down-the-line strokes. The first five successful strokes at each distance were included for further analysis. Participants rested for more than two minutes between trials, or until they reported no fatigue, to ensure they maintained their initial performance levels. After each stroke, participants promptly returned to the starting position. The impact height was adjusted according to the position of each participant’s hip joint by adjusting the launch angle of the machine (Landlinger et al., 2010a; Landlinger et al., 2010b).

Data collection

A motion capture system (OptiTrack, Natural Point, Inc., Corvallis, OR, USA) equipped with 13 high-resolution cameras was used to record the 3D trajectory of the reflective markers during the forehand stroke motion at a frame rate of 240 Hz, from the start position to the return to the start position. Static models were generated for each participant following a standing calibration trial to determine individual body segment parameters. Reflective tape was applied to the tennis ball to accurately capture the moment of impact.

Data processing

The raw data were imported into Visual 3D software (Professional 6.0; C-Motion Inc., Germantown, MD, USA) and low-pass filtered using a fourth-order, zero-lag Butterworth filter with a cutoff frequency of 6 Hz (Li et al., 2023; Li et al., 2025). The pelvis and trunk rotation angles in the transverse plane were calculated by the distal segment relative to the proximal segment using an X-Y-Z cardan rotation sequence (Augustus, Hudson & Smith, 2021). Specifically, the dominant pelvis angle was calculated as right thigh to pelvis, the non-dominant pelvis angle was calculated as left thigh to pelvis, and trunk angle was calculated as pelvis relative to trunk (Augustus, Hudson & Smith, 2021). Angular velocities were calculated from the first-order derivatives of the angles using Visual 3D software. For right-handed players, the right and left sides were defined as the dominant and non-dominant pelvis and trunk, respectively. To standardize the data, global mediolateral data for the two left-handed players were reversed, allowing all players to be analyzed as right-handed (Buszard et al., 2020).

The phase of interest was the period from the beginning of the forward rotation of the racket to the moment of impact, defined as the acceleration phase. This phase was identified by analyzing the trajectories of the markers on the racket head and the reflective markers on the ball (Buszard et al., 2020); both stroke events were visually inspected and confirmed using Visual 3D and video recordings. Additionally, the resultant speed of the racket was computed in Visual 3D as the 3D resultant velocity of the racket in three planes of motion in the one frame prior to impact (Landlinger et al., 2010a; Landlinger et al., 2010b).

A custom MATLAB code (version R2022b; MathWorks, Natick, MA, USA) was used to compute the non-dominant and dominant pelvis-trunk CRP curves. Consistent with a previous study of golf swing (Choi et al., 2016), CRP curves were established using four steps. First, joint data were normalized to 101 data points corresponding to the acceleration phase. Second, a phase plane for each joint was created by plotting the angle (θ) and angular velocity (ω), with values normalized to their relative minimum and maximum, resulting in a range from −1 to 1. Third, the phase angle (∅) was calculated at each data point as ∅ = tan⁻¹ (ω/θ). Finally, the CRP angle was calculated as ∅pelvis - ∅ trunk.

A CRP value of 0° indicated that the two joints were moving in the same direction (“in-phase”), while values of −180° or 180° indicated that the joints were moving in opposite directions (“out-of-phase”). Positive CRP values indicated that the trunk position was ahead of the pelvis, and a positive slope denoted that the trunk was rotating faster than the pelvis. In contrast, a negative slope indicated the opposite (Choi et al., 2016; Lamb & Pataky, 2018). Variability was calculated as the standard deviation of the CRP data points across the acceleration phase for all trials and participants.

Coordination pattern features

To evaluate the relationship between pelvis-trunk coordination and racket speed at impact, the following features of the CRP were extracted: (a) the mean value of the CRP, to quantify the average difference in rotation movement between the pelvis and trunk segments; (b) the peak value of the CRP, to identify the timing of changes in the dynamic pelvis-trunk coordination pattern; (c) the maximum positive CRP slope, to measure the rate at which the trunk rotates faster than the pelvis; and (d) the maximum negative CRP slope (slope = [Y2-Y1]/[X2-X1]), to measure the rate at which the pelvis rotates faster than the trunk. Four features were extracted from the CRP curves for both the non-dominant and dominant sides. Microsoft Excel (version 2019; Microsoft Corp., Redmond, WA, USA) was used for all feature extractions and calculations. To determine the relationship between inter-segment coordination features and racket speed, based on previous research (Chardonnens et al., 2013), players were categorized into two subgroups based on more homogeneous racket speeds at impact observed at each movement distance.

Statistical analysis

To test the hypothesis that movement distances have no effect on the pelvis-trunk CRP, the entire time series was statistically examined using Statistical Parametric Mapping (SPM) one-way repeated-measures ANOVA. First, the test statistics (SPM{F} and SPM{t}) were calculated. The F and t-statistics are qualitatively similar to the effect sizes, and can be used as indicators of practical significance. Random field theory was applied to control the Family-wise type I error rate via a smoothness-dependent correction for multiple comparisons, with the family-wise type I error rate set at 0.05. The critical threshold (F* or t*) was calculated, and if the test statistic trajectory exceeded the critical thresholds, the null hypothesis was rejected, and the difference was considered statistically significant. Finally, the p-value was calculated for each suprathreshold cluster. If a significant main effect was identified, paired comparisons were conducted using SPM{t} tests with Bonferroni corrections to determine the location of the differences. The alpha level for pairwise contrasts was adjusted for the number of comparisons per dependent variable (N = 3, α = 0.017). SPM analyses were executed using the open-source code (http://www.spm1d.org) within the MATLAB software.

The normality of discrete CRP feature data was analyzed using the Shapiro–Wilk test, and all data were confirmed to be suitable for parametric analysis. A one-way ANOVA was used to assess whether racket speeds differed when the ball was impacted at the three movement distances. Pearson’s correlation analysis was conducted to examine the relationships between CRP features and racket speed. All statistical analyses were conducted using SPSS (SPSS Inc., Chicago, IL, USA), with a statistical significance level set at p < 0.05.

Pelvis-trunk CRP at different movement distances

Consistent with our hypothesis, a significant main effect of movement distance on pelvis-trunk CRP was observed. As movement distance increased, CRP values for non-dominant pelvis-trunk coordination progressively decreased during the pre-acceleration phase, indicating a reduction in the pelvis’s posterior rotation position relative to the trunk. This finding supports Wagner et al. (2012) and Button et al. (2003) in throwing actions, tennis players also maintain outcome stability through compensatory kinematic movements of the distal segments. Conversely, CRP slopes for dominant pelvis-trunk coordination progressively increased during the post-acceleration phase, suggesting that the pelvis’s anterior rotation was faster than the trunk’s. These variations in pelvis-trunk coordination patterns align with findings from a study by Giles & Reid (2021), which examined the effect of different entry speeds (analogous to movement distance in our study) on professional female players. Their study reported a decrease in trunk rotation prior to ball impact. However, trunk rotation in professional male players was unaffected by entry speed, likely due to differences in skill levels. The college athletes in our study demonstrated hitting strategies resembling those of professional male players when performing more straightforward task. However, in more extreme tasks, the pattern of pelvis-trunk coordination contradicted the typical proximal-to-distal sequence observed in the kinematic chain. The strategy for addressing the “degrees of freedom problem” appeared to depend on the interaction between movement distance and expertise level (Federolf et al., 2014; Giles & Reid, 2021; Bernstein, 1967). Analyzing coordination strategies could reveal how players adapt to the complexity of the human-task system.

Regarding the temporal region where the significant main effect of movement distance occurred, the SPM statistics indicate that the non-dominant pelvis-trunk CRP adopted an in-phase coupling strategy during the pre-acceleration phase (23–41%). This suggests a decrease in the pelvis-trunk separation angle, which can be explained by the stretch-shortening cycle principle (Choi et al., 2016). Comparing the CRP curve more closely to the results of Lamb & Pataky (2018) provides insight into the role of pelvis-trunk coupling in controlling outcome performance. To counteract the increased lateral momentum and ground reaction force generated by greater movement distances, players delay the X-factor stretch to stabilize the pelvis, allowing the trunk to achieve axial acceleration (Lamb & Pataky, 2018). In contrast, the dominant pelvis-trunk CRP was observed during the post-acceleration phase (76–100%). The CRP curves reveal that the slope increases disproportionately compared to the absolute CRP value as movement distance increases, suggesting that muscular strength may play a more critical role than range of motion; i.e., an athlete’s ability to maximize angular velocity is more important for the racket speed. This interpretation aligns with findings by Seeley et al. (2011) who reported that maintaining higher post-impact ball speeds was more dependent on peak joint angular velocity than on peak joint angle, which disproportionately increased. The trunk as the distal link in the kinetic chain during tennis strokes, exhibits minimal relative change compared to the pelvis. This likely ensures stroke accuracy and maintains consistency in distal movement. These findings align with the “leading joint hypothesis”, this hypothesis suggests that the distal joint may be more suitable for the leading role, thereby generating the limb motion characteristics required by the task, including movement direction, accuracy, and so on (Dounskaia, 2010; Buszard et al., 2020). Indeed, since the mean number of strokes required to obtain five successful trials was 5.8 ± 0.6 and 5.9 ± 0.7 at the minimum and maximum movement distances, respectively, accuracy was not negatively impacted by increases in movement distance. This may explain why differences in dominant pelvis-trunk CRP at different movement distances were observed prior to the stroke. The time-series data provided by the SPM approach highlight the role of non-dominant and dominant pelvis-trunk coupling in adapting to task variations across different temporal regions. We suggest that players, particularly those engaged in unilateral motor skills such as tennis, should focus on strengthening core muscles, including the external oblique abdominal muscles, to enhance the non-dominant side’s ability to initiate body rotation and maintain dynamic balance.

Further analysis revealed that coordination variability (standard deviation) in the dominant pelvis-trunk was lower than in the non-dominant pelvis-trunk across different movement distances, consistent with the findings of Brito et al. (2022) regarding a 1.5 m moving distance forehand stroke task. In contrast, a study on golfer’s downswings by Choi et al. (2016) found smaller coordination variability in the dominant pelvis-trunk compared to the non-dominant side. A possible explanation for this difference is the absence of task constraints in the golf downswing. As movement distance increased, coordination variability in both the non-dominant and dominant pelvis-trunk also increased, likely due to the increased task demands on the system, which require more resources to achieve the task goal and consequently results in greater variability for increased flexibility (Bartlett, Wheat & Robins, 2007; Cazzola, Pavei & Preatoni, 2016). Ruiz-Malagon et al. (2023) presented similar findings, they found greater variability in the trunk in the case of the tennis forehand and backhand than the serve, this could be due to the fact that in the forehand and backhand stroke the ball was thrown by a ball throwing machine and there are more sources of variability. Findings from golf support Newell’s ecological dynamics theory, which posits that task constraints (e.g., movement distance) and self-organization (e.g., skill level) should both be considered when analyzing complex temporal and spatial actions (Newell, 1986). Notably, smaller variability might result in a more concentrated distribution of stress across tissues (Abrams, Renstrom & Safran, 2012; Davids et al., 2003), potentially increasing cumulative loads on internal body structures. This could contribute to mechanisms of repeated loading and high torsional stress, leading to overuse injuries in the trunk and lower back of professional athletes (Abrams, Renstrom & Safran, 2012; Bartlett, Wheat & Robins, 2007; Davids et al., 2003). This finding may inform injury prevention strategies, future prospective training research still needs to confirm this hypothesis. Players employing varied techniques to develop racket speed while minimizing injury risk, such as different stances and stroke directions may be used as potential application in the future.

Relationship between CRP features and racket speed

When comparing the CRP curves with racket speed, a trend emerged indicating that players with the fastest racket speeds adopted different pelvis-trunk coordination strategies on the non- dominant side compared to those with the slowest racket speeds, particularly at the minimum movement distance. For players with the slowest racket speeds, the trunk played a dominant role throughout the movement. Conversely, for players with the fastest racket speeds, the pelvis dominated from 0% to the peak CRP, while the trunk took over from the peak CRP to 100% during the acceleration phase. The mean CRP showed a moderate to large negative correlation (Table 1) with racket speed suggests that players with the fastest racket speeds tend to exhibit a greater pelvis-trunk separation angle and a longer backswing distance under simple task constraints. We hypothesize that this phenomenon may aligns with the principle of the stretch-shortening cycle, where active muscles generate a powerful force during the acceleration phase through concentric shortening when immediately preceded by counter rotation, revered by an eccentric muscle action (Fletcher & Hartwell, 2004; Komi, 1984). Additionally, the correlation coefficient for peak CRP on the non-dominant side (r = 0.488, Table 1) likely highlights the importance of precise timing between the stretching and shortening phases in maintaining robust racket speed during non-dominant pelvis-trunk coordination (Elliott, 2006; Landlinger et al., 2010a). These findings underscore the critical role of non-dominant pelvis-trunk coupling in generating racket speed under lower task constraints. This observation supports prior research emphasizing the non-dominant side’s contribution to racket speed in sports such as table tennis and golf (Iino, 2018; Choi et al., 2016). Consequently, future research should focus on training programs aimed at strengthening the rotational muscle groups of the non-dominant pelvis, which may be a potential application to enhance racket speed and overall performance. At the medium movement distance, the increase in the number of coordination features showed a moderately negative correlated with racket speed, likely due to the need for a more complex set of motor instructions. Given that a medium movement distance occurs approximately 80% of the time in a match, players must employ advanced neuromuscular strategies to optimize technique and overcome task constraints while maintaining racket speed (Athalye et al., 2017). When analyzing the CRP curve trends relative to racket speed, the non-dominant pelvis-trunk separation angle consistently contributed to racket speed (r = −0.493). Additionally, the positive CRP slope on both the non-dominant and dominant sides was significantly correlated with racket speed (r = 0.477 and r = 0.514, respectively). This suggests that faster rotational speed of the trunk relative to the pelvis on the non-dominant side during the post-acceleration phase and on the dominant side during the pre-acceleration phase is critical for generating racket speed. Furthermore, the peak CRP on both sides showed significant correlations with racket speed (r = 0.628 and r = 0.551), indicating that the key to utilizing elastic energy lies in the precise timing between the stretching and shortening phases (Elliott, 2006). As the movement distance increased, the reduced correlation between the mean CRP and racket speed on the non-dominant side may have been compensated for by faster trunk rotation (positive CRP slope) and optimized timing (peak CRP). These mechanical compensations represent inherent biomechanical strategies for executing specific motor tasks (Dupuy, Motte & Ripoll, 2000; Whiteside et al., 2013). Therefore, tennis forehand training should prioritize enhancing flexibility and coordination in both non-dominant and dominant pelvis-trunk movements, while also developing adaptable coordination strategies to address varying task constraints.

The current study had several limitations. In terms of the study population, the selection of skilled male athletes was limited to the one college. In addition, racket speed likely distortions by data smoothing at the moment of impact (Knudson & Bahamonde, 2001; Reid, Campbell & Elliott, 2012). When interpreting the results of this study, it is important to acknowledge that pelvis-trunk coordination represents just one component of the complex movements involved in the tennis forehand. This study specifically focused on task constraints and did not consider potential interactions with other factors, such as self-organizing and environment elements (e.g., unanticipated situations, fatigue). Given the dynamic nature of tennis, it is reasonable to hypothesize that the ability to adapt to unanticipated scenarios and the physiological and psychological fatigue associated with match play could influence pelvis-trunk coordination and its relationship to performance. Individual variability in player strategies was not considered in many of the analyzed variables. The decision not to evaluate players individually was made to identify broader trends in pelvis-trunk coordination during forehand strokes, rather than focusing on specific strategies. However, individual differences may play a critical role in performance outcomes. Future research should investigate whether individual coordination patterns are linked to performance metrics, such as racket speed, to deepen our understanding of how specific biomechanical factors contribute to optimal performance.

Key points

• Movement distance influences pelvis-trunk coordination: Significant changes in pelvis-trunk coordination were observed as movement distance increased in skilled male player, particularly during the pre- and post-acceleration phases, highlighting the adaptability of coordination strategies to varying task demands.

• Correlation between coordination features and racket speed: Key coordination features, such as mean and peak CRP values, were significantly correlated with racket speed at different movement distances, emphasizing the role of pelvis-trunk coupling in generating higher racket speeds.

• Confirmed previous reports of increased coordination variability: As movement distance increased, greater variability in pelvis-trunk coordination was observed in skilled male player, indicating the necessity of greater flexibility and neuromuscular strategies to sustain performance under higher task constraints.

• Non-dominant peak CRP tended to be associated with faster racket speed: Non-dominant side coordination, particularly the rotational speed of the trunk relative to the pelvis in transverse plane, was tended to be associated with faster racket speeds, especially under lower task constraints.