Canoe polo is an increasingly popular kayaking discipline with World Championships held every 2 years since 1994. On a 35 × 20 m stretch of water, two teams of five players compete to score a goal (1.2 × 1 m) 2 m above the water. The playing pitch can be in open water or swimming pools. Players can grab and throw the ball by hand or using their paddle (https://www.canoeicf.com/disciplines/canoe-polo).

Canoe polo requires kayaking skills, particularly navigation and stabilization techniques, but also ball-handling skills to successfully manage throws toward the goal. In this sense, some comparisons can be established with other overhead sports, namely, water-polo, which shares the same ball and aquatic environment, handball, baseball, volleyball, or tennis because of the overhead motions. The kinematics of these latter have been widely documented (Anglin and Wyss, 2000; Park et al., 2002; Tillaar and van den Ettema, 2007; Martin et al., 2012; Kaczmarek et al., 2014; Gelber et al., 2018) and compared (Fleisig et al., 1996; Meriç et al., 2009; Wagner et al., 2014). In these studies, a thorough description of the overhead throwing motion in six phases has been provided: wind-up, early cocking, late cocking, acceleration phase, deceleration, and follow-through phase, which can also be observed in Canoe Polo (Figure 1). Furthermore, a similar inter-articular kinematic sequencing is generally observed in overarm motions (Fleisig et al., 1996; Marshall and Elliott, 2000; Fradet et al., 2004; van den Tillaar and Ettema, 2009; Wagner et al., 2014): a proximal-to-distal sequence with the maximum angular velocity being first reached by the pelvis rotation, then trunk rotation, trunk flexion, elbow extension and finally shoulder internal rotation and shoulder flexion, the angular velocity being the highest for the shoulder internal rotation. Despite these sequencing similitudes, discrepancies in maximal angles, angular velocities, and their timing were observed when comparing different overhead sports. For instance, averaged maximal shoulder internal rotations range from 4,520 ± 1,020°/s up to 5,580 ± 2,350°/s depending on the sport (Wagner et al., 2014). Furthermore, differences were observed between jumping vs. standing throws (Wagner et al., 2011), the contact with the floor enabling greater rotations of the trunk and pelvis and thus greater angular velocities for these segments.

In canoe polo, the athletes are in contact with the kayak at both footrest and seat level. Their legs are extended in the longitudinal direction of the boat throughout the throw, their pelvis mobility being constrained to avoid capsizing (Michael et al., 2009). These major constraints may influence the overhead motion. However, to our knowledge, their impact on canoe polo has not been studied yet. The impact of the mobility of the seat, however, has been reported to influence the paddling kinematics (Fohanno et al., 2011; Willmott and López-Plaza, 2014). The swivel seat that allows rotation about the vertical axis was designed to help the athlete's pelvic rotation and fluidify trunk rotation. In paddling, the use of a swivel seat reduced the rotation angles between the pelvis and the trunk and increased the kayaker's knee range of motion (Fohanno et al., 2011; Willmott and López-Plaza, 2014). As the swivel seat seems to modify the transfer of momentum from proximal to distal, we hypothesize that it might impact the kinematics of the canoe polo throw.

The aim of this study was to analyze the canoe polo throw kinematics in terms of angles [pelvis, trunk, and glenohumeral degrees-of-freedom (dof)] and inter-articular kinematic sequencing, considering the constraints imposed by the seated position. A secondary aim was to investigate whether adding pelvis mobility to the seated player (by enabling pelvis rotation) has an impact on its upper body kinematics and the sequencing. We hypothesize that the typical proximal-to-distal sequence found in overarm sports will be modified for canoe polo throws with the trunk reaching its maximal rotation velocity at the same time or before the pelvis. Our second hypothesis is that the pelvis axial rotation limitation is the major cause for this modified inter-articular sequence.

The canoe-polo throw starts with the pelvis and trunk rotating backwards [the cocking phase, from 0 to 68.01 ± 2.93% (fixed pelvis) vs. 67.22 ± 3.37% (free pelvis)], the shoulder is externally rotated (Figure 5). Then, during the acceleration phase, shoulder internal rotation (glenohumeral axial rotation) occurs resulting in shoulder angular velocities up to about 3,500°/s (Figures 5, 6).

Qualitatively, the trunk axial rotation seems to reach its maximal angular velocity first, followed closely by the pelvis rotation, then the trunk flexion-extension, elbow flexion, and the glenohumeral dofs (Figures 6–8).

The kinematics obtained during the two different types of throws (fixed or free pelvis) differ for the pelvis and trunk rotations (pelvis rotation, trunk flexion-extension, and lateral flexion mainly) but only slightly for the shoulder (Figure 5, the details of the t-statistic and p-values can be found in Supplementary Figure 2). The significant differences observed for the pelvis rotation confirm that the participants actually utilized the axial rotation allowed by the seat: when free, the pelvis rotates more backward (about 20° on average) during the cocking phase (Figure 5). As for the trunk, in the “free pelvis” configuration, its flexion was larger (about 10° on average) as well as its lateral flexion (a few degrees on average). However, despite these kinematics differences, the players reached statistically similar magnitudes of angular velocities at the shoulder (~3,500°/s of glenohumeral axial rotation, Figure 6; the details of the t-statistic and p-values can be found in Supplementary Figure 3). Furthermore, no significant differences were found for the ball release speed (Figure 7, the details of the t-statistic and p-values can be found in Supplementary Figure 4), despite some differences being observed at the previous phases of the throw.

The two-way ANOVA yielded a significant effect for the timing of the maximal angular velocities (F = 30.035, p < 0.001) but didn't show any significant effect of the pelvis mobility (F = 1.056, p = 0.365), nor of its interaction with the timing (F = 2.098, p = 0.087, Figure 8). The post-hoc analysis confirmed significant differences between the timing of the maximal angular velocities for all dof combinations (p < 0.001), except for the pairs: trunk flexion-extension vs. elbow flexion, trunk flexion-extension vs. glenohumeral axial rotation, trunk flexion-extension vs. glenohumeral elevation, and elbow flexion vs. glenohumeral elevation (the detail of the p-values can be found in Supplementary Figure 5).

This study investigated the canoe-polo throw and the effects of the pelvis rotation mobility on the kinematics of the upper body and the inter-articular coordination. As hypothesized, the typical proximal-to-distal sequence found in overhead sports is modified for canoe polo throws. However, the pelvis axial rotation limitation isn't the major cause for this modification, as we initially thought.

First, similitudes can be observed between canoe polo throws and other overhead sports. The same phases (wind-up, early cocking, late cocking, acceleration phase, deceleration, and follow-through phase) are found with qualitatively similar kinematics: the pelvis and trunk rotate backwards during the cocking phase, then during the acceleration phase the shoulder internally rotates (Wagner et al., 2014). The maximal shoulder velocity obtained here was about 3,500°/s in accordance with the magnitudes listed in the literature (from 4,520 ± 1,020°/s up to 5,580 ± 2,350°/s). However, the proximal-to-distal sequence is modified during a canoe polo throw with respect to other sports, confirming our first hypothesis. Here, the pelvis kinematics being limited, it reaches its maximal velocity in rotation slightly later than the trunk in rotation: on average about 0.018's later, while Wagner et al. (2014) reported that the maximal angular velocity occurs sooner for the pelvis rotation vs. the trunk rotation: 0.029 s sooner for team handball throw, 0.036's for a tennis serve and 0.05's for the volleyball spike. The rest of the sequence was yet the same as reported in the literature: trunk flexion, elbow extension, and finally shoulder dofs (Fleisig et al., 1996; Marshall and Elliott, 2000; van den Tillaar and Ettema, 2009; Wagner et al., 2014).

We also hypothesized that the major reason for these sequencing discrepancies would be the mobility of the pelvis in rotation. When comparing the upper body kinematics for two pelvis mobility configurations (free or fixed pelvis) using spm, significant differences were actually observed for the pelvis rotation, trunk flexion-extension, and trunk lateral flexion mainly. It might seem trivial to observe pelvis rotation discrepancies. However, this result is actually useful to confirm that canoe polo players really capitalize on the additional pelvis rotation around the vertical axis. With this mobility addition, the pelvis could rotate more backwards, and the trunk dofs were larger (flexion and lateral flexion). These significant differences visible for the trunk but not for the shoulder show that it is the thoraco-pelvis joint—and the trunk itself—that adapt to the pelvis mobility rather than the shoulder. This is in accordance with the results obtained when using the swivel seat for paddling: while its use influenced the rotation angles between the pelvis and the trunk, the upper limb kinematics were less affected by the choice of the seat (Fohanno et al., 2011; Willmott and López-Plaza, 2014). It is also in accordance with the analysis from Murta et al. (2020), who reported the impact of a pelvic mobility reduction on the trunk kinematics but not on the shoulder kinematics. The significant differences observed for the glenohumeral plane of elevation dof were limited to a couple of degrees (<10°) and for limited time periods (compared to the pelvis and trunk significant differences). They may be explained by trunk geometry modification (lumbar and thoracic lordosis or kyphosis) generating scapula posture modifications and thus glenohumeral kinematics changes (Suzuki et al., 2019). However, this assumption should be taken with caution since the trunk geometry wasn't specifically measured here.

Thus, adding pelvis mobility clearly had an impact on the upper body kinematics, however, the inter-articular sequence was not modified. The sequence “trunk axial rotation, pelvis rotation, trunk flexion-extension, and then upper limb dofs” was not changed, which might indicate that the limitation in pelvis axial rotation was not responsible for this specific sequence. However, even with the added pelvis mobility, the maximum pelvic rotation (−20°) in reference to the position at rest remains smaller than those observed for tennis (−99°), handball (−43°), or volleyball (−48°) (Wagner et al., 2014). Thus, it is possible, that despite the pelvis enabled mobility, the legs position still imposes a significant constraint on the pelvic rotation. In other sports with an overarm throw, the proximal-to-distal sequence ensures an optimal transfer of energy from the lower body to the ball. The position of the legs, as well, as the double contact with the kayak (seat and footrest) might disturb the energy chain resulting in a different technique to increase the throw velocity. Further studies that measure the forces at the footrest and the seat rotation axis will be needed to understand the reason for the shift in this sequence.

The pelvis mobility impacted the ball velocity. Particularly, the increased linear velocity in the lateral direction might be related to the increased pelvis rotation and trunk flexion. However, as the ball reached the release point, no statistically significant difference was observed. This indicates that pelvis mobility might not impact the trajectory, nor the distance traveled by the ball after the throw. However, the absence of a difference might be the result of the players' expertise. Indeed, as the players are trained to control their throws despite the boat instability, they might have used a similar technic to counter the effect of the swivel seat. It would be interesting for future studies, to evaluate the effect of the pelvis mobility on the ball velocity when thrown with the paddle, as the increased moment arm might further exacerbate the effect of the increased trunk flexion.

Indeed, our protocol had some limitations. In particular, the mobility of the boat on the water was not reproduced, although it could generate larger core muscle actions from the canoe polo player and thus change the thoraco-pelvis mobility. To provide a thorough investigation, it would be interesting to study the impact of boat instability by performing field studies. Another limitation is that the starting posture before the throw isn't the initial natural position of the kayak polo throw since players were asked to take an anatomical posture so the markers could easily be identified. The right position would be hand in pronation and ball in the water. Furthermore, the ball and hand were not wet which may have affected the throw for some participants. Third, although it wouldn't change the inter-articular sequencing presented here, including the scapula-thoracic joint could have permitted to see if kinematics differences were generated at this joint when the pelvis was free or fixed. Additionally, the order of conditions was not randomized across participants, which might have introduced some bias. However, the variability across participants for each condition was comparable which indicates that the bias was negligible. Finally, it would have been interesting to test a larger panel of players from different levels and to include female players to broaden the present conclusions.

As for the modeling limitations, the trunk model could be improved since it is here modeled by a single segment (Wu et al., 2016). The prospects would be to perform the kinematic analysis with a more precise model of the trunk since the lack of rotation of the spine at different stages might affect the shoulder kinematics.

In terms of short-term perspectives, implementing a dynamic analysis (i.e., inverse dynamics and static optimization to estimate muscular forces) might be interesting. Analyzing articular torques and muscular forces for free and fixed pelvis may shed more light on differences at the shoulder. Indeed, the absence of major significant glenohumeral differences does not necessarily mean an absence of difference in muscles recruitment. Also, a power analysis based on the kinetic chain of the different segments involved in the throw, determined by the amplitude of the articular torque and the angular velocity as described by Martin et al. (2014), analyzing the tennis service, may be of interest.

This study provides a better understanding of the kinematics of the upper body during a canoe polo throw. While similar general patterns as in other overhead sport throws are found, a different kinematic sequence is obtained for the canoe polo throw with a maximal angular velocity occurring sooner for the thorax axial rotation than for the pelvis rotation.

Investigations on the pelvis mobility of the seated player showed that the limitation of rotation of the pelvis around a vertical axis had an influence mainly on the pelvis and thorax kinematics but not on the shoulder ones and that it was not the cause for this modified kinematic sequence. Further studies will be needed to shed a light on the reason for this specific kinematic sequencing.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by n° IDRCB: 2019-A00494-53 from French CPP. The patients/participants provided their written informed consent to participate in this study.

PC and CD: conceptualization. SD, CB-V, and YB: methodology. NA and YB: software. CD, CB-V, and SD: investigation/experimentations. CD and NA: writing—original draft preparation. SD, YB, CB-V, and PC: writing—review and editing. SD and YB: supervision. PC: funding acquisitions. All authors contributed to the article and approved the submitted version.

This study was funded by Ramsay Santé, France.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.