Because of the different situations that occur during a water polo match, the players need to move in both a horizontal direction using front crawl and a vertical direction using a fundamental technique such as the egg-beater kick (20,28–30). There is also another technique, the breast stroke, which is adopted mainly by players from the former Yugoslavia. Although studies on time motion have shown that 50% of this time is spent in the vertical position (6,26), little research has been conducted on this subject, in particular on the difference between intermediate and expert players. The expert players are able to produce a considerable amount of explosive force in a very short propulsion time (expert players 161 milliseconds; intermediate players 253 milliseconds), developing considerable power and velocity with a low temporal variability (noise) in the jumps. In the intermediate players, the exact opposite occurs. We believe that temporal variability may have a decremental effect on the performance of the intermediate players and that the structure of their jump capacity is multifactorial and different from the expert players. Hence, the aim of this study is to (a) identify the multifactorial structure of jumping ability, which differentiates the 2 groups of players and (b) identify the role of temporal variability (noise) in the execution of the 2 groups and verify whether there is a relation between power and velocity and temporal variability in the 2 groups.
Experimental Approach to the Problem
The principal components analysis (PCA) is a mathematical technique that is considered the starting point of the multivariate analysis (18). Its role is to show the multifactorial structure of jumping ability in which the variables are reduced to ≥2 principal components. We also used multiple regression because it identifies and predicts the factors that determine jump performance in both intermediate and expert players. As far as we know, the present data are the first to evaluate the multifactorial structure components of jumps in intermediate and expert male water polo players. This further justifies the primary use of PCA and multiple regression (18). The dependent variables are represented by power, velocity, maximal height, and residence time at maximal height because (tmh), together with the variables which indicate temporal variability in jump actions (jct1, jct2, and jct3), they are the most common markers of jumping ability.
The physical and dynamic characteristics of the 2 groups of expert (n = 15) and intermediate (n = 15) water polo players can be seen in Table 1. The experts, who all play in the first division of the championship, have at least 6 years' experience in resistance strength and power output training, whereas the intermediate players have more than a year's experience. The expert players developed 3 double training sessions during the week devoted exclusively to swimming (horizontal and vertical) and 2 afternoon physical preparation sessions in the gymnasium with a 48-hour recovery period. The intermediate players, on the other hand, developed 4 afternoon training sessions in the water and 2 afternoon physical preparation sessions in the gymnasium. All the tests retests were carried out during the Christmas break. An institutional review board for the protection of human subjects approved the procedure for the study. All the subjects signed informed consent documents before participation. About a week before the study, the players were instructed to eat and drink normally. None of the players reported any unusual weight loss or used any unusual dietary practices before or during the study. The tests were always carried out on the same day at the same time.
The players performed warm-up exercises for 20 minutes (mainly for the legs) before beginning the tests. They were all instructed on how the Teknotrain3 worked and were told not to go out of the range of the photocells, either posteriorly or laterally. The tests were carried out after 6 trials, which enabled the subjects to familiarize with the device. All the players wore white caps during the tests. The starting position was with the body and the head oriented toward the first photocell, with the trunk and the arms bent slightly forward and the hands sculling the water, whereas the initial movement of the legs was the egg-beater kick. The results of the jumps were analyzed in computer graphs, and if they did not conform to the rules, they were repeated (the players were not allowed to thrust their arms out and upward at the moment of the jump). Three jump tests were carried out, and the best one was used for our statistics. The calculation of the variation coefficient percentages was effected on a subset of the 7 best performances of the 2 groups of players in jct1 (jump climbing time 1), jct2, jct3, and jump climbing total time (jctt), using all the tests and retests for a total of 6 trials of which we used 3 for the statistical analysis (intraclass correlation coefficient [ICC]).
An electronic device called the Teknotrain3 (Figure 1) was used to analyze jumps out of the water. The device, which had 4 photocells arranged in a vertical order, was positioned on the edge of the pool; the first photocell was placed 30 cm above the surface of the water, whereas the others were 20 cm from each other. When the vertex player's head passed in front of the first photocell during the vertical jump a temporal circuit opened and then closed when it passed in front of the first photocell again on descent. The first photocell marked the beginning of the measurement of the ascending phase of the jump. In fact, after the breast stroke leg kick, the body was projected upward in negative acceleration until it reached a point of stasis (the conclusion of the ascending phase). At this point, the direction was inverted and the body, in positive acceleration and free fall, went past the first photocell once again, thus closing the descending phase of the flight. The program recorded the time of flight (milliseconds), which compared the correspondence between the time of flight and the height (centimeters) in a table (10). Basically, the longest flight time always corresponded to the greatest height and vice versa. These correspondences had been previously standardized (10). Hence, the time of flight began when the head went past the first photocell positioned 30 cm from the surface of the water. This ensured that the head did not exceed the line of the first photocell before the start of the jump, which would have nullified the test. Hence, the flight time was translated into maximum height to which the distance between the surface of the water and the first photocell was added (30 cm). In addition, the successive activation of the 4 photocells allowed us to measure jct1, jct2, and jct3. This then enabled us to measure dynamic variables such as velocity (milliseconds), power (watts) and force (newtons), mh (centimeters), and the tmh (milliseconds). To calculate the dynamic parameters of the jump, we considered 50% of the land body mass of the players (22).
A week after the first jump tests, the 2 groups were given a retest to evaluate the reliability of the results using in particular the ICC. We used a 2-way blocked analysis of variance (ANOVA) with random blocks in which each player represented a particular block with 3 trials before-after so as to reduce error by subtracting both the variance because of the difference between the trials and that because of the difference between the blocks (the players). This gave us a much lower square mean of error than could be obtained using a 2-way ANOVA and consequently made the results more precise and reliable. We used Pearson correlations to calculate the percentage of common variance. To avoid the problem of outliers owing to the limited number of subjects (15) in the PCA from the SPSS statistical package (version 13.0; SPSS, Inc., Chicago, IL, USA), the following strategy was used: (a) We kept the 2 groups separate, preferring to carry out 2 different PCAs (b) if there was a result of <0.50 after calculating the communalities, it was excluded and the operations started again from the beginning. This strategy ended when the results of all the variables were higher; (c) we then analyzed the components for the presence of complex structures, which appeared when a variable had a value of ≥0.40 on more than 1 component. The Kaiser-Meyer-Olkin (KMO) measure sampling adequacy (MSA) should be >0.50. The probability associated with the Bartlett test of sphericity should be <0.001. Consequently, the values of jct1, jct2, and strength were excluded in the expert players, whereas jct1, tmh, power, and force were excluded in the intermediate players. Two multiple regressions were then calculated, one for the expert players and the other for the intermediate players. We also calculated Pearson's correlations again to establish a relation between the power and velocity developed and the temporal variability of jumps in expert and intermediate players. Finally, Spearman's rho was calculated between mh and tmh. The statistical significance was set to p ≤ 0.05 for all the analysis.
The ICC for expert players gave the following results: 0.94 (0.88–0.98), with no difference between the tests (F 2,28 = 2.69; p > 0.05) and a significant difference between the players (F 14,28 = 2.24; p < 0. 05). The ICC for the intermediate players was 0.89 (0.84–0.95) with no difference between the tests (F 2,28 = 2.54; p > 0.05) and a significant difference between the players (F 14,28 = 2.37; p < 0.05). The expert players achieved the following results in the Pearson correlations and common percentage variance: mh = 0.96, tmh = 0.96, power = 0.96, velocity = 0.97, force = 0,97, with 92.92%, whereas for the intermediate players, the results were as follows: mh = 0.96, tmh = 0.95, velocity = 0.97, power = 0.96, force = 0.95, with 91.78%. The PCA reduced the dimensionality set of the 8 variables to 2 components for both experimental groups. In expert players, the dependent variables accounted for by the first component included mh = 0.988, jct3 = 0.945, velocity = 0.928, power = 0.978, whereas the second component only accounted for the variable regarding tmh = 0.992, which represented 99.59% of the total variance in the set. The communalities represent the proportions of the variance of each single variable accounted for by the component extracted and have high values, namely, mh = 0.98, velocity = 0.914, power = 0.966, tmh = 0.985, and jct3 = 0.909. The KMO measure sampling adequacy MSA was 0.63.The probability associated with the Bartlett test is p < 0.001. In the intermediate players, the first component included jct2 = 0.981 and velocity = 0.966, whereas the second included jct3 = 0.955 and mh = 0.947 and accounted for 98.9% of the total variance. The (KMO) MSA is 0.65, and the probability associated with the Bartlett test is p < 0.001. The communalities had high values such as jct2 = 0.977, jct3 = 0.994, velocity = 0.975, and mh = 0.995. Following the indications provided by the PCA results, we calculated multiple regressions for both the expert and the intermediate players. In the experts, the dependent variables are tmh, whereas velocity, power, jct3, and mh acted as predictors with F 4,14 = 3.94, p < 0.024, adjusted R 2 was 0.89, mh t = 2.73, p < 0.04; velocity t = 4.4, p < 0.007, power t = 2.75, p < 0.04, jct3 t = 3.54, p < 0.016. In the intermediate players, the variable dependent was the mh, whereas the predictors were velocity, jct2 and jct3 with F 3,14 = 4.1, p < 0.034, the adjusted R 2 was 0.70 and jct3 t = –0.803, p > 0.05, jct2 t = –1.52, p > 0.05, velocity t = 2.34, p < 0.04. The coefficients of variation (CVs) = [(SD/mean) × 100], in the expert players' subset, were jct1 = 2.8%, jct2 = 4.2%, jct3 = 6%, and jctt = 4.3%, whereas those of the intermediate players' subset were jct1 = 48.6%, jct2 = 16.9%, jct3 = 7.8%, and jctt = 24.4%. The relations between power, velocity, and temporal variability in the expert and intermediate players were negative, respectively, r = −0.92 (p < 0.01), r = −0.93 (p < 0.01) and r = −0.89 (p < 0.01), r = −0.94 (p < 0.01). Finally, in expert players, Spearman's rho between mh and tmh was 0.939, whereas in intermediate players, it was 0.753 (p < 0.05).
The PCA reduced the 8 variables to 2 components in the expert players. The first includes mh, velocity, power, and jct3, and the second consists only of mh while, at the same time, velocity, power, and mh are highly significant predictors of tmh. The same reduction was seen in the intermediate players. In fact, the first includes jct2 and velocity, and the second consists of jct3 and mh of which velocity, mh, jct2, and jct3 are predictors although the last 2 are not significant. Hence, the different multifactorial composition of the expert and intermediate players is clearly shown.
According to computational models, temporal variability (noise) is produced at the moment of movement planning and during the execution (11,15,17,32) through the motor commands, which reach the motor effectors. Hence, the hypothesis regarding the limit seen in the intermediate players in producing power as an acceleration force is well grounded. This may have a decremental effect on the jump performance determining a greater temporal variability. In fact, by expressing greater power and velocity (expert players), the muscles of the lower limbs can reduce the temporal variability of the motor commands during the execution, so making the jump more effective.
The low temporal variability in the expert players enables them to have greater velocity and power (Table 1) in jctt, confirming their status as expert players. The expert players reached these specific adaptations as a result of resistance strength and power training for >5 years, mainly in the development of specific mechanisms for neural adaptation (1–4). Similar results to ours, above all in the expression of the higher amount of power output, differentiate elite and subelite rugby players (8). The intermediate players also had problems in jct2 and jct3 (the lack of power is associated with a high temporal variability and jct2 and jct3 are predictors, although they are not significant for maximal height). Confirmation of the above-mentioned conclusions also comes from the results of the 3 individual times in the intermediate players 8, 12, and 13 (Figure 2). Although these players developed similar performances to the expert players in jct1 and jct2, they had quite a large jct3, which considerably increased jctt (Figure 2). Hence, it was not sufficient to perform well during the first 2 jump times (jct1 and jct2) to develop an excellent jump. The deterioration of the performance during the jct3 may be because of (a) a lack of constancy in power during the whole movement because of a continuous variation in the lever relationships and (b) maximal momentary power in relation to the resistance to be overcome (body weight) (23). Furthermore, a low rate of force development (3) conditions the development of power and acceleration (5), and the force-velocity relationship produces an increased rate of crossbridge attach-detach cycling, which equates to less time to develop the tension needed to produce force (21). In addition, the variability of the resistance (body weight) to be overcome, which dynamically changes when the body rises out of the water, creates some problems. In fact, measurements of the body weight of football players, divided into defenders and attackers, respectively, on land (121.73 and 97.37 kg) and completely immersed in the water (7.30 and 6.52 kg) gave this difference (24). In the few thousandths of a second of the duration of the jump, the player's body weight can increase about 8 times and considering that the power generated by the intermediate players is only 53.4% of that generated by the expert players, the set accounts for the difference in results (Table 1). However, the differences are not only quantitative but also qualitative because the variation coefficients in jct1, jct2, and jct3 are low (between 2.8 and 6% in expert players and between 48.6 and 7.8% in intermediate players). This confirms how the expert subjects minimize temporal variability. A model was proposed that explains how central nervous system controls a multijoint movement. The leading joint, which has been identified in the hip joint (20), develops the dynamic component for the whole limb, whereas subordinate articulations (13,14) are represented by the interaction torques, which originate from an articulation and are because of the rotation of the adjacent articulations (knees and feet). The interaction torques account for their intersegmental control (9). During a test of isometric force (16), there may increase force variability in response to the demand for force. On the contrary, the strongest muscles (16) show and are able to generate more force with less variability. In general, the muscles of the proximal segment are stronger than the distal muscles (16) and thus may have a leading role (13,14). In fact, with angular rotational velocity, the role of the hip articulation is to control the force and direction of the entire movement (20). Furthermore, a direct but inverse relation is established between the internal rotational angle of the hips and the abduction angle of the feet. During the abduction-adduction of the leg motion, the feet assume an acute angle, enabling the sole of the foot to hit the water and develop an upward push. The great temporal variability that was seen in the intermediate players may be explained also by their inability to manage the interaction torques at the level of the various articulations in an optimal way. For this reason, the movements may be imprecise in using the right degree of orientation of the feet and the legs, the factor on which an optimal upward kick mainly depends (28–30). On the contrary, the ability to use interaction torques to develop greater velocity, power, and control in a multijoint movement is a characteristic of skilled players (12,19,27). In conclusion, the results of the expert players showed the importance of factors such as power, velocity, and better intersegmental control of the lower limbs which, interacting in a balanced way, confirm the results seen in elite female water polo players (25). Furthermore, as the expert players have a consolidated structure of jump capacity, they develop a minimum temporal variability. The intermediate players, on the other hand, have a jumping ability structure limited exclusively to velocity with medium-high temporal variability, which has a decremental effect on their jumping ability.
This study has shown the multifactorial structure of jump capacity, particularly velocity, power, and initial force or rate of force development (RFD). It has highlighted their importance in consideration of the limits imposed by the force-velocity relationship (21). High rates of velocity, power, and initial force in expert players imply explosive force, in particular Verkoshansky's explicative multicomponential model (33,34). It consists of 4 factors: absolute velocity (V o), initial force (Q), acceleration force (G), and absolute force (P o). The brief duration of the propulsive phase of the jump limits the performance of the intermediate players because they are not able to recruit enough motor units in such a short period of propulsion time. A solution may be the use of resistance strength at a high intensity (85–100% 1RM with intended execution) (31), which develops a higher force rate (1–3) or initial force (33,34). This activity also develops neural adaptations (an increase in central descending drive, motor neuron excitability, motor unit firing rates, and a reduction in neuron inhibition) (1–3) and at the same time develops a lower level of muscular hypertrophy (31). This could prove useful when power or acceleration force are developed using a squat jump with a load of 30–45% 1RM or a leg press with a load of between 20 and 50 kp (23). It could also be useful to increase the load by adding heavy chains to the sleeves of the barbell of a Smith's machine (squat jump), whereas the remaining parts are furled upon the floor (7), so that the additional weight of the chains increases in proportion to the extending of the legs, simulating an increase in body weight that occurs on increasing the lifting of the body out of the water. In this way, the players perform an exercise with a high ecological validity. In the water, the players could use a load of between 7 and 10 kg hung on a rope attached to a scuba diver's belt, developing 3 × 3 × 4 jumps with a 10-second recovery between the repetitions and 3 minutes between the sets.
The authors declare that they have had no financial assistance with the project and that there has been no conflict of interests.
1. Aagaard P. Training-induced changes in neural function. Exerc Sport Sci Rev 31: 61–67, 2003.
2. Aagaard P, Meyer F. Neuronal adaptation to strength training. Sports Med 58: 50–53, 2007.
3. Aagaard P, Simonsen EB, Andersen JL, Magnusson P, Dyhre-Poulsen P. Increased rate of development and neural drive of human skeletal muscle following resistance training. J Appl Physiol 93: 1318–1326, 2002.
4. Aagaard P, Simonsen EB, Trolle M, Bangsbo J, Klausen K. Effects of different strength training regimes on moment and power
generation during dynamic knee extension. Eur J Appl Physiol Occup Physiol 69: 382–386, 1994.
5. Andersen LL, Andersen JL, Magnusson SP, Suetta C, Madsen JL, Christensen LR, Aagaard P. Changes in the human muscle force-velocity
relationship in response to resistance training and subsequent detraining. J Appl Physiol 99: 87–94, 2005.
6. Aziz AR, Lee HC, Teh KC. Physiological characteristics of Singapore national water polo team players. J Sport Med Phys Fitness 42: 315–319, 2002.
7. Baker D. Acute negative effect a hypertrofic-oriented training bout on subsequent upper-body power
output. J Strength Cond Res 17: 527–530, 2003.
8. Baker DG, Newton RU. Comparison of lower body strength, power
, acceleration, speed, agility and sprint momentum to describe and compare playing rank among professional rugby league players. J Strength Cond Res 1: 151–158, 2008.
9. Bernstein N. The Coordination and Regulation of Movement. Oxford, United Kingdom: Pergamon Press, 1967.
10. Bosco C. La Valutazione della Forza con il Test di Bosco. Rome, Italy: Società Stampa Sportiva, 1992.
11. Davidson PR, Wolpert DM. Motor learning and prediction in a variable environment. Curr Opin Neurobiol 13: 1–6, 2003.
12. Debecki DB, Gribble PL, Watts S, Hore J. Kinematics of wrist joint flexion in over arm throws made by skilled subjects. Exp Brain Res 154: 382–394, 2004.
13. Dounskaia N. The internal model and the leading joint hypothesis: Implications for control of multi-joint movements. Exp Brain Res 166: 1–16, 2005.
14. Dounskaia N, Swinnen SP, Walter CB. A principle of control of rapid multi joint movement. In: Biomechanics and Neural Control of Posture and Movement, J. Winters, P.E. Crago, eds. New York, NY: Springer-Verlag, 2000. pp. 390–404.
15. Faisal AA, Selen LPJ, Wolpert DM. Noise in the nervous system. Nat Rev Neurosci 4: 292–303, 2008.
16. Hamilton AF, Jones KE, Wolpert DM. The scaling of motor noise with muscle strength and motor unit in humans. Exp Brain Res 157: 417–430, 2004.
17. Harris CM, Wolpert DM. Signal-dependent noise determines motor planning. Nature 394: 780–784, 1998.
18. Heugas AM, Nummela A, Amorim MA, Billat V. Multidimensional analysis of metabolism contribution involved in running track tests. J Sci Med Sport 10: 280–287, 2007.
19. Hirashima M, Kudo K, Watarai K, Ohtsuki T. Control 3D limb dynamic in unconstrained over arm throws of different speeds performed by skilled baseball players. J Neurophysiol 97: 680–691, 2007.
20. Homma M. Three-dimensional analysis of eggbeater kick in synchronized swimming. In: Biomechanics and Medicine in Swimming X. J. P. Vilas-Boas, F. Alves, M. Marques, eds. Porto, Portugal: Porto J Sports Sci 6(Suppl 2): 40–42, 2006. Available at: www.fade.up.pt/rpcd/_arquivo/RPCD_vol.6_supl.2.pdf?
21. Huxley AF. Muscle structure and theories of contraction. Prog Biophys Biophys Chem 7: 255–258, 1957.
22. Klauck K, Daniel K, Bayat M. Goalkeeper's eggbeater kick in water polo: kinematics, dynamics and coordination. In: Biomechanics and Medicine in Swimming X. J. P Vilas-Boas, F. Alves, A. Marques, eds. Porto, Portugal: Porto J Sports Sci 6(Suppl 2): 56–58, 2006.Available at: www.fade.up.pt/rpcd/_arquivo/RPCD_vol.6_supl.2.pdf?
23. Kunz H, Unold E. Speed training. Didat Mov 79: 29–33, 1992.
24. Mc Ardle WD, Katch FI, Katch VL. Fisiologia Applicata allo Sport. Exercise Physiology- Energy, Nutrition and Human Performance. Milan, Italy: Casa Editrice Ambrosiana, 2004.
25. Mc Cluskey L, Lynskey S, Leung CK, Woodhouse D, Briffa K, Hopper D. Throwing velocity
and jump height in female water polo players: Performance predictors. J Sci Med Sport 2: 236–240, 2010.
26. Platanou T, Gelades N. The influence of game duration and playing position on intensity of exercise during match-play in elite water polo players. J Sports Sci 24: 1173–1181, 2006.
27. Salonikidis K, Amiridis IG, Oxyzoglou N, de Villareal ESS, Zafeiridis A, Kellis E. Force variability during isometric wrist flexion in highly skilled and sedentary individuals. Eur J Appl Physiol 107: 715–722, 2009.
28. Sanders RH. A model of kinematic variables determining height in water polo boosts. J Appl Biomech 15: 270–283, 1999.
29. Sanders RH. Analysis of the eggbeater kick used to maintain height in water polo. J Appl Biomech 15: 284–291, 1999.
31. Schmidtbleicher D, Büherle M. Neural adaptation and increase of cross-sectional area studying different strength training methods. In: Biomechanics X-B. B. Johnson, ed. Champaign, IL: Human Kinetics, 1987, pp. 615–620.
32. Van Beers RJ, Haggard P, Wolpert DM. The role of execution noise in movement variability. J Neurophysiol 9: 1050–1063, 2004.
33. Verkoshansky YR. Components and structure of the explosive force effort. Rivista Cult Sport 34: 15–21, 1993.
34. Verkoshansky YR. Means and Methods for Explosive Force Training. Rome, Italy: Società Stampa Sportiva, 1997.