Karate competitions are divided into "kata" (forms) and "kumite" (match). Kata is characterized by individual and team (three athletes) competitions of standardizes routines, while kumite is a free form of sparring against an opponent. Official karate kumite matches are characterized by 3-minute bouts of high intensity activity involving kicking, punching and quick horizontal displacements (WKF, 2009).

Athletes execute an average of 16.3 ± 5.1 of these decisive actions during the whole match resulting in a total high intensity action time of 19.4 ± 5.5 sec in 3-min matches. Although single kicks and punches are anaerobic alactic and dependent on muscle power, the repetition of these motor actions turns the aerobic pathway the major contributor (77.8% ± 5.8%) to the total energy produced during a kumite match (Beneke et al., 2004). Beneke et al., 2004 also reported that the anaerobic alactic contribution (16.6%) closely represents the time percentage of high intensity actions (16%) during an entire kumite match. However, recovery from these high intensity actions is primary based on aerobic metabolism, and thus explain the large contribution of the aerobic energy pathways for the total energy used.

In addition, anaerobic lactic metabolism can also contribute to the total energy during kumite match. This contribution might be estimated by assessing blood lactate (La) before and after the match. Beneke et al., 2004 showed that La increased 5.9 ± 1.6 mmol·l^-1 after a single kumite match (from 1.7 to 7.6 mmol·l^-1). Lehmann, 1997 reported similar post kumite match La values. These increases suggest some contribution of the glicolytic metabolism to the total energy expenditure during a kumite match.

Despite the metabolic characteristics of karate, the main criteria to score in kumite matches is the vigorous application of kicks and punches (WKF, 2009). Both actions are performed without external loads and, usually, as fast and powerful as possible. The unique characteristics of the kumite motor actions require a strength assessment that may be able to categorize athletes at distinct competitive levels. For instance, Zehr et al., 1997 observed higher values of loaded (i.e. 10% of their maximal voluntary isometric contraction) and unloaded elbow extension velocity in experienced compared to novice karate players. Further, taekwondo (Toskovic et al., 2004) and judo athletes (Franchini et al., 2005) of different levels show similar 1RM values. Thus, it is possible that karate performance is more dependent on muscle power at low loads than at high loads or on maximal dynamic strength (1RM). Therefore it seems appealing to identify the association of usual power and maximum strength assessments with the kumite performance.

Even though literature presents scattered information regarding strength, power and physiological data (Beneke et al., 2004; Zehr et al., 1997) it is difficult to build a profile of karate players. Moreover, a link between neuromuscular variables and performance results in a competition-like simulation is still lacking. Thus, the aim of this study was to verify the relationship of strength and power with performance on an international level karate team during official kumite match simulations.

Participants
Fourteen male black belt karate athletes, all members of the Brazilian Karate National Team took part in the study (age 28.0 yrs ± 5.1; height 1.78 m ± 6.6; weight 73.1kg ± 10.5). Participants were tested for power and strength parameters and then paired by weight class for a judged competition-like-simulation. All participants were National Champions and had competed in high level international events (World Championship). These athletes were in a direct dispute for a starting position on the National Team roster, i.e., they were the best two athletes for each category in Brazil. The study was approved by the institutional ethic committee and all subjects were informed of the inherent risks and benefits, before signing an informed consent form. Athletes were tested during their competitive period.

Testing routine
Tests were conducted on two different days at the same time of day. During the first day anthropometric measurements, vertical jump test, and two maximum dynamic strength (1RM) tests (upper and lower limbs) were conducted, while in the second day subjects performed two power tests (upper and lower limbs) and the kumite match simulation. Thirty minutes separated the tests performed on the same day.

Anthropometric measurements
Body weight was measured on a digital scale to the nearest 100g. Skinfold thickness was measured in triplicate with a Harpenden caliper from six body sites (i.e., triceps, brachial, subscapular, suprailiac, abdominal, calf, and thigh). Measurements were made on the right side of the body by the same experienced researcher (10 years experience and less then 1% variation between measurements, with reproducibility determined by and intra-class correlation coefficient of 0.98 within the assessment performance period).

Vertical jump test
Countermovement jumps were performed on a resistive platform attached to a digital timer (JumpTest^®, Belo Horizonte, Brazil). Subjects were instructed to keep their hands on the hips and to jump as high as possible maintaining the same body position during take-off and landing. Before the actual test, subjects were allowed to practice between 6 to 10 submaximal jumps. Subjects performed five maximal vertical jumps with 15s interval between trials.

Maximum dynamic strength tests
Maximum dynamic strength (1RM) for squat and bench press exercises was assessed on a Smith machine (Nakagym^®, São Paulo, Brazil). For the squat exercise, subject's body and feet positioning were determined and recorded with measuring tapes fixed on the bar and on the ground, respectively. Then, subjects performed familiarization trials to get acquainted with the test procedures. The procedures for this test respected ASEP guidelines (Brown and Weir, 2001). In short, subjects ran for five minutes in a treadmill at 9 km/h, followed by lower and upper limbs stretching exercises. Before each exercise, subjects should perform a specific warm-up set of 5 repetitions at approximately 50% of the estimated 1-RM followed by another set of 3 repetitions at 70% of the estimated 1-RM. Warm-up sets were separated by a 2-minute interval. After the completion of the second set, subjects rested for 3 minutes. Subsequent lifts were single repetitions of progressively heavier loads, until failure. Maximum dynamic strength (1RM load) was determined as the maximum weight that could be lifted once with proper technique. The interval between 1RM attempts was set at 3 minutes, and a maximum of 5 attempts were allowed. Test was accompanied by two experienced researchers. Increments in weight were determined according to researchers' perception. Strong verbal encouragement was provided during all lifts.

Bench press 1RM was assessed with the subjects lying on a horizontal bench. Body and hand positioning on the bench and on the bar were recorded and the test was performed according to the same procedures described above.

Power tests
The power produced during the squat and bench press exercises was assessed on a Smith machine (Nakagym^®, São Paulo, Brazil). A linear encoder (Peak Power^®, Cefise, São Paulo, Brazil) was attached to the equipment's bar to register its position through the repetitions at a frequency of 50 Hz. Finite differentiation technique was used to estimate bar velocity and acceleration (variability coefficient <3%). Then, force and power were calculated using standard procedures (Bosco et al., 1995). Body, foot (squat) and hand (bench press) positioning were reproduced using the previously recorded positions. In addition, during the squat a wood seat with adjustable heights was placed behind the subjects to keep bar displacement and knee angle constant on each repetition; for the bench press a 3cm foam was placed on subject's chest to ensure that bar traveling depth was kept constant (since the bar should touch the foam in each repetition). Both squat and bench press power production was assessed through a six repetition set, using two different loads, 30 and 60% of correspondent 1RM values for each exercise.

Kumite simulated match
In order to address the relationship between the physiological parameters and performance, a judged championship-like kumite match was performed. As mentioned before, subjects were paired according to their weight classes. The subjects were instructed to maintain their usual preparation routine for a competition event. After a warm up period, the kumite was designed in an identical way to the corresponding matches of the Pan American Games of 2007 (i.e. 3 minutes bouts, with timing starting and stopping at each referees' signal) (WKF, 2009).

Blood sampling and lactate analysis
Blood samples (25µl) were obtained at rest and immediately after the kumite matches at the earlobe after application of a vasodilator pomade. Lactate was analyzed using a portable lactate analyzer (Accusport, Boehringer Mannheim, Germany), which has been validated by comparing it with usual laboratory methods with similar results (Fell et al., 1998).

Statistical analysis
After the match, subjects were separated by performance (winners vs. defeated) on the kumite simulated match. Independent t- tests were performed for between group comparisons. Correlation between variables was assessed by a Pearson's correlation coefficient. Significance level was set at p < 0.05.

No significant differences were found between winners and defeated for strength, vertical jump height and anthropometric data. Table 1 presents anthropometric values separated by group.

Table 2 presents jump height, strength and power data for both groups. Statistical analysis showed significant differences between winners and defeated athletes for bench press and squat power using 30% 1RM, while no difference was found at 60% 1RM (Table 2). Blood lactate (La) values obtained pre and kumite matches are also presented in Table 2. Statistical analysis did not show significant differences between winners and defeated for La.

The sum of skinfold thickness were significantly and negatively correlated to both jump height (r = -0.69, p < 0.05) and squat power at 60% 1RM (r = -0.54, p < 0.05). Maximum strength was correlated with absolute (30% 1RM r = 0.92; 60% 1RM r = 0.63) and relative power (30% 1RM r = 0.74; 60% 1RM r = 0.11, p > 0.05) for the bench press exercise.

• Muscle power at low workloads seems to be a reasonable predictor of karate performance.
• There are differences in neuromuscular characteristics between winners and defeated karate players among an international level karate team.
• Karate players rely more on muscle power, rather than on muscle strength.