Does Higher Maturation Make Age-Grouped Swimmers Faster? A Study on Pubertal Female Swimmers

Abstract

Background: The main aim of this study was to identify the differences between subgroups of swimmers based on physiological (peak oxygen uptake—VO₂peak), strength (average tethered swimming force—60F_ave), stroke kinematics (v₁₀₀—swimming speed at 100 m front crawl, stroke rate—SR, stroke length—SL), and anthropometrical (i.e.,: biological age—BA, body height—BH, body mass—BM) factors within swimmers at different levels of maturity (BA). Methods: This study involved 39 female swimmers (age: 12.88 ± 0.54 years, BA: 13.98 ± 1.91 years). Cluster analysis (k-cluster) and stepwise multiple regression was performed. Results: Significant correlations were observed between v₁₀₀ and BA, 60F_ave, AS, VO₂peak. Stepwise multiple regression indicated 60F_ave and VO₂peak as the main explanatory variables of v₁₀₀ (R² = 0.60, p < 0.0001). Cluster analysis allowed us to distinguish three groups of swimmers, differing in BA (cluster 1: 14.07 ± 0.96 years, cluster 2: 17.05 ± 1.53, cluster 3: 11.94 ± 0.95) and v₁₀₀, as well as in BH, FFM, AS. Conclusions: There were differences between cluster groups, with early mature swimmers characterized by the highest BH, FFM, AS, 60F_ave, and VO₂peak. Probably, biologically younger late mature swimmers (cluster 3) are slower than the other 2 groups (cluster 1 and 2) because of being less somatically developed. Based on these study results, coaches should ensure further development of aerobic and anaerobic conditioning among normal mature swimmers with simultaneous focus on improving technique skills among early mature ones.

1. Introduction

The performance of young swimmers has been discussed in many studies considering a variety of indices: kinematic [1], physiological [2], biomechanical [3], and anthropometrical [4]. Barbosa et al. [5] claimed that the swimming performance of adolescent swimmers was determined mostly by anthropometrics, biomechanics, and energetics. Dormehl and Osborough [6] stated that when trying to separate performance progress among adolescent swimmers as a result of the training process, the variability caused by growth and development was one of the most important challenges. It is emphasized by the fact that the 100 m freestyle performance of youth swimmers is, by some researchers, believed to be determined mainly by anthropometrics (in boys and girls) [7], but when girls only are considered, it could be that the correlation between swimming performance and somatic indices is not that high [8].

Aerobic power is significantly related to the pubertal development/growth of the adolescent swimmer [9]. Unnithan et al. [10] discovered no correlation between peak oxygen uptake (VO₂peak) and the swimming performance of young female swimmers at a wide range of distances (from 50 to 1000 m). Similarly, Sokołowski et al. [11] reported that $\dot{V}$O₂max values were not significantly correlated to the 100 m front crawl performance of adolescent female swimmers. Higher values of $\dot{V}$O₂ could be connected to higher endurance and speed of swimming. Lätt et al. [12] presented that $\dot{V}$O₂ increased exponentially as a function of swimming velocity; surprisingly, they did not find any correlation between $\dot{V}$O₂ and swimming speed, maybe because of the importance of the anaerobic component. The influence of maturation on anaerobic production has been dealt with in science since a study by Eriksson et al. [13]. During a swimming race, it takes less than a minute before aerobic metabolism contributes to 30–45% of total energy production [5]. Lätt et al. [12] discovered that accumulated blood lactate values in samples collected 3 and 5 min after maximum 100 m front crawl trial were significantly correlated to the time of the mentioned trials (−0.525 and −0.574; p ≤ 0.05).

Swimming technique, described by the indices of stroke rate (SR), stroke length (SL), and stroke index (SI), has been identified as one of the main factors influencing adolescent performance [4,14]. The growing and maturation processes of adolescent swimmers result in anthropometrical changes and, thus, influence technical parameters and performance [15,16]. Lätt et al. [12] concluded that sprint swimming performance was likely to be dependent on the maintenance of high SL and on a stable increase of SR. Improvement in swimming technique is mainly achieved by proper and systematic training.

Tethered swimming constitutes a reliable and often executed test of in-water strength abilities of young swimmers [17]. Papoti et al. [18] linked tethered swimming strength with anaerobic potential and performance at short and middle distances. Taeyements et al. [19], after analyzing data from their longitudinal studies, stated that the peak for static strength among girls interacted with the maturity status from age 11 to 14. It is necessary to constantly monitor the strength abilities of swimmers because the force exerted while stroking is among the main factors providing progress in performance [20].

The aim of this study was to select, from the variety of indices (anthropometrical, physiological, biomechanical), the ones that best explain the 100-m front crawl performance of the involved age-group female swimmers. An attempt was also made to identify, by performing a cluster analysis based on biological age (BA), groups of swimmers characterized by similar sets of indices to examine if there was any diversity in terms of anthropometrics, performance, and metabolic or biomechanical potential in potentially homogenous groups of swimmers (calendar age). We believe that among the measured indices $V{O}_2$peak, 60F_ave, and SR will be the best predictors of 100 m front crawl performance. We also anticipate that cluster analysis will indicate groups of late and early maturing swimmers, with early maturers being at higher performance level (swimming speed), strength level (tethered swimming) and aerobic conditioning level ($\dot{V}$O₂). A study limitation could be that performance was only analyzed on the basis of the 100 m front crawl (short distance).

2. Materials and Methods

2.1. Participants

This study involved 39 female swimmers (age: 12.88 ± 0.54 years, body height [BH]: 159.58 ± 5.28 cm, body mass [BM]: 49.76 ± 7.62 kg, biological age [BA]: 13.98 ± 1.91 years). They were recruited from among the best female swimmers in their age category from the Cracow region, Poland. The participants presented swimming levels which resulted in a mean value of 376 ± 62 World Aquatics points for 100 m front crawl race, which corresponds to the 5th threshold in the Ruiz-Navarro [21] swimming level classification.

All participants were clinically healthy and held a license from the Polish Swimming Federation. They had undergone a 5–6-year systematic training encompassing at least 8 sessions per week and took part in national-level competitions and national swimming championships for their age group.

All subjects and their parents provided informed consent for their participation in exhaustive physical effort during this study (the parents of all participants became acquainted with the study program and a short description of the tests).

2.2. Anthropometrics, Body Composition, and Biological Age

Anthropometric measurements were performed by 2 experienced researchers. To measure BH and arm span (AS), an anthropometer (Sieber Hegner Maschinen AG, Zürich, Switzerland, accuracy of 1 mm) was used. Scales (Tanita BC-418, Tokyo, Japan, accuracy of 0.01 kg) served to determine BM (kg), and body composition (e.g. free fat mass; FFM—kg) with the use of bioelectrical impedance analysis. Bioelectrical impedance analysis is a reliable method of evaluating the tissue composition of the body; its reliability and validity have been recognized in many independent studies [22,23]. BA (BA: 13.98 ± 1.91 years) was calculated by using the following calculation: BA = (BHage + BMage)/2, where BHage was the age obtained from percentile charts (growth charts by the Children’s Memorial Health Institute; 50th percentile was used to align BH with age) on the basis of the participant’s BH, and BMage was the age obtained from percentile charts (growth charts by the Children’s Memorial Health Institute, standardized and validated for the Polish population; 50th percentile was used to align BM with age) on the basis of the participant’s BM.

2.3. The 100 m Front Crawl

The 100 m front crawl trial was performed in a 25 m swimming pool, which met the World Aquatics requirements. The warm-up before the trial consisted of on-land exercises and a 1000 m in-water part. Each trial was carried out by 3–5 swimmers at once to imitate competition conditions. The final results and split times for each participant were measured with an automatic timing device (Omega, Zürich, Switzerland; OCP5, StartTime V, Swiss Timing, Corgémont, Switzerland). The trials were recorded with a camera at a 50 Hz frame rate (GC-PX100BE, JVC, Yokohama-shi, Japan).

The 100 m front crawl velocity (v₁₀₀, m·s⁻¹) was defined as 100 m divided by the final time of the race. The video footage, the placement of the cameras and markers, the video analysis, and the computation of the basic kinematic parameters were performed in the same manner as described in the previous study by the authors [11].

2.4. The 1 min Tethered Swimming Test

The test was conducted in still water, in conditions provided in the previous study [24] (Figure 1). Before the test, the participants underwent a 1000 m swimming warm-up and became familiarized with the test procedure and conditions. During the test, the maximum pulling force 60Fave (N) and the oxygen consumption (L·min⁻¹) for the whole 60 s (breath by breath) were measured. VO₂peak was the average oxygen consumption during the last 30 s of the test.

2.5. Statistical Analysis

The assumptions of normality of residuals, homoscedasticity, and lack of multicollinearity were analyzed with the Shapiro–Wilk and Levene tests and were reasonably met. The descriptive analyses were performed by computing the mean plus one standard deviation maximum and minimum values. To check the association between performance (v₁₀₀) and other variables, Pearson correlation coefficients were calculated.

Nonhierarchical cluster analysis using the k-cluster method was conducted. The following variables were selected for the cluster: v₁₀₀, BA, BH, AS, FFM, SR, SL, 60Fave, VO₂peak. To test for differences between groups/clusters, one-way ANOVA with the Games–Howell post hoc test was performed to check for differences in each variable and between cluster groups. The limit of iterations was set at 10. Partial eta squared (ηp2) was considered as an effect size measure for ANOVA and interpreted in a way proposed by Ferguson [25]: no effect if 0 < |ηp2| ≤ 0.04; a minimum effect if 0.04 < |ηp2| ≤ 0.25; a moderate effect if 0.25 < |ηp2| ≤ 0.64; and a strong effect if |ηp2| > 0.64. The achieved power for the sample of 39 participants was 0.08, 0.25, and >0.93, respectively. In sum, the ANOVA sample size was sufficient to detect large effects.

Multiple linear regression models involving the backward stepwise procedure were developed by entering all variables significantly correlated to the swimming speed. Additionally, indices included in the regression were checked for data redundancy (all > 0.2) and for autocorrelation with the use of the Durbin–Watson test (all ≤ 2.0). Variables were checked for multicollinearity resulting in excluding them from further analysis. The calculations were performed with the PQStat software (version 1.8.6); only power analyses were conducted with the G*Power software (version 3.1). The statistical significance was set at p ≤ 0.05. The study design is presented in Figure 2.

3. Results

The obtained values of the investigated indices are presented in

Table 1. Mean, standard deviation, and minimum and maximum values of the analyzed indices.

	Mean	SD	Min	Max
BA (years)	13.98	1.91	10.80	18.00
60Fave (N)	52.33	9.43	37.28	75.8
VO2peak (L·min−1)	2.18	0.53	1.21	3.6
SR (cycle·min−1)	47.56	9.43	37.28	75.8
SL (m)	1.75	0.17	1.37	2.08
AS (cm)	160.85	51.6	148	173
v100 (m·s−1)	1.38	0.08	1.23	1.53

SD—standard deviation, BA—biological age, 60F_ave—average tethered swimming force, VO₂peak—peak oxygen uptake, SR—stroke rate, SL—stroke length, AS—arm span, v₁₀₀—100 m front crawl swimming speed.

.

The Pearson correlation calculated between all the measured indices revealed the highest correlation between (a) BA and 60F_ave, AS, and VO₂peak; (b) 60F_ave and AS and v₁₀₀; and (c) SL and SR and VO₂peak (

Table 2. Pearson correlation coefficients between selected measured indices.

	BA (Years)	60Fave (N)	AS (cm)	SL (m)	SR (Cycle·min−1)	VO2peak (L·min−1)	v100 (m·s−1)
BA	1.00 p = ---	0.72 p < 0.01	0.76 p < 0.01	0.26 p = 0.11	−0.04 p = 0.81	0.57 p < 0.01	0.40 p = 0.01
60Fave	0.72 p < 0.01	1.0000 p = ---	0.69 p < 0.01	0.15 p = 0.36	0.22 p = 0.17	0.44 p < 0.01	0.69 p < 0.01
AS	0.76 p < 0.01	0.69 p < 0.01	1.00 p = ---	0.27 p = 0.09	−0.06 p = 0.70	0.61 p < 0.01	0.38 p = 0.02
SL	0.26 p = 0.11	0.15 p = 0.36	0.27 p = 0.09	1.00 p = ---	−0.85 p < 0.01	0.51 p = 0.001	0.24 p = 0.13
SR	−0.04 p = 0.8	0.22 p = 0.17	−0.064 p = 0.70	−0.85 p < 0.01	1.00 p = ---	−0.26 p = 0.12	0.30 p = 0.06
VO2peak	0.57 p < 0.01	0.44 p < 0.01	0.61 p < 0.01	0.51 p < 0.01	−0.26 p = 0.115	1.00 p = ---	0.43 p < 0.01
v100	0.40 p = 0.01	0.69 p < 0.01	0.38 p = 0.02	0.24 p = 0.13	0.30 p = 0.06	0.43 p < 0.01	1.0000 p = ---

BA—biological age, 60F_ave—average tethered swimming force, AS—arm span, SL—stroke length, SR—stroke rate, VO₂peak—peak oxygen uptake, v₁₀₀—100 m front crawl swimming speed.

).

The cluster groups with the post hoc Games–Howell test results presenting between-group differences are depicted in

Table 3. Three cluster groups (k-cluster) based on selected indices of swimming performance, anthropometrics, body composition, swimming technique kinematics, tethered swimming force, and peak oxygen consumption (mean and standard deviation) with ANOVA analysis results including the post hoc Games–Howell test.

Variables	Cluster 1 N = 23	Cluster 2 N = 6	Cluster 3 N = 10	F Statistics	p-Value	ηp2	Games–Howell
v100 (m·s−1)	1.39 (0.08)	1.41 (0.05)	1.31 (0.06)	4.30	0.02	0.19	1 vs. 2
							2 vs. 3 *
							1 vs. 3 *
BA (years)	14.07 (0.96)	17.05 (1.53)	11.94 (0.95)	44.03	<0.000001	0.71	1 vs. 2 *
							2 vs. 3 **
							1 vs. 3 **
BH (cm)	160.47 (2.28)	166.02 (2.91)	153.65 (5.62)	25.25	<0.000001	0.58	1 vs. 2 *
							2 vs. 3 **
							1 vs. 3 *
FFM (kg)	38.7 (2.71)	44.73 (3.44)	32.13 (2.98)	37.52	<0.000001	0.66	1 vs. 2 *
							2 vs. 3 **
							1 vs. 3 **
AS (cm)	161.57 (2.13)	168.33 (2.66)	154.7 (4.08)	45.76	<0.000001	0.72	1 vs. 2 *
							2 vs. 3 **
							1 vs. 3 *
SR (cycle·min−1)	46.93 (4.55)	48.42 (5.0)	48.59 (5.35)	0.52	0.60	0.03	1 vs. 2
							2 vs. 3
							1 vs. 3
SL (m)	1.79 (0.13)	1.77 (0.19)	1.65 (0.19)	2.82	0.07	0.14	1 vs. 2
							2 vs. 3
							1 vs. 3
60Fave (N)	53.75 (7.0)	62.5 (10.57)	42.94 (4.6)	15.24	0.00002	0.46	1 vs. 2
							2 vs. 3 *
							1 vs. 3 **
VO2peak (L·min−1)	2.26 (0.45)	2.63 (0.57)	1.71 (0.33)	8.96	0.0007	0.33	1 vs. 2
							2 vs. 3 *
							1 vs. 3 *

v₁₀₀—100 m front crawl swimming speed, BA—biological age, BH—body height, FFM—fat-free mass, AS—arm span, SR—stroke rate, SL—stroke length, 60F_ave—average tethered swimming force, VO₂peak—peak oxygen uptake, ηp²—eta squared (effect size). * p < 0.05, ** p < 0.001.

.

The k-cluster analysis resulted in dividing the participants into three groups. Figure 3 presents the three clusters in a v₁₀₀–BA graph.

The stepwise regression including all the indices significantly correlated to v₁₀₀ resulted in the selection of two indices—VO₂peak and 60F_ave—as significant components of the model. SR, AS, and BA were not significant indices of the model (

Table 4. Stepwise multiple regression based on indices significantly correlated with the 100 m front crawl swimming speed.

	Variables	B	Beta	rsemi	% of Variance	t	p
R2 = 0.603 Adjusted R2 = 0.543 F (5,33) = 10.03; p < 0.00001	60Fave	0.006	0.79	0.47	21.99	4.28	0.0002
	VO2peak	0.058	0.40	0.30	8.91	2.72	0.01
	SR	0.003	0.20	0.18	3.17	1.62	0.11
	AS	−0.0003	−0.23	−0.14	1.85	−1.24	0.22
	BA	−0.009	−0.21	−0.12	1.5	−1.12	0.27

60F_ave—average tethered swimming force, VO₂peak—peak oxygen uptake, SR—stroke rate, AS—arm span, BA—biological age.

).

4. Discussion

In this study, moderate to strong correlations were discovered between v₁₀₀ and BA, AS, SR, VO₂peak, and 60Fave. The stepwise regression analysis including SR, AS, 60Fave, VO₂peak, and BA allows us to explain 54.3% of the variance for v₁₀₀. A decision was taken to exclude SL from the analysis (involving AS) because of (a) a high correlation coefficient between SR and SL and (b) the fact that SL and SR produce swimming speed—both resulting in a potential bias.

The nonhierarchical cluster analysis (k-cluster) resulted in the identification of three groups within all the swimmers taking part in the research. The performed ANOVA confirmed that the three groups differed from one another in all the measured indices (v₁₀₀, BA, BH, FFM, AS, 60Fave, VO₂peak) except for SR and SL. The post hoc Games–Howell tests revealed differences between the clusters. The values of v₁₀₀, 60Fave, and VO₂peak in clusters 1 and 2 were not significantly different despite the large differences in somatic indices, including BA (14.07 vs. 17.05 years). Swimmers in cluster 2 were biologically older than those in clusters 1 and 3 (17.05 vs. 14.07, 11.94), as well as presented larger BH, FFM, and AS. Cluster 3 included late-maturing swimmers, who were significantly slower and exhibited lower values of somatic, strength, and physiological indices than the subjects from clusters 1 and 2. In the authors’ opinion, the most interesting observation from the cluster analysis is that the participants most advanced in maturation (cluster 2), and because of that having an advantage in somatic characteristics, strength level, and aerobic conditioning, were not significantly faster than their biologically less mature peers (cluster 1). This could be due to the high technical aerobic conditioning level of an average maturer (cluster 1).

The results of a multivariate analysis of 100 m adolescent female front crawl performance conducted in the present study, based on physiological, somatic, and biomechanical indices, suggest that 60Fave and VO₂peak are crucial determinant factors of the 100 m adolescent female front crawl performance. Possibly, the biomechanical constraints or proficiency in pulling force generation and energetic proficiency were the main factors that induce a higher swimming speed. Anthropometric indices (AS) did not turn out to be significant components of the stepwise multiple regression model despite the moderate-to-strong linear correlations with swimming speed. Geladas et al. [8] pointed out that anthropometrics with lower body power (BH, hand length, and horizontal jump) explained only 17% of the 100 m front crawl performance.

VO₂peak was identified as a variable strongly connected to v₁₀₀ in this study. Aerobic metabolism is an important factor of successful performance among adolescent swimmers, even in short distances [26]. Troup et al. [27] estimated that in 100 m front crawl, 45–50% of all metabolic energy was aerobic. In the present study, VO₂peak was moderately correlated (r = 0.43; p = 0.006) with v₁₀₀. VO₂peak also constituted a significant component of the presented stepwise multiple regression. Similarly, Sokołowski et al. [11] reported that 100 m front crawl swimming results of young female swimmers (age: 13.4 ± 0.26 years) were correlated with aerobic indices in spite of respiratory compensation points (r = 0.81; p < 0.001) and VO₂peak (r = 0.47; p < 0.04) reached in a step test. As for the VO₂peak values, the cluster analysis revealed significant differences for cluster 1 vs. cluster 3 and for cluster 2 vs. cluster 3, which positioned late maturers as the group with a much lower potential to use the aerobic pathway of muscle energy production. Significant differences in $\dot{V}$O₂ (L·min⁻¹) were also discovered by Baxter-Jones et al. [28] among female swimmers at the age of 12, 13, and 16 years and by Strzała et al. [29] among young boys.

High SR itself does not lead to better performance; it is known that older swimmers note lower SR values than their younger counterparts at submaximal speeds [30]. It is the proper relationship between SR and SL which guarantees a higher swimming speed and younger swimmers are often not able to maintain SL throughout a race [6]. In the present study, SR was close to significant when measuring the linear correlation with the swimming speed (r = 0.30; p = 0.06), while the correlation with SL was not significant. The kinematic indices of technique (SR, SL, SI) are strongly connected to sprint performance in adolescents. Including those indices into a multiple linear regression model could explain 99% (R² = 0.99) of the swimming performance of young swimmers [31]. Morais et al. [1] noted a strong correlation between SI and 100 m front crawl performance of adolescent girls (age: 12.31 ± 1.09 years). SI was not included in the regression model in the current study because SI is a product of SL and v, which in the authors’ opinion could cause bias in the results when v is considered as the performance level. The SL was also excluded from the model (replaced by AS) to avoid multicollinearity. The cluster analysis did not indicate significant differences in SR and SL between the groups; however, the result for SL was close to significant (p = 0.07). Santos et al. [32] examined 53 boys and girls (age: 12.40 ± 0.74 years), creating three tiers on the basis of the swimming speed. They found no significant differences in SR or SL, or in groups with significantly larger BH. More somatically developed (with advanced biological maturation) swimmers due to greater AS and BH are generally faster (at short distances 50–100 m) and perform fewer upper limb cycles over the same distances [33]. It is supposed that the lack of significant differences in performance among standard and early maturers in our study could be due to the well-trained swimming technique of less physically developed swimmers, who are also members of regional competition teams.

Eriksson et al. [13] investigated the influence of maturation on anaerobic energy production and concluded that this pathway of energy production was higher among adults than in children. Our study confirms Eriksson findings, faster maturation process (higher BA) results in higher energy production (60Fave, v₁₀₀); correlation was also found between BA and v₁₀₀ (r = 0.40; p = 0.01). Also, Sokołowski et al. [11] reported moderate-to-strong significant correlations between $\dot{V}$O₂max, Fave, and BA but did not identify a mediation effect of BA with regard to the correlation between v₁₀₀ and SL or SR. The cluster and ANOVA analyses performed in the present study indicated significant differences between the distinguished cluster groups in terms of the individuals’ BA (F = 44.03, p < 0.000001). It is interesting that only late-maturing girls (cluster 3, BA: 11.94 ± 0.95 years) were significantly slower than the biologically older ones (cluster 1: 14.07 ± 0.96, cluster 2: 17.05 ± 1.53), and the differences in v₁₀₀ between clusters 1 and 2 turned out not significant (1.39 ± 0.08 vs. 1.41 ± 0.05 m·s⁻¹).

Conducting cluster analysis is still not performed very often in research regarding swimming. To the best of our knowledge, this is the first study where BA and VO₂peak are included in cluster analysis. Morais et al. [34], performing cluster analysis in groups of age-grouped swimmers (14 boys: 12.70 ± 0.63; 16 girls: 11.72 ± 0.71 years old), came to a conclusion that coaches should design the training and developing programs of swimmers while being aware of the subgroups which exist in the training group. They also presented three clusters based on performance (100 m freestyle time). Similar to our study, the slowest swimmers from the study of Morais et al. [34] were characterized by low BM and BH. Figueiredo et al. [35] also presented a study with cluster analysis of age-grouped swimmers (boys and girls, 11.8 ± 0.8 years-old) focusing on anthropometric and kinematic variables. They concluded that the swimming speed of adolescent swimmers is influenced by anthropometrics, which is consistent with our study results and technique kinematics (SR, SL).

Modeling the age-grouped female swimmers’ performance is believed to be a demanding task [36]; some of the female swimmers compete successfully at the international level at the age of 15 yrs. We claim that researchers and swimming coaches should be very careful in examination of the potential predispositions of adolescent female swimmers to perform at the senior level.

5. Conclusions

In the results, 60Fave and VO₂peak were the only indices that were significant in the multiple linear regression analysis explaining 100 m front crawl swimming speed, which suggests the importance of developing strength and both aerobic and anaerobic conditioning in young swimmers. The early maturers were not always faster than their peers with the standard pace of maturation, which could be due to the lack of significant advantage in technique kinematics and aerobic conditioning. This can lead to a conclusion that among young female swimmers, stroke technique excellence is joined with higher swimming performance and may overcome the “disadvantages” of being at the standard pace of maturation. On the other hand, late maturers were significantly slower swimmers than their peers, which was mainly due to the somatic and physiological indices being at an average, lower level. In our opinion, it is important to preserve the standard maturing individuals in swimming sports at their age group level; often with high technical abilities and aerobic conditioning levels, they are overshadowed by early maturers being at higher performance levels due to their somatic advantage. On the other hand, it often happens that early maturing swimmers (especially females) reach very high performance levels at the junior level and continue to compete in the senior category. Because of that, coaches of early maturing swimmers should also be focused on their further development (mainly technical skills).