Growing up and reaching for the top: A longitudinal study on swim performance and its underlying characteristics in talented swimmers

ABSTRACT

The present study strived to gain a more profound understanding of the distinctions in development between swimmers who are considered to be on track to the elite level at late junior age (males aged 16; females aged 15) compared to those who are not. In this effort, swimmers were followed during their pubertal years (males aged 13–15; females aged 12–14), which marks a period when performance development aligns with maturation. Longitudinal data of 90 talented sprint and middle-distance swimmers on season best times (SBT) and underlying performance characteristics (anthropometrics, maximal swimming velocity, stroke index [SI] and countermovement jump [CMJ]) were collected over three swimming seasons. Based on their SBT at late junior age (males aged 16; females aged 15), swimmers were classified as high-performing late juniors or lower-performing late juniors. Retrospectively studying these swimmers, we found that all but two high-performing late juniors were already on track to the elite level at early junior age (males aged 13; females aged 12), evidenced with faster SBT throughout puberty compared to their lower-performing peers (p < 0.05). Independent sample t-tests revealed that high-performing late juniors significantly outscored their lower-performing peers when they were early juniors on maximal swimming velocity (males aged 13–15 and females aged 12–14), SI (males aged 13 and 14; females aged 12), CMJ (females aged 14) and height (females aged 13 and 14, p < 0.05). Additionally, multilevel models showed faster rates of development for high-performing late juniors on maximal swimming velocity (males and females) and SI (males) compared to lower-performing peers throughout puberty (p < 0.05). Higher initial levels of SBT and underlying performance characteristics at early junior age as well as the faster rates of development on SBT, maximal swimming velocity and SI (males only) during the pubertal years, may be crucial factors in maintaining the trajectory towards the elite level after puberty.

KEYWORDS:

• Growth and maturation
• Talent development
• Acquisition of expertise
• Competitive swimming
• Longitudinal analysis

Introduction

Competitive swimming is a sport where every fraction of a second can make the difference between winning or losing (World Aquatics, 2016). This compels elite swimmers to pursue the perfect race, constantly refining even the smallest details of their performances (ANP, 2017). However, these swimmers did not start out as world-class athletes; they were once aspiring junior swimmers who belonged to a group where only a tiny minority would eventually reach the top (Barreiros et al., 2014; Brustio et al., 2021; Güllich et al., 2023). What characterises their successful development towards swimming expertise compared to their peers who did not make it to the top?

Undoubtedly, a significant element in the progression from competing at local junior meets to excelling at the World Championships is the continuous improvement of swim performance over time. This increase could be attributed to the development of swimmers’ underlying performance characteristics, including anthropometric, physiological, technical, tactical, and psychological factors (Elferink-Gemser & Visscher, 2012). Accordingly, researchers emphasise the importance of conducting multi-dimensional and longitudinal studies to unravel the pathway towards swimming expertise (Cobley & Till, 2017). Yet, such studies are scarce in the literature, leaving a significant gap for further exploration (Morais et al., 2021).

A particularly intriguing period to investigate would be the pubertal years, which marks a period when performance development aligns with maturation (Malina, Bouchard, et al., 2004). Maturation reflects the timing and tempo of progress towards the mature adult state, which highly varies between individuals (Malina, Bouchard, et al., 2004). It is the driving force for many processes, including the adolescent growth spurt, which typically occurs at 12 ± 2 years in girls and 14 ± 2 years in boys (Till et al., 2020). Previous studies have shown a strong relationship between maturation and physical performance indicators such as size, strength, power and speed (Abbott, Hogan, et al., 2021; Lätt et al., 2009; Malina, Eisenmann, et al., 2004; Oliveira et al., 2021). Moreover, Morais et al. (2014, 2022) found that swimmers minimise performance impairment or even progress in periods of detraining due to growth spurts.

Commonly, the pubertal years also signify the time when the initial stages of talent identification processes are carried out (KNZB, 2023). However, due to the potential asynchrony between chronological and biological age (Towlson et al., 2018), accurately assessing a swimmer’s current performance level can be challenging during this key developmental phase (Malina, Bouchard, et al., 2004). Furthermore, given that maturing swimmers undergo natural, yet highly individual and unpredictable improvements in performance, it can be difficult to distinguish between progress resulting from growing up and progress indicative of the potential for future elite-level performances (Malina, Bouchard, et al., 2004). These challenges can create confusion in evaluating a swimmer’s potential and may introduce maturity-related biases that favour early-maturing swimmers and overlook those who mature later in talent selection processes (Malina et al., 2015).

Hence, gaining insight into the development of performance and its underlying performance characteristics (e.g., height, maximal swimming velocity, stroke index and CMJ) throughout puberty, while differentiating between performance levels, is essential to optimise talent identification and development (TID) in swimming. Obtaining a thorough understanding of the developmental pathways during the pubertal years, such as objective insights into skill levels and rates of progression for swimmers who are on track to the elite level in comparison to those who are not, can provide valuable knowledge to contextualise a swimmer’s current performance and future potential. This can facilitate the advancement of science-based, informed decision-making processes, which may lead to more effective and improved strategies in TID.

Therefore, the present study followed swimmers throughout puberty (males aged 12–15; females aged 11–14) and retrospectively analysed their developmental patterns, differentiating by their performance level at the end of puberty (males aged 16; females aged 15). We first examined whether swimmers who are considered to be on track to the elite senior level (referred to as high-performing late juniors) differed from those who are not (referred to as lower-performing late juniors) on levels of swim performance and underlying performance characteristics throughout puberty. Second, we investigated whether developmental differences in swim performance and underlying performance characteristics emerged during the pubertal years based on late junior performance-level attainment. We hypothesised that high-performing late junior swimmers showed better scores and faster rates of development on both swim performance and its underlying performance characteristics than lower-performing late junior swimmers throughout puberty.

Methods

Ethical approval

All participants were informed of the study’s procedures prior to their participation and provided their written informed consent to participate. Informed consent was also obtained from parents of participants who were below 16 years old. All procedures used in the study complied with the Helsinki Declaration and were approved by the research ethics committee of the University Medical Center Groningen, University of Groningen, The Netherlands (202000488).

Participants

Participants were 90 Dutch talented swimmers (47 males, 14.6 ± 1.0 years; 43 females, 13.2 ± 1.1 years) who were followed throughout the junior years of their swimming career (males aged 12–15; females aged 11–14). All swimmers participated in the National Dutch Junior Championships (“Nederlandse Jeugd & Junioren Kampioenschappen”), and were classified as national-level athletes, corresponding to Tier 3 in the classification system proposed by McKay et al. (2022). Swimmers were specialised in sprint (50–100-m; 32 males and 35 females) or middle-distance (200–400-m; 15 males and 8 females) events. According to the age group regulations of the Royal Dutch Swimming Federation (KNZB), swimmers were classified as early junior (males aged 12–13 years; females aged 11–12 years), mid junior (males aged 14–15 years; females aged 13–14 years) and late junior swimmers (males aged 16–17 years; females aged 15–16 years) based on their calendar age on 31st December of the corresponding season (KNZB, 2022). Swimmers’ average performance level at late junior age corresponded to 597 ± 106 World Aquatics Points for males and 571 ± 86 World Aquatic Points for females.

Study design

Longitudinal data on swim performance and underlying performance characteristics were collected over three swimming seasons. Performance data (season best times from all long course swim events) were obtained from (Swimrankings, 2022) at the end of each swimming season. Repeated measures of underlying performance characteristics were conducted during four measurement moments during the National Dutch Junior Championships (see Figure 1). For males and females, the median number of measurements was n = 3, taken over a period spanning from 6 to 18 months.

Figure 1. Timeline for data collection over three seasons. All measurement moments included assessment of height, sitting height, weight, CMJ, and mid-pool sprint tests.

Testing battery

Each measurement moment consisted of land-based tests (anthropometric assessment and the countermovement jump test), followed by a swimming test. Additionally, swimmers provided their date of birth and reported their weekly training hours dedicated to swim (in-water) training using an online questionnaire (see Table 1).

Table 1. Descriptive characteristics of male and female swimmers according to their age category and performance-level group at late junior age.

		Males (N = 47)			Females (N = 43)
		High-performing late juniors	Lower-performing late juniors	Effect sizes	High-performing late juniors	Lower-performing late juniors	Effect sizes
		M ± SD	M ± SD	d	M ± SD	M ± SD	d
All	Age of peak height velocity (APHV)	13.1 ± 0.4	13.2 ± 0.5	−0.3	11.5 ± 0.3*	11.7 ± 0.3	−0.9
	rST at late junior age (%)	112.0 ± 2.3***	123.6 ± 6.4	−2.3	114.1 ± 1.9***	123.3 ± 6.1	−1.7
12 years	Age (years)	-	-	-	12.4 ± 0.3	12.5 ± 0.3	−0.2
	Swim training (hours per week)	-	-	-	7.8 ± 1.4	6.9 ± 2.2	0.4
	Height (cm)	-	-		167.8 ± 7.3	164.1 ± 5.0	0.7
	CMJ (cm)	-	-	-	42.7 ± 5.0	42.0 ± 4.0	0.2
	rMSV (%)	-	-	-	95.6 ± 3.4***	90.0 ± 1.9	2.5
	rSI (%)	-	-	-	82.7 ± 4.3**	75.9 ± 4.6	1.5
	rST at early junior age (%)	-	-	-	122.3 ± 2.5***	126.7 ± 2.2	−1.9
13 years	Age (years)	13.6 ± 0.3	13.7 ± 0.2	−0.4	13.4 ± 0.3	13.6 ± 0.3	−0.7
	Swim training (hours per week)	9.8 ± 2.3	8.4 ± 3.1	0.5	9.3 ± 1.4	9.4 ± 2.3	−0.1
	Height (cm)	175.4 ± 9.2	176.3 ± 7.0	−0.1	171.2 ± 5.7*	167.2 ± 5.3	0.8
	CMJ (cm)	51.1 ± 2.0	50.3 ± 4.6	0.2	44.6 ± 4.9	42.4 ± 4.3	0.5
	rMSV (%)	95.2 ± 3.7***	86.6 ± 4.5	2.0	96.1 ± 2.7 **	92.2 ± 3.1	1.3
	rSI (%)	77.9 ± 6.4*	74.5 ± 4.4	0.6	82.7 ± 4.5	80.7 ± 4.2	0.5
	rST at early junior age (%)	123.1 ± 2.2***	128.4 ± 3.0	−1.9	119.1 ± 2.6***	123.6 ± 2.4	−1.8
14 years	Age (years)	14.6 ± 0.3	14.6 ± 0.2	−0.2	14.5 ± 0.3	14.6 ± 0.3	−0.5
	Swim training (hours per week)	11.7 ± 2.6	10.1 ± 3.3	0.5	11.1 ± 3	10.9 ± 2.8	0.1
	Height (cm)	180.6 ± 9.0	180.1 ± 7.2	0.1	174.0 ± 3.2*	168.8 ± 4.7	1.2
	CMJ (cm)	52.5 ± 2.0	51.3 ± 4.8	0.3	46.5 ± 3.6*	42.7 ± 4.1	1.0
	rMSV (%)	97.6 ± 3.1***	90.4 ± 3.8	2.0	98.5 ± 2.6**	94.2 ± 3.3	1.4
	rSI (%)	82.6 ± 7.1*	78.2 ± 4.2	0.8	86.1 ± 5.4	84.5 ± 2.1	0.5
	rST at early junior age (%)	119.5 ± 2.4 ***	124.3 ± 2.3	−2.1	115.6 ± 1.9***	120.1 ± 2.2	−2.1
15 years	Age (years)	15.7 ± 0.2	15.6 ± 0.2	0.6	-	-	-
	Swim training (hours per week)	11.6 ± 3.0	11.3 ± 3.1	0.1	-	-	-
	Height (cm)	186.9 ± 3.7	183.4 ± 7.3	0.6	-	-	-
	CMJ (cm)	54.4 ± 2.4	52.8 ± 4.5	0.4	-	-	-
	rMSV (%)	99.0 ± 2.9***	92.5 ± 3.4	2.0	-	-	-
	rSI (%)	89.8 ± 6.7**	82.3 ± 5.5	1.3	-	-	-
	rST at early junior age (%)	115.2 ± 1.7***	121.0 ± 1.9	−3.1	-	-	-

Notes. 41 freestyle (20 males); 13 butterfly (7 males); 22 breaststroke (13 males); 14 backstroke (7 males) swimmers. Age refers to the calendar age as of December 31st of the corresponding season. CMJ=countermovement jump; rMSV = relative maximal swimming velocity; rSI = relative stroke index; rST = relative swim time. CMJ, rMSV, rSI, and rST at early and late junior age are reported as predicted values instead of observed values. *p < 0.05 (one-tailed); **p < 0.01 (one-tailed); ***p < 0.001 (one-tailed).

Anthropometric measures

Swimmers were measured for height, sitting height and body mass with 0.1 cm or 0.1 kg precision. Height was assessed using a stadiometer (Seca, 217, Seca GmbH & Co.KG, Germany), and sitting height was measured using a standard box (height 45 cm) positioned at the stadiometer’s base. Body mass was measured using a digital scale (Beurer, GS 300, Beurer GmbH, Germany). Measures were taken twice and conducted by the same two researchers. The mean value was documented. A third measure was taken if the difference between the first two exceeded 0.4 cm. The median was then recorded.

Maturity status was estimated using a non-invasive method developed by Moore et al. (2015). This approach involves sex-specific calculations that determine the maturity offset of young adolescents, expressed in terms of years before or after Peak Height Velocity (YPHV). By subtracting YPHV from a swimmer’s chronological age, the predicted age of PHV (APHV) was calculated. However, it is important to acknowledge that accurately measuring biological maturity remains a challenging task, as highlighted by ongoing discussions in the literature that emphasise the complexities involved in this process (Malina et al., 2021)

Countermovement jump (CMJ) test

Swimmers were instructed to perform two double-leg vertical countermovement jumps (CMJ) with arm swing, which is reported as a valid and reliable test to measure lower body power (Markovic et al., 2004). Lower body power is considered to be of particular importance during starts and turns, as it is in these moments that the lower extremities must generate the greatest impulse to achieve the highest accelerations off the block and wall, respectively (Jones et al., 2018; Keiner et al., 2021; West et al., 2011). The jumps began from an upright position, and there was a 30 s break between each trial to allow the swimmers to return to the starting position. Each trial was recorded with a vertical jump metre (Takei, TKK5406, Takei Scientific Instruments Co., Ltd., Japan). The maximal jump height (in cm) was taken as indicator of lower body power and taken as outcome measure for further analyses (Gajewski et al., 2018).

Mid-pool sprints

Swimmers were instructed to perform one 25-m distance sprint at maximal swimming velocity. They initiated their effort from the midpoint of a 50-m pool, specifically at the 25-m mark. Starting from a static position, they immediately accelerated to full speed and maintained this pace until they touched the wall, signifying the completion of their effort. Swimmers performed the sprint effort in their best stroke, while wearing racing suits. Sprints were recorded with a digital video camera (HC-X1000 Camrecorder, Panasonic Netherlands, Netherlands), positioned on the lateral side of the pool at 15-m from the start. Kinematic data were collected by means of time video analysis. Maximal swimming velocity was defined as the clean swimming velocity (10-m distance divided by time for the 10-m distance, m/s) between the 10- and 20-m segment of the 25-m trial. Regardless of distance and stroke, this parameter is crucial for any swimmer aiming to touch the wall first (Barbosa et al., 2010), given that clean swimming predominates in (long course) swimming events (Gonjo & Olstad, 2020). Stroke rate (Hz) was calculated as the number of strokes completed by the swimmer during this 10-m segment (Poujade et al., 2002), one stroke rate cycle being defined as the time between the entry of one hand until the following entry of the same hand (Huot-Marchand et al., 2005). Stroke length (m) was calculated as the ratio between swimming velocity over the 10-m segment and the corresponding stroke rate (Poujade et al., 2002). Stroke index (SI), an indirect measure of swimming efficiency, was calculated by multiplying swimming velocity by stroke length. The SI measures the ability of the swimmer to complete a given distance with a particular speed in the fewest possible number of strokes (m²/s) (Costill et al., 1985). Maximal swimming velocity and SI were taken as outcome measures for further analyses.

Data processing

To enable meaningful comparisons among swimmers specialised in different strokes and distances, outcomes were related to relevant reference values and expressed as a percentage, rather than absolute values (see Equation 1(1) $\mathit{relative}\mathit{variable}\mathit{x}=\frac{\mathit{absolute}\mathit{variable}\mathit{x}}{\mathit{reference}\mathit{valuex}}\times 100\mathrm{\%}$(1) ). This approach is essential because direct comparisons of absolute values in swim performance and test scores within our sample could potentially lead to misconceptions. For instance, it is widely acknowledged that the breaststroke is inherently slower to perform than the freestyle (Moser et al., 2020). Similarly, when considering distance, it is evident that the duration of an event increases with the length of the distance to travel (Moser et al., 2020). Taking these stroke-specific and distance-related nuances into account ensures a more accurate evaluation of swimmers’ capabilities.

Consequently, swim time was related to the prevailing world record (WR), a method initially introduced by Stoter et al. (2019) in speed skating and subsequently applied in competitive swimming (Post, Koning, Visscher, et al., 2020; Post, Koning, Stoter, et al., 2020). Lower percentages on relative Swim Time (rST) indicated swim performances closer to the WR. Moreover, scores on swimming tests were related to the average start time, turn time, clean swimming velocity and SI of male and female finalists at the European Championships in 2021 (Born et al., 2022). Stroke-specific data of the 100- and 200-m events were used as reference values for sprinters (50–100-m) and middle-distance (200–400-m) swimmers in our sample, respectively (see Appendix A). Higher percentages on relative maximal swimming velocity (rMSV) and stroke index (rSI) indicate scores more close to the European elite level (set to 100%). For example, the maximal swimming velocity of an early junior male freestyle sprinter (1.85 m/s) was related to the average clean swimming velocity of the 100-m freestyle European male finalists (1.98 m/s), resulting in a rMSV of 93.4% ((1.85/1.98)*100%).

(1) $\mathit{relative}\mathit{variable}\mathit{x}=\frac{\mathit{absolute}\mathit{variable}\mathit{x}}{\mathit{reference}\mathit{valuex}}\times 100\mathrm{\%}$(1)

Data selection

In cases where swimmers had multiple data points within a season, the swimmers’ season best rST, rMSV along with the corresponding rSI, CMJ and anthropometric scores were selected for further analyses (see Appendix B for number of measurements by performance-level group and age category). Any other data were excluded, minimising the impact of variations in achievements within a season. The median number of between-season observations was n = 2 in males and females.

Defining performance level groups

A higher- and lower-level performance group were defined according to performance trajectories of international elite swimmers, representing a performance level similar to the top 50 swimmers worldwide of the past 5 years (2017–2022 with the exception of 2020, see Post, Koning, Visscher, et al., 2020). Following the approach adopted in previous studies (Stoter et al., 2019; Post, Koning, Stoter, et al. (2020), the maximum season best rST by age category, sex and swim event of these international elite swimmers was used as performance benchmark (%WR, see Appendix C). Swimmers whose season best rST at late junior age (males aged 16; females aged 15) fell within the corresponding performance benchmark were categorised as high-performing late juniors and considered to be on track to reach the elite level (16 males; 10 females). Conversely, swimmers who did not meet the performance benchmark were classified as lower-performing late juniors and considered to be off track to reach the elite level (31 males; 33 females). To illustrate, consider a 16-year-old male swimmer competing in the 100 m freestyle. If his season best rST is 115.1%, he would be classified in the high-level performance group since it falls within the performance benchmark for 16-year-old males in the 100 m freestyle, which is set at 116.3%. However, if his season best rST is 117.8%, he would be classified in the lower-level performance group as it exceeds the corresponding performance benchmark

Statistics

All data were analysed for males and females separately, using R (R Core Team, 2021). Data were initially screened on outliers (using box plots), normality (using QQ-plots) and homogeneity of variance (using Levene’s test). Outliers (16 in males; 19 in females) were acknowledged as a natural occurrence within the population and, consequently, were not removed from the dataset. Normality was violated in males (rST at early and late junior age) and females (height, rMSV and rST at early and late junior age). Homogeneity of variance was assumed with the exception of CMJ (males) and rST at late junior age (males and females).

Cross-tabulation analyses were performed to analyse the relationship between performance-level group at late junior (males aged 16; females aged 15) and early junior age (males aged 13; females aged 12). For high- and lower-performing late juniors, mean scores and standard deviations were calculated for swim performance and underlying performance characteristics at the beginning of their junior years (males aged 13; females aged 12). Independent sample t-tests were included to examine between-group differences on age, swim training (hours per week), height, CMJ, rMSV, rSI, rST at early junior age and rST at late junior age (to ensure correct definition of our performance groups). Mann–Whitney U tests were included to examine between-group differences on variables in which assumptions were violated. For all tests, p < 0.05 (one-tailed) was considered statistically significant.

To interpret the scores, effect sizes (Cohen’s d values) were calculated. An effect size of approximately 0.20 was considered small, while effect sizes of 0.50, 0.80 and 1.20 were considered medium, large and very large, respectively (Cohen, 1988). A sensitivity power analysis confirmed that our statistical tests were sufficiently sensitive to detect significant differences between performance-level groups with a minimum detectable effect size of 0.8 and 0.9 (males and females respectively) (alpha = 0.05, power = 0.80). Statistical tests for measuring invariance were not performed given the nature of our dataset (relatively few observations for many items).

Longitudinal multilevel models were created to describe development of rST, rMSV, rSI and CMJ (dependent variables) as a function of (chronological) age, using the lmer4 package in R (R version 3.6.0). The age effect (which was used as measure for development over time) was not imposed to be identical between high- and lower performing late juniors. Therefore, a nested interaction between age and performance-level group at late junior age was included. To represent these two performance-level groups in the statistical models, one dummy variable (high-level performance group) was included and the lower-level performance group functioned as reference level. Each swimmer’s individual trajectory was accommodated through the estimation of a random intercept model, allowing the intercept to vary between swimmers while remaining constant within measurements of the same swimmer. In sum, the following multilevel model was adopted:

${\mathrm{Y}}_{\mathrm{is}}={\mathit{\alpha }}_{\mathrm{i}}+{\mathit{\beta }}_1\mathit{\hspace{0.17em}}\times \mathrm{Ag}{\mathrm{e}}_{\mathrm{is}}+\mathit{\hspace{0.17em}}{\mathit{\beta }}_2\times (\mathrm{Ag}{\mathrm{e}}_{\mathrm{is}}\times \mathrm{High}-\mathrm{level}\mathrm{performance}\mathrm{grou}{\mathrm{p}}_{\mathrm{i}})\mathit{\hspace{0.17em}}+\mathit{\hspace{0.17em}}{\mathrm{u}}_{\mathrm{i}}+{\mathit{\epsilon }}_{\mathrm{is}}$

${\mathrm{u}}_{\mathrm{i}}\sim \mathit{\hspace{0.17em}}\mathrm{N}\left(0,{\sigma }_0^2\right)$

(2) ${\mathit{\epsilon }}_{\mathrm{is}}\mathit{\hspace{0.17em}}\sim \mathrm{N}\left(0,{\mathit{\sigma }}^2\right)$(2)

${\mathrm{Y}}_{\mathrm{is}}$ was the dependent variable (e.g., rMSV) for swimming season $\mathrm{s}$ of swimmer $\mathrm{i}$, ${\mathit{\alpha }}_{\mathrm{i}}$ the intercept of swimmer $\mathrm{i}$, $\mathrm{Ag}{\mathrm{e}}_{\mathrm{is}}$ the corresponding age value and $\mathrm{High}-\mathrm{level}\mathrm{performance}\mathrm{grou}{\mathrm{p}}_{\mathrm{i}}$ the dummy variable indicating whether or not swimmer $\mathrm{i}$ was in the high-level performance group. The unexplained information was the sum of ${\mathrm{u}}_{\mathrm{i}}$ (between-subject variance) and ${\mathit{\epsilon }}_{\mathrm{is}}$ (residual variance). The models were validated by using visible patterns in residual plots to check violations of homogeneity, normality and independence. Predictor variables were considered significant if the p value of the estimated mean coefficient was smaller than 0.05.

Results

Table 1 shows the descriptive statistics, including effect sizes, of male and female swimmers according to their performance level at late junior age (males aged 17; females aged 16) by age group. High-performing late juniors outscored lower-performing late juniors on rST at late junior age (p < 0.001; very large effect sizes), confirming a correct definition of performance-level groups in both males and females. Furthermore, high-performing late junior females demonstrated an earlier age of PHV compared to lower-performing females (p < 0.05). No significant differences between high- and lower-performing male and female swimmers on age and weekly swim training hours were found (p > 0.05). Figure 2 provides a visual representation of the mean scores, interquartile range, as well as the minimum and maximum scores for age of PHV, height, and swim training hours.

Figure 2. Mean (indicated with X), interquartile range (indicated with darker colours), minimum and maximum (indicated with lighter colours) observed values on age of peak height velocity (APHV), height and swim training hours for males and females according to performance level at late junior age.

High-performing late junior males scored significantly higher on rMSV at age 13, 14 and 15 (all p < 0.001), rSI at age 13, 14 (p < 0.05) and 15 ((p < 0.01), and rST at age 13,14 and 15 (all p < 0.001), compared to lower-performing peers. The effect sizes for rMSV and rST were found to be very large, while for rSI, they showed an increasing trend from medium to very large. Although not statistically significant, high-performing late junior males had higher scores on CMJ (small-to-medium effect sizes) at age 13, 14 and 15, and height at age 15 (medium effect sizes) compared to lower-performing males. Similar scores were found between groups on height at age 13 and 14 (no effect).

High-performing late junior females scored significantly higher on rMSV at age 12, 13 and 14 (p < 0.001 at age 12 and 13, p < 0.01 at age 14), rSI at age 12 (p < 0.01), CMJ at age 14 (p < 0.05), height at age 13 and 14 (p < 0.05) and rSBT at age 12, 13 and 14 (all p < 0.001) compared to lower-performing peers. The effect sizes for rMSV and rST were found to be very large while for CMJ and height the effect sizes were considered large. Although not statistically significant, high-performing late junior females had higher scores on CMJ (small-to-medium effect sizes) at age 12 and 13, and height at age 12 (medium-to-large effect sizes) compared to lower-performing females.

Table 2 shows the cross-tabulation analyses of the relationship between performance-level group at late and early junior age of male and female swimmers. At late junior age (16 years), 15 of the 45 male swimmers (33%) were classified in the high-level performance group. All fifteen high-performing male late juniors (100%) were also categorised as high-performing early juniors (13 years), whereas 27 out of the 42 (64%) high-performing male early juniors switched to the lower-level performance group at late junior age. For females, ten of the 43 swimmers (23%) were classified in the high-level performance group at late junior age (15 years). Eight high-performing female late juniors (80%) were also categorised as high-performing early juniors (12 years), whereas 21 out of the 29 high-performing early junior females (72%) switched to the lower-level performance group at late junior age.

Table 2. Cross-tabulation analyses of the relationship between performance-level group at early and late junior age of male and female swimmers.

	Total (N)	High-performing early juniors (n)	Lower-performing early juniors (n)
Males	45 (100%)	42 (93%)	3 (7%)
High-performing late juniors	15 (33%)	15 (100%)	0 (0%)
Lower-performing late juniors	30 (67%)	27 (90%)	3 (10%)
Females	43 (100%)	29 (67%)	14 (33%)
High-performing late juniors	10 (23%)	8 (80%)	2 (20%)
Lower-performing late juniors	33 (77%)	21 (64%)	12 (36%)

Notes. Performance groups are based on season best performances at age 12 for males and age 11 for females. Two male swimmers (1 high-performing late junior and 1 lower-performing late junior) could not been classified at early junior age due to missing data and left out of this analysis (N = 47 adjusted to N = 45).

Developmental models according to performance level group at late junior age

Table 3 shows the developmental models on rSBT, rMSV, rSI and CMJ created for males and females. Each model consists of two age effects, which allows for different rates of development between high- and lower-performing late juniors. The “age” term denotes the development of lower-performing late juniors, whereas “age + age × high-level performance group” denotes the development of high-performing late juniors. To illustrate (using the fixed effects of the model only), the rSBT for a high-performing male late junior swimmer at age 14 will be predicted as follows:

(3) $\mathit{rSBT}=176.80+\left(-3.46\ast 14\right)+\left(-0.37\ast 14\right)=123.18$(3)

Table 3. Model estimates for male (N = 47 with 107 observations) and female (N = 43 with 100 observations) swimmers.

	rSBT		rMSV		rSI		CMJ
	Estimates (S.E.)	P-value	Estimates (S.E.)	P-value	Estimates (S.E.)	P-value	Estimates (S.E.)	P-value
Males
Fixed effects
Intercept	172.62 (4.56)	<0.001	55.51 (7.30)	<0.001	10.45 (14.07)	0.230	34.14 (8.92)	<0.001
Age	−3.16 (0.31)	<0.001	2.26 (0.49)	<0.001	4.52 (0.95)	<0.001	1.16 (0.60)	0.029
Age × high-level performance group	−0.34 (0.07)	<0.001	0.53 (0.11)	<0.001	0.33 (0.16)	0.023	0.07 (0.11)	0.264
Random Effects
σ^2	4.73		8.66		45.97		17.37
τ00	8.40		24.18		40.26		20.87
Marginal R^2/Conditional R^2	0.61/0.86		0.40/0.84		0.26/0.60		0.05/0.57
Females
Fixed effects
Intercept	169.03 (4.66)	<0.001	66.86 (5.80)	<0.001	23.87 (13.67)	0.043	41.63 (7.52)	<0.001
Age	−3.31 (0.35)	<0.001	1.84 (0.43)	<0.001	4.15 (1.02)	<0.001	0.04 (0.56)	0.422
Age × high-level performance group	−0.28 (0.08)	<0.001	0.29 (0.10)	0.002	0.14 (0.19)	0.244	0.18 (0.14)	0.098
Random Effects
σ^2	5.35		7.14		51.27		10.90
τ00	6.62		9.81		30.66		21.45
Marginal R^2/Conditional R^2	0.55/0.80		0.30/0.72		0.17/0.48		0.03/0.67

Notes. “Age” term denotes development of lower-performing seniors. “Age + Age × high-level performance group” denotes development of high-performing swimmers. Abbreviations: rSBT (relative season best time); rMSV (relative maximal sprint velocity); rSI (relative stroke index). Significant predictor variables (p < 0.05) are denoted in bold with adjustments made for one-sided hypothesis testing in the calculation of the p value.

Given the study’s primary focus on differences between high- and lower-performing swimmers, particular emphasis will be placed on analysing the interaction term (age × high-level performance group). A significant interaction term would indicate a faster rate of development of high-performing swimmers compared to their lower-performing peers.

In males, high-performing late junior swimmers showed significant faster progression over time on rSBT (+11%, p < 0.001), rMSV (+22%, p < 0.001) and rSI (+7%, p < 0.05) compared to lower-performing late junior swimmers (+). In females, high-performing senior swimmers showed significant faster progression over time on rSBT (+12%, p < 0.001) and rMSV (+20% p < 0.01). No significant developmental differences between groups were found on rSI for females and CMJ in males and females (p > 0.05). Figure 3 (males) and Figure 4 (females) reflect the predicted development of high- and lower-performing late juniors during their pubertal years.

Figure 3. Predicted development as function of age (mean ± SD) of swim performance and underlying performance characteristics in males (N = 47 with 107 observations).

Figure 4. Predicted development as function of age (mean ± SD) of swim performance and underlying performance characteristics in females (N = 43 with 100 observations).

Discussion

The present study investigated the development of swim performance and its underlying performance characteristics throughout puberty, differentiating between swimmers who were on track to the elite level (referred to as high-performing late juniors) and those who were not (referred to as lower-performing late juniors) at late junior age (males aged 16; females aged 15). Retrospectively studying these swimmers, we found that high-performing late juniors outperformed their lower-performing peers on most of the assessed underlying performance characteristics during the pubertal years (males aged 13–15; females aged 12–14). Furthermore, high-performing late juniors were characterised with significantly faster development in season best performances, maximal swimming velocity and SI (males only) throughout puberty.

Performance

Our findings showed that all high-performing late junior swimmers (except for two females) were already on track to the elite level at early junior age (males aged 12; females aged 11). Additionally, these swimmers demonstrated significantly faster season best performances throughout puberty (males aged 13–15; females aged 12–14) compared to their lower-performing peers. This trend aligns with the finding that top-elite swimmers (best 8 world-wide) progressively outperformed their lower-performing peers, starting from the age of 12 (Post et al., 2020). As such, our results suggest that achieving higher levels of swim performance at early junior age may signify a minimal level of proficiency, serving as a prerequisite for further progression towards swimming expertise. This observation aligns with the work of Yustres et al. (2019) in competitive swimming, and research in cycling (Gallo et al., 2022; Mostaert et al., 2022). However, it is important to note that our findings also reveal that only a minority of the high-performing swimmers at early junior age was able to sustain their performance level until the end of puberty (36% in males; 28% in females). This demonstrates that early achievements in itself do not necessarily guarantee successful development towards higher performance levels, a notion supported by previous studies in both competitive swimming and other sports (Barreiros et al., 2014; Brustio et al., 2021; Güllich et al., 2023). Instead, in line with Brustio et al. (2022), our findings underscore that the significantly faster progression in season best performances shown by high-performing late juniors (+11% in males and + 12% in females as reported in our study) holds equal, or perhaps even greater importance than current performance in the advancement towards swimming expertise.

Underlying performance characteristics

Aligning with the progressive trends observed in season best performances, our findings demonstrate that high-performing late junior swimmers had significantly higher levels, as well as faster progression (+22% in males; +20% in females) on maximal swimming velocity compared to lower-performing peers throughout puberty. Furthermore, they exhibited significantly higher levels of stroke index (SI), a measure of technical ability (males aged 12–14, females aged 12), with significantly faster advancements for high-performing late junior males in this aspect as well (+7% in males). Shifting our focus to the land-based tests, we found that high-performing late junior females were significantly taller at age 13 and 14, and demonstrated higher CMJ at age 14 compared to lower-performing peers, whereas surprisingly, no significant between-group differences were found for males on these variables.

Taken together, swimmers who are on track to the elite level at late junior age (males aged 16; females aged 15) are characterised with the ability to attain higher swimming speed with equal (females aged 13–14) or even better levels of technical efficiency than lower-performing juniors throughout puberty. Given that competitive swimming centres around maintaining optimal power output in an efficient and skilful manner throughout the event (Miyashita, 1996), this could be a critical factor in the attainment of swimming expertise. Additionally, being taller, particularly as a female aged 13–14, may be a beneficial characteristic for superior swim performances post-puberty. This advantage can be attributed to the strong relationship between longer lengths, such as height, and increased stroke length and speed (Morais et al., 2021).

Maturation and training

Considering our findings, it is important to acknowledge that performance and its underlying performance characteristics may be influenced by inter-individual differences in timing of PHV as well as training hours. Regarding the former, we found a significantly earlier onset of PHV in high performing late junior females (~2.4 months) as well as within-group variations of 1.0 to 1.5 years in age of PHV (females and males respectively). While we cannot disregard the possibility that relatively early maturing swimmers in our study may have experienced physical advantages compared to relatively later maturing swimmers, we do not expect that these variations significantly affected our findings. This anticipation is grounded in the observation that the between- and within-group differences are considerably small, particularly compared to the 5 to 6 years difference between players’ chronological and biological age reported by Johnson et al. (2009).

In the context of training, we observed that high-performing late junior swimmers tend to engage in slightly more swim training hours per week, with this trend being more pronounced among males compared to females. Interestingly, the minimum training hours per week are consistently higher among high-performing late juniors. Within performance-level groups, we noticed differences of more than 10 hours between swimmers who had the lowest and highest amount of swim training per week. Given that the total amount of (deliberate) training is correlated with attainment (Baker & Young, 2014), it is likely that such notable differences may advantaged swimmers with access to a higher number of training compared to those with fewer hours. However, while we acknowledge that these benefits may be reflected in our results on the individual level, we expect that the impact on our overall findings will be limited given the subtle and minor variations observed between high- and lower-performing swimmers.

Strengths and limitations

The present study comprised a wide range of talented swimmers as we included participants from the Dutch Junior National Championships, rather than solely inviting swimmers from national talent development programme, who are typically the top performers of their age group. We followed this relatively large and heterogenous group of swimmers over time (varying from 6 to 18 months) and monitored their development on swim performance and multiple underlying performance characteristics. Using this multidimensional and longitudinal design, which is scarce in literature, we acquired insights into swimmers’ developmental patterns (skill levels and progression rates) during the pubertal years.

By comparing the scores of maximal swimming velocity and SI to the reference values of international senior elite swimmers, we not only enabled comparisons between swimmers specialising in different events but also gained insights into a swimmer’s position relative to the standard set by these top performers. Furthermore, we created performance-level groups based on a swimmer’s performance level at late junior age relative to the performances of international senior elite swimmers of the same age in the past. While this approach doesn’t make direct predictions about senior performances, our classification of performance-level groups does take future achievements into account. As a result, our findings may provide insight into swimmer’s potential for later success, which offers scientific and practical value for talent development in swimming.

However, when interpreting the results, it is important to acknowledge that our findings pertain to pubertal swimmers (males aged 13–15; females aged 12–14) who qualified for the Dutch National Junior Championships. Considering that our measurements were conducted during this particular event, it is important to recognise that swimmers who did not meet the qualification criteria were unable to participate in the measurements. Furthermore, reflected by the lower mean ages of PHV (13.1 years for males; 11.6 years for females) compared to the average ages of 14 in males and 12 in females (Malina, Bouchard, et al., 2004), we observed an overrepresentation of early-maturing swimmers throughout the sample. In light of the physical advantages associated with early maturation, it is possible that the swimmers in our study were more likely to qualify for the Dutch National Junior Championships compared to late-maturing swimmers. This potentially introduces a survivorship bias in our results (Baker et al., 2022).

Moreover, we examined swimmers who are currently in the midst of their development, analysing a limited set of underlying performance characteristics. Notably, we did not examine potential interactions between variables which are undoubtedly at play (Abbott, Yamauchi, et al., 2021; Barbosa et al., 2010). Furthermore, given the dynamic nature of performance development, the relative contribution of underlying performance characteristics may vary among different specialisations in our sample and change over time (Morais et al., 2015; Vaeyens et al., 2008). For example, lower body power (measured as jump height), is considered to be more critical for swimmers oriented towards sprinting (Keiner et al., 2021). In terms of timing, it may emerge as a more distinguishing factor in later stages of development, particularly after puberty, as training programmes tend to place a greater emphasis on the development of strength and power (KNZB, 2023). Therefore, it is important to highlight that the present study captures merely a glimpse of the long and complex developmental pathway towards swimming expertise, leaving a lot of opportunities for future research to expand upon our findings.

Perspective

The present study enhances our understanding of advancement towards elite-level swimming performance. Specifically, it underscores the significant role of levels and progression of maximal swimming velocity, SI, and season best performances throughout puberty in males aged 13–15 and females aged 12–14. In addition, height and CMJ emerged as noteworthy characteristics in females. Coaches could focus on developing these factors and monitor their swimmers’ progression towards the elite level in relation to the developmental patterns of high-performing late juniors. However, coaches should consider these findings as a starting point for further development rather than an endpoint, and take inter-individual differences in maturation and training into account when evaluating swimmers’ current performance and future potential.

Moreover, our findings show that differences between high- and lower-performing juniors manifest at least at early junior age (males aged 13; females aged 12) and emphasise the difficulty of closing that gap thereafter. Therefore, it would be interesting to further investigate swimmers’ development from the start of their career. Furthermore, given that high-performing late juniors still have a long road to go before reaching the top, it is recommended to continue monitoring swimmers after puberty. In both cases, gaining insight into swimmers’ training programmes, including factors such as the number of sessions, training hours and metres per week, the employed training methods (Nugent et al., 2017), and indicators of the quality of training (Post et al., 2022) would be highly valuable. This is essential to not only further unravel but also ensure sustained progression towards elite-level swimming performance.

Author statement

A. K. P: conceptualization, investigation, data curation, formal analysis, drafting and writing the original manuscript; R. H. K: conceptualization, formal analysis; C. V.: conceptualization, supervision, review, and editing; M.T. E-G: conceptualization, supervision, review, and editing.

Acknowledgments

We would like to thank the swimmers for participating in the measurements as well as the coaches, parents, Royal Dutch Swimming Federation and InnoSportLab de Tongelreep for their support, thereby enabling us to conduct this study.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplemental material

Supplemental data for this article can be accessed online https://doi.org/10.1080/02640414.2024.2322253

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.

Appendix A.

References values of key performance indicators of European male and female finalists (retrieved from Born et al., 2022).

Table

	N swimmers	n swimmers measured for 2 seasons	n swimmers measured for 3 seasons	N obs.	n obs. (at age 11)	n obs. (at age 12)	n obs. (at age 13)	n obs. (at age 14)	n obs. (at age 15)
Males (N = 47)
High-performing late juniors	16	11	5	37	-	0	12	15	10
Lower-performing late juniors	31	23	8	70	-	9	19	23	19
Females (N = 43)
High-performing late juniors	10	6	4	24	2	6	10	6	-
Lower-performing late juniors	33	23	10	76	13	20	27	16	-
Total	90	63	27	207	15	35	68	60	29

Notes. N swimmers = total number of unique swimmers participating in the study; n swimmers = number of swimmers measured in one through three seasons; N obs. = total number of season best observations; n obs. = number of season best observations per age category.

Appendix B.

Number of swimmers measured in one through three seasons and number of season best observations per age category during the pubertal years.

Table

	Clean swimming speed (m/s)	Stroke index
Males (100-m events)
Backstroke	1.77	3.81
Breaststroke	1.60	2.80
Butterfly	1.84	3.64
Freestyle	1.98	4.63
Males (200-m events)
Backstroke	1.62	3.81
Breaststroke	1.48	3.71
Butterfly	1.68	3.41
Freestyle	1.81	4.54
Females (100-m events)
Backstroke	1.58	3.11
Breaststroke	1.43	2.56
Butterfly	1.63	2.84
Freestyle	1.77	3.84
Females (200-m events)
Backstroke	1.48	3.23
Breaststroke	1.34	3.04
Butterfly	1.50	2.63
Freestyle	1.64	3.69

Appendix C.

Performance benchmarks (%WR) by age category, sex and swim event derived from international elite swimmers.

Table

	Event	Age category	Performance benchmark (%WR)
Males	50 Backstroke	16	117.9
	100 Backstroke	16	116.8
	200 Backstroke	16	114.3
	50 Breaststroke	16	118.4
	100 Breaststroke	16	117.6
	200 Breaststroke	16	115.3
	50 Butterfly	16	117.8
	100 Butterfly	16	116.6
	200 Butterfly	16	117.5
	50 Freestyle	16	117.7
	100 Freestyle	16	116.3
	200 Freestyle	16	116.0
	400 Freestyle	16	114.3
	200 Medley	16	114.9
	400 Medley	16	114.5
Females	50 Backstroke	15	113.9
	100 Backstroke	15	114.4
	200 Backstroke	15	115.2
	50 Breaststroke	15	117.7
	100 Breaststroke	15	115.8
	200 Breaststroke	15	117.4
	50 Butterfly	15	120.3
	100 Butterfly	15	118.7
	200 Butterfly	15	114.6
	50 Freestyle	15	115.9
	100 Freestyle	15	115.0
	200 Freestyle	15	112.2
	400 Freestyle	15	112.7
	200 Medley	15	114.1
	400 Medley	15	113.9