Melatonin Levels and How Training and Match Times Affect Sleep, Recovery, and Body Clocks in Young Pro Football Players

This descriptive, comparative, cross-sectional, and non-experimental study examined the effect of training at different times of day on circadian markers and sleep quality in professional male football players under 19 years of age. The participants were drawn from the first division of Spanish youth football, the highest competitive level for this age group. This study compared players with similar characteristics and identical training programs, differing only in their training schedules (morning training (MT) vs. evening training (ET)), using both objective and subjective markers (melatonin rhythm, chronotype, sleep quality, and daytime sleepiness). To control for potential environmental influences on melatonin production, the light intensity was measured at each team’s training location during their respective training times using a lux meter. The MT group was exposed to natural sunlight, while the ET group trained under artificial lighting due to visibility constraints. The light intensity values obtained (53,626 lux for the MT group and 4154 lux for the ET group) are known to suppress melatonin production, with studies showing that exposure to light intensities of between 200 and 400 lux for less than two hours can cause significant suppression [19,20,21]. This is especially relevant for the ET group, as melatonin naturally starts to be produced later in the day. Additionally, we also performed anthropometric and nutritional assessment measurements by training group to account for any differences between the groups. All study protocols and procedures were conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of the University of Granada (protocol code 3340/CEIH/2023, 18 May 2023 and PPJIB2021-22, 16 May 2022). The players were fully informed about the study’s objectives and methods, and each provided written informed consent before participation. For underage players, legal guardians were informed, and they authorized the minor’s participation through the corresponding informed consent.

Two youth-category teams from the same Spanish club participated in the study, with a total of 40 players: 17 in the MT team (age 17. 59 ± 0.62 years) and 23 in the ET team (age 16.61 ± 0.58 years). The inclusion criteria were as follows: (a) being in good health and possessing medical clearance for federated sports; (b) belonging to one of the two youth-category teams in the Spanish league; (c) being a federated football player; (d) training either in the morning (9:00–10:30 a.m.) or in the evening (20:00–21:30 p.m.); and (e) undergoing measurements in March during the competitive phase. The exclusion criteria were (a) not being within the eligible youth-category age range of 16–18 years and (b) taking melatonin supplements.

Evaluations were conducted during each team’s training schedule. On the day of evaluation, the players were instructed to refrain from high-intensity exercise, training, or stretching. Consequently, anthropometric measurements were taken before training. Specifically, the anthropometric measurements were taken on the same day as the melatonin sampling began but before any training activities. The players were instructed to avoid intense physical activity for at least 24 h prior to saliva sampling. No training was conducted on the saliva collection day, and training was resumed the following day. All the players were already familiar with the procedures, as they routinely performed them as part of their usual activities. Afterward, the players were instructed on how to autonomously collect saliva samples and were provided with the necessary materials. The saliva collection took place midweek (between Wednesday and Thursday) to allow at least 72 h of recovery after the higher-intensity weekend match. The samples were then returned to the club and sent to the laboratory for analysis. The various sleep-related and nutritional questionnaires were explained beforehand, and the players completed them during the same session. Any doubts were addressed to ensure clarity and standardization in their completion.

Saliva was used to determine melatonin levels, as it is a non-invasive method that correlates well with blood melatonin levels [22,23]. The participants were instructed to not eat or drink for 30 min before sample collection and to avoid exposure to blue light during night-time collections. The participants were given a verbal and written explanation of what constitutes blue light, including examples of common sources, such as smartphones, tablets, televisions, and LED lighting. They were specifically instructed to avoid screens and bright artificial lights for at least one hour before and during sample collection, especially for night-time samples. The saliva was self-collected using Salivettes^® (Sarstedt AG & Co. KG, Sarstedtstrabe, Nümbrecht, Germany), which contain a neutral cotton swab, at six different time points: 4 and 2 h before bedtime, at bedtime, 2 and 4 h after bedtime, and upon waking. The date and season of collection were recorded. To preserve the sample integrity, the participants were instructed to refrigerate the daytime samples, covered with aluminum foil, until shipment. The night-time samples could be kept at room temperature until the following morning. All the samples were then placed in a padded envelope and returned to the club, from where they were sent to the laboratory via a refrigerated courier service to ensure delivery within 24 h. Upon arrival, the samples were centrifuged, pipetted, and stored at −80 °C until the analysis. The saliva samples were analyzed at the Centro de Investigación Biomédica of the University of Granada, following a method established by our group and previously used in a study with prostate cancer patients [24]. Briefly, the samples were thawed, and melatonin was extracted using a liquid–liquid method with chloroform. For each 250 μL saliva sample, 1 mL of chloroform was added to glass vials, then shaken for 10 min at 2500 rpm and centrifuged at 3900 rpm for 10 min. The organic phase (800 μL) was transferred to new vials, and chloroform was evaporated under a vacuum using a SPD2010 Savant SpeedVac (ThermoFisher Scientific S.L., Alcobendas, Madrid, Spain). The residues were reconstituted in 100 μL of 5% acetonitrile, shaken, and then injected into the UltiMate 3000 UHPLC system (ThermoFisher Scientific S.L., Alcobendas, Madrid, Spain) for chromatographic separation. A volume of 25 μL of the sample was injected into a reverse-phase Hypersil GOLD C18 column (2.1 mm × 100 mm, 1.9 μm) with a mobile phase of 0.1% formic acid in water (A) and acetonitrile (B) at a flow rate of 0.4 mL/min. A linear gradient was applied from 5% to 98% B over 7 min, held at 98% for 2.1 min, and equilibrated for 0.9 min back to 5% B. The total analysis time was 10 min. The autosampler and column oven were set at 10 °C and 45 °C, respectively. Melatonin was detected using an Orbitrap Q-Exactive Focus mass spectrometer (ThermoFisher Scientific S.L., Alcobendas, Madrid, Spain) in positive electrospray ionization (ESI) mode, with selective ion monitoring (SIM) for the protonated melatonin ion (m/z 233.12845) at a retention time of 3.98 min. Melatonin concentrations were quantified using an external calibration curve. Any subjects with undetectable or excessively high melatonin levels were excluded from the analysis. The average melatonin level in pg/mL, amplitude (maximum melatonin peak) in pg/mL, acrophase (time of maximum melatonin peak) in decimal hours, and dim light melatonin onset (DLMO) in decimal hours were determined. The average melatonin circadian rhythm was represented for each group and further stratified by chronotype. The median of the MT group was used as the cutoff to categorize low or high average melatonin levels and amplitudes. The acrophase and DMLO were classified as early or late using the same criteria [24].

Chronotypes were determined using the Horne and Östberg Morningness–Eveningness Questionnaire [25], consisting of 19 questions about morning and evening habits, with a total score ranging from 16 to 86 points. Based on the total score, the participants were classified into the following chronotypes: definitely evening type (16–30 points), moderately evening type (31–41 points), neither type (42–58 points), moderately morning type (59–69 points), and definitely morning type (70–86 points). From this, three chronotypes were determined in our study: (1) moderately evening type (ME), (2) neither type (N), and (3) moderately morning type (MM).

Sleep duration and quality were evaluated using the Pittsburgh Sleep Quality Index (PSQI) [26], based on sleep habits during the month prior to the interview. Sleep duration was calculated by subtracting the sleep onset from the sleep end. The PSQI includes 19 questions grouped into seven components: subjective sleep quality, sleep latency, sleep duration, sleep efficiency, sleep disturbances, the use of medication, and daytime dysfunction, each scored from 0 to 3 points. The total score (0–21) indicates the overall sleep quality, with higher scores reflecting greater difficulty. The participants were classified as having good (≤5 points) or poor sleep quality (>5 points). Daytime sleepiness was assessed using the Epworth Sleepiness Scale (ESS) [27], which measures the likelihood of dozing or falling asleep in various everyday situations. The ESS consists of eight items scored on a scale from 0 (no risk of dozing) to 3 (a high risk of dozing), resulting in a total score ranging from 0 to 24, with higher scores indicating greater daytime sleepiness. The participants were classified based on their total score: a score of ≤9 was considered as normal sleepiness (low-risk sleepiness, LR), whereas a score of ≥10 indicated excessive daytime sleepiness (high-risk sleepiness, HR), possibly indicating a pathological condition.

Anthropometric data were gathered following ISO 7250-1:2017 guidelines [28] and the standards of the International Society for the Advancement of Kinanthropometry (ISAK) [29]. The measurements included two basic parameters (body mass and height), six skinfold thicknesses (triceps, subscapular, supra-iliac, abdominal, thigh, and calf), and three circumferences (relaxed arm, waist, and hip). A CESCORF inextensible metal tape (CESCORF, Tristeza, Porto Alegre/RS, Brazil) was employed to measure the circumferences, and a Holtain mechanical caliper (HOL-98610ND) with 0.2 mm precision was used for the skinfold assessments. Each anthropometric measurement was taken two or three times. Specifically, when the technical measurement error between the first two measurements surpassed 5% for skinfolds or 1% for the other measurements, a third measurement was performed. The mean or median average was then calculated, as appropriate, for subsequent analyses. This protocol helps ensure precision and consistency in the data, thereby reinforcing the validity of the research findings and facilitating a more accurate interpretation of any differences observed. The room temperature was standardized at 24 °C during all the measurements. The Faulkner formula [30] was used to calculate the fat mass as a percentage. Two health indicators were also derived: the waist-to-hip ratio and the body mass index (BMI).

A 24 h recall questionnaire (R24h) was employed to retrospectively assess and quantify food and beverage intake during the 24 h prior to the interview [31,32]. This method enables the estimation of the energy and nutrient content of the subjects’ diets. On this occasion, each participant completed three 24 h recalls—two on weekdays and one on a weekend day—to enhance the accuracy. Experienced interviewers assisted in administering the questionnaire to improve the data quality and minimize any recall bias. The DIAL software (version 3.15, 2021, Alce Ingenieria, Madrid, Spain) was used to process the R24 h results, translating the reported food items into nutrient profiles for further analysis. The basal metabolic rate (BMR) was calculated using the Harris–Benedict formula [33]. The total energy expenditure (TEE) was subsequently estimated in accordance with the method proposed by the FAO, WHO, and the United Nations University [34], which uses the BMR as its basis. A physical activity level (PAL) factor of 2.15—appropriate for men performing heavy physical activity—was applied to determinate the (TEE) estimate.

In the descriptive analysis, the mean and standard deviation were calculated for the continuous quantitative variables, and percentages were determined for the categorical variables, in order to compare characteristics between the MT and ET groups. The categorical variables were examined using either the Chi-square or Fisher’s exact tests to determine the significance of observed differences, whereas continuous variables were analyzed with the Mann–Whitney U test or Student’s t-test, as appropriate. Additionally, the circadian rhythm of melatonin was plotted for both groups—morning and evening—and further stratified according to chronotype. Effect sizes were calculated to complement the interpretation of the statistical analyses. For the continuous variables, Cohen’s d with 95% confidence intervals was computed, where values of 0.2, 0.5, and 0.8 were interpreted as small, moderate, and large effects, respectively. For the categorical variables, Cramér’s V was used to assess the strength of association, with thresholds of 0.1 (small effect), 0.3 (moderate effect), and 0.5 (large effect). All the statistical tests were performed as two-sided, and the statistical significance was defined at p < 0.05. The analyses were conducted using R version 4.0.2 (R Foundation for Statistical Computing, Vienna, Austria), along with ggplot2 version 3.4.2.

A total of 40 the players, 17 in the morning training group and 23 in the evening training group, were evaluated. Table 1 presents the anthropometric measurements by training group. No significant differences were observed in weight, height, BMI, or most skinfold thicknesses. The subscapular skinfold showed a borderline non-significant difference between the groups (p = 0.051). In contrast, the suprailiac skinfold thickness was significantly higher in the morning training group compared to the evening training group (30.95 ± 1.54 mm vs. 28.91 ± 1.68 mm; p = 0.002).

As shown in Table 2, the BMR and TEE were similar between the morning and evening training groups (p = 0.448). Similarly, there were no significant differences in the total energy intake (EI), water intake, or macronutrient intake (protein, carbohydrates, and lipids), although the clinical significance is important because of its similarity to the methodological recommendations to estimates for these population groups [32,33,34]. The proportions of energy from proteins, carbohydrates, and lipids were also similar (all p > 0.30), and no significant differences were found in cholesterol or fiber intakes.

Table 3 shows the differences in the melatonin level, amplitude, DLMO, and acrophase between the two training groups. The MT group showed a higher mean melatonin concentration (52.49 ± 41.81 vs. 29.93 ± 21.19 pg/mL; p = 0.026), with a moderate-to-large effect size (Cohen’s d = 0. 68; CI: 0.06–1.36). A post hoc power analysis was conducted based on the observed effect size for melatonin concentrations. Given the available sample size, the achieved statistical power was 67% (α = 0.05, two-tailed). Although slightly below the conventional 80% threshold, this level of power is considered acceptable given the moderate-to-large effect size observed.

Similarly, the DLMO was also significantly earlier in the MT group (21.20 ± 1.36 vs. 21.93 ± 1.48 h; p = 0.023), with a greater proportion of individuals exhibiting an early DLMO (<20.5 h) than in the ET group (41.1% vs. 8.7%; p = 0.040). No significant differences were observed in melatonin amplitude or acrophase between the groups.

The average melatonin levels were consistently lower in the ET group than in the MT group at almost all time points, except 2 h before bedtime (around 22:30 p.m. for the ET group), which corresponds to the ET group’s acrophase (Figure 1), indicating a phase advance. Furthermore, this group showed a steeper decline in melatonin levels compared to the MT group, possibly due to the inhibition of melatonin caused by light exposure during night-time training. In contrast, the MT group exhibited a normal melatonin rhythm, with a typical acrophase occurring after bedtime (around 2:15 a.m. for the MT group).

When comparing average salivary melatonin rhythms by both training group and chronotype, the differences between the training groups become even more pronounced (Figure 2). In the MT group, the players with the MM and ME chronotypes exhibited higher average melatonin levels compared to those with the N chronotype. In the ET group, the players with the ME chronotype had higher average melatonin levels than the other chronotypes. Regarding circadian melatonin rhythms, all the chronotypes in the MT group showed a defined acrophase, which aligned well with their respective chronotypes. In contrast, the ET group exhibited a complete loss of the melatonin circadian rhythm across all the chronotypes.

Table 4 summarizes the sleep-related variables for both groups. The participants in the MT group woke up significantly earlier (7.22 ± 0.40 vs. 8.58 ± 1.54 h; p < 0.001), slept fewer total hours (6.95 ± 0.40 vs. 7.95 ± 1.31 h; p = 0.014), and reported good sleep quality more frequently than the ET group (70.6% vs. 47.8%; p < 0.001). The chronotype distribution also differed between the groups, with a higher proportion of ME in the ET group (75.1% vs. 47.1%; p < 0.001) and a lower proportion of MM and N (16.6% vs. 35.3% and 8.3% vs. 17.6%, respectively; p < 0.001). Daytime sleepiness did not differ significantly between the groups (p = 0.678).

Figure 3 shows a comparison of means values and 95% confidence intervals for the acrophase, circadian melatonin mean (MESOR), amplitude, DLMO, and total sleep time by the training group, sleep quality, daytime sleepiness risk, and chronotype. There were notable differences between these subgroups, particularly in the training group, sleep quality, and chronotype. Notably, the sleep duration was the shortest in the morning training group and among individuals with a morning chronotype. Those players with good sleep quality had an early acrophase and DMLO, but a lower MESOR and amplitude.

This study has several limitations that should be considered when interpreting the results. The cross-sectional design prevents establishing causal relationships between the training time, melatonin secretion, and sleep quality, highlighting the need for longitudinal studies to assess long-term adaptations. Another important limitation of this study is the imbalance in group sizes between the morning and evening groups. This imbalance may have reduced the statistical power and affected the stability of the effect size estimates. Although a post hoc power analysis suggested an acceptable power for the main comparison, future studies should aim for balanced group sizes to improve the internal validity. Furthermore, the study sample was composed of elite youth athletes from the same team, which limited the number of available participants. While this enhances sample homogeneity and control, it may reduce the generalizability of the results to other athletic populations. Additionally, the sleep parameters were assessed using self-reported questionnaires, which may introduce a recall bias. The absence of objective performance assessments limits our ability to determine whether disrupted circadian rhythms and reduced sleep quality directly translate into performance impairments. In addition, including a recovery perception scale, such as the Total Quality Recovery (TQR) scale, would have enriched the interpretation of the sleep and melatonin data. While environmental and lifestyle factors, such as diet and training load, were controlled, other variables, including screen exposure before bedtime, stress, and individual sleep habits, may have influenced the results. In addition, the absence of any formal control for caffeine intake or psychological stress may have introduced variability into the melatonin secretion and sleep outcomes. Future studies should incorporate standardized screening for caffeine intake and validated tools for assessing stress (e.g., PSS or RESTQ-Sport). Furthermore, although the artificial lighting at training locations was measured, individual variations in light exposure outside training were not considered, which could have impacted melatonin secretion and sleep patterns. In these future studies, the target population should also be instructed about the use of social media [] and screens before going to bed in the case of experimental studies. Lastly, core body temperature, a key factor in sleep onset and recovery, was not measured, warranting further investigation to understand its role in training-related circadian disruptions. Despite these limitations, this study provides valuable insights into the influence of training schedules on circadian rhythms and sleep quality in elite youth football the players, emphasizing the need for further research in this area. ³⁹