Introduction

The gut microbiota constitutes a vital microecological system within the human body and plays a crucial role in the development of the immune system, nutrient absorption, and metabolic regulation1. Studies have indicated that imbalances in the gut microbiota may lead to health issues in children, such as obesity, allergies, and chronic inflammation2,3. Physical activity (PA) is defined as physical exercise that involves energy expenditure and health benefits resulting from skeletal muscle contraction. It is measured in terms of intensity, frequency, duration, and type of program and is commonly assessed using the International Physical Activity Questionnaire Short Form (IPAQ-S) questionnaire for large samples4,5. Physical fitness (PF) is a comprehensive concept that evaluates individuals’ fitness levels based on their morphological characteristics, functional capabilities, and overall quality. China has a nationally standardized measurement tool for the PF of students, the National Standards for Students’ Physical Health (2014 revision)6. An increasing body of research suggests a positive correlation between the diversity of the gut microbiota in children and their PF levels and physical activity (PA)7,8,9,10. Certain beneficial bacterial strains, such as Faecalibacterium prausnitzii, have been shown to have a significant positive correlation with motor skills and PF in children11. While existing studies have examined the relationship between children’s gut microbiota, PF levels, and PA, current research in this area remains relatively scarce and warrants further exploration10. As research progresses, it has been discovered that exercise training can modulate the gut microbiota profile in obese children and inhibit inflammatory signaling pathways, indicating the potential role of gut microbiota in the healthy development of children9. Increasing evidence has demonstrated that PA can positively influence the occurrence of chronic diseases and immune-related conditions through active modulation of the gut microbiota12.

Despite the importance of PA in the development of PF, the relationship between PF and PA is not entirely positive. The development of PF in children is influenced not only by PA but also by a variety of other factors such as home environment, sleep, diet, schoolwork, etc13., resulting in specific scenarios such as high PA-low PF or high PF-low PA14. Microbial communities may affect an individual’s metabolism and physical capabilities, thereby influencing the overall health status15. However, current research often conflates PF and PA levels, lacking separate analyses, particularly in specific pediatric populations, which highlights a research gap in this field. This study aimed to compare the gut microbiota composition and function among four groups of children (low PA-high PF, high PA-high PF, high PA-low PF, and low PA-low PF) to determine whether differences exist, thereby providing a scientific basis for developing more targeted strategies for promoting PF and intervening in the gut microbiota in children.

Methods

Monitoring

Subjects

This study was based on a monitoring population from primary schools across 12 districts (cities and counties) in Weifang City, conducted in 2023. The selected subjects were healthy, developmentally normal children without congenital or hereditary diseases who were capable of self-care and had basic physical capabilities with no contraindications to PA.

Sampling method

Monitoring was conducted using a stratified sampling method according to the GDP rankings of each district (city, county) to ensure the representativeness of the subjects. The sampling task for each district (city, county) was evenly distributed among three sampling areas to ensure an equal number of samples from various demographic groups (gender, age group, urban/rural classification). During random cluster sampling, the number of classes drawn had to satisfy the minimum requirements for investigation and sampling. A total of 6,074 children aged 6–9 years participated in the study.

Testing method

The monitoring included PF indicators and a lifestyle questionnaire. PF indicators included body mass index, lung capacity, 50-meter run, and sit-and-reach, with scoring criteria derived from the National Standards for Students’ Physical Health (2014 revision)6. The tests used designated equipment for national PF monitoring in China. IPAQ-S was used to assess children’s PA levels. Data recording and scoring were conducted by trained personnel, with PF scores (for assessing PF levels) and physical activity (MET-minutes/week) (for assessing physical activity levels) calculated using standard formulas. Ages 6–7: PF = BMI score × 0.15 + Vital Capacity score × 0.15 + 50-meter run score × 0.2 + Sit-and-Reach score × 0.3 + 1-minute jump rope score × 0.2. Ages 8–9: PF = BMI score × 0.15 + Vital Capacity score × 0.15 + 50-meter run score × 0.2 + Sit-and-Reach score × 0.2 + 1-minute jump rope score × 0.2 + 1-minute sit-up score × 0.1. IPAQ-S assesses physical activity levels by asking about the number of days and duration of different intensities of physical activity (including walking) in work, transportation, household chores, and leisure activities during the past 7 days. PA (MET-minutes/week) = (days of vigorous activity × minutes of vigorous activity × 8.0) + (days of moderate activity × minutes of moderate activity × 4.0) + (days of walking × minutes of walking × 3.3).

Sampling

Subjects

In this study, PF and PA were divided into quartiles. Initially, all participants’ PF and PA levels were categorized into quartiles to determine the high (top 25%) and low (last 25%) levels. This resulted in the classification of the subjects into four groups: high PF-low PA, high PF-high PA, low PF-high PA, and low PF-low PA (Fig. 1).

Fig. 1
figure 1

Schematic of the grouping method.

Full size image

Selection method

Utilizing crossover grouping, samples were selected from the four categories: high PF-high PA, high PF-low PA, low PF-high PA, and low PF-low PA for further analysis. In each group, 30 individuals were randomly sampled for a total of 120 participants.

Collection method

Participants were instructed to use specialized fecal collection tubes, following training and relevant guidelines, to collect fecal samples during the second visit. Samples were stored in sterile collection tubes and transported to the microbiome laboratory at −80 °C to ensure the stability of the specimens.

Gut microbiota analysis

This service was provided by Novogene Co., Ltd. (https://cn.novogene.com/). Total DNA was extracted from fecal samples using a Qiagen DNeasy PowerSoil Kit. The purity and concentration of the extracted DNA were measured using a NanoDrop 2000 spectrophotometer16. Then, 16 S rDNA were amplified using the following primers targeting 16 S V3-V4: 515 F and 806R. All PCR reactions were carried out with 15 µL of Phusion® High-Fidelity PCR Master Mix (New England Biolabs), 0.2 µM of forward and reverse primers, and about 10 ng template DNA. Thermal cycling consisted of initial denaturation at 98℃ for 1 min, followed by 30 cycles of denaturation at 98℃ for 10 s, annealing at 50℃ for 30 s, and elongation at 72℃ for 30 s and 72℃ for 5 min. The PCR products were purified using magnetic beads. Samples were mixed in equidensity ratios based on the concentration of the PCR products. After thorough mixing, the PCR products were detected, and target bands were recovered. Sequencing libraries were constructed using the NEBNext Ultra II DNA Library Prep Kit (Cat. no. E7645B). The resulting libraries were quantified using Qubit and qPCR. Following quality control, the libraries were sequenced on an Illumina NovaSeq 6000 platform using the PE250 sequencing strategy. Paired-end reads (2 × 250 bp) were assigned to samples based on their unique barcode and truncated by cutting off the barcode and primer sequence. Paired-end reads were merged using FLASH (V1.2.11, http://ccb.jhu.edu/software/FLASH/)17, which was designed to merge paired-end reads when at least some of the reads overlapped the read generated from the opposite end of the same DNA fragment, and the splicing sequences were called raw tags. Quality filtering of the raw tags was performed using fastp (version 0.23.1) software to obtain high-quality Clean Tags. The tags were compared with the reference database (Silva database, https://www.arb-silva.de/) to detect chimera sequences, and the effective tags were obtained by removing the chimera sequences using the vsearch package (V2.16.0, https://github.com/torognes/vsearch). Denoising was performed with DADA2 in QIIME2 software (Version QIIME2-202202) to obtain initial Amplicon Sequence Variants. Taxonomy was assigned using QIIME2. Specifically, reads were first assigned taxonomy using the classification information file from the official Silva website, and then unassigned reads were assigned using the.dmp file provided by the National Center for Biotechnology Information (USA). Sequence counts before and after quality control and software version information are provided in Supplementary Tables S1 and S2, respectively.

Statistical methods

For alpha and beta diversity analyses, sequences were rarefied to an even depth of 10,000 reads per sample to ensure comparability across samples. Alpha and beta diversity analyses, including PERMANOVA, NMDS, and PCoA visualizations, and correlation heatmaps, were conducted in R (v4.0.3) using the packages ggplot2, vegan, and others (see Supplemental Table S2 for a complete list). Between-group comparisons of alpha and beta diversity indices were performed using Wilcoxon rank-sum tests. The significance of group differences in PERMANOVA was assessed using Benjamini-Hochberg procedure. Pearson’s correlation coefficients were calculated to assess the associations between physical characteristics. Mantel tests were used to assess the significance of the associations between microbial community composition (phylum-level relative abundance) and physical characteristics. For all statistical tests, a p-value threshold of 0.05 was used unless otherwise specified.

Ethical review

This study was approved by the Ethics Committee of the China Institute of Sport Science, No. CISSLA20230110. Informed consent was obtained from all participants and their legal guardians. All research involving humans was carried out in accordance with The Code of Ethics of the World Medical Association (Declaration of Helsinki).

Results

A total of 113 samples were included in the study after excluding individuals who declined second visits or had invalid samples after the test: P1 (low PA-low PF, N = 30), P2 (high PA-low PF, N = 26), P3 (low PA-high PF, N = 30), and P4 (high PA-high PF, N = 27). Across all groups, Firmicutes was the most dominant phylum, exceeding 50% relative abundance in groups P2-P4. Proteobacteria, Bacteroidota, and Actinobacteriota were also identified as dominant phyla, although their ranking varied among groups. Verrucomicrobiota exhibited a higher relative abundance in group P2 than in the other groups. The microbial community composition of groups P2 and P4 displayed a greater degree of similarity than the other group pairings. (Fig. 2). The relative abundances of bacterial families and genera are shown in Supplementary Figures S1 and S2, respectively.

Fig. 2
figure 2

Relative abundance of bacterial phyla. P1 (Low PA-Low PF, N = 30), P2 (High PA-Low PF, N = 26), P3 (Low PA-High PF, N = 30), and P4 (High PA-High PF, N = 27).

Full size image

Alpha diversity

In Fig. 3, intra-level PA comparisons show that children with high PF exhibited a richer and more diverse microbiome within their respective PA groups. Under high PA conditions, the high PF group demonstrated significantly higher Chao1 and Shannon indices, indicating a greater variety of microorganisms and a more even distribution of abundance in this group (P < 0.01). In contrast, within low PA conditions, the Simpson index for the high PF group was significantly higher than that of the low PF group (P < 0.01), whereas no significant difference in the Simpson index was observed under high PA conditions (P = 0.068).

Fig. 3
figure 3

Boxplots illustrating alpha diversity indices. Pairwise comparisons between the groups were conducted using the Wilcoxon rank-sum test. P1 (Low PA-Low PF, N = 30), P2 (High PA-Low PF, N = 26), P3 (Low PA-High PF, N = 30), and P4 (High PA-High PF, N = 27). Asterisks denote statistical significance: *p < 0.05, **p < 0.01, and ***p < 0.001. (a) Chao1 index on the y-axis and group designation on the x-axis. (b) Shannon index on the y-axis and group designation on the x-axis. (c) Simpson index on the y-axis and group designation on the x-axis.

Full size image

PF status comparisons revealed that within the low PF groups, the Chao1 and Shannon indices indicated that children with high PA exhibited richer and more diverse microbial communities (P < 0.01). Conversely, no significant differences were observed between the two PA groups in the high PF status (P > 0.05). Additionally, within each PF level, Simpson indices across different PA levels did not show significant differences (P > 0.05), suggesting similar levels of alpha diversity.

The overall analysis highlighted differences in microbial diversity among the four combinations studied. The Chao1, Shannon, and Simpson indices indicated that group P3 had the highest values, whereas group P1 had the lowest. Comparatively, group P4 exhibited elevated scores for both Chao1 and Shannon but a lower score for Simpson.

Beta diversity

The within-group beta diversity in P1 was significantly different from that in P2, P3, and P4, as indicated by unweighted UniFrac distances (Fig. 4a): specifically between groups P1 and P2 (p < 0.05), between groups P1 and P3 (p < 0.01), and between groups P1 and P4 (p < 0.01). Significant differences were also observed between groups P2, P3 (p < 0.001), and P4 (p < 0.001). Using weighted UniFrac distances (Fig. 4b), similar patterns were observed, with group P3 exhibiting significantly different community structures than groups P2 (p < 0.001) and P4 (p < 0.001).

Fig. 4
figure 4

Beta Diversity Analysis. P1 (Low PA-Low PF, N = 30), P2 (High PA-Low PF, N = 26), P3 (Low PA-High PF, N = 30), and P4 (High PA-High PF, N = 27). (a, b) Boxplots displaying unweighted (a) and weighted (b) UniFrac distances between microbial communities within each group. Significant differences between groups, as determined by Wilcoxon rank-sum tests, are indicated by asterisks (*p < 0.05, **p < 0.01, ***p < 0.001). (c, d) NMDS plots visualizing Unweighted (c) and Weighted (d) UniFrac distances. Each point in the NMDS plot represents a sample, and the distance between points reflects the degree of dissimilarity between the samples. Samples from the same group are represented using the same color. Stress values, ranging from 0 to 1, reflect the degree to which the two-dimensional representation accurately captures the multidimensional relationships between samples; a stress value less than 0.2 indicates that the NMDS plot accurately represents the dissimilarity between samples. PERMANOVA results, including pairwise group comparisons and overall significance (BH-corrected p-values), are shown at the bottom right of each subplot. (e, f) PCoA plots visualizing unweighted (e) and weighted (f) UniFrac distances. The x-axis represents the first principal component (PC1), and the percentage indicates the proportion of variance in the data explained by PC1. The y-axis represents the second principal component (PC2), and the percentage indicates the proportion of variance explained by PC2. Each point in the PCoA plot represents a sample, and samples from the same group are represented using the same color.

Full size image

The NMDS plots (Fig. 4c and d) visually represent beta diversity based on unweighted and weighted UniFrac distances, respectively. The NMDS plot based on unweighted UniFrac distances (Fig. 4c, overall P = 0.013) showed significant intergroup differences between P1 and P3 (P = 0.006) and group P4 (P = 0.033); however, the overall stress value of 0.21 indicated a less-than-ideal fit. The NMDS plot based on weighted UniFrac distances (Fig. 4d, overall P = 0.258) had a good fit (stress = 0.13), but no significant differences were observed between the groups, and community structure differences were less pronounced than in the unweighted plot.

The PCoA plots (Fig. 4e and f) illustrate the distribution of samples along principal components for unweighted and weighted UniFrac distances, respectively. As shown by the PCoA based on weighted UniFrac distances, there was some overlap between the sample groupings and some apparent clusters. However, PERMANOVA analysis indicated no significant differences between groups for either unweighted or weighted distances (P > 0.05). The PCoA results based on Bray-Curtis distances are illustrated in Supplementary Figure S3.

LEfSe analysis

As shown in Fig. 5, distinct microbial taxa exhibited significant enrichment trends (LDA > 4) across the different groups of children. In the low PA, low PF group (P1), there was notable enrichment of c__Gammaproteobacteria. The high PA-low PF group (P2) showed significant enrichment of s__Bacteroides_coprocola and g__Akkermansia. Within the low PA-high PF group (P3), there was a higher abundance of o__Oscillospirales, f__Ruminococcaceae, g__Faecalibacterium, and g__Subdoligranulum. In the high PA-high PF group (P4), the taxa s__Actinomyces_sp_oral_taxon_414, g__Alistipes, and Eubacterium coprostanoligenes groups were significantly more abundant than the other groups.

Fig. 5
figure 5

LEfSe analysis. (a) LDA scores (log 10) for different groups (P1, P2, P3, and P4) across various taxonomic levels. The figure displays differences in abundance for specific species, genera, or families, with higher LDA scores indicating greater significance of that taxon in the corresponding group. P1 (Low PA-Low PF, N = 30), P2 (High PA-Low PF, N = 26), P3 (Low PA-High PF, N = 30), and P4 (High PA-High PF, N = 27). (b) Cladogram illustrating differences in abundance across groups (P1, P2, P3, and P4) at various taxonomic units. This figure uses color-coded regions to represent group-specific abundances, with nodes indicating different taxonomic levels, and line thickness indicating the size of the abundances.

Full size image

Correlation analysis

As shown in Fig. 6, specific microbial taxa were significantly correlated with various PF metrics. Bacteroidetes were notably associated with body weight, BMI, and number of sit-ups performed. Actinobacteria demonstrated a correlation with the sit-and-reach flexibility test. Cyanobacteria were linked to relative grip strength. Both Gemmatimonadetes and Fusobacteria showed a relationship with lung capacity and relative lung capacity. Additionally, Saccharibacteria was associated with standing long jump performance.

Fig. 6
figure 6

Correlation network of various physical fitness metrics and phylum-level abundance. The upper triangle of the matrix displays the Pearson correlation coefficients (r) between various physical characteristics, with the cell color indicating the strength and direction of the correlation. The lower triangle of the matrix represents the association between microbial phyla and physical characteristics. Each line connects a phylum to a physical characteristic; line thickness corresponds to Mantel’s r value, and line color indicates the direction of the association.

Full size image

Discussion

This study aimed to explore the diversity of the gut microbiota in children and its relationship with physical and exercise metrics. Regarding microbial diversity, within groups of the same PA level, children with high PF exhibited greater abundance, and under low PA conditions, those with high PF showed greater evenness; however, under high PA conditions, PF was not related to evenness. Within groups with the same PF level, higher PA was associated with increased alpha diversity and more stable beta diversity, while evenness remained unchanged. Beta diversity analysis indicated that groups P2–P4 were more stable. LEfSe analysis demonstrated enrichment trends in microbial taxa across different child groups and revealed significant correlations between specific microbial taxa and physical and exercise metrics.

In examining the relationship between microbial diversity and PF and PA, we found that in the low PA group, children with high PF exhibited higher microbial abundance and evenness. This finding aligns with the existing literature, suggesting that a robust PF is associated with a more diverse microbiota18,19. Conversely, a lower PF may lead to microbial imbalance, potentially triggering health issues, which is consistent with previous research8. Gut microbiota, through metabolites such as short-chain fatty acids (SCFAs), supports host energy metabolism and immune regulation, thereby influencing overall PF3. Studies indicate that alterations in the microbiome are closely related to health metrics, such as body fat percentage, muscle mass, and insulin sensitivity20. A healthy gut microbiota helps maintain intestinal barrier function and reduce chronic inflammation, which has been identified as an important cause of various metabolic diseases21. Estaki et al. linked cardiorespiratory fitness with gut microbiota diversity, associating it closely with specific metabolic functions18, particularly those related to carbohydrate and fatty acid metabolism, suggesting that higher cardiorespiratory fitness may entail distinct microbial metabolic potentials that enhance energy expenditure and substrate transformation efficiency.

Further exploration of the PA-microbe diversity relationship revealed that higher PA groups consistently showed greater alpha diversity and more stable beta diversity under identical PF conditions. Bressa et al. highlighted distinct gut microbial profiles between active and sedentary women, with aerobic exercise boosting microbial diversity and enhancing gut health22. Exercise interventions have demonstrated significant species diversity enhancements from high-intensity interval training and Tai Chi, albeit less so than aerobic exercise23. Among older adults, exercise does not strongly correlate with gut microbiota diversity10. These findings underscore the complex influences of different exercise forms, age, and sex on the gut microbiota. Nonetheless, in our study, increased PA at equivalent PF levels was beneficial for gut microbiota diversity.

LEfSe analysis revealed notable enrichment trends in microbial taxa among the different child groups. In the low PA-low PF group (P1), there was a significant enrichment of (c__Gammaproteobacteria and p__Proteobacteria). c__Gammaproteobacteria, such as Escherichia coli, are associated with gut microbiota dysbiosis and inflammatory bowel disease24,25. An increased abundance of these taxa is often considered an indicator of gut imbalance and is linked to metabolic disorders25,26. A higher proportion of c__Gammaproteobacteria has been observed in individuals with obesity and is associated with adverse metabolic health, potentially exacerbating inflammatory responses and impairing barrier function27,28. In the high PA-low PF group (P2), there was enrichment of g__Bacteroides coprocola and g__Akkermansia, bacterial taxa known to produce SCFAs such as butyrate, which are beneficial for intestinal health and immune function29. g__Bacteroides are crucial for the breakdown of carbohydrates and fats2 and participate in nitrogen cycling30, thereby promoting intestinal microbiota diversity31,32. Studies have indicated that g__B. coprocola and g__B. plebeius are associated with gut microbiota dysbiosis and an increased risk of colorectal cancer33. g__Akkermansia muciniphila is considered a “keystone species” for gut health, and its abundance can be increased through exercise, conferring benefits for metabolic health31,34,35,36. This group was also enriched in p__Verrucomicrobiota, which may contribute to enhanced stability within the microbial community35,37. The low PA-high PF group (P3) exhibited enrichment in c__Clostridia and g__Faecalibacterium, belonging to p__Firmicutes, which are key anaerobic bacteria involved in the fermentation of dietary fiber1. These taxa produce SCFAs, which are crucial for intestinal health and improved PF11,38. Exercise has been shown to increase the abundance of g__Faecalibacterium prausnitzii11. The atypical abundance of g__Subdoligranulum has been associated with chronic constipation and gut dysfunction39,40. In the high PA-high PF group (P4), the high PF subgroup (P4) exhibited enrichment in s__Actinomyces_sp_oral_taxon_414, which may be associated with atherosclerotic plaque formation41,42. g__Alistipes has been inversely correlated with the rapid progression of mild cognitive impairment43. g__Eubacterium coprostanoligenes can temporarily reduce cholesterol44 and ameliorate chemotherapy-induced intestinal mucositis, showing an inverse relationship with childhood obesity45,46. Combining the between-group differences calculated by PERMANOVA of the weighted versus unweighted UniFrac distance NMDS plots, the differences between the P1 and P2-4 groups are centered on unique species, with differential tributes perhaps related to the flora enriched in the P1 group, as described above.

Significant correlations were observed between specific microbial taxa and the various physical and exercise metrics. p__Firmicutes was positively correlated with weight, BMI, and sit-up performance, indicating the crucial role of this microbe in body composition and PF performance. Studies have highlighted the roles of p__Firmicutes in fermenting dietary fibers and producing SCFAs such as butyrate, which are key to energy metabolism and weight regulation1,47. Additionally, p__Bacteroidetes was negatively correlated with BMI, possibly affecting sprint speed and other athletic abilities. g__Bacteroides spp. effectively utilize dietary components, linking their abundance to favorable metabolic health, body weight management, and exercise capabilities9,48. In contrast, p__Proteobacteria showed positive correlations with lung capacity and standing long jump performance, indicating potential roles in respiratory function and explosive power. Some c__Gammaproteobacteria members associated with enhanced exercise outcomes may boost performance through cellular metabolism and the modulation of inflammation7,49.

This study classified children based on the combination of PF and PA to explore the gut microbiome, a nuanced approach that is relatively uncommon. The focus on the relationship between the gut microbiome of children and PF and activity metrics is timely, especially considering the relative scarcity of research on the gut microbiome in children compared to that in adults. However, given the diversity of the sample, the results may not be generalizable to broader populations. Additionally, the cross-sectional nature of the study limits the conclusions regarding causality and long-term effects. Overall, this study provides a new perspective on children’s health management and supports the development of personalized exercise and dietary intervention strategies to optimize children’s health and development.

Conclusions

This study demonstrates that children’s gut microbiota composition varies significantly based on the combined levels of PA and PF. Children with high PF show greater microbial diversity within similar PA levels, particularly under high PA conditions.