Exercise-Related Personality Traits Are Associated with Gut Microbiome Composition and Meat Quality in Qingyuan Partridge Chickens

Abstract

Chicken meat quality is primarily determined by genetics and nutrition, but recent evidence suggests that animal personality traits and the gut microbiota may also play crucial roles through their influence on muscle development and metabolism. However, the specific mechanisms linking behavior, gut microbiota, and meat flavor remain largely unexplored, particularly in native chicken breeds like the Qingyuan partridge chicken. This study investigated associations between personality traits, gut microbiome composition, metabolite profiles, and meat quality in 200 female Qingyuan partridge chickens. Chickens were monitored for daily step counts from 70 to 120 days of age and divided into three experimental groups: high-exercise (HE), moderate-exercise (ME), and low-exercise (LE). Behavioral tests (open-field, T-maze) revealed HE chickens exhibited enhanced exploration (p < 0.05), reduced latency to move (p < 0.05), and higher learning success rates (p < 0.05) compared to LE counterparts. Meat quality analysis showed HE chickens had brighter breast muscle (L* value, p < 0.05), firmer thigh muscle (shear force, p < 0.05), and reduced intramuscular fat (p < 0.01) relative to LE chickens. Gut microbiome profiling indicated HE chickens showed enrichment of beneficial taxa like Bifidobacterium (p < 0.01) and Intestinimonas (p < 0.05), alongside reduced levels of opportunistic pathogens such as Staphylococcus (p < 0.05). Metabolomic analysis (LC-MS) highlighted upregulated pathways including phenylalanine metabolism and tryptophan–serotonin signaling (VIP ≥ 1.0, p < 0.05). These findings confirm that exercise-related personality traits are associated with improved meat quality and flavor, with correlations linked to gut microbiota and metabolite remodeling, providing novel insights for optimizing poultry meat quality in commercial production.

1. Introduction

In China, chicken meat is the second most popular and most consumed meat. Yellow feather chickens, particularly due to their unique flavor and high quality, occupy a certain market share. The flavor of chicken meat is affected by genetic traits, feed and nutrition, rearing environment, meat processing methods, and slaughter conditions [1,2]. For instance, the calcium-activated neutral proteinase 1 (CANP 1) gene, which encodes calpain, is associated with the tenderness and leanness of chicken meat [3]. Rutin dietary supplements have been demonstrated to improve the meat quality, fatty acid profile, and antioxidant capacity of Qingyuan partridge chicken [4]. In addition, the increased intermuscular and intramuscular fat due to the long rearing period contributes to the retention of flavoring substances such as polyunsaturated fatty acid and linoleic acid [5]. Although genetics and nutrition are well-recognized factors influencing meat quality, the impact of animal personality traits remains largely under-explored. Personality, manifesting as behavioral patterns ranging from vigilance to boldness, fundamentally modulates muscle development and energy metabolism [6,7,8]. Specifically, stress responses and aggressive behaviors trigger the secretion of hormones like adrenaline and cortisol. These physiological changes accelerate postmortem glycogen metabolism, potentially leading to adverse meat acidification and altered color (e.g., PSE meat) [9]. Furthermore, aggressive interactions can cause physical injury and inflammatory responses, directly impairing meat texture and flavor [10]. The Qingyuan partridge chicken is known for retaining innate wild traits similar to its ancestor, the Red Junglefowl (Gallus gallus). These traits include high vigilance, intensified exploration behavior, and distinct stress responses [11]. Unlike commercial broilers that have been selected for docility, these active personality phenotypes may significantly influence their exercise. Exercise is well-documented to enhance meat quality by optimizing muscle fiber composition, reducing abdominal fat, and promoting the accumulation of flavor-related compounds like amino acids [12,13,14]. Thus, elucidating the link between exercise-driven behavioral traits and meat physiology is essential.

On the other hand, the gut microbiota profoundly impacts animal behavior via the brain–gut axis, a complex bidirectional communication system, through multiple pathways including neurotransmitter regulation and inflammatory responses [15]. Exercise can facilitate the augmentation of beneficial microbial flora within the gastrointestinal tract [16]. These microorganisms generate advantageous metabolites, such as short-chain fatty acids, via metabolic processes, which are capable of being transported to skeletal muscle tissue via the circulatory system, where they contribute to the muscular metabolic pathway, ultimately enhancing the quality of meat [17]. In addition, the gut microbiota may synthesize neurotransmitters or indirectly influence the host’s neurotransmitter synthesis through pathways such as inflammatory responses and affecting intestinal endocrine cells, thereby indirectly affecting the behavioral characteristics of animals [18]. However, few studies have been conducted on animal behavior, which may exert an important role in the flavor and quality of chicken meat, and its underpinning mechanism is not yet clear.

In this study, Qingyuan partridge chickens were selected as the animal model, due to their individual variation in personality. The objective of this study was to investigate the effects of animal personality, including activity, anxiety-like, and learning behaviors, on meat quality and flavor, and determine the potential underpinning mechanism through the gut microbiome and metabolites.

2. Materials and Methods

2.1. Experimental Design and Animals

The guidelines of the animal management of Foshan University (FOSU) were followed throughout the study, and the experimental protocols were approved by the Experimental Animal Care and Use Committee of FOSU (Approval number: 2023#06). The experiment was conducted on a farm of Guangdong Tiannong Food Co., Ltd., Qingyuan, Guangdong Province, China. A total of 200 70-day-old female Qingyuan partridge chickens, a dual-purpose breed (initial body weight: 1.15 ± 0.10 kg), were randomly selected from a 2000-hen flock. They were raised in a separate barn (10 m × 10 m × 2.5 m) with a stocking density of 2 birds/m², with rice husk bedding under controlled environmental conditions, maintained at a temperature of 25 ± 2 °C, a relative humidity of 60–70%, and a 16 h light/8 h dark photoperiod. All chickens were fed daily at 8:00 a.m. and had ad libitum access to feed and water.

A pedometer (Model zm20190001, Guangzhou Zhilv Network Technology Co., Ltd., Guangzhou, China) was attached to the right thigh of each chicken to monitor daily physical activity, with step counts synchronized to a cloud platform. The 200 chickens were ranked in descending order based on their cumulative step counts from 70 to 120 days of age. Subsequently, the population was divided into three groups of equal size: the top 33% constituted the high-exercise (HE) group, followed by the middle 33% (moderate-exercise, ME), and the bottom 33% (low-exercise, LE). One-way ANOVA confirmed highly significant differences in step counts among the groups (p < 0.001), verifying distinct exercise intensities between HE (25,492.28 ± 8501.75 steps), ME (12,608.34 ± 1407.78 steps), and LE (5334.77 ± 2277.87 steps). For subsequent analyses, sample sizes were determined based on the experimental methodology: 20 chickens from each of the three groups (HE, ME, and LE; n = 20/group) were selected for behavioral observation and meat quality assessment, whereas a subset of 10 chickens from the two extreme groups (HE and LE; n = 10/group) was selected for multi-omics sequencing.

2.2. Measurement of Behavioral Traits

2.2.1. The Open-Field Test

The open-field test is a widely utilized method for assessing exploratory behavior, general activity, and fear responses as indicators of stress levels in both laboratory and farm animals. At 120 days of age, all birds were subjected to an open-field test within an arena on the same day. Each chicken was moved to the testing room and introduced into the arena measuring 1.6 m × 1.6 m × 2 m. The arena was divided into 16 squares, with the inner 4 squares designated as the Central Zone and the other squares designated as the Peripheral Area. The latency to first move, steps, vocalization duration, number of escape attempts, and other behaviors (definitions of behaviors are detailed in the Supplementary Materials, Table S1) were observed through continuous video monitoring for 10 min and recorded by the same trained experimenter [19].

2.2.2. Learning Ability

The T-maze experiment is designed to evaluate the learning ability and spatial memory of birds [11]. Birds were subjected to a T-maze learning test after being food-deprived for 12 h. The T-maze arena consists of a starting box and two arms. The starting box is 1.0 m in length and 0.5 m in width. At the end of the starting box, there are two 90° turns, either to the left or to the right, leading to two additional arms, each 3.0 m in length. Birds are free to choose which arm to enter. Prior to the test, birds are grouped and habituated to the arena for 30 min, during which food is available in each arm of the maze. During the test, a bird is placed in the starting box located in a dark room. At the beginning of each trial, the room is illuminated, and the guillotine door is opened to allow the bird to enter the maze. Food is placed at the far end of one arm, out of the bird’s sight. Trials are not counted until the bird successfully locates the food for the first time during the test. For each bird, the food is consistently placed in the same arm, with the direction balanced across groups.

Subjects undergo testing for three consecutive days, with two trials per day, separated by a one-hour interval. Experimenters monitor behavior remotely via cameras and record the choices made by the birds during the test, as well as the time spent. The test is terminated when the bird enters the terminal alley of either arm or when the bird spends more than 10 min in the maze. A bird is considered to have solved the task if it selects the correct arm in five out of six consecutive trials and is then returned to its home cage [20].

2.3. Determination of Meat Quality and Flavor

A total of three groups, each consisting of 20 birds, were euthanized humanely at 130 days of age. Prior to slaughter, the birds were fasted for 12 h but had free access to water. Samples of the left breast and thigh muscles were collected for meat quality analysis (HE: n = 20, ME: n = 20, and LE: n = 20). The pH of the meat was measured 45 min post-slaughter using a pH meter (SevenGo Duo SG23, Mettler Toledo, Greifensee, Switzerland). The color of the meat, including lightness (L*), redness (a*), and yellowness (b*), was determined using a colorimeter (OPTO-LAB, Eoptis, Optoi Group, Trento, Italy). The shear force of the meat was measured using a shear force tester (RH-N50, Guangzhou Runhu Instrument Co., Ltd., Guangzhou, China). The meat’s moisture, collagen, salt, fat, and protein content were analyzed using a multifunctional rapid quality analyzer (Foss DS2500, Foss Analytical, Hillerød, Denmark) [21].

Collected breast and thigh muscles were frozen at −80 °C within 2 h post-slaughter and thawed at 4 °C for 10 h before analysis (HE: n = 20, ME: n = 20, LE: n = 20). Next, 10 g of the samples were placed in a 20 mL headspace vial, and then incubated the samples at room temperature for 30 min to equilibrate the volatile compounds. An electronic nose system (PEN3.5, AIRSENSE, Schwerin, Germany) was used to insert the injection needle into a sealed sample beaker and detect the volatile compounds. The system parameters were set to a carrier gas flow rate of 300 mL/min, a sensor self-cleaning time of 70 s, and a measurement time of 60 s. Each sample was tested three times. After selecting the peak values of the curves, the average value was calculated. The final detection data were collected using the software Win Muster (v.1.6.2.22). The performance characteristics of the sensing arrays were shown in the Supplementary Materials Table S2.

2.4. Analysis of Gut Metabolites

On the slaughter day, cecal contents were collected from the HE and LE groups (n = 10 per group) for metabolomic analysis. Metabolites were extracted by mixing 20 µL of thawed sample with 120 µL of precooled 50% methanol buffer. The mixture was vortexed, incubated at room temperature, and stored at −20 °C overnight to ensure protein precipitation. After centrifugation at 4000 g for 20 min, the supernatant was transferred for LC-MS analysis. Chromatographic separation was performed on an ACQUITY UPLC T3 column (Waters, Cheshire, UK) using a mobile phase system of water (0.1% formic acid) and acetonitrile (0.1% formic acid) under a gradient elution program. Mass spectrometry detection was carried out using a Triple TOF 5600 Plus system (SCIEX, Solihull, UK) in both positive and negative ion modes. Data were acquired in Information-Dependent Acquisition (IDA) mode. To ensure system stability, QC samples were analyzed every 10 samples.

Raw data were converted and processed using XCMS Online (version 3.7.1, Scripps Research, La Jolla, CA, USA) for peak alignment, retention time correction, and extraction. Metabolites were annotated using KEGG and HMDB databases. After normalization, differential metabolites were identified using Student’s t-tests (with FDR adjustment) and PLS-DA (VIP > 1.0).

2.5. Gut Microbiota Sequencing and Analysis

Total DNA was extracted from samples using the CTAB method. PCR amplification was performed in a 25 μL reaction mixture containing 25 ng template DNA, 12.5 μL PCR Premix, and specific barcoded primers. The thermal cycling conditions consisted of an initial denaturation at 98 °C for 30 s, followed by 32 cycles of denaturation (98 °C for 10 s), annealing (54 °C for 30 s), and extension (72 °C for 45 s), with a final extension at 72 °C for 10 min. PCR products were verified by 2% agarose gel electrophoresis, purified using AMPure XT beads (Beckman Coulter, Brea, CA, USA), and quantified via Qubit (Invitrogen, Carlsbad, CA, USA). Library quality was assessed using an Agilent 2100 Bioanalyzer (Agilent, Santa Clara, CA, USA).

Sequencing was conducted on an Illumina NovaSeq platform (LC-Bio, Hangzhou, China). Paired-end reads were merged using FLASH (v1.2.8) and quality-filtered with fqtrim (v0.94). Chimeric sequences were removed using VSEARCH (v2.3.4), and feature tables were generated via DADA2 (v1.24.0). Taxonomy was assigned against the SILVA database (v138). Statistical analyses were performed in R (v3.5.2).

2.6. Targeted Tryptophan Metabolomic Analysis of Gut Content and Hippocampus

Hippocampal samples were collected immediately after slaughter. To minimize metabolite degradation, the entire brain was rapidly removed, and the hippocampal tissue was accurately dissected on ice within 3 min. The samples were instantly snap-frozen in liquid nitrogen and stored at −80 °C until analysis. For metabolite extraction, approximately 50 mg of tissue was weighed and homogenized in 400 µL of precooled 50% methanol buffer (containing 0.1% formic acid) to precipitate proteins and extract metabolites. The mixture was vortexed for 1 min, centrifuged at 12,000× g for 15 min at 4 °C, and the supernatant was collected for subsequent LC-MS analysis. Tryptophan-targeted metabolomics was performed on both gut content and hippocampal tissue samples. A total of 20 samples were analyzed for each tissue type, divided into the high-exercise (HE, n = 10) and low-exercise (LE, n = 10) groups. The metabolite extraction followed a standard protocol provided by LC-Bio Technology (Hangzhou, China). Briefly, an appropriate amount of sample (hippocampus or gut content) was weighed and mixed with precooled 80% methanol (aqueous solution). The mixture was homogenized using a bead beater and vortexed to ensure complete lysis. After centrifugation at 20,000× g for 15 min at 4 °C, the supernatant was collected. The solvent was removed using a vacuum concentrator, and the dried residue was reconstituted in 100 µL of solvent (acetonitrile: water) for LC-MS/MS analysis.

The chromatographic separation was performed using an LC-MS/MS system (AB Sciex 4500MD, SCIEX, Framingham, MA, USA). Samples were separated on a reverse-phase column (Agilent Poroshell 120 EC-C18, 3.0 × 150 mm, 2.7 μm; Agilent Technologies, Santa Clara, CA, USA) maintained at 40 °C. The mobile phases consisted of (A) 5 mM ammonium acetate with 0.01% formic acid and (B) acetonitrile. The gradient elution program was set to effectively separate tryptophan and its downstream metabolites (including Kynurenine, Serotonin, Indole, etc.). The mass spectrometer was operated in Multiple Reaction Monitoring (MRM) mode to ensure high sensitivity and selectivity for targeted quantification.

2.7. Data Analysis

Data were managed using Excel 2016 and analyzed using IBM SPSS Statistics 25.0 (IBM Corp., Armonk, NY, USA). Results are presented as the mean ± standard error (SE). The normality of the data distribution and homogeneity of variance were assessed using the Shapiro–Wilk test and Levene’s test, respectively.

For parametric data (e.g., meat quality indices) that met the assumptions of normality and homogeneity of variance, a one-way ANOVA was performed, followed by Tukey’s HSD test for multiple comparisons. When data showed unequal variances, Dunnett’s T3 test was applied to strictly control the family-wise error rate. Behavioral data did not meet the assumptions for parametric analysis, and transformations were unsuccessful. Therefore, the Kruskal–Wallis H test was used to assess treatment differences, followed by Dunn’s post hoc test with Bonferroni correction for pairwise comparisons. Learning ability (success or failure) was analyzed using the Chi-square test. For gut microbiota and metabolomics datasets, p-values were adjusted using the Benjamini–Hochberg false discovery rate (FDR) method to account for high-dimensional multiple testing. In the correlation analysis, Spearman’s rank correlation coefficient was calculated to evaluate the relationships between variables. Differences were considered statistically significant at an adjusted p-value < 0.05. Images were plotted and processed using GraphPad Prism 9.5 and Origin 2022 software.

3. Results and Discussion

3.1. Behavioral Traits

Morphologically and behaviorally, Qingyuan partridge chickens mirror their ancestor, the Red Junglefowl (Gallus gallus), displaying distinct wild-type phenotypes such as a lean physique, heightened vigilance, and robust locomotor activity [22]. The results showed that HE chickens exhibited significantly more steps (p < 0.05, Figure 1a) and spent significantly less time in the center zone (p < 0.0001, Figure 1b) and more time in the peripheral area (p < 0.0001, Figure 1b) than LE chickens, which indicates higher activities of HE chickens regardless of an unfamiliar environment. The above results were in line with findings that the frequency of exploratory behavior, vocalization duration, and latency to first movement were significantly higher in HE chickens than in LE chickens (p < 0.05, Figure 1f). Detailed statistical results for all behavioral metrics are provided in Supplementary Table S3. These findings indicate that HE chickens are more proactive and explorative than LE chickens. No significant differences were observed in the average number of escape attempts among groups (Figure 1c).

Figure 1. Behavioral traits in response to the open-field test. (a) Steps, (b) proportion of movement time in the center and periphery areas, (c) average number of escape attempts, (d) frequency of exploratory behavior (e), vocalization duration (f), latency to first movement, and (g) total duration of the T-maze test for six trials. Error bars represent the standard error of the mean (SEM). Asterisks indicate significant differences (* p < 0.05, **** p < 0.0001). HE refers to the high-exercise chickens, ME refers to the moderate-exercise chickens, and LE refers to the low-exercise chickens, n = 20.

In the T-maze experiment, the task’s success rate was 82.5% for HE, 70% for ME, and 55% for LE chickens, respectively (Supplementary Materials Table S4). As shown in Figure 1g, the average completion time was significantly higher in HE than in LE chickens (p < 0.05), with no significant difference between LE and ME chickens. A previous study has shown that exercise promotes neuronal growth and connectivity in the brain, thereby enhancing neuroplasticity and improving learning and vigilance traits [23]. We speculate that HE chickens display greater neuroplasticity, enabling more effective adaptation to new environments and tasks [24]. This increased vigilance, exploratory behavior, and learning abilities equip HE chickens with superior independent activity capabilities in response to their living environment. Even under standardized rearing conditions, significant individual variation in physical activity was observed. This intraspecific variation is likely driven by social hierarchy formation (where dominant birds may roam more freely) and epigenetic differences that regulate the hypothalamic–pituitary–adrenal (HPA) axis, resulting in distinct proactive (high-exercise) versus reactive (low-exercise) behavioral coping styles.

3.2. Meat Quality and Flavor

The physical properties of chicken meat are one of the important indicators for evaluating its quality [25]. In this study, different exercise levels had no impact on the pH value and shear force of the breast muscle. In terms of meat color, the brightness of the breast muscle in HE chickens was significantly higher than that in LE chickens (p < 0.05, Figure 2c). These observations suggest that exercise levels may be associated with the efficiency of muscle cell energy metabolism in relation to cardiopulmonary function [26], coinciding with lower accumulation of anaerobic metabolites and the maintenance of chicken meat brightness. On the other hand, the high-exercise phenotype aligns with the continuous contraction and relaxation of muscle fibers, which is associated with the thickening and strengthening of fibers, as well as increased muscle density and firmness [27]. Well-developed muscle fibers exhibit specific light reflection and scattering properties, which are statistically linked to the enhanced brightness of chicken meat [28]. For the thigh muscle, the shear force was significantly higher in HE than in LE chickens (p < 0.05, Figure 2e), which aligns with a previous study showing exercise can alter muscle fiber type and structure. This correlates with increased muscle fibers and connective tissue in HE chickens, as well as potential enrichment of slow-twitch fibers (larger with more connective tissue), with these muscle structural traits showing associations with exercise levels [29]. For Qingyuan partridge chickens, moderate shear force aligns with consumer preference for meat with a “firm and resilient texture”. Furthermore, higher exercise levels correlate with enhanced aerobic metabolism, where myoglobin may exhibit higher oxygen-binding capacity [30], consistent with the observed reduction in yellowness in the thigh muscle of HE chickens (p < 0.05, Figure 2f).

Figure 2. Meat quality. (a) Breast muscle pH. (b) Breast muscle shear force. (c) Breast muscle lightness (L*), redness (a*), and yellowness (b*). (d) Thigh muscle pH. (e) Thigh muscle shear force. (f) Thigh muscle lightness (L*), redness (a*), and yellowness (b*). Error bars represent the standard error of the mean (SEM). Asterisks indicate significant differences (* p < 0.05). HE refers to the high-exercise chickens, ME refers to the moderate-exercise chickens, and LE refers to the low-exercise chickens, n = 20.

In this study, the non-significant differences observed in thigh muscle pH and shear force can be attributed to the voluntary nature of the exercise regimen. Unlike forced exercise, which often acts as an intense exogenous stressor triggering severe glycogen depletion and resulting in DFD (Dark, Firm, Dry) meat characteristics [9], the high activity level in HE chickens was a spontaneous behavioral trait. This voluntary movement allows for metabolic compensation mechanisms to take effect, where the animal adapts its energy metabolism to maintain homeostasis and preserve sufficient glycogen reserves. Consequently, the muscles ensured normal postmortem acidification (stable pH) and avoided the stress-induced structural toughening or dehydration often reflected in altered shear force. Thus, the thigh muscle of HE chickens exhibits a commercially desirable texture that benefits from exercise-induced structural remodeling while avoiding the negative quality defects associated with stress-induced DFD meat.

In the breast muscle, the fat content in LE chickens was significantly higher than that in ME chickens (p < 0.05, Figure 3a), while the ME chickens exhibited the highest protein content (p < 0.05, Figure 3b). The variation in intramuscular fat content associated with different exercise levels was more pronounced in the thigh muscle. Specifically, the thigh muscle of HE chickens had a significantly lower fat content compared to both the LE (p < 0.01) and ME chickens (p < 0.05, Figure 3a). Carotenoids, such as lutein and zeaxanthin, are the primary pigments responsible for the yellow value of chicken meat. Consequently, when the fat content is elevated, the carotenoids within the fat contribute to a more pronounced yellow color in the meat. In this study, the higher yellowness (p < 0.05, Figure 2f) and fat content (p < 0.05, Figure 3f) observed in the thigh muscle of LE chickens compared to HE chickens further aligns with this association. Collagen is the most abundant protein in animals and a primary component of connective tissue, playing diverse biological roles in maintaining tissue cell structural integrity, cell differentiation, adhesion, migration, and tissue repair [31]. The increased thigh muscle collagen content in HE chickens (p < 0.05, Figure 3i) may be linked to the stretching of tendons, ligaments, and fascia during activity, which is associated with connective tissue and collagen growth [32]. Proteins are essential for animals, supplying key amino acids for muscle growth and repair. Interestingly, the LE chickens exhibited significantly higher thigh muscle protein content than the HE chickens (p < 0.05, Figure 3g). This finding suggests a potential linkage between increased activity and more frequent minor muscle damage, which coincides with a higher protein requirement for repair and growth. Regarding chemical composition, the moisture and salt contents in both breast and thigh muscles remained stable across all groups (Figure 3c,e,h,j). The lack of significant variation suggests that the exercise intensity in this study did not disrupt the osmotic balance or water-holding capacity of the muscle cells. This indicates that the exercise-induced metabolic shifts were primarily directed towards lipid and protein metabolism rather than altering the fundamental ionic or hydration status of the muscle tissue.

Figure 3. Component contents of breast and thigh muscles. (a) Fat content of the breast muscle. (b) Protein content of the breast muscle. (c) Moisture content of the breast muscle. (d) Collagen content of the breast muscle. (e) Salt content of the breast muscle. (f) Fat content of the thigh muscle. (g) Protein content of the thigh muscle. (h) Moisture content of the thigh muscle. (i) Collagen content of the thigh muscle. (j) Salt content of the thigh muscle. Error bars represent the standard error of the mean (SEM). Asterisks indicate significant differences (* p < 0.05, ** p < 0.01). HE refers to the high-exercise chickens, ME refers to the moderate-exercise chickens, and LE refers to the low-exercise chickens, n = 20.

In Figure 4a, PC1 and PC2 account for 71.3% and 19.8% of the variance in breast muscle E-nose values, respectively, while in Figure 4b, they explain 68.0% and 26.2% of the variance in thigh muscle. These results confirm the reliability of the E-nose measurements. The radar charts (Figure 4c,d) reveal that breast and thigh muscles share similar main odor components, including W5S (nitrogen oxides), W2W (ethanol and some aromatic compounds), W2S (a wide range of substances mainly detecting ethanol and some aromatic compounds), W1W (inorganic sulfides, sulfur-containing organics, and terpenes and pyrazines), and W1S (inorganic sulfides). Specifically, LE chickens exhibited higher odor indices than HE chickens in breast muscles (p < 0.05, Figure 4c) and lower indices in the thigh muscles compared to HE chickens (p < 0.05, Figure 4d). Exercise is associated with higher metabolic activity and the breakdown of fats and carbohydrates [33]. In the high-exercise group, this metabolic profile is characterized by an increase in volatile compounds such as aromatic substances and short-chain alkanes in the thigh muscles. Moreover, exercise levels align with variations in the structure and function of muscle fibers, which correlate with their ability to adsorb and release odor components [34]. High-intensity exercise coincides with potential minor damage to muscle fibers, which is linked to the muscle repair and growth process and the concomitant presence of more metabolic products and odor components. However, an appropriate amount of intramuscular fat is positively associated with the deposition of flavor substances in chicken meat [35]. The breast muscles, which are mainly used for intense flight, have a relatively lower fat content, while thigh muscles are responsible for supporting body weight and daily movement, thus containing more fat. Exercise levels show an inverse association with breast muscle fat (Figure 3a), which correlates with lower levels of flavor compounds.

Figure 4. (a) Breast muscle E-nose PCA. (b) Thigh muscle E-nose PCA. (c) Breast muscle odor fingerprinting radargram. (d) Thigh muscle odor fingerprinting radargram. The description of each sensor is listed in Supplementary Materials Table S2. HE refers to the high-exercise chickens, ME refers to the moderate-exercise chickens, and LE refers to the low-exercise chickens, n = 20.

In summary, chickens with different levels of exercise exhibit distinct personalities including exploratory behavior, activity, and learning ability, all of which are associated with differences in meat quality and flavor.

3.3. Gut Microbiome

Distinct microbial community structures were observed between the HE and LE groups, as illustrated by the clear separation in the Principal Coordinate Analysis (PCoA) plot (Figure 5a). This analysis demonstrated that PCoA1 explained 10.54% of the total variance, while PCoA2 accounted for 7.38%. Together, these two axes represented the primary sources of β-diversity variation across all microbial samples. The 95% confidence ellipses of the two groups showed a certain degree of separation, indicating that gut microbial community compositions differed between chickens with distinct behavioral patterns.

Figure 5. (a) PCoA (Principal Coordinate Analysis) diagram based on Jaccard distance matrix for dimensionality reduction analysis. (b) Taxonomic proportions according to compositions at the phylum level. (c) Taxonomic proportions according to compositions at the genus level. (d) Comparisons of the abundance of microbial community phyla with significant Top 10 differences. (e) Comparisons of the abundance of microbial community genus with significant Top 10 differences. Abundance is expressed as log-transformed (log₂(Abundance + 1)) to reduce data bias and enhance visualization. An asterisk above each bar indicates statistical significance, where “*” indicates p < 0.05, “**” indicates p < 0.01, “***” indicates p < 0.001. HE refers to the high-exercise chickens, and LE refers to the low-exercise chickens, n = 10.

At the phylum level (Figure 5b), Firmicutes and Bacteroidota (formerly Bacteroidetes) were identified as the dominant taxa in both groups. The Firmicutes/Bacteroidota (F/B) ratio was lower in the HE group (1.31) than in the LE group (1.35). These two phyla coexist and maintain a dynamic equilibrium in the intestines of healthy hosts, playing critical roles in nutrient absorption, immune regulation, and gut barrier preservation. Their abundance ratio is known to reflect the host’s metabolic health and obesity status [36]. Notably, HE chickens also exhibited reduced intramuscular fat content, reflecting a potential association with host health and obesity mitigation. Additionally, the relative abundance of Verrucomicrobiota was significantly higher in the HE group (a 122.3% increase) compared to the LE group. This phylum encompasses Akkermansia, a key probiotic genus. At the genus level (Figure 5c), the HE group had higher abundances of beneficial genera, including Bifidobacterium, Megamonas, and Christensenellaceae_R-7_group. Specifically, Bifidobacterium produces short-chain fatty acids (SCFAs) and inhibits pathogenic bacteria, while Christensenellaceae_R-7_group is associated with obesity resistance.

Differential abundance analysis highlighted specific variations in bacterial community composition between the two groups (Supplementary Materials Table S5). Detailed statistical results, including raw p-values and FDR-adjusted q-values for all detected taxa, are provided in Supplementary Materials Table S5. Among all detected phyla (Figure 5d), the top four dominant phyla were Proteobacteria, Acidobacteriota, Dependentiae, and Latescibacterota. The relative abundances of Proteobacteria and Acidobacteriota were significantly higher in LE chickens than in HE chickens (p < 0.01), whereas Latescibacterota was significantly more abundant in HE chickens (p < 0.05). Among the 10 most abundant genera, Bifidobacterium (p < 0.01), Intestinimonas (p < 0.05), Candidatus_Arthromitus (p < 0.05), Veillonella (p < 0.05), and Tuzzerella (p < 0.05) showed significantly higher relative abundances in HE chickens. In contrast, Staphylococcus, Herbaspirillum (p < 0.001), Allorhizobium–Neorhizobium–Pararhizobium–Rhizobium (p < 0.001), and Methylobacterium–Methylorubrum (p < 0.001) were more abundant in LE chickens. No significant intergroup difference was observed in the relative abundance of Pseudoflavonifractor.

While the dominant phyla (i.e., Actinobacteriota, Bacteroidota, and Firmicutes) were shared between the two groups, distinct variations were observed at the genus level within these phyla. Within Actinobacteriota, Bifidobacterium was the dominant genus in HE chickens, whereas Olsenella prevailed in LE chickens. For Bacteroidota, Rikenellaceae_RC9_gut_group was the dominant genus in HE chickens, while Bacteroides and Prevotellaceae_UCG-001 were more prevalent in LE chickens. Within Firmicutes, HE chickens were characterized by the dominance of Faecalibacterium, Lactobacillus, and Megamonas, whereas Streptococcus was the predominant genus in LE chickens. These findings suggest that exercise intensity is associated with variations in cecal microbial composition.

The LDA analysis revealed that 26 distinct bacterial taxa were significantly enriched from the phylum to the species level (Figure 6). Specifically, 10 species were enriched in HE chickens, with the top five being Actinobacteria (LDA score = 3.56), Bifidobacterium (LDA score = 3.48), Arthromitus (LDA score = 3.24), Sutterellaceae (LDA score = 3.10), and Intestinomonas (LDA score = 3.07). Conversely, LE chickens were enriched with 16 species, the top five of which were Rhizobiales (LDA score = 3.77), Herbaspirillum (LDA score = 3.63), Rhizobium (LDA score = 3.55), Staphylococcus (LDA score = 3.70), and Betaproteobacteria (LDA score = 3.51). As a common probiotic, Bifidobacterium has been associated with the regulation of the animal immune system, the enhancement of metabolic processes, and the maintenance of gut microbiota homeostasis in numerous studies [37], thereby showing a positive correlation with the physiological state of the host. The higher abundance in HE chickens coincides with exercise-related intestinal peristalsis and an improved gut environment. In comparison, LE chickens showed enrichment of four species of Staphylococcus and ten species of Proteobacteria, which may exhibit pathogenicity under certain conditions. Opportunistic pathogens can influence stress responses in the animal gut through multiple mechanisms, including toxin secretion, immune regulation interference, and alteration of the gut microenvironment. When animals are exposed to stressors, Staphylococcus may exacerbate the negative effects of stress by enhancing inflammatory responses or disrupting the balance of the normal microbiota [38]. Conclusively, the results of 16S DNA sequencing suggest that exercise level was associated with an altered microbial structure; HE chickens exhibited higher relative abundance of taxa commonly regarded as probiotics, while LE chickens exhibited lower stability and higher susceptibility of the gut microbiota.

Figure 6. (a,b) LEfSe analysis of variance plot with LDA value of 3. HE refers to the high-exercise chickens, and LE refers to the low-exercise chickens, n = 10.

3.4. Gut Metabolites

In Figure 7a, PC1 accounts for 12.26% of the variance, while PC2 explains 12.54%. To ensure the reliability of the model and exclude overfitting, a permutation test with 200 iterations was performed. The robustness of this model was confirmed by permutation testing (Supplementary Materials Figure S1), which yielded a Q² intercept of −0.3878, significantly lower than zero, indicating that the model was not overfitted. The model possessed high explanatory power (R² = 0.97) and good predictive accuracy (Q² = 0.63). The 95% confidence ellipses indicated the range of variation within each group, revealing significant differences in cecum content metabolites between the HE and LE chickens. A post hoc sensitivity analysis (G*Power, version 3.1.9.7, Heinrich-Heine-Universität Düsseldorf, Düsseldorf, Germany) indicated that with the current sample size (n = 10 per group), α = 0.05, and power 1-β) of 0.80, the minimum detectable effect size was Cohen’s d = 1.325. This implies that the study was primarily powered to detect large effects. Consequently, the absence of significant associations for metabolites with small-to-moderate effects should be interpreted with caution, as these may not have been detectable given the current statistical power.

Figure 7. (a) PLS-DA score plot between HE and LE chickens. (b) Top 10 metabolic pathways in KEGG enrichment histogram. (c) Differential metabolite volcano plot. (d) Differential metabolic pathway KEGG enrichment bar-circle plot. HE refers to the high-exercise chickens, and LE refers to the low-exercise chickens, n = 10.

A total of 2103 metabolites were identified in the cecum contents. Among these, the most abundant class was lipids and lipid-like molecules, totaling 1025. This was followed by organic acids and their derivatives (organic acids and derivatives) with 345 metabolites. The counts for other classes were as follows: organoheterocyclic compounds (273), benzenoids (167), organic oxygen compounds (91), phenylpropanoids and polyketides (57), polyketides (57), organic nitrogen compounds (49), nucleosides, nucleotides, and their analogs (35), alkaloids and their derivatives (23), and other compounds (38). KEGG pathway enrichment analysis revealed significant upregulation of multiple critical metabolic pathways in the cecum metabolome of HE chickens (Supplementary Materials Table S6). These alterations reflect the multi-dimensional physiological associations of exercise with chicken health and meat quality. The elevated efferocytosis pathway suggests potential for enhanced cecum immune homeostasis through accelerated clearance of apoptotic cells, coinciding with reduced inflammatory responses—a profile that aligns with maintaining gut integrity and higher feed conversion efficiency [39]. Concurrently, the upregulation of the aldosterone synthesis and secretion pathway was positively associated with exercise level and coincided with higher predicted capacity for gut electrolyte and water reabsorption [40]. This statistical association suggests a potential link between exercise-related microbial profiles and fluid–homeostasis pathways.

In terms of energy metabolism, the coordinated upregulation of thiamine metabolism and 2-oxocarboxylic acid metabolism pathways suggests that exercise-related metabolic profiles may align with optimized carbohydrate and energy utilization, reflecting variations in gut microbiota metabolism that meet increased host energy demands [41]. Such metabolic reprogramming is not only associated with energy supply efficiency but also shows a potential link to improved meat quality through optimized muscle energetic status. The upregulation of protein metabolism-related pathways, including protein digestion and absorption and aminoacyl-tRNA biosynthesis [42], suggests that higher exercise levels correlate with microbial-mediated amino acid absorption efficiency and coincide with protein synthesis rates. Host utilization of these metabolic products, facilitated by intestinal epithelial cell uptake, plays a pivotal role in promoting muscle hypertrophy, repair processes, and ultimately enhancing meat yield and quality parameters. Concurrent activation of insulin secretion and steroid hormone biosynthesis pathways further reflects exercise-associated profiles in glucose homeostasis and steroidogenic pathways [43].

Notably, alterations in phenylalanine metabolism and dopaminergic synapse pathways suggest that exercise is linked to differences in neurotransmitter synthesis and signaling, which align with the personality traits of chickens. Dopamine is associated with neuroplasticity, motor learning, and skill acquisition, and reflects the formation and optimization of movement patterns [44]. The activation of this pathway is connected to the consolidation of motor memory [45], which aligns with the finding that chickens with higher exercise exhibit stronger learning abilities. These finding highlight the potential associations between the gut microbiome and host cognition. Exercise-associated dopamine levels coincide with the stress response during exercise by influencing the hypothalamic–pituitary–adrenal axis, potentially affecting gut microbiome stability [45,46]. Moreover, specific metabolites from the gut microbiome show correlations with the metabolic synthesis of dopaminergic synaptic pathways. In addition to its role in neurotransmission, dopamine levels are associated with gut motility and secretion, suggesting that exercise levels and personality traits show complex associations with the gut–microbiome axis.

3.5. Correlation Between Gut Microbiota, Metabolome, Exercise, and Meat Quality and Flavor

KEGG pathway annotation of differentially expressed gut microbiome and metabolites revealed two convergent pathways: Aminoacyl-tRNA biosynthesis and Phenylalanine metabolism. As an essential amino acid, phenylalanine must be obtained through dietary intake and plays a critical role in maintaining muscle mass and systemic protein homeostasis [47]. This pathway shows a potential linkage to meat quality, particularly umami flavor development, and aligns with synergistic interactions with glutamate, aspartate, and inosinic acid [48]. Intestinal phenylalanine metabolism involves host-microbiome co-metabolism, producing bioactive intermediates that modulate physiological states [49]. Notably, while KEGG analysis indicated downregulated microbial phenylalanine metabolism in HE chickens, absolute phenylalanine concentrations and downstream metabolites were elevated compared to LE chickens. This paradox suggests either selective enrichment of phenylalanine-synthesizing microbiota or reduced microbial competition for this amino acid in exercised animals.

Correlation analysis (Figure 8a) highlighted significant interactions between gut microbiome composition and metabolic profiles, with particular focus on HE-enriched Bifidobacterium. Positive associations with PC(34:2) and FAHFA 5-OH(22:0) suggests this genus may be linked to lipid membrane homeostasis and energy storage regulation, respectively [50]. Conversely, negative correlations with L-leucine and L-alanine imply branched-chain amino acid (BCAA) catabolism, potentially reflecting microbial nitrogen scavenging or host-microbiome substrate competition [51].

Figure 8. (a) Heat map of correlation between differential metabolites and differential microbes at the genus level, “*” indicates p < 0.05, “**” indicates p < 0.01. (b) Heat map of Mantel test network between chicken exercise, meat quality, and differential metabolites.

Mantel test analysis of environmental–metabolite (Figure 8b) interactions revealed exercise-induced remodeling of lipid and amino acid metabolic networks. Key glycerophospholipids (PC(P-18:1/20:4-OH), PC(16:0/18:2)) showed positive correlations with Bifidobacterium abundance and locomotor activity, consistent with murine studies demonstrating improved glucose-lipid homeostasis following Bifidobacterium supplementation [52]. Notably, BCAA metabolites (Dl-valine), lipid intermediates (5-valerolactone), and inflammatory mediators (PGF2α1) exhibited exercise-dependent associations with meat quality parameters. D-ornithine and guanidinoacetic acid showed strong positive correlations with breast muscle protein content, suggesting exercise-mediated protein synthesis enhancement through creatine precursor availability [53].

These findings confirm that exercise is associated with multi-systemic metabolic and microbial patterns in chickens, integrating metabolic shifts and microbiome profiles that correlate with meat quality and flavor.

Accumulating evidence indicates that tryptophan metabolism plays a crucial role in the link between exercise and cognition [54,55,56]. The gut microbiome is closely associated with tryptophan metabolites, such as serotonin and indole derivatives. These metabolites can traverse the intestinal mucosa, enter the systemic circulation, and influence the function of distal organs, including the brain [54]. This aligns with the hypothesis that microbial tryptophan metabolites facilitate communication between the gut and the central nervous system [57].

Tryptophan metabolism primarily occurs through two pathways: the serotonin pathway and the kynurenine pathway [58]. In this study, the tryptophan metabolic pathways in the gut and hippocampus showed consistent results. As exercise levels increased, the concentration of kynurenine and 5-HT tended to rise in both the gut and the hippocampus (p < 0.05). However, L-tryptophan and other tryptophan metabolites, along with their precursors, showed a negative correlation with exercise levels. Gut tryptophan is converted into kynurenine and serotonin through exercise (Figure 9). A prior study demonstrated that Bifidobacterium administration increases plasma tryptophan levels in rats [59]. Consistently, our results showed that HE chickens hosted a higher abundance of Bifidobacterium than LE chickens. Thus, we speculate exercise is linked to the gut environment in association with tryptophan metabolism, while gut microbes reflect the host’s tryptophan metabolic pathways by producing tryptophan or participating in the serotonin synthesis process.

Figure 9. (a) Correlation analysis of exercise levels and tryptophan metabolism in the gut. (b) Correlation analysis of exercise levels and tryptophan metabolism in the hippocampus.

The HE chickens had superior cognitive abilities, learning abilities, and exploratory behavior compared to the LE chickens. Serotonin, a key neurotransmitter, regulates mood, sleep, and cognitive function [60]. Notably, higher exercise levels coincided with elevated serotonin synthesis in the hippocampus, a brain region central to cognition. We speculate that HE chickens possess an innate metabolic profile favoring serotonin synthesis, which supports their enhanced cognitive resilience and exploratory behavior. Furthermore, exercise is associated with enhanced BDNF expression, which improves cognitive function [61]. Gut microbiome may be linked to host cognition through potential associations with neurotransmitters like tryptophan. Many studies indicate that exercise activates dopaminergic pathways linked to reward mechanisms. Recent reports suggest the serotonin system is also involved [62]. Since tryptophan is a precursor to serotonin, its metabolism provides a potential biological basis for the higher activity and cognitive levels observed in HE chickens.

4. Conclusions

This study demonstrates that exercise-related personality traits in Qingyuan partridge chickens are closely associated with improved meat quality and flavor. High-exercise chickens exhibit brighter breast muscle, firmer thigh texture, reduced intramuscular fat, and distinct volatile flavor profiles compared to low-exercise chickens. The potential mechanism may involve the synergistic association between gut microbiota and metabolites: higher activity levels coincide with a reshaped cecal microbiome enriched with beneficial genera and a lower abundance of opportunistic pathogens. These microbial profiles align with variations in key metabolic pathways, which are statistically linked to gut–brain axis communication and behavioral traits.

However, the present study does have several limitations. Although the gut–brain axis hypothesis provides a biologically plausible framework, it remains unclear whether microbial shifts drive personality traits or are a physiological consequence of different activity levels. Future research incorporating longitudinal designs and interventional approaches, such as fecal microbiota transplantation, is required to validate these causal relationships. Despite these limitations, our findings highlight the potential of using spontaneous exercise behavior as a novel selection criterion. Integrating “personality” screening into breeding programs could offer a non-invasive approach to simultaneously optimize poultry meat quality and animal welfare in sustainable farming systems.