The effects of citrulline on horse milk components and fecal metabolism were analyzed based on LC-MS and 16s amplicon Authors: Yumei Ma, Pengshun Liu, Xiaobin Li, Qingsong Hu, Jianwei Lin, Fan Yang, Kailun Yang Published: 2026 ABSTRACT AbstractBackgroundThe lactation period in mares is critical, as it significantly influences foal health and subsequent reproductive outcomes. Developing a precise nutritional strategy for lactating mares is a primary research focus. L-citrulline, a bioactivenon-protein amino acid and precursor to arginine, has been shown to enhance mammary blood flow and reduce oxidative stress, yielding promising results in livestock lactation studies. However, research on its effects in Equus species remains limited. There is a notable gap in data regarding its impact on lactation performance blood biochemistry, and its role in the fecal microbiota and metabolism of mares. This study, therefore, seeks to explore the effects of L-citrulline supplementation on lactation performance, foal growth, fecal microbiota, and metabolomic profiles in grazing lactating mares, aiming to establish a foundation for dietary optimization and he advancement of the horse industry.ResultsIn this study, 32 lactating mares were selected and assigned to a control group (no amino acid supplementation) and experimental groups I, II, and III (15 g/d, 30 g/d, and 45 g/d of L-citrulline respectively) under identical grazing conditions, with 8 mares per group. The results demonstrated a significant increase in milk production for mares in group II (30 g/d) (P < 0.01). Additionally, milk fat percentage, foal weight and body size were higher in all experimental groups compared to the control group. Blood levels of lactate dehydrogenase and creatine kinase were extremely significantly lower in the experimental groups than in the control group. To further explore the effects of L-citrulline on fecal microbiota and metabolites,16 S rDNA sequencing and LC-MS/MS analysis were performed on the control group and the experimental group receiving 30 g/d (which showed the most pronounced lactation response). The microbiota level, analysis revealed Firmicutes as the dominant phylum, with significant differences in the species F082 and the genus Ruminococcaceae_bacterium_AE2021. Tax4Fun functional annotation showed the highest gene function abundance in digestive metabolism, genetic information processing and environmental information processing pathways. LC-MS/MS analysis identified differential metabolites primarily associated with biological systems and metabolism. In conclusion, L-citrulline supplementation in grazing lactating mares increased milk production, improved milk fat quality, and enhanced foal weight and body size.ConclusionL-citrulline supplementation enhances milk production in mares, improves milk fat, quality, reduces lactate dehydrogenase and creatine kinase levels in the blood, and promote the growth and development of lactating foals.Keywords: Lactating mare, L-citrulline, Lactation performance, Microbiota, Metabolism Introduction The formation of milk in lactating livestock is influenced by multiple factors, including genetic background, nutritional status, and environmental conditions [1]. Within a consistent genetic framework, nutrition is a key determinant of both milk yield and quality. Key nutrients involved in regulating milk synthesis include energy, protein (including amino acids), lipids (including fatty acids), vitamins, and minerals [2, 3]. Amino acids, in particular, play a critical role in milk production. As substrates, they contribute to the synthesis of milk proteins, while also servingas regulatory factors, in milk anabolism [4]. Lysine and threonine essential amino acids for Equus species, are identified as the primary andsecondary limiting amino acids in grazing animals. Jing et al. [5] demonstrated that supplementing lysine and threonine during lactation in Ilima mares significantly improved milk yield, milk fat percentage, and milk protein content. Targeted metabolomic analysis of plasma amino acids and free amino acids in milk revealed that lysine and threonine enhance nitrogen metabolism by increasing nitrogen cycling, thus improving milk yield and protein content. Additionally, lysine may promote mammary epithelial cell proliferation and milk protein synthesis through modulation of the STAT signaling pathway [6]. Nonessential amino acids also play a pivotal role as nitrogen sources, significantly contributing to protein metabolism and milk synthesis. Arginine supplementation has been shown to increase daily milk yield in dairy cows by 2.2 kg compared to control and alanine groups, along with enhanced milk fat content. Furthermore, the expression of genes related to milk fat synthesis in the mammary gland was upregulated, resulting in significant improvements in both milk fat content and overall milk yield [7]. Thus, mammalian milk synthesis requires not only substrate components such as amino acids, fatty acids, and glucose, but also non-substrate substances that regulate the overall synthesis process. L-citrulline, a non-essential amino acid, derives its name from its initial extraction from watermelon juice. It is metabolized in the body and functions in concert with arginine and nitric oxide [8]. L-citrulline has diverse regulatory effects, including modulation of intestinal function, antioxidant and anti-inflammatory activities, and enhancement of lactation performance by increasing milk yield and improving milk composition [9-11]. However, it does not directly regulate milk synthesis; rather, it promotes mammary gland health through indirect mechanisms. Under grazing conditions, the nutritional intake of horses is limited by the quality of available. forage, which fluctuates with seasonal changes, ultimately influencing nutrient absorption and consequently the milk yield of lactating mares [12, 13]. In this study, Ili horses were selected as the experimental subjects. Under uniform feeding conditions, the effects of L-citrulline supplementation on lactation performance, production outcomes, fecal microbiota diversity, and metabolomics were analyzed. The objective was to explore differences in fecal microbiota and metabolites among lactating mares and to identify pathways through which L-citrulline may participate in regulating milk fat synthesis. This research aims to provide a theoretical basis and practical guidance for optimizing the nutritional management of lactating mares. Materials and methods Experimental design The experiment was conducted in Coudre Grassland, Zhaosu County, Ili Kazakh Autonomous Prefecture, Xinjiang. A total of 32 healthy Yili lactating mares and their foals were selected as experimental subjects, with each mare and foal forming a dam-offspring pair. The mares, aged 7–9 years and, with 4–5 previous parities, had an average body weight of 415 ± 25.5 kg. They had foaled in May and were in the first 30 days of lactation. The foals approximately 2 months old, had an average body weight of 94.91 ± 14.3 kg, with an equal number of males and females. All mares and foals were dam-offspring pairs. The animals were randomly assigned to four groups, with eight mares per group (foals were grouped with): a control group (no amino acid supplementation), experimental group I (L-citrulline 15 g/d), experimental group II (L-citrulline 30 g/d), and experimental group III (L-citrulline 45 g/d). L-citrulline (≥ 99% purity on a dry matter basis) was purchased from Anhui Runtai Biotechnology Co., Ltd. The study included a 7-day pre-feeding period, followed by a 90-day formal experimental period. Feeding and management During the milking period, all mares underwent 8 h of daytime tied milking, with foals separated and not allowed to nurse. For the remaining 16 h, mares and foals were kept together, permitting the foals to nurse freely. Each day at 08:30, mares and foals were driven from the grazing pasture to the herders’ milking site. At 09:00, experimental mares received their respective L-citrulline supplements via a special feed bag. Simultaneously, mares were separated from their foals for milking. At 17:00, after the final milking session, mares and foals were reunited and returned to the pasture together. During the milking period, foals had unrestricted access to grazing and water. Table 1 provides the nutritional composition of the forage grass. The dry matter, organic matter and total energy are reported on a dry matter basis. Table 1Nutritional composition of pasture (air-dry basis, %)①ItemsContentDM(%)90.77 ± 1.77OM(%)92.20 ± 1.05GE(MJ/kg)17.55 ± 0.48CP(%)10.53 ± 0.79EE(%)2.73 ± 0.63CF(%)36.66 ± 0.94Ca(%)0.83 ± 0.65P(%)0.15 ± 0.01①The nutrient content is the actual measured value. DM Dry matter, OM Organic matter, GE Gross energy, CP Crude protein, EE Ether extract, CF Crude fiber, Ca Calcium, P Phosphorus Nutritional composition of pasture (air-dry basis, %)① ①The nutrient content is the actual measured value. DM Dry matter, OM Organic matter, GE Gross energy, CP Crude protein, EE Ether extract, CF Crude fiber, Ca Calcium, P Phosphorus Determination of feed intake and sample collection Fecal samples were collected from all mares at 08:30 on the 90th day of the experimental period. The samples were placed into 50 mL and 5 mL sampling tubes, Fecal samples in the 50 mL tubes were frozen at -20 ℃ for the analysis of dry matter intake, microbiota and metabolomics. Dry matter intake was measured using the acid-insoluble ash (AIA) method, based on the principle that AIA is not absorbed in the digestive tract. The calculation formula measured is as follows [14]: Where TDI is the total daily fecal output (dry weight) (kg·day⁻¹·mare⁻¹); F is the dry matter mass of feces discharged by the test horse daily (kg·day⁻¹·mare⁻¹); A is AIA content of dry matter in feces (%); B is the content of dry matter AIA in forage grass (%). The results of dry matter intake are presented in Table 2. Table 2Dry matter intake (kg/d)ItemsControl groupExperimental group ⅠExperimental group ⅡExperimental group ⅢTDI12.13 ± 0.6512.19 ± 1.1212.30 ± 1.2412.17 ± 0.95 Dry matter intake (kg/d) Measurement of growth performance of foals Foals were weighed at 08:30 on days 0, 30, 60, and 90 of the experimental period. Body height, body length, chest circumference, and cannon circumference were measured using a measuring stick and tape measure. Total weight gain, average daily weight gain, total growth, and average daily growth were calculated based on these measurements. Measurement of growth performance of foals Throughout the 0–90-day experimental period, milk yield was recorded at each milking session (11:00, 13:00, 15:00, and 17:00 daily) using electronic scales. On days 0, 30, 60, and 90, 25 mL of milk was collected at each milking session. The samples were thoroughly mixed and transferred into two 100 mL sampling tubes, which were then stored at -20 °C for subsequent analysis of milk composition. Determination of blood biochemical indicators Blood samples were collected from the jugular vein of mares and foals after an overnight fast on days 0, 30, 60, and 90 of the experiment, prior to morning feeding. The mare was restrained and separated from the foal, and the foal’s neck was disinfected with alcohol. A disposable blood collection needle was inserted at a 45° angle into the jugular vein, and 5 mL of blood was collected using a heparin sodium vacuum blood collection tube. The samples were then centrifuged at 3,500 rpm for 15 min. The supernatant was transferred to a centrifuge tube and stored at -20 °C for subsequent analysis. Plasma indicators measured in this study included protein-related markers: total protein (TP), albumin (ALB), globulin (GLB), urea (UERA), and creatinine (CREA); sugar and lipid-related markers: glucose (Glu), triglycerides (TG), total cholesterol (TCH), total bilirubin (T-Bil), and direct bilirubin (D-BiL); and enzyme activity-related markers: alanine aminotransferase (ALT), lactate dehydrogenase (LDH), creatine kinase (CK), and alkaline phosphatase (ALP). All assays were performed according to with the methods provided with the respective kits, all of which were purchased from Beijing Huaying Biotechnology Research Institute. The analysis was conducted using an automated biochemical analyzer . Determination of fecal pH and volatile fatty acids Determination of pH On the final day of the experiment, fresh feces were collected from each horse using disposable sterile gloves. The fecal samples were thoroughly mixed, and 10 g of feces were diluted in 10 mL of distilled water (a 1:1 ratio of feces to water) and stirred at room temperature for 3–5 min. The fecal pH was measured using a calibrated portable pH meter with an accuracy of 0.01 (FiveEasy22-Meter, Mettler-Toledo International Trading (Shanghai) Co., Ltd.). Determination of volatile fatty acids The volatile fatty acids (VFAs) in the fermentation medium were measured using a Shimadzu GC-2010 gas chromatograph equipped with a Stabilwax chromatographic, with 4-methylvalproic acid as the internal standard. Chromatographic conditions were as follows: the injection port temperature was set to 230 °C, the initial column oven temperature was 55 °C, and the temperature was increased at 13 °C/min to 200 °C, which was maintained for 0.5 min. The FID detector was set to 240 °C. Sample processing and determination: Ten grams of fecal sample, were mixed with 10 mL of ultrapure water, vortexed oscillate for 2 min, filtered through four layers of gauze, and centrifuged at 10,000 rpm for 10 min. The supernatant was then frozen at 4 °C for 12 h. Afterward, the sample was centrifuged at 15,000 rpm for 15 min, and 1 mL of the filtrate was centrifuged again at 15,000 rpm for 5 min. 0.5 mL of the supernatant was transferred into a 1.5 mL centrifuge tube, to which 0.5 mL of 10%TCA and 0.1 mL of 40 mmoL 4-methylvaleric acid. were added. The mixture was vortexed, left to stand for 20 min, and then centrifuged at 20,000 ×g for 15 min. The supernatant (1.0 mL) was collected into a sample bottle, and 0.5 µL was used for machine analysis to obtain the VFA chromatogram. The internal standard method was applied to establish a regression equation with the VFA concentration ratio as the Y-axis and the peak-to-height ratio as the X-axis, allowing for the calculation of VFA content in the sample. Determination of fecal microbiota Fecal samples from mares were removed from liquid nitrogen, thawed and genomic DNA was extracted using the cetyltrimethylammonium bromide (CTAB) method. (Hexadecyltrimethylammonium bromide) [15, 16]. The DNA purity and concentration were assessed via 0.8% agarose gel electrophoresis, and the DNA was then diluted to 1 ng/µL for subsequent use. Using this diluted genomic DNA as a template, amplification was performed targeting the V3-V4 region of the 16 S rDNA using primers 341 F and 806 R: 341 F: 5 '-CCTAYGGGRBGCASCAG-3' 341 F: 5 '-CCTAYGGGRBGCASCAG-3' 806 R: 5 '-GGACTACNNGGGTATCTAAT-3' 806 R: 5 '-GGACTACNNGGGTATCTAAT-3' The total bacterial DNA extracted was amplified for the 16 S rDNA-V3-V4 region. The correct target band size was confirmed, and the total yield met the requirements for at least two library constructions, enabling subsequent library preparation. Sequencing was performed on the Illumina MiSeq platform. An average of 85,432 ± 12,456 high-quality reads were obtained per sample. The raw reads were subjected to quality filtering (removing reads with an average quality score below Q20 and discarding sequences shorter than 200 bp). Operational taxonomic units (OTUs) were clustered at a 97% similarity threshold using UPARSE software, and taxonomic classification was performed based on the Silva database. PCR amplification was carried out with the following conditions: all PCR mixtures were added to 15 of µL Phusion High-Fidelity PCR Master Mix(New England Biolabs), 0.2 µM primers, and 10 ng genomic DNA templates. The reaction was denatured for 1 min at 98 ℃, followed by 30 cycles of 98 °C for 10 s,50 °C for 30 s, and 72 °C for 30 s. Finally, the reaction was maintained at 72 °C for 5 min. Analysis of fecal metabolomics Extraction of metabolites Fecal metabolites were analyzed using an untargeted metabolomics approach. Fecal samples (100 mg) were ground with liquid nitrogen, and mixed with pre-cooled 80% methanol, followed by thorough vortex mixing. The samples were incubated on ice for 5 min and then centrifuged at 4 °C at 15,000 g for 20 min. The supernatant was diluted with LC-MS grade water to a final methanol concentration of 53%. The samples were then transferred to new Eppendorf tubes and centrifuged again at 4 °C at 15,000 g for 20 min. Finally, the supernatant was injected into the LC-MS/MS system for analysis. UHPLC-MS/MS analysis UHPLC-MS /MS analysis was conducted at Novogene (Beijing, China) using the Vanquish UHPLC system (Thermo Fisher). It was coupled with either the Orbitrap Q ExactiveTM HF mass spectrometer or the Orbitrap Q ExactiveTM HF-X mass spectrometer (Thermo Fisher, Germany). The sample was injected onto a Hypersil Gold chromatographic column (100 × 2.1 mm, 1.9 μm), and a 12-minute linear gradient elution was performed at a flow rate of 0.2 mL/min. Eluents for the positive and negative polarity modes were: eluent A (aqueous solution containing 0.1% formic acid) and eluent B (methanol), respectively. The solvent gradient was as follows: 2% B for 1.5 min; 2–85% B for 3 min; 85–100% B for 10 min; 100-2% B for 10.1 min; 2% B for 12 min. The Q ExactiveTM HF mass spectrometer operated in both positive and negative polarity modes, with a spray voltage of 3.5 kV, a capillary temperature of 320 ° C, a sheath gas flow rate of 35 psi, an auxiliary gas flow rate of 10 L/min, an S-lens RF level of 60, and an auxiliary gas heater temperature of 350 ° C. Statistical analysis Experimental data were organized using Excel. The milk yield, milk composition, foal weight, and body measurements of lactating mares were analyzed with SPSS 27.0 software using two-way ANOVA with repeated measures over time, with treatment (control group, experimental group I, experimental group II, experimental group III) and time (days 0, 30, 60, and 90) as fixed factors, and their interaction was considered. Blood biochemical indices, fecal pH, and volatile fatty acid concentrations in lactating mares were also analyzed using SPSS 27.0, with post-hoc multiple comparisons conducted using Duncan’s method. Results are presented as mean ± standard deviation. Statistical significance was defined as P < 0.05, with P < 0.01 indicating extreme significance. Amplicon sequencing of the 16 S rDNA gene for microbial analysis and untargeted metabolomics were performed on the Nuohe Zhiyuan NuoHe Cloud platform. For downstream analysis, only operational taxonomic units (OTUs) or metabolites with a relative abundance exceeding 0.01% in more than 50% of samples within each group were included. Significant differences in metabolite levels were determined using the following criteria: VIP > 1, P < 0.05, and fold change (FC) ≥ 2 or ≤ 0.5. Correlation analyses were visualized with the Chiplot software package (https://www.chiplot.online/, Hangzhou Meizheng Biotechnology Co., Ltd., Hangzhou, China). Results Effect of supplementing L-citrulline on the lactation performance of mares As shown in Table 3, supplementation of L-citrulline significantly increased milk production in experimental group II with an 8.61% increase compared to the control group (P < 0.05). The Butterfat percentage in experimental group II also increased significantly by 17.09% compared to the control group. Furthermore, the density in experimental group III was significantly higher by 0.10% compared to experimental group I (P < 0.05). The milk protein percentage across experimental groups ranged from 1.70% to 1.74%, and the lactose percentage ranged from 6.20% to 6.28%, with no significant differences among groups (P > 0.05). Analysis of the main effects of grouping and time revealed that time had an extremely significant impact on milk production, milk fat percentage, milk protein percentage, non-fat solids, and freezing point (P < 0.01). Additionally, time had a significant effect on density and salt content (P < 0.05).Quadratic polynomial regression analysis revealed a significant parabolic relationship for both parameters (milk production: y = 3.585 + 0.017x-0.000256x², R²=0.92; butterfat percentage: y = 1.16 + 0.0127x-0.000222x², R²=0.90). According to the formula, the optimal doses were calculated as (Fig. 1A) 33.3 g/d and (Fig. 1B) 28.5 g/d, respectively. Based on the observed peak values and theoretical optima, 30 g/d was identified as the most effective supplementation dose.the optimal doses were calculated as 33.3 g/d and 28.5 g/d, respectively. Based on the observed peak values and theoretical optima, 30 g/d was identified as the most effective supplementation dose. Table 3Effects of dietary L-citrulline supplementation on lactation performance in lactating maresItemsExperimental treatmentsSEM P ControlgroupExperimentalgroup ⅠExperimental group ⅠIExperimental group ⅠIItrtdaytrt*dayMilk production(kg/d)3.60Bb3.74ABab3.91Aa3.82ABa0.0650.009< 0.0010.952Butterfat percentage(%)1.17b1.27ab1.37a1.27ab0.0520.062< 0.0010.672Milk protein ratio(%)1.711.701.721.740.0230.669< 0.0010.760Lactose rate(%)6.206.206.246.280.0340.353< 0.0010.787Non-fat solid(%)8.678.658.748.810.0580.214< 0.0010.945Freezing point(℃)−0.54−0.55−0.55−0.550.0040.648< 0.0010.515Density(kg/L)1.032ab1.031b1.032ab1.032a0.0000.0830.0040.345Salt(%)0.670.670.680.690.0070.3320.0020.974Different superscript lowercase letters (a, b, c) indicate significant differences (P < 0.05); different superscript uppercase letters (A, B, C) indicate highly significant differences (P < 0.01). trt: refers to treatment effect; day: refers to time effect; trt*day: refers to the interaction effect Effects of dietary L-citrulline supplementation on lactation performance in lactating mares Different superscript lowercase letters (a, b, c) indicate significant differences (P < 0.05); different superscript uppercase letters (A, B, C) indicate highly significant differences (P < 0.01). trt: refers to treatment effect; day: refers to time effect; trt*day: refers to the interaction effect Dose-response curves of milk yield (A) and milk fat percentage (B) in lactating mares supplemented with different doses of L-citrulline Effect of supplementing L-citrulline on the growth performance of foals Table 4 shows the effect of L-citrulline supplementation on the growth performance of nursing foals. In the experimental group II, foal weight, chest circumference, and body circumference increased significantly by 9.62%, 5.80%, and 6.04%, respectively, compared to the control group (P < 0.01). The improvement of the nursing foals in experimental group III also showed significant, improvements, with increases of 3.03% and 4.34% in weight and body circumference, respectively compared to the control group and experimental group I (P < 0.01). The body oblique length of the nursing foals in experimental groups II and III increased significantly by 4.30% and 3.79%, respectively, compared to the control group (P < 0.01). The time effect indicated that growth parameters in all treatment groups continued to increase trend over the experimental period (0, 30, 60, and 90 days). Table 4Effects of dietary L-citrulline supplementation on growth performance in suckling foalsItemsExperimental treatmentsSEM P Control groupExperimentalgroup ⅠExperimentalgroup ⅡExperimentalgroup Ⅲtrtdaytrt*dayWeight(kg)105.56Bc109.81bABbc116.79Aa112.39ABab2.1930.004< 0.0010.839Height(cm)113.98Bbc112.44Bc115.67ABab117.54Aa0.871< 0.001< 0.0010.991ObLique Length(cm)97.81b97.71b102.20a101.66a1.1680.007< 0.0010.977Chest size(cm)103.75Bc106.33ABbc110.14Aab108.41ABa1.2620.004< 0.0010.998Pipe circumference(cm)13.70Bb14.13ABab14.58Aa14.35ABa0.1970.0160.0290.997Different superscript lowercase letters (a, b, c) indicate significant differences (P < 0.05); different superscript uppercase letters (A, B, C) indicate highly significant differences (P < 0.01). trt: refers to treatment effect; day: refers to time effect; trt*day: refers to the interaction effect Effects of dietary L-citrulline supplementation on growth performance in suckling foals Different superscript lowercase letters (a, b, c) indicate significant differences (P < 0.05); different superscript uppercase letters (A, B, C) indicate highly significant differences (P < 0.01). trt: refers to treatment effect; day: refers to time effect; trt*day: refers to the interaction effect Effect of supplementing L-citrulline on the blood biochemical indicators of lactating mares The effects of L-citrulline supplementation on the blood biochemical indicators of lactating mares are shown in Table 5. Supplementation had no significant effect (P > 0.05) on TP, ALB, GLB, UREA and CREA, or in mares’ blood. However, GLB content in the blood of mares in experimental group II was 7.27% lower than in the control group. The UREA content in mares from experimental groups II and III decreased by 1.37% and 7.33%, respectively, compared to the control group. CREA content in the mares’ blood increased by 2.97%, 9.97%, and 4.35% in the experimental groups, respectively, compared to the control group. No significant differences (P > 0.05) were observed in blood Glu, TG, T-Bil, or D-BiL levels in foals across groups. However, in experimental group III, these levels were all decreased by 2.75%, 18.75%, 24.57%, 32.45%, and 2.37%, respectively, compared to the control group. The activity of LDH in the blood of lactating mares was significantly reduced in all experimental groupsby 6.99%, 14.82%, and 24.91%, respectively compared to the control group (P < 0.01). CK activity mares’ blood was also significantly reduced in experimental groups II and III by 16.15% and 35.09%, respectively, compared with the control group (P < 0.01). There was no significant differences in the activities of ALT or ALP in the blood of lactating mares across all experimental groups. Table 5Effects of supplementing L-citrulline on blood biochemical indicators in lactating maresItemsControl groupExperimentalgroup IExperimental group IIExperimentalgroup III P TP (g/L)70.62 ± 6.3673.12 ± 8.5668.47 ± 6.8173.93 ± 7.470.45ALB (g/L)23.28 ± 2.9524.90 ± 1.8624.58 ± 2.5125.10 ± 1.410.39GLB (g/L)47.33 ± 5.8848.22 ± 8.2643.89 ± 4.7548.83 ± 7.540.48UREA(mmol/L)8.05 ± 0.878.22 ± 1.657.94 ± 1.147.46 ± 1.460.69CREA(µmol/L)70.65 ± 6.9872.81 ± 17.5978.47 ± 11.9973.72 ± 19.420.76GLU(mmol/L)2.55 ± 1.422.77 ± 1.632.66 ± 1.202.48 ± 0.970.97TG(mmol/L)0.16 ± 0.080.12 ± 0.040.17 ± 0.100.13 ± 0.060.48T-Bil(µmol/L)8.75 ± 2.048.45 ± 1.978.29 ± 2.716.60 ± 2.170.24D-BiL(µmol/L)5.67 ± 1.715.49 ± 1.995.16 ± 2.363.83 ± 1.470.24TC(mmol/L)1.69 ± 0.171.59 ± 0.411.74 ± 0.201.65 ± 0.300.78ALT(U/L)5.19 ± 0.995.60 ± 1.114.98 ± 1.425.42 ± 1.880.82LDH(U/L)451.68 ± 50.62Aa420.13 ± 100.42ABa384.76 ± 36.42ABab339.15 ± 77.25Bb0.02CK(U/L)231.95 ± 29.20Aa229.89 ± 64.24Aa194.50 ± 51.23ABab150.57 ± 42.11Bb0.01ALP(U/L)268.15 ± 47.04258.45 ± 65.47252.22 ± 62.35217.66 ± 58.940.36Values are presented as mean ± SD. Different superscript lowercase letters (a, b, c) indicate significant differences (P < 0.05); different superscript uppercase letters (A, B, C) indicate highly significant differences (P < 0.01) Effects of supplementing L-citrulline on blood biochemical indicators in lactating mares Values are presented as mean ± SD. Different superscript lowercase letters (a, b, c) indicate significant differences (P < 0.05); different superscript uppercase letters (A, B, C) indicate highly significant differences (P < 0.01) Effect of supplementing L-citrulline on pH and volatile fatty acids in the feces of lactating mares The effects of L-citrulline supplementation on the pH and VFAs in the feces of lactating mares are presented in Table 6. Supplementation with L-citrulline in experimental group III led to a significant reduction in fecal pH by 7.36% compared to the control group (P < 0.01). The acetic acid content in experimental groups II and III increased by 19.33% and 8.69%, respectively, compared to the control group. The Propionic acid levels in all experimental groups increased by 10.23%, 17.58%, and 18.32%, respectively, compared to the control group. Total volatile fatty acid concentrations in experimental groups II and III increased by 19.22% and 8.35%, respectively, relative to the control group. Table 6Effects of Supplementary Feeding of L-citrulline on the pH and Volatile Fatty Acids (VFA) in the Feces of Lactating Mares(mmol/L)ItemsControl groupExperimentalgroup ⅠExperimental group ⅡExperimental group Ⅲ P pH6.97 ± 0.37Aa6.70 ± 0.37ABa6.69 ± 0.29ABa6.29 ± 0.44Bb0.01Acetic acid20.91 ± 8.3320.71 ± 3.8225.92 ± 8.0922.90 ± 7.370.44Procionic acid2.72 ± 1.063.03 ± 0.913.30 ± 0.733.33 ± 1.070.53Isobutyric acid0.49 ± 0.470.51 ± 0.310.70 ± 0.440.33 ± 0.160.28Butyric acid1.05 ± 0.511.17 ± 0.811.23 ± 0.601.27 ± 0.450.90Isovaleric acid0.81 ± 0.680.50 ± 0.340.99 ± 0.670.56 ± 0.360.26Valeric acid0.24 ± 0.230.19 ± 0.130.33 ± 0.190.22 ± 0.140.49TVFA26.22 ± 10.1326.11 ± 5.2532.46 ± 9.1228.61 ± 8.341.00Values are presented as mean ± SD. Different superscript lowercase letters (a, b, c) indicate significant differences (P < 0.05); different superscript uppercase letters (A, B, C) indicate highly significant differences (P < 0.01) Effects of Supplementary Feeding of L-citrulline on the pH and Volatile Fatty Acids (VFA) in the Feces of Lactating Mares(mmol/L) Values are presented as mean ± SD. Different superscript lowercase letters (a, b, c) indicate significant differences (P < 0.05); different superscript uppercase letters (A, B, C) indicate highly significant differences (P < 0.01) Effect of L-citrulline supplementation on bacterial diversity in feces of lactating mares As shown in Fig. 2. the Venn diagram (Fig. 2A) based on OTUs reveals that the fecal microbiota of mares in both the control and experimental groups shareof 4,051 common OTUs, indicating substantial overlap between the groups.The control group harbored 3,568 unique OTUs, while the experimental group contained 3,482 unique OTUs. Further analysis of bacterial community structures using principal component analysis (PCA) and principal coordinate analysis (PCoA) is shown in Fig. 2B and C. In the PCA plot, the first principal component (PC1) explained 11.02% of the variation, while the second principal component (PC2) accounted for 8.94%, with overlapping clusters indicating high similarity between the two groups. The PCoA results also showed minimal differentiation between the groups, with PC1 contributing 26.13% and PC2 explaining 14.67% of the variance. Alpha diversity indices—including Chao1, Shannon, Simpson, and Pielou’s evenness, (Fig. 2D), revealed no significant differences between the groups (P > 0.05), indicating comparable species richness and evenness. At the phylum level (Fig. 2E), the top ten dominant taxa in both groups were Firmicutes, Euryarchaeota, Bacteroidota, Verrucomicrobiota, Actinobacteriota, Patescibacteria, Halobacterota, Spirochaetota, Cyanobacteria, and Synergistota, which together accounted for over 99.7% of all sequences, with less than 0.03% of sequences assigned to unclassified taxa. Among these, Firmicutes, Bacteroidota, and Euryarchaeota were the most abundant. At the family level (Fig. 2F), the dominant families included Lachnospiraceae, Methanobacteriaceae, Oscillospiraceae, Rikenellaceae, F082, Christensenellaceae, p-251-o5, Prevotellaceae, Lactobacillaceae, and Akkermansiaceae, which together made up over 74% of the microbial composition, with approximately 26% belonging to unclassified families. Significant differences in relative abundance between the two groups were observed at the family level (P < 0.05). At the species level (Fig. 2G), the dominant species include Lactobacillus_equi, Methanobrevibacter ruminantium, Ruminococcus_sp_HUN007, Lactobacillus hayakitensis, Rhodococcus coprophilus, Ruminococcaceae_bacterium_AE2021, Clostridium butyricum, Clostridium_bacterium_Firm_14, Rumen_bacterium_NK4A65, and Lactobacillus equigenerosi, representing approximately 2% of the total classified sequences, with the remaining 98% unclassified. Significant differences between the groups at both the family and species levels were observed for F082 and Ruminococcus_bacterium_AE2021 (P < 0.05), as shown in Fig. 2H. Functional predictions of the fecal microbiota based on Tax4Fun analysis, shown in Fig. 2I and J, identified major functional categories including digestion and metabolism, genetic information processing, environmental information processing, cellular processes, human diseases, and organismal systems. Digestion and metabolism, along with genetic and environmental information processing, collectively accounted for over 80% of the predicted functions, highlighting their central roles in microbial activity. Effect of L-citrulline (30 g/d) on fecal microbiota of lactating mares (n = 8 per group). A Venn diagram showing shared and unique OTUs between the control group(C) and the experimental group(L). B Principal component analysis (PCA) score plot. C Principal coordinate analysis (PCoA) plot. D Alpha diversity indices (Chao1, Shannon, Simpson, Pielou’s evenness). E Relative abundance of the top 10 phyla. F Relative abundance of the top 10 families. G Relative abundance of the top 10 species. H Taxa with significant differences between groups at family and species levels. I Tax4Fun functional prediction histogram. J Tax4Fun functional prediction cluster heatmap Effect of supplementary feeding of L-citrulline on fecal bacterial metabolism based on non-targeted metabonomics analysis In positive ion mode As shown in Fig. 3, in positive ion mode, metabolites are classified into 14 categories (Fig. 3A). The most abundant categories include lipids and lipid-like molecules (33.42%), organic acids and derivatives (17.78%), organoheterocyclic compounds (16.41%), benzenoids (9.16%), phenylpropanoids and polyketides (8.83%), and organic oxygen compounds (6.11%). Minor categories include alkaloids and derivatives (2.84%), organic nitrogen compounds (2.34%), nucleosides, nucleotides, and analogues (1.04%), lignans, neolignans and related compounds (0.71%), hydrocarbons (0.60%), organosulfur compounds (0.44%), hydrocarbon derivatives (0.22%), organometallic compounds (0.05%), and organic 1,3-dipolar compounds (0.05%). Figure 3B shows the relative abundance and distribution patterns of various metabolites, with each row representing a metabolite and each column corresponding to a sample. The color scale ranges from blue (low relative abundance, as low as − 2.5) to red (high relative abundance, up to 3.8), with color intensity indicating the relative levels of metabolites across samples. As presented in Fig. 3C, the correlation among QC samples exceeds 0.99 in positive ion mode, indicating the high reproducibility and stability of the LC-MS-based metabolic profiling method, thereby confirming the reliability and quality of the data for downstream analysis.PCA revealed that PC1 accounts for 36.13% of the variance, and PC2 explains 12.76% (Fig. 3D), demonstrating clear separation between the control and experimental groups. To further identify potential biomarkers, partial least squares discriminant analysis (PLS-DA) was conducted (Fig. 3E). All model parameters show R2values greater than Q2, and the intercept of the Q2 regression line with the Y-axis is below zero, confirming that the model is robust and not overfitted. Differential metabolite screening (Fig. 3F) identified the top ten significantly upregulated metabolites in the experimental group under positive ion mode. To interpret their biological roles, KEGG pathway enrichment analysis was performed using ClusterProfiler. As shown in the enrichment plot (Fig. 3G), smaller P-values indicate more statistically significant enrichment. The color of each dot reflects the P-value from the hypergeometric test—lower values correspond to higher significance—while the size of the dot represents the number of differential metabolites mapped to the pathway, with larger dots indicating greater representation.Based on KEGG classification (Fig. 3H), the differential metabolites in positive ion mode are distributed across metabolic pathways and biological systems, with 218 metabolites involved in metabolic pathways and 29 in biological system pathways. Furthermore, Pearson correlation analysis was performed to explore inter-metabolite relationships, with the resulting correlation heatmap presented in Fig. 3I. In this heatmap, rows and columns represent individual metabolites, and colors range from blue (negative correlation) to red (positive correlation), with numerical values indicating the strength and direction of correlations.The results show a strong positive correlation between nicotinic acid riboside and pentadecanal (r = 0.75), suggesting coordinated changes in their abundance. Similarly, myriocin and pentadecanal exhibit a high positive correlation (r = 0.71), indicating synchronized variation. Stearidonic acid also shows a strong positive correlation with pentadecanal (r = 0.80), implying similar dynamic trends in their concentrations. Effect of L-citrulline on milk metabolites(In positive ion mode) A: Classification of metabolites in feces; B: cluster analysis of differential metabolites; C: correlation analysis of QC samples based on relative quantification of metabolites; D: PCA analysis of metabolites; E: partial least square discriminant analysis of differential metabolites; F: differential metabolite matchstick diagram; G: differential metabolite KEGG enrichment bubble diagram; H :differential metabolite KEGG classification analysis; I : differential metabolite correlation thermogr In negative ion mode As shown in Fig. 4, in negative ion mode, metabolites are classified into eight categories (Fig. 4A). The predominant categories include lipids and lipid-like molecules (35.46%) and phenylpropanoids and polyketides (14.44%), followed by organoheterocyclic compounds (12.54%), benzenoids (11.56%), organic acids and derivatives (11.56%), and organic oxygen compounds (10.09%). Minor categories include nucleosides, nucleotides, and analogues (1.26%), lignans, neolignans and related compounds (1.26%), alkaloids and derivatives (0.91%), organosulfur compounds (0.35%), organic nitrogen compounds (0.21%), organohalogen compounds (0.21%), and hydrocarbon derivatives (0.14%). Figure 4B illustrates the relative abundance and distribution patterns of metabolites, with each row representing a metabolite and each column corresponding to a sample. The color scale ranges from blue (low relative abundance, as low as − 2.5) to red (high relative abundance, up to 3.8), with color intensity reflecting the relative levels of metabolites across samples. As presented in Fig. 4C, the correlation among QC samples exceeds 0.99 in negative ion mode, indicating high reproducibility and stability of the LC-MS-based metabolic profiling method, confirming the reliability and quality of the data for downstream analysis. PCA revealed that PC1 accounts for 36.86% of the variance and PC2 explains 14.06% (Fig. 4D), showing a clear separation between the control and experimental groups. To identify potential biomarkers, PLS-DA was applied (Fig. 4E). All model parameters show R2 values greater than Q2, and the intercept of the Q2 regression line with the Y-axis is below zero, confirming that the model is robust and free from overfitting.Differential metabolite screening (Fig. 4F) identified the top ten significantly upregulated metabolites in the experimental group in negative ion mode. KEGG pathway enrichment analysis, performed using ClusterProfiler, was used to interpret the functional implications of these metabolites. As shown in the enrichment plot (Fig. 4G), smaller P-values indicate more statistically significant enrichment. The color of each dot reflects the P-value from the hypergeometric test—lower values correspond to higher significance—while the size of the dot represents the number of differential metabolites mapped to the pathway, with larger dots indicating greater representation.Based on KEGG classification (Fig. 4H), the differential metabolites in negative ion mode are distributed across both metabolic pathways and biological systems, with 221 metabolites involved in metabolic pathways and 33 in biological system pathways. Furthermore, Pearson correlation analysis was conducted to explore inter-metabolite relationships, with the resulting correlation heatmap shown in Fig. 4I. In this heatmap, rows and columns represent individual metabolites, and colors range from blue (negative correlation) to red (positive correlation), with numerical values indicating the strength and direction of correlations. The results demonstrate that grewia (Grewin) exhibits strong positive correlations with several metabolites (r ≈ 0.77), and 1-heptadecanoyl-sn-glycero-3-phosphate shows moderate to strong positive correlations with multiple metabolites (r ≈ 0.64). These high positive correlation coefficients suggest coordinated changes in metabolite levels. When the concentration of one metabolite increases, the other tends to increase as well, implying potential synergistic effects in metabolic regulation or biological processes. Effect of L-citrulline on milk metabolites( Under negative ion mode) A: classification of metabolites in feces; B: cluster analysis of differential metabolites; C: correlation analysis of QC samples based on relative quantification of metabolites; D: PCA analysis of metabolites; E: partial least square discriminant analysis of differential metabolites; F: differential metabolite matchstick diagram; G: differential metabolite KEGG enrichment bubble diagram; H:differential metabolite KEGG classification analysis; I: differential metabolite correlation thermogram Correlation analysis of differential metabolites Ten upregulated differential metabolites were identified, and their correlations with lactation performance and differential bacterial populations were analyzed. As shown in Fig. 5, the color intensity indicates the strength of correlation: red denotes a strong positive correlation, while blue signifies a strong negative correlation. The size of the dot, represents the absolute value of the correlation. with larger dots indicating higher values. The asterisk (*) denotes statistically significant correlations (P < 0.05). As depicted in Fig. 5A. 5-valerolactone shows a positive correlation with both milk production and butterfat percentage. Bistratamide F correlates positively with the milk protein ratio, while N, N,N-dicyclohexylurea is positively correlated with non-fat solids. As presented in Figs. 5B and 7-hydroxyl-6-methyl hepta-3,5-diol correlates positively with Ruminococcaceae_bacterium_AE2021. As presented in to Fig. 5C, cis-2-dodecenoic acid is positively correlated with milk production, 3,4-o-isopropidine-shikimic acid is positively correlated with butterfat percentage, and tributyrin is positively correlated with non-fat solids. Finally, as seen in Fig. 5D 3-hydroxysebacic acid positively correlates with Ruminococcus _ Bacillus _ AE2021. Correlation analysis (in positive and negative ion mode) A is the difference metabolite, milk yield and milk composition in positive ion mode; B is differential metabolite and differential bacteria in positive ion mode; C is the difference metabolite, milk yield and milk composition in negative ion mode; D is differential metabolite and differential bacteria in negative ion mode Discussion Amino acids, are essential building blocks of proteins and play a critical role in equine physiology, especially in lactating mares. Inadequate amino acid intake can impair milk production, subsequently hindering foal growth and development [17, 18]. The results of this study show that supplementing lactating mares with L-citrulline significantly boosts milk production. This aligns with findings from a study in Alpine goats, where daily supplementation of 20 g of L-citrulline increased milk yield [19]. One possible mechanism for this effect is that L-citrulline serves as a precursor to arginine. After absorption, L-citrulline is converted to arginine in the kidneys, which promoting mammary gland development, enhancing milk production, and improving milk quality [20, 21]. Additionally, L-citrulline may be converted into oxaloacetic acid, which enters the gluconeogenic pathway, and is ultimately transformed into phosphoenolpyruvate and lactose in the mammary glands, thereby increasing milk yield [22]. Milk yield, however, is not the sole determinant of lactation performance; milk quality is equally crucial. Important quality indicators include milk fat percentage, protein content, and lactose levels [23, 24]. In the present study, L-citrulline supplementation (15, 30, or 45 g/d) during the lactation period resulted in milk protein, fat, and lactose contents ranging from 1.70% to 1.74%, 1.17% to 1.37%, and 6.20% to 6.28%, respectively, across all experimental groups. These values are comparable to those observed in mares supplemented with lysine and threonine [25], indicating that amino acid supplementation can significantly influence the composition of mare’s milk. An et al. [26] also support these findings, showing that L-citrulline supplementation in Hu sheep ewes from day 60 of gestation increased milk fat content during the first 30 days postpartum, consistent with the results of this study. Thus, L-citrulline likely provides essential amino acids for mammary cells, participates in milk fat synthesis, and supports both fat synthesis and mammary tissue growth. L-citrulline, a conditionally essential amino acid, cannot be synthesized in sufficient quantities by the body and must be obtained through dietary sources or supplementation. It serves as a key precursor in the biosynthesis of arginine [27]. Arginine, an essential amino acid, plays a vital role in providing nutritional support enhancing reproductive promoting, and promotes growth and development [28]. The findings of this study indicate that supplementing lactating mares with L-citrulline significantly increases milk production. Consequently, foals’ body weight and physical dimensions also improve in parallel with the enhanced milk yield. Thus, L-citrulline not only enhances lactation performance but also supports the growth and development of offspring. Previous studies have shown that L-citrulline supplementation in the diets of Hu sheep ewes and lambs significantly increased body weight of both groups [29, 26, 30]. This effect is likely due to the conversion of L-citrulline into arginine in the kidneys, where it participates in the urea cycle, generating urea and ornithine. This process may enhance milk quality, milk amino acid content, and promote foal development [31]. Additionally, unmetabolized L-citrulline may be transferred to the foal via breast milk, stimulating the secretion of growth hormone and influencing foal growth and development. Although plasma levels of L-citrulline and arginine were not measured in this study, future experiments will further explore the factors influencing lactation and growth performance based on metabolic profiling. Animal health is influenced by multiple factors, with blood biochemical indicators serving as key parameters for assessing physiological conditions. In this study, protein metabolism-related indicators such as TP, ALB, GLB, UREA and CREA were analyzed to reflect the animal’s ability to absorb dietary crude protein, manage nitrogen metabolism, and deposit nutrients [32]. L-citrulline supplementation had no significant effect on these parameters. Regarding carbohydrate and energy metabolism, Glu and TG are important markers of sugar metabolism and energy conversion, both of which are essential for maintaining normal physiological function. The study found that L-citrulline supplementation did not significantly alter these indicators. Concerning of liver and kidney function as well as energy metabolism-related enzymes, results showed that ALT levels were higher in experimental groups I and III compared to the control group, while ALP levels in experimental group I were lower than those in the control group. The levels of ALP, LDH, and CK all decreased as the L-citrulline dosage increased. Li et al. [33] reported that L-citrulline supplementation in horses raised plasma ALB levels., while Zhang et al. [30] observed significant increases in plasma TP, ALB, and GLB concentrations in lambs. These which contrasts with the findings contrast with our study, where L-citrulline supplementation in lactating mares did not significantly affect plasma protein content. This discrepancy may be attributed to differences in physiological status, metabolic demands, and nutrient utilization efficiency among the experimental animals. VFAs in equine feces are primarily produced through microbial fermentation in the cecum and colon, playing a critical role in maintaining intestinal health and providing energy to the host [34]. Intestinal pH, which affects the activity of digestive enzymes and the balance of gut microbiota, is also an important factor in gastrointestinal function [35]. In the present study, supplementation with L-citrulline (15, 30, or 45 g/d) led to numerical increases in acetic acid, propionic acid, and total VFAs in the feces of lactating mares compared to the control group. However, no statistically significant differences were found (P > 0.05). Notably, fecal pH consistently decreased with increasing L-citrulline dosage. This may be attributed to two factors: first, enhanced VFA production, which likely increases hydrogen ion (H+) concentration in the intestinal environment [36]; second, L-citrulline, acting as a nitrogen source, contributes to protein synthesis and modulates microbial metabolism in the gut. Specifically, it increased the abundance of Ruminococcaceae_bacterium_AE2021, a bacterium potentially involved in the production of short-chain fatty acids such as acetic and butyric acids, thus lowering intestinal pH. L-citrulline, a non-essential amino acid can be synthesized in intestinal epithelial cells via the glutamate-ornithine pathway and glutamine metabolism, using substrates like glutamine, glutamic acid, and proline [37, 38]. It may influence the intestinal microbiota composition and diversity by modulating nitrogen metabolism and the intestinal environment.Microbial analysis revealed that the core intestinal microbiota of mares shared 4,051 OTUs between the control and experimental groups. The control group contained 3,568 unique OTUs, with Chao1 = 2158.68, Pielou_e = 0.85, Shannon = 9.36, and Simpson = 0.99. The experimental group contained 3,482 unique OTUs, with Chao1 = 2138.04, Pielou_e = 0.86, Shannon = 9.46, and Simpson = 0.99. The high overlap between groups suggests that core ecosystem functions were preserved. However, five bacterial taxa exhibited differences at multiple classification levels (class, order, family, genus, species).At the genus level, Schwartzia is involved in amino acid metabolism and fatty acid synthesis, utilizing L-citrulline as a nitrogen source to enhance arginine metabolism. It may also support intestinal health by producing short-chain fatty acids. At the species level, Ruminococcaceae_bacterium_AE2021 is likely involved in cellulose degradation, with its abundance potentially modulated by L-citrulline induced changes in intestinal nitrogen metabolism. At the class level, Gracilibacteria may participate in the utilization of nitrogen sources derived from citrulline in the nitrogen cycle, thereby enhancing intestinal nitrogen compound metabolism. At the order level, Absconditabacteriales_SR1_is associated with the degradation of complex carbohydrates, and L-citrulline may facilitate short-chain fatty acid production by influencing nitrogen metabolism. At the family level, Ethanoligenenaceae, exhibiting ethanol-producing fermentation traits, may activate its fermentation pathway in response to nitrogen substrates or intermediates derived from L-citrulline, thereby regulating intestinal energy balance.Tax4Fun analysis revealed that gene abundance was highest in categories related to digestion and metabolism, genetic information processing, and environmental information processing. At the phylum level, Firmicutes and Bacteroidetes were dominant, though no significant differences were found. The Firmicutes to Bacteroidetes ratio may influence nitrogen source distribution in the gut: when Firmicutes dominate, their robust lipid metabolism and fermentation capabilities can increase amino acid breakdown and nitrogen accumulation; conversely, Bacteroidetes dominance may reduce excessive fat deposition and provide energy for intestinal epithelial cells to synthesize L-citrulline [39]. At the species level, the experimental group exhibited a higher abundance of Ruminococcaceae_bacterium_AE2021 which can produce short-chain fatty acids (e.g., butyrate) by degrading complex carbohydrates, potentially improving the efficiency of L-citrulline as a nitrogen source. Du et al. [40] reported that the presence of Ruminococcaceae_bacterium_AE2021 in the gut microbiota of fattening pigs supplemented with L-citrulline might enhance nutrient digestion and absorption. The intestinal microbiota and its metabolic products play a pivotal role in maintaining host intestinal homeostasis. Certain metabolites have been identified for their functional properties, promoting growth, development, and overall health [41]. Metabolomics analysis identified 208 differential metabolites in both positive and negative ion modes, primarily consisting of lipids and organic acids. KEGG enrichment pathway analysis focused on two main areas: biological systems and metabolism. A total of 17 pathways were detected, including the pentose phosphate pathway, linoleic acid metabolism, and galactose metabolism. Notably, the pentose phosphate pathway is crucial in mammary cells, particularly during lactation, as it supports the proliferation of mammary epithelial cells and the expression of enzymes involved in milk fat synthesis through the oxidative and non-oxidative phases [42]. Linoleic acid, the primary polyunsaturated fatty acid (PUFA) in horse milk, and its metabolic products, such as arachidonic acid, contribute to the structural composition of triacylglycerols (TAG) in milk fat. These metabolites increase the PUFA content of milk fat, which can enhance foal growth performance [43]. Furthermore, amino acids in mammals are often derived from linoleic acid metabolism, where linoleic acid from dietary sources is desaturated to form gamma-linolenic acid, which is then further desaturated and extended to form amino acids, ultimately producing prostatic acids, and other metabolites [44]. Correlation analyses revealed significant positive relationships between various metabolites, milk production indicators, and specific intestinal bacteria. For instance, 5-valerolactone was correlated with milk yield and milk fat percentage, while 3,4-O-isopropylidene-shikimic acid was positively correlated with milk fat percentage. Cis-2-dodecenoic acid was positively associated with milk yield, and tributyrin and N, N-dicyclohexylurea were correlated with non-fat solids. The ratio of double-layer amine F to milk protein was positively linked to these indicators. Additionally, 7-hydroxy-6-methylheptyl-3,5-dienal and 3-hydroxysebacic acid showed positive correlations with Ruminococcaceae_ bacterium_ AE2021. These associations offer insights into the milk quality regulation mechanism and the interaction between intestinal flora and host metabolism, enriching the theoretical understanding of milk quality control, and confirming the role of metabolites regulating in milk yield and composition. In conclusion, L-citrulline supplementation enhances metabolites involved in amino acid and fatty acid metabolism, thereby improving the associated metabolic pathways during lactation. On one hand, it promotes the utilization and metabolism of nutrients like protein and lipids in mares, on the other hand, it stimulates the growth and proliferation of mammary glands, ultimately increasing milk yield and improving milk quality. Conclusion L-citrulline can increase the high milk yield of mares, improve the milk fat quality, reduce the contents of lactate dehydrogenase and creatine kinase in blood, and promote the growth and development of nursing foals. The results of 16s sequencing showed that L-citrulline could change the microbial diversity and flora abundance in mares’ intestines (F082 and Ruminococcaceae_bacterium_AE2021), and Tax4Fun analysis showed that it had the highest functional abundance in digestion and metabolism, genetic information and environmental information processing. The LC-MS/MS results of fecal samples showed that the differential metabolites were mainly distributed in biological system and metabolism, and these differential metabolites mainly concentrated in two significant pathways: Pentose phosphate pathway and Linoleic acid metabolism. These results may provide new insights for L- citrulline in improving lactation performance of lactating mares and growth performance of lactating foals. Acknowledgements The authors would like to acknowledge the Coudre Grassland, Zhaosu County, Ili Kazakh Autonomous Prefecture, Xinjiang racecourse for allowing sample collection from their foal’s, the vet and staff for assistance with for care of the animals and samples collection. Authors’ contributions YM: Contributed significantly to article conception and design, data acquisition, data analysis, and interpretation.PL、JL、QH and FY: Performed animal feeding and sample collection. All authors read and approved the final manuscript.KL: Project administration and funding acquisition.XL: Participated in critical revisions of important knowledge and content in the manuscript. Funding Effect of Citrulline on Reproductive Performance, Milk Yield, and Investigation of Metabolic Mechanisms in Yili Mare Horses(Project No:20230901001). Data availability All data generated or analysed during this study are included in this published article. Declarations Ethics approval and consent to participate All protocols were approved by the Animal Care and Use Committee of Xinjiang Agricultural University (permission number 2022020). Informed consent - Owners gave informed consent for their animals’ inclusion in the study. All methods were carried out in accordance with relevant guidelines and regulations for the use of animal subjects. The study was carried out in compliance with the ARRIVE guidelines. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement All data generated or analysed during this study are included in this published article.