Multi-kingdom fecal microbiome and virus–host interactions associated with growth performance of indigenous beef calves in Guizhou

Abstract

Background

The associations between the gut microbiome and growth performance in calves have been investigated; however, most existing studies have primarily focused on rumen microbiomes. Whether fecal microbiomes in terms of composition and function are altered among calves with different growth rates remains unclear. Therefore, the objective of this study was to investigate how fecal microbiomes influence calf growth rates. A total of 16 beef calves under the same management were recruited and classified into two groups based on their growth rates (average daily gain: ADG; 4-month-old, n = 8 per group x 2 growth rate groups). At 4 months of age, fecal samples were collected from the rectum for the quantification of volatile fatty acids (VFAs) and characterization of microbial communities via metagenomic sequencing.

Results

The VFA profiles did not differ between the two groups. Calves with higher growth rates exhibited lower bacterial and archaeal Shannon diversity, and the overall microbial community structure showed a clear separation between the two groups. Moreover, fecal bacterial and archaeal species associated with improved growth performance were identified, characterized by the enrichment of Alistipes shahii, Alistipes onderdonkii, Bifidobacterium thermophilum, Akkermansia glycaniphila, and Methanobrevibacter sp. AbM4 in calves with higher growth rates. In addition, the metabolic pathways involved in lipid and amino acid metabolism and CAZyme genes linked to carbohydrate degradation were enriched in the calves with better growth performance. The viral community composition and diversity differed between the two groups, with lower diversity observed in calves exhibiting higher growth rates. Additionally, viruses predicted to infect bacterial hosts such as Prevotella and Succinivibrio, which are involved in carbohydrate degradation, were positively associated with ADG. Interestingly, a virus associated with Methanobrevibacter sp017652345 exhibited a positive correlation with ADG. The relationships between fecal microbes and host phenotypic traits were divergent between the two groups.

Conclusions

These findings suggest that fecal microbiomes are associated with calf growth rates through potential multi-kingdom interactions, particularly those between viruses and their prokaryotic hosts, indicating possible avenues to improve animal performance via microbiome modulation.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12866-025-04631-y.

Introduction

Ruminants could convert low-quality plant materials into animal products including meat and milk consumed by humans, which is largely attributed to the myriad microorganisms in the gastrointestinal tract [1]. Since rumen microbiomes are reported to be linked with feed efficiency [2], milk production [3], and the host health [4], they have attracted increasing attention recently. However, the fecal microbiota, which plays important roles in the function of the animal immune system and health [5–7], remains inadequately elucidated in terms of its association with host phenotypes. Fecal microbiota was preferred over rumen microbiota in the early stages of life due to the accessibility and ease of sampling [8], and fecal indices can be treated as a non-invasive proxy for ruminal activity assessment [9]. Primary studies on gut commensal microbiota have shown that early-life fecal microbiota colonization is associated with immune system development, growth, and health in young ruminants [10, 11], and variation in fecal microbiota community of calves may affect host performance in later life [12]. Moreover, accumulated pieces of evidence have documented that fecal microbiota of ruminants could modulate the host nutrient metabolism and immune defense through producing microbial metabolites such as short-chain fatty acids (SCFA), vitamins, and bile acids [13, 14]. In addition, the associations of fecal microbial community structure with methane emissions, feed efficiency, and animal health have been documented [15–17], highlighting the importance of fecal microbiota on the host productivity and health. These investigations specifically emphasized the significance of fecal bacteria on the host phenotypes and most of them are based on amplicon sequencing, which makes it impossible to comprehensively investigate the relationships between compositional and functional profiles of fecal microbiomes with host phenotypes.

Within the gut microbiome, viruses, which are an important component of the microbial community that closely interact with other microorganisms [18], have been neglected due to methodological limitations and a lack of awareness of their importance. Canonically, viruses are clustered into two types (lytic and temperate), both of which can modulate microbial metabolism and composition [19]. Viruses could enhance the degradation of complex plant carbohydrates and promote energy production and harvest via auxiliary metabolic genes (AMGs) [20]. Prior studies have documented that gut viral composition and diversity are associated with the host’s health and growth performance. For example, alterations in human gut viruses have been associated with metabolic and autoimmune diseases [21]. Moreover, differences in gut virome between early-weaned healthy piglets and those with diarrhea have also been illustrated [22]. In addition, fecal virome transplantation can determine mouse phenotypes (lean and obesity) via altering the fecal microbiome [23]. Additionally, close relationships between fecal viruses and host production performance in pigs (i.e., average daily gain and feed efficiency) were revealed [24, 25]. These studies have deepened our understanding of how fecal viruses affect the host phenotypes through interacting with their hosts in monogastric animals, but the knowledge in ruminants is scarce. To date, only a few studies have investigated the correlations between fecal viruses and calf health [26, 27], and the impact of fecal viruses on host phenotypes in calves remains poorly understood.

Viruses that infect bacteria and archaea are the most plentiful on earth [28] and they can modulate microbial metabolism and affect the animals’ outcomes [29], suggesting the possibility to utilize as the modulator to manipulate the bacteria/archaea to improve the animals’ production performance. Moreover, it has been reported that early life is the manipulation window for modulating gut microbiomes and has a long-term effect on animal performance [30]. Therefore, it is necessary to investigate gut viruses to determine whether they can be programmed to improve calf growth performance.

Here, in this study, sixteen postweaning calves (4-momth-old) were selected with different average daily gain (ADG) from a total of 89 young beef calves under the same management to elucidate the relationships between fecal microbes (bacteria, archaea, and viruses) in terms of their composition and function and the animal’s growth rate. This study may uncover ADG-associated fecal microbiome and virus-host profiles that are beneficial for developing effective strategies to improve the animals’ production performance through microbial makers regulation.

Materials and methods

Experimental animals and sampling

A total of 89 healthy newborn beef calves were initially recruited in this experiment from a local farm in Bijie City, Guizhou, China (27°06′28″N 106°12′19″E). The Guizhou cattle breed is well adapted to the high-altitude and mountainous environment of Guizhou and is primarily raised for beef production. The birth weight of all calves was measured immediately after birth and then were assigned to individual pens. The calves were reared together under the same management with free access to fresh water. All calves were fed the same diet after weaning (2-month-old) and were weighed for 5 days consecutively to calculate the average daily gain (ADG) before sampling at the age of 4-month-old (ranges from 118 to 121 days). The ADG (g/day) was calculated as (final body weight − initial body weight)/feeding duration (days). Afterwards, sixteen calves were selected based on their individual ADG, including the top 8 calves classified as high growth rate group (H-group; male: female = 4:4) while the bottom 8 calves classified as low growth rate group (L-group; male: female = 5:3). All selected calves shared a common paternal lineage to minimize potential confounding effects of genetics on the gut microbiome, and no antibiotics or anthelmintic treatments were administered during the experimental period. To weaken the potential effect of feed intake on ADG, the feed intake of all calves was recorded 7 days before sampling to ensure that there is no difference between H and L calves. In brief, feed intake for each calf was monitored continuously for 7 days by recording both the feed offered and the feed refused. Fecal samples (approximately 70 g) from each calf were collected by rectal stimulation using sterilized gloves before the morning feeding and placed into the 100 mL screw cap tube followed by snap frozen in liquid nitrogen [31]. Samples were then transferred to − 80 °C for subsequent analysis in the lab. This routine, minimally invasive procedure, was performed by trained personnel to minimize discomfort, and no animals were euthanized or anaesthetized during fecal collection.

Chemical analysis

Feed samples (Additional file 1: Table S1. Chemical composition of feed ingredients (% of dry matter) was dried at 65 °C for 48 h to measure the dry matter (DM) content. The crude protein (CP) content, ether extract (EE), and ash were determined following the AOAC protocol. The neutral detergent fiber (NDF) and acid detergent fiber (ADF) content were analyzed using an Ankom 2000i automated fiber analyzer (Ankom Technology, Fairport, NY, USA) following the manufacturer’s guidelines.

Fecal VFAs assays

The quantification of volatile fatty acids (VFAs, including acetate, propionate, butyrate, valerate, iso butyrate, and isovalerate) in the fecal samples was implemented using gas chromatograph (Aligent 7820 A, Santa Clara, CA, USA), as previously described [15]. In brief, about 1.0 g fecal sample was vortex-mixed with 3 mL distilled water and then centrifuged at 20,000 g at 4 °C for 20 min. After that, 400 µL supernatant was mixed with 200 µL of 25% metaphosphoric acid (w/v), followed by centrifuging at 16,000 g at 4 °C for 15 min. Then the supernatant was filtered through a 0.22 μm filter prior to determining VFA concentrations by a gas chromatograph.

DNA extraction, metagenomic library preparation and sequencing

The genomic DNA was extracted from fecal samples using QIAamp Fast DNA Stool Mini Kit (QIAGEN) following the manufacturer’s protocol. The quality and concentration of extracted DNA were evaluated using 1% agarose gel electrophoresis and a NanoDrop 2000 spectrophotometer (Thermo Scientific, Wilmington, USA). The qualified DNA was sequenced on the Illumina NovaSeq X Plus platform (150 bp paired-end reads) at Majorbio Bioinformatics Technology Co. Ltd. (Shanghai, China). Raw sequence data were processed for quality control (reads with q < 20 and length < 50 bp or N bases were removed) and adapter removal using Trimmomatic (v.0.38) [32]. Next, host-derived sequences were removed by mapping the quality-filtered reads to the bovine genomic DNA sequences (ARS-UCD1.3) using bowtie2 [33]. After that, the clean data were de novo assembled using MEGAHIT software (v.1.2.9) [34]. Open reading frames (ORFs) were predicted from each contig using the MetaGene [35], and a non-redundant gene catalog was constructed using CD-HIT [36] according to 95% identity and 90% coverage. Taxonomic assignment was performed using Kraken2 [37] followed by calculating the relative abundance of taxa (phylum, genus, and species) in each sample using Bracken [38]. Taxon at the species level with a relative abundance > 0.01% in at least 50% of beef calves within each group were retained for downstream analysis. The non-redundant ORFs with read length more than 300 bp were annotated using DIAMOND [39] against the KEGG database (Release 113.0, accessed 2025‑01‑01) and CAZy database (CAZyDB_2024-07–14), with an E value of 1e-5.

Individual sample assemblies and co-assemblies (> 1 kb) were further binned to the metagenome-assembled genomes (MAGs) using MaxBin, MetaBAT2, and CONCOCT with default parameters in metaWRAP (v1.3.2) [40]. The refinement of the MAGs was conducted using MAGScoT (v1.0.0), and the refined MAGs were then clustered into species-level genome bins (SGBs) using dRep (v3.5.0) with a 95% ANI cutoff. The taxonomy classification for all the SGBs was performed in GTDB-Tk (v2.4.0) [41], including gene identification for each SGB with Prodigal (v2.6.3). The abundance of SGBs was calculated in CoverM (v0.7.0) with the “coverm genome” module. Genes in each SGB were annotated against the KEGG database using DIAMOND, and automated CAZyme annotation for each SGB was performed using run-dbCAN (v4.1.0). The abundances of KEGG pathways and CAZymes were normalized into counts per million reads (CPM), and the frequency was more than 50% within each group was used for further statistical analysis [42].

Identification, taxonomic annotation, and host prediction for viral sequences

All contigs longer than 5 kb were used for downstream analysis in each sample. The viral contigs were identified using VirSorter2 (v2.2.3, max_score > 0.5) [43] and DeepVirFinder (v1.0, Score > 0.9 and P < 0.05) [44], followed by assessing the quality of the potential viral contigs using CheckV v.1.0.1 (database v.1.4) [45]. The viral contigs with medium or higher quality (completeness score > 50%) were used for further analysis. Next, the viral contigs from each sample were pooled and were dereplicated using CD-HIT (95% nucleotide identity and 85% coverage), and the longest contig per cluster was taken as representative (vOTU). After that, the protein-coding sequences (ORFs) of vOTUs were predicted using Prodigal v2.6.3 with the “-p meta” parameter. Taxonomic classification of vOTU was performed using vConTACT2 and combined with ORFs aligned to the NCBI RefSeq Viral genomes according to the previous description [46]. The virus ORFs were first aligned to Pfam, VOGDB, and PHROGs databases to retrieve the large terminase subunit (TerL), after which all TerL proteins were then aligned using MAFFT (v7.505) [47], and trimAI (v1.4.1) was applied to remove positions with 90% or more from the alignments [48]. The virus phylogeny tree was constructed using FastTree (v 2.1.11) [49]. The lifecycle of viruses was predicted using BACPHLIP [50] and phage-host predictions were performed with IPhoP [51].

The relative abundance of each vOTU was calculated by dividing the number of reads belonging to this vOTU by the total number of mapping reads. For the family-level profiles, the relative abundance of each viral family was generated by aggregating the relative abundance of vOTUs belonging to the same family. To reveal the virus-host associations in the community, the correlations between bacterial species (abundance > 0.01%) and vOTUs (abundance > 0.01%) were calculated using Spearman’s correlations (two-sided) in the WGCNA package and P < 0.05 (corrected with Benjamini–Hochberg method) indicated statistically significant.

Statistical analysis

Statistical analysis of birth weight, average daily gain (ADG), feed intake, and fecal VFAs between groups (H vs. L) was conducted using the independent sample T-test in Prism 8.0 (GraphPad Software, USA), and P < 0.05 was considered statistically significant. The potential effects of sex, age, birth weight, and feed intake on fecal microbial composition were considered, and no significant effects were observed; therefore, these factors were not included in subsequent analyses. Microbial alpha diversity indices (Shannon and Chao1 indices for bacterial and archaeal species, observed number and Shannon index for vOTU) were computed using the vegan package, and their differences between the two groups were determined using Wilcoxon rank-sum test, with P-value < 0.05 indicating statistical significance. Dissimilarity of the fecal microbiota profiles (bacteria, archaea, and virus) were assessed using principal coordinate analysis (PCoA) based on Bray-Curtis distance, and the statistical significance between the two groups was tested using permutational multivariate analysis of variance (adonis) in the vegan package with 999 permutations at the species (bacteria and archaea) or vOTU (virus) level. The differences in abundance of microbial phyla, species, vOTU, and CAZymes between H and L calves were assessed using the DESeq2 package [52] based on a two-sided Wald test, with adjusted P < 0.05 indicating statistical significance (Benjamini–Hochberg method). The pathway enrichment analysis was conducted for KO genes using the R package “clusterProfiler”. The LEfSe analysis was conducted to identify significant differences in SGBs between H group and L group, and P < 0.05 and LDA > 2 were treated as significant differences [53]. The correlations between differential microbial species (bacteria and archaea) and vOTU and phenotypic types (ADG, acetate, propionate, and butyrate) were analyzed by LinKET, and P values < 0.05 were indicated as statistically significant. Results were presented as mean ± SEM unless otherwise noted.

Results

Profiles of growth rates, body weight, and fecal fermentation

The birth weight (kg) between H (high average daily gain) and L (low average daily gain) groups showed no significant difference (P > 0.05, Fig. 1A), whereas the average daily gain (ADG) was greater in the H group compared to L group (P < 0.05, Fig. 1A). Moreover, there were no significant differences in total VFA concentration (mmol/L) or molar proportion of acetate, propionate, butyrate, isobutyrate, isovalerate, and valerate between the two groups (P > 0.05, Fig. 1B).

Fig. 1.

Comparison of phenotypes between H and L calves. A Differences in birth weight and average daily gain (ADG) between H and L groups. B Differences in fecal fermentation between H and L groups. H: high average daily gain; L: low average daily gain. Data are expressed as means ± SEM. Significantly different was assessed by an independent sample t-test between H and L groups (n = 8 per group). ***P < 0.001

Compositional profiles of fecal microbiome between H and L calves

A total of 1,313,852,888 reads (82,115,806 ± 3,598,381 reads per sample, mean ± SEM) were generated from 16 beef calves (4-month-old, n = 8 per group). A median of 550, 093 (range from 322,733 to 930,598; N50 of 61,797 ± 71 bp) contigs were obtained per sample after quality control, removal of host reads, and de novo assembly (Additional file 1: Table S2. Summary statistics of raw metagenomic sequencing data).

The bacterial Shannon index in the H group was significantly lower than that in the L group (P < 0.05, Fig. 2A), while Chao 1 index showed opposite trend (P < 0.01, Fig. 2A). A significant difference in the bacterial community was revealed, as reflected by a clear separation between H and L calves (adonis R2: 0.19; P = 0.002, Fig. 2C). The archaeal Shannon index in the L group was greater than that in the H group (P < 0.001, Fig. 2B), and the archaeal community showed a clear separation between these two groups (adonis R2: 0.5; P = 0.001, Fig. 2D). For taxonomic profiles (based on Bracken-derived abundances), the predominant bacterial phyla belonged to Bacillota (49.1% ± 0.03), followed by Bacteroidota (35.05% ± 0.03), Actinomycetota (8.68% ± 0.01), and Pseudomonadota (6.74% ± 0.01) across all calves (Fig. 3A). Among them, the relative abundance of Bacillota was enriched in the L group than H group, while the relative abundance of Bacteroidota showed opposite trend (P < 0.05, Fig. 3A). At the species level, the fecal microbiota was dominated by Bacteroides uniformis (9.74% ± 0.00), Parafannyhessea umbonate (7.26% ± 0.01), Bacteroides xylanisolvens (5.43% ± 0.01), Chordicoccus furentiruminis (5.07% ± 0.01), Lentibacillus daqui (4.72% ± 0.01), Bifidobacterium pseudolongum (4.59% ± 0.009), and Phocaeicola vulgatus (4.09% ± 0.01) (Fig. 3B). Furthermore, 23 differential bacterial species were detected between H group and L group, with 14 species being enriched in the H group (P < 0.05, Fig. 3C). For archaea, more than 99% of archaeal sequences were assigned to phylum Euryarchaeota. At the species level, 9 species were identified, and the top 5 species were Methanobrevibacter sp. AbM4, Methanobrevibacter millerae, Methanobrevibacter sp. YE315, Methanobrevibacter ruminantium, and Methanobrevibacter olleyae, collectively accounting for > 97.99% of the total archaeal sequences (Fig. 3D). Among them, the relative abundance of Methanobrevibacter sp. AbM4 (H vs. L; 56.48% ± 0.12 vs. 2.07% ± 0.02) was greater in the H group, whereas the relative abundances of Methanobrevibacter ruminantium (3.27% ± 0.01 vs. 19.79% ± 0.03), Methanobrevibacter olleyae (2.17% ± 0.004 vs. 14.37% ± 0.03), and Methanobrevibacter sp. YE315 (5.44% ± 0.02 vs. 36.9% ± 0.04) were enriched in the L group (P < 0.05, Fig. 3D).

Fig. 2.

Comparison of the diversity of fecal microbiomes between H and L calves. Boxplot shows bacterial (A) and archaeal (B) alpha diversity (Shannon and Chao1 indices) at the species level. Principal coordinate analysis (PCoA) plot of bacterial (C) and archaeal (D) variation based on the Bray-Curtis distance matrix. Significant differences in alpha diversity indices were assessed by Wilcoxon rank-sum test between H and L groups (n = 8 per group), and the statistical significance of the overall microbial community between the two groups was tested using Adonis. **P < 0.01, and ***P < 0.001

Fig. 3.

Microbial community composition between H and L calves. Bacterial profiles at the phylum (A) and species (B) levels between H and L calves. C Differentially abundant bacterial species (P < 0.05) between H and L calves. For a taxon represented by a single bar, it was detected in only one group and absent in the other. D Archaeal composition at species level between H and L calves. Differences between H and L groups (n = 8 per group) were analyzed using DESeq2. *P < 0.05 and ***P < 0.001

Functional profiles of fecal microbiome between H and L calves

Next, the functional profiles of fecal microbiomes between H and L groups were compared, and a total of 51 KEGG level 3 pathways were found to be significantly different between the two groups (P < 0.05, Fig. 4A). Among them, the abundances (counts per million, CPM) of 31 pathways were enriched in the H group, with most of them involved in amino acid metabolism (e.g., valine, leucine and isoleucine biosynthesis and cysteine and methionine metabolism), carbohydrate metabolism (e.g., propanoate metabolism and pyruvate metabolism), metabolism of cofactors and vitamins (e.g., folate biosynthesis and pantothenate and CoA biosynthesis), metabolism of terpenoids and polyketides (e.g., nonribosomal peptide structures and biosynthesis of 12-, 14- and 16-membered macrolides), and biosynthesis of other secondary metabolites (e.g., carbapenem biosynthesis and biosynthesis of various other secondary metabolites) (P < 0.05, Fig. 4A). Besides, 20 KEGG pathways were enriched in the L group, and most of them were related to amino acid metabolism (e.g., alanine, aspartate and glutamate metabolism and glycine, serine and threonine metabolism), carbohydrate metabolism (e.g., glyoxylate and dicarboxylate metabolism, and pentose and glucuronate interconversions), and metabolism of cofactors and vitamins (e.g., lipoic acid metabolism and vitamin B6 metabolism) (P < 0.05, Fig. 4A).

Fig. 4.

Fecal microbiomes function between H and L calves. A Bubble plot showing the significantly enriched KEGG pathways in the H group and L group. Bubble color gradient reflects the enrichment significance (-log10 (FDR) value; Fisher’s exact test) and bubble size corresponds to the ratio of the number of genes mapped to a certain pathway. B Composition of CAZyme families between H and L calves. C Significantly different (P < 0.05) CAZyme families between H and L calves. The DESeq2 was used to analyze differences in CAZyme families between H and L groups (n = 8 per group)

The functional profiles of genes encoded for CAZymes were further investigated, revealing that genes assigned to glycoside hydrolases (GHs) were the most abundant, followed by glycoside transferases (GTs), carbohydrate binding modules (CBMs), carbohydrate esterases (CEs), polysaccharide lyases (PLs), and auxiliary activities (AA) (Fig. 4B). Eleven differential CAZyme families were identified between the two groups, with 7 enriched in the H calves (P < 0.05, Fig. 4C). Specifically, 4 CAZymes that involved in the degradation of cellulose (GH6 and GH94), hemicellulose (GH97), and sucrose (GH68) were significantly higher in the L group, while 7 CAZymes that play important roles in the degradation of mucin (GH84) and oligosaccharide (GH29), lipopolysaccharide (GT81 and GT41) and polysaccharides (GT10) synthesis, xylan utilization (CBM15), and sugar transportation (GT26) were enriched in the H group (P < 0.05, Fig. 4C).

To better understand the role of fecal microbiome in shaping calf growth performance, a total of 668 SGBs with completeness > 50% and contamination < 5% were reconstructed to facilitate the screening of key metabolic genes within microbial genomes, and over 50% of the microbial genome was affiliated with phylum Bacillota_A (Fig. 5A). The abundances of SGBs (generated from CoverM) were then compared between the two groups, and 5 and 6 SGBs encoding key metabolic genes were found to be enriched in the H and L groups, respectively (P < 0.05, LDA > 2, Fig, 5B). Specifically, the enriched SGBs in the H group were predominantly predicted to related to amino acid metabolism (e.g., valine, leucine and isoleucine biosynthesis and cysteine and methionine metabolism), carbohydrate metabolism (e.g., amino sugar and nucleotide sugar metabolism and glycolysis/gluconeogenesis), and the metabolism of cofactors and vitamins (e.g., folate biosynthesis and thiamine metabolism) (Fig. 5B), whereas the enriched SGBs in the L group primarily encoded the metabolic genes involved in the metabolism of cofactors and vitamins (e.g., porphyrin metabolism and pantothenate and CoA biosynthesis) (Fig. 5B). For CAZyme profiles, the enriched SGBs in the H group were primarily predicted to be involved in hemicellulose degradation, particularly through the activity of GH84 and GH29 families (Fig. 5B). In contrast, the enriched SGBs in the L group encoded metabolic genes related to both cellulose degradation (GH94) and starch degradation (GH97) (Fig. 5B).

Fig. 5.

Phylogenetic tree and distribution of metabolic genes in SGBs. A Phylogenetic tree of SGBs, the inner circle is a phylogenetic tree of 668 SGBs colored according to GTDB phylum-level taxonomic classifications; the heatmaps in the outer layer show the percentage of contamination and completeness, genome size, GC content (%), N50 length (Kb), and MAG number corresponding to each microbial genome. B Heatmap of selected KO genes and CAZyme families in differentially enriched microbial genomes (SGBs) in the H group or L group (n = 8 per group)

Overview of fecal Virome in H and L calves

A total of 4,492 vOTUs were classified as medium-quality or higher based on quality assessment using CheckV, including 828 complete, 1,392 high-quality, and 2,272 medium-quality vOTUs (Additional file 1: Table S3. The quality, phylum hosts, and phage types of vOTUs). Host-prediction of the vOTUs was available for 2753 vOTUs, with the most prevalent hosts identified as phyla Bacillota_A (synonym Firmicutes: 1475; 53.6%), Bacteroidota (651; 23.6%), and Bacillota (302; 10.97%) (Additional file 2: Fig. S1; Additional file 1: Table S3. The quality, phylum hosts, and phage types of vOTUs).

To investigate the changes in the alpha diversity of fecal virome between H and L calves, the observed number of vOTUs and the Shannon index were calculated at the vOTU level. The Shannon index and the number of vOTUs in the H calves were lower compared to L calves (P < 0.05, Fig. 6A). The principal coordinates analysis (PCoA) based on Bray–Curtis distances (at the vOTU level) revealed that fecal virome differed considerably between H and L calves (adonis R2: 0.17; P = 0.001, Fig. 6B).

Fig. 6.

Fecal virome profiles between H and L calves. A Boxplot showing viral alpha diversity (Shannon and Number of vOTU) at the vOTU level. B Principal coordinate analysis (PCoA) plot of virus variation based on the Bray-Curtis distance matrix, and the statistical significance of the overall microbial community between the two groups was tested using Adonis. C Composition of viral families between H and L calves. D Comparison of the temperate and virulent fecal viruses between H and L calves. The Wilcoxon rank-sum test was used to analyze differences between H and L groups (n = 8 per group). *P < 0.05 and ***P < 0.001

The composition of the fecal virome was then profiled at the family level, and a large proportion of the virome in both the H and L groups was found to be assigned to unclassified viruses (Fig. 6C). The majority of classified viral contigs were assigned to families unassigned Caudovirales, Myoviridae, Salasmaviridae, Straboviridae, and Siphoviridae in both H and L calves (Fig. 6C). The life cycle of vOTUs was further visualized based on bacteriophage lifestyle prediction, and a total of 4,492 vOTUs were identified, consisting of 2,231 temperate and 2,261 virulent vOTUs (Additional file 1: Table S3. The quality, phylum hosts, and phage types of vOTUs), and the relative abundances (based on direct read-mapping abundances) of temperate and virulent viruses between H and L groups showed no significant difference (P > 0.05, Fig. 6D). Next, the viral composition was analyzed at the vOTU level, and 698 differential vOTUs were identified between the H and L groups (P < 0.05, Additional file 1: Table S4. The differential abundant vOTUs between H and L calves). Among them, 36 vOTUs could be assigned to a known family, consisting of Straboviridae (11), Siphoviridae (7), Schitoviridae (7), Saparoviridae (3), Stanwilliamsviridae (2), Halomagnusviridae (1), Marseilleviridae (1), Myoviridae (1), Podoviridae (1), Saffermanviridae (1), and Salasmaviridae (1) (Additional file 1: Table S4. The differential abundant vOTUs between H and L calves). Notably, the abundances of vOTU76, vOTU108, vOTU110, and vOTU112, predicted to be linked to Methanobrevibacter based on virus–host prediction using IPhoP, were enriched in the L group compared to H group (Additional file 1: Table S4. The differential abundant vOTUs between H and L calves). Finally, analysis of the microbial–viral interaction networks revealed that the number of significant associations (P < 0.05) between microbiota and viruses was higher in the L group than H group (H vs. L; 85 vs. 123; Additional file 1: Table S5. The relationships between microbiota and viruses between H and L groups). Notably, Streptococcus sp. showed positive correlations with some vOTUs in the H group, while no correlation was observed in the L group (Additional file 1: Table S5. The relationships between microbiota and viruses between H and L groups).

Divergent correlations between microbes and phenotypic traits between H and L calves

Finally, the potential relationships between fecal microbes and phenotypes were explored, and the results showed that the relative abundances of Alistipes shahii (r = 0.88, P < 0.01) and Alistipes onderdonkii (r = 0.39, P < 0.05) correlated positively with ADG in the H calves, while the relative abundances of Alistipes senegalensis (r = 0.46, P < 0.05) and Bifidobacterium adolescentis (r = 0.52, P < 0.05) showed positive correlations with ADG in the L calves (Fig. 7A). In addition, the relative abundance of Flavonifractor plautii correlated positively with the concentrations of acetate (r = 0.54, P < 0.01) and propionate (r = 0.51, P < 0.01) in the H group, while the relative abundances of Alistipes senegalensis (r = 0.42, P < 0.05) and Bifidobacterium choerinum (r = 0.28, P < 0.05) showed positive correlations with butyrate in the L group (Fig. 7A). The significantly positive correlations between differential vOTUs and phenotypic traits were divergent between two groups, but most of the differential vOTU belonged to unassigned Caudovirales (P < 0.05, Fig. 7A, Additional file 1: Table S4. The differential abundant vOTUs between H and L calves). Particularly, four vOTUs (H: vOTU17 and vOTU23; L: vOTU58 and vOTU72) were predicted to be associated with Prevotella involved in carbohydrate degradation correlated positively with ADG in both H and L groups (P < 0.05, Fig. 7B); three vOTUs (vOTU64, vOTU77, and vOTU81), the host of which were bacteria associated with cellulose and hemicellulose degradation, showed positive correlations with ADG in the L group (P < 0.05, Fig. 7). In addition, two vOTUs (H: vOTU19; L: vOTU94) predicted to infect Limivicinus were positively correlated with ADG in the H and L groups, separately (P < 0.05, Fig. 7). Interestingly, vOTU76 predicted to infect Methanobrevibacter sp017652345 showed positive correlation with butyrate and ADG in the L group (P < 0.05, Fig. 7).

Fig. 7.

Relationships between fecal microbiomes and phenotypes between H and L calves. A Significant correlations between fecal microbiome and phenotypes between H and L calves, only statistically significant relationships were plotted (P < 0.05). B The virus-host dyads associated with host phenotypic traits between H and L calves (n = 8 per group)

Discussion

Although fecal microbiota have long been recognized as important components of the ruminant gastrointestinal tract, their potential relationships with calf growth performance remain largely enigmatic. Here, the ADG-associated bacterial and archaeal species and functional profiles were identified through comparison of the fecal microbiome in beef calves with divergent growth rates using metagenomic sequencing. In addition, the potential role of fecal viruses associated with the divergent phenotypes of young ruminants was outlined.

The gut microbial diversity is strongly linked to the host’s health and metabolism [54]. The bacterial Shannon index in the H group was significantly lower than that in the L group, consistent with previous findings in dairy goats and Hu sheep [55, 56], suggesting that lower alpha diversity of gut microbiota may serve as a novel indicator of animal productivity, as a simple microbial community can perform the core functions-producing specialized substrates to meet the host’s energy requirements without diverting resources to redundant microbial growth [57]. However, the bacterial Shannon index was higher in the feces of high ADG cattle compared to low ADG cattle, as reported previously [58]. Furthermore, no divergence of the Shannon index between H and L groups was observed in the hindgut of preweaning Holstein heifer calves and young goats [15, 59]. The inconsistent pattern of fecal microbiota diversity may be caused by the breed, diet, age, and living environment [60, 61]. Although one low-ADG animal displayed a microbial composition closer to the high-ADG group, this deviation reflects normal inter-individual variability and did not alter the overall patterns between the two groups.

In this study, the relative abundances of some beneficial microorganisms, including members of Alistipes (Alistipes shahii, Alistipes senegalensis, and Alistipes finegoldii), Butyricimonas virosa, Akkermansia glycaniphila, Flavonifractor plautii, and Bifidobacterium thermophilum were enriched in the H group. It was reported that Butyricimonas virosa is a butyrate-producing bacterium and can improve intestinal barrier functions by producing SCFAs [62, 63]. Akkermansia glycaniphila plays an important role in maintaining the gut barrier health and thereby regulating immunity [64]. Flavonifractor plautii possesses the ability to produce health-promoting SCFAs (i.e., propionate and butyrate) [65]. Alistipes have been reported to be beneficial to animals’ growth. For instance, its colonization in the cecum of broiler chicken promotes the growth through producing SCFAs [66]; it has also been reported to correlate positively with body weight (BW) and ADG in Hainan black goats [67], which supports the finding in this study where Alistipes correlated positively with ADG in the H group. Collectively, these beneficial microorganisms may correlate with greater growth rates in young ruminants by improving gut health and modulating fermentation profiles. Notably, Alistipes bacteria are involved in the formation of propionate and lipid metabolism [68]. In this study, the proportion of propionate was numerically higher in the H group and functional genes associated with lipid metabolism were enriched in the H group, which may be ascribed to the higher abundance of Alistipes sp. in the H group because microbial composition affects microbial metabolism [69].

Several probiotic species (Bifidobacterium breve, Bifidobacterium longum, Bifidobacterium choerinum, and Bifidobacterium adolescentis) were enriched in the L group relative to the H group. However, a recent study reported that members of Bifidobacterium were enriched in calves with better growth phenotype (ADG and dry matter intake) [70]. A prior study showed that Bifidobacterium spp. was enriched in the fecal microbiome of LADG calves [59]. Additionally, supplementation of Bifidobacterium spp. in pre-weaning Holstein calves had no further improvement in growth performance [71]. These apparently contradictory outcomes in Bifidobacterium’s effect on calf growth may be attributed to the differences in calf age, diet, and rearing conditions [72, 73]. Members of Bifidobacterium are associated with gut health by forming a biofilm to prevent pathogen colonization [74], which is beneficial to maintain healthy gut micro-ecosystems of calves in the L group. Furthermore, Butyrivibrio fibrisolvens, a major butyrate-producing species that plays a critical role in plant cell wall degradation [75], was enriched in the L group, resulting in a higher proportion of butyrate. Of note, Methanobrevibacter sp. AbM4, one of the major methanogens in ruminants, was enriched in the H group in the present study. It was reported that the prevalence of Methanobrevibacter sp. AbM4 was higher in inefficient animals compared to the efficient counterparts [76]. However, Methanobrevibacter sp. AbM4 increased upon inhibition of methane emissions by 3-NOP [77], and the relative abundance of Methanobrevibacter sp. correlated negatively with methane emissions [78]. Moreover, the relative abundance of Methanobrevibacter showed positive linkages with the growth performance of broilers, rabbits, and pigs [79–81]. Additionally, it was reported that walc (alcohol) and wald (aldehyde) dehydrogenase genes were identified in the genome of Methanobrevibacter sp. AbM4 [82] that make them possess the capacity for alcohol-fueled methanogenesis, which favors less hydrogen production during fermentation and potentially reduces the methane emission [83, 84]. Thus, it was speculated that Methanobrevibacter sp. AbM4 in the H group may employ different methanogenesis pathways compared to the L group, as the profiles of methanogens in the dairy calves varied among individual animals [85], making it potentially adapted to different hosts. The specific ecological roles of Methanobrevibacter sp. AbM4 in the hindgut of young ruminants deserve to be further studied, which is beneficial to improve host metabolism and health in early life.

The microbial function profiles are divergent between H and L groups, reflecting differences in microbial composition, as microbial function depends on their taxa [86]. In this study, the KEGG pathway associated with lipid metabolism was enriched in the H group, and the relative abundances of Alistipes shahii and Alistipes onderdonkii were greater in the H group. These two species are SCFA producers that are involved in regulating barrier function and energy homeostasis, which indirectly affects lipid metabolism [87]. Furthermore, they are associated with the production of sulfonolipids that are metabolites involved in lipid metabolism [88], contributing to the enrichment of lipid metabolism pathways. In addition, genes encoding CAZymes associated with cellulose and hemicellulose degradation were enriched in the L group, and the relative abundances of Butyrivibrio fibrisolvens and Ruminococcus bovis were greater in the L group. These two species encoded CAZymes associated with cellulose and hemicellulose degradation [89]. Thus, the higher abundance of cellulose and hemicellulose CAZymes in the L group could be expected. Although functional differences in microbial metabolism were identified, the specific mechanisms through which they affect host energy utilization remain to be fully elucidated.

As reported in other studies, the dominant viral families were assigned to Myoviridae and Siphoviridae in the healthy dairy calves and other ruminant species [90–92], which supports the finding in the current study, indicating that they may be core gut viruses in all ruminants. Among the predicted bacterial hosts, Bacillota_A, Bacteroidota, and Bacillota were the predominant phyla, which have been observed in the gut of humans, pigs, and ruminants [46, 93, 94]. Moreover, some vOTUs were predicted to infect Methanobrevibacter, which is consistent with the finding in other ruminants [46]. These various virus-host dyads imply the widespread viral infections of gut prokaryotes and the significance of viruses in affecting gut microbial ecosystems. Interestingly, ADG-associated vOTUs were identified in this study and the microbial hosts of them varied significantly between H and L groups. For example, viral microbial hosts associated with starch degradation were enriched in the H group, whereas those involved in cellulose and hemicellulose degradation, as well as methane emission, were enriched in the L group. Viruses can affect the structure and function of microbial communities upon infection by lysing bacterial cells and facilitating horizontal gene transfer [95], thereby influencing the host metabolism and phenotypes (growth, lean and obese body) [23, 95, 96]. Therefore, it was postulated that the divergent virus-host pairs between H and L groups might affect the gut health of animals, contributing to divergent nutrient metabolism and consequently impact the host growth performance. Further study with multiple ruminant species and larger animal size could be done to corroborate it.

It is worth mentioning that the abundance of one vOTU predicted to infect Methanobrevibacter sp017652345 was found to be positively correlated with butyrate and ADG. It was reported that Methanobrevibacter sp017652345 was depleted in patients with adenoma compared to their healthy counterparts [97]. However, the specific function of this species is not clearly defined in the gut of ruminants. Future studies with single-cell transcriptomics and metabolomics could be done to deeply investigate the functions of this microorganism.

This study was based on a relatively small cohort of indigenous beef calves, which may limit the generalizability of the findings to other ruminant species raised under different environments. In addition, only fecal samples were analyzed, which may not fully capture the microbial dynamics across the entire intestinal tract. Additionally, virus–host relationships inferred using bioinformatics tools such as IPhoP have not yet been experimentally validated. Although correlations between viral and prokaryotic abundances were used to support these predictions, such relationships do not provide evidence of causation. Future experimental validation and longitudinal studies with larger cohorts will be necessary to confirm the functional roles of specific viruses and their hosts in host metabolism and growth.

Conclusion

This study revealed the potential of fecal microbiome (bacteria, archaea, and viruses) associated with ADG and fecal fermentation profiles in beef calves of Guizhou using metagenomic sequencing. Overall, fecal microbiome characteristics, including composition, diversity, and function, varied between young beef calves with different growth rates. Some specific bacterial species, such as members of Alistipes and Bifidobacterium, were identified for their positive associations with fecal fermentation and calf growth rates. Notably, vOTU predicted to infect Methanobrevibacter sp017652345 correlated positively with ADG. Additionally, many vOTUs in the high ADG group were found to be associated with carbohydrate-degradation bacteria, suggesting that fecal viruses may be directly or indirectly involved in carbon metabolism in the gut of young ruminants. Therefore, our results advance the understanding of the influence of fecal microbes on the ADG of post-weaning beef calves and emphasize the potential of fecal viruses associated with greater growth rates of calves through interacting with their microbial hosts, which aids in developing strategies to improve animals' production performance.

Authors’ contributions

WG conceived and designed the study, performed the formal analysis and interpreted the results, and wrote the manuscript. JY performed data analysis and interpreted the results. WW helped to collect samples. JW helped to do data analysis. MN helped to collect samples. MZ interpreted the results and reviewed and revised the manuscript. XC conceived and designed this study.

Funding

This study was fund by the National Natural Science Foundation of China (No. 32402705) and Guizhou Provincial Science and Technology Projects (No. Qian Ke He Basic- [2024] Youth 104 and 107) and Wei Guo was supported by the China Scholarship Council (CSC). This study was partially funded by the Foundation of Key Laboratory of Animal Genetics, Breeding and Reproduction in The Plateau Mountainous Region, Ministry of Education, Guizhou University (GZUAGBR202302), Key Project of Guizhou Provincial Department of Science and Technology in the Agricultural Field: Exploration and Innovative Utilization of Local Yellow Cattle Germplasm Resources in Guizhou (Qiankehe Support [2022] Key 027), and Guizhou Provincial Science and Technology Program (Qian-Ke-He Basic Project – ZK[2023] General 086).

Data availability

Raw sequencing data were deposited in the NCBI Sequence Read Archive under SRA accession PRJNA1293833. The 668 HQ SGBs used in this study are available in Figshare under the DOI: https://doi.org/10.6084/m9.figshare.29606717.

Declarations

Ethics approval and consent to participate

The study was conducted in line with the protocol of the Animal Care and Use Committee of Guizhou University (EAE-GZU-2024-T228).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.