Maternal rumen and milk microbiota shape the establishment of early-life rumen microbiota in grazing yak calves

Early-life gut microbial colonization and development exert a profound impact on the health and metabolism of the host throughout the life span. The transmission of microbes from the mother to the offspring affects the succession and establishment of the early-life rumen microbiome in newborns, but the contributions of different maternal sites to the rumen microbial establishment remain unclear. In the present study, samples from different dam sites (namely, oral, rumen fluid, milk, and teat skin) and rumen fluid of yak calves were collected at 6 time points between d 7 and 180 post-partum to determine the contributions of the different maternal sites to the establishment of the bacterial and archaeal communities in the rumen during early life. Our analysis demonstrated that the dam’s microbial communities clustered according to the sites, and the calves’ rumen microbiota resembled that of the dam consistently regardless of fluctuations at d 7 and 14. The dam’s rumen microbiota was the major source of the calves’ rumen bacteria (7.9%) and archaea (49.7%) compared with the other sites, whereas the potential sources of the calf rumen microbiota from other sites varied according to the age. The contribution of dam’s rumen bacteria increased with age from 0.36% at d 7 to 14.8% at d 180, whereas the contribution of the milk microbiota showed the opposite trend, with its contribution reduced from 2.7% at d 7 to 0.2% at d 180. Maternal oral archaea were the main sources of the calves’ rumen archaea at d 14 (50.4%), but maternal rumen archaea became the main source gradually and reached 66.2% at d 180. These findings demonstrated the potential microbial transfer from the dam to the offspring that could influence the rumen microbiota colonization and establishment in yak calves raised under grazing regimens, providing the basis for future microbiota manipulation strategies during their early life.


INTRODUCTION
Colonization of gut microbiota during early life has a long-term impact on the host nutrient metabolism, immune system development and function, and the structure and function of the gastrointestinal tract in mammalian animals (Francino, 2014;Jandhyala et al., 2015).Compared with the numerous studies on monogastric animals, research on early-life gut microbiota in ruminants is relatively scarce, especially for grazing animals.Recent studies have revealed that rumen bacteria and archaea are present in 1-to 3-d-old goat kids (Guo et al., 2020) and dairy calves (Jami et al., 2013;Guzman et al., 2015;Malmuthuge et al., 2015), and that the predominant taxa and microbial diversity vary as the animals age.The early-life rumen microbiota of calves has also been reported to drive the development of the rumen, a key organ that determines ruminants' growth, health, and performance (Malmuthuge et al., 2019;Belanche et al., 2020).The rumen development in newborn Holstein bull calves can be enhanced by the rumen microbiome through the regulation and expression of microRNA and genes that are negatively associated with the volatile fatty acids, important energy metabolites for the host (Malmuthuge et al., 2019).In a similar way, the ruminal microbiome can stimulate ruminal epithelium development in young lambs (Lin et al., 2019).Thus, understanding the acquisition mechanisms of rumen microbes in neonates from various sources is vital for the development of manipulation strategies capable of regulating such processes and enhancing the rumen development and its function throughout the life span.
The factors shaping the succession and establishment of gut microbiota in early life have been widely studied, including the prenatal period, type of birth (cesarean section vs. vaginal delivery), feeding mode (Wang et al., 2016), and host genetics (Fan et al., 2020).The environment has also been reported to affect the rumen microbial communities in early life (Fonty et al., 1989;O'Hara et al., 2020).Recently, social interactions with the mother have been proposed to play an important role in affecting the colonization and establishment of the early-life gut microbiota (Chen et al., 2020).In humans, maternal skin and vaginal microbiota colonize the infant gut transiently during the first 4 mo of life (Ferretti et al., 2018).In ruminants, the contribution of maternal microbiota from different body sites (vagina, colostrum, teat skin, feces, and oral) has been studied for early-life fecal microbiota composition of domesticated ruminants (goat, sheep, and dairy cow) raised under intensive systems (Bi et al., 2019;Klein-Jöbstl et al., 2019;Guo et al., 2020).However, the knowledge of maternal microbial sources that contribute to the rumen microbiota of yak calves raised under grazing regimens (an open and not well-controlled animal rearing system) is still limited.
Yaks (Bos grunniens) possess many anatomical and physiological adaptations that enable them to survive in the harsh environment of the Tibetan plateau and to graze on the natural pasture all year without supplementary feed (Qiu et al., 2012).Therefore, the current knowledge of the early-life rumen microbiota in domesticated ruminants may not be applicable to yaks.We hypothesized that the dam microbiota is one of the major sources that affects the development of the early-life rumen microbiota in grazing yak calves because they are raised together with their dams and depend on dam's milk for growth.Therefore, this study aimed to examine to what extent the dam microbiota contributed to the establishment of the rumen bacteria and archaea of the calf.To achieve this goal, rumen fluid samples from yak calves (grazing with their dams) were collected at 6 time points between d 7 and 180 after birth, and oral, teat skin, udder milk, and rumen fluid samples were collected from dams concomitantly to investigate how the microbes from the various sites contributed to the establishment of the rumen microbiota of grazing yak calves.

Animal Experiment and Sample Collection
All experimental procedures were approved by the Animal Ethics Committee of the Chinese Academy of Lanzhou University (permit number SCXK Gan 20140215).A herd of pregnant yak cows (n = 30) that grazed together year-round in an alpine meadow on the Qinghai-Tibetan Plateau was enrolled in this study.From them, 7 lactating yak cows (5 to 6 years old; body weight 230 ± 16 kg; 3 to 4 parities) that calved male calves (15 ± 3 kg) between April 7 and 10, 2017, were selected for the study.The yak calves ran with their mothers and sucked milk until they were weaned naturally (till dam became pregnant).The overall sample collection scheme is shown in Figure 1.The yaks were corralled the night before sampling and the calves were separated from their dams immediately after corralling.The rumen fluid samples were collected from the lactating yak cows (referred to as dams throughout this paper) via mouth using a stainless-steel stomach tube connected to a rumen vacuum sampler before morning grazing, and from their calves at d 7, 14, 60, 90, 120, and 180 of age via mouth using a plastic flexible stomach tube attached with a vacuum sampler as described previously (Guo et al., 2020).The samples from the oral cavity, teat skin, and udder milk (referred to as oral, skin, and milk, respectively, hereafter) of the dams were collected concomitantly with rumen sampling of calves using methods of sample collection described by Ferretti et al. (2018).In brief, sterile swabs were used to collect samples from the upper area of the maternal teat (skin) and from the mouth (oral) and were placed into sterile plastic tubes.For the milk sample, the teat was cleaned with alcohol (75%) and then 13 mL of milk was collected, of which the first 3 mL was discarded.All samples were snap-frozen in liquid nitrogen immediately after collection and then stored at −80°C until DNA extraction.

DNA Extraction
Genomic DNA was extracted using the PowerSoil DNA Isolation Kit (MoBio Laboratories) following the manufacturer's guidelines.The DNA extracted from frozen swabs followed the steps reported by Ferretti et al. (2018).Briefly, the head of each swab was removed and placed in a PowerBead tube with specimen-containing buffer, and then centrifuged at 1,000 g for 5 min for cell lysing.The DNA quantity, purity, and integrity were determined using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific) and 1% agarose gel electrophoresis.

Amplicon Sequencing
To determine bacterial and archaeal profiles, the primer pairs Ba9f (GAGTTTGATCMTGGCTCAG) and Ba515Rmod1 (CCGCGGCKGCTGGCAC) targeting the bacterial partial 16S rRNA gene and the primer pairs Ar915aF (AGGAATTGGCGGGGGAGCAC) and Ar1386R (GCGGTGTGTGCAAGGAGC) targeting the archaeal partial 16S rRNA gene (Henderson et al., 2015) were used to produce amplicons for each sample.The amplicon libraries were then sequenced on an Illumina PE MiSeq 300 platform to generate 2 × 300 bp paired-end reads at Majorbio Company.

Sequence Data Analysis
The raw sequences were analyzed using QIIME2 2020.2 (Bolyen et al., 2019).First, the paired-end sequences were demultiplexed and quality checked using DADA2 (Callahan et al., 2016), and then the qualified sequences were clustered into amplicon sequence variants (ASV).Taxonomy was assigned to the ASV against the SILVA 132 database (Quast et al., 2013) for bacteria and archaea with the "qiime feature-classifier" command using the "classify-sklearn" option (Hakim et al., 2019).For α and β diversity analyses, sequences were aligned with Mafft (Katoh and Standley, 2013) and then the noninformative positions in the alignment were removed via "mask" command before the construction of the phylogenetic tree using the Fast-Tree plugin (Price et al., 2010).Last, clean reads were rarefied to lowest number of sequences within the data set (6,797 bacterial and 4,160 archaeal sequences) to avoid biases resulting from uneven sequencing depth before calculating the α (Shannon and Chao1) and β diversities (Bray-Curtis).

Determination of Sources of Calf Rumen Microbiota
SourceTracker2, a Bayesian approach that predicts the proportion of contaminants in a given community, was employed to determine the origin of microbes from the source environments (Knights et al., 2011) and assess the contributions of different maternal microbial sources to the yak calves' rumen microbiota after birth as described previously with the use of the sourcetrack-er2 (Guo et al., 2020).Samples collected from different maternal body sites were regarded as sources, and rumen fluid samples of calves were treated as sinks.This analysis was performed with default parameters within the SourceTracker2 software.

Statistical Analyses
Statistical differences in α diversity index among different sites within each age group were determined using the nonparametric Kruskal-Wallis test and Pvalues were adjusted using the Benjamini-Hochberg correction for Dunn's multiple comparisons (Ferreira and Zwinderman, 2006).The adjusted P-value <0.05 was considered significant.Principal coordinates analysis based on Bray-Curtis distance was performed using the qiime2R package to compare the profiles of the microbial community among different sites and between yaks and calves (Guo et al., 2020).The pairwise PER-MANOVA tested was used to test significance among sites with default permutations in QIIME2 (Bolyen et al., 2019).A taxon that appeared in at least 50% of the animals in one site within one age group, with a relative abundance >1% at the phylum and genus levels, was considered for further analysis.All statistical analyses were conducted in RStudio (v 3.5.3),unless otherwise noted.

Data Deposition
Sequencing data were deposited at the NCBI Sequence Read Archive under accession number PRJNA838115.

Microbial Diversity of Bacterial Community Detected in Dams and Calves
Amplicon sequences were obtained from a total of 210 samples (168 from dams and 42 from calves).Forty-five samples for bacteria and 99 samples for archaea were removed due to the low quality of extracted DNA or no amplicon being obtained.The detailed sample numbers for each age group within each sample type are listed in Table 1.
Bacteria were amplified successfully in a total of 165 samples from dams (rumen fluid: n = 33; milk: n = 31; oral: n = 30, teat skin: n = 37) and calves (rumen fluid: n = 34) between d 7 and 180 across 6 time points (Table 1).After quality control, a total of 3,191,878 high-quality reads were obtained for bacteria with an average of 19,345 ± 638 (mean ± SEM) per sample, which yielded a total of 43,871 ASV (674 ± 21 per sample).Good's coverage for each sample was higher than 96%, indicating that the sequencing depth was sufficient to characterize the microbial community in each site.The Shannon index of rumen bacteria in calves was lower than that of dams on d 7 and 14 (P < 0.05; Figure 2A), whereas the Chao1 index of rumen bacteria of calves was lower than that of milk and dams' rumen bacteria on d 7, and of oral bacteria on d 14 (P < 0.05; Figure 2A).This is consistent with studies on dairy cows and goats, where the milk bacteria diversity was much higher than those of calf rumen bacteria (Yeoman et al., 2018) and the fecal bacteria of goat kids (Guo et al., 2020).However, these studies did not compare the rumen microbiota of newborns to oral microbiota.Oral microbiota is an important source for the infant gut microbiota colonization (Ferretti et al., 2018), although the influence was short-lived (2 mo).Our results revealed that the effects of milk and the oral bacteria diversity of dams on the rumen bacterial diversity of yak calves in early life can be long term (6 mo).
The comparison of β diversity of microbial communities with principal coordinates analysis and PER-MANOVA based on the Bray-Curtis distance matrix revealed that bacteria of lactating yak cows were clustered according to the sites, and the rumen microbiota of the calves resembled the maternal rumen microbial community consistently, although there was a little fluctuation in the earlier days (e.g., d 7 and 14) (Figure 3A; Supplemental Table S1, https: / / doi .org/ 10 .6084/m9 .figshare.21077284;Guo, 2022).The rumen bacterial profiles of the calves clustered according to dam rumen bacteria, but they were more related to milk bacterial profiles than other sites on d 7 (P < 0.05; Figure 3A; Supplemental Table S2, https: / / doi .org/ 10 .6084/m9 .figshare.21077284;Guo, 2022).This observation is in line with the data presented by Guo et al. (2020), who showed the rumen bacterial community of goat kids in the early life clustered with that of the milk, and Klein-Jöbstl et al. (2019), who observed a 20% sharing in microbial taxa between colostrum and fecal samples of dairy calves during the first 12 h after birth.Together, these suggest that milk bacteria are likely one of the inoculation sources for neonatal yak calf rumen because they were raised with the dams and fed mainly with dam's milk.As calves aged, the rumen bacterial community gradually clustered with the rumen bacteria of the dam (Figure 3A), and from d 90 it was more similar to the rumen bacteria of the dam than to any other site (Supplemental Table S2).This is similar to the findings for the rumen bacteria in dairy calves (Jami et al., 2058(Jami et al., 2013)), buffalo calves (Koringa et al., 2019), and goat kids (Wang et al., 2016), where the similarity of rumen bacteria between calves and adult animals increased with age.In addition, the rumen gradually developed with age and performs similar functions to that of adult postweaning (Arshad et al., 2021), which contributes to the high similarity in rumen microbiota between calves and dams.

Microbial Diversity of Archaeal Community Detected in Dams and Calves
Archaea were amplified successfully in a total of 111 samples from dams (rumen fluid: n = 28; oral: n = 27; teat skin: n = 27) and calves (rumen fluid: n = 29) between d 14 and 180 across 5 time points (Table 1), though they were not detected in the milk. 2 Frequency = the average number of high-quality reads per sample after quality control using the DADA2 algorithm.
3 A total of 1,008,207 high-quality reads were obtained for archaea with an average of 9,083 ± 268 per sample, which generated a total of 468 ASV (58 ± 3 per sample).The Shannon index of rumen archaea in calves was lower than that of skin archaea on d 14 and 60 (P < 0.05; Figure 2B), whereas the Chao1 index of rumen archaea in calves was lower than that of skin archaea at all tested ages (P < 0.05; Figure 2B).Previous studies have demonstrated that early rumen archaea colonization in dairy calves occurred at the first week after birth (Dias et al., 2017;Malmuthuge et al., 2019;Meale et al., 2021).A recent study has revealed that rumen archaea can be found in the rumen of dairy calves as early as 20 min postpartum (Guzman et al., 2015).These findings suggest that the initial seeding of archaea in the rumen of natural grazing yaks differs from that of domesticated dairy cattle, which was only detected after d 14 of age.The detection of oral archaea also adds to the recent findings reported by other researchers who showed that archaea were commensal in the oral samples of Ayrshire cows (Tapio et al., 2016) and sheep (Kittelmann et al., 2015), indicating that archaea are the essential inhabitants in the oral cavity of ruminants.Archaea were found on the teat skin in this study, which is in line with the study on dairy cows (Ekman et al., 2020), in which archaea were found on the skin at the fore udder.In addition, archaea have been detected on human skin as well as on ruminants (bovine, sheep, and goat) and skin of different members of the class Mammalia (Ross et al., 2018), suggesting archaea may be commensal organisms on the skin of different animal species.However, the human skin archaea are distinct from those of nonhuman mammals (Ross et al., 2018), indicating that host species or living environment could affect the diversity and composition of the skin archaea.
The rumen archaeal community of calves clustered according to the dam rumen archaea and oral archaea after d 90 (P < 0.05; Figure 3B; Supplemental Table S2), which is supported by a previous study on yak rumen microbiota, where the rumen archaea community clustered with adult animals after 2 mo of age of the calves (Guo et al., 2020).As calves aged, they gradually changed their diet from milk to the natural grass that their dams consumed in this study, which contributes to the increased similarity in the rumen archaea with the age between calves and dam, as diet affects the archaeal community (Furman et al., 2020).In addition, rumen archaeal composition and function varied with age, and its community structure reached levels similar to those in adult animals after 2 mo (Friedman et al., 2017).Saliva archaea are considered proxies for profiling the rumen microbial composition (Tapio et al., 2016).We reasoned that archaea from rumen and saliva are potentially important sources for the colonization and development of yak calf rumen archaea.

Bacterial Composition in Dam and Calves
A total of 36 bacterial phyla and 116 bacterial genera were identified according to the threshold mentioned in the materials and methods section.Among the detected taxa, 7 phyla were detected in the milk, with Proteobacteria (45.4% ± 0.02), Firmicutes (20.1% ± 0.01), Actinobacteria (17.9% ± 0.02), and Bacteroidetes (10.7% ± 0.01) being the predominant phyla (Figure 4A; Supplemental Table S3, https: / / doi .org/ 10 .6084/m9 .figshare.21077284;Guo, 2022).A total of 43 genera were treated as detectable, and Acinetobacter (8.8% ± 0.02), Corynebacterium 1 (4.7% ± 0.01), and Moraxella (4.3% ± 0.01) were found to be prevalent in the milk (Supplemental Data Set S1, https: / / doi .org/ 10 .6084/m9 .figshare.21077284;Guo, 2022).Proteobacteria are the dominant taxa in the milk among different ruminant species including dairy cows (Yeoman et al., 2018;Zhu et al., 2021), sheep (Bi et al., 2019), water deer, and reindeer (Li et al., 2017), suggesting that these microbes are the core taxa in the milk produced by the different ruminant species.However, this result was not in agreement with those of goat and sheep milk, where Firmicutes was the most dominant phylum (Li et al., 2017;Biçer et al., 2021).The difference in milk microbial composition may be attributed to the host (ruminant species, stage of lactation, and health conditions) and environmental factors (e.g., farm environment, management, and diet) (Oikonomou et al., 2020).By consuming oxygen, Proteobacteria modifies the rumen environment to create suitable anaerobic conditions that are favorable to the growth of strict anaerobes (Rey et al., 2014).Therefore, the prevalence of Proteobacteria in the milk may facilitate the colonization of the rumen microbiota in yak calves.
Acinetobacter was a prevalent bacterial taxon in the milk of yaks in the current study, which is in line with findings for water buffalo (Ercolini et al., 2012) and water deer (Li et al., 2017).However, this observation was not consistent with previous reports that researched the microbiome of cow and sheep milk, which contained a greater abundance of Lactococcus spp.(Quigley et al., 2013a,b).Acinetobacter flourishes during cold storage (4°C, Raats et al., 2011), and the milk samples in this study were snap-frozen in liquid nitrogen, which partially explains this discrepancy.Other studies showed that Corynebacterium dominated in the milk of water deer (Li et al., 2017), bovine (Oikonomou et al., 2020), and caprine hosts (Toquet et al., 2021).However, Staphylococcus and Ralstonia have been demonstrated to be the dominant microorganism in the milk of goats (Zhang et al., 2017;Polveiro et al., 2020) and dairy cows (Addis et al., 2016), respectively.Previous studies revealed that milk microbiota of dairy cows was affected by diet (70% concentrate diet vs. 40% concentrate diet) (Zhang et al., 2015), breed (Esteban-Blanco et al., 2020), host species (Li et al., 2017;Biçer et al., 2021), weather conditions (temperature and humanity) (Li et al., 2018), and management practices (Doyle et al., 2017).The differences in milk microbial composition in this study compared with the aforementioned research could be attributed to the host (breed, species, and health conditions) and environmental factors (temperature, humidity, and diet).Furthermore, some nonpathogenic Corynebacterium species are beneficial in the processing of dairy products, such as amino acid production and flavor improvement of cheese (Hahne et al., 2018).Thus, the prevalence of Corynebacterium in the yak milk may contribute to its higher nutritious values (protein, polyunsaturated fatty acids, and fat) than that observed in cow milk (Nikkhah, 2011).
Nine and 8 phyla were detectable (appeared in at least 50% of the animals in one site within one age group and relative abundance >1%) in the rumen of calves and dams based on the selected criteria, respectively.Among them, Bacteroidetes (calf vs. dam: 53.9% vs. 49.6%)and Firmicutes (33.1% vs. 38%) were equally dominant in the rumen of calves and dams (Figure 4A; Supplemental Table S3).A total of 45 and 38 genera were referred to as detectable in the rumen of calves and dams, respectively.Of these, Prevotella 1 (10.5% ± 0.02), Prevotellaceae UCG-003 (8.4% ± 0.03), and Rikenellaceae RC9 gut group (8.1% ± 0.01) were dominant in the rumen of calves, whereas Rikenellaceae RC9 gut group (7.8% ± 0.03), Christensenellaceae R-7 group (6.6% ± 0.003), and Ruminococcaceae NK4A214 group (5.9% ± 0.005) were dominant in the rumen of dams (Supplementary Data Set 1).In addition, the relative abundances of Prevotella 1 and Rikenellaceae RC9 gut group in the rumen increased with the age of the calves, although there were fluctuations at d 90 or 120 (P < 0.05; Supplementary Data Set 1).It was reported that the relative abundance of Prevotella 1 correlated positively with the rumen weight and total VFA of goat kids (Jiao et al., 2015), and that of Rikenellaceae RC9 gut group correlated positively with VFA production (Wei et al., 2022).The VFA produced by rumen microbes could stimulate the growth of rumen mucosa and development of rumen epithelium (Connor et al., 2013).Therefore, we speculated that the changes in the relative abundance of these 2 genera may have contributed to the rumen development of yak calves, although the rumen morphology and structure were not examined in this study.Previous studies found that Rikenellaceae RC9 gut group was prevalent in the rumen of yak (Xue et al., 2018;Hu et al., 2019;Ren et al., 2020) and other ruminants (Zhou et al., 2015).In addition, Rikenellaceae RC9 gut group played an important role in the digestion of crude fiber (Zhang, 2017).Therefore, these findings suggest that Rikenellaceae RC9 gut group is one of the core microbial taxa that perform basic metabolic functions in yak rumen.

Archaeal Composition in Dam and Calves
Taxonomic analysis revealed that 2 archaeal phyla and 11 archaeal genera were detected according to the selected criteria.Among them, Euryarchaeota (76.8% ± 0.03) and Thaumarchaeota (23.1% ± 0.03) were the dominant phyla in the teat skin (Figure 4B; Supplemental Table S3).Eight genera were regarded as detectable according to the criteria mentioned earlier, and Methanobrevibacter (61.7% ± 0.03), Candidatus Nitrocosmicus (12.5% ± 0.02), and f.Nitrososphaeraceae (9.8% ± 0.02) dominated in the teat skin (Supplemental Data Set S2; https: / / doi .org/ 10 .6084/m9 .figshare.21077284;Guo, 2022).Traditionally, it is believed that methanogen colonization in the skin appears to be transient or results from contamination by the external environment or the feces, as these microbes are strictly anaerobes (Ekman et al., 2020).However, several stud- ies support the hypothesis that the presence of archaea on the skin is not the mere result of an external colonization.It has been reported that Methanobrevibacter and Methanosphaera were the dominant taxa on the skin at the fore udder of dairy cows (Ekman et al., 2020) and nonhuman mammals (Umbach et al., 2021).In addition, archaea associated with human skin have the metabolic potential to perform ammonia oxidation (Probst et al., 2013) and affect the host health (Ekman et al., 2020).Therefore, we speculated that archaea are the true members of the bovine skin microbiota herein and potentially play important roles in functions such as regulating the skin pH, although this area warrants further investigation.
Euryarchaeota (96.7% ± 0.01) was the most dominant phylum in the oral samples (Figure 4B; Supplemental Table S3).Among the 6 detected genera, Methanobrevibacter (81.9% ± 0.02), Methanosphaera (9.5% ± 0.01), and f.Methanobacteriaceae (2.2% ± 0.002) were the predominant archaeal taxa (Supplemental Data Set S2).It has been demonstrated that Methanobrevibacter dominated in the oral samples of Ayrshire cows (Tapio et al., 2016), sheep (Kittelmann et al., 2015), and humans (Zhang et al., 2018), suggesting Methanobrevibacter is one of the core oral archaeal taxa in different animal species.In addition, oral microbiota is related to the host health conditions (Lamont et al., 2018), and methanogenic archaea in the oral cavity with periodontal pathogens potentially act as terminal degraders of host components, supporting a continuous catabolic cascade (Matarazzo et al., 2012).Future studies focusing on archaeal characterization are necessary for better comprehension of the composition and functions of this domain of life in different ruminant species.

Potential Maternal Sources for Rumen Bacteria in Calves
Bacteria from multiple maternal body sites contribute to the development of the rumen bacteria of calves, with maternal rumen bacteria proving the largest contribution (P < 0.05; Figure 5A).This was further supported by the higher proportion of taxa shared between the calf and the rumen of the dam than with other sites (Supplemental Figure S2A, https: / / doi .org/ 10 .6084/m9 .figshare.21077284;Guo, 2022).The contribution of different sites to the rumen bacteria of calves varied with age (Figure 5C).For example, the contributions of milk (2.7%) and skin (5.3%) to the rumen bacteria of the calves were numerically higher than for other sites (oral: 0.6%, dam: 0.2%) at d 7, and then decreased with the age (Figure 5C).These results indicate that the early bacterial colonization in the rumen is affected by the environment of site or source (Perez Perez et al., 2016).
The proportion of rumen bacteria of calves that originated from milk bacteria was much lower than those of dam and oral samples (P < 0.05; Figure 5A), and it decreased with age (from 2.7% to 0.2%; Figure 5C).Milk from the dam not only is an important source of gut microbiota in newborn ruminants (Yeoman et al., 2018;Bi et al., 2019;Guo et al., 2020) but also contributes to gut microbial colonization and development of the immune system (Daft et al., 2015;Pannaraj et al., 2017).It has been reported that a high proportion of rumen microbiota originates from the colostrum in the first 21 d of life in dairy calves (Yeoman et al., 2018), and from the fecal microbiota in beef calves at 4 wk of age (Barden et al., 2020).These results are in contrast with the findings of this study, where the contribution of milk to the rumen microbiota of yak calves was only 0.7%.This is partially due to the low bacterial diversity (Shannon index) in milk in the early developmental stage compared with other maternal bacterial sources, such as the dam's rumen and oral cavity (Bi et al., 2019).In addition, the experimental period of d 7 to 180 in this study was longer than in the previous studies, where the contribution of colostrum microbiota to offspring gut microbiota decreased with age (Drell et al., 2017;Guo et al., 2020).Moreover, the present study started at 7 d of age, whereas the previous studies started immediately after birth when newborns were fed mainly milk.Consequently, the low contribution of milk to yak calves' rumen bacteria in the present study is not surprising.The contribution of teat skin bacteria to the rumen bacteria of calves was less than that of other sites (1.4%, P < 0.05; Figure 5A) and it decreased with age (Figure 5C), supporting the evidence from a study on cattle and yaks in which the udder skin microbiota of suckling dams had low similarity to the calves' fecal microbiota (Zhang et al., 2022).Human infant gut microbiota also displayed a low similarity to the mammary areola microbiota in the first 6 mo of life (Drell et al., 2017).These results indicate that the udder skin microbiota contributes very little to the establishment of the whole gut microbiota of the infant.However, the dams' udder skin bacteria shared the largest similarity (23.9%) with the rumen bacteria of calves in the first 21 d of life (Yeoman et al., 2018), and approximately 50% of the fecal microbiota in 3-d-old sucking lambs originated from the teat skin of the dams (Bi et al., 2019).An obvious difference among the studies was that in the present study, the dams and calves were grazing, whereas in the other studies the dams and offspring were confined.The free-grazing calves had more contact with a more diverse microbiota present in the environment, likely reducing the chance of the udder skin microbiota surviving in the rumen of the calves.However, more research is needed on dams acquiring microbiota from grazing environments, as well as the transfer from the dam to the calf.
The oral bacteria of dams in the present study were an important inoculation source of rumen bacteria in the calves (Figure 5A), and its contribution increased with age (Figure 5C).It has been reported that the fecal microbiota in neonatal beef calves was closely related to the oral microbiota of adult cows (Barden et al., 2020).Because the bovine oral and fecal microbiota reflects the rumen microbiota (Alipour et al., 2018;Uchiyama et al., 2020), it is not surprising that the calves' rumen microbiota could be partially originated from the dams' oral microbiota.In addition, the percentage of shared species between the oral and gut of infants was very high (>20%) in a human study (Ferretti et al., 2018).These results suggest that continuous seeding of the microbiota in the infant gut via oral cavity is an important route for gut microbiota transfer and colonization in the early life.However, as the oral samples of newborns were not collected in the current study, future studies should take oral microbiota of neonatal calves into consideration to explore how the oral-gut axis in grazing yak calves affects the colonization of the rumen microbiota.
The contributions of the dams' rumen bacteria (0.2% to 14.8%) to the rumen bacteria of calves increased with age, with the dam providing the greatest contribution after 14 d of age when compared with other sites (P < 0.05; Figure 5C).This finding is similar to that of the human research indicating that mothers' gut microbiota contributed mostly to the establishment of the gut microbiota of the infants (Ferretti et al., 2018).The similarities between the maternal and infant gut microbiome increased with age (Bäckhed et al., 2015;Korpela et al., 2018).The rumination of dams may contribute to the process of transfer from the rumen microbiota of dams to the rumen of calves through licking in the early life.On the other hand, the rumen microbiota of calves develops into a structure similar to that in dams postbirth that meets the functional requirements of the rumen, which plays an important role in shaping this transferring process.

Potential Maternal Sources for Rumen Archaea in Calves
The rumen archaea of the dam proved to be an important source of archaea to the calves' rumen, contributing approximately 48.4% of the rumen archaea, and its contribution was higher than that of other sites (oral and skin, P < 0.05; Figure 5B).The contribution of different sites to the rumen archaea of calves varied with age (Figure 5C).For example, the contribution of dam rumen archaea increased with age (30.2% to 66.2%, P < 0.05; Figure 5C), whereas that of oral archaea decreased with age (50.4% to 8.8%, P < 0.05; Figure 5C).The archaea from the skin contributed less to the rumen archaea of calves than other sites (P < 0.05; Figure 5B) but accounted for the highest proportion of shared taxa with the calf (Supplemental Figure S2B), and its contribution decreased with age (13.1% to 1.4%, P < 0.05; Figure 5C).Skin microbiota can be transferred to the calves while they are sucking.The calves gradually increase the ingestion of solid feed (natural grass) with age and reduce the frequency of sucking, thus lowering the contribution of the skin microbiota to the rumen archaeal population of the calf.In addition, the archaea microbes are strict anaerobes and cannot remain viable in the environment for a long time (Yeoman et al., 2018).
Rumen methanogens play an essential role in the electron sinks that are a driving force for the complex metabolism of the rumen microbiome (Friedman et al., 2017;Mizrahi et al., 2021).In addition, methanogenesis plays a major role pertaining to the animal's energy-harvesting capability from the consumption and its effect on the environment (Friedman et al., 2017).Given the importance of the rumen methanogens and the enormous significance of early-life gut microbiome manipulation for long-term host health and growth (Malmuthuge and Guan, 2017), future studies focused on comprehensively and deeply exploring the source of rumen archaea in the neonates and their functions in the early life are needed.
A limitation of the current study was that the maternal vaginal microbiota was not determined, and this site was reported to be an important source for the initial colonization of the gut microbiota in newborns, at least in human infants (Matsumiya et al., 2002).Vaginal samples are very difficult to collect from the yaks immediately after calving, as they calve in the remote mountains of the Qinghai-Tibetan Plateau.However, perhaps collecting vaginal samples as soon as possible after calving might indicate the potential of microbiota transfer from the vagina to the calf.This practice is being planned for future studies.

CONCLUSIONS
This is the first study that investigated the potential maternal sources of microbiota to the rumen microbiota of yak calves from d 7 to 180 after birth.The microbial communities of dams clustered according to the sampling sites, which indicated that site-specific taxa existed at each sample site.The rumen bacteria of calves were derived mainly from the milk at d 7, suggesting the importance of milk bacteria to the colonization of rumen bacteria in calves during early life.Although the potential sources of rumen bacteria and archaea varied with the age of calves among sample sites, the rumen microbiota of calves tended to be composed mainly of microbiota from the rumen of dams.Results in the current study provided basic information on the transfer of bacteria and archaea from dam to calf under grazing conditions.These results might differ from those for animals raised in intensive or confined conditions; however, these latter animals could be used to examine microbiota development in the newborn under more controlled conditions.

Figure 1 .
Figure 1.Flowchart describing the sample collection and approach used to profile the microbial community.Samples were taken from the rumen fluid, teat skin, oral cavity, and milk of the dams and from the rumen fluid of the calves from 7 dam-calf pairs.Sampling started at d 7 of age of the calf after birth and continued for up to 180 d.The yak calves ran with their mothers and sucked milk during the trial.

Figure 2 .
Figure 2. Alpha diversity of bacteria (A) and archaea (B) among different sites with age.Nonparametric Kruskal-Wallis and Duncan's tests were used to test for differences among different sites within each age group.A P-value <0.05 was accepted as significant.Means with different lowercase letters within an age group differ from each other (P < 0.05).The box plots show the median (dark line) and upper and lower quartiles.The whiskers display the minimum and maximum values, and the dots represent the outliers.
Figure 4. Compositional profile of bacteria and archaea at the phylum level.(A) Bacterial phylum composition at different sites across ages.(B) Archaeal phylum composition at different sites across ages.

Figure 5 .
Figure 5. Source Tracker depicting the proportion of rumen bacteria (A) and archaea (B) in yak calves originating from different maternal sites (milk, dam, oral cavity, and skin).(C) Proportion of rumen bacteria and archaea in yak calves originating from different maternal sites (milk, dam, oral, and skin) and for different age groups.Nonparametric Kruskal-Wallis and Duncan's tests were used to test for differences among sites within each age group.**P < 0.01; ***P < 0.001; ****P < 0.0001.The box shows the interquartile range below the first quartile and above the third quartile, and the line inside the box indicates the median.The whiskers indicate the minimum and maximum values, and the dots indicate outliers.
Guo et al.: MATERNAL SOURCES OF RUMEN MICROBIOTA IN YAK CALVES

Table 1 .
Guo et al.: MATERNAL SOURCES OF RUMEN MICROBIOTA IN YAK CALVES Sample information and sequencing statistics obtained using the DADA2 algorithm in QIIME2 1 Number of sequences = the average number of raw sequences per sample generated by Illumina sequencing.
Guo et al.: MATERNAL SOURCES OF RUMEN MICROBIOTA IN YAK CALVES