Pregnancy loss and calf death can cost dairy producers more than $230 million annually. While methods involving nutrition, climate, and health management to mitigate pregnancy loss and calf death have been developed, one potential influence that has not been well examined is the reproductive microbiome. I hypothesized that the microbiome of the reproductive tract would influence heifer fertility and calf gut microbial development. The objectives of this dissertation were: 1) to examine differences in phenotypes related to reproductive physiology in virgin Holstein heifers based on outcome of first insemination, 2) to characterize the uterine microbiome of virgin Holstein heifers before insemination and examine associations between uterine microbial composition and fertility related phenotypes, insemination outcome, and season of breeding, and 3) to characterize the various maternal and calf fecal microbiomes and predicted metagenomes during peri-partum and post-partum periods and examine the influence of the maternal microbiome on calf gut development during the pre-weaning phase.

In the first experiment, virgin Holstein heifers (n = 52) were enrolled over 12 periods, on period per month. On -3 d before insemination, heifers were weighed and the uterus was flushed. Flush pH and glucose concentration were measured. Blood was collected from coccygeal vessels on d -3, 15, 18, 21, 24, 27, and 30 relative to insemination and serum progesterone concentration was measured. Ultrasound measurements of dominant follicle diameter and corpus luteum volume. Insemination outcome was determined on d 30 using ultrasound and pregnancy was checked on d 42, 56, 70, and 84. Heifers were clustered based on outcome of insemination at d 30 (not pregnant, NP30, n = 24; pregnant, PS30, n = 28), d 84 (not pregnant, NP84, n = 24; pregnant but lost before d 84, PL84, n = 2; successfully pregnant through d 84, PS84, n = 26). Differences in phenotypes were assessed based on insemination outcome at d 30 and d 84. Weight was greater in NP30 heifer than PS30 heifers. Progesterone was greater in PS30 and PS84 heifers than NP30 or NP84 heifers on d -3 and 18 to 30 and CL volume was greater in PS30 and PS84 heifers than NP30 and NP84 heifers on d 21 and 30. To summarize, traits related to pregnancy maintenance were different in virgin Holstein heifers based on first insemination outcomes and might be able to be used to predict heifer reproductive performance.

Uterine flushes were examined in a subset of heifers (n = 28) based on insemination outcome and period. This subset was also clustered based on season (spring, n = 3; summer, n = 12; fall, n = 8; winter, n = 5). From this subset of heifers, DNA was extracted from uterine flush and 16S amplicons of the V4 region underwent 250 paired-end sequencing via Illumina NovaSeq 6000. Filtered reads were clustered into operational taxonomic units using a 97% similarity and assigned taxonomy using the SSURNA Silva reference version 132. Alpha and beta diversity were measured and differences in alpha and beta diversity measurements were analyzed based on insemination outcome at d 30 or d 84 and season of breeding. Differential abundance analyses were performed at the phylum and genus taxonomic ranks based on insemination outcome at d 30 or d 84 and season of breeding. Bacterial richness was reduced in PL84 heifers than NP84 and PS84 heifers and reduced in heifers bred in spring than those bred in other seasons. Bacterial community structure was different based on insemination outcome at d 30 and d 84 using unweighted Unifrac distances and was different based on season of breeding using weighted Unifrac distances. We observed an increase of Bacteroidetes in PS30 and PS84 heifers compared to NP30 and NP84 heifers. Ureaplasma and Ruminococcus had an increased abundance in PS30 and PS84 heifers than NP30 and NP84 heifers, while Afipia and Gardnerella had an increased abundance in NP30 and NP84 heifers than PS30 and PS84 heifers. Prevotella and Ruminococcus had a reduced abundance in summer bred heifers than winter bred heifers. Proteobacteria had a moderate negative correlation with -3 d progesterone (rp = -0.42) and Actinobacteria had a moderate negative correlation with fetal growth rate (rp = -0.66). Uterine microbiome of virgin Holstein heifers differed based on insemination outcomes and season of breeding and might be a new phenotype to indicate heifer fertility.

In the second experiment, multiparous Holstein cows (n = 12) were placed in individual box stalls 14 d before expected calving. Sterile swabs were used to collect samples from the posterior vagina of the dam approximately 24 h before calving, dam feces, dam oral cavity, and colostrum within 1 h after calving, and cotyledonary placenta within 6 h after calving. Calves (n = 12; bulls = 8, heifers = 4) were isolated immediately after parturition to prevent environmental contamination. Colostrum was fed to calves using a clean bottle that was assigned to the calf for the duration of the study. Calves were individually housed for 60 d until weaning. Sterile swabs were used to collect calf fecal samples at birth, 24 h, 7 d, 42 d, and 60 d of age. A subset of calf-dam pairs (n = 6; bulls = 3, heifers = 3) were selected and DNA was extracted from all samples. Amplicons covering V4-V5 16S rDNA regions were generated using extracted DNA and sequenced using 300 bp paired end sequencing via Illumina MiSeq. Sequences were aligned into operational taxonomic units using the 97% Greengenes reference database. Spearman correlations were performed between maternal and calf fecal microbiomes. Negative binomial regression models were created for genera in calf fecal samples at each time point using genera in maternal microbiomes. Metagenomes were predicted, collapsed into gene pathways and differences in predicted metagenomes were analyzed within STAMP (Statistical Analysis of Metagenomic Profiles). We determined that Bacteroidetes dominated the calf fecal microbiome at all time points (relative abundance ≥ 42.55%) except for 24 h post-calving, where Proteobacteria were the dominant phylum (relative abundance = 85.10%). Colostrum and placenta had low diversity within samples and clustered independently from fecal samples. Each maternal microbiome was a significant predictor for calf fecal microbiome during at least 2 time points. Genes for infectious disease and neurodegenerative disease were greater in colostrum and 24 h calf fecal samples compared to other samples. Results indicated that no one maternal microbiome was a major influence on calf fecal microbiome inoculation and development. Instead, calf fecal microbial development stems from various maternal microbial sources.

Overall, the reproductive microbiome was predictive of heifer pregnancy outcomes and calf fecal microbial development. The virgin heifer uterine microbiome could be used to predict fertility and adaptation to heat stress, but further research including a larger group of pregnancy loss is needed. Maternal microbiomes from the reproductive tract, colostrum, oral cavity, and feces could all be used to predict calf microbial development, but more research including other maternal microbiomes and environmental microbial contributions is needed. However, the results from this dissertation indicate reproductive microbiome composition is a trait that might be predictive of dairy cattle performance.