Calves were selected from a single spring calving, multi-parous herd [1–10 lactations, mean 2.83 (SEM 0.33)] consisting of Holstein–Friesian (n = 31) and Jersey (n = 20) cows. All cows were vaccinated for rotavirus, bovine corona virus, and E. coli (r-c Rotavec®Corona, MSD Animal health, Ireland) 4 to 8 weeks prior to calving and received an annual booster for infectious bovine rhinotracheitis (IBR; Rispoval® IBR-Marker inactivated, Zoetis Belgium S.A., Dublin Ireland), bovine viral diarrhea virus (BVD; Bovilis® BVD, Intervet Ireland Limited, Dublin, Ireland), Leptospirosis (Bovilis® Leptavoid®-H, Intervet Ireland Limited, Dublin, Ireland), and Salmonella (Bovilis® Bovivac S, Intervet Ireland Limited, Dublin, Ireland). Cows calved indoors in a shed away from the main herd, with concrete floors and deep straw bedding.

Colostrum was obtained from each dam using a portable milking unit (MK100 034904 2017 Inv Ref17027151). Within 2 h of birth, each calf received 8.5% of their BW (in liters: L) of colostrum from either their dam or a refrigerated colostrum source. Refrigerated colostrum consisted of colostrum from two or less cows that had calved within the previous 24 h and was stored in a refrigerator (4°C) and is referred to as “mixed.” Mixed colostrum was heated to approximately 38° C using a water bath prior to feeding. Colostrum quality was tested using a digital brix refractometer (Hanna Digital Refractometer for Sugar, HI9681).

Fifty-one spring-born dairy heifer Holstein (HO; n = 29, birth weight (BW) 34.7 (SEM 0.69) kg) and Jersey (JE; n = 22, BW 25.9 (SEM 0.81) kg) calves were enrolled in this study.

Calves remained with their dam for 30 min after calving. Upon removal, each calf was weighed using a calibrated weighing scale and the navel was disinfected (Foradine 10, povidone iodine 10%). The calf was then moved to an individual pen (1.4 m × 0.46 m) until day (d) 3 after birth. From d3 until weaning, calves were housed in group pens (9.1 m × 4.6 m) with 40 calves of similar age.

Calves received, on average, 2.6 L (SE 0.07) of colostrum within 2 h of birth either from the dam (n = 28) or a mixed source (n = 23). Calves were then fed 2 L of transition milk twice daily using a bucket with a teat. From d3 to d14, calves received transition milk twice a day using a milk trough with teats, gradually increasing from 4 L to 6 L per calf per day. From second feeding to d7, calves received paromomycin (Parafor™, BIOVET JSC, Bulgaria) preventatively against Cryptosporidiosis within each milk feed. At d14, calves were transitioned to an allocation of 6 L of milk replacer [Heiferlac GB/IRL, Volac Feeds LTD, Ireland; 26% crude protein (CP)] per day fed via automated calf feeder (Forster Technik, Volac). Calves had access to roughage (straw) and pelleted concentrates (Sweet start calf starter pencils, 20% CP, Southern Milling LTD., Co. Cork, Ireland) from d3 until weaning.

Calves were housed indoors in deep straw-bedded pens and had access to an outdoor area that was bedded with woodchips. At 2 weeks of age, calves received an oral dose of toltrazuril (Bovicox, KRKA, Slovenia) for Coccidiosis and were vaccinated against Clostridia (Tribovax10, Intervet Ireland LTD., Ireland) and respiratory disease pathogens (Pneumovac PLUS, Animal Health Distributors, Ireland) subcutaneously and received booster vaccinations 3 weeks later against the respiratory disease pathogens (Pneumovac PLUS, Animal Health Distributors, Ireland).

Calves were weaned gradually off milk replacer for 7 days based on targeted weights (HO 85 kg; JE 75 kg) and concentrate intake (minimum 1 kg of concentrate per calf per day for 3 consecutive days). Once weaned, calves were turned out to pasture and offered 1 kg of concentrates/calf/day (Super Calf Rearer Nuts, 17% CP, Southern Milling LTD., Ireland).

Sample processing order was determined on a random basis. Using a similar methodology as previously described (McGovern et al., 2018; Smith et al., 2020a), fecal microbial samples from the three sampling time-points (n = 150) were ground to fine powder, under liquid nitrogen, using a pestle and mortar. Microbial DNA was extracted from each sample using approximately 250 mg of ground feces and was subjected to repeated bead beating and column purification using the QIAGEN DNeasy® PowerSoil® Pro Kit (QIAGEN, Manchester, United Kingdom) and stored at −80°C. Microbial DNA extraction was also performed on the ZymoBIOMICS™ Microbial Community Standard (MC; Zymo Research Corp., Irvine, CA, United States), for each extraction kit (n = 9), as an internal positive control. DNA quality was assessed on 0.8% agarose gels with the concentration of extracted DNA quantified on the Nanodrop 1,000 spectrophotometer.

Extracted DNA was sent to Macrogen (Seoul, South Korea) for 16S rRNA amplicon library preparation and sequencing. In brief, the first round of PCR amplification, targeting the V4 hypervariable region of the 16S rRNA gene, was performed using the 515F/806R primers (Caporaso et al., 2011) designed with Nextera overhang adapters and Herculase II Fusion DNA Polymerase (Agilent, Santa Clara, CA, United States). Cycle conditions were as follows: 95°C for 3 min, 25 cycles at 95°C for 30 s, 55°C for 30 s, 72°C for 30 s, and then 72°C for 5 min. PCR amplicon purification was performed using the standard AMpure paramagnetic bead protocol (Beckman Coulter, Indianapolis, IN, United States). Following purification, amplicons were subjected to a second round of PCR to allow the attachment of dual indices and Illumina sequencing adapters using the Nextera XT indexing kit (Illumina, San Diego, CA, United States). Cycle conditions for the second round of PCR were as follows: 95°C for 3 min, 10 cycles at 95°C for 30 s, 55°C for 30 s, 72°C for 30 s, and then 72°C for 5 min followed by an additional PCR purification with the AMpure paramagnetic bead protocol (Beckman Coulter, Indianapolis, IN, United States). Amplicons were pooled together in equal concentration and were subjected to sequencing on the Illumina MiSeq using the 500 cycle version 2 MiSeq reagent kit (Illumina, San Diego, CA, United States) split equally across two flow cells.

Amplicon sequence data were processed in R (version 4.2.0) using DADA2 (version 1.26.0) and submitted to the pipeline as described by Callahan et al. (2016), with minor modifications as outlined by Smith et al. (2020b, 2021). Quality check of both forward and reverse reads was carried out based on the visualization of Q scores, ensuring that the mean Q scores of >30 were upheld for forward and reverse reads. Filtering and trimming of poor quality reads and removal of primer sequences were conducted using the trimLeft function in DADA2. Identical sequences were combined using the de-replication function followed by the merging of forward and reverse reads. An amplicon sequence variant (ASV) table was then constructed following which chimeric sequences were removed and taxonomy was assigned to sequence variants using the SILVA database (version 138.1) downloaded from the DADA2 website.¹ A bootstrapping threshold of 80 was applied for taxonomic classification by incorporating minBoot = 80 as part of the assignTaxonomy function, as described by Smith et al. (2020b). Sample metadata, sequence taxonomy, and ASVs were combined into a phyloseq object using phyloseq (version 1.44.0; McMurdie and Holmes, 2013) and analyzed in RStudio (version 4.3.0).

A rarefication curve was plotted for fecal microbial samples (Figure 2). Based on plateauing of the generated rarefication curve, sequencing was deemed to be conducted to a sufficient depth. Following this, only bacterial ASVs, classified beyond the phylum level, were obtained. Subsequently, alpha (Shannon) diversity was calculated for each sample. For comparisons of beta diversity, as well as differential abundance analysis, ASVs which were not present in >0.05% of the samples were removed before calculating the relative abundance (RA).

Calf performance and clinical health datasets were analyzed using statistical procedures in SAS 9.4. Data for STP and IgG concentrations, average daily gain, and RT were checked for normality and homogeneity of variance by histograms, qqplots, and formal statistical tests as part of the UNIVARIATE procedure. Data were then analyzed using a repeated measures ANOVA (PROC MIXED procedure, SAS 9.4). The model included fixed effects (calf breed, dam breed, health status, and colostrum source) and random effects (shed and lactation number). Differences among means were determined by F-tests using Type III sums of squares. The PDIFF option and Tukey’s test were applied, where appropriate, to evaluate pairwise comparisons between means. Non-significant (p > 0.10) interactions were subsequently excluded from the final model. Mean values were considered statistically significant at p < 0.05, and tendency was considered when p > 0.05 and < 0.10. Pearson’s correlation coefficients were determined using the CORR procedure. Correlations (effect size and the strength of the correlation are denoted as r) were described as follows: 0.00–0.19 “very weak”; 0.20–0.39 “weak”; 0.40–0.59 “moderate”; 0.60–0.79 “strong”; 0.80–1.0 “very strong” (Swinscow, 1997). Correlations with p < 0.05 were considered statistically significant. Clinical health scores were analyzed using a Wilcoxon Rank Sum test (NPAR1WAY procedure).

The effect of health status and time-point on alpha diversity of the bacterial community was assessed using the PROC MIXED procedure in SAS. The model included fixed effects (calf breed, dam breed, health status, and colostrum source) and random effects (shed and lactation number). Differences among means were determined by F-tests using Type III sums of squares. The PDIFF option and results from Tukey’s test were applied to evaluate pairwise comparisons between means. Mean values were considered to be different when p < 0.05 and a tendency when p ≥ 0.05 and < 0.10.

Before assessing the effects of health status and time-point on the overall bacterial community structures, the homogeneity of group dispersion was assessed between groups. Following this, PERMANOVA tests based on Bray–Curtis dissimilarities, 9,999 permutations, and a significance level of p < 0.05 were implemented to determine if the bacterial community structure was impacted by calf health status, time-point, colostrum source, and calf breed. Both the assessment of the homogeneity of group dispersion and PERMANOVA analysis were carried out using the R package vegan (Oksanen et al., 2019; version 2.5.7) implemented through microbiome (Lahti and Shetty, 2017; version 1.12.0). Due to poor species level classification of ASVs, differential abundance analysis was performed at the genus, family, and phylum levels. Data were subset into time-points and then filtered for ASVs with an RA of >0.05% using phyloseq. At each time-point, statistical analysis of ASVs was performed using MaAsLin2 (version 1.14.1; Mallick et al., 2021) with time-point and health status as fixed effects and the Benjamini–Hochberg procedure to correct for false discovery rate. The three time-point files were then merged into one phyloseq object and run through the MaAsLin2 package with time-point and health status as fixed effect and the Benjamini–Hochberg procedure to correct for false discovery rate. The resulting data were used to determine the effect of time-point and health status on ASVs. While taxonomy was assigned using the SILVA database (version 138.1), phyla names were updated according to Oren and Garrity (2021). A Spearman rank-order correlation (r_s) for non-parametric data was performed to determine correlations between ASVs and calf performance data and fecal scores at disease manifestation. Spearman correlations (effect and strength are denoted as r_s) were described using the same methodology as outlined in section 2.6.1, and any correlations with p < 0.05 were considered significant.

Colostrum fed to calves had a mean Brix score of 26% (SE 0.51). Calves were weaned, on average, at d83 (SE 1.04; HO: d79, SE 1.22; JE: d88, SE 1.16) post-birth at an average weight of 82 kg (SE 1.10; HO: 85 kg, SE 1.38; JE: 79 kg, SE 1.50). Colostrum source and calf breed had no effect (p > 0.05) on disease incidence. Of the 51 calves enrolled, diarrheal incidence occurred in 53% (n = 27), with a mortality of 2% (n = 1) from birth to weaning. Day of diarrheal disease manifestation ranged from d18 to d38, with the mean day post-birth for diarrhea being d22 (SE 0.70).

Serum IgG concentrations were strongly correlated (r = 0.61; p < 0.0001) with STP concentrations. There was no effect (p > 0.05) of colostrum source on calf STP, serum IgG concentrations, or AEA. Calf breed had no effect on mean STP or AEA (p > 0.05), whereas serum IgG concentrations were greater (p < 0.04) in Holstein calves than Jersey calves. Healthy calves and diarrheic calves had similar AEA, STP, and serum IgG concentrations (p > 0.05; Table 1).

At d2 (SE 0.23) after birth, calves had a mean RT of 38.8°C (SE 0.07). At perinatal assessment, all calves were healthy. There was calf breed × colostrum source (p = 0.05) for RT where mixed colostrum Jersey calves tended to have a lower temperature than mixed colostrum HO (−0.40°C, SE 0.19; p = 0.05) and dam-fed JE calves (−0.40°C, SE 0.19; p < 0.05). There was no effect of calf breed or colostrum source on perinatal (baseline) health scores [data available on Open Science Framework (OSF)].

At d7, calves had a mean RT of 39.1°C (SE 0.05). There were no calf breed × colostrum source interactions or effects of calf breed or colostrum source on RT prior to disease manifestation (DM; data available on OSF).

Calves were clinically diagnosed with diarrhea at a mean of d22 (SE 0.70) after birth. Health status had an effect on temperature (p < 0.0001), where healthy calves had lower (p = 0.0004) RT [38.9 (SE, 0.07)] than diarrheic calves [39.3 (SE 0.05)].

All health indicators were normal (median score of 0) with the exception of feces, which had a median score of 2 during disease manifestation (Table 2). There was no effect (p > 0.05) of colostrum source on any of the indicators that were used to assess health. Calf breed had an effect on joint scores (p < 0.0001), where HO had higher joint scores than JE. Health status had an effect on ear (p = 0.001), fecal (p < 0.0001), dehydration (p = 0.03), and mucus membrane (p = 0.05) scores such that diarrheic calves scored higher in these categories than healthy calves (Table 2).

At d83 (SE 1.04) after birth, calves had a mean temperature of 39.0°C (SE 0.03). All indicators had a median score of 0 with the exception of ocular discharge, which had a median score of 1 after diarrheal disease. Colostrum source had an effect on calf temperature, where at weaning, dam-fed calves had a lower temperature (−0.17°C, SE 0.07; p = 0.01) than mixed colostrum calves. There were no other effects or interactions on calf temperature at weaning. Health status had an effect on navel score. There were no other interactions or effects on health scores at weaning (data available on OSF).

Based on PERMANOVA analysis, calf breed (p = 0.08), colostrum source (p = 0.31), and passive immune status (p = 0.21) had no effect on the overall fecal microbiota composition. At the genus level, 396 unique ASVs were detected throughout the pre-weaning period across the three time-points (before diarrheal disease, during diarrheal disease, and after diarrheal disease). Genera and the adjusted p-values (adj.p) of ASVs associated with health status across the pre-weaning period are presented in Table 3. At the species level, 4,638 ASVs were detected. Over the pre-weaning period, four phyla dominated the fecal microbiota such as Bacillota (formerly known as Firmicutes—71%), Bacteroidota (15%), Actinomycetota (formerly known as Actinobacteria—5%), and Pseudomonadota (formerly known as Proteobacteria—4%), which accounted for 95% of the RA of bacteria identified. Initially, before diarrheal disease, Actinomycetota (Actinobacteria—33%), Pseudomonadota (Proteobacteria—10%), and Bacteroidota (23%) made up the majority of microbiota present in the feces of all calves. As time progressed throughout the pre-weaning period, microbes within the Bacillota (Firmicutes) phylum began to proliferate, becoming the dominate phylum at disease manifestation (48%) and post-disease (52%) compared with pre-disease (34%).

Before diarrheal disease, during disease manifestation, and after diarrheal disease, alpha diversity changed significantly (p = 0.0001) overtime. The fecal microbiota of calves became progressively more diverse during the pre-weaning period (Figure 3). Health status (p = 0.01) had a significant effect on the overall alpha diversity of the fecal microbiota. A significant interaction between time-point and health status (p = 0.0001) was observed, whereby at the second time-point (d21/DM), diarrheic calves had a reduction in microbial diversity (Figure 4). There was an effect of health status (p = 0.003) on differences in alpha diversity from d7 to d21/DM, and from d21/DM to d83, and none between d7 and d83.

No ASVs were solely associated with health status on a genus level. At the species level, 9 ASVs were associated with health status during the pre-weaning period. Five of these ASVs were unclassified species within the Prevotella_7 genus. The remaining four ASVs included Faecolicoccus pleomorphus, unclassified Lachnoclostridium, unclassified Dialister, and an unclassified microbe in the Ruminococcaceae family. Lachnoclostridium was the only ASV with a greater RA observed in diarrheic calves. All other ASVs were found to have a higher overall RA in healthy calves (Table 4).

Serum total protein and serum IgG concentrations were moderately and weakly correlated with a variety of ASVs associated with health status in all calves, diarrheic calves, and healthy calves. Moderate correlations were observed between ASVs and ADG in diarrheic calves and in healthy calves (Table 6). Across all calves at disease manifestation, correlations between fecal score and all ASVs significantly associated with health status ranged from weak to strong (Table 6). There was a strong positive correlation between fecal score at disease manifestation and Alloprevotella (r_s = 0.61, p ≤ 0.0001). Fecal scores were observed to have a moderate negative correlation with Bifidobacterium, Colinsella, Prevotella_7, Faecalicoccus, ASV 58, Erysipelatoclostridium, Akkermansia, Flavonifractor, ASV 91, Dialister, and Intestinibacter. Moderate positive correlations were observed between fecal scores and Prevotella_9, Ligilactobacillus, ASV 25, and Allisonella.

At the time of disease manifestation, 29 strong correlations (r_s range as absolute value: 0.60–0.77), 72 moderate correlations (r_s range as absolute value: 0.40–0.59), and 72 weak correlations (r_s range as absolute value: 0.27–0.39) of significance (p ≤ 0.05) were observed between the various genera associated with the health status (Table 7).

In the current study, the temporal development of the fecal microbiota observed during the pre-weaning period is in agreement with the findings reported by Song et al. (2018), Hennessy et al. (2020), and Massot et al. (2020), such that overall diversity of the fecal microbiota increased with time, supporting the assertion that microbial colonization of the lower gastrointestinal tract of calves occurs sequentially and progressing rapidly from birth to weaning as has been reported by Alipour et al. (2018).

Neonatal calves are pseudo-monogastric, relying on a milk-based diet, with the main site of digestion, via microbial fermentation, occurring in the small and large intestines (Amin and Seifert, 2021). Inoculation of the hindgut comes from a variety of sources including the placenta and other maternal factors, as well as initial diet and feeding regime (Amin and Seifert, 2021; Du et al., 2023). Milk oligosaccharides are a critical energy source for the initial microbial community (Boudry et al., 2017). The feeding of colostrum has been reported to influence the initital colonization of the hindgut, in that calves that were fed colostrum (fresh and heat-treated) had an increased presence of Bifidobacterium in the first 12 h of life, and feeding heat-treated colostrum may be even more beneficial, even though Bifidobacterium was not present in the colostrum fed (Song et al., 2019). Calves are relied on high-quality colostrum for successful passive transfer of immunity from the dam (McGee and Earley, 2019). The combination of immunoglobulins, antimicrobial factors, growth factors, anti-inflammatories, and nutrients in colostrum are essential for calf health and development (Carter et al., 2021). The nutrients and other bio-active compounds, e.g., lactoferrin, present in colostrum are important for pioneering bacteria that are colonizing the hindgut (Fischer-Tlustos et al., 2021). The feeding of colostrum helps to inhibit pathogen growth, stimulate colonization of the small intestine with commensal microbes, increased weight gain, improved development and function of the gastrointestinal tract, and reduce the risk of diarrhea (Song et al., 2019; Amin and Seifert, 2021; Carter et al., 2021; Fischer-Tlustos et al., 2021). It has been suggested that bacteria found in colostrum may influence the colonization and proliferation of early life microbial communities in the calf hindgut (Gomez et al., 2019). In the current study, calves were handfed colostrum collected using different management practices, colostrum that was collected directly from the calf’s own dam and fed to the calf or colostrum that was sourced from one to two donors, refrigerated and re-heated. Colostrum from a refrigerated, mixed source is at higher risk of environmental contamination (Godden et al., 2019). Increased environmental contamination with microbes could possibly lead to increased risk of disease (Godden et al., 2019) but also increased number of microbes inoculating the hindgut. Thus, one could hypothesize that these different management practices may have had an effect on disease incidence and the composition of the fecal microbiota of the calf. However, the different colostrum sources had neither an effect on diarrheal disease manifestation nor an effect on the diversity or composition of the fecal microbiota at any of the sampling time-points during the pre-weaning period. This may be related to multiple factors in the animal model of the current study. First, if fed quality colostrum, the source of the colostrum (own dam, different dam, pooled) has no effect on serum IgG and passive transfer of immunity, as was reported by King et al. (2020). There was little variation in immune status between diarrheic and healthy calves, and any effect of colostrum source on serum IgG, implying that any changes observed in the fecal microbiota, is neither associated with diarrheal disease manifestation nor poor passive transfer of immunity. Second, the mixed colostrum that was stored in a refrigerator, was collected from the dam and stored as quickly as possible, had been refrigerated for no more than 24 h and was heat-treated, according to best practices prior to feeding. These practices are in line with best practices for minimizing bacterial contamination as suggested by Godden et al. (2019). Additionally, as has been previously discussed, Song et al. (2019) have reported that the feeding of heat-treated colostrum (60°C for 60 min) increases the abundance of beneficial genera such as Bifidobacterium. Lastly, from second feed to d14 all calves were fed re-heated, pooled-transition milk sourced from the maternal herd.

Host genetics has previously been suggested to influence the composition of the hindgut microbiota (Paz et al., 2016; Fan et al., 2020; Slanzon et al., 2022). However, in contrast to the findings by Slanzon et al. (2022), there was no effect of calf breed on the diversity and overall composition of the fecal microbiota in the current study. The calves enrolled in the study by Slanzon et al. (2022) were of similar breeds to those of the present study (Holstein, Jersey, and Jersey-cross) with the exception of beef-crosses and were 4–21 days of age. Differences between the present study and others may be due to a variety of factors, such as multiple sampling time-points vs. singular ones, calf age and life stage, or calf-farm origin and other environmental factors. The difference in the findings between the present study and the study by Slanzon et al. (2022) could be due to calf origin and different diets. Slanzon et al. (2022) found that breed had an effect on the fecal microbiome of calves, all of which originated on different farms and were then reared indoors together. The differences, however, were most evident in calves with or without gastrointestinal disease rather than between breeds (Slanzon et al., 2022). The calves enrolled in the study by Slanzon et al. (2022) were administered two different milk diets depending on enrollment date, which included pasteurized waste milk, whereas calves in the present study were home bred and fed colostrum artificially from within the maternal herd. Additionally, calves were fed transition milk sourced from the maternal herd until d14 when they were transitioned to milk replacer. This difference in management may suggest that breed differences detected in early life may in fact be related to differences in environment and feeding strategies. The lack of effect of breed on diversity and composition in the present study may also be due to age. All calves were enrolled, and all sampling time-points occurred during the pre-weaning period. Fan et al. (2020) found that host genetics did have an effect on composition in animals at 3 months of age, while no differences were detected in calves at a similar age (d83) in the present study. However, Fan et al. (2020) examined the effect of species and crossbreeding on the early gut microbiota of calves ranging from 100% Angus to 100% Brahman, with varying degrees of cross-breeding between the two mid-points. While there were two different breeds (Holstein × and Jersey ×) enrolled in the present study, they were of the Bos Taurus species, suggesting that at the species level, host-genetics may have an effect on the composition of the fecal microbiota. At maturity, an effect of breed has been observed in the composition of the rumen bacterial community, such that the composition of the rumen microbiota varied between lactating Holstein and Jersey cows (Paz et al., 2016) reflecting the findings reported by Fan et al. (2020), regarding the relationship between age, host-genetics, and composition. However, even with a period of acclimation, the Jersey cows sampled by Paz et al. (2016) were from a different farm origin than the Holstein cows that were used for comparison. The lack of breed effect on composition and diversity in the present study may be related to age, although no breed effect was observed at post-disease, where calves were approximately 3 months old. The findings of the current study, however, suggest that in the aforementioned studies, an effect of farm origin may be coming as an apparent effect of breed and host genetics.