Feeding selenium-biofortified alfalfa hay during the preconditioning period improves growth, carcass weight, and nasal microbial diversity of beef calves

We previously reported that feeding Se-biofortified alfalfa hay to weaned beef calves in a preconditioning program decreases morbidity and mortality during the feedlot period. To understand the mode of action by which supranutritional Se supplementation supports calf health, we examined the effect of agronomic Se-biofortification on nasal microbiome and fecal parasites. Recently weaned Angus-cross beef calves (n = 30) were randomly assigned to two groups and fed an alfalfa hay-based diet for 9 weeks in a preconditioning program. Alfalfa hay was harvested from fields fertilized with sodium selenate at a rate of 0 or 90 g Se/ha. Calculated Se intake from dietary sources was 1.09 and 27.45 mg Se/calf per day for calves consuming alfalfa hay with Se concentrations of 0.06 and 3.47 mg Se/kg dry matter, respectively. Feeding Se-biofortified alfalfa hay for 9 weeks was effective at increasing whole-blood Se concentrations (556 ± 11 vs 140 ± 11 ng/mL; P < 0.001) and increasing body weight (PTreatment, = 0.03) in weaned beef calves. Slaughter yield grades were higher for calves that had been fed Se-enriched alfalfa hay during the preconditioning period (PTreatment = 0.008). No significant differences were observed in fecal parasite load, which remained low. The nasal microbiome and microbiota diversity within calves and across calves expanded from weaning (week 0) to the feedlot period (week 12), which was promoted by feeding Se-biofortified alfalfa hay. Especially concerning was the expansion of nasal Mycoplasmataceae in the feedlot, which reached over 50% of the total microbiota in some calves. In conclusion, we identified dietary Se-biofortified alfalfa hay as a potential promoter of nasal microbiome genome and microbiota diversity, which may explain in part high-Se benefits for prevention of bovine respiratory disease complex in beef calves.

Introduction commercial feedlot in Burbank, WA (Simplot Feeders, Pasco WA). Calves were physically examined and nasal swabs and blood and fecal samples were collected 3 weeks after transfer to the feedlot. Health information was available through 5 weeks after transfer to the feedlot. At the end of the 25-week feedlot period, calves were sent to a commercial meat packing plant (Tyson Fresh Meats Inc.; Pasco, WA). Hot carcass weights, carcass quality grades (no-roll, standard, select, choice, prime), and yield grades (1 to 4) were recorded.

Selenium biofortified-alfalfa hay preparation and analyses
Third cutting alfalfa hay was enriched with Se by mixing inorganic sodium-selenate (RETORTE Ulrich Scharrer GmbH, Röthenbach, Germany) with water and spraying it onto the soil and foliage (approximately 10 cm height) of an alfalfa field at application rates of 0 or 90 g Se/ha in July 2017. The application rate was chosen based on a previous study [6]. Thirdcutting alfalfa hay was harvested early October 2017 and then analyzed for Se and nutrient content. A Penn State forage sampler was used to take 25 cores from random bales in each hay source (0 or 90 g Se/ha) prior to beginning the feeding trial. Core samples were mixed well and representative samples selected for Se analysis (Table 1; Utah Veterinary Diagnostic Laboratory, Logan, UT). Plant samples were prepared for Se analysis as previously described [12], and Se was determined using inductively coupled argon plasma emission spectroscopy Potassium, g/kg 9.7 10.0 6.0 Sodium, g/kg 3.9 3.8 3.2 Copper, mg/kg 14 15 6 Iron, mg/kg 627 1196 214 Manganese, mg/kg 39 56 116 Zinc, mg/kg 19 21 28 Selenium, mg/kg 0.06 3.47 0.12 (ICP-MS; ELAN 6000, Perkin Elmer, Shelton, CT). Quantification of Se was performed by the standard addition method, using a 4-point standard curve. A quality-control sample (in similar matrix) was analyzed after every 5 samples, and Se analysis was considered acceptable if the Se concentration of the quality-control sample fell within ± 5% of the standard/reference value for the quality control. Alfalfa hay samples were also submitted to a commercial laboratory for routine nutrient analysis (Table 1; Cumberland Analytical Services, Waynesboro, PA). Alfalfa hay dry-matter determination was completed at a temperature of 105˚C for 12 to 14 h in a forced draught oven. Methods for crude protein (CP), acid detergent fiber (ADF), ash, and minerals were performed according to the Association of Official Analytical Chemists [13]. The neutral detergent fiber (NDF) was determined according to Van Soest et al. [14]. Soluble protein was determined according to Krishnamoorthy et al. [15].

Blood Se collection and analyses
Blood samples were collected from the jugular vein of weaned beef calves, at baseline, after 3, 6, and 9 weeks of alfalfa hay consumption, and 3 weeks after transfer to the feedlot, into evacuated EDTA tubes (2 mL; final EDTA concentration 2 g/L; Becton Dickinson, Franklin Lakes, NJ) and stored on ice until they were frozen at -20˚C to measure WB-Se concentrations. Selenium concentrations were determined by a commercial laboratory (Utah Veterinary Diagnostic Laboratory, Logan, UT) using an ICP-MS (ELAN 6000, Perkin Elmer, Shelton, CT) method as previously described [6].

Fecal collection and analyses
Fecal samples were collected 30 days before the feeding trial began, after 5 and 9 weeks of alfalfa hay consumption, and again 3 weeks after transfer of calves to the feedlot. Fecal samples were extracted manually per rectum using a powder-free latex examination glove (Diamond Grip TM Latex Gloves, Microflex, Reno, NV) while calves were restrained in a chute. Fecal samples were stored in the glove at 4˚C until analysis. The minimum sample amount was 10 grams.
For quantification of fecal trichostrongyle-type egg counts, we followed the McMaster egg count method of Whitlock [16]. In brief, 2 grams of feces were mixed with a mortar and pestle in 13 mL of saturated NaCl solution in order to facilitate the release of the ova from the fecal material and to prepare a homogeneous solution of the feces. The suspension was poured through food grade cheese cloth to remove large particles. Using a disposable Pasteur pipette (GeneMate, BioExpress, Kaysville, UT), we transferred an aliquot of the fecal solution to the chambers of a McMaster slide (Chalex Corp., Wallowa, OR). The slide was examined under low power (100X) by focusing on the etched lines on the underside of the top chamber. Both chambers on the McMaster's slide were counted and the number of trichostrongyle-type eggs was multiplied by 25 to obtain the number of eggs/g of feces. Counting was complete within 2 days of sample collection.
When at least 25 trichostrongyle-type eggs/g feces were observed, fecal samples were submitted to a commercial laboratory (Veterinary Diagnostics Laboratory, Oregon State University) and assayed for the presence of H. contortus eggs using a fluorescein-labeled peanut agglutinin test, following the procedure of Jurasek et al. [17]. In brief, 2 grams of fecal matter were placed in 5 mL of water and crushed to break up the fecal material. Next, 93 mL of water was added (for a total of 98 mL water) and mixed with the fecal material. The sample was then refrigerated overnight to allow the eggs to separate from the fecal matter. The next morning, the sample was thoroughly mixed to form a uniform suspension and immediately decanted into a centrifuge tube, which was spun at 280 × g for 5 minutes. The supernatant was decanted and 2 mL of Stoll's saturated sugar solution (specific gravity = 1.27) [18] was added and mixed with the fecal pellet. The tube was then filled to within a few mm from the top with the sugar solution and centrifuged at 280 × g for 5 minutes. The tube was placed in a stable upright rack and filled with additional sugar solution until a convex meniscus formed, and then a 22 mm 2 coverslip was placed on top. The solution sat for 1 hour to allow eggs to float to the top of the tube and collect on the coverslip. The coverslip was then gently removed and rinsed with 100 mM 7.4 pH phosphate buffered saline (PBS) into a 1.5 mL Eppendorf tube. The PBS solution was added to 1.5 mL, and the sample was rinsed by centrifuging at 280 × g for 5 minutes. The supernatant was removed and the egg sediment resuspended in an additional 1.5 ml of PBS. Eggs in PBS were centrifuged again at 280 × g for 5 minutes, the supernatant removed, and the pellet resuspended in 1mL of fluorescein isothiocyanate-labeled peanut agglutinin (Sigma Cat. No. L-7381 lectin from Arachis hypogaea, reconstituted at 5 μg/1 mL in PBS; Sigma, St. Louis, MO). Eggs then were incubated in the lectin for 1 hour under constant agitation at room temperature. Samples were washed twice more in PBS, as described above, and 5 μL of the egg sediment was transferred onto a glass slide with 3 μL of fluorescent mounting fluid (Veterinary Medical Research and Development Inc., Pullman, WA) and then overlaid with a coverslip. Specimens were examined using a fluorescence microscope with fluorescein isothiocyanate filters (480-490 excitation/527/30 emission). The percentage of total ova that were positively identified as H. contortus was reported. If large numbers of ova were present, this was based on counting 100 eggs.

Nasal microbiota collection and analyses
Nasal swabs were collected from all calves 30 days before the feeding trial began, after 9 weeks of alfalfa hay consumption, and again 3 weeks after transfer to the feedlot. Sterile, individually wrapped, polyester tipped applicators (Puritan1, Guilford, ME) were inserted approximately 10 cm into the ventral meatus of the nares, twirled to collect a mucosal swab, and then placed into individual sterile containers (10 mL, red top, BD Vacutainer collection tubes; Becton Dickinson, Franklin Lakes, NJ) avoiding any contamination or contact with the plastic stick. Swabs in tubes were subsequently frozen at -80˚C within 4 hours of collection. Negative control (sterile) swabs were similarly processed.
Microbial DNA was extracted from nasal swab samples using MoBio Power soil DNA isolation kit (MoBio Laboratories, USA) as per the manufacturer's instructions. The V4 region of the 16S rRNA gene was amplified with primers 515F (5 0 -GTGCCAGCMGCCGCGGTAA-3 0 ) and 806R (5 0 -GGACTACVSGGGTATCTAAT-3 0 ) at the MR DNA Laboratory (Shallowater, TX, USA) and sequenced on an Illumina MiSeq instrument at the MR DNA Laboratory (Shallowater, TX, USA).
The sequencing data were processed using the QIIME 2 v 2018.6 (https://qiime2.org/) platform [19]. The raw reads were deposited in NCBI SRA under the accession number SRP144214. Raw sequence data were de-multiplexed with the q2-demux plugin in QIIME2.
The data was quality filtered by removing low quality regions and chimeric sequences using the Divisive Amplicon Denoising Algorithm 2 (DADA2) with the q2-dada2 plugin [20] to create an amplicon sequence variant (ASV) table. A masked alignment of the sequence variants was conducted using MAFFT [21] with the q2-alignment plugin. A phylogeny tree was created with FastTree2 [22] using the q2-phylogeny plugin. Furthermore, taxonomy was assigned using the QIIME2 naive Bayes feature classifier [23] trained on the Greengenes 13_8 database [24]. The feature table was also filtered to remove sequences that were classified as mitochondria and chloroplasts.
Sequences were rarefied to an even depth of 31,744 sequences per sample to account for unequal sequencing depth across samples in QIIME 2 using the "feature-table rarefy" plugin. The q2-diversity plugin was used to calculate measures of alpha-diversity (observed ASVs, Chao1, Shannon index) and beta diversity (weighted and unweighted UniFrac distances, Bray-Curtis Dissimilarities).
To determine if there were any changes in microbial function in the nasal microbiota, PICRUSt2 [25] (Phylogenetic Investigation of Communities by Reconstruction of Unobserved States) was applied to make functional gene content predictions on the 16S rRNA gene sequencing data. The standard pipeline was implemented. The functional profiles were obtained with PICRUSt2 using standard pipeline and using the default options with picrust2_pipeline.py.

Statistical analyses
Data for Se status and growth characteristics were analyzed as repeated-measures-in-time using PROC MIXED in SAS version 9.2 [26]. Data were averaged within feeding pens. Fixed effects in the model were Se application rate (0 and 90 g Se/ha), baseline values (as linear covariate), time (after 3, 6, and 9 weeks of feeding Se enriched hay), and the interaction between Se application rate and time. An unstructured variance-covariance matrix was used to account for variation of measures within pens. The Kenward-Rogers option was used to account the degrees of freedom for repeated measures within time. Carcass and other data collected after the preconditioning period were analyzed in PROC GLM with treatment as fixed effect.
Fecal parasite egg counts were logarithmically transformed by base 10 to achieve normal distribution. Data were measured over time; thus, repeated-measures-in-time analysis was performed using PROC GLIMMIX in SAS, assuming a negative binomial distribution. Fixed effects in the model were Se application rate (0 and 90 g Se/ha), baseline values (as linear covariate), time (after 5 and 9 weeks of feeding Se enriched hay), and the interaction between Se application rate and time.
Data analysis of nasal swabs was performed on the rarefied, total sum scaling (TSS) normalized ASV table using Calypso v 8.84 [27]. The diversity of bacterial communities within calves (alpha diversity) were determined using the Chao1 index and the observed ASVs, which both measure the presence of bacterial taxa, and the Shannon index, which also accounts for the relative abundance of bacterial taxa [28,29]. Differences in bacterial communities between calves and within calves across time (beta diversity) were determined using the Bray Curtis dissimilarity distance metric for functional gene data and the phylogeny-based unweighted and weighted UniFrac distance metrices for other microbiome and microbiota data [30][31][32]. Unweighted UniFrac distances measure differences in the presence of bacterial taxa, whereas weighted Uni-Frac distances measure not only differences in the presence of bacterial taxa but also in the relative abundance of bacterial taxa. The diversity measures and the number of bacterial taxa and functional genes present were analyzed as repeated-measures-in-time using PROC MIXED in SAS. Fixed effects in the model were Se application rate (0 and 90 g Se/ha), time (weeks 0, 9, and 12), and the interaction between Se application rate and time. An unstructured variancecovariance matrix was used to account for variation of measures within calves. The Kenward-Rogers option was used to account the degrees of freedom for repeated measures within time.
Principal coordinate analysis (PCoA) plots for unweighted UniFrac distances and Bray Curtis dissimilarities were generated with QIIME2. The ANOSIM (Analysis of Similarity) test within PRIMER 6 software package (PRIMER-E Ltd., Luton, UK) was performed on the weighted and unweighted UniFrac and Bray Curtis distance metrics to find significant differences in microbial communities between groups. To compare bacterial taxa between groups and across time, the non-parametric Kruskal Wallis rank sum test was used for bacterial taxa that were present in 14 or more of 15 calves of Control and High-Se groups in SAS. The nonparametric Fisher's Exact test was used for bacterial taxa that were present in less than 14 calves of Control and High-Se groups in SAS. In addition, Linear discriminant analysis effect size (LEfSe) was utilized within Calypso [27] to identify bacterial taxa that were differentially abundant between Control calves and calves fed High-Se enriched alfalfa hay, with an LDA score threshold of > 3.0 (results not shown). All statistical tests were two-sided. Data are reported as least square mean ± SEM. Statistical significance was declared at P � 0.05 and a tendency at 0.05 < P � 0.10.

Effect of feeding weaned beef calves Se-biofortified alfalfa hay in a preconditioning program on Se intake
Fertilizing the alfalfa hay field with sodium-selenate at 90 g/ha increased the Se content of third-cutting alfalfa from 0.06 (non-fertilized control) to 3.47 mg Se/kg DM. Calculated Se intake from alfalfa hay was 0.46 and 26.82 mg Se/calf per day, respectively, for calves consuming hay with Se concentrations of 0.06 and 3.47 mg Se/kg DM. The concentration of Se in the grass hay was 0.12 mg/kg. Calculated Se intake from grass hay was approximately 0.11 mg Se/ calf per day. The measured Se concentration of the grain concentrate was 0.77 mg/kg DM. Calculated Se intake from grain concentrate was 0.52 mg Se/calf per day in weeks 7 through 9 of the preconditioning period. Thus, total dietary Se intake during weeks 7 to 9 was 1.09 and 27.45 mg Se/calf per day, respectively for calves in control and high-Se treatment groups.

Effect of feeding weaned beef calves Se-biofortified alfalfa in a preconditioning program on Se status
Feeding Se-fertilized alfalfa hay was effective at increasing WB-Se concentrations in weaned beef calves (P Treatment, P Week , and P Interaction all � 0.004; Fig 1). The normal reference interval for WB-Se concentrations of adult cows is 120 to 300 ng/mL [5]. The WB-Se concentrations continued to increase throughout the 9-week preconditioning period (week 3: +50%; week 6: +192%; week 9: +292%). During the initial feedlot period (at week 12: +272%), WB-Se concentrations remained higher (P Treatment < 0.001).

Effect of feeding weaned beef calves Se-biofortified alfalfa hay in a preconditioning program on growth characteristics and carcass data
All calves remained healthy during the 9-week preconditioning period and were healthy upon physical examination 3 weeks after transfer to the feedlot. One calf was treated for a fever and hock injury 4 weeks after transfer to the feedlot. Otherwise, no calves experienced clinical signs of BRD through week 5 after transfer to the feedlot.
Feeding Se-fertilized alfalfa hay was effective at increasing BW in weaned beef calves (P Treatment, = 0.03, P Week , < 0.001, and P Interaction = 0.42) and tended to be effective at increasing hot carcass weight at week 34 (P Treatment = 0.07; Fig 2). During the initial feedlot period (at week 12) no significant effect of feeding Se-enriched hay on BW was observed (+2.2%; P Treatment = 0.58). At slaughter, no significant differences were observed for carcass quality grade levels [prime (highest), choice, select, standard, no-roll (lowest)], as both groups had all but 4 animals with choice grade (P Treatment = 0.53). However, yield grade [1 (most desirable trim), 2, 3 (industry average), and 4] were improved in animals that were fed Se-enriched alfalfa hay during the preconditioning period (P Treatment = 0.008), as more calves in the group fed Se-enriched alfalfa hay had a 1 or 2 (86%) compared with calves in the control group (36%). Slaughter data were not available for 4 calves in the control group and 1 calf in the high-Se treatment group because ear tags were lost at slaughter.

Effect of feeding weaned beef calves Se-biofortified alfalfa hay in a preconditioning program on fecal parasite counts
We evaluated fecal parasite load at weeks 0, 5, and 9 (end) of a preconditioning period as well as in the feedlot (week 12). Overall, fecal parasite counts were low; especially counts for Nematodirus, Maniezia, and Trichuris were mostly zero. The range for coccidia oocyst counts were between 0 and 600 oocysts/g and for trichostrongyle-type egg counts were between 0 and 450 eggs/g. Feeding Se-fertilized alfalfa did not significantly alter coccidia (P Treatment = 0.96) and trichostrongyle-type egg counts (P Treatment = 0.81). No significant time, nor treatment × time interactions were observed. At the beginning of the feedlot period, calves were dewormed. As a result, fecal counts were zero or � 25 eggs/g feces in week 12. Comparison of whole-blood Se concentrations (mean ± SEM) in weaned beef calves after 3, 6, and 9 weeks in the preconditioning period, and in the feedlot (week 12). During the 9 week preconditioning period, calves consumed alfalfa hay harvested from a field not fertilized with Se (Control) or from a field fertilized with sodium-selenate (High Se; application rate of 90 g Se/ha; n = 15 calves per group for Control and High Se). Total dietary Se intake during weeks 7 to 9 was 1.09 and 27.45 mg Se/calf per day, respectively for calves in control and High-Se treatment groups. The normal reference interval for whole-blood Se concentrations in beef cattle is 120 to 300 ng/mL. https://doi.org/10.1371/journal.pone.0242771.g001

Effect of feeding weaned beef calves Se-biofortified alfalfa hay in a preconditioning program on nasal microbiome genome
We collected nasal microbiome genome at weeks 0 and 9 (end of the Se-supplementation preconditioning period) as well as in the feedlot (week 12). Sequencing of the bacterial 16S rRNA gene in the samples yielded 10,308,260 quality sequences (n = 90; mean ± SD, 107,377 ± 48,791). Rarefaction analysis was performed at a depth of 31,744 sequences (Fig 3). Expansion of the nasal microbiome diversity was promoted by feeding Se-biofortified alfalfa hay in the Se-supplementation preconditioning period as counts of 16S rRNA genes expanded within High-Se calves during the Se-supplementation preconditioning period ( Fig 3A). Nasal microbiome diversity increased within Control calves during the transition to the feedlot ( Fig  3A).
The microbial diversity increased within High-Se calves during the Se-treatment period, as indicated by Chao1 (+171%) and observed ASVs (+151%). Control calves caught up with High-Se calves in microbial diversity in the feedlot (Fig 3; Table 2). There were no significant shifts in the microbial diversity toward a more even microbial profile across treatment and time, as indicated by a non-significant Shannon index (Table 2).
Nasal microbiome differed between High-Se and Control calves at the end of the Se-treatment preconditioning period (week 9), as shown by the distinct clustering of High-Se and Comparison of baseline-adjusted BW (kg; mean ± SEM) of weaned beef calves (primarily of Angus breeding and ranging in age from 6 to 9 months at baseline). During the 9 week preconditioning period, calves consumed alfalfa hay harvested from a field not fertilized with Se (Control) or from a field fertilized with sodium-selenate (High Se; application rate of 90 g Se/ha; n = 15 calves per group for Control and High Se). Body weights at baseline ranged from 264 to 369 kg (328 ± 5 kg, mean ± SEM), and final body weights at the end of the preconditioning period ranged from 288 to 411 kg (366 ± 5 kg, mean ± SEM). https://doi.org/10.1371/journal.pone.0242771.g002 Control calves within the principal coordinates (Fig 4) and ANOSIM values (R unweighted = +0.30, P = 0.001) of the unweighted UniFrac distances. Moreover, the nasal microbiome differed across time, as shown by distinct clustering for weeks 0, 9, and 12 within Control and High Se calves ( Fig 4A). There were no shifts in major bacteria between treatments, as indicated by non-significant treatment differences in weighted UniFrac distances (ANOSIM R weighted = +0.009, P = 0.32). Treatment groups did not differ in nasal microbiome at week 0 (ANOSIM R unweighted = +0.008, R weighted = -0.011, both P > 0.10) and week 12 (ANOSIM R unweighted = -0.081, R weighted = -0.042, both P > 0.10).
The microbial diversity of less abundant bacteria taxa increased between High-Se calves during the Se-treatment period, as indicated by the unweighted UniFrac distances. Control calves caught up with High-Se calves in microbial diversity in the feedlot (Fig 4; Table 3). There were shifts in major bacteria taxa across time, but not between treatments, as indicated by weighted UniFrac distances ( Table 3).
The nasal microbiome changed during the preconditioning period in all calves, with greater changes in less abundant bacteria taxa observed in High-Se calves compared with Control calves (Table 4). Both the transition from weaning (week 0) to the Se-supplementation preconditioning period (week 9), and the transition from the preconditioning period (week 9) to the feedlot (week 12) promoted changes in nasal microbiome, but the largest differences were observed between weaning (week 0) and feedlot (week 12) for both groups of calves.  Rarefaction analysis of 16S rRNA amplicon sequence variants (ASVs) obtained from nasal swabs after weaning (week 0), at the end of the Sesupplementation preconditioning period (week 9), and in the feedlot (week 12). During the 9 week preconditioning period, calves consumed alfalfa hay harvested from a field not fertilized with Se (Control) or harvested from a field fertilized with sodium-selenate (High-Se; application rate of 90 g Se/ha; n = 15 calves per group for Control and High-Se).

Effect of feeding weaned beef calves Se-biofortified alfalfa hay in a preconditioning program on nasal microbiota
https://doi.org/10.1371/journal.pone.0242771.g003 Table 2. Effect of feeding Se-biofortified hay in a 9-week preconditioning period on nasal microbiome diversity within calves.  contains genera of the Mycoplasmataceae family, increased dramatically after calves entered the feedlot. Some calves had over 50% Tenericutes. The phylum most negatively impacted by the increase in Tenericutes was Bacteroidetes.

Alpha-diversity Measures
The number of nasal microbiota phyla within High-Se calves increased during the Se-treatment period (Table 5). Five bacterial phyla were present only in High-Se calves, which were Acidobacteria, Chlamydiae, OP11, TM7, and WPS-2. Control calves caught up with High-Se calves in phyla diversity after they entered the feedlot; however, Chlamydiae, TM7, and WPS-2 remained present only in High-Se calves.
For the statistical analysis, we partitioned bacterial phyla into those that were detected in at least 14 of 15 calves per group across treatment and time ( Table 6) and those that were detected only in a subset of calves (Table 7). Bacterial taxa present in Control and High-Se treatment calves at � 1% mean relative abundance in Control calves are shown in S1 Table. There were minor effects of feeding Se-biofortified hay on the consistently present phyla, which were present in descending order of prevalence as follows: Firmicutes, Proteobacteria, Bacteroidetes, Actinobacteria, Tenericutes, Euryarchaeota, and Verrucomicrobia ( Table 6). The proportion of Tenericutes increased from weaning to the feedlot, and was accelerated by feeding Se-biofortified hay in the preconditioning period. There was a statistical trend (P = 0.08) for higher proportions of Verrucomicrobia after feeding Se-biofortified hay in the preconditioning period. Table 3. Effect of feeding Se-biofortified hay in a 9-week preconditioning period on nasal microbiome diversity within treatment groups.

PLOS ONE
The major effects of feeding Se-biofortified hay was on the minor bacterial phyla (Table 7). Feeding Se-biofortified hay promoted the presence of minor bacterial phyla. Five bacterial phyla: Acidobacteria, Chlamydiae, OP11, TM7, and WPS-2, were only present in High-Se calves in week 9. In addition, Gemmatimonadetes and Other Bacteria were present in more High-Se than Control calves.
We further queried on the genera level and detected a total of 322 bacterial genera in nasal swabs. The number of nasal microbiota genera within High-Se calves increased during the Se-  Eighteen bacterial genera were present only in High-Se calves in week 9, and in both treatment groups in the feedlot: 5 Bacteroidetes genera; 7 Firmicutes genera, 3 Proteobacteria genera, 2 Actinomycetales genera, and one Tenericutes genus. The Tenericutes genus Ureaplasma of the Mycoplasmataceae family was especially intriguing (Fig 6). At weaning (week 0), Ureaplasma was only detectable at low proportions in one High-Se calf. In week 9, 10 High-Se calves and no Control calves had Ureaplasma (P = 0.0002). In the feedlot, all calves had Ureaplasma and the proportions had dramatically increased: Ureaplasma was the sixth most abundant bacterial genus in Control calves and the third most abundant genus in High-Se calves with no treatment differences observed. All three detected genera of the Mycoplasmataceae family (genus Mycoplasma, Ureaplasma, and Other Mycoplasmataceae) increased dramatically after calves entered the feedlot (Fig 6). Similar changes were not observed in the other 5 detected genera of the phylum Tenericutes (results not shown). In the feedlot, Mycoplasma was the most abundant bacterial genus in both treatment groups with some calves having over 50% of their nasal microbiota being Mycoplasma (Fig 6). In contrast to Ureaplasma, some calves had at weeks 0 and 9 high proportions of Mycoplasma and Other Mycoplasmataceae already at weeks 0 and 9 (Fig 6). Furthermore, proportions of Mycoplasma and Other Mycoplasmataceae were not significantly altered by feeding Se-biofortified hay.

Table 7. Effect of feeding Se-biofortified hay in a 9-week preconditioning period on the presence, and number of calves harboring per Control (C) and High-Se (Se) treatment groups, of the minor nasal microbiota phyla � .
Week 0 Week

Effect of feeding weaned beef calves Se-biofortified alfalfa hay in a preconditioning program on nasal functional microbiome
We detected a total of 452 different predicted metabolic pathways in nasal swabs using PICRUSt2. The effect of feeding weaned beef calves Se-biofortified alfalfa hay was smaller on nasal microbial metabolism compared with its effect on nasal microbiota or total microbiome. There was a minor increase in the number of predicted metabolic pathways in response to feeding Se-biofortified hay (Table 9). Control calves caught up with High-Se calves in the number of predicted metabolic pathways in the feedlot (Table 9). Pathways present in Control and High-Se treatment calves at � 1% mean relative abundance in Control calves are shown in S2 Table. The primary effect of feeding Se-biofortified alfalfa hay was in less abundant predicted metabolic pathways. Eight predicted metabolic pathways were only present in High-Se calves in week 9, whereas no predicted metabolic pathway was only present in Control calves (results not shown). Thirteen predicted metabolic pathways were present in greater numbers of High-Se calves compared with Control calves (week 9), whereas no predicted metabolic pathways were present in greater numbers of Control calves compared with High-Se calves. Of the remaining five significantly different predicted metabolic pathways, two were higher in High-Se calves compared with Control calves and three were lower.
The principal coordinates of Bray Curtis Dissimilarities for predicted metabolic pathways did not differ between Control and High-Se calves at weeks 0, 9, and 12 (Fig 7). The ANOSIM values for treatment differences were R = -0.02 (P = 0.60) and R = -0.02 (P = 0.71) at weeks 0 and 9, respectively. There was a significant effect of time on nasal microbial metabolism, as indicated by PCoA plots of Bray Curtis Dissimilarities (Fig 7) and pairwise ANOSIM tests of Bray Curtis Dissimilarities across times (all P = 0.001; results not shown).

Discussion
Fertilizing alfalfa with sodium selenite at 90 g Se/ha increased the Se content of alfalfa hay over 50 fold. Feeding this Se-biofortified alfalfa hay to calves during the preconditioning period increased WB-Se concentrations almost 300%, improved growth and carcass weights, increased overall scores for slaughter yield grade, and enriched the nasal microbial diversity. There was no effect on parasite load because fecal egg counts remained low throughout.

Selenium intake
We have previously shown a linear relationship between Se fertilizer application rate and forage Se concentration [5,6,33,34]. Our results suggest that Se application rate is the primary determinant of Se content for common forage species. Fertilizing the alfalfa hay field with sodium selenite at 90 g Se/ha resulted in Se content of alfalfa in this study of 3.47 mg Se/kg DM, which was within our previous range of 3.26 mg Se/kg DM [6] to 5.17 mg Se/kg DM [35] at this fertilization rate. This is more than the nutritional requirements of Se in cattle, which are conservatively estimated at 0.1 mg/kg DM [36]. At 3.47 mg Se/kg DM forage Se concentration, calculated Se intake was 26.8 mg Se/calf per day, and represented 98% of the total dietary Se intake. Compared with controls, calves consumed 25× more organic Se with agronomic biofortification. Selenomethionine is the major selenocompound in forage legumes made from Se taken up from the soil [37].

Selenium status
Blood Se concentrations were increasing throughout the 9-week preconditioning period, and remained high relative to control calves after three weeks in the feedlot. These results are consistent with our previous studies. Consumption of Se-biofortified alfalfa hay in this study resulted in WB-Se concentrations after 9 weeks of 556 ± 11 ng/mL, which was consistent with our previous range of 278 ± 7 ng/mL (7-weeks preconditioning period) [6] to 494 ± 11 ng/mL (8 weeks preconditioning period) [35] when fed Se-biofortified alfalfa hay for shorter time periods. An advantage of feeding Se-biofortified forage is that all calves consume the Se treatment such that variation in WB-Se concentrations between calves is small.

Growth characteristics and carcass data
Consumption of Se-fertilized alfalfa hay resulted in increased BW in weaned beef calves and tended to increase hot carcass weight (P = 0.07), similar to what we previously reported [6,9]. In this study, similar to our previous study [9], carcass quality grade levels were not different between control and Se-treatment groups of calves, likely because most calves were choice grade. However yield grades in this study were improved in calves fed Se-biofortified alfalfa hay, unlike in our previous study [9]. More calves in the Se-treatment group (86%) had a 1 (most desirable trim) or 2 (industry average) compared with calves in the control group (36%).
Our results contrast to those reported by Swecker et al. [38] whereby parenteral Se and vitamin E supplementation of weaned beef calves at 0 and 28 days did not affect calf performance during the 42-day preconditioning period. It is likely that the impact and results of Se supplementation depend on Se status of calves (deficient vs adequate), type (organic vs inorganic) and amount of Se administered, and route of Se administration (oral vs injectable) [39]. In this study, third cutting alfalfa hay was enriched with Se by mixing inorganic sodium-selenate with water and spraying it onto the soil and foliage of an alfalfa field at an application rate of 90 g Se/ha. In our previous study [6,9], using similarly enriched second cutting alfalfa hay, alfalfa yield was approximately 4490 kg/ha. Each calf consumed an average of 227 kg of alfalfa during the preconditioning period. We calculated the commercial cost for application of 90 g Se/ha using Selcote Ultra 1 (10 g Se/kg as 1:3 sodium and barium selenate, respectively) to an alfalfa field as $158/ha; thus a calculated cost of $8/head for the Se enriched alfalfa hay consumed during the preconditioning period. Market prices for slaughter calves are currently approximately $3.54/kg hot carcass weight. Applying these market prices to the gain measured in our former study, feeding calves high Se enriched alfalfa hay resulted in an increase of $166/ head over that of the control group of calves. In the current study, at the end of the feedlot period, calves in the high Se group had a 9 kg higher slaughter weight compared with control calves resulting in an increase of $32/head over that of control calves. Thus, it is economically justified to recommend feeding Se-biofortified alfalfa hay during the preconditioning period to beef calves.

Fecal parasite counts
Although fecal parasite counts were assessed at baseline and weeks 5 and 9 of the preconditioning period, and again 3 weeks after entering the feedlot, no significant treatment effects were observed. This was likely because fecal parasite counts were very low throughout the preconditioning period, and essentially zero 3 weeks after entering the feedlot because all calves were dewormed at entry to the feedlot. We had previously shown in sheep that supranutritional Seyeast supplementation may enhance resistance to naturally occurring H. contortus gastrointestinal parasites [40]. Younger ruminants are reported to have immunological hyporesponsiveness and, thus, lower resistance to infections disease, which in calves manifests as respiratory and intestinal infections (viral and bacterial) [41]. In this study, because fecal egg counts remained low throughout, we were unable to test the effect of Se treatment on gastrointestinal parasite infestations in these calves.

Nasal microbiome
The microbiome refers to the collection of genomes from all the microorganisms in the nasopharyngeal environment [42]. In the present study, the nasal microbiome was similar at weaning for Control calves and calves slated for High-Se treatment, as one would expect given that calves from both groups came from the same ranch. During the 9-week preconditioning period, dietary Se-biofortification diversified the nasal microbiome by enriching the number of less abundant nasopharyngeal microbial genomes, evidenced by significant differences in observed ASVs and Chao1 indices. The observed ASVs represent the number of true RNA sequences present in the nasopharynx of each calf and Chao1 is a nonparametric estimate of the minimal number of RNA sequences present [28,29]. Thus, ASV and Chao1 estimate the number of different RNA sequences (richness). The Shannon index is also a measure of alpha diversity. It takes into account not only the number of different RNA sequences, but also the relative abundance of the different RNA sequences (evenness) [29]. Evenness compares the similarity of the population size of each of the species present. Shannon index is significant when both the number of RNA sequences increase as well as the proportion of the major sequences become more equal. Neither High-Se treatment nor transition to the feedlot increased Shannon index, i.e., evenness in our study. Thus, High-Se treatment resulted in a higher number of different RNA sequences, but did not make the different RNA sequences more similar in amounts. We previously reported [35] that increasing Se-status during a preconditioning period resulted in treatment group difference in nasal microbiota. The novel contribution of the present study is that Se-biofortification promoted diversification of the nasal microbiome by allowing the establishment of minor phyla (5 new phyla) and genera (56 new genera) and their unique metabolism pathways.
Beta diversity measured using the phylogeny-based unweighted and weighted UniFrac distances, showed that calves receiving the High-Se-treatment clustered separately from Control calves, although clustering was not significant for weighted UniFrac distances. Beta diversity measures differences between nasopharyngeal samples/calves [31]. Unweighted UniFrac distance measures differences in number of ASVs present, whereas the weighted UniFrac distance measures not only differences in number of ASVs present, but also relative abundance of the ASVs [30]. UniFrac distances measure the proportion of a phylogenetic tree that is not shared between two measures [32]. The more unshared the genome, the bigger the differences. Principal coordinate analysis calculates distances between all the samples and displays the distances in a 2 or 3 dimensional space [31]. Diversity of the less abundant bacteria increased between Control and High-Se calves during the Se-treatment period as indicated by the significantly different unweighted UniFrac distances. Control calves caught up with the High-Se calves in genomic diversity in the feedlot. None-the-less the largest differences were observed between weaning and feedlot in both groups of calves.
Several studies have examined how some management practices decrease diversity within the nasopharyngeal bacterial communities and reported that weaning, transportation methods, on-arrival processing at the feedlot, time in the feedlot [43] and antimicrobial usage [44] decrease nasal microbiome diversity. Those results suggest that stressful events may decrease diversity within nasopharyngeal bacterial communities. However, diversity changes in nasal microbiome differs amongst operations, groups, and time points (spring processing, arrival at feedlot, 40 days in the feedlot), suggesting that the respiratory microbiota of calves may lack a common pattern of evolution from ranch to feedlot [45]. The results in the present study are in line with those findings, as we observed a change over time (weaning, preconditioning period, and feedlot) in nasal microbiome diversity, which differed depending on Sebiofortification.
Another important factor that decreases nasal microbiome diversity is respiratory health status. Alpha-diversity indices revealed a lower bacterial diversity in the nasopharynx of steers with bronchopneumonia compared to healthy pen-mates [46,47]. Bacterial communities in the nasopharynx of cattle that remained healthy during the feeding period were more diverse than those that developed BRD [1], suggesting that management practices that increase microbial diversity may aid in prevention of respiratory diseases in the feedlot, which are especially common on entry to the feedlot. We have shown twice now that High-Se treatment of calves in a preconditioning program was able to upregulate the diversity of the nasopharyngeal bacterial community prior to entering the feedlot, which may ameliorate a risk factor for development of BRD [35]. To date, we are not aware of other management practices that increase diversity of the nasopharyngeal bacterial community, and thereby, potentially improve health status.
The question arises how increasing the diversity of the nasopharyngeal bacterial community improves health status. Microbial biodiversity correlates to the efficiency of nutrient utilization by a bacterial community; a greater microbial diversity means greater diversity in microbial metabolism pathways, which allows for more efficient nutrient utilization. This, in turn, decreases the chances of bacterial respiratory pathogens adhering to and colonizing in the nasopharynx (reviewed in [46]). This becomes especially important when stressed calves enter the feedlot, and are exposed to a variety of new pathogens.
There is limited information how dietary Se may increase the diversity of the nasopharyngeal bacterial community. We have reported over the last decade that Se-supplementation to Se-replete livestock can prevent infectious diseases [8-10, 40, 48, 49]. We proposed that the mechanism by Se prevents infectious diseases is through selenoproteins, which improve the functional immune status of livestock [5,9,10,34,48,49]. The results of this study suggest a potential interplay between the immune system of the host and its nasopharyngeal bacterial community. We hypothesize that Se-biofortification may alter nose-associated lymphoid tissues (NALT), which include T cells, B cells, dendritic cells, and macrophages [50,51], which in turn alters nasopharyngeal bacterial communities. This mechanism has been proposed as one role the immune system plays in regulating the microbiota [52]. Alternatively, Se-biofortification may alter the amount and composition of the nasal mucus layer, which is an important nutrient source for the nasopharyngeal bacterial community [42,52,53]. These are future directions of study for our research.

Nasal microbiota
The microbiota refers to specific microorganisms found within the nasopharynx [42]. At baseline (weaning) the major nasopharynx phyla (with relative abundance >1%) were Firmicutes (35%), Proteobacteria (22%), Bacteroidetes (19%), and Actinobacteria (13%), and Tenericutes (2%). These were similar for Control calves and calves slated for High-Se treatment. We did not have baseline microbiota data in our former study [35], but in Control calves at 8 weeks, the majority of sequences were classified as Proteobacteria (44%). The predominant bacterial phyla in the current study were similar to those reported by others for calves at entry to the feedlot [54], although there is variation in phyla and genera among studies as evidenced by Tenericutes ranging from 47.4% [46] to 22.5% [47] to 11.2% [54] in different studies. We were specifically interested in the phyla Tenericutes, as Tenericutes was the only phylum with relative abundance >1% altered by Se-biofortification in our current study and in our previous study [35].
In the present study, the class Mollicutes of the phylum Tenericutes was the only bacterial class significantly altered by Se-biofortification. Mollicutes are unique in that they lack a cell wall and include the smallest bacterial species with the shortest genome sequences; as a result, Mollicutes species adhere to and sometimes enter nasal epithelial cells and rely heavily on nutrients from the host (reviewed by Parker et al. [55]). Mollicutes bacteria develop a close interaction with the host and employ various mechanisms to evade elimination by both the innate and adaptive immune systems, which makes treatment more challenging [56].
Of the 7 known Mollicutes genera, time and treatment affected primarily the 3 genera of the family Mycoplasmataceae: Ureaplasma, Mycoplasma, and Other Mycoplasmataceae, all of which increased dramatically after calves entered the feedlot. In the feedlot, the predominant genera in the present study were unclassified (24%), Mycoplasma (16%), Moraxella 6%, and Ureaplasma (6%). We did not see high numbers of Acinobacter as reported by others [54]. A similar increase in the family Mycoplasmataceae has been reported previously after calves entered the feedlot [43], which can be prevented by antibiotic treatment with macrolides that interfere with microbial RNA replication [44]. Unique to the present study was the dramatic increase in Ureaplasma after calves entered the feedlot and the effect of Se-biofortification on the presence of Ureaplasma during the preconditioning period.
Ureaplasma and Mycoplasma are very similar in structure and function, and both are linked to a variety of inflammatory diseases in cattle; their differences are in their colony formation structure, the fact that only Ureaplasma can secrete urease, and in the number of species they contain [57,58]. In contrast to over 120 identified Mycoplasma species, there are only 11 identified bovine Ureaplasma species, the best known species being Ureaplasma parvum, which is prevalent in many herds and linked to a high incidence of abortions (i.e., bovine ureaplasmosis) [57,59,60]. Distinct bacterial metacommunities inhabit the upper and lower respiratory tracts of healthy feedlot cattle and those diagnosed with bronchopneumonia [61]. Ureaplasma parvum has been identified in the nose and the lung of cattle and its increased presence has been linked with Mycoplasma to BRD [46, 62,63]. Similar to Mycoplasma species, Ureaplasma species are usually benign; however, their ability to secrete toxins and induce a pro-inflammatory response in the host can cause extensive epithelial tissue damage [64,65]. Moreover, a large number of Ureaplasma or Mycoplasma species could decrease feed efficiency because of their nutrient dependency on the host. The question arises why Ureaplasma was present in 2/3 of High-Se calves, but none of the Control calves at the end of the preconditioning period. One hypothesis is that Ureaplasma species are able to evade detection by the immune system of the host by hiding in epithelial cells [64], which gives this bacterium an edge when invading ecological niches in hosts supplemented with immunomodulatory Se.
In the present study, Mycoplasma was the most abundant bacterial genus in both Control and High-Se treatment calves in the feedlot (over 50% of microbiota in some calves). In contrast to Ureaplasma, some Control and High-Se treatment calves had high proportions of Mycoplasma at baseline and 9 weeks. Others have reported a large increase in Mycoplasma after arrival at the feedlot [45], which they considered to be unrelated to the use of antimicrobials [47]. The genus Mycoplasma contains over 120 species and their differentiation is challenging [55,66,67]. Because Mycoplasma species differ in pathogenicity, polymerase chain reaction (PCR) techniques are the standard method for Mycoplasma diagnosis [55]. In the present study, we did not speciate Mycoplasma using a ASVs approach with DADA2, because assignment to specific variants are based on small differences in ASVs [61] and only differentiate a few Mycoplasma species [46]. We also did not speciate Mycoplasma by PCR techniques so we cannot further characterize the increase in Mycoplasma species.
Bovine respiratory disease is one of the most common causes of health and economic losses in the feedlot and closely linked to an animal's commensal bacteria populations in the respiratory tract including Mycoplasma [68,69]. Several Mycoplasma species are opportunistic pathogens present in symptomatic and asymptomatic cattle with Mycoplasma bovis being the most pathogenic [47, 55,69]. A recent study reported a shared metacommunity between the lung and nasopharynx for Mycoplasma species and suggested the nasopharynx as the best target for management strategies against BRD [61], as it could prevent for example Mycoplasma adherence.
One interesting feature in the present study was the effect of feeding Se-biofortified hay on the number of minor phyla and genera present at the end of the Se-supplementation period, which persisted in the feedlot. After feeding Se-biofortified hay for 9 weeks, the number of phyla within High-Se calves increased compared with control calves from 13 to 19 phyla, whereas Control calves had no change in number of phyla compared with weaning (13 phyla).
The increase in phyla numbers in High-Se calves was the result of increased numbers of the minor bacterial phyla (phyla observed in 13 or fewer of Control and High-Se calves). Five bacterial phyla were present only in High-Se calves at 9 weeks. Although Control calves caught up with High-Se calves in phyla diversity after they entered the feedlot (increased from 13 to 16 phyla), three of the minor bacterial phyla were still only present in the High-Se calves. Similarly, the number of microbiota genera within High-Se calves increased during the High-Se treatment period. A total of 56 bacterial genera (>90% in the major phyla) were present only after High-Se treatment, whereas only 10 genera were present solely in Control calves at 9 weeks. Again, Control calves caught up with High-Se calves in genera diversity after they entered the feedlot. None-the-less, 31 bacterial genera (>85% in the major phyla) remained present only in High-Se calves in the feedlot. Coincidently, WB-Se concentrations remained higher in High-Se treated calves after 3 weeks in the feedlot. We propose that the increased diversity of nasal microbiota present after 9 weeks of consuming Se-fortified alfalfa, and still present 3 weeks after entering the feedlot, may provide a bacterial community component for potentially regulating bacterial overgrowth and respiratory disease caused by Mycoplasmataceae.

Nasal microbiome predicted metabolic pathways
The effect of feeding weaned beef calves Se-biofortified alfalfa hay was smaller on nasal microbial metabolism compared with its effect on the nasal microbiome and microbiota. There was a small increase in the number of predicted metabolic pathways in response to feeding Se-biofortified hay, but Control calves caught up with High-Se calves in the number of predicted metabolic pathways in the feedlot. Again, the primary effect of feeding Se-biofortified alfalfa hay was in the less abundant predicted metabolic pathways.
Bray Curtis dissimilarities measures dissimilarities between samples with no difference being zero and maximum differences being one; this measure only considers differences in abundances and does not account for phylogenetic tree differences [32]. Because High-Se treatment primarily affected the presence of minor phyla, the unweighted Unifrac measures better captured this effect in our study than Bray Curtis Dissimilarities. Thus, Bray Curtis Dissimilarities for predicted metabolic pathways did not differ between Control and High-Se calves at baseline, 9 or 12 weeks. However, there was a significant effect of time on nasal microbial metabolism, and Bray Curtis Dissimilarities were different across time.
These results suggest that the nasopharyngeal microbiota in weaned beef calves can be modified by feeding Se-biofortified alfalfa hay in a preconditioning period prior to entering the feedlot. The nasopharyngeal microbiota is important for overall respiratory health, especially when stressors such as weaning, transportation, and feed adaptation impair the normal respiratory defenses (reviewed in [70]). Because cattle diagnosed with respiratory disease had less bacterial diversity and fewer ASVs at feedlot entry [1], reduced diversity may be a risk factor to developing respiratory disease. Our results from this study, add to those previously reported [35] by following calves into the feedlot, and support feeding Se-biofortified alfalfa hay to weaned beef calves prior to entering the feedlot as an effective strategy for increasing nasopharyngeal microbial diversity.

Conclusions
In summary, Se fertilization of alfalfa fields in a region with Se-deficient soils increases Se content of alfalfa hay. We have now shown in two independent trials that feeding recently weaned beef calves Se-biofortified alfalfa hay during the preconditioning period is an effective management strategy to build Se-body reserves, optimize growth prior to entering the feedlot, and improve feedlot performance as evidenced by higher slaughter yield grades. The novel finding of the present study is that feeding recently weaned beef calves Se-biofortified alfalfa hay during the preconditioning period promotes nasal microbiome and microbiota diversity, which in turn may provide health benefits by ameliorating a risk factor for BRD.
Supporting information S1 Table. Bacterial taxa present at � 1% mean relative abundance in Control calves (n = 15) and High-Se treatment calves (n = 15) after weaning (baseline; week 0), at the end of the Se-supplementation preconditioning period (week 9), and in the feedlot (week 12). Mean and median relative percentages of the most abundant bacterial groups are shown, annotated to the level of phylum, family, and genus, based on sequencing of the 16S rRNA. (XLSX) S2 Table. Functional metabolism pathways present at � 0.5% mean relative abundance in Control calves (n = 15) and High-Se treatment calves (n = 15) after weaning (baseline; week 0), at the end of the Se-supplementation preconditioning period (week 9), and in the feedlot (week 12). Mean and median relative percentages of the most abundant functional metabolism pathways based on sequencing of the 16S rRNA are shown. Bacterial taxa in bold significantly differ between treatment groups. (XLSX)