There is a growing body of literature linking mitochondrial traits to health and performance outcomes due to their dynamic involvement in metabolism, energy production and apoptosis pathways. Despite mitochondria’s biologically active role, little research has been done exploring the effects of mitochondrial function on dairy cow performance due to the invasiveness of traditional methods utilizing liver tissues. The use of peripheral blood mononuclear cells (PBMC) offers a minimally invasive and high throughput model tissue to measure mitochondrial function in commercial dairy herds. Dairy cows face great metabolic pressure to meet the demands of rapid growth, reproduction, and lactation. A greater understanding of mitochondrial traits associated with cow performance could improve dairy farming by allowing for novel selection criteria of highly productive long-living cows. The selection and breeding of high merit cattle would improve profitability and sustainability of the industry by facilitating a reduction of cow numbers without reducing milk production and simultaneously reduce environmental impacts of dairy farming through the reduction of manure, methane, and nitrogen excretion. Therefore, the objectives of this dissertation were to determine if PBMC mitochondrial enzyme activities of citrate synthase, complex I, complex IV and complex V were 1. Different between high and low producing dairy cows 2. Associated with age, growth, reproduction, milk production and survival 3. Impacted by physiological and environmental conditions across multiple dairy farms.

Dairy cows are managed and fed in pens. These pens are constructed for uniform handling and management, and cluster cows on stage of lactation, reproduction status, or milk production. The variation in nutrient requirement of the cows is not accounted for, and different distributions of dry matter intake (DMI) occur depending on cow selection. These pens are fed by a ration supplied collectively. This ration is solved to meet a nutrient requirement and scaled to the number of cows in the given pen. But the solution cannot account for within pen variation because DMI is not individually recorded. The common measurement is the average nutrient or DMI per cow, dividing the quantity of ration consumed by the number of cows present. Another approach is predicting a DMI value for a hypothetical mean cow to represent the pen. Both systems result in one DMI value, rather than a distribution. The only scenario where this would be considered precise or accurate nutrition is if the cows within the pen were completely uniform in their requirements. Failing this, if the distribution of the pen were normal, and the tails symmetrical, the lower requirement cows could have their feed consumed by the higher requirement cows, striking a crude balance.

The first chapter of this work examined this assumption. Its hypothesis was the DMI of cow pens are not normally distributed, and the total pen DMI can be better calculated by describing it with an appropriate distribution shape. Cow DMI was first described by week of lactation for skew and kurtosis, then a large dataset of cow pens representing Fresh, High, and Low milk yield lactation stages were assembled from individual observations. The distribution of DMI for each pen was fit to the best distribution shape, then the total pen DMI was calculated from the area under its curve. This was compared to a second model that calculated the total DMI by applying an empirical equation to the pen’s mean values. The Beta distribution shape was the most common best fit of all pens, and the distribution shape predicted pen DMI with less than 1 % error, and more accurately than the comparison model, with an error of 11 – 22 %.

The second chapter predicted the distribution shape of DMI. In chapter I the distribution was demonstrated as a good calculation of pen DMI, but the shape is only known when described from individual DMI values. To build a prediction model in the case of unknown individual DMI, we trained machine learning algorithms to a known dataset. A large dataset of pens was assembled, and these were described with their best fit distribution, either the Beta or Generalized Normal distribution shape. Machine learning algorithms were trained to classify these pens to their best shape and predict DMI values that fit the distribution; the Distribution Shape Model (DSM). A second model was contrasted to this that predicted pen DMI by applying the mean descriptive statistics of the pen to an empirical equation, the NASEM equation model (NSM). Both models were validated for their DMI prediction performance on a naïve dataset and compared by model diagnostics. The DSM was very precise and accurate, with low error, or bias, and outperformed the NSM in consistently predicting pen DMI. The last chapter of this work performed precision nutrition with the distribution shape of DMI for ration formulation. Three pens were considered: Fresh, High and Low. Each pen had its DMI predicted by the DSM, and by applying an empirical equation to each cow’s characteristics, individual NASEM equation model (iNSM). These DMI values were used to formulate individual rations for every cow, and compared by cost, DMI, and metabolizable energy. As individual rations cannot be practically applied on dairies, precision nutrition pen rations were also solved. Each pen had a nutrient requirement calculated by summing its constituent cow’s requirements, and rations were solved with these totals. The DSM and the iNSM were compared to an ideal ration solved to the observed DMI values (TRU) to evaluate these two precision nutrition approaches. Two imprecise approaches were also considered. One where an empirical equation was applied to the mean of each pen, the average NASEM equation model (aNSM), and one where it was applied to the 75th of the pen, the adjustment factor NASEM equation (fNSM). In the individual, and pen ration solutions, the DSM was the only model that was not significantly different to the TRU. It was also cheaper and less nutrient wasting than all other ration models, as its solution was almost identical to the TRU for individual cows, and pen rations.

Cephalosporin antibiotics are commonly used at dry off to treat and prevent intramammary infections (IMI). Research in murine and other mammalian cell types showed that cephalosporin antibiotics negatively affect mitochondrial respiratory chain activity, which impacts cellular function. The objective of this study was to evaluate the effects of intramammary (IM) cephalosporin (Cefitiofur Hydrochloride and Cephapirin Benzathine) antibiotic treatment administered at dry off on peripheral blood mononuclear cell (PBMC) and milk mononuclear cell (MMC) mitochondrial enzyme activity 4 h prior to dry off (TPT1), 7 d post dry (TPT2), and between 55 – 75 DIM in the subsequent lactation (TPT3). Thirty-seven Holstein cows from a commercial dairy were enrolled at TPT1 and assigned to 1 of 4 treatments: 1) Low SCC control (LCON), 2) High SCC Control (HCON), 3) High SCC Ceftiofur Hydrochloride (CH), (Spectramast DC; Zoetis Inc., Kalamazoo, MI), and 4) High SCC Cephapirin Benzathine (CB), (ToMorrow; Boehringer Ingelheim Vetmedica Inc., St. Joseph, MO). Control treatments received no antibiotics. Low and high SCC cows were defined as < 100,000 cells/mL (low), and > 200,000 cells/mL (high) at TPT1. Whole blood and milk samples were collected at TPT1, TPT2, and TPT3. Enzyme activities of citrate synthase (CS), complex I (CI), complex IV (CIV), and ATP synthase were performed on crude mitochondrial extracts using kits from Abcam (Cambridge, MA). An index representing the ratio of electron transport activity to ATP synthase activity (OXPHOS) was calculated from the sum of CI and CIV enzyme activities divided by ATP synthase enzyme activity. Data were analyzed by REML ANOVA with repeated measures in R (Version 4.0.4). The PBMC CIV enzyme activity was highest at TPT1 in the LCON treatment compared with the HCON, CH, and CB treatments, but was not different at TPT2 and TPT3, indicating that PBMC CIV enzyme activity was influenced by IMI. The PBMC and MMC CS, CI, ATP synthase and OXPHOS changed with time. These data indicate that MMC and PBMC mitochondrial enzyme activities and OXPHOS were not influenced by cephalosporin treatment. The time related changes in OXPHOS and enzyme activities were likely influenced by diet composition, milk production, oxidative stress, and fetal growth.

Fibrolytic enzymes have been shown to enhance nutrient availability by increasing fiber digestibility of dairy cow rations. The objective of this study was to determine if a TMR supplemented with an exogenous fibrolytic enzyme product (EFE; RumaThrive; Wilbur-Ellis, Portland, OR) increased milk production or improved reproduction on a commercial dairy. A total of 2,990 Holstein cows were randomly assigned to 1 of 4 study groups: 1) primiparous control (PC: 2 pens) fed only a TMR; 2) primiparous EFE supplement (PT: 2 pens) fed a TMR with added EFE; 3) multiparous control (MC: 3 pens) fed only a TMR; 4) multiparous EFE (MT: 3 pens) fed a TMR with added EFE. The enzyme product had a declared minimum of 44,750 units/g of xylanase. Supplementation of EFE began 2 wk prior to parturition (pre-calving period) and continued 180 d into their lactation. Cows on PT and MT received EFE at a rate of 2 g/cow/d during the pre-calving period and 4 g/cow/d during the lactation period. Data were collected during the first 180 d of lactation, analyzed as a completely randomized design with repeated measures for milk production using the Mixed procedure of SAS (Statistical Analysis System, v.9.4) and Cox proportional hazard regression model with RStudio (v1.4.1106, RStudio: Integrated Development for R, Boston, MA) for time to conception. Primiparous and multiparous pens were analyzed separately. There was no difference in mean days of exposure to EFE between control and EFE primiparous or multiparous pens. The PC had higher milk fat yield, percentage fat, FCM and ECM compared to PT. The MT also had lower milk fat yield, FCM and higher milk protein percentage compared to MC. The PT had a 18% higher risk of conception when compared to PC pens (P ≤ 0.02). Results of this study indicate that feeding EFE to primiparous dairy cows decreased time to conception.

Some of the most common failures of dietary cation-anion difference (DCAD) TMR to prevent hypocalcemia, are a lack or dilution of acidifying minerals or inadequate mixing of the TMR. Currently, urine pH is used to monitor metabolic acidification in cows 3 wk prior to parturition. However, urine is difficult to collect and does not indicate why the close up TMR failed to acidify the cows. Mineral meters that measure solubilized mineral from the TMR could be used to approximate the amount of mineral in the TMR. The objectives of this study were to 1) evaluate how well meters predicted macromineral concentration in a TMR compared to laboratory macromineral analyses; 2) determine if the meters could estimate DCAD in TMR for close up and lactating cows; and 3) to determine if the meters could measure TMR uniformity. Meters used were Cl (Oakton SaltTestr,Oakton Instruments, Vernon Hills, IL), K (LAQUA Twin K meter, Horiba Scientific, Kyoto, Japan), S (Hanna Instruments Smithfield, RI) Ca and Mg (Hanna Instruments Ca/Mg Photometer , Smithfield, RI). The TMR samples were collected from close up and lactating cow pens at 10 commercial dairies. Ten subsamples of TMR were collected per pen at the time of feeding and sent to Analab (Agri-King Inc., Fulton, IL) for analysis. Time length of soaking each TMR subsample for each mineral was pre-determined by identifying the time of plateau of mineral concentration (Cl for 90 min, Ca for 150 min, K for 180 min and Mg for 210 min). Data were analyzed using PROC REG in SAS (SAS Institute v. 9.4, 2021), regressing lab mineral concentration on meter mineral concentration with pen as a covariate. Prediction of Cl and K concentrations in the TMR on a DM basis were best (R2 = 0.69 and R2 = 0.78, respectively). The combination of both K and Cl meter concentrations were able to predict DCAD in the TMR in mEq (R2 = 0.88), and the Cl meter was also able to predict TMR coefficient of variation (CV) (R2 = 0.89) to evaluate TMR uniformity. All regressions had slopes not different from 1, y-intercepts not different from 0, and normally distributed residual errors. Therefore, Cl and K meters can be used to predict the macromineral concentration, DCAD of TMR, and Cl or K meters can be used to predict overall TMR uniformity of a feed drop.

Intramammary infections, which cause mastitis, can increase treatment and labor costs, decrease milk production, and affect milk quality. Meters that measure quarter somatic cell count (SCC) could be used to make more informed dry cow therapy decisions. The objective of this study was to compare the RT-10 iPhone adapter (RT-10; Dairy Quality Inc., Newmarket, Canada), DeLaval Cell Counter (DSCC; DeLaval, Graiguecullen, Carlow), Porta Check Quick Test (PortaCheck, NJ), California Mastitis Test (ImmuCell, Portland, ME), pH meter (Hanna Instruments, RI), electrical conductivity meter (OHAUS, NJ), and the dual laser infrared temperature thermometer (Klein Tools, IL) for measuring SCC in individual quarters in comparison to a reference standard, the Fourier Transform Spectrometer 600 Combi System (Combi; Bentley Instruments, MN). Meters were evaluated using individual cow quarter samples and bulk-tank samples to measure SCC. To test individual quarter SCC, individual quarter milk samples from 160 cows from 4 commercial dairy herds (40 cows per herd) were collected just before cessation of milking and tested within 4 h of collection. To test bulk-tank SCC, 100 bulk-tank milk samples (25 mL) were collected from UC Davis VMTRC Milk Quality Lab. Meter SCC values were regressed on observed Combi SCC using PROC GLM (SAS Institute, 2021 v. 9.4). Then goodness of fit was evaluated by partitioning of the mean square predicted error (MSPE). For individual quarter SCC, RT-10 had the highest coefficient of determination (R2 = 0.86), lowest mean square predicted error and highest proportion of MSPE due to random variation (96 %). Both the RT-10 and DSCC had the highest sensitivity and specificity for identifying quarter SCC above and below 200,000 cells/mL. For bulk-tank SCC, DSCC had the highest coefficient of determination (R2 = 0.45), lowest mean square predicted error, and highest proportion of MSPE due to random variation (80 %). The RT-10 and DSCC could be used to measure individual quarter SCC to determine which cows to treat at cessation of lactation potentially reducing antibiotic use.KEYWORDS: mastitis, milk meter, somatic cell count

Feeding probiotics has been shown to improve the health and production characteristics of dairy calves. The objective of this trial was to evaluate the effect of a probiotic fed daily on health outcomes, average daily gain, frame growth, feed efficiency, and intake when fed to Holstein heifers from birth to 180 d. A total of 324 Holstein heifer calves from a California dairy were enrolled within 48 h of birth into two treatment groups: 1) control group (CON: 176 calves) were given 0.5 g of lactose added to milk once daily from birth to weaning (60 d) and then 0.75 g in grain from weaning to 180 d and 2) probiotic group (PRO: 153 calves) were given 0.5 g (1.1 × 1010 CFU2/g) of probiotic in milk once daily from birth to weaning and then 0.75 g (1.65 × 1010 CFU2/g) of probiotic in grain from weaning to 180 d of a B. subtilis, B. lichenformis, L. animalis, and P. freudenreichii probiotic (Bovamine Dairy Plus, Chr. Hansen, Milwaukee, WI). The lactose powder was given to CON to balance the lactose content in the probiotic treatment and to ensure no difference in the handling of milk bottles. The PRO ADG was lower during the hutch period (1 - 91 d) and higher fecal shedding of Clostridium at 21 d and 42 d collections. There were no differences in all other health and performance outcomes.