Both whey and skim milk are considered valuable sources of protein in calf milk replacer (MR). However, it is still unclear which source is more beneficial for performance of calves. This study evaluated the effects of whey-based MR and MRs containing 25% or 50% inclusion of skim milk on the growth performance of dairy calves. Ninety Holstein calves (10 d old; 57 females and 33 males; 44.9 ± 4.9 kg) were semi-randomly assigned to one of 3 groups (n = 30/group): 0% skim milk in MR (on as-fed basis; G0), 25% skim milk in MR (G25), and 50% skim milk in MR (G50). Allocation was performed to maintain a balanced distribution of calf sex and calves born to primiparous and multiparous cows among groups. Milk replacers were formulated to contain a similar concentration of protein, fat, and essential amino acids. Calves were fed 7.5 L daily of MR (1200 g of MR powder) in 3 feedings (3 × 2.5 L) and starter ad libitum. Data were analyzed using mixed models, with treatment, calf sex, and dam parity included as fixed effects, and initial BW as a covariate for BW, ADG, starter intake, and feed efficiency. Results were expressed as adjusted least squares means. When justified, means between groups were separated using Tukey's test. Passive immunity and initial BW did not differ between groups. Cumulative MR and starter intake did not differ between groups. Final BW, ADG, and feed efficiency were higher in G25 compared with G50, but did not differ between G0 and G25 or G0 and G50. Fecal scores and health treatments did not differ between groups. In the current study, growth performance of calves fed whey-based MR did not differ from those fed MR containing skim milk powder. Overall, no statistical evidence of a difference was detected between G0 and the skim-milk-containing treatments under the conditions of this study. However, calves fed MR containing 50% skim milk had 8.5% lower ADG and 4% lower final BW than those fed MR containing 25% skim milk.
Olfaction is a key sensory modality in cattle, yet the behavioral repertoire associated with odor exploration remains insufficiently described. Existing studies have primarily focused on sniffing, with limited characterization of additional behaviors that may contribute to olfactory assessment. This study aimed to describe the full range of odor exploration behaviors expressed by dairy heifers and to develop an ethogram specifically adapted for this context. The study was conducted with 21 dairy heifers, each exposed to 10 essential-oil odors in 5-min one-exposure tests performed within their home environment. This resulted in 207 analyzable tests after 3 were excluded due to disturbances during the tests. Video footage was used to identify all behaviors expressed during odor exploration. An initial ethogram based on previous literature was expanded through iterative coding of 40 tests. Behaviors were categorized as odor exploration or other (not odor) exploration using a predefined sniffing criterion. A randomized subsample was used to evaluate intra-observer reliability, and the refined ethogram was applied to the full data set to characterize behavioral distributions. The ethogram comprised a structured and comprehensive set of odor-directed and other exploratory behaviors, including sniffing, scratching, licking or biting, head shaking, head butting, mouth opening, tongue movements, vigilance, kicking, elimination as well as ear and tail postures. Intra- and inter-observer reliability showed moderate to excellent agreement, indicating that the behavioral definitions were sufficiently clear and operational to be applied consistently. This supports the usefulness of the ethogram for future research and its potential to be adapted to other study designs. This study provides a systematically developed and empirically evaluated ethogram describing the behavioral repertoire of dairy heifers during odor exploration. The ethogram expands the methodological tools available for cattle olfaction research and offers a standardized framework for future studies on sensory processing and odor-driven exploratory behavior.
Heat waves (HWs) are increasing in frequency and severity in many regions under climate change, posing challenges to the welfare and productivity of dairy cows. Physiological and environmental indicators are widely used to assess heat stress (HS) in dairy cows; however, the short-term temporal dynamics and lagged responses of molecular stress biomarkers during HW events under production conditions remain poorly characterized. This study aimed to evaluate daily variation in milk HSP70 during a HW and to assess its association with the environmental temperature-humidity index (THI) and sensor-derived reticulorumen temperature (RRT). The research was carried out over 5 consecutive days of HW in a pasture-based dairy farm in Camden, New South Wales, Australia, investigating 20 clinically healthy, third-parity Holstein Friesian cows. Daily milk samples were obtained during afternoon milking for HSP70 assessment using competitive ELISA. Environmental data were recorded at 1-min intervals to calculate THI, while RRT was continuously monitored at 10-min intervals using rumen bolus sensors. Associations between THI, RRT, and milk HSP70 were analyzed using repeated-measures correlations and univariable linear mixed models across lag windows of up to 50 h before sample collection, with adjustment for multiple testing across lag intervals. Exploratory classification performance was evaluated using receiver operating characteristic (ROC) analysis. The results revealed delayed associations between milk HSP70 and both THI and RRT. Associations between milk HSP70 and THI were strongest within an approximate delayed response window of 40-50 h after heat exposure, with repeated-measures correlations ranging from 0.57 to 0.84. Corresponding mixed-model analyses remained significant after adjustment for multiple testing using the Benjamini-Hochberg false discovery rate (FDR) procedure (FDR-adjusted P < 0.001), with R2c values ranging from 43.79% to 75.66%. Similarly, associations with RRT were strongest within a delayed response window of approximately 48-50 h after heat exposure, with repeated-measures correlations ranging from 0.51 to 0.62. Corresponding mixed-model analyses remained significant (FDR-adjusted P < 0.001), with R2c values ranging from 40.92% to 51.71%. In contrast, RRT responded rapidly to environmental heat load, showing the strongest associations with THI within the immediate response window of approximately 0-6 h (repeated-measures r = 0.57-0.67), while the corresponding mixed-model analyses remained significant (FDR-adjusted P < 0.001; R2c = 48.15-57.21%). An exploratory milk HSP70 cut-point of approximately 550 ng/mL was identified, with a sensitivity of 0.782 and specificity of 0.909 for RRT-defined HS. Overall, the findings indicate that milk HSP70 exhibits a delayed response pattern following HW exposure, consistent with cumulative cellular stress, and may provide supplementary insights alongside environmental and physiological indicators for evaluating prolonged HS in dairy cows in pasture-based systems.
The genetic evaluation of female fertility traits in dairy cattle in the United States has progressed over the past 2 decades, with 4 additional traits integrated into the national evaluation system since the introduction of daughter pregnancy rate (DPR) in 2004. However, concerns have arisen in the dairy sector, with reports of producers extending the voluntary waiting period (VWP), which is the time of initial breeding after calving, due to more persistent lactation yields. Since its inception, DPR calculations in the US evaluation have assumed a fixed 50-d VWP, which may not reflect modern reproductive strategies. Furthermore, producers may determine that some of their cows should have their VWP extended while others may follow the standard time. A re-evaluation of the current female fertility evaluation is necessary to ensure traits align with current management practices. Therefore, this study explores the addition of a potential new trait, First Service to Conception (FSC), along with a revised DPR formula that permits more flexibility with VWP. About 32 million records from the National Cooperator Database, covering 5 major dairy breeds (2003-2023), were used for this study. Data for cows calving before 2003 were unavailable because insemination records were not recorded prior. The findings suggest FSC enhances female fertility evaluations, providing a more comprehensive reproductive assessment independent of VWP. Another approach was an adjustment to the DPR calculation (DPRadj) to account for herd VWP on a herd-year and lactation group basis. The mean DPR value was 45.06%, while DPRadj increased the mean value to 52.58%. The mean FSC was 62.64 d. In an analysis using first lactation cows for cow traits and heifers for heifer traits, genetic correlations of FSC with other fertility traits were 0.97 with Cow Conception Rate (CCR) and DPR, 0.98 with DPRadj, 0.40 with Heifer Conception Rate (HCR), and 0.32 with Early First Calving (EFC). Predicted transmitting abilities (PTA) were calculated to assess the implications of adding FSC to, or replacing DPR with DPRadj within the CDCB multi-trait evaluation index, Net Merit $ (NM$). The highest correlation among the top 10% of bulls occurred when replacing DPR with DPRadj (0.99). Adding FSC to the original model reduced the latter correlation to 0.88, which indicates slight re-ranking of top bulls. Among animals with PTA reliability above 50%, the lowest correlation was between the original index and the index replacing DPR with FSC (0.95). This work advances genetic evaluation in dairy cattle, improving reproductive efficiency and productivity.
This longitudinal observational study investigated the impact of somatic cell count (lnSCC) on milk composition and technological traits in 332 clinically healthy cows. Over 2 years, 865 samples were collected at 4 lactation stages. The analysis covered milk yield, major components (fat, protein, lactose, FFA, citric acid, urea contents), and processing-related traits (casein and TS contents, acidity, pH, coagulation time, and curd/whey quality and curd yield). The mixed model included cow as a random effect, while HYS, HYM, parity, and milking time were fixed effects. DIM (quadratic) and lnSCC (linear, quadratic, cubic) served as fixed covariates.Results showed that lnSCC significantly affected 20 out of 26 studied traits, often through nonlinear relationships. While total protein and casein concentrations initially rose, their percentages peaked at approximately 1,187,000 and 348,000 cells/mL, respectively, before declining. The casein-to-protein ratio reached its maximum at only 37,000 cells/mL. A cubic effect was a significant predictor for fat content, titratable acidity, and curd/whey quality. Notably, curd quality began to deteriorate at a threshold as low as 22,600 cells/mL, while total solids (TS) contents reached a maximum at 385,000 cells/mL. Beyond these physiological thresholds, the technological value of milk dropped sharply, alongside a significant linear decline in milk yield and citric acid. These findings demonstrate that early alterations in milk integrity occur well below traditional regulatory limits; however, these values represent model-derived mathematical estimates that require independent validation before any practical diagnostic or regulatory application.
Meteorological and physiological stressors significantly impair dairy farm productivity and profitability. Although their impacts on individual cows have been extensively analyzed, macroscopic herd-level dynamics require further investigation to inform farm-wide decision-making. Therefore, the main objective of this study was to quantify the nonlinear effects of various stressors on herd-level performance and identify the key drivers of herd productivity and variability from a population perspective. We collected high-frequency milk yield records from 45,148 Holstein cows across 5 large-scale commercial farms from 2020 to 2024. Using the herd-day as the evaluation unit, the cow-level records were aggregated into herd-level features (the mean and standard deviation of milk yields and milk losses across all cows per herd-day). By integrating statistical analyses, causal inference modeling, and predictive machine learning algorithms, this study evaluated the nonlinear impact of the temperature-humidity index (THI), daily variation of THI, incidences of 5 diseases (udder health; reproductive, metabolic, and digestive disorders; and hoof health), and the proportions of estrus and artificial insemination (AI) on herd-level features. From a population perspective, increased THI and specific disease incidences were identified as drivers of herd-level milk losses. The herd-level features showed robust stability during periods of low meteorological and physiological stresses. When critical thresholds were exceeded (7-d average THI >68.4, 7-d average incidence of udder health >0.36%, or 7-d average incidence of metabolic disorders >1.30%), the rapid increases in herd-level milk losses were attributable to these stressors, culminating in maximum increases of 1.61%, 1.59%, and 2.90%, respectively. Crucially, elevated disease incidence destabilized the herd, driving significant increases in herd-level variability of both milk yield and milk losses, thereby widening the production gap between healthy and stressed subpopulations. Furthermore, the XGBoost models captured the complex nonlinear dynamics for all herd-level features, explaining 89.3% to 95.2% of the variance for milk yield features and 52.4% to 62.3% for milk loss features. The 7-d average THI and the 7-d average incidence of udder health were the primary risk factors impacting the herd-level milk yield and milk losses. Notably, the 7-d average udder health incidence contributed significantly to all milk loss features, accounting for 8.98% to 17.87% of the total feature importance. Overall, this study highlights the adverse impacts of various stressors on herd-level milk yield and milk loss features, facilitating the improvement of population-centric targeted management practices to mitigate milk losses and enhance the overall resilience of commercial dairy farms.
Subclinical mastitis, identified through elevated somatic cell count (SCC), remains the most prevalent and costly health issue in dairy cattle, yet milk lactose, biologically linked to udder health and routinely measurable by Fourier-transform infrared (FTIR) analyzers, is rarely utilized in routine herd management. We investigated which routine dairy herd improvement (DHI) variables are most consistently associated with elevated SCC, using 72,547 test-day records from 3,562 Holstein cows across 37 Korean dairy farms (2021-2025), where lactose was not available in DHI records. Four interpretable models (logistic regression, generalized additive model, extreme gradient boosting, and random forest with SHapley Additive exPlanations [SHAP]) were compared under farm-level grouped cross-validation. The solids-not-fat minus protein residual (SNF% - protein%), an unvalidated lactose proxy, was the top-ranked contributor to elevated SCC, and the top-3 ranking (residual, parity, milk yield) was consistent across random forest SHAP values, logistic regression standardized coefficients, and extreme gradient boosting SHAP values, as well as across 4 SCC thresholds, 5 lactation stages, and 5 years of data. Farm-level grouped validation reduced discrimination relative to random cross-validation by an amount larger than the entire model-class difference in performance, indicating that validation design matters more than model choice. External corroboration on an independent European open data set with directly measured FTIR lactose supported the inverse lactose-SCC association and ranked directly measured lactose as the top SHAP contributor; given the small single-farm Belgian cohort, this external evidence is supportive rather than confirmatory. Because milk composition and SCC were recorded on the same test day, the results describe concurrent associations rather than prospective prediction; the positive predictive value of 0.530 also precludes stand-alone individual-cow screening.
Geographical origin may be a key factor contributing to variations in dairy product quality. This study focuses on the fatty acid composition and lipidome of camel milk and bovine milk from different regions, aiming to highlight lipid biomarkers that remain unaffected by regional variations in both camel and bovine milk, as well as region-specific lipid biomarkers present in camel milk. The results showed that camel milk consistently exhibited higher levels of long-chain fatty acids and monounsaturated fatty acids, whereas bovine milk showed a higher proportion of medium-chain and short-chain saturated fatty acids, regardless of the region. Lipidomic analysis revealed that phospholipids were more abundant in camel milk, while triglycerides and diglycerides were characteristic lipids in bovine milk. Certain lipid biomarkers were more abundant in specific types of milk, such as PC (21:4/21:6CHO), FA (29:3), and Hex3Cer (d18:0/24:2) in camel milk, and CerPE (d41:6) in bovine milk, and could be used to distinguish camel milk from bovine milk across all 3 regions. Additionally, fatty acids and lipid molecules distinguishing camel milk from different regions were identified, including C18:3n3 and PE (18:2/22:2CHO) in XJNC, C18:2n6c in XJSC, and C14:0 and PS (16:0/18:1) in IMC. These findings provide new insights into developing species-specific nutritional strategies and regional traceability of dairy products.
Physiological effects of delaying cow-calf separation until a few days after birth are of increasing interest as access to the dam makes the rich profile of immunological, nutritive, and bioactive factors of transition milk (TM) available for postnatal development of the calf. Scientific inquiry into opportunities for cow-calf contact may also inform the current cow-calf contact debate. Our objective was to compare preweaning health and growth of calves with access to TM, with or without extended dam contact. Singleton Holstein and Holstein x Limousin male and female calves (n = 81) and their primiparous and multiparous Holstein dams were enrolled at birth in a randomized block design from d 0 to 5 of life to (1) delayed separation until 5 d with unrestricted suckling of TM (URS, n = 26), (2) immediate separation after birth and offered their dam's TM (n = 27) twice daily (up to 10% BW/feeding), or (3) immediate separation and offered unpasteurized whole milk (WM, n = 28) twice daily (up to 10% BW/feeding). All calves received a single meal of maternal colostrum (8% of BW) within 4 h of calving. All calves were moved to a designated calf area on d 5, while cows remained in the experimental stalls until d 7 after which they returned to the herd. Calves were blood sampled, weighed, and assigned health scores according to a rubric (adapted from the Wisconsin Calf Health Scorer) before each morning feeding. On d 0 and 7, calf height was measured at the withers and hips. Calves that were retained as Holstein replacement heifers (n = 38) were raised according to herd protocols after d 7, and were weighed, withers and hip height measured, and health scores assessed once per week until weaning at d 63. Serum IgG concentrations were quantified by radial immunodiffusion and packed cell volume (PCV) by centrifugation. Data analysis was conducted using mixed effects models, or Pearson's chi-squared test as appropriate. Data are reported as LSM ± SEM. Serum IgG concentrations (g/L) 1 d after colostrum feeding (URS: 42.1 ± 2.0; TM: 44.6 ± 1.9; WM: 41.9 ± 1.9) were not different between groups and declined by a similar percentage from d 1 to 7 (22.7 ± 0.6%), whereas PCV (%) was different between groups (URS: 38.3 ± 1.4; TM: 38.7 ± 1.5; WM: 41.5 ± 1.4). There was no difference between groups in BW, ADG, or withers and hip heights through the first 63 d, but BW interacted with group such that URS calves had greater weight during the suckling phase only. There was no association of treatment with fever or diarrhea events from d 1 to 7, or d 7 to 63. Feeding of TM, with and without cow-calf contact, did not result in meaningful differences in growth or health indicators compared with WM feeding, however, the benefits of TM may not be fully explained by serum IgG concentration and short-term growth indicators.
The objective of this experiment was to investigate the effects of sward type on the milk production, intake, digestibility, N excretion and urination characteristics of late-lactation dairy cows offered freshly harvested herbage indoors in a cut-and-carry feeding system during autumn. Nine ruminally cannulated multiparous Holstein Friesian cows averaging (mean ± SD); 181 ± 12 DIM and 526 ± 37 kg liveweight were randomly assigned to 1 of 3 dietary treatments (DT) in a replicated 3 × 3 Latin square experimental design. The experiment included 3 29-d periods, with each period consisting of 21-d of dietary adaptation before an 8-d measurement period. The 3 DT were: 1) tetraploid perennial ryegrass receiving 50 kg of synthetic N/ha/cut (GO); 2) tetraploid perennial ryegrass-white clover receiving 25 kg of synthetic N/ha/cut (GC); and 3) tetraploid perennial ryegrass-white clover-plantain, containing 50% plantain, receiving 25 kg of synthetic N/ha/cut (GCP). All cows were also supplemented daily with 0.89 kg DM/d of a standard concentrate. There was no effect of DT on the majority of milk production outcomes; however, cows fed GCP had lower milk fat yield and tended to have lower milk fat concentration when compared with cows fed GO and GC. There was no effect of DT on DM or OM intake. Cows fed GCP had lower apparent DM, OM and N digestibility when compared with cows fed GO and GC. Urinary N concentration was lower for cows fed GCP (0.36 g/100 g) when compared with cows fed GO and GC (0.63 and 0.58 g/100 g, respectively). Urinary N excretion was greatest for cows fed GO (355 g/d), intermediate for cows fed GC (323 g/d), and lowest for cows fed GCP (267 g/d). Cows fed GCP had greater fecal N output when compared with cows fed GO and GC. Cows fed GCP had lower urinary N output as a proportion of N intake when compared with cows fed GO and GC, whereas cows fed GCP had greater fecal N output as a proportion of N intake when compared with cows fed GO and GC. There was no effect of DT on the concentration of N in milk or milk N output. Cows fed GCP had greater urine output and greater herbage water intake when compared with cows fed GO and GC. Lastly, when compared with cows fed GO and GC, cows fed GCP had a greater number of urinations per day (15.1, 14.9 and 19.3 events/d, respectively) and greater urine event volume (3.7, 3.7 and 4.2 L/event, respectively). Overall, these results support the inclusion of plantain in perennial ryegrass-white clover swards as a practical and effective strategy to reduce urinary N excretion in pasture-based dairy systems.
Parity is a major determinant of transition-cow metabolic load and disease susceptibility, yet hyperketonemia screening commonly applies uniform β-hydroxybutyrate (BHB) cut-points regardless of parity. The objective of this study was to determine whether hyperketonemia-associated metabolomic remodeling in plasma and milk differs among early-lactation Holstein cows of different parities. Cows at 5 to 7 d in milk without apparent clinical disease at routine farm observation were stratified by parity [1 (P1), 2 (P2), or 3 (P3)] and by whole-blood BHB status as control (CON; BHB <1.2 mmol/L), subclinical hyperketonemia (SCK; 1.2 ≤ BHB ≤3.0 mmol/L), or biochemical hyperketonemia (HK; BHB >3.0 mmol/L), resulting in 9 groups (n = 9 per group; n = 81 total). Paired plasma (morning) and milk (same day) were profiled using untargeted LC-MS/MS metabolomics. Within each parity, differential analyses across SCK vs. CON, HK vs. CON, and HK vs. SCK were combined with parity × status interaction screening to quantify parity-conditioned responses and to summarize stage-to-stage trajectories (e.g., early, late, progressive, or reversal patterns). Untargeted profiling detected 5,259 plasma and 8,423 milk features. At the global level, groups defined by BHB status (CON, SCK, and HK) separated more clearly than parities, indicating that BHB status dominated broad variance; however, the timing and magnitude of metabolomic remodeling were strongly dependent on parity. The timing and magnitude of BHB-associated changes in plasma and milk metabolomic features differed by parity and matrix. P2 cows showed relatively limited changes at the SCK stage but more extensive remodeling at the HK stage, whereas P3 cows showed more pronounced early and progressive changes in several summary measures. Plasma pathway enrichment highlighted ovarian steroidogenesis and primary bile acid biosynthesis, suggesting parity-conditioned engagement of endocrine-hepatic modules during early lactation. In milk, hyperketonemia was associated with parity-dependent remodeling of complex lipids and secretion-related metabolism, with enrichment of sphingolipid metabolism, choline metabolism, pantothenate and CoA biosynthesis, and pyrimidine/nucleotide pathways. Plasma-milk association patterns differed by parity: P1 had more cross-matrix correlations meeting the predefined screening criteria, whereas P2 and P3 had fewer such correlations and a greater contribution of lipid-related hub features. Collectively, these findings indicate that similar BHB concentrations can be associated with distinct parity-dependent metabolic signatures. These results suggest that parity should be considered when interpreting BHB-associated metabolomic profiles, particularly candidate plasma and milk features related to acylcarnitine metabolism, bile acid metabolism, complex lipid remodeling, and nucleotide metabolism.
Antibiotic resistance genes (ARGs) are prevalent in livestock farms, with manure serving as a major reservoir. Their distribution and transmission are associated with husbandry practices. To evaluate the presence and abundance of ARGs along the feces-bedding-milk-calves chain, particularly under the practice of recycled manure solids for bedding (RMSB), we sampled from 5 critical stages: fecal slurry before anaerobic digestion, solids after solid-liquid separation, bedding derived from dried solids, unpasteurized mixed milk before feeding calves, and calf feces in 4 large-scale dairy farms (>1,000 lactating cows each). We analyzed ARGs and mobile genetic elements (MGEs) via PCR, characterized bacterial communities using 16sRNA sequencing, measured the concentrations of heavy metals and nutrients (total carbon and nitrogen), and evaluated the correlations between ARGs abundance and these factors using partial least squares path modeling analysis and redundancy analysis. A total of 27 ARGs and 10 MEGs subtypes were detected. Among these, 8 ARGs were found to be prevalent across all groups: blaTEM and cfxA (β-lactamase resistance), dfrA1 (sulfonamide resistance), strA and strB (aminoglycoside resistance), tetH, tet40, and tetQ (tetracycline resistance) besides 7 MGEs. Notably, higher ARG level in calf feces underscores urgent intervention needs. We also found that heavy metals, especially Zn and Pb, were positively correlated with ARG abundance. Most ARGs exhibited co-occurrence and significant correlation with microorganisms (including Enterococcus and Bacteroides), and MGEs (such as IS26, intI1, and Tp614). These findings suggested that the presence of continuous ARGs along the chain may be associated with RMSB, but this association requires further verification. Consequently, improved manure and bedding management, optimized dietary Zn levels, and enhanced removal or inactivation of heavy metals during bedding treatment could substantially lower the ecological and public health risks associated with ARG dissemination.
Whey protein isolates can be obtained from cheese whey or directly from skim milk via microfiltration, producing milk-derived whey protein isolate (mWPI). However, the differences between these protein sources in acidic beverage applications need further studies. Therefore, this study aimed to characterize mWPI functionality by developing stability diagrams in acidic beverage applications. Two lots of mWPI powder were obtained from a commercial manufacturer. A 10% (wt/wt) stock solution was prepared by rehydrating powders at room temperature for 30 min, followed by overnight storage at 4°C to ensure complete hydration. Protein concentrations were adjusted to 2, 4, 6, 8, and 10% (wt/wt). Samples were then adjusted to pH 3.0, 3.5, 4.0, 4.5, and 7.0, followed by the addition of NaCl (0, 5, and 10 mM). Low-pH samples (≤4.5) were heated to 109°C for 3 s, while high-pH samples (>4.5) were heated to 141°C for 6 s, then rapidly cooled. Samples were analyzed for turbidity, particle size, aggregation index (AI), soluble protein content, viscosity, and native PAGE. Beverages transitioned from clear to turbid at pH ≥ 3.5, regardless of salt concentration. Samples at pH 3.0- < 3.5 with protein concentrations of 3-10% exhibited the smallest particle sizes (<150 nm) and remained optically clear. Increased pH and salt concentration promoted protein unfolding, aggregation, and reduced solubility, resulting in higher turbidity and viscosity. In conclusion, mWPI enables formulation of clear, stable acidic beverages at protein concentrations up to ∼8% (wt/wt). These findings provide a practical framework for designing high-protein acidic dairy beverages with improved stability and functionality.
The main objective of this observational study was to assess the sensitivity and specificity of a commercial ATP luminometry swab (AquaSnap Total; Hygiena, CA, USA) and an on-farm bacteriological culture (Petrifilm; 3M aerobic colony count; MN, USA) in identifying bacterial presence in automated milk feeders (AMFs) for calves. The secondary purpose of this study was to examine the association between season, AMF design (Förster-Technik base models, Urban models with non-return valves, and Holm & Laue models with non-return valves and a peristaltic pump), and the type of equipment piece (automatically washed by the system, tube-like parts, other small parts, and nipples) and bacterial presence. Twenty-four dairy farms from Centre-du-Québec (Canada) using AMFs participated in the project. Only one farm used an optional hygiene feature. Each farm was visited 6 times from March 2024 to February 2025. Sterile physiological water was used to rinse each piece of equipment, and the residual liquid was poured in a sterile collection tube. All samples were analyzed with the luminometer (LUM, measured in relative light units; RLU), which detects ATP, and with the Petrifilm (BCT, measured in cfu/mL). We used BCT thresholds of > 10,000, > 20,000, > 50,000, and > 100,000 cfu/mL and selected corresponding thresholds for LUM that maximized sensitivity and specificity. Sensitivity and specificity, with their 95% Bayesian credible intervals (BCIs), were calculated for both tests with Bayesian latent class models assuming conditional dependence between the tests. Risk factors for BCT > 20,000 and > 100,000 cfu/mL were assessed through odds ratios (ORs) with 95% BCIs obtained with Bayesian logistic regression models. A total of 980 samples were collected in winter (n = 183; 18.7%), spring (n = 265; 27.0%), summer (n = 271; 27.7%), and fall (n = 261; 26.6%) from the milk containers (n = 211; 21.6%), nipples (n = 255; 26.0%), tube-like parts (n = 245; 25.0%), and other small parts (n = 269; 27.4%). The median results were 40,000 cfu/mL for BCT (interquartile range = 3,775-340,000) and 334 RLU for LUM (interquartile range = 75-2,184). The correlation between LUM and BCT was strong (rS = 0.83, 95% CI = 0.81-0.86). The LUM thresholds of > 200, > 250, > 350, and > 450 RLU for LUM corresponded to the > 10,000, > 20,000, > 50,000, and > 100,000 cfu/mL thresholds, respectively. The BCT and LUM methods had similar sensitivity and specificity at these thresholds. For example, the sensitivity of BCT and LUM were 0.86 (95% BCI = 0.82-0.90) and 0.81 (95% BCI = 0.75-0.87), and their specificity were 0.80 (95% BCI = 0.74-0.86) and 0.80 (95% BCI = 0.73-0.87) for the > 250 RLU and > 20,000 cfu/mL thresholds, respectively. Samples taken in spring (OR = 1.86; 95% BCI = 1.15-2.74), summer (OR = 1.97; 95% BCI = 1.21-2.87), and fall (OR = 1.69; 95% BCI = 1.09-2.51) had higher odds of being > 450 RLU, as measured by LUM, than the ones collected in winter. Samples taken from designs with non-return valves and a peristaltic pump had lower odds of being > 450 RLU than samples from base models (OR = 0.26; 95% BCI = 0.07-0.59). Manually washed parts had higher odds of being > 450 RLU (nipples: OR = 7.18; 95% BCI = 4.29-10.7; tube-like parts: OR = 3.70; 95% BCI = 2.26-5.46; other small parts: OR = 2.66; 95% BCI = 1.53-3.99) than automatically washed parts (milk container). These results can help investigate hygiene problems of feeding equipment in dairy calf barns.
Early identification of non-pregnant dairy cows is important for minimizing days open, yet current pregnancy diagnosis methods require additional cost, equipment, or veterinary intervention. In this retrospective observational study, we developed an ensemble machine learning model to screen pregnancy status using routine Dairy Herd Improvement (DHI) test-day data together with the date of first insemination recorded in the same DHI database. Test-day records (n = 60,301) from 2,614 Holstein cows across 36 farms in South Korea were used to construct 50,826 observation windows, each comprising 3 consecutive monthly records. Seventy features were engineered: test-day milk traits, days in milk (DIM), and 2 reproductive-context features derived from the first-insemination date. Three base learners (random forest, XGBoost, logistic regression) were combined via soft voting. Under cow-level grouped 5-fold cross-validation (positive class = pregnant), the ensemble achieved an AUC of 0.857 ± 0.006, PR-AUC of 0.949, and a Brier score of 0.119. Leave-one-farm-out cross-validation confirmed generalization to unseen farms within the studied DHI system (AUC = 0.863 ± 0.034). In SHAP analysis, the 6 test-day milk trait groups together formed the largest domain (53.9% combined), reproductive context was the largest single feature group (26.3%), and DIM contributed 17.8% as a time proxy for lactation stage. Even when DIM, reproductive-context, and parity features were removed, a test-day milk traits model retained an AUC of 0.794, and DIM-stratified analysis showed lower milk yield in pregnant cows within every DIM interval (Cohen's d = -0.24 to -0.43). A dual-threshold strategy provided 2 operating points: an F1-maximizing threshold (sensitivity = 0.966, specificity = 0.391) for surveillance and a cost-minimizing threshold (sensitivity = 0.680, specificity = 0.844, precision = 0.934) for non-pregnancy screening. Because 3 mo of data accumulation are required, the model serves as a monthly safety net within existing DHI workflows for cows missed by conventional early pregnancy diagnosis.
Milk mid-infrared (MIR) spectroscopy offers a rapid and cost-effective method for quantifying milk components and has shown potential for predicting physiological and environmental traits in dairy cattle, including methane (CH4) emissions. However, milk MIR-based CH4 predictions remain a black-box, as the biological mechanisms linking milk MIR spectra to CH4 emissions are not understood. This study aimed to predict CH4 emissions in dairy cows, measured using the GreenFeed system, based on milk MIR spectra and milk fatty acids (MFA). By comparing milk MIR-based models with those using MFA as predictors, we intended to disentangle the underlying biological signals associated with CH4 emissions. Data from 16 commercial Austrian dairy farms were available. All farms had a similar feeding system based on grass and corn silage, supplemented by varying amounts of concentrated feed and without grazing. Data included 807 records from 604 cows. Six partial least squares regression models were developed to predict CH4 emissions by using various combinations of predictor variables, including milk MIR spectra, MFA, major milk components, milk yield, days in milk and parity. A nested cross-validation framework was applied using 2 strategies: 5-fold cow-independent and leave-one-farm-out cross-validation. Mean CH4 emissions were 436 g/d. Positive correlations were observed between CH4 and de novo MFA throughout lactation, whereas preformed MFA showed negative associations, particularly in early lactation. These associations reflect signals associated with energy status. In 5-fold cow-independent cross-validation, the model based on milk MIR spectra showed moderate predictive accuracy (r = 0.52, R2 = 0.27, RMSE = 60 g/d), while the model based on milk MFA performed lower (r = 0.42, R2 = 0.18, RMSE = 63 g/d). However, combining MFA with fat%, protein% and lactose% slightly increased accuracy (r = 0.44, R2 = 0.20, RMSE = 62 g/d). The additional consideration of milk yield, days in milk and parity improved accuracy of MIR-based (r = 0.62, R2 = 0.39, RMSE = 55 g/d) and MFA-based models (r = 0.60, R2 = 0.36, RMSE = 56 g/d). The model including milk yield, days in milk and parity showed lower predictive performance (r = 0.49, R2 = 0.24, RMSE = 60 g/d), indicating that both MIR spectra and MFA together with fat%, protein% and lactose% provide additional information. Leave-one-farm-out cross-validation resulted in slightly lower predictive accuracies (MIR-based model r = 0.43, R2 = 0.22, RMSE = 63 g/d and MFA-based model: r = 0.41, R2 = 0.21, RMSE = 66 g/d). Models using all predictor traits performed best (MIR-based model r = 0.56, R2 = 0.33, RMSE = 58 g/d and MFA-based model: r = 0.58, R2 = 0.35, RMSE = 57 g/d). Correlations between predictions from the MIR- and MFA-based models including all predictors were high (r = 0.86), indicating a high agreement between the 2 modeling approaches. This study represents one of the first large-scale evaluations of milk MIR-based CH4 prediction models under commercial farm conditions. Milk MIR spectra and MFA explained only part of the variability in CH4 emissions, largely reflecting metabolic signals related to energy status. Correlations between CH4 emissions and MFA suggest that these models will favor cows with a negative energy balance during early lactation and with consistently lower de novo MFA synthesis throughout lactation. Further research is needed to gain more insight into the use of milk mid-infrared spectra for genetic selection to reduce CH4 emissions.
Subclinical mastitis, one of the most common disorders in dairy cattle farming, affects both milk yield and milk quality. While it causes no visible changes to the udder or milk, it may be either chronic or progress to clinical mastitis. This increases the costs of milk production and herd management, and it is therefore important to devise a means of early identification of cows at risk of subclinical mastitis. The aim of the study was to develop and evaluate models based on artificial neural networks and designed for the identification of cows at risk of subclinical mastitis. Neural networks can recognize complex data patterns; however, the process of building a model, selecting hyperparameters and training a network is a time-consuming and computationally intensive task, and therefore neural network-based models were compared with classical machine learning (ML) models: Support Vector Machines (SVM), Random Forest and Gradient Boosting. The data were collected between 2010 and 2011 as part of routine milk recording procedures. The data set contained 19,768 test day records for 2,225 Polish Holstein-Friesian cows from 3 herds. All the models were trained and evaluated using an 80:20 train-test split ratio according to animal ID. Cows were classified as healthy or at risk of subclinical mastitis based on data from the test day preceding the test day for which the prediction was made. The neural network that achieved the highest mean F1-score in the cross validation (0.763) and for the testing data set (0.760) was optimized using the BayesianOptimization tuner, and the data were rescaled using RobustScaler before calculations. However, the mean F1-scores from the cross validation for 3 classical ML models: SVM, Random Forest and Gradient Boosting, were higher (0.769, 0.769 and 0.766 respectively) than the F1-score obtained for the best performing neural network. SVM and Random Forest models also achieved higher F1-scores (0.763 and 0.761 respectively) for the testing data set as compared with the best performing neural network. As far as ML models are concerned, an increasing importance is also being placed on the interpretation and evaluation of the impact of individual variables on model-derived predictions. This is where SHAP (SHapley Additive exPlanations) values can become useful. For the best performing classical ML models: SVM, Random Forest and Gradient Boosting, and the best performing neural network, somatic cell score (SCS) from previous test day had the greatest impact on predictions, and a higher SCS was associated with a greater probability of subclinical mastitis occurring in subsequent test day. The second variable with the greatest impact on the risk of subclinical mastitis was lactation number. Depending on the model, the third most influential variable was milk yield, lactose percentage or fat percentage. Classical ML models with good predictive performance can support farmers in making decisions concerning the health of cows' udders, and the use of model interpretation tools can help understand why a particular model generates a particular prediction.
A multisite time series investigated effects of nutritional changes in TMR diets and temperature-humidity index (THI) on dairy cow milk yield. Weekly diet samples were collected and tested from 8 farms with a total of 9 strings or herds over an average of 58 weeks. Nutrient components of the diet examined for effects on milk yield were NDF, lignin, crude fat, CP, neutral detergent insoluble crude protein (NDICP), starch, and non-fiber carbohydrates (NFC). Milk yield was measured from daily shipped milk or individual milk volumes. Weather data were measured using on-farm or nearby publicly available weather stations to calculate the weekly average of the daily maximum temperature-humidity index (mTHI). The groups of cows on the diets monitored had mean daily milk yields between 31 and 52 L/cow. Principal components (PC) were created for nutritional covariables. The first PC (PC1) explained over 50% of the variability in nutrition and was positively correlated with slower-fermenting carbohydrates. The PC2 and PC3 were composed mainly of fat and protein variables, though PC3 had a negative loading with NDF. Univariable regression spline models were used to generate residual and smooth values for each variable. Lagged variables up to -5 weeks were generated for each covariable's residual within their farm diet. A Pearson correlation matrix of the milk yield residual and covariable residuals at each week's lag was created to determine the lag of greatest effect. These were included in univariable mixed models before a multivariable mixed model was created through backward stepping with farm diet as a random effect. Changes in milk yield were positively associated with changes in PC1 (carbohydrates) 4 weeks before and PC3 (crude fat/NDICP) at the time of milk yield changes in univariable and multivariable analysis. In the all-diet nutrient univariable model, changes in NDF 4 weeks earlier and NDICP at the time increased milk yield, NFC and starch 4 weeks earlier, and mTHI at the same time decreased milk yield. In the all-diets multivariable model, NFC 4 weeks earlier and mTHI at the time were negatively associated, crude fat 1 week earlier and NDICP at the time were positively associated with milk yield changes. Diets were grouped into low-starch (n = 3) and high-starch diets (n = 6) with a cut point of 24% mean dietary starch. Cows on high-starch diets had a higher average milk yield (high-starch = 41.8 L/cow; low-starch = 34.8 L/cow). The multivariable low-starch diet model showed a positive association between changes in milk yield and starch 1 week earlier and NDICP at the time, while DIM was negatively associated. For the high-starch diets, the multivariable model found positive associations for CP 3 weeks earlier, crude fat 4 weeks earlier and DIM, while starch 2 weeks earlier and mTHI 1 week earlier were negatively associated with milk yield changes. This study highlights the delayed effects of dietary changes, with carbohydrate components having their greatest association with milk yield 4 weeks later. This could have implications for study design and nutritional models. Differing starch levels led to variable associations of nutrients on milk yield.
In large-scale dairy farms, flushing wastewater are directed either into municipal pipelines or manure collection pipeline. These distinct collection methods result in varying concentrations of manure in collection tanks. However, the impact of manure substrate concentration on the reduction of antibiotic resistance genes (ARGs) during anaerobic digestion (AD) remains unclear. To address this gap, a 2-factor experiment involving temperature (37°C and 55°C) and dairy manure concentrations (low and high) was conducted, where manure sampling from collection tanks in 2 same scale farms. The results indicate that the abundance of ARGs in low-concentration substrates (5.15 × 10-1 and 4.64 × 10-2 copies/16S copies) was significantly lower than in high-concentration substrates (1.75 × 10° and 2.66 × 10° copies/16S copies) after AD. Statistical analysis revealed that the effect of substrate concentration on ARG reduction was more significant than temperatures. Further analysis using partial least square models revealed that physicochemical parameters especially NH4+-N levels associated with substrate concentration, played a greater role in influencing the variations of ARGs and their bacterial hosts than temperature. In summary, the study highlights the importance of manure collection method in influencing ARG reduction during AD processes. Combining wastewater with manure collection appears to be more effective in reducing ARGs, while high-concentration manure may require pre-treatments such as dilution and ammonia recovery to mitigate the risks of ARG spread.
Thermal stress alters ruminal function and physiology, compromising health and productivity. Methionine enhances endogenous antioxidant synthesis, whereas guanidinoacetic acid (GAA) supports cellular ATP production. Because methionine serves as a methyl donor for the conversion of GAA to creatine, both nutrients are linked to cellular metabolism and antioxidant defense. The objective was to determine effects of rumen-protected methionine (RPM; Mepron®, Evonik) and GAA (GuanAMINO®, ≥ 97% GAA, ∼1% starch, ∼1% moisture; Evonik), fed alone or in combination during heat stress (HS) on production, thermoregulation, ruminal biopolymer-degrading enzymes and microbiota, ruminal papillae gene expression and cystathionine β-synthase (CBS) activity, and plasma biomarkers. Milk, blood, ruminal fluid and papillae were collected to evaluate lactational, metabolic, microbial, and molecular responses. Ten multiparous Holstein cows (5 intact and 5 with a ruminal cannula; 122 ± 28 DIM, 41.8 ± 3.7 kg milk/d) were used in a replicated 5 × 5 Latin square with 23-d periods (16 d thermoneutral, 7 d HS) and 5 treatments: thermoneutral control (CON-TN), HS control (CON-HS), RPM-HS (RPM at 0.10% of DM), GAA-HS (25 g GAA/d), MIX-HS (RPM at 0.10% of DM plus 25 g GAA/d). Electric heat blankets (EHB) operating at the highest setting (∼40°C) were used to induce HS. Data were analyzed using PROC MIXED (SAS v9.4) with preplanned contrasts: CON-HS vs. CON-TN, RPM-HS vs. CON-HS, RPM-HS vs. GAA-HS, and MIX-HS vs. CON-HS. Compared with CON-TN, CON-HS cows had greater rectal (+0.32°C) and vaginal (+0.34°C) temperatures and respiration rate (+11 beats/min). These variables did not differ among HS treatments. Compared with CON-TN, CON-HS cows had lower milk yield (-3.9 kg/d), milk protein concentration (-0.22 percentage units), and milk protein yield (-0.11 kg/d), whereas MUN was greater (+2.4 mg/dL). Compared with CON-HS, RPM-HS increased milk protein (+0.28 percentage units), casein (+0.11 percentage units), and protein yield (+0.12 kg/d). Milk protein concentration also was greater in RPM-HS than GAA-HS. Total plasma AA concentrations did not differ among treatments (2,056 ± 89 µM), but citrulline concentration was greater in CON-HS than CON-TN (69 vs. 55 µM). Feeding RPM-HS increased plasma cystine concentration compared with CON-HS (24.4 vs. 18.8 µM). At the ruminal level, compared with CON-TN, CON-HS increased the molar proportions of iso-butyrate and isovalerate, and the abundance of Rumicoccus flavefaciens and Succinimonas amylolytica. These variables were not affected by feeding RPM, GAA, or MIX; however, relative to CON-HS, RPM-HS increased cellulase activity and tended to increase xylanase activity. The abundance of Megaspheara elsdenii was greater and Succinimonas amylolytica lower in RPM-HS than CON-HS, whereas GAA-HS increased abundance of Streptococcus bovis approximately 3-fold relative to CON-HS. Ruminal papillae CBS activity more than doubled in CON-HS compared with CON-TN. The HS did not affect mRNA abundance of the 28 target genes evaluated. In contrast, RPM-HS and, to a greater extent, MIX-HS induced coordinated molecular remodeling characterized by upregulation of genes involved in polyamine synthesis, antioxidant responses, and tight-junction signaling. Overall, RPM improved milk protein synthesis and metabolic resilience during HS, whereas MIX primarily enhanced epithelial adaptative responses without additional lactational benefits.
