Genomic prediction of blood biomarkers of metabolic disorders in Holstein cattle using parametric and nonparametric models.

BACKGROUND: Metabolic disturbances adversely impact productive and reproductive performance of dairy cattle due to changes in endocrine status and immune function, which increase the risk of disease. This may occur in the post-partum phase, but also throughout lactation, with sub-clinical symptoms. Recently, increased attention has been directed towards improved health and resilience in dairy cattle, and genomic selection (GS) could be a helpful tool for selecting animals that are more resilient to metabolic disturbances throughout lactation. Hence, we evaluated the genomic prediction of serum biomarkers levels for metabolic distress in 1353 Holsteins genotyped with the 100K single nucleotide polymorphism (SNP) chip assay. The GS was evaluated using parametric models best linear unbiased prediction (GBLUP), Bayesian B (BayesB), elastic net (ENET), and nonparametric models, gradient boosting machine (GBM) and stacking ensemble (Stack), which combines ENET and GBM approaches. RESULTS: The results show that the Stack approach outperformed other methods with a relative difference (RD), calculated as an increment in prediction accuracy, of approximately 18.0% compared to GBLUP, 12.6% compared to BayesB, 8.7% compared to ENET, and 4.4% compared to GBM. The highest RD in prediction accuracy between other models with respect to GBLUP was observed for haptoglobin (hapto) from 17.7% for BayesB to 41.2% for Stack; for Zn from 9.8% (BayesB) to 29.3% (Stack); for ceruloplasmin (CuCp) from 9.3% (BayesB) to 27.9% (Stack); for ferric reducing antioxidant power (FRAP) from 8.0% (BayesB) to 40.0% (Stack); and for total protein (PROTt) from 5.7% (BayesB) to 22.9% (Stack). Using a subset of top SNPs (1.5k) selected from the GBM approach improved the accuracy for GBLUP from 1.8 to 76.5%. However, for the other models reductions in prediction accuracy of 4.8% for ENET (average of 10 traits), 5.9% for GBM (average of 21 traits), and 6.6% for Stack (average of 16 traits) were observed. CONCLUSIONS: Our results indicate that the Stack approach was more accurate in predicting metabolic disturbances than GBLUP, BayesB, ENET, and GBM and seemed to be competitive for predicting complex phenotypes with various degrees of mode of inheritance, i.e. additive and non-additive effects. Selecting markers based on GBM improved accuracy of GBLUP.

Characterization of dry-cured ham microbiota at 12 months of seasoning obtained from different rearing strategies using 16S rRNA profiling.

In this study, we investigated the microbiota of 72 Italian ham samples collected after 12 months of seasoning. The hams were elaborated from pigs fed different rearing methods, including the traditional restricted medium protein diet chosen as control (C group); restrictive low protein diet (LP group); two ad libitum high-protein diet groups (HP9M group: slaughter at 9 months of age; HP170 group: slaughter at 170 kg). A multi-amplicon 16S metabarcoding approach was used, and a total of 2845 Amplicon Sequence Variants were obtained from the 72 ham samples. Main phyla included: Firmicutes (90.8%), Actinobacteria (6.2%), Proteobacteria (2.7%), and Bacteroidota (0.12%). The most common genera were Staphylococcus, Tetragenococcus, and Brevibacterium. Shannon index for α-diversity was found statistically significant, notably for the HP9M group, indicating higher diversity compared to C. PERMANOVA test on ß-diversity showed significant differences in rearing methods between HP170 and C, HP170 and LP, and HP9M vs. C. All three rearing methods revealed associations with characteristic communities: the HP9M group had the highest number of associations, many of which were due to spoilage bacteria, whereas the LP group had the highest number of seasoning-favourable genera.

Novel insights into the associations between immune cell population distribution in mammary glands and milk minerals in Holstein cows.

Udder health has a crucial role in sustainable milk production, and various reports have pointed out that changes in udder condition seem to affect milk mineral content. The somatic cell count (SCC) is the most recognized indicator for the determination of udder health status. Recently, a new parameter, the differential somatic cell count (DSCC), has been proposed for a more detailed evaluation of intramammary infection patterns. Specifically, the DSCC is the combined proportions of polymorphonuclear neutrophils and lymphocytes (PMN-LYM) on the total SCC, with macrophages (MAC) representing the remainder proportion. In this study, we evaluated the association between DSCC in combination with SCC on a detailed milk mineral profile in 1,013 Holstein-Friesian cows reared in 5 herds. An inductively coupled plasma-optical emission spectrometry was used to quantify 32 milk mineral elements. Two different linear mixed models were fitted to explore the associations between the milk mineral elements and first, the DSCC combined with SCC, and second, DSCC expressed as the PMN-LYM and MAC counts, obtained by multiplying the proportion of PMN-LYM and MAC by SCC. We observed a significant positive association between SCC and milk Na, S, and Fe levels. Differential somatic cell count showed an opposite behavior to the one displayed by SCC, with a negative association with Na and positive association with K milk concentrations. When considering DSCC as count, Na and K showed contrasting behavior when associated with PMN-LYM or MAC counts, with decreasing of Na content and increasing K when associated with increasing PMN-LYM counts, and increasing Na and decreasing K when associated with increasing MAC count. These findings confirmed that an increase in SCC is associated with altered milk Na and K amounts. Moreover, MAC count seemed to mirror SCC patterns, with the worsening of inflammation. Differently, PMN-LYM count exhibited patterns of associations with milk Na and K contents attributable more to LYM than PMN, given the nonpathological condition of the majority of the investigated population. An interesting association was observed for milk S content, which increased with increasing of inflammatory conditions (i.e., increased SCC and MAC count) probably attributable to its relationship with milk proteins, especially whey proteins. Moreover, milk Fe content showed positive associations with the PMN-LYM population, highlighting its role in immune regulation during inflammation. Further studies including individuals with clinical condition are needed to achieve a comprehensive view of milk mineral behavior during udder health impairment.

Integrating on-farm and genomic information improves the predictive ability of milk infrared prediction of blood indicators of metabolic disorders in dairy cows.

BACKGROUND: Blood metabolic profiles can be used to assess metabolic disorders and to evaluate the health status of dairy cows. Given that these analyses are time-consuming, expensive, and stressful for the cows, there has been increased interest in Fourier transform infrared (FTIR) spectroscopy of milk samples as a rapid, cost-effective alternative for predicting metabolic disturbances. The integration of FTIR data with other layers of information such as genomic and on-farm data (days in milk (DIM) and parity) has been proposed to further enhance the predictive ability of statistical methods. Here, we developed a phenotype prediction approach for a panel of blood metabolites based on a combination of milk FTIR data, on-farm data, and genomic information recorded on 1150 Holstein cows, using BayesB and gradient boosting machine (GBM) models, with tenfold, batch-out and herd-out cross-validation (CV) scenarios. RESULTS: The predictive ability of these approaches was measured by the coefficient of determination (R2). The results show that, compared to the model that includes only FTIR data, integration of both on-farm (DIM and parity) and genomic information with FTIR data improves the R2 for blood metabolites across the three CV scenarios, especially with the herd-out CV: R2 values ranged from 5.9 to 17.8% for BayesB, from 8.2 to 16.9% for GBM with the tenfold random CV, from 3.8 to 13.5% for BayesB and from 8.6 to 17.5% for GBM with the batch-out CV, and from 8.4 to 23.0% for BayesB and from 8.1 to 23.8% for GBM with the herd-out CV. Overall, with the model that includes the three sources of data, GBM was more accurate than BayesB with accuracies across the CV scenarios increasing by 7.1% for energy-related metabolites, 10.7% for liver function/hepatic damage, 9.6% for oxidative stress, 6.1% for inflammation/innate immunity, and 11.4% for mineral indicators. CONCLUSIONS: Our results show that, compared to using only milk FTIR data, a model integrating milk FTIR spectra with on-farm and genomic information improves the prediction of blood metabolic traits in Holstein cattle and that GBM is more accurate in predicting blood metabolites than BayesB, especially for the batch-out CV and herd-out CV scenarios.

Associations between the detailed milk mineral profile, milk composition, and metabolic status in Holstein cows.

The causes of variation in the milk mineral profile of dairy cattle during the first phase of lactation were studied under the hypothesis that the milk mineral profile partially reflects the animals' metabolic status. Correlations between the minerals and the main milk constituents (i.e., protein, fat, and lactose percentages), and their associations with the cows' metabolic status indicators were explored. The metabolic status indicators (MET) that we used were blood energy-protein metabolites [nonesterified fatty acids, ß-hydroxybutyrate (BHB), glucose, cholesterol, creatinine, and urea], and liver ultrasound measurements (predicted triacylglycerol liver content, portal vein area, portal vein diameter and liver depth). Milk and blood samples, and ultrasound measurements were taken from 295 Holstein cows belonging to 2 herds and in the first 120 d in milk (DIM). Milk mineral contents were determined by ICP-OES; these were considered the response variable and analyzed through a mixed model which included DIM, parity, milk yield, and MET as fixed effects, and the herd/date as a random effect. The MET traits were divided in tertiles. The results showed that milk protein was positively associated with body condition score (BCS) and glucose, and negatively associated with BHB blood content; milk fat was positively associated with BHB content; milk lactose was positively associated with BCS; and Ca, P, K and S were the minerals with the greatest number of associations with the cows' energy indicators, particularly BCS, predicted triacylglycerol liver content, glucose, BHB and urea. We conclude that the protein, fat, lactose, and mineral contents of milk partially reflect the metabolic adaptation of cows during lactation and within 120 DIM. Variations in the milk mineral profile were consistent with changes in the major milk constituents and the metabolic status of cows.

Prediction of detailed blood metabolic profile using milk infrared spectra and machine learning methods in dairy cattle.

The adoption of preventive management decisions is crucial to dealing with metabolic impairments in dairy cattle. Various serum metabolites are known to be useful indicators of the health status of cows. In this study, we used milk Fourier-transform mid-infrared (FTIR) spectra and various machine learning (ML) algorithms to develop prediction equations for a panel of 29 blood metabolites, including those related to energy metabolism, liver function/hepatic damage, oxidative stress, inflammation/innate immunity, and minerals. For most traits, the data set comprised observations from 1,204 Holstein-Friesian dairy cows belonging to 5 herds. An exception was represented by ß-hydroxybutyrate prediction, which contained observations from 2,701 multibreed cows pertaining to 33 herds. The best predictive model was developed using an automatic ML algorithm that tested various methods, including elastic net, distributed random forest, gradient boosting machine, artificial neural network, and stacking ensemble. These ML predictions were compared with partial least squares regression, the most commonly used method for FTIR prediction of blood traits. Performance of each model was evaluated using 2 cross-validation (CV) scenarios: 5-fold random (CVr) and herd-out (CVh). We also tested the best model's ability to classify values precisely in the 2 extreme tails, namely, the 25th (Q25) and 75th (Q75) percentiles (true-positive prediction scenario). Compared with partial least squares regression, ML algorithms achieved more accurate performance. Specifically, elastic net increased the R2 value from 5% to 75% for CVr and 2% to 139% for CVh, whereas the stacking ensemble increased the R2 value from 4% to 70% for CVr and 4% to 150% for CVh. Considering the best model, with the CVr scenario, good prediction accuracies were obtained for glucose (R2 = 0.81), urea (R2 = 0.73), albumin (R2 = 0.75), total reactive oxygen metabolites (R2 = 0.79), total thiol groups (R2 = 0.76), ceruloplasmin (R2 = 0.74), total proteins (R2 = 0.81), globulins (R2 = 0.87), and Na (R2 = 0.72). Good prediction accuracy in classifying extreme values was achieved for glucose (Q25 = 70.8%, Q75 = 69.9%), albumin (Q25 = 72.3%), total reactive oxygen metabolites (Q25 = 75.1%, Q75 = 74%), thiol groups (Q75 = 70.4%), total proteins (Q25 = 72.4%, Q75 = 77.2.%), globulins (Q25 = 74.8%, Q75 = 81.5%), and haptoglobin (Q75 = 74.4%). In conclusion, our study shows that FTIR spectra can be used to predict blood metabolites with relatively good accuracy, depending on trait, and are a promising tool for large-scale monitoring.

Invited review: A comprehensive review of visible and near-infrared spectroscopy for predicting the chemical composition of cheese.

Substantial research has been carried out on rapid, nondestructive, and inexpensive techniques for predicting cheese composition using spectroscopy in the visible and near-infrared radiation range. Moreover, in recent years, new portable and handheld spectrometers have been used to predict chemical composition from spectra captured directly on the cheese surface in dairies, storage facilities, and food plants, removing the need to collect, transport, and process cheese samples. For this review, we selected 71 papers (mainly dealing with prediction of the chemical composition of cheese) and summarized their results, focusing our attention on the major sources of variation in prediction accuracy related to cheese variability, spectrometer and spectra characteristics, and chemometrics techniques. The average coefficient of determination obtained from the validation samples ranged from 86 to 90% for predicting the moisture, fat, and protein contents of cheese, but was lower for predicting NaCl content and cheese pH (79 and 56%, respectively). There was wide variability with respect to all traits in the results of the various studies (standard deviation: 9-30%). This review draws attention to the need for more robust equations for predicting cheese composition in different situations; the calibration data set should consist of representative cheese samples to avoid bias due to an overly specific field of application and ensure the results are not biased for a particular category of cheese. Different spectrometers have different accuracies, which do not seem to depend on the spectrum extension. Furthermore, specific areas of the spectrum-the visible, infrared-A, or infrared-B range-may yield similar results to broad-range spectra; this is because several signals related to cheese composition are distributed along the spectrum. Small, portable instruments have been shown to be viable alternatives to large bench-top instruments. Last, chemometrics (spectra pre-treatment and prediction models) play an important role, especially with regard to difficult-to-predict traits. A proper, fully independent, validation strategy is essential to avoid overoptimistic results.

Role of CSN2, CSN3, and BLG genes and the polygenic background in the cattle milk protein profile.

To devise better selection strategies in dairy cattle breeding programs, a deeper knowledge of the role of the major genes encoding for milk protein fractions is required. The aim of the present study was to assess the effect of the CSN2, CSN3, and BLG genotypes on individual protein fractions (αS1-CN, αS2-CN, ß-CN, κ-CN, ß-LG, α-LA) expressed qualitatively as percentages of total nitrogen content (% N), quantitatively as contents in milk (g/L), and as daily production levels (g/d). Individual milk samples were collected from 1,264 Brown Swiss cows reared in 85 commercial herds in Trento Province (northeast Italy). A total of 989 cows were successfully genotyped using the Illumina Bovine SNP50 v.2 BeadChip (Illumina Inc.), and a genomic relationship matrix was constructed using the 37,519 SNP markers obtained. Milk protein fractions were quantified and the ß-CN, κ-CN, and ß-LG genetic variants were identified by reversed-phase HPLC (RP-HPLC). All protein fractions were analyzed through a Bayesian multitrait animal model implemented via Gibbs sampling. The effects of days in milk, parity order, and the CSN2, CSN3, and BLG genotypes were assigned flat priors in this model, whereas the effects of herd and animal additive genetic were assigned Gaussian prior distributions, and inverse Wishart distributions were assumed for the respective co-variance matrices. Marginal posterior distributions of the parameters of interest were compared before and after the inclusion of the effects of the 3 major genes in the model. The results showed that a high portion of the genetic variance was controlled by the major genes. This was particularly apparent in the qualitative protein profile, which was found to have a higher heritability than the protein fraction contents in milk and their daily yields. When the genes were included individually in the model, CSN2 was the major gene controlling all the casein fractions except for κ-CN, which was controlled directly by the CSN3 gene. The BLG gene had the most influence on the 2 whey proteins. The genetic correlations showed the major genes had only a small effect on the relationships between the protein fractions, but through comparison of the correlation coefficients of the proteins expressed in different ways they revealed potential mechanisms of regulation and competitive synthesis in the mammary gland. The estimates for the effects of the CSN2 and CSN3 genes on protein profiles showed overexpression of protein synthesis in the presence of the B allele in the genotype. Conversely, the ß-LG B variant was associated with a lower concentration of ß-LG compared with the ß-LG A variant, independently of how the protein fractions were expressed, and it was followed by downregulation (or upregulation in the case of the ß-LG B) of all other protein fractions. These results should be borne in mind when seeking to design more efficient selection programs aimed at improving milk quality for the efficiency of dairy industry and the effect of dairy products on human health.

Real-time milk analysis integrated with stacking ensemble learning as a tool for the daily prediction of cheese-making traits in Holstein cattle.

Cheese-making traits in dairy cattle are important to the dairy industry but are difficult to measure at the individual level because there are limitations on collecting phenotypic information. Mid-infrared spectroscopy has its advantages, but it can only be used during monthly milk recordings. Recently, in-line devices for real-time analysis of milk quality have been developed. The AfiLab recording system (Afimilk) offers significant benefits as phenotypes can be collected from each cow at each milking session. The objective of this study was to assess the potential of integrating AfiLab real-time milk analyzer measures with the stacking ensemble learning technique using heterogeneous base learners for the in-line daily monitoring of cheese-making traits in Holstein cattle with a view to developing a precision livestock farming system for monitoring the technological quality of milk. Data and samples for wet-laboratory analyses were collected from 499 Holstein cows belonging to 2 farms where the AfiLab system was installed. The traits of concern were 9 milk coagulation traits [3 milk coagulation properties (MCP), and 6 curd firming traits (CFt)], and 7 cheese-making traits [3 cheese yield (CY) traits, and 4 milk nutrient recovery in the curd (REC) traits]. The near-infrared AfiLab spectral data and on-farm information (days in milk and parity) were used to assess the predictive ability of different statistical methods [elastic net (EN), gradient boosting machine (GBM), extreme gradient boosting (XGBoost), and artificial neural network (ANN)] across different cross-validation scenarios. These statistical methods were considered the base learners, which were then combined in a stacking ensemble learning. Results indicate that including information on the cows (days in milk and parity) in the AfiLab infrared prediction increased its accuracy by 10.3% for traditional MCP, 13.8% for curd firming, 9.8% for CY, and 11.2% for REC traits compared with those obtained from near-infrared AfiLab alone. The statistical approaches exhibited high prediction accuracies (R2) averaged across the cross-validation scenarios for traditional MCP (0.58 for ANN, 0.55 for EN and GBM, 0.52 for XGBoost, and 0.62 for stacking ensemble), CFt (0.55 for ANN, 0.54 for EN and GBM, 0.53 for XGBoost, and 0.61 for stacking ensemble), and similar R2 averages for CY and REC (0.55 for ANN, 0.54 for EN and GBM, 0.53 for XGBoost, and 0.61 for stacking ensemble). The ANN approach was more accurate than the other base learners (EN, GBM, and XGBoost) and improved accuracy across cross-validation scenarios on average by 7% for traditional MCP, 5% for CFt, 8% for CY, and 7% for REC. The stacking ensemble method improved prediction accuracy by 3% to 31% for traditional MCP, 2% to 26% for CFt, 1% to 38% for CY traits, and 2% to 27% for REC traits compared with the base learners. The prediction accuracies of the different approaches evaluated tended to decrease from the 10-fold cross-validation to the independent validation scenario, although there was a smaller reduction in prediction accuracy with the stacking ensemble learning technique across all the cross-validation scenarios. Our results show that combining in-line on-farm information with stacking ensemble machine learning represents an effective alternative for obtaining robust daily predictions of milk cheese-making traits.

Integrating genomic and infrared spectral data improves the prediction of milk protein composition in dairy cattle.

BACKGROUND: Over the past decade, Fourier transform infrared (FTIR) spectroscopy has been used to predict novel milk protein phenotypes. Genomic data might help predict these phenotypes when integrated with milk FTIR spectra. The objective of this study was to investigate prediction accuracy for milk protein phenotypes when heterogeneous on-farm, genomic, and pedigree data were integrated with the spectra. To this end, we used the records of 966 Italian Brown Swiss cows with milk FTIR spectra, on-farm information, medium-density genetic markers, and pedigree data. True and total whey protein, and five casein, and two whey protein traits were analyzed. Multiple kernel learning constructed from spectral and genomic (pedigree) relationship matrices and multilayer BayesB assigning separate priors for FTIR and markers were benchmarked against a baseline partial least squares (PLS) regression. Seven combinations of covariates were considered, and their predictive abilities were evaluated by repeated random sub-sampling and herd cross-validations (CV). RESULTS: Addition of the on-farm effects such as herd, days in milk, and parity to spectral data improved predictions as compared to those obtained using the spectra alone. Integrating genomics and/or the top three markers with a large effect further enhanced the predictions. Pedigree data also improved prediction, but to a lesser extent than genomic data. Multiple kernel learning and multilayer BayesB increased predictive performance, whereas PLS did not. Overall, multilayer BayesB provided better predictions than multiple kernel learning, and lower prediction performance was observed in herd CV compared to repeated random sub-sampling CV. CONCLUSIONS: Integration of genomic information with milk FTIR spectral can enhance milk protein trait predictions by 25% and 7% on average for repeated random sub-sampling and herd CV, respectively. Multiple kernel learning and multilayer BayesB outperformed PLS when used to integrate heterogeneous data for phenotypic predictions.

Bayesian inference of the inbreeding load variance for fertility traits in Brown Swiss cattle.

Our study investigated the inbreeding load for fertility traits in the Italian Brown Swiss dairy cattle breed. Fertility traits included continuous traits (i.e., interval from calving to first service, days open, and calving interval) and categorical traits (i.e., calving rate at first insemination and nonreturn date at d 56). We included only records of the first 3 parities of cows that calved between 2010 and 2018. We traced up the pedigree of the cows with records as far as possible, ending up with a total of 73,246 animals. The final data set consisted of 59,864 records from 34,921 cows. We analyzed all models using a Bayesian approach that included a covariate with total inbreeding in addition to systematic, permanent environment, additive genetic, and inbreeding load effects. We then evaluated the trends in heritabilities and ratios of the inbreeding load using a continuum of partial inbreeding coefficients from 0.001 to 0.100 as reference. Posterior estimates of heritabilities tended to decrease across the continuum, whereas ratios of the inbreeding load tended to increase, more noticeably in categorical traits (calving rate at first insemination and nonreturn date at d 56). From the results obtained, we confirmed the presence of heterogeneity in inbreeding depression. We then predicted the inbreeding load effects, which had a low reliability of prediction, explained by having only 513 ancestors generating inbreeding. However, reliability of prediction was high enough for some of the individuals, obtaining a favorable prediction of inbreeding load for a relevant percentage, which improved the phenotypic performance of their inbred descendants. These results make it feasible to implement breeding and management strategies that select ancestors with a favorable inbreeding load prediction. In addition, it opens the possibility to define a global index for the expected consequences of the inbreeding generated by each individual.

Nonlinear modeling to describe the pattern of 15 milk protein and nonprotein compounds over lactation in dairy cows.

The protein profile of milk includes several caseins, whey proteins, and nonprotein nitrogen compounds, which influence milk's value for human nutrition and its cheesemaking properties for the dairy industry. To fill in the gap in current knowledge of the patterns of these individual nitrogenous compounds throughout lactation, we tested the ability of a parametric nonlinear lactation model to describe the pattern of each N compound expressed qualitatively (as % of total milk N), quantitatively (in g/L milk), and as daily yield (in g/d). The lactation model was tested on a data set of detailed milk nitrogenous compound profiles (15 fractions-12 protein traits and 3 nonproteins-for each expression mode: 45 traits) obtained from 1,342 cows reared in 41 multibreed herds. Our model was a modified version of Wilmink's model, often used for describing milk yield during lactation because of its reliability and ease of parameter interpretation from a biological point of view. We allowed the sign of the persistency coefficient (parameter c) that explained the variation in the long-term milk component (parameter a) to be positive or negative. We also allowed the short-term milk component (parameter b) to be positive or negative, and we estimated a specific speed of adaptation parameter (parameter k) for each trait rather than assumed a value a priori, as in the original model (k = 0.05). These 4 parameters were included in a nonlinear mixed model with cow breed and parity order as fixed effects, and herd-date as random. Combinations of the positive and negative signs of the b and c parameters allowed us to identify 4 differently shaped lactation curves, all found among the patterns exhibited by the nitrogenous fractions as follows: the "zenith" curve (with a maximum peak; for milk yield and 10 other N traits), the "nadir" curve (with a minimum point; for 20 traits, including almost all those expressed in g/L of milk), the "downward" curve (continuously decreasing; for 14 traits, including almost all those in g/d), and the "upward" curve (continuously increasing; only for κ-casein, in % N). Direct estimation of the k parameters specific to each trait showed the large variability in the adaptation speed of fresh cows and greatly increased the model's flexibility. The results indicated that nonlinear parametric mathematical models can effectively describe the different and complex patterns exhibited by individual nitrogenous fractions during lactation; therefore, they could be useful tools for interpreting milk composition variations during lactation.

Effects of breed, farm intensiveness, and cow productivity level on cheese-making ability predicted using infrared spectral data at the population level.

Fourier-transform infrared (FTIR) spectra collected during milk recording schemes at population level can be used for predicting novel traits of interest for farm management, cows' genetic improvement, and milk payment systems. The aims of this study were as follows. (1) To predict cheese yield traits using FTIR spectra from routine milk recordings exploiting previously developed calibration equations. (2) To compare the predicted cheese-making abilities of different dairy and dual-purpose breeds. (3) To analyze the effects of herds' level of intensiveness (HL) and of the cow's level of productivity (CL). (4) To compare the patterns of predicted cheese yields with the patterns of milk composition in different breeds to discern the drivers of cheese-making efficiency. The major sources of variation of FTIR predictions of cheese yield ability (fresh cheese or cheese solids produced per unit milk) of individual milk samples were studied on 115,819 cows of 4 breeds (2 specialized dairy breeds, Holstein and Brown Swiss, and 2 dual-purpose breeds, Simmental and Alpine Grey) from 6,430 herds and exploiting 1,759,706 FTIR test-day spectra collected over 7 yr of milk sampling. Calibration equations used were previously developed on 1,264 individual laboratory model cheese procedures (cross-validation R2 0.85 and 0.95 for fresh and solids cheese yields, respectively). The linear model used for statistical analysis included the effects of parity, lactation stage, year of calving, month of sampling, HL, CL, breed of cow, and the interactions breed × HL and breed × CL. The HL and CL stratifications (5 classes each) were based on average daily secretion of milk net energy per cow. All effects were highly significant. The major conclusions were as follows. (1) The FTIR-based prediction of cheese yield of milk goes beyond the knowledge of fat and protein content, partially explaining differences in cheese-making ability in different cows, breeds and herds. (2) Differences in cheese yields of different breeds are only partially explained by milk fat and protein composition, and less productive breeds are characterized by a higher milk nutrient content as well as a higher recovery of nutrients in the cheese. (3) High-intensive herds not only produce much more milk, but the milk has a higher nutrient content and a higher cheese yield, whereas within herds, compared with less productive cows, the more productive cows have a much greater milk yield, milk with a greater content of fat but not of protein, and a moderate improvement in cheese yield, differing little from expectations based on milk composition. Finally, (4) the effects of HL and CL on milk quality and cheese-making ability are similar but not identical in different breeds, the less productive ones having some advantage in terms of cheese-making ability. We can obtain FTIR-based prediction of cheese yield from individual milk samples retrospectively at population level, which seems to go beyond the simple knowledge of milk composition, incorporating information on nutrient retention ability in cheese, with possible advantages for management of farms, genetic improvement of dairy cows, and milk payment systems.

Evaluation of bull fertility in Italian Brown Swiss dairy cattle using cow field data.

Dairy bull fertility is traditionally evaluated using semen production and quality traits; however, these attributes explain only part of the differences observed in fertility among bulls. Alternatively, bull fertility can be directly evaluated using cow field data. The main objective of this study was to investigate bull fertility in the Italian Brown Swiss dairy cattle population using confirmed pregnancy records. The data set included a total of 397,926 breeding records from 1,228 bulls and 129,858 lactating cows between first and fifth lactation from 2000 to 2019. We first evaluated cow pregnancy success, including factors related to the bull under evaluation, such as bull age, bull inbreeding, and AI organization, and factors associated with the cow that receives the dose of semen, including herd-year-season, cow age, parity, and milk yield. We then estimated sire conception rate using only factors related to the bull. Model predictive ability was evaluated using 10-fold cross-validation with 10 replicates. Interestingly, our analyses revealed that there is a substantial variation in conception rate among Brown Swiss bulls, with more than 20% conception rate difference between high-fertility and low-fertility bulls. We also showed that the prediction of bull fertility is feasible as our cross-validation analyses achieved predictive correlations equal to 0.30 for sire conception rate. Improving reproduction performance is one of the major challenges of the dairy industry worldwide, and for this, it is essential to have accurate predictions of service sire fertility. This study represents the foundation for the development of novel tools that will allow dairy producers, breeders, and artificial insemination companies to make enhanced management and selection decisions on Brown Swiss male fertility.

Purebreeding with sexed semen and crossbreeding with semen from double-muscled sires to improve beef production from dairy herds: Live and slaughter performances of crossbred calves.

The use of sexed semen to produce purebred replacement heifers allows a large proportion of dairy cows to be mated to beef sires, and quantitative and qualitative improvements to be made to beef production from dairy herds. The major dairy and beef breeds are undergoing rapid genetic improvement as a result of more efficient selection methods, prompting a need to evaluate the meat production of crossbred beef × dairy cattle produced using current genetics. As part of a large project involving 125 commercial dairy farms, we evaluated the combined use of purebreeding with sexed semen and crossbreeding with semen from beef sires, particularly double-muscled breeds. A survey of 1,530 crossbred calves revealed that, whereas purebred dairy calves are destined almost exclusively for veal production, beef × dairy crossbred calves are also destined for beef production after fattening on either the dairy farm of birth or by specialized fatteners. In veal production, compared with Belgian Blue-sired calves (taken as the reference), double-muscled INRA 95-sired calves had a lighter slaughter weight (303 vs. 346 kg), but a greater dressing percent (62.3 vs. 58.4%). Limousin (also known as Limousine)-sired calves had a smaller average daily gain (1.26 vs. 1.34 kg/d), and lighter slaughter (314 vs. 346 kg) and carcass weights (182 vs. 201 kg). Last, Simmental-sired calves had a similar growth rate, but lighter carcass weight (177 vs. 201 kg), smaller dressing percentage (55.3 vs. 58.4%), and smaller muscularity scores (3.25 vs. 3.72). In the case of young bulls and heifers fattened on the dairy farm of birth, Belgian Blue-, Piemontese (also known as Piedmontese)-, and Limousin-sired calves performed similarly; the only exception was that Piemontese-sired calves had a greater dressing percentage. Belgian Blue- and Limousin-sired calves performed similarly when fattened by specialized beef producers. In both veal and beef production, the effects of dam breed were less important than sire breed. Considering the entire project, we can conclude that the combined use of sexed semen for purebreeding and conventional beef semen for terminal crossbreeding improves meat production from dairy herds, especially when the sires are double-muscled beef breeds.

Evaluating the performance of machine learning methods and variable selection methods for predicting difficult-to-measure traits in Holstein dairy cattle using milk infrared spectral data.

Fourier-transform infrared (FTIR) spectroscopy is a powerful high-throughput phenotyping tool for predicting traits that are expensive and difficult to measure in dairy cattle. Calibration equations are often developed using standard methods, such as partial least squares (PLS) regression. Methods that employ penalization, rank-reduction, and variable selection, as well as being able to model the nonlinear relations between phenotype and FTIR, might offer improvements in predictive ability and model robustness. This study aimed to compare the predictive ability of 2 machine learning methods, namely random forest (RF) and gradient boosting machine (GBM), and penalized regression against PLS regression for predicting 3 phenotypes differing in terms of biological meaning and relationships with milk composition (i.e., phenotypes measurable directly and not directly in milk, reflecting different biological processes which can be captured using milk spectra) in Holstein-Friesian cattle under 2 cross-validation scenarios. The data set comprised phenotypic information from 471 Holstein-Friesian cows, and 3 target phenotypes were evaluated: (1) body condition score (BCS), (2) blood ß-hydroxybutyrate (BHB, mmol/L), and (3) κ-casein expressed as a percentage of nitrogen (κ-CN, % N). The data set was split considering 2 cross-validation scenarios: samples-out random in which the population was randomly split into 10-folds (8-folds for training and 1-fold for validation and testing); and herd/date-out in which the population was randomly assigned to training (70% herd), validation (10%), and testing (20% herd) based on the herd and date in which the samples were collected. The random grid search was performed using the training subset for the hyperparameter optimization and the validation set was used for the generalization of prediction error. The trained model was then used to assess the final prediction in the testing subset. The grid search for penalized regression evidenced that the elastic net (EN) was the best regularization with increase in predictive ability of 5%. The performance of PLS (standard model) was compared against 2 machine learning techniques and penalized regression using 2 cross-validation scenarios. Machine learning methods showed a greater predictive ability for BCS (0.63 for GBM and 0.61 for RF), BHB (0.80 for GBM and 0.79 for RF), and κ-CN (0.81 for GBM and 0.80 for RF) in samples-out cross-validation. Considering a herd/date-out cross-validation these values were 0.58 (GBM and RF) for BCS, 0.73 (GBM and RF) for BHB, and 0.77 (GBM and RF) for κ-CN. The GBM model tended to outperform other methods in predictive ability around 4%, 1%, and 7% for EN, RF, and PLS, respectively. The prediction accuracies of the GBM and RF models were similar, and differed statistically from the PLS model in samples-out random cross-validation. Although, machine learning techniques outperformed PLS in herd/date-out cross-validation, no significant differences were observed in terms of predictive ability due to the large standard deviation observed for predictions. Overall, GBM achieved the highest accuracy of FTIR-based prediction of the different phenotypic traits across the cross-validation scenarios. These results indicate that GBM is a promising method for obtaining more accurate FTIR-based predictions for different phenotypes in dairy cattle.

The mineral profile affects the coagulation pattern and cheese-making efficiency of bovine milk.

Natural variations in milk minerals, their relationships, and their associations with the coagulation process and cheese-making traits present an opportunity for the differentiation of milk destined for high-quality natural products, such as traditional specialties or Protected Designation of Origin (PDO) cheeses. The aim of this study was to quantify the effects of the native contents of Ca, P, Na, K, and Mg on 18 traits describing traditional milk coagulation properties (MCP), curd firming over time (CFt) equation parameters, cheese yield (CY) measures, and nutrient recoveries in the curd (REC) using models that either included or omitted the simultaneous effects of milk fat and casein contents. The results showed that, by including milk fat and casein and the minerals in the statistical model, we were able to determine the specific effects of each mineral on coagulation and cheese-making efficiency. In general, about two-thirds of the apparent effects of the minerals on MCP and the CFt equation parameters are actually mediated by their association with milk composition, especially casein content, whereas only one-third of the effects are direct and independent of milk composition. In the case of cheese-making traits, the effects of the minerals were mediated only negligibly by their association with milk composition. High Ca content had a positive effect on the coagulation pattern and cheese-making traits, favoring water retention in the curd in particular. Phosphorus positively affected the cheese-making traits in that it was associated with an increase in CY in terms of curd solids, and in all the nutrient recovery traits. However, a very high P content in milk was associated with lower fat recovery in the curd. The variation in the Na content in milk only mildly affected coagulation, whereas with regard to cheese-making, protein recovery was negatively associated with high concentrations of this mineral. Potassium seemed not to be actively involved in coagulation and the cheese-making process. Magnesium content tended to slow coagulation and reduce CY measures. Further studies on the relationships of minerals with casein and protein fractions could deepen our knowledge of the role of all minerals in coagulation and the cheese-making process.

Structural equation modeling for unraveling the multivariate genomic architecture of milk proteins in dairy cattle.

The aims of this study were to investigate potential functional relationships among milk protein fractions in dairy cattle and to carry out a structural equation model (SEM) GWAS to provide a decomposition of total SNP effects into direct effects and effects mediated by traits that are upstream in a phenotypic network. To achieve these aims, we first fitted a mixed Bayesian multitrait genomic model to infer the genomic correlations among 6 milk nitrogen fractions [4 caseins (CN), namely κ-, ß-, αS1-, and αS2-CN, and 2 whey proteins, namely ß-lactoglobulin (ß-LG) and α-lactalbumin (α-LA)], in a population of 989 Italian Brown Swiss cows. Animals were genotyped with the Illumina BovineSNP50 Bead Chip v.2 (Illumina Inc.). A Bayesian network approach using the max-min hill-climbing (MMHC) algorithm was implemented to model the dependencies or independence among traits. Strong and negative genomic correlations were found between ß-CN and αS1-CN (-0.706) and between ß-CN and κ-CN (-0.735). The application of the MMHC algorithm revealed that κ-CN and ß-CN seemed to directly or indirectly influence all other milk protein fractions. By integrating multitrait model GWAS and SEM-GWAS, we identified a total of 127 significant SNP for κ-CN, 89 SNP for ß-CN, 30 SNP for αS1-CN, and 14 SNP for αS2-CN (mostly shared among CN and located on Bos taurus autosome 6) and 15 SNP for ß-LG (mostly located on Bos taurus autosome 11), whereas no SNP passed the significance threshold for α-LA. For the significant SNP, we assessed and quantified the contribution of direct and indirect paths to total marker effect. Pathway analyses confirmed that common regulatory mechanisms (e.g., energy metabolism and hormonal and neural signals) are involved in the control of milk protein synthesis and metabolism. The information acquired might be leveraged for setting up optimal management and selection strategies aimed at improving milk quality and technological characteristics in dairy cattle.

Genomic prediction for latent variables related to milk fatty acid composition in Holstein, Simmental and Brown Swiss dairy cattle breeds.

Genomic selection (GS) reports on milk fatty acid (FA) profiles have been published quite recently and are still few despite this trait represents the most important aspect of milk nutritional and sensory quality. Reasons for this can be found in the high costs of phenotype recording but also in issues related to its nature of complex trait constituted by multiple genetically correlated variables with low heritabilities. One possible strategy to deal with such constraint is represented by the use of dimension reduction methods. We analysed 40 individual FAs from Italian Brown Swiss, Holstein and Simmental milk through multivariate factor analysis (MFA) to study the genetics of milk FA-related latent variables (factors) and assess their potential use in breeding. A total of nine factors were obtained, and their genetic parameters were inferred under a Bayesian framework using two statistical approaches: the classical pedigree best linear unbiased prediction (ABLUP) and the single-step genomic BLUP (ssGBLUP). The resulting factorial solutions were able to represent groups of FAs with common origin and function and can be considered concise pathway-level phenotypes. The heritability (h2 ) values showed relevant variations across different factors in each breed (0.03 ≤ h2  ≤ 0.38). The accuracies of breeding values predicted were low to high, ranging from 0.13 to 0.72 and from 0.18 to 0.74 considering the pedigree and the genomic model, respectively. The gain in accuracy in genetic prediction due to the addition of genomic information was ~30% and ~5% in validation and training groups respectively, confirming the contribution of genomic information in yielding more accurate predictions compared to the traditional ABLUP methodology. Our results suggest that MFA in combination with GS can be a valuable tool in dairy cattle breeding and deserves to be further investigated for use in future breeding programs to improve cow's milk FA-related traits.

Quantitative and qualitative detailed milk protein profiles of 6 cattle breeds: Sources of variation and contribution of protein genetic variants.

Different fractions of milk nitrogenous compounds (not only caseins) have different effects on the nutritional value of milk, its coagulation and curd firming properties, and its cheese-making efficiency. To assess different sources of variation, especially the cows' breed and genetic variants of the main protein fractions, milk samples were collected from 1,504 cows belonging to 3 dairy breeds (Holstein-Friesian, Brown Swiss, and Jersey) and 3 dual-purpose breeds (Simmental, Rendena, and Alpine Grey) reared in 41 multibreed herds. Beyond crude protein, casein (CN), and urea, 7 protein fractions were analyzed using HPLC, and 5 other N fraction traits were calculated. All 15 traits were measured qualitatively (% of milk N) and quantitatively (g/L of milk). The HPLC technique allowed us to discriminate between the main genetic variants of ß-CN, κ-CN, and ß-lactoglobulin and thus to genotype the cows for the CSN2, CSN3, and BLG genes, respectively. Data were analyzed using 2 mixed models, both including the effects of herd-date, breed, parity, and lactation stage, and only one also including the effects of the genotypes of the milk proteins. Breed of cow explained 2 to 36% of phenotypic variability for all the N fractions, with the exception of the urea and total casein contents of milk and the urea and ß-CN proportions of total milk N. Lactation stage had a considerable influence on the amount (g/L) of almost all the protein fractions in milk, but neither the nonprotein N fractions nor the percentage of milk N protein profile were affected. The inclusion of the CSN2, CSN3, and BLG genotypes in the model explained a large part of the total variability in all the milk protein and nonprotein fractions except urea. It also reduced the variance explained by breed and residual factors. An exception was shown by the proportion of αS1-CN variance explained by breed that moved from 13 to 28%. Similarly, for amount (g/L) of ß-CN, the effect of breed became significant (12%), whereas it was almost null before inclusion of genotypes. In terms of percentage of milk N, the genotypes of CSN3 notably affected all the casein fractions, whereas the BLG genotypes had a much greater influence on most noncasein traits. The genotypes of the CSN2 gene exerted an appreciable effect on αS2-CN and not ß-CN, as expected. Comparing the 2 models, we were also able to discriminate the effect of the breed on a milk N fraction, both quantitatively and qualitatively, in 2 quotas: the first due to the milk protein polymorphisms (major genes) and the second due to other genetic factors (polygene), after correcting for the effect of herd-date of sampling, parity, and lactation stage. The knowledge about the detailed milk protein profile of different cattle breeds provided by this study could be of great benefit for the dairy industry, providing new tools for the enhancement of milk payment systems and breeding program designs.