Prediction of key milk biomarkers in dairy cows through milk mid-infrared spectra and international collaborations.

At the individual cow level, suboptimum fertility, mastitis, negative energy balance, and ketosis are major issues in dairy farming. These problems are widespread on dairy farms and have an important economic impact. The objectives of this study were (1) to assess the potential of milk mid-infrared (MIR) spectra to predict key biomarkers of energy deficit (citrate, isocitrate, glucose-6 phosphate [glucose-6P], free glucose), ketosis (ß-hydroxybutyrate [BHB] and acetone), mastitis (N-acetyl-ß-d-glucosaminidase activity [NAGase] and lactate dehydrogenase), and fertility (progesterone); (2) to test alternative methodologies to partial least squares (PLS) regression to better account for the specific asymmetric distribution of the biomarkers; and (3) to create robust models by merging large datasets from 5 international or national projects. Benefiting from this international collaboration, the dataset comprised a total of 9,143 milk samples from 3,758 cows located in 589 herds across 10 countries and represented 7 breeds. The samples were analyzed by reference chemistry for biomarker contents, whereas the MIR analyses were performed on 30 instruments from different models and brands, with spectra harmonized into a common format. Four quantitative methodologies were evaluated to address the strongly skewed distribution of some biomarkers. Partial least squares regression was used as the reference basis, and compared with a random modification of distribution associated with PLS (random-downsampling-PLS), an optimized modification of distribution associated with PLS (KennardStone-downsampling-PLS), and support vector machine (SVM). When the ability of MIR to predict biomarkers was too low for quantification, different qualitative methodologies were tested to discriminate low versus high values of biomarkers. For each biomarker, 20% of the herds were randomly removed within all countries to be used as the validation dataset. The remaining 80% of herds were used as the calibration dataset. In calibration, the 3 alternative methodologies outperform the PLS performances for the majority of biomarkers. However, in the external herd validation, PLS provided the best results for isocitrate, glucose-6P, free glucose, and lactate dehydrogenase (coefficient of determination in external herd validation [R2v] = 0.48, 0.58, 0.28, and 0.24, respectively). For other molecules, PLS-random-downsampling and PLS-KennardStone-downsampling outperformed PLS in the majority of cases, but the best results were provided by SVM for citrate, BHB, acetone, NAGase, and progesterone (R2v = 0.94, 0.58, 0.76, 0.68, and 0.15, respectively). Hence, PLS and SVM based on the entire dataset provided the best results for normal and skewed distributions, respectively. Complementary to the quantitative methods, the qualitative discriminant models enabled the discrimination of high and low values for BHB, acetone, and NAGase with a global accuracy around 90%, and glucose-6P with an accuracy of 83%. In conclusion, MIR spectra of milk can enable quantitative screening of citrate as a biomarker of energy deficit and discrimination of low and high values of BHB, acetone, and NAGase, as biomarkers of ketosis and mastitis. Finally, progesterone could not be predicted with sufficient accuracy from milk MIR spectra to be further considered. Consequently, MIR spectrometry can bring valuable information regarding the occurrence of energy deficit, ketosis, and mastitis in dairy cows, which in turn have major influences on their fertility and survival.

Genome-wide association study and functional annotation analyses for nitrogen efficiency index and its composition traits in dairy cattle.

The aims of this study were (1) to identify genomic regions associated with a N efficiency index (NEI) and its composition traits and (2) to analyze the functional annotation of identified genomic regions. The NEI included N intake (NINT1), milk true protein N (MTPN1), milk urea N yield (MUNY1) in primiparous cattle, and N intake (NINT2+), milk true protein N (MTPN2+), and milk urea N yield (MUNY2+) in multiparous cattle (2 to 5 parities). The edited data included 1,043,171 records on 342,847 cows distributed in 1,931 herds. The pedigree consisted of 505,125 animals (17,797 males). Data of 565,049 SNPs were available for 6,998 animals included in the pedigree (5,251 females and 1,747 males). The SNP effects were estimated using a single-step genomic BLUP approach. The proportion of the total additive genetic variance explained by windows of 50 consecutive SNPs (with an average size of about 240 kb) was calculated. The top 3 genomic regions explaining the largest rate of the total additive genetic variance of the NEI and its composition traits were selected for candidate gene identification and quantitative trait loci (QTL) annotation. The selected genomic regions explained from 0.17% (MTPN2+) to 0.58% (NEI) of the total additive genetic variance. The largest explanatory genomic regions of NEI, NINT1, NINT2+, MTPN1, MTPN2+, MUNY1, and MUNY2+ were Bos taurus autosome 14 (1.52-2.09 Mb), 26 (9.24-9.66 Mb), 16 (75.41-75.51 Mb), 6 (8.73-88.92 Mb), 6 (8.73-88.92 Mb), 11 (103.26-103.41 Mb), 11 (103.26-103.41 Mb). Based on the literature, gene ontology, Kyoto Encyclopedia of Genes and Genomes, and protein-protein interaction, 16 key candidate genes were identified for NEI and its composition traits, which are mainly expressed in the milk cell, mammary, and liver tissues. The number of enriched QTL related to NEI, NINT1, NINT2+, MTPN1, and MTPN2+ were 41, 6, 4, 11, 36, 32, and 32, respectively, and most of them were related to the milk, health, and production classes. In conclusion, this study identified genomic regions associated with NEI and its composition traits, and identified key candidate genes describing the genetic mechanisms of N use efficiency-related traits. Furthermore, the NEI reflects not only its composition traits but also the interactions among them.

Imputation of missing milk Fourier transform mid-infrared spectra using existing milk spectral databases: A strategy to improve the reliability of breeding values and predictive models.

The use of milk Fourier transform mid-infrared (FT-MIR) spectrometry to develop management and breeding tools for dairy farmers and industry is growing and supported by the availability of numerous new predicted phenotypes to assess the nutritional quality of milk and its technological properties, but also the animal health and welfare status and its environmental fingerprint. For genetic evaluations, having a long-term and representative spectral dairy herd improvement (DHI) database improves the reliabilities of estimated breeding values (EBV) from these phenotypes. Unfortunately, most of the time, the raw spectral data used to generate these estimations are not stored. Moreover, many reference measurements of those phenotypes, needed during the FT-MIR calibration step, are available from past research activities but lack spectra records. So, it is impossible to use them to improve the FT-MIR models. Consequently, there is a strong interest in imputing those missing spectra. The innovative objective of this study was to use the existing large spectral DHI database to estimate missing spectra by selecting probable spectra using, as the match criteria, common dairy traits recorded for a long time by DHI organizations. We tested 4 match criteria combinations. Combination 1 required to have equal fat and protein contents between the sample for which a spectrum was to be estimated and the reference samples in the DHI database. Combination 2 also required an equal urea content. Combination 3 requested equal fat, protein, and lactose contents. Finally, combination 4 included all criteria. When more than one spectrum was found during the search, their average was the estimated spectrum for the query sample. Concretely, this study estimated missing spectra for 1,700 samples using 2,000,000 spectral DHI records. For assessing the effect of this spectral estimation on the prediction quality, FT-MIR equations were used to predict 11 phenotypes, selected as their quantification used different FT-MIR regions. They were related to the milk fat and mineral composition, lactoferrin content, quantity of eructed methane, body weight (BW), and dry matter intake. The accuracy between predictions obtained from actual and estimated spectra was evaluated by calculating the mean absolute error (MAE). The criteria in the fourth and second combinations were too strict to estimate a spectrum for most samples. Indeed, for many samples, no spectra with the same values for those matching criteria was found. The third match criteria combination had a poorer prediction performance for all studied traits and spectral absorptions than the first combination due to fewer matched samples available to compute the missing spectrum. By allowing a range for matching lactose content (±0.1 g/dL milk), we showed that this new combination increased the number of selected samples to compute missing spectra and predict better the infrared absorption at different wavenumbers, especially those related to the lactose quantification. The prediction performance was further improved by performing queries on the entire Walloon DHI spectral database (6,625,570 spectra), and it varied among the studied phenotypes. Without considering the traits used for the matching, the best predictions were obtained for the content of saturated fatty acids (MAE = 0.15 g/dL milk) and BW (MAE = 12.80 kg). Yet, the predictions for the unsaturated fatty acids were less accurate (MAE = 0.13 and 0.018 g/dL milk for monounsaturated and polyunsaturated fatty acids), likely because of the poorer predictions of spectral regions related to long-chain fatty acids. Similarly, poorer predictions were observed for the amount of methane eructed by dairy cows (MAE = 47.02 g/d), likely because it is not directly related to fat content or composition. Prediction accuracies for the remaining traits were also low. In conclusion, we observed that increasing the number of relevant matching criteria helps improve the quality of FT-MIR predicted phenotypes and the number of spectra used during the search. So, it would be of great interest to test in the future the suitability of the developed methodology with large-scale international spectral databases to improve the reliability of EBV from these FT-MIR-based phenotypes and the robustness of FT-MIR predictive models.

Validating genomic prediction for nitrogen efficiency index and its composition traits of Holstein cows in early lactation.

Nitrogen (N) use efficiency (NUE) is an economically important trait for dairy cows. Recently, we proposed a new N efficiency index (NEI), that simultaneously considers both NUE and N pollution. This study aimed to validate the genomic prediction for NEI and its composition traits and investigate the relationship between SNP effects estimated directly from NEI and indirectly from its composition traits. The NEI composition included genomic estimated breeding value of N intake (NINT), milk true protein N (MTPN) and milk urea N yield. The edited data were 132,899 records on 52,064 cows distributed in 773 herds. The pedigree contained 122,368 animals. Genotypic data of 566,294 SNP was available for 4514 individuals. A total of 4413 cows (including 181 genotyped) and 56 bulls (including 32 genotyped) were selected as the validation populations. The linear regression method was used to validate the genomic prediction of NEI and its composition traits using best linear unbiased prediction (BLUP) and single-step genomic BLUP (ssGBLUP). The mean theoretical accuracies of validation populations obtained from ssGBLUP were higher than those obtained from BLUP for both NEI and its composition traits, ranging from 0.57 (MTPN) to 0.72 (NINT). The highest mean prediction accuracies for NEI and its composition traits were observed for the genotyped cows estimated under ssGBLUP, ranging from 0.48 (MTPN) to 0.66 (NINT). Furthermore, the SNP effects estimated from NEI composition traits, multiplied by the relative weight were the same as those estimated directly from NEI. This study preliminary showed that genomic prediction can be used for NEI, however, we acknowledge the need for further validation of this result in a larger dataset. Moreover, the SNP effects of NEI can be indirectly calculated using the SNP effects estimated from its composition traits. This study provided a basis for adding genomic information to establish NEI as part of future routine genomic evaluation programs.

Large-scale phenotyping in dairy sector using milk MIR spectra: Key factors affecting the quality of predictions.

Methods and technologies enabling the estimation at large scale of important traits for the dairy sector are of great interest. Those phenotypes are necessary to improve herd management, animal genetic evaluation, and milk quality control. In the recent years, the research was very active to predict new phenotypes from the mid-infrared (MIR) analysis of milk. Models were developed to predict phenotypes such as fine milk composition, milk technological properties or traits related to cow health, fertility and environmental impact. Most of models were developed within research contexts and often not designed for routine use. The implementation of models at a large scale to predict new traits of interest brings new challenges as the factors influencing the robustness of models are poorly documented. The first objective of this work is to highlight the impact on prediction accuracy of factors such as the variability of the spectral and reference data, the spectral regions used and the complexity of models. The second objective is to emphasize methods and indicators to evaluate the quality of models and the quality of predictions generated under routine conditions. The last objective is to outline the issues and the solutions linked with the use and transfer of models on large number of instruments. Based on partial least square regression and 10 datasets including milk MIR spectra and reference quantitative values for 57 traits of interest, the impact of the different factors is illustrated by evaluating the influence on the validation root mean square error of prediction (RMSEP). In the displayed examples, all factors, when well set up, increase the quality of predictions, with an improvement of the RMSEP ranging from 12% to 43%. This work also aims to underline the need for and the complementarity between different validation procedures, statistical parameters and quality assurance methods. Finally, when using and transferring models, the impact of the spectral standardization on the prediction reproducibility is highlighted with an improvement up to 86% with the tested models, and the monitoring of individual spectrometer stability over time appears essential. This list inspired from our experience is of course not exhaustive. The displayed results are only examples and not general rules and other aspects play a role in the quality of final predictions. However, this work highlights good practices, methods and indicators to increase and evaluate quality of phenotypes predicted at a large scale. The results obtained argue for the development of guidelines at international levels, as well as international collaborations in order to constitute large and robust datasets and enable the use of models in routine conditions.

Can unsupervised learning methods applied to milk recording big data provide new insights into dairy cow health?

Among the dairy sector's current concerns, the assessment of global animal health status is a complex challenge. Its multidimensionality means that global monitoring tools are rarely considered. Instead, specific disease detection is often studied separately and, due to financial and ethical issues, uses small-scale data sets focusing on few biomarkers. Several studies have already been conducted using milk Fourier transform mid-infrared (FT-MIR) spectroscopy to detect mastitis and lameness or to quantify health-related biomarkers in milk or blood. Those studies are relevant but they focus mainly on one biomarker or disease. To solve this issue and the small-scale data set, in this study, we proposed a holistic approach using big data obtained from milk recording, including milk yield, somatic cell count, and 27 FT-MIR-based predictors related to milk composition and animal health status. Using 740,454 records collected from 114,536 first-parity Holstein cows in southern Belgium, we performed repeated unsupervised learning algorithms based on Ward's agglomerative hierarchical clustering method to find potential interesting patterns. A divide-and-conquer approach was used to overcome the limitation of computational resources in clustering a relatively large data set. Five groups of records were identified. Differences observed in the fourth group suggested a relationship to metabolic disorders. The fifth group seemed to be related to mastitis. In a second step, we performed a partial least squares discriminant analysis (PLS-DA) to predict the probability of belonging to those specific groups for the entire data set. The obtained global accuracy was 0.77 and the balanced accuracy (i.e., the mean between sensitivity and specificity) of discriminating the fourth and fifth groups was 0.88 and 0.96, respectively. Then, a validation of the interpretation of those groups was performed using 204 milk and blood reference records. The predicted probability associated with the metabolic disorders issue had significant correlations of 0.54 with blood ß-hydroxybutyrate, 0.44 with blood nonesterified fatty acids, -0.32 with blood glucose, -0.23 with milk glucose-6-phosphate, and 0.38 with milk isocitrate. In contrast, the predicted probability of belonging to the mastitis group had correlations of 0.69 with milk lactate dehydrogenase, 0.46 with milk N-acetyl-ß-d-glucosaminidase, -0.18 with milk free glucose, and 0.16 with milk glucose-6-phosphate. Consequently, these results suggest that the obtained quantitative traits indirectly reflect some of the main health disorders in dairy farming and could be used to monitor dairy cows on a large scale. By using unsupervised learning on large-scale milk recording data and then validating the pattern using reference laboratory measures, we propose a new approach to quickly assess dairy cow health status.

Defining a nitrogen efficiency index in Holstein cows and assessing its potential effect on the breeding program of bulls.

The purposes of this study were (1) to explore the relationship between 3 milk mid-infrared predicted features including nitrogen intake (NINT), milk true protein N (MTPN), and milk urea-N yield (MUNY); (2) to integrate these 3 features into an N efficiency index (NEI) and analyses approximate genetic correlations between the NEI and 37 traits (indices) of interest; and (3) to assess the potential effect of including the NEI into breeding programs of bulls. The edited data were 1,043,171 test-day records on 342,847 cows in 1,931 herds and 143,595 test-day records on 53,660 cows in 766 herds used for estimating breeding values (EBV) and variance components, respectively. The used records were within 5 to 50 d in milk. The records were grouped into primiparous and multiparous. The genetic parameters for the included mid-infrared features and EBV of the animals included in the pedigree were estimated using a multiple-trait repeatability animal model. Then, the EBV of the NINT, MTPN, MUNY were integrated into the NEI using a selection index assuming weights based on the N partitioning. The approximate genetic correlations between the NEI and 37 traits of interest were estimated using the EBV of the selected bulls. The bulls born from 2011 to 2014 with NEI were selected and the NEI distribution of these bulls having EBV for the 8 selected traits (indices) was checked. The heritability and repeatability estimates for NINT, MTPN, and MUNY ranged from 0.09 to 0.13, and 0.37 to 0.65, respectively. The genetic and phenotypic correlations between NINT, MTPN, and MUNY ranged from -0.31 to 0.87, and -0.02 to 0.42, respectively. The NEI ranged from -13.13 to 12.55 kg/d. In total, 736 bulls with reliability ≥0.50 for all included traits (NEI and 37 traits) and at least 10 daughters distributed in at least 10 herds were selected to investigate genetic aspects of the NEI. The NEI had positive genetic correlations with production yield traits (0.08-0.46), and negative genetic correlations with the investigated functional traits and indices (-0.71 to -0.07), except for the production economic index and functional type economic index. The daughters of bulls with higher NEI had lower NINT and MUNY, and higher MTPN. Furthermore, 26% of the bulls (n = 50) with NEI born between 2011 to 2014 had higher NEI and global economic index than the average in the selected bulls. Finally, the developed NEI has the advantage of large-scale prediction and therefore has the potential for routine application in dairy cattle breeding in the future.

Estimation of genetic parameters for predicted nitrogen use efficiency and losses in early lactation of Holstein cows.

The objective of this study was to estimate genetic parameters of predicted N use efficiency (PNUE) and N losses (PNL) as proxies of N use and loss for Holstein cows. Furthermore, we have assessed approximate genetic correlations between PNUE, PNL, and dairy production, health, longevity, and conformation traits. These traits are considered important in many countries and are currently evaluated by the International Bull Evaluation Service (Interbull). The values of PNUE and PNL were obtained by using the combined milk mid-infrared (MIR) spectrum, parity, and milk yield-based prediction equations on test-day MIR records with days in milk (DIM) between 5 and 50 d. After editing, the final data set comprised 46,163 records of 21,462 cows from 154 farms in 5 countries. Each trait was divided into primiparous and multiparous (including second to fifth parity) groups. Genetic parameters and breeding values were estimated by using a multitrait (2-trait, 2-parity classes) repeatability model. Herd-year-season of calving, DIM, age of calving, and parity were used as fixed effects. Random effects were defined as parity (within-parity permanent environment), nongenetic cow (across-parity permanent environment), additive genetic animal, and residual effects. The estimated heritability of PNUE and PNL in the first and later parity were 0.13, 0.12, 0.14, and 0.13, and the repeatability values were 0.49, 0.40, 0.55, and 0.43, respectively. The estimated approximate genetic correlations between PNUE and PNL were negative and high (from -0.89 to -0.53), whereas the phenotypic correlations were also negative but relatively low (from -0.45 to -0.11). At a level of reliability of more than 0.30 for all novel traits, a total of 504 bulls born after 1995 had also publishable Interbull multiple-trait across-country estimated breeding values (EBV). The approximate genetic correlations between PNUE and the other 30 traits of interest, estimated as corrected correlations between EBV of bulls, ranged from -0.46 (udder depth) to 0.47 (milk yield). Obtained results showed the complex genetic relationship between efficiency, production, and other traits: for instance, more efficient cows seem to give more milk, which is linked to deeper udders, but seem to have lower health, fertility, and longevity. Additionally, the approximate genetic correlations between PNL, lower values representing less loss of N, and the 30 other traits, were from -0.32 (angularity) to 0.57 (direct calving ease). Even if further research is needed, our results provided preliminary evidence that the PNUE and PNL traits used as proxies could be included in genetic improvement programs in Holstein cows and could help their management.

Development of a subacute ruminal acidosis risk score and its prediction using milk mid-infrared spectra in early-lactation cows.

A routine monitoring for subacute ruminal acidosis (SARA) on the individual level could support the minimization of economic losses and the ensuring of animal welfare in dairy cows. The objectives of this study were (1) to develop a SARA risk score (SRS) by combining information from different data acquisition systems to generate an integrative indicator trait, (2) the investigation of associations of the SRS with feed analysis data, blood characteristics, performance data, and milk composition, including the fatty acid (FA) profile, (3) the development of a milk mid-infrared (MIR) spectra-based prediction equation for this novel reference trait SRS, and (4) its application to an external data set consisting of MIR data of test day records to investigate the association between the MIR-based predictions of the SRS and the milk FA profile. The primary data set, which was used for the objectives (1) to (3), consisted of data collected from 10 commercial farms with a total of 100 Holstein cows in early lactation. The data comprised barn climate parameters, pH and temperature logging from intrareticular measurement boluses, as well as jaw movement and locomotion behavior recordings of noseband-sensor halters and pedometers. Further sampling and data collection included feed samples, blood samples, milk performance, and milk samples, whereof the latter were used to get the milk MIR spectra and to estimate the main milk components, the milk FA profile, and the lactoferrin content. Because all measurements were characterized by different temporal resolutions, the data preparation consisted of an aggregation into values on a daily basis and merging it into one data set. For the development of the SRS, a total of 7 traits were selected, which were derived from measurements of pH and temperature in the reticulum, chewing behavior, and milk yield. After adjustment for fixed effects and standardization, these 7 traits were combined into the SRS using a linear combination and directional weights based on current knowledge derived from literature studies. The secondary data set was used for objective (4) and consisted of test day records of the entire herds, including performance data, milk MIR spectra and MIR-predicted FA. At farm level, it could be shown that diets with higher proportions of concentrated feed resulted in both lower daily mean pH and higher SRS values. On the individual level, an increased SRS could be associated with a modified FA profile (e.g., lower levels of short- and medium-chain FA, higher levels of C17:0, odd- and branched-chain FA). Furthermore, a milk MIR-based partial least squares regression model with a moderate predictability was established for the SRS. This work provides the basis for the development of routine SARA monitoring and demonstrates the high potential of milk composition-based assessment of the health status of lactating cows.

A comparison of 4 different machine learning algorithms to predict lactoferrin content in bovine milk from mid-infrared spectra.

Lactoferrin (LF) is a glycoprotein naturally present in milk. Its content varies throughout lactation, but also with mastitis; therefore it is a potential additional indicator of udder health beyond somatic cell count. Condequently, there is an interest in quantifying this biomolecule routinely. First prediction equations proposed in the literature to predict the content in milk using milk mid-infrared spectrometry were built using partial least square regression (PLSR) due to the limited size of the data set. Thanks to a large data set, the current study aimed to test 4 different machine learning algorithms using a large data set comprising 6,619 records collected across different herds, breeds, and countries. The first algorithm was a PLSR, as used in past investigations. The second and third algorithms used partial least square (PLS) factors combined with a linear and polynomial support vector regression (PLS + SVR). The fourth algorithm also used PLS factors, but included in an artificial neural network with 1 hidden layer (PLS + ANN). The training and validation sets comprised 5,541 and 836 records, respectively. Even if the calibration prediction performances were the best for PLS + polynomial SVR, their validation prediction performances were the worst. The 3 other algorithms had similar validation performances. Indeed, the validation root mean squared error (RMSE) ranged between 162.17 and 166.75 mg/L of milk. However, the lower standard deviation of cross-validation RMSE and the better normality of the residual distribution observed for PLS + ANN suggest that this modeling was more suitable to predict the LF content in milk from milk mid-infrared spectra (R2v = 0.60 and validation RMSE = 162.17 mg/L of milk). This PLS +ANN model was then applied to almost 6 million spectral records. The predicted LF showed the expected relationships with milk yield, somatic cell score, somatic cell count, and stage of lactation. The model tended to underestimate high LF values (higher than 600 mg/L of milk). However, if the prediction threshold was set to 500 mg/L, 82% of samples from the validation having a content of LF higher than 600 mg/L were detected. Future research should aim to increase the number of those extremely high LF records in the calibration set.

Potential of milk mid-infrared spectra to predict nitrogen use efficiency of individual dairy cows in early lactation.

Improving nitrogen use efficiency (NUE) at both the individual cow and the herd level has become a key target in dairy production systems, for both environmental and economic reasons. Cost-effective and large-scale phenotyping methods are required to improve NUE through genetic selection and by feeding and management strategies. The aim of this study was to evaluate the possibility of using mid-infrared (MIR) spectra of milk to predict individual dairy cow NUE during early lactation. Data were collected from 129 Holstein cows, from calving until 50 d in milk, in 3 research herds (Denmark, Ireland, and the UK). In 2 of the herds, diets were designed to challenge cows metabolically, whereas a diet reflecting local management practices was offered in the third herd. Nitrogen intake (kg/d) and nitrogen excreted in milk (kg/d) were calculated daily. Nitrogen use efficiency was calculated as the ratio between nitrogen in milk and nitrogen intake, and expressed as a percentage. Individual daily values for NUE ranged from 9.7 to 81.7%, with an average of 36.9% and standard deviation of 10.4%. Milk MIR spectra were recorded twice weekly and were standardized into a common format to avoid bias between apparatus or sampling periods. Regression models predicting NUE using milk MIR spectra were developed on 1,034 observations using partial least squares or support vector machines regression methods. The models were then evaluated through (1) a cross-validation using 10 subsets, (2) a cow validation excluding 25% of the cows to be used as a validation set, and (3) a diet validation excluding each of the diets one by one to be used as validation sets. The best statistical performances were obtained when using the support vector machines method. Inclusion of milk yield and lactation number as predictors, in combination with the spectra, also improved the calibration. In cross-validation, the best model predicted NUE with a coefficient of determination of cross-validation of 0.74 and a relative error of 14%, which is suitable to discriminate between low- and high-NUE cows. When performing the cow validation, the relative error remained at 14%, and during the diet validation the relative error ranged from 12 to 34%. In the diet validation, the models showed a lack of robustness, demonstrating difficulties in predicting NUE for diets and for samples that were not represented in the calibration data set. Hence, a need exists to integrate more data in the models to cover a maximum of variability regarding breeds, diets, lactation stages, management practices, seasons, MIR instruments, and geographic regions. Although the model needs to be validated and improved for use in routine conditions, these preliminary results showed that it was possible to obtain information on NUE through milk MIR spectra. This could potentially allow large-scale predictions to aid both further genetic and genomic studies, and the development of farm management tools.

Body and milk traits as indicators of dairy cow energy status in early lactation.

The inclusion of feed intake and efficiency traits in dairy cow breeding goals can lead to increased risk of metabolic stress. An easy and inexpensive way to monitor postpartum energy status (ES) of cows is therefore needed. Cows' ES can be estimated by calculating the energy balance from energy intake and output and predicted by indicator traits such as change in body weight (ΔBW), change in body condition score (ΔBCS), milk fat:protein ratio (FPR), or milk fatty acid (FA) composition. In this study, we used blood plasma nonesterified fatty acids (NEFA) concentration as a biomarker for ES. We determined associations between NEFA concentration and ES indicators and evaluated the usefulness of body and milk traits alone, or together, in predicting ES of the cow. Data were collected from 2 research herds during 2013 to 2016 and included 137 Nordic Red dairy cows, all of which had a first lactation and 59 of which also had a second lactation. The data included daily body weight, milk yield, and feed intake and monthly BCS. Plasma samples for NEFA were collected twice in lactation wk 2 and 3 and once in wk 20. Milk samples for analysis of fat, protein, lactose, and FA concentrations were taken on the blood sampling days. Plasma NEFA concentration was higher in lactation wk 2 and 3 than in wk 20 (0.56 ± 0.30, 0.43 ± 0.22, and 0.13 ± 0.06 mmol/L, respectively; all means ± standard deviation). Among individual indicators, C18:1 cis-9 and the sum of C18:1 in milk had the highest correlations (r = 0.73) with NEFA. Seven multiple linear regression models for NEFA prediction were developed using stepwise selection. Of the models that included milk traits (other than milk FA) as well as body traits, the best fit was achieved by a model with milk yield, FPR, ΔBW, ΔBCS, FPR × ΔBW, and days in milk. The model resulted in a cross-validation coefficient of determination (R2cv) of 0.51 and a root mean squared error (RMSE) of 0.196 mmol/L. When only milk FA concentrations were considered in the model, NEFA prediction was more accurate using measurements from evening milk than from morning milk (R2cv = 0.61 vs. 0.53). The best model with milk traits contained FPR, C10:0, C14:0, C18:1 cis-9, C18:1 cis-9 × C14:0, and days in milk (R2cv = 0.62; RMSE = 0.177 mmol/L). The most advanced model using both milk and body traits gave a slightly better fit than the model with only milk traits (R2cv = 0.63; RMSE = 0.176 mmol/L). Our findings indicate that ES of cows in early lactation can be monitored with moderately high accuracy by routine milk measurements.

Prediction of metabolic clusters in early-lactation dairy cows using models based on milk biomarkers.

The aim of this study was to describe metabolism of early-lactation dairy cows by clustering cows based on glucose, insulin-like growth factor I (IGF-I), free fatty acid, and ß-hydroxybutyrate (BHB) using the k-means method. Predictive models for metabolic clusters were created and validated using 3 sets of milk biomarkers (milk metabolites and enzymes, glycans on the immunogamma globulin fraction of milk, and Fourier-transform mid-infrared spectra of milk). Metabolic clusters are used to identify dairy cows with a balanced or imbalanced metabolic profile. Around 14 and 35 d in milk, serum or plasma concentrations of BHB, free fatty acids, glucose, and IGF-I were determined. Cows with a favorable metabolic profile were grouped together in what was referred to as the "balanced" group (n = 43) and were compared with cows in what was referred to as the "other balanced" group (n = 64). Cows with an unfavorable metabolic profile were grouped in what was referred to as the "imbalanced" group (n = 19) and compared with cows in what was referred to as the "other imbalanced" group (n = 88). Glucose and IGF-I were higher in balanced compared with other balanced cows. Free fatty acids and BHB were lower in balanced compared with other balanced cows. Glucose and IGF-I were lower in imbalanced compared with other imbalanced cows. Free fatty acids and BHB were higher in imbalanced cows. Metabolic clusters were related to production parameters. There was a trend for a higher daily increase in fat- and protein-corrected milk yield in balanced cows, whereas that of imbalanced cows was higher. Dry matter intake and the daily increase in dry matter intake were higher in balanced cows and lower in imbalanced cows. Energy balance was continuously higher in balanced cows and lower in imbalanced cows. Weekly or twice-weekly milk samples were taken and milk metabolites and enzymes (milk glucose, glucose-6-phosphate, BHB, lactate dehydrogenase, N-acetyl-ß-d-glucosaminidase, isocitrate), immunogamma globulin glycans (19 peaks), and Fourier-transform mid-infrared spectra (1,060 wavelengths reduced to 15 principal components) were determined. Milk biomarkers with or without additional cow information (days in milk, parity, milk yield features) were used to create predictive models for the metabolic clusters. Accuracy for prediction of balanced (80%) and imbalanced (88%) cows was highest using milk metabolites and enzymes combined with days in milk and parity. The results and models of the present study are part of the GplusE project and identify novel milk-based phenotypes that may be used as predictors for metabolic and performance traits in early-lactation dairy cows.

Standardization of milk mid-infrared spectrometers for the transfer and use of multiple models.

An increasing number of models are being developed to provide information from milk Fourier transform mid-infrared (FT-MIR) spectra on fine milk composition, technological properties of milk, or even cows' physiological status. In this context, and to take advantage of these existing models, the purpose of this work was to evaluate whether a spectral standardization method can enable the use of multiple equations within a network of different FT-MIR spectrometers. The piecewise direct standardization method was used, matching "slave" instruments to a common reference, the "master." The effect of standardization on network reproducibility was assessed on 66 instruments from 3 different brands by comparing the spectral variability of the slaves and the master with and without standardization. With standardization, the global Mahalanobis distance from the slave spectra to the master spectra was reduced on average from 2,655.9 to 14.3, representing a significant reduction of noninformative spectral variability. The transfer of models from instrument to instrument was tested using 3 FT-MIR models predicting (1) the quantity of daily methane emitted by dairy cows, (2) the concentration of polyunsaturated fatty acids in milk, and (3) the fresh cheese yield. The differences, in terms of root mean squared error, between master predictions and slave predictions were reduced after standardization on average from 103 to 17 g/d, from 0.0315 to 0.0045 g/100 mL of milk, and from 2.55 to 0.49 g of curd/100 g of milk, respectively. For all the models, standard deviations of predictions among all the instruments were also reduced by 5.11 times for methane, 5.01 times for polyunsaturated fatty acids, and 7.05 times for fresh cheese yield, showing an improvement of prediction reproducibility within the network. Regarding the results obtained, spectral standardization allows the transfer and use of multiple models on all instruments as well as the improvement of spectral and prediction reproducibility within the network. The method makes the models universal, thereby offering opportunities for data exchange and the creation and use of common robust models at an international level to provide more information to the dairy sector from direct analysis of milk.

Assessing the effect of pregnancy stage on milk composition of dairy cows using mid-infrared spectra.

Changes in milk production traits (i.e., milk yield, fat, and protein contents) with the pregnancy stage are well documented. To our knowledge, the effect of pregnancy on the detailed milk composition has not been studied so far. The mid-infrared (MIR) spectrum reflects the detailed composition of a milk sample and is obtained by a nonexhaustive and widely used method for milk analysis. Therefore, this study aimed to investigate the effect of pregnancy on milk MIR spectrum in addition to milk production traits (milk yield, fat, and protein contents). A model including regression on the number of days pregnant was applied on milk production traits (milk yield, fat, and protein contents) and on 212 spectral points from the MIR spectra of 9,757 primiparous Holstein cows from Walloon herds. Effects of pregnancy stage were expressed on a relative scale (effect divided by the squared root of the phenotypic variance); this allowed comparisons between effects on milk traits and on 212 spectral points. Effect of pregnancy stage on production traits were in line with previous studies indicating that the model accounted well for the pregnancy effect. Trends of the relative effect of the pregnancy stage on the 212 spectral points were consistent with known and observed effect on milk traits. The highest effect of the pregnancy was observed in the MIR spectral region from 968 to 1,577 cm-1. For some specific wavenumbers, the effect was higher than for fat and protein contents in the beginning of the pregnancy (from 30 to 90 or 120 d pregnant). In conclusion, the effect of early pregnancy can be observed in the detailed milk composition through the analysis of the MIR spectrum of bovine milk. Further analyses are warranted to explore deeply the use of MIR spectra of bovine milk for breeding and management of dairy cow pregnancy.

Capitalizing on fine milk composition for breeding and management of dairy cows.

The challenge of managing and breeding dairy cows is permanently adapting to changing production circumstances under socio-economic constraints. If managing and breeding address different timeframes of action, both need relevant phenotypes that allow for precise monitoring of the status of the cows, and their health, behavior, and well-being as well as their environmental impact and the quality of their products (i.e., milk and subsequently dairy products). Milk composition has been identified as an important source of information because it could reflect, at least partially, all these elements. Major conventional milk components such as fat, protein, urea, and lactose contents are routinely predicted by mid-infrared (MIR) spectrometry and have been widely used for these purposes. But, milk composition is much more complex and other nonconventional milk components, potentially predicted by MIR, might be informative. Such new milk-based phenotypes should be considered given that they are cheap, rapidly obtained, usable on a large scale, robust, and reliable. In a first approach, new phenotypes can be predicted from MIR spectra using techniques based on classical prediction equations. This method was used successfully for many novel traits (e.g., fatty acids, lactoferrin, minerals, milk technological properties, citrate) that can be then useful for management and breeding purposes. An innovation was to consider the longitudinal nature of the relationship between the trait of interest and the MIR spectra (e.g., to predict methane from MIR). By avoiding intermediate steps, prediction errors can be minimized when traits of interest (e.g., methane, energy balance, ketosis) are predicted directly from MIR spectra. In a second approach, research is ongoing to detect and exploit patterns in an innovative manner, by comparing observed with expected MIR spectra directly (e.g., pregnancy). All of these traits can then be used to define best practices, adjust feeding and health management, improve animal welfare, improve milk quality, and mitigate environmental impact. Under the condition that MIR data are available on a large scale, phenotypes for these traits will allow genetic and genomic evaluations. Introduction of novel traits into the breeding objectives will need additional research to clarify socio-economic weights and genetic correlations with other traits of interest.

Development of Fourier transform mid-infrared calibrations to predict acetone, ß-hydroxybutyrate, and citrate contents in bovine milk through a European dairy network.

To manage negative energy balance and ketosis in dairy farms, rapid and cost-effective detection is needed. Among the milk biomarkers that could be useful for this purpose, acetone and ß-hydroxybutyrate (BHB) have been proved as molecules of interest regarding ketosis and citrate was recently identified as an early indicator of negative energy balance. Because Fourier transform mid-infrared spectrometry can provide rapid and cost-effective predictions of milk composition, the objective of this study was to evaluate the ability of this technology to predict these biomarkers in milk. Milk samples were collected in commercial and experimental farms in Luxembourg, France, and Germany. Acetone, BHB, and citrate contents were determined by flow injection analysis. Milk mid-infrared spectra were recorded and standardized for all samples. After edits, a total of 548 samples were used in the calibration and validation data sets for acetone, 558 for BHB, and 506 for citrate. Acetone content ranged from 0.020 to 3.355mmol/L with an average of 0.103mmol/L; BHB content ranged from 0.045 to 1.596mmol/L with an average of 0.215mmol/L; and citrate content ranged from 3.88 to 16.12mmol/L with an average of 9.04mmol/L. Acetone and BHB contents were log-transformed and a part of the samples with low values was randomly excluded to approach a normal distribution. The 3 edited data sets were then randomly divided into a calibration data set (3/4 of the samples) and a validation data set (1/4 of the samples). Prediction equations were developed using partial least square regression. The coefficient of determination (R(2)) of cross-validation was 0.73 for acetone, 0.71 for BHB, and 0.90 for citrate with root mean square error of 0.248, 0.109, and 0.70mmol/L, respectively. Finally, the external validation was performed and R(2) obtained were 0.67 for acetone, 0.63 for BHB, and 0.86 for citrate, with respective root mean square error of validation of 0.196, 0.083, and 0.76mmol/L. Although the practical usefulness of the equations developed should be further verified with other field data, results from this study demonstrated the potential of Fourier transform mid-infrared spectrometry to predict citrate content with good accuracy and to supply indicative contents of BHB and acetone in milk, thereby providing rapid and cost-effective tools to manage ketosis and negative energy balance in dairy farms.

Standardization of milk mid-infrared spectra from a European dairy network.

The goal of this study was to find a procedure to standardize dairy milk mid-infrared spectra from different Fourier transform mid-infrared spectrophotometers (different brands or models) inside a European dairy network to create new farm-management indicators (e.g., fertility, health, feed, environmental impact) based on milk infrared spectra. This step is necessary to create common spectral databases, allowing the building of statistical tools, to be used by all instruments of the network. The method used was piecewise direct standardization (PDS), which matches slave-instrument spectra on master-instrument spectra. To evaluate the possibility of using common equations on different instruments, the PDS method was tested on a set of milk samples measured on each machine, and an equation predicting fat content of milk is applied on all. Regressions were performed between master and slaves fat predictions, before and after PDS. Bias and root mean square error between predictions were decreased after PDS, respectively, from 0.3781 to 0.0000 and from 0.4609 to 0.0156 (g of fat/100mL of milk). The stability over time of these results was confirmed by an application of the coefficients created by PDS 1 mo later on the slave spectra. These preliminary results showed that the PDS method permits a reduction of the inherent spectral variability between instruments, allowing the merging of Fourier transform mid-infrared milk spectra from different instruments into a common database, the creation of new types of dairy farm management indicators, and the use of these common calibrations for all Fourier transform mid-infrared instruments of the European dairy network.

Identification of chronic stress biomarkers in dairy cows.

Stress in dairy herds can occur from multiple sources. When stress becomes chronic because of a long duration and inability of animals to adapt, it is likely to deeply affect the emotional state, health, immunity, fertility and milk production of cows. While assessing chronic stress in herds would be beneficial, no real consensus has emerged from the literature regarding the indicators of interest. The goal of this study was to compare and evaluate potential biomarkers for chronic stress after inducing stress over a 4-week period through severe overstocking, restricted access to feed and isolated unusual events. A total of 30 cows were involved in the experiment and two similar groups were constituted. Over a 4-week period, the 15 cows of the stress group were housed in overstocked conditions, with 4.6 m2 per cow, including resting and feeding areas. In this area, only seven individual places at the feeding area were available for the 15 cows to generate competition for feed access. Twice during the trial and over a period of 2 h, an additional stress was induced by moving cows to an unfamiliar barn and diffusion of stressing noises (dog barking). Meanwhile, the 15 cows of the control group stayed in the original barn, with more than 10 m2 per cow and more individual places at the feeding area than cow number. On a weekly basis, several variables considered as potential biomarkers for chronic stress were recorded. Collected data were analysed using single trait linear repeated mixed models. No differences were observed regarding milk yield, BW of cows or body condition score but the milk loss was more pronounced in the stress group. The activity was more heterogeneous and the rumination of cows was lower in the stress group. The heart rate was lower in the stress group and showed more heterogeneity at the end of the stress period. No differences were observed regarding salivary cortisol, blood glucose, ß-endorphin, thyroxine and leucocyte profile. A higher level of hair cortisol and blood fructosamine were observed in the stress group at the end of the stress period. Regarding the practical use of the highlighted biomarkers, milk loss may be an effective and easy way to detect general problems, including stress. The blood fructosamine and the hair cortisol concentrations are promising indicators to assess chronic stress in commercial farms.