Measuring individual cows’ response to heat stress at large-scale is challenging because physiological traits are not recorded routinely, and production traits are unspecific and require environmental data for interpretation. Milk mid-infrared (MIR) spectra, already recorded in routine, offer a potential alternative, as heat stress affects milk composition and is therefore expected to be reflected in MIR spectra. This study thus aimed to develop a MIR prediction equation for individual heat stress response. Surface temperature and milk traits from 399 cows were recorded to develop a combined heat stress response phenotype. This phenotype resulted from two equations: one predicting surface body temperature (R2 = 0.67; RMSE = 0.64 °C) and one classifying records into three heat stress response classes based on surface temperature and milk composition (accuracy = 61%). The final prediction was applied to historical milk recording data associated with weather information to assess external validity. A mixed model was also fitted to identify cow characteristics associated with stronger predicted heat stress responses. As reported in the literature, multiparous cows, in early lactation, with the highest 24 h milk yield tended to be more affected. Overall, the prediction developed in this study shows strong potential for routine heat stress detection.

During the last decade, there has been a growing interest for local and endangered breeds as they are often seen as more resilient and healthier than mainstream breeds. They can also provide high added-value products like meat or dairy products. To better preserve these breeds, it is of main importance to characterize their genetic diversity and have an insight of historical gene flow with more mainstream breeds. In this study, we focused on the genetic relationships of five red-pied cattle breeds: the east Belgian red and white (EBRW); the red-pied of the Ösling (RPO), from Luxembourg; the deep red (DR) and the Meuse-Rhine-Yssel (MRY), both from the Netherlands; and the German red and white dual purpose (RDN). The EBRW, RPO and DR breeds have an official European endangered status. We first investigated the pedigree completeness of available genotyped animals as well as their complex historical relationships through the analysis of common ancestors. We also compared pedigree and SNP-based inbreeding coefficients, defined as the sum of homozygosity-by-descent segments (HBD). We then dived into genomic relationships through a classical multi-dimensional scaling (MDS) of genotyped animals and their admixture. The pedigree analyses showed the complex gene flow between all breeds and that the RPO breed was the most connected to other breeds. Results also showed that the level of inbreeding was so far not an issue in all five breeds even if some animals, for example, in EBRW and MRY breeds, showed higher inbreeding levels than the average of their breeds. Finally, the MDS and admixture analysis also highlighted complex gene flow between the studied breeds and that they may be considered as a genetic continuum. This genomic proximity has the potential to improve genetic evaluations of local breeds by the inclusion of information from more mainstream breeds like MRY and RDN.

Heat stress studies have focused on numerous traits in dairy cows, yet most do not consider that different heat events may trigger distinct responses even in the same animals. A first objective was therefore to follow the dynamics of several traits across different heat waves. To achieve this, weekly sampling was conducted from June to September 2025, with daily monitoring during heat waves, on a total of 57 lactating cows. Mixed models were applied to mitigate the impact of confounding effects and residuals were analysed. Some traits (udder surface temperature, milk protein percentage, milk magnesium concentration, dry matter intake and rumination time) showed relatively consistent direction of variation across heat waves but others, especially those related to energy balance and activity, presented opposite variations depending on the heat wave. This highlight that cows' responses to heat stress may depend on heat intensity and duration but probably also on the resistance or exhaustion levels of cow physiological adaptation mechanisms. A second objective was to identify favourable patterns by comparing cows classified as heat-tolerant or heat-sensitive. Thermotolerant cows exhibited a better energy balance during heat events based on associated markers and tended to better maintain production than thermosensitive cows. They also kept lower activity time during heat wave in which negative energy balance markers increased, which could have helped conserve energy. Overall, this study highlighted the central role of energy balance to cope with heat stress and the importance to account for heterogenous responses across heat events.

With the purpose to organize methodologies found in (recent) papers focusing on the development of genomic breed/population assignment tools, this review proposes to highlight good practice for the development of such tools. After an appropriate quality control of markers and the building of a representative reference population, three main steps can be followed to develop a genomic breed/population assignment tool: 1) The selection of discriminant markers, 2) The development of a model that allows accurate assignment of animals to their breed/population of origin, the so-called classification step, and, 3) The validation of the developed model on new animals to evaluate its performances in real conditions. The first step can be avoided when a mid- or low-density chip is used, depending on the methodology used for assignment. In the case selection of SNPs is necessary, we advise the use of one stage methodologies and to define a threshold for this selection. Then, machine learning can be used to develop the model per se, based on the selected or available markers. To tune the model, we recommend the use of cross-validation. Finally, new animals, not used in the first two steps, should be used to evaluate the performances of the model (e.g., with balanced accuracy and probabilities), also in terms of computation time.

The negative energy balance (NEB) state in dairy cows is a critical factor affecting health, reproduction, and production, particularly during early lactation. Multiple blood and milk biomarkers change when dairy cows are in the NEB state. Direct measurement of NEB is impractical for large-scale use due to costs, necessitating reliance on indirect predictors such as milk mid-infrared (MIR) spectrometry-based predicted biomarkers. However, the genetic relationships between NEB and its potential biomarkers remain unclear. This study aimed to (1) compare measured reference NEB with MIR-predicted NEB (PNEB), a novel energy deficit score (EDS), 15 biomarkers, and 3 production traits; (2) estimate genetic parameters among these traits using a 20-trait repeatability model, quantifying the ability of the 19 other studied traits (logit-transformed EDS (LEDS), 15 biomarkers, and 3 production traits) to genetically predict logit-transformed PNEB (LPNEB); and (3) evaluate the causal effects of LPNEB on the 19 traits through a recursive model. Two datasets were used: dataset I (127 cows, 965 records) provided reference data for objective (1), and dataset II (25,287 first-parity cows, 30,634 records) enabled genetic analysis used for objectives (2) and (3). Traits were analyzed using Pearson correlations, multiple-diagonalization expectation maximization REML-based genetic parameter estimation, and recursive modeling. The studied traits had moderate to moderate-high h2 ranging from 0.16 to 0.38. The genetic correlations between LPNEB and the studied traits ranged from -0.60 for LIGF-1 to 0.85 for MIR-predicted blood nonesterified fatty acids (NEFA). Analysis of genetic predictability of LPNEB traits together explained 89% of the genetic variance of LPNEB, with all 15 biomarkers alone contributing the largest fraction with 82%, LEDS alone 65%, NEFA alone 62%, and all traits except LEDS 85%, indicating that LEDS contains useful additional information. Recursive modeling further identified 8 traits, including NEFA and LEDS, as highly dependent on LPNEB, highlighting their potential as robust biomarkers. This study demonstrates the utility of MIR-predicted traits for understanding the genetic mechanisms of NEB and its potential for integration into breeding programs, while emphasizing cautious interpretation of these results due to limitations of MIR-predictions of studied traits to represent directly measured traits.

This article focuses on the creation of a monitoring tool using routinely collected data from milk payment analyses. Milk samples were analyzed through Fourier Transform mid-infrared spectrometry every 1 to 3 days, and their compositions were predicted using machine learning models. Among the predicted parameters, fatty acid profiles appear to be effective indicators of animal status and management practices. In this research, these profiles were summarized using 31 fatty acids or groups of fatty acids. The methodology consists of four steps: hierarchical clustering to detect patterns in a Belgian spectral dataset (N = 774,781), interpretation of the identified seven clusters, development of predictive models applied to a Canadian dataset (N = 670,165), and validation using management information collected from Canadian farms. The identified clusters revealed significant relationships with feeding management strategies and temporal evolutions, highlighting the potential to develop automated alert systems that assist farmers and advisors in herd monitoring.

Because of lack of pedigree records, the history of the Red-Pied of the Osling (RPO) breed, which is an endangered local red-pied cattle breed from Luxembourg, is not really known. However, it is assumed that there has been exchanges between the RPO and the Meuse-Rhine-Yssel (MRY), another red-pied cattle breed from The Netherlands. To validate this assumption, we estimated the changes in the relationships between the RPO and MRY breeds over time by the definition of cohorts with different birthdates in the MRY breed. The fixation indexes between the RPO breed and the different MRY cohorts, as well as a principal component analysis, showed that the RPO breed was more related to the oldest MRY animals, born before 1990, and to the youngest MRY animals, born after 2009. This confirmed the a priori known pattern of importation of MRY animals to the RPO population over time. Based on results obtained in a genomic relationship matrix and on the proportion of opposing homozygotes, we could infer sire-offspring relationships between an MRY sire and three RPO animals. To complete the pedigree of RPO animals even further and optimize exchanges between the MRY and RPO breeds, it would be interesting to have access to more MRY genotypes.

Using genetic selection for raising intact boars, which improves growth and feed efficiency, is a promising alternative to castration for mitigating boar taint. Selective breeding has the potential to help to identify and select genetic lines with a reduced risk of boar taint. Common phenotypes are laboratory measurements of skatole (SKA) and androstenone (ANON) i.e., the major compounds responsible for boar taint, in backfat. However, an alternative exists: sensory evaluation by human assessors. The objectives of this study were (1) to estimate the genetic relationships among sensory scores (SENS) obtained by different assessors, (2) to correlate these scores with SKA and ANON, (3) to establish the independence of SENS from the causal traits, here SKA and ANON, by recursive modeling, holding those constant, and (4) to combine different assessors to allow an efficient selection against boar taint. Data included up to 1,016 records of SKA, ANON, and SENS (0-5) from 10 trained assessors on the backfat of intact males reared at least until puberty at three performance testing stations testing the products of Pietrain x commercial crossbred sows. Genetic parameters were estimated using restricted estimate maximum likelihood. Traits SKA and ANON were log10 transformed (SKAt and ANONt) and SENS traits were Snell transformed SENS (SENSt). Heritability estimates were 0.52 for SKAt and 0.53 for ANONt, those for SENSt ranged from 0.07 to 0.30. Moderate to high genetic correlations between some SENSt and SKAt (up to 0.87) and ANONt (up to 0.61) were found. Heritabilities and correlations indicated that some SENSt could be used to select against boar taint. Studying the independence of SENSt from SKAt and ANONt based on a posteriori recursive model revealed a large range of reductions of genetic variance: up to 71.08%. However, some SENSt remained moderately heritable (0.04-0.19) indicating independent genetic variance from SKAt and ANONt. This reflects that some heritable compounds potentially not related to SKA or ANON are perceived. Finally, the combination of assessors allowed, here shown with three assessors, to obtain a high heritability of 0.40, associated to high genetic and phenotypic correlations. Moreover, these results demonstrate the potential of using the sensory scores of several trained assessors for selection against boar taint. It seems possible to use sensory evaluations, or combinations of those, as reliable phenotypes for breeding against boar taint risk. In order to draw novel conclusions, we used a recursive model to assess the (in)dependence of sensory scores with respect to the major compounds of boar taint and we identified an optimal combination of assessors for achieving the highest genetic progress. Meat quality can be impacted by different practices during the whole life of pigs. Raising intact boars is interesting to potentially improve the growth rate and feed efficiency of boars. Moreover, castration is a major welfare and health issue. However, in intact boars so called boar taint, a fecal and urinary smell, can occur which can repel consumers. This odor is mainly caused by skatole (SKA) and androstenone (ANON) accumulation in fat tissues. To reduce their impact, genetic selection against these heritable compounds can be applied. However, their analytical measurements are costly, time-consuming, and, in consequence, in low numbers. Alternative routine data collection based on sensory evaluation scores (SENS) has been proposed. These SENS were attributed to heated fat samples by 10 trained assessors to detect SKA and ANON together. Genetic relationships indicated that some SENS could potentially be used for genetic selection against SKA and ANON. Investigations on the origin of attributed SENS demonstrated that some (unknown) compounds probably correlated to SKA and ANON are perceived, too. Finally, SENS from different assessors were combined to select more efficiently against boar taint.

Meat quality traits are economically important in pig production. Breeding strategies can help prevent meat defects such as boar taint, usually characterized by quantified indole, skatole and androstenone (ISA) in back fat. This exploratory study investigated the genetic potential of a novel boar taint phenotype, pooling volatile organic compounds (VOCs), which were recently identified as phenotypically discriminant. Fat samples were collected from 1272 Pietrain x Landrace crossbred boars. Phenotypes for boar taint on these samples were: lab sensory score (LSS; n = 1269), ISA quantification (n = 308), and VOC profiles (n = 127). Given the limited amount of data, a selection index-based approach was used to pool traits in trait groups, ISA and VOC, considering LSS as reference trait. (Co)variance components were estimated with a full multi-trait model, and index equations were adjusted to account for uncertainty in estimated parameters. Index coefficients were then applied to ISA and VOC phenotypes to generate two pooled phenotypes, ISA and VOC indices. Estimates from the 3-trait model (LSS, ISA index and VOC index) confirmed high expected correlations with LSS. Genetic parameter estimates showed higher significance demonstrating the interest of pooling multiple partially informative traits together. Moreover, using the VOC index would generate a higher expected correlated genetic response in LSS (192 %) than the ISA index (160 %) compared to the direct response when using only LSS. Despite limited data, this exploratory study showed the potential of this novel broad phenotype based on pooled VOCs to improve genetic selection for reduced boar taint risk, although further validation in larger populations is required.

Somatic cell count is widely used for large-scale udder health monitoring and remains a proxy for mastitis incidence still used in many genetic evaluation systems. This trait and its log-transformation, SCS, are thus also available to study the effect of heat stress on mammary gland health. Currently, a new trait called differential somatic cell count (DSCC), which represents the percentage of neutrophils and lymphocytes in the total SCC, is increasingly phenotyped simultaneously with SCC. By combining information, SCS and DSCC could more closely reflect the direct trait than SCS alone, providing a better proxy for mastitis incidence including during heat stress events. On this basis, the first objective of this study was to evaluate the interest of DSCC for heat stress assessment with a focus on mammary gland health with SCS as comparison. Additionally, the interest of both traits for genetic evaluation of udder health thermotolerance was explored. Because studies providing basal genetic parameters for DSCC are still rare, they were also estimated in this study. To do so, a random regression model on DIM was performed considering each parity as a different trait. For both SCS and DSCC, similar averaged daily heritability (0.10 to 0.11 for SCS and 0.11 to 0.14 for DSCC) and repeatability (0.61 to 0.64 for SCS and 0.54 to 0.63 for DSCC) were obtained. Moderate averaged daily genetic correlations were also estimated between SCS and DSCC (0.43 to 0.55). From the residuals of the same model, average residual responses with temperature-humidity index (THI) were studied. The results showed that DSCC reaction in mean and in variance with high THI was stronger than SCS. In addition, the reaction with increasing THI seemed to be inconsistent between lactation numbers for SCS conversely to DSCC. In this way, DSCC presented more relevant characteristics than SCS to discriminate thermotolerant and thermosensitive cows for udder health. However, for general heat stress detection, udder health traits seemed not to be the most adapted biomarkers. Low heritability (0.02 to 0.03 for SCS and 0.03 to 0.04 for DSCC) and repeatability (0.12 to 0.18 for SCS and 0.20 to 0.26 for DSCC) values were also obtained for SCS and DSCC newly defined thermotolerance traits.

Negative energy balance (NEB) during early lactation is a critical physiological challenge in high-producing dairy cows, affecting both their health and production performance. The objectives of this study were: (1) to compare the genetic architecture of logit-transformed predicted NEB (LPNEB), a logit-transformed novel energy deficiency score (LEDS), 15 biomarkers, and 3 production traits using SNP-based genomic correlation analysis; (2) to extend this study to a chromosomal level to identify specific genomic regions involved in the regulation of energy metabolism; and (3) to compare the independent contributions of 8 traits to the underlying genetic architecture of LPNEB and LEDS. The SNP effects estimated from single-trait models can be used to quickly calculate genomic correlations for 20 traits. The results indicate strong genomic correlations between LPNEB and LEDS, as well as with key metabolic biomarkers, particularly blood nonesterified fatty acids (NEFA), highlighting their importance in energy metabolism. Furthermore, NEFA was a strong independent contributor to both LPNEB and LEDS. Chromosome regions located on BTA19 and BTA25 were identified as potentially associated with NEB. By combining genomic correlation and contribution analyses, this study provides valuable insights into the genetic basis of NEB and related traits in dairy cows.

This study aimed to compare the genetic architectures of logit-transformed predicted negative energy balance (LPNEB) and a novel logit-transformed energy deficiency score (LEDS) as 2 mid-infrared-derived proxies of negative energy balance in early-lactation dairy cows. A total of 30,634 records from 25,287 first-parity Holstein cows across 508 herds distributed in Walloon region of Belgium were analyzed. Genotypic data of 566,170 SNPs were available for 3,757 animals. Single-step GWAS, combined with a 50-SNP sliding window approach, was employed to explore the genetic architectures of LPNEB and LEDS. The top 10 genomic regions for LPNEB and LEDS were identified across multiple chromosomes, with 3 shared regions (BTA 1, 5, and 16). Despite these overlaps, each trait exhibited unique loci, supporting distinct genetic architectures. Positional candidate gene analyses identified 17 genes for LPNEB and 10 for LEDS, with 6 being in common. Gene Ontology enrichment analyses were then performed to explore their biological functions, although LPNEB was primarily associated with energy metabolism regulation and metabolic adaptation, whereas LEDS integrated neuronal signaling into energy homeostasis. The QTL enrichment highlighted significant associations with fertility and SCS, reinforcing a genetic basis for energy balance. These findings improve our understanding of the genetic background of LPNEB and LEDS, thereby providing new insights into the mechanisms underlying energy balance in dairy cattle.

The adoption of automated milking systems (AMS) across worldwide dairy farms has grown considerably over the last few decades. Automated milking systems contribute to reducing labor costs, increasing milk performance, improving cow welfare, and generating large-scale data on a routine basis that can be used for deriving novel traits for breeding purposes. Therefore, the primary objectives of this study were to (1) derive behavioral traits from AMS data and assess their phenotypic variability during lactation in US Holstein cattle, and (2) estimate variance components and genetic parameters for these traits. Daily AMS records from 5,645 US Holstein cows, collected at 36 robotic milking stations between 2018 and 2021, were analyzed. Evaluated traits included average milking time (AMT, min), total milking time (TMT, min), time interval between milkings (INT, h), number of attempted visits to the AMS (NoV), number of successful entries within the AMS (NSE), percentage of successful milkings (PSM, %), and cow preference consistency score (PCS, score unit). Variance components and genetic parameters were estimated using repeatability models with the restricted maximum likelihood method. The heritability estimates were similar between the 2 models for most traits: 0.46 versus 0.46, 0.27 versus 0.28, 0.08 versus 0.10, 0.10 versus 0.10, 0.10 versus 0.11, and 0.05 versus 0.06, for AMT, TMT, INT, NoV, NSE, and PSM, respectively. However, a notable difference was observed for PCS, with heritability estimates of 0.09 and 0.24 depending on the model fitted. The SE for the heritability ranged from 0.001 to 0.03. Repeatability estimates were 0.74 to 0.71 (AMT), 0.52 to 0.49 (TMT), 0.34 to 0.27 (INT), 0.29 to 0.25 (NoV), 0.29 to 0.30 (NSE), 0.20 to 0.18 (PSM), and 0.55 to 0.53 (PCS). Positive genetic correlations were observed for trait pairs AMT-PSM (0.38-0.35), INT-PSM (0.71-0.64), INT-PCS (0.50-0.40), and PSM-PCS (0.37), whereas other correlations were unfavorable or near zero. All cow behavioral traits related to AMS efficiency evaluated in this study were found to be heritable, suggesting that their inclusion in selection schemes could contribute to improving dairy cow milking efficiency and welfare in dairy farms using AMS. Future studies will model these traits using random regression models and estimate their genetic correlation with other relevant traits in dairy breeding programs.

The increased uptake of sensor technologies and precision farming tools for the dairy cattle sector is enabling real-time monitoring of animal health, welfare, and productivity. These digital advancements provide high-frequency, objective, and large-scale phenotypic data for breeding purposes. This review explores the potential of sensor-derived data to improve genetic and genomic evaluations in dairy cattle and outlines key challenges, opportunities, and approaches associated with their implementation. While these data streams have great potential for genetic evaluations, their integration into national and international breeding programs remains limited due to fragmentation across sensor brands, lack of standardization, and challenges related to data accessibility, data access and portability rights, business interests, and governance. A crucial aspect of leveraging digital technologies in dairy cattle breeding is data harmonization and integration. We highlight the importance of establishing standardized data collection and data sharing protocols, implementing robust quality control and data cleaning methodologies, as well as defining novel sensor-based traits and estimating their genetic background. In this context, we compiled heritability estimates for novel traits derived from data recorded by sensors and other technologies in dairy cattle populations. The development of phenomics in breeding programs, which involves integrating multisource data-including sensor-based, genomic, and management information-will be key to accelerating genetic progress, especially for traits related to animal welfare, health, resilience, and efficiency. This review presents a roadmap for the effective use of sensor-derived data in genetic evaluations, advocating for centralized data infrastructures, transparent data-sharing agreements, and the role of different stakeholders from academia and industry, including organizations such as the International Committee on Animal Recording (ICAR) in establishing global standards and guidelines. By addressing these challenges, dairy breeding programs can fully harness precision dairy farming technologies to enhance production and environmental efficiency, improve animal health and welfare, and drive sustainable genetic advancements in the dairy cattle sector.

Numerous prediction equations have been developed based on mid-infrared (MIR) spectra, and some could be potentially used as biomarkers of heat stress. However, practical experience shows that confusion can easily occur between the effect of heat stress and other effects, such as lactation stage or feeding variation over the year. On this basis, the objective of this study was to identify potential milk components predicted by MIR as biomarkers of heat stress based on a 2-step approach allowing correction for those effects. The first step consisted in the estimation of residuals from test-day random regression models on DIM to remove systematic lactation stage effects. These models also contained, among others, general (i.e., month of production) or specific (i.e., herd x test-day) fixed effects related to feeding and management. During the second step, means and variances of residuals by temperature-humidity index (THI) classes were studied. The models were applied to 611,063 records from 97,042 primiparous Holstein cows from 2015 to 2022 in the south of Belgium. The MIR-predicted milk components with the highest deviations from the mean with increasing THI were protein percentage, casein concentration, magnesium concentration, and (to a lesser extent) PUFA concentration. Concerning residual variances, the highest heteroscedasticity with THI was obtained for milk MIR MUFA, C18:1 cis-9, and citrate concentrations. Conversely, a relative homoscedasticity of variance with increasing THI was observed for several milk MIR components including protein percentage and casein concentration. Based on the criteria of the good biomarkers guidelines, milk protein percentage seems to be the most promising trait of this study, followed by Mg concentration. However, in the context of genetic evaluation, which requires variability, milk MIR MUFA, C18:1 cis-9, or citrate concentration variations, if they are heritable, could be of great interest. Finally, an increase in milk MIR citrate concentration variance could be an early warning for the detection of heat stress in the frame of DHI.

Classification models for rapid discrimination of milk storage time were developed. Buffalo raw milk (BuR) and bovine raw (BoR), pasteurised (BoP) and ultra-high temperature sterilised (BUHT) milk were analysed by mid-infrared (MIR) spectroscopy at different storage times. A total of 175 (44), 96 (24), 252 (64) and 383 (97) samples were used in the developed (validated) models of BuR, BoR, BoP and BUHT milk, respectively. Seven, two, five, and three spectral regions were selected for BuR, BoR, BoP and BUHT, respectively, to develop their models, which reflect changes in milk protein, fat, lactose and urea nitrogen. Accuracies of the optimised models for BuR, BoR, BoP and BUHT milk were 0.95, 1.00, 0.89, and 0.93, respectively, in the validation dataset. All optimal models were obtained by machine learning methods. This preliminarily study shows that MIR combined with machine learning can rapidly identify the storage time of these four types of milk.

This study aims to characterize a complete volatile organic compound profile of pork neck fat for boar taint prediction. The objectives are to identify specific compounds related to boar taint and to develop a classification model. In addition to the well-known androstenone, skatole and indole, 10 other features were found to be discriminant according to untargeted volatolomic analyses were conducted on 129 samples using HS-SPME-GC×GC-TOFMS. To select the odor-positive samples among the 129 analyzed, the selection was made by combining human nose evaluations with the skatole and androstenone concentrations determined using UHPLC-MS/MS. A comparison of the data of the two populations was performed and a statistical model analysis was built on 70 samples out of the total of 129 samples fully positive or fully negative through these two orthogonal methods for tainted prediction. Then, the model was applied to the 59 remaining samples. Finally, 7 samples were classified as tainted.

Regular monitoring of body condition score (BCS) changes during lactation is an essential management tool in dairy cattle; however, the current BCS measurements are often discontinuous and unevenly spaced in time. The imputation of BCS values is useful for two main reasons: i) achieving completeness of data is necessary to be able to relate BCS to other traits (e.g. milk yield and milk composition) that have been routinely recorded at different times and with a different frequency, and ii) having expected BCS values provides the possibility to trigger early warnings for animals with certain unexpected conditions. The contribution of this study was to propose and evaluate potential methods useful to smooth and impute device-based BCS values recorded during lactation in dairy cattle. In total, 26,207 BCS records were collected from 3,038 cows (9,199 and 14,462 BCS records on 1,546 Holstein and 1,211 Montbeliarde cows respectively, and the rest corresponded to other minority cattle breeds). Six methods were evaluated to predict BCS values: the traditional methods of test interval method (TIM), and multiple-trait procedure (MTP), and the machine learning (ML) methods of multi-layer perceptron (MLP), Elman network ( Elman ), long-short term memories ( LSTM) and bi-directional LSTM ( BiLSTM ). The performance of each method was evaluated by a hold-out validation approach using statistics of the root mean squared error ( RMSE ) and Pearson correlation (r). TIM, MTP, MLP, and BiLSTM were assessed for the imputation of intermediate missing values, while MTP, Elman, and LSTM were evaluated for the forecasting of future BCS values. Regarding the machine learning methods, BiLSTM demonstrated the best performance for the intermediate value imputation task (RMSE = 0.295, r = 0.845), while LSTM demonstrated the best performance for the future value forecasting task (RMSE = 0.356, r = 0.751). Among the methods evaluated, MTP showed the best performance for imputation of intermediate missing values in terms of RMSE (0.288) and r (0.856). MTP also achieved the best performance for forecasting of future BCS values in terms of RMSE (0.348) and r (0.760). This study demonstrates the ability of MTP and machine learning methods to impute missing BCS data and provides a cost-effective solution for the application area.

The aims of this study were to estimate genetic parameters and to identify genomic regions associated with eating time (ET) and rumination time (RUT) in Holstein dairy cows. Genetic correlations among ET, RUT, and milk yield traits were also estimated. The data were collected from 2019 to 2022 in 6 dairy herds located in the Walloon Region of Belgium. The dataset consisted of daily ET and RUT records on 284 Holstein cows, from which 41 cows had records only for the first parity (P1), 101 cows had records from both the first and second parities, and 142 cows had records only for the second parity (P2). The number of daily ET and RUT records in the P1 and P2 cows were 18,569 (on 142 cows) and 34,464 (on 243 cows), respectively. Data on 28,994 SNPs located on 29 Bos taurus autosomes (BTA) of 747 animals (435 males) were used. Random regression test-day models were used to estimate genetic parameters through the Bayesian Gibbs sampling method. The SNP solutions were estimated using a single-step genomic best linear unbiased prediction approach. The proportion of genetic variance explained by each 20-SNP sliding window (with an average size of 1.52 Mb) was calculated, and regions accounting for at least 1.0% of the total additive genetic variance were used to search for candidate genes. Mean (standard deviation; SD) averaged daily ET and RUT were 327.0 (85.66) and 559.4 (77.69) min/d for cows in P1 and 316.0 (82.24) and 574.2 (75.42) min/d for cows in P2, respectively. Mean (standard deviation; SD) heritability estimates for daily ET and RUT were 0.42 (0.09) and 0.45 (0.06) for cows in P1 and 0.45 (0.04) and 0.43 (0.02) for cows in P2, respectively. Mean (SD) daily genetic correlations between daily ET and RUT were 0.27 (0.07) for P1 and 0.34 (0.08) for P2. Genome-wide association analyses identified 6 genomic regions distributed over 5 chromosomes (BTA1, BTA4, BTA11, 2 regions of BTA14, and BTA17) associated with ET or RUT. The findings of this study increase our preliminary understanding of the genetic background of feeding behavior in dairy cows; however, larger datasets are needed to determine whether ET and RUT might have the potential to be used in selection programs.

With the rapid development of animal phenomics and deep phenotyping, we can obtain thousands of traditional (but also molecular) phenotypes per individual. However, there is still a lack of exploration regarding how to handle this huge amount of data in the context of animal breeding, presenting a challenge that we are likely to encounter more and more in the future. This study aimed to (1) explore the use of the mega-scale linear mixed model (MegaLMM), a factor model-based approach that is able to simultaneously estimate (co)variance components and genetic parameters in the context of thousands of milk traits, hereafter called thousand-trait (TT) models; (2) compare the phenotype values and genomic breeding value (u) predictions for focal traits (i.e., traits that are targeted for prediction, compared with secondary traits that are helping to evaluate), from single-trait (ST) and TT models, respectively; (3) propose a new approximate method of GEBV (U) prediction with TT models and MegaLMM. We used a total of 3,421 milk mid-infrared (MIR) spectra wavepoints (called secondary traits) and 3 focal traits (average fat percentage [AFP], average methane production [ACH4], and average SCS [ASCS]) collected on 3,302 first-parity Holstein cows. The 3,421 milk MIR wavepoint traits were composed of 311 wave- points in 11 classes (months in lactation). Genotyping information of 564,439 SNPs was available for all animals and was used to calculate the genomic relationship matrix. The MegaLMM was implemented in the framework of the Bayesian sparse factor model and solved through Gibbs sampling (Markov chain Monte Carlo). The heritabilities of the studied 3,421 milk MIR wave- points gradually increased and then decreased in units of 311 wavepoints throughout the lactation. The genetic and phenotypic correlations between the first 311 wavepoints and the other 3,110 wavepoints were low. The accuracies of phenotype predictions from the ST model were lower than those from the TT model for AFP (0.51 vs. 0.93), ACH4 (0.30 vs. 0.86), and ASCS (0.14 vs. 0.33). The same trend was observed for the accuracies of u predictions for AFP (0.59 vs. 0.86), ACH4 (0.47 vs. 0.78), and ASCS (0.39 vs. 0.59). The average correlation between U predicted from the TT model and the new approximate method was 0.90. The new approximate method used for estimating U in MegaLMM will enhance the suitability of MegaLMM for applications in animal breeding. This study conducted an initial investigation into the application of thousands of traits in animal breeding and showed that the TT model is beneficial for the prediction of focal traits (phenotype and breeding values), especially for difficult-to-measure traits (e.g., ACH4).