Use of rumination time in health risk assessment of prepartum dairy cows.

The objectives of this observational cohort study were to evaluate the associations of rumination time (RT) in the last week of pregnancy with transition cow metabolism, inflammation, health, and subsequent milk production, reproduction, and culling. Pregnant nulliparous (n = 199) and parous (n = 337) cows were enrolled 21 d before the expected calving. RT and physical activity were monitored automatically by sensors from d -21 to 15 relative to calving. Blood samples were collected on d -14, -5, 4, 8, and 12 ± 1 relative to calving. Diagnoses of clinical health problems were performed by researchers from calving to 15 d in milk (DIM). In classification 1, cows were ranked based on average daily RT in the last week of pregnancy and classified into terciles as short RT (SRT), moderate RT (MRT), or long RT (LRT) for association analyses. In classification 2, RT deviation from the parity average was used in a receiver operating characteristic curve to identify the best threshold to predict postpartum clinical disease. Cows were then classified as above the threshold (AT) or below the threshold (BT). Compared with cows with LRT, cows with SRT had greater serum concentrations of NEFA (0.47 vs 0.40 ± 0.01 mmol/L), BHB (0.58 vs 0.52 ± 0.01 mmol/L), and haptoglobin (0.22 vs 0.18 ± 0.008 g/L) throughout the transition period, and reduced concentrations of glucose, cholesterol, albumin, and magnesium in a time-dependent manner. Parous cows with SRT had higher odds of postpartum clinical disease (adjusted odds ratio [AOR]: 3.7; 95% confidence interval [CI]: 2.1-6.4), lower odds of pregnancy by 210 DIM (AOR: 0.34; CI: 0.15-0.75), and lower milk production (46.9 vs 48.6 ± 0.5 kg/d) than parous cows with LRT. Deviation in prepartum RT had good predictive value for clinical disease in parous cows (area under the curve [AUC]: 0.65; CI: 0.60-0.71) but not in nulliparous (AUC: 0.51; CI: 0.42-0.59). Separation of parous cows according to the identified threshold (≤-53 min from the parity average) resulted in differences in serum concentrations of NEFA (AT = 0.31 ± 0.006, BT = 0.38 ± 0.014 mmol/L), BHB (AT = 0.49 ± 0.008, BT = 0.53 ± 0.015 mmol/L), and globulin (AT = 32.3 ± 0.3, BT = 34.8 ± 0.5 g/L) throughout the transition period, as well as in serum cholesterol, urea, magnesium, albumin, and haptoglobin in a time-dependent manner. BT parous cows had higher odds of clinical disease (AOR: 3.7; CI: 2.1-6.4), reduced hazard of pregnancy (AHR: 0.64, CI: 0.47-0.89), greater hazard of culling (AHR: 2.1, CI: 1.2-3.6), and lower milk production (46.3 ± 0.7 vs 48.5 ± 0.3 kg/d). External validation using data from 153 parous cows from a different herd and the established threshold in RT deviation (≤-53 min) resulted in similar predictive value, with the odds of postpartum disease 2.4 times greater in BT than AT (37.5 vs 20.1%). In conclusion, RT in the wk preceding calving was a reasonable predictor of postpartum health and future milk production, reproduction, and culling in parous cows but not in nulliparous cows.

Estimates of genetic parameters for rumination time, feed efficiency, and methane production traits in first-lactation Holstein cows.

The large-scale recording of traits such as feed efficiency (FE) and methane emissions (ME) for use in genetic improvement programs is complex, costly, and time-consuming. Therefore, heritable traits that can be continuously recorded in dairy herds and are correlated with FE and ME traits could provide useful information for genetic evaluation. Rumination time has been suggested to be associated with FE, methane production (MeP; ME in g/d), and production traits at the phenotypic level. Therefore, the objective of this study was to investigate the genetic relationships among rumination time (RT), FE, methane and production traits using 7,358 records from 656 first-lactation Holstein cows. The estimated heritabilities were moderate for RT (0.45 ± 0.14), MeP (0.36 ± 0.12), milk yield (0.40 ± 0.08), fat yield (0.29 ± 0.06), protein yield (0.32 ± 0.07), and energy-corrected milk (0.28 ± 0.07), but were low and nonsignificant for FE (0.15 ± 0.07), which was defined as the residual of the multiple linear regression of DMI on energy-corrected milk and metabolic body weight. A favorable negative genetic correlation was estimated between RT and MeP (-0.53 ± 0.24), whereas a positive favorable correlation was estimated between RT and energy-corrected milk (0.49 ± 0.11). The estimated genetic correlation of RT with FE (-0.01 ± 0.17) was not significantly different from zero but showed a trend of a low correlation with dry matter intake (0.21 ± 0.13). These results indicate that RT is genetically associated with MeP and milk production traits, but high standard errors indicate that further analyses should be conducted to verify these findings when more data for RT, MeP, and FE become available.

Comparison of imputation methods for missing production data of dairy cattle.

Nowadays, vast amounts of data representing feed intake, growth, and environmental impact of individual animals are being recorded in on-farm settings. Despite their apparent use, data collected in real-world applications often have missing values in one or several variables, due to reasons including human error, machine error, or sampling frequency misalignment across multiple variables. Since incomplete datasets are less valuable for downstream data analysis, it is important to address the missing value problem properly. One option may be to reduce the dataset to a subset that contains only complete data, but considerable data may be lost via this process. The current study aimed to compare imputation methods for the estimation of missing values in a raw dataset of dairy cattle including 454 553 records collected from 629 cows between 2009 and 2020. The dataset was subjected to a cleaning process that reduced its size to 437 075 observations corresponding to 512 cows. Missing values were present in four variables: concentrate DM intake (CDMI, missing percentage = 2.30%), forage DM intake (FDMI, 8.05%), milk yield (MY, 15.12%), and BW (64.33%). After removing all missing values, the resulting dataset (n = 129 353) was randomly sampled five times to create five independent subsets that exhibit the same missing data percentages as the cleaned dataset. Four univariate and nine multivariate imputation methods (eight machine learning methods and the MissForest method) were applied and evaluated on the five repeats, and average imputation performance was reported for each repeat. The results showed that Random Forest was overall the best imputation method for this type of data and had a lower mean squared prediction error and higher concordance correlation coefficient than the other imputation methods for all imputed variables. Random Forest performed particularly well for imputing CDMI, MY, and BW, compared to imputing FDMI.

Genetic evaluation of heat tolerance in Holsteins using test-day production records and NASA POWER weather data.

Heat stress is a prominent issue in livestock production, even for intensively housed dairy herds in Canada. Production records and meteorological data can be combined to assess heat tolerance in dairy cattle. The overall aim of this study was to evaluate the possibility of genetic evaluation for heat tolerance in Canadian dairy cattle. The 2 specific objectives were (1) to estimate the genetic parameters for milk, fat, and protein yield for Holsteins while accounting for high environmental heat loads, and (2) to determine if a genotype-by-environment interaction causes reranking of top-ranked sires between environments with low and high heat loads. A repeatability test-day model with a heat stress function was used to evaluate the genetic merit for milk, fat, and protein yield under heat stress and at thermal comfort for first parity in 5 regions in Canada. The heat stress function for each trait was defined using a specific temperature-humidity index (THI) threshold. The purpose of this function was to quantify the level of heat stress that was experienced by the dairy cattle. The estimated genetic correlation between the general additive genetic effect and the additive effect on the slope of the change in the trait phenotype for milk, fat, and protein yield ranged from -0.16 to -0.30, -0.20 to -0.44, and -0.28 to -0.42, respectively. These negative correlations imply that there is an antagonistic relationship between sensitivity to heat stress and level of production. The heritabilities for milk, fat, and protein yield at 15 units above the THI threshold ranged from 0.15 to 0.27, 0.11 to 0.15, and 0.11 to 0.15, respectively. Finally, the rank correlations between the breeding values from a repeatability model with no heat stress effect and the breeding values accounting for heat stress for the 100 top-ranked bulls indicated possible interaction between milk production traits and THI, resulting in substantial reranking of the top-ranked sires in Canada, especially for milk yield. This is the first study to implement weather data from the NASA POWER database in a genetic evaluation of heat tolerance in dairy cattle. The NASA POWER database is a novel alternative meteorological resource that is potentially more reliable and consistent and with broader coverage than weather station data increasing the number of animals that could be included in a heat stress evaluation.

Review: When worlds collide - poultry modeling in the 'Big Data' era.

Within poultry production systems, models have provided vital decision support, opportunity analysis, and performance optimization capabilities to nutritionists and producers for decades. In recent years, due to the advancement of digital and sensor technologies, 'Big Data' streams have emerged, optimally positioned to be analyzed by machine-learning (ML) modeling approaches, with strengths in forecasting and prediction. This review explores the evolution of empirical and mechanistic models in poultry production systems, and how these models may interact with new digital tools and technologies. This review will also examine the emergence of ML and Big Data in the poultry production sector, and the emergence of precision feeding and automation of poultry production systems. There are several promising directions for the field, including: (1) application of Big Data analytics (e.g., sensor-based technologies, precision feeding systems) and ML methodologies (e.g., unsupervised and supervised learning algorithms) to feed more precisely to production targets given a 'known' individual animal, and (2) combination and hybridization of data-driven and mechanistic modeling approaches to bridge decision support with improved forecasting capabilities.

Phenotypic analysis of heat stress in Holsteins using test-day production records and NASA POWER meteorological data.

Weather station data and test-day production records can be combined to quantify the effects of heat stress on production traits in dairy cattle. However, meteorological data sets that are retrieved from ground-based weather stations can be limited by spatial and temporal data gaps. The National Aeronautics and Space Administration Prediction of Worldwide Energy Resources (NASA POWER) database provides meteorological data over regions where surface measurements are sparse or nonexistent. The first aim of this study was to determine whether NASA POWER data are a viable alternative resource of weather data for studying heat stress in Canadian Holsteins. The results showed that average, minima, and maxima ambient temperature and dewpoint temperature as well as 4 different types of temperature-humidity index (THI) values from NASA POWER were highly correlated to the corresponding values from weather stations (regression R2 > 0.80). However, the NASA POWER values for the daily average, minima, and maxima wind speed and relative humidity were poorly correlated to the corresponding weather station values (regression R2 = 0.10 to 0.49). The second aim of this study was to quantify the influence of heat stress on Canadian dairy cattle. This was achieved by determining the THI values at which milk, protein, and fat yield started to decline due to heat stress as well as the rates of decline in these traits after the respective thresholds, using segmented polynomial regression models. This was completed for both primiparous and multiparous cows from 5 regions in Canada (Ontario, Quebec, British Columbia, the Prairies, and the Atlantic Maritime). The results showed that all production traits were negatively affected by heat stress and that the patterns of responses for milk, fat, and protein yields to increasing THI differed from each other. We found 3 THI thresholds for milk yield, 1 for fat yield, and 2 for protein yield. All thresholds marked a change in rate of decrease in production yield per unit THI, except for the first milk yield threshold, which marked a greater rate of increase. The first thresholds for milk yield ranged between 47 and 50, the second thresholds ranged between 61 and 69, and the third thresholds ranged between 72 and 76 THI units. The single THI threshold for fat yield ranged between 48 and 55 THI units. Finally, the first and second thresholds ranged between 58 and 62 THI units and 72 and 73 THI units for protein yield, respectively.

Machine learning classification of breeding protocol descriptions from Canadian Holsteins.

Dairy farmers are motivated to ensure cows become pregnant in an optimal and timely manner. Although timed artificial insemination (TAI) is a successful management tool in dairy cattle, it masks an animal's innate fertility performance, likely reducing the accuracy of genetic evaluations for fertility traits. Therefore, separating fertility traits based on the recorded management technique involved in the breeding process or adding the breeding protocol as an effect to the model can be viable approaches to address the potential bias caused by such management decisions. Nevertheless, there is a lack of specificity and uniformity in the recording of breeding protocol descriptions by dairy farmers. Therefore, this study investigated the use of 8 supervised machine learning algorithms to classify 1,835 unique breeding protocol descriptions from 981 herds into the following 2 classes: TAI or other than TAI. Our results showed that models that used a stacking classifier algorithm had the highest Matthews correlation coefficient (0.94 ± 0.04, mean ± SD) and maximized precision and recall (F1-score = 0.96 ± 0.03) on test data. Nonetheless, their F1-scores on test data were not different from 5 out of the other 7 algorithms considered. Altogether, results presented herein suggest machine learning algorithms can be used to produce robust models that correctly identify TAI protocols from dairy cattle breeding records, thus opening the opportunity for unbiased genetic evaluation of animals based on their natural fertility.

Review: Synergy between mechanistic modelling and data-driven models for modern animal production systems in the era of big data.

Mechanistic models (MMs) have served as causal pathway analysis and 'decision-support' tools within animal production systems for decades. Such models quantitatively define how a biological system works based on causal relationships and use that cumulative biological knowledge to generate predictions and recommendations (in practice) and generate/evaluate hypotheses (in research). Their limitations revolve around obtaining sufficiently accurate inputs, user training and accuracy/precision of predictions on-farm. The new wave in digitalization technologies may negate some of these challenges. New data-driven (DD) modelling methods such as machine learning (ML) and deep learning (DL) examine patterns in data to produce accurate predictions (forecasting, classification of animals, etc.). The deluge of sensor data and new self-learning modelling techniques may address some of the limitations of traditional MM approaches - access to input data (e.g. sensors) and on-farm calibration. However, most of these new methods lack transparency in the reasoning behind predictions, in contrast to MM that have historically been used to translate knowledge into wisdom. The objective of this paper is to propose means to hybridize these two seemingly divergent methodologies to advance the models we use in animal production systems and support movement towards truly knowledge-based precision agriculture. In order to identify potential niches for models in animal production of the future, a cross-species (dairy, swine and poultry) examination of the current state of the art in MM and new DD methodologies (ML, DL analytics) is undertaken. We hypothesize that there are several ways via which synergy may be achieved to advance both our predictive capabilities and system understanding, being: (1) building and utilizing data streams (e.g. intake, rumination behaviour, rumen sensors, activity sensors, environmental sensors, cameras and near IR) to apply MM in real-time and/or with new resolution and capabilities; (2) hybridization of MM and DD approaches where, for example, a ML framework is augmented by MM-generated parameters or predicted outcomes and (3) hybridization of the MM and DD approaches, where biological bounds are placed on parameters within a MM framework, and the DD system parameterizes the MM for individual animals, farms or other such clusters of data. As animal systems modellers, we should expand our toolbox to explore new DD approaches and big data to find opportunities to increase understanding of biological systems, find new patterns in data and move the field towards intelligent, knowledge-based precision agriculture systems.