Predicting dry matter intake in Canadian Holstein dairy cattle using milk mid-infrared reflectance spectroscopy and other commonly available predictors via artificial neural networks

Dry matter intake (DMI) is a fundamental component of the animal's feed efficiency, but measuring DMI of individual cows is expensive. Mid-infrared reflectance spectroscopy (MIRS) on milk samples could be an inexpensive alternative to predict DMI. The objectives of this study were (1) to assess if milk MIRS data could improve DMI predictions of Canadian Holstein cows using artificial neural networks (ANN); (2) to investigate the ability of different ANN architectures to predict unobserved DMI; and (3) to validate the robustness of developed prediction models. A total of 7,398 milk samples from 509 dairy cows distributed over Canada, Denmark, and the United States were analyzed. Data from Denmark and the United States were used to increase the training data size and variability to improve the generalization of the prediction models over the lactation. For each milk spectra record, the corresponding weekly average DMI (kg/d), test-day milk yield (MY, kg/d), fat yield (FY, g/d), and protein yield (PY, g/d), metabolic body weight (MBW), age at calving, year of calving, season of calving, days in milk, lactation number, country, and herd were available. The weekly average DMI was predicted with various ANN architectures using 7 predictor sets, which were created by different combinations MY, FY, PY, MBW, and MIRS data. All predictor sets also included age of calving and days in milk. In addition, the classification effects of season of calving, country, and lactation number were included in all models. The explored ANN architectures consisted of 3 training algorithms (Bayesian regularization, Levenberg-Marquardt, and scaled conjugate gradient), 2 types of activation functions (hyperbolic tangent and linear), and from 1 to 10 neurons in hidden layers). In addition, partial least squares regression was also applied to predict the DMI. Models were compared using cross-validation based on leaving out 10% of records (validation A) and leaving out 10% of cows (validation B). Superior fitting statistics of models comprising MIRS information compared with the models fitting milk, fat and protein yields suggest that other unknown milk components may help explain variation in weekly average DMI. For instance, using MY, FY, PY, and MBW as predictor variables produced a predictive accuracy (r) ranging from 0.510 to 0.652 across ANN models and validation sets. Using MIRS together with MY, FY, PY, and MBW as predictors resulted in improved fitting (r = 0.679–0.777). Including MIRS data improved the weekly average DMI prediction of Canadian Holstein cows, but it seems that MIRS predicts DMI mostly through its association with milk production traits and its utility to estimate a measure of feed efficiency that accounts for the level of production, such as residual feed intake, might be limited and needs further investigation. The better predictive ability of nonlinear ANN compared with linear ANN and partial least squares regression indicated possible nonlinear relationships between weekly average DMI and the predictor variables. In general, ANN using Bayesian regularization and scaled conjugate gradient training algorithms yielded slightly better weekly average DMI predictions compared with ANN using the Levenberg-Marquardt training algorithm.     

Predicting the dry matter intake of grazing dairy cows using infrared reflectance spectroscopy analysis

The objective of this study was to compare mid-infrared reflectance spectroscopy (MIRS) analysis of milk and near-infrared reflectance spectroscopy (NIRS) analysis of feces with regard to their ability to predict the dry matter intake (DMI) of lactating grazing dairy cows. A data set comprising 1,074 records of DMI from 457 cows was available for analysis. Linear regression and partial least squares regression were used to develop the equations using the following variables: (1) milk yield (MY), fat percentage, protein percentage, body weight (BW), stage of lactation (SOL), and parity (benchmark equation); (2) MIRS wavelengths; (3) MIRS wavelengths, MY, fat percentage, protein percentage, BW, SOL, and parity; (4) NIRS wavelengths; (5) NIRS wavelengths, MY, fat percentage, protein percentage, BW, SOL, and parity; (6) MIRS and NIRS wavelengths; and (7) MIRS wavelengths, NIRS wavelengths, MY, fat percentage, protein percentage, BW, SOL, and parity. The equations were validated both within herd using animals from similar experiments and across herds using animals from independent experiments. The accuracy of equations was greater for within-herd validation compared with across-herds validation. Across-herds validation was deemed the more suitable method to assess equations for robustness and real-world application. The benchmark equation was more accurate [coefficient of determination (R²) = 0.60; root mean squared error (RMSE) = 1.68 kg] than MIRS alone (R² = 0.30; RMSE = 2.23 kg) or NIRS alone (R² = 0.16; RMSE = 2.43 kg). The combination of the benchmark equation with MIRS (R² = 0.64; RMSE = 1.59 kg) resulted in slightly superior fitting statistics compared with the benchmark equation alone. The combination of the benchmark equation with NIRS (R² = 0.58; RMSE = 1.71 kg) did not result in a more accurate prediction equation than the benchmark equation. The combination of MIRS and NIRS wavelengths resulted in superior fitting statistics compared with either method alone (R² = 0.36; RMSE = 2.15 kg). The combination of the benchmark equation and MIRS and NIRS wavelengths resulted in the most accurate equation (R² = 0.68; RMSE = 1.52 kg). A further analysis demonstrated that Holstein-Friesian cows could predict the DMI of Jersey × Holstein-Friesian crossbred cows using both MIRS and NIRS. Similarly, the Jersey × Holstein-Friesian animals could predict the DMI of Holstein-Friesian cows using both MIRS and NIRS. The equations developed in this study have the capacity to predict DMI of grazing dairy cows. From a practicality perspective, MIRS in combination with variables in the benchmark equation is the most suitable equation because MIRS is currently used on all milk-recorded milk samples from dairy cows.     

Genetic parameters of mid-infrared methane predictions and their relationships with milk production traits in Holstein cattle

Many countries have pledged to reduce greenhouse gases. In this context, the dairy sector is one of the identified sectors to adapt production circumstances to address socio-environmental constraints due to its large carbon footprint related to CH₄ emission. This study aimed mainly to estimate (1) the genetic parameters of 2 milk mid-infrared-based CH₄ proxies [predicted daily CH₄ emission (PME, g/d), and log-transformed predicted CH₄ intensity (LMI)] and (2) their genetic correlations with milk production traits [milk (MY), fat (FY), and protein (PY) yields] from first- and second-parity Holstein cows. A total of 336,126 and 231,400 mid-infrared CH₄ phenotypes were collected from 56,957 and 34,992 first- and second-parity cows, respectively. The PME increased from the first to the second lactation (433 vs. 453 g/d) and the LMI decreased (2.93 vs. 2.86). We used 20 bivariate random regression test-day models to estimate the variance components. Moderate heritability values were observed for both CH₄ traits, and those values decreased slightly from the first to the second lactation (0.25 ± 0.01 and 0.22 ± 0.01 for PME; 0.18 ± 0.01 and 0.17 ± 0.02 for LMI). Lactation phenotypic and genetic correlations were negative between PME and MY in both first and second lactations (?0.07 vs. ?0.07 and ?0.19 vs. ?0.24, respectively). More close scrutiny revealed that relative increase of PME was lower with high MY levels even reverting to decrease, and therefore explaining the negative correlations, indicating that higher producing cows could be a mitigation option for CH₄ emission. The PME phenotypic correlations were almost equal to 0 with FY and PY for both lactations. However, the genetic correlations between PME and FY were slightly positive (0.11 and 0.12), whereas with PY the correlations were slightly negative (?0.05 and ?0.04). Both phenotypic and genetic correlations between LMI and MY or PY or FY were always relatively highly negative (from ?0.21 to ?0.88). As the genetic correlations between PME and LMI were strong (0.71 and 0.72 in first and second lactation), the selection of one trait would also strongly influence the other trait. However, in animal breeding context, PME, as a direct quantity CH₄ proxy, would be preferred to LMI, which is a ratio trait of PME with a trait already in the index. The range of PME sire estimated breeding values were 22.1 and 29.41 kg per lactation in first and second parity, respectively. Further studies must be conducted to evaluate the effect of the introduction of PME in a selection index on the other traits already included in this index, such as, for instance, fertility or longevity.     

Integrating heterogeneous across-country data for proxy-based random forest prediction of enteric methane in dairy cattle

Direct measurements of methane (CH₄) from individual animals are difficult and expensive. Predictions based on proxies for CH₄ are a viable alternative. Most prediction models are based on multiple linear regressions (MLR) and predictor variables that are not routinely available in commercial farms, such as dry matter intake (DMI) and diet composition. The use of machine learning (ML) algorithms to predict CH₄ emissions from across-country heterogeneous data sets has not been reported. The objectives were to compare performances of ML ensemble algorithm random forest (RF) and MLR models in predicting CH₄ emissions from proxies in dairy cows, and assess effects of imputing missing data points on prediction accuracy. Data on CH₄ emissions and proxies for CH₄ from 20 herds were provided by 10 countries. The integrated data set contained 43,519 records from 3,483 cows, with 18.7% missing data points imputed using k-nearest neighbor imputation. Three data sets were created, 3k (no missing records), 21k (missing DMI imputed from milk, fat, protein, body weight), and 41k (missing DMI, milk fat, and protein records imputed). These data sets were used to test scenarios (with or without DMI, imputed vs. nonimputed DMI, milk fat, and protein), and prediction models (RF vs. MLR). Model predictive ability was evaluated within and between herds through 10-fold cross-validation. Prediction accuracy was measured as correlation between observed and predicted CH₄, root mean squared error (RMSE) and mean normalized discounted cumulative gain (NDCG). Inclusion of DMI in the model improved within and between-herd prediction accuracy to 0.77 (RMSE = 23.3%) and 0.58 (RMSE = 31.9%) in RF and to 0.50 (RMSE = 0.327) and 0.13 (RMSE = 42.71) in MLR, respectively than when DMI was not included in the predictive model. When missing DMI records were imputed, within and between-herd accuracy increased to 0.84 (RMSE = 18.5%) and 0.63 (RMSE = 29.9%), respectively. In all scenarios, RF models out-performed MLR models. Results suggest routinely measured variables from dairy farms can be used in developing globally robust prediction models for CH₄ if coupled with state-of-the-art techniques for imputation and advanced ML algorithms for predictive modeling.     

Comparative accuracies of genetic values predicted for economically important milk traits,genome-wide association,and linkage disequilibrium patterns of Canadian Holstein cows

Genomic selection methodologies and genome-wide association studies use powerful statistical procedures that correlate large amounts of high-density SNP genotypes and phenotypic data. Actual 305-d milk (MY), fat (FY), and protein (PY) yield data on 695 cows and 76,355 genotyping-by-sequencing-generated SNP marker genotypes from Canadian Holstein dairy cows were used to characterize linkage disequilibrium (LD) structure of Canadian Holstein cows. Also, the comparison of pedigree-based BLUP, genomic BLUP (GBLUP), and Bayesian (BayesB) statistical methods in the genomic selection methodologies and the comparison of Bayesian ridge regression and BayesB statistical methods in the genome-wide association studies were carried out for MY, FY, and PY. Results from LD analysis revealed that as marker distance decreases, LD increases through chromosomes. However, unexpected high peaks in LD were observed between marker pairs with larger marker distances on all chromosomes. The GBLUP and BayesB models resulted in similar heritability estimates through 10-fold cross-validation for MY and PY; however, the GBLUP model resulted in higher heritability estimates than BayesB model for FY. The predictive ability of GBLUP model was significantly lower than that of BayesB for MY, FY, and PY. Association analyses indicated that 28 high-effect markers and markers on Bos taurus autosome 14 located within 6 genes (DOP1B, TONSL, CPSF1, ADCK5, PARP10, and GRINA) associated significantly with FY.     

Genetic parameters for buffalo milk yield and milk quality traits using Bayesian inference

The availability of accurate genetic parameters for important economic traits in milking buffaloes is critical for implementation of a genetic evaluation program. In the present study, heritabilities and genetic correlations for fat (FY305), protein (PY305), and milk (MY305) yields, milk fat (%F) and protein (%P) percentages, and SCS were estimated using Bayesian methodology. A total of 4,907 lactations from 1,985 cows were used. The (co)variance components were estimated using multiple-trait analysis by Bayesian inference method, applying an animal model, through Gibbs sampling. The model included the fixed effects of contemporary groups (herd-year and calving season), number of milking (2 levels), and age of cow at calving as (co)variable (quadratic and linear effect). The additive genetic, permanent environmental, and residual effects were included as random effects in the model. The posterior means of heritability distributions for MY305, FY305, PY305, %F, P%, and SCS were 0.22, 0.21, 0.23, 0.33, 0.39, and 0.26, respectively. The genetic correlation estimates ranged from −0.13 (between %P and SCS) to 0.94 (between MY305 and PY305). The permanent environmental correlation estimates ranged from −0.38 (between MY305 and %P) to 0.97 (between MY305 and PY305). Residual and phenotypic correlation estimates ranged from −0.26 (between PY305 and SCS) to 0.97 (between MY305 and PY305) and from −0.26 (between MY305 and SCS) to 0.97 (between MY305 and PY305), respectively. Milk yield, milk components, and milk somatic cells counts have enough genetic variation for selection purposes. The genetic correlation estimates suggest that milk components and milk somatic cell counts would be only slightly affected if increasing milk yield were the selection goal. Selecting to increase FY305 or PY305 will also increase MY305, %P, and %F.     

Genetic analysis of milk urea nitrogen and relationships with yield and fertility across lactation

The aim of this project was to investigate the relationship of milk urea nitrogen (MUN) with 3 milk production traits [milk yield (MY), fat yield (FY), protein yield (PY)] and 6 fertility measures (number of inseminations, calving interval, interval from calving to first insemination, interval from calving to last insemination, interval from first to last insemination, and pregnancy at first insemination). Data consisted of 635,289 test-day records of MY, FY, PY, and MUN on 76,959 first-lactation Swedish Holstein cows calving from 2001 to 2003, and corresponding lactation records for the fertility traits. Yields and MUN were analyzed with a random regression model followed by a multi-trait model in which the lactation was broken into 10 monthly periods. Heritability for MUN was stable across lactation (between 0.16 and 0.18), whereas MY, FY, and PY had low heritability at the beginning of lactation, which increased with time and stabilized after 100 d in milk, at 0.47, 0.36, and 0.44, respectively. Fertility traits had low heritabilities (0.02 to 0.05). Phenotypic correlations of MUN and milk production traits were between 0.13 (beginning of lactation) and 0.00 (end of lactation). Genetic correlations of MUN and MY, FY, and PY followed similar trends and were positive (0.22) at the beginning and negative (−0.15) at the end of lactation. Phenotypic correlations of MUN and fertility were close to zero. A surprising result was that genetic correlations of MUN and fertility traits suggest a positive relationship between the 2 traits for most of the lactation, indicating that animals with breeding values for increased MUN also had breeding values for improved fertility. This result was obtained with a random regression model as well as with a multi-trait model. The analyzed group of cows had a moderate level of MUN concentration. In such a population MUN concentration may increase slightly due to selection for improved fertility. Conversely, selection for increased MUN concentration may improve fertility slightly.     

Accuracy of genomic predictions in Gyr (Bos indicus) dairy cattle

Genomic selection may accelerate genetic progress in breeding programs of indicine breeds when compared with traditional selection methods. We present results of genomic predictions in Gyr (Bos indicus) dairy cattle of Brazil for milk yield (MY), fat yield (FY), protein yield (PY), and age at first calving using information from bulls and cows. Four different single nucleotide polymorphism (SNP) chips were studied. Additionally, the effect of the use of imputed data on genomic prediction accuracy was studied. A total of 474 bulls and 1,688 cows were genotyped with the Illumina BovineHD (HD; San Diego, CA) and BovineSNP50 (50K) chip, respectively. Genotypes of cows were imputed to HD using FImpute v2.2. After quality check of data, 496,606 markers remained. The HD markers present on the GeneSeek SGGP-20Ki (15,727; Lincoln, NE), 50K (22,152), and GeneSeek GGP-75Ki (65,018) were subset and used to assess the effect of lower SNP density on accuracy of prediction. Deregressed breeding values were used as pseudophenotypes for model training. Data were split into reference and validation to mimic a forward prediction scheme. The reference population consisted of animals whose birth year was ≤2004 and consisted of either only bulls (TR1) or a combination of bulls and dams (TR2), whereas the validation set consisted of younger bulls (born after 2004). Genomic BLUP was used to estimate genomic breeding values (GEBV) and reliability of GEBV (R²_PEV) was based on the prediction error variance approach. Reliability of GEBV ranged from ∼0.46 (FY and PY) to 0.56 (MY) with TR1 and from 0.51 (PY) to 0.65 (MY) with TR2. When averaged across all traits, R²_PEV were substantially higher (R²_PEV of TR1 = 0.50 and TR2 = 0.57) compared with reliabilities of parent averages (0.35) computed from pedigree data and based on diagonals of the coefficient matrix (prediction error variance approach). Reliability was similar for all the 4 marker panels using either TR1 or TR2, except that imputed HD cow data set led to an inflation of reliability. Reliability of GEBV could be increased by enlarging the limited bull reference population with cow information. A reduced panel of ∼15K markers resulted in reliabilities similar to using HD markers. Reliability of GEBV could be increased by enlarging the limited bull reference population with cow information.     

Short communication: Heterosis and breed effects for milk production and udder health traits in crosses between Danish Holstein,Danish Red,and Danish Jersey

During the last decade, the use of systematic crossbreeding in dairy cattle herds has increased in several countries of the world. The aim of this study was to estimate the effect of breed proportion and heterosis on milk production traits and udder health traits in dairy cattle. The study was based on records on milk yield (MY), protein yield (PY), fat yield (FY), somatic cell score (SCS), and mastitis (MAST) from 73,695 first-lactation dairy cows in 130 Danish herds applying systematic crossbreeding programs. Around 45% of the cows were crosses between Danish Holstein (DH), Danish Red (DR), or Danish Jersey (DJ), and the remaining were purebred DH, DR, or DJ. The statistical model included the fixed effects of herd-year, calving month, and calving age and an effect representing the lactation status of the cow. In addition, the model included a regression on calving interval from first to second lactation, a regression on the proportion of DH, DR, and DJ genes, and a regression on the degree of heterozygosity between DH and DR, DH and DJ, and DR and DJ. Random effects were the genetic effect of the cow and a residual. The effect of breed proportions was estimated relatively to DH. For MY, a pure DR yielded 461 kg milk less than DH, whereas a pure DJ yielded 2,259 kg milk less than a pure DH. Compared with DH, PY was 41.7 kg less for DJ, whereas PY for DR was 4.0 kg less than for DH. For FY, a DR yielded 10.6 kg less than DH, whereas there was no significant effect of breed proportion between DJ and DH. A DR cow had lower SCS (0.13) than DH, whereas DJ had higher SCS (0.14) than DH. There was no significant effect of breed proportion on MAST between the 3 breeds. Heterosis was significant in all combinations of breeds for MY, FY, and PY. Heterosis for crosses between DH and DR was 257 kg (3.2%), 11.9 kg (3.2%), and 8.9 kg (3.2%) for MY, PY, and FY, respectively. Corresponding figures for crosses between DH and DJ were 314 kg (4.4%), 14.3 kg (4.4%), and 10.4 kg (4.0%), whereas heterosis between DR and DJ was 462 kg (6.7%), 19.6 kg (6.7%), and 13.9 kg (5.4%) for MY, PY, and FY, respectively. Heterosis was only significant for SCS in the crosses between DH and DR. Heterosis effects for MAST were nonsignificant for all the crosses. The results obtained in this study demonstrate that in first lactation cows, there is a positive effect of heterosis on milk production traits, but limited effect on udder health traits.     

Methodological guidelines: Cow milk mid-infrared spectra to predict reference enteric methane data collected by an automated head-chamber system

Various methodological protocols were tested on milk samples from cows fed diets affecting both methanogenesis and milk synthesis to identify the best approach for the prediction of GreenFeed system (GF) measured methane (CH₄) emissions by milk mid-infrared (MIR) spectroscopy. The models developed were also tested on a data set from cows fed chemical inhibitors of CH₄ emission [3-nitrooxypropanol (3NOP)] that just marginally affect milk composition. A total of 129 primiparous and multiparous Holstein cows fed diets with different methanogenic potential were considered. Individual milk yield (MY) and dry matter intake were recorded daily, whereas fat- and protein-corrected milk (FPCM) was recorded twice a week. The MIR spectra from 2 consecutive milkings were collected twice a week. Twenty CH₄ spot measurements with GF were taken as the basic measurement unit (BMU) of CH₄. The equations were built using partial least squares regression by splitting the database into calibration and validation data sets (excluding 3NOP samples). Models were developed for milk MIR spectra by milking and on day spectra obtained by averaging spectra from 2 consecutive milkings. Models based on day spectra were calibrated by using CH₄ reference data for a measurement duration of 1, 2, 3, or 4 BMU. Models built from the average of the day spectra collected during the corresponding CH₄ measurement periods were developed. Corrections of spectra by days in milk (DIM) and the inclusion of parity, MY, and FPCM as explanatory variables were tested as tools to improve model performance. Models built on day milk MIR spectra gave slightly better performances that those developed using spectra from a single milking. Long duration of CH₄ measurement by GF performed better than short duration: the coefficient of determination of validation (R²V) for CH₄ emissions expressed in grams per day were 0.60 vs. 0.52 for 4 and 1 BMU, respectively. When CH₄ emissions were expressed as grams per kilogram of dry of matter intake, grams per kilogram of MY, or grams per kilogram of FPCM, performance with a long duration also improved. Coupling GF reference data with the average of milk MIR spectra collected throughout the corresponding CH₄ measurement period gave better predictions than using day spectra (R²V = 0.70 vs. 0.60 for CH₄ as g/d on 4 BMU). Correcting the day spectra by DIM improved R²V compared with the equivalent DIM-uncorrected models (R²V = 0.67 vs. 0.60 for CH₄ as g/d on 4 BMU). Adding other phenotypic information as explanatory variables did not further improve the performance of models built on single day DIM-corrected spectra, whereas including MY (or FPCM) improved the performance of models built on the average of spectra (uncorrected by DIM) recorded during the CH₄ measurement period (R²V = 0.73 vs. 0.70 for CH₄ as g/d on 4 BMU). When validating the models on the 3NOP data set, predictions were poor without (R²V = 0.13 for CH₄ as g/d on 1 BMU) or with (R²V = 0.31 for CH₄ as g/d on 1 BMU) integration of 3NOP data in the models. Thus, specific models would be required for CH₄ prediction when cows receive chemical inhibitors of CH₄ emissions not affecting milk composition.     

Mining data from milk infrared spectroscopy to improve feed intake predictions in lactating dairy cows

Feed intake is one of the most important components of feed efficiency in dairy systems. However, it is a difficult trait to measure in commercial operations for individual cows. Milk spectrum from mid-infrared spectroscopy has been previously used to predict milk traits, and could be an alternative to predict dry matter intake (DMI). The objectives of this study were (1) to evaluate if milk spectra can improve DMI predictions based only on cow variables; (2) to compare artificial neural network (ANN) and partial least squares (PLS) predictions; and (3) to evaluate if wavelength (WL) selection through Bayesian network (BN) improves prediction quality. Milk samples (n = 1,279) from 308 mid-lactation dairy cows [127 ± 27 d in milk (DIM)] were collected between 2014 and 2016. For each milk spectra time point, DMI (kg/d), body weight (BW, kg), milk yield (MY, kg/d), fat (%), protein (%), lactose (%), and actual DIM were recorded. The DMI was predicted with ANN and PLS using different combinations of explanatory variables. Such combinations, called covariate sets, were as follows: set 1 (MY, BW^0.75, DIM, and 361 WL); set 2 [MY, BW^0.75, DIM, and 33 WL (WL selected by BN)]; set 3 (MY, BW^0.75, DIM, and fat, protein, and lactose concentrations); set 4 (MY, BW^0.75, DIM, 33 WL, fat, protein, and lactose); set 5 (MY, BW^0.75, DIM, 33 WL, and visit duration in the feed bunk); set 6 (MY, DIM, and 33 WL); set 7 (MY, BW^0.75, and DIM); set-WL (included 361 WL); and set-BN (included just 33 selected WL). All models (i.e., each combination of covariate set and fitting approach, ANN or PLS) were validated with an external data set. The use of ANN improved the performance of models 2, 5, 6, and BN. The use of BN combined with ANN yielded the highest accuracy and precision. The addition of individual WL compared with milk components (set 2 vs. set 3) did not improve prediction quality when using PLS. However, when ANN was employed, the model prediction with the inclusion of 33 WL was improved over the model containing only milk components (set 2 vs. set 3; concordance correlation coefficient = 0.80 vs. 0.72; coefficient of determination = 0.67 vs. 0.53; root mean square error of prediction 2.36 vs. 2.81 kg/d). The use of ANN and the inclusion of a behavior parameter, set 5, resulted in the best predictions compared with all other models (coefficient of determination = 0.70, concordance correlation coefficient = 0.83, root mean square error of prediction = 2.15 kg/d). The addition of milk spectra information to models containing cow variables improved the accuracy and precision of DMI predictions in lactating dairy cows when ANN was used. The use of BN to select more informative WL improved the model prediction when combined with cow variables, with further improvement when combined with ANN.     

Lactational performance,rumen fermentation,and enteric methane emission of dairy cows fed an amylase-enabled corn silage

This study investigated the effects of an amylase-enabled corn silage on lactational performance, enteric CH₄ emission, and rumen fermentation of lactating dairy cows. Following a 2-wk covariate period, 48 Holstein cows were blocked based on parity, days in milk, milk yield (MY), and CH₄ emission. Cows were randomly assigned to 1 of 2 treatments in an 8-wk randomized complete block design experiment: (1) control corn silage (CON) from an isogenic corn without α-amylase trait and (2) Enogen hybrid corn (Syngenta Seeds LLC) harvested as silage (ECS) containing a bacterial transgene expressing α-amylase (i.e., amylase-enabled) in the endosperm of the grain. The ECS and CON silages were included at 40% of the dietary dry matter (DM) and contained, on average, 43.3 and 41.8% DM and (% DM) 36.7 and 37.5% neutral detergent fiber, and 36.1 and 33.1% starch, respectively. Rumen samples were collected from a subset of 10 cows using the ororuminal sampling technique on wk 3 of the experimental period. Enteric CH₄ emission was measured using the GreenFeed system (C-Lock Inc.). Dry matter intake (DMI) was similar between treatments. Compared with CON, MY (38.8 vs. 40.8 kg/d), feed efficiency (1.47 vs. 1.55 kg of MY/kg of DMI), and milk true protein (1.20 vs. 1.25 kg/d) and lactose yields (1.89 vs. 2.00 kg/d) were increased, whereas milk urea nitrogen (14.0 vs. 12.7 mg/dL) was decreased, with the ECS diet. No effect of treatment on energy-corrected MY (ECM) was observed, but a trend was detected for increased ECM feed efficiency (1.45 vs. 1.50 kg of ECM/kg of DMI) for cows fed ECS compared with CON-fed cows. Daily CH₄ emission was not affected by treatment, but emission intensity was decreased with the ECS diet (11.1 vs. 10.3 g/kg of milk, CON and ECS, respectively); CH₄ emission intensity on ECM basis was not different between treatments. Rumen fermentation, apart from a reduced molar proportion of butyrate in ECS-fed cows, was not affected by treatment. Apparent total-tract digestibility of nutrients and urinary and fecal nitrogen excretions, apart from a trend for increased DM digestibility by ECS-fed cows, were not affected by treatment. Overall, ECS inclusion at 40% of dietary DM increased milk, milk protein, and lactose yields and feed efficiency, and tended to increase ECM feed efficiency but had no effect on ECM yield in dairy cows. The increased MY with ECS led to a decrease in enteric CH₄ emission intensity, compared with the control silage.     

Short communication: Development of an equation for estimating methane emissions of dairy cows from milk Fourier transform mid-infrared spectra by using reference data obtained exclusively from respiration chambers

Evaluation and mitigation of enteric methane (CH₄) emissions from ruminant livestock, in particular from dairy cows, have acquired global importance for sustainable, climate-smart cattle production. Based on CH₄ reference measurements obtained with the SF₆ tracer technique to determine ruminal CH₄ production, a current equation permits evaluation of individual daily CH₄ emissions of dairy cows based on milk Fourier transform mid-infrared (FT-MIR) spectra. However, the respiration chamber (RC) technique is considered to be more accurate than SF₆ to measure CH₄ production from cattle. This study aimed to develop an equation that allows estimating CH₄ emissions of lactating cows recorded in an RC from corresponding milk FT-MIR spectra and to challenge its robustness and relevance through validation processes and its application on a milk spectral database. This would permit confirming the conclusions drawn with the existing equation based on SF₆ reference measurements regarding the potential to estimate daily CH₄ emissions of dairy cows from milk FT-MIR spectra. A total of 584 RC reference CH₄ measurements (mean ± standard deviation of 400 ± 72 g of CH₄/d) and corresponding standardized milk mid-infrared spectra were obtained from 148 individual lactating cows between 7 and 321 d in milk in 5 European countries (Germany, Switzerland, Denmark, France, and Northern Ireland). The developed equation based on RC measurements showed calibration and cross-validation coefficients of determination of 0.65 and 0.57, respectively, which is lower than those obtained earlier by the equation based on 532 SF₆ measurements (0.74 and 0.70, respectively). This means that the RC-based model is unable to explain the variability observed in the corresponding reference data as well as the SF₆-based model. The standard errors of calibration and cross-validation were lower for the RC model (43 and 47 g/d vs. 66 and 70 g/d for the SF₆ version, respectively), indicating that the model based on RC data was closer to actual values. The root mean squared error (RMSE) of calibration of 42 g/d represents only 10% of the overall daily CH₄ production, which is 23 g/d lower than the RMSE for the SF₆-based equation. During the external validation step an RMSE of 62 g/d was observed. When the RC equation was applied to a standardized spectral database of milk recordings collected in the Walloon region of Belgium between January 2012 and December 2017 (1,515,137 spectra from 132,658 lactating cows in 1,176 different herds), an average ± standard deviation of 446 ± 51 g of CH₄/d was estimated, which is consistent with the range of the values measured using both RC and SF₆ techniques. This study confirmed that milk FT-MIR spectra could be used as a potential proxy to estimate daily CH₄ emissions from dairy cows provided that the variability to predict is covered by the model.     

Methane prediction based on individual or groups of milk fatty acids for dairy cows fed rations with or without linseed

Milk fatty acids (MFA) are a proxy for the prediction of CH₄ emission from cows, and prediction differs with diet. Our objectives were (1) to compare the effect of diets on the relation between MFA profile and measured CH₄ production, (2) to predict CH₄ production based on 6 data sets differing in the number and type of MFA, and (3) to test whether additional inclusion of energy-corrected milk (ECM) yield or dry matter intake (DMI) as explanatory variables improves predictions. Twenty dairy cows were used. Four diets were used based on corn silage (CS) or grass silage (GS) without (L0) or with linseed (LS) supplementation. Ten cows were fed CS-L0 and CS-LS and the other 10 cows were fed GS-L0 and GS-LS in random order. In feeding wk 5 of each diet, CH₄ production (L/d) was measured in respiration chambers for 48 h and milk was analyzed for MFA concentrations by gas chromatography. Specific CH₄ prediction equations were obtained for L0-, LS-, GS-, and CS-based diets and for all 4 diets collectively and validated by an internal cross-validation. Models were developed containing either 43 identified MFA or a reduced set of 7 groups of biochemically related MFA plus C16:0 and C18:0. The CS and LS diets reduced CH₄ production compared with GS and L0 diets, respectively. Methane yield (L/kg of DMI) reduction by LS was higher with CS than GS diets. The concentrations of C18:1 trans and n-3 MFA differed among GS and CS diets. The LS diets resulted in a higher proportion of unsaturated MFA at the expense of saturated MFA. When using the data set of 43 individual MFA to predict CH₄ production (L/d), the cross-validation coefficient of determination (R²_CV) ranged from 0.47 to 0.92. When using groups of MFA variables, the R²_CV ranged from 0.31 to 0.84. The fit parameters of the latter models were improved by inclusion of ECM or DMI, but not when added to the data set of 43 MFA for all diets pooled. Models based on GS diets always had a lower prediction potential (R²_CV = 0.31 to 0.71) compared with data from CS diets (R²_CV = 0.56 to 0.92). Models based on LS diets produced lower prediction with data sets with reduced MFA variables (R²_CV = 0.62 to 0.68) compared with L0 diets (R²_CV = 0.67 to 0.80). The MFA C18:1 cis-9 and C24:0 and the monounsaturated FA occurred most often in models. In conclusion, models with a reduced number of MFA variables and ECM or DMI are suitable for CH₄ prediction, and CH₄ prediction equations based on diets containing linseed resulted in lower prediction accuracy.     

Random regression analysis of milk yield and milk composition in the first and second lactations of Murciano-Granadina goats

Records from the milk recording scheme of Spanish Murciano-Granadina goats were studied to estimate genetic (co)variance components and breeding values throughout the first and second lactations. The data used consisted of 49,696 monthly test-day records of milk (MY), protein (PY), fat (FY), and dry matter (DMY) yields from 5,163 goats, distributed in 20 herds, offspring of 2,086 does and 206 bucks. These records were analyzed by 2-trait random regression models (RRM) and a repeatability test-day model (RTDM). At the middle of lactation, heritability estimates for MY, DMY, and FY obtained with RTDM were larger than those estimated with RRM, and the opposite was true for PY. The RRM estimates of heritability for MY, FY, and PY were very similar throughout the trajectories of both lactations. Heritability estimates for DMY decreased through the lactation period. The genetic correlations between the first and second lactation records estimated for all traits by RRM were positive and ranged from 0.43 to 0.80 throughout the lactation curves. The correlation between BV estimated with RTDM and RRM was 0.742 for MY and 0.664 for DMY. The RRM could be a useful alternative to RTDM for the prediction of BV in this breed.     

Predicting methane emissions of lactating Danish Holstein cows using Fourier transform mid-infrared spectroscopy of milk

Enteric methane (CH₄), a potent greenhouse gas, is among the main targets of mitigation practices for the dairy industry. A measurement technique that is rapid, inexpensive, easy to use, and applicable at the population level is desired to estimate CH₄ emission from dairy cows. In the present study, feasibility of milk Fourier transform mid-infrared (FT-IR) spectral profiles as a predictor for CH₄:CO₂ ratio and CH₄ production (L/d) is explained. The partial least squares regression method was used to develop the prediction models. The models were validated using different random test sets, which are independent from the training set by leaving out records of 20% cows for validation and keeping records of 80% of cows for training the model. The data set consisted of 3,623 records from 500 Danish Holstein cows from both experimental and commercial farms. For both CH₄:CO₂ ratio and CH₄ production, low prediction accuracies were found when models were obtained using FT-IR spectra. Validated coefficient of determination (R²_Val) = 0.21 with validated model error root mean squared error of prediction (RMSEP) = 0.0114 L/d for CH₄:CO₂ ratio, and R²_Val = 0.13 with RMSEP = 111 L/d for CH₄ production. The important spectral wavenumbers selected using the recursive partial least squares method represented major milk components fat, protein, and lactose regions of the spectra. When fat and protein predicted by FT-IR were used instead of full spectra, a low R²_Val of 0.07 was obtained for both CH₄:CO₂ ratio and CH₄ production prediction. Other spectral wavenumbers related to lactose (carbohydrate) or additional wavenumbers related to fat or protein (amide II) are providing additional variation when using the full spectral profile. For CH₄:CO₂ ratio prediction, integration of FT-IR with other factors such as milk yield, herd, and lactation stage showed improvement in the prediction accuracy. However, overall prediction accuracy remained modest; R²_Val increased to 0.31 with RMSEP = 0.0105. For prediction of CH₄ production, the added value of FT-IR along with the aforementioned traits was marginal. These results indicated that for CH₄ production prediction, FT-IR profiles reflect primarily information related to milk yield, herd, and lactation stage rather than individual milk fatty acids related to CH₄ emission. Thus, it is not feasible to predict CH₄ emission based on FT-IR spectra alone.     

Reproducibility and repeatability of measures of milk coagulation properties and predictive ability of mid-infrared reflectance spectroscopy

The objectives of the study were to estimate the reproducibility and repeatability of milk coagulation properties (MCP) measured by a computerized renneting meter (CRM) and to evaluate the predictive ability of mid-infrared spectroscopy (MIRS) as an innovative technology for the assessment of rennet coagulation time (RCT, min) and curd firmness (a₃₀, mm). Four samples without addition of preservative (NP) and 4 samples with Bronopol addition (PS) were collected from each of 83 Holstein-Friesian cows. Six hours after collection, 2 replicated measures of MCP were obtained with CRM using 1 NP and 1 PS sample from each cow. Mid-infrared spectra of the remaining NP and PS samples from each animal were recorded after 6 h, 4 d, and 8 d after sampling. Two groups of calibration equations were developed using MIRS spectra and CRM measures of MCP as reference data obtained from analysis of NP and PS, respectively. Reproducibility and repeatability of CRM measures were obtained from REML estimation of variance components on the basis of a linear model including the fixed effects of herd and days in milk class and the random effects of cows, sample treatment (addition or no addition of preservative), and the interaction between cow and sample treatment. Coefficient of reproducibility is an indicator of the agreement between 2 measurements of MCP for the same milk sample preserved with or without addition of Bronopol. Coefficient of repeatability is an indicator of the agreement between repeated measures of MCP. Pearson correlations between MCP measures for NP and PS were 0.97 and 0.83 for RCT and a₃₀, respectively. Reproducibility of CRM measures under different preserving conditions of milk was 93.5% for RCT and 64.6% for a₃₀. Repeatabilities of RCT and a₃₀ measures were 95.7 and 77.3%, respectively. Based on the estimated cross-validation standard errors and coefficients of determination and ratios of standard errors of cross-validation to standard deviation of reference data, the predictive ability of MIRS calibration equations was moderate for RCT and unsatisfactory for a_30. Predictive ability of equations based on spectra and MCP measures of PS was greater than that of equations based on data of NP. The study did not provide conclusive evidence on the effectiveness of MIRS as a predictive tool for MCP and it requires an enlargement of the variability of milk sampling circumstances. Because the relevance of MIRS predictions in relation to breeding programs for MCP based on indicator traits relies on the genetic variation of MIRS predictions and on phenotypic and genetic correlations between MIRS predictions and MCP measures, additional specific investigations on these topics are needed.     

Short communication: Heritability of methane production and genetic correlations with milk yield and body weight in Holstein-Friesian dairy cows

Greenhouse gases originating from the dairy sector, including methane (CH₄), contribute to global warming. A possible strategy to reduce CH₄ production is to use genetic selection. This requires genetic parameters for CH₄ production and correlations with production traits. Data were available on 184 Holstein-Friesian cows. Methane production was measured in the milking robot during milking from December 2009 to April 2010. In total 2,456 observations for CH₄ production were available. Milk yield (MY) and body weight (BW) were obtained at every milking from November 2008 to October 2010. In total 4,567 observations for milk yield and 4,570 observations for BW were available. Restricted maximum likelihood, using random regression models, was used to analyze the data. Heritability (standard error given in parentheses) for CH₄ production ranged from 0.12 (0.16) to 0.45 (0.11), and genetic correlations with MY ranged from 0.49 (0.12) to 0.54 (0.26). The positive genetic correlation between CH₄ production and milk yield indicates that care needs to be taken when genetically selecting for lower CH₄ production, to avoid a decrease in MY at the animal level. However, this study shows that CH₄ production is moderately heritable and therefore progress through genetic selection is possible.     

Estimation of genetic parameters for productive life,reproduction, and milk-production traits in US dairy goats

Heritabilities and correlations for milk yield (MY), fat yield (FY), protein yield (PY), combined fat and protein yield (FPY), fat percentage (F%), protein percentage (P%), age at first kidding (AFK), interval between the first and second kidding (KI), and real and functional productive life at 72 mo (FPL72) of 33,725 US dairy goats, were estimated using animal models. Productive life was defined as the total days in production until 72 mo of age (PL72) for goats having the opportunity to express the trait. Functional productive life was obtained by correcting PL72 for MY, FY, PY, and final type score (FS). Six selection indexes were used, including or excluding PL72, with 6 groups of different economic weights, to estimate the responses to selection considering MY, FY, PY, and PL72 as selection criteria. The main criteria that determined the culling of a goat from the herd were low FS, MY, and FY per lactation. Heritability estimates were 0.22, 0.17, 0.37, 0.37, 0.38, 0.39, 0.54, 0.64, 0.09, and 0.16 for PL72, FPL72, MY, FY, PY, FPY, F%, P%, KI, and AFK, respectively. Most genetic correlations between the evaluated traits and PL72 or FPL72 were positive, except for F% (−0.04 and −0.06, respectively), P% (−0.002 and −0.03, respectively), and AFK (−0.03 and −0.01, respectively). The highest genetic correlations were between FPL72 and MY (0.39) and between PL72 and MY (0.33). Most phenotypic correlations between the traits evaluated and FPL72 and PL72 were positive (>0.23 and >0.26, respectively), except for F% (−0.004 and −0.02, respectively), P% (−0.05 and −0.02), KI (−0.01 and −0.07), and AFK (−0.08 and −0.08). The direct selection for PL72 increased it by 102.28 d per generation. The use of MY, FY, PY, KI, or AFK as selection criteria increased PL72 by 39.21, 27.33, 35.90, −8.28, or 2.77 d per generation, respectively. The inclusion of PL72 as selection criterion increased the expected response per generation from 0.15 to 17.35% in all selection indices studied.     

Use of milk fatty acids to estimate plasma nonesterified fatty acid concentrations as an indicator of animal energy balance

Negative energy balance is an important part of the lactation cycle, and measuring the current energy balance of a cow is useful in both applied and research settings. The objectives of this study were (1) to determine if milk fatty acid (FA) proportions were consistently related to plasma nonesterified fatty acids (NEFA); (2) to determine if an individual cow with a measured milk FA profile is above or below a NEFA concentration, (3) to test the universality of the models developed within the University of Wisconsin and US Dairy Forage Research Center cows. Blood samples were collected on the same day as milk sampling from 105 Holstein cows from 3 studies. Plasma NEFA was quantified and a threshold of 600 µEq/L was applied to classify animals above this concentration as having high NEFA (NEFA_high). Thirty milk FA proportions and 4 milk FA ratios were screened to evaluate their capacity to classify cows with NEFA_high according to determined milk FA threshold. In addition, 6 linear regression models were created using individual milk FA proportions and ratios. To evaluate the universality of the linear relationship between milk FA and plasma NEFA found in the internal data set, 90 treatment means from 21 papers published in the literature were compiled to test the model predictions. From the 30 screened milk FA, the odd short-chain fatty acids (C7:0, C9:0, C11:0, and C13:0) had sensitivity slightly greater than the other short-chain fatty acids (83.3, 94.8, 80.0, and 85.9%, respectively). The sensitivities for milk FA C6:0, C8:0, C10:0, and C12:0 were 78.8, 85.3, 80.1, and 83.9%, respectively. The threshold values to detect NEFA_high cows for the last group of milk FA were ≤2.0, ≤0.94, ≤1.4, and ≤1.8 g/100 g of FA, respectively. The milk FA C14:0 and C15:0 had sensitivities of 88.7 and 85.0% and a threshold of ≤6.8 and ≤0.53 g/100 g of FA, respectively. The linear regressions using the milk FA ratios C18:1 to C15:0 and C17:0 to C15:0 presented lower root mean square error (RMSE = 191 and 179 µEq/L, respectively) in comparison with individual milk FA proportions (RMSE = 194 µEq/L), C18:1 to even short-medium-chain fatty acid (C4:0–C12:0) ratio (RMSE = 220 µEq/L), and C18:1 to C14:0 (RMSE = 199 µEq/L). Models using milk FA ratios C18:1 to C15:0 and C17:0 to C15:0 had a better fit with the external data set in comparison with the other models. Plasma NEFA can be predicted by linear regression models using milk FA ratios.