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.

Estimation of genetic parameters of the productive and reproductive traits in Ethiopian Holstein using multi-trait models.

OBJECTIVE: This study estimated the genetic parameters for productive and reproductive traits. METHODS: The data included production and reproduction records of animals that have calved between 1979 and 2013. The genetic parameters were estimated using multivariate mixed models (DMU) package, fitting univariate and multivariate mixed models with average information restricted maximum likelihood algorithm. RESULTS: The estimates of heritability for milk production traits from the first three lactation records were 0.03±0.03 for lactation length (LL), 0.17±0.04 for lactation milk yield (LMY), and 0.15±0.04 for 305 days milk yield (305-d MY). For reproductive traits the heritability estimates were, 0.09±0.03 for days open (DO), 0.11±0.04 for calving interval (CI), and 0.47±0.06 for age at first calving (AFC). The repeatability estimates for production traits were 0.12±0.02, for LL, 0.39±0.02 for LMY, and 0.25±0.02 for 305-d MY. For reproductive traits the estimates of repeatability were 0.19±0.02 for DO, and to 0.23±0.02 for CI. The phenotypic correlations between production and reproduction traits ranged from 0.08±0.04 for LL and AFC to 0.42±0.02 for LL and DO. The genetic correlation among production traits were generally high (>0.7) and between reproductive traits the estimates ranged from 0.06±0.13 for AFC and DO to 0.99±0.01 between CI and DO. Genetic correlations of productive traits with reproductive traits were ranged from -0.02 to 0.99. CONCLUSION: The high heritability estimates observed for AFC indicated that reasonable genetic improvement for this trait might be possible through selection. The h2 and r estimates for reproductive traits were slightly different from single versus multi-trait analyses of reproductive traits with production traits. As single-trait method is biased due to selection on milk yield, a multi-trait evaluation of fertility with milk yield is recommended.

Long-term effects of early-life rumen microbiota modulation on dairy cow production performance and methane emissions.

Rumen microbiota modulation during the pre-weaning period has been suggested as means to affect animal performance later in life. In this follow-up study, we examined the post-weaning rumen microbiota development differences in monozygotic twin-heifers that were inoculated (T-group) or not inoculated (C-group) (n = 4 each) with fresh adult rumen liquid during their pre-weaning period. We also assessed the treatment effect on production parameters and methane emissions of cows during their 1st lactation period. The rumen microbiota was determined by the 16S rRNA gene, 18S rRNA gene, and ITS1 amplicon sequencing. Animal weight gain and rumen fermentation parameters were monitored from 2 to 12 months of age. The weight gain was not affected by treatment, but butyrate proportion was higher in T-group in month 3 (p = 0.04). Apart from archaea (p = 0.084), the richness of bacteria (p < 0.0001) and ciliate protozoa increased until month 7 (p = 0.004) and anaerobic fungi until month 11 (p = 0.005). The microbiota structure, measured as Bray-Curtis distances, continued to develop until months 3, 6, 7, and 10, in archaea, ciliate protozoa, bacteria, and anaerobic fungi, respectively (for all: p = 0.001). Treatment or age × treatment interaction had a significant (p < 0.05) effect on 18 bacterial, 2 archaeal, and 6 ciliate protozoan taxonomic groups, with differences occurring mostly before month 4 in bacteria, and month 3 in archaea and ciliate protozoa. Treatment stimulated earlier maturation of prokaryote community in T-group before month 4 and earlier maturation of ciliate protozoa at month 2 (Random Forest: 0.75 month for bacteria and 1.5 month for protozoa). No treatment effect on the maturity of anaerobic fungi was observed. The milk production and quality, feed efficiency, and methane emissions were monitored during cow's 1st lactation. The T-group had lower variation in energy-corrected milk yield (p < 0.001), tended to differ in pattern of residual energy intake over time (p = 0.069), and had numerically lower somatic cell count throughout their 1st lactation period (p = 0.081), but no differences between the groups in methane emissions (g/d, g/kg DMI, or g/kg milk) were observed. Our results demonstrated that the orally administered microbial inoculant induced transient changes in early rumen microbiome maturation. In addition, the treatment may influence the later production performance, although the mechanisms that mediate these effects need to be further explored.

Taxon abundance, diversity, co-occurrence and network analysis of the ruminal microbiota in response to dietary changes in dairy cows.

The ruminal microbiome, comprising large numbers of bacteria, ciliate protozoa, archaea and fungi, responds to diet and dietary additives in a complex way. The aim of this study was to investigate the benefits of increasing the depth of the community analysis in describing and explaining responses to dietary changes. Quantitative PCR, ssu rRNA amplicon based taxa composition, diversity and co-occurrence network analyses were applied to ruminal digesta samples obtained from four multiparous Nordic Red dairy cows fitted with rumen cannulae. The cows received diets with forage:concentrate ratio either 35:65 (diet H) or 65:35 (L), supplemented or not with sunflower oil (SO) (0 or 50 g/kg diet dry matter), supplied in a 4 × 4 Latin square design with a 2 × 2 factorial arrangement of treatments and four 35-day periods. Digesta samples were collected on days 22 and 24 and combined. QPCR provided a broad picture in which a large fall in the abundance of fungi was seen with SO in the H but not the L diet. Amplicon sequencing showed higher community diversity indices in L as compared to H diets and revealed diet specific taxa abundance changes, highlighting large differences in protozoal and fungal composition. Methanobrevibacter ruminantium and Mbb. gottschalkii dominated archaeal communities, and their abundance correlated negatively with each other. Co-occurrence network analysis provided evidence that no microbial domain played a more central role in network formation, that some minor-abundance taxa were at nodes of highest centrality, and that microbial interactions were diet specific. Networks added new dimensions to our understanding of the diet effect on rumen microbial community interactions.