Including milk production, conformation, and functional traits in multivariate models for genetic evaluation of lameness

A comprehensive meta-analysis of genetic parameters for resilience and productivity indicator traits in Holstein cattle

Selection for resilience indicator (RIND) traits in Holstein cattle is becoming an important breeding objective as the worldwide population is expected to be exposed to increased environmental stressors due to both climate change and changing industry standards. However, genetic correlations between RIND and productivity indicator (PIND) traits, which are already being selected for and have the most economic value, are often unfavorable. As a result, it is necessary to fully understand these genetic relationships when incorporating novel traits into selection indices, so that informed decisions can be made to fully optimize selection for both groups of traits. In the past 2 decades, there have been many estimates of RIND traits published in the literature, albeit in small populations. To provide valuable pooled summary estimates, a random-effects meta-analysis was conducted for heritability and genetic correlation estimates for PIND and RIND traits in worldwide Holstein cattle. In total, 926 heritability estimates for 9 PIND and 27 RIND traits, along with 362 estimates of genetic correlation (PIND × RIND traits) were collected. Resilience indicator traits were grouped into the following subgroups: Metabolic Diseases, Hoof Health, Udder Health, Fertility, Heat Tolerance, Longevity, and Other. Pooled estimates of heritability for PIND traits ranged from 0.201 ± 0.05 (energy-corrected milk) to 0.377 ± 0.06 (protein content), while pooled estimates of heritability for RIND traits ranged from 0.032 ± 0.02 (incidence of lameness, incidence of milk fever) to 0.497 ± 0.05 (measures of body weight). Pooled estimates of genetic correlations ranged from -0.360 ± 0.25 (protein content vs. milk acetone concentration) to 0.535 ± 0.72 (measures of fat-to-protein ratio vs. milk acetone concentration). Additionally, out of 243 potential genetic correlations between PIND and RIND traits that could have been reported, only 40 had enough published estimates to implement the meta-analysis model. Our results confirmed that the interactions between PIND and RIND traits are complex, and all relationships should be evaluated when incorporating novel traits into selection indices. This study provides a valuable reference for breeders looking to incorporate RIND traits for Holstein cattle into selection indices.

Effects of type traits, inbreeding, and production on survival in US Jersey cattle

Greater longevity is associated with lower replacement costs, higher average milk production, and fewer replacement heifers needed. Longevity data are obtained late in life, and for this reason stayability, defined as the probability of survival from birth until a certain age, can be used as an alternative measure. The objective of this study was to evaluate the effects of different type traits, inbreeding, and production level on the stayability of Jersey cows to various ages, and to assess trends over time. Data consisted of 460,172 to 204,658 stayability records, depending on length of the opportunity period, for survival from birth until 36, 48, 60, 72, or 84 mo of age. Threshold models were used to analyze the stayability traits, including different type traits, inbreeding coefficient, and within-herd production level as explanatory variables. Heritability estimates for stayability traits ranged from 0.05 (36 mo) to 0.22 (84 mo). As expected, the probability of survival decreased as age increased. Highly productive cows were more likely to survive than their poor-producing contemporaries regardless of age and the type trait evaluated. Our data indicate that farmers' selection decisions tend to punish poor production at early ages and reward high production at later stages. Inbreeding negatively affected the probability of survival, especially when inbreeding coefficients exceeded 10%, and this impact was most noticeable at 48 mo of age or later. Some type traits, such as stature and foot angle, had little effect on the probability of survival. Other type traits, such as strength, dairy form, rump width, and rear legs, showed higher probability of survival at intermediate scores, whereas other type traits, such as fore udder attachment, rear udder height, udder depth, and final score, showed higher probability of survival at higher scores. Finally, our results indicate that the probability of survival has decreased in the last decade, probably due to a greater number of heifers available and, therefore, higher culling rates.

Machine learning approaches for the prediction of lameness in dairy cows

Lameness is one of the costliest health problems, as well as a welfare concern in dairy cows. However, it is difficult to detect cows with possible lameness, or the ones that are at risk of becoming lame e.g. in the next week or so. In this study, we investigated the ability of three machine learning algorithms, Naïve Bayes (NB), Random Forest (RF) and Multilayer Perceptron (MLP), to predict cases of lameness using milk production and conformation traits. The performance of these algorithms was compared with logistic regression (LR) as the gold standard approach for binary classification. We had a total of 2 535 lameness scores (2 248 sound and 287 unsound) and 29 predictor features from nine dairy herds in Australia to predict lameness incidence. Training was done on 80% of the data within each herd with the remainder used as validation set. Our results indicated that in terms of area under curve of receiver operating characteristics, there were negligible differences between LR (0.67) and NB (0.66) while MLP (0.62) and RF (0.61) underperformed compared to the other two methods. However, the F1-score in NB (27%) outperformed LR (1%), suggesting that NB could potentially be a more reliable method for the prediction of lameness in practice, given enough relevant data are available for proper training, which was a limitation in this study. Considering the small size of our dataset, lack of information about environmental conditions prior to the incidence of lameness, management practices, short time gap between production records and lameness scoring, and farm information, this study proved the concept of using machine learning predictive models to predict the incidence of lameness a priori to its occurrence and thus may become a valuable decision support system for better lameness management in precision dairy farming.