In recent years, machine learning (ML) algorithms have emerged as powerful tools for predicting and modeling complex data. Therefore, the aim of this study was to evaluate the prediction ability of different ML algorithms and a traditional empirical model to estimate the parameters of lactation curves. A total of 1186 monthly records from 156 sheep lactations were used. The model development process involved training and testing models using ML algorithms. In addition to these algorithms, lactation curves were also fitted using the Wood model. The goodness of fit was assessed using correlation coefficient (r), mean absolute error (MAE), root mean square error (RMSE), relative absolute error (RAE), and relative root mean square error (RRSE). SMOreg was the algorithm with the best estimates of the characteristics of the sheep lactation curve, with higher values of r compared to the Wood model (0.96 vs. 0.68) for the total milk yield. The results of the current study showed that ML algorithms are able to adequately predict the characteristics of the lactation curve, using a relatively small number of input data. Some ML algorithms provide an interpretable architecture, which is useful for decision-making at the farm level to maximize the use of available information.

Energy-demanding processes in animal production rise in tandem with increased nutritional intake. This meta-analysis aimed to quantify the metabolizable energy (ME) requirements for maintenance (kMEm) and metabolizable energy to support metabolism (HiEv), considering genotype and gender, in beef cattle under tropical conditions. A total of 273 individual records of comparative slaughter from seven studies (theses and dissertations) grouped by animal gender (intact males and castrated males), and genotype (Nellore or crossbreed) were used in the current study. We estimated kMEm and HiEv based on the independent variables’ ME intake and full body weight (FBW). The meta-analysis showed no effect of genotype on kMEm (p = 0.66). However, intact males exhibited a 34.41 % lower kMEm compared to castrated males (p = 0.01). Crossbred steers had a 61.33 % higher HiEv than Nellore steers (p < 0.001). Similarly, intact males had a 41.07 % higher energy support demand (p = 0.05) than castrated males. In addition, we observed positive correlations (medium to high) between body protein (r = 0.67; p > 0.001), body fat (r = 0.59; p > 0.001), and retained energy (r = 0.71; p > 0.001) and HiEv. Thus, the greater the deposition of protein (mainly), fat, and body energy, the greater the HiEv. In conclusion, the genotype of the animals did not affect the kMEm. Intact males had lower HiEv than castrated males. Both Nellore steers and intact males utilized HiEv more.

The ability of mathematical models to represent the lactation process varies according with their mathematical structure and database characteristics. The aim of the current study was evaluated the goodness of fit of empirical and mechanistic models applied to dairy sheep lactation curves with different shapes. A total of 4,494 weekly test day records were analyzed. All lactations were individually fitted using two empirical (Wood and Wilmink) and two mechanistic (Dijkstra, and Pollott) models. The Dijkstra model showed the best performance to typical curves and Wood model to atypical curves (without peak lactation). Therefore, the selection of the mathematical model to fit sheep lactation curves must consider the specific patter of milk production.

The primary concern in corn silage is aerobic deterioration. This study aimed to evaluate the impact of essential oil (EO) blends extracted from copaiba (Copaifera langsdorffii) (Co) and Eucalyptus staigeriana Branch (ES) on the fermentation parameters and losses, nutritional value, microbial counts, and aerobic stability of corn silage. A completely randomized design was employed with four treatments and four replicates. The following treatments were employed: control (CON) – corn ensiled without EO; blend 50:50 (50 % ES plus 50 % Co); blend 75:25 (75 % ES plus 25 % Co); blend 25:75 (25 % ES plus 75 % Co). The dose was administered at a rate of 500 mg kg^–1 of ensiled mass (as-fed basis). The chemical composition, in vitro degradability, NH₃-N, pH, acetic, propionic, and butyric acids, aerobic stability, and microbial counts of silages were evaluated. The 75:25 blend reduced (p < 0.038) crude protein (CP) by 7.96 % compared to the 25:75 blend. However, the 50:50 blend reduced (p < 0.03) the hemicellulose (HEM) content of the silage in comparison to the other treatments. The blends were observed to increase (p = 0.046) the concentration of acetic acid. The propionic acid concentration was reduced by 88.13 % with the 25:75 blend compared to the 50:50 blend. The blends increased (p = 0.003) gas losses. The fungal population was drastically reduced (p = 0.048) by the 25:75 blend compared to the other treatments. It can be concluded that the 50:50 blend resulted in high concentrations of acetic and propionic acids, thereby improving the aerobic stability of corn silage.

Recently Latin American countries have developed a dairy sheep industry with an increasing number of specialized dairy-sheep flocks. The objective of this study was to evaluate the goodness of fit of empirical and mechanistic models and sampling interval to model the lactation curve of F1 dairy sheep with different shape of lactation curves in an intensively managed flock of the central highlands of Mexico. A total of 4,494 weekly (7D) test day records (TDR) from 156 lactations were analyzed. Three datasets were generated from the original 7D data set, setting TDR at different sampling intervals: 14 (14D), 21 (21D) and 28 (28D) days. Lactation curves were fitted using two empirical (Wood and Wilmink) and two mechanistic models (Dijkstra and Pollott). The parameters of the empirical and mechanistic models were estimated using the iterative non-linear curve fitting procedure in R. The results showed that the estimation of TMY was not affected by the sampling interval. However, the estimation of peak yield (PY) and day at peak yield (TPY) was affected by sampling interval (P < 0.05), with better estimates for 7D and 14D. Estimates of PY and TPY differed between the empirical and mechanistic models with both methods failing to estimate PY and TPY in atypical curves. The Dijkstra model showed the best fit for typical curves and the Pollott model for atypical curves in all the sampling intervals evaluated.