The effects of heat stress on dairy production can be separated into 2 distinct causes: those effects that are mediated by the reduced voluntary feed intake associated with heat stress, and the direct physiological and metabolic effects of heat stress. To distinguish between these, and identify their effect on milk protein and casein concentration, mid-lactation Holstein-Friesian cows (n = 24) were housed in temperature-controlled chambers and either subjected to heat stress [HS; temperature-humidity index (THI) ~78] or kept in a THI<70 environment and pair-fed with heat-stressed cows (TN-R) for 7 d. A control group of cows was kept in a THI<70 environment with ad libitum feeding (TN-AL). A subsequent recovery period (7 d), with THI<70 and ad libitum feeding followed. Intake accounted for only part of the effects of heat stress. Heat stress reduced the milk protein concentration, casein number, and casein concentration and increased the urea concentration in milk beyond the effects of restriction of intake. Under HS, the proportion in total casein of αS1-casein increased and the proportion of αS2-casein decreased. Because no effect of HS on milk fat or lactose concentration was found, these effects appeared to be the result of specific downregulation of mammary protein synthesis, and not a general reduction in mammary activity. No residual effects were found of HS or TN-R on milk production or composition after THI<70 and ad libitum intake were restored. Heat-stressed cows had elevated blood concentrations of urea and Ca, compared with TN-R and TN-AL. Cows in TN-R had higher serum nonesterified fatty acid concentrations than cows in HS. It was proposed that HS and TN-R cows may mobilize different tissues as endogenous sources of energy.

Livestock welfare and management could be greatly enhanced by the replacement of branding or ear tagging with less invasive visual biometric identification methods. Biometric identification of cattle from muzzle patterns has previously indicated promising results. Significant barriers exist in the translation of these initial findings into a practical precision livestock monitoring system, which can be deployed at scale for large herds. The objective of this study was to investigate and address key limitations to the autonomous biometric identification of cattle. The contributions of this work are fourfold: (1) provision of a large publicly-available dataset of cattle face images (300 individual cattle) to facilitate further research in this field, (2) development of a two-stage YOLOv3-ResNet50 algorithm that first detects and extracts the cattle muzzle region in images and then applies deep transfer learning for biometric identification, (3) evaluation of model performance across a range of cattle breeds, and (4) utilizing few-shot learning (five images per individual) to greatly reduce both the data collection requirements and duration of model training. Results indicated excellent model performance. Muzzle detection accuracy was 99.13% (1024 × 1024 image resolution) and biometric identification achieved 99.11% testing accuracy. Overall, the two-stage YOLOv3-ResNet50 algorithm proposed has substantial potential to form the foundation of a highly accurate automated cattle biometric identification system, which is applicable in livestock farming systems. The obtained results indicate that utilizing livestock biometric monitoring in an advanced manner for resource management at multiple scales of production is possible for future agriculture decision support systems, including providing useful information to forecast acceptable stocking rates of pastures.

To understand the animal welfare impact of virtual fencing stimuli (audio cue ‘beep’ and electrical stimulus) on naïve sheep, it is necessary to assess stress responses during the animal’s first encounters with these stimuli. Eighty Merino ewes were exposed to one of the following treatments (n = 16 animals per treatment): Control (no stimuli), beep, dog bark, manual restraint, and electrical stimulus. Collars were used to apply the audio and electrical stimuli. The restraint treatment showed an elevated cortisol response compared with the control (p < 0.05), but there were no differences between the other treatments and the control. There were no differences between treatments in vaginal temperature (p > 0.05). For behaviors, the sheep receiving the bark and beep treatments were more vigilant compared to the control (p < 0.05), there were more aversive responses observed in the electrical stimulus treatment compared to the control. Together, the responses showed that the beep stimuli were largely benign, the bark stimuli was minimally aversive, the electrical stimuli was acutely aversive, and the restraint was moderately aversive. These data suggest that, for sheep, their first exposure to the virtual fencing stimuli should be perceived as less aversive than a commonly used restraint procedure.

Automated weighing systems to monitor BW and supplement intake (SI) of individual grazing cattle are being developed to better understand the seasonal nutrition and performance of grazing livestock. This study established (1) the accuracy and repeatability of a commercial walk-over weighing (WoW) system for estimating BW and (2) the accuracy of an automatic supplement weighing (ASW) unit for estimating SI based on measuring time spent at the unit. The WoW and ASW units monitored BW and SI of 112 cattle consisting of 55 cows and 57 calves grazed on a 32.5 ha paddock for 41 days, with an average of 258 BW records collected per day. Static BWs were recorded at each mustering event (n = 7) and were compared to repeated measurements collected by the WoW on the day of each mustering event. Body weight was overestimated by the WoW, with the predicted BW of calves and cows averaging 10 and 21 kg heavier, respectively, than actual, and root MS prediction errors (RMSPE) of 5.1% and 5.5% of the static BW, respectively. For both calves and cows, 38% of the MS prediction errors (MSPE) was mean bias (MB) error and 9% of MSPE was slope bias error. The concordance correlation coefficient (CCC; 0.90 v. 0.80) and modelling efficiency (MEF; 0.78 v. 0.62) of WoW BW for calves were higher than for cows, indicating that the predicted values were deviating from a 1 : 1 relationship and in particular as weight increases. A rolling average across five or more consecutive BW measures improved the accuracy of the WoW BW estimates. Regarding estimates of SI, the aggregated time the herd spent at the ASW unit was strongly associated with total SI (R 2 = 0.92; P < 0.001). Further, positive linear relationships (P < 0.001) existed between cumulative weighted time spent at the ASW unit (min) and concentration of fenbendazole (FBZ) used as an intake marker and its derivatives (oxfendazole and oxfendazole sulfone) in the plasma of individual cows, with R 2 of 0.54, 0.73 and 0.75, respectively. Although the WoW overestimated static BW, the low bias in the slope indicated that a linear regression model could be developed to adjust the WoW BW to reduce the MB and improve the estimate of WoW BW. The significant positive relationship between time spent at the ASW unit and individual blood FBZ concentration identified the suitability of the ASW unit for estimating SI by grazing cattle.

Identifying the licking behaviour in beef cattle may provide a means to measure time spent licking for estimating individual block supplement intake. This study aimed to determine the effectiveness of tri-axial accelerometers deployed in a neck-collar and an ear-tag, to characterise the licking behaviour of beef cattle in individual pens. Four, 2-year-old Angus steers weighing 368 ± 9.3 kg (mean ± SD) were used in a 14-day study. Four machine learning (ML) algorithms (decision trees [DT], random forest [RF], support vector machine [SVM] and k-nearest neighbour [kNN]) were employed to develop behaviour classification models using three different ethograms: (1) licking vs. eating vs. standing vs. lying; (2) licking vs. eating vs. inactive; and (3) licking vs. non-licking. Activities were video-recorded from 1000 to 1600 h daily when access to supplement was provided. The RF algorithm exhibited a superior performance in all ethograms across the two deployment modes with an overall accuracy ranging from 88% to 98%. The neck-collar accelerometers had a better performance than the ear-tag accelerometers across all ethograms with sensitivity and positive predictive value (PPV) ranging from 95% to 99% and 91% to 96%, respectively. Overall, the tri-axial accelerometer was capable of identifying licking behaviour of beef cattle in a controlled environment. Further research is required to test the model under actual grazing conditions.

Formulating diets conservatively for minimum crude-protein (CP) requirements and overfeeding nitrogen (N) is commonplace in grain finishing rations in USA, Canada and Australia. Overfeeding N is considered to be a low-cost and low-risk (to cattle production and health) strategy and is becoming more commonplace in the US with the use of high-N ethanol by-products in finishing diets. However, loss of N from feedlot manure in the form of volatilised ammonia and nitrous oxide, and nitrate contamination of water are of significant environmental concern. Thus, there is a need to improve N-use efficiency of beef cattle production and reduce losses of N to the environment. The most effective approach is to lower N intake of animals through precision feeding, and the application of the metabolisable protein system, including its recent updates to estimation of N supply and recycling. Precision feeding of protein needs to account for variations in the production system, e.g. grain type, liveweight, maturity, use of hormonal growth promotants and ß agonists. Opportunities to reduce total N fed to finishing cattle include oscillating supply of dietary CP and reducing supply of CP to better meet cattle requirements (phase feeding).