A time series analysis of milk productivity in US dairy states.

As US dairy cow production evolves, it is important to characterize trends and seasonal patterns to project amounts and fluctuations in milk and milk components by states or regions. Hence, this study aimed to (1) quantify historical trends and seasonal patterns of milk and milk components production associated with calving date by parities and states; (2) classify parities and states with similar trends and seasonal patterns into clusters; and (3) summarize the general pattern for each cluster for further application in simulation models. Our data set contained 9.18 million lactation records from 5.61 million Holstein cows distributed in 17 states during the period January 2006 to December 2016. Each record included a cow's total milk, fat, and protein yield during a lactation. We used time series decomposition to obtain each state's annual trend and seasonal pattern in milk productivity for each parity. Then, we classified states and parities with agglomerative hierarchical clustering into groups according to 2 methods: (1) dynamic time warping on the original time series and (2) Euclidean distance on extracted features of trend and seasonality from the decomposition. Results showed distinguishable trends and seasonality for all states and lactation numbers for all response variables. The clusters and cluster centroid pattern showed a general upward trend for all yields [energy-corrected milk (ECM), milk, fat, and protein] and a steady trend for fat and protein percent for all states except Texas. We also found a larger seasonality amplitude for all yields (ECM, milk, fat, and protein) from higher lactation numbers and a similar amplitude for fat and protein percent across lactation numbers. The results could be used for advising management decisions according to farm productivity goals. Furthermore, the trend and seasonality patterns could be used to adjust the production level in a specific state, year, and season for farm simulations to accurately project milk and milk components production.

Multivariable time series classification for clinical mastitis detection and prediction in automated milking systems.

In this study, we developed a machine learning framework to detect clinical mastitis (CM) at the current milking (i.e., the same milking) and predict CM at the next milking (i.e., one milking before CM occurrence) at the quarter level. Time series quarter-level milking data were extracted from an automated milking system (AMS). For both CM detection and prediction, the best classification performance was obtained from the decision tree-based ensemble models. Moreover, applying models on a data set containing data from the current milking and past 9 milkings before the current milking showed the best accuracy for detecting CM; modeling with a data set containing data from the current milking and past 7 milkings before the current milking yielded the best results for predicting CM. The models combined with oversampling methods resulted in specificity of 95 and 93% for CM detection and prediction, respectively, with the same sensitivity (82%) for both scenarios; when lowering specificity to 80 to 83%, undersampling techniques facilitated models to increase sensitivity to 95%. We propose a feasible machine learning framework to identify CM in a timely manner using imbalanced data from an AMS, which could provide useful information for farmers to manage the negative effects of CM.

A stochastic animal life cycle simulation model for a whole dairy farm system model: Assessing the value of combined heifer and lactating dairy cow reproductive management programs.

This analysis introduces a stochastic herd simulation model and evaluates the estimated reproductive and economic performance of combinations of reproductive management programs for both heifers and lactating cows. The model simulates the growth, reproductive performance, production, and culling for individual animals and integrates individual animal outcomes to represent herd dynamics daily. The model has an extensible structure, allowing for future modification and expansion, and has been integrated into the Ruminant Farm Systems model, a holistic dairy farm simulation model. The herd simulation model was used to compare outcomes of 10 reproductive management scenarios based on common practices on US farms with combinations of estrous detection (ED) and artificial insemination (AI), synchronized estrous detection (synch-ED) and AI, timed AI (TAI, 5-d CIDR-Synch) programs for heifers; and ED, a combination of ED and TAI (ED-TAI, Presynch-Ovsynch), and TAI (Double-Ovsynch) with or without ED during the reinsemination period for lactating cows. The simulation was run for a 1,000-cow (milking and dry) herd for 7 yr, and we used the outcomes from the final year to evaluate results. The model accounted for incomes from milk, sold calves, and culled heifers and cows, as well as costs from breeding, AI, semen, pregnancy diagnosis, and calf, heifer, and cow feed. We found that the interaction between heifer and lactating dairy cow reproductive management programs influences herd economic performance primarily due to heifer rearing costs and replacement heifer supply. The greatest net return (NR) was achieved when combining heifer TAI and cow TAI without ED during the reinsemination period, whereas the lowest NR was obtained when combining heifer synch-ED with cow ED.

Investigating the effect of temporal, geographic, and management factors on US Holstein lactation curve parameters.

We fit the Wood's lactation model to an extensive database of test-day milk production records of US Holstein cows to obtain lactation-specific parameter estimates and investigated the effects of temporal, spatial, and management factors on lactation curve parameters and 305-d milk yield. Our approach included 2 steps as follows: (1) individual animal-parity parameter estimation with nonlinear least-squares optimization of the Wood's lactation curve parameters, and (2) mixed-effects model analysis of 8,595,413 sets of parameter estimates from individual lactation curves. Further, we conducted an analysis that included all parities and a separate analysis for first lactation heifers. Results showed that parity had the most significant effect on the scale (parameter a), the rate of decay (parameter c), and the 305-d milk yield. The month of calving had the largest effect on the rate of increase (parameter b) for models fit with data from all lactations. The calving month had the most significant effect on all lactation curve parameters for first lactation models. However, age at first calving, year, and milking frequency accounted for a higher proportion of the variance than month for first lactation 305-d milk yield. All parameter estimates and 305-d milk yield increased as parity increased; parameter a and 305-d milk yield rose, and parameters b and c decreased as year and milking frequency increased. Calving month estimates parameters a, b, c, and 305-d milk yield were the lowest values for September, May, June, and July, respectively. The results also indicated the random effects of herd and cow improved model fit. Lactation curve parameter estimates from the mixed-model analysis of individual lactation curve fits describe well US Holstein lactation curves according to temporal, spatial, and management factors.

The application of nonlinear programming on ration formulation for dairy cattle.

The objective of this study was to compare the application of iterative linear programming (iteLP), sequential quadratic programming (SQP), and mixed-integer nonlinear programming-based deterministic global optimization (MINLP_DGO) on ration formulation for dairy cattle based on Nutrient Requirements of Dairy Cattle (NRC, 2001). Least-cost diets were formulated for lactating cows, dry cows, and heifers. Nutrient requirements including energy, protein, and minerals, along with other limitations on dry matter intake, neutral detergent fiber, and fat were considered as constraints. Five hundred simulations were conducted, with each simulation randomly selecting 3 roughages and 5 concentrates from the feed table in NRC (2001) as the feed resource for each of 3 animal groups. Among the 500 simulations for lactating cows, 57, 45, and 21 simulations did not yield a feasible solution when using iteLP, SQP, and MINLP_DGO, respectively. All the simulations for dry cows and heifers were feasible when using SQP and MINLP_DGO, but 49 and 11 infeasible simulations occurred when using iteLP for dry cows and heifers, respectively. The average ration costs per animal per day of the feasible solutions obtained by iteLP, SQP, and MINLP_DGO were $4.78 (±0.71), $4.45 (±0.65), and $4.44 (±0.65) for lactating cows; $2.39 (±0.52), $1.48 (±0.26), and $1.48 (±0.26) for dry cows; and $0.98 (±0.72), $0.97 (±0.15), and $0.91 (±0.14) for heifers, respectively. The average computation time of iteLP, SQP, and MINLP_DGO were 0.59 (±1.87) s, 1.15 (±0.62) s, and 58.69 (±68.45) s for lactating cows; 0.041 (±0.070) s, 0.76 (±0.37) s, and 14.84 (±39.09) s for dry cows; and 1.60 (±2.90) s, 0.51 (±0.19) s, and 16.45 (±45.56) s for heifers, respectively. In conclusion, iteLP had limited capability of formulating least-cost diets when nonlinearity existed in the constraints. Both SQP and MINLP_DGO handled the nonlinear constraints well, with SQP being faster, whereas MINLP_DGO was able to return a feasible solution under some situations where SQP could not.

Short communication: Milk urea nitrogen as a predictor of urinary nitrogen and urea nitrogen excretions of late-lactation dairy cows fed nitrogen-limiting diets.

Our objectives were to assess the relationships between milk urea N (MUN), serum urea N (SUN), urine N (UN), and urinary urea N (UUN) in late-lactation cows fed N-limiting diets and compare these relationships with those previously established. Data were from a pen-based study in which 128 Holstein cows had been assigned to 1 of 16 pens in a randomized complete block design to assess the effects of diets containing 16.2, 14.4, 13.1, and 11.8% crude protein (CP, dry matter basis) during a 12-wk period. At least half of the cows in each pen were randomly selected to collect pen-level samples of serum and urine in wk 3, 7, and 11, when wk in lactation averaged 35, 39, and 43, respectively. A mixed model was developed to study the relationship of MUN with SUN, UN, and UUN. Week of lactation did not affect the relation between MUN and SUN across dietary treatments. However, we found a week × MUN interaction, suggesting that between wk 35 and 43 of lactation, UN excretion decreased from 89 to 73 g/d (-17 g/d) when MUN was 6.0 mg/dL (11.8% dietary CP) but increased from 142 to 149 g/d (+7 g/d) when MUN was 13.3 mg/dL (16.2% dietary CP). These effects were essentially due to changes in UUN excretion, which declined from 54 to 37 g/d (-17 g/d) and increased from 112 to 117 g/d (+5 g/d) when MUN was 6.0 and 13.3 mg/dL, respectively. When MUN was 11.2 mg/dL (15% dietary CP), UN and UUN excretions remained constant over time. Based on root mean squared prediction error and the concordance correlation coefficient, these data did not conform to most previously published prediction equations because of both mean and slope biases. The discrepancy could have resulted from difference in study design (cow vs. pen as experimental unit), dietary treatments (energy vs. N-limiting diets), frequency of measurement and duration of adaptation period (single measurement after 1 to 3 wk of adaptation vs. repeated measurements over a 12-wk period), method for determining urine volume (total collection vs. spot sampling), and the assay used to measure MUN. However, our data captured changes in kidney physiology that warrant further studies of long-term renal adaptation to N-limiting diets.

Short communication: Identifying challenges and opportunities for improved nutrient management through the USDA's Dairy Agroecosystem Working Group.

Nutrient management on US dairy farms must balance an array of priorities, some of which conflict. To illustrate nutrient management challenges and opportunities across the US dairy industry, the USDA Agricultural Research Service Dairy Agroecosystems Working Group (DAWG) modeled 8 confinement and 2 grazing operations in the 7 largest US dairy-producing states using the Integrated Farm System Model (IFSM). Opportunities existed across all of the dairies studied to increase on-farm feed production and lower purchased feed bills, most notably on large dairies (>1,000 cows) with the highest herd densities. Purchased feed accounted for 18 to 44% of large dairies' total operating costs compared with 7 to 14% on small dairies (<300 milk cows) due to lower stocking rates. For dairies with larger land bases, in addition to a reduction in environmental impact, financial incentives exist to promote prudent nutrient management practices by substituting manure nutrients or legume nutrients for purchased fertilizers. Environmental priorities varied regionally and were principally tied to facility management for dry-lot dairies of the semi-arid western United States (ammonia-N emissions), to manure handling and application for humid midwestern and eastern US dairies (nitrate-N leaching and P runoff), and pasture management for dairies with significant grazing components (nitrous oxide emissions). Many of the nutrient management challenges identified by DAWG are beyond slight modifications in management and require coordinated solutions to ensure an environmentally and economically sustainable US dairy industry.

Assessment of the temperature cut-off point by a commercial intravaginal device to predict parturition in Piedmontese beef cows.

Dystocic parturitions have an adverse impact on animal productivity and therefore the profitability of the farm. In this regard, accurate prediction of calving is essential since it allows for efficient and prompt assistance of the dam and the calf. Numerous approaches to predict parturition have been studied, among these, measurement of intravaginal temperature (IVT) is the most effective method at the field level. Thus, objectives of this experiment were, 1) to find an IVT cut-off to predict calving within 24 h, and 2) to clarify the use of IVT as an automated method of calving detection in housed beef cows. A commercial intravaginal electronic device (Medria Vel'Phone®) with a sensor that measures the IVT every 12 h was used. Piedmontese cows (n = 211; 27 primiparous and 184 multiparous) were included in this study. One-way analysis of variance was used to assess the temperature differences at 0, 12, 24, 36, 48 and 60 h before parturition. Receiving operator characteristic curves were built to determine the temperature cut-off which predicts calving within 24 h with the highest summation of sensitivity (Se) and specificity (Sp). Binomial logistic regression models were computed to identify factors that may affect the IVT before calving. Mean gestation length was 291.5 ±â€¯13.7 d (primiparous, 292 ±â€¯14.1 d; multiparous, 289 ±â€¯9.2 d). A decrease (P < 0.001) in the average IVT was found from 60 h before calving until the expulsion of the IVT device. A significant (P < 0.05) reduction in the IVT was noticeable from 24 h before until parturition. The IVT drop to predict parturition 24 h before calving was 0.21 °C (area under the curve [AUC] = 0.72; Se = 66%, Sp = 76%). Furthermore, the IVT cut-off value to predict parturition within 24 h was 38.2 °C (AUC = 0.89; Se = 86%, Sp = 91%). None of the evaluated fixed effects (parity, dystocia, season or length of gestation) affected (P ˃ 0.05) the IVT variation from 60 h before and up to calving. To conclude, the IVT average seems to be a better parameter than the drop in temperature to predict parturition within 24 h. In this regard, a cut-off of 38.2 °C showed a high Se and Sp for predicting calving. This study demonstrates the usefulness of a commercially available device to predict calving to improve management in stabled beef farms.

Estimating the energetic cost of feeding excess dietary nitrogen to dairy cows.

Feeding N in excess of requirement could require the use of additional energy to metabolize excess protein, and to synthesize and excrete urea; however, the amount and fate of this energy is unknown. Little progress has been made on this topic in recent decades, so an extension of work published in 1970 was conducted to quantify the effect of excess N on ruminant energetics. In part 1 of this study, the results of previous work were replicated using a simple linear regression to estimate the effect of excess N on energy balance. In part 2, mixed model methodology and a larger data set were used to improve upon the previously reported linear regression methods. In part 3, heat production, retained energy, and milk energy replaced the composite energy balance variable previously proposed as the dependent variable to narrow the effect of excess N. In addition, rumen degradable and undegradable protein intakes were estimated using table values and included as covariates in part 3. Excess N had opposite and approximately equal effects on heat production (+4.1 to +7.6 kcal/g of excess N) and retained energy (-4.2 to -6.6 kcal/g of excess N) but had a larger negative effect on milk gross energy (-52 to -68 kcal/g of excess N). The results suggest that feeding excess N increases heat production, but more investigation is required to determine why excess N has such a large effect on milk gross energy production.

Short communication: Evaluation of nitrogen excretion equations from cattle.

Nitrogen excretion in dairy manure is a precursor for N2O and NH3 formation in livestock housing, manure storage facilities, and after manure is applied to land. Nitrous oxide is a major contributor to greenhouse gas emissions, and reducing N output from dairy production facilities can reduce the amount of anthropogenic N2O entering the atmosphere. The objective of the study was to conduct a comprehensive evaluation of extant prediction models for N excretion in feces and urine using extensive literature data. A total of 45 N excretion equations were evaluated for lactating cows, heifers, and nonlactating cows and steers. These equations were evaluated with 215 treatment means from 69 published studies collected over 20 yr from 1995 to 2015. Two evaluation methods were used: the root mean square prediction error and the concordance correlation coefficient. Equations constructed using a more rigorous development process fared better than older extant equations. Equations for heifers and nonlactating cows had greater error of prediction compared with equations used for lactating cows. This could be due to limited amount of data available for construction and evaluation of the equations. Urinary N equations had greater prediction errors than other forms of excretion, possibly due to high variability in urinary N excretion and challenges in urine collection. Fecal N equations had low error bias and reached an acceptable level of precision and accuracy.

Technical note: Bayesian calibration of dynamic ruminant nutrition models.

Mechanistic models of ruminant digestion and metabolism have advanced our understanding of the processes underlying ruminant animal physiology. Deterministic modeling practices ignore the inherent variation within and among individual animals and thus have no way to assess how sources of error influence model outputs. We introduce Bayesian calibration of mathematical models to address the need for robust mechanistic modeling tools that can accommodate error analysis by remaining within the bounds of data-based parameter estimation. For the purpose of prediction, the Bayesian approach generates a posterior predictive distribution that represents the current estimate of the value of the response variable, taking into account both the uncertainty about the parameters and model residual variability. Predictions are expressed as probability distributions, thereby conveying significantly more information than point estimates in regard to uncertainty. Our study illustrates some of the technical advantages of Bayesian calibration and discusses the future perspectives in the context of animal nutrition modeling.

Predicting nitrogen excretion from cattle.

Manure nitrogen (N) from cattle production facilities can lead to negative environmental effects, such as contribution to greenhouse gas emissions, leaching and runoff to aqueous ecosystems leading to eutrophication, and acid rain. To mitigate these effects and to improve the efficiency of N use, accurate prediction of N excretion and secretions are required. A genetic algorithm was implemented to select models to predict fecal, urinary, and total manure N excretions, and milk N secretions from 3 classes of animals: lactating dairy cows, heifers and dry cows, and steers. Two tiers of model classes were developed for each category of animals based on model input requirements. A total of 6 models for heifers and dry cows and steers and an additional 2 models for lactating dairy cattle were developed. Evaluation of the models using K-fold cross validation based on all data and using the most recent 6 yr of data showed better prediction for total manure N and fecal N compared with urinary N excretion, which was the most variable response in the database. Compared with extant models from the literature, the models developed in this study resulted in a significant improvement in prediction error for fecal and urinary N excretions from lactating cows. For total manure production by lactating cows, extant and new models were comparable in their prediction ability. Both proposed and extant models performed better than the prediction methods used by the US Environmental Protection Agency for the national inventory of greenhouse gases. Therefore, the proposed models are recommended for use in estimation of manure N from various classes of animals.

Evidence of dehydration and electrolyte disturbances in cases of perennial ryegrass toxicosis in Australian sheep.

CASE REPORT: Perennial ryegrass toxicosis (PRGT) is a common disease entity in Australia, presenting as an association of clinical signs including alterations in normal behavioural, ataxia ('staggers'), ill thrift and gastrointestinal dysfunction ('scours'). Clinical signs can range in severity from mild (gait abnormalities and failure to thrive) to severe (seizures, lateral recumbency and death). Presentation across the flock is usually highly variable. PRGT is caused by toxins produced by the endophytic fungus Neotyphodium lolii, a symbiont of perennial ryegrass that is present in pastures across the temperate regions of Australia and Tasmania. A particular feature of PRGT in Australia is the occasional occurrence of large-scale sheep losses, suggesting other factors are influencing mortality rates compared with other PRGT risk zones such as North America and New Zealand. During 2011, producers in the state of Victoria experienced a mild outbreak of PRGT that affected large numbers of animals but with limited mortalities. Clinical samples taken from affected sheep showed a high incidence of dehydration and electrolyte abnormalities. CONCLUSION: We speculate that changes in hydration status may be a contributory aetiological factor in those years in which high numbers of deaths are associated with PRGT outbreaks in Australia.

Dicoumarol toxicity in cattle associated with ingestion of silage containing sweet vernal grass (Anthoxanthum odoratum).

A diagnosis of dicoumarol toxicity in a herd of Friesian cattle was made following investigation of the deaths of three mature cows and eleven yearling heifers. Affected stock had been fed wrapped, bailed silage containing approximately 90% sweet vernal grass (Anthoxanthum odoratum). Sweet vernal grass contains coumarin, which can be converted to dicoumarol, a vitamin K antagonist, through the action of moulds. Most deaths were preceded by lethargy, severe anaemia and subcutaneous and internal haemorrhage. Dicoumarol toxicosis was suspected based on clinical signs, necropsy findings and prolonged prothrombin and activated partial thromboplastin times. Dicoumarol analysis of blood from affected animals and silage confirmed the diagnosis. Activated partial thromboplastin time Haemoglobin Packed cell volume Prothrombin time Red cell count