The objectives of these studies were to assess U.S. dairy nutritionists' approaches toward balancing CP in lactating cow diets, and to leverage existing knowledge of postabsorptive AA metabolism through the application of a mathematical ration-balancing model to predict N efficiency through a more accurate postabsorptive amino acid (AA) delivery. In experiment 1, dairy nutritionists (n = 77) that fed a total of 521,000 lactating dairy cows responded to a questionnaire related to demographic information, feelings toward environmental nitrogen (N) excretion, and dietary CP balancing practices. Eighty-nine percent of nutritionists balanced diets based on one or more individual AA requirements of dairy cows. The primary concern with formulation of lower CP diets was the cost per unit of metabolizable protein (MP). In the second study, three treatments were fed to lactating Holstein cows (n = 48) to test proof of concept of NASEM 2021 and a nonlinear optimizer: a control balanced to fulfill all nutrient needs of lactating dairy cows producing 45 kg milk/d using the NRC (2001) dairy model (NRC01), and two diets balanced with a nonlinear optimizer that fulfilled requirements according to the updated NASEM (2021) dairy model and attempted to either: 1) maximize N efficiency through tailored post-ruminal AA supply (NEFF), or 2) maximize income over feed cost (IOFC). A simulation function was written in RStudio (version 2022.02.3) to predict daily animal performance with NASEM 2021. Dry matter intake, milk, and milk components from both the observed data and the simulation data matrices were analyzed as repeated measures (days) in a mixed model to test for both observed and predicted (simulated) differences in treatment means. Income over feed cost was $4.83, $4.77, and $5.12/cow/d for NRC01, IOFC, and NEFF, respectively (P = 0.96). Nitrogen efficiency (%) was greatest for NEFF (33.7), followed by NRC01 (28.9) and IOFC (23.4; P < 0.05 between all treatments). Based upon simulation data, NASEM 2021 predicted relative performance differences between animals that consumed treatments with differing absorbed EAA supplies, although residual analyses revealed that further progress could be made in milk protein (g/d), milk fat (g/d), milk yield (kg/d), and DMI (kg/d) predictions. / Doctor of Philosophy / Nitrogen (N) is considered one of the major environmental pollutants of the dairy industry. A concerted focus on reducing dietary crude protein (CP) loss by targeting the specific amino acid (AA) needs of cattle will also reduce the amount of N they excrete. Because most dairy farmers hire nutritionists to formulate rations for their cows, the objective of our first study, performed in 2019, was to assess U.S. dairy nutritionists' approaches toward balancing AA in lactating cow diets. The objective of our second study was to apply a mathematical ration-balancing model that specifically optimized AA efficiency for lactating dairy cows by leveraging existing knowledge of dairy cow AA metabolism. In study 1, 89% of nutritionists balanced diets based on one or more individual AA requirements of dairy cows. Most respondents (72%) reported that they are currently formulating diets with lower dietary protein than they were 3 to 5 yr ago. In the first study, the primary concern with formulating lower protein diets was the high cost of protein sources that are highly metabolizable for cows. In study 2, income over feed cost (income made from estimated milk sales minus cost of feed) was similar for cows fed a diet with the lowest protein (relative to two other diets), but cows on the low protein diet also ate the least (kg of dry matter/d) and produced the least amount of milk protein (kg/d) and milk fat (kg/d). Still, relative to the other two diets in the study, N consumed by cows on the lowest CP diet was more efficiently converted to milk protein, which resulted in lesser N excretion in milk. Harnessing the knowledge gained from these studies could allow significant reduction of environmental N excretion at little to no cost to consumers or farmers.

Understanding the particularities of pasture inclusion on cows' diets and their subsequent impact on the digestive processes, together with possible differences in the nutrient utilization of cows managed under grazing conditions, are challenges that must be elucidated in order to design strategies to improve the cows' productive performance. The objectives of this dissertation were: 1) to review the existing literature on ration formulation for dairy cows in pasture-based systems, identifying limitations and potential improvement areas, 2) to evaluate the adequacy of the Molly model predictions of ruminal fermentation, nutrient digestion, and performance from cows consuming fresh ryegrass-based diets, identifying mechanisms that could be used to direct further model improvements, 3) to evaluate the model predictions of milk, fat and protein production using data from Holstein-Friesian and Jersey cows from Chile managed under pasture-based systems, identifying mechanisms that could be used to direct further model improvements, 4) to identify the different dairy breeds and their crosses that are used in the Chilean dairy population, determining if there are specific breed and heterosis effects on productive and some functional traits, and 5) to characterize the ruminal degradation dynamics of crude protein (CP) and individual amino acids (AA) from ryegrass (Lolium perenne) at vegetative stage, generating information to improve the ration formulation in dairy cows in pasture-based systems. From the literature revision in objective 1, different factors that could help to create a more specific classification of dairy pasture-based systems were identified. Maximizing grass inclusion seems to leads to an oversimplification of diets, potentially limiting the performance of medium and high producing cows. Despite the low N use efficiencies reached in pasture-based systems, there is a lack of information in studies that delve into the ruminal degradation of CP and AA, as well as their intestinal absorption, and subsequent post-absorptive utilization. Advances in the understanding of these areas could help to create strategies to face this problem. There is evidence suggesting that cows can perform different under different productive systems, but various breeds, strains, and crossbreds are used in pasture-based systems, being needed first a better characterization of them. To achieve the second objective, a total of 25 studies (n = 115 treatments) including dairy cows consuming ryegrass predominant diets, published from 1970 through 2020, were collected from the literature and used to assess the model accuracy and precision based on root mean squared errors and concordance correlation coefficients (CCC). Predictions of protein and fiber digestion and fiber and organic matter fecal excretion were improved after model reparameterization, while body weight and body condition score predictions were improved after model modifications and reparameterization. Although digestion of nutrient was better represented by the new set of parameters derived, the accuracies of milk, protein, and fat production remained low (CCC of 0.13, 0.12, and 018, respectively), with moderate slope bias. To achieve the third objective, a dataset including 180 group cows' diets (153 from Holstein Friesian and 27 from Jersey cows) from 18 commercial dairy farms with different inclusion of ryegrass was compiled. Chemical composition of feeds ingredients, group intakes and animal performance records, were provided by the Chilean cooperative Colun. Separately model mammary reparameterization of Holstein Friesian and Jersey cows' data, improved both milk production (CCC values of 0.69 and 0.90, respectively) and milk solids predictions (CCC values of 0.68 and 0.92 for milk protein and 0.65 and 0.80 for milk fat, for Holstein Friesian and Jersey cows, respectively). This indicates that there seems to be some differences in mammary cells and cells activity between breeds evaluated, translated into different lactation curves and milk solids synthesis patters. To achieve the fourth objective, a dataset considering 1,429,132 records from 586,624 cows that calved between 1998 and 2018 was compiled. Pedigree information, milk, milk protein, and milk fat, as well as somatic cell score and calving interval records were provided by the Chilean cooperative COOPRINSEM. The proportion of each breed according to the different country origin (strain) was calculated for all the animals, being identified eight genetic strains with enough information to perform separated genetic evaluations: Chilean Friesian, French Holstein-Friesian, US-Holstein, US-Jersey, French Montbeliarde, New Zealander Holstein-Friesian, Swedish Red and White, and British Friesian. Four different genetic models, increasing in complexity (considering breed or strain classification, crossbreeding proportion, and heterosis effects) were tested. Specific effects for some genetic strains, as well as heterosis effects between some strains with Chilean Friesian were identified, so these effects should be considered when performing genetic evaluations in the Chilean dairy population. Also, they can be used to direct future selection programs. To achieve the last objective, an in situ study using three cannulated cows was carried out at the Agricultural Research Station of the Austral University of Chile (Valdivia, Chile). Samples from a perennial ryegrass pasture were collected during winter, spring and summer at vegetative stage. Duplicate bags were ruminally incubated for 4, 8, 12, 24, and 96 h. CP and AA profile of original samples and from incubation residues were determined. Ryegrass samples presented a high crude protein content (averaging 25.8%), however, the RUP content was ~15 to 20% of CP. Significant differences were observed for ryegrass CP and individual AA ruminal degradation kinetics, this seems to be associated to some specific AA. / Doctor of Philosophy / The used of pasture-based systems in dairy production are an important alternative in countries were local environmental and climatological conditions, as well as soil properties are suitable for the growth of high-quality grasslands. The main advantages attributed to the use of pasture-based system are the reduction in the production costs, and potentially improving the nutrition quality of dairy products as well as the cows' welfare. However, they have some important limitations as the natural variation in the quantity and quality of available herbage. Also, there is a lack of information on fundamental nutrition of pasture nutrients digestion and utilization by the cows manage and selected under these environmental conditions. The first two objectives of this dissertation consisted in using the Molly nutritional model, which represents the biology of the whole cow, in order to identify key aspects to better understand the cows manage under pasture-based systems and develop strategies to precise their nutrition. Particularly, topics such as prediction of ruminal fermentation and digestive processes, mobilization of body reserves, mammary glands metabolism, and potential differences between Holstein-Friesian and Jersey cows were analyzed. Overall, work on these objectives resulted in improvements to the Molly model for making predictions of diets that include grass. The diversity of breeds and their country of origin, as well as different crosses in the Chilean dairy population were explored. A more comprehensive genetic model was proposed in this work to precise the calculation of breeding values in the country. Furthermore, our findings could be used as inputs to direct future selection programs. The last objective explored the ruminal degradation of crude protein and each individual amino acid in ryegrass. We demonstrate that indeed there are differences in the degradation parameters for some amino acids. Due to the high degradability of fresh ryegrass, there could be some potential limitations in the supply of some essential amino acids, especially in high producing cows consuming predominant ryegrass diets.

O rebanho brasileiro de bovinos leiteiros é composto predominantemente por vacas mestiças criadas em condições tropicais. Os principais modelos usados no Brasil para a formulação e avaliação de dietas são norte-americanos e baseados nas exigências de vacas Holandesas em clima temperado. A principal contribuição do primeiro capítulo deste estudo foi o desenvolvimento e avaliação de um novo modelo para estimar a ingestão de matéria seca (IMS) de vacas mestiças. A nova equação de predição da IMS foi desenvolvida e avaliada usando um banco de dados com 161 médias de tratamentos provenientes de 38 estudos brasileiros publicados com vacas mestiças [n = 446 vacas, 16,60 ± 5,70 (DP) kg leite/dia]. O modelo proposto foi: IMS (kg/dia): [0,5552 (± 0,06636 EP) × LC4%G + 0,06332 (± 0,009455) × PV0,75] × [1 - e (- 0,7732 (± 0,7019) × (SEM - 1,629 (± 1,9313))) ]. O novo modelo apresentou maior acurácia (QMEP = 1,64, r2 = 0,88) na predição da IMS quando comparado com modelos norte-americanos. O novo modelo de predição da IMS pode ser aplicado na formulação de dietas para vacas leiteiras mestiças em condições tropicais. No capítulo 2, um estudo avaliou as novas atualizações das exigências nutricionais de bezerras leiteiras sugeridas por Van Amburgh e Drackley (2005) e inseridas em programa comercial (AMTS, Agricultural Modeling and Training Systems, AMTS.Cattle.ProTM Calf Model versão 3.5.8.0, 2015). Dados de 16 estudos brasileiros envolvendo 51 dietas para bezerros leiteiros (n = 485, peso ao desmane de 62,02 ± 10,16 kg) foram usados para avaliar os modelos do NRC (2001) e o AMTS. Nas condições do conjunto de dados do presente estudo, ambos os sistemas superestimaram o ganho calculado a partir da proteína disponível. O ganho de peso estimado pela energia disponível da dieta, foi superestimado em 19 g/dia quando calculado pelo NRC e subestimado em 68 g/dia quando calculado pelo AMTS. O ganho de peso possível a partir da energia disponível foi menor quando calculado pelo AMTS em relação ao NRC (2001). Para o conjunto de dados experimentais brasileiros, os ganhos de peso de bezerros foram 32% inferiores aos descritos na literatura para sistemas norte-americanos. As razões para esta discrepância precisam ser compreendidas, pois somente assim novos modelos poderão ser desenvolvidos e parametrizados para estimar o desempenho animal de forma mais acurada e precisa. / The Brazilian dairy herd is mainly made up of crossbred cows fed under tropical conditions. The main models in use to formulate and evaluate diets in Brazil are from North America. These models are based on the requirements of Holstein dairy cows under temperate climate. Therefore, the main contribution of this first Chapter was the development and evaluation of a new model to estimate dry matter intake (DMI) of crossbred dairy cows. The new DMI equation was developed and evaluated using a database with 161 treatment means from 38 published studies of Brazilian crossbred cows [n = 446 dairy cows, mean milk production of 16.60 ± 5.70 (SD) kg/day]. The new model showed higher DMI prediction accuracy (MSEP = 1.64, r2 = 0.88) than American models. Therefore, this model can be used in the formulation of diets for crossbred dairy cows under tropical conditions. In Chapter 2, we evaluated the updates on the nutritional requirements of dairy calves suggested by Van Amburgh and Drackley (2005) and inputted into Agricultural Modeling and Training Systems (AMTS, AMTS.Cattle.ProTM Calf Model version 3.5.8.0, 2015). Data from 16 previous Brazilian studies involving 51 diets for dairy calves (n = 485 calves, weaning weight of 62.02 ± 10.16 kg) were used to evaluate the Nutrient Requirements of Dairy Cattle - NRC (2001) and AMTS models. According to the data set from this study, both systems overestimated the gain calculated from the available protein. The weight gain estimated by the energy available from the diet was overestimated by 19 g/day when calculated by the NRC and underestimated by 68 g/day when calculated by AMTS. Weight gain estimated by the dietary energy of the AMTS system was lower than the NRC (2001). The Brazilian data set showed a calf gain around 32% lower than those described in U.S. dairy farms. The reasons for this discrepancy need to be understood, for only then can new models be developed and parameterized to estimate animal performance more accurately and precisely.

O rebanho brasileiro de bovinos leiteiros é composto predominantemente por vacas mestiças criadas em condições tropicais. Os principais modelos usados no Brasil para a formulação e avaliação de dietas são norte-americanos e baseados nas exigências de vacas Holandesas em clima temperado. A principal contribuição do primeiro capítulo deste estudo foi o desenvolvimento e avaliação de um novo modelo para estimar a ingestão de matéria seca (IMS) de vacas mestiças. A nova equação de predição da IMS foi desenvolvida e avaliada usando um banco de dados com 161 médias de tratamentos provenientes de 38 estudos brasileiros publicados com vacas mestiças [n = 446 vacas, 16,60 ± 5,70 (DP) kg leite/dia]. O modelo proposto foi: IMS (kg/dia): [0,5552 (± 0,06636 EP) × LC4%G + 0,06332 (± 0,009455) × PV0,75] × [1 - e (- 0,7732 (± 0,7019) × (SEM - 1,629 (± 1,9313))) ]. O novo modelo apresentou maior acurácia (QMEP = 1,64, r2 = 0,88) na predição da IMS quando comparado com modelos norte-americanos. O novo modelo de predição da IMS pode ser aplicado na formulação de dietas para vacas leiteiras mestiças em condições tropicais. No capítulo 2, um estudo avaliou as novas atualizações das exigências nutricionais de bezerras leiteiras sugeridas por Van Amburgh e Drackley (2005) e inseridas em programa comercial (AMTS, Agricultural Modeling and Training Systems, AMTS.Cattle.ProTM Calf Model versão 3.5.8.0, 2015). Dados de 16 estudos brasileiros envolvendo 51 dietas para bezerros leiteiros (n = 485, peso ao desmane de 62,02 ± 10,16 kg) foram usados para avaliar os modelos do NRC (2001) e o AMTS. Nas condições do conjunto de dados do presente estudo, ambos os sistemas superestimaram o ganho calculado a partir da proteína disponível. O ganho de peso estimado pela energia disponível da dieta, foi superestimado em 19 g/dia quando calculado pelo NRC e subestimado em 68 g/dia quando calculado pelo AMTS. O ganho de peso possível a partir da energia disponível foi menor quando calculado pelo AMTS em relação ao NRC (2001). Para o conjunto de dados experimentais brasileiros, os ganhos de peso de bezerros foram 32% inferiores aos descritos na literatura para sistemas norte-americanos. As razões para esta discrepância precisam ser compreendidas, pois somente assim novos modelos poderão ser desenvolvidos e parametrizados para estimar o desempenho animal de forma mais acurada e precisa. / The Brazilian dairy herd is mainly made up of crossbred cows fed under tropical conditions. The main models in use to formulate and evaluate diets in Brazil are from North America. These models are based on the requirements of Holstein dairy cows under temperate climate. Therefore, the main contribution of this first Chapter was the development and evaluation of a new model to estimate dry matter intake (DMI) of crossbred dairy cows. The new DMI equation was developed and evaluated using a database with 161 treatment means from 38 published studies of Brazilian crossbred cows [n = 446 dairy cows, mean milk production of 16.60 ± 5.70 (SD) kg/day]. The new model showed higher DMI prediction accuracy (MSEP = 1.64, r2 = 0.88) than American models. Therefore, this model can be used in the formulation of diets for crossbred dairy cows under tropical conditions. In Chapter 2, we evaluated the updates on the nutritional requirements of dairy calves suggested by Van Amburgh and Drackley (2005) and inputted into Agricultural Modeling and Training Systems (AMTS, AMTS.Cattle.ProTM Calf Model version 3.5.8.0, 2015). Data from 16 previous Brazilian studies involving 51 diets for dairy calves (n = 485 calves, weaning weight of 62.02 ± 10.16 kg) were used to evaluate the Nutrient Requirements of Dairy Cattle - NRC (2001) and AMTS models. According to the data set from this study, both systems overestimated the gain calculated from the available protein. The weight gain estimated by the energy available from the diet was overestimated by 19 g/day when calculated by the NRC and underestimated by 68 g/day when calculated by AMTS. Weight gain estimated by the dietary energy of the AMTS system was lower than the NRC (2001). The Brazilian data set showed a calf gain around 32% lower than those described in U.S. dairy farms. The reasons for this discrepancy need to be understood, for only then can new models be developed and parameterized to estimate animal performance more accurately and precisely.

Anaemia, which is caused by iron deficiency, is a global nutritional disorder of utmost concern. It has been assumed that meat, which contains haem as well as non-haem iron, enhances non-haem iron absorption due to the presence of the “meat factor”. In the experiment reported here, 24 semi-anaemic, 3-week-old piglets were utilised as human nutrition models to assess the effects of dietary lactoferrin, meat and LSF (Low Molecular Weight Sarcoplasmic Fraction) on iron bioavailability during a 4-week feeding period. The parameters that were used as measurements of iron bioavailability were changes in haematological indices, haemoglobin iron repletion efficiency, intestinal morphology and mineral balances. Non-significant (p > 0.05) dietary effects were observed for growth performance and for all the haematological and some histological parameters (small intestine villi height, crypt depth and mucosal thickness). Haemoglobin iron repletion efficiency was highest for the control group and was not significantly different between the other 3 diets. The superiority of the control diet in this respect was not consistent with previous trials and can not be explained. Results suggested that increased retention of calcium, magnesium, phosphorous and manganese tended to inhibit iron absorption. However, the LSF and meat diets significantly (p = 0.003) increased the number of goblet cells/100 µm suggesting that mucin secretion was favoured by these two diets. Additionally, all immunological parameters were significantly (p < 0.05) improved by the LSF diet. As such, the LSF diet can be a potential immunobooster feed ingredient for weanling piglets. Overall, the level of LSF in the LSF diet was insufficient to exert a desirable enhancement of iron bioavailability and betterment of 3 iron status of the semi-anaemic piglets relative to the control group. However, the diet containing LSF was as effective as the meat diet with respect to these characteristics.