ABSTRACT Despite tremendous advancements in the livestock sector, additional opportunities exist to improve even further livestock production around the globe. Forecasting is not an exact science and it relies heavily on past and current knowledge. Improvements in the nutritional sciences (both human and animal) include a better understanding of agents that cause deterioration of human health, improving the quality of animal products, applying effective fetal programming, developing new feeds and feeding strategies, and revisiting longstanding technologies. Improvements in the understanding of the rumen microbiome will enable scientists to increase the fermentation efficiency and, hopefully, select microbial species of greater interest. Improvements in remote sensing and ground-based instrumentation, telecommunications, and weather forecasting technologies will aid in the continued improvements of early warning systems to assist livestock producers in reducing risk and adapting to the changing environment. Broad utilization of sensor technologies will allow scientists to collect real-time data and, when combined with mathematical modeling, decision support systems will become an indispensable managerial tool for livestock production with the possibility to automate low-level decisions on the farm, such as supplementation schedules, sorting of animals, and early detection of disease and outbreaks. The identification of feed efficient animals may be the single most impactful advancement towards long-term livestock sustainability and the promise of feeding the world animal products. We contend that education across societal levels is the first step to solve current and future challenges of the livestock industry. The dilemma has been who will take the first step forward.

Protein supply and requirements by ruminants have been studied for more than a century. These studies led to the accumulation of lots of scientific information about digestion and metabolism of protein by ruminants as well as the characterization of the dietary protein in order to maximize animal performance. During the 1980s and 1990s, when computers became more accessible and powerful, scientists began to conceptualize and develop mathematical nutrition models, and to program them into computers to assist with ration balancing and formulation for domesticated ruminants, specifically dairy and beef cattle. The most commonly known nutrition models developed during this period were the National Research Council (NRC) in the United States, Agricultural Research Council (ARC) in the United Kingdom, Institut National de la Recherche Agronomique (INRA) in France, and the Commonwealth Scientific and Industrial Research Organization (CSIRO) in Australia. Others were derivative works from these models with different degrees of modifications in the supply or requirement calculations, and the modeling nature (e.g., static or dynamic, mechanistic, or deterministic). Circa 1990s, most models adopted the metabolizable protein (MP) system over the crude protein (CP) and digestible CP systems to estimate supply of MP and the factorial system to calculate MP required by the animal. The MP system included two portions of protein (i.e., the rumen-undegraded dietary CP - RUP - and the contributions of microbial CP - MCP) as the main sources of MP for the animal. Some models would explicitly account for the impact of dry matter intake (DMI) on the MP required for maintenance (MPm; e.g., Cornell Net Carbohydrate and Protein System - CNCPS, the Dutch system - DVE/OEB), while others would simply account for scurf, urinary, metabolic fecal, and endogenous contributions independently of DMI. All models included milk yield and its components in estimating MP required for lactation (MPl) and calf birth weight and some form of an empirical, exponential equation to compute MP for pregnancy (MPp). The MP required for growth (MPg) varied tremendously among the original models and their derivative works mainly due to the differences in computing growth pattern and the composition of the gain. The calculation of MCP differs among models; some rely on the total digestible nutrient (TDN; e.g., NRC, CNCPS level 1) intake to estimate MCP, while others use fermentable organic matter (FOM; e.g., INRA, DVE/OEB), fermentable carbohydrate (e.g., CNCPS level 2, NorFor), or metabolizable energy (ME; e.g., ARC, CSIRO, Rostock). Most models acknowledged the importance of ruminal recycled N, but not all accounted for it. Our Monte Carlo simulation indicated the prediction of most models for required MPl overlapped, confirming uniformity among models when predicting requirements for lactating animals, but a large variation in required MPg for growing animals exists.

A mechanistic model that predicts nutrient requirements and biological values of feeds for sheep (Cornell Net Carbohydrate and Protein System; CNCPS-S) was expanded to include goats and the name was changed to the Small Ruminant Nutrition System (SRNS). The SRNS uses animal and environmental factors to predict metabolizable energy (ME) and protein, and Ca and P requirements. Requirements for goats in the SRNS are predicted based on the equations developed for CNCPS-S, modified to account for specific requirements of goats, including maintenance, lactation, and pregnancy requirements, and body reserves. Feed biological values are predicted based on carbohydrate and protein fractions and their ruminal fermentation rates, forage, concentrate and liquid passage rates, and microbial growth. The evaluation of the SRNS for sheep using published papers (19 treatment means) indicated no mean bias (MB; 1.1 g/100 g) and low root mean square prediction error (RMSPE; 3.6 g/100g) when predicting dietary organic matter digestibility for diets not deficient in ruminal nitrogen. The SRNS accurately predicted gains and losses of shrunk body weight (SBW) of adult sheep (15 treatment means; MB = 5.8 g/d and RMSPE = 30 g/d) when diets were not deficient in ruminal nitrogen. The SRNS for sheep had MB varying from -34 to 1 g/d and RSME varying from 37 to 56 g/d when predicting average daily gain (ADG) of growing lambs (42 treatment means). The evaluation of the SRNS for goats based on literature data showed accurate predictions for ADG of kids (31 treatment means; RMSEP = 32.5 g/d; r2= 0.85; concordance correlation coefficient, CCC, = 0.91), daily ME intake (21 treatment means; RMSEP = 0.24 Mcal/d g/d; r2 = 0.99; CCC = 0.99), and energy balance (21 treatment means; RMSEP = 0.20 Mcal/d g/d; r2 = 0.87; CCC = 0.90) of goats. In conclusion, the SRNS for sheep can accurately predict dietary organic matter digestibility, ADG of growing lambs and changes in SBW of mature sheep. The SRNS for goats is suitable for predicting ME intake and the energy balance of lactating and non-lactating adult goats and the ADG of kids of dairy, meat, and indigenous breeds. The SRNS model is available at http://nutritionmodels.tamu.edu.

Modelos matemáticos podem ser utilizados para melhorar a performance, reduzir os custos de produção, e minimizar a exceção de nutrientes através de melhores estimativas da exigência e utilização de alimentos em vários cenários produtivos. Modelos matemáticos podem ser classificados em cinco ou mais categorias dependendo da sua natureza. Um dos maiores problemas na construção de modelos matemáticos é o nível de agregação das equações. Os passos mais importantes são o estabelecimento do propósito do modelo, determinação da melhor combinação de equações empíricas e teóricas para representar das funções fisiológicas dado a disponibilidade de banco de dados, informações tipicamente encontradas a nível de campo, e os benefícios e riscos associados com o uso do modelo na produção animal. Nesse artigo são discutidos cinco sistemas de alimentação padrão de ruminantes mais utilizados atualmente. Eles compartilham de conceitos de exigência e disponibilidade de energia e nutrientes, mas diferem na estrutura e como esses conceitos são abordados. Modelos animais podem ser utilizados para vários propósitos, entre eles uma simples descrição de observações, estimativa de respostas à diferentes manejos, e caracterização de mecanismos biológicos. Dependendo dos objetivos, várias alternativas podem ser utilizadas na construção do modelo matemático, entre elas, equações algébricas simples, equações de relação puramente estatísticas, ou até modelos mecanicistas e dinâmicos. Esse último favorece o uso da quantidade crescente de informações cientificas relacionadas à biologia animal. O desenvolvimento contínuo desses tipos de modelos juntamente com as inovações computacionais e de softwares permitem avanços na forma de uso dos conhecimentos fundamentais de nutrição animal de forma que a produção animal possa ser melhor explorada ao mesmo tempo reduzindo-se o impacto ambiental.

Mathematical models can be used to improve performance, reduce cost of production, and reduce nutrient excretion by accounting for more of the variation in predicting requirements and feed utilization in each unique production situation. Mathematical models can be classified into five or more categories based on their nature and behavior. Determining the appropriate level of aggregation of equations is a major problem in formulating models. The most critical step is to describe the purpose of the model and then to determine the appropriate mix of empirical and mechanistic representations of physiological functions, given development and evaluation dataset availability, inputs typically available and the benefits versus the risks of use associated with increased sensitivity. We discussed five major feeding systems used around the world. They share common concepts of energy and nutrient requirement and supply by feeds, but differ in structure and application of the concepts. Animal models are used for a variety of purposes, including the simple description of observations, prediction of responses to management, and explanation of biological mechanisms. Depending upon the objectives, a number of different approaches may be used, including classical algebraic equations, predictive empirical relationships, and dynamic, mechanistic models. The latter offer the best opportunity to make full use of the growing body of knowledge regarding animal biology. Continuing development of these types of models and computer technology and software for their implementation holds great promise for improvements in the effectiveness with which fundamental knowledge of animal function can be applied to improve animal agriculture and reduce its impact on the environment.

O uso do Sistema de Carboidrato e Proteina Líquidos da Universidade de Cornell (CNCPS) tanto para produção de leite como carne tem aumentado durante o últimos anos nas regiões tropicais. Entretanto, a falta de uma caracterização adequada de alimentos tem restringido o seu uso corretamente. Esse trabalho teve como objetivo principal o desenvolvimento e a avaliação de uma tabela de composição de alimentos utilizados nas condições tropicais. Os dados da composição desses alimentos foram baseados nas informações necessárias para o uso do modelo CNCPS desenvolvido pela Universidade de Cornell, USA. A composição desses alimentos foi obtida através de análises realizadas em laboratórios e de experimentos publicados em revistas científicas. Os nutrientes digestíveis totais (NDT) estimados através da composição de carboidratos e proteina dos alimentos pela equação de Weiss et al. (1992) e pelo modelo CNCPS foram comparados com os valores da tabela. O NDT estimado ao nível de mantença (1x) com o modelo CNCPS obteve valores próximos aos estimados pela equação de Weiss et al. (1992) (r² = 92.7% e bias = 0.8%). Entretanto, o r² da regressão entre os valores de NDT da tabela e o estimado pelo CNCPS e por Weiss foram menores (58.1 e 67.5%, respectivamente). Uma comparação completa entre os valores observados e preditos não foi possível devido a falta de caracterização dos alimentos conforme necessário pelos modelos testados. Quando os valores médios de literatura foram utilizados, a correlação entre o NDT estimado e o observado foi muito baixa. Concluímos que os valores de NDT estimados por Weiss e modelo CNCPS fornecem melhores estimativas de NDT do que os valores de tabela. A maioria dos trabalhos publicados que foram avaliados nesse estudo raramente incluíam informações necessárias para modelos como o CNCPS.

The Cornell Net Carbohydrate and Protein System (CNCPS) model has been increasingly used in tropical regions for dairy and beef production. However, the lack of appropriate characterization of the feeds has restricted its application. The objective of this study was to develop and evaluate a feed library containing feeds commonly used in tropical regions with characteristics needed as inputs for the CNCPS. Feed composition data collected from laboratory databases and from experiments published in scientific journals were used to develop this tropical feed library. The total digestible nutrients (TDN) predicted at 1x intake of maintenance requirement with the CNCPS model agreed with those predicted by the Weiss et al. (1992) equation (r² of 92.7%, MSE of 13, and bias of 0.8%) over all feeds. However, the regression r² of the tabular TDN values and the TDN predicted by the CNCPS model or with the Weiss equation were much lower (58.1 and 67.5%, respectively). A thorough comparison between observed and predicted TDN was not possible because of insufficient data to characterize the feeds as required by our models. When we used the mean chemical composition values from the literature data, the TDN predicted by our models did not agree with the measured values. We conclude using the TDN values calculated using the Weiss equation and the CNCPS model that are based on the actual chemical composition of the feeds result in energy values that more accurately represent the feeds being used in specific production situations than do the tabular values. Few papers published in Latin America journals that were used in this study reported information need by models such as the CNCPS.