Modeling in animal science: an introduction to quantitative understanding and prediction

Abstract Text:

In animal science, continuous advances in technology, computing, and engineering result in the generation of data at a rapidly increasing rate. Mathematical models enable quantitative analysis and integration of data to study the behavior and complexity of biological systems. This review highlights several aspects of modeling in the context of understanding, predicting and modifying complex processes in farm animal systems, and offers a current perspective for animal scientists without requiring specialized knowledge of mathematics or bioinformatics.  A mathematical model is an equation or set of equations which represents the behavior of a system, and can be viewed as an idea, hypothesis or relation expressed in mathematics. In animal science, the system may range from the molecules in cells up to herd or flock level, with any level of the system being composed of subsystems lying at a lower level, or being a subsystem of higher level systems itself. In empirical models, experimental data are used directly to quantify relationships based at a single level. Alternatively, mechanistic models are process-based and seek to understand causation in the system of interest by describing a system level in terms of components and associated processes at subsystem levels. Furthermore, models may be static, capturing behavior of the system at a particular point in time, or dynamic, describing how quantities in the system change with time. Several key benefits have been attributed to modeling. Firstly, models can provide an integrative, quantitative understanding of mechanisms and associated relationships between responses of a system at various levels. Secondly, building a model may pinpoint areas where data or knowledge are lacking, and may indicate priorities for further research and development. Thirdly, models provide quantitative assessments of management practices for the animal production sector including policy makers. This aspect becomes particularly important when observations are hardly possible because of time scale (changes emerging after several years or decades only) or technical difficulty of measurements. Two areas are in need of further development. Emerging –omics data on genetic and metabolic regulatory networks at the molecular and cellular level require further modeling methodology efforts to integrate such data with processes at a higher system level. Secondly, further advances in understanding and prediction at integrated levels will be obtained upon combination of models that differ in underlying methodology. Examples include the integration of mechanistic models of animal metabolism with linear programming and life cycle assessment models.

Keywords: modeling; animal science; system