Mechanistic modelling of sustainability metrics and performance in broilers

点击翻译按钮获取中文摘要

As the global population escalates towards an estimated 9.5 billion in 2050 (FAOSTAT, 2013), the need to efficiently produce animal protein (milk, meat, eggs) for human consumption whilst minimizing the impact on the environment (methane, nitrogen, phosphorous, water & land use) will only continue to increase. These animal production systems must also manage increasing societal pressures to reduce production intensity (e.g. pasture-based systems, slow-growing breeds) and remove antibiotics and hormones from animal feed - strategies that may directly counter efforts to improve efficiency and production. Such complexities are difficult to optimize within individual research trials or even to extract from whole bodies of scientific research literature. With complex biological systems, such as the digestive and metabolic pathways that drive nutrient utilization and animal performance, models play a key role in distilling knowledge from information and identifying optimized feeding and mitigation strategies. Mechanistic models (MM) have served as decision-support and opportunity analysis tools within animal production, across species, for decades. However, they must also evolve to meet the wave of the future - utilizing new `big data' streams, hybridizing with machine learning methodologies, as well as aid the movement towards precision nutrition (via capturing additional sources of performance variation) as a strategy to reduce inefficiency in animal production. The objective of this research program is to therefore to develop models that (1) support the efficient, sustainable and ethical production of animal protein for human consumption, and that (2) serve to increase our knowledge of complex biological systems. Anticipated Impact: More efficient, sustainable, ethical and intelligent production of animal protein for human consumption. To this end, this research program proposes the development of a dynamic mechanistic broiler model able to consider sources of performance variation from (1) genetics (including new slow growing broiler breeds, as well as address between-animal variation in genetic potential), (2) nutrient digestion kinetics and non-nutritional feed attributes (large source of performance variation largely not captured or utilized in growth models), and to predict (3) the resulting nutrient excretion (nitrogen, phosphorous, methane) under varying nutritional and management scenarios (slow growing vs conventional breeds, physical feed attributes, etc). This research program will also explore the utilization of current big data streams and machine learning approaches to parameterize the developed model stochastically and capture between animal variation. Anticipated Impact: The development of an openly available mechanistic broiler model that can be used to assess nutritional, environmental, management, and nutrient excretion strategies in order to optimize performance and reduce the environmental impact of broiler production.