Agent-based modelling of grassland grazing: a case study from Zeku, China

The livestock grazing system is one of the most important human natural coupling systems on the earth. More than one-quarter of the global land surface is managed grazing grasslands, and the intensification of animal production and grazing systems is likely to continue worldwide. Managing the grassland grazing system towards a sustainable future is, therefore, an important issue for herders, grassland managers and policymakers. This requires dynamic monitoring and assessment of the grazing system, which consists of the complexities plant growth, livestock dynamics, plant-herbivore interactions and grazing management. Leaf Area Index (LAI) is commonly used as a proxy for grassland condition. However, current studies all focus on the year-round aggregated LAI change or seasonal variation rather than the specific grazing-led LAI defoliation for each pixel, which is the important indicator for quantifying grassland grazing activities. The contribution of this research to grassland grazing management can be summarised through three main components: a new growth function under grazing considering both the growth and senescence of grass with an estimation algorithm; the employment of a LUE-VMP model to estimate Net Primary Productivity (NPP) for improved LAI validation; and a first attempt in building an agent-based model (ABMGG) integrated with patch-specific grazing information for the assessment of various grazing management strategies. It was found that although different grazing management scenarios could not significantly improve or decrease grassland productivity, rotational group grazing performed best in terms of producing a smaller number of degraded grassland patches. Although there are some drawbacks, the agent-based modelling is highly suited to the grassland grazing system that is characterized by individual interactions and contains hierarchical grazing strategies and institutional arrangements. It is also suggested that by improvement of the data quality and extension of the model, ABMGG would be able to predict and analyze the performance of different grazing management scenarios further, and would be an important tool for aiding the sustainable development of the grazing system for both herders and policymakers.