Using mechanistic land-use models to explore the potential of agroforestry for enhancing ecological and economic functions

农林业一直被提倡作为满足日益增长的粮食和能源需求的途径，同时加强提供生态系统服务。过去在这一领域的研究往往只关注地块层面，而科学家和决策者正在呼吁对多功能景观进行研究。因此，这里提出的研究旨在为多功能发展一种机械的土地使用模拟方法，以确定农林业有助于减少土地使用冲突的条件。它的目的是调查在地块、农场和景观层面上哪种程度的多样化是最合适的，以将公众利益和农民的目标结合起来。为此，将进一步开发新的单目标和多目标建模方法来优化不确定条件下的土地利用分配。对于第一个原型模型，该方法将用于巴拿马东部森林边界的典型热带景观。景观层面的方法应首先回答这样一个问题，即在景观组合中需要多少农林业来实现多种生态系统服务，同时减少它们之间的权衡（多目标模型）。农场层面的模型应调查农林业应如何设计，以成为一个有效的农业组合的一部分，满足农民的（最可能的经济）需求（单目标模型）。使用基于WaNuLCAS的过程，可以在不同条件下模拟一系列已经常见和潜在的新农林业系统和布局。其中，具体布局将根据优化方法中确定的条件进行预选。最后，将综合农场模式的土地利用构成和生态系统服务提供与景观模式进行对比，以研究纳入高效农林复合系统是否可以减少潜在差异。拟议的研究具有创新性，因为它将农林业与其他形式的土地利用多样化进行比较，并将不确定性纳入决策过程。新颖的鲁棒优化技术允许降低模型复杂性，从而包含大量的土地使用选项。因此，所提出的研究可能对设计弹性多功能景观和指导未来农林业研究做出重要贡献。

Agroforestry has been advocated as a pathway to satisfy growing food and energy demands, while enhancing the provision of ecosystem services. Past research in this field has often focused only on the plot level, while scientists and policy makers are calling for research on multifunctional landscapes. The research proposed here therefore seeks to develop a mechanistic land-use modelling approach for multifunctionality to identify the conditions under which agroforestry could contribute to reducing land-use conflicts. It aims at investigating which level of diversification will be most appropriate at the plot, farm and landscape level to bring together public interests and farmer s objectives. For this purpose, novel single and multi objective modelling approaches to optimize land use allocation under uncertainty will be further developed. For a first prototype model, the approach will be developed for a typical tropical landscape at the forest frontier in eastern Panama. A landscape level approach shall first answer the question, how much of agroforestry will be desirable in a landscape portfolio to achieve multiple ecosystem services, while reducing trade-offs between them (multi objective model). A farm level model shall investigate how agroforestry should be designed to become an efficient part of a farm portfolio that satisfies the farmer s (most likely economic) needs (single objective model). A range of already common and potentially new agroforestry systems and layouts could be simulated under different conditions using the process based WaNuLCAS. Among these, specific layouts will be pre selected according to the conditions identified during the optimization approach. Finally, the land use composition and provision of ecosystem services of the aggregated farm models shall be contrasted to the landscape model, in order to investigate whether including efficient agroforestry systems may reduce potential differences. The proposed research is innovative in that it compares agroforestry to other forms of land-use diversification and incorporates uncertainties into the decision making process. Novel robust optimization techniques allow for a reduction of model complexity and thus inclusion of a large number of land-use options. The proposed research may therefore make an important contribution to designing resilient multifunctional landscapes and directing future agroforestry research.