Impact of land-use and functional diversity on diversity and stability of grassland communities in fragmented landscapes

随着物种丰富度的急剧下降，预测全球变化（气候变化和土地利用集约化）对草地植物群落组成和服务的长期影响变得越来越重要。然而，到目前为止，我们进行这种预测的能力非常有限。为了提高我们的预测能力，我们必须更多地了解群落结构组成部分和生态系统功能之间的相互作用。基于从功能性状重建群落生态的观点，本文提出了一种新的方法，将植物群落的响应性状多样性（生活史性状多样性）与破碎化景观中物种多样性和生态系统生产力的长期持续性联系起来。我们的方法由三个互补的工具组成：基于方程的元群落模拟模型、物种分布模型和贝叶斯信念网络。利用这个多样化的工具箱，我们试图将有关物种的生活史、分布和地理范围的信息与合适的草原栖息地的空间分布信息结合起来。该方法将应用于生物多样性实验区的草地元群落。机械模型是这个项目的重要组成部分，因为它们能够整合来自不同领域和不同空间尺度的知识。具体来说，使用本文提出的框架，模型将性状和性能指标（如物种多样性的持久性和生态系统生产力）联系起来，如果没有建模，这是不可能的。最终，我们寻求整合概念和经验知识。因此，我们给出了对后续阶段计划的因果模型发展的展望。因果模型易于沟通，并且可以在获得新数据或概念知识时不断更新。然而，为当前项目阶段计划的步骤将显著提高我们对全球变化的元社区响应的预测能力。

With a dramatic decline in species richness, it becomes increasingly important to predict the long-term impact of global changes (climate change and land-use intensification on grassland plant community composition and services. However, to date, our ability for such predictions is very limited. To improve our predictive ability, we have to learn more about the interplay between components of community structure and ecosystem functioning. Motivated by a recent call to rebuild community ecology from functional traits, we propose a new approach to link the relationship between the response trait diversity of plant communities, i.e. diversity of life-history traits, and the long-term persistence of species diversity and ecosystem productivity in fragmented landscapes. Our approach consists of the combination of three complementary tools: an equation-based metacommunity simulation model, species distribution models, and Bayesian belief networks. Using this diverse toolbox, we seek to combine information concerning the life-history, dispersal, and geographic range of species with information about the distribution of suitable grassland habitats in space. This approach will be applied to grassland metacommunities in the Biodiversity Exploratories. Mechanistic models are an essential component of this project because of their ability to incorporate knowledge from disparate domains and different spatial scales. Specifically, using the framework proposed here, models link traits and performance measures such as persistence of species diversity and ecosystem productivity and this would not be possible without modelling. Ultimately, we seek to integrate both conceptual and empirical knowledge. Thus, we give an outlook towards the development of causal models scheduled for a follow-up phase. Causal models are easy to communicate and can be continually updated when new data or conceptual knowledge becomes available. However, the steps planned for the current project phase will already noticeably improve our predictive ability on metacommunity responses to global change.