Aim: Biodiversity and ecosystem productivity vary across the globe, and considerable effort has been made to describe their relationships. Biodiversity and ecosystem functioning research has traditionally focused on how experimentally controlled species richness affects net primary productivity (S → NPP) at small spatial grains. In contrast, the influence of productivity on richness (NPP → S) has been explored at many grains in naturally assembled communities. Mismatches in spatial scale between approaches have fuelled debate about the strength and direction of biodiversity–productivity relationships. Here, we examine the direction and strength of the influence of productivity on diversity (NPP → S) and the influence of diversity on productivity (S → NPP) and how these vary across spatial grains. Location: Contiguous USA. Time period: 1999–2015. Major taxa studied: Woody species (angiosperms and gymnosperms). Methods: Using data from North American forests at grains from local (672 m2) to coarse spatial units (median area = 35,677 km2), we assess relationships between diversity and productivity using structural equation and random forest models, while accounting for variation in climate, environmental heterogeneity, management and forest age. Results: We show that relationships between S and NPP strengthen with spatial grain. Within each grain, S → NPP and NPP → S have similar magnitudes, meaning that processes underlying S → NPP and NPP → S either operate simultaneously or that one of them is real and the other is an artefact. At all spatial grains, S was one of the weakest predictors of forest productivity, which was largely driven by biomass, temperature and forest management and age. Main conclusions: We conclude that spatial grain mediates relationships between biodiversity and productivity in real-world ecosystems and that results supporting predictions from each approach (NPP → S and S → NPP) serve as an impetus for future studies testing underlying mechanisms. Productivity–diversity relationships emerge at multiple spatial grains, which should widen the focus of national and global policy and research to larger spatial grains.

中文翻译：

---

生产力-多样性关系的跨尺度评估

目标：生物多样性和生态系统生产力在全球范围内各不相同，已经做出了相当大的努力来描述它们之间的关系。生物多样性和生态系统功能研究传统上侧重于实验控制的物种丰富度如何影响小空间谷物的净初级生产力（S → NPP）。相比之下，已经在自然聚集的群落中的许多谷物中探索了生产力对丰富度的影响（NPP → S）。方法之间空间尺度的不匹配引发了关于生物多样性-生产力关系的强度和方向的争论。在这里，我们研究了生产力对多样性影响的方向和强度（NPP → S）和多样性对生产力的影响（S → NPP），以及它们如何在空间粒度上变化。地点：美国本土。时间段：1999-2015。研究的主要分类群：木本植物（被子植物和裸子植物）。方法：使用北美森林的谷物数据，从当地（672 平方米）到粗略空间单位（中位数面积 = 35,677 平方公里），我们使用结构方程和随机森林模型评估多样性和生产力之间的关系，同时考虑气候变化，环境异质性、管理和森林年龄。结果：我们表明 S 和 NPP 之间的关系随着空间晶粒而加强。在每个颗粒中，S → NPP 和 NPP → S 具有相似的量级，这意味着 S → NPP 和 NPP → S 下的过程要么同时运行，要么其中一个是真实的，另一个是人工制品。在所有空间颗粒中，S 是森林生产力最弱的预测因子之一，主要受生物量、温度和森林管理和年龄驱动。主要结论：我们得出结论，空间粒度介导了现实世界生态系统中生物多样性和生产力之间的关系，并且支持每种方法（NPP → S 和 S → NPP）的预测的结果可作为未来研究测试潜在机制的动力。生产力-多样性关系出现在多个空间粒度上，这应该将国家和全球政策和研究的重点扩大到更大的空间粒度。