An Empirical Test of Scale‐Dependence of β‐Diversity in a Large Temperate Forest Region

ABSTRACT Aim Variation in species composition among locations (β‐diversity) is of fundamental interest in ecology, yet its dependence on spatial extent and grain has rarely been assessed empirically. This study provides a systematic empirical test of the spatial scaling of β‐diversity and highlights the importance of study design for detecting the role of environmental filtering in shaping regional β‐diversity. Location Northeastern China. Taxon Tree in forest ecosystems. Methods We constructed regional communities to define six gradients of spatial extent and grain. We investigated how two types of spatial scale affect a range of aggregate measures of β‐diversity and corresponding β ‐deviations, as well as the distance‐decay relationship (DDR). Variation in β ‐deviations was further partitioned into habitat heterogeneity, spatial aggregation and their joint effects across scales. Results Raw β‐diversity did not exhibit consistent trends with extent across metrics, but it consistently decreased with increasing grain. Both raw β‐diversity and β ‐deviations were strongly influenced by sampling effort. The decreasing trend in Sørensen differentiation ( q = 0) with increasing extent, predicted by theory, was removed after controlling for sampling effort. Notably, only abundance‐based measures showed a significant increase with spatial extent that was unaffected by sampling effort. Sørensen differentiation also decreased with increasing grain, contrary to theoretical expectations. The DDR was sensitive to grain size, and the proportion of variation in β ‐deviations explained by habitat heterogeneity increased with both spatial extent and grain. Main Conclusion The response of raw β‐diversity to extent depended on the metric used and was sensitive to sampling effort. Pairwise abundance‐based measures may provide a more reliable basis for cross‐scale comparisons. The DDR may only be able to capture scale‐specific ecological processes. Environmental filtering emerges as the dominant driver of community assembly at larger spatial scales, highlighting how analytical choices influence inference about assembly mechanisms.