The illusion of diversity: sampling design drives conflicting estimates of soil bacterial richness

Abstract

Soil microbial diversity is shaped by the spatial scale at which communities are sampled, yet standard sampling practices often homogenize samples, obscuring fine-scale spatial structure and diversity patterns. To better understand how sampling effort, spatial extent, and physical homogenization influence plot-level microbial richness estimates, we sampled 57 forest and grassland sites across three regions in Germany using a 14-core cross-transect design, and performed 16S rRNA gene metabarcoding. We simulated sampling efforts of 1-14 cores and a range of spatial extents, and compared diversity estimates to those from physically homogenized composite samples. Plot-level richness increased continuously with sampling effort and spatial extent, with no evidence of saturation. However, when sequencing depth was held constant, sampling completeness declined with increasing sampling effort, meaning that more diversity is not captured. Composite samples substantially underestimated plot-level richness and altered apparent diversity relationships between ecosystems; individual cores identified forests as richer than grasslands, whereas homogenized samples suggested the opposite relationship. These results demonstrate that sampling effort, spatial extent, and homogenization fundamentally shape soil microbial diversity estimates. Homogenized composite samples cannot substitute for individual cores when the goal is to reliably quantify plot-level richness or compare diversity across ecosystems.