A distributed large‐volume soil sampling approach to address spatial heterogeneity in eDNA data collection

Soil sampling for eDNA is a useful approach to assess ecological communities in specific locations, such as comparing community composition across different site treatments. However, species' dispersal abilities and habitat selection, among many other factors, drive the uneven distribution of organisms across the landscape. Thus, to accurately detect and evaluate community composition using soil sampling, we need to sample the heterogeneity of eDNA distribution in a way that is both rigorous and easy to implement in the field. We developed an edge‐to‐edge high‐volume soil sampling protocol for eDNA data collection and described a calibration approach to identify the amount of soil to collect in a given ecosystem or land use/cover system. This calibration approach allows our sampling strategy to be broadly applicable across many ecosystems and taxa of interest. Our proof of concept examined arthropod communities in Pará, Brazil across multiple land management systems, including a syntropic system, a forest restoration area, second growth forest, and conventional agriculture. The sampling approach successfully distinguished arthropod communities between management systems. Balancing cost and detection rates during calibration resulted in a recommended pooled sample volume of 2.4 L for this system. Further, field technicians found that this sampling approach was easy to learn and could be rapidly executed in multiple land management systems. Many important research questions involve spatially discrete site treatments. This approach is a cost‐effective sampling strategy that is ideal for assessing communities of organisms across site treatments or management units, such as different land management systems. With calibration, it can be used in any ecosystem with accumulated soil and for any target taxonomic group. In addition, this approach could be applied to research questions where the primary sampling unit is volume. While this approach is not efficient for extremely large sites such as entire forests, this high‐volume sampling approach could also be adapted to sample different microclimates, such as sampling forest clearings.