How to take representative samples to quantify microplastic particles in soil?

The amount of plastic particles in terrestrial ecosystems is not well known, not only because it is difficult to extract and identify plastic particles from terrestrial samples, but also because it is challenging to take representative samples from soils or sediments. Here, we numerically simulated how to take representative terrestrial samples to quantify plastic particles, and we evaluated the accuracy (error) of reported plastic concentrations in the literature. Fields with randomly distributed plastic particles (uniform and clustered) were numerically generated and sampled to determine the representative elementary volume (REV) and the required number of samples to quantify plastic concentrations (10 to 106 particles/m2) with different relative errors (5%, 10%, 15%). The REV and the required number of samples decrease hyperbolically as the plastic concentration increases, indicating a strong non-linear relation. For instance, hundreds to thousands of soil cores (8-cm diameter) would be required to quantify plastics at low concentrations (102 particles/m2), while a few cores are sufficient at high plastic concentrations (105 particles/m2). For an accurate measurement of plastic concentrations, the total surface area of samples taken should approach or exceed the REV. We recommend to take replicated samples with each sample as large as possible (e.g., 1 m × 1 m) rather than multiple small cores, and then reduce the soil volume by the quartering method.