| 24 Sep 2026
A global dataset of soil microplastics (2005–2026)
Abstract. Microplastics (<5 mm) pose persistent global ecological risks due to their ubiquitous environmental distribution and negligible biodegradability. Compared with marine and freshwater ecosystems, for which global-scale distribution datasets have been increasingly developed, soil microplastic research still lacks a unified dataset that can support cross-regional comparison and geospatial analysis. Existing syntheses are mostly limited to national or regional scales, and the few global compilations remain constrained by data volume, spatial coverage, or variable harmonization. To address this gap, we systematically compiled published records of soil microplastics worldwide from 2005 to 2026. The resulting database comprises 6,197 harmonized records from 4,901 sampling sites across 49 countries, nearly doubling the record-level evidence base compared with the largest soil-microplastic distribution or risk synthesis identified to date. To enable cross-study synthesis, we developed a standardized schema to harmonize microplastic abundance records, land-use classifications, polymer types, shape categories, and particle-size data. Notably, we implemented particle-size distribution models to correct and harmonize abundance data reported across heterogeneous size fractions. Analysis of the compiled data reveals a highly uneven geographic distribution of current records, with Asia accounting for 89% of records with abundance information and China contributing the largest share. Soil microplastic abundance showed strongly right-skewed distributions across land-use types, and abundance patterns differed markedly between quantification methods. Spectroscopy-based records showed significantly higher median abundance than visual-based records. Differences in particle size, shape, and polymer composition among land-use categories further indicate substantial heterogeneity in soil microplastic characteristics across terrestrial environments. This dataset is archived on the Zenodo platform at https://doi.org/10.5281/zenodo.20806151 (Wu et al., 2026). The standardized data foundation presented here is critical for initializing global monitoring networks and accurately modeling soil microplastic dynamics across the Earth system.
This paper presnets a valuable and timely dataset of MP in soil. to my knowledge, no global soil microplastic compilation has previously been published as a dedicated, openly archived data product.
The dataset is large with 6,197 records from 4,901 sites and 297 primary studies across 49 countries. The authors also retain useful metadata on sampling depth, analytical method, particle characteristics and soil properties, etc. The decision not to convert mass-based concentrations into particle numbers is sensible .
The paper also recognizes an important problem that many previous syntheses avoid that measurement methodology strongly influences reported microplastic abundance. Separating visual-based and spectroscopy-based measurements is therefore a good idea. The authors also cautiously warn that the difference between the two groups.
The authors acknowledge about the geographical imbalance. Asia contributes 89% of abundance records and China alone contributes 4,316 of 5,952 records. but the dataset should be described as a global compilation of currently published observations, not a representative global survey.
The paper should be imrpbed by making the “raw” dataset available (without harmonization), qualify simple land-use rankings, account explicitly for study/site clustering , and depth with linear mixed models with random effects
Details of Some issues that need to be raise
-The particle-size correction
The manuscript assumes a single power-law size distribution to transform measurements made over different particle-size windows onto a common 20–5000 μm basis. The paper needs to discuss and acknowledge that rhe exponent alpha depends on analytical method, environmental setting, degradation history, polymer,etc and that the correction becomes particularly sensitive when measured and target ranges differ greatly.
Also note the Methods says the relationship was built from 54 particle-size distribution datasets, whereas the Results says 84 datasets. Please cgeck
-Data analysis
The manuscript analyses thousands of “records” as observations. But records are nested within sampling sites and sites are nested within papers. Multiple soil depths from the same site are explicitly retained as separate records. They are therefore not independent observations. Thus it needs mixed-effects/hierarchical models with study-level aggregation with rabdom effects. The same with depth variable
-Land-use comparisons
The manuscript reports a medians—for example spectroscopy-based Forest Land at 11,069 particles kg⁻¹ versus Wetlands at 894 particles kg⁻¹. But land use is confounded with country study, analytical method; sampling depth; etc. These should not be presented as land-use effects.
-Need to listing variables of the data, such as units, allowed values and NA conventions.
-Polymer classes (L130–141). PES is defined as polyethylene succinate (L138). In MP literature PES is considred as polyester, Please check it. Biodegradable polymers (PBAT, PLA), which are used in mulch films, are not discussed.
- while this is the first global compilation data paper, the paper says “nearly doubling the record-level evidence base” (L20–21) but did not mention exising soil MP syntheses (e.g., Büks & Kaupenjohann 2020; Wrigley et al. 2024; Ma et al. 2025; Roy et al. 2025; Maggi et al. 2026) and what and how this is different
Minor edits
L175 defines "α and b", but Eq. 1 and Eq. 2 use a.
L212–234: §3.5 is a stratified comparison, not a sensitivity analysis.
L300–302: Rolf et al. (2022) is about flooding in a floodplain, please check
Abstract L26–28: add the median values. Qualify “significantly” given the independence issue.
Typos: L39 “coocurring”; L45 “technics”
L43 Replace the word “bloomed” with increased or other
L134–135 could be condensed.