Investigating microplastic dynamics in soils: Orientation for sampling strategies and sample pre‐procession

Studies on microplastics in soils is currently being established as a new research field. So far, mainly 'explorative studies' have been carried out to detect microplastics in different soil environments. To generate a deeper understanding of microplastics dynamics, 'systematic studies' are required. Such research must built on a targeted sampling strategy and considerate fieldwork and sample handling. From literature enquiry, a five‐stage methodological workflow was deduced for studies on microplastics in soils. In the present review, the spatial representation of soils/soilscapes with microplastics in soils research is conceptually and practically assessed. We discuss judgmental, randomized, and metric soil sampling strategies. Then, we explain sample pre‐processing and give a brief overview of methods for microplastics identification and quantification. We conclude that the establishment of the novel field of research 'microplastic dynamics in soils' requires more intensive consideration of soil sampling strategies. As soil is a complex medium and the soilscape is spatially heterogeneous, we highlight systematic sampling strategies as the best possible options for sophisticated research. However, no overall optimum methodology can be defined because the specific strategy must be in line with the particular research question. For all studies on microplastics in soils, practical improvement is needed to prevent contamination of soil samples with plastics during sampling and sample pre‐processing.

plastics can spread farther in the landscape than previously assumed.
Surprisingly, it is a relatively novel finding that plastic occurs or is Zhang . With regard to polymer types, studies were only able to identify some of the commonly produced plastics, because methods for the analysis of microplastics in soils are in their infancy (Table 1; M. Liu et al., 2018;Piehl et al., 2018;Scheurer & Bigalke, 2018).
Methodological advances are needed to evaluate the presently unknown environmental effects of microplastics in soils. Impacts on soil organisms are possible, including nanoplastics-uptake by plants resulting in entry into the food chain (Huerta Lwanga et al., 2017;Rillig et al., 2019). To evaluate and restrict negative impacts of microplastics in soils, a better understanding of the microplastic-related processes is required (e.g., transport routes and vectors). This can hardly be achieved by 'explorative studies' but requires 'systematic studies' focusing on microplastics' integration in and interaction with their spatial surroundings (i.e., landscape/soilscape). To conduct such studies, we still need to develop methodological foundations for precise and internationally comparable sampling, microplastic detection, and quantification.
Several authors have already dealt with microplastics in soils through reviews (Table 2). However, these publications largely focused on background information on environmental microplastics pollution or on the procedures of microplastics identification/quantification. By contrast, the establishment of adequate sampling strategies, soil sampling, and sample pre-procession has largely been neglected. These aspects are different for soil-related studies than for studies on microplastics in waters. Furthermore, they are crucial if one wants to conduct systematic research on microplastics dynamics in soils. Hence, the present review aims to make three contributions: (a) to differentiate conceptually between explorative and systematic studies to enable the establishment of research on microplastic dynamics in soils; (b) to elaborate on strategies for creating adequate spatial representation in the empirical designs of studies on microplastics in soils; and (c) to critically discuss related sample handling and pre-procession.
From our literature enquiry, a five-stage workflow was deduced for studies on microplastics in soils, which is reflected in the present review's structure ( Figure 2). With regard to the focus of the other topic-related reviews, we deal in particular with the so far underrepresented Stages 1-3 of the Workflow. Still, a short overview of analytical and quantification procedures is given, in combination with a reference list, which might lead the interested readers to further information.

| A GEOSPATIAL APPROACH TO MICROPLASTIC DYNAMICS IN SOILS
To understand microplastic dynamics in soils from a system perspective, we must consider the spatial contexts of microplastics in soils.
First of all, this requires developing a suitable strategy for study site selection and the sampling procedure, in line with the respective research question. Such spatial considerations are significant for investigating possible displacement, transport routes, or environmental risks of microplastics in soils. F I G U R E 2 Five-stage methodological workflow for studies on microplastics dynamics in soils, as derived from literature enquiry (MP = microplastics). (Stages 1-3 are detailed further in the respective sections of the present paper. For Stages 4 and 5, an overview is given) Hence, it is a challenge for microplastic research to generate a process for understanding microplastic dynamics in soils. As a possible solution, we propose a geospatial approach to microplastic dynamics in soils. As microplastic dynamics lead to a certain spatial distribution of microplastics in soil profiles and soilscapes, we can study this spatial distribution in the field and deduce on the processes that formed this distribution. This was proposed by Weihrauch (2019) to investigate soil phosphorus dynamics. In the present paper, we transfer this 'geospatial approach' to microplastic research.

| Developing a sampling strategy
A systematic investigation of the spatial microplastic distribution in soils requires an adequate spatial resolution. Such is achieved when the soil samples sufficiently represent the investigated two-or three-dimensional spatial unit (e.g., a surface area or a soil profile). Spatial (1) Sampling sites might be positioned in the landscape according to (a) subjective interpretation (judgmental sampling), (b) spatial randomization, or (c) metric criteria. (a) If sampling sites are chosen on the basis of interpretation (e.g., background knowledge, visual evaluation of the landscape), they often occupy particular locations related to specific research questions or hypotheses (e.g., depressions, roadside areas as potential zones of microplastic accumulation ;Möller et al., 2020;Wells, 2010). The correctness of the results generated at these locations thus strongly depends on the correctness of the underlying hypotheses plus the validity of the spatial interpretation.
Due to the complexity of the soilscape, the latter might introduce significant bias into interpretative soil sampling. Moreover, on the basis F I G U R E 3 Aspects to be considered for developing a suitable sampling strategy for studies on microplastics dynamics in soils. (a) Soil sampling-based spatial resolution/representation of study areas as a function of sample site number and sample site distribution (triangles mark sample sites). (b) Possible lateral contexts of sampling sites. (c) Possible vertical contexts of samples within one sampling site. Vertical sampling resolution significantly depends on whether one or several samples are taken per sampling site. (Here only shown for mixed sampling from a larger vertical soil section combined into one sample; e.g., 0-20 cm. Alternatively, local samples can be extracted, which relate to a specific soil depth; e.g., 25 cm) of spatially specific hypotheses, one might probably only be able to support or reject the assumptions but not to make systematic unexpected findings.
(b) Randomized sampling means the distribution of several sampling sites within a defined (i.e., limited) area under the premise that all sites have equal opportunity to be selected and that they are selected independently from each other (Möller et al., 2020;Wells, 2010). The positions of the sampling sites depend on a chosen area (mostly on the basis of interpretation or landscape evaluation, e.g., land use) but not on site or soil features of the concrete sites.
The respective study area is thus treated as homogeneous with regard to soil and site features. This is conceptually critical especially because microplastics distribution is unlikely to be homogeneous (Möller et al., 2020). Anyways, this form of spatial generalization might be sufficient for studies, which aim at results representative for certain areas or landscape sections (e.g., comparison of microplastic pollution of two agricultural fields). It might also be plausible when certain statistical tests are planned (e.g., correlation analyses) as the resulting statistical sample will consist of independent data (Wells, 2010). However, randomized sampling is rather inadequate for studies on microplastic dynamics in soils as it ignores the highly relevant landscape and soilscape particularities.
(c) Metric sampling means the positioning of sampling sites on the basis of distances. This type of sampling also ignores particular site and landscape features. It might be useful to generate a good representation of an area or landscape section unbiased by interpretation or subjective landscape evaluation. By contrast to randomized sampling, it is also favourable for the comparison of different areas or landscape sections unbiased by divergent metric dimensions. It is thus favourable for studying and comparing gradients or spatial patterns (e.g., of increasing microplastic accumulation, microplastic hotspots).
However, metric sampling might lead to the integration of uninteresting sites. Moreover, the resulting statistical sample would not consist of independent data. Thus, certain statistical procedures would not be available for data evaluation (Wells, 2010). In consequence, it strongly depends on the research question and the desired form of data evaluation which type of empirical design should be chosen for positioning the sampling sites.
(2) Another important aspect is the number of soil samples. Generally, larger sample numbers lead to higher spatial resolution (i.e., better spatial representation). The best possible spatial representation is achieved when all sampling sites are at an equal distance from each other, that is, when the not investigated spaces in between are smallest. Hence, a sampling strategy must be designed not only just according to the research question but also according to the size of the investigated spatial unit. One option to achieve adequate spatial representation of a particular study area could be the prior calculation of the sample number required for the specific research question and the planned statistical tests on the basis of pilot-sampling and geostatistical analyses (Li, 2019;Li et al., 2020). However, to date, such strategies have not been transferred to research on microplastic dynamics in soils, probably because analytical methods are still very costly. Thus, logistical aspects (e.g., costs, site accessibility) also have to be considered as they would, in most cases, probably decrease spatial resolution.
The first studies on microplastics in soils (Fuller & Gautam, 2016;Huerta Lwanga et al., 2017) did not report the spatial context of the investigated soils. The more recent studies document the spatial sample contexts but not systematically (Corradini et al., 2019;Piehl et al., 2018;Scheurer & Bigalke, 2018;G. S. Zhang & Liu, 2018; Table 2). Hence, as a reader, it is difficult to evaluate if and how well the examined soils represent the respective study areas. Furthermore, it is evident from publications on microplastics in soils that often relatively small sample numbers are processed because the respective analyses are rather cumbersome (see following chapters). For instance, Scheurer and Bigalke (2018)  (3) To develop a suitable sampling strategy, two dimensions of spatial resolution should be considered, the lateral (i.e., soilscape) and the vertical (i.e., soil profile; Weihrauch, 2019). Soils result from and are shaped by pedogenic and environmental processes, which do not just affect one location (i.e., one spot in coordinate space) but larger spatial areas (e.g., landscape sections). Hence, soils are no isolated phenomena but are parts of soilscapes and must be understood in their landscape context (e.g., slope position). This can hardly be achieved, when one single soil is investigated. For instance, the effect of erosion on soil formation cannot be elucidated from one soil on the topslope alone. Several soil profiles along the slope would be required that have a lateral context with each other.
Weihrauch (2019) proposes three options for the lateral context of soil sampling sites: (a) no lateral context (i.e., samples are taken randomly and not interpreted in a genetic context with each other), (b) a linear context (e.g., in transects/catenae), and (c) a two-dimensional context (e.g., mapping of a surface area; Figure 3b). Of the seven published field studies on microplastics in soils, five are based on random sampling, one on transects, and one on area mapping (Table 3). A trend shows for researchers to favour sampling without lateral context. This might suffice for explorative studies. However, systematic studies should rather be based on a linear or two-dimensional lateral sampling site context (according to the research question).
To depict vertically oriented processes (e.g., related to soil water movement or soil horizons), sampling sites need to be investigated by more than just one sample per site. The vertical resolution increases with the vertical sample number per site. The first study on microplastics in soils did not report the vertical sampling context (Fuller & Gautam, 2016). From the other studies, three are based on one sample and three studies are based on two samples per sampling site (Table 2). Hence, the vertical representation of soil profiles has generally been rather poor so far.
There are three options for the vertical context of soil samples within one site (Figure 3c). (a) Samples can be taken randomly, without regard of their depth or pedogenic background (e.g., no attribution to a certain soil horizon). (b) Samples can be taken according to Not reported Fuller and Gautam (2016) pedogenic characteristics (e.g., soil horizons), either from selected soil sections or from the entire profile (i.e., all soil horizons). (c) Samples can be taken from defined metric depths or depth sections, either for some selected depth sections or for all sections of a profile.
No study on microplastics in soils applied a pedogenic vertical sampling strategy (Table 2) adequately (Weihrauch, 2019). Topsoils are particularly critical to study due to a multitude of land use-based alterations (e.g., on agricultural fields), which might distort data comparison between sites, depths, or even across a plot (Li, Dd, Mendoza, & Heine, 2010 In all previous studies, topsoil samples were taken using steel sampling equipment (  The relatively large amount of sample material is logistically needed due to the further analysis steps, especially with respect to the separation of microplastics from the soil matrix ( Figure 2). Probably, larger amounts of soil material were so far also processed, because it has originally not been known if there were any microplastics in the investigated soils and how they were spatially distributed within the material.
Hence, large samples could enlarge the probability of finding microplastic particles at all or in significant number. Alternatively, pile-driving probes with a larger diameter could be used, which would make sampling more time-consuming and cost-intensive.
It is recommended to pay attention to the contamination by plastic components (e.g., plastic caps on pile-driving probes, splinters of plastic hammers) when using drilling equipment. Many of these components can be removed before sampling. Certain textiles might also pose a risk of contamination (e.g., clothing fibers). Contamination can be avoided by reducing contact (quick storage after sampling) or by wearing cotton clothes.
Plastic particles may not just detach from our equipment during sampling but also during transport, and might artificially enrich our Finally, the samples have to be stored until the next methological stage or for longer time. Samples should always be stored in closed containers to prevent contamination by the ambient air (e.g., dusts). If the samples should be stored in plastic containers such as PE bags or PET jars, they should be stored as dry, cool, and dark as possible to prevent potential degradation of the plastics (Napper & Thompson, 2019). In the case of biodegradable bags, it must be examined whether long-term storage is possible. In general, the polymer type of all plastic containers used should be known, the samples should be handled carefully and blank samples should be used as a control to verify any contaminations.

| Sample pre-processing
A main challenge in the analysis of microplastics from environmental media is the separation of microplastics from their respective medium (e.g., water, soil material). The analysis of water samples can usually be carried out by sieving and filtration methods. For soils and sediments, the separation of microplastics from other matter is required.
Various methodological approaches for the analysis of marine, aquatic, and limnic sediments were recently transferred to extract microplastics from soil samples (Bläsing & Amelung, 2018;He et al., 2018;. However, it should be considered if a simple (e.g., beach sands) or a complex sample matrix (e.g., soil material) is at hand (Figure 2). As a result of pedogenesis, the biogeochemical properties of soil samples differ from, for example, beach sediments and form a more heterogeneous sample matrix comprised of several different components (e.g., mineral, organic, and microplastic particles).
In soil science, samples are usually dried before analysis because the results (e.g., heavy metal contents) are mostly reported in relation to soil weight (e.g., in SI of soil). For many questions, it is plausible to use the soil's dry weight. The weight of moist soil is largely influenced by weather conditions (e.g., precipitation). Thus, it gives different results with regard to the timing of soil sampling. Such results would be relative: For instance, element concentrations would appear higher in dry times (because it is related to lower dry weight) than in moist times (because it is related to higher moist weight). To come to absolute (i.e., generalizable) results, the soil moisture is thus eliminated by drying. This is done by air-drying at room temperature or in drying fur- In addition to manual sample homogenization by manually crushing the aggregates with pestle and mortar, it is also possible to use ultrasound techniques, which might be necessary for strongly aggregated soils with high clay contents (Piehl et al., 2018

| Sample matrix separation
In contrast to microplastic analyses in water and sediment samples, processing soil material faces the challenge of separating several components. Presently, no satisfactory methods exist for this purpose. As sample separation can hardly be achieved in one preparation step, a suitable and standardized workflow is required. Three steps of microplastic extraction can be differentiated, which can be applied individually, partially, or consecutively: (a) the removal of the mineral phase, (b) the removal of organics, and (c) the size classification of microplastic particles by sorting and/or sieving. It is crucial to make sure that soil samples are not contaminated with external plastic (clothing fibres, plastic equipment) at any stage of this workflow.

| Removal of the mineral phase
In most studies on microplastics in soils, the principle of density separation is used to remove the mineral phase from the pre-processed samples.
The separation of the heavier mineral components (i.e., sand, silt, finally clay), that sink to the bottom, from the lighter components A new approach is the application of castor oil for separation.
The method was tested with recovery rates of 99 ± 4% for PP, PS, PMMA, and PET, but samples with a high content of organic material require additional treatment for organic matter decomposition A direct separation of microplastics from the mineral and organic soil particles presently only exists for the smallest microplastic particles (<30 μm). For this purpose, a pressurized fluid extraction with methanol and dichloromethane is used (Fuller & Gautam, 2016). However, this method can only be applied for specific research questions.

| Removal of organics
Organics (e.g., humus, peat particles, and plant/root fragments) mostly have a density comparable to microplastics. Thus, density separation usually results in organic components being extracted along with microplastics (Corradini et al., 2019;Felsing et al., 2018;G. S. Zhang et al., 2019). In order to isolate the microplastic particles, they can be removed manually using a stereomicroscope with respect to the detection limits/magnification of the microscope (Crawford & Quinn, 2017;Song et al., 2015). Alternatively, organics and microplastics can be separated technically.
As in marine and aquatic research, enzymatic digestion may be applied using a variety of enzymes in combination with a subsequent H 2 O 2 treatment (Löder, Kuczera, Mintenig, Lorenz, & Gerdts, 2015;Mintenig, Int-Veen, Löder, Primpke, & Gerdts, 2017). During this procedure, plastic is not degraded. However, the method is timeconsuming and still has to be tested for the successful application to soil organic matter (Bläsing & Amelung, 2018 (4:1)]. With all acid treatments, a rapid removal of the organic components is observed (Enders, Lenz, Beer, & Stedmon, 2016). However, structural degradation of plastic particles was often observed.
In contrast to acid treatments, alkaline treatments influence neither the microplastic particles shape nor surface properties (Enders et al., 2016). The treatments apply NaOH, KOH, or both in combination. Both chemicals are suitable for biological samples but have not been applied to soil samples. Although these treatments do not degrade plastics, the methods are unable to remove alkali-insoluble organic matter from soil (Bläsing & Amelung, 2018). Therefore, humins probably remain in the samples after the alkaline treatment, which complicates the later identification of the microplastic particles (Dehaut et al., 2016).
Furthermore, the oxidation of humus is sometimes applied to remove organics (e.g., 30% H 2 O 2 ; He et al., 2018;Scheurer & Bigalke, 2018 Removing organics with just H 2 O 2 causes a digestion of PE and PP (Silva et al., 2018). No negative effects are reported for the application of the Fenton reagent.
Next to the enzymatic and chemical treatments, a separation based on the different electrostatic behaviour of organic and plastic particles was developed (Felsing et al., 2018;Hidalgo-Ruz, Gutow, Thompson, & Thiel, 2012). This method has no negative effect on microplastic particles. However, it was tested only for various sands and sediments, where it gave reliable results in the separation of organic material. The method should be validated for the application to soil samples (Felsing et al., 2018).

| Sieving and sorting of separated microplastic
After microplastics have been separated from the mineral and organic matrix, it is useful to characterize the plastic particles further according to size by sieving and/or sorting. According to the above-  Zhang & Liu, 2018). For particles <1 mm, a filtration with various pump systems (e.g., vacuum) and glass fibre filters is often used (Klein, Worch, & Knepper, 2015;Scheurer & Bigalke, 2018).
Sieving can be conducted after drying and crushing, or intermediately during analysis (i.e., after density separation as dry or wet sieving

| Microplastic quantification and identification
The of the low density and small particle sizes (<500-300 μm). Moreover, the selection of the unit to report the results might depend on the research question. For studies on the general occurrence and abundance of microplastics in the environment, a mass-related specification (e.g., mg kg -1 ) seems to be sufficient and plausible. Instead, when effects on soil functions, relocation processes, or modeling are in focus, informations on particle number (e.g., mpp/kg), size, shape, and type become relevant. For approaches such as simplifying the complex diversity of MP particles through a three-dimensional dimension, also size, density, and shape of each particle is required (Kooi & Koelmans, 2019).
The increasing number of samples and possibly high numbers of microplastic particles in soils lead to large amounts of data, which can be processed in an automated way (image analyses or automatic spectral analysis with databases; Primpke et al., 2017;Primpke et al., 2018). The application of spectroscopic methods for the analysis of plastics has already been discussed in other soil specific reviews as well as reviews of the material sciences and in marine research (Elert et al., 2017;He et al., 2018;Pinto da Costa et al., 2019;Prata et al., 2019;Renner et al., 2018;Ruggero et al., 2020;Silva et al., 2018;B. Zhang et al., 2020) and is not further elaborated in the present review.
The methods commonly applied for soil scientific studies have in common that the heterogeneous soil sample matrices require a more or less complex sample preparation or separation. First approaches to reduce this effort are the pre-scanning of the sample without chemical treatment, based on near-infrared spectroscopy (NIRS) detection (Paul, Wander, Becker, Goedecke, & Braun, 2019). Moreover, a direct quantification of heterogeneous sample matrices by the combination of thermogravimetric analyses (TGA) with thermal desorption system coupled with gas chromatography-mass spectrometry (TDS-GS-MS) and the application of twisters as solid-phase absorbers was demonstrated for PE particles (Dümichen et al., 2015(Dümichen et al., , 2017. Despite the diversity of identification procedures and their different applicability to soil samples, there is still a large demand for research to validate, improve, and develop suitable, comparable, cost-and time-efficient methods-particularly regarding pre-scanning methods.

| CONCLUSIONS
The discovery of microplastic particles as new pollutants in the environment opens up a novel field of research for soil science. Potential hazards posed by microplastics and nanoplastics in soils (e.g., uptake by plants and introduction into the food chain) are theoretically plausible. However, a better understanding of microplastic dynamics is needed to systematically evaluate the effects of soilbound microplastics pollution (e.g., on biota and the food chain) and to develop targeted mitigation strategies. Regarding the current trends in environmental microplastics research, we think that it is specifically required to transition from solely explorative microplastics studies to more systematic investigations. This would especially call for a more intensive consideration of spatially adequate sampling strategies than documented in previous studies. The proposed geospatial approach might thus enable further more sophisticated research.

ACKNOWLEDGMENTS
This study was financially supported by the Hessian Agency for Nature Conservation, Environment and Geology (HLNUG), and the Hessian Environmental Administration. Open access funding enabled and organized by Projekt DEAL.

CONFLICTS OF INTEREST
The authors state that no conflicts of interest exist in association with this publication.