Microplastics negatively affect soil fauna but stimulate microbial activity: insights from a field-based microplastic addition experiment

Abstract

Microplastics are recognized as an emerging contaminant worldwide. Although microplastics have been shown to strongly affect organisms in aquatic environments, less is known about whether and how microplastics can affect different taxa within a soil community, and it is unclear whether these effects can cascade through soil food webs. By conducting a microplastic manipulation experiment, i.e. adding low-density polyethylene fragments in the field, we found that microplastic addition significantly affected the composition and abundance of microarthropod and nematode communities. Contrary to soil fauna, we found only small effects of microplastics on the biomass and structure of soil microbial communities. Nevertheless, structural equation modelling revealed that the effects of microplastics strongly cascade through the soil food webs, leading to the modification of microbial functioning with further potential consequences on soil carbon and nutrient cycling. Our results highlight that taking into account the effects of microplastics at different trophic levels is important to elucidate the mechanisms underlying the ecological impacts of microplastic pollution on soil functioning.

1. Introduction

From 1950 to 2015, approximately 8300 million metric tons of plastic have been produced and have generated about 6300 million metric tons of plastic waste, 79% of which has accumulated in landfills or leaked into the natural environment [1]. Owing to the long-term persistence of plastics in the environment, their accumulation is one of the most widespread and long-lasting anthropogenic changes to the Earth's surface. Once in the environment, large plastic debris gradually degrades into smaller pieces through weathering and other disintegration processes [2–5], which can further increase their dispersion and incorporation into the soil [6–8]. Microplastics are generally defined as plastic particles with diameters less than 5 mm [9]. Given their increasing abundance in the environment, microplastics are increasingly recognized as human-caused pollutants and are now considered as an emerging factor of global change [9–11]. Yet, although an increasing body of literature has shown that the distribution and ecological impacts of microplastics have considerably increased at large spatial scales [12,13], most studies were carried out in aquatic ecosystems and research on microplastics is still particularly limited in terrestrial ecosystems [14–16].

Given that plastics are produced, used and mostly discarded on land, soils represent a large sink for microplastics [4,9,14,15]. For instance, microplastics have been found from high to low latitudes across a variety of terrestrial ecosystems such as agricultural lands, urban and industrial areas, and even in remote mountains [8,17–19]. Recent studies using contaminated soils have demonstrated that pollution by microplastics can alter the coupling between carbon and nutrient cycling through significant increases in nutrients in dissolved organic matter and CO₂ fluxes [20–22]. These findings have triggered increasing concern about how and to what extent microplastics could impair numerous ecosystem processes mediated by soil organisms, such as organic matter decomposition and nutrient cycling [11,23,24]. Yet, little is known if these effects are owing to changes in the structure and function of soil communities at different trophic levels.

Recent studies have reported harmful effects of microplastics on various groups of soil fauna such as earthworms, snails, collembolans and nematodes [25–30]. This is because microplastics can be ingested and induce toxic effects on some groups of soil fauna [25], disturb the symbiotic microbiota in the soil fauna gut [26,27] or inhibit movement of soil microarthropods by filling soil pores [31]. However, the impacts of microplastics on soil microbial communities have led to inconsistent results. For instance, laboratory-based incubations have reported that the addition of microplastic fragments or polyester fibres can have positive or negative effects on soil microbial activity [7,32–34]. This is probably because changes in soil physico-chemical properties such as soil structure, bulk density or water and nutrient dynamics can have different effects on the composition of microbial communities according to the environmental context or the type of microplastic added [7,23,24,32]. Furthermore, most studies investigating the effects of microplastics on soil biota have focused only on a single species or trophic group (e.g. [26,27,29,33,34]). Yet, soil organisms interact within complex food webs in nature, and changes in one or several groups of soil organisms can have further consequences on the abundance, diversity and functioning of other groups in the food web [35–38]. For instance, high-intensity grazing by soil fauna often reduces microbial biomass, whereas low-intensity grazing can have stimulatory effects on microbial activity [36,39,40]. Therefore, because the effect of microplastics within and across trophic levels still remains unclear in soil food webs, this may hinder our ability to develop a conceptual framework to predict the ecological risk of microplastic contaminations in terrestrial biogeochemical cycling (but see [24]).

In this study, our main objective was to assess whether and how microplastic pollution can modify the structure and functioning of soil biota by manipulating microplastics at different concentrations. Specifically, we studied changes in both abundance/biomass and community structures of soil microarthropods, nematodes and microorganisms (i.e. bacteria and fungi), and quantified whether soil microbial functioning (i.e. enzyme activities) changed directly and/or indirectly owing to the microplastic pollution. Because many laboratory-based studies have shown that microplastics can have harmful effects on soil invertebrates (e.g. [25–30]), we hypothesized that addition of microplastics should result in a reduction in the abundance of soil fauna (i.e. microarthropods and nematodes) (H_1a). Given that the effects of microplastics may differ according to the quantity added [25], we expected greater negative effects on soil fauna at higher concentrations of microplastics (H_1b). Then, because the effects of microplastics on soil microorganisms have previously been inconsistent [7,32–34], we evaluated the ‘direct' and ‘indirect' effects of microplastic addition on soil microbial activity by stipulating four alternative hypotheses. First, we hypothesized that the addition of microplastics would result in positive effects on microbial activity, either ‘directly' because microplastics may be used as a source of carbon and modify soil physico-chemical properties (H_2a) [7,20,21,24], or ‘indirectly' because microplastics may reduce the negative effects of high-intensity grazing of microarthropods on soil microbial communities (H_2b) [36,40]. Alternatively, we hypothesized that the addition of microplastics would decrease soil microbial activity, either ‘directly' because microplastics can increase soil toxicity (H_2c) [32,41], or ‘indirectly' because microplastics may reduce the low-intensity grazing by microbivores, thereby weakening the stimulatory effects of microbivores on soil microbial communities (H_2d) [39,42].

2. Materials and methods

(a). Experimental design

A field microplastic addition experiment with a randomized block design (four treatments, replicated in six blocks) was conducted in a flat area in Jinfoshan, Chongqing, China (107.174 E, 29.026 N; 1450 m elevation). The vegetation type is subtropical evergreen broad-leaved forest. This area is remote with low human activity and thus presents very low quantities of microplastic contaminants prior to the experiment. The soil type is a loamy sand soil, with concentrations of organic carbon of 189.19 ± 15.71 mg g⁻¹, nitrogen of 6.8 ± 0.31 mg g⁻¹ and phosphorus of 0.58 ± 0.08 mg g⁻¹.

Microplastics used in this study were low-density polyethylene fragments (small plastic fragments which were purchased from Shanghai Youngling Electromechanical Technology Co., Ltd., Shanghai, China), which are one of the most abundant types of plastics produced worldwide [1]. The microplastic identity and purity were analysed with micro-Fourier transform infrared spectroscopy (electronic supplementary material, figure S1a). Fragment size distribution, measured with a laser particle size analyser revealed a unimodal size distribution (electronic supplementary material, figure S1b), with a volume-weighted mean diameter of 37.13 µm (range from 0.3 to 400 µm, the 10th and 90th percentile were 10.99 µm and 68.23 µm, respectively). Fragment shape was observed with a stereomicroscope (electronic supplementary material, figure S1c).

Treatments consisted of control (no microplastic addition) and three levels of microplastic addition (i.e. +5, +10 and +15 g m⁻²), corresponding to a density of 0, 11,361 ± 354, 23 789 ± 743 and 39 172 ± 1,014 fragments kg⁻¹ of dry topsoil (0–3 cm) at the end of the experiment. These concentrations are within the range of the microplastic density that was previously reported for field soils [18,43], thus representing environmentally realistic concentrations. We refer to the three levels addition as A₅, A₁₀ and A₁₅ hereafter. Each plot was round in shape with a diameter of 1.0 m. Before microplastic addition, we loosened the topsoil (0–3 cm) using an iron-toothed rake after removing recently fallen and partly decomposed litter. To ensure a homogenous addition of microplastics, the microplastic fragments of each treatment level were mixed thoroughly with 1000 ml of soil collected near the study site. The soil-microplastic mixture was then evenly disseminated in each plot. For the control treatment, 1000 ml of soil without microplastics was also evenly disseminated. Finally, the litter was put back on the plots after the microplastic addition.

We harvested soil samples from all plots 287 days after microplastic addition. Five soil samples were collected within each plot by using a steel cylinder (10 cm diameter, 3 cm depth). To minimize potential edge effects, sample collections were confined to the central circular area (0.7 m diameter) within each plot. The five soil cores from each plot were then pooled, placed in a cooler containing ice packets and transported to the laboratory within 3 h. Each sample was homogenized and subsampled for microbial community and enzyme analyses (frozen at −20°C), nematodes (refrigerated at 4°C) and microarthropods extraction and gravimetric water content measurement (kept fresh at 20°C).

(b). Soil fauna extraction and counting

Within 3 days after sampling, microarthropods were extracted from a 200 g (fresh mass) soil subsample of each plot using Tullgren funnels (Burkard Scientific, UK) for 72 h and preserved in 75% ethanol before identification. All specimens were counted and identified to order level under a stereomicroscope. Within 1 day after soil collection, nematodes were extracted from a 100 g (fresh mass) soil subsample of each plot using the sugar centrifugation method [44] and then heat-killed, preserved in 4% formalin and stored in 4°C. Nematodes were counted and identified to five different trophic groups (bacterivores, fungivores, plant feeders, omnivores and predators) based on mouthpart morphology under a microscope with 200× or 400× magnification [45]. Subsamples of the soils were weighed, dried at 105°C for 48 h and then weighed to determine the moisture content. All of the fauna abundances were expressed as number of individuals per kg of dry soil.

(c). Microbial community and extracellular enzyme activities

Microbial biomass carbon and nitrogen was determined by the chloroform fumigation-extraction method [46]. Briefly, for each soil sample, one 10 g subsample was fumigated for 24 h with chloroform vapour, while another was not. Fumigated and non-fumigated samples were extracted by vigorous shaking in 0.5 M K₂SO₄ for 1 h. The carbon and nitrogen present in the extracts were quantified using a total organic carbon/total nitrogen analyser (Shimadzu Corporation, Kyoto, Japan). Microbial carbon and nitrogen were finally calculated as (total carbon or nitrogen in fumigated soil—total carbon or nitrogen in un-fumigated soil)/0.45 [47]. Microbial community composition was determined using phospholipid fatty acids (PLFAs) following the method described by Bossio & Scow [48]. Briefly, total PLFAs were extracted from 3 g of each freeze-dried soil subsample. The resulting PLFAs were analysed on a gas chromatograph equipped with a flame ionization detector (Agilent 6890, Agilent Technologies, Palo Alto, USA) and identified using the MIDI Sherlock microbial identification system (MIDI Inc., Newark DE). Only PLFAs with a carbon chain length greater than 14 and less than or equal to 20, identifiable and present at greater than 0.5 mol % were included for further analyses. The sum of all PLFAs was used to represent total PLFAs, the sum of i14:0, i15:0, a15:0, 15:0, i16:0, 16:1ω9, 16:1ω7c, 10me16:0, i17:0, a17:0, cy17:0, 17:0, 10me17:0, 18:1ω7, 10me18:0 and cy19:0 was used to represent bacterial PLFAs, and the sum of 18:2ω6 and 18:1ω9 was used to represent fungal PLFAs [49]. Microbial functioning was evaluated by determining the potential activity of six hydrolytic enzymes involved in carbon (i.e. α-1, 4-glucosidase (AG), β-1, 4-glucosidase (BG), β-D-cellobiohydrolase (CBH) and β-xylosidase (XYL)), nitrogen (i.e. l-leucine aminopeptidase (LAP)) and phosphorus (i.e. acid phosphatase (AP)) cycling, by employing the method described by Bell et al. [50]. Briefly, 2.75 g of soil was blended in a sodium acetate buffer (pH = 5) for 1 min, incubated in a 96-deep-well microplate for 3 h at 25°C with the appropriate fluorescently labelled substrate, and the fluorescence was measured at an excitation wavelength of 365 nm excitation and 450 nm emission wavelength, using a fluorescence microplate reader (BioTek Synergy LX, BioTek Instruments Inc., VT, USA). The activities were expressed as nmol h⁻¹ g⁻¹ dry soil.

(d). Statistical analyses

Prior to the main analyses, we tested the effect of the block on the abundance of microarthropods (separately for the six most common groups), abundance of nematodes (separately for the total and five different trophic groups), microbial properties (separately for the biomass carbon and nitrogen, total, fungal and bacterial PLFAs, and fungal/bacterial PLFAs ratio (F/B ratio)) and soil enzyme activities (separately for the six hydrolytic enzymes). As the effect of the block was significant for some variables (i.e. total and fungal PLFAs, F/B ratio, activities of CBH, XYL and AP), we included the term ‘block' as a random effect in further analyses. Significant differences in aforementioned variables between the control and each level of microplastic addition were assessed using linear mixed-effects models with ‘treatment' as a fixed factor and ‘block' as a random factor. We considered statistical significance at p < 0.05 for all statistical tests. Percentage of change (%) in each level of microplastic addition treatment relative to the control treatment in the aforementioned variables was calculated separately for each block as (microplastic addition plot – control plot)/control plot × 100. Differences in the composition of microarthropod communities (order level), nematode trophic groups and microbial communities (assessed by PLFAs) after microplastic additions were visualized using principal coordinate analysis (PCoA) based on Bray–Curtis dissimilarity matrices. Permutational multivariate analysis of variance (PERMANOVA, n = 9999 permutations) were then used to assess whether the communities were different among treatments as described previously, i.e. with ‘treatment' as a fixed factor and ‘block' as a random factor.

Finally, structural equation modelling (SEM) was conducted to examine the direct and indirect effects of microplastic additions on various groups of soil organisms and enzyme activities. We used a priori causal model stipulating that the microplastic additions influence microbial functioning directly via microplastic additions and indirectly by decreasing the abundances or shifting the community structures of microarthropods and nematodes (and therefore by altering the grazing effects on microbial communities; electronic supplementary material, figure S2). Microarthropod abundance and microbial biomass data were log-transformed to improve normality. A covariance path between microarthropod community structure and abundance was included in the model owing to the strong relationship between these variables (Pearson's r = −0.81, p < 0.001). We chose a piecewise approach for the SEM because it can accommodate relatively small sample sizes while allowing the inclusion of random effects [51]. Goodness-of-fit of the model was assessed using Shipley's test of directional separation, which yields a χ²-distributed Fisher's C statistic [51]. In order to improve parsimoniousness of the model, a stepwise removal of the least significant paths from the models was conducted manually until there was no remaining path with p > 0.05. All statistical analyses were done using R version 3.6.1 with R packages ‘nlme', ‘vegan' and ‘piecewiseSEM'.

3. Results

(a). Microarthropod response

Overall, the high-density level of microplastic addition (A₁₅) had a negative impact on four of the six microarthropod groups investigated in this study (figure 1). Specifically, we found that the A₁₅ treatment significantly decreased the abundance of oribatid mites (−15.3 ± 5.7%; figure 1a), Dipteran larvae (−30.5 ± 9.3%; figure 1d), Lepidopteran larvae (−41.5 ± 12.2%; figure 1e) and Hymenoptera (ants; −62.5 ± 7.5%; figure 1f) in comparison to the control plots. Although the trends were similar for the low-density levels of microplastics, no significant effect was found (figure 1a–f). The effects of microplastic additions on the abundance of collembolans and non-oribatid mites were non-significant, regardless of the level of microplastic considered (figure 1b,c). The community structure of microarthropods (order level) was also significantly affected by the microplastic treatments (PERMANOVA: F_1,22 = 2.98, R² = 0.12, p = 0.002), with a gradient from control plots to A₁₅ along the second PCoA axis (figure 2a).

Figure 1.

Effects of in situ microplastic addition treatments on microarthropod abundance (a–f), nematode abundance (g–l), microbial properties (m–r), and enzyme activities (s–x) as indicated by the percentage change in microplastic addition treatments (A₅: +5 g m⁻², A₁₀: +10 g m⁻² and A₁₅: +15 g m⁻²) relative to the control plots. Bars represent means ± s.e. (n = 6), and individual data points are shown as black opaque circles. Symbols above the bars indicate the statistical difference between the different level of microplastic addition treatment and the control treatment, which were tested using absolute values (***p ≤ 0.001, **p ≤ 0.01, *p ≤ 0.05, n.s. = non-significant). The corresponding mixed linear effect models are presented in the electronic supplementary material, table S1. (Online version in colour.)

Figure 2.

PCoA plots showing the effects of microplastic addition treatments on (a) soil microarthropod community structure, (b) nematode trophic structure, and (c) microbial community structure. Permutational multivariate analysis of variance (PERMANOVA, n = 9999 permutations) results are shown in the top right corner of each plot. (Online version in colour.)

(b). Nematode response

Total abundance of nematodes was significantly affected by microplastic additions: it was significantly reduced in A₅, A₁₀ and A₁₅ treatments by −15.4 ± 5.9%, −18.2 ± 4.3% and −19.7 ± 3.4%, respectively (figure 1g). However, different nematode trophic groups exhibited divergent responses to microplastic treatments. The abundance of omnivorous nematodes decreased significantly in all levels of microplastic treatments (figure 1h). The predatory nematodes were negatively affected in the A₁₅ treatment (figure 1i), while the plant-feeding nematodes were negatively impacted in the A₁₀ and A₁₅ treatments (figure 1j). By contrast, fungal- and bacterial-feeding nematodes generally did not respond significantly to microplastic additions (figure 1k,l). The trophic structure of the nematode community was significantly affected by the microplastic additions (PERMANOVA: F_1,22 = 8.29, R² = 0.27, p < 0.001), with a clear gradient from control plots to A₁₅ along the first PCoA axis (figure 2b).

(c). Microbial community response

In contrast with the responses of soil fauna, soil microbial communities were not significantly affected by microplastic additions (figure 1m–r). Specifically, no significant microplastic effects on microbial biomass carbon and nitrogen, as well as on total, bacterial and fungal PLFAs, and F/B ratio were detected (figure 1m–r). Accordingly, we did not observe any change in the structure of the soil microbial community in the PCoA plot (PERMANOVA: F_1,22 = 0.24, R² = 0.01, p = 0.61; figure 2c). By contrast, we found that several enzyme activities significantly increased after microplastic additions. For instance, some enzymes involved in carbon cycling such as AG increased by 85.5 ± 22.7% in A₅ treatment (figure 1s), while BG increased by 36.3 ± 14.0% and 52.7 ± 13.8% in the A₁₀ and A₁₅ treatments, respectively (figure 1t). LAP, which is involved in nitrogen cycling, increased by 116.3 ± 29.9% and 82.8 ± 39.5% in A₅ and A₁₅ treatments, respectively (figure 1w). Similarity, AP, which is related to phosphorus cycling, increased by 81.1 ± 20.5% and 37.8 ± 14.0% in A₅ and A₁₅ treatments, respectively (figure 1x).

(d). Direct and indirect effect of microplastic addition

The most parsimonious SEM adequately fitted the data (global goodness-of-fit: Fisher's C = 14.43, p = 0.81, Second order Akaike information criterion (AICc) = −338.08; figure 3). In this model, microplastic additions directly changed the community structure (β = 0.65) and reduced the abundance (β = −0.75) of microarthropods. It also changed the trophic structure of nematodes, both directly (β = 0.70) and indirectly via the changes in the microarthropod community (β = 0.45) or abundance (β = −0.35), whereas the abundance of nematodes was only directly reduced by microplastic addition (β = −0.57). At the same time, microplastic addition promoted microbial functioning, either directly (β = 0.55) or indirectly mediated by changes in the microarthropod community (positively; β = 0.40) and in the nematode trophic structure (negatively; β = −0.61). By contrast, microplastic addition did not affect directly the microbial biomass, which was only impacted indirectly via the nematode trophic structure (β = 0.49).

Figure 3.

Result of structural equation model describing the direct and indirect effects of microplastic additions on soil biota and microbial functioning (Fisher's C = 14.43, p = 0.81, AICc = −338.08). Microarthropod community structure (figure 2a, PCoA axis 2 variation explained of 20.59%) and nematode trophic structure (figure 2b, PCoA axis 1 variation explained of 41.69%) were based on ordination axes derived from principal coordinate analysis. Microbial biomass was based on the value of microbial biomass carbon. We standardized enzyme measurements by calculating the Z-scores for each enzyme, and then the standardized activities of each enzyme were summed to provide an estimate of total hydrolytic enzyme activity, which was used as a proxy of microbial functioning in the model. Experiment block was included as a random effect. Marginal and conditional R² values ($R_{\mathrm{m}}^2$ and $R_{\mathrm{c}}^2$) represent the proportion of variance explained by the fixed factors alone and by all factors (fixed + random), respectively. Numbers on the arrows indicate standardized path coefficients and asterisks mark their significance: *p < 0.05, **p < 0.01 or ***p < 0.001. Positive and negative effects between nodes are connected by blue and red arrows, covariance between variables is connected by a black double-head arrow, and thickness of the arrows is proportional to the standardized regression or covariance coefficients. (Online version in colour.)

4. Discussion

(a). Soil fauna response to microplastic additions

Although recent studies have shown that microplastics can have harmful effects on soil fauna in laboratory experiments [25–27,29,30], very few studies have evaluated their impact on soil invertebrates in natural conditions. In support of our first hypothesis (H_1a), we found that microplastic additions generally decreased the abundance of soil fauna, including soil microarthropods and nematodes. In addition, as we expected, higher concentrations of microplastics generally had greater negative effects (H_1b), leading to significant changes in the composition of soil fauna.

Oribatid mites are widespread globally and play an important role during organic matter decomposition [45]. The only study that tested the microplastic effects on the oribatid mite Oppia nitens in a laboratory experiment found no harmful effect of added polyester fibres [30]. By contrast, probably because we added microplastic fragments but not fibres, we found that the abundance of oribatid mites at the community level was reduced significantly in the high microplastic treatment (A₁₅). Similarly, we observed a significant decrease in the abundance of Dipteran larvae, Lepidopteran larvae and ants in the high microplastic treatment (A₁₅). Several mechanisms including microplastic ingestion, habitat changed by microplastics and avoidance migration may all explain the reduction in the abundance of these microarthropods after microplastic addition [27]. Despite the fact that negative effects of microplastics on collembolans Folsomia candida have previously been reported in laboratory-based studies [26,27], we did not detect such harmful effects on collembolans at the community level in this experiment. The differential effects of microplastics on soil fauna may depend on foraging strategies and life stages [52]. For instance, soil-ingesting fauna such as Lumbricidae or maggots (e.g. Dipteran larvae) may be more likely to inadvertently ingest microplastics, whereas other taxa such as collembolan or social insects that are picky feeders may exhibit avoidance to microplastics in the soil [27,52]. An increase in the microplastic ingestion may in turn affect morphological traits, juvenile development and reproduction in some [27,53] but not all groups of organisms [30,54]. In addition, different polymer types, sizes, shapes and concentrations of microplastics used, the duration of the experiment and/or the experiment setting (laboratory or field site) may all contribute to the different results found among these different studies [55]. Overall, our results generally show important effects of microplastics on many common groups of soil microarthropods, especially at high concentrations (A₁₅).

Soil nematodes also represent a globally important group of invertebrates that contribute to nutrient cycling and organic material decomposition in soils [45,56]. Our result clearly shows that the abundance of total nematodes was significantly reduced by microplastic additions, probably because small fragments were ingested by nematodes [57–59]. For example, a recent study conducted in agar plates using the model species Caenorhabditis elegans has shown that the ingestion of microplastics can cause the blockage of the digestive tract or exert oxidative damage to intestinal tissues [57]. Furthermore, as nematodes depend heavily on water films or water-filled pore spaces in the soils [60,61], microplastics could also indirectly impact nematode performance by changing soil water dynamics [62].

Interestingly, we also found that different functional groups of nematodes responded differently to microplastic additions, leading to significant changes in the nematode trophic structure along the microplastic addition gradient. Notably, the reduction of total nematodes was mainly owing to declines in root-feeding, omnivorous and predatory nematodes, whereas we observed only little change or even marginal increases in the abundance of fungal- and bacterial-feeding nematodes. Because omnivorous and predatory nematodes have a larger body size than bacterial and fungal feeders [63,64], they are probably more vulnerable to the microplastic pollution owing to direct ingestion, than small-bodied nematodes. This is because the buccal cavity size of nematodes is determining factors influencing the microplastic ingestion by nematodes [58,59]. However, it should be noted that direct experimental evidence showing that omnivorous and/or predatory nematodes ingest microplastics is lacking so far. Meanwhile, the non-significant change in the abundance of bacterial and fungal feeders may be explained by the lack of microplastic effects on their food resource (i.e. non-significant effects of microplastics on microbial biomass) or the reduction of predation pressure (i.e. significant reduction in the abundance of omnivorous and predatory nematodes). However, the ingestion of microplastic particles by plant-feeding nematodes is unlikely to occur because of their special mouthparts (i.e. hollow stylet) [58]. Instead, we propose that microplastic additions lead to changes in plant root traits and/or biomass [23,24], therefore probably affecting the quantity and quality of resources available to plant-feeding nematodes. Overall, the divergent responses of various nematode groups to microplastic pollution suggested that pollution by microplastics will have strong impacts on the composition of nematode communities.

(b). Microbial functioning response to microplastic additions

Surprisingly, the soil microbial biomass (assessed by microbial biomass carbon, nitrogen or PLFAs) and microbial community structure (assessed by PLFAs) did not significantly change in the microplastic addition treatments. Nevertheless, microbial functioning as indicated by the activities of several hydrolytic enzymes involved in carbon, nitrogen and phosphorus acquisition increased significantly after microplastic addition. These findings are in accordance with recent laboratory studies, which reported that microplastics led to a significant stimulation of enzyme activities including catalase, urease and acid phosphatase [33,34]. The SEM revealed a direct positive effect of microplastic additions on microbial functioning, which supports our hypothesis H_2a (i.e. positive direct effect) but contrasts with the hypothesis H_2c (i.e. negative direct effect). Two possibilities could explain this pattern. First, the added microplastic fragments may have changed carbon and nutrient availability in the dissolved organic matter [20,21], thereby affecting the microbial community structure towards copiotrophic organisms that can synthesize high quantities of hydrolytic enzymes [33,34]. This hypothesis is supported by several studies using high-throughput sequencing that have demonstrated significant increases in fast-growing bacterial groups such as Bacteroidetes and Proteobacteria [33,34], which are often regarded as copiotrophic in the literature [65,66]. However, it should be noted that we did not observe significant shifts in the structure of soil microbial communities in our study, presumably owing to the relatively low resolution of the PLFA method, which did not allow us to assess changes at a fine taxonomic resolution compared with high-throughput sequencing. Second, changes in soil physical properties may be responsible for the increase in enzyme activities. For instance, de Souza Machado et al. [7,24] found that microplastic additions could increase water holding capacity in soils, therefore maintaining soil moisture for a longer period of time than control treatments. Greater water availability has often been linked to an increase in enzyme activity [67–69], which may thus explain the positive effect of microplastics on enzyme activities.

In support of our hypotheses regarding the ‘indirect effects' of microplastics on enzyme activities (hypothesis H_2b and H_2d), the SEM result revealed that changes in the microarthropod community and in the nematode trophic structure had significant effects on microbial activity, but in opposite directions. Numerous studies have highlighted the potential of soil fauna to influence microbial functioning, with magnitude and direction of the impacts varying dramatically according to the intensity of grazing [35,36,39,40,42,70,71]. High-intensity grazing by microarthropods often reduces microbial activity owing to a strong negative impact on microbial growth and biomass [36], which is more frequently found for large-sized soil organisms [40,72]. By contrast, low-intensity grazing by small-sized soil organisms commonly leads to stimulatory effects on microbial activity, presumably owing to higher nutrient turnover and compensatory effects of the microbial biomass [35,40], which are frequently found in the interactions between nematodes and microorganisms [42,70,71]. Therefore, the results of our experiment highlight that microplastic pollution probably tended to reduce the high-intensity grazing of the soil microarthropods on microbial activity (i.e. positive indirect effect; H_2b), whereas the reduction of low-intensity grazing by nematodes probably tended to weaken the stimulatory effects on microbial activity (i.e. negative indirect effect; H_2d). Interestingly, our result further underlines that the indirect microplastic effect mediated via the nematode trophic structure (β = −0.61) was stronger than the indirect effect mediated by microarthropods (β = 0.40) and even more than that of the direct effect of microplastic additions (β = 0.55). This indicates that changes in the relationship between nematodes and soil microorganisms are the dominant mechanism underlying the observed ecological impacts of microplastic pollution on microbial activity. This further highlights the necessity to consider multiple trophic interactions to better understand the responses of belowground communities to microplastic pollutions.

5. Conclusion

Overall, our field-based microplastic additions resulted in reductions of abundance and shifts in the community composition of soil fauna, especially at the high level of microplastic concentration (+15 g m⁻²). Furthermore, the SEM result suggests that the responses of soil fauna to microplastic additions cascaded through the soil food web, leading to stronger indirect effects on soil microbial functioning than the direct effect induced by microplastics themselves. To our knowledge, this is the first field-based study showing that microplastic additions have important ecological consequences on the structure and functioning of soil communities at different trophic levels. Our results suggest that the abundance and structure of soil biota may serve as useful bioindicators to monitor microplastic pollution in soils. Our results also highlight the importance to consider the effects of microplastics at various trophic levels to elucidate the mechanisms underlying the ecological impacts of microplastic pollution on soil functioning. It is noteworthy that the current study captured the effects of only one microplastic type, and therefore, generalizations need to be made with caution. Considering the increasing microplastic pollution in soils worldwide, the impact of various types of microplastics (e.g. different polymer types, shapes, sizes) on belowground biota and their associated ecological functioning merits critical investigation. Finally, we call for a reduction in the use of plastics and to avoid burying plastic wastes in soils, as this may bring adverse ecological consequences on soil communities and biogeochemical cycling in terrestrial ecosystems.

Data accessibility

All data are available from the Dryad Digital Repository: https://doi.org/10.5061/dryad.1zcrjdfpw [73].

Authors' contributions

D.L. conceived and designed the study. D.L., G.Y. and P.D. collected data. D.L. conducted statistical analyses and wrote the first draft of the manuscript that was amended by N.F. All authors discussed, revised and approved the final version of the manuscript.

Competing interests

We declare we have no competing interests.

Funding

This work was supported by Fundamental Research Funds for the Central Universities (grant no. 2018CDXYCH0014).