Microplastic behaviors and ecological risks in extreme arid hypersaline ecosystems: Qarhan Salt Lake on the Qinghai-Tibet Plateau

MPs (Microplastics), an emerging contaminant defined as particles smaller than 5 mm, have been ubiquitously detected in diverse environments globally, spanning from deep-sea trenches to alpine glaciers and the atmospheric circulation^1,2,3,4,5,6. Owing to their small size and high mobility, MPs can readily adsorb and accumulate toxic substances, including organic pollutants and heavy metals, thereby posing a potential threat to ecosystem health and biosecurity^6,7. Aquatic organisms are prone to ingesting MPs, which can cause physical harm and physiological dysfunction. Furthermore, the associated toxic compounds may accumulate within organisms and be transferred through the food web, with potential implications for human health⁴. Therefore, MP pollution has emerged as a critical environmental issue of international concern.

To date, research on MPs has been far largely focused on marine, river, lake, and urban ecosystems^8,9,10,11. Rivers serve as essential channels, transporting land-based MPs to the oceans, while lakes and reservoirs act as important sinks for these particles^12,13. Their sediments accumulate and preserve MPs over long periods, acting as a geological archive, chronicling the region’s pollution history¹⁴. Furthermore, extensive research has demonstrated a strong link between human activities, particularly population density and wastewater discharge, and the abundance of MPs in aquatic environments^5,10. These findings confirm the pervasiveness of MPs pollution. A critical follow-up question is how to accurately evaluate the ecological risks posed by these ubiquitous contaminants. The evolution of ecological risk assessment methods, has progressed from single-parameter indices, such as Pollution Load Index (PLI), Polymer Hazard Index (PHI), to more comprehensive, multi-level frameworks, such as Potential Ecological Risk Index (PERI), which effectively addresses this need¹⁵. The latest advancements incorporate cross-media risk assessment to deliver a quantifiable, comparable, and actionable foundation for regional management^16,17.

However, research on MP pollution in extreme ecosystems, especially in salt lake ecosystems characterized by severe drought and high salinity, is critically limited and requires immediate attention. A systematic review of the existing literature reveals three key gaps. First, in terms of media coverage, most previous studies have focused on single media (water or sediment) in individual lakes. For instance, Liang et al. examined MPs in water, sediment, and brine shrimp in Kyêbxang Co (Tibetan Plateau)¹⁸, and another study explored MP-associated bacterial communities along a salinity gradient in 11 Turkish lakes—yet none achieved synchronous multi-media sampling at the same sites¹⁹. Second, in terms of environmental context, existing studies have predominantly targeted natural or remote salt lakes (e.g., Spanish Ebro Basin salt lakes, Iranian Bakhtegan-Tashk Lakes), leaving a notable gap in understanding MP pollution under intensive salt lake resource industrial development²⁰. Third, regarding mechanistic understanding, laboratory studies have confirmed the synergistic effects of UV radiation on polymer aging, as well as the inhibitory effect of high salinity on MP transport—but these mechanisms have rarely been validated through systematic field observations in real-world extreme environments²¹. Salt lake ecosystems consistently display low biodiversity and simplified food web structures, meaning that even a slight plastic exposure could result in irreversible damage to the entire ecosystem. Meanwhile, the salt lake system is rich in strategic mineral resources such as potassium, lithium and magnesium, making it an irreplaceable resource treasure trove. The development of these resources serves as a direct pathway for the introduction of anthropogenic MPs. This new form of pollution, introduced by resource development, synergizes with the inherent vulnerability of salt lake ecosystems, heightening the conflict between strategic resource extraction and ecological conservation into a pressing challenge that demands immediate resolution.

Qarhan Salt Lake, situated in the Qaidam Basin on the northern margin of the Qinghai-Tibet Plateau, is the largest salt lake in China and a resource reservoir containing exceptionally abundant resources of salt, potassium, and lithium. Against this backdrop, large-scale exploitation and processing of salt lake resources have been undertaken extensively, resulting in a distinct industrial configuration. Meanwhile, the region experiences an extremely arid climate, characterized by an average annual precipitation far lower than its evaporation rate. This aridity, which, combined with intense solar radiation and persistent wind erosion, has collectively shaped a unique and fragile surface ecosystem. Its primary ecological characteristics include the extensive development of salt crusts, extremely sparse vegetation cover, and salt-tolerant plants confined to limited wetlands and riverbanks, leaving the vast majority of the surface barren. In such habitats, only a limited number of halophilic species, such as brine shrimp, can survive in the saline lake waters²². Against this backdrop of intersecting environmental fragility and intense industrial development, scientifically evaluating the ecological risks is a critical priority. Within this complex setting, the occurrence, transport, transformation, and ecotoxicological effects of MPs across multiple environmental media could exhibit greater complexity. The majority of existing research has concentrated on conventional environments, whereas studies concerning MPs pollution in multi-media systems within extreme environments and regions characterized by intense industrial activity, such as the Qarhan Salt Lake, remain notably lacking. Therefore, this study systematically analyzed the environmental exposure characteristics of MPs in this extreme and human-impacted environment, providing a scientific basis for ecological risk management in salt lake resource development areas.

Accordingly, this study focuses on the Qarhan Salt Lake, (1) to characterize the abundance, morphological properties, and spatial distribution of MPs across multiple environmental media in this extreme salt lake system; (2) to identify the key factors governing the distribution of MPs and their transport pathways among different media; (3) to develop an integrated risk assessment framework that couples pollution levels, ecotoxicological effects, and cross-media migration to enable a systematic evaluation of ecological risks and the accurate identification of key risk sources and high-risk zones. This study is designed to reveal the migration and transformation patterns of MPs under the dual influence of an extreme saline-alkali arid environment and high-intensity industrial activities. Its outcomes are expected to enrich the theoretical framework of MPs geochemical cycles and offer a valuable reference for relevant global studies in similar special environments.

Qarhan Salt Lake is the largest salt lake in China (Fig.1). It boasts the China’s largest reserves of potassium chloride, lithium chloride, magnesium chloride, and sodium chloride. The southeastern part of the study area hosts a concentration of industrial facilities, including a potash fertilizer plant, a lithium carbonate workshop, a chlor-alkali plant, and a comprehensive magnesium/sodium extraction plant. Together, they form China’s largest and most complete salt lake chemical industrial cluster²³.

The land use in the region was classified into six types: salt lake, salt pond, factory, wetland, salt flat, and river. The land use classification map of the study area was generated through manual visual interpretation (Text S1). The study area is predominantly covered by salt crust and is generally barren, with vegetation virtually absent except for isolated halophytic reeds in riparian and wetland zones. Terrestrial animals in the region are predominantly migratory birds. The food web exhibits a relatively simple structure, and the ecosystem’s resilience is limited. Furthermore, ecosystem functioning is highly dependent on the extreme salinity gradient. The placement of chemical clusters characterized by high water consumption, high evaporation rates, and intense halogen cycling on such an environmentally fragile substrate has created a unique research scenario where resource demands and ecological vulnerability are highly intertwined²⁴.

A total of 18 sampling sites were established across the Qarhan Salt Lake (Fig. 1c). At each site, surface water, surface soil (0–5 cm), and sediment (0–5 cm) were synchronously collected (one sample per medium, composited from 3–5 sub-samples). Field sampling was conducted in April 2023, consistent with the acquisition date of the Landsat 9 image (April 9, 2023) used for land use classification (Text S1).

Abundance characteristics and comparison with global and regional backgrounds

Following field sampling and analytical testing, MPs were detected in surface water, surface soil, and sediment samples from the Qarhan Salt Lake. The abundance of MPs in surface water, surface soil, and sediment samples was 30–1021 particles/100 L, 28–2339 particles/kg, and 11–3095 particles/m³, respectively. The average abundance was 340 particles/100 L in surface water, 448 particles/kg in surface soil, and 639 particles/m³ in sediment. The abundance of MPs in surface water observed in our dataset falls within a range considerably higher than the reported averages from Laguna Lake in the Philippines (1.43 particles/100 L)³, La Salada Lake in Argentina (average 14.33 particles/100 L²⁵, and the Ob River in Russia (average 4.42 particles/100 L²⁶. These countries are all characterized by primary product-oriented economies, with incomplete or single industrial systems. They primarily export primary products such as agricultural products, minerals, and energy, and their manufacturing industries are relatively underdeveloped. A similar trend is observed when compared with reported values from the urban estuary of San Francisco Bay (70 particles/100 L)²⁷, Ontario Lake in Canada (average 80 particles/100 L²⁸, and the lakes in northeastern Poland (average 78 particles/100 L²⁹. These countries are all post-industrial developed economies that have completed industrialization and are now dominated by service industries and high-tech industries. In contrast, the abundance range observed in our study area appears relatively lower than the ranges reported for the Pearl River (240–1820 particles/100 L), Taihu Lake (567 particles/100 L), the Chengdu section of the Minjiang River (1588 particles/100 L), and Vellayani Lake in India (410 particles/100 L)^30,31. This can be attributed to the fact that the central and eastern regions of China and the western region of India, as manufacturing-intensive areas, have well-developed systems of both light and heavy industries. These regions are characterized by large-scale processing and manufacturing activities as well as export-oriented economies. This also suggests that the abundance of MPs in surface water is closely related to the development level of light and heavy industries in the region. In areas with a high concentration of light and heavy industries, more MPs are likely to be released into the surrounding environmental media during production, processing, transportation, loading and unloading, and storage, making these areas high-emission sources⁸. There are also several salt lake enterprises engaged in heavy industrial production activities in the Qarhan Salt Lake. The abundance of MPs in its surface water is second only to that in areas with a high concentration of light and heavy industries. Among them, the abundance of MPs is the lowest in the central wetland and the highest on the east side of Dabuson Lake (Fig.2). This is related to the fact that salt lake chemical factories are concentrated in the east. Among the six types of land use, the factory type exhibits the highest abundance of MPs in surface water, reaching 370 particles/100 L. This indicates that industrial activity areas are significant sources and enrichment zones of MPs.

The abundance of MPs in surface soil varied across land‑use types. For example, the average abundance in the suburbs of Buenos Aires, Argentina, was 1280 particles/kg²⁵; in the urban parks of Hamburg, Germany, it was 0.34–1.25 particles/kg³²; along the coast of Hangzhou Bay, it was 6300 particles/kg³³; and in the vegetable fields in the suburbs of Wuhan, it was 320–12560 particles/kg³⁴. This is related to the differences in MPs emission sources, the retention capacity of soil media, and the intensity of human interference³⁵. The differences in emission sources influence the abundance in the surrounding environment, including MPs generated from industrial production, agricultural activities, transportation, and atmospheric deposition⁶. The variations in soil texture, porosity, and organic matter content determine the sedimentation and resuspension of MPs in surface soil, which in turn affect the abundance in surface soil¹⁰. Furthermore, in regions with limited human disturbances, MPs generally exhibit larger particle sizes and lower abundance. In contrast, substantial disturbances result in increased fragmentation and deeper vertical mixing³⁶. In the Qarhan Salt Lake, the abundance of MPs in surface soil is relatively low in the central and northwestern regions, while it is highest in the salt pond and factory areas in the southeast (Fig.2). Among them, the abundance in the factory type area reached 798 particles/kg, while that in the salt pond type area was 550 particles/kg. This is associated with the stronger emission sources and greater human disturbances in the industrial activity zones in the southeast.

Compared with the average abundance of MPs in the sediments of inland water bodies worldwide, which is 25255 particles/m³³⁷, the Qarhan Salt Lake is relatively low. Only the abundance on the east shore of Dabuson Lake and at the estuary where the Golmud River flows into the wetland exceeds 1000 particles/m³. This is closely related to the fact that salt lakes are in a state of evaporative concentration, featuring extremely low deposition rates and a deficiency of adsorption sites^2,14. The eastern shore of Dabuson Lake, densely occupied by factories, brine channels, and roads, exhibited significantly higher sediment MP abundance than other areas (Mann–Whitney U test, p = 0.0019), pointing to industrial activities as a key source of contamination. In contrast, riverine input introduced a greater quantity of sediment, thereby significantly enhancing the burial efficiency of MPs³⁸, while plant roots facilitated the accumulation of MPs³⁹, resulting in an abundance peak at the estuary where the Golmud River discharges into the wetland. Among the six land use types, the wetland type exhibits the highest abundance, reaching 750 particles/m³. This spatial association implies that the abundance of MPs in surface water is associated with the distribution of industrial zones. In areas with a high concentration of industrial activities, more MPs are likely to be released into the surrounding environmental media during production, processing, transportation, and storage, suggesting that these areas serve as major emission sources. However, this inference is based on spatial association rather than direct source apportionment; future studies employing polymer-specific fingerprinting or transport modeling are needed to establish causality.

Driving mechanisms of MP exposure in extremely arid hypersaline environments

Within the 10–500 μm detection window, smaller microplastics dominated in all three media. Particles of 20–50 μm accounted for 71.28% in surface soil, 66.57% in surface water, and 67.44% in sediment (Fig. 3). While the exclusion of particles <10 μm and >500 μm may affect absolute abundance estimates, the observed dominance of fine particles (20–50 μm) is robust, as the inclusion of the sub‑10 μm fraction would only strengthen this trend. Large plastics break down into secondary MPs under ultraviolet radiation, heat, and human disturbance¹¹. In wetland sediments, 20–50 μm MPs occupied 70.48%, due to continuous MP input from the Golmud River and physical fragmentation caused by wet-dry alternation, freeze-thaw cycles, and root compression⁴⁰. Fragments were the predominant MP shape, accounting for 67.93%, 64.87%, and 51.67% in sediment, surface soil, and surface water, followed by granules (26.08%, 31.52%, and 44.79%, respectively) (Fig. 3). The widespread impermeable plastic films used in local salt pans undergo fragmentation under strong UV radiation and salt crystallization, forming dominant fragmented MPs. Granular MPs are mainly derived from industrial adsorbents and flotation agents during chemical production⁷. In conclusion, MP contamination was primarily concentrated in the core industrial zones, associated with human industrial activities. Restricted by extreme regional environments, sediments were not the major MP sink with lower abundance than the global average. Soil MPs can migrate secondarily via wind and surface water. Additionally, recent evidence indicates that residual stress in semicrystalline plastics can drive the release of polymer micropollutants even without external degradation⁴¹. Overall, local MPs are characterized by small particle sizes and fragmented shapes, which are jointly controlled by industrial emissions, environmental aging and natural conditions.

Polymer-specific enrichment and source apportionment across mediaComposition of dominant polymers and their selective enrichment behavior in media

This study identified 33 polymers that constitute MPs. Among them, those with relatively high abundance include Polyethylene (PE), Polyamide (PA), Polypropylene (PP), Methylcellulose (MC), Rubber (RUB), Polyethylene Terephthalate (PET), and Polyurethane (PU). These seven polymers accounted for 69.07%, 62.18%, and 67.72% of the total abundance of MPs in surface water, surface soil, and sediment in the study area, respectively. Based on the abundance distribution characteristics, it was observed that different polymers exhibited significant selective enrichment trends in different media. The relative abundance of PA in surface soil is higher than that in surface water (Fig.S1), which is related to PA’s higher density (1.13–1.15 g/cm³) and its easy adsorption property. This determines that PA is more likely to be adsorbed in the pore structure of nearby surface soil after discharge⁴². In contrast, PP is relatively abundant in surface water and less abundant in surface soil. This is attributed to its lower density (0.89–0.92 g/cm³), which makes it easy to float in water bodies and migrate with water. In addition, the density of PP may increase after aging or bio-mineral attachment, thereby promoting its retention in sediments⁴². Conversely, due to the influence of wind force, it is difficult for PP to remain in surface soil for a long time, and it is more likely to migrate to water bodies and sediment environments⁴³. PET is relatively abundant in surface soil, primarily because it is prone to interweave with soil clay particles and organic matter, thereby becoming fixed within the soil pore structure⁴⁴. However, the relatively low distribution of Rubber (RUB) in surface soil is related to its difficulty in strongly adsorbing soil minerals. In the dynamic and open surface environment, they are more likely to be carried away by wind and water⁷. As can be seen from the above, in addition to differences in sources and sinks, the environmental distribution of MPs is closely related to their physical and chemical properties (such as density and adsorption). This relationship is particularly prominent under high-salt and extreme drought conditions. High salinity can induce gelation and promote sedimentation, significantly altering the environmental behavior of polymers. The weakened hydraulic force due to drought makes wind a key driving force for the secondary migration of MPs, while enhancing their long-term retention potential in surface soil.

Characteristics of salt lake industrial sources and multi-media transport pathways

In terms of spatial distribution (Fig.S1), the peak abundance of PE in surface water is located in the eastern part of Dabuson Lake, while in surface soil is found in the factory and salt pond areas. In surface soil, the abundance of PE in factory land use types reaches 104 particles/kg, which is significantly higher than that in other types of areas (Fig.4b). This indicates that industrial activity-intensive areas, such as factories and salt ponds, are important emission sources and enrichment areas of PE in surface soil. The research also confirmed that PE-based adsorbent particles and PE hollow flotation microspheres are respectively used in the production processes of lithium carbonate and potassium fertilizer⁷. In surface water, the abundance of PE also exhibits significant spatial differentiation due to the distance from areas with dense industrial activities. The peak abundance of PA in surface soil is found in the factory and salt pond areas, as well as in the northwest of Dabuson Lake. As shown in Fig.4c, among the six land use types, the abundance of PA in surface soil is the highest in the factory area, reaching 95 particles/kg, followed by the salt pond area, which has 79 particles/kg. This indicates that PA in surface soil has a strong correlation with the development of salt lake resources and is a type of MPs produced by the development of salt lakes and related human activities. The peak abundance of PA in sediments occurred in the eastern part of Dabuson Lake. Figure4c also indicates that PA is significantly higher in the salt lake type area than in other type areas, at 110 particles/m³. As shown in Fig. Sl, the distribution of PA in sediments is not sourced from the river but is enriched on the side of the salt lake closer to the factory. This further indicates that the PA in sediments is primarily related to the development of salt lake resources. In addition, the abundance of PP in surface water was significantly higher in salt lake and dry salt flat areas than in other types of areas, reaching 124 particles/100 L and 121 particles/100 L, respectively (Fig.4d). This may be related to the production of PP pellets in some areas of the Qarhan Salt Lake, as well as resource development activities. In the surface soil, the peak abundance of MC occurred in the factory and salt pond distribution areas, reaching 62 particles /kg and 47 particles /kg, respectively (Fig.4e). In sediments, the peak abundance of MC occurred in salt lake and salt pond areas, with the abundance in salt lake type areas reaching 50 particles/m³. This indicates that MC is primarily enriched in surface soil and sediment media in industrial activity areas. Studies also show that MC is widely used as a flocculant and thickener in the production processes of potassium fertilizer, lithium carbonate, and other products⁷. Similarly, PE, used as adsorbent particles and hollow flotation microspheres in lithium and potash processing, shows elevated abundances in factory and salt pond areas (Fig. 4b), while PA, which exhibits the highest abundance in factory land-use types (Fig. 4c), is closely associated with salt lake resource development activities. Collectively, these polymer-specific spatial patterns, combined with their known industrial applications, provide stronger source-indicative evidence linking industrial activities to specific MP types. Meanwhile, the abundance peaks of PET and Other in surface water, as well as RUB, PET, PU, and Other in surface soil, also occurred in factory, salt pond, and salt lake areas. This indicates that the polymer types in these media mainly originated from the development and utilization process of salt lake resources and were primarily enriched in industrial activity areas.

In addition, in the topsoil, the peak abundance of PP occurred in the distribution areas of factories and salt pond, while there were smaller peaks in the mouth of the Golmud River flowing into the wetland and Dabusun Lake. In the sediments, the peak abundance of PP was observed in the southeastern salt lake, with the wetlands following closely behind. The abundance of PP in the wetland area reached 90 particles/m³ (Fig.4d). This indicates that a considerable amount of PP MPs sources have been generated in the factory and the salt pond area, while the peak at the estuary is related to the Golmud River carrying PP into the downstream. In surface water, the peak abundance of RUB occurs in salt pond, wetlands, and the eastern salt lake areas. Among the land use types, the abundance of RUB in surface water reached 58 particles/100 L in wetlands and 54 particles/100 L in salt pond, respectively (Fig.4f). In the sediments, the peak abundance of RUB occurred in the southeastern salt pond, wetlands, and salt lakes, among which the abundance in salt pond and wetlands reached 49 particles/m³ and 42 particles/m³, respectively (Fig.4f). This indicates that RUB in surface water and sediment not only originates from industrial production activities but is also closely related to the input from the Golmud River upstream. In addition, in surface water, the abundance peaks of PA, MC, and PU occur not only in industrial activity areas such as salt pans and salt lakes but also in wetland areas. In the sediments, the Other is primarily enriched in the estuary area where the Golmud River flows into the wetland and Dabuson Lake, as well as in the salt pond and the eastern part of Dabuson Lake. This indicates that the polymers in these media originate from industrial activity areas and are also closely related to the input from upstream rivers.

In contrast, in sediments, the peak abundance of PE occurs at the estuary where the Golmud River flows into the wetland, indicating that the river acts as an important channel to transport PE from the upper reaches to this area. When the flow rate of water slows down at the estuary, PE is more likely to settle and accumulate in the sediment. In sediments, the peak abundance of PET occurs at the mouth of the Golmud River where it flows into the wetland and Dabuson Lake. The PET abundance in the wetland and river land use types is 51 particles/m³ and 46 particles/m³, respectively. This indicates that the PET in the sediment was introduced through the Golmud River upstream and deposited at the estuary where the downstream river flows into the wetland and Dabuson Lake, primarily originating from human production activities in Golmud City upstream. Research shows that PET is widely used in packaging, textile fibers, engineering plastics, and other fields^45,46, and it is a type of MPs frequently produced in human daily production activities. In addition, in the sediments, the abundance of PU in the wetland and river areas reached 73 particles/m³ and 66 particles/m³, respectively. PU is commonly used in transportation, construction, textiles, and medical fields, among others⁴⁷. This indicates that these three types of polymers in the sediment are primarily input from the Golmud River and its tributaries upstream, and their sources are closely related to the daily production activities of humans in Golmud City upstream.

Risk assessment and implications for management-from single indicator to comprehensive perspective

Risk assessment of MP exposure has been carried on by preliminary risk diagnosis based on single indicator. More details has been provided in Text S8 in the supplement.

However, the evaluation results based on PLI and PHI show significant differences. The PHI is predominantly at level IV, while the PLI is mainly at levels II and III. To seek a more reasonable evaluation scheme, the study further calculated the PERI. This index simultaneously incorporates the abundance of MPs and the toxicity potential of polymers, thereby providing a more comprehensive risk assessment. The research shows that the PERI of both surface water (Fig. 5g) and sediment (Fig. 5i) is at level I. The PERI of surface soil was mainly at level I, with only one sampling point in the salt pond showing level V (Fig. 5h). This suggests that, based on the occurrence-based risk indices applied here (PLI, PHI, and PERI), MPs do not currently indicate a widespread ecological hazard in the salt lake ecosystem. However, it should be noted that these indices reflect potential risk based on MP occurrence characteristics rather than measured biological responses. In the surface soil, the presence of Level V points in the salt pond indicates the existence of high risk areas. Salt ponds are core industrial areas for salt resource extraction, brine evaporation, and crystallization. Here, brine pumps, transportation pipelines, salt extraction vessels, industrial anti-seepage ground membranes, and various plastic equipment components are densely distributed⁴⁸. During operation, these industrial facilities continuously release MPs fragments, particularly industrial-grade plastic particles with higher density and greater toxicity⁴⁹. Therefore, salt ponds themselves are a highly intense and continuous source of point emissions. The MPs produced by the factory have been continuously accumulated in the nearby salt pond through wind transport, making it a highly enriched area. The existence of high-risk areas is the combined result of high-intensity emissions and enrichment. It reveals the possibility that, although the overall ecological risk of the Qarhan Salt Lake is low, there are localized areas of extremely high risk. This indicates that while developing and utilizing resources, it is also necessary to prevent the potential ecological risks brought by MPs.

The evaluation results of the PLI, PHI, and PERI indices indicate that the MPs pollution load in the surface water and sediment of the Qarhan Salt Lake is mainly at level II. In contrast, the surface soil is already at level III, with some areas reaching level IV. This suggests that surface soil has become the accumulation center for MPs. Meanwhile, surface water, surface soil, and sediments pose a strong potential hazard due to their high content of polymers with a high toxicity weight coefficient. Furthermore, from the perspective of PERI, MPs pollution in surface water, surface soil, and sediment has not yet posed a widespread ecological risk to the salt lake ecosystem. However, there are localized areas of extremely high risk in the surface soil. It should be acknowledged that PLI, PHI, and PERI were originally designed as single-medium indicators addressing distinct risk dimensions; their separate application to surface water, surface soil, and sediment at the same sampling site can produce up to nine divergent risk profiles (3 media × 3 indices). This fragmentation complicates the comprehensive interpretation of site-specific risk and may hinder priority-setting for management. To address this challenge, we developed CERI as a secondary synthesis tool that compresses the multi-dimensional, multi-media information into a single rankable value. Two calculation routes were employed: the Maximum method serves as a worst-case early warning indicator, while the Vector module method provides a holistic compression of the multi-dimensional risk state. It is important to emphasize that CERI is exploratory in nature and its numerical outputs should be interpreted as relative rankings for hotspot identification rather than absolute risk quantifications. The CERI integrates multiple environmental media (surface water, surface soil, and sediment) and multiple evaluation indicators (PLI, PHI, and PERI) at the same sampling point to obtain a more comprehensive and all-round evaluation result.

From the CERI calculated by the Maximum method (Fig.5j), it can be seen that the CERI is mainly at level IV, with one sampling point in the salt pond at level V. This method combines the worst risk assessment results among the nine types (3 media × 3 indices), where level IV is contributed by the PHI, and level V is contributed by the PERI of surface soil. Although the maximum method is relatively extreme in its assessment and does not take into account all indicators and environmental media in a balanced way, it also reminds us that, regardless of the overall average situation, these worst-case scenarios must be given priority attention; otherwise, they may trigger systemic risks. In contrast, the CERI results calculated by the vector module method provide a robust and integrated risk perspective (Fig. 5k). The results show that the CERI is mostly at level III, distributed in factories and salt lake areas. Among them, level V is distributed in salt pond and wetlands, and level IV is distributed in wetlands, salt pond, and the northern side of Dabuson Lake. High-risk sampling points are predominantly distributed around factories and salt pond, followed by the estuary areas where rivers flow into wetlands and Dabuson Lake. This spatial pattern suggests that industrial activities in the salt lake area are the primary risk source in the study area, followed by the input from the Golmud River. However, this inference is based on spatial association rather than direct source apportionment (e.g., industrial fingerprinting or polymer-specific tracing); future studies are needed to confirm this attribution.

Beyond the local context, the mechanisms identified here are likely transferable to other hypersaline systems worldwide. Salinity-driven polymer partitioning (e.g., density-dependent enrichment of PA in soil vs. PP in water) represents a universal process in high-salinity water bodies. The“industrial sources → surface soil accumulation → wind/water-driven secondary migration” pattern is characteristic of arid closed-basin lakes across Central Asia, western North America, and Australia, where similar industrial development pressures exist. Thus, the multi-media, same-site methodological framework developed here offers a reference for comparative MP studies in analogous environments globally, provided that local adjustments account for site-specific industrial portfolios and hydrochemical conditions.

Sample processing and analysis

Environmental samples usually contain large amounts of organic matter, inorganic particles, and biological residues, all of which may interfere with the accurate identification and quantification of MPs. Therefore, prior to MPs counting and characterization, non-plastic components were removed as thoroughly as possible using a standardized sample pre-treatment protocol. Detailed procedures are provided in the Supplementary Information (Text S2). This protocol was designed to enhance the accuracy of MPs identification and quantification while minimizing the potential degradation or loss of plastic particles.

All samples were first subjected to visual inspection and were then analyzed using laser direct infrared imaging (LDIR) to determine MPs abundance and characteristics. Visual examination was conducted using a stereomicroscope (Nikon ECLIPSE Ti equipped with a Digital Sight-U3 camera), following the criteria described in Supplementary Text S3. Prior to LDIR analysis, the filter membranes used for visual inspection were ultrasonically cleaned in ethanol to detach retained particles. The resulting ethanol suspension, containing MPs and other particulates, was concentrated to a final volume of ≤10 μL in an oven at approximately 50°C. The concentrated extract was then transferred onto an infrared-reflective glass slide (7.5 × 2.5 cm; MirrIR, Kevley Technologies, USA). After complete evaporation of ethanol at room temperature, the slide was placed into the LDIR system (LDIR 8700, Agilent Technologies, USA) for automated detection and identification. MP particles were counted and classified using Clarity software (Agilent Technologies, USA). The effective particle size detection range of the LDIR system was approximately 10–500 μm, meaning that particles <10 μm and >500 μm were not included in the quantitative analysis—a limitation inherent to LDIR-based MP characterization and acknowledged in the interpretation of size distribution patterns.

All sampling, processing, and analytical procedures were conducted following strict plastic-free protocols. Quality assurance and quality control (QA/QC) measures applied throughout the study are described in detail in Supplementary Text S4. It should be noted that methodological differences among studies (e.g., sampling devices, pretreatment protocols, and analytical techniques) may influence reported MP abundances and should be considered when comparing absolute values across different investigations.

Characterization of MPs

MPs identified in the salt lake samples were comprehensively characterized in terms of abundance, size, morphology, and polymer composition. MPs abundance was expressed as particle number per unit mass or volume of sample, depending on sample type. Particle size was determined based on the maximum Feret diameter obtained from microscopic observations and instrumental measurements. Morphological classification was conducted according to commonly accepted categories, including fibers, fragments, and pellets, reflecting differences in source, degradation processes, and environmental fate. Polymer types were identified primarily using LDIR analysis by matching infrared spectra with a reference library. Only particles with a high spectral matching confidence were classified as MPs, ensuring the reliability of polymer identification.

Comprehensive risk assessment of MPs across multiple environmental media

To facilitate multi-media, same-site risk comparison, this study proposes the Comprehensive Environmental Risk Index (CERI) as a secondary integration framework. CERI does not replace the independently established PLI, PHI, and PERI indices, but rather synthesizes their outputs across surface water, surface soil, and sediment at each sampling site into a single rankable value. The rationale is that PLI, PHI, and PERI—each validated for specific risk dimensions and typically applied to single media—may yield divergent results when computed separately for multiple media at the same location. CERI serves to compress this multi-dimensional information into a unified metric for spatial hotspot identification and priority ranking. The methods of the determinations of PLI, PHI, and PERI, have shown in Supplementary Text S5–S7.

CERI can more clearly present the comprehensive risk assessment results in the context of the nine combined scenarios (3 media × 3 indices), considering the abundance of MPs, the abundance and proportion of polymers, the toxicity of polymers, and the shape of polymers. In this study, the CERI of the study area was calculated using both the Maximum method and the Vector module method. The Maximum method adopts a conservative risk assessment strategy based on the early warning principle, aiming to identify and highlight potential environmental risks in the“worst-case scenario.” In contrast, the Vector module method provides an objective, comprehensive, and multi-dimensional assessment perspective that reflects the multi-dimensionality of risks. The calculation formulas for these two methods are as follows:

Among them, ({{PLI}}_{1}^{n}) refers to PLI in surface water, surface soil and sediment media; ({{PHI}}_{1}^{n}) refers to PHI in surface water, surface soil and sediment media; ({{PERI}}_{1}^{n}) refers to PERI in surface water, surface soil and sediment media; CERI_max is the CERI calculated using the Maximum method; CERI_Vec is the CERI obtained using the Vector module method. Prior to calculating CERI_Vec, data standardization is necessary. The Maximum method applies the precautionary principle by selecting the highest risk level across all scenarios as a conservative indicator, whereas the Vector module method offers a balanced synthesis by treating the standardized indices as orthogonal components. The assessment criteria for CERI are presented in Table S2. It is worth noting that CERI is designed to integrate MP risk information across multiple media into a site‑specific overall relative differentiation, rather than to serve as an absolute risk index. Its risk levels were determined using the natural breaks method based on the actual data distribution, underscoring that the results reflect relative differences among sampling sites rather than absolute risk classifications.

Statistical analysis

Statistical analyses were performed using ArcGIS 10.8, Excel 2016, and Origin 2021. Spatial interpolation of MP abundance in surface water, surface soil, and sediment was conducted using the inverse distance weighting method. The average abundance of MPs across different land-use types was determined through superposition analysis with the land-use classification map. Spatial distribution patterns were categorized using the natural breaks (Jenks) classification method. For inferential statistics, given that the data did not satisfy the assumption of normal distribution (Shapiro–Wilk test, p < 0.05) and group variances were heterogeneous, non-parametric tests were employed. Differences in MP abundance among land-use types, environmental media, and polymer types were evaluated using the Kruskal–Wallis test for multi-group comparisons, followed by pairwise Mann–Whitney U tests with Bonferroni correction for multiple comparisons. A significance level of α = 0.05 was applied. All figures were generated using ArcGIS 10.8, Origin 2021, and Inkscape.

Spatial interpolation analysis of MPs in surface water, surface soil, and sediment was performed using the inverse distance weighting (IDW) method. IDW was employed as a visualization tool to illustrate abundance trends between sampling points rather than to provide precise predictions for unsampled locations. It should be noted that IDW, as a deterministic interpolation method, does not provide quantitative uncertainty estimates, and the resulting maps are therefore intended to indicate general spatial trends rather than exact concentration values.