Mud and organic content are strongly correlated with microplastic contamination in a meandering riverbed

Microplastics (MPs) are thought to accumulate substantially in river sediments, potentially accounting for a notable fraction of the “missing” plastics not observed at the ocean surface. While large-scale studies show widespread occurrence of MPs in freshwater systems, relationships between MP abundances and source regions are often inconsistent, likely due to hydrodynamics and sedimentary processes. Most prior work samples along long river reaches and often at single within-channel positions (thalweg or bank), leaving within-riverbed spatial variability unresolved. Hydrodynamic theory and observations suggest MP deposition is favored in zones of lower flow velocity (e.g., inner bends/point bars), and that MPs often associate with fine sediments and organic matter. The study addresses the research gap by quantifying how MP concentrations vary across distinct sedimentary environments within a meandering riverbed and by assessing whether sediment characteristics (mud and organic-matter content) can predict MP abundance at small spatial scales.

Since the first report of MPs in river sediments in 2014, numerous studies have documented MPs across freshwater environments and examined drivers such as population density, urbanization, flow velocity, catchment characteristics, and waste management. Findings are often inconsistent, with some studies reporting positive correlations with source regions and many finding none, likely due to hydrodynamic sorting and resuspension. Elevated MP abundances have been predicted and observed in zones of lower flow velocity, including inner bends/point bars. In estuarine and marine settings, MPs often co-occur with finer grain sizes; some freshwater studies also report relations with grain size and organic debris, though large-scale, heterogeneous sampling has limited robust regressions. Enders et al. proposed that correlations improve in small, well-connected areas with uniform MP throughput, reducing source variability. This study builds on that premise by focusing on within-riverbed scales in a meandering reach.

Study area and design: A meandering section of the River Lys near Ghent (Belgium) was investigated within a ~750 m reach bordered mostly by natural banks. Discharge upstream (Deinze) shows seasonal variability with winter peaks (~30–40 m³ s⁻¹) and summer minima (~1–2 m³ s⁻¹). During the study, daily averages were ~4–6 m³ s⁻¹, with short-term fluctuations (0–15 m³ s⁻¹) due to upstream weirs. Three across-channel transects targeted key sedimentary environments: two bend apices (North Bend, NB; South Bend, SB—with a deep pool) and one straight section (East Straight, ES). Within each transect, 5–7 evenly spaced surface sediment samples (0–3 cm) were collected (total n=18). Geophysical mapping: High-resolution multibeam bathymetry (Norbit WBM, 700 kHz, 60 Hz ping, 150° swath) and dual-frequency side-scan sonar (Klein 3000; 100/500 kHz) were used to map channel morphology and identify undisturbed sampling locations. Cross-sectional areas were derived to estimate average flow velocities (discharge divided by cross-section). Water column sampling: MPs were sampled using an in-house filtration device with 150 µm and 51 µm stainless steel filters. Over 6 weeks (Feb–Mar 2022), five replicate filtrations per station were conducted at 8 stations: at 1, 2, and 3 m depth in the thalweg of NB and ES, and at 1 m depth at inner and outer bend positions of NB. Per station, 159–381 L was filtered across replicates. Sediment sampling and sedimentology: Surface sediments (top 3 cm) were collected by UWITEC gravity corer (PVC liners). The outer rim in contact with the liner was removed to minimize contamination; samples were extruded and sealed in aluminum trays. Water content was determined by drying at 60 °C. Organic matter and carbonate contents were estimated by loss-on-ignition (LOI550 for OM, LOI950 for carbonates), with OM corrected by subtracting a field blank LOI550 (1.53%). Clastic grain-size distributions were measured with a Malvern Mastersizer 3000 after removal of organics (H2O2), biogenic silica (NaOH), and carbonates (HCl). Microplastic isolation and identification: For water filters, retained material was removed (5% SDS, ultrasonication), followed by KOH digestion and density separation (per Vercauteren et al.). For sediments, two replicates per site underwent extraction, digestion, and density separation. MPs were identified and quantified via micro-FTIR (Nicolet iN10) after automated optical mapping on PTFE filters; polymer types were assigned against spectral libraries. Lower MP size cutoff was 25 µm. Quality assurance/control: Plastic equipment was minimized; high lab standards were used. Water samples were corrected using LOD/LOQ methods based on blanks. An MP-free field blank (sediment heated to 550 °C for 4 h) was processed through the full workflow for sediment QA/QC. Normalization of MP concentrations: To compare across size ranges and studies, concentrations were normalized to a 1–5000 µm range to produce corrected environmental concentrations (ceMP) using the Koelmans et al. approach and power-law exponents from Kooi et al. (a=3.25 for freshwater; a=2.57 for marine). Estuarine data in comparisons used the freshwater exponent due to low salinities. Data availability and ethics: All sedimentological and MP datasets and source data are available at Zenodo (doi:10.5281/zenodo.11401693). Sampling permission was granted by De Vlaamse Waterweg nv.

• MPs were detected at all 18 surface sediment sites (0–3 cm), with concentrations from 1.50 ± 2.12 × 10^2 to 1.70 ± 0.84 × 10^3 MPs kg⁻¹ DW. After normalizing to a 1–5000 µm size range, ceMP spanned ~2.1 × 10^2 to 2.4 × 10^3 MPs kg⁻¹.
• MPs were present in the water column at all sampled depths and locations, ranging 0.13 ± 0.19 to 25.75 ± 44.31 MPs L⁻¹ (some replicates zero after LOQ correction).
• Polymer distributions differed markedly between water and sediments. Sediments: PS 86%, PET 7%, PP 6% (PE 0.3%, PVC 0.1%, PAM 0.01% sporadic). Water: PP 47%, PE 23%, PVC 21%, PET 3%, PAM 3%, PS 2%, with trace PU (<0.5%).
• Depth trends in the water column: absolute MP concentrations generally increased with depth. Normalized by type, PP (low-density) and, to a lesser extent, PS, PAM, and PE were more abundant at depth; PU, PVC, PET showed little depth relation. The enrichment of PP at depth suggests modified behavior (e.g., fouling or surface treatments increasing hydrophilicity), facilitating sinking and resuspension.
• Cross-transect averages and hydrodynamics: Despite ES having the smallest cross-section (110 m²) and thus higher average velocities (~4.5 cm s⁻¹ during survey; ~24 cm s⁻¹ annual max), it showed the highest average MP concentration (4.78 × 10¹ MPs kg⁻¹ DW). SB (347 m²; ~1.4 cm s⁻¹ during survey; ~8.5 cm s⁻¹ annual max) exceeded NB (132 m²; ~3.8 cm s⁻¹ during survey; ~20 cm s⁻¹ annual max) in average MPs (2.87 × 10¹ vs 1.63 × 10¹ MPs kg⁻¹ DW), indicating lower velocities favor sequestration when environments are comparable.
• Within-transect variability (order-of-magnitude scale): MPs were lowest in thalwegs and outer-bend slopes (high shear), increasing toward banks. Inner-bend/point-bar and straight-section slopes (lower shear) showed higher MPs; the outermost inner-bank site at SB (SB_11) was sandy and had low MPs.
• Predictors of MP abundance in net depositional settings: Excluding outer-bend (erosional) samples, strong positive correlations were observed:
  • log(MP) vs mud content: slope ~0.026, R² = 0.65, p < 0.01.
  • log(MP) vs organic-matter content: slope ~0.27, R² = 0.78, p < 0.001.
  • Combining mud and OM slightly improved prediction (R² = 0.79; n = 12).
• Outer bends are net erosional: sediments there are fine and OM-rich but have low MPs because erosion exposes older, pre-Anthropocene floodplain deposits (low carbonate content), inhibiting permanent MP deposition.
• Proposed grain-size normalization for ceMP: For log(ceMP) vs mud (%), the derived normalization slope is 0.026 ± 0.006 (1σ). This implies a factor ~1.8 increase in MP abundance per 10% mud increase (factor ~390 from 0% to 100% mud). This slope closely matches Enders et al. for estuarine settings (0.030) and is steeper than large-scale marine/lacustrine datasets with non-uniform MP throughput.
• MP size distributions showed weak spatial trends; thalweg samples in bends had relatively more >100 µm MPs, but overall size distributions were stable across sites, supporting the robustness of size-range normalization.
• Mechanistic insights: PS and PET, likely subject to rapid biofouling and higher density, are efficiently sequestered and transported primarily as bedload after deposition, leading to high sediment abundance but low water-column presence; PP shows higher water-column abundance and resuspension propensity.

The study demonstrates that small-scale hydrodynamic variability within a meandering river channel governs MP deposition patterns. Lower shear zones near banks and on inner-bend/point-bar and straight-section slopes favor accumulation, whereas thalwegs and outer-bend slopes experience limited net MP retention due to higher shear and episodic erosion. The strong correlations between MP abundance and mud and organic-matter content in net depositional environments provide predictive capability: natural sediment characteristics can serve as proxies for MP levels where throughput is uniform. Conversely, in outer bends, erosion exposes older, organic-rich but MP-free floodplain deposits, decoupling MP abundance from mud/OM content. Polymer-specific behaviors (e.g., biofouling-driven sinking and bedload transport for PS/PET; hydrophilicity-modified settling and resuspension for PP) help explain the divergence between water-column and sediment polymer profiles. The derived mud-based normalization factor (slope 0.026) offers a practical tool for inter-study comparisons and for normalizing ceMP by grain size under conditions of relatively uniform throughput, aiding both contamination assessment and MP budget estimations. The findings highlight that spatial sampling strategies must account for within-channel heterogeneity to avoid bias and to enable reliable upscaling.

This work elucidates within-riverbed depositional patterns of microplastics in a meandering river and establishes that mud and, locally even more strongly, organic-matter content are robust predictors of MP abundance in net depositional environments. MPs are approximately an order of magnitude higher near banks than in the thalweg, consistent with fluvial hydrodynamics; outer bends remain net erosional, exposing MP-poor floodplain sediments. A grain-size normalization relationship for corrected environmental concentrations (log(ceMP) vs mud; slope 0.026 ± 0.006) is proposed as a broadly applicable factor when MP throughput is uniform, facilitating normalization across aquatic environments. The study provides guidance for targeted sampling (prioritizing depositional zones such as point bars and straight-section slopes) and for estimating total MP sequestration in river sediments. Future research should include temporal monitoring across flow regimes to capture remobilization during high-discharge events, integrate subsurface coring to quantify buried inventories and sedimentation rates, and replicate small-scale, well-connected studies across diverse river types to refine universal normalization factors and assess broader applicability.

• Temporal scope: Sediments and water were sampled over a limited period; depositional and erosional dynamics across seasons and flood events were inferred but not directly monitored. More continuous monitoring is needed to quantify remobilization and net storage.
• Spatial scope and sample size: Although three transects capture key environments, only 18 surface sites were analyzed; broader spatial replication would strengthen generalizability.
• Outer-bend erosional bias: Predictive relationships with mud/organic matter do not apply in net erosional settings where older floodplain sediments are exposed; MP abundances there may vary with recent hydrologic history and approach zero with depth.
• Cross-study normalization: ceMP corrections rely on assumed power-law exponents that differ between freshwater and marine environments; comparisons across systems should be made cautiously.
• Polymer behavior variability: Differences in degradation, fouling, and treatment (e.g., PP hydrophilicity) introduce polymer-specific transport uncertainties not fully resolved here.
• Subsurface inventories: The study focuses on the upper 3 cm; estimating total sequestration requires constraints on burial, mixing, and sedimentation rates.