Plastic pollution in riverbeds fundamentally affects natural sand transport processes

Plastics are now pervasive across Earth surface systems, including air, soils, freshwater and marine environments, with rivers acting as the primary terrestrial conduits transporting an estimated 0.8–2.7 million metric tonnes of plastics to the oceans each year. While riverbed dunes are ubiquitous and their sand-transport mechanics are well understood, the effect of plastic particles on these processes remains poorly constrained. Prior research has examined plastic settling, erosion thresholds, transport, and movement through porous deposits, but the broader impacts on sediment transport mechanics and riverbed morphodynamics are largely unknown. Given plastics’ varied properties, longevity, and increasing abundance—often with densities comparable to or greater than freshwater—many plastics travel near or interact with the bed. The study asks whether and how plastic particles, even at low concentrations, influence riverbed sand-transport processes and dune morphology, and what implications these interactions have for sediment flux, riverbed topography, ecosystem functioning, and representative sampling. The purpose is to provide mechanistic insights using controlled flume experiments, to identify new plastic-sediment interaction processes, and to motivate a nascent sub-branch of process sedimentology focused on anthropogenic particles.

The paper situates the work within a growing body of literature on plastic ubiquity and riverine transport, citing studies on plastic settling velocities, erosion behavior relative to natural sediments, water-column distribution in rivers, and hyporheic exchange leading to subsurface accumulation. It contrasts plastic with natural organic materials (e.g., log jams, charcoal), noting that while both can be less dense than sand and affect bedforms, plastics’ diverse properties and environmental persistence render them non-analogous. The review highlights well-established dune morphodynamics and flow separation over dunes, biotic alterations of bedforms, and the role of experimental flumes in enabling morphodynamic similarity and isolating variables in complex natural systems. Gaps identified include limited understanding of how plastic properties (size, density, shape) influence incorporation into dunes, bedform stability, sediment flux partitioning between bedload and suspended load, and the spatial heterogeneity of plastic within deposits.

A physical experiment was conducted in a 10 m long, 0.5 m wide, recirculating flume (max depth 0.5 m) at the University of Hull. The bed comprised a 0.1 m-thick sand layer (1,241.6 kg) with median grain size D50 = 0.23 mm. The flume was filled to a flow depth of 0.2 m above the sand bed and run at constant discharge Q = 0.05 m3/s, yielding a depth-averaged velocity of ~0.5 m/s. After 12 h, once dunes in the sand bed had equilibrated with the flow, detailed velocity profiles over a migrating dune train (5–7 m from the inlet) were obtained using an Acoustic Doppler Velocimeter. Flow parameters at 18 °C were: Froude number Fr ≈ 0.36, Reynolds number Re ≈ 92,600, bed shear stress τbed ≈ 1.3 Pa (corrected for sidewall effects), and Shields number θ ≈ 0.35. A mixture of 13 plastic types (total mass 1.49 kg; 0.12% of total sediment mass) spanning a range of sizes, shapes, and densities was introduced, including spheres (e.g., polystyrene beads, polycarbonate Mardi Gras beads), elongate cellulose acetate cigarette filter tips, fabric pieces (polyester, fleece, polypropylene baby wipes), PVC powder (0.3 mm), and 1 mm thermosetting plastic fragments. Particle mobility was characterized using RD = R × D, where R is submerged specific gravity and D the equivalent sphere diameter (non-spherical diameters computed from volume; fabrics assumed 0.1 mm thickness). A threshold near RD ≈ 1 mm separated relatively immobile (RD > 1) from mobile (RD < 1) plastics under these flow conditions. After initial observations at Q = 0.05 m3/s for 120 min, discharge was increased to 0.06 m3/s for 120 min to simulate a flood and promote mixing, then returned to 0.05 m3/s for continued observation (>12 h). Processes were documented with video and detailed mapping of transport pathways and bedform interactions. Table 1 reports particle properties and observed mobility relative to sand.

• Plastic is not a passive component: Even at low concentrations (0.12% by mass total plastics; ~0.081% by mass for types regularly interacting with bedforms), plastics actively modified dune morphology and sand transport under constant flow (0.5 m/s).
• RD threshold for mobility and incorporation: A behavioral transition occurred around RD ≈ 1 mm (RD = submerged specific gravity × particle diameter). Plastics with RD > 1 were less mobile and more readily incorporated into dunes; RD < 1 were more mobile, less readily incorporated.
• Lee-side trapping and multi-layered deposition: Plastics accumulated on dune lee slopes via saltation or traction, depositing as individual particles, strings, or lenses. Low-RD particles could be incorporated through multi-layered deposition that sheltered lower layers from recirculation turbulence or by obstacle-induced sheltering (e.g., filter tips forming aligned obstacles with downstream tails of plastic).
• Shape and size effects: Elongate cigarette filter tips aligned with flow and anchored on lee slopes, fostering downstream sheltered deposition. Large spheres (e.g., 14 mm beads) induced local flow acceleration and scour, rolling to the base of the lee slope before being trapped.
• Stoss-side erosion and pit formation: As dunes migrated, incorporated plastics were exhumed on stoss slopes and eroded rapidly, leaving migrating pits/scours that propagated to crests, rounding crests, lowering dune height and sand volume, and sometimes generating superimposed bedforms.
• Dune washout under constant flow: Repeated pit formation and plastic erosion increased stoss erosion and reduced lee deposition, producing symmetrical, unstable dunes and, in cases of lens erosion, catastrophic dune washout and planar bed development—phenomena typically associated with stage changes but observed here due solely to plastic presence.
• Increased sand in suspension: Plastic-induced disruption shifted the local partitioning toward more sediment in suspension relative to bedload, implying increased local turbidity and potentially higher downstream sand transport rates.
• Heterogeneous plastic distributions in deposits: Resulting deposits showed strong vertical and lateral heterogeneity, with plastic-rich lenses/strings and plastic-poor zones centimeters apart. Thermosetting fragment layers preserved lee-slope cross-sets. Older plastic lenses could be overprinted by subsequent dunes, complicating prediction of plastic-rich zones.
• Particle-specific impacts: PVC powder did not noticeably modify sand transport; low-mobility Mardi Gras beads and cigarette filter tips consistently altered dune morphologies.

The experiments demonstrate that plastics fundamentally alter grain-to-grain interactions in sandy riverbeds, directly addressing the research question by identifying mechanisms through which plastics are incorporated into and then destabilize dunes. The RD-based mobility framework connects particle properties to depositional and erosional behaviors, explaining observed lee-side trapping, obstacle-sheltering effects, and stoss-side pit formation that precipitates crest rounding, dune flattening, and potential washout. These process changes locally increase the proportion of sand in suspension, implying shifts in sediment flux partitioning, increased turbidity, and potentially faster downstream transport. Morphological flattening reduces hydraulic roughness and could lower stage for a given discharge, whereas alternative bedform adjustments could increase roughness and flood stage; quantifying such reach-scale hydraulic impacts remains a priority. The strongly heterogeneous plastic distribution underscores challenges for representative sampling and for interpreting the stratigraphic record of anthropogenic particles. While morphodynamic similarity supports broader relevance, scaling to larger rivers suggests that bigger dunes with deeper troughs may be less affected by similarly sized plastics, highlighting the need for parameter-space exploration across particle concentrations, size-density-shape combinations, and flow regimes.

This study provides the first detailed process-level observations of plastic-sand interactions in riverbed dunes under constant flow, establishing that plastics actively modify bedform morphology and sediment transport by promoting lee-side trapping, stoss-side pit erosion, crest rounding, dune flattening, and in some cases washout, while increasing local sand in suspension. It introduces an RD-based framework for anticipating plastic mobility and incorporation, documents highly heterogeneous plastic distributions within deposits, and emphasizes implications for sediment fluxes, hydraulics, ecosystems, and sampling strategies. The authors propose adapting riverbed sampling to target from dune crests to depths commensurate with dune height to capture plastic-rich lenses. Future work should quantify downstream and reach-scale effects, explore broader plastic property and concentration spaces, integrate laboratory, field, and modeling approaches, and develop robust methods for representative monitoring as the sub-discipline of sediment–anthropogenic particle interactions advances.

Findings derive from controlled flume experiments with specific flow (Q = 0.05–0.06 m3/s, velocity ~0.5 m/s), sand size (D50 = 0.23 mm), water depth (0.2 m), and a selected set of plastic types and sizes. Scaling to large natural rivers may reduce sensitivity because larger dunes have deeper troughs; particle concentrations and property distributions in nature vary widely. The study did not quantify reach-scale changes in hydraulic roughness, sediment fluxes, or downstream impacts, and only limited combinations of flow speed and plastic mixtures were tested. Field validation is challenging due to observational constraints near the bed, and representative sampling is complicated by heterogeneous plastic distributions.