{"id":"1d29a8dd-6d2d-4b1e-af61-7756f975a49b","arxiv_id":"2411.14335","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"In a global 3D model, only low-density microplastics of roughly 10 micrometers or larger accumulate in the subtropical garbage patches, while micrometer-scale particles behave like neutrally buoyant tracers.","lead":"This paper simulates where ocean microplastics of different sizes and densities end up using a global 3D computer model. It finds that only larger low-density particles, about 10 micrometers across, build up in the five subtropical garbage patches, while 1-micrometer particles disperse through the upper ocean, and it links seasonal surface changes to the depth of the ocean mixing layer.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Reported positive correlation between surface concentration and mixed-layer depth contradicts the paper's own inverse-proportional mechanism; the seasonal causation claim is internally inconsistent as written.","rationale":"The paper has two central claims: the size-dependent gyre accumulation pattern and the seasonal ML-depth mechanism. I read the Stokes-law treatment as an explicit modeling idealization; shape and biofouling effects would certainly modify real-world terminal velocities, but the paper's size-based classification is internally coherent under its stated spherical-particle assumption. The sparse sampling of only two diameters per material limits threshold precision but does not invalidate the qualitative separation. The seasonal claim, however, contains a direct internal contradiction: a positive tau-h_b correlation is reported while the proposed mechanism requires tau to be inversely proportional to h_b. Because this mechanism is a headline result and is used for CYGNSS validation, the inconsistency must be resolved before the paper can be accepted. The reader's weakest assumption focused on Stokes-law sensitivity and diffusivity fields; their rationale did mention the lack of a mass budget for seasonality, so there is partial overlap, but the sign contradiction is a sharper, independently identifiable issue. The concrete correlation test will settle whether the flaw is a fixable sign error or a substantive failure of the causal explanation.","tokens_in":14093,"tokens_out":8308,"duration_ms":84626,"concrete_test":"Using the model output, compute monthly or climatological time series of PE-10 surface concentration tau and mixed-layer depth h_b at the representative point [30N, 130W] and for all ocean grid cells over 2012-2017. Calculate Pearson r(tau, h_b) and r(tau, 1/h_b). If r(tau, h_b) is negative and near -1, while r(tau, 1/h_b) is positive, the published 'positive correlation' is a sign error and the mechanism survives after correction. If r(tau, h_b) is positive and near +1, the inverse-proportional mechanism is falsified and Section 3.3 must be rewritten. Also inspect the colorbar and sign convention in Fig. 10a against the raw correlation values.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In Section 3.3, the authors report that the correlation between surface PE-10 concentration tau and mixed-layer depth h_b is 'positive ... close to 1' (Fig. 10a), then explain the seasonal cycle by a uniform particle distribution in the ML with conserved total mass and conclude that tau varies 'inversely proportional' to h_b. These two statements cannot both hold: if tau is approximately M/h_b with M conserved, the correlation of tau with h_b is negative, not positive. The same contradiction appears against their own Fig. 7, where tau is higher in summer, while h_b is shallower in summer and deeper in winter. Thus the reported statistic, as written, does not support the causal mechanism; either the text/figure has a sign error, or the correlation was computed against a different quantity such as 1/h_b. This is a load-bearing internal inconsistency in the second central claim, independent of any Stokes-law or shape-idealization caveats.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a global 3D Eulerian advection-diffusion model for microplastics with a size- and density-dependent vertical terminal velocity, forced by ECCOv4r4 currents and coastal mismanaged-waste inputs. Six realistic particle cases plus 2D and neutrally buoyant idealizations are integrated for 26 years. The authors claim that only sufficiently large positively buoyant particles (e.g., PE at 900 kg/m^3 with d >= 10 um) accumulate in the five subtropical gyres, whereas particles near 1 um behave essentially as neutrally buoyant tracers with weaker surface patterns and penetration to roughly 1 km depth. They further report a seasonal cycle of surface PE-10 concentration whose phase agrees with CYGNSS observations, and attribute this cycle to seasonal variation of the mixed-layer depth under an assumed uniform vertical particle distribution with conserved total mass.","tokens_in":14273,"tokens_out":9512,"duration_ms":92351,"significance":"If the results hold, the paper makes a useful contribution by bringing particle size into a global Eulerian microplastic transport framework and proposing a physical mechanism for seasonal surface concentration. The modeling infrastructure is sensible: the buoyancy advection term is written in mass-conservative flux form, the flow and mixing fields come from a data-constrained reanalysis at higher resolution than the previous global Eulerian study, no free parameters are tuned to produce the gyre patterns or seasonal signal, and the code and inputs are intended to be publicly available. The six-case matrix cleanly brackets the neutrally buoyant limit. However, the central size-threshold claim rests on only two diameters per material and on Stokes-law smooth-sphere terminal velocities, and the seasonal mechanism is currently undermined by a sign inconsistency between the reported correlation and the proposed inverse proportionality. These issues do not invalidate the framework but require correction before the conclusions can be accepted as stated.","major_comments":[{"comment":"The seasonal mechanism is internally inconsistent as written. The text reports that the correlation between surface concentration tau and mixed-layer depth h_b is 'positive ... close to 1' (Fig. 10a), and then explains that tau varies inversely with h_b because particles are uniformly mixed in the ML with total mass conserved. If tau is approximately M/h_b with M conserved, the correlation of tau with h_b must be negative, not positive. The paper's own Fig. 7 and Fig. 11 also show high tau in summer when h_b is shallow and low tau in winter when h_b is deep. The authors should correct the sign of the reported correlation, specify that the correlation was computed against 1/h_b or against a transformed variable, or demonstrate that a non-conserved budget term breaks the inverse proportionality. As it stands, the reported statistic contradicts the proposed causal mechanism and must be fixed before the seasonal claim can be evaluated.","section":"Section 3.3, Figs. 7 and 10"},{"comment":"The size-threshold conclusions are not supported by the simulation matrix. Each material is simulated at only two diameters (Table 2), so 'PE with d >= 10 um' is not a demonstrated threshold but rather the statement that the tested 10 um case accumulates while the tested 1 um case does not; for PP, the transition is bracketed only between 10 and 100 um, not located. In addition, Eq. (3) presupposes smooth spheres with CD = 24/Re, while Section 4 defers biofouling and fragmentation to future work; non-spherical shapes, biofouling, and fragmentation can change the effective terminal velocity by orders of magnitude and could shift the claimed 10 um and 1 um values substantially. I request either additional intermediate-size or shape-sensitivity simulations, or a revision that presents the result as a qualitative size-dependent transition rather than precise threshold diameters.","section":"Section 3.2, Table 2 and Figs. 4-6"},{"comment":"The claim that small particles penetrate to about 1 km and that gyre accumulation is weak depends on the GGL and GMRedi diffusivity fields from ECCOv4r4 (Section 2.2), but no sensitivity of the penetration depth or of the 10 um vs 1 um distinction to the vertical mixing intensity is provided. Because in the small-particle limit the residual terminal velocity is tiny, the depth profile is largely controlled by the prescribed K, and the 1 km value is therefore a model-dependent diagnostic. A simple sensitivity experiment (for example, scaling K by factors of 0.5 and 2) would show how robust the central size-based classification is to this external input.","section":"Sections 3.1-3.2, Figs. 3-6"}],"minor_comments":[{"comment":"The GitHub repository URL appears malformed as printed: the text contains spaces and the fragment 'online 68o', and the instruction file is called 'Read me.pdf'. As written, the reproducibility link cannot be used; please provide clean, complete URLs.","section":"Open Research Section"},{"comment":"The ECCOv4r4 period is given as '1992 to 2007' here but as '1992 to 2017' in the Introduction and elsewhere; since ECCOv4r4 spans 1992-2017, the Section 2.1 date appears to be a typo.","section":"Section 2.1"},{"comment":"The global correlation maps are presented without significance testing, sample sizes, or degrees-of-freedom information; given the strong serial correlation of monthly ocean fields, 'close to 1' is not by itself a meaningful statement of agreement unless the correlation is computed on properly detrended or independent data.","section":"Section 3.3, Fig. 10"},{"comment":"The CYGNSS comparison is qualitative: it compares the month of maximum concentration in a single year (2017) and assesses agreement visually. A quantitative phase-error metric (for example, circular correlation or mean absolute month difference) would strengthen the claim, especially because the retrieval is surfactant-based and the paper itself notes this caveat.","section":"Section 3.3, Fig. 8"},{"comment":"The statement that CD = 24/Re 'irrespective of the shape of the particle' is overly strong: at low Reynolds numbers the drag on non-spherical particles still depends on shape through a shape factor, which is one of the reasons the idealization in Eq. (3) should be flagged as a limitation rather than as a shape-independent law.","section":"Section 2.2"}],"recommendation":"major_revision","confidential_remarks":"The model-development contribution is sound and within the scope of JGR: Oceans. The main obstacles are the sign inconsistency in the seasonal correlation analysis and the over-stated size-threshold language; both are fixable in revision. I also recommend that the authors clean up the repository URL. I would be willing to review a revised version."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThe headline: this paper is worth reading for the size-dependent Eulerian modeling, but the second main claim—the ML-depth mechanism—is internally inconsistent as written and needs correction before the paper can be trusted.\n\nWhat's new: it's the first global 3D Eulerian microplastic simulation that puts particle size into the terminal velocity. Stokes settling with CD=24/Re, a mass-conservative buoyancy advection term, 26 years on ECCOv4r4. The gyre results—PE-10 accumulates in the five subtropical gyres while PE-1 behaves like a neutrally buoyant tracer, PP-100 sinks to about 650 m, PVC-10 to the bottom—are plausible and a genuine step beyond Mountford and Morales Maqueda. The neutrally buoyant limit check and the graphical comparisons in the appendix are good practice. The CYGNSS comparison is honest about being phase-only and about the surfactant caveat.\n\nSoft spots. The size threshold rests on only two diameters per material; that's a sampling limitation, not fatal. The bigger problem is Section 3.3. The authors report a positive correlation close to 1 between surface concentration tau and ML depth h_b (Fig. 10a), then explain the seasonal cycle by uniform mixing in the ML with conserved mass, concluding tau varies inversely proportional to h_b. Both cannot be true: if tau ~ M/h_b, the correlation with h_b is negative. Their own Fig. 7 shows higher tau in summer when h_b is shallow. Either the correlation was computed against 1/h_b, or there's a sign error. As written, the statistic contradicts the mechanism, and this is the evidence for the paper's second central claim. Also, the Open Research section has malformed code links (URLs with spaces, a filename like \"Read me.pdf\"); easy to fix, but must be fixed for reproducibility.\n\nBottom line: the modeling and the gyre-size results are solid enough to merit serious review, but the seasonal attribution needs reanalysis and rewriting. Send it to a referee, but expect the authors to sort out the correlation sign and rerun the seasonal analysis before acceptance.","headline":"Worth reading for the size-dependent Eulerian microplastic model, but the ML-depth seasonal mechanism is internally inconsistent as written and needs correction before the claims can be trusted.","tokens_in":14776,"tokens_out":2426,"would_cite":false,"duration_ms":22760,"reading_group":"maybe","serious_thinker":"no","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A 26-year global ocean simulation finds that only larger, low-density microplastic particles collect in the five subtropical garbage patches, while micrometer-scale particles mix down to about 1 km.","keywords":["microplastics","3D Eulerian transport model","terminal velocity","subtropical gyres","mixed layer depth","particle size threshold","ocean reanalysis","CYGNSS"],"falsifier":"Measure the settling velocity of irregular, biofouled, and partially fragmented microplastic particles in the 1–100 µm range: if a 10 µm polyethylene fragment settles materially faster or slower than the Stokes value of about 6 µm/s used here, the paper's size threshold for gyre accumulation does not carry over to real particles. A global set of size-resolved vertical profiles in the upper kilometer would directly test the predicted near-uniform 1 µm particle distribution in the mixed layer and the sharp cutoff below it.","tokens_in":13905,"feed_emoji":"🌊","tokens_out":8941,"duration_ms":78205,"temperature":0.7,"pith_summary":"This paper aims to establish that the fate of microplastic in the ocean is set jointly by particle size and density, not by ocean currents alone. Integrating a 3D transport model for 26 years, it finds that only positively buoyant particles with sufficient size—polyethylene at 900 kg/m³ with diameter around 10 µm or more—accumulate in the five subtropical gyres. Particles near 1 µm, whatever their density, behave like neutrally buoyant tracers: their surface concentration pattern is weak and they penetrate to roughly 1 km depth. The paper also argues that the seasonal rise and fall of floating-particle surface concentration is governed by seasonal changes in mixed-layer depth, with surface concentration inversely proportional to that depth as long as total particle mass is conserved.","feed_headline":"Only larger buoyant microplastics collect in the garbage patches","feed_subtitle":"A 26-year simulation finds tiny plastic escapes the surface gyres and mixes down a kilometer; seasonal surface counts track mixed-layer…","key_machinery":"The argument is carried by a mass-conservative buoyancy flux in the advection-diffusion equation: the concentration $\\tau$ evolves with a vertical term $\\partial_z(\\tau w_r)$ rather than $w_r\\partial_z\\tau$. The terminal velocity entering that term is Stokes' law, $w_r = g(\\rho_w-\\rho_p)d^2/(18\\mu)$, which is what lets particle size enter the problem for the first time: the $d^2$ dependence, together with density contrast, determines whether a particle is surface-trapped, neutrally buoyant-like, or sinking. The flow and mixing fields come from a data-constrained global ocean reanalysis, with a turbulent kinetic energy-based vertical mixing parameterization and an isopycnal mixing parameterization supplying the diffusivity tensor, and coastline sources derived from a global mismanaged-waste estimate.","core_discovery":"The central discovery is a size threshold for gyre accumulation. For each particle density, there is a critical diameter below which the buoyancy-induced terminal velocity becomes so small that turbulent mixing dominates and the particle's distribution converges to the neutrally buoyant one. Above the threshold, low-density polyethylene rises fast enough to remain near the surface and is concentrated by the converging flow into the five subtropical gyres, while high-density PVC sinks to the seafloor and near-neutral-density polypropylene collects near the depth where seawater density matches the particle density. For floating particles, the model's seasonal surface concentration peaks in summer in the subtropical bands and agrees in phase with CYGNSS satellite retrievals; the authors attribute this to the nearly uniform vertical distribution of particles in a mixed layer whose depth changes seasonally, making surface concentration inversely proportional to mixed-layer depth under conservation of particle mass.","pith_inferences":["If real weathered microplastics are nonspherical, biofouled, or fragmented, their effective terminal velocities will differ from Stokes-law values, so an immediate extension would be to replace the smooth-sphere law with shape- and fouling-dependent settling and watch the 1–10 µm thresholds move.","The mixed-layer dilution mechanism should apply to any passively transported floating tracer, including the surfactant signal that underlies CYGNSS retrievals, which may explain part of the satellite seasonal cycle without invoking seasonal changes in plastic input.","The predicted penetration of small particles to about 1 km implies a pathway for microplastics to interact with the biological carbon pump and with deep-ocean food webs; coupling the plastic tracer to sinking organic aggregates would be a direct follow-up.","A targeted test of the size-filter hypothesis would be to measure, in a few gyre transects, whether the size distribution of floating particles is depleted below about 10 µm exactly where the model predicts, and enriched at depth."],"forward_implications":["Surface trawler surveys, which catch particles above roughly 0.2 mm, should see garbage-patch accumulation dominated by larger buoyant particles, while smaller particles are underrepresented because vertical mixing carries them below the surface.","Micrometer-scale plastic of any density should be present throughout the upper kilometer, so global inventories based on surface sampling miss a large subsurface reservoir of microplastic.","Seasonal changes in surface concentration can occur without any change in the amount of plastic in the ocean, simply because the mixed layer deepens and dilutes the same mass over a thicker water column.","For near-neutral-density plastics like polypropylene, the surface filter works in reverse: larger particles settle to a depth of matching seawater density, leaving small particles as the main surface-visible population.","The model gives material-dependent size cutoffs near 1–10 µm that separate surface-accumulating from deeply mixed particles, so particle size data are needed to compare any model or survey with the observed garbage patches."],"supporting_citations":[{"why":"prior 3D Eulerian global microplastic model whose formulation this paper extends and whose surface patterns are the comparison baseline.","marker":"Mountford and Morales Maqueda (2019)"},{"why":"supplies the terminal-velocity balance for buoyant particles that lets density and size control vertical motion.","marker":"Yang et al. (2014)"},{"why":"documents the Stokes terminal fall velocity formula used for the particle settling speed.","marker":"Dey et al. (2019)"},{"why":"provide the data-constrained global ocean state estimate that supplies the 26-year current, density, and mixing fields.","marker":"ECCO Consortium et al. (2021, 2023)"},{"why":"documents the ECCO version 4 inverse-model framework underlying the reanalysis data.","marker":"Forget et al. (2015)"},{"why":"provides the CYGNSS satellite microplastic retrieval whose seasonal phase is compared with the model.","marker":"Evans and Ruf (2022)"},{"why":"supplies the global mismanaged-waste coastline source term used for particle release.","marker":"Jambeck et al. (2015)"},{"why":"defines the vertical mixing parameterization and the mixed-layer depth diagnostic used in the mechanism analysis.","marker":"Gaspar et al. (1990)"},{"why":"supplies the isopycnal eddy-mixing parameterization used in the diffusivity tensor.","marker":"Gent and Mcwilliams (1990)"},{"why":"provides the drifter-based surface transport view of garbage patches that this model reproduces and refines.","marker":"van Sebille et al. (2012)"}],"fun_headline_variants":["Microplastic size decides if it joins ocean garbage patches","Tiny plastics evade surface gyres, sink deeper in ocean","Model reveals key size cutoff for microplastic gyre buildup","Small plastics mix below surface, only big buoyant ones gather","Seasonal ocean mixing drives surface plastic cycles"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The classification of plastic by size rests on Stokes-law terminal velocities for smooth spheres with fixed viscosity; if real weathered, nonspherical, biofouled particles settle even an order of magnitude faster or slower, the claimed micrometer thresholds and the 1 km penetration depth would shift or disappear.","fun_headline_variants_meta":{"raw":{"variants":["Microplastic size decides if it joins ocean garbage patches","Tiny plastics evade surface gyres, sink deeper in ocean","Model reveals key size cutoff for microplastic gyre buildup","Small plastics mix below surface, only big buoyant ones gather","Seasonal ocean mixing drives surface plastic cycles"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000702,"raw_usage":{"total_tokens":3194,"prompt_tokens":999,"completion_tokens":2195,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":615,"completion_tokens_details":{"reasoning_tokens":2116}},"tokens_in":615,"tokens_out":2195,"duration_ms":15622,"temperature":1.0,"reasoning_tokens":2116,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T15:17:21.992970+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the settling velocity of irregular, biofouled, and partially fragmented microplastic particles in the 1–100 µm range: if a 10 µm polyethylene fragment settles materially faster or slower than the Stokes value of about 6 µm/s used here, the paper's size threshold for gyre accumulation does not carry over to real particles. A global set of size-resolved vertical profiles in the upper kilometer would directly test the predicted near-uniform 1 µm particle distribution in the mixed layer and the sharp cutoff below it.","supporting_citations":[{"cited_title":", Chamecki, M","cited_arxiv_id":null,"evidence_quote":"supplies the terminal-velocity balance for buoyant particles that lets density and size control vertical motion."},{"cited_title":", Wang, O","cited_arxiv_id":null,"evidence_quote":"provide the data-constrained global ocean state estimate that supplies the 26-year current, density, and mixing fields."},{"cited_title":", England, M H","cited_arxiv_id":null,"evidence_quote":"provides the drifter-based surface transport view of garbage patches that this model reproduces and refines."}],"review_version":1}