Estimating the size distribution of plastics ingested by animals

The study addresses how animal body size constrains the sizes of plastic particles that can be ingested, aiming to incorporate biological ingestibility into global plastic pollution risk assessments. Contextually, plastic ingestion has been recorded across diverse taxa with implications for individuals, populations, ecosystems, and human health. While models of plastic and organism co-occurrence exist, they assume equal ingestibility across plastic sizes. The authors propose that body length, an easily obtained trait with strong allometric links to biological processes, can predict the maximum ingestible plastic size, thereby enabling more realistic risk modeling and understanding of plastic entry and transfer through food webs.

Prior work has mapped global distributions of floating plastics and species, facilitating co-occurrence and encounter-rate based risk assessments but lacking ingestibility metrics. Factors proposed to affect ingestibility include feeding behavior, prey size distributions, plastic color, degradation state, and odorants/infochemicals (e.g., dimethyl sulfide). Allometry is widely used to relate body size to biological traits, and previous studies have suggested body size influences plastic-biota interactions. However, global plastic models are largely limited to the ocean surface mixed layer, whereas many animals ingesting plastics occupy greater depths, indicating a mismatch between environmental data and animal habitats and motivating biologically informed ingestibility models.

Systematic review: On January 26, 2018, the authors queried Web of Science (v5.27) across SCI-EXPANDED (1900–present), SSCI (1956–present), A&HCI (1975–present), CPCI-S (1990–present), CPCI-SSH (1990–present), and ESCI (2015–present) using the topic search string: ((plastic OR plastics OR microplastic OR mesoplastic* OR macroplastic*) AND (ingest* OR absorb* OR devour* OR eat* OR digest OR consum OR swallow OR ingurgitat OR engorg* OR gorge OR graz* OR masticat* OR ruminat* OR prey OR meal OR nourish OR diet OR sustenance OR gastro* OR stomach OR intest* OR assimil* OR incorporat* OR embody* OR engulf OR envelop*) NOT (consumer)). The 22,205 records returned were sorted by relevance and the top 1,999 exported for screening. Inclusion criteria required peer-reviewed primary field studies reporting ingestion of plastics (any type/size) at natural environmental concentrations; human ingestion reports and reviews were excluded. When uncertain, items were retained for full-text screening. Opportunistic additions after Jan 2018 followed the same criteria.

Data collection: Studies were included if they reported or allowed estimation (including via image analysis) of (1) the length of the longest axis of ingested plastic and (2) the mean or mid-range body length of taxa or individuals containing plastics. When only size-bin means or mid-ranges were available, weighted means were calculated. Approximate total body length definitions included capitulum length, curved carapace length, carapace width, and bivalve shell length; total length was defined as the distance from the most anterior to most posterior body parts. Data were prioritized at the lowest taxonomic level (typically species). For groups, the largest ingested plastic across group members was paired with the mean body length of animals that contained plastics to avoid pseudoreplication. Only linear length measures of plastics were used. Individual-level data were summarized to group-level when both existed. Duplicate records across studies were resolved by retaining the most precise data. Coordinates were approximated from reported values or site descriptions. Habitat classifications (marine/brackish/freshwater) and depth ranges were obtained from FishBase and SeaLifeBase.

Measurement methods: Ingested plastics were those found via necropsy or tissue digestion (e.g., KOH, NaOH, H2O2). Plastics in feces, regurgitates, live animal observations, or behavioral studies were excluded. When necessary, plastic and animal lengths were measured from images using ImageJ (v1.51J8), measuring the longest straight axis (segmented line for coiled pieces), using highest-resolution images and only wholly visible plastics. For coiled plastics that could not be identified as single pieces, the maximum axis of the coil was measured.

Data analysis: The universal allometric relationship between animal body length and maximum ingestible plastic length was modeled as a linear regression on log10-transformed variables in Excel (v16.16.7) and R (v3.6.1) within RStudio (v1.1.463). The fitted relationship is Plastic Size = 10^(0.9341*log10(Body Size) − 1.1200).

Validation: A repeated random sub-sampling validation was performed: in each of 1000 iterations, 90% of observations were used for parameterization and 10% for validation. Predictions were generated using the predict function in R stats, and accuracy assessed with RMSE and a regression between predicted and observed values.

Application to zooplankton risk: The authors combined the allometric relationship with global maps of plastic count concentrations from Eriksen et al. (size classes 0.33–1.00, 1.00–4.76, 4.76–20.00, >20.00 mm) and zooplankton mass distribution from Strömberg et al. Zooplankton size range (0.12–13.5 mm) was derived from COPEPOD biometrics; the model predicts a 13.5 mm animal can ingest ~0.86 mm plastic. Plastic maps were georeferenced and classified in ArcGIS (v10.5.1), transforming exponential bins to linear scales (y = 10^(6/8 x) for plastics, y = 10^(2/5 x) for zooplankton). Ingestible plastic density (0.33–1.00 mm) was divided by zooplankton density (mg C m−3) to map risk; a comparable map using total plastics (sum of all four size classes) was also produced.

• Allometric relationship: Animal body length explains 42% of the variance in the maximum ingested plastic length (log10–log10 linear regression; R^2 = 0.42, F1,634 = 46.06, p = 4.7e-09), implying a ~20:1 ratio between animal length and the largest ingestible plastic. Fitted equation: Plastic Size = 10^(0.9341*log10(Body Size) − 1.1200).
• Dataset: >2000 gut-content surveys synthesized; taxa spanned three orders of magnitude in body length (from 9 mm fish larvae to 10.34 m humpback whales). Records predominantly fish (75%), then mammals (9%), invertebrates (11%), reptiles (5%); environments include marine, brackish, and freshwater. Species depth ranges (25–4000 m) exceed depths covered by common global plastic models focused on surface layers.
• Validation: Predictions and observations were significantly related (R^2 = 0.38, F1,5998 = 59.96, p < 0.001) with RMSE = 0.68; 30.45% of observed maxima fell within 95% CIs due to data sparsity at extremes. Tendency to under-predict for very large animals (>2000 mm) and over-predict for very small animals (<30 mm).
• Smallest fragment detection: Weak relationship between animal length and smallest ingested plastic (R^2 = 0.10, F1,61 = 7.58, p = 0.008), reflecting methodological detection limits that scale with animal size; studies of larger animals often did not report microscope use.
• Risk mapping: Applying the ingestibility filter (0.33–1.00 mm plastics for zooplankton) alters spatial risk patterns compared to using total plastics, highlighting priority regions including the East and South China Seas, Bay of Bengal, Black, Mediterranean and Sargasso Seas, and European North Atlantic coasts.

Incorporating an ingestibility function grounded in body size addresses a key gap in global plastic risk assessments that have relied on co-occurrence alone. The demonstrated allometric relationship provides a simple, broadly applicable predictor of the maximum plastic size animals can ingest, improving ecological realism in exposure estimates across taxa and habitats. Findings emphasize the centrality of particle-to-animal size matching in ingestion processes and underscore potential for plastics to enter basal trophic levels, propagating through food webs. The study also reveals mismatches between animal depth ranges and the surface-constrained plastic models, suggesting underestimation of exposure at depth. Future enhancements should tailor models to specific groups by adding variables such as feeding mode, mouthpart morphology, ontogenetic stage, and habitat preferences, and should integrate improved environmental plastic distributions, especially with depth.

The paper establishes a general allometric model linking animal body length to the maximum size of ingestible plastics, explaining a substantial fraction of variance and yielding an approximate 20:1 body-to-plastic size ratio. This quantifiable ingestibility metric enables estimation of the fraction of environmental plastics accessible to different animals and improves risk mapping, as illustrated for zooplankton. The approach can be integrated with co-occurrence data to refine global assessments, support monitoring, and guide mitigation. Future work should expand environmental plastic datasets beyond the surface layer, include terrestrial taxa, and develop group-specific models incorporating additional life-history and morphological traits to enhance predictive accuracy across food webs.

• Data sparsity at the smallest (<30 mm) and largest (>2000 mm) animal sizes reduces prediction robustness at extremes, leading to under-prediction for very large animals and over-prediction for very small ones.
• Cross-study methodological heterogeneity, particularly in detection of small plastics (e.g., variable use of microscopy), weakens inference on the minimum ingestible sizes.
• The meta-analysis lacks terrestrial animal data, limiting generalizability to non-aquatic systems.
• Environmental plastic models used for applications are largely surface-based; many species ingesting plastics occur at greater depths, potentially biasing exposure estimates.
• Reliance on necropsy datasets and summarized group-level statistics may obscure individual-level variability and introduce measurement uncertainties (e.g., body length definitions, image-based measurements).