Comparison of inventories of plastic in different global marine reservoirs

Abstract Elucidating how much plastic debris by mass is contained in major reservoirs in the marine environment has long been a knowledge gap in the field. Knowledge of major reservoirs and solving the mass balance can inform mitigation and remediation efforts. Our objective was to review the field’s progress in quantifying major reservoirs of plastic pollution in the marine environment, and to assess whether achieving a mass balance is currently feasible with the information available. We compiled estimates of global marine reservoirs to date, and compared the approaches used to generate these estimates. We identified a total of 22 global marine reservoir estimates across 14 studies: four estimates of the coastline reservoir spanning two orders of magnitude, 12 estimates of the ocean surface reservoir spanning three orders of magnitude, one estimate of the ocean water column reservoir (40–90 MMT), one estimate of the ocean floor reservoir (3–11 MMT), and four estimates of the ocean sediment reservoir spanning three orders of magnitude. Studies obtain estimates of global reservoirs in one of two ways: by taking environmental samples and scaling them up in some way to reservoir size or by prescribing inputs and propagating them through the reservoir using computational modelling. The quantity of observational data used to inform reservoir estimates ranged from <100 samples to 10 000+ samples, which were extrapolated globally. While some studies scale based solely on the reservoir size, most attempt to account for heterogeneity using the output of a computational model that takes into account numerous transport processes. No two numerical modelling studies considered the same hydrodynamic processes, particle properties, or compartment-level processes, showing the diversity of model parameterizations used. In addition, the complexity of these models has increased over time. While studies have previously conducted mass balances of plastic across spatial scales, given the current state of knowledge, we conclude that these efforts will be associated with substantial levels of uncertainty and discuss next steps.