The gravitational sinking of organic particles is a vital component of the biological carbon pump. This sinking process is strongly modulated by the spatiotemporally varying eddy field, complicating the interpretation of particle flux measured by deep‐moored sediment traps. By backtracking particles to 200 m depth based on the outputs of a realistic eddy‐resolving simulation, we characterize the origins of particles collected at a long‐term observatory site in the Northeast Atlantic and focus on the impact of mesoscale dynamics on particle transport. Our results show that mesoscale dynamics between 200 and 1,000 m control the statistical funnel. Over the long term, the horizontal sampling scales of traps are estimated as hundreds of kilometers, with containment radius ranging from 90 to 490 km, depending on sinking velocities. Particle travel time suggests that overall vertical flow acts to facilitate the export, with estimated deviations up to 1 ± 2 days for particles sinking at 50 m d−1 to 1,000 m. Statistical analyses of horizontal displacements reveal that mesoscale eddies at the site confine particle sources in a more local area. On average, particles in anticyclonic eddies sink faster to depth than expected from purely gravitational sinking, contrary to their counterparts in cyclonic eddies. The results highlight the critical role of mesoscale dynamics in determining particle transport in a typical open ocean region with moderate eddy kinetic energy. This study provides implications for the sampling design of particle flux measurements during cruises and the interpretation of deep‐ocean mooring observations.

<p>Energetic eddy fields generated by meso and submesoscale dynamics induce tridimensional particle transport pathways, which complicate the interpretation of observed Particulate Organic Carbon (POC) fluxes using sediment traps. It is therefore of importance to understand how horizontal dispersion of particles is structured by these dynamics from surface to depth. In this modelling study, we use a Lagrangian method to backtrack sinking particles collected at various depths ranging from 500 m to 4700 m at the PAP (Porcupine Abyssal Plain) site. Particle trajectories are computed using high-resolution simulations of the Regional Ocean Modelling System (ROMS). Our results show that the horizontal distribution of particles with sinking velocities below 100 m d<sup>-1</sup> presents a large small-scale heterogeneity. Mesoscale eddies act to define the general structure of particle patches while submesoscale features shape particle distributions through convergence/divergence processes. Distribution patterns of particles tracked from different depths suggest regime shifts of particle dispersion between subsurface layers. To identify and quantify these regimes, we perform 2d experiments at specific depths from 100 m to 4000 m and relate the Lagrangian statistics to the characteristics of the different dynamical regimes identified using vertical profiles of eddy energy and Finite Size Lyapunov Exponents (FSLE) approach.                                                                                                                                                               </p>