Samples for the Bermuda Atlantic Time-series Study (BATS) were collected during monthly cruises in July 2021 (AE2112), November 2021 (AE2124), March 2022 (AE2204), July 2022 (AE2214), November 2022 (AE2224), and March 2023 (AE2306) for seasonal comparison. Particle Interceptor Traps (PITs) were deployed at depths of 150 m, 200 m, and 300 m.

Sinking particles were collected using triplicate polycarbonate gel cups per depth, each containing 100 mL of 12% Tissue Tek polyacrylamide gel to preserve particle structure (Durkin et al., 2015; Ebersbach & Trull, 2008). Gel cups were housed in 70 mm diameter PIT tubes overlaid with dense seawater collected below the halocline (~1000 m), which was filtered (0.2 µm capsule filter, Pall Corp.) and fixed with 2% formalin (final concentration).

Triplicate PIT tubes fitted with acid-cleaned polycarbonate membrane filters (0.8 µm pore size) and filled with poisoned seawater brine (50 g NaCl L⁻¹, 0.7% formalin) were deployed alongside gel cups to measure bulk particulate organic carbon (POC) flux. Membrane filters were processed using standard BATS protocols for C/N analysis (Knap et al., 1997). See the related dataset for more information on the BATS Sediment Trap Particle Flux dataset.

After recovery, seawater above the dense brine layer was siphoned off, and the remaining seawater was drained. Excess seawater on gel cup surfaces was removed before storage at -80°C. Gel cups were transported on dry ice from Bermuda Institute of Ocean Sciences (BIOS) to Arizona State University for image analysis.

Gel cup surfaces were imaged with a Zeiss Discovery.V12 Stereo Microscope equipped with a 3.2 MP color camera. A Python-based image analysis pipeline was used to segment particles and quantify sinking particle types and size distributions. For most gel cups, 20 non-overlapping images were collected per cup, divided evenly between two focal planes: one at a higher Z-plane to capture smaller particles near the gel surface and one at a lower Z-plane to capture larger particles near the bottom of the gel. Two scale images were captured for each change in Z. The only exception was the July 2021 sampling, where each of the nine gel cups was imaged with 10 pictures and one scale image per cup.

Particles were categorized as fecal aggregates (dense, dark), phytodetrital aggregates (fluffy, amorphous), crustacean fecal pellets (ovular, dense), euphausiid fecal pellets (cylindrical, dense), or debris (fragments <60 µm, amorphous). Sinking particles like animal tissue, molts, and swimmers were excluded.

Particle areas (µm²) were measured using the Particle Image Analysis tool, then converted to biovolumes (µm³) using shape-specific formulas: spherical for fecal and phytodetrital aggregates, combined spherical/cylindrical for crustacean fecal pellets, and cylindrical for euphausiid pellets. Debris particles were treated as spherical. Biovolumes were converted to carbon content (mg C per particle) using published conversion factors from multiple ocean regions (Alldredge & Gotschalk, 1990; Silver & Bruland, 1981; Durkin et al., 2021).