Location: Laboratory study at Stony Brook University with sediments and organisms from Shinnecock Bay, Long Island, New York. Sediment and animals for all experiments were collected from an intertidal and shallow subtidal sandflat at Old Ponquogue Bridge Marine Park (40.842° N, 72.498° W).

Sampling and Setup for Summer 2019 Antfarms:
Sediment was collected on 6/5/2019 with 14.5 cm diameter cores that were pushed down to a shell layer at approximately 30 cm depth. The bottoms of the cores were sealed with caps and transported submerged in water to the lab. The upper centimeter was separated from the remainder of each core. Top and bottom sediment was combined from multiple cores, sieved, and homogenized. Antfarms were filled to ~3 cm below its top with bottom sediment, dropping sediment through water to avoid air bubbles. It was then topped with ~1 cm of top sediment and allowed to settle overnight with recirculating overlying water in a temperature-controlled room at approximately field temperature (~19°C). In parallel with sediment sampling, maldanids polychaetas were collected by splitting sediment with a fork and carefully selecting intact tubes with animals by hand. Animals were transported to the lab and placed in a bucket with aerated overlying water overnight in the same temperature-controlled chamber as the antfarms. The next day, 7 living individuals were selected from among the animals and were placed on the sediment surface in the antfarms and allowed to burrow. Vacated tubes were removed with forceps. During oxygen imaging, temperature was periodically adjusted to create a temperature ramp, ending at approximately winter temperatures. The timing of these changes is noted in the table below. During the final image series, overlying water was injected manually close to the optode in regions unaffected by worms and oxygen concentration declines were used to determine oxygen consumption rates.

Sampling and Setup of intact Cores in Winter and Summer 2020:
Intact cores were collected on 3/9/2020 and 7/24/2020 with 14.5 cm diameter clear polycarbonate tubes that were pushed down to a shell layer at approximately 30 cm depth. Tubes were prepared with oxygen sensitive foils that were glued to the inside of the tubes with double-sticky foil. The bottom of the cores were sealed with caps and transported submerged in a water bath to the lab and hooked up to a recirculating overlying water system in a temperature-controlled room set to approximately field temperature (6°C and 21 °C in winter and summer, respectively). During oxygen imaging, cores were rotated occasionally to look for activity in different parts of the cores. These rotations are readily apparent in the images. On 8/5/2020 from ~Image 1634 in the 200804 data set to ~ Image 5320, a hypoxic/anoxic event was induced by reducing the oxygen supply to the recirculating water and allowing respiration within the cores to draw down the oxygen concentration.

Determination of Porosity and Permeability:
In July of 2017, four 3.8 cm inner diameter cores were filled with sediment from the study site using the same method as the Summer 2019 Antfarm experiment. These cores were allowed to settle overnight and then measured for permeability by constant head tests (Klute and Dirksen 1986), i.e., measuring the flow rate of water through the cores under at least 3 different pressure heads. Cores were subsampled with a cut-off 10 mL syringes and porosity was estimated via weight loss on dehydration of water-saturated sediment. The results of these measurements are reported as estimates for the hydraulic conductivity and porosity of sediments used in these experiments.

Sampling and analytical procedures:

O2 images were taken over a period of multiple days with slightly varying spatial and temporal resolutions. Specifics on each data set are provided in the supplemental table "Dataset Metadata Log." In all cases, O2 optodes were calibrated using the lifetime values measured in the anoxic sediment and in the air-saturated overlying water.

Instruments:

The luminescence lifetime imaging system is modified after Holst and Grunwald (2001) and comprises a cooled CCD camera (pco.1600MOD, PCO AG, Kelheim, Germany), a pulse delay generator (T560, Highland Technology, San Francisco CA), an array of blue-light emitting diodes (LEDs; lambda max = 455 nm, LXHL-LR5C, Philips Lumileds, San Jose, CA) attached to a heat sink, and a custom-made power supply. The oxygen optodes were prepared as described in Precht et al. (2004). The CCD camera accumulates multiple exposures with a programmable modulation time. Using two intensity images, the luminescence lifetime image is calculated (Holst and Grunwald 2001). The peak current through the LEDs (typically 200–300 mA) and the integration time during which both intensity windows are accumulated (typically 250–1000 ms) were adjusted to optimize data quality. The control of the camera, image acquisition through the IEEE 1394 (firewire) interface and of the delay pulse generator through the RS232 serial interface were done by Borland Delphi and C++ computer software developed by Lubos Polercky (Utrecht University, The Netherlands) and Uli Henne (German Aerospace Center, Göttingen, Germany).

Issue Report:
For routine checks and troubleshooting lights were switched on for short periods, or persons were blocking the camera view during the multi-day imaging series. With lights on, the optical O2 measurements is impaired. These images should be excluded from any future analyses and are easily recognizable due to abrupt jumps in the data series. Also, during the Summer 2019 Antfarms experiment, the O2 imaging system was used to simultaneously image O2 distributions in an unrelated column study. These columns are easy recognizable in the right half of the O2 images. In our experience, keeping data spatially uniform and temporally complete - rather cropping images and chopping time-series - makes data analyses much easier. The time-series data therefore include time periods with non-sense data but are always the full 1600 x 1200 pixel O2 image.