Methods & Sampling:
Data were collected on three cruises conducted as day trips in 2019 (from April to August) aboard the SigmaT, as indicated by the sampling dates. The SigmaT is a local vessel with an appropriate winch for the deployment of coring equipment and Niskins. This vessel is privately owned and operated. All samples were taken at the Compass Station. Drs. Erin Black and Stephanie Kienast can be contacted about all cruises and equipment deployments.

Water column samples were collected with 2- to 10-liter (L) Niskin bottles on a Seabird SBE 19plus CTD (with oxygen and fluorescence sensors) deployed using a winch wire. The bottle volume used was governed by equipment availability. On a given date, the same Niskin volume was used for all samples (i.e., all 2L bottles for one event and all 10L bottles for another).

Total 234Th
When using 10L Niskins, the 4L total Thorium-234 (234Th) samples were immediately taken upon recovery to avoid any settling effects. When 2L bottles were used, the water contained in two 2L Niskins was combined for the same depth. For the total 234Th collection and analysis, 4L of seawater was emptied from the Niskins into volume-mass calibrated bottles for each discrete depth. Total 234Th sample processing (i.e. acidification and precipitation) followed Black et al. 2018. Amendments to these methods are noted here. In summary, the bottles were acidified on the day of sampling in the lab with nitric acid and then spiked with 1 milliliter (mL) of 2.8 decays per minute per gram (dpm/g) 230Th yield monitor using an autopipette. After ~8 hours of equilibration, solutions of KMnO4 and MnCl2 were added to the 4L bottles. The pH was raised above 8 to allow for precipitation. After ~8 hours the custom filter heads were added to the bottles and the total 234Th was collected (as a precipitate) on 25 millimeter (mm) QMAs. All beta counting of the filters was performed at Woods Hole Oceanographic Institution (WHOI). No blank corrections were made to the total 234Th activities because initial process blank counts (0.29 counts per minute (cpm)) were indistinguishable from the average value (± s.d.) for the beta detectors running for 24 hours with no sample (0.28 ± 0.04 cpm). Consistency in beta counter efficiency was monitored using technetium-99 and uranium (via supported 234Th) standards. The total 234Th samples were beta counted first (before the particulate samples) within 4 to 10 days of filtration and then again at least 5 months after filtration. After all beta counting was complete, a Thorium-229 (229Th) recovery yield monitor of 7.7 × 10^3 dpm per gram was added for the filter digestion and ICP-MS analysis. The average recovery was 83% with an s.d. of 13.5% and a median of 85%. The uncertainties on total 234Th are derived from counting statistics and the extrapolation of errors associated with sample processing (i.e. mass and volume measurements, ICPMS recovery analysis).

Particulate, size fractionated 234Th and carbon
~42 liters were collected at each depth with the Niskins and the water was temporarily stored in cleaned plastic barrels prior to filtration. For the particulate 234Th collection, a custom-built 142 mm apparatus was used to vacuum pump sample water from the barrel through a 51 micrometer (µm) pore size nitex filter and a pre-combusted 1 µm pore size quartz microfiber filter (QMA) in sequence. Filters were visually examined for swimmers and a 400 µm pre-filter test found none (August sampling). Photos were taken of all filters in the lab for general assessment of particle loading and color differences, which included the 400 µm screen pre-filter test material. The particulate material from the nitex screens was rinsed onto pre-combusted 25 mm QMA using filtered seawater (<1 µm) and dried overnight. The 142 mm QMA were dried similarly and then a 25 mm punch was collected for successive 234Th and carbon measurements. The particulate samples were beta counted within 6 to 12 days of filtration and then again at least 5 months after filtration. Uncertainties on particulate 234Th activities are derived from counting statistics and error propagation from sampling processing.

Particulate dipped blanks were collected aboard the ship using the 40-meter (m) samples on all dates. A plastic housing (for the dipped blank), containing a 51 µm filter and a 1 µm filter, was submerged in the 40 m barrel until vacuum filtration ended. The plastic housing allowed for the free exchange of sample water. The average beta count for the 51 µm dipped blanks was 0.27 ± 0.2 cpm and for the 1 µm dipped blanks was 0.30 ± 0.4 cpm. As these averages were statistically indistinguishable from empty detector cpm, no blank corrections were required for 234Th analyses.

Particulate carbon content was analyzed in the CERC.OCEAN facility at Dalhousie University. Water column particle filters were analyzed for carbon content using a Costech Instruments Elemental Combustion System 4010. Half of the 25 mm water column QMA filters were analyzed at a time (and both halves were analyzed separately as replicates as indicated). The carbon analysis for the six dipped blanks indicated a small correction was needed for the large and small particle filters. Average dipped blank values were 4% and 9% of the total sample value, respectively. It is important to note that due to the low volume sampled for particulates (~42L) and the subsequent subsampling of half of a 25 mm punch from the small particle QMA filters, there were three samples with blank averages totaling 14-21% of the sample carbon values. With additional instrument uncertainties for the carbon measurements, it is recommended that the full 25 mm filter be used or that a greater volume is collected. The latter may be difficult with high-volume pumps (i.e., McLanes or Challenger pumps), as they have been observed to clog and stop before pumping 40L in coastal waters.

The rinsing of the 142 mm nitex screens to transfer the large particle material onto a 25 mm filter can create a variable distribution on the filter. Thus, the entire 25 mm filter is beta counted for 234Th, but only ½ of the filter is typically used for CHN analysis (due to the size of the filters). The cutting of the 25 mm filter can introduce uncertainty because of inequal halving of the filter and due to the choice of where to place the cut (if there is variable distribution of the particle material). For six of the nitex screens (April and May samples), both halves were analyzed for particulate carbon. The average relative standard deviation (RSD) of the two halves was 17%. For the April and May samples, we could add the two filter values together for the total particulate organic carbon (POC). To be conservative, when we did not have POC concentrations for both halves of the filter in August, we automatically assumed a minimum 20% uncertainty on the single half concentration. This was assumed for the small particle QMA filters as well, although they are much easier to cut. Eliminating the need for cutting the filters or using a partitioned-circle template to cut is recommended in future studies.

Total Particulate RAp234
RAp234 is the residual β activity of particulate 234Th and behaves in a similar manner to lithophile elements (Lin et al. 2016). The residual β activity is found by beta counting the large and small particle filters well after all the shorter-lived and unsupported isotopes have decayed away (e.g. unsupported 234Th). This would be after ~120 days or five half-lives of 234Th. Data from this study has a method limit for RAp234 equivalent to 2 times the s.d. of the dipped blank mean value of RAp234 or 1.54 becquerels per cubic meter (Bq m^-3). The small particle data dominates the RAp234 activity signal (small particle RAp234 makes up 86%-99% of the total RAp234). Any assumed issues with the large particle 234Th data (i.e. completing first counts quickly) do not impact the RAp234 values because the activities of the longer-lived isotopes contributing to the 'background' counts will not change significantly over the counting period. To provide a conservative estimate of uncertainty, a set value of 20% was applied to all total RAp234 values. Propagated counting and measurement uncertainties ranged from 8% to 20%.