The Bedford Basin (BB; 44.6955° N, 63.6330° W; Fig. 1) is a 17 km², semi-enclosed fjord in Nova Scotia, Canada, and forms the northwestern part of the Halifax Harbour. The basin has a volume of 5.6×108 m³ (Burt et al., 2013) and a maximum depth of 70 m. It is connected to the inner Halifax Harbour via a 20 m deep sill and to the open Atlantic Ocean by a 10 km long channel that is less than 400 m wide at its narrowest point (Li and Dickie, 2001). Runoff from the Sackville River (1.5×108 m³ yr⁻¹, Kepkay et al., 1997) and surrounding areas is discharged into the basin, creating a two-layer estuarine circulation – a fresher top layer and a denser bottom layer from the shelf water intruding from the entrance of the Halifax Harbour. The hydrography of the basin is characterized by annual cycles of winter mixing and summer stratification. Stratification is established in the spring (April–May) and persists into the fall and winter (November–February). In most years, the water column fully mixes during the late winter (mid-February to mid-March). Seasonal hypoxia occurs in the deeper basin during the fall, but re-oxygenation takes place during late winter mixing. In spring, a phytoplankton bloom typically occurs as surface stratification increases. Chlorophyll a (chl a) concentrations in surface waters (1–10 m) typically range from ∼3–10 mg m⁻³. Spring blooms typically range from ∼7–28 mg m⁻³ (Fisheries and Oceans Canada, 2025), while extreme short-lived events can exceed 100 mg m⁻³ (Li and Dickie, 2001). Occasional intrusions of dense Scotian Shelf water can replenish either the bottom or mid-depth waters (Shi and Wallace, 2018; Haas et al., 2021). The Bedford Basin Monitoring Program (BBMP), established by the Bedford Institute of Oceanography (BIO) has conducted weekly measurements of water properties since 1992. Despite a large array of oceanographic variables that are frequently sampled by BBMP and other emerging research initiatives (Fennel et al., 2023; Burt et al., 2024), limited information exists on POC export and burial in this important urban coastal region.

Weekly monitoring at the Compass Buoy Station in the centre of the Bedford Basin provides temperature, salinity, oxygen, and chl a concentrations (Fisheries and Oceans Canada, 2025; Seabird CTD SBE-25). To capture the mean conditions near the basin floor, we derived mean values for temperature, salinity, and oxygen between 60 m water depth and the bottom (or the deepest available measurement). We then calculated the anomalies between consecutive profiles, resulting in a time series of anomalies that reflects changes from week to week in each variable in the deep basin (Fig. 2d).

Total (dissolved + particulate) ²³⁴Th (²³⁴Th_tot) was measured six times at the Compass Station between 2021–2024. Water column samples were collected using 10 L Niskin bottles at 9 discrete depths. From each bottle, a 2 L subsample of unfiltered seawater was taken, spiked with a ²³⁰Th yield tracer, and processed according to the MnO₂ co-precipitation technique (Pike et al., 2005). Following co-precipitation, samples were filtered onto quartz fiber filters (QMA, 25 mm diameter, 1 µm nominal pore) following Cai et al. (2006). After filtration, filters were dried (55 °C), mounted on sample holders, covered with Mylar™ film (1 layer) and aluminium foil (2 layers), and their beta emissions were measured on low-level beta GM multi-counters (Risø National Laboratory, Denmark). Samples were counted until the error was typically <3 %. After >120 d from collection (i.e., after 5–6 half-lives of ²³⁴Th had passed), samples were recounted to determine the non ²³⁴Th beta activity stemming from other radionuclides included in the precipitate, which was subtracted from the first count. After final counting was completed, the chemical recovery of ²³⁰Th was determined by inductively coupled plasma mass spectrometry (ICP-MS) as outlined in Clevenger et al. (2021), yielding a mean recovery of 85±12 % (n=63). The net counting rate was corrected for ²³⁴Th decay and ingrowth, counting efficiency, and chemical recovery. The uncertainties in the ²³⁴Th_tot activities are determined from counting statistics and the propagation of errors in relation to sample processing, including mass and volume measurements, and ICP-MS recovery analysis. The parent ²³⁸U activity is derived from salinity following the relationship given by Owens et al. (2011), (Eq. 1). 1238U±0.047dpmL-1=0.0786±0.045×Salinity=(0.315±0.158) Additional work conducted on 15 January 2025 measured salinity (PSU) and dissolved ²³⁸U activities at multiple locations and depths in the Bedford Basin across a salinity range of 0.1–31.2 (n=26) and yielded a strong correlation with salinity (R2=0.99). At the Compass Station specifically, measured dissolved ²³⁸U activities (at 1, 6 and 50 m) were within 1 % of the activities predicted by the U-Salinity relationship, confirming the applicability of the Owens et al. (2011) relationship at this site.

Seawater ²³⁴Th activities are used to calculate cumulative water column ²³⁴Th fluxes (dpm m⁻² d⁻¹) as follows: 2FluxTh@z=∫0zλTh(238U-234Th)+w∂234Th∂z-u∂234Th∂z-v∂234Th∂zdz Here, ²³⁴Th is ²³⁴Th_tot activity (dpm L⁻¹), ²³⁸U is the salinity derived ²³⁸U activity (dpm L⁻¹; Owens et al., 2011), λTh is the decay constant of ²³⁴Th (0.0288 d⁻¹), w is the upwelling velocity, u is the zonal velocity, v is the meridional velocity, and the ∂ terms are the vertical (∂z), west to east (∂x), and south to north (∂y) gradients. Applying the 1D steady state model, Eq. (2) can be simplified to: 3FluxTh@z=∫0z(λTh(238U-234Th)dz Given the temporal resolution of this study, and the fact that ²³⁴Th profiles were not measured during consecutive weeks–months but rather on a quasi-seasonal scale, the use of a non-steady-state model for ²³⁴Th flux is not feasible, and we apply the steady state condition. The validity of the 1D steady-state assumption is discussed further in Sect. 4.3. The uncertainties of the flux are the propagated measurement error.

Size fractionated particulate ²³⁴Th samples (²³⁴Th_p) were collected by in-situ pumping of ∼30 L at 3 depths (∼20, 40, 60 m) using large volume pumps (McLane Laboratories™). Filter heads were equipped with a 51 µm pore size Nitex filter above a 1 µm pore size QMA filter (pre-combusted). The large particle material (>51 µm) captured on the Nitex screens was carefully rinsed onto QMA filters (1 µm pore size, 25 mm diameter, pre-combusted) using 0.2 µm filtered and ultraviolet (UV) treated seawater. The small particulate material (1–51 µm) captured on the QMA filter was subsampled by cutting out a 25 mm diameter punch. After drying (55 °C) all particulate samples were mounted onto sample holders and analyzed on beta counters (see above). Following the second counting, particulate samples were prepared for POC measurement by fumigation under HCl vapor for 6 h inside a glass desiccator to remove the carbonate phase. Samples were dried at 50 °C and packed into tin cups and then measured on an Elemental Analyzer (Elementar MicroCube & Isoprime 100). For each pump cast, an additional filter head was attached to the deepest pump to obtain a seawater process blank (“dipped blank”; Lam et al., 2015). The dipped blanks were exposed to seawater for the same amount of time and processed as regular samples. Dipped blank filters were used as process blanks for blank subtraction, detection limit determination, and uncertainty calculations. The POC flux can be calculated with either (a) the POC : ²³⁴Th ratio on small particles, or (b) the POC : ²³⁴Th ratio on large particles using Eq. (4). The fluxes reported using either (a) or (b) represent two alternative approaches to estimating flux, and should not be considered additive, but rather complementary perspectives on export. 4POC Flux=FluxTh@z×POC234Th(a),(b) To investigate the role of resuspension in this study and in the Bedford Basin as a whole, the residual beta activity of particulate ²³⁴Th (Ra_P234) was analyzed. Ra_P234 is a novel resuspension proxy (Lin et al., 2016) and has previously been applied to investigate particle dynamics in coastal sites. Briefly, the Ra_P234 signal comes from “supported” radionuclides bound to particle components like biogenic or lithogenic materials. Terrigenous particles retain radionuclides such as ²³⁴Th (from ²³⁸U decay) and ²¹²Bi (from ²²⁸Th decay), contributing to β counting due to their long half-lives. In contrast, “unsupported” ²³⁴Th is adsorbed in the water column and is the focus of this (and other studies) using the 234Th/238U method to quantify carbon fluxes. The supported Ra_P234 signal indicates sediment resuspension, potentially biasing particle flux measurements. Ra_P234 was obtained from the second counting rates of ²³⁴Th_p after 5–6 half-lives (>120 d) passed, and calculated as: 5RAP234=np2-n0ε∨ Where np2 is the second β counting rate of ²³⁴Th_p (mean 0.3±0.07 counts per minute (cpm)), n0 is the blank mean (0.24±0.02 cpm), ε is the detector efficiency (0.38), and V is volume of seawater pumped (L). The detection limit is defined here as two times the standard deviation of the blank and corresponds to 128.2 dpm m⁻³.

For the measurement of ²³⁴Th in the sediment, a KC Denmark™ Multicorer was deployed at the Compass Station following water column sampling. Sediments were processed either the same day of retrieval or stored at (4 °C) for 24–48 h prior to processing. Cores were sliced in 0.5–1 cm thick sections and then dried at 55 °C for approximately 24 h. After this period, they were weighed and returned to the oven and reweighed at regular intervals until a constant weight was achieved. To correct for residual salt content, salt contribution from porewater was estimated from the salinity of the overlying water and subtracted from the measured dry weight. Once dry, sediments were manually homogenized and weighed into a petri-dish. Analyses of sedimentary ²³⁴Th and ²³⁸U activities were carried out by non-destructive gamma counting on High-Purity Germanium (HPGe) detectors (Dartmouth College, USA) with ∼15 g of sediment. The raw ²³⁴Th activities (at 63.3 keV) in sediments were decay corrected to time of collection. Unsupported or “excess” ²³⁴Th (²³⁴Th_xs,0) is ²³⁴Th that is not in secular equilibrium with the ²³⁸U parent present in the uppermost sediments, presumably because it was supplied by settling particles. Supported ²³⁴Th, in equilibrium with ²³⁸U present in the sediment, is identified at the core depth below which no further decrease in ²³⁴Th activities is observed (“Equilibrium Depth”; EQ). The supported activity is subtracted from the total activity to derive ²³⁴Th_xs,0 (Eq. 6). 6234Thxs,0=Decay Corrected 234Th-(Supported 234Th)(dpmg-1) Sediment inventories of ²³⁴Th_xs,0 were multiplied by dry bulk density, and the decay constant of ²³⁴Th (0.0288 d⁻¹) and integrated from 0 to the EQ to derive the sediment accumulation flux of ²³⁴Th_xs,0 on the sea floor (Eq. 7).

Finally, to derive the sediment accumulation flux of POC, following Eq. (4), the ²³⁴Th sediment accumulation flux (Eq. 7) are multiplied by the water column (a) small and (b) large particle POC : ²³⁴Th ratios measured at 65 m, the depth which represents particles that are about to settle on the seafloor (70 m). The POC : ²³⁴Th ratios directly measured in the sediment are not used as ²³⁴Th decays away faster than organic carbon is respired in sediments. Total organic carbon (TOC) content in sediments was measured on an Elemental Analyzer (Elementar MicroCube) at Dalhousie University. Previous work on Bedford Basin sediments compared acid-fumigated and non-acid fumigated samples and found that the mean carbon content did not differ between the acid fumigated (5.5±0.5 %, n=29) and non-acid fumigated (5.7±0.8 %, n=39) samples (Black et al., 2023). This published finding is consistent with additional unpublished measurements that were made over the years at Dalhousie University. Collectively, the data imply that most of the sedimentary carbon is organic in nature, and thus acid fumigation was not done here. 7Sediment Accumulation Flux 234Th=∫0EQλTh(234Th-238U)dz=∫0EQλ(234Thxs,0)dz

Hydrographic conditions in the Bedford Basin (2021–2024) generally align with long term trends (Li and Harrison, 2008; Haas et al., 2021). Winter and spring surface cooling reduce summer stratification and drive vertical convection (Fig. 2a). However, 2021 saw higher sea surface temperatures (SSTs), leading to stronger stratification (Rakshit et al., 2023), and a lack of shelf water intrusions (Fig. 2e). That year, bottom waters experienced a six-week oxygen minimum, in contrast with records from 2014–2020 when mixing and intrusions increased dissolved oxygen in the Bedford Basin (Haas et al., 2021). The absence of winter mixing in 2021 may be linked to warmer air temperatures from January–March 2021 (-0.2 °C), compared to the long-term mean (Environment and Climate Change Canada, 2018) at this site (-1.5 °C). A spring phytoplankton bloom followed the cessation of winter mixing (Fig. 2d), with chlorophyll maxima at 5–10 m depth. The 2021 bloom, as denoted by chl a signals, was relatively weak, (11.1 mg m⁻³) compared to stronger blooms observed in 2022 and 2023. The largest spring bloom sampled in this study occurred in 2023 (34.9 mg m⁻³). Deep chl a signals are observed down to 70 m depth in some sampling events (May 2023). A secondary fall bloom occasionally occurred, including one in late summer 2021 (15.4 mg m⁻³ chl a). Two intrusion events were observed throughout 2021–2024 (November 2022, January 2024; Fig. 2e).

Throughout the water column, ²³⁴Th_tot activities displayed large deficits (²³⁴Thtot=0.1–1.1 dpm L⁻¹) compared to ²³⁸U activities (mean = 2.09±0.04 dpm L⁻¹) at all sampling events and all depths (Fig. 3). This indicates that the particle flux in Bedford Basin was large enough in the entire water column to scavenge ²³⁴Th prior to its decay. Similar observations have been made in the few other studies from coastal regions (e.g., Luo et al., 2014; Wei and Murray, 1992; Black et al., 2023; Lepore et al., 2007). This contrasts with open ocean sites where ²³⁴Th deficits are typically restricted to the euphotic zone (Owens et al., 2015). Generally, in each season, consistency is seen between years, with some variation that could be due to differences in bloom timing, productivity, etc. The largest deficit, with ²³⁴Th activities as low as 0.1 dpm L⁻¹, occurred during the large bloom event in May 2023 (Fig. 2d), suggesting enhanced scavenging during this time. Winter profiles (Fig. 3c) are more uniform with depth compared to spring and summer, and despite low productivity during these times (Fig. 2d), still show a pronounced deficit throughout the entire water column.

Along with ²³⁴Th_tot, size fractionated particulate ²³⁴Th (²³⁴Th_p) were obtained at approximately 20, 40, 60 m depths by in-situ pumps. At all depths, the small (1–51 µm) particle fraction dominated the ²³⁴Th_p activities (93 %) and POC concentration (77.4 %), suggesting a large amount of POC is suspended on particles in the water column (Fig. 4a). POC on the small particle fraction had a mean of 4.5±3.1 µM, while POC on the large particle fraction had a mean of 1.0±1.2 µM. Mean ²³⁴Th_p activities on the small size fraction (0.8±0.3 dpm L⁻¹) were higher than those on large particles. Mean ²³⁴Th_p activities on large size fraction particles were 0.06±0.04 dpm L⁻¹ (Fig. 4b) and were highest during bloom events. This value is significantly larger than first measurements of ²³⁴Th_p in the Bedford Basin (0.008±0.006 dpm L⁻¹, Black et al., 2023), likely due to a difference in methods (see Sect. 4.4). Although dissolved ²³⁴Th (²³⁴Th_diss) was not measured in this study, subtracting ²³⁴Th_p from the ²³⁴Th_tot at the same approximate depth provides an estimation. The activity of dissolved ²³⁴Th in Bedford Basin is consistently low and nearly devoid (0–0.6 dpm L⁻¹), a finding that is similar in other coastal fjords (e.g. Luo et al., 2014). Hence, the majority of ²³⁴Th_tot is in the ²³⁴Th_p pool, consistent with other studies in the Bedford Basin (Black et al., 2023; Niven et al., 1995). The residual activity of particulate ²³⁴Th (Ra_P234) reveal that residual beta counts were higher than the detection limit in 38 % of the small particulate samples, but no counts exceeding the detection limit were found on the large particles (Fig. 9).

In general, POC : ²³⁴Th on both the large and small particle fractions are <20 µmolCdpm-1 (Fig. 5) and were variable. On average, higher POC : ²³⁴Th ratios on the larger particles were observed compared to the smaller particles (p<0.01, n=18), consistent with open ocean studies (Charette and Moran, 1999). This trend was most driven by two distinct high ratios on the large fraction in the 0–20 m depth range (May 2023, August 2023), likely driven by a strong chl a signal at these times (Fig. 2d). The POC : ²³⁴Th ratio on the small and large fraction had a mean of 5.8±4.1 (0.2–13.2) µmolCdpm-1 and 16.2±14.3 (range 1.3–67) µmolCdpm-1 respectively. The POC : ²³⁴Th ratio generally decreased with depth on the large fraction, but not on the small fraction which showed no pattern with depth (Fig. 5a). On average, the large fraction POC : ²³⁴Th ratios are a factor of ∼2.8× larger than the small POC : ²³⁴Th ratios.

This study demonstrates the utility of coupled water column-sediment ²³⁴Th measurements in assessing coastal carbon fluxes using a 1D steady-state model. To the best of our knowledge, few studies have applied a coupled water column-sediment approach to explore the 234Th/238U method in support of coastal carbon budget quantifications (Cochran et al., 1995; Charette et al., 2001; Lepore et al., 2007; Turnewitsch and Springer, 2001). The present study, along with the first coupled measurements in Bedford Basin made in 2019 (Black et al., 2023), demonstrates a close match between the flux at the bottom of the water column and the uppermost sediments beneath it. Overall, the mean (± s.d.) 2021–2024 sediment accumulation of ²³⁴Th of 2261±512 matches within error to the mean water column flux (interpolated to 70 m) of 2851±249 (Fig. 6a). A paired t test revealed no statistically significant difference between the water column and sediment fluxes, with a mean difference (± s.d.) of 349±747 dpm m⁻² d⁻¹ (t=1.04; df = 4, p=0.30). When data from 2019 (Black et al., 2023) are included, the mean difference (± s.d.) is even smaller (184.1±836) and not statistically significant (t=0.6; df = 7, p=0.5), suggesting that the observed difference may be due to random variability. The mean ratio between water column and sediment ²³⁴Th fluxes is 1.1 indicating close agreement between the two, with the water column flux being slightly higher than the sediment flux. This finding implies that on the timescales of weeks to ∼100 d, no major particle loss is occurring, and particles are generally not being exported out of the basin. A similar finding was first made by Cochran et al. (1995) who found a water column-sediment inventory match within a factor of 2 in both 1992 and 1993 on the Northeast Water Polynya off Greenland. Likewise, in their comparison of model predicted ²³⁴Th water column inventory vs sediment inventories in the upper 0–5 cm of the Gulf of Maine, Charette et al. (2001) find that inventories were within a factor of 2 of the predicted values, with a ratio of 1.1. These findings suggest that Bedford Basin is a promising location for more complex observational studies of particle dynamics, and opens the opportunity to study other short-lived radioisotopes, such as Bismuth-210 (²¹⁰Bi; t1/2=5 d) to study particle dynamics on a finer temporal resolution (hours to days; see Yang et al., 2022).

A notable feature of this study is the persistent deficit of ²³⁴Th relative to ²³⁸U in all seasons sampled. Similar findings have been reported in other shallow basins (Luo et al., 2014; Charette et al., 2001; Evangeliou et al., 2011; Amiel and Cochran, 2008). This persistent deficit suggests that (i) particle scavenging of ²³⁴Th is occurring throughout the entirety of the water column (0–70 m) and (ii) the scavenging rate is high enough to continually maintain a lower ²³⁴Th activity relative to ²³⁸U. Particles in the Bedford Basin may originate from both biological processes in the upper 30 m (Fig. 3), and from resuspension.

Resuspension from the seafloor could be a source of particles to the water column, though multiple lines of evidence indicate it is likely minimal at the Compass Station. First, sedimentary POC : ²³⁴Th ratio is an order of magnitude higher than those in the water column (Fig. 8b), likely due to the temporal decay of ²³⁴Th exceeding the remineralization of POC. Bottom resuspension would be expected to introduce particles with these elevated POC : ²³⁴Th which is not reflected in deep water column POC : ²³⁴Th ratios from this study. Second, recent measurements of physical properties in the Bedford Basin report very weak near-bottom currents (typically <1–2 cm s⁻¹), even during tidal cycles at the Compass Station. Other coastal studies in shallow systems have indicated that near-bed velocities of 20–30 cm s⁻¹ are generally required to erode bottom sediments (Ziervogel et al., 2021). These findings support negligible resuspension from the seafloor at the Compass Station. However, given the concave bathymetric profile of the Bedford Basin (Fig. 1), resuspended sediments could be “funneled” into the Compass Station from the sides (Trimble and Baskaran, 2005). In this study, we observe a strong signal of resuspension at mid depths in April 2021, as well as a weaker signal of resuspension at 60 m in the summer and winter sampling events (Fig. 9), though only on the small particle fraction. This Ra_P234 signal provides indication that some lateral resuspension of fine particles is likely occurring in the Bedford Basin and is reason to report POC fluxes using the large particle fraction. Furthermore, deep and extensive CTD-derived chl a fluorescence signals (Fig. 2d) suggest that biological processes could also be responsible for particle creation and continued scavenging in deeper waters. These features are not unique (Black et al., 2023), and fluorescent organic material has been visually observed at depths >3000 m in the Arctic as a result of a rapid bloom-collapse cycle (Boetius et al., 2013). While additional data (e.g. Underwater Vision Profilers (UVP), net primary production (NPP), backscatter, etc.) are needed to identify the source for the evident fluorescence signals at depths much deeper than expected given the high particle load (i.e. elevated chl a and POC: ²³⁴Th) in the water column, these features could be the key to understanding the persistent Th deficits in the Bedford Basin.

The persistent deficit is also strong in boreal winter periods, when production is presumed to be at its lowest. As argued by Hargrave and Taguchi (1978), inflows of dense surface water from intrusions could deliver additional organic material that could drive the moderate POC fluxes we see at this time. Therefore, the POC flux in January 2024 could be attributed to the intrusion event occurring around the same time. However, our sampling in January 2022 still yields a modest flux value of 10.6±1.5 mmol C m⁻² d⁻¹ at 65 m. This is comparable to January 2024 (Fig. 7) but took place after a period of water column stability (Fig. 2e). A notable feature of this 2022 winter event is that POC fluxes are higher when using the small fraction to compute flux, yielding 17.7±1.3 mmol C m⁻² d⁻¹ (Fig. 7a). Small phytoplankton (<20 µm) can play an important role in carbon export and biogeochemical cycling and have been shown to contribute substantially to biomass in the North Atlantic, with some species especially dominant in subpolar regions during winter months (Bolaños et al., 2020). Robicheau et al. (2022) have shown that specific phytoplankton types have higher abundance values, especially in the early winter in the Bedford Basin when there are temperature anomalies. Primary production was not measured in this study, but modelling efforts have shown that the NPP in the Bedford Basin is in the 100s of mmol C m⁻² d⁻¹ at peak times, and 10s of mmol C m⁻² d⁻¹ otherwise (Black et al., 2023), suggesting that lower NPP in winter months can still yield modest POC fluxes. This was also observed in the nearby Gulf of Maine (Charette et al., 2001) where POC fluxes of 14±0.8 mmol C m⁻² d⁻¹ were measured in the lower range of their NPP measurements (26.4 mmol C m⁻² d⁻¹). These small cells could possibly be the driver of this distinct winter flux (most prominent in the small particle fraction) that we see in January 2022, though the reasoning behind the winter flux remains ambiguous. These results highlight the dynamic particle processes in Bedford Basin that maintain persistent ²³⁴Th deficits, emphasizing the need for continued year-round, high-resolution particle and radionuclide measurements.

Steady-state (SS) dynamics are often assumed in coastal and open ocean 234Th/238U disequilibrium studies (e.g. Coale and Bruland, 1985; Foster and Shimmield, 2002) due to sampling constraints. Fluxes in coastal zones can however be impacted by lateral advection and non-steady state conditions, though the absolute value of each of these impacts on the total flux of ²³⁴Th is difficult to evaluate without greater spatial and temporal resolution of ²³⁴Th sampling. Thus, the SS model is applied here, and we explore the implications of NSS factors in this section.

In this study, we assume that ∂234Th/∂t is equal to 0 and the system is in steady state. The SS model assumes that (i) ²³⁴Th activities and removal rates are near constant with respect to ²³⁴Th decay, (ii) the contributions of vertical and horizontal advection of ²³⁴Th are minor, and (iii) scavenging is irreversible. Physical studies of the Bedford Basin have generally shown that 1D assumptions hold for most of the year (Burt et al., 2013), and modelling efforts have demonstrated that ∼80 % of particles that originate in the basin stay in the basin (Shan and Sheng, 2012). The circulation pattern of Halifax Harbour is characterized as a two-layer, estuarine-type system, with a fresher upper layer flowing seaward, driven by inflows from the Sackville River, and a saltier deep-water return flow (Fader and Miller, 2008). However, with a mean surface outflow of 0.2 cm s⁻¹, this circulation is weakest in the Bedford Basin (Burt et al., 2013). Despite an inflow from the open Atlantic, the presence of a sill largely prevents mixing below 20 m in the basin (Shan et al., 2011). Although our observations come from a single, centrally located station, we acknowledge that lateral heterogeneity and edge effects may influence ²³⁴Th and POC dynamics in the Bedford Basin, especially from 0–20 m. To assess this, we evaluated a surface lateral ²³⁴Th contribution using our 0–20 m mean ²³⁴Th activity, a 3 km cross-basin scale, and observed basin current velocities (0.2 cm s⁻¹). For a theoretical 50 % cross basin difference in ²³⁴Th activities, the lateral ²³⁴Th flux contribution is ∼346 dpm m⁻² d⁻¹, which is ∼52 % of our mean surface cumulative flux, and ∼14 % of the full water column flux to 65 m. This suggests that even under a strong 50 % contrast, the full water column budget changes only modestly. These lateral ²³⁴Th flux estimates are intended as a first order, illustrative calculation, and real cross-basin gradients ²³⁴Th activities (if any) should be measured for a more robust estimate of lateral contributions in future studies.

NSS impacts (rapid temporal changes in ²³⁴Th activities) have been shown to play a role on flux estimates during phytoplankton blooms, and prior studies have shown that NSS terms during blooms can alter ²³⁴Th export fluxes often by 10 %–50 % (Buesseler et al., 1992; Ceballos-Romero et al., 2018). To assess how these departures from SS may impact our reported fluxes, we applied theoretical NSS sensitivities to our spring profiles. Assuming a modest NSS term (0.003 dpm L⁻¹ d⁻¹), the NSS correction is 7.8 % of our reported SS ²³⁴Th flux. Under an upper bound scenario with a larger NSS term (0.01 dpm L⁻¹ d⁻¹), the resulting NSS correction increases to ∼25.8 % of the spring SS estimate. These theoretical corrections show that though minor, NSS effects can become significant during rapid bloom events. Periodic deep vertical mixing, such as during an intrusion event could increase the potential to disrupt the validity of the SS model. Indeed, we do see some indication of mixing multiple times throughout the study period (Fig. 2e), however, no major lines of evidence in the ²³⁴Th activities are indicated in our study (no major Ra_P234 signals, or significant difference in POC : ²³⁴Th), and despite intrusion events, the ²³⁴Th_tot activities are similar between January 2024 (intrusion) and January 2022 (no intrusion). These observations may suggest that the Bedford Basin rapidly re-establishes a ²³⁴Th deficit following intrusion or mixing events and implies that particle scavenging and export are strong and continuous. If feasible, future studies in the basin should sample on a finer temporal resolution (1–4 weeks) to better understand NSS contributions in the basin.

A key source of uncertainty in estimating POC fluxes using 234Th/238U disequilibrium is the POC : ²³⁴Th ratio used for filtered particles, which has led to inconsistent flux estimates in previous studies (Lalande et al., 2008). Determining the most representative POC : ²³⁴Th ratio remains an active area of research (e.g. Buesseler et al., 2006; Puigcorbé et al., 2020) to ensure accurate flux measurements. In the Bedford Basin, we find the primary source of variability in POC flux estimates is the POC : ²³⁴Th ratio of large (>51 µm) particles. Previously, Black et al. (2023) reported ratios in the Bedford Basin >900 µmoldpm-1, yielding exceptionally large fluxes (∼2100 mmol C m⁻² d⁻¹, Table 2), over an order of magnitude greater than predicted coastal carbon export models (Siegel et al., 2014). By contrast, this present study finds lower, more modest fluxes (<70 mmol C m⁻² d⁻¹), consistent with other coastal studies (<200 mmol C m⁻² d⁻¹, Luo et al., 2014; Cai et al., 2015; Cochran et al., 1995). High POC : ²³⁴Th ratios (>100 µmoldpm-1) have been previously linked to seasonal variability (Charette et al., 2001), however, the extreme values in Black et al. (2023) likely result from the differences in particle collection and delayed counting of beta emissions. Variations in sampling approaches highlight the potential for significant differences in POC measurements. Black et al. (2023) used a vacuum filtration system on Niskin-bottle collected samples, while this study employed McLane™ Large Volume Pumps that filter in-situ. In their comparative study (Graff et al., 2023) found that in-situ pump sampling resulted in substantially different POC concentrations compared to Niskin sampling at similar depths. This difference may arise due to variability in filtration speed, particle solubilization (Buesseler et al., 1995), and size fractionation (Hung et al., 2012). The methodological differences suggest that the elevated large particle POC : ²³⁴Th ratios reported by Black et al. (2023) likely result from sampling artifacts rather than true variability in particle flux. The ratios obtained in this study provide a more representative measure of large particle POC : ²³⁴Th at this site. To improve comparability, future studies in the Bedford Basin and other coastal regions should adopt consistent sampling methods to minimize methodological biases.

The POC export fluxes derived in this study are in good agreement with other estimates of carbon fluxes taken in the Bedford Basin (Table 3). The mean depositional organic carbon flux (at 65 m), derived here from the large particle POC : ²³⁴Th ratio and integrated ²³⁴Th flux was 20±14 mmol C m⁻² d⁻¹ (Fig. 7), two orders of magnitude lower than first estimates of ²³⁴Th derived POC flux (Black et al., 2023). This is a similar range as the nearby Gulf of Maine (34±44 mmol C m⁻² d⁻¹; March, June, September; Charette et al., 2001), Saanich Inlet in western Canada (11.5±4 mmol C m⁻² d⁻¹; Luo et al., 2014), Greenland continental shelf (25±14 mmol C m⁻² d⁻¹; Cochran et al., 1995), and Norwegian fjords (21.9–24.4 mmol C m⁻² d⁻¹; Wassmann, 1984). Our measured values are also similar to model estimates of organic carbon deposition flux in the Bedford Basin (25.2 mmol C m⁻² d⁻¹; Rakshit et al., 2025), and sediment trap estimates at 60 m (17 mmol C m⁻² d⁻¹; Hargrave and Taguchi, 1978). Translated to an annual molar estimate, (assuming our spring, summer, winter mean is representative of the year, and multiplying the mean daily flux by 365), our study reveals that 7.3±5.1 mol C m⁻² yr⁻¹ is delivered to the seafloor annually, similar to annual mean trap estimates in the Bedford Basin of 6.2 mol C m⁻² yr⁻¹ (Hargrave and Taguchi, 1978). The modeled daily benthic carbon flux in the Bedford Basin of 19 mmol C m⁻² d⁻¹ (Black et al., 2023), is equivalent to an annual export of ∼7 mol C m⁻² yr⁻¹, also consistent with our ²³⁴Th-based estimates.

Using the mean wt % POC in the upper 5 cm of the sediment cores (5.7 %), mean sedimentation rate (0.2 cm yr⁻¹; Black et al., 2023) and mean dry bulk density (275 kg m⁻³), we find an organic carbon burial rate of 6.9 mmol C m⁻² d⁻¹. This burial rate, combined with our mean POC depositional flux at 65 m, implies that the carbon burial efficiency is ∼33 %, larger than previous estimates at this site (10 %; Black et al., 2023), and similar to other coastal zones (Faust and Knies, 2019). The remaining ∼67 % of the depositional POC is remineralized before burial, a finding consistent with measured dissolved inorganic carbon (DIC) fluxes from the sediments (15.6 mmol DIC m⁻² d⁻¹; Rakshit et al., 2025). These findings suggest that the Bedford Basin is an important site of organic carbon sequestration. To better understand C dynamics in the ocean, carbon flux metrics have been developed that help define the efficiency of the biological carbon pump in marine environments (Buesseler et al., 2020; Buesseler and Boyd, 2009). The thorium export efficiency ratio (ThE-ratio) is the ratio of POC flux (based on ²³⁴Th measurements) relative to NPP in the euphotic zone. To estimate the ThE-ratio in the Bedford Basin, we divide the annual mean flux of POC measured in this study by the annual mean primary production in the basin. Primary production is rarely measured directly in the Bedford Basin with only one record of NPP from incubation studies (50 mmol C m⁻² d⁻¹; Platt, 1975). Black et al. (2023) approximated an annual mean NPP of 66 mmol C m⁻² d⁻¹ (range 0.4–184 mmol C m⁻² d⁻¹) using a model based on in-situ chl a and PAR data, which falls within regional estimates from the nearby Gulf of Maine (26–399 mmol C m⁻² d⁻¹; Charette et al., 2001). Using this NPP and our mean POC export at the base of the euphotic zone (∼20 m; 29.4 mmol C m⁻² d⁻¹), the export efficiency (ThE) is 44.5 %. Applying a ±20 % sensitivity to NPP shifts ThE to ∼37 %–56 %. This ThE ratio is comparable to other coastal sites (26 %, Charette et al., 2001; 50 %, Falkowski et al., 1988; 60 %, Cochran et al., 1995). A ThE ratio of 44.5 % suggests moderate export efficiency at our site. It is, however, important to note that our estimation relies on modeled NPP values and updated in-situ measurements of NPP would yield a more precise value of these ThE estimates.