Turbidity Currents Can Dictate Organic Carbon Fluxes Across River‐Fed Fjords: An Example From Bute Inlet (BC, Canada)

The delivery and burial of terrestrial particulate organic carbon (OC) in marine sediments is important to quantify, because this OC is a food resource for benthic communities, and if buried it may lower the concentrations of atmospheric CO2 over geologic timescales. Analysis of sediment cores has previously shown that fjords are hotspots for OC burial. Fjords can contain complex networks of submarine channels formed by seafloor sediment flows, called turbidity currents. However, the burial efficiency and distribution of OC by turbidity currents in river‐fed fjords had not been investigated previously. Here, we determine OC distribution and burial efficiency across a turbidity current system within Bute Inlet, a fjord in western Canada. We show that 62% ± 10% of the OC supplied by the two river sources is buried across the fjord surficial (30–200 cm) sediment. The sandy subenvironments (channel and lobe) contain 63% ± 14% of the annual terrestrial OC burial in the fjord. In contrast, the muddy subenvironments (overbank and distal basin) contain the remaining 37% ± 14%. OC in the channel, lobe, and overbank exclusively comprises terrestrial OC sourced from rivers. When normalized by the fjord’s surface area, at least 3 times more terrestrial OC is buried in Bute Inlet, compared to the muddy parts of other fjords previously studied. Although the long‐term (>100 years) preservation of this OC is still to be fully understood, turbidity currents in fjords appear to be efficient at storing OC supplied by rivers in their near‐surface deposits.

Fjord seafloors can, however, be highly heterogeneous (Bianchi et al., 2020;Smeaton & Austin, 2019) and comprise diverse subenvironments including submarine channels, and associated overbanks and lobes (Conway et al., 2012;Pope et al., 2019;Zeng et al., 1991). Such subenvironments are created by submarine sediment density flows, called turbidity currents. These flows distribute sediment and OC across the fjord floor, connecting river mouths to the deeper parts of fjords, and fractionate sediment and OC by grain size and density en-route. Only a handful of studies have discussed how OC is distributed and fractionated within fjords (Bianchi et al., 2020;Cui, Bianchi, Jaeger, et al., 2016;Cui, Bianchi, Savage, et al., 2016;Cui et al., 2017;Hage et al., 2020;Smeaton & Austin, 2019). For example, a recent study of Scottish fjords revealed that muddy (<63 μm) sediments held the largest amounts of OC (∼2.6 Mt) compared to sandy sediment (∼0.26 Mt; Smeaton & Austin, 2019). While the OC budgets are known in the Scottish fjords, the processes determining these budgets remain uncertain, partly because of a lack of bathymetric data and knowledge of submarine flow processes in these settings (e.g., these Scottish fjords may not contain active turbidity currents). In contrast, a study of Bute Inlet (a fjord in British Columbia, Canada) highlighted that the sandy parts of a submarine channel formed by turbidity currents held large amounts of OC . Higher OC content in sandy submarine channels compared to the nonchannelized environments was also observed in Steffen Fjord (Chile; Vanderkerkhove et al., 2020). However, the river-derived OC inputs and burial rates in different subenvironments created by turbidity currents were not considered in that previous study of Bute Inlet, nor in any other fjord. Given the large variation and uncertainty in OC burial efficiency between fjord subenvironments, it is important to assess this variability more accurately to estimate global OC budgets (Burdige, 2007;Smith et al., 2015) and to improve paleoclimate/environmental reconstructions based on sediment cores in fjords (Bianchi et al., 2020).
Here for the first time, we quantify the amount and type of OC supplied by two rivers and its distribution in the surficial sediments (top 2 m) of different subenvironments (e.g., channel floor, overbanks, lobe, and distal basin) across a river-fed fjord. This study then shows how OC is unevenly distributed within a fjord and illustrates how OC distribution and burial is dependent on turbidity current processes. Insights into how OC is distributed have wider implications for understanding OC burial by turbidity currents in locations other than fjords, for which there are also very few detailed OC budgets from river sources to marine sinks. Therefore, we compare our Bute Inlet results with available information on how OC is distributed in sandier or muddier subenvironments of other turbidity current systems, that is, Gaoping (Kao et al., 2014;Liu et al., 2016), Ganges-Brahmaputra (Galy et al., 2007;Lee et al., 2019), and Congo (Baudin et al., 2020). This comparison highlights whether similar or contrasting OC burial patterns emerge more generally.
The main objective of this paper is to constrain OC fluxes from source to sink in Bute Inlet by answering the following questions: (a) how much OC is delivered by the two rivers discharging into Bute Inlet? (b) How much and what type of OC is present in the Bute Inlet subenvironments? (c) What is the terrestrial OC burial efficiency of Bute Inlet and how does it compare with other fjords? Finally, we discuss the study in a wider global context and thus compare our results from Bute Inlet with (nonfjord) deep-sea turbidity current systems.

Homathko and Southgate Rivers
The Homathko and Southgate Rivers feed the head of Bute Inlet (Figure 1). The watersheds of both rivers are supplied by meltwater from snow and glaciers within their catchment (e.g., Homathko has a 19.1% glacier cover, while Southgate has a 33.5% glacier cover) leading to sustained high flow in summer (Giesbrecht et al., 2022). An estimated 1,684 and 2,089 mm of mean annual precipitation (∼60% as snow) occurred between 1981 and 2010  in the Homathko and Southgate Rivers, respectively (Giesbrecht et al., 2022). Rainforest vegetation and soils predominate below treeline on the Pacific side of the Coast Mountains (Meidinger & Pojar, 1991). The river floodplains are populated by coniferous and deciduous forests. These two rivers provide 94% of the freshwater and sediment inputs into the fjord, with the remainder of the freshwater being provided by small streams entering the fjord at its margins . The watershed areas are 5,782 and 1,985 km 2 for Homathko and Southgate Rivers, respectively (Figure 1;Gonzalez Arriola et al., 2018). The Homathko River has an average annual discharge of 250 m 3 /s, with an average summer peak discharge of ∼800 m 3 /s (Water Survey of Canada, 2020). The average annual suspended sediment flux of the Homathko River is 30 kg/s, whereas its bedload discharge is 99 kg/s (Table 1; . There was no gauging station in the Southgate River up to 2021. However, recent (2021) data and the covariation between daily discharge on the two systems allowed the Southgate discharge (water and total suspended load) to be modeled for our years of interest (Table 1; Texts S1-S4 in Supporting Information S1).
During the spring and summer meltwater period (freshet), the rivers create large sediment-laden freshwater plumes at the fjord's head (Syvitski et al., 1985;Tabata & Pickard, 1957). These plumes flow on top of the denser more saline fjord water, thereby bringing terrestrial particles into the fjord and enhancing the heterotrophic activity of bacterioplankton (Albright, 1983). Due to the relatively large tidal range (up to 5.5 m), sediment from these river plumes is likely to be concentrated in a turbidity maximum that is located in between the fresh and saline water (Dyer, 1997). Observations in a similar and nearby fjord setting (Howe Sound, Canada) show that during the freshet these turbidity maxima can hold enough sediment to become denser than the saline fjord water (Hage et al., 2019). During spring tides, the river plume becomes more focused and powerful, so that low tides cause these dense turbidity maxima to migrate onto the steep delta slope, where they can trigger turbidity currents (Hage et al., 2019). These processes, together with submarine failures of the delta slopes, result in turbidity currents frequently being triggered in such river-fed and tide-influenced fjord settings (Clare et al., 2016;Hizzett et al., 2018). It is likely that similar initiating mechanisms for turbidity currents also occur in Bute Inlet (Hughes Clarke et al., 2015).

Bute Inlet Turbidity Current System
Downstream of the Homathko and Southgate Deltas lie two submarine channels, which merge into a single sinuous channel, formed by turbidity currents (Prior et al., 1986(Prior et al., , 1987Figures 2 and 3). The channel is incised into the seafloor to a depth of ∼20 m on average. The channel floor is flanked by terraces that are particularly well developed in the first 10 km of the fjord. Further downstream (430-580 m water depth; Figure 3), the submarine channel becomes less incised and is flanked by levees that are variably developed (Prior et al., 1987). For simplicity, in this paper, we use the term "overbanks" (Figure 3) to describe both the muddy terraces and levees bounding the submarine channel floor in Bute Inlet along its entire length.
The sediment composition of the Bute Inlet turbidity current system has been described in older work, based on sediment cores (Prior et al., 1986;Zeng et al., 1991), yet the OC composition of these cores was not documented until recently. The OC stored within the submarine channel floor was shown to be composed mostly of young terrestrial woody debris buried within fine sands . In this study, we build on the findings of  Hage et al. (2020) from the channel floor, by documenting the distribution of OC across all of the subenvironments within this turbidity current system (i.e., channel floor, overbanks, lobe, and distal flat basin). We also build on previous work on repeated bathymetric surveys and core dating (Heijnen et al., 2022;Syvitski et al., 1988) to derive SRs and OC burial fluxes across the fjord.

River Sampling
We used 22 samples to characterize OC supply from the rivers. These samples cover a range of grain size classes and environments including the rivers' suspended load (i.e., filtered from the river waters; n = 6), the river banks (n = 3), the river bed (n = 1), the delta-top plains (n = 4), and the river plumes at the fjord surface (n = 8; Figure 2). The sampling campaign was conducted on 26 October 2017 (Figures 2 and 3; Figure S3 in Supporting Information S1), when the Homathko River discharge was 390 m 3 /s (Water Survey of Canada, 2020). This is higher than the 250 m 3 /s average annual river discharge and lower than its ∼800 m 3 /s average summer peak discharges (Water Survey of Canada, 2020).

Fjord Sampling and Definition of Subenvironments
Two field campaigns were conducted in June 2016 and October 2016 to sample the fjord sediments. In total, we use a set of 15 sediment cores, which sampled the upper 30-200 cm of seabed sediment. These cores were located across all subenvironments of the fjord's seabed, including the sandy channel floor and lobe, as well as the muddy overbank and distal flat basin ( Figure 3). We have few constraints on the characteristics/extent of turbidity current overbank flows, so we infer the overbank's external limit as the base of the fjord walls (red outlined area in Figure 3). We note that sediment cores from these overbank areas are located relatively close to the sandy channel. A sediment core at the end of the fjord (650 m water depth) is used to characterize the OC composition for the entire distal flat basin (core 15; light blue outlined area in Figure 3). We have no samples on the steep fjord sidewalls; hence, these areas are not included in our OC analyses.
Logging of the sediment cores allowed sandy and muddy facies to be identified (Figures 3e and 3f). A total of 99 subsamples were collected from the fjord sediment cores, to ensure representative sampling of each of the facies that were present in the channel floor, overbanks, lobe, and distal flat basin. All samples were analyzed for carbon geochemistry, as described below.

Geochemistry on River and Fjord Samples
Total carbon (TC) was measured on all river and fjord samples using an LECO elemental analyzer. All samples were also analyzed for total organic carbon (TOC) content after trace amount of inorganic carbon (carbonate) was removed using HCl and rinsing with milli-Q water. No effervescence was observed during this acidification step, confirming minimal carbonates in the system (Figure 4). TOC measurements allowed for quantifying the total amount of OC delivered by the rivers and buried in the different fjord subenvironments ( Figure 5). In order to constrain the type (terrestrial or marine) of OC, all samples were analyzed for carbon stable isotopes (δ 13 C) using an isotope ratio mass spectrometer ( Figure 6). Terrestrial organic matter derived from C3 plants has an average δ 13 C between −26‰ and −28‰ (Hecky & Hesslein, 1995) and can reach values as low as −31‰, which correspond to young woody debris . In contrast, marine organic matter preserved in northeast Pacific sediments has an average δ 13 C of −21.5‰ (MacDonald et al., 1991).
Previous work in Bute Inlet has used a wider range of methods (e.g., radiocarbon Δ 14 C dating and separation of OC mixtures by ramped pyrolysis-oxidation [RPO]) on selected river and sandy channel samples . We complement these previous detailed measurements with new 14 C and RPO measurements on three samples  Giesbrecht et al. (2022). b Syvitski and Farrow (1983). c See Text S1 in Supporting Information S1 for suspended load estimates in both rivers. d No bedload estimate was found in the literature for the Southgate River. The bedload was thus estimated based on the Homathko River using the ratio between the two river watersheds as a scaling factor (Text S2 in Supporting Information S1). TOC data in river samples are included in Figure 5 and Table S7 in Supporting Information S1.

Table 1 Homathko and Southgate Rivers Characteristics and Estimates of Annual Organic Carbon Fluxes
from the distal subenvironment ( Figure 7). The RPO method identifies OC fractions based on thermal lability by heating each sample from 20°C to 1000°C in an oxygenated carrier gas, thus sequentially combusting OC into CO 2 . The CO 2 collected between temperature intervals ("fractions") is measured for carbon stable isotopes (δ 13 C) and radiocarbon activity ( 14 C/ 12 C). These isotope measurements on separated fractions enabled characterization of the distribution of OC source and age within individual samples (Hemingway et al., 2017).

Calculation of River OC Fluxes
In order to determine the particulate OC fluxes from rivers, we use bedload sediment discharge data published in past literature , Text S2 in Supporting Information S1), recent total suspended sediment measurements conducted on the Homathko and Southgate Rivers (Text S3 in Supporting Information S1), and TOC measurements of river sediments collected in October 2017 (Table 1).
Bedload discharge for the Homathko River was estimated at 3,100 Kt/year by Syvitski and Farrow (1983). We scaled this Homathko bedload estimate to the Southgate River using the ratio of watershed areas (Table 1; Text S2 in Supporting Information S1). The yearly mean suspended load discharge for the Homathko and Southgate Rivers was estimated at 7,946 and 221 Kt/year, respectively. This is based on suspended samples collected approximately monthly between 2018 and 2021 for the Homathko River (Text S3 in Supporting Information S1), together with daily water discharge data (Runkel et al., 2004; Text S3 in Supporting Information S1). There was no water 10.1029/2022JG006824 6 of 16 discharge data available for the Southgate River up to 2021. Therefore, a linear regression was used to estimate daily Southgate River water discharge from Homathko River discharge, similar to Hood et al. (2020). The training data set contained 123 daily flow observations spanning a broad range of flows from record high discharge in late June 2021 through moderately low flows in October 2021 (Text S1 in Supporting Information S1).
Finally, we derived each of the river OC fluxes by multiplying the average percent TOC of river bedload and suspended load samples by the annual bedload and suspended sediment discharge, respectively (Table 1). The percent TOC of the bedload is the average between percent TOC measured on coarser samples collected in the river bed, the river banks, and the river deltas. (e, f) Sediment core set; 30-cm-long cores were collected using a box coring system; 200-cm-long cores were collected using a piston coring system.

Calculation of OC Burial Rates in Fjord Sediments
The OC burial rates (Kt/year) in the fjord are calculated using Equation 1 (Table 2; Baudin et al., 2020) as follows: where A is the surface area of a subenvironment (m 2 ), TOC is the total OC content (%) averaged between samples from a given subenvironment, ρ is the grain density (kg/m 3 ), φ is the porosity (%), and SR is the sedimentation rate (m/year).
SRs typically vary when measured over different timescales (Sadler, 1981), especially in a highly active turbidity current system (Heijnen et al., 2020;Prior et al., 1987), making them challenging to quantify. Thus, SRs were estimated herein based on two independent approaches that provide ranges of SRs (Texts S4 and S5 in Supporting Information S1).
The first approach uses differences between two bathymetric surveys obtained in 2008 and 2018 and thus holds for a decennial timescale (Heijnen et al., 2022; Table 2 and Text S4 in Supporting Information S1). This approach highlights zones of erosion and deposition that migrate upstream significantly (100-450 m per year) due to the presence of active knickpoints in the channel (Heijnen et al., 2020). This explains why the channel is net erosive on a decennial timescale (SR = ∼−16 cm/year). The 30-cm-long sediment cores used in this study thus probably represent deposits that were a few days to weeks old, as direct monitoring shows that over 100 turbidity currents can occur in one freshet (Chen et al., 2021;Pope et al., 2022). The top of these deposits was likely reworked again by turbidity currents in the following days to weeks, progressively moving the sediment down the channel. The channel thus acts as a conduit through which sand and associated OC is shuffled to the lobe, in multiple stages over several weeks to several decades (Heijnen et al., 2022).
Time-lapse bathymetry also shows that the terminal lobe is built up from a small number of large magnitude "channel flushing" events (Heijnen et al., 2022), resulting in a locally high SR (SR = 18 cm/year, Table 2) over decennial timescales. It should be kept in mind that this bathymetric method underestimates the depositional volume in areas of slow deposition (overbanks), as cm-thin drapes of sediment cannot be resolved by this method.
The second approach uses 210 Pb and 137 Cs dating methods applied to sediment cores collected in the Bute Inlet overbank and distal flat basin settings (Heerema, 2021;Syvitski et al., 1988; Text S5 in Supporting Information S1) and thus holds for a centennial timescale. This approach postulates that the entire Bute submarine system is net aggrading over centennial timescales, with SR in the overbanks and distal flat basin being 2 ± 1.5 and 1 ± 0.5 cm/year, respectively (Syvitski et al., 1988). 210 Pb and 137 Cs dating was not possible in the channel and lobe subenvironments as they are too sandy. Instead, we used an assumed relationship between the SR obtained in the overbank and that of the channel and lobe (Equations S4 and S5 in Supporting Information S1; Baudin et al., 2020). Thus, the channel floor is assumed to aggrade at a slow, but positive SR (SR = 0.7 ± 0.5 cm/year), rather than being strongly erosional. This is consistent with a rather shallow (∼20 m relief) channel, as prolonged erosion in a submarine channel that is likely at least hundreds of years old would have carved a much deeper conduit. This slow, but positive, SR in the channel is then assumed to be balanced by less aggradation in the lobe (SR = 10 cm/year; Equation S5 in Supporting Information S1).
In total, we use ranges of SRs for each subenvironment as follows. SR in the channel ranges from −16 cm/year (i.e., erosional) to 0.7 cm/year (i.e., slightly depositional). SR in the lobe ranges from 10 to 18 cm/year, implying a large accumulation of sediment over both decennial and centennial timescales. SR in the overbank area varies between 2 and 2.3 cm/year. Finally, SR in the distal flat basin ranges from 1 to 2.3 cm/year.

Results
Below we provide the OC content and composition for both rivers and the fjord sediments separately. We note that comparison between TC and TOC on all samples revealed the absence of carbonates within both river and fjord samples (Figure 4).

Carbon Composition Supplied by Both Rivers; What Is Coming In?
Coarse sand samples collected from the riverbank and delta areas have relatively low TOC (mean TOC = 0.35%; Table S7 in Supporting Information S1), and δ 13 C values (−27‰ to −28‰) indicating a terrestrial origin (Hecky & Hesslein, 1995). TOC is moderately high (mean TOC = 0.8%; Table S7) in the fine sands sampled from the river suspended load, banks, and deltas; whereas δ 13 C values are low (−25‰ to −29‰) and point again to a terrestrial origin (Hecky & Hesslein, 1995). TOC is highest in muddy sediments (mean TOC = 3.1%, Figure 5) collected in the river plume at the fjord head. δ 13 C signatures for these river plume samples are unusually high (−12 to −20 ‰; Figure 6), despite the absence of carbonates (Figure 4). These high δ 13 C values are interpreted to be linked to bacterioplankton producing extracellular polymeric substances (EPS; Albright, 1983), and this will be further discussed in Section 5.1.

Figure 5.
Facies and total organic carbon (TOC) content within rivers, submarine channel and lobe, overbank, and distal flat basin. Pie charts represent the contribution (in %) of each of the facies to a given subenvironment. The TOC content of the river "bedload" includes samples collected in the river bed, banks, and deltas.
In total, we estimate that about 22.7 ± 13.1 Kt OC/year are delivered annually by the Homathko and Southgate Rivers. This is based on the estimated sediment discharge (suspended and bedload) and on the average TOC content measured between samples collected in both rivers in October 2017 (Table 1).   Figure 3). Black lines show distribution of activation energy (E a ; thermogram; Hemingway et al., 2017). Blue squares show radiocarbon ages (in fraction modern, Fm, a measurement of the deviation of the 14 C/ 12 C ratio of a carbon fraction from "modern"). Red dots show carbon stable isotopes (δ 13 C, in ‰). Blue bars represent the E a range to which each red dot and blue square applies.

Facies and Carbon Composition in the Fjord; What Is Preserved and Where?
We now present the OC content and composition within the fjord seabed, across the four previously defined subenvironments (Figure 3). Sediment cores collected from the sandy channel floor and lobe have similar facies and OC composition (Figure 3, Table S7 in Supporting Information S1), thus the OC composition of these subenvironments is described together. The channel floor and lobe contain 63% carbon-poor (mean TOC = <0.05%) coarse sand, 29% carbon-rich (mean TOC = 3%) fine sand, and 8% carbon-moderate (mean TOC = 0.5%) mud ( Figure 5). δ 13 C and 14 C ages revealed that fine sands contain young terrestrial woody debris (visible to the naked eye), buried under older biospheric soil organic matter in the muds (Hage et al., 2020; Figure 6).
In contrast to the channel floor and lobe, the overbanks are dominated (80%) by muddy deposits characterized by lower TOC values (0.35% on average; Figure 5). Fine sands found in the overbank cores have much lower TOC values compared to fine sands in the channel floor and lobe. δ 13 C signatures range between −26‰ and −28‰, pointing to a terrestrial origin of the organic matter ( Figure 6). We note the absence of woody debris visible to the naked eye in the overbank cores, as opposed to the channel floor and lobe cores.
The top 2 m of sediment recovered from the distal flat basin is made exclusively of muddy sediments characterized by homogeneous red (35%) or gray facies (65%). Gray muds have moderate TOC (0.65% on average; Figure 5) and a wide range of δ 13 C signatures, suggesting a mixed terrestrial and marine origin of organic matter  (Heijnen et al., 2022). Sedimentation rates are obtained by dividing the sediment volume by the surface area of a given subenvironment. Volume uncertainties are based on vertical accuracy of the multibeam surveys of 0.5% of the water depth. Hence, the 600-m-deep lobe is greatly affected (Heijnen et al., 2020). b Assumed sedimentation rates in the channel and lobe are based on c and Baudin et al. (2020); see Supporting Information S1. c 100 years sedimentation rates in the overbank and distal basin are based on 210 Pb and 137 Cs dating (Syvitski et al., 1988;Heerema, 2021; Supporting Information S1).

Table 2 Sediment Budget and Organic Carbon (OC) Fluxes in the Subenvironments of the Bute Turbidite System Over 10 and 100 Years Timescales
(−22‰ to −26‰; Figure 6). Red muds have high TOC (2.5% on average; Figure 5) with δ 13 C between −20.5‰ and −23‰ ( Figure 6) that point to a marine-dominated origin (MacDonald et al., 1991). Based on a binary mixing model using δ 13 C compositions of marine (−20.5‰) and terrestrial (−27‰, Hecky & Hesslein, 1995) OC end-members, we estimate that about 54% of the OC found in the upper 2 m of sediment on the distal flat basin is of marine origin (Text S6 and Table S6 in Supporting Information S1). 14 C data and RPO thermograms on the red and gray muds in this distal site (Figure 7) reveal that the OC in this muddy distal site is of similar age (500-1,100 14 C years), compared to the organic matter found in the muddy deposits of the channel (997 14 C years; Figure 6, Hage et al., 2020). Overall, OC associated with muds in the active channel and distal flat basin are older compared to the young OC associated fine sands found in the active channel ( Figure 6, Hage et al., 2020).
In total, we estimate that the top 2 m of sediments in the Bute turbidity current system have an annual OC burial rate ranging from 15.4 to 17.9 Kt OC/year over decennial to centennial timescales, respectively (Table 2). We divide this budget between terrestrial and marine OC contribution based on the mixing model applied to the distal flat basin samples (Text S6 in Supporting Information S1). In total, between 11.7 and 16.3 Kt OC/year of terrestrial origin carbon are found in the fjord, in all of its subenvironments. Between 1.6 and 3.7 Kt OC/year of marine origin carbon are found in the distal flat basin (Figures 6 and 7 and Table 2).

How Much Particulate OC Is Delivered by the Two Rivers Discharging Into Bute Inlet?
We have estimated that Homathko and Southgate Rivers carried 22.7 ± 13.1 Kt OC/year into Bute Inlet based on suspended and bedload sampling conducted in 2017 (Table 1). This total OC flux is interpreted to contain three pools ( Figure 6): young terrestrial biospheric (woody debris), old terrestrial biospheric (soil organic matter), and petrogenic OC (rock-derived OC). Identification of these pools is based on TOC and δ 13 C signatures (Figure 6), and on ramped oxidation data combined with radiocarbon dates previously presented in Hage et al. (2020). To these three pools previously identified, we add a fourth pool made of EPS associated with bacterioplankton that occurs when the rivers enter the fjord and create surface plumes in which the heterotrophic activity of bacterioplankton can be enhanced (Albright, 1983). This activity appears to result in an OC pool characterized by high TOC (>3%) and unusually high δ 13 C signatures ( Figure 6). Such signatures cannot be explained by the presence of carbonates (typically enriched in δ 13 C) because they were removed prior to isotope measurements. These high δ 13 C signatures may also correspond to C4 plants, yet C4 plants are almost exclusively found in tropical and warm environments (Hecky & Hesslein, 1995). The EPS hypothesis is further supported by the stickiness of the plume samples (i.e., samples sticking to the filters), which is typical for EPS (Underwood et al., 1995). This fourth pool of OC is not included in our OC river input flux as it is most likely formed within the fjord water. Further research on the highly enriched δ 13 C composition of these plume samples is required, yet it is beyond the scope of this study. We further note that the EPS-related OC pool is not found in the fjord seabed sediment samples ( Figure 6). We propose this is because OC associated with EPS is labile and not heavy enough to settle, and it thus degrades rapidly in the water column before reaching seabed sediments (Albright, 1983).
We note that there are key assumptions behind these estimates of riverine OC composition. First, we have a limited number of riverine OC samples (n = 22) and these samples were collected during a rain-induced runoff event in fall when river discharge was near the annual average (Text S1 in Supporting Information S1). River discharge peaks driven by the freshet and/or intense rain likely brought more and coarser terrestrial organic debris than the rates estimated here (as in Turowski et al., 2016). We thus expect these river OC fluxes to fluctuate following extreme events such as floods, yet the data presented in this study do not constrain this variability. Second, we have no constraints on the OC content potentially coming from small streams entering the fjord along its margins. However, these streams only represent 6% of the total sediment and freshwater input coming into the fjord .

How Much and What Type of OC Is Buried in the Bute Inlet Subenvironments?
The surficial sediments from the Bute turbidity current system bury a total of 15.4-17.9 Kt OC/year, with 11.7-16.3 Kt OC/year being of terrestrial origin. Thus, 76%-91% of the OC buried in Bute Inlet is terrestrial. This proportion is higher than previous estimates made in other fjords where 42%-65% of the OC was found to be terrestrial ( probably due to the lower riverine inputs compared to Bute Inlet. This highlights the importance of turbidity currents for distributing terrestrial OC in river-fed fjords and the necessity to include turbidity current processes when assessing the burial of terrestrial OC globally. There are a number of uncertainties behind OC rates calculated in Bute Inlet. First, we do not have sediment cores in the distal parts of the overbanks, close to the fjord's sidewalls. Therefore, we extrapolate the SRs and OC composition found in the overbanks close to the channel to the entire overbank area. The more distal overbank sites may either be poorer in terrestrial OC compared to the sites close to the channel fed by the rivers or richer in OC due to smaller rivers and streams and potential landslides sourced from the steep fjord sidewalls. Second, we assume that the 30-cm-thick (i.e., maximum a few weeks old) deposits recovered from the channel are reproduced over centennial timescales despite the rapidly migrating knickpoints in the channel (Heijnen et al., 2020;Prior et al., 1987). We could expect that channel-floor fine sands, which are rich in OC, are excavated several times before reaching their ultimate burial location on the lobe, decreasing their OC content over time (Heijnen et al., 2022;Pope et al., 2022). However, we do not see any evidence of OC content decrease along the channel transect, at least over the very recent (30 cm below seafloor) deposits observed in this study. We suspect that the channel is an efficient conduit for the delivery of terrestrial OC to the lobe, but this needs to be confirmed with longer sediment cores that are difficult to collect in sandy channel floors (as piston corers tend to fail to penetrate sandy seabed).
Despite these uncertainties, the data collected in Bute Inlet allow us to derive for the first time a detailed OC burial budget in fjord sediments controlled by turbidity currents (Figure 8). Because the channel is net-erosional (i.e., negative OC flux, Figure 8, Table 2) and is balanced by deposition on the lobe over a decadal timescale (i.e., any material in the channel will ultimately be shuffled and buried on the lobe, Heijnen et al., 2022), we provide a combined OC burial contribution for these two environments. Together the channel and lobe, dominated by sands (Figures 3 and 5, Table S7 in Supporting Information S1) and only covering 17% of the fjord's seafloor area (Table 2), comprise 48%-77% of the total annual terrestrial OC burial flux in Bute Inlet. This terrestrial OC is predominantly young, associated with fine sands and buried beneath a layer of mud including old biospheric OC fluxes are given as the average value between sediment budgets estimated using two approaches covering two timescales (i.e., decade to century; see Section 3; Heerema, 2021;Heijnen et al., 2022;Syvitski et al., 1988). Error bars correspond to the range between the two approaches. Sediment and OC are shuffled stepwise down the channel before reaching the lobe due to migrating knickpoints. The channel is thus net erosive over decennial timescales, with patches of erosion (E) and deposition (D) between knickpoints. Over longer timescales (>100's year), the channel is interpreted as being neutral to slowly aggrading.
OC  Figure 7). The overbanks, dominated by mud ( Figure 5) and covering 47% of the fjord's seafloor area, comprise 15%-25% of the total annual terrestrial OC burial flux, with low TOC in the fine sands. The settling velocities of waterlogged woody debris have been shown to be relatively rapid due to their size and density (Hoover et al., 2010;A. D. McArthur et al., 2016;Waterson & Canuel, 2008), such that this woody debris is carried mainly along the channel floor. This concentration of woody debris along the channel floor could explain the low TOC values seen within the fine sands of the overbanks. The distal flat basin, exclusively made of mud in the surficial (200 cm) sediment and covering 36% of the fjord's seafloor area, comprises the remaining 6%-22% of the total annual terrestrial OC burial flux. The abundance of mud in the surficial sediment of the distal flat basin is explained by the limited number of turbidity currents reaching this site in present times (Chen et al., 2021;Pope et al., 2022;Prior et al., 1987).

What Is the Terrestrial OC Burial Efficiency of Bute Inlet and How Does It Compare With Other Fjords?
Comparison between OC river input (22.7 ± 13.1 Kt OC/year; Table 1) and terrestrial OC burial rate (11.7-16.3 Kt OC/year; Table 2) suggests that Bute Inlet has a terrestrial OC burial efficiency ranging between 52% and 72% (62% ± 10% on average) when estimated over decennial and centennial timescales, respectively.
No study has closed budgets between river sources and fjord sediments before, so it is not possible to compare this value with other river-fed fjords. However, we can compare our results in Bute Inlet with previous fjord studies, by normalizing OC burial rates to Bute Inlet's total surface area. By doing so, the burial of terrestrial OC in Bute Inlet ranges between 72 and 100 t C/km 2 /year. This range lies between previous estimates of area-normalized OC fluxes of 57-107 t C/km 2 /year in Scottish and Irish fjords that were described as having heterogeneous seafloors (Smeaton & Austin, 2019). Our results from Bute Inlet suggest that fjords characterized by sandy turbidity current systems may bury over 3 times more terrestrial OC compared to global estimates (22 t C/km 2 / year; Cui, Bianchi, Jaeger, et al., 2016;Cui, Bianchi, Savage, et al., 2016;Smith et al., 2015) based on muddy cores from fjords without any identified turbidity current channels (and usually lacking multibeam surveys). It was previously shown that turbidity currents are more active in fjords that are characterized by larger riverine inputs, steeper and deeper offshore slopes (Gales et al., 2019;Pope et al., 2019). Such factors need to be carefully considered when assessing OC burial potential in fjords.

Comparison With Larger, Deep-Sea Turbidity Current Systems
To our knowledge, only one study has provided OC balance budgets between river source and turbidity current subenvironments within a single system. This is the Congo River to the Congo Fan (Baudin et al., , 2020Coynel et al., 2005), where burial rates are calculated over centennial timescales that are comparable to the longer timescales considered in this study. Using a source to sink approach, Baudin et al. (2020) found that 33%-69% of the annual OC delivered by the Congo River is buried in the Congo turbidity current system. This burial efficiency is similar to, if slightly lower than, our estimate in Bute Inlet (52%-72%), highlighting the difficulty in closing budgets between river source and marine sinks fed by turbidity currents (Baudin et al., 2020). Within these two turbidity current systems, both the Congo and Bute Inlet submarine canyon-channels show minimal long-term (>100 years) OC accumulation, potentially indicating remineralization of the OC. However, remineralization was shown to occur in small proportions in the Congo system, based on in situ measurements of total oxygen uptake made on the lobe sediments (Rabouille et al., 2019). Finally, we note that undersampling of the Congo Canyon and Channel occurred in past studies due to the difficulty with sampling sandy deposits (Baudin et al., 2020). This undersampling of the canyon and channel may result in underestimation of OC burial rates in the Congo Channel, similar to the Bute Inlet channel where only 30-cm-long sediment cores could be retrieved.
In both Congo and Bute Inlet systems, the lobe holds most of the total buried OC.
Besides this example from the Congo system, work in Gaoping Canyon offshore Taiwan shows a terrestrial OC preservation efficiency of >70% in marine sediments. The Gaoping submarine canyon is supplied by river floods (hyperpycnal) and dilute surface plume (hypopycnal) inputs (Kao et al., 2014). This high OC burial efficiency results from Taiwan being a mountainous island and a hotspot for sediment and OC production, with frequent typhoon-generated floods and turbidity currents that deliver terrestrial OC (e.g., woody debris associated with sands) through the submarine canyon to the deep-sea ( Another example of high OC burial efficiency comes from the Bay of Bengal where almost no loss of OC was observed between the Ganges-Brahmaputra River inputs and the Bengal Submarine Fan (Galy et al., 2007). Coarse turbidity current deposits from the channel-levees of the Bengal Fan submarine system were also shown to have abundant terrestrial woody fragments throughout the last 19 Myr (Lee et al., 2019). These OC-rich turbidity current deposits are interpreted to result from turbidity currents triggered by high-magnitude, low-frequency events (e.g., floods, cyclones; Lee et al., 2019).
Based on these examples, it is plausible that rare and extreme events may also affect the OC burial efficiency and distribution in Bute Inlet, potentially increasing our burial efficiency estimate. It was recently shown that most turbidity currents (∼90%) dissipate within the shallowest-water (<200 m depth and <12 km along channel from the Homathko Delta) part of Bute Inlet, whereas less frequent (∼10%) events rework this material and progressively shuffle it downstream to the lobe (Heijnen et al., 2022;Pope et al., 2022). We further suspect that rare long runout events can flush material to the distal flat basin, as evidenced by thick sandy accumulations found at >3 m depth in 8-m-long piston cores collected in the distal basin at the location of core 15 (Figure 3, Heerema, 2021). Based on our OC burial efficiency estimates (i.e., 50%-70% over decennial to centennial timescales, respectively), it appears that OC burial efficiency is higher when considering longer timescales, which are more likely to integrate some higher magnitude, less frequent events than shorter timescales.
An example of a rare and large event occurred in the Bute Inlet catchment on 28 November 2020 when a glacial lake outburst flood took place at Elliot Creek, which is a tributary of Southgate River (Figure 1). This outburst flood-caused by an exceptionally large landslide-discharged about 4 million m 3 of terrestrial sediment (including soil organic matter) and a large volume of woody debris into the Southgate River and ultimately Bute Inlet (Geertsema et al., 2022). The impact of this event on the OC burial in Bute Inlet is yet to be determined and compared with the preevent OC fluxes presented in this study, but it may help to understand sediment and OC delivery in infrequent but large magnitude events.

Conclusion
This study provides the first complete source to sink budget for OC in a river-fed fjord turbidity current system, showing how OC is distributed within the upper <2 m of sediment. OC fluxes are estimated for the fjord's two river sources (Homathko and Southgate Rivers, which provide 94% of the water and sediment discharge to the fjord) and compared to OC burial fluxes in different seafloor subenvironments. We estimate that the annual OC export from both rivers to the fjord is 23 ± 13 Kt C/year, although this may not capture the full annual range of streamflow variability. The annual terrestrial OC burial rate of the entire fjord is estimated to range from 12 to 16 Kt OC/year. This suggests that a relatively high terrestrial OC burial efficiency (62% ± 10%) occurs within the fjord. Terrestrial OC is distributed as follows across the fjord seafloor; 63% ± 14% of the total terrestrial OC burial occurs in the sand-dominated subenvironments (covering 17% of the fjord seafloor area), whereas 37% ± 14% occurs in the mud-dominated subenvironments (covering 83% of the fjord seafloor area). Therefore, hydrodynamic fractionation of OC by turbidity currents leads to variable burial efficiencies in different subenvironments of the fjord. This study helps to understand how OC is buried within fjords over short (decennial to centennial) timescales. A comparison to other (nonfjord) turbidity current systems suggests that these systems consistently have high (50%-100%) terrestrial OC burial efficiency (Baudin et al., 2020;Galy et al., 2007;Liu et al., 2016). Turbidity current systems may thus play a globally important role in the highly efficient delivery and burial of mainly terrestrial OC, which affects atmospheric CO 2 levels over geological timescales (Berner, 1982) and food resources for modern benthic ecosystems (Włodarska-Kowalczuk et al., 2019).

Data Availability Statement
Discharge data for the Homathko River and Southgate River are available from https://wateroffice.ec.gc.ca; station 08GD004 and station 08GD010, respectively. Multibeam bathymetric data are held by the Geological Survey of Canada and Canadian Hydrographic Survey. Data and metadata associated with the sediment core and river samples can be found at https://doi.org/10.1594/PANGAEA.944382 and https://doi.org/10.1594/PANGAEA.944379.