Seasonal Particle and Carbon Dynamics in the Eastern Bering Sea

The ocean margins of high-latitude seas, such as the eastern Bering Sea, are recognized as important areas for the enhanced scavenging removal of particle reactive chemicals and for the potential to sequester atmospheric carbon dioxide via photosynthetic conversion to biogenic particles and subsequent downward particle transport to deeper waters. This region is expected to warm in the future, and with the warming will come a reduction in the extent and duration of seasonal sea-ice. This physical process exerts an important control on the timing, location, and magnitude of the spring primary production bloom. Within the context that 2008-2010 are characterized as cold years in the eastern Bering Sea, establishing the mechanistic link between primary production and the seasonal progression in particle export represents a significant challenge because these processes exert a control on organic matter transport from the surface ocean and on the success of economically and culturally important animals. During spring and summer cruises in 2009 and 2010, distributions of 234 Th (t1/2 = 24.1 days) in the water column and sediments were measured at ~60 stations over the middle and outer regions of the shelf and at the shelf break. The inventory of excess 234 Th ( 234 Th which is not produced in the sediments) in shelf sediments was ~1/3 of the total deficit of 234 Th ( 234 Th scavenged by sinking particles) in the overlying water column leading to an average focusing factor of 0.34 ± 0.23. Further, 234 Th export from the shelf was determined to ~30% of the total production of this radionuclide by 238 U decay based on a 234 Th budget. These results, taken together with elevated focusing factors in the offshelf region, suggest that the shelf sediments and lateral transport of particles from the shelf represent significant sinks for biogenic particles produced over the shelf in the spring and summer. The elevated focusing factors at the shelf break are attributed to enhanced particle flux from blooms of primary production in this region, an area commonly referred to as the ‘Green Belt’ for its exceptionally high rates of primary production. Thus, seasonal export of particles from elevated rates of primary production at the shelf break may transfer a significant amount of particulate organic carbon (POC) from the surface waters. POC export in this region demonstrates a clear seasonal progression with low fluxes in the early spring that increase by late spring and early summer. Rates of net primary production (NPP) were high and export fluxes relatively low near the ice-edge in spring, leading to export ratios (e-ratio = POC export/NPP) <0.25. In early summer, POC export exceeded NPP individual stations leading e-ratios >1, which is attributed to a temporal lag, or offset, between the high rates of primary production in spring and export as POC during the early summer. Using a water column-sediment 234 Th budget, the export of POC from the outer shelf to slope water was estimated to be 24±35 mmol C m -2 d -1 , which represents an off-shelf e-ratio of 0.07-0.52 for contemporaneous seasonally averaged and historical monthly averaged daily rates of NPP. In addition to the vertical POC fluxes measured at the shelf break, the imputed off-shelf export flux and e-ratios further suggests that there may be a significant transfer of shelf-derived particles to the slope waters. The high e-ratios and particles fluxes determined at the shelf may be a result of the biological response to the timing of physical processes in spring and summer of cold years. In spring, total chlorophyll a concentrations are generally low; however, localized phytoplankton blooms near the marginal ice zone (MIZ) lead to elevated spring average chlorophyll a concentrations, relative to summer, over the shelf and at the shelf break. Diatoms represent the greatest contribution to total chlorophyll a in spring and summer of cold years. This algal class also represents the majority of total chlorophyll a in particles sinking through the water column. Further, the relatively high proportion of pheophorbide a in sediment trap material indicates that sinking of zooplankton fecal pellets facilitates the export of particles through the water column. In cold years, the emergence of large diatom blooms in the spring MIZ supports the production of abundant large zooplankton. Large zooplankton are a primary food source for juvenile pelagic fishes of economically important species. Therefore, these cold year specific processes may be essential for the transfer of POC from the surface waters and the success of the economically important pelagic fishery. A consequence of a warmer Bering Sea in the coming decades is a reduction in seasonal sea-ice extent and duration. A change in seaice cover may alter the timing and magnitude of spring primary production and the flow of energy through the lower trophic levels.

The elevated focusing factors at the shelf break are attributed to enhanced particle flux from blooms of primary production in this region, an area commonly referred to as the 'Green Belt' for its exceptionally high rates of primary production. Thus, seasonal export of particles from elevated rates of primary production at the shelf break may transfer a significant amount of particulate organic carbon (POC) from the surface waters. POC export in this region demonstrates a clear seasonal progression with low fluxes in the early spring that increase by late spring and early summer. Rates of net primary production (NPP) were high and export fluxes relatively low near the ice-edge in spring, leading to export ratios (e-ratio = POC export/NPP) <0.25. In early summer, POC export exceeded NPP individual stations leading e-ratios >1, which is attributed to a temporal lag, or offset, between the high rates of primary production in spring and export as POC during the early summer. Using a water column-sediment 234 Th budget, the export of POC from the outer shelf to slope water was estimated to be 24±35 mmol C m -2 d -1 , which represents an off-shelf e-ratio of 0.07-0.52 for contemporaneous seasonally averaged and historical monthly averaged daily rates of NPP. In addition to the vertical POC fluxes measured at the shelf break, the imputed off-shelf export flux and e-ratios further suggests that there may be a significant transfer of shelf-derived particles to the slope waters.
The high e-ratios and particles fluxes determined at the shelf may be a result of the biological response to the timing of physical processes in spring and summer of cold years. In spring, total chlorophyll a concentrations are generally low; however, localized phytoplankton blooms near the marginal ice zone (MIZ) lead to elevated spring average chlorophyll a concentrations, relative to summer, over the shelf and at the shelf break.
Diatoms represent the greatest contribution to total chlorophyll a in spring and summer of cold years. This algal class also represents the majority of total chlorophyll a in particles sinking through the water column. Further, the relatively high proportion of pheophorbide a in sediment trap material indicates that sinking of zooplankton fecal pellets facilitates the export of particles through the water column. In cold years, the emergence of large diatom blooms in the spring MIZ supports the production of abundant large zooplankton. Large zooplankton are a primary food source for juvenile pelagic fishes of economically important species. Therefore, these cold year specific processes may be essential for the transfer of POC from the surface waters and the success of the economically important pelagic fishery. A consequence of a warmer Bering Sea in the coming decades is a reduction in seasonal sea-ice extent and duration. A change in seaice cover may alter the timing and magnitude of spring primary production and the flow of energy through the lower trophic levels.
vi ACKNOWLEDGMENTS I compiled a pretty large list of people and experiences worth acknowledging over the past half decade. I should have written them down or something because I'm sure I'll forget some.
I'll start off by thanking my advisor, Dr. Bradley Moran. He gave me an opportunity to succeed in this field. The Ph.D. process is rough and seemingly unending, but years ago Brad mentioned to me that all of the small processes, such as rotating samples on a nightly basis and keeping up with data reduction, will pay off. His pragmatic approach to science led me to an early start on the writing process. For this I'm thankful; I was a terrible writer early on. He's also provided me with, as my predecessor, Scott Stachelhaus, eloquently stated -'food and shelter' and 'gainful employment'. Brad allows for all of his students to have a life outside of the lab and office and to work independently. The Ph.D. process is stressful from time to time, and I'm grateful for the freedom to have a life outside of science. I need to thank Dr. Mike Lomas, my co-advisor, for his guidance over the years, especially in the areas of biology and working at sea. He's a model of tireless hard work, persistence, and productivity. Those are traits I hope to carry with me.
Thank you to my committee members, Dr. Ted Durbin and Dr. Dawn Cardace.
I'm appreciative for their help with comps and for their comments and suggestions that have improved the following manuscripts.
And now for Pat Kelly; without his help, this dissertation would not have been possible. I feel like I'm repeating acknowledgements in past theses from this lab, but his guidance and technical knowledge are the foundation for these manuscripts. His stern vii approach and sound execution while working at sea provided me with a great data set, and for that I cannot be more grateful.
I've had teachers along the way that have shown me the meaning of hard work and to take my passions and run with them. The list of those people is long and I know I will forget to mention some, but there a few that need to be mentioned. I need to thank my parents, girlfriend, and friends for their support over the years.
My sanity depended on it. There have been many who have lent an ear over the years, and I cannot thank them enough.
viii I'll admit I've learned an awful lot over the past 10 or so years of school. One of the most important principles I've gained in going through the Ph.D. process in particular is that there is still way (!) more that I don't know. It's quite humbling to come to grips with that reality actually, and hopefully the tools I've picked up here will allow me to balance the known/unknown out some. That's really all we can ask for, right? Keep moving in a forward direction. The Ph.D. thing is a long, trying, and ultimately less glorious process than I had imagined years ago. That being said, this has been an incredible experience, and I'm thankful for the set of circumstances that moved me along the path to this point. ix

PREFACE
The eastern Bering Sea supports nearly half of the US fishing industry in terms of fish catch revenue (National Marine Fisheries Service) and provides sustenance harvesting for the thousands of natives residing on the islands and Alaskan shore. This region is predicted to warm in the future. A warmer climate will alter the timing of annual physical processes, such as sea-ice advance in the fall and retreat in the spring.
These physical processes are the most important control on the structure of the ecosystem, in particular, at the lowest trophic levels (phytoplankton and zooplankton).
The magnitude and composition of the lowest trophic levels, in turn, have a direct impact on the success of economically important animals. The three years (2008)(2009)(2010) of the field study, from which the three following manuscripts are derived, are considered cold years in this region characterized by extensive sea-ice cover that persists late into spring.
This dissertation is an investigation of the seasonality of particle dynamics in this region, and results are discussed within the context that these observations are from cold years.
The following section provides a brief overview of the objectives of this thesis and the approaches employed to achieve those goals.
The radiochemical balance of 234 Th (t 1/2 = 24.1 days) over the eastern Bering Sea shelf is used to estimate (a) the fraction of particles produced in the water column that are retained in shelf sediments and (b) the net flux of particles, defined here as particulate organic carbon (POC), from the shelf to the slope waters over a seasonal time-scale.
Uranium-series radionuclides are proven tracers of particle transport process in aquatic environments. Previous studies; however, have utilized long-lived tracers, such as 210 Pb (t 1/2 = 22.3 years), 230 Th (t 1/2 = 75, 200 years), and 231 Pa (t 1/2 = 32, 500 years) to x investigate processes over decadal and longer time-scales ). The relatively short half-life, particle reactive nature, and known rate of production of 234 Th suit this radionuclide as a tracer of particle transport over seasonal time-scales, a method which has not be used before. Following the approach described by  and assuming a steadystate of several months for 234 Th, the radiochemical balance over the shelf is defined as: This equation represents the balance between the supply of 234 Th via in situ production from 238 U decay, λA U ( 234 Th decay constant; λ = 0.0288 d -1 ), and the net flux of 234 Th from the Oceanic to the Outer shelf waters, V*ΔTh CS (dpm cm -2 d -1 ), where V (m 2 s -1 ) is the horizontal cross-front exchange rate and ΔTh CS (dpm m -4 ) represents the cross-shelf 234 Th activity gradient between the Oceanic and Outer domains (ΔTh CS can be represented by dTh/dy, where y represents the cross-shelf direction). The removal terms include the in situ decay of 234 Th, λA Th , and the along-shelf transport of 234 Th, T* ΔTh AS (dpm cm -2 d -1 ), where T (m s -1 ) represents the along-shelf water volume transport and ΔTh AS (dpm m -3 ) is defined as 234 Th activity difference between the northern and southern regions. The term J sed represents the net flux of 234 Th to the sediments, and J exp is the net 234 Th export from the shelf to the ocean interior.
Sub-tidal flow fields and cross-front exchange rates are small over the mean life of 234 Th (35 days), therefore the terms V*ΔTh CS and T* ΔTh AS are negligible in the overall 234 Th balance. Thus, Eq. (i.1) is simplified: The production and decay terms are evaluated using spatially averaged, depth integrated activities of 234 Th and 238 U. The difference between the production and decay rates of 234 Th (λ(A U -A Th )) is equal to the total particle removal flux (J sed + J exp ). Once again, assuming a steady-state, the flux of 234 Th into sediments must be balanced by the decay in the sediment column. Therefore, J sed is quantified as the product of the average excess sediment inventory of 234 Th and the 234 Th decay constant : The remainder of Eq. (i.2) is the daily flux of 234 Th from the shelf to the slope waters (J exp ) over seasonal-time scales. Multiplying a POC/ 234 Th by J exp converts the offshelf 234 Th flux to a daily POC flux to off-shelf waters. In this region, the majority of the annual carbon fixation occurs during the spring bloom. Because of the seasonality in this system, constraining the fraction of POC production from the spring bloom event exported from the shelf and, more generally, from the surface ocean has implications in regional carbon budgets.
The shelf break of the eastern Bering Sea is well known for its rich levels of spring primary production . An accurate assessment of the seasonal progression of primary production and subsequent POC export in relation to seasonal sea-ice is necessary for the development of carbon budgets and understanding how carbon flows through the lowest trophic levels. Further, an improved understanding of the seasonality and magnitudes of primary and export production in cold years will provide a framework from which future observations from a warmer eastern Bering Sea may be interpreted.
xii There is no truly unbiased method to estimate POC export from the upper ocean.
A combination of the 234 Th approach and sediment traps are used to provide a range in estimates of POC export. The use of 234 Th to constrain the vertical flux of POC from the upper ocean is a more traditional application of this radionuclide. It has been increasingly utilized in high-latitude system over the past few decades Gustafsson and Andersson, 2012;; however, a multi-year study of the eastern Bering Sea has yet to be presented.
A one-box model is typically used for the calculation of the 234 Th flux through the upper ocean water column : where the change in 234 Th activity over time (dA Th /dt) is equal to production (λA U ) and decay (λA Th ) of the radionuclide in the water column, vertical flux of particulate 234 Th from the upper water column (P Th ), and net transport of 234 Th by advection and diffusion (V Th ). A lack of time-series measurements necessitates the assumption of onedimensional (V Th =0) and steady-state (SS; dA Th /dt=0) conditions, which simplifies Eq. At individual stations, the fraction of net primary production exported from the photic zone is traditionally represented as the export ratio (e-ratio): where POC export is either the sediment trap derived or the 234 Th-derived POC flux.
Each e-ratio calculated at the shelf break represents a single observation or 'snapshot' on the day the measurements were made. Compiling e-ratios over time and space can be used to infer the seasonality in the pulse of primary production in the photic zone and export of particles from the surface waters. More specifically, the seasonal succession of the e-ratio, coupled with proximity to the retreating sea-ice edge and data on the seasonal emergence of zooplankton stocks, provides the basis to interpret carbon cycling within the sunlit upper ocean, the magnitude of particle export from this layer, and perhaps the fraction of carbon available for transfer to higher trophic level, economically important animals.
The composition of the autotrophic community and seasonal emergence of zooplankton may control the seasonal fluxes of POC from the surface ocean in cold years as described above. Algal classes, such as diatoms or prymnesiophytes, produce specific accessory pigments. For example, the pigment fucoxanthin is found in the chloroplasts of brown algae. The primary brown algae in the eastern Bering Sea are diatoms and, to a xiv lesser extent, chrysophytes (silicoflagellates). A more complete description of the specific accessory pigments associated with various algal classes is presented in Mackey et al. (1996). The CHEMTAX program (also described in Mackey et al., 1996) uses the ratio of these indicator pigments to total chlorophyll a from water column samples to calculate the relative contribution of the algal classes to total chlorophyll a. The approach is used to interpret the seasonal evolution of the autotrophic community in spring and summer over shelf and shelf break. The same program is used to evaluate the composition of phytoplankton sinking through the water column. In addition, the pheopigment signature in settling material is used a proxy for the influence that sinking zooplankton fecal pellets exert on the flux of particulate organic matter (POM) from the photic zone.
In the following dissertation, the above approaches are utilized to improve the understanding of carbon cycling in the Bering Sea within the context that these studies are conducted over a multi-year cold period in this region. In Manuscript I, titled " 234 Th balance and implications for seasonal particle retention in the eastern Bering Sea," the radiochemical balance (Eq. i.1) is used to provide estimates of the off-shelf 234 Th flux over seasonal time-scales. Further, retention of water column produced particles in the underlying sediments is evaluated on a station by station basis using the ratio of the excess 234 Th in the sediment to the 234 Th deficit in the water column. These ratios, known as focusing factors (FF Th s), are used as a means to determine whether local areas serve as regions of net sediment gain or loss by lateral transport. This manuscript was published in Deep Sea Research Part II: Topical Studies in Oceanography (October, 2013) in the second special issue on results from BEST-BSIERP (Bering Sea Project). In xix    Table 3.7. Regional averages of percent contribution by algal group to total chlorophyll a.
Primary production values are from . ……………...108  Table 2 are the BEST-BSIERP regions grouped in this study. White (spring) and black (summer) symbols represent water column sampling locations during the field study (see Table 1 for cruise information). Cross symbols are sediment trap deployment locations. ………………………………………………………….162   Continental shelves represent only 10% of the world ocean, yet these regions account for ~20% of global primary production . Ocean margins are recognized to be effective areas for the enhanced removal of reactive chemicals via particle scavenging and for the transfer of organic matter from continental shelves to the deep ocean. These processes are thought to play an important role in shelf-basin exchange of organic carbon and the sediment accumulation of particle-reactive pollutants on a global basis. A significant challenge is quantifying the rates and mechanisms of particle transport in shelf/slope systems.
Uranium-series radionuclides are proven tracers of particle transport processes in aquatic environments.
In particular, previous studies have utilized long-lived radionuclides, such as 210 Pb (t 1/2 = 22.3 y), 230 Th (t 1/2 = 75, 200 y), and 231 Pa (t 1/2 = 32, 500 y) to investigate particle transport in shelf regions on decadal and longer time-scales (e.g., . Over the past several decades, the short-lived, particle-reactive radionuclide 234 Th (t 1/2 = 24.1 d) has been increasingly used as a tracer of POC export from the upper ocean (e.g., . In addition to its utility in quantifying POC export from the upper water column, the disequilibrium between 234 Th and its soluble parent 238 U in seawater and sediments has significant potential in quantifying seasonal particle transport and retention in shelf systems; for example, in a manner similar to that described for 210 Pb (e.g., .

As part of the Bering Ecosystem Study-Bering Sea Integrated Ecosystem
Research Program (BEST-BSIERP), measurements of the water column deficit of 234 Th and sediment excess inventory ( 234 Th xs ) are presented from 2009 and 2010 over the eastern Bering Sea shelf and slope ( Fig. 1.1). These data have been used to investigate several aspects of particle transport, including: seasonal particle retention over the shelf, export of POC, and enhanced particle export and deposition associated with the marginal ice zone (MIZ). Results indicate that ~30% of 234 Th produced over the shelf is exported to the ocean interior, implying that 234 Th and, by inference, associated particles in this system are largely retained over the shelf on a seasonal basis. In addition, relatively large inventories of 234 Th xs observed in the slope and deep ocean sediments during summer are suggested to result from enhanced scavenging removal and deposition of 234 Th associated with the MIZ during the spring sea-ice retreat, and possibly augmented by boundary scavenging removal of 234 Th at the ocean margin.

a. Study area
The broad (~500 km) and extensive (>500,000 km 2 ) seasonally ice-free eastern Bering Sea shelf is bordered on the south by the Alaska Peninsula and to the east by the Alaska mainland ( Fig. 1.1). The shelf break, located approximately at the 170 m isobath, extends northwestward from Unimak Pass and encompasses the Pribilof Islands, St.
Matthew Island, Nunivak Island, and St. Lawrence Island. During the ice-free months, the shelf waters may be subdivided into three cross-shelf domains, separated by three fronts . The Inner Front, located near the 50 m isobath, separates the shallow, well-mixed Coastal Domain (0-50 m) from the twolayered Middle Domain (50-100 m) . The Coastal Domain is well-mixed because the wind and tidally mixed layers overlap. The Middle Domain, characterized by the strongest stratification and the presence of a summer cold pool ( Fig. 1.2; summer temperature section), is isolated from the Outer Domain by the Middle Front, which overlies the 100 m isobath. Recent observations  have indicated north-south variability within the Middle Domain, though these trends are less pronounced than cross-shelf variability. The Outer Domain (100 m-shelf break) is characterized by surface and bottom mixed layers, though is separated by a structured middle layer . The Shelf Break Front separates the Outer Domain from the Oceanic Domain. These fronts affect lateral advection and diffusion, property exchange rates, and mixing between the water masses.

Methods
Water column and sediment core samples were collected in the eastern Bering Sea  Th was radiochemically purified and 230 Th was measured by alpha particle emission . Scavenging efficiencies for the small volume 234 Th method were determined to be 91±4.5% (1 σ).  . Cores were sectioned into 0.5 cm increments in the upper two cm, and into one cm increments for depths of 2 to 5 cm. Sediment samples were dried at 60°C in 125 mL jars, ground, and homogenized prior to analysis. The sediment density (ρ = ρ D (1 -φ)) of each sample analyzed was calculated using an assumed solid particle density (ρ D ) of 2.65 g cm -3  and sediment porosities (φ) determined in the laboratory from measurements of wet and dry sediment weight.
Samples were analyzed for 234 Th using a sea-going Canberra pure Ge planar type detector (GCW3023, 2000 mm 2 ) or on a shore-based Canberra pure Ge well type detector (GL20203, 150 cm 3 ) calibrated for the specific sample geometry. Sample activities were determined by gamma emission at 63.3 keV and decay corrected to the mid-point of collection. Supported levels of 234 Th ( 234 Th produced in the sediment column) were measured after 144 days and subtracted from the total 234 Th activity. Self-absorption corrections were applied according to the method described by .
Detector efficiencies were determined to be 10.8±0.3% and 52.1±1.0% for 234 Th at 63.3 keV for the planar and well type detectors, respectively.

a. Water column 234 Th-238 U disequilibrium
The 234 Th-238 U activity ratio (AR) provides a quantitative measure of the removal of 234 Th relative to 238 U from the water column via particle scavenging and export.
During spring 2009, activity ratios ranged from ~0.4 to 0.8 on the shelf, whereas secular equilibrium (AR = 1) was observed in the slope/oceanic water column ( Fig. 1

a. 234 Th balance over the eastern Bering Sea shelf
The radiochemical balance of 234 Th over the eastern Bering Sea shelf can be used to provide insight into particle transport, including the seasonal retention of particles in shelf sediments. Specifically, particle retention can be evaluated by establishing the balance between the export flux of 234 Th from the water column and the flux of 234 Th into the sediments. This approach follows that described by , who used where T (m s -1 ) represents the along-shelf water volume transport and ΔTh AS (dpm m -3 ) is defined as 234 Th activity difference between the northern and southern regions. The term J sed represents the net flux of 234 Th to the sediments, and J exp is the net 234 Th export from the shelf to the ocean interior. Note that this equation does not explicitly include diffusive transport of 234 Th over the shelf. It is assumed that the contribution of diffusion is negligible in the radiochemical balance for 234 Th, which is justified below. Also, this balance neglects 234 Th input from rivers because 234 Th activities in freshwater are negligible and riverine supply to the Middle and Outer shelf regions is insignificant.

a.1. Diffusive and advective fluxes of 234 Th
The net exchange of 234 Th from the Oceanic Domain to the Outer shelf can be quantified as the product of the exchange rate (V) across the shelf break and the 234 Th activity gradient between the Oceanic and Outer domains (ΔTh CS ). A similar approach was used to examine the diffusive flux of 210 Pb across the MAB frontal zone to the shelf . In their study, strong cross-shelf gradients of 210 Pb and a significant correlation between 210 Pb activity and salinity were used to estimate the flux of 210 Pb from the deep ocean to the shelf. For the eastern Bering Sea, however, there is not a significant correlation between 234 Th and salinity for any season (r 2 = 0.01 to 0.36). The calculated cross-shelf gradient in 234 Th (ΔTh CS ) between the Oceanic and Outer domains ranged from -0.0005±0.0013 to 0.0057±0.0017 (mean: 0.0018 ± 0.0028) dpm m -3 per m of the approximately 100 km width of the Outer Domain. The cross-front exchange rate (V) for the shelf break was determined to be 0.46 m 2 s -1 based on a reported net on-shelf water mass transport rate of 14, 500 km 3 y -1  and a shelf length of approximately 1000 km. This value is similar to cross-front exchange rates recently reported for the Middle Front at the 100 m isobath . Using these values, the cross-shelf exchange of 234 Th from the Oceanic Domain to the Outer shelf was estimated to range from -0.0018±0.0051 to 0.0226±0.0069 (mean : 0.0069±0.0100) dpm cm -2 d -1 spread over the approximate width of the Outer shelf (Table 1.3). These low average lateral fluxes imply that the cross-shelf exchange of 234 Th from the deep ocean to the Outer shelf is a minor component in the 234 Th balance for the eastern Bering Sea shelf ( Fig. 1.8).
Because of the separation imposed by the Middle Front and the difference in water mass circulation, the along-shelf transport of 234 Th may vary for the Outer and Middle domains. Sub-tidal water transport over the Middle and Outer shelf is small relative to boundary currents, ) and along-shelf 234 Th activity differences between the northern and southern regions of the eastern Bering Sea shelf are negligible. Therefore, T is small and variable and ΔTh AS is zero, implying that the alongshelf transport of 234 Th over the shelf is negligible in the 234 Th budget.
A further argument for the negligible transport of 234 Th by advection and diffusion lies in the length scale over which these processes are significant for the relevant spatial and temporal scales. Physical transport of water over the Middle and Outer shelf of the eastern Bering Sea is predominantly due to cross-shelf tidal currents . The eastern Bering Sea experiences mixed semi-diurnal tides, with M 2 tidal current velocities of 15-30 cm s -1 and the K 1 constituent contributing 10 -20 cm s -1 . The length scale (L a ) over which advective transport of 234 Th is significant can be determined using (e.g., : where U h is the tidal velocity (15 cm s -1 ) and t is the mean life of 234 Th (35 d). Using these values, L a is approximately 450 km. However, because daily current velocities are a function of both ebb and flood tides, the net daily cross-shelf water flux over the Middle and Outer shelf is small. Thus, the net advective transport of 234 Th by tides is a minor component in the radiochemical balance of 234 Th.
To examine horizontal diffusion as a mechanism for the transport of 234 Th, the mean diffusive path length in the absence of 234 Th scavenging is calculated as (e.g., : (1.6) Coachman (1982)  (1.7) The production and decay terms are evaluated using spatially averaged, depth integrated activities of 234 Th and 238 U. On a seasonal basis, the areal production of 234 Th by 238 U decay in the water column ranges from 0.55±0.21 to 0.64±0.30 (mean: 0.59±0.04) dpm cm -2 d -1 . The areal decay of 234 Th in the water column is smaller than production, ranging from 0.32±0.18 to 0.36±0.23 (mean: 0.34±0.02) dpm cm -2 d -1 (Table 1.3). The difference between the production and decay rates of 234 Th (λ(A U -A Th )) is equal to the total particle removal flux (J sed + J exp ).
At steady-state, the flux of 234 Th into the shelf sediments must be balanced by decay in the sediment column. Thus, J sed is quantified as the product of the average excess sediment inventory of 234 Th and the 234 Th decay constant : (1.8) The seasonal decay of 234 Th in shelf sediments of the eastern Bering Sea ranges from 0.07±0.04 to 0.10±0.07 (mean: 0.09±0.02) dpm cm -2 d -1 (Table 1.3). By comparison, decay of excess 234 Th in the sediments represents 15±3% of 234 Th production in the water column (Table 1.3). The implication is that shelf sediments are an important sink in the scavenging removal of 234 Th from the overlying water column.

a.3. Off-shelf export of 234 Th
The radiochemical balance of 234 Th in the eastern Bering Sea is summarized in Table 1.3. The seasonal export flux of 234 Th from the shelf to the ocean interior (J exp ) can be calculated from the difference between the supply and removal fluxes of 234 Th (Eq. 1.7). For spring and summer, J exp represents on average 29±2% of the total production of 234 Th ( Fig. 1.8), implying that on a seasonal basis 234 Th is largely retained (i.e., ~70% of 234 Th production) over the eastern Bering Sea shelf. This result is consistent with a previous study conducted in the MAB using 210 Pb, which demonstrated that ~20% of the total 210 Pb supplied to that shelf is removed by particle export into the interior ocean on a time-scale of decades . From the present data set, however, it is not possible to define a mechanism responsible for the transport and removal of 234 Th and associated particles off the shelf over seasonal time-scales. It is interesting to note that  propose that particles are removed from the shelf by a depositionbioturbation-resuspension-redeposition loop over decadal time-scales, and it is possible that such a mechanism exists for the eastern Bering Sea.
In addition, J exp can be used to place an upper bound on the seasonal export flux of POC. Using the average J exp value (Table 1. in press). The estimated POC export flux should be regarded as an upper estimate, due to the uncertainty in converting J exp to a POC export flux using an imputed POC/ 234 Th ratio, which can vary considerably (e.g., , and to possible preferential remineralization of POC.

b. Residence time of 234 Th in the water column
The residence time of total 234 Th (τ t ) can be used to further assess the time-scale of particle retention over the eastern Bering Sea shelf. Specifically, the residence time of total 234 Th provides a quantitative measure of the efficiency of scavenging and transport of 234 Th within the shelf and upper water column of the Oceanic region. A onedimensional, irreversible scavenging model is used for the estimation of total water column 234 Th residence time, which as justified above, sets advective and diffusive transport of 234 Th to zero Wei and Murray, 1992): where λ c is the first-order rate scavenging constant for 234 Th. Eq. (1.9) describes the balance between 234 Th production, decay, and particle scavenging. Assuming steadystate, Eq. (1.9) simplifies to: (1.10) where z is the depth of integration. The residence time of total 234 Th can be estimated for the entire water column by the relationship: (  (Wei and Murray, 1992) and the inner region of the Gulf of Maine (34 to 143 d) . In the Oceanic region, residence times increase to 5 to 6 months, which are similar to those reported for the outer reaches of the Gulf of Maine  and Funka Bay, Japan (Wei and Murray, 1992). By comparison, estimates of water transport onto the shelf of the eastern Bering Sea and flow through the Bering Strait were used to determine water mass residence times of ~3-7 y for the shelf region . Shelf water residence times are much longer than the average τ t , which is consistent with the largely seasonal retention of 234 Th and associated particles over the eastern Bering Sea shelf.

c. 234 Th focusing factors
To further establish the extent to which there is a net seasonal retention of particles over the shelf, the exchange of 234 Th between the water column and sediments can be evaluated on a station-by-station basis. The relationship between the 234 Th deficit in the water column and 234 Th xs inventory (Table 1.2) on the eastern Bering Sea shelf is illustrated in Fig. 1.10. This comparison indicates that for ~65% of stations sampled over the shelf, 234 Th xs inventories are within a factor of ~1.5-4 of the measured water column deficits of 234 Th. This observation is consistent with the conclusion that the flux of 234 Th into the sediment represents an important sink for 234 Th produced over the shelf.
The 234 Th focusing factor (FF Th ) can be used to further quantify the exchange of 234 Th between the water column and sediment. In particular, the FF Th is an empirical relationship that defines the efficiency of 234 Th transport from the water column to the underlying sediment . The sediment 234 Th xs inventory is related to the water column deficit by the relationship : (1.12) A FF Th = 1 implies that the sediment inventory is in balance with water column removal of 234 Th over seasonal time-scales. A FF Th greater, or less, than 1 indicates the redistribution of 234 Th into, or away from, a specific sampling location, respectively.  of sedimentation must be efficient at retaining particulate 234 Th, thereby working against lateral processes that would otherwise largely export 234 Th (J exp ) off the shelf into the interior ocean. As noted above, this result is consistent with a previous study using 210 Pb in the MAB, which indicated a net retention of particles on the shelf for up to several decades .

d. Implications for particle transport and retention
There are several possible qualitative interpretations that may account for the Zhemchung canyons, which may accumulate sediments from large areas of the shelf and slope.

Conclusions
The results reported in this study provide important new insights regarding our understanding of particle transport processes, water column-sediment interaction, and the magnitude of off-shelf export of particles and associated reactive chemicals in a highly dynamic Arctic shelf environment. Prior studies in the northern Bering Sea have reported dramatic shifts in benthic productivity and water column-benthic coupling in the context of a changing climate . The present study utilizes a geochemical tracer method to evaluate the fate of particles over the southeastern Bering Sea shelf, which has broader applicability in quantifying particle transport processes in other complex shelf systems.
Specifically, based on comprehensive measurements of 234 Th in the water column and sediments, it has been determined that on a seasonal basis roughly 2/3 of the supply of 234 Th is balanced by decay and sediment burial over the eastern Bering Sea shelf.
Furthermore, the off-shelf export flux of 234 Th (J exp ) represents ~30% of the total 234 Th supply, implying that 234 Th and associated particles are largely retained on the shelf, rather than exported to the ocean interior. While it not possible to define the mechanism(s) responsible for off-shelf export of 234 Th, it is suggested that this may involve a particle deposition-bioturbation-resuspension-redeposition loop, as described for the off-shelf transport of particles in the MAB . In addition, the results of this study have been used to provide an upper estimate of seasonal POC export from the shelf water column of 18 ± 41 mmol C m -2 d -1 .           Table   1.3.

Publication Status
Manuscript II, titled "Seasonal decoupling of particulate organic carbon export and net primary production in relation to sea-ice at the shelf break of the eastern Bering Sea: implications for off-shelf carbon export" was published in While these observations reveal a seasonal progression in POC export and the e-ratio, there is no direct relationship to sea-ice proximity. Furthermore, based on a water column-sediment 234 Th budget, the off-shelf export of POC during spring-summer is estimated to be 24±35 mmol C m -2 d -1 , which represents an off-shelf e-ratio of 0.07 and 0.21 for contemporaneous seasonally averaged daily rates of NPP and 0.17 and 0.52 for historical monthly averaged daily rates of NPP. An implication is that off-shelf POC transport may represent a seasonal net sink for CO 2 in this and other polar shelf regions.

Introduction
The eastern Bering Sea shelf, as with other polar seas, influences the uptake of atmospheric CO 2 in part by the conversion of dissolved inorganic carbon into biogenic particles and subsequent vertical removal by particle export. The seasonal increase in solar irradiance, coupled with the retreat of sea-ice in spring, allows for exceptionally high rates of primary production in the nutrient rich waters of the shelf break and Outer shelf region of the eastern Bering Sea . The resulting pulse in primary production within the marginal ice zone (MIZ) during spring sea-ice retreat has been reported to decouple from zooplankton growth due to reduced grazing efficiency resulting in the potential for seasonally high rates of particle export in Arctic shelf systems (e.g., Wassmann et al., 2004). Because of the potential for seasonally high particle export fluxes at the shelf break and Outer shelf of the eastern Bering Sea, previous studies have suggested that this region may represent a seasonal net sink for atmospheric CO 2 .
The recent decline in Arctic and sub-Arctic seasonal sea-ice thickness and extent may restructure energy flow through the lowest trophic levels of the ecosystem Grebmeier et al., 2006;, potentially impacting the export of organic carbon from the surface to deeper polar ocean waters. In particular, the seasonal timing of sea-ice retreat exerts an important control on the timing of primary production , but it is still unclear as to how sea-ice retreat influences the overall magnitude of primary production in high-latitude systems. During cold years, the ice-edge can extend as far as the shelf break of the eastern Bering Sea, overlying nutrient rich water upwelled from the deep North Pacific . As sea-ice retreats due to seasonal warming, the upper water column is stratified by the release of freshwater. As a result, cold years typically yield large spring blooms near the MIZ as sea-ice retreats over the shelf break. Zooplankton stocks tend not to respond as quickly to such blooms during the spring, which may allow for a more efficient particle export of primary production at the shelf break following the bloom (e.g., Sakshaug, 2004). Here, we hypothesize that such a temporal offset between the spring primary production event and subsequent export during early summer may allow for the transfer of a large fraction of POC from the surface ocean to depth, and in turn, lead to a seasonal net sink for CO 2 during cold years. Conversely, during warm years characterized by early sea-ice retreat, spring blooms tend to be delayed until the upper water column is thermally stratified and occur primarily under ice-free, open water conditions. These warmer year blooms are more tightly coupled in time with the seasonal emergence of zooplankton stocks, resulting in a greater fraction of carbon retained and recycled within the pelagic ecosystem, reduced export, and hence a reduced transfer efficiency of carbon to benthic organisms. These hypotheses form the basis of the Oscillating Control Hypothesis (OCH), which postulates that walleye pollock (Theragra chalcogramma) recruitment is dictated by bottom-up control during cold years and top-down control during warm years .
While the OCH has been recently modified    , the essential premise remains that the transfer efficiency of export production associated with the spring bloom event may be greater during cold years.
However, no contemporaneous measurements of both net primary and export production exist over multiple years for the eastern Bering Sea, either during cold or warm regimes.
Such information is necessary to testing not only the OCH, but also to address the notion that off-shelf export of POC may represent a seasonal net sink for CO 2 in such dynamic high-latitude shelf systems as the eastern Bering Sea.
A primary goal of this study is to assess seasonal changes in rates of POC export and the export ratio (e-ratio = POC export/NPP) at the shelf break and Outer shelf of the eastern Bering Sea in relation to sea-ice proximity. Because there is no ideal, truly unbiased method to quantify export production from the surface ocean, this study reports POC fluxes determined using both drifting sediment trap arrays and 234 Th-238 U disequilibria. Due to its short half life (t 1/2 = 24.1 d), particle-reactive nature, and known rate of production, 234 Th has been widely used as a tracer to quantify POC export from the upper ocean (Benitez-Nelson et al., 2001a;Bruland, 1985, 1987;. The use of 234 Th as a flux proxy for POC export has been increasingly utilized in the Arctic Gustafsson and Andersson, 2012;, however there is only one previous study of POC export over the eastern Bering Sea shelf, which reports data from a single year . Within the context that 2008 -2010 represents a multi-year cold period, sediment trap and 234 Th-derived POC export, together with seasonal variability in the e-ratio, are compared for the eastern Bering Sea during spring and summer. In addition, an estimate of the off-shelf transport of POC is reported, based on a recent water column-sediment 234 Th budget for the Outer shelf , and the resultant implications for a net seasonal sink for atmospheric CO 2 in this region are discussed.

a. Study area
The eastern Bering Sea shelf may be subdivided into three cross-shelf domains isolated by seasonal fronts: Coastal Domain (0-50 m), Middle Domain (50-100 m), and Outer Domain (100 mshelf break front at ~200 m) . These three shelf domains can be further sectioned into North and South regions separated at 60°N . This present study is focused in the North and South Outer domains and along the shelf break/slope (off-shelf region; ~200-3500 m). For consistency, the study regions will henceforth be referred to as the Outer Domain/shelf or shelf break/off-shelf ( Fig. 2.1).
Seasonal sea-ice extent and duration over the southeastern Bering Sea shelf is highly variable, with distinct temporal changes occurring on the order of less than a decade to several decades . Specifically, the last three decades of the previous century were characterized by a high degree of variability in spring sea-ice cover, which was followed by a five year period (  Water column 238 U activities were calculated from salinity according to the relationship 238 U (dpm L -1 ) = salinity (‰) x 0.0708 . Salinities were obtained from CTD profiles of water column hydrography. This relationship has been previously verified by sector-ICP-MS analysis of unfiltered seawater samples from the Outer shelf and slope water of the Chukchi Sea, a similar high-latitude marginal sea to the Bering Sea  as well as for the high salinity Mediterranean Sea, where possible nonconservative behavior of 238 U may be expected .

e. Analysis of POC
An arc-punch was used to generate a 10 mm diameter subsample from each sediment trap GF/F, which was used for the analysis of POC. It was necessary to subsample the trap GF/F by area rather than weight because of the need to analyze for 234 Th at sea. The POC subsample was frozen until preparation for analysis in a shore- Spring upper water column temperature, salinity, and buoyancy may provide additional, yet qualitative, insight into recent sea-ice cover and melt water release.  (Fig. 2.2c). May and mid-July. During HLY0803, the σ t range was 25.31 to 26.06 (mean: 25.57) kg m -3 , implying a slightly more buoyant <40 m water column than early summer ( Fig.   2.2b).

b. Sediment trap POC and 234 Th fluxes
Along the shelf break during HLY0802, spatial and depth (40 -100 m traps) averaged particle fluxes of 234 Th ( 234 Th trap ) and POC (POC trap ) were relatively low at 759±235 (± 1σ) dpm m -2 d -1 and 4.6±1.6 (± 1σ) mmol C m -2 d -1 , respectively (Fig. 2.3;  (Buesseler, 1991;: (2.5) In this study, the POC/ 234 Th ratio ranged widely, from 1 to 282 (25.5±37.6; avg. ± 1σ) µmol dpm -1 throughout the upper 100 m (Table 2.3); such variability tends to be characteristic of productive shelf waters . Higher POC/ 234 Th ratios were observed during the summer and in areas with greater absolute POC flux from the photic zone (Table 3). Furthermore, during summer and in higher flux environments, POC/ 234 Th ratios decreased with depth (Figs. 2.3, 2.4, 2.5), which may reflect a combination of preferential remineralization of POC, decay of particle-bound 234 Th, or particle aggregation and disaggregation  one-dimensional scavenging models. NSS fluxes take into account changing water column inventories of 234 Th over time, which is likely the case during events of elevated particle flux associated with plankton blooms : where Δt is the time interval between sampling, A Th1 and A Th2 are the activities of 234 Th for the first and second occupations, respectively, A U is the 238 U activity of the first occupation, and is the 234 Th decay constant. The NSS model also assumes that the same water mass is sampled repeatedly, which is reasonable in this study due to the weak subtidal flow fields over the southeastern Bering Sea shelf ).
Here, NSS and SS 234Th fluxes through the 75 m depth horizon are compared.
As described above, the SS 234 Th fluxes calculated during the latter stages of the bloom underestimate 234 Th fluxes determined from sediment traps by up to 85% (

a. Seasonal succession of NPP, POC export, and the export ratio
An overarching goal of this study is to quantify the fraction of POC production that is exported from the photic zone on a seasonal basis and in relation to sea-ice proximity. This objective is central to improving our understanding of carbon cycling and the carbon budget for this shelf region. The exported fraction of primary production from the photic zone is traditionally represented as the e-ratio, or ThE, when calculated using 234 Th-derived POC export ): where NPP is photic zone integrated rate of net primary production and POC export fluxes are determined from the nearest sediment trap below the photic zone depth or calculated at the base of the photic zone using the 234 Th approach. Throughout the course of this study, rates of NPP were discretely estimated by 14 C on-deck incubations. NPP results for 2008 and 2009 are presented elsewhere . During the cold years of this field program, NPP varied greatly between the Northern and Southern regions and between spring and summer. The highest rates of primary production were typically observed in spring and in areas within the MIZ due to ice-edge blooms. Briefly, ice-edge NPP was dominated by >5 µm cells associated with diatom production . During the summer, the northern region was still diatom dominant, however the open water stations in the south were characterized by lower rates of primary production and the emergence of haptophytes (namely Phaeocystis spp.) and microflagellates as major contributors to autotrophic biomass .
It is possible that the measured rates of NPP are underestimated in Lomas et al. The temporal and spatial variability in discrete NPP and POC export observations made during this study prevent the calculation of a statistically meaningful average export ratio characteristic of the entire eastern Bering Sea region. However, on a stationby-station basis both the e-ratio and ThE are typically lower in spring, due mainly to elevated levels of NPP (Fig. 2.9; Table 2.5). Such low export efficiency early in the growing season in high-latitude seas has been previously observed in the Arctic . In this study, spring e-ratio and ThE estimates are below ~0.25 for all but two stations (NP15 in 2008 and NZ11.5 in 2010), which had relatively low rates of NPP (Table 2.5).
As noted previously, nine spring sediment trap deployments were made within ~140 km of the ice-edge (Table 2.2). With the exception of station BL, spring stations nearer the ice pack were characterized by lower fluxes of POC from 40-100 m ( Fig.   2.10a). Therefore, the relatively low export ratios in the spring are the result of a combination of higher spring NPP and lower export production (Figs. 2.9, 2.10b). As with POC export fluxes, export ratios nearer the ice-edge are low relative to the open water and summer estimates ( Fig. 2.10b). Furthermore, the lowest calculated export ratios are evident for station BL (Fig. 2.10b; Table 2.5) though POC export is relatively high at this location. This is because the export ratio is more dependent on the rates of NPP, which exhibit greater variability relative to export production. These low export ratios associated with high rates of NPP are consistent with an inverse relationship between NPP and export efficiency, which has been observed in the Southern Ocean . There is a general increase in POC fluxes for most open water stations relative to traps near the ice-edge, though there is considerable variability between those estimates (Fig. 10a). In fact, during the summer, POC export often exceeded coincident rates of NPP along the shelf break (Figs. 2.9, 2.10b; Table 2.5).
Export ratios >1 are attributed to a temporal decoupling, or offset, between primary production in spring and subsequent export as POC early in summer.
The results of this study reveal a clear seasonal succession in both sub-photic POC export and the export ratio, though with no apparent connection between these estimates and proximity to sea-ice. Specifically, considering station BL as an anomaly, both POC export and export ratios peak in early summer (KN195-10 and TN250) based on cruise average export fluxes and individual export ratios (Fig. 2.11). Changes in the composition of exported material may provide insight into seasonal variations in export production and the e-ratio. Photopigment analysis of exported material reveals that diatoms are the major algal class exported during the early summer (Baumann et al.; Lomas et al., in preparation). In addition, during the late spring (TN249) and early summer, high concentrations of degradation pigments (pheophytin a + pheophorbide a) in trap collected particles indicates the prevalence of sinking senescent cells, or cells that have been repackaged as fecal pellets produced by zooplankton grazing of spring primary production (Baumann et al.; Lomas et al., in preparation). Therefore, export during this period in the seasonal cycle may be driven by the sinking of the enhanced spring PP typically characterized by larger cells  and, perhaps, by increased zooplankton grazing of the spring primary producers. Both of these mechanisms may also contribute to the apparent temporal offset between primary and export production, leading to the elevated export ratios observed in the summer.
The observed seasonal shift in associated particle export and e-ratio is consistent with the revised OCH, which suggests bottom up control on age-0 walleye pollock recruitment during cold years . Recent evidence indicates large spring phytoplankton blooms associated with the MIZ of sea-ice retreat during cold years favor the production of large crustacean zooplankton, such as Calanus marshallae and euphausiids, a common prey of juvenile and adult pollock, while the abundance of smaller copepods is independent of the type of spring bloom . Furthermore, in cold years average energy densities in first year pollock are 33% greater in fall than during warms years, most likely attributed to abundant large zooplankton as a primary diet source . In the absence of large, lipid-rich zooplankton and euphausiids during warm years, juvenile and adult pollock change their diet to age-0 pollock. Therefore, large zooplankton and euphausiids exert a control on age-0 pollock success in late summer and fall by providing an alternate food source for older pollock year classes. In addition to regulating the flow of carbon to economically important pelagic organisms, large zooplankton may facilitate the elevated flux of POC from the photic zone by generating large, rapidly sinking fecal pellets in late spring and early summer during cold years.
The difference in timing between enhanced spring primary production and the emergence of large zooplankton may be an important control on the temporal offset between primary and export production in this region. Moreover, it is possible that large copepods and euphausiids may be important in the transfer of POC to the benthos and/or off the shelf. It remains uncertain how future changes in seasonal sea-ice extent and duration will affect overall NPP, secondary production, POC export, and the export ratio in the eastern Bering Sea, which may impact the structure of the ecosystem. For example, based on historical literature,  suggests that a warmer Bering Sea may favor a regime characterized by reduced annual primary production. Conversely  and  estimate that in years with early seaice retreat there is 40-50% more annual primary production relative to years with late seaice retreat, though these two studies do not comment on autotrophic community composition or POC export.
The temporal offset of primary and export production observed in this study has been reported in other high-latitude oceanic systems. As with the eastern Bering Sea, such regions are characterized by extensive spring MIZ blooms, which are responsible for a large fraction of annual new primary production.  , 1997) and over the Ross Sea shelf .
The spring blooms in these regions are typically comprised of diatoms, although the central Ross Sea is marked by a prevalence of haptophytes . These studies report very little export production during the onset of the spring bloom. As the bloom progressed, export fluxes increased relative to primary production and were attributed to a temporal lag between the elevated rates of production and subsequent export as POC. The enhanced export during the latter stages of the bloom may be attributed to the aggregation and sinking of cells, and a delay in the development of the grazing community. The temporal succession noted in these studies is consistent with the above proposed mechanism that results in export ratios >1 in the eastern Bering Sea (Figs. 2.9, 2.11). Similarly, recent studies conducted in the Chukchi Sea

b. Significance of carbon export from the Outer shelf to the deep ocean
Because of the apparent temporal offset between net primary and export production and the progressive increase in export efficiency of this system over the seasonal cycle, an important question concerns the magnitude of POC that may be exported off-shelf. In this regard, 234 Th can be used as a tracer to constrain the flux of POC from the shelf to the deep ocean during spring and summer. A water columnsediment budget of 234 Th based on in-situ production, decay, sediment burial, and offshelf export of 234 Th for the Outer and Middle domains of the eastern Bering Sea shelf indicates that 29% of 234 Th produced in the water column (equivalent to 1740±3480 dpm m -2 d -1 ) may be exported from the shelf . This water columnsediment 234 Th budget is re-evaluated here using the same data from 2009 and 2010; however, in this case only for the Outer shelf. These results yield a slightly lower estimate for off-shelf export of 24% of the total water column production of 234 Th (2160±2700 dpm m -2 d -1 ). Using this off-shelf export flux for 234 Th and assuming an average POC/ 234 Th ratio of 11±9 µmol dpm -1 (Table 2.3), an upper estimate of the off-shelf export of POC (calculated as the product of the off-shelf 234 Th flux multiplied by the POC/ 234 Th ratio) is 24±35 mmol C m -2 d -1 (Fig. 2.12). It must be emphasized that this is an upper estimate because it is entirely possible that some fraction of this POC is lost to grazing and/or preferential remineralization over the shelf.
By comparison, a recent carbon budget for the Outer Domain developed as part of BEST-BSIERP reports that on an annual basis, ~30% (~100 g C m -2 y -1 ) of total NPP (331 g C m -2 y -1 ) is lost due to lateral transport from this shelf region (J. N. Cross et al., in review). Using the same analytical approach described by Cross et al. (in review), we extrapolate the daily flux of 24±35 mmol C m -2 d -1 to an annual off-shelf POC flux of 66±96 g C m -2 y -1 (270 d year; no winter export). Despite the relatively large uncertainties in these calculations, off-shelf POC export is within a factor of ~2 of the estimated export by lateral transport from this region and ~20% of the total NPP. An additional sink for exported POC is benthic consumption and/or burial of sinking POC over the shelf. However, the measured benthic carbon consumption and estimated sediment burial for the southern Outer Domain represents a relatively small sink for annual NPP of 7±2% (23±7 g C m -2 y -1 ) and 6±3% (19±11 g C m -2 y -1 ), respectively (J. N. Cross et al., in review). Taken together, these results imply that the estimated spring and summer off-shelf POC flux may represent a significant fraction (~20%) of the reported annual NPP for the Outer Domain, and compares reasonably well with the annual loss of POC due to lateral transport from this region of the shelf.
Daily rates of NPP measured over the Outer shelf during this study period range from 17-541 (mean: 304±225) mmol C m -2 d -1 for spring and 47-246 (mean: 111±98) mmol C m -2 d -1 for summer . Using these seasonal average rates of NPP during, an off-shelf e-ratio of 0.07 and 0.21 is determined for the Outer Domain during spring and summer, respectively; however, it should be noted that these average NPP estimates  represent only a few measurements for the Outer Domain.  report monthly mean rates of daily primary production from 1978-1981 and 1997-2001 using 14 C and 13 C incubations, respectively.
For the growing season of March-June, monthly mean rates of primary production range from 45±25 (June) to 139±106 (May) mmol C m -2 d -1 for the Outer shelf, which results in an off-shelf e-ratio of 0.17 and 0.52 for May and June, respectively. Interestingly, these values bracket the fraction of POC production that is annually exported from the Outer shelf of 49% based on an earlier assessment of the carbon budget for the eastern Bering Sea by   (Fig. 2.11). Notwithstanding the estimated uncertainties in the present data set and the historical comparison, this analysis suggests that seasonal offshelf export of POC from the eastern Bering Sea shelf may represent a significant seasonal sink in the carbon budget, as originally postulated by .

Conclusions
Based on a detailed seasonal field study conducted over three consecutive years in the eastern Being Sea, we conclude that temporal decoupling exists between elevated rates of spring NPP and subsequent export as POC in early summer, as evidenced by measured e-ratios >1 in summer. While the measured POC export fluxes and estimates of the export ratio exhibit a seasonal cycle during 2008-2010, there is no apparent connection between these observations and proximity to sea-ice. Elevated rates of export in the summer may be attributed to sinking of the senescent spring bloom and increased fecal pellet production, the latter of which is consistent with the early summer emergence of abundant large zooplankton and euphausiids during cold years . In relation to the OCH, in the absence of large zooplankton, which is a common feature during warm years, juvenile and adult pollock feed on age-0 walleye pollock and thereby threaten the success of this economically important higher trophic species. Thus, an implication of this lower trophic carbon study is that during cold years, populations of large crustacean zooplankton and euphausiids not only facilitate the transfer of carbon to higher trophic animals, but may be important drivers in the enhanced summer export flux of POC, either to the benthos or off-shelf.
A further implication of the apparent temporal offset between primary and export production is the potential for a significant off-shelf transfer of POC on a seasonal basis.
Based on a 234 Th budget for the Outer shelf, we report an upper estimate for off-shelf POC export of 24±35 mmol C m -2 d -1 in spring and summer for this region. When               POC export units are mmol C m -2 d -1 . The vertical arrow represents minimum and maximum averages of P POC for late spring and early summer. Primary production values are from .

Publication Status
Manuscript III, titled "Diatom control of the autotrophic community and particle export in the eastern Bering Sea during the recent cold years ( concentrations, relative to summer, over the shelf and at the shelf break. Diatoms represent the greatest contribution to total chlorophyll a in spring and summer of cold years. This algal class also represents the majority of total chlorophyll a in particles sinking through the water column. Further, the relatively high proportion of pheophorbide a in sediment trap material indicates that sinking of zooplankton fecal pellets facilitates the export of particles through the water column. In cold years, the emergence of large diatom blooms in the spring MIZ supports the production of abundant large zooplankton. Large zooplankton are a primary food source for juvenile pelagic fishes of economically important species. Therefore, these cold year specific processes may be essential for the transfer of POC from the surface waters and the success of the economically important pelagic fishery. A consequence of a warmer Bering Sea in the coming decades is a reduction in seasonal sea-ice extent and duration. A change in seaice cover may alter the timing and magnitude of spring primary production and the flow of energy through the lower trophic levels.

Introduction
The eastern Bering Sea is characterized by some of the highest annual rates of primary production (PP) in the world ocean . Such high levels of primary production over the broad (~500 km wide) and vast (~10,000 km 2 ) shelf support one of the largest fisheries in the United States in terms of fish catch revenue (National Marine Fisheries Service). The seasonal extent and duration of sea-ice represents the most important constraint on the location, timing, and magnitude of spring primary production  and the composition of the autotrophic community . The lowest trophic levels of the ecosystem exhibits marked variability in distribution and abundance in response to changes in sea-ice in this region (e.g., . Because the physical regime exerts a strong control on the spring bloom and carbon flow through phytoplankton and zooplankton , changes in seasonal sea-ice extent and duration are predicted to affect the distribution and abundance of higher trophic level and economically important organisms in both the southeastern and northern regions of the shelf . During cold years characterized by late sea-ice retreat in the eastern Bering Sea, spring phytoplankton production is often dominated by intense diatom blooms . These diatom blooms are initiated by increasing stability of the upper water column that is induced by stratification from melt water released from the retreating ice edge in the adjacent waters, which is more commonly known as the marginal ice zone (MIZ) . More than half of the annual primary production occurs between May and July in this region, a seasonal pulse that is largely controlled by the annual retreat of sea-ice . Ice-edge blooms are terminated by nutrient limitation , with the phytoplankton community shifting from diatoms to smaller cells, including flagellates, dinoflagellates, and haptophytes (namely Phaeocystis pouchetti) . Important questions relate to how newly formed organic carbon from primary production is removed from the surface waters, and whether exported phytoplankton material reflects the actively growing population in the photic zone. Both of these questions bear on carbon linkages between the lower trophic levels and the pelagic and benthic ecosystems, and the control that the physical regime may have on economically significant animals.
The presence of the intense MIZ spring bloom during mid-spring, a characteristic of cold years and late sea-ice retreat, largely supports the production of abundant large copepods and euphausiids that are less prevalent during warm years .
Large zooplankton constitute a lipid rich prey source for young age classes of walleye pollock (Theragra chalcogramma), and the presence of these secondary producers is necessary for the success of the age-0 year class  and other pelagic consumers. Primary production may be exported to deeper waters, to the benthos as particulate organic carbon by the sinking of intact algal cells , or subject to heterotrophic grazing and fecal pellet production by zooplankton . The sinking flux of fecal pellets represents an important pathway for the transfer of POC from the surface waters in high-latitude shelf systems . Therefore, in cold years abundant large zooplankton not only support young age classes of economically important animals, they exert an important control on the export flux of particulate organic carbon (POC) from the upper water column.
This study investigates the seasonal succession of the autotrophic community and the controls that both phytoplankton and zooplankton have on the export of particulate organic matter (POM) during the recent cold years (2008)(2009)(2010) in the eastern Bering Sea.
The specific objectives of this study are to (1) characterize the seasonal evolution of the autotrophic community over the shelf and shelf break using total chlorophyll a (TChl a) concentrations and algal class specific indicator pigment ratios (pigment:TChl a), and (2) determine the essential nutrient composition of the POM sinking from the photic zone along the shelf break and the influence that zooplankton exert on particle export using degraded chlorophyll a (Σpheopigments) and C:N:P flux ratios. These objectives address two central hypotheses. First, particulate matter export is composed primarily of diatoms in spring and summer, despite the autotrophic community shifting from a diatom dominant system in spring to a more heterogeneous phytoplankton assemblage in summer. Second, during such cold years, the presence of zooplankton fecal pellets in exported material implies that large secondary producers are an important component of the particle export flux.   After recovery, the upper seawater layer was siphoned down to the seawater-brine interface, which was indicated by the discontinuity between the layers. Two trap tubes were vacuum filtered onto a pre-combusted 25 mm GF/F. A stainless steel arc-punch was used to generate a 10 mm diameter subsample from each GF/F, which was then frozen at -20°C. The subsample was used for analysis of POC and PON, with the remaining filter material analyzed for 234 Th . One full sediment trap per depth was used for HPLC pigment analysis of sinking material. For both 2010 cruises, a single trap was split into subsamples for POP and bSi analysis.

c. Analysis of POC, PON, POP, and bSi
The subsamples for POC and PON were dried at 60°C in a drying oven, fumed in a desiccator containing concentrated hydrochloric acid for 24 h to oxidize inorganic carbon, and dried again for 24h at 60°C. POC and PON were measured using a Carlo Erba-440 Elemental Analyzer (Exeter Analytical, Inc., North Chelmsford, MA, USA) . Field blanks were prepared for each set of samples by filtering 200 mL of filtered brine. An average blank for each cruise was subtracted from the gross POC and PON concentrations.
Concentrations of POP in both sediment trap and water column samples were measured using the ash-hydrolysis method, followed by orthophosphate measurement using the molybdate technique .
Biogenic silica samples were analyzed by NaOH digestion ; Teflon tubes were used for these analyses to achieve low and consistent blanks . The optical absorption of each sample was measured at 810 nm following the procedure of .

e. Pigment analysis by CHEMTAX
The abundance of specific phytoplankton groups was estimated from indicator pigment concentrations relative to total chlorophyll a (TChl a) using the CHEMTAX program (Mackey et al., 1996). The initial matrix was adapted from two previous studies that characterized relative phytoplankton abundance in the subarctic North Pacific . They determined pigment:TChl a ratios for the seed matrix by averaging minimum and maximum values listed in Mackey et al., (1996), except for diatoms in which they applied a fucoxanthin:TChl a ratio of 0.75 based on observations from a previous study . In this study, the same initial matrix was used for all water column data collected within the photic zone and for pigment fluxes. Based on a limited number of previous studies that relate accessory pigments to algal functional group in the eastern Bering Sea , this study focuses on three phytoplankton groups (diatoms, chlorophytes and prymnesiophytes). However, the CHEMTAX program provides relative abundances for eight algal classes (pelagophytes, prasinophytes, cryptophytes, dinoflagellates, and cyanobacteria).
For most algal classes, the pigment:chlorophyll a ratios used in the initial matrix generally agree to within a factor of ~2 with those calculated in the final matrix for both the water column and trap CHEMTAX analyses (Appendix A.1). By comparison, the final matrix ratios for the water column and sediment trap data fall within range of values reported by Mackey et al., (1996) for the Southern Ocean. A sensitivity analysis was conducted using fucoxanthin to chlorophyll a ratios of 0.35 and 1.1 for diatoms as a means to evaluate the consistency of both the final matrix and the autotrophic percentages of total chlorophyll a for the water column samples (Appendix A.2). For the major pigments and autotrophic groups, the final matrix pigment ratios for the three analyses (diatom fuco:TChl a ratios of 0.35, 0.75, and 1.1) are generally ~99% similar.
The final autotrophic percentages for the individual samples are also ~99% for the three analyses with varying fuco:TChl a ratios for diatoms.

a. Mixed layer hydrography
For the geographic regions assigned in this study, the average mixed layer depth (MLD) is greatest in spring (Table 3.3). The mixed layer is also consistently colder during spring, while average salinity values are indistinguishable between spring and summer (Table 3. (Table 3.4). The most abundant indicator pigment associated with the chlorophyll a measurements is fucoxanthin, which is a primary marker of diatoms. As with TChl a concentrations, the spatial distribution of fucoxanthin demonstrates substantial variability, ranging from negligible to greater than 15 µg L -1 .
Fucoxanthin concentrations that exceed 1 µg L -1 are generally associated with the spring bloom stations (Fig. 3.2b). Fucoxanthin concentrations are highly correlated with TChl a concentrations (m = 0.401x; r 2 = 0.978; p<0.001, Fig. 3 On a station by station basis, average upper water column POC and PON range from less than 7 to greater than 120 µmol C L -1 and 1 to 11 µmol N L -1 during spring (Table 3.5). During summer, depth averaged POC and PON concentrations are considerably lower ranging from 6 to 17 µmol C L -1 and 1 to 3 µmol N L -1 (Table 3.5). In

c. Pigment and particulate organic matter fluxes
The geographic patterns observed in the magnitude of the TChl a flux (mg m -2 d -1 ) are similar to that of pigments in the overlying water column (Table 3. (Table 3.6). The linear regression of fucoxanthin and TChl a in sinking particles (m = 0.259x; r 2 = 0.863; p<0.001, Fig. 3.3b) demonstrates a lower slope than the water column suspended particles. However, the mean ratios (mean: 0.31 for water column particles; 0.35 for sinking particles) are statistically similar, which suggests that material sinking from the photic zone is similar in composition to the autotrophic community with respect to fucoxanthin containing POM ( Fig. 3.3b). The presence of TChl b and 19'Hex is occasionally detected in settling material, though typically to a lesser extent relative to the overlying water column (Table   3.6). The ratio of Σpheopigments to TChl a in sinking particles is usually greater than one, indicating that material sinking through the water column is at least partly degraded due to the presence of senescent cells or zooplankton fecal pellets (Table 3.6).
Sediment trap POC fluxes determined during this field campaign have been presented elsewhere (Baumann et al., in press;. A subset of values used for the present analysis is listed in Table 3 (Table 3.6). POP fluxes are typically <1 mmol m -2 d -1 , with the higher fluxes associated with the higher rates of POC and PON export (Table 3.6). No correlation exists between the flux of bSi with either POC or TChl a export; however, the highest bSi fluxes occur at bloom stations BL (HLY0902) and MN19 (TN249).

a. Characterization of the autotrophic community and vertical export
The observation of a predominantly diatom autotrophic community in the spring and in the MIZ is consistent with previous studies of the ice-edge population  and the presence of abundant resting stage cells in the underlying sediment in this region . In particular, in spring, diatoms consistently represent the greatest contribution to TChl a over the shelf and along the shelf break. Over the shelf, diatoms represent a range of 71.5±10.8 % to 95.7±2.1% (regional mean ± 1σ) of the total chlorophyll a concentration for Regions 1-5. The contribution of diatoms in the northern (Region 6) and southern (Region 7) regions of the shelf break are on average 80.0±18.6% and 65.8±26.5% of the TChl a concentration, respectively (Table 3.7). Throughout the shelf and shelf break, other algal classes, namely prymnesiophytes, chlorophytes and cryptophytes, are present, but to a much lesser degree relative to diatoms in spring (Table 3.7). Smaller cells, such as cyanobacteria, also contribute minimally to the autotrophic community in this subarctic shelf system.
A seasonal shift in the autotrophic community is also apparent from the water column pigment distribution.. In spring, the autotrophic community is dominated by diatoms and characterized by localized blooms and high levels of standing stock chlorophyll a, whereas in summer the shelf and shelf break exhibit lower TChl a levels and a heterogeneous phytoplankton assemblage. Specifically, as the concentration of TChl a markedly decreases in the upper water column in summer, the contribution of diatoms to total chlorophyll a also decreases throughout the system. Algal classes present in relatively smaller proportions during spring are key contributors to total chlorophyll a in early summer (Table 3.7). In particular, prymnesiophytes emerge along the shelf break comprising 54.5±32.3% and 27.1±23.2% of the total chlorophyll a in Regions 6 and 7, respectively. Together with prymnesiophytes, chlorophytes and cryptophytes also become important contributors to the total chlorophyll a in the shelf break regions and over the shelf as well (Table 3.7).
Diatoms are primarily responsible for the elevated levels of total chlorophyll a (e.g., >5 µg L -1 ) in the spring and particularly at the bloom stations ( Fig. 3.4a), while chlorophytes and prymnesiophytes are typically not present during this period ( Fig.   3.4b,c). Interestingly, at lower concentrations of TChl a (e.g., <1 µg L -1 ), in both spring and summer, diatoms frequently still represent the major autotrophic contribution (Figs. 3.4a, 3.5a,b, 3.6). However, when the relative contribution of diatoms is low during the summer, chlorophytes and prymnesiophytes dominate the autotrophic community (Figs. 3.4b,c,3.5a,b As with the geographic distribution of the phytoplankton assemblage in the water column, diatoms are also typically the dominant algal class exported in sinking particles from the photic zone along the shelf break (Figs. 3.6, 3.7). On a station by station basis, there is little vertical variability in the percent composition of the major algal classes.
However, both TChl a and POC fluxes vary substantially over the upper 100 m (Table   3.8), suggesting non-preferential consumption and remineralization of sinking particles.
At three locations, all of which are in Region 6 (PIT1-HLY0803) or the southern reach of Region 7 (T1-HLY0802 and NP15-HLY0902), the average composition of the vertical flux is less than 50% diatoms. The relatively low diatom contribution at stations T1 and NP15 are associated with low TChl a and POC fluxes (Table 3.8). Interestingly, the average POC flux at PIT1 is the highest observed in summer 2008, whereas the average TChl a flux at that station is the lowest observed during the entire field program. At these stations, other algal classes, namely chlorophytes, pelagophytes, and dinoflagellates, represent the largest fraction of the sinking phytoplankton assemblage.
Apart from these few stations, the observed shift in the autotrophic community in the water column is not reflected in the phytoplankton composition of exported particles ( Fig. 3.7; Tables 3.7, 3.8). For all other stations in Regions 2 (BL), 6 and 7, diatoms represent at least 70% of the vertical flux of TChl a. This indicates that, regardless of the total chlorophyll a and POC flux from the photic zone, diatoms are the primary algal class exported from the photic zone ( Fig. 3.7).
The magnitude and seasonal progression of POC export flux, combined with differences in the ratio of pheopigments to TChl a between the upper water column and in sinking particles, provides important insights into the mechanisms controlling the export of diatoms from the photic zone. With regard to the export flux of POC along the shelf break, this region exhibits a progressive increase in the magnitude of particle flux from early spring to late spring and early summer, as noted above (Table 3.6). Based on the observation that the POC flux increases from spring to summer, and that the export population is primarily diatoms (Fig. 3.7), it has been suggested that export in early summer may be characterized by the sinking of spring and MIZ primary production in this system (Baumann et al., in press). A temporal lag in the export of spring primary production as POC in summer has also been observed in other high-latitude systems . The average ratio of the sum of pheopigments to total chlorophyll a (Σpheo:TChl a) is >1 in material exported below the photic zone in late spring and early summer (Table 3.6; 50 m sediment traps). In contrast, the Σpheo:TChl a ratio in phytoplankton derived from integrated pheopigment and total chlorophyll a stocks in the photic zone averages ~0.1. The low ratio of the autotrophic population indicates an actively growing phytoplankton community . In comparison to phytoplankton in the photic zone, the Σpheo:TChl a ratio in sinking particles at 50 m is ~8-75 times greater than in the overlying water column on a station by station basis (Tables 3.4, 3.6). That the Σpheo:TChl a ratio in sinking particles is much greater than those in the upper water column indicates that sinking POM is composed of substantially degraded chlorophyll a and consists of a combination of senescent cells and zooplankton fecal pellets.
Pheophorbide a, the degradation pigment resulting from metazoan digestion, represents ~80% of the total pheopigment concentration in sinking particles (Table 3.6).
This observation suggests that zooplankton grazing of spring primary production and subsequent sinking of fecal pellets is an important control on the vertical export of POC along the shelf break in this region. While microzooplankton abundance and grazing pressure has been reported to be largely unaffected by climate variability , cold years in the eastern Bering Sea, such as during this study, favor the production of abundant large crustacean zooplankton and euphausiids . Therefore, the export of fecal pellets produced by abundant large zooplankton may be an important mechanism controlling the vertical flux of POC from the photic zone along the shelf break in late spring and early summer during cold years.
Assuming that pheopigment fluxes are attributed to zooplankton fecal pellets, and that fucoxanthin fluxes are representative of sinking diatoms, inferences in preferential export of zooplankton fecal pellets and diatoms can be made from the differential loss rates of these pigments compared to those for POC and TChl a. Sinking loss rates for these exported constituents are estimated as the ratio of the measured flux at 50 m compared to the standing stock in the water column . At all trap locations, the loss of fucoxanthin and Σpheopigments from the photic zone represents a larger fraction of the standing stock than for either POC for TChl a (Table 3.9).
Specifically, fucoxanthin and Σpheopigment loss rates range from 14-83% and 5-100% per day, respectively. By comparison, loss rates of TChl a and POC are less than ~7% d -1 (Table 3.9). Because fucoxanthin and pheopigment loss rates greatly exceed those for POC and TChl a, it follows that diatoms are preferentially transported to depth via zooplankton grazing and subsequent sinking of fecal pellets. It must be noted that because the autotrophic community is actively growing, pheopigment concentrations are low, relative to chlorophyll a, which leads to relatively large pheopigment loss rates.
However, the implication of the high loss rates is that almost all degraded chlorophyll a is being rapidly removed by the sinking of zooplankton fecal pellets.
The POC associated with TChl a and pheopigment export can be estimated using generalized POC:TChl a and POC:pheopigment ratios of 50:1 and 15:1 . A POC:pheopigment ratio of 15:1 assumes a 70% carbon assimilation efficiency by zooplankton . POC fluxes associated with TChl a and pheopigment export range from 5-100% and 1.6-44% of the total POC flux at 50 m, respectively (Table 3.9). The total carbon export by zooplankton is likely underestimated because respiration and excretion below the photic zone during daily vertical migration by mesozooplankton represents a significant component of the total C flux ). In addition, ~20% of the POC associated with fecal pellets is likely released below the photic zone or below the deepest sediment trap in this study . These results support the hypothesis that the export of zooplankton fecal pellets represents an important component of POC export along the shelf break in this region.
b. Elemental composition of phytoplankton and zooplankton controlled export As described above, autotrophic biomass in the eastern Bering Sea during this study is composed predominantly of diatoms, and export of this algal group is largely controlled by sinking zooplankton fecal pellets in spring and early summer. A quantitative understanding of the C:P and N:P ratio of phytoplankton in the photic zone and in sinking particles can be used to make further inferences into the degree of zooplankton assimilation of these macronutrients, the efficiency of zooplankton controlled particle export, and the potential stoichiometric ratios resulting from respiration and inorganic and organic excretion by zooplankton (e.g., . On average, phytoplankton in the photic zone are rich in phosphorus relative to both carbon and nitrogen ( Fig. 3.8; Table 3.5). For the eastern Bering Sea in 2010, the average particulate C:P and N:P ratio in the upper water column was 87±44 and 12±5.1, respectively, both of which are less than the Redfield stochiometric relationship of 106:16:1 for C:N:P. Collectively, 86% and 92% of all measurements from 2010 exhibit ratios lower than Redfield for C:P and N:P, respectively ( Fig. 3.8); note that C:P and N:P ratios of phytoplankton are plotted against absolute and relative fucoxanthin (i.e., fuco:TChl a) concentrations. While stoichiometric ratios of phytoplankton cannot be differentiated between spring and summer, the C:P and N:P ratio of suspended particles is invariant with the concentration of fucoxanthin and the fuco:TChl a ratio, suggesting that C:P and N:P do not vary with changing diatom biomass or autotrophic community composition ( Fig. 3.8). In addition, the collective average ratios of C:P and N:P for spring-early summer phytoplankton are below the global mean and consistent with fast growing cells in nutrient-rich, high-latitude systems . Moreover, suspended particles in the Bering Sea are enriched in phosphorus by a factor of ~2, relative to POC compared to phytoplankton in warm, nutrient depleted oligotrophic systems .
The limited number of elevated N:P ratios at relatively low fucoxanthin concentrations (Fig. 8a). Each of these elevated N:P ratios are greater than 20 and observed in different profiles collected during TN250. These few values are not consistent with other ratios from the same profiles and may not be representative of the water column at those locations. However, elevated C:P ratios are present at higher fucoxanthin concentrations and fuco:TChl a ratios ( Fig. 3.8c,d).  .
Elemental ratios in sinking particles are substantially higher than those of water column suspended particles, suggesting either carbon enrichment or phosphorus depletion of sinking zooplankton fecal pellets, relative to suspended particles. Sinking particles (40-100 m traps) exhibit average C:P ratios of 107±71 (TN249) and 156±67 (TN250) and average N:P ratios of 24±14 (TN249) and 14±8 (TN250) for spring and summer in 2010 ( Fig. 3.9; Table 3.6). By comparison, fuco:TChl a ratios are similar in both phytoplankton and trap fluxes. For all six cruises, the average C:N of sinking particles was 7.3±2.8 (Table 3.6). The average C:N ratio is consistent with the Redfield ratio of 7.8, which indicates that phosphorus depletion is likely responsible for the high C:P and N:P values of sinking particles along the shelf break. An alternative explanation for the elevated C:P and N:P ratios is that the processes of inorganic and organic excretion by zooplankton may release relatively high proportions of phosphorus relative to carbon and nitrogen.
Zooplankton in this region demonstrate elemental stoichiometric homeostasis, meaning that their C:N:P stoichiometry does not vary as a function of food source.
Weighted average C:P and N:P ratios of three subarctic copepods (Calanus glacialis, Eucalanus sp., and Metridia pacifica) and an euphausiid (Thysanoessa raschii) are slightly higher (C:P of 176 ±52 and N:P of 35±9; Lomas et al. unpublished data) than in the passively sinking particles these organisms produce. Though the standard deviations of these data sets are large, that the average C:P and N:P of these consumers is greater than the averages of both passively sinking particles and the phytoplankton in the water column suggests that the processes of excretion and respiration by zooplankton are likely enriched in phosphorus relative to carbon and nitrogen.
The C:P and N:P stoichiometry of the combined processes of respiration and inorganic/organic excretion by zooplankton may be important for both nutrient regeneration in the photic zone and export of carbon and nutrients to depth by vertical migration. These ratios may be estimated using a mass balance of the C:P and N:P of four pools: phytoplankton food source (P), zooplankton (Z), particle flux (F), and the combined processes of respiration and inorganic/organic excretion (A). The C:P and N:P ratios of respiration and excretion are calculated independently because carbon must be assimilated at a higher rate than nitrogen based on the high C:P of zooplankton in this region. Also, the C:P and N:P ratios of these pools are inversed (i.e., 1/C:P = P:C and 1/N:P = P:N) because A is calculated with respect to carbon and nitrogen, respectively.
The mass balance equation for these nutrient pools is: where P, Z, F, and A are the P:C or P:N ratios of the four pools listed above, ae is the carbon or nitrogen assimilation efficiency by zooplankton, and f 1 and f 2 are the fractions of carbon or nitrogen available (not assimilated) for particle export or for the combined processes of respiration and inorganic/organic excretion by zooplankton, respectively.
Values of P, Z and F are 87±44, 175±52, and 129±72 for C:P, respectively, and 12±5, 35±9, and 21±14 for N:P, respectively (Tables 3.5, 3.6). With regard to estimating the C:P ratio for A, a range of zooplankton assimilation efficiencies (ae = 0.6 to 0.8; ) and f 1 values (f 1 = 0.1 to 0.3) are used, while the relationship 1-aef 1 is substituted for f 2 . For cases in which ae, f 1 , f 2 sum to 1, the average C:P ratio of A is 21±9.7 (n = 89). The same approach is used for estimating the N:P ratio of A; however, a range of lower assimilation efficiencies are used because a large fraction of the ingested nitrogen is likely excreted immediately (E. Durbin, pers. comm.). For the estimation of the N:P ratio, assimilation efficiencies range from 0.3 to 0.5 and f 1 ranges from 0.1 to 0.3.
These ranges of ae and f 1 result in f 2 values that are generally larger than those used for the C:P ratio estimation, consistent with a larger fraction of excreted nitrogen. Solving equation 1 using these parameters yields an average N:P ratio of 6.3±1.1 (n = 121) for A.
Notwithstanding that the average C:P and N:P ratios calculated for A have large associated uncertainties that include large standard deviations of the P, Z, and F pools, and that the ae, f 1 , and f 2 values are estimates from literature, these relatively low imputed ratios for A suggest that the processes of respiration and excretion are important components of both nutrient cycles in the upper water column and export of carbon and nutrients to depth.

Conclusions
During cold years in the eastern Bering Sea, the autotrophic community is dominated by diatoms in the spring and in the MIZ. Associated with a high percent composition of diatoms in spring is a greater frequency of elevated rates of net primary production and high levels of TChl a and POC in the photic zone. Despite a wide range in the magnitude of particle flux along the shelf break, the diatom algal class represents the majority of the exported total chlorophyll a. Because pheophorbide a is present in large abundance in sinking particulate matter, often at levels much greater than TChl a, the vertical transfer of diatoms is likely mediated by enhanced zooplankton grazing of MIZ primary production and subsequent export of fecal pellets in late spring and early summer. Daily loss rates of fucoxanthin and pheopigments from the photic zone exceed those for both TChl a and POC, supporting the notion that sinking of zooplankton fecal pellets exerts an important control on particle export from the surface waters along the shelf break.
This study provides new evidence of the relationship between the phytoplankton community and zooplankton mediated export during cold years in the eastern Bering Sea.
This region is predicted to warm in the coming decades , resulting in a reduction in maximum sea-ice extent and earlier retreat in spring. As a consequence, a warmer physical regime may restructure the spring autotrophic community to a population consisting of fewer diatoms, similar to summer conditions in this region. Total annual primary production may be greater in years characterized by early sea-ice retreat  however, warm years are unfavorable for large zooplankton . A reduction in large zooplankton may threaten the success of economically important animals. Associated with this climate driven shift of the autotrophic and zooplankton community may be a reduction in POC transfer to deeper waters and greater organic carbon retention within the water column. A further implication of a warming Bering Sea is the magnitude to which this subarctic shelf system sequesters carbon to the deep ocean, which may decrease in the future (Bauman et al., in press).   Table 3.3. Regional averages of mixed layer depth (MLD, m), depth of the photic zone (1% PAR, m), and percent ice at stations within each region (% Ice Cover) together with MLD averages of temperature (°C), salinity (‰), density (σ t , Densit -1000 kg m -3 ), dissolved oxygen (DO, µmol kg -1 ), and dissolved oxygen saturation from equilibrium (ΔDO Saturation, µmol kg -1 ). Table 3.4. Regional averages of primary pigment concentrations (µg L -1 ) in the upper water column: total chlorophyll a (TChl a), fucoxanthin, total chlorophyll b (TChl b), 19'-Hexanoyloxyfucoxanthin, (19'-Hex), Pheophytin a, and Pheophorbide a.   Table 3.7. Regional averages of percent contribution by algal group to total chlorophyll a.  Table 3.9. Photic zone stock of particulate organic carbon (POC, mmol m -2 ), total chlorophyll a (TChl a, mg m -2 ), pheopigment sum (Σpheopigment, Pheophytin a + Pheophorbide a, mg m -2 ), and fucoxanthin (mg m -2 ) together with daily loss rates and fraction of total POC flux associated with TChl a (F Chl a (C)) and pheopigments (F pheo (C)).  Table 2 are the BEST-BSIERP regions grouped in this study. White (spring) and black (summer) symbols represent water column sampling locations during the field study (see Table 1 for cruise information). Cross symbols are sediment trap deployment locations.      Thin bars represent average total chlorophyll a (black), fucoxanthin (grey), and Σpheopigments, and correspond to the right vertical axes.