Biogenic opal, organic carbon, organic matter stable isotope, and trace metal data from a well-dated, high-resolution jumbo piston core (EW0408–85JC; 59° 33.3′N, 144° 9.21′W, 682 m water depth) recovered from the northern Gulf of Alaska continental slope reveal changes in productivity and nutrient utilization over the last 17,000 years. Maximum values of opal concentration (∼10%) occur during the deglacial Bølling-Allerød (B-A) interval and earliest Holocene (11.2 to 10.8 cal ka BP), moderate values (∼6%) occur during the Younger Dryas (13.0 to 11.2 cal ka BP) and Holocene, and minimum values (∼3.5%) occur during the Late Glacial Interval (LGI). When converted to opal mass accumulation rates, the highest values (∼5000 g cm−2 kyr−1) occur during the LGI prior to 16.7 cal ka BP, which points to a strong influence by LGI glacimarine sedimentation regimes. Similar patterns are also observed in total organic carbon and cadmium paleoproductivity proxies. Mid-Holocene peaks in the terrestrial organic matter fraction at 5.5, 4.7, 3.5, and 1.2 cal ka BP indicate periods of enhanced delivery of glaciomarine sediments by the Alaska Coastal Current. The B-A and earliest Holocene intervals are laminated, and enrichments of redox-sensitive elements suggest dysoxic-to-anoxic conditions in the water column. The laminations are also associated with mildly enriched sedimentary δ15N ratios, indicating a link between productivity, nitrogen cycle dynamics, and sedimentary anoxia. After applying a correction for terrestrial δ15N contributions based on end-member mixing models of terrestrial and marine organic matter, the resulting B-A marine δ15N (6.3 ± 0.4 ‰) ratios are consistent with either mild denitrification, or increased nitrate utilization. These findings can be explained by increased micronutrient (Fe) availability during episodes of rapid rising sea level that released iron from the previously subaerial coastal plain; iron input from enhanced terrestrial runoff; and/or the intermittent presence of seasonal sea ice resulting from altered ocean/atmospheric circulation during the B-A in the Gulf of Alaska.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 The North Pacific Ocean contains the largest high-nutrient-low-chlorophyll (HNLC) region in the Northern Hemisphere (Figure 1a), where primary productivity is limited by iron [Martin and Fitzwater, 1988]. In contrast, the coastal regions of both the Northeast and Northwest Pacific are macronutrient-limited (e.g., nitrate and silicic acid [Whitney et al., 2005]). Cross-shelf exchange and vertical mixing of these iron-rich shallow and nitrate-rich deep basinal waters are the chief mechanisms underlying the high seasonal productivity observed in the modern coastal North Pacific (Figure 1b) [Bruland et al., 2001; Childers et al., 2005; Ladd et al., 2005].
 Iron availability controls phytoplankton productivity through its influence on the marine N cycle. Nitrogen-fixation by diazotrophic cyanobacteria requires dissolved Fe [Falkowski et al., 1998], and nitrate assimilation by large-celled phytoplankton is inhibited by low ambient-Fe concentrations [Price et al., 1994]. Iron enrichment experiments in the North Pacific have initiated substantial phytoplankton blooms, particularly among large-celled diatoms, and resulted in significant reductions in nitrate [Boyd et al., 2004; Coale et al., 1996]. Measurements of δ15N in dissolved and particulate nitrate during the SERIES iron enrichment experiment show progressive enrichment of 15NO3 as it was utilized by phytoplankton in the HNLC waters of the Northeast Pacific Ocean [Needoba et al., 2006].
 The linkages between iron availability, the N cycle, and marine primary production suggest that a change in micronutrient input could alter regional productivity through its influence on macronutrient dynamics [Martin, 1990]. Building upon this idea, Davies et al.  recently suggested inundation of subaerially exposed continental shelves during the Late Glacial Interval (LGI) / Holocene transition could serve as a major source of micronutrients to coastal marine ecosystems. To test this hypothesis, we examine a high-resolution record of paleoproductivity (based on biogenic opal, organic carbon, and cadmium accumulation), δ15N and δ13C of bulk organic matter (OM), and trace element concentrations over the last 17 calendar kiloannum before present (cal ka BP) in core EW0408–85JC (59° 33.3′N, 144° 9.21′W, 682 m water depth) from the continental slope of the northern Gulf of Alaska (Figure 1c).
1.1. Modern Gulf of Alaska Oceanographic Setting
 Ocean circulation in the Gulf of Alaska is driven by a dynamic atmosphere-ocean linkage between the Aleutian Low (AL) pressure cell and the Alaska Gyre. Between November to March, the center of the AL is located over the central Gulf of Alaska, resulting in vigorous vertical mixing, reduced sea-surface temperature (SST) and sea level pressure, increased precipitation, and enhanced upwelling in the Alaska Gyre (Figures 1a and 1b) [Trenberth and Hurrell, 1994; Rodionov et al., 2005]. During spring and summer, the AL weakens, and the water column becomes stratified by spring snowmelt and radiative warming of the upper water column. Biological productivity in the central gyre is iron-limited, and residual nitrate at the sea surface defines an HNLC area. Cyclonic Alaska Gyre circulation advects shallow nitrate-rich water toward the continental shelf of the Gulf of Alaska where it mixes with iron-rich coastal waters and contributes to large phytoplankton blooms in the spring and summer [Harrison et al., 1999; Childers et al., 2005]. Concurrent with the spring bloom is the seasonal snowmelt pulse along the margin, when freshwater discharge into the Gulf of Alaska is high, and is entrained in the Alaska Coastal Current [Royer, 2005].
1.2. Paleoceanographic Setting of the Gulf of Alaska and Other North Pacific Marginal Basins
 Relatively few high-resolution records document the paleoceanographic development of the Gulf of Alaska since the Late Glacial Interval (LGI). The complex LGI environment incorporates (1) more than 3,500 km of coastline occupied by the southern Alaska portion of the Cordilleran Ice Sheet [Kaufman and Manley, 2004], (2) complex relative and eustatic sea level changes associated with a tectonically active margin [Gulick et al., 2004], and (3) a re-organization of North Pacific Ocean circulation due to the emergence of central Beringia and closure of the Bering Strait during times of low sea level [Hopkins, 1959].
 Jumbo piston core EW0408–85JC (59° 33.3′N, 144° 9.21′W, depth 682 m) was recovered by the R/V Maurice Ewing from the continental slope along the Gulf of Alaska margin in 2004 (Figure 1c). Following retrieval, the core was cut into 1.5 m sections and analyzed shipboard using a GEOTEK MultiSensor Core Logger (MSCL) to measure high-spatial-resolution seismic velocity, gamma-ray attenuated wet bulk density (WBD), whole-round magnetic susceptibility, and electrical resistivity at 1-cm intervals [Blum, 1997]. Each core section was subsequently split, lithologies described and high-resolution linescan imagery was collected. Computerized tomographic (CT) wet bulk density data were also determined [Davies et al., 2011]. All EW0408 sediment cores are archived at the Oregon State University core repository in Corvallis, Oregon. The chronology of EW0408–85JC is based on 37 AMS 14C dates on planktonic foraminifera [Davies et al., 2011] (see Figure S1 in the auxiliary material).
2.2. Biogenic Opal
 Bulk 1-cm-thick sediment samples of ∼10 cm3 volume were collected from EW0408–85JC at 5 cm intervals. Samples were freeze-dried, homogenized and powdered by hand with a ceramic mortar and pestle, and further subsampled. The first split was treated in 2 N HCl overnight, rinsed with Millipore distilled water three times, and freeze-dried. An aliquot was measured for biogenic silica (opal) using a wet-alkali extraction modified from Mortlock and Froelich . All values are reported as 10% hydrated opal (SiO2 ⋅ 0.4 H2O) using a multiplier of 2.4 on Si. Estimated uncertainty is 4.6% (calculated as the coefficient of variation), based on replicate measurements of two internal opal-rich sediment standards, which is similar to the results of an interlaboratory comparison for this technique [Conley, 1998].
2.3. Organic Matter δ13C, Sedimentary δ15N, Elemental TOC and TN
 A carbonate-free subsample was combusted in a Costech 4010 HCNS elemental analyzer to determine total organic carbon (TOC) and total nitrogen (TN) concentrations. The analyzer was coupled to a Finnigan DeltaplusXP isotope ratio mass spectrometer for δ13C and δ15N measurements. All isotope values are reported in permil units (‰) according to the relationship
where X is the element of interest and R is the measured isotopic ratio. All carbon isotope measurements are relative to the Vienna Peedee Belemnite (VPDB) standard and all nitrogen measurements are relative to atmospheric nitrogen. Molar ratios of TN:TOC (hereafter referred to as molar N/C ratios) were calculated following Perdue and Koprivnjak . Replicate measurements of internal standards run along with TOC and TN yielded coefficients of variation of 4.4%, 6.9%, respectively, while replicate measurements of internal δ13C and δ15N standards yielded 1σ standard deviations of 0.19‰ and 0.20‰, respectively. Isotope measurements were made at the Alaska Stable Isotope Facility at the University of Alaska Fairbanks.
2.4. CaCO3, Lithic Concentrations, and Mass Accumulation Rates
 Total inorganic carbon concentrations were measured on untreated bulk sediment samples by coulometry [Engleman et al., 1985] with an estimated error of ±0.1 wt%, and then multiplied by 8.333 (the molar stochiometric ratio of CaCO3/C) to convert to CaCO3 concentrations. Because CaCO3 analysis was performed on a small subset of the larger bulk inorganic geochemical sample set, a linear regression calculated between the coulometric CaCO3 and total Ca concentrations (section 2.5) was used to develop a composite CaCO3 record (n = 43, r = 0.861, p < 0.01) with an estimated error of ±0.6%. The full suite of major biogenic sediment components was then used to estimate the bulk terrigenous content by difference according to the relationship
where L is the total lithic concentration, 2 ⋅ TOC% represents total organic matter (OM) concentration [Walinsky et al., 2009], and the biogenic phases are assumed to fully account for all other components.
 The flux rates of the biogenic and lithic phases were determined using mass accumulation rate calculations between AMS 14C dates, such that
where MARphase is the mass accumulation rate of a particular phase, DBD is the dry bulk density, and SR is the sediment accumulation rate in cm ky−1 based on the geochronology results of Davies et al. . Each MARphase was averaged between age control points [Francois et al., 2004]. As the MSCL-measured WBD and the CT-derived density show a highly significant correlation (n = 1090, r = 0.912, p < 0.01), DBD was estimated from the CT density scans, according to the relationship
This equation is a modeled linear regression between wet and dry bulk densities, as these quantities tend to be closely related, and assumes the only controls on DBD is the volume of pore water (a function of bulk density, measured by either CT or MSCL), the density of the pore water (predominantly a function of salinity in marine sediments), and the solid grain density [Blum, 1997; Weber et al., 1997]. These quantities were then assigned to be a pore water density of 1.03 g cm−3, a pore water salinity of 32, and a solid grain density of 2.65 g cm−3 (e.g., quartz).
2.5. Inorganic Geochemistry
 Samples were analyzed for bulk inorganic geochemical composition by a combination of inductively coupled plasma optical emission spectrometry (ICP-OES) and ICP mass spectrometry (ICP-MS) at SGS Minerals Services in Toronto, Canada. Samples were digested in a sequential acid leaching method using HNO3, HCl, HF, and HClO4. Samples were then divided into 2 aliquots for analysis by ICP-OES and ICP-MS that yielded a suite of 40 element concentrations. A combination of replicate samples, USGS standards (SGR-1 and PBD-1), and an international standard (NBS-SRM1646) were analyzed within the analytical runs to monitor accuracy and precision. The mean percentage difference between analyzed concentrations and literature values for the standards for all elements was 11%, and most elements are within the reported 1σ range. A mean coefficient of variation for all elements calculated from all replicate samples was 4.1%.
 Water column paleoanoxia conditions were inferred from a suite of redox-sensitive trace elements. Whereas several of these elements (S, Mo, U, and Cd) were discussed by Barron et al. , the coupling between redox biogeochemical cycling and the independent productivity data sets of opal and TOC are of primary interest in this paper [Morel and Price, 2003]. Redox-sensitive elements are predominantly derived from three sources: (1) a lithogenic source associated with terrestrially derived detrital mineral phases; (2) a biogenic source from organic particles via substitution or adsorption processes; (3) a hydrogenous source from precipitation and adsorption reactions from seawater as reduced sulfide phases sensitive to prevailing Eh conditions within the water column or sediment pore waters [Calvert and Pedersen, 1993; Tribovillard et al., 2006; Piper and Calvert, 2009]. To estimate these different contributions, the excess fraction was calculated based on normalization of the detrital fraction to a global average sediment composition, assuming a constant aluminosilicate-hosted terrigenous detrital composition where
where Mmeasured is the measured concentration of the metal of interest M, and (M/Al)avg sed is the global mean sediment Al-normalized ratio for metal M [McLennan, 1995; Van der Weijden, 2002].
2.6. Statistical Analysis
 Bivariate Pearson correlation coefficients were calculated for natural-log-transformed data sets to ensure normal distributions [Davis, 2002]. Natural-log transformation of percentages has the added benefit of circumventing the constant-sum problem [Aitchison, 1986, 1999]. For correlation calculations, all concentration data were ln-transformed prior to analysis.
3.1. Core Lithology and Geochronology
 An analysis of sediment lithology, stratigraphy, and geochronology for core EW0408–85JC is presented by Davies et al.  (Figure S1). The age-depth model indicates that this core contains a continuous record spanning the last 17.3 cal ka BP. The lowermost lithologic unit is a massive dark gray diamict that extends from 831 cm below the seafloor (cmbsf) to the bottom of the core at 1278 cmbsf, and represents rapid ice-proximal sedimentation (613 ± 409 cm ka−1) in the latter part of the LGI. This diamict is overlain by a low-density laminated hemipelagic sediment unit that occurs between 797 and 831 cmbsf (29 ± 19 cm ka−1) that spans the Bølling-Allerød (B-A) warm period identified in the North Atlantic [Rasmussen et al., 2006] and the time of sea level rise associated with Meltwater Pulse (MWP) 1a in Barbados coral records [Fairbanks, 1989; Bard et al., 1990; Deschamps et al., 2009], with an abrupt onset at 14.69 ± 0.85 cal ka BP. Between 760 and 797 cmbsf, EW0408–85JC is composed of a moderately high-density, bioturbated sediment that accumulated at a mean rate of 19 ± 11 cm ka−1, best characterized as a sandy silt containing dispersed coarse sand as ice-rafted debris with a basal age of 12.99 ± 0.19 cal ka BP, coeval with the onset of the Younger Dryas in Greenland [Rasmussen et al., 2006]. The uppermost portion of EW0408–85JC is a hemipelagic mud that extends to 760 cmbsf (or 11.16 ± 0.13 cal ka BP) that accumulated at a mean rate of 65 ± 29 cm ka−1. Although the majority of this upper unit is a massive bioturbated dark gray silty clay, the bottom 15 cm is characterized by a weakly laminated interval of lower-density sediment that was deposited during the early Holocene between 11.14 ± 0.09 and 10.75 ± 0.22 cal ka BP. This interval has been linked to a period of rapid eustatic sea level rise known as MWP-1b [Fairbanks, 1989; Bard et al., 1990].
3.2. Paleoproductivity Proxies
 Biogenic sediment concentrations suggest a high degree of variability in Gulf of Alaska marine productivity since the LGI. Both opal and TOC have higher concentrations in sediments deposited during the B-A and MWP-1b intervals than in sediments deposited during the LGI and YD (Figure 2). Natural-log-transformed opal and TOC concentrations are significantly correlated (n = 165, r = 0.718, p < 0.01), suggesting a preservational bias has not decoupled the accumulation of these biogenic phases given the different conditions under which these components are preserved [Hedges et al., 1999; Ragueneau et al., 2000]. CaCO3 concentrations are low and do not correspond with the concentrations of other biogenic components, as CaCO3 values tend to be high during both the LGI and YD intervals (Figure 2). The lithic concentrations (Figure 2) are highest during the LGI (mean 92.8%), intermediate during the Holocene (89.7%), and lowest during the B-A and MWP-1b intervals (85.0% and 86.6%, respectively). Lithic concentrations are negatively correlated with opal, TOC, and CaCO3, and positively correlated with CT density (all exceed the 99% confidence level).
 Cadmium concentrations are positively correlated with opal (n = 110, r = 0.416, p < 0.01), whereas excess Cd is only present during the B-A and MWP-1b intervals (Figure 2). Unlike most trace elements that are redox-sensitive, Cd accumulation is more sensitive to the biogenic flux than ambient dissolved oxygen concentrations due to its micronutrient status [Lee et al., 1995; Lane and Morel, 2000; Lane et al., 2005]. Cadmium mass balance calculations in Cariaco Basin sediments partitioned between the water column and detrital, biogenic, and authigenic sediment phases also indicate that Cd is dominantly of biogenic origin (86%) [Piper and Dean, 2002]. Elevated concentrations of Cd are found in TOC- and Mo-rich sediments in the Santa Barbara Basin [Ivanochko and Pedersen, 2004], the Santa Lucia continental slope [Hendy and Pedersen, 2005] and on the Pacific margin of southern Baja [Dean et al., 2006], as well as in laminated Holocene sediments along the Mexican Pacific margin, where concentrations of Cd in non-lithogenic settling particles are similar to those of planktonic values [Nameroff et al., 2002]. Authigenic Cd precipitation (probably as CdS) can occur under anoxic conditions [Rosenthal et al., 1995], though the rate of precipitation may not necessarily be controlled by dissolved sulfide concentrations [Sundby et al., 2004]. Based on these collective observations, it seems most appropriate to consider Cd as a semiquantitative indicator of enhanced biogenic sedimentation, rather than an indicator of water column and/or sediment pore water anoxia. When paired with the other measures of productivity in EW0408–85JC, these diverse multiproxy data (e.g., opal, TOC, and Cd) are consistently high during the B-A and MWP-1b intervals, moderate during the Holocene and YD, and low during the LGI.
3.3. Organic Matter δ13C, Sedimentary δ15N, and Molar N/C Ratios
 The sedimentary δ13C, δ15N, and molar N/C ratios vary between −26.3 to −22.1‰, 2.2 to 6.7‰, and 0.04 to 0.11, respectively (Figures 3 and 4). These elemental and isotopic ranges reflect mixing marine and terrestrial organic matter (OM), as well as potential variability in the isotopic composition of marine-derived OM. The LGI samples are similar to sediments derived from modern Bering Glacier outwash, whereas the Holocene sediments in EW0408–85JC are analogous to the surface-sediment OM compositions measured for offshore locations along the modern northern Gulf of Alaska shelf (Figure 3) [Walinsky et al., 2009].
 To estimate the relative proportions of marine and terrestrial OM, a linear mixing model (Table 1) was employed following the results of Walinsky et al.  on surface-sediment samples along the Gulf of Alaska margin. Three end-members are marine phytoplankton, vascular plant detritus, and soil OM [Meyers, 1994; McQuoid et al., 2001; Geider and La Roche, 2002; Gaye-Haake et al., 2005; Walsh et al., 2008]. Here, soil OM was negligible (Figure 3). Using δ13Cterr and δ13Cmarine end-member compositions of −26 and −21‰, respectively, yields mean terrestrial OM contributions (mterr) of approximately 0.50, 0.46, 0.37, 0.43, and 0.77 for Holocene, MWP-1b, YD, B-A, and LGI intervals, respectively (Figure 4). A formulation of
using molar N/C values of 0.025 and 0.125 for the terrestrial and marine end-members yield similar results (Figure 4) (see Table 1 for more information on equations (5), (6), and (7)). The downcore plot of marine organic matter-derived TOC (TOCmarine), calculated using
is similar to those of total TOC and opal (Figure 4), with maxima in the B-A and MWP-1b intervals.
Table 1. Terms and Values Used for Calculating mterr, TOCmarine, and δ15Nmarinea
Equation (5): mterr = [Asample − Amarine] / [Aterr − Amarine], where A refers to the end-member compositions in (i) or (ii) and the corresponding measured value in (iii). Equation (6): TOCmarine = TOCsample − (mterr × TOCsample). Equation (7): δ15Nmarine = [δ15Nsample − (mterr × δ15Nterr)] / mmarine, where mmarine = 1 − mterr.
(i) End-Member Compositions for Equations (5) and (7), Using Organic Matter δ13C
 A comparison of the downcore trends in molar N/C ratio, sedimentary δ13C, and δ15N data show several coherent patterns (Figure 4). LGI sediments are composed of relatively low molar N/C ratios, and depleted δ13C and δ15N values. These characteristics, along with the low opal concentrations (<4.8%) are consistent with predominantly terrestrial-derived OM and low export productivity. This contrasts with the opal maxima seen during the B-A and MWP-1b intervals, both of which are associated with moderate N/C ratios (0.07 to 0.08) and higher δ13C and δ15N (mean values of −23.2‰ and 5.9‰, respectively). The YD interval has opal values similar to Holocene concentrations (mean 6.1%), a lower mean N/C ratio of 0.09, a mean sedimentary δ13C value of −22.8‰, and a mean δ15N value of 5.3‰. Holocene sediments have moderate opal concentrations (mean 8.8%) with little variability, but in-phase shifts in the molar N/C ratio, and sedimentary δ13C and δ15N occur during the mid- and late Holocene intervals (after 7 cal ka BP).
3.4. Mass Accumulation Rates
 When the biogenic and lithic concentrations are included in mass accumulation rate calculations (equation (3a)), these proxies all show a coherent pattern of enhanced accumulation at the base of EW0408–85JC during the LGI prior to 16.7 cal ka BP, reduced accumulation during the LGI termination (16.7 to 15 cal ka BP), latest part of the B-A, and early part of the YD, and heightened peaks within the B-A and MWP-1b intervals (Figure 2). The correspondence between the biogenic fluxes, the lithic MAR, and sediment accumulation rate trends suggest that MAR calculations are mainly controlled by sedimentation rate, and are not truly representative of export productivity, especially during the LGI. Higher opal, TOC, and Cd fluxes during the B-A and MWP-1b intervals are congruent with measured maximum concentrations of these biogenic phases during these same periods, indicating that these deglacial periods were more productive.
3.5. Redox-Sensitive Elements
 Many trace elements are sensitive to reducing or oxidizing conditions in the water column, but we focus on Mn, U, and Mo because these elements are the most sensitive to oxic (Mn) and anoxic/sulfidic (U and Mo) conditions [Calvert and Pedersen, 1993; Tribovillard et al., 2006; Piper and Calvert, 2009]. Concentrations of Co and Cr track the oxic/suboxic boundary [Tribovillard et al., 2006]. High excess concentrations of Mo and U are indicative of bottom water anoxia during deposition of laminated B-A and early Holocene intervals (Figure 5), which are also contemporaneous with high TOC and opal concentrations (Figure 2). High excess Mn and Co, and minimal excess Mo, U, and Cr within the YD interval imply either suboxic or oxygenated bottom waters. The loss of all excess U and Mo during the Holocene, and the monotonic increase in excess Mn and Co concentrations, suggest increasingly oxic water conditions. Excess Cr is also highest in Holocene sediments, which may indicate some suboxia.
4.1. Relationships Between Productivity and Environmental Proxies
4.1.1. The Late Glacial Interval in the Gulf of Alaska
 The biogenic sediment preserved in EW0408–85JC presents the first high-resolution productivity record spanning the LGI to late Holocene for the northeastern Gulf of Alaska continental slope. This unique location captures primary productivity changes, but also is influenced by Cordilleran Ice Sheet dynamics during the LGI and subsequent deglaciation. For example, the order-of-magnitude difference in sedimentation rates between the Holocene and older LGI intervals (17 to 15 cal ka BP) results in mean opal MARs of 511 g cm−2 ky−1 and 5369 g cm−2 ky−1, respectively (Figure 2). The higher values in older LGI sediments seem unlikely to reflect higher productivity within the LGI Gulf of Alaska, and probably instead reflect higher burial efficiency or sediment focusing under higher sedimentation rate conditions. The LGI sediments contain low concentrations of autochthonous marine OM, as evidenced by low opal, TOCmarine, and excess Cd concentrations (Figure 2), but contain high concentrations of terrestrially derived OM (mterr > 0.7; Figure 4). Given the lack of terrestrial vegetation present along the Gulf of Alaska margin during this cold period [Peteet and Mann, 1994], it seems likely that the high concentrations of allochthonous OM are derived primarily from eroding sedimentary bedrock [Walinsky et al., 2009].
 These observations imply the LGI Gulf of Alaska continental slope was functionally similar to one of the many glaciated fjords along the modern coast, with a cold, fresh, and strongly stratified euphotic zone [de Vernal and Pedersen, 1997], high concentrations of suspended sediment, extremely low primary productivity, icebergs, and seasonal sea ice [Barron et al., 2009]. Sedimentation rates associated with the older LGI section, though consistent with glacial delivery, are too low to represent a full proximal glaciomarine sedimentation regime [Powell and Molnia, 1989; Jaeger et al., 1998].
4.1.2. The LGI/Holocene Transition
 The regional Cordilleran Ice Sheet likely retreated from the ocean onto land around 14.8 cal ka BP [Davies et al., 2011]. The majority of sea-ice and sea-ice-related diatom taxa and subarctic silicoflagellate taxa disappear from EW0408–85JC at the end of the Younger Dryas around 11.7 cal ka BP [Barron et al., 2009], and dinocyst-based transfer functions from a nearby site indicate the onset of modern regional surface salinities (>32) by at least 10.2 cal ka BP [de Vernal and Pedersen, 1997]. The onset of full Holocene conditions favored enhanced primary productivity relative to LGI levels (Figure 2), reflected by higher concentrations of opal, TOCmarine, and excess Cd as well as the lower percentage of terrestrially derived OM.
 Holocene sedimentary δ15N and δ13C values (Figure 4) are higher than in LGI sediments. This difference reflects the greater contribution of terrigenous OM contained within the LGI sediments. Holocene δ15N data are similar to modern marine nitrate δ15N (4.1 ± 0.9‰) measured near Vancouver Island [Wu et al., 1997] and global oceanic mean nitrate δ15N (5‰) [Galbraith et al., 2008b], but would be higher when corrected for dilution by terrestrial OM (Table 2). Observations from multiple high-export-productivity margins indicate bulk sedimentary δ15N has an excellent correlation with local sub-euphotic zone nitrate δ15N [Altabet and Francois, 1994; Thunell et al., 2004]. Therefore, early Holocene δ15N values between 10.7 and 7.0 cal ka BP appear to record an increase in fractional nitrate utilization by primary producers. High nitrate utilization (and hence high productivity) is suggested by the correspondence of high opal concentrations and high sedimentary δ15N values.
Table 2. Calculated δ15Nmarine Values for EW0408–85JC Chronozones From Equations (5) and (7), Using Minimum, Mean, and Maximum Terrestrial and Marine End-Member Compositions, as Well as Forcing a Mean Holocene Value of 5‰
Asterisks indicate that values were excluded because they approach end-member composition.
Organic Matter δ13C
7.9 ± 2.9
10.6 ± 0.7
7.3 ± 1.4
8.5 ± 0.9
8.0 ± 1.8
11.5 ± 0.4
8.3 ± 1.3
9.5 ± 0.8
8.4 ± 1.6
12.2 ± 0.3
9.1 ± 1.3
10.3 ± 0.7
11.1 ± 4.7
Holocene δ15N = 5‰
5.0 ± 0.9
7.5 ± 0.2
5.7 ± 0.8
6.3 ± 0.4
5.7 ± 2.0
Molar N/C Ratio
7.3 ± 1.7
5.7 ± 0.3
7.2 ± 2.3
9.1 ± 1.2
8.3 ± 78.5
7.6 ± 0.9
7.7 ± 0.4
8.2 ± 1.4
10.1 ± 0.8
10.4 ± 4.5
8.1 ± 0.8
8.9 ± 0.5
9.0 ± 1.1
10.8 ± 0.7
8.8 ± 1.7
Holocene δ15N = 5‰
5.0 ± 0.6
6.4 ± 0.2
5.8 ± 0.5
6.8 ± 0.4
4.9 ± 0.9
4.1.3. Variations in Organic Matter δ13C, Sedimentary δ15N, and the Molar N/C Ratio: Proxy for Holocene Alaska Coastal Current Variability
 Similar patterns of change in sedimentary δ13C, δ15N, and molar N/C ratio data after 7.0 cal ka BP, during an interval of fairly constant opal MAR, suggest similar changes in organic matter composition independent of marine productivity contributions (Figure 4). Low δ15N values between 6 to 3 cal ka BP can be explained by two mechanisms: (1) increased contributions of terrigenous OM, which is consistent with the sedimentary δ13C and N/C ratio trends; or (2) incomplete nitrate utilization that limits export productivity. Increased N-fixation can also result in decreased δ15N values [Carpenter et al., 1997], but modern circulation patterns in the North Pacific Ocean argue against increased N-fixation as evidenced by: (1) negative N* values throughout the water column of the Subarctic Northeast Pacific [Gruber and Sarmiento, 1997; Sarmiento and Gruber, 2006] and (2) high nitrate concentrations measured directly in the Alaska Gyre and along the adjacent Gulf of Alaska shelf [Childers et al., 2005]. Given these modern N-cycle constraints, and the synchronicity of the mid- and late Holocene fluctuations in sedimentary δ13C, δ15N, and molar N/C ratio data, these cycles likely represent discrete periods of enhanced terrigenous sediment delivery along the Gulf of Alaska slope.
 The proximity of the EW0408–85JC site to both the Copper River delta and the terminus of the Bering Glacier suggests both fluvial and glaciomarine processes contribute terrigenous sediment. While the Bering Glacier is the nearest potential glacial transport system, the Holocene advance and retreat record does not correspond to the enhanced terrestrial OM delivery during the mid-Holocene [Molnia and Post, 1995; Wiles et al., 1999; Calkin et al., 2001]. Modern satellite imagery shows distinctive sediment plumes transported westward by the Alaska Coastal Current along the northern Gulf of Alaska shelf (Figure 1c). These plumes are driven primarily by summer meltwater discharge originating in Icy and Yakutat Bays, as well as winter winds [Stabeno et al., 2004]. Surface sediments from this area are characteristically rich in terrigenous OM [Walinsky et al., 2009]. Therefore, changes in the delivery of glaciomarine sediment by the Alaska Coastal Current appear to be the most likely source of the millennial-scale shifts in terrigenous OM preserved in EW0408–85JC.
4.1.4. Trace Metal Evidence for Enhanced Holocene Vertical Mixing?
 Beginning at 10.7 cal ka BP, a monotonic increase in excess Mn concentrations corresponds with no accumulation of excess U or Mo (Figure 5). These patterns are disrupted during: (1) peaks in U, Mo, and Cd at 10.5 cal ka BP; and (2) a maximum in excess Mn around 3.0 cal ka BP that is synchronous with a CT density peak (Figure 2).
 Significant correlations exist between excess Mn and excess Fe (n = 83, r = 0.57, p < 0.01), excess Co (n = 63, r = 0.47, p < 0.01), and excess Cr (n = 83, r = 0.26, p < 0.05), suggesting association with Fe-Mn oxyhydroxide accumulation [Tribovillard et al., 2006]. Oxic conditions favor dissolved Co2+ adsorption, precipitation of solid Co3O4 [Brookins, 1988] or CoFe2O4 [Glasby and Schulz, 1999], and/or complexation with organic ligands [Saito et al., 2002]. Oxidized Cr6+ persists dominantly as aqueous CrO42- in seawater, but is reduced to Cr3+ under suboxic conditions which then favors either the precipitation of solid Cr2O3 or the particle-reactive (aqua)hydroxyl cations Cr(OH)2+, Cr(OH)3, and (Cr, Fe)(OH)3. [Calvert and Pedersen, 1993; Tribovillard et al., 2006]. While both Cr and Co-bearing sulfides are predicted to be thermodynamically stable in anoxic conditions, empirical data from a variety of marine depocenters suggests the kinetics of such reactions are too slow to result in significant enrichments [Huerta-Diaz and Morse, 1992; Morse and Luther, 1999].
 Increases in oxic-indicator trace elements (e.g., Mn and Co) in the Holocene section (Figure 5) suggest that the Holocene Gulf of Alaska continental slope has become progressively more oxygen-rich since the early Holocene. This increasingly oxic bottom water may be related to a number of factors, including enhanced vertical mixing, better ventilation of deep waters, or reduced remineralization rates of sinking particles. The increases in these oxic and suboxic indicator elements is not paralleled by trends in the productivity indicators (opal, TOC, and excess Cd; Figure 2), indicating the processes leading to the accumulation of these oxic indicators are likely independent of marine productivity. Thus, the lack of Holocene enrichments in Mo and U, and some precipitation of authigenic Mn, Co, and Cr, indicate at least suboxic bottom water conditions [Calvert and Pedersen, 1993; Tribovillard et al., 2006], trending toward more oxic conditions since 10.7 cal ka BP.
4.2. The Bølling-Allerød in the Northern Gulf of Alaska
 The Bølling-Allerød interval begins at about 14.7 cal ka BP in the North Atlantic region [Steffensen et al., 2008]. At this time, biogenic opal and TOCmarine concentrations double abruptly relative to LGI values, and Cd concentrations are at least three times higher (Figure 2). B-A sedimentary δ15N values are also 3‰ higher than during the LGI (Figure 4). The combined presence of elevated concentrations of Mo and U, low concentrations of Mn, Co, and Cr (Figure 5), as well as the presence of laminated sediments, all indicate dysoxic-to-anoxic bottom water conditions.
 High concentrations of opal, TOC, and Cd (Figure 2) in B-A sediments indicate that productivity was higher. This inference is supported by increased percentages of the diatom Thalassionema nitzschioides and the silicoflagellate Distephanus speculum [Barron et al., 2009]. Enhanced productivity during the B-A likely resulted in high export of OM to the benthos, remineralization of sinking particles within the water column and sediments, and low bottom water oxygen concentrations.
4.2.1. N Cycle Variability and Micronutrient Availability During the Bølling-Allerød (MWP-1a) and MWP-1b
Davies et al.  show that two intervals of laminated, opal-rich sediments record episodes of productivity in EW0408–85JC at 14.8 to 13 cal ka BP, and 11.2 to 10.8 cal ka BP, that correspond with the regional expressions of Meltwater Pulses (MWP-1a and −1b) [Fairbanks, 1989; Kienast et al., 2003]. Davies et al. proposed that these periods of enhanced productivity in the Gulf of Alaska and other sites in the North Pacific resulted from episodes of abrupt sea level rise that inundated LGI coastal regions and mobilized labile bio-reactive micronutrients. Because the Gulf of Alaska is a major HNLC region with documented iron limitation of phytoplankton [e.g., Boyd et al., 2004], and the EW0408–85JC core site lies on the modern boundary between the HNLC Alaska Gyre and the macronutrient-limited coastal zone [Childers et al., 2005], we focus the discussion below on relationships between iron, the N cycle, and primary productivity in light of the MWP Fe availability hypothesis.
4.2.2. Distinguishing Between Refractory and Autochthonous δ15N Signals
 The B-A interval in EW0408–85JC contains sedimentary δ15N values of 5.7 ± 0.3‰, which exceed both the modern mean global nitrate δ15N value of 5‰ [Galbraith et al., 2008b] and deep North Pacific nitrate δ15N value of 4.1 ± 0.9‰ [Wu et al., 1997]. The mterr calculations performed in section 3.2 (Figure 4) show that EW0408–85JC sedimentary OM is composed of a combination of terrigenous and marine sources, and because terrestrial δ15N contributions can lower the bulk δ15N, it is necessary to correct the bulk δ15N for better comparison to other North Pacific records to evaluate controls on δ15N variability.
 We revise our earlier linear mixing model,
where mmarine = 1 − mterr, to solve for the δ15N of bulk marine OM (δ15Nmarine), where mterr was determined using equation (5) (Table 1). Estimates of δ15Nmarine can be calculated using either the organic δ13C- or molar N/C-based solutions for mterr. However, an estimate of this term is insufficient to calculate δ15Nmarine, because equation (7) is under-determined with respect to the number of unknown end-member values necessary for a unique solution (i.e., Aterr, Amarine for either δ13C or N/C in equation (5), and δ15Nterr in equation (7)). Another shortcoming of this approach is that these carbon-based estimates of mterr are dependent upon a fixed N:C value for the terrestrial OM end-member [e.g., Meyers, 1994]. Furthermore, as terrestrial OM tends to be depleted in N relative to marine OM, small changes in N content can produce large changes in N/C values [Perdue and Koprivnjak, 2007], and thus large potential variations in mterr.
 To address these complications, δ15Nmarine was calculated using both the OM δ13C and the molar N/C ratio, each in four different ways (Table 1): (1) maximum end-member composition values, so as to calculate the highest potential δ15Nmarine values; (2) minimum end-member compositions to determine the lowest potential δ15Nmarine values; (3) mean end-member values to derive a mean δ15Nmarine record; and (4) end-member values that were optimized to yield a mean Holocene δ15Nmarine of 5‰, which is equivalent to the modern value of mean oceanic nitrate δ15N [Galbraith et al., 2008b]. Given the level of uncertainty inherent in this approach, the δ15Nmarine calculations are best interpreted on a qualitative basis.
 The resultant δ15Nmarine estimations share several features (Table 2), including enriched δ15N values during both the early Holocene and B-A. Some of the estimated values of δ15Nmarine during the LGI and the Holocene are high. In both cases, these unlikely results are due to measured compositions that approach the terrestrial or marine end-member compositions (Figure 4).
4.2.3. Enhanced Denitrification During the Bølling-Allerød? (Hypothesis 1)
 Considering the Holocene optimized δ13C solution as the most conservative estimate for δ15Nmarine (Table 2) implies that the northern Gulf of Alaska continental slope experienced δ15Nmarine ≥ 6.3‰ during the B-A (MWP-1a), whereas the earliest Holocene (MWP-1b) saw δ15Nmarine ≥ 7.5‰. These values are analogous with those seen beneath the central Alaska Gyre at ODP Site 887 [Galbraith et al., 2008a] and along the Vancouver shelf in core MD02–2496 [Chang et al., 2008] during these same time intervals (Figure 6). However, these records have not been corrected for terrestrially derived δ15N contributions, and if significant, would lead to higher δ15N values. For example, in MD02–2496 terrestrially derived organic C may be as high as 30% of the TOC content during the B-A [Chang et al., 2008]. Comparisons with Arabian Sea [Altabet et al., 2005] and eastern Equatorial Pacific δ15N records [Ganeshram et al., 1995; Ganeshram et al., 2000], as well as particulate OM in the Black Sea [Çoban-Yildiz et al., 2006], where water column denitrification has led to δ15N enrichments of >9‰, suggests relatively weak denitrification may be affecting the Northeast Pacific records during both the B-A and early Holocene. In the case of EW0408–85JC, the B-A evidence of high biological productivity, intense water column anoxia, and preserved sedimentary laminations are consistent with water column conditions leading to denitrification [Froelich et al., 1979], and thus elevated sedimentary δ15N values [Galbraith et al., 2008b]. A similar conclusion was reached by Brunelle et al. [2007, 2010] to explain the co-occurrence of suboxia, elevated diatom-bound δ15N, and enriched sedimentary δ15N in both the Bering Sea and the Sea of Okhotsk, with the added caveat that a nitrate utilization response in the δ15N data may be indistinguishable from weak denitrification.
4.2.4. Supply of δ15N-Enriched Nitrate From the Equatorial Pacific to the Gulf of Alaska? (Hypothesis 2)
 A second potential hypothesis to explain elevated δ15N during the B-A and the earliest Holocene has been proposed by Calvert et al. , namely that relatively enriched sedimentary δ15N observed in Saanich Inlet sediment during the Holocene/LGI transition on Vancouver Island at ODP Site 1033 may be due to an isotopically heavier nitrate substrate being imported into the Northeast Pacific from the low latitude Equatorial Pacific margin. This model has also been invoked for linking Pacific sedimentary δ15N maxima during the B-A interval, found on the continental shelves of Mexico, California, Oregon, and Canada [Ganeshram et al., 1995; Kienast et al., 2002; Chang et al., 2008]. The recent recognition of shallow California Undercurrent water near the Aleutian Islands [Thomson and Krassovski, 2010] suggests that such a pathway is possible, but it is only a very small proportion of the total water column (∼20% of the waters between 175 and 225 m depth in the northern Gulf of Alaska). Substantial flow of the California Undercurrent into the Gulf of Alaska in sizable volumes to cause the observed δ15N enrichments seen in EW0408–85JC would require major and unlikely changes to North Pacific circulation, particularly a weakening of the North Pacific Current and the North Pacific Drift wind belt.
 Recent modeling of the North Pacific response to abrupt cooling initiated by freshwater hosing in the North Atlantic suggests there is a potential for weakening of the Aleutian Low and the associated North Pacific Drift during B-A warming [Okumura et al., 2009]. Because this atmosphere-ocean linkage controls the intensity of the offshore “bifurcation” that occurs along the Vancouver coast, and gives rise to both the California Current and the Alaska Current (Figure 1a), weakening of the North Pacific Current would most likely diminish transport of preformed California Undercurrent nitrate northward. This mechanism would also require enhanced Alaska Gyre upwelling and advection to maintain the water mass balance in the Gulf of Alaska, which requires intensification of the Aleutian Low atmospheric cell and seems unlikely given the co-occurrence of seasonal sea-ice at this time (see below). The B-A planktonic foraminiferal δ13C isotope record of Davies et al.  also fails to support this hypothesis, in that the δ13C values become anomalously heavy during this interval, suggesting that there was little influence of deep, upwelled waters at the EW0408–85JC site.
4.2.5. Relaxation of Micronutrient Limitation in the HNLC Gulf of Alaska (Hypothesis 3)
 A final potential hypothesis to explain the enriched δ15N trends seen along the Gulf of Alaska margin is a change in the regional nutrient inventory, possibly forced by global sea level rise. An important component of the sea level hypothesis of Davies et al.  is that during MWP-1a, the 13.5 – 24 m increase in sea level that occurred in less than 500 years [Bard et al., 1990; Blanchon and Shaw, 1995; Hanebuth et al., 2000] released labile bio-limiting micronutrients stored in subaerial LGI coastal zones. These micronutrients were then laterally advected offshore from the inundated continental shelf, and subsequent vertical mixing made these micronutrients available within the euphotic zone [Chase et al., 2007; Lam and Bishop, 2008; Severmann et al., 2010]. Because the Gulf of Alaska is a downwelling margin, and there is no evidence to suggest otherwise during the B-A [Davies et al., 2011], then perhaps the locus of enhanced nitrate uptake favored by the relaxation of micronutrient limitation was driven farther offshore than today. Higher nitrate utilization offshore resulted in an 15N-enriched water mass advected into the coastal zone. This hypothesis favors enhanced nitrate uptake of the residual euphotic zone nitrate pool in areas traditionally considered micronutrient-limited, such as the open Gulf of Alaska.
 A second intense productivity and anoxia peak is observed between 11.2 – 10.7 cal ka BP (Figures 3 and 6). The timing of this peak occurs within MWP-1b identified in Barbados and Greenland [Fairbanks, 1989; Steffensen et al., 2008], and corresponds to a somewhat smaller 12 m sea level rise [Fairbanks, 1990]. Dating of Tahitian corals, however, has recently cast doubt upon the rate and amplitude of MWP-1b [Bard et al., 2010]. Nevertheless, the conceptual model described above to explain the relationship between sea level rise and North Pacific productivity may apply for the abrupt increase in δ15N during MWP-1b (Figure 6). This period is further complicated by the inundation of the exposed Bering Sea continental shelf between 12.4 – 11.3 cal ka BP [Keigwin et al., 2006] which could have affected ocean-atmosphere circulation and Fe release from shelf sediments.
4.3. Reconciling Productivity and Atmospheric Circulation in the Gulf of Alaska During the Bølling-Allerød: A Role for Sea Ice?
 The B-A Gulf of Alaska was probably cooler than today during boreal winter and may have contained seasonal sea ice despite inferred global B-A warming (Figure S2) [de Vernal and Pedersen, 1997; Barron et al., 2009]. At about the same time, terrestrial glacier advances occurred [Reger et al., 2008], and negative benthic foraminiferal δ18O excursions potentially related to brine rejection are observed [Davies et al., 2011]. Therefore, the B-A paleoenvironment along the northern Gulf of Alaska continental slope likely includes (1) enhanced export productivity that forced anoxia within the underlying water column; (2) enriched sedimentary δ15N reflecting either weak denitrification or increased nitrate utilization due to greater micronutrient availability; (3) a dominantly downwelling circulation regime similar to modern conditions; (4) enhanced meltwater flux; and (5) seasonal sea ice. Many of these conditions no longer exist in the modern Gulf of Alaska.
 The presence of seasonal sea ice within the Northeast Pacific Ocean, southward to at least 54°N latitude (core PAR87A-10, Figure S2) [de Vernal and Pedersen, 1997] has potentially major implications for Gulf of Alaska primary productivity. In particular, sea ice would be a physical barrier to vertical mixing by winter storms, which now bring macronutrients present at depth into the euphotic zone for utilization by primary producers [Childers et al., 2005]. The southward expansion of sea ice would also increase the regional albedo, thus altering the atmospheric heat budget and seasonal precipitation cycles. The Aleutian Low pressure cell is particularly sensitive to changes in the latitudinal temperature gradient [Rind, 1998], and high-latitude cooling can alter tropical and mid-tropical atmospheric convergence patterns [Chiang and Bitz, 2005; Broccoli et al., 2006]. It is possible that the winter expansion of sea ice during the B-A in the Gulf of Alaska may have altered regional atmosphere-ocean dynamics, by forcing the Aleutian Low westward into a less intense configuration, as well as contributing to a weakened sea level pressure anomaly. Outputs from the NCAR Community Climate Model suggest that the winter Aleutian Low was indeed weakened during the B-A interval [Bartlein et al., 1998]. Abundant winter sea ice thus may have lead to further deterioration of the Aleutian Low.
 Seasonal weakening of the modern Aleutian Low is associated with intensification and a northward shift of the North Pacific High along western North America [Rodionov et al., 2007]. This mechanism leads to summer coastal upwelling in California and Oregon [Hood et al., 1990]. A weakened Aleutian Low during the B-A interval was accompanied by intensified upwelling in the California Current as inferred from ODP Sites 1017 [Hendy et al., 2004] and 1019 [Barron et al., 2003], and elsewhere along the California margin [Dean, 2007], as a result of a strengthened North Pacific High. Weakened B-A Aleutian Low circulation has also been proposed on the basis of the Cave of the Bells stalagmite δ18O record in the southwestern United States [Wagner et al., 2010]. Dean  proposed a model linking synchronous changes in productivity during the B-A along the western North America margin and in the Cariaco Basin, via atmospheric teleconnections in the Walker and Hadley circulation pathways that would affect the Aleutian Low and the subtropical high-pressure systems in both the North Pacific and the North Atlantic. Although the deglacial productivity peaks in the Gulf of Alaska are not a function of coastal upwelling, regional atmospheric reorganizations could provide a physical link between paleoclimate records in such widespread localities.
 Negative impacts on Gulf of Alaska primary productivity due to inferred B-A shifts in the Aleutian Low and the North Pacific High pressure cells were apparently counter-balanced by enhanced euphotic zone stratification forced by increased meltwater production from the retreating Cordilleran Ice Sheet and the occurrence of seasonal sea ice. Downwelling appears to be persistent throughout the EW0408–85JC record [Davies et al., 2011], and these B-A changes in atmospheric boundary conditions may have increased downwelling relative to earlier LGI time, but less so than during the Holocene. The enhanced B-A marine productivity under weakened AL conditions is thus counter to traditional ecological models for the Gulf of Alaska [Weingartner et al., 2002].
 Comparisons between the modern Bering Sea and the B-A Gulf of Alaska offer some insight into the sea ice – productivity linkage, with the caveat that the physiography of the two basins is markedly different and is reflected in the underlying physical oceanography that controls vertical mixing of deep nutrients into the euphotic zone [Niebauer, 1991; Springer et al., 1996; Stabeno et al., 1998, 2001, 2010]. Along the Bering Sea continental slope, exceedingly high primary productivity is augmented by two mechanisms specifically related to sea ice dynamics: (1) enhanced euphotic zone stratification that occurs due to the reduction in surface salinity during melting, and (2) mixing between macronutrient-rich offshore waters with iron-rich inshore waters, where the iron is derived from particulates released by the melting of sea ice [Aguilar-Islas et al., 2008]. If further micronutrient reservoirs are needed to fully support the B-A Gulf of Alaska productivity maxima and high sedimentary δ15N in EW0408–85JC, then rapid eustatic sea level rise and inundation of LGI coastal regions (with release of iron from the flooded shelf regions) is a potential mechanism.
 Proxies for northern Gulf of Alaska productivity are consistently higher during the Bølling-Allerød and Holocene than during the Late Glacial Interval (LGI) and the Younger Dryas. The Bølling-Allerød interval is laminated and enriched in redox-sensitive elements, indicating productivity-driven dysoxic-to-anoxic bottom water conditions. These laminations are also associated with enriched sedimentary δ15N ratios that indicate a link between productivity and N cycle dynamics.
 The synchronicity of the abrupt Bølling-Allerød transition in Greenland ice core records, several North Pacific continental margin sites, and the Cariaco Basin argues for a global forcing mechanism. Whereas EW0408–85JC was influenced by proximity to the Cordilleran Ice Sheet, most other Bølling-Allerød records of heightened productivity from the North Pacific are far removed from glacial margins. Many of these other records can be explained by changes in upwelling intensity, but the Gulf of Alaska paleocirculation is more consistent with downwelling conditions.
 The sea level hypothesis outlined by Davies et al.  provides a testable framework within which to evaluate our proxy results from EW0408–85JC in the greater context of North Pacific paleoceanographic records. The unique location of EW0408–85JC allows us to simultaneously eliminate both glacial activity and upwelling from the potential mechanisms that lead to the North Pacific productivity increase during the Bølling-Allerød. Instead, the rapid inundation of LGI coastal regions worldwide during periods of abrupt sea level rise appears as a strong candidate for explaining this productivity phenomenon. As labile bio-reactive micronutrients were mobilized from LGI estuaries and advected into the coastal euphotic zone, the micronutrient limitation of North Pacific HNLC water was alleviated and large-scale productivity blooms were initiated along much of the margin. In turn, the export of this autochthonous organic material then led to the onset of anoxic conditions within the underlying water column and sediment, thus enabling the preservation of laminations at slope depths throughout the North Pacific. This model is further supported by our sedimentary δ15N data. The mildly enriched δ15N observed along the Gulf of Alaska margin during the Bølling-Allerød suggests a change in the locus of the micronutrient-limited HNLC front farther seaward, with subsequent advection of 15N-enriched nitrate into the coastal zone, though intense productivity leading to water column denitrification may have also played a role.
 We thank the crew and scientific party of cruise EW0408 onboard the R/V Maurice Ewing, as well as Bobbi Conard and Mysti Weber of the Oregon State University core repository. Analytical assistance provided by Andrea Krumhardt, Tara Borland, Jamie Coon, and Jennifer Addison was invaluable, as was the support of the Alaska Stable Isotope Facility. The manuscript benefited greatly from discussions with Fred Prahl, John Barron, and Lesleigh Anderson. Sedimentary δ15N data from ODP Site 887 and MD02–2496 were provided by Eric Galbraith and Alice Chang, respectively. This publication results in part from a UAF Center for Global Change Student Award to J.A.A. funded by the Cooperative Institute for Arctic Research through cooperative agreement NA17RJ1224 with the National Oceanic and Atmospheric Administration. B.P.F. was supported through NSF grant OCE-0351075. W.E.D. was supported by the USGS Earth Surface Dynamics Program. Editorial comments provided by Rainer Zahn, Alexander van Geen, and an anonymous reviewer greatly improved this manuscript.