Volume 34, Issue 8 p. 1517-1533
Research Article
Free Access

Local and Remote Forcing of Denitrification in the Northeast Pacific for the Last 2,000 Years

Yi Wang

Corresponding Author

Yi Wang

Department of Earth and Environmental Sciences, University of Michigan, Ann Arbor, MI, USA

Correspondence to: Y. Wang,

[email protected]

Search for more papers by this author
Ingrid L. Hendy

Ingrid L. Hendy

Department of Earth and Environmental Sciences, University of Michigan, Ann Arbor, MI, USA

Search for more papers by this author
Robert Thunell

Robert Thunell

School of Earth, Ocean and Environment, University of South Carolina, Columbia, SC, USA

Search for more papers by this author
First published: 09 August 2019
Citations: 8

Abstract

Sedimentary δ15N (δ15Nsed) has been widely applied as a proxy for water column denitrification. When combined with additional productivity proxies, it provides insights into the driving forces behind long-term changes in water column oxygenation. High-resolution (~2 years) δ15Nsed and productivity proxy records (total organic carbon [TOC], Si/Ti, and Ca/Ti) from Santa Barbara Basin, California, were generated from a well-dated Kasten core (SPR0901-03KC). These records reveal the relationship between Southern California upwelling and oxygenation over the past 2,000 years. Inconsistencies between Si/Ti (coastal upwelling proxy) and TOC (total export productivity proxy) suggest wind curl upwelling influenced Southern California primary productivity, especially during intervals of weak coastal upwelling. Coherence between δ15Nsed, TOC, and drought indicators supports a local control of δ15Nsed by atmospheric circulation, as persistent northerly winds associated with an intensified North Pacific High pressure cell lead to enhanced coastal upwelling. In the northeast Pacific, δ15Nsed is used as a water mass tracer of denitrification signals transported north from the eastern tropical North Pacific (ETNP) via the California Undercurrent. A 1,200-year δ15Nsed record from the Pescadero slope, Gulf of California, lies between denitrifying subsurface waters in the ETNP and Southern California. During the Medieval Climate Anomaly, coherence between Pescadero and Santa Barbara Basin δ15Nsed indicates connections between ETNP and Southern California on centennial timescales. Yet an out-of-phase relationship occurred when the Aleutian Low was anomalously strong during the Little Ice Age. We suggest intensified nutrient-rich subarctic water advection might have transported high-15N nitrate into Southern California when the California Undercurrent and ETNP denitrification weakened.

Key Points

  • Wind curl upwelling contributes to Southern California primary productivity, especially during weak coastal upwelling intervals
  • Intensified NPH leads to stronger denitrification through enhanced coastal upwelling and reduced rainfall
  • California receives relatively more tropical water during the Medieval Climate Anomaly and more subarctic water during the Little Ice Age

1 Introduction

Highly active biogeochemical processes in oxygen minimum zones (OMZs) play a significant role in global nutrient cycling through their impact on nitrogen (N) cycles. In the three largest OMZs (eastern tropical North Pacific [ETNP], eastern tropical South Pacific, and the Arabian Sea), water column denitrification and anammox (anaerobic ammonium oxidation) observed at ~200–800 m (Brandes et al., 1998) account for almost a half of the total oceanic N loss (Ganeshram et al., 1995; Ganeshram et al., 2002). Thus, OMZs cast control on global nutrients through the fixed N inventory and ocean nutrient limitation, contributing to the regulation of CO2 levels (Altabet, 2006b; Altabet & François, 1994; Deutsch et al., 2004; Ganeshram et al., 2002; Kienast et al., 2002). OMZs are anticipated to expand in a warming world due to reduced gas solubility and intensified stratification of the water column, yet the brevity of O2 concentration observations in these OMZs (<50 years) makes separating long-term (multidecadal to centennial) natural oceanic variability from anthropogenic influences difficult.

OMZ intensity controls water column denitrification, as nitrate becomes the favorable electron acceptor after dissolved O2 is depleted in microbe-regulated organic carbon (OC) remineralization. δ15N is widely accepted as a proxy for water column denitrification (Altabet et al., 1995; Altabet et al., 1999; Thunell et al., 2004). Preferential removal of 14N by denitrification progressively enriches 15N in the remaining subsurface nitrate pool, which is then advected throughout the ocean. When this δ15N-enriched nitrate is upwelled into the photic zone and incorporated in particulate OC that is exported to the seafloor, the subsurface δ15N signal is preserved in the sediments, leading to elevated sedimentary δ15N (Altabet, 2006b; Altabet & François, 1994; Deutsch et al., 2004; Ganeshram et al., 2002; Kienast et al., 2002). When O2 supplies are reduced (e.g., due to lower solubility in a warmer climate) and/or O2 consumption increases (e.g., due to greater availability of OC for remineralization), sedimentary δ15N increases alongside intensified water column denitrification. To use δ15N as an indicator of O2 concentration, complete nitrate utilization in the photic zone must occur and δ15N cannot be compromised during sedimentary diagenesis (Prokopenko et al., 2006; Thunell et al., 2004). Incomplete nitrate consumption in the photic zone, however, leaves a high δ15Nnitrate signature due to preferential uptake of 14N in photosynthesis, complicating sedimentary δ15N interpretations (Altabet & François, 1994). The relative contribution of different N sources may also alter δ15N of the fixed N pool and sedimentary δ15N records (Figure 1). Although oceanic N is usually sourced from N fixation (δ15N = −2–0‰), atmospheric deposition (δ15N ≈ −2‰), and terrestrial input (δ15N ≈ 4‰; Altabet, 2006b; Sigman et al., 2009a), increasing anthropogenic atmospheric deposition can decrease surface water δ15N (Ren et al., 2017), and δ15N-depleted terrestrial inorganic carbon input can bias sedimentary δ15N toward lower values (Kienast et al., 2005). Additionally, remotely advected water masses with different δ15N signatures supply extra N to the photic zone in upwelling regions (e.g., Southern California margin, Figure 1; Liu & Kaplan, 1989), transmitting unique δ15N signatures to sediments.

Details are in the caption following the image
The nitrogen cycle in Southern California. Nitrogen inputs include dissolved N2 via N fixation, atmospheric precipitation, terrestrial input, and remotely advected water masses (subpolar nutrient-rich waters and eastern tropical North Pacific [ETNP] dentrified waters shown in blue dashed arrows). Nitrogen outputs include water column and sedimentary denitrification. Internal cycling (e.g., remineralization and assimilation) is denoted with white dashed arrows. Kinetic fractionation effects (ε) and isotopic values for major N sources and transformation pathways are labeled.

Previous studies have shown that >1‰ δ15N shifts in the Arabian Sea and the ETNP occurred between glacial and interglacial, as OMZs contracted during cool intervals and expanded as climate warmed (Altabet et al., 1995; Ganeshram et al., 1995; Pride et al., 1999). Although δ15N is assumed to be relatively stable in the late Holocene (Altabet, 2006a), several millennial-scale δ15N records have shown linkages between OMZ variability and the Intertropical Convergence Zone (ITCZ) migration (Agnihotri et al., 2008; Salvatteci et al., 2014). A more southerly ITCZ position associated with centennial/millennial-scale Northern Hemisphere (NH) cooling coincides with lower surface productivity and reduced denitrification off Peru (Agnihotri et al., 2008), supporting an oceanic OMZ response to climate forcing via large-scale atmospheric circulation.

A 2,000-year high-resolution (~2 years) δ15Nsed record from the well-dated Kasten core SPR0901-03KC (34°16.99′N, 120°2.408′W; 586-m depth) in the Santa Barbara Basin (SBB), Southern California, was generated to explore long-term natural variability of δ15N in response to water column oxygenation and/or N flux changes. Paired TOC and the scanning X-ray fluorescence (XRF) elemental analyses constrain the impacts of regional productivity and terrestrial N input on sedimentary δ15N. These records show coherence between δ15Nsed, export productivity, and local precipitation to reveal a local control of the Southern California denitrification. The SBB δ15Nsed was also compared to the 1,200-year δ15Nsed record from the Pescadero slope, Gulf of California (Tems et al., 2016), and the Mount Logan ice core record (Osterberg et al., 2014) to investigate the coherence of northeast Pacific δ15N variability in response to larger-scale (regional or global) processes (e.g., tropical and extratropical forcing). We will demonstrate competing influences of high-δ15N saline water from the ETNP OMZ and oxygenated fresh water from the subarctic ocean following multidecadal to centennial-scale climate change.

2 Background

The SBB is a semiclosed basin located on the Southern California margin. Sedimentation in SBB is associated with annual couplets formed by alternation of biogenic-rich (light laminae under the X-ray radiography) and siliciclastic-rich sediments (dark laminae under the X-ray radiography). The lithogenic component accounts for ~70–80% of SBB sediments (Thunell et al., 1995) and is delivered to the basin via rivers draining the Western Transverse Ranges (Hendy et al., 2015). Under California's Mediterranean climate, the North Pacific High (NPH) weakens and is displaced equatorward in winter, allowing a strengthened Aleutian Low (AL) to steer precipitation toward Southern California, driving increased river runoff and silicilastic sedimentation (Checkley & Barth, 2009; Warrick & Farnsworth, 2009a).

In spring and summer, the NPH strengthens and migrates poleward, resulting in strong east-west pressure gradients that drive northerly upwelling-favorable alongshore winds, generating strong coastal upwelling (Checkley & Barth, 2009; Chelton, 1981). Nutrient-rich subsurface water upwells along the coast, producing spring-summer plankton blooms. In addition to coastal upwelling that is typically restricted within 5–30 km along the coast (Checkley & Barth, 2009), offshore upwelling (up to 200 km) also plays a role in generating the high productivity observed in SBB. In the Southern California Current System (CCS), negative wind curl in the North Pacific subtropical gyre is balanced by the positive wind stress curl near shore (Checkley & Barth, 2009; Pickett, 2003; Rykaczewski & Checkley, 2008). Upwelling driven by the positive wind curl usually has much lower velocities (0.1–0.2 m/day vs. 10–20 m/day for coastal upwelling; Pickett, 2003). However, the volume transport of wind curl upwelling is significant due to greater areal extent (Chelton et al., 2007; Jacox et al., 2014; Münchow, 2000). Together, coastal upwelling and wind curl upwelling in SBB result in the annual formation of biogenic-rich sediment layers during times of NPH dominance. Finally, low-O2 bottom water (<20 μmol/kg) preserves varves formed by the seasonal shift between biogenic and silicilastic sedimentation (Hendy et al., 2015; Schimmelmann et al., 1990; Schimmelmann et al., 1992), while high sedimentation rates (~1 mm/year) minimize sedimentary diagenesis, such that SBB sediments retain the original subsurface δ15N signal (Prokopenko et al., 2006).

SBB waters are affected by the equatorward California Current (CC) and poleward California Undercurrent (CUC; Figure 2). As a part of the North Pacific Gyre, the CC originates in the bifurcation of the North Pacific Current (NPC; Checkley & Barth, 2009). Occupying the upper 500 m and strongest at the surface, the CC transports cold, fresh, and oxygenated water from the subpolar region (Checkley & Barth, 2009; Hickey, 1978). CC strength is connected to large-scale gyral circulation and atmospheric forcing (strength of the trade winds/westerlies). When the NPC is stable, CC transport is generally anticorrelated with Alaska Current strength (Rykaczewski & Checkley, 2008). However, when the NPC intensifies, both CC and Alaska Current transport increase, while when the NPC weakens, transport decreases (Cummins & Freeland, 2007). Most observed low-frequency CC variability is associated with NPC transport changes (Cummins & Freeland, 2007). Satellite altimetry reveals in-phase NPC strength and Ekman pumping variations in the subpolar and subtropical gyres (Cummins & Freeland, 2007), further linking CCS strength to gyral circulation behavior. Finally, where the NPC bifurcates on the North America margin affects SBB water properties, as a poleward displacement leads to transport of fresher, nutrient-rich water from the subpolar gyre into the CCS (Freeland & Cummins, 2005; Sydeman et al., 2011).

Details are in the caption following the image
Core locations. (a) Yellow circle: Core SPR0901-03KC (34°16.99′N, 120°2.408′W; 586-m depth) from the Santa Barbara Basin (SBB); red circle: Pescadero Slope core location from Tems et al. (2016). Red star: ice core location from Mount Logan (Osterberg et al., 2014); red triangle: the YOK-I speleothem record from Belize (Kennett et al., 2012). Summer and winter Intertropical Convergence Zone (ITCZ) positions are indicated in white belts bound by solid and dashed lines, respectively. Ocean currents are shown in white arrows. KaC: Kamchatka Current; Oyashio C: Oyashio Current; KuC: Kuroshio Current; CC: California Current; CUC: California Undercurrent; GoA: Gulf of Alaska; AC: Alaska Current; AS: Alaska Stream. (b) Southern California Bight map corresponding to the black rectangle in a. The cores SPR0901-03KC and SPR0901-04BC (34°16.895′N, 120°2.489′W, 588-m water depth) are shown in the yellow circles. The CalCOFI Station 81.8 46.9 is represented by the purple circle, and the circulation pattern is modified from Hickey (1992). The base maps are generated from the Ocean data View in (a) and the GeoMapApp (http://www.geomapapp.org, Ryan et al., 2009), respectively.

CUC is a subsurface poleward flow with a core depth of ~200–300 m occupying the nearshore region (within 25–40 km off the shelf break), which advects warm, salty, and low-oxygen water from the ETNP up the coast of North America (Hickey, 1978). Nutrient-rich CUC waters are upwelled to the surface along the coast, supporting biological productivity in the southern CCS (Hickey, 1978; McClatchie et al., 2016). Biannual CUC intensification is observed in June and December (Chelton, 1984; Lynn & Simpson, 1987) and was linked to local processes (upwelling-enhanced subsurface flow in spring-summer and the strong Southern California Eddies in winter; Connolly et al., 2014; Hickey, 1978). Recent studies, however, indicate that coastal-trapped Kelvin waves control CUC intensity, which propagate a sea level signal from the equator, providing a further connection between the southern CCS and the Tropics (Gómez-Valdivia et al., 2015; Gómez-Valdivia et al., 2017).

3 Methods

Core SPR0901-03KC was scanned using an ITRAX XRF core scanner (Cox Analytical Instruments) at the Large Lakes Observatory, University of Minnesota, Duluth. The scanner was equipped with a Cr X-ray tube and was operated at 200-μm resolution with an 8-s scan time at 30 kV and 15 mA. The output data are recorded as counts per 8 s and are semiquantitative (Croudace et al., 2006; Hendy et al., 2015). A split of SPR0901-03KC was sampled continuously at 2-mm interval (~2 years per sample). Individual samples were freeze-dried and ground to produce bulk samples for δ15N and TOC measurements. Bulk sedimentary δ15N was measured on a Euro Elemental Analyzer interfaced to a GV Isoprime continuous flow isotope-ratio mass spectrometer at University of South Carolina on unacidified samples. δ15N is defined as [(15N/14Nsample)/(15N/14Nstandard)−1]×1,000 with the standard of atmospheric N2. The reference standards used for data normalization were N-1 (δ15N = 0.4‰), N-2 (δ15N = 20.41‰), N-3 (δ15N = 4.7‰), and USGS-40 (δ15N = −4.52‰). For TOC measurements, ~250-mg aliquots were acidified to remove carbonate with 10-ml 5% HCl on a 50 °C hotplate for three times. Acidified samples were oven-dried at 65 °C for at least 48 hr and then ground for TOC measurements. Eight to 12 mg of acidified samples was loaded into tin capsules and measured on a Costech ECS 4010 Elemental Analyzer at University of Michigan. Acetanilide (C = 71.09 wt. %) and atropine (C = 70.56 wt.%) were used as standards, and the standard deviation of repeated measurements was within 2%. Flood and turbidite layers were removed from the geochemical time series as they do not reflect background marine sedimentation and siliciclastic sediment input causes significant dilution (Hendy et al., 2013).

The age model of SPR0901-03KC was constructed by correlation with nearby cores. Forty nine mixed planktonic foraminifera accelerator mass spectrometry 14C dates (Du et al., 2018; Hendy et al., 2013) from SPR0901-06KC (34°16.914′N, 120°02.419′W, 591-m water depth) were mapped on the master core SPR0901-03KC using sediment fabric characteristics. Varve counted dates of marker layers (e.g., gray layers at 1861–1862 CE and 1761 CE, the Macoma layer at 1841 CE, and a turbidite layer at 1811 CE) were used instead of 14C to constrain the past 300 years (Hendy et al., 2015; Schimmelmann et al., 1992). Instantaneous packets of sediments produced by floods and turbidites were removed from the core depths. An age-depth model for 03KC scanning XRF records was then generated using Bacon 2.2 (Blaauw & Christen, 2011; Du et al., 2018), where 14C ages were converted to calendar ages using the Marine13 calibration curves (Reimer et al., 2013) with variable reservoir ages from Hendy et al. (2013). This age model was then applied to the suite of SBB cores using 31 tie points including known turbidites, flood layers and 12 visually distinct additional marker horizons determined by distinguishable core fabric differences (e.g., varve color and thicknesses).

Geochemical data were interpolated to obtain evenly spaced time series (sampling resolution of 2.28 years after interpolation) prior to statistical analyses. Cross-wavelet analysis was used to calculate squared wavelet coherence and phase differences on a time-frequency plane to reveal regional coherence between individual time series (Grinsted et al., 2004; Torrence & Compo, 1997). All wavelet coherence is calculated using the analytical Morlet wavelet (central frequency ω0 = 6) in MATLAB.

4 Results

Bulk sedimentary δ15N in the core SPR0901-03KC varies between 6.78‰ and 8.33‰ with a mean value of 7.74‰ (Figure 3b). There is no long-term trend through the record; however, several low-δ15Nsed intervals occur, including ~1000–1100 CE during the Medieval Climate Anomaly (MCA) and 1460–1750 CE during the Little Ice Age (LIA). δ15N values decline from ~1800 to the core top (~1900 CE).

Details are in the caption following the image
Local influences on δ15Nsed in the Santa Barbara Basin. (a) The first principal component (PC1) of the scanning X-ray fluorescence elemental data for SPR0901-03KC (black line) used as a proxy for siliciclastic sediment derived from river runoff (Heusser et al., 2015). Higher PC1 indicates wetter conditions while low PC1 indicates drought; (b and c) δ15Nsed (‰; blue line) and total organic carbon (TOC; wt. %; red line) records from SPR0901-03KC, respectively. Blue bars indicate cool intervals: Dark Age Cold Period (DACP) and Little Ice Age (LIA). Red bars indicate warm intervals: Medieval Climate Anomaly (MCA). All instantaneous depositional events (flood and turbidite layers) have been removed.

Bulk TOC concentrations are low at the base of the core (from 170 BCE to 0) with a minimum value of 2.63 wt.% (Figure 3c), and generally increase toward the core top, varying between 4.89 wt.% and 3.77 wt.% (excluding instantaneous depositional events). TOC concentrations are not statistically significant different between cooler (e.g., Dark Age Cold Period at 400–765 CE [Helama et al., 2017] and LIA) and warmer climate intervals (e.g., MCA), yet they have a statistically significant correlation with δ15N (r = 0.0760, p < 0.05). Despite overall coherence, the positive correlation disappears between 950–1550 CE and after 1800s. The elemental, isotopic, and organic geochemical composition of OC in SBB indicates marine sources with significant contributions from terrestrial OC, notably in flood sediments. Background SBB sediment δ13C values of −21.75‰ support a marine source, as terrestrial contributions (primarily from the Santa Clara River (Warrick & Farnsworth, 2009b) are characterized by bedload sediment δ13C values of −28.15‰ (Meyers, 1997).

The first principal component (PC1) of the scanning XRF elements in SPR0901-02KC has been used as a proxy for lithogenic sediment delivery to SBB by river runoff, with higher PC1 corresponding to wetter intervals (Figure 3a; Heusser et al., 2015). PC1 is anticorrelated with TOC (r = −0.4004, p < 0.0001) and has a statistically significant negative correlation with δ15Nsed (r = −0.1617, p < 0.001). Stronger coherence between PC1 and δ15Nsed is observed during wetter intervals (50–100 CE, 950–1130 CE, 1530–1700 CE), whereas the correlation is lost during drier periods (770–1000 CE). This negative correlation also disappears after ~1800 as δ15Nsed decreases monotonically (Figures 3a and 3b).

Scanning XRF elemental records of Ca/Ti and Si/Ti are used here as proxies for inorganic carbon (Hendy et al., 2015) and biogenic silica (Brown et al., 2007), respectively. Ca/Ti varies on a decadal to centennial timescale but no longer-term trend is observed. Several low biogenic silica periods are observed in the Si/Ti record, including 450–890, 1000–1100, 1150–1260, 1310 to ~1370, and 1520–1650 CE (Figure 4e). These intervals correspond to intervals of low upwelling silicoflagellate (Distephanus speculum) and diatom (Rhizosolenia spp.) abundance (Barron et al. (2015), Figure 4d). A significant positive correlation is found between Ca/Ti and TOC (r = 0.3532, p < 0.001). High Si/Ti (high biogenic silica) generally coincides with increased TOC (e.g., ~1100–1170 CE and ~870–1000 CE, Figure 4). Yet a weak anticorrelation is observed between TOC and Si/Ti (r = −0.0941, p < 0.05), notably between 570–970, 1000–1100, 1150–1270, and 1320–1370 CE.

Details are in the caption following the image
Productivity proxies. (a) Scanning X-ray fluorescence (XRF) Br/Cl (proxy for organic carbon, thin purple line). Thick purple line represents the 101-point running mean; (b) total organic carbon (TOC; wt. %; thin red line) with 11-point running mean (thick red line); (c) scanning XRF of Ca/Ti (thin brown line) with the 101-point running mean (thick brown line); (d) upwelling diatoms and silicoflagellates from Barron et al. (2015). The teal and blue line represent the relative abundance of Distephanus speculum and Rhizosolenia spp., respectively; (e) proxy for biogenic silica, Si/Ti, from scanning XRF (thin green line) with the 101-point running mean (thick green line). Blue shaded rectangles indicate intervals of low coastal upwelling indicated from upwelling diatom and silicoflagellate abundance in (d).

5 Discussions

5.1 Export Productivity Proxies and Upwelling

In Southern California, upwelling (coastal and wind curl) brings subsurface high-nutrient denitrified waters to the euphotic zone to support export productivity. Water column denitrification (increasing δ15N) is subsequently intensified by increased export productivity when OC remineralization uses nitrate as the electron acceptor in low-O2 waters. Export productivity and upwelling variability thus need to be reconstructed using biogenic sediments (e.g., TOC, biogenic carbonate, and silica) to understand local water column denitrification.

TOC is commonly employed as a proxy for carbon export from the upper ocean. An overall increase of TOC (~1 wt. %) occurs over the last two millennia, indicating a general increase in export productivity, with significant variability on centennial timescales (Figure 3c). However, the bulk sediment TOC concentration could be subject to sedimentary diagenesis, causing the record to deviate from carbon export. In high sedimentation rate settings (>0.03 cm/year, Canfield, 1994), OC could continue to decompose downcore via anaerobic pathways in reducing porewaters, leading to lower sedimentary TOC (Canfield et al., 1993). Additionally, higher TOC values (enhanced OC preservation) could occur beneath instantaneous depositional events (e.g., flood and turbidite layers) that reduce O2 penetration and aerobic OC degradation (i.e., the “coffin-lid” effect, Schimmelmann, 2011).

To exclude the fore-mentioned complexities, an independent export productivity indicator is needed to account for likely diagenetic processes affecting TOC records. Inorganic carbon determined by scanning XRF Ca/Ti is a productivity proxy representing biogenic carbonate production (primarily foraminifera and coccolithophores, Figure 4c). In SBB, well-preserved inorganic carbon is unaffected by sedimentary anaerobic OC degradation and/or enhanced OC preservation and thus provides an independent measure of export productivity that can be compared to TOC. Ca/Ti and TOC are generally coherent on decadal to centennial timescales. This statistically significant correlation (r = 0.3532, p < 0.01) suggests TOC preservation below instantaneous deposition events was not enhanced nor were there significant changes in TOC anaerobic decomposition downcore. Thus, sedimentary TOC is likely primarily recording export productivity in SBB.

Additionally, biogenic silica (largely produced by diatoms and silicoflagellates) is another major contributor to export productivity, especially during coastal upwelling events. Scanning XRF Si/Ti is used here as a proxy for biogenic silica, with the expectation that higher Si/Ti corresponds to enhanced export productivity. Coherent TOC and Si/Ti maxima are observed (e.g., 1130 and 1300 CE), yet negative correlations between Si/Ti and TOC are shown during intervals of low Si/Ti, during which TOC remains relatively stable (Figures 4b and 4e). Anticorrelations between TOC and biogenic silica have also been recorded in the Gulf of California, where biogenic silica is the primary biogenic sediment component (Pichevin et al., 2012; Thunell, 1998b). Previous studies have related this inverse relationship to either Fe limitation (Firme et al., 2003) and/or enhanced silica preservation during strong coastal upwelling events due to silica supersaturation in porewaters (Pichevin et al., 2012). However, SBB sediments are dominated by lithogenic input (50–80%), with much lower biogenic silica contributions (~15–20% in SBB vs. up to ~75% during upwelling in the Guaymas Basin, Gulf of California; Thunell, 1998a; Thunell, 1998b). Thus, SBB porewaters are always undersaturated for silica, as indicated by the absence of weakly silicified species (Barron et al., 2015; Reimers et al., 1990). Iron limitation due to upwelled Fe-depleted waters and/or low riverine input usually occurs along narrow shelf areas (e.g., the northern and central California coast; Bruland et al., 2001; Firme et al., 2003). However, relatively high dissolved Fe (dFe) concentrations (>1 nM) have been observed at the surface in SBB with increasing dFe in depth (up to ~30 nM at 560 m), arguing against reduced OC production because of Fe limitation (John et al., 2012; King & Barbeau, 2011). Therefore, the lack of TOC and biogenic silica correspondence in SBB requires another explanation.

In the Gulf of California, negative correlations between biogenic silica and TOC usually happen during intervals associated with strong coastal upwelling (Pichevin et al., 2012) In SBB, however, negative relationships are more prominent during weak biogenic silica production intervals indicated by the low abundance of coastal upwelling diatom (Rhizosolenia spp.) and silicoflagellate species (Distephanus speculum; Figure 4d; Barron et al., 2015). Despite undersaturated porewaters leading to biogenic silica dissolution, Si/Ti shows consistent variability with independently measured upwelling diatoms and silicoflagellates. Sedimentary Si/Ti increases when Rhizosolenia spp. and Distephanus speculum become abundant at 1110–1160, 1270–1310, and 1370–1490 CE, supporting the interpretation of intensified coastal upwelling (Figure 4d; Barron et al., 2015). Low Si/Ti is coincident with a scarcity of these upwelling species during 570–870, 1000–1100, 1155–1265, and 1310–1370 CE, suggesting weakened coastal upwelling. Si/Ti thus appears to be an indicator of coastal upwelling but not necessarily export productivity, and can be decoupled from TOC when coastal upwelling is weak.

Nevertheless, anticorrelations between Si/Ti and TOC contradicts sediment trap studies in SBB, where export particulate OC is positively correlated with opal fluxes on an annual basis (Thunell et al., 2007). This may relate to the limitation of the short duration sediment trap study. Weak coastal upwelling in the paleoproductivity record is sustained on decadal to centennial timescales, such that this observed low-frequency variability might be associated with processes that have not yet been observed in the annual trap data.

When biogenic silica indicates weak coastal upwelling, Ca/Ti and TOC support normal to increased productivity (e.g., 1000–1100 and 1310–1370 CE, Figures 4b, 4c, and 4e), indicating that nutrients are being supplied by processes other than coastal upwelling. Discrepancies between high foraminifera production and biogenic silica were also observed in the northern CCS during the Last Glacial Maximum and were attributed to increased wind curl upwelling (Ortiz et al., 1997). Coastal upwelling is usually associated with high nutrient delivery due to high vertical velocity and a shoaling of the nearshore nutricline (Rykaczewski & Checkley, 2008; Taylor et al., 2015), producing diatom blooms (high Si/Ti) and physically larger plankton. In the southern CCS, modeled total upwelling transport (including both coastal and wind curl upwelling), however, is not significantly correlated with the coastal upwelling index calculated from atmospheric sea level pressure (Bakun, 1973). Rather, the nearshore high primary productivity band contains both coastal (<50 km) and wind curl upwelling nutrient contributions (50–200 km; Jacox et al., 2014).

The contribution of wind curl upwelling is difficult to reconstruct, as the bloom-forming taxa associated with strong coastal upwelling events (e.g., upwelling diatoms that favor high nutrient environments) can overprint physically smaller planktonic taxa that dominate offshore and more oligotrophic environments associated with slow and broad wind curl upwelling (Ortiz et al., 1995). Yet the importance of wind curl upwelling on southern CCS planktic biomass is widely reported. Wind stress modeling shows strong wind curl upwelling transport adjacent to coastal promontories (e.g., Point Conception; Pickett, 2003)—a prediction corroborated by wind stress and upwelling rate observations (Enriquez & Friehe, 1995). The biological significance of wind curl upwelling has been observed in seasonal offshore (100 km) zooplankton abundances (Chelton et al., 1982). Chlorophyll a concentrations (proxy for primary productivity) are significantly correlated with the wind curl upwelling but not the coastal upwelling rate and corroborate the active role of curl-driven upwelling in Southern California primary productivity (Rykaczewski & Checkley, 2008). Wind curl upwelling could thus be an important driver of export productivity in SBB over the last 2,000 years, such that intervals of weaker coastal upwelling (low Si/Ti) may have been offset by greater wind curl upwelling to maintain stable/high export productivity (TOC and inorganic carbon; Figure 4).

5.2 SBB δ15N History Over the Past 2,000 Years

Use of δ15Nsed as a proxy for water column denitrification requires both complete nitrate utilization and the absence of sedimentary diagenesis. Diagenetic isotopic alteration of δ15N should be negligible in SBB as high sedimentation rates and low-oxygen bottom waters only allow a small fraction of aerobic OC decomposition before burial (Altabet et al., 1999; Prokopenko et al., 2006). A minimal (<0.5‰) offset between the sediment trap δ15N time series and downcore δ15Nsed records supports this premise (Davis et al., 2019). Preservation of the original δ15N of sinking OC thus allows water column denitrification reconstructions from the δ15Nsed record in SBB. Resolution of our record (~2 years), however, is insufficient to resolve O2 entrainment/solubility shifts induced by seasonal wind-driven mixing/upwelling oscillations. Our δ15Nsed record, therefore, can only reveal water column δ15N variability averaged over decadal or longer timescales.

Despite an ~1 wt.% increase of TOC over the last 2,000 years (Figure 3c), relatively invariant δ15Nsed values (varying within ~1‰) indicate a general stability of water column oxygenation and/or δ15N input from different N sources. Exceptions to this stability occurred, however. Sustained low δ15Nsed values during ~1100–1300, 1460–1750 CE, and after 1800s (>1‰ decline), suggest the presence of more oxygenated waters or increasing δ15N-depleted N inputs (Figure 3b). The post-1800s decreasing δ15Nsed trend has also been observed in the Santa Monica Basin and in the ETNP, where it has been associated with decreasing trade wind strength, reduced equatorial upwelling, and ETNP OMZ contraction (Davis et al., 2019; Deutsch et al., 2014). The stability of the δ15Nsed record suggests that denitrification appears insensitive to sea surface temperature (SST) change during the warm (MCA) and cold periods (Dark Age Cold Period or LIA) in the late Holocene (Figure 3; PAGES 2k Consortium, 2013). This may be related to limited NH mean SST variability (typically <1 °C for LIA, Moberg et al., 2005), resulting in a gas solubility change (~5 μmol/kg given 1 °C SST change at 16 °C) that was insufficient to impact O2 concentrations and thus denitrification.

Low δ15Nsed intervals (~1100–1300 CE and 1460–1750 CE) are usually coincident with high values of the first principle component of the scanning XRF elements (PC1; Figure 3, r = −0.1617, p < 0.001), which have been associated with greater Southern California rainfall (Hendy et al., 2015; Heusser et al., 2015). This anticorrelation may be related to changes in N source between wet and dry climates as enhanced clay-bound N delivery adds δ15N-depleted ammonium (~2–4‰; Schubert & Calvert, 2001; Sigman et al., 2009b) that may substitute for K+ in illite clay structures (Kienast et al., 2005; Müller, 1977; Schubert & Calvert, 2001). Yet the TOC-total nitrogen plot (Figure 5) shows a negative intercept, indicating that the contribution from terrestrial clay-bound N is negligible.

Details are in the caption following the image
Cross plot of total organic carbon (TOC; wt. %) and total nitrogen (TN; wt. %) from SPR0901-03KC. Black line represents linear relationship from the least squares regression.

More likely, the relationship between rainfall (PC1) and δ15Nsed is indirect via the NPH and related to export productivity. The positive correlation between bulk sedimentary TOC (export productivity proxy) and δ15Nsed (r = 0.0760, p < 0.05) indicates a general productivity control on water column denitrification, with enhanced OC export increasing O2 demand and thus denitrification. The negative correlation between TOC and PC1 (r = −0.4004, p < 0.001) further suggests an inverse relationship between export productivity and local precipitation that is controlled by the NPH. A stronger and/or more persistent NPH enhances alongshore northerly winds, which induce stronger coastal upwelling and the upward advection of δ15N-rich subsurface waters to support higher export productivity (higher TOC, Figure 3). Simultaneously, this persistent high-pressure over western North America reduces rainfall in Southern California and lowers PC1, leading to a negative correlation between PC1 and TOC (Hendy et al., 2015; Heusser et al., 2015). Despite the overall local productivity control on δ15Nsed via the NPH, TOC, δ15Nsed, and PC1 still show discrepancies (notably in 750–1250 and 1700–1900 CE), indicating that remote processes/teleconnections must have also played a role.

5.3 Teleconnections to Tropical and High-Latitude Forcing

Inconsistencies between δ15Nsed and local NPH control on export productivity (TOC) and precipitation (PC1) require an alternative explanation involving remotely advected δ15N signals via ocean currents. Currents transport nitrate with unique δ15N signatures from remote water sources. The nitrate is subsequently incorporated into sinking particulate N and preserved in sediments (Figure 2), and thus, δ15Nsed can be used as a water mass tracer to track N contribution changes from different sources. In Southern California, water mass properties are primarily controlled by two competing water mass influences: high-δ15N, warm, saline (spicy) waters from the ETNP OMZ transported by the CUC, as well as nutrient-rich, cold, and fresh subarctic waters advected by the CC (Kienast et al., 2002; Liu & Kaplan, 1989; Figure 2). To investigate impacts of remote water mass advection, the Mount Logan in Yukon Territory (a high-latitude site) and the Pescadero Slope in the Gulf of California (a tropical site) are used as potential regions communicating with the Southern California OMZ (Figure 1). The 1,200-year δ15Nsed record from the Pescadero Slope serves as an end member for ETNP δ15N-rich spicy water, as the site is located at the northern edge of the ETNP OMZ and records water column denitrification in the ETNP (Tems et al., 2016). The competing subpolar end member is represented by the 1,200-year North Pacific Index reconstruction from the Mount Logan ice core (Osterberg et al., 2014). Sodium ion concentrations in the Mount Logan ice core are interpreted as an indicator of the North Pacific sea level pressure and winter AL intensity (Osterberg et al., 2014). Thus, we use the ice core record to link the atmospheric forcing (AL intensity) with the oceanic responses (Ekman pumping and NPC variability; Ishi & Hanawa, 2005) that result in subarctic water transport into the CCS. Here the focus is on decadal to centennial timescales, as the data resolution (~2–3 years for δ15N) is insufficient to resolve interannual variability.

5.3.1 Equatorial Water Influences and Atmospheric Forcing

Spicy and high-δ15N waters from ETNP could be advected to Southern California via CUC to elevate δ15Nsed in SBB. A positive correlation between δ15N from a SBB box core (SPR0901-04BC, 34°16.895′N, 120°02.489′W, 588-m water depth) and measured salinity at the core of CUC (σθ = 26.4–26.5, Gay & Chereskin, 2009) from the CalCOFI station 81.8 46.9 (center of SBB, 34°16′29.64″N, 120°1′30″W) during the last 50 years supports this assertion (Figure 6).

Details are in the caption following the image
Cross plot of δ15Nsed of the Santa Barbara Basin core SPR0901-04BC (34°16.895′N, 120°02.489′W, 588-m water depth) and mean annual salinity of σθ = 26.4–26.5 from the CalCOFI Station 81.8 46.9 (34°16′29.64″N, 120°1′30″W). The black dashed line shows the least squares regression.

Water column denitrification (δ15N) in ETNP has been associated with tropical climate through trade wind strength (Deutsch et al., 2014). Weak easterly trade winds reduce upwelling and lower export productivity, leading to ETNP OMZ contraction that subsequently reduces δ15N. The strength of trade winds is related to ITCZ migration in response to interhemispheric temperature differences. A southward shift of the ITCZ with extratropical cooling in the NH relative to the Southern Hemisphere is usually accompanied by intensified northeast trades in the NH and weakened southeast trades in the Southern Hemisphere (Broccoli et al., 2006; Chiang & Bitz, 2005; Haug et al., 2001; Jacobel et al., 2016; McGee et al., 2018; Meehl et al., 2008; Schneider et al., 2014). Because the ITCZ primarily resides in the NH (Philander et al., 1996; Xie, 1994), a southward ITCZ shift weakens the easterly trades at the equator, deepening the thermocline and reducing upwelling in the eastern equatorial Pacific (Costa et al., 2017; Koutavas & Lynch-Stieglitz, 2004).

The linkage between ITCZ migration and ETNP δ15N is further demonstrated by the relationship between Pescadero Slope δ15Nsed values and a δ18O speleothem record (YOK-I) from Belize that is sited on the northern edge of the ITCZ (Kennett et al., 2012). From 1850 to 2004, YOK-I δ18O values significantly correlate with Pescadero δ15N values (Figure 7, r = −0.5494, p < 0.001). The observed southward displacement of ITCZ since ~1850 (Hwang et al., 2013; Ridley et al., 2015; Rotstayn & Lohmann, 2002) could have slackened east trades and weakened upwelling at the equator, reducing export productivity and decreasing Pescadero δ15Nsed (a record that represents ETNP δ15N). Cross-wavelet coherence between the two records further indicates significant antiphase coherence on decadal (~1100–1300 CE) to centennial timescales (Figure 8e), supporting a persistent linkage between ETNP δ15N and the ITCZ over the past 1,200 years (Figure 8).

Details are in the caption following the image
Comparison of δ15Nsed of the Pescadero Basin (red line) with δ18O of the Yok Balum Cave (YOK-I, black line, Kennett et al., 2012) showing a statistically negative correlation (r = −0.5494, p < 0.001). ITCZ = Intertropical Convergence Zone.
Details are in the caption following the image
δ15N regional comparison with North Pacific teleconnections. (a) Mount Logan North Pacific Index proxy record (purple line, Osterberg et al., 2014). Lower values indicate stronger winter Aleutian Low; (b) δ15Nsed from Santa Barbara Basin core SPR0901-03KC with the thick purple line showing the 11-point running mean to highlight multidecadal variability; (c) δ15Nsed of the Pescadero Basin (Tems et al., 2016) with the thick red line denoting 11-point running mean; (d) YOK-I speleothem δ18O as a proxy for Intertropical Convergence Zone (ITCZ) migration (Kennett et al., 2012). The 41-point running mean is shown in the black solid line. (e) Cross-wavelet coherence between δ18O in (d) and Pescadero δ15Nsed in (c). Arrows show phase differences between the two records when the magnitude-squared coherence is above 0.5 (>95% significance). Arrows pointing left indicate antiphase correlations at that frequency. The Medieval Climate Anomaly (MCA) and the Little Ice Age (LIA) are shown in orange and blue shaded bars, respectively. Grand solar minima are indicated with gray bars. Rectangles shaded in gray show time intervals where Santa Barbara Basin and Pescadero δ15Nsed records are not correlated.

The SBB and Pescadero δ15Nsed records are usually in-phase on decadal to centennial timescales during the MCA (e.g., 980–1120 CE, Figures 8b and 8c). Thus, the ITCZ position increased Southern California (SBB) δ15N through the advection of denitrified ETNP waters in the region via the CUC during the warm NH climate interval. However, this relationship collapses between ~1320–1450 and 1670–1880 CE (Figure 8). The antiphase correlation in 1320–1450 CE could be attributed to 14C dating uncertainties producing a 20- to 30-year offset between the two records. Yet the out-of-phase pattern during 1670–1840 CE (1720–1840 CE in particular) cannot be explained by age model offsets given its duration of centuries. Coincident with NH cooling during the Maunder (1645–1715 CE) and Dalton (1790–1820 CE) solar minimum, the 1670–1840 CE interval stands out as a period of equatorward ITCZ migration (YOK-I δ18O; Kennett et al., 2012) where the weakened ETNP OMZ resulted in low Pescadero δ15Nsed values (Tems et al., 2016; Figures 8c and 8d), while SBB δ15Nsed values increased (Figure 8b). As Pescadero δ15Nsed values are always higher than those in SBB, this out-of-phase pattern reduces the δ15N difference between the two sites. One explanation could be enhanced poleward transport of ETNP waters to Southern California. However, an equatorward ITCZ displacement would have led to similar equatorward migration of the NPH and AL (Christoforou & Hameed, 1997; Lechleitner et al., 2017), suppressing coastal upwelling in Southern California while enhancing upwelling in the Gulf of California (Barron & Bukry, 2007; Pérez-Cruz, 2017). Reduced the sea surface height at the southern boundary of CCS would decrease poleward CUC flow, producing a δ15N result opposite to the one observed (Connolly et al., 2014; Taylor et al., 2015). Another possibility is that during this interval of weak Southern California (SBB) and ETNP (Pescadero Slope) water mass communication, subarctic water influences became more prominent to produce the discrepancy between 1670 and 1840 CE.

5.3.2 Subarctic Water Transport and Atmospheric Forcing

Nutrient-rich subarctic water transported via CC is associated with the latitudinal position and intensity of the NPC (Cummins & Freeland, 2007; Di Lorenzo et al., 2008; Freeland & Cummins, 2005; Sydeman et al., 2011). The CC intensifies following increased NPC transport (Cummins & Freeland, 2007; Douglass et al., 2006). Serving as the boundary between the Gulf of Alaska (GoA) and northeast Pacific subtropical gyre, the strength of NPC is closely linked with in-phase variations of Ekman pumping in both gyres, and directly responds to atmospheric forcing (Cummins & Freeland, 2007). This covariability of subpolar and subtropical gyres is similar to the North Pacific Gyre Oscillation proposed by Di Lorenzo et al. (2008), which has been used to explain nutrient variability in CCS. As no North Pacific Gyre Oscillation reconstruction is available for the last two millennia, the Mount Logan ice core North Pacific Index reconstruction is used instead as an indicator of winter AL intensity and thus is associated with Ekman pumping in the GoA (Osterberg et al., 2014).

When the connection between Southern California (SBB) and ETNP (Pescadero Slope) was weak between 1670 and 1840 CE, an intensified winter AL suggested by the ice core record (Figure 8a; Osterberg et al., 2014) could have resulted in anomalous cyclonic wind stress curl and stronger Ekman upwelling in GoA. Stronger wind stress curl would in turn lead to intensified southward transport of subarctic water into the NPC and subsequently the CCS (Freeland, 2003; Murphree et al., 2003). A similar anomalous southward intrusion of subarctic waters to CCS was observed in the 2002 summer, when the subarctic anomaly extended more than 1500 km along the U.S. West Coast, to at least 33°N (Bograd & Lynn, 2003; Huyer, 2003; Strub & James, 2003). Due to incomplete nitrate utilization in the subarctic ocean, nitrate δ15N in the GoA photic zone is also elevated (up to 11‰ at the surface, Casciotti et al., 2002). Increasing advection of nutrient-rich and high δ15N subarctic water would then serve as another source of high-δ15N nitrate for surface SBB waters. The observed correspondence between intensified AL and increased SBB δ15Nsed thus indicates a stronger connection with the subarctic waters when the tropical influences weakened during the LIA (1670–1840 CE, Figure 8).

6 Conclusions

Marine dissolved oxygen concentrations play an important role in biogeochemical cycles and can be impacted by climate changes. The high-resolution 2,000-year sedimentary δ15N (δ15Nsed) record from SBB reveals natural variability of Southern California water column oxygenation and highlights competition between tropical and subarctic water masses as surface ocean currents respond to climate forcing. Proxies for siliceous plankton bloom events (diatom and silicoflagellate populations, and scanning XRF Si/Ti) and export productivity (TOC) suggests that coastal upwelling and export productivity are not always in phase. This incoherence indicates a potential role for wind curl upwelling in driving SBB primary productivity, especially during intervals of weak coastal upwelling. The correspondence of δ15Nsed to export productivity (TOC) and local precipitation (PC1) indicates that an intensified and/or persistent NPH pressure cell was associated with enhanced export productivity during drought intervals.

Comparison with the Pescadero slope δ15Nsed record in the Gulf of California (Tems et al., 2016) supports subsurface tropical water influences on the surface waters of the Southern California Bight. The Pescadero δ15Nsed record suggests that ETNP denitrification was associated with ITCZ migration on multidecadal to centennial timescales and supports the stronger tropical water transport into Southern California during the MCA. An out-of-phase relationship between the Pescadero and SBB δ15Nsed records during the LIA (1670–1840 CE) occurred during an interval of anomalous intensified AL activity (Osterberg et al., 2014). During this interval we suggest that an intensified NPC advected more subarctic water into the CCS. Enhanced advection of nutrient-rich subarctic waters might have introduced high-δ15N nitrate to SBB, resulting in elevated SBB δ15Nsed values while δ15Nsed decreased at the Pescadero site.

The observed natural variability of water column denitrification is associated with both local (export productivity shifts controlled by NPH) and remote water mass influences related to large-scale atmospheric forcing and thus addresses regional teleconnections within the CCS. A warming climate might lead to a strong, MCA-like connection between the Southern California margin and ETNP, with weaker influences from subpolar regions. Such natural oscillations would continue to be embedded in future climate change with anthropogenic forcing, which will further complicate predictions of future ocean deoxygenation.

Acknowledgments

This work is supported by the National Science Foundation under Grant OCE-1304327 awarded to I. H.; Y. W. acknowledges support from the Rackham Graduate School of the University of Michigan and the Scott Turner Award. We thank Eric Tappa, Alexander Postmaa, and Madeline Parks for lab assistance. δ15N and total organic carbon data are included in the supporting information.