The stable isotope signature of nitrogen in the sedimentary organic matter (δ15Nbulk) is an established proxy for water-column denitrification and for understanding changes associated with nutrient utilization (Thunell et al., 2004; Robinson et al., 2009; Martinez and Robinson, 2010; Dubois et al., 2011, 2014; Tesdal et al., 2013; Wang et al., 2019; Riechelson et al., 2024). An increase (or decrease) in nutrient availability in relation to nutrient demand results in an increase (or decrease) in δ15N values (Wada and Hattori, 1978; Montoya et al., 1990). When the oxygen in the ocean is depleted, either due to global warming or increased remineralization, denitrification rates in the water column are also increasing and so is δ15N (Wang et al., 2019). Therefore, δ15N_bulk can be an important tool for reconstructing past changes in denitrification in the ODZs.

The δ15N records from the bulk sediment can be subject to interlinked processes/or sources which can complicate their interpretation. For example, diagenetic alteration during sinking in the water column and burial in the sediment (Altabet and Francois, 1994; Lourey et al., 2003), as well as terrestrial or shelf sources of organic and inorganic nitrogen (Schubert and Calvert, 2001; Kienast et al., 2005; Meckler et al., 2011), and remotely advected water masses with different δ15N values (for e.g., Southern Californian margin; Liu and Kaplan, 1989), could influence the δ15N signatures in sediments. Nevertheless, Tesdal et al. (2013) proposed that δ15N_bulk can be a reliable indicator for individual locations reflecting the oceanographic conditions of the surrounding environments.

The nitrogen isotopes of organic matter bound and protected within the calcite shell of planktic foraminifera (δ15N_FB) are less subjected to diagenesis or sedimentary contamination than δ15N_bulk and can be used to understand major nitrogen transformations occurring in the ocean (Ren et al., 2012; Studer et al., 2021). There are well-documented disagreements between bulk sediment δ15N and foraminifera-bound δ15N records, particularly in glacial-interglacial comparisons (Studer et al., 2021). While δ15N_bulk suggests strong variability in water-column denitrification between the LGM and deglaciation, δ15N_FB records indicate a more moderate change, with a peak during deglaciation but relatively stable values during the LGM and Holocene. This highlights that δ15N_bulk and δ15N_FB may reflect different aspects of the nitrogen cycle (Studer et al., 2021). Recent studies (Auderset et al., 2022; Hess et al., 2023; Moretti et al., 2024) based on δ15N_FB have shown that water column denitrification decreased and ODZs contracted during warmer-than-present periods of the Cenozoic. In contrast, Riechelson et al. (2024) used δ15N_bulk and hypothesized that the decrease in δ15N_bulk values over the Holocene is related to a decrease in Southern Ocean nutrient utilization and not due to a decrease in denitrification.

Foraminifera account for a major part of benthic denitrification in the ODZs (up to 100 % in some environments) (Piña-Ochoa et al., 2010a, b; Glock et al., 2013; Dale et al., 2016; Choquel et al., 2021; Rakshit et al., 2025). Some species, for example B. spissa, which are abundant in ODZs in and around the Pacific Ocean (Glock et al., 2011; Fontanier et al., 2014) can use NO3- as an electron acceptor (see Fig. 1a) and thus can denitrify (Risgaard-Petersen et al., 2006; Piña-Ochoa et al., 2010a, b). A study by Glock et al. (2019) proposed for some denitrifying foraminifera, denitrification is their preferred respiration pathway. The uptake of NO3- by these foraminifera is likely through pores in the test (see Fig. 1a). Nitrate is completely denitrified to dinitrogen gas (N₂) partly by the foraminifera themselves (Risgaard-Petersen et al., 2006; Woehle and Roy et al., 2018; Orsi et al., 2020; Gomaa et al., 2021), and partly supported by prokaryotic endobionts (Bernhard et al., 2012a, Woehle and Roy et al., 2022). To date, benthic foraminifera are the only eukaryote holobiont known to perform complete heterotrophic denitrification (Risgaard-Petersen et al., 2006; Kamp et al., 2015). Every Bolivina species tested so far (including Bolivina seminuda), can denitrify (Piña-Ochoa et al., 2010a; Bernhard et al., 2012b), suggesting that denitrification is a common survival strategy of Bolivinidae under oxygen-depleted conditions (Glock et al., 2019). This makes species of this genus particularly suitable candidates for reconstructing past nitrate levels using pore characteristics as a proxy. In low-oxygen environments, such as the ODZs off Peru, Costa Rica, and the hypoxic Sagami Bay, B. spissa increase their pore density with decreasing ambient NO3- availability (Govindankutty Menon et al., 2023). Thus, the pore density of several Bolivina species, such as B. spissa, and B. subadvena, is an empirically calibrated proxy that shows the strongest correlation with the bottom-water nitrate concentration (see Fig. 1b) (Glock et al., 2011; Govindankutty Menon et al., 2023) rather than bottom-water oxygen, temperature, water depth, salinity or pore water nitrate.

We used downcore samples from the Eastern Tropical South Pacific, ETSP (Gulf of Guayaquil (M77/2-59-01), Eastern Tropical North Pacific the ETNP (Mexican Margin, MAZ-1E-04), the Gulf of California (Guaymas Basin, DSDP-64-480), and the Sea of Okhotsk (MD01-2415), over the last ∼ 20 000 years (Fig. 3). The Gulf of Guayaquil sediment core M77/2-59-01 (03°57.01^′ S, 81°19.23^′ W, recovery 13.59 m) was collected from the northern edge of the ODZ at a water depth of 997 m during the RV Meteor cruise M77/2 in 2008 (Mollier-Vogel et al., 2013, 2019; Nürnberg et al., 2015). The piston core MAZ-1E-04, Mexican Margin (22.9° N, 106.91° W) was collected on board the RV El Puma at a water depth of 1463 m. The CALYPSO giant piston core MD01-2415 (53°57.09^′ N, 149°57.52^′ E, recovery 46.23 m) was recovered from the northern slope of the Sea of Okhotsk at 822 m water depth during the WEPAMA cruise MD122 of the RV Marion Dufresne (Holbourn et al., 2002; Nürnberg and Tiedemann, 2004). The Deep-Sea Drilling Project core DSDP-480 (27°54^′ N, 111°39^′ W) from the Gulf of California was retrieved at a water depth of 747 m close to the Guaymas Basin. For details on the sampling procedure of foraminiferal specimens, please refer to the Supplement Methods section in the Supplement.

All specimens of B. spissa and B. subadvena were imaged using a Scanning Electron Microscope (Hitachi Tabletop SEM TM4000 series) at Hamburg University, Germany with an accelerating voltage of 15 kV using a back-scattered electron (BSE) detector (Further methodological details are provided in the Supplement).

Following the image analysis, pore density data of benthic foraminifera from the four ODZs were used for the quantitative reconstruction of [NO3-]_BW (Fig. 4). We distinguished five different time intervals, including the Last Glacial Maximum (LGM; 22–17 ka BP), Heinrich Stadial 1 (H1; 17–15 ka BP), Bølling – Allerød (BA; 14.7–12.9 ka BP), Younger Dryas (YD; 12.9–11.7 ka BP), Early Holocene (EH; 11.7–8.2 ka BP) and Middle to Late Holocene (MLH; 8–0 ka BP) to describe the [NO3-]_BW in the East Pacific and the Sea of Okhotsk. We present updated chronostratigraphies of the studied cores, primarily based on accelerator mass spectrometry (AMS) radiocarbon (¹⁴C) dating, as detailed in the Supplement Methods.

The [NO3-]_BW from all cores were calculated using the calibration equation; 1[NO3-]BW=-3896(±350)PD+61(±1) where PD is the pore density of benthic foraminifera (Govindankutty Menon et al., 2023).

The standard error of the mean (SEM) for one sample was calculated using the equation; 2SEM[NO3-]BW=SD[NO3-]BWn where n is the number of specimens analyzed in each sample and SD is 1 standard deviation of mean reconstructed [NO3-]_BW. 3SD[NO3-]BW=(350XPD)2+(-3896XSDPD)2+(1)2 A complete error propagation was done for the calculation of the errors of the reconstructed [NO₃]_BW including both the uncertainty of the mean PD within the samples and the uncertainties of the calibration function. The reconstructed [NO3-]_BW and the calculated SEM and SD of each sample are shown in the Supplement.

We have measured sedimentary nitrogen isotope (δ15Nbulk) rather than δ15N_FB from cores taken from the Sea of Okhotsk, and Gulf of California, because the low abundances of foraminifera were utilized for other analysis. The analysis of bulk sediments allows for high-resolution records. Prior to the δ15N_bulk measurements, the Total Nitrogen (TN %) content of 20 sediment samples from the Sea of Okhotsk and 54 samples from the Gulf of California were measured at the Institute for Geology, Hamburg University, Germany using a flash combustion method with a Eurovector EA-3000 analyzer. The δ15N_bulk measurements for both the Sea of Okhotsk and the Gulf of California were accomplished at the Max Planck Institute for Chemistry (Mainz), Germany using a DELTA V ADVANTAGE Isotope Ratio Mass Spectrometer (IRMS) equipped with a FLASH 2000 Organic Elemental Analyzer. The results were expressed in standard δ-notation (Eq. 4). The standard deviation (± SD) of all individual analysis runs based on a certified international reference standard (USGS65) and internal laboratory standards (L-Phenylalanine and L-Glutamic acid) referenced to certified international reference standards was <0.3‰. The δ15N_bulk data for the Sea of Okhotsk and the Gulf of California are shown in Supplement Table S1. 4δ15N(‰)=[(15N:14Nsample/15N:14Nair)-1]×1000 For the Gulf of Guayaquil core M77/2-59-01, the δ15N_bulk data published by Mollier-Vogel et al. (2019) was used. Their measurements were done on ∼ 5–50 mg of homogenized and freeze-dried bulk sediments using a Carlo-Erba CN analyzer 2500 interfaced directly to a Micromass-Isoprime mass spectrometer at Bordeaux University. Results are expressed in standard δ-notation (Eq. 4) relative to atmospheric dinitrogen gas (N₂).

The reconstructed [NO3-]_BW from each location is subtracted from the modern [NO3-] present at the respective locations from similar water depths the cores were retrieved from. This provided the [NO3-] offset which is the difference (Δ[NO3-] (µM)) between the modern [NO3-] and the past reconstructed [NO3-]_BW. The modern [NO3-] for each location was taken from World Ocean Atlas 2018 (Garcia et al., 2019). The details are given in the Supplement.

The Sea of Okhotsk core MD01-2415 covers the Younger Dryas, (YD, 12.8 ka BP) until the Middle to Late Holocene (MLH, 4.9 ka BP). The reconstructed [NO3-]_BW values range from 32.8 to 44.1 µmol kg⁻¹ (Fig. 4a). A gradual increase in [NO3-]_BW is observed from the Younger Dryas to the Middle to Late Holocene. At the beginning of the Younger Dryas at 12.8 ka BP, [NO3-]_BW were relatively high and then decreased to a minimum value of 32.8 µmol kg⁻¹ at 12.4 ka BP. Since then, [NO3-]_BW steadily increased until the Middle to Late Holocene (MLH, 44.1 µmol kg⁻¹) (Fig. 4a). The [NO3-]_BW during the Middle to Late Holocene (mean =41.2 µmol kg⁻¹) is significantly (t-test, p=0.023) higher than during the Younger Dryas (mean =36.7 µmol kg⁻¹). The sedimentary δ15N_bulk record covers the interval from the Late Heinrich Stadial 1 (H1, 15.4 ka BP) to the Middle Holocene (6.1 ka BP). The δ15N_bulk values were relatively high ranging from 7.1 ‰ to 9.4 ‰ with an average of 8.7 ‰. The δ15N_bulk values increased steadily from the Late Heinrich Stadial 1 (15.4 ka BP) to the Early Holocene (EH, 10 ka BP) with higher values centered between the Late Younger Dryas (11.9 ka BP) and the beginning of the Early Holocene. Since then, the δ15N_bulk values decreased until the Middle to Late Holocene.

The analyzed sections of DSDP Site 480 covered the Last Glacial Maximum (22 ka BP) until the Early Holocene (10.8 ka BP). The reconstructed [NO3-]_BW ranged from 41.4 to 49.1 µmol kg⁻¹. The highest [NO3-]_BW of 49.1 µmol kg⁻¹ occurred during the Last Glacial Maximum (18.2 ka BP). The data points from the Early Holocene (11.6–10.8 ka BP) were the only Holocene data from this core providing the lowest [NO3-]_BW estimate of 42.1 µmol kg⁻¹ during the Early Holocene (10.8 ka BP) (Fig. 4b). A distinct difference in [NO3-]_BW between the glacial period (mean =46.1 µmol kg⁻¹) and the Early Holocene (42.7 µmol kg⁻¹) was observed with [NO3-]_BW found to be substantially higher during the glacial period (t-test, p=0.0067) (Fig. 4b). Accordingly, the [NO3-]_BW followed a decreasing pattern from the glacial period to the Early Holocene. The δ15N_bulk values varied between 6.4 ‰ and 13 ‰ with an average of 10.2 ‰ (Fig. 4b). The δ15N_bulk values from the Guaymas Basin were similar to the δ15N_bulk values (average 9.6 ‰) of Pride (1997) and Altabet et al. (1999). During the last glacial period, the δ15N_bulk values were low ranging from 8.5 ‰ to 9 ‰. At the onset of the deglaciation, the δ15N_bulk values increased by more than 2 ‰ with large-scale changes reaching a maximum of 13 ‰ during the Younger Dryas. Afterward, we observed a gradual decline in δ15N_bulk values throughout the Middle to Late Holocene (mean 10.7 ‰) and this pattern continued to the present.

This core MAZ-1E-04 covered the Last Glacial Maximum (20.5 ka BP) until the Early Holocene (10.47 ka BP). The [NO3-]_BW values range from 37.7 to 43.5 µmol kg⁻¹. We observed the highest [NO3-]_BW during the Younger Dryas. From the beginning to the end of the Last Glacial Maximum, [NO3-]_BW followed a decreasing trend (Fig. 4c). The [NO3-]_BW levels continued to steadily decrease until Heinrich Stadial 1 and consistently stayed low throughout this period. There was a strong change in [NO3-]_BW from the end of Heinrich Stadial 1 to the end of Younger Dryas (Fig. 4c). We observed a peak in [NO3-]_BW from the beginning of Bølling-Allerød, BA (14.29 ka BP) and it continued throughout the Younger Dryas (Fig. 4c). Afterwards, [NO3-]_BW declined during the Early Holocene. The δ15N_bulk values taken from Alcorn et al. (2025) followed an increasing trend from the glacial towards the deglacial period (Fig. 4c).

This core covered the Last Glacial Maximum (18 ka BP) until the Middle to Late Holocene (0.18 ka BP). The reconstructed [NO3-]_BW values range from 40.5 to 46.5 µmol kg⁻¹. The highest [NO3-]_BW occurred during the Last Glacial Maximum (Fig. 4d). The reconstructed [NO3-]_BW levels during the Last Glacial Maximum (mean =45.6 µmol kg⁻¹) were slightly higher than during the Middle to Late Holocene (mean =44.9 µmol kg⁻¹) (t-test, p=0.046). The δ15N_bulk values were relatively low ranging between 4 ‰ and 6 ‰ (Fig. 4d). During the Last Glacial Maximum, the δ15N_bulk values were low, varying between 4.4 ‰ and 4.6 ‰, close to the typical mean range of dissolved nitrate in the ocean (Sigman et al., 1997). Subsequently, the δ15N_bulk values increased from 16.7 ka BP (4.9 ‰), where we observed a decline in [NO3-]_BW to 8.9 ka BP (5.6 ‰). The highest δ15N_bulk values centered at ∼ 14 ka BP (5.9 ‰). From 8.9 ka BP onwards, a long-term decrease in δ15N_bulk (<4.4‰) was observed until the Latest Holocene, consistent with higher [NO3-]_BW levels during the Holocene (Fig. 4d). Despite higher [NO3-]_BW levels, our reconstruction doesn't show any strong variations during the Holocene.

Our data show that [NO3-]_BW levels gradually increased through time and reached modern concentrations during the Middle-Holocene (Fig. 4a). Most of the nutrients in the northwestern Pacific including the Sea of Okhotsk are supplied by the upwelling of the North Pacific Deep Water (NPDW) (Gorbarenko et al., 2014). The weakened Kuroshio current (Ujiié and Ujiié, 1999) and increased sea ice extent (Ternois et al., 2001) weakened the upwelling of NPDW during the Last Glacial Maximum (LGM). Subsequent studies (Gray et al., 2020; Rae et al., 2020) have shown that the expansion of the North Pacific Gyre also resulted in less upwelling of NPDW during the LGM.

During the LGM, the subpolar North Pacific was better ventilated at intermediate depths (Keigwin, 1998) and export productivity was reduced (Ternois et al., 2001; Narita et al., 2002; Seki et al., 2004). This is consistent with a strengthened meridional overturning circulation, with enhanced formation of intermediate waters and advection of nutrient-depleted subtropical waters to high latitudes (Rae et al., 2020). Furthermore, the North Pacific subpolar gyre extended ∼ 3° further south during the LGM (Gray et al., 2020), which shifted the westerly winds southward. This may have resulted in less upwelling of the NPDW during the LGM.

The prolonged ice cover with low biological productivity (Ternois et al., 2001; Narita et al., 2002; Seki et al., 2004; Rae et al., 2020) and well-oxygenated water masses (Keigwin, 1998) might have prevented the formation of an oxygen deficient zone (ODZ) in the Sea of Okhotsk (Bubenshchikova et al., 2015). This is supported by the absence of B. spissa, which are adapted to living in dysoxic conditions, in our records during the LGM.

Deglacial low [NO3-]_BW which correspond to higher δ15N_bulk values (Fig. 4a) could be due to enhanced primary productivity. It is important to note, however, that δ15N_bulk is influenced by diagenetic alteration and the incorporation of allochthonous nitrogen, which can obscure the local denitrification signal. Therefore, interpretations of δ15N_bulk trends should be made cautiously and ideally corroborated with complementary proxies, such as foraminiferal pore density. Increased nutrient supply from the Asian continental shelves and sea-ice retreat (Ternois et al., 2001) strengthened primary productivity. Indeed, the accumulation rate of total organic carbon was relatively higher during the Younger Dryas (Bubenshchikova et al., 2015) in our core (Fig. 4a). The increased oxygen demand and weakened ventilation of intermediate waters in the subarctic Pacific (Lembke-Jene et al., 2018) gradually intensified the ODZ. These poorly oxygenated conditions conceivably strengthened denitrification, resulting in low deglacial [NO3-]_BW levels. However, during the Middle to Late Holocene (MLH) a reorganization in atmospheric circulation favored enhanced formation of oxygenated North Pacific Intermediate Water (NPIW) (Wang et al., 2020). Thus, mid-depth ventilation was closely associated with atmospheric circulation in the Holocene and a weakened ODZ (Ohkushi et al., 2013; Bubenshchikova et al., 2015; Wang et al., 2020). These rising oxygen concentrations probably reduced denitrification (low δ15Nbulk) in the Sea of Okhotsk, resulting in higher [NO3-]_BW comparable to today's conditions (Fig. 4). The δ15N_bulk values show a maximum from 13 ka to 10 ka BP, which indicates increased water-column denitrification during that time. Nevertheless, the [NO3-]_BW increased during this time, which indicates a decoupling from denitrification in the oxygen minimum in the water column and the [NO3-]_BW. This could be related to the sea level rise during that time (Waelbroeck et al., 2008), which increased the vertical distance of the sediments (i.e., bottom water) at the sampling site from the center of denitrification.

The Gulf of California ODZ is influenced by both intermediate and deep-water properties, similar to that of the open Pacific Ocean. Thus, the ODZ intensity in the Guaymas Basin is largely dependent on the oxygen content and ventilation of inflowing NPIW from the Sea of Okhotsk (Pride et al., 1999) and the demand for oxygen at depth. During the glacial period, the dissolved oxygen concentrations were higher due to better-ventilated NPIW at intermediate depths of the Northeast Pacific (Keigwin and Jones, 1990; Ganeshram et al., 1995; Keigwin 1998; Duplessy et al., 1988; Herguera et al., 2010; Cartapanis et al., 2011). Modeling studies show that the Laurentide and Cordilleran ice sheets increased in size (Benson et al., 2003), lowering the temperature of North America (Romanova et al., 2006) during the glacial period. The cold sinking air over the ice sheet established a semi-permanent high-pressure cell (Kutzbach and Wright Jr., 1985; Romanova et al., 2006) causing a substantially weaker North Pacific High (Ganeshram and Pedersen, 1998) or the southward displacement of the Inter Tropical Convergence Zone (Cheshire and Thurow, 2013). This resulted in a weak California Current along the coast and reduced upwelling-favorable winds (Cartapanis et al., 2011) along the North American coastline and reduced primary productivity (Ganeshram and Pedersen, 1998; Hendy et al., 2004; Cartapanis et al., 2011; Chang et al., 2015) within the ETNP and the Gulf of California during the glacial period. The nitrogen isotope ratios in the Guaymas Basin can be affected by subsurface denitrification in the Gulf and in the ETNP (Pride et al 1999). The increase in dissolved oxygen during the glacial period might have reduced water column denitrification (low δ15Nbulk) thereby increasing the [NO3-]_BW (Fig. 4b).

Our study finds a declining trend in reconstructed [NO3-]_BW during the Early Holocene, slowly approaching modern concentrations. This coincides with a maximum in δ15N_bulk values, suggesting elevated denitrification. This agrees with previous studies in the ETNP (Kienast et al., 2002) and within the Gulf of California (Pride et al., 1999), which showed that high denitrification most likely was associated with warming temperatures that occurred during this period. Furthermore, the scarcity of benthic foraminifera after the Early Holocene in our study coincides with laminations of the sediment core (Keigwin and Jones, 1990) below 10.8 ka BP, where reconstructed [NO3-]_BW begins to decrease. It is possible that redox conditions were too hostile for benthic foraminifers in the time periods when laminated sediments formed. We acknowledge limitations in our Holocene reconstruction due to the low abundance of B. subadvena and the limited calibration dataset available for this species, which may introduce a systematic offset (Govindankutty Menon et al., 2023). Bolivina subadvena was used in this core due to the unavailability of B. spissa, and some values fall outside the existing calibration range. We also cannot rule out other factors influencing the proxy signal, such as microhabitat variability. Additional data and further proxy calibration are therefore essential to improve the robustness of Holocene bottom-water nitrate reconstructions.

Our study finds a steep rise in [NO3-]_BW between the Bølling-Allerød (BA) and the Younger Dryas (YD) (Fig. 4c). The transition period from the BA to the Holocene involved rapid oxygenation with increased oxygen levels at the onset of the YD (Jaccard and Galbraith, 2012; Ohkushi et al., 2013; Taylor et al., 2017). This has been linked to active ventilation by increased NPIW production at high latitudes in the North Pacific (Van Geen et al., 1996; Emmer and Thunell, 2000; Okazaki et al., 2010; Cartapanis et al., 2011; Chang et al., 2014). In addition, there was low primary productivity (Hendy et al., 2004; Pospelova et al., 2015), and a higher influx of freshwater (Broecker et al., 1985; Clark et al., 2002) during the YD. However, considering the deep location of the Mexican Margin core below the direct influence of intermediate water masses (Fig. 3), it is less likely to be reflected in the [NO3-]_BW. Bulk sediment δ15N records in the ETNP (Ganeshram et al., 1995; Pride et al., 1999; Emmer and Thunell, 2000; Kienast et al., 2002; Hendy et al., 2004) found a decrease in δ15N_bulk during the YD due to reduced denitrification. Furthermore, a foraminifera-bound nitrogen isotope (δ15N_FB) study (Studer et al., 2021) in the eastern tropical Pacific also found a decrease in δ15N_FB signatures during the Younger Dryas (Fig. 4c). In contrast to this, a bulk sediment δ15N record of MAZ-1E-04 (Alcorn et al., 2025) depicts an increase in water column denitrification during the Younger Dryas. Thus, reduced denitrification may not be the dominant factor that led to the elevated [NO3-]_BW during this time. Instead, the Mexican Margin may be more influenced by the NO3- variability from the Pacific Deep Water, PDW (see Fig. 3). Deep-sea reorganization and ventilation during the deglaciation may have influenced the [NO3-]_BW. At the onset of the deglaciation, deep Southern Ocean ventilation (reduced ¹⁴C ventilation ages) and atmospheric carbon dioxide (CO2) synchronously increased (Robinson et al., 2009; Burke and Robinson, 2012; Rae et al., 2018). This deglacial increase in ¹⁴C ventilation in the Pacific Ocean suggests that most of the increase in atmospheric CO₂ is derived from old carbon in the Southern and Pacific Oceans (Rafter et al., 2022). The increase in reconstructed [NO3-]_BW during the YD may thus reflect the release of sequestered nutrient- and carbon dioxide-rich waters during the deglaciation (Robinson et al., 2009; Rafter et al., 2022).

The relatively high [NO3-]_BW during the glacial period (Fig. 4c), before its decline in Heinrich Stadial 1, is likely indicative of reduced water-column denitrification (Ganeshram et al., 1995, 2000) due to reduced productivity (Ganeshram et al., 1995; Ganeshram and Pedersen, 1998) and low organic matter flux through the oxygen minimum zone (Ganeshram et al., 2000). In the ETNP, including the Mexican Margin, coastal upwelling is driven by trade winds generated by subtropical high-pressure centers. These high-pressure centers largely result from differential heating of the land and the ocean. As a result of glacial cooling on land, these high-pressure systems and the associated trade winds that drive the upwelling have likely been weakened (Ganeshram and Pedersen, 1998).

The core M77/2-59-01 is in a region that is sensitive to changes in subsurface denitrification in the ETSP (Robinson et al., 2007, 2009; Dubois et al., 2011, 2014). The elevated reconstructed [NO3-]_BW levels (Fig. 4d) during the glacial period suggest decreased water-column denitrification (Salvatteci et al., 2014; Erdem et al., 2020; Glock et al., 2022) and relatively low local productivity (Ganeshram et al., 2000; Robinson et al 2007, 2009; Martinez and Robinson et al., 2010; Salvatteci et al., 2016). Nutrient export to the deep Southern Ocean waters increased due to the sluggish Atlantic Meridional Overturning Circulation (Skinner et al., 2010), and increased atmospheric iron (Fe) deposition (Somes et al., 2017) during the glacial period. This reduced the transport of preformed NO3- to the tropics via the Subantarctic Mode Water (SAMW), limiting productivity. In fact, the total organic carbon (Fig. 4d) depicts low productivity during this period. Furthermore, the colder sea surface temperature (SST) and the accelerated formation of SAMW and Antarctic Intermediate Water masses (Russell and Dickson, 2003; Galbraith et al., 2004) and the stronger high-latitude winds in the Southern Hemisphere (Karstensen and Quadfasel, 2002) increased the ventilation rate (Meissner et al., 2005; Jaccard and Galbraith, 2012; Muratli et al., 2010) during the glacial period. The resulting increased oxygen concentrations (Robinson et al., 2005; Robinson et al., 2007) decreased the volume of ODZs, and nitrogen loss processes (lower δ15N_bulk values, Fig. 4d) during the glacial period. In addition, enhanced Fe deposition (Somes et al., 2017), and the glacial low sea level (Clark and Mix, 2002; Wallmann et al., 2016), may have influenced the nitrate inventory in the tropical and subtropical southern hemisphere.

A study by Glock et al. (2018) on core M77/2-52-2 from Peru applying the pore density of B. spissa also shows elevated [NO3-]_BW during the Last Glacial Maximum, a similar decline in [NO3-]_BW during the Heinrich Stadial 1 and thereafter a steady decrease in [NO3-]_BW throughout the Holocene.

The deglacial decline in [NO3-]_BW, especially during Heinrich Stadial 1 in this study (Fig. 4d), indicates a gradual increase in surface productivity and bottom-water deoxygenation. High export production strengthened the expansion of the ETSP ODZ during the deglaciation as compared to LGM and MLH (Salvatteci et al., 2016; Glock et al., 2018; Mollier-Vogel et al., 2019). This is consistent with the denitrification signal in the Eastern Equatorial Pacific through westward advection from the Southeast Pacific margins (Martinez and Robinson, 2010).

The shift towards generally higher reconstructed [NO3-]_BW from the Middle-Holocene, (Fig. 4d), implies a profound change in the climatic state of the Peruvian upwelling system and the associated ODZ during this time. From the deglaciation toward the Late Holocene, there was a general increase in productivity (Mollier-Vogel et al., 2019) as shown by organic carbon accumulation rates (Fig. 4d). This increase in organic matter input and/or preservation was likely related to an increase in upwelling-driven delivery of nutrients towards the surface. The gradual decrease in δ15N_bulk values and higher [NO3-]_BW was likely related to a relaxation in nutrient utilization with a nutrient supply exceeding the biological demand (Riechelson et al., 2024). Moreover, the core M77/2-59-01 was retrieved outside of the core ODZ and is under the strong influence of the oxygen and nutrient-rich Equatorial Under Current subsurface waters (Salvatteci et al., 2019; Mollier-Vogel et al., 2019). These waters might have ventilated the Northern Peruvian margin and deepened the oxycline at this site during the Middle-Holocene. Furthermore, enhanced zonal SST (Koutavas et al., 2006) and a northward shift of the ITCZ strengthened the Pacific Walker and Hadley circulation during the Middle-Holocene across the tropical Pacific (Koutavas et al., 2006; Mollier-Vogel et al., 2013; Salvatteci et al., 2019). These enhanced atmospheric circulations brought oxygen-rich waters to intermediate depths off Peru via the equatorial subsurface countercurrents (Koutavas et al., 2006; Mollier-Vogel et al., 2013; Salvatteci et al., 2019). Hence, increased ventilation of subsurface water masses reduced the strength of nitrogen loss processes and nutrient uptake during the MLH.

At present, the only quantitative reconstruction of bottom-water oxygen from these locations is the core M77/2-59-01 from the Gulf of Guayaquil reported by Erdem et al. (2020). Their record suggests a decline in bottom-water oxygen from the deglacial period to the Holocene. Future more detailed comparisons of the nitrate reconstructions with quantitative bottom water oxygen records at the same cores will further improve our understanding about variability in redox conditions and nitrogen cycling.

The [NO3-] during the present and past are compared to assess the resilience of our chosen study locations towards environmental and ecological impacts of climate change. The generally positive Δ[NO3-] that we found (Fig. 4b) in the Gulf of California (Guaymas Basin) and the Gulf of Guayaquil indicate that today the [NO3-] is lower than in the past. This suggests that today the nitrogen loss processes at these two core sites are stronger, most likely related to ocean warming and a decline in oxygen concentration of bottom waters. The Gulf of California core is within the heart of the oxygen-deficient zone, and thus changes in ODZ oxygenation or denitrification will be more evident in this core than in any other core studied. Under nitrogen limitation, negative feedbacks (e.g., anammox) result in a decline in productivity (Naafs et al., 2019; Wallmann et al., 2022), which will stabilize the oxygen concentration. In the case of the Gulf of California, sediments are enriched in reactive iron (Fe) (Scholz et al., 2019). The decreasing NO3- concentrations in the bottom water reduce the flux of NO3- into the surface sediment. This leads to the release of sedimentary Fe, which enhances nitrogen fixation in the Guaymas Basin (Scholz et al., 2019). Thus, increased denitrification might not act as negative feedback in the Gulf of California because it might be countered by increased nitrogen fixation (White et al., 2013).

In the case of the Gulf of Guayaquil (Fig. 4d), whether today's elevated denitrification could enhance N₂ fixation also depends on the availability of Fe (Pennington et al., 2006). The primary productivity of the Peruvian ODZ is Fe limited due to the reduction of particular Fe oxides in shelf and slope sediments (Scholz et al., 2014). Modeling studies show that primary productivity will be amplified in the Peruvian ODZ due to the release of Fe from shelf and slope sediments (Wallmann et al., 2022). This may induce deoxygenation and drive the expansion and intensification of Peruvian ODZ resulting in a positive feedback loop, like in the Gulf of California. This situation is indicated by lower [NO3-] today compared to the past ∼ 20 000 years.

The negative nitrate Δ[NO3-] in the Sea of Okhotsk and the Mexican Margin (Fig. 4a and c) indicates that modern [NO3-] levels are higher than in the reconstructed past. This suggests that modern nitrogen loss is decreased at these two core sites compared to the last deglaciation. The higher modern [NO3-] in the Sea of Okhotsk is likely associated with less primary productivity and more oxygen in the water column similar to the situation established in the MLH. The higher modern [NO3-]_BW in the case of the Mexican Margin could be associated with sea level rise. The ODZ in the Mexican Continental Margin might have shifted to shallower depths today with less/or no benthic denitrification in intermediate water depths at the core site, resulting in high [NO3-]_BW levels. During the glacial period, continental shelves were exposed due to sea-level lowstands (Clark and Mix, 2002; Kuhlmann et al., 2004; Wallmann et al., 2016), the main areas of primary productivity may have migrated offshore from the shallow shelf towards the continental slope relative to their Holocene positions. A similar situation occurred at the Benguela upwelling system during the LGM: TOC accumulation at the continental slope increased during the LGM in response to the seaward shift of centers of enhanced productivity (Mollenhauer et al., 2002). This offshore shift of the productivity centers and the most likely reduced remineralization rates, due to lower temperatures, indicate that the center of the ODZ at the Mexican Margin before sea level rise was possibly deeper than today. However, with the deglacial eustatic sea-level rise, the ODZ may have shifted to shallower depths. This shifted the main zone of denitrification further away from the seafloor, resulting in the increased modern [NO3-]_BW in comparison to the LGM.