Reduced Nutrient Load Lowers Ocean Carbon Emissions

Shared biogeochemistry in North Pacific and Southern Ocean

A reduced glacial supply of nutrients and carbon to Southern Ocean surface waters is recorded in proxy data over the last glacial cycle (Fig. 2). Reduced glacial biological productivity across the Antarctic Zone of the Southern Ocean is evidenced by proxies reflecting decreased export production (Fig. 2d). Such proxies include reduced fluxes and concentrations of biogenic barium and opal in marine sediments; various elemental ratios in sediments (e.g., reduced Ba/Fe and Ca/Fe reflecting reduced fluxes of biogenic barite and calcite to the seabed); reduced 231Pa/230Th (reflecting reduced particle flux to depth); and reduced calcite fluxes (reflecting reduced export production of calcite-producing organisms)19,30,31,32,33,34,35,36,37,38.

Alongside reduced productivity, the glacial Southern Ocean saw a more complete degree of nitrate utilization, with a greater proportion of the available nitrate pool being consumed by production. This is documented by elevated δ15N values in diatom-, coral- and foraminifera-bound organic matter39,40,41,42 (Fig. 2e). Biological productivity preferentially takes up nitrate with the lighter 14N isotope, enriching the residual seawater nitrate pool in heavy 15N. As nutrient utilization becomes more complete – consuming a greater and greater proportion of the available nitrate pool – the residual heavy signal becomes increasingly incorporated into organic matter. Heavier δ15N values under glacial conditions thus indicate enhanced nitrate utilization in Southern Ocean surface waters during glacial periods39,40,41,42. Critically, greater nitrate utilization (i.e., productivity consuming a greater proportion of the available nitrate pool) during glacial times can only be reconciled with reduced glacial productivity if the nitrate pool itself was smaller, signaling a reduced nutrient supply to the Southern Ocean surface in glacial times.

Remarkably similar glacial trends are also recorded in the far-flung North Pacific. Reduced productivity and export production are recorded throughout the subpolar gyre by reduced barium and opal fluxes43,44,45 (Fig. 2a), as is enhanced nitrate consumption by heavier δ15N values of foram- and diatom-bound organic matter45,46,47,48 (Fig. 2b). This shared “Polar Twins” pattern26 in Southern Ocean and North Pacific proxies has prompted considerable discussion of what process could simultaneously drive the same proxy patterns in two such far-flung regions. Previous explanations have focused on enhanced isolation of surface waters from nutrient-rich deep waters below, invoking the same physical processes (reduced upwelling or enhanced stratification) operating in both regions simultaneously during glacial times26,27. Little consideration, by contrast, has been given to linkages connecting these two regions and the potential importance of a common reduced subsurface nutrient and carbon content. In the North Pacific, reduced nutrient and carbon content in better-ventilated sub-surface waters has recently been demonstrated as a key means of reducing nutrient and carbon delivery to the North Pacific surface during the Last Glacial Maximum (LGM)28. Motivated by contemporary linkages between the North Pacific and Southern Ocean12,13, we here explore the potential of North Pacific ventilation to simultaneously influence carbon and nutrient supply to the Southern Ocean surface.

A reduced sub-surface North Pacific carbon/nutrient supply

Proxy records show North Pacific surface waters exhibiting lower nutrients, higher salinity, and warmer temperatures in the LGM than today, which modeling results show to be consistent with better ventilation of and reduced nutrient content in sub-surface waters28. While reduced upwelling or mixing could also explain reduced surface nutrients, wind-driven upwelling (a dominant mode of nutrient delivery in the modern) was in fact likely increased in the North Pacific during glacial times, with Paleoclimate Modeling Intercomparison Project (PMIP) models consistently indicating substantially (~ 60%) greater wind stress curl over the North Pacific subpolar gyre under LGM conditions than in the pre-industrial49,50. Similarly, evidence of enhanced regional intermediate water formation suggests increased convective mixing at this time25,28, while increases in benthic carbon isotopes (Fig. 2c) reflect enhanced ventilation and influence by heavy-δ13C surface waters51,52. Communication between the sub-surface and surface in the North Pacific, therefore, appears to have been increased under glacial conditions, thus requiring a reduced nutrient content of these sub-surface waters to explain reduced nutrient supply at the surface.

It is a common feature of Earth System Model simulations that such a reduction in sub-surface nutrient content can be achieved by an active meridional overturning circulation operating in the Pacific, as demonstrated in cGENIE28 as well as higher-resolution models such as LOVECLIM and UVic53,54. Across these models, enhanced North Pacific Intermediate Water (NPIW) formation under glacial-like conditions consistently drives an equilibrium state in which well-ventilated sub-surface waters penetrate down to ~ 2000 m depth (hereafter referred to as “mid-depth” waters) (Supplementary Fig. 2 and Supplementary Table 1). Thus, where these mid-depths previously hosted poorly-ventilated carbon-rich waters isolated from the surface, the expanded NPIW cell has replaced these waters with relatively carbon- and nutrient-poor surface waters drawn from the low latitudes (Fig. 3 and Supplementary Fig. 3).

The same pattern of mid-depth carbon/nutrient reduction emerges in response to ventilation in these models, independent of varying configurations, boundary conditions, resolutions, and styles of forcing. Furthermore, mid-depth reductions in carbon and nutrients appear to be a robust response to varying degrees of enhanced ventilation of the North Pacific. The LOVECLIM-LGM simulation54 presented in Fig. 3d, for example, simulates a weaker Pacific overturning than the other simulations (Supplementary Table 1), providing context of that study’s estimated overturning at the LGM (~ 4 Sv). This simulation still sees strong reductions in mid-depth carbon and nutrients, and furthermore, may underestimate the true magnitude of glacial Pacific overturning, which remains poorly constrained. For example, Rae et al. (2020) found a Pacific overturning of ~ 8 Sv yielding the best model-data fit in c-GENIE simulations28, and as such, the LOVECLIM-LGM output should be seen as broadly indicative rather than a definitive representation of LGM-like overturning. More importantly, the emergence of a consistent pattern across a range of strong and weak ventilation strengths (see also Chikamoto et al., 201255) supports North Pacific ventilation as an effective means of reducing sub-surface carbon and nutrient content. Past studies have suggested other mechanisms by which such mid-depth Pacific anomalies can arise, such as changes in southern-sourced water mass characteristics impacting nutrient supply to the basin (Chikamoto et al., 2012). Such dynamics are not expected to be at play here, with the mid-depth anomalies clearly developing from the north due to the onset of northern ventilation (Supplementary Fig. 4).

Reducing the sub-surface carbon and nutrient pool also reduces local surface waters’ carbon and nutrient content (Fig. 3), as it is these sub-surface waters which supply carbon and nutrients to the surface above28. The onset of greater convective mixing in the North Pacific may seem at odds with reduced surface nutrients and carbon, but it is important to distinguish between the transient and equilibrated responses to such ventilation. The initial onset of overturning is expected to generate a short-lived transient peak in surface nutrients and carbon as convection initially taps into nutrient- and carbon-rich sub-surface waters below56. Over time, continued ventilation depletes this sub-surface reservoir, eventually leading to persistently low surface concentrations28. Thus, the long-term equilibrated response to ventilation of the North Pacific is one of persistently low carbon and nutrient concentrations in surface waters, even in the face of sustained convective mixing. The difference between the transient and equilibrated response is also well illustrated in ventilation experiments performed in cGENIE, which show the transient peak in surface nutrients subsiding within the first several hundred years of simulation (see Supplementary Fig. S9 in ref. 28). The same response is observed in the UVic simulations analyzed in this study, where North Pacific ventilation is triggered by North Atlantic freshwater perturbations under an otherwise glacial-like state. Despite the initial mixing up of deep carbon, ventilation reduces outgassing rates in the North Pacific within the first few centuries of the simulation, with strong reductions having developed by the end of 1000 years (Supplementary Fig. 5).

More active NPIW formation and ventilation thus provides a means of reducing surface carbon and nutrients in the glacial North Pacific by reducing the carbon and nutrient load of the subsurface waters that supply them. We propose that these North Pacific changes could also account for the reduced carbon and nutrient load in the glacial Southern Ocean surface. With better ventilation by an expanded “glacial NPIW” (GNPIW) cell, the previously carbon-rich Pacific mid-depths feeding the Southern Ocean would be replaced by carbon-/nutrient-poor waters from the surface. If this well-ventilated low-carbon/nutrient signal in the sub-surface glacial North Pacific reached the Southern Ocean via isopycnal diffusion as it does today23, this would provide an alternative means of reducing the glacial Southern Ocean surface carbon and nutrient load (that could act independently of, or in conjunction with, physical changes like reduced upwelling or mixing), while simultaneously explaining the shared “Polar Twins” proxy pattern between the North Pacific and Southern Ocean. Indeed, this low-carbon low-nutrient signal is seen to propagate southward from the North Pacific in all models, notably along isopycnals outcropping in the Southern Ocean (red and yellow contours in Fig. 3 and Supplementary Fig. 3). This is a consistent pattern persisting across a range of Earth System Models spanning differing configurations, boundary conditions, and ventilation strengths, and thus is an effective demonstration of the ability of a well-ventilated North Pacific to reduce the carbon and nutrient content of waters feeding the Southern Ocean.

Reduced carbon and nutrient delivery to the Southern Ocean

We now quantify the impact of a well-ventilated North Pacific on Southern Ocean carbon and nutrient content in the intermediate-complexity University of Victoria Earth System Climate Model (UVic ESCM) v2.957. While mid-depth carbon/nutrient reduction is a robust response to North Pacific ventilation across the models, independent of model choice or even ventilation strength (Fig. 3, see also Chikamoto et al., 2012)55, we chose this model to analyze in detail as it has the highest resolution (1.8° latitude x 3° longitude) compared to the other models presented in Fig. 3 (LOVECLIM/LOVECLIM-LGM: 3° x 3°; c-GENIE: 5° latitude x 10° longitude). Perhaps linked to this, UVic also most clearly showcases the signal of the dynamics we set out to demonstrate, with the low-carbon, low-nutrient anomaly propagating clearly from the North Pacific and being upwelled in the Southern Ocean. Thus, rather than a comprehensive and definitive representation of LGM overturning and biogeochemistry, we present these results as a test and proof of concept of an unexplored mechanism, one which we hope motivates future study in more targeted sets of model experiments.

The UVic simulation analyzed here was first presented by Menviel et al. (2014), who performed North Atlantic meltwater experiments in the UVic v2.9 and LOVECLIM v1.1 ESMs to investigate marine carbon cycle responses to Atlantic meridional overturning circulation (AMOC) shutdown53. In the North Atlantic meltwater simulations, 0.1 Sv of freshwater was added over the North Atlantic (50–65 °N and 55-10 °W) for 1000 years. After the 1000-year integration, the model has effectively reached equilibrium with respect to its biogeochemistry, with minimal residual trends in its atmosphere (< 0.3ppm/100 yr) and land and ocean (< 1PgC/100 yr) carbon inventories. Conditions after 1000 years are therefore taken to be highly representative of a fully equilibrated state. In both UVic and LOVECLIM, this induced stronger North Pacific Intermediate Water (NPIW) formation (arising from a weaker North Pacific halocline resulting from oceanic and atmospheric teleconnections) and thus stronger ventilation of North Pacific waters. In UVic, this results in better ventilation down to about 2000 m depth and an expanded “GNPIW” (“glacial NPIW”) overturning cell of ~ 14.8 Sv (Supplementary Fig. 2 and Supplementary Table 1)53. We therefore take advantage of this ventilated UVic simulation to study the impacts of a well-ventilated North Pacific on basin-wide biogeochemistry. We acknowledge that North Atlantic-North Pacific teleconnections are a topical area of research, with the strength of North Pacific ventilation and overturning resulting from North Atlantic freshwater perturbations having been demonstrated to differ across models (e.g., Saenko et al., 2004; Chikamoto et al., 2012; Baker et al., 2025)55,58,59. Enhanced North Pacific ventilation resulting from North Atlantic freshwater perturbations, however, remains a robust feature across models (including the simulations of Chikamoto et al., 2012, and Baker et al., 2025), and the details of its establishment do not impinge upon our results. Instead, we focus on the downstream impacts on basin-wide and Southern Ocean biogeochemistry of a ventilated North Pacific, for which there is glacial proxy evidence51,52. Furthermore, strong reductions in mid-depth carbon and nutrients appear to be a robust feature regardless of the magnitude of increase in ventilation strength (Chikamoto et al., 2012)55, further supporting the ideas motivating this study.

We furthermore do not take the meltwater simulation (given its AMOC shutdown) as a precise analog of LGM conditions, but rather we capitalize on its representation of North Pacific ventilation and use it to assess the impacts of a well-ventilated North Pacific on Southern Ocean biogeochemistry. We do note, however, that similar results to those presented here are seen in simulations with LGM-like changes in overturning54, with less extreme AMOC reduction and NPIW formation (Fig. 3d, Supplementary Fig. 2 and Supplementary Table 1, and Rae et al. (2020)28).

Here, we compare Southern Ocean carbon and nutrient content in two simulations of UVic, both run under identical LGM-like boundary conditions (Supplementary Table 2). One simulation features strong North Pacific ventilation driven by North Atlantic meltwater addition and associated enhanced NPIW formation (the perturbed simulation, hereafter referred to as “UVic-NP”), while the other serves as a control simulation, lacking meltwater addition and associated North Pacific ventilation (“UVic-ctrl”) (compare overturning strength and radiocarbon age anomaly between UVic-NP and UVic-ctrl as an indicator of this ventilation, Supplementary Figs. 2 and 6, Supplementary Table 1). In these simulations we quantify carbon content by potential pCO2 (PCO2), which being a function of both alkalinity and dissolved inorganic carbon (DIC) is an effective measure of a water parcel’s outgassing potential12. Nutrient content is traced through phosphate concentrations ([PO43-]). We also quantify differences in Southern Ocean outgassing rates between the two simulations.

In the UVic-NP simulation, a stronger GNPIW overturning cell ventilates and reduces the carbon and nutrient content of sub-surface North Pacific waters – an anomaly which extends southwards through the basin along mid-depths (Fig. 3a and Supplementary Fig. 3a). We see strong reductions in PCO2 in the North Pacific from the near surface down to ~ 3000 m depth, with a mean PCO2 anomaly of − 199 μatm in the sub-surface North Pacific (30–60 °N, 140–240 °E, 500–2000 m) (Fig. 4a). Phosphate concentration is similarly reduced (Supplementary Fig. 3a and Fig. 4a), with a mean anomaly of − 1.3 μmol kg−1 in the sub-surface North Pacific. The low-PCO2 signal extends along isopycnals through mid-depths (~ 1000–2000 m) to southwards of 30 °S, exhibiting a concentrated core (− 78 μatm on average) in the western sector of the basin (140–200–°E, 1000–2500 m), while more diffuse anomalies of − 47 μatm on average pervade the basin (140–280 °E, 1000–2500 m) (Fig. 4a, c). After ~ 50°S, this low-PCO2 signal is translated upwards along isopycnal surfaces (σ2 = 36.0-36.8 kg m−3), outcropping in the Southern Ocean surface (red and yellow contours in Fig. 3a). Again, similar patterns are seen in phosphate concentrations (Fig. 4d), where at 30 °S phosphate is reduced by − 0.3 μmol kg−1 on average across the basin and by − 0.5 μmol kg−1 in the western sector of the basin. This negative phosphate anomaly similarly reaches the Southern Ocean surface along outcropping isopycnals (Supplementary Fig. 3a).

Low PCO2 drives reduced Southern Ocean outgassing

This low-PCO2 low-nutrient signal propagated from the North Pacific translates into quantifiable negative PCO2 and phosphate anomalies throughout the Southern Ocean surface, spanning the Subantarctic and Antarctic Zones (Fig. 5a, b). We find the strongest anomalies in the Pacific-sector (150–280 °E) Subantarctic Zone (SAZ, ~ 40–60 °S), where mid-depth Pacific isopycnals bounding this low-PCO2 water (σ2 = 36.0-36.8 kg m−3, red contours in Fig. 5) outcrop. We observe a mean PCO2 anomaly of approximately − 8 μatm across the Southern Ocean (all longitudes south of 40 °S) and of approximately − 12 μatm across the Indian-Pacific sector (25–280 °E), with maximum anomalies surpassing approximately − 25 μatm (Fig. 5b). Phosphate is reduced by − 0.13 μmol kg−1 on average across the Southern Ocean and by − 0.17 μmol kg−1 in the Indian-Pacific sector, with maximum anomalies surpassing − 0.44 μmol kg−1 (Fig. 5a).

We also note that ventilation could likely induce even greater carbon anomalies than documented here if applied to a system transitioning out of interglacial conditions. Having been run under glacial-like conditions, UVic-ctrl exhibits a low-carbon bias in the North Pacific (PCO2 ~ = 350–450 μatm) relative to a pre-industrial or interglacial-like state (> 1000 μatm24, Fig. 1). Starting from a more carbon-rich baseline in UVic-ctrl could have enabled an even greater reduction in carbon and nutrients at mid-depths of the North Pacific in the perturbed simulation than we observe here, with downstream effects on the Southern Ocean. Thus, in terms of the magnitude of carbon drawdown associated with ventilation change, our results may represent a conservative estimate.

Even so, the negative PCO2 anomalies at the surface of the Southern Ocean reduce CO2 outgassing, chiefly in the Subantarctic Zone (~ 40–60 °S) (Fig. 5c). This translates into a mean Southern Ocean anomaly (in all areas south of 40 °S) of − 0.18 mol m−2 yr−1 (representing a ~ 50% reduction relative to UVic-ctrl’s mean outgassing rate). The Indian-Pacific sector (25–280 °E, south of 40 °S) sees a mean anomaly of − 0.26 mol m−2 yr−1 or ~ 79% relative to the control. Furthermore, reduced outgassing is demonstrably the result of the altered biogeochemistry of the upwelling water. Changes in alkalinity and DIC dominate the outgassing anomaly signal (Fig. 5d), whereas temperature and salinity changes play a relatively minor role (see Supplementary Fig. 7). The density surfaces bounding the core low-PCO2 water mass from the North Pacific (σ2 = 36.0–36.8 kg m−3, red contours in Fig. 5) outcrop at the Southern Ocean surface and delineate the negative air-sea CO2 flux anomaly, further supporting our proposed North Pacific-based mechanism.