Long‐Term Variability in Pliocene North Pacific Ocean Export Production and Its Implications for Ocean Circulation in a Warmer World

Unlike in the high‐latitude North Atlantic, no deep water is formed in the modern subarctic North Pacific. It has previously been suggested that during climate states different from today, this dichotomy did not endure, and the formation of North Pacific Deepwater (NPDW) occurred in the subarctic North Pacific, which supported an active Pacific meridional overturning circulation (PMOC). Here we provide new records of productivity and sedimentary redox conditions from the central subarctic North Pacific spanning the late Miocene to early Pleistocene. These reconstructions indicate greater‐than‐modern and temporally varying North Pacific export production across the interval of ∼2.7–6 Ma. Our time series, combined with previously published data sets and model output for Pliocene North Pacific Ocean dynamics, support the presence of an active PMOC during the Pliocene, and suggest that the characteristics of NPDW formation varied during this warmer interval of Earth's history. This finding of elevated export production at a time of deep water formation presents a conundrum when considering Quaternary North Pacific Ocean dynamics, where subarctic North Pacific productivity declines during intervals when enhanced overturning is posited to occur. We evaluate our data considering the caveats of both (i.e., Pliocene and Quaternary North Pacific circulation) hypotheses, as well as additional mechanisms unrelated to ocean circulation. Because the Pliocene is a possible analogue for near‐future climate, our results and analyses have important ramifications for our understanding of regional and global climate in the coming decades as the planet continues to warm.

3 of 26 Sigman et al., 2004;Studer et al., 2012). After the development of major Northern Hemisphere ice sheets at ∼2.7 Ma, the formation of a modern-like halocline in the subarctic North Pacific reduced the vertical mixing of nutrients, reducing export production overall and enhancing nutrient utilization (G. H. Haug et al., 1999;Sigman et al., 2004).  further suggest that it is the formation of NPDW in the subarctic North Pacific prior to the iNHG (and no formation thereafter) that enhanced export production in the warmer world of the Pliocene.
The proposed changes in circulation and/or the flux of organic matter prior to (and coincident with) ∼2.7 Ma could also drive shifts in bottom water oxygen concentrations, as both have been shown to impact sediment redox conditions (Algeo & Liu, 2020;Emerson et al., 1985;Nameroff et al., 2004;Tribovillard et al., 2006). For example, enhanced subarctic North Pacific bottom water oxygenation relative to today could either reflect changing conditions related to the water mass bathing the region or a reduced flux of organic carbon. Several lines of evidence support shifting redox conditions in the North Pacific during the Quaternary and Pliocene, although the proposed driving mechanisms vary G. R. Dickens & Owen, 1996;Rae et al., 2020;Rafter et al., 2022;Zhai et al., 2021).
Considering that export productivity and sedimentary redox conditions have previously been linked to changes in physical circulation in the subarctic North Pacific, long-term, relatively high-resolution records of these proxies offer the potential to improve our current understanding of NPIW/NPDW formation (or a lack thereof) across the Pliocene, and its relationship to productivity. To that end, here we provide new constant flux proxy (CFP)-derived reconstructions of export productivity as well as redox-sensitive element data from the central subarctic North Pacific spanning the period of ∼2.5-6 Ma ( Figure 1). This study builds on several decades of previous work G. Dickens et al., 1995;G. R. Dickens & Barron, 1997; G. R. Dickens & Owen, 1996;Snoeckx et al., 1995; attempting to derive a consistent and reliable record of Pliocene central subarctic North Pacific productivity. Additionally, in conjunction with other previous geochemical and modeling efforts Ford et al., 2022;G. Haug et al., 1995;G. H. Haug et al., 1999;Rae et al., 2020;Sigman et al., 2004;Thomas et al., 2021), we provide an original evaluation of the connection between productivity and NPDW formation during this warmer-than-present interval of Earth's past, setting up new hypotheses to be tested. Finally, we assess other potential mechanisms that could explain our signal of enhanced and temporally varying export productivity in the Pliocene. Together, our work highlights the likely complex nature of the relationship between North Pacific Ocean circulation, productivity, and global climate.

Core Location, Age Model, and Sampling
All new data produced in this study are from Ocean Drilling Program (ODP) sites 885/886 (44.7°N, 168.2°E, 5,709/5,714 m water depth), which were cored on the southern margin of the modern central subarctic North  (Garcia et al., 2019b). New proxy data presented here is from ODP 885/886 (light brown diamond). Other marine sediment core sites highlighted in the text are indicated by black circles.

of 26
Pacific (D. Rea et al., 1993). A magnetostratigraphic-based age model was developed for a composite depth profile for ODP 885/886 (G. Dickens et al., 1995). We have updated the ages of identified reversal boundaries based on the 2020 geomagnetic polarity time scale (Figure S1a in Supporting Information S1) (Ogg, 2020). Linear sedimentation rates based on the updated ages range from ∼0.48 to 0.89 cm ky −1 ( Figure S1b and Table  S1 in Supporting Information S1). Samples from ODP 885/886 used in this study are from individual cores 885A, 886B, and 886C, with specific sampling intervals recommended from shipboard data (G. Dickens et al., 1995).

Measurement of Helium Isotopes
Helium isotope data for 41 of the 57 samples are from Abell et al. (2021) (Figure S2 in Supporting Information S1). The 16 helium isotope measurements new to this study were produced following the methodology of Winckler et al. (2005) and fill in the lower-resolution section of Abell et al. (2021), specifically in the interval from ∼3.8 to 6.1 Ma. Briefly, samples were weighed into aluminum foil cups, and helium was extracted via heating to ∼1200-1300°C followed by purification through several charcoal traps and a getter. Subsequently, isotopes were measured on a MAP215-50 noble gas mass spectrometer at Lamont-Doherty Earth Observatory (LDEO) and standardized with the "Murdering Mudpot" Yellowstone helium standard (Clarke et al., 1976;Welhan, 1981). Procedural blank corrections were typically <∼3% for both 3 He and 4 He.

Determination of an Extraterrestrial 3 He Component
In marine sediment archives, measured helium isotopes reflect the combination of a terrigenous (He terr ) and extraterrestrial (He ET ) signal. Because it is the 3 He found in interplanetary dust particles (IDPs) delivered to Earth's surface, referred to as extraterrestrial 3 He ( 3 He ET ), that is utilized as a proxy for vertical sediment rain rates (details behind this proxy are further described below) (McGee & Mukhopadhyay, 2013), this component must be deconvolved from the 3 He derived from terrigenous material. To accomplish this, we use the mixing equation: and previously determined 3 He/ 4 He ratio values for the extraterrestrial ( 3 He/ 4 He) ET and terrestrial ( 3 He/ 4 He) terr endmembers (Marcantonio et al., 1995;McGee & Mukhopadhyay, 2013). Following several previous studies (Abell et al., 2021;Middleton et al., 2016Middleton et al., , 2018Winckler et al., 2005), values of 2.4 × 10 −4 and 2.0 × 10 −8 are selected for ( 3 He/ 4 He) ET and ( 3 He/ 4 He) terr , respectively (K. Farley et al., 1997;K. Farley & Patterson, 1995;Nier & Schlutter, 1992;Ozima & Podosek, 1983). Because there has been a plethora of other suggested ( 3 He/ 4 He) terr endmember ratios, including ∼1.0 × 10 −8 (Ozima & Podosek, 1983), ∼3.0 × 10 −8 (K. A. Farley, 2001;Mamyrin & Tolstikhin, 1984), and up to ∼4 × 10 −7 for East Asian dust specifically (McGee et al., 2016), we evaluate the sensitivity of our determined 3 He ET concentrations to these different ratios (Data Set S1, https://doi.org/10.6084/ m9.figshare.21668156). We find that for the lowest three ( 3 He/ 4 He) terr values, >98% of the total measured 3 He is 3 He ET for all samples. Even when a value of ∼4 × 10 −7 is applied, 3 He ET averages ∼92% of the total 3 He, with a minimum of ∼84%. We thus conclude our 3 He ET -derived fluxes are relatively insensitive to the choice of endmember, and as such are robust with respect to the detrital contributor.

Uncertainty Associated With the CFP 3 He ET
Previous work has shown that the reproducibility of determined 3 He ET from sample replicates can differ widely, and is predominantly a function of the statistical representation of the IDPs in a given sediment sample. This can be expressed by the area-time product of the sample aliquot (K. Farley et al., 1997;Patterson & Farley, 1998), defined as the mass of the measured aliquot divided by the sediment mass accumulation rate (MAR). Higher area-time products, for example, represented by larger sample sizes of more slowly accumulating sediments, lead to higher reproducibility between replicates. The overall effect is that this statistical uncertainty associated with sampling far exceeds the typical analytical 3 He ET uncertainties of ∼2%-3%. Quantifying this uncertainty is essential for correctly applying the 3 He ET CFP, and in turn producing accurate estimates of paleo-sediment fluxes.
We follow the same approach of Abell et al. (2021) and others (Middleton et al., 2016Patterson & Farley, 1998) and derive a representative uncertainty using the fractional difference of ODP 885/886 3 He ET replicates (new and previously published). The determined uncertainty, based on the median fractional difference, is 10.1029/2022AV000853 5 of 26 ∼35% (1σ), which is then combined with the sample-specific average analytical uncertainty. For all samples with replicates, we reduce the statistical uncertainty by 1/sqrt(n) prior to propagation.

3 He ET as a CFP
CFPs, such as 230 Th xs0 and 3 He ET , have seen increased usage over the last two decades for the determination of vertical sediment fluxes. This is because they offer two substantial advantages over the typical age model-derived sediment MARs: (a) measurements are performed on individual samples, avoiding the issue of interpolation of constant sedimentation rates between age-depth tie points, and (b) the derived fluxes "see-through" sediment redistribution processes, such as winnowing and focusing, providing vertical rain rates of sediment constituents (Costa et al., 2020;McGee & Mukhopadhyay, 2013).
Because the half-life of 230 Th precludes its utility prior to ∼500 ka (Cheng et al., 2013;Costa et al., 2020;Francois et al., 2004), here we use the CFP 3 He ET to estimate vertical fluxes of our various proxies (ɸ X ) (Marcantonio et al., 1995;McGee & Mukhopadhyay, 2013). This is accomplished by dividing the assumed constant flux of 3 He ET to Earth (F 3HeET ) by the concentration of 3 He ET in a given sample ([ 3 He ET ]) and multiplying by the concentration of the sediment constituent of interest (X): A critical assumption for the use of a CFP is that the flux of said proxy to the sediment archive in question is known and constant over the study interval. Because a direct calibration (independent of an assumption regarding a lack of sediment redistribution (K. A. Farley, 1995;McGee & Mukhopadhyay, 2013)) is missing for F 3HeET in the Pliocene, we use the late Quaternary, 230 Th xs0 -constrained value of (8.0 ± 3.0 1σ) × 10 −13 cm 3 cm −2 ky −1 (McGee & Mukhopadhyay, 2013). We note that even if F 3HeET was different during the last ∼5 Ma, our interpretations are based on relative changes in fluxes, and as such are not dependent on the absolute value of F 3HeET as long as it remains constant over the interval studied. Considering this, the uncertainty associated with F 3HeET is not incorporated into figures. However, for completeness, errors propagated with the F 3HeET uncertainty are provided in Data Set S1 (https://doi.org/10.6084/m9.figshare.21668156). For samples that lack a direct measurement of 3 He ET , we interpolate between available 3 He ET -derived fluxes and apply the average 3 He analytical uncertainty and maximum statistical uncertainty.
With knowledge of the vertical fluxes from 3 He ET and the bulk sediment fluxes based on the age model and dry bulk densities, we calculate focusing factors (Ψ) to provide an estimate of the degree of sediment focusing/ winnowing following Costa et al. (2020): where ρ avg is the average dry bulk density, [ 3 He ET ] avg is the average 3 He ET concentration, Δz is the sediment thickness, and Δt is the elapsed time, all within a given sediment horizon. To be conservative, we use the boundaries defined by the age model of Dickens et al. (1995) updated based on Ogg (2020). For consistency with our He isotope measurements, we use the dry bulk densities interpolated to our sample intervals. Ψ values less than one indicate periods of winnowing, while values greater than one imply focusing.

Major and Trace Elements
Major and trace element concentrations new to this study and those from Abell et al. (2021) were determined through the adapted acid digestion method of Fleisher & Anderson, 2003. Approximately 90-110 mg of bulk sediment was weighed out into Teflon beakers and several ml of milli-Q (MQ) water was added. Samples were digested in a mixture of concentrated HNO 3 , HClO 4 , and HF at temperatures of ∼150-250°C. The dissolved material was then taken up in 0.5 M HNO 3 and diluted to ∼2000x for analysis. Inductively coupled plasma optical emissions spectrometry (ICP-OES) was used for the determination of Al, Fe, Ti, and Mn, while Ba, Sc, La, and Th were analyzed via inductively coupled plasma mass spectrometry (ICP-MS). Concentrations for elements measured through ICP-OES were calculated using external standard solutions, while we utilized a standard solution added to representative samples for elements measured using ICP-MS. All element-specific analytical uncertainties are based on the reproducibility of an internal standard (n = 9), referred to as the VOICE Internal MegaStandard (VIMS), which is comprised of homogenized sediments from the Juan de Fuca Ridge.

Determination of Opal Concentrations
We followed the approach of Snoeckx et al. (1995) and Rea et al. (1998) and calculated opal fractions via subtraction of a measured lithogenic component, under the assumption that the sediments at the site were predominantly a two-endmember system (i.e., biogenic silica and lithogenic material). However, unlike these previous studies, which relied on a sequential leaching method to derive a lithogenic component, we can utilize the concentrations of several elements that are primarily lithogenic in origin (Al, Fe, Ti, and Th). Specifically, we first calculated lithogenic components (Lith %) for each element through the equation: where [X meas ] is the measured concentration of the element and [X Lith ] is an assumed lithogenic endmember concentration. For the endmembers, we used 80,400 ppm, 40,786 ppm, 3,631 ppm, and 14 ppm, for Al, Fe, Ti, and Th, respectively. These are based on either an average upper continental crustal (UCC) value (Al; Taylor & McLennan, 1995), the maximum concentration in our data set (if higher than the UCC value) (Fe and Ti; Abell et al., 2021), or the average in fine-grained dust (Th; McGee et al., 2016). The calculated individual lithogenic components were averaged to produce a final value (Lith avg %). We then determined the opal component (Opal %) by: To calculate the uncertainties for opal, we took the standard deviation (1σ) of the four individual lithogenic fractions. The maximum difference in Opal % between the four elements is 32.9%, with an average of 8.38%.

Determination of Excess Ba
Excess barium (Ba xs ), that is, barium corrected for a lithogenic component, can be utilized as a proxy of past export productivity (Calvert & Pedersen, 2007;Dymond et al., 1992;Fagel et al., 2004;Francois et al., 1995;Winckler et al., 2005). Here we calculate the lithogenic contribution of Ba using the same four lithogenic proxies applied for the determination of opal. Specifically, we first multiply [X meas ] by an assumed ratio of lithogenic Ba to each of the four defined lithogenic elements (Ba/X) lith to determine the Ba derived from lithogenic material. As with opal, we utilize either the average UCC value (Ba and Al), the maximum concentration of the lithogenic element in our data set (Fe and Ti), or the average in fine-grained dust (Th) for the lithogenic endmember concentrations. We then subtracted the lithogenic Ba from the measured total Ba concentration ([Ba meas ]) to derive a Ba xs concentration:

Determination of Excess Mn
Manganese found in open-ocean marine sediments can be derived from lithogenic material, hydrothermal input, and changing redox conditions (Calvert & Pedersen, 2007;Tribovillard et al., 2006). Under typical oxic conditions, dissolved Mn tends to precipitate, typically in the form of Mn oxides, as it is oxidized from Mn(II) to Mn(IV). However, if reducing conditions are encountered, the solubility of Mn increases as it is reduced, and can reprecipitate in shallower sediments where oxygen is present or mobilize out of the sediment into the water column (Burdige & Gieskes, 1983;Calvert & Pedersen, 2007;Froelich et al., 1979). As such, higher values of Mn xs in marine sediments can indicate oxic conditions, while lower values would suggest an anoxic setting, although we note that there are several instances where this proxy does not follow this simplistic process (Tribovillard et al., 2006). While ODP 885/886 is far from substantial hydrothermal settings, the site did receive dust in the Pliocene (Abell et al., 2021;D. K. Rea et al., 1998;Snoeckx et al., 1995). To correct for the lithogenic component of manganese, and produce an excess manganese (Mn xs ) component useable for interpretations of paleo-redox settings, we take the same approach as for Ba xs and utilize Al, Fe, Ti, and Th: Again, the average of all four calculated Mn xs concentrations is taken as the final value, and the associated uncertainty is the standard deviation (1σ)

Mio-Plio-Pleistocene Sediment and Opal Accumulation at ODP 885/886
Over the period spanning ∼2.5-6 Ma, opal concentrations at ODP 885/886 range from ∼8% to 91%. The highest values are found between the late Miocene and mid-Pliocene (∼3.6-5.4 Ma), and generally decrease from ∼3.6 to 2.7 Ma (Figure 2b). The lowest opal concentration values are reached after the iNHG at ∼2.7 Ma. 3 He ET -derived sediment fluxes reach their maximum in the early Pliocene, with values over ∼1 g cm −2 ky −1 found between ∼4.6-4.9 Ma (Figure 2c and Figure S2 in Supporting Information S1). Sediment mass accumulation rates (MARs) are lower before and after this interval, reaching ∼0.2 g cm −2 ky −1 at ∼6 Ma and 3.6 Ma. While variability is present, sediment fluxes fall between ∼0.1-0.4 g cm −2 ky −1 for the period from ∼2.5 to 3.6 Ma.
Combining our opal concentration and CFP-based sediments MARs, we find that 3 He ET -derived opal fluxes display a similar pattern to the concentrations, with greater fluxes occurring prior to ∼3.6 Ma and relatively low and stable fluxes from ∼2.5 to 3.6 Ma ( Figure 2d). During the last ∼1 My of the opal flux record (i.e., ∼2.5-3.6 Ma), two prominent minima are found from ∼2.9 to 3.0 Ma and again after ∼2.7 Ma. We note that in the late Miocene (∼5.3-6 Ma), relatively lower opal fluxes are observed at ∼5.5-5.6 Ma and before ∼6 Ma.

Excess Barium (Ba xs ) and Manganese (Mn xs )
Ba xs fluxes, produced using Ba xs concentrations (see Materials and Methods) and 3 He ET -derived sediment MARs, mirror opal fluxes at ODP 885/886 ( Figure 3c), with elevated values found throughout the late Miocene and early Pliocene from ∼3.6 to 5.4 Ma, a decreasing trend in fluxes from ∼4.6 to 3.6 Ma, and somewhat stable values after ∼3.6 Ma. Two additional, notable features of the Ba xs flux include lower values around 5.1-5.2 Ma, and a substantial peak at ∼3.8-4.0 Ma.
Mn xs concentrations in sediment samples offer the potential to gain additional insight into past redox changes at the sediment-water interface (Calvert & Pedersen, 2007;) (see Materials and Methods). We find little to no Mn xs present for the period of ∼4-6 Ma (particularly when compared to the later portion of our record), followed by a sharp transition to values ranging between ∼2 and 7,045 ppm during ∼2.5-4 Ma ( Figure 3d). Interestingly, after ∼2.6 Ma, there seems to be a stabilization of Mn xs , which coincides with some of the lowest opal and Ba xs fluxes, and is after the iNHG. Finally, aside from a peak spanning ∼3.9-4.0 Ma, the ∼2 million-year low Mn xs interval aligns with the period of elevated Ba xs and opal fluxes.

An Opal Perspective on Central Subarctic North Pacific Productivity in the Pliocene
Our new opal flux record at ODP 885/886 is quite distinct from the only existing opal flux reconstruction from the same site . While we describe the underlying causes of said discrepancy in detail in Text S1 in Supporting Information S1, it is our utilization of the CFP 3 He ET to determine sediment (and in turn opal) fluxes that provides new insight into Mio-Plio-Pleistocene productivity at ODP 885/886. As previously shown by Abell et al. (2021), and confirmed with our new data here, sediment MARs, when normalized with a CFP, are relatively high through the latest Miocene and early Pliocene (∼3.6-5.4 Ma) and lower thereafter ( Figure 2c). In fact, our 3 He ET -derived sediment fluxes, with high fluxes during the early Pliocene and lower fluxes during the mid-and late Pliocene, display the opposite pattern to those produced using the age model. We attribute the anomalously low age model-derived MARs during the period of ∼3.6-5.4 Ma to either sediment winnowing, indicated by a focusing factor of ∼0.3 ( Figure S3 in Supporting Information S1), or previously undescribed hiatuses (Text S1 in Supporting Information S1).
The elevated 3 He ET -derived opal accumulation rates from ∼3.6 to 5.4 Ma overlap with periods of high Ba/Sc ratios (G. R. Dickens & Owen, 1996), the presence of a diatom ooze (G. R. Dickens & Barron, 1997), and the partial dissolution of magnetic minerals   (Figure 4). After ∼3.6 Ma, minimal iron reduction, higher clay fractions, and low Ba/Sc ratios are also consistent with the low opal fluxes in our reconstruction. For context, higher Ba/Sc values are interpreted to reflect increased export production, while the dissolution   Rea et al. (1998). (d) 3 He ET -derived opal fluxes for this study and age model-derived opal fluxes from Snoeckx et al. (1995). The error envelope represents propagated analytical uncertainties for opal concentrations and 3 He ET -derived fluxes. Ages for all previously published data were updated using GPTS 2020 (Ogg, 2020).
of magnetic minerals implies more reducing conditions in the sediment porewaters, which may be related to an enhanced input of organic matter. The interpretation that the dissolution of magnetic minerals in particular reflects reducing conditions at the sediment-water interface in the subarctic North Pacific is supported by evidence from the late Pleistocene (Korff et al., 2016). Overall, the utilization of updated methodologies for the determination of lithogenic contributions and sediment fluxes allows for a reassessment of opal fluxes at ODP 885/886, which, along with additional data from previous studies, all strongly support the notion that the central subarctic North Pacific experienced elevated export productivity during the late Miocene and early Pliocene (∼3.6-5.4 Ma) compared to the late Pliocene and early Pleistocene (post-3.6 Ma).

Additional Evidence of Enhanced Early Pliocene Productivity at ODP 885/886
Considering the similar patterns in opal and Ba xs fluxes (Figure 3), and taking into account that their preservation is influenced by different factors ( Torres et al., 1996;Tréguer & De La Rocha, 2013), close correspondence between the two proxies suggests the observed signal is representative of export productivity in the Mio-Plio-Pleistocene (Figures 3  and 4). Complexities related to these productivity records that merit consideration include (a) the larger magnitude of variability in the opal (∼12-fold from early Pliocene maximum to post-3.6 Ma average) compared to Ba xs (∼4-fold) fluxes, and (b) the impact of Pliocene ocean conditions on preservation. We explore these intricacies in detail in Text S2 in Supporting Information S1, but ultimately we find that both opal and Ba xs fluxes indicate  (Westerhold et al., 2020). Black line indicates a 20-ky smoothing, while the red line denotes a 1-My smoothing. (b) 3 He ET -derived opal fluxes. Error envelope represents propagated uncertainties for opal, analytical uncertainties for 3 He ET , and the statistical uncertainties for 3 He ET . (c) 3 He ET -derived Ba xs fluxes. Error envelope represents propagated uncertainties for Ba xs , analytical uncertainties for 3 He ET , and the statistical uncertainties for 3 He ET . (d) Ba/Sc ratios from this study (light brown) and Dickens and Owen (1996) Snoeckx et al. (1995). Light red shading indicates period of Fe magnetic mineral reduction G. R. Dickens & Owen, 1996). Light green shading represents pennate diatom ooze (G. R. Dickens & Barron, 1997).
subarctic North Pacific export production was higher during the late Miocene and early Pliocene compared to after ∼3.6 Ma, and that Pliocene export production prior to ∼2.7 Ma was higher than modern levels.
Porewater data from shipboard measurements, along with analyses of magnetic mineral grains, provide evidence for Mn and Fe dissolution during the period of ∼4.6-5.8 Ma at ODP 885/886 G. R. Dickens & Owen, 1996;D. Rea et al., 1993). Mn xs concentrations in sediment samples offer the potential to gain additional insight into past redox changes at the sediment-water interface (Calvert & Pedersen, 2007;. Under oxic conditions, dissolved Mn tends to precipitate, typically in the form of Mn oxides, as it is oxidized from Mn(II) to Mn(IV). However, if reducing conditions are encountered, the solubility of Mn increases as it is reduced, and can reprecipitate in shallower sediments where oxygen is present or mobilize out of the sediment into the water column (Burdige & Gieskes, 1983;Calvert & Pedersen, 2007;Froelich et al., 1979). As such, higher values of Mn xs in marine sediments can indicate oxic conditions, while lower values suggest an anoxic setting.
We find that Mn xs concentrations support previous studies at ODP 885/886, which is that there was an interval of high productivity during the latest Miocene and early Pliocene (∼3.6-5.4 Ma) (Figures 3 and 4) G. R. Dickens & Barron, 1997; G. R. Dickens & Owen, 1996). Considering the oxygen concentrations of the deep subarctic North Pacific today (∼150 μmol kg −1 ) (Garcia et al., 2019a), and barring extensive, long-term changes in bottom water oxygen (at least below ∼5,000 m) in the Pliocene prior to ∼3.6 Ma, we posit that the Mn xs record reflects sedimentary reducing conditions driven by an enhanced rate of organic matter input during the early Pliocene. We note that both the low Mn xs values prior to ∼4 Ma and large fluctuations in Mn xs observed after ∼3.6-4 Ma could be partially driven by processes other than the flux of organic carbon (i.e., change in bottom water oxygen due to altered circulation and productivity in the Southern Ocean and/or North Atlantic Ford et al., 2022;Jian et al., 2023;McKay et al., 2012;Yi et al., 2023)), and more work is needed to elucidate the controls (and the magnitude of their impact) on deep North Pacific oxygenation in the Mio-Plio-Pleistocene. Overall, opal and Ba xs fluxes, along with Mn xs concentrations, all support higher export productivity from ∼3.6 to 5.4 Ma (Figures 3 and 4).

The Larger Picture of Pliocene Subarctic North Pacific Export Productivity
We now turn to comparing our new Pliocene reconstructions of export productivity to a subset of available records of productivity and/or water column conditions found throughout the open-ocean subarctic North Pacific ( Figure 1). These data sets include: (a) opal fluxes from ODP 887 in the Gulf of Alaska (D. , and (b) opal and calcium carbonate (CaCO 3 ) fluxes from ODP 882 in the western subarctic North Pacific ( Figure 5) G. Haug et al., 1995;G. H. Haug et al., 1999).
We note that these other flux records are not determined using a CFP, and as such are susceptible to sediment focusing and winnowing as described in the Materials and Methods. Evidence exists for horizontal sediment redistribution at ODP 882 during the late Pleistocene, but the overall pattern of relative fluxes is similar for the last ∼150 ky using both traditional methods and a CFP (Jaccard et al., 2009;Serno et al., 2017). Whether this finding holds for earlier time periods at this site is impossible to determine at present. No CFP data exists for ODP 887. To evaluate the characteristics of Pliocene subarctic North Pacific Ocean export production and its connection to ocean circulation on a regional scale, we compare our new CFP-derived opal records from ODP 885/886 along with the age model-derived export productivity proxy fluxes from ODP 887 and 882. However, we acknowledge that there are key uncertainties related to this analysis, and as such we are conservative when interpreting any observed spatial patterns in Pliocene North Pacific export production.
ODP 885/886, 887, and 882, all generally record elevated siliceous export production during the Pliocene compared to after the iNHG at ∼2.7 Ma ( Figure 5) (G. H. Haug et al., 1999;Raymo, 1994). However, there seems to be an east-west dichotomy in the timing of maximum opal deposition. In the central and eastern subarctic North Pacific, which includes ODP 885/886 and 887, the highest opal fluxes for the interval of ∼2.5-6 Ma predominantly occur prior to ∼3.6 Ma, during the late Miocene and early Pliocene, while in the west, peaks in opal are found at ∼3.0 and 3.5 Ma. Elevated CaCO 3 fluxes at ODP 882 also predominantly appear post-3.6 Ma. Interestingly, new results from IODP U1417, a core located near ODP 887, may lend support for lower productivity in the eastern subarctic North Pacific during the mid-and late Pliocene (Sánchez Montes et al., 2022). These data show relatively lower pelagic diatom fluxes in the Gulf of Alaska after ∼4 Ma (Sánchez Montes et al., 2022), but we note that due to the length of the records, the relevance to our study is tentative. To better constrain the implications of these subarctic North Pacific productivity records, we must consider our current understanding of dynamics in the region, both in the Quaternary and the Pliocene.
As previously detailed in the Introduction, altered stratification in the subarctic North Pacific has been proposed to explain varying productivity observed in various time periods since the Neogene G. H. Haug et al., 1999;Kienast et al., 2004;Ren et al., 2015;Sigman et al., 2004;Studer et al., 2012). For the Pliocene specifically, the breakdown of the modern halocline is suggested to have been driven by an active PMOC and NPDW formation . Elevated opal and CaCO 3 fluxes along with redox element data at ODP 882 (i.e., higher export production and less corrosive waters), a reversed benthic 13 C gradient compared to today in the North Pacific Basin (suggesting northern-sourced waters flowing southward at shallow to mid-depths), and a greater equatorial Pacific pH gradient (upwelling of more corrosive, nutrient-laden waters), all support this hypothesis Ford et al., 2022;Shankle et al., 2021). However, this climate-ocean circulation connection runs counter to a variety of geochemical evidence from the late Pleistocene, which points to enhanced NPIW/NPDW formation and lower subarctic North Pacific export productivity during colder climate states in the Northern Hemisphere (Okazaki et al., 2010;Rae et al., 2014Rae et al., , 2020. Using the Pliocene export productivity results presented here, we provide a novel analysis of the relationship between NPDW formation and biogeochemistry in the North Pacific across this time.

Export Production and NPDW Formation in a Warmer World
The physical basis for PMOC in the warmer world of the Pliocene is that reduced meridional sea surface temperature (SST) gradients (C. M. Brierley et al., 2009;Fedorov et al., 2013;Haywood et al., 2020;McClymont et al., 2020), possibly driven by an altered low cloud-albedo feedback (Burls & Fedorov, 2014a, 2014b, lead to reduced precipitation over the mid-latitude oceans . This process acts to increase sea surface salinity (SSS) in the subarctic North Pacific, decreasing stratification, and ultimately driving the formation of NPDW and development of a PMOC . The connection between regional precipitation/evaporation, the presence of the halocline, and the potential of NPDW (and in turn PMOC) formation have previously been evaluated (Emile-Geay et al., 2003;Menviel et al., 2012;Warren, 1983), and support the findings of  in the northwest subarctic North Pacific.
To assess how NPDW formation may impact nutrient availability, and in turn productivity, at the sites considered here (ODP 885/886, 882, 887, and IODP U1417), we evaluate model output of mixed layer depths Menviel et al., 2012), phosphate concentrations (Menviel et al., 2012), and the pathway of NPDW   Thomas et al., 2021) in which a PMOC is active. When the modern halocline is absent, higher near-surface macronutrient concentrations are found throughout the subarctic North Pacific, arising from enhanced diapycnal mixing, which stimulates enhanced productivity (Menviel et al., 2012). We note here (and elaborate on in Text S3 in Supporting Information S1), that other modeling work has shown lower, not higher, nutrient concentrations under enhanced PMOC scenarios (Menviel et al., 2014;Rae et al., 2020).
For ODP 885/886, the abundant nutrients are likely delivered to the mixed layer via the subpolar gyre from the initial sites of deep convection in the Sea of Okhotsk, Bering Sea, and near Kamchatka . This hypothesized pathway is supported by tracer analysis of the same Pliocene simulation (Figures 6b and 6c) . In the case of ODP 887 and IODP U1417, greater opal and pelagic diatom fluxes could be related to the deeper mixed layer and enhanced mixing Sánchez Montes et al., 2022), as the sites fall within a region of elevated Pliocene SSS, and are in a location of potential deep water formation (Figure 6b), similar to ODP 882 Thomas et al., 2021).
With modeling results supporting an interpretation of Pliocene subarctic North Pacific productivity being driven by enhanced nutrient availability related to NPDW formation, and assuming this relationship holds through time,  Other marine sediment core sites mentioned in the text are indicated by black circles. (b) The origins of late Pliocene simulation mixed layer mean particle transport (CESM v. 1.0.4, configuration T31 gx3v7) . (c) The destinations of late Pliocene simulation mixed layer particle transport . Panels b and c roughly represent the locations of NPDW formation and subsequent upwelling, respectively. All displayed model output is available from Davis et al. (2023) and Thomas et al. (2021).
we can utilize the differences in the three reconstructions to potentially discern the characteristics of the deep water formation regions. Acknowledging that the lack of Pliocene CFP data for cores ODP 887 and 882 makes our analysis of the spatial and temporal variability between the sites more uncertain, we only use the spatial variability in our assessment here assuming a Pliocene "PMOC-on" scenario. Ultimately, more CFP-normalized records are necessary to test the veracity of any assessment of spatially varying export productivity (and its link to NPDW formation) in the Pliocene North Pacific. We emphasize that the CFP-derived export productivity proxy fluxes and redox element concentrations from ODP 885/886 are robust and allow for a novel assessment of the temporal evolution and characteristics of the relationship between North Pacific Ocean circulation and productivity more broadly, which we apply throughout the rest of the Discussion.
Because the highest opal and diatom fluxes in the central and eastern subarctic North Pacific are generally found prior to ∼3.6 Ma, while peak productivity in the western subarctic North Pacific occurs after this interval, we posit that NPDW formation occurred in both the east and west subarctic North Pacific prior to ∼3.6 Ma, and that PMOC was stronger overall during the early Pliocene. From ∼4.6 to 3.6 Ma, the location of NPDW formation then gradually shifted to be located only in the western subarctic North Pacific, and PMOC experienced a weakening across this interval. After ∼3.6 Ma, PMOC either marginally weakened or remained stable until a complete shutdown occurred at ∼2.7 Ma coinciding with the development of expansive Northern Hemisphere ice sheets (G. H. Haug et al., 1999;Kleiven et al., 2002). Support for our multi-phased Pliocene NPDW formation hypothesis can again be found in modeling work and proxy reconstructions. The regions of modeled maximum mixed layer depths and NPDW formation are predominantly found in the western subarctic North Pacific and Bering Sea (i.e., near ODP 882) (Figure 6), suggesting that these locations would still be relevant areas of NPDW formation as PMOC weakened, while any NPDW formation in the eastern subarctic North Pacific (i.e., near ODP 887 and IODP U1417) would likely only occur under stronger PMOC conditions Menviel et al., 2012;Thomas et al., 2021). Lower TOC fluxes in the eastern subarctic North Pacific prior to ∼3.6 Ma point to a more well-mixed and oxygenated water column (Sánchez Montes et al., 2022), which is also observed in model simulations (Figure 6a) (Davis et al., 2023;Menviel et al., 2012), supporting early Pliocene NPDW formation at the site. The opposite is true for the western subarctic North Pacific if CaCO 3 fluxes primarily reflect deep water corrosivity, as ODP 882 CaCO 3 fluxes are the highest after ∼3.6-3.8 Ma . Considering the difference between the PMOC-on and preindustrial simulations of Menviel et al. (2012), lesser NPDW production post-3.6 Ma would likely lead to lower nutrient concentrations in surficial subarctic North Pacific waters overall, and in conjunction with already weakened westerly winds in the Pliocene (Abell et al., 2021), would decrease nutrient delivery to the central subarctic North Pacific and produce the observed decline in productivity at ODP 885/886. Important uncertainties related to the use of ODP 882 CaCO 3 flux as a proxy for NPDW formation in the Pliocene include (a) its direct relation to PMOC strength (which seems to imply a stronger PMOC in the mid-to-late Pliocene, opposite to the combined modeling work of , Thomas et al. (2021), Ford et al. (2022), and Davis et al. (2023)), (b) whether preservation is the only driver of variability in the flux across the entire Pliocene, and (c) if the flux is purely recording bottom water conditions, how much of the subarctic North Pacific basin is integrated into the overall signal. We elaborate on these potential complications in Text S4 in Supporting Information S1.
Our interpreted multi-phase change in North Pacific Ocean circulation would have substantial ramifications for our understanding of the global climate system, particularly through impacts on the carbon cycle. As has previously been described, a stronger surface-to-deep water connection in the North Pacific, in conjunction with decreased overall nutrient utilization (G. H. Haug et al., 1999;Studer et al., 2012), would likely lead to enhanced CO 2 release to the atmosphere Sánchez Montes et al., 2022;Sigman et al., 2004). With the intensity of NPDW formation related to stratification, and in turn the efficiency of the biological pump, then an intensified PMOC in the early Pliocene would have driven higher atmospheric CO 2 levels, followed by a decline during the mid-to-late Pliocene (∼2.7-3.6 Ma), and even lower values after the iNHG. This framework and the records of export production track benthic oxygen isotope records well (Ahn et al., 2017;Lisiecki & Raymo, 2005;Westerhold et al., 2020), and generally follow available pCO 2 estimates from various proxies (Figure 7) (e.g., Guillermic et al., 2022;Rae et al., 2021).
The interpretation that PMOC was a feature of the Pliocene, and varied on million-year timescales, is also consistent with other proxy data related to potential drivers of NPDW formation. Specifically, moisture transport via the East Asian Summer Monsoon to the subarctic North Pacific has been suggested to play a substantial role in the development of the modern halocline (Emile-Geay et al., 2003), and several studies have reported evidence of a weakened East Asian Summer Monsoon during the early Pliocene followed by an intensification toward the  (Westerhold et al., 2020). Black line indicates a 20-ky smoothing, while the red line denotes a 1-Myr smoothing. (b) Compilation of atmospheric pCO 2 records based on either marine alkenone 13 C or carbonate 11 B. Data from Rae et al. (2021) and Guillermic et al. (2022). Error "clouds" represent 1σ uncertainties on the final pCO 2 estimates. (c) 3 He ET -derived opal fluxes. Error envelope represents propagated uncertainties for opal, analytical uncertainties for 3 He ET , and the statistical uncertainties for 3 He ET . (d) 3 He ET -derived Ba xs fluxes. Error envelope represents propagated uncertainties for Ba xs , analytical uncertainties for 3 He ET , and the statistical uncertainties for 3 He ET . (e) Age model-derived opal flux data from ODP 887 (D. . (f) Age model-derived opal flux record from ODP 882 G. H. Haug et al., 1999).
Plio-Pleistocene transition Nie et al., 2014;Sun et al., 2010;Y. G. Zhang et al., 2009;Zheng et al., 2004), although this is heavily debated. The reduced strength of the East Asian Summer monsoon would have led to decreased precipitation delivered to the North Pacific, supporting increased SSS, and in turn NPDW formation, at this time. From ∼3.5−4.0 Ma to the end of the Pliocene, strengthening of the monsoon would have progressively freshened the subarctic North Pacific, reducing SSS and convection, weakening PMOC. Shifts in both Pacific zonal and meridional (C. M. Brierley et al., 2009;Fedorov et al., 2013;Wara et al., 2005) SSTs support large-scale changes in precipitation through the Pliocene, which could influence precipitation and ultimately impact the salinity of the subarctic North Pacific .
While we have provided an original assessment of North Pacific export production and the potential insights it provides on circulation under the assumption that a PMOC was active in the Pliocene, there remains an important caveat to the broader hypothesized scenario of Pliocene NPDW formation and generally greater-than-modern export production. This stems from the basic mechanistic connection between enhanced NPIW or NPDW formation and macronutrient availability in near-surface waters. To this point, regardless of the spatial and temporal variability of North Pacific export production and/or NPDW formation, we have assumed that an active PMOC would lead to enhanced surface ocean nutrient concentrations. However, in the case of an active overturning cell in the subarctic North Pacific during the Pliocene, without global enhancement of diapycnal mixing (Menviel et al., 2012), younger surface waters to be transported to depth would likely contain lower nutrient concentrations because of their origination in the subtropics, similar to the modern North Atlantic (Menviel et al., 2014;Rae et al., 2020). For a situation where no NPDW is formed, but NPIW formation is enhanced, then this intermediate water would reduce upward mixing of nutrients to surface waters, again limiting productivity Worne et al., 2019). We discuss this conundrum in detail below, but in doing so also illuminate the weakness in applying a Quaternary-centric perspective on Pliocene North Pacific Ocean circulation, ultimately further highlighting the complex nature of the subarctic North Pacific productivity-circulation connection.

Assessing a Pleistocene Glacial-Interglacial North Pacific Ocean Circulation-Export Production Regime for the Pliocene
By combining a suite of geochemical data and model simulations, Rae et al. (2020) provide evidence that intermediate waters of the subarctic North Pacific were better ventilated during the LGM via an overturning cell, and posit that this scenario may hold for other glacial periods of the Quaternary and the late Pliocene. In a similar vein, several previous studies suggest that during colder intervals of the Quaternary, NPIW production was enhanced (Keigwin, 1998;Kim & Park, 2008;Knudson & Ravelo, 2015;Rafter et al., 2022;Worne et al., 2019;Zhao et al., 2021). While the proposed mechanisms driving these paleo-circulation changes vary (e.g., sea ice expansion in the Bering Sea, altered position and intensity of wind systems, shifts in regional hydroclimate) Rae et al., 2020;Worne et al., 2019), these findings themselves suggest if an increase in the lateral and/or vertical extent of NPIW (or the formation of NPDW for that matter) was present in the warmer world of the Pliocene, a reduction in overall subarctic North Pacific export productivity should be observed, at least relative to periods of weakened-NPIW or no-NPDW formation, such as today. As shown here and in previous studies at ODP 885/886, Pliocene subarctic North Pacific productivity was higher than today, not lower.
Our analysis of ODP 885/886 data suggests that subarctic North Pacific export productivity was enhanced during much of the Pliocene compared to the latest Pliocene and Quaternary, and the level of enhancement was temporally varying. Additionally, there is other evidence for combined greater export productivity and lower nutrient utilization prior to iNHG G. H. Haug et al., 1999;Sánchez Montes et al., 2022;Sigman et al., 2004;Studer et al., 2012). Considering the available data, for the hypothesis of no overturning/reduced NPIW formation under warmer climate states  to hold, some mechanism(s) must have been altered to reduce stratification (i.e., increase mixing) but still not drive enhanced formation of NPIW/NPDW. Below we assess the proposed drivers of glacial-interglacial changes in subarctic North Pacific Ocean circulation through the Quaternary in the context of their impacts on NPIW/NPDW in the Pliocene.
Lower sea ice production in the warmer Pliocene (Dowsett et al., 2009(Dowsett et al., , 2010(Dowsett et al., , 2016 would lead to reduced brine rejection and in turn lower SSS. Additionally, the effect of poleward-shifted and weakened westerlies could lead to altered storm tracks, ultimately increasing net precipitation (and thus deceasing SSS) in the subarctic North Pacific (Abell et al., 2021). These changes in the winds could also produce reduced Ekman suction, decreasing upwelling of salty waters, at least in the central subarctic North Pacific (Gray et al., 2018. Finally, enhanced moisture transport to the subarctic North Pacific in a warmer world (C. M. Brierley et al., 2009;C. M. Brierley & Fedorov, 2010;Feng et al., 2022;Han et al., 2021) would also decrease SSS. Taken together, these changes would imply a lack of NPIW/NPDW formation in the Pliocene, leading to elevated surface nutrients and/ or exchange with nutrient-laden deep waters (similar to today). At first order, these shifts in various components of the climate system would favor the suppositions of Quaternary studies (i.e., no NPDW formation in a warmer world such as the Pliocene).
However, by applying the new findings presented here (i.e., our temporally varying productivity records from ODP 885/886), we can provide constraints on when these mechanisms would have to shift, and test the feasibility of said changes on NPIW/NPDW formation and productivity. For example, there would have to be an early to mid-Pliocene (∼3.6-4.6 Ma) shift in (a) sea ice production (likely minimal throughout this interval, see references above), (b) characteristics of the westerlies (unlikely, see Abell et al., 2021), and/or (c) regional hydroclimate. For the latter mechanism, if moisture transport to the high latitudes is linked to meridional and zonal temperature gradients, then this may be a plausible mechanism to alter circulation across the early and mid-Pliocene, as changes in zonal/ meridional SSTs (Fedorov et al., 2013) and the East Asian Summer Monsoon Nie et al., 2014;Sun et al., 2010;Y. G. Zhang et al., 2009;Zheng et al., 2004) have been suggested for this interval. However, we note that, as with other drivers, the sign of change for Pliocene North Pacific hydroclimate is still contested.
Additionally, some of these described changes in the climate system could also lead to elevated SSS in this warmer world. For one, while the westerlies over the North Pacific may have been located closer to the poles during the Pliocene, if they were less zonal in nature as they are during the warmer Holocene compared to the LGM , then the cross-gyre export of high-salinity subtropical water may have been increased (potentially counteracting increased net precipitation). Additionally, there is modeling support for reduced net precipitation over the subarctic North Pacific during the mid-Pliocene . We note that again, these works do not provide evidence for changes throughout the Pliocene, and as such are not sufficient to fully evaluate our new productivity records without speculation.
On top of the inability to fully explain the temporal variations in Pliocene productivity, the overwhelming caveat for applying the Quaternary-style circulation scheme (i.e., no NPDW formation in a warmer world)  to the Pliocene is finding an explanation for not just elevated productivity compared to glacials post-2.7 Ma, but also compared to modern warm conditions, where NPIW production is also limited. Again, the delivery of nutrients to subarctic North Pacific surface waters must have been higher than today to explain enhanced export production at ODP 885/886. One possibility is that the iron limitation found in the subarctic North Pacific today was relaxed during the Pliocene. Recent data can be used to suggest that Pliocene dust deposition to the mid-latitude North Pacific may have been similar to or greater than today (Abell et al., 2021;Tang et al., 2022). However, this finding runs counter to a plethora of other dust studies D. K. Rea et al., 1998;Q. Zhang et al., 2020; and nutrient utilization proxies (Ren et al., 2015;Sigman et al., 2004;Studer et al., 2012). Additionally, there is conflicting evidence for dust flux changes through the early and mid-Pliocene (Abell et al., 2021;Anderson et al., 2020;D. K. Rea et al., 1998;Tang et al., 2022;Q. Zhang et al., 2020;W. Zhang et al., 2016), which would be necessary to explain the temporal variability found in our compiled subarctic North Pacific export productivity records. As such, future work is needed to evaluate this hypothesis.
Overall, some of the mechanisms posited for driving changes in subarctic North Pacific glacial-interglacial productivity could lead to greater Pliocene export production compared to cooler intervals thereafter without invoking NPDW formation Rae et al., 2020;Worne et al., 2019). However, they cannot immediately explain greater-than-modern production in the warmer world of the Pliocene, nor is it immediately evident that they are consistent with our postulated temporal variations in central subarctic North Pacific Pliocene export production. As such, we favor the hypothesis of NPDW formation and an active PMOC in the Pliocene to explain our reconstructed changes in opal and Ba xs fluxes at ODP 885/886. Importantly, there are other potential drivers to consider that could bring the Quaternary scheme of North Pacific Ocean circulation more in line with our data. While these mechanisms are either controversial and/or difficult to reconstruct, we nevertheless provide an overview for completeness.

Other Potential Drivers of Elevated Export Production in the Mio-Pliocene North Pacific
Recent work has attempted to explain similarities in the timing of low to mid-latitude productivity shifts during the early Pliocene (∼4.4-4.6 Ma) through an enhanced global nutrient inventory driven by variability of orbital configurations (Karatsolis et al., 2022). Our new export productivity records do show a substantial decline at roughly the same time interval, and as such we cannot rule out an influence from greater nutrient availability. However, recent work has shown that both the temporal and spatial variability of the late Miocene-early Pliocene biogenic bloom does not support enhanced global nutrient inventory driven by increased delivery of sediment and nutrients via rivers from shifts in monsoon strength and/or weathering (Farrell et al., 1995;Filippelli, 1997;Gupta & Thomas, 1999;Hermoyian & Owen, 2001;Karatsolis et al., 2022) when a global compilation of marine sites is utilized (Pillot et al., 2023). Additionally, these authors suggest a potential decline in East Asian monsoon-related precipitation after the 4.4-4.6 Ma event (leading to decreased runoff and nutrient input to the oceans) (Karatsolis et al., 2022), which would lead to higher subarctic North Pacific SSS and in turn enhanced NPIW/NPDW formation in the mid-to-late Pliocene, opposite to the interpretations presented here and previous modeling work Davis et al., 2023;Ford et al., 2022). Further work is needed to evaluate high-latitude productivity data in the context of this low-latitude-based scenario.
Another mechanism to consider is the concept that export productivity proxies may not directly reflect primary production in the past. Specifically, export efficiency out of the photic zone can change under varying conditions, leading to shifts in export productivity without corresponding changes in primary production, or a lack of correlation in general (Cael & Follows, 2016;Henson et al., 2012;Le Moigne et al., 2016;Maiti et al., 2013;Serno et al., 2014). For the Pliocene North Pacific, the notion that export efficiency increases with stratification is particularly relevant (Kemp & Villareal, 2013). If the concept of a more stratified North Pacific during warmer climates (i.e., the proposed Quaternary circulation structure) is correct , then it could be suggested that the greater-than-present Pliocene export production in the central subarctic North Pacific is an artifact of enhanced export efficiency due to stratification, not higher primary production. There are several lines of evidence that may be used to refute this hypothesis. One is that there are numerous proxies pointing to less, not more, stratification under Pliocene conditions. These include combined opal productivity and nitrogen isotopes at ODP 882, which suggest nutrient-rich waters must have been mixed into the surface ocean during at least the late Pliocene (G. H. Haug et al., 1999;Sigman et al., 2004;Studer et al., 2012). Even if the opal flux record is called into question (i.e., due to a lack of CFP-derived sediment accumulation rates), one would need to explain lower nutrient utilization in this warmer climate state. Additionally, the previously published CaCO 3 fluxes from ODP 882 and the compiled Pliocene North Pacific benthic 13 C data support NPDW formation, and thus lower stratification. Finally, SST records from ODP 882 based on two independent proxies along with size specific diatom-bound nitrogen isotopes have been interpreted to reflect an increase in mixed layer seasonality after ∼2.7 Ma, which would be expected to occur with the proposed onset of the modern halocline in the subarctic North Pacific (G. H. Haug et al., 1999;Studer et al., 2012). Ultimately, the hypothesis that Mio-Pliocene North Pacific export production and primary production may be decoupled cannot be completely ruled out, again leaving the application of the Quaternary perspective on North Pacific Ocean circulation for the Pliocene somewhat ambiguous. Future work should aim to address this key uncertainty in our understanding of paleo-productivity proxies and their link to paleo-ocean circulation.

The Role of Paleogeographic Changes in Driving Proposed Pliocene North Pacific Ocean Circulation
Altered paleogeography suggested for the late Miocene and Pliocene, such as open Central American and Indonesian Seaways, a closed Bering Strait, or Tibetan Plateau uplift, could also be important for conditions in the North Pacific Ocean (C. M. Brierley & Fedorov, 2016;Tan et al., 2022). However, (a) considering the lack of age constraints for Indonesian Seaway closure (Cane & Molnar, 2001;Hall, 2002), (b) the copious yet widely debated ages for the closure and opening of the Central American Seaway (G. Haug & Tiedemann, 1998;Molnar, 2008;Montes et al., 2015) and Bering Strait (see Marincovich & Gladenkov, 1999 and references therein), respectively, and (c) ambiguity related to the timing of orographic change in Tibet over the Mio-Pliocene (E. Wang et al., 2012;G. Wang et al., 2011;Yin, 2010), at present, it does not appear likely that these paleogeographic changes can explain the temporal variability of subarctic North Pacific productivity observed at ODP 885/886. As such, we limit our discussion here to the potential changes in ocean circulation and the proposed atmospheric/ocean drivers, but not any potential shifts initiated by paleogeographic alterations.

Conclusions and Outlook
We have provided new paleoceanographic data indicating elevated export productivity in the early Pliocene central subarctic North Pacific. In doing so, this work rectifies longstanding discrepancies at ODP 885/886, and provides novel evidence of subarctic North Pacific-wide shifts in productivity and ocean circulation over the period of ∼2.5-6.0 Ma. Combining our new data sets, previously published opal, diatom, and carbonate records, and recent modeling results, we posit that our data support the formation of NPDW in the Pliocene and offer constraints on the long-term variability of PMOC during this warmer-than-present interval of Earth's history. Specifically, we suggest that NPDW formation may have been enhanced during the latest Miocene and early Pliocene (∼4.6-5.4 Ma), experienced a decline from ∼4.6 to 3.6 Ma, and remained in this weaker state (or continued to marginally weaken) prior to the proposed setup of Quaternary-style alternating glacial-interglacial circulation schemes at ∼2.7 Ma. This hypothesis is generally supported by geochemical proxies for past precipitation and SSTs, the dominant controls on likely mechanics driving changes in PMOC. However, there are several limitations to our analysis. These include (a) the lack of CFP-derived fluxes for ODP sites 887 and 882, as well as (b) the seemingly opposite signals of NPDW formation intensity suggested by our new data set and previous modeling simulations compared to Mio-Plio-Pleistocene CaCO 3 fluxes at ODP 882. It is crucial to resolve these issues in order to validate our proposed spatial variability in export production, and in turn ocean circulation. Regardless, our new data sets combined with previously published export productivity records provide new information on the connection between, and evolution of, Pliocene subarctic North Pacific Ocean circulation and export production with an active PMOC.
While we favor a "NPDW formation-on" scenario to explain our export productivity and sediment redox chemistry data sets at ODP 885/886, there are caveats to the concept of Pliocene PMOC more generally, particularly stemming from evidence supporting a Quaternary-style circulation framework in the North Pacific, that need to be addressed. Comparisons with both the modern North Pacific and North Atlantic as well as North Pacific data from glacials and interglacials of the Quaternary (Menviel et al., 2014;Rae et al., 2020;Rafter et al., 2022) suggests that the impact of overturning on nutrients and productivity is the opposite to what has been proposed for the Pliocene if NPDW forms in this warmer period of Earth's history Ford et al., 2022;Menviel et al., 2012). If all data (i.e., for both the Pliocene and Quaternary) is taken at face value, then enhanced NPIW/NPDW formation could lead to both higher and lower North Pacific productivity, depending on the climate state.
Ultimately, there may not be an a priori reason that we should expect the overturning response to change linearly with global climate, particularly on geologic timescales, as there are a host of feedbacks and boundary conditions that could impact ocean circulation in a non-linear fashion (Ferreira et al., 2018). However, if this is true, then our new data presented here, along with the previously published proxy and modeling work across the Pliocene and Quaternary, point to a complex connection between North Pacific productivity and ocean circulation, a finding that could make it difficult to predict future changes in these systems. With continued warming expected for the next century and beyond, information related to altered Pacific Ocean circulation in the Pliocene is valuable, especially considering the potential ramifications of NPDW formation (or a lack thereof) in the North Pacific for the global carbon cycle.

Conflict of Interest
The authors declare no conflicts of interest relevant to this study.

Data Availability Statement
All new data and results can be found in data set file Data Set S1 stored in the figshare.com data repository at: https://doi.org/10.6084/m9.figshare.21668156. The code files for the analyses performed on the data can be found in the figshare.com data repository at: https://doi.org/10.6084/m9.figshare.21668168. We note that all previously published data from ODP 885/886 that provided sample information on core depth G. R. Dickens & Barron, 1997; G. R. Dickens & Owen, 1996;Snoeckx et al., 1995) were converted to the age model of Dickens et al. (1995) with updated polarity chron age boundaries (Ogg, 2020) to be consistent with the data new to this study.