Effect of an Open Central American Seaway on Ocean Circulation and the Oxygen Minimum Zone in the Tropical Pacific From Model Simulations

The tectonic closure of the Central American Seaway (CAS) during the mid‐Miocene to mid‐Pliocene (∼16–3 Ma BP) is thought of as a key interval for the onset of the present‐day tropical Pacific oxygen minimum zone (OMZ). In this study we investigate the impact of an open CAS on the ocean circulation and the OMZ in the tropical Pacific. We perform a series of sensitivity experiments with the Kiel Climate Model, where we vary the CAS sill depth from shallow to deep. We find that the eastern tropical Pacific OMZ was less developed during the period of an open CAS. This is driven mainly by an enhanced eastward subsurface current that facilitated an increased oxygen supply from the western tropical Pacific. In addition, a small decrease in net marine primary production and subsequent weaker export of particulate organic carbon induced less subsurface oxygen consumption in the Eastern Equatorial Pacific.


Introduction
Oceanic oxygen and so-called oxygen minimum zones (OMZs) have gained attention over the past 50 years, as observations over recent decades show a general tendency toward lower oxygen levels in the ocean (Karstensen et al., 2008;Schmidtko et al., 2017;Stramma et al., 2008).Oxygen is needed by higher trophic marine organisms for respiration, and this circumstance makes oxygen poor waters (known as OMZs) uninhabitable, with critical oxygen levels starting at values of less than 130 μM and typically around 60 μM depending on species (e.g., Rixen et al., 2020).Today, the world's largest OMZ resides in the eastern tropical Pacific where poorly ventilated "shadow zones" (Luyten et al., 1983), associated with stagnant tropical cyclonic gyres (Karstensen et al., 2008), are met by high biological primary production in waters that are nutrient rich due to wind driven coastal and equatorial upwelling.The high biological production and subsequent downward transport of detrital organic matter leads to extensive bacterial consumption of dissolved oxygen due to organic matter remineralization in the water column.
While the future of the EEP OMZ under a warming climate is still uncertain and not consistently projected among Earth System Models (e.g., Busecke et al., 2022), the existence of the EEP OMZ might, in geological terms, be a relatively recent phenomenon.Today, the continental geometry with the closed Central American Seaway (CAS) prevents seawater exchange between the tropical Pacific with relatively low salinity, old and nutrient and carbon rich but oxygen poor waters, and the Atlantic's Caribbean Sea with its saltier, younger and oxygen richer Abstract The tectonic closure of the Central American Seaway (CAS) during the mid-Miocene to mid-Pliocene (∼16-3 Ma BP) is thought of as a key interval for the onset of the present-day tropical Pacific oxygen minimum zone (OMZ).In this study we investigate the impact of an open CAS on the ocean circulation and the OMZ in the tropical Pacific.We perform a series of sensitivity experiments with the Kiel Climate Model, where we vary the CAS sill depth from shallow to deep.We find that the eastern tropical Pacific OMZ was less developed during the period of an open CAS.This is driven mainly by an enhanced eastward subsurface current that facilitated an increased oxygen supply from the western tropical Pacific.In addition, a small decrease in net marine primary production and subsequent weaker export of particulate organic carbon induced less subsurface oxygen consumption in the Eastern Equatorial Pacific.

Plain Language Summary
Oxygen plays a fundamental role for aerobic life and cycling of biogeochemical properties in the ocean.When low rates of oxygen supply combine with high rates of biological oxygen consumption, oxygen minimum zones (OMZs) are generally formed in the ocean.Today the world's largest OMZ is found in the eastern tropical Pacific where the constriction and final closure of the Central American Seaway (CAS) from 16 to 3 Ma BP created specific characteristics of ocean circulation and biogeochemistry.Using simulations with a climate and a marine biogeochemistry model we investigate whether the CAS closure led to the emergence of today's tropical Pacific OMZ.We find that the eastern tropical Pacific OMZ was less developed and oxygen levels were overall higher in the subsurface Eastern Equatorial Pacific (EEP) when the CAS was open.This was driven mainly by an enhanced eastward subsurface current from the northwestern tropical Pacific.In addition, a small decrease in net marine primary production induced less subsurface oxygen consumption from remineralization of particulate organic matter in the EEP.Model simulations using open/closed CAS configurations indicate that this closure significantly affected global climate, through a major reorganization in global ocean circulation, and particularly a strengthened Atlantic Meridional Overturning Circulation (AMOC; Lunt et al., 2008;Maier-Reimer et al., 1990;Mikolajewicz & Crowley, 1997;Schneider & Schmittner, 2006).Based on their model results, Fyke et al. (2015) more recently concluded that the tectonic closure of the CAS led to a stagnation of subsurface waters in the eastern tropical Pacific with impacts on the oceanic carbon reservoir but did not investigate the impact on the EEP OMZ.
Here, we investigate whether the CAS closure led to the emergence of the present-day tropical Pacific OMZ.We also untangle the underlying mechanisms of this emergence by utilizing a set of sensitivity experiments with a global climate model and a marine biogeochemistry model.

Model Description and Experiments
The coupled Kiel Climate Model (KCM) consists of the atmospheric general circulation model ECHAM5 (Roeckner et al., 2003) with a horizontal resolution of T31 (3.75° × 3.75°) and 19 levels, coupled to the ocean-sea ice model NEMO v3.4 (Madec & the NEMO team, 2008) with a horizontal resolution of 2° × 2° and enhanced meridional resolution of 0.5° close to the Equator (known as ORCA2 grid).KCM has been shown to realistically reproduce the present-day climate and its variability (Park et al., 2009).KCM has also been extensively used for paleo-modeling studies, including sensitivity studies of the Pliocene (Krebs et al., 2011;Song et al., 2017;Zhang et al., 2012) and the last two interglacials (Jin et al., 2014;Khon et al., 2010;Salau et al., 2012;Schneider et al., 2010).KCM model output has also previously been used as forcing for the PISCES model to investigate marine biogeochemistry and oceanic oxygen (Segschneider et al., 2018;Xu et al., 2015).
PISCES is a widely used standard marine biogeochemical model that is part of the NEMO ocean model package.
A comprehensive description of PISCES can be found in Aumont et al. (2003).Here the description is restricted to the processes that are most relevant for the simulation of the sources and sinks of oxygen in the ocean.Sources of oceanic oxygen are gas exchange with the atmosphere at the surface and biological production in the euphotic zone.Biological production is simulated by two phytoplankton groups representing nanophytoplankton and diatoms based on the availability of nutrients (where, for the tropics, mostly nitrogen is the limiting nutrient) and light.The sink for oceanic oxygen in the ocean interior is remineralization of organic matter (detritus) from non-living organic material by heterotrophic bacteria.This oxygen-consuming remineralization is simulated over the whole water column with higher rates for warmer waters.There are three components of organic matter (carbon) in PISCES that can be remineralized: non-sinking, semi-labile dissolved organic carbon (DOC), and large and small particulate organic carbon (POC).Here the settling velocity of large detritus is formulated allowing for the ballast effect of calcite and opal shells following Gehlen et al. (2006).This version has been shown to provide good agreement with observations, particularly for oxygen concentration in the EEP (Segschneider et al., 2018).As PISCES uses the same ORCA2 configuration as the NEMO ocean component of KCM, no interpolation of forcing fields is needed.
To simulate the impact of the state (open/closed) of the CAS on ocean circulation and the EEP OMZ, we performed a closed CAS control experiment and a set of sensitivity experiments where the open CAS was implemented at six different depths ranging from 50 to 2,527 m sill depth (Table 1; Figure S1 in Supporting Information S1).The basis for all KCM experiments (control and open CAS series) was a 1,000-years spin-up experiment with modern (closed CAS) bathymetry, modern orbital configuration and preindustrial greenhouse gas (GHG) concentrations in the atmosphere.Eccentricity, obliquity, and precession (ω − 180°) were prescribed at 0.0167, 23.44°, and 102.7°, respectively, atmospheric CO 2 at 286 ppm, CH 4 at 805 ppb, and N 2 O at 276 ppb.GHG concentrations were held constant at pre-industrial levels for all experiments.Starting from the end of the 1,000-years spin-up, a control simulation (CTL) of 5,500 years, utilizing present-day boundary conditions with a closed CAS, was run as a reference experiment.Also initialized from the end of the above 1,000 years closed-CAS spin-up was a 1,500-years spin-up featuring a CAS of 1,212 m depth (described in Song et al. (2017)).From this 1,500-years spin-up, simulations using different CAS sill depths (50,111,197,430,1,212, and 2,500 m; see Table 1) were conducted for 4,000 years each to obtain a close to equilibrium state also for the deep ocean as indicated by the AMOC strength (Figure S2B in Supporting Information S1).To implement the CAS opening we replaced three Full time series of the KCM experiments were then used to force the PISCES model in offline mode using 3d-fields of monthly mean ocean temperature, salinity, and velocity, as well as (2d) sea-ice cover and wind speed on the original NEMO ORCA2 grid.All PISCES experiments were initialized from the end of the transient Holocene simulation, representing modern day pre-industrial conditions as described in Segschneider et al. (2018).

Ocean Circulation Changes in Response to an Open CAS
The analysis of the ocean currents at the region reveals a predominance of Pacific water outflow to the Atlantic Ocean via the open CAS (Figures 1a and 1b).The volume of this throughflow increases with deeper sill depths (Table 1).The CAS throughflow to the Atlantic is thought to represent a geostrophic flow driven by the pressure gradient force resulting from the interoceanic steric height gradient (e.g., Nisancioglu et al., 2003;Schneider & Schmittner, 2006).
Table 1 shows the main components contributing to the Pacific Ocean volume balance as simulated by KCM.For the present-day ocean configuration, the simulated mean transport of the Indonesian Throughflow (ITF; 14.7 Sv), outflow through Bering Strait (1 Sv) and inflow from the Southern Ocean to the Pacific across 30°S (15.7 Sv) are in a good agreement with observations (Lumpkin & Speer, 2007).In the presence of an open CAS, the CAS outflow results in weaker ITF and Bering Strait throughflow, on one hand, and stronger Southern Ocean water inflow to the Pacific, on the other (Table 1).
In our simulations, an open CAS results in an intensification of the eastward subsurface flow in the north-eastern tropical Pacific (Figure 1e).When reaching the Panamanian Gateway (PG), this subsurface current continues into the Caribbean Sea and further extends partly toward the North Atlantic through the Florida Strait and partly toward the South Atlantic via the Antilles Gateway between the Greater Antilles and South America (Figure 1e).The latter current causes a weakening of the northward North Brazil Current and a strengthening of the southward Brazil Current, consistent with earlier modeling studies (e.g., Butzin et al., 2011;Mikolajewicz & Crowley, 1997;Prange & Schulz, 2004;Sentman et al., 2018;Sepulchre et al., 2014).Therefore, less heat is being transported northwards in the North Brazil current than under modern conditions, leading to reduced oceanic heat transport to the North Atlantic (Prange & Schulz, 2004).(1.1 Sv) (Table 1).This westward transport consists of an inflow of Atlantic surface waters driven by trade winds and an inflow of Atlantic subsurface waters driven by the density gradient within ∼300-800 m depth (Figure 1a).
For the deep sill experiment (CAS2500), there is an additional westward inflow of Atlantic deep waters through the CAS into the Pacific below ∼1,200 m of about 2 Sv (Table 1; Figure 1b).Nisancioglu et al. (2003) proposed that CAS depths deeper than the level of North Atlantic Deep Water (NADW) could promote an inflow of NADW into the Pacific.In our simulations this effect is rather moderate (Table 1), and we assume that the shallow sill depth of the Lesser Antilles Arc (∼870 m sill depth in the model bathymetry) may place some restrictions on the strength of this deep Atlantic inflow.
Outflow of relatively fresh Pacific waters through the CAS results in a decrease of North Atlantic surface seawater salinity (Figure 2b) and hence density, with subsequent AMOC slowdown (Table 1; Figure S2 in Supporting Information S1), a result in agreement with other modeling studies (e.g., Sepulchre et al., 2014;Zhang et al., 2012).
The simulated AMOC strength decreases nonlinearly with sill depth with only moderate further weakening for sill depths deeper than 200 m as the current velocity through the CAS is intensified more non-linearly when the opening is shallow (Table 1).
The northward cross-equatorial heat transport in the Atlantic weakens in response to an open CAS creating cooler (warmer) conditions in the North (South) Atlantic.A corresponding pattern can be seen at a global scale (Figure 2a) and is commonly referred to as the interhemispheric "seesaw effect" associated with AMOC changes in climate models (e.g., Zhang et al., 2012).Thus, an open CAS may indirectly (through the AMOC-induced SST anomalies) result in a weaker cross-equatorial SST gradient in the eastern equatorial Pacific (EEP) and associated southward wind anomaly (Figure 1c; see also Song et al., 2017;Zhang et al., 2012).The latter causes water masses to flow eastward (westward) in the southern (northern) tropics leading to strengthening of the South Equatorial Countercurrent and weakening of the North Equatorial Countercurrent (Figure 1d).
The water mass transport through the Straits of Florida is thought to be noticeably (∼50%) sourced by the upper branch of the AMOC (e.g., Johns et al., 2002;Schmitz & Richardson, 1991) and is shown here to be closely coupled to the AMOC strength (Figure S3 in Supporting Information S1 ).An open CAS causes the inflow of Pacific waters into the Caribbean, where they mainly contribute to the Florida Current and to a lesser extent to the southward Brazil Current via the Antilles Gateway (Figure 1f).At the same time, water supply to the Florida Current from the Antilles Gateway drops because it is partly replaced by the Pacific water transport through an open CAS (Figure 1f).As a result, the total transport via the Florida Strait (as well as the AMOC intensity) is also reduced.Our analysis shows that the contribution of the CAS (Antilles) transport to the Florida Current nonlinearly increases (decreases) with deeper sill depths (Figure 1f), a result qualitatively consistent with that of Sepulchre et al. (2014).

Changes in Oxygen Concentration
The simulated mean oxygen concentration is first compared against the World Ocean Atlas observations (Garcia et al., 2013) to evaluate the model's ability to reproduce the mean state of the present-day oxygen concentration (Figures 2c and 2d).Here we show average concentrations over the 100-870 m depth range, which in our simulation, based on a threshold of 30 μM, covers more than 90% of the total hypoxic volume in the EEP, on a common 1° × 1° grid (model data interpolated to WOA13 grid and depths).In general, the KCM-forced PISCES shows a reasonable ability to reproduce the global pattern of the present-day oxygen fields, including the predominance of low-oxygenated areas in the tropical oceans, despite an underestimation of the westward extension of oxygen-depleted waters in the tropical Pacific.
An open CAS and strongly modified ocean circulation have a significant impact on the state of the tropical Pacific OMZ.An open CAS is associated with an overall increase in subsurface water oxygen concentrations in the Pacific with the strongest effect simulated in the EEP north of the equator (Figure 2e).We find that the volume of hypoxic water (<30 μmol L −1 ) in the EEP decreases considerably (by a factor of 7) when the CAS sill is set at 100 m depth (Table 1).When the CAS sill is deeper than this, no hypoxia is developed in the water column (Table 1; Figure S4 in Supporting Information S1).The opposite result is found for the midlatitude North Atlantic where the model shows a tendency toward subsurface oxygen reduction (Figure 2e) despite an increase in oxygen saturation due to colder surface seawater (Figure 2a).Therefore, such a tendency is mainly related to poorer ventilation in the North Atlantic in relation to a weakened AMOC, on one hand, and an increase in oxygen consumption in the midlatitude North Atlantic (Figure 3a) due to a higher biological production in this region, on the other.
To quantify which processes are responsible for a better oxygenated EEP with an open CAS, we calculate the sources and sinks of oxygen in a control volume in the EEP (150°W-75°W; 10°S-10°N, depth range 100-870 m; area indicated by the gray rectangle in Figure 1c).First, the source, namely the convergence of oxygen transport (net transport) was calculated at the western and eastern (southern and northern) boundaries.Similarly, vertical convergence is computed as the difference between the area-integrated vertical oxygen flux at the top (100 m) and bottom (870 m) of the control volume.It should be noted that the northern boundary of the EEP box incorporates the PG (Figure 1c) to account for oxygen transport through the open CAS in our analysis.In a second step the For the present-day closed CAS simulation, the zonal convergence of oxygen transport (net zonal transport) is the dominant source of oxygen for the EEP control volume, which is mainly balanced by the upward outflow at the top boundary (upwelling into the surface ocean) and partly by oxygen consumption from respiration of organic matter (Figure 3b; see also Ito & Deutsch, 2013).In the presence of an open CAS, the horizontal influx of oxygen to the EEP is balanced by its upwelling from the EEP toward the surface, oxygen consumption and newly established oxygen outflux toward the Atlantic for CAS sill depths deeper than 100 m (Figure 3b).
Our simulations show that the net zonal subsurface oxygen transport into the EEP significantly increases in response to an open CAS (Figure 3b).This due to an intensification of the eastward subsurface flow in the northeastern tropical Pacific (Figures 1e and 3c-3f) that improves ventilation and transports oxygen-rich waters toward the eastern tropical Pacific, thereby, increasing oxygen concentration in the EEP (Table 1; Figure 2e).
Simulated changes in the ocean circulation resulting in a ventilation improvement of the EEP are shown in Figures 3c-3f for two depth levels in the Pacific.For the shallower depth level (∼300-400 m), the present-day Mindanao Current in the western Pacific splits into two branches: one feeds the ITF and one moves toward the equatorial Pacific (Figure 3c).In response to an open CAS, the former of these branches is rerouted toward the equatorial Pacific forming the eastward current in the northeastern tropical Pacific which flows further into the Caribbean (Figure 3d).For the deeper level (∼700-800 m), the underlying mechanism is related to the redirection of the ocean currents in the Southern Pacific.The part of the current flowing southward near the eastern coast of Australia at ∼20°S (Figure 3e) under present-day climate, is rerouted equatorward.Then it feeds the transport across the Solomon Sea and further flows to the northwestern tropical Pacific from which it extends toward the northeastern tropical Pacific (Figure 3f).
In addition, marine net primary production weakens overall in the Pacific (not shown) due to a redistribution of surface water nutrients.Figure 2f shows that there is a nutrient depletion (enrichment) of surface waters in the tropical Pacific (North Atlantic) under an open CAS.A similar effect has also been observed in Schneider and Schmittner (2006) and can be explained by stronger advection of nutrient-rich subsurface waters from the EEP to the North Atlantic.For the tropical Pacific this, in turn, leads to a weaker POC export toward the ocean interior, and, therefore, to lower subsurface oxygen consumption in this region (Figure 3a).According to our results, mainly the increased eastward subsurface oxygen transport and to a lesser extent the reduced biological consumption of oxygen contribute to the open versus closed CAS oxygen enrichment in the EEP.

Summary and Conclusions
In this study, we analyze the potential effects of different sill depths of the PG on ocean circulation and marine biogeochemistry using global climate and biogeochemistry models.Particularly, we investigate, for the first time, whether changes in the large-scale ocean circulation caused by the PG closure could have led to the emergence of the present-day tropical Pacific OMZ.
We perform a series of sensitivity experiments with the KCM where land grid cells corresponding to the Isthmus of Panama are replaced by ocean grid cells allowing sea water exchange between the Pacific and Atlantic Oceans.According to our experiments, the CAS closure during the Miocene/Pliocene has led to the termination of relatively low salinity water supply from the tropical Pacific to the northern North Atlantic and therefore to an intensification of the AMOC, a result consistent with numerous modeling studies.
We find that an open CAS is associated with an overall increase of oxygen concentrations in the subsurface eastern tropical Pacific waters.This is because the open CAS is associated with the enhanced eastward subsurface current in the northeastern tropical Pacific that improves ventilation and transports oxygen-rich waters from the western tropical Pacific toward the EEP.We conclude that the increased west-to-east subsurface oxygen transport is the main driver for the open CAS-induced oxygenation in the EEP while the contribution of biological consumption of oxygen is rather minor according to our simulations.
The improved ventilation of the EEP is a result of an open CAS-induced redirection of the ocean currents in the North and South Pacific.The Mindanao Current branch that contributes to the present-day ITF transport is rerouted toward the equatorial Pacific in response to an open CAS.This leads to a weaker total ITF transport and a stronger eastward subsurface current that improves ventilation in the northeastern tropical Pacific for the shallower depths (∼300-400 m).For the deeper level (∼700-800 m), it was found that the current flowing southward along the eastern coast of Australia is partly redirected equatorward, which leads to a stronger inflow of the Southern Ocean waters into the Pacific.This current feeds the transport across the Solomon Sea and further flows to the northwestern tropical Pacific from which it extends toward the northeastern tropical Pacific, thereby improving ventilation in the EEP.
Our study is meant as a sensitivity assessment of the emergence of the EEP OMZ in relation to the CAS sill shoaling, and we do not attempt to provide a fully realistic representation of the Miocene/Pliocene climate evolution.
Further extensive studies are needed to investigate the Miocene/Pliocene evolution in the ocean circulation and the EEP OMZ under more representative boundary conditions and with other independent climate and biogeochemistry models.

•
An open Central American Seaway promotes higher oxygen concentrations in the subsurface Pacific waters, with strongest changes in the EEP • An increased eastward subsurface flow in the northeastern tropical Pacific transports young, oxygen-rich waters in from the western Pacific • The improved EEP ventilation is a result of the ocean current redirection in the North and South Pacific during open Central American Seaway Supporting Information: Supporting Information may be found in the online version of this article.but nutrient and carbon poorer waters.The Isthmus of Panama was formed during mid-Miocene to mid-Pliocene (approximately 16-3 Ma BP) leading to the closure of the CAS (e.g., O'Dea et al., 2016).
-grid land cells representing the Isthmus of Panama by corresponding ocean cells (FigureS1in Supporting Information S1).The total width of the implemented CAS (three cells on the 2° × 2° grid) is approximately 660 km along ∼8°N.
Also, a westward transport through the open CAS is simulated but only makes up, on average, ∼7% of interoceanic exchange between the Pacific and the Atlantic when sill depths range between 50 m (0.1 Sv) to 1

Figure 1 .
Figure 1.Mean northward velocity (cm/s) across the Panamanian Gateway for 1,200 m (a) and 2,500 m (b) CAS sill depths.Anomalies (CAS1200 minus CTL) in wind stress (c), ocean current at the surface (d) and at 270 m depth (e); (f) Water flow contributions to the Florida Strait current (black solid) from the Antilles (blue) and Panamanian (red solid) Gateways as a function of the CAS sill depth; Contribution to the southward Brazil current from the CAS is shown in (d) by red dashed line.Gray lines in (c) indicate the EEP area and the Panamanian, Antilles, and Florida Gateways.In panels (a and b) red colors indicate transport toward the Caribbean and blue toward the tropical Pacific.

Figure 3 .
Figure 3. (a) Anomaly (CAS1200 minus CTL) in oxygen consumption by respiration of particulate organic carbon (POC), calculated as POC flux attenuation between 100 and 870 m depth.(b) Mean oxygen consumption and net oxygen transport for the EEP for different CAS sill depths.Mean ocean current for: CTL experiment at 365 m (c) and 730 m (e) depth; CAS1200 experiment at 365 m (d) and 730 m (f) depth.

Table 1
Note.The rightmost column represents hypoxic (<30 μmol L −1 ) water volume in the eastern equatorial Pacific [150°W-75°W; 10°S-10°N] averaged over the last 100 model years.The AMOC strength (defined as its maximum at 30 N) and water transports are averaged over the last 1,000 model years.Shown are climatological means of net water transport for the preindustrial simulation (CTL) as well as respective anomalies for open CAS experiments (CAS minus CTL).Positive (negative) values of transport correspond to water inflow to (outflow from) the Pacific.List of KCM Experiments Used in This Study Along With Results of Simulations