Equatorial Undercurrent Influence on Surface Seawater δ 18 O Values in the Galápagos

Stable isotopes of oxygen (δ 18 O) in seawater reflect the combined influences of ocean circulation and atmospheric moisture balance. However, it is difficult to disentangle disparate ocean and atmosphere influences on modern seawater δ 18 O values, partly because continuous time series of seawater δ 18 O are rare. Here we present a nearly nine-year, continuous record of seawater δ 18 O values from the Galápagos. Seawater δ 18 O values faithfully track sea surface salinity and salinity along the equator at 50 m depth. Zonal current velocity within the Equatorial Undercurrent (EUC), directly west of the Galápagos, is strongly correlated with Galápagos surface seawater δ 18 O values with a 1-month lag. Reconstructions of Galápagos seawater δ 18 O values could thus provide a window into past variations in the strength of the EUC, an important influence on large-scale tropical Pacific climate.


Introduction
Stable oxygen isotopes in seawater have a long history of utility in defining large-scale water masses and water mass mixing, sea ice melt, and continental runoff to the ocean (i.e., Akhoudas et al., 2021;Biddle et al., 2019;Frew et al., 2000;LeGrande & Schmidt, 2006;Reyes-Macaya et al., 2022;Schlosser et al., 2002). Seawater δ 18 O (hereafter δ 18 O sw ) values also preserve information about precipitation and evaporation over the ocean, which assert a large control on mixed layer δ 18 O sw values. As such, δ 18 O sw values have a strong, linear relationship with salinity, given that freshwater fluxes influence both variables in a similar manner (Conroy et al., 2017;Reed et al., 2022;Thompson et al., 2022). However, δ 18 O sw values reflect more than salinity, as precipitation δ 18 O values vary spatially and temporally, and because atmospheric conditions alter the degree of isotope fractionation during evaporation (Craig & Gordon, 1965). Moreover, advection and upwelling transport waters with varying δ 18 O sw values, complicating interpretation of δ 18 O sw as a simple hydroclimatic indicator in some regions of the ocean. δ 18 O sw information can be extracted from the measured δ 18 O values of marine carbonate climate archives, such as fossil coral skeletons and foraminifera tests. However, in these materials, carbonate δ 18 O values also vary with temperature, given the temperature-dependence of equilibrium stable isotope fractionation. If the temperature component can be isolated, such as with paired measurements of the temperature proxy Sr/Ca in corals (McCulloch et al., 1994;Walter et al., 2022), past δ 18 O sw values can be calculated and used to interpret past salinity, hydroclimate, and ocean dynamics. In some regions, where δ 18 O sw controls a larger portion of carbonate δ 18 O variance, δ 18 O sw and salinity can also be directly inferred from carbonate δ 18 O measurements .
A key challenge in such efforts is the limited understanding of how δ 18 O sw values vary temporally. The majority of modern δ 18 O sw measurements come from sporadic, geographically limited research cruises. The resulting δ 18 O sw data from the last several decades has provided a useful temporal average of spatial δ 18 O sw variability across the global ocean (Figure 1a), but typically cannot define even seasonal δ 18 O sw variability. Few continuous time series of δ 18 O sw from single locations exist, given the challenges in maintaining such seawater collections over long periods of time (Conroy et al., 2017;DeLong et al., 2022). Thus, we still have little idea of how δ 18 O sw varies seasonally and interannually, with our current understanding coming mainly from isotope-enabled ocean model simulations. Models have revealed substantial temporal variability in both δ 18 O sw and the δ 18 O sw -salinity relationship and the varied influence of atmospheric moisture balance, advection, vertical mixing, and diffusion on these values (Stevenson et al., 2018).
Here we present a 9-year, weekly-resolved record of δ 18 O sw from the Galápagos Archipelago in the eastern equatorial Pacific (Figure 1a). The Galápagos are located in a region with complex ocean circulation and strong interannual variability in atmospheric moisture balance due to the El Niño-Southern Oscillation (ENSO). Numerous δ 18 O-based paleoceanographic records from fossil corals and foraminifera have been developed here, which hold critical information on past changes in climate (Dunbar et al., 1994;Koutavas et al., 2006;Lea et al., 2000;Rustic et al., 2015;Shen et al., 1992). Previous research has revealed a relatively weak imprint of atmospheric moisture balance on Galápagos δ 18 O sw (Conroy et al., 2017;Russon et al., 2013;Stevenson et al., 2018;Wellington et al., 1996). Here we more comprehensively assess temporal variability in Galápagos δ 18 O sw values and potential atmospheric and oceanic drivers of this temporal variability using one of the longest available δ 18 O sw time series to date.

Study Site
Galápagos, Ecuador, situated along the equator in the eastern tropical Pacific Ocean (Figure 1a), is under the influence of large-scale atmospheric subsidence and the southeasterly trade winds associated with the zonal Pacific Walker Circulation. The islands are also positioned within the 'cold tongue' of cooler mean sea surface temperatures (SST) in the eastern to central equatorial Pacific. Galápagos climate is arid at sea level, but higher relative humidity and precipitation rates occur at the higher island elevations. The Galápagos seasonal cycle consists of a warm-wet season from December to May and a cool-dry season from June to November (Trueman & d'Ozouville, 2010). Although the Galápagos are consistently south of the Intertropical Convergence Zone (ITCZ), the southward shift of the ITCZ in boreal winter leads to weaker trade winds, reduced upwelling, warmer SST and air temperatures, and enhanced precipitation.
Interannual climate variability associated with ENSO is prominent in the Galápagos. La Niña periods have stronger trade winds and upwelling, cooler SST, and little to no precipitation at sea level in the Galápagos. During El Niño events, weakened trade winds, a deeper thermocline, and warm SST lead to increased precipitation that can be an order of magnitude higher than average (Martin et al., 2018). These conditions are especially pronounced during eastern Pacific -style El Niño events (e.g., Capotondi et al., 2015).
Ocean circulation in and around the Galápagos archipelago is complex, with multiple ocean currents converging in the region (Kessler, 2006;Liu et al., 2014). At the surface, the westward South Equatorial Current (SEC) flows across the northern and southern boundaries of the Galápagos platform. The southwestward Panama Current strongly influences the surface waters of the northeastern archipelago, carrying relatively warm and low salinity waters into the area. The northwestward Peru Coastal Current, or Humboldt Current, transports cooler, upwelled, nutrient-rich waters toward the islands and feeds the southern lobe of the SEC. The relatively cool and saline Equatorial Undercurrent (EUC), or Cromwell Current, flowing from west to east 50-100 m below the sea surface between 1°N and 2°S, intersects the Galápagos and shoals along the western shores of the westernmost islands of Fernandina and Isabela, leading to much cooler SST (Karnauskas et al., 2010). The EUC then splits into northern and southern branches around the islands . These upwelled EUC waters are nutrient-rich, making the EUC a key influence on the Galápagos marine ecosystem (Palacios, 2004 The long-term seawater sampling site of this study is located on the southeastern shore of Santa Cruz Island (Figure 1c), where sampling began in October 2012 (Conroy et al., 2017). The current data set ends in May 2021 and is mostly continuous, apart from one hiatus from January to May 2018. Seawater samples were taken once a week at 6:00 a.m. at La Ratonera beach (0.7437°S, 90.3031°W), off a rocky area of the beach, where wave action is generally intense. Spatial seawater sampling in 2012 and 2015 indicated the δ 18 O sw and salinity values at this site were similar to seawater measured across the platform at the time of sampling (Conroy et al., 2017).
Along with seawater samples, precipitation samples for isotope analysis were collected daily from October 2012 through January 2021 at the Charles Darwin Research Station weather station located ∼100 m from La Ratonera Beach. Samples were collected daily, on days with rain, in a separatory funnel containing mineral oil to hinder water evaporation. An additional seven samples of Santa Cruz groundwater from trips in 2012, 2015, and 2021 were measured for isotopic analysis (Table S1 in Supporting Information S1). These data are assessed along with four samples of previously published Santa Cruz groundwater stable isotope data from Warrier et al. (2012).
The main salinity and current datasets assessed with respect to the new Galápagos data are from the GLORYS12 Reanalysis from the Copernicus Marine Environment Monitoring Service (Jean-Michel et al., 2021). This eddy-resolving, 1/12° data set spans the majority of the sampling period, ending in May 2020. Other datasets analyzed include WHOI OAflux v3 evaporation rates over the ocean (Yu & Weller, 2007), PEODAS 20°C isotherm depth (Yin et al., 2011), ERA5 reanalysis sea level pressure (SLP, Hersbach et al., 2020), UK Met Office Hadley Centre EN4.2.2 salinity (Good et al., 2013), and the NCEP Global Ocean Data Assimilation System (GODAS) zonal and meridional currents (Behringer & Xue, 2004). Charles Darwin Research Station weather station precipitation data are also assessed, as is precipitation at the Bellavista, Santa Cruz, station at 191 masl. We also assessed daily and hourly sea level measurements taken at Puerto Ayora (Caldwell et al., 2015), and sea surface temperature measurements taken daily by station personnel at the seawater sampling site.

Sample Measurements
Precipitation and groundwater samples were stored in 3.5 ml crimp-top glass vials, and seawater and groundwater samples were stored in 60 mL amber bottles capped and sealed with parafilm. Samples from April 2016 onward were analyzed for both δ 18 O and δ 2 H at the University of Illinois at Urbana-Champaign on a Picarro L-2140i water isotope analyzer. Previous samples, reported in Conroy et al. (2017), were measured on a Picarro L-2130i water isotope analyzer. Samples were calibrated with three internal standards, which were in turn calibrated to Vienna Standard Mean Ocean Water 2 (VSMOW2), Greenland Ice Sheet Project (GISP) and Standard Light Antarctic Precipitation 2 (SLAP2) standards. We applied corrections as discussed in Van Geldern and Barth, (2012) to account for memory effects and instrument drift. Data are expressed in standard δ-notation. Total error is <0.1‰ for δ 18 O and <0.8‰ for δ 2 H. Although δ 2 H values are not discussed in this work, these additional values were used to identify and exclude samples affected by evaporation after collection (see results).
From October 2012-September 2020, seawater salinity was measured on a Thermo Scientific Orion Star A212 benchtop conductivity meter. The meter was calibrated every 10 samples with 50.0 mS/cm certified conductivity solution. Twenty mL samples were measured 5 times at a constant temperature of 25.0°C. Salinity values in PSU are reported as the average of the five measurements. Measured standard deviations for individual samples are <0.1 PSU and duplicate measurements on samples indicate a mean absolute difference of 0.1 PSU. From October 2020 onward, salinity was measured on a Guildline Autosal 8400B, using IAPSO standard seawater (34.994 PSU) as a daily standard. Total precision is < ± 0.002 PSU.

Results
A total of 417 seawater samples from October 2012 to May 2021 are analyzed, including data through April 2016 previously published in Conroy et al. (2017). The δ 18 O sw values range from −2.03 to 0.71‰, with a mean of 0.06 ± 0.36‰ (1s, used throughout). Salinity values range from 9.16 PSU to 35.51 PSU, with a mean of 32.05 ± 4.06 PSU ( Figure S1 in Supporting Information S1).
Aperiodic freshwater pulses in the seawater time series are identified using salinity outliers, calculated as 1.5 times the interquartile range below the 25th percentile ( Figure S2a in Supporting Information S1). With this method, salinity values <29 PSU are considered unrepresentative of open ocean salinity and δ 18 O sw . NASA Aquarius v4 daily sea surface salinity measurements from 2011 to 2015 also reveal a minimum value of 29 PSU for the 0.5° × 0.5° grid cell that encompasses our sample site ( Figure S2b in Supporting Information S1, Melnichenko et al., 2016). We identify 51 (12% of the data) of these freshwater pulses. They occur in every month of the year, with more pulses in March (N = 8) and December (N = 6) ( Figure S3 in Supporting Information S1). These pulses are discussed further in the supporting online material, but are excluded from monthly averaged δ 18 O sw values and subsequent analyses that aim to consider the drivers of δ 18 Figure S4 in Supporting Information S1). The linear fit of monthly δ 18 O sw and salinity values, excluding freshwater pulses, produces a slope of 0.13 ± 0.02‰/PSU and an intercept of −4.03 ± 0.49‰ (R 2 = 0.74). Temporal variability in the linear δ 18 O sw -salinity relationship is discussed further in the supporting online material. 610 precipitation samples from October 2012 to March 2021, which include previously published data through 27 March 2016 in Martin et al. (2018) are also reported here ( Figure S5 in Supporting Information S1). Outliers that represent evaporated samples were identified using calculated d-excess values (δ 2 H-8*δ 18 O). Ten samples were excluded from analysis as their d-excess values were lower than 3 sigma below the median precipitation d-excess value (d-excess<1.44‰). The meteoric water line slope (6.75 ± 0.18) and intercept (d-excess value, 7.67 ± 0.25) for this longer data set are slightly higher than those reported in Martin et al. (2018). Precipitation δ 18 O values range from −5.57 to 1.09‰ (mean −1.10 ± 0.88‰). Monthly, amount-weighted averages (N = 94) show a seasonal cycle, with lower δ 18 O values from February to June (−1.33‰ to −2.95‰), when precipitation rates are higher, and higher δ 18 O values from July to January (−0.65‰ to −1.05‰), when precipitation rates are lower ( Figure S5b in Supporting Information S1).  (Table S1 in Supporting Information S1). The groundwater δ 18 O and δ 2 H values fall on the precipitation meteoric water line ( Figure S5a in Supporting Information S1).

Salinity and Atmospheric Controls on δ 18 O sw Values
Galápagos δ 18 O sw is highly correlated not only with measured salinity on the paired water samples (Table S2 in Supporting Information S1, Figure 2a), but with GLORYS12 reanalysis salinity values across the surface eastern-central equatorial cold tongue region (5 m depth) and in the subsurface at 50 m depth (Figures 2b and 2c). We also observe similar correlation patterns using the EN4.2.2 objective analysis salinity product from October 2012 to May 2021 (e.g., extending a year beyond GLORYS12 reanalysis, Figure S6 in Supporting Information S1). Thus, δ 18 O sw is a proxy for regional as well as local surface and sub-surface salinity variability.
Although δ 18 O sw and salinity are often interpreted in the context of hydroclimate, the influence of atmospheric moisture balance on Galápagos δ 18 O sw values is relatively weak ( Figure S7, Table S2 in Supporting Information S1). Higher monthly precipitation and low monthly precipitation δ 18 O at sea level (Puerto Ayora) do not lead to lower δ 18 O sw values. We also assessed δ 18 O sw relative to monthly precipitation measured at 191 masl, in the town of Bellavista. Precipitation δ 18 O values are not only higher in the highlands but also decrease with elevation on Santa Cruz due to orographic rainout-related effects (Martin et al., 2018). However, the relationship between Bellavista precipitation and δ 18 O sw is not significant, up to a lag of 6 months ( Table S3 in Supporting Information S1). We do find a significant, negative relationship between monthly Bellavista precipitation anomalies (after removing the long-term monthly means) and δ 18 O sw at a lag of 5 months (Table S3 in Supporting Information S1). This lagged relationship suggests that groundwater flow from the highlands and subsequent groundwater seepage following periods of higher precipitation rates may play a small role in reducing δ 18 O sw values, but overall, the influence of local precipitation on monthly δ 18 O sw values is weak. A significant, negative correlation is observed with regional ocean evaporation (Table S2 in Supporting Information S1). This direction of the relationship is opposite to the expected relationship (i.e., more evaporation will increase δ 18 O sw ). This suggests another unknown factor is likely influencing both evaporation and δ 18 O sw.

Influence of Advection and the EUC on δ 18 O sw Values
The strong correlations between Galápagos δ 18 O sw and salinity on the equator at 50 m depth, west of the Galápagos, suggests that the higher salinity waters that covary with high δ 18 O sw waters near Santa Cruz may be transported to the site by the EUC. Along with high salinity, waters of the EUC also have relatively high δ 18 O sw values. In the central equatorial Pacific, mean EUC δ 18 O sw values range from 0.41 to 0.51‰ (Conroy et al., 2014). Closer to the Galápagos, data from the NASA GISS δ 18 O sw database reveal a mean δ 18 O sw value of 0.37 ± 0.09‰ from 5 measurements between 95° and 86°W, along the equator, between 50 and 100 m depth (Schmidt et al., 1999).
Along with salinity, an EUC signature in Galápagos δ 18 O sw should also be reflected in correlations with zonal current speed between 0 and 1°S just west of the Galápagos (Johnson et al., 2002;Karnauskas et al., 2010). At a depth of 50 m, west of the archipelago, which approximates the location of the EUC, we find strong, positive correlations between Galápagos δ 18 O sw values and zonal current speed when δ 18 O sw lags zonal current speed by 1 month (Figure 3). That is, higher Galápagos δ 18 O sw values coincide with stronger eastward current speed in the prior month (Figure 3). We also observe a strong, positive correlation with zonal current velocity with depth at 93°W, peaking at 0.67°S-0.75°, 47-55 m depth (Figure 4). Periods of higher δ 18 O sw coincide with higher eastward current velocity when δ 18 O sw lags zonal velocity by 1 month (r = 0.52, p < 0.001, N = 92). A similar relationship is found when using GODAS zonal velocity, which extends to May 2021 ( Figure S8 in Supporting Information S1).
The structure of the correlations in Figure 4 reflects the structure of the EUC itself; however, correlations peak to the south and closer to the surface than the long-term mean zonal current speed (Figure 4b, black contours). This offset provides insight into the conditions that lead the EUC to imprint on Galápagos δ 18 O sw values. When the EUC is faster, closer to the surface and located slightly further south than normal, a greater proportion of EUC waters reaches the southern coast of Santa Cruz. In sum, higher monthly Galápagos δ 18 O sw values are due to a greater proportion of EUC waters in the mixed layer during periods of stronger, more southerly, and shallow EUC flow, and Galápagos δ 18 O sw values appear to be an excellent proxy for regional EUC strength.
It is still unclear how the EUC contributes to sources of upwelled waters on the Galápagos platform. The EUC shoals as it approaches the topographic barrier of the Galápagos, and bifurcates into a northern and southern current around the archipelago . This bifurcation and shoaling is also reflected in the change in sign of the correlation coefficients between Galápagos δ 18 O sw and meridional current speed, both at 50 and 5 m depth (Figure 3). One high-resolution mean current analysis reveals both the southern and northern EUC branches may flow onto the eastern platform at 40-160 m depth (Karnauskas et al., 2010). High-resolution ocean modeling of surface flow in and around the Galápagos archipelago by Liu et al. (2014) also show surface eastward flow onto the Galápagos platform originating from the EUC to the west, in this case even reaching the southern Santa Cruz coast. The 1-month lag in the EUC-Galápagos δ 18 O sw relationships suggests a transit time of this order between the EUC just west of the archipelago and the measurement site on the coast of Santa Cruz. Higher EUC velocity has also been found to lead higher salinity by 30 days south of the equator around the Galápagos (Rudnick et al., 2021).
The relationship between the EUC and Galápagos δ 18 O sw is highly seasonal; When a 1-month lag is applied, the relationship between climatological mean monthly δ 18 O sw and the EUC is strong (r = 0.77, p = 0.004, N = 12). Zonal current speed at 93°W, 0.67°S-0.75°, 47-55 m depth, is highest from March through June in the time period analyzed, consistent with observations of an overall stronger EUC in the boreal spring. δ 18 O sw is highest in April through August ( Figure S4 in Supporting Information S1). The one-month lagged correlation between δ 18 O sw and the EUC timeseries at 93°W is also retained (although weaker) with the removal of the long-term monthly mean values from each time series, indicating that this relationship is interannual as well as seasonal (r = 0.30, p = 0.006, N = 91, Figure S9a in Supporting Information S1). Overall, the EUC is observed to be weaker during El Niño events, although at 93°W, it is not highly correlated with surface temperature anomalies (Johnson et al., 2002;Rudnick et al., 2021). During La Niña periods, as indicated by NIÑO1+2 Index values <0°C, the relationship between anomalies of δ 18 O sw and the 93°W EUC timeseries is stronger ( Figure S9b in Supporting Information S1).

Conclusions and Implications
Many paleoceanographic records from the Galápagos have been developed from fossil coral and foraminifera δ 18 O values over the last three decades (Dunbar et al., 1994;Koutavas et al., 2006;Lea et al., 2000;Rustic et al., 2015;Shen et al., 1992). These records are a mixed contribution of δ 18 O sw values and temperature, with evaluations to date showing the dominance of temperature on coral carbonate δ 18 O values in this region One month lag applied to δ 18 O sw data. Dashed lines represent significant correlations at the 95% confidence level. Black contours represent the mean zonal current speed, 1993-2020. 8 of 10 . However, more recent work pairing foraminifera Mg/Ca and δ 18 O measurements and coral Sr/Ca and δ 18 O measurements from around the archipelago offer the potential for future δ 18 O sw and salinity reconstructions (Cheung et al., 2021;Jimenez et al., 2018;Rustic et al., 2015). Our work provides an important new context for interpreting such data. Although atmospheric moisture balance is the common focus of δ 18 O sw interpretations Walter et al., 2022), our findings show stronger potential for Galápagos δ 18 O sw values to be used to reconstruct the strength of the EUC. Such a reconstruction would be novel and useful, given the critical role of the EUC in supporting the zonal gradient of sea surface temperature across the tropical Pacific and contributing to regional productivity. The EUC has also gained considerable importance as a climate model diagnostic, as it is not always well-simulated, with cascading implications for the tropical Pacific climate system . Ultimately then, this work suggests Galápagos paleo-δ 18 O sw reconstructions could be developed to benchmark the EUC in paleoclimate model simulations.

Data Availability Statement
δ 18 O sw and salinity data are archived on the Seawater Oxygen Isotope Community Database (Conroy, 2023). Precipitation and groundwater isotope data are available through the Water Isotopes Database (https://wateriso. utah.edu/waterisotopes/pages/spatial_db/SPATIAL_DB.html).