Volume 25, Issue 1 e2023GC011246
Research Article
Open Access

Hydrothermal Seepage of Altered Crustal Formation Water Seaward of the Middle America Trench, Offshore Costa Rica

Patrice K. F. Parsons

Patrice K. F. Parsons

Earth and Planetary Sciences Department, University of California, Santa Cruz, CA, USA

Contribution: Methodology, Formal analysis, ​Investigation, Writing - original draft

Search for more papers by this author
C. Geoffrey Wheat

Corresponding Author

C. Geoffrey Wheat

College of Fisheries and Ocean Sciences, University of Alaska Fairbanks, Moss Landing, CA, USA

Correspondence to:

C. G. Wheat,

[email protected]

Contribution: Conceptualization, Methodology, Formal analysis, ​Investigation, Data curation, Writing - review & editing, Supervision, Project administration, Funding acquisition

Search for more papers by this author
Andrew T. Fisher

Andrew T. Fisher

Earth and Planetary Sciences Department, University of California, Santa Cruz, CA, USA

Contribution: Conceptualization, ​Investigation, Writing - review & editing, Supervision, Funding acquisition

Search for more papers by this author
Eli A. Silver

Eli A. Silver

Earth and Planetary Sciences Department, University of California, Santa Cruz, CA, USA

Contribution: Formal analysis, ​Investigation, Writing - review & editing, Funding acquisition

Search for more papers by this author
Michael Hutnak

Michael Hutnak

Earth and Planetary Sciences Department, University of California, Santa Cruz, CA, USA

Contribution: Formal analysis, ​Investigation, Writing - review & editing, Visualization

Search for more papers by this author
First published: 18 January 2024

Abstract

Chemical compositions of sediment pore waters are presented from 13 piston and gravity cores that were collected on ∼24 Ma crust of the Cocos Plate seaward of the Middle America Trench and near the onset of crustal faulting from subduction. Cores were collected mainly within a 1.75 km2 area overlying a buried basement topographic high that supports an elevated heat flux, consistent with seawater transport within the upper volcanic crust. Systematic variations in pore water chemical profiles indicate upward seepage speeds (up to 1.7 cm yr−1 providing a net flux of 0.1 L s−1), constrain the chemical composition of the formation water within the underlying upper basaltic basement, and elucidate diagenetic reactions in the sediment. Relative to seawater, formation water has an elevated temperature (70–80°C) and concentrations or values of Ca, chlorinity, Sr, Ba, Li, Fe, Mn, Si, Cs, D/H, and Mo, and lower concentrations or values of Mg, Na, sulfate, alkalinity, TCO2, K, B, F, phosphate, 87Sr/86Sr, δ13C, δ18O, U, and Rb. Although this site is located only 30 km from the trench axis, there is no chemical evidence for subduction-related hydrologic influences. Instead, the data are explained by a combination of seawater-basalt reactions within the upper basement and diffusive exchange with overlying sediment, as part of a shallow, ridge-flank hydrothermal system. It is unclear why this site has an elevated heat flux relative to neighboring crust, but this may result from variations in crustal properties or modification related to flexural faulting outboard of the trench.

Key Points

  • Pore water chemical profiles reveal sediment diagenesis, seepage speeds up to 1.7 cm yr−1, and discharge from 1.75 km2 of 0.1 L s−1

  • Crustal formation water is warm (∼75°C) and chemically altered, stemming from water-basalt reactions and diffusive exchange with pore water

  • This ridge-flank hydrothermal system is hydrologically isolated from the ventilated crust to the west and the trench to the east

Plain Language Summary

An area of elevated volcanic rock buried below sediment west of the Middle American Trench is unusually warm, with temperatures of 70–80°C compared to more regional-distributed crust that is vigorously cooled (10–20°C) from circulating seawater (formation water) within the upper volcanic crust. Because of the proximity to subduction-related crustal faults, sediment cores were collected from this area of warm crust and interstitial sediment pore waters were extracted and analyzed to assess if and how this area with a high heat flux is linked to subduction processes, potentially providing a source of water that serpentinizes the underlying mantle. Chemical data from sediment pore waters confirm upward seepage, the composition of crustal formation water in underlying volcanic rocks, and the extent of reactions as the formation water ascended the sediment column. Crustal formation water appears to be unaffected by nearby plate subduction. Instead, this warm and chemically altered formation water defines a ridge-flank hydrothermal system, which is distinct from hydrothermal systems to the west and south. How this hydrothermal system became isolated is unknown, but it could be a result of aging and evolving crustal properties or of plate faulting related to subduction processes.

1 Introduction

The amount of water bound in the upper oceanic lithosphere during subduction may influence melting conditions within the overlying mantle wedge, the depth extent of hydrothermal cooling, and the location of critical mineralogical and rheological transformations (i.e., peridotite to serpentinite) (Peacock, 1990). Some crustal hydration occurs along mid-ocean ridges where hydrothermal processes allow seawater to penetrate several kilometers into the upper oceanic crust at intermediate to fast spreading centers and deeper at slow spreading centers (Alt & Bach, 2003). More extensive crustal hydration occurs via low temperature hydrothermal processes on ridge flanks, regions far from the tectonic and magmatic influence of seafloor spreading, where the most intensive alteration is found in the upper 200–600 m of permeable basaltic basement (Alt, 2004; Fisher et al., 2014; Kardell et al., 20192021; Rohr, 1994; Wilkens et al., 1991). It has been proposed that reactivation or development of new normal faults outboard of a trench may allow seawater to penetrate deep into the crust and upper mantle, serpentinizing ultramafic rocks (Acquisto et al., 2022; Allen et al., 2022; Ranero et al., 2003; Shillington et al., 2015; Yamano & Uyeda, 1990), but thus far there has been little geochemical evidence for seawater that penetrates deeply within the crust in such settings.

Extensive normal faulting caused by plate flexure seaward of the Nicoya Peninsula, Costa Rica, coupled with anomalously low heat flow values provides an ideal setting to document processes that govern the flow of crustal formation waters seaward of a subduction zone. Two expeditions were completed using seismic reflection profiling, heat flux measurements, and sediment coring to determine the thermal and chemical state of the Cocos Plate seaward of the Middle American Trench (MAT; TicoFlux expeditions on the R/V Ewing, EW0104 and R/V Melville, Vanc02) (e.g., Fisher et al., 2003; Hutnak et al., 20072008; Wheat & Fisher, 2008). One finding from these studies was an area of ∼14,500 km2 on the Cocos Plate that has a seafloor heat flux of just 10–40 mW/m2, about 10%–40% of the conductive heat flux predicted by standard plate cooling models based on seafloor age (Fisher et al., 2003; Hutnak et al., 20072008).

Another finding identified a small area of seafloor, located on ∼24 Ma seafloor ∼40 km west of the MAT, in which the seafloor heat flux was anomalously elevated, with measured values up to 600 mW/m2, much greater than those calculated using standard lithospheric cooling models. Here we present results of geochemical analyses of pore waters extracted from gravity and piston cores collected on and near this “warm patch” of subducting seafloor and use these data to estimate (a) rates of pore water seepage at the seafloor, (b) the composition of formation waters in the underlying volcanic crust, and (c) the nature of seawater alteration in basement rocks and sediment while formation waters ascend as sediment pore waters. We also discuss potential mechanisms that might explain the presence of this anomalous warm patch of seafloor, including basement formation waters that are unusually altered in comparison to (a) cool hydrothermal waters to the west where the seafloor heat flux is anomalously low, (b) waters recovered from sites to the east, closer to the trench, where the heat flux is lithospheric, and (c) ridge-flank hydrothermal systems on other crustal plates with warm (60–70°C) basaltic crust.

2 Geologic Setting

TicoFlux expeditions sampled 18 to 24 Ma crust (Meschede et al., 1998) on the Cocos Plate immediately west of the MAT off the coast of the Nicoya Peninsula, Costa Rica (Langseth & Silver, 1996; Meschede et al., 1998; Fisher et al., 2003; Hutnak et al., 20072008) (Figure 1). The Cocos Plate comprises a seafloor created at two spreading centers. Crust formed from the fast-spreading East Pacific Rise (EPR), with a half spreading rate of 5.5 cm yr−1, defines the western edge of the plate. The Cocos-Nazca spreading center formed crust on the southern edge of the plate and has a slower half-spreading rate that is highly variable, increasing with distance from the EPR. This asymmetric spreading results in a gradual counterclockwise rotation of the Cocos Plate (Barckhausen et al., 2001; Wilson & Hey, 1995). A fracture zone and adjacent triple junction trace delineate a “suture zone,” where crust from these two spreading centers is joined (Barckhausen et al., 2001). This pattern of seafloor spreading results in a northeastern motion of the Cocos Plate as it subducts beneath the Caribbean Plate at a rate of 70–90 mm yr−1, forming a steeply sloped, non-accretionary trench system. High-angle faults that penetrate deeply within the lithosphere are evident tens of kilometers west of the trench axis and increase in number as the trench is approached (Moritz et al., 2000; Ranero et al., 2003; von Huene et al., 2000).

Details are in the caption following the image

Maps of the study area (modified from Fisher et al. (2003) and Hutnak et al. (2008)). (a) Regional map of the Cocos Plate showing the spreading centers and Middle America Trench west of Central America. (b) Regional map showing major tectonic boundaries (FZ = fracture zone, TJT = triple junction trace, RJ = ridge jump, and MAT = Middle America Trench), mapped volcanic outcrops, and heat flow data from multiple expeditions. Seismic reflection and other data are shown along profiles in Figures 2 and 7. More detailed maps of the focused study site are shown in Figure 3. Area marked in gray is the seafloor with a heat flux ∼10–40% of lithospheric values. The stippled light blue area lacks heat flow data and could connect areas with warm crust.

Typical heat flux values of 100–120 mW/m2 within the CNS-generated crust south of the suture zone are consistent with models for the conductive cooling of 18–24 Ma oceanic lithosphere (e.g., Parsons & Sclater, 1977; Stein & Stein, 1994). In contrast, measurements of seafloor heat flux from a large section of EPR-generated crust are 60%–90% lower than predicted by conductive cooling models (Vacquier et al., 1967; Langseth & Silver, 1996; Fisher et al., 2003; Hutnak et al., 20072008). Importantly, the region of low heat flux contains 11 mapped basement outcrops that rise about the surrounding seafloor, exposing volcanic rocks to the overlying ocean. In contrast, much of the adjacent study region between the fracture zone and the ridge jump (Figure 1b) has lithospheric heat flux and lacks basement outcrops. Heat flux profiles approaching outcrops in the cool area often show patterns indicative of either cool seawater recharge or warm hydrothermal discharge (Hutnak et al., 20072008), and one outcrop through which rapid discharge was inferred to occur was visited and sampled using deep submergence vehicles, confirming the discharge of altered seawater (Hartwell et al., 2018; McManus et al., 2019; Wheat & Fisher, 2008; Wheat et al., 20172019). Curiously, the transition between cool and lithospheric heat flux conditions tends to be abrupt, consistent with the presence and absence of vigorous shallow seawater circulation within the crust, respectively (Fisher et al., 2003; Hutnak et al., 2007; Lauer et al., 2018). Of interest, and the focus of this contribution is an area of higher heat flux within this cool ventilated region, where buried basement highs underlie relatively thin sediment.

This study focuses on a ∼1.75 km2 area located on 24 Ma crust, ∼30 km seaward of the MAT, where the water depth is ∼3,300 m. Here, a locally elevated topographic high (300–400 m across) rises ∼250 m above the surrounding basaltic basement. Sediment cover is uneven, resulting in a subtle seafloor bathymetric high (∼60 m), which is offset slightly to the east of the underlying basement high (Figures 2 and 3a). The seafloor heat flux on this basement high is >200 mW/m2 (highest value is ∼640 mW/m2) (Figure 3a) and is associated with the upward seepage and discharges of warm, altered seawater at the seafloor.

Details are in the caption following the image

Crustal structure and thermal data. (a) Sediment profile delineating major boundaries within the crust is based on the seismic image from line TFI-5 (Figure 1B). Sediment thickness is approximately 350 m at A and 100 m at (b) The solid line shows the area shown in Figure 3. (b) Measured heat flow values are plotted along this transect. Open squares represent values measured by the multipenetration heat flux probe, and open circles represent measurements made by miniaturized data loggers attached to core barrels (Hutnak et al., 2007). Gray dotted line shows the expected conductive heat flux for 24 Ma crust. (c) Calculated temperatures are plotted for the sediment-basement boundary based on heat flux and sediment thickness. Symbols as in panel (b).

Details are in the caption following the image

Local maps of the focused study area. (a) Bathymetric map showing core locations as solid red circles, and heat flow stations as open black squares. Seismic line TFI-5 (solid line) is roughly parallel to the Middle America Trench, providing a NW-SE cross-section of the basement feature shown in Figure 2. Seismic line TFII-10 (dashed line) follows a SW-NE path perpendicular to the trench. (b) Heat flux values (mW m−2) are plotted as open squares for measurements made using the heat flow probe and solid circles for those measurements made by autonomous thermistor outriggers attached to core barrels. Lines denote areas of equal heat flux contoured in 100 mW/m2 intervals. (c) Basement temperatures (°C) are calculated from heat flow measured at each core site (solid circles) and heat flow location (open squares). (d) Upward pore water seepage speeds (cm yr−1) were calculated for each core. Solid contour lines approximate regions of equal pore water upwelling speed.

This seepage occurs through marine sediment, which regionally consists of ∼100–200 m of hemipelagic deposits overlying a thicker pelagic interval (∼200–300 m) of carbonate ooze with a net thickness of 350–400 m (Spinelli & Underwood, 2004). This lower pelagic unit tends to thin and pinch out in proximity to seamounts and appears to be absent above the buried basement high below the warm patch, where the sediment thickness over the basaltic topographical high is less than ∼100 m (Figure 2a). Combined the sediment thickness and seafloor heat flux values indicate that the basement temperature below the warm patch is ∼70–80°C (Hutnak et al., 2007). In addition to the focused study area another sediment-covered basement high with similar chemical and thermal conditions was located ∼12 km to the southwest of the focused study area, defining a much broader area of warm volcanic basement.

3 Methods

We collected six gravity cores (GC) and five piston cores (PC), each with an associated GC, within a 1.75 km2 area of the warm patch (Figure 3) and two additional gravity cores collected ∼12 km to the southwest, above another small-buried basement high (Table 1). Upon retrieval, cores were split, sampled, and visually described. Distinct sediment types were sampled for semi-quantitative x-ray diffraction analyses for bulk mineral abundance and clay mineral abundance (<2 μm size fraction) (Underwood et al., 2003). Wet chemical analyses were completed to determine the amount of biogenic silica (Spinelli & Underwood, 2004). Representative cores were sampled for water content and formation factor every 15 cm (e.g., McDuff & Ellis, 1979). Pore waters were extracted from sediment cores at 10–15 cm intervals in the upper meter and 30–35 cm intervals thereafter. Sediment samples were cooled to 1–4°C and centrifuged. Pore waters were extracted and passed through a 0.45 μm filter into acid-washed high-density polyethylene bottles and glass ampules. Several whole-round sections from one core (PC 44) were sampled in a nitrogen-filled glove bag to minimize oxidation during sample handling. Pore waters were analyzed within several hours of collection using standard techniques: potentiometric titrations (chlorinity, alkalinity, and Ca), ion-selective electrodes (pH and F), and colorimetric analyses (phosphate). Numerous additional analyses were conducted ashore using standard techniques (Table 2).

Table 1. Piston Core (PC) and Gravity Core (GC) Locations, Lengths, and Estimated Core Losses, Sediment Thicknesses, Heat Fluxes, Temperatures in the Upper Basaltic Basement Calculated From the Sediment Thickness and Heat Flux, and Pore Water Seepage Speeds (Positive Is Upward)
Core ID Latitude Longitude Core length Core loss Pore water samples Sediment thickness Heat flux Basement temp Flow rate
N W m m # m mW/m2 °C cm yr−1
TFI-PC 18 (GC 18) 9°40.57ʹ 86°34.12ʹ 8.91 (2.12) 1.0 17 (8) 98 605 75 0.41
TFI-PC 19 (GC 19) 9°40.61ʹ 86°33.68ʹ 5.58 (1.90) 0.8 4 (9) 176 0.01
TFI-PC 38 (GC 38) 9°40.59ʹ 86°34.13ʹ 7.98 (2.08) 0.1 19 (7) 102 570 73 0.36
TFI-PC 39 (GC 39) 9°40.57ʹ 86°34.15ʹ 7.88 (2.43) 0.8 19 (6) 102 597 76 0.34
TFII-GC 11 9°40.77ʹ 86°34.27ʹ 3.80 12 106 538 71 0.87
TFII-GC 12 9°40.79ʹ 86°34.31ʹ 3.50 12 102 556 71 1.68
TFII-PC 44 (GC 44) 9°40.80ʹ 86°34.31ʹ 6.94 (3.00) 0.8 18 (8) 102 633 81 1.09
TFII-GC 62 9°40.89ʹ 86°34.45ʹ 3.58 12 121 516 78 0.36
TFII-GC 63 9°40.92ʹ 86°34.32ʹ 3.90 12 117 503 74 1.21
TFII-GC 64 9°40.84ʹ 86°34.20ʹ 3.37 12 109 635 87 1.28
TFII-GC 65 9°40.67ʹ 86°34.37ʹ 3.86 12 106 406 54 0.13
TFII-CG13 9° 36.44ʹ 86°39.08ʹ 1.24 6 98 539 69 0.13
TFII-GC14 9°36.39ʹ 86°39.14ʹ 3.60 12 94 665 81 0.19
  • Note. TF1 and TF2 refer to two TicoFlux expeditions. Core GC 13 and GC 14 were taken from a small topographic high to the west southwest of the primary coring area.
Table 2. Chemical and Thermal Composition of Crustal Formation Waters From Several Locations
Measurement Units Bottom Seawater Warm Patch ODP Site 1039 ODP Hole 678B Baby bare outcrop
Temperature °C 1.8 75 ∼5 58 64
Mga mmol/kg 52.6 5.25 48 8 0.98
Caa mmol/kg 10.25 98.4 13.6 66 55.2
Sulfateb mmol/kg 27.9 10.7 27.2 17.8 17.8
Chlorinitya mmol/kg 540 572 557 546 554.4
Nac mmol/kg 462 381 477 458 473
Na/Chlorinity mmol/kg 0.856 0.666 0.856 0.839 0.853
Alkalinitya mol/mol 2.51 1.1 1.6 0.6 0.43
TCO2 d mmol/kg 2.34 0.86 0.85
δ13Ce per mil −0.19 −5.1 −0.6
14C agee Years 2,170 35,600 4,300
Kb mmol/kg 10.2 4.9 11 6.5 7.1
18Of per mil −0.7 −1.4 0.3
D/Hf per mil −2.9 −0.9 1
Lib μmol/kg 26.6 34 16 16 9
Srb μmol/kg 87.9 193 310 110
87Sr/86Srf per mil 0.7092 0.7071 0.7089 0.7074
Bab μmol/kg 0.13 0.95 0.8 0.43
Bb μmol/kg 418 240 380 570
Fg μmol/kg 69 30 12
Mnb μmol/kg 0 20 10 2.9
Feb μmol/kg 0 1.5 4 <0.05
pHg --- 7.75 7.5 7.8 8.33
PO4 h μmol/kg 2.89 1 1 0.4 0.3
Sib μmol/kg 170 500 160 360
Rbi μmol/kg 1.34 1.2 0.8 1.12
Csi nmol/kg 2.2 10 5.3
Ui nmol/kg 13 1 0.6
Moi nmol/kg 105 500 297
NH3 h μmol/kg <0.5 20 400 76
  • Note. Ocean Drilling Program (ODP) Site 1039 is the closest borehole to the warm patch and has a composition that is like bottom seawater (Chan & Kastner, 2000; Kimura et al., 1997; Kopf et al., 2000; Silver et al., 2000). Data are provided for two similar ridge flank hydrothermal systems; ODP Hole 678B (Mottl, 1989) and Baby Bare outcrop (Butterfield et al., 2001; Elderfield et al., 1999; Sansone et al., 1998; Wheat et al., 2002; Wheat & Mottl, 2000).
  • a Potentiometric titration.
  • b Inductively coupled plasma (ICP) emission spectrometry.
  • c Charge balance.
  • d Gas chromatography.
  • e Mass spectrometer facilities at WHOI.
  • f Mass spectrometer.
  • g Ion selective electrode.
  • h Colorimetric titration.
  • i ICP-mass spectrometry.

The composition of the formation water at the site was compared to formation waters recovered from two ridge-flank hydrothermal settings for which formation water reacts with basaltic crust at similar temperatures: Ocean Drilling Program (ODP) Site 678B, drilled into 5.9 Ma crust along the southern flank of the Costa Rica Rift (Mottl, 1989), and Baby Bare outcrop, located on ∼3.5 Ma crust on the eastern flank of the Juan de Fuca Ridge (Wheat et al., 2022). Compositions of formation waters were used as inputs to thermodynamic calculations using the WORM portal to assess mineral solubility (i.e., saturation index (SI), which is the log of the ratio of the reaction quotient (Q) relative to the equilibrium constant (K); SI = Log (Q/K)) (Boyer, 2021).

Bottom seawater was collected from 85 to 100 m above the seafloor (3,200 m water depth) during two hydrocasts, each using messenger-tripped 5-L Niskin bottles. Bottom seawater samples were filtered and aliquoted for comparison with pore water compositions and to assess the error for each analytical method.

The heat flux at core locations was determined by an array of autonomous outrigger temperature loggers strapped onto core barrels, and thermal conductivity measurements were made on recovered cores and in situ with a multi-penetration heat-flow probe (Fisher et al., 2003; Hutnak et al., 2007). Sediment thickness was calculated at core locations from multi-channel seismic and swath mapping surveys, using the estimated two-way travel time between the sediment-water and sediment-basalt contacts and depth-dependent seismic velocities reported from nearby ODP Site 1039 (Hutnak et al., 2007; Kimura et al., 1997). Sediment thickness and measured thermal conductivity from ODP Site 1039 were used to calculate the temperature at the sediment-basement interface and interpreted to represent conditions in the uppermost volcanic crust. Uncertainties in these temperature calculations result mainly from uncertainties in determining sediment thickness from seismic reflection data, and are typically 5–10°C.

4 Results

Core recovery of surficial sediment ranged from 3.37 to 8.91 m (Table 1). Sediment was dominated by dark, olive gray, diatom-bearing hemipelagic mud with interbeds of volcanic ash. Mud contained a large fraction of clay minerals (59%–84% by weight, mean 76%), 3%–6% quartz, 14%–29% plagioclase (mean of 20%), and trace amounts (4%) of calcite. Clay minerals were composed of smectite (76%–89%, mean of 86%), kaolinite (10%–19%, mean of 13%) and illite (0%–5%). The opal content (biogenic silica) was 6%–14% (mean of 9%). These compositions are typical of the hemipelagic mud unit throughout the broader study area (Spinelli & Underwood, 2004).

Several cores contained ash layers, which were generally 2–7 cm thick. Two distinct types of ash layers were observed: (a) white to light gray in color, consisting mostly of unaltered shards of clear glass, with minor euhedral plagioclase crystals; and (b) dark gray to black in color, composed of heavily altered glass and volcanic rock fragments, with some clay minerals, plagioclase and pyroxene crystals. These altered sediments could be pyroclastic turbidites rather than atmospheric deposits, with most of the observed alteration occurring on land. Such deposits are still reactive, and the transformation from glass to smectite is an ongoing reaction. The formation factor of these sediments ranged from 1.4 to 2.0 (average ∼1.7). The average porosity was 80%.

Pore water compositions from the study area are highly altered relative to bottom seawater (Tables S1–S4). Relative to seawater, pore waters have lower concentrations of Mg, alkalinity, B, TCO2, F, phosphate, K, Na, Rb, U, and sulfate. Pore waters also have considerably lower δ13C, Sr87/Sr86, and Na/Cl ratios, as well as a slightly lower pH. In contrast, pore waters are enriched in Ba, Ca, Chlorinity, Fe, Li, Mn, Si, Mo, Cs, and Sr and have a slightly higher D/H relative to bottom seawater. The 14C age of overlying bottom water is 2170 ± 40 years, yet pore waters within these sediments have a 14C age of 35,600 ± 400 years. There was no difference between Fe-depth profiles from glove-bag and non-glove-bag samples. The glove-bag samples have at most 10% more Mn than non-glove-bag samples.

Chemical-depth profiles of pore waters typically display monotonic increasing or decreasing trends, and in some cases, an asymptotic value is reached downcore (Figure 4). For example, concentrations of Ca display systematic increases with depth in most cores toward an asymptotic value of 98.4 mmol/kg, whereas concentrations of Mg decrease toward an asymptotic value of 5.3 mmol/kg. Similarly, there is a range of profile shapes for each of the other chemical species, with asymptotic values being reached for most species in one or more cores. We use these systematic changes in composition to adjust the depth scale of the piston core data to account for the loss of sediment due to over-penetration by aligning the corresponding GC and PC data to maximize continuity with depth (Table 1).

Details are in the caption following the image

Pore water depth profiles of Ca, Mg, and Li concentrations. Hollow symbols represent cores taken during TicoFlux I: circles (PC 18), squares (PC 19), diamonds (PC 38) and triangles (PC 39). Solid symbols show cores acquired during TicoFlux II: circles (GC 11), squares (GC 12), diamonds (PC 44), inverted triangles (PC 62), triangles (GC 63), arrow pointed lower left (GC 64) and arrow pointed lower right (GC 65). Purple symbols and lines are from cores GC13 (circle) and GC 14 (square), which are to the west southwest of the primary coring area. The large, solid inverted triangle represents bottom water concentrations. Thick red dotted-dashed lines are modeled profiles for specified upward seepage speeds.

Rates of vertical pore water seepage are estimated by fitting observations to a simplified one-dimensional, time-dependent, advection-diffusion-reaction model (e.g., Berner, 1980). Since our focus is on systematic differences in profile shape and more importantly on the composition of formation waters, we make several simplifying assumptions. We assume a steady state, conservative behavior of solutes, and uniform porosity, temperature, and formation factor with depth. Bioturbation or irrigation processes may only affect the upper 10–15 cm (shallowest sample) below the seafloor and are not included in the calculation. We use ion-specific diffusion coefficients (Li & Gregory, 1974) to model depth profiles of pore water Ca and Mg. These two elements were chosen because they are generally the most conservative in this setting (Table 1), an assumption tested below. Chlorinity data were not used with this simplified approach because chlorinity is influenced by time-dependent processes related to changes in ocean chemistry between glacial and interglacial periods (e.g., McDuff, 1985; Wheat & McDuff, 1995).

Calculated upward pore water seepage speeds range from about 0.01 cm yr−1 for PC 19, which was collected several hundred meters from the peak of the buried basement high, to 1.68 cm yr−1 near the center of the basement topographic high, where sediment is thinnest (Table 1, Figures 3d and 4). Upward seepage speeds are >1 cm yr−1 for more than a third of the cores in the warm patch. At these speeds, the advective solute flux of formation water through the sediment column could greatly exceed solute sources and/or sinks associated with diagenetic reactions, increasing reliably for our estimated formation water composition (e.g., Wheat & Mottl, 2000). However, at slower upward seepage speeds, diagenetic reactions can shape chemical profiles. The seepage flux integrated over the warm patch where most of the cores were collected (1.75 km2) is 0.1 L s−1. This net flux of water is not thought to be significantly larger when considering the flux from the broader warm patch given the slow seepage speeds in cores collected to the southwest. For comparison, the flux of altered formation water that discharges from Baby Outcrop on the eastern flank of the Juan de Fuca Ridge is 4–13 L s−1 (Mottl et al., 1998), the flux from one vent area on the west side of Dorado Outcrop, 80 km southwest of the warm patch, is 200 L s−1 (Wheat et al., 2019), and the flux from a 5-cm-diameter, hydrothermal black smoker that vents at 1 m s−1 is 2 L s−1 (e.g., MacDonald et al., 1980; Schultz et al., 1992).

As noted above, the seafloor heat flux within the warm area is >200 mW/m2, with the highest measured values >600 mW/m2 (Figure 3). The calculated temperature at the top of the volcanic crust ranges from ∼65 to ∼90°C, with temperatures of 75–80°C in areas with the thinnest sediment (Figure 3c). In comparison, the heat flux to the west is 20–40 mW/m2, with a corresponding upper basement temperature of 10–20°C, and to the south and southeast, the heat flux is typically 100–120 mW/m2, with a corresponding upper basement temperature of 50–60°C (Hutnak et al., 20072008). Within each of these areas, relatively homogeneous upper basement temperatures are consistent with vigorous local hydrothermal circulation (e.g., Davis et al., 1989; Fisher et al., 1990; Fisher & Harris, 2010).

5 Discussion

We use systematic variations in pore water chemical compositions to assess where and at what speeds sediment pore waters seep upward from the volcanic crust. The absolute rates of seepage depend on several physical and chemical parameters, but the relative values and their distribution are the focus of this study. Importantly, pore waters flowing with the fastest seepage rates are the least affected by chemical reactions during transport resulting from sedimentary diagenetic processes. Such pore water-depth profiles are used to define the chemical composition of the formation water, which is the asymptotic concentration with depth. For many solutes, there is a single asymptotic value (e.g., Ca and Mg in Figure 4). In contrast, some solute-depth profiles display a range of asymptotic concentrations at the deepest depth sampled (e.g., Li in Figure 4). This range of profile shapes is not a consequence of bottom seawater being entrained (e.g., Wheat et al., 1998), but results from diagenetic reactions within the sediment but deeper than the cored sediment.

In contrast, cores from areas with slow upward seepage provide a measure of diagenetic reactions as formation water ascends. Ion-ion relationships define the type and extent of diagenetic reactions during upward seepage within the sampled section and elucidate processes that likely occur deeper below the sampled section but still within the sediment. The type and extent of diagenetic reactions is assessed by comparing solute concentrations to those of a conservative solute, and drawing a mixing line defined by seawater and the pore water composition at the base of the core that has experienced the fastest upward seepage speed. Diagenetic reactions within the sampled section cause deviations from this mixing line and may lead to underestimates or overestimates of the solute concentrations in the crustal formation water.

The estimated composition of the formation water and thermodynamic data are then used to infer processes and reactions within the basaltic basement and the sampled section. For the latter, we use the asymptotic concentration and a temperature of 2°C and we compare calculated output with observed solute changes in the hemipelagic muds of recovered sediments, consistent with regional stratigraphic and seismic data (e.g., Hutnak et al., 2007; Kimura et al., 1997; Silver et al., 2004; Spinelli & Underwood, 2004). Finally, we assess the significance of the composition of the formation water and pore water seepage in a hydrologic context, given a much cooler and ventilated crust to the west and subduction to the east.

5.1 Chemical Composition of Formation Water: Major Ions

Ca and Mg tend to be conservative in other ridge-flank hydrothermal systems (e.g., Mottl & Wheat, 1994), but there is evidence for non-conservative behavior at this site. Plots of Ca-Mg show data that fall below the mixing line defined by seawater and the deepest concentration from a profile that reaches an asymptotic concentration (thick red line in Figure 5a). This result is consistent with a loss of one of these solutes relative to the other in the sampled section. We use the sulfate data to determine whether Ca or Mg is more reactive (Figures 5b and 5c). In the sulfate-Mg plot, much of the pore water data are slightly above the mixing line, indicating either a sink for Mg or a source for sulfate. Processes likely to affect sulfate concentrations within the sampled sediment section are microbially mediated reduction of sulfate or the dissolution of barite, which is undersaturated at 75°C but saturated at 2°C (SI = −0.4 and 0.8, respectively, using data from Table 2; Table S5). Sulfate reduction consumes sulfate. The dissolution of barite would add sulfate to pore waters; however, barite dissolution is slow and Ba concentrations are orders of magnitude lower than sulfate concentrations; thus, any dissolution would not be observed in the sulfate data. Because there are no other known sources for sulfate in the sampled section, we conclude that there must be an Mg sink, probably involving the formation of smectite within observed ash layers (e.g., ODP Site 1039; Kimura et al., 1997).

Details are in the caption following the image

Plots of pore water Ca and sulfate versus Mg and sulfate, Na/ chlorinity, alkalinity and K versus Ca. Symbols and nomenclature are the same as Figure 4 except thick solid red lines represent mixing lines between the composition of bottom seawater and that of the projected formation water (Table 2). Linear relationships imply that the two ions plotted are conservative in the sampled section. A series of curved trends is consistent with diagenetic reactions within the sampled sediment, and linear but different trends are consistent with reactions deeper than the sampled section.

In contrast, the range of linear trends in the sulfate-Ca plot is consistent with sulfate reduction in sediment deeper than the cored section, and conservative behavior (e.g., linear within the range of the analytical measurements) of both solutes within the sampled sediment section (i.e., Wheat & Mottl, 2000). Data from cores with >1 cm yr−1 upward pore water seepage speeds approach a single mixing line, indicating that sulfate and Ca fluxes resulting from the reaction within the entire sediment column are small compared to those associated with advective fluxes. Thus, our estimate for the sulfate concentration in the formation water may be minimum (Table 2) with the actual concentration being up to ∼0.5 mmol/kg greater and still provide a linear mixing line that fits the data. This interpretation does not preclude Ca from reacting in the sampled section or deeper in the sediment column; however, the extent of such a reaction that removes Ca must be minimal such that it does not affect solute-Ca relationships within the range of the measured data.

Chlorinity increases with depth to a maximum of 572 mmol/kg, ∼6% higher than the concentration in bottom seawater (Table 2). When plotted against Ca, data scatter about the mixing line, consistent with changes in bottom seawater chlorinity during the last glacial period (3%–4%; e.g., McDuff, 1985). Past oceanic conditions can account for only a portion of the observed change in the chlorinity of the formation water. The remainder likely results from crustal hydration. Crustal hydration and glacial-interglacial changes in seawater composition also affect the stable isotopic composition of water. Thus, given the potential for temporal changes in chlorinity and basalt hydration, Ca is considered the most “conservative” solute within the sediment section in this area and is used as the benchmark to assess the reactivity of other solutes.

Na data deviate significantly from a mixing line with Ca; this deviation is shown in a plot of Na/Cl versus Ca to account for the glacial/interglacial effect (Figure 5d). Excess Na in pore waters likely results from reactions that occur in the sampled section and under closer inspection the range of linear Na/Cl versus Ca trends point to reactions deeper than the sampled section, but still within the sediment, given a well-mixed volcanic crust as inferred by the heat flow data.

Several trends are apparent in the plots of alkalinity versus Ca (Figure 5e). Some trends are linear, consistent with sulfate reduction deeper than the sampled section. Other trends are non-linear, consistent with reactions in the sampled section that result in an increase in alkalinity. This increase likely results from sulfate reduction, where a maximum of 1 mmol/kg change in alkalinity requires a maximum decrease in sulfate of 0.5 mmol/kg, which is difficult to detect in the sulfate versus Ca plot (Figure 5c). Alternatively, the increase in alkalinity could result from the dissolution of carbonates; however, primary carbonate minerals are saturated in the sampled section (e.g., calcite has a SI = 0.39 and aragonite has a SI = 0.24 at 2°C). Interestingly, the formation water (Table 2) is more saturated at 75°C (SI = 1.3 and 1.2, respectively for calcite and aragonite). A similar extent of over saturation is observed in formation waters from similarly altered ridge flank hydrothermal systems. For example, Baby Bare (64°C), ODP Sites 1028 (50.5°C), 1029 (58.7°C), 1032 (57.1°C), Isita Bare (64°C) and Hole 678B have calculated SIs for calcite (1.3, 0.8, 1.2, 0.9, 1.7, and 0.7 respectively) and aragonite (1.1, 0.7, 1.0, 0.8, 1.6, and 0.6 respectively) (Tables S1–S4) (Mottl, 1989; Wheat et al., 2022). Thus, our estimated alkalinity in the formation water is likely a maximum, given that the sulfate value is likely a minimum.

Dissolved carbon dioxide concentrations (TCO2) in formation water from the warm area on the Cocos Plate and Baby Bare on the Juan de Fuca flank are similar; however, the δ13C of the TCO2 is significantly more negative at the warm area than at Baby Bare, where fractionation is not evident (Sansone et al., 1998). Preferential removal of the heavier carbon isotope is expected with the precipitation of calcium carbonate (Romanek et al., 1992). However, the magnitude of the shift in δ13C is too large to result only from carbonate precipitation (Zeebe & Wolf-Gladrow, 2001). Thus, a portion of the δ13C must result from the oxidation of organic matter (Nissenbaum et al., 1972), an interpretation that is consistent with the old 14C data (presented below). Similarly, our estimate for the concentration of dissolved carbon dioxide (TCO2) in the formation water is also likely a maximum (Table 2).

The sampled section is a source of K to pore waters (Figure 5f). The K versus Ca trend roughly defines two mixing lines that intersect at the approximate depth of several ash layers. Alteration of this volcanic material is the likely source of K in pore waters. A similar increase in K is observed in surficial sediments at ODP Site 1039 (Kimura et al., 1997). This interpretation is consistent with the observed diagenetic behavior of Mg, which is presumably removed during smectite formation within such ash layers. Our estimate for the concentration of K in the formation water is therefore a maximum (Table 2).

5.2 Chemical Composition of Formation Water: Minor and Trace Ions

Li data show a complex relationship with Ca (Figure 6). There are distinct mixing lines for each core, with exception of the two cores with the slowest pore water seepage speeds. Linear mixing lines are consistent with conservative behavior within the sampled section, and differences in these trends likely result from reactions at greater depths but within the sediment. The extent of Li enrichment/removal within pore waters is most closely correlated with the pore water seepage speed and results from the interplay between advective and reactive fluxes (Figure 6). The two cores with the slowest speeds (0.01–0.13 cm yr−1) are consistent with typical surficial diagenetic reactions, resulting in the removal of Li from pore water in the sampled section. In contrast, the maximum Li enrichment occurs in cores with moderate seepage speeds (0.34–0.41 cm yr−1), requiring deep sediments to be a source of Li. Similarly, pore waters from nearby ODP Site 1039 have subtle Li enrichment (∼5 μmol/kg) and an associated increase in δ6Li (Chan & Kastner, 2000). The alteration of ash layers coupled with ion exchange within clay minerals in the sediment is likely responsible for this enrichment (Chan & Kastner, 2000). For example, ammonium desorbs isotopically light Li from clays, resulting in increased Li concentrations and elevated δ6Li. However, the asymptotic Li concentration in cores with the fastest pore water seepage speed is nearly 50 μmol Li/kg less than the concentration in cores with moderate flow rates (Figure 4c). The asymptotic Li concentration for these cores is ∼34 μmol/kg, which is thought to reflect the composition of the formation water most closely, but is an upper bound. Processes influencing Li profiles are summarized as follows. As the warm formation water from the upper basaltic basement seeps upward, it reacts with sediment, resulting in a source of Li while the pore water cools conductively as it ascends. If the ascending seepage speed is sufficiently slow, reactions that occur within deep sediments are masked by the removal of Li in shallower sediments. If the seepage speed is slightly faster (∼0.3–0.5 cm yr−1), then the removal mechanisms are less effective. If seepage speeds are even greater (e.g., >1 cm yr−1), then the advective flux overwhelms sediment reactive fluxes, maintaining a concentration that is intermediate to the two other scenarios and more similar to the composition in the formation water.

Details are in the caption following the image

Plots of pore water Sr, Li, and Ba versus Ca. Symbols and nomenclature are the same as Figure 5, except a dashed line is used to represent conservative mixing. Linear relationships imply that the two ions are conservative in the sampled section. A series of trends is consistent with diagenetic reactions within the sediment, but deeper than the sampled section.

The Sr-Ca relationships are linear, suggesting that diagenetic reactions in the sampled section are minimal. Cores with the slowest pore water seepage speed have the greatest slope except for one core (triangle in Figure 6a). These data are consistent with the addition of Sr to pore waters deeper than the sampled section, perhaps because of diagenesis within pelagic carbonate sediments (e.g., Stout, 1985). Like other solutes, the rate of the reaction is slow enough such that the advective flux begins to swamp the diagenetic flux at upward seepage speeds exceeding ∼1 cm yr−1. Our estimate for the concentration of Sr in the formation water is technically a maximum; however, given the shape of the profiles and the 87Sr/86Sr of selected pore waters, our estimate is probably close to the actual concentration—the Sr isotopic composition is similar to that of Baby Bare springs and ODP Sites 1028, 1029, and 1032 (Butterfield et al., 2001; Elderfield et al., 1999).

Like alkalinity, the Ba-Ca trends are nonlinear and fall below the mixing line, consistent with the removal of Ba within the sampled section (Figure 6c). Barite formation would remove Ba from the solution; however, the formation water at 2°C is saturated (SI = 0.8). Interestingly, the formation water is undersaturated with barite at 75°C (SI = −0.4). Removal of Ba is likely associated with sulfate reduction and the precipitation of barium sulfide in the sampled section. As noted above, sulfate reduction would have a minimal effect on the concentration of sulfate but could impact Ba concentrations, which are three to four orders of magnitude less than that of sulfate. Our estimate of the Ba concentration in the formation water is a minimum.

Additional dissolved solutes were analyzed (B, Cs, Rb F, Mn, Fe, phosphate, Mo, U, Si, and pH) and concentrations were calculated for these solutes in the formation water (Table 2, Figures S1–S5 in Supporting Information S1). The B concentration in our deepest sample is about half of the value in bottom seawater. Like the case for Li, B is removed in surficial diagenetic reactions but added at depth, resulting in an estimate for the formation water that is a maximum. The Cs and Rb profiles are strongly influenced by low temperature diagenetic reactions in the sampled section. In general, Cs and Rb in pore waters are consumed during these reactions, yet concentrations of Cs in the formation water are elevated relative to those in bottom seawater. In contrast, Rb is removed in formation water relative to bottom seawater. Surficial and deep diagenetic reactions remove F, resulting in a minimum estimate of the concentration in formation water. Mn and Fe data are consistent with diagenetic profiles influenced by microbial activity. Both solutes show characteristic maxima near the sediment-seawater interface, followed by a decrease in the concentration below these maxima. Concentrations of phosphate increase with depth in surficial sediments as a product of microbial consumption of organic matter and then decrease with depth. Profiles are typical of those found in this area (Wheat & Fisher, 2008) and in other ridge flank hydrothermal systems in general (Wheat et al., 2022). Profiles of Mo and U are also influenced by biological diagenetic reactions. The rate of these reactions is sufficiently fast to affect profiles from cores with the fastest seepage speeds. Silica profiles are impacted by the dissolution of amorphous silica in the sediment section. The concentration of Si at the base of the recovered core with the fastest pore water seepage speed is ∼500 μmol/kg, resulting in a concentration that is barely undersaturated with respect to quartz (SI = −0.06) and chalcedony (SI + −0.3) in the formation water at 75°C. Lastly, pH values decrease with depth to a value of ∼7.5, the chosen value for the formation water.

5.3 Source of Altered Formation Water

There are three possible sources for the composition of altered formation water at this site, two of which are associated with subduction-related processes. The third explanation, and the one that we consider to be likely, is that the composition of the formation water results from ridge-flank hydrothermal circulation within the upper basaltic basement, independent of subduction.

It has been suggested that formation water near subduction zones may migrate along deep crustal faults, possibly penetrating the upper mantle, reacting with peridotite to generate serpentinite (e.g., Ranero et al., 2003; Yamano & Uyeda, 1990). If the formation water sampled in the study area was subjected to serpentinization at depths of several to tens of kilometers below the seafloor, water-rock reactions and high temperatures at depth should result in (a) solute concentrations of K that exceed those of bottom seawater, consistent with water-rock experimental data at temperatures in excess of 150°C (Seyfried & Bischoff, 1979), (b) a water depleted of sulfate resulting from anhydrite formation at temperatures much less than 150°C given the high Ca concentrations relative to seawater (Bischoff & Seyfried, 1978; Janecky & Seyfried, 1986), (c) high pH and depleted Si concentrations relative to bottom seawater (Hulme et al., 2010; Janecky & Seyfried, 1986; Kelley et al., 2001), and (d) enriched B concentrations relative to bottom seawater (Chan & Kastner, 2000; Menzies et al., 2022; You et al., 1996). Data from the study area seaward of the MAT are inconsistent with this explanation.

Another possibility is that waters from the subduction system, originating either above, within, or below the plate-boundary fault, could migrate from the upper basaltic basement of the incoming plate and discharge at basaltic outcrops outward of the trench. This scenario is also inconsistent with observations. First, there is no evidence of elevated heat flux or basement temperatures in association with normal faults that formed as the plate approaches the MAT (Figure 7). These faults do not appear to be hydrothermally active, as there are no associated heat flux anomalies at the seafloor, in contrast to an expected sharp increase in heat flux as a fault is approached if there was extensive discharge (Hutnak et al., 2007). Second, at ODP Sites 1039 and 1253, which were drilled in the trench and penetrated the incoming plate (Figure 1), the estimated temperature in uppermost basaltic basement is ∼6°C, and crustal formation waters are relatively unaltered from their original seawater composition (Chan & Kastner, 2000; Kimura et al., 1997; Morris et al., 2003; Silver et al., 2000) (Table 2). In contrast, east (landward) of these boreholes, ODP Sites 1040 and 1255 yielded sediment pore waters that were transported by subduction processes. These waters have low chlorinity, sulfate and Ca concentrations relative to bottom seawater (Chan & Kastner, 2000; Kimura et al., 1997; Silver et al., 2000; Zuleger et al., 1996). Geophysical and geotechnical analyses demonstrate that subduction-related dewatering of the sediment occurs mainly east of the trench at this subduction system, and there is no evidence that this water reaches the incoming plate west of the trench (McIntosh & Sen, 2000; Saffer et al., 2000). Chemical data are also inconsistent with the migration of subduction-related water west toward ODP Sites 1039 and 1253. For example, there is no source of sulfate along a potential flow path from deeply sourced subduction-related water at ODP Sites 1255 to ODP Sites 1039 and 1253.

Details are in the caption following the image

Seismic reflection and heat flux profiles along a profile oriented perpendicular to the MAT and east of the warm patch on the Cocos Plate (location of line shown in Figure 1B). The yellow band shows the range of lithospheric heat flux expected for 18–24 Ma seafloor, and the green band shows the range of typical observations for the seafloor of this age (modified from Hutnak et al. (2007)). Normal faults are apparent in the seismic reflection data from the east side of the line, going down and into the trench, but no faults are evident to the west. Seafloor heat flux is low in the trench area, even directly above fault traces, and basement temperatures are consistently 10–20°C. In contrast, to the southwest, in an area of expected lithospheric heat flux, basement temperatures range from 50–60°C.

The chemical composition of formation water within the buried basement high is most consistent with water-rock interaction within a localized ridge-flank hydrothermal system. Our data are generally consistent with trends in seawater-basalt experiments at 70°C except for the lack of sulfate and chlorinity changes during the period of experimentation (e.g., Seyfried & Bischoff, 1979). The extent of alteration is consistent with that from two other well-studied ridge-flank hydrothermal systems with similar temperatures in the upper basaltic basement (Table 2; Costa Rica Rift (Mottl, 1989) and Juan de Fuca Ridge (Wheat et al., 2022)). In general, the compositions of the formation waters from these three sites are similar (Table 2), but there are some noticeable differences.

The Ca enrichment within the study area is equivalent to about twice the Mg depletion. In contrast, formation water with similar temperatures that range from 20°C to 65°C on the eastern flank of the Juan de Fuca Ridge show a one-to-one relationship between the gain in Ca and the loss in Mg (Wheat et al., 2022). An additional source of Ca in formation waters includes the exchange of Ca for Na. Ca-Na exchange can occur (a) during in situ seawater-rock reactions at temperatures greater than those estimated at the top of basement in this location (e.g., Magenheim, Bayhurst, et al., 1992; Magenheim, Gieskes, et al., 1992) and (b) in basal sediments within which Ca concentrations can reach 300 mmol/kg (e.g., ODP Site 1201; Shipboard Scientific Party, 2002) followed by diffusive exchange with formation water in uppermost basaltic basement (additional examples are listed in Table S6). This exchange, should it occur at this site, is likely localized near the basement high where clay-rich basal sediments are present, in contrast to the surrounding basal sediments that are carbonate-rich. Such exchange in basal sediments is well documented in other settings (e.g., Gieskes, 1983; Mottl, 1989; Wheat et al., 2000).

Diffusive exchange between overlying sediment pore waters and formation waters appears to be the primary sink for sulfate in formation water on ridge flanks (Lever et al., 2013; Wheat et al., 2000). For example, deep-sea sediment pore waters can be depleted in sulfate due to microbial activity, resulting in a diffusive flux of sulfate from the formation water to the overlying sediment pore waters. Continued diffusion of sulfate from the basaltic basement to the sediment ultimately decreases the sulfate concentration in the formation water, allowing sulfate to be used effectively as a hydrologic tracer (e.g., Hulme & Wheat, 2019; Wheat et al., 20002022). The lower sulfate concentration within the study area relative to the other two ridge flank hydrothermal sites (Table 2) likely results from more extensive diffusive exchange; however, the elevated Ca concentration also makes anhydrite precipitation in the upper volcanic crust a possibility. Given the composition in Table 2, the SI of anhydrite is 0.13 at 75°C compared to lower values for Baby Bare (0.03), Site 1028 (−0.3), Site 1029 (−0.03), Site 1032 (0.02), Isita Bare (−0.3) and Hole 678B (0) (Table S5). Even though the SI for Baby Bare is greater than 0, the sulfate data can be explained by diffusional losses to the overriding sediment pore waters alone. Data from ODP Sites 1028,1029,1032, and Isita Bare are likely to have a long residence time of water in upper basaltic basement, resulting in more substantial impacts from diffusional exchange with overlying sediment pore waters, as appears to be the case in the study area (Wheat et al., 2022).

Additional differences between the study area, Hole 678B, and the Juan de Fuca Flank can be explained by somewhat warmer basement conditions and a longer residence time in the warm patch, such that formation waters become more altered by reaction and diffusive exchange with the overlying sediment. Long resident times favor lower K and B concentrations in formation water as observed (Table 2) (e.g., Wheat et al., 2022).

The suggestion that formation water within the study area has undergone a more extensive diffusive exchange with overlying pore waters is consistent with a 14C age of 35,600 ± 400 years, considerably older than that estimated for basement water recovered on the eastern flank of the Juan de Fuca Ridge or at Dorado Outcrop (Elderfield et al., 1999; Wheat & Fisher, 2008). However, at 1 cm yr−1 it takes ∼10,000 years for formation waters to transit through a 100 m thick sediment column; thus, the 14C age of formation water is at most 25,000 yrs. This is still considered a maximum age because the diffusion of older carbon during transport can contribute significantly to the apparent age of subsurface water transport (e.g., Bethke & Johnson, 2002; Sanford, 1997) and sulfate reduction in the sediment likely adds old carbon during ascent. However, the depth of the sample for 14C age dating was 6.08 m, thus not impacted by reactions within the shallower section, but could be impacted by the unsampled section. Given the sulfate-Ca relationship and similar SI values for calcite among similar settings, it is unlikely that the alkalinity in the formation water is more than 0.5 mmol/kg lower than the estimated value in the formation water. Any increase in alkalinity during ascent would add old carbon to the formation water and prior to collection.

5.4 Hydrologic Interpretations

Waters collected from the study area represent the most altered endmember sampled to date along a continuum of ridge flank hydrothermal systems, from cool, nitrate-rich relatively unaltered seawater that extracts much of the crustal heat flux to warm, reduced highly altered formation water that removes little of the local and global heat flux (e.g., Table 2; Figure 8) (e.g., Wheat et al., 2022). This transition occurs when basement exposures become blanketed with sediment, impeding water and solute transport across the seafloor (Spinelli et al., 2004; Wheat et al., 2019). This sediment barrier also acts as a reactive source or sink for solutes, resulting in diffusive exchange between sediment pore water and formation water. This exchange can bring additional Ca to the formation water, resulting in the precipitation of carbonates within the upper basaltic crust, which in turn decreases basement permeability and likely the flux of seawater transport through the crust (Wheat et al., 2022).

Details are in the caption following the image

The general pattern for seawater flow through the warm patch on the Cocos Plate. This cartoon demonstrates the interaction of the advective, reactive, and diffusive processes that control the composition of seawater as it is cycled through the upper basaltic crust prior to seeping through the sediment and discharging at the seafloor.

Pore waters that seep through sediments within the study area must comprise a small fraction of water stored in the underlying hydrothermal reservoir. What is most unusual about this hydrothermal system, and the associated seafloor seepage area, is that it is surrounded by a much more open and regionally vigorous hydrothermal system. This regional hydrothermal system is responsible for reducing the seafloor heat flux on EPR-generated seafloor to 10%–40% of lithospheric input (Fisher et al., 2003; Hutnak et al., 20072008). The more localized system in this study must be isolated from this larger regional system because (a) upper basement temperatures are much greater than those in the surrounding area (70–80°C vs. 5–20°C), and (b) formation water is considerably more altered than formation waters in the surrounding area, with the latter closely resembling seawater (McManus et al., 2019; Wheat et al., 2017). Thus, the two systems must be hydrogeologically distinct; a similar separation of distinct basement formation waters was also found on the eastern flank of the Juan de Fuca Ridge, suggesting that the effective permeability is geologically controlled (Wheat et al., 2022). In each of these areas, the separation of warmer and cooler regions is likely a consequence of variations in basement permeability, which must remain relatively high to allow thermal homogenization, and to gather sufficient heat to elevate the seafloor heat flux immediately above the basement high.

One indication as to the size of the hydrothermal reservoir below the seepage site comes from the calculated seepage flux and the apparent water age. The total seepage flux from an area of 1.75 km2 is ∼0.1 L s−1, equivalent to ∼4 × 103 m3 yr−1. If we assume that the 14C age is an indication of the typical water residence time, on the order of 2.5 × 104 yr (accounting for the time required to transit the sediment column), and that the system is currently operating at steady state, then the volume of water in the reservoir is ∼108 m3 (0.09 km3). Assuming an effective porosity (the porosity of the rock matrix through which most of the water flows) within the basement reservoir of <1% (Neira et al., 2016), then the reservoir has a volume of just ∼10 km3. This reservoir volume fits easily within the upper 200–500 m of the basement in the area defined by elevated seafloor heat flow surrounding the seepage site. This is an upper limit to the volume of the hydrothermal reservoir because the apparent 14C age is probably greatly overestimated.

What remains uncertain is why the hydrothermal reservoir below the study site is hydrogeologically isolated from the surrounding crust. Multiple abrupt thermal transitions were documented during heat flux surveys across this region, so the same processes responsible for these juxtapositions of warmer and cooler seafloor in close proximity (kilometers) may also be responsible for isolating the warm patch. Given that the two cores to the southwest have a similar thermal and chemical composition in the underlying formation water relative to the area of focused sediment coring, the area of warm crust may be quite substantial and may be hydrologically connected to crust to the south (Figure 1), given no heat flow data were collected between these two regions. Thus, the warm crust could extend to the warm heat flow area to the south, suggesting that crustal properties within the study area and in the south are similar and limit the extent of seawater circulation in the upper crust relative to the cooler crust to the west.

Another possibility is that this is a site of a waning ridge-flank hydrothermal system, as sediment covers what were once efficient flow pathways between the crust and the ocean, leading formation waters to be trapped in a reheating seafloor. Another possibility is related to the proximity of this portion of the plate to the MAT. In this scenario, normal faulting as the plate descends forms hydrologically distinct regions or compartments. In this latter case, there would be an indirect influence of subduction processes on this hydrothermal system, one that does not involve a water contribution from subduction.

Because the hydrothermal reservoir is small, it is unlikely that this area with an elevated heat flux and highly altered formation water will have a significant influence on subduction processes when this portion of the crust enters the trench, but in general, thermal and alteration properties of the subducting crust have important ramifications for processes within the seismogenic zone (Figure 7) (e.g., Harris & Wang, 2002). Drilling and hydrogeologic testing will be required to directly assess the nature of this seepage site and the hydrothermal reservoir beneath it.

6 Conclusions and Implications

We present geochemical data from samples collected from a seafloor seepage site above a ridge flank hydrothermal system. The system is located within and surrounding a hydrological basement high on 24 Ma crust outboard of the MAT. The formation water in the upper basaltic basement has an estimated temperature of ∼75°C near the peak of the basement high and a chemical composition that is highly altered compared to that of bottom seawater. Seawater-rock reactions within volcanic basement and diffusive exchange with overlying sediment result in formation water that has less Mg, alkalinity, B, K, F, phosphate, Na, TCO2, U, Rb, and sulfate and more Ca, chlorinity, Mn, Fe, Li, Si, Ba, Cs, Mo, and Sr relative to bottom seawater. The stable isotopic compositions of water, dissolved Sr, and TCO2 are also altered. As this highly altered formation water in basaltic basement ascends through the sediment column at speeds of 0.01–1.68 cm yr−1 the chemical composition changes because of diagenetic reactions. Concentrations for most of these solutes are affected by reactions that occur either within the sampled section, below the sampled section, or both. These reactions include, but are not limited to, microbial activity (e.g., reduction of sulfate), the precipitation of barite, alteration of volcanic ash, smectite formation, carbonate recrystalization, and ion exchange. This site represents the most altered endmember sampled to date along a continuum of ridge flank hydrothermal systems.

We have considered the possibility that the composition of the formation water at this site results from either deep circulation of lithospheric water outboard of the trench, or waters from the subduction zone, but neither of these explanations is consistent with geochemical nor geophysical observations. The lack of evidence for subduction-related water at this site is consistent with an analysis of the global heat flow data set, which shows a lack of significant thermal anomalies near trenches (Stein, 2003). This finding suggests that water and chemical fluxes from subduction zones or associated with flow along deeply penetrating faults are either insignificant or are masked by the ridge flank hydrothermal process (Stein, 2003). Instead, the formation of water within the study area appears to result mainly from ridge-flank hydrothermal processes that have been documented in numerous settings. Relative to other ridge flank hydrothermal systems, differences in the formation water composition in the warm patch of the Cocos Plate are attributed to elevated temperatures and a longer water residence time, both of which allow for the cumulative effects of diffusive exchange with overlying sediments and water-rock reactions to significantly alter the composition of the formation water. In addition, the water reservoir below this seepage site is relatively small (perhaps 1–10 km3 in total volume) and may be hydrogeologically isolated from surrounding reservoirs.

Acknowledgments

We thank the TicoFlux scientific parties, crews of the R/V Ewing and Melville, and Chris Moser for their dedicated efforts in helping us obtain sediment samples. This paper benefited from anonymous reviews. This project was funded by the NSF MARGINS program (OCE 0002031 to CGW, OCE 00001892 to ATF and EAS, OCE-1924384 to ATF, and the NSF REU Program. Sr isotopic measurements were made at the University of Hawaii and the Scripps Institute of Oceanography by M. Kastner. Measurements of TCO2 and δ13C were made by Dan McCorkle at WHOI, and an aliquot from these analyses was measured for radiocarbon at the NOSAMS Facility in Woods Hole.

    Data Availability Statement

    The pore water chemicals used in this study are archived (Wheat, 2023) and included in Supporting Information S1 for easy access.