Geochemistry, Geophysics, Geosystems

Fluid sources and pathways of the Costa Rica erosional convergent margin



[1] The margins community has only relatively recently begun to examine the tectonics and associated hydrologic systems of erosive convergent margins, which are substantially different as compared with accretionary margins. In this respect, the type example erosive margin is the Costa Rica system, which has been the subject of numerous recent large-scale investigations. Here pore fluids expelled at the wedge toe and at midslope mounds and mud volcanoes have been interpreted to have a common deep source of dehydrating clays, analogous to that at accretionary margins. However, we report unusually high B/Li molar ratios in pore fluids from a recent mudflow on Mound 11, offshore Costa Rica, which, together with unusually low B/Li ratios previously reported at the wedge toe, reveal that alternative fluid sources and/or processes must be operating at the Costa Rica margin. As serpentine formation is the only subduction zone process that significantly fractionates B and Li, we propose that the difference in fluid chemical composition is the result of erosion of upper plate serpentinites, ongoing serpentinization, and serpentine mineral phase transitions in the subduction channel. These processes provide both a source of fluids and fluid pathways that lead to the unique geochemical signature observed at this erosional margin. This conclusion is compatible with, and supported by, the current view of the tectonics, geology, and hydrogeology of the Costa Rica margin and the similarity of the pore fluid to that of two other convergent margins, both with known fluid/serpentinite interactions.

1. Introduction

[2] Most of our knowledge of hydrogeologic systems at convergent margins results from over two decades of studying accretionary margins, such as Barbados, Cascadia, and Nankai [e.g., Carson and Screaton, 1998; Moore and Vrolijk, 1992]. Only recently, however, have we begun to examine the substantially different tectonics and associated hydrologic systems of erosive convergent margins [Ranero and von Huene, 2000; Ranero et al., 2000, 2008; Vannucchi et al., 2003]. The Middle American margin off Nicaragua and Costa Rica is considered an end-member-type example in this respect. Consequently, the section from the Nicoya Peninsula to the Osa Peninsula has been the focus of multiple international research programs, including the Ocean Drilling Program, the National Science Foundation MARGINS Program, and the German SFB574 (SonderForschungsBereich 574, Volatiles and Fluids in Subduction Zones). These programs have included a strong focus on understanding the hydrogeological system through deep and shallow sediment coring, high-resolution mapping, active and passive seismic experiments, and visual bottom surveys (see synthesis by Ranero et al. [2008]). One outcome of this work has been the discovery of numerous mound structures on the margin slope in water depths of 1000–3000 m that appear to be mud volcanoes or diapirs and chemoherms [Bohrmann et al., 2002; Weinrebe and Flüh, 2002]. Significant volumes of water and gas emanate from many of these structures and current estimates indicate that much, if not most, of the chemically bound water released from this subduction system may be discharged via focused seepage though these mounds [Ranero et al., 2008]. Missing from such estimates, however, is the contribution by sources other than subducting sediment. In particular, subduction erosion systems can include a contribution from hydrated igneous material from the upper plate to the hydrologic system.

[3] Evidence for the expulsion of chemically bound water was discovered during ODP Leg 170, offshore Nicoya Peninsula [Chan and Kastner, 2000; Kimura et al., 1997]. Exotic pore fluids were found at ODP Sites 1040 and 1041 where sediment coring intersected the décollement and major faults near the toe of the wedge. On the basis of the fluid composition of pore waters extracted from cored sediment, Silver et al. [2000] concluded that these fluids include a significant component of deeply sourced fluids derived from clay dehydration at ∼10 km depth where the temperature is ∼100°C. These deeply sourced fluids have low chlorinities and high Li concentrations, typical indicators of clay dehydration. Atypically, they also exhibited low B concentrations (as low as 73 μmol B/kg (Table 1)). In contrast, gravity and multicoring of mud mounds 20–30 km arcward of the Costa Rican trench and south of ODP Leg 170 produced pore fluid with a low chlorinity, a high concentration of B, and an isotopic composition of water that is consistent with a source from the dehydration of underthrust clay-rich sediments at a formation temperature of 85–130°C [Hensen et al., 2004]. These findings were augmented in June 2005 by DSV Alvin surveys, push cores, and fluid sampling at Mound 11 and 12 [Tryon et al., 2006]. A notable feature of the fluids sampled at Mound 11 was that they also exhibited unusually low concentrations of Li, which had not been observed previously. The unusually low B/Li molar ratio at the wedge toe but high B/Li molar ratio at the arcward mud mounds presents a conundrum, since both fluids have been interpreted to originate by the same process and at the same temperature of ∼100°C. This paper presents a hypothesis to account for the difference in fluid composition between these two localities that is compatible with the emerging view of the geology and tectonics of this erosional convergent margin.

Table 1. A Comparison of Chlorinity, Boron, and Lithium From Mud Volcanoes With the Costa Rica Sitesa
SiteSourceCl (mM)Li (uM)B (uM)B/Li
  • a

    Seawater is included for reference.

Barbados MVs     
   AtalanteDia et al. [1995]250177411023
   CyclopeDia et al. [1995]29855535906
   CyclopeManonGodon et al. [2004]4502043022
Eastern Mediterranean MVs     
   Milano domeODP 160-970D8016∼5000313
   Napoli domeODP 160-971D4100146∼500034
Caucasus MVs     
   GeorgiaKopf [2003]425339530016
   TamanKopf [2003]1574092150053
Taiwan MVs     
   Average of reported valuesYou et al. [2004]211216343016
Mariana MVs     
   ConicalMottl et al. [2003]2601.639002438
   S. ChamorroMottl et al. [2003]5100.432008000
Black Sea MVs     
   DvurechenskiiAloisi et al. [2004]835150021701.4
Gulf of Cadiz MVs     
   Captain ArutyunovHensen et al. [2007]5806501400022
   BonjardimHensen et al. [2007]340500800016
   OthersHensen et al. [2007]440150420028
Costa Rica MVs     
   Mound 11this study2138.11786220
Costa Rica     
   Wedge toeODP 170-1040C5131481390.9
   Wedge toeODP 170-1041B3512521570.6
   Wedge toeODP 205-1254A480235730.3

2. Observations and Sampling at Mounds 11 and 12

[4] During R/V Atlantis cruise AT11-28, 5 DSV Alvin dives were carried out on Mounds 11 and 12 (Figure 1) to conduct visual surveys, collect push cores, and deploy CAT fluid flowmeters and geochemical samplers. The CAT flowmeters [Tryon et al., 2001] were recovered one year later and the sites resurveyed with DSV Alvin. The mounds have distinctive characteristics, both morphologically and chemically. The northernmost, Mound 12, has a cone-shaped summit several hundred meters in diameter that rises ∼50 m above the seafloor and is densely covered with massive carbonate in the summit area. The current hydrologic activity at the site supports dense biological communities dominated by clams and mussels in the carbonate areas and microbial mats in sedimented areas. Mound 11 lies ∼1 km SSE of Mound 12 and is significantly smaller with two distinct summits rising ∼25 m (11b) and ∼15 m (11a) (Note that we are adopting the naming convention of Klaucke et al. [2008] for 11a and 11b and that this is the opposite of that in our earlier presentations of this work). Massive carbonates cap Mound 11b, but are less densely spaced compared to Mound 12. Mound 11a has very little carbonate exposed on the seafloor. The morphology of Mound 11 is rugged with sharp scarps in sediment many meters high. These scarps are particularly evident at Mound 11a where numerous features, which we interpret as fracture scarps, were observed. The morphology suggests an inflating dome that is fracturing and extending. A NW-SE trending valley divides Mound 11a in its SW quadrant where sediment with a diffuse cover of microbial mat and a moussy texture cover much of the valley floor. A 1 m long heat flow probe, which normally requires significant force to insert into sediment, fell full length into this sediment under its own weight, indicating little cohesiveness. The lack of cohesiveness, morphology of the mound, and side-scan sonar data [Klaucke et al., 2008] are consistent with this being a relatively recent mudflow. On the basis of these observations, we conclude that Mound 11 is a newly forming mound with the current focus of activity at Mound 11a.

Figure 1.

Map of Mounds 11a and 11b indicating the location of the mudflow and the locations of the cores and CAT meters. Inset shows regional setting.

[5] Push cores were taken and CAT meters deployed on the SW flank of Mound 12 and in the mudflow area of Mound 11a at what appeared to be locations where seepage of crustal fluids was greatest, based primarily on the extent and density of microbial mats. Additional push cores, including background cores, were taken at other sites on both mounds. Cores were sectioned, cooled to 1–4°C, and centrifuged at sea. Extracted pore fluids were analyzed for chlorinity (titration), and alkalinity, sulfide, and ammonia (colorimetry) and stored for subsequent shore-based analysis. Na, Ca, Mg, K, Sr, Ba, B, Mn, Fe, Li, and Si were analyzed by ICP-AES. After CAT meter recovery, the fluid sample coils were sectioned and analyzed for chlorinity by titration and Na, Ca, Mg, K, S, Sr, Ba, B, and Li by ICP-AES. A summary of the results are listed in Table 2 with complete data available in the auxiliary material.

Table 2. Pore Fluid Chemistry at Mounds 11 and 12a
Instrument/CoreSiteEnvironmentAlk (meq)H2SCl (mM)Ca (mM)K (mM)Mg (mM)Na (mM)B (μM)Li (μM)Sr (μM)Si (μM)
  • a

    CAT chemistry reported is the most altered fluid end-member in the temporal record for each instrument. Core chemistry is the deepest 2 cm section of the core for which we have complete analyses and that do not appear contaminated on recovery (i.e., some core bottoms were not used). Typical depth was 15–25 cm. The complete record is available in the auxiliary material.

CAT-211amudflow  2464.64.812.7211146010.823.2 
CAT-2911amudflow  3976.07.129.732288118.144.6 
CAT-2511amudflow  3074.46.017.6261129614.633.0 
CAT-2812mat  55410.210.453.247839725.686.5 
CAT-1812mat  54910.110.353.747539325.885.9 
CAT-2712mat  54210.510.453.847840826.889.5 
CAT-2212mat  55110.310.552.547342225.480.5 
CAT-2412mat  5499.810.553.548740223.888.5 
CAT-2112mat  5375.49.949.546941522.877.9 
Core 4125-1012background2.30.05369.910.852.045936024.486.4293
Core 4125-612background6.70.154210.09.950.846538826.185.7320
Core 4125-712background2.30.053710.910.250.946036924.586.6993
Core 4127-212background2.50.05389.810.050.446538624.686.8248
Core 4127-612background4.40.25399.611.649.646852329.387.0310
Core 4125-412mat margin31.510.45396.010.147.347334621.868.5286
Core 4125-812mat margin19.21.35397.310.048.946536121.774.0276
Core 4127-112mat margin30.16.35387.010.646.247444025.869.71362
Core 4125-1212mat35.99.95424.310.346.848231620.264.1309
Core 4125-212mat35.921.95437.19.749.948337724.680.7689
Core 4125-912mat34.013.95306.610.247.246541826.173.2829
Core 4127-1012mat28.57.85377.110.449.248141223.580.4288
Core 4127-312mat30.16.35387.010.346.548141326.774.7910
Core 4128-212mat33.79.05385.310.246.847837120.167.4323
Core 4128-312mat35.912.55357.09.846.448138419.867.7359
Core 4128-512mat38.513.35355.610.147.048343820.868.2335
Core 4128-612mat40.516.05315.89.647.547636420.667.2291
Core 4129-112mat4.60.55349.210.646.946450425.881.3483
Core 4130-311abackground22.64.44895.49.737.142075418.762.8472
Core 4130-A11abackground12.11.05107.610.045.743750219.775.4384
Core 4126-1011abackground2.50.053910.711.247.946847327.086.0306
Core 4126-1211amat29.68.54676.78.938.840262221.863.6653
Core 4126-911amat32.79.64816.610.539.741661123.070.1309
Core 4130-1011amat7.90.32706.35.713.5238153910.430.1557
Core 4130-1111amudflow5.10.02214.94.28.319615368.520.7543
Core 4130-1211amudflow8.00.12364.64.79.021314789.622.8548
Core 4130-411amudflow5.30.02165.04.19.318916458.821.3539
Core 4130-511amudflow5.00.01974.44.08.217416968.218.1545
Core 4130-611amudflow5.20.02134.24.38.318917868.120.4599
Core 4126-211bbackground14.93.94047.58.727.2355111018.153.8409
Core 4126-611bbackground18.73.04626.38.336.339662922.061.0326
Core 4126-411bmat5.20.02257.85.89.5191179510.226.1541
Core 4130-1311bmat19.20.25395.96.616.1509172914.132.51054
Core 4130-1411bmat6.60.02458.06.310.4210183110.625.1621
Core 4130-1711bmat9.80.62466.25.912.8215198111.329.9486
Core 4130-1811bmat11.62.02765.66.415.2248173112.232.2415

[6] There are clear differences in the fluid composition of pore waters from Mounds 11 and 12. For example, fluids from Mound 12 are little different from seawater, except for those ions involved in near-surface microbial activity and carbonate formation (e.g., sulfide, sulfate, alkalinity, and Ca). In contrast, the composition of pore waters from Mound 11 are highly altered from seawater values and, in particular, have very low Cl and Mg concentrations, and very high B concentrations and B/Li molar ratios. A clear mixing line is evident between seawater and a low chloride end-member fluid (Figure 2). The high B/Li molar ratio at Mound 11 has not been observed at other mounds along this margin, including Mound 12, only 1 km away. Other mounds exhibit B and Li concentrations close to seawater values [Hensen et al., 2004].

Figure 2.

Plots of the major ions and B, Li, Sr, and Si versus chloride at Mound 11. A clear mixing line is indicated between seawater and a low-chlorinity fluid enriched in Si and greatly enriched in B. Note that extrapolation to zero chlorinity results in nonzero results for other ions (e.g., Mg = −20 mM and K = 1.3 mM) so simple dilution by, e.g., hydrate dissociation is not indicated.

3. Marine Geochemical Cycle of Li and B

[7] There are three possible sources for Li and B in fore-arc settings: subducting sediment, the igneous crust and serpentinized upper mantle that underlies it, and the overriding margin wedge. Volcanic ash that has weathered to smectite is a common component of convergent margin sediment. It is considered the primary source of the low-Cl water that seeps from active margins. It also hosts Li and B which are adsorbed in the interlayer region of expandable smectite minerals. Li is preferred because its charge and size are comparable to the common interlayer elements K and Na [Williams and Hervig, 2005]. At temperatures in the range of 80–130°C, smectite transforms into illite through the loss of interlayer water along with some of the K, Li, and B. Some B and Li remains in the silicate framework at tetrahedral and octahedral sites, respectively [Calvet and Prost, 1971; Williams et al., 2001]. Laboratory hydrothermal experiments with sediment have shown that Li and B are released over the temperature range from ∼60°C through 300°C with Li increasing roughly exponentially and B linearly, with the B/Li ratio below seawater values for all heating steps [You and Gieskes, 2001].

[8] Low temperature alteration of extrusive oceanic crust is similar to ash with the most abundant alteration products being ∼80% saponite (a smectite) and ∼20% celadonite (a mica) [Alt and Honnorez, 1984; Donnelly et al., 1979; Jarrard, 2003]. While no low-temperature experimental data exist, low-temperature ridge flank hydrothermal data indicate that B and Li are removed from pore water and incorporated within alteration products such as smectite [e.g., Wheat et al., 2003]. At temperatures below ∼65°C concentrations for B and Li remain around seawater values, but tend to have a B/Li molar ratio that is greater than seawater. We anticipate B and Li will be liberated upon heating altered crust, in a fashion analogous to high-temperature hydrothermal vents (300–400°C) in which seawater reacts with basaltic crust. These fluids have high Li concentration relative to B, resulting in lower than seawater B/Li molar ratios [e.g., von Damm, 1990].

[9] The alteration of ultramafic rocks in the oceanic slab or overriding wedge may influence crustal fluids in subduction zones should they be exposed in faults and fractures. Here, seawater-rock reactions produce serpentinite. The alteration of peridotite or other ultramafics to serpentinite greatly enriches the B content of the rock, with B content increasing with water/rock ratio [Agranier et al., 2007; Lee et al., 2008; Vils et al., 2008]. There are minor changes in the Li concentrations with some studies indicating leaching of Li from the host rock during alteration [Lee et al., 2008; Snyder et al., 2005; Vils et al., 2008] and others suggesting uptake of Li [Benton et al., 2004]. For example, serpentinite mud from Conical Seamount has a B content of about 10–80 ppm and Li content of 1–8 ppm while protolith mantle peridotite values are less than 1 ppm and 3–7 ppm [Benton et al., 2001, 2004]. In contrast to these petrological findings, field results indicate that Li concentrations in pore waters in active serpentinite mud volcanoes are lower than seawater values and B concentrations are typically significantly higher [Mottl et al., 2004].

4. Discussion

[10] The constraints outlined above on the B and Li cycles in subduction environments leads one to conclude that if the Mound 11 pore fluid origin was sediment or hydrated basalt then the resultant fluid would have elevated B and Li concentrations, with the B/Li molar ratio decreasing with increasing temperature. If this fluid reacted with sediment or basaltic crust on its ascent to the mounds then one would expect B and Li to be removed. This removal may not necessarily be at a one-to-one molar ratio, but we would not expect large deviations from this ratio. This process is observed in subduction environments around the world where B/Li molar ratios seldom differ from seawater by more than a factor of 2. In contrast, a source fluid with a gross difference in B and Li contents should maintain that difference to a significant degree as it ascends through sediment or basalt and reacts, unless it is strongly diluted by other fluid sources (e.g., seawater, consolidation dewatering). Consequently, we are led to the conclusion that fluids must be enriched in B relative to Li to a great extent prior to transport through the basaltic and sedimentary fore-arc wedge. We also suspect that serpentinite may be involved in the fluid genesis as serpentinite contains the only common subduction zone mineral assemblage capable of producing large elemental fractionations between B and Li.

[11] In order to understand the genesis of this unusual pore fluid composition, we first look at potential source regions and fluid pathways. We then compare the Costa Rica mounds area to other localities worldwide where similar fluid compositions have been observed and discuss the emerging hypotheses. The discussion concludes with a reconciliation of the different pore fluid compositions of Mounds 11 and 12.

4.1. Sources and Pathways

[12] Generally, the difference in pore fluid composition at the wedge toe and upper slope mounds is attributed to either different fluid sources or to different reactions along fluid pathways subsequent to fluid generation, or to some combination thereof. The source fluid may differ along strike (i.e., different inputs with distinct physical or chemical characteristics) or across strike (i.e., upper or lower plate or higher or lower along the subduction thrust). Here, we present structural differences that may account for the observed chemical differences between data from Mound 11 and ODP Leg 170 sites.

[13] There are significant along strike changes in the origin, morphology, and thermal structure of the incoming plate off Costa Rica. In the ODP Leg 170 area off Nicoya, ∼24 Ma crust formed at the fast spreading East Pacific Rise (EPR) has a relatively smooth seafloor topography with a trench-parallel abyssal hill pattern. This crust is anomalously cold for its age and the pattern of heat flow has been interpreted to indicate a vigorous fluid flow system within the upper oceanic crust, with entry and exit points through basaltic outcrops (seamounts) [Hutnak and Fisher, 2007]. The tectonic boundary between EPR crust and intermediate spreading Cocos-Nazca Spreading Center (CNS) crust is ∼20 km southwest of ODP Leg 170 with an age of ∼22.7 Ma. Further to the southwest lies Mound 11 where the crust is ∼18 Ma. This region has about 40% coverage with seamounts, other igneous structures, and sills younger than 16 Ma resulting from off-axis volcanism from the Cocos Ridge/Galapagos hot spot trace [von Huene et al., 1995]. Abyssal hill orientation is at a steep angle to the trench, with plate bending and subduction being accommodated by new faulting on the outer rise. ODP Leg 170 and mound sites are also on opposite sides of a major arc chemical transition where moderate 10Be, moderate Ba/La, and high δ15N are observed to the north, all indicative of high sediment input to the source of arc magmatism. In contrast, southern sites are consistent with little sediment input [e.g., Carr et al., 1990; Gazel et al., 2009; Morris et al., 1990]. This transition also roughly coincides with the Quesada Sharp Contortion (QSC), a sharp decrease in slab subduction angle in the south thought to be a slab tear below ∼70 km depth. In contrast to these significant differences, sediment cover along the entire margin is similar in thickness and composition and consists of typical pelagic and hemipelagic sediment between 200 and 500 m in thickness [Spinelli and Underwood, 2004]. The incoming sedimentary section off Nicoya in the ODP Leg 170 area is ∼380 m thick, composed of an upper unit of diatomaceous ashy ooze with local sands (∼80 m), an intermediate unit of silty clays and ash layers (∼80 m), and a lower biogenic unit largely comprising calcareous oozes and some siliceous oozes and calcareous clays (∼220 m).

[14] Across strike source differences in fluid origins might be a consequence of a migrating décollement or different depth-temperature source regions. A downward migrating décollement would underplate sediment allowing it to dewater through the upper plate while fluids from higher-temperature sources further downdip pass below the underplated sediment to the toe along the décollement. A downward migrating décollement would also thin the high-impedance sediment layer between the igneous slab and the décollement, potentially allowing greater fluxes of slab-sourced fluids. An upward migrating slab, as would be expected of an erosive margin, would both increase the hydraulic conductivity of the décollement through introduction of highly fractured upper plate rock and also transport upper plate material to higher pressures and temperatures.

[15] The difference in pathway between fluid exiting near the toe via the décollement and that exiting the upper slope are clear, yet the nature of toe/décollement pathways are potentially different for an erosive margin than for other well studied accretionary convergent margins. In accretionary margins, the décollement can remain at the subducted sediment/wedge boundary or progressively migrate downward as subducted sediment is underplated. One consequence of a downward migrating décollement is that the oceanic basaltic crust can, on occasion, contribute to fluid expulsion if the décollement nears or reaches the crust, essentially breaching the sedimentary aquatard at the top of the extrusives. Another pathway for ventilating the downgoing oceanic basaltic crust is through seamounts that penetrate the sediment burden. In erosive margins, the décollement progressively migrates upward as a result of erosion of the upper plate via hydrofracture and/or mechanical means such as seamount subduction. In this case, the décollement pathway lies in typically igneous or mantle wedge materials with these materials being transferred to the lower plate and subducted to greater depths.

4.2. Other Locations With High B/Li Molar Ratios

[16] High B/Li molar ratios are rarely seen in association with fluids from mud volcanoes (Table 1). The most common occurrence is at active serpentinite mud volcanoes such as Conical Seamount in the Mariana subduction zone [Mottl et al., 2004]. High B/Li molar ratios in fluids of a composition similar to Mound 11 are also found at Milano Dome in the Hellenic subduction zone [de Lang and Brumsack, 1998; Haese et al., 2006]. These two systems seemingly have little in common; however, in both cases, serpentinites are thought to be present in the subduction channel and fluid pathway.

[17] In the Mariana fore arc, pore water from massive serpentinite mud volcanoes indicates an arcward trend of progressive devolatilization of the subducting plate [Fryer et al., 1999; Mottl et al., 2004]. This progression generally shows an increasing B/Li molar ratio ending with the most arcward serpentinite volcanoes, Conical and South Chamorro Seamounts, with 3–4 mM B and ∼1 μM Li. Fluids exiting these volcanoes are thought to be derived from dehydration of hydrous minerals in the subducting slab (sediment and upper basaltic crust) that subsequently interact with the overlying mantle wedge, producing serpentinite muds and mud volcanoes [Fryer et al., 1995]. The source of the high B seen in the expelled fluids is considered to be subducted material, such as sediment, altered oceanic crust, and/or eroded upper plate serpentinite [Barnes et al., 2008; Benton et al., 2001; Mottl et al., 2003].

[18] In addition to high B fluid concentrations, serpentinites from Conical Seamount also have high δ11B values (∼+13‰) that are essentially derived from advecting subduction fluids since mantle peridotites have a very low B content and negative δ11B values [Benton et al., 2001]. In toeward portions of a subduction system serpentinites with high δ11B values may exist from reaction with seawater (∼40‰) released from pore spaces by consolidation of the sedimentary material. Deeper-sourced fluids, however, originate from dehydration reactions within the incoming crustal unit. Estimates of the sedimentary and altered oceanic crustal δ11B input for deep-sea subduction zones are −3 and +4‰, respectively [Benton et al., 2001], too low to produce the observed serpentinite δ11B values at Conical Seamount. The high boron isotope fractionation values could result from reactions that preferentially partition 11B into the fluid phase [Palmer et al., 1987, 1992; You et al., 1995], yet production of the observed high δ11B fluid values require a large degree of fractionation of 9–16‰ [Benton et al., 2001]. Alternatively, subduction of upper plate serpentinite to the depth of the Conical Seamount source would not only increase the B content of the fluids, from the underlying crustal source, but also greatly increase the δ11B value of this deep source serpentinite material and, thus, the fluid derived from it, as observed in the upper cored sediments of serpentinite mud volcanoes [e.g., Mottl et al., 2004]. Chloride isotopes further suggest that subducted serpentine is the source for the high B of the arcward mud volcano fluids. Barnes et al. [2008] concluded that high δ37Cl values of the muds and clasts are consistent with these materials being derived from the lizardite-antigorite phase transition. The results of the above studies on the Mariana mud volcanoes suggest that the arcward increase in B content of expelled pore fluids may be greatly enhanced by the subduction of eroded upper plate material and the subsequent reaction of this material with water released by additional dehydration reactions.

[19] Another location where high B/Li molar ratios are observed is the Hellenic subduction system where the uplift of the island of Crete is driven by subcrustal dehydration [e.g., Meier et al., 2007]. On the basis of recent geodetic, geophysical and structural data, the devolatilizing mantle wedge is believed to favor material transfer through the subduction channel updip along the plate boundary thrust. Here the ascent of serpentinized peridotite/dunite or eclogite is likely responsible for the present-day uplift rates observed on Crete. Both seismological investigation and numerical simulation favor a model where the return flow occurs along a corridor of hydrated material atop the subducting slab that reaches crustal levels or even shallower [Gerya et al., 2002; Meier et al., 2007].

4.3. Hypothesis: Serpentinite in the Costa Rica Mounds Fluid Source Region and Pathways

[20] The hypothesis emerging from the chemical composition of the Mound 11 fluids, the similarity of that fluid to fluid from two other margins, and the current view of the tectonics and hydrogeology of the Costa Rican Pacific margin, is that serpentinite minerals are involved in the fluid genesis. The formation of serpentinite is the only common subduction zone process that acts to significantly fractionate B and Li. Here there are two possible sources for serpentinite: the subducting oceanic lithosphere and the base of the Nicoya Complex (upper plate).

[21] The subducting oceanic lithosphere clearly contains altered peridotites at outer rise and abyssal hill faults as well as serpentinized olivine within the gabbro layer. Fluids associated with these assemblages may make their way through to the upper plate. Normally, such a flow direction is difficult due to the great hydraulic impedance of the overlying sediment layer; however, the incoming plate at this location is densely covered with seamounts. These seamounts may provide a conduit of high-permeability fractured rock for the crustal/lithospheric fluids to ascend.

[22] The Nicoya Complex is the western edge of the Caribbean Large Igneous Province (CLIP) that is thought to mark the initiation of the Galapagos hot spot [Hauff et al., 2000; Kerr et al., 1998]. Most of the exposed CLIP is composed of basaltic pillow lavas, flows, and sills; however, picrites and less abundant komatiites are distributed near the base of the upper crustal sequence [Kerr et al., 1998] (Figure 3a). These features result from excess magma supply leading to emplacement of sills and high-level magma chambers. The dense and chemically heterogeneous ultramafic magmas intrude the base of the pile and undergo fractional crystallization to form ultramafic dunite and pyroxenite cumulates [Kerr et al., 1998]. In the fore-arc wedge, these assemblages would be intruded by high-pressure-derived fluids from sediment compaction/dehydration, altering mineral assemblages present including serpentinizing the ultramafics (Figures 3 and 4). Another potential indicator of upper plate serpentinite is the Santa Elena complex, adjacent to and north of the Nicoya Peninsula, which contains extensive outcrops of serpentinized mantle peridotites and may be a portion of an exhumed subduction assemblage [Hauff et al., 2000]. Such assemblages are not uncommon along convergent margins past and present [e.g., Shervais et al., 2004]. Thus, although no ultramafics have been directly associated with the Nicoya Complex, it is reasonable to assume that they are incorporated within the upper plate.

Figure 3.

Hypothetical simplified cross section through the CLIP and subsequent initiation of subduction, volcanism, and subduction erosion through the present (not to scale). (a) Redrawn from Kerr et al. [1998], copyright 1998, with permission from Elsevier. (b and c) Based on Vannucchi et al. [2003].

Figure 4.

Simplified cartoon of the Costa Rica fore arc. Sediment enters the subduction channel at the toe and is consolidated, expelling fluid that passes up to the toe via the décollement. Further consolidation and early diagenesis expels fluids that pass via fractures in the upper plate, contributing to alteration of the fore-arc mineral assemblage. As more high-pressure fluids intrude the fore arc and seamounts are subducted, subduction erosion begins to carry upper plate materials into the subduction channel, ultimately dominating the flux. Some of these upper plate materials include serpentinite, which releases B into the fluids, leading to a high B/Li molar ratio fluid passing through the most arcward faults to ultimately produce mud volcanoes on the fore-arc surface. Figure modified from von Huene et al. [2004] and Goss and Kay [2006].

[23] This upper plate material is also being eroded and transferred to the lower plate [Ranero et al., 2008] (Figures 3b and 3c). As erosion continues, the ratio of this eroded material to lower plate sediment increases downdip toward the seismogenic zone, at which point the flux of eroded material is an order of magnitude greater than that of sediment (Figure 4) [Vannucchi et al., 2003]. Consequently, the deepest sourced fluids will have a chemical signature dominated by fluids in equilibrium with the mix of sediment, hydrated basalt, and hydrated ultramafic minerals (e.g., serpentine) at temperatures of ∼125–150°C.

[24] On the basis of the arguments above, we suggest that serpentinite is present in the subduction channel and that fluid-serpentinite reactions are responsible for the unique composition of the Mound 11 pore fluids as well as the large difference in B/Li molar ratio between the ODP Leg 170 décollement region and Mound 11. This hypothesis does not require serpentine dehydration. Serpentine mineralogy and B content are strongly tied to temperature, with temperature and B content inversely related [Bonatti et al., 1984; Wenner and Taylor, 1971]. Orthochrysotile is the dominant serpentine phase in the 120–200°C range, lizardite and “polygonal serpentine” are dominant in the 80–120°C range, and clinochrysotile and lizardite are dominant in the 30–80°C range [Bonatti et al., 1984]. The B content is thought to result from lizardite's structure being more accommodating to B than other phases [Bonatti et al., 1984]. Thus, phase changes associated with serpentine at increasing temperature may lead to an increased B content of the fluid. Subduction of B-rich serpentinite, whether of upper plate or lower plate origin, is problematic for the observed low B content of the central Costa Rican arc [Leeman et al., 1994]. However, the coincidence of fluid release through sediment dehydration and B release through the serpentine low- to high-temperature phase transitions may transport much of the B to the fore arc rather than to subarc depths.

[25] The presence of serpentinites has been proposed as one control on the downdip limit of the seismogenic zone that often coincides with the décollement intersecting the serpentinized mantle. In the case presented here, however, serpentinites make up a relatively small and inhomogeneous portion of the seismogenic region and are insufficient to limit seismogenic behavior of the subduction thrust. Therefore, seismogenesis does not preclude the presence of serpentinite. This distribution of serpentinite leads to the intriguing possibility that serpentine minerals may be dehydrated by frictional heating, either coseismic or aseismic, introducing a potentially more deeply sourced B-rich fluid.

4.4. Mound 11 Versus 12

[26] The proximity of mounds 11 and 12 virtually insures that they must have a common source region for the fluids that created them and share at least the lowermost portion of the fluid pathway. Therefore, the marked differences in fluid composition must have a relatively near-surface origin. Pore fluids from Mound 12 have high alkalinities (∼35 meq) and sulfides (7–16 mM), in contrast to those of Mound 11 (∼10 meq and 0–5 mM, respectively). The total dissolved inorganic carbon (TDIC) content and δ13C∑CO2 values also differ with Mound 12 at −60 to −68‰ (VPDB) and Mound 11 at −11 to −23‰ [Füri et al., 2010; this study]. Typically, very low δ13C∑CO2 values in such environments result from microbially mediated anaerobic oxidation of methane via sulfate reduction according to reaction [Boetius et al., 2000]:

equation image

This process transfers the C isotopic signature of the methane to the HCO3 and increases the alkalinity and sulfide concentrations. This observation, combined with the other features of the mounds described earlier, suggests that Mound 11, and particularly the Mound 11a mudflow, is a younger and more active feature while Mound 12 is older and possibly in the waning portion of its cycle. The increased age and lower activity level allows for more methane oxidation imparting the observed dissolved carbon and sulfide characteristics. We suggest that the greater age of Mound 12 provides the means for B in the pore fluid to be adsorbed by clay minerals whereas fluids at Mound 11 retain a composition much closer to the original source. The greater age of Mound 12 also allows time for greater mixing of seawater with formation fluids as a result of cycles of inflow and outflow driven by free gas expulsion [Tryon et al., 1999]: this mixing not only accounts for the lower B/Li but also its overall seawater-like composition. These processes are analogous to the situation at Torishima Seamount in the Marianas which is an inactive serpentinite mud volcano. The lack of fluid flow has allowed the continued formation of serpentinite to greatly reduce the B content of the pore fluids to below seawater values. The influx of seawater also has altered the pore fluid content of Li and other species to more seawater-like values.

5. Conclusion: Mound 11 Versus the Wedge Toe

[27] The evidence of Mound 11 is insufficient to determine whether the fluids discussed here are derived from the subducted component of the eroded fore-arc wedge or if they originate in the slab and enter the subduction channel through subducted seamounts. However, the low B/Li molar ratio at the wedge toe strongly suggests ongoing serpentinization in the shallow, cooler portions of the subduction channel. This observation, and the foregoing discussion, leads us to the following model for the genesis of the deep-sourced fore-arc fluids.

[28] The fluid source for Mound 11 lies just above the seismogenic zone at ∼120–150°C and is composed of up to 90% altered igneous and ultramafic material rich in serpentinite mixed with ∼10% lower plate sediment. In this zone, dehydration and illitization of smectite drives fluid production and some B and Li is released with this water with a B/Li molar ratio lower than but similar to seawater. The phase change in serpentine in the same region, from low- to high-temperature forms, releases significantly more B into the fluid, raising the B/Li molar ratio. These deeply sourced B-rich fluids pass upward through the fractured upper plate to the slope sediment layer where they fluidize mud before extruding onto the seafloor (Figure 4).

[29] This deeply sourced fluid represents a relatively minor component of fluids reaching the wedge toe via the décollement. Dilution by dewatering occurs toeward and likely volumetrically masks a deeper-sourced fluid signal. The dominant fluid source at the toe originates farther updip along the décollement at cooler temperatures (<100°C). Consolidation and early diagenetic reactions, including the loss of interlayer water in smectite, drives fluid production and releases B and Li into the fluid. This fluid passes up the subduction channel along the décollement and loses much of its B to low-temperature serpentinization of the eroded material passing down the subduction channel. This loss of B results in the observed low B/Li molar ratio observed in the ODP Leg 170 area.

[30] The deep fluid source may be unique to the Mound 11 area due to the heterogeneity of the source region or the lack of extensive exploration of the fore arc, but, more likely, is only evident here due to a recent eruptive cycle. Fluid sampled at other mounds may have had a longer residence time in the slope sediments where B would be adsorbed and the upwelling fluid diluted with seawater.


[31] Funding for this project was provided by NSF grant OCE-0242034 and OCE-0242091. We would like to thank the captain, expedition leader, and crews of the R/V Atlantis and DSRV Alvin for their continued excellence and professionalism.