PRIMARY RESULTS OF ODP LEGS 170 AND 205 AND RELATED EXPERIMENTS

Approximately 170 m of igneous rock was cored during Leg 205 at Site 1253, with two primary objectives in mind. One was to use core descriptions and borehole logging to identify zones of higher alteration and fracture density within the drilled section, which would be targets for long-term monitoring. These horizons are candidates for localized fluid flow, which could be responsible for the significantly lower heat flow in the region and for the return to near-seawater compositions seen in basal sediment pore water profiles, discussed above. A second goal was to characterize the composition and alteration history of the igneous basement entering the Middle America Trench (MAT) to better constrain the flux of volatiles and other elements through the seismogenic zone and beyond to the depths of magma generation beneath the Central American volcanic arc (CAVA). This goal can be accomplished through a combination of Legs 205 and 206 (Wilson, Teagle, Acton, et al., 2003) and IODP Expeditions 309 and 312 (Teagle, Alt, Umino, Miyashita, Banerjee, Wilson, et al., in press). The latter three focused on deep drilling of the oceanic section at Site 1256 (see Fig. F1). This site is located along approximately the same flow line from the EPR to the MAT as is Site 1253. At Site 1253 itself, drilling bottomed in what proved to be a thick complex (>150 m) with chemical compositions typical of OIB rather than mid-ocean-ridge basalt (MORB). Leg 205 thus sampled a large amount of OIB overprinting on the subducting plate and can speak to its origin and significance but cannot address the nature of the crust generated at the EPR, or its alteration history.

Igneous Character and Origin of Leg 205 Gabbros

Igneous rocks were recovered at Site 1253 between ~400 and 430 mbsf (Subunit 4A) and from 450 mbsf to the bottom of the hole at 600 mbsf (Subunit 4B). As seen in Figure F10, Subunit 4A is clearly a sill, being relatively flat lying and bounded by sediments above and below. It is comparable to similar units cored in Leg 170 Holes 1039B and 1039C, drilled ~1 km seaward of Site 1253, and in Hole 1040C (Kimura, Silver, Blum, et al., 1997). Calcareous nannofossil dating of sediments above and below the sill indicates a minimum age for the sediment of 15.6 Ma and a maximum age of 18.2 Ma (Muza, this volume), implying sill emplacement more recently than 18.2 Ma and perhaps since 15.6 Ma. Carbon and oxygen isotope stratigraphy suggest that the age of these sediments may be <15.7 Ma (see "Biostratigraphy" and Strasser et al., this volume). These younger ages are in contrast to the age of the Cocos plate in this vicinity, estimated at ~24 Ma, on the basis of marine magnetic anomalies (Barckhausen et al., 2001). The lower igneous complex (Subunit 4B) is composed of at least seven distinct petrologic units (Morris, Villinger, Klaus, et al., 2003). At all depths, recovered rocks include sparse microcrystalline intervals interspersed with more abundant fine- to medium-grained rocks with occasional coarse-grained horizons. True glassy horizons are extremely rare. Whether described as microgabbros, diabases, or gabbros, the textures of the cored rocks speak to generally slow cooling rates appropriate for large intruded bodies (formed by multiple magma injections in this case) or the deep interior of thick ponded lava flows. With extremely good recovery overall, and in the absence of almost any glassy selvages, we describe the lower unit as also being intrusive. Baked sediments at the top of Subunit 4B are in the same age bracket described by Muza (this volume) for Subunit 4A, implying similar ages for the two intrusive units. Radiometric dating is under way.

In terms of primary mineralogy and composition, the rocks from Subunits 4A and 4B are quite similar. They have abundant plagioclase and clinopyroxene phenocrysts, with ilmenite and magnetite microphenocrysts, often in a largely holocrystalline matrix; olivine and orthopyroxene phenocrysts occur rarely. Samples from Subunit 4A typically have higher concentrations of the more incompatible elements than do those from Subunit 4B, at comparable Mg numbers. Similarly, postcruise work (Dreyer et al., this volume) shows that samples from Subunit 4A have higher rare earth element (REE) concentrations and slightly steeper REE patterns than do samples from Subunit 4B, although all have relatively flat heavy REE (HREE) patterns. These two geochemical groupings are also seen in the rocks recovered from Leg 170, although they were not identified at the time.

Nd isotopic work by Chavagnac, reported in Dreyer et al. (this volume), shows a very restricted range of composition (atom ratios of 0.51293–0.51299, with one sample at ~0.51304), which is important in several regards. Samples from Subunit 4A overlap with those from Subunit 4B, indicating that the differences in element concentrations just discussed are mostly, although not entirely, the result of variations in extent of partial melting and crystal fractionation. The Leg 205 samples have isotopic ratios lower than typical for MORB but similar to OIB values in general and to a subset of Galápagos lavas in particular. Lavas erupted above the Galapagos plume show a wide range of chemical variation and have been characterized as the result of mixing between a depleted MORB mantle (DMM, presumably entrained upper mantle) and plume components. Following Harpp and White (2001) those plume components can include what is inferred to be a long-term enriched lower mantle (PLUME), an incompatible trace element enriched mantle (FLO), and a high U/Pb mantle with elevated Pb isotope ratios (sampled along the so-called Wolf-Darwin lineament). As shown in figure F10 in Dreyer et al. (this volume), samples from other off-axis volcanism in the region (e.g., Fisher Seamount and Cocos-track tholeiites) have either similar isotopic compositions (Fisher) or a wider range of variation encompassing that shown by the Leg 205 samples. The Nd isotopic composition of the mantle source for the Leg 205 samples can be generated by mixing 30%–50% DMM with 50%–70% enriched Galápagos plume mantle (PLUME). There is a hint of a possible FLO contribution to gabbros sampled from Leg 170, Subunit 4A (see Dreyer et al., this volume), although this could also be the consequences of seafloor alteration. Pb isotopic analyses are under way to further test these identifications and mixing models and to test for the presence of the high U/Pb contribution seen in the Wolf-Darwin lineament (Harpp and White, 2001).

Dreyer et al. (this volume) also use geochemical ratios such as La/Sm and Hf/Ta, which are little affected by varying amounts of crystal fractionation, to constrain mantle melting models for the Leg 205 samples. When using the mixture of enriched and depleted mantle sources calculated from the Nd isotopic results, trace element modeling suggests 2%–7% partial melting to form the Leg 205 samples, with Subunit 4A formed from lower degrees of melting than Subunit 4B. The gentle slope of the REE patterns at the HREE end indicate that little if any mantle melting occurred at depths of garnet stability (>60 km), in contrast to models for the majority of the Galápagos Islands, Galápagos Seamounts, and the aseismic Cocos and Carnegie Ridges (White et al., 1993; Hauff et al., 1997; Harpp and White, 2001). Constraints imposed by Nd isotopic compositions, trace element ratios, and REE abundances and patterns are simultaneously satisfied by models that mix depleted and enriched mantles as solids followed by melting; the timing of melting subsequent to mantle mixing is not constrained by the data.

Significant magmatism is present in this section of the Cocos plate, away from the ridges and hotspot tracks. As shown in Figure F4, the incoming plate just south of the Leg 205 sites is described as a seamount province; Fisher Seamount, at 14 Ma, is the northernmost of these. Using the regional distribution of a high-amplitude seismic reflector, which correlates with drilled gabbro sills at ODP Sites 1039, 1040, and 1253, Silver et al. (2004) propose a widespread area of sill intrusion at ~8–10 Ma. Based on results to date for the Leg 205 samples, the modeling suggests upwelling of enriched mantle to depths shallower than 60 km and extensive mixing with ambient depleted upper mantle. This mixed mantle is melted at some later time to produce the lavas erupting at the surface. If additional isotopic work currently under way shows that this upwelling mantle is indeed similar to that of the Galápagos hotspot, then significant input of enriched mantle is seen at distances several hundred kilometers beyond the surface expression of typical hotspot activity. Speculation is that ridge jumps at ~22.7 and 19.5 Ma may have facilitated more distal emplacement of enriched mantle (e.g., Barckhausen et al., 2001) The presence of 14-Ma and perhaps 8- to 10-Ma igneous rock in the region suggests that later events in the plate history trigger melting. One possibility is that decompression melting could be related to a changing stress regime as the site approaches the trench or as the Cocos Ridge intersected the MAT (Silver et al., 2004; Abratis and Woerner, 2001). A speculative alternative is that serpentinization of the mantle of the incoming plate (e.g., Hacker et al., 2003) outboard of the trench could help lower melting temperatures locally. Preexisting zones of weakness in the lithosphere, resulting from prior plate rearrangements, could then provide conduits for magma emplacement. Detailed consideration of paleolocations and forthcoming ages for the Leg 205 rocks will aid this debate.

Alteration of the Cored Section

Alteration in Subunits 4A and 4B is generally low in abundance (1–5 vol%), except at specific horizons (Morris, Villinger, Klaus, et al., 2003). Within the upper sill, alteration is characterized primarily by palagonitization of glass in the groundmass and in veins and primary mineral replacement (most olivine and typically 10% of plagioclase phenocrysts) with clay minerals (saponite and, rarely, a chlorite-group clay). Voids and veins, typically <1 mm wide, are primarily filled by clay with minor zeolites. In the lower igneous subunit, alteration is generally more abundant, particularly below ~509 mbsf. Clay (saponite, a chlorite-group clay, and possibly nontronite) replacement of glass, olivine, and plagioclase is more typically 10%–20% and may be locally as high as 50%. Zeolites are found filling voids and veins. Above ~509 mbsf, the zeolites are laumontite, mesolite, thomsonite, and scolecite. Only below this depth are phillipsite and stilbite identified. Calcite is also present in veins in the lower subunit. A lack of oxyhydroxides and only very minor presence of celadonite could be related to more reducing conditions.

In addition to being more altered, Subunit 4B is also more fractured below ~509 mbsf (Fig. F11). Also below this depth, Formation MicroScanner images show more closely spaced conductive features that are continuous across all four pads at specific horizons (e.g., 513 mbsf). Pfender and Villinger (this volume) use Stoneley wave data from the Dipole Shear Sonic tool used in logging Hole 1253A to identify regions of energy loss in borehole sections with relatively smooth walls. These horizons of reduced Stoneley wave energy are good candidates for high-permeability regions and higher fracture density and correspond well to regions of high fracture density observed in the cores. Based on their analysis, they proposed that fluid flow in the igneous section is most likely at depths of ~468, 492, 500, 508, and 518 mbsf.

Strontium isotope data from Chavagnac for bulk rock samples (Dreyer et al., 2005) further show that significant alteration is restricted to specific horizons. Figure F12 shows that Sr isotope compositions for the Leg 205 samples are generally in the range of 0.7033–0.7035, values that are higher than those of MORB but within the range shown by OIBs (Zindler and Hart 1986). However, pronounced excursions to elevated Sr isotopic ratios (as high as 0.705) are observed at restricted horizons (~10–20 m wide) including those at ~513 and 565 mbsf. Increased uptake of elements such as Cs and Li are also observed at these depths. Slight shifts in Sr isotopic ratios are seen at shallower levels. Following the methods of Albarede (2003), Dreyer et al. (2005) calculated that a seawater:rock ratio of ~6 was necessary to produce the extreme shifts in Sr isotopic composition. The modeling further suggests that the fluids responsible for alteration are modified from modern seawater compositions as a result of their residence time in, and previous interactions with, the igneous section.

The coring, logging, and chemical data suggesting localized intervals for fluid flow are particularly interesting in combination with heat flow data from this region, which averages ~30% of the value appropriate for the plate age and is locally as low as 1% of predicted near the Leg 205 sites and more typically 10%–15% in the vicinity of the drill sites (Langseth and Silver, 1996; Fisher, 2003a, Hutnak et al., 2006). Modeling of both thermal and chemical data suggest vigorous flow of cold fluids through high-permeability horizons in the subducting plate. Not clear is the hydrologic relationship, if any, between the thermal anomaly and the sills. The thermal anomaly is a large regional feature which likely extends beyond the region of sills inferred from seismic imaging and drilling (Langseth and Silver 1996; Silver et al., 2004), although it is greatest in the vicinity of the trench, in the region studied during Legs 170 and 205.

Implications for the Seismogenic Zone and the Subduction Factory

Although not successful in recovering EPR-generated basement of the incoming plate, the results from Leg 205 do have implications for the seismogenic zone and subduction factory. Modeling of thermal contours on the subducting plate shows that equivalent isotherms are shallower on EPR crust than on the adjoining CNS crust (i.e., the thermal perturbations caused by flow of cold fluid through the EPR segment may persist to significant depth and could affect seismogenesis) (Newman et al., 2002; Hacker et al., 2003; Spinelli and Saffer, 2004).

Significant OIB (?Galápagos) overprinting of the incoming EPR crust is important for understanding the flux of mass and elements into the subduction zone and to the depths of magma generation. It is not currently possible to arrive at any reasonable mass balance between MORB and OIB in the downgoing plate. Additional analysis of MCS data (e.g., Silver et al., 2004) in light of drilling results and improved velocity models may provide better constraints. However, we do know that the uppermost part of the incoming igneous section, over perhaps large areas, is composed of rocks with higher abundances of many trace elements (e.g., K, Rb, Cs, Sr, Ba, U, Pb, B, Li, and Be) than is typical of MORB. These shallowest igneous rocks, subjected to higher temperatures during subduction than are rocks deeper within the slab, also have more enriched isotopic compositions. Moreover, the localization of clays, zeolites, and calcites on preferred alteration horizons may make these intervals particularly susceptible to dehydration reactions, which generate fluids that move from slab to the mantle wedge and contribute to mantle melting and magma generation. Although we are not able to predict precisely how high-permeability horizons in the incoming plate would transform during subduction, it is conceivable that they could play a role in fluid transport from the slab to the upper plate across a wide range of depths.

Implications for Long-Term Monitoring

Figure F8 shows that the OsmoSamplers deployed during Leg 205 are located at depths of 497–504 and 512–519 mbsf. They are both within the open hole beneath a packer centered at 473 mbsf. They are thus well positioned to capture fluids inferred to flow along localized horizons, from both the Stoneley wave interpretations and the Sr isotopic results. Mixing of fluids from different horizons in the open hole will certainly occur, meaning that compositions of fluids in the samplers cannot be associated with unique horizons. In recovered fluids, any departure of compositions from seawater values would, however, contain indirect information about fluid-rock reactions in the oceanic crust.

With an emphasis on installing subseafloor observatories, Leg 205 cored only at selected horizons. Samples and results from Legs 170 and 205 together, however, provide a good stratigraphic framework. Figure F3 provides a general overview of the stratigraphy of the oceanic sediment section as sampled on the incoming plate at Sites 1039 and 1253. It also shows the repeated sediment section beneath the décollement at Site 1043 (Site 1255) and Site 1040 (Site 1254), along with the lithology of the sedimentary wedge above the décollement. Results following Leg 170 highlighted the poor age control in the basal sedimentary section on the incoming plate and in the prism above the décollement, both of which are discussed here.

Age Constraints and Accumulation Rates for Mid-Miocene Carbonates

Leg 170 shipboard and postcruise biostratigraphic work at Site 1039 (Ibaraki, 2000; White, 2000; Muza, 2000) show very limited age control through the middle Miocene carbonate section (between ~250 and 350 mbsf). Using Leg 205 coring, approximately one sample per cored sediment section from Site 1253 was analyzed biostratigraphically (Muza, this volume). High abundances and good preservation were typical for this site, except for samples nearest the igneous intrusions, which were either barren or highly recrystallized. Fossil assemblages in sediments both above and below the sill correspond to Martini's (1971) nannofossil Zone NN4, which has a minimum age of 15.6 Ma and a maximum of 18.2 Ma (Berggren et al., 1995), spanning the early/middle Miocene boundary. This same time bracket applies to the baked sediments located just above igneous Subunit 4B.

Strasser et al. (this volume, their fig. F8) point out a large discrepancy between biostratigraphic and magnetostratigraphic age estimates and accumulation rates through the middle Miocene. At a depth of ~360 mbsf, diatom, nannofossil, and foraminifer datums generally agree with each other, with age estimates from ~15.8 to 16.2 Ma as shown in the compilation in Strasser et al. (this volume). Interpolation between biostratigraphic datums suggests ages of 14–16 Ma at depths of 250–350 mbsf. As these same depths, however, magnetostratigraphic ages are 12–14 Ma. Figure F13 shows bulk sediment carbon isotope stratigraphy for Site 1039 (a companion oxygen isotope stratigraphy is shown in Strasser et al., this volume, their fig. F7), where ¹³C maxima are matched to the carbon isotope maxima CM 1–CM 6 from Woodruff and Savin (1991) and John et al. (2003). The carbon and oxygen isotope stratigraphies provide age constraints through the interval that has little or no biostratigraphic control. All isotopic age picks fall along an age-depth curve derived by interpolating between the biostratigraphic ages from above 250 mbsf and below 350 mbsf. They are thus consistent with sedimentation rates in the middle Miocene carbonate-rich Subunits U3B and U3C being approximately constant at ~50 m/m.y. This contrasts with the magnetostratigraphic estimates that suggest rates of ~18 m/m.y. between 16 and 13 Ma, with much higher rates at ~12.7 Ma. Note that the magnetostratigraphy could have been reset by sediment heating associated with sill emplacement.

Constraints on Sill Emplacement

In addition to constraining sedimentation rates, the isotope stratigraphy also tightens age estimates for sill emplacement. Calcareous nannofossils from just above and below the sill yield the same age bracket of 15.6–18.2 Ma (Muza, this volume). A sample from a few meters above the sill in Hole 1039B is identified as CM 2 (= 15.8 Ma) and ¹⁸O Event B (15.75 Ma), suggesting that the younger age of the nannofossil bracket may be appropriate. Subunit 4A, with clear evidence for intrusion into the Miocene carbonate sediments, is thus likely younger than 15.8 Ma. Radiometric dating for the gabbro sill is under way and will test this interpretation. Radiometric ages will show whether sill intrusion here is associated with ~14-Ma volcanism as at Fisher Seamount or an 8- to 10-Ma event proposed by Silver et al. (2004). Dating is also in progress for the multiple sections within the lower Subunit 4B, to define the age spectrum for this lower igneous section. Once in hand, these ages relative to those of the overlying sediments should constrain the current interpretation of the lower subunit as an intrusive event and its temporal relationship to other magmatic events in the region.

Age of the Forearc Sediment Wedge

Age control for forearc sediments above the décollement was extremely difficult using samples from Leg 170 and generally remains so following Leg 205. This difficulty results from sparse abundance of nannofossils, poor preservation, and mixed species of different ages, consistent with a slumping origin for the prism sediments. In Hole 1254A, however, nannofossil assemblages are more diverse and generally more abundant than at Site 1040, especially within the upper cored interval (151–222 mbsf) and between ~351 and 359 mbsf. Although a precise nannofossil zonation for Site 1254 is not possible, Muza (this volume) offers a tentative biostratigraphy. His work suggests that sediments between 150 and 161 mbsf are of early Pleistocene age, sediments between 161 and 216 mbsf are late Pliocene, and those from 219 to 222 mbsf are early Pliocene, being no younger that 3.75 Ma. The lack of marker fossils in most of the sediments in the décollement zone precludes age determinations, but sediments from ~352 to 360 mbsf are early Pliocene in age but no younger than 3.75 Ma. Profiles of ¹⁰Be vs. depth provide indirect age constraints (Morris et al., 2002), due to the ¹⁰Be half-life of 1.5 m.y. As shown in that paper and Morris, Villinger, Klaus, et al. (2003), incoming sediments outboard of the trench and those below the décollement have very high ¹⁰Be concentrations typical of marine sediments. Sediments above the décollement have uniformly low ¹⁰Be concentrations, consistent with the particles being older than 3–5 Ma.

Leg 205 piggybacked extensively on the stratigraphy from Leg 170, shown in Figure F3 (Kimura, Silver, Blum, et al., 1997; Silver, Kimura and Shipley, 2001). Despite the limited coring during Leg 205, a significant amount of geochemistry, sedimentology, and biogeochemistry is emerging from the papers in this volume, combining results for samples from both legs.

Metalliferous Sediments on the Incoming Plate

Basal metalliferous sediments were identified above the gabbro sill (Subunit 4A) at Sites 1039 and 1253 on the incoming plate and in the underthrust section at Site 1040, in the sedimentary prism. At Site 1039, liesegang structures and other visual indicators of hydrothermal fluid flow were noted; enrichments of trace elements such as Cu, Ni, Zn, and V were observed for at least 50 m above the sill (Kimura, Silver, Blum, et al., 1997).

At Site 1253, visual indicators of hydrothermal activity were absent, but chemical signatures are clear. Chavagnac et al. (this volume) use the Ti/Al ratio to show that the bulk of carbonates cored during Leg 205 from Subunit U3C, both above the sill and below it, have similar if not identical detrital sources and proportions. With that established, these authors use the metalliferous index (MI = 100 x Al/Al + Fe + Mn) to assess the major element signature of hydrothermal enrichment. Deep-sea pelagic clays of the Miocene typically have MI > 51 (Kyte et al., 1993), with metalliferous sediments from the EPR and Juan de Fuca Ridge having much lower values, typically 1–10 (German et al., 1997, 1999). The Leg 205 Miocene carbonates show a wide range of MI, from 3.6 to 51.6, with a majority between 24 and 43; lowest values are typically closest to the igneous rocks. Concentrations of metals such as Cu, Co, Zn, Fe, V, and Ti are typically 2–3 times the values measured in background carbonates away from the igneous units, comparable with results from Leg 170.

It is likely that the metal enrichments seen in the basal sediments were imparted soon after sill emplacement, given that the most extreme metal enrichments are typical of those seen in hydrothermal systems. If so, the signature would be unrelated to the reversal in gradient seen in many aspects of the pore water chemistry discussed earlier. These reversals, which generally show a return toward compositions of modern seawater in the basal 120 m, would diffuse away if unsupported by flow in the last 15,000 yr (Silver et al., 2000). There is a hint that hydrothermal activity that affected the sediments can have minor impact on their associated pore fluids, in that Mn concentrations just above and below the sill are about 3 times those measured 30 m above the sill (Morris, Villinger, Klaus, et al., 2003). Sr isotope compositions in the pore fluids just above the sill, however, show no trend or deflection toward values typical of the gabbros, limiting the extent of recent interaction with hydrothermal fluids.

Tephra Chemistry, Arc Evolution, and Sediment Dynamics

Legs 170 and 205 cored abundant ash horizons throughout the entire section. Basal ashes, estimated to be ~15 Ma from biostratigraphy, have chemical characteristics typical of OIB lavas and may reflect explosive activity from the Galapagos hotspot. Clift et al. (2005) focus on chemical analysis of tephra from air fall ash horizons younger than 2.5 Ma to study evolution of the CAVA. The younger age of the tephra makes them less vulnerable to alteration, and all samples in their study were visibly unaltered. Tephra chemistry from cored ash deposits frequently allows study of the largest explosive eruptions and permits the history of arc volcanism to be read further back in time than is easily feasible in land-based studies and with age constraints from biostratigraphy. Direct correlation between tephra and lava chemistry is generally quite difficult, however, in that most lava studies focus on the more mafic eruptives and most tephras are from magma chambers that have evolved to very silicic compositions.

With that caveat clearly in mind, Clift et al. (2005) use tephra chemistry to look at temporal evolution in the CAVA and its link to the balance between sediment subduction and forearc erosion through time. In order to do this, they first use the generally tholeiitic character of the tephras, together with their light REE (LREE) abundances and patterns, to propose that the bulk of the ashes are derived largely from Costa Rica and possibly Nicaragua, rather than Panama.

In considering temporal evolution, several features stand out. The youngest ashes are the most mafic of the entire section. Whether this reflects a trend toward more mafic compositions in the last 15,000 yr or the increased likeliness of depositing more mafic ashes as the incoming plate approaches nearest the trench and arc is unclear. At ~1.45 Ma, an ash horizon is characterized by the lowest Nd isotope ratios and highest ⁷Li yet measured for the CAVA, best explained by large amounts of sediment subduction and recycling to the arc, significantly higher than that recorded by either older or younger ashes (Clift et al., 2005). Ashes from 1.45 Ma to ~250 ka have ⁷Li that are still elevated above values for modern CAVA lavas (Chan et al., 1999, 2002), seen even for bulk tephra samples that were acid washed prior to analysis. Clift et al. (2005) use the detailed Nd-Li isotopic systematics of the ashes to argue that a third component is needed in magmagenesis, in addition to depleted MORB mantle and subducted sediments. This third component must have high Nd and high ⁷Li, like either extremely altered subducted oceanic basalt crust or, as likely for the Nicoya ophiolite complex, which makes up the forearc basement (McIntosh et al., 1993; Kimura Silver, Blum, et al., 1997). In the latter case, tectonic erosion of the forearc basement (e.g., Ranero et al., 2000b; Vannucchi et al., 2001, 2003; Meschede et al., 1999; Clift and Vannucchi, 2004) could deliver materials of the right composition to the depths of magma generation. Recent Costa Rican lavas and the youngest tephras show evidence for a much smaller subducted component than seen in the older ashes (Carr et al., 1990; Patino et al., 2000; Carr et al., 2003; Morris et al., 1990), even though active forearc erosion has been imaged along the Costa Rica–Nicaragua margin. Consideration of the tectonic evidence together with the arc geochemistry suggests that periods of enhanced sediment subduction and subduction erosion may be episodic rather than steady-state features of the margin, perhaps engendered by seamount subduction (von Huene and Scholl, 1991; Morris and Ryan, 2003; Clift et al., 2005).

Pore Fluid Chemistry and Fluid Flow

Pore fluid chemistry determined for samples from Legs 170 and 205 is useful in a number of different ways. At each site (1039/1253, 1043/1255, and 1040/1254), the compositions place constraints on local diagenetic reactions that change pore fluid chemistry, as well as on the larger fluid flow systems affecting each region, as discussed above. Comparison of pore fluid profiles, gradients, and concentrations for selected species between the incoming plate and the prism sites also illuminates the character of the deeply sourced fluid, thought to have advected from depths where temperatures are in excess of 80°–90°C and perhaps higher. They also constrain the fluid-mobile behavior at shallow levels of elements such as Li, Ba, Rb, and Cs, frequently used as important tracers for understanding sediment recycling in volcanic arcs. In the absence of firm constraints otherwise, such treatments in the literature typically assume that the composition of sediments delivered to the depths of magma generation is that subducted beneath the trench, a topic investigated here and in the following section. Numerical modeling of reactive transport, using selected aspects of pore fluid compositions, also provides estimates of flow rates and fluid residence times, linking to the hydrology of the margin. All set the stage for later discussion of long-term monitoring.

Solomon et al. (this volume) investigate changes in pore water Ba and sulfate between the incoming and underthrust sections. Ba in marine sediments may be tied up in barite (BaSO₄), other biogenic phases such as refractory organic matter and carbonate, or inorganic phases such as detrital silicates and Fe-Mn oxides and oxyhydroxides. Ba in aluminosilicates is generally regarded as immobile during diagenesis, but Ba in barite is strongly affected by pore fluid sulfate concentrations (Dymond et al., 1992, 1996; McManus et al., 1998; Torres et al., 1996b). Depletion of pore fluid sulfate via microbially mediated oxidation of organic carbon or methane can greatly increase the solubility of barite, and dissolved Ba²⁺ concentrations can increase by several orders of magnitude (Brumsack and Geiskes, 1983; Torres et al., 1996a; Dickens, 2001). Figure F14C and F14D show that the uppermost part of the incoming sediment section (with 13–25 mM SO₄^2– at Site 1039) undergoes sulfate depletion during earliest subduction (SO₄^2– = 0 in the upper 30 m of the underthrust section at Site 1040/1254). Solomon et al. (this volume) suggest that initial underthrusting severed the supply of sulfate from seawater to surface sediment via diffusion. In the ~17 k.y. required to move between the two sites, sulfate-reducing bacteria utilized the pore fluid SO₄^2– in the upper 30 m, depleting it to zero. This zero pore fluid sulfate horizon would presumably deepen with increasing time and downdip subduction. Figure F14A and F14B shows changes in Ba concentrations in pore fluids in the uppermost sediments between the incoming and underthrust sections. These changes are strikingly complementary to those seen in the sulfate profiles. Where pore fluid sulfate has not been exhausted, Ba pore fluid concentrations are similar at comparable depths in the incoming and underthrust sections. Where pore fluid sulfate concentrations have dropped to zero in the underthrust section, Ba pore fluid concentrations have increased by a factor of ~50 over comparable depths in the incoming section. As the zero sulfate zone progresses downward during subduction, Ba would be liberated from greater depths in the subducting sediment section.

Haeckel (this volume) models the downward propagation of the zero sulfate zone shown in Figure F14 using measured methane and sulfate concentrations above and below the décollement, measured porosity data, and an anaerobic methane oxidation reaction. In the ~17-k.y. time step since the underthrust section passed below the deformation front, Haeckel estimates that 1264 mol/m² of methane is oxidized by the reaction, equivalent to ~15 kg/m² of carbon. The model predicts a barite dissolution rate of 1.2 x 10^–5 mM/yr, equivalent to a Ba²⁺ flux of 4025 mol/m² and sulfur dissolution of 130 kg/m² in the 17 k.y. of earliest subduction.

Kastner et al. (this volume) expand the spectrum of elements measured in pore fluids from the incoming and underthrust sediments and add Sr, Li, and Cl isotope measurements to the data set. Sr isotope ratios measured from basal carbonates on the incoming plate, cored during Leg 205, show the same trend seen in Leg 170 pore fluids and discussed earlier. Pore fluids in basal sediments show a shift toward higher ⁸⁷Sr/⁸⁶Sr values, intermediate between modern seawater and Miocene seawater ratios (see also Kastner et al., this volume, their fig. F4A). Li isotope ratios in basal carbonates show similar systematics, as do profiles for Ca²⁺, Sr²⁺, Li⁺, NH₄⁺, PO₄^3–, alkalinity, and sulfate (Morris, Villinger, Klaus, et al., 2003). Higher Ca concentrations in pore fluids from the basal carbonates are observed relative to pore fluids from higher in the carbonate section; there is a general but not perfect tendency for pore fluid Ca and F concentrations to covary positively. As noted previously, these gradients indicate recent to contemporaneous fluid flow (Kastner et al., 2000; Silver et al., 2000).

Haeckel (this volume) uses reactive transport modeling of sulfate, ammonium, and phosphate profiles vs. depth to investigate two models that could support the gradients seen in the basal pore fluids. One invokes upward advection of distinctive pore fluids through the basal sedimentary section just above the sill. The other envisions lateral fluid flow in the basement, with upward diffusion of solutes through the basal sediments. As the author notes, the models are strongly dependent on the sulfate profile assumed as a starting condition. An advective model can fit the observed data across a wide range of initial sulfate conditions. The best fit to the observed profiles is met with a vertical fluid velocity of 0.15 cm/yr for the past 200–240 k.y., which seems hydrologically unlikely for reasons discussed below. A diffusive transport model can fit the data if the system has been running for ~1 m.y. and if pore fluids through the entire sediment column initially had sulfate concentrations typical of seawater. At the high basement fluid discharge rates modeled by Silver et al. (2000) and Fisher et al. (2003a), either model can easily supply enough sulfate to the basal sediments. The precise nature of permeability distribution in the igneous section, required to link lateral fluid flow in the basement with pore fluid gradients in the basal sediments, remains unclear.

Initial results from Legs 170 and 205 highlighted fluid advection along the décollement at Sites 1043/1255 and 1040/1254. Propane, which is a strong indicator of deeply source fluids because of its thermogenic origin (>80°–90°C required) showed strong peaks within the upper fault zone at ~216 mbsf and in the basal part of the décollement zone at ~355 mbsf. Kastner et al. (this volume) document that many other pore fluid species show dramatic variation in the regions of fluid advection. The halogens F and Br are strongly enriched above seawater values in the upper fault and décollement zone and throughout the region in between; F/Cl and Br/Cl show the same patterns, indicating that variations are not simply a consequence of brine concentration in the fluids associated with local gas hydrate formation. Ca concentrations are highest in this same region and are associated with highest F concentrations in the pore fluids. Lowest ⁸⁷Sr/⁸⁶Sr ratios measured at Site 1040/1254 (Kastner et al., this volume, their fig. F4C) and ⁷Li (Kastner et al., this volume, their fig. 6B) values are seen in the upper fault and décollement. Both isotopic tracers show values in those horizons well below anything measured in the incoming sediment section, indicating a fluid source that is not primarily derived from local compacting sediments. Rather, it is advected from a source with high Ca and lower Sr isotopes than appropriate for fluids in equilibrium with marine sediments. Possible sources for this fluid include the dehydration of subducted altered basaltic crust or of the Nicoya ophiolite complex further downdip. Sharp gradients between the underthrust sediment and the décollement zone, seen in K, Li, Ba, F, and Br concentrations and Sr isotopic composition (Kastner et al., this volume) limit the role of fluids derived by local compaction dewatering and vertical advection from below. Any mixing between deeply sourced and local fluids must preserve the very sharp gradients observed.

Haeckel (this volume) modeled the propane profiles in the upper fault zone and in the décollement zone using data from Leg 205 (Morris, Villinger, Klaus, et al., 2003). The two peaks seen at these intervals are asymmetric, with very abrupt bounds at greater depths and more gradual bounds at the shallower depth. This asymmetry requires a vertical advection rate of ~0.4 cm/yr to match. A lateral flow rate cannot be constrained by modeling, as the width of the zone of fluid advection, rate of propane production at depth, and fluid flow rate were combined into a single source term. A constant propane source term of 22 ppm V/yr across a 10-cm-wide conduit could produce the peaks observed, as could lower rates across wider conduits. In the former case, modeling suggests that the upper conduit has been active for ~2000 yr and the lower for ~4000 yr. These results are consistent with suggestions of transient dewatering at other prisms (Bekins et al., 1995; Saffer and Bekins, 1998), albeit on much shorter timescales. As discussed below, long-term monitoring at Site 1255 includes an experiment for determining fluid flow rates.

Composition of Incoming and Subducting Sediments

Good recovery of the entire incoming sediment column at Site 1039/1253 makes this an excellent reference section for determining the composition of the sediments carried into the subduction zone. Legs 170 and 205 also showed that the entire sediment section of the incoming plate is present beneath the décollement at Sites 1043/1255 and 1040/1254 (see Fig. F3), compacted and dewatered somewhat, but intact and allowing no frontal sediment accretion (Kimura, Silver, Blum, et al., 1997; Morris, Villinger, Klaus, et al., 2003). Building on samples cored from these sites, authors in this volume determine the initial composition of the incoming sediments as well as investigate how the earliest stages of plate subduction, with associated diagenesis and compaction dewatering, may change the composition of the subducting sediments. The correlative changes in pore fluid chemistry are addressed above.

In order to make the comparisons between equivalent horizons on the incoming plate and beneath the décollement, two things are necessary. First is an assumption that sediments at the two sites were of identical composition at the time of deposition (i.e., steady state pertained in sediment deposition and composition over the space and time represented by Sites 1039 and 1040/1254). For comparison, Site 1039 is ~1 km seaward of the deformation front, with Site 1040/1254 being ~1.5 km arcward of the deformation front. At a convergence rate of 88 mm/yr (DeMets et al., 1990) the time difference between the two sites is ~27 k.y. with ~17 k.y. spent in transit beneath the décollement. The second necessity is that the differences in thickness of correlative lithologic units is due only to compaction, justified by previous discussions of lithologic repetition of the lithology outboard of the trench and below the décollement. Estimates of compaction dewatering between Sites 1039 and 1040 were made during Leg 170 and subsequently (Kimura, Silver, Blum, et al., 1997; Saffer et al., 2000). The two estimates are in generally good agreement for Unit U1 (36% compaction) and Subunits U2A (36%), U2B, and U3C (62%). Correlative depths at Site 1040/1254 for the carbonate section Unit 3 are ~15 m deeper using the method of Kimura, Silver, Blum, et al. (1997). In what follows, each set of authors uses their preferred method, meaning that caution should be used in directly comparing the different data sets.

Li and Bebout (this volume, 2005) examine the downhole variations in nitrogen concentrations and isotopes and in carbon abundances and isotopes (in both total organic carbon [TOC] and carbonate) in sediments from both the incoming plate and prism sites to constrain sources for organic matter and paleoproductivity. A first issue to address is possible alteration of original characteristics through diagenesis. Using constraints from the literature and the restricted range of variation in TOC/TN in sediments with large systematic downhole variations in concentration and isotopic composition, the authors evaluate diagenesis as being largely unimportant. Measured C and N isotope compositions imply a dominantly marine source for the organic matter at Site 1039, with mixed marine and terrestrial input of organic matter to the sediments above the décollement.

At Site 1039, TN and TOC increase significantly in the upper 150 m, largely reflecting a change in lithology from carbonate to hemipelagic sediments. Within the hemipelagics, variations in abundance and isotopic composition are noted and attributed to variation in productivity since the Pliocene. Li and Bebout (2005) speculate that these productivity changes relate to closing of the Central American Seaway, most rapid from 4.6 to 2.6 Ma and completed by ~2 Ma, which resulted in changing ocean circulation and upwelling patterns. The nearer approach of Site 1039 to the trench over time could also bring the site into areas of higher productivity nearer the coastline. These authors calculate that sediment subduction feeds ~1.3 x 10¹⁰ g/yr N (mean ¹⁵N = 5.7), 1.4 x 10¹¹ g/yr TOC (mean ¹³C = –22), and 1.5 x 10¹² g/yr oxidized C (mean ¹³C = 1.9) into a section of the MAT ~1100 km long.

N and C isotope systematics of the forearc sediments are reported in Li and Bebout (this volume). TN and TOC in the wedge sediments above the décollement are similar to those measured in the hemipelagic section of Site 1039, although the depth-dependent variation is quite different between the two sites. N isotopes from Site 1040 show a more restricted range than at Site 1039 but with similar average values; ¹³C in the TOC is more negative (1–2) at Site 1040, leading to the identification of a larger proportion of TOC from terrestrial sources. In detail, the downhole variations are different at Site 1040, with maxima in TN, ¹⁵N, and pore fluid ammonium concentrations peaking at ~130–150 mbsf (see Li and Bebout, this volume, their fig. F2). These maxima occur below the lithologic boundary between Subunits P1A and P1B. Aside from pore fluid ammonium, no other obvious changes in sediment or pore fluid chemistry occur at this depth (Kimura, Silver, Blum, et al., 1997). Above this depth, the C-N systematics for the wedge sediments at Site 1040 are more similar to those seen farther upslope at Site 1041 than those at Site 1039, again suggesting that the wedge sediments at the very nose of the prism are most likely derived by slumping or debris flows rather than from paleoaccretion (Li and Bebout, this volume). Considerable scatter of the data points about the best fit lines for downhole variation in C and N concentrations and isotopes makes it difficult to identify small excursions in the vicinity of the upper faults and the décollement. In the décollement zone, TN, ¹⁵N, and TOC increase slightly, toward values typical of the top of the incoming section. In the top of the décollement zone, TOC ¹³C drops to the most negative values measured in either hole.

Strasser et al. (this volume) also examine stable isotope compositions, measuring ¹³C and ¹⁸O in bulk carbonate sediments and comparing values between correlative horizons in the incoming sediments at Site 1039/1253 and the underthrust sediments at Site 1040. ¹³C values measured by Strasser et al. are similar to those measured in carbonate by Li and Bebout (2005). The records of downhole variation at Sites 1039 and 1040 generally overlap each other within analytical error (see Strasser et al., this volume, their fig. F5), with several exceptions. Using decompacted depths for Site 1040 that are correlative to Site 1039 depths, the underthrust sediments have higher ¹⁸O between 228 and 246 mbsf, 288 and 296 mbsf, and 350 and 372 mbsf and lower values between 310 and 330 mbsf. Carbon isotopes for the two cores are different only between 300 and 330 mbsf. The horizons where measurable change in ¹⁸O occurs between incoming and underthrust sections do not differ appreciably from adjacent intervals in terms of lithology (e.g., weight percent CaCO₃) or physical properties but are areas of local maxima in Cl content (Kimura, Silver, Blum, et al., 1997).

As a complement to the study of Ba and sulfate in pore fluids, Solomon et al. (this volume) also examine Ba concentrations in bulk samples of correlative sediments at Sites 1039 and 1040 (Solomon et al., this volume, their fig. F5). In Units U2 and U3, values measured for correlative horizons generally scatter around the 1:1 line, indicating similar values in the incoming bulk sediments and their underthrust equivalents. In Unit U1, four samples show Ba concentrations in the underthrust sediments that are much lower than in their correlatives in the incoming section; one sample shows the opposite relationship. Given the pore fluid Ba profiles, one might expect to see maximum departure from the 1:1 line (i.e., greatest Ba loss from the sediment) in the shallowest underthrust sediments, but no simple systematics between Ba loss and depth are observed. The authors speculate that the variability could be due to varying amounts of barite in the different bulk sediment samples; sequential barite extraction and analysis is under way to test this idea.

Implications for Subduction Recycling

The recognition that barite solubility linked to progressive pore water sulfate exhaustion could liberate Ba to the fluid as the sediments subduct is provocative, particularly in combination with the literature on Ba recycling in arcs. Average Ba concentrations in a number of different arcs show good correlation with the subducting Ba flux, as determined from the composition of the incoming sediments outboard of each margin (Plank and Langmuir, 1993, 1998). Studies of prograde metamorphic suites exhumed from the hanging wall of paleosubduction zones typically show marked decreases in the abundance of elements such as B, Cs, and Pb but not Ba with increasing metamorphic grade and volatile loss (Bebout et al., 1999; Morris and Ryan, 2003). Taken together, the metamorphic and arc chemistry studies suggest that Ba mobility at depth is not a significant effect, whereas Solomon et al. (this volume) show Ba liberation to a fluid during subduction. Understanding the relative amounts of Ba in barite vs. other less soluble phases in the incoming sediments will be important in resolving this apparent contradiction, along with understanding the partitioning of any liberated Ba between the fluid and other Ba hosts in the sediments such as white micas. The possibility that high F concentrations in subduction zone fluids may help complex and transport trace metals normally regarded as immobile also needs investigation (Kastner et al., this volume).

Plate subduction along ~40,000 km of modern-day convergent margins carries large amounts of volatiles into the deeper Earth. Their devolatilizaton behavior as the plate subducts, their contributions to arc magmatism, and their impact on the deep mantle beyond the arc are largely unconstrained at present. Li and Bebout (2005) compared the sedimentary C and N fluxes into the subduction zone along the MAT with volatile fluxes out of the CAVA to better understand the C and N cycling through subduction zones. Estimates for the N flux out of the arc range from 3.6 x 10⁸ g/yr to 8.2 x 10⁹ g/yr, with C flux estimated at 6.9 x 10¹¹ g/yr (Zimmer et al., 2004; Hilton et al., 2002). As expected, there are large uncertainties in the volcanic flux estimates, and the subducted flux calculated above does not include any N or C bound in alteration phases in the oceanic crust or in any material that may be eroded from the forearc. Even so, the sediment flux into the subduction zone is calculated to be at least 2–5 times greater than the volcanic flux out of it. This gives rise to the question of what happens to the rest of the subducting C and N. Shaw et al. (2003) report and use data from Site 1039 and the volcanic arc to argue that ~86% of subducted C is carried beyond the arc to the deep mantle. Data from subduction zone and ultra-high-pressure (UHP) metamorphic rocks suggest that N may also subduct to >90 km depth.

Biomarkers at Leg 205 Sites

With the documented flow of methane, propane, and higher hydrocarbons along the upper fault zones and the décollement, the prism sites from Leg 205 in particular are interesting targets for biogeochemical studies. Cardace et al. (this volume) carried out a variety of biomarker analyses on cellular components extracted form lyophilized sediment to identify deoxyribonucleic acid (DNA) and screen for methanogens, quantify eubacterial and Archeal biomass, and evaluate microbial respiratory strategies (aerobic vs. anaerobic). The bulk of samples reported are from Site 1254, cored only around the upper fault zone and décollement region.

The most notable result is that biomarker indicators of microbial activity range from very low to absent. DNA was identified in half of the samples analyzed; methanogen-specific genes were detected in four samples from the sedimentary prism. Working near detection limits for many analyses, it is difficult to pick out meaningful downhole variations at Site 1254 or possible correlations with sediment or pore fluid chemistry or variations in physical properties such as porosity. Highest concentrations of quinones, sometimes treated as a proxy for total biomass (Hiraishi et al., 1998), are associated with the décollement zone in a region where highest Li concentrations suggest most vigorous flow of deeply sourced fluids. Thus, it's likely that viable methanogenic populations exist within the sediment wedge and décollement, but the extremely low temperatures in this region limit their growth.

The hydraulic and physical properties of the incoming and overlying sediments and décollement zone were important objectives for Legs 170 and 205. Their link to the structural character of the wedge and fluid flow processes were explicitly explored in the Leg 170 Scientific Results volume (Silver, Kimura, and Shipley, 2001; Bolton et al., 2000).

Vertical permeabilities of subducted sediments were measured on 14 samples from whole rounds from Legs 170 (Site 1040) and 205 (Sites 1253, 1254, and 1255) by Screaton et al. (this volume) and McKiernan and Saffer (this volume). The results confirm the low permeabilities of the hemipelagic mud sediments on the order of 10^–16 to 10^–20 m² with on average slightly higher values up to 10^–15 m² for pelagic carbonate samples. Porosities range between 26% and 71% and show a clear inverse correlation with permeability. McKiernan and Saffer (this volume) determined consolidation behavior on samples from Sites 1253, 1254, and 1255 (Leg 205) by constant rate of strain tests.

Screaton and Saffer (2005) use one-dimensional (1-D) and two-dimensional (2-D) fluid flow modeling to explore the impact of rapid loading on fluid expulsion from underthrust sediments as they pass beneath the overlying sediment wedge. In their modeling, they used the measured permeability-porosity relationships and pore pressures inferred from consolidation tests. Their 1-D simulations match observed values if the estimated permeability of the wedge and décollement is greater than or equal to 5 x 10^–17 m². Their 2-D models indicate decreasing pore pressure upward in the underthrust hemipelagic sediments and seaward in the carbonate section, which would tend to drive flow vertically toward the décollement in the hemipelagic sediments and laterally toward the deformation front in the carbonate section. The authors note that some of the pore fluid chemical data discussed previously are not consistent with fluid drainage into the décollement and indicate the need for detailed flow, transport, and reaction modeling to resolve the inconsistencies. Screaton and Saffer (2005) note the lower permeabilities of the Costa Rica margin, in contrast to those of Nankai or northern Barbados. In the latter two cases, high décollement permeabilities correspond to modeled and observed fluid flow rates that are too high to be sustained continuously because fluid loss quickly exceeds incoming fluid supply. In Costa Rica, fluid supplies are much greater, in part because of the high convergence rate (~2 times faster than Nankai and nearly 4 times faster than Barbados). Faster rates and complete sediment subduction beneath Costa Rica mean significantly greater fluid flux (24 m³/yr vs. ~5 m³/yr at Nankai and Barbados). In combination with lower permeabilities, Costa Rica has the potential to maintain continuous high flow rates along the décollement.

Following installation of two CORK-II observatories at Sites 1255 and 1253 during Leg 205 (Morris, Villinger, Klaus, et al., 2003; Jannasch et al., 2003) in October–November 2002, there have been several return visits to collect data and samples. An Alvin dive in November 2002 corrected minor problems that had occurred during installation, ensured proper functioning of the pressure monitoring system, and downloaded initial pressure records. A series of Alvin dives in March 2004 downloaded pressure data from both sites. Dives had also been planned to recover the OsmoSampler packages and deploy new ones at the two sites. For a variety of reasons (Shipboard Scientific Party, 2004), these efforts failed. During an IODP transit in September 2004, IODP Expedition 301T spent 4 days recovering the originally deployed OsmoSamplers and replacing them with new packages including temperature loggers intended to monitor temperature and collect water samples for 3.5 yr. At Site 1253, OsmoSamplers were lost during recovery but could be located on the seafloor close to Site 1253; the upper sampler was safely returned. Given the timing of the various operations, pressure records run from November 2002 to March 2004, whereas temperature and fluid chemistry data are available from November 2002 to September 2004. Details of the installation (Morris, Villinger, Klaus, et al., 2003) and of methods of long-term fluid sampling and pressure and temperature monitoring are described in Jannasch et al. (2003) and in Heesemann et al. (this volume). Interpretation of the first pressure records (Davis and Villinger, 2006) and temperature time series (Heesemann et al., this volume) demonstrate that the installations deliver a wealth of data whose integrated interpretation especially in conjunction with chemical data is still under way.

Objectives for the monitoring are slightly different at the two sites. At Site 1253, the primary goals are to (1) determine temporal variations in pressure and temperature and the state of the igneous basement in order to constrain the role of the basement as a pathway for fluid flow; (2) determine the origin of the basement fluid (seawater recharge, ridge-flank hydrothermal fluids, or from the deeper subduction zone); and (3) characterize fluid fluxes responsible for the low heat flow at this site. Related to the last are goals of determining solute and gas fluxes to seawater and the residual fluxes to the subduction factory. At the décollement Site 1255, the primary goals are to (1) determine pressure and temperature state of the décollement and the hanging wall of the thrust; (2) determine temporal variation in pressure, temperature, and hydrogeochemical fluxes and their response to tectonic events; and (3) determine the source(s) of fluid in the décollement, again with implications for fluxes to seawater and the deeper subduction zone.

Monitoring on the Incoming Plate at Site 1253

Initial borehole results from Site 1253 are compatible with the scenario discussed earlier, of vigorous flow through permeable horizons in the igneous section, with such flow being recent to contemporary. As shown in Figure F15, pressure observations show a relatively rapid decrease in the first 2 months followed by a more gradual decrease throughout the 1.5-yr record. After the initial recovery the upper and lower sensors recorded virtually the same values, indicating that basement is in near-hydrostatic condition despite the proximity to the consolidating subduction complex (Davis and Villinger, 2006). This suggests that it is probably highly permeable and well ventilated to the ocean, consistent with inferences from thermal observations made across the prism and incoming plate (Fisher et al., 2003b; Hutnak et al., 2006) that show the seafloor heat flux to be only a small fraction (as little as 1%) of that expected from the underlying lithosphere. High basement permeability and efficient ventilation at distant basement outcrops are suggested also by geochemical observations made in sediments above the basement contact that show basement water composition to be very similar to seawater (Kastner et al., this volume). Rapid fluid flow within the sediment-buried oceanic crust and efficient advective exchange with the ocean are implied.

Given its ability to maintain a low-pressure state, basement has the capacity to serve as an efficient hydrologic "drain," channelling fluids expelled from the prism and subducted sediments.

The temperature record shows a smooth recovery to a formation fluid temperature of 7.944°C at the end of the end of the recording period (Heesemann et al., this volume). An excursion to lower temperatures occurred when the attempted recovery with the Alvin broke the seal and exposed the borehole to the cooler temperatures of bottom seawater. Heat flow at this site derived from the equilibrium downhole temperature is quite low, ~10 ± 0.6 mW/m² (Heesemann et al., this volume) and in good agreement with seafloor measurements (Hutnak et al., in press).

Temporal variations in fluid composition at Site 1253 yield several provocative results (Kastner et al., 2005). Postcruise measurements show that the formation fluid (Fig. F15) has steady-state Mg concentrations at approximately half of seawater value; Ca is 3 times seawater value. Sr isotopic compositions measured in the fluid (0.70963) are greater than seawater values, past or present. These compositional features show that the fluid could be derived by mixing local bottom water with altered basement fluids sampled by the TicoFlux cruises; Fisher et al., 2003a) just outboard of Site 1253 or by mixing with fluids derived from deeper in the subduction zone (Kastner et al., 2005). In either case, the Sr isotopes suggest that the fluid must also have interacted with sediments containing some older continental component. Such elevated ratios are not measured in pore fluids anywhere in the overlying sediment section at Site 1253. As seen in Figure F15, Ca and Mg concentrations show sharp changes following the original sealing of the borehole in November 2002, and again following the Alvin recovery attempt, which may be used to calculate flow rates. Fluid velocities are estimated at 45 m/yr immediately following deployment and 7 m/yr following the Alvin attempt. It should be noted, however, that additional hydrologic testing and modeling and carbon dating of the formation fluids is necessary before these estimates can be regarded as robust.

The pressure, temperature, and chemical records from Site 1255, shown in Figure F16, are quite remarkable. The décollement and overlying prism are observed to be superhydrostatic during the monitoring period (Fig. F16) but only moderately so. Even the maximum pressure observed early in the observation period, 250 kPa above hydrostatic, was only ~25% of a lithostatic level. Long-term transient decays are observed at both monitoring levels in Hole 1255A. The origin of the initially elevated pressures can only be speculated about. The last major thrust earthquake that occurred in this region in 1950 is an unlikely candidate; some other more recent deformational event, either aseismic (like those witnessed by Protti et al., 2004), or with associated seismicity (as in the case of a nearby Mw = 6.4 earthquake in July 2000), could be responsible. Except at the base of the sediments (which must remain at a near-hydrostatic state for some distance landward of the prism toe), no constraints are provided about levels of overpressure in the underthrust section, where greater overpressures are expected (given their greater hydrologic isolation) and inferred (on the basis of consolidation test results). Targeting the underthrust sediment section for monitoring was precluded during Leg 205 because of the difficulty and time needed to install instrumentation with the technology employed. Consolidation tests (Saffer, 2003; Saffer et al., 2000) suggest the presence of relatively high pressures, but they provide only indirect estimates and provide no information about temporal variations.

Formation temperatures at Site 1255 are 3.58°C at the base of the overthrust prism sediment section and 3.64°C at a level 5 m deeper within the décollement, resulting in a heat flow of ~11 mW/m². Temperature equilibrates at Site 1255 more rapidly than at Site 1253; this could reflect differences in the CORK configurations at the two sites or could be due to a much greater and longer inflow of cold seawater during drilling at Site 1253. In contrast, drilling operations at Site 1255 were much shorter, permeability is much lower, and the formation pressures are superhydrostatic. Unfortunately, the last submersible visit to download pressure data occurred during early March 2004; therefore, no pressure data currently exist for the end of the fluid sampling and temperature records.

Two small stepwise pressure offsets superimposed on the generally smooth decay in pressure occurred during the observation period at Site 1255 in the upper and lower screens (Fig. F16). The first event occurred in late May to early June 2003 and is manifested by a sharp increase in pressure within the décollement of ~30 kPa and a smaller pressure transient in the overlying sediments of opposite polarity. After the first event, the long-term transient decline in pressure in the décollement disappeared and temperatures remained elevated for 2 months. The second event occurred during mid-September to late October 2003. Detailed aspects of these (including inconsistent signs and offsets in timing; an end of decreasing pressure at the décollement at the time of the first step; correlative temperature and geochemical transients) do not lead to any obvious explanation regarding their origin, although a clear association between these pressure steps and two deformational events was observed during a GPS strain monitoring experiment (see Fig. F16) that was under way on the Nicoya Peninsula, directly onshore from the borehole sites, during much of the CORK monitoring epoch described here. Similar correlations, albeit at different times, are observed between diffuse fluid flow and seismic activity in this region (Brown et al., 2005). In both cases, the prism pressure offsets (24 May and 12 October 2003) occurred 2–3 weeks after the initiation of the onshore strain events (Fig. F16). If their origin is local strain, not hydrologically transmitted pressure, the magnitude of volumetric strain suggested would be somewhat greater than 10^–6 (Davis and Villinger, 2006).

The chemical results at Site 1255 are also tantalizing. Figure F17 uses temporal variation in NH₄ concentrations to show that the events affecting pressure and temperature in June and October 2003 also affect the chemistry of the fluids. The decreasing NH₄ during Event 1 is accompanied by increasing Cl at values above seawater, indicating Event 1 is not simply leaking of seawater through the packer seal (Kastner et al., 2005). The changes in fluid composition indicate that these events are not just changes in hydraulic properties of the wedge in the vicinity of the samplers but must include changes in the fluid sources or in mixing proportions between sources. As shown, Event 1 could include a greater admixture of fluids with NH₄ contents similar to those squeezed from the underthrust sediments, and Event 2 incorporates a greater proportion of fluids derived either from prism sediments above the décollement or advected from farther downdip. Strikingly, salinity contents for formation fluids are above seawater values and increase through time (Kastner et al., 2005), which they attribute to in situ gas hydrate formation. Preliminary results from these same authors suggest that the flow meter experiment deployed at this site (Jannasch et al., 2003) worked well and that Events 1 and 2 are associated with significant changes in relative fluid flow rates.