The character of the incoming plate subducting at convergent margins and the processes affecting it as it passes below the shallow forearc may play a major role in the nature and extent of hazardous interplate seismicity as well as the magnitude of volcanism and the chemistry of lavas produced in the overlying volcanic arc. The fate of incoming sediments and ocean crust, and of their associated volatiles, as they pass through the shallow levels of a subduction zone (0–50 km depth) has profound effects on the behavior of the seismogenic zone, which produces most of the world's destructive earthquakes and tsunamis. Fluid pressure and sediment porosity influence fault localization and deformation style and strength and may control the updip limit of the seismogenic zone (e.g., Scholz, 1998; Moore and Saffer, 2001). Fluids within both fault zones and sediments underthrust at the trench affect early structural development and are a key agent in transport of chemical species. The mineralogy and chemistry of any subducted sediments and their dehydration reactions during subduction may control the physical properties of the deeper subduction interface and, hence, downdip limits of the seismogenic zone.
The escape of fluids to the surface from the downgoing plate at depth (return flow) may support a deep biosphere, contributes methane and higher molecular weight hydrocarbons for gas hydrate formation, affects seawater chemistry for selected elements, and is intimately linked to deformation, faulting, and the evolution of the décollement. The distillation and loss of some volatiles and fluid-soluble elements from the shallow slab not only record reactions and processes within the seismogenic zone, but they also play a central role in the supply of residual volatiles to the deeper Earth and change the composition of the slab delivered to the depths of magmatism beneath volcanic arcs. Processes operating in the shallow subduction zone thus affect the way the slab contributes to continent-building magmatism, explosive volcanism, ore formation and, ultimately, the evolution of the mantle through time (collectively known as the subduction factory). The subduction signature recorded in the chemistry of arc volcanics constrains the nature and sometimes the volume of the sediments transported through the seismogenic zone to the depths of magmatism. The arc thus acts as a flow monitor for the transport of sediments to depths greater than those that can be drilled or imaged seismically.
The Ocean Drilling Program (ODP) has identified deformation at convergent margins, fluid flow in the lithosphere, and subduction zone geochemical fluxes as important aspects of the JOIDES Long Range Plan (JOIDES Planning Committee, 1996). The Initial Science Plan for the Integrated Ocean Drilling Program includes an initiative focused on the seismogenic zone. The Central American convergent margin (see Fig. F1) has been a focus area for a number of national and international programs studying the seismogenic zone and subduction factory for several reasons. First, it is one of the few modern subduction zones that is subducting a significant carbonate section and thus provides an opportunity to investigate CO2 cycling through convergent margins. Second, along strike from Nicaragua to Costa Rica (Fig. F2), the style and extent of seismicity and plate coupling changes. Third, along the same section, the style of arc volcanism changes as do volumes and the chemistry of the arc lavas. Changes in both the seismicity and volcanic chemistry have been proposed to result from changes in the balance between sediment underplating, erosion, and subduction (collectively referred to here as sediment dynamics), perhaps related to changing bathymetry, thermal structure, and hydrological behavior along the margin.
Leg 205, building on Leg 170 coring and logging while drilling (LWD) at the same sites (Fig. F3), is designed to investigate the composition of the downgoing plate together with the thermal structure and hydrological activity across the Costa Rica margin. This will be done through a combination of downhole measurements and long-term sampling of fluids and gases as well as monitoring of fluid pressure, temperature, and flow rate at critical horizons. First observations of temporal variations of fluid and gas chemistry will be available once the fluid and gas samples have been recovered 1 to 2 years postcruise. During the leg, we also drilled and cored into the subducting igneous section to characterize the mass and fluid fluxes to the volcanic arc together with their chemical compositions. The high recovery during coring through the décollement zone in the prism sites provided the opportunity to evaluate local heterogeneity in the development of the décollement (in conjunction with Leg 170 results) and to integrate the location and magnitude of pore fluid anomalies with structural fabrics observed in the cores.
A large body of work shows that there are differences in seismicity and arc magmatism along the length of the Central American margin (Fig. F1), with sharp contrasts seen between Nicaragua and immediately adjacent parts of Costa Rica. The Nicoya section of the Costa Rica margin appears to have earthquakes with a moment magnitude (Mw) of 7 or greater at a 40- to 50-year recurrence interval, with the last such event in 1950 (Guendel, 1986). Coupling between the downgoing and overriding plates is estimated from Global Positioning System data to be 40%–60% (T. Dixon, pers. comm., 2001) and appears to start ~15 km arcward of the trench. Nicaragua is characterized by a greater frequency of Mw of 7 or larger earthquakes, including the 1992 tsunamogenic earthquake. In detail, the updip limit of seismicity appears to be at ~20 km depth north of the fracture zone trace shown in Figure F4 and at ~10 km depth to the south (Newman et al., 2002).
There are also significant changes in volcanism between Nicaragua and Costa Rica. Figure F2 shows an offset in the volcanic chain just north of the northernmost Nicoya Peninsula. In Nicaragua, the arc-trench gap is 180–190 km and the volcanoes lie ~180–200 km above the Waditi-Benioff Zone of the downgoing slab. In Costa Rica, the arc-trench gap narrows to ~165 km and the seismic zone is ~120–130 km below the volcanoes (Protti et al., 1994). Nicaraguan volcanoes tend to be smaller than those of Costa Rica as shown in Figure F2; when averaged over the last 100–130 ka, magma production rates, compiled in Patino et al. (2000), are much lower in Nicaragua than Costa Rica (~14 and 44 km3/km arc length per million years, respectively). In the chemistry of the arc volcanics, 10Be data, radiogenic isotopes, and trace element studies of Nicaragua lavas (Tera et al., 1986; Carr et al., 1990; Reagan et al., 1994; Patino et al., 2000) suggest that the entire sediment section is subducting to the depths of magma generation, producing in the lavas a strong signature from the hemipelagic sediments at the top of the incoming sediment section. In contrast, the Costa Rican lavas have a much smaller sediment signature, little contribution from the uppermost hemipelagic sediments of the incoming plate, and a proportionally larger contribution from the basal carbonate section.
The differences in arc chemistry along strike cannot be explained easily by variations in the lithology or chemistry of the incoming plate. Figure F5 shows the lithologic section entering the trench off Guatemala (DSDP 495) and Costa Rica (ODP Sites 1039 and 1253). The two sections show 150–180 m of diatom-rich hemipelagic sediments overlying 230–250 m of calcareous nannofossil oozes and chalks. Given the similarity of the subducting sediment sections along the Middle America Trench, differences in the sediment signature between the Nicaraguan and Costa Rican volcanoes are likely caused by variations in sediment dynamics.
Various workers have suggested that the changing nature of seismicity and arc volcanism along the Middle America Trench may be due to variations in the incoming plate, the fate of incoming sediments as they traverse the forearc of the overriding plate, or a combination of the two. The bathymetry and thermal structure of the incoming plate and the active fluid flow both outboard and inboard of the trench may play a key role in deformation and sediment dynamics across the margin. In addition, differences in origin and chemistry of the subducting oceanic crust may also contribute to changing chemistry of the arc lavas. These issues are addressed in more detail in the following section.
The structure of the Middle America Subduction Zone (Fig. F1) has been intensively studied by North-American and German geophysicists since the early 1980s. The collected data sets comprise bathymetric and side-scan sonar surveys, seismic refraction and reflection studies, and passive seismological experiments (Christeson et al., 2000; McIntosh et al., 1993; Ranero et al., 2000a, 2000b; von Huene et al., 2000). Potential field investigations were mostly done in conjunction with seismic studies and bathymetric mapping. In Barckhausen et al. (1998, 2001), all marine magnetic data from Nicaragua south to the Carnegie Ridge were interpreted to unravel the tectonic history of the Cocos-Nazca spreading center (CNS) and the adjacent Cocos plate off Costa Rica. In addition, recent heat flow surveys at the margin of Costa Rica and Nicaragua (METEOR 54-2, 2002) and seaward on the Cocos plate off Nicoya Peninsula (Ticoflux I, 2001; Ticoflux II, 2002) complement older existing data sets by Langseth and Silver (1996). The regional tectonic history with special emphasis on the subduction erosion is discussed by Meschede et al. (1999a, 1999b), Ranero and von Huene (2000), Abratis and Woerner (2001), and Vannucchi et al. (2001). Pore pressures and the accompanying fluid and energy fluxes as well as the presence of gas hydrates in the area of Costa Rica are investigated in numerous papers (McIntosh and Sen, 2000; Kopf et al., 2000; Silver et al., 2000; Vannucchi and Tobin, 2000; Saffer et al., 2000; Ruppel and Kinoshita, 2000; Pecher et al., 2001, and Saffer, in press). This compilation of published results comprise only recently published papers; more references can be found in Kimura, Silver, Blum, et al. (1997) and Silver, Kimura, and Shipley (2001).
There are significant variations in the origin, morphology, and thermal structure of the incoming plate through the Nicaragua-Costa Rica segment of the Central American convergent margin (Figs. F1, F2). Large-scale tectonic features, such as the Carnegie and Cocos Ridges and the subduction trench, are clearly reflected in the regional bathymetry as shown in Figures F1 and F2. The topography of the incoming plate changes along the length of the margin (Fig. F1) (Ranero et al., 2000b): crust subducting beneath Nicaragua, formed at the East Pacific Rise (EPR), is pervasively faulted with offsets of up to 700 m on back-tilted normal faults (Kelly and Driscoll, 1998; Kelly et al., pers. comm., 2002), possibly associated with extensional tectonics caused by the flexure of the crust as it is subducted. The offsets become smaller as one approaches Nicoya Peninsula, being <200 m. Farther to the southeast, seafloor relief in general is more pronounced (being southeast of the "rough-smooth boundary" after Hey, 1977) and the seafloor is covered by numerous seamounts (Figs. F1, F2). Crustal thickness increases slightly from ~5 km off Nicaragua to ~6 km off Nicoya Peninsula (Ranero and von Huene, 2000). The thickness of the incoming sediments is generally in the range of 400–500 m.
Analysis of marine magnetic measurements (Hey, 1977; Lonsdale and Klitgord, 1978; Barckhausen et al., 2001) shows (Fig. F4) that ~20 km southeast of the Leg 205 transect, a fracture zone trace (FZT) separates lithosphere formed at the EPR from that formed at the CNS. This means that the drill holes of Leg 205 are underlain by crust that formed at the EPR at ~24 Ma. Analysis of magnetic anomaly data (Wilson, 1996) indicates that the crust at this location was formed at a full spreading rate of ~130 mm/yr. Seismic images of the FZT (Barckhausen et al., 2001) confirm the location of the boundary and reveal a jump up in basement depth of ~100–200 m from the EPR to CNS crust. The lithosphere to the southeast of the FZT formed at the CNS, with decreasing age to the southeast. The oldest crust directly at the FZT is 22.7 Ma, which corresponds to the break up age of the Farallon plate. The location for ODP Leg 206 (6°44.19'N, 91°56.06'W) is located on crust generated at the EPR at a full spreading rate of ~200 mm/yr, in the immediate vicinity of the "rough-smooth boundary."
Convergence parameters are broadly similar along the margin. The convergence rate of the Cocos plate vs. the Caribbean plate increases only slightly from 83 mm/yr off Guatemala to 85 mm/yr off Nicaragua and reaches 88 mm/yr off southernmost Costa Rica (De Mets et al., 1990). The convergence direction off Nicoya Peninsula is almost perpendicular to the trench with the subducting plate dipping to the northeast (N25°–30°E). The maximum depth of seismicity gradually becomes shallower from Nicaragua (~200 km) to southern Costa Rica (~45 km). The dip angle of the slab in the upper 100 km is closely similar from Nicaragua to Central Costa Rica (~30°) and becomes steeper at 100 km depth with a value of ~80° (Protti, 1994).
The part of the Cocos plate that is presently being subducted offshore Costa Rica was overprinted by hotspot-related volcanism between 14 and 12 Ma (Barckhausen et al., 2001). This is most evident in the area of the Cocos Ridge off southern Costa Rica, but rock samples dredged from seamounts have proven that the overprinting extended at least as far north as the southern tip of the Nicoya Peninsula (Fisher Seamount). It seems possible that the sills encountered at the base of some of the ODP Leg 170 holes and at Site 1253 of Leg 205 and also observed in nearby seismic reflection profiles may be related to this volcanic event. One other interpretation (U. Barckhausen, pers. comm., 2002) would be that the FZT acted as a "leaky fault" and intruded magma into the bathymetrically lower sediments north of the FZT.
The seismic profile BGR-99-44, shown in Figure F6, reveals the general structure of the seaward side of the trench, the trench itself, and the lowermost seaward part of the prism. All drill sites of Leg 205 are located on this multichannel seismic (MCS) line (C. Reichert and C. Ranero, pers. comm., 2001) shot in 1999. The data were acquired with a 6-km, 1024-channel digital streamer using a 3-m3 tuned air gun array and differential Global Positioning System navigation. Data shown in Figure F6 are a near-trace (171-m offset) time-migrated section of the complete profile. The incoming pelagic sediments with a thickness of ~400 m show faults with offsets of 50–100 m seaward of the trench. A prominent reflector at 0.25-s two-way traveltime (TWT) below the seafloor marks the base of lithologic Subunit U3A (Fig. F5) (Kimura, Silver, Blum, et al., 1997) of late Miocene age. Beneath the sedimentary sequence, the strong reflector at 0.5 s (TWT below seafloor) images the top of a gabbro sill as revealed by drilling results from Leg 170 at Site 1039. The top of oceanic basement below the sill is very difficult if not impossible to identify based on the seismic records.
Through extensive work, the structure of the forearc has been imaged. Limited ODP Leg 170 coring, drilling, and seismic data show that the bulk of the Pacific margin is a wedge-shaped high-velocity body probably made of rocks similar to the igneous oceanic rocks (Nicoya Ophiolite Complex) cropping out along the coast (Shipley et al., 1992; Kimura, Silver, Blum, et al., 1997; Ranero and von Huene, 2000), which precludes the existence of any significantly large sediment mass being recently accreted. Only a small sediment prism (<10 km wide) is located at the front of the margin wedge. Initially, MCS images were interpreted in terms of sediment accretion to the Costa Rica margin (Shipley et al., 1992). More recently, however, and in the wake of Leg 170 drilling, seismic images have been interpreted to show that basically the entire sediment cover of the ocean plate is currently underthrust beneath the margin and that the frontal sediment prism can store very little, if any, of the incoming material (Kimura, Silver, Blum, et al., 1997; Christeson et al., 1999, 2000; McIntosh and Sen, 2000; Moritz et al., 2000; Silver et al., 2000; Ranero et al., 2000a; von Huene et al., 2000).
Part of the prism relevant to Leg 205 is imaged in the seismic line shown in Figure F6. Northeast of the deformation front (shotpoint 3210) the décollement is clearly visible up to 5 km arcward as a boundary separating the underthrust sediment sequence from the overlying poorly structured prism sediments. The detailed analysis of a three dimensional seismic data set (Shipley et al., 1992) shows that across the 8.5-km-wide coverage of the lowermost part of the prism the décollement structure is quite diverse. Shipley et al. (1992) were able to identify numerous thrust faults mostly in the lower part of the prism acting as possible fluid conduits, but clear tectonic structures are less evident approaching the deformation front. Only one of these faults is imaged offsetting the underthrust sequence (Fig. F6) (CMP 3155) and appears to continue up into the prism sediments.
Prior to Leg 170, the Nicaragua margin was believed to be a nonaccretionary margin, whereas Costa Rica was believed to be a site of sediment accretion (von Huene and Scholl, 1991; Shipley and Moore, 1986; Shipley et al., 1992). Leg 170 drilling and subsequent research show no current or recent frontal sediment accretion off Costa Rica. The sediment section beneath the décollement at Site 1040 repeats the complete lithology and sequence of the incoming section cored at Site 1039 (Fig. F5), allowing little sediment accretion to the front of the prism at present (Kimura, Silver, Blum, et al., 1997). Cosmogenic 10Be, which decays with a 1.5-m.y. half-life, also shows that there has been little, if any, frontal accretion at this site over the last several million years (Fig. F7) (Morris et al., 2002). The very high 10Be concentrations in the incoming sediment section beneath the décollement are typical of young marine sediments. Were these incoming sediments to be frontally accreted, the prism sediments above the décollement would have measurable 10Be enrichments. The very low concentrations of 10Be in the sediments above the décollement indicate that they are older than several million years and preclude construction of the prism from accretion of imbricate thrust packets over the last several million years. The prism is thus either a paleoaccretionary prism or is composed largely of slumped slope sediments rather than accreted marine sediments. Sedimentological and chemical data (Kimura, Silver, Blum, et al., 1997) are more consistent with the latter interpretation. The thinning of the underthrust section seen between Sites 1039 and 1040 (Figs. F5, F6) must then reflect compaction and dewatering, rather than sediment offscraping. As noted earlier, the chemical differences in the arc lavas from Nicaragua and Costa Rica suggest that the entire sediment section is subducting to the depths of magma generation beneath Nicaragua, with only the lower part subducting beneath Costa Rica. The seismic and lithologic observations indicate complete sediment subduction past the prism front in both regions. The arc and prism observations can be reconciled if sediments are underplated to the base of the prism beneath Costa Rica or if greatly enhanced subduction erosion occurs beneath the Nicoya segment.
There is evidence for underplating to the base of the prism landward of the Leg 170 and 205 coring area. Christeson et al. (1999) use seismic reflection and refraction data to show stacked velocity duplicates, interpreted as repeated stratigraphic sections because of underplating. The low, but real, 10Be enrichments in the Costa Rican lavas could be explained if the upper 80–100 m of the incoming sediment section was underplated (Valentine et al., 1997).
In addition to evidence for sediment subduction and underplating beneath the Nicoya segment, the seismic stratigraphy of the slope off Nicaragua and the tectonic structure off Costa Rica indicate extension and subsidence of the margin during much of the Miocene (Ranero et al., 2000a; Ranero and von Huene, 2000; Walther et al., 2000). Multibeam bathymetry along the continental slope displays structures that indicate significant mass wasting off Nicaragua and a rugged morphology off Costa Rica (Ranero et al., 2000a; von Huene et al., 2000), which is consistent with tectonic erosion and thinning of the overriding plate. These results are further substantiated by Leg 170 coring and postcruise science. Coring at Site 1042, 7 km landward of the Middle America Trench, encountered a ~30-m-thick sequence of fossiliferous well-lithified calcarenite breccia at a depth of ~4000 meters below sea level (mbsl) (Kimura, Silver, Blum, et al., 1997). Fossil, textural, cement paragenesis, and sedimentological observations document that the calcarenite was formed, brecciated, and cemented in a nearshore setting (Vannucchi et al., 2001). Sr isotope ratios place the depositional age at 16–17 Ma (latest early Miocene) and establish that the breccia section is stratigraphically upright. It is overlain by ~320 m of unconsolidated slope mud showing the complete Pleistocene to Miocene sequence where benthic foraminifers indicate the subsidence of the margin from the upper bathyal to abyssal depths (Meschede, 1999b; Vannucchi et al., 2001). Unfortunately, erosion rates over the last several million years are not well constrained. Speculation is that tectonic erosion has been controlled by the roughness of the subducting plate. The thinning of the overriding plate and the continental margin morphology suggest that subduction erosion increases in intensity from Nicaragua to southern Costa Rica (Ranero and von Huene, 2000).
For the purposes of Leg 205, an important result is that the margin off Nicoya is not currently accreting sediments and has not done so over the last several million years. The conclusion that all sediments fed to the trench over this time frame have been subducted past Site 1040 greatly simplifies estimating the mass and element fluxes. Specifically, changes in underthrust sediment thickness between Site 1039 and 1040 are due to compaction. To calculate the flux of elements out of the compacting sediment section and updip from the deeper décollement, steady-state conditions may be assumed.
Heat flow data from the Global Heat Flow Database within the seaward area off Costa Rica and Nicaragua are sparse with a spacing of typically 60 to 100 km. However even with this very limited coverage, a clear trend in the overall picture is very clear. Values north of the FZT heat flow average ~30 mW/m2 (about one-third of the expected value from lithospheric cooling after Stein and Stein, 1992) and increase to an average value of ~110 mW/m2 south of the FZT. Plates north and south of the FZT have approximately the same age, so one would expect them to show similar average heat flow. This indicates that crust south of the FZT, created at the CNS, is close to the expected value from lithospheric cooling, whereas crust north of the FZT is significantly cooler, which means that a substantial amount of its heat is probably being removed by hydrothermal circulation within the crust. This conceptual model still needs confirmation with more heat flow data and hydrogeological modeling in order to understand the driving forces of the observed heat flow variations.
A detailed heat flow study prior to leg 170, focusing on the trench and the prism offshore Nicoya Peninsula (Langseth and Silver, 1996), confirmed the trend of a cool plate subducting under Costa Rica. Two recent heat flow surveys (Ticoflux I and II) in the Leg 205 area investigated in detail the thermal structure of the incoming plate seaward of the trench by mapping heat flow along seismic lines. Three major conclusions can be drawn: (1) small and isolated areas show either very high or very low heat flow, indicative of active recharge or discharge and (2) profiles across the FZT confirm the increase of heat flow as one steps from the EPR-created crust to the CNS-created crust as suspected from an analysis of the Global Heat Flow Database and (3) a profile seaward of the deepest part of the trench across major extensional faults show no indication that these faults act as major fluid conduits. New heat flow values at the prism from northern Nicaragua to southern Costa Rica (METEOR cruise 54-2) clearly support the idea that the FZT is a major thermal boundary, not only seaward of the trench but also underneath the prism. One profile runs along the transect of holes drilled during ODP Legs 170 and 205 and coincides with seismic line BGR99-44. Figure F8A shows the locations of all seafloor and borehole heat flow measurements along the drilling transect, starting at the trench and up to 34 km away from the deformation front. Data from Leg 170, Langseth and Silver (1996), and METEOR 54-2 are projected along this profile. All data sets generally show a similar pattern of an increase from a low heat flow of ~7–15 mW/m2 to values of ~20–40 mW/m2 passing from the trench to the prism (Fig. F8B). Only the extremely low mean value of 6.7 mW/m2 at ODP site 1040 is not confirmed by recent measurements during METEOR 54-2 that show values twice as high that fit in the general trend of observed heat flow up prism.
At the Costa Rica subduction zone, coring during ODP Leg 170 and subsequent postcruise studies identified three fluid flow systems: active flow of modified seawater within the upper oceanic crust, updip fluid flow within underthrust sediments, and deeply sourced fluid expulsion along the décollement and through faults in the margin wedge (e.g., Silver et al., 2000; Kastner et al., 2000; Saffer et al., 2000; Saffer, in press).
The existence of active fluid flow in the oceanic basement of the incoming plate is evidenced by both the heat flow anomalies discussed earlier and the chemistry of pore fluids sampled during Leg 170. Postcruise studies of pore fluid chemistry (Fig. F9) show chemical characteristics (e.g., 87Sr/86Sr) typical of modern seawater in the uppermost sediments. Below this, values depart increasingly from seawater at increasing depth as a result of diagenesis, largely ash alteration. The consequence is that measured pore fluid isotope ratios at a particular depth-age interval are lower than those of seawater at the comparable time, as determined from the paleoseawater Sr isotope curve. The deepest sediments, however, have pore fluid chemistry that trends back toward modern seawater values. This is particularly evident in the Sr isotope profile, where pore fluid values exceed those intrinsic to the Miocene carbonates in a section almost totally lacking terrigenous input other than young volcanic ash. The heat flow anomaly and pore fluid profiles have been modeled in terms of active fluid flow at rates of ~1–5 m/a (Silver et al., 2000). Although certainly model dependent, these results do indicate extensive contemporary flow of seawater to basement and require high permeability horizons thought to be within the uppermost basement based on results from off-axis drilling along the Juan de Fuca and Mid-Atlantic Ridges (Davis and Becker, 1994; Becker et al., 1997, 1998; Fisher, 1998; Davis et al., 2000; Kopf et al., 2000). In addition to cooling the uppermost part of the plate, the flow may further alter the basaltic crust. Any subducted high-permeability horizons may provide conduits for fluids leaving the deeper subduction zone.
The décollement and faults cutting through the margin wedge of the upper plate serve as pathways for updip flow of deeply sourced fluids. Structural observations across the décollement indicate a zoned fault, with heavily fractured fault rocks above a ductile, plastic zone that may act as an aquitard to segregate the flow systems above and below the plate boundary (Vannucchi and Tobin, 2000; Tobin et al., 2001). At Site 1040, the deformation increases gradually downward, and the shipboard structural geologists placed the top of the décollement zone at 333 meters below seafloor (mbsf). The change from brittle to ductile deformation is present at ~355 mbsf, and the base of the décollement is a sharp boundary at 371 mbsf (Fig. F10). Pore fluid chemistry studies during and since Leg 170 at Site 1040 show three distinct intervals. Above ~190 mbsf, Li, propane, and Ca concentrations are relatively uniform, as they are again in the underthrust sediments below the décollement. Generally high Li, propane, and Ca concentrations are observed between ~200 mbsf and the décollement, with peaks typically coincident with fault zones. Similar chemical anomalies are also seen along the décollement zone at Site 1043 (~130–150 mbsf). Extremely high Li concentrations are observed along the thrust and décollement at Site 1040 (Fig. F10). Enrichments in Ca and Sr, changing Sr and Li isotopic compositions, as well as higher concentrations of thermogenic heavy hydrocarbons (propane to hexane) are also observed. Collectively, these data indicate that some fraction of the fluids sampled along these localized horizons are derived from depths great enough that temperatures are at least 150°C. The chemical variations shown in Figure F10 suggest that the zone between 200 and 370 mbsf is heavily infiltrated by deeply sourced fluids; the sharp peaks at ~200 and 350–360 mbsf indicate that the anomalies are supported by relatively recent advective flow. Note that the décollement zone anomaly approximately coincides with the base of the brittle fracture zone, just above the region of ductile deformation. The composition of deeply sourced fluids from the updip limit of the seismogenic zone may be useful in constraining the mineralogy of sediments at depth and the dehydration reactions that are thought to be important in governing the rheological properties of the subduction interface in the region of seismogenesis. Given that most seismogenic zones will be forever beyond the reach of even a riser drill ship, development of chemical proxies for "remote sensing" of processes occurring in the seismogenic zone is an important adjunct of Leg 205.
Structural and chemical studies indicate a third hydrologic system, which drains fluids from the underthrust sediment section. Results from Leg 170 documented the complete underthrusting of the incoming sedimentary section at the trench, with the important implication that observed changes in sediment thickness and porosity directly reflect the evolution of effective stress. Laboratory consolidation tests, combined with LWD data from Leg 170, show that the subducting sediments are effectively undrained at Site 1043 (Fig. F11) (Saffer et al., 2000; Saffer, in press). At Site 1040, the lower carbonate section (Unit III) remains essentially undrained, whereas the upper hemipelagic units (Units I and II) are partially drained, as shown in Figure F11. Pore fluid profiles from sediments immediately below the décollement at Sites 1040 and 1043 show a distinct chemistry from that of the décollement. A Ba spike immediately below the décollement (Fig. F11) reflects sulfate reduction in the uppermost underthusting section that mobilizes Ba out of barite and into the pore fluids. The very low Ba values in the décollement zone itself indicate that these fluids are not primarily draining into the décollement, indicating little, if any, communication between the hydrologic systems above and below the décollement. Other broad anomalies below and separate from the décollement also imply updip advective flow of locally derived fluids below the décollement (Kastner et al., 2000).
The differences in pore pressure development downsection reflect nonuniform fluid escape (Saffer et al., 2000; Saffer, in press). More rapid drainage of the uppermost ~100 m (Units I and II) than that of Unit III may result from (1) more abundant coarse-grained high-permeability ash layers that focus flow, (2) higher permeability within the hemipelagic sediments, or (3) significant permeability anisotropy within the hemipelagic sediments. The nonuniform dewatering also has important mechanical implications. An inferred minimum in effective stress developed near the top of Unit III between Sites 1043 and 1040 results in a mechanically weak horizon and suggests that detachments may form below the décollement there. This is consistent with the down-stepping of the décollement at 2–3 km from the trench observed in this region (McIntosh and Sen, 2000) and illustrates the role of fluid pressure in mediating structural development. Based on observed changes in porosity, volumetric fluid sources (in Vfluid/Vsediment/s) range from ~10–12/s at the top of the section to ~10–13/s at the base. These values are one to two orders of magnitude larger than those calculated for underthrust sediments at the Nankai and Barbados subduction zones (e.g., Zhao et al., 1998; Screaton et al., in press). This dramatic difference likely reflects higher sediment permeabilities at Costa Rica, resulting in a more active fluid flow system.
Overall, hydrological and geological modeling (e.g., Saffer et al., 2000; Silver et al., 2000) suggests relatively high permeabilities in the oceanic basement, décollement, and underthusting section, with the décollement being locally more permeable than the underthrusting sediments.