BACKGROUND AND STRUCTURAL SETTING
Recent studies of the processes occurring at convergent plate boundaries have established beyond doubt that fluids play a major role in how accretionary prisms and mountain belts evolve (e.g., Carson et al., 1990; Henry et al., 1989). Tectonic stresses at convergent margins lead to the expulsion of intergranular fluids through compaction of unlithified sediments (e.g., tectonic dewatering [Moore et al., 1986] or shear dewatering [Bray and Karig, 1988], whereas thermal alteration of primary minerals produces excess fluids by dehydration (Moore and Vrolijk, 1992). It is inferred that these fluids play an important role in the initiation and development of the basal décollement—the plate boundary fault—beneath an accretionary prism, commonly deep within subducting sediments. The décollement apparently can act as both a barrier to and a conduit for fluid flow, as it tends to have a strongly anisotropic permeability, being much more permeable along than across the plane of the fault zone. Even deeper than décollement, the possible role of fluid flow in subducted oceanic basement remains to be assessed; where young crust is subducted (as at Nankai Trough), basement should be expected to be quite permeable (e.g., Fisher, 1998) and may support an important component of the overall fluid-flow system. Finally, these fluids are also inferred to be intimately related to earthquake cycles and tectonic stress relief in subduction systems, but in a poorly understood manner.Scientific ocean drilling into accretionary prisms has provided important insights into the mechanisms by which accretionary prisms dewater. Pore-water chemistry, temperature anomalies, and structural observations, particularly at the more mud-dominated Barbados prism, suggest that fluids move mainly through zones with high fracture permeability, specifically faults. In other prisms like Nankai, fluids also move by intergranular flow, probably along stratigraphic horizons (e.g., Hyndman et al., 1993). At the presently nonaccretionary Costa Rica Trench, similar data clearly demonstrate the importance of fluid circulation in the relatively young subducted oceanic basement (Silver et al., 1997). Results of Barbados drilling indicate that the décollement is an active fluid conduit; the fluid geochemistry indicates that the fluid source is deep seated. Rates and directions of fluid flow are generally inferred indirectly from heat-flow measurements and the thermal, geochemical, and isotopic composition of pore fluids (e.g., Fisher and Hounslow, 1990; Gieskes et al., 1990; Vrolijk et al, 1991; Langseth et al., 1990; Le Pichon et al., 1990). Pore-water chemistries from Nankai and several other complexes, which are otherwise quite different, show a strikingly similar pattern of anomalously low chloride concentrations (Taira et al., 1992; Moore et al., 1991; Kastner et al., 1990, 1991; Shipboard Scientific Party, 2000). These low-chloride zones occur as local minima along what appear to be zones of preferential fluid flow, either faults or sedimentary unconformities. At Nankai, they appear as broad minima within the lower Shikoku formation, centered roughly on the décollement (Fig. 4). The source of this fresher water and the mechanism for its transport over long distances in discrete zones is a topic of active debate and modeling efforts (e.g., Bekins et al., 1995; Saffer and Bekins, 1998).