Cocos Ridge extends from the Galapagos hotspot to Osa Peninsula (Figs. 1 and 3). We summarize our discussion in a compilation of Central American studies (Ranero et al., 2007) and then summarize the extensively studied vertical motion on land associated with this collision (cf. Gardner et al., 1992).
Observations
At the base of the Costa Rican margin, the 20-km- to 25-km- thick Cocos Ridge crust was produced ~14 m.y. ago by intru- sion of magma and extrusion of volcanic materials from the Galapagos hotspot (Werner et al., 1999). Hotspot magmas were underplated and vented through an ocean crust that formed at the Cocos-Nazca spreading center (Barckhausen et al., 2001). Near the base of the margin, the ridge crest is a graben trending par-
allel to the ridge with up to 1-km-high normal fault scarps (at bottom of Fig. 4). The crestal graben indicates extension normal to the ridge axis, and it can be traced 200 km seaward toward the Galapagos Islands. Sediment in the graben is ~1 km thick adjacent to the slope and is little deformed (Hinz et al., 1996). On its northwest fl ank, the seafl oor has small ridges, seamounts, and the Quepos Plateau (Fig. 4). Samples from seamounts and the plateau that were isotopically dated are 15–13 m.y. old (Werner et al., 1999) and were emplaced on 21–18 m.y. old lithosphere (Barckhausen et al., 2001). Heat fl ow recorded with multiple probe measurements shows an elevated temperature on the ridge compared with the host lithosphere (Grevemeyer et al., 2004). The age and geochemistry of the seamounts are essentially the same as the rock forming the Cocos Ridge crest, indicating coeval Galapagos hotspot emplacement. Laterally adjacent to the ridge, the normal 7–8-km-thick ocean crust (Ye et al., 1996; Walther, 2003) bends into the trench more steeply than thickened crust of the ridge crest. This explains the extensional tecton- ics forming the graben parallel to the ridge axis. Where Cocos Ridge subducts beneath the margin, the base of slope is 1 to 2 km deep, whereas 200 km northwest, it reaches 5 km depth. Ridge collision has introduced a broad swell ~2 to 3 km high into the subduction zone. However, the area of signifi cantly thickened crust represents only part of the ocean crust affected by hotspot magma, and a sill of Galapagos composition was found at ODP Site 1039 some 250 km northwest of the crest (Kimura et al., 1997). Despite the lack of clear topographic expression, the hot- spot magma forming the ridge has affected a much broader area
Figure 3. Regional topography and bathymetry off Costa Rica looking ENE showing subduction of Cocos Ridge opposite Osa Peninsula. Cocos Ridge and its associated seamounts are in the foreground. The high mountains opposite the collid- ing ridge (Talamanca Cordillera) are landward of the collision area.
of the adjacent seafl oor than is apparent from morphology and geophysical data. Cocos Ridge in its present form is estimated to have arrived at the southern Costa Rica margin ~3–5 m.y. ago based on paleontological and geochemical studies of unroofi ng and uplift (Gräfe et al., 2002; Collins et al., 1995). Because it trends at only a small angle to the convergence vector, the area of collision has been essentially stationary for roughly 3–5 m.y.
Seamounts on the ocean plate associated with the ridge are nominally 20 km across and 2 to 2.5 km high. Seismic images of seamounts subducted to 10 km depth are not resolved well enough to give a precise height, but the seamount’s relief appears to be roughly equivalent to that of the exposed ones. Earthquakes and their tightly clustered aftershocks have been recorded above sub- ducted seamounts and on a projection of the subducting Quepos
Plateau, indicating that ocean relief remains attached to the sub- ducting Cocos Ridge at least to the shore (DeShon et al., 2003). An extension of the plateau remains attached in the seismogenic zone, as do seamounts such as the one that was the asperity for the 1992 Cobano earthquake (Bilek et al., 2003). Ocean fl oor relief constructed on crust with only thin pelagic sediment as off Costa Rica could be more fi rmly attached to the lower plate than relief built on thick ocean fl oor sediment. Most seamounts sampled returned largely pyroclastic materials, and subduction of an edifi ce constructed of such weak materials indicates a weak interplate fault (von Huene et al., 2000).
On land, studies of the Quaternary geology along the coast document uplift above subducted seamounts, and a very well- documented example is the uplift at the southern end of the
84° 20' 84° 00' 83° 40' 83° 20' 8° 00' 8° 20' 8° 40 ' 9° 00' 9° 20' Lake Nicaragua Nicoya Peninsula Costa Rica Osa Peni nsula volcan ic front bend-faults Quepos Plateau Osa Peninsul a Fila Costeña Talamanca Cordillera Cocos Rid ge Paleo- Plate Boundary San Jose
Figure 4. Perspective of multibeam ba- thymetry and topography looking NW. This shows the retreat of the continental slope where abundant relief associated with the Cocos Ridge subducts as com- pared to the NW area of horst and gra- ben, where little relief is subducted. The Cocos Ridge crest is marked by the sed- iment-fi lled graben paralleling the ridge axis. Since convergence is essentially parallel to the ridge, it has not migrated signifi cantly in the past 3–5 m.y.
Nicoya Peninsula (Gardner et al., 2001). Uplift rates of up to 8 mm/yr near the inferred top of the seamount have been reported. Tilted wave-cut erosional terraces are produced even where the upper plate is 25 km thick; however, the tilt is attenuated because of the upper-plate thickness above the subducted relief.
Above the subducted Cocos Ridge, there is the broad uplifted arch of the Osa Peninsula and its adjacent seafl oor (von Huene et al., 2000). Uplift is also well defi ned by uplifted wave-cut ter- races on land (Fisher et al., 2004; Gardner et al., 1992; Sak et al., 2004; Sitchler et al., 2007, 2009). A topographic lineament across the peninsula on a projection of the crestal graben reveals that large features of the ridge crest, despite attenuation, are per- haps still visible in the topography of the peninsula (Fig. 4, bot- tom). Extensive identifi cation and dating of raised wave-cut plat- forms on Osa Peninsula show uplift rates of 8 mm/yr, as well as a period of subsidence (Sak et al., 2004). Uplift and erosion expose the Osa mélange (cf. Vannucchi et al., 2006), which was buried at depths where temperatures are 110 °C. Rapid uplift began ca. 3 Ma and is greatest where the crest of Cocos Ridge subducts. About two-thirds of the 3- to 5-km-thick uplifted sections have been eroded, exhuming the mélange. Rates of uplift are greatest on a projection of the subducting crest of the ridge.
Landward of the peninsula and opposite the full width of the hotspot-affected ocean crust, the Fila Costeña fold-and- thrust belt displays active contraction along as many as fi ve thrust slices. The Fila Costeña extends parallel to the regional trend and rises against the Talamanca Cordillera (Figs. 3 and 4). The Talamanca mountains reach 3700 m elevation (Gard- ner et al., 1992) and expose midcrustal granodiorite plutons and abundant Miocene intrusive rocks (de Boer et al., 1995). The maximum elevation of the Talamanca, the maximum deforma- tion of the Fila Costeña, and the highest crest of Osa Penin- sula are aligned with the crest of Cocos Ridge. Arc volcanism is now almost completely inactive. Large earthquakes of Mw 7–7.5 have nucleated here, but in the past 20 yr, interseismic earthquakes are fewer than in adjacent segments (Protti et al., 1995). As would be expected, subduction of thickened crust is associated with more rapid uplift than above the thinner crust under the lower seafl oor relief on the northwest fl ank.
An ~230-km-long wide-angle seismic transect across south- ern Costa Rica just north of Osa Peninsula extends from the Pacifi c basin to the Caribbean slope (Stavenhagen et al., 1998). Margin wedge seismic velocities increase landward in the com- mon fashion. A moderate velocity reversal occurs within the wedge above the plate interface, probably from a subduction channel containing fl uid-rich material (Ranero et al., 2008). The plate interface can only be followed to a depth of nearly 40 km beneath the Talamanca Cordillera. The Moho could not be identi- fi ed unambiguously. The refraction section (Stavenhagen et al., 1998) and earthquakes (Protti et al., 1995) show no evidence of a fl at slab as occurs along other margins where ridges subduct, and a volcanic front is absent. A receiver function study (Dzierma et al., 2007) is consistent with the refraction interpretation. How- ever, beyond the 40-km-deep plate interface where the junction
between the Fila Costeña and Talamanca Cordillera occurs, plate interface dip increases from 16º to 30º and is imaged to 170 km depth. Under these conditions, the usual explanation for extin- guished volcanism from a fl at slab is problematic.
Ideas Regarding Tectonism
The pronounced landward shift in the continental slope along which Cocos Ridge subducts indicates accelerated retreat of the central Costa Rica margin opposite the subducting ridge fl ank and its associated seamount-studded seafl oor. If so, the margin has retreated ~25 km more than the Central America margin to the northwest in ~3–5 m.y. (Fig. 4). Coeval with accelerated ero- sion, uplift of Osa Peninsula above sea level has occurred. Before that uplift, the peninsula probably had a cover represented by the section now exposed on the peninsula. The cover of younger sediment and other materials has been eroded, exposing Miocene to Eocene rocks (Vannucchi et al., 2006); however, the depth to which these rocks were once buried is still debated (Buchs and Baumgartner, 2003). The Cocos Ridge crest subduction appears to be accommodated by block tilting without obvious compres- sional faulting and folding of the margin slope, except for a small fold at the deformation front where graben sediment subducts. The collision area of the margin opposite Cocos Ridge is disrupted less than that where seamounts subduct, indicating a low-friction environment (Fig. 4). The thick low-velocity layer, thought to be extrusive volcanic material, may have higher than normal poros- ity. One explanation for low friction is an abundance of fl uid.
The lower-plate relief subducting beneath the Costa Rica margin is of various heights and lengths. Smaller features show the beginning stages of collision. Subducting seamounts leave a trail of local uplift and subsidence across the slope and shelf that are quickly obscured by sedimentation. They uplift the land sur- face even where the upper plate is 15–20 km thick, and attenua- tion of deformation above a subducting seamount reduces the sur- face uplift to ~1/10 of the seamount height. Subducted seamounts associated with the Cocos Ridge form asperities for earthquakes up to M 6.5 (Bilek et al., 2003).
Ridges also form asperities for earthquakes like the 1999 M6.9 earthquake north of the Osa Peninsula (DeShon et al., 2003). This earthquake nucleated on the probable subducted extension of Quepos Plateau. The plateau subducts without much disruption of the slope and causes local slides at the shelf edge associated with an indentation indicating loss of material from subsurface tectonism. On the shelf, it raises the seafl oor in a very low ridge that is not visible with conventional bathymetry.
The crestal graben morphology of Cocos Ridge subducts easily at least at the scale of multibeam bathymetry (~100 m), and its morphology is refl ected in topography across the penin- sula where the upper plate is relatively thin. The ridge’s thick axial crust bends less than thinner ridge fl ank crust as they enter the subduction zone, as shown by the shallower dip beneath the crest than the fl ank (Protti et al., 1995). Farther inland, compres- sional deformation of the upper plate is greater above the ridge
crest than above its fl anks. Compressional structure begins at the Fila Costeña, indicating where subduction zone friction becomes strong enough to signifi cantly deform the upper plate. As this col- lision indicates, when positive ocean relief has subducted in one spot for 3–5 m.y., signifi cant upper-plate deformation and moun- tain building results, as long as the ridge does not leave the area.