Cracking of lithosphere north of the Galapagos triple junction

Hans Schouten Department of Geology and Geophysics, Woods Hole Oceanographic Institution, Woods Hole, Deborah K. Smith Massachusetts 02543, USA Laurent G.J. Montési Department of Geology, University of Maryland, College Park, Maryland 20742, USA Wenlu Zhu Emily M. Klein Division of Earth and Ocean Sciences, Duke University, Durham, North Carolina 27708, USA

ABSTRACT evolution that probably affect terrestrial as The Galapagos triple junction is a ridge-ridge-ridge triple junction where the Cocos, Nazca, well as oceanic rifts. and Pacifi c plates meet around the Galapagos microplate. Directly north of the large scarps of Lonsdale (1988) proposed that the GTJ is the Cocos-Nazca Rift, a 250-km-long and 50-km-wide band of northwest-southeast–trending made up of a set of four rifts: the East Pacifi c cracks with volcanics at their western ends crosscuts and blankets the north-south–trending Rise, Incipient Rift, Cocos-Nazca Rift, and abyssal hills of the East Pacifi c Rise. It appears that the roughly northeast-southwest exten- Galapagos-Nazca Rift (referred to herein as Dietz sion of East Pacifi c Rise–generated seafl oor has been accommodated by a succession of minor Deep Rift), which enclose a rotating Galapagos rifts that, during at least the past 4 m.y., had their triple junctions with the East Pacifi c Rise microplate (Fig. 1A). The current tip of the at distances of 50–100 km north of the tip of the propagating Cocos-Nazca Rift. We propose Cocos-Nazca Rift (Hess Deep) does not intersect that the rift locations are controlled by stresses associated with the dominant Cocos-Nazca the East Pacifi c Rise; instead there are two ridge- Rift, and scaled by the distance of its tip to the East Pacifi c Rise. We speculate that similar ridge-ridge (RRR) intersections at 1°10′N and ephemeral rifts occurred south of the Cocos-Nazca Rift and were instrumental in the origin of 2°40′N, where Dietz Deep Rift and Incipient the rotating Galapagos microplate ca. 1.5 Ma. Rift, respectively, meet the East Pacifi c Rise. Lonsdale’s (1988) model for the GTJ has a Keywords: Galapagos triple junction, plate boundaries, lithospheric stress. simple major plate RRR triple junction until ca. 1.5 Ma, when Dietz Deep Rift developed at INTRODUCTION lithosphere responds to stress. In this paper we the East Pacifi c Rise, forming a short, east-west– Distributed deformation at oceanic triple investigate the kinematic history and nature trending spreading center. Over time, Dietz junctions shows that the lithospheric plates of distributed deformation at the Galapagos Deep Rift propagated northeast, approaching undergo signifi cant internal deformation triple junction (GTJ) (e.g., Bird et al., 1999; the southern scarps of the Cocos-Nazca Rift. as their boundaries rapidly evolve (e.g., Klein et al., 2005; Lonsdale, 1988; Searle and The Galapagos microplate took on its own Lonsdale, 1988; Mitchell, 1991; Mitchell and Francheteau, 1986; Zonenshain et al., 1980), motion, a clockwise rotation about a vertical Livermore, 1998). The nature of the deforma- a region that reveals the fundamental inter- axis, ca. 1 Ma; Incipient Rift also began opening tion provides important constraints on how the actions between stress and plate boundary around that time.

100W 80W A 10N Cocos B 102°00′W 101°40′W 101°20′W 3°00′N 1 My

0 Pacific 10S Nazca

102°W 101°W 100°W 2°50′N

ER EPR 3°N EPR C TJ IR B Figure 1. A: Location maps showing major NGMP C-N 2°40′N 2°N TJ tectonic features of Galapagos triple junction GMP (after Karson et al., 2002). GMP—Galapagos Rift scarps microplate, NGMP—North Galpagos micro- ′ 1°N TJ DDR 2°30 N plate, ER—Extinct Rift, IR—Incipient Rift, DDR—Dietz Deep Rift, C-N—Cocos-Nazca spreading center, EPR—East Pacifi c Rise, C 101°20′W 101°00′W 100°40′W 100°20′W TJ—triple junction. B: Multibeam bathymetry ′ of Incipient Rift. C: Multibeam bathymetry of 3°20 N Extinct Rift. Shading—regions of volcanic highs and fl ows. White dashed lines— graben extending southeast from the vol- 3°10′N ? canic highs. White solid line—1 m.y. iso- ? chron (from Lonsdale, 1988). Red boxes show locations of maps in B and C.

3°00′N Extinct Rift

2°50′N

10 km m –3500 –3000

© 2008 The Geological Society of America. For permission to copy, contact Copyright Permissions, GSA, or [email protected]. GEOLOGY,Geology, May May 2008; 2008 v. 36; no. 5; p. 339–342; doi: 10.1130/G24431A.1; 4 fi gures. 339 A 102°W 101°W 100°W the Cocos-Nazca Rift and was instrumental in the development of Dietz Deep Rift and the origin of the Galapagos microplate ca. 1.5 Ma.

EPR NORTHERN TRIPLE JUNCTIONS At the western end of Incipient Rift (Fig. 1B), 3°N 1 My Extinct Rift a broad volcanic high extends east from its triple junction with the East Pacifi c Rise for ~15 km, where it changes to a narrow graben (~200 m deep, ~4 km wide) extending southeast (Figs. TJ Incipient Rift 1B and 2). Excess volcanism blankets East Pacifi c Rise–generated crust to the north and south of the volcanic high (Klein et al., 2005). The overall shape of Extinct Rift is similar to that of Incipient Rift: a volcanic high and fl ows 2°N blanketing East Pacifi c Rise–generated abyssal Water depth (m) hills changing to a graben trending southeast 50 km B –4000 –2000 (Figs. 1C and 2). This suggests that the west- ern end of the volcanism marks the location of the former junction of Extinct Rift with the EPR East Pacifi c Rise. Because Extinct Rift is deeper (to 600 m) and broader (~14 km) than Incipi- 3°N ent Rift, we infer that it opened for a longer Extinct Rift period of time (>0.5 m.y.) before it was aban- Rift scarps doned ca. 1.5 Ma (inferred from the location of its tip). If Extinct Rift bounded an edge-driven TJ microplate, the lithosphere to its south should have been noticeably rotated. As at the Incipi- ent Rift, however, no evidence of seafl oor fabric HD Cocos-Nazca SC rotation is observed, which challenges the model proposed by Klein et al. (2005) of a succession of rotating edge-driven microplates north of the 2°N 102°W 101°W 100°W Cocos-Nazca Rift. Between Incipient and Extinct Rifts and far- Figure 2. A: Multibeam bathymetry and global seafl oor topography (Smith and Sandwell, ther to the east, we recognize multiple volcanic 1997) north of Cocos-Nazca Rift. Red dashed line—inferred trace of northern triple junctions. ridges (Figs. 1B, 1C) that change to graben along B: Interpretation of features from A. Dark shading—outline of excess volcanism. Light gray shading—Cocos-Nazca Rift. Bold black lines—rift systems. Light black lines—lineations. their strike. The western edge of volcanism that Red dashed line—same as in A. SC—spreading center, HD—Hess Deep, TJ—triple junction, blankets and abuts the north-south–trending abys- EPR—East Pacifi c Rise. sal hills (Fig. 2) is identifi ed as the trace of many short-lived rifts where they intersected the East Pacifi c Rise. The associated graben have different It has been diffi cult to understand what role Klein et al. (2005) also described Extinct widths, and thus presumably opened for varying Incipient Rift plays in the triple junction kine- Rift, a trough that trends parallel to the Incipi- lengths of time. In places where volcanic highs matics. Lonsdale (1988) and Lonsdale et al. ent Rift, ~100 km to the northeast. They inter- are observed and graben are not well defi ned, (1992) concluded that Incipient Rift was a preted Extinct Rift to have been a larger version we infer that volcanism occurred along an initial small westward-propagating rift. In contrast, of Incipient Rift and suggested that the GTJ has crack, which was then quickly abandoned. Klein et al. (2005) concluded that the Incipient undergone a long and complex history of plate Thus it appears that Incipient Rift is just the Rift opens about a pivot at its eastern end. reorganization with the development and aban- latest and Extinct Rift perhaps the largest of a And, what was previously thought to be a donment of rotating microplates north and south sequence of more than 20 southeast-trending rifts single Galapagos microplate could, in fact, be of the Cocos-Nazca Rift. or cracks that progressively stepped southward composed of two counterrotating microplates: We present a new interpretation of the evolu- and westward and successively accommodated the northern portion of Lonsdale’s Galapagos tion of the GTJ. We observe that during the past approximately north-south extension of East microplate (North Galapagos microplate), 4–5 m.y., north-south–trending abyssal hills Pacifi c Rise–generated crust. At ~101°20′W, and the remaining portion of the microplate. of the Cocos plate have been cut and volcani- the strike of the inferred triple junction trace Applying the edge-driven microplate model of cally overprinted by a succession of short-lived changes abruptly from westward to southward. Schouten et al. (1993) to their proposed dual northwest-southeast–trending minor rifts and At ~101°35′W it changes back to westward. We microplate system, Klein et al. (2005) estimated associated triple junctions, but without signifi cant observe that a more westward trend implies a that the North Galapagos microplate would rotation. We suggest that the rift locations are con- relatively stationary triple junction, giving rise have been subjected to ~7° of rotation since its trolled by stresses associated with the dominant to more prolonged rifting at a steady location, initiation ca. 0.5 Ma, an angle considered to be Cocos-Nazca Rift, and scaled by the distance of and that a southward trend implies that rift- too small to be resolved conclusively from the its tip to the East Pacifi c Rise. We speculate that ing and triple-junction vol canism rapidly and seafl oor fabric. similar short-lived rifting also occurred south of repeatedly jumped south.

340 GEOLOGY, May 2008 AB102°W 101°W 100°W of reduced tension and perhaps more diffi cult magma extraction. The East Pacifi c Rise in this region does not have the typical fast-spreading Extinct Rift 3°N axial rise morphology (e.g., Lonsdale, 1977, 1988; Searle and Francheteau, 1986). Instead,

Free boundary Incipient Rift it looks more like a slow-spreading ridge Maximum TJ with horst and graben terrain. It remains to 1.4D be determined whether this morphology is a Crack product of the stresses induced by the propa- D 2°N gating Cocos-Nazca Rift. Maximum DISCUSSION ep Rift Directly north of the large scarps that mark Dietz De 50 km the propagation of the Cocos-Nazca Rift, a TJ 1°N 250-km-long and 50-km-wide band of roughly Water northwest-southeast–trending cracks with vol- depth (m) –4000 –2000 canics at their western ends, crosscuts and Figure 3. A: Stress enhancement for a simple crack model. Warm colors (red through yellow) blankets East Pacifi c Rise–generated abyssal are enhanced tensile stress, and cool colors (cyan through blue violet) are reduced tensile hills (Fig. 2). This indicates that 50–100 km stress. Stress enhancement is the stress fi eld normalized by the stress expected in the north of the Cocos-Nazca Rift, East Pacifi c absence of the crack. Contours range from 0 to 0.4 at intervals of 0.02. D—distance of tip of Rise–generated lithosphere has been extended the crack from the free boundary. Tensile stress maxima occur at ±1.4D. B: Contours of stress roughly north-south by a succession of minor enhancement overlain on topography. The tip of model crack is positioned over tip of Cocos- Nazca Rift, and the model is rotated to match the orientation of the spreading center. Incipient rifts since at least 4 Ma. Rift is located at a tensile stress maximum at the East Pacifi c Rise. TJ—triple junction. A schematic evolution of the GTJ for the past 2 m.y. is shown in Figure 4 (A–E). Based on the results of our numerical model, we envi- CRACK INTERACTION MODEL A similar model has been used to explain sion that each of the cracks and graben north We hypothesize that the cracking at Incipi- the development of graben above non-erupting of the Cocos-Nazca Rift initiates as a crack at ent Rift and the other extinct rifts relates to the magmatic dikes (e.g., Mastin and Pollard, 1988; the East Pacifi c Rise. As the crack propagates westward propagation of a Cocos-Nazca Rift Pollard and Holzhausen, 1979). The diking eastward, eruptions build a localized volcanic that never reached the East Pacifi c Rise. We use models assume plane strain and loading caused high and fl ows blanket the seafl oor. Currently, a simple crack interaction model to understand by the internal pressure of the dike; in contrast, the volcanic high at Incipient Rift is shallower the controls on the location of the northern triple we assume plane stress and loading caused by than the adjacent rise crest, suggesting that junctions. An initial numerical model is con- remote displacement. In both cases off-axis melt may be diverted preferentially from the structed in map view with the commercial fi nite stress maxima develop. East Pacifi c Rise to Incipient Rift. If the tip of element software COMSOL® Multiphysics A zone of reduced tension develops ahead the Cocos-Nazca Rift remains at a steady dis- (COMSOL, 2006) (Fig. 3A). The model fol- of the crack and tensile stress maxima develop tance D from the East Pacifi c Rise, the loca- lows the plane stress approximation appropri- along the East Pacifi c Rise ~±1.4D from the rift tion of the maximum tensile stress stays steady ate for the lithosphere in the absence of surface tip. It is at these locations that the fi rst crack not at 1.4D north of the Cocos-Nazca Rift while loads over long time scales (e.g., Turcotte and connected to the propagating Cocos-Nazca Rift a stable triple junction moves away from the Schubert , 2002). This approximation differs is expected to appear. Near the new crack, the Cocos-Nazca Rift tip with roughly half from the study of stress interactions at Hess stress along the East Pacifi c Rise decreases and the Cocos-Nazca spreading rate. At some point, Deep by Floyd et al. (2002), who considered the crack propagates to the southeast following a stable triple junction will have moved too far an infi nite half space to understand these inter- the tensile stress maxima. As seen in Figure 3B, north for continued opening and a new crack actions at a time scale of <1000 yr, for which the distance 1.4D along the East Pacifi c Rise will initiate to the south (at 1.4D), forming a viscous relaxation does not occur. from the tip of the crack corresponds closely to new RRR triple junction. The abandonment In our model, the Cocos and Nazca plates are the location of Incipient Rift with respect to the of the triple junction will occur more quickly approximated by rectangular areas. Their west- Cocos-Nazca Rift. when the Cocos-Nazca Rift tip moves closer ern boundary, the East Pacifi c Rise, is stress free. Plate thickness is constant in our model. When to the East Pacifi c Rise (decreasing D) and The plate is cut along an axis perpendicular to the we include plate thickening away from the East the location of maximum tensile stress along East Pacifi c Rise, representing the Cocos-Nazca Pacifi c Rise, it mainly enhances the stress at the the East Pacifi c Rise moves closer to the rift. Rift. The cut stops at a distance D from the East East Pacifi c Rise without signifi cantly changing When, instead, the tip retreats from the East Pacifi c Rise. Plate velocities are imposed on the the stress maxima locations. We have also con- Pacifi c Rise (increasing D), the triple junction edges of the model to the north and south, and sidered a model in which the stress-free bound- remains the optimum geometry for continued the stress fi eld is examined after an arbitrary time ary more closely follows the geometry of the opening of the crack, resulting in a graben increment. The amplitude of stress scales pro- East Pacifi c Rise. This results in even stronger simi lar in size to Incipient Rift or Extinct Rift. portionally with time and Young’s modulus, and tensile stresses at Incipient Rift because a Figure 4F shows the predicted distance of the inversely with plate thickness. The stress pattern, change in strike of the East Pacifi c Rise there rift tip to the East Pacifi c Rise for the past however, remains unchanged. We present this (Fig. 1) (Lonsdale et al., 1992) helps focus and 5 m.y. based on the estimated north-south dis- pattern as a map of stress enhancement, which is intensify the tensile stresses. tance of the trace of the northern triple junc- the stress fi eld normalized by the stress expected We note that in our numerical model the tions from the trace of the Cocos-Nazca Rift tip in the absence of the Cocos-Nazca Rift. zone ahead of the Cocos-Nazca Rift tip is one (rift scarps, Fig. 2).

GEOLOGY, May 2008 341 Figure 4. A–E: Schematic A D v. 27, p. 911–914, doi: 10.1130/0091–7613 evolution of Galapagos (1999)027<0911:ETJMBR>2.3.CO;2. triple junction region for COMSOL, 2006, User’s guide: COMSOL, http:// past 2 m.y. DDR—Dietz ER ER comsol.com/products/multiphysics/. Deep Rift, IR—Incipient IR Floyd, J.S., Tolstoy, M., Mutter, J.C., and Scholz, Rift, ER—Extinct Rift, light proto DDR C.H., 2002, Seismotectonics of mid-ocean ridge shading—Cocos-Nazca DDR propagation in Hess Deep: Science, v. 298, Rift, dark shading—zone p. 1765–1768, doi: 10.1126/science.1077950. of excess volcanism, bold 0 Ma 1.5 Ma Karson, J., Klein, E., Hurst, S., Lee, C., Rivizzigno, lines—spreading rifts. P., Curewitz, D., Morris, A., Miller, D., Varga, Near-continuous jumping B E R., Christeon, G., Cushman, B., O’Neill, J., of Incipient Rift-type rifts Brophy, J., Gillis, K., and Stewart, M., 2002, occurs north of Cocos- Structure of the uppermost fast-spread oceanic Nazca Rift. In contrast, ER crust exposed at the Hess Deep Rift: Implica- IR the southern triple junc- tions for subaxial processes at the East Pacifi c tion and Dietz Deep Rift DDR Rise: Geochemistry, Geophysics, Geosystems, have been stable and v. 3, doi: 10.1029/2001GC000155. have bounded a rotating 0.5 Ma 2.0 Ma Klein, E.M., Smith, D.K., Williams, C.M., and Galapagos microplate for Schouten, H., 2005, Counter-rotating micro- ~1.5 m.y. Note that proto plates at the Galapagos triple junction, east- F Extinct Dietz Deep Rift and C 80 Rift ern equatorial Pacifi c Ocean: Nature, v. 433, Extinct Rift occupied sym- p. 855–858, doi: 10.1038/nature03262. metrical positions about 60 ‘1.4D’ Lonsdale, P., 1977, Regional shape and tectonics of Cocos-Nazca Rift ca. 1.5 ER the equatorial East Pacifi c Rise: Marine Geo- Ma. F: Black line (1.4D)— D physical Researches, v. 3, p. 295–315. DDR 40 Incipient Rift estimated north-south dis- Lonsdale, P., 1988, Structural patterns of the tance between the trace of Distance (km) 20 Dietz Deep Galapagos microplate and evolution of the northern triple junc- ? the Galapagos triple junction: Journal of Geo- 1.0 Ma GMP Rift tions (red dashed line, 0 physical Research, v. 93, p. 13,551–13,574. Fig. 2) and rift scarps mark- 0 12345 Lonsdale, P., Blum, N., and Puchelt, H., 1992, The Time (Ma) ing the trace of Cocos- RRR triple junction at the southern end of the Nazca Rift tip (dashed line, Fig. 2). Dashed line (D)—predicted distance of Cocos-Nazca Rift tip Pacifi c–Cocos East Pacifi c Rise: Earth and to the East Pacifi c Rise. This suggests that Extinct Rift (and Incipient Rift) were longer lived Planetary Science Letters, v. 109, p. 73–85, because the Cocos-Nazca Rift tip was slowly retreating from the East Pacifi c Rise. Alterna- doi: 10.1016/0012–821X(92)90075–7. tively, other factors acting at the East Pacifi c Rise may have controlled its stability. GMP— Mastin, L.G., and Pollard, D.P., 1988, Surface defor- Galapagos microplate. mation and shallow dike intrusion processes at Inyo Craters, Long Valley, California: Journal of Geophysical Research, v. 93, p. 13,221–13,235. The numerical model predicts a mirror-image slow-spreading Southwest Indian Ridge meets Mitchell, N.C., 1991, Distributed extension at the band of cracks and graben south of the Cocos- the intermediate spreading Central and South- Indian Ocean Triple Junction: Journal of Geo- Nazca Rift before the formation of the Galapagos east Indian Ridges, Mitchell (1991) reported physical Research, v. 96, p. 8019–8043. Mitchell, N.C., and Livermore, R.A., 1998, Spiess microplate. What precipitated the development an area of “transverse” lineations that crosscut Ridge: An axial high on the slow spreading of the Galapagos microplate is still unclear. abyssal hills generated at the Central Indian Southwest Indian Ridge: Journal of Geophysi- Lonsdale (1988) suggested that Dietz Deep Ridge, which may be similar to that described cal Research, v. 103, p. 15,457–15,471, doi: Rift was triggered by a volcano forming near here. The geology of the GTJ and our mechani- 10.1029/98JB00601. Pollard, D.D., and Holzhausen, G., 1979, On the the East Pacifi c Rise. We note that Extinct Rift cal modeling of the stresses associated with the mechanical interaction between a fl uid-fi lled existed at roughly the same time and same dis- propagating Cocos-Nazca Rift have revealed a fracture and the Earth’s surface: Tectonophys- tance to the Cocos-Nazca Rift as the incipient fundamental mechanical load associated with a ics, v. 53, p. 27–57, doi: 10.1016/0040–1951 proto Dietz Deep Rift (Fig. 4E). This supports major propagating rift that controls the develop- (79)90353–6. a relationship with deformation caused by the ment of plate boundaries at this triple junction Schouten, H., Klitgord, K.D., and Gallo, D.G., 1993, Edge-driven microplate kinematics: Journal of Cocos-Nazca Rift, and in turn suggests that simi- and perhaps at others. Geophysical Research, v. 98, p. 6689–6701. lar jumping cracks may have occurred south of Searle, R.C., and Francheteau, J., 1986, Morphology the Cocos-Nazca Rift prior to ca. 1.5 Ma, when ACKNOWLEDGMENTS and tectonics of the Galapagos Triple Junc- the last active crack became Dietz Deep Rift. We thank the captain and crew of the R/V Melville tion: Marine Geophysical Researches, v. 8, p. 95–129. The Dietz Deep Rift decoupled the Galapagos (Vancouver Leg 01) for their help in collecting data at Incipient Rift. This study was funded by National Smith, W.H., and Sandwell, D.T., 1997, Global microplate from the Nazca plate and facilitated Science Foundation grants OCE-0096360 (Smith), seafl oor topography from satellite altimetry the start of edge-driven microplate rotation. The OCE-0096154 (Klein), OCE-0649103 (Montési), and and ship depth soundings: Science, v. 227, decoupling may have been assisted by the for- OCE-0221436 (Zhu). We had fruitful conversations p. 1956–1962. mation of the large seamount (Lonsdale, 1988), with M. Tivey, J. Cann, C. Williams, and R. Katz. Turcotte, D.L., and Schubert, G., 2002, Geodynamics: Cambridge, Cambridge University Press, 472 p. effectively pinning the triple junction at Dietz N. Mitchell, D. Naar, and D. Wilson provided con- structive comments on our interpretation of the Zonenshain, L.P., Kogan, L.I., Savostin, L.A., Deep Rift and facilitating magmatic spreading Galapagos triple junction region. R. Searle and an Golmstock, A.J., and Gorodnitskii, A.M., and propagation of Dietz Deep Rift to the north- anonymous reviewer helped to improve the manu- 1980, Tectonics, crustal structure and evo- east. At present, there are no data to determine script. We thank J. Gee and S. Cande for use of their lution of the Galapagos triple junction: ® Marine Geology, v. 37, p. 209–230, doi: whether there is a symmetrical occurrence of unpublished multibeam bathymetry. COMSOL is a registered trademark of COMSOL AB. 10.1016/0025–3227(80)90102–4. multiple crosscutting rifts predating the initia- tion of Dietz Deep Rift. Manuscript received 5 October 2007 REFERENCES CITED Distributed seafl oor deformation has been Revised manuscript received 21 December 2007 Bird, R.T., Tebbens, S.F., Kleinrock, M.C., and Naar, Manuscript accepted 1 January 2008 reported at other triple junctions. For example, D.F., 1999, Episodic triple-junction migration at the Rodriguez triple junction, where the by rift propagation and microplates: Geology, Printed in USA

342 GEOLOGY, May 2008