Segmentation of transform systems on the East Pacific Rise: Implications for earthquake processes at fast-slipping oceanic transform faults

Patricia M. Gregg Massachusetts Institute of Technology/WHOI Joint Program in Oceanography, Woods Hole, Massachusetts 02543, USA Jian Lin   Woods Hole Oceanographic Institution, Woods Hole, Massachusetts 02543, USA Deborah K. Smith 

ABSTRACT Per®t et al., 1996) indicate that ITSCs are Seven of the eight transform systems along the equatorial East Paci®c Rise from 12؇ N magmatically active, implying that the regions -to 15؇ S have undergone extension due to reorientation of plate motions and have been beneath them are hotter, and thus the litho segmented into two or more strike-slip fault strands offset by intratransform spreading spheric plate is thinner than the surrounding centers (ITSCs). Earthquakes recorded along these transform systems both teleseismically domains. To explore the effect of segmenta- and hydroacoustically suggest that segmentation geometry plays an important role in how tion on the transform fault thermal structure, slip is accommodated at oceanic transforms. Results of thermal calculations suggest that we use a half-space steady-state lithospheric the thickness of the brittle layer of a segmented transform fault could be signi®cantly cooling model (McKenzie, 1969; Abercrom- reduced by the thermal effect of ITSCs. Consequently, the potential rupture area, and bie and Ekstrom, 2001). The temperature thus maximum seismic moment, is decreased. Using Coulomb static stress models, we within the crust and mantle, T, is de®ned as T ϭ ␬ Ϫ1/2 illustrate that long ITSCs will prohibit static stress interaction between transform seg- Tmerf [y(2 t) ], where Tm is the mantle ments and limit the maximum possible magnitude of earthquakes on a given transform temperature at depth, assumed to be 1300 ЊC; ␬ system. Furthermore, transform earthquakes may have the potential to trigger seismicity y is the depth; is the thermal diffusivity, Ϫ6 2 Ϫ1 on normal faults ¯anking ITSCs. assumed to be 10 m s ; and t is the age of the lithosphere obtained by dividing distance Keywords: seismology, earthquake stress triggering, Siqueiros transform fault, transform faults, from the ridge axis by half the spreading rate. East Paci®c Rise, Clipperton transform fault.

INTRODUCTION servations of earthquakes recorded on East Segmented transform systems are com- Paci®c Rise transform faults indicate that seg- posed of several fault strands offset by short mentation is an important factor in¯uencing ridges or rifts referred to as intratransform rupture of large earthquakes at oceanic trans- spreading centers (ITSCs) (Menard and At- forms. While it has been shown that segmen- water, 1969; Searle, 1983; Pockalny et al., tation and fault steps play an important role in 1997), where active sea¯oor spreading and controlling the earthquake behavior of conti- crustal accretion are occurring (Fornari et al., nental strike-slip faults (e.g., Harris and Day, 1989; Hekinian et al., 1992; Per®t et al., 1993), the in¯uence of segmentation and 1996). Along the equatorial East Paci®c Rise ITSCs on earthquake processes at an oceanic between 15Њ S and 12Њ N (Fig. 1), the Siquei- transform system has not been studied in ros, Quebrada, Discovery, Gofar, Yaquina, detail. Wilkes, and Garrett transform systems have In this paper, we use teleseismically and hy- all undergone transtension due to changes in droacoustically recorded seismicity data from plate motions, and each of these transforms is the equatorial East Paci®c Rise and Coulomb segmented by at least one ITSC (Searle, 1983; static stress models to explore the effect of Fornari et al., 1989; Lonsdale, 1989; Goff et ITSCs on static stress interaction between al., 1993; Pockalny et al., 1997). The Clip- transform fault segments. We investigate perton transform system has undergone sev- whether adjacent fault segments can behave eral periods of transpression (Pockalny, 1997), independently of one another, and how the in- and is the only unsegmented transform system teraction between segments depends on their along the equatorial East Paci®c Rise. offset distance. The global de®ciency of seismic moment release on oceanic transform systems has led TRANSFORM SEGMENTATION researchers to hypothesize that a signi®cant Segmentation of the transtensional trans- portion of oceanic transform slip is accom- form systems at the equatorial East Paci®c Figure 1. Regional map of the equatorial modated aseismically (e.g., Boettcher and Jor- Rise has resulted in individual strike-slip fault East Paci®c Rise showing large transform dan, 2004). However, global seismicity studies strands with lengths of 18±89 km, with an av- and nontransform offsets. Segmentation ge- have yet to consider the prevalence of trans- erage of ϳ37 km. The ITSCs separating the ometry is included based on previous geo- form fault segmentation. Dziak et al. (1991) fault strands have lengths of 5±20 km, with logical mapping of the transform systems ϳ (e.g., Fornari et al., 1989; Lonsdale, 1989). observed that earthquake sizes generally cor- an average length of 11 km. Fresh lavas col- Inset: Regional map showing location of the relate with the lengths of individual fault seg- lected from the ITSCs within the Siqueiros full array of NOAA Paci®c Marine Environ- ments at the Blanco transform fault. Our ob- and Garrett transforms (Hekinian et al., 1992; mental Laboratory hydrophones.

᭧ 2006 Geological Society of America. For permission to copy, contact Copyright Permissions, GSA, or [email protected]. Geology; April 2006; v. 34; no. 4; p. 289±292; doi: 10.1130/G22212.1; 7 ®gures. 289 Figure 2. Comparison of estimated areas of Figure 4. Schematic models showing the ge- brittle lithosphere using a one-dimensional, ometry of two transform segments bisected steady-state lithosphere cooling model by a single ITSC of variable length, L. The (McKenzie, 1969) for the Clipperton (B) and source earthquake is located on the bottom Siqueiros (C) transforms. A: The 90 km Clip- right transform segment with its left edge perton transform system (X±X') and the 150 located at a distance, d, from the ITSC- km Siqueiros transform system (Y±Y'), transform intersection. The source earth- which is broken into ®ve major segments quake is assumed to be a strike-slip event S1, S2, S3, S4, and S5 separated by four on a vertical plane parallel to the transform ITSCs SA, SB, SC, and SD (Fornari et al., segment. A: A scenario in which the receiv- 1989). B: Calculated area of brittle litho- er fault is a strike-slip fault located on the -sphere for temperatures <600 ؇C (shaded re top left transform segment, which is as- gion) for the Clipperton transform. C: Com- sumed to have the same dip, strike, and parison of the calculated areas of brittle Figure 3. A: The predicted maximum mo- rake as the source earthquake. B: A scenar- lithosphere for the Siqueiros transform for ment magnitude, M , of earthquakes W io in which the receiver fault is a normal a model of unsegmented geometry (area (curved lines) for a given transform segment fault located along the ITSC, which is as- above the dotted line) versus a model con- area and a constant slip of 0.1, 0.3, and 1.0 sumed to have a dip of 60؇ and is parallel to sisting of ®ve individual segments offset by m. The calculations assume that the earth- the ITSC. steady-state ITSCs (shaded area). quake ruptures the entire transform seg- ment. Black dots mark the observed maxi-

mum MW recorded on the transform Figure 2 compares the calculated areas of segments of the equatorial East Paci®c lihood of rupturing multiple transform seg- ؍ brittle deformation, de®ned as regions with Rise. The rightmost data point, MW 6.6, ments during a single earthquake. According calculated temperatures Յ600 ЊC, for the ge- corresponds to the Clipperton transform. B: to Coulomb failure criteria, when an earth- The predicted maximum M assuming a ometries of the Clipperton and Siqueiros W quake occurs on a source fault, changes in constant stress drop of transform earth- ⌬␴ transform systems. The calculated area under quakes of 1, 10, and 100 bar. Coulomb failure stress ( f) on a receiver Њ ⌬␴ ϭ⌬␶ ϩ␮Јϫ⌬␴ the 600 C isotherm for the Clipperton trans- fault are expressed as f s n, 2 2 ⌬␶ ⌬␴ form fault is 326 km , compared to 710 km where s and n are changes in shear and for a model of a single unsegmented fault with to a speci®c transform segment using hydro- normal stresses, on the receiver fault, and ␮Ј the cumulative length of the Siqueiros trans- acoustically recorded earthquakes, which have is the apparent friction coef®cient adjusted for form system. When the actual segmentation smaller location errors (Ͻ6 km) (Fox et al., the pore pressure effect (King et al., 1994). geometry of the Siqueiros transform system is 2001). The maximum earthquake magnitudes We consider a simple geometry in which considered, however, the integrated area of the observed on each of the transform fault seg- two adjacent transform segments are offset by calculated brittle deformation region is de- ments at the equatorial East Paci®c Rise from an ITSC of variable length, L, for two scenar- creased by ϳ60% to 277 km2. 1996 to 2001 are plotted in Figure 3. Assum- ios assuming the receiver faults are either Seismic moment (Mo), which re¯ects the ing the complete rupture of a given individual strike-slip events along the adjacent transform energy released by an earthquake, is a func- fault segment, we can estimate the amount of segment (Fig. 4A) or normal-faulting events tion of the rupture area of the fault. Speci®- slip or the stress drop for a given earthquake. located along the ITSC (Fig. 4B). The rupture ϭ ϫ ϫ cally, Mo G D S, where G is the shear For example, the largest earthquake observed size for the source earthquake in both cases is modulus, estimated to be 27 GPa from seismic on the Clipperton transform has a MW of 6.6 varied to re¯ect typical earthquake magnitudes velocities (Canales et al., 2003), D is the av- (Fig. 3). If the entire brittle area of Clipperton observed along the segmented transforms of erage slip, and S is the estimated brittle area. (326 km2) was ruptured during this earth- the equatorial East Paci®c Rise. ϭ The resulting moment magnitude is MW quake, the estimated average slip is 1 m, or For the ®rst scenario, the calculated maxi- ϫ Ϫ ͦͦ⌬␴ ͦͦ (2/3) log(Mo) 10.73. For a model of con- the estimated average stress drop is 53 bar or mum change in static stress, f , transferred stant stress drop during pure strike-slip earth- 5.3 MPa. to the receiver fault is plotted versus L for MW ϭ ␲ ϫ⌬␴ϫ ϫ ϭ quakes, Mo ( /2) w S, where 5.0, 5.5, and 6.0 (Fig. 5). As the separation ⌬␴ is the earthquake stress drop, and w is the COULOMB STRESS CALCULATIONS distance between the two transform segments fault width, estimated from S divided by fault Evidence for potential earthquake interac- increases, the predicted maximum induced length (Stein and Wysession, 2003). tions along oceanic transform faults has been Coulomb stress change on the receiver fault ϭ ϭ Curves in Figure 3 show the predicted noted in several investigations (e.g., Toda and decreases. For example, if Mw 6.0, d 0 earthquake magnitudes for a given fault area, Stein, 2000; Bohnenstiehl et al., 2002, 2004; km, and L is increased from 5 to 15 km, the ͦͦ⌬␴ ͦͦ assuming models of constant fault slip (Fig. McGuire et al., 2002; Forsyth et al., 2003). We calculated f decreases from 1.35 bar to 3A) or constant stress drop (Fig. 3B) during utilize the methods of King et al. (1994) and 0.25 bar (Fig. 5A). The proximity of the earth- earthquakes. Earthquakes recorded teleseis- Toda et al. (1998) to calculate how static stress quake to the ITSC-transform intersection (ITI) mically as listed in the Harvard Centroid- from a moderate-sized earthquake is trans- is also very important: the closer the source Moment Tensor (CMT) catalog were relocated ferred to adjacent faults, and assess the like- earthquake is located to the ITI (i.e., smaller

290 GEOLOGY, April 2006 170 aftershocks were recorded hydroacousti- cally within 24 h and 50 km of the epicenter of the main shock. The ®rst 45 aftershocks (Fig. 7C) occurred along the fault segment S3 within 10 h of the main shock, and were spa- tially truncated by the ITSCs SB and SC. These aftershock locations correspond well with areas of predicted static stress increase along the source fault and on secondary strike- slip receiver faults along segment S3. Fur- thermore, the termination of the aftershocks east of the source fault at ITSC SC agrees with predicted estimates for segment interaction. Figure 6. Calculated maximum Coulomb The subsequent 125 aftershocks (Fig. 7D) stress changes on a secondary normal re- ceiver fault along the ITSC caused by a occurred on ITSC SB and the eastern end of strike-slip source earthquake on the adja- the adjacent fault segment S2. The static stress cent transform segment with geometry models assuming transform-parallel strike-slip shown in Figure 4B. The maximum change receiver faults do not predict this pattern of in Coulomb stress is taken from the point seismicity (Fig. 7C). Geologic interpretations on the ITSC where Coulomb stress reaches a maximum value. The results shown are for of the Siqueiros transform by Fornari et al. calculations assuming a tapered slip distri- (1989) indicate several ITSC-parallel faults bution along the source earthquake. ¯anking ITSC SB to the east, west, and north (Fig. 7B). As the latter 125 aftershock loca- tions correspond well with areas of increased strike-slip earthquakes and areas of calculated Coulomb stress for normal receiver faults, we stress changes Ͼ0.2±1.0 bar (Toda et al., hypothesize that these aftershocks might be 1998). associated with triggered seismicity on the Ն In the second scenario (Fig. 6), for MW normal faults mapped by Fornari et al. (1989). 5.0, a source earthquake with relatively small The ITSC-parallel tectonic fabric was created d is calculated to cause signi®cant Coulomb by ITSCs SA and SB, which appear to have static stress increases on ITSC-parallel sec- slow-spreading ridge morphology (Fornari et ondary normal faults. Such Coulomb stress al., 1989); this may account for the develop- changes correspond to a decrease in normal ment of seismically active normal faults. An- con®ning pressure across the ITSC axis, other possibility is that these aftershocks may which may result in triggering of normal- re¯ect a diking event near ITSC SB resulting Figure 5. Calculated maximum Coulomb faulting earthquakes or magmatic diking from main-shock-triggered decreases in con- stress changes on a secondary strike-slip events along the ITSCs. The predicted Cou- receiver fault caused by a strike-slip source ®ning pressure. Dynamic stress changes might earthquake (geometry shown in Fig. 4A) for lomb stress changes on secondary normal also trigger aftershocks but are dif®cult to faults along the ITSC are a strong function of ؍ source earthquake MW 6.0, 5.5, and 5.0. evaluate due to the lack of detailed main- Note that the vertical scale is different for the location of the source earthquake. For ex- each panel. All stress calculations were ϭ shock rupture models. ample, if Mw 6.0 and d is increased from carried out using a three-dimensional 5 to 10 km, the calculated ͦͦ⌬␴ ͦͦ decreases boundary-element model, Coulomb 2.6 f CONCLUSIONS from 2.5 bar to 0.5 bar. (Toda et al., 1998), assuming that both the Detailed analysis of earthquakes on trans- source and receiver faults extend to a depth form systems at the East Paci®c Rise suggests of 5 km. For each Coulomb calculation, we EXAMPLES OF POSSIBLE STRESS used a Young's modulus of 62.5 GPa, a Pois- INTERACTION that segmentation geometry plays an impor- son's ratio of 0.25, an apparent friction co- Hydroacoustic monitoring of the East Pa- tant role in how slip is accommodated at fast- ,(ef®cient, ␮؅, of 0.4 (e.g., King et al., 1994 ci®c Rise (Fox et al., 2001) has allowed us to slipping oceanic transforms. Results of Cou- and a tapered slip distribution. Stresses are lomb stress models suggest that the length of sampled on a horizontal plane at a depth of investigate several moderate-sized earth- 2 km. The maximum change in Coulomb quakes to determine the role of transform seg- the ITSC that offsets two transform fault stress is taken directly from the point on the mentation in earthquake processes. Here, we strands will determine whether the adjacent receiver fault where Coulomb stress reach- ϭ fault segments will interact by static stress have chosen one MW 5.7 teleseismically re- es a maximum value. Calculations were car- transfer. If the ITSC is suf®ciently long, the corded earthquake that occurred in April 2001 ؍ ؍ ried out for L 1to20kmandd 0, 2.5, adjacent segments will be decoupled and be- 5, and 10 km. ITSCÐintratransform spread- along the S3 segment of the Siqueiros trans- ing center. form fault, which appears to have triggered have independently of each other. This is par- seismicity on the S2 segment as well as on the ticularly important in studies of earthquakes at ITSC SB (Fig. 7). Coulomb stress models oceanic transforms, since a long oceanic trans- d), the greater the predicted maximum Cou- were calculated using the source earthquake form system could be composed of several de- lomb stress change on the receiver fault (Fig. focal mechanism recorded in the Harvard coupled fault segments. Moreover, we illus- ϭ ϭ 5). For example, if Mw 6.0, L 5 km, and CMT catalog, and the earthquake source lo- trate that the thermal effect of ITSCs may d is increased from 0 to 10 km, the calculated cation was taken from the hydroacoustic reduce the thickness of the brittle layer, thus ͦͦ⌬␴ ͦͦ f decreases from 2.5 bar to 0.1 bar. Pre- earthquake catalog. decreasing the potential rupture area and the vious studies have shown statistically signi®- The main shock ruptured the 24 km long maximum seismic moment of an oceanic cant correlations between regions of seismic- S3 fault segment ϳ10 km from its intersection transform fault system. Finally, we suggest ity rate changes following continental with the 8 km long ITSC SB. Approximately that transform earthquakes may have the po-

GEOLOGY, April 2006 291 Figure 7. Coulomb stress Forsyth, D.W., Yang, Y.J., Mangriotis, M.D., and Shen, models for a teleseismi- Y., 2003, Coupled seismic slip on adjacent oce- cally recorded earth- anic transform faults: Geophysical Research Let- .ters, v. 30, doi: 10.1029/2002GL016454 ؍ quake (Mw 5.7, 26 April 2001) on the Siqueiros Fox, C.G., Matsumoto, H., and Lau, T.-K.A., 2001, transform system. Earth- Monitoring Paci®c Ocean seismicity from an au- quake location is shown tonomous hydrophone array: Journal of Geophys- by white star on each ical Research, v. 106, p. 4183±4206. Goff, J.A., Fornari, D.J., Cochran, J.R., Keeley, C., and panel. A: Location map Malinverno, A., 1993, Wilkes transform system shows the segmentation and nannoplate: Geology, v. 21, p. 623±626. geometry of Siqueiros. Harris, R.A., and Day, S.M., 1993, Dynamics of fault Outlined region indicates interactionÐParallel strike-slip faults: Journal of area investigated in B, C, Geophysical ResearchÐSolid Earth, v. 98, and D. B: Bathymetric p. 4461±4472. map overlain by the geo- Hekinian, R., Bideau, D., Cannat, M., Francheteau, J., logic interpretations (thin and Hebert, R., 1992, Volcanic activity and crust white lines) of Fornari et mantle exposure in the ultrafast Garrett transform al. (1989). White circles fault near 13-degrees-28s in the Paci®c: Earth and indicate the locations of Planetary Science Letters, v. 108, p. 259±275. the 170 aftershocks. C: King, G.C.P., Stein, R.S., and Lin, J., 1994, Static stress Calculated Coulomb stat- changes and the triggering of earthquakes: Bul- ic stress changes on sec- letin of the Seismological Society of America, ondary strike-slip receiv- v. 84, p. 935±953. Lonsdale, P., 1989, Segmentation of the Paci®c-Nazca er faults with the same Њ Њ dip, strike, and rake as Spreading Center, 1 N±20 S: Journal of Geo- physical Research, v. 94, p. 12,197±12,225. the source earthquake. McGuire, J.J., Jordan, T.H., and Lin, J., 2002, Com- Source fault parameters plexities of transform boundaries in the oceans, from Harvard CMT focal in Stein, S., and Freymueller, J., eds., Plate ؍ mechanism: strike boundary zones, volume 30: Washington, D.C., .81؇, and sca- American Geophysical Union, p. 425 ؍ 263؇, dip -؋ 1024 McKenzie, D.P., 1969, Speculations on the consequenc 4.49 ؍ lar moment dyne-cm. We assume es and causes of plate motions: Geophysical Jour- ,km, nal of the Royal Astronomical Society, v. 18 5 ؍ rupture length .p. 1±32 ؍ km, and slip 5 ؍ width 0.36 m. We used a ta- Menard, H.W., and Atwater, T., 1969, Origin of fracture pered slip distribution, zone topography: Nature, v. 222, p. 1037±1040. and stresses were sam- Per®t, M.R., Fornari, D.J., Ridley, W.I., Kirk, P.D., Cas- pled on a horizontal plane ey, J., Kastens, K.A., Reynolds, J.R., Edwards, at a depth of 2 km. Fox M., Desonie, D., Shuster, R., and Paradis, S., et al. (2001) estimate a 1996, Recent volcanism in the Siqueiros trans- lower threshold for earth- form fault: Picritic basalts and implications for -The ®rst 45 aftershocks MORB magma genesis: Earth and Planetary Sci .1.8±1.0 ؍ quakes recorded within the hydrophone array of M b ence Letters, v. 141, p. 91±108. (shown as white circles) occurred within ten hours of the main shock and fall along fault Pockalny, R.A., 1997, Evidence of transpression along segment S3. The 0.15 bar and 0.05 bar contours are shown as solid black lines. We observe Ն the Clipperton transform: Implications for pro- that ~31% of the ®rst 45 aftershocks occurred in regions with stress increases 0.15 bar, cesses of plate boundary reorganization: Earth Ն and ~56% in areas with stress increases 0.05 bar. D: Calculated Coulomb static stress and Planetary Science Letters, v. 146, .changes on secondary normal faults dipping 60؇ and parallel to the ITSC SB. The later 125 p. 449±464 aftershocks (shown as white circles) correspond well with predictions of increased normal Pockalny, R.A., Fox, P.J., Fornari, D.J., Macdonald, stress, thus suggesting that they might be associated with normal receiver faults. We ob- K.C., and Per®t, M.R., 1997, Tectonic reconstruc- serve that ~63% of the latter 125 aftershocks occurred in regions with stress increases Ն0.15 tion of the Clipperton and Siqueiros fracture bar, and ~90% occurred in areas with stress increases Ն0.05 bar. zones: Evidence and consequences of plate mo- tion change for the last 3 Myr: Journal of Geo- physical ResearchÐSolid Earth, v. 102, tential to trigger seismicity on secondary nor- and Smith, D.K., 2002, Aftershock sequences in p. 3167±3181. mal faults ¯anking ITSCs. the mid-ocean ridge environment: An analysis us- Searle, R.C., 1983, Multiple, closely spaced transform ing hydroacoustic data: Tectonophysics, v. 354, faults in fast-slipping fracture zones: Geology, p. 49±70. v. 11, p. 607±610. ACKNOWLEDGMENTS Bohnenstiehl, D.R., Tolstoy, M., and Chapp, E., 2004, Stein, S., and Wysession, M., 2003, An introduction to This work was supported by an NSF (National Sci- Breaking into the plate: A 7.6 Mw seismology, earthquakes, and earth structure: ence Foundation) Graduate Research Fellowship fracture-zone earthquake adjacent to the Central Malden, Massachusetts, Blackwell Publishing, (Gregg), NSF grant OCE-0221386 (Smith and Lin), and Indian Ridge: Geophysical Research Letters, 498 p. the Woods Hole Oceanographic Institution Deep Ocean v. 31, p. B2, doi: 10.1029/2003GL018981. Toda, S., and Stein, R.S., 2000, Did stress triggering Exploration Institute (Lin). We are grateful for helpful Canales, J.P., Detrick, R., Toomey, D.R., and Wilcock, cause the large off-fault aftershocks of the 25 discussions with C. Williams, M. Behn, M. Boettcher, S.D., 2003, Segment-scale variations in the crust- March 1998 Mw ϭ 8.1 Antarctic plate earth- J. Cann, D. Fornari, D. Forsyth, J. McGuire, H. Schou- al structure of 150±300 kyr old fast spreading quake?: Geophysical Research Letters, v. 27, ten, and the WHOI geophysics group, and to D. Boh- oceanic crust (East Paci®c Rise, 8Њ15Ј N±10Њ5Ј p. 2301±2304. nenstiehl, E.A. Okal, and S. Toda for constructive and N) from wide-angle seismic refraction pro®les: Toda, S., Stein, R.S., Reasenberg, P.A., Dieterich, J.H., insightful reviews. Geophysical Journal International, v. 152, and Yoshida, A., 1998, Stress transferred by the p. 766±794. 1995 Mw ϭ 6.9 Kobe, Japan, REFERENCES CITED Dziak, R.P., Fox, C.G., and Embley, R.W., 1991, Re- shock: Effect on aftershocks and future earth- Abercrombie, R.E., and Ekstrom, G., 2001, Earthquake lationship between seismicity and geologic struc- quake probabilities: Journal of Geo- slip on oceanic transform faults: Nature, v. 410, ture of the Blanco transform zone: Marine Geo- physical ResearchÐSolid Earth, v. 103, p. 74±77. physical Researches, v. 13, p. 203±208. p. 24,543±24,565. Boettcher, M.S., and Jordan, T.H., 2004, Earthquake Fornari, D.J., Gallo, D.G., Edwards, M.H., Madsen, scaling relations for mid-ocean ridge transform J.A., Per®t, M.R., and Shor, A.N., 1989, Manuscript received 26 August 2005 faults: Journal of Geophysical ResearchÐSolid Structure and topography of the Siqueiros Revised manuscript received 21 November 2005 Earth, v. 109, p. B12, doi: 10.1029/2004JB transform-fault systemÐEvidence for the devel- Manuscript accepted 22 November 2005 003110. opment of intra-transform spreading centers: Ma- Bohnenstiehl, D.R., Tolstoy, M., Dziak, R.P., Fox, C.G., rine Geophysical Researches, v. 11, p. 263±299. Printed in USA

292 GEOLOGY, April 2006