Geochemical and age constraints on the formation of the Gorda Escarpment and Mendocino Ridge of the Mendocino in the NE Pacifi c

J.M. Kela† D.S. Stakes‡ Monterey Bay Aquarium Research Institute, 7700 Sandholdt Road, Moss Landing, California 95039, USA R.A. Duncan College of Oceanic and Atmospheric Science, Oregon State University, Corvallis, Oregon 97331, USA

ABSTRACT southernmost Gorda Ridge, or on a series of period of the breakup of the short intratransform spreading-center seg- and development of the San Andreas fault zone, The Mendocino transform fault is an ments during plate reorganization. Thus, the forming the boundary of the transform regime active, dextral strike-slip zone that separates Mendocino transform fault provides a record to the south and subduction regime to the north the from the Pacifi c plate in the of ridge migration, abandonment, and resid- along the margin. NE Pacifi c Ocean. The compositions of the ual volcanism of the southern Gorda Ridge Prior to this study, the Gorda Escarpment igneous rocks exposed along the southern spreading system from 23 to 11 Ma. section of the Mendocino transform fault margin of the Mendocino transform fault has not been systematically sampled beyond include tholeiitic and alkaline basalts. Major- Keywords: transform faults, geochemistry, dredge samples collected in 1964 by Krause et element, trace-element, and radiometric mid-ocean ridges, NE Pacifi c, 40Ar-39Ar dating. al. (1964). For this study, basement exposures data suggest that the rocks were generated along the entire Gorda Escarpment and eastern through fractionation of different parental INTRODUCTION part of Mendocino Ridge were examined and melts, derived by varying degrees of partial sampled during a series of remotely operated melting from different depths, at or near the Transform faults have an important role in the vehicle (ROV) dives to determine the lithol- intersection of the Mendocino transform dynamics of the global mid-ocean-ridge system. ogy, age, and origin of these transverse ridges. fault with the Gorda Ridge. There is evidence Transform faults are strike-slip faults that offset Using fi eld observations, geochemistry, and for extensive cooling and fractionation remi- active spreading centers, thus creating a change age data, we ascertained whether these rocks niscent of the transform-fault effect of Lang- in the melting regime along the mid-ocean are the result of (1) mid-ocean-ridge processes, muir and Bender (1984). Alkaline and high- ridge. In the vicinity of the fault, cold litho- near the transform zone; (2) tectonic slivering Al compositions also argue for melts from a sphere is introduced adjacent to the hot, upwell- of the Pacifi c plate; (3) relicts of the rift propa- deeper source than a normal mid-ocean-ridge ing mantle of the mid-ocean-ridge axial region. gation that created the Juan de Fuca–Gorda environment. The preferred geochemical Extensive cooling results in an increased degree plate; or (4) rocks derived from ephemeral analogue for the Mendocino transform fault of fractional crystallization and increases the intratransform spreading axes. We conclude is a failed rift system where mid-ocean-ridge depth of melting (Langmuir and Bender, 1984). that crustal formation at the southern end of basalt (MORB) compositions likely repre- Transform faults also provide tectonic windows the Gorda Ridge was complicated by waning sent basalts created at a waning spreading into crustal processes. Fault zones may expose magmatic activity associated with the chang- center before its abandonment. The MORB crustal sections of variable ages in uplifted trans- ing tectonics of the adjacent Mendocino trans- compositions were subsequently buried by verse ridges where younger, more buoyant crust form-fault boundary. younger enriched (E-MORB) and alkaline is adjacent to older, denser crust. It is within basalts derived from deeper melting and/or these transform zones that relicts of the com- GEOLOGICAL SETTING a more enriched source. We suggest that a plexities of plate tectonics may be preserved and period of rift failure, abandonment, and con- sampled. In the northeast Pacifi c Ocean (Fig. 1), The Mendocino transform fault is the major tinued alkaline volcanism occurred on the the Mendocino transform fault provides such a plate boundary between the Gorda plate (south- record into the tectonic and magmatic history ern Juan de Fuca plate; e.g., Stoddard, 1987; †Present address: Department of Earth Sciences, of the area with the preservation of rocks and Denlinger, 1992) and the Pacifi c plate (Fig. 1). University of California, Santa Cruz, Santa Cruz, structural features associated with changing It is an active zone of dextral strike-slip motion California 95064, USA. spreading-center regimes. This major structure separating the 6–8 Ma crust of the Gorda ‡Corresponding author present address: Division of Science and Environmental Policy, California consists of two transverse ridges (Gorda Escarp- plate from the 28–30 Ma crust of the Pacifi c State University, Monterey Bay, Monterey, Califor- ment to the east, and Mendocino Ridge to the plate (Atwater, 1970, 1989). The eastern part nia 93955, USA; e-mail: [email protected]. west; Fig. 1), and has existed during the entire of the Mendocino transform fault consists of

GSA Bulletin; January/February 2007; v. 119; no. 1/2; p. 88–100; doi: 10.1130/B25650.1; 7 fi gures; 3 tables.

88 For permission to copy, contact [email protected] © 2006 Geological Society of America The geochemistry of the Mendocino transform fault

A major plate reorganization of the Pacifi c- Farallon spreading system occurred at ca. 30 Ma (Atwater, 1989). On the basis of seafl oor mag- netic anomaly data, Wilson (1988) concluded that north of the Mendocino transform fault there was a transition from stable ridge-trans- form fault confi gurations to periods of rift propagation (Hey and Wilson, 1982), which led to abandonment of parts of the spreading ridge at 30 Ma and at 19 Ma. The Mendocino transform fault experienced a period of transten- sion between 24 and 19 Ma, slowly changing to transpression after 19 Ma. The Mendocino transform fault has likely experienced intervals with intratransform spreading centers, in the same style as the Blanco (Embley and Wilson, 1992), Siqueiros (Perfi t et al., 1996), and Gar- rett transform faults (Wendt et al., 1999). The Gorda Ridge has a complex history due to changes in spreading rate, intraplate deforma- tion of the Gorda plate, and reorientation of the ridge axis (Wilson, 1986, 1989). Given the com- plex tectonic history, either normal spreading or abandoned ridge segments within the transform system are plausible possibilities for the interval of 30–10 Ma. Multichannel seismic data (Trehu et al., 1995, 2003) and bathymetric data characterize the Mendocino transform fault as a series of east-west crustal slivers with signifi cant vertical relief between the Gorda and Pacifi c plates. The Figure 1. Regional map of the NE Pacifi c with major tectonic boundaries. The Mendocino Mendocino Ridge is thought to be an uplifted transform zone between the Escanaba Trough and the Mendocino triple junction is com- transverse ridge of oceanic basalt that formed prised of the Mendocino Ridge (just south of the Escanaba) and the Gorda Escarpment at at the Gorda Ridge (Fisk et al., 1993), with the eastern end of the transform. The partially subducted Monterey plate is preserved off- lithology and age relationships characterized shore central California with the failed spreading center capped by the Davidson Seamount by previous ROV and submersible investigation adjacent to the Morro Ridge on the fossil Morro . results (Fisk et al., 1996; Duncan et al., 1994). Previous bathymetric and petrologic studies have shown that the Mendocino Ridge is com- two shallow transverse ridges that parallel the system (Atwater, 1989). The eastern end of posed of deformed basalts and crystalline rocks transform. A change in morphology occurs at the present-day transform fault meets the San with a fl attened summit created by wave erosion ~126°W; for ~150 km to the west, the south- Andreas fault at the Mendocino triple junction during a time when the ocean crust was uplifted facing Mendocino Ridge rises to 1 km above (MTJ; Fig. 1). The triple junction is thought to to sea level or above (Krause et al., 1964; Fisk et the Pacifi c plate. This vertical offset is consis- have formed at ca. 27 Ma when the Farallon al., 1993). The Mendocino Ridge is presumably tent with the more than 20 m.y. age difference spreading ridge was subducted underneath the derived from the Gorda plate (Fisk et al., 1993; between the Gorda plate and the Pacifi c plate. North American continental margin. The anom- Duncan et al., 1994; Krause et al., 1998) and To the east of 126°W, the north-facing Gorda alously shallow NE corner of the Pacifi c plate was transferred to the Pacifi c plate by northward Escarpment is ~80 km long and consists of sedi- at the Mendocino triple junction is the faulted migration of the Mendocino transform fault. mented, faulted basement blocks. At the Gorda “Vizcaino block,” thought to represent an accre- The eastern end of the Mendocino transform Escarpment, the older Pacifi c plate is elevated tionary wedge that was formed during Farallon– fault, the Gorda Escarpment, has been seismi- up to 1.5 km above the younger Gorda plate, a North American plate collision prior to 27 Ma cally imaged as a series of sedimented crustal depth relationship opposite to what would be and subsequently transferred to the Pacifi c plate slivers. The Gorda Escarpment has been inter- expected of normal conductive cooling with after the initiation of the San Andreas transform preted to be a relatively young (younger than age. The Mendocino transform fault continues fault (e.g., Griscom and Jachens, 1989). To the 6 Ma) structural feature (Leitner et al., 1998) beyond the southern terminus of the Gorda west, at 127°30′W, the Mendocino transform produced by compression across the Mendocino Ridge to form the , fault meets the Escanaba Trough, the southern- transform fault. The transpressional environment which extends ~3000 km into the Pacifi c Basin. most segment of the Gorda Ridge (Fig. 1). The is the result of misalignment of the Mendocino The evolution of the Mendocino transform Gorda Ridge is a 300-km-long active spreading transform fault compared to current regional fault can be traced back to ca. 85 Ma as a dis- center (Clague and Holmes, 1987), which cre- plate motions that can be related to the nonpar- continuity within the Farallon-Pacifi c spreading ated the crust of the Gorda plate. allel movement of the Blanco fracture zone in

Geological Society of America Bulletin, January/February 2007 89 Kela et al. the north (between the Juan de Fuca and north- recovered samples did not include lithologies can be found at http://www.wsu.edu/~geology/ ern Gorda Ridges) compared to the Mendocino of the continental margin just to the south. geolab/note.html (Knaack et al., 1994; Johnson transform fault in the south (at the southern end Conglomerates and carbonate-welded brec- et al., 1999). of the Gorda Ridge). cias were present as vertical sheets separating Six of the XRF analyses from the Mendocino more intact basalt, representing lithologies that Ridge reported in Fisk et al. (1993) are incorpo- SAMPLING METHODS AND might be expected within an active fault zone. rated for comparative purposes in this study, but SAMPLING SITES Dive T208 contained extremely well-preserved only major-element data are available for those pillow basalts overlying massive basalts with samples. The XRF and ICP-MS analyses for the In 2000 and 2001, eleven dives with the Mon- columnar jointing. Dive T353 located gabbros Morro Ridge samples (Tables 1 and 2) were also terey Bay Aquarium Research Institute (MBARI) overlain by sheeted dikes. These observations completed at the GeoAnalytical Laboratory of ROV Tiburon were carried out to study the geo- suggest that at least locally the Gorda Escarp- Washington State University. logical evolution of the Mendocino transform ment includes relatively intact slices of oceanic Sawn pieces of seven whole-rock basaltic fault and to collect samples for petrologic, geo- crust separated by near vertical shear zones of samples, and plagioclase and groundmass con- chronologic, and geochemical analysis. All new highly deformed rocks. This picture is consis- centrates separated from an eighth sample from samples were collected using the robotic manip- tent with the published seismic transects across the Mendocino transform fault, were prepared ulator arm of the ROV. The ROV dive tracks the Gorda Escarpment (Trehu et al., 2003). at Oregon State University for 40Ar-39Ar incre- on the Mendocino transform fault were based To supplement the Mendocino transform mental heating age determinations (Table 3). on a Simrad EM300 30 kHz multibeam sonar fault data set, six basaltic samples collected Small cores (200 mg) cut from the freshest por- mapping survey performed by MBARI in 1998 from the Morro Ridge, using the ROV Tiburon tions of the sawn surfaces, and plagioclase and (Fig. 2). ROV tracks were targeted on erosional were also analyzed (T143; Tables 1, 2, 3). The groundmass wrapped in Cu-foil were loaded in gullies that were expected to provide exposures Morro Ridge is a transverse ridge on the Morro evacuated quartz vials for neutron irradiation. of the underlying basement rocks. Bright refl ec- transform zone and thus is a remnant of the par- Samples were irradiated for 6 h at 1 MW power tors observed on the backscatter data (see inset tially subducted Monterey plate located offshore at the Oregon State University TRIGA reactor. on Fig. 2; Stakes et al., 2002) were interpreted central California (Fig. 1). The Morro Ridge is The total fl uence of fast neutrons in produc- to represent exposed bedrock and were targeted ~20 km from the intersection of the Morro trans- ing 39Ar from 39K was monitored with biotite as promising dive sites. Two of the dives were form with the fossil Pacifi c spreading center at standard FCT-3 (28.03 ± 0.16 Ma; Renne et al., located on the Mendocino Ridge: one near the Davidson Seamount (Fig. 1; Davis et al., 2002). 1998). After decay of short-lived radionuclides, intersection of the Mendocino Ridge and the Thus, Morro Ridge provides samples of the the samples were loaded in a glass manifold Gorda Escarpment (Dive T203), and another Pacifi c plate at 27 Ma and a Pacifi c equivalent above a low-blank, double-vacuum resistance near the Escanaba Trough (Dive T347). These of the fossil ridge-transform environment. furnace, where they were dropped one at a time. two dives were placed on the western Men- Each sample was heated incrementally (6–9 docino Ridge near the 1994–1995 dive sites ANALYTICAL METHODS steps) from 600 °C to fusion. The isotopic com- of the Navy ATV (Advanced Tethered Vehicle) position of Ar (masses 36, 37, 38, 39, and 40) where outcrops of gabbro and diabase were A total of 72 igneous rock samples was col- at each temperature step was measured with an sampled using the ROV manipulator (ATV81; lected and examined in thin section. Of these, MAP 215/50 mass spectrometer. Further details ATV154; Duncan et al., 1994; Fisk et al., 1996). 35 basaltic samples were chosen for X-ray fl uo- of the experimental procedure are described in Historical data from dredged samples include rescence (XRF) analyses of major elements, and Duncan (2002). basalts from Fisk et al. (1993: CASC8, 9; FAN inductively coupled plasma–mass spectrometer 25, 33, 36). Many of the historical samples do (ICP-MS) analyses of rare earth elements (REEs) CLASSIFICATION OF MAJOR- AND not have complete geochemistry published, but based on hand specimen and microscopic exami- TRACE-ELEMENT GEOCHEMISTRY comparisons are made where possible. nation. Samples were cleaned in distilled water The Gorda Escarpment was sampled during in a sonic bath to remove vesicle and vein fi ll- The basaltic rocks from the Mendocino Ridge eight ROV Tiburon dives (T202, T204, T205, ings prior to any chemical analyses. The initial and the Gorda Escarpment exhibit a broad range T207, T208, T348, T349, and T353). These 35 samples showed primary mineralogy with of compositions, from tholeiitic basalts (T202, dive tracks were spaced as evenly as possible less than 5% secondary minerals observed in T203, T204, T207, T208, T347, T348, T349, along the escarpment, targeting unsedimented thin section (see petrography). Of these samples, and T353) to alkaline basalts (T208 and T353), areas as inferred from high-refl ective backscat- only 12 had total major-element concentrations based on their total alkali versus silica content ter based on the EM300 data. Sampling from above 98 wt% during the XRF analysis (Table 1). (TAS) (Fig. 3). Based on Karsten et al.’s (1990) bottom to top of the fault scarp revealed gab- The 23 remaining samples had totals between 96 and the TAS classifi cation schemes, the tholei- bros, dolerites, basalts, and sediments, which wt% and 98 wt%. All 35 were selected to char- itic basalts rocks can be further divided into nor- together comprise a complete oceanic crustal acterize the igneous processes using the more mal (N) mid-ocean-ridge basalts (MORBs) (Zr/ sequence in three out of the eight dives. Most immobile trace-element concentrations obtained Nb >25), transitional T-MORBs (Zr/Nb 16–25), of the sampling locations yielded only normal by ICP-MS (Table 2). REEs are considered to and enriched E-MORBs (Zr/Nb 9–16). Basalts tholeiite samples, while T208 and T353 yielded be relatively insensitive to alteration. The 12 with Zr/Nb of ~8 are generally considered to both tholeiites and alkali basalts (Fig. 2). The samples that yielded totals >98 wt% were used be mildly alkalic (Davis et al., 2002). However, samples collected from the Gorda Escarpment to describe the major-element characteristics of in this study, the E-MORBs and mildly alkalic did not include any potential accretionary com- the sample set. The XRF and ICP-MS facilities (based on TAS) basalts of dive T208 both have plex or continental material from the Vizcaino at the GeoAnalytical Laboratory of Washington Zr/Nb concentrations from 8 to 9. Dive T353 block (e.g., Franciscan sediments, serpentine State University were used in the study. Detailed contained the most alkalic enrichment with or metabasalt, Salinian granite). Thus, the description of methods, precision, and accuracy Zr/Nb values of <8.

90 Geological Society of America Bulletin, January/February 2007 The geochemistry of the Mendocino transform fault

A

B

C

Figure 2. (A) SeaBeam bathymetry (National Oceanographic and Atmospheric Administration [NOAA] public data) with Monterey Bay Aquarium Research Institute (MBARI) ROV dive tracks (yellow and red) and numbers from the Gorda Escarpment and the western part of the Mendocino Ridge on the Mendocino transform fault in 2000 and 2001. FAN, CASC, and ATV refer to 1994–1995 dive sites of the Navy ROV (Duncan et al., 1994; Fisk et al., 1996). (B and C) Simrad EM300 30kHz multibeam bathymetry (B) and backscatter (C) maps of the Gorda Escarpment. Areas of high acoustic refl ectivity were targeted as promising dive sites. Dive tracks are shown in yellow and red. Letters N, T, E, and A refer to the basalt types present: N (normal), T (transitional), E (enriched) mid-ocean-ridge basalt (MORB) and A (alkali basalt). WF—Western Flyer.

Geological Society of America Bulletin, January/February 2007 91 Kela et al.

TABLE 1. MAJOR-ELEMENT DATA

Sample Longitude Latitude Type SiO2 Al2O3 TiO2 FeO MnO CaO MgO K2ONa2OP2O5 Total (°W) (°N) T203-G7 126.41 40.43 N 49.53 13.99 2.07 10.67 0.252 11.12 6.89 0.62 3.07 0.23 98.45 T203-G9 126.41 40.43 N 50.44 14.22 1.99 11.4 0.207 12 6.54 0.38 2.61 0.23 100.01 T204-G10 125.39 39.42 N 48.98 13.12 2.88 13.08 0.236 8.71 6.29 0.57 4.03 0.28 98.17 T207-G6 125.5 40.39 N 51.44 15.26 1.37 9.31 0.188 9.38 8.12 0.7 3.67 0.13 99.57 T347-G16 127.61 40.42 N 50.29 14.13 2.43 11.23 0.167 10.81 6.81 0.14 2.95 0.23 99.18 T204-G5 125.38 38.48 T 50.02 18.41 1.49 7.62 0.226 8.71 7.38 0.2 3.98 0.19 98.23 T208-G3 125.22 40.36 E 51.08 16.89 2.09 9.67 0.191 10.47 4.6 0.86 3.63 0.48 99.96 T208-G5 125.22 40.36 E 51.7 16.69 2.21 9.28 0.222 10.5 4.35 0.72 3.7 0.42 99.79 T208-G19 125.21 40.36 E 50.99 16.24 2.25 9.76 0.148 11.05 4.66 0.64 3.69 0.41 99.83 T143-R10 122.67 35.22 E 50.88 13.32 2.58 14.04 0.25 9.76 5.56 0.21 3.16 0.26 100.01 T208-G4 125.22 40.36 A 50.19 16.75 2.09 11 0.175 10.09 4.35 1.01 3.5 0.43 99.59 T208-G11 125.21 40.36 A 51 16.12 2.04 9.5 0.131 9.9 3.79 1.36 3.68 0.5 98.02 T208-G18 125.21 40.36 A 50.67 17.76 2.25 7.95 0.142 10.81 2.93 1.08 3.99 0.42 98 Note: Selected major-element X-ray fl uorescence (XRF) data in wt%. Letters N, T, E, and A for rock classifi cation refer to normal (N) mid-ocean-ridge basalt (MORB), transitional (T)-MORB, enriched (E)-MORB and alkali basalt, respectively. Analyses were completed at Washington State University.

TABLE 2. TRACE-ELEMENT DATA (PPM) Sample ID Type La Ce Nd Sm Eu Yb Nb Y Zr Zr/Nb Zr/Y Nb/Y Nb/Y(N) Ce/Y(N) Ce/Yb(N) T208-G4 A 16.02 34.3 19.89 5.92 2.09 3.21 25.51 37.01 223.45 8.76 6.038 0.69 4.4 2.38 2.96 T208-G11 A 19.37 38.64 21.78 6.32 2.17 3.47 25.98 40.05 228.82 8.81 5.713 0.65 4.14 2.47 3.1 T353-G7 A 16.52 34.33 19.18 5.22 1.67 2.07 20.85 25.05 162.35 7.79 6.481 0.83 5.31 3.51 4.61 T353-G15 A 15.82 32.48 18.4 4.89 1.7 1.91 19.84 23.69 150.25 7.57 6.342 0.84 5.34 3.51 4.71 T203-G15 E 10.48 23.87 15.6 4.79 1.76 2.53 12.87 29.88 148.27 11.52 4.962 0.43 2.75 2.05 2.62 T205-G4 E 10.79 24.36 14.99 4.41 1.48 2.36 15.22 26.93 141.8 9.32 5.266 0.57 3.61 2.32 2.87 T205-G5 E 15.65 29.36 14.77 3.87 1.25 2.03 11.98 22.38 119.4 9.97 5.335 0.54 3.42 3.36 4.02 T208-G9 E 17.54 37.07 21.67 6.35 2.23 3.49 26.35 39.09 230.37 8.74 5.893 0.67 4.3 2.43 2.95 T208-G17 E 20.56 42.06 24.32 7.01 2.37 3.17 23.02 41.76 200.77 8.72 4.808 0.55 3.52 2.58 3.68 T208-G19 E 16.77 36 21.14 6.3 2.19 3.44 26.97 39.56 234.17 8.68 5.919 0.68 4.35 2.33 2.91 T353-G4 E 14.02 29.39 18.02 5.24 1.77 2.75 13.95 32.75 150.82 10.81 4.605 0.43 2.72 2.3 2.97 T353-G17 E 12.68 27.48 16.7 4.8 1.62 2.31 16.4 27.77 148.29 9.04 5.340 0.59 3.77 2.54 3.31 T208-G5 E 16.28 35.04 20.57 6.11 2.1 3.35 26 37.98 227.61 8.76 5.993 0.68 4.37 2.37 2.9 T208-G9 E 17.54 37.07 21.67 6.35 2.23 3.49 26.35 39.09 230.37 8.74 5.893 0.67 4.3 2.43 2.95 T348-G10 T 10.12 26.25 20.08 6.44 2.25 4.24 10.09 46.39 197.85 19.62 4.265 0.22 1.39 1.45 1.72 T204-G5 T 7.92 17.69 11.21 3.54 1.37 2.09 6.82 23.3 125.35 18.39 5.380 0.29 1.87 1.95 2.35 T202-G5 N 9.4 29.11 27.77 10.64 3.49 9.14 6.59 95.79 300.9 45.68 3.141 0.07 0.44 0.78 0.88 T203-G9 N 4.5 12.93 11.73 4.6 1.66 4.26 2.78 43.35 110.19 39.57 2.542 0.06 0.41 0.76 0.84 T203-G10 N 1.91 5.57 5.61 2.35 0.94 2.43 1 24.68 49.5 49.28 2.006 0.04 0.26 0.58 0.64 T203-G11 N 4.2 13.09 12.43 4.89 1.76 4.52 2.42 46.16 118.35 48.86 2.564 0.05 0.33 0.73 0.8 T204-G10 N 6.16 19.02 17.84 6.76 2.43 6.08 3.77 64.42 183.34 48.58 2.846 0.06 0.37 0.76 0.87 T207-G11 N 3.68 9.47 7.13 2.49 1.04 2.5 2.18 25.69 78.73 36.07 3.065 0.08 0.54 0.95 1.05 T347-G16 N 5.02 15.32 14.25 5.55 1.91 4.89 3.05 48.49 142.85 46.84 2.946 0.06 0.4 0.81 0.87 T143-R7A E 10.38 22.04 15.71 5.11 1.75 4.11 12.45 44.19 129.22 10.38 2.924 0.28 T143-R8A E 3.81 9.14 6.74 2.43 0.93 2.28 4.64 22.87 60.05 12.94 2.626 0.20 T143-R13 E 8.29 18.29 12.07 3.79 1.29 2.72 10.9 29.03 99.85 9.16 3.440 0.38 Note: Selected rare earth element inductively coupled plasma–mass spectrometry (ICP-MS) data in ppm. Letters N, T, E, and A for rock classifi cation refer to normal (N) mid-ocean-ridge basalt (MORB), transitional (T)-MORB, enriched (E)-MORB and alkali basalt, respectively. Analyses were completed at Washington State University.

TABLE 3. 40Ar-39Ar INCREMENTAL HEATING AGES FOR THE MENDOCINO TRANSFORM FAULT AND MORRO RIDGE Sample Material Total fusion age 2σ Plateau age 2σ N%39Ar MSWD Isochron age 2σ MSWD 40Ar/ 36Ar Comments (Ma) error (Ma) error in plateau (Ma) error initial T203-G9 Whole rock 10.70 0.28 None developed None developed Recoil Ar T205-G2 Whole rock 13.04 0.54 13.36 0.34 6/8 89 1.08 13.34 0.71 1.35 296 ± 7 T207-G6 Whole rock 21.03 0.32 None developed 19.51 0.34 2.37 319 ± 5 Excess Ar T208-G5 Whole rock 11.91 0.22 12.17 0.19 3/6 79 1.80 12.12 0.18 0.72 299 ± 5 T208-G19 Whole rock 12.53 0.29 12.75 0.35 4/7 73 1.55 12.51 0.38 0.16 298 ± 3 T347-G16 Whole rock 23.14 1.01 None developed 18.05 0.43 0.79 306 ± 2 Excess Ar T348-G10 Plagioclase 716.7 2.40 None developed None developed Excess Ar T348-G10 Groundmass 15.58 0.10 16.27 0.14 4/8 1.55 16.14 0.23 1.2 299 ± 6 T353-G4 Whole rock 23.10 0.30 None developed None developed Recoil Ar T143–18a Plagioclase 46.13 4.37 28.97 3.87 5/6 88 1.08 27.19 5.94 1.18 297 ± 5 T143–18b Plagioclase 35.81 3.46 30.03 4.05 5/6 91 1.35 26.53 5.70 1.02 298 ± 4 Note: Ages calculated using biotite monitor FCT-3 (28.04 Ma) and the total decay constant is 5.543 × 10–10 yr–1. N is the number of heating steps (defi ning plateau/total); MSWD is an F-statistic that compares the variance within step ages with the variance about the plateau age (mean square of weighted deviates).

92 Geological Society of America Bulletin, January/February 2007 The geochemistry of the Mendocino transform fault

6 clinopyroxenes. Subophitic intergrowth of pla- Davidson Seamount ALKALIC Alkalic gioclase and clinopyroxene can be observed in E-MORB the coarser-grained rocks, and quench textures 5 T-MORB dominate the fi ne-grained rocks. The gabbroic N-MORB rocks contain mainly plagioclase, clinopyrox-

O From Fisk 1993 2 ene, and minor olivine. Some gabbros contain 4 primary amphibole where olivine is absent, O+K 2 which is not atypical for seafl oor gabbros

Na (Stakes, 1991). 3 Escanaba A majority of the alkali basalts are vesicular Morro Ridge Trough and contain abundant plagioclase; however, only samples from T208 have large (up to 6 mm) THOLEIITIC sub- to euhedral plagioclase feldspar xenocrysts 2 (corroded), and phenocrysts occur mainly as 46.2 47.2 48.2 49.2 50.2 51.2 52.2 crystal aggregates in these samples. These large

SiO2 (wt%) crystals exhibit complex zoning patterns. The Figure 3. Total alkali versus silica (TAS) diagram for basaltic rocks on the Mendocino Ridge plagioclase phenocrysts are set in a matrix of and the Gorda Escarpment. These rocks show a range of compositions from tholeiitic basalts plagioclase, clinopyroxene, and minor olivine. to alkaline basalts. Comparisons for the alkaline series are from the California seamounts, The gabbroic and diabasic rocks typically e.g., Davidson Seamount (Davis et al., 2002). The tholeiitic series is analogous with the contain greenschist-grade secondary minerals southern Gorda Ridge–Escanaba Trough (Davis et al., 1998), and the Morro Ridge, 27 Ma that refl ect their early hydrothermal alteration by remnant of the Monterey plate. Abbreviations: N (normal), T (transitional), E (enriched) seawater. These minerals include varying pro- mid-ocean-ridge basalt (MORB). portions of epidote, chlorite, amphibole, zeolite, smectite, and prehnite. The most abundant sec- ondary minerals are chlorite/smectite replace- ment of mafi c phenocrysts and albitization of The rocks appear to form two chemical groups depleted in light REE with respect to heavy plagioclase. Volcanic rocks more typically have based on their major-element chemistry (Fig. 4). REE (LREE/HREE < 1; La/Sm < 1; Ce/Yb < 1) low-temperature oxidative replacement of mafi c The alkali basalts and E-MORBs are low in compositions (N-MORB) to enriched light REE phases and glass to smectite/chlorite, along with

FeO (8–11 wt%), low in TiO2 (2–2.3 wt%), compared to heavy REE (LREE/HREE > 1; La/ vesicle fi llings of clay, carbonate, and (less com-

and high in Al2O3 (16–18.5 wt%). Their MgO Sm > 1; Ce/Yb > 1) compositions (T-MORB, monly) sulfi de. Highly deformed volcanic rocks concentration varies from 3 wt% to 4.7 wt%. E-MORB, alkaline samples) (Table 2; Fig. 5). collected by diamond drilling from the ROV The N-MORBs (including data from Fisk et The Mendocino Ridge samples are N-MORBs contain extensive mineralization within the al., 1993) show a variety of TiO2 concentrations and are more depleted in LREEs than HREEs, brittle fracture network. Veins of carbonate and

(1.1–3.9 wt%), low Al2O3 (12.5–15.5 wt%), and with the exception of T203-G15, which is an sulfi de are abundant in some of the deformed high FeO (from 9 wt% to 16 wt%). Their MgO E-MORB. The samples from the Gorda Escarp- samples. More extensively altered volcanic concentration varies from 5.3 wt% to 8.2 wt%. ment show a variety of different compositions rocks contained epidote and chlorite replace- The only T-MORB in the major-element data from N-MORBs to alkaline basalts. The REE ment of phenocrysts along with secondary veins

set has 7.5 wt% MgO, high Al2O3 (19 wt%), patterns of the N-MORBs have a crosscutting of carbonate. Samples containing more than

low TiO2 (1.5 wt%), and low FeO (7.8 wt%). relationship with the T-MORBs, E-MORBs, 5% of visible secondary phases were generally

The K2O content increases systematically from and alkaline basalts. The REE patterns of the excluded from chemical analyses. Given that N-MORBs to alkali basalts with the exception T-MORBs also have crosscutting relationships chlorite/smectite and carbonate were the most of the T-MORB and the most mafi c N-MORB. with each of the other chemical groups. The E- signifi cant secondary phases, we also excluded

The Na2O content is approximately the same in MORBs and alkaline basalts share similar REE samples with high volatile contents. the alkali basalts and E-MORBs (3.4–4 wt%) patterns. Many of the N-MORBs have a nega- and lower in N-MORBs (2.6–3.3 wt%). Two tive Eu anomaly. GEOCHRONOLOGY

N-MORBs form an exception with high Na2O (4.11 wt% and 3.7 wt%). The N-MORB with PETROGRAPHY Age determinations from the 40Ar-39Ar exper- the highest Na2O (4.11 wt%) plots within the iments were calculated in three ways (sum- alkali fi eld on the TAS diagram (Fig. 3). Trace- In thin section, the Mendocino transform marized in Table 3). Full data sets and plots element data (discussed below) confi rm that this fault rocks display an unusually broad variation from experiments are provided in the EarthRef sample is indeed an N-MORB, and therefore the in texture and mineralogy from basalts to gab- Digital Archive (ERDA) at http://earthref.org/.

high Na2O is not likely to be a primary igneous bros. The main crystallizing phases present in First, age spectra (step ages versus temperature, 39 feature. The T-MORB also has high Na2O of the groundmass of the holo- and hypocrystalline represented by % Ar released) were exam-

4.1 wt%. SiO2 and CaO are moderately variable tholeiitic basalts are plagioclase, clinopyroxene, ined for evidence of concordant step ages for within each geochemical group across the range Fe-Ti oxides ± apatite and olivine. Sub- to euhe- a majority of the Ar released in each sample of MgO values. dral plagioclase and clinopyroxene are the main (called a plateau). In four samples, T205-G2 (E- The chondrite-normalized rare earth ele- phenocryst phases present, with minor amounts MORB diabase), T208-G5 (E-MORB), T208- ments (REE) refl ect great diversity in trace-ele- of olivine. Zoning is present in some of the pla- G19 (E-MORB), and T348-G10 (T-MORB) ment chemistry, with patterns that vary from gioclase phenocrysts, and less frequently in the groundmass, good plateaus were apparent and

Geological Society of America Bulletin, January/February 2007 93 Kela et al.

52 20 (wt%) 3 O 2 Al (wt%) Alkalic 2 16 E-MORB SiO T-MORB N-MORB From Fisk et al. 1993 LLD 48 12 MgO (wt%) MgO (wt%) 246810246810

4 16

LLD (wt%) FeO (wt%) FeO 2 11 TiO

1 MgO (wt%) 6 MgO (wt%) 246810246810

14 4.5 O (wt%) 2 Na CaO (wt%) 11 3.5

8 2.5 MgO (wt%) MgO (wt%) 246810246810 1.5

Figure 4. Selected major-element plots for the Mendocino transform fault; LLD (liquid line

O (wt%) of descent) modeled using Petrolog. Abbrevia- 2

K tions: N (normal), T (transitional), E (enriched) mid-ocean-ridge basalt (MORB). 0.0 MgO (wt%) 246810

defi ned an age range from 12.2 to 16.3 Ma. In crystallization. In these cases, age calculations reliability of the plateau ages. The isochron ages whole-rock samples T203-G9 (N-MORB) and were derived from the correlation of the step Ar have slightly larger fi tting uncertainties. Our T353-G4 (E-MORB), step ages decreased with compositions (40Ar/36Ar vs. 39Ar/36Ar isochrons). new ages show that Mendocino Ridge rocks increasing temperature as a result of irradiation- For sample T207-G6 the isochron age is 19.51 were erupted between 11 and 18 Ma, while induced 39Ar and 37Ar recoil from K- and Ca- ± 0.34 Ma; for sample T347-G16, the isochron Gorda Escarpment rocks formed between 12 rich sites within these fi ne-grained basalts. In age is 18.05 ± 0.43 Ma. These correlations also and 23 Ma, which is signifi cantly younger than such cases, the best estimate of crystallization allowed us to determine the initial composition the age of adjoining Pacifi c plate crust (28– age was the total fusion age, obtained by sum- of Ar in the sample at crystallization, which was 30 Ma), estimated from the identifi cation of ming all the step compositions as if the sample greater than the atmospheric value (40Ar/36Ar = marine magnetic anomalies (Atwater, 1989). had been heated to fusion in one step, compa- 295.5). The suspected nonatmospheric initial Ar rable to a conventional K-Ar age. Hence, we in samples T207-G6 and T347-G16 was con- PETROGENESIS report an age of 10.70 ± 0.28 Ma for T203-G9 fi rmed by 40Ar/36Ar intercepts of 319 ± 5 and 306 and 23.10 ± 0.30 Ma for T353-G4. ± 2. In these cases, we accept the isochron ages The rocks from the Mendocino transform Whole-rock samples T207-G6 (N-MORB) as better estimates of the crystallization ages. fault are exceptionally heterogeneous in com- and T347-G16 (N-MORB) produced con- For the four samples that produced acceptable position, including tholeiites and alkalic basalts cave-up age spectra with no clear plateaus, plateau ages, the isochron ages were concor- with Zr/Nb = 7–47 and Ce/Yb(N) = 0.58–3.53. which indicates contributions of undegassed dant with the plateau ages, and initial 40Ar/36Ar The alkaline nature of some of the basalts is (mantle-derived, “excess”) Ar at the time of compositions were atmospheric, confi rming the refl ected both in the major-element data (high

94 Geological Society of America Bulletin, January/February 2007 The geochemistry of the Mendocino transform fault

100 100 A B

10 10 Alkali Alkali E-MORB E-MORB T-MORB Siqueiros Seamount Morro Ridge Davidson Seamount Davidson Seamount Endeavour 1 1 La Ce Pr Nd Sm Eu Tb Dy Ho Er Tm Yb La Ce Pr Nd Sm Eu Tb Dy Ho Er Tm Yb

100 100 C D

10 10

T-MORB N-MORB N-MORB N-MORB N. Gorda Ridge S. Gorda Ridge Escanaba Trough Cleft RTI 1 1 La Ce Pr Nd Sm Eu Tb Dy Ho Er Tm Yb La Ce Pr Nd Sm Eu Tb Dy Ho Er Tm Yb

Figure 5. Selected rare earth element (REE) patterns for the Mendocino transform fault. (A) Selected T (transitional)- and E (enriched)-mid- ocean-ridge basalts (MORBs) and alkali basalts of the Mendocino transform fault. (B) Comparison of the Mendocino transform fault data with other volcanic provinces in the NE and E Pacifi c, including the less alkalic members of the Davidson Seamount series and the adjacent Morro Ridge, Endeavor segment of the , and data from a seamount at the intersection of the East Pacifi c Rise and the Siqueiros transform fault. (C) Selected normal (N)-MORBs and a T-MORB from the Mendocino transform fault. (D) Comparison of the Mendocino transform fault data with Gorda Ridge N-MORBs and the Juan de Fuca Ridge Cleft segment–Blanco ridge-transform intersection (RTI).

K2O + Na2O), and the trace-element data (low Mendocino Ridge trend. We suggest that this Figures 4B and 4C. The most mafi c basalt of Zr/Nb), which support the argument that this is a too is the result of extreme fractional crystalliza- the sample set could not be related to the rest primary igneous feature. The observed complex tion. For example, high fO2 values in the magma of the N-MORB samples. We were also unable zoning patterns and corrosion of the plagioclase (from the source or crustal contamination) would to relate the T-MORB to the rest of the samples phenocrysts suggest magma mixing with peri- result in an early crystallization of ilmenite and using Danyushevsky’s liquid line of descent cal- ods of disequilibrium and resorption. magnetite, driving the fractional crystallization culation model. Based on the nonoverlapping The major-element variations (Fig. 4; Table 1) trend toward lower concentrations of FeO and REE patterns in Figure 5, the T-MORB with in Al2O3, FeO, Na2O, TiO2, and K2O, are consis- TiO2, as observed in these rocks. Simple models lower REE concentrations (T204-G5) could tent with the crystallizing phases (olivine + pla- of crystal fractionation would similarly predict have potentially come from the same source gioclase + clinopyroxene) observed in thin sec- a systematic increase in total REE concentra- as the E-MORBs and alkali basalts from dive tion. The wide range of MgO values (2.99–8.15) tions, as is also seen in each of the geochemical T208. However, there is a crosscutting rela- and lack of olivine as a major mineral phase sug- groups in Figure 5. The N-MORBs also exhibit tionship between the T-MORB (T204-G5 and gest extensive crystal fractionation. The normal an increasing negative Eu anomaly due to pla- T353 data [alkaline basalts]), and therefore the MORB fractionation path is refl ected in the sys- gioclase fractionation. two cannot share the same mantle source. The tematic increase in FeO and TiO2 with decreas- Liquid lines of descent were calculated using second T-MORB sample presented on the REE ing MgO (Perfi t and Chadwick, 1998), a varia- a computational crystal fractionation model plots cuts across all of the E-MORBs and alka- tion that can be seen in the Mendocino Ridge Petrolog (Danyushevsky, 2001). The less-mafi c line basalts, and therefore cannot be related by samples (at low MgO) and most of the Gorda N-MORBs (MgO < 7) appear to be related to a simple crystal fractionation model. The major- Escarpment samples (at high MgO). However, each other by fractional crystallization of oliv- element compositions of alkaline basalts and E- at low MgO the alkalic basalts and E-MORBs ine and plagioclase at 1 × 105 Kpa (5 kbar). The MORBs could not be produced by fractionation

are depleted in FeO and TiO2 compared to the proposed liquid lines of descent are shown in of the more mafi c samples from the Mendocino

Geological Society of America Bulletin, January/February 2007 95 Kela et al. transform fault, and therefore we rely on the be attributed to fractional crystallization, as the (Cleft-Blanco system; Stakes et al., 2006) are REE patterns to interpret their relative relation- total REE compositions would increase with presented in Figure 5. The highly fractionated ships. The majority of the alkali basalt data cross increasing fractionation of plagioclase and oliv- basalts, andesites, and dacites from the south- over E-MORB data on the REE plots, with the ine. Two of the Mendocino samples, T348-G10 ernmost Cleft segment at its intersection with exception of T208. and T204-G5, plot within the depleted source, the Blanco transform fault are all related by The process of crystal fractionation clearly low melt fraction in Figure 6A, but fall into the extended crystal fractionation and cooling from cannot explain the full range of compositional enriched source region on both Figures 6B and a common source (Stakes et al., 2003, 2006; variation in these Gorda Escarpment suites. 6C. The log-log plot of Figure 7A might better Perfi t et al., 2003; Cotsonika et al., 2005). The High fi eld strength elements (HFSE), such as distinguish between small percentages of partial abundance of E-MORBs and alkali basalts, Nb, Zr and Y, are highly incompatible, relatively melting versus distinct mantle sources, although the crossing REE patterns, and high Ce/Yb insensitive to secondary alteration, and their it is clear that both processes are required to populations for the Mendocino transform fault, ratios should remain constant during fraction- explain the full range of Mendocino samples. however, suggest that magma may have been ation. Variations in the content of these elements The trends observed on the Ce/Yb (N) versus derived from melting of a deeper, more hetero- can therefore be used to assess different mantle Ce, the large variation in Ce/Yb, and the crossing geneous, source to variable extents. This is the source regions and/or extent of partial melting. rare earth patterns require additional processes, other mechanism implicit in the transform-fault The variation of log Nb/Y versus log Zr/Y was such as variable melt percentages or different effect—the proximity of the transform perturbs exploited by Fitton et al. (1997) to distinguish source regimes, to explain their origins. Further, the mantle melting regime to greater depths. N-MORB versus enriched mantle source regions the alkalic and high-Al character of the basalts beneath Iceland, with the plume-related com- suggests an increased depth of melt segregation DISCUSSION positions from the neovolcanic zone bounded compared to MORB. Such large variations in by the two parallel lines shown in Figure 6A. major- and trace-element chemistry have also A majority of the rocks collected along the The lower line (referred to as ΔNb = 0) sepa- been noted for seamounts on the fl anks of the Mendocino transform fault were probably rates the enriched mantle source (+ΔNb) from East Pacifi c Rise (Niu and Batiza, 1997), which formed in an ridge-transform intersection envi- the MORB source (–ΔNb). We have added the have been attributed to deeper heterogeneous ronment, in which normal mid-ocean-ridge average normal MORB (N-MORB), enriched source regions (Niu et al., 2002). magmatism was modifi ed by extensive cool- MORB (E-MORB), and oceanic-island basalt Langmuir and Bender (1984) suggested that ing and fractionation. However, other processes (OIB) from an enriched mantle source, from the generation of magmas at an oceanic spread- were probably involved in order to explain the Sun and McDonough (1989), to this diagram ing center is profoundly impacted by proximity full range of compositions. The abundance for comparison; the MORB composition falls to a large-offset transform zone: Not only does and extent of enrichment along the Mendocino below, and both the E-MORB and OIB fall the “cold edge effect” result in smaller magma transform fault appear to be anomalous for mid- above, the ΔNb = 0 line. The Mendocino com- bodies with more extensive cooling and crystal- ocean-ridge or even ridge-transform intersec- positions similarly fall along two parallel arrays lization within the axial magmatic system, but tion environments (Perfi t and Chadwick, 1998). on the log Nb/Y versus log Zr/Y plot (Fig. 6A). the presence of a transform zone also cools the Cousens (1996) reported a variety of compo- The offset between the two arrays suggests two subcrustal mantle, expressed by lower extents sitions from the Juan de Fuca Ridge. These mantle source regions, while the variation within of melting and perhaps deeper melting near ranged from highly depleted basalts from the the array refl ects variation in melt fractions, the ridge-transform intersection. The “trans- Heck and Heckle Seamounts adjacent to the with smaller melt fractions occurring at higher form-fault effect” would be refl ected in mixed northern part of the ridge, to alkali basalts in values of Nb and Zr. One array of samples (E- magmas near the ridge-transform intersection the Pratt-Welker Seamount chain in the Gulf of MORBs and alkaline basalts) falls on or slightly that display a greater extent of fractionation and Alaska. A variety of MORB compositions from Δ above the Nb = 0 line, suggesting a more cooling (e.g., high FeO, TiO2, Zr) combined N- to E-MORBs has also been reported from the enriched source region. Most of these samples with elevated incompatible element ratios (e.g., Endeavor segment of the Juan de Fuca Ridge fall in the region between the average Sun and alkali content, La/Sm, and Ce/Yb). Basalts that (Gill et al., 2005). Cousens (1996) concluded

McDonough (1989) E-MORB and OIB, sug- are high in FeO and TiO2 but have Ce/Yb < 1 that the enrichment present in the NE Pacifi c gesting that there must be an enriched mantle and Zr/Nb > 25 include N-MORBs from the oceanic rocks may be due to the presence of component. The second array of samples falls Mendocino Ridge (Fig. 5) and from the north- hydrated, subducted oceanic crust, stored in the below the ΔNb = 0 line, suggesting a depleted ern Gorda Ridge axial valley (Fig. 5) (Davis and mantle. The LREE-enriched Endeavor samples source typical for N-MORB. Over half of the Clague, 1987; Keaten et al., 2001). Basalts that compare well with some of the enriched Men- samples from the Gorda Escarpment plot in the we refer to as T-MORB (Zr/Nb 16–25) are close docino transform fault samples (Fig. 5) but do enriched source fi eld, and all of the rocks from in composition to basalts from the magmatically not reach LREE concentrations as high as in the the Mendocino Ridge fall in the depleted source waning Escanaba Trough at the ridge-transform Mendocino transform fault. fi eld, with the exception of T203-15, which is intersection (Davis et al., 1998) (Figs. 3 and 5). Other possibilities for generating the alka- more enriched in Nb and Zr. Rocks of intermediate composition, however, lic magmatic compositions observed along the Similar separation into enriched and depleted are less common in our suite than the strongly Mendocino transform fault include the greater mantle source groups is depicted on the Nb/ enriched E-MORBs and alkalic types. depths of melting resulting in alkaline composi- Y(N) versus Ce/Y(N) plot (Fig. 6B), and also on Clearly the transform-fault effect has played tions associated with seamount volcanism, pos- the Ce/Yb(N) versus Ce plot (Fig. 6C). In addi- a major role in the formation of the basalts sibly near the intersection of the Gorda Ridge tion to the depleted and enriched source regions, high in FeO, TiO2, and LREE, and possibly and the Mendocino transform fault in the past. the Ce/Yb(N) versus Ce plot also shows sev- even the alkali, enrichment. Comparative data This confi guration would be similar to the sea- eral subtle positive trends at different Ce/Y(N) of an andesite pillow basalt from the southern- mount observed now at the intersection of the values as Ce increases to the right. These can most Juan de Fuca ridge-transform intersection East Pacifi c Rise and the Siqueiros transform

96 Geological Society of America Bulletin, January/February 2007 The geochemistry of the Mendocino transform fault

A10 Alkali Increasing degree E-MORB Nb/Y of partial melting Figure 6. (A) Log (Nb/Y) versus log (Zr/Y) T-MORB Nb+ Enriched source 1 plot distinguishes variations in mantle source N-MORB characteristics using highly incompatible elements that are relatively unaffected by N-MORB alteration. Parallel lines mark the limits of Sun and McDonough 1989 Icelandic plume source lavas compared to 0.1 E-MORB depleted mid-ocean-ridge basalt (MORB), Sun and McDonough 1989 each of which was empirically found to form OIB tight linear arrays on this diagram defi ned Sun and McDonough 1989 by ΔNb = 1.74 + log (Nb/Y) – 1.92 log(Zr/Y) N Nb- Depleted source (Fitton et al., 1997). The lowermost line (ΔNb 0.01 = 0) separates depleted mantle sources from enriched mantle sources. Basalts derived 110Zr/Y from each distinct source fall on a separate co-parallel array, with variation contained B within each array determined solely by vari- 6 able degrees of mantle melting and source depletion through melt extraction, as these Toward OIB variables are insensitive to crystal fraction- (alkali) ation. Average MORB, enriched (E)-MORB, Nb/Y(N) 4 compositions and oceanic-island basalt (OIB) composi- tions from Sun and McDonough (1989) are shown for comparison. The Mendocino data form two linear arrays with normal (N)- MORB source (ΔNb < 0) at moderate to high 2 melt fractions and slightly enriched (ΔNb = 0) at moderate to low melt fractions. (B) Nb/ Enriched Y(N) versus Ce/Y(N) plot similarly uses Depleted the variation in highly incompatible trace elements to distinguish different mantle 0 source regions. Nb/Y(N) > 1 is characteristic 01234Ce/Y(N) of a more fertile (enriched) mantle source, and Nb/Y(N) < 1 is typical for a less fertile C 5 (depleted) mantle source. Ce/Y(N) (light Toward OIB FC rare earth element [LREE]/heavy rare earth element [HREE]) is insensitive to crystal (alkali) 4 fractionation, and thus covariation in these compositions two parameters can be attributed to differ- Ce/Yb(N) FC ent melt percentages in the source region or different mantle sources. (C) Plot of Ce/Yb 3 Decreasing % FC N versus Ce distinguishes basalts that are of melting derived from a depleted (low Ce/Yb) mantle 2 source as linear arrays of variable Ce. The positive trends at distinct Ce/Y(N) values Enriched source with increasing Ce can be related to crystal 1 fractionation (FC). FC Depleted source 0 0153045Ce (ppm)

Geological Society of America Bulletin, January/February 2007 97 Kela et al. fault (Batiza and Johnson, 1980; Natland and rift is now capped by younger alkaline volcanics A Melson, 1980; Niu and Batiza, 1997). The alka- associated with Davidson Seamount. Batiza and 24-19 Ma

line enrichment observed in our study compares Vanko (1985) suggested that the process of rift dge well with the most evolved samples from the failure for the Mathematician spreading center Siqueiros Seamount (Fig. 5). Melankholina et resulted in alkalic magmas for up to 10–15 m.y. al. (1994) reported a compositional range from after the spreading center was abandoned and

N-MORBs to E-MORBs in rocks collected the waning mantle melting regime retreated to Ri JdF along the Mendocino fracture zone far west of greater depths. Perhaps a similar period of rift the Mendocino Ridge (165°W and 145°W). failure occurred either on the southernmost These compositions appear to fall within or Gorda Ridge, or on a series of short intratrans- close to the same compositional range as in the form spreading segments during plate reorgani- rocks collected on the Gorda Escarpment and the zation. The abandoned segments may have gone Mendocino Ridge. Interestingly, the E-MORBs through an extended period of low-volume alka- reported by Melankholina et al. (1994) were all lic volcanism, similar to that described for the B collected on a small seamount adjacent to the Mathematician Ridge and Davidson Seamount, e fault. However, no alkaline compositions were before it was slivered along the transform by 19-12 Ma reported at this location, and there is no direct tectonic processes and ultimately transferred to evidence to support the presence of a seamount the Pacifi c plate by the northward movement of at the intersection of the Gorda Ridge and the the Mendocino transform fault (Fig. 7). Mendocino transform fault. The Mendocino transform fault was in exten- Periods of transtension during regional plate sion from 24 to 19 Ma, during which time mag- Ridg JdF reorganization might produce intratransform matism on the Gorda Ridge may have retreated spreading centers. Such magmatism, described from the transform fault leaving behind a seg- at the Siqueiros (Perfi t et al., 1996), the Gar- ment of waning magmatism and extensive rett (Hekinian et al., 1992; Wendt et al., 1999), crystal fractionation. This was followed by a and the Blanco transform faults (Gaetani et al., more stable period from 19 to 10 Ma, when 1995), has generally been associated with primi- the stress regime across the transform changed tive magmas (picrites). Such picritic lavas have into transpression. During this period of com- been explained by off-axis remelting of upper pression, alkalic magmatism could have con- C mantle that had previously been depleted in tinued in decreasing abundance until the failed 12-6 Ma incompatible element–enriched heterogeneities spreading segment was slivered onto the trans- during melting beneath the ridge axis (Wendt et form zone. Thus, the Gorda Escarpment does al., 1999). The active spreading centers within not provide a window into the Vizcaino block, the Siqueiros transform are now producing but rather a record of the ridge migration, aban- magmas that are mainly N-MORB in composi- donment, and residual volcanism of the south- tion (Fornari et al. 1989). ern Gorda Ridge–Mendocino transform system Ridge JdF However, within an environment of active from 23 to 11 Ma. ridge migration, short intratransform spreading centers might tap deeper, less-depleted mantle CONCLUSIONS sources. It is tectonically feasible that samples from the Gorda Escarpment formed on short, ROV observations and systematic collection Figure 7. (A) The Mendocino transform fault intratransform spreading segments. Magnetic of seafl oor rocks have constrained the origin of was in extension from 24 to 19 Ma, during anomalies east of 127°W become diffi cult to oceanic crust exposed in the transverse ridges which time magmatism on the Gorda Ridge follow, and north of 40.2°N, which is also the of the Mendocino transform zone. A study of (Juan de Fuca Ridge [JdF]) fi rst extended southern limit of the Mendocino fracture zone, major- and trace-element chemistry supple- into the transform fault as a curved ridge seafl oor ages are not constrained by mapped mented by radiometric age determinations has (much like what is observed at the Cleft seg- magnetic anomalies. Many patterns of seafl oor led us to conclude that: ment of the Juan de Fuca at present day). age from 40.2°N to the northern limit of the 1. The slabs of ocean crust from either the (B) Toward 19 Ma, ridge magmatism may Gorda Escarpment are possible under an intra- Mendocino Ridge or the Gorda Escarpment did have moved away from the transform fault, transform spreading model. not originate from the Pacifi c plate, and cer- leaving behind a segment of ceasing magma- A compelling geochemical analogue for the tainly those from the Gorda Escarpment do not tism within the transform zone due to the Gorda Escarpment chemical variations is the represent an accretionary wedge. The range of extensional regime. The period was followed Morro Ridge–Davidson Seamount system in ages (11–23 Ma) is younger than rocks from the by a more stable period from 19 to 12 Ma, the central California borderland (Figs. 1, 3, and adjacent Pacifi c plate. when the stress regime across the transform 5). The Morro Ridge samples are basalts cre- 2. The geochemistry of Gorda Escarpment changed into transpression. (C) 12–6 Ma ated 29–30 Ma (Table 3) at a Farallon-Pacifi c rocks shows much more variability than the transpression continued and became stron- spreading center near the intersection with the Mendocino Ridge; this is especially apparent in ger. Magmatism completely ceased at the transform zone. These MORB compositions the E-MORB and alkalic basalt compositions. ridge-transform intersection (RTI), and the likely represent the basalts created at the spread- Variations in alteration-resistant trace-elements failed spreading segment was slivered onto ing center before its abandonment. The failed Nb, Y, Ce, and Zr indicate at least two mantle the transform zone.

98 Geological Society of America Bulletin, January/February 2007 The geochemistry of the Mendocino transform fault

Atwater, T.M., 1989, Plate tectonic history of the northeast Fornari, D.J., Edwards, M.H., Gallo, D.G., Madsen, G.E., sources with variable degrees of partial melting Pacifi c and western North America, in Winterer, E.L., Perfi t, M.R., and Shor, A.N., 1989, Structure and from each source. et al., ed., The eastern Pacifi c Ocean and Hawaii: Boul- topography of the Siqueiros transform fault system: 3. There is evidence of extensive cooling and der, Geological Society of America, The Geology of Evidence for the development of intra-transform North America, vol. N, p. 21–72. spreading centers: Marine Geophysical Researches, fractionation predicted by the transform-fault Batiza, R., and Johnson, J.R., 1980, Trace element and iso- v. 11, p. 263–299, doi: 10.1007/BF00282579. effect of Langmuir and Bender (1984). Alkaline topic evidence for magma mixing in alkalic and tran- Gaetani, G.A., DeLong, S.E., and Wark, D.A., 1995, Petro- sitional basalts near the East Pacifi c Rise at 8°N, in genesis of basalts from the Blanco Trough, northeast and high-Al compositions also argue for melts Rosendahl, B.R., and Hekinian, R., eds., Initial Reports Pacifi c: Inferences from off-axis melt generation: Jour- from a greater depth and smaller degrees of par- of the Deep Sea Drilling Project, Volume 54: Washing- nal of Geophysical Research, v. 100, p. 4197–4214, tial melting than are characteristic of a normal ton, D.C., U.S. Government Printing Offi ce, p. 63–69. doi: 10.1029/94JB02774. Batiza, R., and Vanko, D., 1985, Petrologic evolution of large Gill, J.B., Kela, J.M., Ramos, F., Michael, P., Woodcock, J., mid-ocean-ridge environment. failed rifts in the eastern Pacifi c: Petrology of volcanic and Stakes, D.S., 2005, The geology and geochemistry 4. The formation of an alkaline seamount at a and plutonic rocks from the Mathematician Ridge area of the Endeavour segment of the Juan de Fuca Ridge ridge-transform intersection that was then cap- and the Guadalupe Trough: Journal of Petrology, v. 26, [abs.]: Eos (Transactions, American Geophysical p. 564–602. Union), v. 75, p. 475. tured within the transform zone would explain Clague, D.A., and Holmes, M.L., 1987, Geology, petrology, Griscom, A., and Jachens, R.C., 1989, Tectonic history of some of the geochemical variability. and mineral potential of the Gorda Ridge, in Scholl, the north portion of the San Andreas fault system, Cali- D.W., Grantz, A., and Vedder, J.G., eds., Geology and fornia, inferred from gravity and magnetic anomalies: 5. The transform-fault effect is a realistic Resource Potential of the Continental Margin of West- Journal of Geophysical Research, v. 94, p. 3089–3099. explanation for the extensive fractionation ern North America and Adjacent Ocean Basins—Beau- Hekinian, R., Bideau, D., Cannat, M., Francheteau, J., and needed to produce compositions high in FeO, fort Sea to Baja California: Houston, Circum-Pacifi c Hebert, R., 1992, Volcanic activity and crust-mantle Council for Energy and Mineral Resources, Earth Sci- exposure in the ultrafast Garrett transform fault near TiO2, and REE. However, the alkaline composi- ence Series, p. 563–580. 13 degrees 28′S in the Pacifi c: Earth and Planetary Sci- tions with high Ce/Yb suggest derivation from a Cotsonika, L.A., Perfi t, M.R., Stakes, D.S., and Ridley, ence Letters, v. 108, p. 259–275, doi: 10.1016/0012- failed spreading segment that was slivered onto W.I., 2005, The occurrence and origin of andesites and 821X(92)90027-S. dacites from the southern Juan de Fuca Ridge [abs.]: Hey, R.N., and Wilson, D.S., 1982, Propagating rift explana- the transform zone between 23 and 11 Ma. The Eos (Transactions, American Geophysical Union), tion for the tectonic evolution of the northeast Pacifi c; the Mendocino transform fault was in extension Spring Meeting, no. 86(18), abstract V13A-04. psuedomovie: Earth and Planetary Science Letters, v. 58, Cousens, B.L., 1996, Depleted and enriched upper mantle p. 167–188, doi: 10.1016/0012-821X(82)90192-3. from 24 to 19 Ma, when magmatism may have sources for basaltic rocks from diverse tectonic envi- Johnson, D.M., Hooper, P.R., and Conrey, R.M., 1999, XRF waned adjacent to or within the nodal basin of ronments in the northeast Pacifi c Ocean: The genera- Analysis of rocks and minerals for major and trace the transform zone. From 19 to 11 Ma, when the tion of oceanic alkaline vs. tholeiitic basalts, in Basu, elements on a single low dilution Li-tetraborate fused A., and Hart, S., eds., Earth processes: Reading the iso- bead, Advances in X-ray Analysis: GeoAnalytical Lab, stress regime across the transform changed into topic code: American Geophysical Union Geophysical Washington State University, v. 41, p. 843–867, http:// transpression, alkalic compositions continued Monograph 95, p. 207–231. www.wsu.edu/~geology/geolab/note/xrf.html. to erupt after the spreading was abandoned and Danyushevsky, L., 2001, The effect of small amounts of Karsten, J.L., Delaney, J.R., Rhodes, J.M., and Liias, R., H2O on crystallization of mid-ocean ridge and backarc 1990, Spatial and temporal evolution of magmatic until the failed spreading segment was slivered basin magmas: Journal of Volcanology and Geother- systems beneath the Endeavour segment, Juan de Fuca onto the transform zone. mal Research, v. 110, p. 265–280, doi: 10.1016/S0377- Ridge: Tectonic and petrologic constraints: Journal of 0273(01)00213-X. Geophysical Research, v. 95, p. 19,235–19,256. Thus, the crustal slices exposed on the trans- Davis, A.S., and Clague, D.A., 1987, Geochemistry, mineralogy Keaten, R., Davis, A.S., and Clague, D.A., 2001, An along verse ridges of the Mendocino transform fault and petrogenesis of basalt from the Gorda Ridge: Journal axis study of basaltic glass from the northern Gorda provide a record of the history of the south- of Geophysical Research, v. 92, p. 10,467–10,483. Ridge [abs.]: Eos (Transactions, American Geophysi- Davis, A.S., Clague, D.A., and White, W.M., 1998, Geo- cal Union), v. 82, p. 47. ernmost Gorda Ridge during a period of ridge chemistry of basalt from Escanaba Trough: Evidence Knaack, C., Cornelius, S.B., and Hooper, P.R., 1994, Trace migration, abandonment, and residual volca- for sediment contamination: Journal of Petrology, element analyses of rocks and minerals by ICP-MS: nism from 23 to 11 Ma. v. 39, p. 841–858, doi: 10.1093/petrology/39.5.841. GeoAnalytical Lab, Washington State University, http:// Davis, A.S., Clague, D.A., Bohrson, W.A., Dalrymple, G.B., www.wsu.edu/~geology/geolab/note/icpms.html. and Greene, G.H., 2002, Seamounts at the continen- Krause, D.C., Menard, H.W., and Smith, S.M., 1964, Topog- ACKNOWLEDGMENTS tal margin of California: A different kind of oceanic raphy and lithology of the Mendocino Ridge: Journal intraplate volcanism: Geological Society of America of Marine Research, v. 22, p. 236–249. We are grateful for the skill and patience of the Bulletin, v. 114, p. 316–333, doi: 10.1130/0016- Krause, D.C., Duncan, R.A., and Fisk, M.R., 1998, Origin ROV Tiburon pilots and the crew of the R/V Western 7606(2002)114<0316:SATCMO>2.0.CO;2. of the Mendocino Ridge, NE Pacifi c, through obduc- Flyer for the two highly successful fi eld programs. Denlinger, R.P., 1992, A model for large-scale plastic yield tion of the Juan de Fuca Plate [abs.]: Eos (Transactions, Mike Perfi t provided extensive suggestions and con- of the Gorda deformation zone: Journal of Geophysical American Geophysical Union), v. 79, p. 859. Research, v. 97, p. 15,415–15,423. Langmuir, C.H., and Bender, J.F., 1984, The geochemistry stant encouragement for this paper. Alicé Davis pro- Duncan, R.A., 2002, A time frame for construction of the Ker- of oceanic basalts in the vicinity of transform faults: vided suggestions for the interpretation of the geo- guelen Plateau and Broken Ridge: Journal of Petrology, Observations and implications: Earth and Planetary chemical data. John Chadwick, Doug Wilson, and v. 43, p. 1109–1119, doi: 10.1093/petrology/43.7.1109. Science Letters, v. 69, p. 107–127, doi: 10.1016/0012- an anonymous reviewer greatly improved this paper. Duncan, R.A., Fisk, M.R., Carey, A.G., Jr., Lund, D., Doug- 821X(84)90077-3. The authors thank Associate Editor Rodney Metcalf las, L., Wilson, D.S., and Krause, D., 1994, Origin and Leitner, B., Trehu, A.M., and Godfrey, N.J., 1998, Crustal for the signifi cant contribution of his time and ideas emergence of the Mendocino Ridge [abs.]: Eos (Trans- structure of the Vizcaino block and Gorda Escarpment, that guided this paper through the revision process. actions, American Geophysical Union), v. 75, p. 475. offshore northern California, and implications for post- Support for Kela was provided by a Monterey Bay Embley, R.W., and Wilson, D.S., 1992, Morphology of the subduction deformation of a paleoaccretionary margin: Blanco transform fault zone, NE Pacifi c: Implications for Journal of Geophysical Research, v. 103, p. 23,795– Aquarium Research Institute (MBARI) student its tectonic evolution: Marine Geophysical Researches, 23,812, doi: 10.1029/98JB02050. internship. The fi eld program was jointly supported v. 14, p. 25–45, doi: 10.1007/BF01674064. Melankholina, E.N., Lyapunov, S.M., Baranov, B.V., by MBARI funds from the David and Lucile Packard Fisk, M.R., Duncan, R.A., Fox, C.G., and Witter, J.B., 1993, Kononov, M.V., Rudnik, G.B., Saidova, Kh.M., Tik- Foundation (to Stakes) and from the National Ocean- Emergence and petrology of the Mendocino Ridge: honov, L.V., and Shmidt, O.A., 1994, Compositional ographic and Atmospheric Administration (NOAA) Marine Geophysical Researches, v. 15, p. 283–296, variations of oceanic basalts from the Mendocino fault, Undersea Research Program (NURP) (to Duncan). doi: 10.1007/BF01982386. Pacifi c Ocean: Geotectonics (English translation), T. Ramirez and A. Gough assisted with illustrations Fisk, M.R., Duncan, R.A., Carey, A.G., Jr., Nielsen, R.L., v. 28, p. 226–237. and graphics. Chen, Y.J., Sours-Page, R., Sprtel, F., and Weber, M., Natland, J.H., and Melson, W.G., 1980, Compositions of 1996, Plate boundary effects at Mendocino Ridge: Eos basaltic glasses from the East Pacifi c Rise and Siqueiros (Transactions, American Geophysical Union), v. 77, fracture zone, near 9°N, in Initial Report of Deep Sea REFERENCES CITED Supplement, p. S270–271. Drilling Project, Volume 54: Washington, D.C., U.S. Fitton, J.G., Saunders, A.D., Norry, M.J., Hardarson, B.S., Government Printing Offi ce, p. 705–723. Atwater, T.M., 1970, Implications of plate tectonics for the and Taylor, R.N., 1997, Thermal and chemical struc- Niu, Y., and Batiza, R., 1997, Trace element evidence from Cenozoic tectonic evolution of western North Amer- ture of the Iceland plume: Earth and Planetary Sci- seamounts for recycled ocean crust in the eastern ica: Geological Society of America Bulletin, v. 81, ence Letters, v. 153, p. 197–208, doi: 10.1016/S0012- Pacifi c: Earth and Planetary Science Letters, v. 148, p. 3513–3536. 821X(97)00170-2. p. 471–483, doi: 10.1016/S0012-821X(97)00048-4.

Geological Society of America Bulletin, January/February 2007 99 Kela et al.

Niu, Y., Regelous, M., Wendt, I.J., Batiza, R., and O’Hara, Taylor, H.P., Jr., O’Neil, J.R., and Kaplan, I., eds., Sta- Godfrey, N., Hole, J., Klemperer, S., Clarke, S., Luet- M.J., 2002, Geochemistry of near-EPR seamounts: ble isotope geochemistry: A tribute to Samuel Epstein: gert, J., and Mooney, W.D., 1995, Pulling the rug out Importance of source vs. process and the origin of Geochemical Society [London] Special Publication 3, from under California: Seismic images of the Men- enriched mantle component: Earth and Planetary Sci- p. 77–90. docino triple junction region [abs.]: Eos (Transactions, ence Letters, v. 199, p. 327–345, doi: 10.1016/S0012- Stakes, D.S., Trehu, A.M., Goffredi, S.K., Naehr, T.H., and American Geophysical Union), v. 76, no. 38, p. 369, 821X(02)00591-5. Duncan, R., 2002, Mass wasting, methane venting and 380–381. Perfi t, M.R., and Chadwick, W.W., 1998, Magmatism at mid- biological communities on the Mendocino transform Trehu, A.M., Stakes, D.S., Bartlett, C., Chevalier, J., Duncan, ocean ridges: Constraints from volcanological and geo- fault: Geology, v. 30, p. 407–410, doi: 10.1130/0091- R.A., Goffredi, S., Potter, S.M., and Salamy, K.A., 2003, chemical investigations, in Buck, W.R., Delaney, P.T., 7613(2002)030<0407:MWMVAB>2.0.CO;2. Seismic and seafl oor evidence for free gas, gas hydrates Karson, J.A., and LaGabrielle, Y., eds., Faulting and Stakes, D., Perfi t, M., Wheat, C., Ramirez, T., Koski, R., and fl uid seeps on the transform margin offshore Cape magmatism at mid-ocean ridges: American Geophysi- and Hein, J., 2003, Evidence of off-axis volcanism Mendocino: Journal of Geophysical Research, v. 108, cal Union Geophysical Monograph 106, p. 59–116. and hydrothermal venting along the Cleft segment of B5, p. 2263, doi: 10.1029/2001JB001679. Perfi t M.R., Fornari, D.J., Ridley, W.I., Kirk, P.A., Casey, D.J., the southern Juan de Fuca Ridge, in Contributions of Wendt, J.I., Regelous, M., Niu, Y., Hekinian, R., and Coller- Kastens, K.A., Edwards, M., Shuster, R., and Paradis, S., the EGS/AGU/EUG Joint Assembly: Nice, France, son, K.D., 1999, Geochemistry of lavas from the Gar- 1996, Recent volcanism in the Siqueiros transform fault: 6–11 April 2003, Geophysical Research Abstracts, v. 5, rett transform fault: Insights into mantle heterogeneity Picritic basalts and implications for MORB magma gen- abs. EAE03-A-04666. beneath the eastern Pacifi c: Earth and Planetary Sci- esis: Earth and Planetary Science Letters, v. 141, p. 91– Stakes, D.S., Perfi t, M., Tivey, M.A., Caress, D.W., Ramirez, ence Letters, v. 173, p. 271–284, doi: 10.1016/S0012- 108, doi: 10.1016/0012-821X(96)00052-0. T.M., and Maher, N., 2006, The Cleft revealed—Geo- 821X(99)00236-8. Perfi t, M.R., Stakes, D.S., Tivey, M., Kulp, S., Ridley, W.I., logic, magnetic and morphologic evidence for construc- Wilson, D.S., 1986, A kinematic model for the Gorda defor- and Ramirez, T.M., 2003, Magma genesis and crustal tion of upper oceanic crust along the southern Juan de mation zone as a diffuse southern boundary of the Juan formation of the southern Juan de Fuca Ridge (JdFR): Fuca Ridge: Geochemistry, Geophysics, Geosystems, de Fuca plate: Journal of Geophysical Research, v. 91, Results of fi ne-scale sampling, in Contributions of v. 7, Q04003, doi: 10.1029/2005GC001038. p. 10,259–10,269. the EGS/AGU/EUG Joint Assembly: Nice, France, Stoddard, P.R., 1987, A kinematic model for the Gorda plate: Wilson, D.S., 1988, Tectonic history of the Juan de Fuca 6–11 April 2003, Geophysical Research Abstracts, v. 5, Journal of Geophysical Research, v. 92, p. 11,524– Ridge over the last 40 million years: Journal of Geo- abs. EAE03-A-07287. 11,532. physical Research, v. 93, p. 11,863–11,876. Renne, P.R., Swisher, C.C., Deino, A.L., Karner, D.B., Sun, S.S., and McDonough, W.F., 1989, Chemical and iso- Wilson, D.S., 1989, Deformation of the so-called Gorda Owens, T.L., and DePaolo, D.J., 1998, Intercalibra- topic systematics of oceanic basalts: Implications for plate: Journal of Geophysical Research, v. 94, tion of standards, absolute ages and uncertainties in mantle composition and processes, in Saunders, A.D., p. 3065–3075. 40Ar/39Ar dating: Chemical Geology, v. 145, p. 117– and Norry, M.J., eds., Magmatism in the ocean basins: 152, doi: 10.1016/S0009-2541(97)00159-9. Geological Society [London] Special Publication 42, MANUSCRIPT RECEIVED 28 MAY 2004 Stakes, D.S., 1991, Oxygen and hydrogen isotope compo- p. 313–345. REVISED MANUSCRIPT RECEIVED 18 JANUARY 2006 sitions of oceanic plutonic rocks: High temperature Trehu, A.M., Lendl, C., Leitner, B., Meltzer, A., Gulick, S., MANUSCRIPT ACCEPTED 19 MARCH 2006 deformation and metamorphism of oceanic layer 3, in Holl, J., Levander, A., Henstock, T., Beaudoin, B., Printed in the USA

100 Geological Society of America Bulletin, January/February 2007