Geochemistry, Geophysics, Geosystems

Petrology and Sr-Nd-Pb-He isotope geochemistry of postspreading lavas on fossil spreading axes off Baja California Sur, Mexico



Postspreading volcanism has built large seamounts and volcanic ridges along the short axes of a highly segmented part of the East Pacific Rise crest that ceased spreading at the end of the middle Miocene, offshore Baja California Sur, Mexico. Lava samples from Rosa Seamount, the largest volcano, are predominantly alkalic basalts, mugearites, and benmoreites. This lavas series was generated through fractional crystallization and is compositionally similar to the moderately alkalic lava series in many oceanic islands. Samples from volcanic ridges at three adjacent failed spreading axes include mildly alkalic, transitional, and tholeiitic basalts and differentiated trachyandesites and andesite. The subtle but distinct petrologic and isotopic differences among the four sites may be due to differences in the degree of partial melting of a common, heterogeneous source. Postspreading lavas from these four abandoned axes off Baja California Sur together with those from other fossil spreading axes and from seamount volcanoes that grew on the East Pacific Rise flanks define a compositional continuum ranging from normal mid-ocean ridge basalt (NMORB)-like to ocean island basalt (OIB)-like. We propose that the compositional spectrum of these intraplate volcanic lavas is due to different degrees of partial melting of the compositionally heterogeneous suboceanic mantle in the eastern Pacific. A large degree of partial melting of this heterogeneous mantle during vigorous mantle upwelling at an active spreading center produces NMORB melts, whereas a lesser degree of partial melting during weak mantle upwelling following cessation of spreading produces OIB-like melts. The latter melts have a low (<8 RA) 3He/4He signature indicating their formation is different from that of OIBs from major “hot spot” volcanoes in the Pacific with high 3He/4He ratios, such as Hawaii and Galapagos.

1. Introduction

The study of oceanic volcanism during the last four decades or so has primarily focused on the origin of lavas erupted along active spreading centers and linear volcanic island and seamount chains. Results of studies have led to the current popular views that the geochemically depleted, tholeiitic mid-ocean ridge basalts (MORB) are produced in response to plate separation and passive upwelling and adiabatic decompression melting of the mantle, and that the relatively enriched tholeiitic and alkalic ocean island basalts (OIB) that comprise the bulk of linear hot spot volcanoes are melts originating from either rising plumes of deep mantle materials [e.g., Morgan, 1971, 1972; Hofmann, 2003] or heterogeneities in the upper mantle sampled during fracturing of the lithosphere [e.g., Meibom and Anderson, 2004; Natland and Winterer, 2005; Foulger, 2007]. Dredge sampling of many of the thousands of large and small volcanoes not in linear chains and scattered across the ocean floor has proved that they commonly have at least a carapace of alkalic lava compositionally akin to alkalic OIB, as first reported by Engel et al. [1971]. These authors also recognized a pattern of tholeiitic central volcanoes transitioning to eruption of more alkalic lava, and while suggesting this evolution was mainly a result of fractionation of a tholeiitic melt they also considered decreasing extent of partial melting as a possible alternative. The latter explanation is more consistent with recent geochemical data, leading to a commonly accepted model in which a low degree of partial melting of a heterogeneous oceanic mantle, i.e., less than that responsible for tholeiitic MORB at and around active spreading axes, produces OIB-like alkalic magma at large and small volcanoes located on the flanks of the East Pacific Rise (EPR) [Batiza and Vanko, 1984; Zindler et al., 1984].

Batiza and Vanko [1985] likewise proposed such an origin (low degree melting of a compositionally heterogeneous mantle) for the young alkalic lavas dredged from seamounts and volcanic ridges along the crest of Mathematician Ridge, where spreading ceased during the Pliocene when the Pacific plate captured the Mathematician microplate [Mammerickx and Klitgord, 1982]. Two former spreading segments on the northern part of Mathematician Ridge continue to erupt alkali basalt and more silicic differentiates, sometimes subaerially, at Isla San Benedicto [Richards, 1966] and Isla Socorro [Siebe et al., 1995; Bohrson and Reid, 1995, 1997]. Long-continued postspreading ridge-building eruption of alkalic lava has also occurred along several other failed spreading axes in the eastern Pacific. Examples include Davidson Seamount [Clague et al., 2009; Castillo et al., 2010] and Guide Seamount [Davis et al., 2002, 2010], which were both built along axes that had stopped spreading in the Early Miocene when the Pacific plate captured the Monterey microplate [Lonsdale, 1991], and narrow volcanic ridges that grew along extinct segments of the Galapagos Rise [Batiza et al., 1982; Haase and Stroncik, 2002] following late Miocene capture of the Bauer microplate by the Nazca plate [Eakins and Lonsdale, 2003].

This paper presents the petrology and geochemistry of postspreading lavas on fossil spreading axes between 27°30′N and 25°20′N in the eastern Pacific off Baja California Sur, Mexico (Figure 1) and evaluates the source of fossil spreading center volcanism. Major and trace element chemistry and Sr-Nd-Pb-He isotope geochemistry were obtained for whole rock and a few glass samples from volcanic ridges built along four of the seven short EPR axes abandoned in the middle Miocene (Figure 1). Compared to the relatively uniform composition of MORB, postspreading lavas from these ridges span a much larger range of composition, similar to lavas erupted along other fossil spreading axes [e.g., Batiza et al., 1982; Batiza and Vanko, 1985; Bohrson and Reid, 1995; Davis et al., 2002; Castillo et al., 2010]. Like many riseflank central volcanoes [Engel et al., 1971], there is evidence that lavas erupted along abandoned spreading axes evolve through time from tholeiitic to alkalic composition.

Figure 1.

Simplified map showing the study area. On the large-scale map, the patterns of abandoned plate boundaries, fracture zones, and the extents of postspreading axial ridges are inferred from multibeam bathymetry; the pattern of crustal isochrons (updated from Lonsdale [1991]) is inferred from analysis of magnetic anomalies. The regional map also locates two other abandoned Pacific-Magdalena spreading axes (J, Juanita; M, Magdalena), and some other abandoned EPR axes with large postspreading axial ridges: G, Guide; D, Davidson; Ne, Newton (sampled by Batiza and Vanko [1985]); Ga, Galileo (sampled by Richards [1958]). Labeled fracture zones are as follow: Mu, Murray; Gu, Guadalupe; Mo, Molokai; Cl, Clarion.

2. Geologic Setting and Sample Description

During the Early Miocene, the crest of the EPR, west of Baja California Sur, hosted a straight (almost unsegmented) Pacific-Cocos spreading axis, striking north-northwest (340°) for 900 km between Clarion and Molokai transforms (Figure 1) [Lonsdale, 1991]. The westward moving North American plate was converging on this spreading center, narrowing the northern part of the Cocos plate, which was being subducted at the Baja California and Middle American Trenches. Magnetic anomalies show [Lonsdale, 1991] that just before Chron C5Bn, i.e., ∼15.5 Ma, the young lithosphere entering the Baja California Trench detached from the Cocos Plate to become the Magdalena microplate. Only two thirds of the Cocos plate north of Clarion fracture zone was included in this microplate (the Pacific-Magdalena-Cocos triple junction, and a transform linking this junction to the North American plate, was 300 km north of Clarion fracture zone); the southern third continued to be subducted as part of the Cocos plate at the Middle America Trench, then after ∼5 Myr [DeMets and Traylen, 2000] was detached as the still surviving Rivera microplate. Freed from the slab pull of old, dense Cocos lithosphere entering the Middle American Trench, the Magdalena plate stagnated, though overthrusting of North America persisted along its eastern margin. Along its western EPR margin, relative plate motion changed from rapid divergence between two subducting oceanic plates to separation of the Pacific plate from a stagnating microplate; i.e., the relative motion approached the absolute motion of the northwest moving Pacific plate. As a result, the spreading rate almost halved, and the spreading direction rotated clockwise by 45°–60°. The Pacific-Magdalena risecrest followed the usual response of the EPR crest to a change in spreading direction [Nelson, 1981; Lonsdale, 1985]: it segmented to form a chain of spreading axes, each of which rotated to stay orthogonal to the changing spreading direction while the overall strike of the chain remained ∼340°. As the spreading direction grew ever closer to this azimuth, the spreading axes became shorter, and the links between them (northwest striking shear zones forming transform and nontransform offsets) lengthened. Eventually, 60–100 km long shear zones linked the 7 short (10–60 km long) northeast striking Pacific-Magdalena spreading axes named in Figure 1. A similar process earlier produced slow, northeast striking spreading axes along the western margin of the Monterey microplate (e.g., Guide and Davidson axes, Figure 1), and in the past 2.5 Myr along the highly segmented, still active Pacific-Rivera EPR [Lonsdale, 1995].

Slow Pacific-Magdalena spreading built a segmented oceanic rise for just a few Myr after 15 Ma, accreting igneous crust assumed to be of typical tholeiitic basalt composition. However, the only direct support for this assumption comes from the single locality where igneous basement of the rise flank has been sampled: Deep Sea Drilling Project (DSDP) Site 472, which recovered normal abyssal tholeiite [Shibata et al., 1982]. Then the microplate was captured by the Pacific plate, so the intervening spreading axes became extinct, tectonically if not volcanically. The interactions of continental and oceanic plates that caused this event, and similar captures of, for example, the Monterey and Guadalupe microplates, are poorly understood and beyond the scope of this paper. The dynamics of most oceanic microplates that are bounded by two or more oceanic plates can be explained by a simple model [Schouten et al., 1993] in which the microplate is propelled by drag forces imposed on it by the surrounding major plates, and is captured by one of the major plates once it has grown to a size and shape that resists these driving forces [e.g., Eakins and Lonsdale, 2003]. The dynamics of the Magdalena microplate were more complex, because as well as having a shearing boundary with the Cocos plate and a partly (but decreasingly) accreting boundary with the Pacific plate it had a long converging boundary where the North American continent was overthrusting its northeast margin, and at least initially the pull of a subducted slab provided another driving force on microplate motion. The microplate began to shrink rather than grow as most of its Pacific boundary became a shear zone rather than a zone of plate accretion, and any slab-pull force diminished as continued overthrusting brought the North American margin closer to its short rotated spreading axes, but it is not clear why the microplate was captured by the Pacific plate before being completely overthrust by North America. Even the timing of Magdalena microplate capture is controversial [e.g., Michaud et al., 2006], mainly because seafloor-spreading magnetic anomalies near most of the abandoned risecrest have been obscured by local anomalies from volcanic landforms built after spreading ceased. From analysis of profiles across those axes with the smallest amount of postspreading additions, we estimate spreading ceased soon after Chron C5An, i.e., 12–11.5 Ma.

The rocks we describe here were obtained by dredging four of the volcanic ridges built along the abandoned risecrests off Baja California Sur, Mexico (Figure 1) during Scripps Institution of Oceanography's (SIO) Phoenix 03 cruise in 1992 and Rosa 01 cruise in 1993. Our dredge sampling (Table 1) was preceded by magnetic profiling and multibeam bathymetry to identify the axes of spreading and to locate anomalous axial volcanic topography of likely postspreading origin. Two of the extinct Pacific-Magdalena axes have been smothered by large seamounts: the (unsampled) 2700 m high Magdalena Seamount and the 3300 m high Rosa Seamount, named “Rosa Bank” by Chase et al. [1968], each covering almost the entire length of their respective axes (Figure 1). Sara and Nithya axes have narrow 2 km high fissural volcanic ridges along their entire lengths; Rosana and unsampled Teresa and Juanita (Figure 1) each have a row of low (<1 km high) volcanic cones.

Table 1. Dredge Locations for Postspreading Lavas on Fossil Spreading Axes off Baja California Sur, Mexico
LocationCruiseDredge StationLatitude (N)Longitude (W)Depth (m)

Most of the rocks recovered from these volcanic landforms are ferromanganese oxide-encrusted fragments of pillow lava and other types of lava flows, commonly with glassy rinds, and encrusted slabs of altered hyaloclastite. Most lava fragments have been affected by secondary alteration, but relatively fresh cores were found in many cases. Some of the samples from the summit of Rosa Seamount (dredges 9 and 10) have very fresh glass and lack a ferromanganese crust; we infer that they were erupted recently. The recovered lavas are lithologically variable, ranging from mafic basalt and basaltic andesite, to mugearite, benmoreite, trachyandesite, and andesite (Figure 2). Many of the samples are highly vesicular with small, round or elongate vesicles that typically show a pronounced flow lineation. The majority of the moderately mafic volcanic rocks are dense, hypocrystalline to holocrystalline and aphyric to phyric, the latter containing olivine, clinopyroxene and plagioclase phenocrysts. In general, the more mafic samples have higher phenocrysts contents. Dominant groundmass minerals are plagioclase, clinopyroxene, Fe-Ti oxides, and glass. The dredges also collected several round, unencrusted cobbles judged to be exotics, probably kelp-rafted from the nearby coast [Emery and Tschudy, 1941]; lithologies include metamorphic and granitic rocks.

Figure 2.

Silica (SiO2) versus total alkalis (Na2O + K2O) diagram for samples from volcanic ridges along Sara, Rosa, Rosana, and Nithya axes. Field for lavas from Davidson Seamount [Davis et al., 2002; Castillo et al., 2010] is shown for reference.

3. Analytical Methods

The samples dredged from postspreading volcanic ridges along Sara, Rosana, Rosa, and Nithya axes were analyzed for major and trace element chemistry and some were also analyzed for Sr-Nd-Pb-He isotopic compositions. For the chemical and Sr-Nd-Pb isotope analyses, fine powders were prepared from the fresh cores of individual samples using an alumina ceramic mill. Details of the sample preparation procedure are described by Solidum [2002].

Major element analyses were carried out on glass, fused from rock powders, by a fully automated, wavelength-dispersive ARL 840 X-ray fluorescence spectrometry (XRF) instrument using the procedure described by Janney [1996] and on solution prepared from sample powders by inductively coupled plasma optical emission spectrometry (ICP-OES) using the procedure described by Murray et al. [2000] with some modifications. An estimate of the volatile (e.g., H2O and CO2) contents in the rock samples was determined by weighing loss on ignition (LOI) following the procedure described by Solidum [2002]. Concentrations of trace elements Rb, Sr, Y, Ba, Pb, Th, U, Zr, Nb, Ta, and rare earth elements (REEs) were determined by inductively coupled plasma mass spectrometry (ICP-MS) following the methods modified from Janney and Castillo [1996], whereas concentrations of compatible elements Cr and Ni were determined by ICP-OES. All the analyses were carried out at the SIO Analytical Facility.

Some of the whole rock samples were analyzed for Sr, Nd and Pb isotopes at SIO following the methods described by Janney and Castillo [1996, 1997] and Solidum [2002]. For Nd and Pb isotopic analysis, sample powders were digested with a double-distilled, 2:1 mixture of concentrated HF: HNO3 acid, then Pb was first separated by redissolving the dried samples in 2N HBr and then passing the solutions through a small ion exchange column in an HBr medium. Residues from Pb extraction were collected and then passed through ion exchange columns using HCl as the eluent to collect REE. Finally, Nd was separated from the rest of the REE in an ion exchange column using alpha-hydroxyisobutyric (α-HIBA) acid as the eluent. To minimize alteration effects on bulk isotope analyses, leached sample powders were used for Sr isotopic analysis. The leaching procedure used was similar to that described by Janney and Castillo [1996, 1997]. The leached samples were also dissolved in a mixed HF: HNO3 (2:1) acid as before and Sr was subsequently separated from the solution in an ion exchange resin using HCl as the eluent. The Sr, Nd and Pb isotopes were measured using a nine-collector, Micromass Sector 54 thermal ionization mass spectrometer (TIMS) at SIO. Strontium isotopes are reported relative to NBS 987 87Sr/86Sr = 0.71025, and Nd isotopes are reported relative to 143Nd/144Nd = 0.51185 for the La Jolla Nd standard.

Helium (He) isotopes and concentrations were measured from the few available glass chips and one clinopyroxene separate. Helium gas trapped in vesicles and fluid inclusions in the glasses and mineral grain, respectively, was extracted using online vacuum crushing, and 3He/4He ratios and He concentrations were measured using a MAP 215E noble gas mass spectrometer at the SIO Fluids and Volatiles Laboratory. More details of the He isotope analytical procedure are given by Shaw et al. [2006]. Standard aliquots of air (1 RA where RA = 1.4 × 10−6) and Yellowstone Park He (16.45 RA) were used to determine the abundance and isotopic composition of the samples. Typical crusher blanks were ∼6 × 10−11cm3STP 4He. Details of the analytical accuracy and precision of the major and trace elements and isotopic measurements are described in the footnotes of Tables 2, 3 and 4, as appropriate.

Table 2. Major and Trace Element Compositions of Postspreading Lavas on Fossil Spreading Axes off Baja California Sur, Mexico
 SampleInternational Rock Standard
  • a

    Major elements oxides were measured by XRF.

  • b

    AB, Alkalic Basalt; TrB, Transitional Basalt; ThB, Tholeiitic Basalt; An, Andesite; TrAn, Trachy-Andesite; Mug, Mugearite; Ben, Benmoreite; Bas, Basanite; P-T, Phonolitic Tephrite.

  • c

    Major elements oxides in wt %. Fe2O3T is total Fe expressed as Fe3+. Each sample was analyzed in duplicates and reproducibility was better than 5% based on repeated analyses of known rock standards analyzed as unknowns. The accuracy of the standard is within 5% of the suggested values (AGV-1, n = 33; BHVO-1, n = 16) [Govindaraju, 1994], except for TiO2, MnO, CaO, and Na2O, which are ≤10%. Several samples were analyzed through both XRF and ICP-OES to evaluate the major element analytical methods, and the discrepancy is within 4% (CaO < 1%; TiO2, <2%; SiO2, Al2O3 and MgO < 3%; Fe2O3T, MnO, Na2O, K2O < 4%).

  • d

    Trace elements in ppm. Each sample was analyzed in duplicates and reproducibility was better than 5% based on repeated analyses of known rock standards analyzed as unknowns. The accuracy of the standard is within 10% of the suggested values (AGV-1, n = 42; BHVO-1, n = 33) [Govindaraju, 1994], but generally better than ±5% for rare earth elements (REEs).

P2O5        0.66 0.59        0.59      0.64      0.670.720.730.540.810.560.610.560.530.63 0.660.611.501.481.380.49 0.273 
Crd13310878346781961271398121012111511 119151133136394689121113113125361026585132134 114106105.9 20722722741916225610.1 289293
Ni181.1111.037.7181.47.625.062.519.5141.068.11618.514.825.120.522.616.442.443.8 45.619.4159.6119.4151.659.748.091.1126.1114.0120.6137.152.7132.090.0115.0229.0212.4 177.1177.5154.7 107.6101.545.4374.1330.4133.016 121121
Rb11.9 58.555.464.34.462.669.122.519.555.640.246.538.660.045.427.318.053.457.560.469.816.422.76.691.085.314.421.726.919.
Y42.111.175.490.757.2 62.295.749. 30.530.332.133.923.825.1 22.631.317.917.926.638.827.533.541.029.114.329.42021.727.627.0
Zr178157446444520 498542219212395424428428385409413336347 343427281249294474476252246248263272505220224212228218221217199198224212159171400199294227230179175
Nb49.7 132.5134.6146.1 146.0161.145.660.581.090.890.292.181.786.488.361.457.6 59.996.181.474.285.6116.4100.379.178.172.576.579.7124.
La32.86.777.176.976.7 92.790.529.538.154.052.852.554.851.653.847.328.457.751.259.456.740.041.427.
Ce43.117.1151.9148.7153.5 191.7181.952.774.3105.3106.0104.0110.0102.2106.594.960.9109.698.0109.0112.384.785.758.3144.4139.087.882.191.784.688.2144.876.198.367.723.192.276.465.866.975.352.756.056.959.968.531.755.76769.63939.7
Nd27.410.273.470.764.2 78.784.429.635.044.945.943.647.043.945.641.530.150.440.551.247.938.940.627.753.849.440.938.242.338.840.653.438.550.734.8 38.640.333.433.831.925.426.225.330.431.113.926.53332.825.225.5
Sm5.72.314.013.511.8 13.916.
Dy6.72.914.014.511.5 13.815.
Lu0.6 0.30.4
Ta3.23 8.789.109.60 8.429.512.703.70 5.695.645.924.595.355.683.883.71 3.766.505.374.745.627.86 4.874.894.765. 3.792.262.484.803.692.600.90.831.231.17
Pb5.3 2.63.3
Th2.91 5.966.9211.110.2312.168.332.744.068.197.307.806.607.987.734.60 9.416.358.809.184.894.62 11.278.685.405.035.835.075.4210.265.005.885.93 4.994.392.662.573.394.312.503.794.024.221.503.296.56.481.081.13
U1.250.143.413.54 0.142.332.801.331.152.362.472.352.372.222.292.56 2.671.832.882.42 1.762.012.742.542.711.982.011.861.982.711.301.681.68 1.361.721.10 0.971.251.361.
Table 3. Strontium, Nd and Pb Isotopic Compositions of Representative Samples on Fossil Spreading Axes off Baja California Sur, Mexicoa
LocationSample Name87Sr/86Sr2σ143Nd/144Nd2σ206Pb/204Pb2σ207Pb/204Pb2σ208Pb/204Pb2σ
  • a

    Strontium isotopic ratios were measured through dynamic multicollection and fractionation corrected to 86Sr/88Sr = 0.1194. Repeated measurements of NBS 987 yielded 87Sr/86Sr = 0.710256 ± 0.000017 (n = 18); 2σ indicates in-run precisions. Neodymium isotopic ratios were measured in oxide form through dynamic multicollection and fractionation corrected to 146NdO/144NdO = 0.72225 (146Nd/144Nd = 0.7219). Repeated measurements of the La Jolla Nd standard yielded 143Nd/144Nd = 0.511856 ± 0.000010 (n = 17). Lead isotopic ratios were measured through static multicollection and were fractionation corrected using the repeated measurements of NBS SRM 981 (n = 22; 206Pb/204Pb = 16.899 ± 0.007, 207Pb/204Pb = 15.445 ± 0.010, and 208Pb/204Pb = 36.550 ± 0.026), relative to those of Todt et al. [1996]. Total procedural blanks are negligible: <10 picograms (pg) for Nd, <35 pg for Sr, and <74 pg for Pb.

Table 4. Helium Isotope Composition of Representative Samples on Fossil Spreading Axes off Baja California Sur, Mexico
LocationSample NameWeight (g)3He/4He (R/RA)aX3He/4He (RC/RA)a[He]C (10−9 cm3 STP g−1)
  • a

    Measured 3He/4He (R) ratios are corrected for the addition of atmospheric helium to air-corrected values (RC) and are reported normalized to RA, which is the air 3He/4He value (1.4 × 10−6); X (column 5) is the measured air-normalized He/Ne ratio. Measured helium concentrations are also corrected for the effects of atmospheric He to air-corrected values [He]C, (column 7). The correction details are reported by Hilton [1996].

  • b

    Repeat analysis.

  • c

    Clinopyoxene separate. All other samples are glasses.

RosaROSA D9-11.12781.151.063.65 ± 0.950.26
 ROSA D9-1rep.b0.83107.0443.067.18 ± 0.211.5
 ROSA D9-81.42831.543.441.76 ± 0.161.6
 ROSA D9-90.75495.5643.425.67 ± 0.183.7
 ROSA D9-101.45481.391.282.78 ± 0.300.56
 ROSA D9-110.65692.322.992.98 ± 0.201.6
NithyaPX 127D-20c0.96581.2311.941.25 ± 0.0423.4

4. Results

Major element, trace element and Sr, Nd, Pb, and He isotope data for samples from Sara, Rosana, Rosa, and Nithya are presented in Tables 2, 3, and 4. The majority of volcanic rocks from Rosa Seamount, the largest of these four features, are alkalic basalts, mugearites and benmoreites. These samples define a broad, but coherent trend that plots slightly above the boundary between alkalic and tholeiitic fields in the silica versus total alkali diagram (Figure 2), and hence they define a broad, moderately alkalic lava series. Our samples do not include hawaiite, suggesting a “Daly gap” or paucity of intermediate-composition lavas, as in many oceanic islands [Chayes, 1963, 1977; Yoder, 1973]. The remainder of Rosa samples includes basanites, a phonolitic tephrite and a trachyandesite. Lava samples from the high Sara volcanic ridge also include alkalic basalts in addition to transitional basalts, trachyandesites, and an andesite. The andesite and almost all of the trachyandesites (except sample ROSA D6-10) actually belong to the tholeiitic lava series although, as will be discussed below, they are not geochemically depleted like NMORB, and thereby resemble tholeiitic lavas from intraplate eruptions at many volcanic chains. Fewer samples were dredged from the volcanic ridges along Rosana and Nithya axes. Of the four samples from Rosana, two are alkalic basalts and two are tholeiitic basalts. The three samples from Nithya are alkalic basalts to transitional basalts similar to enriched MORB (EMORB).

As a whole, the data display broad correlations between MgO and major elements (Figure 3). For example, concentrations of SiO2, Al2O3, Na2O, and K2O in Rosa lavas increase whereas Fe2O3T and to a lesser degree CaO decrease with decreasing MgO concentrations. Some of the major element oxides of Sara lavas behave similarly as those of Rosa lavas, but the former have lower Na2O and K2O and higher CaO for given MgO contents than the latter. Moreover, the correlation of Al2O3 with MgO of Sara lavas is opposite to that of Rosa lavas. The few samples from Rosana and Nithya do not define clear differentiation trends; nevertheless, they generally plot together with the basaltic lavas from Sara.

Figure 3.

Wt % MgO versus major elements oxides SiO2, Al2O3, Fe2O3T, CaO, Na2O, and K2O for samples from Sara, Rosa, Rosana, and Nithya. Field and fractional crystallization (FC) trend (solid curve with arrows) computed from a primitive basalt using the MELTS software package [Ghiorso and Sack, 1995; Asimow and Ghiorso, 1998] from Davidson Seamount [Castillo et al., 2010] are shown for reference.

All samples are enriched in highly incompatible relative to moderately incompatible trace elements (Figure 4). Rosa has the most enriched pattern among the four axes, save for differentiated lavas from Sara that also have the most distinct negative Sr anomalies. Moreover, individual Rosa samples show subparallel trace element patterns, which collectively are more enriched in highly incompatible trace elements than an average alkalic OIB [Sun and McDonough, 1989]. They do, however, display slight positive spikes in high field strength elements (HFSEs) such as Nb and Zr, which is typical in OIB [Hofmann, 2003]. Basaltic lavas from Sara, Rosana and Nithya have lower highly incompatible but higher moderately incompatible trace element contents than Rosa lavas. In other words, the trace element patterns of the basaltic lavas from these three ridges cut across those of Rosa lavas.

Figure 4.

Incompatible trace element concentration diagrams for samples from Sara, Rosa, Rosana, and Nithya. The average OIB and normalizing values are from Sun and McDonough [1989].

The samples have moderate ranges in 87Sr/86Sr (0.7027 to 0.7035), but a constant 143Nd/144Nd ratio of ∼0.51295 (total range = 0.51289 to 0.51300; Figure 5). As a whole, the Sr and Nd isotope values of these four volcanoes fall within the range of other fossil spreading axes west of California [e.g., Davis et al., 2002; Castillo et al., 2010] and overlap with the values for basaltic rocks from Socorro and San Benedicto islands along the fossil Mathematician Ridge [e.g., Bohrson and Reid, 1995, 1997]. Furthermore, they overlap with the radiogenic, high 87Sr/86Sr and low 143Nd/144Nd end of the field for seamounts on the EPR flanks [e.g., Zindler et al., 1984; Graham et al., 1988; Niu et al., 2002]. Interestingly, their 87Sr/86Sr and 143Nd/144Nd isotopic compositions also overlap with some of the OIB from the Galapagos archipelago and Iceland, save for one mugearite from Rosa Seamount, which overlaps with the OIB from the Hawaiian volcanic chain. The OIB from the large Galapagos, Iceland and Hawaiian hot spot volcanoes that overlap with the fossil spreading lavas in the Sr and Nd isotopic space have among the highest (≥15 RA) terrestrial 3He/4He ratios [e.g., Stuart et al., 2003; Class and Goldstein, 2005].

Figure 5.

(top) 87Sr/86Sr versus 143Nd/144Nd and (middle and bottom) 206Pb/204Pb versus 207Pb/204Pb and 208Pb/204Pb for representative samples from Sara, Rosa, Rosana, and Nithya. Data for Davidson Seamount [Davis et al., 2002; Castillo et al., 2010], Guide Seamount [Davis et al., 2002], and Socorro Island [Bohrson and Reid, 1995, 1997] are shown for reference. Also shown are fields for some seamounts on the EPR flanks [Niu et al., 2002], EPR MORB, Galapagos Islands, Hawaii, and Iceland (GEOROC database: HIMU, EM1, and EM2 are proposed mantle end-components, and NHRL is the northern hemisphere reference line from Hart [1984].

The Pb isotopes of the samples are relatively radiogenic and span a wide range of values (e.g., 206Pb/204Pb = 18.86 to 19.53). Consistent with Sr and Nd isotopes, their Pb isotopes overlap the available isotope data for other fossil spreading axes and the radiogenic end of riseflank seamounts (Figure 5). Moreover, the Pb isotopic composition of the majority of lavas from these four axes overlaps with the OIB from the large Galapagos, Iceland and Hawaiian hot spot volcanoes, again those with high 3He/4He ratios.

Measured 3He/4He ratios and He concentrations are corrected for the effects of air contamination following procedures described by Hilton [1996]. The air-corrected 3He/4He ratios and 4He concentrations in vesicles and fluid inclusions in fresh glasses and clinopyroxene from the axes of Rosa and Nithya, respectively, range from 1.25 to 7.18 RA (where RA = 3He/4He of air) and 0.26 to 23.4 × 10−9 cm3 STP g−1 (Table 4 and Figure 6). Thus, to a first order, none of the fossil spreading center OIB-like basalts has 3He/4He ratio higher than the canonical MORB value of 8 ± 1 RA [Graham, 2002]. It is important to note, however, that four out of seven analyses are dominated by air-derived He (X < 3.5) and their He concentrations are low ([He]c < 1.6 × 10−9 cm3STP/g). Under these circumstances, it is possible that either the correction procedure fails to completely resolve air from trapped He or any trapped He includes a radiogenic component produced within the sample after eruptive. Therefore, in the following discussion, we give greater weight to those three samples where X > 10, i.e., there is little air correction. Thus, two Rosa samples give air-corrected 3He/4He ratios of 7.18 RA (ROSA D9-1rep.) and 5.67 RA (ROSA D9-9) while the clinopyroxene sample from Nithya gives a 3He/4He value of 1.25 RA (PX 127D-20).

Figure 6.

Relationships between 3He/4He ratios and 4He abundances for representative samples from Rosa Seamount, Nithya ridge, and Davidson Seamount. Shown for reference are data for axial ridges on Mathematician Ridge [Graham et al., 1988], EPR riseflank seamounts [Graham et al., 1988; Hahm et al., 2009], Galapagos Islands, Hawaii, and Iceland [Abedini et al., 2006; USGS-NoGaDat:].

The two Rosa samples are alkalic lavas that have relatively high concentrations of U and Th, so their 3He/4He ratios at the time of eruption may have been higher than present-day values. However, a substantial decrease in vesicle-sited 3He/4He values can only occur with almost complete diffusion of 4He* from the glass matrix (where it is produced) to the vesicle phase. For example, Graham et al. [1987] predicted that a maximum of 10% of the 4He* would diffuse from the glass matrix to the vesicles if the vesicularity was 1% and the vesicle radius was between 100 and 300 mm. The lavas in this study are highly vesicular so that the 3He/4He ratios of the larger vesicles, which are less affected by 4He* diffusion, would dominate the measured 3He/4He [Graham et al., 1987]. Moreover, although the fossil spreading lavas of this study could be as old as 11.5 Ma, based on the tectonic setting, we note earlier that the samples collected from the summit of Rosa Seamount, including ROSA D9-1 and ROSA D9-9, consist of pristine glass and appear to have been erupted recently. Thus, we believe that the 3He/4He ratios of samples ROSA D9-1 and ROSA D9-9 from Rosa Seamount can be considered representative of their mantle source at the time of eruption. Notably, volcanoes along axes of the fossil Mathematician Ridge [Graham et al., 1988] and riseflank seamounts studied by Hahm et al. [2009] have similarly low (≤8 RA) 3He/4He ratios. In contrast, the clinopyroxene sample from Nithya has clearly incorporated posteruptive radiogenic He, consistent with the antiquity of the volcano.

5. Discussion

5.1. Geochemical Variations of Postspreading Lavas on Fossil Spreading Axes off Baja California Sur

The volcanic lavas erupted along the four fossil spreading axes display two types of compositional variation: (1) variations in the major and trace element contents of the alkali basalt to benmoreite lava series, e.g., from Rosa Seamount, that can be ascribed to fractional crystallization and (2) intra-axis and inter-axis geochemical variations due to a compositionally heterogeneous mantle source. The broad linear correlations between MgO and other major element oxides exhibited by the Rosa Seamount lava series (Figures 2 and 3) are consistent with differentiation trends through fractional crystallization. The same can be said for the subparallel incompatible trace element concentration patterns of these Rosa Seamount lavas (Figure 4). Specifically, the concentrations of incompatible elements such as Rb, Ba, Th, U, REEs, and HFSEs generally increase from primitive basalts to differentiated benmoreites (Figure 7). The reverse is true for compatible trace elements such as Ni and Cr, which suggests that the differentiated lavas were derived from similar, parental primitive basalts through fractionation of minerals that included olivine and clinopyroxene. More importantly, basalts, mugearites and a benmoreite from the Rosa Seamount lava series have a limited range of Sr and Nd plus identical Pb isotope ratios (Figure 5 and Table 3), clearly indicating that this lava series came from a single mantle source. In detail, however, the lava series defines bimodal groups in plots of Ba versus Zr (Ba/Zr = ∼1.4. and 1.86) and Nb versus Zr (Nb/Zr = ∼0.2 and 0.3) (Figure 8), suggesting that the lava series was actually generated by a complex fractional crystallization process that likely occurred within the mantle at pressures that increased as the volcano grew, similar to the generation of the main lava series at Davidson Seamount [Clague et al., 2009; Castillo et al., 2010]. The presence of a few Rosa Seamount lavas that do not belong to this lava series (Figures 3 and 7) and have different Sr, Nd and Pb isotopic ratios (Figure 5 and Table 3) indicates that the mantle source of Rosa Seamount lavas was not compositionally homogeneous.

Figure 7.

Wt % MgO versus ppm trace elements Ba, Th, La, Nb, Cr, and Ni for samples from Sara, Rosa, Rosana, and Nithya.

Figure 8.

(a) Zr versus Ba, (b) La versus Sm, (c) Zr versus Nb, and (d) Ta versus Nb for samples from Sara, Rosa, Rosana, and Nithya. Shown for reference are data for other fossil spreading axes and riseflank seamounts. Sources of data as in Figure 5.

As noted earlier, the basaltic and andesitic lavas from Sara ridge range from alkalic to tholeiitic in composition (Figure 2) and thus, unsurprisingly, their major element data do not form a single linear trend with MgO (Figure 3). Moreover, their trace element concentration patterns appear to form two separate groups (Figure 4) and their Sr, Nd and Pb isotopic compositions are variable and are, on average, different from those of Rosa Seamount (Figure 5). Thus, unlike the bulk of Rosa Seamount lavas, Sara lavas cannot simply be related to one another through fractional crystallization of similar parental basalts. They originate from a compositionally heterogeneous mantle source, as do the basaltic samples from Rosana and Nithya ridges. As a whole, the trace element and isotopic signatures of Rosana and Nithya are overlapping with that of Sara; altogether, these are less geochemically enriched than those of Rosa Seamount (Figures 4 and 5).

5.2. Comparison With Postspreading Lavas on Other Fossil Spreading Axes in the Eastern Pacific

As described earlier, there are many volcanic ridges or seamounts located along fossil spreading axes in the eastern Pacific, and alkalic volcanism is a common feature of these failed axes. Samples dredged from an axial ridge on fossil Galapagos Rise are alkalic basalts that have experienced variable degrees of crystal accumulation/fractionation and seawater alteration [Batiza et al., 1982; Haase and Stroncik, 2002]. Although Batiza et al. [1982] reported sampling tholeiitic lava from the crest of the Galapagos Rise, they had mislocated the former spreading axis by ∼35 km; the geophysical mapping of Eakins and Lonsdale [2003] showed the axis to lie along the 3 km high “elongate seamount” from which Batiza et al. [1982] reported alkali-olivine basalt (their dredges 4 and 5). Samples from volcanic ridges along the Mathematician Ridge crest (Figure 1) are vesicular alkalic basalts with large crystals of olivine and plagioclase [Batiza and Vanko, 1985], benmoreites [Richards, 1958] and trachyandesite [Moore, 1970]. Note that Batiza and Vanko [1985] originally described sampling their alkalic basalts from abandoned transform faults, but subsequent multibeam bathymetry (SIO, archival data) showed them to be from axial volcanoes on failed Mathematician spreading segments.

Socorro and San Benedicto Islands at the northern terminus of the Mathematician Ridge provide more evidence of postabandonment alkaline magmatism that continues to the present, as indicated by recent eruption in Socorro Island in 1993 [McClelland et al., 1993] and San Bendicto Island in 1952–1953 [Richards, 1959]. The bulk of postcaldera volcanism in Socorro Island spans the fractional crystallization trend from alkalic basalt to hawaiite and mugearite [Bohrson and Reid, 1995]. The alkalic basalts have chemical characteristics (e.g., negative Ce anomaly and P2O5 enrichment) which are indicative of assimilation of hydrothermal sediments or apatite accumulated during previous episodes of magmatism. San Benedicto Island has an alkalic basalt to mugearite and trachyte lava series [Richards, 1966] which almost completely overlaps with the Socorro Island alkalic lava series in major and trace element composition [Bohrson and Reid, 1995]. Significant volumes of silicic peralkaline volcanic rocks (dominantly peralkaline trachytes and rhyolites) also occur on Socorro and San Benedicto Islands (Figure 1), and these may have been formed by partial melting of basaltic basement [Bohrson and Reid, 1997, 1998].

Samples collected from Guide Seamount (Figure 1) are mainly hawaiites, but also include alkalic basalts and mugearites [Davis et al., 2002, 2010]. The broad spectrum of major element compositions of the lavas, especially among the hawaiites, reflects the chemical diversity of melts erupted at Guide Seamount. Samples from Davidson Seamount (Figure 1) belong to an alkalic basalt to trachyte lava series generated through a complex fractional crystallization process [Clague et al., 2009; Castillo et al., 2010]. A diverse assemblage of small mafic and ultramafic xenoliths are present in the alkalic lavas from Davidson Seamount and these represent both trapped melts from the mantle and alkalic cumulates from the margins of magma chambers [Davis et al., 2002, 2007]. Sampled Davidson Seamount lavas spanned five million years and the older, more differentiated and alkali-rich lavas occur near the summit of Davidson Seamount [Clague et al., 2009].

Relatively smaller volumes of tholeiitic to transitional basalts have erupted at Socorro Island, the northern rift zone of San Benedicto Island [Bohrson and Reid, 1995], and at a small volcano near the fossil spreading axis-transform intersection south of Davidson Seamount [Castillo et al., 2010]. Most of these tholeiitic basalts, however, are slightly more alkalic than NMORB [Bohrson and Reid, 1995]. Castillo et al. [2010] propose that the few transitional basalts from south of Davidson Seamount were produced from the same source of the Davidson Seamount alkalic lava series, although at higher degree of partial melting.

Similar to Rosa Seamount, alkalic lavas from large volcanoes built along other fossil spreading axes in the eastern Pacific have trace element abundances and ratios resembling many alkalic ocean island lavas [e.g., Castillo et al., 2010]. The alkalic lavas from fossil spreading centers have higher abundances of incompatible elements and higher highly/moderately incompatible trace element ratios relative to tholeiitic basalts from riseflank seamounts (e.g., higher Ba/Zr, La/Sm and Nb/Zr ratios; Figure 8). The highly incompatible trace element ratios of some of the transitional lavas are overlapping with those of the alkalic lavas [e.g., Castillo et al., 2010]. There are also similarities or overlaps in the Sr, Nd and Pb isotopic compositions of all alkalic lavas on all fossil spreading axes (Figure 5), and these compositions overlap the radiogenic end of Pacific MORB and trend toward to an enriched mantle component that contributes to the source of OIBs.

5.3. Petrogenesis of Postspreading Lavas on Fossil Spreading Axes

Rosa, Sara, Rosana, and Nithya lavas were erupted in identical tectonic settings, along short spreading axes that had failed simultaneously, but show significant compositional variation. From this variation and observations at better-dated counterparts along other fossil spreading axes, we infer temporal change in the composition of postspreading eruptions.

The bulk of the lavas sampled from the largest volcanic ridge, Rosa Seamount (i.e., the one that probably has the longest history of volcanic eruption, with apparently recent flows on its summit) are alkalic in composition. As discussed above, most superficial lavas from prominent volcanic ridges and islands covering segments of fossil spreading axes in the eastern Pacific are also alkalic. In comparison, superficial lavas from the smaller volcanic features along Sara, Rosana and Nithya axes are less alkalic on average, with several samples having tholeiitic to transitional compositions, similar to EMORB. Lavas from a small seamount located at the axis of a fossil spreading center in Drake Passage in the southern Pacific Ocean, erupted <2 Myr after the cessation of spreading, are also EMORB [Choe et al., 2007; Choi et al., 2008]. The evidence suggests that the composition of risecrest lavas generally evolves from typical tholeiitic NMORB while actively spreading to EMORB and eventually to alkalic basalts, similar to OIB, and their differentiates, after spreading has stopped.

A better understanding of the mantle melting process beneath active spreading centers can provide clues as to why the composition of postspreading erupted lavas evolve with time. The mantle rises beneath active risecrests in response to plate separation, decompresses adiabatically, begins to melt when it intersects the solidus, and continues melting to produce voluminous MORB. The whole process occurs in a melting column, which is defined by the mantle potential temperature that sets the depth of intersection between the solidus and the melting curve or by the melting interval between initial (deeper and hence higher pressure) and final (shallower and hence lower pressure) depth of melting, which in turn is defined by the balance between conductive cooling on the seafloor and adiabatic upwelling of hot mantle [e.g., Langmuir et al., 1992; Niu and Hékinian, 1997; Rubin and Sinton, 2007; Niu and O'Hara, 2008]. Melting columns heights correlate with variations in axial topography, which are controlled by either mantle temperature [e.g., Langmuir et al., 1992], mantle source mineralogy [e.g., Niu and O'Hara, 2008] or plate separation rate [e.g., Niu and Hékinian, 1997; Rubin and Sinton, 2007]. More relevant to the current discussion is variations in spreading rate. Compared to faster spreading centers, slower spreading ridges have shorter melting columns that produce less melt, which results from lower degrees and, to a limited extent, higher pressure of decompression partial melting of the mantle [Langmuir et al., 1992; Niu and O'Hara, 2008]. By analogy, when spreading slows to a stop at dying risecrests, the amount of upwelling mantle also diminishes, resulting in even shorter melting columns and smaller degree of decompression partial melting. In general, basalts formed at lower degrees and greater pressures of melting are more alkalic, silica undersaturated and olivine normative compared to tholeiites that are formed at higher degree and lower pressure of melting [e.g., Langmuir et al., 1992; Dasgupta et al., 2007; Pilet et al., 2008, and references therein]. Thus, lowering the degree of partial melting surely plays a major role in creating the general evolution of fossil spreading axes from NMORB to alkalic OIB-like basalts [see also, Choe et al., 2007; Choi et al., 2008; Castillo et al., 2010].

It is important to note, however, that Rosa Seamount alkalic lavas are generally more enriched (i.e., higher Ba/Zr, La/Sm, Nb/Zr, 87Sr/86Sr, 206Pb/204Pb, but lower 143Nd/144Nd) than the less alkalic and tholeiitic lavas from Sara, Nithya and Rosana, which themselves still are geochemically more enriched than NMORB (Figures 4, 5, and 8). Moreover, the apparently recent flows on Rosa Seamount indicate that it is still volcanically active ∼11–12 Myr after the cessation of spreading; Davidson Seamount remained volcanically active ∼8 Myr after spreading stopped there [Clague et al., 2009]. Thus, decreasing degree of decompression melting alone cannot fully explain the observed compositional evolution and duration of magmatism along fossil spreading axes. Here we infer that a heterogeneous suboceanic mantle source is the other factor, along with a decreasing degree of partial melting, can explain the evolution of lava types in fossil spreading axes. The nature of such a heterogeneous mantle will be discussed in detail below.

5.4. Nature of the Mantle Source

Castillo et al. [2010] note that major and trace element concentration and radiogenic isotope data for fossil spreading axes and riseflank seamounts in the eastern Pacific define a coherent, broad compositional spectrum that is bounded by tholeiitic NMORB-like and alkalic OIB-like lavas. They then adopt the generally accepted model for the origin of riseflank seamount lavas [e.g., Batiza and Vanko, 1984; Zindler et al., 1984; Graham et al., 1988] to explain this broad compositional spectrum. The model claims that the mantle source of MORB is heterogeneous, consisting of easily meltable, enriched components embedded in a depleted lherzolitic matrix. The OIB-like lavas originate from small degree of partial melting that preferentially fuses the enriched components in the heterogeneous source, producing a small amount of melt that bypasses the magma transport processes beneath the EPR that tend to homogenize MORB compositions. On the other hand, the NMORB-like lavas most likely result from large degree of partial melting that fuses both the enriched and more dominant lherzolite components. The enriched signature in the resultant melt is buffered/diluted by the depleted signature during melt aggregation and mixing within crustal magma chambers. We adopt this model to explain postspreading magmatism in the fossil spreading axes off Baja California Sur as well. As spreading slows down and/or even after it has ceased, mantle upwelling still occurs but at a diminishing rate; this scenario generally favors progressive sampling of the enriched, easily meltable components in the heterogeneous suboceanic mantle at the expense of the depleted lherzolitic matrix, producing a compositional continuum in the fossil spreading axis lava record.

We have also emphasized the tendency for temporal change from tholeiitic to alkalic eruptions at ridges built along abandoned risecrests and at central volcanoes that form riseflank seamounts. This temporal evolution combined with the aforementioned variable degrees of partial melting model for fossil spreading center magmatism defines an evolutionary scheme that is akin to the more recent alternative explanation for the origin of linear hot spot volcanoes that have traditionally been interpreted as products of deep mantle plumes [e.g., Morgan, 1971, 1972; Hofmann, 2003]. The alternative explanation invokes partial melting of upper mantle heterogeneities that happen to underlie lithospheric fractures [e.g., Meibom and Anderson, 2004; Natland and Winterer, 2005; Foulger, 2007]. Does this mean that magmas erupted along the healing lithospheric fractures along the axes of abandoned risecrests may have the same mantle source as the magmas that built, e.g., Galapagos and the Hawaiian volcanic chain? We conclude that their sources are different, from the absence in our samples of the high 3He/4He ratios (>8 +/− 1 RA) that are distinctive features of these major volcanic systems [e.g., Graham, 2002]. One may argue that fossil spreading magmas may be coming from the HIMU mantle source, which produces some OIB with low 3He/4He ratios [e.g., Eiler et al., 1997; Hanyu et al., 1999]. However, the Sr and Nd isotopic signature of the bulk of fossil spreading lavas, and Rosa samples in particular, do not trend to the HIMU end-component (Figure 5). Moreover, the “classic” HIMU islands are relatively rare [Stracke et al., 2005] and some actually have high 3He/4He ratios (>8 +/− 1 RA) [Hilton et al., 2000]. Thus, if high 3He/4He ratio is considered diagnostic of the deep mantle plumes, we conclude that no such plumes were involved in building ridges along the abandoned EPR crest that we studied.

The production of the range of lava compositions along Sara, Rosa, Rosana, and Nithya axes by decreasing degree of partial melting of a heterogeneous upper mantle is illustrated in the Rb/Sr versus 87Sr/86Sr and La/Sm versus 143Nd/144Nd diagrams (Figure 9). For simplicity, mixing of melts is assumed although actual mixing of sources prior to melting is also a possibility. In order to evaluate the model quantitatively, the depleted MORB mantle (DMM) and an enriched mantle component (EC) are chosen as the putative two end-member sources for the primary magmas of these volcanoes. The composition of DMM is obtained from that of the average depleted (D)MORB by Salters and Stracke [2004]. The composition of EC is not well defined except that it must be more enriched than our samples (e.g., 87Sr/86Sr > 0.703536 and 143Nd/144Nd < 0.512894; Figure 5 and Table 3); here we back-calculated its composition from a geochemically enriched lava from the nearby Flint (off-axis) seamount [Davis et al., 1995; Konter et al., 2009]. The lithology of the EC is assumed to be an amphibole-rich lherzolite [Pilet et al., 2008]. To avoid any effect of fractional crystallization, only primitive basalts are used in the model calculations. Changes in Rb/Sr and La/Sm ratios (subhorizontal solid curves) represent mixing lines between variable degrees of partial melt (0.05% to 2%) from DMM and 10% melt from EC. On the other hand, changes in 87Sr/86Sr and 143Nd/144Nd ratios (subvertical dashed curves) represent changes in the proportions of the two sources for a given mixture. Although the exact amounts are different, the most enriched basalts (highest Rb/Sr and 87Sr/86Sr plus relatively higher La/Sm and lower 143Nd/144Nd ratios) from Rosa generally contain the lowest amount of low-degree partial melt from DMM. In contrast, the alkalic to tholeiitic basalts from Sara, Rosana and Nithya contain more of higher-degree partial melt from DMM. In other words, the geochemically enriched alkalic lavas from Rosa produced by smaller degree of partial melting contain lesser amount of DMM, but larger proportion of enriched component than the relatively less enriched alkalic to tholeiitic Sara, Rosana and Nithya. Rosa lavas also contain less of the lower-degree partial melt from DMM than most of the seamounts on the EPR flanks. Tholeiitic basalts sampled from the EPR axis have, of course, the highest content of the highest-degree partial melt from DMM.

Figure 9.

(a) Rb/Sr abundance ratio versus 87Sr/86Sr and (b) La/Sm abundance ratio versus 143Nd/144Nd for primitive basalts from Sara, Rosa, Rosana, and Nithya. The vertical line with numbers from depleted MORB mantle (DMM) indicates Rb/Sr and La/Sm abundance ratio at 0.05%–2% melting of a depleted MORB mantle source (Rb = 0.088 ppm, Sr = 9.80 ppm, 87Sr/86Sr = 0.7026; La = 0.234 ppm, Sm = 0.270 ppm, Nd = 0.713 ppm, 143Nd/144Nd = 0.51311) [Salters and Stracke, 2004]. Here the initial mode of the depleted MORB mantle source was assumed to be olivine:orthopyroxene:clinopyroxene:garnet (60:25:10:5 mixture) lherzolite. An equilibrium batch partial melting model is used for calculation; bulk D (partition coefficient) for Rb is 0.0003, for Sr 0.0093, for La 0.0015, for Sm 0.0406, and for Nd 0.0153 in the lherzolite. The subhorizontal linear solid curves are mixing lines between the various DMM melts (F%, degree of partial melting) and 10% melt of an enriched mantle component (EC) (Rb = 6.45 ppm, Sr = 127.5 ppm, 87Sr/86Sr = 0.704162; La = 4.61 ppm, Sm = 3.85 ppm, Nd = 10.64 ppm, 143Nd/144Nd = 0.512869). The trace element abundances of EC are back-calculated by assuming that a primitive basalt from Flint seamount [Davis et al., 1995] was produced by 20% partial melting of this source; its Sr and Nd isotopic values are from Konter et al. [2009], normalized relative to NBS 987 87Sr/86Sr = 0.71025 and 143Nd/144Nd = 0.51185 for the La Jolla Nd standard. The EC source is assumed to be an amphibole:olivine:orthopyroxene:clinopyroxene:garnet (55:20:05:15:05) lherzolite. Bulk D for Rb is 0.1981, for Sr 0.1990, for La 0.089, for Sm 0.579, and for Nd 0.282 in the amphibole lherzolite. Vertical dashed curves represent various DMM:EC melt mixtures. Detailed predictions of this model depend on the initial mineralogy of two end-members, source composition, style of melting, and the chosen D values, but the applicability of the model results is not affected by these parameters. Fields for basalts from Davidson Seamount, Socorro Island, selected riseflank seamounts, and adjacent EPR axis are shown for reference. Data sources are as in Figure 5.

6. Summary and Conclusions

A highly segmented part of the EPR crest west of Baja California Sur, Mexico, ceased spreading at the end of the middle Miocene, but continued eruptions built postspreading volcanic ridges along the former spreading axes.

1. The bulk of the lavas dredged from Rosa Seamount, the largest volcanic ridge and therefore the one most likely to have developed a long-lived magma chamber, belong to an alkalic rock series generated through fractional crystallization. Samples from the smaller Sara, Rosana and Nithya ridges mainly include mildly alkalic, transitional and tholeiitic basaltic and andesitic rocks.

2. The petrologic and geochemical data from the four abandoned axes provide new evidence that postabandonment alkalic volcanism is a common feature of large failed axes in the eastern Pacific. Their compositions overlap with those of lavas on other fossil spreading axes and riseflank seamounts, defining a compositional continuum ranging from MORB-like to OIB-like.

3. Geochemical differences among lavas erupted along the four abandoned axes result from different degrees of partial melting of a compositionally heterogeneous suboceanic mantle and from different amounts of fractional crystallization. We infer that a large extent of partial melting may continue for a geologically brief period after spreading ceases, initially producing tholeiitic MORB-like melts, then alkalic OIB-like melts as partial melting wanes. This temporal transition mirrors that inferred from the spatial pattern of riseflank seamounts, where near-axis volcanoes commonly erupt tholeiitic MORB-like lava, succeeded by alkalic OIB-like lava further down the riseflanks.

4. Although the Sr, Nd and Pb isotopic signatures of the alkalic OIB-like lavas from fossil spreading axes in the eastern Pacific are similar to those of OIB from the Galapagos archipelago, Iceland and Hawaiian volcanic chain, they differ in lacking high 3He/4He ratios. This indicates that the origin of fossil spreading center magma is different from that of OIB magma comprising major “hot spot” volcanoes.


We are grateful to R. Solidum for analysis of some major elements data by XRF at SIO, C. MacIsaac, G. Lugmair, and X. Liu for their assistance with the TIMS analysis, and J. Kluesner for help with Figure 1. We also thank two anonymous reviewers for their constructive comments and suggestions which significantly improved the paper and Joel Baker for thorough editorial handling. This work was funded by NSF OCE0550237 grant to PRC and DRH, NSF OCE9116493 grant to PFL, and the University of California Ship Fund.