Lithospheric control on geochemical composition along the Louisville Seamount Chain



Major and trace element and Sr, Nd, and Pb isotope data for lavas from 12 seamounts along the western (older) 1500 km section of the Louisville Seamount Chain in the southwest Pacific show remarkably uniform compositions over a ∼30–40 Myr period of volcanism. All 56 samples analyzed are alkalic to transitional in composition. Unlike Hawaiian volcanoes, Louisville volcanoes appear not to pass through a sequence of evolutionary stages characterized by older tholeiitic basalts overlain by incompatible element enriched alkalic and silica-undersaturated lavas. The youngest lavas from a given Louisville seamount tend to have the least enriched incompatible element compositions. This unusual chemical evolution may be the result of re-melting of heterogeneous hot spot mantle that was partially depleted during the earlier, age progressive stages. The oldest Louisville seamounts were constructed close to the extinct Osbourn Trough spreading center, located north of the chain, but age-progressive lavas from these older seamounts are not significantly different to lavas from younger seamounts. This may indicate that spreading at this fossil ridge ceased several tens of millions of years before the oldest Louisville seamounts were constructed. Large fracture zones apparently had no significant effects on the composition of Louisville magmatism. However, lavas from the central part of the Louisville Seamount Chain, where volcanoes are smaller and more widely spaced, tend to have more variable and more enriched compositions. We suggest this may reflect smaller degrees of melting resulting from greater lithosphere thickness, and hence a shorter melting column for this section of the Louisville Seamounts.

1. Introduction

Changes in the chemical and isotopic composition of lavas erupted from volcanoes along long-lived seamount chains can provide important insights into the origin of intraplate magmatism. Along-chain (temporal) variations in lava composition may arise from variations in the composition of the mantle source, changes in the melting process, and/or variability in the thickness of the underlying lithosphere, which influences the degree and depth of melting beneath intraplate volcanoes [Basu and Faggart, 1996; Haase, 1996; Humphreys and Niu, 2009; Ito and Mahoney, 2005; Regelous et al., 2003]. These effects can be examined most easily by comparing the lavas erupted along a single seamount chain. In the Pacific, the Hawaiian-Emperor and Louisville Seamount Chains preserve a >70 Ma history of continuous, predominantly age-progressive volcanism [Duncan and Keller, 2004; Koppers et al., 2004; Sharp and Clague, 2006; Watts et al., 1988; A. A. P. Koppers et al., New location of the present-day Louisville hot spot based on 40Ar/39Ar geochronology, submitted to Geochemistry, Geophysics, Geosystems, 2011; J. O‘Connor et al., Past mantle and plate motion from hot spot trails and bends: 1. A combination of age progressive and secondary volcanism in the Louisville Seamount Chain, submitted to Geochemistry, Geophysics, Geosystems, 2011]. Whereas the chemical evolution of volcanism along the Hawaiian-Emperor Chain is reasonably well known from analysis of drilled and dredged samples [e.g., Basu and Faggart, 1996; Keller et al., 2000; Regelous et al., 2003], the Louisville Seamount Chain is much less well sampled, and few geochemical data exist for Louisville seamount lavas [Cheng et al., 1987; Hawkins et al., 1987].

Here we present new major element data for 56 samples of volcanic rock, together with trace element and Sr, Nd and Pb isotope data for selected samples, from 12 seamounts in the westernmost 1500 km of the Louisville Seamount Chain (LSC) between 175°W and 168°W. Together with new data for the younger part of the LSC between 168°W and 156°W [Vanderkluysen et al., 2007, L. Vanderkluysen et al., Geochemical evolution of the Louisville Seamount Chain, manuscript in preparation, 2011], our data can be used to examine the chemical and isotopic evolution of magmatism on the LSC over a distance of approximately 2850 km and an age range between 79 Ma and 25 Ma, and evaluate the possible influence of fracture zones and variations in lithosphere thickness on the composition of intraplate magmatism.

2. Tectonic Setting and Sample Locations

The LSC is an approximately 4200 km long chain of seamounts and guyots on the Pacific Plate, extending southeastward from the 80 Ma Osbourn Guyot (located close to the Tonga Trench at 25°52.3S, 175°05.4W; Figure 1), to a seamount at 50°26.4S, 139°10.0W (about 300 km west of the Pacific-Antarctic spreading center), where a sample of alkali basalt (MTHN7–1) dated at 1.11 Ma may represent the most recent Louisville volcanism [Cheng et al., 1987; Hawkins et al., 1987; Koppers et al., 2004; Lonsdale, 1988; Watts et al., 1988]. The Hollister Ridge, a NW-SE trending aseismic ridge immediately west of the Pacific-Antarctic Ridge, lies approximately 250 km south of sample MTHN7–1 and has been proposed to be part of the LSC [Wessel and Kroenke, 1997]. However, geochemical data for Hollister Ridge lavas do not support this hypothesis [Vlastelic et al., 1998]; instead the Hollister Ridge may be the result of intraplate deformation [Géli et al., 1998]. The oldest sampled Louisville seamounts are situated close to the Osbourn Trough (Figure 1), an east-west trending structure that is thought to represent a fossil spreading ridge that became inactive at some time before 87 or 93 Ma, depending on when the Manihiki and Hikurangi plateaus split apart [Downey et al., 2007; Worthington et al., 2006]. At approximately 167°W, the LSC crosses the West Wishbone Scarp, a north-south trending ridge that has been variously interpreted as a fracture zone, a fossil spreading center between the Phoenix and Pacific Plates, or a former intraoceanic arc [Billen and Stock, 2000; Luyendyk, 1995; Mortimer et al., 2006]. The East Wishbone Scarp crosses the LSC at approximately 163° (Figure 1) and has been interpreted as a remnant of a transform fault belonging to the Phoenix-Moa Ridge system [Sutherland and Hollis, 2001]. The youngest, easternmost Louisville seamounts tend to be smaller and more widely spaced, so that the easternmost end of the volcanic chain is less well defined [Lonsdale, 1988]. Louisville seamounts between 148°W and 155°W were constructed close to the large Tharp and Heezen Fracture Zones.

Figure 1.

Bathymetric map of the western Pacific, showing the western part of the Louisville Seamount Chain, together with the location of SO167 dredge samples analyzed in this study (circles), those from the AMAT02RR cruise (diamonds) studied by Vanderkluysen et al. [2007, also manuscript in preparation, 2011] and literature data [Cheng et al., 1987; Hawkins et al., 1987]. Map prepared using GMT [Wessel and Smith, 1991, 1998]. Black and white inset shows simplified tectonic map of the western Pacific (modified from Vlastelic et al. [1998]), showing location of Louisville Seamount Chain, Tonga-Kermadec Trench, major fracture zone systems and Pacific spreading ridges, and the location of sample MTHN7–1 dated at 1.11 Ma [Koppers et al., 2004]. Lonsdale [1988] proposed naming the Louisville seamounts according to their latitude or longitude. Labels correspond to seamount numbers, i.e., L37 is Louisville seamount 37. The seamounts sampled during the SO167 cruise were named using a different system [Stoffers et al., 2003], but the latitude and longitude of all samples are listed in Table 1, allowing easy comparison with the earlier sample set.

Early total fusion 40Ar-39Ar ages for lavas from eight locations along the length of the LSC [Watts et al., 1988] appeared to indicate a highly linear age progression. However, Koppers et al. [2004] re-dated the same samples using the incremental heating method, and found that the oldest lavas are in fact almost 10 Myr years older and thus closer to 79 Ma and that the age-progression along the seamount chain was not as linear as previously thought. More recent 40Ar-39Ar ages for new samples collected during the AMAT02RR and SO167 cruises (Koppers et al., submitted manuscript, 2011; O'Connor et al., submitted manuscript, 2011) have confirmed these findings. These new studies have found, in addition, that several of the westernmost Louisville seamounts were apparently volcanically active for more than 10 Myr, and at least one volcano was active over a 23 Myr period (O'Connor et al., submitted manuscript, 2011). In contrast, most Hawaiian volcanoes have an active lifetime of at most 6 Myr between the shield and post-erosional (rejuvenated) stages [Clague and Dalrymple, 1989]. Louisville volcanoes to the east of the West Wishbone Scarp show a more regular age progression, and no evidence for anomalously young, rejuvenated magmatism (Koppers et al., submitted manuscript, 2011).

Currently, there are few geochemical analyses of LSC lavas. Major element and limited trace element data for 22 samples from 11 volcanoes along the LSC were presented by Hawkins et al. [1987], and Cheng et al. [1987] reported Sr, Nd and Pb isotope data (Sr and Nd isotope data were partially age-corrected; Pb isotope ratios were uncorrected) for 15 of these samples. These studies found that lavas from Louisville volcanoes are dominantly alkalic or transitional in composition and have moderately radiogenic Pb isotope compositions with present-day 206Pb/204Pb ratios between 19.1 and 19.5. Chemical and isotopic variation along the LSC appeared to be limited with initial ɛNd values varying only by 1.1 ɛ-units (from +4.9 to +6.0) compared, for example, to Hawaiian-Emperor lavas erupted over a similar 80 Ma period (206Pb/204Pb of ∼17.5–18.7 and ɛNd of 0 to +11) [Basu and Faggart, 1996; Keller et al., 2000; Regelous et al., 2003].

Here, we report major and trace element and Sr, Nd and Pb isotope data for 56 new samples from 12 different volcanoes along the older, western end of the LSC between Osbourn Guyot and Seamount 40, located just east of the West Wishbone Scarp at 168°W (Figure 1). The samples were collected in 2002 by dredging during RV Sonne expedition SO 167 [Stoffers et al., 2003]. Our data are combined with data for lavas from the younger portion of the LSC (156 to 168°W), which was dredged during the AMAT02RR expedition of RV Revelle in 2006 (Figure 1). Geochemical results for the latter samples are described in detail by Vanderkluysen et al. (manuscript in preparation, 2011), but selected data for AMAT02RR samples are shown here in order to illustrate the whole along-chain variability and evolution of magmatism on the LSC. Data for the samples previously analyzed by Hawkins et al. [1987] and Cheng et al. [1987] for this section of the LSC are also shown in some diagrams. The combined data sets considered here thus cover 2850 km of the LSC and 50 Myr of its history, although the youngest 1350 km section (from 139°10.0W to 155°9W) of the LSC remains poorly sampled (Figure 1).

3. Petrography and Mineralogy

Most samples are variably vesicular and aphyric or olivine ± clinopyroxene phyric; plagioclase is a rare phenocryst mineral, occurring only in a few samples mostly from the westernmost volcanoes (dredges 134, 135, and 139 on seamounts 32 and 35). Fresh volcanic glass was not recovered. Conglomerates containing rounded pebbles and boulders of volcanic rock were present in many of the dredges, and presumably represent beach deposits, which together with the guyot morphology of many volcanoes indicates that they were once emergent [Lonsdale, 1988].

All samples have undergone some degree of low-temperature alteration and hydrothermal alteration. Olivine phenocrysts are typically completely iddingsitized, and carbonates, zeolites and clay minerals partially or completely fill vesicles. Sample surfaces are often coated with Fe-Mn oxyhydroxide crusts up to several centimeters thick. A detailed description of the mineralogy of the samples analyzed, and an estimate of the degree of alteration based on petrography and geochemistry (see section 5.1), is provided in the auxiliary material (Table S1).

4. Analytical Methods

Altered rinds were removed by sawing. Samples were then coarsely crushed and the least altered fragments handpicked, avoiding secondary minerals as far as possible. The picked chips were washed in an ultrasonic bath with deionised water, and powdered in an agate ball mill. Major element concentrations of sample powders (Tables 1 and S2) were measured on fused glass beads using a Philips 1400 XRF spectrometer at the Institut für Geowissenschaften, Universität Kiel. Precision and accuracy were better than 0.8% and 1% (2σ) for most elements and <0.5% for SiO2. Trace element analyses were carried out using an Agilent 7500cs inductively coupled plasma mass spectrometer at the Institut für Geowissenschaften, Universität Kiel, using methods described by Garbe-Schönberg [1993]. Accuracy was checked using international rock standards (BHVO-1, BCR-2, OU-6, AGV-1), and the external precision for repeated analyses and duplicate digests was better than 3% for all elements (see also standards given in the studies by Beier et al. [2008] and Worthington et al. [2006]).

Table 1. Major Element, Trace Element and Sr-Nd-Pb Isotope Data for Representative Lavas From the Louisville Seamount Chain Collected on Cruise SO-167a
135-02139-01147-01152-01154-02157-01159-06162-03165-01167-01168-03171-01172-03173-01BCR-2 Average ± Standard DeviationOU-6 Average ± Standard DeviationBHVO-1 Average ± Standard DeviationBHVO-1 Average ± Standard DeviationBIR-1 Average ± Standard DeviationAGV-1 Average ± Standard Deviation
  • a

    The 40Ar-39Ar ages from O'Connor et al. (submitted manuscript, 2011) are in bold type; ages marked with a star are inferred ages from the same seamount, age marked with a plus is inferred from a nearby seamount (see also main text). The ɛNd values are calculated assuming a present-day chondritic 143Nd/144Nd value of 0.512638.

VolcanoLouisville 32Louisville 33Louisville 34Louisville 35Louisville 36Louisville 37Louisville ‘x’Louisville 38Louisville 39Louisville 39Louisville 39Louisville 40Louisville 40Louisville 42      
Latitude [°N]26°43.3127°35.1928°36.8031°29.3033°44.2135°27.8236°04.4637°01.5237°32.5637°31.9937°31.9637°38.1837°43.4838°14.90      
Longitude [°W]174°38.20174°01.58173°19.73172°16.43171°25.90170°24.72169°34.87169°46.79169°25.32169°18.49169°18.42169°09.80168°55.72168°08.49      
Waterdepth [m]23452541236322391984216521882221267719501925228323332192      
Distance to MTHN7-1 [km]40663954382234893230301829212846278527782778276127382192      
Age [Ma]68.0*51.0*50.254.357.052.0*49.0+48.0**50.342.0*46.7      
[wt.%]              XRF (n = 8)XRF (n = 9)XRF (n = 20)ICP-MS (n = 11)ICP-MS (n = 3)ICP-MS (n = 10)
SiO247.1245.5842.9846.5541.7146.9535.9937.8137.4241.8536.6242.2826.1653.0354.05 ± 0.0757.07 ± 0.1249.96 ± 0.11   
TiO25.093.532.712.033.943.173.843. ± 0.010.99 ± 0.002.76 ± 0.01   
Al2O318.3314.4013.8414.3616.7716.7013.0413.2213.6915.9510.4214.969.7917.6613.39 ± 0.0320.42 ± 0.0413.59 ± 0.04   
Fe2O311.1613.5211.2611.5412.7511.8813.5313.9016.3113.0515.6713.647.788.8513.66 ± 0.019.03 ± 0.0312.20 ± 0.02   
MnO0. ± 0.000.27 ± 0.000.17 ± 0.00   
MgO1.365.256.406.794. ± 0.022.44 ± 0.037.19 ± 0.07   
CaO4.8710.7812.9210.5510.498.3113.5811.9312.0510.3212.889.7026.104.537.20 ± 0.010.77 ± 0.0011.45 ± 0.02   
Na2O3.833.092.292.872.674.132.421.611.853.921. ± 0.021.94 ± 0.022.38 ± 0.11   
K2O1.701.390.980.501.041.741.851.571.031.380.861.322.042.631.76 ± 0.013.06 ± 0.010.52 ± 0.01   
P2O50.840.670.460.340.991.084.582.022.321.230.910.803.880.480.36 ± 0.000.13 ± 0.000.28 ± 0.01   
Li11.67.3652.725.836.013.217.436.118.424.341.39.9819.49.79   4.57 ± 0.2213.32 ± 0.29410.8 ± 0.487
Sc17.930.729.623.411.39.8328.724.029.29.6927.316.94.703.35   31.8 ± 1.2444.6 ± 1.8512.50 ± 0.323
V23235230420824013529730138420931823761.63.40421 ± 4.34 355 ± 4.68319 ± 6.41321 ± 6.93122 ± 3.52
Cr3.2326.11773910.5292.322184744410.94278148.95.620.97219.5 ± 4.96 280 ± 6.46290 ± 7.71386 ± 4.188.66 ± 2.20
Co12.743.739.044.634.018.339.946.366.532.780.239.714.07.34   44.6 ± 0.68652.5 ± 1.8215.4 ± 0.651
Ni8.6345.713718329.912.472.911514621.034331.339.21.879.80 ± 3.11 111 ± 21.3126 ± 4.35184 ± 4.1416.2 ± 0.481
Cu22.743.875.856.426.66.0181.312712420.768.926.920.88.89   135 ± 9.59126 ± 1.2261.3 ± 1.99
Zn235135179140184155180173231172172137127135137 ± 5.78 105 ± 5.38105 ± 2.4070.7 ± 2.23089.4 ± 3.31
Mo  0.837         1.32    4.39 ± 1.570.088 ± 0.0082.33 ± 0.323
Ga29.625.620.421.327.725.119.219.725.629.121.926.712.228.8   22.0 ± 0.53915.8 ± 0.92021.3 ± 0.853
Rb13.625.229.55.8517.426.130.128.519.533.316.125.735.654.448.4 ± 1.06 8.90 ± 1.599.27 ± 0.2640.257 ± 0.20168.7 ± 2.00
Sr122361241246713431294114733175714944161160968745342 ± 2.07 398 ± 1.93398 ± 7.09110 ± 4.25695 ± 23.9
Y22.134.327.923.036.939.797.427.429.538.025.735.127.640.6   25.6 ± 0.45415.4 ± 0.45419.1 ± 0.449
Zr334294213108358342231219269440254372323572181 ± 1.04 190 ± 1.57173 ± 9.74913.9 ± 0.131240 ± 5.20
Nb51.842.027.022.365. ± 1.39 19.0 ± 1.5717.2 ± 0.3630.548 ± 0.05513.4 ± 0.163
Sn2.922.151.631.172.642.470.7751.492.052.772.072.692.013.63   1.76 ± 0.0850.703 ± 0.0144.41 ± 0.229
Sb1.750.2300.6220.2790.6720.5853.704.114.470.2391.110.0862.640.167   0.112 ± 0.0240.502 ± 0.0104.44 ± 0.307
Cs0.2110.5212.300.3000.6420.6150.6670.6880.6881.880.6860.3850.7531.07   0.108 ± 0.0070.001 ± 0.0071.35 ± 0.045
Ba365239146150449362313168269534104416346707682 ± 18.4 142 ± 25.6135 ± 4.566.16 ± 0.6961231 ± 45.0
La50.734.123.916.351.244.967.739.633.165.449.650.645.271.3   15.4 ± 0.3130.607 ± 0.01038.3 ± 0.618
Ce10176.751.233.311497.981.278.271.615110611793.415377.3 ± 4.50 10.2 ± 4.7738.2 ± 0.8491.98 ± 0.12770.0 ± 1.06
Pr14.   5.60 ± 0.1300.391 ± 0.0098.84 ± 0.124
Nd55.943.930.020.158.756.157.041.341.273.753.   25.3 ± 0.5612.47 ± 0.07432.6 ± 0.412
Sm11.19.486.725.0712.   6.25 ± 0.1471.12 ± 0.0345.92 ± 0.087
Eu3.973.172.351.924.144.173.733.043.034.803.474.022.813.92   2.19 ± 0.0660.568 ± 0.0131.26 ± 0.432
Gd8.948.796.645.3710.911.   6.35 ± 0.1361.85 ± 0.0564.95 ± 0.054
Tb1.221.280.9800.8181.501.581.671.111.161.711.221.461.021.39   0.958 ± 0.0210.368 ± 0.0080.68 ± 0.011
Dy6.057.005.494.667.868.349.975.826.188.556.137.665.367.77   5.47 ± 0.1392.66 ± 0.0423.66 ± 0.044
Ho0.9721.260.9980.8561.341.462.060.9971.071.400.9971.310.9391.46   0.992 ± 0.0230.581 ± 0.0110.68 ± 0.008
Er2.243.202.512.153.213.605.692.362.573.   2.49 ± 0.0611.67 ± 0.0381.80 ± 0.017
Tm0.2770.4280.3320.2890.4070.4680.7900.3000.3310.3930.2720.4010.3160.558   0.335 ± 0.0080.253 ± 0.0060.26 ± 0.003
Yb1.622.652.071.792.442.845.021.801.962.281.592.401.983.63   2.06 ± 0.0541.71 ± 0.0551.68 ± 0.021
Lu0.2080.3730.2870.2490.3330.3880.7770.2450.2690.2990.2070.3220.2820.521   0.282 ± 0.0080.253 ± 0.0040.25 ± 0.003
Hf9.306.694.783.057.737.785.   4.52 ± 0.1560.603 ± 0.0145.14 ± 0.033
Ta3.832.621.641.394.053.381.202.942.405.173.964.063.535.84   1.12 ± 0.0430.041 ± 0.0020.80 ± 0.012
W  0.116         0.479    0.282 ± 0.0640.017 ± 0.0020.74 ± 0.184
Tl0.1100.0130.0450.0240.1630.0330.1110.0490.7340.0530.0860.0530.0760.087   0.024 ± 0.0070.004 ± 0.0040.33 ± 0.007
Pb2.511.991.461.053.383.085.502.082.103.842.703.242.595.16   1.69 ± 0.1382.97 ± 0.07336.3 ± 1.73
Th5.373.282.251.735.484.564.364.773.236.915.955.204.409.10   1.25 ± 0.0270.031 ± 0.0026.51 ± 0.073
U1.600.5810.5130.4270.9400.5692.162.141.931.841.511.641.582.71   0.426 ± 0.0100.010 ± 0.0011.94 ± 0.019

The Sr and Nd isotope measurements were carried out on leached sample powders. Between 200 and 250 mg of sample was leached for 60 min in hot 6M HCl, then placed in an ultrasonic bath for 30 min before the waste acid was removed and the sample was washed thoroughly with deionised H2O before digestion. The Pb isotope measurements were carried out on unleached sample powders, in order to avoid problems associated with variable mobilization of primary, magmatic Pb and uncontrolled fractionation of U/Pb ratios during leaching [Hauff et al., 2000]. Both leached and unleached powders of three samples (116–04, 120–34 and 139–01; two fresh and one moderately altered sample) were analyzed for Sr, Nd and Pb isotope ratios for comparison (see Table S1 and discussion below). Total procedural blanks for Sr, Nd were better than 0.2 and 0.1 ng respectively. Pb blanks were less than 0.25 ng which is approximately 1‰ of the Pb extracted from the most depleted sample, and therefore negligible.

Lead and Sr isotope measurements were carried out on a Finnigan MAT262 thermal ionisation mass spectrometer in static mode at the IFM-GEOMAR, Leibniz Institut für Meereswissenschaften, Kiel, whereas Nd isotope data were acquired using a TRITON instrument at IFM-GEOMAR. Mass bias corrections for Sr and Nd were applied assuming 86Sr/88Sr and 146Nd/144Nd ratios of 0.1194 and 0.7219 respectively. The SRM987 standard gave an average 87Sr/86Sr of 0.710273 ± 7 (n = 13); data in Table 1 are normalized to a value of 0.710238 for direct comparison with the data of Vanderkluysen et al. (manuscript in preparation, 2011). The in-house Nd monitor SPEX yielded a 143Nd/144Nd value of 0.511718 ± 5 (n = 7); the data in Table 1 are normalized to a value of 0.511843 for the La Jolla standard. Instrumental mass fractionation of Pb isotope ratios was monitored using the NBS981 standard, which yielded average values of 16.899 ± 7, 15.437 ± 9 and 36.525 ± 27 for 206Pb/204Pb, 207Pb/204Pb and 208Pb/204Pb, respectively (n = 11). Lead isotope data in Table 1 have been normalized to NBS981 values of 16.941, 15.497, 37.722, so that they can be compared directly with the Pb isotope data for Louisville lavas obtained using a double-spike method by Vanderkluysen et al. (manuscript in preparation, 2011).

The oldest samples analyzed in this study are approximately 79 Ma [Koppers et al., 2004], and the measured isotope ratios therefore require an age correction. We have used the 40Ar-39Ar ages for the same samples (O'Connor et al., submitted manuscript, 2011) for the age correction (see ages in Tables 1 and S2), where these are available (marked in bold type in Table 1). In cases where a particular sample has not been dated, we have used either an average of ages obtained for that seamount (marked with star in Table 1), or an age inferred from data for adjacent seamounts (marked with a plus in Table 1). As we do not have parent-daughter ratios for leached sample powders, we have age-corrected the Sr isotope data using the average Rb/Sr ratio of 0.035 measured for mildly leached powders of the Louisville lavas analyzed by Vanderkluysen et al. (manuscript in preparation, 2011). The age correction for 87Sr/86Sr is small compared to the total range within the data set, even for the oldest samples (less than 0.0001), and this approach is therefore unlikely to lead to significant inaccuracies in the age-corrected ratios. Measured Nd isotope ratios were age-corrected using Sm/Nd ratios of unleached sample powders, since alteration and the type of acid leaching we performed generally results in little change in the Sm/Nd ratio of rock powders, and leached and unleached powders of samples 116–04, 120–34 and 139–01 yielded similar 143Nd/144Nd ratios (see Table S1). For the particular case of the LSC we thus consider this approach to be appropriate.

The Pb isotope data were age-corrected using the measured U, Th and Pb concentrations of unleached powders following the method of Hauff et al. [2000]. This approach will yield accurate initial Pb isotope ratios provided that any alteration-induced fractionation of U/Pb and Th/Pb occurred soon after eruption of the lavas. It is important to note that the measured Pb isotope compositions of the leached powders for samples 116–04, 120–34 and 139–01 are not systematically more or less radiogenic than those of the unleached samples (Tables 1 and S2). After age-correction using the average U/Pb and Th/Pb ratios for leached LSC powders measured by Vanderkluysen et al. (manuscript in preparation, 2011; 0.506 and 1.53 respectively), initial Pb isotope ratios for the leached powders are comparable to initial ratios calculated for the unleached powders of these samples. The differences are smaller than the range in Pb isotope ratios observed in the data set as a whole (maximum difference in 206Pb/204Pb between leached and unleached sample 116–04 is 0.13, while the data set covers a range from 18.85 to 19.34). We therefore consider that the age-corrected isotope ratios in Table 1 are close to those of the samples at the time of their eruption, with the exception of the Pb isotope values for some of the most highly altered samples (see discussion in section 5.1.).

5. Results

5.1. Effects of Alteration

As described in section 3 and in the Appendix, all samples analyzed in this study have undergone some degree of seafloor alteration and some have also experienced subaerial weathering, both of which may have affected chemical and isotopic compositions. The presence of carbonate and hydrous secondary minerals is reflected in the high loss-on-ignition values of many samples (6–14.6 wt.% LOI), together with high CaO, Na2O, and K2O concentrations. Several samples have undergone phosphatisation, leading to P2O5 contents >4% in eight samples. Concentrations of the more mobile trace elements have also been affected by alteration. Fresh oceanic basalts have relatively uniform Ba/Rb, Nb/U and Ce/Pb ratios of 11 ± 2, 45 ± 10 and 24 ± 5 [Hofmann and White, 1983; Hofmann et al., 1986], whereas the SO167 samples display much more variability (see Figures S1 and S2). Ba/Rb, Nb/U and Ce/Pb ratios range from 2.8 to 27, 11 to 93 and 4.6 to 44 respectively, indicating that Rb, Ba, U, Ce and Pb were variably mobile during seafloor alteration and subaerial weathering. Data for these elements are reported in Table 1, but we have not included them in the discussion below. The high field strength elements (HFSE) and rare earth elements (REE) are generally much more resistant to alteration; however, interaction with seawater appears to have affected REE concentrations in several of the samples, which have relatively high Lu and Yb concentrations, lower La/Yb and Hf/Lu ratios and a marked negative Ce anomaly in their rare earth patterns. We have therefore excluded from the following discussion all trace element data for those samples which have Ce/Ce* (Ce/Ce* = CeN/(La)N0.5 × (Pr)N0.5) ratios outside the range 0.9–1.1 (see also Table S2).

Alteration may result in introduction of seawater-derived Sr and Pb and, in rare cases, Nd into samples, although the leaching procedure is expected to have removed secondary carbonate, phosphate, and Fe-Mn oxyhydroxides. Age-corrected (87Sr/86Sr)i and ɛNd values of leached LSC lavas define a negative correlation lying parallel to the mantle array and within the range of previous analyses of leached LSC lavas [Cheng et al., 1987]. The most altered sample, with Ce/Ce* ratios outside the range 0.9–1.1, does not have significantly different Sr or Nd isotope compositions than the least altered samples, which suggests that the leaching procedure has also effectively removed seawater-derived Sr. However, several of the most highly altered samples (lowest Ce/Ce*) have among the highest 207Pb/204Pb ratios. The Pb isotope composition of Pacific seawater is variable, but is typically characterized by high 207Pb/204Pb for a given 206Pb/204Pb compared to Pacific mid ocean ridge basalts (MORB) and LSC lavas [Ling et al., 1997]. The Pb isotope compositions of samples with 207Pb/204Pb > 15.62 are therefore likely to be influenced by the presence of secondary minerals containing seawater-derived Pb, and are not included in the following discussion.

5.2. Major and Trace Element Chemistry

We have divided the sample set into three groups defined on the basis of distance along the chain from the MTHN7–1 dredge site of the youngest sample dated by Koppers et al. [2004] and Watts et al. [1988], in order that any chemical and isotopic variations with longitude and age can be seen more clearly. Group 1 seamounts include Osbourn Guyot at 4148 km from MTHN7–1 to Seamount X at 2921 km, just west of the West Wishbone Scarp (>171°W); Group 2 includes Seamount 38 at 2846 km to Seamount 168.6 at 2694 km (167°–171°W), and Group 3 includes Seamount 167.4 at 2548 km to Seamount 155.9 at 1318 km (<167°W).

Most of the SO167 samples were dredged from fault scarps, in an attempt to sample older lavas from deeper in the stratigraphy, yet no unambiguously tholeiitic lavas were recovered. In a total alkalis - silica diagram (Figure 2a), all samples exhibit alkalic compositions, and immobile trace element compositions of Louisville lavas are more similar to those of Hawaiian alkalic lavas (Figure 2b) than Hawaiian tholeiites. There also is no clear differentiation of the SO167 lavas into distinct, non-overlapping groups in Figure 2b, unlike that in Hawaiian shield, post-shield and post erosional lavas. Hawkins et al. [1987] noted pebbles of tholeiitic basalt in Deep Sea Drilling Project Site 204, which was drilled on the flanks of the LSC, and suggested that Louisville volcanoes may pass through a tholeiitic shield stage. If this is the case, then the lack of tholeiitic samples in the dredge collection suggests the volume of later alkalic lavas must be proportionately greater than in Hawaii, perhaps more similar to Pacific intraplate volcanoes such as those on the Samoan, Marquesas and Cook-Austral chains, which also erupt relatively large volumes of alkalic-transitional basalt [e.g., Jackson et al., 2007a, 2007b; Natland, 1980; Workman et al., 2004].

Figure 2.

(a) Na2O + K2O versus SiO2 plot after Le Maitre [1989]. All SO167 data lie above the alkalic-tholeiitic dividing line of Macdonald and Katsura [1964] for lavas from Hawaii. (b) Zr/Nb versus La/Yb for the LSC lavas and Hawaiian shield, post-shield and post-erosional lavas (note that the LSC lavas display some similarity to the Hawaiian post-shield lavas). Data for Hawaii are from Huang et al. [2005], Shafer et al. [2005], Yang et al. [2003] and Pietruszka and Garcia [1999].

There is no systematic along-chain variation in average major element composition over the chain as a whole when considering the major element composition averaged over a given seamount. Although our samples were collected from different volcanoes of various ages and cannot therefore represent a single liquid line of descent, the major element variations are consistent with a role for crystal fractionation of an assemblage of olivine and clinopyroxene for samples with MgO greater than about 5 wt.%, and plagioclase, clinopyroxene and magnetite at lower MgO values. Several of the most MgO-rich lavas contain cumulus olivine; for example, samples 171–02 and 171–03 with 14.7 wt.% and 16.8 wt.% MgO, respectively, which contain approximately 30% olivine phenocrysts. Therefore, these MgO-rich lavas do not represent primitive liquid compositions but rather represent cumulates (see Table S1).

The incompatible trace element variability of the LSC lavas is relatively small compared to that in many other intraplate volcanic chains such as the Hawaiian chain. La/Yb ratios are 9.1 to 31.1 (Figure 2b), within the range for Hawaiian post-shield alkalic lavas, and higher than for either Hawaiian tholeiites or MORB. Figure 3 displays the combined REE and HFSE patterns of the LSC lavas. They show convex upward patterns, which are commonly observed in ocean island basalts. The HFSE and REE of the samples discussed below do not show evidence for alteration (Figure 3).

Figure 3.

Primitive mantle-normalized trace element patterns (REE and HFSE only). PRIMA after Lyubetskaya and Korenaga [2007]. Note that sample 141–07 is shown as an example for those samples that have been excluded as a result of alteration (see text and auxiliary material).

5.3. Sr, Nd and Pb Isotope Compositions

Age-corrected Sr and Nd isotope compositions of the Louisville lavas analyzed in this study and by Vanderkluysen et al. (manuscript in preparation, 2011) range from 0.702848 to 0.704626 87Sr/86Sr, and +3.8 to +6.2 ɛNd, and extend the known compositional range of LSC lavas. However, the majority of samples have initial ɛNd values in a relatively narrow range between +4.5 and +6.2 (Figure 4), within the range of Hawaiian post-shield lavas. Louisville Seamount Chain lavas have higher 87Sr/86Sr, lower ɛNd, and more radiogenic Pb isotope compositions than lavas from the Hollister Ridge and the adjacent Pacific-Antarctic Rise (Figure 4). Lavas from volcanoes east of the West Wishbone Scarp generally have slightly higher 87Sr/86Sr than lavas from the older volcanoes to the west (Figure 5d). Age-corrected 206Pb/204Pb ratios for most samples are between 18.8 and 19.4, and many of the more radiogenic samples are also from east of the West Wishbone Scarp (Figures 4 and 5d).

Figure 4.

Age-corrected (a) 87Sr/86Sr, (b) ɛNd, (c) 207Pb/204Pb, and (d) 208Pb/204Pb versus 206Pb/204Pb of LSC lavas. Louisville literature data are from Cheng et al. [1987], Osbourn Trough data from Worthington et al. [2006] and data for the Pacific-Antarctic Ridge and Hollister Seamount from Castillo et al. [1998]. Note that age correction was only applied to LSC data presented here. Grey arrow shows maximum influence of age correction between measured and age corrected values.

Figure 5.

Variation of (a) age, (b) La/Yb, (c) ɛNd, and (d) 207Pb/204Pb with distance from sample MTHN7–1. Samples labeled with sample numbers are post-shield and ‘very-late’ rejuvenated stage samples dated by O'Connor et al. (submitted manuscript, 2011) (see also Figure 6). Maximum errors on La/Yb ratios are indicated with error bar at lower left. Curve in (a) shows variation in age of the underlying seafloor from Müller et al. [2008]. Bend in LSC is assumed to be located at 167° W, close to the West Wishbone Scarp. Literature data for Louisville lavas in Figures 5a and 5b are from Cheng et al. [1987] and Hawkins et al. [1987]. Data from Cheng et al. [1987] are not shown in Figures 5c and 5d because they are not age- corrected.

6. Discussion

6.1. Geochemical Variation Within Individual Louisville Seamounts

The chemical and isotopic variation among lavas from individual seamounts, and the differences between lavas from different seamounts are summarized in Figures 58. The compositional range within an individual seamount is similar to that observed along the chain as a whole, with a few exceptions (Osbourn Guyot lavas extend to lower La/Yb and Nb/Zr; Seamounts 38 and 39 lavas extend to higher Nb/Zr; Figure 5). However, compared to Hawaiian volcanoes, individual Louisville volcanoes display less variation (Figure 2b) and they appear to not pass through compositionally distinct shield, post-shield and post-erosional magmatic stages that characterize most Hawaiian volcanoes. Nevertheless, new 40Ar-39Ar ages for SO167 samples (O'Connor et al., submitted manuscript, 2011) show that magmatism at several Louisville volcanoes took place over a protracted period of time, with temporal chemical variations among the lavas erupted on a particular seamount. They divided the dated samples into ‘age-progressive’, ‘classic post-shield’ or ‘very-late’ rejuvenated types, according to whether the ages of the samples are within 5 Myr of the age of the oldest lavas for a given seamount along the entire chain (either based on measured ages or inferred from the age-progression of the oldest volcanism along the chain as a whole), between 5 and 10 Myr younger, or between 10 and 23 Myr younger than the oldest lavas, respectively. Although 23 Myr is much longer than the duration of volcanism reported for any other intraplate volcano on the fast-moving Pacific Plate (typically <5 Myr, occasionally up to 10 Myr [Hirano et al., 2002]), Lonsdale [1988] documented morphological evidence from several of the Louisville seamounts for relatively late, ‘rejuvenated stage’ magmatism. In particular, volcanic cones or small guyots built upon the eroded summits of several of the main volcanoes show that significant volumes of magma were erupted after the main ‘shield-building’ stage had ended and the volcano summit had been eroded and subsided several hundred meters below sea level. The time elapsed for such erosion and subsidence, formation of the guyot and eruption of post-erosional volcanic deposits has been calculated to be on the order of 20–30 Myr for the Magellan seamount trail [Koppers et al., 1998], implying that the Louisville seamounts may have a similar age range. On the Cook-Austral and Samoan seamount chains, protracted volcanism and a complicated age progression have been suggested to result from the presence of more than one hot spot [Chauvel et al., 1997; Jackson et al., 2010]. However, this model cannot easily explain either the location or age of Louisville non-age-progressive magmatism, which apparently occurred only during the period 70–30 Ma and is restricted to seamounts west of the West Wishbone Scarp (Figure 5a).

Figure 6.

Variation of (a) Nb/Zr, (b) La/Yb and (c) ɛNd with age difference from estimated age progression (ΔTime). Lavas within 5 Myr of the age progression line (Figure 5a) are age-progressive, lavas erupted within 5–10 Ma are ‘post-shield’, and lavas erupted >10 Myr after age progression are ‘very late’ stage lavas. Ages for SO167 data are from O'Connor et al. (submitted manuscript, 2011). Ages for AMAT02RR data are from Koppers et al. (submitted manuscript, 2011). Literature data for Louisville lavas are from Cheng et al. [1987] and Hawkins et al. [1987]. Data from Cheng et al. [1987] are not shown in Figure 6c because they are not age- corrected.

Figure 7.

Variation in Th/Nb and Th/Zr with age and distance along the LSC from sample MTHN7–1 [Hawkins et al., 1987], located at 50°26.4′S, 139°10.0′W. Note relatively depleted compositions of some lavas from Osbourn Guyot (most distant site from MTHN7–1; lower Nb/Zr, La/Yb (Figure 4), Th/Zr), and more variable compositions of lavas from seamounts between 2600 and 3000 km from MTHN7–1 (Seamounts, 38, 39).

Figure 8.

Variation in cumulative lava volumes [Lonsdale, 1988], lithosphere thickness at the time of seamount formation, and initial 87Sr/86Sr and Nb/Zr with distance from MTHN7–1. The more variable Louisville lavas were erupted at a time of relatively low volcanic output, on relatively thick oceanic lithosphere. Thickness of oceanic lithosphere at time of seamount formation is calculated from (age lithosphere - age seamount)0.5 × 11 [Humphreys and Niu, 2009]. See text for discussion.

O'Connor et al. (submitted manuscript, 2011) also noted that among the dated samples from SO167 those belonging to the ‘very-late’ stage tend to have lower incompatible trace element concentrations, and lower Nb/Zr and La/Yb, than ‘age-progressive’ samples (Figure 6). For example, of the SO167 samples that have been dated, the three with the lowest Nb/Zr (0.100–0.125) are ‘very-late’ lavas from Osbourn Guyot. Two samples from Osbourn Guyot belonging to the ‘age-progressive’ group (dated by Koppers et al. [2004]) have higher Nb/Zr (0.132 to 0.147 [Hawkins et al., 1987]). We have isotope data for eight samples that have been dated by O'Connor et al. (submitted manuscript, 2011); based on these data, there is no systematic difference in Sr, Nd or Pb isotope composition between ‘age-progressive’ and ‘very-late’ lavas (Figure 6). The apparent evolution of individual Louisville volcanoes toward less enriched incompatible element compositions with time contrasts with most Hawaiian volcanoes, where the post-shield and post-erosional lavas are more enriched in incompatible trace elements and have higher La/Yb and Nb/Zr than the earlier, shield-stage lavas. The large age range of 23 Myr for lavas from Seamount 32 (O'Connor et al., submitted manuscript, 2011) is too great for the later volcanism to result from decompression due to lithospheric uplift above a flexural bulge, as has been proposed to explain Hawaiian post-erosional volcanism [Watts et al., 1985; Wessel, 1993]. At 45 Ma, which is the age of the youngest lavas recovered from Seamount 32, volcanism had already started at Seamount 38 about 1250 km to the southeast. The differences in trace element composition and timing therefore suggest that LSC ‘very-late’ volcanism has a different origin from that of Hawaiian post-erosional magmatism. O'Connor et al. (submitted manuscript, 2011) speculated that ‘very-late’ magmatism on the oldest Louisville seamounts might have been a consequence of the relatively young age of the underlying seafloor, which was formed at the Osbourn Trough in the Late Cretaceous. Subsidence, thermal contraction and cracking of this initially relatively warm, thin lithosphere over a period of several tens of millions of years after the oldest Louisville volcanoes were constructed may have allowed ‘very late’ magmas to reach the surface.

The comparatively depleted compositions of Louisville ‘very-late’ lavas could result either from replacement of Louisville hot spot mantle by Pacific MORB-source mantle during the >9 Myr long period between ‘age-progressive’ and ‘very-late’ magmatism, or from re-melting of hot spot mantle that was previously depleted during the earlier, main stage of magmatism [e.g., Jackson et al., 2010]. In the former case, data for the ‘very-late’ lavas would be expected to lie on mixing lines or curves between data for ‘age-progressive’ Louisville lavas and Pacific MORB. However, although the lower Nb/Zr and La/Yb ratios of ‘very-late’ lavas could be explained by mixing of Louisville mantle and more depleted MORB mantle, the isotope compositions of ‘age-progressive’ and ‘very-late’ lavas overlap (Figure 6). Thus, we favor re-melting of slightly heterogeneous hot spot mantle that was depleted during the earlier, age progressive stage of magmatism as the origin of the relatively depleted trace element compositions of ‘very-late’ lavas.

6.2. Geochemical Variations Along the Louisville Seamount Chain

In agreement with previous studies, we find that lava compositions display little variation along most of the LSC, particularly along the younger part of the chain east of East Wishbone Scarp (Figures 5 and 7). Nevertheless, some significant chemical and isotopic variations are observed: (1) Lavas erupted on seamounts in the vicinity of the West and East Wishbone Scarps tend to have more variable compositions, (2) lavas from Osbourn Guyot extend to less enriched compositions (lower ratios of highly to moderately incompatible elements, e.g., Nb/Zr, La/Yb and Th/Zr) than lavas from younger seamounts (Figures 5 and 7) and, (3) lavas from seamounts between the West and East Wishbone Scarp have higher 87Sr/86Sr than older lavas between the West Wishbone Scarp and Osbourn Guyot, although there are no other obvious systematic geochemical variations with distance along the LSC. This difference between lavas from east and west of the West Wishbone Scarp cannot be due to the influence of the Osbourn Trough because most of these seamounts were formed too far from the Osbourn Trough, and long after it ceased active spreading, as discussed elsewhere in the manuscript. The variations that are observed could result from changes with time in the composition of asthenosphere passively sampled by a propagating crack in the lithosphere, or in the composition of mantle material arising from greater depths via a mantle plume, or these may instead be linked to differences in the age and structure of the oceanic lithosphere underlying the seamount chain. These possibilities are discussed in more detail in section 6.2.4 below.

6.2.1. No Influence of the Osbourn Fossil Spreading Center? Evidence From Osbourn Guyot

Osbourn Guyot was constructed at about 79 Ma on seafloor that was formed at the Osbourn Trough fossil spreading center during the Late Cretaceous [Billen and Stock, 2000; Downey et al., 2007; Lonsdale, 1986; Worthington et al., 2006]. Much of the seafloor in this region was formed during the long normal magnetic polarity Chron C34 (121–83 Ma), and its exact age is therefore poorly known. Estimates for the time at which spreading at the Osbourn Trough ceased range from 71 Ma to 115 Ma [Billen and Stock, 2000; Downey et al., 2007; Mortimer et al., 2006; Taylor, 2006]. Most recently, Downey et al. [2007] have argued that the minimum age for the end of spreading at the Osbourn Trough is 87 Ma to 93 Ma. The seafloor beneath Osbourn Guyot could therefore have been only 8 Myr old at the time this volcano formed.

The age and thickness of the oceanic lithosphere are thought to have an important influence on the compositions of intraplate lavas [Haase, 1996; Humphreys and Niu, 2009; Ito and Mahoney, 2005; Regelous et al., 2003]. Intraplate volcanoes built on young, thin oceanic lithosphere close to spreading centers appear to erupt lavas with relatively depleted compositions. For example, the oldest Emperor seamounts were constructed on thin oceanic lithosphere close to a spreading center, and their comparatively depleted incompatible element compositions have been ascribed to either ‘plume-ridge interaction’ [Keller et al., 2000] or a ‘lithospheric lid’ effect [Frey et al., 2005; Huang et al., 2005; Regelous et al., 2003; Shafer et al., 2005]. Osbourn Guyot was built closest to the fossil Osbourn Trough spreading center, and by analogy with the Emperor seamounts, an effect on the composition of lavas of the oldest Louisville seamounts could be expected.

As discussed above, lavas from Osbourn Guyot, the seamount closest to the Osbourn Trough, extend to less enriched compositions than any other LSC lavas. For example, Nb/Zr, La/Yb and Th/Zr ratios reach lower values than do other LSC lavas (Figures 5 and 7). Although 40Ar-39Ar ages are not available for all of the Osbourn Guyot lavas we have analyzed, it is apparently only the ‘very-late’ lavas that have these less enriched compositions (116–04, 120–34; Figure 6). The older, ‘age-progressive’ lavas from Osbourn Guyot have trace element compositions within the range of other Louisville lavas [Hawkins et al., 1987; Koppers et al., 2004], and the ɛNd of the two ‘age progressive’ lavas from Osbourn Guyot analyzed by Cheng et al. [1987] lie within the range of other LSC lavas (Figure 6). Lavas from Seamount 32 also have compositions within the range for other, younger LSC lavas despite being erupted onto crust only 50 km southeast of the Osbourn Trough. Because the ‘very-late’ Osbourn Guyot lavas were erupted as much as 23 Myr after the earliest ‘age-progressive’ lavas, and because the latter do not have compositions significantly different from those of other LSC lavas, it appears that the presence of the Osbourn Trough did not significantly influence the chemical composition of Louisville seamount lavas. The absence of an effect could perhaps be explained if ridge-plume interaction [Keller et al., 2000] requires active spreading, if the Osbourn spreading axis was extinct at the time Osbourn Guyot formed, whereas the Pacific-Izanagi spreading axis remained active at the time of emplacement of the oldest Emperor seamounts [Steinberger and Gaina, 2007]. Alternatively, as we propose below, active spreading at the Osbourn Trough ceased long enough before the formation of Osbourn Guyot that the underlying oceanic lithosphere had essentially reached its ‘mature’ thickness of at least 40 km [Stein and Stein, 1992].

6.2.2. No Influence of the Osbourn Fossil Spreading Center? Evidence From Osbourn Trough

The oldest lavas recovered from Osbourn Guyot are 79 Ma; any older Louisville seamounts that may have existed have since been subducted at the Tonga Trench. Geochemical studies of the Tonga Arc have shown that the Pb isotope signature of lavas from the LSC can be identified in lavas from the northernmost Tongan islands of Tafahi and Niuatoputapu [Regelous et al., 2010, and references therein]. The LSC therefore appears to have extended at least 1500 km farther to the northwest of its current western limit, but seamounts westward of Osbourn Guyot have been subducted beneath the Indo-Australian Plate. From the general age-progression along the LSC, active Louisville volcanoes are predicted to have been situated very close to the Osbourn Trough at 85–95 Ma. If the youngest lavas on this fossil spreading center were erupted in this time period, they would be expected to have relatively enriched incompatible element compositions and relatively low ɛNd and high 87Sr/86Sr as a result of mixing with Louisville-type mantle. As discussed above, the exact time at which the Osbourn spreading center became extinct is not known precisely, but 87–93 Ma is thought to represent a minimum age for the youngest lavas [Downey et al., 2007]. Worthington et al. [2006] analyzed a suite of samples dredged from the axial rift of the Osbourn Trough, and found that both trace element and isotope compositions are those of normal MORB, with no evidence for a contribution from Louisville-type mantle. There is no systematic difference in composition from west to east along the Osbourn Trough that could result from more enriched, Louisville-type mantle flowing into the sub-Osbourn Trough melting zone from the west. Instead, Nb/Zr ratios of the Osbourn Trough samples are <0.04 [Worthington et al., 2006], giving no evidence for mixing with enriched mantle sources.

This result suggests that the Osbourn Trough samples analyzed by Worthington et al. [2006] may be significantly older than the younger limit of 87–93 Ma proposed by Downey et al. [2007]. If instead spreading at Osbourn Trough ceased at 115 Ma [Mortimer et al., 2006], then by the time the Osbourn Guyot formed, the local oceanic crust would have had an age of approximately 35 Myr. At 115 Ma, the Louisville hot spot would have been located approximately 1750 km to the northwest of the Osbourn Trough (assuming a volcanic propagation rate of 5 cm/y along the Louisville Seamount Chain). This distance is greater than the predicted maximum distance of hot spot-ridge interaction (about 1250 km [Mittelstaedt and Ito, 2005]), and so the youngest lavas erupted at the Osbourn Trough would not be expected to show geochemical evidence for interaction with Louisville-type mantle. This scenario would explain both the lack of a ‘thin lithosphere effect’ for the oldest Louisville seamounts as well as the normal MORB compositions of the youngest Osbourn Trough lavas. On the Emperor Seamount Chain, the effect of variable lithospheric thickness is observed only in lavas from seamounts built on crust younger than about 30 Myr [Regelous et al., 2003; Shafer et al., 2005].

6.2.3. Influence of Fracture Zones on Louisville Magma Compositions?

Basu and Faggart [1996] noted that lavas from the Hawaiian islands of Lanai and Oahu, which are constructed above the Molokai Fracture Zone, tend to have more variable isotopic compositions and lower ɛNd than lavas from elsewhere along the Hawaiian Seamount Chain. They speculated that the presence of the fracture zone influences the shape of the mantle melting region, and hence the composition of the lavas. The younger portion of the LSC, between 148°W and 155°W, crosses the Tharp and Heezen Fracture Zones. In addition, at longitude 163°W and 167°W, the LSC crosses the East and West Wishbone Scarps, respectively. These are two major bathymetric features that separate older oceanic crust formed at the Osbourn Trough to the west from younger crust created at eastern Pacific spreading centers to the east. Lavas from the LSC therefore offer an opportunity to study the influence of lithospheric age and structure on intraplate mantle melting.

The Tharp Fracture Zone intersects the LSC between the 155.9°W and 154.3°W seamounts, and separates oceanic crust with an age difference of about 8 Myr [Lonsdale, 1988]. Seamounts 157.7 and 155.9, studied by Vanderkluysen et al. (manuscript in preparation, 2011), were constructed within about 70 km of the Tharp FZ. Lavas from these two seamounts have trace element compositions within the range of other LSC lavas (Figure 5). The age offset across the Tharp Fracture Zone is similar to that across the Molokai Fracture Zone, but there is apparently no effect of the former on the composition of Louisville lavas. The oceanic crust west of the West Wishbone Scarp was formed at the Osbourn Trough spreading center, whereas crust formed east of the East Wishbone Scarp was formed at the Pacific-Antarctic Rise [Billen and Stock, 2000; Watts et al., 1988]. The age offset across the East Wishbone Scarp is predicted to be more than 50 Myr from the Müller et al. [2008] seafloor age data set (Figure 5a). However, there is apparently no effect on the compositions of the lavas from the two seamounts formed closest to the East Wishbone Scarp (Figures 5 and 7).

Louisville volcanoes between 167°W and 169°W are built upon relatively shallow seafloor associated with the West Wishbone Scarp (Figure 1). On the basis of zircon ages of dacite lavas with subduction-related trace element signatures dredged from the ocean crust exposed on the scarp south of its intersection with the LSC at 168°W, Mortimer et al. [2006] suggested that a short period of plate convergence across the West Wishbone Scarp, and accompanying arc magmatism, took place at around 115 Ma. If this were the case, then contamination of the upper mantle by fluids and melts released from the subducting plate might be expected. This might explain the more variable lava compositions on the Louisville seamounts just west of the West Wishbone Scarp, which were erupted at 40–50 Ma (Koppers et al., submitted manuscript, 2011; O'Connor et al., submitted manuscript, 2011). However, we find no evidence for elevated arc-like Th/Nb in Louisville lavas from Group 2, all of which have Th/Nb ratios <0.105, within the typical range for intraplate oceanic basalts and MORB, and similar to other Louisville lavas (Figure 7). In contrast, most subduction-related lavas are characterized by much higher Th/Nb ratios (>0.30 [e.g., see Kelemen et al., 2003, and references therein]).

Group 2 lavas in the central region of the LSC, between the East and West Wishbone Scarps, are generally more variable in composition than Group 1 or Group 3 lavas. For example, Nb/Zr and La/Yb ratios of lavas from Seamounts 38 and 39 cover almost the entire range observed in LSC lavas (Figure 5). It is possible that magmas erupted in the vicinity of large fracture zones may be more susceptible to contamination by seawater-altered oceanic crust [Lassiter et al., 2002], but there is no evidence that any of the Louisville lavas have undergone coupled assimilation of altered oceanic crust and fractional crystallization, and these process are in any case unlikely to influence Nb/Zr or La/Yb ratios significantly because the Nb/Zr and La/Yb ratios of the assimilant and the magma would be similar. Instead, we suggest that variations in the age and thickness of the underlying lithosphere are responsible for the more variable compositions of lavas from volcanoes in this section of the LSC, as discussed in detail below.

6.2.4. Mantle Melting Effects, and Implications for the Origin of Intraplate Magmatism

The central part of the LSC (between 2600 and 3100 km from MTHN7–1, in the vicinity of the West Wishbone Scarp), where lava compositions are apparently most variable, corresponds to a segment of the chain characterized by a relatively low volcanic output rate, as inferred from the volumes of individual volcanoes and their spacing [Lonsdale, 1988, Figure 8]. If Louisville magmatism results from the passage of the Pacific Plate over a fixed thermal or chemical mantle anomaly with constant melt productivity, then the volcanic flux per km of seamount chain will be inversely proportional to the rate of relative motion between the lithosphere and the underlying mantle. The rate of migration of volcanism along the central and eastern parts of the LSC has not varied greatly [Watts et al., 1988; Koppers et al., submitted manuscript, 2011; O'Connor et al., submitted manuscript, 2011] (see Figure 5a), which implies that the volcanic output rate is likely to have been significantly lower during the period that this part of the chain formed.

A decrease in melt supply rate could result from a decrease in mantle temperature or mantle upwelling rate (plume model), or less fertile underlying mantle, or a lower degree or slower rate of lithospheric extension (‘plate’ model), or from variations in the thickness of the lithosphere (the ‘lithosphere-lid’ effect). As noted above, there is no significant change in the rate of migration of volcanism along the LSC during this time period, which would appear to rule out changes in lithosphere extension rate as the origin of the lower melt supply rate inferred for this part of the LSC.

Volcanoes from the central section of the LSC, which are characterized by the most variable lava compositions and lowest eruption rates, are also underlain by the oldest oceanic crust (70–90 Myr old at the time the seamounts were formed (Figure 8) [Worthington et al., 2006]). Smaller degrees of melting of heterogeneous mantle beneath older, thicker oceanic lithosphere would tend to result in more variable and more enriched lava compositions (e.g., higher and more variable La/Yb, Nb/Zr and Th/Zr ratios, Figures 5, 7, and 9). Our results show that lithosphere structure and thickness may have an important influence on the compositions of intraplate lavas, an effect which must be taken into account when using such lavas to infer the composition and origin of mantle sources.

Figure 9.

Lithosphere thickness at the time of seamount magmatism and La/Yb and Nb/Zr ratios of the LSC lavas. Thickness of oceanic lithosphere at time of seamount formation is calculated from (age lithosphere - age seamount)0.5 × 11 [Humphreys and Niu, 2009]. Lavas erupted on seamounts situated on older lithosphere tend to have higher and more variable Nb/Zr and La/Yb. See text for discussion.

7. Summary and Conclusions

All dredged LSC lavas are alkalic and our sampling therefore suggests that unlike Hawaiian volcanoes, LSC volcanoes either did not evolve through an early tholeiitic stage, or that older tholeiitic lavas have been completely covered over by younger alkalic lavas. Volcanism on several Louisville volcanoes apparently occurred over protracted intervals of up to 23 Myr, and the later lavas tend to be less enriched in the highly incompatible elements than earlier lavas from the same volcano but display homogenous Sr-Nd-Pb isotope ratios. This chemical evolution, which is unlike that observed on Hawaiian and many other intraplate volcanoes, likely results from re-melting of isotopically similar mantle that was previously depleted during melting of Louisville mantle (resulting in the observed trace element signatures).

Chemical and isotopic variations were limited during the 55 Myr period represented by the 2850-km-long length of the LSC studied here. There is no evidence that the presence of the Osbourn Trough fossil spreading center influenced the composition of LSC magmas, and we suggest that this is because spreading at the Osbourn Trough ceased as much as 35 Myr before the nearest (oldest remaining) Louisville seamounts were constructed. Neither the East and West Wishbone Scarps, which separate oceanic lithosphere of very different age, nor the major Tharp Fracture Zone, obviously influenced the compositions of lavas from adjacent Louisville volcanoes. However, lavas from the central section of the LSC, between about 167 and 171°W, tend to have more variable trace element and isotope compositions and extend to higher Nb/Zr, La/Yb values. This part of the LSC is characterized by the lowest eruption rates, and is also underlain by the oldest oceanic crust (70–90 Ma old at the time of the seamounts were constructed). We therefore suggest that smaller degrees of melting of heterogeneous mantle beneath older, thicker oceanic lithosphere are responsible for the more variable and more enriched lava compositions and the lower eruption rates along this section of the LSC.


We thank John O'Connor for providing data from his submitted manuscript. We are grateful to the crew, captain and scientific party of RV Sonne Cruise 167 for sample collection. We thank Tim Worthington, Peter Appel and Folkmar Hauff for their help during major element, trace element and isotope analyses. We gratefully acknowledge the very detailed and constructive reviews by Matt Jackson, Ivan Vlastelic and the Associate Editor Anthony Koppers. ChB also acknowledges Anthony Koppers, Jörg Geldmacher and the Scientific Party of IODP Leg 330 for an insightful and entertaining cruise to the Louisville Seamount Chain, and thanks MacAllan for inspiration. ChB also thanks C. Mardin for still being able to read this manuscript.