Geochemistry, Geophysics, Geosystems

Magmatic processes under mid-ocean ridges: A detailed mineralogic study of lavas from East Pacific Rise 9°30′N, 10°30′N, and 11°20′N

Authors

  • Yucheng Pan,

    1. Institute of Oceanology, Chinese Academy of Sciences, Qingdao, Shandong Province, China
    2. Department of Geology and Geophysics, University of Hawaii, 1680 East West Road, Honolulu, Hawaii, 96822, USA
    Search for more papers by this author
  • Rodey Batiza

    1. Marine Geology and Geophysics, National Science Foundation, Arlington, Virginia, USA
    2. Department of Geology and Geophysics, University of Hawaii, 1680 East West Road, Honolulu, Hawaii, 96822, USA
    Search for more papers by this author

Abstract

[1] Detailed petrologic study has been made for lavas from the northern East Pacific Rise (EPR) 9°30′N, 11°20′N, and 10°30′N. 9°30′N and 11°20′N have robust magma supply and a shallow melt lens while 10°30′N has nonrobust magma supply and no shallow melt lens. Lavas from all three localities are sparsely phyric and glassy, containing plagioclase ± olivine ± pyroxene. Typically, the lavas contain several to many (up to seven) distinct chemical groups of plagioclase that are not always distinct texturally. The lavas may also contain up to three chemically distinct groups of olivine and two groups of pyroxene. The lavas contain both individual crystals and groups comprising reticulate and dendritic clots that we interpret to represent bits of crystal networks forming in mushy zones of an axial magma chamber. 21 of the 23 samples studied in detail have diverse crystal compositions and require mixing, which most likely occurs when mostly liquid magma passes through mushy zones. We find no significant systematic differences between robust and nonrobust segments in terms of their crystal content, proportion of texturally distinct xenocrysts, crystal size, aspect ratio, roundness, modal abundance, magma residence time, number of diverse mineral-chemical groups, and characteristics of mixing like the total range of composition of disequilibrium minerals present, and the magnitude of the chemical gaps between disequilibrium and equilibrium compositions. This unexpected and remarkable similarity suggests that the presence or absence of a seismically imaged shallow melt lens has essentially no effect on the mineralogy of erupted lavas. We conclude that the fundamentally important magmatic process under mid-ocean ridges, from slow to fast, is the formation of crystal networks and their subsequent compaction; the seismically detected shallow melt lenses likely contain highly evolved magma, formed by expelled interstitial melt during crystal network compaction, and play very little role in crustal accretion. Our conclusion implies that the thick (>3 km) low velocity zone near the base of the crust produces the thin (<30 m) shallow melt lenses, not the other way around.

1. Introduction

[2] Magmatic processes at mid-ocean ridges (MORs) are chiefly responsible for the accretion of the earth's oceanic crust. The crust so accreted usually has three layers: extruded basalts at the top, sheeted dikes below, and gabbros at the base. Much understanding about the magmatic processes of crustal genesis has been obtained by 1) extensive sampling of the extruded basalts and more limited sampling of the intruded dikes and gabbros at MORs [e.g., Klein and Langmuir, 1987; Grove et al., 1992; Niu and Batiza, 1993; Natland and Dick, 1996], 2) by study of all three layers from ophiolites which are thought to be preserved ocean crust on land [e.g., Coleman, 1977; Gass et al., 1984; Nicolas, 1989; Peters et al., 1990; Parson et al., 1992; MacLeod and Yaouancq, 2000], and 3) by geophysical experiments at MORs [e.g., Detrick et al., 1987; Hooft et al., 1997; Crawford et al., 1999; Carbotte et al., 2000]. Nevertheless, a number of important questions about MOR processes are yet to be resolved.

[3] The concept of magma chambers under MORs has undergone great changes during the last few decades, from extremely large liquid magma chambers extending from the Moho all the way to the dikes [e.g., Cann, 1974], to composite chambers consisting mainly of crystal mush, commonly with a shallow melt lens at fast spreading ridges [Sinton and Detrick, 1992; Barth et al., 1994]. In addition to the shallow melt lens, a deep magma lens at Moho depth has been detected [Garmany, 1989; Dunn and Toomey, 1997; Crawford et al., 1999]. Crustal accretion under MORs is thought to be accomplished by either the shallow melt lens [e.g., Sleep, 1975; Nicolas et al., 1988; Quick and Denlinger, 1993; Phipps Morgan and Chen, 1993; Henstock et al., 1993; Chen, 2001], or by both the shallow and deep magma lenses [e.g., Boudier et al., 1996; Schouten and Denham, 1995], or largely by the repeated intrusion of magma sills throughout the gabbro layer [e.g., Kelemen et al., 1997; Korenaga and Kelemen, 1997, 1998; Kelemen and Aharonov, 1998; Koga et al., 2001].

[4] The shallow melt lens has received extensive attention because it has been widely detected by geophysical experiments and is thought to play a central role in crustal accretion especially at fast spreading ridges [e.g., Sinton and Detrick, 1992]. Recent detailed seismic studies showed that this magma lens has both solid roof and floor [Hussenoeder et al., 1996; Singh et al., 1999] and abrupt lateral and vertical discontinuities [Carbotte et al., 2000]. The properties of the magma lens, such as its depth, width, and crystal content, show little or only poor correlation with supposed indicators of magma supply such as axial depth and cross-sectional area [Kent et al., 1994; Hussenoeder et al., 1996; Hooft et al., 1997; Carbotte et al., 2000]. These observations seem to be more consistent with the interpretation that the shallow melt lens is either (1) segregation melt [Natland and Dick, 1996; Philpotts et al., 1996; Hussenoeder et al., 1996] or (2) newly injected magma body [Hussenoeder et al., 1996; Hooft et al., 1997; Carbotte et al., 2000]. To place additional constraints on the role of the shallow melt lens in crustal accretion and the nature of magmatic processes under mid-ocean ridges, we have made detailed mineralogical and petrological studies on mid-ocean ridge basalts (MORBs) from East Pacific Rise (EPR) 9°30′N, 10°30′N, and 11°20′N. Our studies do not reveal significant systematic differences between segments with or without a shallow melt lens, suggesting that the presence or absence of the shallow melt lens has little effect on the mineralogy of erupted lavas. Combining our study on mid-ocean ridge lavas with prior studies of cooling lava lakes, thick continental lava flows, ophiolites, and oceanic gabbros, we conclude that the most important magmatic process under mid-ocean ridges is the formation of crystal networks and their subsequent compaction; the shallow melt lenses seismically detected along mid-ocean ridges contain highly evolved magma, formed by expelled interstitial melt during the compaction, and play very little role in crustal accretion.

2. Petrologic and Mineralogic Studies of Lavas From East Pacific Rise at 9°30′N, 10°30′N, and 11°20′N

2.1. Chemical and Modal Abundances

[5] Samples were collected by dredging and rock coring along flow lines (±1 km) of the EPR axis at 9°30′N, 10°30′N, and 11°20′N on both the Cocos and Pacific plates out to 40–50 km from the axis [Batiza et al., 1996]. Sample locations and glass electron microprobe analyses are presented in Table 1. Table 2 gives trace element data and sources of isotope data. 9°30′N, 10°30′N, and 11°20′N have statistically different MgO contents: 10°30′N has the lowest MgO of 5.6 ± 1.4 wt% (one standard deviation, used consistently throughout the paper), 11°20′N has the highest (7.0 ± 0.7 wt%), and 9°30′N is intermediate (6.6 ± 0.6 wt%). By “statistically” different we mean, here and throughout the paper, that a Student t-test of the two populations show differences that are significant at the 95% confidence level.

Table 1. Representative Electron Microprobe Analyses of Glasses From East Pacific Rise at 9°30′N, 10°30′N, and 11°20′Na
SampleLat.Lon.Begin DepthEnd DepthSiO2TiO2Al2O3FeOMnOMgOCaONa2OK2OP2O5SumK2O/TiO2Mg#
  • a

    Major element abundances were determined by electron microprobe at LDEO, Y. Niu analyst. Lat., beginning latitude of the dredge, Lon., beginning longitude, Mg# = Mg/(Mg + 0.9 Fe).

9°30′N
PH2-39.518104.1882888280550.222.4014.1012.190.205.7210.613.380.430.3099.560.1848.19
PH4-19.520104.1723018287950.491.6115.1210.370.187.4211.492.640.120.1599.580.0758.61
PH6-19.524104.1432980281650.841.7014.2811.010.197.0211.532.750.120.1599.590.0755.83
PH6-29.524104.1432980281650.731.7114.3610.850.187.0711.662.720.130.1699.580.0856.37
PH6-39.524104.1432980281650.691.6914.2011.040.207.1011.632.760.130.1599.590.0856.02
PH7-19.526104.1292949273750.541.7914.5910.770.176.9511.642.740.220.1699.580.1256.12
PH7-29.526104.1292949273750.361.8214.6510.780.177.0311.642.750.210.1699.570.1256.37
PH7-39.526104.1292949273750.201.8214.4910.880.186.9911.752.760.220.1999.480.1256.00
PH7-49.526104.1292949273750.431.8114.5710.650.187.0811.722.730.210.1799.530.1256.84
PH7-69.526104.1292949273750.501.8114.6910.740.187.0211.622.660.210.1799.600.1256.41
PH7-89.526104.1292949273750.171.7814.6010.700.207.0611.802.720.220.1899.440.1256.66
PH8-19.526104.1182963297250.861.7314.3310.950.206.9511.552.770.110.1599.590.0655.67
PH8-29.526104.1182963297250.381.7914.3611.030.207.0411.682.810.120.1699.550.0755.82
PH8-39.526104.1182963297250.451.7314.3411.090.226.9911.692.810.100.1699.580.0655.51
PH14-19.533104.0262880299750.621.3014.999.180.177.7112.632.750.090.1099.540.0762.45
PH19-19.541103.8793014297050.412.9312.7914.990.265.329.542.850.170.2899.550.0641.25
PH19-39.541103.8793014297050.942.8812.6414.470.235.309.473.070.180.2899.460.0642.06
PH19-59.541103.8793014297050.612.9412.7614.840.265.369.572.770.180.2899.570.0641.69
PH19-69.541103.8793014297050.622.6413.1513.710.255.749.923.080.140.2699.500.0545.34
PH19-79.541103.8793014297050.762.5913.0913.660.235.809.863.080.150.2699.480.0645.70
PH19-89.541103.8793014297050.832.6213.1113.700.235.719.803.040.140.2899.450.0545.23
PH19-89.541103.8793014297050.822.6113.1013.670.245.709.863.060.150.2799.480.0645.23
PH19-99.541103.8793014297050.792.6613.1113.640.235.729.863.060.150.2699.500.0645.38
PH20-19.545104.0052953297350.692.8712.6914.750.255.249.503.050.180.3099.510.0641.32
PH20-29.545104.0052953297350.802.0913.8412.270.246.3810.692.960.120.1899.570.0650.76
PH23-19.508104.3022627279850.891.5414.7510.090.207.1012.062.660.100.1399.530.0658.22
PH23-29.508104.3022627279850.851.6014.6810.140.187.1912.042.650.110.1499.580.0758.39
PH24-19.509104.3112729281351.191.7014.7510.210.216.6111.442.900.320.2199.540.1956.18
PH24-29.509104.3112729281351.011.7214.5910.510.236.4811.582.910.310.2199.550.1855.00
PH24-39.509104.3112729281351.141.6914.6810.330.196.5711.532.930.310.1999.560.1855.76
PH26-29.506104.3322984282651.121.8214.0911.190.216.6511.452.790.120.1599.580.0754.07
PH30-19.496104.3812782297450.962.1313.7312.170.226.0911.012.920.140.1899.550.0749.79
PH30-29.496104.3812782297451.012.1613.6612.210.246.0610.952.900.130.2099.520.0649.57
PH30-39.496104.3812782297450.982.2113.6712.100.236.1510.992.870.150.2099.550.0750.17
PH30-49.496104.3812782297450.992.1413.7012.050.246.1211.062.900.140.1999.530.0750.16
PH31-19.498104.4002891307050.761.9314.3511.170.216.6311.482.810.120.1799.630.0654.02
PH31-29.498104.4002891307050.991.9414.0511.320.216.5011.472.780.130.1799.570.0753.22
PH31-39.498104.4002891307051.181.8814.0011.140.216.5911.422.810.120.1799.530.0653.96
PH31-49.498104.4002891307051.171.9414.1411.190.206.6011.252.770.130.1599.550.0753.88
PH31-59.498104.4002891307051.181.9214.0711.010.216.6511.412.810.130.1599.540.0754.45
PH31-69.498104.4002891307050.921.9214.0911.250.216.6911.402.810.120.1699.560.0654.07
PH31-79.498104.4002891307051.031.9214.1511.180.216.6211.342.810.120.1899.550.0653.98
PH31-89.498104.4002891307051.031.9314.0911.370.216.5011.342.820.120.1799.570.0653.10
PH31-99.498104.4002891307051.021.9214.2311.140.216.6011.392.810.130.1799.620.0753.98
PH32-19.493104.4072929295851.371.5314.5010.130.207.0012.002.600.100.1499.580.0757.79
PH32-29.493104.4072929295851.311.5914.5510.200.217.0011.912.610.100.1399.600.0657.62
PH32-39.493104.4072929295851.261.7714.2210.800.196.8011.462.780.120.1799.560.0755.49
PH32-49.493104.4072929295851.041.7814.2810.990.216.8211.412.760.120.1699.590.0755.13
PH32-59.493104.4072929295851.281.7514.2910.630.226.8011.562.770.120.1599.570.0755.91
PH32-89.493104.4072929295851.281.9114.0411.170.226.5211.362.790.140.1699.590.0753.62
PH33-39.489104.4292920292951.111.6614.4810.540.196.9011.712.770.130.1499.620.0856.46
PH33-59.489104.4292920292951.051.6614.4210.450.207.0511.712.800.120.1599.610.0757.20
PH35-29.487104.4632971302051.202.4613.2313.320.255.4510.093.110.200.2499.560.0844.78
PH36-19.487104.4822787303551.321.5414.4810.320.217.0511.912.590.100.1199.630.0657.50
PH36-29.487104.4822787303551.371.5514.4110.260.196.9811.972.640.100.1299.590.0657.38
PH36-39.487104.4822787303551.261.5914.4210.350.196.9412.002.640.120.1099.610.0857.05
PH36-49.487104.4822787303551.321.5514.5710.330.196.9911.852.650.100.1199.670.0657.27
PH36-59.487104.4822787303551.431.6014.5610.230.206.8611.962.610.100.1299.660.0657.02
PH36-69.487104.4822787303551.341.5414.5210.150.217.0611.932.650.100.1199.600.0657.92
PH36-79.487104.4822787303551.321.5614.4910.180.197.0511.932.670.100.1299.620.0657.84
PH38-29.478104.5153117295550.952.1814.0611.580.216.3411.032.900.160.1999.590.0752.03
 
10°30′N
PH41-310.549103.2263215313656.262.4913.0612.840.272.916.773.950.620.3499.510.2531.01
PH43-110.515103.3213166316657.461.9912.3713.930.281.845.904.240.790.6899.470.4020.73
PH44-210.526103.3433249305651.412.2413.4412.930.255.7310.053.130.180.2199.570.0846.76
PH45-110.536103.3593257308051.603.6311.7917.010.323.277.613.410.380.5399.540.1027.56
PH45-210.536103.3593257308051.703.5711.8116.810.313.257.633.460.410.5699.520.1127.70
PH46-110.520103.3743050308154.212.7312.0715.830.332.546.773.690.570.7399.490.2124.12
PH47-210.518103.3813292309152.052.4213.1613.690.254.899.173.320.310.3099.560.1341.43
PH49-410.518103.4213211321652.092.6313.3113.420.235.039.443.040.290.4299.910.1142.58
PH51-310.516103.4643200310850.832.5313.4613.320.275.7110.372.980.180.2599.910.0745.94
PH51-510.516103.4643200310851.642.4413.4212.860.255.6610.192.910.190.1999.740.0846.58
PH51-810.516103.4643200310850.852.5713.4313.440.265.7410.262.960.200.2099.910.0845.81
PH52-310.502103.4873060304551.181.5114.969.730.197.3511.992.500.130.1599.650.0959.96
PH54-310.498103.5133029310950.951.7714.4610.840.216.9911.562.690.120.1399.730.0756.09
PH54-410.498103.5133029310950.681.7714.5910.880.217.0611.662.650.130.1399.760.0756.23
PH55-110.487103.5323254308550.942.3313.8012.440.206.3610.732.750.160.2099.910.0750.32
PH55-310.487103.5323254308550.822.3413.9812.450.236.3110.772.710.160.1999.960.0750.09
PH56-210.488103.5483247310950.802.2814.2311.940.226.6711.092.270.160.2199.890.0752.54
PH59-110.488103.5842893285450.912.6013.7312.570.226.0410.642.940.220.20100.060.0848.76
PH59-210.488103.5842893285451.042.6113.7612.560.236.0110.502.900.200.24100.060.0848.66
PH60-110.483103.5982828283350.273.2012.8015.420.314.959.103.320.260.3099.910.0838.87
PH60-210.483103.5982828283350.273.2112.9115.330.254.939.123.290.260.2899.870.0838.93
PH62-210.480103.6132832279251.272.3313.7612.210.235.9210.623.050.200.2099.800.0949.00
PH62-610.480103.6132832279251.122.2313.9112.030.216.1610.633.080.190.2399.790.0950.35
PH63-210.476103.6192959270550.682.5913.3313.750.235.6310.053.250.200.2199.930.0844.78
PH63-310.476103.6192959270550.682.7213.2813.730.215.599.983.270.210.2499.910.0844.66
PH65-110.454103.6372973301950.792.5113.6013.570.245.5310.113.150.190.2599.940.0844.67
PH65-210.454103.6372973301952.372.4713.2013.500.245.008.973.440.330.3499.860.1342.32
PH66-110.475103.6483064308550.262.9713.5113.970.225.929.732.980.190.2499.990.0645.64
PH66-410.475103.6483064308550.212.9313.4213.980.245.949.813.000.210.2799.990.0745.69
PH67-110.450103.6593123310250.272.7513.4614.050.245.959.752.990.200.3099.960.0745.61
PH68-110.462103.6762947294750.432.0714.6511.100.206.8711.402.710.190.2599.870.0955.09
PH6910.472103.6983068294650.401.8714.6111.550.226.6611.402.850.160.1499.870.0953.33
PH70-210.472103.6983068294650.542.2013.7512.910.256.0910.852.900.150.2599.910.0748.32
PH72-1310.446103.7503151300551.611.6214.5010.120.186.9911.652.700.180.1999.740.1157.79
PH76-110.431103.8263323313250.350.9814.8510.070.208.2413.141.810.010.0799.730.0161.84
PH76-310.431103.8263323313253.462.6312.9615.730.312.497.043.850.640.7599.860.2423.85
PH77-110.433103.8473051305050.442.6013.5613.430.245.9010.342.860.200.2199.780.0846.53
PH77-410.433103.8473051305051.162.5213.5012.950.255.8810.163.020.190.2199.840.0847.37
PH78-210.427103.8633142297452.723.2012.4214.740.263.907.993.680.380.4899.760.1234.37
PH78-310.427103.8633142297451.983.0512.6814.800.244.048.473.790.390.4799.920.1335.11
PH80-310.409103.9813168294350.692.6113.0214.270.275.389.783.300.210.2999.820.0842.77
PH81-210.401104.0113140301751.292.1114.3310.410.136.7911.332.930.160.1999.690.0856.35
PH81-310.401104.0113140301750.632.0014.0711.800.216.4711.292.980.170.2099.830.0952.04
PH81-710.401104.0113140301750.431.9214.3711.290.216.8211.552.850.170.1899.790.0954.48
PH81-910.401104.0113140301750.612.0714.4211.890.226.2910.933.010.160.1499.750.0851.16
 
11°20′N
PH82-111.427103.3852946295850.991.6215.249.330.187.2711.882.870.200.2099.770.1260.69
PH83-311.423103.4302996303150.181.6415.609.510.207.4911.712.790.230.1899.530.1460.95
PH84-411.412103.5293051295250.411.8515.2110.200.177.2211.343.030.230.1899.840.1258.39
PH84-511.412103.5293051295251.251.8814.839.990.196.7511.402.980.280.1899.740.1557.24
PH86-111.395103.5633025297450.461.8215.459.890.187.2611.672.780.270.1899.960.1559.26
PH86-311.395103.5633025297450.631.8315.399.900.177.2111.612.770.260.1899.940.1459.07
PH87-311.389103.5823036293450.121.8115.1510.220.187.5311.902.690.150.1899.920.0859.34
PH88-111.384103.5973008295751.341.8514.5710.140.197.1111.462.730.170.1999.770.0958.13
PH88-611.384103.5973008295750.361.8914.4111.100.206.9211.672.950.150.2099.860.0855.26
PH89-311.389103.6122979294149.951.6816.029.420.177.4511.742.910.280.2199.830.1761.03
PH90-211.374103.6393003293350.612.1514.9710.540.196.1710.953.440.540.3399.880.2553.70
PH91-111.380103.6563101302450.161.4715.799.510.167.5812.192.760.130.1399.870.0961.21
PH92-311.382103.6782900290350.161.9416.079.430.197.0311.133.210.520.2299.900.2759.61
PH92-411.382103.6782900290349.991.9616.089.440.197.0811.183.230.510.2599.900.2659.75
PH93-111.378103.6902960290850.331.8215.1710.310.196.9711.732.910.200.2099.830.1157.27
PH93-411.378103.6902960290850.541.9014.9910.020.207.0611.782.970.200.1999.840.1158.25
PH93-611.378103.6902960290850.381.5815.089.810.217.5112.092.830.210.1699.860.1360.24
PH93-711.378103.6902960290850.631.8215.2810.220.206.9111.532.920.200.1699.890.1157.25
PH94-111.369103.7092863286450.341.6215.269.890.207.4211.982.730.210.1599.790.1359.76
PH95-311.366103.7192883288250.081.7915.179.870.217.5712.012.920.170.1499.930.0960.31
PH95-711.366103.7192883288250.821.8514.9610.320.216.6611.373.120.290.2399.830.1656.12
PH96-111.365103.7342826277150.221.7215.1210.090.187.0111.912.970.170.1999.580.1057.89
PH97-111.362103.7422848273350.211.9814.7811.040.206.7111.443.090.180.1999.820.0954.63
PH98-111.361103.7522910272049.881.6115.789.840.187.5612.022.730.150.1599.890.0960.33
PH99-111.359103.7602830289050.412.1214.7011.030.196.7211.313.020.210.2299.940.1054.67
PH99-211.359103.7602830289049.982.0114.8711.510.206.7011.362.990.200.19100.020.1053.56
PH99-411.359103.7602830289049.982.0714.9111.490.226.6811.342.950.200.19100.040.1053.54
PH100-111.357103.7632623262850.682.0414.3511.070.226.5011.373.150.230.1999.810.1153.78
PH100-211.357103.7632623262850.682.0614.4211.110.226.6211.372.950.250.2299.910.1254.13
PH101-211.357103.7712505254950.392.1514.6911.060.236.5611.393.020.220.1899.890.1054.01
PH102-611.358103.7782625254950.171.9414.9010.980.226.8311.472.990.230.2099.930.1255.17
PH103-211.368103.7802570254952.402.4113.7812.480.214.788.973.860.570.3599.810.2443.15
PH103-311.368103.7802570254950.351.8914.9710.630.197.1211.442.910.180.1799.850.1057.02
PH104-111.344103.7772364254550.361.8815.0610.510.207.1211.502.930.230.1999.990.1257.29
PH104-411.344103.7772364254552.212.4714.2012.160.224.789.343.610.570.4299.980.2343.77
PH105-SG11.354103.7882652263849.682.2116.369.650.197.2110.323.320.750.2899.970.3459.67
PH106-111.378103.7852534252850.131.9015.1310.290.197.3011.562.940.190.1599.780.1058.44
PH108-111.342103.7942724263849.902.2215.9310.160.176.4610.923.190.600.2799.810.2755.73
PH109-111.352103.7992796277849.621.6116.309.330.187.5711.822.970.200.1499.740.1261.64
PH109-211.352103.7992796277849.221.7016.479.540.187.5111.942.960.190.1799.880.1160.93
PH111-311.349103.8222891281150.841.6615.2610.160.166.9311.302.890.330.2199.740.2057.46
PH112-111.346103.8322929287650.241.6715.479.840.197.3311.692.950.270.1899.810.1659.58
PH112-211.346103.8322929287650.111.6515.579.840.187.3611.732.920.270.1999.820.1659.72
PH113-111.340103.8492942292150.901.6114.8110.390.186.8611.812.920.180.1699.840.1156.67
PH113-211.340103.8492942292150.711.6414.8710.310.206.9211.902.880.190.1999.830.1257.07
PH114-511.343103.8702951293249.701.6915.9110.070.167.4411.642.940.130.1899.860.0859.39
PH115-211.334103.8883014302150.681.7614.8110.440.197.0311.732.910.160.1499.850.0957.14
PH115-611.334103.8883014302150.711.7114.7510.650.227.0511.552.920.160.1499.860.0956.75
PH117-311.325103.9222978309450.921.4815.109.510.207.4512.142.630.120.1299.680.0860.82
PH117-311.325103.9222978309450.373.0514.9112.700.254.408.933.861.040.50100.010.3440.67
PH119-111.317103.9723006298650.651.8414.9510.340.216.6611.683.070.190.1799.780.1056.06
PH119-711.317103.9723006298650.691.7414.7010.730.206.9011.413.100.190.1799.830.1156.03
PH120-111.309103.9893064300651.061.7314.5610.080.207.2911.892.710.120.1699.800.0758.89
PH120-211.309103.9893064300650.441.7215.709.320.197.3311.712.890.270.2299.780.1660.89
PH120-711.309103.9893064300649.951.8716.508.960.187.1511.342.810.490.2599.510.2661.25
PH121-211.314104.0082993298250.381.7015.479.750.177.4811.452.890.260.2299.770.1560.29
PH122-111.311104.0283057296349.761.8816.299.330.177.4311.043.070.640.3199.940.3461.19
PH122-411.311104.0283057296349.431.3016.188.930.188.7412.302.530.090.1399.800.0765.95
PH125-411.264104.1742398237749.071.2917.178.530.178.7811.822.670.120.1199.720.0967.08
PH125-511.264104.1742398237749.101.3017.108.610.148.5511.972.660.120.1399.670.0966.29
PH125-611.264104.1742398237748.951.2517.058.500.159.2111.592.780.130.1199.730.1068.21
Table 2. Trace Element Data and Sources of Isotope Dataa
SampleScCrCoNiSrLaCeNdSmEuTbYbLuHfTaThULa/Sm NIsotopic data and ICPMS-UQ
  • a

    The trace element data are all in ppm and were collected by Instrumental Neutron Activation Analysis (INAA) at Washington University [Korotev, 1991, 1996]. The samples in this table are those selected for detailed mineral chemical and CSD study. In some cases, trace element data for these particular samples was unavailable, but was available for a sample from the same dredge with very similar major element chemistry (Tables 1 and 2). In such cases, data for the latter samples appears, and the sample number of the sample analyzed is in parentheses in the first column. La/Sm is normalized to chondritic values [Haskin et al., 1968]. Isotopic data for the 9°30′N is not available, however ICPMS trace element data was collected at the University of Queensland, Australia, either for the sample listed in column 1 or a sample from the same dredge with similar major element chemistry, for which the sample number is given. For both the 10°30′N and 11°20′N areas, both isotopic data and ICPMS trace element data have been published in Regelous et al. [1999] and Niu et al. [1999] respectively. Samples from the 9°30′N and 10°30′N areas are NMORB, whereas samples from the 11°20′N area are both NMORB and EMORB.

9°30′N Area
PH6-24518844801404.0212.7143.841.4213.690.552.980.20.110.080.57yes
PH14-14239341801102.618.982.771.10.692.510.372.120.110.060.020.52yes
PH19-9 (19-7)434545401206.5119.6226.62.181.546.120.925.50.240.190.120.54 
PH23-2                  23-1
PH30-4 (30-1)449445 904.8714.7134.731.671.164.530.653.810.220.150.030.56 
PH32-84416845901204.1612.1123.941.460.993.90.583.130.210.090.20.58 
PH35-2 (35-3)424744501305.7816.3115.121.781.254.730.74.270.310.190.20.6235-1
 
10°30′N Area
PH51-3 (51-5)44314760805.9517155.251.831.254.890.744.090.320.230.040.6251-5
PH52-3                  52-2
PH54-4 (54-1)44166441001103.6411.1103.891.440.983.770.593.10.170.11 0.5154-3
PH55-1 (55-3)431814470705.3616.1145.131.81.244.970.744.220.30.210.20.5755-3
PH60-2                  yes
PH69-3                  69-2
PH76-14328448100500.672.7 1.910.810.572.60.391.360.03  0.19 
PH78-3 (78-2)371737 16013.1138.43410.993.252.519.451.389.220.690.650.30.65 
 
11°20′N Area
PH90-242143383025010.2824.6135.141.791.043.670.534.150.810.790.351.14yes
PH91-14033643 1503.3710.1133.251.270.792.950.432.460.190.120.020.57yes
PH92-3 (92-1)38248391202009.1822.3194.411.620.893.150.453.510.70.630.321.192-1
PH93-4 (93-6)4327041501805.214.8124.151.50.963.660.533.30.270.240.20.6993-6
PH94-142323421101204.8113.3143.481.290.843.060.462.690.340.250.20.76yes
PH99-1                  99-2
PH100-14412742601405.6715.8124.551.641.0940.63.610.340.30.110.68yes
PH108-137219409024010.4925.6164.881.680.973.250.493.660.840.730.361.18yes

[6] Chemical variation diagrams and liquid line of descent (LLD) modeling results using MELTS [Ghiorso and Sack, 1995] are shown in Figures 1, 2, and 3 for glasses from EPR 9°30′N, 10°30′N, and 11°20′N respectively. Regressions for N-MORBs with MgO > 5 wt% show that 9°30′N and 10°30′N are very similar for most elements while 11°20′N has higher K2O, Na2O and P2O5 contents at similar MgO; higher Al2O3, FeO, TiO2 at high MgO and lower FeO and TiO2 at low MgO; lower and higher CaO at high and low MgO respectively. These differences are readily explained in terms of the occurrence of larger numbers of EMORB samples at 11°20′N. E-MORBs (higher K2O/TiO2 (≥0.14), K2O, Na2O, P2O5, and Al2O3, and lower CaO and FeO at similar MgO compared to N-MORB) comprise 7%, 0.0%, and 36% of the lavas from 9°30′N, 10°30′N, and 11°20′N respectively.

Figure 1.

Liquid lines of descent of glasses from EPR 9°30′N. Black circles represent N-MORBs, blue crosses E-MORBs (K2O/TiO2 > 0.14), and red solid squares the selected samples, which are Ph14-1, Ph23-2, Ph6-2, Ph32-8, Ph30-4, and Ph35-2 respectively from high to low MgO. The solid curves with green squares are the best fit LLDs we found after many trials using MELTS (0.1 kbar, 0.4% H2O, QFM-1 buffer, and fractionate solid). The parental magma composition is estimated from regression.

Figure 2.

Liquid lines of descent of glasses from EPR 10°30′N. Black circles represent N-MORBs and red solid squares the selected samples, which are Ph76-1, Ph52-3, Ph54-4, Ph69-3, Ph55-1, Ph51-3, Ph60-2, and Ph78-3 respectively from high to low MgO. The solid curves with green squares are the best fit we found after many trials using MELTS (1 kbar, 0.2% H2O, Fe-FeO buffer, and fractionate solid). The parental magma composition is estimated from regression.

Figure 3.

Liquid lines of descent of glasses from EPR 11°20′N. Black circles represent N-MORBs, blue crosses E-MORBs, and red solid squares the selected samples, which are Ph122-4, Ph91-1, Ph94-1, Ph93-4, Ph99-1, and Ph100-1 respectively from high to low MgO for N-MORBs and Ph92-3, Ph108-1, and Ph90-2 from high to low MgO for E-MORBs (shown more clearly in panels for Al2O3, FeO, and K2O). The solid curves with green squares are the best fit we found after many trials using MELTS (0.1 kbar, 0.4% H2O, QFM-1 buffer, and fractionate solid). The parental magma composition is estimated from regression.

[7] Modal abundances in more than 150 samples were measured by point counting (Table 3). 11°20′N has statistically lower total crystal content (1.4 ± 1.5 vol%) than both 9°30′N (2.6 ± 2.1 vol%) and 10°30′N (2.7 ± 2.8 vol%), which are not statistically different (Figure 4). The abundances of xenocrysts, crystals with obvious disequilibrium texture, are 0.57 ± 0.69 vol%, 0.96 ± 1.2 vol%, and 1.6 ± 1.2 vol% for 9°30′N, 10°30′N, and 11°20′N respectively. 9°30′N has statistically lower xenocryst content than 11°20′N.

Figure 4.

Crystal contents (plagioclase + olivine + pyroxene) versus MgO contents for MORBs from EPR 9°30′N, 10°30′N, and 11°20′N. The red squares are N-MORB samples and green diamonds are E-MORB samples selected for detailed crystal assemblage study. For sample number, see captions of Figure 1, Figure 2, and Figure 3. The right panel plots are for all lavas from EPR 9°30′N, 10°30′N, and 11°20′N. (a) Total crystal (including xenocryst) versus MgO. (b) Xenocryst, defined as having obvious disequilibrium texture, versus MgO. (c) Total crystal minus xenocryst, regarded as total phenocryst, versus MgO. The maximum or upper limit of total phenocryst content decreases as MgO content decreases. This trend of the upper limit is robust in experiments of randomly drawing smaller and uniform sample size to remove sampling bias at lower MgO.

Table 3. Point-Counted Modal Abundances for MORBs From EPR 9°30′N, 10°30′N, and 11°20′Na
SampleVes., %Ol, %Pl, %Cpx, %Xtall, %Total pts
  • a

    Ves.%, volume percent vesicles; crystal abundances are calculated vesicle-free: Ol, olivine, Pl, plagioclase, Cpx, clinopyroxene, Xtall, total crystals; Total pts, total points counted. The point counting excludes microlites and includes xenocrysts.

9°30′N
PH2-31.01.10.31.41000
PH4-11.40.90.80.21.91000
PH6-15.01.00.21.3910
PH6-22.30.30.31000
PH6-31.40.60.41.01593
PH7-12.00.74.85.51000
PH7-21.70.43.84.21000
PH7-31.40.12.50.12.71527
PH7-42.20.12.62.71256
PH7-66.02.20.32.61580
PH7-81.40.21.51.61320
PH8-13.00.50.51153
PH8-20.00.50.5936
PH8-31.10.80.81385
PH14-12.82.26.30.99.41209
PH19-15.10.30.31362
PH19-34.51.21.21000
PH19-52.20.80.81527
PH19-66.81.70.92.61271
PH19-75.00.40.10.51214
PH19-86.02.10.42.51217
PH19-91.02.60.63.32411
PH20-11.21.30.11.41313
PH20-21.80.70.71261
PH23-11.10.21.60.01.82864
PH23-26.21.54.15.5790
PH24-15.00.21.80.52.51562
PH24-23.71.01.32.21500
PH24-33.60.11.90.72.71575
PH26-23.80.21.30.31.81475
PH30-18.50.40.10.51123
PH30-211.20.30.31323
PH30-35.00.40.42268
PH30-410.00.30.10.4878
PH31-11.80.12.51.03.62896
PH31-20.40.82.61.44.81699
PH31-32.80.24.51.36.01778
PH31-41.40.91.00.82.71821
PH31-54.40.23.10.74.01632
PH31-65.10.21.40.11.71000
PH31-72.40.41.11.51446
PH31-81.71.42.23.61500
PH31-92.41.11.02.11851
PH32-15.00.23.63.81231
PH32-25.00.02.32.42360
PH32-34.00.60.61282
PH32-41.00.80.81141
PH32-54.52.02.02071
PH32-83.51.46.61.89.81318
PH33-32.81.00.40.82.22374
PH33-52.00.73.74.42035
PH35-22.70.93.71.76.31366
PH36-14.60.21.10.61.81389
PH36-24.10.43.70.34.31476
PH36-31.21.01.01500
PH36-44.30.83.91.15.82746
PH36-52.90.74.31.66.62523
PH36-62.20.51.60.22.41381
PH36-72.20.63.00.23.81589
PH38-21.80.00.80.61.42697
 
10°30′N
PH41-31.31.01.01000
PH43-10.61.90.12.01000
PH44-20.50.40.41000
PH45-11.10.01000
PH45-21.70.10.11000
PH46-10.50.10.20.31000
PH47-20.70.40.41000
PH49-41.10.01000
PH51-30.20.61.82.34.71000
PH51-50.81.13.81.76.71000
PH51-80.11.35.42.08.71000
PH52-30.91.11.40.12.61500
PH54-30.70.20.21000
PH54-43.80.21.71.9500
PH55-12.80.61.70.22.51500
PH55-31.20.80.60.41.81000
PH56-20.80.20.21000
PH59-10.31.36.00.27.51000
PH59-20.21.43.90.35.61000
PH60-11.20.55.00.76.21000
PH60-22.41.43.71.26.31500
PH62-20.30.52.72.35.51000
PH62-60.30.23.32.15.61000
PH63-20.60.10.70.91.71000
PH63-30.80.41.20.21.81000
PH65-11.10.54.42.06.91000
PH65-20.50.86.92.210.01000
PH66-11.10.30.31000
PH66-40.70.10.11000
PH67-10.40.90.11.0800
PH68-10.00.40.4982
PH69-30.91.64.31.37.31000
PH70-21.20.10.11000
PH72-131.80.24.10.14.41000
PH76-10.01.94.26.12000
PH76-30.70.01000
PH77-10.70.30.31000
PH77-41.11.00.31.31000
PH78-20.50.20.21000
PH78-30.00.40.20.61000
PH80-31.00.40.40.8984
PH81-20.60.20.40.61000
PH81-30.20.11.61.22.91000
PH81-70.20.40.60.21.21000
PH81-90.20.41.10.21.71000
 
11°20′N
PH82-11.20.41.31.71000
PH83-30.50.85.46.21000
PH84-41.30.60.61000
PH84-52.30.20.21000
PH86-11.10.30.81.11000
PH86-30.31.42.23.61000
PH87-32.20.60.61000
PH88-10.20.20.21000
PH88-60.61.41.41000
PH89-30.50.82.02.81000
PH90-21.20.20.21000
PH91-11.10.71.01.71500
PH92-30.00.10.20.31000
PH92-40.70.20.90.11.21000
PH93-10.20.20.70.91000
PH93-40.20.51.41.91000
PH93-60.40.70.91.61000
PH93-70.70.60.81.41000
PH94-12.20.43.10.13.61500
PH95-30.70.01000
PH95-70.80.01000
PH96-11.60.51.01.51000
PH97-10.60.11.00.21.31000
PH98-10.20.63.64.21000
PH99-10.40.20.50.71000
PH99-22.40.10.80.11.01000
PH99-41.90.11.11.21000
PH100-11.70.10.40.30.81000
PH100-20.70.11.10.31.51000
PH101-20.50.41.21.61000
PH102-61.20.30.31000
PH103-21.10.30.20.51000
PH103-30.90.01000
PH104-10.50.80.81000
PH104-40.40.10.60.41.11000
PH105-SG1.20.40.10.5870
PH106-11.40.01000
PH108-11.90.40.81.21500
PH109-11.50.10.11000
PH109-21.30.20.20.41000
PH111-30.90.61.82.41000
PH112-10.60.21.51.71000
PH112-20.10.20.70.91000
PH113-13.40.25.86.01000
PH113-23.10.16.46.81000
PH114-50.60.90.91000
PH115-21.40.10.50.61000
PH115-61.00.70.71000
PH117-30.43.23.21000
PH117-30.43.23.21000
PH119-10.00.40.41000
PH119-70.30.20.81.01000
PH120-10.61.41.41000
PH120-20.20.10.11000
PH120-70.10.60.10.71000
PH121-22.90.01000
PH122-10.50.41.52.01000
PH122-41.20.70.71000

[8] Figure 5 shows that crystals of EPR lavas from 9°30′N, 10°30′N, and 11°20′N are equimodal, namely their phase proportions are similar to those seen in low-pressure crystallization experiments. Other EPR segments and Atlantic mid-ocean ridges also have equimodal crystal assemblages [Michael and Chase, 1987; Bryan, 1983], suggesting that widespread and significant fractionation of one solid phase from another is not significant [Bryan, 1983].

Figure 5.

Comparison of phase proportions between 1 bar experiments (circles) and lavas from EPR 9°30′N, 10°30′N, and 11°20′N (triangles). Only samples with more than 1 vol.% total phenocryst were plotted. The larger scatters seen in EPR lavas are probably due to low total phenocryst content (average 3.1 vol.%) compared to experiments (average 33 vol.%). EPR lavas are equimodal.

[9] On the basis of their chemical (Figures 1 to 3) and petrographic (Figure 4) representativeness, 23 samples were selected for detailed mineralogic study.

2.2. Crystal Assemblages

[10] Because our samples are generally sparsely phyric, it was necessary to make 5–10 thin sections for each of the 23 samples in order to have enough crystals for mineralogic study. Probe analyses, from a few hundreds to a few thousands, were collected for each mineral phase present in each of the 23 samples selected. We analyzed every crystal (large enough to be analyzed by probe) in a thin section or in an area of a thin section due to considerations of representation, crystal scarcity, and efficient use of probe time. For most crystals a cross section from crystal center to rim was analyzed, with denser or extra probe points near the rim as volume is proportional to cubic power of radius or linear dimension. Table 4, Table 5, and Table 6 show representative analyses for plagioclase, pyroxene, and olivine respectively. Note the fairly wide range of compositions present within the data set, consistent with the wide range of MgO among the samples (Table 1). Also note the wide range of compositions present even within a single sample, evidence for mixing, as discussed below.

Table 4. Representative Electron Microprobe Analyses of Plagioclasea
SamplePointSiO2Al2O3FeOCaONa2OTotalAn
  • a

    We used the Cameca SX-50 microprobe at the University of Hawaii. The probe was run with a beam current of 10 nA, a voltage of 15 kv, a beam size of ∼1 μm and count times of 30 seconds s. Na, Al, and Si are analyzed first to minimize evaporation. For olivine and pyroxene, the beam current is 20 nA.

9°30′N
6-2T1151.530.30.813.63.7100.066.6
6-2T28-449.532.00.514.92.899.773.3
6-2T56-548.532.60.615.92.4100.077.9
14-1N25-354.329.00.711.84.7100.457.2
14-1N7-1351.830.90.314.03.6100.667.8
14-1N19-148.133.50.416.42.1100.680.3
23-2P6-454.328.70.711.54.8100.055.5
23-2k15-349.831.90.415.13.0100.273.6
23-2K1-548.133.20.516.12.4100.478.9
23-2k44-347.234.20.417.01.7100.683.4
30-4B50-752.728.90.912.24.499.160.0
30-4B2950.931.20.613.83.399.867.6
30-4B76-348.532.30.715.12.599.274.7
32-8L38-352.529.30.711.94.598.958.4
32-8L5-851.230.50.613.73.699.667.1
32-8L10-947.832.40.615.72.498.877.9
35-2P27-855.228.30.610.85.2100.252.1
35-2P19-353.729.20.712.14.5100.258.5
35-2P37-252.030.60.613.33.8100.364.7
35-2P5-250.931.50.614.53.2100.770.4
 
10°30′N
51-3D4159.325.01.57.57.0100.436.1
51-3D36-155.727.80.710.55.3100.050.9
51-3D62-250.631.10.714.23.499.969.3
52-3C42-752.529.90.713.04.0100.163.0
52-3C8-750.331.30.514.53.099.671.3
52-3C22-247.433.30.416.91.999.983.5
54-4U21-1357.425.61.48.96.399.643.0
54-4U60-450.431.50.514.53.2100.270.8
54-4U21-146.134.20.417.71.499.887.6
55-1E2-156.526.41.29.46.099.545.8
55-1E40-1452.329.70.712.94.099.663.1
55-1E35-347.433.20.616.22.199.680.2
55-1E43-3745.634.50.517.91.299.788.6
60-2E12-356.425.92.39.35.999.945.2
60-2E25-154.728.50.911.15.2100.453.7
60-2E2-452.629.40.712.24.699.559.9
76-1D10-150.531.40.714.52.9100.170.8
76-1D4-149.431.80.515.22.799.674.8
76-1D23-148.232.70.516.12.199.779.2
69-3V79-252.929.50.512.54.399.861.1
69-3V69-247.533.40.416.71.899.882.1
69-3V198-345.135.70.318.90.8100.992.9
78-3H43-759.024.01.97.76.599.137.4
78-3H23-256.627.10.89.46.1100.045.1
78-3H31-655.227.90.710.55.499.951.0
78-3H46-1049.031.80.615.52.499.276.4
 
11°20′N
90-2small57.127.11.19.46.1100.844.9
90-2N37-152.329.70.812.94.299.862.8
90-2N53-247.533.30.516.82.1100.182.5
91-1L14-451.830.30.513.13.899.664.1
91-1L19-648.232.70.615.62.599.777.1
91-1L91-146.434.00.517.11.799.684.5
92-3K15-453.828.40.911.74.899.657.2
92-3N18-149.432.30.415.22.7100.074.5
92-3K13-147.333.30.416.72.099.782.6
93-4G106-954.128.40.811.54.999.855.9
93-4G122-850.131.70.514.33.299.970.2
93-4G4946.434.50.417.21.6100.184.6
94-1N9-352.729.20.612.54.299.361.3
94-1M4-350.730.80.614.03.499.568.5
94-1M41-146.933.50.416.91.999.683.3
99-1y73-1355.228.11.010.85.3100.252.0
99-1y29-452.229.50.913.03.999.563.7
99-1y30-348.432.60.515.72.599.677.1
100-1C27-154.328.40.711.64.999.956.2
100-1C21-251.130.80.514.13.6100.068.6
100-1C47-448.133.10.515.82.4100.077.8
108-1N36-352.929.90.712.54.3100.260.6
108-1N48-151.130.60.513.53.799.466.0
108-1N6-1249.232.60.515.62.7100.575.9
122-4p84-954.227.51.211.35.099.255.6
122-4P30-250.231.10.614.53.399.671.1
122-4p129-946.234.60.417.91.4100.587.8
76-1D10-150.531.40.714.52.9100.170.8
76-1D4-149.431.80.515.22.799.674.8
76-1D23-148.232.70.516.12.199.779.2
69-3V79-252.929.50.512.54.399.861.1
69-3V69-247.533.40.416.71.899.882.1
69-3V198-345.135.70.318.90.8100.992.9
78-3H43-759.024.01.97.76.599.137.4
78-3H23-256.627.10.89.46.1100.045.1
78-3H31-655.227.90.710.55.499.951.0
78-3H46-1049.031.80.615.52.499.276.4
 
11°20′N
90-2small57.127.11.19.46.1100.844.9
90-2N37-152.329.70.812.94.299.862.8
90-2N53-247.533.30.516.82.1100.182.5
91-1L14-451.830.30.513.13.899.664.1
91-1L19-648.232.70.615.62.599.777.1
91-1L91-146.434.00.517.11.799.684.5
92-3K15-453.828.40.911.74.899.657.2
92-3N18-149.432.30.415.22.7100.074.5
92-3K13-147.333.30.416.72.099.782.6
93-4G106-954.128.40.811.54.999.855.9
93-4G122-850.131.70.514.33.299.970.2
93-4G4946.434.50.417.21.6100.184.6
94-1N9-352.729.20.612.54.299.361.3
94-1M4-350.730.80.614.03.499.568.5
94-1M41-146.933.50.416.91.999.683.3
99-1y73-1355.228.11.010.85.3100.252.0
99-1y29-452.229.50.913.03.999.563.7
99-1y30-348.432.60.515.72.599.677.1
100-1C27-154.328.40.711.64.999.956.2
100-1C21-251.130.80.514.13.6100.068.6
100-1C47-448.133.10.515.82.4100.077.8
108-1N36-352.929.90.712.54.3100.260.6
108-1N48-151.130.60.513.53.799.466.0
108-1N6-1249.232.60.515.62.7100.575.9
122-4p84-954.227.51.211.35.099.255.6
122-4P30-250.231.10.614.53.399.671.1
122-4p129-946.234.60.417.91.4100.587.8
Table 5. Representative Pyroxene Chemical Compositions by Electron Microprobea
SamplePointSiO2TiO2Al2O3Cr2O3FeOMnOMgOCaONa2OTotalWoEnFsMg/Mg + Fe
9°30′N
6-2small53.00.51.90.27.70.219.716.60.199.933.454.312.381.9
6-2D9051.40.53.20.75.50.217.420.60.299.741.949.38.884.9
6-2D7051.50.63.10.75.30.117.121.00.299.643.048.48.585.1
14-1v62-353.50.32.20.86.10.121.016.20.1100.332.358.29.586.0
14-1v70-651.90.53.61.15.10.118.119.40.2100.039.852.08.286.3
14-1v57-150.20.84.61.34.70.116.321.10.399.544.048.27.885.9
30-4C11152.10.52.10.47.10.218.418.10.299.137.151.611.482.2
30-4C9251.80.73.20.56.90.217.419.30.3100.339.849.111.081.8
30-4C10151.60.63.00.56.40.216.820.30.299.641.947.810.482.3
32-8H15-752.21.25.20.111.30.315.513.61.1100.433.646.819.670.8
32-8H23-752.00.72.60.18.30.217.418.60.2100.138.248.513.378.8
32-8c13-650.91.16.00.36.90.115.219.40.6100.743.245.211.679.6
35-2B6746.31.510.00.015.60.217.26.81.799.321.551.726.966.1
35-2B3-250.91.23.70.18.90.216.717.90.299.937.847.614.776.8
35-2B8250.61.33.60.19.30.215.119.50.3100.041.243.415.374.3
 
10°30′N
51-3E25-553.50.41.40.111.20.322.111.00.0100.021.760.617.777.8
51-3E24-350.61.33.60.09.90.316.018.10.099.737.646.016.574.2
51-3E37-452.10.72.50.26.80.117.119.90.299.741.147.911.081.6
52-3P33-153.30.52.10.39.50.321.512.40.1100.125.259.615.180.1
52-3P32-252.80.62.30.47.40.218.917.00.299.834.753.411.982.0
52-3P36-350.71.04.50.26.90.115.920.40.3100.043.045.711.380.4
54-4S3452.40.52.20.36.80.118.618.00.299.136.852.310.982.9
54-4S41-749.41.46.80.28.70.215.018.00.6100.340.644.514.975.3
54-4S15-451.60.62.90.55.50.117.021.30.399.743.447.98.784.7
55-1F23-1152.10.61.90.19.80.317.117.30.099.335.448.516.175.5
55-1F23-752.30.61.90.09.50.217.217.60.099.335.948.615.576.2
55-1F23-251.80.72.00.09.80.316.717.90.099.336.747.316.175.1
60-2F16-253.50.41.00.017.10.523.15.20.0100.710.462.726.970.6
60-2F19-450.01.33.30.012.80.316.415.60.3100.032.946.320.869.5
60-2F17-249.31.63.60.010.80.214.519.00.399.340.541.617.970.3
69-3c43-452.80.51.60.27.30.219.117.40.299.335.353.011.782.4
69-3N24-453.40.41.90.35.40.218.319.90.2100.140.450.98.785.6
69-3N14-251.90.82.70.36.70.116.220.90.399.943.445.810.981.0
78-3P5752.40.61.20.113.20.418.312.90.299.226.851.621.671.1
78-3P8951.00.82.40.112.30.316.515.70.399.533.146.820.170.5
78-3P1251.01.02.50.110.50.315.318.60.399.739.243.517.372.0
 
11°20′N
90-2A10-251.70.92.90.27.60.218.617.40.299.835.652.112.381.3
90-2A11-449.02.05.20.18.30.215.918.70.399.740.146.013.977.2
90-2A2-147.72.36.70.37.60.214.220.10.499.544.442.613.076.8
94-1F27-252.80.52.50.57.00.219.916.50.1100.033.355.511.283.4
94-1F31-150.81.14.50.36.20.216.719.70.299.741.448.310.482.6
94-1F15-150.01.04.81.45.30.215.921.00.299.744.047.48.784.3
100-1D4-753.40.51.70.28.40.220.315.00.2100.030.556.113.481.1
100-1D9-252.50.62.20.37.00.218.518.00.299.536.751.911.482.3
100-1D8-750.31.69.00.26.70.113.518.70.7100.845.342.612.078.1
Table 6. Representative Olivine Chemical Compositions by Electron Microprobea
SamplePointSiO2FeOMgOCaOTotalFo
32-8H7-238.621.440.10.3100.476.7
32-8H31-1338.918.941.30.399.579.2
32-8H37-539.117.542.50.399.580.9
32-8c1139.416.543.30.399.582.1
32-8c439.516.144.00.399.982.7
6-2D5738.616.742.50.498.281.6
6-2D5939.316.743.80.4100.182.0
6-2D2039.416.443.80.399.982.4
100-1D52-239.118.241.60.399.280.1
100-1D11-238.918.042.30.399.580.4
100-1D23-238.517.842.70.399.280.7
52-3P25-139.216.442.90.398.882.1
52-3P19-439.515.244.30.399.483.5
52-3P30-339.614.645.10.399.684.3
52-3P18-239.813.945.30.399.385.0
108-1small39.517.143.60.3100.581.7
108-1B3039.315.944.50.3100.183.0
108-1B54-1039.415.245.00.399.983.8
35-2B62-138.223.937.90.3100.373.5
35-2B87-138.022.438.90.399.775.2
35-2B72-438.621.239.70.399.876.7
14-1v34-239.614.845.70.3100.484.4
14-1v36-139.413.646.30.499.785.5
14-1v35-139.912.747.30.3100.286.6
60-2F24-437.328.634.40.3100.668.1
60-2F9-337.327.735.30.3100.569.3
60-2F32-337.626.835.90.3100.570.4
78-3*F7-437.327.635.20.3100.469.0
99-1L86-538.219.041.70.399.279.6
99-1L45-139.615.743.80.399.483.2
99-1L5-839.914.444.70.399.384.6
122-4A4-239.416.043.70.499.482.9
122-4A47-140.212.646.50.399.586.8
122-4A22-140.311.747.40.399.787.8
122-4A59-440.311.247.00.398.988.2
92-3E5-139.815.644.70.3100.483.5
92-3E18-439.315.144.90.399.584.1
92-3E13-139.714.545.40.3100.084.7
23-2x23-639.517.642.20.399.680.7
23-2x19-539.216.943.10.399.481.7
23-2R11-639.814.545.00.499.784.6
23-2R23-339.914.145.10.399.485.0
55-1F16-338.621.040.40.3100.277.1
55-1F7-1238.619.740.90.399.478.4
55-1F49-1939.018.341.80.399.480.0
69-3c19-441.815.732.23.292.978.2
69-3c37-138.917.742.90.399.880.9
69-3c90-239.813.346.10.399.585.9
69-3540.710.548.40.3100.088.9
69-3N4-340.610.148.30.399.389.4
30-4C6938.919.440.60.399.278.6
30-4C6538.919.241.00.399.478.8
30-4C5839.018.042.00.399.380.4
30-4C4439.017.542.20.399.080.8
91-1small39.915.143.90.399.283.6
91-1D2-440.513.546.10.3100.485.7
91-1D1340.312.946.20.399.786.3
76-1M28-139.814.545.50.3100.184.6
76-1M66-639.714.045.60.399.685.1
76-1M76-139.713.446.10.399.685.7
93-4D8539.617.542.80.3100.281.0
93-4A16639.815.644.00.399.883.1
93-4A9339.615.144.90.3100.083.9
54-4s1739.117.243.10.399.881.4

[11] We did quantitative statistical analysis to identify the number of significant chemical groups for each phase in each of the 23 samples. Because repeated measurements of a single composition are normally distributed, many measurements of a mineral with n different compositional groups should produce n normal distributions. So essentially we try to fit the measured data with the least number of normal distributions. We used Kolmogorov-Smirnov test [Press et al., 1989] to measure the probability of n normal distributions being the same as the measured data (goodness of fit). A probability of less than 5% means that at significance level α = 0.05 the n normal distributions are significantly different from the measured data. Therefore our fitting goal was to find the least number of normal distributions that have a goodness of fit (or probability) of more than 5%. Any higher than 5% cutoff of goodness of fit, like 70%, has no clear statistical meaning and is not warranted as we are seeking the least number of normal distributions. We have tested our quantitative statistical analysis by forward and backward modeling: we first combined one to many normal distributions together and then tried to break the combined distribution into separate normal distributions. We always recovered correctly all the major input normal distributions. Minor (less weight proportion) input normal distributions might not be recovered due to the 5% goodness of fit cutoff. Owing to our small analytical errors (or standard deviations) and wide range of mineral compositions, uniformly and continuously distributed data cannot be fitted using limited number of normal distributions.

[12] The above quantitative statistical analysis gives the number of normal distributions, their means, standard deviations, and weight proportions for each mineral in the 23 samples, shown in Table 7. During optimization, we set the upper limit of standard deviation based on our analytic uncertainty. The plagioclase standard for our electron microprobe analysis is An94.5 [An = 100 × Ca/(Ca + Na)] and our analyzed results give An94.6±0.5. The olivine standard (San Carlos) has Fo90.0 [Fo = 100 × Mg/(Mg + Fe)] and our analyzed results are Fo89.9±0.3. The pyroxene standard is Wo39.9 [Wo = 100 × Ca/(Ca + Mg + Fe)], En52.2 [En = 100 × Mg/(Ca + Mg + Fe)], and Mg#86.8 [Mg# = 100 × Mg/(Mg + Fe)], and our analyzed results are Wo39.2±0.9, En52.8±0.9, and Mg#86.8±0.8. The standard deviations shown above are for all repeated analyses of the standards during our one-year period of analysis. For any single probe session, the standard deviations were smaller. We set the upper limit for the standard deviation to be about 2 times the analytical standard deviations, namely An1.0, Wo1.8, En1.8, Mg#1.8, and Fo0.6 to insure that the number of identified normal distributions would be reasonable as larger standard deviations would lead to smaller number of normal distributions. Our t-test and χ2 (chi-square) test showed that when two or more normal distributions were identified for a mineral in a sample, they are statistically different at 99.9% confidence level.

Table 7. Equilibrium Composition and Number of Normal Distributions for Plagioclase, Olivine, and Pyroxene for the 23 Samples From EPR 9°30′N, 10°30′N, and 11°20′Na
SampleMineralEqui CompChemical GroupsLiquidus
MeanStdWt%Fit%
  • a

    Equi comp: the mineral composition in equilibrium with host glass obtained from MELTS under 1 kbar pressure, 0.2 wt% H2O, QFM-1 buffer, and fractionate solid; Std: standard deviation; Wt%: the weight percent of a chemical group; Fit%: goodness of fit measured by Kolmogorov-Smirnov test with a goodness of fit of 6% considered to be significant [Press et al., 1989]; Liquidus: liquidus minerals. The number of identified groups of minerals, when more than one, is indicated in parentheses. Asterisk indicates estimated from simple statistics.

6-2An(2)7371.80.84482Pl + Cpx
  74.51.056 
Wo4141.90.810018
En4849.00.610041
Mg#(2)8179.80.81797
  85.00.583 
Fo82.082.0*0.3*  
14-1An7767.81.01007Pl + Cpx
Wo(2)4341.71.04899
  37.51.852 
En(2)4850.51.77787
  55.81.823 
Mg#8485.70.910079
Fo85.585.50.410092
23-2An(2)7573.41.08428Pl + Cpx
  77.31.016 
Fo83.484.50.210087
30-4An(2)7066.31.04019Pl + Cpx
  68.51.060 
Wo4039.61.710037
En4749.01.310037
Mg#7781.70.510093
Fo78.279.50.61009
32-8An(4)7260.90.93624Pl + Cpx
  64.61.024 
  67.31.030 
  71.40.510 
Wo(2)4141.50.88074
  39.21.020 
En(2)4848.60.95739
  45.90.743 
Mg#(2)8178.60.85633
  83.40.944 
Fo(3)80.578.10.61285
  80.90.651 
  81.70.637 
35-2An(3)6558.11.05157Pl + Cpx
  62.10.723 
  66.01.026 
Wo(2)3838.51.78478
  30.71.516 
En4747.61.81009
Mg#7676.81.810072
Fo74.375.20.610044
51-3An(2)6760.41.04512Pl + Cpx
  65.51.055 
Wo(2)3839.41.17358
  34.61.827 
En(2)4748.41.0647
  49.10.736 
Mg#(2)7681.10.66099
  77.91.940 
52-3An(3)7671.41.02218Pl + Cpx
  78.40.928 
  80.60.949 
Wo4036.81.811 
En4950.51.86 
Mg#8380.41.2100100
Fo84.084.00.610040
54-4An(4)7466.20.5933Pl + Cpx
  71.11.062 
  80.40.69 
  85.70.921 
Wo4142.20.41006
En4848.60.410089
Mg#8184.30.610020
Fo82.181.3*0.1*  
55-1An(7)6955.10.5814Pl + Cpx
  62.71.034 
  66.91.031 
  72.40.54 
  77.60.55 
  81.50.57 
  84.90.511 
Wo3735.90.6100100
En4948.21.010022
Mg#7875.70.6100100
Fo78.478.40.510064
60-2An(2)6053.20.84832Pl ↓ OL
  58.10.552 
Wo(2)1434.01.46642
  38.71.534 
En5446.51.810012
Mg#(2)6370.21.55893
  74.91.142 
Fo69.369.30.510050
69-3An(6)7364.81.01512Pl ↓ OL
  68.91.034 
  72.71.019 
  79.51.09 
  83.70.612 
  88.70.610 
Wo(2)4042.00.68739
  38.80.413 
En4648.80.81003
Mg#7784.11.110017
Fo(3)80.680.60.63312
  85.60.537 
  88.80.530 
76-1An8174.91.010010Pl + Cpx
Fo85.185.10.210085
78-3An(3)5639.30.5166Pl ↓ OL
  47.41.025 
  52.01.059 
Wo(2)3833.21.36780
  38.30.733 
En(2)4044.10.83995
  47.41.161 
Mg#6471.40.710045
Fo65.969.1*0.5*  
90-2An(3)7056.20.51063Pl ↓ Cpx
  62.81.078 
  81.40.812 
Wo4142.01.810046
En4746.21.810065
Mg#8080.21.610054
91-1An(4)7770.30.91152Pl ↓ Cpx
  75.31.027 
  78.01.037 
  81.11.025 
Fo85.085.70.410098
92-3An(3)7467.31.0349Pl + OL
  74.01.041 
  78.70.626 
Fo84.184.10.3100100
93-4An(5)7462.00.6811Pl ↓ OL
  65.30.826 
  70.41.046 
  76.50.613 
  81.20.58 
Fo83.183.10.410078
94-1An(4)7668.90.71535Pl + Cpx
  71.40.53 
  75.71.042 
  79.91.040 
Wo4241.41.810034.3
En4848.81.610086.8
Mg#8383.51.210095.3
99-1An(4)7263.50.81436Pl ↓ OL
  69.11.014 
  71.31.065 
  75.90.67 
Fo(2)81.082.90.56541
  83.90.635 
100-1An(3)7164.01.05191Pl ↓ OL
  69.01.035 
  74.41.014 
Wo(2)4140.51.676100
  34.71.624 
En(2)4648.01.68543
  54.20.515 
Mg#7880.61.110096
Fo80.480.4*0.2*  
108-1An(2)7365.01.05525Pl + OL
  67.81.045 
Fo82.083.00.510063
122-4An(3)7970.81.01335Pl + OL
  81.31.035 
  83.81.051 
Fo(2)87.386.80.61899
  87.80.282 

[13] We provide here the descriptions and interpretations of crystal assemblages for a variety of representative and illustrative samples starting with the simplest mineralogy and progressing to more complex: 1) Ph76-1 with one group of plagioclase and olivine; 2) Ph6-2 with two groups of plagioclase and one group of olivine and discordant pyroxene (pyroxene with discordant or unequal numbers of mineral groups of Wo, En, and/or Mg#); 3) Ph99-1 with four groups of plagioclase and one group of olivine, and 4) Ph69-3 with 6 groups of plagioclase, one group of discordant pyroxene, and three groups of olivine. Detailed descriptions of the remaining 19 samples are available from the authors. Plagioclase and pyroxene compositions in equilibrium with their host lava (Table 7) are calculated using MELTS [Ghiorso and Sack, 1995]. Overall, the samples display an unexpected range of mineralogic complexity more like what is seen in highly phyric samples from the Mid-Atlantic Ridge [Kuo and Kirkpatrick, 1982; Rhodes et al., 1979; Sinton and Detrick, 1992] in terms of the large range of compositional diversity, zoning, and disequilibrium textures.

2.2.1. Ph76-1

[14] Plagioclase is An72.0–79.2 with one identified group at An74.9 (Figure 6). Most plagioclase grains are small (average width about 90 μm) microlites with large aspect ratio and are distributed in clear or dark glass. There are large, low-aspect ratio, diamond-shaped plagioclase grains (width 300 μm) that generally lack twinning and have the same composition as other plagioclase grains. These large grains, cut parallel to (010), are the same as others. Olivine has Fo84.6–85.7 with one identified group at Fo85.1 (Figure 6). Olivine grains are usually small; some large skeletal grains form glomerophyric clusters with small plagioclase grains. Under 1 kbar pressure, 0.2 wt% H2O, QFM-1 buffer, and fractionate solids (these parameters are used consistently in this paper), MELTS predicts that plagioclase in equilibrium with the host glass has An81, close to the upper limit of plagioclase found. Olivine is not a liquidus phase for Ph76-1 according to MELTS. For Fo85.1 to be in equilibrium, the required Kd (Fe-Mg exchange partition coefficient between olivine and host glass, see Roeder and Emslie [1970]) is 0.28 with 90 mole% total iron in glass as Fe2+ (a convention used consistently in this paper).

Figure 6.

Histograms of mineral compositions for Ph76-1. The positions and number of yellow arrows indicate the mean compositions and number of identified chemical groups respectively for each phase. For olivine, the horizontal green bar indicate Fo range for Kd = 0.27 to 0.29.

[15] On the basis of their size and shape, all crystals of Ph76-1 appear to have crystallized during the late stages of magmatic development with relatively fast cooling [e.g., Lofgren, 1983]. This sample shows no evidence for magma mixing or crystal capture and has a simple mineralogy consistent with a simple history of cooling and eruption that was not immediately preceded by a mixing event. Only two of 23 samples we studied have such simple mineralogies (Table 7) and are thus exceptional.

2.2.2. Ph6-2

[16] Much more common are samples with 2–4 distinct compositional groups of plagioclase, such as Ph6-2. Plagioclase has An66–78 with two identified groups at An71.8 and An74.5 (Table 7, Figure 7). Most grains are euhedral to subhedral. Most smaller grains (average width <70 μm) have weak normal zoning and higher An73.7 (average) while larger grains (average width 160 μm) commonly have weak reverse zoning and lower An72.2 (average) interior with An74.0 (average) rims. All the larger grains combined have statistically lower An (An72.7) than all smaller grains combined (An73.7) with a probability less than 10−5 that the two populations are the same. MELTS predicts that equilibrium plagioclase has An73. We have considered several explanations why the smaller size grains have higher An and normal zoning, while the larger size grains have lower An and reverse zoning. Scenarios such as decompression and water loss cannot consistently explain the observed differences. Crystal capture of the large plagioclase grains as an explanation is viable only if they co-existed with substantial amount of magma before being captured because of their mostly euhedral crystal form and the open structure of their glomerophyric clusters. A simple and viable explanation is mixing: magma crystallizing the smaller crystals mixed with slightly more evolved magma crystallizing the larger crystals, with the mixed magma crystallizing the plagioclase rim with a composition in between those of the small and large grain interiors.

Figure 7.

Histograms of mineral compositions for Ph6-2. See Figure 6 for symbols.

[17] Pyroxene grains are euhedral to subhedral and are commonly clustered with large plagioclase grains. Pyroxene compositions have one identified group for Wo41.9 and En49.0 and two identified groups for Mg#79.8 and Mg#85.0 (Figure 7). The Mg#79.8 group could be an artifact of the small sample size, together with the tail formed by groundmass: the long tails for both Mg# and Wo came from tiny pyroxene grains in the groundmass. MELTS predicts pyroxene equilibrium composition has Wo41, En48, and Mg#81.

[18] There are a few tiny olivine grains in the groundmass with Fo81.6–82.4. For Fo82.0 to be in equilibrium with the host glass, the calculated Kd is 0.28, consistent with equilibrium [Roeder and Emslie, 1970; Gee and Sack, 1988].

2.2.3. Ph99-1

[19] Another typical sample is Ph99-1, with An57.1–77.1 and four identified groups at An63.5, An69.1, An71.3, and An75.9 (Table 7, Figure 8). Large (160 μm) plagioclase crystals generally cluster with each other or with large olivine crystals and have core composition in the An71.3 group and rim composition in the An63.5 group (Figure 9). The An75.9 group crystals are less abundant and have a somewhat rounded shape. Other, quite rare large (110μ) plagioclase crystals with (low) An58–60 core and (high) An68–70 mantle/rim have mesh and other resorbed textures and are more densely clustered than the other large crystals. The small (<40μ) plagioclase microlites in the groundmass have the composition of the An63.5 group. MELTS predicts equilibrium plagioclase has An72.

Figure 8.

Histograms of mineral compositions for Ph99-1. See Figure 6 for symbols.

Figure 9.

Microscope images showing crystal networks. Plain light. The length of each image is 2.6 mm. The yellow lines indicate connected or nearly connected crystals. OL, olivine; PL, plagioclase; Py, pyroxene. (a) From Ph23-2; (b) from Ph99-1; (c) from Ph32-8; (d) from Ph91-1.

[20] Olivine is Fo80.1–84.6 with two identified groups at Fo82.9 and Fo83.9. Large olivine crystals have composition in the Fo83.9 group while microlites in the groundmass have composition close to the Fo82.9 group (Figure 8). The rim composition of large olivine crystals generally has higher Fo than the composition of the microlites. For Kd = 0.28, the equilibrium olivine has Fo81.2.

[21] Mixing is mostly likely responsible for the crystal assemblages of Ph99-1.

2.2.4. Ph69-3

[22] This sample represents the extreme of mineralogic complexity we observe, with six identified plagioclase groups at An64.8, An68.9, An72.7, An79.5, An83.7, and An88.7 (Table 7, Figure 10). An unexpected observation is that the microscopic difference among crystals of these groups is surprisingly small, and it is often impossible to tell them apart using textural criteria. For example, very narrow, elongated grains with aspect ratio as large as 17 would be expected to be sodic, yet in this rock they are among the most calcic group. A wide variety of textures are observed in the plagioclase crystals, but the lack of systematic correlation between chemistry, zoning patterns, and textural characteristics hints at a complex history of the mineral assemblages.

Figure 10.

Histograms of mineral compositions for Ph69-3. See Figure 6 for symbols.

[23] Some calcic plagioclase grains show normal zoning which is observable optically. A few grains with An86 show resorbed mesh texture. Grains with An68–72 often have higher An rims or layers adjacent to rims (Figure 11), similar to those of Kuo and Kirkpatrick [1982]. We interpret this zoning to indicate high temperature overgrowth on a lower temperature crystal following mixing. A lot of plagioclase grains, regardless of composition, have An65 to An62 rims. Grains of An65 can be tiny as well as large and are mostly not clustered. Plagioclase grains clustered with pyroxene grains have An68–81, mostly An68–70. An61–63 are from tiny plagioclase microlites clustered with tiny pyroxene grains as well as quench growth on many more calcic crystals. MELTS predicts the equilibrium plagioclase has An73.

Figure 11.

High An near rims for grain V8 in Ph69-3. Each solid circle represents one probe Analysis. We interpret the high An zones to be the result of mixing and the very low An at the rims to be the result of quench growth.

[24] There are three identified groups of olivine at Fo88.8, Fo85.6, and Fo80.6 (Table 7, Figure 10). The Fo85.6 and Fo88.8 groups have normal zoning. The Fo80.6 group comes from the rims of large crystals, tiny skeletal and hopper-shaped grains, and large reversely zoned and rounded crystals. For this Fo80.6 composition to be in equilibrium with the host glass, the required Kd is 0.28.

[25] There are two identified pyroxene groups at Wo38.8 and Wo42.0, and only one identified group at En48.8 and Mg#84.1, so variation of Wo is independent of Mg#, most likely due to sector zoning usually ascribed to rapid growth [e.g., Hollister and Gancarz, 1971; Bryan, 1972]. A single pyroxene crystal may include plagioclase grains of very diverse chemistry (An68–81).

[26] An88.7 group plagioclase grains are usually clustered with Fo88.8 group olivine, An83.7 group grains with Fo85.6 olivine, An68.9 group grains with Fo80.6 and pyroxene, and An64.8 group grains with small pyroxene. Although we do not find Fo88.8 (high) olivine clustered with An83.7 (low) plagioclase, we do find plagioclase grains with a core composition of An88.7 (high) clustered only with Fo85.6 (low) olivine grains, which are likely the result of diffusive re-equilibration of Fo88.8 grains as Fe-Mg diffusion in olivine is much faster than coupled An-Ab diffusion in plagioclase [Morse, 1984; Grove et al., 1984; Chakraborty, 1997].

[27] Mixing is certainly needed to explain the very diverse crystal assemblage of Ph69-3. Multistage mixing is probably needed to explain the complex crystal assemblage observed. Crystal networks very likely had formed before mixing suggested by the common occurrence of crystal clusters.

2.3. Summary

[28] While most of the samples are sparsely phyric and superficially appear to have very simple mineralogies, it is clear that EPR samples are as complex mineralogically as samples from the MAR [e.g., Frey et al., 1974; Rhodes et al., 1979; Kuo and Kirkpatrick, 1982], albeit generally less phyric. Descriptions of the samples not discussed in this section are available from the authors. Our main observations and conclusions from the detailed study of the crystal assemblages are as follows:

[29] 1. Plagioclase crystals have the largest number of chemical composition groups, averaging 3.2 (±1.5). Olivine crystals mostly (80%) have one chemical group, with an average of 1.3 (±0.7) groups. For samples with pyroxene, the average groups for En, Mg#, and Wo are 1.3, 1.3, and 1.5 respectively, all with 1σ = 0.5. Unlike olivine, however, different groups of En, Mg#, and Wo mostly come from the same pyroxene crystals with only one exception. We infer that the number of chemical groups for a mineral is likely determined by two factors: the rate of chemical diffusion or re-equilibration and the order of crystallization. The diffusion of CaAl-NaSi in plagioclase is several orders of magnitude slower than Mg-Fe and MgFe-Ca diffusion in olivine and pyroxene [e.g., Brady and McCallister, 1983; Morse, 1984; Grove et al., 1984; Chakraborty, 1997]. Therefore diverse groups of plagioclase are retained for much longer periods, whereas diverse olivine and pyroxene groups would tend to diffusively re-equilibrate with time. In addition, plagioclase has the longest crystallization history as it is often a liquidus phase and is almost always saturated during crystallization. Slower re-equilibration and longer crystallization history of plagioclase could explain why the number of compositionally distinct plagioclase groups is always greater than or equal to the number of olivine or pyroxene groups.

[30] 2. The presence of diverse groups of plagioclase and olivine provides strong support for multiple magma mixing events [Rhodes et al., 1979] that bring together diverse crystals to the hybrid host. The common occurrence of crystal clusters with open structure suggests that the mixing involved both crystal networks and their interstitial melt like those from mush zones, where in situ crystallization [Langmuir, 1989] might be fostered resulting in the over enrichment of strongly incompatible elements like K and P compared to simple fractionation (Figures 1–3). Lavas without mineralogic evidence of mixing are very rare, and even in these, the lack of multiple groups of crystals does not necessarily imply that mixing did not occur. In rare cases crystal aggregates without open structure, often deformed, were observed. These could be captured solid wall rock with little magma mixing.

[31] 3. Pyroxene is the last mineral to crystallize and is almost always close to equilibrium with the host glass. In samples with two distinct groups of pyroxene crystals like Ph32-8, we found singular pyroxene crystals are close to equilibrium with the host glass while pyroxene crystals in clusters with plagioclase are more evolved.

[32] 4. It is common for primitive crystals to have normal zoning and evolved crystals to have reverse zoning. This is consistent with textural evidence that indicates there is a delicate balance between resorption/dissolution of evolved minerals in hot liquids and growth of primitive overgrowths on evolved mineral cores. Crystals predicted to be in equilibrium with their host glass mostly have weak or absent chemical zonation.

[33] 5. The differences in crystallization conditions like cooling rate and pressure may not explain the chemical difference between phenocrysts (large crystals) and microlites. Mixing is often required after the crystallization of phenocrysts and before the crystallization of microlites. Sometimes small crystals are the primitive ones.

[34] 6. Large and often dense aggregates of crystals are more evolved crystals with no exception in the 23 samples studied. These large aggregates probably were in equilibrium with evolved interstitial melt before their incorporation into the current, more primitive host glass.

[35] 7. An differences or gaps between primitive and equilibrium plagioclase show no correlation with MgO contents of host glasses while those between evolved and equilibrium plagioclase show poor correlation (Figure 12). We interpret these results to indicate that evolved minerals are more apt to be preserved in liquids of moderate to low temperature (lower MgO wt%), as expected, while the dissolution of primitive minerals is minimal. This is supported by the significant dissolution of evolved plagioclase crystals observed in lab experiments [Tsuchiyama, 1985]. Crystals from relatively evolved mixing end-members are much more likely to dissolve than those of relatively primitive end-members. An additional effect may be that the relatively evolved end-members may have few loose phenocrysts to contribute to mixed magmas because most of their fractionated solid forms crystal networks that are difficult to incorporate in a hybrid.

Figure 12.

Correlation between MgO content and An gap or difference of nonequilibrium plagioclase from equilibrium plagioclase. (a) For more primitive plagioclase, there is no correlation. (b) For more evolved plagioclase, the probability of no correlation between MgO content and An gap is less than 3%; if the point with An gap <−12 is removed then the probability of no correlation between MgO content and An gap becomes 20%.

[36] 8. Lavas from EPR 9°30′N, 10°30′N, and 11°20′N are equimodal, suggesting that widespread fractionation of one solid phase from another is not significant.

2.4. Differences Between Robust and Nonrobust Segments

[37] Given the observed petrologic differences between robust and nonrobust segments of the EPR [Batiza et al., 1996], we expected to find significant petrographic and mineral chemical differences in the lavas that could be used to constrain how the eruptive behavior of the EPR might be regulated by the shallow melt lens. We hypothesized that robust segments with a well-defined shallow melt lens would show abundant evidence for magma mixing in the lens, whereas nonrobust segments might show more evidence of mush-dominated petrogenesis or mixing at deeper levels. Surprisingly, we were not able to find ANY significant mineralogic differences between the robust (9°30′N and 11°20′N) and nonrobust (10°30′N) segments.

[38] The average size (width), aspect ratio, and roundness of plagioclase from EPR 9°30′N, 10°30′N, and 11°20′N, computed for the largest 100 crystals in each sample, are listed in Table 8. Among the three segments there are no statistically significant differences in size, aspect ratio, and roundness. The average number of chemical groups of plagioclase in each sample (Table 7) for EPR 9°30′N, 10°30′N, and 11°20′N are 2.3 ± 1.0, 3.5 ± 2.1, and 3.4 ± 0.9 respectively. 10°30′N is not statistically different from 9°30′N and 11°20′N although the two robust segments are statistically different from each other. The average An gaps between primitive and equilibrium plagioclase composition (Table 7) for EPR 9°30′N, 10°30′N, and 11°20′N are An3.7±0.5, An7.3±4.4, and An7.5±4.3 respectively. 9°30′N is statistically different from 10°30′N and 11°20′N but again there are no systematic differences between robust and nonrobust segments. The average An gaps between evolved and equilibrium plagioclase compositions (Table 7) for EPR 9°30′N, 10°30′N, and 11°20′N are An2.2±1.3, An5.1±2.3, and An2.2±1.5 respectively. In this case, there is a statistical difference between 10°30′N (nonrobust) and the robust 9°30′N and 11°20′N probably due to the fact that 10°30′N lavas are on average more evolved. The average An gaps between groups with the highest and lowest An (Table 7) within each sample are An4.5±3.9, An13±10, and An13±6.2 for EPR 9°30′N, 10°30′N, and 11°20′N respectively. 9°30′N is statistically different from 10°30′N and 11°20′N which are not statistically different from each other.

Table 8. Average Plagioclase Crystal Size and Shape for EPR 9°30′N, 10°30′N, and 11°20′Na
SegmentsWidth, mmAspect RatioRoundness
  • a

    We first took high quality digital images from thin sections and then used Zeiss Image version 3.0 to measure width, aspect ratio, and roundness of crystals. Roundness is defined as (perimeter2)/(4 * π * area), with circular objects as 1 and other shapes >1. Numbers after ± are one standard deviation. The average is computed from data on 7 samples from each segment.

9°30′N0.20 ± 0.22.7 ± 0.52.3 ± 0.3
10°30′N0.10 ± 0.032.9 ± 0.42.4 ± 0.5
11°20′N0.13 ± 0.052.7 ± 0.62.5 ± 0.3

[39] The averaged proportions of plagioclase, olivine, and pyroxene pointed-counted in thin sections are listed in Table 9. 9°30′N has statistically lower olivine abundance than 10°30′N and 11°20′N which are not statistically different from each other; 10°30′N has statistically lower plagioclase than 9°30′N and 11°20′N which are not statistically different from each other. And Finally, 11°20′N has statistically lower pyroxene than 9°30′N and 10°30′N which are not statistically different from each other (Table 9). The difference in plagioclase amount between the robust and nonrobust segments is difficult to interpret because it is not matched by other expected mineralogic differences, e.g., in the amounts of Cpx. This difficulty is complicated by the fact that the mineral assemblages we observe are the result of mixing processes and the addition of accidental pieces of crystal networks, not simply minerals crystallizing from melts in a simple cooling scenario. In any case, the average compositions of plagioclase for EPR 9°30′N, 10°30′N, and 11°20′N are An68±5, An67±10, and An72±5, not statistically different from each other.

Table 9. Point-Counted Modal Proportion in EPR Lavasa
MineralSegmentsMean, %Std, %
  • a

    Std, one standard deviation.

Plagioclase9°30′N7819
10°30′N6230
11°20′N8020
Olivine9°30′N812
10°30′N1927
11°20′N1720
Pyroxene9°30′N1315
10°30′N1623
11°20′N39

[40] In summary, we find no systematic significant differences in phase proportions, crystal sizes, or compositions between segments with and without a shallow melt lens, other than those ascribable to the known differences in glass composition discussed before. The differences in glass composition, however, cannot be easily explained by the presence or absence of a magma lens. For example, a magma lens might facilitate crystallization and lead to more evolved lavas [Sinton and Detrick, 1992], but it is the segment without a magma lens that has the most evolved glass composition. The presence of a dynamic shallow melt lens might buffer and homogenize the diverse magmas from below; but this is inconsistent with the observation that significant proportions of E-MORBs were recovered at both 9°30′N and 11°20′N and none at 10°30′N. In a separate study [Pan and Batiza, 2002], the residence time of magma in a mid-ocean ridge magma chamber, based on extensive diffusion study and a mathematical model, was 1 to 3 months with no statistical difference between robust and nonrobust segments.

3. Mid-Ocean Ridge Magma Chamber Processes: Discussion

[41] Our petrological study failed to show any significant role of the shallow melt lens in the erupted lavas at EPR 9°30′N, 10°30′N, and 11°20′N, echoing the work of Van Avendonk et al. [2001], who found that the crustal thickness and seismic structure south and north of the Clipperton transform, with and without shallow melt lens respectively, show no significant difference and concluded that “a steady state magma lens is not required to form normal East Pacific Rise type crust”. What our petrological study revealed is the existence of diverse crystals in EPR lavas, from singular crystals, to crystal clusters (with open structure), to crystal networks (with open structure, Figure 9), and to dense crystal aggregates (without open structure). These crystals represent the entire spectrum from liquid magma to solid rock and may provide insight into magmatic processes under mid-ocean ridges. We will argue below that the fundamentally important magmatic process under mid-ocean ridges, from slow to fast, is the formation of crystal networks and their subsequent compaction; the shallow melt lenses detected along fast and medium spreading ridges contain highly evolved magma and play very little role in crustal accretion.

[42] In a series of studies of lava flows, Philpotts and coworkers [Philpotts and Carroll, 1996; Philpotts et al., 1996, 1998, 1999; Philpotts, 1998; Philpotts and Dickson, 2000] have shown that a fundamental process in crystallization of these flows is the formation of crystal networks and the segregation of interstitial melt. Philpotts and Carroll [1996] and Philpotts et al. [1999] showed experimentally that at ∼25% crystallization, crystal networks form. Shaw et al. [1968] and Marsh [1988, 1989] estimated a similar amount of crystallization for the formation of crystal mush. As little as 20% elongated plagioclase grains might be sufficient to form crystal networks [Burgers, 1938]. Saar et al. [2001] showed that continuous crystal networks could be formed at about 8% to 20% crystallization of plagioclase. The maximum amount of crystallization needed to form continuous crystal networks (randomly oriented) is 22% [Saar et al., 2001]. Compaction of the crystal networks, initiated by the density difference between the crystal networks and their interstitial melt and overlying pressure, expels interstitial (evolved) melt, which, if not erupted, moves upward to form evolved intrusive magma bodies [Philpotts, 1998]. In this way, formation of crystal networks and their subsequent compaction can result in efficient chemical differentiation and physical separation of crystals and their interstitial melt [Philpotts et al., 1996, 1999]. Natland and Dick [1996] found that the proportion of trapped residual melt in gabbro rocks from EPR is extremely low, attesting the efficiency of melt expulsion from crystal networks.

[43] The extent of crystallization is critical in determining whether crystal networks had been formed in EPR lavas. One way to estimate the minimum amount of crystallization is to regard the highest-MgO glass (Ph125-6, Table 1) as a parental magma. The actual parental magmas most likely had significantly higher MgO contents [e.g., O'Hara, 1968; Donaldson and Brown, 1977; Duncan and Green, 1980; Kinzler and Grove, 1992; Sobolev and Shimizu, 1993; Clague et al., 1995; Perfit et al., 1996; Shimizu, 1998]. On the basis of a least squares mass balance calculation involving fractionation of plagioclase, olivine, and pyroxene, the average amounts of crystallization are 34 ± 11 mole%, 47 ± 12 mole%, and 37 ± 10 mole% for 9°30′N, 10°30′N, and 11°20′N respectively, with an overall average of 39 ± 13 mole%, possibly enough to form crystal networks. The total misfit of all elements is 1.7 ± 1.2 mole% averaged for all the samples (notice the purpose here is just to get a broad estimate of the amount of crystallization, not to imply that all samples are daughter melts of Ph125-6). If the glass composition of D20-20, which has 10.47 wt% MgO [Perfit et al., 1996] is used as a parental magma composition, then the average amounts of crystallization are 54 ± 10 mole%, 63 ± 10 mole%, and 55 ± 9 mole% for 9°30′N, 10°30′N, and 11°20′N respectively, with an overall average of 57 ± 10 mole%, and an averaged total misfit of 1.4 ± 1.0%. If the parental magma end-members H and L of Elthon et al. [1992] are used, the overall average of crystallization are 49 ± 5 mole% and 58 ± 9 mole% respectively. There is little doubt that lavas from EPR 9°30′N, 10°30′N, and 11°20′N experienced more than 25% crystallization and could have formed crystal networks.

[44] The formation of crystal networks fits well with our petrological observations. Crystal clusters, some of which preserve a three-dimensional network structure (Figure 9), might be captured pieces of crystal networks during mixing event. Mixing events might be difficult to avoid when a mostly liquid magma passes through crystal networks with significant interstitial melt. Mixing of relatively primitive, mostly liquid magma and evolved, interstitial melt in crystal networks, as proposed by Nakagawa et al. [2002] for Ruapehu volcano, could well explain the complicated mineral assemblages of EPR lavas. As crystallization increases (MgO content decreases), the crystal networks become stronger and harder to break, making the incorporation of crystals in erupted lavas more difficult. The formation of crystal networks could explain the data shown in Figure 4. If most lavas formed crystal networks, their phenocryst contents would be low. The upper limit of total phenocryst contents decreases as MgO contents decrease, the opposite to what would be expected if all crystals were retained with melt during crystallization. This trend is present for all data sets for which there are both modal and chemical data and significant variation in MgO content [Marsh, 1981; Schilling et al., 1982; Michael and Chase, 1987; Michael et al., 1989; K. Swanson and M. O. Garcia, personal communication, 2000].

[45] When pyroxene starts to crystallize from parental magmas of most MORBs, the amount of crystallization is already high. Therefore most pyroxenes might be “locked” in crystal networks. Mixing between primitive magma and interstitial melt could produce a hybrid whose liquidus phases do not include pyroxene but whose chemical fingerprint, inherited from the interstitial melt, could indicate extensive pyroxene fractionation [e.g., Rhodes et al., 1979].

[46] The formation of crystal networks may commonly be followed by compaction and expulsion of interstitial melt, which in turn may move upward to form shallow melt lenses with evolved composition. We argue that it is the expelled and evolved melt, which is near the end of magmatic processes, that forms the shallow melt lenses under mid-ocean ridges.

[47] We believe that the role of the melt lens in crustal accretion has been overestimated. It has been argued that the melt lens is the major locus of “eruptible” magma [e.g., Sinton and Detrick, 1992]. However, because nonrobust fast spreading segments, as well as many intermediate and slow spreading segments have no seismically defined shallow melt lens, yet still possess an upper volcanic layer and experience eruptions, the notion that eruptions can only originate in the melt lens is open to serious question. The shallow melt lens may be analogous to segregation lenses in lava flows, and may contain highly evolved melt, whose role in crustal accretion is very limited.

[48] Given that the subaxial low velocity zone as a whole can be thought of as a large basaltic intrusion several kilometers thick, it is reasonable to ask what is known about the manner in which such bodies solidify. We start with studies of cooling lava lakes such as those of the Kilauea volcano [e.g., Helz et al., 1989; Barth et al., 1994]. In all cases studied, extensive segregation veins, mostly horizontal, formed in the upper crust of cooling lava lakes [e.g., Helz, 1980]. These segregation veins are distinctly more fractionated than their host rocks or original magma [Helz et al., 1994]. Because the segregation veins were formed by the intrusion of fractionated melt into cracks in rigid upper crust [Wright and Okamura, 1977; Chouet, 1979; Helz, 1980], the segregation melt lenses have a solid roof and floor.

[49] The distribution of melt within the crystallizing lava lakes was constrained by drilling into the melt zone and by geophysical experiments [Helz and Wright, 1983; Aki et al., 1978; Hardee et al., 1981; Smith et al., 1977]. The composition of the drilled melts and glasses is clearly that of evolved segregation veins [Helz and Wright, 1983; Helz et al., 1994]. Seismic and electromagnetic experiments mapped a thin melt lens at the boundary between low and high temperature zones, the same position of the drilled-into melt lenses. The main part of the lava lake showed high Vp andversusand high electrical resistivity [Aki et al., 1978; Smith et al., 1977; Barth et al., 1994]. The geophysically mapped low velocity and low resistivity zone is highly likely a zone of segregation melt lenses, consisting of multiple layers of melt lenses because each melt lens is usually less than 1 m thick with a few meters of separation between [Helz, 1980]. The efficient transportation of residual interstitial melt from the lower part to the upper part of the lava lakes is responsible for the formation of segregation veins, as well as for depleting the lower part in elements like Ti that partition into residual melt and enriching the upper part in such elements [Helz, 1989]. These findings in cooling lava lakes are similar to what has been inferred about the shallow melt lenses at the EPR. For example, the high velocity roof and floor [Singh et al., 1999] and the abrupt lateral discontinuities and multilayered melt lenses detected at the EPR [Hooft et al., 1997; Carbotte et al., 2000] are consistent with the idea that the melt lenses are analogous to segregation veins in lava lakes.

[50] Evolved layers, lenses and veins are common features in the upper part of thick (over a few meters) basaltic igneous bodies [e.g., Cornwall, 1951; Lindsley et al., 1971; Page, 1972; Coleman and Peterman, 1975; Coleman, 1977; Helz, 1980; Pallister and Hopson, 1981; Gass et al., 1984; Lippard et al., 1986; Juteau et al., 1988a, 1988b; Nicolas, 1989; Peters et al., 1990; Dostal and Greenough, 1992; Greenough and Dostal, 1992; Parson et al., 1992; Puffer and Horter, 1993; Carman, 1994; Larsen and Brooks, 1994; Thordarson, 1995; Philpotts et al., 1996; Natland and Dick, 1996; Constantin et al., 1996; MacLeod and Yaouancq, 2000; Dick et al., 2000; Coogan et al., 2001]. Of particular interest are oxide gabbro bodies under modern spreading centers at the Pacific Ocean [e.g., Natland and Dick, 1996; Constantin et al., 1996], the Indian Ocean [e.g., Pettigrew et al., 1999; Dick et al., 2000; Coogan et al., 2001], the Atlantic Ocean [e.g., Elthon, 1987; Cannat et al., 1992; Casey, 1997; Ross and Elthon, 1997; Coogan et al., 2000], and the Western Woodlark Basin [Taylor et al., 1999]. To familiarize readers that are new to seggregation veins, Figure 13 shows quite a few of them in a basaltic lava flow outcrop besides the tennis court of University of Hawaii at Manoa. Under mid-ocean ridges, thick magma and slow cooling (forming crystalline gabbros rather than the cryptocrystalline lava flow in Figure 13) make the separation between interstitial melt and crystal networks more efficient, enhancing the formation of large seggregation veins.

Figure 13.

Segregation veins in a basaltic lava flow outcrop besides the tennis court of University of Hawaii at Manoa. The arrows, pointing to the top of the lava flow, indicate the positions of segregation veins. Notice the lateral discontinuity, multiple layer structure, and sharp boundaries of the segregation veins. The cooling of the segregation veins was relatively slow as indicated by their relatively large mineral grain size even though their thickness, which is positively correlated with the thickness of lava flows (or intrusions), is very small. The segregation veins, which have low liquidus temperature due to their evolved chemical composition, intruded their wall rocks, which have high solidus temperature, when the wall rocks were still hot (but already rigid).

[51] The time needed for the formation of crystal networks may be short. The first 25% crystallization requires less than ∼25°C of cooling, estimated from MELTS runs [Ghiorso and Sack, 1995] and from Weaver and Langmuir [1990]. Kelemen and Aharonov [1998] suggested for every 1°C of the first 20°C to 40°C cooling, 1% to 3% solids will crystallize. Magma cools relatively fast immediately after its intrusion into a lower-temperature environment. Efficient heat convection before the formation of crystal networks may exist to facilitate cooling [Hess, 1960; Wager and Brown, 1968; Marsh, 1988; Philpotts and Dickson, 2000]. The formation of crystal networks is therefore quick and consequently the window to detect primitive liquid magma body via seismic imaging is short.

[52] The window to detect evolved segregation melt lenses via seismic imaging, on the other hand, is much longer. Evolved melt can exist within a very wide range of cooling, from ∼20°C to over 200°C below the liquidus. Also, cooling rates at low temperatures may be significantly slower. The large amount of interstitial melt in thick mush zones under mid-ocean ridges [Sinton and Detrick, 1992; Dunn et al., 2000; Crawford and Webb, 2002] provides the source for the formation of evolved segregation melt lenses at shallow depths.

[53] A schematic illustration of what we interpret to be the magmatic processes active under the EPR is presented in Figure 14. Following Boudier et al. [1996] and Kelemen et al. [1997], we show multiple sill injection throughout the gabbro layer, with more primitive sills in the lower part. The formation of crystal networks in the sills and their compaction result in the expulsion of interstitial melt. The expelled melt is either erupted or intruded as sills at a higher level, and then starts the process of crystal network formation and compaction again, similar to the processes envisioned by Kelemen et al. [1997]. The thick cooling subaxial LVZ gabbro produces the thin segregation lenses at the top, in distinction to the gabbro glacier model that the shallow melt lenses produce the thick LVZ by crystallization and downward flow [e.g., Sleep, 1975; Nicolas et al., 1988; Quick and Denlinger, 1993; Phipps Morgan and Chen, 1993; Henstock et al., 1993; Chen, 2001]. Mixing demonstrated by crystal assemblages in erupted lavas occurs in the mush zone when interstitial melts from different intrusive sills or different parts of a sill are brought together or when newly injected magma passes through this crystal network-interstitial melt media, most likely via dikes [e.g., Kelemen and Aharonov, 1998]. The residence time of melts after mixing is short for erupted EPR lavas [Pan and Batiza, 2002], on the order of one to a few months, to keep the hybrid magma eruptible. The deep magma lens, if there is one, likely has primitive (high temperature) magma and probably is the true liquid melt lens in a magma chamber system.

Figure 14.

Schematic illustration of magmatic processes under mid-ocean ridges (not to scale). The crust has three layers: extruded basalt, sheeted dikes, and gabbro. Within the gabbro layer, there are the melt-rich mush zone (dark shading), the melt-free gabbro (no shading), and the transition zone (light shading) in between. The short line segments, which dip away from ridge axis [Nicolas, 1994], indicate the foliation of the gabbro layer formed when crystal mush moves upward in response to ridge opening. Magma from the mantle or the optional magma lens at the Moho would intrude as sills, mostly in the lower gabbro layer. Crystal networks form after <25°C cooling and their subsequent compaction expels their interstitial melt, which either gets erupted or intrudes to a higher level at which crystal network formation and compaction are repeated. The formation of segregation veins from interstitial melt at the top of the gabbro layer is likely and would approach a steady state process where the LVZ has a significant interstitial melt fraction. The shallow melt lenses detected under mid-ocean ridges are very likely segregation veins with evolved composition (low temperature) and play a limited and passive role in crustal accretion.

4. Conclusion

[54] The mineral chemistry and textures of lavas from the northern East Pacific Rise (EPR) 9°30′N, 11°20′N, and 10°30′N have been studied in detail. 9°30′N and 11°20′N have robust magma supply and a shallow magma lens while 10°30′N has nonrobust magma supply and no shallow magma lens. Chemically, the lavas from 9°30′N and 11°20′N exhibit steady state compositions over timescales of ∼1 Ma, whereas those of 10°30′N exhibit departures from chemical steady state and are, as a group, more evolved, lower temperature lavas. Lavas from all three localities are sparsely phyric and glassy, containing plagioclase (Pl) ± olivine (Ol) ± clinopyroxene (Cpx). Typically, the lavas contain several to many (up to seven) distinct chemical groups of Pl that are not always distinct texturally. The lavas may also contain up to three chemically distinct groups of Ol and two groups of cpx. We interpret the origin of the diverse chemical groups of Pl and Ol as the result of magma mixing, with preservation of many more diverse groups of Pl being the result of the much slower rates of diffusion in Pl compared to Ol. The lavas contain both individual crystals and groups comprising reticulate and dendritic clots that we interpret to represent bits of rigid crystal networks forming in mushy zones of magma chambers. The formation of this network may result in the observed negative correlation between extent of fractionation and crystal content. Mixing may occur between relatively fractionated melt batches that occur as interstitial melt in crystal networks and more primitive melt present in sill-like bodies prior to the formation of a rigid crystal networks. We find no significant systematic differences between robust and nonrobust segments in terms of their crystal content, proportion of texturally distinct xenocrysts, crystal size, aspect ratio, roundness, modal abundance, residence time, number of diverse mineral-chemical groups, and characteristics of mixing like the total range of composition of disequilibrium minerals present, and the magnitude of the chemical gaps between disequilibrium and equilibrium compositions. This unexpected and remarkable similarity suggests that the presence or absence of a seismically imaged shallow melt lens has essentially no effect on the mineralogy of erupted lavas.

[55] Mass balance calculation suggests that prior to eruption, most EPR lavas could have been interstitial melt, because their compositions indicate prior amounts of crystallization of over 25%, at which point rigid crystal networks could form. Interstitial melt in relatively thick igneous layers commonly forms segregation veins in the upper part of the layers, as indicated by studies of segregation veins in cooling lava lakes, thick continental lava flows and gabbro plutons. In this context, studies of ophiolites and drilled and dredged gabbros from modern mid-ocean ridge crust, as well as recent seismic experiments, seem to indicate that the seismically detected shallow magma lens represents segregation melt bodies with evolved (low temperature) composition that play only a limited and mostly passive role in crustal accretion. In contrast to the shallow magma lens, the deep, Moho-level magma lens detected by seismic and seafloor compliance experiments likely contains primitive (high temperature) melts. The injection of sill-like bodies throughout the subaxial crustal low-velocity zone may play an important role in the formation of the gabbroic layer. We interpret the chemical differences between 10°30′N (nonrobust) and 9°30′N and 11°30′N (robust) together with their mineralogic similarity as a result of greater thermal vulnerability at 10°30′N due to a smaller overall mush zone or low velocity zone, without any other significant differences in melt delivery, melt migration or eruptive/mixing processes.

Acknowledgments

[56] This work was supported by NSF grant OCE 9711294 to R. Batiza and by the “Bairen Plan” of the Chinese Academy of Sciences to Y. Pan. We acknowledge valuable discussions with many individuals. Formal reviews by P. Kelemen, J. Natland, B. White, and an anonymous reviewer greatly improved this contribution.

Ancillary