Plumbing and Plumes: Submarine Picritic Magmas from the Southwest Rift Zone of Mauna Loa Volcano.

J. M. Rhodes (Univ. of Massachusetts, Amherst, MA 01003); M.O. Garcia (Univ. of Hawaii, Honolulu, HI 96822); H. Guillou (LSCE-CNRS, 91198 Gif sur Yvette Cedex, France) and M.. D. Kurz (WHOI, Woods Hole, MA 02543)

In order to understand magma generation in the Hawaiian plume it is imperative that we have information on the long-term compositional variation of lavas from the tholeiitic shield-building stage of the volcanoes. For Mauna Loa volcano this information includes the following suites of subaerial lavas: historical lavas (1843-1984), relatively young 14C dated lavas (less than 36 ka), Hawaii Scientific Drilling Pilot Hole samples (1.3-132 ka), and lavas exposed in landslide scarps of uncertain age. Here we discuss our geochemical and isotopic data for lavas from the submarine extension of Mauna Loa's southwest rift zone. Sampling sites include four dredge hauls from the axis of the rift zone at depths between 2550-1925 m (Garcia et al., 1989; Gurriet, 1988) and four dives using the Pisces V submersible (Garcia et al., 1995; Kurz et al., 1995). Three of the Pisces dives (182,183, 184) stratigraphically sampled lavas exposed in a giant 1.5 km scarp resulting from the east Ka Lae landslide. The fourth dive (185) sampled lavas at depths between 1825-1505 m along the rift zone axis. With the exception of one dike and a tephra layer at 755 m, all of the samples are pillow lavas.

Nine samples were dated using the unspiked K-Ar method (Guillou et al., 1997). Six of these yielded unrealistic old ages (0.73-4.46 Ma), probably due to excess argon. Three samples, however, gave realistic ages (133 +/-12 ka; 179 +/- 7 ka; 243 +/- 16 ka) that were consistent with their stratigraphic position and a geologically reasonable accumulation rate of about 4 mm/year. At the times these submarine lavas formed, Mauna Loa was probably sampling different parts of the Hawaiian plume, erupting lavas with compositional characteristics that differ from those of modern lavas. This is shown schematically by the radially-zoned "bulls-eye" plume cartoon (Fig. 1) (modified from Rhodes and Hart (1995)). The important point is that at about 200 ka, Mauna Loa might have been 24 km closer to the plume axis than it is today. These dates also imply that the east Ka Lae landslide was younger than 133 ka.

Figure 1.
The submarine lavas exhibit a very wide range in composition (MgO 6.2-34 %) and, in contrast with subaerial lavas, picrites (MgO greater than14%) and olivine-rich tholeiites (MgO greater than10%) are the dominant rock type. Also, there is no cluster of "steady-state reservoir lavas" (MgO from 6.5-8.0%; Rhodes, 1988), which are typical of historical and young prehistoric lavas (less than30 ka). Major element data show that the compositional range is controlled by accumulation and fractionation of olivine. FeO vs. MgO and Al/(Fe+Mg) vs. Mg/(Fe+Mg) relationships (Fig. 2) indicate that the olivine-control is dominated by Fo90, and that the parental magma was picritic with about 15% MgO. Note that the most forsteritic olivine phenocrysts in these lavas are Fo91, which imply a maximum MgO in the melts of up to about 17% (Garcia et al., 1995). In contrast, olivine-control in historical picrites is dominated by Fo87-89, implying parental magmas with lower MgO contents of about 13% (Rhodes, 1995). These differences require that the parental magmas for the submarine lavas were hotter than for the subaerial magmas. We attribute this to supply from a compositionally zoned and fractionated magma column beneath the summit magma reservoir. This inference has other important implications. It may mean that we have a better chance of identifying diverse parental magma compositions among the submarine lavas, such as those proposed by Sobolev et al. (in press), before they become mixed and homogenized in the magma reservoir. Apart from the higher MgO contents, major element abundances in the submarine lavas are nearly indistinguishable from those of historical and young prehistoric lavas, even for oxides (e.g. SiO2, TiO2, CaO) that typically differ between individual Hawaiian volcanoes (Fig. 3). There is no indication of alkalic lavas throughout the entire 2km thick section.

Most trace element abundances at a given MgO content and trace element ratios (e.g. K/Y, Ba/Sr, Zr/Y, Zr/Sr) are also similar to historical and young prehistoric lavas. Ni tends to be higher at a given MgO value, supporting the inference that the olivines are more forsteritic, crystallizing from a hotter, more MgO-rich, picritic magma.

In contrast, Sr and Pb isotopic data and Zr/Nb ratios show wide variations. Sr isotopes range from values that are slightly less than modern Mauna Loa lavas to values that overlap with those of Kilauea lavas (Fig. 4). Zr/Nb ranges from typical modern Mauna Loa values to Kilauea-like values. In part, lower Zr/Nb values can be attributed to less extensive melting, but this is also source-related. Lavas with isotopic and Zr/Nb ratios comparable with, or close to, historical and young prehistoric Mauna Loa lavas are found at all water depths (fig. 5). The lavas with isotopic and Zr/Nb ratios that are close to Kilauea lavas, or intermediate between Kilauea and Mauna Loa values, are found at depths below 1000 m. They are more prevalent among the dive 185 and dredge haul samples. These data show very clearly that at around 200 ka Mauna Loa was erupting lavas with widely differing isotopic and Zr/Nb ratios, indicating that a wider range of source components were involved in the melting process than in more recent times. Despite the range in isotopic ratios in these older Mauna Loa lavas, there is no correlated change in SiO2 at a fixed MgO content. As noted above, all resemble modern lavas in MgO-SiO2 relationships. This observation is at odds with the findings of Hauri and Kurz (1997).

The relationship between source and melting is explored further (Fig. 6) by comparing the data with melting trends, based on an accumulated incremental melting model, for primitive mantle (Norman and Garcia, 1999). The numbers along the melting vector represent the accumulated melt. Although melting of a single source may account for variations in trace element ratios for many of the submarine lavas, as well as historical and young prehistoric lavas, it cannot explain all of the data. A source similar to that of modern Kilauea lavas is required for some of the deeper and older submarine lavas. More extensive melting of this Kilauea-like source component, or melting of an already depleted Kilauea source component, is required to account for these lavas.

Garcia et al., J. Geophys. Res., 94, 10525-10538 (1989);
Garcia et al., AGU monograph 92, 219-239 (1995);
Guillou et al., J. Volc. Geotherm. Res. 78, 239-250, (1997);
Gurriet, unpubl. M.S. thesis, MIT (1988);
Hauri & Kurz, Earth Planet. Sci Lett. 153, 21-36 (1997);
Kurz et al., AGU monograph 92, 289-306 (1995);
Norman & Garcia, Earth Planet. Sci Lett. 168, 27-44, (1999);
Rhodes, J. Geophys. Res. 93, 4453-4466 (1988);
Rhodes, AGU monograph 92, 242-262 (1995);
Rhodes & Hart, AGU monograph 92, 263-288 (1995);
Sobolev et al., Nature (in press).