GEOCHEMISTRY

The Indian-Pacific mantle boundary was originally identified within the AAD on the basis of isotopic ratios in seafloor lavas, and these ratios, particularly ²⁰⁶Pb/²⁰⁴Pb, continue to be the only definitive discriminants between Pacific and Indian mantle provinces (Fig. 23). In planning for Leg 187, we worked exhaustively with our data from zero-age and young (<7 Ma) dredge samples throughout the AAD and Zone A to identify a reliable discriminant that could be analyzed onboard the JOIDES Resolution. A single element, barium, appeared to be both reliable as a discriminant and relatively easy to determine with the necessary accuracy. For ease of use, we settled upon a single diagram, Zr/Ba vs. Ba, as providing a clear, visual discrimination between Indian-type and Pacific-type lavas from our 0- to 7-Ma data set. A second element ratio, Na/Ti (expressed as Na₂O/TiO₂ throughout this report), also appeared to have potential as a discriminant, although both elements are susceptible to fractionation by a variety of magmatic processes. As it turned out, the 0- to 7-Ma dividing line on this diagram does not apply to most Leg 187 sites, so we relied almost exclusively on Ba-Zr systematics in our decision-making process. Nevertheless, the Ba Zr systematics that we are relying on are an imperfect discriminant, which means that later isotopic analysis may lead to definitive assignment of some sites that are currently classified as transitional. It is also possible, though less likely, that some sites are currently misidentified and may be reassigned when the isotope data become available.

The glasses from Leg 187 sites are all relatively primitive in composition, with MgO ranging from 9.4 to 7.2 wt%. With few exceptions, whole-rock samples have significantly lower MgO contents than the associated glasses. In the most extreme cases, whole-rock MgO values are lower by >3.0 wt%; in most cases these are lower by 0.5-1.0 wt%. Variations in MgO are seldom accompanied by coherent variations in other elements and, therefore, cannot be attributed to crystallization or other magmatic processes. A particularly clear example of this phenomenon is provided by Hole 1160B, which recovered seven lithologic units, three of which are massive flows interspersed with pillow flows. Each massive flow is overlain by a pillow flow of the same lithology. Whole-rock samples from all three massive flows and glasses from two of the overlying pillow flows are essentially identical in composition with ~8.9-9.1 wt% MgO (Fig. 24). Whole rocks from the pillow flows, however, have lower MgO contents, differing from the glasses by as much as 3-4 wt% MgO. Most other elements in the pillow whole rocks are similar in abundance to the glasses and massive flows. We conclude from these observations that MgO is selectively removed from pillow interiors as a result of the pervasive low-temperature alteration that has affected all our sites.

The data for elements not discussed above have received only cursory examination, but apparently there has been significant temporal variation in primary magma compositions in all three segments. For Segments B4 and B5, these variations are likely related to the westward migration of the depth anomaly relative to the segment boundaries. In Zone A, temporal variability appears to have been modulated by repeated rift propagation.

Mantle Domain Recognition
Barium and Zirconium
Throughout Leg 187, we used ICP-AES Ba and Zr data to determine mantle domain based on the 0- to 7-Ma data fields in the Zr/Ba vs. Ba diagram. Because of the potential for alteration of Ba content by seawater alteration, we used only hand-picked basalt glasses in making this determination. Figure 25 shows all available glass data from the AAD region. There is a clear dividing line between glasses from the Indian mantle domain and those from the Pacific domain. A tie line connecting all the glasses from the transitional domain beneath Segment B5 cuts across both fields. The only 0- to 7-Ma lavas that do not fall within the clearly defined Indian and Pacific fields are a small group of primitive lavas from propagating rift tips in Zone A.

Leg 187 glass data are plotted in relation to the 0- to 7-Ma Ba-Zr fields and the B5 tie line in Figure 26A. Although many of the Leg 187 data plot within or very close to the 0- to 7-Ma fields and can readily be assigned to one of the 0- to 7-Ma fields, there is an important group of data that cannot. These data plot in the area below the Pacific field (i.e., toward lower Zr/Ba and slightly higher Ba than the majority of Pacific type lavas). We refer to these lavas as Transitional-Pacific (TP) type because they appear to represent an extension of the Pacific field toward the compositions of the Zone A propagating rift tip lavas (Fig. 26A). This extension may reflect a temporal shift in any of several source parameters, including source composition and overall extent of melting or a shift in the mixing proportions of one or more mantle end-member components. It could also reflect some form of addition of Ba to the samples from seawater, either directly by low-temperature alteration or indirectly by assimilation of altered crust into the magma. In terms of the Zr/Ba vs. Ba diagram, basaltic liquids plotting in the TP field could also have been derived, by normal crystal fractionation processes, from Indian type parents. Such processes would be expected to produce trends that cut across the observed data array at a high angle. Cursory examination of the data does not, however, support a fractionation origin for the TP glasses, as there is no apparent progression to decreasing MgO across the Indian field or into the TP field. Finally, the TP field is distinct from the B5 tie line and is therefore unlikely to reflect mixing between Indian and Pacific domains.

Sodium and Titanium
For the 0- to 7-Ma glasses, a plot of Na₂O/TiO₂ vs. MgO effectively discriminates Indian domain glasses from those of the Pacific domain, again with the B5 tie line crossing the divide between domains (see Fig. 25B). Unfortunately, this diagram does not apply in a straightforward way to the Leg 187 glasses. In Figure 26B, Leg 187 glass data are plotted on this diagram and coded as Indian, Pacific or Transitional Pacific, according to their position on the Zr/Ba vs. Ba diagram (Fig. 26A). According to this division, no Leg 187 Indian-type lava plots in the 0- to 7-Ma "Indian" field, and lavas of all three types are intermixed in the "Pacific" field.

Although it is offset to lower Na₂O/TiO₂ than the 0- to 7-Ma field, Leg 187 Indian lavas do define a discrete field, although this field also encompasses several TP and Pacific lavas. The three Pacific lavas are, however, all from Site 1160; these also have the three lowest Ba contents of the Leg 17 Pacific lavas. In Figure 26A, they plot to the left of the B5 tie line. The only other sample with comparable low Ba is the borderline Pacific sample from Site 1164 that is plotted as a red triangle in the figure.

The offset of the Pacific and Indian fields appears to reflect the generally low Na₂O contents of Leg 187 glasses relative to those of the AAD. Whether this apparently fundamental shift in primary magma chemistry reflects a temporal change in mantle temperature and/or extent of melting, a change in the source composition or some other source parameter cannot be readily evaluated without further data.

Distribution of Mantle Domains
Most of the Leg 187 sites are distributed along three north-south transects, one each in Segments B4 and B5 and one in western Zone A (Fig. 27, Fig. 28). There are two additional sites in eastern Zone A.

Segments B4 and B5
Indian-type mantle was present beneath all three B5 sites and one B4 site (Site 1163) at their time of eruption. At two sites, basalts derived from two distinct mantle types were erupted in close proximity. Glasses from Hole 1155A are of TP type, whereas those from Hole 1155B, 200 m away, are of Indian type. Glasses from Holes 1164A and 1164B are of Pacific and Indian types, respectively, although the Pacific glass plots very close to the boundary between the two fields. (This glass is represented by a red triangle in Fig. 26A.) Only TP-type glass was recovered at Site 1152.

Zone A West
Six sites form a transect between crustal ages 28-14 Ma in western Zone A (Fig. 27, Fig. 28). Along this transect, Indian-type mantle was present at ~25 and 14 Ma, but in three out of four cases the transition between Indian- and Pacific-type is constrained to have taken place at no more than 2-3 Ma. (The fourth transition, north of Site 1157, is not constrained as there is no site in this area.)

Zone A East
Sites 1154 and 1160 in eastern Zone A are distinctly of Pacific character (Fig. 27, Fig. 28).

Summary
Taking the shipboard identifications of mantle domain at face value, three fundamental observations can be made:

1. No Indian-type mantle occurs east of the 500-m contour on the regional depth anomaly. At ~6 Ma, this contour is very close to the rough-smooth terrain boundary that marks the isotopic boundary in Segment B5.
   
2. Pacific- and especially TP-type mantle occurs throughout the region of the depth anomaly, at least in the older part of the study area.
   
3. Between ~25 and 14 Ma, Indian and Pacific mantle types alternated in western Zone A on a time scale of a few millions of years, comparable to the time scale of the recent migration of the Pacific mantle across Segment B5.
   

A discrete mantle boundary comparable to the present-day boundary in the AAD cannot be mapped through the entire 14- to 28-Ma time interval encompassed by Leg 187, although comparable boundaries may have existed for relatively short, discrete time intervals. For the longer term, it appears likely that the eastern limit of the Indian-mantle province corresponds closely to the eastern edge of the depth anomaly. The locus of this boundary is unconstrained, but it must lie close to the 500-m residual depth contour. West of this boundary, and perhaps coinciding with the >500-m region of the depth anomaly, Indian mantle predominates, but occurrences of Transitional-Pacific- and even Pacific-type mantle are not uncommon. The western limit of Pacific or TP mantle is not well defined by our data, but it cannot be farther east than the 400-m residual depth contour.

The alternation of Indian-type sites with Pacific- and TP-type sites along the western Zone A transect from 14 to 25 Ma suggests a rapid alternation of discrete mantle types on time scales of a few million years, comparable to that of the current migration in the AAD. These occurrences can be interpreted as discrete incursions, either of Indian mantle beneath Zone A or, equally plausibly, of Pacific mantle into the dominantly Indian region of the depth anomaly. If either interpretation is correct, then a discrete Indian-Pacific boundary likely existed for much of that time.

TP-type mantle is not represented in our 0- to 7-Ma data set, but it appears to be an extension of the Pacific-mantle field, rather than a transition between Pacific and Indian types. This type may manifest a mantle component that cannot be seen in the well-mixed, high-flux magma systems of Zone A. In the region of the depth anomaly, magma flux is greatly reduced and small magma batches with a variety of individual source signatures may be erupted without severe modification.