During coring at Site 1253, we encountered two igneous units. The upper unit, Subunit 4A, is a gabbro sill between 400 and 431 mbsf and is further subdivided into two subunits on petrologic grounds. Below this unit, ~30 m of sediment was recovered. Subunit 4B (450 mbsf curated depth to the bottom of the hole at 600 mbsf) was subdivided into seven subunits. Both Subunits 4A and 4B are plagioclase- and clinopyroxene-phyric with microcrystalline, fine-grained, and, rarely, medium-grained groundmasses. At 513 mbsf, a thin basaltic (cryptocrystalline) interval was recovered, and Subunit 4B is more glass rich and altered below this depth. These features are generally similar to those described during Leg 206 (Holes 1256C and 1256D) when thick ponded lava flows were encountered in basement (Shipboard Scientific Party, 2003, in press). The data suggest that Subunit 4B is either a series of thick and slowly cooled lava flows formed at the East Pacific Rise, oceanic crust intruded by mutliple gabbro sills generated by the Galapagos hotspot, or a combination of the two. It is possible that the drill string passed from a sill complex to basement at a depth of ~513 mbsf. Postcruise dating and detailed geochemical and isotope analysis will be necessary to discern between the possibilities.
An igneous formation of gabbroic texture was first encountered at 401 mbsf in Hole 1253A during RCB coring. We recovered gabbro from 401 to 432 mbsf before intercepting a layer of clastic, granular limestone, defined as packstone with clay.
Subunit 4A is composed of microcrystalline to fine-grained plagioclase-pyroxene gabbro with plagioclase aggregates or, more rarely, plagioclase with pyroxene aggregates in a holocrystalline groundmass (Fig. F19). The contact between sediment and gabbro was not recovered in place. The maximum age of the gabbroic sill is constrained by the Miocene ooze and clastics from Subunit 3C and based on paleomagnetism data at post-16.726 Ma in age (see "Paleomagnetism").
The principal features observed in core hand specimen of the gabbroic sill are summarized in Figure F20. This is presented in terms of structures such as veins, fractures, magmatic contacts, and voids and textures along with the proportion of each mineral phenocryst and the groundmass characteristics. Several observations can be made: (1) voids are solely located at the top of Sections 205-1253A-4R-CC, 5R-1, and 8R-2; (2) veins are more frequent for Core 205-1253A-5R and the bottom of 8R; (3) magmatic contacts are located at the top of Core 8R; and (4) plagioclase phenocrysts are more abundant than pyroxene phenocrysts except at the top of Core 8R. We tentatively interpret these textural and structural variations as features defining two different parts of the sill. The upper Subunit 4A-1 extends from 401 to 415.2 mbsf (Sections 205-1253A-4R-CC to 7R-2) and the second Subunit 4A-2 from 415.2 to 430.8 mbsf (Sections 205-1253A-8R-1 to 10R-1). The core recovery is >85% within the first subunit, whereas it drops to ~50% in the second subunit. In addition, the ratio of the natural remnant magnetization (NRM) over the magnetic intensity varies from 0 to 1.5 within Subunit 4A-1, whereas it remains close to 0 within Subunit 4A-2 (see Fig. F65). This indicates that the magnetization is stronger and more stable within the upper part than the lower part of the gabbro sill. This could be related to the variation of gabbro grain size as illustrated in Figure F20, where it is possible to distinguish, on a centimeter scale, changes in phenocryst abundance and groundmass grain size. The downhole logging data indicate high Th, U, and K gamma ray counts throughout the gabbro sill.
Eighteen thin sections from the gabbro sill were examined, and Figure F21 shows a typical example of microcrystalline gabbro. Plagioclase (An>50) is present either as aggregates up to 7 mm in size or as several millimeter-sized laths. The proportion of laths and aggregates are related to the groundmass grain size, where smaller phenocrysts are observed in microcrystalline groundmass and larger ones in fine- to medium-grained groundmass. Phenocrysts of clinopyroxene (augite) up to 1.6 mm in size show anhedral to subhedral morphology and are present sometimes with plagioclase aggregates. The groundmass is mainly composed of ~300-µm euhedral to subhedral plagioclase, ~100-µm anhedral clinopyroxene, a rare amount of glass (<2%), palagonite, and clay. Rarely, we observed <1% of anhedral orthopyroxene and/or olivine phenocrysts. Ilmenite and opaque minerals, such as magnetite, up to 15% in total, are observed as either phenocrysts or as small minerals within the groundmass. Finally, we observed elongated and xenomorphic glass, altered glass, and mineral inclusions within plagioclase and sometimes within clinopyroxene as shown in Figure F22. Most of the inclusions are located in phenocryst centers, indicating their incorporation at early stages of mineral crystallization. Detailed thin section descriptions are presented in the thin section description log (see the "Site 1253 Thin Sections").
Alteration in the gabbro sill is characterized by weathering, replacement of primary minerals, and precipitation of secondary minerals in voids, veins, and the groundmass. Voids are elongated (up to 1.2 mm wide) and are present only at the top of each "subunit" and close to veins that are located throughout the full thickness of the sill. Veins, <1 mm wide, are also filled with a mixture of very fine grained cryptocrystalline groundmass (Fig. F23) and altered glass (palagonite). Because of thin section preparation difficulties, most of the vein and void filling material was lost and could not be identified by microscopic studies. However, macroscopic observations show that clay is the main secondary mineral precipitating in voids and veins. Zeolites could not be identified in the XRD pattern of samples taken from veins; however, macroscopic observations show the presence of white secondary minerals in voids that are believed to be zeolites. XRD analyses identified a trioctahedral smectite (saponite) in Sections 205-1253A-5R-2, 5R-3, and 6R-3. Based on thin section observations, it is difficult to clearly identify the presence of minerals from the chlorite group (corrensite and chamosite?, although they are tentatively identified as present). An overview of the minerals identified by XRD analyses is given in Table T5. The main alteration feature observed in thin sections is the replacement of primary minerals by clay, as shown in Figure F24A. Olivine is identified by the shape of the crystal (Fig. F24B), which is completely replaced by clay and remnants of the primary mineral (Fig. F24C). An example of altered plagioclase is shown in Figure F24D, where it is replaced by a mixture of secondary minerals, partly clays. Similar alteration is also observed for other primary minerals. However, most of the phenocrysts remain unaltered and the groundmass shows only a modest amount of clays (~5%-10%).
A second igneous formation is present at 452 mbsf (curated depth) in Hole 1253A below a >30-m-thick layer of clastic, granular limestone or packstone with clay. Logging results indicate that Subunit 4B begins at ~461 mbsf, where the depth difference reflects the standard curatorial practice of moving any recovered material to the top of the core.
Subunit 4B is composed of microcrystalline to fine-grained and, occasionally, medium-grained plagioclase-pyroxene gabbro with plagioclase aggregates or, more rarely, plagioclase with pyroxene aggregates in a holocrystalline groundmass. Very rare horizons of cryptocrystalline igneous rock, with clear basaltic texture, are present within the 148 m of recovered igneous rocks (Cores 205-1253A-12R to 43R). However, we observed textural, structural, and mineralogical features identical to those of Subunit 4A, a gabbroic sill. We terminated Hole 1253A at 600 mbsf without reaching the lower edge of Subunit 4B. As a consequence, it remains unclear whether the gabbro of Subunit 4B is an igneous intrusion within the basement related to the magmatic activity of the Galapagos hotspot or represents a coarser part of the basement of the Cocos plate formed by slow cooling of thick lava flows, similar to the ponded flows observed in basement samples from Holes 1256C and 1256D during ODP Leg 206 (Shipboard Scientific Party, 2003, in press).
The main structural, textural, and mineralogical characteristics seen in core hand specimen of Subunit 4B are summarized in Figures F25, F26, F27, F28, F29, and F30. Representative close-up photographs of cores are added to the figures to illustrate the main macroscopic features. We described the 33 cores (205-1253A-12R through 43R) following the scheme used for Subunit 4A. We used the following criteria to subdivide the cored section:
Using these criteria, we distinguish seven different subunits, which are characterized as follows.
Subunit 4B-1 occurs in Cores 205-1253A-12R through 15R, extending from 450.6 to 470.9 mbsf. It is a 20.3-m-thick microcrystalline gabbro with 2 m of fine-grained gabbro at the top of Core 205-1253A-13R. Voids filled with sediment and/or clays are present throughout Subunit 4B-1 as shown in Figure F25. Recovery was generally low (below 20%) except for Core 205-1253A-14R in which recovery was 87%. Paleomagnetism data indicate that the formation of Subunit 4B-1 took place during an interval of reversed polarity.
Subunit 4B-2 extends from Cores 205-1253A-16R through 17R (470.9 to 480.5 mbsf) (see Figs. F25, F26). A 3.5-m-thick fine-grained gabbro is present at the center part of the subunit and is overlain and underlain by microcrystalline gabbro. Veins <1 mm wide are present throughout the subunit, whereas magmatic contacts are concentrated at the top of Core 205-1253A-16R and at middle of Core 17R. Recovery was excellent (>89%), and paleomagnetism data indicate a reversed polarity for the emplacement of Subunit 4B-2.
Subunit 4B-3 is present in Cores 205-1253A-18R and 19R (480.5 to 490.33 mbsf) and is composed of 10-m-thick microcrystalline gabbro. By comparison, Subunits 4B-1 and 4B-2 exhibit a higher proportion of voids, a larger number of magmatic contacts at the top (see Fig. F26A), a higher amount of veins toward the bottom (Fig. F26B), and a higher proportion of pyroxene phenocrysts. Coring recovery was variable (32.6% and 115.9%).
Cores 205-1253A-20R through 24R constitute Subunit 4B-4, which extends from 490.3 to 513 mbsf. It is essentially composed of fine-grained gabbro with some interlayering of microcrystalline gabbro. At the bottom of Core 205-1253A-22R, we observed a 0.5-m-thick medium-grained gabbro. Magmatic contacts are not frequent. However, veining and void presences are more concentrated at the top and bottom of Subunit 4B-4. Coring recovery was excellent, averaging 83%. Downhole logging data show a resistivity and velocity decrease at 513 mbsf, which is the depth of the petrological boundary between Subunits 4B-4 and 4B-5.
Subunit 4B-5 (513-543 mbsf; Cores 205-1253A-25R through 31R) is characterized by an upper 1.26-m-thick cryptocrystalline basalt (Figs. F27A, F27B, F28). The contact between cryptocrystalline basalt (Subunit 4B-5) and the overlying microcrystalline gabbro (Subunit 4B-4) was not recovered. The cryptocrystalline basalt grades progressively into microcrystalline gabbro without any compelling evidence of contact (Fig. F28). In Piece 8 of Section 205-1253A-25R-1, we observed a dark gray 0.5-cm-wide margin mainly composed of very cryptocrystalline groundmass rather than a glassy chilled margin. Fine-grained gabbro is the most abundant grain size groundmass with intercalation of microcrystalline gabbro at the top (apart from the 1.26 m of cryptocrystalline basalt) and bottom of Subunit 4B-5. The number of magmatic contacts increases within the coarser grained part of the subunit (Fig. F29). The lowest coring recovery (35.2%) was obtained in Core 205-1253A-29R, where the recovered material is mainly composed of the coarsest groundmass of the entire Subunit 4B, which does not contain any veins or fractures. Apart from the 1.26-m-thick cryptocrystalline groundmass, Subunit 4B-5 does not differ from previous subunits. No downhole logging data are available below 550 mbsf. However, paleomagnetism data indicate that a magnetic polarity reversal occurs at the transition between Subunits 4B-4 and 4B-5.
Cores 205-1253A-32R through 36R compose Subunit 4B-6 which extends from 543.2 to 566 mbsf (Figs. F29, F30). It is essentially microcrystalline gabbro with a few intervals of fine-grained gabbro. Subunit 4B-6 shows an increase in small-scale veining (Fig. F30B, F30C) and clustering of magmatic contacts within Core 205-1253A-33R (549.2-552.9 mbsf). The recovery is lower (53.2%) at the top of Subunit 4B-6, whereas it averages ~86% to its base.
Subunit 4B-7 extends from 566 to 600 mbsf and is mainly composed of microcrystalline and fine-grained gabbro (Figs. F30, F31). The first four cores (Cores 205-1253A-37R through 40R) are mainly composed of microcrystalline gabbro, with a larger number of fractures on a centimeter scale (Fig. F30C) and low recovery between 30.8% and 52%. In contrast, Cores 205-1253A-41R, 42R, and 43R are very homogeneous fine-grained gabbro that exhibits some magmatic contacts as well as large veins as illustrated in Figure F31A.
Figure F32 shows the variation of core recovery along with the lithostratigraphy from 370 to 600 mbsf, as well as the main features in each subunit. These observations were used to summarize observations from 180 m of igneous rock and to guide the emplacement of two OsmoSamplers. The first OsmoSampler is centered at 500.37 mbsf in the microcrystalline to fine-grained gabbro of Subunit 4B-4. The second OsmoSampler is centered at 515.8 mbsf, within the upper 10 m of Subunit 4B-5, including the basaltic horizon. Their positions are sketched in Figure F32.
We examined 61 thin sections from Subunit 4B, and we present six photomicrographs of microcrystalline (Fig. F33A, F33B, F33E), fine-grained (Fig. F33C, F33D) and medium-grained gabbro (Fig. F33F) to illustrate the groundmass grain size observed over the entire Subunit 4B. The full set of thin section descriptions are presented as a thin section description log (see the "Site 1253 Thin Sections"). Phenocrysts are plagioclase, clinopyroxene, olivine, and orthopyroxene. Plagioclase (An>55) is present either as aggregates up to 1 cm in size in medium-grained gabbro or as several-millimeter-sized laths. We also observed plagioclase growth zoning, indicating magma differentiation during crystallization of plagioclase phenocryst (Fig. F34). The proportion of laths and aggregates are related to the groundmass grain size. Phenocrysts of clinopyroxene (augite to titanoferrous augite) up to 2.6 mm in size show anhedral to subhedral morphology. Orthopyroxene phenocrysts are less abundant than 1% and are present as up to 1.3-mm-sized anhedral crystals within the groundmass. The abundance of olivine phenocrysts (up to 1.2 mm in size) varies significantly throughout Subunit 4B. This mineral is typically altered as explained in detail in "Alteration" and its shape is generally anhedral. We encountered difficulties in certain thin sections distinguishing olivine from clinopyroxene, as we did not systematically observe clinopyroxene cleavage and olivine morphology has been destroyed by alteration. However, we distinguished each mineral based on crystallographic features such as pleochroic color, extinction sector, and biaxial figure. Ilmenite and magnetite (<7% in total) are present either as anhedral phenocrysts up to 0.7 mm in size or within the groundmass. Groundmass is mainly composed of ~300-µm euhedral to subhedral plagioclase, ~200-µm anhedral clinopyroxene, glass, altered glass (palagonite), and clays. Subunit 4B contains a greater abundance of glass (several percent) than Subunit 4A (<2%).
Melt as glass or altered glass is found as inclusions at the center of plagioclase (Fig. F35), pyroxene, and olivine phenocrysts. They do not present any specific morphology, but they are always located at the center of phenocryst before extensive magma differentiation as shown from plagioclase growth zoning.
Another feature of Subunit 4B is the large number of magmatic contacts. The contacts are defined by a sharp increase in groundmass grain size and phenocrysts. Plagioclase and clinopyroxene exhibit euhedral morphology in the smaller groundmass size, whereas they are anhedral and intergrown with each other in the coarser groundmass size as illustrated in Figure F36. Such features are consistent with injection/intrusion of melt within microcrystalline gabbro, with recrystallization of the groundmass on the microcrystalline side and slow growth of plagioclase and clinopyroxene within the medium-grained groundmass. Such characteristics do not fit with crystallization segregation processes, as intergrowth of minerals will not occur in such a case. In addition, we also observed a higher abundance of glass or altered glass (palagonite) at the magmatic contact than within the gabbro groundmass. However, glass does not define a continuous chilled margin but is present as small accumulations <0.5 mm across. Within these areas, ilmenite and opaque minerals are highly concentrated (up to 20%). Such features support an injection crystallization/recrystallization rather than a crystallization segregation magmatic process. It is also possible that the magmatic contacts represent the location of fluid addition or circulation at high temperature, which implies coarse mineral crystallization.
The degree of alteration is determined by macro- and microscopic observations combined with onboard XRD analyses. Because of the relatively rare abundance of veins and voids and, as a consequence only small amounts of discrete alteration material, only a few XRD analyses of separated vein and void material were conducted. Additionally, whole-rock samples with and without void and vein material were analyzed. An overview of the results is given in Table T5. The identification of different types of zeolites has to be confirmed by onshore microprobe analyses because many relevant peaks occur close to each other, making conclusive identification by XRD difficult.
Subunit 4B is characterized by a higher amount of alteration compared to the gabbro sill (Subunit 4A). However, the discrete alteration still remains low, giving the rocks a relatively fresh appearance in hand specimen. Below Core 205-1253A-24R, which also marks the border of a "subunit," thin sections show an increased percentage of glass as fragments within the groundmass (up to 10% in Sections 205-1253A-34R-2 and 41R-2 in contrast to very rare amounts of glass in the first subunits). Glass remains fresh or is partly altered to palagonite. Clay minerals, originating from palagonite or as alteration products from primary minerals, are common secondary minerals (10%-20%) and can make up as much as 50% in Section 205-1253A-33R-1.
Observations of the primary minerals indicate that olivine is the most altered mineral (up to 100%). In many cases only the shape of olivine can be identified, filled by clay, which is believed to be saponite. Saponite precipitates also in veins as a green mineral. XRD patterns show clear evidence of saponite. The (060) reflection at 1.541 Å and peaks at 1.749, 2.65, 3.14, and 4.64 Å demonstrate the presence of a trioctahedral smectite (Sample 205-1253A-42R-2, 94-99 cm, separated vein material). A glycol-saturated sample shows a shift of the 13.0-Å peak to 14.6 Å, also indicating the presence of saponite. Because of the similar occurrence of peaks, the presence of nontronite is also possible.
Plagioclase is generally more altered than pyroxene; the average of plagioclase alteration is ~10%, and pyroxene alteration is ~5%. However, in rare cases, both minerals can be altered up to 30%-40% or even totally replaced by clay. The alteration of primary minerals starts at glass inclusions or intramineral fissures and, along the rim, weathering to clay minerals. The surfaces of fissures cause the mineral to be divided into several fragments, where the alteration continues until the whole primary mineral is replaced (e.g., Fig. F37 for olivine). Clays are also abundant close to opaque minerals (magnetite and ilmenite), suggesting that some of the opaque minerals that are also found in replaced primary minerals could be alteration products. Chlorite is identified in only a few thin sections (e.g., Sample 205-1253A-42R-3), but XRD analyses indicate the presence of minerals from the chlorite group (probably chamosite and corrensite). The lack of iron oxyhydroxides and the weak appearance of celadonite in XRD patterns could be related to more reducing conditions or their replacement during a later alteration stage.
Only trace amounts of zeolites are found in cavities, voids, and veins in thin sections, as most material is lost because of difficulties in thin section preparation. The identification of zeolites in Section 205-1253A-22R-2 is shown in Figure F38. XRD patterns show distinct reflections at 6.56, 5.91, 5.42, 4.39, and 2.87 Å, indicating the presence of mesolite and scolecite. Other zeolites that are in this sample are thomsonite and phillipsite, although their identification is equivocal in terms of XRD pattern. Nevertheless, laumontite, mesolite, thomsonite, and scolecite, which all belong to the same zeolite group, are common in the upper part of Subunit 4B above Core 205-1253A-25R, whereas phillipsite and stilbite can be identified only below Core 24R (Fig. F39). Stilbite is also identified in Sample 205-1253A-33R-1 (Piece 8A, 132-134 cm) by the characteristic sheaflike aggregates and radiating masses (Fig. F40A, F40B, F40C). Other zeolites are found in thin sections from the upper part of the unit (Fig. F40D, F40E, F40F). The reason for the different types of zeolites in relation to depth could be changing alteration conditions due to changes in fluid flow, temperature, abundance and chemical composition of primary minerals, chemical composition of the groundmass, and different redox conditions.
The presence of calcite is restricted to veins and is clearly identified by XRD pattern with representative peaks at 3.86, 3.03, 2.495, 2.281, 2.090, 1.908, and 1.869 Å in Samples 205-1253A-39R-1, 130-131 cm, and 27R-1, 94-98 cm. In thin sections, calcite could not be identified. However, not all veins were selected for XRD or thin section observations, allowing the possibility that calcite may also be present in other veins of Subunit 4B.
The presence of different secondary mineral assemblages probably reflects different stages in the progressive alteration. Clay minerals are the most abundant secondary minerals and are believed to precipitate first. It is unclear how many clay alteration stages are present. Empty alteromorphs of primary minerals are sometimes replaced with an isotropic alteration mineral and clay, which in terms of olivine alteration is believed to be saponite. At an advanced weathering stage, primary minerals are partly replaced by chlorite, whereas zeolite and calcite precipitate in voids and veins.
Geochemistry of the gabbro sill (Subunit 4A) and the gabbro (Subunit 4B) has been investigated by ICP-AES analyses. The results are presented in Table T6. Issues of data quality are discussed in the "Explanatory Notes" chapter. Major element totals vary between 96.45 and 102.12 wt% with loss on ignition ranging from 0.17 to 3.16 wt%. The principal objectives are (1) to decipher primary geochemical characteristics from those produced by fluid/rock interaction, (2) to examine the extent to which fractional crystallization processes have modified the chemical composition of the melt before emplacement, and (3) to begin the identification of the mantle source of Subunits 4A and 4B.
Figure F41 shows the variation of major and trace element concentration vs. depth (in mbsf). Geochemical variation within and between the gabbro sill and the gabbro is generally small, which is not unusual for rocks of 46-49.5 wt% SiO2. The gabbro sill is nearly holocrystalline and may not represent original liquid composition, whereas we observe more glass and palagonite in Subunit 4B, suggesting that some of the samples (below Core 205-1253A-25R; 513 mbsf) more closely approximate liquid compositions and, thus, better reflect the original mantle geochemical characteristics. In addition, we observed melt inclusions at the center of phenocrysts, and these inclusions may help to constrain the primary chemical composition of the magma. But first of all, it is important to distinguish between geochemical features related to magmatic processes and those induced by fluid-rock interaction. Most of the subunit boundaries, established on macroscopic and microscopic observations, are also highlighted in the geochemistry by concentration changes. This is true for the following subunit transitions: 4A-1 to 4A-2, 4B-2 to 4B-3, and 4B-6 to 4B-7; however, for the remaining three boundaries, additional data are needed. Boundaries characterized by significant changes of major element content such as TiO2, Na2O, or Fe2O3 and changes of concentrations for both an immobile trace element such as Zr and mobile elements like Sr or Ba should be considered as magmatic. In contrast, the boundary between Subunits 4B-6 and 4B-7 may be a zone of fluid-rock interaction, reflected in a systematic increase in all mobile element concentrations relative to immobile ones.
The effects of fractional crystallization and removal of minerals such as plagioclase, clinopyroxene, olivine, and ilmenite/opaque minerals on the geochemical composition of the melt must be evaluated. This could be assessed at first glance in Figure F42. For example, the covariation of TiO2 and V throughout Subunits 4A and 4B indicates the control of ilmenite and/or titanomagnetite on the distribution of TiO2 and V content. This is confirmed in Figure F42, which shows a positive linear variation of V content as a function of TiO2. Both subunits present similar trends, but the gabbro sill is shifted to higher TiO2 contents and lies along a trend with lower V concentration. Therefore, the two units were not generated at the same time and probably not from the same mantle source, as a common linear trend would have been expected in case of a single melting event of a mantle source.
Decreasing MgO and Cr contents for increasing Fe2O3 content at the same depth as shown in Figure F41 suggest that clinopyroxene fractionation occurred and modified the chemical composition of the melt previous to emplacement. Mg# is calculated according to the following equation:
where FeO = 0.8998 x Fe2O3. Figure F43 presents the variation of CaO/Al2O3 as a function of Mg# to assess the effect of clinopyroxene (Ca[Mg, Fe]Si2O6), olivine ([Fe, Mg]2SiO4), and plagioclase ([Ca, Na]Al1-2Si3-2O8) fractionation on the melt composition. Subunits 4A and 4B are separated into two groups: a group with CaO/Al2O3 ratio ~0.9 for variable Mg# values (0.28-0.4) and a second one with CaO/Al2O3 ratios ~0.7 for similar Mg#. The first group may have undergone a significant amount of olivine crystallization and removal at an early stage of melt formation. This is shown in the Ni vs. CaO/Al2O3 diagram, as Ni content decreases with decreasing CaO/Al2O3 values (Fig. F44). In contrast, the low CaO/Al2O3 value of the second group might be related to early clinopyroxene crystallization and removal. This is consistent with a negative linear correlation between Cr and Zr contents of Subunits 4A and 4B (Fig. F45). Plagioclase fractionation seems to take place during melt crystallization and differentiation for the gabbro sill as emphasized by a slight CaO/Al2O3 decrease for decreasing Mg# values. Possible plagioclase control on Sr content is highlighted in Figure F46, where we observe a positive linear correlation between Na2O and Sr. As Sr is a compatible element in plagioclase, the fractionation of this mineral would lower the Sr and Na2O content of the melt.
Overall, the chemical composition of melt derived from the mantle has been progressively changed by mineral fractionation and removal before the emplacement of the gabbro sill and the gabbro of Subunit 4B. From trace element systematics, Subunits 4A and 4B were not generated by the same melting episode. Regarding the mantle source, we do not have at the present the key analytical data (such as Nb, Th, and rare earth elements) to answer this question. It is unclear whether the mantle source is the Galapagos hotspot (an enriched mantle source), the East Pacific Rise (a depleted mantle source), or a combination of the two.