APPENDIX

Assessing the Parentage of Altered Rocks

Aphyric and sparsely porphyritic glassy lavas from Pual Ridge ranging from basaltic andesite to rhyodacite display highly constrained geochemical fractionation trends for most major and trace elements (Binns et al., 2002b). Systematic increase in Zr content and decrease of TiO₂ content plotted, for example, against SiO₂ in Harker-type diagrams make the Zr/TiO₂ ratio a sensitive measure of fractionation status. These elements are widely considered immobile during hydrothermal alteration. In Leg 193 altered cores the concept of Ti and Zr immobility is supported by their presence in resistate minerals rutile and zircon, respectively, and by consistency in Zr/TiO₂ ratios across zoned samples with varying alteration assemblages (see Table T2).

Algorithms for calculating precursor compositions, assuming consanguinity of seabed and subseafloor lavas at Pual Ridge, are listed in Table AT1. These have been derived from the glassy lava database (~70 samples, recalculated to 100% anhydrous totals) by plotting each element against Zr/TiO₂ and selecting the better of logarithmic or exponential functions calculated to illustrate the trend. Zirconium is expressed in parts per million and TiO₂ in weight percent. Resultant precursor compositions are also in 100% anhydrous format.

Relative precision of the algorithms for different elements is indicated by cited variances in the ratio of calculated to actual element abundances obtained by reapplying the algorithms to the fresh lava database itself. Algorithms for Cr, Ni, S, and certain other low-abundance elements (e.g., Te) that display scatter in fresh lavas are useful only to assess gross changes in composition and should be applied with due consideration of potential errors. For Cu, the algorithm is derived using only fresh lavas with >60 wt% SiO₂ to avoid the abrupt change in fractionation behavior displayed by this element in more mafic rocks (Sun et al., 2004), but even so scatter remains in Cu contents, leading to an elevated variance. Of the major elements, only Mg shows significant variance, but this is caused by a small number of anomalous samples in the fresh lava database and the algorithm is considered valid overall.

In comparing altered rock compositions with their (anhydrous) precursors it is necessary to take account of water contents of the (clay dominated) former. Structurally combined water has not been separately analyzed for CSIRO samples from Leg 193, and "loss on ignition" determinations for sulfide-bearing samples are an unreliable measure of water. To remove the effects of hydration in comparative plots (e.g., Figure AF1) a factor, calculated from the average for Zr and TiO₂, respectively, of the ratios between actual and algorithm-derived precursor contents of these two constituents, must be applied. Thereby we are comparing a fixed weight of precursor rock with whatever its weight has become after alteration. On diagrams like Figure AF1, Ti and Zr will plot exactly at the unity ratio as required by the immobility concept, except for any minor analytical errors in either element. By restricting the study of alteration geochemistry to CSIRO samples analyzed by the same methods as used for glassy lavas, we avoid interlaboratory variability.

To assess mass changes on a volume basis, it is necessary to use measured or estimated SGs of altered rocks, and precursor SGs (at calculated precursor compositions) derived from measurements on Pual Ridge glasses. This was necessary, for example, in assessing volume changes during alteration (see Fig. F10).

Summary of Results

Abbreviations used below for alteration categories are set out in Table T1 of the main chapter.

Lithophile Elements

• Si is preserved at precursor levels irrespective of alteration style, except for moderate to severe depletion in Category Z samples (e.g., from the Stockwork Zone of Site 1189) and for mild enrichments in a few samples that show petrologic indications of silicification or extensive quartz growth in cavities (Fig. AF2).
• Al behaves overall in an immobile fashion.
• Total Fe is severely depleted in pyrite-poor selvages (Subcategory S) from Site 1188, but is highly enriched (along with S) in most Category Z wallrocks from the Stockwork Zone of Hole 1189B, where much pyrite has deposited in vesicles and microcavities. Elsewhere Fe varies between moderate enrichment and depletion in a manner suggesting overall sulfidation of precursor Fe but with local redistribution of resultant pyrite.
• Mn is everywhere depleted except in the Lower Sequence of Hole 1189B.
• Mg and K show the greatest diversity in behavior across different alteration styles, reflecting respective clay mineralogies (Figs. AF3, AF4).
• Selvages and most pyrophyllite-bearing rocks at Site 1188 are extremely depleted in Mg. At both Site 1188 and 1189 kernel lithologies are moderately to significantly enriched, but this is less pronounced for samples with relict plagioclase (Fig. AF2). Kernels lacking plagioclase below ~190 mbsf at Site 1188 show a slight downhole increase in Mg enrichment. Transitional samples (Subcategory T) vary considerably from depletion to low-level enrichment in Mg.
• Plagioclase-free kernel lithologies at Site 1188 are only slightly enriched in K relative to precursors, the levels increasing slightly downhole. Their equivalents retaining plagioclase are highly K-depleted in the cristobalite domain and less depleted in the quartz domain (Fig. AF3). Bleached samples from the upper pyrophylltic interval in Hole 1188A are severely depleted in K, whereas those from the lower interval of Hole 1188F are only mildly depleted by comparison. Most selvages and transitional samples are more enriched in K than associated kernels.
• Higher levels of K enrichment typify all alteration categories at Site 1189, reflecting presence of hydrothermal K feldspar. In the Lower Sequence of Hole 1189B, cristobalite-bearing samples are less enriched than those containing quartz.
• Rb closely follows the behavior of K. So does Cs, except its downhole profiles are displaced to lower altered/precursor ratios. Lithium is universally depleted.
• At Site 1188, Ba displays a similar behavior pattern to K across differing alteration styles. At Site 1189, however, Ba departs from conformity with K, varying from slightly to extremely enriched. Except in the most extremely enriched samples where barite is probably present, bulk Ba contents are explicable by levels present in illite and potassium feldspar.
• Ca and Sr behave similarly but vary considerably across alteration styles depending on relative abundances of anhydrite and relict plagioclase. Kernel lithologies lacking plagioclase are slightly depleted in the cristobalite domain at Site 1188, highly depleted in the quartz domain, and highly to extremely depleted at Site 1189.
• Contrary to reported immobile behaviors in many ancient metamorphic and hydrothermal environments, rare earth elements (REEs) are variably depleted within the PACMANUS hydrothermal system, the effect being least for plagioclase-bearing material and increasing from La to Yb across the REE series. Typically the light REEs (LREEs; La-Sm) are approximately conserved or mildly depleted, while from Gd onward to Yb the heavy REEs (HREEs) become progressively depleted (to ~0.1 for Yb). Illite selvages (Subcategory S) and transitional alteration styles (Subcategory T) tend to overlap kernels in behavior, whereas pyrophyllite-bearing bleached rocks (Subcategory B) are even more highly depleted.
• At Site 1188 Eu/Eu* anomalies range from mildly negative in most kernels to positive in anhydrite-bearing alteration styles such as illitic selvages. In wallrocks from Site 1189, where anhydrite is subordinate, the Eu/Eu* anomalies are almost exclusively negative.
• Scandium and Y display depletion levels and relationships with alteration style similar to those of the LREEs. Thorium behaves in a very similar fashion to the HREEs.
• Uranium ranges from precursor levels to highly depleted at Site 1188 and from precursor levels to very highly enriched at Site 1189. Below ~150 mbsf at Site 1188 U becomes increasingly depleted with depth. Except that in most plagioclase-bearing samples the levels of depletion or enrichment, respectively, are subdued, there is no clear relationship to alteration styles. This contrasted behavior of U between Sites 1188 and 1189 was observed also in downhole gamma ray logging (Shipboard Scientific Party, 2002b, 200c) and constitutes one of the most prominent lateral geochemical differences between Snowcap and Roman Ruins. A major uranium anomaly (Shipboard Scientific Party, 2002b) in downhole spectral gamma at 197–209 mbsf in Hole 1188F (an uncored interval) remains unexplained.

Chalcophile Elements

• Chalcophile metals show less tendency than lithophiles to vary with alteration style (Fig. AF5), except they are particularly depleted in illitic selvages (Subcategory S) compared to associated kernels (Subcategory K) and transition zones (Subcategory T).
• Sulfur (total) is consistently enriched by two to three orders of magnitude for most alteration styles at Sites 1188 and 1189. Some category Z wallrocks in the Stockwork Zone of Hole 1189B are even more enriched relative to precursors, by 4 orders or more: parallel Fe enrichments indicate mineralization by pyrite. In the Lower Sequence of Hole 1189B, many samples with plagioclase are enriched in S by only 1 to 2 orders of magnitude.
• Tellurium is less dramatically and less systematically enriched than S.
• The behavior of copper is obscured by occasional samples with extreme "nugget effect" enrichments. If the extreme values are disregarded, then at Site 1188 Cu is generally depleted, increasingly so at depth. Most samples in Hole 1189A are mildly depleted. Wallrocks from the upper Stockwork Zone in Hole 1189B vary widely from depletion to enrichment in Cu, whereas in the Lower Sequence it is essentially conserved in category X rocks but enriched in most category Y samples.
• Zinc displays a more coherent downhole profile at Site 1188. It falls progressively from precursor levels at ~50 mbsf to significant depletion at ~250 mbsf, then the profile reverses back toward precursor levels at ~370 mbsf. In Hole 1189A Zn varies from moderate depletion to enrichment in a nonsystematic manner. In Hole 1189B, Zn is mildly to moderately depleted in wallrocks from the Stockwork Zone, whereas in the Lower Sequence it varies from very depleted to highly enriched, the latter being possibly another nugget effect reflecting sphalerite presence. Cadmium closely reflects the behavior of Zn at Site 1189, but at Site 1188 it becomes increasingly depleted at depth without the reversal shown by Zn.
• Lead at Site 1188 exhibits a marked downhole trend from distinct enrichment in the upper cristobalite zone to progressively increasing depletion below ~150 mbsf in quartz-bearing lithologies. In Hole 1189A, Pb is enriched in all but a few samples. In Hole 1189B, Pb spans precursor levels in wallrocks from the Stockwork Zone and varies from slightly depleted to distinctly enriched in the Lower Sequence.
• At Site 1188, both arsenic and antimony show progressive downhole change from enrichment just below the capping of fresh lava to depletion at depths below 120–150 mbsf. This behavior closely follows that of Pb, suggesting that Pb in the altered rocks (as in PACMANUS chimneys) is associated mainly with traces of disseminated sulfosalts. In Holes 1189A and 1189B these elements vary from moderate enrichment to moderate depletion, but only Sb appears closely related to Pb.
• Bismuth is highly enriched at Site 1189 and in the cristobalite domain of Site 1188. Indium is depleted at Site 1188 but mostly enriched at Site 1189. Thallium shows variable behavior at Site 1188 but is slightly enriched at Site 1189.
• Cobalt, a trace element typically hosted in pyrite, behaves similarly to Fe at Site 1188. At Site 1189 it is mostly enriched in Hole 1189A, distinctly enriched in wallrocks from the Stockwork Zone of Hole 1189B, and essentially conserved relative to precursors in the Lower Sequence.