Provenance and Geochemical Variations in Shales of the Mesoarchean Witwatersrand Supergroup

New major- and trace-element data are presented for low-grade metamorphosed marine shale units of the highly auriferous Mesoarchean Witwatersrand Supergroup in the Kaapvaal Craton, South Africa. Comparison of the most altered with the least altered samples revealed that postdepositional modification of shale chemistry was significant only near contacts with over- and underlying coarser-grained siliciclastic rocks and along cross-cutting faults, veins, and dikes. Away from such zones, the shale composition can be used to draw inferences on sediment provenance and paleoweathering in the source region and/or on intrabasinal erosion surfaces, as well as on the likely tectonic setting of the depositional basins. The shale composition is highly variable, both laterally within and vertically across stratigraphic units. The shales are characterized by a high chemical index of alteration, chemical index of weathering, and index of compositional variability, all of which suggest that the source area was lithologically complex and subjected to intense chemical weathering. Relatively immobile trace-element ratios confirm derivation from both mafic and felsic source rocks, as indicated by previous geochemical studies on arenitic and ruditic units as well as heavy minerals, and support the contention of a passive-margin setting for the >2.98-2.91 Ga West Rand Group and an active continental margin source for the overlying >2.90-2.78 Ga Central Rand Group-in agreement with a foreland basin setting for the latter. Compositional profiles through the only major marine shale unit within the Central Rand Group, which is host to the greatest portion of Witwatersrand gold, indicate progressive unroofing of a granitic source in an otherwise greenstone-dominated hinterland in the course of sedimentation. They also point to huge differences in the mafic/felsic proportion of source rocks for the different goldfields, with a dominance of mafic to ultramafic rocks in the hinterland of the eastern parts of the basin and of felsic rocks toward the west. It can be inferred from this variation in the provenance of contemporaneous sediments in different parts of the basin that the gold endowment in the various goldfields was controlled by factors other than a specific source rock type. These results have significant implications for the genetic model for Witwatersrand gold deposits and support the hypothesis of leaching of background concentrations of gold from the hinterland, rather than erosion of discrete Paleoarchean to Mesoarchean gold deposits from a specific host rock type. Online enhancements: appendixes, supplemental tables.