Multivariate analysis of an LA-ICP-MS trace element dataset for pyrite

Abstract

Application of multivariate statistics to trace element datasets is reviewed using 164 multi-element LA-ICP-MS spot analyses of pyrite from the Moonlight epithermal gold prospect, Queensland, Australia. Multivariate analysis of variance (MANOVA) is used to demonstrate that classification of pyrite on morphological and other non-numeric factors is geochemically valid. Parallel coordinate plots and correlation cluster analysis using Spearman’s coefficients are used to discover unexpected elemental relationships without making assumptions a priori. Finally, principal component analysis and factor analysis are used to demonstrate the presence of sub-classes of pyrite. Corroborated with geological data, statistical analysis provides evidence for successive generations of hydrothermal fluids, each introducing specific metals, and for partial or complete replacement of different minerals. The data permit reinterpretation of Moonlight as a telescoped system where epithermal-Au (± base metals) is superposed onto early porphyry-Mo mineralization.