Solving geological mixing problems with Bayesian tracer models: A demonstration of the method applied to Carlin-type pyrite

Abstract

Mixing problems are common in ore deposit geochemistry, since numerous geological sources can contribute metals and other elements during mineralization. Here we demonstrate how Bayesian tracer models from the field of ecology can be used to solve geochemical mixing problems in the study of ore deposits. The model MixSIAR (Mixing Stable Isotope Analysis in R) was developed by ecologists to quantify the proportional contributions to a consumer's diet, using stable isotopic tracers for the consumer and its potential diet sources. In a novel application of MixSIAR, we adapt the method to solve mixing problems in ore deposit genesis. We treat hydrothermal ore minerals as the mixtures and the potential fluids as the sources, enabling us to model the probability distributions of source contributions to the ore minerals. We use Carlin-type pyrite as our example, since the Au enrichment in the pyrite has been alternately suggested to record the circulation of meteoric fluids through a metalliferous package of sedimentary rocks, or the exsolution and ascent of Au-rich fluids from Eocene magmas. Using δ34S as a tracer and Au as a covariate, we model the contributions of four potential sources to Carlin-type hydrothermal pyrite: local sedimentary pyrite, unmineralized Popovich Formation stratigraphy, Jurassic and Cretaceous granitoids, and Eocene magmatic fluids. The modeling indicates that all these sources likely contributed during hydrothermal pyrite growth, and the proportional contribution of Eocene magmatic fluid is positively correlated with Au. We briefly compare the model to other methods, in order to illustrate how Bayesian tracer modeling is ideally suited to study mineralization and other geological processes.