Bulk compositional influence on diverse metapelitic mineral assemblages in the Whetstone Lake area, Ontario

Abstract Understanding the interplay between bulk composition and metamorphic grade underpins our interpretations of metamorphism in orogenic belts. The focus of this study is the regional garnet-staurolite-kyanite-sillimanite metamorphic sequence of the Whetstone Lake area, southeastern Ontario. In the kyanite and lower sillimanite zones of this area, there is exceptional diversity in metapelitic mineral assemblages that cannot be accounted for by differences in metamorphic grade. We present a dataset of petrographic observations, phase proportions, whole-rock geochemical compositions, and mineral compositions, from thirty-two samples which encapsulate the range of assemblages found in these zones. Wide, as well as quite subtle, differences in bulk composition are the primary control on mineral assemblage development. Whole-rock XMg = molar MgO/(MgO+FeO) and ${\mathrm{X}}_{\mathrm{Fe}3+}=\mathrm{molar}\ 2\times {\mathrm{Fe}}_2{\mathrm{O}}_3/\left(2\times {\mathrm{Fe}}_2{\mathrm{O}}_3+\mathrm{FeO}\right)$ exert the greatest control on the observed mineral assemblages, whilst MnO, K2O, and Al2O3 have a secondary influence. We use a set of quality factors (Duesterhoeft and Lanari, 2020) to test the ability of thermodynamic models to reproduce the observed mineral assemblages, modal abundances, and mineral compositions in the diverse bulk compositions at Whetstone Lake. Eight samples were selected for phase equilibrium modelling, for which two bulk compositions were calculated for each sample: (1) a whole-rock bulk composition based on an X-ray fluorescence analysis and (2) a carefully considered local bulk composition based on combining mineral proportions with representative mineral compositions, as obtained from a single thin section. Our modelling uses thermodynamic dataset 6.2 (Holland and Powell, 2011) and the solution models of White et al. (2014a, 2014b) that incorporate several Fe3+ end members needed to model the natural data. Modelling in both types of bulk composition broadly predicted mineral assemblages matching those observed. In addition, predicted mineral assemblage fields overlap within uncertainty between 620-675°C and 6.5-7.5 kbar, consistent with the limited range of grade represented by the natural rocks. Predicted modal abundances better match those observed when phase diagrams are constructed using local bulk compositions compared to whole-rock bulk compositions. Despite the acceptable agreement between predicted and observed mineral assemblages, consistent discrepancies are found between predicted and observed mineral compositions. These include overestimation of XMg in garnet, staurolite, and cordierite, overestimation of Ti in staurolite and biotite, underestimation of Si in biotite, and overestimation of Al and underestimation of Fe3+, Fe2+, and Mg in muscovite. The Whetstone Lake suite of this study will be useful to test the predictive capability of future thermodynamic models.