GRANITE (AND RHYOLITE) FINGERPRINTING REVISITED

The aim of this presentation is to revisit the methodologies for granite fingerprinting that were published by myself and co-workers in 1984 and 1996 and which are still in common use for evaluating past tectonic settings of granite genesis. A useful starting point is the detailed and independent assessment made by Förster et al. in 1997. They used data from 250 localities to endorse the key fingerprinting projection of Rb v (Nb+Y) and its discriminant boundaries and to highlight precautions needed for optimum application, notably: 1) that there are ambiguities in classification of granites and rhyolites formed within accretionary and collisional orogenies and within the extensional regimes resulting from subsequent orogenic collapse; 2) that differentiation can sometimes produce compositional trends which cross discriminant boundaries; and 3) that the diagram should not be used alone but in conjunction with other parameters such as dating and geology. As will be shown, precaution 1) may be taken by petrogentic modeling of the fingerprinting projection, precaution 2) by using less felsic, non-cumulate members of co-genetic series, and precaution 3) by using the Bayes Decision Rule to integrate the various geochemical parameters with dating and geology.

It is relevant to this Session that Förster et al. [3] were able to greatly extend the database of granitic compositions from known tectonic settings by including data from evolved volcanic rocks, primarily dacites and rhyolites. Given that the time interval between crystallization and surface exposure makes granites from known settings hard to find, this was an important enhancement. A caveat is that some processes relevant to granite geochemistry, such as crystal cumulation, filter-pressing, and melt-restite mixing may not be represented by erupted liquids. In addition, some intracrustal granite-forming magmas never reach the surface. Thus, volcanic rock geochemistry cannot be used alone for fingerprinting granites, but a combined volcanic-plutonic database is more effective than granites alone. This database also enables the same fingerprinting methodologies to be applied to granites and rhyolites, as will be demonstrated using examples from the focus region of this Session.