Role of batholith residues

Our best conceptual framework for understanding partial melting in a deep crustal sub-arc zone comes from dehydration (vapor-absent) melting experiments. These experiments show that mafic or intermediate protoliths are the most likely sources that generate the compositions observed in the large Cordilleran batholiths (Rushmer, 1991; Johannes and Holtz, 1996). All available high-pressure dehydration melting experiments using basaltic to intermediate starting materials indicate that at pressures exceeding 1 GPa, silicic melt should equilibrate with granulitic or eclogitic residues (Rapp and Watson, 1995, and references therein). Olivine typically is not a residual phase in these assemblages. Melting experiments indicate that feldspar-rich assemblages are replaced by feldspar-free, garnet- and clinopyroxene-dominated assemblages with increasing depth (e.g. Wolf and Wyllie, 1994). This result is consistent with observations from exposed deep crustal rocks (e.g. Barboza and Bergantz, 1996) as well as thermobarometric observations on Sierra Nevada xenoliths (Ducea and Saleeby, 1996), suggesting a transition from granulite facies rocks to garnet pyroxenites at a depth of 40±5 km beneath the arc.

Material balance calculations indicate that the ratio of melt to residue in Andean-type arcs is 1:1 to as much as 1:3 (Kay and Kay, 1991). In the Sierra Nevada , granitoid arc thickness (e.g. Fliedner et al., 2000) and xenolith data (Ducea and Saleeby, 1998b) support a melt to residue ratio of 1:1 to 1:2, which would correspond to a bulk composition of high-Mg basaltic andesite (Ducea, 2002) (Figure 4). The thicker a felsic arc, the more likely it is that it needed to develop an eclogite facies residue. Residual pyroxenites are extremely dense rocks, due to their garnet-rich (50% by volume compared to about 15% in typical subduction-related basaltic eclogites (Carswell, 1990) and Fe-rich nature (Ducea, 2002). Figure 5 shows the calculated densities of Sierra Nevada arc-related residues as a function of depth at their equilibration pressures and temperatures. Rocks that equilibrated at pressures in excess of ~1.5 GPa are garnet pyroxenites, have densities of 3,450-3,550 kg/m 3 , approximately 5 to 10% higher than that of typical mantle peridotites (~3,300 kg/m 3 ) and in agreement with model calculations by Jull and Kelemen (2001).

Therefore, it appears that a non-peridotitic residue of thickness similar to or larger than the calc-alkaline batholiths themselves is required beneath Cordilleran arcs. The residual, low silica material can be in either granulite or eclogite facies. In the case of a major arc with granitoid thickness in excess of 25-30 km, most of the residue is predicted to be a dense, foundering-prone, garnet pyroxenite with a subordinate granulite residue. Residues of thinner arcs would be predominantly granulitic (Clemens and Vielzeuf, 1987) and would not develop a gravitationally unstable root (Ducea, 2002).