DIFFERENTIATION OF STRONGLY PERALUMINOUS, PERPHOSPHORUS GRANITES - THE PEDROBERNARDO PLUTON, CENTRAL SPAIN

The Pedrobemardo pluton (295 +/- 2 Ma; initial Sr-87/Sr-86 = 0.7122) is a layered, sheet-like granite body with distinct vertical zoning: biotite +/- muscovite-bearing granites in the lower zone, muscovite +/- biotite leucogranites in the middle zone, and muscovite-bearing aplites and pegmatites in the upper zone. The magma was anatectically generated, intruded almost free of restite, and evolved as a closed system by crystallization in situ. Geochemical modelling indicates that the lower and middle zones followed an equilibrium crystallization path, with no separation between solids and melt, and the upper zone followed a fractional crystallization path. Magmatic fractionation produced leucocratic differentiates with high P concentrations and a P/Ca ratio higher than in apatite. The excess P is located in plagioclase and K-feldspar. Monazite (+apatite) fractionation in the lower and middle zones produced extreme depletions in Th, U, Y, and the REEs, and chondrite-normalized REE patterns with low LREEs/HREEs and a slight discontinuity between Nd and Sm. About 90% of the Th, U, Y, and REEs reside in accessory minerals, especially in monazite. Biotite and muscovite have very low REE contents with small positive or no Eu anomalies. K-feldspar and plagioclase in both the lower and upper zones have moderate REE contents with positive Eu anomalies. Feldspars from the upper zone, in contrast, have negative Eu anomalies and have very low REE contents. Compared to available data from literature, D(Sr,Ba,Th,U,Y,REFs)bio/mineral are lower by one to three orders of magnitude, and D(REEs)plag/Kfsp are also lower by one order of magnitude. These huge discrepancies are attributed not only to contamination with REE-rich inclusions in mineral concentrates sed in previous studies, but to real differences in geochemical behavior of trace elements in peraluminous systems, with respect to subaluminous systems. Biotite/melt and, probably, K-feldspar/melt distribution coefficients for Th, U, Y, and the REEs in peraluminous magmas must be much lower than previously considered.