ELECTRONIC SUPPLEMENTARY MATERIAL

Analytical methods Dome-lava and enclave samples have been cut and selected for bulk rock analysis. Dome-lava samples were as much as possible cleaned from micro-enclaves dispersed into the groundmass, whereas enclaves were chosen excluding those including blobs of lava-dome or having porphyritic index higher than about 10% at the hand specimen estimation. Major, trace elements and Sr, Nd isotope ratios of the whole rock were analyzed at the Department of Earth Sciences of the University of Florence. Major elements were determined by X- ray fluorescence (Si, Al, Ti, Ca, K, Mn, and Fetot) using the method of Franzini et al. (1972), while traditional wet chemical techniques were used for Na, Mg and Fe2+. Loss on ignition (LOI) was determined after heating the sample powder at 950° C. Some trace elements (V, Cr, Co, Ni, Rb, Sr,

Zr, Nb, Ba, La, Ce, Nd) were also measured by X-ray fluorescence correcting for the matrix effects according to De Vries and Jenkins (1971). Repeated analyses of reference rock standards allowed estimating the uncertainty for trace elements as <10% for Nb, Rare Earth Elements (REE) and Cr and <5% for all other trace elements. A complete set of trace element analysis on selected samples has been performed by ICP-MS at Department of Earth Science of Perugia; the uncertainty is <10% for trace element contents higher than 2 ppm (except for Pb, <15%) and <15% for all the other trace elements (Petrelli et al. 2007). Sr and Nd isotope ratios were determined on a selection of whole-rock samples; measurements were performed at the Department of Earth Sciences of the University of Florence, using standard chemical separation techniques and a Thermal Ionisation Mass Spectrometer Triton-Ti® by Thermo-

Finnigan, equipped with 9 movable collectors. 87Sr/86Sr and 143Nd/144Nd values were measured dynamically and corrected using an exponential mass fractionation law to 86Sr/88Sr = 0.1194 and 146Nd/144Nd = 0.7219, respectively, as described by Avanzinelli et al. (2005). Replicate measurements of NBS 987 and La Jolla standards (0.710249 and 0.511856, respectively (Thirlwall 1991) gave mean values of 87Sr/86Sr = 0.710248 ± 0.000013 (2s, n = 86) and 143Nd/144Nd = 0.511846 ± 0.000007 (2s; n = 67). The Sr analytical blank measured during the course of these analyses was 270 pg (average value of seven analytical blanks lower than 400 pg), which is in the blank range of our lab (Avanzinelli et al. 2005). References

Avanzinelli R, Boari E, Conticelli S, Francalanci L, Guarnieri L, Perini G, Petrone CM, Tommasini S, Ulivi M (2005) High precision Sr, Nd, and Pb isotopic analyses using the new generation Thermal Ionisation Mass Spectrometer ThermoFinnigan Triton-Ti®. Per Mineral 74: 147-166 De Vries JL, Jenkins R (1971) Spettrometria a raggi X in pratica. Biblioteca Tecnica Philips, pp. 193 Franzini M, Leoni L, Saitta M (1972) A simple method to evaluate the matrix effect in X-ray fluorecence analyses. X-Ray Spectrom 1: 151-154 Petrelli M, Perugini D, Alagna KE, Poli G, Peccerillo A (2007) Spatially resolved and bulk trace element analysus by Laser Ablation-Inductively Coupled Plasma Mass Spectrometry (LA-ICP- MS). Geostandard and Geoanalytical Res 31: 13-25 Thirwall MF (1991) Long-term reproducibility of multicollector Sr and Nd isotope ratio analysis. Chem. Geol. Isot Geosci Sect 94: 85-104