Hornblende- and Phlogopite-Bearing Gabbroic Xenoliths from Volcán San Pedro (36°S), Chilean Andes: Evidence for Melt and Fluid Migration and Reactions in Subduction-Related Plutons

Texte intégral

(1)JOURNAL OF PETROLOGY. VOLUME 43. NUMBER 2. PAGES 219–241. 2002. Hornblende- and Phlogopite-Bearing Gabbroic Xenoliths from Volcań San Pedro (36°S), Chilean Andes: Evidence for Melt and Fluid Migration and Reactions in Subduction-Related Plutons F. COSTA1∗, M. A. DUNGAN1 AND B. S. SINGER2 SECTION DES SCIENCES DE LA TERRE, UNIVERSITE´ DE GENE`VE, 13 RUE DES MARAIˆCHERS, 1211 GENEVA,. 1. SWITZERLAND 2. DEPARTMENT OF GEOLOGY AND GEOPHYSICS, UNIVERSITY OF WISCONSIN–MADISON, 1215 W. DAYTON ST.,. MADISON, WI 53706, USA. RECEIVED JUNE 19, 2000; REVISED TYPESCRIPT ACCEPTED AUGUST 7, 2001. Two groups of gabbroic xenoliths (I and II) containing large proportions of late-crystallized hornblende (up to 50 vol. %) and Na-rich phlogopite (up to 15 vol. %), were brought to the surface by a late Holocene eruption of Volcań San Pedro, the youngest edifice of the Tatara–San Pedro Volcanic Complex (36 °S, Chilean Andes). Group I are inferred to be fragments of partially solidified Holocene plutons because they contain residual interstitial glass, whereas exsolution and deformation textures in Group II indicate that they are fragments of pre-Quaternary plutonic basement. On the basis of textural relations plus the mineral and whole-rock compositions of both groups of xenoliths, we suggest that hornblende and phlogopite with high mg-numbers and Cr contents have formed by reactions between refractory cumulus minerals (olivine, Cr-spinel, pyroxenes or plagioclase) and evolved melts ± aqueous fluids that migrated through partly solidified crystalline frameworks. Thus, the hydrous minerals are not early-crystallized phases in the basaltic magmas from which the cumulus minerals precipitated. The high proportions of hornblende in many subduction-related gabbroic plutons and xenolith suites compared with its paucity in basaltic or basaltic andesitic lavas may be partially explained by multistage plutonic crystallization histories involving reaction and migration of evolved melt ± aqueous fluids that either could have originated within the cumulus pile of the mafic intrusion or were derived externally, from broadly contemporaneous felsic magmas.. Hornblende-bearing gabbroic xenoliths found at subduction-related volcanoes have been inferred to reflect differentiation processes that have operated on magmas of these arcs (Itinome-gata, Japan, Aoki, 1971; Lesser Antilles, Arculus & Wills, 1980; Adak, Aleutian arc, Conrad et al., 1983; Conrad & Kay, 1984; DeBari et al., 1987; Medicine Lake, California, Grove & DonnellyNolan, 1986; Japan, Yagi & Takeshita, 1987; Martinique, Lesser Antilles, Fichaut et al., 1989; Arenal volcano, Costa Rica, Beard & Borgia, 1989; Mt. St. Helens, Cascades, Heliker, 1995; Calbuco volcano, Southern Chile, HickeyVargas et al., 1995; compilation from various sites, Beard, 1986). Many such xenoliths have been interpreted as crystal fractionation residues, and their mineral abundances and compositions have been used to constrain fractional crystallization models of arc magmas. The presence of hornblende in these gabbroic xenoliths and certain apparently refractory compositional characteristics (e.g. high Cr contents) have led to inferences. Extended dataset can be found at http://www.petrology.oupjournals.org ∗Corresponding author. Present address: ISTO, 1A rue de la Ferollerie, 45071 Orleans, France. Telephone: +33-238255213. Fax: +33238636488. E-mail: costaf@cnrs-orleans.fr.  Oxford University Press 2002. KEY WORDS: Tatara–San Pedro; gabbroic xenolith; hornblende; phlogopite;. melt migration. INTRODUCTION.

(2) JOURNAL OF PETROLOGY. VOLUME 43. that (1) hornblende formed as an early-crystallizing mineral in water-bearing mafic magmas (e.g. Conrad & Kay, 1984; Yagi & Takeshita, 1987; Beard & Borgia, 1989), and (2) fractionation of near-liquidus hornblende contributes to the calc-alkaline differentiation trend (Cawthorn & O’Hara, 1976; Beard, 1986; Grove & Kinzler, 1986; Yagi & Takeshita, 1987). Despite the plausibility of a relation between hornblende stability in water and alkali-rich, high- fO2 mafic magmas (e.g. Sisson & Grove, 1993) and early silica enrichment along the calc-alkaline trend, the high modal abundances of hornblende in arcrelated gabbros (xenoliths and plutons) are in marked contrast to the rarity of hornblende-bearing basaltic or basaltic andesitic magmas erupted from subduction-zone volcanoes. This could be explained by the instability of amphibole at low pressure in mafic magma, or by protracted closed-system crystallization reactions in gabbroic plutons. However, the association of hornblende and Na-rich phlogopite in the San Pedro gabbroic xenoliths suggests that reactions in mafic plutons might be triggered also by migrating water-rich melts and aqueous fluids, as has been found in experiments involving interactions between evolved liquids and mafic cumulates (e.g. Prouteau et al., 2001; Costa et al., in preparation). We propose that two groups of hornblende- and phlogopite-bearing gabbroic crustal xenoliths from the Holocene Volcań San Pedro (Tatara–San Pedro Volcanic Complex, Chilean Andes) are distinct in terms of age and origin, but that both suites are the result of multistage differentiation histories involving migration of evolved melts ± aqueous fluids through cumulate piles. Reactions between early-crystallized refractory minerals (olivine, Cr-spinel, pyroxenes or plagioclase) and percolating melts and fluids have produced substantial proportions of hornblende (up to 50 vol. %) plus Na-rich phlogopite (up to 15 vol. %) with features such as high Cr2O3 contents, which have been interpreted elsewhere as indicating early crystallization from hydrous basaltic magmas. The processes of melt and aqueous fluid migration and reaction-replacement proposed for the San Pedro gabbros could be analogous to those described in tholeiitic intrusions (e.g. Muskox intrusion, Canada, Irvine, 1980; Skaergaard intrusion, Greenland, McBirney, 1995; Bushveld Complex, South Africa, Mathez, 1995; Stillwater Complex, USA, Boudreau, 1999) or those caused by interactions between mafic and felsic magmas (e.g. Sha, 1995).. NUMBER 2. FEBRUARY 2002. low-grade metavolcanic and metasedimentary rocks intruded by two shallow-level granitoid plutons dated at 6·2–6·4 Ma (Davidson & Nelson, 1994; Nelson et al., 1999). Gabbroic rocks are not observed at the surface, although gabbro and olivine gabbro (including troctolite and norite) with anhydrous mineralogy are the most common xenolith lithologies and the sources of minor (<10 vol. %) but widespread xenocrystic contamination in mafic to intermediate lavas. The TSPC (>55 km3) consists mainly of basaltic andesitic lavas, although erupted magmas range from primitive basalt to high-SiO2 rhyolite that define calc-alkaline medium- to high-K trends (e.g. Singer et al., 1997; Dungan et al., 2001). Dacitic lavas at the TSPC typically contain minor hornblende phenocrysts that are rarely accompanied by biotite, but no lava with <65 wt % SiO2 contains hornblende phenocrysts.. Volcań San Pedro and xenoliths The principal phases of volcanic construction at Volcań San Pedro (1·5 km3) are divided into a main cone-building stage comprising andesitic and dacitic lavas, and a late event triggered by sector collapse of the eastern flank. The latter comprises an explosive eruption that produced air-fall dacitic tephra (Singer & Dungan, 1992; Singer et al., 1995) followed by extrusion of a succession of lavas from the eastern flank: (1) 0·2 km3 of biotite–hornblende dacite containing up to 10% mafic xenoliths plus minor quenched mafic inclusions (QMI); (2) 0·5 km3 of twopyroxene dacite with abundant QMI; (3) 0·1 km3 of twopyroxene andesite with rare QMI. The last volcanic activity at this cone consisted of 0·2 km3 of basaltic andesitic magma erupted from the summit crater. The majority of the xenoliths are gabbroic (22 samples), although scarce granites and metamorphic rocks (hornfels) similar to exposed basement are also present. Small xenoliths are rounded to subrounded, but larger fragments (up to 45 cm in diameter) tend to be angular. The observation that the xenoliths are found exclusively in the first lava flow following structural failure of the east flank of Volcań San Pedro suggests that they are fragments of the conduits or upper parts of the margins of the San Pedro magma chamber that were shattered during the eruption (in a similar fashion to the 18 May 1980 Mount St. Helens eruption; Heliker, 1995).. GEOLOGICAL SETTING. TEXTURES, AND MINERAL AND GLASS COMPOSITION OF THE XENOLITHS. The Quaternary Tatara–San Pedro Complex (TSPC; 36°S, 71°51′W) is a long-lived frontal arc volcanic centre (930 ka; Holocene) of the Southern Volcanic Zone of the Andes (Fig. 1). Exposed basement lithologies are mainly. Major and minor element compositions of minerals and glass were determined by electron microprobe (CAMECA SX-50; see the Appendix). Mineral names, structural formulae and end-members were determined. 220.

(3) COSTA et al.. ´ N SAN PEDRO GABBROIC XENOLITHS, VOLCA. (2) Group II have subsolidus exsolution and deformation textures (19 samples). These observations and 40 Ar/39Ar data from two xenoliths (Fig. 3) indicate that they are fragments of the pre-Quaternary, plutonic basement of the volcano. These have been further subdivided into: (a) Group IICL, which are mainly clinopyroxene leuconorites; (b) Group IIHN, which are hornblende norites.. GROUP I xenoliths: olivine–hornblende norites and melanorites. Fig. 1. Simplified geological map of Central Chile showing the location of the Tatara–San Pedro Volcanic Complex. Μ, main Quaternary volcanic centres. Pz, Palaeozoic rocks; Mz, Mesozoic rocks. Grey shaded areas indicate Tertiary plutons. Figure modified from Hildreth & Moorbath (1988) and from Dungan et al. (2001). The location of Tertiary plutons is from Mapa Geolo´gico de Chile (Servicio Nacional de Geologıá y Minerıá, 1982).. following Morimoto et al. (1988) for pyroxenes; Leake et al. (1997) for amphiboles; Rieder et al. (1998) and Deer et al. (1962) for micas; Deer et al. (1992) for olivine, plagioclase and apatite; and Stormer (1983) for spinel and ilmenite. Tables with the complete electron microprobe analyses of spinel, ilmenite, pyroxenes, olivine, plagioclase and glass can be downloaded from the Journal of Petrology Web site at http://www.petrology.oupjournals.org. None of the samples shows evidence of low-temperature hydrothermal alteration. Partial melting along grain boundaries or post-entrainment modification as a result of interaction with the host lava are minimal. On the basis of textures and modal mineralogy (Table 1 and Fig. 2), we have divided the xenoliths into two groups [nomenclature after Streckeisen (1976) and LeMaitre (1989)]: (1) Group I are olivine–hornblende norites and melanorites (three samples) with interstitial glass bounded by euhedral crystal faces. These samples were probably dislodged from partly solidified crystal-rich zones of an active conduit or reservoir system.. These samples consist of olivine, orthopyroxene, hornblende, plagioclase and phlogopite forming a mediumgrained (1–5 mm) crystal network with interstitial dacitic to rhyolitic glass (SiO2 >67–72 wt %; K2O >3·7– 8·6 wt %) bounded by euhedral crystal faces suggesting that the glass is residual from crystallization and not due to partial melting (Fig. 4a). Cr-spinel (Ulv0·01–0·31; Cr2O3 >10–18 wt %) occurs as inclusions in olivine, orthopyroxene, hornblende and rarely in plagioclase cores. Olivine (Fo86–76; NiO Ζ0·05−0·4 wt %) is typically resorbed and surrounded by hornblende, orthopyroxene and phlogopite that formed in reaction relationship with olivine (Fig. 4b). Less commonly, euhedral olivine showing no reaction is in contact with glass (Fig. 4a). Rare clinopyroxene (Wo46–45En46–44Fs9–11; Cr2O3 Ζ0·3– 0·8 wt %) is also resorbed and has largely been replaced by hornblende. Plagioclase is euhedral, whether it occurs as free crystals or as inclusions within orthopyroxene, hornblende or phlogopite. Most plagioclase crystals consist of a normally zoned core (An86–78) surrounded by a normally zoned rim (An45–26) with an >35 An mol % gap between cores and rims (Fig. 5). The high Fo and NiO contents of olivine, the high Wo and Cr2O3 contents of clinopyroxene and the high An contents of plagioclase cores suggest that these minerals are near-liquidus crystallization products of a water-bearing basaltic magma (e.g. Gaetani et al., 1993; Sisson & Grove, 1993). Subhedral to euhedral orthopyroxene, hornblende (magnesiohastingsite) and phlogopite are late-crystallizing minerals, as they are commonly in contact with interstitial glass (Fig. 4a). Despite this, the three minerals are characterized by high mg-numbers [mg-number = 100Mg/ (Mg + Fet) in mols, where Fet is total iron] ranging from 77 to 82, and Cr2O3 contents from <0·1 to 1·2 wt % (Tables 2–4). The Cr2O3 concentrations in these three minerals vary irregularly within and between crystals, and they are as high as, or higher than those of clinopyroxene (Fig. 6). This is in accord with the textural relations indicating that orthopyroxene, hornblende and phlogopite are the products of reactions between cumulus minerals (olivine, Cr-spinel and clinopyroxene) and liquid. Most phlogopite is characterized by Na2O contents. 221.

(4) Rock type. Grain size. Ol norite. Ol melanorite. Hx14k. Hx14n. medium. medium. medium. Hbl norite. Cpx leuconorite. Cpx leuconorite. Cpx leuconorite. Cpx leuconorite. Cpx leuconorite. Hbl norite. Hbl leucogabbro fine–medium. Cpx leuconorite. Hbl norite. Cpx leuconorite. Cpx leuconorite. Hbl leuconorite. Hbl gabbro. Hx14c. Hx14d. Hx14e. Hx14h. Hx14i. 222. Hx14j. Hx14l. Hx14m. Hx14q. Hx14s. Hx14u. Hx14w. Hx14y. Hx14x. Hbl melanorite. Hx14z. 22·6. 0·0. medium–coarse 10·3. coarse. fine. 0·5. 0·0. 0·0. 0·0. 0·5. 0·0. 0·0. 0·0. 0·0. 3·8. 7·5. 0·0. 8·8. 0·0. 0·0. 0·0. 0·0. 0·6. 3·0. 0·2. 3·5. 7·5. 0·0. 9·3. 4·9. 1·5. 0·0. 6·1. 0·0. 9·7. 6·0. 0·0. 5·8. 7·2. 25·9. 8·0. 16·7. 7·2. 0·6. 19·2. 13·9. 14·5. 13·4. 10·7. 11·7. 16·1. 13·0. 14·1. 16·8. 20·7. 8·0. Opx. 48·1. 31·6. 30·2. 10·4. 4·5. 0·1. 28·0. 8·1. 16·3. 37·9. 2·2. 8·3. 1·3. 8·3. 5·2. 23·1. 3·6. 13·8. 1·7. 33·1. 21·1. 11·5. Hbl. 25·3. 35·5. 61·4. 55·3. 81·3. 64·1. 57·0. 55·9. 65·5. 50·8. 62·3. 66·8. 70·2. 68·5. 70·8. 50·0. 65·9. 58·5. 68·1. 23·4. 25·3. 37·2. Pl. 3·9. 4·0. 1·5. 15·0. 1·4. 3·9. 0·0. 0·0. 0·0. 0·0. 0·9. 0·6. 1·9. 1·8. 0·0. 0·3. 0·1. 0·0. 0·7. 3·0. 7·9. 1·6. Phl. 0·0. 0·0. 0·0. 13. 0·5. 0·0. 2·7. 11·7. 0·0. 2·1. 1·9. 2·7. 3·8. 0·0. 0·0. 9·4. 0·8. 1·6. 0·2. 0·0. 0·0. 0·0. Phl(lt). 2·7. 0·4. 6·9. 0·1. 1·3. 2·9. 3·1. 3·7. 3·4. 6·4. 4·3. 2·8. 3·0. 0·6. 2·6. 1·0. 4·3. 4·0. 4·0. 0·3. 1·4. 0·9. Opaque. 0·0. 0·0. 0·0. 0·0. 3·2. 0·0. 1·2. 0·0. 0·0. 2·2. 0·0. 0·0. 0·0. 0·0. 4·6. 4·6. 0·0. 6·5. 2·7. 2·7. 0·6. 12·7. Glass. Pl, Opx, Phl. Pl, Opx, Phl. Pl. Pl. Pl. Pl. Pl. Pl. no. Pl. Pl. Pl. Pl. Pl, Phl. Pl. Pl. Pl. Pl. Pl. no. no. no. Opx, Pl, Hbl, Phl. no. no. Opx, Hbl, Phl. Opx, Hbl, Phl. Hbl. Opx, Pl, Hbl. Hbl, Phl (1 mm wide). no. no. no. Hbl, Phl, opaque. Hbl. no. Opx, Hbl, Phl, opaque. Opx, Pl. Opx, Hbl, Phl, opaque. no. no. no. no. no. Microfractures. Modes based on counting >2000 points except for fine-grained rocks (>1000 points). Average grain size: coarse >5 mm; medium 1–5 mm; fine <1 mm. Rock names follow Streckeisen (1976) and LeMaitre (1989). Mineral symbols after Kretz (1983). Phl (lt) refers to late phlogopite.. Hbl norite. Hx14v. medium–coarse. medium–coarse. 0·6 0·0. 2·3. 8·4. 0·3. 0·0. 0·0. Cpx. Deformation. NUMBER 2. medium. medium. medium. medium. medium. medium–coarse. medium–coarse. coarse. medium. medium–coarse. fine–medium. 0·4. 0·0. 20·6. 22·9. 28·2. Ol. Modal mineralogy (vol. %). VOLUME 43. Group IIHN. Cpx leuconorite. Hx14a. medium. Hbl leuconorite. Hx12b. medium. Cpx leuconorite. Hx12a. Group IICL. Group II: xenoliths with subsolidus textures. Ol norite. Hx14b. Group I: partially crystallized xenoliths. Sample. Table 1: Modes, names, and main textural features of the xenoliths. JOURNAL OF PETROLOGY FEBRUARY 2002.

(5) ´ N SAN PEDRO GABBROIC XENOLITHS, VOLCA. COSTA et al.. Table 2: Ranges of mg-numbers and Cr2O3 (wt %) contents of the mafic minerals from the three groups of xenoliths Ol. Cpx. Opx. Hbl. Phl. Group I mg-number. 86–76. 84–81. 82–77. 80–72. 82–77. Cr2O3. n.a.. Ζ0·9. Ζ0·5. Ζ1·2. Ζ0·6. Group IICL mg-number. 81–72. 81–73. 80–65. 77–64. 81–70. Cr2O3. n.a.. Ζ0·2. Ζ0·2. Ζ0·7. Ζ0·4. In microfractures mg-number. —. —. 78–73. 77–64. 76–70. Cr2O3. —. —. Ζ0·2. Ζ0·2. Ζ0·2. Group IIHN mg-number. 79–78. 89–81. 81–77. 80–72. 84–77. Cr2O3. n.a.. Ζ0·4. Ζ0·2. Ζ0·6. Ζ0·4. mg-number = 100Mg/(Mg + Fet), in mols, where Fet is total iron. Mineral symbols after Kretz (1983). n.a., not analysed.. Fig. 2. Mineral modes of the xenoliths compared with other gabbroic xenoliths from subduction-related volcanoes. Group IICL xenoliths have similar modes to other xenoliths, whereas Group I and Group IIHN have less common compositions. Data sources: Mt. Pele´ e xenoliths (Fichaut et al., 1989), Mt. St. Helens (Heliker, 1995), Medicine Lake (Grove & Donnelly-Nolan, 1986), Calbuco (Hickey-Vargas et al., 1995) and Lesser Antilles [including quartz-bearing gabbros of Arculus & Wills (1980)]. Figure modified from Arculus & Wills (1980).. (2–3·4 wt %; Table 4 and Fig. 7) that are higher than those for most biotite or phlogopite reported in the literature. Such high sodium contents have been interpreted by Costa et al. (2001) as the result of (1) crystallization from a liquid with high Na2O contents, (2) crystal-chemical effects, as incorporation of Na in biotite or phlogopite is enhanced by high Mg/Fe of the mica (Volfinger et al., 1985), and (3) the presence of a. solvus between phlogopite and aspidolite (synonymous with the sodium phlogopite end-member; Rieder et al., 1998). On the basis of the data discussed above, we propose a two-stage crystallization sequence for these xenoliths: (1) Cr-spinel + olivine + clinopyroxene ± plagioclase crystallized from a water-bearing basaltic magma; (2) reaction occurred between the mafic minerals with an. 223.

(6) JOURNAL OF PETROLOGY. VOLUME 43. Fig. 3. Results of 40Ar/39Ar analyses of two hornblende separates (samples Hx14v and Hx14z) of Group IIHN xenoliths. Incremental furnace heating 40Ar/39Ar analyses were performed at the University of Geneva following the methods described by Singer & Pringle (1996). Both hornblendes show comparable, but discordant apparent age spectra. The presence of phlogopite in the mineral separate could explain the first high K/Ca steps, but afterwards the K/Ca remains low (0·04–0·05) and within the values of hornblende obtained by electron microprobe analyses (0·03–0·08). Determining the cause(s) of the discordant 39Ar release spectra is beyond the scope of this paper, but as the samples are xenoliths, heat from the host lava may have partially degassed the hornblendes. We interpret these 40Ar/39Ar data as indicating that the xenoliths are certainly older than 1 Ma and may be up to 8 Ma. The total fusion age is >5 Ma, which is >1 my younger than the age of the Huemul and Cerro Risco Bayo plutons that form the basement of the TSPC (6·2–6·4 Ma; Nelson et al., 1999).. evolved water-rich liquid that produced hornblende + orthopyroxene + phlogopite, accompanied or followed by plagioclase + apatite crystallization. The question of whether these reactions are due to progressive closedsystem crystallization or were triggered by ingress of an evolved water-rich liquid is addressed in the Discussion.. GROUP IICL xenoliths: clinopyroxene leuconorites with subsolidus textures The majority of these samples are characterized by mosaic, seriate textures, with subsolidus textural re-equilibration along grain boundaries between plagioclase crystals, and between plagioclase and pyroxenes (i.e. constant grain boundary dihedral angles between crystals; Hunter, 1987). Plagioclase is dominantly normally zoned, and there are compositional modes at An85−80 and An60–55. NUMBER 2. FEBRUARY 2002. (Fig. 5), which reflect that some samples have mainly anorthite-rich plagioclase (Hx14e, Hx14y) whereas in others, plagioclase is more albitic (Hx14a). Olivine (Fo81–72; NiO Ζ0·25 wt %) is commonly anhedral and surrounded by rims of orthopyroxene, Fe–Ti oxide symplectites and occasionally by small flakes of phlogopite, which we refer to as late phlogopite (Table 1). Rare Crspinel inclusions (Ulv0·23; Cr2O3 >18 wt %) are present in olivine. Anhedral to subhedral diopsidic to augitic clinopyroxene (Wo40–48En40–44Fs10–17) and orthopyroxene (mg-number 80–65) containing exsolution lamellae commonly occur as clusters interstitial to plagioclase crystals. Cr-poor magnetite (Ulv0·04–0·44) and ilmenite (Ilm0·74–0·96) are typically exsolved, and occur as euhedral inclusions in pyroxenes, or as anhedral oikocrysts surrounding anhedral plagioclase, pyroxenes and occasionally hornblende and phlogopite. Anhedral apatite is the only accessory mineral, and it occurs between grain boundaries of plagioclase crystals or inside phlogopite. Halogen contents and ratios in apatite are variable between samples (Table 5, Fig. 8), which suggests that they crystallized from melts with different halogen compositions (e.g. Boudreau, 1995). Subhedral to anhedral hornblende and phlogopite are late-crystallizing minerals, as they typically occur as small (<1 mm) poikilitic crystals surrounding resorbed, anhedral pyroxene, olivine and, in contrast to Group I xenoliths, partly resorbed plagioclase (Fig. 4c). Hornblende is commonly magnesiohastingite, but rare tschermakitic hornblende and magnesiohornblende are also present in sample Hx14y (Table 3). Hornblende and phlogopite have mg-numbers (Table 2) that overlap with those of pyroxene and olivine, whereas their Cr2O3 contents are typically higher (Fig. 6). As previously argued for Group I xenoliths, we interpret the high mg-numbers and Cr2O3 contents of the hydrous minerals as indications of reactions between Mg-rich, Cr-bearing minerals (olivine, pyroxenes and Cr-spinel) and water-rich evolved liquid, and not as evidence for crystallization from mafic magma. However, in contrast to Group I, plagioclase is resorbed in most xenoliths of this group. This could be explained by an increase in the water contents of the interstitial melt (e.g. Sisson & Grove, 1993). The Na2O contents of phlogopite are lower (>1–2 wt %; Table 4 and Fig. 7) than those of Group I xenoliths, but are none the less higher than in most biotite or phlogopite analyses reported in the literature (Costa et al., 2001). Low proportions (Ζ6·5 vol. %) of rhyolitic glass (SiO2 >71–74 wt %; K2O >5·7–6·7 wt %) are present exclusively along resorbed plagioclase–plagioclase, plagioclase–orthopyroxene and plagioclase–hornblende grain boundaries, indicating that glass is a secondary partial melting product that formed after xenolith entrainment rather than as a primary residual melt as is the case for Group I xenoliths.. 224.

(7) COSTA et al.. ´ N SAN PEDRO GABBROIC XENOLITHS, VOLCA. Fig. 4. Photomicrographs of the xenoliths. Group I: (a) vesiculated interstitial SiO2-rich glass (>66–72 wt %) in contact with euhedral orthopyroxene (Opx), hornblende (Hbl) and phlogopite (Phl). [Also note the olivine (Ol) crystal (Fo82) in contact with the silica-rich glass.] (b) Hbl enclosing resorbed Ol, but plagioclase (Pl) is euhedral. (Note also Phl surrounding anhedral Ol.) Group IICL: (c) poikilitic Hbl surrounding resorbed Pl crystals, many of which are bent or cracked (black arrows). This textural relation suggest that Hbl crystallized after deformation and dissolution of Pl. (d) Subvertical microfracture cutting across Pl and filled with Hbl, and Phl. (Note that where the microfracture intersects pyroxenes, larger Hbl crystals are present.) Group IIHN: (e) large poikilitic Hbl surrounding resorbed Pl and Ol, suggesting a reaction relation between the interstitial liquid, Pl and Ol to produce Hbl. The bent twins of the Pl crystal in the upper right corner of the picture (black arrow) indicate that reaction took place after deformation of the cumulate pile. (f ) Poikilitic Phl surrounding anhedral Ol and subhedral Pl.. Deformation and microfracturing Plagioclase commonly shows bent twins and microcracks (Fig. 4c), whereas hornblende and phlogopite do not display textural evidence of deformation other than kinkbands in phlogopite. In many samples, discontinuous microfractures (Ζ0·5 mm in width; Fig. 4d) filled with. 225. Fe–Ti oxides, hornblende and phlogopite (and occasionally also orthopyroxene) cut across all minerals except poikilitic hornblende and phlogopite. Where microfractures intersect pyroxene–plagioclase contacts, both minerals are resorbed, and anhedral, and they are mantled by a rim of hornblende or phlogopite (Fig..

(8) JOURNAL OF PETROLOGY. VOLUME 43. NUMBER 2. FEBRUARY 2002. Fig. 5. Histograms of plagioclase composition of the three xenolith groups. Plagioclase of Group I xenoliths that is not included in other minerals shows a bimodal composition, with cores at An85–75 and rims at An45–30. Those Pl included in Opx, Hbl and Phl lack the compositional gap and have similar compositions, suggesting that the three minerals co-crystallized. Most samples of Group IICL have Pl of An60–40 composition, except for samples Hx14y and Hx14e, which have Pl with higher An contents (>An88−80). Group IIHN xenoliths have mainly Pl with high An contents.. 4d). Orthopyroxene, hornblende and phlogopite filling microfractures have mg-numbers and Cr2O3 contents that overlap with or are lower than those of poikilitic minerals (Table 2). These compositional and textural observations suggest that the microfractures hosted evolved waterbearing melts or aqueous fluids that reacted with pyroxenes, olivine and plagioclase to produce the poikilitic hydrous minerals with mg-numbers and Cr2O3 contents that are higher than those hosted by microfractures.. GROUP IIHN xenoliths: hornblende norites with subsolidus textures Samples from this group are texturally heterogeneous and are characterized by large anhedral hornblende oikocrysts ([1 cm; Fig. 4e) that surround resorbed olivine (Fo79–78; NiO Ζ0·20 wt %), diopsidic to augitic clinopyroxene (Wo42–48En45–47Fs6–11) and orthopyroxene (mgnumber 81–77). Plagioclase (An88−80, occasionally An50) is also resorbed when it occurs as inclusions in hornblende. Oxide minerals are Cr-spinel (Ulv0·2; Cr2O3 >9 wt %),. magnetite (Ulv0·07–0·76) and ilmenite (Ilm0·89). Phlogopite is also present as oikocrysts that include resorbed olivine, orthopyroxene and plagioclase, and occasionally it also occurs inside hornblende oikocrysts (Fig. 4f ). Rare anhedral apatite is present along plagioclase grain boundaries. Late-crystallized hornblende (magnesiohastingite) and phlogopite have mg-numbers and Cr2O3 contents (Table 2) that overlap with, or are higher than those of olivine and pyroxenes (Fig. 6). As in Group IICL xenoliths, resorbed plagioclase inside hornblende could be explained by an increase in the water content of the interstitial melt before or during hornblende (and phlogopite) crystallization-reaction. Phlogopite has extremely high Na2O (>1·5–5 wt %), approaching the composition of aspidolite (Table 4 and Fig. 7). Plagioclase and orthopyroxene inside hornblende oikocrysts are deformed (Fig. 4e), whereas hornblende oikocrysts do not show textural evidence of deformation, and phlogopite occasionally shows kink bands. From these textural relations, we infer that deformation occurred before crystallization of the hydrous minerals.. 226.

(9) COSTA et al.. ´ N SAN PEDRO GABBROIC XENOLITHS, VOLCA. Fig. 6. Concentrations of Cr2O3 wt % of mafic minerals from the three groups of xenoliths. It should be noted that the Cr2O3 contents of Hbl, Phl, and occasionally Opx (only in Group I xenoliths) are as high as or higher than those of clinopyroxene (Cpx).. Fig. 7. Phlogopite composition (atoms per formula unit) from the three groups of xenoliths are characterized by high Na/K. (See text for discussion.) For comparison are also shown the biotite compositions of the host dacite lava, and those of low-pressure (Ζ0·3 GPa) water-bearing crystallization experiments of basaltic to andesitic composition. Data sources: SG, 1993 is Sisson & Grove (1993); RC, 1996 is Righter & Carmichael (1996).. 227.

(10) JOURNAL OF PETROLOGY. VOLUME 43. NUMBER 2. FEBRUARY 2002. Table 3: Representative hornblende analyses Group I. Group IICL. Group IIHN. wt %. n-1-17. n-III-2-5. 14a-VII-1p. e-III-1-5p. w-VIII-1p. y-I-1p. y-V-4v. w-IX-3v. v-II-7. z-I-3. 42·02. SiO2. 43·61. 43·31. 42·67. 45·10. 42·50. 42·47. 43·26. 42·66. 42·40. TiO2. 2·22. 3·54. 3·50. 1·03. 3·34. 1·99. 0·17. 2·91. 3·47. 3·20. Al2O3. 11·32. 11·13. 10·41. 10·60. 10·88. 11·82. 10·93. 9·83. 12·63. 12·13. Cr2O3. 0·64. 0·06. 0·01. 0·41. 0·00. 0·36. 0·04. 0·10. 0·27. 0·25. FeO∗. 9·31. 8·99. 13·51. 10·11. 12·75. 11·71. 10·95. 11·94. 9·35. 9·17. MnO. 0·15. 0·08. 0·22. 0·05. 0·17. 0·24. 0·31. 0·16. 0·10. 0·17. MgO. 15·96. 15·91. 13·54. 16·71. 13·85. 14·62. 16·08. 15·07. 15·46. 15·97. CaO. 11·60. 11·63. 11·37. 11·01. 11·66. 11·46. 11·50. 11·22. 11·26. 11·20. Na2O. 2·81. 2·97. 1·98. 2·53. 2·15. 2·29. 2·10. 2·29. 3·00. 2·92. K2O. 0·36. 0·34. 0·89. 0·27. 0·65. 0·43. 0·44. 0·64. 0·42. F. 0·09. 0·10. 0·10. 0·08. 0·00. 0·00. 0·00. 0·00. n.a.. 0·10. Cl. 0·02. 0·02. 0·05. 0·13. 0·09. 0·10. 0·13. 0·20. n.a.. 0·08. F=O. 0·04. 0·04. 0·04. 0·04. 0·00. 0·00. 0·00. 0·00. —. 0·05. Cl=O. 0·01. 0·00. 0·01. 0·03. 0·02. 0·02. 0·03. 0·04. —. Tot. 98·08. 98·06. 98·20. 97·96. 98·01. 97·46. 95·87. 96·98. 98·36. 97·58. Si. 6·24. 6·22. 6·22. 6·37. 6·19. 6·14. 6·27. 6·22. 6·05. 6·03. Al. 1·91. 1·88. 1·79. 1·76. 1·87. 2·01. 1·87. 1·69. 2·12. 2·05. Cr. 0·07. 0·01. 0·00. 0·05. 0·00. 0·04. 0·00. 0·01. 0·03. 0·03. Ti. 0·24. 0·38. 0·38. 0·11. 0·37. 0·22. 0·02. 0·32. 0·37. 0·35. Fe3+c. 0·65. 0·43. 0·74. 1·17. 0·66. 0·97. 1·31. 0·94. 0·64. 0·82. Mg. 3·41. 3·42. 2·94. 3·52. 3·01. 3·15. 3·48. 3·28. 3·29. 3·42. Fe2+c. 0·47. 0·65. 0·91. 0·03. 0·89. 0·44. 0·02. 0·52. 0·47. 0·28. Mn. 0·02. 0·01. 0·03. 0·01. 0·02. 0·03. 0·04. 0·02. 0·01. 0·02. Ca. 1·78. 1·79. 1·77. 1·67. 1·82. 1·77. 1·79. 1·75. 1·72. 1·72. Na. 0·78. 0·83. 0·56. 0·69. 0·61. 0·64. 0·59. 0·65. 0·83. 0·81. K. 0·06. 0·06. 0·16. 0·05. 0·12. 0·08. 0·08. 0·12. 0·08. 0·08. mg-no.. 75·4. 76·0. 64·1. 74·7. 65·9. 69·0. 72·4. 69·2. 74·7. 0·43. 0·02. 75·6. ∗Total iron as Fe2+. The first letter or number of the analysis label indicates the sample. p, poikilitic mineral; v, mineral in microfractures; n.a., not analysed; c, calculated. Structural formula (23 oxygens and OH + F + Cl = 2, and cations = 13 – Na – Ca – K) calculated as in Leake et al. (1997). mg-number = 100Mg/(Mg + Fet), in mols, where Fet is total iron.. WHOLE-ROCK CHEMICAL COMPOSITIONS. GROUP I xenoliths: olivine–hornblende norites and melanorites. It is well established that the bulk-rock compositions of many plutonic rocks are not representative of liquids, but are the result of varying degrees of mineral accumulation or/and melt ± fluid migration (e.g. McBirney, 1995). In the following sections we describe the major, minor and trace element compositions of the xenoliths with respect to the mean composition of 10 basalts that we think are representative of mafic liquids of the TSPC (see the Appendix for the analytical methods and precision of the analyses).. Compared with the mean basaltic composition of the TSPC (Table 6), the xenoliths have higher MgO (>20– 21 wt %) and Ni (446–643 ppm), and lower SiO2 (>46– 47 wt %) and incompatible elements (e.g. K2O, Zr) (Fig. 9). This, together with the high modal proportions of olivine, suggests that the low incompatible element abundances are mainly due to accumulation of olivine. Ratios of elements that are not affected by olivine accumulation, such as P/Zr (9–10), or Rb/Y (1·1–1·5) fall within the range of TSPC basalts (Fig. 10).. 228.

(11) COSTA et al.. ´ N SAN PEDRO GABBROIC XENOLITHS, VOLCA. Table 4: Representative phlogopite analyses Group I. Group IICL. Group IIHN. wt %. b-tr1-1. k-tr2-6. n-5-1. a-VI-6-1p. e-I-3p. y-V-1p. w-V-5p. a-VI-2-1v. w-IV-6v. v-III-8. z-II-3. 42·02. SiO2. 37·16. 38·92. 37·80. 39·40. 36·87. 36·90. 36·42. 38·77. 37·08. 42·40. TiO2. 2·56. 1·05. 1·49. 4·05. 0·73. 1·30. 4·78. 3·70. 5·11. 1·94. 2·35. Al2O3. 16·13. 16·46. 16·30. 13·27. 16·18. 16·11. 14·63. 13·90. 13·83. 17·32. 16·30. Cr2O3. 0·08. 0·33. 0·05. 0·12. 0·29. 0·08. 0·04. 0·06. 0·08. 0·19. FeO∗. 10·11. 9·04. 9·51. 12·09. 9·23. 10·63. 11·21. 12·69. 10·76. 8·35. 9·02. MnO. 0·02. 0·08. 0·06. 0·09. 0·09. 0·08. 0·18. 0·17. 0·09. 0·06. 0·08. MgO. 19·68. 21·55. 20·30. 18·09. 21·17. 19·65. 17·63. 17·82. 17·86. 20·89. 20·56. n.a.. CaO. 0·03. 0·02. 0·04. 0·01. 0·04. 0·02. 0·00. 0·01. 0·11. 0·03. 0·00. Na2O. 2·21. 3·36. 2·58. 1·24. 1·92. 1·13. 1·36. 1·23. 1·21. 4·08. 1·94. K2O. 7·53. 5·70. 7·03. 8·51. 8·25. 8·98. 8·63. 8·54. 8·26. 5·06. 7·83. BaO. 0·47. 0·25. 0·30. 0·16. 0·07. 0·28. 0·47. 0·18. 0·21. 0·37. 0·14. F. 0·18. 0·06. 0·13. 0·16. 0·00. 0·00. 0·00. 0·10. 0·38. 0·13. 0·14. Cl. 0·02. 0·05. 0·07. 0·12. 0·08. 0·14. 0·20. 0·09. 0·17. 0·06. 0·11. F=O. 0·08. 0·03. 0·05. 0·07. 0·00. 0·00. 0·00. 0·04. 0·16. 0·05. 0·06. Cl=O. 0·00. 0·01. 0·02. 0·03. 0·02. 0·03. 0·04. 0·02. 0·04. 0·01. 0·02. Tot. 96·10. 96·83. 95·58. 97·14. 94·72. 95·50. 95·54. 97·18. 94·93. 96·92. 97·29. Si. 5·39. 5·50. 5·47. 5·69. 5·40. 5·42. 5·38. 5·62. 5·49. 5·43. 5·49. Ti. 0·28. 0·11. 0·16. 0·44. 0·08. 0·14. 0·53. 0·40. 0·57. 0·21. 0·25. Al. 2·75. 2·74. 2·78. 2·26. 2·79. 2·79. 2·55. 2·37. 2·41. 2·87. 2·72. Cr. 0·01. 0·04. 0·01. 0·01. 0·03. 0·01. 0·00. 0·01. 0·01. 0·02. Fe∗. 1·23. 1·07. 1·15. 1·46. 1·13. 1·31. 1·39. 1·54. 1·33. 0·98. 1·07. Mn. 0·00. 0·00. 0·01. 0·01. 0·01. 0·01. 0·02. 0·02. 0·01. 0·00. 0·00. Mg. 4·25. 4·54. 4·38. 3·90. 4·63. 4·30. 3·89. 3·85. 3·94. 4·38. 4·35. Ca. 0·00. 0·00. 0·01. 0·00. 0·01. 0·00. 0·00. 0·00. 0·02. 0·00. 0·00. Na. 0·62. 0·92. 0·72. 0·35. 0·54. 0·32. 0·39. 0·35. 0·35. 1·11. 0·53. K. 1·39. 1·03. 1·30. 1·57. 1·54. 1·68. 1·63. 1·58. 1·56. 0·91. 1·42. Ba. 0·03. 0·01. 0·02. 0·01. 0·00. 0·02. 0·03. 0·01. 0·01. 0·02. 0·01. t. Mg/Fe. mg-no. Na/K. 3·47 77·6 0·45. 4·25 81·0 0·90. —. 3·81 79·2 0·56. 2·67 72·7. 4·09. 3·30. 80·3. 0·22. 76·7. 0·35. 0·19. 2·80 73·7 0·24. 2·50 71·5 0·22. 2·96 74·7 0·22. 4·46 81·7 1·23. 4·06 80·3 0·38. ∗Total iron as Fe2+. The first letter or number of the analysis label indicates the sample. p, poikilitic mineral; v, mineral in microfractures; n.a., not analysed. Structural formula calculated with 22 oxygens. mg-number = 100Mg/(Mg + Fet), in mols, where Fet is total iron.. GROUP IICL xenoliths: clinopyroxene leuconorites with subsolidus textures The major element compositions of most of these xenoliths are comparable with those of high-alumina basalts typical of subduction zones (e.g. Gust & Perfit, 1987), with >49–52 wt % SiO2, >5·2–8·2 wt % MgO and >17·2–19 wt % Al2O3 (Table 6). Samples Hx14e, Hx14h and Hx14y stand out from the rest by their lower. concentrations of SiO2 (>46–47 wt %), and generally higher MgO (>7·6–10·7 wt %) and Al2O3 (>21·8– 23 wt %) (Fig. 9). Minor and trace element abundances are highly variable. Most xenoliths have concentrations of incompatible elements (e.g. K2O, Zr, Y) that range from those of the TSPC basalts to much lower values (Fig. 9). Concentrations of compatible elements (e.g. Sr and Ni) of most xenoliths are, however, within the ranges defined by TSPC basalts (Fig. 9), and thus their low. 229.

(12) JOURNAL OF PETROLOGY. VOLUME 43. NUMBER 2. FEBRUARY 2002. Table 5: Representative apatite analyses Group IICL. Group IIHN. wt %. 12a-1c. 14a-1c. e-1c. 4w-1c. y-4c. z-1c. z-4r-5p. SiO2. 0·10. 0·14. 0·05. 0·25. 0·05. 0·05. FeO∗. 0·66. 0·57. 0·63. 0·75. 0·43. 0·42. 0·13 0·71. CaO. 53·83. 53·79. 53·47. 53·51. 53·71. 53·94. 52·61. Ce2O3. 0·41. 0·21. 0·11. 0·27. 0·10. 0·11. 0·07. Na2O. 0·06. 0·13. 0·16. 0·23. 0·21. 0·20. 0·50. P2O5. 41·03. 41·61. 41·67. 41·41. 41·94. 42·02. 41·92. SO3. 0·00. 0·00. 0·00. 0·17. 0·06. 0·06. 0·10. MnO. 0·13. 0·08. 0·08. 0·09. 0·09. 0·10. 0·03. SrO. 0·00. 0·00. 0·00. 0·00. 0·00. 0·00. 0·00. F. 1·42. 0·79. 0·45. 0·55. 0·19. 0·54. 0·18. Cl. 1·50. 1·55. 2·29. 2·63. 2·17. 1·90. 2·77. F=O. 0·60. 0·33. 0·19. 0·23. 0·08. 0·23. 0·08. Cl=O. 0·34. 0·35. 0·52. 0·59. 0·49. 0·43. 0·63. 98·29. 98·16. 98·19. 99·03. 98·39. 98·67. 99·33. Tot Si. 0·03. 0·02. 0·01. 0·04. 0·01. 0·01. 0·02. Fe∗. 0·09. 0·08. 0·09. 0·11. 0·06. 0·06. 0·10. Ca. 9·77. 9·73. 9·70. 9·65. 9·70. 9·70. 9·52. Ce. 0·03. 0·01. 0·01. 0·02. 0·01. 0·01. 0·00. Na. 0·02. 0·04. 0·05. 0·07. 0·07. 0·06. 0·16. P. 5·88. 5·95. 5·97. 5·90. 5·99. 5·97. 5·99. S. 0·00. 0·00. 0·00. 0·02. 0·01. 0·01. 0·01. Mn. 0·02. 0·01. 0·01. 0·02. 0·01. 0·01. 0·00. Sr. 0·00. 0·00. 0·00. 0·00. 0·00. 0·00. 0·00. F. 0·38. 0·21. 0·12. 0·15. 0·05. 0·14. 0·05. Cl. 0·22. 0·22. 0·33. 0·38. 0·31. 0·27. 0·40. OHc. 0·40. 0·57. 0·55. 0·48. 0·64. 0·59. 0·55. Total. 16·84. 16·85. 16·83. 16·83. 16·85. 16·84. 16·82. Cl/F. 0·57. 1·05. 2·73. 2·57. 6·18. 1·87. 8·17. ∗Total iron as Fe2+. The first letter or number of the analysis label indicates the sample. Structural formula calculated with 25 O, OH, F, Cl. c, calculated.. incompatible element concentrations cannot be solely explained by plagioclase or olivine accumulation, and suggest that loss of interstitial melt rich in incompatible elements is a more plausible explanation. Some xenoliths show a positive Eu anomaly (Eu/Eu∗ up to 1·75) when normalized to primitive mantle (McDonough et al., 1992). This could be due to plagioclase accumulation, although it could also occur if the parent liquid initially had a positive Eu anomaly or, as will be discussed later, by loss of interstitial liquid with a negative Eu anomaly. Samples Hx14e, Hx14h and Hx14y also have low concentrations. of some incompatible elements (e.g. Y, Zr), although K2O (0·34–2·48 wt %) and Rb (10–80 ppm) are highly variable. Their Ni and Sr concentrations are within the values or higher than those of the TSPC basalts, and thus accumulation of olivine + plagioclase may be a contributing factor for the low abundances of some incompatible elements in these three xenoliths. Ratios of incompatible elements (e.g. K/P, P/Zr, Rb/Y) of all xenoliths of this group are highly variable, ranging from much higher to lower than those of the TSPC basalts (Fig. 10). Such a large range of incompatible element. 230.

(13) COSTA et al.. ´ N SAN PEDRO GABBROIC XENOLITHS, VOLCA. Fig. 8. Apatite halogen compositions of Group II xenoliths. Apatite from some samples has subequal contents of Cl (1·4–1·6 wt %) and F (1·4–1·7 wt %), whereas that from other samples has higher Cl (1·4–2·9 wt %) than F (0·2–0·9 wt %). The field of layered intrusions includes analyses from Skaergaard, Jimberlana, Dufek, Munni Munni, Penikat, Great Dyke, Mt. Thirsty, Ora Banda Sill, Windimurra, Stillwater and Bushveld (not below major platinum group element bearing zones). References have been given by Boudreau (1995). Figure redrawn from Boudreau (1995). OHc indicates calculated OH from structural formula.. ratios further indicates that processes other than mineral accumulation (e.g. migration of interstitial melt and aqueous fluids) have contributed to bulk compositional variations among the xenoliths.. GROUP IIHN xenoliths: hornblende norites with subsolidus textures Compared with the mean basaltic composition of the TSPC these samples have low concentrations of SiO2 (>45 wt %) and incompatible elements (e.g. K2O, Zr), and high MgO (>13·3–16·5 wt %), Ni (226–335 ppm) and Ca/Na values (5·2–6·3), and thus olivine and plagioclase accumulation could partly explain their low incompatible element abundances (Fig. 9). However, ratios of incompatible elements such as K/P (7–25) and Rb/ Y (2–3) range from those of the TSPC basalts to higher (Fig. 10), suggesting that apart from mineral accumulation, migration of interstitial melt and fluids are important processes for understanding the petrogenesis of these xenoliths.. DISCUSSION Melt migration and reaction in Group I xenoliths Textural relations and compositions of orthopyroxene, hornblende and phlogopite (e.g. high mg-number and. Cr2O3 contents) indicate that they are probably the result of reactions between early-crystallized mafic minerals (Cr-spinel, olivine and clinopyroxene) and an evolved, water-rich liquid. Reactions between clinopyroxene or olivine and liquid to produce hornblende, and between olivine and liquid to produce orthopyroxene have been reported in low-pressure (Ζ0·3 GPa) crystallization experiments using basaltic to andesitic starting compositions (e.g. Holloway & Burnham, 1972; Heltz, 1973; Sisson & Grove, 1993; Grove et al., 1997; Moore & Carmichael, 1998). However, co-crystallization of hornblende, orthopyroxene and phlogopite in reaction relationship with olivine or clinopyroxene has never been reported. In particular, the high Na2O contents of the phlogopite are an uncommon compositional feature, which points to a more complex differentiation history for these xenoliths (Costa et al., 2001). Recent experimental work (Prouteau et al., 2001; Costa et al., in preparation) involving interactions between evolved water-rich liquids and mafic minerals (e.g. forsteritic olivine) have co-crystallized hornblende, orthopyroxene and phlogopite as reaction products, and thus an alternative possibility to closed-system crystallization is that these minerals in the San Pedro xenoliths are due to ingress of a differentiated melt (e.g. dacitic) by displacement of the mafic interstitial liquid in equilibrium with olivine, clinopyroxene and highanorthite plagioclase. This process could explain several textural and mineralogical features of the xenoliths such as the compositional gap between plagioclase cores (An85–70) and rims (An45–20) [a detailed discussion of trace element composition of plagioclase has been given by Costa (2000)], and the coexistence of forsteritic olivine and rhyolitic glass, both of which are atypical of closedsystem crystallization (e.g. Brophy et al., 1996), and are more characteristic of mixing or mingling between felsic and mafic magmas (e.g. Feeley & Dungan, 1996). In the next section we derive some constraints on the amount and composition of the reacting liquid.. Mass-balance constraints on the amount and composition of the reactive melt To test the hypothesis of melt migration, we have undertaken least-squares mass-balance calculations using the composition of the cumulus minerals (high-anorthite plagioclase, clinopyroxene, Cr-spinel and olivine) and a representative dacitic composition of Volca´ n San Pedro as an analogue for the replacive interstitial melt. The results show that (Table 7): (1) the residuals (R2) are low (<1) and thus do not preclude the melt migration hypothesis; (2) before reaction the three xenoliths consisted of large amounts of olivine (42–45 wt %), so that for sample Hx14n the calculated amount of olivine before reaction is more than twice the observed amount; (3) the amount of clinopyroxene that was consumed. 231.

(14) JOURNAL OF PETROLOGY. VOLUME 43. NUMBER 2. FEBRUARY 2002. Table 6: Whole-rock major and trace element analyses Group I: olivine norites. Group IICL: Cpx and Hbl leuconorites. Hx14b. Hx12a. Hx12b. Hx14a. Hx14c. Hx14d. Hx14e. Hx14h. Hx14i. Hx14j. 52·0. 51·7. 50·9. 52·2. 52·3. 46·2. 46·9. 51·1. 50·4. Hx14k. Hx14n. X-ray fluorescence (wt %) SiO2. 46·4. 45·8. 47·3. TiO2. 0·54. 0·64. 0·58. 1·30. 0·81. 1·00. 0·47. 0·91. 0·26. 0·70. 1·06. 0·92. Al2O3. 11·72. 10·63. 10·94. 18·17. 18·02. 17·75. 18·90. 18·96. 23·41. 21·59. 18·86. 17·30. Fe2O3∗. 13·25. 13·10. 11·73. 9·72. 8·93. 9·78. 7·72. 8·85. 6·54. 8·91. 8·95. 9·95. 0·20. 0·19. 0·18. 0·16. 0·15. 0·16. 0·13. 0·16. 0·10. 0·11. 0·15. 0·20. MnO MgO. 5·47. 7·26. 7·09. 7·95. 5·21. 9·74. 7·64. 5·91. 8·17. CaO. 19·6 5·26. 21·4 5·07. 20·8 5·44. 8·77. 9·10. 9·20. 8·17. 9·26. 11·59. 10·24. 10·16. 10·40. Na2O. 2·16. 2·06. 1·67. 3·63. 3·57. 3·25. 3·11. 3·72. 1·57. 2·25. 3·16. 2·62. K2O. 0·56. 0·52. 0·71. 0·72. 0·49. 0·29. 0·78. 0·37. 0·34. 0·57. 0·35. 0·23. P2O5. 0·13. 0·13. 0·11. 0·35. 0·12. 0·19. 0·12. 0·24. 0·08. 0·09. 0·09. Sum. 99·9. 99·6. 99·5. 100·3. 100·2. 99·6. 99·6. 100·0. 99·8. 99·0. 99·8. 0·04 100·2. X-ray fluorescence (ppm) Nb Zr. 3·0. 3·0. 56. 2·0. 4·9. 2·0. 56. 53. 84. 50. 2·7. 0·9. 0·2. 35. 11. 14 616. 381. 333. 316. 603. 692. 647. 725. 734. 685. Zn. 106. 109. 94. 83. 78. 85. 57. 74. 65. 81. Ni. 446. 643. 621. 39. 107. 21. 161. 133. 46. 64. Cr. 707. 1065. 1308. 63. 160. 72. 189. 213. 67. 219. V. 109. 109. 142. 195. 188. 157. 46. 193. 265. 214. Ce. 19. 12. 9. 33. 19. 23. 12. 14. 6. 13. Ba. 181. 167. 186. 350. 165. 245. 106. 222. 188. 123. La. 8. 3. 6. 15. 8. 3. 6. 3. Y. 8. 9. 17·8. 4. 4·3. Rb. 12. 10. 14·8. Ga. 11. 11. 20. 7 10·6. 5·3. 3·8. 27·4. 9·9 17. 672. 1·5. 34. Sr. 10·5. 675. 1·7. 38. 19. 17. 11·5 6·2 20. 9·5 15. 3 7·5. 14 19. 3 17. ICP-AES (ppm) La. 6·4. 6·5. 14·3. 6·0. 9·5. 4·1. 5·4. 4·9. 3·4. Ce. 15·5. 12·4. 34·8. 15·2. 22·5. 10·0. 11·1. 10·2. 7·0. Pr. 2·1. 1·6. 4·5. 2·1. 3·0. 1·3. Nd. 7·6. 7·1. 19·0. 8·7. 12·1. 4·8. 4·7. 6·1. 4·6. Sm. 2·1. 1·9. 4·7. 2·6. 3·2. 1·2. 1·2. 1·6. 1·3. Eu. 0·63. 0·59. 1·39. 1·04. 1·32. 0·42. 0·57. Gd. 1·5. 1·7. 3·7. 2·3. 2·5. 1·0. Dy. 1·4. 1·6. 3·1. 1·9. 2·1. Ho. 0·26. 0·35. 0·60. 0·41. 0·38. Er. 0·8. 0·9. 1·4. 1·1. 0·9. b.d.. b.d.. b.d.. Tm. 0·11. 0·13. 0·20. 0·15. 0·13. b.d.. b.d.. b.d.. Yb. 0·6. 0·8. 1·2. 0·8. 0·8. b.d.. b.d.. b.d.. Lu. 0·10. 0·11. 0·18. 0·11. 0·11. b.d.. b.d.. 3·14. b.d. 0·18. b.d.. b.d.. b.d.. 0·85. 0·80. b.d.. 1·5. 1·5. b.d.. 1·6. 1·4. b.d.. 0·33. 0·34. 0·10. b.d. 0·13 b.d. 0·11. mol Ca/Na. 2·35. 2·37. K/P. 8·2. 7·6. P/Zr. 10·1. 10·1. Rb/Y. 1·5. 1·1. Eu/Eu∗. 1·03. —. 2·33. 2·46. 2·73. 12·3. 3·9. 7·8. 2·9. 9·1. 18·2. 10·5. 0·8. 0·9. — 0·98. 0·98. —. —. 2·53 12·4 —. 0·4 1·27. 232. 5·2 —. 2·40. 7·11. 2·9. 8·1. 12·0. 4·38. 27·6. 10·3. 11·2. 0·5. 2·4. 3·3. 1·38. 1·14. —. 3·10. 3·83. 7·4. 10·9. 35·7. 12·5. —. 0·4 1·65. 1·75.

(15) COSTA et al.. ´ N SAN PEDRO GABBROIC XENOLITHS, VOLCA. Group IICL: Cpx and Hbl leuconorites Hx14l. Hx14m. Group IIHN. Basalts of TSPC. Hx14q. Hx14s. Hx14u. Hx14w. Hx14y. Hx14x. Hx14v. Hx14z. Mean. 49·4. 49·2. 51·7. 49·4. 46·1. 49·1. 44·9. 45·8. 50·7. 2. X-ray fluorescence (wt %) SiO2. 49·6. TiO2. 1·32. 48·9. 0·6. 1·20. 0·78. 0·87. 0·94. 1·31. 0·12. 1·41. 0·28. 0·34. 1·92. 0·05 0·88. Al2O3. 18·29. 19·04. 18·16. 18·19. 17·66. 18·31. 21·76. 18·98. 17·43. 19·50. 17·75. Fe2O3∗. 10·68. 10·90. 9·80. 9·60. 9·60. 10·74. 6·87. 10·04. 9·78. 9·50. 9·30. 0·34. MnO. 0·16. 0·15. 0·16. 0·18. 0·16. 0·16. 0·11. 0·18. 0·14. 0·14. 0·15. 0·02. MgO. 6·07. 5·70. 8·24. 7·98. 7·59. 6·11. 10·75. 5·19. 16·45. 13·31. 8·15. 0·60. CaO. 9·76. 10·57. 10·09. 9·90. 8·94. 9·79. 10·57. 9·69. 8·99. 9·24. 8·86. 0·69. Na2O. 3·45. 3·38. 2·88. 3·00. 3·40. 3·53. 1·33. 4·10. 1·37. 1·72. 3·27. 0·13. K2O. 0·69. 0·41. 0·80. 0·57. 0·37. 0·68. 2·48. 0·52. 0·26. 0·47. 0·85. 0·14. P2O5. 0·22. 0·15. 0·02. 0·12. 0·13. 0·22. 0·06. 0·31. 0·07. 0·04. 0·17. 0·05. Sum. 100·3. 100·4. 100·3. 99·6. 100·5. 100·2. 100·2. 99·5. 99·7. 100·1. 100·1. 0·17. X-ray fluorescence (ppm) Nb. 4·5. 2·4. 1·0. 1·2. 1·6. 2·1. 1·0. 3·9. 1·0. 1·4. 2·7. 1·1. Zr. 79. 42. 7. 20. 32. 40. 23. 70. 30. 28. 83. 28. Sr. 562. 633. 594. 591. 592. 868. 586. 693. 573. 632. 614. 86. Zn. 78. 75. 68. 84. 72. 32. 75. 88. 73. 78. 77. 7. Ni. 45. 34. 95. 78. 91. 36. 237. 21. 335. 226. 109. 31. Cr. 95. 69. 212. 229. 204. 76. 161. 51. 252. 305. 252. 103. V. 234. 244. 208. 212. 154. 58. 33. 242. 57. 91. 197. 21. Ce. 25. 17. 7. 11. 18. 12. 9. 35. 13. 10. 23. 7. Ba. 213. 178. 171. 171. 247. 180. 206. 213. 88. 140. 233. 43. 4. 4. 13. 2. 4. 11. 3. 6·7. 2·9. 23·4. 3·5. 4·2. 14·9. 1·6. 11. 79·5. 7·2. 7·2. 14·3. 16·1. 6·3. 19. 14. 15. 19. 1. La. 12. Y. 17·6. Rb. 15·4. Ga. 20. 6. 1. 3. 4. 6·6. 8. 10·6. 39·5. 14·1. 6·4. 18. 17. 12 6·1 20. 20. 21. 13. ICP-AES (ppm) La. 7·5. 3·3. Ce. 15·5. 6·7. Pr. 1·9. Nd. 9·5. 3·6. 4·1. Sm. 2·7. 1·0. 1·2. Eu. 0·88. 0·36. Gd. 2·3. b.d.. b.d.. Dy. 2·10. b.d.. b.d. Ho. 0·37. Er. 1·0. b.d.. b.d. Tm. 0·14. b.d.. b.d. Yb. 0·9. b.d.. b.d. Lu. 0·13. b.d.. b.d. b.d.. 4·2 7·9 b.d.. 0·45. 0·16. 0·18. mol Ca/Na. 2·73. 3·01. 3·18. 2·53. 2·67. 2·28. 6·32. K/P. 6·0. 5·2. 76·1. 9·0. 5·4. 5·9. 78·6. 3·2. 7·1. 22·4. 9·7. 3·1. P/Zr. 12·2. 15·6. 12·5. 26·2. 17·7. 24·0. 11·4. 19·3. 10·2. 6·2. 8·7. 3·8. 0·9. 0·5. Rb/Y Eu/Eu∗. —. 1·05. 3·38. 6·0 —. 1·8 —. 0·6 —. 1·6. 7·66. 27·4. —. —. 0·3 —. 2·1 —. ∗Total iron as Fe3+. b.d., below determination. (For methods and precision of the analyses, see the Appendix.). 233. 5·18. 2·6. 3·4 —. 0·2. 1·1 —. 0·4 —.

(16) JOURNAL OF PETROLOGY. VOLUME 43. NUMBER 2. FEBRUARY 2002. Fig. 9. Major and trace element variation diagrams of the three groups of xenoliths. Also shown is the mean composition of 10 TSPC basalts. Black arrows indicate the effects of accumulation of Pl and Ol to the TSPC composition. Ni concentration is that of Ol analysis n-V-1c. The Sr concentration of An88 and An60 is taken from the isotope dilution analyses of Pl from samples Hx14v (Sr 1130 ppm) and Hx14a (Sr 1030 ppm), respectively. (See text for discussion.). (4·5–6·5 wt %) is also significant; (4) the quantity of reactive liquid is almost the same for all three xenoliths (30–33 wt %). As an alternative means of obtaining a rough estimate of the amount of interstitial melt that could have been displaced, we have used the proportions of post-cumulus minerals (hornblende, orthopyroxene, phlogopite, plus plagioclase rims) and glass present in the xenoliths. However, as hornblende, orthopyroxene and phlogopite are the products of reactions that consumed liquid and minerals, their observed modal abundances do not directly correspond to the porosity at the time of melt migration. As an approximation, the proportions of liquid and minerals that participated in the reactions were taken from the literature: for the hornblende reaction we have used the stoichiometry (wt %) determined by Sisson & Grove (1993): 100 Hbl = 22 Ol + 38 Cpx + 42 liquid. For the orthopyroxene reaction we have estimated the stoichiometry (wt %) suggested by Kelemen (1990): 100 Opx = 60 Ol + 40. liquid. No experimentally determined stoichiometry for the phlogopite reaction was found in the literature, so we assumed that the quantity of liquid consumed by this reaction equals the observed modal proportion of phlogopite. Lastly, we have estimated that plagioclase rims are one-third of the plagioclase in the xenoliths. The calculations (Table 8) give similar results to those obtained from mass-balance constraints, i.e. (1) large amounts of olivine and clinopyroxene were consumed by the reactions and (2) the amount of liquid that was consumed is 32–33 wt %. Although the calculations might not be very accurate because the stoichiometry of the reactions depends on the liquid composition, it illustrates how estimating the amount of ‘trapped melt’ or porosity in cumulate rocks by considering only the amount of post-cumulus minerals could be misleading if there are reactions involved. For example, if we take the amount of post-cumulus minerals and glass as representative of ‘trapped melt’ the values vary between 44. 234.

(17) COSTA et al.. ´ N SAN PEDRO GABBROIC XENOLITHS, VOLCA. Fig. 10. Variation diagrams of element ratios of the three groups of xenoliths. The legend is the same as in Fig. 9. (See text for discussion.). and 58 wt %, up to 1·8 times higher than the calculated amount (32–33 wt %).. Textural, mineralogical and compositional evidence for melt and fluid migration in GROUP II xenoliths The low concentrations of incompatible elements (e.g. Zr) of most xenoliths are not correlated with high concentrations of compatible elements (e.g. Sr, Ni), suggesting that loss of evolved interstitial liquid rather than mineral accumulation is responsible for the low incompatible element abundances. Only six samples have major and trace element abundances that could be partly explained by accumulation of plagioclase (Hx14w) or plagioclase + olivine (Hx14e, Hx14h, Hx14v, Hx14y and Hx14z) (Fig. 9). However, the large range of P/Zr values cannot simply be the result of melt loss from a crystal pile consisting of plagioclase, pyroxenes or olivine, as the partition coefficients of P and Zr in these minerals are <0·1 (e.g. Rollinson, 1993). Thus, we propose that expulsion of interstitial liquid occurred both before and after apatite crystallization. Samples with high P/Zr lost melt after apatite crystallization, whereas xenoliths with P/Zr values within the range of the TSPC basalts (but with low Zr and P2O5 concentrations) lost melt before apatite crystallization (Fig. 10). Xenoliths that lost melt before apatite crystallization are commonly those that show positive Eu anomalies, which suggests that the Eu. anomalies could be due to loss of interstitial liquid rich in REE with a negative Eu anomaly. The large ranges of ratios of incompatible elements involving Rb and K suggest that an aqueous fluid has also modified the bulk-rock composition of some xenoliths. For example, many samples have Rb/Y values similar to or lower than those of the TSPC basalts, whereas others have very high Rb/Y. As Y is highly compatible in apatite (e.g. Pearce & Norry, 1979), the low Rb/Y values of some xenoliths could be explained by loss of interstitial liquid after apatite crystallization. However, the high Rb/Y of other xenoliths suggests that they have gained Rb with respect to Y. Decoupling of K and Rb from the rest of incompatible elements can be produced by the involvement of an aqueous fluid phase, as fluid–melt partition coefficients of K and Rb are much higher than those of Y and Zr (e.g. Keppler, 1996). Accordingly, the high Rb and K2O concentrations and the high Rb/Y (or low P/Rb) of some xenoliths (Fig. 10) could be explained by the arrival of an aqueous fluid phase rich in alkalis (e.g. K, Rb, Na) that dissolved into the remaining melt. Further evidence for the arrival of a fluid is the variable halogen contents of apatite. Fluid addition to a melt can be recorded as high Cl/F in apatite (e.g. Boudreau & McCallum, 1989) because Cl tends to partition into the fluid phase, whereas F remains in the melt (e.g. Candela & Piccoli, 1995; Villemant & Boudon, 1999). Apatite from one xenolith (Hx12a) has low Cl/F (Fig. 8), whereas the rest have. 235.

(18) JOURNAL OF PETROLOGY. VOLUME 43. NUMBER 2. FEBRUARY 2002. Table 7: Least-squares mixing model using the cumulate mineralogy (Group I xenoliths) and a dacitic composition of Volca´ n San Pedro. SiO2. Hx14b. Calc. mixa. 47·1. 47·1. Residual. Hx14k. Calc. mix. 0·00. 45·9. 45·9. Residual. Hx14n. Calc. mix. 0·03. 47·5. 47·4. Residual. Dacite (H23). 0·03. 63·5. TiO2. 0·55. 0·36. 0·20. 0·64. 0·38. 0·26. 0·58. 0·34. 0·24. 0·64. Al2O3. 11·90. 11·88. 0·02. 10·66. 10·67. 0·00. 10·98. 10·94. 0·03. 16·70. FeO∗. 12·10. 11·88. 0·22. 13·14. 12·91. 0·24. 11·77. 11·61. 0·15. 5·00. MnO. 0·20. 0·17. 0·03. 0·19. 0·18. 0·01. 0·18. 0·17. 0·01. 0·09. MgO. 0·02. 2·28. CaO. 19·9 5·34. 19·9 5·24. 0·10. 5·09. 5·00. 0·09. 5·46. 5·42. 0·04. 4·79. Na2O. 2·19. 1·82. 0·37. 2·07. 1·64. 0·42. 1·68. 1·75. 0·07. 4·34. K2O. 0·57. 0·82. 0·25. 0·52. 0·75. 0·23. 0·71. 0·80. 0·09. 2·45. P2O5. 0·13. 0·06. 0·06. 0·13. 0·06. 0·07. 0·13. 0·06. 0·07. 0·19. Cr2O3. 0·10. 0·75. 0·64. 0·15. 0·93. 0·78. 0·19. 0·59. 0·40. 0·00. R2. 0·03. 21·5. 21·5. 0·05. 0·73. 20·9. 20·9. 0·98. 0·26. Cumulate minerals (wt %) Spl (n-1-27). 4·5. 5·5. 3·4. Fo83 (n-V-1c). 41·8. 44·8. 43·3. An80. 16·6. 13·5. 14·4. 3·9. 5·7. 6·5. Total of crystals. 66·8. 69·6. 67·6. Dacite (wt %). 33·2. 30·4. 32·4. Cpx (n-III-2r). Calc. mix is wt % of cumulate + wt % of dacite. ∗Total iron as Fe2+. R2, squared sum of the residuals; Spl, spinel; Fo, forsterite; An, anorthite; Cpx, clinopyroxene.. a. much higher Cl/F, suggesting that it has crystallized from melts enriched in Cl by fluid addition. Because it is not apparent from Fig. 9, it is worth noting that samples that lost melt before apatite crystallization also have apatite with high Cl/F (e.g. Hx14y and Hx14e), suggesting that fluid arrival might have post-dated melt migration. The bent laths and microcracks displayed by the plagioclase and the microfractures filled with hornblende, phlogopite, orthopyroxene and magnetite (Fig. 4c–e) are interpreted as textural evidence of expulsion of interstitial liquid by compaction of a crystal-rich magma. Microfractures commonly cut across bent plagioclase crystals, suggesting that rock deformation changed from ductile to brittle (e.g. Kronenberg & Shelton, 1980), or that the behaviour of plagioclase changed from plastic to cataclastic (e.g. Hacker & Christie, 1990). In many samples poikilitic hornblende contains plagioclase with bent twins (Fig. 4c and e). Thus, deformation of the crystal pile and expulsion of interstitial liquid seem to have occurred before arrival of aqueous fluids. Perhaps during initial ductile deformation of the crystal matrix, melt migrated through the pore spaces, and later, during. brittle deformation, melt migration occurred mostly through microfractures. The remaining interstitial melt was enriched by aqueous fluids (and alkalis), which caused dissolution of olivine, pyroxenes and plagioclase and crystallization of poikilitic hornblende and Na-rich phlogopite with high mg-numbers and Cr2O3. The available stable isotope analyses (sample Hx14h, whole rock: 18O = 5·4; sample Hx14v: hornblende, 18O = 5·3; bulk rock, D = −62, all values relative to SMOW; B. S. Singer, unpublished data, 1993) suggest that the fluids that fluxed the xenoliths were magmatic and not meteoric (e.g. Taylor & Forester, 1979).. 236. IMPLICATIONS OF THE PRESENCE OF HORNBLENDE AND PHLOGOPITE IN SUBDUCTIONRELATED GABBROIC ROCKS Most of San Pedro gabbroic xenoliths have significant proportions of hornblende and phlogopite, either as small.

(19) COSTA et al.. ´ N SAN PEDRO GABBROIC XENOLITHS, VOLCA. Table 8: Estimate of the mineral and liquid proportions consumed by reactions (Group I xenoliths; values are all wt %) Calculated mineralogy before Mode. Hx14b. reaction. Hx14k. Hx14n. Hx14b. Hx14k. Hx14n. Hx14b. Hx14k. Hx14n. Ol. 32·0. 24·7. 22·4. Liquid + Ol = Opx∗. 39·9. 42·5. 40·5. An80. 22·8. 14·8. 13·8. Ol. 5·3. 13·1. 10·7. Cpx. 4·4. 7·7. 12·5. liquid. 3·5. 8·7. 7·1. An80. 22·8. 14·8. 13·8. 1·4. 2·2. 0·4. 68·5. 67·2. 67·1. An40. 11·1. 7·2. 6·7. Opx. 8·8. 21·8. 17·8. Cpx. 0·0. 0·0. 0·3. Hbl. 12·2. 21·5. 33·9. Phl. 1·6. 7·3. 2·8. Spl. 1·4. 2·2. 0·4. 10·1. 0·5. 2·1. Glass. Ol. liquid + Cpx + Ol = Hbl†Spl Ol. 2·7. 4·7. 7·5. Cpx. 4·4. 7·7. 12·2. liquid. 5·1. 9·0. 14·2. Total of reacted minerals and liquid Ol Cpx Tot min liquid. 8·0. Total. Calculated amount of replaced melt An40. 11·1. 7·2. Phl. 1·6. 7·3. 2·8. 17·7. 21·3. 17·8. 18·1. liquid. 8·7. 6·7. 4·4. 7·7. 12·2. Glass. 10·1. 0·5. 2·1. 12·4. 25·5. 30·3. Total. 31·5. 32·8. 32·9. 8·7. 17·7. 21·3. ∗Opx reaction, 100 Opx=60 Ol + 40 liquid (in wt %). Estimated from Kelemen (1990). †Hbl reaction, 100 Hbl= 22 Ol + 36 Cpx + 42 liquid (in wt %). From Sisson & Grove (1993). Mineral symbols after Kretz (1983). (See text for discussion.). crystals filling microfractures, or as large poikilitic postcumulus crystals that can make up >50 vol. % of the rock. In this respect they are not unusual, and a survey of the literature shows that the majority of subductionrelated gabbroic xenoliths and plutons have important amounts of hornblende, and occasionally phlogopite. In some localities, at least one generation of hornblende is an early-crystallizing mineral (Ulmer et al., 1983, Adamello batholith, Italy; Sisson et al., 1996, hornblende gabbro sill, California), whereas in others it is a late phase in a reaction relation with other minerals (Smith et al., 1983, Peninsular Ranges batholith, California; Regan, 1985, Coastal batholith of Peru; Whalen, 1985, Uasilau–Yau Yau Intrusive Complex, New Britain; Himmelberg et al., 1987, Yakobi intrusion, Alaska; Beard & Day, 1988, Smartville Complex, California; DeBari & Coleman, 1989, Tonsina Complex, Alaska; Springer, 1989, Pine Hill Complex, California; Kepezhinskas et al., 1993, Kamchatka; DeBari, 1994, Fiambala´ intrusion, Argentina; Tepper, 1996, Chilliwack batholith, Washington; Roberts et al., 2000, Que´ rigut Complex, French Pyrenees). The high proportions of hydrous minerals in gabbroic plutons and xenolith suites is in marked contrast with the rare occurrences of hornblende phenocrysts in basalts or basaltic andesites in arc volcanoes world wide (Sigurdsson & Shepherd, 1974, Kick’em-Jenny Volcano, Lesser Antilles; Arculus, 1976, Grenada, Lesser Antilles;. Arculus et al., 1976, Bogoslof Volcano, Alaska; Luhr & Carmichael, 1985, Cerro la Pilita, Mexico; Peterson & Rose, 1985, Ayarza caldera, Guatemala; Rose, 1987, Santa Marıá Volcano, Guatemala). Apart from the possibility that the hydrous minerals are produced by reactions triggered by protracted closed-system crystallization, the petrological, mineralogical and geochemical characteristics of the San Pedro xenoliths suggest that the large proportions of hornblende with high mg-numbers and Cr2O3 in gabbroic rocks can be the result of reactions between early-crystallized refractory minerals (olivine, Cr-spinel, pyroxenes and plagioclase) and evolved melt ± aqueous fluids that percolate through crystal-rich mafic magmas, and do not necessarily imply hornblende crystallization from water-rich mafic magmas. These melts and aqueous fluids could be derived from within the cumulate pile of the intrusion, as has been documented for the Skaergaard intrusion (e.g. McBirney, 1995) or from broadly contemporaneous felsic magmas (e.g. Sha, 1995). Interactions between partially solidified mafic cumulate piles and invading water-rich silicic magmas may provide a means of stabilizing substantially larger quantities of amphibole and mica (e.g. >50% mode) in subvolcanic reservoirs or plutons than closedsystem evolution of either mafic or silicic magmas. Partial melting and recycling of these hydrous and alkali-enriched plutonic facies could have important implications. 237.