Contfib Mineral Petrol (1981) 77:337- 354 Contributions to Mineralogy and Petrology (~ Springer-Verlag 1981 Petrology and Geochemistry of the Island of Sarigan in the Mariana Arc; Calc-Alkaline Volcanism in an Oceanic Setting Arend Meijer and Mark Reagan Department of Geosciences, University of Arizona, Tucson, Arizona 85721, USA Abstract. Isotopic studies of rocks from oceanic island arcs such as the Marianas indicate that little, if any, recycling of continen- tal material (e,g. oceanic sediments) occurs in these arcs. Because oceanic arcs are on the average more mafic than the dominantly andesitic continental arcs, an important question is whether the andesites of continental arcs are produced by a fundamentally different (more complex?) mechanism than the lavas of oceanic arcs. An excellent opportunity for study of this question is pro- vided by the island of Sarigan, in the Mariana active arc, on which calc-atkaline andesites (including hornblende-bearing types) are exposed along with more mafic lavas. Available isotope data suggest the Sarigan lavas (including the andesites) were derived from mantle material with little or no involvement of continental components. Ratios of incompati- ble elements suggest that most of the Sarigan lavas were derived from similar source materials. Absolute abundances of incompat- ible elements vary irregularly within the eruptive sequence and indicate at least 5 distinct magma batches are represented on Sarigan. Major element data obtained on the lavas and mineral phases in them, combined with modal mineral abundances, suggest that the calc-alkaline nature of thc volcanic rocks on Sarigan results from the fractional crystallization of titanomagnetite in combina- tion with other anhydrous phases. Amphibole, although present in some samples, is mainly a late-crystallizing phase and did not produce the catc-alkline characteristics of these lavas. Gab- broic samples found in the volcanic sequence have mineralogic and geochemical characteristics that would be expected of residu- al solids produced during fractional crystallization of the Sarigan lavas. When combined, data on the lavas and the gabbros suggest the following crystallization sequence: olivine - plagioclase - clinopyroxene - titanomagnetite - orthopyroxene + hornblende, biotite and accessory phases, These results lead to the conclusion that calc-alkaline magmas can be generated directly from mantle sources. Introduction Studies of the ,origin (i.e. source materials) of calc-alkaline andes- ites from continental arcs have been hampered by uncertainty concerning the degree to which magmas derived from below the crust may have been "contaminated" by reaction with the crust during ascent. This is particularly evident in the isotopic studies (e.g. Ewart and Stipp 1968 ; Armstrong and Cooper 1971 ; Church 1976). To sidestep this uncertainty, isotopic studies have been carried out on Iavas from oceanic arcs well away from continental areas (e.g. Sinha and Hart 1972; Oversby and Ewart 1972; Meijer 1976; Kay et al. 1978; Hawkesworth et al. 1979). These studies have generally found little, if any, evidence for the involvement of continental material in the genesis of lavas erupted in these arcs. The question remains as to what implica- lions these results have for the genesis of magmas in continental arcs. According to recent reviews (Miyashiro 1974; Ewart 1976), lavas erupted in continental and oceanic arcs show some consis- tent differences. Oceanic arc lavas generally have lower alkali and SiOz contents, more pronounced Fe enrichment trends, a greater proportion of basalts and basaltic andesites and they less commonly contain hydrous phases. The important question in the present context is whether these differences reflect distinct origins for continental and oceanic arc lavas or simply variations on the same basic magma generation process. We have studied a sequence of lavas on the island of Sarigan in the Mariana arc which bears on this question. The lavas of this sequence have many of the characteristics attributed to lavas erupted in continental arcs even though they were clearly erupted in an oceanic arc. The results of our studies should help bridge the gap in our understanding of the origin of oceanic and continenlaI arc lavas. Regional Setting Sarigan Island is part of the active volcanic arc associated with the Mariana arc system (Karig 1971). Subaerially, the arc stretches 500 km from Anatahan to Uracas (Fig. 1). Arc-related seamounts (Stern and Bibee t980) extend this trend north of Uracas and south of Anatahan to a total length of 1,000 km. A striking feature of this arc is its near perfect curvature, clearly lacking the segmentation noted in other circum-Pacific arcs (Carr 1976; Marsh 1979), Most of the subaerial cones and several of the seamounts are known to have been active within this century (Kuno 1962; Meijer 1981). Sarigan and Anatahan appear to be exceptions although they have undoubtedly been active in the last 10,000 years judging from the sparse vegetation on the most receut flows, Although the crustal thickness in the central part of the arc is not well defined (La Traille and Hussong 1980), the crust is clearly of oceanic character. Recent seismic studies suggest the active arc volcanoes are in part superimposed on the yotmg (0-8 m.y.b.p.; Hussong and Uyeda et al. 1981) inter-arc basin crust of the Mariana Trough (Bibee et al. 1980). The crustal O010-7999/81/0077/0337/$03.60