¹DEPARTMENT OF EARTH AND SPACE SCIENCES, UCLA, LOS ANGELES, CA 90095, USA ²DEPARTMENT OF GEOLOGICAL SCIENCES AND INSTITUTE FOR CRUSTAL STUDIES, UNIVERSITY OF CALIFORNIA AT SANTA BARBARA, SANTA BARBARA, CA 93106, USA

Magmatism at ocean islands typically has been associated with deep-rooted mantle plumes. Rarely, these are coincident with oceanic spreading centres (e.g. Iceland), but more commonly they form intra-plate hotspots (e.g. Hawaii, Canary Islands). If magmatism at ocean islands is indeed intimately linked with mantle plumes and these plumes are derived from great depths (670 km discontinuity or core-mantle boundary), effectively constituting the major mode of deep upwelling in the global mantle convection system, then their surface manifestation should provide some of our best constraints on the composition and nature of the deep mantle. In contrast, the dominant volume of present-day volcanic activity, which is concentrated along plate boundaries, provides little definitive information on the deep mantle.

Over the past 20 years considerable progress in our understanding of the composition of the upper mantle has been inextricably linked to advances in analytical geochemistry, particularly the use of isotope and incompatible trace element ratios in oceanic basalts as tracers of their mantle sources. Early studies have now been followed up by a number of very detailed volcanological-petrological-geochemical studies of individual oceanic islands. Although Hawaii has traditionally served as our de facto `model' of ocean-island magmatism by virtue of the sheer volume of geochemical and geophysical data accumulated there, we now realize that a spectrum of volcano types and compositions exists, and these have been used to map mantle domains, to infer processes of partial melting, and to calculate recycling budgets between crust and mantle.

The use of ocean-island basalt (OIB) geochemistry to define mantle sources and partial melting processes depends implicitly on the assumption that the primary magmas are not significantly modified, other than by closed system fractional crystallization, en route to the surface. Some of the most detailed and comprehensive recent studies of individual ocean-island magma systems suggest, however, that many of the important criteria that we use to infer mantle sources may be compromised by interaction between magmas and the oceanic and island crust through which they ascend. The occurrence of contamination at ocean islands has been deemed less likely than in continental settings, where magmas pass through thicker, lower density crust. This contention is apparently supported by the numerous case studies of contamination of continental magmas. The difference in the observable chemical consequences of contamination in the two environments must be appreciated in the context of the compositional distinctions of the crust. The compositional contrast between continental crust and mantle-derived magma is commonly large, and therefore the effects of contamination, in terms of standard indicators such as isotope ratios, are readily appreciated. In contrast, basaltic magmas erupted on ocean islands pass through dense basaltic crust, and any compositional differences between magma and contaminant may be subtle. It is important to point out, however, that the potentially subtle effects of contamination do not imply that such effects are irrelevant in the grand picture of mantle geodynamics, which itself is dependent on compositional subtlety.

Given the large volume of high-quality data, and the large number of studies of individual ocean islands that now exist, the time has come to ask what effects shallow-level processes at ocean islands may have in modifying magma compositions. This question was addressed at a Chapman Conference sponsored by the American Geophysical Union from 9 to 16 November 1996 in Tenerife, Canary Islands. The conference was attended by 45 scientists from seven nations, representing interests varying from isotope geochemistry to geophysics. This special issue of the Journal of Petrology includes a series of short papers summarizing some of the major findings of the meeting.

The first series of papers in this issue includes case studies of ocean-island magmatism that highlight the diversity of conclusions regarding the importance of shallow-level processes. Garcia et al., Gee et al. and Davis & Clague provide evidence for crustal contamination of basaltic magmas. Assimilation of hydrothermally altered basement is reflected in various aspects of the geochemistry of basalts from Hawaii (Garcia et al.) and Iceland (Gee et al.). The role of hydrothermally altered crust as an assimilant at ocean islands was a recurring theme, no doubt reflecting the magma-crust compositional contrast in characteristics, such as fluid-mobile trace element contents and stable isotope ratios, that facilitate the identification of contamination. The Escanaba Trough basalt data discussed by Davis & Clague, although not strictly from an ocean island, illustrate the compositional effects of contamination by seafloor sediments that might be expected in any oceanic environment. The basalt-sediment interface, in principle, represents a level of strong density contrast at which basalts might be expected to pond and differentiate. Hoernle provides a counterpoint example from the Canary Islands, in which the composition of the associated oceanic basement is found not to be a suitable contaminant of basalts from the islands. Class et al. provide evidence for interaction between plume-derived magmas and the oceanic lithosphere at Grand Comore, although the correlation of these effects with pressure-sensitive indices of amphibole involvement places these processes in the lithospheric mantle rather than the crust. Case studies from Tenerife (Ablay et al.), French Polynesia (Dostal <