Modification of an oceanic plateau, Aruba, Dutch Caribbean: Implications for the generation of continental crust

Abstract

The generation of the continental crust may be connected to mantle plume activity. However, the nature of this link, and the processes involved, are not well constrained. An obstacle to understanding relationships between plume-related mafic material and associated silicic rocks is that later tectonic movements are liable to obscure the original relationships, particularly in ancient greenstone belts. Studies of younger analogous regions may help to clarify these relationships. On the island of Aruba in the southern Caribbean, a sequence of partly deformed mafic volcanic rocks intruded by a predominantly tonalitic batholith is exposed. The mafic lavas show geochemical and isotopic affinities with other basaltic, picritic and komatiitic rocks that crop out elsewhere in the Caribbean—these are well documented as belonging to an 88–91 Ma plume-related oceanic plateau, which is allochthonous with respect to the Americas, and is thought to have been formed in the Pacific region. The ∼85 to ∼82 Ma tonalitic rocks share some geochemical characteristics (high Sr and Ba, low Nb and Y) with Archaean tonalite–trondhjemite–granodiorite (TTG) suites. Field relationships suggest that deformation of the plateau sequence, possibly related to collision with a subduction zone, was synchronous with intrusion of the Aruba batholith. New incremental heating ${}^{40}\text{Ar/}{}^{39}\text{Ar}$ dates, combined with existing palaeontological evidence, show that cooling of the batholith occurred shortly after eruption of the plateau basalt sequence. Sr–Nd isotopic data for both rock suites are uniform (${}^{87}\text{Sr/}{}^{86}\text{Sr}{}_{\mathrm{i}}\text{≈0.7035}$, εNd_i≈+7), whereas Pb isotopes are more variable (Plateau sequence: ${}^{206}\text{Pb/}{}^{204}\text{Pb}\text{=18.6–19.1}$, ${}^{207}\text{Pb/}{}^{204}\text{Pb}\text{=15.54–15.60}$, ${}^{208}\text{Pb/}{}^{204}\text{Pb}\text{=38.3–38.75}$; Aruba batholith: ${}^{206}\text{Pb/}{}^{204}\text{Pb}\text{=18.4–18.9}$, ${}^{207}\text{Pb/}{}^{204}\text{Pb}\text{=15.51–15.56}$, ${}^{208}\text{Pb/}{}^{204}\text{Pb}\text{=38.0–38.5}$). This suggests that there has been a minor sedimentary input into the source region of the batholith. However, the limited time interval between basaltic and tonalitic magmatism makes a normal subduction-related origin for the tonalites improbable. Instead, models involving derivation of tonalite from partial melting of the plateau sequence, or alternatively, genesis in an unusual subduction zone environment, are investigated.

Introduction

The question of how continental crust was (and is) generated is of fundamental importance to geologists and planetary scientists. Several authors have proposed that the growth of the continental crust may be connected to the presence of mantle plumes and oceanic plateaux (e.g., Stein and Hofmann, 1994; Abbott and Mooney, 1995; Stein and Goldstein, 1996), although the nature of this link is not well-constrained and many questions remain unanswered. Growth of the continental crust can arise directly from a plume ascending beneath continental lithosphere, particularly if accompanied by rifting (e.g., White et al., 1987); additionally, plumes which rise beneath oceanic lithosphere can result in the formation of thick oceanic plateaux, which because of their buoyancy may resist subduction (Saunders et al., 1996), becoming accreted to the edges of continents. A major difficulty is unequivocally identifying examples of oceanic plateaux in the ancient geological record. Modern plateaux are recognised by their crustal thickness, composition, and sedimentary record (e.g., Coffin and Eldholm, 1994), but even when all these criteria are used to evaluate an environment of formation, the interpretation is rarely irrefutable. The disparity between the abundance of oceanic plateaux in the modern oceans and their relative paucity in the continental geological record suggests that, if the rate of production of plateaux has not been anomalously high in the last 150 m.y., either plateaux are eventually subducted, or they are processed and converted into material which is no longer easily recognisable as oceanic plateau crust (Stein and Goldstein, 1996; Saunders et al., 1996).

Several important comparisons can be made between the two generally accepted models of continental growth: volcanic arc accretion and oceanic plateau accretion. Firstly, oceanic plateaux represent enormous volumes of material (Schubert and Sandwell, 1989). The crust of the Ontong Java plateau (OJP), for example, occupies a volume of 44 to 56 million km³ (Gladczenko et al., 1997), which could potentially be accreted to make continental lithosphere. The oceanic plateau model may therefore be able to explain periods of rapid continental growth better than the volcanic arc accretion model; additionally, as evidence mounts for episodic crustal growth (e.g., McCulloch and Bennett, 1994), it has been suggested that the sporadic nature of mantle plume activity and oceanic plateau magmatism is responsible for this episodicity (e.g., Condie, 1997). Secondly, neither oceanic plateau nor oceanic arc accretion can, on their own, explain the geochemical characteristics of the bulk continental crust, because both are too mafic in composition (Tarney and Windley, 1977). Differentiation and modification of primitive mantle-derived arcs or plateaux must therefore be an important process in creating the enrichment of silica and various incompatible trace elements observed in continental crust. We need to know how these enrichment processes operate, and on what timescale. Finally, it must be stressed that there are unambiguous examples of accretion of both oceanic plateaux (e.g., the Caribbean plateau in Colombia: Millward et al., 1984; Kerr et al., 1997a) and oceanic arcs (Soja, 1996) to pre-existing continents. Accordingly, our task should now be to constrain the means by which conversion of oceanic plateau or arc into continental crust occurs, by considering both modern and ancient examples.

Some fragments of oceanic plateau crust have been identified as accreted members of greenstone belts (e.g., Kusky and Kidd, 1992; Desrochers et al., 1993), although other greenstone belts are interpreted as accreted volcanic arc terranes (e.g., the Slave Province, Canada: Kusky, 1989; Davis and Hegner, 1992) or autochthonous eruptive sequences (Bickle et al., 1994). In spite of this dissent, however, it seems that studies of Phanerozoic oceanic plateaux and their subsequent histories may provide useful constraints on Archaean processes.

The island of Aruba, located in the complex southern plate boundary zone of the Caribbean region, may provide the answers to some of these controversies. Here, a predominantly tonalitic batholith dated as ∼85–82 Ma intrudes a sequence of mafic lavas, volcaniclastic sediments and intrusive rocks belonging to the predominantly 88 to 91 Ma Caribbean–Colombian Cretaceous Igneous Province (CCCIP; Fig. 1) (Kerr et al., 1997b; Sinton et al., 1998). The association of mafic rocks that are thought to be plume-related members of the CCCIP and tonalites that share some characteristics with Archaean tonalites means that Aruba is an ideal place to study the relationships between initial material extracted from the mantle and the silicic material intimately associated with it in both space and time. The swift transmutation in magmatic environment observed on Aruba suggests that conversion of plateau material into continental crust may begin at a much earlier stage than has been previously suspected (e.g., Stein and Goldstein, 1996).