June 2004 LIP of the Month

(corresponds to event 11 in LIP record database)

Caribbean-Colombian Oceanic Plateau

Andrew C Kerr
School of Earth, Ocean and Planetary Sciences,
Cardiff University, Park Place Cardiff,
CF10 3YE, UK
kerra@cf.ac.uk

webpages:
http://www.earth.cf.ac.uk/people/personal-info-page.asp?id=116
http://www.earth.cf.ac.uk/people/default.asp?id=116

The ~90 Ma Caribbean Colombian Oceanic Plateau (CCOP) is exposed around the margins of the Caribbean and along the northwestern continental margin of South America (Figure 1). The thickened nature of the bulk of the Caribbean plate (8-20 km; (Edgar et al., 1971; Mauffret and Leroy, 1997) testifies to its origin as an oceanic plateau. The preserved water-covered area of the CCOP is about 6 x 105 km2; however, because a significant portion of the plateau appears to have accreted onto the western continental margin of Colombia and Ecuador (and some may have subcreted), and thus tthe oceanic plateau may originally have been more than twice this size (Burke, 1988).

Duncan and Hargraves (1984) and Hill (1993) have proposed that the CCOP was produced by melting during the initial ‘plume head’ phase of the Galápagos hotspot. Eastward movement of the Farallon Plate in the Late Cretaceous brought the northern part of the plateau into the continental gap which had been opening up between North and South America since the Jurassic. The eastward moving plateau appears to have been too buoyant (due to remnant heat and crustal thickness) to be subducted (Burke et al., 1978; Hill, 1993). This ‘clogging’ of the subduction zone led to a ‘flip’ in the direction of subduction from east to west, and the Atlantic Plate began to be consumed by subduction as opposed to the Farallon/Caribbean Plate (Figure 2a). In addition to subduction polarity reversal, subduction of proto-Pacific crust also commenced behind the trailing edge of the plateau, eventually forming the current Central American arc (Figure 2a).

In contrast, the southern part of the Caribbean-Colombian plateau began to interact with the northwestern continental margin of South America (Figure 2b). In this region, docking of the plateau with the continent resulted in the jamming of the subduction zone, but rather than subduction flip this led to progressive, westward back-stepping of the subduction zone (Figure 2b), leading to the formation of accreted oceanic plateau terranes along the northwestern margin of South America (e.g., Kerr et al., 1997b).

In the Caribbean, the plateau has been drilled by DSDP Leg 15 and ODP Leg 165 (Figure 1; Bence et al., 1975; Sinton et al., 2000). The accreted plateau material in Colombia, Ecuador, Costa Rica and Hispanola consists of fault-bounded slices of basaltic, and occasionally picritic, lavas and sills with relatively few intercalated sediments and ash layers (Kerr et al., 1997a). Although they preserve layered and isotropic gabbros and ultramafic rocks, unlike accreted ophiolites generated at spreading centres, these obducted sequences of oceanic plateau do not possess sheeted dyke complexes (Kerr et al., 1998).

Several exposures of the CCOP are worthy of special mention: Firstly, the 5 km-thick section on the island of Curaçao, 70 km north off the coast of Venezuela (Figure 1). The sequence consists of pillowed picrites (Figure 3) low in the succession that gradually give way to more basaltic pillow lavas nearer the top. These pillow lavas are intercalated with hyaloclastite horizons and intrusive sheets (Klaver, 1987; Kerr et al., 1996b). The second noteworthy locality is the Island of Gorgona, 50 km off the western coast of Colombia (Figure 1). This small island (2.5 x 8 km) is the site of the youngest spinifex-textured komatiites yet found (Echeverría, 1980; Kerr et al., 1996a; Arndt et al., 1997). The formation of these Cretaceous komatiites in the CCOP has led to the suggestion that pre-Cambrian komatiites formed in ancient oceanic plateaus.

The most magnesian lavas found in the CCOP contain up to 28 wt. % MgO (Figure 4). However, it is likely that these lavas contain substantial accumulated olivine and so the whole rock compositions of these high MgO rocks cannot represent parental mantle melts. Estimates of the MgO content of the parental melts for various parts of the province vary from 18 wt. % MgO to about 12 wt. % MgO (Kerr et al 1996a,b; Révillon et al 1999). Although picritic lavas are more common than in other Cretaceous oceanic plateaus, basalts are by far the most common rock type preserved in the CCOP, with the vast majority of samples containing between 6-10 wt. % MgO (Figure 4). Al2O3 contents broadly increase with decreasing MgO, reflecting the importance of the addition and removal of olivine during the petrogenetic history of the CCOP lavas. CaO increases with decreasing MgO until 8-10 wt. % MgO where the CCOP lavas display a scattered but discernible downward trend. Fe2O3(t) and TiO2 display broadly horizontal trends until about 8-10 wt. % MgO, where they both increase markedly. These trends can be modelled by the initial fractional crystallisation or accumulation of olivine (+ minor Cr-spinel) followed by the commencement of crystallisation of plagioclase and clinopyroxene between 8 and 10 wt. % MgO (Kerr et al., 1996b).

Trace element data (Figure 4) also support the proposed fractional crystallisation model: Ni and Cr contents fall with decreasing MgO contents. Despite some scatter, incompatible trace element contents, e.g., Nb and Zr, generally increase with decreasing MgO contents (Figure 5).

One of the most interesting aspects of the trace element data for the CCOP is that the basaltic lavas possess a narrower range of incompatible trace element ratios, than the picrites. For instance, well over 80% of the basaltic samples (<12 wt. % MgO) from the CCOP possess La/Y between 0.05 and 0.2 and Zr/Nb between 7 and 20 (Figures 5e-f). In contrast, the picritic and komatiitic lavas possess much more variable ratios of incompatible trace elements, with La/Y ranging from 0.05 to 0.45 and Zr/Nb from 5 to 85. This is also shown on primitive mantle normalised multi-element plots where it can be seen that the CCOP basalts possess broadly flat patterns whereas the high-MgO picrites and komatiites are generally much more variable, with some being more depleted, and some more enriched than the basalts, particularly for the most highly incompatible trace elements such as Th, Nb, La, Ce & Nd (Figure 6).

The heterogeneity of the high-MgO rocks is also reflected in the radiogenic isotope ratios, particularly eNd (Figures 7 and 8). Virtually all the analysed basalts from the CCOP possess initial eNd ranging from +6 to +9, whereas the high MgO lavas generally fall outside this range (eNd >+9 and <+6; Figure 7). Elevated initial 87Sr/86Sr ratios found in several parts of the province have been attributed either to contamination with altered oceanic crust (Curaçao: Kerr et al., 1996b) or to secondary alteration (Gorgona: Révillon et al., 1999).The wide range of isotopic data for the CCOP reveals that the enriched and depleted lavas are not simply formed by variable melting of a homogeneous source region, but rather reflect melting of long-term depleted and enriched components from a markedly heterogeneous plume source region. (Kerr et al., 1996a; 2002; Arndt et al., 1997; Hauff et al., 2000a; Thompson et al., 2004).

The greater heterogeneity of the high MgO rocks in comparison to the basalts, has been interpreted to reflect the formation of these lower MgO magmas through mixing and fractional crystallisation of the high-MgO magmas in large magma chambers. The heterogeneous high-MgO rocks thus represent magmas that passed relatively quickly through the lithosphere without being trapped in magma chambers (Kerr et al., 1998). The extent of partial melting required to produce the parental magmas of the CCOP has been calculated to be of the order of 20% (e.g. Kerr et al., 1997a).

References