Lesser Antilles arc lavas have trace element and radiogenic isotope characteristics indicative of a continent-derived contribution. It is debated vigorously whether this continental signature represents terrigenous sediment that has been subducted with the Atlantic plate and added to the magma sources in the mantle wedge, or portions of the subarc crust that are assimilated during magma ascent. Here we present Mo isotope data for Lesser Antilles arc lavas and sediments offboard the Lesser Antilles trench. Sequences of black shales, present in the subducting sediment piles, are highly enriched in Mo and have unusually high 98Mo/95Mo. Despite their low mass fraction in the sediment package (<10% in Deep Sea Drilling Project Site 144), they dominate the Mo content and isotopic composition of the bulk sediment subducting at the Lesser Antilles trench. We show that lavas from the southern part of the Lesser Antilles arc also have high 98Mo/95Mo ratios, implicating the addition of Mo derived from the subducted black shales to their mantle sources. This establishes a new link between the composition of subducted material and the arc lava output.
Molybdenum isotope ratios provide an important means of tracing paleo–redox conditions in the ocean. Among the many different oceanic sediment types, black shales are of special interest for Mo isotope studies because they have unusually high Mo concentrations and are associated with heavy Mo isotope ratios that record the composition of contemporaneous seawater when depositional conditions are pervasively euxinic (Barling et al., 2001, Gordon et al., 2009). Therefore, black shales constitute a high 98Mo/95Mo end member among oceanic sediments. Under more oxidizing conditions, lighter Mo isotopes are preferentially incorporated into sediments (Barling et al., 2001).
The unique signature of high Mo concentrations and high 98Mo/95Mo ratios provides an attractive means to trace the fate of black shales after they have been subducted. Sequences of black shales deposited in the Atlantic Ocean as a result of Cretaceous oceanic anoxic events (OAEs) are currently being subducted at the Lesser Antilles trench. Black shales from OAE 2 and OAE 3 (93–84 Ma) have been sampled at Deep Sea Drilling Project (DSDP) Site 144. Carpentier et al. (2008) concluded that a component derived from these black shales is likely transported into the Lesser Antilles arc magma sources, based on their highly radiogenic Pb isotope ratios. However, the view that the isotopic compositions of the Lesser Antilles lavas are related to subducted sediments is contended (e.g., Thirlwall and Graham, 1984; Davidson, 1986; White and Dupré, 1986; Thirlwall et al., 1996; Turner et al., 1996; Carpentier et al., 2008; Labanieh et al., 2010). Most recently, it has been suggested that crustal assimilation could explain the isotopic variability in lavas from Saint Lucia, assuming that the southern Lesser Antilles arc is built upon the older Aves Ridge (Bezard et al., 2014). In addition, the highly radiogenic Pb isotope composition of some Lesser Antilles arc lavas could be related to melting sediments similar to those found on the overriding plate in Barbados, rather than subducting black shales (Carpentier et al., 2008). The possibility of combining a novel tracer for components derived from subducted black shales in Lesser Antilles arc magmas with other geochemical fingerprints could therefore offer important new information on the long-standing controversy regarding the importance of subducted sediments in Lesser Antilles magmas.
MOLYBDENUM ISOTOPE COMPOSITION OF SEDIMENTS SUBDUCTING AT THE LESSER ANTILLES ARC
Extensive work geochemically characterizing the sediments on the Atlantic plate near the Lesser Antilles trench has been done, with the aim of constraining subduction inputs (White et al., 1985; Carpentier et al., 2008, 2009), including analyses of samples drilled at DSDP Sites 144 and 543 (Fig. DR1 in the GSA Data Repository1). DSDP Site 543 is located close to the Lesser Antilles arc on oceanic crust of Campanian (ca. 80 Ma) age. DSDP Site 144 is located further from the trench than DSDP Site 543, on the Demerera rise southeast of the Lesser Antilles trench (Fig. DR1). Black shales deposited during OAE 2 and OAE 3 (93–84 Ma) have been sampled at DSDP Site 144 but are not present at the younger DSDP Site 543, where the oceanic crust formed after the deposition of the black shales. For the purposes of this study, we therefore focus on sediments of DSDP Site 144.
Our measurements of Mo isotope ratios and Mo concentrations in representative sediment samples from the different lithological units of DSDP Site 144 (reported in Table DR1) are shown in Figure 1, together with sediment samples from Ocean Drilling Program (ODP) Leg 129, Sites 800, 801, and 802. The ODP sites are located on ca. 167 Ma Pacific crust near the Mariana Trench and are the only other oceanic sediment columns that have been analyzed for Mo isotope ratios (Freymuth et al., 2015). Of all the sediments analyzed from the Pacific and Atlantic sites, black shales have the highest δ98/95Mo (the permil variation in 98Mo/95Mo relative to the National Institute for Standards and Technology 3134 standard) and the highest Mo concentrations. The Mo isotope composition of the DSDP Site 144 black shales (δ98/95Mo ∼0.6) is within the range of previously reported data for OAE 2 black shales from DSDP Site 367 and drill site S57 in the northeast Atlantic (δ98/95Mo is ∼0.45–0.85; Westermann et al., 2014; Goldberg et al., 2016).
The calculated Mo isotope ratio of the bulk sediment at DSDP Site 144 (see Fig. 1 caption) is dominated by the black shale contribution and it is ∼0.8‰ higher in δ98/95Mo than the bulk ODP Leg 129 sediment. It also greatly exceeds the estimated ranges in δ98/95Mo of the upper mantle (Freymuth et al., 2015; Greber et al., 2015), bulk silicate earth (Burkhardt et al., 2014; Greber et al., 2015), and the continental crust (Siebert et al., 2003; Voegelin et al., 2014) (Fig. 1).
TRACING SUBDUCTED BLACK SHALES WITH MO ISOTOPES
Within the Lesser Antilles arc lavas there is a substantial, and in detail complex, gradient of radiogenic isotope ratios with more continental compositions in the southern islands than in the northern islands; e.g., Sr and Pb isotope ratios become more radiogenic and Nd isotope ratios become less radiogenic southward (White and Dupré, 1986; Turner et al., 1996; Macdonald et al., 2000; Carpentier et al., 2008) (Fig. DR2). We analyzed samples from along the arc that are already well characterized for Sr, O, and He isotopic compositions (van Soest et al., 2002). The sample set encompasses much of the isotopic variability along arc from depleted signatures in the north to more continental compositions in the south (Fig. DR2).
With the exception of the sample from Martinique, Mo isotope ratios in the Lesser Antilles lavas are unrelated to the degree of differentiation (Fig. 2A), but correlate with radiogenic Sr and Pb isotope ratios and Ce/Mo (Figs. 2B–2D). As with the radiogenic isotopes, the Mo isotope and Ce/Mo ratios of the Lesser Antilles arc lavas show a regional variation with higher δ98/95Mo and Ce/Mo in the southern islands than in the northern islands.
We focused on mafic samples; the most primitive magmas, from southern islands, have the highest δ98/95Mo (Fig. 2A). Thus there is no scope to explain the unusually high δ98/95Mo in the southern islands as a result of fractionation of hydrous phases (Voegelin et al., 2014), which are absent as phenocryst phases in most of our samples (Table DR1). Assimilation of crustal basement has been suggested to explain some of the geochemical heterogeneity in the Lesser Antilles, in particular in the central islands (e.g., Davidson, 1986; Macdonald et al., 2000; Bezard et al., 2014). The highest δ98/95Mo ratios are present in samples from the southern islands (Fig. 2A) and include the most primitive lava sample from the Lesser Antilles (van Soest, 2000). This sample has mantle-like Os isotope ratios and was unaffected by crustal assimilation (Bezard et al., 2015a). It is therefore highly unlikely that the elevated Mo isotopic ratios documented here are related to assimilation of subarc crust. An exception might be our most differentiated, andesitic sample from Martinique that has the lowest δ98/95Mo (Fig. 2A). A role for crustal assimilation has been documented for lavas from Martinique in order to account for their extreme Sr and Nd isotopic compositions (Fig. DR2). The composition of the subarc crust is largely unknown, but has been suggested to be partly formed by an ancient accretionary prism of the Aves Ridge system (Macdonald et al., 2000). The Caribbean plate on which the Aves Ridge is built originates from the eastern Pacific (Sykes et al., 1982; Burke, 1988), which remained oxic at least during OAE 2 (Takashima et al., 2011), suggesting that the subarc crust has a composition more similar to typical oceanic sediments with low δ98/95Mo (Fig. 1) and so could account for the isotopically light Mo isotope composition of our Martinique sample.
Molybdenum is similarly incompatible during mantle melting to Ce (Newsom et al., 1986). Ratios of Ce/Mo in arc lavas below upper mantle values should therefore reflect preferential slab addition of Mo to the arc lava source (Freymuth et al., 2015). Freymuth et al. (2015) showed that the Mo budget of Mariana arc lavas is dominated by the addition of low Ce/Mo fluids with high δ98/95Mo (∼0.05) derived from the subducted mafic oceanic crust and high Ce/Mo melts with similar or lower δ98/95Mo. Arc lavas with a more dominant influence of sediment melts are typically characterized by lower δ98/95Mo values (König et al., 2016) that can be as low as −0.7‰. Radiogenic isotope ratios of the northern Lesser Antilles lavas indicate they are less influenced by continental, sediment-like material (Figs. 2B and 2C; Fig. DR2) and so should have a more fluid-dominated signature. By analogy with the Mariana arc lavas and supported by the high fluid-mobility of Mo compared to the rare earth elements (Green and Adam, 2003; Bali et al., 2012), we therefore interpret their low Ce/Mo ratios to reflect the addition of Mo-rich fluids to their mantle sources. The Mo isotopic composition of the northern Lesser Antilles lavas suggests a fluid composition of δ98/95Mo ∼−0.15 (Fig. 2D), slightly lower than that inferred for the Mariana arc (Freymuth et al., 2015).
The high Ce/Mo and radiogenic Sr and Pb isotope compositions in the southern Lesser Antilles (Fig. 2) clearly require another component with high δ98/95Mo in addition to slab-derived fluids. The altered top part of the mafic oceanic crust (AOC) has a moderately heavy Mo isotope composition, yet a major contribution by the AOC is inconsistent with its unradiogenic Pb isotope composition (Fig. 2C). Among all sediment types, only the black shales have suitably high δ98/95Mo to constitute this component, suggesting that the high δ98/95Mo is derived from the black shales.
Figure 2D shows that the high δ98/95Mo end member required for the Lesser Antilles arc lavas is also characterized by high Ce/Mo ratios (>35), which is initially surprising, given the low Ce/Mo of the black shales and bulk DSDP Site 144 sediment. However, Ce concentrations in hydrous melts derived from certain sediment types can be increased relative to starting compositions if the residual assemblage lacks a host that is as efficient in retaining Ce (e.g., Skora and Blundy, 2010; Martindale et al., 2013). Although the Ce/Mo of the subducted bulk sediment is low, associated experimental work on the Lesser Antilles sediment documents that its melts will likely have higher Ce/Mo (Skora, personal data).
Carpentier et al. (2008), White and Dupré (1986), and Bezard et al. (2015b) argued for a changing sediment composition toward the northern Lesser Antilles to explain some of the compositional variation in the Lesser Antilles arc magmas. In particular, Carpentier et al. (2008) argued that the black shale sequence is absent beneath the northern Lesser Antilles arc because the crust in the northern part appeared to be younger than the age of the black shales. Such a scenario is principally consistent with our new data set given that a black shale component is not prominent in that section of the arc. A closer examination of the seafloor magnetic anomalies in the area (Fig. DR1), however, shows that the oceanic crust underneath some of the northern islands is likely older than 93 Ma and could thus carry the entire sequence of OAE 2 and OAE 3 black shales sampled at DSDP Site 144.
The lower δ98/95Mo and Ce/Mo in combination with less radiogenic Sr and Pb isotope ratios in the northern Lesser Antilles islands therefore argue for the model of Turner et al. (1996), who proposed a lower contribution of sediment melts to the sources of the northern islands. This could be achieved, for example, if the slab-top temperature beneath the northern Lesser Antilles was lower than beneath the southern arc section, thus inhibiting melting of the slab beneath the northern Lesser Antilles. This scenario is supported by current models of the subduction zone thermal structure that suggest that slab-top temperatures are lower by ∼50 °C in the northern part of the arc (Syracuse et al., 2010). Alternatively, melting of the sediment section of the slab beneath the northern Lesser Antilles could be inhibited by a low supply of H2O, as suggested to explain variable degrees of slab melting beneath the Izu arc (Freymuth et al., 2016). Regardless of the underlying process, the along-arc compositional variation in the Lesser Antilles lavas likely reflects changes in the physical parameters of the subduction zone rather than variable input compositions.
This study was funded by Natural Environment Research Council grants NE/J009024/1 and NE/H023933/1 to Elliott. Skora acknowledges funding by Swiss -National Science Foundation Ambizione grant PZ00P2_142575/1. We thank Marion Carpentier for sharing her Deep Sea Drilling Project Site 144 samples, David Schlaphorst for help with the preparation of the Lesser Antilles map, and Jenny Collier for sharing her data and knowledge of seafloor magnetic anomalies in the Caribbean. Samples used for this study were collected by van Soest, who was supported by a grant from the Netherlands Organization of Scientific Research to David R. Hilton, whom we thank. Reviews from three anonymous reviewers helped to improve the manuscript.
- Received 10 July 2016.
- Revision received 14 September 2016.
- Accepted 15 September 2016.
- © 2016 Geological Society of America