mwr212-08 1st pgs page 139 The Geological Society of America Memoir 212 2015 Foundering-driven lithospheric melting: The source of central Andean mafi c lavas on the Puna Plateau (22°S–27°S) Kendra E. Murray Department of Geosciences, University of Arizona, 1040 E. 4th Street, Tucson, Arizona 85721, USA Mihai N. Ducea Department of Geosciences, University of Arizona, Tucson, Arizona 85721, USA, and Facultatea de Geologie Geofi zica, Universitatea Bucuresti, Strada N. Balcescu Nr 1, Bucuresti, Romania Lindsay Schoenbohm Department of Chemical and Physical Sciences, University of Toronto–Mississauga, Mississauga, ON L5L 1C6, Canada ABSTRACT Investigations of lithospheric foundering and related magmatism have long focused on the central Andes, where there are postulated links between the eruption of mantle-derived lavas and periodic loss of the lower lithosphere. Whole-rock ele- mental and Nd-Sr-Pb isotopic results from a suite of late Miocene–Quaternary mafi c lavas erupted onto the Puna Plateau clarify the relationship between this hypothe- sized process and lava composition. Zinc and Fe provide a critical perspective because they are partitioned differently during the melting of asthenospheric and lithospheric 4 mantle. All Puna lavas have Zn/FeT (×10 ) values >13, which requires clinopyroxene and perhaps garnet to be the dominant phase(s) in the melt source; this precludes a melt source of typical mantle asthenosphere. This result is contrary to classic models of delamination magmatism that suggest asthenospheric peridotite melts to gener- ate these lavas. Pyroxenite (±garnet)–bearing lithospheric materials in the central Andes are likely common and heterogeneous in age, volatile content, and mineralogi- cal composition, and if they are the melt source, this can explain the diversity in the elemental (La/Yb = 11–45; La/Ta = 22–40) and isotopic (87Sr/86Sr = 0.7055–0.7080; ε − Nd = -1 to 7) compositions of these mafi c magmas (MgO > 8%, Mg number > 60). We propose that compositionally diverse, gravitationally unstable pyroxenites both drive “dripping” of the lower lithosphere and are the source of the resulting melt. We also postulate that mantle-derived lavas erupted on the Puna Plateau were generated during localized foundering and melting of these materials. The cumulative effect of these drip events is a modern Puna Plateau with geodynamic anomalies including thin lithosphere and anomalously high surface elevation. Murray, K.E., Ducea, M.N., and Schoenbohm, L., 2015, Foundering-driven lithospheric melting: The source of central Andean mafi c lavas on the Puna Plateau (22°S–27°S), in DeCelles, P.G., Ducea, M.N., Carrapa, B., and Kapp, P.A., eds., Geodynamics of a Cordilleran Orogenic System: The Central Andes of Argentina and Northern Chile: Geological Society of America Memoir 212, p. 139–166, doi:10.1130/2015.1212(08). For permission to copy, contact [email protected]. © 2014 The Geological Society of America. All rights reserved. 139 mwr212-08 1st pgs page 140 140 Murray et al. INTRODUCTION If lithospheric loss generated the Puna mafi c lavas, then the scale of individual foundering events, the mantle material(s) that In the central Andes, extensive differentiation and crustal melted in response, and the relationship between these parame- mixing in Neogene arc magmas has obscured the composition ters and the composition of resulting melts remain equivocal. Kay of the regional mantle (Davidson et al., 1991), which restricts our et al. (1994) proposed a model for “delamination magmatism” understanding of Andean lithospheric evolution and arc petrogen- based on trace-element compositions of the Puna mafi c lavas, esis (Thorpe et al., 1984; de Silva, 1989; Hildreth and Moorbath, regional geophysical data, and structural observations. They used 1988; Rogers and Hawkesworth, 1989; Davidson et al., 1991; geographic trends in K and La/Ta values as an indicator of (1) the Francis and Hawkesworth, 1994; Beck et al., 1996; Haschke et relative proportions of asthenospheric melt generated below dif- al., 2006; Lucassen et al., 2007; Kay and Coira, 2009; Mamani ferent parts of the Puna Plateau, and/or (2) variability in the “arc- et al., 2010; Risse et al., 2013; Kay et al., 2013; Ducea et al., like” chemistry of the mantle source as a result of delamination. 2013). Mantle-derived mafi c lavas erupted onto the Puna Plateau This trace-element interpretation required that the lavas were (Fig. 1) since 10 Ma are the most direct evidence for late Ceno- generated from a relatively homogeneous asthenospheric mantle zoic upper-mantle composition. These lavas are thought to be the peridotite that adiabatically upwelled and melted in the wake of a product of late Cenozoic loss of dense lower lithosphere (Kay et large foundering block of lower lithosphere (Kay and Kay, 1993). al., 1994), a process commonly referred to as lithospheric foun- However, if K and La/Ta values fi ngerprint extensive replacement dering or delamination (Kay and Kay, 1993; Göğüş and Pyskly- of the lower lithosphere by upwelling asthenosphere, then there wec, 2008). Recycling of lithospheric material into the convect- should be other systematic trends in the elemental and isotopic ing upper mantle is required by chemical and mass balance in values in Puna primitive lavas. This has been observed in other cordilleran orogens like the central Andes (Rudnick, 1995) and orogens (e.g., Iberian Massif—Gutierrez-Alonso et al., 2011; is hypothesized to play a key role in Andean geodynamic evolu- Canadian cordillera—Manthei et al., 2010). Instead, Kay et al. tion (Kay et al., 1994; Beck and Zandt, 2002; Schurr et al., 2006; (1994) and subsequent studies of primitive Puna lavas (Kraemer Garzione et al., 2006; DeCelles et al., 2009; Pelletier et al., 2010). et al., 1999; Drew et al., 2009; Risse et al., 2008, 2013) have 70°W 68°W 66°W 64°W 68°W 67°W 20°S A Salar B W de Uyuni 03060 24°S CHORRILLOSCHORRILLOS E km ALTIPLANO MC11 S MC12 T 22°S E R Bolivia MC18 MC17 N AZ7 SALARSALAR Argentina RG1 MC13 AZ4 DEDE Or Salar Chile ARIZAROARIZARO Pacific Ocean de Atacama 25°S C FA AZ1 24°S O R HM1 D I la L PUNA l HM2 tofa L n 26°S A NS3 AF2 E AF1 r de NS1 NS2 CerroCerro a R l GalanGalan a ALB-01 26˚S Figure 1B S A 0 100 200 ANTOFAGASTAANANTOFAGASAGASTA BASINBASIN J1 km 28°S Figure 1. (A) Digital elevation map of the south-central Andes. The PV2 PV4 active arc stratovolcanoes are currently located in the Western Cordil- PV1 PPASTOASASTO PV8 PV3 VENTURAVENTURA lera. Gray dotted line shows the regional extent of the Miocene Alti- PV7 PV5 PV6 plano-Puna volcanic complex ignimbrites after de Silva et al. (2006). 27°S (B) Simplifi ed geologic map of the Puna Plateau modifi ed from basalt, basaltic andesite Oligo-Miocene - Schnurr et al. (2006), with sample locations and regional names from salar Quaternary sediments this study. andesite, dacite, rhyolite Pre-Neogene ignimbrite basement rocks mwr212-08 1st pgs page 141 Source of central Andean mafi c lavas on the Puna Plateau (22°S–27°S) 141 reported a striking decoupling of whole-rock major-element, bly in Las Conchas valley (~26°S; Lucassen et al., 2005, 2002; trace-element, and isotopic compositions. This compositional Viramonte et al., 1999) located east of the Puna Plateau. Their complexity has been attributed to variable re-enrichment of the compositions suggest signifi cant mantle diversity beneath the mantle source by metasomatism or subduction erosion (Risse et central Andean margin during the Cretaceous. At that time, the al., 2013, and references therein), preserved pieces of enriched active Andean magmatic arc, the active Andean magmatic arc Paleozoic subcontinental lithospheric mantle (Drew et al., 2009), was centered ~100 km west of its modern position, and this con- or minor to moderate amounts of crustal contamination (up to tinental rift system was active in the back arc and foreland (Vira- 20%–25%; Kay et al., 1994; Risse et al., 2013). If we consider monte et al., 1999). The Las Conchas xenolith suite presented by that the lower crust and lithospheric mantle beneath long-lived Lucassen et al. (2005) is composed of peridotites (spinel lher- cordilleran systems would tend toward and experience founder- zolites and harzburgites with 2%–18% calculated modal clino- ing because of profound heterogeneities in age, composition, pyroxene) and minor pyroxenites; there are no garnet-bearing and temperature (Elkins-Tanton, 2007), we evidently need better xenoliths. Although many of these xenoliths and the basanite ways to clarify the links between lava composition and the mate- fl ows that host them are clearly sourced from the depleted mantle ε 87 86 rials driving foundering events. ( Nd > 0, Sr/ Sr < 0.704), there are also many peridotite and Here, we fi nd that the compositional complexity of Puna pyroxenite samples with elemental and isotopic variability mafi c lavas offers a critical insight into the processes driv- suggesting that diverse mantle sources were tapped by the rift- ing melting in the Puna mantle. Considered together with high induced melting (Lucassen et al., 2005). These heterogeneities Zn/FeT values in the most primitive lavas (Le Roux et al., 2010), are likely more profound in closer proximity to the continental which require dominantly clinopyroxene (±garnet)–bearing margin to the west. source compositions, the trace-element and Nd-Sr isotopic data suggest that heterogeneous lithospheric material, and not asthe- Lithospheric Foundering and Melt Generation nosphere, is the primary source of these lavas. This contribution expands upon Ducea et al. (2013) by presenting and interpret- Foundering of dense lithosphere into the asthenosphere has ing the full trace-element and Nd-Sr-Pb isotopic data set for the been invoked to explain “missing” lower crust and lithospheric 26 Puna lava samples used in that study.