Versão online: http://www.lneg.pt/iedt/unidades/16/paginas/26/30/185 Comunicações Geológicas (2014) 101, Especial I, 99-104 IX CNG/2º CoGePLiP, Porto 2014 ISSN: 0873-948X; e-ISSN: 1647-581X

Island arc-magmatism fingerprint in the geochemistry of tephras from Deception Island, Antarctica Assinatura do magmatismo tipo arco vulcânico na geoquímica dos piroclastos da Ilha Deception, Antarctica P. Ferreira1*, R. Calvo2, R. Santos2, A. Mão de Ferro3, J. Canário4, A. M. Mota4

© 2014 LNEG – Laboratório Nacional de Geologia e Energia IP

Abstract: Deception Island is located in a highly complex tectonic setting, in a back arc basin at the transition from rifting to spreading and genetically related to subduction of Phoenix plate under the Antarctic plate, assumed to be inactive for the last 3.3 Ma. Geochemical interpretation of recent (Quaternary age) volcanic rocks (tephras) collected in this island supports the important role of subduction on the mantle chemical characteristics (low HFSE / LILE ratios and Nb-Ta depletion) thought to be the source of recent magmatism in this region. Inter-element correlations obtained in this study suggest that fractional crystallization is the dominant process relating tephra compositions.

Keywords: Geochemistry, Deception Island, Tephras, Island arc-magmatism. Resumo: A ilha Deception encontra-se num contexto tectónico bastante complexo, estando localizada numa bacia tipo “back arc” que se encontra numa fase transicional entre a abertura e o alastramento do fundo oceânico; esta bacia está geneticamente relacionada com a subdução da placa Phoenix sob a placa Antárctica, que se assume como estando inactiva desde há 3.3 milhões de anos. A interpretação geoquímica de rochas vulcânicas (piroclastos) recentes (idade quaternária), amostradas nesta ilha, corrobora o papel importante dos efeitos da subducção no estabelecimento das características químicas do manto (baixas razões HFSE / LILE e empobrecimento em Nb-Ta) que constitui a fonte dos magmas gerados nesta região. As correlações obtidas para os vários elementos químicos analisados neste trabalho sugerem que a cristalização fraccionada é o processo dominante durante a evolução magmática que gerou estas rochas piroclásticas.

Palavras-chave: Geoquímica, Ilha Deception, Piroclastos, Magmatismo tipo arco vulcânico. 1Laboratório Nacional de Energia e Geologia, I.P., Estrada da Portela, Apartado 7586- Alfragide, 2610-999 Amadora. 2Laboratório Nacional de Energia e Geologia, I.P., Rua da Amieira, Apartado 1089, 4466-901 S. Mamede de Infesta. 3Instituto de Tecnologia Química e Biológica, Univ. Nova de Lisboa, 2780-157 Oeiras. 4CQE, Instituto Superior Técnico, Universidade de Lisboa, Av. Rovisco Pais, 1, 1049-001 Lisboa. *Corresponding author / Autor correspondente: pedro.ferreira@lneg.pt

1. Introduction

Rocks formed by volcanic activity have specific chemical fingerprints reflecting the composition and evolution of the mantle domains, from where they originate, and the processes occurring during the magmatic ascension towards earth’s surface. The determination of certain major and trace elements and specific radiogenic isotopic ratios, both highly discriminant chemical parameters, allow studying the fractionation processes acting at distinct stages of magmatic evolution and the Earth’s mantle composition and evolution. The importance of studying the volcanic rocks geochemistry from Deception Island comes from its highly complex and hybrid plate tectonic setting in the Antarctic peninsula (Fig.1), where a subduction zone, a back-arc basin and 3 lithospheric plates co-exist (e.g. Barker et al., 1991). It is known that magma compositions and volcanism types in such hybrid tectonic settings show more complex characteristics (e.g. Gvirtzmann & Nur, 2001) than those normally found in the Mid-Ocean Ridges or in intraplate settings. Given this tectonic scenario, and being Deception Island located in the spreading centre of the young (< 4 Ma) Brandsfield Basin, but distant by only 120 km from South Shetland Trench (where the convergence ceased 3.3 Ma ago) we can ask: Is Deception island’s volcanism genetically associated to the subduction of Phoenix plate under the Antarctic plate or, alternatively, related to the rifting leading to the opening of the marginal basin? To address this question, this work presents geochemical data (major elements, REE, Nb, Hf, Ta, Th and U) characterizing the recent (Quaternary age) volcanism occurred in Deception Island, obtained from the study of 31 tephras samples, collected in 2011. Additionally, insights into magmatic differentiation processes and chemical characteristics of the mantle source will be addressed.

2. Geologic setting

The Scotia Arc is a megastructure composed of seamounts, submarine ridges, and islands (Fig.1a) that connects South

Artigo Curto Short Article

100 P. Ferreira et al. / Comunicações Geológicas (2014) 101, Especial I, 99-104

America to the Antarctic Peninsula and encloses the Scotia Plate (Barker et al., 1991). In Antarctic Peninsula, a magmatic arc developed from late Triassic to recent times as part of the Andean-West Antarctic continental margin (e.g. Barker et al., 1991). During the Cretaceous, the focus of arc magmatism shifted towards the western edge of the Antarctic Peninsula, leading to the development of a separate magmatic arc (e.g. Smellie et al., 1984) as a consequence of the Phoenix plate subduction under the Antarctic plate.

The opening of the Scotia Sea, as the result of the initiation of Antarctic / Phoenix Ridge (also called Aluk Ridge), is considered to be a Tertiary event, that involved spreading from about 29 Ma to 3.3 Ma (e.g. Lawver et al.,

1996). The Bransfield Basin, a 500 km long and 100 km wide extensional structure (Fig.1b) is considered a back-arc basin developed since the Pliocene and associated with subduction of the former Phoenix Plate below the South Shetland Islands (e.g. Dalziel, 1984). The opening of the Bransfield rift generated a new microplate, the South Shetland Plate, bounded by the Shackleton and Hero Fracture Zones (east and west, respectively), by the South Shetland Trench (north) and by the Bransfield rift (south). The edge of the Bransfield basin is defined by a spreading centre in which Deception Island and a number of submerged volcanic vents are associated. Younger volcanism is not directly connected to subduction but is related to rifting and back-arc basin formation.

Fig. 1. a) Simplified regional tectonic framework of the Scotia Arc, north western Antarctic Peninsula regions and location of the South Shetland Islands. b) Tectonic and geographical location of Deception Island along the spreading axis of Bransfield Through (figures a) and b) from Martí et al. (2013). c) Deception Island orthophotomap, showing the three main sampling sites (black rectangles) performed during CONTANTARC campaign from 1 to 20th of December 2011 (Mão de Ferro et al., 2013). Fig. 1. a) Enquadramento tectónico regional simplificado do Arco Magmático de Scotia, da região NW da Península Antárctica e localização das Ilhas Shetland do Sul. b) Localização tectónica e geográfica da Ilha Deception ao longo do Rift de Bransfield (figuras a) e b) de Martí et al. (2013). c) Ortofotomapa da Ilha Deception, indicando-se as três principais zonas de amostragem realizadas durante a campanha CONTANTARC, entre 1 e 20 de Dezembro de 2011 (Mão de Ferro et al., 2013).

Geochemistry of tephras from Deception Island 101

3. Geochemistry

Integrated in the CONTANTARC project (Mão de Ferro et al., 2013), a set of 32 volcanic samples (tephra, having dimensions corresponding to lapilli), were collected in three regions of Deception Island (Fig.1c), all belonging to Post-caldera deposits formation, defined by Smellie et al. (1984). The samples were analyzed in terms of their major (by X-Ray Fluorescence Spectroscopy) and trace element compositions (REE, Nb, Hf, Ta, Th and U by Inductively coupled plasma mass spectrometry) at the laboratory facilities of LNEG in Porto.

The Deception rocks form a compositional series that is almost limited to basaltic andesites, having a restricted SiO2 variation (between 52.5 and 55.5 wt%, Fig.2); only three samples are plotted in the basaltic trachyandesite field (compositional fields from Le Bas et al. (1986)).

Fig. 2. Total alkalis vs. SiO2 diagram (TAS) for Deception tephras and other published data for the island. Compositional fields from Le Bas et al. (1986) for volcanic rocks with a dashed line (MacDonald & Katsura, 1964) separating the alkali and sub alkali fields. Symbology: tephras from this work – gray circles; from Kraus et al. (2013) – crosses; lavas from Weaver et al. (1979) – open triangles. Compositional fields: a) basalts; b) basaltic andesites; c) trachiandesites. Fig. 2. Diagrama SiO2 vs. álcalis (TAS) para os piroclastos analisadas neste trabalho, com inclusão de outros dados publicados. Campos composicionais de Le Bas et al. (1986) para rochas vulcânicas e separação das séries alcalinas das subalcalinas através da linha a tracejado definida por MacDonald & Katsura (1964). Simbologia: piroclastos deste trabalho - círculos cinzentos; dados para os piroclastos de Kraus et al. (2013) - cruzes; dados para as lavas de Weaver et al. (1979) – triângulos abertos. Campos composicionais: a) basaltos; b) basaltos andesíticos; c) traquiandesitos.

The tephras samples of Deception Island are subalkaline according to MacDonald (1968) definition, but they are plotted close to that dividing line. On the A(Na2O+K2O) – F(FeO*) – M(MgO) diagram (Fig.3a), the trend produced by Deception tephras implies an initially moderate Fe-enrichment followed by a decrease in Fe content through magma differentiation, with data plotting close to the line proposed by Irvine & Baragar (1971), which divides calc-alkaline from tholeiitic rocks. This is opposite to the pronounced Fe-enrichment trends characteristic of abyssal tholeiites and island arc tholeiites (e.g. Ringwood, 1974).

The low-K nature of all Deception tephras could be shown in the well-known SiO2-K2O diagram (not shown), with most primitive samples lying in the transition between the Low-K (tholeiite) and the calc-alkaline series, while the most silica enriched tephras are within the calc-alkaline field. One of the main characteristics of Deception lavas, formerly reported in the literature (e.g. Weaver et al., 1979), is their high Na2O content (basalts with over 4%); tephras composition obtained in this study (where Na2O ≈ 4-5 wt%) are consisted with these data. Consequently, Na2O/K2O ratios are also high (5.4 to 7.6) and are well within the variation obtained for Deception lavas (4.4 -14, Weaver et al., 1979).

Fig. 3. Discrimination diagrams for tholeiitic (TH) and calc-alkaline (CA) magma series. a) A(Na2O+K2O) – F(FeO*) – M(MgO) diagram for Deception tephras studied in this work; other published data for the island are included for comparison (data sources and symbols as in figure 2). The solid curved line divides tholeiitic and calc-alkaline fields of Irvine & Baragar (1971). b) FeO*/MgO vs. SiO2 (FeO*- total iron expressed as FeO) discrimination diagram with dividing line from Miyashiro (1974). Note the ambiguity conveyed, using both diagrams, in the magma series classification of Deception volcanic rocks. Fig. 3. Diagramas de discriminação para as séries magmáticas toleíticas (TH) e calco-alcalinas (CA). a) Diagrama A(Na2O+K2O) – F(FeO*) – M(MgO) para as rochas piroclásticas da Ilha Deception analisadas neste trabalho; outros dados publicados para esta ilha estão incluídos para comparação (fontes dos dados e símbolos como na figura 2). A linha sólida curva separa os campos toleíticos e calco-alcalinos (Irvine & Baragar, 1971). b) Diagrama discriminante FeO*/MgO vs. SiO2 (FeO* - ferro total expresso sob a forma de FeO) em que a linha separação das duas séries é de Miyashiro (1974). Notar a ambiguidade obtida na classificação das séries magmáticas a que correspondem as rochas vulcânicas da Ilha Deception.

102 P. Ferreira et al. / Comunicações Geológicas (2014) 101, Especial I, 99-104

The magnesium number (Mg#) varies between 53 and 41 (typical values for basaltic andesites). Co-variations among several major oxides and Mg# produce distinct correlations: a) decreasing CaO, Al2O3 and MgO contents with differentiation; b) well correlated increase of alkalis and silica, P2O5 and all incompatible trace element concentrations with decreasing Mg#’s; c) Fe2O3

T and TiO2 increase until Mg# ~ 43, followed by a significant decrease of both ratios, with decreasing Mg# (Fig.4).

Fig. 4. Various element concentrations (Al2O3, CaO, Fe2O3

T, TiO2. Na2O e La) plotted against Mg# for Deception Island tephra obtained in this study. In CaO, TiO2 and La diagrams, other published data for Deception are also included for comparison (data source and symbols as in figure 2). Mg# = Mg/(Mg+Fe2+) in mol%, where Fe2+ = 0.9FeO, representing a differentiation index. All oxides in wt% and La in ppm. Note the greater differentiation interval found for the lavas compared to most tephra from Deception and the general agreement in the correlation trends obtained having all data. Fig. 4. Diagramas de variação para as concentrações de Al2O3, CaO, Fe2O3

T, TiO2. Na2O e La versus Mg#, obtidos para os piroclastos da Ilha Deception analisados neste trabalho. Nos diagramas do CaO, TiO2 e La, estão incluídos para comparação outros dados publicados para esta ilha (fontes dos dados e símbolos tal como na figura 2). Mg# = Mg/(Mg+Fe2+) em mol%, em que Fe2+ = 0.9FeO, e representa um índice de diferenciação. Todos os óxidos em wt% e o La em ppm. Notar o maior intervalo de diferenciação existente para os dados das lavas, comparativamente à maioria dos piroclastos, e a significativa correlação obtida no conjunto com a inclusão de todos os dados.

Representative Masuda-Coryell plots (Fig.5) of the REE data for Deception Island tephras indicate that in all samples the light REE (LREE) are enriched with respect to heavy REE (HREE), compared to chondrites. The critical ratios commonly used to express the degree of fractionation among REE highlight the similar fractionation (although slightly higher for LREE) between LREE / MREE (middle REE) and MREE / HREE ((La/Sm)cn= 1.50-1.70 and (Gd/Yb)cn=1.31-1.42, respectively); this relatively constant enrichment in both REE groups gives rise to an overall REE fractionation expressed by (La/Yb)cn ratios from 2.34 to 2.65.

A spidergram, frequently used to identify the source environment of basalts, summarizing the analytical results produced in this study for Deception Island tephras is

shown in figure 6. In this diagram, the element concentrations are normalized to N-MORB (values from Sun & McDonough, 1989) and a general enrichment in all the represented incompatible elements compared to N-MORB values is noticeable. Moreover, this multi-element diagram shows generally moderate to steep patterns, meaning significant fractionation between the most (Th, U, K2O) and the least (Yb, Lu) incompatible elements. Highly significant are the negative peaks in Nb and Ta normalized concentrations.

Fig. 5. Chondrite-normalized rare earth element (REE) patterns in Deception Island tephras (solid lines). Normalizing values from Sun & McDonough (1989). REE data for other South Shetland Islands (light gray field) and Bransfield Strait rift (dark gray field) are also plotted for comparison (data source as in table 1). Also shown is the typical normalized REE patterns for N-MORB (dotted line) and E-MORB (segmented line) (from Sun & McDonough, 1989). Fig. 5. Padrões de terras raras (REE) normalizadas em relação aos condritos (McDonough & Sun, 1995) para as rochas piroclásticas da Ilha Deception estudadas neste trabalho (linhas sólidas). Igualmente projectados para comparação, encontram-se outros dados de REE para algumas das ilhas Shetland do Sul (campo cinzento-claro) e do rift da bacia de Bransfield (campo cinzento-escuro) (fontes referidas na tabela 1). São também mostrados os padrões normalizados de REE típicos para os MORB-N (linha ponteada) e MORB-E (linha segmentada) (ambos os dados de Sun & McDonough, 1989).

4. Discussion

The limited geochemical variation of the studied tephras, in terms of silica and alkalis (basaltic-andesites; Fig.2), is also supported by the recent published data from Kraus et al.,(2013); however, chemical data obtained in lavas show a broader variation, ranging from basalts to trachydacites (Weaver et al., 1979) (Fig.2). Nevertheless, the Deception Island basalts are more evolved than those from MORB, having lower MgO contents, which is characteristic of island-arc tholeiitic series. The tephras have a variable, but normally high, amount of vesicles (up to ~ 50% volume) that could have resulted from magmas with high volatile content. If this is a primary characteristic associated to magma composition, it is envisaged as a zonation in the magma chamber, where the volatile-rich magmas (already vesiculating) could preferentially occupy its top through density contrast to the remaining volatile-poor magmas. This magma separation could have contributed to generate magmas having limited compositional variation.

Geochemistry of tephras from Deception Island 103

Fig. 6. MORB-N normalized (Sun & McDonough, 1989) trace element patterns for Deception tephras (solid lines) and for volcanics from other locations (as referred in figure 5, with same symbology). Note the highly significant negative anomalies in Nb-Ta, and the positive spike in K2O obtained for Deception Island tephras, supporting the contribution of subducted material, along South Shetland trench, to magma geochemical characteristics associated to quaternary volcanism in Deception Island. To highlight the differences to MORB magmatism, typical normalized values for E-MORB (segmented line) are shown. Fig. 6. Diagrama multielementar para as concentrações normalizadas (relativamente aos MORB-N de Sun & McDonough, 1989) de alguns elementos-traço para as rochas piroclásticas da Ilha Deception (linhas sólidas) e rochas vulcânicas de outros locais (referidos na figura 5, com a mesma simbologia). Notar as significativas anomalias negativas em Nb-Ta, e o pico positivo em K2O, obtido para os piroclastos de Deception corroborando a contribuição do material subductado (ao longo da zona de subdução das ilhas Shetland do Sul) para as características geoquímicas dos magmas percursores do vulcanismo quaternário da Ilha Deception. Encontram-se igualmente representados os valores normalizados para os MORB-E (Sun & McDonough, 1989) para destacar as diferenças relativamente ao magmatismo típico das cristas oceânicas.

Despite the limited tephra compositional variation (e.g. Mg#= 53-41), their geochemical characteristics imply the occurrence of crystal fractionation; in fact, the regular decrease in CaO, Al2O3 and CaO/Al2O3, coupled with an increase of Fe2O3

T with decreasing Mg#’s, supports the fractionation of olivine ± plagioclase ± clinopyroxene (all these mineral phases forming phenocrysts in Deception lavas (e.g. Smellie, 1990). The sharp decrease in iron and titanium concentrations (at ~Mg# =43), indicates titanomagnetite crystallization. The crystal fractionation trend for Deception tephras (also supported by the systematic increase in incompatible elements with differentiation) is well integrated in this more extensive magmatic process defined for lavas (basalts to trachydacites) (Fig.2), suggesting that it is the dominant process relating volcanic rock compositions.

Geochemical data integration from Deception in the magma series’ terminology shows some ambiguity. Taking into consideration the AFM diagram (Fig.3a), the data from this study and other compiled data are projected in the calk-alkaline field; however, the defined trend is identical to that of the tholeiitic magmatic suite which

evolves towards the FeO apex before eventually enriching in the alkalis and evolving toward the alkali apex. Conversely, a calk-alkaline magmatic suite plots on a trend through the middle of the ternary diagram, towards the alkali apex, without showing a trend towards FeO; furthermore, this is coupled with the decrease of total iron and titanium at a late stage (at the level of dacites), which contrasts to our data (decreasing at early stage – basaltic andesite). Using the FeO/MgO vs. SiO2 discrimination diagram (Fig.3b) with dividing line from Miyashiro (1974), almost all data points are plotted in the tholeiitic field.

To obtain a better understanding of the general enrichment in LREE with respect to HREE, shown by Deception tephras, figure 5 also include typical REE patterns characterizing island arc magmatism (volcanics from South Shetland Islands - SSI), REE data from Bransfield Strait ocean floor (BSOF) basaltic rocks (representing the quaternary volcanism in the basin) and a representative E-MORB REE pattern. The Deception patterns have intermediate REE fractionation between those from SSI and BSOF (Tab. 1) and are similar to E-MORB pattern (although having slightly higher values for MREE / HREE fractionation).

Table 1. Significant incompatible element ratios obtained in this study for Deception Island (col. 1), from South Shetland Island arc volcanics (data from Kraus et al. (2013) for Inott Point and Penguin Island) (col. 2), basalts from Bransfield Strait rift (data from Keller et al., 2002), for sample dredges performed near Deception Island) (col. 3) and the typical values for enriched MORB’s (Sun & McDonough, 1989). La/Yb, La/Sm and Gd/Yb are chondrite normalized ratios. K/La, La/Nb and Th/Ta are N-MORB normalized ratios (N-MORB compositions from Sun & McDonough, 1989). Note the intermediate ratios obtained for Deception samples in comparison to those from other South Shetland Islands and those from Bransfield Strait. Tabela 1. Algumas razões significativas entre elementos incompatíveis obtidas neste estudo para as amostras da Ilha Deception (col. 1), para as rochas vulcânicas de outras ilhas das Shetland do Sul (dados de Kraus et al. (2013) para as ilhas Inott Point e Penguin) (col. 2), para os basaltos do rift da bacia de Bransfield (dados de Keller et al., 2002) para as amostras dragadas perto da Ilha Deception) (Col. 3), e os valores característicos dos MORB enriquecidos (from Sun & McDonough, 1989). As razões La/Yb, La/Sm e Gd/Yb são razões normalizadas relativamente aos condritos. As razões K/La, La/Nb e Th/Ta são razões normalizadas à composição dos MORB-N (valores de Sun & McDonough, 1989). Notar as razões intermédias obtidas para as amostras de Deception relativamente às apresentadas pelas amostras das ilhas Shetland do Sul e do estreito de Bransfield.

The clear geochemical distinction between Deception and MORB-type rocks is highlighted when normalizing element concentrations to N-MORB (Fig.6); the resulting strong Nb-Ta negative anomaly, coupled with a positive

104 P. Ferreira et al. / Comunicações Geológicas (2014) 101, Especial I, 99-104

spike in K2O normalized values, are both key geochemical signatures also present in SSI and BSOF, all noticeably distinct from E-MORB pattern. Regarding trace element systematic, arc magmas are characterized by low HFSE contents with respect to LILE and LREE, and most magmas also show selective enrichments in LILE with respect to LREE (e.g. Turner & Hawkesworth, 1997). Selected discriminating ratios are shown in table 1 (K/La, La/Nb and Th/Ta, all ratios normalized to N-MORB) supporting the multi-element diagram interpretation (Fig.6) and emphasizing the important role of subducted contribution material to the melting sources of the recent (quaternary age) volcanism, not only in Deception Island but also in the Bransfield Strait ocean floor.

References

Barker, P.F., Dalziel, I.W.D., Storey, B.C., 1991. Tectonic development of the Scotia Arc Region. In: R.J. Tingey, (Ed.). Antarctic Geology. Oxford University Press, Oxford, UK, 215-248.

Dalziel, I.W.D., 1984. Tectonic evolution of a Forearc Terrane. Southern Scotia Ridge, Antarctica. Geological Society of America Special Papers. 200, 32 p.

Gvirtzmann, Z., Nur, A., 2001. Residue to topography, lithospheric structure andsunken slabs in the central Mediterranien. Earth and Planetary Science Letters, 187, 117-130.

Irvine, T.N., Baragar, W.R.A., 1971. A guide to the chemical classification of the common volcanic rocks. Canadian Journal of Earth Sciences, 8, 523-548.

Keller, R.A., Fisk, M.R., Smellie, J.L., Strelin, J.A., Lawver L.A., 2002. Geochemistry of back arc basin volcanism in Bransfield Strait, Antarctica: Subducted contributions and along-axis variations. Journal of Geophysical Research, 107(B8), 2171, doi:10.1029/2001JB000444.

Kraus, S., Kurbatov, A, Yates, M., 2013. Geochemical signatures of tephras from Quaternary Antarctic Peninsula volcanoes. Andean Geology, 40, 1-40.

Lawver, L.A., Sloan, B.J., Barker, D.H.N., Ghidella, M., Von Herzen, R.P., Keller, R.A., Klinkhammer, G.P., Chin, C.S., 1996. Distributed, active extension in Bransfield Basin, Antarctic

Peninsula: evidence from multibeam bathymetry. GSA Today, 6, 1-6.

Le Bas, M.J., Le Maitre, R.W., Streckeisen, A., Zanettin, B., 1986. A chemical classification of volcanic rocks based on the total alkali-silica diagram. Journal of Petrology, 27, 745-750.

MacDonald, G.A., 1968 Composition and origin of Hawaiian lavas. In: R.R. Coats, R.L. Hay, C.A. Anderson, (Eds.). Studies in Volcanology: A Memoir in Honour of Howell Williams. The Geological Society of America Memoir, 116, 477-522.

MacDonald, G.A., Katsura, T., 1964. Chemical compositions of Hawaiian lavas. Journal of Petrology, 5, 82-133.

Mão de Ferro, A., Mota, A.M., Canário, J., 2013. Sources and transport of As, Cu, Cd and Pb in the environmental compartments of Deception Island, Antarctica. Marine Pollution Bulletin, 77, 341-348.

Martí, J., Geyer, A., Aguirre-Diaz, G., 2013. Origin and evolution of the Deception Island caldera (South Shetland Islands, Antarctica). Bulletin of Volcanology, 75, 1-18.

Miyashiro, A., 1974. Volcanic rock series in island arcs and active continental margins. American Journal of Science, 274, 321-355.

Ringwood, A.E., 1974. The petrological evolution of island arc systems. Journal of the Geological Society of London, 130, 183-204.

Smellie, J.L., 1990. Graham Land and South Shetland Islands. In: W.E. LeMasurier, J.W. Thomson, (Eds.). Volcanoes of the Antarctic Plate and Southern Oceans. AGU Antarctic Research, 48, 302-353.

Smellie, J.L., Pankhurst, R.J., Thomson, M.R.A., Davies, R.E.S., 1984. The geology of the South Shetland Islands, VI, Stratigraphy, geochemistry and evolution, Br. Antarctic Survey Scientific Report, 87, 85 p.

Sun, S.-S., McDonough, W.F., 1989. Chemical and Isotopic Systematics of oceanic basalts: implications for Mantle Composition and Processes. In: A.D. Saunders, M.J. Norry, (Eds). Magmatism in the Ocean Basins. Geological Society, London, Special Publications, 42, 313-345.

Turner, S.P., Hawkesworth, C.J., 1997. Constraints on flux rates and mantle dynamics beneath island arcs from Tonga-Kermadec lava geochemistry. Nature, 389, 568- 573.

Weaver, S.D., Saunders, A.D., Pankhurst, R.J., Tarney, J., 1979. A geochemical study of magmatism associated with the initial stages of back-arc spreading. Contributions to Mineralogy and Petrology, 68, 151-169.