<<

Giving birth to volcanoes: Distribution and composition of young from the seafloor near and Pitcairn

C.W. Devey   Fachbereich 5—Geowissenschaften, Universita¨t Bremen, D-28334 Bremen, Germany K.S. Lackschewitz  D.F. Mertz Institut fu¨r Geowissenschaften, Universita¨t Mainz, Becherweg 21, D-55099 Mainz, Germany, and Max Planck Institut fu¨r Chemie, D-55128, Mainz, Germany B. Bourdon Laboratoire de Ge´ochemie, Institut de Physique du Globe Paris, place Jussieu, F-75252 Paris Cedex 05, France J.-L. ChemineÂe   Institut de Physique du Globe, place Jussieu, F-75252 Paris Cedex 05, France J. Dubois  C. Guivel Laboratoire Pe´trologie Structurale, Universite´ de Nantes, F-44322 Nantes, France R. HeÂkinian Institut Francais de Recherche pour l’Exploitation de la Mer, F-29280 Plouzane´, France P. Stoffers Institut fu¨r Geowissenschaften, Universita¨t Kiel, D-24118 Kiel, Germany

ABSTRACT samples produced by the hotspot. These youn- Apart from being popular holiday destinations, oceanic- volcanoes such as , ger volcanoes can be further subdivided into Tahiti, or the Canaries provide that yield valuable information about the interior small isolated volcanoes (height above sea- of our planet. Until recently, studies have concentrated on the easily accessible, subaerial ¯oor, Յ500 m), large isolated volcanoes parts of the volcanoes, largely ignoring their earlier-formed, submarine parts. These sub- (height above sea¯oor, Ͼ1500 m) and small marine parts, however, provide critical information about how the begins to melt parasite cones on the ¯anks of the large edi- and about the lowest-melting-point mantle componentsÐinformation not available from ®ces. In the Pitcairn area, eight recently the subaerial volcanoes but highly relevant for the chemical evolution of the whole mantle. formed small volcanoes and three large vol- We present here compositional information from small (Ͻ500 m) volcanoes on the sea¯oor canoes were mapped and sampled over an area near Tahiti and and show that these small volcanoes erupt only highly of ϳ50 ϫ 70 km. At the , ®ve differentiated magmas. These early melts are derived exclusively from the most trace small volcanoes and ®ve larger volcanoes element±enriched, isotopically extreme mantle component, evidence that this component have been mapped and sampled over an area has the lowest melting temperature and is the ®rst product of melting of a new batch of of 150 ϫ 80 km. Two of these volcanoes, Ro- mantle. The geochemical mantle components (enriched mantle EM-I, EM-II) proposed in card and Turoi, rise some 1000 m above the the 1980s to explain the compositional variations among oceanic volcanoes worldwide sea¯oor, occupying a transitional place in our appear in reality to represent distinct masses in the mantle. size classi®cation, a situation also re¯ected by their compositions, as outlined Keywords: , geochemistry, plume, enriched mantle, EM-I, EM-II, melting, plum pud- subsequently. ding, evolution, hotspot. Major element analyses on fresh glass chips show a clear distinction in degree of magmatic INTRODUCTION of present-day submarine volcanic activity. differentiation between the large and small Many chains of intraplate oceanic-island Although such sampling has been carried out volcanoes (see Fig. 2A). This is particularly volcanoes are built as lithospheric plates move on Loihi in the Hawaii chain, this striking for the Society hotspot, where all over stationary melt sources (hotspots) in the seamount is already relatively large and pres- large-volcano magmas have MgO Ͼ 3%, . A hotspot is probably main- ently erupting magmas similar to those found whereas the smaller volcanoes all yield te- tained over long periods by an adiabatically on the islands (e.g., Garcia et al., 1995) and phritic or magmas with upwelling mantle diapir or plume (Morgan, so is not ideal for studying the onset of vol- MgO Ͻ 2% characterized by high volatile 1971). The degree of partial melting of the canism. We present here geochemical analyses contents. Rocard and Turoi volcanoes have plume mantle will vary both laterally (owing from samples obtained from the submarine ed- yielded many and a handful of ba- to radial gradients in plume temperature from i®ces of the Pitcairn and Society hotspots saltic samples. A similar situation, with more a hot center to a cool rim; Loper and Stacey, (e.g., Stoffers et al., 1988; Stoffers et al., basic magmas being found on the larger vol- 1983) and vertically (owing to the effect of 1990a, 1990b; Binard et al., 1992a) and use canoes, is seen at Pitcairn, although here some pressure on the solidus; e.g., Farnetani and them to develop a petrogenetic model for the evolved magmas are also found on the larger Richards, 1995). The volcanoes will therefore initial stages of hotspot . edi®ces, most notably as a trachytic dome cap- be fed by melts formed at different tempera- ping the apparently extinct Adams volcano tures and pressures during their growth. Par- RESULTS AND DISCUSSION (Stoffers et al., 1990a). Nevertheless, no sam- ticularly interesting in this respect are the ini- On bathymetric maps (Fig. 1) we distin- ple with MgO Ͼ 3.5% has ever been recov- tial phases of volcano growth, because they guish two major types of volcanoes in these ered from the small Pitcairn volcanoes. Liquid should be fed by melt formed either at the rim hotspot areas. The ®rst type yields only Mn- line of descent modeling at both hotspots is of the plume as it passes beneath previously encrusted samples (e.g., Glasby et al., 1997; compatible with derivation of evolved mag- unaffected (Frey et al., 2000) or Puteanus et al., 1989), with mid-oceanic-ridge mas by extensive crystal fractionation of basic deep in the plume as the upwelling mantle ®rst (MORB) compositions (e.g., Devey et parent magmas similar to those found on the crosses the solidus. Although deep drilling on al., 1990). These volcanoes are old and not adjacent large volcanoes (Devey et al., 1990). oceanic islands (e.g., Stolper et al., 1996) can related to hotspot activity and will not be dis- The Pitcairn and Society hotspots are typi- start to examine the initial phases of volcano cussed further in this paper. The second type cal examples of EM-I and EM-II hotspots, re- growth, it must be complemented by sampling comprises volcanoes yielding young, fresh spectively (EM is enriched mantle; see Zindler

᭧ 2003 Geological Society of America. For permission to copy, contact Copyright Permissions, GSA, or [email protected]. Geology; May 2003; v. 31; no. 5; p. 395±398; 3 ®gures; Data Repository item 2003055. 395 and Hart, 1986). Lava compositions from these volcanoes form linear arrays in a Sr ver- sus Nd isotope diagram (White and Hofmann, 1982; Woodhead and Devey, 1993; Woodhead and McCulloch, 1989). Isotopic data from the seamounts (Fig. 2B) show the evolved mag- mas from the small volcanoes and Rocard to lie at the extreme high 87Sr/86Sr and low 143Nd/144Nd end of the EM-I and EM-II arrays for the present-day Pitcairn and Society hot- spots, respectively. We note that older parts of the Society chain (e.g., Tahaa Island; White and Duncan, 1996) have yielded with more extreme isotopic compositions than those presently erupting on the small Society volcanoes. We attribute this to long-term var- iations in the composition of the enriched So- ciety source component and as such not in con¯ict with our observations; we would ex- pect small volcano trachytes erupted at the time of formation of Tahaa to show 87Sr/86Sr ratios close to 0.707.

MODEL FOR THE INITIATION OF HOTSPOT VOLCANISM Seismological observations from the Soci- ety area show that both large basaltic and small trachytic volcanoes are currently active (Talandier, 1989; Talandier and Kuster, 1976; Talandier and Okal, 1984). The systematic freshness of samples recovered at Pitcairn im- plies that the same is true there. The large number of small compared to large or subaer- ial volcanoes currently active at both hotspots suggests that many more volcanic systems are initiated than ever reach maturity. The mag- mas entering these initial systems from the mantle are most probably basic; the eruptive products appear, however, to be exclusively highly fractionated and trachytes. This ®nding points to extensive fractionation in the lithosphere, either because the conduit system is initially cold and saps the magmas of heat before they can reach the sur- face, or because, in the absence of a continu- ous magma supply from the mantle, the mag- mas can reach the surface only when fractionation has suf®ciently lowered their density or increased their volatile pressure. Whatever the reason for their extreme differ- Figure 1. Bathymetry of sea¯oor southeast of (A) Pitcairn Island (modi®ed after HeÂkinian entiation, the initial magmas seem to be de- et al., 2003) and (B) Tahiti Island. Volcano shading codes: Light grayÐold non-hotspot rived, without exception, from the mantle volcanoes. SpeckledÐlarge, recently active volcanoes associated with hotspot. Dark grayÐsmall, recently active volcanoes associated with hotspot. WhiteÐhotspot volca- source with the highest time-integrated incom- noes of intermediate height. A: Dredge sampling stations (PNDRÐfrom POLYNAUT patible trace element enrichment. If these ini- cruise, DSÐfrom cruise 65 of FS Sonne) are shown. Note that volcano at 25؇35؅S, tial magmas re¯ect the onset of melting in a -129؇30؅W is large enough to be classi®ed as a large volcano (and hence falls in same particular mantle volume, then this observa size classi®cation as Bounty and Adams); however, it has yielded only old samples. Volcano names and numbers after Binard et al. (1992b). Samples PNDR13 and 50DS come tion implies that the most trace element± from a on southern ¯ank of Bounty volcano. Inset: Location of Polynesian enriched mantle component has the lowest hotspots in South Paci®c. East Paci®c Rise (EPR) and are visible near melting point. This inference lends some di- right-hand margin. B: Dredge sampling stations (DTHÐfrom 1989 CYAPOL submersible rect observational support to the suggestion, cruise, DRÐfrom 1985 cruise of NO Charcot, 47 and 65 from cruises 47 and 65 of FS based either on theoretical or trace-element Sonne, respectively) and dive tracks (TH) are shown. Volcano names and numbers after Binard et al. (1992a). Asterisk on the summit of Teahitia is the location of samples 65- modeling considerations (e.g., Batiza, 1984; 119GTV, 65-120GTV, 47-5DS, 47-20DS, 47-22GTV. Zindler et al., 1984; Graham et al., 1988;

396 GEOLOGY, May 2003 Phipps Morgan and Morgan, 1999) that melt- ing in the mantle initially fuses a low-melting- point, trace element±enriched component pre- sent as ``plums'' in the mantle. Furthermore, the correlation of initial melting with extreme isotopic compositions implies that the mantle end members EM-I and EM-II originally pro- posed to explain hotspot (Zindler and Hart, 1986) do actually corre- spond to distinctive mantle rock types with relatively low solidus temperatures. Figure 3 shows our model for the and mantle around the Polynesian hotspots. At the onset of melting deep in the plume, trace element±enriched melts from the enriched mantle plums are formed. These melts either (1) rise into the lithosphere where they stag- nate at a density boundary (Moho?) or (2) are channeled into preexisting magma-pathways related to the larger volcanoes in the area. In case 1, the magmas will fractionate, and the residual magma will eventually erupt as tra- chytes or phonolites on the small volcanoes as previously outlined. In case 2, the enriched magmas become mixed with melts generated shallower in the plume at larger degrees of melting. They provide the enriched mantle end member for the range of isotopic values seen in the larger volcanoes. The facts that the larger volcanoes are (1) fewer in number than the small volcanoes and (2) predominantly basaltic or basanitic in composition suggest that at some point, one or a few of the initial magma pathways will become more established. Magma will then be channeled into them regularly, and the thermal Figure 2. Composition of volcanic samples from submarine Pitcairn and Society hotspots. conditions for extensive fractionation will no A: SiO vs. MgO plot using previously published data (Binard et al., 1992a, 1992b; Devey et 2 longer (or in the case of Pitcairn less often) al., 1990; HeÂmond et al., 1994; Woodhead and Devey, 1993) and our own analyses (Table DR11). Different volcano types are described in text and are visible in Figure 1. Fields show be met. With time, one of these volcanoes may range of compositions of large volcanoes from Pitcairn and Society hotspots. B: 87Sr/86Sr vs. 143Nd/144Nd isotope diagram. As previously shown (White and Hofmann, 1982; Woodhead 1GSA Data Repository item 2003055, micro- and McCulloch, 1989), Pitcairn and Society hotspots lie on EM-I and EM-II isotopic trends probe analyses of volcanic glasses from the Pitcairn (Zindler and Hart, 1986) as marked by trends of samples from neighboring islands (Duncan and Society hotspots, is available on request from et al., 1994; White, 1985; White and Duncan, 1996), respectively. Data from Devey et al. Documents Secretary, GSA, P.O. Box 9140, Boul- (1990), HeÂmond et al. (1994), Woodhead and Devey (1993), and present work (Table DR2; der, CO 80301-9140, USA, [email protected], see footnote one). or at www.geosociety.org/pubs/ft2003.htm.

Figure 3. Model for magma-plumbing system beneath Society and Pitcairn hotspots. See text for discussion.

GEOLOGY, May 2003 397 become the focus of almost all eruptionsÐin Binard, N., HeÂkinian, R., and Stoffers, P., 1992b, and Kunzendorf, H., 1989, Distribution, inter- this case an island may be formed. The oc- Morphostructural study and type of volcanism nal structure and composition of manganese of submarine volcanoes over the Pitcairn hot crusts from seamounts east of Teahitia- casional, more evolved magmas occurring spot in the South Paci®c: Tectonophysics, Mehetia hotspot, S.W. Paci®c: Marine Mining, even on the larger Pitcairn edi®ces may re¯ect v. 206, p. 245±264. v. 8, p. 245±266. a generally lower rate of magma supply at Pit- Devey, C.W., AlbareÁde, F., ChemineÂe, J.-L., Mi- Stoffers, P., Botz, R., Hartmann, M., KoÈgler, F., cairn (as indicated by the intermittent nature chard, A., MuÈhe, R., and Stoffers, P., 1990, MuÈhe, R., and Puteanus, D., 1988, Recent hot- Active submarine volcanism on the Society spot-related volcanism in the Austral Islands: of the hotspot trace since 15 Ma [Duncan et hotspot swell (west Paci®c): A geochemical The geological structure of the Macdonald al., 1974] when compared to the Societies) study: Journal of Geophysical Research, v. 95, Seamount: Meyniana, v. 40, p. 21±29. that allows extensive cooling and fractionation p. 5049±5066. Stoffers, P., and the Scienti®c Party, 1990a, Active of the magma to occur even in a large vol- Duncan, R.A., McDougall, I., Carter, R.M., and found: Marine Geology, v. 95, cano. It is interesting to note, however, that Coombs, D.S., 1974, Pitcairn IslandÐAnother p. 51±55. Paci®c hot spot: Nature, v. 251, p. 679±682. Stoffers, P., Hekinian, R., and the Scienti®c Party, the and trachytes on these Duncan, R.A., Fisk, M.R., White, W.M., and Nielsen, 1990b, Cruise Report SONNE 65ÐMidplate large Pitcairn volcanoes do not show the ex- R.L., 1994, Tahiti: Geochemical evolution of a II. Hotspot volcanism in the central South Pa- treme isotopic values characteristic of evolved French Polynesian volcano: Journal of Geo- ci®c: Berichte-Reports Geologie-PalaÈontolo- magmas from the smaller edi®ces; they are physical Research, v. 99, p. 24,341±24,357. gie Institut University zu Kiel, v. 40, 223 p. Farnetani, D.G., and Richards, M.A., 1995, Thermal Stolper, E.M., DePaolo, D.J., and Thomas, D.M., isotopically indistinguishable from the entrainment and melting in mantle plumes: 1996, Introduction to special section: Hawaii erupted on the same volcano (Woodhead and and Planetary Science Letters, v. 136, Scienti®c Drilling Project: Journal of Geo- Devey, 1993). Our model for the evolution of p. 251±267. physical Research, v. 101, p. 11,593±11,598. Polynesian hotspot volcanoes therefore in- Frey, F.A., Clague, D., Mahoney, J.J., and Sinton, Talandier, J., 1989, Submarine volcanic activity; volves a two-stage processÐthe initiation of J.M., 2000, Volcanism at the edge of the Ha- Detection, monitoring and interpretation: Eos waiian plume: Petrogenesis of submarine al- (Transactions, American Geophysical Union), many small-volume volcanic systems derived kalic lavas from the North Arch volcanic ®eld: v. 70, p. 561, 568±569. from melting of the enriched mantle plums, Journal of Petrology, v. 41, p. 667±691. Talandier, J., and Kuster, G.T., 1976, Seismicity and followed by focusing of magmatic activity in Garcia, M.O., Foss, D.J.P., West, H.B., and Maho- submarine volcanic activity in French Poly- only a few of these systems. Whether the ney, J.J., 1995, Geochemical and isotopic evo- nesia: Journal of Geophysical Research, v. 81, lution of Loihi volcano, Hawaii: Journal of Pe- p. 936±948. small volcanoes that we currently observe at trology, v. 36, p. 1647±1674. Talandier, J., and Okal, E.A., 1984, The volcanoseis- both hotspots are coeval with the initiation of Glasby, G.P., StuÈben, D., Jeschke, G., Stoffers, P., mic swarms of 1981±1983 in the Tahiti- the larger volcanoes or whether they represent and Garbe-SchoÈnberg, C.-D., 1997, A model Mehetia area, French Polynesia: Journal of a more recent renewed phase of magmatic sys- for the formation of hydrothermal manganese Geophysical Research, v. 89, p. 11,216±11,234. tem initiation is at present not known. The So- crusts from the Pitcairn Island hotspot: Geo- White, W.M., 1985, Sources of oceanic basalts: Ra- chimica et Cosmochimica Acta, v. 61, diogenic isotopic evidence: Geology, v. 13, ciety volcanoes Rocard and Turoi, from which p. 4583±4597. p. 115±118. some basalts and numerous trachytes have Graham, D.W., Zinlder, A., Kurz, M.D., Jenkins, White, W.M., and Duncan, R.A., 1996, Geochem- been recovered, may currently be in the pro- W.J., Batiza, R., and Staudigel, H., 1988, He, istry and geochronology of the Society Is- cess of focusing magmas and will perhaps be Pb, Sr and Nd isotope constraints on magma lands: New evidence for deep mantle recy- genesis and mantle heterogeneity beneath cling, in Basu, A., and Hart, S.R., eds., Earth the next large underwater Society volcanoes. young Paci®c seamounts: Contributions to processes: Reading the isotopic code: Ameri- Mineralogy and Petrology, v. 99, p. 446±463. can Geophysical Union Geophysical Mono- ACKNOWLEDGMENTS HeÂkinian, R., ChemineÂe, J.-L., Dubois, J., Stoffers, graph 95, p. 183±206. The research cruises that yielded the samples and P., Scott, S., Guivel, C., Garbe-SchoÈnberg, D., White, W.M., and Hofmann, A.W., 1982, Sr and Nd maps forming the basis for this project were funded Devey, C., Bourdon, B., Lackschewitz, K., isotope geochemistry of oceanic basalts and by the German Bundesministerium fuÈr Bildung und McMurtry, G., and Le Drezen, E., 2003, The mantle evolution: Nature, v. 296, p. 821±825. Forschung and Deutsche Forschungsgemeinschaft Pitcairn hotspot in the South Paci®c: Distri- Woodhead, J.D., and Devey, C.W., 1993, Geochem- and the French Centre National de la Recherche bution and composition of submarine volcanic istry of the Pitcairn seamounts: I. Source char- Scienti®que. We thank the captains and crews of the sequences: Journal of Volcanology and Geo- acter and temporal trends: Earth and Planetary research vessels Sonne and Atalante and the Nautile thermal Research (in press). Science Letters, v. 116, p. 81±99. submersible team for their un¯agging assistance. R. HeÂmond, C., Devey, C.W., and Chauvel, C., 1994, Woodhead, J.D., and McCulloch, M.T., 1989, An- Duncan and P. Castillo provided thoughtful, con- Source compositions and melting processes in cient sea¯oor signals in Pitcairn Island lavas structive reviews that helped improve the manu- the Society and Austral plumes (South Paci®c and evidence for large amplitude, small script markedly. We also thank A.W. Hofmann of Ocean): Element and isotope (Sr, Nd, Pb, Th) length-scale mantle heterogeneities: Earth and the Max-Planck-Institut fuÈr Chemie, Mainz, for ac- geochemistry: Chemical Geology, v. 115, Planetary Science Letters, v. 94, p. 257±273. cess to isotope facilities. p. 7±45. Zindler, A., and Hart, S., 1986, Chemical geodyn- Loper, D.E., and Stacey, F.D., 1983, The dynamical amics: Annual Review of Earth and Planetary Sciences, v. 14, p. 493±571. REFERENCES CITED and thermal structure of deep mantle plumes: Zindler, A., Staudigel, H., and Batiza, R., 1984, Iso- Batiza, R., 1984, Inverse relationship between Sr Physics of Earth and Planetary Interiors, v. 33, tope and trace element geochemistry of young isotope diversity and rate of oceanic volca- p. 304±317. Paci®c seamounts: Implications for the scale nism has implications for mantle heterogene- Morgan, W.J., 1971, Convection plumes in the low- of heterogeneity: Earth and Plan- ity: Nature, v. 309, p. 440±441. er mantle: Nature, v. 230, p. 42±43. etary Science Letters, v. 70, p. 175±195. Binard, N., HeÂkinian, R., ChemineÂe, J.-L., and Stof- Phipps Morgan, J., and Morgan, W.J., 1999, Two- fers, P., 1992a, Styles of eruptive activity on stage melting and the geochemical evolution Manuscript received 2 October 2002 intraplate volcanoes in the Society and Austral of the mantle: A recipe for mantle plum- Revised manuscript received 20 January 2003 hot spot regions: Bathymetry, petrology, and pudding: Earth and Planetary Science Letters, Manuscript accepted 28 January 2003 submersible observations: Journal of Geo- v. 170, p. 215±239. physical Research, v. 97, p. 13,999±14,015. Puteanus, D., Glasby, G.P., Stoffers, P., Mangini, A., Printed in USA

398 GEOLOGY, May 2003