Journal of Volcanology and Geothermal Research 136 (2004) 159–163 www.elsevier.com/locate/jvolgeores Discussion Comments on ‘‘Epiclastic deposits and ‘‘horseshoe-shaped’’ caldeiras in () and Ua Huka (Marquesas Archipelago), ’’ by Cle´ment et al. (2003)$

Pierre-Yves Gillota,*, Anthony Hildenbranda,1, Vale´rie Clouardb

a Lab. de Ge´ochronologie UPS-IPGP, Dpt. Sciences de la Terre, UPS, 91405 Orsay, b Departamento de Geofı´sica, Univ. de Chile, Santiago, Chile Received 9 July 2003; accepted 8 March 2004

1. Introduction ent et al. (2003) in the north of Tahiti-Nui, several issues can be raised. They focus on: (1) the level and the One of the main features in the evolution of dimension of the northern landslide; (2) the nature and oceanic shield volcanoes is the occurence of huge facies of the breccia deposits; (3) the age of the Tahiti- landslides affecting their flanks. It is thus relevant to Nui northern landslide; (4) the cause of the Tahiti-Nui accurately determine the morphology, the extension landslides; and (5) the post-sliding evolution. and the related breccia deposits in order to constrain the conditions of sliding. 1.1. Morphological level and dimension of Tahiti-Nui Cle´ment et al. (2003) recently described epiclastic northern landslide deposits in the north of Tahiti-Nui (Society Islands) and in the island of Ua Uka (Marquesas archipelago). On Tahiti-Nui, Cle´ment et al. (2003) describe a They claim that it is the first time that such products ‘‘debris avalanche’’ unit which covers most of the floor are interpreted as debris avalanches resulting from a of the present central Maroto depression and the bottom huge landslide event. However, debris avalanche of the Papenoo river (Fig. 1). They present the mor- products from a southern landslide have been already phological system of Maroto basin and Papenoo river described by Clouard et al. (2001) to the south of as a horseshoe-shaped caldera resulting from a huge Tahiti-Nui. Moreover, a large landslide event in the landslide. The epiclastic deposits they described con- north of Tahiti-Nui has been previously proposed and tain poorly sorted breccia lava blocks ( < 10 m in size) dated (Gillot et al., 1993; Le Roy, 1994). within a matrix constituted by clay and sand. From the interpretation of the landslide and the The level described by these authors, which out- related epiclastic breccia deposits described by Cle´m- crops at the surface of the central Maroto basin and Papenoo river, represents obviously a recent morpho- $ doi of original article 10.1016/S0377-0273(02)00366-9. logical level. Indeed, erosion is very intense in such a * Corresponding author. Laboratoire de Ge´ochronologie, De- wet tropical climate and it has been shown that the partment des Sciences de la Terre, Universite Paris-Sud, Batiment central part of the island has experienced a vertical 504, Orsay F-91405, France. loss of thousands of meters over the past 500 ka E-mail address: [email protected] (P.-Y. Gillot). 1 Present address: Lab. de Ge´odynamique des Rifts et des (Gillot et al., 1993; Le Roy, 1994). Thus, the present Marges Passives, Universite´ du Maine, Avenue Olivier Messiaen geometry of the depression cannot fossilize the struc- 72085 Le Mans Cedex 9, France. tural scar of a landslide generated 450 ka ago: nor its

0377-0273/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jvolgeores.2004.03.011 160 P.-Y. Gillot et al. / Journal of Volcanology and Geothermal Research 136 (2004) 159–163

Fig. 1. Shaded relief map of Tahiti-Nui (sunlight from North). Bold solid lines indicate the main E–W rift zone from Gillot et al. (1993). Dotted lines for the inferred rims of the landslides, respectively, to the North (Gillot et al., 1993) and to the South (Clouard et al., 2001). White line for the rims of the present central Maroto depression. Ages are indicated in Ma from Le Roy (1994): bold characters for the primitive shield; normal characters for the post-collapse northern shield; italic character for the post-erosion, valley-filling flows. V.: Vaitamanu valley; Vir.: Viriviriterai plateau; Tam.: Tamanu plateau. walls, neither its floor. The current shape of the bank of Vaitamanu river (Fig. 1), much higher than depression, which significantly differs from those of the present floor of the Maroto basin. a ‘‘horseshoe-shaped caldera’’, results obviously from a recent ongoing morphological evolution. 1.2. Nature and facies of the breccia deposits A huge northern landslide has been previously identified and described by Gillot et al. (1993) and Erosion of the volcanic structure of Tahiti-Nui is Le Roy (1994). Its extension is much larger than that responsible for the development of widespread scree proposed by Cle´ment et al. (2003). It has a clear and mud flow deposits (Deneufbourg, 1965), such as horseshoe shape. It is limited by the main E–W rift lahars (e.g. Boutault, 1985). Surprisingly, Cle´ment et zone to the south, whereas the lateral rims are under- al. (2003) do not refer to any previous study, but these lined by secondary concentration of dykes in the units have been already recognized and mapped in Tipaerui river to the north-west and between Tahaute most of the largest valleys (Deneufbourg, 1965; river (Le Roy, 1994) and Onoheha river (Hildenbrand, Brousse et al., 1985; Brousse and Gelugne, 1987; 2002) to the north-east (Fig. 1). Soon after the slide, a Brousse et al., 1990). In the Papenoo valley (Fig. 1), second shield immediately built into and filled up the similar units and post-erosional valley-filling flows, northern depression. Its remnants correspond today to dated around 200 ka (Gillot et al., 1993; Le Roy, the highest relief of the island (Aorai, Orohena and 1994), stand at various topographic levels, indicating Pito Iti). This second shield construction began with a a multi-stage evolution. The digging of the central thick brecciated lava sequence. The scarp of the Maroto depression has thus occurred through a grad- sliding is located between this sequence and the lavas ual and rather complex way, probably influenced by of the first shield stage. These lavas are preserved at the eustatic variations of the sea-level (Becker et al., the base of the southern wall, i.e. along the southern 1974). P.-Y. Gillot et al. / Journal of Volcanology and Geothermal Research 136 (2004) 159–163 161

The epiclastic deposits described by Cle´ment et al. of the post-collapse activity. On Tahiti-Nui, this has (2003) may typically result from the rapid dissection been the approach followed by Gillot et al. (1993) and and erosion of the volcanic structure, induced by the Le Roy (1994) to accurately date a major catastrophic intense tropical precipitation on the high and young northern landslide between 870 and 850 ka, i.e. 400 volcanic relief. This can explain the high energy of the ka before the age proposed by Cle´ment et al. (2003) breccia flow deposits, which led to the confusion with for the landslide. large landslide deposits. In fact, the deposits described by Cle´ment et al. 1.4. Cause of the Tahiti-Nui landslides (2003) correspond to water saturated breccias with various facies including debris flows, hyperconcen- Cle´ment et al. (2003) suggest that the debris trated flows and muddy streamflows (Vallance, 2000). avalanche occurred at the end of the shield-building The lithics range from metric blocks to infra-millimet- stage as a consequence of the emplacement of the ric grains inside a muddy matrix. The water circulation plutonic body which occupies the central part of the is responsible for a cementation of the deposits. A Maroto depression. large proportion of the blocks have clearly been rolled; This interpretation appears rather unrealistic as the there is no clear evidence of jigsaw cracks in the larger coarse-grained plutonic rocks are generated by a blocks, but most of them are indurated breccia blocks gentle cooling at depth, which implies the presence with small pebbles reworked in the lahar deposit (as of a thick overlying volcanic pile. The sudden de- shown in the photo, Fig. 3, in the work of Cle´ment et compression due to a major landslide induces, on the al., 2003). None of the blocks are larger than a few contrary, a rapid and voluminous magma extrusion as meters. discussed by Le Roy (1994). The major northwards destructive event (Gillot et 1.3. Age of the landslide al., 1993), as well as the southern landslide (Clouard et al., 2001), have been favored by the concentration In order to date the event, Cle´ment et al. (2003) use of the magmas through a main E–W rift zone. Gillot a mean K–Ar age of 565 F 9 ka which they obtained et al. (1993) and Le Roy (1994) have related this from a lava block included in the breccia deposits, and event to the presence of hundreds of dykes, which the age of 440 F 50 ka proposed by Becker et al. induced a strong dilatation of the primitive shield, (1974) for a valley flow covering similar breccia 10 probably triggering the two slides in opposed direc- km downwards along the Papenoo river. The latter tions, oriented perpendicularly to the E–W rift zone. flow has been more accurately dated at 227 F 3ka(Le They affected the northern and the southern flanks Roy, 1994). of the primitive shield-volcano, for which the ex- Because pebble dating provides a maximum age posed activity ranges from 1.4 to 0.87 Ma (Fig. 1). for the related erosion process, the age of 565 ka is Most of the northern depression was then rapidly then not surprising since the main shield-building filled by the second shield products (Fig. 1), until phase on Tahiti-Nui ended about 500 ka ago (e.g. the erupted lavas overflowed the depression walls Gillot et al., 1993; Hildenbrand and Gillot, 2001). The about 650 ka ago (Gillot et al., 1993; Le Roy, 1994; lower temporal bound inferred by Cle´ment et al. Hildenbrand, 2002). Then the flows draped the (2003) for their landslide event relies on the hypoth- preserved slopes of the primitive shield or filled esis that the so-called ‘‘debris avalanches’’ deposits existing valleys. Finally, late lavas were extensively correspond to a ‘‘lahar’’ formation observed by trapped by the southern landslide depression Becker et al. (1974) several kilometers away, in the (Clouard, 2000; Clouard et al., 2001), about 550 lower part of the Papenoo river. However, the lack of ka ago (Hildenbrand, 2002). geometrical and facies continuity between the two geological units renders this assumption doubtfull. 1.5. Post sliding evolution of Tahiti-Nui We rather think that in order to date the landslide event, one needs to sample (1) the upper flows from Cle´ment et al. (2003) assume that the extension of the volcano affected by the collapse, and (2) the base the caldera and the landslide scar with epiclastic 162 P.-Y. Gillot et al. / Journal of Volcanology and Geothermal Research 136 (2004) 159–163 breccia deposits would have been wholly preserved in proposed by Cle´ment et al. (2003). Thus, the above their original geometry and remained at the surface arguments lead us to conclude that the interpretation since at least 390 ka, despite the intense erosional of Cle´ment et al. (2003) about the extension, the age conditions prevailing on Tahiti. and the cause of the northern landslide in Tahiti-Nui is We rather consider that the present Maroto basin, not supported by geological data and does not take which cuts the main volcano-structural units, includ- into account most of the geological studies previously ing the post-collapse northern shield, results from a performed on Tahiti. recent evolution, due to the erosion. The dissection This is LGMT contribution no. 50. of the volcanic structure has been favored along the backward contact, between the two nested volcanoes (Le Roy, 1994) and reached the coarse-grained References rocks of the magma reservoir. This explains why pebbles of these plutonic rocks are present inside Becker, M., Brousse, R., Guille, G., Bellon, H., 1974. Phases the breccias. d’e´rosion-comblement de la valle´e de la Papenoo et vol- The intrusions of the main rift zone (Gillot et al., canisme subre´cent a` Tahiti, en relation avec l’e´volution desˆ ıles de la Socie´te´ (Pacifique Sud). Mar. Geol. 16, 1993) controlled the E–W development of the pres- 71–77. ent-day drainage network within the Maroto depres- Boutault, G., 1985. Tahiti (Polyne´sie Francßaise, Archipel de la sion, as well as the E–W elongation of the Socie´te´): volcanologie, pe´trographie et ge´otechnie du secteur Vaitamanu valley (Fig. 1). Note that downward, the coˆtier nord-est. The`se 3j cycle, Universite´ Paris-Sud, Orsay. main Papenoo river brutally diverts, at an almost 262 pp. Brousse, R., Gelugne, P., 1987. Tahiti-Nui, carte ge´ologique (1/25 right angle. This peculiar pattern cannot realistically 000), carte des formations de surface (1/25 000), carte des insta- be the consequence of a northward landslide process bilite´s et risques naturels (1/50 000), feuille -. but, on the contrary, to a modification of the draining Ministe`re de l’e´quipement, de l’ame´nagement, de l’energie et of the Maroto erosional cirque, which was first des mines. drained by the Papeiha river to the East and the Brousse, R., Boutault, G., Eisenstein, A., Gelugne, P., 1985. Tahiti- Nui, carte ge´ologique (1/25 000), carte des formations de sur- Punaruu river to the West (Fig. 1). The effusive and face (1/25 000), carte des instabilite´s et risques naturels (1/50 erosional processes filled up these valleys, with 000), feuille . Ministe`re de l’e´quipement, de l’ame´nage- alternating lavas and scree and mud flows, and form ment, de l’e´nergie et des mines. the present-day Viriviriterai and Tamanu plateaus Brousse, R., Lavest, P., Le´otot, C., 1990. Tahiti-Nui, carte ge´ologi- (Fig. 1). The water was thus deviated and escaped que (1/25 000), carte des formations de surface (1/25 000), carte des instabilite´s et risques naturels (1/50 000), feuille — to the north, opening the Papenoo river (Hilden- Teva I Uta—Taravao. Ministe`re de l’e´quipement, de l’ame´nage- brand, 2002). ment, de l’energie et des mines. Cle´ment, J.P., et al., 2003. Epiclastic deposits and ‘horseshoe- shaped’ calderas in Tahiti (Society Islands) and Ua Huka (Mar- 2. Conclusion quesas Archipelago), French Polynesia. J. Volcanol. Geotherm. Res. 120 (1–2), 87–101. Clouard, V., 2000. Etude ge´odynamique et structurale du volca- The huge landslide, which developed a horseshoe- nismedelaPolyne´sie francßaise de 84 Ma a` l’actuel. The`se shape structure open to the north, does exist, but with doctorat, Universite´ de la Polyne´sie francßaise, Tahiti. 214 pp. a much larger extension than reported by Cle´ment et Clouard, V., Bonneville, A., Gillot, P.Y., 2001. A giant landslide on al. (2003): from Tipaerui river North-west to Onoheha the southern flank of Tahiti Island, French Polynesia. Geophys. Res. Lett, 2253–2256. river North-east (Gillot et al., 1993; Le Roy, 1994; Deneufbourg, G., 1965. Carte ge´ologique de Tahiti, notice explica- Hildenbrand, 2002). The sole of the collapse appears tive BRGM, Orle´ans. 97 pp. much higher than the modern erosional surface of Gillot, P.Y., Talandier, H., Guillou, H., Le Roy, I., 1993. Temporal Maroto cirque bottom; it corresponds to the southern evolution of the volcanism at Tahiti Island: geology and struc- bank of the Vaitamanu river some hundreds of meters tural evolution. International Workshop on Intraplate Volcanism: The Polynesian Plume Province, August 1993. Universite´ de higher. The breccia products have been identified Polyne´sie francßaise, , Tahiti, p. 29. offshore (Hildenbrand, 2002). The landslide occurred Hildenbrand, A., 2002. Etude ge´ologique de l’ıˆle volcanique de between 870 and 850 ka ago, much earlier than Tahiti-Nui (Polyne´sie francßaise): e´volution morpho-structurale, P.-Y. Gillot et al. / Journal of Volcanology and Geothermal Research 136 (2004) 159–163 163

ge´ochimique et hydrologique. The`se doctorat, Universite´ Paris- Le Roy, I., 1994. Evolution des volcans en syste`me de point chaud: Sud, Orsay. 202 pp. Ile de Tahiti, Archipel de la Socie´te´ (Polyne´sie francßaise). The`se Hildenbrand, A., Gillot, P.Y., 2001. Evidence for a differentiated doctorat, Universite´ Paris-Sud, Orsay. ignimbritic activity ending the building-stage of Tahiti (French Vallance, J.W., 2000. Lahars. In: Sigurdsson, H. (Ed.), Encyclopedia Polynesia). EUG XI. Terra Nova, Strasbourg, p. 85. of Volcanoes. Academic Press, San Diego, CA, pp. 601–616.