Explosive eruption, flank collapse and megatsunami at Tenerife ca. 170 ka

Abstract

Giant mass failures of oceanic shield volcanoes that generate tsunamis potentially represent a high-magnitude but low-frequency hazard, and it is actually difficult to infer the mechanisms and dynamics controlling them. Here we document tsunami deposits at high elevation (up to 132 m) on the north-western slopes of Tenerife, Canary Islands, as a new evidence of megatsunami generated by volcano flank failure. Analyses of the tsunami deposits demonstrate that two main tsunamis impacted the coasts of Tenerife 170 kyr ago. The first tsunami was generated during the submarine stage of a retrogressive failure of the northern flank of the island, whereas the second one followed the debris avalanche of the subaerial edifice and incorporated pumices from an on-going ignimbrite-forming eruption. Coupling between a massive retrogressive flank failure and a large explosive eruption represents a new type of volcano-tectonic event on oceanic shield volcanoes and a new hazard scenario.

ARTICLE

Received 25 May 2016 | Accepted 13 Mar 2017 | Published 15 May 2017

Giant mass failures of oceanic shield volcanoes that generate tsunamis potentially represent a high-magnitude but low-frequency hazard, and it is actually difcult to infer the mechanisms and dynamics controlling them. Here we document tsunami deposits at high elevation (up to 132 m) on the north-western slopes of Tenerife, Canary Islands, as a new evidence of megatsunami generated by volcano ank failure. Analyses of the tsunami deposits demonstrate that two main tsunamis impacted the coasts of Tenerife 170 kyr ago. The rst tsunami was generated during the submarine stage of a retrogressive failure of the northern ank of the island, whereas the second one followed the debris avalanche of the subaerial edice and incorporated pumices from an on-going ignimbrite-forming eruption. Coupling between a massive retrogressive ank failure and a large explosive eruption represents a new type of volcano-tectonic event on oceanic shield volcanoes and a new hazard scenario.

DOI: 10.1038/ncomms15246 OPEN

Explosive eruption, ank collapse and megatsunami at Tenerife ca. 170 ka

Raphal Paris1, Juan J. Coello Bravo2, Mara E. Martn Gonzlez3, Karim Kelfoun1 & Franois Nauret1

1 Universit Clermont Auvergne, CNRS, IRD, OPGC, Laboratoire Magmas et Volcans, F-63000 Clermont-Ferrand, France. 2 Instituto Volcanolgico de Canarias (INVOLCAN), Antiguo Hotel Taoro, Parque Taoro s/n, 38400 Puerto de la Cruz, Spain. 3 Museo de Ciencias Naturales de Tenerife, Calle Fuente Morales, s/n, 38003 Santa Cruz de Tenerife, Spain. Correspondence and requests for materials should be addressed to R.P. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 8:15246 | DOI: 10.1038/ncomms15246 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15246

Massive ank failures of oceanic shield volcanoes are often an order of magnitude larger (tens to hundreds of km3) than the largest debris avalanche affecting other

types of volcanic edices such as stratovolcanoes, and represent a potential source of megatsunami1,2. With a volume of 5 km3 the collapse of Ritter Island stratovolcano (Papua New Guinea) in 1888 was the largest historical volcano ank failure and it produced a 1015 m tsunami on the coasts of the Bismarck Sea3. There is an abundant literature on volcanic instability and the possible role of internal versus external factors1,46. Models of evolution of oceanic shields and their rift-zones focus on the relationships between gravitational spreading, the intrusive system and the formation of shallow magma reservoirs711. However, the failure mechanisms and dynamics of ank failures of oceanic shield volcanoes are still poorly documented, thus resulting in great uncertainties on related tsunami hazards2,12,13. Marine conglomerates and megaclasts found at unusually high elevations in Hawaii, Cape Verde, Mauritius and Canary Islands were interpreted as being the result of tsunami waves generated by massive ank failures of oceanic shield volcanoes1420. These conglomerates are preserved at different elevations, often out of the range of marine highstand deposits. Their composition reects a mixing of different sediment provenance and assemblages of bioclasts (planktonic, benthic, littoral and subaerial), which are never found in growth position.

Here we document tsunami deposits (marine gravels with pumices) on the north-western anks of Tenerife, Canary Islands, at altitudes up to 132 m a.s.l. (Fig. 1). The stratigraphy of the tsunami deposits and characterization of pumice clasts found in these deposits allows us to distinguish at least two successive tsunamis, identify their possible source, and demonstrate the association between a massive ank failure (Icod collapse) and an explosive ignimbrite-forming eruption (El Abrigo) both about 170 kyr ago. A unique example of coupled slope instability and explosive activity at an oceanic shield volcano is the relatively small debris avalanche (o200 106 m3) triggered during the

Helecho explosive eruption on the south-eastern ank of Tenerife 733 kyr ago21. Apart for this example, such causal links remain speculative, especially for larger-scale ank failures22.

ResultsCharacteristics of the tsunami deposits. Tsunami deposits are preserved at different locations around Isla Baja and Teno Bajo (Fig. 1). The stratigraphy of the sedimentary sequences slightly differs from one site to another (Fig. 2), due to the nature of the substratum underlying the tsunami deposits, and to the lateral and longitudinal discontinuity of the different tsunami units. At Playa de la Arena, two units of tsunami deposits can be distinguished within a paleo-valley (Fig. 3a). Lower unit A is a massive gravel that is clast-supported, very coarse-grained and poorly sorted (from ne pebbles to 1 m large boulders). The dominant population of clasts comes from basaltic lavas that forms nearby coastal cliffs and platforms, but there are also numerous marine bioclasts (Fig. 3b: bivalve and gastropod shells, foraminiferas, calcareous algae, coral fragments), rare rounded pebbles from the beach, and rare pumices. The thickness of unit A ranges from 40 cm to 1.5 m, but its base is not visible. Upper unit B is also a coarse gravel, but it is slightly ner than unit A, matrix-supported, and considerably enriched in pumices relative to unit A. Unit B is thus a pumiceous gravel, and its composition is much more varied than unit A: local-derived basalts are mixed with phonolites, hydrothermally altered rocks, syenites, obsidian and subangular pumices. The dominant type of pumices is light-green coloured, highly vesiculated, and brous. The matrix is sand to silt-sized, and contains abundant bioclasts. The contact between units A and B displays large scour-and-ll features.

Unit B is particularly thick (up to 2 m) and crudely structured into subunits with local variations of grain size and composition. The subunits are often separated by a thin layer of very ne material (ash size). There is no vertical size grading in the individual units and subunits, except in some inversely-graded clast-supported lenses. The two units of tsunami deposits are overlain by a consolidated lithic-rich pumiceous gravel that is less than 1 m thick. No marine bioclasts could be found in this gravel, which could correspond to the reworked facies of a lithic-rich ignimbrite (Fig. 3a). Another coastal section of tsunami deposits was found near El Puertito, on the northern coast of Isla Baja (Fig. 1), where tsunami unit A overlies a palaeosoil on a lava ow dated 194 ka (ref. 20).

Tsunami deposits are also exposed on the Teno Bajo peninsula at altitudes between 15 and 50 m a.s.l. (Fig. 3c). Some outcrops have been briey described23,24 and a tsunami origin was proposed25,26. Post-depositional erosion reduced the deposits to small patches on top of a basaltic lava ow dated 178 ka (ref. 27). A 40 cm thick sand layer is intercalated between the lava ow and the tsunami deposits. The stratigraphy of the deposits slightly differs from one section to another, but a synthetic log can be traced as follows (Fig. 2). Lower unit A is a coarse gravel ning landward (from very coarse to medium pebbles, in a very coarse sand matrix), with rare pumice clasts, and particularly rich in fragments of bivalve shells. As in Playa de la Arena, Unit A has no internal structure. The majority of the clasts are angular to subangular fragments of basaltic lavas coming from the underlying lava ows. Unit A is clearly residual here and its thickness ranges between 5 and 50 cm (Fig. 3d). There is no dominant orientation of the elongated clasts. The basal contact between unit A and the sand layer is erosional. Compared to unit A, upper unit B is considerably enriched in pumice clasts, thus forming a 0.51.5 m thick pumiceous gravel where different subunits can be distinguished (depending on variations of grain size and proportion of pumices). The indurated matrix is essentially made of ne pumice fragments. Marine bioclasts are also present in various abundances from one subunit to another. Mean grain size ranges from ne to coarse pebbles, but larger clasts up to metric boulders can be found. The lower part of unit B is characterized by a crude stratication forming discontinuous lenses or trains of imbricated and inversely graded clasts. The orientation of the elongated or imbricated clasts is alternatively landward and seaward (Fig. 3d), except in the uppermost subunit where clast imbrication is clearly seaward. It is worth noting that this uppermost subunit is also particularly rich in rounded pumice lapilli and marine bioclasts. Units A and B can be traced respectively up to 320 and 700 m inland (21 and 50 m a.s.l.).

Traces of the tsunami deposits were identied at several locations around a volcanic cone named Taco and dated 706 ka (ref. 27; Fig. 1). On the western ank of the cone, tsunami unit B was identied at altitudes between 115 and 132 m a.s.l. The main section (quarry at 132 m a.s.l.) displays tsunami unit B as a massive pumiceous gravel eroding a succession of pyroclastic deposits, colluvial deposits and palaeosoils (Fig. 2). The tsunami also truncates a layer of cream-coloured pumice lapilli interpreted as a Plinian fall deposit (Fig. 4a). Thickness of the tsunami gravel is irregular, ranging from 50 cm to 3 m. Erosion of the substratum is illustrated by 2040 cm large rip-up clasts of the underlying palaeosoil (Fig. 4b). The base of the tsunami deposit is characterized by a 25 cm thick ne-grained traction carpet. Composition of the gravel is similar to unit B downslope (Playa de la Arena section): fragments of lava ows of different lithology (from basalts to phonolites), pumices and rare obsidians, and marine bioclasts. Different facies of pumices are present: light green pumices (dominant facies), cream pumices, and banded crystal-rich (feldspars) pumices.

2 NATURE COMMUNICATIONS | 8:15246 | DOI: 10.1038/ncomms15246 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15246 ARTICLE

b a

Atlantic

Canary

Islands

18 W

16 14

30 N

29

28

20 W 20 E 40 N

0

Lanzarote

Fuerteventura

Gran canaria

20 N

0 100 200 km

Tenerife

El Abrigo - DHF III vent area

Teide volcano (< 170 ka)

Las Caadas volcano

c

Tigaiga

Orotava

Icod collapse ~170 ka

194 ka

1

706 ka

153 ka

Isla Baja Teno Alto

178 ka

3

Tsunami deposits

1: Teno Bajo (1550 m a.s.l.)

2: Playa de la Arena (57 m a.s.l.)

3: El Puertito (4 m a.s.l.)

4: Los Pajales (40 m a.s.l.)

5: Lomo de la Campanas (50 m a.s.l.)

6: Taco (115132 m a.s.l.)

4

N

5

Teno Bajo

Figure 1 | Location and altitude of tsunami deposits on the north-western coast of Tenerife, Canary Islands. (a) Location map of the Canary Islands in the Atlantic Ocean; (b) Shaded relief view of Tenerife island, showing the Icod debris avalanche on the north anks of the island; (c) 3d view of North-West Tenerife. Tsunami deposits are located on the Isla Baja and Punta de Teno platforms (white dots with numbers). Black squares with numbers correspond to wave gauge used for numerical simulations (Supplementary Table 4). Note limits of the Icod ank failure and Diego HernandezEl Abrigo caldera in the background. The scar of the failure is lled by lava ows of the Teide volcanic complex.

At Lomo de las Campanas (50 m a.s.l.) tsunami unit A is also absent and tsunami unit B directly scours a clast-supported breccia (Fig. 4c,d), which corresponds to the uppermost subunit of the Abrigo ignimbrite (last eruption of the third Diego Hernandez cycle2832). The dominant facies of the Abrigo eruption on Isla Baja is a massive lithic-rich ignimbrite, as observed along the coast, but the Abrigo breccia outcrops locally28. Tsunami unit B and the Abrigo breccia share the same heterogeneous composition (basaltic and phonolitic lavas, hydrothermally altered lavas, syenites, obsidian, crystal-rich juvenile clasts), but the tsunami gravel is matrix supported and cemented by carbonates. The contact between tsunami unit B and the Abrigo breccia is clearly erosional (Fig. 2), but its interpenetrative geometry (amalgamated contact) suggests that the two events are closely spaced in time, if not simultaneous (Fig. 4d). There is no weathering horizon between the Abrigo breccia and tsunami unit B.

The abundance of bioclasts differs from one site and subunit to another (upper tsunami unit A being richer than lower unit B), and tends to decrease landward. Bioclasts are never in growth position. The terrestrial fauna is represented by rare gastropod shells and two bones of a giant endemic lizard (Gallotia goliath). More than 1,000 marine bioclasts were analysed and 123 taxons were determined: 85 gastropods, 31 bivalves, 6 corals and

1 scaphopod (Supplementary Table 1). Bivalve shells such asG. glycymeris and Anadora gibbosa, and scaphopods Laevidentalium caudani are particularly abundant. Biodiversity of the marine fauna is particularly rich and represents a mixing of faunas from different environments (depth, substratum), species of the infra-circalittoral zones being dominant. All taxons can be found nowadays in the Canary Islands. Fragmentation of shells is moderate for the gastropods (25%) and high for the bivalves (46%). Bioturbation affects 28% of the marine bioclasts (incrustation, bioperforation).

Age of the tsunami. Tsunami deposits are younger than the 178 ka (Teno Bajo, Playa Arena) and 194 ka lavaows (El Puertito) on which they rest (Fig. 1). The 40 cm thick sand layer intercalated between the tsunami deposits and the 178 ka lava ow at Teno Bajo suggests that the tsunami did not occur immediately after the emplacement of the lava ow. We could not nd evidence of tsunami deposits on the 153 ka lava ow of El Palmar volcano27, which is very close to Playa Arena (Fig. 1). A chronological link between the tsunami, the last major explosive eruption in Tenerife (El Abrigo2832) and a massive collapse of the north ank of the Las Canadas central edice (Icod collapse3335) can be established. The age of the uppermost

NATURE COMMUNICATIONS | 8:15246 | DOI: 10.1038/ncomms15246 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15246

Playa de la Arena 5 m a.s.l. UTM

316754/3139895

Lomo Campanas 50 m a.s.l. UTM

318563/3140871

Taco 132 m a.s.l. UTM

319923/3140360

Scale (m)

Notes

Scale (m)

Notes

Scale (m)

Notes

Pumice gravel

Scale (m)

Notes

Mud

Clay

Silt

Mud

Clay

Silt

vf

Sand

f

mc

vc

Gravel

Gran

Pebb

Cobb

Boul

Mud

Clay

Silt

Sand

vf f c

m vc

Gravel

Gran

Pebb

Cobb

Boul

Mud

Clay

Silt

vf

Sand

f

mc

vc

Gravel

Gran

Pebb

vf

Sand

f

mc

vc

Gravel

Gran

Pebb

Cobb

Boul

Cobb

Boul

Pumice

gravel

Breccia

(Abrigo reworked)

3

2

1

Teno Bajo 18 m a.s.l. UTM

312454/3138658

3

2

1

Pumice gravel

5

4

3

2

1

5

4

3

2

1

Tsunami

Tsunami

unit B

Tsunami

unit A

Tsunami

unit B

Tsunami

Tsunami

DHF III lapilli

(pre-Abrigo)

Aeolian

Lava flow

178ka

Palaeosoil

El Abrigo

breccia

Lava flow

706 ka

Lava flow

Figure 2 | Stratigraphic logs of the main sections. Tsunami unit A is a very coarse-grained gravel that is clast-supported, and poorly sorted (from ne pebbles to boulders). Composition of tsunami unit A is dominated by locally-derived basaltic lavas, but there are also numerous marine bioclasts (bivalve and gastropod shells, foraminiferas, calcareous algae, coral fragments) and rare pumices. Tsunami unit B is a coarse gravel, matrix-supported, and considerably enriched in pumices relative to unit A. Unit B and the underlying El Abrigo breccia have a similar composition: local-derived basalts are mixed with phonolites, hydrothermally altered rocks, syenites, obsidian, pumices, and few marine bioclasts. The contact between tsunami units A,B displays large scour-and-ll features. Unit B is crudely structured into subunits with local variations of grain size and composition. Note that tsunami unit A is absent at elevations higher than 21 m a.s.l. See Fig. 1 for location of the sections.

a

Pumice gravel

Tsunami unit A

b

0 2 mm

c

Tsunami unit B

Lava 178 ka

d

Tsunami unit B

Tsunami unit A

Figure 3 | Tsunami deposits at Playa Arena and Teno Bajo. (a) Succession of tsunami units A,B at Playa de la Arena (Isla Baja); (b) Thin section of tsunami deposits at Playa Arena (tsunami unit A). Note the numerous marine bioclasts (fragments of bivalve shells, bryozoans, coralline algae, and foraminifers); (c) General view of the tsunami deposits at Teno Bajo (altitude 18 m a.s.l.) with a lava ow dated 178 ka outcropping on the lower right corner; (d) Tsunami unit A eroded by tsunami unit B with oating boulders from the underlying 178 ka lava ow at Teno Bajo. Note the difference between tsunami units A,B in terms of texture, mostly due to the abundance of pumices in unit B. Scale bars in centimeters.

4 NATURE COMMUNICATIONS | 8:15246 | DOI: 10.1038/ncomms15246 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15246 ARTICLE

a

Tsunami unit B

DH III pumice fall (pre-Abrigo)

b

Rip-up clasts

c

El Abrigo breccia

Tsunami

d

El Abrigo breccia

El Abrigo breccia

Figure 4 | Tsunami deposits at Taco and Lomo de las Campanas. (a) Tsunami deposit (pumiceous marine gravel, corresponding to tsunami unit B) at 132 m a.s.l. on the anks of Taco volcanic cone, eroding pumice fall deposits of the Diego Hernandez III Formation (pre-Abrigo eruption); (b) Rip-up clasts of soil as evidence of substrate erosion at the base of tsunami unit B (Taco outcrop); (c) Contact between the Abrigo breccia and tsunami deposits (tsunami unit B) at Lomos de las Campanas (50 m a.s.l.). Both units have a similar composition but tsunami unit B is matrix-supported and cemented by carbonates; (d) Detailed view of the erosional contact between the Abrigo breccia and tsunami unit B. The contact is characterized by downward injections of the tsunami in the breccia, suggesting basal amalgamation. Scale bars in centimeters.

breccia of the Abrigo ignimbrite, which is stratigraphically concomitant with the tsunami, is still controversial. 40Ar/39Ar ages of feldspars range between 1966 ka (ref. 29) and 1691 ka (ref. 36), with discrepancies related to the presence of partly degassed xenocrysts29. Nepheline syenites of the breccia were dated at 17911 ka (ref. 37) (K-Ar), 1838 ka (ref. 37) (40Ar/39Ar) and 1753 ka (K-Ar)38. The Icod submarine debris avalanche has been dated B170 ka from shallow seismic33. The age of the Icod turbidite in the Agadir Basin was estimated at 155175 ka (ref. 39) (coccolithophore biostratigraphy) and 16515 ka (ref. 40) (Oxygen Isotope Stages). This is concordant with the 1615 ka (ref. 38) and 1585 ka (ref. 41) ages of the oldest post-collapse lavas lling the Icod embayment onshore (collected in two different water galleries below the Teide volcanic complex). Thus, published ages point to a major suite of large-magnitude events (explosive eruption, ank collapse and tsunami) affecting Tenerife ca. 170175 kyr ago. Note that the present-day altitudes of the different tsunami outcrops are out of the range of documented MIS 5.5 marine highstands in the Canary Islands (o12 m a.s.l.)42, except for the Playa de la Arena and El Puertito outcrops.

Origin of the pumices. The pumice clasts found in the tsunami deposits resemble the great variety of pumices of the third Diego Hernandez Formation (DHF III), in terms of colour (light green, grey, cream pumices), crystallinity (from near-aphyric pumice to crystal mush), and texture (brous, coarsely to nely banded). The phenocryst assemblage is similar to the DHF phonolites (alkali feldspars, clinopyroxene, biotite, magnetite, titanite). Major and trace elements analyses (and especially the Si/Al and Nb/Zr ratios) are consistent with a DHF III origin of the pumices incorporated in the tsunami deposits (Fig. 5). Despite small differences between the DHF phonolitic units in terms of mineralogy and geochemistry, Si/Al and Nb/Zr values allow us to distinguish two lineages43,44. DHF III products have signicantly

higher Si/Al and higher Nb/Zr ratios than DHF I and DHF II (Fig. 3). Complex history of titanite fractionation, melting of pre-existing syenite plutons, and mingling with mac magma explains these variations in trace elements within the different DHF units43,44.

Light green and cream pumice collected in the tsunami deposits both have a typical DHF III signature (Fig. 5). The green pumice is tentatively interpreted as the northern counterpart of plinian fall deposits identied at the base of the Abrigo ignimbrite in the south of the island28. Both deposits share the same composition, colour and texture (high-Nb/Zr, brous, near-aphyric, light green pumice). The occurrence of a plinian fall phase at the onset of the Abrigo eruption remains controversial because fallout deposits are absent (or not preserved) below the ignimbrite for most of the outcrops28,30,31. The cream pumice in the tsunami deposits comes from a Plinian fall deposit that is clearly eroded by the tsunami on the western ank of Taco (Fig. 3a: pumice lapilli). The upper part of the cream pumice fall deposit displays traces of pedogenesis suggesting that it is not synchronous with the tsunami and Abrigo eruption (it was thus possibly deposited by a DHF III pre-Abrigo eruption such as Benijos29). The pumice gravel overlying tsunami unit B at Playa de la Arena, Lomo Campanas and Taco (Fig. 2) represents a pumice-rich reworked facies of the Abrigo breccia.

DiscussionHigh-Nb/Zr phonolitic magmas of Tenerife are often correlated with large caldera-forming eruptions leading to the destruction of the shallow magma reservoirs, such as the Granadilla36,45 and El Abrigo eruptions43,44. The Abrigo eruption was triggered by input of fresh mac magma into a crystal-rich, water-undersaturated phonoliticsyenitic reservoir46. The roof of this reservoir was at 13050 MPa (ca. 45 km below the surface or 12 km below the ocean surface). The Abrigo eruption has an estimated volume of 20 km3 (with only 1.8 km3 now exposed

NATURE COMMUNICATIONS | 8:15246 | DOI: 10.1038/ncomms15246 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15246

onshore29), a maximum thickness of 20 m in the Bandas del Sur area (southern anks of the Las Canadas volcano), 35 m in the Orotava valley and Tigaiga massif (North), and 23 m in Isla Baja (Northwest). The uppermost vent-derived lithic breccia of the Abrigo ignimbrite, which is associated with tsunami unit B in Isla Baja, is interpreted to record the onset of caldera collapse31. Up to 40% of the lithic clasts come from the pre-existing hydrothermal complex. The eruption led to a near-complete evisceration of the shallow magma system46. Note that the interpretation of the Las

Canadas caldera walls (ank failure versus caldera subsidence) and occurrence of a caldera collapse s.s. during the El Abrigo eruption is beyond the scope of this paper.

However, we demonstrate that there is converging evidence that the Abrigo eruption and the Icod ank collapse are linked. In the proximal eld the Icod submarine debris ow deposits are up to 20 km wide and 105 km long, with an estimated volume of 11040 km3 (refs 34,35). Different depositional lobes can be distinguished, suggesting a multistage collapse34,47. The Icod debris ow is correlated to a large turbidite (21020 km3) in the Agadir and Madeira abyssal plains at distances up to 1,000 km away from Tenerife48,49. The Icod turbidite consists of a series of seven vertically stacked volcaniclastic subunits separated by erosive discontinuities and/or short mud intervals49. The lower three subunits have larger volumes (8090 km3), palagonitized glass and higher biogenic carbonate content (2050% in the lowermost subunit), compared to the upper four subunits (B20 km3 each). There is a vertical trend in the composition of volcanic glass, from mac-intermediate lavas in the lower subunits to phonolitic-trachytic glass in the upper subunits. Glasses of El AbrigoDH III composition appears in the last subunit only. The Icod event was thus a multistage retrogressive failure, affecting successively the submarine anks of the island and the basaltic shield, and then the phonolitic-trachytic series of the Las Canadas edice49. The onset of the rst Icod failures pre-dates the Abrigo eruption, which is only recorded in the last turbidite subunit.

On the northern submarine ank of Tenerife, the Icod ank collapse and Abrigo eruption are thus recorded by three proximal lobes of debris ow, seven distal turbidite subunits, and two tsunami units onshore. The spatial distribution of facies and characteristics of newly discovered tsunami deposits in Tenerife allows us to propose a complete scenario of ank collapse, major explosive eruption and subsequent tsunamis 170 kyr ago. The event started with successive submarine-to-subaerial ank failureslarge enough to generate tsunami wavesinundate Isla Baja and Teno Bajo, and leave a massive gravel rich in marine bioclasts at altitudes up to B20 m a.s.l. (tsunami units A). This rst stage detached more than 100 km3 from the north ank of Tenerife, including the submarine anks, and might have triggered a signicant decompression of the magma reservoir. The northwestern part of Tenerife was mantled by pumice fall deposits and lithic-rich ignimbrite before the second tsunami occurred and incorporated large amount of pumices (tsunami units B). This second pumice-rich tsunami reached higher altitudes (deposits at 132 m a.s.l. in Taco and 50 m a.s.l. in Teno Bajo) than the rst one, and it was generated at an advanced stage

a

2,200

200 100 150 200 250 300 350 400 450

Tsunami (this work)

DHF III Isla Baja (this work)

DHF III North (this work)

DHF III El Abrigo

DHF III pre-Abrigo

DHF II Caleta

DHF II Arafo

DHF I Poris

DHF I Fasnia

DHF I Aldea

1,700

Zr

1,200

700

Nb

b

3.4

100 150 200 250 300 350 400 450

3.3

3.2

3.1

3.0

Si/Al

2.9

2.8

2.7

2.6

2.5

Nb

Figure 5 | Pumice geochemistry. Nb versus Zr (a) and Nb versus Si/Al (b) of pumices collected in the Tenerife tsunami deposits, compared to pumices of the Diego Hernandez Formation43,44. Error bars (2s for major elements and 3% for trace elements) are within the size of the symbols.

100 80 60 40 200 m 20 40 60 80 100

a b

3,220,000

3,200,000

3,180,000

3,160,000

3,140,000

3,120,000

3,220,000

3,200,000

3,180,000

3,160,000

3,140,000

3,120,000

100 80 60 40 200 m 20 40 60 80 100

Submarine failure - 41 km3 Subaerial failure - 12 km3

380,000

360,000

340,000

320,000

300,000

280,000

380,000

360,000

340,000

320,000

300,000

280,000

Figure 6 | Numerical simulations of failures on the northern anks of Tenerife and related tsunami. Snapshots after 500 s of simulation. Submarine failure (a) and subaerial failure (b), respectively, correspond to scenarios 1b and 3b in Supplementary Table 4.

6 NATURE COMMUNICATIONS | 8:15246 | DOI: 10.1038/ncomms15246 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15246 ARTICLE

of the eruption, that is, at a late and exclusively subaerial stage of the Icod failures49. The inferred source is a debris avalanche of the Las Canadas edice at the end of the eruption, which generated a larger tsunami on the coasts of Tenerife (tsunami units B), compared to the previous submarine failures (tsunami units A).

Thus, the volume of the successive turbidites and ank failures is not the only parameter controlling the amplitude of tsunamis impacting the coasts of Tenerife. Indeed, the characteristics of tsunamis generated by ank failure do not depend only on the volume of the sliding mass, but also on its origin (subaerial, submerged or both), dynamics (for example, initial acceleration, maximum velocity, retrogressive behaviour, deformation, single block or granular), and water depth (the height of the leading wave being limited by water depth in open ocean)50,51. Numerical simulations of failures on the northern ank of Tenerife (Supplementary Table 4) conrm that a 1215 km3 en masse failure of the subaerial ank of the Las Canadas edice generate tsunami waves high enough to submerge the cone of Taco (Fig. 6), where tsunami deposits (unit B) were preserved at an altitude of 132 m. Whatever the scenario and rheology of the failure, the Teno Bajo peninsula is ooded until altitudes higher than 50 m a.s.l. (that is, higher than the tsunami deposits preserved).

Marine conglomerates at high altitudes (up to 132 m a.s.l.) on the northwestern anks of Tenerife are thus an evidence of tsunami generated by the Icod ank failures ca. 170 ka. The stratigraphy and composition of the deposits suggest at least two successive tsunamis, the second one reworking pumices from the DHF III pre-Abrigo and Abrigo explosive eruptions. The proposed scenario agrees with the turbiditic record of the events and demonstrates that the retrogressive ank failure and explosive eruption are linked. The rst submarine-to-subaerial failures were tsunamigenic (tsunami units A) and might have participated to the triggering of the eruption. The last subaerial failure is associated to the largest tsunami (tsunami units B) and nal destruction of the Las Canadas edice. This is the rst evidence of coupled large explosive eruption (ignimbrite-forming) and massive ank failure at an oceanic shield volcano. Further investigations might reveal similar associations in Tenerife (link between the Roques de Garcia debris avalanche35 and previous explosive phases of Las Canadas volcano?) or the Cape Verde Islands (Santo Antao52, Fogo53). The scenario inferred from this study has major implication in terms of hazard assessment since the tsunamis extends the impact of the collapse and eruption to a regional scale. Considering the high density of population and economic infrastructures along the coasts, such a scenario would have a devastating impact. The monitoring and warning systems are unsuited to dealing with such events, but this study represents an important step towards an integrated volcano-tsunami risk evaluation.

Methods

Taphonomic and geochemical analyses. For taphonomic analysis of the bioclasts found in the tsunami deposits (Supplementary Table 1) we followed the methodology used by Yesares-Garcia and Aguirre (2004)54. Major element compositions (Supplementary Tables 2 and 3) were analysed by ICP-AES (Jobin-Yvon ULTIMA C) at LMV (Laboratoire Magmas and Volcans, Clermont-Ferrand France) after XRF dissolution. For trace elements analysis (Supplementary Tables 2 and 3), samples were dissolved in a HNO3HF mixture, heated for 24 h and then evaporated. After dissolution, uoride precipitateswere dissolved with several cycles of additions of 7N HNO3 and 6N HCl and evaporations. Whole-rock trace elements were obtained by solution Inductively-Coupled Plasma Mass-Spectrometry (ICP-MS, Agilent7500, Agilent Technologies) at LMV. A standard-sample bracketing was used: the BIR-1standard was measured every two samples and the sample measurements where normalized by linear interpolation to the Geo-ReM (http://georem.mpch-mainz.gwdg.de/

Web End =http://georem.mpch-mainz.gwdg.de/) preferred values of the BIR-1.

Numerical modelling. Numerical simulations of debris avalanches and tsunamis were realized using Volcow numerical package, which is able to dynamically model any kind of mass ow and related tsunami generation, propagation and inundation inland12,18,55. Solving for the non-linear shallow water equations on a regular Cartesian grid is employed using an Eulerian explicit upwind numerical scheme. Open boundaries are prescribed at the limits of the computational domain. Second order processes such as wave breaking near the shore and physical dispersion are neglected. Parameters of the different scenarios of ank failure simulated are summarized in Supplementary Table 4.

Data availability. All data generated or analysed during this study are included in this published article (and its Supplementary Information les) or available from the corresponding author on reasonable request.

References

1. Keating, B. H. & McGuire, W. J. Island edice failures and associated tsunami hazards. Pure Appl. Geophys. 157, 899955 (2000).

2. Paris, R. Source mechanisms of volcanic tsunamis. Philos. Trans. R Soc. A 373, 20140380 (2015).

3. Ward, S. N. & Day, S. Ritter Island volcanolateral collapse and the tsunami of 1888. Geophys. J. Int. 154, 891902 (2003).

4. Siebert, L. Large volcanic debris avalanches: characteristics of source areas, deposits, and associated eruptions. J. Volcanol. Geoth. Res. 22, 163197 (1984).

5. McMurtry, G. M., Watts, P., Fryer, G., Smith, J. R. & Imamura, F. Giant landslides, mega-tsunamis, and paleo-sea level in the Hawaiian Islands. Mar. Geol. 203, 219233 (2004).

6. Boudon, G., Le Friant, A., Komorowski, J. C., Deplus, C. & Semet, M. Volcano ank instability in the Lesser Antilles arc: Diversity of scale, processes, and temporal recurrence. J. Geophys. Res-Sol. EA 112, B8205 (2007).

7. Walker, G. P. L. Coherent intrusion complexes in large basaltic volcanoes: a new structural model. J. Volcanol. Geoth. Res. 50, 4154 (1992).

8. Walter, T. R. et al. Rift zone reorganization through ank instability in ocean island volcanoes: an example from Tenerife, Canary Islands. Bull. Volcanol. 67, 281291 (2005).

9. Manconi, A., Longpr, M. A., Walter, T. R., Troll, V. R. & Hansteen, T. H. The effects of ank collapses on volcano plumbing systems. Geology 37, 10991102 (2009).

10. Delcamp, A. et al. Vertical axis rotation of the upper portions of the north-east rift of Tenerife Island inferred from paleomagnetic data. Tectonophysics 492, 4059 (2010).

11. Chaput, M., Famin, V. & Michon, L. Deformation of basaltic shield volcanoes under cointrusive stress permutations. J. Geophys. Res-Sol. EA 119, 274301 (2014).

12. Giachetti, T., Paris, R., Kelfoun, K. & Prez Torrado, F. J. Numerical modelling of the tsunami triggered by the Gmar debris avalanche, Tenerife (Canary Islands): comparison with eld-based data. Mar. Geol. 284, 189202 (2011).

13. Hunt, J. E., Wynn, R. B., Talling, J. & Masson, D. G. Multistage collapse of eight western Canary Island landslides in the last 1.5 Ma: sedimentological and geochemical evidence from subunits in submarine ow deposits. Geochem. Geophys. Geosyst. 14, 21592181 (2013).

14. Moore, J. G. & Moore, G. W. Deposit from a giant wave on the island of Lanai, Hawaii. Science 226, 13121315 (1984).

15. Moore, J. G., Bryan, W. B. & Ludwig, K. R. Chaotic deposition by a giant wave, Molokai, Hawaii. Geol. Soc. Am. Bull 106, 962967 (1994).

16. McMurtry, G. M. et al. Megatsunami deposits on Kohala volcano, Hawaii, from ank, collapse of Mauna Loa. Geology 32, 741744 (2004).

17. Prez Torrado, F. J. et al. The Agaete tsunami deposits (Gran Canaria): evidence of tsunamis related to ank collapses in the Canary Islands. Mar. Geol. 227, 137149 (2006).

18. Paris, R., Giachetti, T., Chevalier, J., Guillou, H. & Frank, N. Tsunami deposits in Santiago Island (Cape Verde archipelago) as possible evidence of a massive ank failure of Fogo volcano. Sediment. Geol. 239, 129145 (2011).

19. Paris, R., Kelfoun, K. & Giachetti, T. Marine conglomerate and reef megaclasts at Mauritius Island (Indian Ocean): evidences of a tsunami generated by a ank collapse of Piton de la Fournaise volcano, Runion Island? Sci. Tsunami Hazards 32, 281291 (2013).

20. Ramalho, R. S. et al. Hazard potential of volcanic ank collapses raised by new megatsunami evidence. Sci. Adv. 1, e1500456 (2015).

21. Dvila Harris, P., Branney, M. J. & Storey, M. Large eruption-triggered ocean-island landslide at Tenerife: onshore record and long-term effects on hazardous pyroclastic dispersal. Geology 39, 951954 (2011).

22. McMurtry, G. M. et al. Stratigraphic constraints on the timing and emplacement of the Alika 2 giant Hawaiian submarine landslide. J. Volcanol. Geoth. Res. 94, 3558 (1999).

23. Garca Talavera, F. G., Paredes, R. & Martn, M. Catlogo-Inventario: Yacimientos Paleontolgicos 76 (Provincia de Santa Cruz de Tenerife. Instituto de Estudios Canarios, 1989).

NATURE COMMUNICATIONS | 8:15246 | DOI: 10.1038/ncomms15246 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15246

24. Zazo, C. et al. Registro de los cambios del nivel del mar durante el Cuaternario en las Islas Canarias Occidentales (Tenerife y La Palma). Estud. Geol. 59, 133144 (2003).

25. Criado, C. & Yanes, A. in Geomorfologia Litoral y Quaternary. Homenatge al Professor Vicen Rosell I Verger (eds Santjaume, E. & Mateu, J. F.) 113122 (Universitat de Valncia, 2005).

26. Ferrer, M. et al. in Landslide Science and Practice. Proceedings of the Second World Landslide Forum, Rome, 2011. International Programme on Landslides (IPL), Vol. 5 (eds Margottini, C., Canuti, P. & Sassa, K.) 27-34 (Springer-Verlag, 2013).

27. Carracedo, J. C. et al. The Teide volcano and the rift-zones of Tenerife, Canary Islands: eruptive and structural history. Geol. Soc. Am. Bull 119, 10271051 (2007).

28. Pittari, A. et al. The inuence of palaeotopography on facies architecture and pyroclastic ow processes of a lithic-rich ignimbrite in a high gradient setting: the Abrigo Igrimbrite, Tenerife, Canary Islands. J. Volcanol. Geoth. Res. 152, 273315 (2006).

29. Edgar, C. J. et al. The late Quaternary Diego Hernandez Formation, Tenerife: volcanology of a complex cycle of voluminous explosive phonolitic eruptions.J. Volcanol. Geoth. Res. 160, 5985 (2007).30. Bryan, S. E., Mart, J. & Cas, R. A. F. Stratigraphy of the Bandas del Sur Formation: an extracaldera record of Quaternary phonolitic explosive eruptions from the Las Canadas edice, Tenerife (Canary Islands). Geol. Mag. 135, 605636 (1998a).

31. Bryan, S. E., Cas, R. A. F. & Mart, J. Lithic breccias in intermediate volume phonolitic ignimbrites, Tenerife (Canary Islands): constraints on pyroclastic ow depositional processes. J. Volcanol. Geoth. Res. 81, 269296 (1998b).

32. Ancochea, E. et al. Evolution of the Canadas edice and its implications for the origin of the Las Canadas Caldera (Tenerife, Canary Islands). J. Volcanol. Geoth. Res. 88, 177199 (1999).

33. Watts, A. B. & Masson, D. G. A giant landslide on the north ank of Tenerife, Canary Islands. J. Geophys. Res. 100, 48724,498 (1995).

34. Watts, A. B. & Masson, D. G. New sonar evidence for recent catastrophic collapses of the north ank of Tenerife, Canary Islands. Bull. Volcanol. 63, 819 (2001).

35. Masson, D. G. et al. Slope failures on the anks of the western Canary islands. Earth Sci. Rev. 57, 135 (2002).

36. Brown, R. J., Barry, T. L., Branney, M. J., Pringle, M. S. & Bryan, S. E. The Quaternary pyroclastic successions of southeast Tenerife, Canary Islands: explosive eruptions, relative caldera-subsidence, and sector collapse. Geol. Mag. 140, 265288 (2003).

37. Mart, J., Mitjavila, J. & Arana, V. Stratigraphy, structure and geochronology of the Las Canadas caldera, Tenerife, Canary Islands. Geol. Mag. 131, 715727 (1994).

38. Boulesteix, T., Hildenbrand, A., Gillot, P. Y. & Soler, V. Eruptive response of oceanic islands to giant landslides: new insights from the geomorphologic evolution of the TeidePico Viejo volcanic complex (Tenerife, Canary). Geomorphology 138, 6173 (2012).

39. Wynn, R. B., Weaver, P. E., Stow, D. A. V. & Masson, D. G. Turbidite depositional architecture across three interconnected deep-water basins on the northwest African margin. Sedimentology 49, 669695 (2002).

40. Frenz, M. et al. Provenance and pathways of late Quaternary turbidites in the deep-water Agadir Basin, northwest African margin. Int. J. Earth Sci. 98, 721733 (2009).

41. Carracedo, J. C. et al. in Los Volcanes del Parque Nacional del Teide(ed. Carracedo, J. C.) 338 (Naturaleza y Parques Nacionales, Serie Tcnica, 2006).

42. Muhs, D. R., Meco, J. & Simmons, K. R. Uranium-series ages of corals, sea level history, and palaeozoogeography, Canary Islands, Spain: an exploratory study for two Quaternary interglacial periods. Palaeogeogr. Palaeo. Palaeo. 394, 99118 (2014).

43. Wolff, J. A., Grandy, J. S. & Larson, B. Interaction of mantle-derived magma with island crust? Trace element and oxygen isotope data from the Diego Hernandez Formation, Las Canadas, Tenerife. J. Volcanol. Geoth. Res. 103, 343366 (2000).

44. Olin, H. Magma dynamics of the phonolitic Diego Hernndez Formation, Tenerife, Canary Islands. 416 (PhD thesis, Washington State University, 2007).

45. Bryan, S. E. Petrology and geochemistry of the Quaternary caldera-forming, phonolitic Granadilla eruption, Tenerife (Canary Islands). J. Petrol. 47, 15571589 (2006).

46. Andjar, J., Costa, F., Mart, J., Wolff, J. A. & Carroll, M. R. Experimental constraints on pre-eruptive conditions of phonolitic magma from the caldera-forming El Abrigo eruption, Tenerife (Canary Islands). Chem. Geol. 257, 173191 (2008).

47. Ablay, G. J. & Hrlimann, M. Evolution of the north ank of Tenerife by recurrent giant landslides. J. Volcanol. Geoth. Res. 103, 135159 (2000).

48. Weaver, P. E., Rothwell, R. G., Ebbing, J., Gunn, D. & Hunter, M. Correlation, frequency of emplacement and source directions of megaturbidites on the Madeira Abyssal Plain. Mar. Geol. 109, 120 (2011).

49. Hunt, J. E., Wynn, R. B., Masson, D. G., Talling, J. & Teagle, D. A. H. Sedimentological and geochemical evidence for multi-stage failure of volcanic island landslides: a case-study from Icod landslide on north Tenerife, Canary Islands. Geochem. Geophys. Geosyst. 12, Q12007 (2011).

50. Harbitz, C. B., Lvholt, F., Pedersen, G. & Masson, D. G. Mechanisms of tsunami generation by submarine landslides: a short review. Norw. J. Geol. 86, 255264 (2006).

51. Lvholt, F., Pedersen, G., Harbitz, C. B., Glimsdal, S. & Kim, J. On the characteristics landslide tsunamis. Philos. Trans. R Soc. A 373, 20140376 (2015).

52. Eisele, S. et al. Stratigraphy of the Pleistocene, phonolitic Cao Grande Formation on Santo Antao, Cap Verde. J. Volcanol. Geoth. Res. 301, 204220 (2015).

53. Eisele, S. et al. Pleistocene to Holocene offshore tephrostratigraphy of highly explosive eruptions from the southwestern Cape Verde Archipelago. Mar. Geol. 369, 233250 (2015).

54. Yesares-Garcia, J. & Aguirre, J. Quantitative taphonomic analysis and taphofacies in lower Pliocene temperate carbonate-siliciclastic mixed platform deposits (Almeria-Nijar Basin, SE Spain). Palaeogeogr. Palaeo. Palaeo. 207, 83103 (2004).

55. Kelfoun, K., Giachetti, T. & Labazuy, P. Landslide-generated tsunamis at Reunion Island. J. Geophys. Res. 115, F04012 (2010).

Acknowledgements

We are grateful to Tim Druitt, Mhammed Benbakkar, Krzysztof Suchorski, Christophe Constantin, Claire Fonquernie, and Jean-Luc Devidal (Laboratoire Magmas and Volcans), Mario Cachao (Universidade de Lisboa) and to Juan Carlos Carracedo (Universidad Las Palmas de Gran Canaria). The rst version of this contribution also beneted from the helpful comments of three reviewers. This research is funded by ASTARTE collaborative project FP7-ENV2013 6.4-3 and French National Research Agency (ANR project VITESSS 08-JCJC-0042). This is Laboratory of Excellence ClerVolc contribution no 240.

Author contributions

All authors contributed to sampling, lab analyses and preparation of the manuscript.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Paris, R. et al. Explosive eruption, ank collapse and megatsunami at Tenerife ca. 170 ka. Nat. Commun. 8, 15246 doi: 10.1038/ncomms15246 (2017).

Publishers note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional afliations.

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

r The Author(s) 2017

8 NATURE COMMUNICATIONS | 8:15246 | DOI: 10.1038/ncomms15246 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications