A geological record of multiple Pleistocene tsunami inundations in an oceanic island: The case of Maio, Cape Verde

A geological record of multiple Pleistocene tsunami inundations in an oceanic island: The case of Maio, Cape Verde JOS  EMADEIRA*†, RICARDO S. RAMALHO*†‡§, DIRK L. HOFFMANN¶, JO ~ AO MATA*†and M  ARIO MOREIRA*** *Instituto Dom Luiz, Faculdade de Ci^ encias, Universidade de Lisboa, Edif ıcio C1, Campo Grande, 1749-016 Lisboa, Portugal (E-mail: [email protected]) †Departamento de Geologia, Faculdade de Ci^ encias, Universidade de Lisboa, Edif ıcio C6, Campo Grande, Lisboa 1749-016, Portugal ‡School of Earth Sciences, University of Bristol, Wills Memorial Building, Queen’s Road, Bristol BS8 1RJ, UK §Lamont-Doherty Earth Observatory, Columbia University, Comer Geochemistry Building, PO Box 1000, Palisades, NY 10964-8000, USA ¶Department of Human Evolution, Max Planck Institute for Evolutionary Anthropology, Deutscher Platz 6, Leipzig 04103, Germany **Instituto Superior de Engenharia de Lisboa, R. Conselheiro Em ıdio Navarro 1, Lisboa 1959-007, Portugal Associate Editor – Pedro Costa ABSTRACT In the Central Atlantic archipelagos –the Canaries, Cape Verde, Madeira and the Azores –tsunamihazardisoftenregardedaslow,whencomparedwithother extreme wave events such as hurricanes and storms. The geological record of many of these islands, however, suggests that tsunami hazard may be underestimated, notwithstanding being lower than in areas adjacent to subduction zones, such as the margins of the Pacific and Indian oceans. Moreover, tsunamis in oceanic islands are generally triggered by local large-scale volcanic flank collapses, for which little is known about their frequency, making it difficult to estimate the probability of a new occurrence. Part of the problem lies in the fact that tsunami deposits are usually difficult to date, and few islands in the world exhibit evidence for repeated tsunami inundation on a protracted timescale. This study reports on the presence of abundant tsunami deposits (conglomerates and sandstones) on Maio Island (Cape Verde) and discusses their stratigraphy, sedimentological characteristics, probable age and tsunamigenic source. Observations indicate that four distinct inundation events of variable magnitude took place during the Pleistocene. One of the tsunami deposits yielded a high-confidence U/Th age of 78809 ka, which overlaps within error with the 73 7 ka age proposed for Fogo volcano’s flank collapse, an event known to have had a significant tsunami impact on nearby Santiago Island. This shows that the Fogo tsunami also impacted Maio, resulting in runups in excess of 60 m above coeval sea-level at ca 120 km from the source. Two older deposits, possibly linked to recurrent flank collapses of the Tope de Coroa volcano in Santo Ant~ ao Island, yielded lower-confidence ages of 479 to 390 ka and 360 to 304 ka. A younger deposit (<78 ka) remains undated. In summary, the geological record of Maio exhibits well-preserved evidence of repeated tsunami inundation, reinforcing the notion that tsunami hazard is not so low at volcanic archipelagos featuring prominent and highly-active volcanoes such as in Cape Verde. Keywords Cape Verde, frequency, ocean islands, sediments, tsunami. 1 ©2019 The Authors. Sedimentology ©2019 International Association of Sedimentologists Sedimentology (2019) doi: 10.1111/sed.12612 INTRODUCTION Oceanic islands are exposed to various natural hazards depending on their geodynamic setting and geographical location; these include volcanic eruptions, earthquakes, floods, droughts, storms, landslides and tsunamis. Island population centres are usually located at the coast and often at low elevations, and are thus highly vulnerable to marine inundations resulting from storm waves and surges or from tsunami impact. In the Central Atlantic archipelagos, tsunami hazard is often regarded as low, but the geological record suggests that this hazard has been considerably underestimated. Despite being much lower than on coastlines of oceans surrounded by active subduction zones capable of producing high-magnitude submarine earthquakes, examples from Macaronesia (a biogeographic region including the Central Atlantic archipelagos of the Azores, Madeira, Selvagens, Canaries and Cape Verde; Fig. 1A) show that the risk imposed to populations by this type of natural hazard may, in fact, be much higher than is usually considered. Accounts of historical events and geological evidence of past tsunami inundations affecting the islands, triggered by both local and distal sources, exist in all of Macaronesia (Ward & Day, 2001; Pararas-Carayannis, 2002; Andrade et al., 2006; Gisler et al., 2006; Perez-Torrado et al., 2006; Giachetti et al., 2011; Paris et al., 2011, 2017, 2018; Ferrer et al., 2013; Hunt et al., 2013; Ramalho et al., 2015; Madeira et al., 2016; Omira et al., 2016). For example, chaotic conglomerates featuring marine fossils occur dispersed along the slopes of Aga€ ete valley in Gran Canaria, from 41 to 188 m in elevation, attesting to significant tsunami inundation resulting from the G€ u ımar volcanic flank collapse(s) in the nearby island of Tenerife, which took place sometime before 083 Ma (Perez-Torrado et al., 2006; Giachetti et al., 2011; Paris et al., 2018). Another example in the Canary Islands concerns the Isla Baja and Teno Bajo deposits in north-west Tenerife, found up to 132 m in elevation, which attest to the successive impact of two consecutive tsunamis triggered respectively by the Icod flank collapse and Abrigo ignimbrite eruption ca 170 ka; these deposits indicate a scenario involving a coupled flank collapse and major explosive eruption, resulting in extreme tsunami inundation of adjacent coastlines (Paris et al., 2017, 2018). The collapse of Fogo volcano, in Cape Verde, is yet another example of a tsunamigenic event that resulted in a devastating impact to adjacent coastlines. Effectively, fields of stranded megaclasts (up to 243 m 3 ) and chaotic marine conglomerates occur in the adjacent island of Santiago, up to 220 m of elevation, attesting to the impact of a tsunami triggered by this collapse at ca 73 ka (Paris et al., 2011, 2018; Ramalho et al., 2015; Mart ınez-Moreno et al., 2018). These deposits imply tsunami runups exceeding 270 m along the northern shore of Santiago, once again confirming the hazard potential of volcanic flank collapses as near-field sources of highly devastating tsunamis (Ramalho et al., 2015; Paris et al., 2018). Despite the existence of ample evidence for tsunami impact on several volcanic archipelagos, a key question in the debate about the hazard and risk posed by tsunami inundation along the coastlines of these archipelagos –by either locally-sourced collapse-triggered tsunamis or remotely-generated tsunamis –concerns the frequency of such events. The estimation of its frequency is crucial to assess the susceptibility/ vulnerability of coastal populations to tsunami inundation. A lack of a reasonable knowledge concerning the frequency of such events is derived from several factors, namely the poor preservation of many of the deposits (particularly deposits from smaller-magnitude events), difficulty in dating several of those deposits (a problem exacerbated by the intrinsic extreme lateral variation of facies), variable coastal morphology providing different sediment accommodation space and, perhaps as the result of the previous factors, the rarity of outcrops that clearly attest to multiple tsunami inundation of the same coastline on a protracted timescale. So far, Aga€ ete valley (Gran Canaria) constitutes one of the few locations at oceanic islands worldwide with evidence for multiple extreme tsunami inundations over a period that was long enough to allow the deposition of alluvial conglomerates and the formation of palaeosols in between events (Madeira et al., 2011; Paris et al., 2018). Successful dating of individual deposits at Aga€ ete valley, however, has proved elusive, mostly owing to the lack of dateable material or to the poor preservation of the deposits themselves, a fact that has so far hampered a solid estimation of the frequency of tsunami inundation at a single island or island group (Perez-Torrado et al., 2006; Madeira et al., 2011). The geological record of Maio in the Cape Verdes, however, may provide the perfect place to study the frequency of tsunami inundation at ©2019 The Authors. Sedimentology ©2019 International Association of Sedimentologists, Sedimentology 2J. Madeira et al. a single island. Owing to its arid climate, low coastal morphology and very slow uplift rate, Maio offers the perfect conditions to host and preserve tsunami deposits, allowing also for more thorough comparative studies between tsunamigenic and marine terrace deposits, given that the latter are also well-preserved on the island. Accordingly, this study analyzes the geological record of Maio and demonstrates that the island exhibits well-preserved evidence of multiple tsunami inundation on a protracted timescale, possibly making it a reference locality amongst oceanic islands to investigate the frequency of such extreme events. As the current Fig. 1. Location, morphology and geology of Maio. (A) Location of the Macaronesian archipelagos, bathymetry/ altimetry from ETOPO1 Global Relief Model (National Geophysical Data Center/NESDIS/NOAA/U.S. Department of Commerce, 2011). (B) The location of Maio Island within Cape Verde Archipelago, bathymetry from General Bathymetric Chart of the Oceans (GEBCO, 2014) and altimetry from Shuttle Radar Topography Mission, 1 arc sec scene (SRTM3). (C) Hillshade from a digital elevation model (DEM) of Maio Island (original dataset at 1 : 5000 scale; MAHOT, 2010), with main toponymy. (D) Geological map of Maio (adapted from Serralheiro, 1970, and Stillman et al., 1982) over hillshade from a DEM (MAHOT, 2010). ©2019 The Authors. Sedimentology ©2019 International Association of Sedimentologists, Sedimentology Multiple tsunami inundations in Maio Island 3 study demonstrates, the stratigraphic relations between the abundant outcrops of tsunamigenic sediments (conglomerates and sandstones) in this island indicate the occurrence of four distinct inundation events of variable magnitude, which took place during the Pleistocene. This study documents these deposits including their stratigraphic relations, textural characteristics, basal features and palaeocurrent markers. The age of the deposits is addressed by U/Th dating of corals sampled from these deposits. Possible sources for these tsunamis are also discussed within the context of Cape Verde. MORPHOLOGY AND GEOLOGICAL SETTING OF MAIO The island of Maio is one of the oldest in the Cape Verde Archipelago (Fig. 1A and B), possibly having emerged above sea-level during the early Miocene (Serralheiro, 1970; Mitchell et al., 1983; Ramalho et al., 2010a). It belongs to the Leeward or Southern Group of the Cape Verdes and is located between the islands of Boavista and Santiago (Fig. 1B). Maio is a relatively small (269 km 2 ), elliptically-shaped island, 25 km long in the north to south direction by 15 km east to west. The island, however, has been significantly reduced by marine abrasion and is presently approximately half of its original area. The extension of the island’s north shelf indicates an original north–south dimension of ca 44 km. Presently, the island edifice exhibits a razed morphology characterized by low-lying littoral platforms, surrounding residual reliefs in the centre of the island (Fig. 1C). The littoral platforms exhibit a staircase morphology that corresponds to Holocene littoral salt pans and a set of Pleistocene raised beaches at elevations of 2to6m,8to12m,15to20m,30to40m,55 to 65 m, 65 to 75 m and 80 to 100 m above present sea-level (apsl). The marine terraces are interpreted as open-shelf to wave-dominated beach deposits (Ramalho et al., 2010a), in every way similar to the present-day beaches that surround the island, but which were consolidated and raised by uplift, which differentially affected most Cape Verdean Islands (Ramalho et al., 2010b). The residual reliefs rise to low elevations, for example Monte Penoso with 430 m, the maximum elevation in the island, Monte Batalha with 294 m and Monte Branco with 253 m. Coastlines are usually sandy, comprising extensive beaches laterally confined by low rocky bluffs composed of either basalt (s.l.), limestones, or consolidated Pleistocene beach and dune deposits, alternating with stretches of rocky coastline (particularly on the eastern shore), characterized by low cliffs (<5 m). Beach sand composition is highly bioclastic, composed mostly of shell and rhodolith debris, often with <10% of lithoclasts/mineroclasts in the mixture (Johnson et al., 2013; Ramalho et al., 2013). Geologically (Fig. 1D), the island exhibits a basement formed by an uplifted Jurassic–Cretaceous ocean floor sequence comprising uplifted Normal Mid Ocean Ridge Basalts (N-MORB) pillow lavas of the Batalha Formation, intensely intruded by dykes, overlain by limestones of the Morro Formation and shales of the Carqueijo Formation (Serralheiro, 1970; De Paepe et al., 1974; Bernard-Griffiths et al., 1975; Stillman et al., 1982; Az ema et al., 1990). Conglomerates (Coruja Formation) representing shoaling of the island overlie the sea floor sequence. This basement was intruded and deformed (folded and faulted) during the Palaeogene by the successive intrusion of plutons, forming the Central Igneous Complex (Represas et al., 2012), comprising gabbros (essexites and pyroxenites) and syenites. Probably of the same age are some rare carbonatite dykes. A basaltic (s.l.) volcanic sequence of Miocene age, comprising hyaloclastites and pillow lavas (Casas Velhas Formation) representing the seamount and emergent stages, subaerial conglomerates (Pedro Vaz Formation) and lava flows (Malhada Pedra and Monte Penoso formations), unconformably overlies the Central Igneous Complex and the Mesozoic sea floor sequence. The lowlands around the residual reliefs are extensively covered by calcarenites and conglomerates from the raised Pleistocene beaches, aeolian sandstones (consolidated dune fields), alluvial fans, active sand dunes and salt flats (sabkha). The deposits reported in this study have been formerly classified and mapped as raised Pleistocene beach deposits by previous authors (e.g. Serralheiro, 1970; Stillman et al., 1982; Ramalho et al., 2010a), notwithstanding their different and distinctive textural and stratigraphic characteristics. On closer inspection, however, these characteristics set them apart from a typical ‘normal’ beach deposit. This work demonstrates the tsunamigenic nature of these deposits, presenting also evidence for a multistage origin (based on a textural and stratigraphical analysis), and documenting the relationship between these and the adjacent ‘normal’ raised beach deposits. ©2019 The Authors. Sedimentology ©2019 International Association of Sedimentologists, Sedimentology 4J. Madeira et al. METHODS Field observations The sediments were surveyed through detailed fieldwork and mapped at the 1 : 25 000 scale. The texture, structure, stratigraphy and thickness of the deposits were described and measured, and the nature of their individual components analyzed. Stratigraphic columns were produced for selected outcrops, using standard methods. Whenever palaeocurrent markers were observed, these were measured to infer transport direction. Attention was also dedicated to the basal features displayed by the outcrops, i.e. the geometry and nature of the stratigraphic contact between the deposits and the underlying substrate, as well as distinctive textures at the base (for example, inclusion of ‘rip-up’ clasts of the substratum within the tsunami deposit). The deposits were sampled for palaeontological content, but its study will not be addressed here. Sedimentological criteria used to identify gravelly tsunami deposits A detailed analysis of the sedimentological criteria used for identifying tsunami conglomerates and gravel beds on oceanic islands can be found in the recent review by Paris et al. (2018). In general, tsunami deposits typically mantle the topography (even on steep slopes) and are usually preserved as patches (lenticular geometry) at different elevations (Moore & Moore, 1984; Moore, 2000; McMurtry et al., 2004a,b; PerezTorrado et al., 2006; Paris et al., 2011, 2017, 2018; Ramalho et al., 2015). Additionally, tsunami conglomerates and gravel beds almost invariably exhibit chaotic textures, being either clast or matrix-supported but generally featuring a mixing of sediments from different sources redistributed both inland and offshore (i.e. terrestrial and marine) (Perez-Torrado et al., 2006; Ramalho et al., 2015; Paris et al., 2017, 2018). Grain-size distribution is commonly highly heterogeneous, often including large boulders ‘floating’ in finer conglomeratic or sandy matrixes and frequently displaying a mixture of both well-rounded and angular boulders/pebbles (Moore & Moore, 1984; Moore, 2000; McMurtry et al., 2004a,b; Perez-Torrado et al., 2006; Paris et al., 2011, 2017, 2018). Deposits are often organized in subunits separated by erosional unconformities or by diffuse but abrupt transitions; extreme lateral variation of facies may be present (Perez-Torrado et al., 2006; Ramalho et al., 2015; Paris et al., 2017, 2018). Landward fining and thinning of sediment is a key feature of tsunami deposits but is not always easy to identify given the deposits’ extreme variation in grain sizes. Tsunami deposits usually rest over irregular erosive basal contacts and may exhibit clastic dykes injected downward into the substratum (Perez-Torrado et al., 2006; Paris et al., 2017, 2018). The preservation of friable subaerial substrates beneath tsunami deposits and the presence of rip-up clasts of this substratum, as well as reverse graded traction carpets, are particularly distinctive features linked to tsunami inundation (Paris et al., 2018). In terms of palaeontological content, tsunami deposits also generally exhibit a mixture of taxa with different bathymetrical ecological zonation, different habitats and different types of substrate (e.g. Massari et al., 2009; Coello Bravo et al., 2014; Paris et al., 2017), often resulting in higher biodiversity indexes (Paris et al., 2018). The only bioturbation found in tsunami sediments deposited onshore corresponds to root concretions, formed by subsequent plant colonization at the surface of the deposits. In contrast to the aforementioned characteristics, raised beach deposits typically show great lateral continuity alongshore, resting on wave-cut low-angle shore platforms, often close to or even abutting against the former shore angle that marks the maximum reach of the high/still sea-level stand. Basal shore platforms almost always correspond to hard substrates, given that soft substrates are rarely preserved by everyday processes of wave-cutting (Ramalho et al., 2013; Paris et al., 2018). Sedimentary sequences typically exhibit normal grading and, despite the presence of conglomeratic basal layers interpreted as ‘transgressive lag’, rarely exhibit boulders or pebble-sized sediments except in well-stratified beds of wellrounded boulders/pebbles associated with clear palaeo-channels or seaward-directed terrigenous debris flows. Sedimentary structures are those typical of wave-beaten sandy beaches, exhibiting even and cross-bedding and lamination (depending on which zone of the beach it corresponds to), or swaley and/or hummocky cross-lamination in preserved storm/higher energy sediments deposited further offshore (Ramalho, 2011; Mayoral et al., 2013). Transition to aeolian structures is frequently found in the upper part of raised terraces, and it is often possible to follow the entire typical beach profile along continuous exposures upstream present-day drainage ©2019 The Authors. Sedimentology ©2019 International Association of Sedimentologists, Sedimentology Multiple tsunami inundations in Maio Island 5 channels (Ramalho, 2011). Bioturbation is paramount in preserved raised beach profiles, as described in detail in Mayoral et al. (2013). As shall be demonstrated in the following sections, the studied deposits of Maio Island exhibit characteristics that conform with those referred above as typical of tsunamiites and are at odds with those of beach deposits. Key sedimentological criteria used to identify these deposits as tsunamiites are also here illustrated through field photographs, palaeocurrent information and stratigraphic sections. Uranium/thorium (U/Th) dating Well-preserved corals were chosen for dating purposes and small sections were cut and polished for visual inspection. The chosen pristine coral samples showed no visible indication for recrystallization. Subsamples for U/Th age determinations were milled from pristine, dense sections of the corals using a hand-held dental drill. The masses of individual subsamples were in the range between 2 mg and 5 mg. Chemical preparation followed a standard protocol (Hoffmann, 2008) and U/Th measurements were performed on a multicollector inductively-coupled-plasma massspectrometer (MC–ICP–MS; Neptune; Thermo Fisher Scientific, Waltham, MA, USA) at the CENIEH (Centro Nacional de Investigaci on sobre la Evoluci on Humana, Burgos, Spain) following procedures outlined in Hoffmann et al. (2007). All ages were calculated using the half-lives reported in Cheng et al. (2000) and are here reported within 2rof uncertainty. RESULTS Geographical distribution of the tsunami deposits Discontinuous tsunami deposits were identified along most of Maio’s littoral coastline and up to 7 km inland from the present-day coastline (Fig. 2). Outcrop elevations (obtained from 1 : 25 000 scale topographic maps and confirmed with an altimeter) range from present-day sea-level up to 70 m apsl, as attested by a tsunami deposit exposed in a quarry located south of the main road 44 km to the ENE of Vila de Porto Ingl^ es (site 1 in Fig. 2; at N15.15515°, W23.17500°). The tsunami deposit located furthest inland (7 km) is exposed in the upper reaches of Ribeira do Lugar (site 2 in Fig. 2; at N15.25953°, W23.14935°), at ca 40 m in elevation. Local discontinuity of the outcrops is interpreted to be either the result of erosion or of coverage by younger sediments. This is the case along the south-west coast, the longest stretch without visible outcrops, owing to the existence of extensive salt flats and a continuous and wide sandy beach without rocky exposures. Although the base of the tsunami deposits is often sub-horizontal, in some places they cover littoral and fluvial valley slopes (Fig. 3A to D). Generally, the deposits can only be recognized in cross-section exposures (at coastal cliff faces, stream banks, road cuts and quarries). At surface exposures, a pervasive carbonated pedogenic alteration (calcrete) surface makes it impossible to distinguish tsunami from other carbonate deposits such as raised beach and consolidated aeolian sandstones (Fig. 3E). Stratigraphy and structure of the deposits The tsunami deposits are amongst the most recent geological units in the island and are only covered by Holocene or Upper Pleistocene formations (active or consolidated dune fields, alluvial and salt-flat deposits, and beach sands in the case of the older tsunami deposit; Fig. 4A and B). They overlie all other geological formations (Jurassic–Cretaceous sea floor sequence, Miocene subaerial and submarine lava flows, Pleistocene raised beaches, consolidated dunes, lagoon sequences, palaeosols and alluvial fan conglomerates; Fig. 4C to G). The individual tsunami deposits consist of one to three layers representing successive flow units, and typically start with matrix-supported chaotic conglomerates (Fig. 4G). The thickness of individual units ranges from a few decimetres up to several (4 to 6) metres. Four different deposits were recognized in the island. In one key coastal section north-east of the Barreiro river mouth (site 3 in Fig. 2; at N15.12817°, W23.14037°), three tsunami deposits are superposed in a continuous outcrop (Fig. 5A and B). All three deposits exhibit erosional bases. The lower tsunami deposit (t1 in Fig. 5A) is discontinuous and is represented by two lenticular bodies filling palaeochannels on a basement of subaerial basalts from the Casas Velhas Formation; it is a matrix-supported, fining-upward, conglomerate composed of angular basalt clasts (1 to 30 cm in diameter), chaotically set in a bioclastic sandy-silty matrix, and exhibiting marine fossils. There are remains of a palaeosol developed on ©2019 The Authors. Sedimentology ©2019 International Association of Sedimentologists, Sedimentology 6J. Madeira et al. top of this deposit. The lower and middle (t1 and t2 in Fig. 5A and B) tsunami deposits are separated by a 06to09 m thick red alluvial fan deposit, which overlies a 08to16 m thick beach sandstone layer. The middle tsunami deposit (t2) is composed of two layers (layers 4a and 4b in Fig. 5A). The base is a discontinuous lenticular layer unconformably overlying the alluvial fan and palaeosol, or resting directly on the beach sandstone, where the alluvial deposit was eroded. It is a fossiliferous (rhodoliths and molluscs) matrix-supported to clast-supported conglomerate made of angular basalt clasts chaotically set in a clay to sand matrix (probably the result of mixing between the alluvial fan sediments and marine sands) and contains rip-up clasts of the beach sandstone. It is less consolidated than the upper layer, which stands out as an erosion-resistant bed. The upper layer of the middle tsunami (unit 4b of t2 in Fig. 5) is a fossiliferous upward-fining matrix-supported conglomerate composed of angular basalt fragments (<20 cm) chaotically set in a bioclastic sand matrix. Locally it displays a palaeosol developed on top. This deposit is covered by a 07 m thick red alluvial fan deposit, which in turn is overlain by the upper tsunami deposit (t3 in Fig. 5A and B). This tsunami deposit presents a wavy base that corresponds to palaeochannels carved on the underlying alluvial sediments. It is a single fossiliferous conglomerate layer with angular basalt clasts supported by a bioclastic sand matrix. This deposit is richer in fossils and slightly less consolidated than the lower and middle tsunami deposits. The sequence is covered by modern red alluvial fan conglomerates. A fourth, younger tsunami deposit (t4) occurs at the coast just north of the mouth of Ribeira Funda de Cima (site 4 in Fig. 2; at N15.15102°, W23.10458°). This deposit covers a slope that extends from the base of the sea cliff at 1 to 2 m up to 9 to 10 m apsl and cuts the platform (at an elevation of 9 to 13 m) that supports one of the older tsunami deposits Fig. 2. Simplified map of the tsunami deposits recognized in the Island of Maio (shaded in yellow: darker areas –observed deposits; lighter areas –inferred deposits) and main toponymy, over hillshade from a digital elevation model (MAHOT, 2010), numbers mark the sites referred to in the text and in Table 1. U/Th ages of dated corals are assigned to the respective outcrops. ©2019 The Authors. Sedimentology ©2019 International Association of Sedimentologists, Sedimentology Multiple tsunami inundations in Maio Island 7 (probably t1); it is a matrix-supported chaotic conglomerate with angular clasts of MORB basalt and aeolian sandstone set in a weakly consolidated bioclastic sand matrix (Fig. 3C). This deposit is very fossiliferous and some of the gastropod shells still display colour. The geometric relation with the tsunami deposit that crops out on the 9 to 13 m apsl platform, the weak degree of consolidation and the good preservation of the fossil content suggest a younger age when compared with the three older tsunami deposits. A BC DE Fig. 3. Field photographs of tsunami deposits and their relationship with the topography. Map shows the location of the depicted outcrops. (A) Tsunami deposit overlapping a slope at the right bank of Casas Velhas creek (base marked by dotted line), in the far field the horizontal white deposit is a Pleistocene raised beach (base marked by dashed line). Length of marked tsunami deposit is ca 80 m. (B) Tsunami deposit overlapping a slope at the right bank of a creek on the east coast (base marked by dotted line, 14 m walking stick for scale). (C) Photograph looking upslope of the younger tsunami (t4) deposit (base marked by dotted line) overlapping a coastal slope on the east coast, the flat surface in the background is covered with an older tsunami (probably t1; base marked by dashed line) and the white cliff behind is cut on a Pleistocene raised beach deposit (the visible part of walking stick is 12m). (D) Geological sketch showing the relation between tsunamis t1 and t4. (E) Tsunami conglomerate exposed at the coast south of Calheta, with a 20 cm thick calcrete developed on top, the deposit is covered by a sandy colluvium. ©2019 The Authors. Sedimentology ©2019 International Association of Sedimentologists, Sedimentology 8J. Madeira et al. At another location in the south-east coast (site 5 in Fig. 2; at N15.13639°, W23.12778°), an outcrop at the sea cliff exposes two superposed tsunami deposits (Fig. 5C). The lower tsunami sequence is formed by four layers (units 1a to 1d in Fig. 5C) dipping 16°to the north (S 0 – N80W,16N), i.e. landward, resting on an erosion surface cut on MORB pillows, intruded by dykes, from the Batalha Formation. The landward dip of the lower tsunami sequence indicates that it fossilized a northward-dipping palaeotopography, a feature that is never observed in the case of marine terraces, which invariably rest on gentle seaward-dipping shore platforms. This tsunami deposit starts with a chaotic coarse conglomerate, mostly clast-supported, comprising angular basalt clasts set in a bioclastic sand matrix containing rhodoliths and oysters; it is followed by three fining-upward layers of fossiliferous conglomeratic silty-sands. The upper tsunami deposit stands directly, through an erosive undulated surface, on the lower tsunami deposit. This deposit consists of two layers (units 2a and 2b in Fig. 5C). The basal layer is a fossiliferous fining-upward matrix-supported conglomerate with a bioclastic sandy matrix. Most of the larger clasts are at the base filling the erosive palaeochannels, but some are floating in the sandy matrix, in chaotic positions. The larger clasts consist of igneous rocks, aeolian sandstone, and rip-up clasts from the underlying tsunami conglomerate. The base of upper layer is also erosive, and the conglomerate displays reverse grading with the larger blocks on top. Igneous clasts are dominant, and the texture is mostly matrix-supported. In all other locations only one tsunami deposit is present, making it difficult to determine to which event it corresponds. There are, nonetheless, some indicators such as the degree of lithification, dissolution of mollusc shells, or coverage by consolidated aeolian sands all pointing to an older age of the deposits. The attribution of the outcrops to a determined tsunami event is in most cases dependent on further isotopic age determinations. The base of the deposits is erosional in most cases (Fig. 6A to D), although laterally the contact may be apparently conformable with the underlying units. The erosional features include palaeochannels and trenches truncating the basement formations in both soft and hard rock, or downward injected sediment dykes. At Ponta Preta beach, the process of quarrying the substrate by the tsunami was preserved in the texture of the conglomerate, with rip-up clasts of the underlying beds floating in the conglomeratic matrix in the vicinities of the erosive step (Fig. 6D). There, the tsunami deposit overlies a 25to28 m thick lagoon sequence consisting of terrigenous sandstones and consolidated mudstone layers exhibiting bioturbation by fine roots. The tsunami sediment can be found injected into the layers of the lagoon sequence along mechanical discontinuities (subvertical joints and stratification surfaces). These injections act as wedges that plucked blocks from the underlying layer. Blocks of these characteristic sedimentary rocks are found incorporated within the overlying tsunami conglomerate (Fig. 6D). Soft intraclasts of palaeosol, volcanic tuffs, aeolian sandstone and weathered lava are occasionally observed but are not abundant (Fig. 6E). Texture and composition of the deposits The tsunamigenic sediments exhibit a bimodal granulometry corresponding to very coarse conglomerates and sandstones containing floating pebbles, boulders and blocks (Fig. 7A to G). Some sandstones exhibit well-developed undulating thin lamination and include floating (i.e. matrix-supported) boulders (Fig 7D). The texture of the conglomerates can be either clastsupported or matrix-supported depending on the amount of matrix. The matrix of both the conglomerates and sandstones is a medium to coarse biogenic sand (Fig. 7C) dominantly composed of fragments of calcareous algae (but also containing fragments of echinoderm spicules, mollusc shells and occasional foraminifera tests) with a minor fraction of lithoclasts (basalt and calcarenite) and mineroclasts (mostly fragments of pyroxene crystals). The nature of the boulders and blocks is variable and depends on the geological units that crop out in the surrounding areas. These may include blocks of volcanic rocks (subaerial and submarine basalts and dykes), various types of sedimentary rocks (lagoon mudstones and sandstones, calcarenites from raised beaches, consolidated aeolian sandstones, or Mesozoic limestones) and plutonic rocks (gabbro). The coarse fraction includes both well-rounded beach pebbles and boulders and angular rock fragments (Fig. 7B and D). Some of the rounded boulders were transported inland from the littoral as evidenced by the presence of attached oysters and bioerosion features ©2019 The Authors. Sedimentology ©2019 International Association of Sedimentologists, Sedimentology Multiple tsunami inundations in Maio Island 9 A BC DE FG Fig. 8. Examples of the fossiliferous content of the tsunami deposits. Map shows the location of the depicted outcrops. (A) Large shattered block of reef coral (white dotted contour) at Ribeira do Lugar creek (geological hammer as scale). (B) Small coral head at tsunami outcrop at the east coast (coral is 10 cm in diameter). (C) and (D) Examples of fossil gastropods: fossil (C) is 25 cm in diameter; the limpet in (D) belongs to genus Fissurella and is 35 cmindiameter.(E)Fossilofanechinoderm spicule (3 cm long). (F) and (G) Two different species of bivalves presenting articulated, closed valves. (F) Sample from Ponta Pedernau on the north coast (23 cm in diameter). (G) Sample from the west littoral south of Calheta (7 cm in diameter). ©2019 The Authors. Sedimentology ©2019 International Association of Sedimentologists, Sedimentology 16 J. Madeira et al. deposits indicates that the waves of the tsunami events were refracted around the island and that the inundations penetrated significantly inland across flat land and climbing low-gradient slopes. Sediment sources and sediment transport along the shore When the geology of the coastline varies rapidly, the tsunami deposits may contain boulders of lithologies that do not crop out in that specific locality, implying along-shore, landward or seaward transport. Along the Ribeira Preta valley (site 8 in Fig. 2; at N15.14503°, W23.19338°) the tsunami deposit consists of two layers: a basal conglomerate up to 14 m thick, overlain by a 4 to 6 m thick massive sandstone containing big blocks of columnar-jointed submarine basaltic sheet flows. The dimensions of 41 parallelepipedal blocks were measured, with major axes ranging from 07to46 m. In most cases the longer axis is parallel to the jointing indicating the minimum thickness of the lava flow from which they were eroded. The sandy matrix is bioclastic (carbonated), and thus of marine origin. The basement here corresponds to the Casas Velhas Formation, which consists of hyaloclastites and isolated pillow lavas and does not contain sheet flows. This type of lava presently does not crop out anywhere onshore, so the blocks must have been transported northward along the valley from outcrops presently below sea-level, or from below the deposit itself. This assumption is consistent with the flow directions observed at the coast. An additional indicator of the inundation direction is given by the nature of boulders included in the tsunami conglomerates along the east shoreline. In fact, on three locations the dominant nature of the boulders does not correspond to the local basement lithologies. Conversely, the dominant composition of the clasts represents lithologies that consistently only crop out further south along the coast, implying a northward transport, slightly oblique to the coast. Effectively, inland from the dune of Ponta do Morro da Areia (site 14 in Fig. 2; at N15.15417°, W23.10167°), the tsunami deposit stands on limestones of Morro Formation, whilst the dominant clasts in the tsunami conglomerate are basalts from the Batalha Formation, which only crops out further south. On that same outcrop the tsunami deposit injects the basement limestones along a narrow crack that dips 30°to the NNE, a feature consistent with the deduced inundation direction. Further north of Ponta do Morro da Areia (site 15 in Fig. 2; at N15.15639°, W23.09972°), the relation is the same; in deposits overlying limestones of the Morro Formation, the dominant conglomerate boulders are basaltic. On the north bank of Cimidor lagoon (site 16 in Fig. 2; at N15.16861°, W23.09778°), the tsunami deposit overlies aeolian sandstones that cover a basement of shales belonging to the Carqueijo Formation. At this location, about 50% of the conglomerate boulders are limestone blocks from the Morro Formation, which only crops out southward of this location. These three outcrops therefore exhibit evidence for alongshore sediment transport from south to north, compatible with the orientation of the conglomerate boulders at Ribeira de Santa Clara. This means that along the southern half of the east shore of Maio, the tsunami inundated the island towards a northward direction. On the north coast, the tsunami deposit is composed of marine bioclastic sand (with abundant marine fossils) and a coarse fraction mostly composed of angular basaltic boulders from the Monte Penoso Formation. However, a few blocks of basaltic submarine sheet flows also occur in the conglomerate. This lithology does not presently crop out in the region, so these blocks of submarine lavas must have been picked up by the tsunami from a location below present-day sea-level. The sand and fossils were, thus, transported inland from a sandy shore, and the angular boulders were picked up from the inundated subaerial topographic surface where those basalts crop out. Tsunami sources The most probable source for the tsunami deposit yielding an age of ca 79 ka is the flank collapse of Fogo volcano (Paris et al., 2011; Ramalho et al., 2015). This age of 78809is well within the 65 to 84 ka age interval and compatible, within error, with the mean age of 73368 ka reported by Ramalho et al. (2015) on the basis of cosmogenic exposure dating of the fields of megaclasts of northern Santiago, which are attributed to the impact of a megatsunami resulting from Fogo’s flank collapse. Because the island of Santiago stands between Fogo and Maio, the tsunami waves had to be refracted around the considerable obstacle formed by the island of Santiago to hit Maio and then around Maio to leave deposits at the east coast of the island. Given that sea-level during ©2019 The Authors. Sedimentology ©2019 International Association of Sedimentologists, Sedimentology Multiple tsunami inundations in Maio Island 17 65 to 84 ka (the age interval reported for this tsunami; Ramalho et al., 2015) was 50 to 60 m below present-day level, the elevation (at least 8 m) of this tsunami deposit implies a minimum runup in excess of 60 m above coeval sea-level, at a location that is at a distance of ca 120 km A BC DE FG ©2019 The Authors. Sedimentology ©2019 International Association of Sedimentologists, Sedimentology 18 J. Madeira et al. (as the crow flies) from the source of this tsunami. The same tsunami produced a minimum runup of 270 m at the north coast of Santiago, at a distance of 70 km from Fogo. There are no confirmed sources for the three remaining tsunami deposits (t1, t2 and t4) found on Maio. However, onshore and offshore evidence for large flank collapses abound in the archipelago, namely at Santo Ant~ ao, S~ ao Vicente and S~ ao Nicolau Islands (Le Bas et al., 2007; Masson et al., 2008; Ancochea et al., 2010; Eisele et al., 2015). Accordingly, the two older tsunami deposits could be associated with the two known flank collapses of the Tope de Coroa volcano in south-west Santo Ant~ ao, which may have taken place at around 200 to 400 ka and >500 ka (Masson et al., 2008). Bearing in mind the probable ages reported by Masson et al. (2008), the other major landslides in the Cape Verde Islands (north and northeast Santo Ant~ ao, north and south S~ ao Vicente, and north and south S~ ao Nicolau) are too old to account for the two older tsunami deposits documented on Maio, assuming a maximum age of 479 ka for these deposits. The ages obtained for Maio’s tsunami event t2 is compatible with the younger collapse of Santo Ant~ ao (Tope de Coroa 1), tentatively dated at 200 to 400 ka by Masson et al. (2008). Moreover, the age range displayed by the coral samples collected on this deposit (Fig. 11), roughly coincides with marine isotope stage 9 (MIS 9; 337 to 300 ka; Lisiecki & Raymo, 2005). This suggests that this tsunami occurred during or soon after this interglacial. In a similar fashion, the age of the older tsunami partially overlaps with MIS 11c (three peaks of temperature maxima at 425 ka, 420 to 412 ka and 407 ka; Candy et al., 2014). This interglacial corresponds to a long (20 to 40 ka) warm period during which the sea was close to Holocene levels (Candy et al., 2014; Past Interglacials Working Group of PAGES, 2016). The fact that the climate was warm, and the interglacial period long, may account for the development of coral reefs on the Maio coast. This could explain the large (metresized) blocks of coral included in the Ribeira do Lugar tsunami deposits. Presently no coral reefs exist in Maio and the rocky outcrops in the coast only display small encrusting and occasional branching corals. Another significant aspect observed in one of the Ribeira do Lugar outcrops is that the tsunami deposit seems to cover a rocky abrasion platform. This is suggested by a dyke protruding from that surface that displays bioerosion features, blocks of which were incorporated in the tsunami deposit and transported eastward. The bioerosion borings in the basaltic dyke indicate that the surface on which the tsunami deposit stands was at or slightly below sea-level. Since the outcrop is presently at an elevation of ca 40 m apsl, it indicates an average uplift of the island of 01 mm year 1 during the last 400 ka, which is a rate compatible with the uplift reconstructions for Maio (Ramalho et al., 2010a,b,c). The trigger for this tsunami is still uncertain, but the older flank collapse of Topo de Coroa (Tope de Coroa 2 of Masson et al., 2008) is a good candidate, given the proximity (but no overlap) between the age interval yielded by t2 (479 to 390 ka) and the age inferred (>500 ka; Masson et al., 2008) for this collapse, particularly taking into account the large uncertainty in the age estimates reported for these landslides. The beach–lagoon–tsunami sequence of Ponta Preta beach (sites 6 and 7 in Fig. 2) suggests either a sea-level drop or coastal progradation. The beach sandstone crops out in the presentday sandy beach and sea-cliff from sea-level up to ca 10 m apsl, while the top of the lagoon sequence is at ca 14 m apsl. Considering the age of 3971283 ka yielded by the coral sampled from the beach sandstone, the sea-level drop following the MIS 11c high-stand could account for this regressive sedimentary sequence. This age also means that the overlying tsunami deposits are related either to events t2 or t3. Fig. 9. Photographs of the different aspects of directional structures displayed by the tsunami deposits. Map shows the location of the depicted outcrops. (A) and (B) Imbricated clasts at the base of tsunami conglomerates indicating inflow (coast is to the left) at the right bank of Casas Velhas creek –portion of walking stick visible in (A) is 1 m long; geological hammer for scale in (B) is 33 cm long. (C) and (D) Imbricated flat clasts in tsunami conglomerates indicating uprush (C) and backwash (D) at the left bank of Cumeazo creek –coast to the right in both photographs; yellow ruler in (C) is 20 cm long; geological hammer for scale in (D) is 33 cm long. (E) Oriented flat clasts in tsunami conglomerate indicating currents from the north (region of Laje Branca Islet, geological hammer for scale). (F) Grooves at the base of the tsunami conglomerates of Ponta Preta beach indicating a north–south current (13 cm long pen for scale and to indicate current direction). (G) Lineations in laminated sandstone layer from the tsunami deposits of Casas Velhas creek valley interpreted as flow structures (pen for scale and to indicate current direction). White arrows in photographs (A) to (E) indicate current direction. ©2019 The Authors. Sedimentology ©2019 International Association of Sedimentologists, Sedimentology Multiple tsunami inundations in Maio Island 19 Fig. 10. Stereographic plots of the oriented and imbricated boulders, grooves and lineations described in the text (and in Appendix S1) and respective locations over the hillshade from a digital elevation model of Maio (MAHOT, 2010). (A) Ponta Preta beach data with oriented clasts in blue (blue arrow indicates current sense and associated azimuth) and grooves and flute casts in red (black arrow). (B) and (C) Oriented clasts (B) and lineations (C) measured at Casas Velhas valley at the outcrop closer to the shore. (D) to (F) Stereographic plot of data from two nearby outcrops upstream from (B) and (C) along Casas Velhas valley –(D) and (E) correspond to the lower and upper part of the same outcrop, showing inflow and backwash. (G) Data from oriented flat clasts (red dots and black arrow) and imbricated clasts (blue dots and blue arrow) from Cumeazo valley. (H) Data from the coast near Pil~ ao C~ ao. (I) Data from oriented flat clasts from the coast east of Laje Branca islet with the rose diagram showing a dominant SSW–NNE direction and a fainter NNW–SSE direction. (J) Data from flat clasts at Ponta Pedernau coast with the large dispersion corresponding to inflow from both the NNW and from the WSW (red dots) and some blocks overturned by backwash (blue dots). Dip and dip direction of flat clasts, grooves and lineations were projected as lines (represented as points in the plots), lower hemisphere stereographic projections produced with GEORIENT, version 9.5.1; GEORIENT, 2015. ©2019 The Authors. Sedimentology ©2019 International Association of Sedimentologists, Sedimentology 20 J. Madeira et al. Finally, the youngest of the tsunami deposits (t4) only crops out at one location. Therefore, it seems to be the result of a very localized event and may have been produced by a local submarine landslide (on the south-east submarine flank of Maio?), or by a smaller landslide at Fogo, following the main event at 73 to 79 ka. Final remarks Notwithstanding the still incomplete knowledge of the likely sources and the need for more precise ages for these events, the geological record shows that the island of Maio was impacted by four tsunami inundations in the last (four?) hundred thousand years, three of which produced significant inundation. This constitutes evidence for multiple tsunami inundations within a protracted timescale, with a mean frequency of one event every 100 ka. Combined with examples from other archipelagos (see Paris et al., 2018, for a review), the findings reported here reinforce the notion that tsunami hazard is high at most volcanic archipelagos, including those located in oceans surrounded by dominantly passive margins. Indeed, volcanic islands can grow quickly producing steep morphologies and abrupt flanks, which are gravitationally unstable, being prone to fail. This is especially true for archipelagos that still feature prominent and highly active volcanoes that may one day collapse (again) as is the case of Fogo and Tope de Coroa in Cape Verde, Tenerife, La Palma and El Hierro in the Canary Islands, or Pico in the Azores. Islands with extensive coastal lowlands such as Maio are particularly vulnerable to Table 1. Location of the samples analyzed by U/Th. Site No. Field Reference N Lat. W Long. Elevation (m) 1 CVP116 15.15528°23.17500°70 2 CVP71a 15.25917°23.14983°39 2 CVP100a 15.25944°23.14944°40 2 CVP142a 15.25944°23.14944°40 2 CVP142b 15.25944°23.14944°40 4 CVP128 15.15111°23.10472°4 7 CVP77 15.12608°23.20255°1 17 CVP135 15.30944°23.12888°1 18 CVP95 15.31222°23.12111°4 18 CVP138 15.31222°23.12111°6 19 CVP122 15.12250°23.14722°3 19 CVP123 15.12250°23.14722°4 20 CVP127 15.20500°23.09416°10 21 CVP129 15.21805°23.09444°3 Table 2. Results of the U/Th analyses of corals from Maio: n.d. = no data. Lab ID Sample 238 U (ng g 1 ) 232 Th (ng g 1 ) 230 Th (ng g 1 ) [ 230 Th/ 232 Th] activity ratio ( 232 Th/ 238 U) activity ratio ( 230 Th/ 238 U) activity ratio ( 234 U/ 238 U) activity ratio Age (ka) ( 234 U/ 238 U) initial activity ratio UTO221 CVP129 2478013039489 0573 2353E-02 124E-03 4630441253E-03 1310E-05 5801E-01 4239E-03 1117E+00 2117E-03 7880911457 00026 UTO197 CVP123 2850316292790 0173 4789E-02 275E-03 320532753202E-04 3071E-06 1026E+00 6557E-03 1065E+00 2423E-03 316112311597 00062 UTO218 CVP138a 2747021421162 0179 4629E-02 362E-03 7436710431384E-04 8272E-06 1029E+00 4821E-03 1061E+00 2634E-03 332011611555 00057 UTO192 CVP127 2511421120096 0042 4298E-02 371E-03 83588951401251E-05 1967E-06 1046E+00 6203E-03 1071E+00 3125E-03 338515111848 00077 UTO219 CVP135 2705315900202 0064 4610E-02 268E-03 42579032762445E-05 4323E-06 1041E+00 5150E-03 1067E+00 2470E-03 339312511749 00062 UTO220 CVP95 23959179213241 0987 4130E-02 311E-03 5823441808E-03 1587E-05 1053E+00 5576E-03 1075E+00 2348E-03 344113511976 00071 UTO216 CVP122 3078911821806 0124 5234E-02 222E-03 541035721920E-04 5261E-06 1039E+00 4074E-03 1062E+00 1937E-03 348810911655 00050 UTO196 CVP77 165508807210 0412 2662E-02 131E-03 6893551425E-03 1198E-05 9825E-01 6145E-03 1007E+00 2487E-03 397128310204 00070 UTO195 CVP71a 233499690097 0015 3783E-02 180E-03 73075953401354E-05 9697E-07 9897E-01 5445E-03 1008E+00 2326E-03 422332010261 00068 UTO198 CVP142a 260267080172 0010 4227E-02 119E-03 45911543702161E-05 6896E-07 9921E-01 6255E-03 1009E+00 2403E-03 423736110314 00071 UTO217 CVP128 2674210881589 0096 4658E-02 176E-03 547496941944E-04 5047E-06 1064E+00 4662E-03 1063E+00 2061E-03 424724012085 00121 UTO194 CVP100a 2545515654274 0276 4188E-02 256E-03 182951405494E-04 4994E-06 1005E+00 4995E-03 1017E+00 2480E-03 442835910577 00073 UTO193 CVP116 241910732846 1444 4089E-03 184E-04 232024443E-02 3540E-04 1033E+00 7987E-03 1021E+00 5552E-03 Out of range UTO222 CVP142b 277861005 n.d. 4665E-02 165E-03 1026E+00 3716E-03 1015E+00 1701E-03 Out of range ©2019 The Authors. Sedimentology ©2019 International Association of Sedimentologists, Sedimentology Multiple tsunami inundations in Maio Island 21 tsunami inundation, which may affect areas up to several kilometres inland. In Cape Verde this is the case for Boa Vista, Sal and S~ ao Vicente (in addition to Maio itself), which are islands with growing coastal occupation as a result of a steady increase in tourism. CONCLUSIONS Research on tsunami deposits from the geological record of volcanic oceanic islands, including their probable age and source, is crucial to estimate the frequency and magnitude of these lowprobability, high-impact events, and to assess the threat posed by such extreme geohazards to island populations. In this study, the occurrence of abundant tsunami deposits on the geological record of Maio Island is reported, with key findings that are summarized here: 1Sedimentary deposits assigned to tsunami deposition occur at multiple places along the coast of Maio island, being found up to 70 m in elevation, and up to ca 7 km inland from the present-day coastline. 2The geological record of Maio attests to the presence of possibly four distinct events of tsunami inundation at a protracted timescale, during the Pleistocene. 3Based on U/Th dating of corals sampled in these deposits, one of the events yielded an age of 78 09 ka, which overlaps within error with the ca 73 7 ka megatsunami triggered by the flank collapse of Fogo volcano, an event that left ample evidence for its impact on the nearby Santiago Island (Paris et al., 2011; Ramalho et al., 2015). Two older tsunami events are tentatively dated at 479 to 390 ka and 360 to 304 ka, dates that are compatible with flank collapses at Tope de Coroa in Santo Ant~ ao (Masson et al., 2008), thus suggesting that this volcano is a possible source for these events. A younger (<79 ka), smaller event remains undated. 4The presence of tsunami deposits correlative of the Fogo flank collapse implies a local runup for this event in excess of 60 m above coeval sea-level, at a distance of ca 120 km from the tsunamigenic source. 5The geological record of Maio Island thus exhibits well-preserved evidence of repeated tsunami inundation, making it another reference locality amongst oceanic islands to investigate the frequency of such extreme events. It also highlights that low-lying islands are particularly vulnerable to tsunami inundation, and how tsunami hazard assessments for volcanic archipelagos should consider such low-probability, highimpact events. ACKNOWLEDGEMENTS This work benefitted from discussions with colleagues M ario Cach~ ao, C esar Andrade and Pedro Costa. This is a contribution from projects CVPLUME (PTDC/CTE-GIN/64330/2006), MEGA WAVE (IF/01641/2015) and UID/GEO/50019/ 2013 to Instituto Dom Luiz, funded by FCT – Fundacß ~ ao para a Ci^ encia e Tecnologia (Portugal). The support of colleagues from the Instituto Nacional para a Gest~ ao do Territ orio (INGT) and Instituto Nacional de Meteorologia e Geof ısica (INMG) of Cabo Verde, as well as the support of colleague Jair Rodrigues, is much appreciated. We acknowledge the contributions by the Associate Editor Pedro Costa, and Francisco PerezFig. 11. Plot of the set of U/Th coral dates by age and elevation; sampling sites are identified by colours shown in the inset legend, horizontal bars show the age uncertainty at the 2-sigma level, elevation range of the deposits is depicted by vertical bars and circles are positioned at the sampling elevation. The age of site 4 (marked with an asterisk in the inset legend) represents a coral sampled from tsunami t4, which was probably reworked from tsunami t1 and therefore is not indicative of the true age of deposition of t4. ©2019 The Authors. Sedimentology ©2019 International Association of Sedimentologists, Sedimentology 22 J. Madeira et al. Torrado, Raphael Paris and an anonymous reviewer for their careful revision that helped to substantially improve the original manuscript. REFERENCES Ancochea, E.,Huertas, M.J.,Hern an, F. and Br€ andle, J.L. (2010) Volcanic evolution of S~ ao Vicente, Cape Verde Islands: the Praia Grande landslide. J. Volcanol. Geoth. Res.,198, 143–157. Andrade, C.,Borges, P. and Freitas, M.C. (2006) Historical tsunami in the Azores archipelago (Portugal). J. Volcanol. Geoth. Res.,156, 172–185. Az ema, J.,Fourcade, E. and De Weaver, P. (1990) D ecouverte de Valanginian inferieur  a Calpionelles  a Maio (R epublique du Cap Vert): discussion de l’age des s ediments associ es aux laves de type MORB de ce s ecteur de l’Atlantique Central. CR Acad. Sci. Paris,310,s  erie II, 277–283. Bernard-Griffiths, J.,Cantagrel, J.-M.,Matos-Alves, C.A., Mendes, C.A.,Serralheiro, A. and Macedo, J.R. (1975) Donn ees radiometriques potassium-argon sur quelques formations magmatiques des iles de I’archipel du Cap Vert. CR Acad. Sci. Paris,280, 2429–2432. Candy, I.,Schreve, D.C.,Sherriff, J. and Tye, G.J. (2014) Marine isotope stage 11: Palaeoclimates, palaeoenvironments and its role as an analogue for the current interglacial. Earth-Sci. Rev.,128,18–51. Cheng, H.,Edwards, R.L.,Hoff, J.,Gallup, C.D.,Richards, D.A. and Asmerom, Y. (2000) The half-lives of uranium234 and thorium-230. Chem. Geol.,169,17–33. Coello Bravo, J.J.,Mart ın Gonz alez, M.E. and Hern andez Guti errez, L.E. (2014) Tsunami deposits originated by a giant landslide in Tenerife (Canary Islands). Vieraea,42, 79–102. De Paepe, P.,Klerkx, J.H. and Plinke, P. (1974) Oceanic tholeiites on the Cape Verde Islands: petrochemical & geochemical evidence. Earth Planet. Sci. Lett.,22, 347–354. Eisele, S.,Freundt, A.,Kutterolf, S.,Ramalho, R.S., Kwasnitschka, T.,Wang, K.-L. and Hemming, S.R. (2015) Stratigraphy of the Pleistocene, phonolitic C~ ao Grande Formation on Santo Ant~ ao, Cape Verde. J. Volcanol. Geoth. Res.,301, 204–220. Ferrer, M.,Gonz alez de Vallejo, L.,Seisdedos, J.,Coello, J.J.,Garc ıa, J.C.,Hern andez, L.E.,Casillas, R.,Mart ın, C., Rodr ıguez, J.A.,Madeira, J.,Andrade, C.,Freitas, M.C., Lomoschitz, A.,Yepes, J.,Meco, J. and Betancort, J.F. (2013) G€ u ımar and La Orotava mega-landslides (Tenerife) and tsunamis deposits in Canary Islands. In: Landslide Science and Practice: Volume 5: Complex Environment (Eds C. Margottini, P. Canuti and K. Sassa), pp. 27–33. Springer-Verlag, Berlin, Heidelberg. GEBCO (2014) The GEBCO_2014 Grid, version 20141103. Available at: http://www.gebco.net. GEORIENT (2015) GEOrient, version 9.5.1 of April 9th, 2015. Available at: http://www.holcombcoughlinoliver.com/holc omb/. Giachetti, T.,Paris, R.,Kelfoun, K. and P erez-Torrado, F.J. (2011) Numerical modelling of the tsunami triggered by the G€ u ımar debris avalanche, Tenerife (Canary Islands): comparison with field-based data. Mar. Geol.,284, 189–202. Gisler, G.R.,Weaver, R.P. and Gittings, M.L. (2006) Sage calculations of the tsunami threat from La Palma. Sci. Tsunami Hazards,24, 288–301. Hoffmann, D.L. (2008) 230 Th isotope measurements of femtogram quantities for U-series dating using multi ion counting (MIC) MC-ICPMS. Int. J. Mass Spectrom.,275,75–79. Hoffmann, D.L.,Prytulak, J.,Richards, D.A.,Elliott, T., Coath, C.D.,Smart, P.L. and Scholz, D. (2007) Procedures for accurate U and Th isotope measurements by high precision MC-ICPMS. Int. J. Mass Spectrom.,264,97–109. Hunt, J.E.,Wynn, R.B.,Talling, P.J. and Masson, D.G. (2013) Turbidite record of frequency and source of large volume (>100 km 3 ) Canary Island landslides in the last 1.5 Ma: implications for landslide triggers and geohazards. Geochem. Geophys. Geosyst.,14, 2100–2123. Johnson, M.E.,Gudveig Baarli, B.,Marques da Silva, C., Cach~ ao, M.,Ramalho, R.S.,Ledesma-V azquez, J., Mayoral, E.J. and Santos, A. (2013) Coastal dunes with high content of rhodolith (coralline red algae) bioclasts: Pleistocene formations on Maio and S~ ao Nicolau in the Cape Verde archipelago. Aeolian Res.,8,1–9. Le Bas, T.P.,Masson, D.G.,Holtom, R.T. and Grevemeyer, I. (2007) Slope failures of the flanks of the southern Cape Verde Islands. In: Submarine Mass Movements and Their Consequences, Advances in Natural and Technological Hazards Research (Eds V. Lykousis, D. Sakellariou and J. Locat), vol. 27, pp. 337–345. Springer, Dordrecht, the Netherlands. Lisiecki, L.E. and Raymo, M.E. (2005) A Plio-Pleistocene stack of 57 globally distributed benthic d 18 O records. Paleoceanography,20, PA1003. Madeira, J.,Ferrer Gij on, M.,de, Gonz ales Valejo, L.,Andrade, C.,Freitas, M.C.,Lomoschitz, A. and Hoffmann, D.L. (2011) Agaete revisited: new data on the Gran Canaria tsunamiites. Geophysical Research Abstracts,13.8thGeneralAssemblyof the European Geosciences Union, Vienna. Madeira, J.,Ramalho, R.S.,Hip olito, A.,Mata, J.,Moreira, M.,Andrade, C.,Freitas, M.C.,Ferrer, M.,de, Gonz alez Vallejo, L. and Gaspar, J.L. (2016) Urban risk from tsunami hazard at volcanic oceanic islands: examples from Macaronesia. Proceedings of ICUR 2016 –1st International Conference on Urban Risks, Lisboa, 30 June-2 July, 8 pp. MAHOT (2010) Digital elevation model of Maio Island, at 1:5,000 scale. Unidade de Coordenacß ~ ao do Cadastro Predial (UCCP) do Minist erio do Ambiente Habitacß ~ ao e Ordenamento do Territ orio (MAHOT), Cabo Verde. Mart ınez-Moreno, F.J.,Monteiro-Santos, F.A.,Madeira, J., Pous, J.,Bernardo, I.,Soares, A.,Esteves, M.,Ad~ ao, F., Ribeiro, J.,Brum da Silveira, A. and Mata, J. (2018) Investigating the collapse structures of Fogo Island (Cape Verde) by a magnetotelluric survey. J. Volcanol. Geoth. Res.,357, 152–162. Massari, F.,D’Alessandro, A. and Davaud, E. (2009) A coquinoid tsunamite from the Pliocene of Salento (SE Italy). Sed. Geol.,221,7–18. Masson, D.G.,Le Bas, T.P.,Grevemeyer, I. and Weinrebe, W. (2008) Flank collapse and large-scale landsliding in the Cape Verde Islands, off West Africa. Geochem. Geophys. Geosyst.,9, Q07015. Mayoral, E.,Ledesma-Vazquez, J.,Baarli, B.G.,Santos, A., Ramalho, R.S.,Cach~ ao, M.,da Silva, C.M. and Johnson, M.E. (2013) Ichnology in oceanic islands; case studies from the Cape Verde Archipelago. Palaeogeogr. Palaeoclimatol. Palaeoecol.,381,47–66. McMurtry, G.M.,Watts, P.,Fryer, G.,Smith, J.R. and Imamura, F. (2004a) Giant landslides, mega-tsunamis, and paleo-sea level in the Hawaiian Islands. Mar. Geol.,203, 219–233. ©2019 The Authors. Sedimentology ©2019 International Association of Sedimentologists, Sedimentology Multiple tsunami inundations in Maio Island 23 McMurtry, G.M.,Fryer, G.J.,Tappin, D.R.,Wilkinson, I.P., Williams, M.,Fietzke, J.,Garbe-Schoenberg, D. and Watts, P. (2004b) Megatsunami deposits on Kohala volcano, Hawaii, from flank, collapse of Mauna Loa. Geology,32, 741–744. Mitchell, J.,Bas, M.L.,Zielonka, J. and Furnes, H. (1983) On dating the magmatism of Maio, Cape Verde Islands. Earth Planet. Sci. Lett.,64,61–76. Moore, A.L. (2000) Landward fining in onshore gravel as evidence for a late Pleistocene tsunami on Molokai, Hawaii. Geology,28, 247–250. Moore, J.G. and Moore, G.W. (1984) Deposit from a giant wave on the island of Lanai, Hawaii. Science,226, 1312– 1315. National Geophysical Data Center/NESDIS/NOAA/U.S. Department of Commerce (2011) ETOPO1, global 1 arcminute ocean depth and land elevation from the US National Geophysical Data Center (NGDC). Research Data Archive at the National Center for Atmospheric Research, Computational and Information Systems Laboratory. https://doi.org/10.5065/d69z92z5. Omira, R.,Quartau, R.,Ramalho, I.,Baptista, M.A. and Mitchell, N.C. (2016) The tsunami effects of a collapse of a volcanic island on a semienclosed basin: the Pico-S~ ao Jorge Channel in the Azores Archipelago. In: Plate Boundaries and Natural Hazards (Eds J.C. Duarte and W.P. Schellart), Geophysical Monograph, 219, pp. 271–287. American Geophysical Union, John Wiley & Sons, Inc., Washington, DC. Pararas-Carayannis, G. (2002) Evaluation of the threat of mega tsunami generation from postulated massive slope failures of island stratovolcanoes on La Palma, Canary Islands, and on the Island of Hawaii. Sci. Tsunami Hazards,20, 251–277. Paris, R.,Giachetti, T.,Chevalier, J.,Guillou, H. and Frank, N. (2011) Tsunami deposits in Santiago Island (Cape Verde archipelago) as possible evidence of a massive flank failure of Fogos volcano. Sed. Geol.,239, 129–145. Paris, R.,Coello Bravo, J.J.,Mart ın Gonz alez, M.E.,Kelfoun, K. and Nauret, F. (2017) Explosive eruption, flank collapse and mega-tsunami at Tenerife ca. 170 ky ago. Nature Commun.,8, 15246. Paris, R.,Ramalho, R.S.,Madeira, J.,  Avila, S.P.,May, S.M., Rixhon, G.,Engel, M.,Br€ uckner, H.,Herzog, M., Schukraft, G.,P erez Torrado, F.J.,Rodr ıguez Gonz ales, A.,Carracedo, J.C. and Giachetti, T. (2018) Megatsunami conglomerates and flank collapses of ocean island volcanoes. Mar. Geol.,395, 168–187. Past Interglacials Working Group of PAGES (2016) Interglacials of the last 800,000 years. Rev. Geophys.,54, 162–219. Perez-Torrado, F.J.,Paris, R.,Cabrera, M.C.,Schneider, J.L., Wassmer, P.,Carracedo, J.C.,Rodriguez Santana, A. and Santana, F. (2006) The Agaete tsunami deposits (Gran Canaria): evidence of tsunamis related to flank collapses in the Canary Islands. Mar. Geol.,227, 137–149. Ramalho, R.S. (2011) Building the Cape Verde Islands. Springer Science & Business Media, Berlin; Heidelberg, NY, 210 pp. Ramalho, R.,Helffrich, G.,Schmidt, D.N. and Vance, D. (2010a) Tracers of uplift and subsidence in the Cape Verde archipelago. J. Geol. Soc. London,167, 519–538. Ramalho, R.S.,Helffrich, G.,Cosca, M.,Vance, D., Hoffmann, D. and Schmidt, D.N. (2010b) Vertical movements of ocean island volcanoes: insights from a stationary plate environment. Mar. Geol.,275,84–95. Ramalho, R.,Helffrich, G.,Cosca, M.,Vance, D.,Hoffmann, D. and Schmidt, D.N. (2010c) Episodic swell growth inferred from variable uplift of the Cape Verde hotspot islands. Nature Geosci.,3, 774–777. Ramalho, R.S.,Quartau, R.,Trenhaile, A.S.,Mitchell, N.C., Woodroffe, C.D. and  Avila, S.P. (2013) Coastal evolution on volcanic oceanic islands: a complex interplay between volcanism, erosion, sedimentation, sea-level change and biogenic production. Earth-Sci. Rev.,127, 140–170. Ramalho, R.S.,Winckler, G.,Madeira, J.,Helffrich, G.R., Hip olito, A.R.,Quartau, R.,Adena, K. and Schaefer, J.M. (2015) Hazard potential of volcanic flank collapses raised by new megatsunami evidence. Sci. Adv.,1, e1500456. Represas, P.,Catal~ ao, J.,Montesinos, F.G.,Madeira, J., Mata, J.,Antunes, C. and Moreira, M. (2012) Constraints on the structure of Maio Island (Cape Verde) by a 3D gravity model: imaging partially exhumed magma chambers. Geophys. J. Int.,190, 931–940. Serralheiro, A. (1970) Geologia do Ilha de Maio (Cabo Verde). Junta Investigacß ~ ao do Ultramar, Lisbon, 103 pp. Stillman, C.J.,Furnes, H.,Lebas, M.J.,Robertson, A.H.F. and Zielonka, J. (1982) The geological history of Maio, Cape Verde Islands. J. Geol. Soc.,139, 347–361. Ward, S.N. and Day, S. (2001) Cumbre Vieja Volcano – potential collapse and tsunami at La Palma, Canary Islands. Geophys. Res. Lett.,28, 3397–3400. Manuscript received 28 September 2018; revision accepted 2 April 2019 Supporting Information Additional information may be found online in the Supporting Information section at the end of the article: Appendix S1. Details of the directional structures observed in Maio’s tsunami deposits. ©2019 The Authors. Sedimentology ©2019 International Association of Sedimentologists, Sedimentology 24 J. Madeira et al.