Internal igneous growth, doming and rapid erosion of a mature ocean island: the Miocene evolution of Maio (Cabo Verde)

Maio Island (Cabo Verde Archipelago) is composed of uplifted Early Mesozoic MORB-type pillow lavas and deep-sea sediments, unconformably overlain and intruded by Miocene igneous rocks. Combined structural analyses and 40Ar–39Ar dating were used to constrain the Miocene evolution of Maio. Structures and ages of uplifted Mesozoic sequences and crosscutting Miocene dykes showed that numerous intrusive events were associated with the intense growth of an igneous core complex in the middle to upper crust, causing semi-circular doming and partial disruption of the Mesozoic strata. Two nosean nephelinite dykes cut the Valanginian Batalha Formation and yielded phlogopite 40Ar–39Ar ages of 10.405 ± 0.033 Ma and 10.570 ± 0.053 Ma (2σ errors). A nosean nephelinite dyke that cuts the overlying Valanginian to Early Aptian Morro Formation yielded an age of 9.273 ± 0.020 Ma. Combined with existing K–Ar and 40Ar–39Ar ages, this confirmed a main period of island growth between ~ 16 and 8.7 Ma. We re-interpreted extensive polymict conglomerates, which occur below the Late Miocene Monte Penoso Formation, as landslide deposits. A nephelinite lava clast yielded a phlogopite 40Ar–39Ar age of 8.666 ± 0.0274 Ma, which represents a maximum age for these landslides and thus confined a period of large-scale flank collapses and erosion to between 8.7 and 6.7 Ma. Flank collapses and further mass wasting during this period may have rejuvenated the igneous activity, i.e., resulting in the formation of the Tortonian/Messinian Monte Penoso and Malhada Pedra Formations, due to decompression-induced melting at upper mantle depths. Such interaction between flank collapses and rejuvenated volcanism may be a key to better understand ocean island evolution worldwide.


Introduction
Volcanic ocean islands show a typical pattern of growth stages, starting with a submarine seamount stage, followed by a shield building stage, collapse of parts of the shield volcano, final volcanism during a rejuvenation stage, and in many cases the final guyot phase (Clague and Dalrymple 1987). The volcanic Cabo Verde Archipelago hosts islands and seamounts in various growth stages, ranging from the seamount stage (e.g., Cadamosto Seamount; Samrock et al. 2019), to the rejuvenation stage (e.g., São Vicente; Ancochea et al. 2010). Maio Island, situated in the E of the archipelago, is one of the oldest islands in the archipelago and is volcanically inactive (Fig. 1).
Large-scale landslides or the collapse of island flanks are typical for volcanic edifices and have a large hazard potential, especially if associated with catastrophic tsunamis (Masson et al. 2008;Watt et al. 2014;Ramalho et al. 2015; van Wyk de Vries and Davies 2015; Madeira et al. 2020). Several landslide and collapse events have been discovered in the Cabo Verde Archipelago, for example, on Santo Antão and Fogo (Masson et al. 2008;Ramalho et al. 2015;Day et al. 1999). The interplay between volcanic growth and destruction by flank collapses is still poorly understood (Cornu et al. 2021).
Similar to the neighbouring island of Boa Vista, Maio is deeply eroded and mostly dominated by gentle hills, with only a few topographic highs up to 436 m in elevation (e.g., Monte (Mte) Penoso; Figs. 1, 2). Maio comprises a basement of uplifted Mesozoic mid-ocean ridge basalt (MORB) lavas and dykes (Batalha Formation). Cretaceous deep-sea sedimentary rocks (Morro and Carqueijo Formations) are overlain by a series of Miocene to Pliocene volcanic rocks and 1 3 conglomerates of the Casas Velhas, Pedro Vaz, Malhada Perdra and Mte Penoso Formations and are partially displaced by the Miocene Central Intrusive Complex (CIC) (Paepe et al. 1974;Stillman et al. 1982;Fig. 1). Large volumes of the Miocene volcanic rocks are eroded, exposing the uplifted seafloor and intrusive basement of the island. The Miocene evolution and processes of doming, collapse and erosion on Maio are still debated and poorly understood, although this deeply eroded island allows the study of different phases of growth, intensive erosion and collapse in great detail.
During two field campaigns on Maio in 2017 and 2018, we found new evidence for an intense erosive phase, predating the youngest volcanic activity on Maio. In this study, we present new field observations of the Miocene intrusive and erosive units, their structural and stratigraphical relationships, and new 40 Ar-39 Ar ages. This allows re-evaluation of  Serralheiro (1970) and Stillman et al. (1982) the ages of intrusive growth of the island, phases of erosion and collapse, and structural processes that shaped the current topography on Maio. In addition, we present a refined model of the stratigraphy and geological evolution of Maio.

Geological setting
The Cabo Verde Archipelago, which comprises 10 islands, several islets and seamounts in two volcanic chains, is situated on one of Earth's most prominent bathymetric anomalies, the Cape Verde Rise (Fig. 1). It rises ~ 2000 m from the surrounding seafloor and is associated with geoid, gravity, and thermal anomalies that are thought to originate in the deep mantle (Dash et al. 1976;Crough 1982;Courtney and White 1986;Ali et al. 2003;Pim et al. 2008;Wilson et al. 2010). The archipelago has been volcanically active since the Oligocene, with the oldest subaerial volcanic activity reported from Sal at ~ 26 Ma (Torres et al. 2002).
The eastern islands in both volcanic chains are deeply eroded, while the western islands have higher topographies (e.g., Ramalho  Gradstein 2012). The islands show a flexural moat due to volcanic loading that has been filled with sediments, which are tilted away from the islands due to uplift of the edifices (Ali et al. 2003).
The geological units on Maio have been mapped by Serralheiro (1970) and further refined and named by Stillman et al. (1982). The oldest rocks on Maio are MORB-type pillow lavas and dykes of the Early Cretaceous (Valanginian) Batalha Formation (Paepe et al. 1974;Stillman et al. 1982;Robertson 1984;Casson et al. 2020). The upper part of the Batalha Formation contains intercalated sediments that include Early Valanginian calpionellids Fourcade et al. 1990;Casson et al. 2020).
The Batalha Formation rocks are overlain by Cretaceous siliceous pelagic nannofossil limestones and marls of the Morro Formation (Valanginian to Barremian/Early Aptian), which were formed in a deep-sea environment (Figs. 1,2,3). Based on the assemblages of macrofossils (e.g., ammonites and aptychi) and nannofossils (e.g., radiolaria and Nannoconus) the Morro Formation is attributed to the Maiolica facies (Stahlecker 1934;Robertson and Bernoulli 1982;Robertson 1984;Casson et al. 2020). The Albian Carqueijo Formation appears to conformably overlie the Morro Formation sequences; however, biostratigraphic analyses suggest a hiatus between these formations (Fig. 3;Robertson 1984;Casson et al. 2020). The Carqueijo Formation is seen as the equivalent to Mid-Cretaceous black shale deposits in the Central Atlantic Ocean ( Fig. 1; Stillman et al. 1982;Robertson 1984;Fourcade et al. 1990;Casson et al. 2020). All these Cretaceous units formed on the seafloor prior to island formation, underwent uplift, and now form a circular structure dipping away from the Miocene CIC. Thrusting results in local repetition of the Cretaceous sedimentary units, especially in the NE of Maio (Monte Branco area; Robertson and Bernoulli 1982).
Intrusive and volcanic activity on Maio started at ~ 19 Ma with the formation of the CIC and dyke intrusions in the Cretaceous sedimentary rocks (Mitchell et al. 1983). The Mesozoic units of the Basement Complex are apparently conformably overlain by volcaniclastic sediments (tuffs, sandstones, and conglomerates) of the Coruja Formation (Figs. 1, 2, 3;Stillman et al. 1982;Robertson 1984). Although rare planktonic foraminifera suggest a Paleocene to Early Miocene age for the Coruja Formation, it is difficult to determine the onset of extrusive volcanism on Maio (Robertson 1984;Bernoulli et al. 2007). The formations of the Basement Complex are unconformably overlain by ankaramite lavas of the subaerial Miocene Casas Velhas Formation, in the SW of Maio, and by sedimentary and volcaniclastic units of the Pedro Vaz Formation in the NE of Maio (Fig. 3;Serralheiro 1970;Stillman et al. 1982). The Casas Velhas Formation is dated by K-Ar at 9.8 ± 0.8 Ma (all errors are quoted at the 2σ level; Mitchell et al. 1983), and by 40 Ar-39 Ar at 11.9 ± 1.0 Ma (Ramalho 2011). Volcanic activity continued with the eruption of Malhada Pedra lava flows, dated by K-Ar at 7.3 ± 0.4 Ma and 6.5 ± 1.2 Ma, and the formation of the Monte Penoso stratovolcano, dated by K-Ar at 6.9 ± 0.4 and 6.7 ± 0.4 Ma ( Fig. 3; Stillman et al. 1982;Mitchell et al. 1983). The remnants of this volcano now form the highest elevation on the island, at ~ 436 m above sea level (a.s.l.). Extensive Pliocene and Quaternary marine terraces, with elevations up to 100 m a.s.l., cover large areas, especially on the coastal regions of Maio. These marine terraces occur at different elevation levels and suggest several phases of uplift and subsidence (Serralheiro 1970;Ramalho et al. 2010a;Samrock et al. 2018). The coastal areas of Maio additionally exhibit numerous Pleistocene tsunami deposits (Madeira et al. 2020).

Methods
This study integrated field observations with 40 Ar-39 Ar geochronology and structural analyses. Structural measurements were taken with the Lambert App (Appel 2017), quoted with dip direction/dip angles. Structural plots were created with Stereonet v.11 (Allmendinger et al. 2012;Cardozo and Allmendinger 2013).
Phlogopite crystals from three dyke samples within the Morro and Batalha Formations, and phlogopite from a nephelinite lava clast in a conglomerate from the landslide deposits directly below Mte Penoso were dated using the 40 Ar-39 Ar technique. The 40 Ar-39 Ar sample locations are detailed in Table 1 (Figs. 2,3,4,5). Sample descriptions are summarized in Appendix A. The samples were prepared for dating by crushing, sieving and washing in distilled water in an ultrasonic bath. Phlogopite separates in the 0.25-0.5 mm and 0.5-1 mm size fractions were separated and handpicked under a binocular microscope, and cleaned in acetone before drying in an oven at 50 °C overnight. The samples were loaded into an aluminum cannister, and irradiated with fast neutrons in a Cadmium-Lined In-Core Irradiation Tube (CLICIT) reactor for four hours at the Oregon State University nuclear reactor, Corvallis, USA. The samples were analyzed at the GEOMAR Argon Geochronology in Oceanography (ArGO) Laboratory in Kiel, Germany. The isotopic analyses were corrected using the total K decay constant and atmospheric air ratio of Steiger and Jäger (1977); using the recommendation of Fleck et al. (2019), and J values were determined using the Taylor Creek sanidine (TCR-2) age standard (28.344 ± 0.0114 Ma, 1σ; Fleck et al. 2019).
A summary of the 40 Ar-39 Ar data are shown in Table 2, and the full 40 Ar-39 Ar data are shown in Appendix B. All our 40 Ar-39 Ar ages in the text and figures are quoted with 2σ internal errors, and include analytical, J value, and irradiation correction factor errors. A complete description of the laser step-heating and single-/multi-grain laser total fusion 40 Ar-39 Ar methods are given in Samrock et al. (2019).

Batalha Fm
Pillow lavas, hyaloclastites; red intercalated sediments and breccia in the top part of the formation.

Mte Penoso Fm (E-flank)
Ankaramite lavas with interbedded tuffs.  Fig. 3 Simplified stratigraphic column of the Maio sequences, separated by the eastern and western flanks of Maio. K-Ar and 40 Ar-39 Ar ages for the Miocene formations are notated in the lithology and fossil columns and were based on this study and data from Bernard-Grif-fiths et al. (1975) and Mitchell et al. (1983). The geological timescale and Chrons were modified after Gradstein (2012). The arrow between the Morro and Carqueijo Formations indicates a hiatus of apparently conformably overlying rocks 1 3 The 40 Ar-39 Ar corrections, age determinations, and plateau and inverse isochron plots were made using the ArArCALC Excel macro (v. 2.5.2) developed by Koppers (2002). The single-crystal total-fusion weighted mean plots were obtained using the Excel Isoplot macro v. 4.15 (Ludwig 2011), and the 40 Ar-39 Ar data and plots are presented in Appendix B. Plateau ages were determined using the criteria of Lanphere and Dalrymple (1978), where the plateau was comprised of at least 3 consecutive heating steps that were within 2σ errors of each other, and where a significant proportion of the 39 Ar release (> 50%) is represented in the plateau steps. 36 Ar/ 39 Ar alteration index (AI) values were used as a guide to gauge the alteration/freshness of the phlogopite samples, calculated using the criteria of Baksi (2007), with a cut-off value of 36 Ar/ 39 Ar AI < 0.0013 for the single-/multi-crystal totalfusion analyses, and < 0.0005 for the step-heated samples.

Results
In the following section, we (a) present field observations from the Pre-Miocene Basement Complex and the Miocene sequences, and (b) combine the structural relationships for four dykes intruded into the Basement Complex at two key locations (see Fig. 2b, c) with new 40 Ar-39 Ar data (Table 2). In addition, we (c) describe a newly discovered mass wasting deposit, which we ascribe to the Mid-to Late Miocene Pedro Vaz Formation.

Pre-Miocene Basement Complex
The Basement Complex comprises Early Cretaceous pillow lavas of the Batalha Formation, the Early Cretaceous pelagic limestones of the Morro and Carqueijo Formations,   Values in bold indicate preferred ages ID # identification number, SH step-heating, SCTF single crystal total-fusion, MCTF multi-crystal total-fusion (2 grains maximum), High-T hightemperature, WMA weighted mean age, MSWD mean square weighted deviation, P probability, SF spreading factor, Wt% weight % *The combined SCTF/SH weighted mean age for sample 17-16/3 was calculated using the 40 Ar*/ 39 Ar ratios, after the approach of Heath et al. (2018), and includes the analytical and J value errors red intercalated sediments that contain angular fragments and altered sedimentary rocks, which may have originated from the underlying red sediments ( Fig. 3; Appendix C). Brecciation of lithified sediments during magmatic growth of the Batalha Formation suggests that MORB-type magmatic activity in this area covered a considerable time span.
The limestones of the Morro and Carqueijo Formations are observed in some of the hills surrounding the CIC, e.g., Mte Branco and Mte Carqueijo to the NE and N, respectively (Fig. 1). Limestone beds are tilted away from the CIC, seemingly uplifted and tilted by CIC intrusions during the Miocene, and dip towards the NE at Mte Branco, SW at Mte Esgrovere, W at Ribeira (Rib) do Morro, and E at Rib da Baia (Figs. 1, 2). At Mte Esgrovere, the units dip with an average value of 49° towards the SW, but dip angles are steeper (up to 70°) closer to the CIC (Fig. 2d). At Rib do Morro in the W, the units have an average dip angle of 51°, but numerous meter-scale upright folds are observed with fold axes trending to the SE. At the top of Mte Branco, the limestone beds of the Morro Formation dip almost vertically and are thrust towards the NE (Mte Branco Thrust; Fig. 2b). Other outcrop locations are heavily intruded by dykes or covered by alluvium.
The Morro Formation consists of up to ~ 3 dm-thick beds of grey limestones that contain cm-sized chert lenses and bands, and show fine and swirly lamination. Stylolites are common and bioturbation is observed in the form of burrows, up to 2 cm in diameter and often associated with diagenetic dolomite rhombs (Appendix C). Thin sections of the Morro Formation rocks reveal radiolaria (in the form of calcite-filled solution-molds), and rare fragments of gastropods, sponge spicules, echinoid fragments, radiolaria (Nasselaria) and foraminifera. At Mte Esgrovere, we found ammonites and aptychi, ~ 150 m above the stratigraphic base of the Morro Formation (Latitude/Longitude: 15.162936, − 23.182217). Some of these ammonites are well preserved and were identified as Phylloceras cf. infundibulum (d'Orbigny 1840) and Neocomites neocomiensis (d'Orbigny 1841) (Appendix C), which indicates a Valanginian to early Hauterivian age (136.4-130 Ma; Fossilworks.org 2019) and supports similar ammonite findings at Mte Esgrovere by Casson et al. (2020). At Rib do Morro, we also found ammonites exposed on a bedding surface ~ 140 m above the Morro Formation stratigraphic base. These ammonites are unfortunately not well preserved, but confirm the first descriptions of ammonites from this location by Stahlecker (1934) that were revised by Casson et al. (2020). Additionally, the Morro Formation rocks contain Early Cretaceous nannofossils of biostratigraphic value (listed in Appendix C), which support Valanginian to Early Aptian ages as quoted by Casson et al. (2020).
The Carqueijo Formation overlies the Morro Formation rocks and consists of cm-thick limestone beds, alternating with cm-thick elongated chert bands (usually 2-3 cm thick, locally up to 10 cm thick), soft marls, and marly and shaley limestones. Bright purple and green coloration of these rocks suggest hydrothermal alteration, which is commonly observed, as well as the occurrence of dyke intrusions. Macrofossils are rare, but thin sections reveal abundant calcite filled solution molds of radiolaria, and some samples yield nannofossils of biostratigraphic value (listed in Appendix C).
The Coruja Formation successions contain tuffs and conglomerates, which overlay the Carqueijo Formation and were mapped for large areas in the N and NW of the Basement Complex, and W of Mte Penoso (Serralheiro 1970;Stillman et al. 1982). However, we found that especially flat areas are covered by Quaternerary sediments, plus reforestation and deposition of recent sediments cover many of the originally described outcrop locations. We were able to confirm the occurrence of Coruja Formation outcrops at the southern foot of Mte Penoso, and at the SW flank of Lombo Vermelho, where the formation is composed of orange weathered tuffs, bioturbated sandstones, and conglomerates (Appendix C). The conglomerates contain wavy-crossbedding and channel structures. Conglomerate clasts are strongly altered, and the original mineralogy was replaced by secondary minerals, despite partly retaining the original igneous textures. Unfortunately, these clasts were too altered to be used for 40 Ar-39 Ar dating.

Miocene deposits
The Early Miocene CIC is exposed in the southern central part of Maio island, where it forms a relatively flat, almost bowl-shaped, topography (Figs. 1, 2). It is composed of intrusions of essexites, syenites, and pyroxenites, cut by numerous basanitic, ankaramitic, trachytic, and carbonatite dykes. These Miocene units intruded the Early Cretaceous pillow lavas of the Batalha Formation, and the Early Cretaceous pelagic limestones of the Morro and Carqueijo Formations, which were uplifted and partly tilted during magmatic intrusion activity. The rocks of the Basement Complex are also discordantly overlain by Miocene volcanic deposits. The Casas Velhas Formation, composed of ankaramite lava flows, is exposed in river sections in the S and SW of Maio (Fig. 1). In the NE of Maio, the sediment-dominated Pedro Vaz Formation is exposed (Fig. 1), and is typically composed of conglomerates that contain (sub-) rounded igneous clasts, with clast diameters of up to 40 cm. Locally, pyroclastic deposits (tuffs and ignimbrites) and ankaramite lavas appear as lenticular bodies in sections within this formation. This agrees well with observations of Stillman et al. (1982), suggesting that the Pedro Vaz Formation crops out mainly between the Mte Penoso and Mte Santo António.
In the W and SE of Maio, the olivine-nephelinites and melilitites of the Malhada Pedra Formation are exposed (Fig. 1). These rocks are thickest at Mte Batalha and inclined to the SE. In the NE of Mte Esgrovere, these rocks are exposed in an open pit mine, where they form columns. Along the road towards Figueira da Horta, pillow lavas are also exposed, suggesting partial formation during periods of sea-level highstands.
The youngest volcanic rocks on Maio belong to the Mte Penoso Formation, which is exposed on Mte Penoso and Mte Santo António in the N of Maio (Fig. 1). Ankaramite lava flows are interbedded with tuffs and other pyroclastic deposits. The Mte Penoso Formation once formed a stratovolcano, with units dipping towards the NE (Stillman et al. 1982). The Mte Penoso Formation rests unconformably on rocks of the Coruja Formation at the southern flanks of Mte Penoso and Lombo Vermelho (Fig. 2c), and on conglomerates and pyroclastics of the Pedro Vaz Formation. Feeder dykes to the former stratovolcano are observed in the northern part of the island, where they cross-cut underlying rock units.
All formations on Maio are cross-cut by dykes, especially in the older parts of the island, where the dyke density is very high and obscures former rock units. Carbonatite dykes are present, especially in the pelagic limestones and intrusive rocks of the CIC. A larger complex of carbonatites is present E of Mte Vermelho and NW of Mte Penoso (Fig. 1).

Ar-39 Ar ages of dykes and tectonic structures in the Basement Complex
A small area of Batalha Formation rocks is exposed at the NE flank of Mte Branco (Figs. 2b, 4). At this location, the shape of the pillows is still visible, but they were heavily intruded by dykes (Fig. 4a). These dykes are all truncated by the Mte Branco Thrust (Fig. 4b, c), and faults and fractures parallel to this thrust were subsequently filled with calcite.
We obtained 100 structural measurements (dip direction/ dip) of dykes intruding the Batalha Formation (Fig. 4d-f; Appendix D). Dip angles vary between 25° and 90°. One third of the measured dykes are almost vertical with dip angles of > 80°, and the majority (60%) have dip angles between 50° and 79° (Fig. 4). The steep dykes (> 80°) dip towards the NW, NNE and SSE (Fig. 4d, f). The moderately dipping dykes (25-79°) have a larger variation in orientations, but also show a similar NW dipping component (Fig. 4d, e). However, there is also an additional group of ENE and E dipping dykes, and a minor SSE dipping component (Fig. 4d, e). Dykes dipping towards the SW are completely absent (Fig. 4d). In two cases, the age relationships of the dykes were recorded by cross-cutting relationships. The older dykes are inclined towards the W and NW (275/58 and 341/63), while the younger dykes cross-cut the old dykes and dip towards the NE (053/53 and 044/59).

Miocene mass wasting deposit and its 40 Ar-39 Ar age
At the base of Mte Penoso and at the southern flank of Lombo Vermelho, conglomerates are found that had not previously been described in detail (Figs. 2c, 5). The two main outcrop areas are separated by more than 1 km across an erosional valley. Below Mte Penoso, exposed to the W and S, this ~ 50 m-thick conglomerate sequence is exposed from ~ 200 m a.s.l. in the valley to ~ 250 m asl and was laterally traced for ~ 1 km along the base of the hill. It is overlain by lava flows and pyroclastic deposits of the Mte Penoso Formation (Fig. 5a). The conglomerate is polymict and poorly sorted, with clasts typically ranging from centimetres to several decimetres in diameter. Clast sizes vary by more than one order of magnitude at any location. The clast contents range between ~ 25 and 50 volume %, resulting in both matrix-supported and clast-supported fabrics on an outcrop scale (Fig. 5b-d). Some areas comprise subunits or flow units identified by difference in dominant grain size, typically separated by a thin indistinct transition zone. The individual clasts range from angular through sub-angular and sub-rounded to well rounded and are set in a weathered matrix that locally shows signs of hydrothermal alteration. Similar sequences have been characterized as debris flow or flank collapse deposits (e.g., Nemec and Steel 1984). The clasts represent a wide range of rock fragments, including alkali-basaltic and nephelinitic fragments ranging from dense to vesicle-rich, ankaramite, phonolite lava, and subordinate alkaline plutonic fragments, which closely resemble the Miocene igneous rocks on Maio. Many clasts are weathered and some show possible hydrothermal alteration. Locally, pyroclastic, lenticular units and occasional small lava flows are also identified and the deposit is therefore multi-phase, testifying to intermittent volcanism during this erosional period.
At Lombo Vermelho, such conglomerates also crop out discordantly above units of the Carqueijo and Coruja Formations, which were in turn intruded by numerous dykes. The Lombo Vermelho conglomerate is also overlain by lava flows and pyroclastic deposits of the Mte Penoso Formation (Fig. 5e), but at this location the conglomerates are thinner (~ 20 to 30 m thick) compared to the Mte Penoso outcrop area (~ 50 m thick). Clasts have been derived from various lithologies as described above for Mte Penoso, but additionally include subordinate limestone and sedimentary rocks from the underlying Carqueijo Formation. The lower part of this conglomerate was strongly hydrothermally altered and shows a network of secondary mineralized cracks.

Age of intrusive growth, dyke emplacement and tectonic activity at Mte Branco
Two of our nephelinite dyke samples were found intruding Batalha Formation rocks below the Mte Branco Thrust (Fig. 4), yielding 40 Ar-39 Ar ages of 10.570 ± 0.053 Ma and 10.405 ± 0.033 Ma (Table 2). On the southern coast of Maio, dykes intruding the Batalha Formation yield various K-Ar ages ranging from 12.2 to 9.1 Ma (12.2 ± 0.4 Ma, 11.1 ± 0.6 Ma, 10.8 ± 0.4 Ma and 9.1 ± 0.4 Ma; Mitchell et al. 1983), which partly overlap with our 40 Ar-39 Ar ages for the Monte Branco area further north (Fig. 6). Our new 40 Ar-39 Ar phlogopite ages overlap within 2σ errors with 2 of the 3 literature K-Ar ages (11.0 ± 0.6 Ma, 10.8 ± 2.2 Ma and 9.9 ± 0.4 Ma) from alkaline lamprophyre dykes cutting the nearby Albian Carqueijo Formation (Mitchell et al. 1983) and 2 of the youngest K-Ar ages from dyke intrusions in the CIC (10.9 ± 0.8 Ma to 10.0 ± 0.8 Ma; Fig. 6). In addition, they overlap within 2σ errors with a K-Ar age from a Casas Velhas Formation ankaramite lava (9.8 ± 0.8 Ma; Mitchell et al. 1983).
The preferred 40 Ar-39 Ar age of 9.273 ± 0.020 Ma for our Morro Formation dyke sample from the Mte Branco region overlaps within 2σ errors with all 4 K-Ar ages (9.3 ± 0.4 to 8.1 ± 1.2 Ma) obtained from two samples of Mte Branco basanite and ankaramite intrusions presented by Mitchell et al. (1983;Fig. 6). However, Mitchell et al. (1983) obtained an older 40 Ar-39 Ar plateau age (10.5 ± 0.6 Ma) from a split of the same Morro Formation/Mte Branco basanite sill that also yielded younger 3 K-Ar sample split ages of 9.3 ± 0.5 to 8.1 ± 1.2 Ma. This 10.5 Ma older age is similar to a Bernard-Griffiths et al. (1975) K-Ar age of 10.6 ± 1.0 Ma for micas from a carbonatite dyke sample at Mte Branco (Fig. 6). Mitchell et al. (1983) suggest that the formation age of their basanite sill is 10.5 Ma and their three younger K-Ar ages may have been reset by a younger event, even though their samples are fresh and show no evidence of thermal overprinting. It is difficult to confirm the Mitchell et al. (1983) 10.5 ± 0.6 Ma 40 Ar-39 Ar age, as no accompanying isochron or inverse isochron data are presented to confirm this plateau age. Given the very fresh nature of the phlogopites from our Morro Formation nosean nephelinite dyke sample and reproducible 40 Ar-39 Ar single-crystal total-fusion and inverse isochron ages, we are confident that the 9.273 ± 0.020 Ma is a magmatic formation age.
The context of multiple phases of intrusive growth is verified by the new and published geochronology ages. The overall distribution of new and published dyke K-Ar and 40 Ar-39 Ar ages suggest the preservation of several phases of intrusive growth in the different rock formations of Maio ( Fig. 6; Grunau et al. 1975;Bernard-Griffiths et al. 1975;Mitchell et al. 1983;Ramalho 2011). Gabbros, essexites, pyroxenites and syenites from the CIC yield K-Ar ages of 8.2 ± 0.4 up to 21.6 ± 12.6 Ma (Grunau et al. 1975;Mitchell et al. 1983). However, the 2σ errors are large for some of these ages, therefore we chose to reject the three Grunau et al. (1975) K-Ar ages with the largest 2σ errors of 12.5 ± 4.8 Ma, 14.0 ± 8.0 Ma and 21.6 ± 12.6 Ma, as these cannot be used for any time discrimination in such a young age interval. Based on the combined 40 Ar-39 Ar and K-Ar age data of dyke and sill intrusions we tentatively suggest different periods of intense intrusive activity and by inference correspondingly high volcanic activity in Maio: (1) a first phase from ~ 16 to 13.5 Ma is recorded in dykes that intrude CIC and Coruja Formation rocks; (2) a second phase from ~ 12 to 10 Ma is obtained primarily in dykes that intruded the CIC, Batalha and Carqueijo Formations; and (3) a final phase from ~ 9.5 to 8 Ma is recorded mainly in dykes that intrude Morro Formation rocks (Fig. 6). We emphasise that the second and third phases partly overlap in time, but represent a geographical shift in dyking and intrusion activity. All three phases are recorded in dykes/sills and overlap Fig. 6 Compilation of available Miocene K-Ar and 40 Ar-39 Ar ages (all with 2σ errors) from Maio, Cabo Verde (Bernard-Griffiths et al. 1975;Mitchell et al. 1983;Ramalho 2011), including new 40 Ar-39 Ar ages presented in this study (marked by black arrows). The colors of the data points are based on the colors of the rock formations in the geological map (Fig. 1). Grey shaded areas indicate suggested phases of volcanic activity. Data that originated from one sample but yielded different K-Ar or 40 Ar-39 Ar ages are encircled. The small 2σ errors for our new 40 Ar-39 Ar ages are smaller than the symbol sizes within 2σ errors with the K-Ar and 40 Ar-39 Ar ages recorded in the CIC intrusive rocks (Fig. 6). The second (~ 12-10 Ma) phase is previously mentioned by Mitchell et al. (1983) as a period of intense intrusive activity on Maio. It is interesting to note that this ~ 12-10 Ma phase mainly records dyke/sill emplacement in the CIC, Batalha, and Carqueijo Formations, whereas the third phase (~ 9.5-8 Ma) is mainly recorded in dykes that intrude Morro Formation rocks in the Mte Branco area. The different geographical and lithological distributions of ages could be explained by a geographical shift in volcanic activity or be caused by tectonic activity. Both the second and third phases overlap with K-Ar and 40 Ar-39 Ar ages of ankaramite lavas from the Casas Velhas Formation (11.9 ± 1.0 Ma and 9.8 ± 0.8 Ma; Mitchell et al. 1983;Ramalho 2011).
At one of our study locations, N of Mte Branco, the pillow lavas are intensely intruded by numerous dykes that are all truncated by the Mte Branco Thrust. The variation in dip angles from vertical to 25° suggests different groups of intrusions, varying from steep dykes to inclined sheets. The dip directions of the moderately dipping dykes (25° to 79°) suggest conjugate extension (dipping twoards NW and ESE; Fig. 4d, e), which may be due to a doming process caused by magmatic underplating and thermal uplift (e.g., Klügel et al. 2005). However, we cannot exclude the possibility that the orientations of the Batalha Formation rocks were also changed due to tectonic activity.
In general, the dykes that intruded the Morro Formation are younger than the Batalha Formation dykes below the Mte Branco Thrust and they also show different orientations: while the dykes in the Morro Formation dip towards the North and are relatively flat (336/10, 007/24 and 358/21), the dykes in the Batalha Formation below the thrust show various orientations (NW, ENE and SSE for the moderately dipping dykes and NW, NNE and SSE for the almost vertical dykes) The different orientations of dykes suggest different pulses of magma and therefore different phases of intrusive activity, which could be related to various growth stages in the CIC. However, the dyke orientations may have changed due to the thrusting and tectonic activity and, therefore, cannot be used to establish former eruption centres.

Miocene erosive activity: conglomerates
The conglomerates at the base of Mte Penoso and at the southern flank of Lombo Vermelho are variable in composition, rounding and grain sizes of clasts, and in facies types. This variability is typical for complex volcanic landslide and debris avalanche deposits (e.g., Bernard et al. 2009;Roverato et al. 2014;van Wyk de Vries and Delcamp 2015). In general, overloading, over-steepening, uplift and subsidence, magma emplacement in the crust and erosion can destabilize a volcanic edifice. Furthermore, debris avalanches and flank collapse events may be triggered by magmatic (intrusive or extrusive), seismogenic (tectonic/volcanic earthquakes), or environmental (sea level changes, precipitation) events (McGuire 1996).
According to Serralheiro (1970) and Stillman et al. (1982), the Pedro Vaz Formation crops out mainly between Mte Penoso and Mte Santo António and in small patches in the SW of Maio (e.g., Rib do Morro, close to Barreiro; Fig. 1). It contains "rudites" or matrix-supported conglomerates, interbedded with fine "tuffs" and ankaramite lava flows, separated by thin "arenites", and is intruded by numerous ankaramite dykes (Stillman et al. 1982). Locally confined, it apparently includes material from the underlying Casas Velhas Formation (Stillman et al. 1982). The deposits presented in this study are mainly composed of polymict conglomerates and contain dispersed pyroclastic rocks and lava flows, which agrees well with the descriptions of Stillman et al. (1982). Based on their stratigraphic position and the resemblance to previous descriptions, we therefore suggest that the conglomerates presented in our study are part of the Pedro Vaz Formation. An overlap in ages between the Pedro Vaz and Casas Velhas Formations is also noted by Casson et al. (2020;Fig. 3).
The Pedro Vaz Formation multi-phase deposits represent a period of both rapid erosion (landslides and flank collapses) and simultaneous moderate volcanic growth (pyroclastic flows and lavas), which most likely coincides with the onset of rejuvenated magmatic activity. Such periods of simultaneous volcano growth and destruction are common worldwide (e.g., Tutucapa volcano, Peru (Valderrama Murillo 2016); Stromboli, Italy (Romagnoli et al. 2009);Mount St. Helens, USA (McEwen and Malin 1989); Augustine volcano, Alaska (Begét and Kienle 1992); Tenerife, Canary Islands (Hunt et al. 2011(Hunt et al. , 2013(Hunt et al. , 2018; and Montserrat (Cassidy et al. 2014)). The shield stage of an ocean island volcano is defined as the period with highest magma supply rates and thus the most rapid growth. Intense erosion processes dominate the morphologic evolution of the mature edifice, as magma extrusion and intrusion rates diminish towards the end of the shield stage (Carracedo 1999;Ramalho et al. 2013). Such erosional phases are well-documented worldwide and include gravity sliding on both local and regional scales (e.g., Hunt et al. 2018;Urlaub et al. 2018). For example, in the Cabo Verde Archipelago, several large landslides and flank collapses have been described (Day et al. 1999;Le Bas et al. 2007;Masson et al. 2008;Ancochea et al. 2010;Martínez-Moreno et al. 2018;Barrett et al. 2020). Land slide deposits on São Vicente can also be compared to the conglomerates on Maio (Ancochea et al. 2010).
Our new 8.7 Ma 40 Ar-39 Ar age data combined with literature K-Ar and 40 Ar-39 Ar ages of the overlying Malhada Pedra and Monte Penoso Formations (Bernard-Griffiths et al. 1975;Mitchell et al. 1983) confine the Miocene erosional phase at Maio to between 8.7 and 6.7 Ma (base of Monte Penoso Formation), coinciding with dispersed and rejuvenated volcanism of the temporally overlapping Malhada Pedra and Pedro Vaz Formations. We additionally suggest that flank collapse activity and rapid mass wasting in the Tortonain/Messinian (8.7-6.7 Ma) may have enhanced the rejuvenated igneous activity on Maio, due to decompression-induced melting at upper mantle depths (Manconi et al. 2009;Cassidy et al. 2015;Hunt et al. 2018). It remains unclear what caused the presented multiphase flank collapse events. They may have been emplaced contemporaneously with fault activity along the Mte Branco Thrust, which was possibly active between 9 and 8.7 Ma, following the final dyke intrusions at Mte Branco. However, the collapse events could also have been influenced by early rejuvenated activity of Malhada Pedra and/or Monte Penoso Formations, which may have formed the local pyroclastic sequences within the conglomerate sequences.
The well-constrained 40 Ar-39 Ar phlogopite age of 8.666 ± 0.027 Ma obtained from a nosean nephelinite clast within the conglomerate directly underlying the Mte Penoso Formation is regarded as a maximum age for this conglomerate deposit and a minimum age for previous igneous units (i.e., the Casas Velhas Formation). The presence of two older age groups of phlogopites within this sample may indicate the presence of older phlogopite antecrysts and/or xenocrysts, which are distinguished by their ages and wt.% K contents (Table 2; Appendix B). Xeno-/antecrysts in igneous rocks are typical for Cabo Verde rocks and the plumbing system below the islands and seamounts (e.g., Samrock et al. 2019;Barker et al. 2019Barker et al. , 2021 and is further supported by the presence of rounded phlogopites in the sample (Appendix A).

Refined geological model of Maio
Based on field observations and new 40 Ar-39 Ar dating of dykes and a conglomerate clast, we present a refined model of the geological evolution of Maio (Fig. 7).
Initially, Early Cretaceous pillow lavas of the Batalha Formation and Early Cretaceous pelagic limestones from the Morro and Carqueijo Formations are deposited on the sea floor ( Fig. 7a; e.g., Robertson 1984, Casson et al. 2020. We confirm descriptions of red intercalated sediments that include calpionellids in the upper Batalha Formation at Mte Esgrovere Casson et al. 2020). Our ammonite findings and microfossil assemblages in the Morro Formation matches the previously proposed Valanginian to Early Aptian ages for the Morro Formation and the Albian age for the overlying Carqueijo Formation (Casson et al. 2020).
It is unclear when subaerial volcanism on Maio started, as there are little data for the rare outcrops of the Coruja Formation. Mitchell et al. (1983) published K-Ar ages for Coruja Formation igneous clasts in a conglomerate, yielding Late Jurassic to Late Cretaceous ages (157.8 ± 8.0,99.8 ± 4.4,and 73.2 ± 2.4 Ma). Rare planktonic foraminifera suggest a Paleocene to Early Miocene age for the Coruja Formation, but the precise age of these units, and therefore the age of the first subaerial volcanism on Maio, remains weakly constrained (Robertson 1984). Gravity data suggest that an eruptive center is located in the SE of Maio (Represas et al. 2012). In this area, dyke density is high and rocks of the Miocene CIC are mainly preserved, and most successions of sedimentary units of the Basement Complex have been removed by erosion. However, early volcanic units are absent; therefore, it is not clear if there is an early volcanic center in SE Maio. Volcanic ash layers in marine sediments NE of the Cabo Verde Archipelago suggest the onset of volcanism on the Cabo Verde Islands during the early Middle Miocene (DSDP Site 368; Lancelot et al. 1978).
During the Early Miocene (at or before 19.3 ± 1.2 Ma; Mitchell et al. 1983) the CIC begins to form with the emplacement of shallow intrusions in the middle and upper crust, i.e., the formation of an intrusive core complex (Fig. 7b, c). Such pronounced intrusion activity is typically associated with volcanic activity and we also assume this in this case for Maio. Emplacement of plutonic rocks led to mechanical and thermal uplift and doming, which induced the semi-circular dip of the former seafloor units ( Fig. 7b; Serralheiro 1970), and stoping probably led to the disruption of strata above the intrusive complex (e.g., Klügel et al. 2005Klügel et al. , 2015Montanari et al. 2017).This intrusive period may have lasted only a few million years, leading to a rapid flare up and accompanying uplift of the edifice (Klügel et al. 2005). Such comparatively rapid uplift has also been recorded from other Cabo Verde islands, e.g., Santiago (Marques et al. 2020). It has been shown that locally confined intrusive activity at crustal levels is responsible for distinct island uplift in the region (Ramalho et al. 2010b, c;Ramalho 2011). The intrusive activity also caused crustal thickening, with the thickest crust being present below the older eastern islands of the Cabo Verde Archipelago, such as Sal and Maio (Lodge and Helffrich 2006).
At a later stage of intrusive island growth (from ~ 12 to 8 Ma; Fig.7d, e), dyke emplacement plays a more important role. Some dykes probably stalled, leading to edifice growth at depth, while others reached the surface, feeding eruptions and extrusive growth and emplacement of the Casas Velhas Formation. The importance of dyke emplacement to the growth of volcanic edifices has been shown previously (e.g., Annen et al. 2001;Kervyn et al. 2009). The Casas Velhas Formation may be regarded as rocks from a 'Lower Volcanic Complex' and is possibly associated with a shield building stage, which may have started prior to ~ 11.9 Ma and overlaps with the last intrusive activity in the CIC that phased out at ~ 8 Ma (Fig. 7d, e; Ramalho 2011). The three high activity intrusive phases recorded in dykes (16-13.5 Ma, 12-10 Ma, and 9.5-8 Ma; Fig. 6) overlap with ages recorded in intrusive rocks from the CIC, which may partly be due to the overprinting of older rocks (Bernard-Griffiths et al. 1975;Mitchell et al. 1983).
Intrusive activity not only led to edifice growth from the inside, but also to extrusive activity and growth on the outside. Dyke K-Ar and 40 Ar-39 Ar ages from the second and third intrusive phases (12-10 Ma and 9.5-8 Ma) correspond with K-Ar ages from the subaerial Casas Velhas Formation lava flows and pyroclastic rocks, which are preserved in the S and SW parts of the island, mainly in river sections (Mitchell et al. 1983; Fig. 1). Casas Velhas Formation rocks are not preserved in the N and NE parts of Maio, possibly due to the later flank collapse events, but dykes of similar ages are present in these regions. An ankaramite lava from the base of the Casas Velhas Formation in the S of Maio yields a K-Ar age of 9.8 ± 0.4 Ma (Mitchell et al. 1983), although the authors suggest that the age may be older as the dyke may have been thermally overprinted due to the high dyke density at the sampling location. Mitchell et al. (1983) additionally observed petrological similarities between the youngest group of plutonic rocks from the CIC in the NW of Maio (Mte Vermelho) and lavas from the Casas Velhas Formation. This suggests that parts of the CIC and dykes cutting the Basement Complex rocks are the intrusive counterpart of the extrusive volcanics in the Casas Velhas Formation, with some of the dykes feeding surface eruptions. A period of strong erosion and low volcanic activity followed the deposition of the Casas Velhas Formation lava flows, as shown by the emplacement of massive conglomerates and sparse pyroclastic deposits and the removal of large volumes of the previous volcanic edifice (Figs. 2c, 3, 7f). The 8.666 ± 0.027 Ma 40 Ar-39 Ar phlogopite age from a nephelinite clast from the Pedro Vaz Formation conglomerate, and the 6.7 ± 0.4 Ma K-Ar age from an ankaramite from the base of Mte Penoso (Mitchell et al. 1983), can be used to confine this erosive period to the Tortonian/Messinian. The sedimentdominated Pedro Vaz Formation is described as locally overlying the volcanic-dominated Casa Velhas Formation and incorporated rocks from the Casas Velhas Formation rocks in its base (i.e., in the western part of Rib do Morro and N of Barreiro;Serralheiro 1970;Stillman et al. 1982). Similar sedimentary deposits, which are locally confined and heterogenous, are difficult to differentiate and therefore it is hard to constrain their relative ages. This is especially true in this sequence of laterally heterogeneous units. We suggest that the younger parts of both the volcanic-dominated Casas Velhas Formation in the SW and the sediment-dominated Pedro Vaz Formation in the NE show similar features (i.e., conglomerates with pyroclastic layers) and they may overlap in age.
Previously published ages of the Pedro Vaz Formation are based on foraminifera findings from siltstones collected in the Pedro Vaz area (Rigassi 1972), who suggested an open sea environment and a planktonic zone N12 of Middle Miocene age. Reexamination of the Rigassi (1972) list of foraminifera reveals that only Globorotalia fohsi robusta and G. fohsi lobata are currently relevant for biostratigraphy and are now listed as modifications of Fohsella fohsi (s.l.) (J. Schönfeld, pers. comm.; Wade et al. 2011). This supports planktonic zone N12 (reclassified as M9a-M9b) with ages of 13.41-11.79 Ma (Wade et al. 2011). The Rigassi (1972) Pedro Vaz Formation microfossils have been previously used to confirm Miocene (> 12 Ma) ages for the underlying Casas Velhas Formation. However, Stillman et al. (1982) states that the intercalation of open-sea rocks with subaerial lavas and conglomerates is problematic and thus these interpretations should be treated with caution.
Tectonic activity on the Mte Branco Thrust must postdate the intrusive activity in the Batalha Formation, as the dykes are all truncated by the thrust, which must therefore be younger than 10.405 ± 0.033 Ma. It is not clear if one of the Mitchell et al. (1983) samples from Mte Branco, yielding a K-Ar age of 9.0 ± 0.4 Ma, comes from the hanging or the foot wall of the thrust, but Mitchell et al. (1983) mention that it is truncated by the thrust, therefore the age of this thrust must also be younger than 9.0 Ma. Such reverse faults are common in volcanic settings around the world and typically occur at the end of the doming process caused by shallow magmatic intrusions (Montanari et al. 2017). Tectonic activity may have triggered the major Late Miocene erosive events, thus leading to the formation of sedimentary deposits overlying the Basement Complex Rocks (below Mte Penoso, in the Pedro Vaz Formation) and may therefore be contemporaneous in age (Tortonian/Messinian; Fig. 7f).
The eruption of Malhada Pedra Formation rocks (olivine melilitites) and Mte Penoso Formation nephelinite lavas mark the final phase of volcanism (Fig. 7g). Both formations are physically separated due to the deep erosion levels in the central part of Maio, but have overlapping ages: The Malhada Pedra olivine melilitite lavas yield K-Ar ages of 7.3 ± 0.4 Ma and 6.5 ± 1.2 Ma, whereas ankaramites from the base of the Mte Penoso lava pile and from Mte Santo António give K-Ar ages of 6.7 ± 0.4, and 6.9 ± 0.4 Ma, respectively ( Fig. 6; Mitchell et al. 1983). While the Mte Penoso lavas dip towards the NE, the Malhada Pedra lavas dip towards the SE, indicating that they may originate from a restricted area in the W-central part of Maio. A missing systematic record of dyke directions, as well as faulting and thrusting of units complicates the effort to establish the positions of former eruptive centers. An eruptive center origin in the NW central region of Maio would fit well with a source area from the NW part of the deeper intrusive complex or island core, as deduced from gravity data (Represas et al. 2012), which again implies that this area is a younger part of the CIC. This is consistent with the youngest K-Ar ages (8.6 ± 0.4 Ma and 9.2 ± 0.4 Ma) of CIC intrusive rocks from the Monte Vermelho area (Mitchell et al. 1983). We additionally suggest that flank collapse activity in the Tortonian/Messinian (8.7 to 6.7 Ma) could have enhanced rejuvenated igneous activity of Malhada Pedra and Monte Penoso Formations on Maio, due to decompressioninduced melting in the upper mantle.
After the final phase of volcanism (Fig. 7g), erosion was the dominant process shaping Maio. The topography has been flattened and only competent rocks remain as hills (Fig. 7h). Lava flows of the final volcanic phase capped the underlying conglomeratic units, protecting the last remnants from erosion. Quaternary dunes, alluvium and beach sediments are the youngest sediments on Maio.

Conclusions
Based on field observations, structural data and 40 Ar-39 Ar dating, we can refine the Miocene evolution of Maio Island in the Cabo Verde Archipelago.
Intrusive magmatic growth of Maio probably started in the Early Miocene with the emplacement of shallow plutonic rocks. Edifice growth was dominated by several phases of dyke intrusions and accompanying effusive activity in the Middle Miocene, leading to rapid growth of the island. The associated flare up and uplift is recorded in dyke orientations and resulted in semi-circular dips, deformation and disruption of Mesozoic marine sedimentary successions.
A newly discovered series of conglomerates, interwoven with local pyroclastic units, is interpreted as a flank collapse dominated phase of the Pedro Vaz Formation. The flank collapse events mark a period of rapid erosion in the Tortonian/ Messinian (8.7-6.7 Ma), was possibly triggered by tectonic activity (i.e., the Mte Branco Thrust) and corresponds to the onset of rejuvenated volcanic activity. Mass wasting may have induced rejuvenated volcanism of the Malhada Pedra and Monte Penoso Formations due to decompressioninduced melting in the upper mantle.