Constraints on the structure of Maio Island (Cape Verde) by a three-dimensional gravity model: imaging partially exhumed magma chambers

Summary

We propose a 3-D gravity model for the volcanic structure of the island of Maio (Cape Verde archipelago) with the objective of solving some open questions concerning the geometry and depth of the intrusive Central Igneous Complex. A gravity survey was made covering almost the entire surface of the island. The gravity data was inverted through a non-linear 3-D approach which provided a model constructed in a random growth process. The residual Bouguer gravity field shows a single positive anomaly presenting an elliptic shape with a NW–SE trending long axis. This Bouguer gravity anomaly is slightly off-centred with the island but its outline is concordant with the surface exposure of the Central Igneous Complex. The gravimetric modelling shows a high-density volume whose centre of mass is about 4500 m deep. With increasing depth, and despite the restricted gravimetric resolution, the horizontal sections of the model suggest the presence of two distinct bodies, whose relative position accounts for the elongated shape of the high positive Bouguer gravity anomaly. These bodies are interpreted as magma chambers whose coeval volcanic counterparts are no longer preserved. The orientation defined by the two bodies is similar to that of other structures known in the southern group of the Cape Verde islands, thus suggesting a possible structural control constraining the location of the plutonic intrusions.

Inverse theory, Gravity anomalies and Earth structure, Oceanic hotspots and intraplate volcanism, Crustal structure, Africa, Atlantic Ocean

1 Introduction

Within-plate ocean islands are the result of long-lived (10⁶–10⁷ yr) magmatic activity giving rise to highly complex buildings whose emerged part usually corresponds to a very small proportion (usually <5 per cent) of the total volume of the insular volcanic edifice. Geology and volcanostratigraphy give us information about the shallow distribution/structure of the different formations/lithologies based on the direct observation of the emerged outcropping rocks. However, the determination of the underground structure of a volcanic edifice requires the use of indirect geophysical methods like gravity and magnetic field modelling or seismic tomography. Despite its propensity to non-unicity problems, the gravity data required to determine a 3-D subsurface mass/density distribution is relatively easy to acquire. Thus, it is widely used to study the internal structure of ocean islands (e.g. Minshull & Brozena 1997; Malengreau 1999; Montesinos et al. 2005; Gottsmann et al. 2008; Camacho et al. 2009; Montesinos et al. 2011).

The Cape Verde archipelago has been the subject of geophysical surveys using seismic, gravimetric and magnetic techniques to study the mechanisms responsible for the Cape Verde rise (e.g. Dash et al. 1976; Ali et al. 2003; Lodge & Helffrich 2006; Pim et al. 2008; Wilson et al. 2010). The only geophysical study made with the purpose of resolving the internal structure of each individual islands was carried out in the 1960s (1961–1968) and consisted of a land gravity survey covering all islands of the archipelago (Mendes-Victor 1970). However, these readings are sparse, and only allow for a very rough interpretation of the underground structures. Despite the existence of a complete report of Mendes-Victor's campaigns, it was not possible to accurately identify and locate the stations measured on Maio Island due to a lack of precision in the coordinate records. Hence, data sparseness, in addition to the lack of location accuracy, has limited its usefulness for more detailed studies. Thus, a new gravity survey was performed in 2008 September covering almost the entire surface of Maio Island. The impossibility to find common references to accurately perform the adjustments needed for compatibility, made it unfeasible to conjugate the old and the new gravity data sets. The new data set was used to calculate a 3-D density contrast model of the subsurface structure of the island of Maio by means of an inversion methodology (Camacho et al. 2002) that has already been successfully applied elsewhere in similar volcanic contexts (e.g. Montesinos et al. 2003; Nunes et al. 2006; Gottsmann et al. 2008).

2 Geological Setting

2.1 The Cape Verde archipelago

The Cape Verde archipelago (14°48′–17°12′N; 22°44′–25°28′W) is located in the Atlantic Ocean, some 600–900 km offshore the coast of Senegal, West Africa (Fig. 1). Magnetic anomaly and bathymetric data suggest that the Cape Verde archipelago was built on Early Cretaceous (120–140 Ma) oceanic crust, unusually shallow for its age (Williams et al. 1990; Müller et al. 2008). Indeed, the islands are located astride the ≈2.2 km high and 1400–1600 km wide Cape Verde mid-plate topographic swell (Cape Verde Rise), the largest within-plate bathymetric anomaly in ocean basins (e.g. Lodge & Helffrich 2006). The archipelago comprises 10 islands and several islets forming three lineaments: (1) a WNW–ESE trending alignment formed by the islands of Santo Antão, São Vicente, Santa Luzia and São Nicolau; (2) an almost N–S alignment that includes Sal, Boavista and Maio and (3) a WSW–ENE trending alignment encompassing Brava, Fogo, Santiago and Maio.

Figure 1

(a) Geographical location of the Cape Verde Archipelago; (b) location of the island of Maio in the archipelago and (c) simplified geological map compiled from Serralheiro (1970) and Stillman et al. (1982) indicating the toponyms referred to in the text. Ages from Mitchell et al. (1983).

The Cape Verde Rise is associated with pronounced geoid (7.6 ± 0.3 m), gravimetric and heat flow (16 ± 4 mWm⁻²) anomalies (e.g. Dash et al. 1976; Courtney & White 1986) compatible with those believed to be originated by the impingement of mantle plumes on lithospheric plates (e.g. Sleep 1990). This is consistent with a high geoid height to topography ratio (4.4 ± 0.23) suggesting the re-heating of the lower lithosphere (Grevemeyer 1999). Also, suggesting the action of a mantle plume in the formation of Cape Verde is the flexural moats flanking the islands, which are less pronounced than would be expected from surface volcanic loading. This implies subsurface, upward-acting, pressures counteracting the surface downward loading (Ali et al. 2003).

The theory that Cape Verde originated from a deeply rooted mantle plume seems to be supported by seismic tomography studies (Montelli et al. 2006; Zhao 2007) and by unradiogenic He isotope signatures obtained both from silicate and carbonatitic rocks (R/Ra up to 15.7 and 15.5, respectively; Doucelance et al. 2003; Mata et al. 2010; Mourão et al. 2012). However, this is still a debatable question given that a recent analysis of P-to-S receiver functions showed that the time separation between the 410 and 660 km discontinuities does not change beneath Cape Verde as would be expected in the presence of a mantle plume (Helffrich et al. 2010), whereas Vinnik et al. (2012) consider that at the Cape Verde region the transition zone is up to ≈30 km thinner than the ambient mantle.

The crustal thickness is anomalously large beneath the islands (up to 22 km at Maio Island). This is attributed to magmatic transfer from the mantle source(s) to crustal levels (Lodge & Helffrich 2006). However, between the islands, the crust is only ≈7 km thick, a normal value for oceanic settings (Ali et al. 2003; Pim et al. 2008). Beneath the islands, the shallowest levels of the mantle are characterized by P-wave velocities ranging from 8.1 ± 0.5 to 8.6 ± 0.3 km s^–1 (8.4 ± 0.4 km s^–1 for Maio) which overlies a zone of low shear wave velocity starting at depth ≈80 km (Lodge & Helffrich 2006). Still, it should be noted that seismic tomography suggests that part of the Cape Verde archipelago is underlain by a shallow (<100 km deep) high S-wave velocity (V_S) region (Begg et al. 2009) interpreted as a fragment of ancient subcontinental lithosphere left stranded in the oceanic mantle during the opening of the Atlantic Ocean (Coltorti et al. 2010). This hypothesis is supported by Os isotope analyses of sulphide grains in xenoliths (Coltorti et al. 2010) and by the geochemistry of lavas from some of the islands (Gerlach et al. 1988; Doucelance et al. 2003; Martins et al. 2010).

For the Cape Verde hotspot, the most ancient age (25.6 Ma) was obtained in submarine basalts from the basal complex of the Island of Sal (Torres et al. 2002). The youngest eruption occurred on Fogo Island in 1995 (Madeira et al. 1997; Torres et al. 1997). However, the evidences for K–Ar age resetting at 60–40 Ma for Mid-Ocean Ridge Basalts (MORB) pillow lavas cropping out at Maio (Mitchell et al. 1983), led to the suggestion that the Cape Verde plume-related magmatism may have started some 25 Ma before the genesis of the first emerged testimonies (Gerlach et al. 1988). An ageing pattern can be found in the WSW–ENE island alignment from the west (Brava; Madeira et al. 2010) to the east (Maio), through Fogo (e.g. Madeira & Brum da Silveira 2005) and Santiago (e.g. Barker et al. 2009) consistent with a northeastward movement of the oceanic plate over a stationary plume.

2.2 The geology of the island of Maio

Maio, one of the smallest and oldest islands of the archipelago, is 25 km long, 15 km wide and has a surface of 269 km². It has been volcanically inactive for the last ≈7 Ma (Mitchell et al. 1983). Consequently, its morphology is strongly dominated by the effects of exogenous processes, and it is clearly distinct from that of the younger islands in the archipelago (e.g. Serralheiro 1970; Holm et al. 2008; Madeira et al. 2008;). Mostly due to marine abrasion, Maio presents a flat morphology, surrounding a few residual hills that rise to a maximum altitude of 437 m (Monte Penoso, see Fig. 1c).

A simplified geological map of Maio is shown in Fig. 1(c). The island is formed by a basement of uplifted seafloor, which comprise normal-type MORB pillow lavas (de Paepe et al. 1974; Gerlach et al. 1988) and breccias (Batalha Formation, hereafter Fm.) underlying a Lower Cretaceous sedimentary suite of limestones (Morro Fm.), shales (Carqueijo Fm.) and volcaniclastic breccias (Coruja Fm.) (e.g. Serralheiro 1970; Stillman et al. 1982; Azéma et al. 1990). This sequence is intruded by alkaline plutonic bodies of essexite/pyroxenite (and minor nepheline syenite) forming a Central Igneous Complex (CIC), dating from before 20 Ma (Mitchell et al. 1983). The seafloor sequence was intensely deformed by the plutonic intrusions. This deformation is particularly evident in the stratified sequence of deep marine sediments which are folded and faulted. Submarine basement and plutonic intrusions are crossed by a dense dyke (sensu lato) swarm of alkali lamprophyric nature which attains a frequency of almost 100 per cent in the central area of CIC (Furnes & Stillman 1987). K–Ar ages for these dykes range from 15.4 ± 0.3 Ma to 8.1 ± 0.6 Ma (Mitchell et al. 1983), with a major period of dyke intrusion at around 11 Ma. A Mid- to Late-Miocene (∼15 7 Ma; Mitchell et al. 1983) sequence of hyaloclastites and pillow lavas (Casas Velhas Fm.), subaerial lavas (Malhada Pedra and Monte Penoso Fms.) and related volcaniclastic sediments (Pedro Vaz Fm.) rests unconformably on an erosion surface cut on the previous units. The magma chambers feeding the younger volcanic events on the island are undetermined, raising the question of whether there is a hidden plutonic centre beneath northern Maio (Mitchell et al. 1983). This question will be addressed below (see 5. Discussion).

All sequences are partially covered by a staircase of Quaternary beach deposits, raised up to 70 m above present sea level (a.p.s.l.), and Holocene subaerial sediments (consolidated and active dunes, sabkha salt flats and alluvial fans). The raised seafloor sequence cropping out above present sea level, the uplifted Miocene pillow lavas and hyaloclastites, and the Quaternary beach deposits all indicate that the island underwent a significant uplift. At Cape Verde archipelago, important vertical movements are also testified by the occurrence of MORB lavas in Santiago Island (de Paepe et al. 1974; Gerlach et al. 1988; Millet et al. 2008), by the occurrence on several islands of uplifted submarine alkaline lavas (up to 450 m a.p.s.l. in Santiago Island and 400 m a.p.s.l. in Brava Island), and raised Quaternary beaches (Serralheiro 1976; Zazo et al. 2007; Holm et al. 2008; Madeira et al. 2008; Madeira et al. 2010; Ramalho et al. 2010a,b). Although, all evidence point to very important vertical movements at the scale of the archipelago, it has been shown that the amount of uplift and the uplift rates vary significantly from island to island (Madeira et al. 2010; Ramalho et al. 2010c).

3 Gravity and Height Data

The gravity measurements were made in 2008 using a Lacost Romberg 1052-G gravimeter (Micro-g LaCoste, Lafayette, CO, USA) with an electronic reading system. A total of 144 stations were measured covering almost the entire surface of the island, distributed as homogeneously as possible with an average spacing of 1000 m (Figs 2b and 3b). Only the northern extremity of the island was left uncovered due to inaccessibility during rainy season. Most of the highest points (Monte Santo António, Monte Batalha, Monte Vermelho and Monte Penoso, see Figs 1c and 2b) are almost inaccessible and were also not measured. The adjustment of gravity data, corrected by drift, jumps and tides, produced residuals with a standard deviation of 0.10 mGal. All gravity values refer to the ISGN71.

Figure 2

(a) Bathymetry from altimetric satellite data collected from the Smith & Sandwell (1997) seafloor model. The black circle indicates the data used for topographic correction (80 km radius from the centre of the island of Maio); (b) topography of the island of Maio, constructed from the 1:10 000 topographic cartography, with a planimetric resolution of 20 m and an estimated height accuracy of 2 m. The dots represent the location of the gravity stations.

Figure 3

Bouguer anomaly map: (a) from satellite data collected from Andersen et al. (2010). The black circle indicates the data used for crustal density determination (80 km radius from the island of Maio); (b) from the land survey data. The dots represent the location of the gravity stations.

The geodetic coordinates of the stations were determined by differential GPS observations and by RTK (real time kinematics) technique. All the obtained positions have a mean precision greater than 5 cm. The ellipsoidal heights were converted to orthometric heights using the EGM08 global geopotential model, and a local station was used to calibrate the EGM08 model. The 144 estimated gravity values and corresponding geodetic positions and orthometric heights were used to compute the gravity anomalies. A free-air gradient of −0.3086 mGal m^–1 and the normal gravity referred to the GRS80 (Moritz 1980a) were considered.

The digital terrain model depicted in Fig. 2(b) was constructed from the 1:10 000 topographic map. This model has a planimetric resolution of 20 m and an estimated height accuracy of 2 m, which was calculated from the orthometric heights of the stations. Although the island is almost flat (the maximum height is 437 m and more than 70 per cent of its surface is below 80 m a.s.l.), the terrain correction was computed, mainly because of the surrounding marine areas, where the ocean floor reaches depths of −4000 m. Two terrain data sets were used; the digital terrain model covering the land area, and another, covering the marine area, with grid resolution of 1', collected from Smith & Sandwell (1997) seafloor global model.

The determination of crustal density in the island of Maio using the land gravity signal is not an easy task, as there are no significant short wavelength gravity anomalies and the large positive gravity anomaly that dominates the gravity field is clearly overlapping the area where the terrain is more elevated. Thus, calculations were made over a larger region, considering a radius of 80 km from the centre of the island and using DNSC08 global gravity model (Andersen et al. 2010). The Bouguer gravity anomaly was calculated using several densities. The density value of 2.52 g cm^–3 was chosen as it was the one that resulted in a lower correlation between Bouguer gravity anomaly and topography. This value was considered as representing the average density for the rocks of Maio Island. Indeed, density values measured on samples collected on the island vary between 2.77 and 3.25 g cm^–3 for the essexites and pyroxenites of the CIC, 2.64 g cm^–3 for the Lower Cretaceous limestones of Morro Fm, and 2.10–2.71 g cm^–3 ( = 2.36 g cm^–3) for Quaternary raised beach calcarenites, and lagoonal sandstones/mudstones.

The mean density value of 2.52 g cm^–3 is comparable to the values which have been used for Cape Verde and other volcanic islands. A value of 2.4 g cm^–3 was used by Dash et al. (1976) as mean rock density of island edifices for the archipelago. On the other hand, Ali et al. (2003) found that a value of 2.7 g cm^–3 gave the best results for flexure modelling of the crust beneath the island loads. Pim et al. (2008) referred to the findings by Watts et al. (1997) regarding the Canary Islands and considered a density of 2.6 g cm^–3 for Cape Verde. This value was also used by Kauahikaua et al. (2000) for the island of Hawaii, and the density studies performed on that island by Moore (2001) do not contradict this choice. All of these considerations point to an average density value of about 2.52 g cm^–3. Thus, lacking a more definite assessment of the mean crustal density, the value of 2.52 g cm^–3 was used to carry out the Bouguer and terrain corrections. The result is the complete Bouguer gravity anomaly map depicted in Fig. 3.

The Bouguer gravity anomaly map is dominated by an elliptic positive anomaly, oriented NW–SE and located at the mid-southern area of the island, where the CIC crops out (Fig. 1c). It reaches a maximum of 60 mGal and no other Bouguer gravity anomalies are visible that would indicate the presence of other anomalous structures. Ship-based marine gravity data sets around the island of Maio, which would provide more information about the gravity anomalies offshore, are not available. However, the satellite Bouguer gravity data shown in Fig. 3(a) suggests that this high Bouguer gravity anomaly is confined onshore. Taking into account that the resolution of the satellite gravity data is worse than that of the land gravity data, it was considered that the use of satellite gravity data would not change the geometry of Maio's structural model onshore and it was decided to use only the land data for this study.

To estimate the quality of the data, we assumed that the gravity signal and the noise correspond to the correlated and uncorrelated parts of the data, respectively. So, the Bouguer gravity data was analysed applying the following iterative approach. Using the autocorrelation analysis of the data and calculating a suitable analytical covariance function (e.g. Catalão 2006), the usual formulae for least-squares prediction (Moritz 1980b) provides us with a first estimation of the signal at this level of covariance and the corresponding prediction error matrix. A residual signal is obtained by subtracting the estimated signal from the observed data. This process is repeated to detect, if possible, correlated signals of minor amplitude (levels). This iterative procedure ends when no further signal level can be detected. By eliminating the sum of the signal levels collected from observed data, the distribution of the resulting residual values displays the non-correlatable noise (e.g. Montesinos et al. 2005). The final residual of the estimated signal with respect to the observed data is the non-correlated noise. The standard deviation of the residual shows the quality of the data and the obtained value of 0.776 mGal (Fig. 4) indicates that the quality of the data is adequate for this study. This noise may be related to the observation process itself, to very local anomalies, or to uncertainties in the previous corrections (such as terrain corrections using a unique density value).

Figure 4

Histogram of the residuals of the filtering process. This non-correlated noise of the data follows a normal distribution of mean 0.026 mGal and standard deviation 0.776 mGal.

The distribution and geometry of the sources of the gravity field were modelled by a 3-D-inversion technique with the objective of studying the crustal structure of the island. It is assumed that the main Bouguer gravity anomaly is associated with the CIC and, because it is confined to the onshore area, it was determined that the land data is sufficient so that the inversion procedure results in the definition of the main anomalous structure. Also, the spatial distribution of the measurements seems to sample adequately the Bouguer gravity anomaly in question. This, together with the previous noise analysis made on the data, show that the survey is suitable to perform this task.

4 Gravity Inversion Model

The gravimetric data-inversion technique used in this work aims at the determination of the geometry of the sources producing the observed gravity field, by adjusting a 3-D model composed of prismatic cells which assume a priori values of density contrast (positive and negative). Similar inversion model has been applied and published in several previous papers (e.g. Camacho et al. 2002; Montesinos et al. 2003).

This gravity inversion method looks for an anomalous sources model formed by the 3-D aggregation of M parallelepiped cells which are filled, in a growth process, by the prescribed positive, , and/or negative, , density contrasts. The non-uniqueness problem, inherent to the inversion of potential field data, is addressed by applying two fitness considerations; the minimization of the l2 adjustment of calculated and observed gravity data and the smoothness of the model. This is carried out through a controlled 3-D growth of the anomalous bodies by means of an exploratory approach. The process starts with an empty model made of cells with density contrast set to 0. For each step of the growth process there is one cell that changes its property, and by the kth step, k cells have been filled with one of the prescribed density contrasts originating a certain gravity response. At step (k+ 1), a new cell is filled by testing the prescribed density contrasts on every empty cell. A linear regional trend is calculated simultaneously to the inversion process and, for each density contrast essayed for the new cell, the residual of the model response is tested against the observed gravity anomaly, weighted by a corresponding scale factor (f_s > 1), and smoothness constraints. Finally, the combination cell/density contrast which provides the best fit is selected as the optimal change for this (k+ 1) growth step. The process continues with further expansions of the anomalous body, and a correspondent decrease of the scale factor. The algorithm stops when the value of this scale factor is very close to 1.

A suitable choice of the a priori density contrasts (positive and negative) is necessary to properly apply this inversion method. Higher density contrast values will produce an anomalous model with smaller and more compact structures than if lower contrast values are chosen. Logically, the adjusted geometry for these anomalous bodies can appear too sharp, as they are controlled by the selected density contrasts. To produce a model with a smoother geometry that can diminish the uncertainty of the inversion problem, a smoothing technique is employed (Camacho et al. 2002). This appears in the final model structures as a small gradient of density contrasts that blurs the borders of the anomalous bodies.

As there are no previous deep structural studies made on the island of Maio, the constraints that can be imposed on the inversion process are very limited. Nevertheless, because gravity data inversion is highly prone to equivalence problems, some a priori considerations were required. Due to the absence of significant minimums in the gravity field, and because the known geology of the island gives no indications otherwise, it was assumed that there were no negative contrasts. However, for mathematical purposes regarding the architecture of the inversion tool, the minimum value for the density contrast was set at −0.01 g cm^–3. As for the maximum value, there were no concrete indications concerning a preferred density contrast to use. According to Wohlenberg (1982), essexites density lies between 2.69 and 3.14 g cm^–3 with a mean value of 2.95 g cm^–3 and pyroxenites density lies between 2.93 and 3.34 g cm^–3 with a mean value of 3.22 g cm^–3, which are consistent with the values obtained from samples of Maio's CIC (see 3. Gravity and Height Data). Density measurements were also made on samples from dykes whose frequency can reach almost 100 per cent of the cropping out area of the CIC (Fig. 1c). The obtained values range between 2.69 and 3.04 g cm^–3 ( = 2.83 g cm^–3). Considering that dykes and essexite/pyroxenite rocks are representative of the material that generates the large positive Bouguer gravity anomaly and bearing in mind the 2.52 g cm^–3 set for average crustal density, this provided a starting point suggesting values around 0.2–0.5 g cm^–3 for the maximum density contrast. A qualitative analysis of models calculated using this range of values points towards a contrast of 0.5 g cm^–3. Lower values result in instabilities that are only resolved applying a too high regularization, and therefore the model obtained is not realistic.

The calculated model produces a response (Fig. 5) that adjusts quite well to the observed data in Fig. 3(b). A rms of 0.4 mGal was obtained and the larger misfits are located mainly in the border of the island and in regions where the gravity gradient is steeper.

Figure 5

Gravity response calculated from the determined density model.

5 Discussion

This study revealed the image of an anomalous density body that has a good correlation with the surface geology of Maio (Figs 6 and 7). It coincides with the CIC, mainly formed by essexites/pyroxenites/nepheline syenites and associated dense dyke swarm cropping out in the central–southeastern region of the island. The high density of the predominant essexites and pyroxenites accounts for the density contrast shown by the model.

Figure 6

Calculated density contrast model: (a) Horizontal slice at a depth of 1.5 km, the blue lines mark the limits of the Central Igneous Complex outcrops; (b) vertical slice at x = 268.5 km; (c) vertical slice at y = 1680.5 km; (d), (e) and (f) horizontal slices at depths 2, 3 and 4 km, respectively. The blue dotted lines represent a rough outline of two possibly individual bodies.

Figure 7

3-D representation of the calculated anomalous density body.

The coarse granularity (clinopyroxene and kaersutite crystals up to 1 cm long) of essexite/pyroxenite/nepheline syenite rocks indicates that they crystallized under a low cooling rate regime, typical of intrusive rocks. Magmatic intrusions can be subdivided in tabular/sheet-like and non-tabular bodies. The association of mafic, ultramafic and felsic rocks in the same intrusive body is usually interpreted as the result of magmatic evolution processes (e.g. gravitational segregation-assisted fractional crystallization) affecting a significant magma volume. As examples, the tabular intrusion of Bushveld (South Africa) and the Skaergaard pluton (Greenland) can be referred.

The obtained 3-D model shows a large volume (≈800 km³) of higher density related to the positive Bouguer gravity anomaly that dominates the survey. Its estimated mass is about 0.38 × 10¹⁵ kg. As shown in Fig. 6, the centre of mass is located at a depth around 4500 m, and the anomalous body probably extends below −9000 m. It is characterized by steep sides and by aspect ratios of thickness/length clearly higher than those typical of tabular intrusions, that is higher than 10⁻²–10⁻⁴. These characteristics, the outcropping area (≈20 km²) and the volume of the magmatic body allow for its classification as stock(s). The non-tabular character of the intrusion is also endorsed by the strong deformation induced on the country rocks.

As revealed from the shape of the Bouguer gravity anomaly, the anomalous body is oriented approximately NW–SE (Fig. 6). It has a regular smooth ellipsoidal shape at depths below −3000 m. At this depth there are indications suggesting the existence of at least two distinct, but coalescing, anomalous bodies that jointly form the large anomalous mass revealed by the model. Individually, none of these bodies show a particularly defined orientation, and the general NW–SE trend of the anomalous mass seems to be determined by the relative position of the distinct bodies. Above −3000 m, and up to the surface, several smaller-scale structures can be identified. The vertical profiles show that these shallower structures are connected to the large anomalous mass, apparently deriving from it.

Despite the fact that thermal resetting by later magmatism cannot be completely excluded, the significant range of K–Ar ages (>20–8.2 ± 0.2 Ma; see Mitchell et al. 1983), for the CIC, confer some support to the presence of multiple intrusive bodies, which we consider as remnants of magma chambers, part of the plumbing system which fed the volcanism in Maio Island.

Ocean islands are the subaerial consequence of mantle melting and magma transfer from mantle depths to the Earth surface. Magmas are less dense than their solid sources and melting residues and thus they tend to migrate upwards to a neutral buoyancy level, which favours magma storage. In the oceanic crust a crossover occurs at depths <3 km, between basaltic (s.l.) melt and crustal densities (e.g. Ryan 1994). It can be inferred that, depending on the magma composition and volatile content, the depth at which alkaline magma density equals that of the surrounding host rocks can be shallower than 600 m. Examples of such high level magma chambers have been described elsewhere (e.g. Wanless et al. 2006; Massin et al. 2011). For Maio Island, it is clear that plutons imaged by 3-D gravimetric inversion represent magma ponding at very shallow oceanic crustal levels, because their emplacement strongly deformed (folded and faulted) the abyssal Lower Cretaceous marine sediments. For these sediments, Storetvedt & Løvlie (1983) detected two axes of magnetization. One of them, striking almost N–S and concordant with the Tertiary–Quaternary palaeomagnetic data for Africa and Canary Islands, is clearly of secondary origin. The observed depth-dependent increase of the frequency of this magnetization is interpreted as the thermal/chemical effect of the magma intrusion at shallow depth under the Lower Cretaceous sediments.

The volcanic counterparts of these plutonic rocks were removed by erosion due to continuous uplift of the insular edifice. This is expressed by a major erosive discontinuity separating the deformed seafloor sequence and plutonic complex from the overlying volcanic sequences.

The 3-D model depicts a main NW–SE trending high-density mass as the source of the positive Bouguer gravity anomaly that dominates the island. The trend of this anomaly and related intrusive complex is not coincident with the N–S elongated shape of Maio, which is aligned with the easternmost islands of Cape Verde (Sal, Boavista, João Valente seamount and Maio). It may, thus, represent some kind of structural control of the intrusions making up the NW–SE elongated body.

Although there is no defined volcano-tectonic pattern expressed in the surface geology of Maio, NW–SE trending structures are known in the other islands forming the southern group of Cape Verde (Brava, Fogo and Santiago). In Brava, the most conspicuous tectonic structure is the NW–SE trending Cachaço graben defined by the Vigia and Cachaço faults (Madeira et al. 2010). The analysis of the volcano-tectonic structure of Fogo Island (Brum da Silveira et al. 1997a,b; Madeira & Brum da Silveira 2005) shows a NW–SE trending set of faults with geomorphic expression on the northwestern flank of Fogo (Galinheiros and S. Jorge scarps). The island of Santiago, on the other hand, presents a NW–SE elongated shape and a Bouguer gravity anomaly with similar direction (Mendes-Victor 1970).

These structures, together with the Maio Bouguer gravity anomaly, may suggest a regional structural control of volcanism in the southern group of islands, whose location is probably constrained by a set of ‘en-echelon’ megatension cracks, compatible with a dextral shear zone. This could represent the presence of a major discontinuity in the oceanic crust beneath these islands, probably an old transform fault zone. Maio would lie in the intersection of such a structure with the N–S alignment of the older eastern islands, whose nature is unclear.

The gravity field obtained from satellite data covering the marine area surrounding the island shows that the anomalous mass does not extend beyond the shoreline of the island (Fig. 3a). However, to the north of Maio, there is another Bouguer gravity anomaly that has its maximum offshore, and which appears to reach the northern shore of the island. Also, the calculated density model shows a high-density body which cannot be well defined due to the lack of data in that area. This is in agreement with the probable existence of another volcanic centre to the north as suggested by Mitchell et al. (1983), which could be the magma source of the younger extrusive sequences in Maio. The analysis of this Bouguer gravity anomaly will only be possible if more detailed data is obtained both on land and at sea.

6 Conclusions

A gravity survey covering almost the entire surface of the island of Maio was performed. The resulting gravity Bouguer anomaly field shows a high positive gravity anomaly oriented NW–SE and located at the mid-southern area of the island, where the CIC crops out. Satellite data suggests that this anomaly is confined to this islands edifice.

The data-inversion procedure confirms the existence of a high-density body that is interpreted as remnants of magma chambers which fed volcanism not preserved nowadays in Maio Island. The configuration of this anomalous density volume suggests that it is formed by at least two coalescent bodies, suggesting the occurrence of two volcanic phases before the volcanic sequences that presently crop out in Maio Island.

The NW–SE orientation defined by these two bodies is similar to that of other structures known in the southern group of the Cape Verde islands. This suggests that the location of the intrusions is possibly constrained by a structural control (‘en-echelon’ cracks related with a dextral shear zone?).

Acknowledgments

This work is mainly a contribution from FCT project CV-Plume, PTDC/CTE-GIN/64330/2006, funded by Fundação para a Ciência e Tecnologia, Portugal. It was also partially supported by Pest-OE/CTE/LA0019/2011 (IDL) and Pest-OE/CTE/UI0263/2011 (CeGUL).