1 Scientific and programmatic context of the cruise

The COLMEIA (COLd Mantle Exhumation and Intra-transform Accretion) project aimed to study in detail the temporal evolution of the complex plate boundary of St. Paul, and the origin of the peridotite massif of St. -Pierre-St. Paul, in the Equatorial Atlantic. It mapped with bathymetry, magnetism, gravimetry and seismic much of the St. Paul intra-transform accretionary system. Dredging and sampling of the water column were also carried out in order to map the extension of the outcrops of peridotites, basalts and gabbros in the area, to study the magmatic processes, to characterize the mantle and to identify possible hydrothermal vents. The project was designed in close collaboration with Brazilian colleagues, the ridge axis being located in the Brazilian EEZ. The cruise took place in January/February 2013. A second cruise, COLMEIA-HYDRO, took place aboard the NPo Araguari of the Brazilian Navy in May 2014 to recover the moorings of the hydrophones.

Figure 1: Ship tracks of the cruise, showing the acquired bathymetry, as well as the dredging sites, CTDs and hydrophone moorings.

2 Cruise objectives

This part of the Atlantic has been recognized for many years as a "cold" mantle region, i.e., with low basalt production and low rates of partial melting (e.g. Schilling et al., 1995; Bonatti et al. 1993). The "abnormal" zone would encompass all the major equatorial fracture zones, including the Saint Paul system, which is characterized by the presence of a large peridotite ridge reaching sea level (islets of St. Pierre and St. Paul, formed by mylonitized and serpentinized peridotites). The islets were studied by Brazilian teams, which revealed an uplift of about 1 mm/year (Campos et al., 2010). The work of Schilling and Bonatti suggested that the low melting rates resulting from the analysis of the basalts sampled at the axis and the large quantity of peridotites present in the area would not be the result solely of the thermal effect of the large offsets of the axis of the ridge, but also linked to the presence of an "abnormal" mantle. Analysis of the Rh-Os system on peridotite samples obtained from the flanks of the St. Pierre-St. Paul would suggest that the anomalous mantle results from the presence of residual fragments of lithosphere from the ancient Japetus ocean, which have entered into subduction (Sichel et al., 2008). A recent study on the petrology of the same samples (Brunelli and Seyler, 2010) seems to support this model. However, a model using fragments of delaminated subcontinental lithosphere incorporated into the mantle cannot be ruled out.

The formation of the oceanic lithosphere at slow spreading ridges is a highly variable process in both time and space, with portions dominated by magmatism and others where deep rocks are exposed. The portions of ridge where these rocks outcrop are generally associated with axial discontinuities, although not necessarily transform faults (eg Dannowski et al., 2010). Despite the presence of peridotites on the flanks of these structures, called "core complexes", IODP drillings have shown that their core is composed of gabbros. The model of Dannowski et al (2010) offers an explanation for these apparently contradictory observations, formulating a hypothesis of spatial variability in the structure of these bodies linked to the presence or absence of magma at depth. The "core complexes" are therefore not the result of totally amagmatic accretion, since the main volume of these structures is composed of gabbros. The presence of large outcrops of peridotites on the northern edge of the St. Paul system suggests, unlike the "core complexes", a strongly amagmatic system (linked to a cold mantle). One of the questions was therefore whether the St. Pierre-St. Paul was a localized event, linked to the presence of the transform fault, or if outcrops of peridotites constituted the majority of the floor of this part of the Atlantic. In the latter case, we would have here an "end member" of the processes of slow accretion.

3 Main results

Bathymetric data and rocks dredged during the COLMEIA cruise show that the idea of ¿¿an amagmatic accretionary system do not apply to the entire Saint-Paul system. The significant variability of accretionary styles within the St. Paul system shows that the formation processes of the oceanic lithosphere are very variable and complex there, depending on the nature of the mantle, the neighboring lithosphere, and local constraints (presence of large offsets, obliquity).

The intra-transform (ITR) segments are short and narrow, with exceptionally deep axial valleys. Axial depths are below 4000 m on average, and reach 5400 m in some nodal basins. Neovolcanic ridges on the floor of the axial valley are poorly defined, and a few circular volcanoes were observed in the axial valleys. The off-axis reliefs are very variable from one segment to another. The northern segment is the most volcanic in the St. Paul system (Figure 2). It presents a West-East asymmetry, with a series of faulted OCCs and a thin gravimetric crust on the North-West flank, in position of Inside corner, and a series of abyssal hills and a normal crust on the North-East flank. The southwestern part is associated with narrow, curved, poorly defined abyssal hills and normal crust. The southeast flank is associated with thin magmatic crust, few faults and irregular morphology, but no detachments have been identified. It is therefore a weakly magmatic segment, but presenting a volcanic style which is surprising, considering its size and the length of the bounding axial offsets (~300 and 190 km to the north and to the south).

Figure 2. Bathymetry and structural interpretation of the north segment (top) and associated crustal thickness derived from gravity (bottom).

The central segment (Figure 3) is the least magmatic of the system (Bebin, M2 2018). It presents oblique peridotite hills on the two flanks of the axial valley, associated with a thin crust. Dredging confirms the presence of abundant peridotites on these structures and the almost absence of gabbros. On both flanks, further away from the axis, an area of thin crust is associated with areas of flat topography. This morphology suggests a "smooth sea floor" spreading style, with a very low melting rate, a near-absent magmatic crust, a strong obliquity of the structures and alternating detachment faults between the two flanks of the ridge. To the south, this segment is very asymmetric with a long and narrow OCC on the southwest flank, bordered by oblique faults near the axis, associated with slightly higher than average crustal thicknesses. This detachment is old and currently inactive. The southeast flank shows normal thickness crust but very few faults and no abyssal hills. This type of functioning had not yet been described for a slow ridge, where the less magmatic styles correspond to OCCs and the presence of gabbro at depth.

Figure 3. Bathymetry and structural interpretation of the central segment (top) and associated crustal thickness derived from gravity (bottom).

The southern segment is strongly asymmetrical. It suggests spreading through asymmetric detachment faults for about five million years or even longer, with a succession of detachment surfaces, some intensely faulted and deformed (Vincent, 2016; Vincent et al., 2023). Each detachment spans the entire length of the segment. The asymmetric accretion is compensated by a series of small relocations of the accretion axis at the level of the old detachment surface, with a minor and relatively short-lived magmatic phase and formation of a new detachment fault (Figs. 4 and 5). Spreading here is tectonically dominated, as expected, with very little magmatism and no well-defined abyssal hills. We have highlighted a gradual reduction in magmatic production at the axis, with a shift from an operating mode in Oceanic Core Complex with a gabbroic core to hills of peridotite, similar to those observed in the South-West Indian ridge. The recent phase is characterized by detachments on both flanks of the ridge, highlighting the onset of flip-flop functioning (Vincent et al., 2023).

Figure 4. Bathymetry and structural interpretation of the south segment (top) and associated crustal thickness derived from gravity (bottom).

Figure 5. Main structural features of the south segment. Each detachment surface is indicated by a color code.

A major result is the demonstration of compressive stresses over a large area, causing the uplift of the Saint Paul massif. Intense tectonic deformation is evidenced by the significant presence of mylonites of ultramafic and mafic origins, obtained by dredging over a very wide shear and compression zone (Figure 6). Reverse faults have been imaged by seismic and gravity models confirm the presence of a broad layer of low densities, reaching in places 12 km thick, interpreted as being partly serpentinized mantle. At the heart of this transpressive system, gravimetry reveals a zone of slightly serpentinized mantle at the level of the islets of St. Pierre and St. Paul, in agreement with the observed rocks. The transpressive stresses are linked to shifts of the transform fault in its western portion, forming a succession of step overs (Figure 6). These shifts are related to the propagation of the segment of the Mid-Atlantic Ridge located north of St. Paul into the lithosphere of the transform system, an area where the age contrast reaches 20 million years (Figure 7). The spread was triggered by a kinematic shift of about 5 degrees ten million years ago and sustained by particularly volcanic and robust accretion (Maia et al., 2016). This particular behavior is due to the presence of a fertile mantle north of St. Paul, due to the presence of the Sierra Leone thermo-chemical anomaly. This same mantle feeds the northern intra-transformant segment, which explains its very volcanic morphology. This pattern has been confirmed by the analysis of basalts sampled during the cruise and persists for at least ten million years.

Figure 6. Bathymetry and petrology of the samples recovered on Atoba ridge and on the north intra-transform segment.

Figure 7. Sketch of the evolution ont the north transform fault of the St. Paul system, showing the different steps of the formation of Atoba ridge.

Prigent (M1, M2 2016, 2017) highlighted the distinction between the three compositional domains of the mantle source: the northern ITR samples a mantle with traces of the Sierra Leone hot spot and it was even more so a few millions of years ago, as revealed by samples from abyssal hills (Figure 8). The slope on the heavy REEs indicates garnet at the source, thus a partial melting in a deeper mantle, as for a hot spot. The southern segment resembles the MAR south of St. Paul. The central ITR is distinct from the other two. It is intermediate in traces but depleted in isotopes. So, there is a recent addition in the upper mantle which has enriched the trace elements but does not yet translate into the isotopes. The St. Paul system therefore does not lie over a zone of cold mantle, but rather at the transition between a fertile mantle domain to the north and a "cold" mantle to the south. The nature of the underlying mantle, enriched under the northern segment and more depleted under the central and southern segments, explains to the first order the variations in the style of accretion. The central segment presents, in addition, a probable cumulative influence of the two large transform offsets which border it. This results in a particularly cold segment (Maia et al., in preparation)

Figure 8. Isotopic composition of axial basalts from St. Paul's ITR segments and the MAR south and north of the system.

The analysis of mylonites and ultramyonites, sampled along the Atoba ridge shows that the deformation linked to the formation of this structure was acquired partly at great depths and high temperatures (Bickert et al., in review). The deduced depths are in excellent agreement with the thickness of the low-density layer derived from gravity and with the maximum depth of earthquakes (work in progress). This work has defined new deformation laws for mantle ultramylonites at high temperatures and shows how pyroxenes and olivines share a common deformation mechanism and crystal lattice orientation at high strain rates and temperatures in presence of fluids. This work also made it possible to define the temperature profile and to establish the mechanical stresses on the deformation along a transform domain crossing the entire lithospheric section.

Hydroacoustic data from the COLMEIA network (3 operational instruments out of the 5 deployed) were used by marine biologists (teams from the Chisé observatory, France, the Museum of Natural History in Rio and the UFRGN, Natal ) for studying the migration patterns of large whales. The results show the presence of the Antarctic blue whale in the Equatorial Atlantic (Samaran et al., 2019) as well as the Omura whale (Moreira et al., 2020). These two identifications constitute a "first" for these two types of large whales in the Equatorial Atlantic region. These data were also used in a study on the distribution of seismic velocities in the Equatorial Atlantic (Sampaio de Melo et al., 2020). The small number of instruments did not allow us to use these data for studies of seismicity. However, we are working on the recordings of the seismological station of the archipelago in collaboration with the UN. Fed. Northern Rio Grande (Brazil) and SCRIPPS. A work, recently submitted (de Melo et al., submitted) deals with the cyclicity of large earthquakes in the St. Paul system and shows the influence of the geometry of the system and the length of the transform faults in the duration of the seismic cycles. Another work on the depth of earthquakes is being finalized.

The seismic profiles acquired during COLMEIA are in the final processing phase, in collaboration with M. Ligi (CNR Bologna, Italy). Lines across the Atoba mylonite ridge show thrust along the southern edge of the structure, with faulting and the associated positive flower structure. In some lines, the deformation of the oceanic crust of the northern segment, caught in the compressive system, appears as a succession of faulted and deformed blocks, a pattern which is also visible in the structural interpretation of the bathymetry. Evidence of recent extension is visible in the southern part of the system (chapter in final phase of writing).

We have recently finished the work on the temporal evolution of the St Paul system (coll. with the CNR Bologna and Rome). After the onset of the transtensive phase (about 40 Ma) the system evolved rapidly, with the establishment of small "pull-apart" basins within the transforming domain, which then evolved into small spreading segments( Figure 9). The configuration remained unstable for a long period of time (about 25 million years) and during this phase, we observe propagations and axial displacements, with the birth and disappearance of non-transform discontinuities inside the system. The current configuration with three segments separated by well-defined transform faults, was acquired about 15 Myr ago (Maia, 2019, chapter in final phase of writing). It is difficult to place exact ages, because the system is close to the magnetic equator, which makes it difficult to use magnetism. The results of the thermal calculations show that in the periods of transtension, the temperatures inside the transform system increase slightly and it is probable that the conjunction of the extension, of a fertile mantle and this slight increase in the temperature of the mantle allow the start of accretion, even in the context of a slow ridge (chapter in the final phase of writing coll. Brest-Bologne-Rome).

Figure 9. Basement topography, after correction for the sediment thickness. The magenta lines show the location of the spreading axes. Gray lines show transform faults and white lines follow fracture zones and traces of spatial discontinuities in the past.