Widespread volcanism Southeast of Futuna Island (SW Pacific Ocean): Near-seafloor magnetic dating and regional consequences

formatting this that, during the process, may be discovered which could affect the content, and all legal disclaimers that to the pertain. ABSTRACT Near-seafloor bathymetric and magnetic data have been collected by Autonomous Underwater Vehicle (AUV) and manned submersible (DSS) over a volcanically active area southeast of Futuna Island, French Territory of Wallis-and-Futuna, in the Southwestern Pacific Ocean. Located at the edge of the Lau and North Fiji basins, at the convergence of the Pacific and Australian plates, this area is characterized by intense volcanic and tectonic activity. Direct observation by submersible reveals that the seafloor is covered by recent lava flows, volcanoes and large, active or inactive calderas filled by smooth lava flows and eventually hosting hydrothermal sites. We investigate the volcanic chronology by combining a Bayesian inversion of the AUV-gridded magnetic data with an inversion of the submersible data specifically designed to estimate the rock magnetic polarity and absolute magnetization. We show that some volcanoes predate the last (Brunhes-Matuyama) magnetic polarity reversal 780 kyr ago whereas their neighbors exhibit a normal polarity and

appear to be recent. This result suggests that the seafloor in this region has undergone continuous and sparse volcanic activity over the last few million years.

1) Introduction
The most active volcanic area in the world, known as the Pacific Ring of Fire, encompasses the Pacific Ocean and marks the limits of the Pacific Plate. To the South-West, this plate subducts beneath the Australian plate and forms the Tonga Trench. West of the trench, the Kermadec volcanic arc and the Lau back-arc basin are made of the several inferred small plates (e.g., Bird, 2003). Near its northern end, the Tonga Trench winds Westwards and connects to a complex system of strike-slip faults and volcanic ridge segments within the Northern Lau Basin (Pelletier et al., 2001;Wright et al., 2000). Southeast of Futuna Island lies a volcanically complex area characterized by relatively shallow depths and intense volcanic and hydrothermal activity (e.g., Fouquet et al., 1991;Fouquet et al., 1993;Baker et al., 2005, Labanieh et al., 2011Lupton et al., 2015, Sleeper andMartinez, 2016). This area does not present the morphology of a spreading center but defines a roughly 100x100 km 2 ( Fig. 1) diffuse volcanic province instead (Pelletier et al., 2001). We use a combination of regional and high-resolution bathymetric and magnetic data respectively collected at the sea surface by R/V L'Atalante and near seafloor by Autonomous Underwater Vehicle (AUV) Idef-X or Deep-Sea manned Submersible (DSS) Nautile to investigate some major volcanic edifices and unravel the chronology of their emplacement. Combining a Bayesian inversion of the AUVgridded magnetic data (Honsho et al., 2012) with an inversion of the submersible data specifically designed to estimate the rock magnetic polarity and absolute magnetization (Szitkar et al., 2015a), we show that the volcanic activity started before the last magnetic J o u r n a l P r e -p r o o f polarity reversal (780 ka) and has been continuous ever since, with no signs of waning. Such "intraplate" (i.e. off the well-defined plate boundaries) volcanic activity may result from melting of the underlying Indian ocean-type mantle and an additional input driven by the Samoa hotspot (Jackson et al., 2010;Jenner et al., 2012). Such intense volcanic activity likely plays an important role in remodeling the small plates constituting the Northern Lau Basin.

2) Geological context
Futuna island is part of the French overseas territory of Wallis-and-Futuna, in the South Pacific, roughly halfway between Samoa and Fiji. Wallis and Futuna islands are separated by a fossil subduction zone, the Vitiaz Trench, where water depths largely exceed 4500 m, in the direct continuation of the Tonga Trench (Brocher et al., 1985). This fossil subduction is a key element to the geological evolution of the Southwestern Pacific, as it delineates the border between the Cretaceous Pacific Plate to the North and the Cenozoic North-Fiji and Lau Basins to the South (e.g., Pelletier et al., 1998).
Previous research carried out on Futuna island has revealed the existence of volcanic structures dated Upper Pliocene and made of pillow lavas, breccias, hyaloclastites and massive lava flows (Grzesczyk et al., 1987(Grzesczyk et al., , 1988(Grzesczyk et al., , 1991a. A switch in the volcanic activity at the end of the Pliocene is seen as a marker of a tectonic change from a subducting (Vitiaz-Tonga subduction) to a transform regime (North-Fiji transform fault).
In March 2000, cruise ALAUFI (Pelletier et al., 2001) discovered the Futuna Ridge, a spreading center extending over a distance of more than 200 km from the North of the Fiji platform to the Northwest of Futuna Island. It also revealed a 30 km-wide area with a WSW-ENE orientation between 176°30' and 177°25'W centered at 14°45'S associated with a J o u r n a l P r e -p r o o f shallower bathymetry and many volcanoes (Fig. 1). This region, now known as the South-East Futuna Volcanic Zone (SEFVZ), is defined by bathymetric lineations with a 80 -90° orientation in its northern part and 50° in the Southwest, and is associated with a strong volcanic activity (Pelletier et al., 2001). Although the SEFVZ may be interpreted as another active spreading center in the northern Lau Basin, the lack of a clear topographic expression may also suggest diffuse magmatic activity over a wide area (Pelletier et al., 2001). While the Futuna Ridge exhibits a series of magnetic chrons 2 to 1 (Pelletier et al., 2001), the SEFVZ lacks any recognizable seafloor spreading anomalies and its age remains unconstrained. As shown below, deep-sea magnetic anomaly data provide some clue on this aspect.

3) Acquisition and Processing
Two scientific cruises, FUTUNA 2010 and 2012, were carried out by R/V L'Atalante, AUV Aster X (in 2010) and Idef X (in 2012) and DSS Nautile. They collected shipboard and highresolution (AUV) bathymetry as well as shipboard and high-resolution (AUV and DSS) magnetic data. Three-component magnetometers were installed on both submersibles.
The initial processing of the raw magnetic data is comparable to that described in several other papers (Isezaki, 1986;Honsho et al., 2009;Szitkar et al., 2015b;Szitkar et al., 2017) (Supplementary Material). AUV dives were designed to survey large areas along regularlyspaced, 100 m-apart parallel profiles, and the data underwent a quick daily processing to identify geological features for DSS Nautile to dive on. Conversely, DSS Nautile explores the seafloor and carries out direct observation, geological mapping and collect rock, fluids and J o u r n a l P r e -p r o o f biological samples. Its path is therefore not suitable for gridding the magnetic data. Due to the proximity of the seafloor, these data also show shorter-wavelengths and higher amplitudes than those collected by AUV dives at higher altitudes above the seafloor (on average 70 m above the seafloor). Since the depth of sources depicted by magnetic anomalies directly depends on the altitude of the measurements, DSS Nautile data mostly depict outcropping small sources whereas AUV data reflect slightly deeper sources.
We take advantage of the peculiarities of DSS Nautile dives, including its strongly varying altitude above the seafloor, to estimate the absolute magnetization and magnetic polarity of the shallow subsurface magnetic sources (Honsho et al., 2009;Szitkar et al., 2015a Constraining magnetic polarity represents a key parameter for constraining the age of volcanic edifices beyond the freshness of the topographic expression, backscatter intensity, and direct evidence.
For the AUV data, we applied the Honsho et al. (2012) Bayesian inversion that considers the varying AUV altitude above the seafloor and preserves the full-wavelength content of the signal to compute an equivalent magnetization and a rigorous Reduced-to-the-Pole (RTP) anomaly (i.e. in the geometry of the experiment and assuming a vertical geomagnetic field; Supplementary Material) (Fig. 2,3,4,5).

4.1) Kulo Lasi Caldera
The first major target to be investigated during cruise Futuna 2010 was the Kulo Lasi Caldera (Fouquet et al., 2018;Fig. 2). This roughly circular caldera, ∼5 km in diameter, is flanked by two small volcanoes to the Northeast and to the West. The caldera floor lies at ~1500 m bsl J o u r n a l P r e -p r o o f and exhibits a smooth bathymetry, apart from a central, faulted resurgent dome ( Fig. 2A). A total of six AUV dives surveyed the whole caldera and six DSS Nautile dives investigated specific targets. The equivalent magnetization (Fig. 2B) and RTP magnetic anomaly (Fig. 2C) reveal highly magnetized areas, mostly on top of the two nearby volcanoes and at the top of the caldera walls. The magnetization is generally weaker on the caldera floor and the Northeastern part presents a lower RTP magnetic anomaly. The amplitude of RTP magnetic anomaly spans 50 000 nT.
On the caldera floor, DSS Nautile discovered abundant high and low-temperature hydrothermal activity with numerous small active and inactive sites. High CH 4 / 3 He and CH 4 /TDM ratios were reported in the hydrothermal plumes, these values being characteristic of an intense volcanic event in 2010 (Konn et al., 2016) followed by the widespread occurrence of extremely young high-temperature hydrothermal chimneys at the time of the dives, a few month later (Fouquet et al. 2018, Konn et al 2018. The sites are always associated with a negative RTP magnetic anomaly on the AUV map ( Fig. 2B, 2C), in accordance with previous studies conducted over basalt-hosted or andesite-hosted hydrothermal systems in other (e.g., Tivey et al., 1993, Szitkar et al., 2014, Szitkar et al., 2015b, Fujii et al., 2015 and similar (Caratori Tontini et al., 2012;Honsho et al., 2013) geological contexts.

4.2) Ono Caldera
The Ono Caldera is located roughly 30 km WSW from the Kulo Lasi Caldera (Fig. 3). The highresolution bathymetry reveals a uniformly smooth topography (Fig. 3A), suggesting that it has been covered by a significant accumulation of sediments. A group of N30°E faults appears on the bathymetric highs and cuts across the whole structure ( The only DSS Nautile dive devoted to this caldera roughly followed the crests (Fig. 3D). The absolute magnetization is significantly weaker on Ono (0-12 A/m) than on Kulo Lasi and its polarity is systematically reversed.

4.3) Isolated volcanoes on Fatu Kapa area
The Fatu Kapa area is located 30 km North of the Kulo Lasi Caldera and has been surveyed by 15 AUV dives. The high-resolution bathymetry at the center of the Fatu Kapa area, (Fig. 4A) reveals a wide and relatively flat volcanic dome, 6-km in diameter, cut across by a series of roughly East-West normal faults and grabens. Successive tectonic and volcanic episodes result in a network of intersecting faults which favor the circulation of hydrothermal fluids (Fig. 4A). The magnetic data from AUV and DSS Nautile result in weak RTP anomalies, equivalent magnetization (AUV data) and absolute magnetization (DSS Nautile data) over J o u r n a l P r e -p r o o f the newly discovered hydrothermal fields, as expected for basalt-hosted hydrothermal sites (Szitkar et al., 2015;Fouquet et al., 2015;Konn et al., 2018). The area is characterized by a strong backscatter (Fouquet at al., 2018, their Figure 3), and therefore fresh and young lava.
At the west and East of the main volcanic areas (Fig 3) several isolated volcanic edifices have been observed on the graben floor, two to the Southwest and a third one to the East (Fig. 4).
Only AUV data are available there, as no DSS Nautile dive has been carried out on these volcanoes. On the high-resolution bathymetry, the two Southwestern volcanoes show no evidence of faulting whereas the Eastern volcano has E-W fissures and small-offset normal faults (Fig. 4A). The three edifices display positive RTP magnetic anomalies and equivalent magnetization (Fig. 4B, 4C).

4.4) Tasi Tulo area
The Tasi Tulo area extends Northeast of the Fatu Kapa area (Fig. 5). It exhibits a number of volcanoes and has been surveyed by two AUV and two DSS Nautile dives. This area is characterized by a superposition of two types of volcanoes: six large, flat isolated edifices on one hand, and series of small conical mounds forming two volcanic ridges trending N60°E and N70°E on the other hand, both roughly aligning with the graben located further Northeast (Fig. 1) (Fig. 5). The magnetic signature of volcano C is unclear, and is probably hidden by the strong positive signature of the nearby volcanic ridge. Finally, the two ridges display a strong positive equivalent magnetization and RTP anomaly in the shallowest part of the survey (Fig. 5).
Direct observations from DSS Nautile as well as rock sample analyses indicate that the seafloor is made of recent basalt with little sediment cover. Magnetic data were only acquired during one of the two DSS Nautile dives and was mostly inconclusive. Our analysis therefore relies on the AUV data (Fig. 5).

5.1) Kulo Lasi Caldera
The Kulo Lasi Caldera presents all the evidence of an active volcano. The smooth caldera floor is a direct consequence of its resurfacing by successive lava flows. Moreover, the AUV and DSS Nautile discovered several small active basalt-hosted hydrothermal systems. These sites are all associated with a lack of magnetization and even the smaller ones are depicted on the equivalent magnetization and RTP magnetic anomaly. An alignment of hydrothermal sites has been observed to the Northeast of the caldera, resulting in the elongated negative magnetic anomaly visible in this area (Fig. 2B, 2C). The limited size of these sites also supports the hypothesis of frequent resurfacing as sites do not have time to reach larger polarities with a few anomalous zones probably related to fallen blocks. Such a signature may be representative of a recent / active caldera formed during the Brunhes normal period.

5.2) Ono Caldera
The uniform sediment accumulation, the faults cross-cutting the whole caldera, the lack of recent volcanism and active hydrothermalism (Fig. 3a) suggest that the Ono Caldera is significantly older than the Kulo Lasi Caldera.
The RTP magnetic anomaly amplitude at Kulo Lasi reaches roughly 50.000 nT whereas that at Ono barely reaches 2000 nT, suggesting that the basalt magnetization at Ono is weaker than at Kulo Lasi. This observation could be explained by the progressive alteration of older rocks by seawater and would therefore support older rocks at Ono with respect to the fresh basalt at Kulo Lasi. For both calderas, the AUV was navigated at an average altitude of 70 m above the seafloor, suggesting that the comparison is meaningful. Deep-sea sediments are weakly magnetized and their contribution to the observed magnetic anomalies is negligible (e.g., Szitkar et al., 2015b). However, the unknown thickness of sediments at Ono is not considered in the inversion to equivalent magnetization. The real top of the magnetized layer is likely significantly deeper than the bathymetry, especially on the caldera floor. In terms of the RTP magnetic anomaly, this corresponds to an upward continuation which attenuates the anomaly amplitude and more specifically its short-wavelength content.

J o u r n a l P r e -p r o o f Journal Pre-proof
Comparing the RTP magnetic anomaly amplitude is therefore hazardous as long as the sediment thickness is unknown. Conversely, the shape of the RTP magnetic anomaly and equivalent magnetization distribution is of particular interest: they both are positive on the floor and negative on the crests of the Ono Caldera, systematically opposite to those observed on the Kulo Lasi Caldera. Moreover, the magnetization polarity deduced from DSS Nautile data (Fig. 3D) is uniformly reversed. These observations concur in demonstrating that the Ono Caldera was formed during a reversed polarity interval, most likely the Matuyama Period (2.59-0.78 Ma).

5.3) Fatu Kapa area
The strong backscatter of the Fatu Kapa area, similar and in continuity to that observed around the Kulo Lasi Caldera on the shipboard data (Fouquet et al., 2018;their

5.4) Tasi Tulo area
The equivalent magnetization and RTP magnetic anomaly display strong variations among the different volcanic structures of the area. The small flat volcano A associated with a positive RTP anomaly resembling that of the three Fatu Kapa volcanoes were clearly formed during a period of normal polarity, most likely the Brunhes period (0-0.78 Ma). Such is also the case of the large flat volcanoes B and E displaying a positive magnetic ring similar to J o u r n a l P r e -p r o o f those of the Kulo Lasi Caldera and volcanoes. We cannot rule out the hypothesis that these volcanoes could be older than the Brunhes period, for instance from the Olduvai episode within the Matuyama period; however, the unfaulted aspect of these volcanoes makes this possibility very unlikely. Conversely, the large flat volcanoes D and F displaying a negative magnetic ring similar to that of the Ono Caldera were formed during a period of reverse polarity, most likely the Matuyama period (2.59-0.78 Ma). The two volcanic ridges topping the flat volcanoes exhibit a strong positive equivalent magnetization and RTP magnetic anomalies and are obviously recent, formed during the Brunhes normal polarity period. The Tasi Tulo area is therefore made of a succession and a superposition of volcanoes that are older and younger than the last magnetic polarity reversal, suggesting a continuous volcanic activity over at least the last million years, and possibly up to 2.5 million years.

5.5) Regional consequences
The magnetic analysis of volcanic edifices in four areas of the SEFVZ suggests the presence of recent and active volcanoes at Kulo Lasi, Fatu Kapa, and Tasi Tulo areas, whereas older inactive volcanoes are observed at Ono and Tasi Tulo areas. The structural expression of the active areas does not support the presence of a continuous spreading center but instead of isolated magmatic centers which tectonic connection is complex and mostly hidden by the recent lava flows (Fig. 1). The observation of two generations of volcanoes at Tasi Tulo supports episodic and unfocused magmatic events over a relatively wide (~100 km) area.
The northern North Fiji and Lau basins exhibit a variety of small plates whose definition and delimitation are still to be completed. A discontinuous belt of major volcanic edifices exists within 250 km from the fossil Vitiaz Trench, where the Pacific Plate was subducting below the Australian plate (see Ruellan and Lagabrielle, 2005, their Fig. 2) (Fig. 6). These edifices J o u r n a l P r e -p r o o f are observed between 169 and 173°E north of the Hazel-Holmes and South Pandora spreading centers (Lagabrielle et al., 1996); between 173 and 178°E at the axis and north of the South Pandora and Tripartite spreading centers (Lagabrielle et al., 1996); between 179°E and 178°W north of the Futuna and North Cikopia spreading centers (Pelletier et al., 2001); between 179 and 177°W in our study area; between 177 and 175°W at the axis and north of the North-West Lau and Niuafo'ou spreading centers (the latter also known as Rochambeau Rifts; Lupton et al., 2012); and between 175 and 174°W on the Mangatolu Triple Junction and North-East Lau Spreading Center (e.g., Lupton et al., 2012). Although the location of these edifices suggests the slow melting of the remnant Pacific slab and magma ascent within the North Fiji and Lau basins, their geochemistry is more consistent with the melting of the underlying Indian ocean-type mantle and an additional input driven by the Samoa hotspot (Jackson et al., 2010;Jenner et al., 2012). The absence of a Nb anomaly characteristic of recycled lithospheric material in subduction zones and the major and trace element contents closer to those observed in Oceanic Island Basalts (OIB) indeed favors the presence of OIB material, either from the Samoa chain (Guivel et al., 1997) or the Cook-Austral chain (Price et al., 2016).
The occurrence of episodic magmatic events likely affects the distribution and stability of the plate boundaries and diffuse volcanic zones, with new potential boundaries initiating on the most prominent zones of weakness. This in turn explains the presence of many small plates in the northern Lau Basin and offers a mechanism to maintain the tectonic complexity of this area. The northern North Fiji and Lau basins are clearly complex and unstable in terms of plate tectonics, and may represent an analog of Archean seafloor dynamics with a hotter mantle, as suggested by Lagabrielle et al. (1997).
J o u r n a l P r e -p r o o f

6) Conclusion
The northern North Fiji and Lau basins are made of a mosaic of small plates that remain difficult to precisely delineate and that lie on top of the fossil Vitiaz subduction zone.  Northeast, an elongated negative magnetic anomaly corresponds to a hydrothermally active area, in accordance with previous studies in basaltic (e.g., Tivey et al, 1993) and andesitic (Fujii et al, 2015) contexts.    High-resolution magnetic data reveal the existence of calderas older and younger than the last geomagnetic polarity reversal.

Regional
Continuous and sparse volcanic activity has occurred for the last few million years.
Volcanic activity is driven by the melting of the underlying mantle with a contribution of the nearby Samoa hotspot.