Seamounts and oceanic igneous features in the NE Atlantic: a link between plate motions and mantle dynamics

Abstract A new regional compilation of seamount-like oceanic igneous features (SOIFs) in the NE Atlantic points to three distinct oceanic areas of abundant seamount clusters. Seamounts on oceanic crust dated 54–50 Ma are formed on smooth oceanic basement, which resulted from high spreading rates and magmatic productivity enhanced by higher than usual mantle plume activity. Late Eocene–Early Miocene SOIF clusters are located close to newly formed tectonic features on rough oceanic crust in the Irminger, Iceland and Norway basins, reflecting an unstable tectonic regime prone to local readjustments of mid-ocean ridge and fracture zone segments accompanied by extra igneous activity. A SOIF population observed on Mid-Miocene–Present rough oceanic basement in the Greenland and Lofoten basins, and on conjugate Kolbeinsey Ridge flanks, coincides with an increase in spreading rate and magmatic productivity. We suggest that both tectonic/kinematic and magmatic triggers produced Mid-Miocene–Present SOIFs, but the Early Miocene westwards ridge relocation may have played a role in delaying SOIF formation south of the Jan Mayen Fracture Zone. We conclude that Iceland plume episodic activity combined with regional changes in relative plate motion led to local mid-ocean ridge readjustments, which enhanced the likelihood of seamount formation. Supplementary material: Figures detailing NE Atlantic seamounts and SOIF distribution, and the location of earthquake epicentres are available at https://doi.org/10.6084/m9.figshare.c.3459729

The NE Atlantic oceanic basins have been formed since Early Eocene times following the break-up between Eurasia and Greenland. Various types of volcanic edifices (including seamounts) were emplaced on stretched continental crust before final break-up and seafloor spreading (Jones et al. 1994;Marty et al. 1998;O'Connor et al. 2000). The formation of oceanic crust was preceded by high magmatic activity, which resulted in additional igneous material being emplaced at the base and on top of stretched continental margins (Storey et al. 2007). Seamount volcanism and the emplacement of igneous centres on oceanic crust continued after seafloor spreading was established in various basins between Greenland and Eurasia. A considerable number of volcanic edifices have been identified in the NE Atlantic, mainly on remote sensing data including bathymetry and gravity data derived from satellite altimetry (e.g. Hillier & Watts 2007;Kim & Wessel 2011;Yesson et al. 2011) (Fig. 1).
Seamount volcanism is attributed to magmatic processes connected to the formation of new ocean floor/oceanic crust (seafloor spreading), or to the modification of this crust by subsequent intra-plate volcanism. A classic example of intra-plate volcanism is the plume-related creation of linear chains of age-progressing volcanic edifices on oceanic or continental crust (e.g. Morgan 1971). Intra-plate volcanism may also be the result of local processes such as lithosphere cracking or melt extraction from heterogeneous mantle (e.g. Forsyth et al. 2006), small-scale sublithospheric convection (e.g. Ballmer et al. 2009), or shear-induced melting of low-viscosity pockets of asthenospheric mantle located along the base of the lithosphere (Conrad et al. 2010).
This study aims to evaluate the correlations between a new database of oceanic volcanic features (seamounts and other small igneous edifices), and the oceanic crust morphology and evolution as established within the international NAG-TEC project ). We will first present the current knowledge of seamount and volcanic feature distribution in the NE Atlantic. The occurrence of these volcanic features on oceanic crust of various ages and structure is subsequently described. Possible links between NE Atlantic variations in seafloor spreading, mantle dynamics and seamount formation since the Eocene is also discussed. Our results may help in understanding the spatial and temporal interplay between volcanism and tectonics in a region that has also been heavily influenced by a pulsating mantle plume since the inception of oceanic crust formation.

Regional distribution of seamounts and volcanic edifices in NE Atlantic oceanic basins
According to the International Hydrographic Organization (IHO 1994, pages 211 and 121), a seamount is 'an isolated or comparatively isolated elevation rising 1000 m or more from the seafloor and of limited extent across the summit', whereas a knoll is 'a relatively small isolated elevation of a rounded shape rising less than 1000 m from the seafloor and of limited extent across the summit'. The first regional count of seamounts in the North Atlantic was carried out by Epp & Smoot (1989), who used multibeam data to identify approximately 800 seamounts between the equator and Iceland. More recently, seamount-like features interpreted on shiptrack bathymetry data (Hillier & Watts 2007), gridded bathymetric data (Yesson et al. 2011), and satellite-derived gravity anomaly data and its vertical gradients (Wessel 2001;Kim & Wessel 2011) were catalogued in regional and global databases.
The igneous centres from the NE Atlantic identified in the NAG-TEC study ) is a collection of 429 features that has been divided into six subunits: offshore seamounts; igneous complexes; inactive calderas; active calderas; inactive central volcanoes; and active central volcanoes (Horni et al., this volume, in review). They occur both onshore and offshore, on oceanic and on continental crust (Figs 1 & 2). An overview of the complete NE Atlantic igneous centre compilation is presented in .
In this contribution, we will focus on seamounts and igneous edifices situated on NE Atlantic oceanic crust (Fig. 1). These features were identified on published multichannel and single-channel seismic reflection profiles as mounded or bank features with dipping flanks and commonly erosional features on top (Fig. 3). In areas with no seismic control, bathymetry (SRTM30_PLUS: Becker et al. 2009), gravity (Andersen 2010) and magnetic gridded data (Gaina et al., this volume, in review) were used for locating seamounts and other volcanic-like edifices, which are usually characterized by circular or elliptical anomalies in potential field data (Fig. 2). The seamount-like features were first manually identified on bathymetry and gravity data, and the interpretation was cross-checked with the magnetic anomaly maps. It has been assumed that a magnetic source will result in a distinct magnetic anomaly, and therefore only features with clear signatures on bathymetry, gravity and magnetic data have been considered in this database. Only features that rise more than 500 m above their surroundings and have a subcircular or well-defined base were included in the NAG-TEC database. Their structure varies and some are flat-topped, while others are more peaked. In addition, the volcanic features described in the EarthRef database (Earthref.org/SC) were also included (Fig. 3). Altogether, 175 identified features have an elevation of more than 500 m (therefore they fall into the 'knoll' category), but only 12 of them are over 1000 m in height (and can be called 'seamounts'). We suggest labelling the volcanic edifices discussed in this paper as 'seamount-like oceanic igneous features' (SOIFs), an acronym that is used in the rest of the paper.
For studying the geodynamic context of SOIF formation in various sub-basins of the NE Atlantic, we have scrutinized four main regions that display various volcanic activity patterns (Fig. 4). The structure and evolution of these sub-basins differ depending on their geographical location relative to the Iceland plume and on the proximity to additional plate boundaries -including the ones created by the formation of the Jan Mayen microcontinent (JMMC) (Fig. 1) (e.g. Gaina et al. 2009). In this study, we do not discuss in detail the magmatic history of Iceland or volcanism formed on extended continental crust.
Region I: south of Iceland and north of the Bight Fracture Zone The identification of SOIFs in region I (Fig. 5a) is based on bathymetry (in most cases, satellitederived altimetry and multibeam data for a few cases: see EarthRef.org/SC) only. The sediment thickness ) in these basins is less than 2 km (Fig. 2d). The majority of SOIFs are located in three distinct areas. Few edifices (seven out of 42) are on Early Eocene crust (c. 52 -54 Ma), close to the identified continent-ocean boundary (COB). Most of the volcanic features (28 out of 42) are located on Late Eocene -Oligocene crust (38-20 Ma), and are distributed almost symmetrically on both conjugate oceanic ridge flanks. SOIFs are not identified on the conjugate European flank in the northern Iceland Basin, at approximately 638 N, in a region with higher sediment thickness than on the Greenland flank. Note that part of that thick sediment succession constitutes Late Miocene and younger drift deposits resulting from the onset of deep-water circulation in the NE Atlantic. Two recent seismic reflection lines crossing the northern Irminger and Iceland basins from west to east ( fig. 1 in Parnell-Turner et al. 2015) imaged prominent high bathymetric features, including V-shaped ridges, under the sediment pile on both flanks of the Reykjavik Ridge (Parnell-Turner et al. 2015).
In region I, less than 20% of seamounts/volcanic edifices (seven out of 42) are on Mid-Miocene-Recent oceanic crust. Here, we do not discuss the volcanic edifices observed in the Rockall region on continental or extended continental crust: they have been described in detail in previous studies (e.g. Jones et al. 1994;O'Connor et al. 2000).
In the Irminger and Iceland basins, a clear change in the seafloor spreading regime occurred at C17 time (c. 38 Ma). The seafloor spreading direction changed by 208 -258 counterclockwise and the spreading rate decreased by about 30% (see Gaina et al., this volume, in review). As a result, the oceanic crust was transformed from a linear, fracture-zone-free fabric to a 'stair-case'-like fabric due to the appearance of small offset fracture zones, especially in the area south of 608 N and north of the Bight Fracture Zone (Fig. 1).
A closer look at the second group of SOIFs described above reveals that the volcanic edifices are mostly elliptical in shape, and some of them coincide with the intersection between fracture zones and palaeo-mid-ocean ridges (identified as magnetic isochrons) that formed between C13 and C6. Further observations related to the oceanic crust characteristics in the regions linked to SOIF occurrences are summarized in Table 1. The crustal thickness obtained with two different methods (Fig. 5), and the seafloor spreading rates and asymmetry ( Fig. 4), are described in detail in Funck et al. (2016) and Gaina et al. (this volume, in review).

Region II: the Norway Basin
The NAG-TEC SOIF database contains 34 seamounts in the Norway Basin (Fig. 5b), which were identified on the gravity, magnetics and bathymetry gridded data, with five of them cross-checked on 2D seismic reflection data. In addition, 13 SOIFs were identified as igneous centres, four of them on the eastern JMMC, in the vicinity of the COB Blischke et al., this volume, in press), and therefore linked to break-up volcanism. The emplacement of these igneous centres occurred during and immediately after the initial formation of seawards-dipping reflectors (SDRs). The igneous centres cut through the SDR section and are located close to fracture/fault zones. Peron-Pinvidic et al. (2012) and Blischke et al. (this volume, in press) suggest that the igneous centres located in the vicinity of the JMMC eastern margin are related to break-up and volcanic margin formation. Six igneous centres were identified on old oceanic crust (c. C24) close to the JMMC and four along the Norwegian margin (Table 1).
The majority of SOIFs are on Late Eocene -Early Oligocene crust (C20-C18 to C13: i.e. , flanking the Aegir extinct spreading ridge. A small number of large, elongated seamount chains or isolated rounded seamounts are also visible along the Jan Mayen Fracture Zone (JMFZ) situated in the northern Norway Basin (Table 1). One large feature was identified at the southernmost tip of the Aegir Ridge as a possible central volcano, formed by ridge propagation just prior to its extinction (Vogt & Jung 2009).
The SOIF production in the Norway Basin may have started in post-C21 (c. 47 Ma) time and continued until the Early Oligocene (C13, c. 33 Ma). This is illustrated by the fact that only a few isolated seamounts were identified on Early Eocene crust (C24-C22). Seamounts in the Norway Basin can also be seen along the oblique SSE-NNW pseudofaults, features visible on gravity anomaly maps (e.g. Fig. 2), and described by Breivik et al. (2006) and Gernigon et al. (2012). Note that in Region II, the appearance of SOIFs in Late Eocene time coincides with a change in the spreading regime, when a drop in the spreading rate and a change in the spreading direction resulted in a fan-shaped basin geometry.

Region III: Kolbeinsey Ridge and associated oceanic basin
A continuous mid-ocean ridge (MOR) was established west of the JMMC about 20 myr ago (Nunns 1983;Kuvaas & Kodaira 1997;Gaina et al., this volume, in review), and oceanic crust continued to form until today along the Kolbeinsey Ridge (Fig. 5c). A few igneous centres (five out of 22) were identified on seismic reflection, bathymetry, gravity and magnetic anomaly gridded data along the Greenland margin (and may be linked to the Oligocene break-up processes). One of these igneous centres is located on continental crust. No igneous features are visible on the conjugate western JMMC margin.
Two prominent SOIF populations are distinguished on Late Miocene -Pliocene crust (6 Ma and younger) (Fig. 5c), both of which are located in the vicinity of fracture zones (four SOIFs near the Spar Fracture Zone) or a ridge propagator tip (four SOIFs next to the southern propagator). The third distinct population (seven out of 22) is grouped SW and west of Jan Mayen Island, and includes the Eggvin Bank -a plateau with young (,1 Ma), scattered volcanic peaks (Mertz et al. 2004).

Region IV: Mohn's Ridge and associated oceanic basin
The SOIFs are divided in two distinct groups within the Greenland and Lofoten basins, which have formed along the Mohn's Ridge since C24 (c. 54 Ma) onwards (Fig. 5d). The first group (22 out of 70) is distributed along the Greenland margin on both extended continental crust and Early Eocene oceanic crust. The second group comprises 43 out of 70 SOIFs, and is scattered on oceanic crust younger than 28 Ma on both conjugate flanks of Mohn's Ridge. Only one large seamount is located on the present-day MOR in the southern Mohn's Ridge, referred to as the Troll Wall-Soria Moria (Pedersen et al. 2010), and four SOIF are located on the JMFZ.
Four seamounts are outside these two SOIF groups and are located on 44-33 Ma crust on the Greenland side. One of these seamounts is the Vesteris Seamount (Cherkis et al. 1994;Haase & Devey 1994), a young, large intra-plate volcano of nonplume origin.
North of Region IV, the compilation by Yesson et al. (2011) shows a few seamounts along the Knipovich Ridge, mostly in the Boreas Basin on the Greenland Plate. The NAG-TEC study has not included this area into its SOIF database and we will not discuss them further, as there is sparse information about volcanic centres in that region.

SOIFs and oceanic crust formation
The distribution of SOIFs in the NE Atlantic oceanic basins, and links to the age of oceanic crust, seafloor spreading rates, asymmetry of oceanic crustal accretion and oceanic crustal thickness, are summarized in Table 1. A new grid for the oceanic lithospheric age has been constructed based on updated magnetic anomaly identification in the NE Atlantic (Gaina et al., this volume, in review). The oceanic lithospheric age grid model, together with rotation parameters describing the opening of the NE Atlantic, have been used to compute seafloor spreading rates, directions and deviations from symmetrical oceanic crust formation at various intervals, as constrained by the kinematic model (see Gaina et al., this volume, in review) (Fig. 4). Seafloor spreading asymmetry can be described as the percentage of crustal accretion (values from 0 to 100%) on conjugate flanks along a MOR (Müller et al. 2008). Symmetrical seafloor spreading is expressed as 50% asymmetry, values smaller than 50% indicate less oceanic crust on one flank, which is compensated for on the conjugate flank with a crustal accretion percentage greater than 50%. Besides the age of oceanic crust, crustal thickness and seafloor spreading parameters, we also inspected the oceanic basement seismic reflection characteristics, as described by Horni et al. (this volume, in review). The NE Atlantic oceanic basement has been divided into six categories: smooth, transitional, rough, very rough, rubbly and igneous provinces   (Fig. 6). We observe that most SOIFs (75%) are associated with rough basement, and only a few with smooth basement. The rough basement type, as described by Horni et al. (this volume, in review), is present in areas with significant basement relief of the order of 1 s two-way travel time (TWT) on seismic reflection sections. They suggest that rough basement may result from tectonic processes (e.g. faulting) and volcanic processes (e.g. intrusions, seamounts or locally robust volcanism). The smooth basement type is distinguished by long continuous, high-amplitude, seismic reflections and often appears as a single reflection, although sometimes there may be packages of strong subplanar continuous reflections. The two types of basement morphology are also reflected in bathymetry (Fig. 1) and gravity data (Fig. 2).
Note that the transition from smooth to rough oceanic crust has been mainly associated with the boundary between areas affected by higher magma supply from the Iceland plume (situated on V-shaped regions south and north of Iceland) and 'colder' areas, which were less affected, or unaffected, by the hotter mantle (e.g. Poore et al. 2009). Hey et al. (2010) postulated that a seafloor spreading asymmetry-producing mechanism is rift propagation. Rift propagation produces V-shaped ridges, a 'ridge and trough' geometry within the V-shaped elevated area, and crustal thickness variations. According to Hey et al. (2010), this mechanism can be triggered independently of the presence of a mantle plume, and better explains the relief of V-shaped regions in the NE Atlantic. More recently, Jones et al. (2014) used geophysical and geochemical data and modelling to confirm the original idea of Vogt (1971) that the V-shaped ridges are generated by the plume stem pulses that spread radially from the central plume location. The model of Jones et al. (2014) explains both the seafloor spreading roughness and the geochemical signatures revealed by oceanic crust samples.
SOIFs or oceanic core complexes (OCC)? We indicated that some SOIFs are located at the intersection between fracture zones and former MORs (as depicted by isochrons). The occurrence of some of these features also coincides with a decrease in spreading rate (Fig. 7a -c). These conditions would suggest that the identified bathymetric features might be oceanic core complexes (OCC). Oceanic core complexes are bathymetric features composed Our criteria to identify SOIFs included height above 500 m and relatively high total magnetic field values (.100 nT on the NAG-TEC magnetic map, which shows the total magnetic field 2 km upwards continued from the original measurement position). Therefore, if any of the identified SOIFs in this study happen to be an OCC, then that feature was formed in a tectonic regime able to generate rounded to elliptical bathymetric structures that are surrounded, covered or intruded by basaltic rocks that have remanent magnetization.
Previous detailed studies on seamount volcanism in the NE Atlantic focused on the enigmatic Vesteris Seamount (Mertz & Renne 1995) located in the Greenland Basin (Fig. 5d), and the igneous centres situated in the Rockall region (Fig. 5a). These seamounts have been dredged and a few petrological studies have been published (e.g. Cherkis et al. 1994;Haase & Devey 1994;O'Connor et al. 2000). The young episodic alkaline volcanism of the Vesteris Seamount has been attributed to intra-plate  stresses, as the volcanic edifice is situated on older oceanic crust, at the intersection of major structural features (Haase & Devey 1994). In contrast, the episodic volcanism (from the Late Cretaceous to the Mid-Eocene) that formed the igneous edifices situated on extended continental crust in the Rockall region was linked to the Iceland plume pulsations which may have occurred at 5-10 myr intervals (O'Connor et al. 2000). The third group of NE Atlantic seamounts/submarine volcanic edifices documented by petrological studies is located in the Eggvin Bank region, situated between Jan Mayen Island and the Kolbeinsey Ridge (Fig. 5c). The off-axis, transitional to alkaline lavas are similar to the Jan Mayen Island basalts, but the near-axis tholeiites resemble the SE Iceland lavas, indicating a change in the mantle composition (Trønnes et al. 1999;Mertz et al. 2004), or different magma sources along the Kolbeinsey Ridge. Most of the NE Atlantic SOIFs located on oceanic crust lack absolute ages, as only few seamounts or volcanic edifices have been dredged and dated.    Inset map in the upper panels show the location of 'seed points' along the plate boundaries used as starting points to compute seafloor spreading rates and directions (look for matching colours). The magmatic crustal production is calculated by multiplying gridded data of oceanic spreading rates (Gaina et al., this volume, in review) and crustal thickness (Haase et al., this volume, in press). Inset maps on the lower panels show the selected corridors selected for this calculation. The results from different corridors illustrate similarities and contrasts in the crustal productivity variations. Different colours are used for conjugate sides in each corridor to illustrate asymmetries. Time intervals of SOIF formation are marked in various shades of pink (darker for higher SOIF production). Seamount ages from the Rockall region (O'Connor et al. 2000)  The amount of lithospheric flexure due to the formation of a volcanic edifice can reveal the age of loading and, indirectly, the approximate age of the volcanic feature built on oceanic crust. We have not attempted to establish accurate ages for the oceanic volcanic edifices in the oceanic basins of the NE Atlantic, but consider that they were formed not long after the underlying oceanic crust. This hypothesis is probably more accurate in the case of SOIFs developed as a result of tectonic changes accommodated at ridge-fracture zone junctions or as ridge propagation.
Based on our data compilations (Table 1) and SOIF description (Figs 5 & 6), we infer that: (1) 75% of SOIFs are associated with rough basement formed at low spreading rates, and with seafloor spreading asymmetries; (2) 35% of NE Atlantic SOIFs cluster on Late Eocene -Early Oligocene crust and 38% on Miocene-Present oceanic crust; and (3) SOIFs are associated with tectonic changes accommodated at ridge-fracture zone junctions or at ridge propagators. These observations lead to the conclusion that the SOIF magmatic activity is mostly linked to the normal oceanic crust production at times of spreading direction readjustments. We note that SOIFs occur on crust produced at intermediate spreading rates and on smooth to transitional basement, but most of them are associated with low spreading rates, asymmetry in crustal accretion and rough basement (Figs 5 & 6; Table 1). Batiza (2001) postulated that off-ridge volcanism is mainly encountered in regions with intermediate to high spreading rates and abundant melt supply. However, Standish & Sims (2010) described young off-axis volcanism along the ultraslow spreading SW Indian Ridge and suggest that this volcanism is the result of magma rising along faults, which contributes to off-axis accretion of oceanic crust. Some SOIFs can be OCCs formed in a slow seafloor spreading regime that was affected by episodes of higher than normal magma injection supply.
SOIFs and the Iceland plume. The NE Atlantic region has been strongly influenced by the Iceland mantle plume before, during and after continental break-up and subsequent seafloor spreading. Massive volcanism that predated and assisted the continental break-up formed the North Atlantic Igneous province (NAIP) in two pulses, at approximately 62 and 55 Ma, and was spread along the Greenland and NW European margin (e.g. Storey et al. 2007). Subsequently, the Iceland plumerelated magmatic activity formed the Greenland-Faroe province (GIR and FIR in Fig. 1) from the Eocene onwards (Soager & Holm 2009). However, age-progressing seamount chains are absent from the NE Atlantic region.
About 100 igneous centres have been identified in Iceland and the surrounding regions (Fig. 1), and most of them show the evolution of volcanism connected to mid-ocean ridge-plume interactions. All other seamount and igneous feature clusters described in this study do not show distinct linear trends, and we infer that they cannot fall into the hotspot seamount chain category. Some of the NW-SE-trending seamounts in the Greenland Basin may resemble linear seamounts formed in the direction of plate motion relative to the mantle (so-called 'hot lines' or melting anomalies elongated in the direction of plate motion). The motion of Greenland relative to the mantle for the last 40 myr (shown by the white arrow in Fig. 5d) appears to have been sub-parallel to these seamount trends in the Greenland Basin.
Plume-related volcanism has been invoked to explain the formation of seamounts in the Rockall region (O'Connor et al. 2000), and the elevated bathymetry and V-shaped ridges south of Iceland in the Iceland and Irminger basins (White et al. 1995;Jones et al. 2002). In order to study any possible correlations between episodic pulsations of the Iceland plume (Jones et al. 2002;Parkin et al. 2007;Parnell-Turner et al. 2014) and the formation of seamounts in the basins situated south and north of Iceland, we examined the evolution of oceanic crust production. We computed the amount of oceanic crustal accretion in our selected regions (Figs 4 & 5) by taking into account the crustal thickness derived from gravity anomaly inversion (Haase et al., this volume, in press) and the spreading rates (Gaina et al., this volume, in review), and compared the magmatic production and spreading rates fluctuations with suggested episodes of high plume activity (Fig. 7). According to recent studies based on high-quality seismic reflection data, the Iceland plume activity had peaks every 3 myr from 55 to 35 Ma, and every 8 myr from 35 Ma to the present day (Parnell-Turner et al. 2014). We observed that the identified periods of SOIF production south and north of Iceland and the Greenland -Faroe Ridge (GIR and FIR in Fig. 1) fall within pulses of higher activity of the Iceland plume, but not every pulse resulted in SOIF cluster production. The three periods of seamount formation in the Rockall region, dated at 52, 47 and 42 Ma by O'Connor et al. (2000), coincide with pulses in the magmatic crustal production in all oceanic basins (mostly asymmetrical on conjugate flanks), decreases in spreading rates and changes in spreading directions, but have no obvious connection with SOIF production (Fig. 7).
We notice that SOIF formation in the basins north of the JMFZ coincides with an increase in spreading rates and magmatic production, reflected in higher crustal thickness. South of the JMFZ, Late Eocene -Early Miocene SOIF formation occurred when the spreading rates and magmatic crustal production first decreased and became highly asymmetrical (Fig. 7). Parnell-Turner et al. (2014) observed that the Iceland plume fluctuations are superimposed on a rapidly cooling temperature structure manifested by a northwards shift from smooth to rough crust in the Iceland and Irminger basins. The NE Atlantic oceanic basement morphology model ) and kinematic history (Gaina et al., this volume, in review) also show changes at the times of Iceland plume pulsations, as postulated by Parnell-Turner et al. (2014). We infer that SOIF formation south of Iceland occurred as plume activity decreased, but coincided with a considerable change in plate motion, which resulted in transtensional motion that allowed localized additional melting along the newly formed fracture zones and ridge propagators. Interestingly, recent seismic activity (from ISC catalogue: http://www.isc.ac.uk); appears to cluster in some of the SOIF locations (Fig. 6). Apart from the seismicity associated with active mid-ocean ridges, we observed seismic activity in Region I in the proximity of the SOIF cluster located on Oligocene oceanic crust, in Region II in the Norway Basin and in Region IV close to SOIFs situated on Oligocene-Miocene oceanic crust. A more conservative seismic event catalogue, the EHB Bulletin (http:// www.isc.ac.uk/ehbbulletin), shows much less intraplate seismic activity in the southern part of the NE Atlantic Ocean. Note that the seismic events extracted from the EHB Bulletin indicate only earthquakes with magnitude greater than 3, which occurred before 2008. We suggest that a link may exist between seismic events and SOIF distribution, and indicates that lithosphere weakening due to tectonic activity combined with subsequent volcanic loading may leave a long-lasting imprint and facilitate subsequent crustal deformation.

'Paired' SOIFs
Although SOIF formation in all NE Atlantic subbasins occurs on conjugate flanks of the same age, Fig. 8. Plate reconstruction at C6 (19.7 Ma). Background gridded data are the free-air gravity curvature (Sandwell & Smith 2009) in the grey palette and the spreading asymmetry (colour palette as in Fig. 4). SOIF outlines are in black; the pink ellipse shows the position of possible paired SOIFs along Reykjanes Ridge at C6 time.
there are very few 'paired basement ridges', which are volcanic edifices built at mid-ocean ridges (MORs) and equally split on both flanks by subsequent MOR evolution. Vogt & Jung (2005) described a series of paired basement ridges in the North Atlantic realm, including V-shaped ridges of Reykjanes Ridge, and suggested that small-scale conjugate ridge pairs are generated by axial magmatic centres that are not fixed to a mantle frame. They also postulate that the magma centres pulsate at 0.2-1.0 myr intervals and may be active for at least 1 myr, creating several off-axis ridge pairs. Among our identified SOIF, we observed two cases of small-scale conjugate volcanic pairs. Figure 8 shows the reconstructed locations of one pair of such conjugate volcanic features at the intersection between the Reykjanes MOR and a non-offset fracture zone at approximately 20 Ma. The underlying oceanic crust shows asymmetry on conjugate flanks (Fig. 8), and was formed at the time when the seafloor spreading direction changed to clockwise and the spreading rate increased (Fig. 7a). According to the Parnell-Turner et al. (2014), the model of the Iceland plume pulsation episodes, the 19-20 Ma formation of the paired SOIF along the Reykjanes axis, coincides with the increase in mantle plume activity. However, this may only be a coincidence, as the size and number of paired SOIF cannot justify a clear link between the two processes. Changes in plate motion and transtension at the MOR-fracture zone intersection seem to be a more realistic explanation for the formation of these small-scale paired SOIFs along the Reykjanes Ridge.

Conclusions
We have inspected the new NE Atlantic database of oceanic volcanic edifices including seamounts and igneous centres (abbreviated as SOIFs in this study) in four different regions situated south and north of the Greenland-Faroe Ridge. SOIF occurrences are distributed differently in the four regions, but we have identified three distinct 'pulses' of abundant seamount cluster formation. SOIFs on older oceanic crust (54-50 Ma) are situated on smooth oceanic basement. If the seamounts were formed at the time of, or shortly after, seafloor spreading, then their emplacement coincides with an increase in spreading rates and higher magmatic productivity. Large seamounts were formed on the Rockall plateau and in the Rockall Trough around 52 Ma, and the Iceland plume activity increased at 55 and 52 Ma, which could indicate that the Early Eocene SOIF formation in the NE Atlantic may have resulted from higher than usual mantle plume activity.
The second SOIF group is located on Late Eocene -Early Miocene oceanic crust of the Irminger, Iceland and Norway basins. This group is located on rough oceanic basement, in the proximity of newly formed tectonic features, such as fracture zones, ridge propagators and V-shaped ridges. In the Late Eocene, seafloor spreading rates dropped and the crustal production became highly asymmetrical. We suggest that the formation of these volcanic edifices is mostly related to kinematic changes, which led to local readjustments of MOR segments and fracture zones. The third SOIF population is observed on Mid-Miocene -Present oceanic crust, mostly located in the Greenland and Lofoten basins, but also on conjugate flanks of the Kolbeinsey Ridge. In addition, these SOIF clusters are associated with rough oceanic basement and fracture zones, V-shaped ridges, and ridge propagators. Early-Mid-Miocene (c. 27 Ma) oceanic crust registered an increase in spreading rate and magmatic productivity, which appears to coincide with a burst of seamount formation north of the JMFZ. South of the JMFZ, this activity was delayed until about 11 Ma, as shown by the higher number of seamounts emplaced on crust of this age and younger. It is not clear which processes initiated and dominated this third period of SOIF formation, as it seems that both tectonic/kinematic and magmatic triggers were present. However, the Early Miocene relocation of the mid-ocean ridge from the AEgir Ridge to the Kolbeinsey Ridge, west of the Jan Mayen microcontinent, may have played a role in delaying SOIF formation south of the JMFZ.
We note that the identified periods of seamount production south and north of Iceland and the Greenland -Faroe Ridge fall within Iceland plume pulses of higher activity, but not every pulse resulted in SOIF cluster production. We conclude that the episodic activity of the Iceland plume, combined with regional changes in relative plate motion, led to local MOR readjustments that enhanced the likelihood of seamount and other igneous feature formation. Further studies aimed at dating the identified SOIFs will help in understanding the connection between tectonic and magmatic activity in the NE Atlantic.
We acknowledge the support of the NAG-TEC industry sponsors (in alphabetical order): Bayerngas Norge AS; BP Exploration Operating Company Limited, Bundesanstalt für Geowissenschaften und Rohstoffe (BGR); Chevron East Greenland Exploration A/S; ConocoPhillips Skandinavia AS; DEA Norge AS; Det norske oljeselskap ASA; DONG E&P A/S; E.ON Norge AS; ExxonMobil Exploration and Production Norway AS; Japan Oil, Gas and Metals National Corporation (JOGMEC); Maersk Oil; Nalcor Energy -Oil and Gas Inc.; Nexen Energy ULC, Norwegian Energy Company ASA (Noreco); Repsol Exploration Norge AS; Statoil (U.K.) Limited, and Wintershall Holding GmBH. C.G. acknowledges support from the Research Council of Norway through its Centres of Excellence funding scheme, project number 223272. GSK publishes with permission of the Executive Director of the British Geological Survey (Natural Environment Research Council). The authors thank Thomas Funck, Christian Berndt and an anonymous reviewer for their suggestions that greatly improved the manuscript.