The Amirante Ridge and Trench System is located south of the Seychelles within the Mascarene Basin (Figure 1). It consists in a ∼600-km-long, up to 5300-m-deep arcuate trench, flanked on its eastern side by a basaltic ridge, which supports some reef atolls [Johnson et al. 1982]. The Amirante Ridge and Trench System is still one of the most enigmatic features of the Indian Ocean, with various origins proposed so far, from an extinct plate boundary [Masson 1984; Damuth and Johnson 1989; Mart 1988; Mukhopadhyay et al. 2013; Eagles and Hoang 2014] to a remnant of a meteorite impact [Hartnady 1986]. Although the hypothesis of a crater has been dismissed [Plummer 1996], the nature and origin of the fossil plate boundary preserved at the Amirante Ridge and Trench System remain unclear [Damuth and Johnson 1989; Mukhopadhyay et al. 2013], as the structure ambiguously shares some morphological and structural characteristics of subduction zones [Johnson et al. 1982; Mart 1988], transform faults [Plummer 1996] and mid-oceanic ridges [Eagles and Hoang 2014]. The age of basalts dredged at the Amirante Ridge spans the Late Cretaceous–Early Paleocene period [Fisher et al. 1968; Johnson et al. 1982], a period marked by drastic changes in the configuration of the Indian Ocean’s plate boundaries (i.e. curvature of fracture zones at the SW Indian Ridge, [Cande and Patriat 2015]; migration of the India–Africa transform boundary; [Rodriguez et al. 2016, 2020]; extinction of the Mascarene and East Somali Basin spreading centers and rifting between Seychelles and India; [Gaina et al. 2015; Yatheesh 2020]).
The objective of this paper is to determine how the Amirante Ridge and Trench System fits in the complex framework of the Indian Ocean’s fabric during the Late Cretaceous–Early Paleocene interval in order to unravel its tectonic origin. I provide a comprehensive paleogeographic reconstruction of the NW Indian Ocean from the Late Cretaceous to the Early Paleocene (Figure 2), based on magnetic anomalies recorded in the oceanic basins [Dyment 1993, 1998; Bernard and Munschy 2000; Royer et al. 2002; Chaubey et al. 2002; Yatheesh 2020] and the distribution of ophiolites surrounding the Arabian Sea [Gnos et al. 1998; Mahoney et al. 2002; Khan et al. 2007b,a; Kakar et al. 2012, 2014; Siddiqui and Ma 2017]. I show that prior to the eruption of the Deccan Traps and India–Seychelles breakup circa 65 Ma [Yatheesh et al. 2009; Calvès et al. 2011], the Amirante Ridge and Trench System was in line with a ∼1500-km-long subduction zone running along NW India, nowadays preserved in the ophiolites running from Bela to Waziristan in Pakistan (Figure 1). I finally propose that the Amirante Ridge and Trench System constitutes the southern termination of the NW India subduction zone and discuss the implications for the evolution of the NW India passive margin during the India–Seychelles breakup episode and prior to the India–Eurasia continental collision.
2. Geological background
2.1. Opening of the NW Indian Ocean
The opening of the NW Indian Ocean results from the fragmentation of Gondwana since the Late Jurassic [McKenzie and Sclater 1971; Norton and Sclater 1979; Schlich 1982; Besse and Courtillot 1988], in response to complex interactions between subduction zones and mantle plumes [Coltice et al. 2009; Gaina et al. 2013; Matthews et al. 2012], which led to the separation of the India–Seychelles–Madagascar block from Africa and Australia [Gibbons et al. 2013]. The initiation of the Southern Neotethys Subduction documented around 110–105 Ma from metamorphic soles of the related ophiolites [Guilmette et al. 2018; Pourteau et al. 2018] drove the India’s drift toward Eurasia, while the Marion and Deccan plumes acted as rheological facilitators over the breakup of the Madagascar–Seychelles–India continent [Torsvik et al. 2000; Minshull et al. 2008; van Hinsbergen et al. 2011], which occurred in three major steps (see paleogeographic reconstructions of [McKenzie and Sclater 1971; Norton and Sclater 1979; Schlich 1982; Calvès et al. 2011; Gaina et al. 2015; Yatheesh et al. 2019; Yatheesh 2020]).
First, the ophiolites scattered from Bela to Waziristan in Pakistan [Tapponnier et al. 1981] testifies that a mid-oceanic ridge separated the Kabul block from NW India during Albian–Senonian (Figure 2; [Gnos and Perrin 1997; Gaina et al. 2015]).
Second, the opening of the Mascarene and the East Somali Basins records the Late Cretaceous breakup between Madagascar and the India–Laxmi–Seychelles continental blocks (Figure 2; [Schlich 1982; Bernard and Munschy 2000; Calvès et al. 2011; Gibbons et al. 2013, 2015; Bhattacharya and Yatheesh 2015; Shuhail et al. 2018]).
Third, the breakup between the Seychelles–Laxmi block and India formed the Gop Basin (between magnetic chrons 31 (68 Ma) and 30 (67 Ma; [Minshull et al. 2008; Yatheesh et al. 2009]). Following the culmination of the Deccan trap volcanism at ∼65.5 Ma [Courtillot and Renne 2003; Hooper et al. 2010; Verma and Khosla 2019], the rifting migrated between the Laxmi Ridge and the Seychelles block (Figure 2) and led to the formation of the Carlsberg mid-oceanic ridge since chron 28 (63 Ma) [Dyment 1998; Chaubey et al. 2002; Royer et al. 2002].
2.2. The Amirante Ridge and Trench System: geological observations
The Amirante Ridge and Trench puzzle only relies on a few ground-truth geological constraints. The Amirante Ridge is capped by tholeiitic basalts, dated at 82 ± 16 Ma according to K/Ar method [Fisher et al. 1968]. This age reflects the formation of the oceanic lithosphere exposed at the Amirante Ridge, but does not necessarily indicate the age of uplift of the ridge. Pelagic chalk cored at the Amirante Passage, located at the southwestern termination of the Amirante Trench, gives a minimum age of 73–66 Ma [Masson 1984; Johnson et al. 1982]. Ponded turbidites, mass transport deposits, and pelagic sediments filled in the Amirante Trench [Masson 1984]. There, the sedimentary record does not reveal any major tectonic deformation, which makes the tectonic origin of the system difficult to decipher. The Seychelles Island displays some calcalkaline dolerite dykes and syenite batholiths, formed until 50–46 Ma [Mart 1988].
2.3. Previous interpretations of the Amirante Ridge and Trench System
The presence of a trench on the seafloor morphology (Figure 1), together with the calcalkaline lavas at the Seychelles suggest that the Amirante Ridge and Trench System may correspond to a fossil subduction zone, as initially proposed by Fisher et al. . However, the geochemistry of basalts dredged at the Amirante Ridge itself does not reflect the influence of a subduction (Mid-Oceanic Ridge Basalt signature, [Mart 1988]). The free-air gravity field of the Amirante Ridge and Trench System can be reproduced by considering only a very limited length of the slab, no more than 100 km [Miles 1982]. The lack of deformation within the trench does not necessarily rule out the interpretation as a subduction zone, as numerous subduction zones do not display any accretionary wedge. Moreover, the sedimentation rates were quite low during the Late Cretaceous–Paleocene in the NW Indian Ocean (mainly pelagic sediments, [Shipboard Scientific Party 1989; Rodriguez et al. 2016]), or poorly preserved since then [Willenbring and von Blackenburg 2010], making the record of tectonic events difficult.
As a result, the Amirante Ridge and Trench System is either interpreted as a remnant of a subduction segment, which did not consume an important amount of oceanic lithosphere [Miles 1982], or a mixed system of mid-oceanic ridge and transform fault surrounding the Seychelles microplate [Damuth and Johnson 1989; Plummer 1996], which was formed during its isolation from India subsequent to the underplating of the Deccan plume [Eagles and Hoang 2014]. Rotation of microplates may indeed induce local compression or extension leading to formation of structures comparable to the Amirante Ridge and Trench System [Hey 2004; Matthews et al. 2015]. However, the rotation of such microplates affects the fabric of the surrounding oceanic lithosphere (e.g. rotation of nearby ridges), which is not observed in the vicinity of the Amirante System for the Late Cretaceous fabric [Bernard and Munschy 2000]. Yet, the interpretation of the Amirante Ridge and Trench System as a subduction zone suffers from the difficulty to determine how a 600-km-long subduction initiated south of the Seychelles during the Late Cretaceous, coeval with the rifting and oceanic accretion episodes that contributed to the fragmentation of the India–Seychelles–Madagascar block.
3. Reconstructions of NW India Margins during the Late Cretaceous–Paleocene
The basin that used to be located between the NW Indian margin and the Kabul block (Figure 2) has been subducted during the Late Cretaceous–Early Paleocene interval, giving birth to a series of ophiolites including, from North to South, the Waziristan-Kost, Zhob, Muslim Bagh, and Bela ophiolites (Figure 1). The slab penetrated down to at least ∼1000 km depth in the mantle, according to tomography [Gaina et al. 2015]. The obduction of eastern Pakistan ophiolites initiated around 70–65 Ma, prior to the onset of the Deccan traps, and ended no later than 55 Ma, according to the dating of the carbonate platform that sealed the ophiolites [Beck et al. 1995] and detrital remnants of these ophiolites encountered in the earliest Indus River sediments [Zhuang et al. 2015].
Most of these ophiolites display metamorphic soles at their base. Metamorphic soles correspond to thin metamorphosed tectonic slivers detached from the downgoing plate during the first few million years of intra-oceanic subduction [Agard et al. 2018]. The dating of the amphibolite facies of the metamorphic soles provides clues over the timing of subduction initiation. In addition, most of these ophiolites display a supra-subduction zone signature, with sometimes traces of Oceanic Island Basalts corresponding to a pre-Deccan plume [Mahoney et al. 2002]. Supra-subduction zone basins correspond to the oceanic lithosphere accreted in the upper plate, at the corner of the subduction interface. These basins are generally composed of Mid-Oceanic Ridge Basalts, marked by a negative Nb and Ta anomaly in rare-earth elements patterns. The direction of thrusting of the ophiolites can be deciphered from the related structures. This set of observations constrains the vergence of the subduction zone.
Here, we provide an update of the detailed reconstructions of the India–Seychelles breakup based on magnetic anomalies of the seafloor and seismic profile analysis [Dyment 1998; Chaubey et al. 2002; Calvès et al. 2011; Yatheesh 2020], which includes the configuration of the NW India subduction deduced from the ophiolite record (Figure 2). We do not address the problem of the shape of Greater India and choose a configuration where the entire India is composed of continental lithosphere, with a northern passive margin of about 1000 km in length [Ali and Aitchison 2008; Guillot et al. 2008], although alternatives may be considered [van Hinsbergen et al. 2012].
The amphibolite facies of the metamorphic soles document a series of ages of subduction initiation (Figure 2) no later than 90 Ma at the latitude of the Waziristan ophiolite (K/Ar dating of the amphibolite facies; [Gnos et al. 1998]), 70.7 ± 5 Ma at the latitude of the Muslim Bagh ophiolite (amphibolite facies [Mahmood et al. 1995], and ∼65 Ma at the latitude of the Bela Ophiolite (Ar/Ar dating of hornblende; [Gnos et al. 1998]). The accretion of the Muslim Bagh ophiolite started at 80 Ma from the dating of plagiogranites preserved in the sequence. Considering the supra-subduction origin of this basin would imply subduction initiation at this latitude around 80 Ma [Kakar et al. 2012]. The configuration of supra-subduction zone basins recorded between the Waziristan and Bela ophiolites suggests a northwestwards subduction of India beneath the plate supporting the Kabul block, with ophiolites thrusted over the Mesozoic Indian passive margin [Kakar et al. 2012]. The NW Indian subduction therefore propagated from north (Waziristan) to south over a period of ∼30 Myrs and reached the area of India–Seychelles breakup around 73–63 Ma. A slab break-off event is recorded at Bela ophiolites around 70–65 Ma [Gnos et al. 1998].
In these reconstructions, the Amirante Ridge and Trench System is in line with the NW Indian subduction during the 73–63 Ma period. The configuration of the Amirante Ridge and Trench System is in agreement with a vergence of the subduction to the North East, that is, a subduction beneath the NW Indian margin.
The maturation of the Gop Basin and the accretion of oceanic lithosphere within the East Somali Basin from 73 to 68 Ma [Gaina et al. 2015] are coeval with the slab break-off event and mark the disconnection of the Amirante Ridge and Trench System from the segment of the subduction nowadays preserved in the Eastern Pakistan ophiolites. The Amirante segment of the subduction remained active until the Eocene, feeding the calkalkaline volcanism of the Seychelles [Mart 1988], while a subduction segment remained active until 55 Ma at the suture zone marked by eastern Pakistan ophiolites.
In this framework, the Dalrymple Trough and the Murray Ridge (Figure 1) are located at the transition between the NW and NE dipping subduction. The present-day configuration of the Dalrymple Trough and the Murray Ridge results from a series of Late Miocene–Early Pleistocene tectonic events related to the structural evolution of the India–Arabia plate boundary (i.e. the Owen Fracture Zone, [Rodriguez et al. 2014, 2018]). This series of vertical motions of the lithosphere opened a window to the lithospheric mantle. Peridotites dredged at the Murray Ridge–Dalrymple Trough reveal a supra-subduction signature [Burgath et al. 2002]. The area of the Dalrymple Trough and the Murray Ridge consists in a thinned continental lithosphere [Minshull et al. 2015], commonly attached to the segment of the Indian margin formed during the Late Jurassic–Early Cretaceous opening of the Western Somali Basin [Calvès et al. 2011; Minshull et al. 2015]. In this configuration, the supra-subduction signature of the peridotites dredged at Murray Ridge records the NE dip of the subduction. An alternative would be to consider that the area of the Dalrymple Trough derives from the Kabul Terrane, accreted to India since Paleocene [Tapponnier et al. 1981]. In this case, the supra-subduction zone preserved there is in line with the supra-subduction domains marking the NW dip of the subduction.
4.1. The nature of the Amirante Ridge and Trench System
The paleogeographic reconstructions (Figure 2) suggest that the Amirante Ridge and Trench System is the southern termination of more than 1500-km-long subduction that used to run along NW India. However, several alternative interpretations remain for the precise nature of the Amirante Ridge and Trench System. The source of the uncertainties relies in (1) the lack of documentation of the potential supra-subduction zone signature of the Amirante Ridge, which makes difficult to discriminate if it is related to a subduction zone or another type of plate boundary; (2) the lack of the precise dating of the formation of the Amirante Ridge and Trench System; (3) the difficulty to decipher the nature of the connection between the Mascarenes spreading center and the NW Indian subduction.
Ramana et al.  suggest that a major transform was running along India and Seychelles prior to their breakup, hence acting as a boundary with the plate supporting the Kabul block, which connected the Mascarenes spreading center to the NW Indian subduction (Figure 2). Such a transform could have progressively turned into a subduction during the southwestwards propagation documented in the metamorphic soles of the Pakistan ophiolites (Figure 2).
If the formation of the Amirante Ridge and Trench System pre-dates the dislocation of the NW Indian Subduction documented around 73–68 Ma, then the Amirante Trench probably reflects a short subduction zone segment connecting the transform proposed by Ramana et al.  and the Mascarenes spreading center, similar to the present-day configuration of the Hjort trench and ridge system at the Pacific–Australia boundary [Meckel et al. 2003]. In this case, the transform connects two subduction zones: the subduction at the Amirante System and the NW Indian Subduction. The transform progressively vanishes as the NW subduction propagates southwestwards. This interpretation goes along with a mixed transform and subduction origin for the Amirante Ridge and Trench.
If the formation of the Amirante Ridge and Trench system post-dates the dislocation of the NW Indian Subduction, then the dip reversal of the subduction in the Amirante area reflects the slab break-off event recorded at the Bela ophiolites around 65–70 Ma [Gnos et al. 1998]. Most of the ophiolites exposed between Bela and Waziristan display a Mid-Oceanic Ridge Basalt geochemical signature, in addition to some traces of Ocean Island Basalts due to interactions with the Deccan plume and Island Arc Basalts formed within the back-arc domain [Kakar et al. 2012; Siddiqui and Ma 2017]. In this framework, the Amirante Ridge and its Mid-Oceanic Ridge Basalt signature may represent a natural example of the ridge formed during the earliest stages of an intra-oceanic subduction in thermo-mechanical models, prior to the full development of an obduction [Duretz et al. 2016].
4.2. Relationships between the Amirante Ridge and Trench system and the Indian passive margin
The segment of the NW Indian subduction zone running from Bela to the Amirante was dipping beneath the Indian margin. It explains the supra-subduction zone signature recognized at the Dalrymple Trough (Figure 2; [Burgath et al. 2002]), which used to be a distal part of the Indian passive margin [Edwards et al. 2000, 2008]. A supra-subduction zone signature has been suggested as far as the Laxmi Ridge [Pandey et al. 2019], although many ambiguities need to be deciphered [Clift et al. 2020]. In our framework, the supra-subduction zone signature identified along the NW Indian margin at the Dalrymple Trough and Murray Ridge reflects the influence of the NW Indian Subduction, deactivated during the Paleocene–Eocene, instead of a subduction induced by the enhanced activity of the Deccan plume around 65 Ma, as proposed by Pandey et al. .
4.3. The Amirante Ridge and Trench system in the frame of the 73–63 Ma plate reorganization event
During the 73–63 Ma interval, the Western Indian Ocean’s plate boundaries experienced a series of drastic changes in their configuration, recorded by the ridge jump from the Mascarene Basin to the Gop Basin and the Carlsberg Ridge [Yatheesh 2020], the curvature of fracture zone at the SW Indian Ridge [Cande and Patriat 2015], and a major migration of the India–Africa transform boundary from Chain Ridge to the Chain Fracture Zone offshore Somalia [Rodriguez et al. 2020]. This episode of plate reorganization is commonly attributed to the enhanced activity of the Deccan plume, through the action of plume head forces [Cande and Stegman 2011] or the coupling of the subducting plates with large-scale mantellic conveyor belts [Jolivet et al. 2016]. Other invoked driving mechanisms involve the penetration of the Indonesian Slab in the lower mantle during the Late Cretaceous/Early Tertiary period, which affects the slab pull force [Faccenna et al. 2013], or collision of continental terranes, such as the collision of the Woyla Arc [Wajzer et al. 1991; Gibbons et al. 2015] and Burma block by the Late Maastrichtian [Socquet and Pubellier 2005] or the Early Tertiary [Searle et al. 2007] with southeastern Eurasia.
The NW Indian subduction zone, and its segment running from Bela to the Amirante, may have played a key role in this episode of plate reorganization event. While the subduction branch of the Neotethys running along Africa deactivated at Troodos (Cyprus) ∼75 Ma [Robertson 1977], the NW Indian subduction deactivated ∼63 Ma, following the slab break-off recorded at the Bela ophiolites [Gnos et al. 1998]. India was therefore bounded to its northern and northwestern boundaries by major subduction zones, which likely contributed to the faster motion of India compared to Africa during the Late Cretaceous. The slab break-off at Bela, together with the separation of the Amirante segment in the wake of the India–Seychelles breakup events, occurred around 63–65 Ma and likely contributed to the end of the episode of plate reorganization affecting the Indian Ocean at this period.
In detail, the short (∼100 km) amount of subducted material deduced from gravity models [Miles 1982] may be explained either by a low subduction rate after the isolation of the Amirante subduction, or by the formation of a slab tear, resulting in the detachment of a longer slab, as suggested from backward mantle tomography [Glisovic and Forte 2017]. The plate kinematics analysis by Cande et al.  requires a diffuse area of compression in the area of the Amirante System and along NW India to accommodate at least ∼100 km of convergence during the Paleocene–Early Eocene. Considering the NW Indian subduction and the Amirante Ridge and Trench System as part of the same plate boundary (between India and the microplate supporting the Kabul block) may provide an explanation for the origin of this enigmatic episode of convergence.
Paleogeographic reconstructions of the NW Indian margin reveal that the Amirante Ridge and Trench System was the southern termination of a major subduction zone running along the NW Indian margin. The NW Indian subduction split in two parts in the wake of a slab break-off event recorded around 70–65 Ma and the series of breakup and seafloor spreading events at the origin of the Gop Basin and a part of the East Somali Basin, prior to the formation of the Carlsberg Ridge 63 Ma. Considering the Amirante subduction further explains the supra-subduction signature identified along the NW Indian passive margin.
The reconstructions suggest that the Amirante Ridge may represent either a rare case of a ridge formed during the earliest stages of an intra-oceanic subduction or a mixed transform and subduction feature, similar to the present-day Hjort trench at the Australia–Pacific boundary. A more complete geochemical analysis of the basement rocks outcropping at the Amirante Ridge may help to buttress the scenario of its origin.