Refertilization of Mantle Peridotites from the Central Indian Ridge: Response to a Geodynamic Transition Open Access

The tectonic and magmatic processes contributing to the heterogeneous nature of the upper mantle posit important constraints on the composition and differentiation of the earth at various scales. Mid oceanic ridges (MOR) of intermediate, slow, and very slow spreading rates across the world’s oceans are characterized by different crustal configurations portraying the compositional diversity of the upper mantle. The two viable processes that are attributed to chemical and isotopic heterogeneity of the mantle are: (i) multiple episodes of melt replenishment, melt-rock interaction, and mantle refertilization in open magma systems, and (ii) recycling of oceanic and continental crustal materials and components of subcontinental lithospheric mantle. Geochemical signatures of magmas in oceanic realms, owing to minimal effects of contamination during ascent, record imprints of multiple episodes of melt replenishment, melt-rock interaction, fractional crystallization in open magma systems, and crustal recycling from paleotectonic events. These features provide a natural archive to probe into the compositional anomalies of the melting domains within the mantle. The mid oceanic ridge system (MOR) is a dynamic expression of the mantle convection cycle that governs the horizontal motion and interactions of lithospheric plates across the globe. Magmas generated at MORs form the oceanic crustal sequence by decompression melting of the upper mantle in response to plate divergence and hence render the upper mantle refractory in nature owing to successive melt extraction processes [1–3]. The oceanic crustal sequence consists of mid oceanic ridge basalts (MORB) including dykes, lower crustal cumulate gabbros, and peridotites, whereas the serpentinized abyssal peridotites tectonically exhumed along the fracture zones, within transform faults, and locally exposed on rift valley floors at some slow-spreading ridges represent the residual upper mantle section of the oceanic lithosphere [4]. The geochemical and petrogenetic attributes of abyssal peridotites have yielded conspicuous evidence for the compositional diversity of suboceanic lithospheric mantle at divergent settings. Mantle heterogeneities from the global MOR systems have previously been addressed by researchers like Niu [4], Seyler et al. [5], and Brunelli et al. [6]. The chemical heterogeneities of the Indian Ocean MOR mantle have been propounded by Hellebrand et al. [7], Murton et al. [8], and Saha et al. [9].

This study presents petrology, mineral chemistry, bulk-rock major, trace, and PGE data for the abyssal mantle peridotites exposed along the Vema and Vityaz fracture zones of the northern portion (between 5 and 10°S) of the Central Indian Ridge (CIR) located in the Indian Ocean to address their petrogenetic evolution, geodynamic implications, and factors contributing to the refertilization of the Indian Ocean ridge mantle. The studied peridotites from the CIR represent the upper mantle section of the oceanic lithosphere tectonically exhumed on the ocean floor along the Vema and Vityaz transform faults and fracture zones. Therefore, the mineralogical and geochemical attributes of these mantle peridotites explicitly track the magmatic response to melt-percolation, melt-rock interaction, and mantle replenishment processes operative beneath the ridge system in an extensional tectonic regime.

The MOR system comprises about 70000 km long globe-encircling underwater continuous mountain chain of constructive plate boundary spreading throughout the major world oceans. The MORs have an average width of 300 km, rise up to ~2000 m above the seafloor, and constitute about 23% of the earth’s surface [10]. The MORs are categorized into five types depending on the rate at which new crust is being created at the ridge axis: (a) ultraslow-spreading MOR with $rateoffullspreading<20mm/yr$ (e.g., Gakkel Ridge and Southwest Indian Ridge [11, 12]), (b) slow-spreading MOR with rate of full spreading between 20 and 40 mm/yr (e.g., Mid-Atlantic Ridge and Northern Central Indian Ridge [10]), (c) intermediate-spreading MOR with rate of full spreading 40-80 mm/yr (e.g., Juan de Fuca, Southern Central Indian Ridge, and Southeast Indian Ridge [13]), (d) fast-spreading MOR with rate of full spreading between 80 and 120 mm/yr (e.g., Northern East Pacific Rise [11]), and (e) ultrafast-spreading MOR with rate of full spreading between 120 and 160 mm/yr (e.g., Southern East Pacific Rise [11]). The CIR is a part of the Indian Ocean Ridge System (IORS) that includes four ridges; namely, Carlsberg (CR), Central Indian (CIR), Southwest Indian (SWIR), and Southeast Indian (SEIR) ridges. The CIR bifurcates at 25°30$′$S and 70°E at the Indian Ocean Triple Junction (IOTJ) in an apparently inverted Y shape to form SWIR and SEIR. The continuity of all these ridges is often offset by several NNE-SSW trending transform faults (also variously described as fracture zones), among which from north to south—Owen, Mahabiss, Sealark, Vityaz, Vema, Argo, Marie Celeste, Egeria and Gemino, Prince Edward, Indomed, Amsterdam, and St. Paul—are of significance.

The present configuration of the Indian Plate evolved through diverse tectonothermal, geomorphological, and paleoclimatic conditions marked by supercontinent dispersal, seafloor spreading, birth of ocean basins, mantle plume activity, transform fault movements, subduction-accretion-collisional processes, and orogenic events during a 9000 km northward journey spanning over 160 Ma timeframe. The breakup of Gondwana at 167 Ma and the Phanerozoic evolution of the Indian subcontinent with its adjacent ocean basins was predominantly influenced by the ongoing subduction of the Tethyan ocean floor beneath the southern margin of Eurasia [9, 14]. The Indian subcontinent got detached from the Gondwana supercontinent at 130 Ma and eventually amalgamated with the Eurasian supercontinent at 52 Ma. Thus, the opening of the Indian Ocean basin at 125 Ma and its subsequent evolutionary process were ubiquitously linked to the disintegration and drifting of African, Indian, Australian, Antarctic, and Arabian continental landmasses from the Gondwana supercontinent. The dispersal of Indian and African continental plates at 84 Ma led to the formation of the Northwest Indian Ocean that represents a potential zone to decipher ocean basin evolution, oceanic plate divergence, crustal accretion, propagation, and segmentation of ridges. Consequent to this, the movement of Africa towards west and separation of Australia and Antarctica around 53 Ma aided the northward journey of the Indian Plate and its collision with Eurasia at 50 Ma [14].

CIR is the northern arm of the Indian Ocean Ridge System (IORS) and extends approximately from latitude 2°S to 25°30$′$S and longitude from 62 to 67°E. At the Indian Ocean Triple Junction, the CIR bifurcates into two additional ridges such as the South East Indian Ridge (SEIR) and South West Indian Ridge (SWIR) [15, 16]. The CIR is about 2550 km long linear chain of active submarine volcanoes and tectonic regions, often displaced by numerous NNE-SSW trending transform faults and fracture zones [17]. The CIR is a slow- to intermediate-spreading ridge and seems to have attained its present geometry during the plate reorganization as a result of the Indian subcontinent colliding with Asia [18]. The formation of CIR is estimated about 38 Ma [19, 20] or the Late Cretaceous [17]. The present configuration of the Indian Ocean was attained by 36 Ma, and almost the entire Indian Ocean basin is <80 Ma old [14]. The interplay of tectonic and magmatic processes along the CIR significantly controls the oceanic crust generation, its thickness, and evolution of the oceanic core complexes (OCC) [21–25]. The CIR shows a typical morphology of a slow- (to intermediate) spreading ridge, having an axial valley of 500 to 1000 m deep. Kamesh Raju et al. [26] presented a 750 km long bathymetric map of CIR and demarcated the magmatic and less magmatic sections associating them with the transform faults and the nontransform discontinuities, respectively, and also confirmed the presence of megamullion structures along the aseismic less magmatic lineations on the CIR ocean floor. The axial volcanic ridges are around 1-2 km wide and up to 15 km in length. The general depth of the ridge axis is around 3000 m; however, it increases up to 4000 m at the southernmost portion near the Rodrigues Triple Junction (RTJ) [8].

The NE-SW trending VMFZ (also known as Vema transform fault or fracture zone) right laterally offsets the NNW-SSE oriented CIR [27, 28]. The offset of the CIR has been variously estimated as 320 km [29], 300 km [30], and 240 km [31]. Detailed single and multibeam data found that the sea floor depth along the length of the VMFZ varies from 1500 to 6200 m [27, 28]. The principal transform displacement zone of the VMFZ consists of many closely spaced faults that can be traced for more than 150 km [32]. The mean full spreading rate of the CIR also varies between the ridge and segment located north of the VMFZ (3.6 cm/yr) and that to the south of the fracture zone (4.2 cm/yr). The segment north of the VMFZ displays intense deformation comprising alternating troughs (4800 m deep) and crests (2300 m). The average width of the CIR axial valley is about 15 km that varies from 12 km in the north to 17 km in the central part where it interacts with VMFZ [28]. The rock assemblage along VMFZ is very peculiar, consisting of both pillow and columnar basalts. The CIR is generally associated with smaller magma cells, displaying a characteristic low “bulls eye” Mantle Bouguer Anomaly (MBA) and gravity anomaly. This is in agreement with plume-like mantle upwelling and crustal density structure generally found along slow-spreading ridges [26]. The CIR segment to the north of VMFZ displays a well-defined, ridge-parallel topographic fabric. However, the southern segment appears to be undergoing a magmatic phase of accretion as understood from topographic fabric and rift valley morphology [26, 33]. Characterized, in general, by significantly low free air gravity anomalies, the VMFZ displays free air gravity anomalies to the order of 150 mGal [27]. It is maximum (200 mGal) at the steepest part of the transform fault [26, 33]. The VMFZ also shows a distinct seismicity pattern. It is characterized by low moment release (21.82$×$1025 dynes/cm) and low strain rate suggesting that a significant part of the plate motion in this segment is aseismic [34].

The sampled rocks for this study were collected during the cruise SK-195 of the ORV Sagar Kanya. A total of nine dredge operations were made for the sampling of rocks on the ridge and fracture zones. These successful dredges recovered a variety of rocks ranging from basalts, gabbros, serpentinized peridotites, ferromanganese encrusted rocks, and calcareous sediments. Studied serpentinized peridotites were recovered from two dredges targeted at the fracture zones (Figure 1).

The samples devoid of surface alteration were selected, and thin sections were prepared for optical microscopy. After petrographic studies, twelve (12) samples were selected for whole rock major oxides, trace elements and platinum group elements (PGE) analyses. Samples of around 1 kg were crushed and powdered to a definite size (~120 mesh) using agate mortar to avoid contamination or loss of homogeneity at all stages.

The mineral chemical composition of the Cr-spinel, olivine, and orthopyroxene grains along with the serpentines was determined with CAMECA SX Five Electron Probe Micro Analyser at CSIR-National Institute of Oceanography (NIO), Goa, India. The analyses were performed with an acceleration voltage of 15 kV and beam current of 10 nA. Both natural mineral and synthetic standards were used during calibration. The precision of the analysis is better than 1% for major oxides from the repeated analyses of standards. Representative mineral chemical data are given in supplementary tables (Tables S1–S4).

Major oxide compositions were determined using X-ray fluorescence (XRF) spectrometer at CSIR-National Institute of Oceanography (NIO), Goa, India. Fusion beads of the powdered samples were used for the determination of major oxides by XRF. PANalytical’s Axios WD, microprocessor controlled, sequential XRF covering the complete elemental range was used for the analyses of major oxides of representative rock samples. Certified reference materials BE-N and JP-1 were also analyzed along with the ultramafic rock samples to assess the accuracy of the data. Accuracy and precision are better than 5% RSD (% relative standard deviation).

Trace element compositions of representative whole rock samples were determined using High-Resolution Inductively Coupled Plasma Mass Spectrometer (HR-ICPMS) (Nu Instruments Attom®, UK) at CSIR-National Geophysical Research Institute (NGRI), Hyderabad, India. All rock samples were analyzed for a total of 34 trace elements including REE. A mixture of doubly distilled acids (HF : HNO₃) mixed in the ratio 7 : 3 was employed for the closed digestion method and the diluted solutions were analyzed by HR-ICPMS [35]. Certified reference materials UB-N (ANRT, France), PCC-1 (USGS, USA), and JP-1 (GSJ, Japan) [36] were used to validate the results obtained, and the precision for most of the elements was better than 5% RSD (% relative standard deviation).

Platinum group element concentrations were estimated by using the nickel sulphide fire-assay preconcentration method followed by High-Resolution Inductively Coupled Plasma Mass Spectrometer (HR-ICPMS) (Nu Instruments Attom®, UK) at CSIR-NGRI, Hyderabad, India. The analytical procedures of Balaram et al. [37] were followed during the digestion and analyses. Certified standards like WMG-1, WMS-1 (CANMET, Canada), and PTM-1 (IGGE, China) were used to check the quality of the analyses. The precision and reproducibility obtained for most of the platinum group of elements were better than 8% RSD.

The peridotites recovered from fracture zones of the Central Indian Ridge are highly altered and serpentinized wherein at most of the places primary mineralogy is overprinted by hydrothermal alteration. The primary mineralogy of these peridotites is depicted by olivine, pyroxene, and opaque minerals. The olivine grains are altered to the serpentine group of minerals showing fibrous nature (Figure 2(a)). This serpentinization took place along the grain boundaries and cross fractures of olivine grains due to infiltration of sea-water and late-stage percolation of hydrothermal fluids (Figure 2(b)). In few places, relict olivine grains are also present (Figure 2(c)). Orthopyroxene is the second dominant mineral in these rocks (Figure 2(d)). In most places, pyroxene grains are also serpentinized, but the degree of alteration is much lesser than that in olivine. Clinopyroxene is also present in minor amounts. In these rocks, two types of opaque oxides are present. These opaques occur as primary magmatic minerals and as inclusions within the main rock forming silicates. Chromite and chromium-bearing spinels are the most dominant opaque phases present in these rocks. In the high-resolution microscopic study, the chromium-bearing spinels sometimes show compositional zoning due to reaction failure at the time of their crystallization (Figure 2(e)). Apart from primary opaques, secondary Fe-rich opaques (~magnetites) are present, and these are products of alteration and formed mainly during late-stage hydrothermal activity (Figure 2(f)). These opaques occur along the secondary veins and grain boundary of olivine grains.

Serpentines from the studied samples are classified as antigorite and lizardite (Figure 3(a)). SiO₂ contents of these serpentines are almost similar to that of the other global serpentines and range from 34.73 to 44.93 wt.% (avg: 40.39 wt.%) and 26.04 to 45.49 wt.% (avg: 41.98 wt.%) for the mesh serpentines and vein serpentines, respectively (Table S1). MgO in both types ranges from 29.05 to 43.98 wt.% and 26.99 to 42.57 wt.%, respectively. Al₂O₃ content is also low (0.24 to 1.05 wt.%) for the vein serpentines, compared to that (0.08 to 3.69 wt.%) for the mesh serpentines. Antigorite is most stable at 680°C at 1.9 GPa, and the increase in Al and Cr in them expands the serpentinite stability by 60-70°C [38]. The higher Al and Cr contents of the antigorite of mesh serpentines relate to its greater stability in that temperature range. Average Mg# (Mg/Fe²⁺+Mg) values are 0.96 for the former and 0.95 for the latter. These values do not correlate with the bulk rock compositions and are much higher thereby suggesting early crystallization of the precursor minerals in a distinct pressure-temperature condition.

The Cr-spinels from the restitic mantle peridotites of CIR are characterized by 22.8 to 45.6 wt.% Cr₂O₃ with high MgO content (5.44 to 17.32 wt.%) (Table S2). The high Al₂O₃ content of spinels ranging from 18.7 to 34.8 wt.% suggests its aluminous nature. The wide variation of Cr# (Cr/Cr+Al; molar ratio), ranging between 0.37 and 0.59 (<0.60; except one 0.61) indicates their affinity towards oceanic crust [47] (Table S2). Mineral chemical compositions characterize the chrome-spinels as Al-chromite, picotite, and pleonaste (Figures 3(b) and 3(c)). The chrome-spinel for the CIR mantle peridotite samples reflects transitional character from depleted MORB to an enriched SSZ regime in terms of Mg# vs. Cr# (Figure 3(d)) and TiO₂ vs. Al₂O₃ (Figure 3(e)) variations. The studied chrome-spinels largely correspond to the overlapping domain of ocean floor and alpine-type peridotites on a cationic Fe-Cr-Al diagram (Figure 3(f)) thereby corroborating the role of a subduction-related arc component. This conjecture is substantiated by their prominent arc signatures as reflected by the Al₂O₃ vs. Cr₂O₃ (Figure 3(g)) plot.

The olivine and orthopyroxene grains of the studied samples are extensively serpentinized. In few places, olivine and orthopyroxene pseudomorphs occur as relict. Few relict olivine and orthopyroxene grains are found suitable for mineral chemical analyses (Tables S3 and S4). Olivine grains are highly magnesian (avg: Fo_~90) in nature. Orthopyroxene chemistry (Wo_4.7En_86.3Fs₉) shows enstatitic composition. Single pyroxene thermometry [48] (figure not shown) indicates an equilibration temperature of ~1000-1400°C, thereby indicating that the peritectic reactions of Cr-spinels to form orthopyroxenes started at a high temperature regime, probably at crust-mantle transition zone.

Major oxide compositions for the studied mantle peridotites (Table S5) reveal a restricted SiO₂ content ranging from 37.5 to 40.84 wt.% with elevated abundances of MgO (35.42-39.45 wt.%). Al₂O₃ (0.02-0.42 wt.%), TiO₂ (0.4-0.97 wt.%), CaO (0.05-0.35 wt.%), and MnO (0.07-0.21 wt.%) characteristically show low concentrations with moderate values for Fe₂O₃^(t) (7.32-10.75 wt.%). Mg# varies largely from 74 to 82, attributing to the Mg-loss phenomena. CaO, FeO, and Al₂O₃ variations with respect to MgO classify the studied samples as harzburgites (Figures 4(a)–4(c)). They are characterized by high concentrations of Ni (1160-2084 ppm) and Cr (2158-2990 ppm) (Table S5). Chondrite-normalized REE patterns (Figure 5(a)) exhibit a U-shaped profile marked by pronounced LREE and HREE enrichment over MREE complying with La/Sm_N: 1.47-3.30, Dy/Yb_N: 0.40-0.91, and Gd/Yb_N: 0.53-1.53 (Table S5). Positive chondrite-normalized Eu anomalies (Figure 5(a)) suggest a plagioclase free source. $LREE/MREE>1$ in conjunction with $MREE/HREE<1$ reflects boninitic signatures of these rocks which in turn conform to subduction-derived melt percolation in the depleted ridge mantle. The primitive mantle-normalized trace element abundances (Figure 5(b)) display marked positive anomalies at U and negative anomalies at Nb.

The CIR mantle peridotites display moderate to high concentrations of total PGE (26.14–113.40 ppb) in which Pt concentration ranges from 9 to 29.37 ppb, Pd varies between 4.95 and 27.48 ppb, and Rh shows a range of 0.24-1.11 ppb among PPGE (Table S5). Among IPGE, Ir, Os, and Ru concentrations range from 2.76 to 15.06 ppb, 0.35-4.78 ppb, and 3.39-15.37 ppb, respectively. PGE abundances of the studied samples are marked by PPGE (18.83-76.28 ppb) enrichment relative to IPGE (7.31-37.12) (Table S5). Chondrite-normalized PGE patterns of CIR mantle peridotites (Figure 5(c)) are marked by a relative enrichment in Ir, Ru, Pt, Pd, and Cu and depletion in Os and Rh. These rocks have Pd/Ir (0.47-3.26) and Pd/Pt (0.36-1.3) (Table S5) ratios higher than those of primitive mantle (Pd/Ir: 1.01; Pd/Pt: 0.53) and chondrite (Pd/Ir: 1.0; Pd/Pt: 0.53).

The diverse thermotectonic conditions imprinted on peridotitic rocks of oceanic setting result in redistribution of elements through time via various post magmatic processes including hydrothermal alterations, metamorphism and related deformation, late-stage intrusions, metasomatic alterations, and carbonation [52–55]. Hence, identification of the primary geochemical characters is a prerequisite to address the petrogenesis and geodynamic implications of mantle peridotitic rocks. Chemistry of primary minerals comprises one of the most reliable proxies to understand the genesis and parent melt characters. Several studies have demonstrated that major elements like Al, Ti, Fe, and P, high field strength elements (HFSEs), REEs (except Eu and Ce), and transition metals like Cr, Ni, Sc, V, and Y are relatively immobile during hydrothermal alteration and greenschist to amphibolite grade metamorphism [56, 57]. Bulk major oxide chemistry of the peridotitic protolith remains unchanged by serpentinization and low grade metamorphism [53]. Peltonen et al. [58] demonstrated that transition metals like Sc, V, Cr, Co, and Ni are concentrated in early crystallizing mineral phases in the magma and suffer minimal fractionation, and they preserve the protolith signature in serpentinized peridotites. In this view, primary mineral chemistry, some of the major oxides, HFSEs, transition metals, and REE chemistry of whole rocks are used to decipher the petrogenetic processes and geodynamic conditions for the CIR peridotites in this study. None of the elements is showing any linear relationship with increasing LOI content of the studied samples nullifying the possibility of their mobility with the post magmatic serpentinization processes. Bulk rock PGE chemistry has been evaluated to underpin the mantle processes and melt generation mechanisms since IPGEs (Os, Ir, and Ru) are refractory in their behavior, unaffected by post magmatic alterations, and overall PPGE/IPPGE is controlled by primary partial melting conditions, degree of melt percolation, and metasomatic processes in the mantle [59, 60].

Global geochemical correlations of MORB invoke a complex interplay of magmatic and tectonic processes including variations in mantle temperature, degree of melting, rate of magma supply, and oceanic plate divergence involved in the generation of oceanic lithosphere at midoceanic ridge-rift systems with preponderant control on the crustal thickness and axial depth of the magma chamber beneath the ridge [61, 62]. The ascent and subsequent cooling of melts, resultant of decompression melting of asthenospheric mantle upwelling below midoceanic ridges, form the oceanic crust broadly comprising 0.5 km thick MORB successively underlain by ~1.5 km of sheeted dykes and 4 to 5 km of cumulate gabbros with ultramafic rocks, while the residual melts form the depleted, refractory upper mantle [62–64]. The crustal and mantle sections of the oceanic lithosphere at midoceanic ridge-rift settings record different stages of elemental depletion or enrichment of subridge mantle through melt extraction and mantle refertilization mechanisms, respectively [65]. Mantle peridotites are generally envisaged to represent solid residues left after variable extents of melt extraction during oceanic crust formation. Consequently, the compositional variations reflected by these mantle relicts constrain the geodynamic conditions, magmatic episodes, partial melting, melt migration, and melt-rock interactions [66, 67]. Likewise, the peridotitic upper mantle section uplifted and exposed at Vema and Vityaz fractures zones of CIR provide a window to probe into the chemical heterogeneity of the Indian Ocean subridge upper mantle. The LILE, HFSE, and REE geochemistry reflects a deviation from typical, depleted N-MORB composition to an enriched intraoceanic forearc mantle further substantiating melt transfer from enriched asthenosphere to depleted lithospheric mantle and refertilization of the upper mantle peridotite by reactive melt percolation and melt-peridotite interactions [64, 68]. The compositional traits and elemental variations for the chrome-spinel from the CIR peridotites correspond to the overlapping domains of abyssal MOR peridotites and oceanic arc-related SSZ peridotites. Cr# ranging from 0.38 to 0.59 (except one data) for the chromian spinels in the studied abyssal peridotites from CIR classifies them as type I alpine-type peridotites thereby fingerprinting an evolving mantle character from MORB to a boninitic compositional trend [47]. Further, these mantle harzburgites with chromian spinels having $Cr#>30$ conform to melt infiltration and melt-peridotite interaction [69]. The chrome-spinel chemistry (Figures 3(e) and 3(f)) for the studied samples implies a compositional transition from depleted MOR to a hydrated and enriched SSZ regime. The SSZ affinity for the studied samples can be equated with an intraoceanic forearc mantle wedge developed in response to incipient subduction. Multiple phases of melting, melt percolation, impregnation, and mantle refertilization under extensional conditions are potential indicators for a tectonic transition from ridge-rift to spontaneous subduction [70]. The mantle harzburgites exposed at Vema and Vityaz fracture zones of CIR record multiple stages of melting and refertilization as implicated by (i) ridge-associated melting rendering depleted upper mantle harzburgites and (ii) fluid-fluxed melting of juvenile forearc mantle triggering melt impregnation, melt-peridotite reaction, and refertilization of refractory peridotites. The evolved and enriched composition of oceanic mantle peridotites and their deviation from a typical N-MORB type DMM chemistry can be ascribed to (i) infiltration of plume-derived melts of OIB chemistry into a MOR magma chamber inducing incompatible trace element enrichment in the case of the proximity of ridge segments to plumes, (ii) subduction and recycling of OIB-like pyroxenite, (iii) refertilization of depleted subridge mantle by impregnation of subducted slab fluids and melts in the case of distal relation between plume and ridge at the onset of spontaneous subduction at ridge transform faults, and (iv) melt replenishment in upwelling heterogeneous mantle by melting of residual eclogite derived from recycled oceanic crust [64, 68, 71, 72]. The CIR peridotites in this study yield unequivocal evidence for the compositional inhomogeneity of the Indian Ocean subridge mantle marked by interaction between depleted and enriched melt components. The absence of primitive mantle-normalized positive Nb anomalies coupled with extreme LREE enrichment over MREE and HREE discards the possibility of infiltration of OIB-type mantle components carried by plume-derived melts or recycled subducted slab materials into the refractory mantle beneath CIR. Rather, selective abundance of fluid-mobile elements (FME), HFSE depletion, and U-shaped chondrite-normalized REE patterns corroborate input from subduction-derived fluids and melts. The Nb/Zr vs. Th/Zr plot (Figure 6(a)) distinctly substantiates the fluid-related enrichment for the studied samples.

Higher values of Zr/Hf (avg: 45.16) and Zr/Sm (avg: 73.78) in addition to lower Nb/Ta (avg: 4.75) with respect to primitive mantle compositions (Zr/Hf: 36.25, Zr/Sm: 25.23, and Nb/Ta: 17.39) conform to a metasomatized, enriched mantle signature. Lower Nb/U (avg: 0.19) values with higher Ba/Th (avg: 213.16) and Ba/Nb (avg: 316.27), compared to OIB (Nb/U: 47.06, Ba/Th: 87.5, and Ba/Nb: 7.29) and N-MORB (Nb/U: 49.57, Ba/Th: 52.5, and Ba/Nb: 2.7), attest to contributions from subduction-derived fluids. The role of a recycled arc mantle could have been propounded as a key factor accounting for the enriched geochemical features of CIR mantle peridotites. However, the boninitic compositions of these rocks reflected on the V vs. Ti plot (Figure 6(b)) attest to melt input via melting of a juvenile, metasomatized forearc mantle during initiation of intraoceanic subduction. In general, boninite melt generation marks the compositional transition from anhydrous, depleted to hydrous, enriched mantle triggered by progressive influx of slab-derived fluids and melts during the nascent stage of oceanic subduction. The studied samples preserve ubiquitous signatures of boninitic melt-harzburgite interactions subsequent to spontaneous subduction initiation, generation, and infiltration of boninitic melts into the depleted mantle. The CIR mantle harzburgites reflect a geochemical and geodynamic transition from MOR to a SSZ regime developed in response to incipient subduction, and these mantle units constitute a geochemical record of mantle refertilization beneath CIR and melt-harzburgite interaction in a juvenile subduction-modified mantle. PGE chemistry marked by an abundance of PPGE over IPGE attests to infiltration of metasomatized, subduction-processed enriched components into depleted upper mantle. Pd/Ir vs. Ni/Cu variations for the studied samples (Figure 6(c)) conform to a mantle origin, while the Pd vs. Cu plot (Figure 6(d)) distinctly corroborates a sulphur undersaturated boninitic affinity. (Pt/Ir)_N vs. Cu (Figure 6(e)) variations display a melt percolation trend for the studied samples. Thus, these are infiltrated mantle peridotites pervasively impregnated by boninitic melts derived through juvenile forearc mantle melting in response to spontaneous subduction initiation at submarine transform faults and fracture zones in CIR. Subduction initiation marks the first descent of oceanic lithosphere into the upper mantle prior to the inception of self-sustaining subduction. One of the primary triggers for this phenomenon comprises spontaneous sinking or gravitational collapse of older and denser oceanic lithosphere at transform faults and fracture zones [78–81]. Submarine transform faults and fracture zones are sites of pre-existing weakness or mechanical instability. Intraoceanic subduction zones can initiate along submarine transform faults and fracture zones and extensional detachment faults associated with slow- to ultraslow-spreading ridges [78, 82, 83]. Serpentinized oceanic detachments at midoceanic ridge-rift systems owing to their weak and fragile nature develop favorable conditions conducive for nucleation of incipient, spontaneous subduction parallel or close to spreading ridges [84]. This study infers the subduction initiation and prevalence of a juvenile, metasomatized forearc mantle associated with Vema and Vityaz fracture zones of CIR. This phenomenon generates boninitic melts that subsequently impregnate the depleted mantle peridotites. The Th_N vs. Nb_N plot (Figure 6(f)) substantiates the forearc affinity of the studied samples thereby corroborating the inception of spontaneous subduction beneath CIR. The serpentinized harzburgitic mantle sections exhumed at the transform fault sections of the CIR are potential archives of diverse magmatic features testifying to an evolving mantle composition from MORB to boninitic chemistry comparable to those developed in infant intraoceanic arc systems in response to multiple phases of fluid and melt impregnation associated with flux melting at initial stages of subduction [85, 86].

Mantle sections of oceanic lithosphere at midoceanic ridge-rift settings are characteristically restitic or refractory in nature owing to multiple episodes of melt extraction and consequent impoverishment in incompatible trace elements. Thus, refractory mantle peridotites from which oceanic crust has been extracted at MOR exhibit an overall depleted chemistry with respect to primitive mantle and chondrite compositions [4, 87, 88]. However, MOR lithologies showing a deviation from this conventional depleted trend account for the geochemical heterogeneity of suboceanic mantle. Mantle refertilization and melt replenishment by subduction-derived melt percolation, plume-derived melt input, and episodic recycling of oceanic and continental crust are principal proponents for the chemical diversity of the upper mantle [88, 89]. These thermotectonic conditions render inevitable imprint on the mineralogical and whole rock geochemical compositions of oceanic crust and mantle lithologies [9, 68, 88, 90]. The unambiguous influence of mantle plume and recycled and stagnant subducted slabs on the composition of ridge mantle beneath oceanic spreading centers across the globe has been widely discussed [91]. One of the distinguishing features of the mantle peridotites exposed at the fracture zones of CIR, as inferred from the study, is the transitional geochemical and tectonic spectrum from depleted MOR to enriched SSZ. The distinct boninitic composition featured by the U-shaped chondrite-normalized REE patterns (Figure 7) for the studied CIR mantle harzburgites is not discernible for their crustal counterparts which exhibit typical LREE depleted MORB trend (Figure 7). This boninitic signature of the studied rocks, their forearc-related SSZ tectonic affinity (Figure 3(e)), and sulphur undersaturated nature (Figure 6(d)) as comprehensively revealed by mineral chemical compositions, whole rock, and PGE chemistry cannot be correlated with mantle plume activity or contributions from recycled/stagnant oceanic slab. Rather, the geochemical features inferred for the CIR peridotites can be validated in terms of a gradual transition in a magmatic and tectonic regime from decompression melting of a depleted lherzolitic mantle to fluid fluxed hydrous melting of fertile harzburgitic mantle in response to inception of intraoceanic subduction. This progressive change is reflected in the melt generation process which is governed by combined effects of melt depletion and enrichment; mantle metasomatism and refertilization; and pervasive melt impregnation and melt-rock interactions. The favorable conditions for the generation of boninitic magmas include subduction of active spreading ridge or a young oceanic slab and initiation of spontaneous subduction along transform fault in an active ridge transform system. The depleted nature of the precursor mantle and unusual melting conditions involving elevated temperature and volatile influx at shallow depth substantiate the transitional melt depletion and reenrichment involved in boninite genesis [92, 93]. The dynamics of subduction initiation invokes two contrasting scenarios, where an older, denser, negatively buoyant oceanic lithosphere (at least 10 Ma old) subducts spontaneously into the underlying asthenosphere while the descent of younger (<1 Ma age), positively buoyant oceanic lithosphere requires external forces to initiate forced subduction [82]. A young, buoyant oceanic plate near a divergent boundary can only be forcefully subducted, whereas self-sustained, spontaneous subduction initiation is a more common phenomenon at mechanically weak zones associated with older and denser oceanic lithosphere. The magmatic response to this tectonic transition is imprinted within the lower oceanic crust and mantle sections in terms of elemental concentrations modified through pervasive melt-peridotite interactions. Thus in this case, spontaneous subduction initiation stands as the plausible proponent for the observed magmatic features of the relatively older and denser oceanic lithosphere existing away from the ridge axis at CIR (Figure 8). In coherence with this, the boninitization of CIR mantle peridotites can be interpreted to be linked with spontaneous subduction initiation at the fracture systems of CIR induced by gravitational collapse at the zones of lithospheric weakness and mechanical instability. Therefore, the mineral chemical, trace, and platinum group elemental signatures of the CIR mantle harzburgites collectively envisage the following: (i) replenishment of depleted subridge upper mantle by enriched subduction-derived melts and fluids indicates the role of a recycled arc mantle component or a tectonic transition from midoceanic ridge-rift to an embryonic suprasubduction zone, (ii) boninitic melt percolation underpins an incipient stage of oceanic subduction (Figure 8), and (iii) network of submarine transform faults and fracture zones along slow-spreading CIR owing to their weakness and mechanical instability develop favorable conditions to serve as potential precursors for initiation of spontaneous subduction (Figure 8) beneath Vema and Vityaz fracture zones of CIR.

• (i)

  Peridotites exposed at the Vema and Vityaz fracture zones of the Central Indian ridge-rift system represent a section of the Indian Ocean ridge mantle

• (ii)

  The geochemical attributes of the CIR peridotites mark a shift from MORB to arc signature that cannot be solely equated with the effects of serpentinization and hydrothermal alteration

• (iii)

  The absence of primitive mantle-normalized positive HFSE anomalies, $LREE>MREE<HREE$⁠, and low Zr/Nb values collectively discarded the possibility of infiltration of low-degree plume-derived melts, OIB-type components, or recycled subducted slab materials and substantiate percolation of subduction-derived fluids and melts into the refractory mantle beneath the CIR

• (iv)

  The geochemical heterogeneities can be construed in terms of refertilization of depleted mantle by boninitic melts generated through flux melting in response to spontaneous subduction initiation beneath CIR

• (v)

  Chrome spinel chemistry in conjunction with trace and PGE chemistry of the studied peridotite reconciles the phenomenon of boninitization of upper mantle beneath CIR and pervasive melt-peridotite interactions

• (vi)

  The fracture zones associated with CIR serve as a potential site for spontaneous subduction initiation owing to the weakness and mechanical instability of older, denser, and negatively buoyant oceanic lithosphere

All data have been provided in Supplementary Materials.

The authors do not report any conflicts of interest regarding the contents of this paper.

The authors are grateful to the Director of CSIR-National Institute of Oceanography for his encouragement, support, and permission to publish this work. The authors sincerely acknowledge all participating scientists and crew members of SK-195. The authors are thankful to Drs. K.A. Kameshraju, R. Mukhopadhyay, and S.D. Iyer for sharing the samples. The authors sincerely acknowledge the facilities received from Central Analytical Facility, CSIR-NIO. AS acknowledges the funding received from the Council of Scientific and Industrial Research (CSIR) and Ministry of Earth Sciences (MoES), Government of India, under MLP 2017 and GAP 2175, respectively. AH acknowledges CSIR for Senior Research Fellowship (Grant no. RF83931). This work is a part of Ph.D. of AH. This is NIO’s contribution no. 6779.