Zircon Xenocrysts from the Shaka Ridge Record Ancient Continental Crust: New U-Pb Geochronological and Oxygen Isotopic Data

Over the past two decades, a significant number of discoveries of ancient zircon xenocrysts in ocean-floor magmatic rocks have been reported. These findings provide compelling evidence for the presence of ancient continental crust within young oceanic lithosphere. Almost all finds of ancient zircon xenocrysts within oceanic crust are from the Mid-Atlantic Ridge. For other localities, however, similar data are very limited. This report presents the first age determinations (U-Pb, SHRIMP-II) and isotope-geochemical data (oxygen, trace and rare earth elements) for zircon xenocrysts from gabbro-diorites of the Shaka Ridge, in the vicinity of the Shaka fracture zone, near the western end of the Southwest Indian Ridge. The work is based on a study of bottom rock material dredged during expeditionary research on the R/V “Akademik Fedorov” (Russia) in 2016. The U-Pb isotope system of the zircon xenocrysts recorded a crystallization age of ∼2.8 Ga at an upper discordia intercept and an age of ∼600 Ma interpreted as the timing of a superimposed thermal event at a lower discordia intercept. The zircon xenocrysts show geochemical signatures of magmatic origin, i.e., fractionated REE distribution spectra with an increase in chondrite-normalized values from light to heavy REE, positive Ce anomalies and negative Eu anomalies, and high Th/U ratios (0.59–7.77). In discrimination diagrams based on a series of inter-element relationships, zircon compositions fall into the fields of zircons from rocks of continental crust, mostly granitoids. The Li content of the zircons is high (1.8 ppm–50 ppm), adding further evidence to their derivation from rocks of continental crust. During their residence within young oceanic crust, the zircon xenocrysts experienced alterations under the influence of submarine high-temperature hydrothermal fluids, which selectively affected the distribution of trace elements in the zircons and reduced the δ18O values to 1.75‰–3.15‰. The presence of obviously older zircons in Shaka gabbro-diorites clearly demonstrates the presence of ancient continental fragments and their recycling into the mantle at the western end of the Southwest Indian Ridge.


INTRODUCTION
The finding of ancient zircons (Paleozoic to Proterozoic) in oceanic gabbros from the Mid-Atlantic Ridge near the Kane fracture zone (Pilot et al., 1998) triggered extensive discussion that is still under way. The primary issue is the mechanism of incorporation of these zircons into young oceanic crust. Initially, attempts were made to explain the presence of obviously older zircons either by contamination during sampling of rocks or separation of minerals. However, a significant number of such discoveries (about twenty publications, links below), and a wide range of laboratories in which sample preparation was performed, suggest that this phenomenon is not uncommon, and zircons should be considered as true xenocrysts (Bea et al., 2020). The results of these studies provide convincing evidence for the presence of ancient continental crust in young oceanic lithosphere (e.g., Bortnikov et al., 2019;Ashwal et al., 2017;Kamenetsky et al., 2001). Sergey G. Skublov, Natalia A. Rumyantseva, Qiuli Li, Boris G. Vanshtein, Dmitriy I. Rezvukhin and Xianhua Li Zircon is a unique indicator and a common representative of ancient continental lithosphere. It retains isotope-geochemical signatures and records information on the reworking of ancient crust during various geodynamic processes (Leontev et al., 2020;Zhao et al., 2019;Xu and Zhang, 2018). Almost all finds of ancient zircons within oceanic crust come from the Mid-Atlantic Ridge (Bea et al., 2020;Bortnikov et al., 2019Bortnikov et al., , 2008Shulyatin et al., 2019;Kostitsyn et al., 2018Kostitsyn et al., , 2015Kostitsyn et al., , 2009Kremenetskiy et al., 2018;Aranovich et al., 2013;Skolotnev et al., 2010;Sharkov et al., 2004;Pilot et al., 1998;Belyatskii et al., 1997). This is probably due to the fact that this area of the world ocean has been systematically studied for several decades. For the Indian Ocean, however, the available data are scarce (Ashwal et al., 2017;Cheng et al., 2016;Torsvik et al., 2013). In this contribution we report the first age determinations and isotope-geochemical data for zircons from the Shaka Ridge, adjacent to the Shaka fracture zone, near the western end of the Southwest Indian Ridge. We provide evidence for the continental derivation of the zircons and discuss possible mechanisms of their transportation into young oceanic crust.

GEOLOGICAL BACKGROUND AND SAMPLING
The Southwest Indian Ridge (SWIR) is a Meso-Cenozoic ocean floor uplift. Its formation began as a result of the Gondwana supercontinent breakup in the Mesozoic (Norton and Sclater, 1979). The ridge extends~8 000 km from the Bouvet triple junction in the South Atlantic to the Rodrigues triple junction in the southern Indian Ocean. The SWIR is an ultraslow spreading ridge with spreading rates of only 14-18 mm/ yr. Bouvet Island, a hotspot, is located~300 km east of the Bouvet triple junction and 55 km south from the nearest SWIR segment (Lin et al., 2003;Georgen et al., 2001). Within the SWIR, there is an intensely dissected topography with elevation differences of up to 4 km or more as well as a significant number of transform faults having their own names. One of such transform faults is the Shaka fault. The namesake Shaka Ridge is located on the northeastern flank of the latter; it is an underwater upland extending to the east. The Shaka Ridge is thus not a part of the SWIR, but is located in its immediate vicinity ( Fig. 1).
Currently, there is a lack of reliable information about the origin and formation of the Shaka Ridge (Rumyantseva et al., 2021). The first detailed geological and geophysical study of the Shaka Ridge area was carried out by the staff of the Woods Hole Oceanographic Institution (WHOI) during December and January, 2001. This research was part of the geophysical mapping and dredging program of the southwestern part of the Indian Ridge between E9°and E16° (Lin et al., 2003. As a result of these studies, pronounced mantle heterogeneity of the area was established, which is expressed in differing indicators of the gravitational field. The largest negative mantle anomaly (-50 mGal) was recorded in the area of a number of relict volcanic cones at the junction of the Shaka fault and the Shaka Ridge (S52.25°; E10.5°). Dredging of the cones showed that they were composed of alternating layers of ash and lapilli tuff. This evidence confirmed the researchers' assumption that the Bouvet hotspot interacted with this area at~20 Ma . The magnetic age of Shaka Ridge rocks is between 35-53 Ma (Sauter and Cannat, 2010;Fisher and Sclater, 1983).
Our work is based on a study of bottom rock material dredged during expeditionary research within the Shaka Ridge on the R/V "Akademik Fedorov" (Russia) in the spring of 2016. The research was performed in two stages. At the first stage, a bathymetric survey of seabed topography using a multibeam echo sounder was carried out. Based on the results, a geomorphological map was prepared, and locations for dredging stations were selected. In geomorphological terms, the study area has a substantial topographic relief with a major northeaststriking fault and individual volcanic edifices. At the second stage,~696 kg of bottom rock materials were sampled at the selected locations using a bottom dredge sampler. The dredged material represents rock fragments of different composition, from sedimentary siltstones to basaltoids. All these samples are weakly rounded or angular with some being covered by Fe-Mn  Zircon Xenocrysts from the Shaka Ridge Record Ancient Continental Crust: New U-Pb Geochronological crusts. After lifting the dredge aboard the vessel, the bottom rock materials waere extracted from the dredge and sorted according to rock types for further geological description, photographing and weighing. Among the bottom rock materials sampled from dredging station FB-2016-3-2 on the northeastern flank of the middle part of the Shaka Ridge (S51.057 8°; E13.164 9°), gabbro, diorite, dolerite, basaltic andesite, trachybasalt and gabbro-diorite are the predominant lithologies. A representative sample of gabbro-diorite from this station (FB-2016-3-2-55) was taken for a further analytical study.

ANALYTICAL METHODS
Zircon grains were recovered from the gabbro-diorite (sample FB-2016-3-2-55) by conventional magnetic and heavy liquid techniques. Once separated, the zircon grains were mounted in epoxy together with the TEMORA-2 (Black et al., 2004) and 91500 (Wiedenbeck et al., 1995) zircon standards. Spots for in-situ analyses were selected by using both transmittedand reflected-light images in order to avoid cracks and inclusions. Cathodoluminescence (CL) and back-scattered electron (BSE) images were obtained at the Centre for Isotopic Research of the All-Russian Geological Research Institute (St. Petersburg) using a CamScan MX2500S scanning electron microscope (SEM) coupled with a QLI/QUA2 CL and operated at 12 kV, 5-7 nA and a working distance of 31-33 mm.
The U-Pb dating of zircons was performed on a SHRIMP-II ion microprobe at the Centre of Isotopic Research of the All-Russian Geological Research Institute. The U-Pb measurements were carried out by the technique described in Bröcker et al. (2014), Rodionov et al. (2012) and Williams (1998). The intensity of primary O 2 beam was 4 nA, and the spot (crater) was 20 μm across. The obtained data were processed using the SQUID program (Ludwig, 2001). The U/Pb ratios were normalized to the value of the TEMORA zircon standard, 0.066 8, corresponding to an age of 416.75 Ma (Black et al., 2004). The errors of single analyses (U/Pb ratios and ages) were at the ±1σ level, and the errors of calculated concordant ages and intercepts with concordia were at the 2σ level. Error in standard calibration was 0.36%. The concordia plots were constructed using the ISOPLOT/EX program (Ludwig, 2003).
Oxygen isotope compositions of zircon were determined at the Institute of Geology and Geophysics of the Chinese Academy of Sciences (Beijing) on a Cameca IMS-1280 ion microprobe. Analytical procedures are similar to those described in Tang et al. (2015) and Li et al. (2010). The Cs + primary beam was accelerated at 10 kV with an intensity of ca. 2 nA. The spot size was about 20 µm in diameter (10 µm beam diameter + 10 µm raster). The measured 18 O/ 16 O ratios were normalized to the standard VSMOW ( 18 O/ 16 O = 0.002 005 2). We performed oxygen isotope measurements at spots located directly adjacent to pits after U-Pb isotope analyses. Before measuring, the sample was re-polished to remove the pits from previous analyses. The instrumental mass fractionation (IMF) was corrected using the TEMORA-2 zircon standard. The second reference zircon 91500 was analyzed as an unknown to ascertain the veracity of the IMF. Four measurements of the 91500 zircon during the course of the present study yielded an average 18 O value of 10.07‰ ± 0.08‰ (2SD). It is in agreement with the value of 9.9‰ for the 91500 zircon standard (Wiedenbeck et al., 2004).
Measurements of zircon trace element composition were performed using a Cameca IMS-4f ion microprobe at the Yaroslavl' Branch of the Institute of Physics and Technology of the Russian Academy of Sciences. We mainly followed the analytical procedure described in Dokukina et al. (2014), Fedotova et al. (2008) and Hinton and Upton (1991). The primary O 2ion beam spot size was~20 µm. Each analysis was averaged from 3 measurement cycles. Concentrations of trace elements were calculated from the normalized to 30 Si + secondary ion intensities using calibration curves based on a set of reference glasses (Jochum et al., 2006(Jochum et al., , 2000. NIST-610 reference glass (Rocholl et al., 1997) was used as a daily monitor for trace element analyses. Accuracy of trace element measurements was up to 10% and 20% for concentrations of more than 1 ppm and between 0.1 ppm and 1 ppm, respectively. To construct REE distribution spectra, the composition of zircon was normalized to that of chondrite CI (McDonough and Sun, 1995). The zircon crystallization temperature was determined by the Ti-in-Zrn thermometer (Watson et al., 2006).

Petrogeochemical Features of the Host Rock
On macroscopic examination, sample F-B-2016-3-2-55 is a rounded, slightly angular rock fragment weighing 8.5 kg and 18 × 17 × 13 cm 3 in size. It is a coarse-grained crystalline rock with a massive structure. Sample petrography is shown in Fig.  2. Rock-forming minerals are plagioclase (An 45-50 ; 55%; 0.6-4.0 mm in size), hornblende (25%; 0.2-1.5 mm), quartz (10%; 0.4-0.6 mm), K-feldspar (5%;~0.2 mm), and biotite (5%; 0.4-1.0 mm). Accessory minerals are zircon, apatite and titanite with epidote and chlorite occurring as secondary minerals. As is evident from its mineralogy and degree of alteration, the rock appears to have undergone greenschist-facies metamorphism. The structure of groundmass is hypidiomorphic granular with poikilitic domains. Plagioclase is saussuritized and crosscut by thin chlorite veins; it is also densely populated by inclusions of apatite, to a lesser extent biotite and hornblende. Hornblende forms irregularly-distributed elongated crystals with inclusions of biotite. Quartz is mostly xenomorphic and fills interstices between plagioclase crystals and colored minerals, although irregular-shaped individual grains are recognized as well. Biotite is present as individual laths, intergrowths and inclusions in principle rock-forming minerals. Zircon appears as irregular-shaped grains or prismatic crystals showing square cross-sections.

Zircon Characterization
The zircon grains separated from the gabbro-diorite are elongated (with aspect ratios 1 : 2-1 : 3) or isometric with pitted and corroded surfaces (Fig. 3). The grain size generally does not exceed 150-200 µm. In CL imaging, the zircon population is generally dark with shades from almost black to dark gray. Occasionally, individual domains bright in CL are recognized in both the interior and rim parts of the grains (e.g., grain 6.1, Fig. 3). Some examples show sector-zoned CL responses (e. g., grain 3.1, Fig. 3), where domains with different internal textures randomly alternate with each other. Fine-scale oscillatory zoning is revealed in several zircons (e.g., grain 7.1, Fig.  3), which is commonly considered as being a magmatic signature (e.g., Corfu et al., 2003).
All 15 dated zircon grains plot on a discordia line with an upper intercept with a concordia curve at 2 810 ± 4 Ma and a lower intercept at 590 ± 83 Ma (MSWD = 1.5; Fig. 4). With the exception of three spots (6.1, 14.1 and 11.1 with discordance values of 9%, 15%, and 20% respectively), all other points are clustered in the region of the upper discordia intercept and are considered subconcordant with the values of discordance not greater than 6% (Table 1).  The content of U in the zircon suite ranges from 280 ppm to 2 093 ppm (with an average of 860 ppm). It should be noted that the three most discordant zircons (spots 6.1, 14.1 and 11.1) show U abundances (318 ppm to 499 ppm) not exceeding those established in the subconcordant zircons. Thorium varies in a wider range than U (170 ppm-5 857 ppm, with an average of 1 324 ppm). The Th contents of the three most discordant grains are also far from the established maximum values and do not exceed the average value, varying from 277 ppm to 381 ppm. The Th/U ratios in the examined zircons are between 0.59 and 7.77, averaging 1.51.
The isotopic composition of oxygen in zircons is characterized by relatively low values of δ 18 O from 1.75‰ to 3.15‰, averaging 2.71‰ (Table 1).
The total abundance of REE in the studied grains varies from 960 ppm to 4 815 ppm, with an average of 2 591 ppm (Lu N /La N ratio averages 4 659; Table 2). In general, the REE distribution patterns are fractionated with a gradual increase in chondrite-normalized values from light to heavy REE (Fig. 5). Note that the REE patterns exhibit a very similar shape in the HREE region, differing only in the overall HREE content. The Lu N /Gd N ratio characterizing the degree of HREE slope is fairly consistent and varies from 14.5 to 28.6 with an average value of 18.8. By contrast, the concentrations of LREE show much more variability. The total LREE content varies from 23 ppm to 252 ppm, with an average of 75 ppm. In zircons with the highest LREE content (spots 4.1 and 7.1), the REE distribution spectra in the LREE region show a flattened (unfractionated)  The REE patterns of the zircons exhibit positive Ce anomalies and negative Eu anomalies. Note that a negative correlation is observed between the values of a positive Ce anomaly (Ce/Ce* ratio) and the total LREE abundances (r = -0.5). At spot 4.1 having the highest LREE content of 252 ppm, there is virtually no positive Ce anomaly (Ce/Ce* = 1.15). By contrast, in zircons showing low LREE levels, the positive Ce anomaly is most pronounced. For instance, spot 13.1 with the total LREE concentration of only 23 ppm shows the most prominent positive Ce anomaly (Ce/Ce* = 50.2).
The magnitude of a negative Eu anomaly varies from 0.01 to 0.29 with an average Eu/Eu* value of 0.09. In zircons with a high content of LREE and flat LREE distribution patterns, small negative Eu anomalies are observed, whereas zircons with a fractionated distribution of both REE (total) and LREE demonstrate well-developed negative Eu anomalies. Hence, a positive correlation is observed between the values of the posi-

10
Zircon Xenocrysts from the Shaka Ridge Record Ancient Continental Crust: New U-Pb Geochronological tive Ce and negative Eu anomalies (r = 0.57); i.e., zircons with a well-developed positive Ce anomaly also demonstrate a prominent negative Eu anomaly.
Hafnium abundances in the analyzed zircons are rather high and vary in a narrow range from 9 355 ppm to 11 751 ppm, with an average of 10 491 ppm. The content of Y ranges from 1 418 ppm to 8 483 ppm (with an average of 4 110 ppm) and shows a strong positive correlation with the total HREE content (r = 0.99). The concentration of P, varying from 259 ppm to 1 022 ppm, positively correlates with Y (r = 0.92) and HREE (r = 0.88) levels, which is consistent with the xenotimetype substitution (Y, REE) 3+ + P 5+ = Zr 4+ + Si 4+ (Finch and Hanchar, 2003).
Some zircon grains display increased abundances of nonformula elements. Calcium content varies from 12 ppm to 185 ppm, averaging 57 ppm. An increased Ca is characteristic of zircons with small positive Ce and negative Eu anomalies as well as with higher LREE content (r = 0.64). Strontium varies from 0.9 ppm to 5.5 ppm, with an average of 1.9 ppm. Ti content is between 14 ppm and 54 ppm, with an average of 24 ppm, which corresponds to a crystallization temperature range of 774-917°C (average value of 821°C) using the equation of Watson et al. (2006). Grimes et al. (2015Grimes et al. ( , 2009Grimes et al. ( , 2007 demonstrated that the origin of zircons that crystallized from melts in continental and oceanic settings can be revealed based on their Yb, U, Y and Hf abundances. The examined zircon xenocrysts from gabbrodiorites of the Shaka Ridge (sample F-B-2016-3-2-55) have been plotted in the discrimination diagrams with outlined fields for oceanic and continental zircons (Fig. 6). In the Y vs. U/Yb binary plot (Fig. 6a), all the zircon compositions fall within the continental field. About half of the grains are located within the compositional area of zircons from continental granitoids, identified according to Ballard et al. (2002) and Belousova et al. (2006). In the Yb vs. U (Fig. 6b) and Hf vs. U/Yb (Fig. 6c) discrimination diagrams, all the zircon points plot into the area of zircons of continental origin as well. In the discrimination diagrams proposed by Belousova et al. (2002), the zircon population belongs to the compositional field of zircons from granitoids (in Yb/Sm vs. Y plot, Fig. 7a), or to the field of granitoids and partially to the field of zircons from syenites and associated pegmatites (in Ce/Ce* vs. Y plot, Fig. 7b).

DISCUSSION
Another effective criterion for the discrimination of zircon origin is the concentration of Li (Bouvier et al., 2012;Ushikubo et al., 2008). Zircons from rocks of oceanic crust are characterized by an extremely low Li content (less than 0.01 ppm), while continental crust zircons, as a rule, show much higher Li abundances in the range of 1 ppm-100 ppm (Grimes et al., 2011;Ushikubo et al., 2008). In the studied zircons, Li varies from 1.8 ppm to 50 ppm with an average of 16 ppm (Table 2), which makes it possible to unambiguously assign our samples to zircons of continental crust.
In the La vs. Sm N /La N space (Fig. 8) with the outlined compositional fields of unaltered magmatic, hydrothermal, and porous zircons (Bouvier et al., 2012;Fu et al., 2009;Grimes et al., 2009;Hoskin, 2005), the studied zircon xenocrysts com-prise a single trend with a clear negative correlation between these two geochemical parameters. Of these, eight spots (1.1,  2.1, 5.1, 9.1, 10.1, 11.1, 13.1, 15.1) unambiguously plot into the field of unaltered magmatic zircons, while six spots are located either within the overlap between the field of unaltered magmatic zircons and the field of porous zircons (spots 3.1, 6.1, 14.1) or in the area of porous zircons (spots 7.1, 8.1, 12.1). One grain with the highest La content (spot 4.1, Table 2) is located at the border between the fields of porous and hydrother- Sergey G. Skublov, Natalia A. Rumyantseva, Qiuli Li, Boris G. Vanshtein, Dmitriy I. Rezvukhin and Xianhua Li mal zircons. Taking into account that porous zircons are considered by some authors as a variety of hydrothermal zircons (Fu et al., 2009;Kirkland et al., 2009), the proportion of altered zircons among the xenocrysts may be more significant.
Therefore, a series of crucial geochemical signatures of the zircons from gabbro-diorites of the Shaka Ridge region (fractionated REE distribution patterns with a positive slope in chondrite-normalized values from light to heavy REE, positive Ce anomalies and negative Eu anomalies, high Th/U ratios, trace-element geochemistry) indicate the zircons to be of magmatic origin and have been derived from rocks of continental crust, mostly from granitoids (e. g., Balashov and Skublov, 2011;Hoskin and Schaltegger, 2003). These zircons were inherited from much older rocks (~2.8 Ga) with respect to the gabbro-diorite and thus represent genuine xenocrysts. Before being transported into young oceanic crust, the zircons had undergone a thermal influence (probably, metamorphism) with an age of~600 Ma recorded by a lower discordia intercept (Fig.  4). The increased LREE contents and the flattened normalized LREE distribution spectra in some zircon grains are explained by that metamorphic fluids perhaps exerted some effect on the geochemical characteristics of the zircons during this thermal event. The fluids also increased the Ca concentration in the zircons; this signature of fluid impact on zircon has been reported previously (e.g., Geisler and Schleicher, 2000).
Alternatively, such a marked increase in Ca and LREE concentrations in some zircon grains could have been caused by high-temperature hydrothermal alterations within young oceanic crust. There are also other indicators of zircon alteration during their residence in the oceanic crust. It is likely hydrothermal fluid influence that affected the La vs. Sm N /La N systematics (Fig. 8) and decreased the δ 18 O values down to 1.75‰-3.15‰ (Table 1). Submarine hydrothermal solutions related to volcanic activity have high LREE/HREE and low Sm/La ratios (Bau and Dulski, 1999;Michard and Albarède, 1986) and thus could have contributed to the observed increase of LREE concentrations in some zircon grains. The values of δ 18 O in the examined zircons are lower than typical values for mantle zircons of about 5.3‰ (Valley et al., 1998), which strongly implies the influence of hydrothermal fluid flow (Bindeman, 2008). δ 18 O values in the range of 0-6‰ are typical for minerals from rocks belonging to the lower layer of oceanic crust that is composed of gabbro (Korolev et al., 2018;Valley et al., 2005;Eiler, 2001). Previously, it has been suggested that high-temperature (over 300 ºC) hydrothermal alterations may lead to a decrease of δ 18 O in minerals, including zircon, from the rocks of this layer with respect to mantle values (e.g., Korolev et al., 2018;Eiler, 2001;Gregory and Taylor, 1981). By contrast, zircon from oceanic plagiogranites and gabbros in the Mid-Atlantic and SWIR areas that is unaffected by submarine hydrothermal solutions is characterized by a rather narrow δ 18 O interval with an average value of 5.2‰ ± 0.5‰, indicating isotopicgeochemical equilibrium with MORB (Grimes et al., 2011).
It is challenging at this stage of research to give an unambiguous answer to the urgent question-how were the continentderived zircons delivered into oceanic igneous rocks in the vicinity of the SWIR? Over the past two decades, a number of hypotheses have been put forward to explain the unusual find-

12
Zircon Xenocrysts from the Shaka Ridge Record Ancient Continental Crust: New U-Pb Geochronological ings of continent-derived zircons in oceanic-floor magmatic rocks. A pioneer study of Pilot et al. (1998) proposed two plausible interpretations to explain the occurrence of unusually old zircons (~330 and~1 600 Ma) near the Kane fracture zone in the Mid-Atlantic Ridge. During the Atlantic opening, fragmented crustal material sank into small circulation cells that developed in the shallow mantle at each side of the ridge axis, and this material was transported through these cells to the ridge axis (Pilot et al., 1998). Alternatively, continental crustal material has been entrapped in the Kane fracture zone since the Atlantic opening as a result of repeated ridge jumping and transform migration (oscillatory spreading), with some of this material subsequently migrating down the ridge axis (Pilot et al., 1998).
The study of Bea et al. (2020), which summarizes a significant number of older-than-host zircon xenocryst samples (190 Ma to 3 Ga) in oceanic rocks of the Atlantic, suggested that the source of ancient xenocrysts are fragments of continental crust preserved in the vicinity of the mid-ocean ridge. Upwelling mafic magma can scavenge continental xenoliths, disintegrating them, and thus releasing zircon xenocrysts. Alternatively, it is assumed that hot mafic magma penetrating zircon-containing oceanic plagiogranites or fragments of continental crust may locally heat these rocks above the solidus. As a result of this process, patches of low-temperature silicate magma enriched with zircon are formed, which then migrate and mix with the rapidly rising magma of mafic composition (Bea et al., 2020 and references therein).
For the Indian Ocean, there are few reports dealing with findings of ancient zircon xenocrysts. The occurrences of Proterozoic (660-840 and >1 971 Ma; Torsvik et al., 2013) and Archaean (2.5-3.0 Ga; Ashwal et al., 2017) zircons on the Island of Mauritius were ascribed to zircon assimilation from ancient fragments of continental lithosphere beneath Mauritius and subsequent transportation to the surface by plume-related lavas (Ashwal et al., 2017;Torsvik et al., 2013). In the work by Cheng et al. (2016), two possibilities for the origin of Jurassic (~180 Ma) zircons in gabbroid rocks of the SWIR are considered: (1) the preservation of an intact Gondwana fragment on the ocean floor; (2) the occurrence of a partially altered continental relic that once resided in shallow regions of the upper mantle, and was entrained and transported to the vicinity beneath the axis.
From these examples，it is obvious that a convincing and unambiguous interpretation of the origin of unusually old zircons in young oceanic rocks has not yet appeared; it is also difficult to verify the explanations (Cheng et al., 2016). Note that the zircons examined in these papers span a very broad range of crystallization age (from 180 Ma to 3 Ga), which indicates their derivation from very different continental sources. In general, there is a series of major factors possibly responsible for the zircon transportation, such as a long-lived subduction zone, generation of mantle plumes, Atlantic opening, Gondwana supercontinent breakup, and further melting of small continental blocks of the sub-Gondwana lithosphere containing an ancient zircon component. However, many authors agree that the preservation of ancient continental fragments in the vicinity of a ridge axis is a necessary requirement for the transportation of zircons to occur. Taking into account the complex geological history of the study area, geochemical characteristics of the zircon xenocrysts examined here, and petrochemical characteristics of the host gabbro-diorite sample (high LILE, low HFSE), it appears that the zircons were captured as xenocrysts during the formation and migration of parental for the gabbro-diorite melts. Some authors provided strong evidence for the presence of the remains of the sub-Gondwana lithospheric mantle in the form of xenoliths and schlieren within modern oceanic lithosphere (Frey et al., 2002;le Roux et al., 2002;Kamenetsky et al., 2001;Douglass et al., 1999). We suggest that small blocks of continental crust after the Gondwana breakup could have survived in younger oceanic lithosphere and became involved in melting processes. The exact mechanism of crystal recycling into the mantle, however, is yet to be constrained. The continuing discoveries of ancient zircon in the mid-ocean ridges and deep ocean basins will shed additional light on this urgent issue.

CONCLUSIONS
(1) For the first time, zircon in gabbro-diorites of the Shaka Ridge was described. The U-Pb isotope system of the zircons records a crystallization age of~2.8 Ga and an age of a superimposed thermal event (metamorphism?) of~600 Ma.
(3) In discrimination diagrams based on the concentrations of several indicator elements (Yb, U, Y, Hf, etc.), the zircons fall into the fields of zircons from rocks of continental crust, mostly granitoids. Our results show the zircons to be derived from ancient continental crust and thus being genuine xenocrysts with respect to the host gabbro-diorite.
(4) During their residence within young oceanic crust, the zircons xenocrysts experienced the influence of a hightemperature hydrothermal fluid, which reduced the δ 18 O values of the zircons to 1.75‰-3.15‰.

ACKNOWLEDGMENTS
The study was funded by the National Key R & D Program of China (No. 2016YFE0203000) and the National Natural Science Foundation of China (No. 41673018). This study was also conducted under state contract (Institute of Precambrian Geology and Geochronology of Russian Academy of Sciences) (No. FMNU-2022-0005). Dmitriy I. Rezvukhin was supported by the state assignment project of IGM SB RAS. We thank Pavel Lvov for his help with U-Pb isotopic analyses, and Guoqiang Tang and Yu Liu for help with O isotopic analyses. Sergey Simakin and Evgeny Potapov assisted with measuring of zircon trace element compositions. Three anonymous reviewers provided valuable comments that improved the quality of the manuscript.The final publication is available at Springer via https://doi.org/10.1007/s12583-021-1422-2 Electronic Supplementary Material: Supplementary material (Table S1) is available in the online version of this article at https://doi.org/10.1007/s12583-021-1422-2.
Open Access: This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creativecommons.org/licenses/by/4.0/.