Archaean zircons in Miocene oceanic hotspot rocks establish ancient continental crust beneath Mauritius

Abstract

A fragment of continental crust has been postulated to underlie the young plume-related lavas of the Indian Ocean island of Mauritius based on the recovery of Proterozoic zircons from basaltic beach sands. Here we document the first U-Pb zircon ages recovered directly from 5.7 Ma Mauritian trachytic rocks. We identified concordant Archaean xenocrystic zircons ranging in age between 2.5 and 3.0 Ga within a trachyte plug that crosscuts Older Series plume-related basalts of Mauritius. Our results demonstrate the existence of ancient continental crust beneath Mauritius; based on the entire spectrum of U-Pb ages for old Mauritian zircons, we demonstrate that this ancient crust is of central-east Madagascar affinity, which is presently located ∼700 km west of Mauritius. This makes possible a detailed reconstruction of Mauritius and other Mauritian continental fragments, which once formed part of the ancient nucleus of Madagascar and southern India.

ARTICLE

Received 20 Sep 2016 | Accepted 25 Nov 2016 | Published 31 Jan 2017

DOI: 10.1038/ncomms14086 OPEN

Archaean zircons in Miocene oceanic hotspot rocks establish ancient continental crust beneath Mauritius

Lewis D. Ashwal1, Michael Wiedenbeck1,2 & Trond H. Torsvik1,2,3,4,w

A fragment of continental crust has been postulated to underlie the young plume-related lavas of the Indian Ocean island of Mauritius based on the recovery of Proterozoic zircons from basaltic beach sands. Here we document the rst UPb zircon ages recovered directly from 5.7 Ma Mauritian trachytic rocks. We identied concordant Archaean xenocrystic zircons ranging in age between 2.5 and 3.0 Ga within a trachyte plug that crosscuts Older Series plume-related basalts of Mauritius. Our results demonstrate the existence of ancient continental crust beneath Mauritius; based on the entire spectrum of UPb ages for old Mauritian zircons, we demonstrate that this ancient crust is of central-east Madagascar afnity, which is presently located B700 km west of Mauritius. This makes possible a detailed reconstruction of Mauritius and other Mauritian continental fragments, which once formed part of the ancient nucleus of Madagascar and southern India.

1 School of Geosciences, University of the Witwatersrand, Private Bag 3, WITS, Johannesburg 2050, South Africa. 2 Deutsches GeoForschungsZentrum GFZ, Telegrafenberg, D14473 Potsdam, Germany. 3 Center for Earth Evolution and Dynamics (CEED), University of Oslo, 0316 Oslo, Norway. 4 Geodynamics, Norges Geologiske Underskelse (NGU), N-7491 Trondheim, Norway. w Present address: Deutsches GeoForschungsZentrum GFZ, Telegrafenberg, D14473

Potsdam, Germany. Correspondence and requests for materials should be addressed to L.D.A. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 8:14086 | DOI: 10.1038/ncomms14086 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14086

Evidence is accumulating that old continental crust may occur beneath some young ocean-island volcanoes, contributing contaminating components to their chemical and

isotopic compositions. For example, the Sr, Pb and Nd isotopic compositions of basalts to rhyolites in therfajkull volcano of southeastern Iceland were modelled to have incorporated 26% of Precambrian continental crust1. This, combined with inversion of gravity anomaly data that indicate unusually thick crust of 430 km, has been used to infer that the Jan Mayen

Microcontinent extends southwestward to underlie southeast Iceland1. Similarly, a fragment of continental crust was suggested to underlie the young plume-related lavas of the Indian Ocean island of Mauritius, on the basis of gravity inversion modelling (crustal thickness) and the recovery of Proterozoic (6601,971 Ma) zircons from basaltic beach sands2. The island of

Mauritius (2,040 km2) is the second youngest member of a hotspot track extending from the active plume site of Runion, through the Mascarene Plateau, the Laccadive-Chagos Ridge and into the 65.5 Ma Deccan Large Igneous Province3,4. Three phases of Mauritian basaltic volcanism have been named the Older (9.04.7 Ma), Intermediate (3.51.66 Ma) and Younger(1.00.03 Ma) Series5,6. Minor volumes of trachytic rocks occur as intrusive masses or plugs o300 m across and are associated with Older Series basalts, as conrmed by UPb zircon thermal ionization mass spectrometry dating, which yielded at one location an age of 6.8490.012 Ma (ref. 7). Mauritian trachytes are variably altered from probable primary phonolitic magmas7, forming a prominent Daly Gap when plotted with coeval basalts; this suggests formation either by extreme fractional crystallization from the basalts8 or by direct partial melting of

40 E

45 E

57 2000

2010

2020

2030

35 S

40 S

Figure 1 | Simplied geology of Madagascar and India reconstructed to 9085 Ma. Mauritius (M) is reconstructed in a likely location near ArchaeanNeoproterozoic rocks in central-east Madagascar just prior to break-up2. The exact size and geometries of Mauritius and other potential Mauritian continental fragments (collectively known as Mauritia, including SM Saya de Malha; C, Chagos; CC, Cargados-Carajos Banks; LAC, Laccadives; N, Nazreth; see present location in Fig. 6) are unknown, and are generously drawn in the diagram. We propose that Mauritia is dominantly underlain by Archaean continental crust, and part of the ancient nucleus of Madagascar25,46 and India20,21 (stippled black line). A Large Igneous Province event (linked to the Marion plume) occurred from 92 to 84 Ma, and most of Madagascar was covered with ood basalts (full extent not shown for simplicity). Blue stippled line indicates the site of Cretaceous pre-breakup strike-slip faulting. AG, Analava gabbro (91.6 Ma); LR, Laxmi Ridge; S, Seychelles;

SM, St Mary rhyolites (91.2 Ma)41. The blackwhite box (geology of Madagascar) is enlarged in the inset to Fig. 5. Inset map shows simplied geology of Mauritius, including trachyte plugs7. Star symbol marked MAU-8 is the sampling area for the present study and black bars indicate locations of zircons recovered from beach sand samples2.

2 NATURE COMMUNICATIONS | 8:14086 | DOI: 10.1038/ncomms14086 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14086 ARTICLE

metasomatized mantle7. Isotopic compositions of the trachytes show relatively constant eNd of 4.030.15, with highly variable

ISr of 0.704080.71034, interpreted as reecting small amounts(0.43.5%) of contamination by Precambrian continental crust, successfully modelled using as a proxy B750 Ma granitoid composition from the Seychelles. It has been noted that other compositions, including older Proterozoic or even Archaean components, can also be plausibly modelled7. It is from such a

Mauritian trachyte that a population of zircons was extracted and analyzed for this paper. A subset of these zircons yields Archaean UPb ages, conrming the existence of a fragment of Precambrian continental crust beneath Mauritius, and we propose here that Mauritius and other Mauritian continental fragments are dominantly underlain by Archaean continental crust, and that these originally formed part of the ancient nucleus of Madagascar and India.

Grain 3

3/9: 3022

Ma

Qtz

3/1: 2930

Ma

3/7: 2874

Ma

3/8: 2950

Ma

3/10: 2849

Ma

3/11: 2967

Ma

Ksp

3/3: 2934

Ma 3/4:

2897

Ma 3/5:

3003 Ma

3/2: 3030

Ma

3/6: 3006

Ma

a

d

10 m

Grain 5

Qtz +

Ksp

5/3:

2810

Ma

5/2:

2830

Ma

5/1:

2834

Ma

e

6 m

b

Grain 8

8/4: 2552

Ma

8/5: 2447

Ma

Qtz

8/3: 2836

Ma

8/1: 2887

Ma 8/2: 2807 Ma

8/6: 2766

Ma

Mon

c

f

10 m

Figure 2 | Scanning electron microscope images of three Archaean zircon grains. These three grains were recovered from the MAU-8 trachyte sample. Backscattered electron (BSE) images (ac) of the three grains taken after completing all UThPb isotopic analyses. Cathodoluminescence (CL) images (df) taken prior to acquiring our SIMS data. The indicated analysis numbers correspond to those in Supplementary Data 1. The indicated ages are the radiogenic 207Pb/206Pb ages for the corresponding craters.

NATURE COMMUNICATIONS | 8:14086 | DOI: 10.1038/ncomms14086 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14086

0.8

ResultsUPb systematics. A single trachyte sample (MAU-8; location given in inset to Fig. 1) was selected for zircon separation based on available sample mass (B1 kg) and Zr concentration (1,165 mg g 1); extreme care was taken during sample processing to avoid any risk of contamination. Thirteen zircon grains were recovered, from which we report 68 individual point analyses acquired using the Cameca 1280-HR SIMS (secondary ion mass spectrometer) instrument at GFZ Potsdam; details of the sample processing and analytical methods are provided in the Methods section. Three zircon grains show uniquely mid- to late-Archaean UPb systematics, with no evidence for Phanerozoic components. Cathodoluminescence (CL) and backscattered electron (BSE) images (Fig. 2) show that these three crystals contain internal structures such as metamict cores, partially resorbed idiomorphic banding and numerous mineral inclusions. We conducted 20 individual UThPb spot analyses on these grains, where the concordant or near-concordant 207Pb/206Pb ages range from 3,0305 Ma to 2,55211 Ma (Fig. 3 and Supplementary Data 1). We interpret these data as indicating the crystallization ages of a complex Archaean xenocrystic component; no over-growths of young zircon are evident in any of the three Archaean grains, despite great efforts to identify such rims. The mineral inclusions in these ancient zircons include quartz, K-feldspar and monazite, as determined by EDS measurements; this assemblage would be consistent with crystallization from granitic or syenitic magmas. As the three grains are distinct in terms of their crystallization ages as well as their Th/U systematics, we conclude that trachytic magma traversed through and incorporated silicic continental crustal material that preserves a record of several hundred Myr of Archaean evolution. This is consistent with SrNd isotopic systematics of the Mauritian trachytes7, as discussed above.

Ten of the 13 grains differ from the older zircons in that they are featureless, with no internal structures visible in CL or BSE images (Fig. 4); these were determined to have late Miocene UPb systematics with no traces of inherited components, yielding an unweighted mean 206Pb/238U age of 5.70.2 Ma (1 s.d., N 48,

Fig. 3). We interpret these results as representing the crystallization age of the MAU-8 trachyte magma, which is consistent with their association with the Older Series of Mauritian basaltic volcanism (9.04.7 Ma, as constrained mainly by 40Ar/39Ar dating6).

Th/U ratios. U and Th concentrations of all measured individual spots in the zircon grains are given in Supplementary Data 1. For the three Archaean grains, Th/U ratios show distinct values, with Grains 3 and 5 yielding Th/U between 0.05 and 0.47, which is typical of variably metamorphosed and/or recrystallized igneous rocks9,10. In contrast, Grain 8 is very depleted in Th, with consistently low Th/U values around 0.01, more typical of high-grade metamorphic rocks11,12; we note that among the Archaean zircons, Grain 8 contains the youngest 207Pb/206Pb spot age (2,55211 Ma, excluding analyses that partly overlapped epoxy or mineral inclusions). In Grains 3 and 8, there is a tendency for spots with lower Th/U to occur on grain margins. These features suggest that the Archaean zircons were derived from source rocks that experienced a complex history of magmatic and metamorphic events. The young (Late Miocene) zircon grains all show much higher Th/U ratios of 0.72.6 (average Th/U 1.40.6, N 48) typical of magmatic zircons in

alkaline rocks10,13.

Oxygen isotopes. In order to further characterize the sources of the zircons, we undertook spot analyses of oxygen isotopes by

SIMS for some of the grains (Supplementary Data 2). For the Archaean zircons, the results yield d18OSMOW values between5.5% (Grain 3) and 9.79.9% (Grain 8). The lower value is comparable to the average d18OSMOW value of 5.50.4% for Archaean zircons in tonalitetrondhjemitegranodiorite rocks in the Superior Province of Canada14. The higher values are outside the range of typical d18OSMOW for magmatic zircons15, although grains with d18OSMOW48.5% have been reported in Precambrian granitoids from Fennoscandia16. In contrast, oxygen isotope data for 29 spot analyses of the Miocene age zircons give a mean d18OSMOW value of 4.590.20% (1 s.d.)

(Supplementary Data 2). These values are lower than the average d18OSMOW of mantle zircons (5.30.3%; ref. 15), but are within the range of zircons recovered from some kimberlites (d18OSMOW 3.44.7%; ref. 17). This would be consistent

with an origin for Mauritian trachytes as low-degree partial melts of fertile metasomatized mantle7.

Comparison of UPb ages with adjacent continents. Can the spectrum of UPb ages for old Mauritian zircons be correlated with exposed Precambrian terranes in nearby continental entities? We considered major continental masses like India, as well as large (e.g., Madagascar, 587 103 km2) and smaller (e.g.,

Seychelles, 459 km2) continental fragments as potential correlatives of the sub-Mauritius continental crust. Granitoid rocks of the Seychelles18 range from B700800 Ma, with the vast majority of ages B750 Ma (ref. 19); no Archaean components have been identied there. The Dharwar Craton of southern India (Fig. 1) consists of a nucleus of Palaeoarchaean to Neoarchaean(3.42.5 Ga) migmatitic orthogneisses anked by juvenile Neoarchaean (2.72.5 Ga), dominantly granitoid gneisses20. Palaeo- and Mesoproterozoic rocks are present south of the Dharwar Craton in India21. Neoproterozoic igneous or metaigneous rocks are rare to absent in this region, although the B750 Ma Malani Igneous Suite of Rajasthan, some 1,000 km to the NNW, has been correlated with the granitoids of the Seychelles22,23. The best match to the age spectrum of Precambrian zircons recovered from Mauritius occurs in east-

2

0.6

207 Pb/238 U

0.4

0.2

0.0 0 4 8 12 16 20 24

207Pb/235U

Figure 3 | Concordia plot. Includes all 20 data points from the three Archean zircons found in the MAU-8 trachyte sample. The ellipses indicate the 1 s.d. analytical uncertainties for each of the SIMS determinations. Also shown as red symbols are the TIMS wet chemical results reported by Torsvik et al.2 for eight Proterozoic zircons recovered from Mauritian basaltic beach sands. Not shown are the 48 SIMS determinations on the10 Miocene zircons, here indicated with the arrow. Tick marks on the Concordia curve are in Ma. TIMS, Thermal Ionization Mass Spectrometry.

4 NATURE COMMUNICATIONS | 8:14086 | DOI: 10.1038/ncomms14086 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14086 ARTICLE

Grain 6

a

Grain 1

Grain 4

b

1/54.9 Ma

4/35.3 Ma

1/25.9 Ma

6/25.7 Ma

6/35.5 Ma

6/45.6 Ma

4/25.8 Ma

1/35.6 Ma

6/55.7 Ma

1/65.7 Ma

4/15.7 Ma

6/16/6 5.9 Ma

1/15.4 Ma

20 m

1/45.7 Ma

20 m

c

20 m

Grain 7

d

f

e

Grain 9

11/35.8 Ma

11/55.4 Ma

11/26.2 Ma

9/55.5 Ma

9/65.5 Ma

11/15.8 Ma

7/25.6 Ma

9/25.8 Ma

9/35.9 Ma

7/55.8 Ma

9/15.8 Ma

9/45.6 Ma

11/65.6 Ma

7/16.0 Ma

9/75.5 Ma

7/36.0 Ma

11/45.6 Ma

Grain 11

20 m

20 m

g

Grain 13 Grain 14

i

12/35.5 Ma

12/55.5 Ma 12/15.9 Ma

12/2

5.7 Ma

12/45.8 Ma

13/15.8 Ma

14/35.6 Ma

13/45.4 Ma

14/25.5 Ma

14/55.2 Ma

13/25.7 Ma

14/45.3 Ma

h

14/15.9 Ma

20 m

20 m

20 m 20 m

30 m

Grain 12

Grain 15

15/15.7 Ma

j

15/45.5 Ma

15/25.8 Ma

15/35.6 Ma

Figure 4 | Backscattered electron images of the 10 Miocene age zircons recovered from the MAU-8 sample. Analysis numbers in BSE images (aj) correspond to those given in Supplementary Data 1. The indicated ages are the radiogenic 206Pb/238U ages for the corresponding crater.

central Madagascar, where Palaeoarchaean to Mesoarchaean(3.43.1 Ga, AntongilMasora domain) gneisses have been correlated with lithostratigraphic units in the Dharwar Craton, and juvenile Neoarchaean rocks of the Antananarivo domain can be correlated with similar belts in the Eastern Dharwar24,25. In addition, gabbroic, syenitic and granitic rocks of Neoproterozoic age (0.850.70 Ga) are well represented in Madagascar as the ImoronaItsindro Suite24,25, which has been suggested to represent the products of an Andean-type continental arc26,27. The entire Precambrian zircon age spectrum found in young Mauritian volcanic rocks could have been derived from a

B2,000 km2 area in east-central Madagascar, as illustrated in Fig. 5 (inset map).

DiscussionMicrocontinental fragments are a natural consequence of plate plume interaction1,2,28; an obvious example is the Seychelles, where evidence from surface exposures18 and seismic studies29,30 support the existence of a partly emergent 44.5 104 km2

fragment of continental crust with a thickness of B33 km. Less clear examples include the Laxmi Ridge (Fig. 6), which has been interpreted as a sliver of thinned continental crust31,32, or

NATURE COMMUNICATIONS | 8:14086 | DOI: 10.1038/ncomms14086 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14086

90

80

70

60

50

40

30

20

10

0

= 386)

6

2

Madagascar frequency (N)

Mauritius frequency (N)

4

Basalt beach sand (N=8) NW Mauritius SE Mauritius

Trachyte zircons (N=20)

Archean

Proterozoic Neoproterozoic

Mesoproterozoic

Paleoproterozoic

Neoarchean

Mesoarchean

Paleoarchean

Eoarchean

Ma

1,000 1,500 2,000 2,500 3,000 3,500

Figure 5 | Precambrian timescale and histogram of UPb ages from Madagascar and Mauritius. The diagram includes 368 UPb ages older than 600 Ma from Madagascar25, Mauritius trachyte UPb (207Pb/206Pb) ages of the present study (red bars) and ages from an earlier study of beach sand in Mauritius2 (blue bars). Arrows indicate that UPb ages are minimum ages (Pb-loss). Left-hand scale frequency for Madagascar ages and right-hand frequency (N number of ages) for Mauritius ages. Inset shows geological map of Madagascar in the Masora region (see Fig. 1, main text) depicting the

current size of Mauritius (thick red line); this illustrates that if underlain by continental crust of Madagascar afnity, Mauritius could well sample a spectrum of Neoproterozoic and Archean zircon xenocrysts. AIHSZ, Angavo-Ifanadiana High-Strain Zone in central-east Madagascar.

composed entirely of underplated oceanic crust33. In the Indian Ocean, contiguous continental crust of thickness 42530 km beneath the Seychelles and northern Mascarenes, and extending south-westwards towards Mauritius (Fig. 6) has been predicted2, but admittedly crustal thickness determinations from gravity inversion alone cannot distinguish thinned continental crust from anomalously thick oceanic crust. For Mauritius, receiver function data were used34 to infer a Moho depth of B21 km; those authors favoured entirely oceanic crust thickened by magmatic underplating, although they could not rule out the presence of embedded continental crustal relics. Clearly, geophysical data, in isolation, yield equivocal interpretations of oceanic crustal structures and compositions, but if used in conjunction with geochemical and/or isotopic data (e.g., ref. 7), or with recovered mineral grains that demand the presence of deep continental rocks (this study), the results are more secure.

Our ndings conrm the existence of continental crust beneath Mauritius and document, for the rst time, the presence of Archaean zircons as xenocrysts in young volcanic rocks from ocean basin settings. Archaean zircons (B3,000 Ma) were reported in young lavas and beach sands from the Galapagos, but so far only mentioned in a conference abstract35. Younger zircons (B330 and B1,600 Ma) have previously been described from Mid-Atlantic Ridge gabbros36, and Proterozoic (B1,800 Ma)37 and Mesozoic zircons (126242 Ma)38 have been found in young Icelandic basalts. Proterozoic zircons (2,430, 1,279, 807 Ma) were recovered in an early Paleogene (64.9 Ma) trachyte from Silhouette Island, Seychelles39, a result that is relevant to our work because it further improves our understanding of the nature and distribution of Precambrian continental fragments in the Indian Ocean. We propose that Mauritius and other potential Mauritian continental fragments, collectively named Mauritia2 (Fig. 1; see present location in Fig. 6), are dominantly underlain by Archaean continental crust, and formed part of the ancient nucleus of Madagascar and India.

Conversely, the Seychelles (and probably also the Laxmi Ridge) is likely underlain by Palaeoproterozoic rocks39, as in northern Madagascar, but was later affected by extensive Neoproterozoic arc magmatism18.

Mauritia acted as a buffer zone between the western Indian subcontinent and eastern Madagascar, and was fragmented by numerous tectonic and volcanic events that occurred in that region since the early Cretaceous. Due to the early opening between India and East Antarctica in the Enderby Basin, a sinistral shear zone must have acted between India and Madagascar, with India displaced northward relative to Madagascar. That caused convergence and perhaps subduction between northwestern India and Somalia/Arabia40. Between 100 and 90 Ma, those displacements were reversed. Soon afterwards, a late Cretaceous Large Igneous Province event (linked to the Marion plume) occurred from about 92 to 84 Ma. It is noteworthy that hardly any volcanism related to the early break-up is found in western margin of India (except the91.2 Ma St Mary Islands41), while most of Madagascar was covered with ood basalts. This suggests that Madagascar and western India were separated at that time by at least B200 km (Fig. 1), a region occupied by Mauritia. The bulk of this continental separation probably existed back in Precambrian times, although Late Jurassic (Tithonian) volcanism in the Seychelles, North Madagascar and Southern India may have been related to a failed rift system39 that contributed to continental extension between Madagascar and India. Cretaceous pre-breakup strike-slip faulting occurred immediately to the east of Madagascar (blue stippled line in Fig. 1), followed by the opening of the Mascarene Basin, when India, together with Mauritia and the Seychelles/Laxmi Ridge, broke away from Madagascar. Mauritia was subsequently fragmented into a ribbon-like conguration by a series of mid-ocean ridge jumps2, triggered by the proximity of the Marion, and thereafter the Runion plume. Subsequent volcanism blanketed most of the

6 NATURE COMMUNICATIONS | 8:14086 | DOI: 10.1038/ncomms14086 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14086 ARTICLE

Figure 6 | Location of possible continental fragments in the Indian Ocean. These include Mauritia (brownish shading), Laxmi Ridge and the Seychelles (yellowish shading). During the opening of the Mascarene Basin at about 84 Ma, India, together with most of Mauritia and the Seychelles/Laxmi Ridge, broke away from Madagascar. Mauritia was subsequently fragmented into a ribbon-like conguration because of a series of mid-ocean ridge jumps2, which were partly related to the Marion plume and later the Reunion plume (after 66 Ma). The current conguration with Mauritius/Cargados-Carajos Banks/Nazareth/Saya de Malha forming the Southern Mascarene Plateau (part of the African/Somali Plate), and with the Laccadives and Chagos being part of the Indian Plate, arose at 41 Ma (black arrow shows where Chagos originated at 41 Ma). North of Mauritia, seaoor spreading was initiated between the Laxmi Ridge and the Seychelles at around 6263 Ma, and the Seychelles became part of the African/Somalian plate after 61 Ma when seaoor spreading in the Mascarene Basin ceased2. Background bathymetry is ETOPO147 and continental plate polygons with continental-ocean boundaries are from Torsvik & Cocks48.

Mauritian continental fragments, and in the case of Mauritius, the ancient materials were sampled by plume-related volcanic rocks.

Methods

SIMS UPb methods. A key objective of this investigation was to establish whether any older components were present within the late Miocene Mauritian trachytic rocks. The selected sample (MAU-8, Fig. 1) was collected in situ from outcrop located at 20.32840 S; 57.71312 E (ref. 7). Approximately 1 kg of material was sent to GFZ Potsdam for sample processing and zircon recovery. Prior to processing, both the sample and the entire crushing facility were carefully cleaned and inspected in order to preclude any risk of cross-contamination from previously processed samples. The MAU-8 sample was friable, so the initial crushing was accomplished by hand using a metal rolling pin on a steel surface. No crushing or grinding apparatus was used. Repeated crushing and sieving resulted in nearly all of the material being reduced to the o500 mm grain size fraction. The resulting material was then panned in tap water, both to remove ne-grained dust and to concentrate the heavy mineral fraction. The dried material was passed through a Frantz magnetic separator, followed by concentration using bromoform, and then methyl iodide heavy liquids. The resulting material was washed, and ultimately 13 zircons could be recovered from the starting material. We exclude any possibility of having contaminated sample MAU-8 with foreign zircons during laboratory procedures: during the preceding decade, no samples of Archaean age had ever been processed in GFZ Potsdam, and the vast majority of material that had been crushed and concentrated was of Phanerozoic age.

The 13 grains were cast in Epox 2-component cold-set epoxy, along with the 91500 and Temora2 zircon reference materials42,43. The sample was then polished to a atness of o5 mm, as conrmed using white light prolometry. The 1 inch diameter sample block was then twice cleaned for 5 min using a high-purity ethanol ultrasonic bath, prior to carbon coating. We used the Potsdam Zeiss Ultra 55 plus eld emission scanning electron microscope to image each of the 13 grains

in both backscattered electron and monochromatic cathode luminescence modes. The sample was again cleaned in high-purity ethanol prior to being argon sputter coated with a 35-nm-thick, high-purity gold lm. The sample was then placed in a low magnetic susceptibility SIMS sample holder, being held in place with tension springs. The sample was then placed in the airlock of the Cameca 1280-HR SIMS instrument.

Our initial analyses employed a 13 nA 16O2 primary beam, employing Khler illumination, and focused to B10 mm diameter at the surface of the sample. Oxygen ooding, at a pressure of B2 10 3 Pa, was used in order to improve Pb

sensitivity. The mass spectrometer was operated in mono-collection mode, using a mass resolution of M/DM B5,000, using an ETP 133H electron multiplier for ion detection, to which a synthetic 46.2 ns dead time was applied to the preamplier circuit. A single analysis took 32 min, and the data were mostly collected in fully automated mode, using the Cameca point-logger software package. The UPb fractionation factor was dened using the Pb/UO versus UO/UO2 relationship, employing a power law t as dened using the 91500 reference material, which has a 206Pb/238U age of 1062.4 Ma (ref. 42). A total of N 29 determinations were

conducted on the 91500 calibrant during the initial UPb session (3 days spread over a week), yielding an unweighted mean 206Pb/238U age of 1062.36.6 Ma(1 s.d.). We conrmed the accuracy of the UPb fractionation correction by measuring N 21 determinations on the Temora2 zircon Quality Control

Material. Here we determined a mean, unweighted 206Pb/238U age of 421.12.8 Ma (1 s.d.), which is in reasonable agreement with the published age of 416.50 Ma (ref. 43), conrming that no signicant systematic bias is present in our U/Pb determinations. Common Pb corrections, where needed, were based on the observed 204Pb/206Pb ratio (older grains) or 207Pb/206Pb ratio (younger grains), in conjunction with a recent common Pb composition.

Our initial survey of UPb ages identied three grains with Archaean ages. We decided to survey the two larger of these grains (Grains 3 and 8, Fig. 2) in detail, using a small beam diameter. A 100 pA 16O2 primary beam employing a Gaussian beam distribution was focused to a B2 mm diameter at the sample surface; this lower primary beam current resulted in a much lower ion collection rate. In order to produce a at-bottomed sputtering cater, a 5 5 mm raster was applied to the

primary beam, and this was compensated for using the dynamic transfer capability of the 1280-HRs secondary ion optics. All other aspects of the second data acquisition series were the same as those for earlier 2 nA analyses. Measurements of the Temora2 Quality Control Material gave an unweighted 206Pb/238U age of 4105.3 Ma (1 s.d., N 6); despite the low primary ion current used, this is in

reasonable agreement with the assigned 416.50 Ma age43 of the Temora2 zircon suite. After completion of this second analytical series, we again imaged the sample in BSE mode using the Potsdam SEM, thereby documenting the exact locations of each analysis point (Figs 2 and 4). This imaging required that an additional 20 nm of gold be sputter coated onto the sample mount. Finally, SIMS crater dimensions were determined using a Zygo Lot 7100 white light prolometer. These reveal crater depths of 2.7 and 1.3 mm for the 2 nA and 100 pA analyses, respectively.

From the prolometry results, we estimate test portion masses of B3.2 ng and B150 pg for the two machine settings, respectively.

Results of all SIMS spot analyses of zircons in trachyte sample MAU-8 are given in Supplementary Data 1, and are plotted on a conventional Concordia diagram in Fig. 3, along with the thermal ionization mass spectrometry analyses of Proterozoic zircons recovered from basaltic beach sands2. All results for Precambrian zircons from Mauritius are plotted on a histogram in Fig. 5, along with a compilation of Precambrian zircons from Madagascar (N 386)25.

SIMS oxygen isotope methods. With the intent of further characterizing the nature of our suite of Mauritian zircons, we undertook additional oxygen isotope determinations by SIMS. Our analytical technique closely followed that described for the Potsdam SIMS laboratory44, using the 91500 zircon reference material for calibration, which has an assigned d18OSMOW value of 9.860.11%, based on gas source mass spectrometry45. Due to the fact that the zircons were both small in size and thin, due to earlier sample polishing, it was decided that a repolishing of the mount was imprudent. Furthermore, due to the numerous UThPb determinations, which had been previously conducted and which would have implanted 16O from the SIMS primary beam, there were only three locations on the Archean grains that could be analyzed. These results, along with the results from 29 locations on Miocene zircons, are given in Supplementary Data 2.

Imaging methods. Backscattered electron images show the presence of numerous mineral inclusions in 2 of the 3 Archaean grains (Fig. 2), whereas all 10 of the Miocene grains were inclusion-free (Fig. 4). In order to identify the nature of these inclusions, we removed the gold coating by gently rubbing with an ethanol-saturated tissue. The mount was then cleaned in high-purity ethanol for 5 min using an ultrasonic bath. The sample was sputter coated with carbon, followed by EDX analyses using the GFZ Potsdam Zeiss SEM instrument. From the major element abundances we identied quartz, K-feldspar and monazite.

Data availability. The authors declare that all relevant data are available within the article and its Supplementary Data les. Other pertinent data are available from the authors upon request.

NATURE COMMUNICATIONS | 8:14086 | DOI: 10.1038/ncomms14086 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14086

References

1. Torsvik, T. H. et al. Continental crust beneath southeast Iceland. Proc. Natl Acad. Sci. USA 112, E1818E1827 (2015).

2. Torsvik, T. H. et al. A Precambrian microcontinent in the Indian Ocean. Nat. Geosci. 6, 223227 (2013).

3. Duncan, R. A. in Proc. Ocean Drilling Program Scientic Results 115 (eds Duncan, R.A., Backman, J., Peterson, L.C.) 310 (1990).

4. Richards, M. A., Duncan, R. A. & Courtillot, V. E. Flood basalts and hotspot tracks: plume heads and tails. Science 146, 103107 (1989).

5. Baxter, A. N. Magmatic Evolution of Mauritius, Western Indian Ocean. PhD thesis, University of Edinburgh, (1972).

6. Moore, J. et al. Evolution of shield-building and rejuvenescent volcanism of Mauritius. J. Volcanol. Geothermal Res. 207, 4766 (2011).

7. Ashwal, L. D. et al. A mantle-derived origin for Mauritian trachytes. J. Petrol. 57, 16451676 (2016).

8. Baxter, A. N. Petrology of the Older Series lavas from Mauritius, Indian Ocean. Geol. Soc. Am. Bull. 86, 14491458 (1975).

9. Hoskin, P. W. O. & Black, L. P. Metamorphic zircon formation by solid-state recrystallization of protolith igneous zircon. J. Metamorphic Geol 18, 423439 (2000).

10. Kirkland, C. L., Smithies, R. H., Taylor, R. J. M., Evans, N. & McDonald, B. Zircon Th/U ratios in magmatic environs. Lithos 212215, 397414 (2015).

11. Rubatto, D. Zircon trace element geochemistry: partitioning with garnetand the link between U-Pb ages and metamorphism. Chem. Geol. 184, 123138 (2002).

12. Hoskin, P. W. O. & Schaltegger, U. The composition of zircon and igneous and metamorphic petrogenesis. Rev. Mineral Geochem. 53, 2762 (2003).

13. Sutherland, F. L. et al. Multiple felsic events within post-10 Ma volcanism, Southeast Australia: inputs in appraising proposed magmatic models. Austr. J. Earth Sci. 61, 241267 (2014).

14. King, E. M., Valley, J. W., Davis, D. W. & Edwards, G. R. Oxygen isotope ratios of Archean plutonic rocks from granite-greenstone belts of the Superior Province: indicator of magmatic source. Precambrian Res. 92, 4767 (1998).

15. Valley, J. W. Oxygen isotopes in zircon. Rev. Mineral. Geochem. 53, 343385 (2003).

16. Lahtinen, R. et al. Paleoproterozoic magmatism across the Archean-Proterozoic boundary in central Fennoscandia: geochronology, geochemistry and isotopic data (Sm-Nd, Lu-Hf, O). Lithos 262, 507525 (2016).

17. Valley, J. W., Kinny, P. D., Schulze, D. J. & Spicuzza, M. J. Zircon megacrysts from kimberlite: oxygen isotope variability among mantle melts. Contrib. Mineral. Petrol. 133, 111 (1998).

18. Ashwal, L. D., Demaiffe, D. & Torsvik, T. H. Petrogenesis of Neoproterozoic granitoids and related rocks from the Seychelles: the case for an Andean-type arc origin. J. Petrol. 43, 4583 (2002).

19. Tucker, R. D., Ashwal, L. D. & Torsvik, T. H. U-Pb geochronology of Seychelles granitoids: a Neoproterozoic continental arc fragment. Earth Planet. Sci. Lett. 187, 2738 (2001).

20. Jayananda, M., Chardon, D., Peucat, J.-J., Tushipokla & Fanning, C. M. Paleo- to Mesoarchean TTG accretion and continental growth in the western Dharwar craton, Southern India: constraints from SHRIMP U-Pb zircon geochronology, whole-rock geochemistry and Nd-Sr isotopes. Precambrian Res. 268, 295322 (2015).

21. Ghosh, J. G., de Wit, M. J. & Zartman, R. E. Age and tectonic evolution of Neoproterozoic ductile shear zones in the Southern Granulite Terrain of India, with implications for Gondwana studies. Tectonics 23, TC3006 (2004).

22. Torsvik, T. H. et al. Rodinia rened or obscured: palaeomagnetism of the Malani Igneous Suite (NW India). Precambrian Res. 108, 319333 (2001).

23. Torsvik, T. H., Ashwal, L. D., Tucker, R. D. & Eide, E. A. Neoproterozoic geochronology and palaeogeography of the Seychelles microcontinent: The India link. Precambrian Res. 110, 4759 (2001).

24. Tucker, R. D. et al. Neoproterozoic extension in the Greater Dharwar Craton: a reevaluation of the Betsimisaraka suture in Madagascar. Can. J. Earth Sci. 48, 389417 (2011).

25. Tucker, R. D., Roig, J. Y., Moine, B., Delor, C. & Peters, S. G. A geological synthesis of the Precambrian shield in Madagascar. J. Afr. Earth Sci 94, 930 (2014).

26. Handke, M. J., Tucker, R. D. & Ashwal, L. D. Neoproterozoic continental arc magmatism in west-central Madagascar. Geology 27, 351354 (1999).

27. Bybee, G. M., Ashwal, L. D. & Wilson, A. H. New evidence for a volcanic arc on the western margin of a rifting Rodinia from ultramac intrusions in the Andriamena region, north-central Madagascar. Earth Planet. Sci. Lett. 293, 4253 (2010).

28. Gaina, G., Mller, R. D., Brown, B. J. & Ishihara, T. Microcontinent formation around Australia. Geol. Soc. Am. Spec. Pap. 372, 405416 (2003).

29. Davies, D. & Francis, T. J. G. The crustal structure of the Seychelles Bank. Deep-Sea Res. 11, 921927 (1964).

30. Laughton, A. S., Matthews, D. H. & Fisher, R. L. in The Sea, Vol. 4 (eds Maxwell A. E., Hill, M. N. & Munk W. H.) 543586 (Wiley, 1970).

31. Collier, J. S. et al. Factors inuencing magmatism during continental breakup: new insights from a wide-angle seismic experiment across the conjugate Seychelles-Indian margins. J. Geophys. Res. 114 (2009).

32. Nemcock, M. & Ryba, S. Rift-Drift Transition in a Magma-Rich System: The Gop Rift-Laxmi Basin Case Study, West India, Vol. 445 (Geological Society London Special Publications, 2016).

33. Misra, A. A., Sinha, N. & Mukherjee, S. Repeat ridge jumps and microcontinent separation: insights from NE Arabian Sea. Mar. Petrol. Geol. 59, 406428 (2015).

34. Fontaine, F. R. et al. Crustal and uppermost mantle structure variation beneath La Runion hotspot track. Geophys. J. Int. 203, 107126 (2015).

35. Rojas-Agramonte, Y. et al. in International Geological Congress, Abstracts, Vol. 35 (Cape Town, South Africa, 2016).

36. Pilot, J., Werner, C.-D., Haubrich, F. & Baumann, N. Palaeozoic and Proterozoic zircons from the mid-Atlantic Ridge. Nature 393, 676679 (1998).

37. Schaltegger, U. et al. Contamination of OIB by underlying ancient continental lithosphere: U-Pb and Hf isotopes in zircons question EM1 and EM2 mantle components. Geochim. Cosmochim. Acta 66, A673 (2002).

38. Paquette, J., Sigmarsson, O. & Tiepolo, M. in EOS Trans. AGU, Fall Meeting 2006. Abstract #V33A-0642 (American Geophysical Union, 2006).

39. Shellnutt, J. G., Lee, T.-Y., Chiu, H.-Y. & Wong, J. Evidence of Middle Jurassic magmatism within the Seychelles microcontinent: Implications for the breakup of Gondwana. Geophys. Res. Lett. 42, 20710,215 (2015).

40. Gaina, C., van Hinsbergen, D. & Spakman, W. Tectonic interactions between India and Arabia since the Jurassic reconstructed from marine geophysics, ophiolite geology, and seismic tomography. Tectonics 34, 875906 (2015).

41. Torsvik, T. H. et al. Late Cretaceous India-Madagascar t and timing of break-up related magmatism. Terra Nova 12, 220224 (2001).

42. Wiedenbeck, M. et al. Three natural zircon standards for U-Th-Pb, Lu-Hf, trace element and REE analysis. Geostand. Newslett. 19, 123 (1995).

43. Black, L. P. et al. Improved 206Pb/238U microprobe geochronology by monitoring of a trace-element-related matrix effect: SHRIMP, ID-TIMS, ELA-ICP-MS and oxygen isotope documentation for a series of zircon standards. Chem. Geol. 205, 115140 (2004).

44. Nasdala, L. et al. Zircon M127a homogeneous reference material for SIMS U-Pb geochronology combined with hafnium, oxygen and, potentially, lithium isotope analysis. Geostand. Geoanal. Res. 40, 457475 (2016).

45. Wiedenbeck, M. et al. Further characterisation of the 91500 zircon crystal. Geostand. Geoanal. Res. 28, 939 (2004).

46. Archibald, D. B. et al. Towards unravelling the Mozambique Ocean conundrum using a triumvirate of zircon isotopic proxies on the Ambatolampy Group, central Madagascar. Tectonophysics 662, 167182 (2015).

47. Amante, C. & Eakins, B. W. ETOPO1, 1 Arc-Minute Global Relief Model: procedures, data sources and analysis, NOAA Technical Memorandum NESDIS NGDC-24 (2009).

48. Torsvik, T. H. & Cocks, L. R. M. Earth History and Palaeogeography 317 (Cambridge University Press, 2017).

Acknowledgements

Zircon extraction in Potsdam was overseen by J. Glodny. F. Coufgnal was responsible for the SIMS data acquisition parts of this project. A. Rocholl and Z. Jin supported the data reduction and presentation of the UPb results. I. Schapan provided both the SEM images and the EDS chemical analyses. The Research Council of Norway, through its Centres of Excellence funding scheme, project number 223272 (CEED), and the European Research Council, under the European Unions Seventh Framework Programme (FP7/20072013/ERC Advanced Grant Agreement Number 267631, Beyond Plate Tectonics), are acknowledged for nancial support. L.D.A. acknowledges support from the National Research Council (South Africa) and the University of the Witwatersrand. We thank Chris Harris for advice on oxygen isotope interpretations, and Carmen Gaina, Alan Collins, Greg Shellnutt and Jean-Louis Paquette for constructive and helpful comments.

Author contributions

L.D.A. collected and characterized the samples petrologically and geochemically, and initially recovered the rst zircons in the trachytic rocks. M.W. oversaw the SIMS and SEM laboratory experiments, and played the leading role in data reduction and the interpretation of the SIMS isotopic data. T.H.T. provided the plate reconstructions. All authors contributed to the manuscript.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

8 NATURE COMMUNICATIONS | 8:14086 | DOI: 10.1038/ncomms14086 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14086 ARTICLE

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Ashwal, L. D. et al. Archaean zircons in Miocene oceanic hotspot rocks establish ancient continental crust beneath Mauritius. Nat. Commun. 8, 14086 doi: 10.1038/ncomms14086 (2017).

Publishers note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional afliations.

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

r The Author(s) 2017

NATURE COMMUNICATIONS | 8:14086 | DOI: 10.1038/ncomms14086 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9