Neoproterozoic evolution of the eastern Arabian basement based on a refined geochronology of the Marbat region, Sultanate of Oman

Abstract New high spatial resolution secondary ion mass spectrometry (SIMS) U–Pb zircon data from the Sadh gneiss complex and the intruding Marbat granodiorite of the Marbat region, southern Sultanate of Oman, yield Cryogenian magmatic protolith ages for gneisses ranging from c. 850 to 830 Ma. Zircon ages record a c. 815–820 Ma period of deformation and migmatization, followed by intrusion of a hornblende gabbro/diorite and the undeformed Marbat granodiorite at c. 795 Ma. Following break-up and rifting of Rodinia at c. 870 Ma, crustal growth in the Marbat region occurred via arc accretion at c. 850–790 Ma, possibly in the easternmost part of the Mozambique Ocean based on earlier cessation of accretion here compared to the Arabian–Nubian Shield. Similarity of the new zircon geochronology to peaks of detrital zircon ages in the unconformably overlying Ediacaran Marbat sandstone suggests relatively local derivation from uplifted basement for the latter. Supplementary material: Detailed petrographic descriptions and photographs of hand specimens and thin-sections are available at http://www.geolsoc.org.uk/SUP18685.

The Marbat region of southern Oman (also referred to as the Salalah area in some studies, e.g. Mercolli et al. 2006) (Fig. 1a), located c. 50-130 km east of the city of Salalah, provides the best-exposed section of Precambrian high-grade gneisses in the southeastern part of the Arabian Peninsula (Platel et al. 1987a, b, c). Significant lack of exposure east of the ANS raises a fundamental question regarding the relationship between the isolated exposures of the eastern Arabian basement in Oman (i.e. the Marbat region of this study as well as smaller inliers at Jebel Ja'alan and Qalhat) and those of the ANS itself: do the accretionary terranes of the ANS extend across the Arabian Peninsula to the exposed basement windows in Oman, or do the latter represent a completely different tectonic regime(s)?
Previous geochronological studies of predominantly magmatic rocks and high-grade gneisses in the Marbat region ( Fig. 1b) used K -Ar, Rb -Sr, Sm -Nd, conventional U -Pb and Pb/Pb leaching methods (e.g. Mercolli et al. 2006; Table 1). Such methods applied to rock-forming minerals with different closure temperatures (e.g. mica, garnet, hornblende, etc.) or to potentially complex zircons do not necessarily provide a robust chronology of magmatic events because of resetting of the isotopic systems or mixing of different age domains.
In order to better understand the complex evolution of these high-grade rocks, new U -Pb zircon age data were generated using high spatial resolution secondary ion mass spectrometry (SIMS). Gneisses belonging to the heterogeneous Sadh complex together with the intruding Marbat granodiorite (Fig. 1b) are used to better constrain the timing of crust-forming events in the crystalline basement of the Marbat region and permit comparison with the crustal evolution of the ANS.   Fig. 1. (a) Present-day tectonic setting of the Arabian-Nubian Shield and the exposed Precambrian crystalline basement (after Mercolli et al. 2006;Schlüter 2006;Stoeser & Frost 2006;Bowring et al. 2007). The study area in Oman (indicated by the square) is in the SE Arabian Peninsula. The collage of gneissic terranes (e.g. Abas and Al Mahfid) and island-arc terranes (e.g. Al Bayda and Al Mukalla) in the SW Arabian Shield are shown. (b) Geological map of crystalline basement in southern Oman, mainly compiled from the geological map sheets of Marbat (NE 40-9E) and Sadh (NE 40-9F) (after Platel et al. 1987a, c). The Hadbin tonalite is after Mercolli et al. (2006). The sample locations (*) and sample numbers for the Sadh complex and the Marbat granodiorite are shown.

Geological setting and previous work
The Arabian Shield contains a wide range of Archaean and Proterozoic rocks (Quick 1991;Agar et al. 1992;Windley et al. 1996Windley et al. , 2001Whitehouse et al. 1998Whitehouse et al. , 2001aJohnson et al. 2001;Stoeser & Frost 2006;Stern & Johnson 2010). In the ANS of Saudi Arabia, Stoeser & Frost (2006) distinguish three different terranes using Nd, Pb, Sr and O isotopic data: the western and eastern arc terranes and the Khida terrane (Fig. 1a). The western and eastern terranes comprise rocks of oceanic arc affinity while the Khida terrane exposes reworked older continental crust (Stacey & Agar 1985;Stoeser & Stacey 1988;Whitehouse et al. 2001b;Stoeser & Frost 2006). Terrane boundaries within the western and eastern terranes are generally represented by complex fault zones (Fig. 1a) that commonly contain ophiolitic remnants (Pallister et al. 1988), although the northwestern border of the eastern terrane remains poorly defined (Stoeser & Frost 2006;Stern & Johnson 2010). The evolution of these terranes is based on a variety of age (Rb -Sr, K -Ar and U -Pb zircon) and isotopic (Sm-Nd depleted mantle model ages or T DM ; Stoeser & Frost 2006) data. The western arc terranes were likely formed and assembled at c. 870-700 Ma (Pallister et al. 1988;Quick 1991;Stoeser & Frost 2006). The eastern arc terranes are somewhat younger, spanning the period c. 740 -620 Ma (Stoeser & Frost 2006). While generally considered to represent largely juvenile Neoproterozoic crustal growth, inherited pre-Neoproterozoic zircons have been documented in both the western (Hargrove et al. 2006) and eastern arc terranes (Calvez et al. 1985). In the Khida terrane, metasediments have yielded U -Pb zircon age populations (both conventional and ion-probe) of c. 2600-2400, 1900-1650and 950 -800 Ma (Agar et al. 1992, while the Muhayil granite located in the Khida terrane yields a SIMS U -Pb zircon age of 1660 + 10 Ma (Whitehouse et al. 2001b).
In the southern parts of the Arabian shield in Yemen, several distinct terranes have been recognized on the basis of age and isotope data ( Fig. 1a; Windley et al. 1996;Whitehouse et al. 1998). Granitoid gneisses of the Abas gneiss terrane with U-Pb zircon ages of c. 760 Ma (Whitehouse et al. 1998) yield T DM ages of c. 2300-1300 Ma (Windley et al. 1996). The Al-Bayda island-arc terrane comprises granitoids with T DM ages of c. 2500-2000 Ma and a gabbro with a T DM of c. 1200 Ma (Windley et al. 1996). Gneisses of the easternmost Al-Mahfid gneiss terrane yield T DM ages of c. 3000-2700, 2100, 1800 and 1300 Ma (Windley et al. 1996) and U -Pb zircon ages of c. 760 and 2900-2550 Ma (Whitehouse et al. 1998). Unambiguous correlation of the terranes in Yemen with those in Saudi Arabia is not possible, but the broadly eastward increasing continental affinity evidenced by Archean Nd model ages, elevated Pb isotopes (Whitehouse et al. 2001a) and even crystallization ages for some rocks of the Al Mahfid terrane point to a significant difference in character of the ANS at its exposed eastern margin.
In the Marbat region of southern Oman, Precambrian crystalline basement exposed over an area of c. 1500 km 2 has been subdivided into several lithological suites (Platel et al. 1987a, b;Mercolli et al. 2006). The oldest of these are the predominantly meta-sedimentary Juffa and meta-igneous Sadh complexes. These are succeeded by the undeformed Tonalite Group (which includes the dioritictonalitic Fusht complex, the tonalitic/granodioritic Hadbin complex that contains a major body of netveined co-mingled tonalite -diorite and the gabbroic Hasik complex) and cross-cutting intrusive rocks. The latter comprise the Marbat granodiorite, the extensive calc-alkaline Shaat dyke swarm (basalt, andesite, dacite and rhyolite in composition; Worthing 2005), pegmatites, and the Leger granite (Table 1; Fig. 1b). The Sadh and Juffa complexes were metamorphosed under amphibolite facies and partly retrogressed to greenschist facies (Mercolli et al. 2006). Phlogopite-bearing and hornblendebiotite-olivine-bearing lamprophyre dykes cutting the basement rocks are undated but probably relate to the Oligocene opening of the Gulf of Aden.
The Sadh gneiss complex, which accounts for a major part of the basement exposure in the Marbat region, has been interpreted to represent a subduction-related island arc on the basis of its geochemical signature (Platel et al. 1987b;Würsten 1994;Briner 1997). The gneiss complex is lithologically heterogeneous, consisting predominantly of mafic and intermediate gneiss with subordinate felsic gneiss and amphibolite. The mafic gneiss has locally intruded earlier amphibolites, some with preserved gabbroic textures. The mafic gneiss, and also 100-m-sized sheets of leucogranitic gneiss, were then intruded by many basic dykes and later folded, sheared and metamorphosed to amphibolite faces; this has resulted in the gneisses containing many folded amphibolite dykes that are still discordant to the foliation of the gneiss. These discordant amphibolite dykes serve to separate and distinguish the pre-dyke and post-dyke periods of deformation and amphibolite facies metamorphism. The earliest gabbroic amphibolites, the mafic gneisses and the amphibolite dykes were later intruded by homogeneous granitic sheets and veins. Hornblende, biotite-hornblende and biotite-gneiss varieties are generally considered to be of magmatic origin, although occasional small calc-silicate nodules in the biotite-gneiss suggest that at least some of these may be meta-sedimentary (Mercolli et al. 2006). Granitic gneiss mainly occurs in the southern part of the study area (Fig. 1b), primarily in the hinge of a major antiform (Platel et al. 1987a, b). Polyphase deformation of the gneiss is recorded by at least two generations of folds. The gneiss is layered at the metre-scale, banded at the centimetrescale and contains a penetrative foliation that is parallel or discordant to these larger structures (Mercolli et al. 2006). Notably, the heterogeneous biotite and biotite -hornblende gneiss has undergone partial melting to generate a migmatite with numerous thin and laterally persistent leucosome bands that have accumulated locally in lowstrain fold hinges. Sm-Nd model ages (T DM ) of c. 960 -910 Ma have been reported from rocks of the Sadh complex (e.g. Mercolli et al. 2006; Table 1).
The original dioritic to tonalitic plutons of the Mahall complex intrude throughout the Sadh gneiss complex (Mercolli et al. 2006). In contrast to the Sadh gneisses, rocks of the Mahall complex have only experienced a single penetrative deformation event creating the main foliation in the area (Mercolli et al. 2006). During deformation, the Mahall bodies recrystallized or were partially melted to amphibolite, biotite gneiss, biotite-hornblende and hornblende -biotite gneiss (Würsten 1994;Briner 1997). The Mahall complex yielded conventional U-Pb zircon ages of c. 800 Ma (hornblende biotite gneiss and biotite gneiss) and T DM ages of 950 and 880 Ma (Mercolli et al. 2006; Table 1).
The c. 2 km 2 Marbat granodiorite is generally massive and cross-cuts the penetrative foliation of surrounding gneisses (Platel et al. 1987a, b;Mercolli et al. 2006). It is occasionally foliated along its margins. Xenoliths of dark gneiss are presumably fragments of the host Sadh gneiss complex, partly to almost wholly assimilated (Platel et al. 1987a, b;Mercolli et al. 2006). Occasional coarse-grained K-feldspar-rich pegmatite veins in the Marbat granodiorite are thought to be of anatectic origin and not genetically related to the Marbat granodiorite (Platel et al. 1987b Bowring et al. 2007). Mercolli et al. (2006) reported a Rb -Sr whole-rock plus muscovite age of c. 745 Ma and a K -Ar biotite age for the Marbat granodiorite of 718 + 12 Ma; these were interpreted as dating the cessation of greenschist facies metamorphism in the region.
The Precambrian basement of the Marbat region is unconformably overlain by the Marbat Sandstone Formation of probable Ediacaran (630-542 Ma) age (Platel et al. 1987b;Mercolli et al. 2006;Bowring et al. 2007). Although the depositional age of the Marbat sandstone is not known, the detrital zircon population contains major age peaks at c. 870, 840, 810 and c. 745 Ma (Bowring et al. 2007;Rieu et al. 2007), the youngest age defining a maximum age of deposition of ≤745 Ma.

Sample descriptions
Six samples representing the range of rock types present in the Sadh complex were collected for this study. Latitude and longitude of sample locations shown on Figure 1b are provided in Table 2.

Sadh complex: granitic gneiss (OM05-14)
This sample is from an outcrop on an unpaved (in 2005) road c. 14 km ESE of Marbat (Fig. 1b). The gneissic foliation strikes east -west and is nearly vertical. Dark intermediate gneiss and granitic gneiss are interlayered as lens-like bodies which together follow the regional-scale folding defined by granitic veins and gneiss (Fig. 1b). The granitic gneiss contains thin quartz-and feldspar-rich leucosomes (Fig. 2a).
Sadh complex: biotite gneiss (OM05-25) intruded by hornblende gabbro/diorite (OM05-26) The Sadh complex to the NE of Marbat is represented by interlayered intermediate gneiss and biotite gneiss (Fig. 2b). A biotite gneiss sample was collected c. 1 m from the contact, c. 5 m west of the hornblende gabbro/diorite sample location. The biotite gneiss contains elongated thin feldsparand quartz-rich leucosomes hosted by dark intermediate gneiss (Fig. 2b). The penetrative common foliation of both components of the biotite gneiss varies across the outcrop from striking SW with a steep dip (70-908) to striking north-south and dipping c. 408 to the west.

Marbat granodiorite (OM05-33)
The Marbat granodiorite sample was collected from a road-cut situated c. 3 km along the main asphalt road north of Marbat (Fig. 1b). Adjacent to the sample location, the Marbat granodiorite entrains a partially resorbed gneissic xenolith c. 1 m 2 in size that resembles the Sadh intermediate banded gneiss (Fig. 2d). Elsewhere, the granodiorite clearly cross-cuts the Sadh complex. Semi-brittle fractures and thin quartz-filled shear zones are seen in the same road-cut.

Metamorphism
Petrographic analysis of the Sadh complex gneiss samples indicates peak metamorphism under amphibolite facies conditions accompanied by partial melting, migmatization and formation of heterogeneous gneisses. Original magmatic features are limited to relict pyroxenes in hornblende gabbro/diorite (OM05-26) and preserved primary igneous textures in the Marbat granodiorite (OM05-33). Pervasive retrogression is documented by alteration of relict pyroxenes to amphibole and hornblende to lower-grade metamorphic minerals biotite, chlorite and epidote. Feldspar is commonly altered to saussurite, sericite and occasionally chlorite, while chlorite rimming subhedral biotite and sericite indicates later hydration. In the post-tectonic Marbat granodiorite, normally zoned plagioclase (Ca-rich cores and more albitic rims) together with saussuritization and myrmekite indicates crystallization during cooling.

Methods
Zircons were extracted from bulk-rock samples using conventional methods (cleaned, divided, crushed, milled, Wilfley table, magnetic separation and heavy liquid separation using lithium heteropolytungstate). A single sample (hornblende gabbro/ diorite OM05-26) was further processed with methylene iodide. Zircons were handpicked and mounted in epoxy, together with zircon standard 91500 (1065 Ma, Wiedenbeck et al. 1995). The mount was polished to expose the interior of the zircons and coated with gold. Imaging of the zircons was performed at Stockholm University with a Philips XL30 FEG scanning electron microscope (SEM) using secondary electron (SE) and cathodoluminescence (CL) detectors. Zircon descriptions are summarized in Table 2 and representative CL images presented in Figure 3. Zircon U-Pb analyses were performed at the NordSIM facility (Swedish Museum of Natural History, Stockholm) using a CAMECA IMS 1270 SIMS. Analytical procedures closely follow those of Whitehouse et al. (1999) and Whitehouse & Kamber (2005), using a c. 6 nA O 2 2 primary ion beam to sputter an elliptic crater of c. 20-25 mm. Secondary ions were measured at a mass resolution sufficient to resolve Pb ions from molecular interferences (M/DM c. 5400) using a single ioncounting electron multiplier and a peak hopping routine. All analyses were run in fully automated chain sequences. The U/Pb ratio calibration was based on regularly interspersed analyses of the 91500 zircon standard. Common lead corrections were made using the terrestrial Pb-isotope composition of Stacey & Kramers (1975) when the count rate of 204 Pb exceeded detection limit (3 × standard deviation of the detector background over the analytical session). Ages were calculated with Isoplot (version 3.16; Ludwig 2004) using the decay constant recommendations of Steiger & Jäger (1977). Ages are presented at the 95% confidence (or 2s) level and, when concordia ages are calculated, the quoted mean square of weighted deviates (MSWD) value is that of combined concordance and equivalence following the recommendation of Ludwig (2004).
All analyses were subject to detailed postanalytical optical microscope and CL imaging. In cases where the ion beam had clearly sampled two distinct growth zones or partially sampled epoxy, these analyses have been excluded from consideration and are not reported.

Results and age interpretations
SIMS analytical data are presented in Table 2. All results are plotted as 207-corrected ages (Ludwig 2004) in which analyses are coded into their broadly identified internal structure groups based on CL imaging; those which appear to have experienced lead loss are highlighted (Fig. 4). Data selected for concordia age calculations are presented (Fig. 5) using inverse concordia diagrams ( 207 Pb/ 206 Pb v. 238 U/ 206 Pb). All age uncertainties are quoted at 2s (or 95% confidence level as appropriate) and the MSWD for concordia ages is that of combined concordance and equivalence.

Sadh complex: granitic gneiss (OM05-14)
The majority of zircons from the granitic gneiss are translucent, inclusion-free, pinkish and euhedral to subhedral. The zircons range in length from c. 50 to 300 mm and have aspect ratios (length: width) of 2:1 to 4:1; only grains .100 mm in length were picked for analysis. A few zircons in this group have dark inclusions. Subordinate populations of euhedral brownish zircons with dark inclusions as well as rounded colourless to pink crystals also occur, but were not selected for analysis because their internal growth zones were generally smaller than the ion probe spot. We attribute rounding of the zircon to partial resorbtion during migmatization (e.g. Hoskin & Black 2000), particularly in grains that were likely present in the thin leucosome layers. CL imaging reveals oscillatory zoned zircon interiors that are generally considered (e.g. Corfu et al. 2003) to be characteristic of a magmatic origin (Fig. 3a). The oscillatory zoning in some zircon grains is cut by a homogenous phase (Fig. 3a). A total of 23 analyses are reported with U concentrations in the range 26-1231 ppm (Table 2). On the cumulative 207-corrected age plot of the individual analyses, the youngest ages are attributed to lead loss and therefore excluded from the concordia age calculation (Fig. 4a). Four analyses of CLhomogeneous regions are ambiguous with respect to magmatic CL morphology and are therefore excluded from the calculation. In addition, two analyses (one with high common lead, the other with a large analytical error; Table 2) are also excluded from the concordia age calculation. The remaining 15 analyses of zircon interiors combine to yield a concordia age of 831 + 7 Ma (MSWD ¼ 1.4; Fig. 5a), interpreted as the crystallization age of granitic gneiss protolith.

Sadh complex: biotite gneiss (OM05-25)
Zircons from the biotite-rich gneiss range in length from 100 to 250 mm and in aspect ratio from 2:1 to 6:1. The majority of the zircon grains are subhedral to occasionally euhedral, pink in colour, with a few dark inclusions. As with sample OM05-14, a subordinate population of rounded zircons (mostly translucent and colourless) are attributed to partial resorbtion during migmatization; due to their small size and/or complex internal structures, these grains were not analysed. The larger grains often contain inclusions. CL images reveal zoned central portions with oscillatory zoned rims. It is unclear whether the latter represent a new zircon growth phase or a continuum in the same magmatic event as there are no clear rim/core truncations (Fig. 3b).
A total of 23 analyses were made. U concentrations vary within the range 29-689 ppm ( Table 2). On the 207-corrected age plot there are six analyses with apparently younger ages, which are interpreted to represent Pb-loss (Fig. 4b). Omitting these and the three possible 'rims' yields a concordia age of 835 + 6 Ma (MSWD ¼ 1.2; Fig. 5b), interpreted as the magmatic protolith age. This group selected for calculation also includes two analyses which might have partially sampled across a 'core/rim' boundary (Table 2) although, given the absence of unambiguous evidence for rim growth outlined above and the dominance of the 'core' portion in these analyses, they have been retained in our preferred age. Their omission results in a statistically indistinguishable concordia age of 833 + 6 Ma (MSWD ¼ 1.0; not plotted). There are too few unambiguous 'rims' to determine a meaningful concordia age, but a weighted average 207-corrected age of the three 'rim' analyses (Table 2, Fig. 4b) gives an age of 821 + 16 Ma (MSWD ¼ 0.24) which might indicate the time of leucosome genesis.

Sadh complex: hornblende gabbro/diorite (OM05-26)
Zircons from OM05-26 are c. 50-180 mm wide and c. 50 -430 mm long, with aspect ratios of 2:1 to 3:1. The majority of the zircons in this sample are broken, elongate, inclusion-rich, CL-dark fragments with a cloudy interior which were unsuitable for analysis. The nine analysed zircons are moderately CL-bright fragments with a more homogeneous appearance (Fig. 3c). The uranium concentrations vary from 166-3066 ppm. On the 207-corrected age plot there is a clear tendency to Pb-loss in the youngest grains (Fig. 4c). Of the remaining analyses, one is discordant and five analyses combine to yield a concordia age of 796 + 14 Ma (MSWD ¼ 1.7; Fig. 5c). This age is within error of the reported ages from the Sadh complex and is interpreted to indicate the age and time of intrusion of the hornblende gabbro/diorite.

Sadh complex: intermediate gneiss (OM05-39)
This sample contains predominantly inclusion-free, pink, subhedral to rounded zircons. Subordinate rounded zircons vary from colourless to pink and brown while large brown subhedral zircon crystals contain dark inclusions and cracks. Zircon lengths range from 50 to 500 mm and aspect ratios from 2:1 to 5:1. The zircons selected for analysis are between 100 and 250 mm long. The crystals exhibit complex structures in CL such as irregular featureless (homogeneous) both CL bright and dark areas and areas with non-planar zoning. The external part of one zircon (Fig. 3d, grain 14) shows convoluted ('folded') zoning. These textures are typical of metamorphic zircon (e.g. Hoskin & Schaltegger 2003). There are also subhedral zircons with rounded or irregular, weakly zoned cores cut by CL bright irregular homogeneous areas, sometimes overgrown by a zoned or homogeneous, featureless overgrowth. There are only a few convincing truncated rim/core relationships (e.g. grains 8 and 11, Fig. 3d) that suggest a true later metamorphic overgrowth v. magmatic development during a single growth phase in these zircons.
A total of 18 analyses are reported and record a range in U concentration from 51 to 516 ppm (Table 2). On the cumulative 207-corrected age plot there is again a clear tendency towards younger ages which are interpreted as Pb-loss (Fig. 4d). Excluding the four younger analyses as well as the two oldest obviously truncated cores, the ages of the remaining analyses are indistinguishable (within 2s error) from each other and combine (n ¼ 12) to yield a concordia age of 816 + 5 Ma (MSWD ¼ 0.9; Fig. 5d). This age is interpreted as dating leucosome formation in the intermediate gneiss.

Sadh complex: felsic vein (OM05-38)
The zircons from this felsic vein range in length from c. 100 to 250 mm and have aspect ratios from 2:1 to 5:1. The majority are inclusion-free, pink, translucent and subhedral to euhedral. A few zircons have black inclusions and one pink-brownish zircon was observed. CL-imaging reveals cores that are either oscillatory zoned or CL-dark and homogeneous (Fig. 3e). Zircon cores are mantled by CL-bright, irregular regions and younger, mainly zoned, outer rims. In many cases, structures in the zircon cores are clearly truncated by the rims (Fig. 3e), suggesting that the cores are an inherited, partially resorbed, older zircon component.
A total of 46 analyses are reported, differentiated on the basis of CL images into centres/cores and outers/rims (Table 2). Both groups have a similarly wide range in U concentration from a few tens to several hundreds of parts per million, while the cores generally have higher Th/U ratios from 0.25 to 2.3 compared to the range in the rims of 0.04 -0.44. On the cumulative 207-corrected age plot (Fig. 4e) there is a marked tendency towards younger ages in both cores and rims which is attributed to Pb-loss, although the oldest 12 207-corrected ages are consistently represented by cores. These oldest core analyses yield a concordia age of 841 + 6 Ma (MSWD ¼ 1.5; Fig. 5e) after omission of a single, slightly reverse discordant point that is possibly over-corrected for common Pb. The remaining six core analyses reflect Pb-loss.
Defining an age from 28 analyses of zircon rims is complicated by the apparent pervasive but subtle Pb-loss affecting this group (Fig. 4e). The approach taken here is to combine minimization of the MSWD value with maximization of the number of grains included. The most inclusive concordia age that can be calculated from this group is 810 + 5 Ma (MSWD ¼ 1.8; not plotted) based on the oldest 19 analyses in this group. As shown in Figure 6, successive rejection of the youngest analysis results in a marked lowering of MSWD with values of c. 1 obtained from populations of between 14 and 8 analyses. These correspond to concordia ages of 815 + 4 Ma and 819 + 5 Ma respectively, with the younger age shown in Figure 5f preferred because of its higher inclusivity of analyses. The oscillatory zoned nature of most of these rims as well as their clear truncation of core structures supports an interpretation of the rim age as dating crystallization of the felsic vein with the cores representing an inherited component, possibly from more than one source given their wide Th/ U range.

Marbat granodiorite (OM05-33)
Zircons from this sample are predominantly euhedral, translucent and colourless to pink, with no visible inclusions. Occasional subhedral to rounded grains commonly have sharply faceted, very thin overgrowths (e.g. Fig. 3f, grain 11). There are also a few brownish zircons which contain inclusions. Zircon lengths range from c. 70 to 375 mm, with aspect ratios from 1:1 to 4:1. CL imaging reveals cores with a range of morphologies, from uniform and rounded to zoned or aggregate (Fig. 3f). Oscillatory zoning varies from broad to narrow and is considered to indicate a magmatic origin. Outer regions also display oscillatory zoning and are similarly considered magmatic in nature (Fig. 3f).
A total of 28 analyses are reported with uranium concentrations ranging from 36 to 495 ppm. In a plot of 207-corrected ages, the three youngest analyses show clear evidence for a Pb-loss younging trend ( Fig. 4f; Table 2) with the remaining 25 analyses falling into three groups separated by small steps in age. While CL-images do not clearly distinguish cores and rims, the six oldest analyses define a distinct group with a concordia age of 853 + 10 Ma (MSWD ¼ 1.1; not plotted). The next 14 youngest analyses (group 2, Fig. 4f) combine to yield a concordia age of 826 + 7 Ma (MSWD ¼ 1.2; Fig. 5g) while the final group of 5 analyses (group 3, Fig. 4f) yield a concordia age of 791 + 10 Ma (MSWD ¼ 1.1; Fig. 5h). Given the clearly magmatic nature of all internal zircon structures revealed in CL and the presence of older, likely xenocrystic, grains (probably from the adjacent Sadh gneiss complex), assignment of the crystallization age of the Marbat granodiorite to either of the two younger groups is problematic. This is considered further in the following discussion.

Refined geochronology of the Marbat region
A summary of the U -Pb zircon geochronological data obtained in this study is presented in Figure 7. Two of the investigated samples from the Sadh gneiss complex have magmatically zoned zircon cores that yield broadly similar U/Pb concordia ages and are interpreted as protolith ages, namely granitic gneiss (OM05-14) at 831 + 7 Ma and biotite gneiss (OM05-25) at 835 + 6 Ma. Additionally, the inherited cores of zircons analysed from the intruding felsic vein (OM05-38) and the Marbat granodiorite (OM05-33) yield ages of 841+6 and 853 + 10 Ma, respectively. These ages are similar to those of crystalline rocks documented elsewhere in Oman (Al Jobah, Jabal Akhdar; Bowring et al.  Fig. 1a), which was proximal to the Marbat region prior to Oligocene opening of the Gulf of Aden. Of the gneissic samples, the intermediate gneiss (OM05-39) appears anomalous in yielding zircon with a somewhat younger age of 816 + 6 Ma. The hornblende-dominated melanosome of this banded migmatitic gneiss however suggests a rather mafic precursor, which may not have contained much zircon; the analysed zircons therefore likely derive from the thin (millimetrescale) leucosome layers and date partial melting rather than the age of the gneiss protolith. Rims on zircons from the biotite gneiss (OM05-25) and felsic vein (OM05-38) yield ages of 821 + 16 and 815 + 4 Ma respectively and are similar to the previously mentioned 816 + 6 Ma age of likely migmatite-generated zircon in the intermediate gneiss . Together, these c. 815-820 Ma ages are interpreted as representing a period of partial melt genesis and emplacement associated with penetrative deformation(s). On the basis of field relationships alone, the folded felsic vein must be younger than the formation of migmatitic leucosome in the intermediate gneiss it intrudes and this age relationship is permissible at the 2s uncertainty level of the ages obtained.
The Marbat granodiorite (OM05-33), which demonstrably cross-cuts the Sadh gneiss complex, preserves a primary magmatic texture and lacks evidence for penetrative deformation in both outcrop and thin-section. Setting aside the c. 850 Ma clearly xenocrystic group mentioned above, assignment of the crystallization age to either the 826 + 7 Ma or 791 + 10 Ma age groups obtained from the analysed zircon population must be consistent with the ages observed in the Sadh gneiss complex and the field relationships. Critical to this interpretation from our new data is the 815 + 4 Ma age obtained from the late-intrusive but deformed felsic vein (OM05-38). At the 2s uncertainty level, the 826 + 7 Ma age from the Marbat granodiorite does not overlap with the crystallization age obtained from the felsic vein (Fig. 7) and is too old to reconcile with the observed deformational state of both samples. While obtained on a smaller subset of the analysed zircon grains, the younger age of 791 + 10 Ma accords better with observed field relationships; these also require the Marbat granodiorite to post-date the folded Mahall complex intrusions into the Sadh gneisses, dated by conventional U -Pb zircon ages of c. 800 Ma (Table 1; Mercolli et al. 2006). In this favoured age interpretation, the presence of xenoliths of apparent Sadh complex in the Marbat granodiorite (some of which are partially resorbed; Fig. 2d) may explain the preponderance of xenocrystic zircon in the analysed population.
Five zircon fragments from the hornblende gabbro/diorite yield an age of 796 + 14 Ma and are tentatively suggested to document a postdeformation intrusion age (Fig. 7). Younger K -Ar and Rb -Sr ages for the Marbat granodiorite (Mercolli et al. 2006; Table 1), previously interpreted to place it within a suite of late (750 -770 Ma) intrusions into the basement, are reinterpreted to reflect cooling ages.

Crustal evolution and tectonic implications
The oldest component of the Marbat basement is the meta-sedimentary Juffa gneiss complex, which records T DM model ages in excess of 1.3 Ga and metamorphic zircon at c. 815 Ma (Mercolli et al. 2006). Our new zircon geochronology from the predominantly meta-igneous Sadh complex records Cryogenian magmatic and metamorphic events occurring during the relatively short period of crustal evolution in the Marbat region of c. 850-780 Ma. The calc-alkaline nature of the Sadh complex (Platel et al. 1987b;Würsten 1994;Briner 1997) together with the previously documented absence of evolved Nd isotopes (Mercolli et al. 2006) is consistent with an episode of juvenile oceanic island-arc magmatism between c. 850 Ma (the age of the oldest recorded zircon cores) and c. 800-780 Ma. The latter period includes intrusion and folding of magmatic rocks of the c. 800 Ma Mahall complex into the Sadh gneiss complex, as well as the likely age of intrusion of the undeformed Marbat granodiorite (this study) and the Fusht and Hadbin complexes (Mercolli et al. 2006). Deformation and syntectonic migmatization, evidenced by leucosome accumulation in lowstrain areas (e.g. fold hinges, boudin necks and in cracks) occurred at c. 815-820 Ma and probably records a period of arc amalgamation in the region. The coincidence of these ages with those of metamorphic zircon growth in the Juffa gneiss complex supports the suggestion (Mercolli et al. 2006) that the Juffa and Sadh complexes were a contiguous crustal block by at least this time, if not earlier.
The Marbat region basement rocks experienced retrograde greenschist facies metamorphism associated with cooling, uplift and erosion followed by the unconformable deposition of the Marbat Sandstone Formation in late Neoproterozoic time (Bowring et al. 2007;Rieu et al. 2007). While the precise time of deposition of the Lower Member/ Ayn Formation (Marbat Sandstone Formation) remains uncertain (e.g. Bowring et al. 2009), peaks at c. 810, 840 and 870 Ma in the detrital zircon spectrum (Rieu et al. 2007) closely correspond to the ages presented here and are consistent with relatively local derivation.
Our new data from the Marbat region may be viewed in the broader context of models for the Neoproterozoic evolution of the ANS, which suggest that break-up of the supercontinent Rodinia at c. 870 Ma (Li et al. 2008) was followed by a prolonged period of predominantly juvenile arc generation at c. 870-690 Ma, final arc collision and accretion at c. 630 -600 Ma and uplift and cooling at c. 600 -540 Ma . Compared to this long period of documented events, the Marbat region records a short period of juvenile arc accretion and deformation that is temporally correlative with similar events in the ANS. The absence of any substantially older inherited zircons (c. .850 Ma, this study) together with documentation of juvenile Nd model ages (Mercolli et al. 2006) shows that at least this small part of the east Arabian basement exposed in the Marbat region has more in common with the older juvenile western arc terranes of the Arabian shield in Saudi Arabia than with the younger but more evolved eastern arc terranes of Saudi Arabia, or with the continental terranes exposed at the eastern margin of the ANS (Khida, Saudi Arabia and Yemen terranes). This may reflect arc accretion at the opposing margin of the Mozambique Ocean, that is, along East Gondwana. However, the complete absence of basement exposure across more than 800 km continues to frustrate attempts to provide more than a simple chronological and petrogenetic correlation between the eastern Arabian basement and the exposed ANS.

Conclusions
New zircon U -Pb (SIMS) geochronology ages from basement rocks in the Marbat region, southern Oman, indicate that the earliest phase of crustal development began at c. 850 Ma and continued until at least c. 830 Ma as recorded by protolith ages of calc-alkaline rocks (deformed to gneisses) in the Sadh gneiss complex and xenocrystic zircon in the Marbat granodiorite. At c. 815-820 Ma pervasive deformation and partial melting formed migmatites in the Sadh complex. This was followed by intrusion of the Mahall complex bodies at c. 800 Ma (Mercolli et al. 2006), which were deformed prior to intrusion of post-tectonic c. 790 -800 Ma hornblende gabbro and Marbat granodiorite. The zircon age interpretation of the Marbat granodiorite is complicated by apparent extensive contamination from the host Sadh complex gneisses, but is nonetheless considerably older than previously inferred from other geochronometers (Mercolli et al. 2006) which likely record cooling of the region. Together with previously published juvenile Nd model ages from the Sadh complex, the .850 Ma to c. 790 Ma evolution of the Marbat region is consistent with development of an island arc, similar in age and nature to some of those of the Arabian Nubian Shield. Direct correlation is neither implied nor possible to demonstrate however, given the .800 km separation of the Marbat region from the nearest shield outcrops.
Subsequent cooling and uplift of the Marbat region was accompanied by intrusion of the Leger granite and the Shaat dyke swarm at c. 770-726 Ma (Mercolli et al. 2006;Bowring et al. 2007), with cessation of retrograde greenschist facies metamorphism at c. 718 Ma recorded by a biotite K -Ar age (Mercolli et al. 2006). This final uplift and cooling of the Marbat region considerably predates that of the southeastern Arabian Shield at c. 600 Ma . This may indicate that arc accretion in the eastern Arabian Peninsula reflects earlier juvenile crust formation in the Mozambique Ocean occurring close to or at the margin of East Gondwana. Reviews by Y. Be'eri-Shlevin and an anonymous referee, comments by J. P. Liégeois and R. J. Stern on an earlier version of this manuscript and discussions in the field with M. Ba-Bttat are gratefully acknowledged, as is technical assistance from staff at the Institute for Geological Sciences, Stockholm University and at the Swedish Museum of Natural History. This project was supported by grants from the Swedish Research Council (VR) to VP and MJW and from Stiftelsen Stockholms geologer (SSG) to NR. The Knut and Alice Wallenberg Foundation is acknowledged for financial support of the IGV SEM and the NordSIM facility. NordSIM is jointly funded by the Nordic countries. This is NordSIM contribution number 333.