Capturing the Mesoarchean Emergence of Continental Crust in the Coorg Block, Southern India

The emergence of Earth's continental crust above sea level is debated. To assess whether emergence can be observed at a regional scale, we present zircon U‐Pb‐Hf‐O isotope data from magmatic rocks of the Coorg Block, southern India. A 3.5‐Ga granodiorite records the earliest felsic crust in the region. Younger phases of magmatism at 3.37–3.27 and 3.19–3.14 Ga, comprising both reworked crust and juvenile material, record successive crustal maturation. We interpret an elevation in δ18O through time as an increase in both the amount of sediment recycling and hence, crustal thickening, as well as an increase in the emerged area of continental crust available for weathering. Geochemical signatures do not point to any apparent change in geodynamic regime. We interpret the isotopic evolution of these rocks as solely reflecting regional emergence and thickening of the continental crust, assisted by the increasing strength of the lithosphere.


Introduction
There is a continuing debate over both the onset of modern-day-style plate tectonics and the emergence of continental crust above sea level. The onset of modern-day-style plate tectonics has been long debated. Estimates range from the Hadean (Harrison, 2009) to the Neoproterozoic (Stern, 2008), but many occur in the range of 3.2 to 2.5 Ga (see Hawkesworth et al., 2018;Roberts & Spencer, 2015). Although it is common to quote a single age as a transitional point in Earth evolution, for example, 3.0 Ga (Dhuime et al., 2012), it is more likely that the gradual secular cooling in mantle temperatures may have led to a protracted transition from preplate tectonic to plate tectonic geodynamic regimes (Condie, 2016;Spencer et al., 2017). Similarly, the emergence of large areas of continental crust above sea level is also predicted to occur around the late Archean, based on both numerical models (Flament et al., 2008;Vlaar, 2000), and geochemical evidence (Pons et al., 2013;Szilas et al., 2016). However, numerical models for continental emergence are built on several wide-ranging assumptions such as Earth's thermal evolution and continental crustal growth rate and have wide-ranging estimates from the Hadean to the Neoproterozoic (Lee et al., 2018;Maruyama & Ebisuzaki, 2017).
Our knowledge of early Earth geodynamics has been significantly improved through the advent of complex numerical models (e.g., Fischer & Gerya, 2016;Rozel et al., 2017;Sizova et al., 2010); however, these models require ground truthing within the limited rock record. There are several cratons that preserve Earth's Eoarchean to Mesoarchean geological history, and these indicate that both subduction-dominated and plume-dominated processes occurred from at least the Mesoarchean, with different cratons being dominated by different settings (e.g., Van Kranendonk, 2010). Detrital mineral records provide an important archive of the continental crust, but removal from their source means they lack accompanying geochemical ROBERTS AND SANTOSH 7444 and geophysical constraints. Here we present a combined U-Pb-Hf-O zircon isotope data set of Palaeoarchean to Mesoarchean magmatic rocks from the Coorg Block of southern India. We interpret a systematic increase in zircon δ 18 O VMSOW through time, coincident with an increasing range in zircon 176 Hf/ 177 Hf i and Th/U ratios, as reflecting a combination of a regional increase in the area of continental emergence and of increasing crustal thickening and burial, from 3.5 to 3.1 Ga.

Geological Setting
Southern India hosts a belt of crustal blocks that represent an arc accretionary belt formed during the Mesoarchean to Neoarchaen (Jayananda et al., 2015(Jayananda et al., , 2018Peucat et al., 2013;Santosh et al., 2015Santosh et al., , 2016 and accreted to the Dharwar Craton at circa 2.5 Ga (Figure 1). The Coorg Block, the northwesternmost arc terrane, comprises the oldest known rocks from this belt; it therefore provides an ideal locality to study early crust formation. The Coorg Block is dominated by upper amphibolite to granulite-facies charnockite and tonalite-trondhjemite-granodiorite gneisses but also contains less abundant mafic-ultramafic rocks, now in the form of tremolite-actinolite schists (Santosh et al., 2015. Metasediments are rare within the core of the block but are found as tens-of-meter-thick bands of khondalites (granulite-facies metapelites) along its periphery; these are younger than the magmatic rocks based upon their detrital zircon ages . Based on the lithological and geochemical characteristics, Santosh et al. (2015Santosh et al. ( , 2016 interpret the block to represent the midcrust of a continental arc.

Samples and Methods
Seven metaigneous samples previously described by Santosh et al. (2015Santosh et al. ( , 2016 were selected for detailed work. Zircon was analyzed in the following order, oxygen isotopes (Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing), U-Pb isotopes (Natural Environment Research Council Isotope Geosciences Laboratory, Nottingham), and Lu-Hf isotopes (Natural Environment Research Council Isotope Geosciences Laboratory). Detailed methods and the data set are located in the supporting information (Fielding et al., 2017;Horstwood et al., 2016;Li, Long, et al., 2010Spencer et al., 2014).

U-Pb Geochronology
Crystallization ages of the seven samples range from circa 3502 to 3140 Ma, overlapping those previously obtained by Santosh et al. (2015Santosh et al. ( , 2016. We split the samples into Phase 1, comprising a 3502-Ma granodiorite gneiss; Phase 2, comprising three samples from circa 3.37 to 3.27 Ga (charnockite and granodiorite); and Phase 3, comprising three samples at circa 3.19 to 3.14 Ga (charnockite and a meta-tuff). Isotope data are plotted with individual 207 Pb/ 206 Pb ages, since the LA-ICP-MS method lacks resolution at this age range to confidently determine magmatic, anticrystic, and xenocrystic zircons and to derive a precise crystallization age of each unit. ZirconTh/U ratios are shown in Figure 2a. Although the discriminator of magmatic versus metamorphic zircon is arbitrary and not always accurate (Rubatto, 2017), the data show a general trend of an increasing range of Th/U values through time, importantly with a larger population of low Th/U zircon domains (metamorphic) within Phase 3 magmatism.

Lu-Hf Isotopes
Hf isotope data are plotted in Figure 2b as initial age-corrected 176 Hf/ 177 Hf ratio ( 176 Hf/ 177 Hf i ) versus 207 Pb/ 206 Pb age. For comparison, also shown are evolution trends of Lu/Hf = 0, 0.012, and 0.029. The horizontal evolution (Lu/Hf = 0) is equivalent to lead loss. The trend of Lu/Hf = 0.012 corresponds to the estimated value of average mafic tonalite-trondhjemite-granodiorite gneiss-forming crust (Gardiner et al., 2018). The trend of Lu/Hf = 0.029 corresponds to the value required to produce the most radiogenic values of Phases 2 and 3 magmatism from a source equivalent to the average of Phase 1.
The oldest sample (Phase 1) represents juvenile magmatism with an average 176 Hf/ 177 Hf i of 0.28062. Younger zircon domains in this sample likely reflect lead loss during subsequent magmatic phases. This sample represents the earliest formed felsic crustal component, derived directly or through a multistage melt differentiation process from the earliest formed mafic protocrust. Of Phase 2 magmatism, the circa 3.37-Ga rocks were derived mostly from reworking of the early formed protocrust, assuming a composition similar to Phase 1 magmatism with an Lu/Hf around 0.012. The 3.27-Ga sample is likely recording renewed juvenile magmatic addition since it lies at a higher 176 Hf/ 177 Hf i . Phase 3 magmatism may also represent reworking of the earliest formed 3.5-Ga protocrust, if such crust is mafic in composition with a high Lu/Hf ratio. However, most data plot above the evolution trend displayed (Lu/Hf = 0.012) and therefore reflect some juvenile input to the crust at this time. A spread in 176 Hf/ 177 Hf i equivalent to 10 epsilon units across individual Phase 3 samples likely reflects mixed crust-mantle magmatism, although disequilibrium melt processes cannot be excluded (e.g., Tang et al., 2014;Wang et al., 2018). Data from previous studies of these and other samples from the Coorg Block (see Figure 2b) overlap those presented here, although with slightly larger scatter.
We calculated simple two-component mixture modeling to provide an indication of the potential mantle contributions through time. A full description of the end-member compositions and results are given in the supporting information. We assume that Phase 1 presents mantle-derived magmatism, although it is not constrained whether the first mafic protolith that melted to form this granodiorite was extracted around the crystallization age of circa 3.5 Ga or earlier (the two-stage Hf model age is 3.8 Ga). The juvenile Phase 2 sample (PKD-1) corresponds to a mixture of 50% depleted mantle and 50% Phase 1 protocrust. The most juvenile Phase 3 sample (TKLD-1) requires a mixture of 37% depleted mantle and 63% Phase 1 protocrust. The least juvenile Phase 3 sample (NDK-1) requires 15% depleted mantle.

Oxygen Isotopes
Oxygen isotope data are plotted in Figure 2c as zircon δ 18 O VMSOW versus 207 Pb/ 206 Pb age, choosing only the concordant (to <10%) data. For comparison, we plot the mantle composition of +5.3 ± 0.3‰ (Valley et al., 1998). The oldest samples fall exactly on this mantle composition, averaging +5.26‰. The next phase of magmatism (at circa 3.37 to 3.27 Ga) records an increase in both the mean and maximum of the δ 18 O VMSOW values to +5.96‰ and +6.37‰, respectively. The youngest phase of magmatism records another increase in the mean and maximum to +6.28‰ and +6.89‰, respectively. The dashed gray line corresponds broadly to the increasing maximum values through Phase 1 to Phase 3 magmatism. Values greater than the mantle range reflect a contribution from material that has been altered at low temperature, that is, weathering at or near the Earth's surface (see Valley, 2003). In magmatic rocks this  can be achieved by incorporation of weathered sediments to the magma source. The data presented here, at face value, therefore correspond to an increasing sedimentary component to the magmatism through time.

Geochemistry
Petrography, mineral chemistry, and whole-rock geochemistry of these samples were discussed in detail in Santosh et al. (2015Santosh et al. ( , 2016; notable features are their metaluminous nature, low K 2 O/Na 2 O ratios, enriched light rare earth elements and large-ion lithophile elements, and low Nb/Zr. These authors ascribed the rocks to typical continental arc rocks. Figure 3 presents primitive mantle (Sun & McDonough, 1989) normalized trace elements and a Quartz-Alkali Feldspar-Plagioclase classification plot; notably, all samples exhibit broadly similar large-ion lithophile element and light rare earth element enrichment, and distinctive negative Nb anomalies. These characteristics are typical of magmas formed in a suprasubduction type setting; however, it is well documented that such characteristics are not exclusive and may represent hydrous melting of mafic rocks in other settings (e.g., Johnson et al., 2017;Moyen, 2011;Moyen & Laurent, 2017;Nagel et al., 2012). The notable feature is that there is no apparent change in geochemical signature that may also reflect a change in geodynamic regime, that is, from plume-derived to subduction-derived magmas.

Magma Evolution
Magmatism in the Coorg Block records circa 0.4 Gyrs of crustal evolution. We interpret the granodioritic gneiss at 3.5 Ga as representing a rare record of the earliest felsic crust formation along this entire accretionary belt. Juvenile hafnium and mantle-like oxygen isotope signatures of this sample are compatible with its formation from a mantle-derived protolith. Hf isotope data of subsequent magmatic rocks imply that this protolith formed the crustal substrate for successive periods of magmatism to be formed within and upon. A significant range in Hf isotope values for Phases 2 and 3 magmatism suggests mixed source contributions, that is, both reworked older crust and more juvenile material. This suggests a maturing crust through time. The oxygen isotope signatures record a similar maturation through, ranging from pure mantle, to increasingly elevated signatures above the mantle average, implying an increasing sedimentary component to magmatism through time. Conversely, the geochemistry does not record any apparent change through time.
The combined zircon isotope and geochemical data lend support to a simple model involving crustal maturation through time, from early-formed mantle-derived intermediate-felsic magmatism to later mixed crust-mantle felsic magmatism, with crust-mantle interaction and sedimentary incorporation increasing during each period of magmatism. Because the geochemical signatures do not show any marked change, we speculate that each magmatic phase is formed in a similar geodynamic regime. In concordance with the interpretation of previous studies (Santosh et al., 2015, we interpret this to be related to a subduction setting. However, we accept that subduction may have differed in its mechanics and that crust formation and reworking in a vertically dominated tectonic setting is also permissible (Johnson et al., 2017). To account for the evolving magmatic compositions through time, we present a unifying model below and depicted in Figure 4.

A Model of Crustal Maturation and Continental Emergence
Phase 1 of the model represents formation of the earliest felsic crust. The Phase 1 granodiorite likely formed from melting of a mafic protolith, which may have been extracted from the mantle at 3.5 Ga or slightly earlier (<3.8 Ga). Phase 2 represents maturation into an ensialic crust (a volcanic arc?) with a range of compositions. Phase 3 represents further maturation into a thicker crust, with a potential role of crustal accretion. We interpret the oxygen isotope data to provide important constraints on emergence of this continental crust. Phase 1 lacks any sedimentary input and is therefore compatible with submergence. Phase 2 has an increasing sedimentary component; therefore, the crust has thickened enough to allow for incorporation of sediment into the melts through burial. To produce weathered sediments in the local area, we interpret the crust as being emergent above sea level. Phase 3 records a further increase in sedimentary incorporation and therefore increased crustal thickening and burial. The Th/U data lend additional support to increasing tectonothermal activity during Phase 3, and this is further supported by a circa 3.0-Ga age of metamorphism in the northern periphery of the Coorg Block, recording its accretion to Dharwar crust to the north (Amaldev et al., 2016). Such crustal-accretion tectonic processes would drive increasing sedimentary incorporation into mid-to deep-crustal magma source regions through burial. Crustal thickening during/preceding Phase 3 would also lead to further continental emergence, which would subsequently drive an increase in the δ 18 O of the regional sedimentary volume through increased weathering. According to our model, several parameters are evolving simultaneously: crustal thickness, continental emergence, and tectonothermal activity. Crustal thickness and continental emergence are linked but not directly correlative, as they are controlled by global hypsometry. It is difficult to disentangle the signals of crustal thickness versus continental emergence from geochemistry alone. Although crustal thickness can be estimated from magma compositions (Chiaradia, 2015;Mantle & Collins, 2008;Profeta et al., 2015), this requires a large geochemical data set, and it is unclear how applicable these correlations will be for Archean rocks.
The evolving magmatic oxygen isotope signatures could reflect changes in either one or more parameters. Firstly, the δ 18 O signature of the sedimentary component could be static through time, in which case the data would likely reflect an increasing sedimentary contribution to each phase of magmatism. Alternatively, the δ 18 O signature of the sedimentary component may be increasing through time, in which case the amount of sediment incorporated in the magmas may be static. Although it is difficult to assess this precisely, particularly without the ability to sample the sedimentary component locally, the global data set may provide some insight. Global oxygen isotope compilations of sedimentary rocks for this period show a steady increase through time (see Figure 5a; Bindeman et al., 2016;Payne et al., 2015). Thus, the global data are compatible with more crust being weathered over time, that is, a result of increasing continental emergence. Results of simple two component mixture modeling using the mean and maximum global shale data (Bindeman et al., 2016) are shown in Figure 5a. The results show that using either the mean or maximum, the required contribution from the sedimentary component is increasing for each phase of magmatism (see . Cartoon tectonic model highlighting the key changes in continental composition during each magmatic phase. It should be noted that the behavior of subducting lithosphere versus upwelling mantle is speculative for Archean tectonics, and thus, we place little emphasis on this aspect of the model. Our preferred model is that of horizontal tectonics and plate boundary convergence being responsible for magmatism, rather than a rising plume. TTG = tonalite-trondhjemite-granodiorite. supporting information for details), that is, from approximately 0% to 10.5% to 13.7% (using the maximum) or from approximately 0% to 15.5% to 21% (using the mean). We postulate that this indicates (1) an increasing sedimentary contribution through time and (2) that emergence of continental crust is increasing on both the local and global scale, elevating the δ 18 O signature of the sedimentary budget.

Global Versus Regional Signals
Our model of Coorg Block evolution implies an increase in both crustal thickness and continental emergence in the 3.5-to 3.1-Ga period. This contrasts with global models based on numerical modeling of continental emergence that are younger (Neoarchean, Flament et al., 2008;Neoproterozoic, Lee et al., 2018) but is broadly correlative to models of increasing crustal thickness around 3.0 Ga (Dhuime et al., 2015). We do not constrain the area of emerged crust, but we expect it to be large enough to affect the composition of the regional sedimentary budget. To allow for the dichotomy between the numerical models cited and our results, we suggest that (1) our data represent a regional rather than global effect, that is, not all cratons were emergent by 3.1 Ga and (2) crustal thickening and continental emergence are unlikely to be directly correlated across all cratons at the same time, that is, plume-dominated and arc-dominated regions likely have differing crustal thicknesses. The global thickening of the continental crust through the Mesoarchean (e.g., Dhuime et al., 2015) may reflect an increase in a particular tectonic process (i.e., subduction); however, we suggest that this could also be the result of an increase in the strength of the lithosphere, due to decreasing mantle potential temperatures (Herzberg et al., 2010), so that it can support higher topography through time .
The fact that some cratons likely formed through plume-dominated processes, whereas others are dominated by arc processes (e.g., Van Kranendonk, 2010) argues for heterogeneity amongst cratonic evolution. Of note, zircon oxygen isotope data supporting Mesoarchean continental emergence exist from other cratons. The Pilbara craton shows a marked increase in zircon δ 18 O at 3.2 Ga (Van Kranendonk et al., 2015); these authors interpret this change to mark a shift in global geodynamics whereby modern-day-style steep subduction and associated crustal recycling occurs after 3.2 Ga. In West Greenland, the same 3.2-Ga timeframe is also interpreted as the onset of modern-accretionary-style plate tectonics, based on an increase in zircon δ 18 O and a change to wide-ranging zircon ɛHf-reflecting both mantle input and crustal reworking (Naeraa et al., 2012). Importantly, these interpretations are different to our model, whereby we do not infer a change in (regional) geodynamics through the 3.5-to 3.1-Ga period.
Compilations of detrital zircon data with higher than mantle values through the Eoarchean to Neoarchean would imply that our regional data, and that of the cratons mentioned above, are not globally significant (see Figure 5b). These data would imply that significant continental reworking of altered sediments was prevalent since the dawn of the Archean (Ge et al., 2014). However, it should be noted that not all detrital data confidently represent primary values for their isotopic signatures. Oxygen isotope signatures can be modified by fluid alteration at high temperature, and this may not be recognized in all detrital data sets (Roberts et al., 2018). Some detrital data sets, particularly those form Jack Hills, have been repeated and scrutinized (e.g., Whitehouse et al., 2017), and therefore, the interpretation of sedimentary incorporation is likely to have some validity (e.g., Cavosie et al., 2005). However, splitting the published data into igneous and detrital zircons for the 3.6-to 3.0-Ga time period, shows that in general, the igneous data mostly fall within the constraints recorded by the Coorg Block. We postulate that a large number of detrital data may be erroneous, either in their age assignment or in their assignment of primary rather than secondary oxygen isotope values. We also note that none of the detrital data are taken from the Indian Craton, and thus, there are no conflicting data from this craton itself.
In summary, our comparison of literature data for zircon oxygen isotopes from detrital and igneous rocks highlights the need to compare the evolution of different cratons with different histories around the Earth using the magmatic record (cf. Fisher & Vervoort, 2018). This will allow us to better constrain what timescale global changes are occurring over and how they are represented in the global sedimentary archive. Furthermore, only by ground truthing in the preserved rock record can we test the important constraints provided by numerical models and large detrital data sets.

Conclusions
Zircon U-Pb-Hf-O isotope data from the Coorg Block in southern India record evolving crust during three phases of magmatism. The earliest felsic melt recorded at circa 3.5 Ga formed from a protolith extracted from the mantle at circa >3.8 to 3.5 Ga. Phase 2 (circa 3.37 to 3.27 Ga) and Phase 3 (circa 3.19 to 3.14 Ga) record mixed magma sources, with juvenile input during each phase. Evolving oxygen isotope signatures reflect both an increase in the incorporation of sediments and an elevation of the oxygen isotope signature of those sediments; critically, these result from increases in crustal thickening and continental emergence, respectively. No apparent geodynamic change is recorded by whole-rock geochemistry; our preferred interpretation is that all phases formed in a subduction-related setting and that crustal thickening was assisted by increasing strength of the lithosphere. The timing of continental emergence of the Coorg Block is similar to that recorded by the Pilbara Craton but may not be representative of the global record. As such, detailed magmatic records from each craton are needed to further understand the evolution of the Earth's continental crust (as well as to ground truth numerical models), rather than relying on detrital data sets alone.