Origin of the Paleoproterozoic basaltic dikes from the central and eastern Dharwar Craton and sills and volcanics from the adjoining Cuddapah Basin, southern India

Abstract

Reverse fractionation modeling considering energy-constrained assimilation-fractional crystallization is performed to estimate primary magma compositions, degree of crustal contamination, pressure–temperature of equilibrium with mantle, and potential temperatures for the origin of the Paleoproterozoic (~ 2.37–1.88 Ga) basaltic dikes in central and eastern Dharwar Craton and sills and volcanics in the adjoining Cuddapah Basin, southern India. Mineral thermobarometry indicates that the dikes crystallized at upper crustal conditions (~ 1–6 kbar/ ~ 1120–1210 °C). Hence, the reverse fractionation calculations are performed at low pressures by adding olivine + plagioclase + clinopyroxene, olivine + plagioclase and only olivine in equilibrium with melt, and simultaneously subtracting an upper crustal partial melt in small steps until the melt is multiply saturated with lherzolite at a high pressure. The results indicate that the basalts are 5–30% contaminated, and their enriched light rare earth element (REE) patterns can be attributed to upper crustal assimilation. The upper crust was pre-heated to 665–808 °C during dike emplacement. The primary magmas of all basalts were last equilibrated with spinel lherzolite at 10–16.5 kbar/1291–1366 °C, and they resemble pooled polybaric incremental melts generated along a ~ 1450 °C adiabat. The estimated mantle potential temperatures (1293–1515 °C) are similar to Paleoproterozoic ambient mantle temperatures. All basalts and their primary magmas show lower chondrite-normalized Dy_N/Yb_N ratios than the plume-derived mid-Proterozoic Mackenzie dikes of Canadian Shield, and the primary magmas show flat REE patterns indicating spinel lherzolite melting. The low estimated potential temperatures, low Dy_N/Yb_N ratios, and a spinel-bearing mantle source are at odds with an origin of the basalts from mantle plumes.

Introduction

Flood basalt volcanism is thought to result from high degrees of melting in the upper mantle caused by upwelling of a hot mantle plume from depth (Morgan 1971; Sleep 1990; Davies 1999). The hallmark of a mantle plume is a high potential temperature, T_P, defined as the temperature of the mantle if it were to adiabatically decompress and reach the Earth’s surface without melting. The excess potential temperature associated with a mantle plume relative to the ambient mantle, ∆ T_P, is ~ 100–250 °C (McKenzie and Bickle 1988; Watson and McKenzie 1991; Kinzler and Grove 1992a,b; Presnall et al. 2002; Herzberg et al. 2007; Putirka et al. 2007; Krein et al. 2021). Herzberg et al. (2010) concluded from petrological modeling that the T_P of ambient mantle was ~ 1500–1600 °C at 2.5–3.0 Ga that decreased to the present-day value of ~ 1350 °C in accordance to the Earth’s thermal history model of Korenaga (2008). High potential temperatures of ~ 1700 °C estimated for the Archean and Paleoproterozoic komatiites (Herzberg 2022, and references therein) have been used as evidence to support their origin from mantle plumes.

The origin of radiating mafic dike swarms such as the mid-Proterozoic Mackenzie swarm of the Canadian Shield has been linked to mantle plumes (Ernst and Baragar 1992; Baragar et al. 1996). Several researchers contend that the different Paleoproterozoic (~ 2.37–1.89 Ga) basaltic dike swarms intruding the central and eastern Dharwar Craton (CDC and EDC, Fig. 1) of the Indian Shield, believed to be remnants of ancient flood basalt provinces, also originated from mantle plumes (Halls et al. 2007; French et al. 2008; Ernst and Srivastava 2008; French and Heaman 2010; Kumar et al. 2012a,b; Belica et al. 2014; Mishra 2015; Stark et al. 2019). Their conclusions are based on geometric reconstructions of the piercing points of radial dike swarms across continents using geochronological and paleomagnetic data. However, considering oroclinal bending due to later tectonic deformation, Söderlund et al. (2019) showed that the pre-2.08 Ga swarms were originally linear, and the plume center reconstructions based on the current orientations of the dikes are incorrect. Furthermore, Anand et al. (2003) estimated a T_P of ~ 1500 °C for the ~ 1.89 Ga old sills and volcanics within the adjoining Cuddapah Basin that are genetically related to the ~ 1.89–1.88 Ga old EDC dikes, and explained their result by secular cooling of the Earth without invoking a plume. Sheppard et al. (2017) also presented geological arguments to preclude the involvement of a plume in the origin of the Cuddapah Basin sills and volcanics. In addition, Shellnutt et al. (2018) concluded from isotopic and trace element data that the ~ 1.88 Ga old dikes from the neighboring Bastar Craton, genetically related to the ~ 1.89–1.88 Ga old EDC dikes and Cuddapah Basin sills and volcanics, originated from a subcontinental lithospheric mantle source, not from an asthenospheric source. Srivastava et al. (2015) demonstrated the futility of trace element discrimination diagrams that often indicate incorrect or ambiguous tectonic settings. This study attempts to estimate mantle potential temperatures for the origin of the CDC/EDC dikes and Cuddapah Basin sills and volcanics using their major element compositions. The primary magmas of the basalts and their pressure–temperature (P–T) conditions of equilibrium with mantle are modeled with the reverse fractionation technique that has been previously used for mid-ocean ridge, ocean island, arc, and flood basalts (Till et al. 2012, 2013; Grove et al. 2013; Chatterjee and Sheth 2015; Till 2017; Chatterjee 2021; Krein et al. 2021). There is trace element evidence of upper crustal contamination in the basalts. Hence, the energy-constrained assimilation-fractional crystallization (EC-AFC) formulation of Spera and Bohrson (2001) and Bohrson and Spera (2001) is incorporated in the modeling that provides estimates of the degree of crustal contamination as well as the temperature of the upper crust during magmatism. Abundances of the trace elements including Ni, Rb and the rare earth elements (REE) in the primary magmas are also modeled, providing further insight into the origin the CDC/EDC and Cuddapah Basin basalts.

Fig. 1

Geological map of a part of the Dharwar Craton and Cuddapah Basin (box in inset) after Geological Survey of India (GSI 1998) with boundaries between western (WDC), central (CDC) and eastern (EDC) parts of the craton (thick dashed lines) and for sanukitoids and anatectic granites from Chadwick et al. (2000) and Jayananda et al. (2018), dike swarms (grey lines) from Halls et al. (2007), and Moho depth contours (km, red dashed lines) from Das et al. (2015). Location of the studied samples of dikes, sills and volcanics are shown (letter in parentheses after the sample names refer to sources reporting bulk compositions: a—Murty et al. 1987, b—Rao et al. 1995, c—Chatterjee and Bhattacharji 1998, d—Anand et al. 2003, e—Halls et al. 2007, f—French and Heaman 2010, g—Kumar et al. 2012b, h, i and j—Srivastava et al. 2014a,b, 2015). Abbreviations for the cratons are: AC Aravalli, BC Bastar, BuC Bundelkhand, EDC/CDC/WDC eastern/central/western Dharwar, and SC—Singhbhum

Geological setting

The Dharwar Craton (~ 600,000 km²) in southern India is one of the several Archean cratonic blocks that comprise the Indian shield (Naqvi and Rogers 1987) (Fig. 1). It is composed of ~ 3.4–3.0 Ga old tonalite-trondhjemite-granodiorite (TTG) gneisses and Neoarchean greenstone belts with basaltic volcanics that are intruded by late Neoarchean calc-alkaline and potassic granitoids (Friend and Nutman 1991; Chardon et al. 2011; Manikyamba and Kerrick 2012; Jayananda et al. 2013a,b). The oldest rocks representing the cratonic nucleus occur in the western Dharwar Craton (WDC). A steep mylonitic shear zone along the eastern margin of the Chitradurga schist belt has been traditionally considered the eastern boundary of WDC (Swami Nath et al. 1976; Gupta et al. 2003). Based on recent petrologic, geochronologic and isotopic data, the region to the east of the shear zone has been divided into the central and the eastern Dharwar cratonic blocks (CDC and EDC) along the Kolar-Kadiri-Hungund belt (Peucat et al. 2013; Jayananda et al. 2013a). Charnockites near the southern margin of CDC/EDC originated by metamorphism of magmatic protoliths at ~ 2.48 Ga as a result of a collision between the CDC/EDC block and the Southern Granulite Terrane (SGT, Ghosh et al. 2004; Clark et al. 2009).

The EDC is separated from the Eastern Ghats Belt to the east by the crescent-shaped Cuddapah Basin that contains a sequence of gently east-dipping, Proterozoic sedimentary rocks (Nagaraja Rao et al. 1987) (Fig. 1). Near the base of the sequence in the eastern part of the basin, the sediments are intercalated with basaltic pillow lavas in the Vempalle Formation, and basaltic sills and tuffs in the overlying Tadpatri Formation (Sheppard et al. 2017, and references therein). The basaltic exposures are parallel to the arcuate southwestern margin of the basin and coincide with an elliptical region of gravity high (− 55 mGal) surrounded by gravity lows (− 100 mGal) that indicate the presence of a dense (~ 3.0 g cm⁻³) lopolithic intrusion in the upper crust (NGRI 1978; Bhattacharji and Singh 1984; Bhattacharji 1987; Singh et al. 2004). A sill (1899 ± 20 Ma, Anand et al. 2003; 1885.4 ± 3.1 Ma, French et al. 2008) in the lower part and a felsic tuff (1862 ± 9 Ma, Sheppard et al. 2017) in the upper part of the Tadpatri Formation provide evidence for a protracted ~ 30 Myr period of volcanism during sediment deposition. The basin may have started forming before ~ 1.9 Ga by rifting between the EDC and the Napier complex of East Antarctica (Mohanty 2011), and it evolved into a foreland basin after collision with the Eastern Ghats Belt in the late Paleoproterozoic (Collins et al. 2015).

Paleoproterozoic mafic dike swarms of different trends and ages are widespread in the CDC and the EDC (Halls 1982; Murty et al. 1987; Radhakrishna and Joseph 1996; Poornachandra Rao 2005; Halls et al. 2007; French and Heaman 2010; Söderlund et al. 2019; Samal et al. 2021) (Fig. 1). The individual dikes can be traced in length from a few meters to hundreds of kilometers, and their widths vary between 1 m and ~ 400 m. At least nine different dike swarms have been identified (Söderlund et al. 2019; Samal et al. 2021). The dominantly ENE- to NE-trending (some ESE-trending at the SGT contact) ~ 2.37 Ga old dikes of the Bangalore-Karimnagar swarm orthogonally cut across the NNW-SSE structural grain of the craton. These dikes were emplaced within a short interval of < 5 Myr (2369–2365 Ma, Halls et al. 2007; French and Heaman 2010; Kumar et al. 2012a; Liao et al. 2019; Söderlund et al. 2019). The N- to NNE-trending ~ 2.25 Ga old dikes of the Ippaguda-Dhiburahalli swarm to the north and south of Cuddapah Basin were also emplaced within a short, ~ 6 Myr time interval (2257–2251 Ma, Nagaraju et al. 2018a; Söderlund et al. 2019). The N- to NNW-trending ~ 2.22 Ga old dikes of the Kandlamadugu swarm includes the well-known Kandlamadugu dike to the southwest of Cuddapah Basin and the ~ 400 km long, ~ 2.21–2.22 Ga old, arcuate Nelahalu dike parallel to the western margin of the CDC (French and Heaman 2010; Kumar et al. 2012b; Söderlund et al. 2019). The subparallel NW- to WNW-trending ~ 2.21 Ga old dikes of the Anantapur-Kunigal swarm including the Somala dike, and the NW- to WNW-trending ~ 2.18 Ga old dikes of the Mahabubnagar-Dandeli swarm including the Dandeli and Bandepalem dikes were emplaced within ~ 30 Myr each other, and the younger dikes probably intruded through some of the older magma pathways (French and Heaman 2010; Nagaraju et al. 2018a,b; Söderlund et al. 2019). Several NE-, NW- and N-trending ~ 2.08 Ga old dikes of the Devarabanda swarm around the Cuddapah Basin form a radial swarm with the center inside the basin (Kumar et al. 2015; Söderlund et al. 2019). Two ENE-trending ~ 1.89–1.88 Ga old dikes of the Hampi swarm to the west of the basin (Chatterjee and Bhattacharji 2001; Halls et al. 2007) are coeval with the Tadpatri sill within Cuddapah Basin (Anand et al. 2003; French et al. 2008). Some of the ~ 2.08 Ga old dikes and the ~ 1.89–1.88 Ga old dikes are parallel to the older ~ 2.37 Ga and ~ 2.21 Ga old dikes, indicating that the younger dikes were perhaps emplaced by reactivation of older magma pathways and preexisting fractures (cf. Bhattacharji 1987). In addition, several NW-trending ~ 1.84 Ga old dikes of the Dharmapuri swarm (Belica et al. 2014) and NW- to WNW-trending ~ 1.79 Ga old dikes of the Pebbair swarm (Söderlund et al. 2019) are subparallel to the ~ 2.21 Ga old dikes, and may have also intruded through preexisting magma pathways.

Several researchers have correlated the CDC/EDC dike swarms with swarms on other continents using U–Pb geochronological and paleomagnetic data, and have reconstructed the location of plume centers from which the dike swarms supposedly originated (Halls et al. 2007; French et al. 2008; French and Heaman 2010; Kumar et al. 2012a,b; Stark et al. 2019). These studies suggest that the ~ 2.37 Ga (Bangalore-Karimnagar), ~ 2.21–2.18 Ga (Anantapur-Kunigal and Mahabubnagar-Dandeli) and ~ 1.89 Ga (Hampi) old dike swarms originated from mantle plumes with centers located ~ 300 km west, ~ 1000 km NNW, and ~ 600 km east of the CDC/EDC, respectively. Halls et al. (2007) proposed the location of the ~ 2.37 Ga old plume center to the west of the Dharwar Craton by correlating the Dharwar dikes with the ~ 2.41 Ga old dikes of the Yilgarn Craton. However, Belica et al. (2014) argued against a link between the Yilgarn Craton and the Dharwar Craton at 2.41–2.37 Ga citing age disparity and a ~ 25° latitudinal separation between the two cratons. French and Heaman (2010) also proposed that the Dharwar and Yilgarn dikes are unrelated, and they were emplaced on different continental masses through discrete events. Thus, the locations of the purported plume centers are highly uncertain. Furthermore, Söderlund et al. (2019) argued against the involvement of plumes for the pre-2.08 Ga old CDC/EDC dikes by showing that the swarms were initially linear, and their fan-shaped orientations originated by later tectonic deformation. Hence, the plume center reconstructions are flawed.

Previous petrological work

The Cuddapah Basin sills and volcanics mostly comprise subalkaline tholeiitic basalts (Chatterjee and Bhattacharji 1998; Anand et al. 2003). A few of the Vempalle samples are alkalic. The major sill complex in the Tadpatri Formation consists of basaltic rocks with mafic xenoliths composed of olivine + orthopyroxene + clinopyroxene + plagioclase (Ol + Opx + Cpx + Pl) near the base, grading into leucocratic gabbro near the top. The minor sills consist of non-cumulate, differentiated basalts. In general, the Cuddapah Basin basalts can be related by fractional crystallization of olivine, clinopyroxene and plagioclase. Thermobarometry shows that the basalts crystallized at a pressure of ~ 5 kbar (~ 18 km depth) and temperatures of 1019–1154 °C (Chatterjee and Bhattacharji 1998). These P–T conditions are consistent with differentiation within a crustal magma chamber under Cuddapah Basin postulated from gravity data (NGRI 1978). The basalts have low loss-on-ignition values indicating minor hydrothermal alteration (Anand et al. 2003). They show negative Nb–Ta anomalies in their incompatible element patterns and high La/Nb (1.3–3.8, cf. primitive mantle: 0.99) and Th/Nb (0.23–1.45, cf. primitive mantle: 0.12) ratios (Anand et al. 2003; McDonough and Sun 1995). Mixing models based on La/Nb and Ce/Y ratios indicate that the non-cumulate basalts comprising most of the Cuddapah Basin lavas and sills are ~ 10–15% contaminated and some minor sills are 20–35% contaminated by the local granitic crust (Anand et al. 2003). The isotopic compositions of the basalts (εNd(t) =  −  10 to + 1, ⁸⁷Sr/⁸⁶Sr_i = 0.7056 to 0.7082) are also consistent with upper crustal contamination (Anand et al. 2003). In addition, based on Fe-Nd and REE modeling, Anand et al. (2003) suggested ~ 10–15% partial melting of spinel lherzolite for the generation of the primary magmas of the basalts. They estimated a mantle potential temperature of ~ 1500 °C, and an initially 120 km-thick lithosphere that was thinned to 70 km during magma generation. Moreover, Anand et al. (2003) concluded that the ~ 1500 °C potential temperature can be explained by secular cooling of the Earth without the involvement of a hot mantle plume.

The CDC and EDC dikes are mostly composed of basalts and basaltic andesites. A few have picritic compositions likely due to their cumulate nature. The dikes of all ages show LILE and LREE enrichment and negative Nb–Ta anomalies in normalized incompatible element plots, and high Th/Nb ratios (0.23–0.73) (Kumar et al. 2012a,b; Srivastava et al. 2014a,b, 2015; Liao et al. 2019). In a Th/Yb versus Nb/Yb diagram, the ENE- to NE-trending dikes of the ~ 2.37 Ga old Bangalore-Karimnagar swarm and the N- to NNW-trending dikes of the ~ 2.22 Ga old Kandlamadugu swarm plot above the MORB-OIB array and the data trend toward upper continental crust (Kumar et al. 2012a, b). These characteristics have been attributed to AFC-type fractionation of low-Th/Nb parental melts, which are particularly susceptible to Th enrichment and Nb depletion through crustal contamination (Pearce 2008; Kumar et al. 2012a). In addition, the εNd(t) values of the Karimnagar dikes (Bangalore-Karimnagar swarm, − 0.7 to + 0.6, Liao et al. 2019) and the Nelahalu dike (Kandlamadugu swarm, − 2.9 to − 1.7, Kumar et al. 2012b) have been attributed to crustal contamination or an isotopically heterogeneous mantle source.

The origin of the ENE- to NE-trending dikes of the ~ 2.37 Ga old Bangalore-Karimnagar swarm, N- to NNW-trending dikes of the ~ 2.22 Ga old Kandlamadugu swarm, and NW- to WNW-trending dikes of the ~ 2.21 Ga old Anantapur-Kunigal swarm were modeled by ~ 15–25% batch melting of primitive mantle sources (Srivastava et al. 2014a, b, 2015). Furthermore, Srivastava et al. (2015) used chondrite-normalized Dy_N/Yb_N ratios and available petrogenetic models to suggest that the ~ 2.37 Ga old Bangalore-Karimnagar swarm originated by melting of spinel lherzolite, whereas the ~ 2.21 Ga old Anantapur-Kunigal, ~ 2.18 Ga old Mahabubnagar-Dandeli, and ~ 1.89–1.88 Ga old Hampi swarms originated by melting of transitional spinel-garnet lherzolite.

Sample locations, age and bulk compositions

Sixteen dike samples were studied in detail for the purpose of mineral thermobarometry. Nine of these samples (D84, D86, D88, 1A, 2/14, D9A, D11, D28, and D47) are from ENE-trending dikes near the western, southwestern and southern margins of Cuddapah Basin (Fig. 1). One of the dikes (D11) has been previously dated at 2369 Ma (dike JEF-99–7, French and Heaman 2010). All of the dikes in this group are inferred to be of the same age and members of the ~ 2.37 Ga old Bangalore-Karimnagar swarm. Samples D86, D88 and D9A are tholeiitic basalts (Rao et al. 1995), and D88 has a similar bulk composition to several samples (EDC9/6,9,13) analyzed by Srivastava et al. (2014a). D84 is a basaltic andesite, D28 is a picrobasalt, and D47 is a trachyandesite (Murty et al. 1987). In addition, the tholeiitic basalt sample D8A is also from an ENE-trending dike west of the basin (Fig. 1), but it belongs to the younger ~ 1.89–1.88 Ga old Hampi swarm (1879 Ma, Chatterjee and Bhattacharji 2001). It has a bulk composition (Murty et al. 1987) similar to the Dike-25 sample studied by Halls et al. (2007).

All of the other samples are also tholeiitic basalts. One sample (D75) from an NNE-trending dike near the southwestern margin of Cuddapah Basin (Fig. 1) belongs to the ~ 2.25 Ga old Ippaguda-Dhiburahalli swarm, though it has a bulk composition (Murty et al. 1987) similar to some ~ 2.22 Ga old N- to NNW-trending dikes of the Kandlamadugu swarm (Kumar et al. 2012b). Three samples (D89, D77 and D30) are from NW-trending dikes parallel to the southwestern margin of Cuddapah Basin (Fig. 1). These dikes cross-cut the Bangalore-Karimnagar and Ippaguda-Dhiburahalli swarms, and are members of the ~ 2.21 Ga old Anantapur-Kunigal swarm (French and Heaman 2010). Sample D30 has a variable bulk composition with MgO contents between 8.3 wt% (Murty et al. 1987) and 11.2 wt% (Chatterjee and Bhattacharji 2001). In addition, two samples (6/33 and K88) are also from NW-trending dikes, but they are located near the northwestern margin of Cuddapah Basin (Fig. 1) and they belong to the younger, ~ 2.08 Ga old Devarabanda radial dike swarm (2083–2080 Ma, Kumar et al. 2015; Söderlund et al. 2019).

Analytical methods

Textural studies and mineral analyses were performed on a JEOL JXA-733 Superprobe electron probe microanalyzer (EPMA) at Massachusetts Institute of Technology, Cambridge, MA, USA operating with a 15 kV accelerating voltage, a 10 nA beam current, and 1–10 μm beam diameter. Typical counting times were 20–40 s per element that yielded accumulated counts with 1σ standard deviations of 0.3–1.0% for major elements and 1–5% for minor elements from counting statistics. The raw data were corrected for matrix effects with the CITZAF package (Armstrong 1995).

Petrography and mineral chemistry

The analyzed samples primarily consist of augite and plagioclase with ilmenite and magnetite as common accessory minerals (Fig. 2, Table 1). Minor orthopyroxene and pigeonite are present in some samples. Olivine and Mg-rich amphibole are rare. Most of the samples have equigranular, sub-ophitic and hypidiomorphic textures (Fig. 2a, d).

Fig. 2

Back-scattered electron images of basaltic dike samples (names at upper right corners) from the CDC and EDC showing a, d ophitic to sub-ophitic texture dominated by plagioclase laths and clinopyroxene crystals, b orthopyroxene between olivine and clinopyroxene, c cumulus clinopyroxene with interstitial plagioclase, e olivine overgrowth on orthopyroxene, and f plagioclase and amphibole inclusions in olivine

Table 1 Mineral assemblages in the CDC and EDC basaltic dikes

ENE- to NE-trending dikes of the ~ 2.37 Ga old Bangalore-Karimnagar swarm

These dikes are dominantly composed of normally zoned augite and plagioclase with accessory ilmenite, magnetite, titanite, apatite and pyrite. The augite cores in samples D86, D88, 1A, 2/14 and D11 have a restricted composition range (En_41-49Fs_12-16Wo_35-46), whereas the augite cores are relatively Fe-rich in D84 (En₄₆Fs₂₁Wo₃₃) and Mg-rich in D9A and D28 (En_49-50Fs_10-14Wo_37-41) (Fig. 3a, Table S1). Samples D9A and D28 also contain Mg-rich orthopyroxene (En_64-71Fs_26-31Wo_4-5) and plagioclase with cores that are more calcic (An₇₉Ab₂₁) than in the other samples (An_52-68Ab_32-49) (Fig. 3a, b), and D28 contains olivine with Fo₇₀ composition (Fig. 2a, b, Table S1). The high bulk MgO (13.8 wt%) and low Al₂O₃ and CaO contents of sample D9A (Murty et al. 1987) are related to excess accumulation of Mg-rich orthopyroxene. Orthopyroxene (Fe-rich) also occurs in D86 (En₂₈Fs₆₇Wo₄), and pigeonite (core: En_60-64Fs_26-39Wo_7-10) occurs in D84, D86, D88 and D11 (Fig. 3a). The jadeite and aegirine contents of the pyroxenes are < 2.5%. Minor quartz is present in samples D84 and D86. Clinopyroxene is commonly rimmed by actinolite and ferrohornblende. Secondary chlorite, epidote and albite are also present. Clinopyroxene is absent in sample D47 that contains ferroedenite, Fe-rich orthoamphibole and Ca-poor plagioclase (An₃₀Ab₇₀).

Fig. 3

Composition of minerals in the CDC and EDC basaltic dikes: a, c pyroxenes, and b, d plagioclase. c, d Compositional variation from core to rim of orthopyroxene and plagioclase in sample D30

N- to NNE-trending dikes of the ~ 2.25 Ga old Ippaguda-Dhiburahalli swarm

The NNE-trending dike D75 consists of augite with thick rims of actinolite and ferrohornblende, plagioclase, and accessory epidote, ilmenite and pyrite. The augite and plagioclase cores have compositions of En₄₂Fs₂₃Wo₃₄ and An₇₂Ab₂₈, respectively (Fig. 3a, b, Table S1). The sample also contains pigeonite (En₆₁Fs₂₈Wo₁₁) and minor quartz.

NW- to WNW-trending dikes of the ~ 2.21 Ga old Anantapur-Kunigal swarm

Dikes D89 and D77 locally cross-cut the ~ 2.25 Ga old dike D75. They are mineralogically similar to D75, and are also composed of augite with thick rims of actinolite and ferrohornblende, plagioclase, and accessory epidote, ilmenite and pyrite. However, compared to D75, the augite (En_46-47Fs₁₆Wo_37-38) and plagioclase (An_60-64Ab_36-40) cores in D89 and D77 are Mg-rich and Ca-poor, respectively (Fig. 3a, b, Table S1). Sample D30 has an equigranular sub-ophitic texture, and it consists of reverse-zoned augite (En_52-55Fs_9-15Wo_34-36) and pigeonite (En_59-67Fs_22-31Wo_10-11), oscillatory zoned orthopyroxene (core: En_73-80Fs_15-22Wo₅, rim: En₆₀Fs₃₆Wo₄), and normally zoned plagioclase (core: An₆₈Ab₃₂, rim: An₂₅Ab₇₅) (Figs. 2d, 3a-d). It also contains secondary orthoamphibole, chlorite and mica, and accessory ilmenite and magnetite.

Radial dikes of the ~ 2.08 Ga old Devarabanda swarm

The NW-trending dike K88 is medium-grained and has an ophitic texture. It dominantly consists of augite (En₄₇Fs₁₅Wo₃₉) and plagioclase (core: An₇₁Ab₂₉) with minor olivine (Fo₆₇), orthopyroxene (En₆₈Fs₂₉Wo₄) and amphibole (magnesiohastingsite, Mg/(Mg + Fe) = 0.65), and accessory chlorite, ilmenite and magnetite (Fig. 3a, b, Table S1). Orthopyroxene occurs as discrete crystals, exsolution lamellae in augite, and at the rims of olivine and clinopyroxene. Olivine-hosted inclusions of plagioclase, pyroxene and amphibole (Fig. 2e, f) belong to an older generation of basalt, and the olivine overgrowth probably formed through influx of a younger batch of primitive magma. This suggests that dike K88 was probably emplaced by reactivation of an old magma pathway. Another NW-trending dike 6/33 contains augite with lower Mg (En₃₈Fs₂₅Wo₃₈) and less calcic plagioclase (core An₆₃Ab₃₇) compared to K88 (Fig. 3a, b). It also contains Mg-rich pigeonite (En₇₁Fs₁₉Wo₁₀), but it lacks olivine, orthopyroxene and hornblende.

ENE- to NE-trending dikes of the ~ 1.89–1.88 Ga old Hampi swarm

The ENE-trending dike D8A contains abundant cumulus augite (core: En₅₀Fs₁₄Wo₃₆) and compositionally zoned plagioclase (core: An₄₉Ab₅₁). Augite accounts for > 50% of the rock volume (Fig. 2c). The sample also contains minor pigeonite (En₅₁Fs₄₁Wo₈) and hornblende (Mg/(Mg + Fe) = 0.64), secondary albite and mica, and accessory ilmenite and magnetite.

Thermobarometry

Methods

The P–T conditions of crystallization were determined with the clinopyroxene-anhydrous liquid thermobarometer of Putirka et al. (1996) using mineral compositions in Table S1 and bulk compositions in Murty et al. (1987) and Rao et al. (1995). The quoted uncertainties in the P–T calculated with this thermobarometer are ± 1.4 kbar and ± 27 °C. Application of this thermobarometer requires that the Cpx is in equilibrium with the bulk (liquid). Equilibrium is assessed from the Cpx-bulk Fe²⁺-Mg distribution coefficient, K_D(Fe²⁺-Mg), the equilibrium value of which is 0.28 ± 0.08 (Putirka 2008). So, a knowledge of the bulk Fe²⁺ content (or the bulk Fe^3+/Fe ratio) is necessary. For samples that contain equilibrium olivine and the equilibrium temperature is independently known, the bulk Fe^3+/Fe ratio can be calculated with Eq. 8 of Blundy et al. (2020). Olivine-melt equilibrium is assessed by comparing the observed value of the olivine-bulk Mn-Mg distribution coefficient, K_D(Mn-Mg), which is relatively constant over a wide range of P–T-fO₂ conditions, with the equilibrium value predicted by the lattice strain model (Blundy et al. 2020). If the observed K_D(Mn-Mg) shows disequilibrium, the bulk composition is adjusted by adding or subtracting olivine until olivine is in equilibrium with the bulk (Blundy et al. 2020). This method was attempted on the olivine-bearing samples K88 and D28, the latter with a temperature of 1172 °C determined from Cpx composition only (see below). However, the olivines (Fo_67-70) in these samples are not in equilibrium with the bulk, as indicated by the higher observed K_D(Mn-Mg) values (0.8–1.0) than predicted (0.26–0.27) at 1172 °C, and Eq. 8 of Blundy et al. (2020) yields negative Fe³⁺/∑Fe ratios (− 1.35 and − 1.65). Very large corrections (40–45% olivine subtraction) are required to the bulk to bring it in Mn-Mg equilibrium with olivine, and the Fe³⁺/∑Fe ratios after correction are unreasonably high (0.36–0.58). So, Blundy et al.’s method was not applied to determine bulk Fe³⁺/∑Fe ratios. Instead, the bulk Fe³⁺/∑Fe ratios were determined using the spinel-ilmenite equilibrium. First, oxygen fugacity values were calculated from coexisting spinel and ilmenite compositions (program QUILF4, Andersen and Lindsley 1988). Then, Eq. 6b of Putirka (2016a) was used to calculate ln(X_Fe2O3/X_FeO) of the melt. Fe³⁺/Fe²⁺ equals 2*X_Fe2O3/X_FeO, and Fe³⁺/∑Fe is 1/(1 + 1/(Fe³⁺/Fe²⁺)). For samples in which coexisting spinel and ilmenite are absent, the average bulk Fe³⁺/∑Fe ratio of the other samples was used in the calculations.

The P–T were also calculated with the Cpx-composition thermobarometer of Putirka (2008). The T-dependent barometric expression (Eq. 32a) and P-dependent thermometric expression (Eq. 32d) of Putirka (2008) were solved simultaneously to obtain P–T. These equations are based on multiple regression of clinopyroxene compositions obtained from partial melting experiments on basalts in the P–T range of 1 bar-75 kbar/800–2200 °C. The equations use the enstatite-ferrosilite and diopside-hedenbergite components and cation proportions of clinopyroxene calculated on the basis of 6 oxygen atoms. The quoted uncertainties are ± 3.1 kbar and ± 58 °C for clinopyroxene crystallizing from anhydrous melts. In addition, the P–T were also calculated with the random forest machine learning-based algorithms of Higgins et al. (2022) and Jorgenson et al. (2022) that provided independent estimates of the uncertainties. Jorgensen et al. (2022) use the same methodology as Higgins et al. (2022), but they use an expanded dataset that includes Cpx in equilibrium with alkalic liquids. These thermobarometers are also based on clinopyroxene compositions obtained from partial melting experiments on basalts that cover a P–T range of 0.002–30 kbar/750–1250 °C.

The two-pyroxene thermobarometer of Putirka (2008) (uncertainties: ± 3.7 kbar and ± 60 °C) was applied to calculate P–T in sample D30 that contains Cpx and Opx showing equilibrium Fe–Mg distribution (K_D(Fe²⁺-Mg) = 1.09 ± 0.14). In addition, samples K88 and D8A contain Mg-rich amphibole, whose compositions were used to calculate temperatures with the thermometer of Putirka (2016b) (uncertainty: ± 30 °C).

Results

In the ENE-trending ~ 2.37 Ga old Bangalore-Karimnagar dikes, clinopyroxene crystallized at P–T conditions of 0.8–1.5 kbar and 1119–1158 °C, as estimated with the Cpx-anhydrous liquid formulations of Putirka et al. (1996) (Table 2). Using only Cpx compositions, Putirka’s (2008) formulations yielded P–T of 1.0–4.9 kbar (± 3.1 kbar) and 1166–1192 °C (± 58 °C) (Table 3). The Higgins et al. (2022) and Jorgenson et al. (2022) methods also yielded low pressures (1 bar-2 kbar), but with lower uncertainties (± 1.1–1.5 kbar, and ± 0.3–2.0 kbar). The temperature estimates are lower with both the methods of Higgins et al. (2022) (1016–1150 °C, ± 20–89 °C) and Jorgenson et al. (2022) (1115–1160 °C, ± 27–88 °C) compared to Putirka (2008), but all three methods have overlapping uncertainties (Table 3). Spinel and ilmenite equilibrated at subsolidus temperatures (457–661 °C) and oxygen fugacity values below the fayalite-magnetite-quartz buffer (∆FMQ between -0.4 and -3.2), and the calculated range of Fe³⁺/∑Fe ratios is 0.06–0.12.

Table 2 Mineral-liquid and mineral thermobarometry of the CDC and EDC basalts

Table 3 Cpx thermobarometry of the CDC and EDC basalts

In the NNE-trending dike D75 from the ~ 2.25 Ga old Ippaguda-Dhiburahalli swarm, the estimated P–T of clinopyroxene crystallization are 0.9 kbar and 1119 °C with the Cpx-only formulations of Putirka (2008). The methods of Higgins et al. (2022) (2.0 ± 1.1 kbar, 1032 ± 82 °C) and Jorgenson et al. (2022) (1 bar, 1112 ± 21 °C) yielded similar results.

In the NW-trending ~ 2.21 Ga old Anantapur-Kunigal dikes, clinopyroxene crystallized at P–T conditions of 4.0–5.7 kbar and 1169–1204 °C, as estimated with the Cpx-anhydrous liquid formulations of Putirka et al. (1996) (Table 2). The Cpx-only formulations of Putirka (2008) yielded P–T of 2.6–5.4 kbar (± 3.1 kbar) and 1169–1211 °C (± 58 °C) (Table 3). The Higgins et al. (2022) method yielded similar results (2–7 kbar, ± 0.3–4.0 kbar, 1016–1200 °C, ± 29–76 °C). The Jorgenson et al. (2022) method also yielded similar results (1 bar-2 kbar, ± 0.5–10.0 kbar, 1135–1210 °C, ± 29–79 °C), but very high uncertainties in pressure for sample D30 (Table 3) that may be related to the high Cr₂O₃ (> 1 wt%) content of the clinopyroxene (Table S1). In sample D30, the Cpx-Opx thermobarometer of Putirka (2008) yielded P–T of 5.5–5.7 kbar (± 3.7 kbar) and 1150–1178 °C (± 60 °C) (Table 2). Thus, the P–T results for sample D30 with the Cpx-anhydrous liquid, Cpx-only and Cpx-Opx formulations are consistent with each other. Sample D30 also registered a spinel-ilmenite equilibration temperature of 638 °C and ∆FMQ of -1.06 with a corresponding Fe³⁺/∑Fe ratio of 0.11.

In the NW-trending dike K88 from the ~ 2.08 Ga old Devarabanda swarm, amphibole crystallized at a temperature of 986 °C, and in dike 6/33 from the same swarm, spinel and ilmenite equilibrated at a temperature of 791 °C and a ∆FMQ value of − 0.49 (Table 2).

In the ENE-trending dike D8A from the ~ 1.89–1.88 Ga old Hampi swarm, clinopyroxene crystallized at P–T conditions of 5.1 kbar and 1202 °C, as estimated with the Cpx-anhydrous liquid formulations of Putirka et al. (1996) (Table 2). With the Cpx-only formulations of Putirka (2008), the P–T are 3.8 kbar (± 3.1 kbar) and 1192 °C (± 58 °C) (Table 3). The Higgins et al. (2022) (5.0 ± 2.4 kbar, 1150 ± 48 °C) and Jorgenson et al. (2022) methods (0.3 ± 2.7 kbar, 1148 ± 37 °C) yielded similar results with lower uncertainties. Amphibole crystallized at a temperature of 785 °C, and spinel-ilmenite equilibrated at a temperature of 665 °C and ∆FMQ of -0.54 with a corresponding Fe³⁺/∑Fe ratio of 0.16.

In summary, clinopyroxene crystallized at P–T conditions of 0.8–5.7 kbar and 1119–1211 °C in all samples, estimated using the formulations of Putirka et al. (1996) and Putirka (2008). The machine learning-based algorithms of Higgins et al. (2022) and Jorgenson et al. (2022) yielded similar pressures (1 bar-5 kbar) with lower uncertainties (except for sample D30), but a larger range of temperatures (1016–1210 °C). Spinel and ilmenite equilibrated at subsolidus temperatures (~ 460–665 °C, excluding 6/33) and ∆FMQ values between -0.5 and -3. The average Fe³⁺/∑Fe ratio of the samples is 0.1. These results clearly indicate that the dikes crystallized within the upper crust.

Primary magma and crustal contamination modeling

Sample selection

At low pressures such as the ≤ 6 kbar pressures of crystallization calculated above, primitive basalts evolve by crystallizing olivine, followed by Ol + Pl, and then by Ol + Pl + Cpx with decreasing temperature (Kinzler and Grove 1992a). The olivine control line, the Ol-Pl cotectic, and the Ol-Pl-Cpx cotectic together define the fractionation path of the basalt. Any basalt whose composition has been modified by excess crystal accumulation (i.e., a “cumulate” sample) would show displacement from its fractionation path when plotted in the Ol-Pl-Cpx (from Qz) and Ol-Cpx-Qz (from Pl) pseudoternary projections of the basalt tetrahedron according to the methods of Tormey et al. (1987) and Grove (1993). For the purpose of modeling, a total of 88 basalt samples from the literature were considered (Murty et al. 1987; Rao et al. 1995; Chatterjee and Bhattacharji 1998; Anand et al. 2003; Halls et al. 2007; French and Heaman 2010; Kumar et al. 2012a,b; Srivastava et al. 2014a,b, 2015). The bulk composition of each sample and its plagioclase lherzolite multiple saturation point (PL-MSP) at 1 bar-10 kbar pressures predicted by the parameterized expressions of Kinzler and Grove (1992a) were plotted in the pseudoternary projections. The position of the PL-MSP and the constraints of Yang et al. (1996) define the position of the Ol-Pl-Cpx cotectic at different pressures in the projections. Six of the samples plot on their corresponding Ol-Pl-Cpx cotectics, indicating that they represent unmodified basaltic liquids. The other samples are variably displaced from their cotectics. The bulk compositions of these samples were adjusted by subtracting normative olivine so that they plotted on their respective Ol-Pl-Cpx cotectics. Thirty (including the six that plot on their cotectics) of the 88 samples required ≤ 4% normative olivine subtraction, and these were selected for primary magma modeling (Table S2, locations in Fig. 1). The other samples were not selected as their excess crystal contents were deemed too high for meaningful adjustment to the bulk composition.

Six of the selected samples are from inside the Cuddapah Basin: three from the Tadpatri sills (minor sills near Krishnagiri and Yeraguntla-Vempalle, and top part of the major sill near Pulivendla), and three from the Vempalle volcanics near Gattimanikonda (Chatterjee and Bhattacharji 1998; Anand et al. 2003). They plot on their respective Ol-Pl-Cpx cotectics at 1 bar-6.5 kbar pressures (Table 4, S3), consistent with the ~ 5 kbar pressure of crystallization estimated by Chatterjee and Bhattacharji (1998).

Table 4 P–T of crystallization and primary magma equilibration with lherzolite for the CDC, EDC and Cuddapah Basin basalts

From the remainder of the selected samples, 14 are from the ENE- to NE-trending dikes of the ~ 2.37 Ga old Bangalore-Karimnagar swarm (four from CDC, 10 from EDC, Rao et al. 1995; Halls et al. 2007; Srivastava et al. 2014a), five are from the N- to NNW-trending dikes of the ~ 2.22 Ga old Kandlamadugu swarm (three from CDC, two from EDC, French and Heaman 2010; Kumar et al. 2012b; Srivastava et al. 2014b), and five are from the NW- to WNW-trending dikes of the ~ 2.21 Ga old Anantapur-Kunigal swarm (three from CDC, two from EDC, Srivastava et al. 2015). Srivastava et al. (2014a) identified two groups of basalts among the Bangalore-Karimnagar dikes based on chondrite-normalized La_N/Lu_N ratios (~ 2 and > 2). The 14 selected samples of Bangalore-Karimnagar dikes include 11 from the low La_N/Lu_N group and 3 from the high La_N/Lu_N group. The selected dike samples from all swarms plot on their 1 bar-3 kbar Ol-Pl-Cpx cotectics (Table 4, S3), indicating crystallization pressures similar to those obtained from thermobarometry.

The selected samples are all subalkaline tholeiitic basalts (Mg# 37–55) according to the total alkali versus silica classification (Fig. 4a). There is a broad negative correlation between MgO and FeO^T (Fig. 4b). The samples show chemical index of alteration (CIA) values of 37–41, and they are tightly clustered around average unaltered basalt and gabbro in an A-CN-K plot (Nesbitt and Young 1982; Babechuk et al. 2014) (Fig. 4c). Hence, these samples have not been altered by kaolinitization of feldspar, and their bulk K contents have largely remained unchanged since formation. In a Th/Yb versus Nb/Yb diagram (Fig. 4d), the samples plot above the MORB-OIB array and toward upper continental crust, and the trend of the data coincides with a model AFC trend of Pearce (2008). A similar (but less obvious) trend toward upper continental crust is also observed in a Zr/Y versus Nb/Y diagram (Fig. 4e). In a Ce/Y versus La/Nb diagram (Fig. 4f), the data plot along a mixing line between melts of primitive mantle and an average CDC/EDC anatectic granite, as shown previously for the Cuddapah Basin basalts by Anand et al. (2003). The CDC/EDC samples show similar Th/Nb (0.20–0.60) and La/Nb (1.4–3.7) ratios to the Cuddapah Basin samples (0.23–0.69 and 1.9–2.6) indicating that upper crustal contamination may be similar in the two groups (Fig. 4d, f, Table S2).

Fig. 4

Bulk composition of the selected CDC/EDC and Cuddapah Basin basalts (data from Rao et al. 1995, Halls et al. 2007, and Srivastava et al. 2014a for ENE- to NE-trending dikes, French and Heaman 2010, Kumar et al. 2012b, and Srivastava et al. 2014b for N- to NNW-trending dikes, Srivastava et al. 2015 for NW- to WNW-trending dikes, and Chatterjee and Bhattacharji 1998, and Anand et al. 2003 for Vempalle volcanics and Tadpatri sills within Cuddapah Basin), a Total alkali versus silica after Le Bas et al. (1986), subalkalic-alkalic boundary from Macdonald and Katsura (1964), b MgO versus total FeO, c A-CN-K plot, average gabbro and basalt (red dots) from Nesbitt and Young (1982) and Babechuk et al. (2014), d MORB-OIB array and dashed line with arrow representing AFC trend from Pearce (2008), e Plume array after Condie (2005), f Average CDC and EDC anatectic granites (green dots) from Jayananda et al. (2018), dashed line with arrow shows mixing trend of primitive melts with average CDC/EDC granite. d-f PM – primitive mantle (McDonough and Sun 1995), and UCC and LCC – upper and lower continental crust (Taylor and McLennan 1995)

In the Th/Yb versus Nb/Yb and Ce/Y versus La/Nb diagrams (Fig. 4d, f), the data plot beyond the range of mixing between primitive melts and lower continental crust, indicating that the lower crust was not the contaminant. A good match with the model AFC trend of Pearce (2008) (Fig. 4d) shows that the primitive melts were probably contaminated with the upper crust through an AFC-type process. However, contamination of the primitive melts during passage through lithosphere containing fusible crustal material acquired by previous subduction cannot be ruled out. In this scenario, the contamination occurred before the primary magma crystallized.

Calculation method: reverse FC model

Assuming that the CDC/EDC and Cuddapah Basin primary magmas were already contaminated before crystallization during passage through lithosphere, their composition and P–T of equilibration with the metasomatized mantle were determined through simple low-pressure reverse fractional crystallization (Reverse FC) calculations (Till et al. 2012; Chatterjee and Sheth 2015; Till 2017; Krein et al. 2021) (Fig. S1). Starting from the composition of the sample, its low-pressure fractionation path was modeled backward by adding Ol + Pl + Cpx (stage 1), Ol + Pl (stage 2) and Ol-only (stage 3) in small steps (step size < 0.5%) until the melt reached its lherzolite MSP at a high pressure. Equilibrium Fe²⁺-Mg distribution between olivine-liquid (K_D(Fe²⁺-Mg) = 0.3, Roeder and Emslie 1970) and Cpx-liquid (K_D(Fe²⁺-Mg) = 0.25), and equilibrium Ca-Na distribution between plagioclase-liquid (Grove et al. 1992) were maintained at each step of the calculation. In stage 1, the melt moved toward the Ol-Cpx sidebar in the Ol-Cpx-Qz projection following the constraints of Yang et al. (1996) that define the intersection of the Ol-Pl-Cpx cotectic with the Ol-Pl cotectic. In stage 2, the melt moved toward the olivine apex in the Ol-Cpx-Qz projection and toward the Ol-Pl sidebar in the Ol-Pl-Cpx projection. In stage 3, the melt moved toward the olivine apex in both projections. The phase proportions and the switching points between stages 1–2 and 2–3 were adjusted so that the melt moved toward its spinel lherzolite MSP (SL-MSP) at high pressures predicted by Till et al.’s (2012) parameterized expressions of experimental data. At the end of the calculation, the melt was in equilibrium with Fo₉₀ olivine, and it plotted exactly on its SL-MSP at a specific high pressure. The result was unique, as any deviation from the phase proportions and switching points between stages would result in a melt not on its lherzolite MSP at any pressure, though it may show equilibrium with Fo₉₀ olivine (Fig. S1). The trace elements were modeled using the mineral-melt partition coefficients listed in Table S3.

Calculation method: Reverse EC-AFC model

In AFC, the magma mass changes through mass gained by assimilation of partial melt of wall rock and mass lost by crystallization of the magma. The masses of crystallized and assimilated material depend on the heat budget, which is balanced by heat lost by the magma through cooling and crystallization when the temperature drops below the liquidus, and heat gained by the wall rock through heating and partial melting when the temperature exceeds the solidus. The mass and heat balance constraints are combined together in the AFC formulation (e.g. Thompson et al. 2002). The efficiency of assimilation critically depends on the initial temperature of the crust. Thompson et al. (2002) showed that a picritic magma assimilates tonalitic crust much more efficiently if the crust is initially at 800 °C than at 200 °C. The AFC process is complicated by localized convective heat transfer and dynamic processes including melt segregation by compaction, deformation, and buoyancy instabilities, but these local processes do not have large effects on the overall heat sharing and melting relationships (Annen and Sparks 2002).

Spera and Bohrson (2001) considered energy-constrained AFC (EC-AFC) within an adiabatically sealed system that can be described in terms of a set of coupled ordinary differential equations expressing conservation of energy (enthalpy), total mass, and trace element abundances and isotopic ratios. In this study, the primary magmas were modeled by incorporating the EC-AFC formulation of Spera and Bohrson (2001), and Bohrson and Spera (2001) in the reverse fractionation calculations (Reverse EC-AFC). As in the Reverse FC model, the low-pressure fractionation path was modeled backward by adding Ol + Pl + Cpx (stage 1), Ol + Pl (stage 2) and Ol-only (stage 3) in small steps while maintaining equilibrium Fe²⁺-Mg distribution between olivine-liquid and Cpx-liquid, and equilibrium Ca-Na distribution between plagioclase-liquid. In addition, a partial melt of the upper crustal assimilant was subtracted in each reverse fractionation step as described below.

The EC-AFC calculations were carried out using the updated RK07A_2011_1.xlsm spreadsheet (Spera and Bohrson 2001; Bohrson and Spera 2001). This is a forward modeling approach that requires knowledge of the initial composition and temperature of the magma (i.e., temperature of the primary magma after rising to the upper crust, T_m⁰), and the initial temperature of the upper crustal assimilant (T_a⁰). These parameters for each sample were determined through an iterative approach using the K content of the magma. In the first iteration, the input K content of the primary magma was calculated through the Reverse FC method. The bulk compositions of this primary magma and the sample were used to determine T_m⁰ of the primary magma and T_m of the sample with Eq. 16 of Putirka (2008). The starting value of T_a⁰ was 300 °C. The M_a*/M_c values (ratio of mass of assimilated crustal melt to mass crystallized) as a function of temperature obtained from the EC-AFC forward model were used to subtract the appropriate amounts of partial melt of crustal assimilant in each step of the reverse fractionation. This yielded a tentative estimate of the primary magma considering EC-AFC, whose K content, a recalculated T_m⁰ (which changed slightly), and an adjusted T_a⁰ were used as input in the second iteration. The iterations were continued until the K contents of the sample (at T_m of the sample) and the melt in each step of the reverse fractionation matched the K contents predicted by the EC-AFC forward model (Fig. S2). The constraints of the reverse fractionation method ensured that the primary magma plotted on its SL-MSP and the evolving magma remained on its fractionation path. Although the K content was used in the modeling, the Rb contents (where available for the sample) also showed a good match with the Rb contents predicted by the EC-AFC forward model (Fig. S2). The calculations were carried out with an average CDC or EDC TTG gneiss (Jayananda et al. 2018) (Table S2), depending on sample location, as the upper crustal assimilant (Table S2). Partial melts of the assimilant were subtracted in each step using the M_a*/M_c value that varies according to energy constraints (in contrast to a constant “r” ratio, DePaolo 1981) during fractionation. In the calculations, the solidus and liquidus temperatures of the upper crustal assimilant were 900 °C and 1000 °C, and the energy parameters were: crystallization enthalpy, ∆h_cry = 396 kJ/kg, isobaric specific heat of magma, C_p,m = 1.484 kJ/kg per K, fusion enthalpy, ∆h_fus = 270 kJ/kg, and isobaric specific heat of assimilant, C_p,a = 1.37 kJ/kg per K (Table 1 of Bohrson and Spera 2001). The equilibration temperature (T_eq) was assumed as 1000 °C. The gneiss-melt partition coefficients of the trace elements were estimated from their contents in average TTG gneiss and anatectic melt in Jayananda et al. (2018) (Table S2, S3). The basalt-melt elemental partition coefficients were calculated from the estimated proportions of fractionating phases and mineral-melt elemental partition coefficients (Table S2, S3).

Reverse FC modeling results

The estimated compositions and P–T conditions of equilibrium of the primary magmas with lherzolite for the Reverse FC models are provided in Tables 4 and S2. The primary magmas are high-Mg basalts with MgO contents ranging 10–12 wt%. The fractionation trends for all samples are similar. Stage 1, i.e., the last stage of fractionation, was the longest during which the melts fractionated by an average of 39.9% (from 23.5 to 63.4%, Table S2). Phases were added in average proportions of Ol:Pl:Cpx = 12:49:39 that resulted in an increase in the average Mg# from 44 to 66 (Table S2). During Stage 2 (intermediate stage), the melts fractionated by an average of 21.2% (from 2.3 to 23.5%). Phases were added in average proportions of Ol:Pl = 29:71, resulting in an increase in the average Mg# from 66 to 71. During Stage 3, i.e., the earliest stage of fractionation, the primary magmas fractionated by an average of 2.3%. Only olivine was added, and the average Mg# increased from 70 to 73.

The calculations show that the ~ 1.89 Ga old samples from inside the Cuddapah Basin are 57–72% fractionated, and they were last equilibrated with spinel lherzolite at P–T of 8–12 kbar and 1266–1315 °C (Table 4, S2). The ENE- to NE-trending dike samples from the ~ 2.37 Ga old Bangalore-Karimnagar swarm are 58–73% fractionated, and they were last equilibrated with spinel lherzolite at P–T of 8–12.5 kbar and 1268–1320 °C. The N- to NNW-trending dike samples from the ~ 2.22 Ga old Kandlamadugu swarm are 53–72% fractionated, and they were last equilibrated with spinel lherzolite at P–T of 10–12.5 kbar and 1290–1314 °C. The NW- to WNW-trending dike samples from the ~ 2.21 Ga old Anantapur-Kunigal swarm are 50–74% fractionated, and they were last equilibrated with spinel lherzolite at P–T of 10.5–13 kbar and 1297–1323 °C. All of these results significantly overlap considering their uncertainties (see below), and there are no systematic differences among the different swarms. It is concluded that the samples were last equilibrated with metasomatized spinel lherzolite in the P–T range of 8–13 kbar/1266–1323 °C.

Reverse EC-AFC modeling results

The estimated compositions and P–T conditions of equilibrium of the primary magmas with lherzolite and the amounts of crustal contamination during fractionation for the Reverse EC-AFC models are provided in Tables 4 and S2, and shown in Figs. 5, 6, and 7. The primary magmas are high-Mg basalts and picrites with MgO contents ranging 10.7–13.2 wt%. The fractionation trends for all samples are similar. The basalts evolved by fractionating and assimilating crustal melt with maximum M_a*/M_c ratios between 0.16 and 0.64. Stage 1, i.e., the last stage of fractionation, was the longest during which the melts fractionated by an average of 38.1% (from 24.1 to 62.2%, Table S2). In this stage, phases were added in average proportions of Ol:Pl:Cpx = 13:49:38 that resulted in an increase in the average Mg# from 44 to 65 with concomitant decrease in SiO₂ and Na₂O, and increase in CaO and Al₂O₃ (Fig. 6, Table S2). During Stage 2 (intermediate stage), the melts fractionated by an average of 19.2% (from 4.9% to 24.1%). Phases were added in average proportions of Ol:Pl = 31:69, resulting in an increase in the average Mg# from 65 to 70, decrease in SiO₂, CaO and Na₂O, and increase in Al₂O₃ (Fig. 6). During Stage 3, i.e., the earliest stage of fractionation, the primary magmas fractionated by an average of 4.9%. Only olivine was added, and the average Mg# increased from 70 to 73 as the SiO₂, Al₂O₃, CaO and Na₂O contents decreased (Fig. 6). The trace element variations of the melts are shown in Fig. 7. The compatible trace element Ni increased with increasing MgO contents (Fig. 7a). The incompatible trace elements Rb and the REE decreased with increasing MgO as also observed for TiO₂ and K₂O (Figs. 6b,h, 7b-d).

Fig. 5

Ol-Pl-Cpx and Ol-Cpx-Qz pseudoternary projections from Qz and Pl showing compositions of the selected CDC/EDC and Cuddapah Basin basalts and their primary magmas estimated through Reverse EC-AFC calculations. Also shown are the average fractionation path (black lines, only the average is shown for clarity) and plagioclase lherzolite (PL) MSPs at 1 bar-10 kbar in 2 kbar intervals for the average basalt, and the spinel lherzolite (SL) MSPs at 9–21 kbar in 3 kbar intervals for its primary magma

Fig. 6

Bivariate plots showing the variation of major oxides and CaO/Al₂O₃ with MgO for the selected CDC/EDC and Cuddapah Basin basalts and their parental and primary magmas estimated through Reverse EC-AFC calculations. The fractionation path of each basalt is distinct but similar. For clarity, only an average fractionation path (black lines) is shown. Melts labeled “First Pl” and “First Cpx” correspond to initiation of plagioclase and Cpx crystallization

Fig. 7

Bivariate plots showing the variation of trace elements with MgO for the selected CDC/EDC and Cuddapah Basin basalts and their parental and primary magmas estimated through Reverse EC-AFC calculations. The fractionation path of each basalt is distinct but similar. For clarity, only an average fractionation path (black lines) is shown. Melts labeled “First Pl” and “First Cpx” correspond to initiation of plagioclase and Cpx crystallization

The calculations show that the ~ 1.89 Ga old samples from inside the Cuddapah Basin were last equilibrated with spinel lherzolite at P–T of 10–15.5 kbar and 1291–1359 °C (Table 4, S2). They are 56–72% fractionated and 6–23% contaminated with crust that was initially at a temperature of 757–801 °C, and assimilation started when the magma was at 1159–1224 °C. The ENE- to NE-trending dike samples from the ~ 2.37 Ga old Bangalore-Karimnagar swarm were last equilibrated with spinel lherzolite at P–T of 10.5–16 kbar and 1297–1363 °C. They are 58–73% fractionated and 7–30% contaminated with crust that was initially at a temperature of 665–808 °C, and assimilation started when the magma was at 1138–1194 °C. The N- to NNW-trending dike samples from the ~ 2.22 Ga old Kandlamadugu swarm were last equilibrated with spinel lherzolite at P–T of 13.5–15.5 kbar and 1333–1352 °C. They are 51–71% fractionated and 13–20% contaminated with crust that was initially at a temperature of 710–760 °C, and assimilation started when the magma was at 1147–1177 °C. The NW- to WNW-trending dike samples from the ~ 2.21 Ga old Anantapur-Kunigal swarm were last equilibrated with spinel lherzolite at P–T of 12–16.5 kbar and 1315–1366 °C. They are 48–74% fractionated and 5–18% contaminated with crust that was initially at a temperature of 680–778 °C, and assimilation started when the magma was at 1153–1165 °C.

All of these results significantly overlap considering their uncertainties (see below), and there are no systematic differences among the different swarms. Considering all samples, the P–T range for last equilibration with spinel lherzolite was 10–16.5 kbar/1291–1366 °C. These P–T conditions are higher than the P–T estimated through the Reverse FC models. The samples are mostly ≤ 20% contaminated except for three samples, one from the Vempalle volcanics of Cuddapah Basin and two from the ENE- to NE-trending ~ 2.37 Ga old Bangalore-Karimnagar dikes, that are 23–30% contaminated. The upper crust was already at high temperatures (665–808 °C) at the time the dikes, sills and volcanics were emplaced.

Uncertainties

The uncertainties in the estimated compositions and P–T of primary magmas arise from the uncertainties in the equilibrium mineral-melt K_D(Fe²⁺-Mg) coefficients, and the mantle olivine composition that may vary between Fo₈₈ and Fo₉₂. If the primary magma is allowed to equilibrate with olivine compositions between Fo₈₈ and Fo₉₂ using fixed K_D(Fe²⁺-Mg) values, the uncertainty in the Mg# of the primary magma is ± 6% (Mg# range: 68.5–77.5), and the uncertainties in the MgO and FeO contents of the primary magma are ± 12% and ± 10% (represented by error bars in Figs. 5–7). These uncertainties are larger than the uncertainties calculated by changing the K_D(Fe²⁺-Mg) values. For example, a ± 10% change in the Ol-melt and Cpx-melt K_D(Fe²⁺-Mg) values results in a ± 3% change in MgO and ± 6% change in FeO of the primary magma for the most fractionated (73%) sample EDD09/14. For less fractionated samples such as GD24 (51% fractionated), a ± 10% change in the Ol-melt and Cpx-melt K_D(Fe²⁺-Mg) values results in a ± 2% change in MgO and ± 3% change in FeO of the primary magma. Using higher values of K_D(Fe²⁺-Mg) results in an increase in FeO and decrease in MgO, and the primary magma (Mg# 71.2–71.9) equilibrates with Fo_88.2–88.6 olivine. Using lower values of K_D(Fe²⁺-Mg) results in a decrease in FeO and increase in MgO, and the primary magma (Mg# 74.7–74.0) equilibrates with Fo_91.6–91.3 olivine.

Multiple regression of lherzolite saturated melt compositions in experiments indicate that the P–T values at SL-MSP are accurate within ± 1.5 kbar and ± 11 °C (Till et al. 2012; Krein et al. 2021). However, the uncertainties in the P–T of multiple saturation are larger if the primary magma is allowed to equilibrate with olivine compositions between Fo₈₈ and Fo₉₂. In this case, the average uncertainties in pressure and temperature are ± 15% and ± 3%, respectively (e.g., 15 ± 2.3 kbar, 1350 ± 40 °C), and the average uncertainty in the estimated crustal assimilation is ± 20% (e.g., 10 ± 2%).

Discussion

Crustal contamination

Application of EC-AFC in the reverse fractionation modeling shows that the amount of upper crustal contamination in the Cuddapah Basin basalts is 6–23% (Table 4). This is in good agreement with the model mixing of 10–35% granitic crust with the Cuddapah Basin primary melts based on La/Nb and Ce/Y ratios by Anand et al. (2003). The CDC/EDC dikes also show 5–30% upper crustal contamination. An important result of the Reverse EC-AFC modeling is that the upper crust was already heated to high temperatures (665–808 °C) during emplacement of the various dike swarms. Such thermal priming of the crust has been attributed to persistent magmatism for the Steens basalts in the Columbia River flood basalt province (Moore et al. 2018, 2020). According to field observations, the ~ 2.37 Ga old Bangalore-Karimnagar dikes cross-cut an older set of mafic dikes (Padmakumari and Dayal 1987; Kumar and Bhalla 1983). The CDC crust was probably pre-heated to high temperatures by the intrusion of the older dikes. Furthermore, there is evidence of a major thermal pulse at ~ 2.5 Ga followed by slow cooling to ~ 2.4 Ga in the CDC/EDC (Jayananda et al. 2011, 2013b; Peucat et al. 2013). For example, in the southern part of CDC, garnet dated at 2439 ± 36 Ma and 2435 ± 56 Ma in metapelite and calc-silicate gneiss grew under granulite facies conditions (Jayananda et al. 2013b). Thus, the crust was at high temperatures ~ 65 Myr (at least ~ 10 Myr considering uncertainties) before the emplacement of the ~ 2.37 Ga old dikes. The CDC/EDC crust was also thermally primed by the intrusion of the Ippaguda-Dhiburahalli swarm (~ 2.25 Ga) ~ 5 Myr before the emplacement of the Kandlamadugu swarm (~ 2.22 Ga), and the Anantpur-Kunigal swarm (~ 2.21 Ga) was emplaced another ~ 10 Myr later. The prevailing high temperatures of the crust is evident from the presence of anatectic granites such as the 2221 ± 99 Ma old Yelagatti granitoid in northern EDC (Rogers et al. 2007). The EDC crust was also probably thermally primed by the intrusion of the Devarabanda swarm at ~ 2.08 Ga before the eruption of the Vempalle volcanics (> 1.89 Ga) and intrusion of the Tadpatri sills (~ 1.89 Ga) within Cuddapah Basin.

Potential temperatures

The apparent potential temperatures (T_P*, Krein et al. 2021) for the origin of the CDC/EDC and Cuddapah Basin primary magmas, estimated from the P–T of multiple saturation with spinel lherzolite with the Reverse EC-AFC models (10–16.5 ± 2.3 kbar, 1291–1366 ± 40 °C) and an adiabatic slope (dT/dP) of 1.5 °C/kbar, are 1233–1385 °C. The Reverse FC models indicate lower P–T of multiple saturation with spinel lherzolite, reflecting equilibrium with a metasomatized mantle. The Reverse FC results were not used to estimate potential temperatures. The true potential temperature (T_P) depends on the style and degree of melting (Krein et al. 2021). To assess the approximate degree of melting, the compositions of the calculated CDC/EDC and Cuddapah Basin primary magmas based on the Reverse EC-AFC models are compared with model melts of spinel lherzolite (Behn and Grove 2015) (Fig. 8). The N- to NNW-trending dike samples from the ~ 2.22 Ga old Kandlamadugu swarm show a restricted range of FeO and SiO₂, indicating ~ 10–20% melting at 15–20 kbar pressure according to the isobaric batch melting models. The NW- to WNW-trending dike samples from the ~ 2.21 Ga old Anantapur-Kunigal swarm show a restricted range of Na₂O and CaO/Al₂O₃, indicating 15–20% batch melting at pressures between ~ 15 kbar and > 20 kbar. Most of the ENE- to NE-trending dike samples from the ~ 2.37 Ga old Bangalore-Karimnagar swarm are similar to the NW- to WNW-trending Anantapur-Kunigal dike samples. Hence, their pressure and degree of batch melting are probably similar. The Cuddapah Basin samples show a negative correlation between FeO and Na₂O, and positive correlations between SiO₂ and Na₂O, and between FeO and CaO/Al₂O₃. These trends also suggest 10% to > 20% batch melting at pressures between ~ 12 kbar and > 20 kbar. However, all CDC/EDC and Cuddapah Basin primary magmas estimated above probably represent pooled melts as their average compositions are similar to a pooled melt generated by polybaric incremental melting at pressures between 26 and 9 kbar along a 1450 °C adiabat (Behn and Grove 2015). Assuming ~ 10–20% melting, T_P may be ~ 60–130 °C higher than the estimated T_P* for narrow to fully pooled melts (Table S2 of Krein et al. 2021). Thus, the range of T_P may be 1293–1515 °C.

Fig. 8

Compositions of the CDC/EDC and Cuddapah Basin primary magmas compared with model melts of spinel lherzolite (Behn and Grove 2015). Dashed lines: isobaric batch melts at 10, 15 and 20 kbar; red dot in a and b: pooled polybaric incremental melt generated along a 1450 °C adiabat

The estimated T_P values of all samples are similar to predicted ambient mantle temperatures in the Paleoproterozoic according to various thermal history models of the Earth (see Fig. 10 of Herzberg 2022). Even the highest estimated T_P (1515 °C) is not higher than the Paleoproterozoic ambient mantle temperature predicted by the model of Korenaga (2008). Thus, there is no indication of a mantle plume based on the above estimates of T_P. However, magma may lose heat during its passage through the lithosphere, especially in dike swarms where lateral flow may be important (Ernst and Baragar 1992; Ernst et al. 2019). In this case, the above T_P estimates may be lower than the actual T_P. To further assess the origin of the CDC/EDC and Cuddapah Basin basalts, their REE patterns are compared in the following section with the Mackenzie dikes of Canadian Shield that are evidently derived from a mid-Proterozoic mantle plume (Ernst and Baragar 1992; Baragar et al. 1996).

Plume origin?

The Mackenzie dikes are composed of subalkaline tholeiitic basalts with Mg# ranging 41–50 (averages in Table 6 of Baragar et al. 1996), similar to the CDC/EDC and Cuddapah Basin basalts considered above (Mg# 37–55). They also consist of plagioclase, pyroxene, and Fe-Ti oxides with rare olivine, and their estimated P–T of emplacement are ≤ 5 kbar/ ~ 1025–1225 °C (Baragar et al. 1996). Thus, the major element compositions and emplacement conditions of the Mackenzie dikes are similar to the CDC/EDC dikes and Cuddapah Basin sills.

The incompatible trace element patterns of the Mackenzie basalts show LILE enrichment relative to HFSE, LREE enrichment relative to HREE, enriched Nb-Ce plateaus, and negative anomalies for Sr, Ti and K (Baragar et al. 1996). Their εNd(t) values are mostly between 0 and + 2, indicating little contamination with the continental crust. By comparison, all CDC/EDC and Cuddapah Basin basalts show prominent negative anomalies for Nb, but not for K (Srivastava et al. 2015), and they show evidence of upper crustal contamination in incompatible trace element ratio diagrams (Fig. 4d, e, f). Their εNd(t) values (–10 to + 1) also suggest variable degrees of contamination (Anand et al 2003; Kumar et al. 2012b; Liao et al. 2019). Compared to the Mackenzie basalts, the REE abundances of the CDC/EDC and Cuddapah Basin basalts are lower, but they also show LREE enrichment relative to HREE (Figs. 9, S3). When the basalts are corrected only for fractionation (primary magmas with Reverse FC, Table S2), their chondrite-normalized La_N/Sm_N and Dy_N/Yb_N ratios remain largely unchanged (Fig. S3). But when they are corrected for fractionation and upper crustal assimilation (primary magmas with Reverse EC-AFC, Table S2), their La_N/Sm_N ratios decrease while their Dy_N/Yb_N ratios remain similar (Figs. 9, 10, S3). High La_N/Sm_N ratios of basalts may result from upper crustal contamination or through clinopyroxene-dominated fractional crystallization occurring at high pressures. Since the thermobarometric calculations do not show evidence of high-pressure crystallization, the elevated La_N/Sm_N ratios of the CDC/EDC and Cuddapah Basin basalts are probably due to upper crustal contamination.

Fig. 9

Chondrite-normalized (McDonough and Sun 1995) REE patterns of basalts and their primary magmas (Reverse EC-AFC models) from a ENE- to E-trending, b N- to NNW-trending, and c NW- to WNW-trending dikes of the CDC/EDC, and d sills and flows within the Cuddapah Basin. The range for the Mackenzie dikes (Baragar et al. 1996) is shown by two dashed lines

Fig. 10

Bivariate plot of chondrite-normalized (McDonough and Sun 1995) REE ratios for the CDC/EDC dikes, Cuddapah Basin sills and flows, and Mackenzie dikes (Baragar et al. 1996). The filled symbols represent primary magmas (Reverse EC-AFC models)

Compared to the CDC/EDC and Cuddapah Basin basalts, the Dy_N/Yb_N ratios of the Mackenzie basalts are distinctly higher (Fig. 10), indicating a deeper origin from a garnet-bearing source (Baragar et al. 1996). By contrast, the lower Dy_N/Yb_N ratios of all CDC/EDC and Cuddapah Basin basalts and their primary magmas, and the flat REE patterns of most of the primary magmas (Reverse EC-AFC models) (Figs. 9, S3) indicate melting of a primitive spinel lherzolite source (see modeling by Shellnutt et al. 2018). Some of the primary magmas even show La_N/Sm_N < 1 (Fig. 10), indicating the possibility of a depleted source. This does not preclude origin from a plume because plumes are chemically heterogeneous, often containing both enriched and depleted components (White 2010). However, neither the CDC/EDC and Cuddapah Basin basalts nor their primary magmas exhibit a deep melting signature such as the high Dy_N/Yb_N ratios observed in the plume-derived Mackenzie basalts (Fig. 10). Thus, there is no indication from the REE data of a plume-related origin for any of the CDC/EDC and Cuddapah Basin basalts.

The non-plume origin for all CDC/EDC dikes and Cuddapah Basin sills and volcanics agrees with the studies of Anand et al. (2003), Sheppard et al. (2017), Shellnutt et al. (2018), and Söderlund et al. (2019), whose conclusions are based on petrological modeling, and geological, isotopic, trace element, and field structural data. The ~ 1300–1500 °C potential temperatures estimated in this study (Reverse EC-AFC models) and Anand et al. (2003) are lower than the estimated temperatures (> 1575 °C) for the late Archean komatiitic amphibolites from Kolar schist belt (CDC-EDC boundary, Rajamani et al. 1985) that may have originated from a mantle plume. Herzberg (2022) also concluded that Archean and Paleoproterozoic komatiites with temperatures of ~ 1700 °C may have originated from mantle plumes, but the ~ 1550 °C temperatures estimated for the ~ 1.88 Ga tholeiitic intrusives of the Circum-Superior LIP indicate that a plume was not involved in their origin. Dike emplacement is controlled by factors such as crustal stress, crustal heterogeneity, magma viscosity and intrusion rates (Rivalta et al. 2015; Kjøll et al. 2019). A discussion of these mechanisms is beyond the scope this paper.

Conclusions

Mineral thermobarometry indicates that the CDC/EDC dikes around Cuddapah Basin crystallized at upper crustal P–T conditions of ~ 1–6 kbar and ~ 1120–1210 °C. Spinel and ilmenite equilibrated at subsolidus temperatures (~ 460–660 °C) and oxygen fugacity values below the fayalite-magnetite-quartz buffer (∆FMQ values of -0.5 to -3) that correspond to bulk Fe³⁺/∑Fe ratios of 0.06–0.16. Thirty selected CDC/EDC and Cuddapah Basin basalts of different ages from the literature (with minor correction in bulk composition for crystal accumulation) plot on their Ol-Pl-Cpx cotectic boundaries at 1 bar-6.5 kbar pressures, consistent with the pressure estimates from mineral thermobarometry. Primary magmas modeled through reverse fractionation calculations incorporating crustal assimilation (Reverse EC-AFC) show that they were last equilibrated with spinel lherzolite at P–T conditions of 10–16.5 kbar (± 2.3 kbar) and 1291–1366 °C (± 40 °C). Considering EC-AFC, the models show that the basalts are mostly ≤ 20% contaminated with the upper crust except for three samples that are 23–30% contaminated. The upper crust was thermally primed to high temperatures (665–808 °C) at the time of emplacement of the different dike swarms. A comparison with model melts of spinel lherzolite (Behn and Grove 2015) shows that basalts can be generated by ~ 10% to > 20% batch melting at ~ 12–25 kbar pressures. The estimated primary magmas all basalts probably represent polybaric incremental pooled melts generated along a ~ 1450 °C adiabat. The estimated range of mantle potential temperatures is 1293–1515 °C. These potential temperatures are not higher than ambient mantle temperatures in the Paleoproterozoic according to different thermal history models of the Earth, and are inconsistent with an origin of the basalts from mantle plumes. The incompatible element and REE patterns of the basalts are distinct from the plume-derived mid-Proterozoic Mackenzie basalts of the Canadian Shield, the latter showing higher chondrite-normalized Dy_N/Yb_N ratios indicative of melting of a garnet-bearing mantle source. By contrast, the lower Dy_N/Yb_N ratios of all CDC/EDC and Cuddapah Basin basalts and flat REE patterns of their primary magmas indicate origin from a shallower, spinel-bearing mantle source. The basalts show LREE enrichment over HREE, and their La_N/Sm_N ratios are higher than the La_N/Sm_N ratios of the primary magmas, which can be attributed to upper crustal contamination. The estimated low potential temperatures, melting in the spinel lherzolite stability field, and the REE characteristics of the basalts and their primary magmas do not support an origin of the CDC/EDC and Cuddapah Basin basalts from mantle plumes.