Introduction

The Indian Ocean Ridge system (IORS) is an upside-down “Y”-shaped ridge system with three main oceanic ridges: the northern Indian Ocean Ridge, the Southwest Indian Ridge (SWIR), and the Southeast Indian Ridge (SEIR) (Fig. 1). These three ridges intersect at the Rodriguez Triple Junction (RTJ; 25°30′S, 70°E; Humler and Whitechurch 1988; Munschy and Schlich 1989) (Fig. 1). The northern Indian Ocean Ridge can be further subdivided into four ridge segments: from north to south they are the Gulf of Aden spreading center, the Carlsberg Ridge (CR), Northern Central Indian Ridge (NCIR), and Southern Central Indian Ridge (SCIR) (Patriat et al. 1997; Ray et al. 2013) (Fig. 1). CR has the lowest half spreading rate (~11–16 mm/year) in the northern Indian Ocean Ridge system (Chaubey et al. 1993; Ramana et al. 1993; Raju et al. 2008); there are relatively few studies on CR (e.g. Mahoney et al. 1992; Ray et al. 2013), even compared to the other Indian Ocean spreading centers.

Fig. 1
figure 1

Bathymetric map showing the ridge system of the northern Indian Ocean. Sample locations of this study are shown as green circles and the inset shows detailed information of the sampling area. Other stations of published basalts (green diamonds for the Gulf of Aden, circles for CR, and squares for CIR-RTJ) are also shown. Yellow circles are used to discriminate stations with basalts having isotopic data from other CR stations (pink circles). Fracture zones (white lines) and hot spots (red diamonds) are also shown. CR Carlsberg Ridge, CIR Central Indian Ridge, RTJ Rodrigues Triple Junction, SWIR Southwest Indian Ridge, SEIR Southeast Indian Ridge

Full size image

Previous studies confirmed that mid-ocean ridge basalts (MORB) in the Indian Ocean differ from Atlantic and Pacific MORB in terms of Sr, Nd, and Pb isotopic composition (e.g. Dupré and Allègre 1983; Hart 1984). MORB in the Indian Ocean display a greater degree of isotopic heterogeneity per given length of ridge than in the Atlantic and far more than in the Pacific (Mahoney et al. 1992; Meyzen et al. 2007), for example, a well-defined upper mantle domain with higher 87Sr/86Sr and lower 143Nd/144Nd were identified in the Indian Ocean mantle. This domain extends under the CR-CIR (Rehkämper and Hofmann 1997; Nauret et al. 2006), and its northern arm extends into the Gulf of Aden (Rooney et al. 2012). Several different hypotheses have been proposed to explain the distinct isotopic characteristics of the upper mantle domain. Isotopic evidences of MORB from the Gulf of Aden indicate that sub-ridge mantle was contaminated by the Afar plume (Rooney et al. 2012). Studies of the CR-CIR suggest that the influence of the Reunion plume on the southern CIR accounts for the highly enriched compositions that exist between latitude 18°S and 20°S (Dyment et al. 2001; Escrig et al. 2004; Murton et al. 2005; Nauret et al. 2006; Füri et al. 2011). Additionally, Ray et al. (2013) proposed that the geochemistry of the CR-CIR MORB was governed by the oceanic upper mantle contaminated by ancient continental crust.

At present, few papers have been published on CR MORB (e.g. Subbarao et al. 1979; Mahoney et al. 1992; Ray et al. 2013), especially on their Sr–Nd–Pb isotopic composition; moreover, there is no study that compares MORB from different segments of the internal CR. CR MORB are regarded as compositionally distinct from IORS MORB having typical N-MORB characteristics (Subbarao et al. 1979). However, previously acquired detailed geochemical data of MORB from several locations on the CR are still inadequate to understand their petrogenetic evolutionary process (e.g. Dupré and Allègre 1983; Mahoney et al. 1989, 1992; Ray et al. 2013). Since the CR mantle is an important part of the distinct Indian Ocean mantle domain, more geochemical data of MORB are required to investigate the mantle source and melt evolution beneath the poorly studied CR. Comparison of MORB from different segments of the CR will give a new and detailed sight to the diversity of mantle composition and magmatic process within the CR. Additionally, a general comparison of MORB from different parts of the northern Indian Ocean Ridge is helpful to investigate the different influencing factors during the genesis of the Indian Ocean MORB. So in this study, we present new geochemical data of MORB from four different locations on CR (Fig. 1), including Sr–Nd–Pb isotope data, and perform a geochemical comparison between MORB from different segments of the CR and the northern Indian Ocean Ridge to investigate their petrogenetic processes and mantle source composition.

Sampling and petrography

The Carlsberg Ridge (ridge length ~2000 km) is characterized by rugged topography, steep inner valley walls and a wide rift valley floor similar to the slow-spreading ridge segment of the northern mid-Atlantic Ridge (MAR). CR has a NW–SE orientation that becomes more N–S at its southern end (Fig. 1), with non-transform discontinuities and along-axis bathymetric deeps (Ramana et al. 1993; Ray et al. 2013). The study area is located in the middle portion of the CR; the samples were collected from four sites (3.70°N, 63.81°E; 3.67°N, 63.75°E; 3.69°N, 63.72°E; and 3.69°N, 63.83°E) by Television Grab during the DY125-26 cruise in 2012 (Fig. 1). These samples are all pillow basalts and have a black surface with patches of brown iron and manganese (Fig. 2a, c). Except for one sample (CR03-2) with ~1 cm thick glassy rim (Fig. 2a), other samples are fine-grained basalts largely lacking fresh glassy rims due to alteration (e.g. Fig. 2c). Thin sections of the fresh samples show that some samples consist of aphanitic glass with a few tiny plagioclases and olivine phenocrysts (Fig. 2b), while others have typical basaltic structure with interstitial pyroxene in triangular lattices composed of acicular plagioclase (e.g. Fig. 2d).

Fig. 2
figure 2

Hand specimens (a, c) and photomicrographs (b, d) of glassy MORB from the CR; b tiny plagioclase and olivine phenocrysts in glasses; d quenched plagioclase laths in glassy MORB

Full size image

Analytical techniques

The fresh samples were selected and crushed to powder of 200 mesh in agate mortars at the Second Institute of Oceanography, State Oceanic Administration, Hangzhou, China. Glass and whole-rock major and trace element analysis was performed at ALS Minerals in Guangzhou, China. The major elements were analyzed by X-ray fluorescence (XRF); the analytical precision was better than ±2–5%. The trace elements were determined by inductively coupled plasma mass spectrometry (ICP-MS) after complete dissolution. The analytical precision for most of the trace elements was better than ±5%. Two rock standards (kinzingite SARM-45 and kiorite Gneiss SY-4) and two randomly selected repeat samples were chosen to monitor the data quality and reproducibility.

Sr–Nd–Pb isotope analysis was carried out by multicollector ICP-MS (Neptune MC-ICP-MS, Thermo Scientific) using internal standardization and external calibration with bracketing isotope SRMs (ALS Scandinavia, Luleå, Sweden). The samples were digested by HNO3 + HF in closed Teflon vials at 160 °C for 48 h; the digests were then dried and the excess F component was removed by addition of concentrated HNO3 and evaporation until dry (the procedure was repeated three times). The solid residue was re-dissolved with HCl by heating in closed vials at 160 °C for 1 h, and then evaporated until dry. All the samples were digested in duplicates, one for further process by Pb–Sr column separation, and one for further process by Nd column separation. For samples with high Sr and Nd concentrations where no Pb IRMs was required, alkali fusion was used for sample preparation. For Sr and Pb column separation, the residues from the above digestion-drying process were taken up in 70% HNO3 followed by Sr and Pb separation using Sr-specific Eichrom ion-exchange columns where the Sr was eluted in 0.01 M HNO3 and the Pb in a EDTA solution. After purification, each element fraction was dried and re-dissolved in 5% HNO3, ready for isotope analyses. In the process, ICP-SFMS was used for testing concentration and MC-ICP-MS was used to test isotopes. Where possible, the Sr and Pb concentrations were adjusted to 0.2–0.1 ppm, respectively, by dilution. For Nd column separation, the residues from the above digestion–drying process were taken up in 1 M HCl, then the rare earth elements (REEs) as a group were separated from the sample matrix solution by AG50-X8 cation exchange resin, evaporated until dry, re-dissolved in 0.4 M HNO3, and the Nd separated from the Sm by using a Ln Resin (Eichrom) chromatography column. After purification, the Nd fraction solution was analyzed by ICP-SFMS and the analyte concentration was adjusted to a Nd concentration of 0.1 ppm. The Sr and Nd isotope ratios were normalized to 86Sr/88Sr = 0.1194 and 146Nd/144Nd = 0.7219, respectively. During the analyses, the NBS987 Sr isotope standard yielded a 87Sr/88Sr value of 0.710266 ± 0.000006 (2σ, n = 2). The measured Nd isotope standard Thermo Nd yielded a 143Nd/144Nd ratio of 0.512420 ± 0.000009 (2σ, n = 2). The measured 206Pb/204Pb, 207Pb/204Pb, and 208Pb/204Pb ratios of standard NBS981 were 16.9374 ± 0.0085, 15.4916 ± 0.0076, and 36.7219 ± 0.0182 (2σ, n = 2), respectively. The standard deviations (SD) are reported from repeated analysis of each sample.

Geochemistry

Major elements

The major and trace element data of the CR basalts are shown in Table 1. The samples analyzed in this study belong to the low-K series with MgO contents of 7.52–9.51 wt%, SiO2 contents of 49.40–50.60 wt%, K2O contents of 0.09–0.27 wt%, and Na2O contents of 2.55–2.90 wt%. The samples were relatively fresh with low LOI and a narrow range of Mg# (0.60–0.68) (Table 1), and belong to the mid-Ti basalts (1.22–1.64 wt%) (Natland 1991). The Na2O, K2O, and TiO2 content of the samples decreased with increasing MgO content, while the content of SiO2, Al2O3, and CaO are relatively uniform in all samples (Fig. 3). Different segments of CR from north to south (10°N to 2°S) showed different composition characteristics, and the major elements varied widely with the lowest alkali in basalts from 9 to 10°N segment (Fig. 3). However, the MgO content of all the samples from CR is relatively high (about 7–10 wt%), indicating a lower degree of magma fractionation. The major element compositions of the samples from this study are consistent with those previously reported from the 3–4°N region of CR (Fig. 3).

Table 1 Representative major and trace element compositions of basalts from the CR
Full size table
Fig. 3
figure 3

Major elements versus MgO (all in wt%) diagrams. The CR basalts data for comparison with our samples are from Hekinian (1968), Cann (1969), Cohen and O’Nions (1982), Banerjee and Iyer (1991), Melson et al. (2002), Jenner and O’Neill (2012), and Ray et al. (2013)

Full size image

Trace elements

The chondrite normalized REE diagram (Fig. 4a) shows that all the CR samples have light REE (LREE)-depleted patterns typical of N-MORB, but the samples from the 9 to 10°N segment are most depleted in LREE, while the CR samples of this study and samples from the 3 to 4°N segment are somewhat less depleted in LREE [(La/Sm)N = 0.55–0.68] and have flat heavy REE (HREE) patterns [(Ce/Yb)N = 0.84–1.08)].

Fig. 4
figure 4

Chondrite-normalized rare earth element (REE) diagram (a) and N-MORB-normalized spider diagram (b) of CR basalts. The compositions of chondrite, E-MORB, and N-MORB are from Sun and McDonough (1989). Other data are from Jenner and O’Neill (2012), Gale et al. (2013), and Ray et al. (2013)

Full size image

The N-MORB-normalized incompatible element diagram (Fig. 4b) shows that CR basalts from different segments have different patterns. The 9–10°N segment samples and some 1–2°S segment samples are more depleted in some large-ion lithophile elements (LILE) than N-MORB, but the composition of the samples from this study and from the 3 to 4°N segment are similar to those of N-MORB, except for some samples that are enriched in LILE (such as Rb and K) (Fig. 4b); however, their (La/Sm)N ratios are far less than 0.15, which suggests that they cannot belong to E-MORB [(La/Sm)N > 0.15 Gale et al. 2013].

Sr–Nd–Pb isotope

The Sr–Nd–Pb isotope data is shown in Table 2. The isotopic compositions of MORB from this study have the highest radiogenic Pb and lowest Nd isotopic ratios within the CR basalts, which extends the published composition range of CR basalts (Fig. 5). Comparison with the Pacific and Atlantic MORB shows that the isotopic compositions of the northern Indian Ocean Ridge basalts have little overlap with those of the Pacific MORB, with relatively higher 87Sr/86Sr and lower 143Nd/144Nd (Fig. 5b). Additionally, the 87Sr/86Sr of the CR basalts is closer to that of NCIR and RTJ basalts with a range of 0.70257–0.70298, while the 87Sr/86Sr of the SCIR and Gulf of Aden MORB has a larger variation, from 0.70268 to 0.70411 (Fig. 5b). In the 206Pb/204Pb versus 207Pb/204Pb and 206Pb/204Pb versus 208Pb/204Pb diagrams (Fig. 5c, d), most of the northern Indian Ocean Ridge basalts are distributed on the mixing lines between the depleted mantle (DM) and upper continental crust (UCC) or enriched mantle II (EM2) whereas some basalts are near the mixing line defined by the DM and lower continental crust (LCC). Different segments of CR have different isotopic ratios (Fig. 5): the 9–10°N segment basalts are characterized by the lowest Pb isotopic ratios compared to the northern Indian Ocean Ridge basalts; basalts from this study (near 3.7°N) extend the CR MORB to higher radiogenic Pb and are distributed along the mixing line between the DM and UCC data; other basalts from the 3 to 6°N and 1 to 2°S segments are closer to the DM compositions despite the slightly higher Pb isotopic ratios. These observations suggest that the isotopic variation of CR MORB is wide and extends to the least radiogenic Pb end member within the northern Indian Ocean Ridge basalts.

Table 2 Sr–Nd–Pb isotopic compositions of CR basalts
Full size table
Fig. 5
figure 5

Data sources Pelagic sediments: Othman et al. (1989); DM: Hart (1984); HIMU, EM1 and EM2: Armienti and Gasperini (2007); LCC: Escrig et al. (2004); UCC: Hemming and McLennan (2001); CR: as shown in Table 2; the Gulf of Aden: Schilling et al. (1992); NCIR: Ito et al. (1987), Mahoney et al. (1989), Escrig et al. (2004), and Ray et al. (2014); SCIR: Mahoney et al. (1989), Rehkämper and Hofmann (1997), Escrig et al. (2004), and Nauret et al. (2006); RTJ: Michard et al. (1986), Price et al. (1986), and Ito et al. (1987). Additional data are from PetDB

Sr–Nd–Pb isotopes diagrams for basalts from the northern Indian Ocean Ridge.

Full size image

Discussion

The isotopic composition of the Indian Ocean MORB is different from that of the Pacific MORB and Atlantic MORB which have, on average, relatively higher 87Sr/86Sr and lower 143Nd/144Nd and 206Pb/204Pb ratios (Dupré and Allègre 1983; Hart 1984). Diversity in mantle source composition was thought to be the major reason for these differences (Ito et al.1987; Mahoney et al. 1992; Hofmann 2003). MORB samples from different ridges in the northern Indian Ocean show obvious variation in isotopic composition (Fig. 5), which is consistent with the observation that the mantle domain of the Indian Ocean MORB is more compositionally heterogeneous than that of the Pacific MORB (Mahoney et al. 1989; Rehkämper and Hofmann 1997; Escrig et al. 2004; Hofmann 2003; Sang et al. 2014). In the CR basalts, significant compositional variations can also be observed among the different ridge segments (Figs. 3, 5). Since mantle source is not the only factor that influences the basalt composition, in the following sections we examine the magmatic differentiation and partial melting of the CR basalts and the mantle source and discuss the cause of the compositional variations in CR basalts.

Fractional crystallization

The CR basalts in this study have small amounts of plagioclase phenocrysts and olivine phenocrysts (Fig. 2). However, most of the CR basalts have MgO content higher than 7 wt% (Fig. 3), indicating that the fractional crystallization of the parent magma is not extensive. The negative anomaly of Eu is also slight in CR basalts (Fig. 4), which excludes the possibility of significant crystallization of plagioclase. The fractional crystallization of clinopyroxene phenocrysts can decrease the content of HREE in basaltic magma (White 1991); however, the flat HREE pattern in our data suggests no major fractionation of clinopyroxene from the CR parent magmas (Fig. 4). In addition to the REE results, Fig. 6 shows that the Ni and Cr concentrations in the CR basalt specimens tend to decrease with increasing Y and Zr, which suggests crystallization of olivine and clinopyroxene of the parent magmas (Winter 2001). In contrast, Sr shows no significant change with Y and Zr, indicating weak fractional crystallization of the plagioclase (Fig. 6). Our results are consistent with published CR basalt data although the compositional trend is not so obvious in the published data. Generally, the combination of the data of the major and trace elements suggests that the CR basalts experienced minor rather than large-scale magmatic differentiation, and probably experienced only a moderate amount of olivine crystallization. This is consistent with the spreading rate (half spreading rate of 11–16 mm/year) of CR (Chaubey et al. 1993) because no large, stable magma chambers are present beneath the slow-spreading ridge (Dunn and Forsyth 2007), leading to less magmatic fractionation; thus, the MORB are more likely to preserve the information of the source mantle composition and mantle melting.

Fig. 6
figure 6

Data sources are as in Fig. 4

Ni/Cr/Sr–Y and Ni/Cr/Sr–Zr diagrams for CR basalts. All elements are in ppm.

Full size image

Mantle melting

The geochemistry of MORB is dominated by the source mantle composition, mantle melting, and fractional crystallization. In the Mg# versus (La/Sm)N plot (Fig. 7a), the majority of CR MORB have (La/Sm)N values of ~0.5 to ~0.7 except for the 9–10°N MORB (~0.3); these data do not show any variation with decreasing Mg#, suggesting a DM source for the CR MORB. To eliminate the influence of fractional crystallization, we use Na8 (Na2O content corrected for fractional crystallization to a common MgO content of 8 wt%; Klein and Langmuir 1987) to investigate the influence of mantle melting on the magmatic evolution of CR MORB. As a moderately incompatible element, Na concentration in the melt decreases as the degree of melting increases (Klein and Langmuir 1987; Elthon 1992). Figure 7b shows a positive correlation between Na8 and CaO/Al2O3, a further indicator of the degree of mantle melting (e.g. Condie 2003; Standish et al. 2008), suggesting that mantle melting plays a major role in controlling the major elements of the CR MORB (Fig. 7b). In the Zr versus Zr/Y (Fig. 7c) and Nb versus Nb/Y plots (Fig. 7d), the CR MORB samples show a strong positive correlation between the highly incompatible elements (Zr and Nb) and the highly incompatible/less incompatible element ratios (Zr/Y and Nb/Y), suggesting that Zr and Nb have more variability during mantle melting than Y, which is consistent with their higher incompatibility. This emphasizes the important role of mantle melting in controlling the trace element concentrations in CR MORB: fractional crystallization of MORB magma would cause similar variability of Zr and Y (Hofmann 2003); thus, Zr/Y will not show significant variation with rising Zr values.

Fig. 7
figure 7

Data sources are as in Fig. 4

Mg#–(La/Sm)N (a), CaO/Al2O3–Na8 (b), Zr–Zr/Y (c) and Nb–Nb/Y (d) diagrams for CR basalts.

Full size image

In Na8 versus CaO/Al2O3 plots (Fig. 7) the Na8 of CR MORB show a large variation (1.4–4.2), indicating a great variability of the melting degree of the CR mantle. The mantle source composition greatly affects the initial melting depth; thus, the degree of melting depends on the fertility of the source mantle (Cannat et al. 2008; Meyzen et al. 2005). The initial compositional characteristics of MORB are derived from the mantle source even though subsequent compositional trends are controlled by mantle melting. However, mantle composition is thought to be heterogeneous beneath slow- and ultraslow-spreading ridges that contain enriched or/and DM component (e.g. Murton et al. 2005; Liu et al. 2008). The refractory mantle domains may be difficult to infer from MORB composition analysis because they would have been too depleted to melt during the most recent mantle melting. However, the addition of an enriched component to normal MORB mantle should be detected by MORB composition analysis, especially by isotope analysis (Meyzen et al. 2005; Nauret et al. 2006).

Implication for source contamination

The large variation in the major and trace elements of the CR MORB and the similarity of their characteristics to those of enriched MORB (E-MORB) in CR 3–4°N MORB (including the data obtained in this study) (Fig. 4b) suggest that the MORB mantle beneath CR may incorporate some enriched components. Previous studies indicated that the MORB mantle beneath the Indian Ocean was contaminated by various enriched components, such as hotspot material (Nauret et al. 2006), recycled subducted altered oceanic crust and/or sediment (Dupré and Allègre 1983; Cohen and O’Nions 1982; Rehkämper and Hofmann, 1997), ancient subcontinental lithospheric mantle (Mahoney et al. 1989; Escrig et al. 2004), and delaminated continental crust (Meyzen et al. 2005; Ray et al. 2013). The enriched components beneath CR can be further identified by the isotope data of the CR basalts.

Sr–Nd–Pb isotope data of MORB are very sensitive indicators for detecting mantle source composition and have been used to constrain the anomalous upper mantle beneath the Indian Ocean which shows higher 87Sr/86Sr, 207Pb/204Pb, 208Pb/204Pb and lower 143Nd/144Nd and 206Pb/204Pb compared to those ratios in the Atlantic and Pacific Oceans (Dupré and Allègre 1983; Hart 1984). In addition, the heterogeneity of the mantle is more obvious in both space and time in the Indian Ocean than in the other oceans (e.g. Meyzen et al. 2005; Mahoney et al. 1998; Janney et al. 2005), which may be related to the spreading rate and the presence of hotspots close to the ridge. The influence of hotspots on the ridge may be enhanced at slow- to ultraslow-spreading ridges (Mahoney et al. 1992) and the heterogeneity of the mantle may be reduced by the magma extraction processes associated with fast-spreading ridges (Cohen and O’Nions 1982). However, it is generally accepted that the variation of the MORB isotopic composition is mainly caused by the heterogeneity of the mantle source.

Basalts from different segments of the northern Indian Ocean Ridge have different isotopic compositions (Fig. 5). The Gulf of Aden and SCIR MORB have the largest variation in isotopic composition among the northern Indian Ocean Ridge basalts, having the highest 87Sr/86Sr and radiogenic Pb and the lowest 143Nd/144Nd; Rooney et al. (2012) and Nauret et al. (2006) proposed that the mantle sources of these basalt groups may be influenced by material from the Afar and Reunion hotspots, respectively. In contrast, NCIR basalts show a narrower range of isotopic composition which may be due to the longer distance between the Reunion hotspot and NCIR compared to SCIR. The basalts from CR and RTJ which are relatively far from the Afar and Reunion hotspots (Fig. 1) show small variations in Sr and Nd isotope ratios, closer to the composition of DM compared with basalts from other ridges (Fig. 5b). However, these basalts still have higher 87Sr/86Sr and lower 143Nd/144Nd than DM (Fig. 5b), which may indicate an enriched component in the mantle source. The enriched component which contaminates the CR mantle is less likely to be hotspot material because compared with the northern Indian Ocean Ridge basalts, the CR MORB has the lowest 87Sr/86Sr (<0.7030) which is far less than the SCIR basalts (Nauret et al. 2006) influenced by the Reunion hotspot (87Sr/86Sr >0.7035; Escrig et al. 2004) (Fig. 5b). The ridge area that is most likely influenced by the hotspot is at the Marie Celeste Fracture Zone and on the adjacent ridge segment to the south; it is ~1100 km from the Reunion hotspot (Mahoney et al. 1989; Nauret et al. 2006), which is much less than the distance between the CR and the Reunion hotspot (>2000 km) (Fig. 1).

As shown in the Pb isotopic plots (Fig. 5c, d), most of the CIR MORB specimens lie on the mixing line between the DM and the UCC, which is consistent with contamination by the UCC component in the CIR mantle (Ray et al. 2013). The RTJ mantle may be contaminated by the LCC component as the basalts lie on the mixing line between the DM and the LCC in the 206Pb/204Pb versus 207Pb/204Pb plot; however, the 207Pb/204Pb values of the RTJ MORB are higher than those of the DM and LCC. In contrast, the CR MORB show large variability in the Pb isotopic composition, consistent with both mixing lines, which may suggest a more heterogeneous mantle contaminated by both the UCC and LCC components (Fig. 5c, d). The UCC contamination signatures are supported by the CR MORB data obtained in this study with the highest radiogenic Pb and lowest 143Nd/144Nd compared with the published CR MORB data (Fig. 5). Whereas the 9–10°N MORB which plot closest to the LCC have the least radiogenic Pb compared with the northern Indian Ocean Ridge MORB (Fig. 5c, d). Even though the extremely high 87Sr/86Sr and low 143Nd/144Nd, which characterize the LCC and UCC, are not observed in the CR MORB, the compositional trend defined by the majority of the northern Indian Ocean Ridge MORB do show that the CR MORB (and some RTJ MORB) which deviate from the trend have higher 87Sr/86Sr and lower 143Nd/144Nd values than the compositional trend (Fig. 5e, f). The influence of the hotspot material can also increase the radiogenic Pb and Sr; however, the UCC is the most plausible contaminant because the CR 3–4°N MORB with highest radiogenic Pb and Sr in CR MORB are far from the Reunion and Afar hotspots (>2000 km) and the CR 9–10°N MORB which are closer to the Afar hotspots have lowest radiogenic Pb and Sr compared with other northern Indian Ocean Ridge MORB (Fig. 5). Precambrian continental peridotite xenoliths often have unradiogenic Pb similar to that found in LCC (Cohen and O’Nions 1982; Sgualdo et al. 2015); hence, the possibility that the sub-ridge mantle may be contaminated by the continental lithospheric mantle cannot be excluded. However, no matter what the specific contaminant is, the evidence of contamination by the UCC, LCC, and/or continental lithospheric mantle suggests that the CR MORB mantle contains continental lithospheric material.

Basalts from different ridge segments of CR have distinct isotopic compositions with the lowest Pb isotope ratios found in CR 9–10°N and the highest ones in CR 3–4°N (Fig. 5); this is consistent with mantle contamination by continental lithospheric material. In contrast, the contamination signatures in basalts from other CR ridge segments are not significant (Fig. 5). This indicates that the MORB mantle beneath the CR is extremely heterogeneous. The heterogeneity of the CR mantle is strongly expressed by the presence of basalts having 206Pb/204Pb lower than 17.5 (Fig. 5). These basalts combined with the low Pb isotope ratios MORB from the RTJ (Price et al. 1986) (Fig. 5) and the SWIR 39–41°E (Mahoney et al. 1992; Meyzen et al. 2005) represent the least radiogenic Pb end member of all Indian Ocean MORB; their low 206Pb/204Pb values rule out a plume origin. Old subducted marine sediments also fail to explain the low 206Pb/204Pb, while an origin from LCC was proposed for extremely low radiogenic Pb of MORB from the SWIR 39–41°E (Meyzen et al. 2005). Ancient continental lithospheric mantle was also proposed as a source of MORB from the SWIR 39–41°E in several studies (Mahoney et al. 1992; Janney et al. 2005). Hence, we suggest that the low Pb isotope ratios in the CR MORB are the result of the melting of mantle contaminated by continental lithospheric materials. Involvement of continental materials in the oceanic mantle can be induced by either destabilization of the ancient continental lithospheric mantle by Mesozoic plume activity (Mahoney et al. 1992) or delamination of the continental lithosphere related to the breakup of Gondwana (Escrig et al. 2004; Ray et al. 2013). These two mechanisms can work effectively in the Indian Ocean ridge because the Indian continental lithosphere is unusually thin (down to ~50 km) (Negi et al. 1986). Evidence indicating the presence of continental lithosphere material in the mantle beneath different ridges in the Indian Ocean suggests that interaction between continental and oceanic lithospheres was prevalent during the opening of the Indian Ocean. Given that the continental lithospheric material was introduced into the asthenosphere before or during the breakup of Gondwana (Mahoney et al. 1992; Janney et al. 2005; Escrig et al. 2004; Ray et al. 2013), it could have been preserved in the asthenospheric mantle for a long time rather than being erased by mantle convection. Thus, the long-term preserved continental lithosphere may play a significant role in producing the isotopic anomaly of the Indian Ocean MORB.

Conclusions

Major and trace element analysis suggest that variations in the composition of MORB from the Carlsberg Ridge are dominated by partial melting variations other than fractional crystallization. Melting of a DM-like mantle with the addition of some enriched components may be responsible for the enrichment of incompatible elements such as K and Rb. The Sr–Nd–Pb isotopic compositions of CR basalts show that the CR mantle is contaminated by both upper and lower continental crust (or continental mantle) components, leading to a heterogeneous mantle on a regional scale (within ~2000 km). These results confirm that significant interactions occurred between the continental lithosphere and oceanic mantle during the opening of the Indian Ocean, suggesting that continental material played a significant role in producing the isotopic anomaly observed in the Indian Ocean MORB.