Interactions of the Greater Ontong Java mantle plume component with the Osbourn Trough

The Ontong Java-Manihiki-Hikurangi plateau (OJMHP) is considered to have originated from a starting mantle plume, and have been rifted apart by two spreading ridges. However, the ages of these spreading ridges and their possible interactions with the presumed mantle plume are unclear. The Manihiki-Hikurangi plateau has been rifted apart by the Osbourn Trough which formed the southwestern Pacific crust to the east of the Tonga-Kermadec trench. Here we report Pb-Hf-Os isotopes of the basaltic crust (Site U1365 of IODP Expedition 329) formed by the Osbourn Trough. Linear regression of Re-Os isotopes results in an age of 103.7 ± 2.3 Ma for Site U1365 basalts, indicating that the Manihiki-Hikurangi plateau was rifted apart by the Osbourn Trough with a spreading rate of ~190 mm/yr. The superfast spreading rate supports the Osbourn as an abandoned segment of the early Pacific spreading ridge, which initially overlapped with the giant starting plume. Moreover, the Pb-Hf isotopes of some of Site U1365 basalts show distinct differences from those of the Pacific mid-ocean ridge basalts, while they are similar to the basalts of the Ontong Java and Manihiki plateaus. We suggest that the OJMHP mantle plume components has been involved by the Osbourn spreading center.


Samples and Results
Basalt samples from Site U1365 of IODP Expedition 329, which is located ~250 km to the north of the Osbourn Trough ( Fig. 1), were analyzed for Hf-Pb-Os isotopes in this study. Only the fresh samples based on thin section observations are selected for analyses. Results of Re-Os isotopes, Hf and Pb isotopes in this study are shown in Supplementary Table. These basalt samples have very low but variable contents of Os (0.81-6.3 ppt), while contents of Re are less variable (0.65-1.56 ppb), indicating strong Re/Os fractionation during magmatic processes (Fig. 2). These basalt samples have wide ranges of 187 Re/ 188 Os (689-3395) and 187 Os/ 188 Os (1.368-6.098) (Supplementary Table). Although these samples have ( 206 Pb/ 204 Pb) t (18.05-18.54) within the range of the Pacific normal mid-ocean ridge basalts (N-MORBs) (Fig. 3), several samples are decoupled from the East Pacific Rise (EPR) N-MORBs on the plot of ( 208 Pb/ 204 Pb) t vs. ( 206 Pb/ 204 Pb) t (Fig. 4a). Their Ɛ Hf t are also distinctly lower than those of EPR N-MORBs (Fig. 4b).

Re-Os Systematics and Age
The concentration of Re tends to increase with decreasing MgO and Al 2 O 3 and increasing SiO 2 , TiO 2 , Nb and La (Fig. S1), which are consistent with the moderate incompatibility of Re during magma fractionation as suggested by previous studies 17,18 . These correlations corroborate that these samples are fresh and exclude the 'nugget' effect 19 , which is further excluded by replicated analyses of Re (see Supplementary Table). Unlike Re, Os is strongly compatible during mantle melting and magma fractionation 20 , which can result in an elevated ratio of 187 Re/ 188 Os in this study. The low concentrations of Os in this study and their lack of correlation with concentrations of Re (Fig. 2b), MgO and Ni (Fig. S2) indicate a role of either mantle source heterogeneity or 'nugget' effects, however, replicated analyses of Os also indicate a negligible 'nugget' effect (Supplementary Table). Mixing of intraplate magma with ancient oceanic crust with a high 187 Os/ 188 Os ratio might increase the ratio of 187 Os/ 188 Os. However, the depleted trace element patterns of these samples and geologic setting of Site U1365 point to MORBs and exclude them as ocean island basalts 13 , thus, contamination of ancient crust during magma processes is unlikely. Because Re and Os behave differently during magma processes, the highly variable concentrations of 187 Os and their positive correlation with Re ( Fig. 2) are related to post eruption radiogenic ingrowth. The extremely low abundance of 188 Os indicates low initial 187 Os, thus, the post-eruption radiogenic 187 Os resulted from the high 187 Re/ 188 Os ratios accounts for a majority of the total 187 Os abundance.
These samples show excellent correlation and linearity (R 2 of 0.999) between 187 Re/ 188 Os and 187 Os/ 188 Os (Fig. 3). Samples in this study have much higher ratios of 187 Os/ 188 Os than those of global modern MORBs, which cannot be related to mantle source heterogeneity. The positive correlation of 187 Os/ 188 Os vs. 1/ 188 Os was resulted from radiogenic ingrowth in these samples with variable initial 187 Re/ 188 Os ratios. The well-correlated ratios of   Os/ 188 Os results in an apparent age of 103.7 ± 2.3 Ma for Site U1365 basalts (Fig. 3). An initial 187 Re/ 188 Os ratio of 0.196 is obtained for this suite of samples based on the linear regression, which is much higher than those of EPR N-MORBs (Fig. 3), e.g., ~0.133 ± 0.009 in average based on Gannoun et al. 21 (this average value would be 0.1327 after correction to 103.7 Ma according to the average Re/Os ratio of peridotite in Liu et al. 22 ). The calculated initial 187 Os/ 188 Os ratio of 0.196 is much higher than the global modern MORBs, implying a mantle source with long-term enrichment of Re relative to Os. The regression age from Re-Os isotopes is interpreted to reflect the formation age of Site U1365 basalts.

Nature and Spreading History of the Osbourn Trough
The cessation age and spreading rate of the Osbourn Trough have long been debated because the Osbourn Trough was active during the Cretaceous normal superchron (125-84 Ma) [11][12][13] . Estimated cessation age of the Osbourn Trough varies between 105 Ma-71 Ma 11,12,14,23,24 . The Osbourn Trough has been shown to have a medium to slow spreading rate (60-80 mm/yr) according to the morphology of the fossil ridge 11 , which, however, could also have been a fast ridge with decreasing rate before cessation. It was also considered as a fast ridge which belongs to a segment of the early Pacific spreading center 12 , which could have originally overlapped with the OJMHP 25,26 and rifted apart the Manihiki-Hikurangi plateau immediately after its formation. Based on the age of 103.7 Ma for the basalt of Site U1365, if the rifting of the joint Manihiki-Hikurangi plateau by the Osbourn spreading center occurred immediately after its formation (e.g., ~119 Ma) 1 , the Osbourn Trough would have a minimum full spreading rate of 190 mm/yr according to the distance of ~1500 km between Site U1365 and the Manihiki plateau. This spreading rate exceeds that of the fastest southern East Pacific Rise (160 mm/yr) 27 . This also indicates that the medium to low spreading rate based on the fossil ridge morphology 11 could have resulted from the decreases in spreading rate before cessation.
This superfast spreading rate supports the prediction that the Osbourn Trough belongs to an extinct section of the Cretaceous Pacific-Phoenix ridge 12,23 . Thus, the joint Manihiki-Hikurangi plateau would have overlapped with the Cretaceous Pacific-Phoenix ridge during its initial formation. Moreover, if a constant spreading rate is assumed for the Osbourn Trough before cessation, it would have ceased at ~101 Ma according to its distance from Site U1365. This age is in concert with the collision time of the Hikurangi plateau with the Chatham rise ( Fig. 1) at ~100 Ma 14 , which is considered as the cause of cessation of the Osbourn spreading 12,14 .

Interactions with the OJMHP Mantle Plume Component
The initial 187 Os/ 188 Os ratio of 0.196 calculated for this suite of samples is much higher than those of the Pacific N-MORBs (Fig. 3) (0.126 to 0.148, based on Gannoun et al. 21 ). The 187 Os/ 188 Os ratios in this study are systematically higher than the isochron line of 103.7 Ma with an initial 187 Os/ 188 Os ratio of 0.133 (Fig. 3), which corroborates that the samples in this study have systematically higher ( 187 Os/ 188 Os) t than the Pacific N-MORBs. Such high initial 187 Os/ 188 Os ratios (0.196) have never been observed in the EPR MORBs, indicating secular enrichments of Re relative to Os in the mantle source or contamination from a mantle source having long-term Re-enrichment relative to Os. Because Re and Os are not sensitive to metasomatism caused by melts or fluids 25 , potential mantle sources enriched in Os isotope (with high 187 Os/ 188 Os) would not have been derived from metasomatized oceanic/continental lithospheric mantle. Magmatic processes can cause strong enrichment of Re relative to Os 28,29 , thus, a mantle source with anomalous high ( 187 Os/ 188 Os) t could have been related to contamination of crustal materials, e.g., recycled continental/oceanic crust and terrigenous sediments. Origin of such high initial 187 Os/ 188 Os ratio could be either recycled components imbedded in depleted asthenosphere, or contamination from an external component during mantle melting/evolution. However, the unusually high initial 187 Os/ 188 Os ratio of Site U1365 basalts relative to the EPR MORBs (and also global MORBs) imply that it is unlikely resulted from melting of normal asthenosphere under the spreading ridge.
Based on the unique geologic setting of Site U1365, contamination of a mantle source with high ( 187 Os/ 188 Os) t , e.g., through interactions with nearby mantle plumes, should be evaluated. The Louisville seamount chain intersects the Osbourn Trough at its western end (Fig. 1). Previous studies showed that the Louisville seamount chain might have influenced the Osbourn Trough magmatism through plume/ridge interactions 30 . However, according to the cessation time at ~101 Ma for the Osbourn Trough the Louisville seamount chain would not have interacted with the Osbourn Trough magmatism, because the oldest seamount that intersects the Osbourn Trough is ~79 Ma 31 . A recent study showed homogeneous and normal compositions of the LSC basalts in Os isotope (( 187 Os/ 188 Os) t (0.1245-0.1314) 32 , which further rules out the possibility of contamination from the Louisville seamount chain.
The mantle plume components of OJMHP are potential sources that have contaminated the lavas formed by the Osbourn Trough 15 . Based on the exposed crust in Solomon islands, the Ontong Java plateau is composed of two isotopically distinct groups of volcanic rocks, the Kawaimbaita-Kroenke type and the Singgalo type 7,33 . An increasing number of studies show that these two groups are widely distributed in the Ontong Java, Manihiki and Hikurangi plateaus 2,3 . The Singgalo type basalts have lower 206 Pb/ 204 Pb and Ɛ Hf (Enriched Mantle 1-or EM1-like component) than those of the Kawaimbaita-Kroenke type (Fig. 4) 7,33,34 . The basalts of both types from the Ontong Java plateau have exceptionally high ( 187 Os/ 188 Os) t (e.g., up to 0.4 for Singgalo type and 0.26 for Kawaimbaita type), which are considered to have been derived from recycled continental crust 33 . Publication on Re-Os isotopes of the Manihiki and Hikurangi basalts has long been lacking. However, a recent study of Schaefer 35 reported high ratios of ( 187 Os/ 188 Os) t (up to > 0.17) for the Manihiki low-Ti basalts (ranging isotopically from the Kawaimbaita-Kroenke type on the OJP to a HIMU component).
The OJMHP Singgalo/Kawaimbaita type components with high ( 187 Os/ 188 Os) t ratios are likely the potential contaminant in the source of Site U1365 basalts. Additionally, the low-Ti Kawaimbaita basalts from Manihiki were also shown to have high ( 187 Os/ 188 Os) t ratios 35 . Site U1365 basalts have different ( 206 Pb/ 204 Pb) t and distinct ƐHf t from EPR N-MORBs and their initial 187 Os/ 188 Os ratio is between the OJMHP samples and EPR N-MORBs, indicating that the OJMHP mantle source might have contributed to melts of the Osbourn spreading center. Contamination of a N-MORB type mantle source by the OJMHP mantle source is sensitive in ( 187 Os/ 188 Os) t ratios because of the higher Os contents in plateau samples (> 20 ppt in average) than Site U1365 samples.
Site U1365 basalts have variable ( 208 Pb/ 204 Pb) t ratios for a given ratio of ( 206 Pb/ 204 Pb) t from typical EPR MORBs to the domain of OJMHP basalts (Fig. 4a). These samples also deviate from the EPR MORBs and extend to the range of OJMHP basalts on the plot of ( 206 Pb/ 204 Pb) t vs. Ɛ Hf (Fig. 4b). These also imply contamination of the OJMHP mantle source on the Site U1365 basalts. One sample in this study has relatively high ( 206 Pb/ 204 Pb) t and ( 208 Pb/ 204 Pb) t and low Ɛ Hf t , which is similar to the Kawaimbaita basalts of the Ontong Java plateau.
The oceanic crust formed by the Osbourn spreading center is currently subducting into the Kermadec-Tonga trench (Fig. 1). Castillo et al. 36 reported Pb isotopes of basalt samples dredged from the incoming plate of Kermadec-Tonga trench. The data on the N-MORBs reported by Castillo et al. 34 are plotted in Fig. 4 for comparison. Similar to Site U1365 basalts, several samples reported by Castillo et al. 36 also deviate from the EPR MORBs and extend towards the OJMHP domain on plot of ( 208 Pb/ 204 Pb) t vs. ( 206 Pb/ 204 Pb) t , one of which is solely in the domain of Singgalo-type basalts (Fig. 4). These data and the new data in this study jointly support the contamination of the mantle source of Osbourn spreading center by the OJMHP mantle plume component.
One likely explanation for the contamination of Site U1365 basalts by the OJMHP mantle plume component is plume/ridge interations 15 . The volcanism caused by this starting mantle plume is distributed far more than on these plateaus. The crust in the Nauru basin and the Pacific crust to the east of the Mariana trench also show similar formation age and geochemistry to the OJMHP basalts 37 . This is analogous to effects of the Iceland mantle plume on the mid-Atlantic ridge. The effects of the OJMHP mantle plume on the Pacific spreading centers are expected to be stronger than the Iceland mantle plume on the mid-Atlantic ridge, because the superfast spreading ridge would have caused more consumption of mantle plume material for a given period.
Detailed sampling and age-dating studies on the Ontong Java, Manihiki and Hikurangi plateaus showed that they were formed in at least two stages, 125-117 Ma and 96-88 Ma, respectively 2,34 . This is evidence for continued volcanic activities in the Manihiki and Hikurangi plateaus after the major volcanic stage (125-117 Ma) on the OJMHP. It is difficult to infer the location of the presumed plume center over the two volcanic stages, however, the geochemical similarities of basalts formed by the two stages indicate a common mantle source 7,34 . Thus, the volcanic activities on the OJMHP have continued to later than 100 Ma. However, it is not clear if these volcanic activities after the major stage of OJMHP formation are related to the presumed mantle plume tail after the starting mantle plume melting. Additionally, there is lack of clear seamount chains between the Osbourn Trough and the Manihiki and Hikurangi plateaus, which indicate flows of mantle plume components to the ridge.
A starting mantle plume has been supposed to spread horizontally to a diameter exceeding 2000 km in the shallow mantle 38,39 , which melts and produces a giant oceanic plateau in a short period (i.e., 5 Myr). Both the Kawaimbaita-and the Singgalo-type basalts are distributed on the Ontong Java, Manihiki and Hikurangi plateaus, corroborating the widespread distribution of plume component in the southwestern Pacific. The starting mantle plume materials might have stagnated in the asthenospheric mantle after formation of the overlying oceanic plateau. If the residual mantle plume material of the OJMHP has stagnated in the asthenosphere, it would have been entrained by the Osbourn spreading center and contributed to the MORB magmas of the Osbourn Trough.

Conclusion
We report Pb-Hf-Os isotopes of Site U1365 basalts from IODP Expedition 329 to investigate the mantle source nature and spreading history of the Cretaceous Osbourn Trough. The basalt samples show much higher 187 Os/ 188 Os ratios than global modern MORBs that is obviously caused by post-eruption radiogenic ingrowth. The good linearity (R 2 of 0.999) of 187 Os/ 188 Os vs. 187 Re/ 188 Os gives an apparent regression age of 103.7 ± 2.3 Ma for Site U1365 basalts. This age is consistent with the prediction that the Manihiki-Hikurangi plateau was rifted apart by the Osbourn spreading center with a superfast rate (~190 mm/yr). This also indicates a cessation time of ~101 Ma for the Osbourn trough if a constant spreading rate is assumed, which is in concert with the time of collision between the Hikurangi plateau and Chatham Rise. Moreover, Site U1365 basalts have an initial 187 Os/ 188 Os of 0.196 based on regression of 187 Os/ 188 Os vs. 187 Re/ 188 Os, which is much higher than those of EPR MORBs. The Pb-Hf isotope compositions of Site U1365 reflect contamination from mantle plume components from the OJMHP. We propose that the mantle source of the Cretaceous Osbourn spreading ridge could have been contaminated by the detached lithosphere of the OJMHP during its rifting.

Methods
Pb-Hf isotope analyses. Sample powders used for analyses of Pb-Hf isotopes in this study were leached in hot 6 N HCl for ~30 minutes to remove any potential seawater contamination, rinsed in water purified in a Milli-Q reverse osmosis system to remove any residual traces of acid and dried. For analyses of Pb isotopes, sample powders were dissolved with 1 mL HNO 3 plus 4 mL HF. Then, the dissolution was dried and added with HNO 3 for three times until there was no HF left. The dissolution was dried and dissolved by 1 mol/L HBr and transferred to the microtube for centrifugation. The AG1-8 anion resin was used for Pb separation using the standard procedure. Pb isotopes were analyzed using a Nu instrument MC-ICP-MS at Institute of Geology Chinese Academy of Geological Sciences (IGCAGS). Samples were "spiked" with a Tl standard ( 203 Tl-205 Tl isotopes) to correct for mass-dependent isotopic fractionation. Blank of Pb of the procedure was 0.14 ppb. The entire procedure was monitored using standards SRM-981, and multiple analyses (n = 6) of SRM-981 yielded 206  Hf isotopic data were obtained using a Neptune plus (Thermo Finnigan) multi-collector inductively coupled plasma mass spectrometer (MC-ICP-MS) at Nanjing University. 100 mg powders were leached for 12 h in warm 2.5 N HCl, and dissolved in 15 ml Teflon beakers in an HF-HClO 4 acid mixture at 120 °C for more than 5 days. After evaporation to dryness, all samples were dried at 200 °C in order to break CaF bonds. Finally, the samples were dissolved in 3 N HCl. Hafnium was separated from the rock matrix by ion exchange procedures using Eichrom ® Ln-Spec resin. The detailed analytical procedure for the Hf isotopic measurement can be seen elsewhere 40 . Hf isotopic ratios were normalized to 179 Hf/ 177 Hf = 0.7325. The results were then normalized to a 176 Hf/ 177 Hf value of 0.282160 using the daily average of the JMC 475 Hf standard. The JMC 475 Hf standard analyzed over the period of the analyses gave an average value of 176 Hf/ 177 Hf = 0.282157 ± 0.000005. In addition, international standards BHVO-2 and BCR-2 were also tested in this method. Measured values for BHVO-2 and BCR-2 were 0.283082 ± 0.000004 and 0.282857 ± 0.000006, respectively (Reference values are 0.283101 ± 0.000026 and 0.282867 ± 0.000018, respectively 41 ). Re-Os isotope analyses. Re-Os isotope analyses were performed in the National Research Center for Geoanalysis, Chinese Academy of Geological Sciences (NRCG-CAGS). Detailed procedures of chemical separation for Re-Os can be referred to Du et al. 42 and Li et al. 43,44 . Samples were loaded in a Carius tube through a thin neck funnel. The mixed 190 Os and 185 Re spike solutions and 3 mL HCl and 6 mL HNO 4 were loaded while the bottom of the tube was frozen at − 50 °C to − 80 °C in an ethanol-liquid nitrogen slush with the top sealed by an oxygen-propane torch. The tube was then heated for 24 h at 230 °C. The bottom part of tube was kept frozen during the cooling. The Os was separated through distillation from carius tube for 50 min and was trapped in 5 mL 1:1 HBr. Micro distillation was used for N-TIMS (Triton) determination of Os isotope ratio. The residual Re-bearing solution was saved in a 150 mL Teflon beaker for Re separation.
The residual Re-bearing solution was heated to near-dryness. Then 10 mL of 50% NaOH were added to the residue followed by Re extraction with 10 mL of acetone in a 120 mL Teflon funnel. The acetone phase was transferred to a 100 mL beaker that contains 1 mL water. It was evaporated to dry and picked up in 2% HNO 3 that was used for the N-TIMS determination of Re isotope ratio. The purified Os and Re were loaded to Pt filaments respectively, and analyzed via negative ion thermal ionization mass spectrometry (NTIMS) using a second electron multiplier (SEM) in peak-hopping mode for Os and by static Faraday collectors for Re. The Re and Os isotope ratios were corrected for mass fractionation using 185 Re/ 187 Re = 0.59738, and 192 Os/ 188 Os = 3.08271. The concentrations of Re and Os were determined by isotopic dilution method, while Os was calculated based on the concentration and natural abundance of 188 Os in Os. The total procedural blanks of this study are approximately 0.27 ± 0.01 ρ g for Os, 0.93 ± 0.04 ρ g for Re, and 187 Os/ 188 Os of the blank is 0.205 ± 0.054. The standard material, GBW04477 (JCBY, a net-textured sulfide ore from Jinchuan Cu-Ni sulfide deposit, China), was used to monitor the accuracy of the method. The Re and Os contents of the JCBY determined are 38.28 ± 0.11 ppb and 16.12 ± 0.05 ppb, respectively, which are within uncertainty of the reference values (Re = 38.61 ± 0.54 ppb, Os = 16.23 ± 0.17 ppb).