Mantle plume and rift-related volcanism during the evolution of the Rio Grande Rise

Large oceanic bathymetric swells can form from excess magmatism related to mantle plumes, subduction processes, or by rifting of continental fragments. The Rio Grande Rise (RGR) and Walvis Ridge are major South Atlantic features commonly attributed to the Tristan-Gough mantle plume, with parts forming near the Mid-Atlantic Ridge (MAR) between ~120 and ~80 Ma. However, recent petrological and geophysical observations have suggested parts of the RGR could be detached continental fragments (a microcontinent). Additionally, the Cruzeiro do Sul Rift (CdSR) and the unsampled Jean Charcot Seamount Chain (JCSC) complicate the tectono-magmatic history. This study investigates whether the RGR formed predominantly by plume-related magmatism at or near a spreading axis, by later intraplate rift-related volcanism, or whether continental fragments underlie it. The authors present new major and trace element data from the ERGR, WRGR, JCSC, and DSDP Site 516 to constrain petrogenesis, volcanic setting, relationship to the Tristan-Gough plume, and assess evidence for continental crust beneath the RGR.

Previous models attribute the RGR-Walvis Ridge system to the Tristan-Gough plume interacting with the MAR, forming thick oceanic crust up to 35 km. Tholeiitic lavas at DSDP Site 516 (80–87 Ma) support co-formation with Walvis Ridge near a plume-influenced spreading center. Subsequent tectonic evolution involved separation of the Walvis Ridge and WRGR (80–60 Ma) and development of the N–S trending ERGR. Around ~60 Ma, ridge jumps isolated the RGR on the South American Plate, with plume volcanism continuing along the Walvis Ridge–Guyot Province toward Tristan and Gough. A later WRGR volcanic event at ~46 Ma with distinct geochemistry was identified, coincident with development of the NW–SE CdSR and possibly linked to major plate reorganization. Satellite altimetry reveals the JCSC aligned with the CdSR, but it had not been sampled previously. In contrast to plume-only models, some studies argue for continental crustal fragments within the WRGR based on negative Bouguer anomalies and Proterozoic ages of dredged granitic/metamorphic boulders, suggesting possible Gondwanan affinity. These interpretations challenge purely volcanic models and motivate a comprehensive geochemical reassessment.

Sampling: During RV Maria S. Merian cruise MSM-82 (Mar/Apr 2019), 47 volcanic rocks were dredged from the ERGR (seven sites) and WRGR (six sites) across steep, sediment-poor flanks at 915–5266 m water depth. An additional 31 samples were dredged from six volcanoes of the JCSC. Five drill core samples from DSDP Hole 516F (core 128) were obtained from the IODP Bremen repository. Sample preparation: Onboard, samples were cut; altered rims and Mn crusts removed; fresh interiors retained. Onshore, pieces were ultrasonically cleaned, dried (60 °C, 12 h), coarsely crushed, and unaltered fragments milled in agate. Powders were dried at 105 °C for 12 h. DSDP samples were prepared similarly. Analytical selection: 83 total samples were analyzed; to mitigate alteration effects, only samples with loss on ignition (LOI) <5 wt% were used (56 of 83), with two slightly above threshold (5.3–5.8 wt%) included due to macroscopic freshness. Major elements: XRF on fused glass beads using Spectro XEPOS He at GeoZentrum Nordbayern (precision better than 1.2%; accuracy better than 4.2% except P2O5 ~12.8% on BE-N standard). Trace elements: Solution ICP-MS (Thermo X-Series2) with HF-HNO3 digestion, perchloric fuming, re-dissolution and dilution; desolvating nebulizer (Cetac Aridus II); tuning for high sensitivity and low oxides. Accuracy and precision better than 5% and 3% respectively on BHVO-2. Geochemical approach: Focused on fluid-immobile elements (Ti, Y, Zr, Nb, Hf, Th, REE) to minimize seawater alteration effects. Used discrimination diagrams (Nb/Y vs Zr/Ti; Nb/Yb vs TiO2/Yb), primitive mantle-normalized multi-element patterns, and trace element ratios (e.g., Nb/Zr, Nb/Th, Hf/Th, Nb/La) to classify tectono-magmatic settings and assess crustal contamination. Modelling: Forward partial melting modeling with REEBOX Pro (v1) to estimate degrees and pressures of melting and mantle potential temperatures (Tp). Assumed heterogeneous mantle sources: ERGR and DSDP Site 516 modeled with 80% depleted MORB mantle (DMM) + 20% primitive mantle; WRGR alkaline lavas with 70% primitive mantle + 30% DMM. Active upwelling mixing functions appropriate for plume-influenced ridge settings were used. Garnet-spinel transition field considered (garnet >2.7 GPa; spinel <3.0 GPa). Only samples with MgO ≥4 wt% were used in classification and modeling to minimize fractional crystallization effects.

• ERGR and older WRGR lavas are tholeiitic basalts with moderately enriched, E-MORB-like signatures consistent with formation above the Tristan-Gough plume near a spreading center (overlap with Walvis Ridge and DSDP Site 516 on Nb/Y vs Zr/Ti and Nb/Yb vs TiO2/Yb diagrams).
• Younger WRGR and JCSC lavas are alkaline, intraplate OIB-like, geochemically similar to Tristan da Cunha and Gough, indicating low-degree partial melting beneath thicker lithosphere, likely linked to rifting along the Cruzeiro do Sul Rift (CdSR).
• No geochemical evidence for assimilation of continental crust: RGR lavas lack decreasing Nb/Th or Nb/Zr with decreasing MgO; Hf/Th, Nb/Th, Nb/La ratios are unlike contaminated continental-influenced suites (e.g., Florianópolis Dyke Swarm) and more like Walvis Ridge plume-related basalts.
• Temporal-geochemical heterogeneity in WRGR: tholeiitic stage at 87–80 Ma (DSDP 516) vs alkaline stage at ~46 Ma (published and new dredges along CdSR), implying prolonged magmatic evolution with a switch from plume–ridge to intraplate rift-related volcanism.
• Partial melting conditions from REEBOX Pro: • ERGR and DSDP 516: degrees of melting ~2–5%; pressures ~2.2–2.7 GPa; Tp ~1400–1440 °C; modeled crustal thickness ~15–19 km, matching geophysical estimates (~15–17 km). • WRGR alkaline lavas: degrees of melting ~0.5–2%; pressures ~3.0–3.2 GPa; Tp ~1420–1440 °C; modeled crustal thickness ~21–24 km, consistent with thick WRGR crust (~25 km).
• JCSC geochemistry broadly resembles WRGR alkaline lavas but with variations in Nb/La and Nb/Zr suggesting small-scale mantle heterogeneity and potentially distinct enriched mantle sources.
• Dredging across RGR slopes (>3000 m) recovered exclusively angular volcanic rocks; combined with lack of crustal contamination signals and seismic velocity structure, indicates RGR is dominantly volcanic rather than a microcontinent.
• Proposed linkage: Eocene rifting along CdSR thinned lithosphere and triggered low-degree melts that fed both WRGR late-stage alkaline volcanism and JCSC formation; possible tectonic connectivity to Cabo Frio High and analogies with Cameroon Line.

The new major and trace element data resolve the origin of the RGR as primarily volcanic, formed initially by plume–ridge interaction with the Tristan-Gough mantle plume producing tholeiitic, E-MORB-like basalts (ERGR and older WRGR). Subsequently, intraplate rifting along the CdSR induced low-degree, deeper partial melting of enriched mantle beneath thickened lithosphere, generating alkaline OIB-like volcanism on the WRGR and along the JCSC. The absence of crustal assimilation trends and the recovery of only volcanic lithologies across varied depths and locations argue against significant continental basement beneath the sampled areas, countering recent microcontinent hypotheses for the WRGR. Melt modeling quantifies the shift in melting regime: moderate degrees at shallower pressures near a plume-influenced ridge during the tholeiitic stage versus very low degrees at higher pressures during the alkaline stage. The geochemical differences between WRGR and JCSC alkaline lavas, despite their spatial and temporal association, imply small-scale mantle heterogeneity beneath the western South Atlantic. These results refine the tectono-magmatic evolution of the RGR, emphasizing a multi-stage development over tens of millions of years involving both plume–ridge processes and later lithospheric rifting, and clarify that late-stage volcanism on the WRGR was not simply a counterpart to late events on the Walvis Ridge but had distinct mantle sources and tectonic drivers.

The study demonstrates that the Rio Grande Rise is a dominantly volcanic large igneous province. The ERGR and the older WRGR formed as tholeiitic basalts above the Tristan-Gough plume near the Mid-Atlantic Ridge, while younger alkaline volcanism on the WRGR and the JCSC resulted from low-degree melting beneath thick lithosphere during Eocene rifting along the Cruzeiro do Sul Rift. Geochemical proxies show no evidence for assimilation of continental crust, and dredging recovered only volcanic rocks across the plateau flanks, supporting a volcanic, not microcontinental, origin. Quantitative melt modeling constrains mantle potential temperatures (~1400–1440 °C) and melting conditions consistent with plume influence and subsequent intraplate rifting. Future work should include precise geochronology for ERGR and JCSC samples, comprehensive radiogenic isotope studies to resolve mantle source heterogeneity, expanded geophysical imaging to refine crustal architecture, and integrated tectonic modeling to further link rifting processes with magmatism across the South American margin.

• Many new WRGR and JCSC samples lack direct radiometric ages; ages for some WRGR alkaline lavas are inferred by geochemical similarity to previously dated (~46 Ma) samples dredged along the CdSR.
• Interpretations of mantle sources rely on trace element systematics; comprehensive isotopic datasets for all new samples were not presented here.
• Melt modeling results depend on assumed source compositions (DMM/primitive mantle mixes), upwelling/mixing functions, and mineral stability fields; alternative source lithologies or temperature structures could yield different quantitative estimates.
• Alteration is mitigated but not entirely eliminated; although fluid-immobile elements were emphasized, some samples had LOI up to ~5.8 wt% and two such samples were included due to macroscopic freshness.
• Spatial sampling, while broad (ERGR, WRGR, JCSC), cannot capture the full heterogeneity of such a large province.