A new plate boundary started to form 160 million years ago, splitting the Gondwana super-continent into the American plate to the West and the African plate to the East. Today, the American and African rifted margins, that were once neighbors, are separated by the Atlantic Oceanic domain produced by the Mid-Atlantic oceanic ridge.

Both the Afar hot spot to the north of the East African Rift and the Tristan hot spot in the southernmost part of the American and African conjugated margin system, are thought to have strongly impacted the continent that began to break, Africa and Gondwana, respectively. The rising Tristan hot plume is thought to have triggered the Paraná-Etendeka igneous province (a large area with massive continental flood basalts), and further weakened and thinned the Gondwana lithosphere to ultimately lead to breakup. However, we know from seafloor spreading magnetic anomalies that seafloor spreading started more than 2000 km south of the Paraná-Etendeka igneous province and propagated northward, reaching the Walvis Ridge, the inferred plume tail of the Tristan hot spot, ~110 million year ago, well after the last sporadic continental flood basalt events ~120 million years ago. This northward unzipping of the South Atlantic, starting far away from the plume head, led thus to question the triggering role of the Tristan plume impingement.

 Volcanic rifted margins, like the southernmost American and African margins on both sides of the Southern Atlantic Ocean, are typically associated with a thick magmatic layer. These large melt volumes were often interpreted as mainly due to melting of anomalously hot sublithospheric material. Although hot plume-sourced mantle is not the unique ingredient, it is conventionally considered as the key one to create these large melt volumes during rifting, and thicker than average oceanic crust during subsequent seafloor spreading. This large volumes of magma have been used early on to estimate the mantle thermal anomaly thought to be responsible for this excess melt production. However, one of the main pitfalls of this approach lies precisely in determining the volume of magma at volcanic rifted margins.

 At volcanic rifted margins, continental crust might be highly intruded resulting in a hybrid crust within the ocean-continent transition. Intrusions into thinned continental crust cannot be properly resolved using geophysical methods (e.g. seismic reflection and refraction). By contrast, the seismic crustal thickness within the oceanic domain is a proper and widely used estimation of the melt supply per increment of plate separation. It can be readily obtained by picking the top and base (oceanic Moho) of the magmatic oceanic crust on seismic reflection profiles. This approach requires access to high-quality deep-penetrating seismic profiles.

 We used 24 such profiles provided to us by ION Geophysical^TM, to investigate how much melt was produced at the onset of seafloor spreading, after the formation of the large melt volumes along the rifted margins and the complete rupturing of the continental lithosphere. We found that, along ~75% of the length of the initial spreading centre, the crustal thickness is similar to regular oceanic thickness. Thus, most of the southernmost Atlantic Ocean opened without anomalously hot or fusible mantle, high magma supply being restricted to the Walvis Ridge area, the inferred plume tail of the Tristan hot spot. We suggest that the magma budget along the northern part of the South Atlantic is the result from a trade‐off between the activity and the distance to the Tristan hot spot.

 Our results may have important implications for the origin of the large melt volumes in the more proximal parts of the South Atlantic margins and other volcanic rifted margins. Because there is no anomalously hot mantle beneath the initial South Atlantic spreading ridge away from the Walvis ridge area, one could argue that hot mantle potential temperatures (with a 150-300°C thermal anomaly relative to the ambient mantle) still triggered the production of large volumes of melt along the rifted margins (while volcanic activity climaxed in the Paraná-Etendeka igneous province) and then went back to normal 8 million years later at the onset of seafloor spreading. This mantle potential temperature variation may be compared to the one along the Mid-Atlantic ridge south of Iceland, a classic example of a ridge-centred plume. There, the mantle potential temperature variation inferred to produce the crustal thickness variations along the Reykjanes ridge is an order of magnitude smaller for the same timescale (3-8 million years). This suggests that such a fast cooling of the mantle is rather unlikely.

 Alternatively, we propose that either the magma volume along the South Atlantic volcanic margins has to be re-evaluated downward and/or explanations other than a hotter mantle, prior to the onset of seafloor spreading, have to be favored to explain massive magmatic production there. This is in line with recent results from a numerical approach demonstrating that the large volumes of magma at volcanic rifted margins can be explained by depth dependent extension and very moderate excess mantle potential temperature significantly smaller than previously suggested. Therefore, we suggest that hot plume-sourced mantle, that is conventionally considered as the key parameter to create large melt volumes during rifting and thicker than average oceanic crust during initial seafloor spreading, may not play such a role of booster away of the plume head.