The Siberian craton was affected by flood basalt volcanism at least twice during the Devonian (Yakutsk-Vilyui province) and Permian–Triassic (Siberian province) periods. In both cases volcanism appeared as brief pulses of flood basalt eruptions, followed by kimberlitic (and lamproitic) emplacement. Pressure estimations for the kimberlite-entrained mantle xenoliths reflect that the lithosphere was 190–230 km thick at the time of the Devonian flood basalt volcanism. Differently from Devonian kimberlites, the majority of Triassic kimberlites are diamond free, but at least one Triassic kimberlite pipe and some lamproites are diamondiferous, suggesting that the Siberian lithosphere remained thick during the Permian–Triassic flood basalt volcanic activity. If both the lithosphere and the asthenosphere were volatile poor, thick cratonic lithosphere prevented melting even at an elevated geotherm. During the Paleozoic, Siberia was surrounded by subduction systems. The water deep cycle in association with fast subduction and slab stagnation in the mantle transition zone is proposed to cause fluxing of the asthenosphere by water plus other fluids via wet diapir formation in the mantle transition zone. Such diapirs started to melt in the asthenosphere beneath thick cratonic lithosphere, producing voluminous melts. Mafic melts probably accumulated beneath cratonic lithosphere and rapidly erupted on the surface in response to stress-induced drainage events, as assumed for some other cratonic flood basalts.

Various types of evidence, including the size and volume of the Siberian Traps, the timing and duration of eruptions, paleotectonic and paleogeographic reconstructions, lithospheric structure, heatflow, and the trace-element and radiogenic isotope compositions of lava, are reviewed in this chapter. The major evidence may be summarized as follows. The Siberian Traps erupted in a number of brief volcanic events from the Late Permian until the end of the Middle Triassic. They occupied a vast region (∼7 × 10 6 km 2 ) in a back-arc tectonic setting. The overall volume of erupted rocks was as much as ∼4 × 10 6 km 3 , with most of the volume erupted within the Tunguska syncline. This syncline experienced long-term subsidence before initiation of the volcanism, and the region is now underlain by a relatively thin lithosphere, which is ∼180 km thick. Two types of trace-element patterns are observed in the Siberian Traps: subordinate high-Ti ocean island basalt–like patterns and dominant low-Ti island arc basalt–like patterns. In radiogenic isotope and trace-element coordinates, mixing trends between these two types of magma are absent, or at least not evident. Some volcanic rocks contain primary magmatic mica. These are considered in light of different models. Each model can explain, or was thought to explain, particular observations. However, some evidence can be fatal for some models. For example, the enormous size and volume of the Siberian Traps cannot be explained in the framework of impact and edge-driven convection models and are problematic for lithospheric delamination models. Plume models face problems in explaining the uplift and subsidence pattern and the absence of mixing curves between expected high-Ti primary plume melts and contaminated low-Ti melts. Therefore, a model that relates Siberian Trap magmatism and subduction is suggested. In this model, subducting slabs brought significant amounts of water into the mantle transition zone. Consequent release of water from the transition zone lowered the solidus of the upper mantle, leading to voluminous melting. Major supporting observations for this model include (1) the tectonic position of the Siberian Traps in a back-arc setting of Permian subduction systems, (2) island arc basalt–like trace-element patterns for the majority of the erupted basalts, (3) primary mica found in volcanic rocks, and (4) experimental data on the high water capacity of the mantle transition zone and its recharging via the subduction process.

Understanding of the style of mantle convection and determining the mass flux across the upper–lower mantle boundary is dependent on several critical questions, including whether the mantle is chemically stratified, where the boundary between chemically distinct mantle regions lies, what the compositions of these mantle regions are, to what depths slabs subduct, and what mechanism of mass and heat transfer is used. Answers to these questions are expected in geochemical and geophysical data. Many geochemical observations that have been interpreted as evidence for vigorous mass flux from the lower mantle and the core via plumes to the upper mantle (e.g., He-Sr-Nd-Pb-Os isotopic ratios in oceanic basalts) can be interpreted as involvement of recycled lithospheric material without lower-mantle plumes. Geophysical observations provide no strong constraints on penetration of slabs deep into the lower mantle. Deep slab penetration, if any, could occur during agglomeration of super-continents. Most subducting slabs remain in the upper mantle or at the boundary between the upper and the lower mantle. The primary mechanism of convection in the upper mantle is cooling from above. However, ancient slabs can become buoyant after radiogenic heating if they contain sufficient sediments bearing radioactive elements. These heated slabs will rise, contributing to upward convective flow. Mass flux between the upper and the lower mantle is either absent or highly limited, but further studies are required to resolve this question.