A relative timing framework

Recent high-resolution U/Pb geochronology provides a detailed chronology of Siberian Traps magmatism and the end-Permian mass extinction4, 8, and the ability to directly compare the relative timing of the two events. Building on the schema presented in ref. 4, we construct a framework to guide identification of the causative LIP subinterval. In this framework, the Siberian Traps magmatic activity can be segmented into three distinct emplacement stages. Stage 1, beginning just prior to 252.24 ± 0.1 Ma, was characterized by initial pyroclastic eruptions followed by lava effusion (Fig. 2). During this stage, an estimated 2/3 of the total volume of Siberian Traps lavas was emplaced (>1 × 106 km3). Stage 2 began at 251.907 ± 0.067 Ma, and was characterized by cessation of extrusion and the onset of widespread sill-complex formation4. These sills are exposed over a >1.5 × 106 km2 area, and form arguably the most aerially extensive continental sill complex on Earth. Intrusive magmatism continued throughout stage 2 with no apparent hiatus. Stage 2 ended at 251.483 ± 0.088 Ma, when extrusion of lavas resumed after an ~420 ka hiatus, marking the beginning of stage 3. Both extrusive and intrusive magmatism continued during stage 3, which lasted until at least 251.354 ± 0.088 Ma, an age defined by the youngest sill dated in the province4. A maximum date for the end of stage 3 is estimated at 250.2 ± 0.3 Ma6.

Fig. 2 Stages of Siberian Traps LIP magmatism relative to timing of mass extinction. Stages are color coded to the dominant style of magmatism at the time, purple for extrusive, and green for intrusive. Weighted mean 206Pb/238U zircon dates are shown at 2-sigma, and are also color coded purple for extrusive and green for intrusive4. Carbonate carbon record from refs. 8, 38. Stars located on the carbonate carbon record are stratigraphic locations of U/Pb dates8 Full size image

Integration of LIP stages with the record of mass extinction and carbon cycle at the Permian-Triassic Global Stratotype Section and Point (GSSP)4 shows three notable relationships. (1) Extrusive eruption during stage 1 of Siberian LIP magmatism occurs over the ~300 kyr prior to the onset of mass extinction at 251.941 ± 0.037 Ma4, 8. During this interval, the biosphere and the carbon cycle show little evidence of instability, although high-latitude environmental stress prior to the mass extinction has been observed22 (Fig 2). (2) The onset of stage 2, marked by the oldest Siberian Traps sill4, and cessation of lava extrusion, coincides with the onset of mass extinction and the abrupt (2–18 kyr) negative δ13C excursion immediately preceding the extinction event4. The remainder of LIP stage 2, which is characterized by continued sill emplacement, coincides with broadly declining δ13C values following the mass extinction. (3) Stage 3 in the LIP begins at the inflection point in δ13C composition, whereupon the reservoir trends positive, toward pre-extinction values.

Identifying the smoking gun

The disparity in duration between Siberian LIP magmatism and the end-Permian mass extinction, and the observation that pre-extinction eruption of an estimated 2/3 of LIP lavas resulted in limited deleterious global forcing on the biosphere23, suggests that only a restricted subinterval of LIP magmatism triggered the mass extinction. Siberian Traps lava eruptions appear to have ceased immediately prior to the extinction, and these lavas likely lack the dissolved CO 2 contents required to drive a significant global heating event24, both of which suggest the causative aliquot was probably not extrusive. Initial impingement of a plume head on the lithosphere, and related melting, is also hypothesized to generate significantly higher volatile yields at depth than those released during accompanying LIP lava eruptions14. However, our preferred extinction age model4 indicates that the end-Permian event did not occur during this period of theoretically higher plume-related devolatization (inferred to have initiated at 252.8 Ma14), but postdated it by ~900 ka.

Fig. 3 Time series of Siberian Traps LIP emplacement. a Pre-emplacement basin. b Emplacement of a volcanic load during stage 1. The feeder system is unresolved, and most likely situated below lavas in f. c Beginning of stage 2, with lateral sill complex growth, widespread heating, and greenhouse gas generation. d Continued sill emplacement during stage 2. e Renewed extrusive magmatism during stage 3. Geochronology defining time-steps from ref. 4. Map inset f modified from ref. 12 Full size image

A striking temporal coincidence is instead observed between the extinction event and the emplacement of Siberian Traps sills, which intruded the thick Tunguska basin, composed of evaporite, clastic, carbonate, and hydrocarbon-bearing rocks12. Heating of sediments over the large area encompassed by the sill complex (>1.5 × 106 km2) likely liberated massive volumes of greenhouse gasses12, 25. However, sills were intruded over an interval of ~500 kyr4, while the extinction interval and carbon isotope anomaly are both an order of magnitude shorter. This disparity necessitates that a currently unidentified subinterval of intrusive magmatism during stage 2 is the extinction-triggering aliquant. The oldest sills dated in the magmatic province mark the onset of stage 24, and are the only sills whose emplacement timing is coincident with both the carbon isotope excursion and the onset of mass extinction; all other dated sills postdate these events (Fig. 2), and are thus disqualified as the trigger. Therefore, we suggest that only intrusions emplaced at the beginning of stage 2, during the initial lateral growth of the Siberian Traps sill complex, satisfy all the necessary criteria to qualify as the trigger of the end-Permian mass extinction.

The exact cause for the observed transition from lava eruptions to widespread sill complex formation during stage 2 is challenging to pinpoint. However, emplacement of these sills occurred immediately after cessation of stage 1 lava extrusion, possibly in response to or as a result of construction of a thick volcanic load atop the lithosphere during stage 1 magmatism (Fig. 3). Numerical modeling studies illustrate that volcanic loads can impose a compressive stress state in the underlying crust that impedes magma ascent26, and in some instances, will promote sill formation27. Transitions from dominantly sub-vertical lava feeder systems to horizontal modes of intrusion (i.e., sills) are also attributed to rotation of the least compressive stress direction due to high magma flux into the feeder network28, and contrasting mechanical properties between layers of varying density and rigidity29. Once the Siberian sill complex was established, sill intrusion continued into stage 3, coincident with renewed lava effusion. The impetus for stage 3 extrusive magmatism in the northern region of the LIP is unclear, but may be the result of a new eruption center30, or eclipsing of a critical magmatic overpressure threshold at depth26.

Prior to growth of the sill complex, magmas feeding stage 1 lavas likely transited the crust rapidly through sub-vertical conduits within a narrow area presently buried either beneath the Siberian LIP lavas, or represented as sparsely exposed (possibly radiating) dikes in northern parts of the province30. Consequently, only a small volume of basin sediments experienced contact metamorphism during stage 1 and thus environmental impacts were likely relatively negligible. Not until the Siberian LIP emplacement style changed at the onset of stage 2 from being dominantly extrusive to intrusive was the “volatile-fertile” Tunguska basin subjected to widespread contact metamorphism, thereby maximizing the likelihood of heat transfer and massive volatile yield (Fig. 3)12.

Lateral sill-complex emplacement continued, subsequent to the initial pulse, but into a basin depleted of its initially high volatile content. Heat transfer to sediments from these later sills still generated greenhouse gasses; their effects are evident from the prolonged isotopically negative δ13C interval at the GSSP, and the sustained elevation in global sea surface temperature8, 10. The biosphere’s ability to buffer itself against prolonged, rather than initially large and abrupt greenhouse gas input, however, likely mitigated many of the deleterious effects of this gas generation24. Sluggish biotic recovery following extinction may well be attributed to stages 2 and 3 magmatism, but the rapid loss in global biodiversity characterizing the mass extinction suggests an equally rapid trigger. Initial intrusion of the Siberian Traps sill complex represents such a trigger, as it was short lived, coincident with the mass extinction, and capable of producing the large amount of climate-altering gasses required to drive biosphere collapse.

Making a deadly LIP

Our model suggests that LIPs characterized by sill-complex formation are more likely to trigger mass extinction than their flood basalt- and/or dike-dominated counterparts. The composition of sediment into which sills are emplaced is also critically important31, 32, as a sill network built within a volatile-poor substrate will not result in volatile generation on the scale necessary to drive extinction. Some of the largest LIP-related extinction events (>13% genus-level extinction33) in the last 300 Myr are demonstrably associated with widespread sill emplacement into sedimentary basins during Siberian Traps, Central Atlantic magmatic province, and Karoo-Ferrar LIP magmatism4, 15, 17 (Fig. 1). The notable exception over this period is the flood basalt-/dike-dominated Deccan Traps LIP, which is temporally associated with the ~66 Ma Cretaceous-Paleogene (K-Pg) extinction event19, 34. The model presented here suggests a limited role for Deccan magmatism in triggering the K-Pg event, which would shift burden to the contemporaneous Chicxulub impact34. It is plausible that the Deccan LIP and Chicxulub impact shared the causal burden34, and it is therefore possible that neither would have driven extinction of K-Pg magnitude acting alone35. The model presented here suggests that LIPs characterized by a pulse of widespread sill emplacement into a volatile-fertile basin are lethal on a global scale.

Data availability

Geochronology supporting the model presented here is available through refs. 4, 8. Treatment of this data is available from the corresponding author, S.D.B.