The Permian-Triassic mass extinction: the Siberian Traps, the 'Great Dying,' and the collapse of the Paleozoic biosphere

Intuition Beginner

About 252 million years ago, at the end of the Permian period, the Earth experienced the most severe mass extinction in its history. Paleontologists call it the "Great Dying." Roughly 96 percent of all marine species and 70 percent of land vertebrate families disappeared. Among the mammal lineage, only a few small burrowing species survived.

The trigger was a colossal volcanic eruption in what is now Siberia: the Siberian Traps. Over roughly two million years, the eruptions poured out about four million cubic kilometers of lava, enough to bury all of Russia under a layer one kilometer thick. The magma burned through coal-bearing sediments, releasing enormous quantities of carbon dioxide and methane.

The planet warmed by five to eight degrees Celsius. The oceans acidified and lost their oxygen, the ozone layer may have been damaged, and most species could not adapt in time. Recovery took five to ten million years, far longer than after the asteroid-driven extinction 66 million years ago. The Permian-Triassic event ended the Paleozoic era and cleared the way for the Mesozoic dinosaur radiation.

Visual Beginner

Stage	What happens	Timescale
Siberian Traps LIP	~4 million km³ of basalt intrude coal-bearing sediments	~2 Myr beginning ~252 Ma
Carbon release	CO₂ and methane from coal and mantle, ~10,000-100,000 Gt C	pulsed over ~60 kyr
Climate cascade	5-8°C sea-surface warming, ocean acidification, marine anoxia, ozone loss	~100 kyr
Extinction pulse	96% marine species, 70% terrestrial vertebrate families	~60 kyr, two pulses
Recovery	Marine diversity rebounds; reefs rebuild	5-10 Myr

The cascade runs from deep-mantle trigger to surface-biosphere collapse, with the carbon cycle as the load-bearing link between the two.

Worked example Beginner

At Meishan, China, the Global Stratotype Section for the Permian-Triassic boundary exposes the extinction horizon in marine limestones.

Step 1. Below the boundary clay layer, late-Permian beds hold diverse faunas: brachiopods (lamp shells), bryozoans, crinoids, ammonoids, and the last trilobites. Species counts in a single bed number in the hundreds.

Step 2. Above the boundary, the earliest-Triassic sediment holds only a handful of disaster taxa, most notably the opportunistic bivalve Claraia. The trilobites, having survived 270 million years since the Cambrian, are gone. The ammonoids lose roughly 98 percent of their diversity.

Step 3. High-precision dating brackets the main extinction pulse at about 60,000 years in this section. Recovery to pre-extinction marine diversity takes about 5 million years; reef ecosystems take about 10 million years to rebuild.

What this tells us: the extinction was geologically sudden, but the recovery was one to two orders of magnitude slower than the kill event itself.

Check your understanding Beginner

Formal definition Intermediate+

Definition. A mass extinction is a geologically rapid (spanning less than a few million years, often much less) global increase in the extinction rate sufficiently elevated above the background rate to eliminate a substantial fraction of Earth's species. Raup and Sepkoski in 1982 ^{[Raup-Sepkoski 1982]} identified the Phanerozoic's "Big Five": the Ordovician-Silurian (445 Ma, ~85% marine species), Late Devonian (375 Ma, 75%), Permian-Triassic (252 Ma, 96%, the most severe), Triassic-Jurassic (201 Ma, 80%), and Cretaceous-Paleogene (66 Ma, ~76%, asteroid-driven).

Definition. The Permian-Triassic extinction cascade is the chain of Earth-system consequences linking the Siberian Traps large igneous province to the P-Tr biosphere collapse:

1. LIP emplacement. Approximately 4 million km³ of basaltic lava erupted over ~2 Myr (Reichow 2009 ^{[Reichow 2009]}), beginning at ~252 Ma.
2. Carbon amplification. Traps sills intruded the Tunguska Basin coal-bearing sediments, thermally metamorphosing coal and organic-rich evaporites, releasing ~10,000-100,000 Gt C as CO₂ and methane.
3. Climate cascade. Global sea-surface temperature rose ~5-8°C; the carbonate compensation depth shoaled (acidification); extensive marine anoxia and euxinia developed, recorded as global "black shale" deposits (Wignall-Twitchett 1996 ^{[Wignall-Twitchett 1996]}); ozone depletion is indicated by anomalous sulfur-mass-independent-fractionation signatures.
4. Extinction. Two pulses (Song 2013 ^{[Song 2013]}): the first at the P-Tr boundary, the second 60 kyr later. Marine: trilobites eliminated, ammonoids nearly lost (98%), all reef builders gone. Terrestrial: ~70% vertebrate families lost; the dicynodont Lystrosaurus dominated post-extinction assemblages.
5. Recovery. Marine diversity rebounded over ~5 Myr; reef ecosystems over ~10 Myr (Payne 2004 ^{[Payne 2004]}), much slower than after the K-Pg.

Counterexamples to common slips Intermediate+

• "An asteroid caused the P-Tr." The evidence is much weaker than for the K-Pg. No confirmed P-Tr-age impact crater of sufficient size exists. The proposed Bedout structure on the northwest Australian shelf and the Antarctic Wilkes Land anomaly remain debated. The sulfur-MIF ozone-depletion signal begins before any candidate impact horizon, pointing to volcanogenic halogens instead. An impact may have contributed, but the Traps-driven cascade is sufficient on its own.

• "All mass extinctions are volcanic." The K-Pg is asteroid-driven (Chicxulub), with the Deccan Traps contributing ancillary stress. The P-Tr and likely the T-J (Central Atlantic Magmatic Province) are volcanically-driven. The Ordovician-Silurian and Late Devonian appear driven by glaciation and oceanographic change rather than volcanism.

• "The Siberian Traps killed life directly, by lava." The lava covered ~7 million km² of Siberia, but the kill mechanism was global: carbon-cycle warming, acidification, and anoxia. Organisms thousands of kilometers from the Traps died from the cascade, not from lava. Direct burial is a local effect; the mass extinction is a carbon-cycle event.

• "The extinction took millions of years." The main pulse was ~60 kyr in high-resolution sections (Bowring 1998; Burgess-Bowring-Shen 2014), possibly structured as two pulses separated by ~60 kyr (Song 2013). The total duration of the crisis is short on geologic timescales.

• "Life recovered quickly." Marine diversity took ~5 Myr and reefs ~10 Myr to return to pre-extinction levels, an order of magnitude slower than after the K-Pg. The "sluggish" recovery reflects repeated carbon-cycle instability through the Early Triassic.

• "Trilobites were already extinct." Trilobites had declined in diversity from their Cambrian peak but remained extant through the entire Paleozoic. The P-Tr eliminated the last surviving order (Proetida), ending a 270-million-year lineage that had weathered the four earlier Big Five extinctions.

Key result: the Siberian Traps-extinction cascade Intermediate+

Result. The Siberian Traps large igneous province triggered the P-Tr extinction via a cascade of CO₂-driven warming, ocean acidification, and marine anoxia. The Traps' lava volume (4 million km³ over ~2 Myr), combined with carbon amplification from intruded coal-bearing sediments (10,000-100,000 Gt C), is sufficient to drive the observed P-Tr carbon-isotope excursion (~5‰) and the climate cascade. The high-precision timing match between the Traps eruption and the extinction is the central evidence.

The timing match

Renne and Basu in 1991 ^{[Renne-Basu 1991]} dated the Siberian Traps to ~251 Ma using high-precision ⁴⁰Ar/³⁹Ar geochronology, matching the P-Tr boundary within uncertainty. Bowring, Erwin, and colleagues in 1998 ^{[Bowring 1998]} applied U/Pb zircon dating to volcanic ash beds interlayered with the fossil-bearing marine section at Meishan, China, constraining the main extinction pulse to roughly 60,000 years. Burgess, Bowring, and Shen in 2014 ^{[Burgess-Bowring-Shen 2014]} refined both dates and demonstrated synchrony: the Traps eruption began just before the extinction, with the main extinction confined to a ~60 ± 48 kyr window.

The geochemical evidence

Three independent geochemical signatures support the cascade. Marine carbonate δ¹³C drops by ~5‰ across the P-Tr boundary, indicating a massive injection of isotopically light carbon (organic/coal-derived, δ¹³C ~ -25‰); mass-balance requires release of ~10,000 Gt C at the lower bound, consistent with coal-intrusion amplification. Marine δ¹⁸O records indicate a ~5-8°C sea-surface temperature rise across the P-Tr, consistent with greenhouse warming from the carbon release. Anomalous mass-independent sulfur-isotope fractionation in P-Tr sections indicates ozone-layer depletion, consistent with volcanogenic halogen release from the Traps and from intruded evaporites.

The two-pulse structure

Song and colleagues in 2013 ^{[Song 2013]} resolved the extinction into two pulses: the first at the P-Tr boundary (the Clarkina meishanensis conodont zone), eliminating approximately 57 percent of marine species, and the second approximately 60 kyr later (the Isarcicella isarcica zone), eliminating approximately 71 percent of the survivors. The two-pulse structure is consistent with pulsed carbon release from the Traps.

Limitations and residual debate

The Traps-trigger hypothesis is the consensus, but residual debate concerns: (i) the Traps' own trigger mechanism (mantle plume versus non-plume mantle dynamics); (ii) whether an asteroid impact contributed (the evidence is weaker than for the K-Pg and no confirmed crater exists); (iii) the relative importance of direct halogen-mediated ozone destruction versus indirect warming-driven anoxia. None of these uncertainties overturn the central cascade; they refine its details.

Derivation. The carbon-isotope mass-balance is the load-bearing quantitative constraint. Let $M_0$ be the pre-event ocean-atmosphere carbon reservoir with isotopic composition ${\delta }_0\approx 0\text{permil}$, and let $m$ be the mass of isotopically light carbon (${\delta }_{\text{add}}=-25\text{permil}$) released instantaneously by the Traps. The mixed reservoir's isotopic composition is the mass-weighted average:

$${\delta }_1=\frac{M_0{\delta }_0+m\cdot {\delta }_{\text{add}}}{M_0+m}.$$

Setting ${\delta }_1=-5\text{permil}$, ${\delta }_0=0\text{permil}$, ${\delta }_{\text{add}}=-25\text{permil}$ and solving:

$$-5(M_0+m)=-25m⟹-5M_0=-20m⟹m=M_0/4.$$

For $M_0\sim 40,000\mathrm{Gt}\mathrm{C}$ (modern scale, comparable to late-Permian estimates), the required release is $m\sim 10,000\mathrm{Gt}\mathrm{C}$, consistent with the lower-end coal-intrusion estimate. Larger releases (up to $100,000\mathrm{Gt}\mathrm{C}$) remain consistent if the pre-event reservoir or isotopic contrast differs. Basaltic degassing alone, at $\sim 1,000\mathrm{Gt}\mathrm{C}$, is insufficient by roughly an order of magnitude, which is why the coal-intrusion amplification is load-bearing. $\square$

Bridge. The Traps-extinction cascade builds toward 27.01.04 mantle plumes and LIPs, where the Siberian Traps appear as the canonical LIP-extinction pair, and appears again in 27.08.01 the geologic time scale, where the P-Tr boundary is defined by the extinction horizon itself. The foundational reason the P-Tr is the most severe of the Big Five is that the Traps intruded coal-bearing sediments, multiplying the carbon release; this is exactly the amplification that distinguishes the P-Tr from the smaller T-J (CAMP) and K-Pg (Deccan) events. The pattern generalises to every LIP-extinction pair in the Phanerozoic, and the bridge is the recognition that volcanically-driven mass extinctions are carbon-cycle events, not direct-kill events.

Exercises Intermediate+

Advanced results Master

Newell 1967: the P-Tr as a distinct extinction

Norman Newell's 1967 Geological Society of America Special Paper ^{[Newell 1967]} first recognized the P-Tr as a distinct and exceptionally severe mass extinction, separate from the background extinction rate. Newell compiled Permian and Triassic marine fossil ranges and demonstrated a sharp diversity drop at the boundary that exceeded the background rate by more than an order of magnitude. The Newell paper is the founding document of P-Tr extinction study, establishing the phenomenon as real and worth dedicated investigation.

Raup-Sepkoski 1982: the Big Five

Raup and Sepkoski's 1982 Science paper ^{[Raup-Sepkoski 1982]} analyzed the global marine animal fossil record and identified five stand-out extinction pulses exceeding a statistical threshold above the background rate. The P-Tr is the most severe at approximately 96 percent marine species loss, followed by the Ordovician-Silurian (85%), Triassic-Jurassic (80%), K-Pg (76%), and Late Devonian (75%). The Big Five framework remains the standard Phanerozoic extinction taxonomy and is the organizing structure for all comparative extinction research.

Renne-Basu 1991: dating the Siberian Traps

Renne and Basu's 1991 Science paper ^{[Renne-Basu 1991]} applied high-precision ⁴⁰Ar/³⁹Ar geochronology to Siberian Traps lavas, yielding an age of approximately 251 Ma, statistically indistinguishable from the then-accepted P-Tr boundary age. This was the first robust temporal link between the Traps and the extinction, transforming the Traps from a coincident geographic feature into the leading causal candidate. Subsequent work has refined the date but confirmed the temporal coincidence.

Bowring-Erwin-Isozaki 1998-1999: the rapid-extinction timing

Bowring and colleagues' 1998 Science paper ^{[Bowring 1998]} applied U/Pb zircon dating to volcanic ash beds interlayered with the fossil-bearing marine section at Meishan, China. The main extinction pulse was constrained to roughly 60,000 years, far shorter than earlier estimates of millions of years. The rapid timing ruled out gradualistic explanations (slow climate drift) and required a catastrophic trigger, consistent with the Traps' pulsed carbon release.

Reichow 2009: Traps volume

Reichow and colleagues' 2009 Geology paper ^{[Reichow 2009]} synthesized areal extents and stratigraphic thicknesses across the Siberian Traps outcrop and subsurface, yielding a total volume estimate of approximately 4 million km³ of lava. This makes the Siberian Traps the largest well-preserved continental large igneous province of the Phanerozoic. The volume estimate, combined with the intruded-coal amplification, provides the carbon budget sufficient to drive the observed isotopic and climatic excursions.

Wignall-Twitchett 1996: marine anoxia and euxinia

Wignall and Twitchett's 1996 Science paper ^{[Wignall-Twitchett 1996]} documented widespread marine anoxia (oxygen-depleted bottom waters) and euxinia (hydrogen-sulfide-rich waters) across the P-Tr boundary, recorded as "black shale" deposits in sections globally from South China to Greenland to Italy. The anoxia-euxinia framework explains the selective extinction pattern: benthic and deep-water taxa suffered the heaviest losses, while some surface-water and opportunistic species survived. Persistent Early-Triassic anoxia also explains the "sluggish" recovery documented by Payne and colleagues.

Burgess-Bowring-Shen 2014: synchronicity

Burgess, Bowring, and Shen's 2014 PNAS paper ^{[Burgess-Bowring-Shen 2014]} combined high-precision U/Pb dating of both the Siberian Traps and the Meishan extinction section, demonstrating that the Traps eruption began just before the extinction and that the main extinction was confined to a ~60 ± 48 kyr window. The temporal coincidence is the strongest evidence for the Traps-trigger hypothesis: the probability of two such events coinciding by chance is negligible, and no alternative trigger (asteroid, climate threshold) has matching timing.

Song 2013: the two-pulse structure

Song and colleagues' 2013 analysis ^{[Song 2013]} of the Meishan and Shangsi sections resolved the extinction into two pulses approximately 60 kyr apart, with the first eliminating approximately 57 percent of marine species and the second eliminating approximately 71 percent of the survivors. The two-pulse structure matches the pulsed character of Traps volcanism (discrete eruptive episodes separated by quiescent intervals) and rules out a single-impact cause.

Synthesis. The P-Tr framework builds toward a unified carbon-cycle catastrophe picture in which the Siberian Traps LIP, the coal-intrusion amplification, the warming/acidification/anoxia cascade, and the two-pulse extinction structure are layered, not alternative, descriptions of a single Earth-system event. The foundational reason the P-Tr is the most severe Phanerozoic extinction is that the Traps' lava volume (4 million km³) and coal-derived carbon release (10,000-100,000 Gt C) exceed every other LIP of the past 300 million years; this is exactly the content of the Burgess-Bowring-Shen 2014 synchronicity and the Song 2013 two-pulse structure. The central insight is that putting these together, the same volcanically-driven carbon-cycle physics explains the Paleozoic-Mesozoic transition, the elimination of the trilobites, and the Lystrosaurus-dominated post-extinction landscape. The pattern appears again in 27.01.04 the Deccan and CAMP LIPs at weaker amplitude, and generalises to the modern anthropogenic carbon release, where the same cascade mechanisms operate at a smaller magnitude; the bridge is the recognition that without the coal-intrusion amplification, the Traps alone would have produced a smaller extinction.

Full proof set Master

Proposition (P-Tr carbon-isotope mass balance). Let $M_0$ denote the pre-event ocean-atmosphere carbon reservoir with isotopic composition ${\delta }_0$, and let $m$ denote the mass of isotopically distinct carbon (composition ${\delta }_{\text{add}}$) released instantaneously. The post-release isotopic composition ${\delta }_1$ of the combined reservoir satisfies

$${\delta }_1=\frac{M_0{\delta }_0+m{\delta }_{\text{add}}}{M_0+m},$$

and the fraction $f=m/(M_0+m)$ of the post-release reservoir that is added carbon is

$$f=\frac{{\delta }_1-{\delta }_0}{{\delta }_{\text{add}}-{\delta }_0}.$$

For the observed P-Tr excursion (${\delta }_0=0\text{permil}$, ${\delta }_1=-5\text{permil}$, ${\delta }_{\text{add}}=-25\text{permil}$), $f=1/5$, corresponding to $m/M_0=1/4$.

Proof. The isotopic composition of a mixture is the mass-weighted average of the components, giving the first stated relation. Solving for $m$:

$${\delta }_1(M_0+m)=M_0{\delta }_0+m{\delta }_{\text{add}}⟹m({\delta }_{\text{add}}-{\delta }_1)=M_0({\delta }_1-{\delta }_0)⟹\frac{m}{M_0}=\frac{{\delta }_1-{\delta }_0}{{\delta }_{\text{add}}-{\delta }_1}.$$

Substituting ${\delta }_0=0$, ${\delta }_1=-5$, ${\delta }_{\text{add}}=-25$:

$$\frac{m}{M_0}=\frac{-5-0}{-25-(-5)}=\frac{-5}{-20}=\frac{1}{4}.$$

The added fraction of the post-release reservoir is $f=m/(M_0+m)=(M_0/4)/(5M_0/4)=1/5$. For $M_0\sim 40,000\mathrm{Gt}\mathrm{C}$, the required release is $m\sim 10,000\mathrm{Gt}\mathrm{C}$, consistent with the lower-end coal-intrusion estimate and with the isotopic excursion observed in marine carbonates globally. $\square$

Proposition (extinction-rate scaling). If a fraction $f$ of $N$ species goes extinct uniformly over a time interval $\Delta t$ under a constant per-unit-time extinction hazard $p$, then $p=1-(1-f{)}^{1/\Delta t}$ and the mean species survival time during the interval is $1/p$.

Proof. Under a constant hazard rate $p$ per unit time, the probability that a given species survives $k$ time units is $(1-p{)}^k$. Setting the survival probability over the full interval $\Delta t$ equal to the observed survival fraction $1-f$:

$$(1-p{)}^{\Delta t}=1-f⟹1-p=(1-f{)}^{1/\Delta t}⟹p=1-(1-f{)}^{1/\Delta t}.$$

For the P-Tr main pulse, $f=0.96$ and $\Delta t=60,000\mathrm{yr}$. With time units of millennia ($\Delta t=60$):

$$p=1-(0.04{)}^{1/60}\approx 1-e^{\ln (0.04)/60}\approx 1-e^{-0.0537}\approx \mathrm{0.052.}$$

The mean survival time during the pulse is $1/p\approx 19$ millennia. The background extinction rate is approximately $10^{-4}$ to $10^{-3}$ per species per million years, corresponding to $p_{\text{bg}}\sim 10^{-9}$ per millennium, so the P-Tr pulse elevated the per-species hazard by roughly five to six orders of magnitude above background. $\square$

Connections Master

• Earth history and the geologic time scale 27.08.01. The chapter survey introduces the Phanerozoic eon and its three eras. This unit supplies the load-bearing event at the Paleozoic-Mesozoic boundary: the P-Tr extinction defines the boundary itself, and the Burgess-Bowring-Shen 2014 high-precision dating anchors the boundary's absolute age at approximately 252 million years. Every Phanerozoic timescale published since 1998 uses the P-Tr extinction horizon as a primary age tie-point.

• Mantle plumes, hot spots, and large igneous provinces 27.01.04. The Siberian Traps are the canonical LIP-extinction pair: the largest well-preserved continental flood basalt of the Phanerozoic, coincident with the most severe mass extinction. This unit takes the LIP concept from the plumes unit and shows the downstream consequence — the carbon-cycle cascade that links a deep-mantle thermal event to a surface biosphere collapse. The Deccan Traps at the K-Pg and the CAMP at the T-J are the smaller-amplitude analogs.

• Fossil trends 19.08.03 pending. The P-Tr extinction is the largest-magnitude data point in the Phanerozoic diversity curve. The selectivity pattern (trilobites eliminated, ammonoids nearly lost, Lystrosaurus dominating the aftermath) exemplifies the macroevolutionary concepts of disaster taxa, post-extinction faunal turnover, and the evolutionary release of surviving clades. The 5-to-10-million-year recovery is the canonical "sluggish recovery" against which smaller extinction recoveries are measured.

• Stellar nucleosynthesis 28.02.05. The P-Tr geochemical proxies rest on isotopes whose nucleosynthetic origins are catalogued in the stellar-nucleosynthesis unit. The carbon-isotope excursion relies on the ${}^{12}\mathrm{C}{/}^{13}\mathrm{C}$ fractionation set by the CNO cycle and helium burning in stellar interiors; the sulfur-MIF ozone-depletion signal relies on the ${}^{33}\mathrm{S}$ and ${}^{36}\mathrm{S}$ produced in supernovae and explosive nucleosynthesis. Connecting the surface extinction record to its nucleosynthetic isotope inventory closes a loop between paleobiology and stellar physics.

Historical & philosophical context Master

Norman Newell, in a 1967 Geological Society of America Special Paper ^{[Newell 1967]}, first recognized the Permian-Triassic as a distinct and exceptionally severe mass extinction, separating it from the background extinction rate on the basis of compiled Permian and Triassic marine fossil ranges. The modern quantitative framework crystallized with Raup and Sepkoski in 1982 ^{[Raup-Sepkoski 1982]}, whose Science paper established the "Big Five" — the Ordovician-Silurian, Late Devonian, Permian-Triassic, Triassic-Jurassic, and Cretaceous-Paleogene — as the Phanerozoic's stand-out extinction pulses, with the P-Tr the most severe at approximately 96 percent marine species loss.

The volcanic-trigger hypothesis became testable with Renne and Basu's 1991 Science paper ^{[Renne-Basu 1991]}, which dated the Siberian Traps to approximately 251 Ma using high-precision ⁴⁰Ar/³⁹Ar geochronology, matching the P-Tr boundary within uncertainty. Bowring, Erwin, and colleagues in 1998 ^{[Bowring 1998]} tightened the extinction timing to roughly 60,000 years using U/Pb zircon geochronology at the Meishan stratotype section. Burgess, Bowring, and Shen in 2014 ^{[Burgess-Bowring-Shen 2014]} confirmed the extinction's synchrony with the Traps eruption, and Song and colleagues in 2013 ^{[Song 2013]} resolved the two-pulse structure. The modern synthesis integrates the Traps trigger, the coal-intrusion carbon amplification, the warming/acidification/anoxia cascade documented by Wignall and Twitchett in 1996 ^{[Wignall-Twitchett 1996]}, and the prolonged recovery documented by Payne and colleagues in 2004 ^{[Payne 2004]}.

Bibliography Master

@article{Newell1967,
  author = {Newell, Norman D.},
  title = {Revolutions in the history of life},
  journal = {Geological Society of America Special Paper},
  number = {89},
  pages = {63--91},
  year = {1967},
}

@article{RaupSepkoski1982,
  author = {Raup, David M. and Sepkoski, J. John},
  title = {Mass extinctions in the fossil record},
  journal = {Science},
  volume = {215},
  number = {4541},
  pages = {1501--1503},
  year = {1982},
}

@article{RenneBasu1991,
  author = {Renne, Paul R. and Basu, Asish R.},
  title = {Rapid eruption of the {Siberian} {Traps} flood basalts at the {Permo}-{Triassic} boundary},
  journal = {Science},
  volume = {253},
  number = {5016},
  pages = {176--179},
  year = {1991},
}

@article{Bowring1998,
  author = {Bowring, Samuel A. and Erwin, Douglas H. and Jin, Yugan G. and Martin, Mark W. and Davidek, Katherine and Wang, Wei},
  title = {{U}/{Pb} zircon geochronology of the end-{Permian} mass extinction},
  journal = {Science},
  volume = {280},
  number = {5370},
  pages = {1039--1045},
  year = {1998},
}

@article{BurgessBowringShen2014,
  author = {Burgess, Seth D. and Bowring, Samuel A. and Shen, Shu-zhong},
  title = {High-precision timeline for {Earth}'s worst extinction},
  journal = {Proceedings of the National Academy of Sciences},
  volume = {111},
  number = {9},
  pages = {3316--3321},
  year = {2014},
}

@article{Song2013,
  author = {Song, Haijun and Wignall, Paul B. and Tong, Jinnan and Yin, Hongfu},
  title = {Two pulses of extinction during the {Permian}-{Triassic} crisis},
  journal = {Nature Geoscience},
  volume = {6},
  pages = {52--56},
  year = {2013},
}

@article{Reichow2009,
  author = {Reichow, Marc K. and Pringle, Malcolm S. and Al'Mukhamedov, Andrey I. and Allen, Mark B. and Andreichev, Valery L. and Buslov, Mikhail M. and Davies, Christopher E. and Fedoseev, Gennady S. and Fitton, J. Godfrey and Inger, Simon and Medvedev, Aleksandr Ya. and Mitchell, Claire and Puchkov, Valery N. and Safonova, Inna Yu. and Scott, Robert A. and Saunders, Andrew D.},
  title = {The timing and extent of the eruption of the {Siberian} {Traps}, large igneous province: implications for the end-{Permian} environmental crisis},
  journal = {Geology},
  volume = {37},
  number = {9},
  pages = {819--822},
  year = {2009},
}

@article{WignallTwitchett1996,
  author = {Wignall, Paul B. and Twitchett, Richard J.},
  title = {Oceanic anoxia and the end-{Permian} mass extinction},
  journal = {Science},
  volume = {272},
  number = {5265},
  pages = {1155--1158},
  year = {1996},
}

@article{Payne2004,
  author = {Payne, Jonathan L. and Lehrmann, Daniel J. and Wei, Jiayong and Orchard, Michael J. and Schrag, Daniel P. and Knoll, Andrew H.},
  title = {Large perturbations of the carbon cycle during recovery from the end-{Permian} extinction},
  journal = {Science},
  volume = {305},
  number = {5683},
  pages = {506--509},
  year = {2004},
}

@book{Erwin2006,
  author = {Erwin, Douglas H.},
  title = {Extinction: How Life on {Earth} Nearly Ended 250 Million Years Ago},
  publisher = {Princeton University Press},
  year = {2006},
}

@book{Benton2005,
  author = {Benton, Michael J.},
  title = {When Life Nearly Died: The Greatest Mass Extinction of All Time},
  publisher = {Thames and Hudson},
  year = {2005},
}