The end-Permian mass extinction: a still unexplained catastrophe

The end-Permian mass extinction is widely regarded as the largest mass extinction in the past 542 million years with loss of about 95% of marine species and 75% of terrestrial species. There has been much focus and speculation on what could have caused such a catastrophe. Despite decades of study the cause or causes remain mysterious. Numerous scenarios have been proposed, including asteroid impact, Siberian flood basalt volcanism, marine anoxia and euxinia, sea-level change, thermogenic methane release and biogenic methane release due to explosive growth of a methanogenic microbe. It is now clear that a number of major environmental perturbations are approximately coincident with the end-Permian mass extinction. These include global negative excursions of both  13 Ccarb and  13 Corg near the extinction interval (see a review by [1] and a recent study by [2]); distinctive calcium isotope excursions [3]; a sudden expansion of microbialites [4]; a rapid temperature rise of ~8 o C in the extinction interval [5] followed by a long “hothouse” period in the Early Triassic [6], large regression followed by rapid transgression [7], evidence for wildfires and cyanobacteria blooms [8] etc. There remains disagreement over the nature, timing, and duration of the environmental perturbations and how they relate to detailed patterns of extinction, resolution of which are critical for understanding the causative mechanism(s).

The end-Permian mass extinction: a still unexplained catastrophe Shu-zhong Shen 1, * and Samuel A. Bowring 2 The end-Permian mass extinction is widely regarded as the largest mass extinction in the past 542 million years with loss of about 95% of marine species and 75% of terrestrial species. There has been much focus and speculation on what could have caused such a catastrophe. Despite decades of study, the cause or causes remain mysterious. Numerous scenarios have been proposed, including asteroid impact, Siberian flood basalt volcanism, marine anoxia and euxinia, sea-level change, thermogenic methane release and biogenic methane release due to explosive growth of a methanogenic microbe.
It is now clear that a number of major environmental perturbations are approximately coincident with the end-Permian mass extinction. These include global negative excursions of both δ 13 C carb and δ 13 C org near the extinction interval (see a review by Korte and Kozur [1] and a recent study by Shen et al. [2]); distinctive calcium isotope excursions [3]; a sudden expansion of microbialites [4]; a rapid temperature rise of ∼8 • C in the extinction interval [5] followed by a long 'hothouse' period in the Early Triassic [6], large regression followed by rapid transgression [7], evidence for wildfires and cyanobacteria blooms [8], etc. There remains disagreement over the nature, timing and duration of the environmental perturbations and how they relate to detailed patterns of extinction, resolution of which is critical for understanding the causative mechanism(s).

TIMING AND DURATION
The timing and duration of the end-Permian extinction at the Meishan sections in South China has been studied for over two decades. In many ways, successive publications track the evolution of high-precision U-Pb geochronological techniques that have led to increasingly precise and accurate constraints on the extinction.
Burgess et al. [9] review the evolution of increasing precise and accurate geochronological constraints on the age and duration of the extinction. For example, the published ages of Bed 25 at Meishan have varied from 251.4 ± 0.3 to >254 Ma and the duration of the extinction from 500 to 61 ± 48 kyr. The latter reflects the latest work using EARTHTIME protocols [9]. In addition to U-Pb zircon geochronology of ash beds, the floating astronomical time scale (ATS) has been applied to the extinction interval at Meishan and Shangsi. The ATS is based on recognizing astronomically forced stratigraphy and the latest effort has yielded an extinction duration of 112 kyr at Meishan and 83 kyr at Shangsi in South China [10]. Thus, both the most recent high-precision CA-TIMS U-Pb dates and the ATS are consistent with a catastrophic event that occurred in less than 100 kyr. Higher temporal resolution estimates from Shangsi and Meishan will likely be limited by the condensed nature of the sections.

EXTINCTION PATTERN
In addition to timing of the extinction, another fundamental issue is how to reconstruct and understand patterns of biological diversity. A single catastrophic event between Bed 25 and 28 was proposed based on detailed paleontological study of the Meishan sections using a confidence interval approach [11]. More recently, a new analysis of the diversity patterns based on a much larger database, including tens of sections in South China and the peri-Gondwanan region, using the constrained optimization (CONOP) approach also supports a single event [12,13]. In contrast, Song et al. [14, fig. 1] proposed a two-phase extinction at Meishan, one at Bed 25 and the other at Bed 28 and suggest that these two pulses can be recognized in other sections. Moreover, many previous studies reported multiple pulses of extinction based on different fossil groups from single horizons and sections (see those summarized in [13], fig. 8). These two different views of the extinction patterns likely reflect different inventories of fossils in collections from different sections and different analytical/statistical approaches.
Although the fossil record can be a spectacular archive of patterns of biodiversity providing a vital deep-time perspective on extinction patterns, the record is commonly biased by the incompleteness of fossil preservation, facies changes and time averaging of fossil collections. For example, the observed local end of a fossil taxon in a section is almost always a truncated false last occurrence of the true extinction. In assessing extinction, it is imperative to not just look at the total number of species lost within a certain interval or on a bedding plane, but more importantly to elucidate the true pattern after eliminating sampling biases caused by facies change, incorrect or imprecise biostratigraphical correlation, incomplete preservation and collecting intensity with different analytical approaches across a wide geographic area.
At present, many agree that the main pulse of end-Permian extinction occurred at Bed 25 at the Meishan section, but it is controversial whether the interval between Bed 25 and 28 that contains some relict Permian taxa as well as the occurrences of rare Triassic taxa (e.g., Hypophiceras, Claraia, Eumorphotis) represents continued extinction during severe environmental stress [12,13] or a new ameliorated stage followed by another extinction pulse at Bed 28 [14]. Strong evidence for euxinic condition with cyanobacterial blooms [15], oceanic acidification [3], microbialite expansion [4] and the distinct Lilliput effect [16] indicates that the entire extinction interval was characterized by extreme environmental stress (Fig. 1). Accurate patterns of biodiversity throughout the extinction interval must come from multiple PERSPECTIVES sections and especially ones that reflect higher sediment accumulation rates than are preserved at Meishan and Shangsi.

CONTEXT OF THE EXTINCTION
A deeper understanding of the end-Permian extinction must rely on integration of palaeontological, geochronological, geological and geochemical data from multiple sections and palaeoenvironments so that the magnitudes and patterns of extinction can be correlated within the context of multiple proxies from before, during and after the event (Fig. 1). The onset of the globally recognized δ 13 C carb excursion began about 60 kyr below Bed 25 at the Meishan section, and the abrupt decline of δ 13 C carb marking the onset of the extinction is estimated to have lasted less than 10 kyr [9]. A protracted but gradual negative downturn of δ 13 C carb refined by Burgess et al. [9], beginning at Bed 30 of the Yinkeng Formation in the lower Triassic, was proposed to correspond to a second phase of the extinction based on carbon isotope and cyanobacterial bloom of Xie et al. [15]. Burgess et al. [9] show that this interval is about 50 kyr younger than the end of the main extinction interval ( Fig. 1) and lasted nearly half a million years until Bed 39 at Meishan. The double negative shifts of δ 13 C org of Cao et al. [17] are before the onset (between Bed 23 and 24) and within the extinction interval (Bed 26) at the Meishan section, but only one negative shift of δ 13 C org has been recorded in southwest China [12]. The first evidence of a cyanobacterial bloom at Meishan is in Bed 26 within the extinction interval, but the second interval of the cyanobacterial bloom is in the Triassic, ca 300 kyr after the end of extinction (Fig. 1). While the rapid rise in sea surface temperature reported by Joachimski et al. [5] post-dates the negative spike in δ 13 C carb and is approximately coincident with the onset of the extinction at Meishan section, but a more precise relationship with the extinction requires additional study in multiple sections, especially those with higher sediment accumulation rates. For example, if the temperature increase lagged the on-set of the extinction, high temperatures could not be invoked as a major killing mechanism.
Previous studies have noted the approximate coincidence between the timing of the extinction and the eruption of the Siberian flood basalt lava flow province, but given the very short duration of the extinction, testing a precise relationship is not yet possible. Many mafic rocks in the region are younger than the end-Permian mass extinction based on dates from the Meishan section in South China. However, the available, relatively imprecise, geochronological data from the Siberian Traps cannot be used to determine whether the eruption predates and/or overlaps the mass extinction interval.
Given the apparent brevity of the extinction interval, two issues must yet be clarified before a definitive link can be made. (1) What is the temporal relationship between eruption of the Siberian Traps and the mass extinction? (2) What is the source of isotopically light carbon associated with the extinction interval? Did the eruption begin before the extinction? And if the eruption of the Siberian Traps lasted ca 1-2 Myr, much longer than the mass extinction, is the extinction related to one pulse of Siberian magmatism or the cumulative effects of the eruption up to the time of the extinction? The last question is crucial as it is possible that Late Permian ecosystems were undergoing stress well before the sudden collapse as evidenced by the progressive negative shift before the abrupt decline of δ 13 C carb [2,8,12,17], marine anoxic and euxinic conditions before the extinction [17] and changes of rates of taxon turnover and diversification before the sudden extinction [13]. In this scenario, once a critical threshold is reached, total collapse of marine and terrestrial ecosystems result.
It has long been argued based on Triassic palaeoclimate, evidence for a calcification crisis at the extinction, palaeophysiology of the ecosystem collapse and a sudden perturbation of the carbon cycle with addition of isotopically light carbon, that elevated atmospheric CO 2 played a major role in the extinction. Although volcanism undoubtedly contributed to the elevated CO 2 con-tents in the atmosphere, it is doubtful if volcanism alone could have released enough magmatic CO 2 to cause the higher temperatures. Debate continues regarding the source of the isotopically light carbon. It has been suggested that Siberian large igneous province (LIP) magmatism triggered the massive release of greenhouse gases from thick organicrich deposits or rapid venting of coalderived methane or massive combustion of coal [18]. Another recent hypothesis is that the dramatic rise of greenhouse gas was produced by rapid expansion of a methanogenic microbe fuelled by addition of nickel to the oceans from the eruption [19].
There is still, much unknown about the exact cause(s) of the extinction, but the timing constraints leave little doubt that it was a catastrophic event.
Although the Meishan section in South China contains two Global Stratotype Section and Points (GSSPs) that have been intensively studied, one obvious drawback is that the section is highly condensed. The extinction interval is 26 cm thick with distinct hiatuses and facies changes making accurate calculation of sediment accumulation rates and internal data difficult, even with closely spaced ash beds. Much more work is needed in other more expanded sections.

TRIASSIC RECOVERY
Marine ecosystems underwent a revolutionary reorganization from Late Permian benthic shelly-dominated communities in shallow marine carbonate environments to widespread microbial communities with low-oxygen conditions across the Permian-Triassic boundary [20]. However, a recent analysis challenges the severity of the ecological effect of the end-Permian extinction. Statistical analysis of benthic marine invertebrate fossils indicates that there is no significant loss of functional diversity at the global scale [21], although fossil diversity was massively reduced during the extinction. A key question is how rapidly biota and ecosystems recovered and how evolutionary processes led to occupation of vacated ecospaces in the Triassic. It has been traditionally postulated that the Early Triassic recovery took at least 5-6 Myr or until the early Middle Triassic (Anisian). However, new and abundant palaeontological data suggest that different fossil groups had different recovery patterns. Triassic disaster taxa such as Claraia began to occur in the latest Permian before the onset of the extinction in Guangxi, South China, whereas conodonts experienced turnover at species and genus levels only. Triassic ammonoids first appear in the extinction interval, but rapidly diversify after the extinction, and bivalves, foraminifers also recovered rapidly. Fishes and marine reptiles show apparently explosive diversification very early in the Early Triassic. Brachiopods and ostracods took a longer time to recover until late Early Triassic to early Middle Triassic.
There are also wider debates about the dynamics of the Triassic recovery. The recovery was associated with a number of repeated large carbon isotope excursions [22], but their age and causal-effect relationship remains unclear. Apparently, the large perturbations in carbon isotope signatures are not correspondingly reflected in biodiversity change during Early and Middle Triassic. Triassic carbon isotope profiles indicate a rapid recovery before the Dienerian. In contrast, the whole Early Triassic is characterized by high sea surface temperatures with little fluctuation in temperature [6]. In addition to global warming, elevated CO 2 also likely led to ocean acidification, increased erosion rates and raised inputs of nutrients into the oceans that led to widespread anoxia. If high temperatures and anoxic conditions led to a prolonged recovery, it is still unclear how to explain the latitudinal and palaeogeographic control on the timing and patterns of recovery. Similar to the extinction, the recovery must be also examined using a number of temporally calibrated sections from a broad range of paleogeographic and paleoenvironemental settings with high-quality palaeontogical data. Again, fossil incompleteness and sampling biases must be considered carefully. A new effort will most likely lead to insights into the recovery as well as the extinction.