Pattern of Marine Mass Extinction Near the Permian-Triassic Boundary in South China

See allHide authors and affiliations

Science  21 Jul 2000:
Vol. 289, Issue 5478, pp. 432-436
DOI: 10.1126/science.289.5478.432


The Meishan section across the Permian-Triassic boundary in South China is the most thoroughly investigated in the world. A statistical analysis of the occurrences of 162 genera and 333 species confirms a sudden extinction event at 251.4 million years ago, coincident with a dramatic depletion of δ13Ccarbonate and an increase in microspherules.

The end-Permian mass extinction eliminated over 90% of all marine species and had a significant impact on land species as well (1, 2). However, geochronologic results from South China reveal that the main extinction occurred over a period of less than 500,000 years (3), coincident with the eruption of the Siberian flood basalts (4, 5) and with a sharp shift in δ13Ccarb (6). Although there are claims for multiple pulses of extinction, including at least three at the classic Meishan sections in South China (7,8) [probably the most thoroughly studied Permian-Triassic (P-T) marine boundary section in the world], the cause of the extinction remains enigmatic. Here we examine sampling and preservation effects (9) using a statistical analysis of species' stratigraphic ranges (10,11) to demonstrate the extreme rapidity of the extinction.

We studied fossils systematically collected from five sections at the Meishan locality (12), at an average sample spacing of 30 to 50 cm. Smectite-rich clay beds occur frequently through the Changhsing and Yinkeng formations and can be traced across all outcrop areas. With the clay beds used as reference, a standard composite section was established for all fossil occurrences. We recorded a total of 333 species in 162 genera of 15 marine fossil groups, including foraminifera, fusulinids, radiolarians, rugosan corals, bryozoans, brachiopods, bivalves, cephalopods, gastropods, ostracods, trilobites, conodonts, fish, calcareous algae, and others from 64 horizons; the collection intensity was uniform, without any sampling gaps (Fig. 1).

Figure 1

Stratigraphic ranges of fossil species (indicated by vertical gray lines) from the latest Permian to the Early Triassic in the Meishan sections projected onto the composite section. Species numbers are shown on thex axes. (A) Fossil range scaled to rock thickness, with an abrupt faunal change near the base of bed 25. (B) Fossil range scaled to time. Faunal change appears gradual except around 251.4 Ma. The positions of volcanic ash beds and isotopic ages are from (3). The δ13Ccarb profiles integrate all available data from the Meishan sections (8,20, 21). Three previously proposed extinction levels are shown (indicated by A, B, and C) (24).

About 161 species became extinct below the P-T boundary beds (beds 24 to 27). The extinction rate (extinct species divided by total species at the same level) does not exceed 33% in any bed of the Changhsing Formation below bed 24. The remaining species mostly disappeared within a short interval around the P-T boundary (beds 25 and 26), including the base of bed 25, where the extinction rate is 94%. We used confidence intervals on the end points of stratigraphic ranges (10, 14) and simulations for abrupt, gradual, and stepwise extinction scenarios (15) to evaluate the empirical pattern.

Confidence intervals were calculated for the stratigraphic ranges, measured in rock thickness, of the 93 genera (265 species; 80% of the total species) with multiple occurrences. The range end points of 95% confidence intervals for more than 95% of the genera extend into bed 34 of the Yinkeng Formation. Calculated depositional rates between the radiometrically dated ash beds (3) are 0.03 cm per 1000 years for the transitional beds versus 0.4 cm per 1000 years for the upper part of the Changhsing Formation and 1.3 cm per 1000 years for the basal part of the Yinkeng Formation. The idea that deposition occurred by a slow sedimentation process is supported by extensively burrowed hardgrounds within bed 27 (16). Thus the forward smearing of the confidence intervals was likely produced by the condensed transitional sequence from beds 24e to 27.

To assess the true timing of the events, we assumed a linear deposition rate in the intervals bracketed by the dated ash beds, which accounts for these variations in sedimentation rates. We calculated the ages of each fossil occurrence, the number of fossil horizons, the time interval from the last observed occurrence to the postulated extinction horizon (α), and confidence levels (ρ) associated with each α value for all genera (10, 14). A Kolmogorov-Smirnov test does not reject the null hypothesis: The confidence levels are independently uniformly distributed between ρ = 0 and ρ = 1. The distribution of last occurrences suggests an extinction peak between 251.2 and 251.4 million years ago (Ma). The 50% confidence intervals on the stratigraphic ranges of all 93 genera are consistent with a sudden extinction around 251.4 Ma (13) (Fig. 2A), with a predicted true extinction level near 251.3 Ma [94% of genera are included in a 0.1-million-year (My) interval spacing]. A more reasonable conclusion, equally consistent with the statistical analysis, is a sudden extinction at 251.4 Ma, followed by the gradual disappearance of a small number of surviving genera over the next 1 million years.

Figure 2

(A) Fifty percent confidence intervals [the interval between red dots (confidence end points) and last occurrences (upper black dots)] for 93 genera of 10 different groups (foraminifera, fusulinids, bryozoans, brachiopods, rugosan coral, bivalves, cephalopods, ostracods, conodonts, and calcareous algae) with multiple occurrences at the composite P-T section at Meishan. Fifty percent confidence intervals for the foraminifer (B), ostracod (C), cephalopod (D), and brachiopod (E) genera are shown separately. Time (in Ma) is shown on the y axes; numbers on the x axes indicate species number. Contours indicating the predicted extinction level are shown for each group (blue bars); the probability that the extinction lies within each interval is given below the intervals. The probability was defined by iteratively eliminating genera that fall beyond the extinction level at the 99.99% level. The apparent brachiopod extinction at 250.6 Ma may reflect the scarcity of fossils below the boundary beds; the occurrences of ammonoids, conodonts, bivalves, and other groups are also spotty. Ammonoids exhibit stepped declines at 253.0 and 251.4 Ma.

We repeated the analysis for the 38 genera that cross the P-T boundary to test for a second extinction peak. The predicted extinction is at 250.6 Ma, with a confidence of 95.1% for a 0.5-My interval spacing, declining to 55% for a 0.1-My spacing; the broad spacing of final occurrences does not support a sudden extinction after the P-T boundary. A similar analysis for genera that disappeared before 251.4 Ma provides no support for an earlier extinction step at bed 24e (17).

Several groups were analyzed individually. Fifty percent confidence intervals for 15 of 22 foraminifer genera and 13 of 21 ostracod genera indicate a sudden extinction near 251.4 Ma. Fifty percent confidence intervals for conodont genera (platform elements) and bivalves show a gradual change. The predicted position of the true extinction horizons for the foraminifera, ostracods, and cephalopods is around 251.4 Ma (Fig. 2, B through D). In contrast, 50% confidence intervals for 6 of 13 brachiopod genera display a sudden extinction at 250.6 Ma (Fig. 2E). Despite this apparent latter extinction of brachiopods, they are not well preserved in this slope facies. A P-T boundary section 27 km east of Meishan in a shallow shelf facies contains 18 brachiopod genera between the sequence boundary and the boundary clay. As with the foraminifera, ostracods, and cephalopods at Meishan, the brachiopods suffer dramatic decline at the boundary clay and a secondary extinction at the level corresponding to 250.6 Ma.

Simulations of sudden extinctions produce an accelerating decline in the number of taxa, whereas gradual extinctions display a constant decline, and stepwise extinctions display a stepped decline (15). We plotted the age of last occurrences versus stratigraphic abundance (percentage of the time intervals during which a genus occurs divided by the total number of occurrences) for all 162 genera as well as for each group. The results for all 162 genera show a sudden extinction near 251.4 Ma, followed by a gradual decline from 251.4 to 250.6 Ma. The results for foraminifera and ostracods display the hollow shape typical of a sudden extinction (Fig. 3A). Individual exceptions include the ostracod genera Acratia (30% abundance) and the foraminiferPseudonodosaria (20%), which extend beyond the P-T boundary elsewhere in South China (18). The number of last occurrences by time has a single mode (Fig. 3B). The fossil occurrences at Meishan are best explained by a major extinction around 251.4 Ma. No support is found for previously proposed extinction steps related to the end of the Paleozoic reef system correlative with bed 24e or the extinction of relic Paleozoic brachiopods at bed 28. However, an increase in fungal spores, a possible indicator of massive disturbance in the terrestrial ecosystem, begins below bed 24 (19).

Figure 3

Results for simulations of 162 genera from the Meishan sections. (A) Frequency distribution of last occurrences for 162 genera and for foraminifer, ostracod, cephalopod, and brachiopod genera, respectively. (B) Histograms plotted by the total number of time intervals at which a genus occurs versus the age of last occurrence. Only genera of less than 15% stratigraphic abundance tend to have last occurrences well below the major extinction time of 251.4 Ma; genera of exception are indicated by names in (A). The genera that extend upward beyond 250 Ma are not plotted in the figure. Each dot may represent more than one genus with the same coordinate position.

Significant negative δ13C anomalies have been reported from the extinction interval at Meishan (6, 20). These include a drop in δ13C value from +2 to −4 per mil (‰) within bed 26, and to −6‰ at the basal levels of bed 27 (20). New δ13Ccarb data from section B (21) confirm a depletion of δ13C in beds 25 and 26 but not the reported depletion in bed 27 (Fig. 4). The accompanying δ18O values, ranging from −8 to −11‰ (21), indicate that the previously reported unusually low δ13C values in basal bed 27 might reflect strong weathering. The carbon isotope shift from the top of bed 24e to the lower part of bed 26 is consistent with a major extinction event around 251.4 Ma and the addition of light carbon.

Figure 4

Carbon isotope profile of P-T boundary interval at section B of Meishan. Blue triangles represent replicate results of the same samples, provided by S. D'Hondt (28).

Unlike the illite ash clay beds above and below the P-T boundary, the boundary clay beds are dominated by interstriatified montmorillonite-illite clay (22). P-T boundary levels in South China are accompanied by concentrations of microspherules that are 102 to 103 times those of other horizons (8, 23); many of these appear to be volcanic, although the mechanism remains uncertain (7). The peak microspherule abundance is not necessarily associated with an ash-clay layer. In the Shangsi section in Sichuan Province, the microspherule concentration increases rapidly from the white ash-clay layer coincident with the mass extinction (bed 27b) to a peak in the black shale of bed 27c (23).

In the Meishan sections, three successive transgression surfaces are closely associated with hypothesized biotic downturns (24). Evidence for anoxia at Meishan includes pyritic laminae beneath bed 25 and a reduction in trace fossils in P-T boundary beds (13). Other, more reliable evidence of bottom oxygenation does not indicate drastic anoxia in association with the severe extinction in bed 25. Framboidal pyrites are rare, and total organic carbon (TOC) is relatively low in beds 26 and 27 (25), although well-developed framboidal pyrites and high percentages of TOC appear in organic-rich levels of beds 24 and 29. Bed 26 contains the trace fossil Planolites and bed 27 was thoroughly burrowed by Thalassinoides(16). The scenario of transgression with anoxia appears unable to explain the severe extinction in bed 25.

Both pyroclastic and flood basalt volcanism have been invoked as causes for the extinction (4, 7). The temporal overlap between the Siberian Trap volcanism (251.2 ± 0.3 Ma) and the P-T boundary (<251.5 and >250.5 Ma) suggests a causal relation (3, 5, 26). The Siberian flood basalt may have released large amounts of CO2, and possibly sulfates, triggering a brief volcanic winter, followed by a period of global warming (5,27). The frequent acidic volcanism from the latest Permian to the earliest Triassic in South China has been invoked as a causal factor (7, 26), but the lack of extinction at most ash beds augurs against any simple relation (1). The rapid marine extinction in bed 25 is consistent with the dramatic shift of carbon isotopic data and coincides with the microspherule anomaly. Despite the lack of compelling evidence for extraterrestrial impact, the rapidity of the extinction and the associated environmental changes are also consistent with the involvement of a bolide impact in this most severe biotic crisis in the history of life.

  • * To whom correspondence should be addressed. E-mail: ygjin{at}


View Abstract

Navigate This Article