Large Perturbations of the Carbon Cycle During Recovery from the End-Permian Extinction

See allHide authors and affiliations

Science  23 Jul 2004:
Vol. 305, Issue 5683, pp. 506-509
DOI: 10.1126/science.1097023


High-resolution carbon isotope measurements of multiple stratigraphic sections in south China demonstrate that the pronounced carbon isotopic excursion at the Permian-Triassic boundary was not an isolated event but the first in a series of large fluctuations that continued throughout the Early Triassic before ending abruptly early in the Middle Triassic. The unusual behavior of the carbon cycle coincides with the delayed recovery from end-Permian extinction recorded by fossils, suggesting a direct relationship between Earth system function and biological rediversification in the aftermath of Earth's most devastating mass extinction.

The most severe extinction since the advent of animal life on Earth occurred at the end of the Permian Period, 251 million years ago (Ma) (13), with global loss of marine species estimated near 90% (4, 5). Organisms with heavy calcification and limited elaboration of circulatory and respiratory systems were most severely affected, whereas those with more active control of circulation, elaborated structures for gas exchange, and lightly calcified or uncalcified skeletons survived in much higher proportions (6). The ensuing Early Triassic was an interval of delayed biotic recovery characterized by continued low diversity (5); the persistence of a cosmopolitan fauna in the oceans (7); the absence of metazoan reefs (8), calcareous algae (9), calcareous sponges (10), and corals (11); the apparent absence of marine taxa recorded in both Permian and Middle Triassic rocks (12); a reduction in the size of invertebrates (13); and, on land, a hiatus in coal deposition (14). Sustained recovery of marine diversity and ecology began primarily in the early part of the Middle Triassic, some 4 to 8 million years after the extinction itself. The apparent delay of biological renewal could reflect the time scale necessary for reintegration of ecosystems (5, 12) or poor Early Triassic fossil preservation (12), but it has also been widely interpreted as a consequence of persistently unfavorable environmental conditions through part or all of the Early Triassic (5, 15, 16).

Aside from the extinction itself, the strongest evidence for environmental disturbance at the Permian-Triassic (P-Tr) boundary is a sharp negative excursion of 2 to 4 per mil (‰) in the carbon isotopic composition (δ13C) of marine carbonate (δ13Ccarb) (17, 18) and one or more similar excursions in the δ13C of organic matter (δ13Corg) (1921). Although the apparent stabilization of δ13Ccarb above the boundary interval (17) has been used to suggest a stable but reduced fraction of organic carbon burial in the Early Triassic (22), the carbon isotopic record of Early and Middle Triassic rocks has received relatively little study. Positive carbon isotopic excursions have been noted at the Dienerian-Smithian (23), Smithian-Spathian (24), and Spathian-Anisian boundaries (25), but, to date, the only complete Early and Middle Triassic carbon isotopic data set previously compiled from a single stratigraphic section is in an unpublished dissertation (23).

We sampled the Great Bank of Guizhou (GBG), an isolated Late Permian to Late Triassic carbonate platform in the Nanpanjiang Basin of Guizhou Province, southern China (Fig. 1A), to obtain high-resolution profiles of δ13Ccarb from the Late Permian through the Middle Triassic. The exposure of sections in both platform interior and basin margin settings further allows us to compare data across a range of depositional environments (Fig. 1B). A detailed stratigraphic framework for the platform has been developed from sequence stratigraphic (26), biostratigraphic, and geochronologic studies (27).

Fig. 1.

(A) Early Triassic paleogeographic map, modified from (26). Inset: Hash pattern indicates the Nanpanjiang Basin and brick pattern indicates the Yangtze. (B) Schematic cross section of the GBG, illustrating the locations of stratigraphic sections within the platform architecture.

Our results show that the P-Tr boundary carbon isotope excursion was not an isolated event. Rather, it was the first in a series of (mostly larger) excursions that continued throughout the Early Triassic and into the early part of the Middle Triassic Period (Fig. 2). The excursions ended early in the Anisian (Bithynian) and were followed by an extended interval of stable values around 2‰. The interval of carbon-isotopic stability continued through the remainder of the Middle Triassic and into the Carnian (Fig. 3), demonstrating that the large fluctuations in δ13Ccarb are confined to Early Triassic strata.

Fig. 2.

Carbon isotopic data from all sections and conodont ranges from Guandao section. The time scale is based on conodont and foraminiferal biostratigraphy from the Guandao section. Correlations from the basin margin to the platform interior are constrained by the occurrence patterns of Late Permian foraminifera (e.g., Palaeofusulina and Colaniella) and Griesbachian conodonts (Hindeodus parvus and Isarcicella isarcica) in the P-Tr boundary interval of the platform interior and by the occurrence of the Smithian conodont Parachirognathus (indicated by an asterisk) in the uppermost cyclic interval of Dawen and Dajiang sections. Gd., Gladiogondolella; Cs., Chiosella; Ns., Neospathodus; Ns. ex. gr., Neospathodus ex grupo; Ng., Neogondolella; Bith., Bithynian; Aeg., Aegean.

Fig. 3.

Composite carbon isotopic curve for the Changxingian-Carnian (Cordevolian) compared to the pattern of biotic recovery from the end-Permian extinction. Lithostratigraphy and conodont ranges are shown for the Upper Guandao section. The conodont ranges are used to constrain the Middle and Late Triassic time scale. Radiometric dates are from (1, 27, 44), dasyclad algal diversity data from (9), coral and sponge diversity data from (45), and global diversity data from (46). Gastropod data are shown in table S1. Carbon isotope data: red, Guandao; blue, Dajiang; gold, Dawen; green, Upper Guandao. Fass., Fassanian; Long., Longobardian; Ni., Nicoraella; Pg., Paragondolella; Bv., Budurovignathus; M., Metapolygnathus; gen., number of genera; sp., number of species.

Absolute ages of 251.4 ± 0.3 Ma (1) or ∼253 Ma (2) for the P-Tr boundary and ∼247 Ma for the Early-Middle Triassic boundary (27) constrain the entire Early Triassic to ∼4.5 to 6 Ma and individual isotopic excursions to less than 1 million years in each case. The drops in δ13Ccarb at the P-Tr boundary and at the base of the overlying cyclic interval in the platform interior (Fig. 2) occur within 10 m of section, suggesting a very short time scale for these events, whereas the positive and negative excursions observed in the upper part of the platform interior and on the basin margin occur over longer stratigraphic intervals (>50 m), implying that they developed more slowly, although still in less than 1 million years. The symmetry of the positive and negative excursions indicates that the time scales of increases and decreases in δ13Ccarb were similar.

Our data (28) reproduce both the gradual carbon-isotopic drop in Late Permian oceans and the rapid negative excursion at the P-Tr boundary (17). The negative excursion at the base of the platform interior cyclic interval and the subsequent positive excursion are hinted at by basal Triassic δ13Corg profiles for nonmarine successions (19), and they appear to correlate with data from the Italian Dolomites reported in an abstract (29). Our data also corroborate previous reports of positive excursions at the Smithian-Spathian (24) and Spathian-Anisian (25) boundaries, as well as the general shape of composite isotopic profiles developed from sections in Anhui Province, China (30), and multiple profiles reported from across the Tethys (23). The consistency of our findings with the more limited results from other areas, the sharp contrast between the Early and Middle Triassic records on the GBG, and the consistency of the platform interior and basin margin records indicate that these data reflect global instability of the Early Triassic carbon cycle rather than local, diagenetic, or facies-specific effects. Confirmation of the global nature of the signal will depend upon additional high-resolution studies at other localities.

What physical mechanisms could have generated positive and negative isotopic shifts in δ13Ccarb of up to 8‰ in marine carbonates over time scales of tens to hundreds of thousands of years? Although the rapidity of the extinction (31) and isotopic excursion at the P-Tr boundary is compatible with emerging evidence of bolide impact (32, 33), the longer time scale of the subsequent excursions suggests another cause. Eruption of the Siberian Traps provides another possible explanation for prolonged instability in the carbon cycle. Available radiometric dates restrict Siberian trap basalts to the boundary interval itself (34), but the limited geographic scope of dated sections relative to the known area of the volcanic flows and the presence of >1000 m of flows above the youngest dated horizon (34) leave open the possibility that episodic eruption occurred through much of the Early Triassic. More age data are needed to test this scenario.

Massive methane release from sea-floor gas hydrate reservoirs has also been suggested as an explanation for the P-Tr boundary excursion (5, 19, 21, 35). Unlike the Late Paleocene event for which gas hydrate release was first proposed (36), however, the more gradual and roughly symmetric increases and decreases in δ13Ccarb during the later part of the Early Triassic would require extended, alternating intervals of methane storage and release. The long time scale [>100 thousand years (ky), assuming constant sedimentation rates] of the negative shifts is difficult to account for under the scenario of methane release, because as the time scale of the isotopic shift increases, so too does the amount of methane needed to produce the same excursion. The ∼8‰ drop from the late Dienerian to Smithian (Fig. 2) over 100 to 500 ky would require the release of much more than 10,000 Gt of methane (1 Gt = 107 kg) (37), more than five times the amount suggested to account for the Late Paleocene thermal maximum (36). Furthermore, the dependence of methane production on organic carbon burial precludes rapid (i.e., <1 million years) replenishment of the methane hydrate reservoir in the absence of extraordinarily high rates of organic carbon burial (38). Methane release is an attractive hypothesis for the P-Tr carbon-isotopic event viewed in isolation, but the full Early Triassic record is not easily reconciled with a methane-driven scenario.

Another explanation for the positive and negative excursions is that there were massive changes in the burial of organic carbon relative to carbonate carbon (forg). Such an explanation requires periodic episodes of extraordinarily high organic carbon burial (forg > 0.5, resulting in a positive shift in δ13Ccarb) alternating with episodes of a much lower fraction of organic carbon burial (forg < 0.2, resulting in a negative shift in δ13Ccarb), a scenario that may also account for what has been observed in the carbon isotope record for the Neoproterozoic and Cambrian (39, 40). One hypothesis that has been proposed to explain very high δ13Ccarb values at those times was a more tropical continental configuration, such that tropical river deltas became anoxic, greatly increasing phosphate recycling and allowing very high rates of organic carbon burial (37). If correct, then perhaps the Early Triassic, tropical Tethys ocean basin alternated between oxic and anoxic conditions, with very large changes in organic burial rates. Alternatively, large variations in carbon burial fluxes may have resulted from mechanisms related to the low diversity of Early Triassic ecosystems. Regardless of the explanation, the close resemblance of the Early Triassic carbon isotope record with the repeated excursions recorded near the Neoproterozoic-Cambrian boundary (39, 40) may reflect similar forcing mechanisms.

The Early Triassic interval marked by repeated large and rapid isotopic excursions coincides with the paleontologically observed interval of limited biological recovery. Fossils provide several independent metrics of recovery. After the decimation of Late Permian ecosystems, global diversity began to rise in the Smithian, with the largest and most rapid increase occurring in the early Anisian. Much anecdotal evidence indicates that body size in many invertebrate groups remained small throughout the Early Triassic (13). Figure 3, which includes a global compilation of maximum size in gastropods, provides quantitative confirmation of this phenomenon for one group. The reappearance of calcified green algae (41) also began in the Early Anisian (Aegean and Bithynian), coincident with carbon isotope stabilization. Scleractinian corals appeared, and heavily calcified sponges reappeared shortly thereafter, in the Pelsonian (42). The considerable Anisian diversity of dasyclad algae, scleractinian corals, and sponges reflects the nearly simultaneous resumption or acquisition of calcification in several lineages within these groups (43). Differences in physiological mechanisms of biomineralization may explain the slower time scale of skeletal (re)acquisition in previously uncalcified lineages of scleractinians and sponges versus algae.

The stratigraphic coincidence of large carbon isotope fluctuations with the protracted delay of biotic recovery from the end-Permian mass extinction suggests two classes of explanations. One possibility is that the carbon isotopic variations represent repeated environmental disturbances that directly inhibited biotic recovery. If correct, then the disturbances must have followed one another with sufficient rapidity to preempt any visible recovery in the fossil record. This is consistent with the known paleontological pattern and can be tested by future refinements in the timing of first appearances of Middle Triassic scleractinian corals and sponges (i.e., they should post-date isotopic stabilization by an interval longer than the time between Early Triassic isotopic excursions). Another possibility is that the carbon isotope variations are, themselves, a consequence of decimated Early Triassic ecosystems, reflecting ecological controls on organic carbon burial. Whichever explanation is correct, the association between the most extreme carbon isotope excursions observed in the Phanerozoic Eon and the delayed recovery in the aftermath of Earth's largest mass extinction challenges our understanding of the ways that biological diversity and carbon cycling have interacted through Earth history.

Supporting Online Material

Materials and Methods

Table S1

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article