It is generally assumed that the δ ¹³C of carbonate ultimately reflects the carbon isotopic composition of the dissolved HCO₃ ^2– in the water mass from which the carbonate precipitated. However, where carbonates are rare or absent, some have substituted the δ ¹³C of bulk organic matter and assumed a fractionation factor, usually ~25 to 28‰ (reflecting the isotopic offset imparted by photosynthetic phytoplankton common in marine environments), to compare with δ ¹³C variations in carbonate-rich sections. This approach works well for open ocean settings far from the influence of continents and for geologically older sections that were deposited before the advent of a volumetrically significant terrestrial biomass (e.g., Precambrian). Unfortunately, Permo-Triassic sections that were deposited near continents may contain a mix of terrestrial and marine organic carbon (e.g., [29], [32] and [90]). Terrestrial and marine carbon may have vastly different δ ¹³C compositions and the mixing of the two may result in an averaged δ ¹³C value for bulk carbon that does not directly reflect the CO₂ fixed by photoautotrophs in the marine environment. Thus, the δ ¹³C of bulk organic carbon where continental influence may be present cannot be confidently interpreted to reflect whole ocean shifts in δ ¹³C without intensive compound specific isotopic studies. For example, Foster et al. [28] demonstrated that a negative δ ¹³C_organic excursion in the marine Woodada-2 borehole, Australia, was caused by the switch from a preponderance of isotopically heavy reworked woody material in the Latest Permian to isotopically lighter acritarch-dominated material in the Early Triassic and not by a change in the δ ¹³C of marine CO₂. Therefore, the δ ¹³C profiles generated from organic matter alone are considered less robust than those from unaltered carbonates and correlations based solely on shifts in bulk organic carbon should be treated with caution.

In general, the Permo-Triassic transition is characterized by a stratigraphically sharp negative δ ¹³C excursion from a Latest Permian peak of ~+2 to +4‰ to an isotopic nadir of ~ –1‰ (Fig. 2). The stratigraphic position of the δ ¹³C excursion varies from study to study, although some of this confusion may stem from the fact that the GSSP for the basal Triassic was only recently approved and now coincides with the first occurrence of the conodont H. parvus [107]. Currently, most workers place the isotopic nadir within the Latest Permian, in association with the Latest Permian Changxingian extinction event. The excursion precedes the base of the Triassic at the type section in Meishan [50]. Another good example of this would be the Dajiang section in South China [80]. Here, the base of the Triassic, coincident with the first occurrence of H. parvus, occurs tens of meters above the isotopic nadir, where δ¹³C values are relatively constant at ~+2‰. According to the new data of Richoz [88], the C isotope shift appears stepwise: the first shift occurred in the Late Changxingian before the mass extinction and the second shift follows the mass extinction (but precedes the first occurrence of of H. parvus).

In some sections, the stratigraphic duration of the negative anomaly spans many tens of meters. In the Gartnerkofel core (Austria), for example, the δ ¹³C anomaly spans approximately 30 m; in detail, the profile declines from ~+3‰ to +1‰ over the first ~25 m and then declines more abruptly to ~–1‰ over ~5 m [43]. An examination of the literature would suggest that most workers consider the abrupt decline to represent the characteristic “negative excursion” associated with the boundary transition. As stated above, the decline does not appear as abrupt in some sections; instead, it appears stepwise and spans at least two conodont subzones [88]. On the South China Platform [80], the anomaly spans 50 m at the Dajiang section, from +4‰ to ~–1‰, with an increase in the rate of isotopic change in the last 5 meters. At the Guryul section in Kashmir [1] and [5], the anomaly spans ~30 m from +2‰ to –4‰, without an oversteepening of the trend in the upper part of the section. Such discrepancies likely originate from stratigraphic completeness and sampling issues (cf. [69]).

The temporal duration of the anomaly is also debated [13], [65], [75] and [82]. If we take a simplistic approach and compare the stratigraphic duration of the anomaly versus average sedimentation rate for carbonate platformal sections for these examples (17–200 m/Ma) (e.g., [11]), the decline from positive δ ¹³C values in the Latest Permian to the isotopic nadir would appear to last ~1 Myr, with the most rapid decrease encompassing the last 160 000 years. Bowring et al. [13] studied the geochronology of the type section at Meishan and determined that the δ ¹³C excursion lasted less than 165 000 years, in good agreement with the average sedimentation rate model. Alternatively, Rampino et al. [82] employed astronomically tuned cyclostratigraphy of the Gartnerkofel core to suggest a much shorter duration, less than 30 000 years. More recently, Mundil et al. [75] recalibrated the Permo-Triassic boundary interval using different geochronologic methods than Bowring et al. [13] in South China and suggest that the negative anomaly may have lasted as long as 2 Myr, given the uncertainties in their data. It is critical to know the duration of the anomaly: shorter durations favor catastrophic processes while longer durations preclude such processes. Thus, the uncertainty surrounding its duration hampers our ability to constrain the cause of the negative anomaly.

The extreme magnitude of the end-Permian extinctions attracts much attention, such that the δ ¹³C record for the remaining Early Triassic has been virtually ignored and the available data originate from only a few sections (see Fig. 1 and Table 1). Indeed, few studies report useful δ ¹³C profiles beyond a few tens of meters of the base of the Triassic. Baud et al. [5] were the first to use systematically sampled complete sections spanning the Lower Triassic (composed of the Griesbachian, Dinerian, Smithian, and Spathian substages, respectively) in its entirety, although the isotopic profile is not continuous. By studying strata from northern Gondwana, now located on the northern margin of the Indian subcontinent in Pakistan, Indian Kashmir, Spiti (India) and Nepal (Fig. 1), it became apparent that the Early Triassic as a whole was isotopically anomalous [1], [2] and [5]. At Nammal Gorge in the Salt Ranges (Pakistan), the δ ¹³C record is characterized by a return to mildly positive δ ¹³C values above the ‘pre-boundary excursion’ followed by a mildly negative δ ¹³C excursion through Upper Griesbachian strata. A minor positive trend is noted at the Griesbachian–Dienerian boundary, followed by a negative excursion that culminates in Middle Dienerian beds. Carbon isotope values return positive for the remaining Dienerian stage and culminate in a strong positive δ ¹³C excursion best recorded in the Losar section (Spiti, India) at or near the Dienerian–Smithian boundary. Smithian strata return to mildly negative δ ¹³C values. A second strongly positive excursion occurs across the Smithian–Spathian boundary; it is pronounced at the Landu Nala section (–2 to +4‰), and even more pronounced in the Losar section (+5‰, Spiti, India). Lower Spathian strata record a final mildly negative δ ¹³C excursion, followed by a third pronounced positive excursion across the Spathian–Anisian boundary. Thus, the δ ¹³C record oscillates by at least 10‰ during the Early Triassic, as demonstrated by the northern Gondwana sections. In sum, the upper Lower Triassic δ ¹³C record is punctuated by three extraordinarily large positive shifts that correlate well with the recent results from China [80], discussed below. The δ ¹³C record from Oman (Batain region) reveals 10‰ δ ¹³C oscillations, as well [88].

The most recent dataset to date originates from the South China region [80], which represents an isolated carbonate platform on the far eastern margin of the Paleotethys during Early Triassic time (Fig. 2). The stratigraphic succession spans the Upper Permian, through the Middle Triassic, and into the Upper Triassic. The advantages of the South China section include relatively continuous carbonate platform sedimentation, the appropriate biostratigraphy, and the ability to compare pre- and post-Lower Triassic δ ¹³C profiles to the Lower Triassic profiles. Broadly, the results are strikingly similar to the Lower Triassic sections from the northern Indian subcontinent, but with greater sampling resolution. Notable differences are apparent during the Smithian-Spathian positive excursion: the Losar section, discussed above, reveals a more strongly positive excursion to δ ¹³C values than the South China sections. It should be noted that the biochronological calibration in the Losar section [1] is based on ammonoid biostratigraphy, whereas the South China section is based upon conodont biostratigraphy, and minor biostratigraphic offsets may be present. Regardless, the Lower Triassic succession from South China reveals remarkable δ ¹³C oscillations on the order of 10‰ over a 4- to 6-million-year period. Notably, the South China data reveal that the Middle Triassic δ ¹³C record is strikingly calm when compared to the Lower Triassic record.

Hypotheses that explain the negative δ ¹³C anomaly in association with the Permian–Triassic transition fall into two categories: temporally instantaneous or temporally protracted. Unfortunately, the timing of the boundary δ ¹³C anomaly is unclear, as discussed above, and the radiometric calibration of the remaining Early Triassic is not available. It should be noted that most of the hypotheses discussed below were formulated before it was clear that the entire Early Triassic δ ¹³C record was so unusual and characterized by massive 10‰ changes. Thus, most workers likely assumed that the δ ¹³C record returned to ‘normal’ shortly after the boundary excursion. We now know this is not the case (Fig. 2), which clearly questions the validity of the instantaneous causes.

Instantaneous events include (1) the catastrophic release of methane from the gas hydrate reservoir with δ¹³C ~–60‰, (2) the input of CO₂ from volcanic out-gassing with δ ¹³C ~–6‰, and (3) the release of CO₂ from rotting terrestrial biomass after a catastrophic die-off with δ ¹³C ~–25‰. Many proponents of the methane source suggest the gas hydrate reservoir was destabilized by a bolide impact – for which there is little unambiguous evidence [26] and [48]. Modern estimates would suggest that the marine clathrate reservoir requires ~10 Myr to ‘recharge’ after a catastrophic release (e.g., [19]). Thus, if destabilization of clathrates caused the boundary excursion, alternate causes must be found to explain the Griesbachian, Smithian, and Spathian negative δ ¹³C excursions that followed, insofar as the entire Early Triassic may span only 4 to 6 Myr (cf. [56]). In addition, the shape of the isotopic profile for the post-boundary Early Triassic does not support a methane trigger. In general, clathrate destabilization is thought to produce an asymmetric δ ¹³C profile – cf. Paleocene–Eocene event [55] whereas the post-boundary profile is virtually symmetric going into and out of each δ ¹³C excursion – as noted by [80]. The Siberian Traps, a large magmatic province formed at the same time as the negative δ¹³C anomaly (within radiometric error), is commonly cited as the prime volcanogenic culprit [84]. As with the gas hydrate model, it is unclear whether the Siberian Traps continued to erupt for the entire Early Triassic. On the other hand, the dating with respect to the upper limit of the volcanic flows is incomplete and the Siberian traps may remain a possible source of isotopically light carbon. The mass mortality followed by the mass rot-off is hypothesized to have occurred because of a bolide impact or other extraterrestrial cause; this instantaneous hypothesis suffers from the same deficiencies as the others. There is no evidence, for example, that the terrestrial biosphere was decimated and regenerated several times over the course of the Early Triassic.

Protracted causes for the δ ¹³C anomaly would include (1) ocean stratification/turnover and (2) the reorganization of the carbon cycle. The ocean stratification model would suggest that isotopically light organic matter produced in the photic zone would sink and become remineralized in the deeper portions of a stratified ocean, creating an isotopically heavy surface ocean and an isotopically depleted deep ocean. Carbonates deposited in surface waters during stratification would record isotopically heavy δ ¹³C values. The episodic overturn, or destratification, of such an ocean would return light carbon to the surface ocean and resulting in the negative δ ¹³C anomaly. Abundant evidence exists for anoxia during the Early Triassic, implying that a certain degree of stratification is likely [30], [36], [47], [52], [53], [99], [100], [101] and [103]. Models of ocean circulation and chemistry, however, differ on whether or not whole ocean stratification is possible, and mass balance considerations are problematic (see below). The reorganization of the carbon cycle hypothesis involves the reduction in the burial flux of terrestrial biomass versus marine biomass. In this scenario, the reduction of the terrestrial biomass is shown by the noted ‘coal gap’ in the stratigraphic record and the change from high biomass arborescent plants in the Permian to low biomass herbaceous plants in the Early Triassic. Such a reorganization of the carbon cycle would change the burial proportion of terrestrial versus marine organic carbon and cause a long-term negative shift in the δ ¹³C of the oceans. Essentially, the burial of organic carbon would shift from terrestrial + marine to marine alone, causing the negative shift. This scenario is appealing for long-term changes in the carbon cycle, but is less appealing for rapid δ ¹³C shifts.

The negative δ ¹³C anomaly associated with the Permo-Triassic boundary has been modeled in an attempt to better understand the processes responsible for the δ ¹³C anomalies [10], [14] and [31]. The models are restricted by the ambiguity surrounding the duration of the negative anomaly and similarly fall in to ‘instantaneous’ versus ‘protracted’ categories. Broecker and Peackock [14] modeled the reorganization of the carbon cycle, as outlined above. Their scenario fits both the δ ¹³C and the δ ³⁴S record well, assuming that a long duration for the anomaly is chosen. Berner [10] assumes a short duration (~20 000) for the decline from positive to negative δ ¹³C values. Importantly, Berner's model demonstrated that the reservoir of isotopically depleted carbon available in the mass-mortality and overturn hypotheses was much too small to cause the observed δ ¹³C shift, regardless of the duration, and thus these hypotheses were rejected as sole causes. Even with such a short duration, methane release alone was not found to be sufficient to cause the boundary δ ¹³C anomaly, given reasonable rates of clathrate decomposition. Rather, the best model fit to the data occurred when multiple sources of isotopically depleted carbon were combined (cf. [24]). Such a combination of factors seems inevitable if we seek to explain the unusually variable Early Triassic δ ¹³C record. At any rate, the lack of consensus on the duration of the δ ¹³C boundary anomaly clouds such studies.

The Phanerozoic δ ¹³C record is somewhat tranquil with respect to multiple large and rapid swings in δ ¹³C when compared to the Early Triassic, although exceptions are clearly evident as high-resolution datasets are collected (e.g., [96]). The Early Cambrian, however, displays many of the same carbon isotopic features as the Early Triassic. As many as eight excursions are noted in Lower Cambrian strata over the course of approximately 20 Myr – see compilation in [57] – (Fig. 3); the magnitude of the excursions decreases from a maximum of ~10 ‰ at the Precambrian-Cambrian boundary to less than a few per mil at the Lower-Middle Cambrian boundary. It may not be wise to compare the two periods based solely on a perceived similarity in the magnitude and rapidity of their δ ¹³C records. However, comparative studies may reveal features not originally noted in one system or the other, so with caution, we will examine the similarities and differences between the Early Cambrian and Early Triassic biospheres.

Notable ecological confluences between the Early Cambrian and the Early Triassic set both periods apart from the remaining Phanerozoic. Similarities would include the importance (or lack thereof) of the terrestrial biosphere, the depth of infaunal tiering (e.g., [3]), and the average ichnofabric index [23]. The importance of a eukaryotic terrestrial biosphere as a carbon cycle moderator has been discussed above, and both the Cambrian and Triassic are characterized by a reduced terrestrial biosphere. The Ordovician and Silurian likely lacked a well-developed terrestrial biosphere, as well, but both record advanced ichnofabrics and relatively well-developed infaunal tiering [3] and [12]. In contrast, the Early Cambrian and the Early Triassic share a reduced average ichnofabric index and similarly reduced maximum depth of infaunal tiering [12]. The consequences of reduced sediment churning/oxygenation via bioturbation/infaunalization could include a change in the way carbon is stored and recycled on the continental shelves. The bulk of modern organic carbon is buried in continental shelf environments. Perhaps a reduction in the biogenic sediment churning would enhance organic carbon burial, and thus drive the δ ¹³C of seawater to much heavier values, as noted in the Dienerian-Smithian boundary [80]. Such an enhanced organic carbon reservoir would return isotopically light carbon to the system if exhumed. These processes would clearly operate on longer time scales, consistent with the current δ ¹³C record for the Early Triassic, and would be enhanced by the presence of anoxic bottom waters.