Prolonged Late Permian–Early Triassic hyperthermal: failure of climate regulation?

1. Introduction

In the context of the archetypal hyperthermal (the Palaeocene–Eocene Thermal Maximum or PETM), the Late Permian–Early Triassic (P-Tr) event presents an extreme end-member. Unlike the relatively benign effects of the PETM on the biota [1], the P-Tr event was the largest mass extinction of animals in Earth history [2]. Extinction during the PETM was restricted to the benthic foraminifera; other groups, including the mammals, experienced range shifts and diversifications. In contrast, during the Permian–Triassic event virtually all major groups of plants and animals suffered extinction, especially sessile and heavily calcified organisms [3]. The two events left similar records in the carbon isotope composition of limestones (figure 1a), reflecting potentially similar carbon cycle perturbations, both likely triggered by volcanism [8,9] and involving amplifying carbon-cycle feedbacks. Yet while the global average temperature rose approximately 5°C during the PETM [10,11], there is growing but still relatively sparse evidence that equatorial temperatures increased by as much as 16°C during the P-Tr [7]. Moreover, extreme warmth persisted for up to 5 million years during the P-Tr, whereas recovery occurred within 200 thousand years during the PETM (figure 1b).

• Download figure
• Open in new tab
• Download PowerPoint

One might question the assumption of equivalent forcing: similar carbon isotope excursions (CIEs) can be generated in an infinite number of ways, through combinations of the amount of low δ¹³C carbon added to the ocean/atmosphere system and the difference between the δ¹³C of that source and the ambient ocean/atmosphere mean δ¹³C; to a crude first approximation, the magnitude of the CIE scales with the product of those two terms (in detail, rates and durations matter). So it is possible that the P-Tr event was driven by emissions of carbon with considerably higher δ¹³C (closer to the magmatic end-member value of −5‰ to −6‰). By adding the additional constraint on pH from boron isotopes, Clarkson et al. [12] and Gutjahr et al. [13] were able to estimate both the amount of carbon added and its isotopic composition for the P-Tr and PETM events, respectively. Clarkson et al. [12] estimated that the two phases of P-Tr extinction were driven first by a small addition of isotopically light C (e.g. methane) and then by a massive addition of isotopically heavy C (e.g. from decarbonation of limestones intruded by Siberian Traps dikes and sills), with a total emission of from 30 000–40 000 Pg C (consistent with an independent estimate by Svensen et al. [9]). A similar analysis of the PETM, consistent with boron and carbon isotope records, indicates 10 000 Gton C was released [13]. There is significant uncertainty in the boron isotope pH proxy, especially in shallow-water sedimentary carbonate archives like that employed for the P-Tr estimate (e.g. [14]), and thus so too in the carbon emissions estimated from them. Nevertheless, it seems likely that the total amount of CO₂ emitted during the P-Tr event could have exceeded that of the PETM by a factor of 3–4.

This paper explores the possibility that the silicate weathering feedback was overwhelmed by the 30 000–40 000 Pg C released by Siberian Trap volcanic activity during the P-Tr vent, and that this, together with other factors that minimized resilience of the climate regulating system, led to failure of regulation (persistently high levels of atmospheric CO₂) for up to millions of years.

2. Climate regulation through silicate weathering

Planetary temperature regulation is generally thought to arise from feedback involving atmospheric pCO₂, surface temperature, rainfall and runoff, and chemical weathering of silicate rocks exposed on Earth's surface [15,16]. An increase in atmospheric pCO₂ resulting from, for example, an interval of intense volcanic activity, will warm the planet and increase rainfall rates. Both of these changes in climate will tend to increase rates of silicate weathering (e.g. [17]), a process that consumes CO₂ and releases neutralized carbon in the form of carbonate and bicarbonate ions to soil waters, eventually with transport to the ocean and sequestration into sedimentary carbonates. This series of relationships creates a negative feedback that can damp large fluctuations in atmospheric pCO₂ and climate as long as weathering rates respond positively and sufficiently strongly (i.e. aren't limited by other factors). The thickness of the soil layer depleted in primary silicate minerals determines the extent that soil-water flowpaths intersect unweathered material: when soils are too thick, flowpaths do not penetrate to the unweathered bedrock, and CO₂ neutralization does not occur and the system becomes transport limited [18]. When soils are too thin, the residence time of waters is too short. Optimal conditions depend on various factors [19] but ultimately, the maximum rate of silicate weathering is set by the rate at which rock uplift transports new, fresh bedrock minerals to the surface [20,21].

3. Past variations in silicate weathering rates

Many studies have provided estimates of global silicate weathering rates for various intervals of Earth history, and some have looked specifically at times when factors limiting silicate weathering may have driven global warming events. Kump & Arthur [22], Kent & Muttoni [23], Li & Elderfield [24], Froelich & Misra [25] and Caves et al. [26] all suggest that increasing silicate rock weatherability (susceptibility to weathering, influenced by uplift, exposure, and geographical position) likely drove the Cenozoic global cooling trend. Similarly, Kump et al. [27] and Godderis et al. [28] invoke similar reasoning for the onset and termination of the Late Ordovician and Permian/Carboniferous ice ages, respectively. Only Froelich & Misra [25] explored a failure of silicate-weathering feedback as an explanation for a sustained ‘hothouse’ climate interval: they argue that the Eocene warmth was brought on and sustained by anomalously slow rates of tectonic uplift. Very slow uplift presumably led to thick, unreactive residual soils that severely diminished the maximum potential silicate weathering rate below the volcanic CO₂ outgassing rate.

4. The maximum silicate weathering rate

One can make a simple estimate of this maximum rate (F_sw,max) if it is assumed that all silicate minerals brought to the earth surface through uplift are completely depleted of their soluble cations (Ca, Mg, Na, K) requiring the maximum carbonate alkalinity to balance their charge (i.e. the maximum CO₂ neutralization rate). Then

where E is the global average denudation rate, today estimated to be about 5 cm kyr⁻¹ (if estimates of global denudation from Larsen et al. [29] of 28 Gtons per year are distributed across the total continental area, a rate consistent with that of Willenbring et al. [30,31]) presumed equal to the average uplift rate at a global steady state, ρ is the density of average upper crustal materials (2.65 g cm⁻³; [29]), A is the area of continental exposure, today less than or equal to the total continental area (210 × 10⁶ km²), and Ψ is the CO₂ demand (4.6 mol CO₂/kg; table 1; [32]) based on the composition of average upper continental crust [33]. The product of these various factors is (correcting for units) is 130 × 10¹² mol CO₂ yr⁻¹. The largest uncertainty in this estimate of the modern day maximum silicate weathering rate is likely the estimated uplift rate; Larsen et al. [29] state that the uncertainty of the total continental denudation rate is from a factor of 2 higher to a factor of 2 lower than their estimate. If we accept this as the main source of uncertainty, F_sw,max ranges from 65–260 × 10¹² mol CO₂ yr⁻¹.

The estimate of F_sw,max is approximately 20(10–40) times the canonical modern volcanic CO₂ degassing rate of 6 × 10¹² mol CO₂ yr⁻¹ (e.g. [34]). In comparison, the maximum weathering rate estimated by Foley [21] for an arbitrary uplift rate of 10 mm yr⁻¹, scaled to our estimate of the modern global average of 5 cm kyr⁻¹, is approximately 50 × 10¹² mol cation yr⁻¹, which when converted to CO₂ demand is within the range of estimates presented here (other smaller differences include land area and crustal density).

It is important to emphasize that 65–260 × 10¹² mol CO₂ yr⁻¹ is a likely upper limit for a rock uplift rate of 5 cm kyr⁻¹, given that all other factors are maximized. A more realistic upper limit would factor in the area of basement exposure where primary silicate minerals are preferentially exposed (about 129 × 10⁶km²; [33]), reduced by the area of desert (approximately 25% of the land surface), to yield a rate of about 60 × 10¹² mol CO₂ yr⁻¹, or about 10 times the canonical outgassing rate. Variations on the million-year time scale in volcanic outgassing are generally thought to have been no more than 80% greater than today and never less than today during the Phanerozoic [35], although estimates of sustained Cretaceous outgassing are as high as 5.5× the modern rate [36]. Even given these higher outgassing rates, long-term secular variations in volcanic outgassing fell well within the range in which the silicate weathering feedback was operative (i.e. in which global weathering rates we well below the uplift-limited maximum), presuming that rock uplift rates today are representative of those of the past (see below).

By comparison, modern fossil-fuel burning is 10 Pg C yr⁻¹, or 830 × 10¹² mol yr⁻¹ and peak PETM and P-Tr emission rates are estimated to be about 80–120 × 10¹² mol yr⁻¹ [37]. Thus CO₂ emission during our incipient and these two geo-historical hyperthermals meet or exceed the maximum ability of the silicate weathering feedback to regulate atmospheric CO₂. So during the emissions event itself, the silicate weathering regulatory mechanism was ineffective. Current thinking, however, is that once the emissions stopped, atmospheric CO₂ levels were drawn back toward their pre-perturbation level through a variety of feedback mechanisms, including ocean uptake, seafloor carbonate dissolution, and carbonate and silicate weathering on land, with a wide range of time scales ultimately completed through warm-climate enhanced silicate weathering on an order 100 kyr time scale [38,39]. The PETM seems to have played out this way, whereas the P-Tr remained hot for millions of years. What delayed the climate recovery?

5. Low resilience of climate regulation during the P-Tr event

An average weathering zone thickness of around 4.4 m [20] has a residence time (at 5 cm kyr⁻¹ of steady state rock uplift) of approximately 100 kyr. Thus, a volcanic event that released (55 × 10¹² mol CO₂ yr⁻¹ × 10⁵ yr=) 5.5 × 10¹⁸ mol or 66 000 Pg C would completely deplete the weathering zone of primary silicate minerals, creating a globally transport-limited weathering regime that would be unable to regulate atmospheric CO₂ and climate. Recovery would take 100 000 yr under today's uplift rate. The P-Tr estimate of 30 000–40 000 Pg C falls a bit shy of this threshold.

However, if uplift rates were an order of magnitude slower in the Late Permian than today, the recovery time scale would be order 10⁶ yr. Is there any reason to suspect that Permian–Triassic uplift rates were anomalously slow? One gauge of tectonic activity is the area of continental contractional deformation, the areas likely to have been rapidly uplifting, as reconstructed for various times in the geologic past using palaeogeographical techniques. Richter et al. ([40]; figure 2) found that such a reconstruction, used to scale silicate weathering rates, reproduces the Phanerozoic Sr isotope evolution to a good first approximation. The low areas of deformation in the late Permian and early Triassic, together with the very low values of marine ⁸⁷Sr/⁸⁶Sr at this time (the Palaeozoic and near Phanerozoic minima) and an apparent Phanerozoic minimum in global sedimentation [42], suggest that uplift rates could have been at or near the Phanerozoic minimum. If uplift rate scales with contractional area, then rates at this time could have been 10% of modern rates. Interesting, a broad minimum occurs in the late Cretaceous through early Cenozoic, consistent with Eocene warmth and the ‘flat Earth’ hypothesis of Froelich & Misra [25]. However, other minima and maxima are not closely tied to extreme climate states; one factor may be the challenge of accurately assessing the timing of the palaeotectonic activity reconstructions, to which the authors ascribe a 20 Myr uncertainty. The Ordovician minimum corresponds with a time of global warmth [43] and increased tectonism through the middle to late Ordovician has been invoked to explain Late Ordovician glaciation [44].

• Download figure
• Open in new tab
• Download PowerPoint

Another factor that reached a minimum in the Permian–Triassic was the degree of continental fragmentation [45]. Low fragmentation (high continentality) during the Pangean supercontinental configuration of the Permian and Triassic was accompanied by dry tropical and subtropical continental interiors and a strong mosoonal climate with widespread evaporite and aeolian sand deposition [46,47]. With rainfall focused on the continental margins, the likelihood of developing transport-limited weathering conditions would have been enhanced. Vast regions of the cratonic interiors would have remained unweathered despite increasing warmth as atmospheric CO₂ increased.

These inferences are supported by numerical modelling of the coupling between climate and weathering. Gibbs et al. [48] used output from the atmospheric general circulation model (AGCM) GENESIS together with a model of chemical weathering as a function of temperature, runoff and lithology to calculate changes in global weathering from 250 Ma to the present. They found that palaeogeography was a strong driver of global weathering, characterized by two states, Pangean and Tethyan, with high continentality/low weathering rate and highly dispersed continents/ high weathering rate, respectively. Similarly, Donnadieu et al. [49] found, through a series of simulations representing the break-up of Pangea, that the transition from supercontinent to dispersed continents increased global runoff and thus silicate weathering rates. Their modelling, coupling the FOAM AGCM and their GEOCLIM weathering model, generated persistent aridity, low weathering, and thus high atmospheric CO₂ during the Late Permian/Early Triassic.

These two factors, slow uplift and high continentality, mean that even before the Siberian Traps eruptions, maximum rates of CO₂ uptake during silicate weathering were much depressed and could have approached the canonical volcanic emission rate. In other words, the carbonate silicate climate feedback mechanism could have been poised very close to, or come to failure before (and after) the eruptive event.

The hypothesis that Late Permian–Early Triassic silicate weathering rates were overwhelmed by volcanism is a testable hypothesis: early Triassic terrigenous clastic sediments should be intensely weathered, and this intense weathering should be reflected in low marine Li isotope values, a phenomenon demonstrated for the Eocene [25]. Indeed, while this manuscript was under review, Sun et al. [50] published a study of the Li isotopic composition of marine carbonates from the P-Tr boundary, revealing values that are considerably lower than today's and indicating extreme weathering intensity (essentially congruent dissolution) during the boundary event, consistent with the hypothesis proposed here. Although carbonate δ⁷Li remained low in the earliest Triassic at Meishan (South China), high clay content complicates the determination of original seawater isotopic composition: indeed, Sun et al. [50] interpret this interval to have returned to modern-like weathering intensities, although their observations and simulations allow for persistently intense weathering rates.

6. Conclusion

Several factors seem to have contributed to the Late Permian to Early Triassic climate regulation failure and prolonged hyperthermal conditions and distinguish its response to volcanism from that of the PETM. The agglomeration of the continents into the supercontinent Pangea culminated in late Palaeozoic time with high continentality with arid continental interiors and subdued average global rock uplift. This predisposition toward failure of the carbonate-silicate climate regulator was then put to the test with the emplacement and eruption of the Siberian Traps, which provided a sustained high emission of carbon dioxide sufficient to fully deplete the weathering zone of soluble cations and thus of CO₂ neutralizing capacity. The heavier δ¹³C of this volcanic flux led to a deceptively modest CIE, similar to that of the PETM. Only with the gradual shedding of this weathered material over the ensuing hundreds of thousands to millions of years did fresh silicate material become available for the climate regulator. In contrast, the PETM seems to be an event that exceeded regulatory capacity during the period of CO₂ emissions but did not create such an extensive weathered regolith, allowing the expected climate recovery to occur. Subsequent drivers of hyperthermal conditions seem to have occurred following both the PETM and P-Tr events, creating smaller hyperthermals following the PETM [51] but perhaps sustaining hothouse conditions well into the Early Triassic [3]. Again, the pre-existing conditions of the Late Palaeozoic–Early Mesozoic seem to have reduced the effectiveness of the carbonate-silicate weathering feedback.

Data accessibility

This article has no additional data.

Competing interests

We declare we have no competing interests.

Funding

This research was supported by a grant from the Heising-Simons Foundation.

Footnotes

One contribution of 11 to a discussion meeting issue ‘Hyperthermals: rapid and extreme global warming in our geological past’.