Earth-Science Reviews: Ying Cui, Lee R. Kump
Earth-Science Reviews: Ying Cui, Lee R. Kump
Earth-Science Reviews: Ying Cui, Lee R. Kump
Earth-Science Reviews
journal homepage: www.elsevier.com/locate/earscirev
a r t i c l e i n f o a b s t r a c t
Article history: The mass extinction event that occurred at the close of the Permian Period (~252 million years ago) represents
Received 9 October 2012 the most severe biodiversity loss in the ocean of the Phanerozoic. The links between the global carbon cycle,
Accepted 22 April 2014 climate change and mass extinction are complex and involve a whole range of often inter-related geochemical,
Available online 8 May 2014
biological, ecologic and climatic factors. It has become widely accepted that the end-Permian mass extinction
was associated with a global warming event, because the age of the Siberian Trap eruption, a potentially massive
Keywords:
End-Permian extinction event
source of carbon dioxide, coincides within error with the extinction event. However, geologic data that are in sup-
Climate model port of this global warming event are relatively sparse. The chain of events and the causal relationship between
Geochemical proxies the eruption of Siberian Traps and mass extinction is not well established. Global warming, in particular, has only
Temperature been reported from limited proxy data and climate models, for which the pCO2 in the atmosphere just before and
pCO2 during the end-Permian extinction event is poorly known. This study critically assesses both the proxy climate
data and the existing paleoclimate simulations with the goals of assessing our current understanding of the
link between mass extinction and climate change and providing some guidance for future studies. Proxies indi-
cate that prior to the end-Permian extinction event tropical sea surface temperatures ranged from ~22 to 25 °C,
and possible pCO2 values ranged from ~500 to ~4000 ppm before the extinction event. During the peak extinc-
tion, tropical temperatures rose up to ~30 °C while pCO2 perhaps increased up to ~8000 ppm. Climate models
that use different pre-event pCO2 values show similar amount of CO2 doubling to replicate the observed carbon
isotope excursions. We find that the temperature anomaly during the end-Permian extinction is consistent with
~1.5 doublings of atmospheric pCO2, and that the implied climate sensitivity is 5–6 °C, within the upper range of
values produced by climate models.
© 2014 Elsevier B.V. All rights reserved.
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2. Climate change during the Permian–Triassic transition — Geochemical proxy perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1. Temperature proxies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.1. Oxygen isotope paleothermometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.2. Clumped isotopes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2. Atmospheric pCO2 proxies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2.1. Plant stomatal density and index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2.2. Carbon isotopes of carbonate nodules in paleosol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3. Climate change during the Permian–Triassic transition — Modeling perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.1. Conceptual models and EBMs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.2. Geochemical box models used to investigate hypotheses associated with the end-Permian extinction event . . . . . . . . . . . . . . . . 12
3.3. Early general circulation models (GCMs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.4. Ocean general circulation models (OGCMs) and the oceanic anoxia hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.5. Earth system models of intermediate complexity (EMICs) and fully coupled atmosphere–ocean general circulation models (AOGCMs) . . . . 16
3.5.1. EMICs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.5.2. AOGCMs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
4. Climatic impact from the Siberian Traps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
⁎ Corresponding author at: 512 Deike Building, the Pennsylvania State University, University Park, PA, 16802, United States.
E-mail address: [email protected] (Y. Cui).
http://dx.doi.org/10.1016/j.earscirev.2014.04.007
0012-8252/© 2014 Elsevier B.V. All rights reserved.
6 Y. Cui, L.R. Kump / Earth-Science Reviews 149 (2015) 5–22
1. Introduction forming a global coal gap. Expansion of subtropical desert belts indicates
arid conditions during the Late Permian (Fluteau et al., 2001).
The earth system experienced a series of fundamental upheavals Other lines of evidence, including extremes of the oceanic isotopic
through the Permian–Triassic (P–T) transition (ca. 252 Ma), an interval compositions of C, S, and Sr suggest that the global environment
marked by the largest mass extinction in the geologic record (Burgess reached a critical state in the late Paleozoic for which perturbation,
et al., 2014; Erwin, 2006). The configuration of the supercontinent namely massive volcanism associated with the eruption of the Siberian
Pangea, formed through collision of the southern supercontinent of Traps, could pitch it over a threshold into near uninhabitability
Gondwana with the northern supercontinent of Laurasia during the (e.g. Kiehl and Shields, 2005; Erwin, 2006; Luo et al., 2010; Fig. 1).
late Paleozoic, has been proposed to be responsible for the icehouse to Some researchers have proposed that the end-Permian mass extinction
greenhouse transition during the Late Permian (Erwin, 1993). The ex- event was a geologically instantaneous catastrophic event (e.g. Bowring
tent of continental flooding (perhaps reflecting eustatic sea level) et al., 1998; Rampino et al., 2000), but evidence for prolonged extreme
reached its minimum during the Late Permian as the supercontinent environments following the event seems to argue against an instanta-
Pangea was established, leading to extreme continentality (Fig. 1 and neous perturbation (e.g. Romano et al., 2012; Sun et al., 2012). Associated
reference therein). Reduced availability of fresh silicate rock for ocean acidification and deoxygenation, similar to that projected for
weathering due to a long orogenic gap might have caused increased at- the future, was the likely consequence of volcanic CO2 emissions and
mospheric CO2 concentrations and warmed up the Late Permian global warming (e.g. Clapham and Payne, 2011; Montenegro et al.,
(Kidder and Worsley, 2004). The pole–equator temperature gradient 2011; Payne and Clapham, 2012). Following the end-Permian extinction
was probably lower based on the observed high-paleolatitude floras event are widespread microbialite sequences in shallow water settings,
adapted to warm climates (Taylor et al., 1992; Rees et al., 2002). Thick reflecting global expansion of microbial ecosystem (Xie et al., 2005,
and widespread coals up to the Permian–Triassic boundary (Veevers 2007, 2010) and perhaps an anomalously low oceanic sulfate concentra-
et al., 1994; Retallack et al., 1996) disappeared for 10 million years, tion in an anoxic ocean (Luo et al., 2010).
Scaled Atmospheric
Ma Sea level (Scaled) 87
Sr/86Sr δ34S (‰) δ13C (‰) δ18O (‰) Genera CO2 (ppm)
0 A B C D E F G
Pg. Ng.
Cold
GEOCARB III
Paleosols
Stomata
Warm
Ordov. Sil. Devon. Carbonif. Perm. Trias. Jurassic Cretaceous
100
Cold
200
Warm
PTB
Cold
300
Warm
400
Warm Cold
500
2000
4000
6000
0.707
0.708
0.709
Fig. 1. Long-term environmental variations from the Late Cambrian through the Pliocene. (A) Resampled global continental flooding estimates of 5-million-year bin averages at the bin
midpoints of 80 time intervals; (B) to (E) Geochemical analyses from marine carbonates. (F) Total number of marine genera based on North American fossil occurrences; (G) Atmospheric
pCO2 level based on δ13C of pedogenic carbonate, plant stomatal density and index and GEOCARB III.
Panels (A) to (F) are modified from Hannisdal and Peters (2011). δ13C and δ18O records are from low-latitude based on Prokoph et al. (2008). Solid black lines are averages for the time bins
used in Hannisdal and Peters (2011), and the shadowed region brackets the data used in Prokoph et al. (2008). Panel (G) is modified from Royer et al. (2004).
Y. Cui, L.R. Kump / Earth-Science Reviews 149 (2015) 5–22 7
As we review below, over the past 20 years a number of climate depth and secondary effects from diagenesis. More than 20 years ago,
proxy and modeling efforts have been put forward to investigate Holser et al. (1989) reported rapid shifts in stable carbon and oxygen iso-
environmental change during the mass extinction event. Geologic data topes observed in a continuous drill core known as the Gartnerkofel-1
that are in support of this global warming event are relatively sparse core on the north slope of Gartnerkofel near Nassfeld in the Carnic Alps
(e.g. Kidder and Worsley, 2004; Joachimski et al., 2012; Sun et al., of Austria. Key features of this event were a rapid onset (which subse-
2012), and estimates of the background CO2 level are few and discor- quently estimated to be a few thousand years, see Rampino et al.,
dant (e.g. Ekart et al., 1999; Retallack, 2001; Royer, 2001; Royer et al., 2000), a negative shift in the δ13C of carbonate (~− 3‰), a decline in
2004; Retallack, 2013). In some ways the modeling has outpaced the δ18O (~−3‰ to − 4‰ excursion, interpreted as ~ 5 °C warming), and
proxy data collection (e.g. Kiehl and Shields, 2005; Meyer et al., 2008; anomalous concentration of iridium and other metals indicating anoxia.
Cui et al., 2013), and there has been little effort to perform detailed All these changes were synchronous with the known extinction event.
model-data comparison, including geological, paleontological, sedimen- However, the magnitude of sea surface temperature change and the
tological and geochemical data. key greenhouse gas concentration in the atmosphere associated with
This review focuses on the status of modeling investigations into the the extinction event are all subjects of current debate.
Late Permian climate and its association with the end-Permian mass Oxygen isotopic signals preserved in conodont calcium phosphate
extinction event. We begin with a review of proxies used to identify have been used as proxy for paleotemperature because of their resis-
high pCO2 and warm climate in the Late Permian. We then compare tance to diagenetic alteration. Joachimski et al. (2012) measured oxygen
the boundary conditions and results of various modeling studies. We isotopes on monogeneric conodonts from both Meishan and Shangsi
explore the critical uncertainties in the current paleoclimatic models, sections in south China, including the taxa Clarkina and Hindeodus. The
focusing on uncertainties in pCO2. We also evaluate the climatic conse- measured δ18O values of conodonts at Meishan section, converted to
quences of Siberian Trap volcanism from the release of sulfuric acid paleotemperatures of P–T seawater, range from 23 to 27 °C in beds 24
aerosol and greenhouse gases. We then provide analysis of the proposed and 25, and reach a maximum of ~ 35 °C in bed 29 (Fig. 2, Table 2). It
hypotheses for the killing mechanisms of the mass extinction event. Fi- is important to note that the high temperature (35 °C) derived from
nally, we suggest improved modeling approaches and model-data com- these conodont data as well as the sharp temperature increase both
parison to test the hypotheses invoked to explain the mass extinction postdate the end-Permian mass extinction event (bed 25) so cannot
event. be the cause of it (Fig. 2). A question remains though, whether
conodonts are reliable indicators of seawater temperature. Pucéat
2. Climate change during the Permian–Triassic transition — et al. (2010) challenged the idea that conodont is resistant to diagenetic
Geochemical proxy perspective alteration, and suggested that the assumption that conodonts incorpo-
rated oxygen isotopes into phosphate in equilibrium with their ambient
Geological evidence in support of warming immediately prior to and seawater is not well substantiated (Brand et al., 2012). Sun et al. (2012)
during the Permian–Triassic boundary mainly comes from proxy stud- built upon the methodology developed in Joachimski et al. (2012) and
ies including oxygen isotopes on well-studied stratigraphic sections extended the δ18O measurements of biogenic apatite to the entire
and paleo-pCO2 thermometers including paleosol carbonate nodules Early Triassic (Payne and Kump, 2007; Retallack, 2013; Sun et al.,
and plant stomatal indices. Indirect evidence of global warming comes 2012). The minimum δ18O value of conodonts during late Smithian indi-
from mathematical modeling based mainly on the carbon isotopes of cates a maximum sea surface temperature as high as 40 °C. Variations in
carbonate rocks. We first review proxy studies on temperature and δ18O of condonts covary with δ13C, suggesting that warming events re-
pCO2 prior to and during the extinction event, and then consider the curred into the Early Triassic (Sun et al., 2012), which could be associat-
modeling studies. ed with carbon releases and significant buildup of CO2 from Siberian
Trap eruptions or the sources of methane (Payne and Kump, 2007).
2.1. Temperature proxies Overall, temperature and CO2 proxies and modeling all suggest that
CO2 buildup and warming events occurred episodically into the Early
2.1.1. Oxygen isotope paleothermometry Triassic. The relationships between Siberian Trap volcanism, CO2 and
Temperature reconstruction is extremely important in studies of warming are tentative, however, and need to be further refined with
paleoclimatic change during critical geological transitions. The available spatially and temporally extensive proxy data collection. δ18O measure-
proxies include δ18O from bulk carbonates, or, preferentially, diageneti- ments of biogenic apatite provide important P–T temperature records,
cally unaltered brachiopod calcite, fossil conodont (mostly calcium but because the data are sparse, the reconstructed temperature increase
phosphate), vertebrate bones and clumped isotopes on unaltered could be further refined with better spatially and temporally resolved
brachiopods. Oxygen isotope paleothermometry is based on the theory proxy data collected from stratigraphically complete sections in higher
that partitioning of isotopes between two phases is a function of latitude regions.
temperature. The observational data fit the theory quite well, and
thus there has been widespread application of oxygen isotopes as a 2.1.2. Clumped isotopes
paleotemperature proxy. However, separating the contributions of tem- The analysis of clumped isotopes is a relatively new technique that
perature from variations in the water δ18O caused by ice-sheet size, local emerged less than 10 years ago. It is based on the theory and experi-
salinity variations, and long-term evolution of seawater δ18O requires ments that the proportions of the “rare–rare” bonds in minerals, such
additional constraints to uniquely interpret calcite δ18O variations in as 18O–13C in carbonate minerals, are dependent on temperature and
terms of temperature. There are also diagenetic effects, namely recrys- independent of the oxygen isotopic composition of the fluid from
tallization and secondary calcification of calcite shells, which complicate which the mineral is precipitating (Eiler, 2007). The variable Δ47 is
the use of oxygen isotopes as a paleotemperature proxy. Studies on in- defined as the measurement of the excess of mass-47 isotopologues
dividual shells of well-preserved brachiopods have been more success- relative to the abundances expected for a random distribution of all
ful than those using bulk carbonate. However, the isotope composition isotopes among all possible isotopologues. Δ47 is calculated using the
of primary shells also depends on biological “vital” effects and burial following formula:
conditions, including lithology, temperature, pore water chemistry
and burial depth.
Oxygen isotope paleothermometry has been applied to the Late !
R47
measured
Permian, but this has proven to be more problematic than applications Δ47 ¼ −1 1000 ð1Þ
to the Mesozoic or Cenozoic records due to the generally deeper burial R47
stochastic
8 Y. Cui, L.R. Kump / Earth-Science Reviews 149 (2015) 5–22
Permian Tr.
22 23 24 27 29
35 Clarkina
Hindeodus
30
T (oC)
25
20 A
43
42 Sass de Putia
41 Val Brutta
40
39
T (oC, Δ47)
38
37
36
Extinction event
35
34
33
32
B -100 cm -50 0 50
Bulla Member Tesero Member
Bellerophon Formation Werfen Formation
Changhsingian
Permian
Fig. 2. (A) Sea surface temperature (SST) reconstruction based on δ18O of conodont Clarkina and Hindeodus from GSSP Meishan section S. China; (B) SST reconstruction based on clumped
isotopes from Italy, blue dots are data from Sass de Putia and red dots are data from Val Brutta.
Panel (A) is modified from Joachimski et al. (2012). Panel (B) is modified from Brand et al. (2012).
where R47 measured is the measured abundances of mass 47 CO2, and prior to the extinction event (Fig. 2). One advantage of clumped iso-
R47
stochastic can be expressed approximately by the following equation: topes over the traditional oxygen isotopes in carbonate is that this
paleotemperature proxy is thermodynamically based and is indepen-
47 2 ½18 ½16 ½13 þ ½172 ½13 þ 2 ½18 ½17 ½12 dent of the δ18O of seawater and dissolved inorganic carbon from
Rstochastic ¼ ð2Þ which carbonate grew, and thus is suitable for interpreting the geologic
½162 ½12
record at times, such as the Late Permian, when sea water δ18O is more
where [16], [17], and [18] are the concentrations of 16O, 17O, and 18O in difficult to constrain (Brand et al., 2012). However, because of the chal-
the pool of all oxygen atoms, and [12] and [13] are the concentrations of lenging techniques involved and persistent issues including discrepan-
12
C and 13C in all carbon atoms contributing to a given population of CO2 cies between previously published synthetic carbonate calibrations
molecules (Eiler and Schauble, 2004). Δ47 can be converted to carbonate (Huntington et al., 2009; Dennis et al., 2011; Eiler, 2011), clumped iso-
growth temperature based on the following equation (Ghosh et al., topes have not been applied widely despite the unique advantages the
2006): technique provides.
different CO2 levels relative to the pre-industrial atmospheric level although a significant increase in pCO2 across the P–T boundary is con-
(PAL;~280 parts per million by volume, or ppmv). sistent with the global warming suggested by oxygen isotopes
The paleo-pCO2 proxies for times prior to the Mesozoic include δ13C (Joachimski et al., 2012).
from pedogenic carbonates (Cerling, 1991; Mora et al., 1991; Mora et al., Because of the perplexing factors that affect ambient pCO2 such as
1996; Ekart et al., 1999; Tabor et al., 2004) and trace carbonate in elevation and respiration/photosynthesis within the canopy, and differ-
goethite (Yapp and Poths, 1992, 1996), and plant stomatal index ent responses among species to changes in pCO2 (Royer, 2001), caution
(Retallack, 2001; Royer, 2001; Royer et al., 2004). In the following sec- needs to be exercised in selecting the most appropriate fossil plant for
tion, we discuss these two paleo-pCO2 approaches and their application establishing the proxy. There are significant disagreements between
to the Late Permian to Early Triassic pCO2 reconstruction. pCO2 reconstructions during the same time period at the same location
using different fossil plant species (Kürrschner et al., 1996; Retallack,
2.2.1. Plant stomatal density and index 2001). These discrepancies highlight the need to apply multiple proxies
The conductance of water vapor per unit area of leaf is regulated by in various sites across the Permian–Triassic boundary to detect pCO2
the opening of plant stomata, and is sensitive to a number of environ- level prior to, during and after end-Permian extinction event to disen-
mental factors including atmospheric pCO2 (Royer, 2001). Two mea- tangle the cause and effect of the extinction event.
sures of plant conductance have been used as proxies of atmospheric
pCO2, including stomatal density and stomatal index (SI). Stomatal den-
sity is the numbers of stomata per unit area, a feature that is established 2.2.2. Carbon isotopes of carbonate nodules in paleosol
during an early stage of leaf development (Woodward, 1987) (Eq. (4)). Cerling (1991) first proposed that the δ13C of pedogenic carbonates
Besides atmospheric pCO2, stomatal density of modern plants has been could be used to estimate the pCO2 at the time of pedogenic carbonate
shown to be sensitive to irradiance, water stress, and the position of the formation, based on the fact that the stable isotopic composition of
pore within a cell (Royer, 2001). In contrast, SI minimizes these effects soil CO2 is strongly influenced by atmospheric CO2 level. Pedogenic car-
by relating atmospheric pCO2 to the proportion of pores relative to epi- bonates in paleosols are the most widely applied atmospheric pCO2
dermal cells of fossil leaf cuticles (Royer, 2001) (Eq. (5)). proxy for pre-Cenozoic times because they are common in sedimentary
records (Cotton and Sheldon, 2012) and the presumed error for pCO2
Stomatal density ¼ number of pores per unit area ð4Þ estimate is relatively small (Royer et al., 2001). The atmospheric pCO2
level over the last 400 Myr has been estimated following the method
of Cerling (1991). One conclusion is that the inferred pCO2 levels of
stomatal density the late Paleozoic are significantly higher than today (Mora et al.,
Stomatal index ¼ 100 ð5Þ
stomatal density þ epidermal cell density 1991). However, the δ13C of pedogenic carbonate can be influenced by
a number of factors, including the soil respiration rate, porosity, δ13C
Observations and experiments indicate that plant stomatal density of respired CO2, the temperature and depth of carbonate formation,
and index decreases in response to increase in atmospheric pCO2 and the type of vegetation providing organic matter to the sediment
(Woodward, 1987; Woodward and Bazzaz, 1988). Application of SI as (Cerling, 1991). The equation based on a diffusion–reaction model
a pCO2 proxy, however, requires calibration to a SI–pCO2 relationship used to calculate the atmospheric pCO2 is as follows:
under known atmospheric pCO2. Under ideal conditions, the calibration
would be based on the same species of plant as that found in the fossil
record. However, the reconstruction of paleo-pCO2 is by necessity δ13 Cs −1:0044δ13 CΦ −4:4
based on calibration datasets from modern tree and herbaceous species Ca ¼ SðzÞ ð7Þ
δ13 Ca −δ13 Cs
that differ from their fossil counterparts. Moreover, its sensitivity dimin-
ishes as atmospheric pCO2 increases, although an upper limit for effec-
tive use of the proxy has yet to be determined (Royer, 2001). Applying where δ13Cs, δ13CΦ, and δ13Ca is the isotopic composition of soil CO2, soil
the SI paleo-pCO2 proxy, Retallack (2001) reconstructed the paleo- respired CO2, and the atmospheric CO2 respectively (Ekart et al., 1999).
pCO2 level during the past 300 million years from fossil plant leaf The term S(z) is soil respired CO2 (in units of ppmv), expressed as the
cuticles. The SI paleo-pCO2 proxy has also been applied to the Permian– difference between the actual soil CO2 and the atmosphere CO2 concen-
Triassic interval recently by Retallack et al. (2011) and Retallack (2013). tration, which is presumed to asymptote with depth when the calcite
Using an empirical relationship derived from Ginkgo leaves grown in horizon is below depth 20–30 cm (Cerling and Quade, 1993).
greenhouse experiment (Wynn, 2003; Retallack, 2009), pCO2 was calcu- Like many other proxies, assumptions and limitations exist for the
lated using the following equation: paleosol carbonate paleobarometer. Modern and recent soil studies
" # show that the isotopic composition of soil carbonate is constant below
1 ~20 cm, and thus it is essential to sample pedogenic carbonate below
pCO2 ¼ 294:1 þ : ð6Þ
4:84 10−10 SI7:93 b20 cm depth, which can be difficult to determine in paleosols that
have undergone compaction and erosion (Ekart et al., 1999). Carbon
This inverse relationship between SI and pCO2 suggests plant leaves isotope analysis on 251 Ma pedogenic carbonate from the Quartermas-
have fewer stomates when pCO2 is higher (Retallack, 2001; Botha and ter Formation (Ochoan; time equivalent to Changhsingian in strati-
Smith, 2007; Krassilov and Karasev, 2009; Retallack, 2009; Retallack graphic age) from Texas, US indicates that the atmospheric pCO2 level
et al., 2011; Retallack, 2013). We compiled SI–pCO2 proxy data for the was about 1000 ppmv (the range of pCO2 level varied from 600 to
interval of 255 Ma to 245 Ma, representing estimates of how atmo- 1500 ppmv) (Ekart et al., 1999) (Fig. 3). However, the pCO2 level prior
spheric pCO2 evolved through the Permian–Triassic transition based to the extinction is indiscernible from that during and after the event
on Retallack (2013) (Fig. 3). SI has been calculated using the fossil plants due to the low temporal resolution of paleosol sampling. There is also
Tatarina and Lepidopteris (pteridosperm genera), showing values a discrepancy with pCO2 estimate using plant stomatal indices (Royer
ranging from 4.9 to 11, which indicates pCO2 condition varying from et al., 2004) beyond the P-T interval; the pCO2 reconstructed using car-
300 to N7000 ppm, with uncertainties as high as N2000 ppm bon isotopes of pedogenic carbonate seems to be lower than that de-
(Retallack et al., 2011; Retallack, 2013). The coarse sampling of fossil rived from plant stomatal indices. Uncertainties that might lead to this
plants and uncertainties in age from terrestrial sections leads to signifi- discrepancy include assumptions regarding S(z) in Eq. (7), a factor
cant uncertainty in atmospheric pCO2 immediately prior to, during that is dependent on climate, especially mean annual precipitation
and after the end-Permian extinction event in these studies (Fig. 3), (Cotton and Sheldon, 2012), or the δ13C of soil respired CO2.
10
Table 1
Overview of the modeling boundary conditions and key results.
Model type Paleogeography Paleo-elevations Solar luminosity pCO2 Resolution Key findings Reference
(km) (W m−2) (ppmv)
Model type Paleogeography & paleobathymetry Salintiy Solar luminosity Resolution Run length pCO2 Reference
(W m−2) (years) (ppmv)
8-Level dynamical ocean model Idealized Pangean ocean; Dmax = 4 km, 34.6 1% reduced 11° × 11° horizontal 400 for exp. I, 250 for 5× PAL Kutzbach et al. (1990), Kutzbach et al.
Dmin = 0.5 km exp. II, III, and IV. (1990), Kutzbach et al. (1990),
Kutzbach
et al. (1990)
MOM linked to biogeochemical Wordian geography from Rees et al. N/A 1336 4° × 4° horizontal, 2700 2760 Hotinski et al. (2001)
model with O and P cycle (1999); 5.15 km depth flat ocean bottom 16 vertical level
3D biogeochemical model + 3- From D. B. Rowley 34.6 N/A 2.8° × 2.8° horizontal, N/A 5× PAL Zhang et al. (2001)
box 15 vertical level
model of O and P cycle
LSG/EBM + HAMOCC3 Wordian Geography from Rees et al. 35? 1336 (2.1% 3.5° × 3.5° horizontal, 2000 1×, 2×, 4× and Winguth and Maier-Reimer (2005)
(1999)? reduced 11 vertical level 8× PAL
CCSM3 From D. B. Rowley; flat bottom ocean 35 1338 3.75° × 3.75° 2700 12.7× PAL Kiehl and Shields (2005)
bathymetry horizontal, (3550)
25 vertical level
GENIE-1 Same as Kiehl and Shields (2005) 33.9 1339 36 × 36 equal area grid, 10,000 12× PAL (3360) Meyer et al. (2008)
8 vertical level
UVic ESCM and a terrestrial Same as Kiehl and Shields (2005) Present Present day 1.8° × 3.6° horizontal, 8000 to 12,000 years 300, 3000 and Montenegro et al. (2011)
carbon model day (~1365) 15 vertical level 4500
CCSM3 linked to a carbon From D. B. Rowley; flat bottom 35 1338 3.75° × 3.75° 2500 additional 12.7× PAL (3550) Winguth and Winguth (2012)
cycle model ocean bathymetry horizontal, years
25 vertical level using K&S 2005
boundary conditions
cGENIE Same as Kiehl and Shields (2005) 33.9 1339 36 X 36 equal area grid, 110,000 2800 Cui et al. (2013)
16 vertical level
Model type
The Pangea climate simulation from Kutzbach and Gallimore (1989) I. Prescribed surface forcing N/A From Kutzbach N/A Kutzbach et al. (1990)
II. Same as 1, but with increased salinity and
III and IV. Same as I but two different shapes Gallimore
for Tethys basin (1989)
Scenario 1: prescribed zonal mean forcing from Rees et al. (1999); I. High T gradient N80 (S.H.), N/A Large scale anoxia Hotinski et al. (2001)
15 (N.H.)
Scenario 2: T. from Paleocene–Eocene simulations in Bice et al. (2000) II. Low T gradient 20 (S.H.),
15 (N.H.)
Scenario 1: Prescribed T. and fresh water flux from Kutzbach and I. Thermal mode N80 (S.H.), 1.8; 0.4 No large scale anoxia Zhang et al. (2001)
Gallimore (1989),e at 5× PAL and 1% reduction in solar luminosity II. Haline mode 15 (N.H.);
20 (S.H.),
15 (N.H.)
Freshwater fluxes, momentum fluxes from Gibbs et al. (2002) and heat fluxes I. 1×, 2×, 4× and 8× PAL ~60 1.5 to 2.4 No large scale anoxia Winguth and Maier-
by EBM II. Fresh water input from Southern polar Reimer (2005)
III. CO2 increases to 18× PAL
CO2: 3550 ppmv (10× PAL); CH4: 0.7 ppmv, N2O: 0.275 ppmv 10× PAL 10 N/A Large-scale anoxia/euxinia Kiehl and Shields (2005)
11
12 Y. Cui, L.R. Kump / Earth-Science Reviews 149 (2015) 5–22
3. Climate change during the Permian–Triassic transition — Late Permian Early Triassic
10000
6000
and the model techniques advanced, increasingly sophisticated and
computationally complicated models have been applied to the end-
(252.17
Permian extinction event. We categorize the paleoclimatic models
into four types, including (1) Conceptual models and process modeling,
2
PTB (2
2000
including energy balance models (EBMs) and geochemical box models;
(2) Atmosphere general circulation models (AGCM) typically with a
highly simplified “slab ocean,” and ocean GCMs (OGCMs) with a simple
0
atmosphere; (3) Earth system models of intermediate complexity
255 254 253 252 251 250 249 248 247 246 245
(EMICs); and (4) Fully coupled atmosphere–ocean general circulation
models (AOGCMs). These models vary in complexity and sophistication, Age (Ma)
and they represent the processes being parameterized based on the un-
Fig. 3. Atmospheric pCO2 reconstruction based on two proxies from 255 to 245 Ma. Red
derstanding of the climate system at the time the studies were being
circle is pCO2 value calculated based on δ13C of pedogenic carbonate (modified from
conducted. Most studies that we review included sensitivity analyses, Ekart et al., 1999) and blue squares are pCO2 reconstruction based on plant stomatal
in which key parameters were varied to test the response of the models index (modified from Retallack, 2013). The error bars represent upper and lower limit
to changes in these variables. of the estimated pCO2 using both methods. Blue shadow indicates pCO2 range estimated
in GEOCARB III.
We begin by summarizing the earlier Pangean climate studies,
which focused on the effect of large continents on monsoons. We then
summarize the climate modeling studies chronologically, each driven
by a specific hypothesis related to the end-Permian mass extinction include paleogeography (idealized Pangea and Pangea A and B), topog-
event. The hypotheses being explored to explain the extinction event raphy, atmospheric pCO2, solar luminosity, and distributions of extend-
are constantly being refined (if not refuted) with the appearance of ed lakes or seas within the supercontinent (Table 1 and references
new geological evidence as new outcrops, cores or techniques become therein). One common finding from these models was extreme seasonal
available. We do not attempt to review all the paleoclimatic studies temperature contrast with a hot summer (N 40–50 °C) and a cold winter
during the Permian–Triassic transition, but rather provide an overview (b− 15 °C) within the large continent during the Late Permian and a
of how our quantitative understanding of the Permian–Triassic climate strong monsoon circulation characterized by high precipitation along
system has progressed over the last two decades. the Tethys coast and arid continent interior. These simulations were
Key questions of interest in this review include: (1) How did the broadly consistent with the distribution of climate indicators. However,
global mean temperature and the equator–pole temperature gradient they failed to simulate high latitude warmth that was indicated by well-
respond to this proposed greenhouse gas-driven climate event? preserved forests from the Upper Permian of Antarctica (Taylor et al.,
(2) How sensitive are the climate modeling studies to uncertainties of 1992, 2000).
greenhouse gas concentrations, paleogeography, paleobathymetry and
topography? (3) How sensitive was the Late Permian climate to changes
in orbital configuration? (4) What was the distribution of vegetation, 3.2. Geochemical box models used to investigate hypotheses associated
and how important was this feedback to Late Permian climate? with the end-Permian extinction event
3.1. Conceptual models and EBMs It has been widely accepted that there was a negative carbon isotope
excursion (CIE) in the global ocean–atmosphere system, as recorded in
Early climate modeling study of the Late Permian was focused on both marine carbonate/organic matter and terrestrial organic material.
the effect of the Pangea supercontinent, a landmass distribution funda- The CIE recorded in marine carbonate shows a long-term, slowly
mentally different than today, using conceptual models (Parrish, 1993), decreasing trend, followed by a sharp drop (about 4‰ on average;
energy balance models (EBMs) (Crowley et al., 1987, 1989) and atmo- see Korte and Kozur, 2010 for review) immediately at the extinction
spheric general circulation models (Kutzbach and Gallimore, 1989; horizon (e.g. Zhang et al., 2005) and a slight increase after the main
Kutzbach et al., 1990; Kutzbach and Ziegler, 1993; Kutzbach, 1994). extinction, whereas the terrestrial CIE is complicated by the organic C
Conceptual or modern-analog modeling studies have been conducted source variation. Absolute age determinations suggest that the duration
by Nairn and Smithwick (1976), Parrish (1982) and Parrish et al. (1986). of the CIE (contemporaneous with the main extinction event) was less
One common result from these modern-analog models is that strong than 165 kyr (Bowring et al., 1998). Newly published U–Pb zircon
seasonality exists during the Late Permian due to the amalgamation of ages indicate that the sharp decline of δ13C occurred within about
Pangea. A strong monsoonal circulation is also the expected consequence 20 kyr (Shen, et al., 2011; Burgess et al., 2014). One commonly pro-
of the large land mass. posed interpretation of the rapid, large magnitude CIE is the release of
EBMs estimate the changes in the climate system using the energy a large amount of 13C-depleted carbon, either in the form of oxidized
budget of the Earth; they determine global temperature based on a bal- coal or organic matter (CO2 with δ13C of about − 25‰), biogenic
ance between the net absorbed solar radiation and the net emitted ter- methane clathrate and permafrost on the seafloor (with δ13C of about
restrial radiation (Budyko, 1969; Sellers, 1969). They do not include b−60‰), or thermogenic methane (with δ13C of about −40‰) associ-
ocean circulation, topography, vegetation and other important variables ated with dike and sill intrusion of coal during the Siberian Trap
(Crowley and North, 1991) which limit their application in realistic emplacement. To test the 13C-depleted carbon source, geochemical
paleoclimatic studies. However, due to their simplicity and fast compu- box models have been widely applied. Box models involve solving a
tational speed, and the general lack of spatially resolved observations series of ordinary differential equations associated with the carbon cycle
to justify more sophisticated models, EBMs were applied to the Late (e.g., Slingerland and Kump, 2011) reflecting the time-dependent
Permian to investigate the temperatures of Pangea (Crowley et al., mass balance equations for some combination of total carbon, alkalinity,
1989). Key forcing factors that were investigated in these models phosphorus, oxygen and carbon isotopes.
Y. Cui, L.R. Kump / Earth-Science Reviews 149 (2015) 5–22 13
10000
Berner (2002) utilized a version of the “GEOCARB” carbon cycle Rampino and Caldeira (2005)
model based on Beerling and Berner (2002) to evaluate several hypoth- Grard et al. (2005)
Table 2
Comparison of sea surface temperature estimates based on proxy and pCO2 based on modeling across the Permian–Triassic boundary. The inferred climate sensitivity is also shown.
(a). Sea surface temperature estimates based on proxy
Amount of doubling Reference Model pCO2 pre- pCO2 peak- ΔpCO2 Climate sensitivity given 8 °C
pCO2a method extinction extinction warming
(ppm) (ppm)
In contrast, Payne and Kump (2007) demonstrated, again using a the conceptual models of the Pangean climate, suggesting strong
simple box model, that Siberian Trap volcanism could have been re- monsoonal circulation, equatorial and continental interior aridity
sponsible for the large carbon isotope fluctuations observed in the latest and extreme seasonal temperature variations. Their model results
Permian and Early Triassic, but that other factors, including oxidation of highlighted the importance of geologic boundary conditions, including
crustal carbon, also likely played a role. The box model they used was geography, topography and atmospheric greenhouse gas composition
similar to that of Grard et al. (2005), but explicitly balanced carbonate (key boundary conditions are summarized in Table 1).
alkalinity, atmospheric oxygen and phosphate. Carbonate alkalinity In order to improve on the previous study of Kutzbach and Gallimore
was delivered to the ocean from carbonate and silicate weathering (1989), Kutzbach and Ziegler (1993) simulated the Late Permian
and removed from the ocean through carbonate burial. Their model uti- climate using the community climate model version 1 (CCM1) of the
lized a somewhat smaller equilibrium atmospheric pCO2 level National Center for Atmospheric Research (NCAR). This model also in-
(1500 ppmv) following Berner and Kothavala (2001), compared to cluded a 50 m thick mixed layer ocean with a prescribed poleward
that of Grard et al. (2005) (~ 3000 ppm) (Fig. 4). They tested the transport of heat by the oceans. The advantages of this model in com-
model using three eruption scenarios, namely release of 3 × 1018 mol parison to Kutzbach and Gallimore (1989) were that the horizontal
of volcanogenic CO2 (with δ13C = − 5‰) over timescales of 100 kyr, and vertical resolution was higher (12 levels atmosphere), and there
300 kyr and 600 kyr respectively. Any single source, from methane were improvements in parameterizations of radiation, clouds, snow
clathrate, oxidized organic carbon or volcanic carbon source alone failed cover, and vertical diffusion. The boundary conditions being prescribed
to account for the series of observed CIEs. Several pulses of CO2 release in their model included (1) land–ocean distribution and topography
from volcanic eruption or organic carbon oxidation were required in from the maps of Ziegler (1990), showing a relatively coarse resolution,
their model to account for the Early Triassic large fluctuations of δ13C a maximum elevation of 2300 m, and an average of 480 m (4.4° latitude
records. × 7.5° longitude model grid); (2) a prescribed land surface albedo,
ranging from 0.13 to 0.24 using the vegetation biome maps for the
3.3. Early general circulation models (GCMs) Kazanian (Ziegler, 1990); (3) a prescribed poleward zonal average
ocean heat flux; (4) 1% reduced solar irradiance (1356 W m− 2 vs.
Unlike EBMs or geochemical box modeling, general circulation modern value at 1370 W m−2), modern axial tilt (23.5°) and a circular
models (GCMs) are spatially resolved and include calculations of the orbital; and (5) atmospheric pCO2 prescribed at 5 × PAL (Table 1).
important state variables at each grid cell using partial differential equa- Kutzbach and Gallimore (1989) tested the effect of the inclusion of
tions solving the equations of motion of the atmosphere (AGCMs) and extensive inland seas and lakes, and the results indicated that these
oceans (OGCMs). The first climate models that were applied to the lakes make a considerable difference in local temperature because of
Permian climate problem were uncoupled GCMs of two sorts. One the differences in low-level heating of the atmosphere and the damping
was a model with a 3D atmosphere and a simple 2D “surface slab” of local seasonal temperature range. Modeled climate and biome distri-
ocean without ocean circulation (e.g. Kutzbach and Gallimore, 1989; bution agreed well with the observed Late Permian distribution. How-
Gibbs et al., 2002), in which the temperature of the ocean was either ever, there were some important limitations of this type of AGCMs,
prescribed or allowed to adjust through energy exchanges with the at- including the coarse modeling resolution, the oversimplification of
mosphere. The other was a model with a 3D ocean and no atmosphere treatment of the ocean and the lack of an interactive vegetation model
(e.g. Hotinski et al., 2001; Zhang et al., 2001), in which the ocean model that might bias the regional climate. Another key uncertainty was
was forced with prescribed winds, surface temperature and salinity. topography, which had been demonstrated to be a critical parameter
The first general circulation model used to simulate the idealized as shown by the importance of the uplift of Tibetan plateau on global cli-
Pangean climate was a low-resolution, global GCM coupled to a 50 m mate and Asian monsoon (Kutzbach et al., 1989, 1993; Ramstein et al.,
deep mixed-layer ocean using idealized land–ocean distribution and 1997).
no topography (Kutzbach and Gallimore, 1989). The authors conducted To understand the patterns of Early Triassic ocean circulation, tem-
four suites of sensitivity experiments, in which two different mean ele- perature and salinity, a dynamical ocean model with eight vertical levels
vations (0 m vs. 1000 m), two different snowcovers, two atmospheric driven by prescribed surface forcings from Kutzbach and Gallimore
pCO2 levels (1 × PAL vs. 5 × PAL) and two solar luminosities (modern (1989) was run by Kutzbach et al. (1990). Their idealized Pangean
vs. 1% decreased) were used. The simulations performed confirmed ocean model had a horizontal resolution of 5° latitude × 5° longitude
Y. Cui, L.R. Kump / Earth-Science Reviews 149 (2015) 5–22 15
and 4 km ocean depth. Four experiments were conducted to investigate land–sea distribution and the paleoelevations of mountain belts were
the ocean temperature distribution and circulation response to varia- tested in a number of sensitivity analyses. Their results indicated that
tions in the depth and salinity of the Tethys Ocean. The model results the Pangea B configuration seems to be more consistent with the Late
indicated a much warmer high-latitude surface and deep-water Permian climatic status.
temperature in comparison to the present, subtropical anticyclonic
(~ 30° latitude) and subpolar cyclonic (~ 60°) latitude gyres in the 3.4. Ocean general circulation models (OGCMs) and the oceanic
Pathalassic ocean, eastward and poleward flowing equatorial currents anoxia hypothesis
and a relatively cool and fresh polar water and warmer and saline
Tethys. Unfortunately, there was little, if any geological data that could Ocean general circulation models coupled to a simple atmosphere
be directly compared with the model results. With the exception of with biogeochemistry components have been used to test the following
deep sea sediments of Permian–Triassic age in Japanese and North two hypotheses invoked to explain the end-Permian extinction event:
American subduction–accretion complexes (Isozaki et al., 1990), no (1) anoxic water propagated into shelf regions and epicontinental seas
deep-sea sediments of Late Permian age remain. Several other limitations and caused the extinction (Wignall and Twichett, 1996); and (2) cata-
to OGCM used by Kutzbach et al. (1990) include its coarse resolution, the strophic release of deep ocean CO2 upon invigorated ocean circulation
idealized Pangean geography and ocean bathymetry, its short integration and upwelling resulted in CO2 toxicity (hypercapnia) and led to the
time, prescribed surface forcing from previous OGCM simulations, and extinction (Knoll et al., 1996). Both hypotheses presume a causal
the possibility that dynamic coupling between ocean and atmosphere relationship between ocean stagnation and the development of large
might lead to quite different circulations for the Pangean ocean. chemical gradients, a relationship that was intuitive but when subject
The Wordian (ca. 265 Ma) climate was simulated by Gibbs et al. to quantitative evaluation using numerical models, proved largely
(2002) using the AGCM GENESIS2 (Global Environmental and Ecologi- incorrect.
cal Simulation Interactive Systems version 2; Thompson and Pollard, The ocean anoxia hypothesis was first proposed about 20 years ago
1997), and the results compared with the geographic distribution of by Wignall and Hallam (1992), who found syngenetic pyrite associated
climate sensitive sediments, including coals, evaporites, eolian sands, with the maximum flooding surface at the P–T boundary in sediments
carbonate, organic-rich shales, dropstones, tillites and phosphorites. from a couple of locations. This finding led to the proposition that anoxia
GENESIS2 consisted of an 18 level atmosphere with a land-surface might have caused the end-Permian mass extinction event. Wignall and
model and a simple 50 m slab ocean, and had a horizontal resolution Twichett (1996) later extended the evidence of anoxia to a more global
of 3.75° latitude × 3.75° longitude. Like other A/OGCMs, boundary con- extent and into shallow water. Supporting evidence of global deep
ditions needed to be specified for GENESIS2 (Table 1), including (1) the ocean anoxia came from a number of studies from the obducted accre-
Ziegler et al. (1997) land–sea distribution paleogeography and topogra- tionary complexes in Japan, which were interpreted to represent pelagic
phy map; (2) a 2.1% decrease in solar luminosity (1336 W m− 2 vs. sediments deposited in an anoxic, deep Panthalassic ocean (Isozaki
1365 W m−2 used in GENESIS2); (3) a uniformly distributed vegetation et al., 1990; Kajiwara et al., 1994; Isozaki, 1997). This proposal of a
consisting of mixed tree, savanna and grassland; (4) atmospheric pCO2 prolonged “superanoxic” event that significantly preceded the Perm-
values of 4 × and 8 × present levels (345 ppmv was the assigned ian–Triassic boundary and continued into the Early Triassic prompted
“present level”); and (5) a modern axial tilt (23.5°) and a circular orbit- a number of studies, both modeling and data gathering, aimed at evalu-
al, similar to that in Kutzbach and Gallimore (1989), although two ating the extent and duration of anoxia and the mechanisms that might
extreme values for the Pleistocene from Berger (1978) were adopted have generated and sustained it.
to investigate the possibility of ice-sheet reappearance, including a In order to investigate whether warming the poles would induce
warm summer orbit (eccentricity = 0.06 and obliquity = 24.5°) and a ocean stagnation and if circulation changes would result in anoxia
cold summer orbit (eccentricity = 0.06 and obliquity = 22.0°). Similar and high CO2 in the Late Permian deep ocean, Hotinski et al. (2001)
to the findings in previous modeling studies, their model simulated conducted numerical experiments using a 3-D OGCM, the Geophysical
extreme seasonality and aridity in the continental interior during Fluid Dynamic Laboratory's Modular Ocean Model (MOM version 2),
the Wordian. The northward movement of land from Sakmarian connected to a biogeochemical model of phosphate and oxygen cycling.
(ca. 280 Ma) to Wordian might have led to a cool high latitude winter, The model had 4° × 4° horizontal resolution, 16 vertical level and fixed
stronger winter westerlies and a southward shift of the storm-track pre- values of horizontal and vertical mixing coefficients. Hotinski et al.
cipitation. The model simulations were in generally good agreement (2001) ran the model with two climatic forcings, including a high-
with the climate sensitive sediments, except that the simulated gradient case which used sea surface temperatures derived from 8 ×
extremely cold high-latitude temperatures were inconsistent with the PAL GENESIS simulation for the Late Permian (Gibbs et al., 2002), and
observations, a situation that was attributed to either an incorrect a low-gradient, zonally averaged case using the reduced-temperature
interpretation of the paleoclimate data or limitations of the model gradient field from the Paleocene–Eocene Thermal Maximum (a period
(Gibbs et al., 2002). with extreme polar warmth) simulation by Bice et al. (2000). The high
As a key boundary condition, the configuration of Pangea has been pole-to-equator gradient (−2–33 °C) case corresponded to a relatively
strongly debated for decades (Van der Voo and French, 1974; Irving, cool polar ocean (b 0 to 5 °C), and the low pole-to-equator gradient
1977; Livermore et al., 1986; Torcq et al., 1997). Two paleogeographic (12–28 °C) experiment showed a warmer polar ocean (12 °C), which
configurations were proffered: Pangea A and Pangea B, and the differ- was similar to that during the Paleocene–Eocene and more consistent
ences between them were mainly in the different relative position of with geological observations (Ziegler, 1990; Yemane, 1993). A vigorous
Gondwana and Laurussia. Fluteau et al. (2001) conducted a number of ocean circulation and an oxic ocean was observed in their high-gradient
climate modeling experiments to help resolve this debate, using a 3D experiment, whereas a substantially decreased pole to equator surface
LMD 5.3 (Laboratoire de Météorologie Dynamique version 5.3) AGCM density gradient was found in the low-gradient experiment. This
based on that presented in Harzallah and Sadourny (1995), coupled to suggested that warming the poles slows ocean overturning (although
a soil vegetation model. This model had a mid-latitude horizontal reso- the circulation does not stagnate), and oxygen values are significantly
lution of 400 × 400 km and 11 vertical levels including 8 in troposphere reduced. However, the low-gradient model result did not support lethal
and 3 in stratosphere. The boundary conditions included a prescribed build up of H2S and CO2 in the deep ocean without invoking a large
3 × PAL atmospheric pCO2 (~ 900 ppm), a similar-to-present pole increase in oceanic nutrient level.
(0 °C) to equator (25 °C) temperature gradient, and present values of At the same time, two circulation scenarios were being tested by
orbital parameters and solar constant (1365 W m− 2) (Table 1). The Zhang et al. (2001), who used the MIT 3-D ocean circulation model
paleogeographic configuration including Pangea A and Pangea B, the and a box model of P and O cycles to investigate the issue of anoxia in
16 Y. Cui, L.R. Kump / Earth-Science Reviews 149 (2015) 5–22
deep ocean. The two experiments included a vigorous “thermal mode” atmosphere model simulations become computationally prohibitive.
driven by cooling in southern polar latitudes (sinking in high latitudes, The EMICs used to investigate the late Permian climate systems include
“weak” meridional temperature gradients) and a weaker “haline GENIE (Meyer et al., 2008; Cui et al., 2013), LSG/EBM (large-scale geo-
mode” driven by an enhanced hydrological cycle (in other words: strophic ocean general circulation model and coupled atmospheric en-
evaporation from the subtropics) and/or weakened diapycnal mixing ergy balance model; Winguth and Maier-Reimer, 2005), PLASIM
in the ocean. The boundary conditions used in these two scenarios (PlanetSimulator Model; Roscher et al., 2011) and the UVic ESCM (Uni-
were otherwise similar: surface temperatures were prescribed, and versity of Victoria Earth System Climate Model; Montenegro et al.,
the mean salinity of the ocean was assumed to be 34.6 (Table 1). The 2011). The first two EMICs were used to investigate carbon cycle pertur-
“thermal” mode had weaker surface wind stress and weaker tempera- bations and oceanic chemistry changes, such as oxygen or H2S re-
ture in comparison to their “haline” mode. Zhang et al. (2001) did not sponses to increased nutrient contents and pCO2 levels. The later two
get any significant reduction in deep ocean oxygen in the thermal EMICS included a biome module; Montenegro et al. (2011) emphasized
mode, whereas deep-sea anoxia was observed in the haline mode the sensitivity of ocean circulation and oxygen levels to uncertainties in
with enhanced hydrological cycle, which is an expected consequence paleogeography and paleobathymetry, and Roscher et al. (2011) fo-
of the projected global warming (IPCC, 2007). cused on estimating climate sensitivity during global warming versus
Differences between the Hotinski et al. (2001) and the Zhang et al. global cooling.
(2001) studies prompted a discussion about their somewhat different
definitions of “weak” and “reduced” meridional temperature gradients 3.5.1. EMICs
(Hotinski et al., 2002; Zhang et al., 2003), in which Hotinski et al. Winguth and Maier-Reimer (2005) (hereafter W&M) used the
(2002) argued that the “weak” T gradient in Zhang et al. (2001) was Hamburg large-scale geostrophic ocean GCM coupled to an atmospheric
similar to the “high” T gradient case in Hotinski et al. (2001). The “weak- energy balance model (LSG/EBM) to investigate the sensitivity of
er” meridional temperature gradient in Zhang et al. (2001) was based ocean's thermohaline circulation to a massive methane release and
on an atmospheric GCM of Kutzbach and Gallimore (1989), where the ensuing global warming. In order to study the response of export pro-
high-latitude average surface temperature was less than 0 °C at 70° duction and oxygen distribution to ocean circulation changes, W&M
South. The “reduced” pole–equator temperature gradients in Hotinski also incorporated a simplified marine carbon cycle component known
et al. (2001) were based on a Paleocene–Eocene Thermal Maximum as the Hamburg model of ocean carbon cycle (HAMOCC3). The W&M
modeling study (Bice et al., 2000) indicating a high latitude temperature study was largely a continuation of Winguth et al. (2002) with the
about 12 °C. These different definitions of “low” and “weak” gradient addition of a carbon cycle model and additional sensitivity analyses by
led to different rates of meridional overturning (zonally integrated varying boundary conditions. The boundary conditions of the W&M
mass transport in the north–south plane) and thus differing rates of model were similar to that in Winguth et al. (2002), examining the ef-
ventilation of the deep ocean with oxygen. It is important to note that fects of changes in topography in the polar regions on ocean circulation
in contrast to Zhang et al. (2001), negative oxygen levels were allowed during the middle Permian. W&M first explored the model sensitivity of
in the Hotinski et al. (2001) model under the assumption that they ocean circulation in response to changes in thermal and haline bound-
represent organic matter decomposition using alternative oxidants ary conditions by adding a southern hemisphere freshwater source of
(nitrate or sulfate) when oxygen is depleted; Hotinski et al. (2002) 2 Sv (Sverdrups, 106 m3 s−1) and doubled the pCO2 from 4 × PAL to
argued that the resetting of negative oxygen values to zero in the 8× PAL. In their baseline 4× PAL experiment, the deep sea was well ox-
Zhang et al. (2001) model led to an artificial oxygenation of the deep ygenated and lowest oxygen concentration was seen above regions
ocean. with high export production, which was found in eastern Panthalassic
Kump et al. (2005) used a simple two-box ocean model coupled to a Ocean and east of S. China due to the upwelling of nutrient-rich bottom
1-D atmospheric photochemical model to explore the consequence of water. A prominent oxygen minimum zone (OMZ) was observed
chemocline upward excursion, a process that is triggered by the desta- around ~ 1000 m in the eastern equatorial Panthalassa ([O2] as low as
bilization of the chemocline (the transition from oxygenated surface 20 μmol kg−1). Simply doubling the pCO2 from 4× PAL to 8× PAL did
waters to anoxic and sulfidic deepwaters) by sulfide accumulation in not change the results significantly due to the similar poleward
the deep sea under severe conditions of eutrophication. Their 1-D atmo- heat transport and circulation strength in the deep sea. By inducing
spheric photochemical modeling suggested that large fluxes of H2S freshwater input in the southern hemisphere and doubling the pCO2,
(≥ 2000 × present levels) to the atmosphere might have led to lethal the oxygen-depleted area expanded horizontally in response to a re-
levels of H2S in the atmosphere, the destruction of the ozone layer, duced circulation in the tropical region. This supported the idea that a
and increased atmospheric methane levels, providing a killing mecha- “more sluggish circulation may lead to extended anoxic conditions in
nism for the end-Permian extinction event in both marine and terrestri- the deep-sea”, although the [O2] simulated in W&M was much higher
al realms. A key finding in Kump et al. (2005) was that in regions than that in the low-gradient scenario of Hotinski et al. (2001) and
with elevated upwelling rate, chemocline upward excursion could be the haline mode of Zhang et al. (2001). Doubling the atmospheric
reached when H2S was high (~1 mmol kg−1) even under modern atmo- pCO2 and introducing freshwater input did change the oxygen distribu-
spheric pO2 level. Lower levels of H2S were required for chemocline tions in the deep ocean, but these changes were insufficient to generate
upward excursion under lower levels of pO2, such as the proposed low widespread anoxia. W&M also did sensitivity analyses by reducing the
pO2 atmosphere in the Late Permian (Berner, 2006a). export production and adding methane-derived carbon to the atmo-
Subsequent work showed that the high H2S flux envisioned by sphere. They reduced the export production by 10% and found that
Kump et al. (2005) was likely insufficient to cause toxic H2S levels in the oxygen content increased in response to a reduction of organic mat-
the troposphere and ozone destruction in the stratosphere (Beerling ter remineralization. Two methane release scenarios were proposed
et al., 2007), unless a significant CH4 flux accompanied the high H2S with the amount of carbon released of 7500 Gt C and 4200 Gt C to inves-
flux (Lamarque et al., 2007). tigate the ocean circulation changes in response to massive carbon re-
lease. They also allowed an atmospheric pCO2 increase to 18 × PAL in
3.5. Earth system models of intermediate complexity (EMICs) and fully response to an instant 7500 Gt C release (methane converted instanta-
coupled atmosphere–ocean general circulation models (AOGCMs) neously to CO2). Increased atmospheric pCO2 induced an increase in
deep-sea temperature, but did not result in a sluggish thermohaline
EMICs are particularly useful for investigating long term, large-scale circulation or widespread deep-sea anoxia. Finally W&M compared
climate changes driven by forcings with durations of millennia or longer their modeling results from three simulations, including the 4 × PAL
(IPCC, 2007), timescales for which high-resolution coupled ocean– baseline experiment, the 8× PAL with southern hemisphere freshwater
Y. Cui, L.R. Kump / Earth-Science Reviews 149 (2015) 5–22 17
input experiment, and the 4× PAL along with massive methane release of euxinia (as much as 25% decrease in surface [H2S]). They proposed
(18× PAL) experiment to climate-sensitive sediments and the recon- that the presence of H2S in the surface ocean due to increased phosphate
structed water mass from Ziegler et al. (1997). The water mass in concentration (N3× modern levels) could lead to the extreme euxinia
Ziegler et al. (1997) was classified in eight categories, including glacial, that caused the mass extinction in the end-Permian given the fact that
cold temperate, wet temperate, temperate, upwelling, dry subtropical, H2S is fatal to most marine eukaryotes. The spatial distribution of
tropical, and wet tropical (also see Winguth et al., 2002). All the exper- surface-water euxinia in the Meyer et al. (2008) model was in broad
iments seemed broadly consistent with the observed geological data, agreement with geological evidence, including the known occurrences
while the 8 × PAL simulation generated tropical climate zones that of the green-sulfur phototroph biomarker isorenieratane (Grice et al.,
were more consistent with the reconstructed water masses between 2005) and framboidal pyrite size distribution (Wilkin et al., 1996; Bond
30°N and 30°S. and Wignall, 2010).
Montenegro et al. (2011) evaluated oceanic anoxia and ocean acidi- While H2S provides a potent killing mechanism for the end-Permian
fication as the killing mechanisms for the end-Permian mass extinction, mass extinction event, the degree of euxinia due to phosphate buildup is
using the UVic ESCM (University of Victoria Earth System Climate uncertain due to the uncertainty about phosphate inventory. Winguth
Model), which consisted of an energy-moisture balance mode, the and Winguth (2012) proposed that only a small amount of phosphate
MOM ocean model, a terrestrial carbon cycle, model and an ocean increase likely occurred prior to the end-Permian extinction, invoking
carbon cycle component. The model had a horizontal resolution of the arguments of Algeo and Twitchett (2010) that weathering intensity
1.8° × 3.6° and 15 vertical levels in the ocean. The boundary conditions, (interpreted as increased riverine phosphate delivery) did not increase
including paleogeography, topography and winds were adopted from until the Early Triassic. Moreover, the evidence for global deepwater
Kiehl and Shields (2005) (see below). Present-day solar insolation, or- anoxia, a feature of the Meyer et al. (2008) simulations above 3 ×
bital parameters, atmospheric albedo and zonally averaged atmosphere modern [PO4], has been challenged recently by studies in central Japan
diffusion were used in the model. Nine experiments were conducted to (e.g. Algeo et al., 2010), indicating an expansion of the oxygen minimum
investigate the effect of changing atmospheric pCO2 (300, 3000 and zone in Panthalassa but no deep basinal anoxia. Other studies in S. China
4500 ppmv), paleogeography and bathymetry (Table 1). Sensitivity also indicate that the anoxia occurred time-equivalent or slightly prior
analyses were performed to understand the impact on ocean circulation to the end-Permian extinction event (Brennecka et al., 2011), and did
of the opening and closing of the seaway between Angara and not last millions of years as suggested by Isozaki (1997).
Euroamerica. Three different ocean bathymetry configurations were Instead of investigating the extent of euxinia in the ocean, Cui et al.
evaluated, including a flat bottom with fixed depth at 4 km, arbitrarily (2013) focused on the paleoclimatic consequences of the addition of
distributed meridional ridges, and the best guess of realistic ridge three distinctive CO2 sources, including the release of methane hydrate
distribution. The modeled temperature and precipitation patterns and and thermogenic methane, and the oxidation of organic matter. The
biome distributions were in general agreement with the observed sed- GENIE model simulations were forced by δ13C of atmospheric CO2
imentary records and paleontological data (Montenegro et al., 2011 and based on the Meishan record. The rate and total amount of carbon
reference therein). The ocean saturation state dropped dramatically being released was calculated and compared with modern fossil fuel
(undersaturated with respect to calcite) in response to an increase in burning. Their favored scenario, the oxidation of fossil organic matter,
pCO2, but the widespread anoxia was not observed in the model when shows a maximum rate of C release six times smaller than the modern
ocean bottom topography and paleogeography were changed. fossil fuel burning rate, the total amount of C much larger and the dura-
Meyer et al. (2008) used GENIE-1 (Grid Enabled Integrated Earth tion of emission more protracted than those estimated for the current
system model, version 1) to examine the distribution of euxinia in the fossil fuel reservoir. The climatic consequence is that global temperature
Latest Permian ocean. GENIE is an EMIC, with 36 × 36 equal area rises up to 6 °C globally, similar to that estimated by Joachimski et al.
horizontal grids and 8 or 16 vertical levels, a frictional–geostrophic (2012) and Sun et al. (2012). The atmospheric pCO2 almost triples at
ocean model and a coarse-resolution 2-D energy moisture balance the peak of the event from 2800 ppm to about 8000 ppm (Fig. 4),
atmospheric model (Edwards and Marsh, 2005). The key tracers incor- while the peak value falls in the range of previous modeling results
porated in the Meyer et al. (2008) version of GENIE included phosphate (e.g. Berner, 2002; Grard et al., 2005; Rampino and Caldeira, 2005)
(PO4), dissolved inorganic carbon (DIC), alkalinity (ALK), sulfate (SO2−
4 ), (also see Table 2).
and hydrogen sulfide (H2S). In Meyer et al. (2008), the atmosphere
model in GENIE-1 was forced by the wind stress fields equilibrated at 3.5.2. AOGCMs
12× PAL (pre-industrial atmospheric level) pCO2 from the fully coupled The fully coupled AOGCM that had been applied in the latest
modeling study of Kiehl and Shields (2005), and the same paleogeogra- Permian is CCSM3 (Community Climate System Model, Version 3.0;
phy and paleobathymetry as in Kiehl and Shields (2005) (Table 1). The Kiehl and Shields, 2005; Winguth and Winguth, 2012), which includes
atmospheric O2 was set at the modern value (~0.2095 atm), similar to a 3-D ocean coupled to a 3-D atmosphere, and land and sea-ice using re-
most other Permian model studies. It has been proposed that the atmo- alistic geography and topography (Table 1). Kiehl and Shields (2005)
spheric oxygen levels during the Late Permian were on its minimum in used ideal water-mass age, which is an indicator of ocean mixing effi-
the Phanerozoic (Berner, 2005) (possibly as low as 15%), so the estimates ciency, to infer deep ocean oxygenation because the model did not in-
of oceanic anoxia extent in the modeling study of Meyer et al. (2008) are clude oxygen as an explicit tracer. The larger ideal age indicated less
possibly a minimum. Meyer et al. (2008) conducted two series of simu- efficient ocean mixing, hence less oxygen. The model simulation sug-
lations to investigate the oceanic H2S buildup in response to increased gested low oxygen levels at depth in the tropical Panthalassic Ocean,
phosphate supply (from 1 × to 10 × modern phosphate content) and which led the authors to conclude that the simulation was in support
the metabolic effect of sulfur phototrophs on the surface ocean chemis- of the observed geological data (Wang et al., 1994; Isozaki, 1997). The
try. Consistent with most other studies, at modern phosphate levels warm, less-dense water at high latitudes retarded deep-water forma-
the high global average surface temperature does not result in wide- tion and in turn reduced the efficiency of surface oxygen transport to
spread anoxia. However a strong oxygen minimum zone (OMZ) was the deep ocean, presumably resulting in widespread anoxia. The simu-
observed in the Paleo-Tethys Ocean, while the Panthalassic Ocean re- lated extreme daily temperatures in the dry subtropical regions were
mains well oxygenated ([O2] N ~ 150 μmol kg−1). Meyer et al. (2008) consistent with some studies suggesting warming as a killing mecha-
found that as PO3−4 concentration increased, the extent of anoxia and ul- nism for terrestrial life during the P–T transition. Their ocean model in-
timately euxinia increased. Upwelling region in Paleo-Tethys formed a dicated warmth over the high-latitude Panthalassic Ocean (zonal mean
“nutrient trap” that enhanced sulfide accumulation. However, including sea surface temperature up to 8 °C) and at depth (up to 5 °C at 3 km),
sulfur phototrophy in their experiment significantly reduced the extent indicating a weak pole–equator temperature gradient in the ocean.
18 Y. Cui, L.R. Kump / Earth-Science Reviews 149 (2015) 5–22
Table 3
Summary of Siberian Trap volcanism and its environmental consequences.
Additionally, their simulated ocean circulation pattern suggested a modern phosphate level in combination with a high Fe dust supply
weak overturning circulation (about 10 Sv) in the Latest Permian (10× modern Fe supply, so no Fe limitation) didn't change the O2 distri-
ocean, and an extremely stratified Tethys sea. bution significantly relative to the one without elevated Fe levels, indi-
In order to further test the hypothesis of widespread anoxia due to cating that export production wasn't severely limited by Fe in their 1×
elevated pCO2 and nutrient supply, Winguth and Winguth (2012) con- PO4 experiment. However, with a more deeply penetrative biological
tinued the CCSM study of Kiehl and Shields (2005) by adding a carbon pump in the 10× modern phosphate experiments, widespread anoxic
cycle model to their existing 2700 year simulation. The carbon cycle conditions were observed throughout the central deep Panthalassa
model they used tracks seven variables that are transported in the (3800 m depth), all deep Tethys Oceans, and eastern Boreal Ocean. Un-
ocean, including phosphate (PO4), total dissolved inorganic iron (Fe), fortunately, Winguth and Winguth (2012) didn't perform the experi-
dissolved organic phosphorus and iron, dissolved inorganic carbon ment that would have been most like the 10× PO4 case of Meyer et al.
(DIC), total alkalinity, and dissolved oxygen (O2). They use boundary (2008), i.e., 10 × Fe and 10 × PO4. Winguth and Winguth (2012) con-
and initial conditions from Kiehl and Shields (2005) to initiate and cluded that the end-Permian ocean was better represented by the ex-
integrate the model for an additional 2500 years, allowing the tracer periments with modern or a small increase in phosphate level, with
distributions to reach equilibrium. They conducted 3 series of simula- the elevated nutrient delivery that caused anoxic conditions confined
tions to investigate the anoxia distribution due to elevated nutrient or to the Early Triassic due to strong weathering in response to terrestrial
dust input to the ocean and intensification of biological pump. ecosystem destruction. Because of this, they suggested anoxia was not
Similar to previous studies (Kiehl and Shields, 2005; Meyer et al., widespread enough to have caused the end-Permian mass extinction
2008), at 1 × modern phosphate level, the Panthalassic and Tethys event.
Ocean were well ventilated in Winguth and Winguth (2012). However,
the pole to equator sea surface temperature gradient was 5 °C higher 4. Climatic impact from the Siberian Traps
than the “weak” gradient scenario studied by Hotinski et al. (2001),
which showed a widespread anoxia in the deep sea. The greater export Investigation of the link between Siberian Trap volcanism and the
productivity induced by nutrient-rich coastal upwelling caused an ex- end-Permian mass extinction event has been a topic of debate for the
pansion and intensification of the OMZ in their model. Similar to last 20 years (e.g. Campbell et al., 1992; Basu et al., 1995; Renne et al.,
Meyer et al. (2008), Winguth and Winguth (2012) found that as phos- 1995; Bowring et al., 1998; Mundil et al., 2004; Saunders and Reichow,
phate content increased from 1× to 10× modern levels, higher produc- 2009). The Siberian Trap large igneous province (LIP) (ca. 252 Ma) con-
tivity induced higher O2 demand at depth and both vertical and tains perhaps the largest eruption of continental basaltic magma known
horizontal expansion of OMZ in their model. Their model included Fe on Earth (Reichow et al., 2002, 2009), with its volume being estimated
limitation on primary productivity, but their experiment with 1 × between 2 and 5 × 106 km3 (Renne and Basu, 1991). The Siberian LIPs
intruded into and erupted onto the thick sedimentary sequences of
30,000 the Tunguska basin, including Cambrian evaporites with 1 to 2 km of
CIE = -3‰
Volcanism anhydrites, halite and dolostone, and Ordovician to Permian marls,
CIE = -4‰
25,000 limestones, sandstones and coal, which contained abundant carbon
CIE = -5‰
Mass of C added (Gt C)
CIE = -6‰
(C), sulfur (S), fluorine (F) and chlorine (Cl). Geochronological evidence
20,000 suggested that Siberian Trap eruptions and mass extinction were
low C inventory: DIC = 38,000 Gt C synchronous (Reichow et al., 2009; Burgess et al., 2014).
15,000 Less attention has been paid in numerical modeling studies to the
potential short-term climatic consequences from the eruption of the
Siberian Traps, including a decadal scale cooling induced by sulfate
10,000
aerosol particles produced from the SO2 emissions and a long-term
warming due to the large amount of volcanic CO2 emissions (Renne
5,000 Organic Matter et al., 1995). The amount of released volcanic sulfur, halogens, and mag-
Thermogenic methane matic waters has been difficult to measure and is highly dependent on
Biogenic methane
0 the compositions of magma. Measurement of sulfur emission on melt
0 -10 -20 -30 -40 -50 -60 -70 -80 inclusions from the Siberian Traps indicates that the total S degassing
δ13C of carbon source (‰) was approximately ~ 6300 to 7800 Gt S (Black et al., 2012). Modern
basalt volcanic eruption observations, such as the 1783–1784 Iceland
Fig. 5. The calculated total amount of carbon required generating the carbon isotope
Laki volcanic eruption that have caused temperature decrease of about
excursion observed in marine carbonate records, assuming the source of 13C-depleted
carbon is volcanic CO2 (−6‰ to −12‰), organic matter (−25‰), thermogenic methane 1 °C in the northern hemisphere overall, combined with the effect of
(− 40‰) and biogenic methane (−60‰) and assuming an initial DIC reservoir of acid rain formed through reactions between atmospheric water and
38,000 Gt C. sulfur-rich gases, killing livestock and crops (Thordarson and Self,
Y. Cui, L.R. Kump / Earth-Science Reviews 149 (2015) 5–22 19
Table 4
Mass balance calculations for the total amount of carbon required generating the carbon isotope excursion (CIE) observed in marine carbonate.
Initial DIC Source of 13C-depleted carbon δ13C Total amount if CIE = −3‰b Total amount if CIE = −6‰
(Gt Ca) (‰) (Gt C) (Gt C)
2003). The estimated emitted SO2 from this relatively small-scale flood (Broecker and Peng, 1982; Kump and Arthur, 1999) can be used to
lava eruption is ~120 megatons (Mt) (1 Mt = 106 t), about four orders estimate the air–sea partition of added CO2, assuming no change in
of magnitude smaller than that estimated from the Siberian Traps, and the Ca concentration of seawater:
an order of magnitude smaller than the proposed S release during the
Deccan Traps event (Sigurdsson et al., 1992). Both theoretical and ob- h i
Ca2þ ½HCO3 − 2
servational data confirm the cooling effect from sulfuric acid aerosols: K¼ ð8Þ
such aerosols reduce the direct sun light reaching Earth surface, back- pCO2
scatter short-wave radiation, altering the energy balance between in-
coming solar radiation and the outgoing infrared radiation for periods and the amount of CO2 remaining in the atmosphere can be determined
up to several years (Sigurdsson et al., 1992; Self, 2005). from the following equation:
Any cooling on the short term would ultimately have been followed
by warming driven by volcanic CO2 degassing. The total CO2 released
2
from volcanic degassing, and the extent to which metamorphic reac- ½HCO3 − t ½pCO2 t
¼ : ð9Þ
tions driven by dike and sill penetration of organic-rich sedimentary ½HCO3 − 0 ½pCO2 0
layers released additional CO2, are key uncertainties in constraining
the global atmospheric CO2 (and temperature) rise. There may also
have been a catastrophic biogenic methane clathrate or permafrost Assuming that the initial ocean DIC reservoir is 38,000 Gt C, and the
methane release triggered by the initial warming from Siberian Traps initial atmospheric pCO2 is 10× PAL, the total C being added is 30,000 Gt
through climate feedbacks, thus intensify the global warming effect C, solving for Eq. (9) gives [pCO2]t/[pCO2]0 ≈ 3, equivalent to ~ 1.6
(Krull and Retallack, 2000; Retallack and Jahren, 2008). It has been a doublings of pCO2. The calculation provided above is an extreme
challenge to estimate the total amount of CO2 release due to a number simplification of the complex climate system that involves a number
of factors, including the uncertainty in the original volume of the Siberian of feedbacks and other climatic factors, but it does indicate that the
Trap volcanism (Saunders and Reichow, 2009), contributions from magnitudes of CO2 addition are in the range of what would be needed
various carbon sources (e.g. metamorphism of dolomite and organic car- to drive the observed warming, within large uncertainties. This amount
bon), and the uncertainty in gas generation potentials (Svensen et al., of pCO2 doubling from simple mass balance calculation is also consistent
2009). The CO2 production potential from contact metamorphism of with the results from more sophisticated model (Table 2). If we adopt
sediments during the Siberian Trap eruption in Tunguska Basin during the suggested 8–9 °C warming for this time period from oxygen isotope
pipe formation and the sill emplacement has been estimated (Svensen paleothermometer, the temperature rise would be consistent with a cli-
et al., 2009). Assuming that the volume of Siberian Trap volcanism is mate sensitivity of 5–6 °C per doubling of pCO2 (Table 2). The implied
1 × 106 km3 distributed across an area of 2.3 × 106 km2 (Reichow climate sensitivity should be used with caution because of the following
et al., 2002) with a degassing rate of ~0.2 Gt C yr−1 (McCartney et al., reasons. Firstly, the available Permian–Triassic paleotemperature proxy
1990; Self et al., 2005), Svensen et al. (2009) estimated that basin scale data come from limited geographic coverage, thus do not provide an
C production potential reaches as high as N30,000 Gt C (corresponding accurate representation of global mean surface warming. Secondly,
to N100,000 Gt CO2) from a combination of some volcanic degassing the inferred pCO2 increases throughout the extinction event are mostly
but a significant amount of contact metamorphism of coal (see Table 3 based on C isotope inversion, lacking independent proxy-data at similar
for a summary of the Siberian Trap volcanism). resolution.
One can perform a simple mass balance calculation to derive the
total amount of carbon release from various sources needed to replicate 5. Conclusions and future challenges
a − 3‰ carbon isotope excursion in the ocean–atmosphere system.
Allowing the duration of carbon release to vary from instantaneous 5.1. Concluding points
(0 kyr), to 20 kyr and 100 kyr, one finds that more carbon is needed
to produce the same amount of CIE as the duration of carbon release is (1) In summary, from our review of the temperature and pCO2 prox-
prolonged (Fig. 5 and Table 4). There are also uncertainties in the initial ies that are applicable to the pre-Mesozoic sedimentary records,
total carbon inventory in the ocean–atmosphere system, which one can we conclude that Late Permian ocean temperature in the tropics
address by assuming a range of initial DIC conditions (in our case, could have been as high as 35 °C based on multiple temperature
38,000 and 53,000 Gt C). One also finds that the calculated total amount proxies study, and the temperature could have risen up to 5–9 °C
of C needed to generate −6‰ CIE is a bit less than twice as large than across the P–T transition. We also find that the atmospheric
the amount needed to create a −3‰ CIE (Table 4). Assuming that the pCO2 prior to the end-Permian extinction event could have
ocean remains near saturation with respect to calcite, the equilibrium been as high as 4000 ppm, and pCO2 might have risen up to
equation for the saturation state between calcite and atmospheric CO2 N8000 ppm during the event. We find that the temperature
20 Y. Cui, L.R. Kump / Earth-Science Reviews 149 (2015) 5–22
Eiler, J.M., Schauble, E., 2004. 18O13C16O in Earth's atmosphere. Geochim. Cosmochim. Kutzbach, J., 1994. Idealized Pangean climates: sensitivity to orbital change. In: Klein, G.
Acta 68, 4767–4777. (Ed.), Pangea: Paleoclimate, Tectonics, and Sedimentation During Accretion, Zenith,
Ekart, D.D., Cerling, T.E., Montanez, I.P., Tabor, N.J., 1999. A 400 million year carbon isotope and Breakup of a Supercontinent. Geological Society of America Special Paper,
record of pedogenic carbonate: implications for paleoatmospheric carbon dioxide. Boulder, 288, pp. 41–55.
Am. J. Sci. 299, 805–827. Kutzbach, J., Gallimore, R., 1989. Pangaean climates: megamonsoons of the
Erwin, D.H., 1993. The Great Paleozoic Crisis: Life and Death in the Permian. Columbia megacontinent. J. Geophys. Res. 94, 3341–3357.
University Press, New York (327 pp.). Kutzbach, J., Ziegler, A., 1993. Simulation of Late Permian climate and biomes with
Erwin, D.H., 2006. Extinction: how life on earth nearly ended 250 million years ago. an atmosphere–ocean model: comparisons with observations. Philos. Trans. R. Soc.
Princeton University Press, Princeton, NJ (360 pp.). B—Biol. Sci. 341, 327–340.
Fluteau, F., Besse, J., Broutin, J., Ramstein, G., 2001. The Late Permian climate. What can be Kutzbach, J., Guetter, P., Ruddiman, W., Prell, W., 1989. Sensitivity of climate to late
inferred from climate modelling concerning Pangea scenarios and Hercynian range Cenozoic uplift in Southern Asia and the American West: numerical experiments.
altitude? Palaeogeogr. Palaeoclimatol. Palaeoecol. 167, 39–71. J. Geophys. Res. 94, 18393–18407.
Ghosh, P., Adkins, J., Affek, H., Balta, B., Guo, W., Schauble, E.A., Schrag, D., Eiler, J.M., 2006. Kutzbach, J., Guetter, P., Washington, W., 1990. Simulated circulation of an idealized
13
C–18O bonds in carbonate minerals: a new kind of paleothermometer. Geochim. ocean for Pangaean time. Paleoceanography 5, 299–317.
Cosmochim. Acta 70, 1439–1456. Kutzbach, J., Prell, W., Ruddiman, W.F., 1993. Sensitivity of Eurasian climate to surface
Gibbs, M.T., Rees, P.M.A., Kutzbach, J.E., Ziegler, A.M., Behling, P.J., Rowley, D.B., 2002. uplift of the Tibetan Plateau. J. Geol. 101, 177–190.
Simulations of Permian climate and comparisons with climate sensitive sediments. Lamarque, J.-F., Kiehl, J.T., Orlando, J.J., 2007. Role of hydrogen sulfide in a Permian–
J. Geol. 110, 33–55. Triassic boundary ozone collapse. Geophys. Res. Lett. 34, 1–15.
Grard, A., Francois, L., Dessert, C., Dupré, B., Godderis, Y., 2005. Basaltic volcanism Livermore, R., Smith, A., Vine, F., 1986. Late Palaeozoic to early Mesozoic evolution of
and mass extinction at the Permo-Triassic boundary: environmental impact and Pangaea. Nature 322, 162–165.
modeling of the global carbon cycle. Earth Planet. Sci. Lett. 234, 207–221. Luo, G., Kump, L., Wang, Y., Tong, J., Arthur, M., Yang, H., Huang, J., Yin, H., Xie, S., 2010.
Grice, K., Cao, C., Love, G., Bottcher, M., Twitchett, R., Grosjean, E., Summons, R., Turgeon, Isotopic evidence for an anomalously low oceanic sulfate concentration following
S., Dunning, W., Jin, Y., 2005. Photic zone euxinia during the Permian–Triassic end-Permian mass extinction. Earth Planet. Sci. Lett. 300, 101–111.
superanoxic event. Science 307, 706–709. McCartney, K., Huffman, A., Tredoux, M., 1990. A paradigm for endogenous causation of
Hannisdal, B., Peters, S.E., 2011. Phanerozoic Earth system evolution and marine biodiver- mass extinctions. In: Sharpton, V.L., Ward, P.D. (Eds.), Global Catastrophe in Earth
sity. Science 334, 1121–1124. History; An Interdisciplinary Conference on Impacts, Volcanism, and Mass Mortality.
Harzallah, A., Sadourny, R., 1995. Internal versus SST-forced atmospheric variability as Geological Society of America Special Paper, Boulder, 247, pp. 125–138.
simulated by an atmospheric general circulation model. J. Clim. 8, 474–495. Meyer, K.M., Kump, L.R., Ridgwell, A., 2008. Biogeochemical controls on photic-zone
Holser, W.T., Schonlaub, H.P., Aetrep Jr., M., Boeckelmann, K., Klein, P., Magaritz, M., Orth, euxinia during the end-Permian mass extinction. Geology 36, 747–750.
C.J., 1989. A unique geochemical record at the Permian/Triassic boundary. Nature Montenegro, A., Spence, P., Meissner, K., Eby, M., Melchin, M., Johnston, S., 2011. Climate
337, 39–44. simulations of the Permian–Triassic boundary: ocean acidification and the extinction
Hotinski, R., Bice, K., Kump, L., Najjar, R., Arthur, M., 2001. Ocean stagnation and end- event. Paleoceanography 26, 1–19.
Permian anoxia. Geology 29, 7–10. Mora, C.I., Driese, S.G., Seager, P.G., 1991. Carbon dioxide in the Paleozoic atmosphere:
Hotinski, R.M., Kump, L.R., Bice, K.L., 2002. Comment on “Could the Late Permian deep evidence from carbon-isotope compositions of pedogenic carbonate. Geology 19,
ocean have been anoxic?” by R. Zhang et al. Paleoceanography 17, 1–2. 1017–1020.
Huntington, K., Eiler, J., Affek, H., Guo, W., Bonifacie, M., Yeung, L., Thiagarajan, N., Passey, Mora, C.I., Driese, S.G., Colarusso, L.A., 1996. Middle to late Paleozoic atmospheric CO2
B., Tripati, A., Daëron, M., 2009. Methods and limitations of ‘clumped’ CO2 isotope levels from soil carbonate and organic matter. Science 271, 1105–1107.
(Δ47) analysis by gas-source isotope ratio mass spectrometry. J. Mass Spectrom. 44, Mundil, R., Ludwig, K.R., Metcalfe, I., Renne, P.R., 2004. Age and timing of the Permian
1318–1329. mass extinctions: U/Pb dating of closed-system zircons. Science 305, 1760–1763.
IPCC, 2007. In: Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M., Avery, K.B., Tignor, Nabbefeld, B., Grice, K., Summons, R., Hays, L., Cao, C., 2010. Significance of polycyclic ar-
M., Miller, H.L. (Eds.), Climate Change 2007: Physical Science Basis — Contribution of omatic hydrocarbons (PAHs) in Permian/Triassic boundary sections. Appl. Geochem.
Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on 25, 1374–1382.
Climate Change. Cambridge University Press, Cambridge (996 pp.). Nairn, A., Smithwick, M., 1976. Permian paleogeography and climatology. In: Falke, H.
Irving, E., 1977. Drift of the major continental blocks since the Devonian. Nature 270, (Ed.), The Continental Permian in Central, West, and South Europe. D. Reidel, Boston,
304–309. pp. 283–312.
Isozaki, Y., 1997. Permo-Triassic boundary superanoxia and stratified superocean: records Newell, A.J., Sennikov, A.G., Benton, M.J., Molostovskaya, I.I., Golubev, V.K., Minikh, A.V.,
from lost deep sea. Science 276, 235. Minikh, M.G., 2010. Disruption of playa-lacustrine depositional systems at the
Isozaki, Y., Maruyama, S., Furuoka, F., 1990. Accreted oceanic materials in Japan. Permo-Triassic boundary: evidence from Vyazniki and Gorokhovets on the Russian
Tectonophysics 181, 179–205. Platform. J. Geol. Soc. 167, 695–716.
Joachimski, M.M., Lai, X., Shen, S., Jiang, H., Luo, G., Chen, B., Chen, J., Sun, Y., 2012. Climate Parrish, J.T., 1982. Upwelling and petroleum source beds, with reference to Paleozoic.
warming in the latest Permian and the Permian–Triassic mass extinction. Geology 40, AAPG Bull. 66, 750–774.
195–198. Parrish, J.T., 1985. Latitudinal distribution of land and shelf and absorbed solar radiation
Kajiwara, Y., Yamakita, S., Ishida, K., Ishiga, H., Imai, A., 1994. Development of a largely an- during the Phanerozoic. U.S. Geological Survey Open File Report, 803–827, pp. 1–21.
oxic stratified ocean and its temporary massive mixing at the Permian/Triassic Parrish, J.T., 1993. Climate of the supercontinent Pangea. J. Geol. 101, 215–233.
boundary supported by the sulfur isotopic record. Palaeogeogr. Palaeoclimatol. Parrish, J., Parrish, J., Ziegler, A., 1986. Permian–Triassic paleogeography and paleoclima-
Palaeoecol. 111, 367–379. tology and implications for therapsid distribution. In: Hotton, N., MacLean, P.D.,
Kamo, S., Czamanske, G., Krogh, T., 1996. A minimum U–Pb age for Siberian flood-basalt Roth, J.J., Roth, E.C. (Eds.), The Ecology and Biology of Mammal-like Reptiles.
volcanism. Geochim. Cosmochim. Acta 60, 3505–3511. Smithsonian, Washington, p. 326.
Kamo, S., Czamanske, G., Amelin, Y., Fedorenko, V., Davis, D., Trofimov, V., 2003. Rapid Payne, J.L., Clapham, M.E., 2012. End-Permian mass extinction in the oceans: an ancient
eruption of Siberian flood-volcanic rocks and evidence for coincidence with the analog for the twentyfirst century. Annu. Rev. Earth Planet. Sci. 40, 89–111.
Permian–Triassic boundary and mass extinction at 251 Ma. Earth Planet. Sci. Lett. Payne, J., Kump, L., 2007. Evidence for recurrent Early Triassic massive volcanism from
214, 75–91. quantitative interpretation of carbon isotope fluctuations. Earth Planet. Sci. Lett.
Kidder, D.L., Worsley, T.R., 2004. Causes and consequences of extreme Permo-Triassic 256, 264–277.
warming to globally equable climate and relation to the Permo-Triassic extinction Prokoph, A., Shields, G., Veizer, J., 2008. Compilation and time-series analysis of a marine
and recovery. Palaeogeogr. Palaeoclimatol. Palaeoecol. 203, 207–237. carbonate δ18O, δ13C, 87Sr/86Sr and δ34S database through Earth history. Earth Sci. Rev.
Kiehl, J., Shields, C., 2005. Climate simulation of the latest Permian: implications for mass 87, 113–133.
extinction. Geology 33, 757. Pucéat, E., Joachimski, M.M., Bouilloux, A., Monna, F., Bonin, A., Motreuil, S., Moriniére, P.,
Knoll, A.H., Bambach, R.K., Canfield, D.E., Grotzinger, J.P., 1996. Comparative earth history Hénard, S., Mourin, J., Dera, G., 2010. Revised phosphate-water fractionation equation
and Late Permian mass extinction. Science 273, 452–457. reassessing paleotemperatures derived from biogenic apatite. Earth Planet. Sci. Lett.
Knoll, A.H., Bambach, R.K., Payne, J.L., Pruss, S., Fischer, W.W., 2007. Paleophysiology and 298, 135–142.
end-Permian mass extinction. Earth Planet. Sci. Lett. 256, 295–313. Rampino, M.R., Caldeira, K., 2005. Major perturbation of ocean chemistry and a
Korte, C., Kozur, H.W., 2010. Carbon-isotope stratigraphy across the Permian–Triassic ‘Strangelove Ocean’ after the end-Permian mass extinction. Terra Nova 17,
boundary: a review. J. Asian Earth Sci. 39, 215–235. 554–559.
Krassilov, V., Karasev, E., 2009. Paleofloristic evidence of climate change near and beyond Rampino, M.R., Prokoph, A., Adler, A., 2000. Tempo of the end-Permian event: high-
the Permian–Triassic boundary. Palaeogeogr. Palaeoclimatol. Palaeoecol. 284, 326–336. resolution cyclostratigraphy at the Permian–Triassic boundary. Geology 28, 643–647.
Krull, E.S., Retallack, G.J., 2000. δ13C depth profiles from paleosols across the Permian– Ramstein, G., Fluteau, F., Besse, J., Joussaume, S., 1997. Effect of orogeny, plate motion and
Triassic boundary: evidence for methane release. Bull. Geol. Soc. Am. 112, 1459–1472. land–sea distribution on Eurasian climate change over the past 30 million years.
Kump, L., 2002. Reducing uncertainty about carbon dioxide as a climate driver. Nature Nature 386, 788–795.
419, 188–190. Rees, P.M., Gibbs, M.T., Ziegler, A.M., Kutzbach, J.E., Behling, P.J., 1999. Permian climates:
Kump, L.R., Arthur, M.A., 1999. Interpreting carbon-isotope excursions: carbonates and evaluating model predictions using global paleobotanical data. Geology 27, 891–894.
organic matter. Chem. Geol. 161, 181–198. Rees, P.M., Ziegler, A.M., Gibbs, M.T., Kutzbach, J.E., Behling, P.J., Rowley, D.B., 2002.
Kump, L.R., Pavlov, A., Arthur, M.A., 2005. Massive release of hydrogen sulfide to the sur- Permian phytogeographic patterns and climate data/model comparisons. J. Geol.
face ocean and atmosphere during intervals of oceanic anoxia. Geology 33, 397–400. 110, 1–31.
Kürrschner, W.M., van der Burgh, J., Visscher, H., Dilcher, D.L., 1996. Oak leaves as biosen- Reichow, M.K., Saunders, A.D., White, R.V., Pringle, M.S., Al'Mukhamedov, A.I., Medvedev,
sors of late neogene and early pleistocene paleoatmospheric CO2 concentrations. Mar. A.I., Kirda, N.P., 2002. 40Ar/39Ar dates from the West Siberian Basin: Siberian flood
Micropaleontol. 27, 299–312. basalt province doubled. Science 296, 1846–1849.
22 Y. Cui, L.R. Kump / Earth-Science Reviews 149 (2015) 5–22
Reichow, M., Pringle, M., Almukhamedov, A., Allen, M., Andreichev, V., Buslov, M., Davies, Torcq, F., Besse, J., Vaslet, D., Marcoux, J., Ricou, L., Halawani, M., Basahel, M., 1997. Paleo-
C., Fedoseev, G., Fitton, J., Inger, S., 2009. The timing and extent of the eruption of magnetic results from Saudi Arabia and the Permo-Triassic Pangea configuration.
the Siberian Traps large igneous province: implications for the end-Permian environ- Earth Planet. Sci. Lett. 148, 553–567.
mental crisis. Earth Planet. Sci. Lett. 277, 9–20. Van der Voo, R., French, R., 1974. Apparent polar wandering for the Atlantic-bordering
Renne, P.R., Basu, A.R., 1991. Rapid eruption of the Siberian Traps flood basalts at the continents: Late Carboniferous to Eocene. Earth Sci. Rev. 10, 99–119.
Permo-Triassic boundary. Science 253, 176–179. Veevers, J.J., Conaghan, P.J., Shaw, S.E., 1994. Turning point in Pangean environmental
Renne, P.R., Zhang, Z., Richardson, M.A., Black, M.T., Basu, A.R., 1995. Synchrony and causal history at the Permian/Triassic (P/Tr) boundary. In: Klein, G.D. (Ed.), Pangea:
relations between Permian–Triassic boundary crises and Siberian flood volcanism. Paleoclimate, Tectonics, and Sedimentation During Accretion, Zenith, and Breakup
Science 269, 1413–1416. of a Supercontinent. Geological Society of America Specical Paper, 288, pp. 187–196.
Retallack, G.J., 2001. A 300-million-year record of atmospheric carbon dioxide from fossil Visscher, H., Looy, C.V., Collinson, M.E., Brinkhuis, H., van Konijnenburg-van Cittert, J.H.A.,
plant cuticles. Nature 411, 287–290. Kurschner, W.M., Sephton, M.A., 2004. Environmental mutagenesis during the
Retallack, G.J., 2009. Greenhouse crises of the past 300 million years. Geol. Soc. Am. Bull. end-Permian ecological crisis. Proc. Natl. Acad. Sci. U. S. A. 101, 12952–12956.
121, 1441–1455. Visscher, H., Sephton, M.A., Looy, C.V., 2011. Fungal virulence at the time of the
Retallack, G.J., 2013. Permian and Triassic greenhouse crises. Gondwana Res. 24, 90–103. end-Permian biosphere crisis? Geology 39, 883–886.
Retallack, G.J., Jahren, A.H., 2008. Methane release from igneous intrusion of coal during Wang, C., Visscher, H., 2007. Abundance anomalies of aromatic biomarkers in the
Late Permian extinction events. J. Geol. 116, 1–20. Permian–Triassic boundary section at Meishan, China — evidence of end-Permian
Retallack, G.J., Veevers, J.J., Morante, R., 1996. Global coal gap between Permian–Triassic terrestrial ecosystem collapse. Palaeogeogr. Palaeoclimatol. Palaeoecol. 252, 291–303.
extinction and Middle Triassic recovery of peat-forming plants. Geol. Soc. Am. Bull. Wang, K., Geldsetzer, H.H.J., Krouse, H.R., 1994. Permian–Triassic extinction: organic 13C
108, 195–207. evidence from British Columbia, Canada. Geology 22, 580–584.
Retallack, G.J., Sheldon, N.D., Carr, P.F., Fanning, M., Thompson, C.A., Williams, M.L., Jones, Ward, P.D., Montgomery, D.R., Smith, R., 2000. Altered river morphology in South Africa
B.G., Hutton, A., 2011. Multiple Early Triassic greenhouse crises impeded recovery related to the Permian–Triassic extinction. Science 289, 1740–1743.
from Late Permian mass extinction. Palaeogeogr. Palaeoclimatol. Palaeoecol. 308, Wignall, P., 2001. Large igneous provinces and mass extinctions. Earth Sci. Rev. 53, 1–33.
233–251. Wignall, P.B., Hallam, A., 1992. Anoxia as a cause of the Permian/Triassic extinction: facies
Romano, C., Goudemand, N., Vennemann, T.W., Ware, D., Schneebeli-Hermann, E., evidence from northern Italy and the western United States. Palaeogeogr.
Hochuli, P.A., Brühwiler, T., Brinkmann, W., Bucher, H., 2012. Climatic and biotic up- Palaeoclimatol. Palaeoecol. 93, 21–46.
heavals following the end-Permian mass extinction. Nat. Geosci. 6, 57–60. Wignall, P.B., Twichett, R.J., 1996. Ocean anoxia and the end Permian mass extinction.
Roscher, M., Stordal, F., Svensen, H., 2011. The effect of global warming and global cooling Science 272, 1155–1158.
on the distribution of the latest Permian climate zones. Palaeogeogr. Palaeoclimatol. Wilkin, R., Barnes, H., Brantley, S., 1996. The size distribution of framboidal pyrite in
Palaeoecol. 309, 186–200. modern sediments: an indicator of redox conditions. Geochim. Cosmochim. Acta
Royer, D., 2001. Stomatal density and stomatal index as indicators of paleoatmospheric 60, 3897–3912.
CO2 concentration. Rev. Palaeobot. Palynol. 114, 1–28. Winguth, A.M.E., Maier-Reimer, E., 2005. Causes of the marine productivity and oxygen
Royer, D.L., Berner, R.A., Beerling, D.J., 2001. Phanerozoic atmospheric CO2 change: changes associated with the Permian–Triassic boundary: a reevaluation with ocean
evaluating geochemical and paleobiological approaches. Earth Sci. Rev. 54, 349–392. general circulation models. Mar. Geol. 217, 283.
Royer, D.L., Berner, R.A., Montañez, I.P., Tabor, N.J., Beerling, D.J., 2004. CO2 as a primary Winguth, C., Winguth, A.M.E., 2012. Simulating Permian–Triassic oceanic anoxia
driver of Phanerozoic climate. GSA Today 14, 4–10. distribution: implications for species extinction and recovery. Geology 40 (2),
Saunders, A., Reichow, M., 2009. The Siberian Traps and the End-Permian mass extinction: 127–130.
a critical review. Chin. Sci. Bull. 54, 20–37. Winguth, A.M.E., Heinze, C., Kutzbach, J., Maier-Reimer, E., Mikolajewicz, U., Rowley, D.,
Self, S., 2005. Effects of volcanic eruptions on the atmosphere and climate. In: Marti, J., Rees, A., Ziegler, A., 2002. Simulated warm polar currents during the middle Permian.
Ernst, G.G.J. (Eds.), Volcanoes and the Environment. Cambridge University Press, Paleoceanography 17, 1–18.
Cambridge, pp. 152–174. Woodward, F., 1987. Stomatal numbers are sensitive to increases in CO2 from pre-
Self, S., Thordarson, T., Widdowson, M., 2005. Gas fluxes from flood basalt eruptions. industrial levels. Nature 327, 617–618.
Elements 1, 283–287. Woodward, F., Bazzaz, F., 1988. The responses of stomatal density to CO2 partial pressure.
Sellers, W.D., 1969. A global climatic model based on the energy balance of the earth– J. Exp. Bot. 39, 1771–1781.
atmosphere system. J. Appl. Meteorol. 8, 392–400. Wynn, J.G., 2003. Towards a physically based model of CO2-induced stomatal frequency
Sephton, M.A., Looy, C.V., Brinkhuis, H., Wignall, P.B., de Leeuw, J.W., Visscher, H., 2005. response. New Phytol. 157, 394–398.
Catastrophic soil erosion during the end-Permian biotic crisis. Geology 33, 941–944. Xie, S., Pancost, R.D., Yin, H., Wang, H., Evershed, R.P., 2005. Two episodes of microbial
Shen, S.-z, Crowley, J.L., Wang, Y., Bowring, S.A., Erwin, D.H., Sadler, P.M., Cao, C.-q, change coupled with Permo/Triassic faunal mass extinction. Nature 434, 494–497.
Rothman, D.H., Henderson, C.M., Ramezani, J., 2011. Calibrating the end-Permian Xie, S., Pancost, R.D., Huang, J., Wignall, P.B., Yu, J., Tang, X., Chen, L., Huang, X., Lai, X.,
mass extinction. Science 334, 1367–1372. 2007. Changes in the global carbon cycle occurred as two episodes during the
Sigurdsson, H., D'Hondt, S., Carey, S., 1992. The impact of the Cretaceous/Tertiary bolide Permian Triassic crisis. Geology 35, 1083–1086.
on evaporite terrane and generation of major sulfuric acid aerosol. Earth Planet. Sci. Xie, S., Pancost, R.D., Wang, Y., Yang, H., Wignall, P.B., Luo, G., Jia, C., Chen, L., 2010.
Lett. 109, 543–559. Cyanobacterial blooms tied to volcanism during the 5 m.y. Permo-Triassic biotic
Slingerland, R., Kump, L., 2011. Mathematical Modeling of Earth's Dynamical Systems: A crisis. Geology 38, 447–450.
Primer. Princeton University Press, Princeton (231 pp.). Yapp, C., Poths, H., 1992. Ancient atmospheric CO2 pressures inferred from natural goe-
Steiner, M.B., Eshet, Y., Rampino, M.R., Schwindt, D.M., 2003. Fungal abundance spike and thites. Nature 355, 342–344.
the Permian–Triassic boundary in the Karoo Supergroup (South Africa). Palaeogeogr. Yapp, C.J., Poths, H., 1996. Carbon isotopes in continental weathering environments and
Palaeoclimatol. Palaeoecol. 194, 405–414. variations in ancient atmospheric CO2 pressure. Earth Planet. Sci. Lett. 137, 71–82.
Sun, Y., Joachimski, M.M., Wignall, P.B., Yan, C., Chen, Y., Jiang, H., Wang, L., Lai, X., 2012. Yemane, K., 1993. Contribution of Late Permian palaeogeography in maintaining a
Lethally hot temperatures during the Early Triassic greenhouse. Science 338, 366–370. temperate climate in Gondwana. Nature 361, 51–54.
Svensen, H., Planke, S., Polozov, A., Schmidbauer, N., Corfu, F., Podladchikov, Y., Jamtveit, Zeebe, R.E., Westbroek, P., 2003. A simple model for the CaCO3 saturation state of the
B., 2009. Siberian gas venting and the end-Permian environmental crisis. Earth Planet. ocean: the “Strangelove,” the “Neritan,” and the “Cretan” Ocean. Geochem. Geophys.
Sci. Lett. 277, 490–500. Geosyst. 4, 1–26.
Tabor, N.J., Yapp, C.J., Montañez, I.P., 2004. Goethite, calcite, and organic matter from Zhang, R., Follows, M.J., Grotzinger, J.P., Marshall, J., 2001. Could the Late Permian deep
Permian and Triassic soils: carbon isotopes and CO2 concentrations. Geochim. ocean have been anoxic? Paleoceanography 16, 317–329.
Cosmochim. Acta 68, 1503–1517. Zhang, R., Follows, M., Marshall, J., 2003. Reply to Comment by Roberta M. Hotinski, Lee R.
Taylor, E.L., Taylor, T.N., Cúneo, N.R., 1992. The present is not the key to the past: a polar Kump, and Karen L. Bice on “Could the Late Permian deep ocean have been anoxic?”.
forest from the Permian of Antarctica. Science 257, 1675–1677. Paleoceanography 18, 1–2.
Taylor, E.L., Taylor, T.N., Cúneo, N.R., 2000. Permian and Triassic high latitude Zhang, H., Tong, J., Zuo, J., 2005. Lower Triassic and carbon isotope excursion in West
paleoclimates: evidence from fossil biotas. In: Huber, B.T., McLeod, K.G., Wing, S.L. Guangxi, southwest China. Albertiana 33, 103–104.
(Eds.), Warm Climates in Earth History. Cambridge University Press, Cambridge, Ziegler, A., 1990. Phytogeographic patterns and continental configurations during the
pp. 321–350. Permian period. Geol. Soc. Lond. Mem. 12, 363–379.
Thompson, S.L., Pollard, D., 1997. Greenland and Antarctic mass balances for present and Ziegler, A.M., Hulver, M.L., Rowley, D.B., Martini, I.P., 1997. Permian world topography and
doubled atmospheric CO2 from the GENESIS version-2 global climate model. J. Clim. climate. In: Martini, I.P. (Ed.), Late Glacial and Postglacial Environmental Changes:
10, 871–900. Quaternary, Carboniferous–Permian, and Proterozoic. Oxford University Press,
Thordarson, T., Self, S., 2003. Atmospheric and environmental effects of the 1783–1784 New York, pp. 1–37.
Laki eruption: a review and reassessment. J. Geophys. Res. 108, 1–29.