Six-fold increase of atmospheric pCO2 during the Permian–Triassic mass extinction

Abstract

The Permian–Triassic mass extinction was marked by a massive release of carbon into the ocean-atmosphere system, evidenced by a sharp negative carbon isotope excursion. Large carbon emissions would have increased atmospheric pCO2 and caused global warming. However, the magnitude of pCO2 changes during the PTME has not yet been estimated. Here, we present a continuous pCO2 record across the PTME reconstructed from high-resolution δ13C of C3 plants from southwestern China. We show that pCO2 increased from 426 +133/−96 ppmv in the latest Permian to 2507 +4764/−1193 ppmv at the PTME within about 75 kyr, and that the reconstructed pCO2 significantly correlates with sea surface temperatures. Mass balance modelling suggests that volcanic CO2 is probably not the only trigger of the carbon cycle perturbation, and that large quantities of 13C-depleted carbon emission from organic matter and methane were likely required during complex interactions with the Siberian Traps volcanism. The Permian–Triassic mass extinction was accompanied by a massive release of carbon into the ocean-atmosphere system, but the magnitude of change is not well known. Here, the authors present a new record of C3 plants from southwestern China which shows that atmospheric pCO2 increased by a factor of six during this event.

Full text

ARTICLE
Six-fold increase of atmospheric pCO
2
during the
Permian–Triassic mass extinction
Yuyang Wu 1,2, Daoliang Chu 1✉, Jinnan Tong1, Haijun Song 1, Jacopo Dal Corso 1, Paul B. Wignall3,
Huyue Song1, Yong Du1& Ying Cui2✉
The Permian–Triassic mass extinction was marked by a massive release of carbon into the
ocean-atmosphere system, evidenced by a sharp negative carbon isotope excursion. Large
carbon emissions would have increased atmospheric pCO
2
and caused global warming.
However, the magnitude of pCO
2
changes during the PTME has not yet been estimated.
Here, we present a continuous pCO
2
record across the PTME reconstructed from high-
resolution δ13CofC
3
plants from southwestern China. We show that pCO
2
increased from
426 +133/−96 ppmv in the latest Permian to 2507 +4764/−1193 ppmv at the PTME within
about 75 kyr, and that the reconstructed pCO
2
significantly correlates with sea surface
temperatures. Mass balance modelling suggests that volcanic CO
2
is probably not the only
trigger of the carbon cycle perturbation, and that large quantities of 13C-depleted carbon
emission from organic matter and methane were likely required during complex interactions
with the Siberian Traps volcanism.
https://doi.org/10.1038/s41467-021-22298-7 OPEN
1State Key Laboratory of Biogeology and Environmental Geology, School of Earth Sciences, China University of Geosciences, Wuhan, China. 2Department of
Earth and Environmental Studies, Montclair State University, Montclair, NJ, USA. 3School of Earth and Environment, University of Leeds, Leeds, UK.
✉email: chudl@cug.edu.cn;cuiy@montclair.edu
NATURE COMMUNICATIONS | (2021) 12:2137 | https://doi.org/10.1038/s41467-021-22298-7 | www.nature.com/naturecommunications 1
1234567890():,;
Content courtesy of Springer Nature, terms of use apply. Rights reserved

The Permian–Triassic mass extinction (PTME; ca. 252 Ma)
coincided with rapid global warming that produced one of
the hottest intervals of the Phanerozoic1–5, which was
likely triggered by a massive release of greenhouse gases6,7. The
emplacement of the Siberian Traps large igneous province has
been widely suggested as the ultimate trigger for the extinction of
~90% of marine species and ~70% of terrestrial vertebrate species
at the Permian–Triassic boundary8, with major losses amongst
plants (e.g. refs. 9,10). Alongside volcanic degassing, CO
2
,SO
2,
and halogen volatiles were likely released due to thermal meta-
morphism by Siberian Traps’intrusions into organic-rich
sediments6,7,11. The global negative carbon isotope excursion
(CIE) found in both marine and terrestrial settings at the PTME
(for a review, ref. 12) indicates a major carbon cycle perturbation
in the ocean-atmosphere system, which implies a rise in the
atmospheric CO
2
levels (pCO
2
). However, pCO
2
changes during
the PTME still remain poorly constrained.
On the one hand, records of pCO
2
from proxies (stomata
index, palaeosol carbonates, and biomarkers) are mainly focused
on the late Permian and/or Phanerozoic long-term trends without
detailed pCO
2
data for the earliest Triassic (refs. 13–17). On the
other hand, various models show large variability of peak pCO
2
estimates, because of the different assumed background pCO
2
levels (e.g. refs. 18–20). Hence, there is a pressing need for a
continuous proxy-based and high-resolution record of pCO
2
during the PTME. Understanding the magnitude of pCO
2
changes during past hyperthermals is indeed crucial to under-
stand the possible imminent environmental effects of today’sCO
2
increase: pCO
2
has risen from 280 to more than 400 ppmv in the
last 150 years and is projected to go higher21.
Experiments on living C
3
plants (in the field and in growth
chambers) suggest that carbon isotope fractionation (Δ13C)
during photosynthesis increases with increasing CO
2
levels,
lowering the carbon isotope signature of C
3
plants (δ13C
p
)22.
Based on this relationship, Δ13C calculated from δ13C
p
measured
in fossil C
3
plants remains can be used as a proxy for past pCO
2
23.
This proxy successfully reproduced ice-core records of pCO
2
for
the Last Glacial Maximum23, and has been applied to reconstruct
pCO
2
during Early Eocene hyperthermals24, the Cretaceous
Period25, and the Toarcian Oceanic Anoxic Event26.
Here, we present high-resolution δ13C records of fossil C
3
plant
remains from sedimentary successions of southwestern China.
Using the δ13C data of C
3
plants, we calculated a six-fold increase
of atmospheric pCO
2
during the PTME, from 426 +133/−96
ppmv to 2507 +4764/−1193 ppmv. Furthermore, the pCO
2
estimates are compared with carbon isotope mass balance cal-
culations showing that in addition to volcanic CO
2
, large quan-
tities of 13C-depleted carbon emission from organic matter and
methane were likely required to trigger the observed global
negative CIE in the exogenic carbon pool.
Results and discussion
High-resolution terrestrial carbon isotope records. We present
high-resolution terrestrial organic carbon isotope records
(δ13C
org
) from plant cuticles, wood and bulk organic matter
(OM) together with our previous work10 from four terrestrial
Permian–Triassic boundary sections (Chahe, Jiucaichong, core
ZK4703 and Chinahe) in southwestern China (Supplementary
Fig. 1; Supplementary Fig. 2). The δ13C of bulk OM and C
3
plant
remains from the four study sections exhibit nearly identical
secular trends (Fig. 1). Each profile can be divided into four
stages: (1) a pre-CIE stage, (2) an onset of the negative CIE (onset
of CIE) stage, (3) a prolonged CIE body stage and (4) a post-CIE
stage. In the pre-CIE stage, δ13C
org
records from the Xuanwei
Formation are characterized by steady values around −25.0%
(Fig. 1). The synchronous, prominent onset of CIEs with peak
values of −32% occurs at the bottom of the Kayitou Formation.
Subsequently, the onset of the CIE stage is followed by a pro-
longed interval with sustained low values (ca. −30%) through the
whole Kayitou Formation, interrupted by a slight positive shift
immediately after the onset of CIE. A recovery to slightly higher
δ13C
org
values (−28% to −26%) starts in the uppermost part of
the Kayitou Formation and the Dongchuan Formation. Pre-
viously published terrestrial δ13C
org
profiles in southwestern
China (e.g. refs. 27,28) all belong to mixed organic carbon source
Siltstone
Silty mudstone
Mudstone
Ash bed
Coal bed
Sandstone
Charred wood
Cuticle
Bulk organic matter
Non-charred wood
floraGigantopteris
δ13C()
org ಽ
−30 −28 −26 −24 −22
−32
TOC(%)
0123
no
itam
r
oF
u
ot
i
yaK .
m
F C
D
.mF iew
n
a
u
X
0
70
75
80
85
90
95
100
105
110
5
10
15
20
25
(m)
(m)
Jiucaichong
Chahe
δ13C()
org ಽ
−30 −28 −26 −24 −22
−32
TOC(%)
0123
noitamroF
uot
i
yaK
noitamroF i
ew
n
auX
690
685
680
675
670
665
660
(m)
ZK4703
onset of CIE
pre-CIE
post-CIE
CIE body
Southwestern China Pakistan Global marine
−32 −30 −28 −26 −24 −22
δ13C()
org ಽ
n
o
it
a
m
r
oF
u
oti
ya
K
n
o
i
t
a
mroF
ie
wn
a
u
Xno
it
am
r
oF
n
a
uh
c
g
n
o
D
LAD of coal
−36
−34 8.
Eastern Paleotethys Western Paleotethys
Central Paleotethys
Northern Neotethys
Southern Neotethys
Bálvány North
0HLVKDQ
1KL 7DR
<DQJRX
=DO
.XK H $OL %DVKL
6KDKUH]D
$EDGHK
:DGL 6KDKKD
:HQEXGDQJVDQJ
U/Pb dates at Meishan
δ13C ( )
carb ಽ
sisne
g
n
i
xna
h
c
C
C yini
C m
H c
C t
H parvus
ie
hc
s
e
a
ts
Iacicr
as
i
I
and
QD
LPUH3H
WD/ FLVVDLU7
\
OUD
(
 

  
 
   
 

 
     
)aM(eg
A
30°
30°
0°
60°
60°
South China
Pakistan
Arabia
Tibet
Alps
Iran
Neotethys
Paleotethys
Panthalassa
−30 −28 −26 −24 −22
−32
0123
TOC (%) δ13C()
org ಽ
noitamroF
u
o
tiyaK n
oit
amr
oF
nauhcgn
oD
n
oi
t
amroF
i
e
wnauX
0
10
20
30
40
50
60
70
80
Chinahe
(m)
−32 −30 −28 −26 −24 −22
δ13C( )
pಽ
(m)
Amb







P)XUGLKK
&P)LO
DZ
Q
DL0
Fig. 1 Carbon isotope excursion trend recorded in global terrestrial C
3
plants and marine carbonates. The secular carbon isotope excursion (CIE) trend
can be divided into four stages (i.e. pre-CIE, onset of CIE, CIE body and post-CIE) in terrestrial bulk organic matter, C
3
plants and marine carbonates, and
are shown as different color fields. The last appearance datum (LAD) of coal beds and Gigantopteris flora distributions represent the coal gap and collapse
of tropical peatlands respectively10,45. Carbon isotope (δ13C) data source: Chahe (δ13C of bulk organic matter from ref. 27;δ13C of plants data from this
study), Jiucaichong (this study), ZK4703 core and Chinahe (δ13C data in this study together with our previous work10), Amb (Pakistan)32 and global
marine carbonate δ13C (Methods). The locations of marine and terrestrial carbon isotope profiles are shown in the late Permian palaeogeographic map.
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-22298-7
2NATURE COMMUNICATIONS | (2021) 12:2137 | https://doi.org/10.1038/s41467-021-22298-7 | www.nature.com/naturecommunications
Content courtesy of Springer Nature, terms of use apply. Rights reserved

in bulk OM. Few unusually negative values (< −34%) observed in
the upper Kayitou Formation, e.g., in a published record from
Chahe27, are statistical outliers and local signals, as such negative
values are not observed in our high-resolution study. These
outliers may be caused by local 13C-depleted samples possibly
containing an algal and/or bacterial component29.
The four-stage terrestrial δ13C
org
trend is also seen in the
marine carbonate carbon isotope (δ13C
carb
) records (Fig. 1). A
total of 10 global-distributed marine Permian–Triassic boundary
sections with both high-resolution δ13C
carb
and conodont
biostratigraphy were integrated as a global marine δ13C
carb
profile, using the age model from the Meishan Global Stratotype
Section and Point (GSSP)30 (Supplementary Fig. 3; Supplemen-
tary Fig. 4). These newly compiled global δ13C
carb
records are
nearly identical to those published previously (e.g. ref. 12).
pCO
2
estimates based on Δ13C of fossil plants. Constraining the
magnitude of the CIE is crucial to estimate accurate mass, rate,
and source of the 13C-depleted carbon released during the
PTME20. CIE magnitudes show large variations between different
localities and substrates because they can be affected by multiple
factors31.δ13C
p
profiles from southwestern China (low latitude)
and Pakistan32 (middle latitude) exhibit CIE magnitudes of ca.
−7% and ca. −5.5% respectively, which are significantly larger
than the ca. −3.5% marine CIE magnitude estimated from global
marine δ13C
carb
records (Fig. 1). Data compilation confirms this
discrepancy: terrestrial CIE magnitudes range from −3.6% to
−6.1% (bulk OM, 25th percentile to 75th percentile, n=29), and
from −5.2% to −7.1% (C
3
plants, n=9), whereas marine CIE
magnitudes range from −3.0% to −4.7% (n=69) (Fig. 2;Sup-
plementary Table 1). As shown both in modern and fossil plants,
elevated atmospheric pCO
2
was likely responsible for amplifying
the magnitude of the CIE in the terrestrial δ13C
p
record due to an
increase of Δ13C22,33. Therefore, following the relationship
between Δ13C and pCO
2
developed by Cui and Schubert24
(Methods), we could calculate the pCO
2
across the PTME. The
Δ13C was calculated using the δ13C
p
records of four study sections
from southwestern China, and the δ13C
CO2
(the δ13C of atmo-
spheric CO
2
; Supplementary Fig. 5) calculated from the global
marine δ13C
carb
compiled in this study. As explained above, this is
possible because the marine and terrestrial δ13C records are clo-
sely comparable and can be readily correlated (Fig. 1), the corre-
lation being supported also by biostratigraphy (flora and
conchostracans), and radioisotope dating (Supplementary Fig. 6;
Supplementary Information). The initial, background late Per-
mian pCO
2(t
=
0)
is set in our calculations at 425 ± 68 ppmv based
on the late Changhsingian pCO
2
estimates calculated by Li et al.16
using stomatal ratio method and mechanistic gas exchange model
for fossil conifers from the Dalong Formation in South China,
with good age control and reliable taxonomy.
Our estimates (Fig. 3) show that pCO
2
was moderately low (426
+133/−96 ppmv) at 252.1 Ma within the pre-CIE stage (upper part
of conodont Clarkina changxingensis zone). Subsequently, the pCO
2
began to increase rapidly in the Clarkina yini zone, reaching a
maximum level (2507 +4764/−1193 ppmv), immediately after the
Permian–Triassic boundary (Hindeodus parvus zone). This near
six-fold increase of atmospheric pCO
2
occurred within ~75 kyr and
coincided with the onset of the global CIE. The pCO
2
remained
high (ca. 1500 to 2500 ppmv) immediately after the onset of the
CIE, with only one transient drop (down to ca. 1300 ppmv).
Coupled to the recovery of δ13C, pCO
2
dropstoca.700ppmvat
the top of Isarcicella isarcica zone. Atmospheric CO
2
levels show a
close coupling with estimated sea surface temperatures (r=+0.60,
p< 0.001, n=173; Supplementary Fig. 7), implying that CO
2
was
likely the dominant greenhouse gas across the PTME, although the
contribution of other greenhouse gases such as methane and water
vaporcannotbeexcludedhere.Thesix-foldincreaseofatmospheric
pCO
2
, together with a 10 °C increase in sea surface temperatures
estimated from low latitude conodont oxygen isotope (Fig. 3)
implies Earth system sensitivity (ESS) of 3.9 °C per doubling of CO
2
if we assume ESS equals to ΔT/log
2
[pCO
2(peak)
/pCO
2(background)
]34.
This is consistent with a previous estimate of the Permian–Triassic
ESS35 and the IPCC equilibrium climate sensitivity range of 1.5 to
4.5 with a median of 3.036, suggesting slow feedbacks operated in
the geologic past. However, climate model simulations reveal that
the increase of SST in high latitude should be higher than low
latitude37. As a result, the 10 °C SST increase in low latitude might
underestimate the global SST increase, which leads to an under-
estimate of the Earth system sensitivity during the PTME.
Comparison with previous studies and uncertainty. Previous
pCO
2
estimates around the Permian–Triassic boundary (Fig. 3;
Supplementary Table 2) come from stomatal proxies16,17,
palaeosol carbonates13,14, phytane15 and carbon cycle modelling
(e.g. refs. 18–20). Published proxy-based pCO
2
reconstructions are
mostly for the late Permian, within long-term and very low-
resolution Phanerozoic records. Stomata-based estimates from
modified fossil Ginkgo stomatal index method17 gave pCO
2
around 400–800 ppmv in the latest Permian, but with poor age
constraint and high taxonomic uncertainty16. Latest Permian
pCO
2
from δ13C of palaeosol carbonates from Texas, US13, was
calculated at 400 ppmv38,39 (re-calculated by ref. 38 correcting the
assumed soil respired CO
2
concentration), but latest Permian
palaeosol carbonate record from the Karoo Basin14 shows higher
n = 69 n = 29 n = 9
-10.0
-7.5
-5.0
-2.5
0.0
Carbonate Bulk OM C plant
3
marine
terrestrial
Maximun
Minimum
Mean
Median
15th quantile
75th quantile
᧤‰᧥
Fig. 2 Boxplot of carbon isotope excursion magnitudes for three
substrates. Carbon isotope excursion (CIE) magnitudes of marine
carbonate, terrestrial bulk organic matter (OM), and terrestrial C
3
plant
compiled from the literature and this study. The magnitude of the terrestrial
CIE is larger compared to the marine CIE magnitude. The Wilcoxon test
suggests that the CIE magnitude between marine and terrestrial substrate
is statistically different (Supplementary Table 1, p < 0.001). A Kruskal-
Wallis test further shows the significant difference of CIE among marine
carbonate, terrestrial bulk organic matter and C
3
plant groups (p< 0.001).
The “n”value represents the number of δ13C profiles.
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-22298-7 ARTICLE
NATURE COMMUNICATIONS | (2021) 12:2137 | https://doi.org/10.1038/s41467-021-22298-7 | www.nature.com/naturecommunications 3
Content courtesy of Springer Nature, terms of use apply. Rights reserved

values (883–1325 ppmv). Similarly, the δ13C
phytane
-based pCO
2
estimates show that CO
2
levels during Changhsingian could have
ranged from 873 to 1085 ppmv15. The few earliest Triassic peak
pCO
2
estimates from stomatal17 and phytane15 proxies show
significant variation (600–2100 ppmv). Simulations with various
climate models (e.g. carbon cycle box modelling18,19 and
cGENIE20) show major variability of peak pCO
2
values
(1000–9380 ppmv; Supplementary Table 3), using a large range of
assumed background pCO
2
.
Several effects, especially diagenesis40, chemical treatment41,
plant taxonomic changes42 and precipitation43,44 can influence
the δ13C
p
and consequently affect pCO
2
estimates. The original
signals of δ13C
p
values can potentially be altered by diagenesis
during burial40 and chemical treatment during sample
preparation41. However, the method we use to calculate
palaeo-pCO
2
considers a relative change of the Δ13Cthat
minimizes these biases (Methods). A dramatic plants turnover
occurred in southwestern China during the PTME, with a
Gigantopteris flora (spore plant) in the Xuanwei and basal
Kayitou formations replaced by an Tomiostrobus (spore plant)
and Peltaspermum (seed plant) dominated flora45. Experiments
on modern plants indicate lower Δ13C in seed plants than in
spore plants42. Using a plant assemblage including a mixture of
different taxa and plant tissues is better than using single
species and plant remains when using δ13C
p
as pCO
2
proxy33.
In this study we used a mixture of different plant tissues (i.e.
cuticle, charred wood and non-charred wood), which very likely
includes different plant taxa.
An increase of the mean annual precipitation (MAP) can also
increase Δ13C44,46. This effect is negligible in sites experiencing
high precipitation (>1500 mm/yr)47, such as the studied area in
southwestern China, which was a humid, equatorial peatland
during the PTME45. The plant community changed from
Gigantopteris flora-dominated rainforest ecosystem to isoetalean-
dominated (lycophyte) herbaceous vegetation that inhabited the
surrounding margins of coastal oligotrophic lakes, which indicate
fairly constant precipitation regimes during the PTME
interval48,49. The sedimentology of the Xuanwei and Kayitou
formations suggests there was no significant precipitation change
across the mass extinction (Supplementary Fig. 8; ref. 50). In
contrast, low MAP can explain the smaller magnitude of the CIE
(<3%) recorded at the PTME in the semi-arid locations of Karoo
Basin and North China31. In summary, the persistently humid
condition in southwestern China was unlikely to have affected
plant Δ13C, thus the pCO
2
estimates are considered robust. A
Monte Carlo method has been applied to evaluate the uncertain-
ties (Supplementary Information; Supplementary Fig. 9), which
reveals that the uncertainty in the pCO
2
increases with increasing
pCO
2
, as seen in the previous studies51.
Potential source of 13C-depleted carbon during the PTME.The
ultimate source of 13C-depleted carbon capable to trigger the
observed negative CIE, is widely debated. Several climate models of
varying complexities (e.g. simple box models18,19 and cGENIE20)
use different light carbon sources to fittheδ13C of marine carbo-
nates (Supplementary Table 3). Proposed 13C-depleted carbon
sources include biotic or thermogenic methane (δ13C≈−60% to
−40%;e.g.ref.18), CO
2
from thermal metamorphism or rapid
oxidation of organic-rich rock (δ13C≈−25%;e.g.ref.6,19,52,53), and
Cc
Cy
Cm
Hc
Ct
Hp
Istaeschei and Iisarcica
/DWH3HUPLDQ (DUO\7ULDVVLF
     
           
FXWLFOH
FKDUUHGZRRG
QRQFKDUUHGZRRG
      
6RXWK&KLQD
,UDQ
$UPHQLD
    
VWRPDWD
SDODHRVRO
SK\WDQH
    
δ&FDUEδ&Sp&2SSPY
Age(Ma)
ಽ)(ಽ)667 ഒ 6SHFLHVULFKQHVV
Conodont
zone
Fig. 3 Summary of Permian–Triassic boundary proxy data and reconstructed pCO
2
changes. The radiometric ages are from the Meishan section30.
Conodont zones are those of the Meishan section. Global marine carbonate carbon isotope (δ13C
carb
) compiled from ten sections (Methods). Land C
3
plant
carbon isotopes profile (δ13C
p
) is from the four study sections in southwestern China (Fig. 1). Sea surface temperature (SST) was calculated based on
conodont δ18O values from South China (Meishan and Shangsi)1–3, Iran (Kuh-e-Ali Bashi and Zal)4and Armenia (Chanakchi)5. The blue, green and red
lines represent the LOESS fit curve for δ13C
carb
,δ13C
p
and SST, respectively, while light blue, green, and red shaded area represent 68% confidence intervals
(standard errors calculated from LOESS). Reconstruction of atmospheric pCO
2
based on carbon isotope fractionation in C
3
land plant (on a log scale).
Median values of the 10,000 re-samplings determined by Monte Carlo uncertainty propagation are shown as dark gray line. The 68% confidence intervals
for pCO
2
are showed as light gray shaded area (lower limit and upper limit represent the 16th and 84th percentiles respectively). Previous reported pCO
2
estimates based on stomata16,17, palaeosol carbonates13,14 and phytane15 are shown as points with error bar (Supplementary Table 2). Marine species
richness data show the two pulse mass extinction8.I.―Isarcicella;C. c―Clarkina changxingensis;C. y―Clarkina yini;C. m―Clarkina meishanensis;
H. c―Hindeodus changxingensis;C. t―Clarkina taylorae;H. p―Hindeodus parvus.
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-22298-7
4NATURE COMMUNICATIONS | (2021) 12:2137 | https://doi.org/10.1038/s41467-021-22298-7 | www.nature.com/naturecommunications
Content courtesy of Springer Nature, terms of use apply. Rights reserved

volcanic CO
2
(δ13C≈−6%;e.g.ref.54)oracombinationofthese
sources20. We performed a simple carbon isotope mass balance to
evaluate the most likely 13C-depleted carbon source55,56.Underthe
assumption of four possible 13C-depleted carbon sources (i.e. vol-
canic CO
2
,organicmatter,thermogenic methane, and biogenic
methane), the mass of released carbon was calculated (Fig. 4)
and compared with our pCO
2
rise estimates (2081 +4764/−1193
ppmv). Our calculated pCO
2
mostly falls within the range of model
results for organic matter and methane release scenarios (Fig. 4),
supporting the hypothesis that these more 13C-depleted sources
than volcanic CO
2
are required to contribute to the global carbon
cycle perturbation. There are some U-Pb dating7and field
evidence6,57 show that the organic-rich sediment intruded by
Siberian Traps sill likely provided massive 13C-depleted CO
2
and
thermogenic methane, which may have been the ultimate trigger of
the global CIE and significant increase in atmospheric CO
2
.How-
ever, due to the limitation of the C
3
plant proxy, the uncertainty of
pCO
2
is significantly larger at high CO
2
levels (Supplementary
Fig. 9). Therefore, volcanic CO
2
source could still have made a
contribution to the global carbon cycle perturbation.
The best estimates for mass of added carbon based on a 3.5%
carbonate CIE magnitude and a source with δ13Cof−25% to
−60%, suggest that at least 3900~12,000 Gt carbon were added
into the ocean-atmosphere system during the PTME. Previous
estimates (15,000–20,000 Gt C) were based on an assumed ca.
5.5% negative shift of C
3
plant in simple mass balance
calculations32 and might therefore have overstated the amount
of added carbon. The ca. 7% CIE in C
3
plants, amplified by pCO
2
increase, also produces an over-estimate in the mass balance
calculation (Fig. 4). Our estimates of the amount of injected
carbon are also smaller than those calculated by the cGENIE
climate model (7,000~22,400 Gt C)20, because the ~5%
magnitude of δ13C
carb
from Meishan used in the calculations is
too large compared to the global carbonate records. However,
simple mass balance calculations don’t consider global carbon
cycle fundamental processes and changes through time, like
carbon weathering and burial rates during the studied interval. In
addition, the size of the DIC reservoir is usually assumed to be the
size of the background surface carbon reservoir, because of poor
understanding of atmosphere carbon reservoir size58. These
limitations might lead to an underestimate of the total mass of
added CO
2
.
Carbon emission caused prolonged high pCO
2
and high
temperature (ca. 35 °C) during the earliest Triassic (H. parvus and
I. isarcica zones) and may have lasted for > 500 kyr (Fig. 3). This
lengthy phase of extreme warmth likely implies prolonged carbon
emissions into ocean-atmosphere system from continued erup-
tion of the Siberian Traps volcanism, and/or reduced carbon
sequestration rate, potentially due to lower consumption of
atmospheric CO
2
through reduced organic carbon burial and the
possible failure of the silicate weathering thermostat59.
Methods
Sample treatment and analysis. In total, 68 samples from Chinahe, 41 samples
from ZK4703 and 40 samples from Jiucaichong were analyzed for bulk organic
carbon isotopes. Samples were crushed to fine powder (<200 mesh), and ∼2g
powder were weighed, placed into a centrifuge tube and treated with 3 mol/L HCl
for 24 h to remove the carbonate. Then the treated samples were rinsed with
ultrapure water repeatedly until neutralized and finally dried at 35 °C. For C
3
plants
δ13C analysis, 45 samples from Chinahe, 26 samples from ZK4703, 30 samples
from Jiucaichong and 13 samples from Chahe were treated with concentrated HCl
and HF, then sieved over 500 μm and a 100 μm mesh screen to get the 100~500 μm
particles. C
3
plant fragments, including cuticle, non-charred wood and charred
wood (charcoal), were picked under the microscope. The δ13C
org
analyses were
performed by using an elemental analyzer (EA) coupled to an isotope ratio mass
spectrometer (Thermo Delta V Advantage) at the State Key Laboratory of Bio-
geology and Environmental Geology of the China University of Geosciences
(Wuhan). The results were calibrated using certified secondary references stan-
dards: USGS40 (δ13C=−26.39%) and UREA (δ13C=−37.32%) and given in per
mil (%) relative to Vienna Pee Dee Belemnite (VPDB) with analytical precision
better than ± 0.2%. A Multi EA 4000-analyzer was used for TOC at China Uni-
versity of Geosciences (Wuhan), yielding an analytical precision of 1.5%.
Carbon isotope compilation and estimate of CIE magnitude. In order to esti-
mate a reliable magnitude of the CIE, the carbon isotope profiles that record a
roughly complete CIE shape with pre-CIE and CIE body are selected in our study.
The compilation consists of 69 marine carbonate carbon isotope (δ13C
carb
) profiles
and 38 terrestrial δ13C
org
profiles. The δ13C
carb
profiles recording complete nega-
tive CIE are from Eastern Palaeotethys (n=29), Western Palaeotethys (n=19),
Central Palaeotethys (n=4), Northern Neotethys (n=4), Southern Neotethys
(n=10) and Panthalassa (n=3). Bulk marine organic matter δ13C records were
not included, because they often represent a mix of various organic components
(both marine and terrestrial). The sedimentary facies belong to a range of shallow
shelf, deep shelf, and slope environments. The few reported δ13C
carb
profiles from
deep basins are ignored in this compilation (e.g. Shangsi section), because the
elevated water stratification and large vertical δ13C DIC gradients at deep basin
sites during Permian–Triassic crisis could cause large CIE magnitudes60. A total of
38 terrestrial δ13C
org
records are reviewed from eight terrestrial basins including
western Guizhou and eastern Yunnan in southwestern China (n=14), Junggar
Basin (n=3), Turpan Basin (n=1), North China (n=3), Central European Basin
(n=1), Bowen Basin (n=2), Sydney Basin (n=7), Karoo Basin (n=1) and three
oceanic regions where organic matter (OM) in samples are C
3
plants or a mix of
organic matter dominated by C
3
plants including South China (n=1), Boreal
realm (n=2) and South Neotethys (n=3). Among these terrestrial δ13C
org
pro-
files, there are nine records of δ13C records from C
3
plant (5 δ13C
wood
and 4
δ13C
cuticle
) from Meishan section, Amb section in South Neotethys, southwestern
China, and others are all bulk δ13C
org
profiles.
In order to demonstrate the difference of marine and terrestrial CIE
magnitudes, carbon isotope values immediately before the CIE (δ13C
background
) and
peak values (δ13C
peak
) are used to calculate the magnitude of the CIE (δ13C
peak
−
δ13C
background
). Note that 31 pairs of δ13C
background
and δ13C
peak
values are from
marine sections that are well constrained by latest Permian conodont occurrences
(e.g. C. changxingensis,H. praeparvus,H. latidentatus zones) and earliest Triassic
conodont (H. parvus and I. isarcica zones) occurrences. To test if the discrepancy
of the CIE magnitude in different substrates (marine carbonate, terrestrial bulk
OM, terrestrial C
3
plant tissues) is statistically significant, we used a non-parameter
Kruskal-Wallis test (function kruskal.test), using R software. The Wilcoxon
(function wilcox.test) test was performed in R software to determine whether
means of two independent groups (marine vs. terrestrial) are equal or not. Boxplots
were drawn to visualize discrepancy in CIE magnitude of different substrates. All
statistical analyses and graphing functions were undertaken using R.
C
3
plant proxy. The carbon isotope fractionation in C
3
plants (Δ13C) and atmo-
spheric pCO
2
is described as a hyperbolic relationship22–24,61:
Δ13C¼AðÞBðÞpCO2þC

AþBðÞpCO2þC
 ð1Þ
C3 plant proxy
0
20000
40000
60000
80000
100000
120000
140000
0
6000
12000
18000
24000
30000
36000
42000
−1 −2 −3 −4 −5 −6 −7 −8
CIE magnitude (‰)
Mass of added carbon (Gt)
pCO2 increase (ppmv)
Volcanic CO2 (−6‰)
Organic matter (−25‰)
Thermogenic methane (−40‰)
Biogenic methane (−60‰)
Fig. 4 Mass of added carbon estimated from carbon isotope mass
balance calculation. Four different scenarios including volcanic CO
2
(δ13C=
−6%), organic matter (δ13C=−25%), thermogenic methane (δ13C=−40%)
and biogenic methane hydrate (δ13C=−60%). Second y-axis converts the
massofaddedcarbontoanincreaseinatmosphericpCO
2
based on the earth
system model (1 Gt C =0.3 ppmv CO
2
)55,56. Gray shaded area represents the
68% confidence intervals of pCO
2
increase (2081 +4764/−1193 ppmv)
estimated from the C
3
plant proxy.
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-22298-7 ARTICLE
NATURE COMMUNICATIONS | (2021) 12:2137 | https://doi.org/10.1038/s41467-021-22298-7 | www.nature.com/naturecommunications 5
Content courtesy of Springer Nature, terms of use apply. Rights reserved

The original δ13C signals in C
3
plant can be altered by several effects (e.g.
diagenesis40, chemical treatments41), that potentially influence pCO
2
calculations.
In order to minimize this effect, the data set is analyzed by a relative change in the
Δ13C value between the time of interest (t) and a reference time (t=0), designated
as Δ(Δ13C):
ΔΔ
13C

¼Δ13CtðÞ Δ13Ct¼0ðÞ ð2Þ
which can be expanded as:
ΔðΔ13CÞ¼
A
ðÞ
B
ðÞpCO2t
ðÞþC

AþBðÞpCO2tðÞþC


A
ðÞ
B
ðÞpCO2t¼0
ðÞ
þC

AþðBÞpCO2t¼0ðÞ
þC

ð3Þ
By rearranging Eq. (3), pCO
2(t)
at any given time can be calculated by
where A,B,Care curve fitting parameters. Values for Aand Bare 28.26 ± 0 and
0.22 ± 0.028, respectively51, which could produce more robust pCO
2
estimates
compared with other parameter values in subsequent research33. The C is the
function of the Aand Bvalues [C=A×4.4/((A−4.4) × B)]. The pCO
2(t=0)
is equal
to the pCO
2
level before the negative CIE, determined from independent stomatal
proxies based on fossil conifers from the Dalong Formation in South China16.
Because of the good age control (Clarkina changxingensis conodont zone), reliable
taxonomy and calculation method, these stomatal estimates are considered as
robust pCO
2
estimates before CIE. The mean value for the stomatal estimates is
425 ± 68 ppmv set as pCO
2(t=0)
. The Δ13C is the carbon isotope fractionation
between atmospheric CO
2
and plant organic carbon (Δ13C¼ðδ13CCO2δ13CpÞ=
ð1þδ13Cp=1000Þ). Thus, the Δ(Δ13C) can be calculated by
ΔΔ
13C

¼δ13CCO2ðtÞδ13 CpðtÞ

=1þδ13CpðtÞ=1000

δ13CCO2ðt¼0Þδ13 Cpðt¼0Þ

=1þδ13Cpðt¼0Þ=1000

ð5Þ
where δ13Cpðt¼0Þand δ13C
p(t)
are δ13C values in C
3
plant at reference time (t=0)
and the time of interest (t). The values for δ13 Cpðt¼0Þis determined as −24.42 ±
0.5%, whose age equals to Clarkina changxingensis conodont zone that occurred
slightly earlier than the onset of the CIE16. We suggest that a mixture of δ13CinC
3
plant cuticle, charred wood and non-charred wood from southwestern China
(without δ13C of bulk OM) provides the best choice as δ13C
p(t)
input data for three
reasons. Firstly, using the δ13C values of C
3
plant tissues (e.g. cuticle, wood) can
minimize the influence of varying OM sources from mixed soils and sediments.
Several previous reports on terrestrial δ13C
org
in western Guizhou and eastern
Yunnan, South China have recorded the negative CIE27,28,62,63, but all the data are
not δ13C from C
3
plants and not suitable for pCO
2
calculation. Secondly, Eq. (1) is
based on the combination of carbon isotope from stem and leaf tissues of chamber
plants. Thirdly, the mixture of different micro plant tissues (i.e. cuticle, charred
wood and non-charred wood) would contain different plant fossils species that is
suggested to be better than a single species approach when applying this proxy33.
The δ13CCO2ðt¼0Þand δ13 CCO2ðtÞare δ13C values in atmospheric CO
2
at
reference time (t=0) and the time of interest (t). The temperature (T) dependent
carbon isotope fractionation between dissolved inorganic carbon (DIC) and
atmospheric CO
2
64 can be used to calculate δ13CCO2.
δ13CCO2¼δ13 CDIC 0:91´0:1141 ´Tþ10:78ðÞþ0:08 ´0:052 ´Tþ7:22ðÞðÞ
ð6Þ
where Tis the sea surface temperature determined from oxygen isotopes of conodont
fossils1–5.δ13C
DIC
can be estimated from marine δ13C
carb
(δ13C
DIC
=δ13C
carb
−1%),
because the carbon isotope fractionation between marine carbonate (δ13C
carb
)and
dissolved inorganic carbon (δ13C
DIC
) is constant and independent of temperature
(~1%)65. Among the global CIE compilations, 10 marine, high-resolution, non-
basinal δ13C
carb
profiles are well constrained by detail conodont zones including
Meishan30, Nhi Tao66, Yangou67 in eastern Palaeotethys; Zal3, Kuh-e-Ali Bashi3in
central Palaeotethys; Bálvány North68 in western Paloetethys; Shahreza12,Abadeh
12
in northern Neotethys; Wadi Shahha69, Wenbudangsang70 in Southern Neotethys.
Thus, we integrated these 10 δ13C
carb
profiles together as a global marine δ13C
carb
profile combined with a U-Pb age model30 and high-resolution conodont zones
(conodont zones from Meishan are selected as standard71,72). LOESS curves with
0.002 Myr spacing were fitted to the integration of global marine δ13C
carb
.Ateach
0.002 Myr time step, the probability maximum value and standard error are identified
and served as δ13C
carb
input parameters in calculations. The best degree of smoothing
for LOESS fitting was determined using cross-validation method in package
fANCOVA. To eliminate the potential for an uneven distribution of δ13C
carb
data, we
also applied a LOESS fitting based on an 80% subsample of all data. In addition, the
δ13C
carb
data from Meishan (n=199) is the most abundant of all the δ13C
carb
data
(n=707), thus, we performed a LOESS fitting based on δ13C
carb
data without
Meishan data (Supplementary Fig. 4).
In order to calculate pCO
2(t)
, we need to align global marine δ13C
carb
profiles
and δ13C
p
based on same age model. The nearly same CIE curves were divided into
four stages in carbonate and C
3
plants records to ensure the correlation between
marine and terrestrial carbon isotope profiles. The age model for four sections is
showed in Supplementary Information. In addition, the LOESS method with 0.002
Myr spacing was also performed in δ13C
p
and temperature data to get the
probability maximum value and standard error at each 0.002 Myr time step. The
Monte Carlo method was employed to propagate input error51 by the propagate
package in R. All the input parameters were assumed to be Gaussian distributed
with mean and standard deviations listed in Supplementary Table 4. 10,000 values
for each input parameters were randomly sampled to calculate 10,000 values for
each pCO
2(t)
. The invalid pCO
2(t)
values (i.e., pCO
2(t)
< 0 or >106ppmv) were
excluded. The 16th and 84th percentiles of the remaining estimates were determined
to construct the 68% confidence interval. The positive error of the reconstructed
pCO
2(t)
value represents the difference between the 84th percentile value and the
median, and the negative error represents the difference between the 16th
percentile value and the median. The sensitivity analysis of C
3
plant proxy is
discussed in Supplementary information and Supplementary Fig. 9.
Carbon isotope mass balance. This model used to evaluate the light carbon
source, following mass balance equation is modified from McInerney and Wing55:
Madded ¼
CIE ´Mbackground

δ13Cpeak δ13 Cadded
 ð7Þ
where M
added
is the mass of carbon added into atmosphere-ocean system carbon
emission. The M
background
represents initial carbon reservoir size during
Permian–Triassic including ocean and atmosphere carbon inventory, but domi-
nated by the ocean reservoir. Thus, the M
background
is assumed to be the initial
marine DIC reservoir size ranging from 66,000 to 82,000 Gt58,73. CIE represents
the global magnitude of CIE controlled only by release of light carbon effect, which
is set as a series values from −1% to −8%. The peak δ13C value (δ13C
peak
) at the
event is calculated by initial isotopic composition of global carbon reservoir
(δ13C
background
) and CIE (δ13C
peak
=δ13C
background
+CIE). The δ13C
background
is
assumed to be the initial isotopic composition of DIC reservoir 2.2% that is esti-
mated from global marine δ13C
carb
profiles (age >252.104 Ma). The δ13C
added
is the
δ13C value of the carbon source causing the CIE. Four kinds of carbon sources are
involved including biogenic methane buried in permafrost or seafloor (δ13C=
−60%), thermogenic methane (δ13C=−40%), thermal metamorphism or rapid
oxidation of organic-rich rock (δ13C=−25%) and CO
2
released from direct
volcanic eruption (−6%). Finally, the increased pCO
2
is estimated from M
added
(1 Gt C =0.3 ppmv; ref. 56), and compared with reconstructed atmospheric CO
2
levels from C
3
plant proxy.
Data availability
The authors declare that all data supporting the findings of this study are available within
the paper and its supplementary file.
Code availability
R code to run the model is available from D.L. Chu on request.
Received: 4 May 2020; Accepted: 26 February 2021;
References
1. Joachimski, M. M. et al. Climate warming in the latest Permian and the
Permian-Triassic mass extinction. Geology 40, 195–198 (2012).
2. Sun, Y. et al. Lethally hot temperatures during the Early Triassic greenhouse.
Science 338, 366–370 (2012).
3. Chen, B. et al. Permian ice volume and palaeoclimate history: oxygen isotope
proxies revisited. Gondwana Reso 24,77–89 (2013).
4. Schobben,M.,Joachimski,M.M.,Korn,D.,Leda,L.&Korte,C.
Palaeotethys seawater temperature rise and an intensified hydrological cycle
following the end-Permian mass extinction. Gondwana Reso 26, 675–683
(2014).
5. Joachimski, M. M., Alekseev, A. S., Grigoryan, A. & Gatovsky, Y. A. Siberian
Trap volcanism, global warming and the Permian-Triassic mass extinction:
pCO2ðtÞ¼ΔΔ
13CðÞA2þΔΔ
13CðÞABpCO2t¼0ðÞ
þ2ΔΔ
13CðÞABCþΔΔ
13CðÞB2CpCO2t¼0ðÞ
þΔΔ
13CðÞB2C2þA2BpCO2t¼0ðÞ
A2BΔΔ
13CðÞABΔΔ
13CðÞB2pCO2t¼0ðÞ
ΔΔ
13CðÞB2Cð4Þ
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-22298-7
6NATURE COMMUNICATIONS | (2021) 12:2137 | https://doi.org/10.1038/s41467-021-22298-7 | www.nature.com/naturecommunications
Content courtesy of Springer Nature, terms of use apply. Rights reserved

new insights from Armenian Permian-Triassic sections. Geol. Soc. Am. Bull.
132, 427–443 (2020).
6. Svensen, H. et al. Siberian gas venting and the end-Permian environmental
crisis. Earth Planet. Sci. Lett. 277, 490–500 (2009).
7. Burgess, S. D., Muirhead, J. D. & Bowring, S. A. Initial pulse of Siberian Traps sills
as the trigger of the end-Permian mass extinction. Nat. Commun. 8, 164 (2017).
8. Song, H. J., Wignall, P. B., Tong, J. & Yin, H. Two pulses of extinction during
the Permian-Triassic crisis. Nat. Geosci. 6,52–56 (2013).
9. Fielding, C. R. et al. Age and pattern of the southern high-latitude continental
end-Permian extinction constrained by multiproxy analysis. Nat. Commun.
10, 385 (2019).
10. Chu, D. et al. Ecological disturbance in tropical peatlands prior to marine
Permian-Triassic mass extinction. Geology 48, 288–292 (2020).
11. Black, B. A. et al. Systemic swings in end-Permian climate from Siberian Traps
carbon and sulfur outgassing. Nat. Geosci. 11, 949–954 (2018).
12. Korte, C. & Kozur, H. W. Carbon-isotope stratigraphy across the Permian-
Triassic boundary: a review. J. Asian Earth Sci 39, 215–235 (2010).
13. Ekart, D. D., Cerling, T. E., Montanez, I. P. & Tabor, N. J. A 400 million year
carbon isotope record of pedogenic carbonate: Implications for
paleoatmospheric carbon dioxide. Am. J. Sci. 299, 805–827 (1999).
14. Gastaldo, R. A., Knight, C. L., Neveling, J. & Tabor, N. J. Latest Permian
paleosols from Wapadsberg Pass, South Africa: implications for
Changhsingian climate. Geol. Soc. Am. Bull. 126, 665 (2014).
15. Witkowski, C. R., Weijers, J. W. H., Blais, B., Schouten, S. & Sinninghe
Damsté, J. S. Molecular fossils from phytoplankton reveal secular Pco
2
trend
over the Phanerozoic. Sci. Adv. 4, eaat4556 (2018).
16. Li, H., Yu, J., McElwain, J. C., Yiotis, C. & Chen, Z. Reconstruction of
atmospheric CO
2
concentration during the late Changhsingian based on fossil
conifers from the Dalong Formation in South China. Palaeogeogr.
Palaeoclimatol. Palaeoecol. 519,37–48 (2019).
17. Retallack, G. J. & Conde, G. D. Deep time perspective on rising atmospheric
CO
2
.Global Planet. Change 189, 103177 (2020).
18. Berner, R. A. Examination of hypotheses for the Permo-Triassic boundary
extinction by carbon cycle modeling. Proc. Natl. Acad. Sci. U.S.A. 99,
4172–4177 (2002).
19. Payne, J. L. & Kump, L. R. Evidence for recurrent Early Triassic massive
volcanism from quantitative interpretation of carbon isotope fluctuations.
Earth Planet. Sci. Lett. 256, 264–277 (2007).
20. Cui, Y., Kump, L. R. & Ridgwell, A. Initial assessment of the carbon emission
rate and climatic consequences during the end-Permian mass extinction.
Palaeogeogr. Palaeoclimatol. Palaeoecol. 389, 128–136 (2013).
21. Keeling, R. F. & Keeling, C. D. Atmospheric Monthly In Situ CO
2
Data-Mauna
Loa observatory, Hawaii. Scripps CO
2
Program Data (UC San Diego, Library
Digital Collections, 2017).
22. Schubert, B. A. & Jahren, A. H. The effect of atmospheric CO
2
concentration
on carbon isotope fractionation in C
3
land plants. Geochim. Cosmochim. Acta
96,29–43 (2012).
23. Schubert, B. A. & Jahren, A. H. Global increase in plant carbon isotope
fractionation following the Last Glacial Maximum caused by increase in
atmospheric pCO
2
.Geology 43, 435–438 (2015).
24. Cui, Y. & Schubert, B. A. Towards determination of the source and magnitude
of atmospheric pCO
2
change across the early Paleogene hyperthermals. Global
Planet. Change 170, 120–125 (2018).
25. Barral, A., Gomez, B., Fourel, F., Daviero-Gomez, V. & Lécuyer, C. CO
2
and
temperature decoupling at the million-year scale during the Cretaceous
Greenhouse. Sci. Rep. 7, 8310 (2017).
26. Ruebsam, W., Reolid, M. & Schwark, L. δ13C of terrestrial vegetation records
Toarcian CO
2
and climate gradients. Sci. Rep. 10, 117 (2020).
27. Shen, S. Z. et al. Calibrating the End-Permian Mass Extinction. Science 334,
1367–1372 (2011).
28. Cui, Y. et al. Carbon cycle perturbation expressed in terrestrial Permian-
Triassic boundary sections in South China. Global Planet. Change 148,
272–285 (2017).
29. Hermann, E. et al. A close-up view of the Permian-Triassic boundary based on
expanded organic carbon isotope records from Norway (Trøndelag and
Finnmark Platform). Global Planet. Change 74, 156–167 (2010).
30. Burgess, S. D., Bowring, S. & Shen, S. High-precision timeline for Earth’smost
severe extinction. Proc. Natl Acad. Sci. USA 111, 3316–3321 (2014).
31. Wu, Y. et al. Organic carbon isotopes in terrestrial Permian-Triassic boundary
sections of North China: Implications for global carbon cycle perturbations.
Geol. Soc. Am. Bull. 132, 1106–1118 (2020).
32. Schneebeli-Hermann, E. et al. Evidence for atmospheric carbon injection
during the end-Permian extinction. Geology 41, 579–582 (2013).
33. Porter, A. S. et al. Testing the accuracy of new paleoatmospheric CO
2
proxies
based on plant stable carbon isotopic composition and stomatal traits in a
range of simulated paleoatmospheric O
2
:CO
2
ratios. Geochim. Cosmochim.
Acta 259,69–90 (2019).
34. Knobbe, T. K. & Schaller, M. F. A tight coupling between atmospheric pCO
2
and sea-surface temperature in the Late Triassic. Geology 46, 43 (2017).
35. Cui, Y. & Kump, L. R. Global warming and the end-Permian extinction event:
Proxy and modeling perspectives. Earth-Sci. Rev. 149,5–22 (2015).
36. Stocker, T. F. et al. Climate change 2013: The physical science basis.
Contribution of Working Group I to the Fifth Assessment Report of the
Intergovernmental Panel on Climate Change 1535 (2013).
37. Penn, J. L., Deutsch, C., Payne, J. L. & Sperling, E. A. Temperature-dependent
hypoxia explains biogeography and severity of end-Permian marine mass
extinction. Science 362, eaat1327 (2018).
38. Foster, G. L., Royer, D. L. & Lunt, D. J. Future climate forcing potentially
without precedent in the last 420 million years. Nat. Commun. 8, 14845
(2017).
39. Breecker, D. O., Sharp, Z. D. & McFadden, L. D. Atmospheric CO
2
concentrations during ancient greenhouse climates were similar to those
predicted for A.D. 2100. Proc. Natl Acad. Sci. USA 107, 576–580 (2010).
40. Lukens, W. E., Eze, P. & Schubert, B. A. The effect of diagenesis on carbon
isotope values of fossil wood. Geology 47, 987–991 (2019).
41. Barral, A., Lécuyer, C., Gomez, B., Fourel, F. & Daviero-Gomez, V. Effects of
chemical preparation protocols on δ13C values of plant fossil samples.
Palaeogeogr. Palaeoclimatol. Palaeoecol. 438, 267–276 (2015).
42. Porter, A. S., Yiotis, C., Montañez, I. P. & McElwain, J. C. Evolutionary
differences in Δ13C detected between spore and seed bearing plants following
exposure to a range of atmospheric O
2
:CO
2
ratios; implications for
paleoatmosphere reconstruction. Geochim. Cosmochim. Acta 213, 517–533
(2017).
43. Lomax, B. H., Lake, J. A., Leng, M. J. & Jardine, P. E. An experimental
evaluation of the use of Δ13C as a proxy for palaeoatmospheric CO
2
.Geochim.
Cosmochim. Acta 247, 162–174 (2019).
44. Schlanser, K. et al. On geologic timescales, plant carbon isotope fractionation
responds to precipitation similarly to modern plants and has a small negative
correlation with pCO
2
.Geochim. Cosmochim. Acta 270, 264–281 (2020).
45. Chu, D. et al. Biostratigraphic correlation and mass extinction during the
Permian-Triassic transition in terrestrial-marine siliciclastic settings of South
China. Global Planet. Change 146,67–88 (2016).
46. Diefendorf, A. F., Mueller, K. E., Wing, S. L., Koch, P. L. & Freeman, K. H.
Global patterns in leaf 13C discrimination and implications for studies of past
and future climate. Proc. Natl. Acad. Sci. U.S.A. 107, 5738–5743 (2010).
47. Basu, S., Ghosh, S. & Sanyal, P. Spatial heterogeneity in the relationship
between precipitation and carbon isotopic discrimination in C
3
plants:
Inferences from a global compilation. Global Planet. Change 176, 123–131
(2019).
48. Yu, J. et al. Vegetation changeover across the Permian–Triassic Boundary in
Southwest China: Extinction, survival, recovery and palaeoclimate: a critical
review. Earth-Sci. Rev. 149, 203–224 (2015).
49. Feng, Z. et al. From rainforest to herbland: New insights into land plant
responses to the end-Permian mass extinction. Earth-Sci. Rev. 204, 103153
(2020).
50. Wignall, P. B. et al. Death in the shallows: the record of Permo-Triassic mass
extinction in paralic settings, southwest China. Global Planet. Change 189,
103176 (2020).
51. Cui, Y. & Schubert, B. A. Quantifying uncertainty of past pCO
2
determined
from changes in C
3
plant carbon isotope fractionation. Geochim. Cosmochim.
Acta 172, 127–138 (2016).
52. Dal Corso, J. et al. Permo–Triassic boundary carbon and mercury cycling
linked to terrestrial ecosystem collapse. Nat. Commun. 11, 2962 (2020).
53. Jurikova, H. et al. Permian–Triassic mass extinction pulses driven by major
marine carbon cycle perturbations. Nat. Geosci. 13, 745–750 (2020).
54. Gales, E., Black, B. & Elkins-Tanton, L. T. Carbonatites as a record of the
carbon isotope composition of large igneous province outgassing. Earth
Planet. Sci. Lett. 535, 116076 (2020).
55. McInerney, F. A. & Wing, S. L. The Paleocene-Eocene Thermal Maximum: a
perturbation of carbon cycle, climate, and biosphere with implications for the
future. Annu. Rev. Earth Planet. Sci. 39, 489–516 (2011).
56. Panchuk, K., Ridgwell, A. & Kump, L. R. Sedimentary response to Paleocene-
Eocene Thermal Maximum carbon release: a model-data comparison. Geology
36, 315–318 (2008).
57. Elkins-Tanton, L. T. et al. Field evidence for coal combustion links the 252 Ma
Siberian Traps with global carbon disruption. Geology 48, 986–991 (2020).
58. Payne, J. L. et al. Calcium isotope constraints on the end-Permian mass
extinction. Proc. Natl. Acad. Sci. USA 107, 8543–8548 (2010).
59. Kump, L. R. Prolonged Late Permian–Early Triassic hyperthermal: failure of
climate regulation? Phil. Trans. R. Soc. A: Math. Phys. Eng. Sci. 376, 20170078
(2018).
60. Song, H. Y. et al. Large vertical δ13C DIC gradients in Early Triassic seas of the
South China craton: implications for oceanographic changes related to
Siberian Traps volcanism. Global Planet. Change 105,7–20 (2013).
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-22298-7 ARTICLE
NATURE COMMUNICATIONS | (2021) 12:2137 | https://doi.org/10.1038/s41467-021-22298-7 | www.nature.com/naturecommunications 7
Content courtesy of Springer Nature, terms of use apply. Rights reserved

61. Schubert, B. A. & Jahren, A. H. Incorporating the effects of photorespiration
into terrestrial paleoclimate reconstruction. Earth-Sci. Rev. 177, 637–642
(2018).
62. Zhang, H. et al. The terrestrial end-Permian mass extinction in South China.
Palaeogeogr. Palaeoclimatol. Palaeoecol. 448, 108–124 (2016).
63. Shen, J. et al. Mercury evidence of intense volcanic effects on land during the
Permian-Triassic transition. Geology 47, 1117–1121 (2019).
64. Tipple, B. J., Meyers, S. R. & Pagani, M. Carbon isotope ratio of Cenozoic CO
2
:
a comparative evaluation of available geochemical proxies. Paleoceanography
25, PA3202 (2010).
65. Romanek, C. S., Grossman, E. L. & Morse, J. W. Carbon isotopic fractionation
in synthetic aragonite and calcite: Effects of temperature and precipitation
rate. Geochim. Cosmochim. Acta 56, 419–430 (1992).
66. Algeo, T. J., Ellwood, B., Nguyen, T. K. T., Rowe, H. & Maynard, J. B. The
Permian–Triassic boundary at Nhi Tao, Vietnam: Evidence for recurrent
influx of sulfidic watermasses to a shallow-marine carbonate platform.
Palaeogeogr. Palaeoclimatol. Palaeoecol. 252, 304–327 (2007).
67. Song, H. et al. The large increase of 13C
carb
-depth gradient and the end-
Permian mass extinction. Sci. China Earth Sci 55, 1101–1109 (2012).
68. Schobben, M. et al. Volatile earliest Triassic sulfur cycle: a consequence of
persistent low seawater sulfate concentrations and a high sulfur cycle turnover
rate? Palaeogeogr. Palaeoclimatol. Palaeoecol. 486,74–85 (2017).
69. Clarkson, M. O. et al. A new high-resolution δ13C record for the Early
Triassic: Insights from the Arabian Platform. Gondwana Res. 24, 233–242
(2013).
70. Ji, C. et al. 13C-18O Isotopic anomalous study of the carbonate rock at the
Wenbudangsang PTB Section,Tibet. Acta Geologica Sin. 92, 2018–2027
(2018).
71. Yin, H. et al. The end-Permian regression in South China and its implication
on mass extinction. Earth-Sci. Rev. 137,19–33 (2014).
72. Yuan, D. et al. Revised conodont-based integrated high-resolution timescale
for the Changhsingian Stage and end-Permian extinction interval at the
Meishan sections, South China. Lithos 204, 220–245 (2014).
73. Ridgwell, A. & Zeebe, R. E. The role of the global carbonate cycle in the
regulation and evolution of the Earth system. Earth Planet. Sci. Lett. 234,
299–315 (2005).
Acknowledgements
Many thanks to Linhao Fang for the help of sample treatment method. We thank
Wenchao Shu and Yao Wang for sample treatment. This study was supported by the
National Natural Science Foundation of China (42030513, 41821001, 41530104,
42072025, 41888101), the US National Science Foundation grant no. EAR-1603051
and EAR-2026877, and also benefited from Natural Environment Research Council
(UK) grant, ‘Ecosystem resilience and recovery from the Permo-Triassic crisis’
(grant NE/P013724/1), which is a part of the Biosphere Evolution, Transitions and
Resilience (BETR) Program. This is Center for Computational & Modeling Geos-
ciences (BGEG) publication number 2.
Author contributions
D.L.C., Y.C., and Y.Y.W. designed this study with in-depth inputs from H.J.S., P.B.W., J.D.C.,
and J.N.T. Y.Y.W. and D.L.C completed the data preparation and analysis with the help from
H.Y.S.,Y.D., and J.D.C. Y.Y.W. and Y.C. performed the calculations. Y.Y.Wand D.L.C wrote
the main text and supplementary materials with inputs from all authors.
Competing interests
The authors declare no competing interests.
Additional information
Supplementary information The online version contains supplementary material
available at https://doi.org/10.1038/s41467-021-22298-7.
Correspondence and requests for materials should be addressed to D.C. or Y.C.
Peer review information Nature Communications thanks Christopher Fielding and the
other, anonymous reviewer(s) for their contribution to the peer review of this work. Peer
review reports are available.
Reprints and permission information is available at http://www.nature.com/reprints
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons
Attribution 4.0 International License, which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long as you give
appropriate credit to the original author(s) and the source, provide a link to the Creative
Commons license, and indicate if changes were made. The images or other third party
material in this article are included in the article’s Creative Commons license, unless
indicated otherwise in a credit line to the material. If material is not included in the
article’s Creative Commons license and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this license, visit http://creativecommons.org/
licenses/by/4.0/.
© The Author(s) 2021
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-22298-7
8NATURE COMMUNICATIONS | (2021) 12:2137 | https://doi.org/10.1038/s41467-021-22298-7 | www.nature.com/naturecommunications
Content courtesy of Springer Nature, terms of use apply. Rights reserved

1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com