Content uploaded by Ralf Tappert
Author content
All content in this area was uploaded by Ralf Tappert on Sep 15, 2018
Content may be subject to copyright.
Stable carbon isotopes of C3 plant resins and ambers
record changes in atmospheric oxygen since the Triassic
Ralf Tappert
a,b,⇑
, Ryan C. McKellar
a,c
, Alexander P. Wolfe
a
, Michelle C. Tappert
a
,
Jaime Ortega-Blanco
c,d
, Karlis Muehlenbachs
a
a
Department of Earth and Atmospheric Sciences, University of Alberta, Edmonton, Alberta, Canada
b
Institute of Mineralogy and Petrography, University of Innsbruck, Innsbruck, Austria
c
Division of Entomology (Paleoentomology), Natural History Museum, University of Kansas, Lawrence, KS, USA
d
Departament d’Estratigrafia, Paleontologia i Geocie
`ncies Marines, Facultat de Geologia, Universitat de Barcelona, Barcelona, Spain
Received 28 September 2012; accepted in revised form 10 July 2013; Available online 24 July 2013
Abstract
Estimating the partial pressure of atmospheric oxygen (pO
2
) in the geological past has been challenging because of the lack of
reliable proxies. Here we develop a technique to estimate paleo-pO
2
using the stable carbon isotope composition (d
13
C) of plant
resins—including amber, copal, and resinite—from a wide range of localities and ages (Triassic to modern). Plant resins are par-
ticularly suitable as proxies because their highly cross-linked terpenoid structures allow the preservation of pristine d
13
Csignatures
over geological timescales. The distribution of d
13
C values of modern resins (n= 126) indicates that (a) resin-producing plant fam-
ilies generally have a similar fractionation behavior during resin biosynthesis, and (b) the fractionation observed in resins is similar
to that of bulk plant matter. Resins exhibit a natural variability in d
13
C of around 8&(d
13
Crange:31&to 23&,mean:
27&), which is caused by local environmental and ecological factors (e.g., water availability, water composition, light exposure,
temperature, nutrient availability). To minimize the effects of local conditions and to determine long-term changes in the d
13
Cof
resins, we used mean d
13
C values (d13 Cresin
mean) for each geological resin deposit. Fossil resins (n= 412) are generally enriched in
13
C
compared to their modern counterparts, with shifts in d13 Cresin
mean of up to 6&. These isotopic shifts follow distinctive trends through
time, which are unrelated to post-depositional processes including polymerization and diagenesis. The most enriched fossil resin
samples, with a d13Cres in
mean between 22&and 21&, formed during the Triassic, the mid-Cretaceous, and the early Eocene. Exper-
imental evidence and theoretical considerations suggest that neither change in pCO
2
nor in the d
13
C of atmospheric CO
2
can
account for the observed shifts in d13 Cresin
mean. The fractionation of
13
C in resin-producing plants (D
13
C), instead, is primarily influ-
enced by atmospheric pO
2
, with more fractionation occurring at higher pO
2
. The enriched d13 Cresin
mean values suggest that atmospheric
pO
2
during most of the Mesozoic and Cenozoic was considerably lower (pO
2
= 10–20%) than today (pO
2
= 21%). In addition, a
correlation between the d13Cresin
mean and the marine d
18
O record implies that pO
2
,pCO
2
, and global temperatures were inversely
linked, which suggests that intervals of low pO
2
were generally accompanied by high pCO
2
and elevated global temperatures. Inter-
vals with the lowest inferred pO
2
, including the mid-Cretaceous and the early Eocene, were preceded by large-scale volcanism dur-
ing the emplacement of large igneous provinces (LIPs). This suggests that the influx of mantle-derived volcanic CO
2
triggered an
initial phase of warming, which led to an increase in oxidative weathering, thereby further increasing greenhouse forcing. This pro-
cess resulted in the rapid decline of atmospheric pO
2
during the mid-Cretaceous and the early Eocene greenhouse periods. After the
cessation in LIP volcanism and the decrease in oxidative weathering rates, atmospheric pO
2
levels continuously increased over tens
of millions of years, whereas CO
2
levels and temperatures continuously declined. These findings suggest that atmospheric pO
2
had
a considerable impact on the evolution of the climate on Earth, and that the d
13
C of fossil resins can be used as a novel tool to
assess the changes of atmospheric compositions since the emergence of resin-producing plants in the Paleozoic.
Ó2013 Elsevier Ltd. All rights reserved.
0016-7037/$ - see front matter Ó2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.gca.2013.07.011
⇑
Corresponding author at: Institute of Mineralogy and Petrography, University of Innsbruck, Innsbruck, Austria. Tel.: +43 512 492 5519.
E-mail address: ralf.tappert@uibk.ac.at (R. Tappert).
www.elsevier.com/locate/gca
Available online at www.sciencedirect.com
ScienceDirect
Geochimica et Cosmochimica Acta 121 (2013) 240–262
1. INTRODUCTION
Our knowledge of the composition of the atmosphere
through time and of its impact on climate change and
the evolution of life is based on evidence preserved in
the fossil record. For the Pleistocene and Holocene, a di-
rect and continuous record of the partial pressures and
isotopic compositions of atmospheric gases is preserved
in the form of air vesicles entrapped in polar ice sheets
(e.g.,Petit et al., 1999). With increasing age, the record
of atmospheric compositions becomes less directly accessi-
ble, and inferences about the composition of the atmo-
sphere in the geological past rely on indirect measures
or proxies. The quality of such information depends di-
rectly on the ability of proxies to reliably capture the com-
position of ancient atmospheres.
Attempts have been made to reconstruct atmospheric
pCO
2
and pO
2
using mass balance calculations and flux
modeling (Berner, 1999; Berner and Kothavala, 2001; Ber-
ner et al., 2007). Other approaches to estimate the paleo-
pCO
2
have utilized the density of stomata on fossil leaf cuti-
cles (Retallack, 2001, 2002; Beerling and Royer, 2002) and
the carbon stable isotopic composition of pedogenic car-
bonates (Cerling and Hay, 1986; Ekart et al., 1999;Bowen
and Beerling, 2004). One of the richest sources of informa-
tion on ancient atmospheric compositions comes from the
marine sedimentary record, in particular from the stable
isotope composition (d
13
C, d
18
O, d
11
B) of biogenic carbon-
ates (e.g.,Clarke and Jenkyns, 1999; Pearson and Palmer,
2000; Royer et al., 2001; Zachos et al., 2001). This informa-
tion has been used to infer changes in the isotopic compo-
sition of atmospheric CO
2
through time and to derive
paleo-temperature estimates. More recently, the d
13
Cof
marine phytoplankton biomarkers have been used to esti-
mate the pCO
2
of Cenozoic atmospheres (Pagani et al.,
2005). Additional constraints on the composition of ancient
atmospheres were obtained from the marine record using
the biochemical cycles of sulfur (Berner, 2006; Halevy
et al., 2012).
For terrestrial environments, attempts to estimate paleo-
pCO
2
have frequently exploited the d
13
C of fossil organic
matter, primarily plant organic matter (Strauss and Pe-
ters-Kottig, 2003; Jahren et al., 2008). A range of plant con-
stituents have been targeted for d
13
C analysis, including
wood, foliage, and cellulose extracts, as well as the coals
to which they contribute (Stuiver and Braziunas, 1987;
Gro
¨cke, 1998, 1999, 2002; Lu
¨cke et al., 1999; Peters-Kottig
et al., 2006; Poole and Van Bergen, 2006; Poole et al., 2006;
Bechtel et al., 2008). However, three key issues need to be
addressed when plant organic matter is used for stable car-
bon isotopic analyses in geochemical and chemostrati-
graphic studies:
(1) Plants are composed of a complex range of organic
compounds, including polysaccharides (e.g., cellu-
lose), lignin, lipids, cuticular waxes, and photosynth-
ates. Due to differential fractionation of carbon
isotopes during biosynthesis, the isotopic composi-
tion of individual plant fractions typically differ from
each other and from that of the bulk plant (Park and
Epstein, 1960; Brugnoli and Farquhar, 2000). In
addition, some plant groups use distinctive metabolic
pathways that fundamentally affect their isotopic
composition. For example, C4 and CAM plants are
generally more enriched in
13
C compared to C3
plants (O’Leary, 1988; Farquhar et al., 1989). There-
fore, if stable isotopes are used in chemostratigraphic
studies, it is crucial to determine the organic compo-
sition of the material under investigation and its
botanical source.
(2) Environmental and ecological factors can play an
important role in the isotopic fractionation of plants,
and individual isotope values of fossil or modern
plant material may reflect local instead of large-scale
environmental conditions.
(3) Plant constituents can undergo substantial composi-
tional changes during and after burial and diagenesis.
These changes have the potential to alter the isotopic
composition of the fossil plant material, to the extent
of preventing a meaningful paleoenvironmental
interpretation.
To determine environmentally mediated changes in the
d
13
C of fossil plant organic matter through time, we ana-
lyzed modern and fossil resins from a wide range of locali-
ties and ages. Compared to other terrestrial plant
constituents, resins have chemical properties that make
them particularly suitable as proxies of environmental
changes over geological time. Resins can polymerize into
highly cross-linked hydrocarbons that have considerable
preservation potential in the geological record (Fig. 1). This
means that a reliable terrestrial stable isotope history can
potentially be reconstructed well into the Paleozoic, consid-
ering that the earliest known fossil resins are from the Car-
boniferous (White, 1914; Bray and Anderson, 2009).
Unlike many other fossil plant organic constituents,
resins can often be assigned to a specific source plant fam-
ily, and only a few plant families produce significant
amounts of preservable resins, all of which have C3
metabolism. Furthermore, resins are chemically conserva-
tive, which means that the composition of the resins did
not change significantly as plants evolved. These factors
limit not only the chemical variability of fossil resins,
but also any potential family-specific biases, which could
influence the isotopic composition of fossil resins in
chemostratigraphic studies.
Although large samples of gem-quality fossil resins, or
amber, are found in fewer than 50 locations worldwide,
small pieces (<5 mm) of non-precious resin, which are
sometimes referred to as resinites, are found frequently,
and are often associated with coal deposits (Stach et al.,
1982). Previous studies of the carbon isotopic composition
of resins were limited because they were conducted on only
a few localities, and in most cases, using only a few samples
(Nissenbaum and Yakir, 1995; Murray et al., 1998; Nissen-
baum et al., 2005; McKellar et al., 2008; DalCorso et al.,
2011). Our study is the first attempt to integrate the d
13
C
data of resins from a wide range of localities and ages,
and to determine their potential as proxies for changes in
atmosphere composition through geological time.
R. Tappert et al. / Geochimica et Cosmochimica Acta 121 (2013) 240–262 241
2. SAMPLES AND METHODS
In total, 538 d
13
C measurements were conducted on
modern (n= 126) and fossil resins (n= 412) from various
locations worldwide, ranging in age from recent to Triassic
(Fig. 2,Table 1). Analytical work was performed on hard-
ened resins that were visibly free of inclusions and vesicles.
Modern resins were collected primarily from the trunks of
native trees growing under a range of climatic conditions
(i.e., tropical to subarctic), although specimens from culti-
vars were also analyzed in some cases. We restricted the
sampling of modern resins to representatives of the families
Pinaceae (Pseudotsuga,Picea, Pinus), Cupressaceae (Meta-
sequoia, Thuja), Araucariaceae (Agathis), and Fabaceae
(Hymenaea), as these families represent the most important
resin producers in modern environments and the fossil re-
cord (Langenheim, 2003).
The fossil resin samples include many small, non-pre-
cious varieties, in addition to genuine ambers. To simplify
the terminology, all fossil resins are referred to herein as
amber, with the exception of sub-recent resins (>200 years
old), which are referred to as copal. The sample set includes
specimens from many well-known deposits (e.g., Baltic am-
ber, Bitterfeld amber, Dominican amber, Myanmar or Bur-
mese amber, and Lebanese amber), but also material from
newly discovered occurrences (Table 1).
For many amber deposits, the age of amber formation
can be tightly constrained (±2 Ma or better) by indepen-
dent stratigraphic controls or radiometric dating (Table 1).
However, for some of the deposits, the age constraints are
more provisional, especially for localities that show evi-
dence of re-working or re-deposition (e.g., Baltic and Bitter-
feld ambers). For these deposits, the age estimates have
relatively large uncertainties (±4 Ma).
To examine the chemical composition of the resins, rep-
resentative samples of modern and fossil resins from each
locality were analyzed using micro-Fourier transform infra-
red (FTIR) spectroscopy. The resulting absorption spectra
were used to (1) classify the resins into distinct composi-
tional groups (see below), and (2) determine the botanical
affinity of the fossil resins. Particular emphasis was placed
on identifying the botanical source of fossil resins from
BA
C
E
D
Fig. 1. Examples of modern and fossil resins. (A) Fresh resin oozing out of the trunk of Araucaria cunninghamii, Adelaide Botanical Garden,
Adelaide, Australia. Field of view (FOV): 20 cm. (B) In situ lenticular resinite fragment (arrow) within a coal seam, Horseshoe Canyon
Formation (Campanian), Edmonton, Canada (forceps for scale). (C) Selection of amber samples from the Horseshoe Canyon Formation
(Campanian), Drumheller, Canada. The largest specimen is 5 cm. (D) Inclusions of foliage from Parataxodium sp. (an extinct cupressaceous
conifer) in amber, Foremost Formation (Campanian), Grassy Lake, Canada. FOV: 1 cm. (E) Triassic resin droplets in carbonate matrix,
Heiligkreuzkofel Formation (Carnian), Dolomites, Italy. FOV: 10 cm.
242 R. Tappert et al. / Geochimica et Cosmochimica Acta 121 (2013) 240–262
previously undescribed localities. Absorption spectra were
collected between 4000 and 650 cm
1
(wavenumbers) using
a Thermo Nicolet Nexus 470 FTIR spectrometer equipped
with a Nicolet Continuum IR microscope, with a spectral
resolution of 4 cm
1
. A total of 200 interferograms were
collected and co-added for each spectrum. Spectra were col-
lected from inclusion-free, freshly broken resin chips placed
on an infrared-transparent NaCl disc. To avoid the effects
of oversaturation of the spectra, sample thickness was kept
below 5 lm. Depending on the quality of the sample, the
spot size was set to values between 50 50 and
100 100 lm (square aperture). Further details concerning
the FTIR analytical procedures are presented in Tappert
et al. (2011).
The stable carbon isotope compositions of the resins were
determined using a Finnigan MAT 252 dual inlet mass spec-
trometer at the University of Alberta. Samples of untreated
resin (1–5 mg) were combusted together with 1g CuO
as an oxygen source in a sealed and evacuated quartz tube
at 800 °Cfor12 h. The d
13
C analyses were normalized to
NBS-18 and NBS-19, and the results are given with respect
to V-PDB (Coplen et al., 1983). Instrumental precision and
accuracy was on the order of ±0.02&.Repeatedanalyses
of individual samples were within ±0.1&.
3. RESULTS
3.1. Infrared spectroscopy
The comparison of infrared spectra from modern and
fossil resins (Fig. 3) shows that resins, irrespective of their
age and botanical origin, produce spectra that are very sim-
ilar. This similarity is due to the dominance of diterpenoids
in the composition of the resins. However, some variability
in the spectra exists and differences in the position and
height of specific absorption features can be used to distin-
guish different resin types. These subtle spectroscopic differ-
ences primarily reflect variations in the structure of the
dominant terpenoid structures, but they can also be influ-
enced by the presence of additional organic components
and the degree of polymerization.
Four compositional types of resins are distinguished in
this study. These resin types were produced by specific plant
families and are, accordingly, classified into the following
four categories:
(1) Pinaceous resins are exclusively produced by conifers
belonging to the family Pinaceae. The infrared spec-
tra of pinaceous resins are characterized by distinc-
tive absorption peaks at 1460 and 1385 cm
1
(Fig. 3). The amplitude of the peak at 1460 cm
1
is
generally lower than the peak at 1385 cm
1
. Another
trait is the absence or the weak development of a
peak at 2848 cm
1
. This peak is typically well devel-
oped in cupressaceous-araucarian resins (see below).
Further details about the spectral characteristics of
pinaceous resins can be found in Tappert et al.
(2011). Based on results of gas chromatography–
mass spectrometry (GC–MS), pinaceous resins consist
primarily of diterpenoids that are based on pimarane
and abietane structures (Langenheim, 2003).
(2) Cupressaceous-araucarian resins are mainly pro-
duced by conifers of the families Cupressaceae and
Araucariaceae, but also by Sciadopitys verticillata,
the sole extant member of the family Sciadopytaceae
(Wolfe et al., 2009). The infrared spectra of cupressa-
ceous-araucarian resins are characterized by distinc-
tive absorption peaks at 1448, 887, and 791 cm
1
(Fig. 3). In fossil resins, the 887 cm
1
peak is often
reduced or absent as a result of the fossilization pro-
cess. The amplitude of the absorption peak at
1385 cm
1
is typically lower than that of the adjacent
peak at 1448 cm
1
. Further details about the spectral
characteristics of cupressaceous-araucarian resins and
their distinction from pinaceous resins can be found
in Tappert et al. (2011). Cupressaceous-araucarian
pinaceous
cupressaceous-araucarian
fabaceous
dipterocarpaceous
Resin Type
Neogene
Paleogene
Late Cretaceous
Early Cretaceous
Age
Triassic
Fig. 2. Locations of amber deposits and types of amber analyzed in this study.
R. Tappert et al. / Geochimica et Cosmochimica Acta 121 (2013) 240–262 243
Table 1
Overview of the fossil resins analyzed in this study, including their age and botanical source.
No.Sample Location Stratigraphic
setting
Age/Epoch
of formation
Absolute
Age
[Ma]
Compositional
Type (based on
FTIR
spectroscopy)
Class
a
Botanical
source
References to age
and/or botanical
source
1 Colombian
copal
Colombia Undetermined Pliocene–
Holocene
0.0002–
2.5
Fabalean Ic Hymenaea sp. Ragazzi et al. (2003),
Lambert et al. (1995)
2 Malaysian
amber
Borneo Merit-Pila coal
field
Middle
Miocene
12 DipterocarpaceousII DipterocarpaceaeLangenheim and Beck
(1965),
Schlee and Chan (1992)
3 Mexican
amber
Chiapas La Quinta Fm.–
Balumtun Fm.
Early Miocene 13–19 Fabalean Ic Hymenaea sp. Langenheim and Beck
(1965),
Cunningham et al., 1983,
Solo
´rzano Kraemer (2007)
4 Dominican
amber
Dominican
Rep.
La Toca Fm.–
Yanigua Fm.
Late Oligocene–
Early Miocene
15–20 Fabalean Ic Hymenaea sp. Langenheim and Beck
(1965),
Iturralde-Vinent and
MacPhee (1996)
5 Baltic
amber
Baltic
coast
Blaue Erde Fm.
(redeposited)
Late Eocene–
Early Oligocene
33–40 Cupressaceous-
araucarian
suc
Ia Sciadopitys sp. Langenheim (2003),
Wolfe et al. (2009)
6 Bitterfeld
amber
Germany Bitterfeld coal
field (redeposited)
Late Eocene–
Early Oligocene
33–40 Cupressaceous-
araucarian
suc
Ia Sciadopitys sp. Fuhrmann (2005),
Wolfe et al. (2009)
7 Giraffe
kimberlite
amber
Canada
(NT)
Peat sequence in
kimberlite crater
Eocene 39–38 Cupressaceous-
araucarian
Ib Metasequoia sp. Tappert et al. (2011)
8 Tiger
Mountain
amber
USA (WA) Tiger Mountain
Fm.
Middle Eocene 47–48 Cupressaceous-
araucarian
Ib Metasequoia sp. Mustoe (1985)
9 Ellesmere
Island amber
Canada
(NU)
Eureka Sound
Fm.
Middle Eocene 50 Pinaceous
SUC
VPseudolarix sp. Francis (1988), this study
10 Tadkeshwar
amber
India Cambay shale Eocene 50–52 DipterocarpaceousII DipterocarpaceaeRust et al. (2010)
11 Panda
kimberlite
amber
Canada
(NT)
Kimberlite crater
infill
Early Eocene 53 Cupressaceous-
araucarian
Ib Metasequoia sp. Wolfe et al. (2012)
12 Alaskan
amber
USA (AK) Chickaloon Fm. Paleocene–Early
Eocene
53–55 Cupressaceous-
araucarian
Ib Metasequoia sp. Triplehorn et al. (1984),
this study
13 Evansburg
amber
Canada
(AB)
Upper Scollard
Fm.
Paleocene 59–62 Cupressaceous-
araucarian
Ib Metasequoia sp. This study
14 Genesee
amber
Canada
(AB)
Scollard Fm. Early Paleocene 65 Cupressaceous-
araucarian
Ib Metasequoia sp. This study
15 Albertan
amber
Canada
(AB)
Horseshoe
Canyon Fm.
Campanian–
Maastrichtian
70–73 Cupressaceous-
araucarian
Ib Parataxodium sp.McKellar and Wolfe
(2010)
16 Grassy Lake
amber
Canada
(AB)
Foremost Fm. Campanian 76–78 Cupressaceous-
araucarian
Ib Parataxodium sp.McKellar and Wolfe
(2010)
17 Cedar Lake
amber
Canada
(MT)
Foremost Fm.
(redeposited)
Campanian 76–78 Cupressaceous-
araucarian
Ib Parataxodium sp.McKellar and Wolfe
(2010)
18 New Jersey
amber
USA (NJ) Raritan Fm. Turonian 90–94 Cupressaceous-
araucarian
Ib Cupressaceae Grimaldi et al. (1989),
Anderson (2006)
19 Burmese
amber
Myanmar Hukawng Basin Late Albian–
Early
Cenomanian
96–102 Cupressaceous-
araucarian
Ib Cupressaceae Grimaldi et al. (2002),
Cruickshank and Ko
(2003)
20 Spanish
amber
Spain,
various
Eschucha Fm. Albian 100–110 Cupressaceous-
araucarian
Ib CheirolepidiaceaeAlonso et al. (2000)
21 Lebanese
amber
Lebanon Gre
`s de Base Fm.Late Barremian–
Early Aptian
123–127 Cupressaceous-
araucarian
Ib Cupressaceae Nissenbaum and Horowitz
(1992),Azar et al. (2010)
22 Wealden
amber
UK (Isle
of Wight)
Hastings beds
(Wealden Group)
Berriasian–
Valanginian
138–142 Cupressaceous-
araucarian
Ib Cupressaceae Nicholas et al. (1993)
23 Italian
amber
Italy
(Dolomites)
Heiligkreuzkofel
Fm.
Carnian 220–225 Cupressaceous-
araucarian
Ib CheirolepidiaceaeRoghi et al. (2006),
Ragazzi et al. (2003)
SUC
: contains succinic acid esters (succinate).
a
Classification based on Anderson et al. (1992) and Anderson and Botto (1993).
244 R. Tappert et al. / Geochimica et Cosmochimica Acta 121 (2013) 240–262
resins consist primarily of diterpenoids that are based
on labdane skeletal structures, such as communic
acid polymers (Thomas, 1970; Anderson et al., 1992).
(3) Fabalean resins are produced by the tropical angio-
sperm Hymenaea (Fabaceae, Fabales). The infrared
spectra of these resins superficially resemble spectra
from pinaceous and cupressaceous-araucarian repre-
sentatives (Fig. 3). Similar to pinaceous resins, faba-
lean resins produce an absorption peak at 1460 cm
1
,
but the amplitude of the 1385 cm
1
peak is lower or
equivalent to the 1460 cm
1
peak. A distinctive fea-
ture of fabalean resin spectra is the lack of strong
absorption features between 970 and 1050 cm
1
.
Fabalean resins also produce an absorption peak at
887 cm
1
. Like cupressaceous-araucarian resins,
fabalean resins are primarily based on labdane struc-
tured diterpenoids, with zanzibaric and ozic acid
polymers being typical components (Cunningham
et al., 1983).
(4) Dipterocarpaceous resins are produced by members
of the angiosperm family Dipterocarpaceae, which
is a family of diverse and widespread tropical rainfor-
est trees. The infrared spectra of dipterocarpaceous
resins are quite distinct from other resin spectra
(Fig. 3). The most characteristic spectroscopic fea-
tures are the distinctive absorption triplet between
1360 and 1400 cm
1
, and an additional absorption
peak at 1050 cm
1
. Dipterocarpaceous resins are
primarily based on sesquiterpenoids, such as cadin-
ene, but triterpenoids can also be present (Ander-
son et al., 1992). The resin produced by
dipterocarpaceous trees is colloquially referred to as
Dammar.
Some of the resins, including Baltic amber and resins
from Ellesmere Island, were found to produce an additional
and distinctive spectral feature at 1160 cm
1
, which is asso-
ciated with a broad shoulder or a broad peak between 1200
and 1300 cm
1
(Fig. 3). These spectral features generally re-
late to the presence of esterified succinic acid (succinate) as
an additional component in the resin (Beck et al., 1965;
Beck, 1986). Although succinate has been identified in some
pinaceous resins—namely in resins from Pseudolarix amabi-
lis and in some fossil resins of Pseudolarix-affinity from the
Canadian Arctic (Anderson and LePage, 1995; Poulin and
Helwig, 2012)—it is the hallmark constituent of Baltic am-
ber, which itself is a cupressaceous-araucarian resin. The
presence of succinate has important implications in the
assessment of the botanical source of fossil resins, and con-
sequently, succinate-bearing resins are marked separately in
Table 1.
Most of the fossil resins analyzed for this study are of
cupressaceous-araucarian type. Along with fabalean res-
ins (e.g., Dominican and Mexican ambers), cupressa-
ceous-araucarian resins make up the most productive
commercial amber deposits (Table 1). The predominance
of these resins in the fossil record is due to their terpe-
noid composition because labdane-based resins polymer-
ize rapidly and are highly resistant to chemical
breakdown (Fig. 1B).
Although pinaceous resins are produced in extremely
high abundance in northern-hemisphere forests, their pres-
ervation potential in the fossil record is limited. The main
constituent of pinaceous resins—diterpenes with pimarane
and abietane skeletal structures—do not possess the func-
tional groups required to form stable polymers. Therefore,
they are prone to decomposition. Nevertheless, under
favorable conditions, even pinaceous resins can persist for
millions of years (Wolfe et al., 2009).
In accordance with previous studies (Langenheim and
Beck, 1965; Broughton, 1974; Tappert et al., 2011), all res-
ins of a given type produce very similar FTIR absorption
spectra, irrespective of whether the samples are modern
or fossil, which indicates that chemical transformations
during burial and maturation are minor. This observation
is important because it supports the premise that resins
can resist isotopic exchange over geological timescales.
3.2. Carbon stable isotopes
3.2.1. Modern resins
The d
13
C values of modern resins analyzed in this study
range from 31.2&to 23.6&. These values are consistent
with previously published d
13
C analyses of modern resins
from comparable plant taxa (Nissenbaum and Yakir,
1995; Murray et al., 1998; Nissenbaum et al., 2005). The
d
13
C values of the modern resins follow a near Gaussian
distribution, with a mean value of 26.9&(Fig. 4). It is
notable that resins from individual plant families have sim-
ilar d
13
C distributions and similar mean d
13
C values
(Fig. 5). This indicates that the isotopic fractionation that
occurs during resin biosynthesis is nearly identical for the
plant families considered in this study, despite differences
in the terpenoid profile of individual resin types.
Fig. 4 also shows the carbon isotopic compositions of
bulk leaf material from modern C3-plants using data com-
piled by Ko
¨rner et al. (1991), Diefendorf et al. (2010), and
Kohn (2010). The leaf material was sampled from a wide
range of plant taxa growing under diverse climatic condi-
tions, from alpine and polar to tropical. Therefore, the iso-
topic composition should approximate the average
composition of modern C3 plant organic matter. However,
samples from extremely dry environments and samples that
were potentially influenced by the canopy effect (see below)
were excluded. It is notable that the isotope values of the
leaf material (32.8&to 22.2&, mean: 27.1&) and
the distribution of isotope values are very similar to those
reported for modern resins. In conclusion, we find no con-
sistent differences between the distributions of d
13
C values
in modern leaves and modern resins.
3.2.2. Fossil resins
Carbon isotope values of fossil resins were found to
range from 28.4&to 18.2&, which means that some
fossil resins are more enriched in
13
C compared to their
modern counterparts. Data for individual samples are
shown in Fig. 6. Similar to modern resins, the fossil resins
at each deposit produce a range of d
13
C values. These
ranges are similar to those observed for modern resins
(i.e.,8&) provided that a comparable number of samples
R. Tappert et al. / Geochimica et Cosmochimica Acta 121 (2013) 240–262 245
were analyzed. For some of the amber deposits, only a
small number of samples were analyzed due to sample
availability. Consequently, the range of isotope values for
these deposits tends to be narrower. Despite differences in
the number of samples for different deposits, a considerable
shift in d
13
C between deposits is observed (Figs. 6 and 7).
To quantify the isotopic shift through time, and to mini-
mize the effect of sample size, we used the mean d
13
C values
of the fossil resins (d13Cresin
mean) as a representative measure for
each deposit (Table 2).
The d13Cresin
mean of subrecent Colombian copal (27.5&)is
nearly identical to modern resins, but with increasing age,
Wollemia nobilis
-modern-
Tiger Mountain aber (USA)
-Middle Eocene-
Ellesmere Island amber (Canada)
-Middle Eocene-
Genesee amber (Canada)
-Early Paleocene-
Pseudolarix amabilis
-modern-
Dominican amber
-Late Oligocene-Early Miocene-
Hymenaea courbaril
-modern-
Shorea rubriflora
-modern-
Malaysian amber
-Middle Miocene-
Pinaceous resins
Cupressaceous-
araucarian resins
Fabalean
resins
Dipterocarpaceous
resins
Wavenumbers (cm )
-1
8001000120014001600180020002200240026002800
3000320034003600
887
1460
1385
791
1050
succinate
region
1160
1448
2848
fo(ecnabrosbA)ytiralcroftsef
Mexican amber
-Early Miocene-
Single bonds (x-H)
O-H C-H
Double bonds Single bonds/Skeletal vibrations
C=C, C=O C-O, C-C
Fig. 3. Micro-FTIR spectra of different resin types distinguished in this study (selected modern and fossil examples). Dashed lines mark
spectral features discussed in the text. Yellow backgrounds highlight the most distinctive spectral regions. Characteristic spectral regions for
molecular vibrations relevant to resins are marked in blue. (For interpretation of the references to color in this figure legend, the reader is
referred to the web version of this article.)
246 R. Tappert et al. / Geochimica et Cosmochimica Acta 121 (2013) 240–262
resins become continuously more enriched in
13
C and con-
sequently less negative in their d13Cresin
mean values. This trend
towards more enriched isotopic compositions in fossil res-
ins continues through the Neogene and into the middle Eo-
cene. Some of the most productive amber deposits,
including the Dominican (d13Cresin
mean =24.8&) and the Bal-
tic (d13Cresin
mean =23.5&) amber deposits, formed during this
time interval. Eocene ambers from Tiger Mountain (Wash-
ington, USA) and Ellesmere Island (Nunavut, Canada) are
more than 5.3&enriched in
13
C compared to modern res-
ins, with a d13Cresin
mean of 21.7&and 21.6&, respectively.
The latter are the isotopically most enriched Cenozoic res-
ins in the present sample set.
The early Eocene is characterized by a dramatic shift to-
wards more depleted resin isotope compositions, as demon-
strated by amber from the Tadkeshwar coal field of India
(d13Cresin
mean =25.8&), and amber preserved in the Panda
kimberlite pipe in Canada (d13Cresin
mean =24.8&). The earli-
est Eocene amber from the Chickaloon Formation in Alas-
ka marks a return to slightly more
13
C-enriched
compositions, with a d13Cresin
meanof 23.6&.
Paleocene amber from Evansburg (d13Cresin
mean ¼22:3&)
and Genesee (d13Cresin
mean ¼23:1&) in Alberta, Canada,
are more enriched than their early Eocene counterparts.
However, from the Paleocene into the Late Cretaceous
(Campanian), resin compositions become increasingly more
mean: -26.9
= 1.39
n = 160
‰
Frequency
Modern resins
mean = -27.1
= 1.67
n = 891
‰
Frequency
Bulk C3 leaf
Growth conditions (interpreted)
optimal typical poor
Fractionation efficiency
wolhgih
A
B
13C (‰ vs PDB)
Fig. 4. Distribution of d
13
C values in (A) modern resins (data from this study with additional data of Hymenaea resins from Nissenbaum
et al., 2005), and (B) bulk leaf matter from a wide range of C3 plant taxa (data from Ko
¨rner et al., 1991; Diefendorf et al., 2010; Kohn, 2010).
R. Tappert et al. / Geochimica et Cosmochimica Acta 121 (2013) 240–262 247
depleted in
13
C. The most depleted Cretaceous resins with a
d13Cresin
mean of 23.9&were found in the Horseshoe Canyon
Formation (late Campanian) of southern and central Al-
berta. From the late Campanian and into the Early Creta-
ceous, this trend reverses and the d13Cresin
mean shifts to more
13
C-enriched compositions, as shown by the coeval Grassy
Lake and Cedar Lake amber (d13Cresin
mean ¼23:6&), and the
New Jersey Raritan amber (d13Cresin
mean ¼22:1&)(Fig. 7).
Resins that formed between the Cenomanian and Barre-
mian are characterized by even more enriched d13 Cresin
mean
compositions. During this time interval, Burmese amber
(d13Cresin
mean ¼21:3&), Spanish amber (d13Cresin
mean ¼
21:4&), and Lebanese amber (d13Cresin
mean ¼21:1&) were
formed. The oldest Cretaceous amber samples were recov-
ered from the Wealden Group of England (UK). Although
the d13Cresin
mean (22.8&) of the Wealden amber is more de-
pleted than that of the Burmese, Spanish, and Lebanese
amber, its d13Cresin
mean is not well constrained because only
two samples were available for analysis.
Pre-Cretaceous ambers were limited to samples from the
Triassic (Carnian) Heiligkreuzkofel Formation of the Dol-
omites, Italy. With a d13Cresin
mean of 20.9&, these ambers are
the most
13
C-enriched in the current sample set. In accor-
dance with previously published d
13
C values of amber from
the same locality (d13Cresin
mean ¼20:12&;DalCorso et al.,
2011), our results confirm that these Triassic resins
are > 6&more
13
C-enriched relative to modern resins.
4. DISCUSSION
4.1. Influences on the d
13
C of modern plants
4.1.1. Local environmental factors
Experimental studies, including growth chamber experi-
ments, have shown that plant d
13
C is influenced by a wide
range of local environmental factors, such as water avail-
ability, water composition (e.g., salinity, nutrients), light
exposure, local temperature, altitude, humidity, presence
of parasites, etc. (Park and Epstein, 1960; Guy et al.,
1980; O’Leary, 1981; Ko
¨rner et al., 1991; Arens and Jahren,
2000; Edwards et al., 2000; Dawson et al., 2002; Gro
¨cke,
2002; McKellar et al., 2011)(Fig. 8). Each of these factors
has a direct influence on the efficiency of fractionation in
plants during photosynthesis, and each can cause an isoto-
pic shift on the order of several permil in the d
13
Cofa
plant.
Every plant community will produce a range of d
13
C val-
ues because the environmental conditions (i.e., growth con-
ditions) for each plant are unique. However, each plant
taxon has its own range of growth conditions under which
13
C fractionation is maximized. The observation that the
d
13
C of modern resins follows a Gaussian distribution
(Fig. 4), suggests that the majority of the resin-producing
plants grew under conditions that were intermediate be-
tween optimal (i.e., most efficient fractionation) and poor
(i.e., least efficient fractionation). Therefore, the relative
d
13
C values of plant matter within a local plant community
can be viewed primarily as a measure of the growth
conditions.
For most plants, including the copious resin producers,
it can be assumed that the metabolized CO
2
was sourced
from isotopically undisturbed air, which had a d
13
C compo-
sition approximating a global atmospheric average. How-
ever, local isotopic effects can alter the d
13
C of the CO
2
in
the ambient air. Limited air circulation, for example, can
lead to a reprocessing of respired air, which is depleted in
13
C(i.e., canopy effect). Metabolizing this depleted air
causes the d
13
C of plants to shift towards more negative val-
ues. In extreme cases, it can result in a depletion in
13
Cof
plant matter of >6&(Medina and Minchin, 1980). The
canopy effect primarily affect plants in the understory of
dense tropical rainforests. Among the plant families consid-
ered in this study, however, only the Dipterocarpaceae
grow in such dense rainforest environments, and only the
Miocene amber from Borneo and the Eocene amber from
Tadkeshwar, India, are of dipterocarpacean origin.
-31 -30 -29 -28 -27 -26 -25 -24 -23
0
5
10
15
20
25
30
35
Frequency
Pinaceae
(mean=-26.9‰, =1.12, n=86)
Cupressaceae-Araucariaceae
(mean=-26.6‰, =1.77, n=50)
Fabaceae (angiosperm)
(mean=-27.3‰, =1.28, n=24)
13C (‰ vs PDB)
Fig. 5. Distribution of d
13
C values from different types of modern resins, separated according to their source plant families. Data include
previously published analyses of Hymenaea (Fabaceae) resins (n= 17) from Nissenbaum et al. (2005).
248 R. Tappert et al. / Geochimica et Cosmochimica Acta 121 (2013) 240–262
4.1.2. Regional and global environmental factors
The results of recent studies on the global distribution of
d
13
C values in bulk leaf matter indicate that the fraction-
ation of
13
C in plants is not only influenced by local envi-
ronmental factors, but also by regional precipitation
patterns (Diefendorf et al., 2010; Kohn, 2010;Fig. 8). It
was found that the average d
13
C of bulk leaf matter and
the calculated fractionation values (D
13
C) show some corre-
lation with the mean annual precipitation (MAP) at a given
location. The
13
C fractionation thereby increases with
increasing MAP (Diefendorf et al., 2010; Kohn, 2010).
However, the correlation is strongly influenced by plants
that were sampled from very dry environments
(MAP < 500 mm/year), in which
13
C fractionation is re-
duced, and those sampled from tropical rainforests
(MAP > 2000 mm/year), in which fractionation is in-
creased. For plants that grew in environments with an inter-
mediate MAP, average D
13
C values only show a weak
correlation with MAP. Since most of the resin-producing
plant families investigated here grow typically in environ-
ments with intermediate MAP, the effect of MAP on their
D
13
C can be considered minor. This notion is supported
by the observation that the d13Cresin
mean values of resins from
different modern plant families investigated in this study
are very similar. However, the Dipterocarpaceae represent
an exception, because they commonly grow in areas with
high MAP. Therefore, they are predisposed towards in-
creased
13
C fractionation. This ecological preference may
explain the low d
13
C values of dipterocarpaceous resins ob-
served by Murray et al. (1998).
-32 -30 -28 -26 -24 -22 -20 -18 -16
Pinaceae
Cupressaceae-Araucariaceae
Fabaceae
Columbian copal
Malaysian amber
Mexican amber
Dominican amber
Baltic amber
Giraffe kimberlite ambe
r
Tiger Mountain amber
Ellesmere Island amber
Panda kimberlite amber
Alaskan amber
Evansburg amber
Genesee amber
Albertan amber
Grassy Lake amber
New Jersey amber
Burmese amber
Spanish amber
Lebanese amber
Wealden amber
Italian amber
13C (‰vs VPDB)
Cedar Lake amber
Bitterfeld amber
modern
Tadkeshwar amber
Fig. 6. d
13
C of modern and fossil resins analyzed for this study. Modern resin data are separated according to their botanical source, i.e., resin
type. Data from fossil resins (grey) are shown for individual deposits. The data are arranged according to the relative age of the fossil resins
(see Table 1). Black rectangles represent one standard deviation. Black squares mark mean values and vertical bars median values.
R. Tappert et al. / Geochimica et Cosmochimica Acta 121 (2013) 240–262 249
The d
13
C of atmospheric CO
2
(d13Cresin
mean) has a direct im-
pact on the d
13
C of plant organic matter, and analyses of
gas inclusions in ice cores (Francey et al., 1999) and of cor-
alline carbonates (Swart et al., 2010) indicate that the global
average d13Cresin
mean has evolved by at least 1&(from around
6.5&to <7.5&) over the last 200 years. This rapid
Pleistocene
Pliocene
Mioocene
Paleocene
Eocene
Oligocene
Maastrichtian
Campanian
Santonian
Coniacian
Turonian
Cenomanian
Albian
Aptian
Barremian
Hauterivian
Valanginian
Berriasian
enegoeN
ene
go
e
la
P
suoec
a
terCetaL
su
o
ecaterCy
l
ra
E
13C (‰vs PDB)
-28 -26 -24 -22 -20
Modern Resins
10
20
30
40
50
60
70
80
90
100
110
120
130
140
Age (Ma)
Albertan amber
New Jersey amber
Grassy Lake/Cedar Lake ambers
Giraffe kimberlite amber
Baltic/Bitterfeld ambers
Dominican amber
Mexican amber
Spanish amber
Lebanese amber
Wealden amber
Columbian copal
Pinaceae
Fabaceae
Cupressaceae/Araucariaceae
Tiger Mountain amber
Alaskan amber
Ellesmere Island amber
Burmese amber
Panda kimberlite amber
Evansburg amber
Genesee amber
c
i
s
sairT
Carnian Italian amber
220
Ladinian
230
Malysian amber
Tadkeshwar amber
Fig. 7. Temporal variations in the mean d
13
C of fossil resins (amber). Horizontal bars mark one standard deviation (see Fig. 6). Modern resin
values are shown for comparison. Different resin types are distinguished by color. Pinaceous: red; cupressaceous-araucarian: blue; fabalean:
yellow; dipterocarpaceous: green. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of
this article.)
250 R. Tappert et al. / Geochimica et Cosmochimica Acta 121 (2013) 240–262
change in d13Cresin
mean is generally linked to the increasing in-
flux of anthropogenic CO
2
from fossil fuel combustion,
which is depleted in
13
C (Suess effect). Although the post-
industrial decrease in d13Cresin
mean has been detected in the
d
13
C of bulk plant material (e.g., tree rings; Loader et al.,
2003), it was not possible to identify a similar shift in
d13Cresin
mean between modern and pre-industrial resins (i.e.,
Colombian copal), possibly due to the relatively small num-
ber of copal samples analyzed.
4.1.3. Compositional factors
In addition to the isotopic variability that is caused by
environmental factors, some plant compounds are isotopi-
cally distinct from bulk plant matter. Cellulose, for exam-
ple, is slightly more enriched in
13
C(2–3&), whereas
lignin is slightly more depleted (3–4&) compared to the
bulk plant (Park and Epstein, 1960; Benner et al., 1987;
Schleser et al., 1999; Brugnoli and Farquhar, 2000)
(Fig. 9). The fractionation that causes the isotopic deviation
of these plant compounds, consequently, must occur after
the initial photosynthetic fixation of CO
2
by ribulose-1,5-
bisphosphate carboxylase oxygenase (Rubisco). Unlike cel-
lulose and lignin, the d
13
C values of plant resins overlap
with those of primary plant metabolites, such as leaf sugars
and bulk leaf tissues (Fig. 9), which indicates that fraction-
ation during resin biosynthesis (i.e., after the initial CO
2
fix-
ation) is minor, typically < 1&(Diefendorf et al., 2012).
This conclusion is important because it suggests that
d13Cresin
mean d13Cbulk plant
mean , and shifts in d13Cresin
mean reflect shifts
in the d13Cbulk plant
mean of C3 plants through time.
The observation that the d
13
C of individual plant com-
pounds can deviate from the d
13
C of the bulk plant, needs
to be considered whenever non-specific plants remains are
used in chemostratigraphic studies, since differential preser-
vation of the individual compounds can result in consider-
able shifts in d
13
C. For example, mean d
13
C values of bulk
organic matter (i.e., coal, coalified tissues, and cuticles)
from the Cretaceous and earliest Cenozoic (Strauss and Pe-
ters-Kottig, 2003) are 2.5&lower compared to d13Cresin
mean
values from the same time interval. This difference can be
readily explained by the preferential preservation of
13
C-de-
pleted lignin and the breakdown of
13
C-enriched cellulose
in fossil wood, from which the d13 Cresin
mean is unaffected.
4.2. Causes for the shift in 13Cresin
mean through time
The different types of modern resins investigated in this
study show a very similar distribution of d
13
C values
(Fig. 5), which indicates that the structure of their terpenoid
constituents has little or no effect on their d
13
C. This con-
clusion, however, contradicts claims that systematic differ-
ences in the fractionation behavior between gymnosperms
and angiosperms exist, with plant matter from angio-
sperms, including resins, generally being enriched in
13
C
by 2–3&(Diefendorf et al., 2012). Although it is conceiv-
able that certain plant taxa are prone to isotopic biases, due
to their ecological preferences (e.g., dipterocarpaceous trees
growing in areas with high MAP and closed canopy condi-
tions), there is no conclusive evidence that the fractionation
behavior of C3 plants is affected by their phylogenetic
position.
The tendency of fossil resins to be isotopically more en-
riched in
13
C compared to modern resins could be inter-
preted as diagenetic overprinting. After exudation, resins
generally lose a fraction of their terpenoid constituents—
mainly volatile mono- and sesquiterpenes—in addition to
water, whereas the non-volatile fractions polymerize rap-
idly. Once initial polymerization has occurred, any
additional maturation of the resins during fossilization
causes only minor structural changes (Lambert et al.,
Table 2
Mean d
13
C values of fossil resins from individual deposits,
including standard deviations, and number of samples analyzed
(arranged by age).
No Sample d
13
C
mean
(&)r(&)n
1 Columbian copal 27.5 0.6 6
2 Malaysian amber 26.1 0.3 12
3 Mexican amber 25.5 0.9 15
4 Dominican amber 24.8 1.6 32
5 Baltic amber 23.5 1.0 37
6 Bitterfeld amber 23.7 1.0 15
7 Giraffe kimberlite amber 22.5 1.9 23
8 Tiger Mountain amber 21.7 0.4 12
9 Ellesmere Island amber 21.6 1.5 12
10 Tadkeshwar amber 25.8 0.8 10
11 Panda kimberlite amber 24.8 0.2 5
12 Alaskan amber 23.6 1.2 15
13 Evansburg amber 22.3 0.8 14
14 Genesee amber 23.1 2.0 14
15 Albertan amber 23.9 1.8 39
16 Grassy Lake amber 23.7 1.4 34
17 Cedar Lake amber 23.2 1.4 12
18 New Jersey amber 22.1 1.1 36
19 Burmese amber 21.3 1.0 6
20 Spanish amber 21.4 1.5 21
21 Lebanese amber 21.1 1.2 21
22 Wealden amber 22.8 0.5 2
23 Italian amber 20.9 0.7 19
water availability
light exposure
local temperature
altitude
water composition (salinity etc.)
nutrient availability
local 13Cair
(e.g., canopy effect)
humidity
presence of parasites
Global
13Cair
pCO2
pO2
Regional
mean annual precipitation
Local
~7-8‰variability
13C effect
shift
shift
shift
shift
shift
etc.
combined effect:
increase
increase
increase
increase
increase
Fig. 8. Overview of global, regional, and local factors and their
effect on the d
13
C of plant organic matter.
R. Tappert et al. / Geochimica et Cosmochimica Acta 121 (2013) 240–262 251
2008; Tappert et al., 2011). These changes are unlikely to
influence the d
13
C of the resins significantly. In fact, if an
isotopic exchange had taken place during diagenesis, fossil
resins from different localities should show an erratic distri-
bution of the d13Cresin
mean arising from different diagenetic con-
ditions at each locality, but this is not observed.
Furthermore, if resins displayed a tendency to become
unstable and prone to carbon isotopic exchange through
time, they should become continuously more enriched in
13
C with increasing age, which is also not the case. Diage-
netic resetting should also diminish the natural variability
in d
13
C of resins, and this would result in samples from
individual localities converging on similar d
13
C values in-
stead of retaining their original range of d
13
C values.
As previously mentioned, an important factor that influ-
ences the d
13
C of plant organic matter, including resins, is
the carbon isotopic composition of the CO
2
in the atmo-
sphere (d13Catm
CO2). A shift in d13Catm
CO2, in fact, would result
in a comparable shift in the d
13
C of plant organic matter.
Although changes in d13Catm
CO2have been invoked to explain
the d
13
C variations observed in fossil plant organic matter
(Arens et al., 2000; Jahren, 2002;Strauss and Peters-Kottig,
2003; Hasegawa et al., 2003;Jahren et al., 2008), subse-
quent studies have questioned whether changes in d13Catm
CO2
alone can be responsible the observed d
13
C variability
(e.g.,Gro
¨cke, 2002; Beerling and Royer, 2002; Lomax
et al., 2012; Schubert and Jahren, 2012). If changes in
d13Catm
CO2were to be the primary cause for the d
13
C variabil-
ity of fossil plant organic matter, shifts in d13Cresin
mean should
correlate with independent proxies for d13Cresin
mean, including
those that are based on the d
13
C of marine carbonates
(e.g., calcareous tests of benthic foraminifera) (Zachos
et al., 2001;Tipple et al., 2010). However, the comparison
of the marine d
13
C record with the resin record shows very
little similarity (Fig. 11). A notable exception is the pro-
nounced decline in d
13
C values, which started in the late
Paleocene and lasted well into the early Eocene (57–
52 Ma, Fig. 11). This decline, which coincides with times
of peak Cenozoic temperatures (i.e., Early Eocene climatic
optimum, EECO), is discernible in both the marine and the
resin d
13
C record and it is only interrupted by short-lived
negative excursions in d
13
C and d
18
O(e.g., Paleocene Eo-
cene Thermal Maximum, PETM) (Zachos et al., 2008).
The cause for the rapid change in atmosphere composi-
tion during the PETM and other similar short-lived
(<20 ka) events remains a matter of debate (McInerney
and Wing, 2011), but the observation that these events
are associated with negative excursions in d
13
C and d
18
O
indicates that a considerable amount of
13
C-depleted CO
2
was added to the atmosphere during these intervals, causing
peak warming (hyperthermals). Proposed mechanisms to
create such a rapid influx of
13
C-depleted CO
2
into the
atmosphere include the dissociation and oxidation of meth-
ane hydrates in marine sediments (Sloan et al., 1992; Dick-
ens et al., 1995, 1997), the release of thermogenic CH
4
from
sediments through magmatism and its subsequent oxida-
tion (Svensen et al., 2004; Westerhold et al., 2009), an in-
crease in oxidation of organic matter caused by the
drying of epicontinental seas (Higgins and Schrag, 2006),
the release and oxidation of CH
4
stored in polar permafrost
(DeConto et al., 2012), and the rapid burning of Paleocene
terrestrial carbon (Kurtz et al., 2003). Since the effects of
these processes are considered to be short-lived, they are
unlikely to be responsible for the more gradual decline in
the d
13
C between the late Paleocene and early Eocene.
Therefore, other, more fundamental processes, such as
changes in the rate of oxidation of organic matter, must
be invoked as driving forces for this d
13
C decline.
Despite the exceptional d
13
C excursions in the early
Paleogene, the shifts in d
13
C of benthic foraminifera
through the Cenozoic are only moderate (2&) even at
their extremes. In fact, between the Middle Eocene and
the Miocene, d
13
C values of marine carbonates remain rel-
atively constant, with only minor shifts of 61&. The shifts
in d13Cresin
mean over the same time interval are approximately
6&, and they record a long-term trend towards more neg-
ative values (Fig. 11). In summary, changes in d13Catm
CO2may
influence the d
13
C of plant organic matter to some extent,
but their influence is insufficient to account for the magni-
tude of the observed shifts in d13 Cresin
mean.
Further comparison of the d13Cresin
mean with the marine sta-
ble isotope record shows that for much of the Cenozoic,
negative shifts in d13Cresin
mean broadly correlate with positive
shifts of a similar magnitude in the d
18
O of marine carbon-
ates (Fig. 11). The only pronounced exception is the late
Paleocene to early Eocene isotopic decline, which resulted
in a shift in both d13Cresin
mean and d
18
O to more negative values.
The marine d
18
O record is primarily a temperature record;
with less negative d
18
O values reflecting higher average
ocean surface temperatures and consequently higher global
temperatures (Emiliani, 1955; Bemis et al., 1998). The cor-
relation between the marine d
18
O and the d13Cresin
mean record,
therefore, indicates that resins that formed during intervals
of high global temperatures tend to be more enriched in
13
C. Since high global temperatures are generally linked
to high atmospheric pCO
2
(Royer, 2006), variations in
atmospheric pCO
2
seem to be an obvious alternative to ex-
plain shifts in the d
13
C of plant matter, including resins.
The results of recent growth experiments under controlled
pCO
2
and water/moisture availability suggest a hyperbolic in-
crease of plant fractionation (D
13
C) with increasing pCO
2
(Schubert and Jahren, 2012). These results are consistent with
most previous studies on
13
C fractionation in plants, which
also reported positive correlations between D
13
CandpCO
2
(Park and Epstein, 1960; Beerling and Woodward, 1993; Tre-
ydte et al., 2009). However, some studies found negative cor-
relations or no correlations (Beerling, 1996;Jahren et al.,
2008). The discrepancy between these studies is most likely
the result of differences in the control of the growth conditions
(i.e., water availability, light exposure, etc.) because variations
in the growth conditions can mask the isotopic effect caused
by increasing pCO
2
.
Irrespective of the extent of the pCO
2
impact, the obser-
vation that
13
C fractionation in plants increases with
increasing pCO
2
cannot account for the d
13
C shifts in the
resin record (Fig. 7). In the resin record, the least depleted
d13Cresin
mean compositions occur precisely during intervals with
predicted high atmospheric pCO
2
. We conclude, once
again, that some other process must be responsible for
the shifts in d13Cresin
mean.
252 R. Tappert et al. / Geochimica et Cosmochimica Acta 121 (2013) 240–262
Another important factor of the atmosphere that influ-
ences the isotopic fractionation in plants is the oxygen con-
tent. Preliminary experimental studies on C3 plants indicate
that fractionation of carbon during photosynthesis will in-
crease if the pO
2
in the ambient air is significantly higher
than modern values (pO
2
= 21%), resulting in plant bio-
mass that is isotopically more depleted in
13
C(Berner
et al., 2000). However, the opposite effect (i.e., a decrease
in fractionation) has been observed under lower than
modern pO
2
in ambient air (Beerling et al., 2002). This
means that plants growing under low O
2
conditions will
be more enriched in
13
C than those growing under high
-25-20 -30 -35
-15
Pectin
Asparatate
Hemicellulose
Aminoacids
Sap sugars
Cellulose
Starch
Leaf sugars
Proteins
-carotene
Isoprene
Lignin
Fatty acids
Total lipids
Bulk leaf
Resin
(this study)
13C (vs PDB)
Fig. 9. Ranges in d
13
C of different plant compounds, modified from Brugnoli and Farquhar (2000). The d
13
C range for resin is based on the
analyses of modern resins in this study; the range for bulk leaf matter is based on the compilations from Ko
¨rner et al. (1991), Diefendorf et al.
(2010), and Kohn (2010) (Fig. 4).
)aM
(egA
13C= 0
pCO2
Stomatal density
GEOCARB III
Pedogenic carb.
Atmospheric O (%)p2
Fig. 10. Comparison of calculated pO
2
values using different models to quantify the impact of pCO
2
on plant fractionation (dD13CpCO2).
Values for dD13CpCO2were calculated based on the relationship between pCO
2
and D
13
C proposed by Schubert and Jahren (2012),in
combination with pCO
2
estimates from (a) the density of stomata on fossil leaf cuticles (Retallack, 2001), (b) mass-balance modeling
(GEOCARB III; Berner and Kothavala, 2001), and (c) the d
13
C of pedogenic carbonates (Ekart et al., 1999). pO
2
values calculated for
dD13CpCO2¼0 (no effect of pCO
2
on D
13
C) are shown for comparison.
R. Tappert et al. / Geochimica et Cosmochimica Acta 121 (2013) 240–262 253
O
2
conditions. The strong influence of O
2
on the carbon
isotope fractionation in C3 plants is consistent with theoret-
ical considerations of leaf-gas exchange processes and the
fixation of CO
2
by plants. During photosynthesis, atmo-
spheric CO
2
is converted in a carboxylation reaction
through the enzyme Rubisco to energy-rich metabolites,
such as glucose. At the same time, atmospheric O
2
com-
petes with CO
2
for reaction sites on Rubisco. The reaction
of Rubisco with O
2
(oxygenation) partially offsets photo-
synthesis and results in the release of CO
2
in the process
of photorespiration. With higher atmospheric pO
2
, more
oxygen is used through photorespiration, which leads to
an increase of CO
2
within the plant cell and hence an in-
crease in
13
C fractionation. The relationship between
13
C-
fractionation in plants (D
13
C) and pCO
2
has been described
in general terms by Farquhar et al. (1982) as:
D13C¼aþðbaÞci=cað1Þ
where c
i
denotes the intercellular pCO
2
, and c
a
the pCO
2
in
the ambient atmosphere. The factors aand brepresent
fractionations that occur during diffusion through the sto-
mata (a= 4.4&) and during fixation of CO
2
by Rubisco
(b=27&), respectively.
The observed effect of atmospheric pO
2
on the d
13
Cof
plant tissues means that the d
13
C of plant organic matter
should be explored as a proxy for pO
2
in ancient atmo-
spheres (paleo-oxybarometer). Given that the d
13
C distribu-
tions of modern plant organic matter and resins are very
similar (Fig. 4), we conclude that amber has all the prere-
quisite qualities to be used as a pO
2
-proxy.
To estimate pO
2
from plant d
13
C it is necessary to quan-
tify the
13
C fractionation in the plant at varying atmo-
spheric pO
2
. This information is typically derived from
experimental work on plants in growth chambers under
controlled pO
2
conditions. However, only a few studies to
date have attempted to systematically determine the influ-
ence of pO
2
on plant fractionation, and the results of these
studies show large discrepancies (e.g., compare Smith et al.,
1976; Berry et al., 1972 and Beerling et al., 2002). These dis-
crepancies may also be the result of differences in the con-
trol of the experimental growth conditions, similar to
those observed in the CO
2
growth experiments, since even
small variations in the growth conditions (water and nutri-
ent availability, light exposure etc.) can mask the effects of
varying pO
2
. Irrespective of the cause of the discrepancies,
none of the proposed relationships that are based on exper-
imental data (e.g.,Berner et al., 2000; Beerling et al., 2002)
can account for the magnitude of the shifts in d13Cresin
mean.In
fact, calculated pO
2
values for most of the Cretaceous
and early Paleocene would be close to 0% if the relationship
between pO
2
and D
13
C proposed by Berner et al. (2000) and
Beerling et al. (2002) was applied to fossil resins. More real-
istic pO
2
values would require much larger empirical scaling
factors than those proposed by these authors.
Due to the inconsistencies in the experimental determi-
nation of the
13
C fractionation factor, we used a direct rela-
tionship between resin-derived D
13
C and atmospheric pO
2
as a first order approximation. Under this assumption,
the fractionation effect from photorespiration is propor-
tional to the pO
2
in the ambient air. The use of a scaling
factor, in this case, is not necessary. A direct relationship
may not be extended to extremely low values
(pO
2
< 10%), but we believe a direct relationship is a rea-
sonable assumption for moderate pO
2
levels as long as
physiological adaptations of plants (e.g., changes in metab-
olism) do not come into play. In our model, paleo-pO
2
at
the time of resin formation can be estimated as:
pO2¼D13Cresin dD13 CpCO2
D13CpOmodern
2ð2Þ
where D
13
C
resin
denotes the
13
C fractionation of resin-bear-
ing plants at the time of resin formation, and D
13
C repre-
sents the carbon isotope fractionation of plants in the
current atmosphere
(pOmodern
2¼21%;pCOmodern
2¼0:039%), which is approxi-
mately 21&. Based on Farquhar et al. (1982),D
13
C
resin
can be calculated as:
D13C13
resin ¼ðd13Catm
CO2d13Cmean
resin Þ1000
1000 þd13Cmean
resin
ð3Þ
In this equation, d13Catm
CO2denotes the isotopic composi-
tion of CO
2
in the ambient air. For modern plants,
d13Catm
CO2can be directly measured. For samples from the
geological past, values for d13Catm
CO2were calculated from
the d
13
C and d
18
O of benthic foraminiferal tests, following
the approach of Tipple et al. (2010). The d
18
O composition
is thereby used to correct for the temperature-dependency
of the fractionation between dissolved inorganic carbon
and CO
2
. For the Cenozoic, the d
13
C and d
18
O data to cal-
culate d13Catm
CO2were taken from Zachos et al. (2001). For the
late and mid-Cretaceous, d
13
C and d
18
O values from ben-
thic foraminifera were taken from Li and Keller (1999) (late
Campanian) and Huber et al. (1995) (Campanian–Albian).
Due to the lack of high-quality stable isotope data from
benthic foraminifera for the Early Cretaceous and the Tri-
assic, we used average d
13
C and d
18
O values from compila-
tions of Weissert and Erba (2004) and Korte et al. (2005),
which are based on bulk carbonate rocks and carbonaceous
macrofossils. Although this approach introduces some
additional uncertainties, most of the error on the d13Catm
CO2
calculations is due to uncertainties in the exact age place-
ment of the fossil resin samples.
The factor dD13CpCO2in Eq. (2) denotes the change of
D
13
C at varying pCO
2
relative to the fractionation at mod-
ern pCO
2
. As previously mentioned, the influence of pCO
2
on plant fractionation has recently been investigated in
growth experiments under systematically controlled growth
conditions by Schubert and Jahren (2012). These authors
observed a hyperbolic relationship between the fraction-
ation in C3 plants and pCO
2
, and formulated a best-fit
function, using their own and previously published data.
Based on this best-fit function proposed by Schubert and
Jahren (2012), we calculated dD13CpCO2as:
dD13CpCO2¼ð28:26Þð0:21ÞðpCO2þ25Þ
28:26 þð0:21ÞðpCO2þ25ÞD13 Cð4Þ
whereby pCO
2
is given in ppm. Although there is a general
consensus that the pCO
2
for most of the Cenozoic and
Mesozoic was higher than at present, considerable dispute
remains about the absolute pCO
2
values in the geological
254 R. Tappert et al. / Geochimica et Cosmochimica Acta 121 (2013) 240–262
past. To address the uncertainties in the paleo-pCO
2
esti-
mates and their impact on the pO
2
results, we calculated
dD13CpCO2using pCO
2
values derived by three independent
approaches: (1) mass balance calculations (GEOCARB III,
Berner and Kothavala, 2001), (2) the abundance of stomata
on leaf surfaces (Retallack, 2001), and (3) the d
13
C of ped-
ogenic carbonates (Ekart et al., 1999). Despite considerable
differences in the proposed pCO
2
levels between these three
models, the calculated pO
2
values show only minor devia-
tions (<±1.5%) for most of the Cenozoic and Mesozoic.
Larger discrepancies (<±3%) are restricted to the Paleocene
and early Eocene (Fig. 10). This indicates that most of the
uncertainty in the calculated pO
2
values rests on the accu-
racy of the relationship between D
13
C and pCO
2
as deter-
mined by Schubert and Jahren (2012).
4.2.1. Mesozoic and Cenozoic pO
2
evolution
A compilation of the calculated variations in pO
2
through time is compared to d13Cresin
mean, the marine stable iso-
tope record, and important environmental events in Fig. 11.
It is apparent from this compilation that for most of the
Cenozoic and Mesozoic pO
2
was considerably lower com-
pared to today. This is consistent with a reduced
13
C frac-
tionation (i.e., higher d13Cresin
mean values) of resin-producing
plants in the geological past. The particularly high
d13Cresin
mean values of early to mid-Cretaceous amber samples
suggest that the pO
2
at the time of their formation was as
low as 10–11% (Fig. 11). Such low pO
2
levels are regarded
to be below the limit of combustion and seem to contradict
the presence of charcoal in Cretaceous sedimentary se-
quences (Jones and Chaloner, 1991; Belcher and McElwain,
2008). However, the presence of charcoal in sedimentary
rocks itself does not preclude low atmospheric pO
2
, since
the charring process (i.e., pyrolysis) only occurs under
low-O
2
conditions. Furthermore, experiments to establish
the lower limit of atmospheric pO
2
required for combustion
did not consider certain factors that influence the combus-
tion in nature, such as the effect of wind, which may lower
the required pO
2
considerably.
Despite some uncertainties about absolute pO
2
values,
our low-pO
2
estimates for the mid-Cretaceous are consis-
tent with pO
2
estimates based on the abundance and mass
balance of carbon and pyritic sulfur in sedimentary rocks
during that same time interval (Berner and Canfield,
1989; Berner, 2006,Fig. 12). Our results, on the other hand,
question recently proposed paleo-pO
2
estimates that are
enec
o
tsi
e
lP
e
n
ec
oilP
e
ne
c
o
o
i
M
e
n
ec
oela
P
eneco
E
e
neco
g
i
l
O
naithcirts
aaM
n
a
in
a
pmaC
n
a
in
otnaS
n
aicainoC
nainoru
T
n
a
in
a
moneC
n
a
i
b
l
A
naitpA
naimerraB
n
a
ivire
t
ua
H
nainignalaV
naisairreB
nainraC
nain
i
daL
Deccan
North Atlantic
Ontong Java
Parana-Etendeka
Igneous
ytivitca
)sPIL(
MTEP
s
u
oecat
er
C
-
diM
greenhouse
stnevecitamilC
mret-gnoldna
sdn
ert-T
citcra
t
nA
steehseci
gnimrof
steehsecici
t
crA
gnimrof
g
nimra
w
gnilooc
gnilooc
g
nimr
aw
15
10 20
13 )
‰(C 18 )‰(O
15
10 20
O
foegatnecrePereh
psom
tani
2
drocerepot
osielbatseniraM
K
gP
OC
E
E
-28 62- 42-
82- 62- 42
-
13 )‰(C
)eulavnaem(sniserli
ssoF
-1 5
4
3
2
1
0
Ice-free temperature
0
4
812
012
-22
-22
NeogenePaleogeneLate CretaceousEarly Cretaceous
10
20
30
40
50
60
70
80
90
100
0
11
120
130
140
Age (Ma)
Triassic
220
230
-20
Fig. 11. Variations in predicted atmospheric pO
2
(red) and d13Cresin
mean (black) through time. The pO
2
estimates represent mean values for
different pCO
2
models, with the outline indicating the extreme values (Fig. 10). Error bars on d13Cresin
mean mark one standard deviation. The
marine stable isotope record for the Cenozoic is adopted from Zachos et al. (2001), and illustrates variations in d
13
C and d
18
O of benthic
foraminifera. Major climatic events, long-term temperature trends, and periods of large-scale igneous activity are shown for comparison. (For
interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
R. Tappert et al. / Geochimica et Cosmochimica Acta 121 (2013) 240–262 255
based on the abundance of charcoal in the fossil record,
which predict a higher-than-modern pO
2
for the Cretaceous
and the Cenozoic (Glasspool and Scott, 2010; Brown et al.,
2012,Fig. 12).
As previously mentioned, shifts in the Cenozoic d13 Cresin
mean
broadly resemble those in the marine d
18
O record, with the
exception of the negative d13Cresin
mean excursion in the early
Paleogene (Fig. 11). This indicates that changes in plant
fractionation, i.e., changes that are linked to varying pO
2
,
may also be linked to changes in global temperatures, and
thus pCO
2
. In fact, our pO
2
estimates for the Cretaceous
and Cenozoic show moderate to strong negative correla-
tions with pCO
2
estimates derived by independent methods
(Fig. 13). These correlations suggest that times of low pO
2
,
on a broad scale, were characterized by high pCO
2
and high
global temperatures, and vice versa.
For the Early Cretaceous, it appears that an early phase
of low pO
2
(12–13% O
2
) was followed by an interval of even
lower pO
2
levels (10–11% O
2
) during the Aptian to Turo-
nian. This phase of extremely-low pO
2
coincides with the
mid-Cretaceous greenhouse—a time interval characterized
by high atmospheric pCO
2
, high global temperatures, and
extensive ocean anoxic events (Jenkyns, 1980; Huber
et al., 1995; Wilson and Norris, 2001;Leckie et al., 2002).
From the Turonian to the end of the Cretaceous, atmo-
spheric pO
2
continuously rose to >15%, whereas pCO
2
and global temperatures declined.
The early Paleogene appears to have been characterized
by moderate pO
2
levels in the atmosphere (14–18% O
2
),
which declined to a Cenozoic minimum of 12% O
2
imme-
diately after the EECO (Fig. 11). The EECO marks the time
of highest global temperatures and pCO
2
during the Ceno-
zoic, as indicated by the exceptionally light d
18
O composi-
tion of marine carbonates deposited during that time
(Fig. 11). As previously mentioned, the time interval just
prior to the EECO (i.e., late Paleocene to early Eocene)
was marked by a steady decline in the d
13
C of marine car-
bonates and in d13Cresin
mean, indicating an increased influx of
13
C-depleted CO
2
into the atmosphere and an associated
decrease in d13Catm
CO2that lasted for 5 million years. In view
of our pO
2
estimates, this continuous influx of
13
C-depleted
CO
2
was most likely the result of an increase in the rate of
oxidation (oxidative weathering), caused by the rising glo-
bal temperatures and by elevated pO
2
levels during that
time interval. The increased rate of oxidative weathering
would have dramatically accelerated the oxidation of or-
ganic matter and let to a further increase in pCO
2
and
greenhouse forcing. At the same time, it would have re-
sulted in a continuous depletion of O
2
in the atmosphere,
which would explain the low pO
2
following the EECO.
After the height of the early Eocene greenhouse phase,
atmospheric pO
2
went into a long-term phase of continuous
recovery until it reached modern levels. The recovery in pO
2
was most likely the result of a decreased rate of oxidative
weathering and an associated increase in carbon burial.
This phase was also characterized by a continuous decrease
in pCO
2
and global temperatures, which culminated in the
glacial inception in the southern and northern hemispheres
during the terminal Paleogene.
4.2.2. Volcanism and its effects on pO
2
It is noteworthy that the intervals with lowest resin-
inferred pO
2
during the Cretaceous and the Cenozoic
were preceded by episodes of intense volcanism and the
Fig. 12. Comparison of predicted atmospheric pO
2
from this study with previously proposed models that are based on mass balance
calculations (GEOCARBSULF; Berner, 2006) and the abundance of charcoal in the rock record (Glasspool and Scott, 2010).
256 R. Tappert et al. / Geochimica et Cosmochimica Acta 121 (2013) 240–262
emplacement of some of the largest known igneous prov-
inces (LIPs). These include the Ontong-Java Plateau and
the Parana-Etendeka Province (Early Cretaceous), as well
as the Deccan and North Atlantic Provinces (Late Creta-
ceous to Paleocene) (Fig. 11,Larson and Erba, 1999).
The emplacement of these LIPs was associated with the
expulsion of vast amounts of mantle-derived volcanic gases,
which are typically dominated by H
2
O and CO
2
, but also
contain sulfur dioxide (SO
2
) and other trace gases (Sy-
monds et al., 1994). The expulsion of such large quantities
of volcanic gases, particularly of CO
2
, provides a valid
mechanism to cause a rapid increase in atmospheric pCO
2
and an associated increase of greenhouse forcing and global
temperatures during these time intervals. Although there is
still some dispute over the long-term effect of volcanogenic
CO
2
emissions (Self et al., 2005), the hypothesis that LIP
magmatism contributed to changes in atmospheric compo-
sition and global climate is supported by the marine d
13
C
record, which indicates that the late Paleocene decline of
d
13
C was preceded by a notable rise in d
13
C values that
started in the mid-Paleocene. This rise in d
13
C coincides
with phases of intense volcanism in the Deccan area of In-
dia, and in the North Atlantic Igneous Province (Fig. 11).
Marine carbonates deposited at that time reached d
13
C val-
ues of around +2&(Fig. 11;Zachos et al., 2001). These
unusually positive values are consistent with a strong influx
of mantle-derived CO
2
into the atmosphere because mantle
CO
2
is enriched in
13
C relative to CO
2
derived from organic
matter and has a rather restricted d
13
C range (8to3&,
mean: 5&)(Exley et al., 1986; Javoy and Pineau, 1991).
Although the episodes of LIP volcanism in the Early
Cretaceous and the Late Cretaceous/Paleocene may have
not been sufficient to sustain long-term greenhouse condi-
tions, they may have served as a trigger for the subsequent
increase of oxidative weathering. This, in turn, would have
provided a feedback for the further rise in pCO
2
and the
prolonged greenhouse episodes that followed the LIP volca-
nism. In this scenario, the elevated pO
2
that prevailed be-
fore the onset of the LIP volcanism (i.e., in the Early and
Late Cretaceous) would have been an important factor that
determined the extent of the subsequent greenhouse phases.
After the cessation of LIP volcanism and the end of the
greenhouse periods in the Cretaceous and the Cenozoic,
atmospheric pO
2
was at its minimum, but slowly increased
over the following tens of millions of years, as the result of a
decrease in oxidative weathering (associated with an in-
crease in carbon burial) and the absence of the CO
2
-influx
from LIP magmatism. After the mid-Cretaceous, the cycle
of increasing pO
2
and decreasing pCO
2
lasted for 40 mil-
lion years, but was interrupted by renewed LIP volcanism
at the end of the Cretaceous. The Cenozoic decline of
pCO
2
and increase in pO
2
, on the other hand, continued
uninterrupted for 50 million years from the Middle Eo-
cene into modern times. The prolonged absence of large-
scale volcanism during that time period may have been a
prerequisite for pCO
2
and global temperatures to decrease
sufficiently to allow sea ice to form on the polar oceans,
ultimately triggering the onset of glaciation in both hemi-
spheres in the latest Paleogene.
4.2.3. Effects of pO
2
on animal evolution
Although it has been proposed that atmospheric pO
2
has
a profound effect on the physiology and evolution of ani-
mals, and that gigantism in animals is linked to high pO
2
(Graham et al., 1995; Clapham and Karr, 2012), our data
preclude such a link for the evolution of giant theropod
and sauropod dinosaurs. Considering the low pO
2
that pre-
vailed during the Cretaceous and, at least, part of the Tri-
assic, it is more reasonable to assume that a high primary
photosynthetic productivity (i.e., food availability) caused
by high pCO
2
and high global temperatures was the key
factor in the evolution of dinosaurs and their tendency to
Atmospheric O (%)p2
ci
rehpsomtA)vmpp(OCp2
Stomatal density
GEOCARB III
Pedogenic carbonates
Fig. 13. Plot of predicted atmospheric pO
2
against pCO
2
estimates based on the density of stomata on fossil leaf cuticles (Retallack, 2001),
mass-balance modeling (GEOCARB III; Berner and Kothavala, 2001), and the d
13
C of pedogenic carbonates (Ekart et al., 1999).
R. Tappert et al. / Geochimica et Cosmochimica Acta 121 (2013) 240–262 257
develop large body sizes. A similar conclusion can be drawn
for the Eocene radiation of large placental mammals, which
has previously been linked to high pO
2
levels as well (Fal-
kowski et al., 2005). Based on our data, however, this time
interval was also characterized by low pO
2
.
5. CONCLUSIONS AND FUTURE DIRECTIONS
Due to their ability to polymerize into highly resistant
organic macromolecules that are demonstrably representa-
tive of whole-plant carbon fractionation, resins can pre-
serve pristine isotopic signatures over geological time.
This makes them ideal natural biomarkers for chemostrati-
graphic studies. As with all plant organic constituents, the
d
13
C of C3 plant resins is influenced by a range of local
environmental and ecological factors, which result in con-
siderable isotopic variability. The variability seen in resins
is directly comparable to that of primary plant biomass,
which suggests that the d
13
C of resins can be used as a mea-
sure of the d
13
C of bulk plant organic matter. To minimize
statistical biases caused by variations in the growth condi-
tions of individual plants, it is necessary to analyze suffi-
ciently large sample populations and to compare their
mean isotopic values (d13Cresin
mean). This is particularly impor-
tant for the comparison of resins of different ages. Since
major diagenetic influences on the d
13
C of fossil resins are
negligible, variations in d13Cresin
mean can be related to long-term
global environmental changes. In fact, our present compila-
tion and analysis of d
13
C values from exhaustively sampled
resins, which is the largest data set of its kind, strongly sug-
gests that the pO
2
of ancient atmospheres is reliably repre-
sented in the isotopic composition of these secondary
metabolites.
Resins can be used as proxies of atmospheric composi-
tion for much of the Cenozoic and Mesozoic, and perhaps
even into the Paleozoic, given that resin-producing plants
have existed since at least the Carboniferous. Resins from
different localities, different botanical sources, and different
ages are directly comparable because they consist of similar
macromolecules, and they are resistant to post-depositional
isotopic exchange. These advantages are complemented by
the fact that the analysis of resins does not require extensive
sample treatment, which limits biases introduced by labora-
tory procedures. Therefore, resins can provide a consistent
and highly resolved terrestrial d
13
C record, particularly
when samples from stratigraphically and chronologically
well-characterized sedimentary units are analyzed.
To improve our understanding of the fractionation
behavior of plant organic matter in relation to atmospheric
composition, it is also necessary to conduct additional, sys-
tematically controlled growth chamber experiments on a
wider range of plants, including resin-producing plant taxa.
This information will help to more accurately quantify
compositional changes of the atmosphere that occurred
throughout Earth’s history. Additional constraints on the
composition of ancient atmospheres may also be gained
by combining isotopic information with careful analysis
of biological inclusions within amber—e.g., insects, plants,
and fungi—which are frequently preserved with exquisite
detail.
ACKNOWLEDGMENTS
We are grateful to Jim Basinger, Madelaine Bo
¨hme, Xavier
Delclo
`s, Michael Engel, George Mustoe, Philip Perkins, Guido
Roghi, Alexander Schmidt, and Chris Williams for generously pro-
viding samples used in this study. Thomas Stachel is thanked for
providing access to the FTIR spectrometer housed in his labora-
tory. We also thank Stacey Gibb and Gabriela Gonzalez for assis-
tance at various stages, and we would like to thank the three
anonymous reviewers, whose constructive comments have im-
proved this publication considerably. Much of this research was
supported by the Natural Science and Engineering Research Coun-
cil of Canada (Discovery awards to A.P. Wolfe and K. Muehlenb-
achs, Postdoctoral fellowship to R.C. McKellar). J. Ortega was
supported by the Ministerio de Economı
´a y Competitividad, Ful-
bright Espan
˜a, and FECYT.
REFERENCES
Alonso J., Arillo A., Barron E., Corral J. C., Grimalt J., Lopez J.
F., Lopez R., Martinez-Delclos X., Ortuno V., Penalver E. and
Trincao P. R. (2000) A new fossil resin with biological
inclusions in lower Cretaceous deposits from Alava (Northern
Spain, Basque-Cantabrian Basin). J. Paleontol. 74, 158–178.
Anderson K. B. (2006) The nature and fate of natural resins in the
geosphere—XII. Investigation of C-ring aromatic diterpenoids
in Raritan amber by pyrolysis–GC-matrix isolation FTIR-MS.
Geochem. Trans. 7,2.
Anderson K. B. and Botto R. E. (1993) The nature and fate of
natural resins in the geosphere—III. Re-evaluation of the
structure and composition of Highgate Copalite and Glessite.
Org. Geochem. 20, 1027–1038.
Anderson K. B. and LePage B. A. (1995) Analysis of fossil resins
from Axel Heiberg Island, Canadian Arctic. In Amber, Resinite,
and Fossil Resins (eds. K. B. Anderson and J. C. Crelling).
American Chemical Society, Washington, pp. 170–192.
Anderson K. B., Winans R. E. and Botto R. E. (1992) The nature
and fate of natural resins in the geosphere—II. Identification,
classification and nomenclature of resinites. Org. Geochem. 18,
829–841.
Arens N. C. and Jahren A. H. (2000) Carbon isotope excursion in
atmospheric CO
2
at the Cretaceous-Tertiary boundary: evi-
dence from terrestrial sediments. Palaios 15, 314–322.
Arens N. C., Jahren A. H. and Amundson R. (2000) Can C3 plants
faithfully record the carbon isotopic composition of atmo-
spheric carbon dioxide? Paleobiology 26, 137–164.
Azar D., Ge
`ze R. and Acra F. (2010) Lebanese amber. In
Biodiversity of Fossils in Amber from the Major World Deposits
(ed. D. Penney). Siri Scientific Press, Manchester, pp. 271–298.
Bechtel A., Gratzer R., Sachsenhofer R. F., Gusterhuber J., Lu
¨cke
A. and Pu
¨ttmann W. (2008) Biomarker and carbon isotope
variation in coal and fossil wood of Central Europe through the
Cenozoic. Palaeogeogr. Palaeoclimatol. Palaeoecol. 262, 166–
175.
Beck C. W. (1986) Spectroscopic investigation of amber. Appl.
Spec. Rev. 22, 57–110.
Beck C. W., Wilbur E., Meret S., Kossove D. and Kermani K.
(1965) The infrared spectra of amber and the identification of
Baltic amber. Archaeometry 8, 96–109.
Beerling D. J. (1996)
13
C discrimination by fossil leaves during the
late-glacial climate oscillation 12–10 ka BP: measurements and
physiological controls. Oecologia 108, 29–37.
Beerling D. J. and Royer D. L. (2002) Fossil plants as indicators of
the Phanerozoic global carbon cycle. Ann. Rev. Earth Planet.
Sci. 30, 527–556.
258 R. Tappert et al. / Geochimica et Cosmochimica Acta 121 (2013) 240–262
Beerling D. J. and Woodward F. I. (1993) Ecophysiological
responses of plants to global environmental change since the
last glacial maximum. New Phytol. 125, 641–648.
Beerling D. J., Lake J. A., Berner R. A., Hickey L. J., Taylor D. W.
and Royer D. L. (2002) Carbon isotope evidence implying high
O
2
/CO
2
ratios in the Permo-Carboniferous atmosphere. Geo-
chim. Cosmochim. Acta 66, 3757–3767.
Belcher C. M. and McElwain J. C. (2008) Limits for combustion in
low O
2
redefine paleoatmospheric predictions for the Mesozoic.
Science 321, 1197–1200.
Bemis B. E., Spero H. J., Bijma J. and Lea D. W. (1998)
Reevaluation of the oxygen isotopic composition of planktonic
foraminifera: experimental results and revised paleotempera-
ture equations. Paleoceanography 13, 150–160.
Benner R., Fogelt M. L., Sprague E. K. and Hodsong R. E. (1987)
Depletion of
13
C in lignin and its implications for stable carbon
isotope studies. Nature 329, 708–710.
Berner R. A. (1999) Atmospheric oxygen over Phanerozoic time.
Proc. Natl. Acad. Sci. U.S.A. 96, 10955–10957.
Berner R. A. (2006) GEOCARBSULF: a combined model for
Phanerozoic atmospheric O
2
and CO
2
.Geochim. Cosmochim.
Acta 70, 5653–5664.
Berner R. A. and Canfield D. E. (1989) A new model for atmospheric
oxygen over Phanerozoic time. Am.J.Sci.289, 333–361.
Berner R. A. and Kothavala Z. (2001) GEOCARB III: a revised
model of atmospheric CO
2
over Phanerozoic time. Am. J. Sci.
301, 182–204.
Berner R. A., Petsch S. T., Lake J. A., Beerling D. J., Popp B. N.,
Lane R. S., Laws E. A., Westley M. B., Cassar N., Woodward
F. I. and Quick W. P. (2000) Isotope fractionation and
atmospheric oxygen: implications for Phanerozoic O
2
evolu-
tion. Science 287, 1630–1633.
Berner R. A., VandenBrooks J. M. and Ward P. D. (2007) Oxygen
and evolution. Science 316, 557–558.
Berry J. A., Troughton J. H. and Bjo
¨rkman O. (1972) Effect of
oxygen concentration during growth on carbon isotope dis-
crimination in C3 and C4 species of Atriplex.Carnegie Inst.
Washington Yearb. 71, 158–161.
Bowen G. J. and Beerling D. J. (2004) An integrated model for soil
organic carbon and CO
2
: implications for paleosol carbonate
pCO
2
paleobarometry. Global Biogeochem. Cycles 18, GB1026.
Bray P. S. and Anderson K. B. (2009) Identification of Carbon-
iferous (320 million years old) class Ic amber. Science 326, 132–
134.
Broughton P. L. (1974) Conceptual framework for geographic-
botanical affinities of fossil resin. Can. J. Earth Sci. 11, 583–594.
Brown S. A. E., Scott A. C., Glasspool I. J. and Collinson M. E.
(2012) Cretaceous wildfires and their impact on the Earth
system. Cretaceous Res. 36, 162–190.
Brugnoli E. and Farquhar G. D. (2000) Photosynthetic fraction-
ation of carbon isotopes. In Photosynthesis: Physiology and
Metabolism (eds. R. C. Leegood, T. D. Sharkey and S. Von
Caemmerer). Kluwer Academic, Publishers, pp. 399–434.
Cerling T. E. and Hay R. L. (1986) An isotopic study of paleosol
carbonates from Olduvai Gorge. Quat. Res. 25, 63–78.
Clapham M. E. and Karr J. A. (2012) Environmental and biotic
controls on the evolutionary history of insect body size. Proc.
Natl. Acad. Sci. U.S.A. 109, 10927–10930.
Clarke L. J. and Jenkyns H. C. (1999) New oxygen isotope
evidence for long-term Cretaceous climatic change in the
Southern Hemisphere. Geology 27, 699–702.
Coplen T. B., Kendall C. and Hopple J. (1983) Comparison of
stable isotope reference samples. Nature 302, 236–238.
Cruickshank R. D. and Ko K. (2003) Geology of an amber locality
in the Hukawng Valley. Northern Myanmar. J. Asian Earth Sci.
21, 441–455.
Cunningham A., Gay I. D., Oehlschlager A. C. and Langenheim J.
H. (1983)
13
C NMR and IR analyses of structure, aging and
botanical origin of Dominican and Mexican ambers. Phyto-
chem. 22, 965–968.
DalCorso J., Preto N., Kustatscher E., Mietto P., Roghi G. and
Jenkyns H. C. (2011) Carbon-isotope variability of Triassic
amber, as compared with wood and leaves (Southern Alps,
Italy). Palaeogeogr. Palaeoclimatol. Palaeoecol. 302, 187–193.
Dawson T. E., Mambelli S., Plamboeck A. H., Templer P. H. and
Tu K. P. (2002) Stable isotopes in plant ecology. Ann. Rev.
Ecol. Syst. 33, 507–559.
DeConto R. M., Galeotti S., Pagani M., Tracy D., Schaefer K.,
Zhang T., Pollard D. and Beerling D. J. (2012) Past extreme
warming events linked to massive carbon release from thawing
permafrost. Nature 484, 87–92.
Dickens G. R., Castillo M. M. and Walker J. C. G. (1997) A blast
of gas in the latest Paleocene; simulating first-order effects of
massive dissociation of oceanic methane hydrate. Geology 25,
259–262.
Dickens G. R., O’Neil J. R., Rea D. K. and Owen R. M. (1995)
Dissociation of oceanic methane hydrate as a cause of the
carbon isotope excursion at the end of the Paleocene. Paleoce-
anography 10, 965–971.
Diefendorf A. F., Freeman K. H. and Wing S. L. (2012)
Distribution and carbon isotope patterns of diterpenoids and
triterpenoids in modern temperate C3 trees and their geological
significance. Geochim. Cosmochim. Acta 85, 342–356.
Diefendorf A. F., Mueller K. E., Wing S. L., Koch P. L. and
Freeman K. H. (2010) Global patterns in leaf
13
C discrimina-
tion and implications for studies of past and future climate.
Proc. Natl. Acad. Sci. 107, 5738–5743.
Edwards T. W. D., Graf W., Trimborn P., Stichler W., Lipp J. and
Payer H. D. (2000) D
13
C response surface resolves humidity and
temperature signals in trees. Geochim. Cosmochim. Acta 64,
161–167.
Ekart D. D., Cerling T. E., Montanez I. P. and Tabor N. (1999) A
400 million year carbon isotope record of pedogenic carbonate:
implications for paleoatmospheric carbon dioxide. Am. J. Sci.
299, 805–827.
Emiliani C. (1955) Pleistocene temperatures. J. Geol. 63, 538–578.
Exley R. A., Mattey D. P., Clague D. A. and Pillinger C. T. (1986)
Carbon isotope systematics of a mantle hotspot: a comparison
of Loihi seamount and MORB glasses. Earth Planet. Sci. Lett.
78, 189–199.
Falkowski P. G., Katz M. E., Milligan A. J., Fennel K., Cramer B. S.,
Aubry M. P., Berner R. A., Novacek M. J. and Zapol W. M.
(2005) The rise of oxygen over the past 205 million years and the
evolution of large placental mammals. Science 309, 2202–2204.
Farquhar G. D., Ehleringer J. R. and Hubick K. T. (1989) Carbon
isotope discrimination and photosynthesis. Ann. Rev. Plant
Physiol. Plant Mol. Biol. 40, 503–537.
Farquhar G. D., O’Leary M. H. and Berry J. A. (1982) On the
relationship between carbon isotope discrimination and the
intercellular carbon dioxide concentration in leaves. Aust. J.
Plant Physiol. 9, 121–137.
Francey R. J., Allison C. E., Etheridge D. M., Trudinger C. M.,
Enting I. G., Leuenberger M., Langenfelds R. L., Michel E. and
Steele L. P. (1999) A 1000-year high precision record of d
13
Cin
atmospheric CO
2
.Tellus 51B, 170–193.
Francis J. E. (1988) A 50 million-year old fossil forest from
Strathcona Fiord, Ellesmere Island, Arctic Canada: evidence
for a warm polar climate. Arctic 41, 314–318.
Fuhrmann R. (2005) Die Bernsteinlagersta
¨tte Bitterfeld, nur ein
Ho
¨hepunkt des Vorkommens von Bernstein (Succinit) im
Tertia
¨r Mitteldeutschlands. Z. Deutsch. Gesell. Geowiss. 156,
517–529.
R. Tappert et al. / Geochimica et Cosmochimica Acta 121 (2013) 240–262 259