Introduction
The Eocene (55.8–33.9 Ma) generally was much warmer
than present-day, although temperatures varied significantly across this time
interval (Zachos et al., 2008). Climate of the early Eocene was extremely
warm, particularly during the early Eocene Climatic Optimum (EECO;
51–53 Ma), and the Paleocene-Eocene Thermal Maximum (PETM;
∼ 55.9 Ma). However, global climatic conditions cooled
significantly by the middle to late Eocene (40–36 Ma). Indeed,
small, ephemeral ice-sheets and Arctic sea ice likely existed during the
latest Eocene (Moran et al., 2006; Zachos et al., 2008).
Many authors have suggested that changes in temperature during the
Phanerozoic were linked to atmospheric pCO2 (Petit et al., 1999;
Retallack, 2001; Royer, 2006). Central to these discussions are records
across the Eocene, as this epoch spans the last major change from a
“greenhouse” world to an “icehouse” world. The Eocene pCO2
record remains incomplete and debated (Kürschner et al., 2001; Royer et
al., 2001; Beerling et al., 2002; Greenwood et al., 2003; Royer, 2003). Most
pCO2 reconstructions have focused on the Cretaceous–Tertiary and
Paleocene–Eocene boundaries (65–50 Ma) and the middle Eocene. In
particular, there are few reconstructions for the late middle Eocene (Pagani
et al., 2005; Maxbauer et al., 2014). In addition, the pCO2
reconstruction results have varied based on different proxies. Various
methods having been used in pCO2 reconstruction mainly include the
computer modeling methods: GEOCARB-I, GEOCARB-II, GEOCARB-III, GEOCARB-SULF
and the proxies: ice cores, paleosol carbonate, phytoplankton, nahcolite,
Boron, and stomata parameters.
The abundance of stomatal cells can be measured on modern leaves and
well-preserved fossil leaves. Various plants show a negative correlation
between atmospheric CO2 concentration and stomatal density (SD),
stomatal index (SI), or both. As such, these parameters have been determined
in fossil leaves to reconstruct past pCO2; examples include
Ginkgo (Retallack, 2001, 2009a; Beerling et al., 2002; Royer, 2003;
Kürschner et al., 2008; Smith et al., 2010), Metasequoia (Royer,
2003; Doria et al., 2011), Taxodium (Stults et al., 2011),
Betula (Kürschner et al., 2001; Sun et al., 2012),
Neolitsea (Greenwood et al., 2003), and Quercus
(Kürschner et al., 1996, 2001), Laurus and Ocotea
(Kürschner et al., 2008). Recently, positive correlations between
stomatal index or stomatal frequency and pCO2 have been reported
based on fossil Typha and Quercus (Bai et al., 2015; Hu et
al., 2015). However, the tropical and subtropical moist broadleaf forest
conifer tree Nageia has not been used previously in paleobotanical
estimates of pCO2 concentration.
Herein, we firstly document correlations between stomatal properties and
atmospheric CO2 concentrations using leaves of the extant species
Nageia motleyi (Parl.) De Laub. that were collected over the last
2 centuries. This provides a training data set for application to fossil
representatives of Nageia. We secondly measure stomatal parameters
on fossil Nageia leaves from late middle Eocene of South China to
estimate past CO2 levels. The work provides further insights for
discussing Eocene climate change.
Background
Stomatal proxy in pCO2 research
Stomatal information gathered from careful examination of leaves has been
widely used for reconstructions of past pCO2 concentrations
(Beerling and Kelly, 1997; Doria et al., 2011). The three main parameters are
stomatal density (SD), which is expressed as the total number of stomata
divided by area, epidermal density (ED), which is expressed as the total
number of epidermal cells per area, and the stomatal index (SI), which is
defined as the percentage of stomata among the total number of cells within
an area [SI=SD×100/(SD+ED)]. Woodward (1987)
considered that both SD and SI had inverse relationships with atmospheric
CO2 during the development of the leaves. Subsequently, McElwain
(1998) created the stomatal ratio (SR) method to reconstruct pCO2.
SR is a ratio of the stomatal density or index of a fossil
[SD(f) or SI(f)] to that of
corresponding nearest living equivalent [SD(e) or
SI(e)], expressed as follows:
SR=SI(e)/SI(f).
The stomatal ratio method is a semi-quantitative method of reconstructing
pCO2 concentrations under certain standardizations. An example is
the “Carboniferous standardization” (Chaloner and McElwain, 1997), where
one stomatal ratio unit equals two RCO2 units:
SR=2RCO2
and the value of RCO2 is the pCO2 level divided by the
pre-industrial atmospheric level (PIL) of 300 ppm (McElwain, 1998) or
that of the year when the nearest living equivalent (NLE) was collected
(Berner, 1994; McElwain, 1998):
RCO2=C(f)/300or RCO2=C(f)/C(e).
The estimated pCO2 level can then be expressed as follows:
C(f)=0.5×C(e)×SD(e)/SD(f)orC(f)=0.5×C(e)×SI(e)/SI(f),
where C(f) is the pCO2 represented by the fossil leaf,
and C(e) is the atmospheric CO2 of the year when the leaf
of the NLE species was collected (McElwain and Chaloner, 1995, 1996; McElwain
1998). The equation adapts to the pCO2 concentration prior to
Cenozoic.
Another standardization, the “Recent standardization” (McElwain, 1998), is
expressed as one stomatal ratio unit being equal to one RCO2 unit:
SR=1RCO2.
According to the equations stated above, the pCO2 concentration can
be expressed as
C(f)=Ce×SD(e)/SD(f)orC(f)=Ce×SI(e)/SI(f).
This standardization is usually used for reconstruction based on Cenozoic
fossils (Chaloner and McElwain, 1997; McElwain, 1998; Beerling and Royer,
2002).
Kouwenberg et al. (2003) proposed some special stomatal quantification
methods for conifer leaves with stomata arranged in rows. The stomatal number
per length (SNL) is expressed as the number of abaxial stomata plus the
number of adaxial stomata divided by leaf length in millimeters. Stomatal
rows (SRO) are expressed as the number of stomatal rows in both stomatal
bands. Stomatal density per length (SDL) is expressed as the equation
SDL=SD×SRO. True stomatal density per length
(TSDL) is expressed as the equation TSDL=SD× band
width (in millimeters). The band width on Nageia motleyi leaves was
measured as leaf blade width.
Map showing the distribution of extant and fossil Nageia
and their mean annual temperature (modified after the map from
http://nelson.wisc.edu/sage/data-and-models/atlas/maps/avganntemp/atl_avganntemp.jpg).
Review of extant and fossil Nageia
The genus Nageia, including seven living species, is a special group
of Podocarpaceae, a large family of conifers mainly distributed in the
Southern Hemisphere. Nageia has broadly ovate-elliptic to
oblong-lanceolate, multi-veined (without a mid-vein), spirally arranged or in
decussate, and opposite or sub-opposite leaves (Cheng et al., 1978; Fu et
al., 1999). Generally, Nageia is divided into two sections,
Nageia Sect. Nageia and Nageia Sect.
Dammaroideae (Mill 1999, 2001). Both sections are mainly distributed
in southeast Asia and Australasia from north latitude 30∘ to nearly
the equator (Fu, 1992; Fig. 1). Four species of the N. section
Nageia – Nageia nagi (Thunberg) O. Kuntze,
N. fleuryi (Hickel) De Laub., N. formosensis (Dummer) C. N.
Page, and N. nankoensis (Hayata) R. R. Mill – have hypostomatic
leaves where stomata only occur on the abaxial side. One species of this
section – N. maxima (De Laub.) De Laub. – is characterized by
amphistomatic leaves, but where only a few stomata are found on the adaxial
side (Hill and Pole, 1992; Sun, 2008). Both N. wallichiana (Presl)
O. Kuntze and N. motleyi of the N. section
Dammaroideae are amphistomatic with abundant stomata distributed on
both sides of the leaf. This is especially true for N. motleyi,
which has approximately equal stomata numbers on both surfaces (Hill and
Pole, 1992; Sun, 2008).
The fossil record of Nageia can be traced back to the Cretaceous.
Krassilov (1965) described Podocarpus (Nageia)
sujfunensis Krassilov from the Lower Cretaceous of Far East Russia.
Kimura et al. (1988) reported Podocarpus (Nageia)
ryosekiensis Kimura, Ohanaet Mimoto, an ultimate leafy branch
bearing a seed, from the Early Barremian in southwestern Japan. In China, a
Cretaceous petrified wood, Podocarpus (Nageia)
nagi Pilger, was discovered from the Dabie Mountains in central
Henan, China (Yang et al., 1990). Jin et al. (2010) reported an upper Eocene
Nageia leaf named N. hainanensis Jin, Qiu, Zhu et Kodrul
from the Changchang Basin of Hainan Island, South China. Recently, Liu et
al. (2015) found another leaf species N. maomingensis Jin et Liu
from upper middle Eocene of Maoming Basin, South China. Although some of the
Nageia fossil materials described in the above studies (Krassilov,
1965; Jin et al., 2010; Liu et al., 2015) have well-preserved cuticles, these
studies are mainly concentrated on morphology, systematics, and
phytogeography.
Here we try to reconstruct the pCO2 concentration based on stomatal
data of Nageia maomingensis Jin et Liu. Among the modern
Nageia species mentioned above, N. motleyi was considered
as the NLE species of N. maomingensis (Liu et al., 2015). However,
because of the species-specific inverse relationship between atmospheric
CO2 partial pressure and SD (Woodward and Bazzaz, 1988), it is
necessary to explore whether the SD and SI of N. motleyi show
negative correlations with the CO2 concentration before applying the
stomatal method. Both N. maomingensis and N. motleyi are
amphistomatic, suggesting that both upper and lower surfaces of leaves might
be used to estimate the pCO2 concentrations.
Material and methods
Extant leaf preparation
We examined 12 specimens of extant Nageia motleyi from different
herbaria (Table 1). We removed one or two leaves from each specimen, and took
three fragments (0.25 mm2) from every leaf (Fig. 2a) and numbered
them for analysis.
The numbered fragments were boiled for 5–10 min in water.
Subsequently, after being macerated in a mixed solution of 10 % acetic
acid and 10 % H2O2 (1:1) and heated in the thermostatic water
bath at 85 ∘C for 8.5 h, the reaction was stopped when the
specimen fragments turned white and semitransparent. The cuticles were then
rinsed with distilled water until the pH of the water became neutral. After,
the cuticles were treated in Schulze's solution (one part of potassium
chlorate saturated solution and three part of concentrated nitric acid) for
30 min, rinsed in water, and then treated with 8 % KOH (up to
30 min). The abaxial and adaxial cuticles were separated with a hair
mounted on needle. Finally, the cuticles were stained with 1 % Safranin T
alcoholic solution for 5 min, sealed with Neutral Balsam and observed
under LM.
Modern Nageia motleyi (Parl.) De
Laub. samples and atmospheric
CO2 values of their collection dates from ice core data (Brown,
2010).
Herbarium
Collection number
Collecting locality
Collectors
Number of leaf
Collection date
CO2 (ppmv)
samples
LE
No. 2649
Malaysia
O. Beccari,
1
1868
289.23
A/GH
No. bb. 17229
150 m, Riau on Ond. Karimon
Neth. Ind. For. Service
2
1932
306.19
A/GH
No. bb. 18328
5 m, Z. O. afd. v. Borneo Tidoengsche Landen
Neth. Ind. For. Service
2
1934
306.46
A/GH
No. bb. 21151
500 m, Z. O. afd. Borneo, Poeroek Tjahoe Tahoedjan
Neth. Ind. For. Service
2
1936
306.76
KEP
No. 30887
Kata Tinggi, Johor, Malaysia
E. J. H. Corner
1
1936
306.76
KEP
No. 57329
Batang Padang, Perak, Malaysia
Unkonwn
2
1947
309.82
KEP
No. 57330
Batang Padang, Perak, Malaysia
Unkonwn
2
1947
309.82
KEP
No. 55897
Batang Padang, Perak, Malaysia
Unkonwn
2
1947
309.82
KEP
No. 61064
Batang Padang, Perak, Malaysia
Syed Woh
2
1947
309.82
E
No. bb. 40798
51 m, Kuala Trengganu-Besut Road, Bukit Bintang Block,
Gunong Tebu Forest reserve, Malaysia
J. Sinclair and K. bin Salleh
2
1955
313.73
KEP
No. 80548
Gombak, Selangor, Malaysia
Rahim
1
1965
320.04
KEP
No. 33343
Jelebu, Negeri Sembilan, Malaysia
S. K. Yap
2
1987
348.98
Note: A/GH: Harvard
University Herbarium, Harvard University, 22 Divinity Avenue, Cambridge,
Massachusetts 02138, USA (www.huh.harvard.edu). E: The
Herbarium of Royal Botanic Garden, Edinburgh EH3 5LR, Scotland, UK
(www.rbge.org.uk). LE: The Herbarium of the V. L. Komarov
Botanical Institute of the Russian Academy of Sciences, Prof. Popov Street 2,
Saint Petersburg 197376, Russia (www.binran.ru). KEP: Kepong
Herbarium, Forest Research Institute Malaysia, 52109 Kepong, Selangor,
Malaysia (http://www.frim.gov.my/).
Sampling areas and counting rules are shown.
(a) Nageia motleyi (Parl.) De Laub. leaf. Black squares in the middle of the leaf show the sampling areas
for preparing the cuticles. (b) The abaxial side of the cuticle from
N. motleyi leaf. Black circles show the counted stomatal complexes.
(c) N. maomingensis Jin et Liu. Red squares in the middle
of the leaf indicate the sampling areas. (d) The abaxial side of the
fossil cuticle. Red circles show the counted stomatal complexes. Scale bars:
(a) and (c) =1 cm; (b) and
(d) =50 µm.
Fossil leaf preparation
Maoming Basin (21∘42′33.2′′ N, 110∘53′19.4′′ E) is
located in southwestern Guangdong, South China including Cretaceous and
Tertiary strata. Tertiary strata are fluvial and lacustrine sedimentary
units, divided into the Gaopengling, Laohuling, Shangcun, Huangniuling and
Youganwo formations in descending order, aged from late Eocene to early
Oligocene (Wang et al., 1994).
Four fossil leaves of Nageia maomingensis were recovered from the
Youganwo (MMJ1-001) and Huangniuling (MMJ2-003, MMJ2-004 and MMJ3-003)
formations of Maoming Basin, South China. Further information on the sections
is provided by Liu et al. (2015). Importantly, the formations span a
depositional age of approximately 42.0–38.5 Ma which was considered
as late Eocene by Wang et al. (1994), but it can be recognized as late middle
Eocene according to Walker and Geissman (2009).
Macrofossil cuticular fragments were taken from the middle part of each
fossil leaf (Fig. 2c) and directly treated with Schulze's solution for
approximately 1 h and 5–10% KOH for 30 min (Ye, 1981).
The cuticles were observed and photographed under a Carl Zeiss Axio Scope A1
light microscope (LM). All fossil specimens and cuticle slides are housed in
the Museum of Biology of Sun Yat-sen University, Guangzhou, China.
Summary of stomatal parameters of the adaxial surface from modern
Nageia motleyi (Parl.) De Laub.
Collection number
Collection date
CO2 (ppmv)
SD (mm-2)
SI (%)
x
σ
s.e.
t*s.e.
n
x
σ
s.e.
t*s.e.
n
No. 2649
1868
289.23
78.60
15.44
1.41
2.76
120
3.44
0.66
0.06
0.12
120
No. bb. 17229
1932
306.19
62.14
17.20
1.78
3.50
93
2.89
0.68
0.07
0.14
93
No. bb. 18328
1934
306.46
64.57
15.05
1.58
3.11
90
3.39
1.01
0.11
0.21
90
No. bb. 21151
1936
306.76
65.45
11.14
1.17
2.30
90
3.94
0.74
0.08
0.15
90
No. SFN30887
1936
306.76
66.90
16.10
1.27
2.49
161
3.61
0.92
0.07
0.14
161
No. 61064
1947
309.82
56.71
16.81
1.95
3.83
74
3.27
1.26
0.15
0.29
74
No. 57330
1947
309.82
67.37
15.97
2.04
4.01
61
3.70
0.82
0.10
0.20
61
No. 57329
1947
309.82
67.85
15.61
1.70
3.34
84
3.50
0.90
0.10
0.20
84
No. 55897
1947
309.82
66.74
14.10
1.78
3.48
63
3.18
0.66
0.08
0.16
63
No. 40798
1955
313.73
45.89
13.81
1.12
2.20
151
3.03
0.87
0.07
0.14
151
No. KEP80548
1965
320.04
52.94
11.25
0.85
1.67
175
2.81
0.61
0.05
0.09
175
No. FRI33343
1987
348.98
52.25
12.05
0.77
1.51
242
2.87
0.69
0.04
0.09
242
Mean
–
–
62.28
14.54
1.45
2.85
117
3.30
0.52
0.08
0.16
117
Note: x: mean, σ: standard deviation, s.e.: standard
error of mean, n: numbers of photos counts (200×), t*s.e.:
95 % confidence interval.
Summary of stomatal parameters of the abaxial surface from modern
Nageia motleyi (Parl.) De Laub.
Collection number
Collection date
CO2 (ppmv)
SD (mm-2)
SI (%)
x
σ
s.e.
t*s.e.
n
x
σ
s.e.
t*s.e.
n
No. 2649
1868
289.23
82.71
12.23
1.02
2.00
144
3.89
0.58
0.05
0.09
144
No. bb. 17229
1932
306.19
69.16
14.23
1.48
2.90
93
3.13
0.58
0.06
0.12
93
No. bb. 18328
1934
306.46
69.92
14.38
1.52
2.97
90
3.99
1.08
0.11
0.22
90
No. bb. 21151
1936
306.76
75.68
15.74
1.66
3.25
90
4.66
0.88
0.09
0.18
90
No. SFN30887
1936
306.76
76.18
12.51
0.99
1.93
161
4.42
0.89
0.07
0.14
161
No. 61064
1947
309.82
60.93
11.02
1.39
2.72
63
3.05
0.62
0.08
0.15
63
No. 57330
1947
309.82
75.82
14.14
1.82
3.58
60
4.38
0.84
0.11
0.21
60
No. 57329
1947
309.82
71.74
16.84
1.75
3.42
93
3.72
0.62
0.06
0.13
93
No. 55897
1947
309.82
78.63
13.41
1.75
3.42
59
4.41
1.00
0.13
0.26
59
No. 40798
1955
313.73
53.22
13.88
1.12
2.19
155
3.71
0.93
0.07
0.15
155
No. KEP80548
1965
320.04
67.22
13.97
1.07
2.09
171
3.70
0.80
0.06
0.12
171
No. FRI33343
1987
348.98
59.09
12.10
0.79
1.55
233
3.69
0.86
0.06
0.11
233
Mean
–
–
70.03
13.70
1.36
2.67
118
3.90
0.81
0.08
0.16
118
Note: x: mean, σ: standard deviation, s.e.: standard
error of mean, n: numbers of photos counts (200×), t*s.e.:
95 % confidence interval.
Summary of stomatal parameters of the combined data of the adaxial
and abaxial surfaces from modern Nageia motleyi (Parl.) De Laub.
Collection number
Collection date
CO2 (ppmv)
SD (mm-2)
SI (%)
x
σ
s.e.
t*s.e.
n
x
σ
s.e.
t*s.e.
n
No. 2649
1868
289.23
80.84
13.74
0.85
1.66
264
3.69
0.66
0.04
0.08
264
No. bb. 17229
1932
306.19
65.65
16.13
1.18
2.32
186
3.01
0.64
0.05
0.09
186
No. bb. 18328
1934
306.46
67.24
14.92
1.11
2.18
180
3.69
1.08
0.08
0.16
180
No. bb. 21151
1936
306.76
70.57
14.53
1.08
2.12
180
4.30
0.89
0.07
0.13
180
No. SFN30887
1936
306.76
71.54
15.12
0.84
1.65
322
4.01
0.99
0.05
0.11
322
No. 61064
1947
309.82
58.65
14.54
1.24
2.43
137
3.17
1.02
0.09
0.17
137
No. 57330
1947
309.82
71.56
15.61
1.42
2.78
121
4.03
0.89
0.08
0.16
121
No. 57329
1947
309.82
69.90
16.33
1.23
2.41
177
3.62
0.77
0.06
0.11
177
No. 55897
1947
309.82
72.49
14.95
1.35
2.65
122
3.77
1.04
0.09
0.18
122
No. 40798
1955
313.73
49.60
14.31
0.82
1.60
306
3.37
0.96
0.05
0.11
306
No. KEP80548
1965
320.04
60.00
14.53
0.78
1.53
346
3.25
0.84
0.05
0.09
346
No. FRI33343
1987
348.98
55.61
12.53
0.58
1.13
475
3.28
0.88
0.04
0.08
475
Mean
–
–
66.14
14.77
1.04
2.08
235
3.60
0.89
0.06
0.12
235
Note: x: mean, σ: standard deviation, s.e.: standard
error of mean, n: numbers of photos counts (200×), t*s.e.:
95 % confidence interval.
Correlation between SD and SI versus CO2 concentration for
modern Nageia motleyi. (a) Trends of SD with CO2
concentration for the adaxial surface. (b) Trends of SD with
CO2 concentration for the abaxial surface. (c) Trends of SI
with CO2 concentration for the adaxial surface. (d) Trends
of SI with CO2 concentration for the abaxial surface.
(e) Trends of SD with CO2 concentration for the combined
data of both leaf surfaces. (f) Trends of SI with CO2
concentration for the combined data of both leaf surfaces.
Stomatal counting strategy and calculation methods
The basic stomatal parameters, SD, ED, and SI were counted based on analyzing
pictures taken with a light microscope (LM). A total of 2816 pictures
(200× magnification of Zeiss LM) of cuticles from 21 leaves of
N. motleyi were counted. Each counting field was 0.366 mm2.
We used a standard sampling protocol (Poole and Kürschner, 1999),
counting all full stomata in the image plus stomata straddling the left and
top margins, as presented in Fig. 2b and d.
The SNL, SRO, SDL, and TSDL were also determined based on LM images. A total
of 2293 pictures (200× magnification of Zeiss LM) of the cuticles from
21 leaves of N. motleyi were counted. Each counting field was
0.366 mm2. None of the aforementioned counting areas overlapped and
they were larger than the minimum area (0.03 mm2) for statistics
(Poole and Kürschner, 1999). In this study, the stomatal data of both
surfaces are applied in pCO2 reconstruction because both the fossil
and NLE species are amphistomatic.
Results
Correlations between the CO2 concentrations and stomatal
parameters of Nageia motleyi
The SD and SI data of the adaxial sides of N. motleyi leaves are
presented in Table 2. The SDs and SIs average 62.28 mm-2 and
3.30 %, respectively. However, the SDs and SIs data of the abaxial sides,
summarized in Table 3, give higher average values (70.03 mm-2 in
SDs and 3.90 % in SIs) than those from the adaxial sides. The combined SD
and SI of the adaxial and abaxial surfaces average 66.14 mm-2 and
3.60 %, respectively (Table 4).
Summary of stomatal parameters from modern Nageia motleyi
(Parl.) De Laub. (Kouwenberg et
al., 2003).
Collection
Collection
CO2
SNL
SDL
TSDL
n
number
date
(ppmv)
No. 2649
1868
289.23
11.64
394.38
1455.10
264
No. bb. 17229
1932
306.19
9.19
337.98
1280.12
186
No. bb. 18328
1934
306.46
8.71
378.92
1277.63
180
No. bb. 21151
1936
306.76
9.62
376.93
1517.21
180
No. SFN30887
1936
306.76
10.55
325.08
735.38
240
No. 61064
1947
309.82
8.19
282.04
1200.66
133
No. 57330
1947
309.82
9.67
397.83
1397.33
119
No. 57329
1947
309.82
10.13
350.98
1672.50
176
No. 55897
1947
309.82
10.48
379.06
1486.13
122
No. 40798
1955
313.73
10.29
175.14
933.85
305
No. KEP80548
1965
320.04
9.36
266.16
585.72
263
No. FRI33343
1987
348.98
9.84
252.20
1181.51
125
Mean
–
–
9.81
326.39
1226.93
191
Correlation between SNL, SDL, and TSDL versus CO2
concentration for modern Nageia motleyi. (a) Trends of SNL
with CO2 concentration for the adaxial surface. (b) Trends
of SDL with CO2 concentration for the adaxial surface.
(c) Trends of TSDL with CO2 concentration for the adaxial
surface.
Figure 3 shows the relationships between the stomatal parameters (SD and SI)
of modern N. motleyi and the atmospheric CO2 concentration
(SD-CO2 relationship and SI-CO2 relationship). R2 values
in the SD-CO2 relationship from the adaxial and abaxial surfaces of
N. motleyi are up to 0.4667 and 0.3824 (Fig. 3a, b), suggesting that
the stomatal densities of N. motleyi are inverse to the CO2
concentrations. However, Fig. 3c and d indicate no relationship between the
SIs and CO2 concentrations for the extremely low level of the R2
values (0.2558 and 0.0248). Figure 3e and f based on the combined data also
show that SD inversely responds to the atmospheric CO2 concentration
(R2=0.4421), while SI has almost no relationship with the atmospheric
CO2 concentration (R2=0.1177).
The mean values of SNL, SDL, and TSDL are 9.81, 326.39 and
1226.93 no.mm-1, respectively (Table 5). Figure 4 shows the
relationships between SNL (SDL, TSDL) and CO2 concentrations. The low
R2 values in Fig. 4a and c indicate that SNL (R2=0.0643) and TSDL
(R2=0.0788) have no relationship with the CO2 concentration in
this study. Figure 4b shows that there is a weak reverse relevance between
SDL and the CO2 concentration (R2=0.3154).
Compared with the SDL method, the SD-based method shows a larger R2 value,
indicating a stronger relevance between the SD and CO2
concentrations. In this study, the pCO2 is reconstructed based on
the regression equations of SD-CO2 relationship. Additionally, the
stomatal ratio method can be also used in estimating pCO2
concentration of late middle Eocene based on stomatal densities (SDs) of the
fossil species N. maomingensis and extant species N. motleyi. The SD results of specimen No. 18328 are selected to reconstruct
the pCO2 concentration, because they are closest to the fitted
equations in Fig. 3. This specimen was collected by the Netherlands Indies Forest Service
from Borneo Tidoengsche Landen, in 1934 at an altitude
of 5 m and CO2 concentration of 306.46 ppmv (Brown,
2010).
Summary of stomatal parameters of the adaxial surface of fossil
Nageia and pCO2 [C(f)] estimates results.
Species
Age
SD (mm-2)
SI (%)
SR
pCO2 (ppmv)
C(f) (ppmv)
x
σ
s.e.
n
x
σ
s.e.
n
x
t*s.e.
x
t*s.e.
x
t*s.e.
MMJ1-001
Late Eocene
52.5
17.1
3.1
30
2.08
0.7
0.1
30
1.40
0.20
333.6
13.9
428.7
62.0
MMJ2-003
Late Eocene
42.3
12.9
2.4
30
1.80
0.6
0.1
30
1.82
0.41
356.8
10.5
557.6
126.2
MMJ2-004
Late Eocene
39.9
13.6
2.5
30
1.66
0.6
0.1
30
1.88
0.33
362.4
11.0
576.5
101.9
MMJ3-003a
Late Eocene
43.2
17.7
3.2
30
1.67
0.7
0.1
30
1.92
0.44
354.8
14.4
587.2
135.7
Mean
Late Eocene
44.5
16.3
1.5
120
1.80
0.7
0.1
120
1.75
0.18
351.9
6.6
537.5
56.5
Note: x: mean, σ:
standard deviation, s.e.: standard error of mean, n: numbers of photos
counts (400×), t*s.e.: 95 % confidence interval,
pCO2: the result based the regression approach, C(f):
the result based on the stomatal method.
Summary of stomatal parameters of the abaxial surface of fossil
Nageia and pCO2 [C(f)] estimates results.
Species
Age
SD (mm-2)
SI (%)
SR
pCO2 (ppmv)
C(f) (ppmv)
x
σ
s.e.
n
x
σ
s.e.
n
x
t*s.e.
x
t*s.e.
x
t*s.e.
MMJ1-001
Late Eocene
47.7
17.7
3.2
30
2.11
0.8
0.2
30
1.68
0.24
368.6
16.2
515.6
72.3
MMJ2-003
Late Eocene
50.9
18.3
3.3
30
2.12
0.8
0.1
30
1.59
0.23
360.9
16.6
486.0
70.7
MMJ2-004
Late Eocene
48.2
15.8
2.9
30
2.14
0.7
0.1
30
1.65
0.25
367.4
14.5
504.6
77.3
MMJ3-003a
Late Eocene
48.9
12.6
2.7
22
1.85
0.5
0.1
22
1.54
0.19
365.4
13.5
472.3
59.0
Mean
Late Eocene
48.9
16.2
1.5
112
2.07
0.7
0.1
112
1.62
0.12
365.6
7.7
496.1
35.7
Note: x: mean, σ:
standard deviation, s.e.: standard error of mean, n: numbers of photos
counts (400×), t*s.e.: 95 % confidence interval,
pCO2: the result based the regression approach, C(f):
the result based on the stomatal method.
Summary of stomatal parameters of the combined data of the adaxial
and abaxial surfaces of fossil Nageia and pCO2
[C(f)] estimates results.
Species
Age
SD (mm-2)
SI (%)
SR
pCO2 (ppmv)
C(f) (ppmv)
x
σ
s.e.
n
x
σ
s.e.
n
x
t*s.e.
x
t*s.e.
x
t*s.e.
MMJ1-001
Late Eocene
50.1
17.5
2.3
60
2.09
0.8
0.1
60
1.54
0.16
349.7
10.6
471.2
47.8
MMJ2-003
Late Eocene
46.5
16.3
2.1
60
1.96
0.7
0.1
60
1.71
0.25
358.3
9.8
524.1
75.7
MMJ2-004
Late Eocene
44.0
15.8
2.0
60
1.90
0.7
0.1
60
1.77
0.17
364.3
9.5
542.9
52.6
MMJ3-003a
Late Eocene
45.6
16.1
2.2
52
1.75
0.6
0.1
52
1.78
0.29
360.4
10.4
544.6
88.3
Mean
Late Eocene
46.6
16.4
1.1
232
1.93
0.7
0.1
232
1.70
0.11
358.1
5.0
519.9
35.0
Note: x: mean, σ:
standard deviation; s.e.: standard error of mean; n: numbers of photos
counts (400×); t*s.e.: 95 % confidence interval.
pCO2: the result based the regression approach; C(f):
the result based on the stomatal method.
The pCO2 estimates results
The regression approach
The summary of stomatal parameters of the fossil Nageia and
reconstruction results are provided in Tables 6–8. The mean SD and SI values
of the adaxial surface are 44.5 mm-2 and 1.8 %, respectively
(Table 6). The mean SD and SI values of the abaxial surface are
48.9 mm-2 and 2.07 %, respectively (Table 7).
Based on the regression approach, the pCO2 was reconstructed as
351.9±6.6 ppmv and 365.6±7.7 ppmv according to the
SD of adaxial and abaxial sides. The combined SD value is an average of
46.6 mm-2 (Table 8), giving the reconstructed pCO2 of
358.1±5.0 ppmv.
The stomatal ratio method
Mean SR value of the adaxial side (SR=1.75±0.18) is a little
larger than that of the abaxial side (SR=1.62±0.12) in fossil
Nageia leaves (Tables 6, 7). The pCO2 reconstruction
results are 537.5±56.5 ppmv (Table 6) and 496.1±35.7 ppmv (Table 7) based on the adaxial and abaxial cuticles,
respectively. Based on the combined SD of both leaf sides, the pCO2
result is 519.9±35.0 ppmv.
The partial pressure of CO2decreases with elevation (Gale, 1972).
Jones (1992) proposed that the relationship between elevation and partial
pressure in the lower atmosphere can be expressed as P=-10.6E+100, where
E is elevation in kilometers and P is the percentage of partial pressure
relative to sea level. Various studies corroborate that SI and SD of many
plants have positive correlations with altitude (Körner and Cochrane,
1985; Woodward, 1986; Woodward and Bazzaz, 1988; Beerling et al., 1992;
Rundgren and Beerling, 1999) while they are negatively related to the partial
pressure of CO2 (Woodward and Bazzaz, 1988). Therefore, it is
essential to take elevation calibration into account during pCO2
concentration estimates. However, Royer (2003) pointed out that it is
unnecessary to provide this conversion when trees lived at < 250 m
in elevation. In this paper, the nearest living equivalent species,
Nageia motleyi, grows at 5 m in elevation with P=99.9,
suggesting that CO2 concentration estimates were only underestimated
by 0.1 %. Consequently, no correction is needed for the reconstruction
result in this study. After being projected into a long-term carbon cycle
model (GEOCARB III; Berner and Kothavalá, 2001), the results of this
study compare well with CO2 concentrations for corresponding age
within their error ranges (Fig. 5).
The pCO2 reconstruction results and extant CO2
concentrations are projected onto the long-term carbon cycle model
(GEOCARB III; Berner and Kothavalá, 2001). The pCO2 results
based on the regression approach and stomatal ratio method are represented by
red and blue squares, respectively.
Discussion
Stomatal parameters response to CO2
For modern Nageia, we find that SD decreases as atmospheric
CO2 concentrations increase, but that SI does not. Generally, SI is
more sensitive in response to the atmospheric CO2 concentration than
SD (Beerling, 1999; Royer, 2001). However, the reverse case has been observed
from some flora. For example, Kouwenberg et al. (2003) reported that SD is
better than SI in reflecting the negative relationships with CO2 in
conifer needles, accounting for the special paralleled mode of the ordinary
epidermal and stomatal formation. Although Nageia is broad-leaved
rather than needle-leaved, it also has well-paralleled epidermal cells.
Compared with SD, the SDL has weaker correlation with CO2 at a
smaller R2. The SNL and TSDL have no response to the change of
CO2. The insensitivity of SNL, SDL, and TSDL might account for the
characters of broad-leaved leaf shape and paralleled epidermal cells. The SNL
should be applied to conifer needles with single file of stomata (Kouwenberg
et al., 2003). The SDL and TSDL were considered as the most appropriate
method when the stomatal rows grouped in bands in a hypo- or amphistomatal
conifer needle species (Kouwenberg et al., 2003). Considering all the
stomatal parameters above, SD appears to be the most sensitive to
CO2.
The SD-CO2 correlation shows one value from leaf No. 40798 offset
from the others. The SI-CO2 correlation shows different offset values
in different leaf sides. The offset values might be affected by leaf maturity
and light intensity. However, it is hard to distinguish whether a fossil leaf
was young or mature, or grew in a sunny or shady environment.
The R2 value (0.5) of SD-CO2 based on the adaxial side is higher
than from the abaxial side and the combination of both sides, indicating that
the correlation of SD-CO2 is stronger than the other parameters
herein. Therefore, the SD on the adaxial side is the best in reconstructing
pCO2. The reconstruction result based on the regression approach is
351.9±6.6 ppmv lower than the one based on the stomatal ratio
method (Table 6), and it is relatively lower than the results based on the
other proxies (Fig. 6; Freeman and Hayes, 1992; Pagani et al., 2005; Maxbauer
et al., 2014). However, the result based on stomatal ratio method is
537.5±56.5 ppmv, which is fairly close to GEOCARB III
predictions (Fig. 5) and historical reconstruction trends (Fig. 6).
Atmospheric CO2 estimates from proxies over the past
60 million years. The horizontal dashed line indicates monthly atmospheric
CO2 concentration for March 2015 at Mauna Loa, Hawaii
(401.5 ppmv) (Pieter and Keeling, 2015). The vertical lines show the
error bars. The data are from the supporting data of Beerling and Royer
(2011) and references in Table 9. The lower blue star shows the reconstructed
result based on the regression approach. The higher one presents the result
of stomatal ratio method.
pCO2 estimates proxies and corresponding references.
Proxies
References
Boron
Pearson et al. (2009), Seki et al. (2010)
B / Ca
Tripati et al. (2009)
Phytoplankton
Freeman and Hayes (1992), Stott (1992),
Pagani et al. (1999, 2005),
Henderiks and Pagani (2008), Seki et al. (2010)
Nahcolite
Lowenstein and Demicco (2006)
Liverworts
Fletcher et al. (2008)
Paleosols
Cerling (1992), Koch et al. (1992),
Ekart et al. (1999), Royer et al. (2001),
Nordt et al. (2002), Retallack (2009b),
Huang et al. (2013)
Stomata
Van der Burgh et al. (1993),
Kürschner et al. (1996, 2001, 2008),
McElwain (1998), Royer et al. (2001, 2003),
Greenwood et al. (2003), Beerling et al. (2009),
Retallack (2009a), Smith et al. (2010),
Doria et al. (2011),
Roth-Nebelsick et al. (2012; 2014),
Grein et al. (2013), Maxbauer et al. (2014)
Paleoclimate reconstructed history
The pCO2 levels throughout the Cenozoic were generally lower than
during much of the Cretaceous, but probably also decreased significantly from
the early to late Eocene. However, there is a wide range of estimates for the
Eocene (Koch et al., 1992; Sinha and Stott, 1994; Ekart et al., 1999;
Greenwood et al., 2003; Royer, 2003; Pagani et al., 2005; Wing et al., 2005;
Lowenstein and Demicco, 2006; Fletcher et al., 2008; Zachos et al., 2008;
Beerling et al., 2009; Bijl et al., 2010; Smith et al., 2010; Doria et
al., 2011; Kato et al., 2011; Maxbauer et al., 2014).
Smith et al. (2010) reconstructed the value of the early Eocene
pCO2 ranging from 580±40 to 780±50 ppmv using the
stomatal ratio method (recent standardization) based on both SI and SD. A
climatic optimum occurred in the middle Eocene (MECO): the reconstructed
CO2 concentrations are mainly between 700 and 1000 ppmv
during the late middle Eocene climate transition (42–38 Ma) using
stomatal indices of fossil Metasequoia needles, but concentrations
declined to 450 ppmv toward the top of the investigated section
(Doria et al., 2011). Jacques et al. (2014) used CLAMP to calibrate climate
change in Antarctica during the early-middle Eocene, suggesting a seasonal
alternation of high- and low-pressure systems over Antarctica during the
early-middle Eocene. Spicer et al. (2014) also reconstructed a relatively
lower cool temperature than δ18O records (Keating-Bitonti et
al., 2011) in the middle Eocene of Hainan Island, South China using CLAMP,
indicating a not uniformly warm climate in the low latitude during the
Eocene. An overall decreasing trend of the pCO2 level was presented
after the middle Eocene (Fig. 6; Retallack, 2009b). The ice-sheets started to
appear in the Antarctic during the Late Eocene (Zachos et al., 2001), then
the temperature suffered an apparent further decrease from the late Eocene
onwards (Fig. 6).
In conclusion, although various results were made by different pCO2
reconstruction proxies at the same time, their entire decreasing tendency of
pCO2 level is remarkably consistent with each other since the
Eocene (Fig. 6). Figure 6 shows that during the Eocene the temperature was
higher than at present. Comparing to the estimates of late middle Eocene
pCO2 by Doria et al. (2011), the present result of 351.9±6.6 ppmv based on the regression approach shows a remarkably lower
pCO2 level, while the one based on the stomatal ratio method of
537.5±56.5 ppmv is within the variation range of
500–1000 ppmv, which is closely consistent with the pCO2
changes over the geological ages (Fig. 6). The world was dynamic in the
Paleogene, including in the late middle Eocene, when the MECO occurred. Thus,
the exact age matters, and it is possible that the values may differ because
of slight offsets in time.