Professional Documents
Culture Documents
JESSICA E. TIERNEY
ABSTRACT.—Microbial communities adjust the chemical structure of their cell membranes in response to
environmental temperature. This enables the development of lipid-based paleothermometers such as the
glycerol dialkyl glycerol tetraether (GDGT) proxies described here. Surface-sediment calibrations estab-
lish a strong empirical relationship between the relative distribution of GDGTs and temperature. GDGT
proxies can be used in marine, lacustrine, and paleosol sequences as long as the organic material is not
thermally mature. Thus far, GDGT proxies have been applied to sediments dating back to the middle Ju-
rassic. Many of the key uncertainties of these proxies are related to our emerging understanding of ar-
chaeal (and for the branched GDGTs, bacterial) ecology.
THE LAST twenty-five years have witnessed the that are mid-Paleogene in age or younger, and has
growth and expansion of a new technique for es- been a mainstay of paleoclimate reconstruction
timating paleotemperatures based on the use of since its development in the late 1980s (Herbert,
biomarkers: organism-specific, fossil lipids (fats) 2003). There are, however, some limitations to the
found in sedimentary organic matter. Biomarkers alkenone proxy. Extant haptophyte species that
enhance our ability to reconstruct paleoenviron- the modern calibrations of the alkenone proxy are
ments in geologic history, in part because they are based on evolved relatively recently (e.g., 0.28
subject to a different set of diagenetic alterations Ma for Emiliania huxleyi, Thierstein et al., 1977),
than hard parts of fossils. They may be well pre- so alkenones found in older sedimentary se-
served in sedimentary environments where tradi- quences can only be presumed to be produced by
tional skeletal temperature proxies (e.g., calcare- ancient relatives with a similar temperature sensi-
ous foraminifera) typically are not, thereby en- tivity. Also, the alkenone calibrations to SST are
hancing the range of depositional conditions from limited to ~28°C, such that the proxy cannot be
which paleotemperature information can be ex- used in some areas of the tropical oceans, nor in
tracted. ancient hothouse climates.
Biomarker-based paleothermometers are built In the last decade, a new biomarker paleo-
upon the biochemical principle that microorgan- temperature approach based on the relative abun-
isms adjust the rigidity of their cell membrane dances of glycerol dialkyl glycerol tetraethers
structures by altering the number of double bonds, (GDGTs) has been developed that complements
rings or branches in response to environmental and extends the capabilities of the alkenone proxy.
temperature (e.g., Ray et al., 1971; Kita et al., GDGTs are relatively large (up to 86 carbon at-
1973; Nozawa et al., 1974). The first biomarker- oms) membrane lipids produced by Archaea and
based paleotemperature proxy was based on the some Bacteria. They are common, typically abun-
relative distribution of unique long-chain (C37) tri- dant, easy-to-analyze lipids found in a variety of
and di-unsaturated ketones (alkenones) produced environments (lakes, soils, oceans). Such attrib-
by marine haptophyte algae, such as the ubiqui- utes make GDGTs a promising universal tool for
tous coccolithophorid Emiliania huxleyi (Volkman estimating temperatures in the geologic past, es-
et al., 1980; Marlowe et al., 1984). The alkenone pecially in deep geologic time.
In Reconstructing Earth’s Deep-Time Climate—The State of the Art in 2012, Paleontological Society Short Course,
November 3, 2012. The Paleontological Society Papers, Volume 18, Linda C. Ivany and Brian T. Huber (eds.),
pp. 115–131. Copyright © 2012 The Paleontological Society.
THE PALEONTOLOGICAL SOCIETY PAPERS, VOL. 18
678!9:4;./04<=15!)*)+: 6-8!>.1023/=!)*)+:
)*)+,$
!"#!"#$% brGDGT-III !"#!"$I$
)*)+,"
!"#!"#$$ +H;/!999 brGDGT-IIIb !"#!"$J'
!"#!"%&%
brGDGT-IIc !"#!"$#%
-./01.231/45
./B<4<:4?/.
!"#!"%&%G
brGDGT-I !"#!"$%%
6>8!-4??40!@451.!A/1=B.4C;:!,!9:4;./04<=15!)*)+:
+H;/!9 brGDGT-Ib
!"#!"$%$
brGDGT-Ic !"#!"$"'
FIGURE 1.—A) Core structures of the isoprenoidal glycerol dialkyl glycerol tetraethers (isoGDGTs), with mass-to-
charge ratios (m/z) on the right associated with analysis via HPLC/positive ion APCI. B) Common polar head-
groups of intact isoGDGTs, which would replace one or both of the terminal hydroxyl groups indicated in (A). C)
Core structures of branched glycerol dialkyl glycerol tetraethers (brGDGTs).
116
TIERNEY: RECONSTRUCTING PALEOTEMPERATURES USING GDGT THERMOMETRY
!"#$%&#"
smaller amounts of a crenarchaeol regioisomer 4;'-@-AB?L ...
(Figure 1; oft-abbreviated as cren’) that plays a
role in the temperature predictability of the
isoGDGT paleothermometer: relatively more of .
the regioisomer is observed at higher tempera-
tures.
BrGDGTs share many structural characteris-
tics with the isoGDGTs, including the ether bonds
to terminal glycerol groups, a feature in mem-
brane lipids typically seen only in Archaea (with .K
some exceptions, e.g., see Weijers et al., 2006a, .:
and references therein). Yet both the chain archi- '()"
tecture (branched vs. isoprenoidal) and the stereo-
chemistry of the glycerol groups on the brGDGTs FIGURE 2.—A) HPLC trace of isoGDGTs from a
tropical marine sediment. TEX86 value is calibrated to
are diagnostic of Bacteria (Weijers et al., 2006a).
SST using the Kim, et al. (2010) TEXH86 calibration.
BrGDGTs are commonly found in soils, peats, (B) HPLC trace of brGDGTs from a freshwater lake.
lakes, and marginal/deltaic environments, and are MBT and CBT values are calibrated to MAAT using
not present in pelagic marine environments the Tierney et al. (2010b) lakes calibration. All num-
(Hopmans et al., 2004), suggesting that they are bers correspond to the structures in Figure 1.
associated with terrestrial organisms. However, in
spite of their ubiquity, efforts to identify the or- al., 1997), but were difficult to analyze because
ganism(s) responsible for these compounds have intact GDGTs are too large for gas chromatogra-
not been highly successful (e.g., Weijers et al., phy. It was not until the relatively recent devel-
2009). Recently, Sinninghe Damsté et al. (2011) opment of a high-performance liquid-
demonstrated that a strain of Acidobacteria pro- chromatography mass-spectrometry method
duces one type of brGDGT (brGDGT-I in Figure (Hopmans et al., 2000) that GDGTs became a fo-
1), confirming suspicions that this phylum of Bac- cus of both organic geochemical and paleoclima-
teria, which is diverse, commonly found in peats tological research. Figure 2 illustrates some typi-
in soils, but not well characterized, may be a cal chromatograms of both iso- and br-GDGTs
source of brGDGTs (Weijers et al., 2006a, 2009). using high-performance liquid chromatography-
However, the other eight brGDGTs utilized in atmospheric pressure chemical ionization-mass
brGDGT-based temperature proxies are, at pre- spectrometry (HPLC-APCI-MS), which separates
sent, “orphan” compounds. compounds by mass and polarity. The run time for
GDGTs—or chemical remnants of them— a single LC/MS analysis is 60–70 minutes, and
have been observed in both modern and ancient preparative procedures are relatively minimal (ex-
sedimentary deposits and oils for some time (e.g., traction followed by column chromatography for
Chappe et al., 1980; Brassell et al., 1981; Hoefs et
117
THE PALEONTOLOGICAL SOCIETY PAPERS, VOL. 18
purification), making GDGT analysis sufficiently tion given above, efforts have been made to ex-
rapid for use in paleoclimate studies. Details of pand the collection of modern core-top sediments
the GDGT LC/MS methodology may be found in available to constrain the TEX86 proxy and better
Schouten et al. (2007b). understand its predictive ability (Kim et al., 2008;
Liu et al., 2009; Kim et al., 2010). The current
HOW DO GDGTS RECORD calibration dataset includes 350+ core tops from
PALEOTEMPERATURES? around the global ocean, though the distribution
of those core tops is not uniform (Figure 3). The
isoGDGTs: the TEX86 paleothermometer relationship between TEX86 and SST remains
In response to temperature variations, the mi- largely linear (Figure 4), although non-linear rela-
crobial producers of GDGTs produce compounds tionships to SST have been proposed (Liu et al.,
with different numbers of rings in their isoprenoid 2009; Kim et al., 2010) based on observations that
chains: a GDGT with more rings has a higher TEX86 is more sensitive to SST changes at the
melting point and is more stable at warm tempera- high end of the temperature range and less sensi-
tures (Gliozzi et al., 1983). Although mesophilic tive at the lower end of the calibration range (e.g.,
Thaumarchaeota have not yet been cultured in Schouten et al., 2003). Both linear and non-linear
different temperature environments, thermophile regressions through the modern core-top dataset
relatives of Thaumarchaeota that are more readily perform comparably (Figure 4).
manipulated in culture produce more cyclic The core-top studies also indicate that there
isoGDGTs at higher temperatures (de Rosa et al., are certain areas in the marine realm where TEX86
1980; Uda et al., 2001). Similarly, mesocosm is unreliable, or necessitates a different calibra-
studies (experiments in which natural seawater is tion. These areas include the polar oceans, where
manipulated in the laboratory) show that more of the relationship between TEX86 and SST is poor
the cyclic isoGDGTs are present in seawater at (Kim et al., 2008, 2010). This could reflect a re-
higher temperatures (Wuchter et al., 2004; duced sensitivity of TEX86 in cold waters, or con-
Schouten et al., 2007a). Schouten et al. (2002) versely, inaccurate mean annual SST estimates at
developed an index to mathematically represent high latitudes. Kim et al. (2010) proposed an al-
this degree of cyclization called TEX86 (the Tet- ternate functional form of TEX86, called “TEXL86”
raEther indeX of 86 carbons; numbers correspond (L for low-temperature) that has better predict-
to compounds labeled in Figure 1; cren′ stands for ability at low temperatures, but is less precise
the crenarchaeol regioisomer): given the entire global SST range (Table 1). Inter-
estingly, TEXL86 excludes the regioisomer:
118
TIERNEY: RECONSTRUCTING PALEOTEMPERATURES USING GDGT THERMOMETRY
!"#$%&'()$*+,-.+/$*0123,0.2+4$/+24.5
o
80 N
40oN
0o
40oS
70,24-:$!;<=>
?0@-5:$!;<<
80oS
180oW 120oW 60oW 0o 60oE 120oE 180oW
!6#$76%896%$*+,-.+/$*0123,0.2+4$/+24.5
o
80 N
40oN
0o
40oS
A+215:$!;>B
?0@-5:$!;CDC
80oS
180oW 120oW 60oW 0o 60oE 120oE 180oW
FIGURE 3.—A) Locations of sediment core-top calibration points for TEX86, including data from Kim et al. (2010),
Ho et al. (2011), Trommer et al. (2009), Chazen (2011), Leider et al. (2010), Seki et al. (2009), Castañeda et al.
(2010), Shintani et al. (2010), Powers et al. (2010), Blaga et al. (2009) and Tierney et al. (2010a) and excluding
duplicates. (B) Locations of sediment core-top calibration points for MBT/CBT, including data from Tierney et al.
(2010b), Blaga et al. (2010), Zink et al. (2010), Sun et al. (2011), and Pearson et al. (2011).
chaea performing methanogenesis or anaerobic rithimic transform of TEX86) for marine SST, and
oxidation of methane (AOM) may produce the Powers et al. (2010) calibration for lake LST.
isoGDGTs (see below), compromising the Thau-
marchaeotal TEX86 signature (Blaga et al., 2009; brGDGTs: the MBT/CBT paleothermometer
Powers et al., 2010). In small lakes receiving Because the producers of most of the
large amounts of soil organic matter, some of the brGDGTs have not been identified, and no species
isoGDGTs may derive from soil Thaumarchaeota, that produce any of these lipids other than
again complicating the sedimentary signal (Blaga brGDGT-I have been identified in culture, the
et al., 2009). As with the marine TEX86 calibra- physiological relationship between brGDGT cyc-
tion, the relationship between TEX86 and LST is lization and/or methylation to temperature is not
generally linear, although a non-linear form is understood. We might expect that such a relation-
also possible. Tierney et al. (2010a) proposed a ship exists, but to date, it relies solely on empiri-
“warm lakes” calibration with a shallower linear cal evidence in the form of core-top/topsoil stud-
slope (see Table 1) to account for the apparent ies. Weijers et al. (2007c) studied a global collec-
increased sensitivity of lacustrine TEX86 at higher tion of topsoils and identified two major environ-
LSTs, indicating that a nonlinear model could be mental parameters that had a statistical relation-
appropriate (Figure 4). ship to brGDGT distributions: soil temperature
Presently, there are a number of calibrations (approximated by mean annual air temperature;
for TEX86, reflecting ongoing attempts to refine MAAT) and soil pH. The degree of cyclization of
the relationship between TEX86 and temperature the brGDGTs seemed to be related only to soil pH
(Table 1). The most common calibrations used to in a non-linear fashion, implying a power law re-
date are the Kim et al. (2010) “TEXH86” calibra- lationship. Thus, Weijers et al. (2007c) devised an
tion (where TEXH86 is simply the base-10 loga- index representing the cyclization of the
119
THE PALEONTOLOGICAL SOCIETY PAPERS, VOL. 18
TABLE 1.—Sediment core top TEX86 calibrations. *= calibration applies just to the Red Sea, where TEX86 distribu-
tions are different from the open ocean, ‡=TEXL86 is an alternate functional form of TEX86 arguably better for pre-
diction at low SSTs, see main text and Kim et al. (2010) for more details, †=lake calibrations.
5–30 T = −10.78 + 56.2 × T E X86 223 0.94 1.7 Kim et al. (2008)
-3–30 T = 50.475 − 16.332 × (1/T E X86) 287 0.82 3.7 Liu et al. (2009)
-3–30‡ T = 49.9 + 67.5 × T E X L86 396 0.86 4.0 Kim et al. (2010)
5–30 T = 38.6 + 68.4 × log(T E X86) 255 0.86 2.5 Kim et al. (2010)
brGDGTs called the CBT (Cyclization of and CBT (as a surrogate for the pH influence) to
Branched Tetraethers) index: predict MBT, which could be inverted to solve for
MAAT:
120
TIERNEY: RECONSTRUCTING PALEOTEMPERATURES USING GDGT THERMOMETRY
#! #!
)% )%
;3<2415/6372*82092/385/2
123415/6372*82092/385/2
)! )!
(% (%
(! (!
!*+*#%,-*")*+*!".$-*/012*+*)". !*+*##-*")*+*!".!-*/012*+*#"'
% %
!*+*#%,-*" *+*!".%-*/012*+*)".
)
!*+*##-*")*+*!"'.-*/012*+*#",
!"# !"$ !"% !"& !"' !") !"# !"$ !"% !"& !"' !".
82:.& 82:.&
FIGURE 4.—TEX86 regressions to SST and LST respectively, with best-fit linear and non-linear models. The marine
TEX86 data shown exclude data from the Red Sea (Trommer et al., 2009) and the polar oceans (SST<2, Kim et al.,
2010) where the relationship between TEX86 and SST is different and/or poor. RMSE = root mean square error.
2009; Zink et al., 2010; Tyler et al., 2010); thus, EXAMPLE APPLICATIONS OF GDGT
the soil calibration was not applicable. This may PROXIES
suggest that lacustrine brGDGTs are produced in
situ with a different sensitivity to lake tempera- GDGTs have been used to reconstruct both air
ture—a hypothesis also supported by the presence and water temperatures on a variety of geologic
of intact polar brGDGT in lakes (Tierney et al., timescales, but have been especially useful for
2012). However, recent attempts to develop lake- estimating temperatures in the Mesozoic and early
specific brGDGT distributions are promising and Cenozoic, where there are few applicable methods
demonstrate that lake sediments perform similar for estimating paleotemperature. In one of the
to or better than soils in predicting MAAT (Figure earliest applications of the TEX86 proxy, Schouten
5). et al. (2003) reconstructed low-latitude SSTs dur-
For both soils and lakes, the MBT/CBT tem- ing select intervals in the middle Cretaceous (Al-
perature proxy is considerably less accurate in bian–Cenomanian–Turonian), a hothouse period
predicting temperature than TEX86, with a root in Earth history where temperatures are relatively
mean square error near 5°C. Alternate functional poorly constrained. They estimated low-latitude
forms representing the distribution of brGDGTs SSTs of 32–36°C, in general agreement with
may improve predictive performance and are be- δ18O-based estimates from well-preserved Ceno-
ginning to be explored. Tierney et al. (2010b) manian foraminifera (Schouten et al., 2003).
proposed a calibration based on the fractional GDGTs also have been used to provide both
abundance (relative to the sum of all nine low- and high-latitude constraints on temperature
GDGTs) of the three major non-cyclic brGDGTs excursions during the Paleocene–Eocene Thermal
that provides more precise prediction of MAAT in Maximum (PETM), a brief interval when massive
tropical East African lakes (Figure 5, Table 2). amounts of greenhouse gases were injected into
Similarly, Pearson et al. (2011) proposed a cali- the atmosphere ~55 million years ago, causing
bration based on the fractional abundances of rapid and extreme warming. Zachos et al. (2006)
brGDGT-III, brGDGT-II, and brGDGT-Ib for a applied TEX86 to a section from Wilson Lake,
more globally distributed set of lakes (Table 2). A New Jersey (paleolatitude of 36°N), and found
list of brGDGT-based calibrations is provided in that SSTs were ~25–30°C prior to and after the
Table 2. PETM, but reached a maximum of ~35°C during
the event (Figure 6). Sluijs et al. (2006) and
Weijers et al. (2007b) applied TEX86 and the
121
THE PALEONTOLOGICAL SOCIETY PAPERS, VOL. 18
:1./,;/6&2/-5/,829,/
:1./,;/6&2/-5/,829,/
$! $! $!
#! #! #!
! ! !
FIGURE 5.—brGDGT regressions to MAAT for (A) soils (data from Weijers et al., 2007c), (B) lakes (data from
Tierney et al., 2010b; Blaga et al., 2010; Zink et al., 2010; Sun et al., 2011), and (C) East African lakes, using an
alternative functional form (Tierney et al., 2010b, see Table 2 for the equation).
MBT/CBT proxy, respectively, to the PETM sec- anthropogenic climate change has had a strong
tion of a drill core taken from the Lomonosov impact on the thermal structure of Lake Tangany-
Ridge (paleolatitude of 85°N), and reconstructed ika.
surprisingly warm temperatures for the high Arc- BrGDGTs have not been used as widely as
tic (Figure 6). Taken at face value, these PETM TEX86 for paleoclimate reconstruction because
GDGT data suggest an extremely warm world their dependence on temperature was only de-
with a surprisingly minimal latitudinal gradient in scribed in 2007 and, to date, is not well con-
temperature (~10°C) compared to present-day strained. In addition to the PETM study, Weijers
gradients (20°C). However, it should be noted that et al. (2007a) used MBT/CBT to reconstruct
because the TEX86 values at Wilson Lake exceed MAAT in the Congo Basin, using deltaic sedi-
those observed in the modern ocean, the inferred ments from the Congo Fan. Their data indicate a
temperatures rely on an extrapolation of the cali- LGM–present change of ~3.5–4°C, a reasonable
bration regression lines; thus, the absolute values finding for the deep tropics, and in agreement
are highly uncertain. Nevertheless, such data pro- with independent proxy data from Africa. Fawcett
vide a hypothesis of how Earth systems responded et al. (2011) used MBT/CBT to reconstruct tem-
to what were likely very high levels of atmos- perature within a Pleistocene lacustrine deposit
pheric carbon dioxide during the PETM. from the Valles Caldera, and estimated glacial/
GDGT proxies also have utility on more re- interglacial thermal variability on the order of
cent timescales, either as a complement to other ~8°C, which is in general agreement with pollen-
paleotemperature proxies or (as is sometimes the based estimates. Neither brGDGTs nor isoGDGTs
case with deep-time studies) because it is the only have been used, as of yet, to reconstruct tempera-
proxy method available in a given region or depo- tures in ancient lake basins.
sitional setting. For example, Tierney et al.
(2010a) used TEX86 in Lake Tanganyika to recon- LIMITATIONS ON THE USE OF GDGT
struct temperatures for the past 1500 years in East PALEOTHERMOMETERS
Africa, a region where there are few options for
recent temperature reconstruction (there are no Preservation
developed tree-ring chronologies, and stable iso- GDGTs are relatively refractory compounds,
topes generally reflect hydroclimatic variability). but they do degrade over time, with isoGDGTs
TEX86 shows evidence for warm conditions dur- apparently degrading faster than brGDGTs (Hu-
ing the Medieval Warm Period, followed by cool/ guet et al., 2008). However, diagenetic degrada-
variable temperatures during the Little Ice Age tion does not affect the TEX86 index within ana-
(Figure 7). In the last 150 years, TEX86 indicates a lytical error (Sinninghe Damsté et al., 2002a;
warming of ~2°C, an event unprecedented in the Schouten et al., 2004; Kim et al., 2009; Huguet et
last 1500 years (Figure 7). These data suggest that al., 2009). When sedimentary sequences undergo
122
TIERNEY: RECONSTRUCTING PALEOTEMPERATURES USING GDGT THERMOMETRY
TABLE 2.—Topsoil (Weijers et al., 2007c) or lacustrine (all others) core top brGDGT calibrations. *=pH calibration
from CBT, †=uses African lakes data only, ‡=uses New Zealand lakes only, §=This calibration predicts mean sum-
mer temperature (MST) not MAAT.
0–26† MAAT = 11.84 + 32.54 × MBT − 9.32 × CBT 39 0.89 3.0 Tierney et al. (2010b)
0–26† MAAT = 50.47 − 74.18 × fbrGDGT III − 39 0.94 2.2 Tierney et al. (2010b)
31.60 × f brGDGT II − 34.69 × f brGDGT I
6–15‡ MAAT = 55.01 × MBT − 6.055 10 0.74 N/P Zink et al. (2010)
-5–30 MAAT = 3.949 − 5.593 × CBT + 38.213 × MBT 100 0.73 4.3 Sun et al. (2011)
5–30§ MST = 20.9 − 20.5 × f brGDGT III − 12 × 85 0.88 2.0 Pearson et al. (2011)
fbrGDGT II + 98.1 × f brGDGT Ib
catagenesis, GDGTs break down more rapidly and (2010) observed that anoxic lakes yield a “colder”
do so preferentially: hydrous pyrolysis experi- MBT/CBT signature than oxic lakes, indicating
ments suggest that they degrade between 240– more methylated brGDGTs in the sediments.
300°C, and that TEX86 values decline, indicating These observations may suggest preferential dia-
preferential destruction of the more cyclic com- genetic loss of Type III brGDGTs upon exposure
pounds (Schouten et al., 2004). This experimental to oxygen, but alternatively, could reflect a sensi-
evidence suggests that TEX86 is generally not ap- tivity of the microbial producers to anoxia
plicable in more mature sediments (i.e., those (Tierney et al., 2012). Further research is needed
with high amounts of 22S hopanes; 17α,21β (H)- to assess the relative diagenetic stability of the
hopane 22S/(22S+22R) ratios > 0.1). Thus, while nine brGDGTs.
isoGDGT likely have an ancient biosynthetic ori-
gin (Thaumarchaeota have been extant for billions Ecological constraints
of years, Spang et al., 2010, although this may or As mentioned previously, the TEX86 proxy is
may not apply to the lipids), and theoretically predicated on members of the phylum Thaumar-
TEX86 could be used throughout much of geo- chaeota acting as the primary producers of
logic history, the application of the proxy is lim- isoGDGTs. This is a reasonable assumption in the
ited to rock and sediment sequences that have global ocean where Thaumarchaeota dominate,
been minimally thermally altered. although the contribution of isoGDGTs from
Diagenetic/catagenic effects on the brGDGT mesophilic Euryarchaeota, also abundant in ma-
distributions (MBT/CBT) are not as well con- rine environments, remains a matter of some un-
strained due to the novelty of the proxy. In a study certainty and debate (Turich et al., 2007; Schouten
of brGDGTs in a suburban lake that became sea- et al., 2008b). Even if the source of isoGDGTs in
sonally anoxic and eutrophic as a result of human the modern ocean can be reasonably constrained
disturbance, Tierney et al. (2012) found that there to Thaumarchaeota, it is unlikely that isoGDGTs
were more “Type III” (see Figure 1) brGDGTs in are predominantly produced by a single species
lake sediments that were recently deposited in the (unlike alkenones that select marine species pro-
anoxic environment, and less in deeper sediments duce). The lack of regioisomer in cultures of a
that were deposited in the pre-anthropogenic oxy- dominant marine archaeon, Nitrosopumilus mari-
genated environment. Similarly, Bechtel et al. timus (Schouten et al., 2008a), and the observa-
123
THE PALEONTOLOGICAL SOCIETY PAPERS, VOL. 18
124
TIERNEY: RECONSTRUCTING PALEOTEMPERATURES USING GDGT THERMOMETRY
125
THE PALEONTOLOGICAL SOCIETY PAPERS, VOL. 18
A potentially severe complicating factor for straining absolute temperature error is less impor-
application of brGDGT paleothermometers in tant than relative changes, error can be estimated
lakes is the likelihood that brGDGTs are associ- on the calibration slope (e.g., Tierney et al.,
ated with both allochthonous soil OM and in-situ 2010a). In sum, the RMSE values shown in Ta-
production, each of which appear to have a differ- bles 1 and 2 should not be viewed as immutable
ent relationship to temperature (Tierney and Rus- estimates of the error of GDGT-based temperature
sell, 2009). However, brGDGT concentrations inferences. As with any proxy, error needs to be
(per gram of organic matter) tend to be much rigorously assessed within the geological context
higher in lakes than in soils—sometimes by as at hand, and in light of the climatological hy-
much as an order of magnitude (Tierney and Rus- potheses being investigated.
sell, 2009; Tierney et al., 2010b)—and the newest
lake calibrations based on alternate functional CONCLUSIONS
forms (i.e., those that are not based on the MBT
or CBT indices) have remarkably good predict- Although new to the paleotemperature proxy
ability (Tierney et al., 2010b; Pearson et al., 2011, scene, GDGTs significantly expand the toolkit
and see Table 2). Detailed down-core studies will available to geologists for reconstructing past
help address these issues. temperatures. Since GDGTs are exceptionally
ubiquitous compounds found in marine sedi-
A note on calibration uncertainty ments, ancient sedimentary sequences, lakes,
Both TEX86 and brGDGT paleothermometers soils, hot springs and even speleothems (Yang et
rely on surface-sediment calibrations to convert al., 2011), they offer opportunities to reconstruct
relative abundances to temperature. Global sur- temperatures in geographic places and geological
face calibrations have an advantage in that they time intervals where other proxies may not be
cover a wide diversity of natural environments available. Many uncertainties, especially ecologi-
and thus incorporate a large range of sources of cal uncertainties, need to be better constrained so
variability leading to error. However, the implicit that GDGT proxies can be applied with more con-
assumption of this approach is that the Earth sys- fidence. Nevertheless, as the data in Figure 6 il-
tem is ergodic, having the same type of variability lustrate, GDGTs have already fundamentally in-
in modern geographical space as it does in geo- fluenced our thinking about climate sensitivity
logical time. Depending on the paleoenvironment, during ancient hothouse worlds, highlighting their
this assumption may or may not always be valid. utility as paleothermometers.
For example, the TEX86 temperatures estimated
for the Cretaceous and for the mid to low latitudes REFERENCES
during the PETM (discussed above) rely on an
extrapolation of the modern calibration equations BECHTEL, A., R. SMITTENBERG, S. BERNASCONI, AND
to temperatures higher than observed in the mod- C. SCHUBERT. 2010. Distribution of branched and
ern oceans. They are also contingent on assuming isoprenoid tetraether lipids in an oligotrophic and
that Thaumarchaeota living in (e.g.) the Creta- a eutrophic Swiss lake: Insights into sources and
GDGT-based proxies. Organic Geochemistry,
ceous oceans experienced a similar sensitivity to
41:822–832.
temperature and non-temperature effects as in the BLAGA, C., G. REICHART, S. SCHOUTEN, A. LOTTER, J.
modern oceans. As the latter is very hard to con- WERNE, S. KOSTEN, N. MAZZEO, N., G. LACEROT,
strain with confidence, the error associated with AND J. S. SINNINGHE DAMSTÉ. 2010. Branched
the absolute temperature estimates in ancient hot- glycerol dialkyl glycerol tetraethers in lake sedi-
house environments is almost certainly quite a bit ments: Can they be used as temperature and pH
larger than the root mean square error (RMSE) of proxies? Organic Geochemistry, 41:1225–1234.
the modern core-top calibrations. Conversely, a BLAGA, C. I., G.-J. REICHART, O. HEIRI, AND J. S. SIN-
Late Quaternary TEX86 application in a single NINGHE DAMSTÉ. 2009. Tetraether membrane lipid
tropical locale is unlikely to incorporate sources distributions in water-column particulate matter
of error associated with the polar oceans (and vice and sediments: a study of 47 European lakes along
a north-south transect. Journal of Paleolimnology.
versa), in which case the global core-top calibra-
41 (3), 523–540.
tion error may be an overestimate. In this situa- BLUMENBERG, M., R. SEIFERT, J. REITNER, T. PAPE,
tion, local calibrations can be devised to better AND W. MICHAELIS. 2004. Membrane lipid pat-
constrain sources of error relevant to the region of terns typify distinct anaerobic methanotrophic
interest (e.g., Shevenell et al., 2011) or, if con-
126
TIERNEY: RECONSTRUCTING PALEOTEMPERATURES USING GDGT THERMOMETRY
consortia. Proceedings of the National Academies coastal NW Mediterranean Sea (Blanes Bay Mi-
o f S c i e n c e s U S A , 1 0 11 : 11111 – 1111 6 . crobial Observatory). Limnology and Oceanogra-
doi:10.1073/pnas.0401188101 phy, 55:2117–2125.
BRASSELL, S., A. WARDROPER, I. THOMSON,J. MAX- GLIOZZI, A., G. PAOLI, M. DE ROSA, AND A. GAMBA-
WELL, AND G. EGLINTON. 1981. Specific acyclic CORTA. 1983. Effect of isoprenoid cyclization on
isoprenoids as biological markers of methanogenic the transition temperature of lipids in thermophilic
bacteria in marine sediments. Nature, 290:693. archaebacteria. Biochimica et Biophysica Acta
BROCHIER-ARMANET, C., B. BOUSSAU, S. GRIBALDO, (BBA)—Biomembranes, 735:234–242.
AND P. FORTERRE. 2008. Mesophilic Crenar- HERBERT, T. D., 2003. Alkenone paleotemperature de-
chaeota: proposal for a third archaeal phylum, the terminations. Treatise on Geochemistry, 6:391–
Thaumarchaeota. Nature Reviews Microbiology, 432.
6:245–252. HERNDL, G., T. REINTHALER, E. TEIRA, H. VAN AKEN,
CASTAÑEDA, I., E. SCHEFUß, J. PÄTZOLD, J. S. SIN- C. VETH, A. PERNTHALER,AND J. PERNTHALER.
NINGHE DAMSTÉ, S. WELDEAB, AND S. SCHOUTEN. 2005. Contribution of Archaea to total prokaryotic
2010. Millennial-scale sea surface temperature production in the deep Atlantic Ocean. Applied
changes in the eastern Mediterranean (Nile River Environmental Microbiology, 71:2303–2309.
Delta region) over the last 27,000 years. Paleo- HO, S., M. YAMAMOTO, G. MOLLENHAUER, AND M.
ceanography, 25:PA1208. MINAGAWA. 2011. Core top TEX86 values in the
CHAPPE, B., W. MICHAELIS, AND P. ALBRECHT. 1980. south and equatorial Pacific. Organic Geochemis-
Molecular fossils of archaebacteria as selective try, 42: 94–99.
degradation products of kerogen. Physics and HOEFS, M., S. SCHOUTEN, J. DE LEEUW, L. KING, S.
Chemistry of the Earth, 12:265–274. WAKEHAM, AND J. S. SINNINGHE DAMSTÉ. 1997.
CHAZEN, C., 2011. Holocene climate evolution of the Ether lipids of planktonic archaea in the marine
eastern tropical Pacific told from high resolution water column. Applied Environmental Microbiol-
climate records from the Peru margin and equato- ogy, 63:3090–3095.
rial upwelling regions. Ph.D. thesis, Brown Uni- HOLLIBAUGH, J., S. GIFFORD, S. SHARMA, N. BANO,
versity. AND M. MORAN. 2010. Metatranscriptomic analy-
CHURCH, M., B. WAI, D. KARL, AND E. DELONG. 2010. sis of ammonia-oxidizing organisms in an estuar-
Abundances of crenarchaeal amoA genes and tran- ine bacterioplankton assemblage. The ISME Jour-
scripts in the Pacific Ocean. Environmental Mi- nal, 5:866–878.
crobiology, 12:679–688. HOPMANS, E. C., S. SCHOUTEN, R. D. PANCOST, M. T. J.
DE LA TORRE, J., C. WALKER, A. INGALLS, M. KÖN- VAN DER MEER, AND J. S. SINNINGHE DAMSTÉ.
NEKE, AND D. STAHL. 2008. Cultivation of a ther- 2000. Analysis of intact tetraether lipids in ar-
mophilic ammonia oxidizing archaeon synthesiz- chaeal cell material and sediments by high per-
ing crenarchaeol. Environmental Microbiology, formance liquid chromatography/atmospheric
10:810–818. pressure chemical ionization mass spectrometry.
DE ROSA, M., E. ESPOSITO, A. GAMBACORTA, B. NICO- Rapid Communications in Mass Spectrometry
LAUS, AND J. BU’LOCK. 1980. Effects of tempera- 14:585–589.
ture on ether lipid composition of Caldariella aci- HOPMANS, E. C., J. W. H. WEIJERS, E. SCHEFUß, L.
dophila. Phytochemistry, 19:827–831. HERFORT, J. S. SINNINGHE DAMSTÉ, AND S.
E DER , W., M. S CHMIDT , M. K OCH , D. G ARBE - SCHOUTEN. 2004. A novel proxy for terrestrial
SCHÖNBERG, AND R. HUBER. 2002. Prokaryotic organic matter in sediments based on branched
phylogenetic diversity and corresponding geo- and isoprenoid tetraether lipids. Earth and Plane-
chemical data of the brine–seawater interface of tary Sciences Letters, 224:107–116.
the Shaban Deep, Red Sea. Environmental Micro- HUBER, M., 2010. Why improving molecular and iso-
biology, 4: 758–763. topic proxy paleoclimate records is the most im-
FAWCETT, P., J., J. P. WERNE, R. S. ANDERSON, J. M. portant problem in science today: A climate mod-
HEIKOOP, E. T. BROWN, M. A. BERKE, S. J. SMITH, eling perspective. In Gordon Research Confer-
F. GOFF, L. DONOHOO-HURLEY, L. M. CISNEROS- ence. Holderness, NH, USA.
DOZAL, S. SCHOUTEN, J. S. SINNINGHE DAMSTÉ, Y. HUGUET, C., G. J. DE LANGE, O. GUSTAFSSON, J. J.
HUANG, J. TONEY, J. FESSENDEN, G. WOLDEGAB- MIDDELBURG, J. S. SINNINGHE DAMSTÉ, AND S.
RIEL, V. ATUDOREI, J. W. GEISSMAN, AND C. D. SCHOUTEN. 2008. Selective preservation of soil
ALLEN. 2011. Extended megadroughts in the organic matter in oxidized marine sediments (Ma-
southwestern United States during Pleistocene deira Abyssal Plain). Geochimica et Cosmo-
interglacials. Nature, 470:518–521. chimica Acta, 72: 6061–6068.
GALAND, P., C. GUTIÉRREZ-PROVECHO, R. MASSANA, HUGUET, C., J. KIM, G. DE LANGE, J. S. SINNINGHE
J. GASOL, AND E. CASAMAYORA. 2010. Inter- DAMSTÉ, AND S. SCHOUTEN. 2009. Effects of long
annual recurrence of archaeal assemblages in the term oxic degradation on the TEX86 and BIT or-
127
THE PALEONTOLOGICAL SOCIETY PAPERS, VOL. 18
ganic proxies. Organic Geochemistry, 40:1188– VERSTEEGH. 2010. Core-top calibration of the
1194. lipid-based UK0 37 and TEX86 temperature prox-
HUGUET, C., A. SCHIMMELMANN, R. THUNELL, L. ies on the southern Italian shelf (SW Adriatic Sea,
LOURENS, J. S. SINNINGHE DAMSTÉ, AND S. Gulf of Taranto). Earth and Planetary Sciences
SCHOUTEN. 2007. A study of the TEX86 paleo- Letters, 300:112–124.
thermometer in the water column and sediments of LEININGER, S., T. URICH, M. SCHLOTER, L. SCHWARK,
the Santa Barbara Basin, California. Paleoceanog- J. QI, G. NICOL, J. PROSSER, S. SCHUSTER, AND C.
raphy, 22:PA3203. doi:10.1029/2006PA001310 SCHLEPER. 2006. Archaea predominate among
INGALLS, A., S. SHAH, R. HANSMAN, L. ALUWIHARE, ammonia-oxidizing prokaryotes in soils. Nature,
G. SANTOS, E. DRUFFEL, AND A. PEARSON. 2006. 442:806–809.
Quantifying archaeal community autotrophy in the LIU, X., A. LEIDER, A. GILLESPIE, J. GRÖGER, G. VER-
mesopelagic ocean using natural radiocarbon. Pro- STEEGH, AND K. HINRICHS. 2010. Identification of
ceedings of the National Academies of Sciences polar lipid precursors of the ubiquitous branched
USA 103:6442–6447. GDGT orphan lipids in a peat bog in Northern
IONESCU, D., S. PENNO, M. HAIMOVICH, B. RIHT- Germany. Organic Geochemistry, 41:653–660.
MAN,A. GOODWIN,D. SCHWARTZ, L. HAZANOV, LIU, Z., M. PAGANI, D. ZINNIKER,R. DECONTO, M.
M. CHERNIHOVSKY, A. POST, AND A. OREN. 2009. HUBER, H. BRINKHUIS, S. SHAH, R. LECKIE, AND
Archaea in the Gulf of Aqaba. FEMS Microbiol- A. PEARSON. 2009. Global cooling during the
ogy Ecology, 69: 425–438. Eocene-Oligocene climate transition. Science,
KARNER, M., E. DELONG, AND D. KARL. 2001. Ar- 323:1187–1190.
chaeal dominance in the mesopelagic zone of the LLIRÓS, M., F. GICH, A. PLASENCIA, J. AUGUET, F.
Pacific Ocean. Nature, 409:507–510. DARCHAMBEAU, E. CASAMAYOR, J. DESCY, AND C.
KIM, J., J. VAN DER MEER, S. SCHOUTEN, P. HELMKE, BORREGO. 010. Vertical distribution of ammonia-
V. WILLMOTT, F. SANGIORGI, N. KOC, E. HOP- oxidizing crenarchaeota and methanogens in the
MANS, J. S. SINNINGHE DAMSTÉ. 2010. New indi- epipelagic waters of Lake Kivu (Rwanda-
ces and calibrations derived from the distribution Democratic Republic of the Congo). Applied En-
of crenarchaeal isoprenoid tetraether lipids: Impli- vironmental Microbiology, 76:853–6863.
cations for past sea surface temperature recon- LOPES DOS SANTOS, R., M. PRANGE, I. CASTAÑEDA,E.
structions. Geochimica et Cosmochimica Acta, SCHEFUß, S. MULITZA, M. SCHULZ, E. NIEDER-
74:4639–4654. MEYER, J. SINNINGHE DAMSTÉ, AND S. SCHOUTEN.
KIM, J.-H., C. HUGUET, K. A. F. ZONNEVELD, G. J. M. 2010. Glacial-interglacial variability in Atlantic
VERSTEEGH, W. ROEDER, J. S. SINNINGHE DAM- meridional overturning circulation and thermo-
STÉ, AND S. SCHOUTEN. 2009. An experimental cline adjustments in the tropical North Atlantic.
field study to test the stability of lipids used for the Earth and Planetary Sciences Letters, 300:407–
TEX86 and UK0 37 palaeothermometers. Geo- 414.
chimica et Cosmochimica Acta, 73:2888–2898. MARLOWE, I., J. GREEN, A. NEAL,S. BRASSELL, G. EG-
KIM, J.-H., S. SCHOUTEN, E. C. HOPMANS, B. DONNER, LINTON, AND P. COURSE. 1984. Long chain (n-
AND J. S. SINNINGHE DAMSTÉ. 2008. Global sedi- C37–C39) alkenones in the Prymnesiophyceae.
ment core-top calibration of the TEX86 paleother- Distribution of alkenones and other lipids and their
mometer in the ocean. Geochimica et Cosmo- taxonomic significance. British Phycological
chimica Acta72 (4), 1154–1173. Journal, 19:203–216.
KITA, M., S. AIBARA, M. KATO, M. ISHINAGA, AND T. MASSANA, R., A. MURRAY, C. PRESTON, AND E. DE-
HATA. 1973. Effect of changes in fatty acid com- LONG. 1997. Vertical distribution and phylogenetic
position of phospholipid species on the b -galacto- characterization of marine planktonic Archaea in
side transport system of Escherichia coli K-12. the Santa Barbara Channel. Applied Environmen-
Biochimica et Biophysica Acta (BBA)- tal Microbiology63 (1), 50–56.
Biomembranes 298:69–74. NOZAWA, Y., H. IIDA, H. FUKUSHIMA, K. OHKI, AND S.
KÖNNEKE, M., A. E. BERNHARD, J. R. DE LA TORRE, C. OHNISHI. 1974. Studies on Tetrahymena mem-
B. WALKER, J. B. WATERBURY, AND D. A. STAHL. branes: Temperature-induced alterations in fatty
2005. Isolation of an autotrophic ammonia- acid composition of various membrane fractions in
oxidizing marine archaeon. Nature, 437:543–546. Tetrahymena pyriformis and its effect on mem-
LEE, K., J. KIM, I. WILKE, P. HELMKE, AND S. brane fluidity as inferred by spin-label study. Bio-
SCHOUTEN. 2008. A study of the alkenone, TEX86, chimica et Biophysica Acta (BBA)-
and planktonic foraminifera in the Benguela Up- Biomembranes, 367:134–147.
welling System: Implications for past sea surface OCHSENREITER, T., D. SELEZI, A. QUAISER, L. BONCH-
temperature estimates. Geochemistry Geophysics OSMOLOVSKAYA, AND C. SCHLEPER. 2003. Diver-
Geosystems, 9:Q10019. sity and abundance of Crenarchaeota in terrestrial
LEIDER, A., K. HINRICHS, G. MOLLENHAUER, AND G. habitats studied by 16S RNA surveys and real time
128
TIERNEY: RECONSTRUCTING PALEOTEMPERATURES USING GDGT THERMOMETRY
129
THE PALEONTOLOGICAL SOCIETY PAPERS, VOL. 18
130
TIERNEY: RECONSTRUCTING PALEOTEMPERATURES USING GDGT THERMOMETRY
131