Introduction
Long-chain n-alkanes and fatty acids are important components of the
epicuticular leaf waxes of terrestrial plants (Eglinton and Hamilton, 1967; Samuels et
al., 2008). As leaf waxes can be preserved in sedimentary archives over a
long time, they serve as valuable biomarkers for paleoenvironmental and
paleoclimate reconstructions (Eglinton and Eglinton, 2008; Zech M. et al.,
2011b). The δ2H isotopic composition of leaf waxes is of
particular interest in this regard because, at least to the first order, it
reflects the isotopic composition of precipitation δ2Hprec (Sauer et al., 2001; Huang et al., 2004; Sachse et al., 2004; Schefuss et
al., 2005; Pagani et al., 2006; Tierney et al., 2008; Rao et al., 2009),
which in turn depends on temperature, the amount of precipitation, atmospheric
circulation, etc. (Dansgaard, 1964; Rozanski et al., 1993; Gat, 1996;
Araguas-Araguas et al., 2000). While there is probably no fractionation of
hydrogen isotopes during water uptake by roots (Ehleringer and Dawson,
1992), several studies have shown that leaf water is enriched in 2H
compared to source water or precipitation
(Flanagan et al., 1991;
Yakir, 1992; Sachse et al., 2006; Smith and Freeman, 2006; Farquhar et al.,
2007; Feakins and Sessions, 2010). This 2H enrichment, which is also
recorded in leaf waxes (Kahmen et al., 2013a, b), can be explained through
evapotranspiration and is mainly controlled by relative air humidity (RH),
temperature and the isotopic composition of atmospheric water vapor. Indeed,
creating a robust reconstruction of δ2Hprec from soils and
sedimentary records increasingly has increasingly turned out to be quite challenging, because
it is hitherto difficult to disentangle past changes in δ2Hprec and changes in evapotranspirative enrichment of leaf water
(Zech R. et al., 2013; Zech M. et al., 2015).
Compared to compound-specific δ2H analyses, compound-specific
δ18O analyses are by far less adopted by the scientific
community as of yet
(Hener et
al., 1998; Juchelka et al., 1998; Jung et al., 2005, 2007; Greule et
al., 2008). However, compound-specific δ18O
analyses of hemicellulose-derived sugar biomarkers (δ18Osugars) extracted from plants, soils and sediments, in particular,
have been proposed to have large potential, especially in
paleoclimate/paleohydrological
research (Zech M. and Glaser, 2009; Zech M. et al., 2012b). Similar to leaf
waxes, hemicellulose-derived sugars record the isotopic composition of water
used for metabolism, i.e., the isotopic composition of precipitation altered
by evapotranspirative 18O enrichment of soil and leaf water (Tuthorn et
al., 2014; Zech M. et al., 2014a). Recently, Zech M. et al. (2013) proposed
a conceptual coupled δ2Hn-alkane-δ18Osugar model for paleoclimate research and suggested that this
coupling allows for overcoming the above defined limitation of single δ2Hn-alkane approaches. Accordingly, the coupled δ2Hn-alkane–δ18Osugar approach
allows for reconstructing (i) δ2H/δ18Oleaf water
values, (ii) deuterium excess (d-excess) of leaf water, which mainly
reflects evapotranspirative enrichment and can be used to reconstruct
relative air humidity (RH) and (iii) δ2H/δ18Oprec values.
The study presented here aimed at evaluating the coupled δ2Hn-alkane–δ18Osugar biomarker approach by
applying it to a modern topsoil climate transect from Argentina. More
specifically, we aimed at (i) analyzing and comparing the δ2H
values of n-alkanes and fatty acids, (ii) modeling 2H leaf water
enrichment along the transect and comparing of δ2Hleaf water values with δ2Hn-alkane and δ2Hfatty acid values, (iii) reconstructing d-excess of leaf
water using the coupled δ2Hn-alkane-δ18Osugar approach and evaluating the potential for reconstructing
RH and (iv) reconstructing “biomarker-based” δ2H/δ18Oprec values and comparing them with actual δ2H/δ18Oprec values.
Materials and methods
Transect description and samples
The investigated transect in Argentina spans ∼ 32 to 47∘ S, and encompasses 20 sampling locations
spanning a large climate and altitudinal (22–964 m) gradient (Fig. 1).
Mean annual temperature ranges from 11.4 to 18.0 ∘C
and mean annual precipitation from 185 to 1100 mm (GeoINTA, 2012).
Precipitation shows a systematic southward trend towards more negative
δ18O and δ2H values (δ18Oprec
and δ2Hprec, respectively; Bowen, 2012).
Sampling locations along the investigated transect in
Argentina. The colors illustrate the gradient in δ2Hprec,
and mean annual temperature and precipitation are shown below.
The transect is described in detail by Tuthorn et al. (2014) and Ruppenthal
et al. (2015). Briefly, it is characterized by warm humid subtropical
conditions in the north (Zárate, Buenos Aires Province), pronounced arid
conditions in the middle part of the transect and cool temperate conditions
in the south (Las Heras, Santa Cruz Province). These markedly contrasting
climate conditions are reflected in the vegetation zones of the study area,
changing from humid/dry Pampas (with dominance of Triticum, Setaria, Eragrostis, Andropogon, Panicum and Festuca species) in the north
to the Espinal vegetation zone (with dominance of Festuca and Larrea species) which prevails
under semi-arid climate (Burgos and Vidal, 1951), Low Monte
semidesert/desert (with dominance of Larrea species) in the most arid region of
Argentina (Fernández and Busso, 1997) and Patagonian Steppe (with
dominance of Stipa species) in the southernmost part of the transect (Le
Houérou, 1996; Paruelo et al., 1998).
During a field campaign in March and April 2010, mixed topsoil samples
(Ah-horizons) from maximum 51 cm depth were collected in triplicate
replication from the 20 sample sites along the transect (for soil type and
total organic carbon contents please see Table 1 of Tuthorn et al., 2014).
The soil samples were air-dried in the field and later in an oven at
50 ∘C for several days. The sampling site heterogeneity was
checked for the δ18Osugar analyses and in most cases did
not exceed the analytical uncertainty (Table 2 in Tuthorn et al., 2014).
Therefore, the field replications were merged into one composite sample per
study site for the δ2Hlipid analyses.
δ2H values of individual leaf wax n-alkanes and fatty acids.
Measurements were carried out in at least triplicate (SD = standard deviation).
δ2Hn-alkanes
δ2Hfatty acids
sampling
C29
C31
C22
C24
C26
C28
C30
locality
mean (‰)
SD
mean (‰)
SD
mean (‰)
SD
mean (‰)
SD
mean (‰)
SD
mean (‰)
SD
mean (‰)
SD
1
-157
2
-164
2
-155
1
-157
1
-151
1
-153
1
-153
2
2
-166
0
-166
1
-150
0
-155
1
-165
1
-163
1
-161
3
3
-175
1
-179
1
-162
0
-161
1
-165
1
-159
1
-155
0
4
-176
1
-176
1
-162
2
-163
1
-166
1
-165
1
-158
2
5
-178
1
-180
2
-164
0
-165
1
-168
2
-162
1
-159
1
6
-171
2
-172
0
-166
0
-165
2
-169
1
-161
1
-158
1
7
-179
0
-182
0
-170
0
-172
1
-177
0
-169
1
-157
0
8
-162
1
-167
1
-161
1
-161
1
-166
1
-161
1
-158
2
9
-173
1
-168
1
-163
1
-164
0
-168
1
-169
0
-156
1
10
-173
2
-170
2
-159
1
-167
1
-168
0
-159
1
-137
2
11
-170
2
-156
2
-158
0
-169
0
-167
2
-153
4
-147
4
12
-155
1
-176
0
-158
1
-168
1
-172
1
-148
1
-133
1
13
-157
2
-161
1
-158
1
-153
0
-140
1
-135
1
-128
1
14
-158
1
-166
0
-168
1
-183
0
-181
2
-160
2
-147
1
15
-194
2
-193
1
-194
0
-197
0
-191
1
-176
2
-168
2
16
-203
1
-211
1
-204
1
-198
0
-201
0
-193
0
-189
1
17
-218
1
-217
1
-219
1
-220
1
-217
0
-205
1
-204
1
18
-213
1
-202
4
-211
0
-203
1
-204
0
-196
0
-194
0
19
-222
1
-222
1
-220
0
-210
0
-225
1
-212
1
-204
1
20
-220
1
-212
1
-225
0
-221
1
-211
1
-193
3
-195
2
Compound-specific δ2H analyses of n-alkanes and fatty acids
For δ2H analyses of n-alkane and fatty acid biomarkers, an
Accelerated Solvent Extractor (Dionex ASE 200) was used to extract free
lipids from the dried soil samples with dichloromethane (DCM) and methanol
(MeOH; 9:1) according to Zech R. et al. (2013). The total lipid extracts
were separated over pipette columns filled with ∼ 2 g
aminopropyl. n-alkanes were eluted with hexane, more polar lipids with
DCM / MeOH (1:1) and free fatty acids with diethyl ether / acetic acid (19:1).
The n-alkanes were further purified using zeolite (Geokleen) pipette columns.
The zeolite was dried and dissolved in hydrofluoric acid after eluting branched- and
cyclo-alkyl compounds with hexane, and the straight-chain (n-alkyl)
compounds were then recovered by liquid–liquid extraction with hexane. For
samples 1–12, an additional purification step with silver nitrate columns
was carried out in order to eliminate unsaturated compounds. The
chromatograms of the other samples displayed no requirement for this
purification step.
Fatty acids were methylated using 5 % HCl in methanol at 80 ∘C
for 12 h. Subsequently, liquid–liquid extraction with 5 % NaCl and
hexane was used to retrieve fatty acid methyl esters (FAMEs). FAMEs were
purified by elution with dichloromethane over SiO2 columns
(∼ 2 g).
5α androstane and hexamethylbenzene was used for quantification of
the compounds using an Agilent Technologies 7890A gas chromatograph (GC)
equipped with a VF1 column (30 m, 0.25 mm i.d., 0.25 µm film thickness)
and a flame ionization detector (FID). Compound-specific δ2H
values of the long-chain n-alkanes and FAMEs were determined based on at
least triplicate analyses using a gas chromatograph–pyrolysis–isotope ratio
mass spectrometer (GC–pyrolysis–IRMS, Delta V, ThermoFisher Scientific,
Bremen, Germany). The A4 standard mixture (provided by Arndt Schimmelmann,
Indiana University, USA) was run three times per sequence at three different
concentrations. All results are reported after normalization using
multi-linear regression (Paul et al., 2007) and simple mass-balance
correction of the FAMEs for the isotopic composition of the methanol used
for derivatization. Long-term precision of the analyses was monitored using
a laboratory standard (oak, n-C29). The standard was analyzed in every
sequence and yielded a mean value of -147.2 ‰ with a
standard deviation of ±1.7 ‰ across all sequences
run for this study.
Modeling of leaf water 2H enrichment
The empirical data analyses were combined with mechanistic model simulations
of δ2Hleaf water in order to better detect and evaluate
how the dominant climate variables (air temperature and relative air
humidity) influence 2H enrichment in lipids. The 2H
enrichment of leaf water due to evapotranspiration can be predicted by using
mechanistic models originally developed for isotope fractionation processes
associated with evaporation from water surfaces by Craig and Gordon (1965).
These models were adapted for plants by Dongmann et al. (1974) and
subsequently by Flanagan et al. (1991) and Farquhar and Lloyd (1993).
Evaporative 2H enrichment of the leaf water (Δ2He) at
the evaporative surface in the mesophyll is given by the equation
Δ2He=ε++εk+Δ2HWV-εkeaei,
where ε+ is the equilibrium fractionation between liquid
water and vapor at the air–water interfaces, εk is
the kinetic fractionation during water vapor diffusion from leaf
intercellular air space to the atmosphere, Δ2HWV is the
isotopic difference of the water vapor and the source water and
ea/ei is the ratio of ambient to intercellular vapor pressure
(Farquhar and Lloyd, 1993). This basic calculation was modified by including
a Péclet effect that accounts for opposing fluxes of source water
entering the leaf through the transpiration flow and the back-diffusion of
isotopically enriched water from the sites of evaporation (Farquhar and
Lloyd, 1993):
Δ2Hleafwater=Δ2He(1-e-φ)EL/CD.
The quotient of EL/CD represents the Péclet number (φ) where E is the transpiration rate, L is the effective path length, C is
the molar concentration of water and D is the diffusivity of
1H2HO. The model approach we used followed that of Kahmen et al. (2011b), where the Péclet-modified Craig–Gordon (PMCG) model is reduced to
three input variables: air temperature, atmospheric vapor pressure and
source water δ2H. This simplified model is based on the
assumption that throughout the season leaf temperature equals air
temperature and that atmospheric vapor δ2H is generally in
equilibrium with source water δ2H (Kahmen et al., 2011b).
Transpiration rates are estimated using relative humidity and air
temperature (retrieved from GeoINTA, 2012) and assuming a mean stomatal
conductance of 0.15 mol m-2 s-1. Based on reports for a large number of
species in the literature (Kahmen et al., 2008, 2009; Song et
al., 2013), we used an average value of 20 mm for L and kept it constant
across the transect. For our simulation of leaf water δ2H
values we obtained the model input variables air temperature and atmospheric vapor pressure from GeoINTA (2012)
and source water δ2H from Bowen (2012), respectively.
The isotopic composition of the leaf water can be estimated according to
Eq. (3):
δ2Hleaf water=Δ2Hleaf water+δ2HSW,
where Δ2Hleaf water is the bulk leaf water evaporative
enrichment and δ2HSW is the hydrogen isotope ratio of
source / xylem water.
Conceptual model for the coupled δ18O–δ2H biomarker approach
The conceptual coupled δ2Hn-alkane–δ18Osugar model was introduced
previously by Zech M. et al. (2013). In brief, it is based on the following fundamentals. Precipitation
world-wide typically plots along/close to the so-called global meteoric water
line (GMWL, δ2H = 8 × δ18O + 10) in a
δ18O–δ2H diagram (Dansgaard, 1964; Fig. 5). Due
to fractionation processes, evaporation/transpiration causes water vapor to
be isotopically depleted in 18O and 2H, whereas residual (leaf)
water (δ2H/δ18Oleaf water) is isotopically
enriched. In a δ18O–δ2H diagram, leaf water
therefore does not plot on the GMWL but on an evaporation line (EL). The
distance of leaf water to the GMWL can be
described as deuterium excess (d=δ2H – 8×δ18O). Using a Craig-Gordon model adapted by Gat and Bowser (1991), the
d-excess of leaf water can be used to calculate RH values normalized to the
temperature of leaf water (M. Zech et al., 2013):
RH=1-Δd(ε2*-8⋅ε18*+Ck2-8⋅Ck18),
where Δd represents the difference in d-excess between leaf water and
source water. According to Merlivat (1978), experimentally determined
kinetic isotope fractionation equals 25.1 and 28.5 ‰ for Ck2 and Ck18, respectively,
considering that these are the maximum values of kinetic fractionation
during molecular diffusion of water through stagnant air. Equilibrium
isotope enrichments ε2* and ε18* as functions of temperature can be calculated using
empirical equations of Horita and Wesolowski (1994). Hence, provided
that n-alkanes and sugars in plants and soils reflect (albeit with a constant
offset caused by biosynthetic fractionation) the isotopic composition of
leaf water, a coupled δ2Hn-alkane–δ18Osugar approach allows for reconstructing RH values. Furthermore, the
biomarker-based “reconstructed” δ2H/δ18Oleaf water values allow reconstructing the isotopic composition of
plant source water, which can be considered in an approximation to reflect
δ2H/δ18Oprec (illustrated as intercepts of
the individual ELs with the GMWL in Fig. 5). Assuming a slope of
∼ 2.82 seems reasonable based both on model considerations and
on field observations and laboratory experiments (Allison et al.,
1985; Walker and Brunel, 1990; Bariac et al., 1994). For further details on
modeling coupled δ18O–δ2H biomarker results, the
reader is referred to Zech M. et al. (2013).
Results and discussion
Comparison of δ2Hn-alkanes and δ2Hfatty acids
The C29 and C31n-alkane homologues were sufficiently abundant in
all samples to be measured for their hydrogen isotopic composition. The
δ2H values range from -155 to -222 ‰ and
reveal a similar trend between n-C29 and n-C31 along the investigated
transect (Table 1 and Fig. 2). While the northern and middle part of the
transect is characterized by relatively high δ2H values
(∼ -160 ‰), the southern part of the
transect is characterized by considerably more negative δ2H
values (∼ -210 ‰).
Comparison of δ2H results of individual leaf wax
n-alkanes and n-alkanoic (fatty) acids along the investigated transect.
The δ2H values of the fatty acids n-C22, n-C24,
n-C26, n-C28 and n-C30 range from -128 to -225 ‰ (Table 1 and Fig. 2). In general, there is a good
overall agreement between the n-alkanes and the fatty acids (R= 0.96,
p < 0.001, n= 20; for the weighted means), both showing more
negative δ2H values in the southern than in the northern and
middle portions of the transect (Table 1, Fig. 2). Interestingly, the longer
homologues n-C28 and n-C30 are systematically more enriched, by 3 to 43 ‰, compared to the
n-alkanes. The same was observed by Chikaraishi and Naraoka (2007), reporting
on n-alkanes being depleted in 2H relative to the corresponding
n-alkanoic acid. The reasons for this trend remain vague at this point, but may
be related to metabolic pathways, seasonal differences in homologue
production or differences in homologue sources. Roots, for example, have
also been suggested as a source of long-chain n-fatty acids (Bull et al.,
2000). Shorter homologues, have been suggested to be not only plant-derived,
but also of bacterial origin (Matsumoto et al., 2007; Bianchi and Canuel,
2011). Similarly, soil microbial overprinting of long-chain n-alkanes and
fatty acids cannot be excluded (Nguyen Tu et al., 2011; Zech M. et al.,
2011a). By contrast, there is strong evidence suggesting that n-alkanes are
not significantly introduced into soils/subsoils by roots (Häggi et al.,
2014), which seems reasonable given the significantly lower amounts of
n-alkanes being produced by roots compared to above-ground plant organs (Zech
M. et al., 2012a; Gamarra and Kahmen, 2015).
The consistent δ2H pattern revealed by the n-alkanes and fatty
acids along the north–south climate transect does not solely reflect the
δ2H isotopic composition of precipitation. δ2H of the lipid biomarkers shows
a pronounced offset, especially in the middle part of the transect (Fig. 3). Given that n-alkanes are considered to
primarily reflect leaf signals and are most widely applied in paleoclimate
and paleohydrological studies, we will principally refer to δ2H
of long-chain n-alkanes in further discussion and calculations.
Comparison of measured δ2Hn-alkanes
(weighted mean of n-C29 and n-C31) and δ2Hfatty acids (weighted mean of n-C22,
n-C24, n-C26, n-C28 and
n-C30) patterns with δ2Hprec (Bowen, 2012) along the
north–south climate transect (xmin and +max representing annual
minimum and maximum value at the sampling sites). Additionally, assuming a
biosynthetic fractionation of -160 ‰ for the n-alkane and
fatty acid biosynthesis in plants the biomarker-based reconstructed
isotopic composition of leaf water is shown.
Evapotranspirative 2H enrichment of leaf
water
Assuming a constant biosynthetic fractionation of -160 ‰
for the n-alkane and fatty acids biosynthesis in plants (Sessions et al.,
1999; Sachse et al., 2006), we estimated the isotopic composition of leaf
water using our n-alkane and fatty acids δ2H values along the
transect/gradient (Fig. 3). Note that an average biosynthetic fractionation
factor of ∼ -200 ‰ was reported by
Sessions et al. (1999) for short- and mid-chain fatty acids synthesized
mostly by unicellular/multicellular marine algae. By contrast, there are
hardly any biosynthetic fractionation factors reported for long-chain
fatty acids of higher plants. Given that our δ2H n-alkanes and
fatty acids values are very similar, using a biosynthetic fractionation
factor of -160 ‰ for both lipids seems appropriate.
Estimated leaf water δ2H values suggest a pronounced greater 2H
enrichment of leaf water compared to precipitation (up to +62 ‰). This finding highlights the role of aridity for
evapotranspiration and isotopic enrichment of leaf waxes, in good agreement
with prior studies (Sachse et al., 2006; Feakins and Sessions, 2010; Douglas
et al., 2012; Kahmen et al., 2013a).
Figure 4 illustrates the overall good agreement between δ2Hleaf water values inferred from the measured n-alkanes and
fatty acids, and δ2Hleaf water values calculated using
the PMCG model. The correlations are highly
significant (r= 0.88, p < 0.001, n= 20, for n-alkanes and r= 0.93,
p < 0.001, n= 20 for fatty acids), suggesting that the model
correctly implements the most relevant processes related to
evapotranspirative enrichment of leaf water. While predicting the overall
trend in leaf water δ2H along the transect with reasonable
accuracy, the model does not capture site-to-site excursions in the
n-alkane-derived leaf water δ2H values from this overall trend.
As such, additional influences that are not captured by the model, such as
possible evaporative 2H enrichment of soil water (see, e.g., Dubbert et
al., 2013), could explain the underestimation of the modeled δ2Hleaf water values in the middle part of the transect (Fig. 4).
In contrast, the model might overestimate δ2Hleaf
water in the southern part of the transect. The corresponding ecosystem,
the Patagonian Steppe, is a grassland, whereas the middle part of the
transect is dominated by shrubland. Grass-derived lipids have been shown to
be less strongly affected by evaporative leaf water 2H enrichment than
those of trees or shrubs (McInerney et al., 2011; Yang et al., 2011; Sachse
et al., 2012; Kahmen et al., 2013b), and hence the overestimation of the
model may be due to plant species effects (Pedentchouk et al., 2008; Douglas
et al., 2012). The more pronounced offsets in Patagonia could additionally
be attributed to a seasonality effect. The growing season in Patagonia is
not year-round but mainly in spring.
Results of δ2Hleaf water model simulations
and comparison with biomarker-based reconstructed (assuming a biosynthetic
fractionation factor of -160 ‰) isotopic composition of
leaf water based on n-alkanes and fatty acids, respectively. Sensitivity
tests for δ2Hleaf water are shown for changes in RH
and air temperature for all 20 sites along the transect.
In order to assess the sensitivity of the model to the input parameters, we
varied vapor pressure of air by ±5 hPa and mean annual temperature by
±5 ∘C. Changing ea in Eq. (1) by ±5 hPa
corresponds to changes of RH from ca. 94 to 46 % at the beginning of
the transect and 89 to 15 % at the end of the transect. While changes
in temperature have only negligible effects on the modeled δ2H
isotopic composition of leaf water, changes in RH yield difference in
δ2Hleaf water of up to ∼ 30 ‰ (Fig. 4). Different climatic conditions during the
spring growing season in Patagonia could thus explain the overestimation of
the evapotranspirative enrichment in the model.
Evapotranspirative enrichment of leaf water has also been observed in
δ18O values of hemicellulose-derived arabinose, fucose and
xylose analyzed in topsoils along the investigated transect (Tuthorn et
al., 2014). Model sensitivity tests of 18O enrichment of leaf water
using the PMCG model corroborate the observations presented here that air
humidity is the key factor defining the 18O–2H enrichment of leaf
water.
Coupling of the δ2Hn-alkane and δ18Osugar biomarker results
The conceptual model for the coupled δ2Hn-alkane–δ18Osugar biomarker approach is
illustrated in Fig. 5. The model is based on the assumption that the
investigated n-alkane and hemicellulose biomarkers are primarily leaf-derived
and reflect the isotopic composition of leaf water. With regard to the
topsoil transect investigated here, this assumption is reasonable and
supported by leaf water modeling (for δ2H in Sect. 3.2, and
for δ18O see Tuthorn et al., 2014). Accordingly,
biomarker-based reconstructed δ2H/δ18Oleaf water values can be calculated from the biomarkers by applying
biosynthetic fractionation factors εbio. For our
reconstructions, we applied εbio factors of -160 ‰ (Sessions et al., 1999; Sachse et al., 2006)
and +27 ‰ (Sternberg et al., 1986; Yakir and DeNiro, 1990;
Schmidt et al., 2001; Cernusak et al., 2003; Gessler et al., 2009) for
δ2H and δ18O, respectively (Fig. 5).
δ18O–δ2H diagram illustrating the
conceptual model of the coupled δ2Hn-alkane-δ18Osugar approach (modified after Zech M. et al., 2013).
δ2Hn-alkane (mean of n-C29 and n-C31) and
δ18Osugar (mean of arabinose, fucose and xylose) results
are used to reconstruct δ2H/δ18Oleaf water
by subtracting the biosynthetic fractionation factors. The deuterium excess
(d=δ2H – 8⋅δ18O) of leaf water serves
as proxy for RH, and δ2H/δ18Oprec is
calculated as the intersection of the individual evaporation lines (ELs, slope of
2.82) with the GMWL.
Reconstructed RH values along the climate transect and
comparison with actual RH values
The reconstructed d-excess values of leaf water along the investigated
transect range from -67 to -178 ‰ and reveal a
systematic trend towards more negative values in the south (Fig. 6). The
reconstructed RH values calculated using the leaf water d-excess values
according to the above-described coupled δ2Hn-alkane–δ18Osugar approach range from 16 to 65 %, with one
extremely low value of 5 % (Fig. 6). Reconstructed RH values follow the
systematic d-excess trend and correlate highly significantly (r= 0.79,
p < 0.001, n= 20) with the actual mean annual RH values retrieved
from GeoINTA (2012) for all investigated sites.
Comparison of biomarker-based reconstructed relative
humidity (RH) values with actual RH values (mean annual RH retrieved for all
investigated sites from GeoINTA, 2012; mean summer daytime RH for six
stations retrieved from www.ncdc.noaa.gov). Deuterium excess values were
calculated using δ18Oleaf water reconstructed from
terrestrial sugars (Tuthorn et al., 2014) and δ2Hleaf water reconstructed from n-alkanes.
Correlation of biomarker-based reconstructed δ18Oprec and
δ2Hprec values with modern
“actual” δ18Oprec and δ2Hprec values (from Bowen,
2012).
However, as depicted by Fig. 6, the reconstructed RH values systematically
underestimate the actual mean annual RH values. This is especially
pronounced for the three southernmost locations (18–20) and may be
attributed to several causes. First, the applied model calculations do not
account for evaporative enrichment of soil water. In the δ18O-δ2H diagram, the soil water enrichment shifts the
source water (simplified to “reconstructed precipitation” in Fig. 5 and our
model) along the evaporation line and thus leads to too-negative d-excess
values and an underestimation of RH. Second, given that leaf waxes are
considered to be formed mostly during early stages of leaf ontogeny
(Kolattukudy, 1970; Riederer and Markstaedter, 1996; Kahmen et al., 2011a;
Tipple et al., 2013), they may not necessarily reflect the mean annual
isotopic composition of precipitation in regions with pronounced
seasonality, but rather the isotopic composition of precipitation during the
growing season. Furthermore, mean annual RH values likely overestimate the
RH values actually seen in photosynthetically active leaves. Indeed,
when comparing the biomarker-based reconstructed RH values with mean
summer daytime RH values (available for six stations along the investigated
transect from www.ncdc.noaa.gov), satisfactory agreement between
reconstructed and actual RH values is obtained, with the exception of the
southern portion of the transect (Fig. 6). Third, the δ18O
biosynthetic fractionation factor of ∼ +27 ‰, which has been reported for newly assimilated sugars
and cellulose, underestimates in our opinion the actual fractionation factor
of hemicellulose sugars (Tuthorn et al., 2014; Zech M. et al., 2014a). This
results in reconstructed leaf water values plotting too far to the right in
the δ18O–δ2H diagram (Fig. 5), and in turn leading to the
observed underestimated RH values (Fig. 6). We argue with the loss of a
relatively 18O-depleted oxygen atom attached to C-6 during pentose
biosynthesis (C-6 decarboxylation; Altermatt and Neish, 1956; Harper and
Bar-Peled, 2002; Burget et al., 2003) and point to a recent study of
Waterhouse et al. (2013), who determined the position-specific δ18O values in cellulose. Further experimental studies, like those suggested and
encouraged by Sternberg (2014) and Zech M. et al. (2014b), are urgently
needed to determine an improved biosynthetic fractionation factor for
hemicellulose-derived sugars.
Comparison of reconstructed and actual δ2Hprec and δ18Oprec values
Values of δ18Oprec and δ2Hprec
reconstructed as the intercepts of the individual evaporation lines (EL)
with the GMWL in the δ18O–δ2H diagram (Fig. 5)
range from -7 to -22 ‰ and from -47 to -166 ‰, respectively. They correlate highly significantly
(Fig. 7; r= 0.90, p < 0.001, n= 20, and r= 0.88, p < 0.001,
n= 20 for δ18Oprec and δ2Hprec,
respectively) with the “actual” δ2Hprec and δ18Oprec values as derived from Bowen (2012). While the
reconstructed δ18Oprec and δ2Hprec
values, like the reconstructed RH values, generally validate the conceptual
coupled δ18O–δ2H approach, they appear to
systematically underestimate the actual δ18O and δ2H values of the precipitation water (Fig. 7).
The uncertainties discussed above for the observed offset of
reconstructed
versus actual RH values can also affect the accuracy of reconstructed
δ18Oprec and δ2Hprec values. Hence, the
actual δ2H/δ18Oprec values used for our
comparison with the biomarker-based reconstructed values can be assumed to
be one of the possible sources of uncertainty. While Bowen (2012) reported
a confidence interval (95 %) ranging from 0.2 to
1.2 ‰ and from 2 to
11 ‰ for δ2Hprec and δ18Oprec, respectively, future climate transect studies that will
be carried out with actual precipitation being sampled for δ2H–δ18O analyses are encouraged. Moreover, we would also like
to emphasize here the very likely influence of seasonality. As reported
for sugar biomarkers (Tuthorn et al., 2014), we also suggest that leaf waxes
mainly reflect the humidity and the isotopic composition of spring and
summer precipitation rather than mean annual values.
Conclusions
The hydrogen isotopic composition of leaf wax n-alkanes and n-alkanoic (fatty)
acids extracted from topsoils along a transect in Argentina varies
significantly, with δ2H values ranging from -155 to -222 ‰ and -128 to -225 ‰, respectively.
These δ2H values broadly parallel variations in the hydrogen
isotopic composition of precipitation, but are modulated by evaporative
2H enrichment of leaf water. A mechanistic leaf water model correctly
simulates the overall trends. Sensitivity tests show that relative humidity
exerts a much stronger influence on evaporative enrichment than temperature.
In order to evaluate the conceptual coupled δ2Hn-alkane–δ18Osugar approach proposed by
Zech M. et al. (2013), we interpreted the biomarkers extracted from our
Argentinean climate transect as such:
Assuming that the n-alkanes and hemicellulose-derived sugars are primarily leaf-derived, we reconstructed δ2Hleaf water and δ18Oleaf water.
This in turn allowed us assessing the d-excess of leaf water. The large calculated range in d-excess along the transect (-67 to -178 ‰) can be used to calculate biomarker-based
reconstructed RH values. Reconstructed RH values correlate highly significantly with actual mean annual RH values along the transect. Despite this highly significant correlation,
reconstructed RH values systematically underestimate actual mean annual RH values. However, this discrepancy is largely reduced when reconstructed RH values are compared with actual mean summer daytime RH values.
Similarly, biomarker-based reconstructed δ18Oprec and δ2Hprec values correlate highly significantly with actual δ18Oprec
and δ2Hprec values, but reveal systematic offsets, too.
We conclude that compared to single δ2Hn-alkane or
δ18Osugar records, the proposed coupled δ2Hn-alkane–δ18Osugar approach will allow
more robust δ2H/δ18Oprec reconstructions in
future paleoclimate studies. Additionally, it allows for establishing a
“paleohygrometer”, more specifically the reconstruction of mean summer
daytime RH changes/history using d-excess of leaf water as proxy. However,
further studies are needed to determine an improved biosynthetic
fractionation factor for hemicellulose-derived sugars. Also, in the light of
strong diurnal variations of δ2H and δ18O of leaf
water, it is important to determine which portion of this diurnal
signal is actually incorporated in the n-alkanes and sugars being synthesized
in the leaves.