HAL Id: hal-02962491
https://hal.archives-ouvertes.fr/hal-02962491
Submitted on 7 Jan 2021HAL is a multi-disciplinary open access
archive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
Validation and calibration of soil δ2H and brGDGTs
along (E-W) and strike (N-S) of the Himalayan climatic
gradient
Iris van der Veen, Francien Peterse, Jesse Davenport, Bernd Meese, Bodo
Bookhagen, Christian France-Lanord, Ansgar Kahmen, Hima
Hassenruck–gudipati, Ananta Gajurel, Manfred Strecker, et al.
To cite this version:
Iris van der Veen, Francien Peterse, Jesse Davenport, Bernd Meese, Bodo Bookhagen, et al.. Validation and calibration of soil δ2H and brGDGTs along (E-W) and strike (N-S) of the Hi-malayan climatic gradient. Geochimica et Cosmochimica Acta, Elsevier, 2020, 290, pp.408-423. �10.1016/j.gca.2020.09.014�. �hal-02962491�
Iris van der Veen, Francien Peterse , Jesse Davenport , Bernd
Meese , Bodo Bookhagen , Christian France-Lanord , Ansgar
Kahmen , Hima J. Hassenruck–Gudipati , Ananta Gajurel ,
Manfred R. Strecker , Dirk Sachse
2020
Validation and calibration of soil δ2H and brGDGTs along (EW)
and strike (NS) of the Himalayan climatic gradient.
Geochimica Et Cosmochimica Acta, 290, 408–423.
Validation and limits of soil n-alkane δ2H values and brGDGT abundances
1
for paleoaltimetry in the Himalayas
2 3
Iris van der Veen1, Francien Peterse2, Jesse Davenport3, Bernd Meese1, Bodo
4
Bookhagen1, Christian France-Lanord3, Ansgar Kahmen4, Hima J. Hassenruck –
5
Gudipati6, Ananta Gajurel7, Manfred R. Strecker1, Dirk Sachse5
6 7
1Universität Potsdam, Institut für Erd- und Umweltwissenschaften,
8
Potsdam-Golm, Germany 9
(*Correspondence: Veen@geo.uni-potsdam.de) 10
2University of Utrecht, Department of Earth Sciences, the Netherlands
11
3Geochemistry, Centre des Recherche Petrographiques et Geochimiques
(CNRS-12
CRPG), Vandœuvre-lés-Nancy France 13
4University of Basel, Botanical research, Basel, Switzerland
14
5GFZ German Research Centre for Geosciences, Section 4.6: Geomorphology,
15
Organic Surface Geochemistry Lab Potsdam Germany 16
6Department of Geological Sciences, The University of Texas at Austin, 2275
17
Speedway, M.S. C9000, Austin, Texas 78712, USA 18
7Tribhuvan University, Tri-Chandra College, Central Department of Geology,
19 Kathmandu, Nepal 20 21 22 *Manuscript
Abstract
23
Reconstructing the timing of the uplift of mountain ranges and the evolution of 24
high-altitude plateaus is important to understand potential feedbacks between 25
tectonics and climate at geological timescales. This requires proxies that can 26
accurately reconstruct elevation during different time periods in the past. Often, 27
the sensitivity of climatic parameters to elevation gradients, recorded in 28
geological archives such as soils, is used to estimate paleoelevations. However, 29
most proxies reflect an indirect response to elevation change, adding 30
uncertainties to reconstructions. In this study we aim to identify sources of
31
uncertainty in paleoelevation reconstruction and test if the combined application 32
of two such proxies, i.e. stable hydrogen isotope ratios (expressed as δ2H values)
33
of plant waxes in modern soils and bacterial membrane lipids (brGDGTs) in soils
34
can potentially reduce uncertainties in the estimation of (paleo-) elevation. We
35
performed this study in four major Himalayan river catchments along the
36
mountain range from west to east: Sutlej, Alaknanda, Khudi and Arun. Each
37
individual catchment is subject to a unique precipitation regime, relative
38
influences of moisture sources, and vegetation cover. In total, we have analyzed
39
275 surface water samples, 9 precipitation samples, 131 xylem water samples
40
and 60 soil samples, which were collected between 2009 and 2014.
41
We find that δ2H values (δ2Hwax) of soil derived nC31-alkanes in the Sutlej,
42
Alaknanda, Khudi and Arun catchments generally record surface water δ2H
43
values, confirming that the first-order control on plant wax isotopic signature is
44
precipitation δ2H and as such elevation in orogenic settings. We identified aridity
45
as an additional parameter to influence this relationship and hence introducing
46
uncertainty, particularly important in the driest catchment (Sutlej).
derived Mean Annual Temperature (MATmr) correlated in a statistically
48
significant manner with sample site elevation and a 14-year annual average of
49
remote sensing (MODIS) land-surface temperature in all four catchments,
50
showing that the main process influencing the distribution of brGDGTs is the
51
adiabatic cooling of air.
52
In an effort to combine these proxies to improve uncertainties in
53
elevation reconstruction, elevations were inferred independently from both the
54
δ2Hwax and brGDGT distributions. Samples from arid, high elevation sites
55
underestimated actual sample site elevations derived from δ2Hwax values, while
56
sites subject to high (>23-25°C) annual temperatures overestimated actual
57
sample site elevation inferred from brGDGT distributions. Therefore, elevations
58
inferred from both proxies under very arid and/or very warm paleoclimatic
59
conditions should be interpreted with caution.
60
In turn we suggest that the difference in the elevation estimate between
61
the two proxies, described by the proposed ΔElevation parameter, can provide
62
information on hydrological conditions of the soil’s depositional environment. In
63
conclusion, we emphasize that knowledge of the sample site’s climatic conditions
64
are essential in order to reconstruct elevation from paleoarchives. In particular,
65
knowledge of moisture availability and annual air temperatures are important,
66
as these have been found to cause the largest scatter in the observed data.
1. Introduction
68
The uplift of mountain ranges has been a key interest in climate and tectonic 69
studies, due to its impact on atmospheric circulation patterns, erosion and 70
precipitation patterns (Zhisheng et al., 2001; Clift et al., 2008). For example, the 71
uplift of the Himalayan mountain range has played a key role the onset and 72
intensification of the Indian (ISM) and East Asian (EASM) monsoons, as well as in 73
changing the global carbon cycle due to increasing weathering of freshly exposed 74
bedrock (Molnar and England, 1990; Garzione et al., 2000a; Dettman et al., 2003; 75
Quade et al., 2007; Poulsen, 2011). In order to test these scenarios, uplift 76
histories and comparison with coeval climatic records are needed. 77
In order to reconstruct mountain range paleoelevation, proxies that record 78
the persistent hydrological and climatological gradients of mountain ranges 79
preserved in geological archives, are employed. These proxies include numerous 80
stable isotope approaches, such as δ18O values of pedogenic carbonates
81
(Garzione et al., 2000b; Quade et al., 2007), leaf wax δ2H values (Peterse et al.,
82
2009; Polissar et al., 2009; Zhuang et al., 2014; Bai et al., 2015) and branched 83
tetraether membrane lipid distributions from paleosols and sediments (Ernst et 84
al., 2013; Wang et al., 2017). These aforementioned proxies are all based on
85
climatic parameters that are observed to co-vary with elevation, and hence,
86
record elevation indirectly: Rayleigh distillation processes during rainout result
87
in a negative relationship between the isotopic composition (δ18O and δ2H) of
88
precipitation and elevation (Dansgaard, 1964; Gat et al., 2000) and temperature
89
generally decreases with increasing elevation.
90
Even though δ18O and δ2H values in precipitation and temperature primarily
91
correlate with elevation (citations, can use Meese at al for this and some others),
locally or seasonally varying processes and climatic conditions such as
93
heterogenic precipitation patterns, complex topography, and the relative
94
influence of multiple moisture sources can distort the general linear elevation
95
relationships (e.g. Rohrmann et al. 2014; Hren et al. 2009; Galewsky 2009).
96
Similarly, mean annual air temperature (MAT) generally decreases with
97
elevation due to the adiabatic cooling of air, and is therefore an indirect measure 98
of elevation as well. Sediments and soils record changes in MAT, for example, 99
through the distribution of soil microbe derived branched glycerol dialkyl 100
glycerol tetraethers (brGDGTs) (Weijers et al., 2007). BrGDGTs are membrane 101
lipids produced by soil bacteria, which vary in the number (4-6) of methyl 102
branches attached to their alkyl backbone, the position of these methyl branches 103
(5 or 6 position), and the number (0-2) of internal cyclizations, depending on the 104
MAT and pH of the soil in which they are produced (Weijers et al., 2007). 105
Nevertheless, the latest global temperature transfer function still contains a 106
substantial amount of scatter, indicating that this proxy is influenced by 107
additional climatic parameters, such as soil moisture content (SMC) and 108
precipitation amount/aridity (Dirghangi et al., 2013; Ernst et al., 2013; Menges et 109
al., 2014; Wang et al., 2014; Dang et al., 2016; Naafs et al., 2017), but possibly 110
also by soil and vegetation type (Davtian et al., 2016). These effects can increase 111
the uncertainties in the relationship of elevation with temperature, and thus the 112
proxy’s robustness in recording elevation. 113
In this study, we validate each of these proxies separately by using a
114
combination of field and remote-sensing data. In a second step, the proxies are
115
combined and we investigate whether this combination is capable of reducing
116
potential uncertainties in the elevation estimates. Specifically, we analyze stable
isotope ratios (δ2H) of surface waters, xylem waters (i.e. the organisms H source)
118
and terrestrial lipid biomarkers in modern soils, and brGDGTs in soils along four
119
climatically distinct altitudinal gradients along the Himalayan orogeny. These
120
altitudinal transects are subject to varying precipitation regimes, different
121
degrees of aridity, relative influences of moisture sources, and vegetation cover,
122
which allows a thorough investigation of the impact of different environmental
123
factors on δ2Hwax and brGDGTs. Combining the multiproxy data with
satellite-124
derived climate products, we aim to identify the controlling secondary factors
125
that potentially alter the relationship between source water δ2H and δ2Hwax, as
126
well as physically measured MAT and brGDGT-derived MAT. Ultimately, we
127
discuss how the offset between these parameters can potentially support
128
paleoelevation studies and aid the interpretation of these proxies in sedimentary
129
archives.
2. Study area
131
The Himalayan mountain range separates the Tibetan Plateau from the Indian 132
subcontinent, and traverses 2,700 km from the Karakoram in Pakistan in the 133
northwest, through India, Nepal and Bhutan into the Arunachal Pradesh in the 134
southeast. The range varies in width between ~ 400 km in the west, and ~ 150 135
km in the east, and contains several significant mountain peaks. 136
The Himalayan range acts as an orographic barrier separating the humid 137
regions in the foreland and arid sections in the rain shadow, resulting in varying 138
precipitation patterns along the Southern Himalayan Front (SHF). Two, 139
somewhat independent, but interfering climatic circulation systems dominate 140
the precipitation regime in the Himalaya: the Indian Summer Monsoon (ISM) 141
system, and the Western Disturbances (WD)(Bookhagen and Burbank, 2010; 142
Cannon et al., 2014). The ISM moves along the SHF and transports moisture from 143
the Bay of Bengal toward the northeast, causing heavy rainfall along the 144
southern slopes of the mountain front during summer (Bookhagen et al., 2005). 145
The second major moisture source is the WD, originating from the Caspian, Black 146
and Mediterranean seas, bringing winter precipitation between December and 147
March (Dash et al., 2009; Wulf et al., 2010; Cannon et al., 2014). Spatially, the 148
dominance of the ISM decreases from east to west, whereas the WD become 149
particularly influential west of 78°E (Bookhagen and Burbank, 2010). The 150
Himalayan range creates an extreme gradient between the humid tropical 151
climate in the foreland and alpine conditions at higher elevations. The two major 152
climatic gradients that strongly influence vegetation cover and type along the 153
orogen are the decreasing air temperature from low to high elevation, and the 154
decreasing amount of moisture from east to west (Singh and Singh, 1987). 155
3. Methods
156
3.1. Elevation transects
157
Four transects from the northeast to the southwest were selected in the 158
Sutlej, Alaknanda, Khudi Khola (further referred to as Khudi) and Arun 159
catchments, of which the first two are located in the western Himalaya in India, 160
and the latter in central and eastern Nepal (Fig.1A). These four transects each 161
cover a large altitudinal gradient, and consequently also span large ranges in 162
precipitation, vegetation and temperature (Fig. 2). 163
The Sutlej catchment elevation transect ranges from 475 – 3,533 m (table 164
1, Supplemental Material A) over a horizontal distance of ~170 km. The lower 165
sites (<3,000 m asl) receive 1,500 – 2,000 mm rainfall annually (cf. TRMM 2B31, 166
Bookhagen & Burbank 2010), decreasing to <500 mm/year at the higher sample 167
sites (>3,000 m asl) (Fig 2A). Mean Annual Temperature (MAT) in the catchment 168
ranges between 9.7 °C and 25.4 °C, with an average of 19.7 °C (2000-2014 yearly 169
average, MOD11C3, Wan & Hulley 2015; Wulf et al. 2016). Concurrently with the 170
decrease in rainfall amount and the increase in elevation, the Enhanced 171
Vegetation Index (EVI), shows that vegetation cover in the catchment decreases 172
with elevation (Fig.2A). 173
The Alaknanda catchment transect ranges from 346 to 3,155 m with a 174
horizontal distance of 160 km. Annual rainfall varies between 300 and 2,600 175
mm/year and consists of two rainout belts (at ~500 and ~2,000 m, Fig. 2B) due 176
to a two-step rise in relief (Fig. 2B), as described in (Bookhagen and Burbank, 177
2006). The 14-year average MAT in the catchment is 19.8 °C, ranging between 178
7.1 °C and 26.8 °C (Wan and Hulley, 2015). 179
The Khudi, where the transect ranges from 2,155 – 4,085 m asl with a
180
horizontal transect of ~13 km, receives rainfall amounts of >3,900 mm/year
181
(Fig. 2C). There is little variation in the amount of rainfall along the altitudinal 182
transect due to its short distance. MAT varies between 7.5 °C and 19.5 °C with an 183
average temperature of 13.1 °C. 184
The transect in the Arun valley ranges from 225 to 2,580m over a 185
horizontal transect of 160 km. Annual rainfall amounts along the transect varies 186
between 2,000 – 4,000 mm/year (Fig. 2D), and shows a similar two-belt rainout 187
pattern as in the Alaknanda transect (at ~500 and ~2,000 m, Fig. 2D). MAT 188
varies between 3.5 °C and 26.7 °C, with an average of 16.3 °C. 189
190 191 192
3.2 Sample collection
193
Soil (0-5 cm, without litter) and water samples in the Alaknanda and the 194
Sutlej catchment were obtained in September/October 2014, whereas soils and 195
water in the Arun catchment were collected both in September 2011 and 196
October/ November 2012 (Hoffmann et al., 2016). In the Khudi Khola catchment 197
only soils were sampled at various depths in September 2009, as indicated in
198
Table 1 (Supplemental Material A). In total, 107 water samples and 17 soil
199
samples in the Sutlej catchment, 52 water samples and 21 soil samples in the 200
Alaknanda catchment, 10 soil samples and 9 precipitation samples in the Khudi 201
Khola catchment, and 116 water samples and 12 soil samples in the Arun 202
catchment were collected. In addition, 35 xylem water samples from dominating 203
vegetation at sample sites in the Alaknanda, 33 in the Arun, and 63 in the Sutlej 204
catchments were collected. At every xylem water sampling location, 9 branches 205
were collected from 3 individuals of the same species. Per sample location, the 2 206
most dominant, angiosperm species were sampled. Due to the large climatic 207
gradient, it was not possible to sample the same species along the entire 208
elevation transect. After collecting the branches, the bark was removed to 209
prevent mixing between xylem water and phloem water, after which the peeled 210
branches were placed in airtight containers. Xylem waters were extracted at a 211
vacuum line the university of Basel (Newberry et al., 2017). Extracted xylem 212
water was measured on a TCEA-IRMS for δ18O and δ2H values.
213 214
3.2. Analysis of δ2H and δ18O values of water samples 215
The surface water δ2H and δ18O values sampled in the Alaknanda and
216
Sutlej catchment were measured on a Picarro Cavity Ringdown Spectrometer 217
L2140-I at GFZ Potsdam, with a precision of 0.08 ‰ for δ18O and 0.5 ‰ for δ2H.
218
All samples were filtered through a 0.45µm syringe filter and stored in 2ml vials. 219
The samples were injected 10 times, with a volume of 1 µl, and the first three 220
injections were discarded for each individual sample, to avoid any memory 221
effect. The measurements were normalized using a two-point correction using 222
VSMOW2 and SLAP2 standards, provided by the IAEA. 223
The stable isotope composition is reported using the δ-notation relative to the
224
Vienna Standard Mean Ocean Water (VSMOW) standard as:
225
Where R is the ratio of heavy isotopes relative to light isotopes (18O/16O and
226
2H/1H). All values are reported in per mille ‰ (implying a factor of 1000).
227
The xylem water δ2H values were measured on a TC-EA system at the University
228 of Basel. 229 230 3.3. Lipid extraction 231
The soil samples were freeze-dried and stored in pre-combusted (500°C) 232
glass vials. Before extraction, all soil samples were sieved with a 2 mm sieve, and 233
any leaves and large roots were removed to avoid any contamination from 234
modern organic material. A total lipid extract (TLE) was extracted from the soil 235
samples using an Accelerated Solvent Extractor (ASE) (Type Dionex ASE 350), by 236
using 9:1 Dichloromethane:Methanol as a solvent. Samples were initially 237
separated into two fractions using columns containing 1,5 – 2 gram 238
precombusted silica gel. Samples were added to the top of the column, and then 239
rinsed with 12 mL hexane to obtain the hydrocarbon fraction. 240
241
3.4. Lipid analysis
242
The n-alkane fraction was analysed on an Agilent GC MSD (Agilent 5975C 243
MSD, Agilent 7890A GC with Agilent J&W HP-5 MS column, 30 m×0.25 m×0.25 244
µm film) coupled to a FID. A Thermo Scientific Delta V Plus IRMS coupled to a 245
Trace 1310 GC (Agilent GC MSD (RESTEK, Rtx-5 Crossbond ms column, 30 246
m×0.25 mm×0.25 µm df. 5% diphenyl, 95% dimethyl polysiloxane) via an Isolink 247
pyrolysis furnace operated at 1420 °C was used at Potsdam University for 248
measurement of n-alkane δ2H values (δ2Hwax). The H3+ factor was determined at
249
the beginning of every sequence, and was constant (3.9±0.7) throughout the 250
entire duration of the measurements, indicating stable conditions in the ion 251
source. All samples were measured in duplicates, with a standard deviation < 252
3‰, and a C10 – C40 standard measured in between every 10 samples. The δ2Hwax
253
values were all normalized to the VSMOW-SLAP scale, with the use of an external
254
standard containing C16 to C30 alkanes (A-Mix, A. Schimmelmann, Indiana
255
University, Bloomington)
256
A known amount of internal standard (C46-GDGT) was added to the GDGT
257
fraction, after which it was filtered through a 0.45µm PTFE filter using 99:1 258
Hexane:2-propanol. Samples were measured on an Agilent 1260 Infinity ultra 259
high performance liquid chromatography (UHPLC) coupled to an Agilent 6130 260
single quadrupole mass detector (MS), according to the method described by 261
(Hopmans et al., 2016). The separation of the brGDGTs has been performed on 262
two silica Waters Acquity UPLC HEB Hilic (1.7µm, 2.1mm x 150mm) columns in 263
tandem, preceded with a guard column of the same material. The [M+H]+ ions
264
were detected in selected ion monitoring mode. 265
266
3.5. Proxy calculation
267
The apparent fractionation (Sauer et al., 2001) εapp, reflecting net fractionation
268
between source water (in this case the surface waters) and soil lipids, is defined
269 as: 270
All values are reported in per mille ‰ (implying a factor of 1000). Surface water
271
δ2H was used as a catchment integrated value of precipitation δ2H (Kendall and
272
Coplen, 2001; Hren et al., 2009; Bershaw et al., 2012).
273
The Branched and Isoprenoid Tetraether (BIT) index was calculated according to
274
the methods described in Hopmans et al. (2004), although adjusted to include
275
both 5 and 6-methyl brGDGT isomers:
276
Where the roman numerals represent different GDGT structures (See De Jonge et
277
al. 2014). Based on the fractional abundances of the brGDGTs, MATmr was
278
determined according to the methods described in De Jonge et al. (2014):
279
(n = 222, r2 = 0.68, RMSE = 4.6°C, p<0.01).
280
The isomerisation Ratio IR was determined according to the methods in De Jonge
281 et al. (2014b): 282 IR =
(5) 283
3.6. Remote sensing
284
MAT was derived by averaging the monthly 0.05° Land Surface 285
Temperature (LST) from 2000 to 2014 from the Moderate Resolution Imaging 286
Spectroradiometer (MODIS) MOD11C3 product, from its first availability in 2000 287
until the fieldwork period (Wan & Hulley 2015). Seasonal and annual rainfall 288
were determined using the Tropical Rainfall Measurement Mission (TRMM) 289
2B31 data product averaged over 1998-2010, with a spatial resolution of ∼5km 290
(Bookhagen and Burbank, 2010; Huffman et al., 2014). Seasonal and annual 291
Enhanced Vegetation Index (EVI) was determined using the MOD13C2 (MODIS), 292
averaging data products from 2000 – 2017 (Didan, 2015). Soil moisture in the 293
root zone (0 – 100 cm) in m3/m3 was determined using the SMAP soil moisture
294
SPL4SMGP data product (Reichle et al., 2016). Data was averaged over the 295
months March – April – May (growing season), and have a spatial resolution of 9 296
x 9 km. The aridity index was determined using the following equation: 297 298 (6) 299 300
where the mean annual precipitation derives from the aforementioned (TRMM 301
2B31) data product and mean annual potential evapotranspiration was 302
determined using the Global PET dataset (Trabucco and Zomer, 2009). 303
304
4. Results
305
4.1. Surface water and xylem water δ2H values
306
In the Sutlej catchment, surface water δ2H values ranged from −50‰ in the
307
foreland to -112‰ in the high elevation catchments (Fig. 3). The Alaknanda
surface water δ2H values ranged from -51‰ to -102‰. The Arun water data
309
(Meese et al., 2018) was sampled together with the soils and additional river
310
sediments (Hoffmann et al., 2016). A negative correlation was observed between
311
δ2H and mean catchment elevation in the Sutlej, Alaknanda and Arun river
312
catchments (R2 = 0.79, p < 0.001 and R2 = 0.82, p < 0.001, R2=0.73, p<0.01,
313
respectively). Moreover, a significant negative relationship was observed
314
between surface water δ2H and sample site elevation in both the Sutlej and
315
Alaknanda (R2 = 0.61, p < 0.01 and R2 = 0.72, p < 0.01, respectively). The sample
316
site elevation represented a more local δ2H signal, while the mean catchment
317
elevation also integrated the upstream catchments of the surface waters. The
318
isotopic lapse rates of the surface water in the Sutlej and Alaknanda catchments
319
were -15.7‰ km-1 and -8.8‰ km-1 respectively, assuming a linear relationship
320
between δ2H and mean catchment elevation. In the Arun catchment, no
321
significant correlation between sample site elevation and surface water δ2H was
322
observed.
323
In the Khudi catchment there was no surface water δ2H dataset obtained, but
324
a precipitation δ2H dataset (Fig. 3C). A significant relationship was observed
325
between sample site elevation and precipitation δ2H (R2 = 0.85, p < 0.01).
326
Xylem water δ2H of the Sutlej samples varied between -18.3‰ at 475
327
masl and -119.4‰ at 3,533 masl (Fig. 3), but showed an absolute difference
328
between the minimum and maximum of 28‰ to 82‰ between individual
329
sample sites. The smallest range in isotope values was observed at the second
330
highest sample site (28‰), while the largest range in isotope values was
331
observed at the highest sample site (55‰). In the Alaknanda, xylem water
332
ranged between 7.1‰ at 346 masl and -74.3‰ at 651 masl. High variability in
the xylem water δ2H was observed along the Arun catchment, where values
334
ranged between -39,7‰ and – 96,07‰.
335 336
4.2. Soil n-alkane isotopic composition
337
Along the Sutlej transect, δ2Hwax ranged between -148‰ and -183‰,
338
over an altitudinal transect from 475 – 3,371 masl, in the Alaknanda catchment
339
δ2Hwax ranged between -129‰ and -202‰, over an altitudinal transect from
340
346 – 3,100 masl, and finally in the Arun catchment values ranged between
-341
125‰ - -180‰ (Fig. 3). The Khudi Khola transect is significantly shorter and
342
steeper than the other three catchments, and δ2Hwax ranged between -169‰ and
343
-229‰ with an altitudinal transect from 1,750 – 4,085 masl over a distance of
344
~13km, whereas the Alaknanda, Sutlej, and Arun transects are each between 160
345
– 175 km long.
346
We observed a significant negative correlation between δ2Hwax and
347
sample site elevation along the Sutlej, Alaknanda, Khudi, and Arun transects
348
(R2Sutlej = 0.48, p = 0.001, R2Alaknanda = 0.32, p = 0.01, R2Khudi = 0.94, p = 0.001,
349
R2Arun = 0.63, p = 0.01; Fig. 3). The δ2Hwax elevation isotopic lapse rate was the
350
lowest in the Sutlej at –6.7 (± 1.8)‰ km–1, –9.6 (± 3.5)‰ km–1 for the
351
Alaknanda, –15.6 (± 3.9)‰ km–1 for the Arun, and –26.3 (± 2.6)‰ km–1 in the
352
Khudi Khola(Fig. 3).
353
The comparison of δ2H of surface waters and δ2Hwax in soils yielded a
354
significant correlation in the Sutlej and Alaknanda catchment (R2 = 0.31, p =0.01
355
and R2 = 0.34, p = 0.01 respectively)(Fig. 4A). In the Arun catchment, no
356
significant correlation (R2 = 0.14, p > 0.01) between δ2H of surface waters and
357
δ2H of soils was observed. No values on surface water δ2H and εapp (see below) in
the Khudi are reported, since no surface water samples were collected to pair the 359 soils. 360 361 4.3. Apparent fractionation 362
Values of the apparent fractionation εapp (equation 2) in the Alaknanda,
363
Sutlej and Arun catchments have average values of 108.3‰, 102.1‰ and
-364
112.3‰ respectively (Table 1, Supplemental Material A), but varied between
-365
138.3‰ and -73.3‰ in the three transects.
366
A significant relationship was observed between εapp and soil moisture
367
content in the root zone during spring (March-April-May, Reichle et al. 2016), in
368
both the Sutlej and the Arun catchment (Fig 4B). In the Sutlej, correlation
369
between soil moisture content and εapp for the entire transect showed a
370
significant correlation (R2 = 0.51, p = 0.001). When only considering the sample
371
sites that with predominant angiospermae vegetation cover, which are the main
372
producers of n-alkanes (Diefendorf et al., 2011; Bush and McInerney, 2013) as
373
opposed to gymnospermae, a higher correlation between soil moisture content
374
and εapp was observed (R2 = 0.65, p = 0.001). In the Alaknanda, where εapp was
375
relatively stable along the entire transect, no significant relationship between
376
soil moisture in the rootzone and εapp was observed (Fig 4B).
377 378 379 380 381
4.4. BrGDGT thermometry
382
MATmr derived from brGDGTs (equation 4) ranged between 5.6°C –
383
17.7°C in the Sutlej, 6.3°C - 22.7°C in the Alaknanda, and -3.4°C and 15.3°C in the 384
Khudi Khola catchments. One of the lowest sample sites (AK6) was located on a 385
very steep north-facing slope, receiving little to no sunlight. As a result, the 386
MATmr for this location was lower than expected (7.8°C) for a sample at this
387
elevation (523 m). We therefore identified this sample as an outlier (studentized 388
residual t(21) = -10.13, adjusted Bonferroni p = 0.02). In all transects, a 389
significant negative correlation between MATmr and sample site elevation was
390
observed (R2Sutlej = 0.76, p = 0.001, n=17, R2Alaknanda = 0.65, p = 0.001, n = 20 and 391
R2Khudi = 0.73, p = 0.001, n= 10)(Fig. 5A). The associated temperature lapse rates 392
in the Sutlej, Alaknanda and Khudi were -3.2±0.5°C km-1, -4.6±1.1°C km-1,
-393
5.0±1.1°C km-1, respectively (Fig. 7B).
394
When correlating MATmr with the 14-year average annual MODIS derived
395
MAT (from its first availability in 2000 until 2014, when our fieldwork was 396
carried out, Wan & Hulley 2015) for each sample location (Fig. 5B), a positive 397
correlation was found in all three catchments:. The Sutlej catchment and the 398
Khudi catchment both showed a highly significant correlation (R2 = 0.72, p <
399
0.001, R2 = 0.76, p < 0.001, respectively), while the Alaknanda catchment showed
400
a lower correlation between MATmr and MAT (MODIS) (R2 = 0.64, p = 0.001).
401
The modern lapse rates in the three catchments were determined by 402
correlating the 14-year MODIS derived MAT with sample site elevation (Fig. 5C). 403
The associated temperature lapse rates in the Sutlej, Alaknanda and Khudi were -404
5.1±0.6°C km-1, -5.4±0.3°C km-1, -6.5±0.4°C km-1, respectively.
In addition to annual temperatures, MATmr from all sample sites was also
406
compared with the 14-year average summer and winter temperatures 407
(Supplemental Material C). Nevertheless, in the Alaknanda and Sutlej the MODIS 408
mean annual temperature showed the highest correlation with MATmr, although
409
winter temperatures correlated slightly better with MATmr in the Khudi
410
catchment compared with the mean annual temperatures (R2 = 0.84, p<0.001,n=
411
10 and R2 = 0.76, p<0.001,n= 10, respectively).
412 413 414
4.5. δ2H and BrGDGT-derived elevation
415
To estimate elevation from δ2Hwax data, we applied a Rayleigh distillation
416
model to the dataset (Rowley et al., 2001; Rowley and Garzione, 2007). This
417
model describes the progressive isotopic depletion of a reservoir in atmospheric
418
moisture during transport (Rowley et al., 2001) and hence depicts the ideal
419
scenario at the point where a moisture packet hits an orographic barrier,
420
resulting in isotopic depletion with altitude. The model variables and a more
421
extensive description of the model can be found in the Supplementary Material
422
B.
423
The δ2Hwax -based elevation reconstructions correlate significantly with
424
actual sample site elevation along all three transects (Fig. 6C). Due to the large
425
uncertainties associated with the Rayleigh model, the standard deviations of the
426
elevation estimates were in a magnitude of up to 2,000 meters. Most samples
427
plotted above the 1:1 line, indicating that the δ2Hwax proxy generally
428
overestimated sample site elevation. The extent of overestimation was the
429
largest in the lower Alaknanda (<2,000 m asl), where reconstructed elevation
430
can be up to 2,500 meters higher than the actual sample site elevation (Fig. 6C).
431
BrGDGT-derived MATmr was translated to elevation by using the MODIS
432
14-year average temperature lapse rate of each elevation transect (Wan and
433
Hulley, 2015). BrGDGT-based elevations were subsequently compared with the
434
actual sample site elevation, and showed a close resemblance in all three
435
transects (Fig. 6B). However, at lower elevations (<1,500 m asl) the
brGDGT-436
derived elevation generally overestimated (~500 m up to~3,000 m) actual
437
sample site elevation, which was particularly evident in the Sutlej and Alaknanda
438
catchments.
440
The difference in estimated elevation from both the δ2Hwax and brGDGT
441
proxies can be visualized by the ΔElevation parameter (Fig. 6D). Soils that were
442
sampled at arid sites (low aridity index) generally plotted under the 0-line, while
443
most of the Alaknanda and all of the Khudi samples plotted above this line.
444 445 446
5. Discussion
447
5.1. Relationship between elevation and surface water δ2H values,δ2Hwax,
448
and brGDGTs
449
In order to assess the robustness of δ2Hwaxand brGDGTs as a proxy for
450
paleoelevation, the relative importance of all factors influencing these proxies
451
other than changes in elevation need to be determined. The four elevational
452
transects along the southern Himalayan front are all characterized by different
453
precipitation amounts, vegetation cover, and moisture sources, and thus allow us
454
to investigate the impact of these variable parameters.
455
The δ2H values of surface waters in the Sutlej, Alaknanda and Arun all show a
456
significant correlation with mean catchment elevation (R2 = 0.79, p < 0.001 and
457
R2 = 0.82, p < 0.001, R2=0.73, p<0.01, respectively). Comparable lapse rates are
458
observed in the Alaknanda and Arun surface waters (Sutlej = -15.7‰ km-1,
459
Alaknanda = -8.8‰ km-1, Arun = -8.8‰ km-1)(Fig. 3), while the higher lapse rate
460
in the Sutlej can be explained by a larger relative contribution of snow and
461
glacial melt from tributaries in the higher elevation regions of the catchment
462
(Karim and Veizer, 2002; Bookhagen and Burbank, 2010; Maurya et al., 2011;
463
Wulf et al., 2016; Varay et al., 2017). The uniform direction of the lapse rates
464
indicates that the main process controlling surface water δ2H values is the
465
progressive rainout of a monsoonal moisture source, i.e. the altitude effect (Gat
466
and Confiantini, 1981). The remaining scatter in the relationship between
467
surface water δ2H and mean catchment elevation is possibly due to a
468
combination of processes such as: evaporation, mixing of moisture sources with
469
a different isotopic signature, blocking of moisture by topography, convective
470
storms, seasonality or contribution of snow and glacial melt, which have been
observed in high elevation systems (e.g. Gat 1996; Dansgaard 1964; Rohrmann
472
et al. 2014; Hughes et al. 2009; Lechler & Niemi 2012).
473
The δ2Hwax shows the expected negative correlation with sample site
474
elevation in all four transects, suggesting that the first order control on the plant
475
wax isotopic signature is precipitation δ2H (Fig. 3). However, the different
476
degrees of correlation between δ2Hwax and elevation in the different transects
477
suggests that this relationship is subject to additional processes (see section
478
5.2.1. for more detail).
479
The significant correlation between MATmr and both sample site elevation
480
and MODIS-derived MAT indicates that the adiabatic cooling of air mainly 481
controls the distribution of brGDGTs in the soils (Fig. 5A,B). Nevertheless, 482
brGDGT-based MATmr is at times below the expected modern temperature
483
(derived from the MODIS MAT remote sensing product) at low elevation sample 484
sites (Fig. 5B). This offset may in part be explained by the absence of field 485
measured temperature data, instead replaced by a 14 year average remotely 486
sensed MODIS MAT product (Wan and Hulley, 2015). However, there is an 487
ongoing discussion on what temperature the brGDGT proxy actually reflects (e.g. 488
soil or air temperature, mean growing season temperature; (Naafs et al., 2017). 489
We note that the 14-year average temperature remote sensing data set used here 490
provides a consistent estimate of temperatures in all trans-Himalayan transects, 491
although the brGDGTs in the soils of this study may have a different turnover 492
rate (Weijers et al., 2010; Huguet et al., 2013), possibly introducing an additional 493
degree of uncertainty. The accuracy of the MOD11C3 temperature product has 494
been estimated to be < 1 K in the range from -10°C to + 58°C range (Wan and Li, 495
2011), and the MODIS MAT and ground station lapse rates have been found to 496
correlate well in previous studies in the Sutlej (Wulf et al., 2016). 497
The underestimation of MATmr compared with the MODIS MAT, and a
498
higher amount of scatter is mainly observed in the warmer, low elevation 499
sections (>20°C, <1500m) of the Sutlej and Alaknanda elevation transects. 500
Similarly, De Jonge et al. (2014) find that brGDGTs in the global soil calibration 501
dataset also underestimate MAT at the high end of the temperature range, and 502
that the maximum MAT that can confidently be reconstructed is ~23-25°C, due 503
to the saturation of the proxy (De Jonge et al., 2014a; Naafs et al., 2017). Hence, 504
our data support the notion of an upper temperature limit of the proxy, 505
explaining the underestimated MATs in the lower regions of these Himalayan 506
study transects. 507
508
5.2. Influence of water availability on soil n-alkane δ2H and brGDGTs
509
5.2.1. Soil n-alkane δ2Hwax
510
While surface water δ2H values have a significant linear relationship with
511
elevation, δ2Hwax values along our altitudinal transect show a higher degree of
512
scatter when compared to sample site elevation (Fig. 3). We can examine the
513
environmental factors that influence plant δ2Hwax and soil δ2Hwax at two different
514
levels: First, processes influencing the plant’s moisture source (precipitation, and
515
to a lesser degree soil water) before it enters the plant. Second, processes that
516
affect the isotopic composition of δ2Hwax through evaporation of leaf water from
517
leaves (Smith and Freeman, 2006; Lai et al., 2006; Liu and Yang, 2008; Feakins
518
and Sessions, 2010; Sachse et al., 2012; Kahmen et al., 2013a).
In order to characterize the offset between source water and lipid δ2H, the
520
apparent fractionation (εapp) was calculated (equation 2). The apparent
521
fractionation incorporates the influence of plant and soil evapotranspiration and
522
plant physiology, and is directly linked to variation in relative humidity and
523
precipitation in the study area (Smith and Freeman, 2006; Sachse et al., 2006;
524
Kahmen et al., 2013b). However, the εapp in this study is used to describe the
525
offset between surface water δ2H and δ2Hwax. Surface water δ2H is assumed to be
526
an annual integrated precipitation δ2H signal, but glacier/snowmelt, seasonality
527
in precipitation and evaporation could influence the isotopic signature of surface
528
water δ2H. Moreover, the apparent fractionation integrates both evaporative
529
effects and biosynthetic fractionation, complicating attempts to decipher the
530
degree of impact from both factors individually (Sessions et al., 1999; Smith and
531
Freeman, 2006; Sachse et al., 2006; Feakins and Sessions, 2010; Kahmen et al.,
532
2013b).
533
In the Sutlej, Alaknanda, and Arun catchments, the apparent fractionation
534
changes with elevation (Fig. 3A, B, D). Unfortunately, along the Khudi elevation
535
transect, no surface water samples coupled to the soil sampling locations were
536
taken (Fig. 3C). Values of apparent fractionation in the Sutlej, Alaknanda and
537
Arun are similar to εapp between soil alkane δ2H and surface waters observed in
538
other elevation transects on the SE Tibetan Plateau (Bai et al., 2015), but lower
539
than other surface soil studies located on the Tibetan Plateau (Jia et al., 2008;
540
Luo et al., 2011; Wang et al., 2017).
541
Soil evaporation enriches δ2H in soil water, and causes a decrease in εapp in
542
arid regions (Smith and Freeman, 2006; Feakins and Sessions, 2010; Polissar and
543
Freeman, 2010). A decrease in apparent fractionation with increasing elevation
is observed along the Sutlej and Alaknanda transect, which are the two transects
545
that receive lowest amounts of precipitation at high elevation sample sites (Fig.2,
546
3). This decrease in εapp with increasing elevation suggests that aridity has an
547
effect on soil δ2Hwax values in the high elevation Sutlej and Alaknanda samples.
548
This hypothesis is supported by the observation of a significant correlation
549
between εapp and soil moisture content (Fig. 4B)(SMC, Reichle et al. 2016), i.e. a
550
decrease of apparent fractionation under drier conditions in both the Sutlej and
551
Arun transects (Polissar and Freeman, 2010; Kahmen et al., 2013a; Schwab et al.,
552
2015). Moreover, a stronger correlation was observed between εapp and SMC in
553
the rootzone, when only considering the soil sample sites with angiospermae in
554
the Sutlej transect (Fig 4.B, circles). Gymnospermae modify their leaf waxes in a
555
different way (having higher photosynthetic discrimination) when subject to
556
specific environmental stress, and display lower stomatal conductance for CO2
557
and H2O vapor in comparison with angiospermae, resulting in a different
558
relationship between surface water δ2H and δ2Hwax (Pedentchouk et al., 2008;
559
Diefendorf et al., 2011; Tipple et al., 2013).
560
However, in the Alaknanda transect, no significant correlation between εapp
561
and any of the aforementioned hydrological parameters was observed. This
562
suggests that aridity is not the main controlling factor determining εapp, or that
563
the surface water δ2H values were not representative for the local conditions at
564
the soil sampling sites.
565 566
5.2.2. BrGDGTs
567
In a similar manner as with the lipid δ2Hwax proxy, confounding factors that may
568
alter the brGDGT distribution in soils also influences the brGDGT temperature
proxy. The most important factors that were found to have an impact on the 570
relationship between brGDGTs and temperature are soil pH and moisture 571
availability (Weijers et al., 2007). These influences become especially important 572
in arid regions (MAP<500 mm), where moisture availability appears to explain a 573
larger part of the variation in brGDGT distribution than temperature (Dirghangi 574
et al. 2013; Wang et al. 2014; Menges et al. 2014; Peterse et al. 2012; Dang et al., 575
2016). Based on brGDGT distributions in a soil transect with a large range in 576
moisture content (0-61%), Dang et al. (2016) suggested that especially 6-methyl 577
brGDGTs respond to variations in moisture content rather than to MAT. 578
Consequently, it was proposed that only sites where the contribution of 6-methyl 579
brGDGTs is low, defined by an isomerisation ratio (IR) <0.5 can be used to 580
reliably reconstruct MAT (Dang et al., 2016; Naafs et al., 2017). Interestingly, for 581
most sites (>75%) in the Alaknanda and Sutlej catchments the IR is well above 582
0.5, whereas the relationship between MATmr and elevation, as well as between
583
MATmr and MODIS MAT is good (Fig. 5). Moreover, there is no trend between
584
SMC and the relative distribution of brGDGTs, suggesting that moisture content
585
does not influencing brGDGT distributions in Himalaya transects, and, more 586
importantly, that the use of IR is generally not a valid method for discarding 587
samples for (paleo-) temperature reconstruction (Supplemental Material A, 588
Table 4). 589
590
5.3. Combination of brGDGTs and n-alkane δ2H as a more robust elevation
591
proxy
592
The dual application of brGDGTs and δ2Hwax has been used to assess the
593
effect of varying environmental conditions on their potential as elevation proxies
(Ernst et al. 2013; Nieto-Moreno et al. 2016; Wang et al. 2017; Hren et al. 2010;
595
Peterse et al., 2009). Although these studies analyzed both proxies in the same
596
set of samples, their performance has so far only been assessed separately. Here,
597
we test the potential of the actual combination of brGDGTs and δ2H in an
598
elevation context, with the aim to improve their use as a reliable paleoelevation
599
proxy.
600
Both the δ2Hwax and brGDGT proxy are used to reconstruct elevation, but
601
the individual proxies are recording different processes that indirectly cause
602
their change with increasing elevation (i.e. Rayleigh distillation and adiabatic
603
cooling of air). The strongest correlation between these proxies is found along
604
the Khudi transect, suggesting that both proxies are suitable for elevation
605
reconstruction in this catchment (Fig.6A). The Khudi transect is relatively short
606
and does not encompass large changes in hydrology (i.e. it is generally wet with
607
mean annual precipitation above 3,000 mm/year), and environmental
608
conditions vary less than in the other transects. The Sutlej shows significant
609
correlation between δ2Hwax and MATmr but with a substantial amount of scatter,
610
while in the Alaknanda no significant correlation was observed. The
611
aforementioned climatic conditions and processes influencing either δ2Hwax or
612
MATmr could be the cause of this scatter.
613
In an attempt to combine these two proxies, and observe to what extent
614
this can improve elevation reconstruction reliability, we assessed the differences
615
in estimated elevation between the proxies (Fig. 6D), where the difference is
616
indicated as ΔElevation:
617 618
619 620
To assess the influence of changing hydrological conditions, ΔElevation is compared
621
to the aridity index (equation 6) of each sample location. The majority of the
622
sites plot within a 1,250m range from the 0-line, indicating that both proxies
623
yield comparable elevation estimates that are less than 1,250m offset (Fig. 6D).
624
In all three trans-Himalayan transects of this study, the brGDGT-derived
625
elevation can be considered a good predictor of elevation. However, δ2Hwax is
626
associated with large error bars, and primarily subject to influences from aridity
627
and soil moisture availability.
628
Generally, samples with negative ΔElevation valuesare located in areas that are
629
either experiencing arid conditions, or located behind an orographic barrier,
630
resulting in moisture blocking (Galewsky, 2009; Hughes et al., 2009). This
631
suggests that in a multiproxy study using both δ2Hwax and MATmr, not only
632
elevation can be reconstructed, but also information can be provided on the
633
hydrological conditions of the (paleo-) soils using the ΔElevation parameter. For
634
example, the formation of soils during arid conditions can result in large offsets
635
between reconstructed elevation and real sample site elevation. We therefore
636
suggest that soils that show large negative offset (ΔElevation value outside of the
637
1,250 m error bars) between the δ2Hwax and brGDGT proxy to be interpreted
638
with caution, since these could be influenced by arid conditions during
639
formation.
5. Conclusion
641
Both leaf wax n-alkane δ2H values and brGDGTs primarily record a climatic
642
parameter that changes with elevation, although additional processes also
643
influence them. We found that the δ2Hwax values in the Sutlej, Alaknanda, and
644
Arun generally record surface water δ2H values. Scatter in this relationship is
645
attributed to the possibility that surface water δ2H values did not represent the
646
local conditions at the soil sampling sites, as well as the influence of aridity on
647
the δ2Hwax signature stored in soils.
648
The brGDGT-based MATmr in the same three transects showed high
649
correlation with sample site elevation and MODIS-derived MAT, but showed an
650
underestimation of the remotely sensed field temperature. This underestimation
651
is likely caused by the inability of the MATmr proxy to robustly record
652
temperatures above ~23-25°C (De Jonge et al., 2014a; Naafs et al., 2017).
653
In order to improve the accuracy of paleoelevation studies, a combined
654
approach between the δ2H and brGDGT proxy could be applied. Large offsets (up
655
to 1,250 m) between elevations inferred from both proxies (ΔElevation ) should be
656
interpreted with caution, and could be affected by moisture availability/aridity,
657
or originating from locations with high annual temperatures. In case of a large
658
offset, where MBT values are <1, the proposed ΔElevation parameter can provide
659
information on the hydrological setting of the depositional environment of these
660
soils.
661
In conclusion, the results of this study show that both the δ2H and brGDGT
662
proxies are optimal in a relatively stable climate (as the shorter Khudi transect)
663
but are significantly influenced by variable hydrology, i.e. increasing aridity (as
664
often the case in orogenic settings with high altitude plateaus). Our results
contribute to the existing literature on organic proxies, showing that the
666
application of a combined proxy approach could provide information on the
667
hydrological circumstances of the depositional environment. The uncertainties
668
in elevation estimates due to aridity, moisture availability and high temperatures
669
need to be taken into account when interpreting paleoelevations. We emphasize
670
that prior knowledge of the climatic and tectonic setting is crucial in order to
671
reconstruct reliable elevations.
672 673
Acknowledgements
674
A Marie Curie ITN (iTECC) funded Iris van der Veen, Dirk Sachse was funded by a 675
DFG Emmy Noether (SA1889/1-1) grant and the ERC Consolidator grant 676
STEEPclim (grant no. 647035). NWO grant no. 834.11.006 enabled the purchase 677
of the UHPLC–MS system used for GDGT analyses at the UU. Francien Peterse 678
acknowledges financial support from NWO-Veni grant no. 863.13.0016. We 679
thank Viktor Evrard and Thomas Rigaudier for technical lab support, Guillaume 680
Morin for sampling soils in Nepal, and Tashi Jigmet for field assistance. 681
682
Figure 1: Topographic overview of the Himalaya (SRTM V3). White circles indicate the sample
683
sites of the soils along 4 altitudinal gradients. White rectangles indicate areas over which the
684
swath profiles were determined, and labeling corresponds to the swath profiles shown in Fig. 2.
685 686
Figure 2: 10-km wide swath profiles along the four altitudinal transects in the Himalaya (cf.
687
Figure 1). 12-year average JJA mean TRMM 2B31 precipitation in blue (Bookhagen and Burbank,
688
2010), 17-year average (2000-2017) MODIS derived EVI in green (Didan, 2015), and topography
689
in black with 1-sigma standard deviation (gray)(USGS, 2006). Yellow circles indicate soil sample
690
sites along the elevation transects.
692
Figure 3: Soil δ2H of nC31 (red, grey, blue and black circles) and δ2H of tributary surface waters
693
(white circles) and xylem waters (green diamonds) in the Sutlej, Alaknanda, Khudi and Arun
694
catchment. εapp is calculated at different elevations according to equation 2. The precipitation δ2H
695
in Fig. C is from Gajurel et al. 2019 (in prep).
696 697
Figure 4: A. δ2Hwax C31 plotted against of surface water δ2H of the tributaries closest to the soil
698
location. Correlation coefficients and p-values were determined with a linear regression (red and
699
grey lines). Only regressions that are statistically significant are shown. B: Apparent fractionation
700
versus soil moisture content in the root zone (m3/m3) from March - May, derived from the Soil
701
Moisture Active Passive (SMAP) soil moisture SPL4SMGP data product (Reichle et al., 2016).
702
Circles indicate sample site with a predominant angiosperm vegetation cover, and triangles
703
indicate sample sites with predominantly gymnosperms. Red and black lines are the linear
704
regressions. Only regressions that are statistically significant are shown
705 706
Figure 5: BrGDGT-derived MATmr for the Sutlej, Alaknanda and Khudi catchments. A: MATmr (De
707
Jonge et al., 2014a) versus sample site elevation. B: MATmr versus a 14 year average MODIS
708
derived MAT (2000-2014, Wan & Hulley 2015). C: MODIS MAT versus sample site elevation.
709
Error bars represent the standard deviation from the 14-year mean MODIS temperature.
710
Regressions were determined by using a least-square weighted linear regression.
711 712
Figure 6: A: MATmr versus δ2Hwax. B. BrGDGT-derived elevation versus actual sample site
713
elevation. C. δ2H derived elevation versus actual sample site elevation. Regressions were
714
determined by least square weighted linear regression. D. Absolute differences between
brGDGT-715
derived and δ2H-derived elevation plotted versus the Aridity index (equation 6).
716 717
References
718
Bai Y., Fang X., Jia G., Sun J., Wen R. and Ye Y. (2015) Different altitude effect of 719
leaf wax n-alkane δD values in surface soils along two vapor transport 720
pathways, southeastern Tibetan Plateau. Geochim. Cosmochim. Acta 170, 721
94–107. Available at: http://dx.doi.org/10.1016/j.gca.2015.08.016. 722
Bershaw J., Penny S. M. and Garzione C. N. (2012) Stable isotopes of modern 723
water across the Himalaya and eastern Tibetan Plateau: Implications for 724
estimates of paleoelevation and paleoclimate. J. Geophys. Res. Atmos. 117, 1– 725
18. 726
Bookhagen B. and Burbank D. W. (2006) Topography, relief, and TRMM-derived 727
rainfall variations along the Himalaya. Geophys. Res. Lett. 33, L08405. 728
Available at: http://doi.wiley.com/10.1029/2006GL026037 [Accessed 729
November 6, 2013]. 730
Bookhagen B. and Burbank D. W. (2010) Toward a complete Himalayan 731
hydrological budget: Spatiotemporal distribution of snowmelt and rainfall 732
and their impact on river discharge. J. Geophys. Res. Earth Surf. 115, F03019. 733
Available at: http://doi.wiley.com/10.1029/2009JF001426 [Accessed 734
January 28, 2014]. 735
Bookhagen B., Thiede R. C. and Strecker M. R. (2005) Abnormal monsoon years 736
and their control on erosion and sediment flux in the high, arid northwest 737
Himalaya. Earth Planet. Sci. Lett. 231, 131–146. Available at: 738
http://linkinghub.elsevier.com/retrieve/pii/S0012821X0400696X 739
[Accessed November 6, 2013]. 740
Bush R. T. and McInerney F. A. (2013) Leaf wax n-alkane distributions in and 741
across modern plants: Implications for paleoecology and chemotaxonomy. 742
Geochim. Cosmochim. Acta 117, 161–179. Available at:
743
http://dx.doi.org/10.1016/j.gca.2013.04.016. 744
Cannon F., Carvalho L. M. V, Jones C. and Bookhagen B. (2014) Multi-annual 745
variations in winter westerly disturbance activity affecting the Himalaya. 746
Clim. Dyn., 441–455.
747
Clift P. D., Hodges K. V., Heslop D., Hannigan R., Van Long H. and Calves G. (2008) 748
Correlation of Himalayan exhumation rates and Asian monsoon intensity. 749
Nat. Geosci. 1, 875–880.
750
Dang X., Yang H., Naafs B. D. A., Pancost R. D. and Xie S. (2016) Evidence of 751
moisture control on the methylation of branched glycerol dialkyl glycerol 752
tetraethers in semi-arid and arid soils. Geochim. Cosmochim. Acta 189, 24– 753
36. 754
Dansgaard W. (1964) Stable isotopes in precipitation. Tellus 16, 436–468. 755
Dash S. K., Kulkarni M. a., Mohanty U. C. and Prasad K. (2009) Changes in the 756
characteristics of rain events in India. J. Geophys. Res. 114, D10109. 757
Available at: http://doi.wiley.com/10.1029/2008JD010572 [Accessed 758
November 19, 2013]. 759
Davtian N., Menot G., Bard E., Poulenard J. and Podwojewski P. (2016) 760
Consideration of soil types for the calibration of molecular proxies for soil 761
pH and temperature using global soil datasets and Vietnamese soil profiles. 762
Org. Geochem. 101, 140–153.
763
De Jonge C., Hopmans E. C., Zell C. I., Kim J. H., Schouten S. and Sinninghe Damsté 764
J. S. (2014a) Occurrence and abundance of 6-methyl branched glycerol 765
dialkyl glycerol tetraethers in soils: Implications for palaeoclimate 766
reconstruction. Geochim. Cosmochim. Acta 141, 97–112. 767
De Jonge C., Stadnitskaia A., Hopmans E. C., Cherkashov G., Fedotov A. and 768
Sinninghe Damsté J. S. (2014b) In situ produced branched glycerol dialkyl 769
glycerol tetraethers in suspended particulate matter from the Yenisei River, 770
Eastern Siberia. Geochim. Cosmochim. Acta 125, 476–491. 771
Dettman D. L., Fang X., Garzione C. N. and Li J. (2003) Uplift-driven climate 772
change at 12 Ma: A long δ18O record from the NE margin of the Tibetan 773
plateau. Earth Planet. Sci. Lett. 214, 267–277. 774
Didan K. (2015) MOD13C2 MODIS/Terra Vegetation Indices Monthly L3 Global 775
0.05Deg CMG V006. NASA EOSDIS Land Processes DAAC. , 5067. 776
Diefendorf A. F., Freeman K. H., Wing S. L. and Graham H. V. (2011) Production of 777
n-alkyl lipids in living plants and implications for the geologic past. Geochim. 778
Cosmochim. Acta 75, 7472–7485. Available at:
779
http://dx.doi.org/10.1016/j.gca.2011.09.028. 780
Dirghangi S. S., Pagani M., Hren M. T. and Tipple B. J. (2013) Distribution of 781
glycerol dialkyl glycerol tetraethers in soils from two environmental 782
transects in the USA. Org. Geochem. 59, 49–60. 783
Ernst N., Peterse F., Breitenbach S. F. M., Syiemlieh H. J. and Eglinton T. I. (2013) 784
Biomarkers record environmental changes along an altitudinal transect in 785
the wettest place on Earth. Org. Geochem. 60, 93–99. Available at: 786
http://dx.doi.org/10.1016/j.orggeochem.2013.05.004. 787
Feakins S. J. and Sessions A. L. (2010) Controls on the D/H ratios of plant leaf 788
waxes in an arid ecosystem. Geochim. Cosmochim. Acta 74, 2128–2141. 789
Available at: http://dx.doi.org/10.1016/j.gca.2010.01.016. 790
Galewsky J. (2009) Orographic precipitation isotopic ratios in stratified 791
atmospheric flows: Implications for paleoelevation studies. Geology 37, 792
791–794. 793
Garzione C. N., Dettman D. L., Quade J., De Celles P. G. and Butler R. F. (2000a) 794
High times on the Tibetan Plateau: Paleoelevation of the Thakkhola graben, 795
Nepal. Geology 28, 339–342. 796
Garzione C. N., Quade J., Decelles P. G. and English N. B. (2000b) Predicting 797
paleoelevation of Tibet and the Himalaya from N 18 O vs . altitude gradients 798
in meteoric water across the Nepal Himalaya. 183. 799
Gat J. R. (1996) Oxygen and Hydrogen Isotopes in the Hydrologic Cycle. Annu. 800
Rev. Earth Planet. Sci. 24, 225–262. Available at:
801
http://www.annualreviews.org/doi/abs/10.1146/annurev.earth.24.1.225. 802
Gat J. R. and Confiantini R. (1981) Stable Isotope Hydrology: Deuterium and 803
Oxygen-18 in the Water Cycle. IAEA Tech. Reports Ser. No. 210, 339. 804
Gat J. R., Mook W. G. and Meyer H. a J. (2000) Observed isotope effects in 805
precipitation. Vol. II Atmos. Water 2, 43–62. 806
Hoffmann B., Feakins S. J., Bookhagen B., Olen S. M., Adhikari D. P., Mainali J. and 807
Sachse D. (2016) Climatic and geomorphic drivers of plant organic matter 808
transport in the Arun River, E Nepal. Earth Planet. Sci. Lett. 452, 104–114. 809
Available at: http://dx.doi.org/10.1016/j.epsl.2016.07.008. 810
Hopmans E. ., Schouten S. and Sinninghe Damsté J. S. (2016) Hopmans, E.C., 811
Schouten, S., Sinninghe Damsté, J.S., 2016. The effect of improved 812
chromatography chromatography on GDGT-based palaeoproxies. Org. 813
Geochem. 93, 1–6. , 2016. 814
Hopmans E. C., Weijers J. W. . H., Schefuß E., Herfort L., Sinninghe Damsté J. S. and 815
Schouten S. (2004) A novel proxy for terrestrial organic matter in sediments 816
based on branched and isoprenoid tetraether lipids. Earth Planet. Sci. Lett. 817