Impacts of natural and human-induced hydrological variability on

Subscriber access provided by EKU Libraries

Characterization of Natural and Affected Environments

Impacts of natural and human-induced hydrological variability on particulate organic carbon dynamics in the Yellow River Meng Yu, Timothy I. Eglinton, Negar Haghipour, Daniel B. Montluçon, Lukas Wacker, Pengfei Hou, Hailong Zhang, and Meixun Zhao Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.8b04705 • Publication Date (Web): 09 Jan 2019 Downloaded from http://pubs.acs.org on January 12, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 34

Environmental Science & Technology

1

Impacts of natural and human-induced hydrological

2

variability on particulate organic carbon dynamics in

3

the Yellow River

4

Meng Yu1,2,3, Timothy I. Eglinton3,* , Negar Haghipour3, Daniel B. Montluçon3, Lukas Wacker4,

5

Pengfei Hou1, Hailong Zhang1, Meixun Zhao1,2*

6

1

Key Laboratory of Marine Chemistry Theory and Technology, Ministry of Education /Institute

7 8

for Advanced Ocean Studies, Ocean University of China, Qingdao 266100, China 2

Laboratory for Marine Ecology and Environmental Science, Qingdao National Laboratory for

9 10 11

Marine Science and Technology, Qingdao 266237, China 3

Geological Institute, Department of Earth Sciences, ETH Zürich, 8092 Zürich, Switzerland 4

Laboratory for Ion Beam Physics, Department of Physics, ETH Zürich, 8093 Zürich,

12

Switzerland

13 14

Abstract

15

Natural and human-induced hydrological changes can influence OC composition in fluvial

16

systems, with biogeochemical consequences in both terrestrial and marine environments. Here, we

17

use bulk and molecular carbon isotopes (13C and

18

particulate OC (POC) composition and age from two locations along the course of the Yellow

14C)

to examine spatiotemporal variations in

ACS Paragon Plus Environment

1


Page 2 of 34

19

River during 2015–2016. Dual carbon isotopes enable deconvolution of modern, pre-aged

20

(millennial age) soil and fossil inputs, revealing heterogeneous OC sources at both sites. Pre-aged

21

OC predominated at the upstream site (Huayuankou) throughout the study period, mostly

22

reflecting the upper riverine OC. Strong downstream (Kenli) intra-annual variations in modern and

23

pre-aged OC were caused by increased contributions from modern aquatic OC production under

24

the drier and less turbid conditions during this El Niño year. The month of July, which included

25

the human-induced water and sediment regulation (WSR) event at Kenli, accounted for 82% of

26

annual POC flux, with lower modern OC contribution compared with periods of natural seasonal

27

variability. Both natural and human-induced hydrological events clearly exert strong influence on

28

both fluxes and composition of Yellow River POC which, in turn, affect the balance between OC

29

remineralization and burial for this major fluvial system.

30

Abstract art

31 32

Introduction

33

Rivers play an important role in regulating atmospheric CO2 over varying timescales through

34

the export of large fluxes of particulate organic carbon (POC) from the continent to the ocean.1-4

35

Pre-aged soil-derived OC and petrogenic OC derived from erosion of sedimentary rocks comprise

36

a significant fraction of the OC in river-dominated margin sediments, and both influence long-

37

term marine and terrestrial carbon cycles,2, 3, 5 with the balance between burial and oxidation of

38

these different carbon pools influencing atmospheric CO2. It is thus important to constrain source


2

Page 3 of 34


39

inputs and timescales of export of OC from rivers to improve our understanding of the terrestrial

40

OC cycle and its links to OC burial in marine environments. However, increasing anthropogenic

41

perturbations (e.g., land-use modification, waterway impoundent, eutrophication) to river systems

42

and their watersheds have markedly altered the natural hydrological regime and associated

43

transport of OC from land to sea, with biogeochemical consequences throughout the land-ocean

44

aquatic continuum.6-9 Assessment of seasonal and interannual variations with respect to OC

45

composition, age and quantity are therefore essential to delineate natural and anthropogenic

46

influences on carbon cycling.

47

The Yellow River (Figure 1), the second largest river in China, is characterized by high sediment

48

load, low POC content (POC%), and an extremely low dissolved to particulate OC ratio

49

(DOC:POC, 1:9).10, 11 It serves as the major contributor of sediment and OC to the adjacent Bohai

50

Sea. Because of climate change and human activities such as reservoir construction, soil-

51

conservation practices and irrigation, the annual water and sediment fluxes of the Yellow River

52

have decreased dramatically, currently corresponding to about 70% and 90% of the 1950s level.12-

53

14 The practice of water and sediment regulation (WSR) started in 2002. This artificial hydrological

54

intervention was designed to mitigate water and sediment imbalances in the Yellow River

55

following the construction of the Xiaolangdi Reservoir (Figure 1), and shifted the lower Yellow

56

River to a highly regulated fluvial system.14 Consequently, the majority of sediment, OC, nutrient

57

and pollutant transport all occurs within this short human-controlled hydrological event that

58

typically is less than 20 days in duration, which has influenced both the morphology of the river

59

mouth and estuarine and coastal ecosystems.15-19

60

The Yellow River serves an excellent system to investigate interactions between hydrological

61

conditions and OC composition under the combined forcing of both natural and anthropogenic


3


Page 4 of 34

62

processes. Previous investigations have focused on the WSR period in order to better constrain its

63

effects on spatiotemporal variability in the flux and composition of riverine materials, including

64

sediments, OC, nutrients, and polycyclic aromatic hydrocarbons along the Yellow River.10, 11, 20-22

65

However, the paucity of spatial and temporal 14C data limits our understanding of its effects on

66

OC sources and ages. Bulk and molecular 14C studies near the estuary have shown that the Yellow

67

River mainly transports millennial-age POC derived from soils, and this characteristic is one of

68

the contributing factors leading to high burial efficiencies of terrestrial OC in the adjacent China

69

marginal seas.20, 23-25 A two-year time-series study in the lower Yellow River revealed largely

70

invariant geochemical characteristics, including proportions of pre-aged OC, despite large

71

variations in OC flux resulting from natural and human-induced changes in hydrological

72

conditions.26 However, this two-year study was too short to evaluate the effects of inter-annual

73

hydrological variations. For example, under the influence of the El Niño in 2015,27 lower

74

precipitation resulted in a dramatic decrease of annual water discharge of Yellow River compared

75

with previously studied interval (2011–2013). The 2015 El Niño event provided an opportunity to

76

investigate the effects of climate variability on the flux and composition of POC transported by

77

the Yellow River. Moreover, there has been no previous investigation of seasonal variability in

78

POC composition synchronously from different sections of the river. Such a comparison is

79

important for understanding the influence of upstream supply of sediments on temporal OC

80

variations at downstream sites.

81

To address above questions, this study examines seasonal variations in stable carbon (δ13C) and

82

radiocarbon (Δ14C) isotopic composition of POC and of source-specific biomarker compounds

83

from a location in the Yellow River mainstream just downstream from the Xiaolangdi Reservoir

84

(Huayuankou site, Figure 1), at the divide between the middle and lower reach, and a location near


4

Page 5 of 34


85

the estuary (Kenli site, Figure 1) during 2015–2016. Thus, spatiotemporal variations of OC

86

composition between Huayuankou and Kenli enable more in-depth comparison of relationships

87

between upper stream and downstream OC dynamics and transport. Building on prior two-year

88

study (2011-2013) at Kenli,26 we examine a three-year time-series compound-specific

89

POC dataset for the Yellow River. In this way, we compare and contrast intra-annual and inter-

90

annual variability in OC composition in order to examine the influence of hydrological changes

91

resulting from natural and human-induced events on POC.

92

Materials and Methods

14C

and

93

Study area and sampling. The Yellow River originates from the Qinghai-Tibet Plateau at an

94

elevation of 4500 m, stretching 5464 km from its source to the Bohai Sea with a drainage area of

95

75.2×104 km2 (Figure 1). Field sampling was carried out at two sites, Huayuankou (HYK, 34.91°N,

96

113.71°E, ca. 700 km upstream of the river mouth) and Kenli (KL, 37.61°N, 118.54°E, ca. 50 km

97

upstream of the river mouth), seasonally during 2015–2016. Because WSR was not implemented

98

in July 2016, we also obtained one sample of the highest discharge day (July 9) during the WSR

99

period in July 2015 at KL. During the higher discharge period, large amounts of water were

100

released from the Xiaolangdi Reservoir, causing much greater resuspension of riverbed sediments

101

in comparison to normal periods (Figure 1, Table 1). The HYK site, located between the middle

102

and lower reach, is ca. 120 km downstream from the Xiaolangdi Reservoir. The latter, which was

103

built in the late 1990s and has a capacity of 12.65 km3, now plays the key role in regulating water

104

and sediment in the lower Yellow River. Further downstream from the HYK site, the river is

105

confined to a narrow basin, characterized by a river channel up raised to 10 m above the

106

surrounding area as a result of severe siltation. The HYK site is thus representative of supply of

107

sediments from the upper reach to the lower reach. The KL site, located near the river mouth, is


5


Page 6 of 34

108

taken to represent Yellow River OC delivery to the adjacent Bohai Sea. Surface (upper 0.5 m) total

109

suspended particles (TSP) samples collected are considered to represent overall water column

110

characteristics because of the shallow water depth of the main channel at both sites (ave., ~1.5m)

111

that exhibits well mixed characteristics almost year-round. About 100 to 150 liters of surface water

112

were filtered through pre-combusted and pre-weighed glass fiber filters (Whatman GF/F, 0.7μm,

113

150mm) and then freeze-dried for further laboratory analysis. Water temperature (T) and salinity

114

(S) were measured by a portable multi-parameter probe (WTW Multi 3420). The in situ suspended

115

sediment solid concentration (TSS) was determined based on the dry weight of particles on the

116

filters and the volume of water filtered.

117 118

Figure 1. Sampling locations of Huayuankou (HYK, 34.91°N, 113.71°E) and Kenli (KL, 37.61°N,

119

118.54°E) sites in the Yellow River. Maps were generated by Ocean Data View software.28

120

Bulk analyses. An aliquot of homogenized TSP was used for elemental (TOC, TN) and TOC

121

isotopic (13C and 14C) analysis after removal of inorganic carbon with 12 N HCl fumigation (60 °C,

122

72h) and NaOH neutralization (60 °C, 72h).29 TOC, TN contents and TOC δ13C and Δ14C

123

composition were determined by the coupled EA-IRMS-AMS online system at the Laboratory for

124

Ion Beam Physics, ETH Zürich.30 Another aliquot of TSP samples were heated at 350 °C for 12h


6

Page 7 of 34


125

to remove organic matter (OM) and then measured for specific surface area (SA) and median grain

126

size by NOVA 4000 surface area analyzer and Malvern Mastersizer 2000, respectively.

127

Extraction and purification of lipid biomarkers. Methods for the extraction, purification and

128

isolation of lipid biomarkers were modified from Tao, et al.31 Briefly, about 50–100 g of freeze-

129

dried

130

dichloromethane/methanol (DCM/MeOH 9:1, 25 min at 100 °C). After saponification with KOH

131

in MeOH (0.5 M, 2 h at 70 °C), the neutral fraction and acid fraction (after acidification to pH =

132

2) were obtained by solvent extraction. Straight-chain n-alkanes were further isolated and purified

133

from the neutral fraction by eluting through SiO2 column, AgNO3-SiO2 and zeolite column with

134

hexane.32 Fatty acids in the acid fraction were derivatized to corresponding fatty acid methyl esters

135

(FAMEs) with MeOH:HCl (95:5, 12 h at 70 °C). Saturated FAMEs were further purified by eluting

136

through SiO2 column with DCM/hexane (2:1) and AgNO3-SiO2 column with DCM. n-Alkanes

137

and FAMEs were quantified against external standards on a gas chromatograph with flame

138

ionization detection (GC-FID; Agilent Technologies 7890A) equipped with an Agilent VF-1 ms

139

column (30 m × 250 µm i.d., 0.25 µm film thickness). The temperature program started with a 1

140

min hold time at 50 °C, followed by a 10 °C/min ramp to 320 °C and a 5 min hold time 320 °C.

141

13C

and

homogenized

suspended

sediment

was

microwave-extracted

with

and 14C analyses of lipid biomarkers. A small aliquot of each lipid fraction was measured

142

for compound-specific 13C by GC-isotope ratio mass spectrometry (GC-IRMS), on a HP 6890 GC

143

coupled with a Thermo Delta-V system. Results were reported as δ13C values relative to VPDB

144

standard (‰) with a standard deviation of < 0.3‰. The remaining purified lipids were subject to

145

preparative capillary gas chromatography (PCGC) using a GC system coupled to a Gerstel

146

preparative fraction collector.33 The resulting isolated compounds were measured for radiocarbon

147

contents by the gas-ion source mini carbon dating accelerator mass spectrometry (MICADAS)


7


Page 8 of 34

148

system at the Laboratory for Ion Beam Physics, ETH Zürich.34 All radiocarbon values are

149

corrected for procedural blanks with error propagations. Procedural blanks yielded 0.74 ± 0.10 μg

150

C with an Fm value of 0.18 ± 0.06 (n = 12) for 64 PCGC injections of lipid compounds in this

151

study. The carbon isotopic values of n-FAMEs were also corrected for the derivative carbon from

152

MeOH. The analytical uncertainty for the whole procedure of compound specific

153

ranged from 5 to 31‰ (ave., 13‰). All the radiocarbon data were reported as Δ14C (‰) values

154

and corresponding conventional

155

Polach.35

14C

14C

analysis

ages (years before 1950 A.D.) according to Stuiver and

156

Monte Carlo simulations. Since bulk δ13C and Δ14C values are generally insufficient to

157

distinguish the pre-aged soil OC from modern/contemporary and fossil OC because of the

158

overlapping end-member isotopic compositions, we applied a dual carbon isotope mixing model

159

based on the δ13C and Δ14C values of both TOC and source-specific biomarkers in order to

160

constrain the proportions of the three OC components.23, 26, 31, 36 The relative fractional contribution

161

of modern/contemporary (fM), pre-aged soil (fS), and fossil (fF) OC to Yellow River POC are

162

assessed by following equations:

163

Δ14CPOC = fM × Δ14CM + fS × Δ14CS + fF × Δ14CF

(1)

164

δ13CPOC = fM × δ13CM + fS × δ13CS + fF × δ13CF

(2)

165

1 = fM + fS + fF

(3)

166

where, δ13CPOC and Δ14CPOC are the measured δ13C and Δ14C values of riverine POC sample.

167

The end-member δ13C and Δ14C values of modern, pre-aged soil and fossil components are

168

constrained based on carbon isotopic compositions of source-specific biomarkers. In applying δ13C

169

values of source-specific biomarkers to trace the bulk OC, the range of lipid-to-biomass δ13C

170

offsets (δ13Cbulk – δ13Clipid) as a consequence of biosynthetic fractionation and related effects must


8

Page 9 of 34


171

be considered. Although the nature and potential variability of these offsets remain only partially

172

constrained, available studies suggest a range of 5-10‰ for algae and 7-10‰ for vascular plants,

173

and up to 7‰ for petrogenic materials, with lipid biomarkers showing greater 13C depletion.37-41

174

Because of the absence of data for δ13C offsets between biomarkers and bulk biomass for typical

175

source materials in the Yellow River basin, we adopt an offset of 5-7‰ to measured biomarker

176

δ13C values in order to derive the corresponding end-member bulk OC values, as previously

177

adopted for a prior Yellow River POC source apportionment study.26, 31 Detailed information on

178

the end-member value assignments are shown in the Results and Discussion section, as well as in

179

the Supporting Information (Table S1). We use a Monte Carlo (MC) simulation to minimize errors

180

from arbitrary assignments of end-member values.26, 42 The MC simulations were performed using

181

Matlab (version R2013a, Math Works, USA) based on script described by Yao, et al.,43 where

182

1,000,000 out of 100,000,000 random samples from the normal distribution of each end-member

183

within the given mean and standard deviation were taken to simultaneously fulfill the given system

184

(eqs 1–3). The mean relative contributions and the standard deviation of the three OC sources were

185

then calculated.

186

Results and Discussion

187

Spatiotemporal variations in sediment and bulk OC characteristics. Bulk geochemical

188

properties of Yellow River suspended samples are listed in Table 1. Yellow River suspended

189

sediments were characterized by relatively low POC values (POC%, ave., 0.31 ± 0.17%), which

190

are attributed to the dilution of OM by mineral materials from loess as well as limited aquatic

191

production due to high turbidity.10, 11, 20 The C/N ratios at HYK and KL were similar, and are also

192

in line with previously reported values for the Yellow River.10, 25 The δ13CPOC (ave., –24.5 ± 0.2‰)

193

and Δ14CPOC (ave., –401 ± 41‰) values at HYK were slightly lower and less variable than those


9


Page 10 of 34

194

at KL (ave., –24.0 ± 0.6‰ and –391 ± 52‰, respectively; Table 1). δ13CPOC values were within

195

those previously reported for the river basin and lower reach,20, 24-26, 44 and also close to the δ13C

196

values of modern soils from the loess plateau,44 which reflected C3 plant-dominant vegetation

197

sources. The slightly 13C-enriched values of POC relative to that of C3 plants (ave., –27.3‰, n =

198

39) may reflect some minor C4 plant OC inputs (ave., –12.6‰, n = 17), with the latter potentially

199

contributing about 20% OC in the upper and lower basin.44, 45 Δ14CPOC values of the lower Yellow

200

River (ave., –395 ± 46‰, ~4000 14C yr) are significantly lower than the global riverine median

201

Δ14CPOC value of –203‰ (~1800 14C yr),4 implying that the Yellow River mainly mobilizes and

202

exports the millennial-age OC to marginal seas.24, 26

203

Table 1. Sediment and POC characteristics at Huayuankou (HYK) and Kenli (KL) stations during

204

2015-2016

Site

Sampling date

Season

m3/s

℃

S

TSS

SA

Median grain size

POC

kg m-3

m2 g-1

μm

%

C/N ratio

OC/SA mg C m-2

δ13CPOC

Δ14CPOC

‰

‰

summer

1090

23.4

0.4

0.20

22.1

13.5

0.26

7.6

0.12

-24.8

-352±37

Nov 12, 2015

autumn

335

10.5

0.5

0.13

19.6

15.1

0.27

7.9

0.14

-24.5

-451±8

Jan 7, 2016

winter

505

5.1

0.5

0.17

18.7

14.7

0.34

6.6

0.18

-24.4

-390±7

Apr 24, 2016

spring

660

/

0.5

0.29

13.4

19.4

0.25

7.3

0.19

-24.3

-410±7

Average

Kenli (KL)

T

Jun 9, 2015

　

Huayuankou (HYK)

Daily discharge

　

/

/

/

0.20±0.07

18.4±3.7

15.7±2.6

0.28±0.04

7.3±0.5

0.16±0.03

-24.5±0.2

-401±41

Jun 4, 2015

early summer

796

23.8

0.5

1.03

13.7

23.7

0.16

9.3

0.12

-23.3

-412±28

Jul 9, 2015

WSR

2950

26.3

0.5

6.54

11.0

25.2

0.27

10.5

0.25

-23.9

-409±10

Aug 4, 2015

summer

560

28.0

0.4

0.68

19.0

18.5

0.29

8.5

0.15

-23.5

-351±8

Nov 24, 2015

autumn

123

3.3

0.5

0.19

13.1

28.8

0.12

4.7

0.09

-24.3

-429±76

Jan 21, 2016

winter

233

0.4

0.5

0.26

17.7

15.4

0.35

8.2

0.20

-24.3

-439±7

May 7, 2016

spring

86.5

20.2

0.5

0.18

30.1

7.9

0.75

5.8

0.25

-24.9

-304±7

/

/

/

1.48±2.50

17.4±6.9

19.9±7.6

0.32±0.23

7.8±2.2

0.18±0.07

-24.0±0.6

-391±52

Average

205 206

Although selected bulk OC characteristics (POC%, C/N ratio, δ13CPOC and Δ14CPOC) of HYK

207

and KL samples indicate some spatiotemporal variation, there are no statistically differences (t

208

test, p > 0.05). This implies that OC characteristics do not undergo significant alteration during


10

Page 11 of 34


209

long-range (~ 650 km) transport through the lower reach of the river (between HYK and KL).

210

Normalizing TOC concentrations to the mineral-specific SA (i.e., OC loading, OC/SA) reduces

211

the influence of physical sorting processes on OC characteristics during transport from the middle

212

to the lower river.3, 46 However, we find no differences in sediment characteristics (SA, median

213

grain size) and OC/SA values between HYK and KL samples (p > 0.05), further implying minimal

214

OC alteration during transport in the lower Yellow River. One likely reason for similar OC

215

loadings at both sites is the dominance of weathered loess-derived material, the latter also serving

216

as a likely explanation for the much lower OC/SA values in the Yellow River than those of other

217

rivers and sediments (0.4-1.1 mg C m-2).3 Furthermore, comparisons of KL sample bulk OC

218

(POC%, δ13CPOC and Δ14CPOC) and sediment characteristics (SA, median grain size and OC/SA)

219

in this study with previous samples of 2011–2013 (n = 12) 26 do not reveal differences (p > 0.05).

220

Taken together, these bulk-level observations suggest relatively consistent OC inputs and transport

221

dynamics in the lower Yellow River, irrespective of time or location. However, bulk level

222

measurements may mask source heterogeneity,47 and hence molecular carbon isotopic

223

measurements are used to provide additional information on specific source contributions to POC.

224

Heterogeneous OC sources constrained by molecular δ13C and Δ14C. The homologue

225

distribution of n-fatty acids (FAs) at HYK and KL was similar, with an even-carbon number

226

predominance and a bimodal pattern (Figure S1) that indicates mixed autochthonous and

227

allochthonous OC sources. Corresponding carbon isotopic compositions are characterized by

228

higher δ13C and Δ14C values of short-chain n-FAs than the longer-chain homologues at both sites

229

(Figures 2a, 2b). Short-chain (C16 and C18) n-FAs may derive from aquatic algae, microbial

230

activity, vascular plant leaf tissues and soil OM.48-50 The abundance-weighted average δ13C16+18FAs

231

and Δ14C16+18FAs values at HYK (–29.7 ± 1.0‰ and –43 ± 83‰, respectively) are similar to those


11


Page 12 of 34

232

at KL (–29.6 ± 1.2‰ and –42 ± 68‰). These relatively high average Δ14C16+18FAs values imply

233

rapid turnover times and an origin from recently fixed OC. Lower Δ14C16+18FAs values were

234

returned for samples collected in June at both HYK (−165‰) and KL (−144‰), as well as in May

235

at KL (−108‰). These values were even lower than those for corresponding fatty acid homologues

236

previously reported for samples collected between April and July (up to −83‰).26 Lower

237

Δ14C16+18FAs values in the warm season (May and June) may reflect an increased contribution from

238

aquatic production utilizing

239

October),51,

240

(Δ14CDOC at KL= –158 ~ –195‰).53 Irrespective of whether aquatic primary or secondary microbial

241

production is responsible, the carbon isotopic compositions of n-C16+18FAs appear to primarily

242

reflect contemporary/modern OC inputs to the Yellow River POC (Figure 3). For the longer-chain

243

homologues, n-C24FA was more 13C- and 14C-enriched than n-C26+28+30FAs, and similar to those of

244

short-chain n-FAs in some seasons (Figures 2a, 2b), suggesting that this homologue is not

245

exclusively derived from vascular plant waxes.48, 54 With respect to the latter, focus is therefore

246

placed on long-chain (≥ C26) n-FAs. The low δ13C26+28+30FAs (–32.6 ± 0.7‰) and Δ14C26+28+30FAs (–

247

198 ± 15‰) values observed at HYK indicate inputs of pre-aged higher plant-derived OC,

248

consistent with those at KL (–33.0 ± 0.6‰ and –219 ± 33‰). Corresponding 14C ages (1470-2300

249

14C

250

river. This process has been previously observed in rivers, marine sediments as well as continental

251

aerosols.55-59 In contrast to the much younger 14C age of lignin phenols reported at KL, which are

252

also established terrestrial plant biomarkers and primarily delivered to river via surficial erosion

253

of soil and fresh plant detritus, the pre-aged OC traced by older 14C ages of long-chain n-FAs is

254

inferred to reflect inputs from erosion of deeper soil.26, 56 Furthermore, the lower variability in

52

14C-depleted

DIC (Δ14CDIC

at KL:

–164‰ in April and –130‰ in

or contributions from heterotrophic organisms which could utilize pre-aged DOC

yr) of n-C26+28+30FAs imply storage in intermediate reservoirs (e.g., soil) prior to its delivery to


12

Page 13 of 34


255

carbon isotopes values of n-C26+28+30FAs than n-C16+18FAs implies a more constant contribution of

256

pre-aged soil OC.

257 258

Figure 2. Compound-specific δ13C and Δ14C values of HYK (blue) and KL (red) samples. (a) δ13C

259

and (b) Δ14C compositions of n-FAs; (c) δ13C and (d) Δ14C compositions of n-alkanes. The

260

horizontal line and small square inside each box represent median and mean values, respectively.

261

The δ13C29+31Alkanes values at HYK (–32.2 ± 0.5‰) and KL (–31.8 ± 0.3‰) also imply an origin

262

predominantly from C3 plants,37 while their corresponding

263

2190-4920 yr; KL, 3190-5620 yr) also indicate millennial-scale storage of this terrestrial OC pool

264

in soils before fluvial transport. However, Δ14C29+31Alkanes values are lower than corresponding

14C

ages of n-C29+31 alkanes (HYK,


13


Page 14 of 34

265

Δ14C26+28+30FAs values at both HYK and KL, which may reflect different terrestrial OC storage and

266

mobilization behavior, and/or fossil hydrocarbon contributions from petrogenic or anthropogenic

267

sources.36, 60-62 Shorter-chain odd-carbon-numbered Δ14C25+27Alkanes (–475 ± 126‰) values at KL

268

were

269

(Δ14C26+28+30+32Alkanes, –774 ± 144‰) were most strongly depleted in

270

(Figures 2c, 2d). The latter values suggest a predominantly fossil OC input to C26+28+30+32Alkanes and

271

to a lesser extent to C25+27Alkanes. The Δ14C17+19+21+23Alkanes value (–913‰) for one winter sample at

272

HYK was much more depleted than that of C29+31Alkanes (–373 ± 106‰), suggesting almost

273

exclusive fossil sources. The molecular 13C and 14C compositions of alkanes thus reveal at least

274

two older OC sources: pre-aged soil OC and fossil OC sources, the latter likely eroded from

275

sedimentary rock in the watershed.63

lower

than

Δ14C29+31Alkanes,

while

those

of

even-carbon-numbered 14C,

n-alkanes

and enriched in

13C

276

Taken together, the average molecular δ13C and Δ14C values of n-FAs and n-alkanes reveal

277

different sources of Yellow River OC that are heterogeneous with respect to age:

278

contemporary/modern OC derived from fresh higher plant detritus and autochthonous OC, pre-

279

aged mineral-bound soil OC with millennial-scale turnover times emanating from erosion of

280

deeper soil horizons, and fossil OC mainly originating from bedrock erosion (Figure 3). The

281

molecular isotopic characteristics of the HYK and KL samples investigated here, together with

282

previously reported Δ14CFAs data for KL samples26 reveal only minor spatiotemporal variations (p

283

> 0.05, Figure 2). We do note, however, that δ13CFAs values in this study were 1.7-2.2‰ lower

284

than previously reported data from 2011–2013, potentially reflecting variations in the contribution

285

of C3 plant-derived soil OC to the long-chain n-FAs, and fresh C3 plant, and planktonic OC to the

286

short-chain n-FAs examined in this study (Figure 3).


14

Page 15 of 34


287 288

Figure 3. Cross-plot of average δ13C versus Δ14C values of different organic components for POC

289

from the lower Yellow River and other sources. Different colors represent different data sets

290

including KL (red) and HYK (blue) from this study, as well as previously reported KL data of

291

2011–2013 (purple).26 The δ13C and Δ14C values of source-specific biomarkers correspond to

292

abundance-weighted values of combined compounds, and δ13C values of source-specific

293

biomarkers were corrected by +6‰ in order to compensate for δ13Cbulk δ13Clipid fractionation.26, 31

294

δ13C values of bulk C3 and C4 plant tissue samples and rock OC from Chinese Loess Plateau are

295

also shown.64, 65 δ13C and Δ14C values of aquatic phytoplankton were deduced from DIC values

296

of the Yellow River with consideration of

297

sources and ancient sedimentary rocks were assigned as 100 ± 50‰ and -1000‰ respectively.

13C

fractionation.51, 66 Δ14C values of modern plant


15


Page 16 of 34

298

Interannual variations of OC sources and fluxes. For quantitative apportionment of different

299

aged OC contribution to the Yellow River POC pool, we constrain end-member values using δ13C

300

and Δ14C values of corresponding source-specific biomarkers in each sample (applying a +5 to

301

7‰ correction for δ13Cbulk δ13Clipid fractionation). All possible solutions derived from MC

302

simulations of the three end-member mixing model are shown in Table S1 and Figure 4.

303

During the June 2015–May 2016 period of this study, fM, fS and fF were found to range between

304

22-59% (ave., 40 ± 14%), 20-43% (ave., 30 ± 9%) and 19-35% (ave., 30 ± 6%), respectively. On

305

average, the modern OC contribution was the highest over this period, however strong intra-annual

306

seasonal variations are evident, particularly for fM and fS (Figure 4a). These findings contrast

307

sharply with a prior investigation of the 2011–2013 period, during which fM, fS and fF were

308

respectively 13-22% (ave., 18 ± 3%), 44-57% (ave., 51 ± 3%) and 29-35% (ave., 32 ± 2%). In

309

addition, there was substantially less intra-annual variability, with pre-aged OC predominating

310

throughout.26 These characteristics were inferred to reflect the POC sources dominated by deeper

311

(soil) erosion processes.26 The new measurements spanning 2015–2016 also revealed larger

312

seasonal variations of OC composition, with higher fM (the dominant fraction) occurring in spring

313

and summer (45-59%, June and August 2015, May 2016) and the lowest fM value (22%) in winter

314

(January 2016). Higher fM in spring and summer may because of increasing contributions from

315

aquatic primary production or from enhanced surface runoff entraining fresh terrestrial OC during

316

wet seasons. Specifically, the higher fM (50%) in August in concert with higher Δ14C16+18FAs values

317

than those in spring sample suggests increasing inputs of fresh higher plant detritus that contributed

318

to a relatively high Δ14CPOC value. In contrast, the highest fM (59%) that is accompanied by the

319

highest POC% and Δ14CPOC values in May 2016 is likely a consequence of increased aquatic

320

primary production because the lower turbidity (lowest water discharge and TSS) during this El


16

Page 17 of 34


321

Niño year and suitable water temperature in spring. A higher aquatic contribution was also inferred

322

for April 2013, but in this case contemporary OC only accounted for a relatively minor proportion

323

(13-22%) of POC.26 The decrease in fM and increase in fS during WSR is attributed to

324

remobilization of riverbed sediments which increases turbidity and inhibits aquatic primary

325

productivity. The lowest fM and highest fS values both occurred in winter (January), likely due to

326

enhanced erosion of deeper soil OC, as well as diminished inputs of fresh terrestrial OC from

327

surface runoff and lower aquatic contribution because of low water temperature (Table 1).

328

The marked difference in POC age and compositional heterogeneity between the 2011–2013

329

and 2015–2016 sampling intervals is likely caused by the sharply contrasting hydrological

330

conditions. Water discharge was higher and punctuated by high-frequency floods in 2011–2013,

331

whereas relatively arid conditions and hence muted hydrological patterns prevailed in 2015–2016

332

(Yellow River Sediment Bulletin, 2015, 2016, http://www.yellowriver.gov.cn/nishagonggao/) as

333

a consequence of the influence of El Niño,27 when higher temperatures and lower precipitation

334

resulted in a dramatic decrease of annual water discharge and suspended sediment flux. The annual

335

water discharge and suspended sediment flux between June 2015 and May 2016 (91.3  108 m3

336

and 22.1  106 t/yr, respectively) corresponds to only 33-40% and 12-22% of the values during the

337

June 2011–May 2013 period (Table S2). Lower suspended sediment concentrations likely

338

facilitated increased aquatic primary production, thereby contributing to higher fM values in 2015–

339

2016 (Figure S2a), due to alleviation of light limitation in the normally highly turbid Yellow River.

340

11, 20

341

reduced resuspension of riverbed sediment would also lead to diminished inputs of aged OC in the

342

2015–2016 period. Thus, climate-driven differences in hydrological conditions between 2015-

Decreased delivery of sediments from the upper and middle reaches of the river as well as


17


Page 18 of 34

343

2016 and 2011-2013 translate to marked changes in the composition of POC in downstream

344

sections of the Yellow River.

345

With respect to anthropogenic influences, WSR practices are known to exert strong control on

346

sediment transport dynamics, and on OM and nutrient fluxes to the sea.11, 20-22 Under lower flow

347

conditions arising from the 2015–2016 El Niño period, changes in OC composition and flux during

348

human-induced WSR are likely to be more pronounced. Increased riverbed scouring during WSR

349

would result in higher TSS concentration in the lower reaches, and this process results in decreased

350

fM and increased fS compared with samples from periods both immediately before (June) and

351

afterwards (August, Figure 4a), albeit without significant influence on bulk 14C age (Figure S2b).

352

The fluxes of the three different OC components each showed significant positive correlations

353

with monthly discharge (Figure S3). By extrapolation, we calculated that the Yellow River

354

delivered 1.89  104 t/yr, 1.95  104 t/yr and 1.94  104 t/yr of modern/contemporary, pre-aged

355

soil and fossil OC, respectively, to the China marginal seas during the period of June 2015–May

356

2016 under the influence of El Niño. These fluxes correspond to only 19-31%, 7-11% and 11-17%

357

of corresponding fluxes in the June 2011–May 2013 period. The corresponding total POC flux of

358

5.78  104 t/yr we derive for the 2015–2016 hydrological year (June 2015–May 2016) is similar

359

to that estimated by Xue et al.,24 for 2015 (4.12  104 t/yr), but represents only 8-17% (ave., 12 ±

360

4%) of that estimated for the 2008–2013 period (Table S2). Because of the low discharge and

361

suspended sediment load during the 2015–2016 sampling period, water discharge and TSS during

362

the month of July (WSR period) account for 35% and 79% of annual discharge and TSS. However,

363

the WSR period contributed 82%, 76%, 85% and 84% of annual POC, modern, pre-aged soil and

364

fossil OC fluxes, significantly higher than previously reported values (28-50% of annual POC flux

365

during 2012–2013).20, 22 As a result, the annual flux-averaged fM, fS and fF values (33%, 34% and


18

Page 19 of 34


366

33%) were similar to those of the sample collected during the WSR interval (31%, 35% and 34%).

367

These results thus suggest that both natural and human-dominated hydrological events exert strong

368

influence on the POC composition and fluxes of each component.

369 370

Figure 4. The variation of fM, fS, fF and monthly TSS flux during 2015–2016 at (a) KL and (b)

371

HYK sites and comparisons with reported results during 2011–2013.26 The asterisk represents

372

samples collected during WSR period.


19


Page 20 of 34

373

Spatial variations of OC sources in the lower Yellow River. Although sample numbers are

374

limited, this study for the first time examined the seasonal variations in carbon isotopic

375

characteristics and composition of OC at HYK station, which lies at the transition between the

376

middle and lower reach of the river, and ~120 km downstream of the Xiaolangdi Reservoir. The

377

range of fM, fS and fF values at HYK during June 2015–May 2016 was 20-46% (ave., 28 ± 12%),

378

33-50% (ave., 42 ± 8%) and 21-37% (ave., 30 ± 7%), respectively (Figure 4b), with the three OC

379

components exhibiting smaller variability relative to that at KL. The highest fM (46%) also

380

occurred in June, coupled with the lowest Δ14C16+18FAs value, suggesting greater contribution from

381

aquatic primary production during the warm season. However, in other seasons, pre-aged OC

382

predominated, followed by fossil OC, suggesting larger OC contributions from deeper soils and

383

ancient sedimentary rock erosion. The highest fS and lowest fM occurred in winter (January 2016),

384

similar to results at KL, suggesting enhanced contributions from erosion of deep soils and

385

proportionally less input from modern sources of OC. Seasonal patterns for HYK and KL are

386

similar. For example, fM is highest in warm seasons, fS is highest in winter (January), and fS and fF

387

are similar in autumn (November). This coherence implies that changes in upstream supply of

388

suspended particulate matter influences downstream temporal variability in OC. However, side-

389

by-side comparison of the results from corresponding sampling intervals reveals that the

390

proportions of pre-aged OC at HYK are systematically higher than those at KL, indicating that

391

additional processes must modify OC composition during transit between the middle to the lower

392

river. The predominance of pre-aged OC at HYK echoes the composition of POC at KL station

393

during the 2011–2013 period. This suggests that during relatively high discharge, as encountered

394

during 2011–2013, there is more efficient transport of sediment characterized by pre-aged OC

395

from upper to lower reaches of the river. The OC composition at HYK is thus more strongly


20

Page 21 of 34


396

influenced by the POC composition from the upper reach of Yellow River, consistent with a

397

slightly lower Δ14CPOC value reported for Xiaolangdi Reservoir than at HYK.24 Particle settling

398

occurs during transport, increasing riverbank height and sedimentation in the lower reach.

399

Although there is no significant input of particles from surrounding areas from HYK due to the

400

raised nature of the river channel, multiple source inputs of OC with different age could occur at

401

lower site near river estuary. Riverbed sediments and associated OC are prone to remobilization,

402

particularly during WSR, which could increase pre-aged OC inputs, and other contributions could

403

be augmented by that draining wetland, agricultural fields on the narrow riverbank and coastal

404

plain as well as direct anthropogenic inputs such as nutrients from wastewater discharge or

405

fertilizer near river estuary,45, 67 which may lead to different OC sources at KL than HYK.

406

Although a previous study reported that proportions of aged OC could increase during the

407

resuspension and deposition cycles along transport pathways in coastal environments on passive

408

continental margins,3,

409

middle (HYK) to the lower (KL) Yellow River. However, the inherently heterogeneous

410

composition of POC was apparent from dual carbon isotopic measurements, enabling

411

disentanglement of OC sources within bulk OM pools, and illuminating spatiotemporal variations

412

of riverine OC characteristics and sources.

68

our study did not reveal substantial differences in POC age from the

413

Implications for marginal sea OC burial. The China marginal seas (Bohai Sea, Yellow Sea

414

and East China Sea) bury 13.0 Mt/yr of OC,69, 70 accounting for about 10% of global shelf sea OC

415

burial.71 A recent study revealed that, on a 100-year time scale, the Bohai and Yellow Sea sediment

416

OC comprised of 24-49% pre-aged OC and 7-26% fossil OC, that majority of which was derived

417

from the Yellow River.23 The high flux of non-modern OC from the Yellow River contribution to

418

the high burial efficiency of terrestrial OC in adjacent marginal seas,23 and exerts important


21


Page 22 of 34

419

influence on the coastal carbon cycle. In the present study, we found that lower discharge

420

stemming from drier conditions results in a higher proportion of modern OC inputs, and this

421

difference in the nature of exported OC is likely to influence its subsequent biogeochemical

422

cycling in estuarine and coastal ecosystems. In contrast, human-induced flooding (i.e., WSR)

423

during drier conditions could enhance export of pre-aged OC from the river, with this short period

424

of human intervention dominating annual OC fluxes to coastal seas.

425

As a consequence of climate change and direct human interventions, sediment and water

426

discharge from the Yellow River to the coastal ocean has dramatically decreased over the past 60

427

years, and this trend is likely to continue.9, 15 This is also the case for POC export during recent

428

years (2008–2013) compared to 1950s levels, which showed a sharp drop (Table S2). The distinct

429

climate conditions encompassed during this study (2015–2016), which overlapped with an El Niño

430

period, resulted in a decrease in POC flux to the sea by up to 90% compared with those of 2011-

431

2013, and in a ~90% decrease in corresponding pre-aged and fossil OC fluxes. We conclude,

432

therefore, that both natural (climate-driven) and human-made perturbations (i.e., WSR flooding

433

event) which modulate hydrological variability on both inter- and intra-annual timescales exert

434

strong influence on the sources, compositions and fluxes of OC exported by the Yellow River.

435

This variability in turn is likely to influence the balance of oxidation versus burial of OC in the

436

adjacent China marginal seas. The limited observations currently available render it challenging

437

to disentangle those impacts due to climate variability versus human activities. We note that, on

438

average, 60% of OC delivered by the Yellow River to China marginal seas is millennial in age,

439

and other large rivers have been reported to export suspended sediments dominated by pre-aged

440

OC.59, 72 While burial of this type of OC would exert little influence on short-term (

Impacts of natural and human-induced hydrological variability on

Recommend Documents