A 400-year record of atmospheric mercury from tree-rings in

Subscriber access provided by - Access paid by the | UCSB Libraries

Characterization of Natural and Affected Environments

A 400-year record of atmospheric mercury from tree-rings in northwestern Canada Sydney Clackett, Trevor Porter, and Igor Lehnherr Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.8b01824 • Publication Date (Web): 02 Aug 2018 Downloaded from http://pubs.acs.org on August 3, 2018

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 31

Environmental Science & Technology

TOC art 47x26mm (300 x 300 DPI)

ACS Paragon Plus Environment


1

A 400-year record of atmospheric mercury from tree-rings in northwestern Canada

2

Sydney P. Clackett1, Trevor J. Porter1,*, and Igor Lehnherr1

3 4 5

1

Department of Geography, University of Toronto, Erindale Campus, Mississauga, Canada, L5L 1C6 *Corresponding author. Tel. +1-905-828-5314; [email protected]

6 7 8 9

Abstract Tree-rings are a promising high-resolution archive for gaseous atmospheric mercury (comprised primarily of Hg0) reconstruction, but the influence of cambial age (ring number from

10

pith) and tree-specific differences are uncertainties with potential implications for interpreting

11

tree-ring Hg signals. We address these uncertainties and reconstruct the last 400 years of Hg0

12

change using a tree-ring Hg dataset from 20 white spruce (Picea glauca) trees from a pristine site

13

in central Yukon. Cambial age has no significant influence on tree-ring Hg concentration, but

14

tree-specific differences in mean concentration are prevalent and must be normalized to a

15

common mean to accurately constrain long-term trends in the mean tree-ring Hg record. Our

16

record shows stable, low Hg0 concentrations prior to ~1750 CE, a persistent rise from ~1750-

17

1950 (increasing more rapidly post-1850), a pause from ~1951-1975, and then a resumed

18

increase to record-high levels at present. This general pattern is reflected in other proxy-based

19

Hg reconstructions worldwide. This study provides a novel long-term Hg0 reconstruction in the

20

Western subarctic from one of the most widely distributed boreal tree species in North America

21

and, therefore this proxy may also hold potential for investigating broader spatial patterns in Hg0

22

cycling across the subarctic and northern boreal forest.

23

1 ACS Paragon Plus Environment

Page 2 of 31

Page 3 of 31

24

25


1.0 INTRODUCTION Gaseous elemental mercury (Hg0) has a long atmospheric residence time which allows

26

for long-range dispersal of natural and anthropogenic Hg emissions to remote and otherwise

27

pristine areas on Earth.1 Eventual deposition of Hg in terrestrial and aquatic ecosystems, and

28

conversion to toxic forms (e.g., methylmercury), is a major environmental and human health

29

concern.1 The United Nation’s Minamata Convention on Mercury came into force in 2017 to

30

address Hg pollution, leading to international efforts to monitor and mitigate atmospheric Hg

31

emissions. However, monitoring of Hg in the environment only began ~40 years ago,2 well after

32

the rise of anthropogenic industrial emissions. Natural archives for deposited Hg such as ice

33

cores,3,4 peat cores5,6 and sediment cores6–8 offer a longer-term perspective on changes in Hg

34

cycling from the pre-industrial era to present. Such a long-term perspective is needed to evaluate

35

the efficacy of international mitigation efforts. All proxies have unique strengths and

36

uncertainties, for example, related to chronology6 and non-atmospheric Hg inputs.9 However,

37

conventional wisdom from the paleo-sciences dictates that knowledge of past environments can

38

be enhanced through replication and a multi-proxy approach.10

39

Recent studies suggest that annually resolved tree-rings are a promising archive for Hg0

40

at the time of assimilation.11–17 A combination of field- and growth-chamber studies reveal that

41

tree-ring Hg is derived primarily from gaseous atmospheric Hg0, via air-to-foliage (stomata and

42

cuticle) pathways and phloem translocation to where Hg is assimilated into woody tissue.13,14,16,18

43

Neither uptake of soil Hg through the roots, or exposure to different gaseous oxidized Hg

44

compounds appears to significantly influence tree-ring Hg concentrations.18 Tree-rings have

45

several advantages including the broad distribution of forests globally, potential for centennial-

46

to millennial-length chronologies, annual resolution, and absolute dating control (annual ring 2 ACS Paragon Plus Environment


47

counting and cross-dating verification) that does not rely on radiometric chronometers. The

48

advantage of absolute dating in tree-rings cannot be overstated, since radiometric (e.g., 14C)

49

chronology uncertainties for Late Holocene sedimentary records are commonly > 50 years,

50

which has major implications for constraining the timing of Hg trends and calculated fluxes.

51

As Hg dendrochemistry is a relatively young field of research, there are few long tree-

52

ring Hg records worldwide. Most applied tree-ring Hg studies have focused on recent (20th

53

century) local industrial signals in the tree-ring Hg,12,14,17 which provide important proof-of-

54

concept that tree-rings reliably track local atmospheric Hg0 variability. However, virtually all

55

tree-ring Hg studies are based on small sample sizes (e.g., < 5 trees)11,12,14,19 or pooled, multi-tree

56

records,16,20 which has precluded assessments of the influence of ‘internal factors’ such as

57

cambial age (ring number since the pith, or first year of growth) and tree-specific differences on

58

mean tree-ring Hg concentrations. As tree-specific and age-related factors are known to

59

influence other physical tree-ring variables (e.g., ring-width, density and 13C/12C),21–25 recent

60

studies have suggested the need to understand the potential influence of internal factors on the

61

long-term trends in tree-ring Hg.16

62

In this study, we address these key uncertainties using a well-replicated dataset of tree-

63

ring Hg developed from 20 mature white spruce (Picea glauca) trees from a pristine boreal site

64

in central Yukon, northern Canada, ~140 km south of the Arctic Circle. We then use this dataset

65

to estimate the common Hg signal, and use this record to infer the last ~400 years of Hg0 change

66

in this region.

67

68


Page 4 of 31

Page 5 of 31


69

2.0 METHODS

70

2.1 Field sampling and cross-dating

71

Fifty-five mature white spruce trees (living and dead) were sampled in 2015 and 2016

72

from a woodland site we call Scree Hill in central Yukon (65.071°N, 138.157°W) (Fig. 1). All

73

55 trees were included in our cross-dating analysis, and a subset of 20 trees was selected for tree-

74

ring Hg analysis (Section 2.2). Bark-to-bark cores (i.e., dual radii) were collected from living

75

trees (n = 28) perpendicular to the trunk at ~1 m height using 5.1 mm or 10 mm diameter Haglof

76

increment borers (10 mm cores were collected for Hg analysis). The cores passed through or near

77

the pith for most of the sampled trees. Disks were cut by chainsaw from dead trees (n = 27), also

78

at ~1 m height or higher if rot was encountered lower in the stem.

79

Samples were air dried for several weeks before further preparation. Cores were fixed to

80

mounts with a small bead of wood glue and then lightly sanded with increasingly finer sandpaper

81

(220 to 400 grit) to polish the surface for tree-ring measurement and dating. For living samples,

82

calendar years were assigned to each ring by counting back from the outermost growth ring (i.e.,

83

year of sampling) to the pith or the innermost growth ring. Ring-width measurements (Velmex

84

measuring system, ± 0.001mm precision) were collected for the core samples, and the ring-width

85

patterns were cross-dated using the software program COFECHA26 to verify the age of each ring

86

and ensure there are no locally absent or false rings. Ring-width series from deadwood samples

87

were matched to the ‘master’ living tree chronology, also using COFECHA, to determine their

88

positions in time, and then cross-dated against all other samples. The principle of cross-dating

89

allows for the detection of misaligned ring-width series caused by random growth anomalies

90

expressed in some trees (e.g., locally absent or false rings), and is the basis for absolute,



91

calendar year tree-ring chronologies that underpin parts of the 14C timescale27. The Scree Hill

92

ring-width chronology benefits from high replication (n = 55 trees) and the individual ring-width

93

series that define it show no evidence of misaligned series based on the cross-dating analysis (see

94

SI – COFECHA report). Therefore, we assume the Scree Hill tree-ring chronology is absolute

95

and free of dating errors. White spruce is regarded as a species of high importance in

96

dendrochronology, owing to its tendency to cross-date strongly and unambiguously across local

97

to regional scales28.

98

2.2 Sub-sampling for Hg analysis

99

The 20 trees selected for Hg analysis include 11 living trees and 9 dead trees. The bark-

100

to-bark core samples (10 mm bore diameter) from living trees capture ‘duplicate’ pith-to-bark

101

sections from opposite sides of the tree. Only one of the two pith-to-bark section was analyzed

102

for Hg. The disks from dead trees were cut into ~2 cm wide cross-sections using a band saw, to

103

facilitate dissection of the rings. Similar to the core samples, only one pith-to-bark section from

104

the dead wood samples was analyzed for Hg (see next paragraph for sub-sampling details). We

105

avoided samples with distorted growth rings due to branch intrusions or reaction wood. We also

106

made an effort to include samples with a wide range of initial growth dates to avoid over-

107

representation of trees that began growing during a particular period of time when atmospheric

108

Hg0 concentrations may have been naturally high or low, which could result in a ‘temporal bias’

109

when characterizing cambial age-related trends in tree-ring Hg. Initial growth rings of the

110

selected samples are relatively evenly distributed across a range of calendar years from 1605-

111

1860 CE (Common Era), and with cambial ages that range from 135-385 years. For core samples

112

that did not pass through the pith, a tree-ring image analysis program CooRecorder (Cybis


Page 6 of 31

Page 7 of 31


113

Elektronik) was used to estimate the number of missing growth rings by concentric-ring curve fit

114

in order to assign the most accurate cambial age possible to each sample.

115

The core and disk tree-ring series selected for Hg analysis were sub-sampled into 5-year

116

blocks (i.e., 5 annual rings per block) corresponding with the first and last halves of the

117

Gregorian calendar decades (e.g., 1901-1905, 1906-1910, etc.). The goal of this strategy was to

118

optimize inter-tree replication when organizing the individual tree-ring Hg series by calendar

119

year. This strategy also meant that the earliest block of rings for some trees did not always

120

include 5 rings; in all but one case the first block of rings included 3 or more rings (4.25 on

121

average). A clean scalpel blade was used to dissect the 5-year blocks, and shave the outer surface

122

to remove potential contamination since the time of collection.

123

2.3 Mercury Analysis

124

Samples were oven dried at 60°C for 24 hr to remove adsorbed and interstitial water.

125

Twenty milligrams of dry wood was sub-sampled from each 5-year block with a clean scalpel,

126

cut perpendicular to the ring boundary to maintain the natural weighting of rings in the 5-year

127

block. Blocks that weighed less than 20 mg were not analysed. Total Hg concentrations were

128

quantified with a Milestone tri-cell Direct Mercury Analyzer (DMA-80) by atomic absorption

129

following thermal decomposition and amalgam pre-concentration. Our quadratic calibration

130

curve was constrained by a 10-point dilution series of a certified aqueous Hg standard (High

131

Purity Standards 1000 µg mL-1 Hg, 2% HCl), bracketing Hg amounts in the tree-ring samples.

132

The detection limit, calculated as 3× the standard deviation of the blank, is 0.0042 ng Hg, which

133

corresponds to a concentration of 0.21 ng g-1 since the samples analyzed in this study were

134

weighed to 20 mg (dry weight). Instrument accuracy was assessed daily using the NIST 1575a



135

Pine Needle Standard, for which we obtained a mean concentration of 40.0 ± 0.6 ng g-1 (1.6%

136

RSD, N = 38), which is consistent with the certified value (39.9 ± 0.7 ng g-1). Blanks, sample

137

duplicates (1σ = 0.1 ng g-1), and an internal powdered wood standard (4.7 ± 0.2 ng g-1, 4.8%

138

RSD, N = 84) were analyzed before every 9th sample for quality assurance and to monitor

139

instrument stability and precision. The DMA-80 program was as follows: 250°C maximum

140

starting temperature; 200°C drying temperature; 60 s drying time; 650°C decomposition

141

temperature; 180 s decomposition time; and 60 s dwell time.

142

143

3.0 RESULTS AND DISCUSSION

144

3.1 Cambial age-related trends

145

The influence of a tree’s cambial age on ring-width and other tree-ring variables (e.g.,

146

δ13C and maximum ring-density) is well described in the literature.22,25,29 However, to date, there

147

are no published studies that have investigated cambial age-related effects in tree-ring Hg;

148

although previous studies have highlighted the need for such investigations to better understand

149

the potential implications for interpreting environmental signals in long tree-ring Hg series.16 of

150

the focus of this section is our analysis of the relation between cambial age and tree-ring Hg.

151

However, we begin with a brief discussion of the influence of cambial age on the traditional

152

ring-width variable from our sample trees.

153

A negative exponential trend was observed in mean ring-width when all series were

154

aligned by cambial age (Fig. 2a). Large ring-widths (~0.57 mm) are typical in the first years of

155

growth, followed by an initially rapid decrease between years 1-101 (mean ring-width = ~0.24

156

mm by age 101), a smaller rate of change between years 101-201 (mean ring-width = ~0.17 mm 7 ACS Paragon Plus Environment

Page 8 of 31

Page 9 of 31


157

by year 201), and no significant change thereafter (trends beyond year 270 are not considered

158

due to low replication). Negative exponential trends are common in open-canopy boreal forests,

159

owing to a general lack of within-stand disturbance (e.g., gap dynamics) and low competition for

160

resources during the juvenile growth years, and subsequently slower rates of growth as the tree

161

approaches a maximum size in later years.25,30 Paleo-environmental signals are encoded in ring-

162

width series, but are secondary to the cambial age-related trend in terms of variance, which is

163

why raw ring-width data are usually ‘detrended’ prior to climate-proxy analysis and paleoclimate

164

reconstruction.

165

We also evaluated possible cambial age-related trends in the average tree-ring Hg series,

166

but restrict our analysis to cambial ages defined by three or more trees, and we use the robust bi-

167

weight mean to estimate the average tree-ring Hg record since it is insensitive to outliers.31 We

168

calculated two cambial age-aligned mean Hg records (Fig. 2), one based on all trees and another

169

that excludes post-1850 CE tree-rings to avoid the confounding impact of the rapid rise in global

170

atmospheric Hg0 concentrations since industrialization,32 which could dominate variability in the

171

cambial age-aligned mean record and be mistaken as an age-related trend. When post-1850 CE

172

tree-rings are excluded (Fig. 2b, red line), a slightly negative slope (-0.005 ng g-1 decade-1) is

173

observed in the mean Hg concentration with increasing cambial age (cambial ages 1-216) but

174

which is not significantly different from a flat line (p = 0.87). However, if post-1850 CE data are

175

included (Fig. 2b, black line), a persistent rise in mean Hg concentration is observed between

176

cambial ages 1-281 at a rate of +0.032 ng g-1 decade-1 (p = 5.6 ×10-21). This comparison

177

demonstrates that industrial-era tree-rings strongly bias the result, probably due to a higher Hg0

178

background in the industrial era,32,33 and should be excluded from cambial age-trend analysis if

179

the objective is to isolate Hg trends solely related to cambial age, but we also recognize that



180

doing so reduces the total sample size and range of cambial ages that can be effectively

181

characterized (Fig. 2c).

182

We conclude that tree-ring Hg concentrations are not significantly influenced by a tree’s

183

cambial age and, thus, tree-ring Hg series do not require the same adaptive curve-fitting

184

standardization procedures (with potential low-frequency information loss; Cook et al.34) that are

185

typically applied to traditional tree-ring variables prior to paleoenvironmental analysis. This

186

finding holds when other transformations of our dataset are used, including ‘bias-adjusted’ series

187

as we discuss in Section 3.2. However, this being the first study of age-related trends in tree-ring

188

Hg, additional studies from other sites and species are needed to verify this result. Ideally future

189

studies that investigate cambial age effects on tree-ring Hg will include samples from a wider

190

range of pre-historic intervals to better control for period-specific variability in atmospheric Hg0.

191

The northern boreal forest in particular has great potential for such investigations, as

192

demonstrated by millennial-length tree-ring records reported in nearby Alaska35 and Northwest

193

Territories.36

194

3.2 Inter-tree variability

195

Both systematic differences in mean Hg concentration and random variability (‘noise’)

196

are apparent in the individual tree-ring Hg series. We begin this discussion by focusing on the

197

systematic differences in mean concentration that are evident in some of the individual series

198

(Fig. 3). For example, tree-ring Hg concentrations of tree SH1-15-01a are on average 0.64 ng g-1

199

higher than the mean record (‘site average’) defined by all other trees, and tree-ring Hg

200

concentrations of tree SH1-16-01b are on average 0.34 ng g-1 lower than the site average (Fig. 3).

201

Mean differences between individual Hg series and the site average are small in absolute terms

202

(ranging from 0.01 to 0.64 ng g-1), but have implications for calculating a site averaged tree-ring 9 ACS Paragon Plus Environment

Page 10 of 31

Page 11 of 31


203

record that most accurately depicts the common, long-term Hg trend shared by the sample

204

population. These implications are discussed later in this section, but first we discuss possible

205

reasons for tree-specific differences in mean tree-ring Hg concentration.

206

Tree-specific differences in mean tree-ring Hg concentration (hereafter ‘tree-specific

207

bias’) implies that tree-specific factors may influence tree-ring Hg concentrations, although the

208

source of tree-specific bias is unknown. Possible factors might include differences in stomatal

209

conductance or in phenology (timing and duration) of tree-ring growth. Growth chamber

210

experiments of trees growing under different relative humidity treatments (to induce different

211

stomatal conductance) suggest that higher stomatal conductance results in higher Hg0 diffusion

212

rates into leaves, and higher tree-ring Hg concentrations.18 Differences in stomatal conductance

213

between trees under the same ambient conditions can be explained by stomatal phenotypes (e.g.,

214

density and aperture)18 or localized soil moisture variability which can elicit a stomatal response

215

when moisture is limited.37,38 Differences in soil moisture availability across a boreal site could

216

be influenced by heterogeneous drainage due to complex microtopography. The Scree Hill site

217

has a gentle slope, but with complex microtopography owing to the underlying parent material

218

(frost shattered rock and colluvium from the adjacent slope), frost mounds and hollows, and a

219

spatially ubiquitous moss cover of variable thickness that insulates the ground. Site-level

220

heterogeneity of soil moisture in nearby white spruce boreal woodlands is thought to be a major

221

factor driving variable growth responses between trees at the same site.36 The phenology of tree-

222

ring growth can also vary between trees due to variable moss thickness, which influences the

223

onset and rate of springtime ground thaw and, therefore, availability of liquid water to plants.

224

Dendrometer monitoring of active white spruce growth in the nearby Mackenzie Delta region

225

confirms that the onset, intensity and duration of active growth varies between trees at the same



226

site.39 Since atmospheric Hg0 concentration is known to vary seasonally (highest in the

227

winter/spring and lowest in the late summer),40–42 it is plausible that slight phenological

228

differences in active tree growth could lead to differences in mean tree-ring Hg concentration. If

229

this mechanism is important, trees that grow proportionally more in the late summer (when

230

atmospheric Hg0 concentration is lowest) would have lower mean tree-ring Hg concentrations.

231

However, these possibilities remain speculative and future studies that control for these

232

uncertainties are needed to advance knowledge of tree-specific bias in tree-ring Hg records.

233

Systematic differences in mean tree-ring Hg concentration have potential implications for

234

reconstructing long-term trends in atmospheric Hg0 based on a composite (site average) record of

235

tree-ring Hg series from multiple individual trees. In cases where all sample trees contribute

236

equally to each time step of the mean tree-ring Hg record (i.e., perfect temporal overlap of all

237

contributing trees), then tree-specific bias should not have a disproportionate influence on overall

238

trends in the average Hg record. But in cases where the mean record is the product of individual

239

tree-ring series with unequal temporal coverage (as is true for the Scree Hill dataset), trends in

240

the mean Hg concentration record can be skewed by tree-specific bias as the number of

241

individual Hg series contributing to the mean record changes through time (see SI – ‘Erroneous

242

trend’ cartoon illustrating pitfall of averaging individual tree-ring Hg records that cover different

243

periods of time without first correcting for tree-specific bias). This problem is often avoided in

244

traditional dendrochronology studies where cambial age trends in ring-width series are first

245

detrended by dividing the raw ring-width series by its line of best-fit, which produces

246

standardized indices normalized to a mean value of one.25 However, traditional curve-fit

247

detrending methods are unable to perfectly separate the independent effects of cambial age

248

(usually the dominant mode of variability) and climate signals in ring-width series, which results


Page 12 of 31

Page 13 of 31


249

in some loss of low-frequency climate information in the standardized series.34 Progress to better

250

separate the independent effects of cambial age and climate in tree-ring width series is described

251

by Melvin and Briffa43 based on a method called ‘signal-free standardization’. The premise of

252

the signal-free method is to define an initial ‘best estimate’ of the ‘common tree-ring signal’

253

shared by the sampled population, which is then accounted for during the standardization

254

procedure to more efficiently separate cambial age and other tree-specific trends in the raw ring-

255

width series. In theory, and as demonstrated by pseudo-data simulations, the signal-free method

256

optimizes removal of tree-specific trends from the individual series, which allows for a more

257

accurate estimation of the primary environmental signal that is common to the sample population

258

of trees.43

259

The signal-free method can also be applied to the Scree Hill tree-ring Hg dataset to

260

improve recovery of the common Hg signal, but in simplified form since there are no cambial

261

age-related trends that would otherwise confound the initial best estimate of the common signal.

262

The site average of all raw tree-ring Hg series (calculated here as the robust mean of all tree-ring

263

Hg series) is a reasonable approximation of the common signal (Fig. 3). The mean differences

264

between each raw tree-ring Hg series and the common signal (see ‘diff’ in Fig. 3), therefore, can

265

be regarded as best estimates of tree-specific bias for each tree. These differences can then be

266

subtracted from each of the raw tree-ring Hg series to account for tree-specific bias, which

267

effectively normalizes to each individual series a common mean defined by all other trees during

268

the period of overlap. Ultimately this procedure minimizes inter-tree Hg variance in the bias-

269

adjusted dataset.

270 271

Adjusting (i.e., normalizing) for tree-specific bias has no effect on the overall mean Hg concentration (1.22 ng g-1), but greatly reduces the spread of data around the site average record. 12 ACS Paragon Plus Environment


272

The mean inter-tree variance (σ2), averaged over all 5-year time-steps, is 0.47 ng g-1 for the raw

273

dataset (Fig. 4a) and 0.32 ng g-1 for the bias-adjusted series (Fig. 4b). Alternatively, the improved

274

spread of residuals can be quantified using the root-mean-squared-error (RMSE) statistic

275

averaged over all 5-year time-steps; the raw dataset is characterized by an average RMSE of 1.49

276

ng g-1 and the bias-adjusted dataset is characterized by an average RMSE of 1.05 ng g-1,

277

indicating that the bias-adjusted site-average record better characterizes the full dataset than the

278

raw site-average record. In general, the two site averages calculated from the raw and bias-

279

adjusted Hg series are strongly correlated (r = 0.98) due to similar trends, but there is one notable

280

difference with respect to the slope of the post-1950 CE trend. The raw tree-ring Hg average

281

record shows a slightly positive, but statistically insignificant increase after 1950 CE (+0.01 ng

282

g-1 decade-1, p = 0.09), whereas the bias-adjusted average record shows a significant positive

283

trend (+0.04 ng g-1 decade-1, p < 0.01). These differences in trend are not trivial if the research

284

objective is to characterize recent trends in atmospheric Hg0, and highlights the need to account

285

for tree-specific bias in order to most accurately constrain the average record and trends related

286

to atmospheric change. The bias-adjusted site average record (Fig. 4b) is superior in this regard

287

and we base our subsequent discussions of long-term trends on this record.

288

Aside from tree-specific bias in mean value, individual series also contain a significant

289

fraction of random noise that is unrelated to the common signal. This is demonstrated by a less

290

than perfect average correlation between each of the individual tree-ring Hg series and the site

291

average tree-ring Hg record defined by all other trees (rmaster = 0.32, p < 0.02, mean series length

292

= 42 data points). A similar correlation analysis based on first differences calculated for the bias-

293

adjusted series, thereby retaining only high-frequency variance, yields a lower rmaster of 0.07

294

(n.s.). This reveals that high-frequency (sub-decadal) variability is generally uncorrelated


Page 14 of 31

Page 15 of 31


295

between trees and tends to pull down the overall correlation. Conversely, rmaster increases when

296

the dataset is smoothed (e.g., rmaster increases to 0.41 when all series are first smoothed with a 40-

297

year cubic spline). One likely reason for the high-frequency noise in the Scree Hill dataset is our

298

sub-sampling strategy, specifically that annual ring increments within each 5-year block were not

299

forced have the same proportional contribution; rather, they reflect the natural weighting of the 5

300

rings as they are represented in the individual core samples. Interannual growth ring patterns can

301

vary around the circumference of an individual tree and between trees for various reasons,

302

including phenotype disposition or localized disturbance. Since atmospheric Hg0 is not constant

303

from year to year, unequal representation of annual increments in the 5-year blocks could lead to

304

seemingly random differences between individual tree-ring Hg series. Finally, it is important to

305

recognise that a small fraction of the noise in the Scree Hill dataset is likely due to analytical

306

precision (e.g., 4.8% RSD for the powdered wood standard) and potentially intra-tree

307

heterogeneity in Hg concentration. For our samples measured in duplicate (every 9th sample, or

308

~11% of all samples), an average standard deviation of 0.1 ng g-1 was observed. The 1σ of 0.1 ng

309

g-1 for duplicate measurements is slightly higher than the expected 1σ analytical uncertainty of

310

0.06 ng g-1 (assuming a mean tree-ring Hg concentration of 1.22 ng g-1 and analytical precision

311

of 4.8%), which suggests intra-tree-ring Hg heterogeneity may have some (albeit a fairly small)

312

effect on tree-ring Hg series.

313

Although high-frequency coherence is generally low in our dataset, there is some

314

evidence of high-frequency coherence at ca.1698 CE where several trees (n = 10) document a

315

short-lived spike in local Hg0 (Fig. 4b). The 1698 CE data point represents an average Hg

316

concentration of the tree-rings dating to 1696-1700 CE and, therefore, the putative increase in

317

local Hg0 may have occurred in any one or several of these years. The average tree-ring Hg



318

concentration for the 1698 CE data point is 0.45 ng g-1 higher than the previous 5-year data point

319

(1693 CE). This anomaly is ~4.5 standard deviations greater than the average difference between

320

consecutive data points and, therefore, the 1698 CE data point is a statistical outlier in the

321

context of the 400-year record. The origin of this putative Hg0 increase at ca.1698 CE could be

322

from a local wildfire or volcanism in Yukon or Alaska; however, there are no historical records

323

dating to back this far to confirm either of the possible sources. Industrial activity is unlikely the

324

cause, since central Yukon remained undeveloped until the late 19th century. Tree-ring Hg

325

records from other sites in the region are needed to better understand the spatial extent of the

326

ca.1698 CE event, which may provide more evidence on the cause.

327

Regardless of cause, the ca.1698 CE anomaly demonstrates that tree-ring Hg series do

328

preserve some high-frequency information that can be recovered with sufficient replication. This

329

example also demonstrates that tree-ring Hg trends are preserved over long time-scales. In a

330

recent growth chamber study, Arnold et al.18 presented some evidence of radial translocation of

331

Hg between adjacent rings, with the implication being that tree-ring Hg trends may not be stable

332

over long periods. However, the existence of common Hg trends such as the ca.1698 CE

333

anomaly suggests otherwise. Radial translocation had more than 300 years to redistribute Hg to

334

adjacent rings, and the fact that these rings (1696-1700 CE) retained such a high Hg content

335

relative to adjacent rings after so three centuries implies that radial translocation is negligible.

336

3.3 Tree-ring mean Hg concentrations

337

The Scree Hill dataset is characterized by raw Hg concentrations for individual trees

338

ranging from 0.3 to 3.2 ng g-1; the mean tree-ring Hg record spans a range of concentrations from

339

0.7 to 1.9 ng g-1 with an overall mean concentration of 1.22 ng g-1. These observations fall on the

340

lower end of published tree-ring Hg concentrations11–17,19,20,44–47 which range from 0.2 to 644 ng 15 ACS Paragon Plus Environment

Page 16 of 31

Page 17 of 31


341

g-1. However, it is important to note that the upper range of the published values is defined by

342

sites in close proximity to major Hg emission sources. Wood Hg concentrations less than 5 ng g-1

343

are more typical for Pinaceae trees from unpolluted sites in North America.16,44,48

344

The nearest industrial Hg0 emission source to Scree Hill would have been the Klondike

345

gold mining district near Dawson City (Fig. 1), which has a legacy of Hg pollution at some sites

346

(e.g., Bear Creek) where fine gold was recovered from placer ore with the Hg amalgam method.

347

However, the Klondike is ~130 km SW of Scree Hill and likely too far to significantly influence

348

regional atmospheric concentrations. In a study of Scots pine in the Czech Republic, Navrátil et

349

al.14 found that tree-ring Hg concentrations were significantly elevated within a radius of 4 km of

350

a major chlor-alkali plant (~5× higher than values at a control site 37 km away), but that tree-ring

351

Hg concentrations beyond that perimeter (e.g., 4-9 km) were no different from the control site.

352

Considering the great distance between Scree Hill and the Klondike, and that the upper range of

353

Scree Hill tree-ring Hg values is consistent with wood Hg values observed in Picea glauca trees

354

at pristine sites44, it seems more likely that the Scree Hill record (Fig. 3b) captures the untainted

355

signal of regional atmospheric change, integrating the global background as well as Hg emission

356

sources that are atmospherically upstream from the site, including outgassing volcanoes in

357

southern Alaska and industrial emission sources in Eurasia.

358

3.4 Long-term trends in tree-ring Hg (1606-2015 CE)

359

Here, we focus our discussion on long-term trends in the mean record which are most

360

relevant for comparison with other long natural Hg records, including lake sediments, ice cores

361

and peat cores. However, it is important to note that while all of these proxies are generally

362

interpreted to reflect long-term changes in local atmospheric Hg, they do not all accumulate Hg

363

inputs in the same manner. For example, tree-rings and peat cores are primarily proxies for 16 ACS Paragon Plus Environment


364

atmospheric Hg0 concentrations, while lake sediments record atmospheric deposition of oxidized

365

Hg(II) to the lake and its watershed. Therefore, differences and similarities between different

366

proxy records should be interpreted with some degree of caution.

367

The Scree Hill bias-adjusted tree-ring Hg record (Fig. 4b) is defined by 6 or more trees

368

for all 5-year increments after 1643 CE, and 10.3 trees on average for the period of record. The

369

mean record shows little change in Hg concentration prior to ca. 1750 CE, with a mean of 0.90 ±

370

0.12 ng g-1. From 1751-1850 CE, Hg concentrations increase at a rate of 0.025 ± 0.006 ng g-1

371

decade-1 (p = 3.3×10-4), and then at a faster rate of 0.035 ± 0.005 ng g-1 decade-1 (p = 4.1 ×10-6)

372

from 1851-1950 CE. There is brief plateau in the Hg trend during the mid-20th century, and then

373

a continued increase at a rate of 0.037 ± 0.010 ng g-1 decade-1 from 1951-2015 CE (p = 0.003).

374

The mean Hg concentration post-1951 CE was 1.61 ± 0.10 ng g-1, with a maximum of 1.85 ng g-1

375

for the terminal 5-year increment (2011-2015).

376

The increasing trend in tree-ring Hg concentrations from ca. 1750-1850 CE pre-dates the

377

large, post-1850 increase observed in most natural archives of atmospheric Hg deposition, such

378

as lake sediments,49 peat cores5 and ice cores3,4, although a small increase in Hg accumulation in

379

the late 1700s is apparent in the nearby Mt. Logan ice core record from SW Yukon.3 The mid-

380

1700s also marks the start of a sustained increase in a long tree-ring record from Nevada.16 As

381

this early increase is observed in both tree-rings and ice cores from Yukon, it seems likely that

382

Hg0 levels were regionally elevated at the time. The timing of this increase coincides with the

383

later period of colonial silver mining in Spanish America (1572-1833 CE) that used Hg

384

amalgamation50,51 which resulted in locally elevated Hg accumulation in lake sediments7,52 and

385

trace metal deposition in glacier ice.53 It is unclear if this is the emission source reflected in the

386

Yukon ice core and tree-ring records. Many lake sediment cores in North America do not show a 17 ACS Paragon Plus Environment

Page 18 of 31

Page 19 of 31


387

significant Hg increase during this period, which has been suggested as evidence that Hg

388

emissions associated with early South American colonial mining did not significantly impact the

389

North American continent.49 Regardless of source, our data suggest that the assumption of

390

constant anthropogenic emissions from 1570-1850 CE, as in some global biogeochemical

391

models,32,33,54 may not be valid and may in fact underestimate Hg emissions upwind of NW

392

North America. Although additional tree-ring records from other locales are needed to

393

substantiate this.

394

From 1851 to 1950 CE, tree-ring Hg concentrations continue to increase, likely due to a

395

combination of Hg emissions associated with silver and gold mining in North America in the late

396

19th century, and continued industrialization in Europe and North America.33 Increases in tree-

397

ring Hg concentrations through this interval are also observed in several tree-ring Hg records

398

from sites in the eastern Sierra Nevada Mountain Range, characterized by trends of +0.033 to

399

+0.16 ng g-1 per decade over the period 1900-200016 Over the same period, the Scree Hill record

400

shows a smaller rate of increase of +0.013 ng g-1 per decade. The larger rate of increase in the

401

eastern Sierra Nevada Range may reflect locally elevated atmospheric Hg0 concentrations from

402

local mining sources,16 or simply regional differences in Hg cycling.

403

The small plateau observed in the Scree Hill record after 1950 could be the result of

404

declining anthropogenic Hg emissions during the 1940s.33 Previous estimates suggest that

405

anthropogenic Hg emissions globally peaked in 197055 and stabilized thereafter.33 However, our

406

record shows a continued rise in tree-ring Hg concentrations up until the present day, likely

407

because emitted Hg can continue to cycle actively between various compartments (e.g.,

408

atmosphere, surface ocean, fast terrestrial reservoir) for years to decades,54 thereby smoothing

409

out the emission signal recorded in tree-rings. It is interesting to note that the large (up to 1-2% 18 ACS Paragon Plus Environment


410

year-1) decline in atmospheric Hg0 concentrations reported for parts of temperate North America

411

and western Europe since 199056 is not supported by the Scree Hill tree-ring record, nor is it

412

supported by lake sediment core records from nearby Alaska49. This discrepancy may be due to

413

atmospheric Hg0 in Yukon and Alaska being mainly influenced by Asian sources,57 which have

414

continued to increase to present day,33 thus resulting in smaller declines in atmospheric Hg0

415

concentrations and deposition compared to temperate North America.57 However, it remains the

416

case that measurements of atmospheric Hg in northern Canada in general are showing decreasing

417

Hg0 concentrations in the past 20 years or so, 42 while proxy archives such as sediment cores,8,49

418

and now tree-rings, are suggesting a sustained increase in atmospheric Hg concentration and

419

deposition to the present.

420

Proxy records are also useful for quantifying the enrichment of mercury in ecosystems

421

due to anthropogenic emissions. Enrichment Factor (EF) estimates are typically calculated as the

422

ratio of the modern to pre-industrial Hg concentration (or flux). Amos et al.32 recently calculated

423

median EF values (pre-industrial baseline = 1760-1880) for lake sediment and peat core records,

424

which range from 3 (lake sediments) to 4.3 (peat cores) and broadly agree with model-based EF

425

estimates. In nearby Alaska, Engstrom et al.49 calculated EFs (pre-industrial baseline = pre-1500

426

CE) for 8 lake sediment core records with EF values ranging from 2.2-5.2 (median = 2.9). Based

427

on the Scree Hill record, we estimate an EF of 1.9 when comparing post-1991 and pre-1750 CE

428

tree-ring Hg concentrations (excluding the outlier 1698 CE data point). We used the post-1991

429

period since we did not observe a distinct mid-20th century maximum as other studies have noted

430

(e.g., Horowitz et al.55).

431

The overall EF calculated for the Scree Hill record is similar to the low-end of EF values

432

from lake sediment records in nearby Alaska, but less than the median of 2.9 from these Alaskan 19 ACS Paragon Plus Environment

Page 20 of 31

Page 21 of 31


433

records.49 This discrepancy is possibly explained by differences in pre-industrial baseline period,

434

where the pre-1500 CE baseline assumed by Engstrom et al.49 would favor higher EF values on

435

average due to lower Hg concentrations for most Alaskan lakes further back in time. Estimates of

436

EF are particularly sensitive to the pre-industrial baseline,32 and our record does not extend as far

437

back in time as many sediment records do. Alternatively, some differences may be explained by

438

spatial variability in Hg deposition due to continentality. Models show highly variable patterns of

439

Hg concentrations and deposition in northwestern Canada57, although recent temporal trends in

440

central Yukon have not been monitored empirically. The nearest sites where gaseous elemental

441

Hg has been monitored semi-continuously are at Little Fox Lake (~500 km away) near the city of

442

Whitehorse in southern Yukon, and Barrow, coastal northern Alaska (~1000 km away), both of

443

which have only a couple years of data and cannot be used to constrain long-term trends42. The

444

lack of long-term direct measurements of Hg in this region highlights the need to develop proxy-

445

based Hg records from tree-rings and other natural archives to fill knowledge gaps.

446

Seasonality of Hg0 uptake in trees may also influence long-term enrichment in our record.

447

Since stomatal uptake of Hg0 and fixation in tree-rings is limited to the summer months, winter

448

signals may be effectively ‘missed’ by tree-rings, which may lead to some differences in Hg

449

trends between tree-rings and other proxy types such as lake sediments or peatlands which

450

integrate year-round Hg deposition. Seasonal differences in mean Hg0 concentration in the

451

Northern Hemisphere are higher in the winter due to a combination of elevated anthropogenic

452

emissions and longer lifetime against oxidation.41 However, it is not yet clear whether temporal

453

trends differ by season, and highlights a strength of tree-rings which can be used to isolate long-

454

term trends in summertime Hg0. Finally, it is also important to note that sedimentary proxies may

455

integrate non-atmospheric inputs, including remobilized legacy Hg from the catchment9, which



456

again is another factor that may explain differences between lake sediment Hg and tree-ring Hg

457

records. Indeed, the fact that trees maintain a fixed position on the landscape throughout their

458

lifetime and are thought to uptake Hg0 primarily from air and not soil18,20,45 make tree-rings an

459

ideal proxy for isolating temporal changes in atmospheric Hg0.

460

Globally, the all-time EF (pre-1550 CE baseline) has been estimated to be significantly

461

larger than the pre-industrial EF (1760-1880 CE)5,32, which demonstrates a significant

462

anthropogenic influence on atmospheric Hg deposition that predates the industrial era.51,52 Tree-

463

ring records can provide valuable insights on the all-time anthropogenic Hg enrichment, but this

464

will require longer tree-ring Hg datasets than Scree Hill. Millennial length tree-ring width and

465

tree-ring density chronologies have been developed in neighboring Alaska and Northwest

466

Territories35,36 and, therefore, this particular region holds potential for developing sufficiently

467

long tree-ring Hg-based estimates of the all-time EF.

468

469

Acknowledgements

470

We thank Jonny Vandewint, Gerard Otiniano, Aleesha Bakkelund and Avneet Ghotra for their

471

assistance in the field and lab. Financial support for this project was provided by Natural

472

Sciences and Engineering Research Council of Canada Discovery Grants to TP and IL,

473

Connaught New Researcher Awards to TP and IL, and a Canadian Foundation for Innovation

474

John Evans Leadership Fund Award to IL and TP.

475

476

Supplementary Information


Page 22 of 31

Page 23 of 31


477

One illustration of ‘erroneous trend’ caused by tree-specific bias (SI – erroneous trend.pdf); and

478

one zip-file (Supporting Information files.zip) containing tree-ring Hg data (SI - Tree-ring Hg

479

data.xlsx) and a COFECHA cross-dating report (SI – COFECHA report.txt).

480

481

References

482 483

(1)

Lehnherr, I. Methylmercury Biogeochemistry: A Review with Special Reference to Arctic Aquatic Ecosystems. Environ. Rev. 2014, 22 (3), 229–243.

484 485

(2)

AMAP. AMAP Assessment 2011: Mercury in the Arctic; Arctic Monitoring and Assessment Programme (AMAP): Oslo, 2011.

486 487

(3)

Beal, S. A.; Osterberg, E. C.; Zdanowicz, C. M.; Fisher, D. A. Ice Core Perspective on Mercury Pollution during the Past 600 Years. Environ. Sci. Technol. 2015, 49 (13), 7641–7647.

488 489 490 491

(4)

Schuster, P. F.; Krabbenhoft, D. P.; Naftz, D. L.; Cecil, L. D.; Olson, M. L.; Dewild, J. F.; Susong, D. D.; Green, J. R.; Abbott, M. L. Atmospheric Mercury Deposition during the Last 270 Years: A Glacial Ice Core Record of Natural and Anthropogenic Sources. Environ. Sci. Technol. 2002, 36, 2303–2310.

492 493 494

(5)

Enrico, M.; Le Roux, G.; Heimbürger, L.-E.; Van Beek, P.; Souhaut, M.; Chmeleff, J.; Sonke, J. E. Holocene Atmospheric Mercury Levels Reconstructed from Peat Bog Mercury Stable Isotopes. Environ. Sci. Technol. 2017, 51 (11), 5899–5906.

495 496 497

(6)

Lamborg, C. H.; Fitzgerald, W. F.; Damman, A. W. H.; Benoit, J. M.; Balcom, P. H.; Engstrom, D. R. Modern and Historic Atmospheric Mercury Fluxes in Both Hemispheres: Global and Regional Mercury Cycling Implications. Global Biogeochem. Cycles 2002, 16 (4), 51-1–11.

498 499

(7)

Cooke, C. A.; Hintelmann, H.; Ague, J. J.; Burger, R.; Biester, H.; Sachs, J. P.; Engstrom, D. R. Use and Legacy of Mercury in the Andes. Environ. Sci. Technol. 2013, 47 (9), 4181–4188.

500 501 502 503

(8)

Muir, D. C. G.; Wang, X.; Yang, F.; Nguyen, N.; Jackson, T. A.; Evans, M. S.; Douglas, M.; Köck, G.; Lamoureux, S.; Pienitz, R.; et al. Spatial Trends and Historical Deposition of Mercury in Eastern and Northern Canada Inferred from Lake Sediment Cores. Environ. Sci. Technol. 2009, 43 (13), 4802–4809.

504 505 506

(9)

Rydberg, J.; Klaminder, J.; Rosén, P.; Bindler, R. Climate Driven Release of Carbon and Mercury from Permafrost Mires Increases Mercury Loading to Sub-Arctic Lakes. Sci. Total Environ. 2010, 408 (20), 4778–4783.

507 508 509

(10)

Emile-Geay, J.; McKay, N. P.; Kaufman, D. S.; von Gunten, L.; Wang, J.; Anchukaitis, K. J.; Abram, N. J.; Addison, J. A.; Curran, M. A. J.; Evans, M. N.; et al. A Global Multiproxy Database for Temperature Reconstructions of the Common Era. Sci. Data 2017, 4, 170088.

510 511 512

(11)

Abreu, S. N.; Soares, A. M. V. M.; Nogueira, A. J. A.; Morgado, F. Tree Rings, Populus Nigra L., as Mercury Data Logger in Aquatic Environments: Case Study of an Historically Contaminated Environment. Bull. Environ. Contam. Toxicol. 2008, 80 (3), 294–299. 22 ACS Paragon Plus Environment


513 514 515

(12)

Hojdová, M.; Navrátil, T.; Rohovec, J.; Žák, K.; Vaněk, A.; Chrastný, V.; Bače, R.; Svoboda, M. Changes in Mercury Deposition in a Mining and Smelting Region as Recorded in Tree Rings. Water, Air, Soil Pollut. 2011, 216 (1–4), 73–82.

516 517 518 519

(13)

Maillard, F.; Girardclos, O.; Assad, M.; Zappelini, C.; Pérez Mena, J. M.; Yung, L.; Guyeux, C.; Chrétien, S.; Bigham, G.; Cosio, C.; et al. Dendrochemical Assessment of Mercury Releases from a Pond and Dredged-Sediment Landfill Impacted by a Chlor-Alkali Plant. Environ. Res. 2016, 148, 122–126.

520 521 522

(14)

Navrátil, T.; Šimeček, M.; Shanley, J. B.; Rohovec, J.; Hojdová, M.; Houška, J. The History of Mercury Pollution near the Spolana Chlor-Alkali Plant (Neratovice, Czech Republic) as Recorded by Scots Pine Tree Rings and Other Bioindicators. Sci. Total Environ. 2017, 586, 1182–1192.

523 524

(15)

Siwik, E. I. H.; Campbell, L. M.; Mierle, G. Distribution and Trends of Mercury in Deciduous Tree Cores. Environ. Pollut. 2010, 158 (6), 2067–2073.

525 526 527

(16)

Wright, G.; Woodward, C.; Peri, L.; Weisberg, P. J.; Gustin, M. S. Application of Tree Rings [Dendrochemistry] for Detecting Historical Trends in Air Hg Concentrations across Multiple Scales. Biogeochemistry 2014, 120, 149–162.

528 529 530

(17)

Jung, R.; Ahn, Y. S. Distribution of Mercury Concentrations in Tree Rings and Surface Soils Adjacent to a Phosphate Fertilizer Plant in Southern Korea. Bull. Environ. Contam. Toxicol. 2017, 99 (2), 253–257.

531 532 533

(18)

Arnold, J.; Gustin, M. S.; Weisberg, P. J. Evidence for Nonstomatal Uptake of Hg by Aspen and Translocation of Hg from Foliage to Tree Rings in Austrian Pine. Environ. Sci. Technol. 2018, 52 (3), 1174–1182.

534 535 536

(19)

Becnel, J.; Falgeust, C.; Cavalier, T.; Gauthreaux, K.; Landry, F.; Blanchard, M.; Beck, M. J.; Beck, J. N. Correlation of Mercury Concentrations in Tree Core and Lichen Samples in Southeastern Louisiana. Microchem. J. 2004, 78 (2), 205–210.

537 538

(20)

Zhang, L.; Qian, J.-L.; Planas, D. Mercury Concentration in Tree Rings of Black Spruce (Picea Mariana Mill. BSP) in Boreal Quebec, Canada. Water, Air Soil Pollut. 1995, 81, 163–173.

539 540 541 542

(21)

Gagen, M.; McCarroll, D.; Loader, N. J.; Robertson, I.; Jalkanen, R.; Anchukaitis, K. J. Exorcising the “Segment Length Curse”: Summer Temperature Reconstruction since AD 1640 Using NonDetrended Stable Carbon Isotope Ratios from Pine Trees in Northern Finland. The Holocene 2007, 17, 435–446.

543 544 545

(22)

Young, G. H. F.; Demmler, J. C.; Gunnarson, B. E.; Kirchhefer, A. J.; Loader, N. J.; McCarroll, D. Age Trends in Tree Ring Growth and Isotopic Archives: A Case Study of Pinus Sylvestris L. from Northwestern Norway. Global Biogeochem. Cycles 2011, 25, GB2020.

546 547 548

(23)

Esper, J.; Frank, D. C.; Battipaglia, G.; Büntgen, U.; Holert, C.; Treydte, K.; Siegwolf, R.; Saurer, M. Low-Frequency Noise in δ13C and δ18O Tree Ring Data: A Case Study of Pinus Uncinata in the Spanish Pyrenees. Global Biogeochem. Cycles 2010, 24 (4), GB4018.

549

(24)

McCarroll, D.; Loader, N. J. Stable Isotopes in Tree Rings. Quat. Sci. Rev. 2004, 23, 771–801.

550

(25)

Fritts, H. C. Tree-Rings and Climate; The Blackburn Press: Caldwell, New Jersey, 1976.

551 552

(26)

Holmes, R. L. Computer-Assisted Quality Control in Tree-Ring Dating and Measurement. TreeRing Bull. 1983, 43, 69–78.

553

(27)

Leavitt, S. W.; Bannister, B. Dendrochronology and Radiocarbon Dating: The Laboratory of Tree23 ACS Paragon Plus Environment

Page 24 of 31

Page 25 of 31


Ring Research Connection. Radiocarbon 2009, 51, 373–384.

554 555 556

(28)

Grissino-Mayer, H. D. An Updated List of Species Used in Tree-Ring Research. Tree-Ring Bull. 1993, 53, 17–43.

557 558 559

(29)

Loader, N. J.; Robertson, I.; McCarroll, D. Comparison of Stable Carbon Isotope Ratios in the Whole Wood, Cellulose and Lignin of Oak Tree-Rings. Palaeogeogr. Palaeoclimatol. Palaeoecol. 2003, 196, 395–407.

560 561

(30)

Cook, E. R.; Peters, K. The Smoothing Spline: A New Approach to Standardizing Forest Interior Tree-Ring Width Series for Dendroclimatic Studies. Tree-Ring Bull. 1981, 41, 45–53.

562 563 564

(31)

Cook, E.; Shiyatov, S.; Mazepa, V. Estimation of the Mean Chronology. In Methods of Dendrochronology: Applications in the Environmental Sciences; Cook, E. R., Kairiukstis, L. A., Eds.; Kluwer Academic Publishers: Boston, 1990; pp 123–132.

565 566 567

(32)

Amos, H. M.; Sonke, J. E.; Obrist, D.; Robins, N.; Hagan, N.; Horowitz, H. M.; Mason, R. P.; Witt, M.; Hedgecock, I. M.; Corbitt, E. S.; et al. Observational and Modeling Constraints on Global Anthropogenic Enrichment of Mercury. Environ. Sci. Technol. 2015, 49 (7), 4036–4047.

568 569 570

(33)

Streets, D. G.; Horowitz, H. M.; Jacob, D. J.; Lu, Z.; Levin, L.; ter Schure, A. F. H.; Sunderland, E. M. Total Mercury Released to the Environment by Human Activities. Environ. Sci. Technol. 2017, 51 (11), 5969–5977.

571 572 573

(34)

Cook, E. R.; Briffa, K. R.; Meko, D. M.; Graybill, D. A.; Funkhouser, G. The “segment Length Curse” in Long Tree-Ring Chronology Development for Paleoclimatic Studies. The Holocene 1995, 5 (2), 229–237.

574 575 576

(35)

Anchukaitis, K. J.; D’Arrigo, R. D.; Andreu-Hayles, L.; Frank, D.; Verstege, A.; Curtis, A.; Buckley, B. M.; Jacoby, G. C.; Cook, E. R. Tree-Ring-Reconstructed Summer Temperatures from Northwestern North America during the Last Nine Centuries. J. Clim. 2013, 26 (10), 3001–3012.

577 578 579

(36)

Porter, T. J.; Pisaric, M. F. J. J.; Kokelj, S. V.; DeMontigny, P. A Ring-Width-Based Reconstruction of June–July Minimum Temperatures since AD1245 from White Spruce Stands in the Mackenzie Delta Region, Northwestern Canada. Quat. Res. 2013, 80 (2), 167–179.

580 581

(37)

Barber, V. A.; Juday, G. P.; Finney, B. P. Reduced Growth of Alaskan White Spruce in the Twentieth Century from Temperature-Induced Drought Stress. Nature 2000, 405, 668–673.

582 583 584

(38)

Porter, T. J.; Pisaric, M. F. J.; Kokelj, S. V; Edwards, T. W. D. Climatic Signals in δ13C and δ18O of Tree-Rings from White Spruce in the Mackenzie Delta Region, Northern Canada. Arctic, Antarct. Alp. Res. 2009, 41, 497–505.

585 586

(39)

King, G. M. Factors Influencing the Growth of White Spruce (Picea Glauca) in the Mackenzie Delta, NT. Unpublished MSc Thesis; Carleton University: Ottawa, Canada, 2009.

587 588 589 590

(40)

Kellerhals, M.; Beauchamp, S.; Belzer, W.; Blanchard, P.; Froude, F.; Harvey, B.; McDonald, K.; Pilote, M.; Poissant, L.; Puckett, K.; et al. Temporal and Spatial Variability of Total Gaseous Mercury in Canada: Results from the Canadian Atmospheric Mercury Measurement Network (CAMNet). Atmos. Environ. 2003, 37 (7), 1003–1011.

591 592 593

(41)

Jiskra, M.; Sonke, J. E.; Obrist, D.; Bieser, J.; Ebinghaus, R.; Myhre, C. L.; Pfaffhuber, K. A.; Wängberg, I.; Kyllönen, K.; Worthy, D.; et al. A Vegetation Control on Seasonal Variations in Global Atmospheric Mercury Concentrations. Nat. Geosci. 2018, 11 (4), 244–250.

594

(42)

Steffen, A.; Lehnherr, I.; Cole, A.; Ariya, P.; Dastoor, A.; Durnford, D.; Kirk, J.; Pilote, M. 24 ACS Paragon Plus Environment


Atmospheric Mercury in the Canadian Arctic. Part I: A Review of Recent Field Measurements. Sci. Total Environ. 2015, 509–510, 3–15.

595 596 597 598

(43)

Melvin, T. M.; Briffa, K. R. A “Signal-Free” Approach to Dendroclimatic Standardisation. Dendrochronologia 2008, 26, 71–86.

599 600 601

(44)

Friedli, H. R.; Radke, L. F.; Payne, N. J.; McRae, D. J.; Lynham, T. J.; Blake, T. W. Mercury in Vegetation and Organic Soil at an Upland Boreal Forest Site in Prince Albert National Park, Saskatchewan, Canada. J. Geophys. Res. 2007, 112 (G1), G01004.

602 603

(45)

Fleck, J. A.; Grigal, D. F.; Nater, E. A. Mercury Uptake by Trees: An Observational Experiment. Water. Air. Soil Pollut. 1999, 115 (1/4), 513–523.

604 605 606

(46)

Reimann, C.; Arnoldussen, A.; Finne, T. E.; Koller, F.; Nordgulen, Ø.; Englmaier, P. Element Contents in Mountain Birch Leaves, Bark and Wood under Different Anthropogenic and Geogenic Conditions. Appl. Geochemistry 2007, 22 (7), 1549–1566.

607 608 609

(47)

Nóvoa-Muñoz, J. C.; Pontevedra-Pombal, X.; Martínez-Cortizas, A.; García-Rodeja Gayoso, E. Mercury Accumulation in Upland Acid Forest Ecosystems Nearby a Coal-Fired Power-Plant in Southwest Europe (Galicia, NW Spain). Sci. Total Environ. 2008, 394 (2–3), 303–312.

610 611

(48)

Yang, Y.; Yanai, R. D.; Montesdeoca, M.; Driscoll, C. T. Measuring Mercury in Wood: Challenging but Important. Int. J. Environ. Anal. Chem. 2017, 97, 456–467.

612 613 614 615

(49)

Engstrom, D. R.; Fitzgerald, W. F.; Cooke, C. A.; Lamborg, C. H.; Drevnick, P. E.; Swain, E. B.; Balogh, S. J.; Balcom, P. H. Atmospheric Hg Emissions from Preindustrial Gold and Silver Extraction in the Americas: A Reevaluation from Lake-Sediment Archives. Environ. Sci. Technol. 2014, 48, 6533–6543.

616 617

(50)

Hylander, L. D.; Meili, M. 500 Years of Mercury Production: Global Annual Inventory by Region until 2000 and Associated Emissions. Sci. Total Environ. 2003, 304, 13–27.

618

(51)

Nriagu, J. O. Legacy of Mercury Pollution. Nature 1993, 363, 589.

619 620

(52)

Cooke, C. A.; Balcom, P. H.; Biester, H.; Wolfe, A. P. Over Three Millennia of Mercury Pollution in the Peruvian Andes. Proc. Natl. Acad. Sci. U. S. A. 2009, 106 (22), 8830–8834.

621 622 623

(53)

Uglietti, C.; Gabrielli, P.; Cooke, C. A.; Vallelonga, P.; Thompson, L. G. Widespread Pollution of the South American Atmosphere Predates the Industrial Revolution by 240 Y. Proc. Natl. Acad. Sci. U. S. A. 2015, 112 (8), 2349–2354.

624 625 626

(54)

Amos, H. M.; Jacob, D. J.; Streets, D. G.; Sunderland, E. M. Legacy Impacts of All-Time Anthropogenic Emissions on the Global Mercury Cycle. Global Biogeochem. Cycles 2013, 27 (2), 410–421.

627 628 629

(55)

Horowitz, H. M.; Jacob, D. J.; Amos, H. M.; Streets, D. G.; Sunderland, E. M. Historical Mercury Releases from Commercial Products: Global Environmental Implications. Environ. Sci. Technol. 2014, 48 (17), 10242–10250.

630 631 632 633

(56)

Zhang, Y.; Jacob, D. J.; Horowitz, H. M.; Chen, L.; Amos, H. M.; Krabbenhoft, D. P.; Slemr, F.; St Louis, V. L.; Sunderland, E. M. Observed Decrease in Atmospheric Mercury Explained by Global Decline in Anthropogenic Emissions. Proc. Natl. Acad. Sci. U. S. A. 2016, 113 (3), 526– 531.

634 635

(57)

Dastoor, A.; Ryzhkov, A.; Durnford, D.; Lehnherr, I.; Steffen, A.; Morrison, H. Atmospheric Mercury in the Canadian Arctic. Part II: Insight from Modeling. Sci. Total Environ. 2015, 509– 25 ACS Paragon Plus Environment

Page 26 of 31

Page 27 of 31


510, 16–27.

636 637

(58)

MATLAB 2014b. The MathWorks, Inc.: Natick, Massachusetts, United States 2018.

638

(59)

M_Map v1.4j. Available at: https://www.eoas.ubc.ca/~rich/map.html.

639 640

Figure captions

641 642 643

Fig 1. Regional map showing the location of the Scree Hill (SH1) site. This map was created using MATLAB 2014b with the open-source M_Map v1.4j mapping package.58,59

644 645 646 647 648 649 650 651

Fig 2. (a) Scree Hill ring-width series aligned by cambial age (grey lines – all trees; thin black line – average record), and a negative exponential curve fit to years 1-270 of the average record (thick black line; dashed line is an extrapolation of the negative exponential curve beyond cambial age 270). (b) Treering Hg series aligned by cambial age (grey lines), and average records calculated based on all data (black line) and pre-1850 CE data only (red line); average records were calculated for years defined by 3 or more trees. (c) Sample replication with respect to the ‘all data’ (black) and ‘pre-1850 CE only’ (red) average Hg records in sub-plot ‘b’. This figure was created using MATLAB 2014b.58

652 653 654 655 656 657

Fig 3: Comparisons of the individual raw tree-ring Hg series (red lines; tree ID is indicated) against the ‘site average’ tree-ring Hg record (black line) calculated from all other tree-ring Hg series (grey lines). The mean concentration difference (diff.) between each of the individual series and the site average is indicated. Tree IDs from living trees are marked with an ‘*’. This figure was created using MATLAB 2014b.58

658 659 660 661 662 663 664 665

Fig 4: (a) Raw and (b) bias-adjusted tree-ring Hg series (grey lines), site average records (black lines) and 40-year cubic smoothing splines (red lines). Site averages are calculated for all years defined by 3 or more trees. Natural breaks in temporal trend are marked with dashed lines, corresponding to 1750, 1850 and 1950 CE. The 1σ confidence interval (yellow area; excludes outliers defined as points that are 1.5× the inter-quartile range from the mean) is indicated for the bias-adjusted series. (c) Sample replication (black line) and coverage for each tree is indicated (horizontal bars; white dots indicate missing data). This figure was created using MATLAB 2014b.58



Fig 1. Regional map showing the location of the Scree Hill (SH1) site. This map was created using MATLAB 2014b with the open-source M_Map v1.4j mapping package.58,59 249x199mm (300 x 300 DPI)


Page 28 of 31

Page 29 of 31


Fig 2. (a) Scree Hill ring-width series aligned by cambial age (grey lines – all trees; thin black line – average record), and a negative exponential curve fit to years 1-270 of the average record (thick black line; dashed line is an extrapolation of the negative exponential curve beyond cambial age 270). (b) Tree-ring Hg series aligned by cambial age (grey lines), and average records calculated based on all data (black line) and pre1850 CE data only (red line); average records were calculated for years defined by 3 or more trees. (c) Sample replication with respect to the ‘all data’ (black) and ‘pre-1850 CE only’ (red) average Hg records in sub-plot ‘b’. This figure was created using MATLAB 2014b.58 177x199mm (300 x 300 DPI)



Fig 3: Comparisons of the individual raw tree-ring Hg series (red lines; tree ID is indicated) against the ‘site average’ tree-ring Hg record (black line) calculated from all other tree-ring Hg series (grey lines). The mean concentration difference (diff.) between each of the individual series and the site average is indicated. Tree IDs from living trees are marked with an ‘*’. This figure was created using MATLAB 2014b.58 348x199mm (300 x 300 DPI)


Page 30 of 31

Page 31 of 31


Fig 4: (a) Raw and (b) bias-adjusted tree-ring Hg series (grey lines), site average records (black lines) and 40-year cubic smoothing splines (red lines). Site averages are calculated for all years defined by 3 or more trees. Natural breaks in temporal trend are marked with dashed lines, corresponding to 1750, 1850 and 1950 CE. The 1σ confidence interval (yellow area; excludes outliers defined as points that are 1.5× the inter-quartile range from the mean) is indicated for the bias-adjusted series. (c) Sample replication (black line) and coverage for each tree is indicated (horizontal bars; white dots indicate missing data). This figure was created using MATLAB 2014b.58 152x198mm (300 x 300 DPI)


A 400-year record of atmospheric mercury from tree-rings in

Recommend Documents