Page 1

Mapping aerial metal deposition in metropolitan areas from

tree bark: a case study in Sheffield, England

E. Schellea, B. G. Rawlins∗,b, R. M. Larkc, R. Websterc, I. Statona, C. W. Mcleod∗,a

aCentre for Analytical Sciences, Department of Chemistry, University of Sheffield,

Sheffield S3 7RF, UK

bBritish Geological Survey, Keyworth, Nottingham NG12 5GG, UK

cRothamsted Research, Harpenden, Hertfordshire AL5 2JQ, UK

∗Corresponding authors:

B. G. Rawlins

British Geological Survey

Keyworth

Nottingham NG12 5GG

UK Telephone: +44 (0)

Fax: +44 (0)

e-mail: bgr@bgs.ac.uk

C. W. Mcleod

Centre for Analytical Sciences

Department of Chemistry

University of Sheffield

Dainton Building

Sheffield S3 7HF

UK

Telephone: +44 (0) 114 222 3602

Fax: +44 (0) 114 222 9379

1

Manuscript

Click here to download Manuscript: shefr.pdf

Page 2

e-mail: c.w.mcleod@sheffield.ac.uk

2

Page 3

Abstract

1

We investigated the use of metals accumulated on tree bark for mapping their

2

deposition across metropolitan Sheffield by sampling 642 trees of three common species.

3

Mean concentrations of metals were generally an order of magnitude greater than

4

in samples from a remote uncontaminated site. We found trivially small differences

5

among tree species with respect to metal concentrations on bark, and in subsequent

6

statistical analyses did not discriminate between them. We mapped the concentrations

7

of As, Cd and Ni by lognormal universal kriging using parameters estimated by residual

8

maximum likelihood (reml). The concentrations of Ni and Cd were greatest close to

9

a large steel works, their probable source, and declined markedly within 500 metres

10

of it and from there more gradually over several kilometres. Arsenic was much more

11

evenly distributed, probably as a result of locally mined coal burned in domestic fires

12

for many years. Tree bark seems to integrate airborne pollution over time, and our

13

findings show that sampling and analysing it are cost-effective means of mapping and

14

identifying sources.

15

Capsule: Multi-element analysis of tree bark can be effective for mapping the deposition

16

of metals from air and relating it to sources of emission.

17

Keywords: Multi-element analysis; Arsenic; Cadmium; Nickel; Geostatistics; REML;

18

Universal kriging

19

20

3

Page 4

1. Introduction

21

22

Inhalation of atmospheric aerosols, particularly of the fine size-fraction, can cause

23

lung diseases, and regulatory standards exist to ensure that air quality meets interna-

24

tionally defined standards. Airborne particulate matter (APM) for PM2.5and PM10is

25

now widely monitored, particularly in urban environments. Nevertheless, government

26

agencies and local authorities rarely have the resources to install equipment at the

27

many sites that would be needed to map the spatial distribution of airborne particles.

28

Normal practice for monitoring metals in APM is to establish installations at a few

29

fixed locations, as in the Heavy Metals Monitoring Networks in the United Kingdom

30

(Brown et al., 2007).

31

Typical of this approach is the study of Moreno et al. (2004) who analysed APM

32

at five sites in England and Wales. They showed that the air in Sheffield contained

33

many metal-bearing particles in the <PM2.5size-fraction. Those containing Cd and

34

Ni are likely to derive from large steel works (Buse et al., 2003) and to impair health

35

when inhaled. Attempts to apportion the particles to particular sources of metals in

36

APM include chemical mass balance (Wang et al., 2006; Samara et al., 2003) and

37

multivariate statistical analyses (Kim et al., 2006; Shah et al., 2006). Thomaidis et

38

al. (2003) incorporated meteorological variables in their multivariate analysis because

39

they found these influenced the concentrations of Cd, Ni and As in the APM in Athens.

40

Sweet and Vermette (1993) studied anthropogenic emissions in urban Illinois based

41

on trace metal data from three sites; they reported much temporal variation in the

42

quantity of metal in the APM. They attributed this to variations in wind strength

43

and direction, and the degree of atmospheric mixing. Atmospheric particulate matter

44

in towns and cities is readily resuspended, and this process contributes significantly

45

to temporal (Vermette et al., 1991) and spatial variation in the contents of metals

46

(Kuang et al., 2004). So mapping the spatial distribution from direct measurements

47

would require many permanent sampling installations to integrate concentrations over

48

4

Page 5

time.

49

An attractive approach for mapping the long-term spatial distribution of elements

50

in APM is by biomonitoring. The underlying idea is to let plants accumulate atmo-

51

spheric depositions over time and then to analyse chemically the plant tissue. The

52

scope for exploiting plants in this way is diverse and includes plant leaves, lichens,

53

mosses and tree bark (Markert et al, 1993; Walkenhorst et al. 1993). The outer layer

54

of tree bark, in particular, has been found to be an effective passive accumulator of

55

airborne particles in both rural (Bohm et al., 1998) and urban (Tanaka and Ichikuni,

56

1982) environments. The particles in question settle on the outer bark by wet and

57

dry deposition, and they remain there until the tree sheds its bark, or are leached or

58

washed away by rain, or a combination of the two.

59

Smooth-barked trees in the northern temperate zone begin to shed their bark

60

only when mature (after about 50 years); trees with rougher bark tend to shed theirs

61

somewhat earlier. The metal species deposited in the outer bark are separated physi-

62

cally from trace elements taken up in solution from the soil in the trees and their xylem

63

by a layer of phloem and cambium (Martin and Coughtrey, 1982). Further, extraneous

64

contamination from the soil itself is limited to lowest 1.5 m of the trunk. So pollutants

65

in the bark above this height are almost entirely derived from the air (Wolterbeek and

66

Bode, 1995).

67

Determination of the metal contents of tree bark cannot lead to a direct as-

68

sessment of air quality because such measurements are retrospective and integrate as

69

averages over long times. Nevertheless, because trees are widespread in most towns

70

and cities, sampling their bark for subsequent chemical analysis and then noting pre-

71

cise locations mean that the elemental concentrations in the barks can be mapped.

72

Such maps, whether simple displays of measured concentrations or ones made by more

73

elaborate interpolation could point to the emitter(s) of the metals, and identify regions

74

where much (and little) metal is deposited. To date there have been few published

75

attempts to map the distributions of metals from the analysis of tree bark. One was by

76

5

Page 6

Lotschert and Kohm (1978) who drew isarithmic (‘contour’) maps of Pb and Cd based

77

on samples from 34 ash trees (Fraxinus excelsior) throughout Frankfurt. A similar

78

approach, adopted by Bellis et al. (2001) to map airborne emissions in the vicinity of

79

a lead smelter, was based on plotting data on the enrichment in Pb. On a national

80

scale Lippo et al. (1995) drew a ‘pollution’ map of Finland detailing anthropogenic

81

emissions for cities and industrial regions.

82

We have investigated the potential of tree bark for mapping the accumulated

83

deposition of airborne metals across metropolitan Sheffield, a city which has more trees

84

per unit of population than any other in Europe. We measured the concentrations of

85

18 metal and metalloid elements in bark at 642 locations in the region and compared

86

them with those at a virtually uncontaminated site (Mace Head, western Ireland) to

87

determine the magnitude of contamination in the former. We collected bark samples

88

from three tree species to determine whether there were any substantial differences in

89

metal contents between them. We did a principal component analysis on the elements

90

to establish the relationships between the elements and to discover whether there were

91

particular groups of them that might behave differently from one another. We then

92

chose three potentially toxic elements, namely Cd, As and Ni, as representatives and

93

analysed their data spatially (a) to determine regional trends, (b) to estimate their

94

spatial covariances, and (c) to interpolate and map their distributions by kriging. We

95

have used these maps to identify likely local sources of atmospheric pollution. We

96

discuss the wider implications of our findings for the use of tree bark in environmental

97

monitoring.

98

2. Materials and methods

99

2.1 Study region, tree bark survey and analysis.

100

Sheffield has a long tradition of iron smelting and the production of steel. The invention

101

of the crucible process in 1740 sparked a massive expansion of the industry in the city,

102

relying in part on coal from local mines, which continues to this day. This in turn led to

103

6

Page 7

severe air pollution before measures to combat it were introduced under the Clean Air

104

Act of 1956. In 1963 the company British Steel opened a large works at Tinsley in the

105

north east of the city (Figure 1) to make special steels. Its production, including that

106

of stainless steel (ferrochrome), continues and emits significant quantities of Cr and Ni

107

into the atmosphere. Gilbertson et al. (1997) reported on the long-term significance

108

of metal emissions from steel manufacturing from their study of concentrations of Co,

109

Cu, Fe, Ni, Pb, and Zn in a peat monolith from close to the works. They found

110

extraordinarily large concentrations (in mg kg−1) of Cu (472), Ni (320), Pb (827) and

111

Zn (613) compared to concentrations in soil from an urban survey of the city for which

112

Rawlins et al. (2005) presented data for Pb and Ni. The study of Gilbertson et al.

113

attributed the greatest enrichment of Cu and Zn in the uppermost layers to the works.

114

The Environment Agency of the UK had compiled an inventory of pollution

115

(Environment Agency, 2003) in which it registered the locations and quantities of

116

atmospheric particulate metal emissions from static sources. The inventory included

117

emissions exceeding the reporting thresholds of 100 g for Cd, 1 kg for As and 10 kg

118

for Ni, each per year. It did not include sources of smaller amounts for which no

119

information is available on metal composition. We collated the data for the Sheffield

120

region and to 3 km beyond its boundary for the years for which data were available

121

prior to our collection of the bark samples (1998–2002). We calculated the sum of

122

emissions for the five years so that they could be presented as total emission figures

123

(in kg) for each particular source.

124

Below we discuss the significance of these sources in relation to the distributions

125

of the metals in bark.

126

In establishing the region for our current study we wished to encompass the

127

major sources of metal emissions, including industry to the north and east of the City,

128

whilst also estimating the spatial extent of metal deposition. We therefore surveyed

129

an area extending across the city from the suburbs of Whirlow and Greenhill south

130

and west of the centre and from which the prevailing wind blows (see wind rose inset

131

7

Page 8

in Figure 1) to industrial Brinsworth and Ecclesfield to the north-east of the city

132

centre (see Figure 1). The number of people living in the region, estimated from

133

the 1991 census, is approximately 271 500. This figure was calculated from the UK

134

EDINA database of population-weighted centroids defined for all the 1991 enumeration

135

districts (Bracken and Martin, 1989.). The total population of greater Sheffield is

136

around 550 000.

137

Samples of bark were collected from 642 trees of the three species (with pro-

138

portions of each shown in parentheses): sycamore (Acer pseudoplatanus — 68%), oak

139

(Quercus robur — 22%) and cherry (Prunus serrula — 10%); their locations are shown

140

in Figure 1. Both sycamore and cherry have fairly smooth bark, whereas the bark of

141

oak is rougher. The samples were collected between April and November, 2003. In

142

a series of local neighbourhoods, trees belonging to the three species occurring in a

143

public space were identified. From these a subset was sampled to provide, as far as

144

possible, an even spatial distribution.

145

Approximately 10 g of the external outer bark (1–2 mm depth) was removed from

146

each target tree with a clean scraping tool at 1.5 m above the ground. Sample sites

147

spanned an altitude range of 271 m, from 33 to 301 m above mean sea level. The mean

148

altitude was 106 m.

149

Trees in the temperate zone of the UK enlarge their diameters by approximately

150

1 cm per year, so a tree’s circumference in cm divided by π gives an approximate age

151

of the tree (P. Casey, personal communication). The ages of the trees from which bark

152

was sampled ranged from 25 to 45 years (circumferences of 78 to 141 cm). Bark with

153

moss, lichen or paint was excluded from the sample. The orientation of the location for

154

sampling was random. The samplers wore polythene gloves to avoid contaminating the

155

samples, which were stored in sealed brown paper envelopes at 4◦C. The geographic

156

co-ordinates and altitude of each site were obtained by GPS (Garmin International,

157

Inc., USA). A further nine bark samples were collected from sycamore trees at Mace

158

Head on the west coast of Ireland (53◦20’ N, 9◦54’ W) so that we could estimate

159

8

Page 9

background concentrations where there is negligible pollution.

160

Each bark sample was crushed into a fine powder in a Tema mill. The bark

161

powder was then passed through a sieve (0.5 mm mesh) to remove any large lumps.

162

Thereafter all the equipment was cleaned thoroughly to prevent cross-contamination of

163

the following sample. Tree bark powder (4.0 g) was thoroughly mixed with polystyrene

164

co-polymer binder (0.9 g) (Hoechst Wax, Spectro Analytical, UK) and pressed for 1

165

minute to produce powder pellets. The powder pellets were analysed by EDXRF

166

spectrometry (X-LAB, Spectro Analytical, UK). The instrument was calibrated for 18

167

elements (Ag, Al, As, Ba, Cd, Co, Cr, Cu, Fe, Mn, Ni, Pb, Sb, Se, Sn, Ti, V, Zn) for

168

a wide range of standard biological reference materials which included poplar leaves,

169

lichen, human hair and tea leaves. Typical analytical performance has been published

170

previously (Schelle et al., 2002). The concentrations of Cd were less than the detection

171

limit for 19% of the samples, and so we set their values to half that limit for subsequent

172

statistical analysis. Fifty two percent of the samples contained less than the detection

173

limit of Ag, and so we do not consider it further.

174

2.2 Summary statistics

175

Table 1 summarizes the data for all 17 elements. The distribution of most was posi-

176

tively skewed, some strongly, and so to stabilize variances for subsequent analyses we

177

transformed the values to logarithms. The table lists the transformations we made.

178

As expected, there is a huge range in the mean values. Aluminium, which is the

179

most abundant metal in the rocks and soil, is most abundant in the bark also. Iron

180

appears in large amounts, and given Sheffield’s history we might expect such results

181

too. The concentrations of the other elements do not immediately stand out. With the

182

exception of Al, the mean concentrations in Sheffield were much larger than those those

183

at Mace Head (Table 2); for Cr, Mn, Ni and Ti they were an order of magnitude larger.

184

Anthropogenic sources are almost certainly the reason for the greater concentrations

185

of metal in the tree bark in Sheffield.

186

What is highly significant is that the distributions of all the elements are strongly

187

9

Page 10

positively skewed, with skewness coefficients ranging from 1.7 to almost 10. We found

188

that all could be described well by a three-parameter log-normal distribution, which

189

has the probability density function:

190

g(z) =

1

σ(z − α)√2πexp

?

−

1

2σ2{ln(z − α) − µ}2?

, (1)

where z is the variable of interest, µ and σ are the mean and standard deviation

191

of the transformed variable, and α is the shift in the original scale to maximize the

192

goodness of fit. The shift and the mean and standard deviations in natural logarithms

193

are listed in Table 1. In the final column of Table 1 are the skewness coefficients

194

of the logarithms from which it is evident that the transformations have made the

195

distributions symmetric. This is important for stabilizing the variances, and we have

196

done all our further analyses on these transformed scales.

197

Analysis of variance revealed little differences among species; they accounted for

198

less than 5% of the variance for any of 16 metals and for only 8.5% for As. We have

199

therefore disregarded differences between species in our subsequent multivariate and

200

spatial analysis.

201

2.3 Selection of variables; principal component analysis.

202

For the purpose of this paper we wanted to select a few elements from the 17 listed

203

above that would illustrate both the feasibility of analysing APM in bark and mapping

204

the distribution of elements in it and produce maps interesting in their own right.

205

To help in the selection we did a principal component analysis on the correlation

206

matrix of the logarithms. We hoped thereby to see any clusters of strongly correlated

207

elements from which we could choose representatives and any other elements that were

208

clearly uncorrelated with others and should be treated in their own right. Table 3 lists

209

the leading eigenvalues of the correlation matrix. The first component accounts for

210

almost half the variance, and second and third together account for more than half

211

the remainder. Pursuing the analysis, we computed the correlation coefficients, rij,

212

10

Page 11

between the principal component scores and the (logarithms of the) original variables

213

as

214

rij = νij

?

λj/σ2

i, (2)

where νijis the ith entry in the jth eigenvector, λjis the jth eigenvalue, and σ2

iis the

215

variance of the ith original variable. We then plotted the results in the unit circles for

216

pairs of the leading components. We show two such circles in Figure 2 in which we

217

have plotted the correlation coefficients (a) for component 2 against component 1 and

218

(b) for component 3 against component 1. In general, the closer the points lie to the

219

circumference of one of these circles the better are they represented in that projection.

220

We note first that all of the plotted points fall in the right hand halves of the

221

graphs: component 1 is essentially one of size. Component 2 discriminates, separating

222

the siderophile (Fe, Mn, Co, Ni) and lithophile (Cr and V) elements from the calcophile

223

group (Pb and Zn) and their associates. Arsenic appears nearest the centre in circle

224

(a) and the least correlated with the other elements. This is confirmed in circle (b)

225

in which the point for As lies close to the circumference and away from the other

226

elements. Somewhat surprisingly Zn lies near the bottom of axis 3. The siderophiles

227

remain clustered in this projection.

228

From this examination of the data we have chosen three elements for our spatial

229

analysis. We have chosen Ni as representative of the siderophiles and because it is a

230

key element in steel production. We chose Cd because of its potential toxicity and

231

again used in manufacturing. Third, we chose As, another poison, but from Figure

232

2(b) clearly dissociated from the other elements.

233

2.4 Spatial modelling by REML

234

Our objective is to display the spatial variation of the three selected elements on

235

tree bark across Sheffield as isarithmic (‘contour’) maps having first estimated the

236

concentrations at the nodes of a fine grid. We used kriging for the estimation, following

237

closely the technique we used to map the distribution of metals emitted from a smelter

238

11

Page 12

and described recently in this Journal (Rawlins et al., 2006).

239

Ordinary kriging is based on two assumptions.

240

1.A variable of interest, y, at locations xi, i = 1,2,... , is a realization of an

241

intrinsically stationary correlated random function Y (x) such that

242

E[Y (x) − Y (x + h)] = 0for all x, h ,(3)

where E[·] denotes the statistical expectation of the term in brackets, and h is a

lag vector, a displacement in space from the location x.

243

244

2. The expected squared difference between Y (x) and Y (x + h) depends only on h:

245

E

?

{Y (x) − Y (x + h)}2?

= 2γ(h) .(4)

The quantity γ(h) is the variance perpoint at lag h and as a function of h is the

246

variogram.

247

A preliminary display of the data for Ni and Cd at least suggested that the

248

assumption in Equation (3) was not tenable; there were evident trends from small

249

concentrations far from the steel works in the south west of the city to large ones close

250

to the works in the north east, as we expected. This situation requires more complex

251

geostatistical analysis in which the trend is separated from the random component

252

and the estimates are made by universal kriging (Matheron, 1969), or ‘kriging with

253

trend’ as it is now more generally known. Saito and Goovaerts (2001) encountered

254

a similar problem in a study on the distribution of metal pollutants in two urban

255

areas in the United States. In each case there were clear trends in the distribution

256

of these contaminants, which could be accounted for by the wind direction and the

257

location of sources (one smelter in one of the areas, and two adjacent smelters in the

258

second). They used this information to produce simple trend models, based on physical

259

principles, which predict the amount of metal that has been deposited from the sources

260

at any location. This constituted the trend in their universal kriging. In order to model

261

the spatial dependence of the random component, the residual from the trend, they

262

12

Page 13

estimated variograms of the pollutant from paired comparisons between sites at which

263

the trend was deemed to be similar. This crudely filters the trend from the variogram

264

that is obtained. It also discards the information about the random component of

265

variation that could be obtained from comparisons between points where the trend is

266

very different. To do this requires a more sophisticated analysis.

267

Recent developments in numerical analysis linked to modern computing power

268

enable us to use Residual Maximum Likelihood (reml) for the purpose, and we must

269

now regard this as best practice. We described the procedure fully in Rawlins et al.

270

(2006), and we shall not repeat the detail here. In this respect, then, our analysis

271

was more sophisticated than that of Saito and Goovaerts (2001). In another respect

272

it was more primitive, because we did not attempt to use a physically-based model

273

for the trend in metal content of the bark. This was because, by contrast to the two

274

regions studied by Saito and Goovaerts (2001), Sheffield has multiple sources of metal

275

pollutants, and not only current or recent ones, but also many others from the distant

276

past about which we have no detailed information. For our trend models therefore we

277

considered only simple functions of the spatial coordinates.

278

We treat the transformed data as the outcome from a mixed model:

279

Y (x) =

K

?

k=0

βkfk(x) + ε(x) . (5)

It consists of K + 1 fixed effects (which explain the trend in terms of known functions

280

of the spatial co-ordinates) and a spatially dependent random variable ε(x) with mean

281

zero and variogram γ(h). In order to apply reml to estimate the variance of the

282

random variable and its spatial dependence we make stronger assumptions of station-

283

arity than the intrinsic hypothesis stated in Equations (3) and (4) above. We require

284

that the random variable is second-order stationary, which means that the variogram

285

is bounded by the a priori variance of the process. This is not a serious constraint in

286

practice once we have separated out the fixed effects, and is met by most of the popular

287

variogram models used in geostatistics. Our task is to estimate the contributions of

288

the fixed and random components simultaneously, minimizing the estimation variance.

289

13

Page 14

The separate contributions need not be explicitly computed when we use universal

290

kriging, but they should be inspected to assess the weight of evidence for a trend in

291

the variable.

292

We first chose a few plausible models for the trend in Equation (5) by inspection

293

of the data. We then separated these trends from the data and computed experimental

294

variograms of the residuals by the usual method of moments:

295

? γ(h) =

1

2m(h)

m(h)

?

j=1

{y(xj) − y(xj+ h)}2,(6)

where y(xj) and y(xj+h) are the values of y at sampling points xjand xj+h separated

296

by the lag h and m(h) is the number of paired comparisons at that lag. We fitted several

297

of the standard simple models to these variograms by weighted least squares and chose

298

the ones that fitted best in the least squares sense.

299

This estimation of the trend ignores the spatial correlation of the residuals, but

300

is acceptable for exploratory purposes. We found that we could describe the trend in

301

the transformed data simply by the distance from a reference site in the north-east of

302

the region, so that our full model for the variation was

303

Y (x) = β0+ β1||x − xR|| + ε(x) ,(7)

where || · || denotes the Euclidean norm of the enclosed vector. The vector xR is

the reference site close to the steel works in the north-east of the region with British

304

305

National Grid co-ordinates (441945.8, 390339.4). We chose this model in preference to

306

a more conventional linear function of the co-ordinates because it achieved at least as

307

good an ordinary least-squares fit to the data with one fewer terms.

308

We then computed the experimental variograms of the ordinary least-squares

309

residuals and found that an isotropic exponential model with nugget gave a satisfactory

310

fit. Its equation is

311

γ(h) = c0+ c

?

1 − exp

?

−h

a

??

,(8)

in which c0is the nugget variance, c is the sill of the correlated variance, a is a distance

312

parameter and h = ||h|| is now a scalar in distance only. This model, which is widely

313

14

Page 15

used in geostatistics, increases asymptotically to its maximum, with an effective range

314

of 3 × a.

315

We then used the ASreml program (Gilmour et al., 2002) to fit the model in

316

Equation (7) to each variable. We specified an exponential correlation function, which

317

corresponds to the exponential variogram in Equation (8). The program provides reml

318

estimates of the parameters c0, c and a, and generalized least-squares estimates of the

319

fixed effects. We tested the null hypothesis that the true value of the fixed effect for the

320

trend, β1, is zero by computing the Wald statistic. This statistic is equivalent to the

321

variance ratio for the predictor in an analysis of variance for an ordinary least-squares

322

regression. However, we used the method of Kenward and Roger (1997) to compute an

323

adjusted Wald statistic, and adjusted degrees of freedom in the denominator for the F

324

test to allow for the spatial dependence of the residuals.

325

2.5 Lognormal universal kriging

326

For reasons described above we transformed the raw data, z(x), to approximately

327

normally distributed variables, which we have denoted by y(x). These values were

328

used to obtain predictions at points on a fine grid over the region by universal kriging.

329

The universal kriging (UK) uses the specified fixed effects in the prediction and the

330

covariance parameters estimated by reml. Note that for arsenic, for which the trend

331

was effectively constant, the universal kriging predictions are the same as those from

332

ordinary kriging since we estimate one fixed effect, β0, which is the mean.

333

Universal kriging returns an estimate of the transformed random variable Y (x);

334

but we require estimates on the scale of the original data z(x). As with any estimate

335

derived from log-transformed data, we cannot simply back transform the estimates on

336

the logarithmic scale; we must also correct for bias. Cressie (2006) has shown that the

337

UK estimate of a log-normal variable˜Z?(x0), based on the UK estimate˜Y (x0) of the

338

corresponding Y , is

339

˜Z?(x0) = exp

?

˜Y (x0) +1

2σ2

UK− ψ0−

K

?

i=1

ψifi(x0)

?

,(9)

15