Abstract

This study evaluated temporal and spatial variations in water quality to understand the characteristics of the Jialing River watershed in Guangyuan City, China. Data on 17 parameters obtained from seven sites from 2012 to 2015, with a total of 329 samples for each parameter, were analyzed using multivariate statistical techniques. Observation months were grouped into two periods (Period A, May–November; Period B, December–April) according to similarities in water quality characteristics through time analysis and cluster analysis (CA). Water temperature (TEMP), flow rate (Q), dissolved oxygen (DO), oils, fluoride (F) and cadmium (Cd) were the most significant parameters for discriminating between the two periods. Through a spatial analysis, the sites were classified into two groups (Groups 1 and 2). Q, total phosphorus (TP), oils, F and fecal coliform bacteria (F. coli) were the most significant parameters for discriminating between the two groups. Results suggested that TEMP, DO and Cd as functions of time, TP and F. coli as functions of space, and Q, oils and F as functions of both time and space should be monitored closely. The main sources of water pollutant were surface runoff and industrial wastewater, of time, and wastewater from agricultural irrigation, industrial wastewater, and municipal sewage, of space.

INTRODUCTION

The Jialing River watershed is a major tributary of the Yangtze River (Wu et al. 2012; Yan et al. 2015). It flows from north to south through major cities, including Baoji, Hanzhong, Guangyuan, Nanchong, and Chongqing (Wu et al. 2012). The main economic source of the Jialing River watershed is agriculture, and the growth of crops depends on irrigation (Li 2015). Jialing River covers a wide area and is an important source of drinking water. Therefore, the quality of water in the river significantly affects the life and ecological environment of residents (Yan et al. 2015).

The water quality of Jialing River has become a popular research subject due to recent industrial and urban construction developments in the area (Li 2015). River water quality assessment is necessary, especially in areas where river water serves as a drinking water source and is threatened by pollution caused by various human activities along the river course (Amadi 2011; Batayneh & Zumlot 2012). The temporal and spatial variation characteristics of river water quality can provide dynamic information for the effective management of the water environment and have become an important means of water environment management (Shrestha & Kazama 2007; Shrestha & Muangthong 2014; Muangthong & Shrestha 2015).

In recent decades, the quality of water has been heavily affected by human agricultural and industrial activities. Ali (2013) used bioassessments to evaluate the biological conditions and identify the degree of water quality degradation in the Suez Irrigation Canal and the results showed the significant impact of human activities along the canal banks on the canal ecosystem. Many researchers have also combined the water quality index (WQI) and GIS technology to reveal the patterns of different heavy metals in aquatic environments and analyze their spatial distributions through maps. Multivariate statistical techniques, such as cluster analysis (CA), discriminant analysis (DA), and principal component analysis, are used to characterize and evaluate surface water quality because they can verify temporal and spatial variations caused by natural and anthropogenic factors linked to seasonality (Helena et al. 2000; Singh et al. 2004, 2005; Muangthong & Shrestha 2015). In the present study, data sets obtained from the Guangyuan section of Jialing River were subjected to different multivariate statistical techniques to explore temporal and spatial variations in water quality. The results provide a scientific basis for ecological system management, environmental protection, and water pollution control in the Jialing River basin.

EXPERIMENTAL SITE, MATERIALS AND METHODS

Study site and monitored parameters

Guangyuan City (31° 31′ N–32° 56′ N, 104° 36′ E–106° 45′ E) is in the northern region of Sichuan, China (Figure 1). The Guangyuan section of Jialing River is located in the middle of the basin, and water quality in the river is affected by upstream water quality and significantly influences the water quality downstream (Li 2015). The population in Guangyuan is 3.05 million. The entire basin has a total area of approximately 16,319 km2, the forest cover rate of the land is 55.3%, and the cultivable land is 1,688.98 km2. The basin experiences a humid subtropical monsoon climate and annual average temperature, sunshine, and rainfall of 17 °C, 1,398 h and 1,000 mm, respectively (Guangyuan Statistics Intranet 2010). The precipitation of the Jialing River basin is unevenly distributed, with 70%–90% rainfall distributed in June–September, and the runoff mainly originates from precipitation (Long et al. 2008). Water samples were collected from seven sites, namely, Bamiaogou (BMG), Shangshipan (SSP), Zhangjiayan (ZJY), Anjiawan (AJW), Nandu (ND), Juguocun (JGC), and Yaodu (YD), at monthly intervals between January 2012 and November 2015 from the Guangyuan Environmental Monitoring Center (GEMC). BMG, SSP, and ZJY are the monitoring sites for Jialing River. ND and AJW are the monitoring sites for Nan River, which is one of the branches of Jialing River. YD and JGC are the monitoring sites for Bailong River, which also is one of the branches of Jialing River. Both Nan River and Bailong River flow into Jialing River. SSP, ZJY, ND, and AJW are located in the urban area. BMG, YD, JGC, and ZJY are located in the country, but BMG is polluted by industrial wastewater and material.

Figure 1

Map of the study area and surface water quality monitoring stations in Jialing River watershed (Guangyuan).

Figure 1

Map of the study area and surface water quality monitoring stations in Jialing River watershed (Guangyuan).

Seventeen parameters were selected based on their sampling continuity at all the selected monitoring sites. These parameters included water temperature (TEMP), flow rate (Q), pH, electrical conductivity (EC), dissolved oxygen (DO), 5-day biochemical oxygen demand (BOD5), chemical oxygen demand (CODMn), fecal coliform bacteria (F. coli), total nitrogen (TN), total phosphorus (TP), oils, ammonia nitrogen (NH3–N), fluoride (F), copper (Cu), lead (Pb), cadmium (Cd), and zinc (Zn). All of these parameters, except for Q (m3/s), pH, EC (μS/cm), TEMP (°C), and F. coli (cfu/L), are expressed in mg/L. GEMC had sampled, preserved, and analyzed these water quality parameters according to the Technical Specifications Requirements for Monitoring of Surface Water and Waste Water (HJ/T 91-2002, SEPA 2002). Environmental quality standards for surface water (GB3838-2002, in China) are shown in Table 1.

Table 1

The standard of surface water environment quality (GB3838-2002, China)

DOCODMnBOD5NH3–NTPTNCuZnFCdPboilsF. coli
7.5 0.15 0.02 0.2 0.02 0.05 0.001 0.01 0.05 200 
II 0.5 0.1 0.5 0.005 0.05 0.05 2,000 
III 0.2 0.005 0.05 0.05 10,000 
DOCODMnBOD5NH3–NTPTNCuZnFCdPboilsF. coli
7.5 0.15 0.02 0.2 0.02 0.05 0.001 0.01 0.05 200 
II 0.5 0.1 0.5 0.005 0.05 0.05 2,000 
III 0.2 0.005 0.05 0.05 10,000 

Analytical methods

Most multivariate statistical methods require variables to obey the normal distribution. Before multivariate statistical analysis, the normality of the distribution of each variable was checked by analyzing kurtosis and skewness statistical tests (Johnson & Wichern 1992; Zhou et al. 2007a). The original data demonstrated values of kurtosis ranging from −1.188 to 173.975 and skewness ranging from −0.907 to 11.942, indicating that distributions were far from normal with 95% confidence. Since most of the values of kurtosis or skewness were greater than zero, the original data were transformed in the form x’ = log10(x) (Kowalkowski et al. 2006; Papatheodorou et al. 2006). After log-transformation, the kurtosis and skewness values ranged from −1.176 to 1.801 and −0.588 to 1.406, respectively. But the kurtosis and skewness values of TN were 2.612 and −0.455, respectively, indicating that the distribution of the log-transformed TN was also non-normal, therefore it was not considered in the following study.

Multivariate analysis of the water quality data set was performed with two techniques: CA (Jarvie et al. 1998; Singh et al. 2005; Zhou et al. 2007a; Varol & Şen 2009) and DA (Johnson & Wichern 1992; Singh et al. 2005; Zhou et al. 2007a). To avoid misclassifications, CA was applied to experimental data standardized through z-scale transformation using Ward's method of linkage with squared Euclidean distance as the measure of similarity. Raw data were processed through DA and the stepwise method (Liu et al. 2003; Simeonov et al. 2003). Non-parametric tests were often used to evaluate the correlation structure between water quality parameters with non-normal distributions (Wunderlin et al. 2001; Singh et al. 2004; Shrestha & Kazama 2007). All mathematical and statistical computations were performed with IBM SPSS 23.0.

CA is an exploratory analysis and classifies objects so that each of them can be similar to others in the cluster with respect to a predetermined selection criterion. The resulting clusters of objects should exhibit high internal (within-cluster) homogeneity and high external (between-cluster) heterogeneity (Varol & Şen 2009). In the evaluation of water quality, CA is often based on sampling points and time to analyze the spatial and temporal variations in water quality or based on the evaluation index to analyze the similarity between the indexes (Singh et al. 2005; Zhou et al. 2007a, 2007b). In this study, CA was performed on the standardized data sets by using Ward's method with squared Euclidean distances as a measure of similarity (Jarvie et al. 1998; Singh et al. 2005).

DA is a multivariate statistical method that generally uses a stepwise model to construct the discriminant functions (DFs) that present important information for each group (Johnson & Wichern 1992; Singh et al. 2005). In this study, DA operates on the original data using a stepwise mode to confirm the groups found by CA and evaluate spatiotemporal variations in terms of discriminant variables. The monitoring periods (temporal) and sites (spatial) were regarded as grouping variables, and the measured parameters were regarded as independent variables (Zhou et al. 2007a). Further details of the method can be found in the study of Johnson & Wichern (1992).

RESULTS AND DISCUSSION

Temporal similarity and period grouping

Temporal CA was used to generate a dendrogram (Figure 2) that grouped the months into two clusters with similar physiochemical water quality characteristics. Period A covered May–November, which included the high-flow period (July–September) and the mean-flow period (April–May and October–November) of Jialing River. Period B covered December–April, which included the low-flow period (December–March) of the river.

Figure 2

Dendrogram showing the clustering of monitoring periods in Jialing River watershed (Guangyuan).

Figure 2

Dendrogram showing the clustering of monitoring periods in Jialing River watershed (Guangyuan).

As shown in Table 2, the value of Wilks’ lambda for the discriminant function was small (0.305), the chi-square value was high (384.064) and the p level (0.000) was below 0.05. These figures suggested that the temporal DA in this study was significant (Zhou et al. 2007a). The DFs and classification matrices (CMs) obtained from DA are shown in Tables 3 and 4, respectively. DA produced a CM with almost 91% correct assignments, a percentage that is higher than most of the correct assignments produced by backward stepwise DA (Shrestha & Kazama 2007; Zhou et al. 2007b). The temporal DA results suggested that TEMP, Q, DO, oils, F and Cd were the most significant parameters for discriminating between Periods A and B and accounted for most of the expected temporal variation in water quality.

Table 2

Wilk's lambda and chi-square values from DA of temporal variations in water quality

Fun. (s)RWilks’ lambdaChi-squareSig.
Temporal 0.834 0.305 384.064 0.000 
Fun. (s)RWilks’ lambdaChi-squareSig.
Temporal 0.834 0.305 384.064 0.000 
Table 3

Classification function coefficients for DA of temporal variation

ParametersPeriod APeriod B
TEMP 1.678 1.086 
−0.018 −0.025 
DO 11.965 13.118 
oils −29.830 23.768 
16.089 22.348 
Cd 2,101.702 263.842 
Constant −67.492 −70.701 
ParametersPeriod APeriod B
TEMP 1.678 1.086 
−0.018 −0.025 
DO 11.965 13.118 
oils −29.830 23.768 
16.089 22.348 
Cd 2,101.702 263.842 
Constant −67.492 −70.701 
Table 4

Classification matrix for DA of temporal variation

Percent correctPeriod assigned by DAa
Period APeriod B
Period A 92.9 182 13 
Period B 90.2 14 120 
Total 90.8 196 133 
Percent correctPeriod assigned by DAa
Period APeriod B
Period A 92.9 182 13 
Period B 90.2 14 120 
Total 90.8 196 133 

aChecked by cross-validation method.

The results showed that TEMP was significantly higher (p= 0.000) in Period A (21.49 °C) than in Period B (9.68 °C) and revealed a clear-cut temporal effect (Figure 3(a)) because Period A included all the warm months. A clear inverse relationship was observed between TEMP and DO. This relationship is natural because warm water saturates easily with oxygen and contains minimal DO (Varol & Şen 2009). DO has been used as an indicator of the ecological health of aquatic ecosystems (Qadir & Malik 2009). The DO content during most of the period was higher than the second national standard limit (6 mg/L, GB3838-2002) required for the protection of aquatic life (DeZuane 1997). The average Q and Cd in Period A (98.33 m3/s and 0.00019 mg/L) were higher than those in Period B (83.06 m3/s and 0.00008 mg/L, p(Q) = 0.028, p(Cd) = 0.000). Rainfall was also higher in Period A than in Period B because Period A covered the rainy months. The main sources of the water pollutant Cd were surface runoff and industrial wastewater (Li et al. 2013). The high Cd content in Period A can be related to urban runoff, traffic density, and automobile washing, which is common along stream banks. The maximum concentration of Cd (0.004 mg/L) was within the second national standard limit but higher than the WHO standard of 0.003 mg/L. Thus, Cd concentration should be controlled, and attention should be focused on waste disposal during Period A. The average concentrations of oils and F under the same pattern were lower in Period A (0.0157 mg/L and 0.0259 mg/L) than in Period B (0.0186 mg/L and 0.0287 mg/L, p(oils) = 0.017, p(F) = 0.014) and within the second national standard limit in both periods. The pollution source of oils and F may be from industrial wastewater. The government should strengthen monitoring of industrial wastewater and develop the waste disposal industries whose facilities should comply with state standards.

Figure 4

Dendrogram showing the clustering of monitoring sites in Jialing River basin (Guangyuan).

Figure 4

Dendrogram showing the clustering of monitoring sites in Jialing River basin (Guangyuan).

Spatial similarity and site grouping

The dendrogram showed that the monitoring locations may be grouped into two main clusters for the analysis of spatial variation (Figure 4). JGC, YD, and ZJY formed Group 1, and the remaining sites formed Group 2. The classification of these groups changed with the significance level. Group 2 is characterized by a larger squared Euclidean distance compared with Group 1. In Group 1, JGC, YD, and ZJY, which are in the country, exhibited lower levels of pollution than the other sites. Most of these stations receive pollution from agricultural and farm effluents. Group 2 included AJW, ND, BMG, and SSP, which have higher pollution levels than the other sites. Most of these stations receive pollution from domestic wastewater and industrial effluents located in urban areas.

Figure 3

Temporal variations: (a) temperature, (b) Q, (c) DO, (d) oils, (e) F, and (f) Cd.

Figure 3

Temporal variations: (a) temperature, (b) Q, (c) DO, (d) oils, (e) F, and (f) Cd.

As shown in Table 5, the values of Wilks’ lambda and chi-square for the discriminant function were 0.560 and 187.976, respectively, and the p level (0.000) was below 0.05. These values indicated that the spatial DA in this study had a good discriminatory ability and was significant (Zhou et al. 2007b). Tables 6 and 7 respectively show the DFs and CMs obtained from DA. Spatial DA produced CMs with 83% correct assignments, which is close to those of other reports (Shrestha & Kazama 2007; Zhou et al. 2007b). DA showed that Q, oils, TP, F, and F. coli were the discriminating parameters in space.

Table 5

Wilk's lambda and chi-square values of DA of spatial variation in water quality

Fun. (s)RWilks’ lambdaChi-squareSig.
Spatial 0.663 0.560 187.976 0.000 
Fun. (s)RWilks’ lambdaChi-squareSig.
Spatial 0.663 0.560 187.976 0.000 
Table 6

Classification function coefficients for DA of spatial variation

ParametersGroup 1Group 2
0.003 −0.015 
oils 137.139 189.239 
TP 27.843 56.590 
32.516 42.080 
F. coli 0.001 0.002 
Constant −17.913 −24.663 
ParametersGroup 1Group 2
0.003 −0.015 
oils 137.139 189.239 
TP 27.843 56.590 
32.516 42.080 
F. coli 0.001 0.002 
Constant −17.913 −24.663 
Table 7

Classification matrix for DA of spatial variation

Percent correctRegions assigned by DAa
Group 1Group 2
Group 1 75.2 106 21 
Group 2 88.8 35 167 
Total 83.0 141 188 
Percent correctRegions assigned by DAa
Group 1Group 2
Group 1 75.2 106 21 
Group 2 88.8 35 167 
Total 83.0 141 188 

aChecked by cross-validation method.

The discriminating parameters identified by spatial DA were used to evaluate different patterns associated with spatial variations in water quality (Figure 5). The average Q was higher in Group 1 (142.03 m3/s) than in Group 2 (54.74 m3/s, p= 0.000, Figure 5(a)). The average concentration of F. coli in Group 2 (3,085.11 cfu/L) was higher than that in Group 1 (2,470.92 cfu/L, p= 0.000), and both averages were above the second national standard limit (2,000, GB3838-2002). These results indicated that water quality was negatively affected by anthropogenic and animal activities (Whitlock et al. 2002). The concentrations of oils, TP and F were within the second national standard limits and lower than those reported by other authors (Zhou et al. 2007a; Varol & Şen 2009; Muangthong & Shrestha 2015). The trends of these parameters indicated that the average concentrations in Group 2 were higher than those in Group 1 (p(oils) = 0.000, p(TP) = 0.000, p(F) = 0.045, Figure 5(b)5(d)). Therefore, the management of rational pesticides and the principles of rational fertilization should be strengthened at the sites in Group 2. The sites in Group 1 were relatively far from pollution sources, and the concentration of pollutants was diluted by river confluence. Sites in Group 2 were more affected by anthropogenic activities and characterized by relatively turbulent flow, washing activities along the stream banks, urban runoff, and municipal effluents, compared with the sites in Group 1. Group 2's discharge was easily influenced by wastewater from agricultural irrigation, industrial wastewater, and municipal sewage because most of its monitoring sites were near urban areas or ‘unsewered’ villages (Shrestha & Muangthong 2014; Muangthong & Shrestha 2015). Therefore, F. coli should be monitored and controlled. Moreover, the construction and management of many municipal intercepting sewer pipes, wastewater, and garbage treatment plants should be supervised well. The wide application of agricultural machines was the source of oils and F.

Figure 5

Spatial variations: (a) Q, (b) oils, (c) TP, (d) F, and (e) F. coli.

Figure 5

Spatial variations: (a) Q, (b) oils, (c) TP, (d) F, and (e) F. coli.

CONCLUSIONS

The temporal similarity analysis of Jialing River basin showed that months can be divided into two periods: May–November (Period A) and December–April (Period B). TEMP, Q, DO, oils, F, and Cd were the discriminant variables. The temporal analysis assigned these groups with 90.8% accuracy. The average values of TEMP, Q, and Cd in Period A were higher than those in Period B. Therefore, the source of Cd should be controlled, and attention should be focused on waste disposal during Period A. The average values of DO, oils, and F in Period B were higher than those in Period A. During Period B, management should focus on managing oils and F.

The spatial similarity analysis showed that the sampling stations can be divided into two clusters. AJW, ND, BMG, and SSP, which were highly polluted regions, formed Group 2, and the remaining stations formed Group 1. Q, oils, TP, F, and F. coli were discriminant variables with 80.3% correct assignments. The average Q in Group 1 was higher than that in Group 2. The average values of oils, TP, F, and F. coli in Group 2 were higher than those in Group 1. These parameters, especially F. coli, should be controlled. Management can then establish a network of municipal intercepting sewers effectively.

In summary, management should concentrate on monitoring TEMP, Q, DO, oils, F, and Cd as functions of time and Q, oils, TP, F, and F. coli as functions of space. There were four main pollution sources including domestic sewage, industrial wastewater, agricultural fertilizer and pesticide inputs, and livestock and poultry-raising waste. The main sources of water pollutant were surface runoff and industrial wastewater as functions of time and wastewater from agricultural irrigation, industrial wastewater, and municipal sewage as functions of space. This study provides an improved understanding of the temporal and spatial variations in the Guangyuan section of the Jialing River watershed.

ACKNOWLEDGMENTS

The authors sincerely thank Guangyuan Environmental Protection Department for permission to use the data. This paper was supported by the Department of Science and Technology of Sichuan Province (CN) (2015JY0094). The opinions in this paper do not reflect the views or policies of GEMC.

REFERENCES

REFERENCES
Ali
M. M.
2013
An analysis of the impact of human activities on water quality and ecological responses in the Suez irrigation canal
.
Management of Environmental Quality
19
(
3
),
377
401
.
Amadi
A. N.
2011
Assessing the effects of aladimma dumpsite on soil and groundwater using water quality index and factor analysis
.
Australian Journal of Basic & Applied Sciences
5
(
11
),
763
770
.
Batayneh
A.
&
Zumlot
T.
2012
Multivariate statistical approach to geochemical methods in water quality factor identification: application to the shallow aquifer system of the Yarmouk basin of north Jordan
.
Research Journal of Environmental & Earth Sciences
4
(
7
),
170
177
.
DeZuane
J.
1997
Handbook of Drinking Water Quality
.
Wiley
,
New York, USA
.
GB3838-2002
2002
Environmental quality standards for surface water. In: Inspection and Quarantine of the People's Republic of China
.
Guangyuan Statistics Intranet
2010
Natural resources of Guangyuan. Guangyuan Statistics Intranet. http://www.scgytjj.gov.cn/info/23.htm.
Helena
B.
,
Pardo
R.
,
Vega
M.
,
Barrado
E.
,
Fernandez
J. M.
&
Fernandez
L.
2000
Temporal evolution of groundwater composition in an alluvial aquifer (Pisuerga River, Spain) by principal component analysis
.
Water Research
34
(
3
),
807
816
.
Jarvie
H. P.
,
Whitton
B. A.
&
Neal
C.
1998
Nitrogen and phosphorus in east coast British rivers: speciation, sources and biological significance
.
Science of the Total Environment
210–211
,
79
109
.
Johnson
R. A.
&
Wichern
D. W.
1992
Applied Multivariate Statistical Analysis
, 5th edn.
Prentice-Hall
,
Englewood Cliffs, NJ, USA
.
Kowalkowski
T.
,
Zbytniewski
R.
,
Szpejna
J.
&
Buszewski
B.
2006
Application chemometrics in river water classification
.
Water Research
40
,
744
752
.
Li
W. X.
2015
Investigation of water quality and pollution discharge in Jialing river basin
.
Water Resources Planning and Design
8
,
37
38
.
Li
C. L.
,
Hu
Y. M.
,
Liu
M.
,
Xu
Y. Y.
&
Sun
F. Y.
2013
Urban non-point source pollution: research progress
.
Chinese Journal of Ecology
32
(
2
),
492
500
.
Long
T. Y.
,
Li
J. C.
&
Liu
N. M.
2008
Adsorbed non-point source pollution load of Jialing River Basin
.
Environmental Science
29
(
7
),
1811
1817
.
SEPA
2012
Technical Specifications Requirements for Monitoring of Surface Water and Wastewater HJ/T 91-2002
.
China Environmental Science Press
,
Beijing, China
.
Shrestha
S.
&
Muangthong
S.
2014
Assessment of surface water quality of Songkhram River (Thailand) using environmetric techniques
.
International Journal of River Basin Management
12
(
4
),
341
356
.
Simeonov
V.
,
Stratis
J. A.
,
Samara
C.
,
Zachariadis
G.
,
Voutsa
D.
,
Anthemidis
A.
,
Sofoniou
M.
&
Kouimtzis
Th
, .
2003
Assessment of the surface water quality in northern Greece
.
Water Research
37
(
17
),
4119
4124
.
Wu
L.
,
Long
T. Y.
,
Liu
X.
&
Mmereki
D.
2012
Simulation of soil loss processes based on rainfall runoff and the time factor of governance in the Jialing River Watershed, China
.
Environmental Monitoring & Assessment
184
(
6
),
3731
3748
.
Wunderlin
D. A.
,
Díaz
M. d. P.
,
Amé
M. V.
,
Pesce
S. F.
,
Hued
A. C.
&
Bistoni
M. d. l. A.
2001
Pattern recognition techniques for the evaluation of spatial and temporal variations in water quality. A case study: Suquía River Basin (Cordoba, Argentina)
.
Water Research
35
,
2881
2894
.
Yan
J. L.
,
Jiang
T.
,
Gao
J.
,
Wei
S. Q.
,
Lu
S.
&
Liu
J.
2015
Characteristics of absorption and fluorescence spectra of dissolved organic matter from confluence of rivers: case study of Qujiang River-Jialing River and Fujiang River-Jialing River
.
Environmental Science
36
(
3
),
869
878
.
Zhou
F.
,
Huang
G. H.
,
Guo
H.
,
Zhang
W.
&
Hao
Z.
2007a
Spatio-temporal patterns and source apportionment of coastal water pollution in eastern Hong Kong
.
Water Research
41
(
15
),
3429
3439
.