adsorptive behaviour of arsenic in a karst subterranean stream and principal components analysis of its influencing variables: a case study at the lihu subterranean stream, guangxi province, china
adsorpcijsko obnašanje arzena v kraških podzemnih tokovih in analiza glavnih komponent spremenljivk, ki vplivajo nanj: primer podzemnega toka lihu, provinca
guangxi, kitajska
yang HUI1 & Zhang LIANKAI1,2,*
Abstract	UDC 551.444:546.19(510)
Yang Hui & Zhang Liankai: Adsorptive behaviour of arsenic in a karst subterranean stream and principal components analysis of its influencing variables: A case study at the Lihu subterranean stream, Guangxi province, China Arsenic (As) pollutants are serious threat to water ecological security and human health, especially in karst areas because of their unique hydrogeological characteristics. Physical-chemical analyses of karst water and its sediments at the Lihu subterranean stream, southwest China, were conducted by ICP-MS and xRF to elucidate the reaction mechanisms of arsenic in karst subterranean streams. The results show that inorganic arsenic comprise most of the total arsenic, while organic arsenic including monomethylated arsenic (MMA) and dimethyl arsenic (DMA) are not detected or infinitesimal. The reducing environment in the subterranean stream makes As(III) dominant and accounts for 53 % of the inorganic species. Adsorptive behaviour of arsenic occurred and the removal rates of As, As(III) and As(V) in the Lihu subterranean stream are 51 %, 36 % and 59 % respectively after a 25.6 km underground distance. To find out the main influencing factors on arsenic adsorptive process in this underground river, principal component analysis in SPSS and Minitab were applied. Seven main factors, i.e. sediment Fe (Fesed), sediment Al (Alsed), sediment Ca (Cased), particulate organic matter (POM), sediment Mn (Mnsed), water Ca2+ (Ca2+) and water HCO- (HCO-) are extracted fsfom thirteen indicators. The rank of those factors for total arsenic and As(III) is Cased > Fesed > Ca2+ > POM > Mnsed > Alsed > HCO-, while it is Fe^'J^ > Ca^'J^ > Ca2+ > Mnsed > POm';> Alsed HCO- for As(V). Of these seven factors, Fe >, Al >, Ca >, POM, Mn > and
Izvleček	UDK 551.444:546.19(510)
Yang Hui & Zhang Liankai: Adsorpcijsko obnašanje arzena v kraških podzemnih tokovih in analiza glavnih komponent spremenljivk, ki vplivajo nanj: primer podzemnega toka Lihu, provinca Guangxi, Kitajska
Arzen (As) je nevaren onesnaževalec, ki ogroža ekološko stanje voda in zdravje ljudi; še posebej v krasu zaradi njegovih edinstvenih hidrogeoloških značilnosti. Z namenom pojasnitve reakcijskih mehanizmov arzena v kraških podzemnih tokovih so bile z uporabo ICP-MS in XRF opravljene fizikalno-kemijske analize kraške vode in sedimentov v podzemnem toku Lihu na jugozahodu Kitajske. Rezultati kažejo, da večino skupnega arzena predstavlja anorganski arzen, organski arzen, vključno z monometü arzenovo kislino (MMA) in dimetil arzenovo kislino (DMA), pa ni bil zaznan. Zaradi redukcijskega okolja v podzemnem toku prevladuje As(III), ki predstavlja 53 % anorganskega tipa. Pojavilo se je adsorpcijsko obnašanje arzena in deleži As, As(III) in As(V) so se zmanjšali za 51 %, 36 % in 59 % na 25,6 km dolgem podzemnem toku Lihu. Z namenom določitve najpomembnejših faktorjev, ki vplivajo na procese adsorpcije arzena v tej podzemni reki, je bila uporabljena anli-za glavnih komponent v SPSS in Minitab. Sedem glavnih faktorjev, to so Fe v sedimentu (Fese>), Al v sedimente (Alse>), Ca v sedimentu (Cased), suspendirana organska snov (POM), Mn v sedimentu (Miis>>), Ca2+ v vodi (Ca2+) in HCO- v vodi (HCO-), je bilo povzeto iz trinajstih indikatorjev. Zaporedje teh faktorjev za skupni arzen in As(III) je Ca > > Fe > > Ca2+ > POM > Mnsed > Alsed > HCO3-, za AS(V) pa l^esed > ^-ased > Ca2+ > Mnse> > POM > Alse> > HC03-. Od teh sedmih faktorjev Fese>, Alse>, Case>, POM, Mnse> in Ca2+ spodbujajo adsorpcijo, HCO- pa
1	Yang H, Institute of Karst Geology, Chinese Academy of Geological Sciences, 50 Qixing Road, Guilin 541004, China, Karst Dynamics Laboratory, Ministry of Land and Resources & Guangxi Zhuang Autonomous Region, 50 Qixing Road, Guilin 541004, China, E-mail: yanghui-kdl@karst.ac.cn
2	Zhang L (Corresponding author), Institute of Karst Geology, Chinese Academy of Geological Sciences, 50 Qixing Road, Guilin 541004, China, Institute of Geochemistry, Chinese Academy of Sciences, 46 Guanshui Road, Guiyang 550002, China, Karst Dynamics Laboratory, Ministry of Land and Resources & Guangxi Zhuang Autonomous Region, 50 Qixing Road, Guilin 541004, China, E-mail: zhangliankai@karst.ac.cn
Received/Prejeto: 10.12.2013
Ca2+ are promoting factors for arsenic adsorption while HCO-is an inhibiting factor. Calcium and bicarbonate turn out to be the main influencing factors for water arsenic adsorption in the study area, largely because the high calcium and alkaline values in karst water. This finding is an obvious distinction compared with the research findings at a non-karst area. Keywords: karst subterranean stream, sediment, arsenic, influencing factors, principal component analysis.
jo zavira. V vodah proučevanega območja sta kalcij in hidro-genkarbonat glavna faktorja, ki vplivata na adsorpcijo arzena, predvsem zaradi visoke vsebnosti kalcija in alkalnosti v kraških vodah. Ta ugotovitev je značilna razlika v primerjavi z raziskovalnimi rezultati za nekraško območje.
Ključne besede: kraški podzemni tok, sediment, arzen, faktorji vpliva, analiza glavnih komponent.
INTRODUCTION
As an ubiquitous element in the environment, arsenic (As) is a carcinogen to both humans and animals. The arsenic contamination for water, air and soil is a significant environment health concern because of its toxicity (Ng et al. 2003). Arsenic mine drainage, a common type of pollution that forms at non-ferrous metal mining districts, is one of the most important anthropogenic arsenic sources. Gross arsenic discharged by mining activity has reached up to 72.6 % of man-made sources in the world (Han et al. 2003).
The arsenic storage in China is rich and it accounts for about 70 % of the world's total storage (xiao et al. 2008). Guangxi province, located in southern China, reserves 41.5 % of China's total arsenic (Wei & Zhou 1992). Arsenic pollutants produced by mining, mineral processing and metal chemical process in this area have polluted the soil, vegetation, surface water and ground-water through runoff, leaching and wind transportation (Segura et al. 2006, Li & Su 2001, Li et al. 2010, Zhai et al. 2008). Jian et al. 2012 has reported that the arsenic concentration in sediments of the Diao River in northern Guangxi hit 1000 mg/kg (67 times more than the background level).
The total area of the karst globally is about 22 million km2 and covers 15 % of the Earth's land surface (Nguyet & Goldscheider 2006). Guangxi province is just located in the world's largest karst contiguous distribution area (Li & Luo 1983). Karst aquifers are becoming an increasingly important resource in many countries and currently contribute one quarter of world-wide drinking water supply (Nguyet & Goldsc-
heider 2006), predicted to rise to almost 50 % in the near future (Kollarits et al. 2006). Karst systems are more complicated than sand and gravel aquifers, because of the strong karst development, slow soil forming process, substantial uplift in Cenozoic (in China), and the duplex structure in surface and underground (yuan & Cai 1988). Karst problems such as rock desertification, soil erosion, and soil degeneration cause loss of natural protective and filtration layer and therefore, the groundwater is vulnerable to being polluted (Ford & Williams 1989). Many studies on the geo-chemical behavior of arsenic in closed basin ground-water or surface water have been conducted by scholars (Ahmed et al. 2004, Bissen & Frimmel 2003, Guo et al. 2003, Guo et al. 2008, Guo et al. 2011, Jay et al. 2005, Redman et al. 2001, Savage et al. 2000, Segura et al. 2006, Smedley & Kinniburgh 2002, Smedley et al. 2002, Wang et al. 2010). Compared with those homogeneous aquifer systems, karst subterranean streams have special geological background, spatial structure, hydrodynamic conditions and water chemical characteristics, i.e. the carbonate rock geological background, uneven distribution of underground spatial structure, the strong karst dynamic process and unique hydro-chemical characteristics. These features would affect the migration of arsenic in groundwater. Unfortunately, few studies have been done in this field. In-depth research is needed and it is also conducive to reveal the arsenic environmental geochemical behavior in this well-known fragile environment.
study area and methods
SITE DESCRIPTION The Lihu subterranean stream catchment, with the karstification of 31.67 %, is located in NW Guangxi
province, southwest China (Fig.1). This river originates from Layi village and Guanxi village, wherein the peak cluster, peak forest and uvala are the main physiographic
Fig.1: Location of study sites and the geological background of Lihu subterranean stream catchment.
types. It discharges at a steep cliff of the Dagouhe River after flowing through 25.6 km of limestone hills and depressions. The basin area is 517.4 km2, and the average water flow is 4594 L/s in rainy season (June to August), while 3340 L/s in dry season (December to February). There are three tributaries in this underground river, i.e. Lizhai, Jihou and Badi subterranean streams. The mean annual precipitation in study area ranges from 1261 mm to 1628 mm. Geologically, The Lihu subterranean stream basin consists of various rocks from Carboniferous, Permian, Triassic, to Quaternary fluvial sediments. Cherty limestone is widely distributed in this basin. Geography and geological background of the study area are shown in Fig. 1.
SAMPLES AND EXPERIMENT The river surface sediments (0 to 5 cm) and water samples were collected in June 2012. Three to five equally spaced distributed samples were obtained on each sampling point. Branches, leaves, roots and other debris were picked out. Collected sediment was put into polyethylene bags that were rinsed with HNO3 (5 %) and preserved in a refrigerator (at approximately 4 °C) instantly, then freeze dried in laboratory and sieved with 2 mm nylon mesh after being triturated in agate
mortar. Concentrations and speciation of arsenic were measured by inductively coupled plasma mass spectrometer (ICP-MS) and inductively coupled plasma-atomic emission spectrometer (ICP-AES) at National Research Center for Geoanalysis, Beijing, China. The contents of major elements (Fe, Mn, Ca, Mg and Al) in sediment samples were determined by x-ray fluorescence spectrometer (xRF, Axios x). Water anions and cations were monitored by plasma spectrometer (IRIS Intrepid II xSP) at the karst geology resources environment supervision and inspection center, Ministry of Land and Resources, PRC. In situ measurement of pH, redox potential (Eh), temperature and electrical conductivity were performed by portable probes (Multi 3420 Set, Water Test Meters). All apparatus and beakers utilized throughout the study were cleaned using acid reagents and deionized water and chemicals in the analysis were of analytical grade.
DATA PROCESSING Principal component analysis and correlation analysis were performed by SPSS 13.0 software. Maps were generated by ArcGIS 9.0, and Patero diagram was drawn by Minitab 15 software.
MULTIVARIATE STATISTICAL ANALYSIS
There are many influential factors such as sediment physical-chemical properties, coexisting ions, etc., that could affect the migration and transformation of arsenic in groundwater (Singh et al. 1999, Su et al. 2009). In order to clarify the reaction mechanisms of arsenic in karst underground river, water chemical characteristics and sediment physical-chemical properties were measured in this study. Multivariate statistical analysis methods were also applied to the correlation analysis and principal component analysis (PCA). The aim was to identify the main hydrogeological factors on arsenic adsorption in karst subterranean stream.
PRINCIPAL COMPONENT ANALYSIS Impact factors, i.e. sediment iron (Fesed), sediment aluminum (Alsed), sediment magnesium (Mgsed), sediment calcium (Cased), particulate organic matter (POM), sediment manganese (Mnsed), water potassium (K+), water sodium (Na+), water calcium (Ca2+), water magnesium (Mg2+), water chloride (Cl-), water sulfate ion (SO^2-), water bicarbonate ion (HCO-) and sediment arsenic (Assed, As(III)sed, As(V)sed) were selected in this research for PCA. The contents of those indicators were listed in Tab. 1. To ensure their comparability among these different dimensions of data, dimensionless normalization was used in data processing before PCA. Through normalization processing, a new 16 x 17 matrix were attained. The weight in each column of the new matrix is the same, with a mean value of 0 and standardized deviation of 1. SPSS 13.0 software was then used for correlation analysis. The results are listed in Tab. 2. Most indexes have high Pearson correlation coefficient (r > 0.3, p < 0.05), indicating a suitable factor analysis and extractable common factors from the matrix.
In PCA, rotated eigenvalues that are higher than 1 were chosen as explanatory variables according to the total variance interpretation instead of without varimax rotation. Accumulative variance contribution of the first four principal components (PC1-PC4) that were extracted from the PCA was 76.4 % (Tab. 3). Namely, most of the information of the total variance of the original variables has been explained by the top four common factors. Correlation of common factors and the original variables before rotation are shown in Tab. 4. The correlations of the first four common factors are not statistically significant. In order to distinguish the relationship between various factors clearly, initial factor loading matrix was rotated in SPSS software. After rotation, the original variables variances were redistributed with the accumulative variance invariably. The changing of variance contribution makes common factors clear and easy to explain.
As can be seen from Tab. 4, seven variables i.e. As .,
*	sed*
As(III)sed, As(V)sed, Fesed, Alsed, Cased, POM, Mnsed, Csa2+
and HCO3- have high load on the first common factor PC before and after rotating, illustrating the high inter-correlation between those seven variables. Those seven main influencing variables that impact arsenic contents and its speciation in karst groundwater were drawn from thirteen indicators.
Fig. 2: Pareto sorts of influencing variables on a) total As; b) As(III); c) As(v).
g ö

-a C
C

-a C

b'

5


-S
Ol Ji


IS

<s
l
o		m	CO	00
r^			c^	
r^		r^	CD	
m		LO	Ln	rs
				
	LO	rs	ro	
	r^		r^	
		cK	uri	uS
rs	m		(N	m
rs	LO	no	^	(ji
r^		o^		
		r^	rri	
		rs	Ln	
			c^	
iJD	iJD	oi	uri	^
r^	00	rs	o	o
		c^		
	uS	c^		OD
			ro	
	00		O	rs
				
^		c^	ci	
O		00		
				
C3	^	^^		
r^	rs	o	Ln	CO
VD	Ol	c^		
CO	ijD	rö		MD
Ln			rN	ro
i-v
o
m		rs	rs	rs	rs	rs
r^ LO	uri ui	OD o rs	c^	r^ r^	iJD ro rs	VD no rs
rs r^	rs r^	00		r^	r^	r^
r^ m	CD	cK rs	r^	r^ rs	c^ m	cK m
r^		00 0^	(ji	ro		
uri		ori		r^		^^
o		r^ r^		rs 0^		u^
iJD	iJD					^
o r^	o	o	O r^	o		
ro	iJD	uri	r^	ijD ro		ori
rs	u^			00		
^	^		Ö			^
m	u^		rs	00 L^		
^	^^	^	Ö	CD	CD	c^
o	o r^	o	o		r^	r^
iJD	iJD rs	Ö	CO Ln	uri		
m	00	rs		rs		o	O	o		O
r^	c^			C^		r^				r^
c^		cK	c^	rs	Ln	r^	c^	iJD		
ro						rs	rs			
o	o	rN	^	O	^	O	o	O		o
			r^			r^				
		ro	c^	cK	00	00	uri			
		rN	^	^	ro	ro				rs
										
o	o	O	O	o	o	O	o	O	o	o
c^	c^	C^	c^	c^	c^	C^	c^	C^	c^	c^
r^		MD		00		CD			cK	
		Ln				ro	o			
m	un	ro	rN	ro		ro	m		r^	
ui	uo				ro	^	rs			m
c^	r^		c^	c^	r^			n^		
uri			ro		uri	r^			ori	
ro		uo	rs	ro	Ln	^	m			
r^		c^		r^						L^
uri	MD	CD		r^	r^	r^			ori	
^		ro	rs	Ln		ro				
	m	u^	u^	r^	m	r^		C^	r^	m
CD	CD	CD	CD		CD	CD	CD	CD	c^	CD
Ln			rs	Ln	^	ro	rs		rs	
r^		C^				m		r^		
uri		uri	uri	uri	cK	00	uri	uri		00
	o	ro					r^	00	ro	o
r^	c^		r^	c^	C^				C^	
CD	r^	c^	C^	CD	cK		0^	IJD	MD	
^	^			^						^
	rN	ro	^	Ln	^	l-v	00		O	
CD	O	O	O	o	o	O	o	o		1—
X	X	X	X	X	X	X	X	X	□Ž	X
					_1					
LO fN (N
■vT ro ro _ Vl^ o^ rq ^^ C^ cK O^
^^	^^	^^
r^	<—	<—
rs	CO	CO
es	CN	CN
^ ^ ^
5 o
"U
m m
TO
U
C C
ro
CT
-F
fü
fü
O U

	<u	
		
	CT	
	cz	
	'x	
3	CT	
o	TD	
	O ^	
	X 02	LJ
CT	CT
C	C
o	o
T3	T3
CT	CT
£=	£=
(3
>< ><
3 "t?
ši S O
s Ö
CO	O		u^
u^	^		
			
CD	CD	CD	CD
			
m	u^	c^	m
	m	m	
c^	t—		o^
	CD	CD	CD
m
r^ CO o r^ CO u^
^ ci ci ci

^ Ö
S^
O po CN m Tf Lo r^ m
r—
c^ c^ c^ \ c^
r^	VO O
u^	m
CD	o
f		f	f	f						
	*		*						CN	
			CD	m	m	CO			CN	
CO				CO					r^	
u^	u^	c^			CD				CD	CD
CD	CD	CD	O	CD	1	CD	CD	CD	1	1
^ i
m	m
w	LO	LO
r^ ^^ c^	^^ Ö ci C3 ci
											
	rN	LlO	O	m	vjD	VID		CN	LO		O
		vo				O		O	m		o
	r^	u^				LZ^			r^		
	CD	CD	O	o	CD	o	o	O	o	Ö	o
											
		*			*		*			*	
LO	CD	VJD		CD	CD	00			LO		^
LO	rN	c^		r^	rN		o		CN	ro	
					c^			«—	«—		
CD	o	CD	CD	CD	CD	CD	CD	CD	CD	CD	CD
Jl ro^ O
Tä

o
m CN
c^ c^
cB it
vO r^ u^
^^^^ <5 <5 <5
«—oil	I
c^	vo	Lo	Lo	m	r^
CN	Lo	Tf	r^	vo	m
C^	r^
r— c^	c^	c^	c^	c^	I
5
r^	r^
vo-^o^oc^«—	LO
LOVOfN'^tOO«— CN	Tf
^^^ c^ c^	c^
c^ c^ c^ c^ c^ I I	I
VID vo ro vo o o I o CD o I I I
r^ vo o m LO CN CN LO
VO	LO vo
o LO r^	LO
rN m o	o PO ^^ ^^
vo lo
r-^ r-^ vo o

r^ ro
Tf
CN CN CN m
un rs
< < < z
U
1.J uo
o
D
C
Tab. 3: Eigenvalues, variances and cumulative contribution rate of principal components before and after rotation.
Principal components
Before rotating
After rotating
Eigenvalues
Variance contribution (%)
Accumulating contribution (%)
Eigenvalues
Variance contribution (%)
Accumulating contribution (%)
PC1	7.07	44.16		44.16	6.34	39.65		39.65
PC2	2.43	15.20		59.36	2.26	14.14		53.79
PC3	1.54	9.62		68.99	2.13	13.29		67.08
PC4	1.18	7.37		76.35	1.48	9.27		76.35
tab. 4: Factors loading matrix before and after rotation.								
	Common factors before rotating				Common factors after rotating			
Variables	PC1	PC2	PC3	PC4	PC1	PC2	PC3	PC4
Fesed	0.89	0.09	-0.22	0.11	0.78	0.15	0.47	-0.02
Alsed	0.67	0.51	0.00	0.13	0.57	0.56	0.28	0.14
MSsed	0.41	0.12	-0.74	-0.28	0.13	-0.05	0.89	0.03
Cased	0.87	-0.19	0.23	-0.03	0.89	-0.12	0.05	0.20
POM	0.64	-0.23	0.38	0.16	0.76	-0.08	-0.22	0.05
Mn d sed	0.68	-0.05	-0.30	0.44	0.67	0.12	0.34	-0.41
Assed	0.96	-0.19	0.04	0.06	0.95	-0.10	0.22	0.06
As(lll)sed	0.92	-0.16	0.16	0.14	0.95	-0.03	0.09	0.04
AS(V)sed	0.92	-0.21	-0.10	-0.04	0.87	-0.17	0.36	0.08
K+	0.08	0.76	0.00	0.29	0.01	0.81	0.09	-0.02
Na+	0.24	0.34	0.21	-0.72	0.06	0.08	0.17	0.84
Ca2+	0.85	0.03	0.12	-0.27	0.75	-0.01	0.25	0.42
Mg2+	0.46	0.32	0.55	-0.19	0.46	0.31	-0.21	0.55
Cl-	-0.14	0.67	0.36	0.29	-0.10	0.76	-0.31	0.07
so4-	-0.12	0.72	-0.14	0.00	-0.27	0.64	0.23	0.13
hco3-	-0.39	-0.35	-0.62	-0.21	-0.61	-0.44	0.33	0.15
SORTING ANALYSIS Pareto diagram is a queuing method in some scientific statistical work. Minitab 15 software was applied to sort the seven variables mentioned at "Principal component analysis" section. The results were shown in Fig. 2. The influence degree of each factor on the arsenic shows a gradient descent. This suggests none of the factors occupy a dominant position. Arsenic in the sediment was affected by the coactions of those seven variables. The
impact on As(III) and total As decreased in the order Ca , > Fe , > Ca2+ > POM > Mn , > AK > HCO:; for
sed	sed	sed	sed	3
As(V), the order is Fesed > Cased > Ca2+ > Mnsed > POM > Al . > HCO-. What should be noted about these seven
sed	3
variables is that HCO- has negative relationships with total As, As(III) and As(V). The correlation coefficient of HCO- with total As, As(III) and As(V) are -0.51, -0.46 and -0.53, respectively as shown in Tab. 2.
ARSENIC ABSORPTIVE PROCESS IN KARST SUBTERRANEAN STREAM
The concentrations of total arsenic, As(III) and As(V) in the Lihu subterranean stream are listed in Tab. 5. Most of the sampling points' arsenic content were rich and exceed 10 ^g/L (the guideline concentration for drinking water set by the World Health Organization) (Ahmed et al. 2004). The average value is 35.76 ^g/L, slightly above the value 20.71-27.05^g/L in the Diaojiang river sediment detected by Jian et al. (2010) near study area. Mining and metallurgy processes upstream are the main
causes of arsenic pollutants. Field investigation completed by our group found that there is a small quarry and a coal mine running into the Layi and Badi cave inlet, respectively, which make the arsenic content of these two points higher than that of other's (Tab. 5). So the arsenic content in the underground river is closely related to human activities. Due to high dissolved oxygen and high Eh in surface rivers, As(V) is dominant in inorganic arsenic at surface points. The average is 76.3 %. At karst under-
tab. 5: The concentration of total As, As(III) and As(V) in Lihu subterranean stream.
Sampling points	No.	Total As /^g-L-1	As(III) /^g-L-1	As(V) /^g-L-1
Xiaochang	LH01	42.32	15.10	27.22
Layi cave	LH02	86.30	25.89	60.41
Layi Karst window	LH03	35.15	19.28	15.87
Liangfeng Cave	LH04	15.93	7.50	8.43
Qiaocun	LH05	33.60	6.32	27.28
Ganhe spring	LH06	12.50	6.00	6.50
Bachuanhe	LH07	17.22	11.05	6.17
Gantianba	LH08	25.26	15.25	10.01
Jihou	LH09	32.51	12.60	19.91
Hongxinghe	LH10	16.45	5.68	10.77
Badi	LH11	126.19	1.76	124.43
Lizhai	LH12	22.06	3.80	18.26
Xiaolongdong	LH13	15.60	7.33	8.27
tab. 6: Average content of elements in parent material and soil (Cao & Yuan 2005, Chen et al. 1999) (Wb /10 6).
Elements	Mean value in Earth's crust	Limestone parent material (Qingxudong Fm)	Dolomite parent material (Aoxi Fm)	Yellow soil (basalt)	Red soil (basalt)	Calcareous soil
Al	84100	700	3600	121700	140100	88500
Fe	70700	670	2100	170500	170200	59900
Mn	1400	160	300	1400	600	700
Ca	52900	390400	234600	700	700	16600
Mg	32000	3300	102100	2700	100	8900
ground rivers, Eh may decrease and thus the reducing environments formed, so As(V) is reduced to As(III). The average percentage of As(III) in whole underground river system reaches up to 53 %.
The line chart shown in Fig. 3 represents the changes of arsenic along the main streams of Lihu subterranean stream. Arsenic concentrations from upstream to
downstream manifested a declining trend, except for LH02 point, which is influenced by quarry waste residue. The decline of arsenic contents in the subterranean stream implies a water self-purification process in karst groundwater.
For example, the concentration of total As, As(III) and As(V) in sample point LH01 are 42.32 ^g/L,
Fig. 3: variation of inorganic arsenic along the Lihu subterranean stream.
15.10 ^g/L and 27.22 ^g/L, respectively. Those values are 2.7, 2.0 and 3.2 times higher than that of LH04, and 2.7, 2.1 and 3.3 times higher than that of LH13. The sum fluxes of total As, As(III) and As(V) at upper reaches of LH11, LH10, LH12 and LH08 (Fig. 1) are 589 kg/h,
212 kg/h, 377 kg/h and the fluxes at outlet of xiaolong-dong (LH13) are 290 kg/h, 136 kg/h and 154 kg/h, respectively. The concentrations of those three forms arsenic decreased by 51 %, 36 % and 59 % respectively after a 25.6 km distance in subterranean stream.
DISCUSSION
Compared to the research findings at a non-karst area, calcium and bicarbonate turned out to be the main influence factors for water arsenic adsorption largely because the high calcium and alkaline value in karst water. Therefore, the discussion is mainly focus on those two factors.
CALCIUM FOR ARSENIC ADSORPTION Generally speaking, the cations such as Fe, Al, Mn have a strong As retention ability and show a remarkable correlation relationships with arsenic (Manning & Goldberg 1997). The Ca can also form complexes with arsenic and then be adsorbed to the sediments surface. Calcium has a promoting effect on arsenic sorption according to Goh and Lim (Goh & Lim 2005). With those ion concentration increases, the sorption function gradually strengthens. Calcium was the most sensitive cations because it can enhance electropositivity at the adsorbent surface. Thus, it strengthens the electrostatic interactions between the arsenic anion, causing more arsenic to be adsorbed. The coexistence of cations consolidate this process (Goh & Lim 2005). Previous research results indicated that the main species of arsenic in water deposits around the Lihu subterranean stream are Fe-As and Ca-As besides residual arsenic (Res-As). The proportion of Ca-As is higher than Al-As and Fe-As (Jian et al. 2010), which is different from fluvial sediment in the non-karst area (Cui & Liu 1988, Wei et al. 1999). The main lithologic chemical composition in karst area is CaCO3. Carbonate rock can react with arsenic in weak alkaline environment and generate calcium arsenate which precipitates within the stream bottom sediments over time (Bhumbla & Keefer 1994, Jekel & Nriagu 1994).
Compared with the average chemical composition of the Earth's crust, Fe, Al and Mn contents in limestone
and dolomite are significantly lower than average crust-al elements (only 0.8 % to 21 % of the crustal median). However, the level of calcium and magnesium in karst area is 319 %~738 % of mean value of crust (Tab. 6). Moreover, Fe, Al, Mn contents in calcareous soil is only 0.35 to 1.2 times of red and yellow soil derived from basalt. yet the corresponding Ca, Mg contents is 3 ~ 89 times higher than that corresponding basalt soil. This may be the reason for increased Ca activities in karst area, and hence it can explain the Ca elements become one of the most important factors on arsenic migration in karst subterranean stream.
BICARBONATE FOR ARSENIC ADSORPTION The previous research revealed that bicarbonate (HCO-) has a negative relationship with arsenic concentration (Anawar et al. 2004, Jay et al. 2005, Smith et al. 2002, Su et al. 2009). They thought that anions such as Cl-, F-, SO42-, HCO-, H2PO- and SiO- have prohibitive function on arsenic adsorption, and this prohibitive function would be amplified with the anions concentration increasing (Jay et al. 2005, Smith et al. 2002; Su et al. 2009).
The inhibitional effect of bicarbonate for arsenic adsorption is mainly caused by the competitive adsorption between bicarbonate and arsenic. The higher competitive ability, the more restraining performance. The carbonate weathering by atmospheric CO2 at karst areas lead to the high bicarbonate concentration in water. The HCO3-can be chelated with adsorption sites and consequently hinder arsenic from being adsorbed (Smith et al. 1999). Meanwhile, the alkaline environment would slowdown the arsenic adsorption. That is why arsenic concentration in the study area expresses a negative relationship with bicarbonate.
CONCLUSIONS
There are some reasons for arsenic adsorption in karst subterranean stream. According to the study, cations (Fe,
Al, Mn, and Ca) and organic matter have acceleration effect on arsenic adsorption, which could separate the ar-
senic from water and reduce the risk of arsenic contamination. Anions (Cl- SO2- and HCO-), especially HCO-, have inhibitory effect on arsenic removal. Calcium and bicarbonate in karst terrain revealed an important role
during arsenic transportation and transformation. ttere-fore, the unique characteristics of karst should be considered during arsenic treatment in karst underground water.
ACKNOWLEDGEMENTS
The work was funded by the Natural Science Fund Project of Guangxi (2013GXNSFBA019218, 2013GXNS-FBA019217), and the Project of the China Geological Survey (12120113052500, 12120113005200). We also
thank the two anonymous peer-reviewers and the journal editors for their constructive suggestions that improved the manuscript.
REFERENCES
Ahmed, K. M., Bhattacharya, P., Hasan, M. A., Akhter,
5.	H., Alam, S., Bhuyian, M., Imam, M. B., Khan, A. A. & O. Sracek, 2004: Arsenic enrichment in groundwater of the alluvial aquifers in Bangladesh: an overview.- Applied Geochemistry, 2, 181-200.
Anawar, H. M., Akai, J. & H. Sakugawa, 2004: Mobilization of arsenic from subsurface sediments by effect of bicarbonate ions in groundwater.- Chemosphere,
6,	753-762.
Bhumbla D. K. & R. F. Keefer, 1994: Arsenic mobilization and bioavailability in soils. -In: Hutchinson T.C. & K. M. Meema (eds.) Lead, Mercury, Cadmium and Arsenic in the Environment. Scope, pp. 31-51, New York.
Bissen, M. & F. H. Frimmel, 2003: Arsenic-a review. Part II: oxidation of arsenic and its removal in water treatment.- Acta Hydrochimica et Hydrobiologica, 2, 97-107.
Cao, J. & D. Yuan, 2005: Karst ecosystem of southwest China constrained by geological setting.- Geological Publishing House, pp. 103, Beijing.
Chen, J., Hong, S., Deng, B. & M. Pan, 1999: Geographical tendencies of trace element contents in soils derived from Granite, Basalt and Limestone of Eastern China.- Soil and Environmental Sciences, 3, 161-167.
Cui, C. & Z. Liu, 1988: Chemical speciation and distribution of arsenic in water, suspended solids and sediment of Xiangjiang River, China.- tte Science of the Total Environment, 1, 69-82.
Ford, D.C. & P. W. Williams, 1989: Karst Geomorphology and Hydrology.- Academic Division of Unwin Hy-man Ltd, pp. 601, London.
Goh, K. H. & T. T. Lim, 2005: Arsenic fractionation in a fine soil fraction and influence of various anions on its mobility in the subsurface environment.- Applied Geochemistry, 2, 229-239.
Guo, H., Wang, Y. & Y. Li, 2003: Analysis of factors resulting in anomalous arsenic concentration in groundwaters of Shanyin, Shanxi province.- Environment Science, 4, 60-67.
Guo, H., Yang, S., Tang, X., Li, Y. Z. & Shen, 2008: Groundwater geochemistry and its implications for arsenic mobilization in shallow aquifers of the Hetao Basin, Inner Mongolia.- Science of the Total Environment, 1, 131-144.
Guo, H., Zhang, B., Li, Y., Berner, Z., Tang, X., Norra, S. & D. Stüben, 2011: Hydrogeological and biogeo-chemical constrains of arsenic mobilization in shallow aquifers from the Hetao basin, Inner Mongolia.- Environmental Pollution, 4, 876-883.
Han, F. X., Su, Y., Monts, D. L., Plodinec, M. J., Banin, A. & G. E. Triplett, 2003: Assessment of global industrial-age anthropogenic arsenic contamination.-Naturwissenschaften, 9, 395-401.
Jay, J., Blute, N., Lin, K., Senn, D., Hemond, H. & J. Du-rant, 2005: Controls on arsenic speciation and solid-phase partitioning in the sediments of a two-basin lake.- Environmental Science & Technology, 23, 9174-9181.
Jekel, M. & J. Nriagu, 1994: Em Arsenic in environment. Cycling and Characterization, pp 119, New York.
Jian, L., Huang, Z., Liu, Y. & Z. Yang, 2010: Particle size distribution and arsenic partitioning in sediments from a river polluted by mining.- Acta Scientiae Circumstantiae, 9, 1862-1870.
Jian, L., Li, H., Wu, L., Liu, J. & Z. Huang, 2012: Discussion on heavy metal pollution treatment of Diaoji-ang River.- Environment Science and Management, 148, 108-111.
Kollarits, S., Kuschnig, G., Veselic, M., Pavicic, A., Soc-corso, C. & M. Aurighi, 2006: Decision-support systems for groundwater protection: innovative tools for resource management.- Environmental Geology, 6, 840-848.
Li, D. & Y. Luo, 1983: Measurement of carbonate rocks distribution area in China.- Carsologica sinica, 2, 61-64.
Li, L., Zhang, G., Liu, H., Xiang, M., Wei, X. & H. Li, 2010: Distribution and mobility of Sb and As in topsoils and plants in the Dachang multi-metalliferous mine area, Guangxi, China.- Acta Scientiae Circumstantiae, 11, 2305-2313.
Li, X. & Z. Su, 2001: tte geochemical characteristics of soil elements of strata in Southwestern Guangxi.-Guangxi Sciences, 4, 301-307.
Manning, B. A. & S. Goldberg, 1997: Adsorption and stability of arsenic (III) at the clay mineral-water interface.- Environmental Science & Technology, 7, 2005-2011.
Ng, J. C., Wang, J. & A. Shraim, 2003: A global health problem caused by arsenic from natural sources.-Chemosphere, 9, 1353-1359.
Nguyet, V. T. M. & N. Goldscheider, 2006: A simplified methodology for mapping groundwater vulnerability and contamination risk, and its first application in a tropical karst area, Vietnam.- Hydrogeology Journal, 8, 1666-1675.
Redman, A., Macalady, D. & D. Ahman, 2001: A preliminary study of various factors influencing arsenic mobility in porous media: Proceedings USGS Workshop on Arsenic in the Environment.-[Online] Available from: http:/wwwbrr.cr.usgs.gov/Arsenic/ workshop.htm [Accessed 13'd April 2013].
Savage, K. S., Tingle, T. N., O'Day, P. A., Waychunas, G. A. & D. K. Bird, 2000: Arsenic speciation in pyrite and secondary weathering phases, Mother Lode gold district, Tuolumne County, California.- Applied Geochemistry, 8, 1219-1244.
Segura, R., Arancibia, V., Zuniga, M. & P. Pasten, 2006: Distribution of copper, zinc, lead and cadmium concentrations in stream sediments from the Mapo-cho River in Santiago, Chile.- Journal of Geochemi-cal Exploration, 1, 71-80.
Singh, A., Hasnain, S. & D. Banerjee, 1999: Grain size and geochemical partitioning of heavy metals in sediments of the Damodar River-a tributary of the lower Ganga, India.- Environmental Geology, 1, 90-98.
Smedley, P. & D. Kinniburgh, 2002: A review of the source, behaviour and distribution of arsenic in natural waters.- Applied Geochemistry, 5, 517-568.
Smedley, P., Nicolli, H., Macdonald, D., Barros, A. & J. Tullio, 2002: Hydrogeochemistry of arsenic and other inorganic constituents in groundwaters from La Pampa, Argentina.- Applied Geochemistry, 3, 259-284.
Smith, E., Naidu, R. & A. Alston, 1999: Chemistry of arsenic in soils: I. Sorption of arsenate and arsenite by four Australian soils.- Journal of Environmental Quality, 6, 1719-1726.
Smith, E., Naidu, R. & A. M. Alston, 2002: Chemistry of inorganic arsenic in soils: II. Effect of phosphorus, sodium, and calcium on arsenic sorption.- Journal of Environmental Quality, 2, 557-563.
Su, C., Win, H., Wang, Y., Xu, F. & S. Zhang, 2009: Arsenic adsorption behavior and influence factors in sediments of endemic arsenism diseased areas from Datong Basin.- Geological Science and Technology Information, 3, 120-126.
Wang, Y., Sun, C., Xie, X. & Z. Xie, 2010: tte genesis of high arsenic groundwater: a case study in Datong basin.- Geology in China, 03, 771-780.
Wei, L. & W. Zhou, 1992: Development of arsenic mineral resources and environment control.- Hunan Geology, 3, 259-262.
Wei, X., Wang, X., Liu, Y. & J. Tan, 1999: tte study of the adsoptive behaviour of arsenic in soil and its form distribution.- Journal o Agricultural University of Hebei, 3, 28-30, 55.
Xiao, X., Chen, T., Liao, X., Wu, B., Yan, X., Zhai, L., Xie, H. & L. Wang, 2008: Regional distribution of arsenic contained minerals and arsenic pollution in China.- Geographical Research, 1, 201-212.
Yuan, D. & G. Cai, 1988: The science of karst environment.- Chongqing Publishing House, pp. 65, Chongqing.
Zhai, L., Chen, T., Liao, X., Yan, X., Wang, L. & H. Xie, 2008: Pollution of agricultural soils resulting from a tailing spill at a Pb-Zn mine: A case study in Huan-jiang Guangxi Province.- Acta Scientiae Circum-stantiae, 6, 1206-1211.