Distribution and partitioning of heavymetals in large anthropogenically im pacted river,the Pearl River,China

Silan Liu·ZhongweiW ang·Yuanyuan Zhang·Yulong Liu·

W ei Yuan1,2·Ting Zhang3·Yujie Liu1·Ping Li4·Li He5·Jiubin Chen1

Abstract In order to evaluate the distribution and partitioningcharacteristics of heavymetals inthe large anthropogenically impacted PearlRiver Basin,the contents of‘‘anthropophile''elements(Cr,Ni,Cu,Zn,Cd and Pb,which are clearly inf luenced by human activities)were determ ined,and theirpartitioning coeff icients(Kd)between water and sediments and enrichment factors(EF)were calculated for samples collected atdifferent locationsalong the Pearl Rivermain stream.Themodif ied BCR sequential extraction procedure(proposed by the European Community Bureau of Reference in 1993),which involves the successive extraction of metals in a decreasing order of reactivity,was applied.Sediment samples from the upper,middle,and lower reacheswere included in this study.The resultsshowed that the contentofmostmetals inwaterand sediment samples gradually increases from upstream to downstream,suggesting a possible input from human activities as shown by their increasing high EF,ranged from 1.4 to 3.9 for Cu,from 1.4 to 6.7 for Zn,from 2.5 to 59.1 for Cd,and from 1.7 to 8.9 for Pb,respectively.The higher partition coeff icients(Kd)for Cr,Zn,and Pb(105—106)indicated that they were mainly transported in solid phase,while parts of Ni,Cu,and Cd were transported in dissolved phase as they display relatively lower Kd in the range of 104—105.According to the results of the BCR leaching,the percentage of non-residual fraction of heavy metals in the sediments showed a decreasing order of Cd＞Pb＞Zn＞Cu＞Ni＞Cr,implying that Cd and Pb weremore active and bioavailable compared to the other fourmetals,and thuswould be potentiallymore harm ful to the watershed ecosystem.

Keywords Pearl River·Water and sediment·Heavy metals·Partitioning·Distribution

1 Introduction

Heavy metals such as Cr,Ni,Cu,Zn,Cd,and Pb have attracted increasing attention because of their persistence geochemistry,toxicity,and possible bio-accumulation in aqueous environments and ecosystems(Niu et al.2009;Zhang etal.2017;Meng et al.2016;Liet al.2017;Odukoya and Akande 2015).As critical environmental pollutants and potential‘‘anthropophile''elements(Chen et al.2014),thesemetals can be discharged into rivers by both natural and anthropogenic processes,and be transported in rivers through various carriers such as in dissolved phase,suspended particulate matters,sediments,and organisms(Yang etal.2017a,b).During transport,heavymetalsmay undergo complex exchanges due to adsorption/desorption,precipitation/dissolution,complexation,redox reactions,and biouptake processes(Lin et al.1984;Xie et al.2012;Islam et al.2015;Xiong et al.2017),which in turn affect theirenvironmentalgeochem icalbehaviorandbioavailability.Understanding the distribution of metals among different phases is thus crucial for better evaluating the environmental impacts of these metals.Though the distribution of heavy metals has been w idely studied in largeriversdrainingintorelativelypristinebasins,including the M ississippi River(Shiller 1997;Piper et al.2006),the Amazon River(Gaillardet et al.1997;Guinoiseau etal.2017)and the Congo River(Dupre etal.1996),and small rivers like the Seine River(Chen et al.2014;Meybeck etal.2007),the geochem ical characteristics,and especially the distribution of metals in large anthropogenically impacted rivers,are rarely reported,which lim its our know ledge on the riverine geochem istry of heavy metals and their transport to the ocean.

In this study,the Pearl River,a large anthropogenically impacted river located in the south of China,wasstudied in order to characterize the distribution and the partitioning of six typical‘‘anthropophile''elements(Cr,Ni,Cu,Zn,Cd,and Pb)among different phases.Several studies reported the spatial distribution,pollution assessment,and sources of heavy metals in the water and sediments of the Pearl River.However,most studies focused only on the variations of total concentrations of certain metals either in a certain section of the river(Xu et al.2009;Tang et al.2015;Deng etal.2017;Liu etal.2017),or in one tributary(Zhou etal.2013;Liu etal.2011;Zhao etal.2012)or in theestuary(Wang etal.2004;Ip etal.2007;Lietal.2007;Fan et al.2010;Pan and Wang 2012).Although these studies allowed for a better understanding of the geochem icalbackground and the contentsand the transportsof metals in the PearlRiver(Wang etal.2007,2013;Caietal.2016),little is known about the distribution and partitioning ofmetalsamongstdifferentphases(Lietal.2007;Wu etal.2016).Therefore,further research is needed to better characterize the distribution and the transport dynamics of metals and better constrain their ecological effect.In this study,we systematicallyinvestigatedthe distribution characteristics and chem ical fractions of six typical‘‘anthropophile''metals between water and suspended sediments,and amongst different sediment phases along the downstream gradient of the Pearl River,by means of the partitioning coeff icient calculation and the BCR sequential digestion of solid samples(Tessieretal.1979;Rauretetal.1999;Quevauviller et al.1997),which can provide useful information on thegeochem icalactivity and bioavailability of heavy metals in rivers(Zhuang 2015;Yang et al.2016a,b,c;Wang et al.2016;Lin et al.2017;Qiao et al.2013;Yang et al.2016a,b,c).

2 M ethods and materials

2.1 Background of the Pearl River

The Pearl River is the third longest river in China,w ith an area of 454,000 km2w ithin a range of E102°14′—115°53′/N21°31′—26°49′(Lu et al.2009)(Fig.1).From west to east,it drains six provinces and two special adm inistrative regions in southern China,including Yunnan,Guizhou,Guangxi,Guangdong,Hunan,Jiangxi,Hong Kong,and Macao.The annual mean precipitation is 1470mm and mean annual runoff is about320 billionm3yr-1.Recently,with the implementation of China's reform and opening-up policy,thewhole basin hasencountered rapid urbanization andindustrialization,accompanyingrapideconomic development and rapid population grow th.These changes have also led to a seriesof environmentalproblems such as heavy metal pollution(Wang et al.2013;Han et al.2014;Zhao et al.2017),which potentially threatens the local ecosystem.

2.2 Sam p ling

The distribution of sampling locationswasshown in Fig.1.Twelve samples were collected from upstream to downstream in themain stream of the Pearl River during a high f low cruise in June 2015.According to the watershed features(Zhen etal.2016),the PearlRiverwasdivided into the upper reaches(fromthe springtoShilongzhen,Guangxi Province,including sampling M 1-4),m iddle reaches(from Shilongzhen to Wuzhou,GuangxiProvince,including sampling M 5-9)and lower reaches(from Wuzhou to estuary,including sampling M 10-12).Water samples and surface sediment samples were collected from each location.

The pH wasmeasured in situ by portable Multiparameter(Multi3430,WTW).Water sampleswere immediately f iltered after collection through a 0.22μm m ixed cellulose estersmembrane installed in a Tef lon f iltration system and then into previously acid-washed polypropylene(Nalgene)bottles.The f iltered solutions for traceelementsand cations analyseswere acidif ied to a pH＜2 w ith ultra-pure HNO3and stored at4°C.The sedimentsamplewas collected into a 50m l polypropylene centrifuge tube and freeze-dried by a vacuum freeze dryer(FD-1,Beijing Tianlin Hengtai Technology Co.,Ltd).

2.3 M aterials and reagents

All reagentsused in theexperiments,including HNO3,HF,CH3COOH,HONH2·HCl,H2O2,and CH3COONH4were analytical grade(Sinopharm Chem ical Reagent Co.,Ltd.,China).Deionized water obtained from the M illipore-Q water(Deionized Advantage A10,Merck M illibo)system was used in all preparations and experiments,including preparing0.11mol·L-1CH3COOH,0.5mol·L-1HONH2·HCl,8.8 mol·L-1H2O2and 1 mol·L-1CH3-COONH4for sequential extraction.In addition,sediment standard referencematerials GSD-11(GBW 07311),GSD-3a(GBW 07303a)and soil standard reference material GSS-7(GBW 07407)provided by the State Technology Supervision Adm inistration Bureau,China were used in this study.

Fig.1 Location of samples in the Pearl River,southern China

All polypropylene centrifuges tubeswere soaked in 5%HNO3for 24 h and then rinsed repeatedly w ith deionized water.The PTFE digestion bombsand Tef lon beakerswere soaked in 20%HNO3for 24 h,then soaked in deionized water and rinsed repeatedly,and f inally were placed in a ventilation cabinet to dry before using.

2.4 Solid sam ple digestion

The large mineral particles and plant impurities were removed and the particlem ixturewasuniform ly ground to 200 mesh using an agate mortar.Twelve samples were chosen for entire digestion.50 mg of sediment(and standards GSD-11,GSD-3a,GSS-7)materials were placed in pre-treated PTFE digestion bombs.0.8 m l HF and 1m l HNO3were added to the sealed bombs before heating at 170°C for48 h.Afterdigestion,0.8m lH2O2wasadded to the samples before evaporation at 120°C.Afterward,the residue was dissolved in 5m l 40%HNO3and heated at 140°C for 5 h.After cooling,the sampleswere diluted by adding deionized water for analysis.The concentration results derived from thisentire digestion were compared to those of BCR leaching experiments for recovery and quality control(see below).

2.5 Sequential extraction of heavy m etal in sedim ents

Three sedimentsamples from the upper,m iddle,and lower reaches of the Pearl River Basin(as well as standard materials GSD-11,GSD-3a,GSS-7)were selected for the modif ied BCR sequential extraction experiments(Rauret et al.1999).This procedure led to the extraction of four separate fractions:exchangeable(or weak acid solubleextractable),reducible(or iron-magnesium oxides),oxidizable(or organic-sulf ide bound),and residual.The details of the procedure are given as follows:

1.The exchangeable fraction(F1):250mg of sediment and 10m lof0.11mol·L-1acetic acid werem ixed in a 50 m l polypropylene centrifuge tube and stirred by a water bath oscillator(SHA-B,Changzhou Runghua Electric Co.,Ltd)for 16 h at constant temperature of 30°C before centrifugation(by a desktop high speed centrifuge,TG1650-WS,Shanghai Luxiangyi centrifuge instrument Co.,Ltd)at 5750 r·m in-1for 20 m in.The supernatant was carefully transferred toa Tef lon beaker.The residue was then washed w ith 10 m l of deionized water and centrifuged.The water supernatant was then recuperated and combined w ith the acetic acid supernatant.The f inal solution was evaporated to dryness at 100°C.Finally,the residue was dissolved in 10m l of 10%HNO3and transferred to a polypropylene centrifuge tube and stored at 4°C for analysis.

2.The reducible fraction(F2):10m l of 0.5mol·L-1hydroxylam ine hydrochloridewasadded to the residue derived from step(1)and the particles adhereding to the inner wall of the centrifuge tube were detached using sonication.The whole tube was stirred for 16 h at 30°C.The separation of the extract,collection of the supernatant,and rinsing of residueswere the same as described in step(1).

3.The oxidizable fraction(F3):2.5m l of 8.8mol·L-1hydrogen peroxide was slow ly added to the residue from step(2),and the tubeswere then covered and the residue was digested for 1 h at room temperature.The samples were then heated at 85°C w ith a lid for 1 h and then evaporated until the volume of solution was less than1m l.Subsequently,another 2.5m l of 8.8mol·L-1hydrogen peroxide was added to allow digestion at85°Cwith a lid for1 h beforeevaporation to near dryness.Finally,12.5m l of 1mol·L-1ammonium acetate(adjusted to pH 2.0±0.1 with HNO3)was added to the residue and immediately stirred at 30°C for 16 h.Then the extraction procedure was performed as described in step(1).

4.The residual fraction(F4):4m l HF and 4m l HNO3were added to the residue obtained from step(3)and transferred to PTFE digestion bombs.The bombswere sealed and heated at 170°C for 48 h.After cooling,4m lof hydrogen peroxidewereadded and then heated at 120°C until near dryness.Finally,10m l 10%HNO3wasadded and the f inal solution was transferred to a polypropylene centrifuge tube and stored at 4°C for analysis.

2.6 Concentration measurement and extraction recovery control

The contents of trace elements Cr,Ni,Cu,Zn,Cd,and Pb in water and different fractions derived from digestion and our sequential extraction were determ ined by inductively coupled plasma mass spectrometer(ICP-MS,Nex ION 300X,PE),and cations were determ ined by inductively coupled plasma optical em ission spectrometer(ICP-OES,Wasst-mpx,Agilent)at the State Key Laboratory of EnvironmentalGeochemistry(SKLEG),Chinese Academy of Sciences,Guiyang.The qualityfor concentration analysis was controlled by adding an external standard(Rh)andmeasurements of thewell-calibrated international standards(1640a,NIST).The relative standard deviation was less than 5%for all elements.

Thequalitycontrol for entiredigestionandthe sequential extraction procedure was assessed using blank samples,sample replicates,and standard reference materials(GSD-11,GSD-3a,GSS-7).Blank samples were treated follow ing the same proceduresw ith entire digestion and sequential extraction,and their contents of the six interest elements were negligible compared to the total mass of the six elements in samples.The standard deviations of the replicate samples were＜10%.The concentrationsofheavymetals from entire digestion demonstrated a good agreement with their reference values,and the recovery rate of Cr,Ni,Cu,Zn,Cd,and Pb were 101%±3%,101%±1%,99%±1%,91%±7%,96%±3%,and 93%±7%,respectively.

The recovery of the sequentialextraction was calculated as follows:

where C represented themass of heavymetals in different extracted fractionsof the sediment.Ctotalcontentrepresented themass of heavy metals in entire digestion.The overall recoveries ranged from72.0%to 108.1%,and were 84%±10%,91%±3%,92%±4%,83%±1%,99%±5%,and 99%±4%for Cr,Ni,Cu,Zn,Cd,and Pb,respectively.

3 Results

3.1 Concentration ofm etals in water and sediments

Total concentrations ofmetals in water and sediments are listed in Table 1 and shown in Fig.2.

Concentration in water The concentrations of heavy metals in water ranged from 0.40 to 2.17μg/kg(average:0.80μg/kg)for Cu,from 0.12 to 2.31μg/kg(average:0.62μg/kg)for Zn,from 0.17 to 0.90μg/kg(average:0.42μg/kg)for Ni,from 0.16 to 0.40μg/kg(average:0.24μg/kg)for Cr,from 0.01 to 0.08μg/kg(average:0.03μg/kg)for Pb,and from 0.005 to 0.096μg/kg(average:0.025μg/kg)for Cd in the Pearl River.

Concentration of Cr in water slightly increased before the eff luent Liujiang River(M 4),then decreased overall downstream.Concentration of the other metals in water generally increased from upstream to downstream,w ith relativelyhigherconcentrationsobserved after the conf luence of the Beijiang River(M 11)and at the estuary(M 12)(Fig.2).Of particular interest was that the Pb concentration showed a peak after the eff luent Beipanjiang River(M 2).These variations likely indicated the direct impact of tributary input on metal concentrations.

Fig.2 Heavymetal contents of water and sediments in the Pearl River,southern China

Concentration in sediments The total concentrations of heavy metals in sediments ranged from 48 to 259mg/kg(average:162mg/kg)for Zn,from 49 to 101mg/kg(average:75 mg/kg)for Cr,from 13 to 95mg/kg(average:47 mg/kg)for Pb,from 19 to 69mg/kg(average:38 mg/kg)for Cu,from 18 to 41mg/kg(average:30 mg/kg)for Ni,and from 0.2 to 3.3 mg/kg(average:1.67mg/kg)for Cd in the PearlRiver Basin.This resultwas consistentw ith the values reported in Xiong et al.(2017)on the Nanpan River in the upper reachesof the Pearl River and Caietal.(2016)on Shunde waterways in the lower reaches of the Pearl River.Individual samples or the averages in the watershed systematically exhibited lower values than previous studies(Xiong etal.2017;Caietal.2016;Niu etal.2009;Liu etal.2017;Xie etal.2012;Yang etal.2017a,b;Zhu and Wang 2012)(Table 2).These valueswere higher relative to the average values in the upper continental crust(UCC)(Rudnick and Gao 2003),w ith the exception for Cr and Ni.

The spatialdistribution of heavymetals in the sediments of the Pearl River Basin(Fig.2)showed that from upstream to downstream(M 1 to M 12),the concentration of Cu,Zn,Cd,and Pb gradually increased,while those of Cr and Nigenerally remained constant.

3.2 Enrichm ent factor(EF)of heavy metals in sedim ents

There were different factors that potentially inf luenced the enrichment or depletion of trace elements in river sediments(Chen et al.2014).The enrichment factor(EF)was widely used to discriminate between anthropogenic and natural sources and to ref lect the extent of environmental contam ination(Zhao etal.2017;Viersand Dupre 2009).In this study,the EFwas calculated by comparing w ith UCC according to:where X was the concentration of interestelement(Cr,Ni,Cu,Zn,Cd and Pb),using A l for normalization here.The concentrationsof Zn,Cu,Cd,and Pb in the sedimentswere higher relative to the average values in the upper continental crust(UCC;Rudnick and Gao 2003),with the EF values increasing fromupstreamto downstream.The concentrations of Cr and Ni in the sediments are generally comparable to the average values in the upper continental crust,w ith the EF values remaining constant from upstream to downstream.Overall,most heavy metals have enrichment factors(EF)over 1(Table 1).

3.3 Speciation distributions of heavy metals in sedim ents

The distribution of six heavy metals in different fractions of sedimentswas shown in Fig.3.Cr,Ni,Cu,and Zn were mainly distributed in residual fraction,w ith the proportion of 73%—88%,59%—67%,36%—68%,and 33%—66%,respectively.The proportion of residual fraction decreased gradually from upstream to downstream,meanwhile the total proportion of exchangeable fraction,reducible fraction,and oxidizable fraction increased gradually downstreamfor Ni,Cu,Zn,and Cd.However,the total concentrations of these four elements in the residual fraction did not decrease signif icantly downstream(Table 3),and therefore the decreasing proportion of residual fraction wasprobably caused by the increased presence ofmetals in the other three fractions.

The proportions of non-residual fraction of elements in sediments showed large variation towards downstream,w iththeorder of Pb(84.2%)＞Cd(82.2%)＞Zn(34.3%)＞Ni(32.6%)＞Cu(32.0%)＞Cr(14.6%)in the upstream,ofCd(94.7%)＞Pb(81.1%)＞Zn(65.6%)＞Cu(47.4%)＞Ni(39.4%)＞Cr(11.8%)in the midstream,andof Cd(96.0%)＞Pb(83.7%)＞Zn(67.5%)＞Cu(63.6%)＞Ni(40.6%)＞Cr(27.4%)in the downstreamof the Pearl River,respectively(Fig.3).According to these values,Cd,Pb,and to a lesserextent Zn and Cu in the sedimentsweremoreactiveatm idstream and downstream than upstream w ith a clearly increasing proportion of non-residual fractions.

4 Discussion

4.1 Potential sources of heavy metals in the Pearl River

The spatial concentration variations and relatively lower enrichment factors(＜2)of Crand Niwere suggested to be mainly controlled by natural processes rather than anthropogenic contribution.The Pb,Cd,Zn,and Cu,however,were suggested to be dom inantly enriched by anthropogenic activities,as they displayed much higher enrichment factors(＞1.5)toward downstream(Fig.4).The enrichmentof these heavymetals could be attributed to the importing of the tributaries,including the Liujiang River,Guijiang River,Hejiang River,and Beijiang River,wherea large amount of the m ining and chem ical industries and concentrated(Zhou et al.2013;Dang et al.2016;Ning et al.2017;Wang et al.2014;He et al.2015;Xu et al.2009).A lthough the main bedrock lithology in the basin changed from carbonate to silicate afterM 4(Fig.1),which would partially explain the Cr and Nibudgets in the Pearl River sediments,this lithological contribution with an increaseof claym inerals(from silicate bedrocks)would be limited to the other fourmetals as they havemuch higher enrichment factors relative to the UCC(Rudnick and Gao 2003)(Fig.4)which indeed wasmainly composed of silicates(Gaillardetetal.2003;Chen etal.2014).The abrupt increase of several metals[e.g.EF(Cr)=2.8 at the sampling pointM 3]likely indicated a local contamination or a tributary contribution into the river section(Fig.4).

Table 2 Local and regional comparison of heavy metal contents in surface sediments of Pearl River Basin w ith other studies

Fig.3 Distribution of heavy metals in different chem ical fractions in the sediments of the Pearl River,southern China

The results of Pearson correlation analysis showed that Crwasonly positively correlated w ith Ni,and therewasno signif icant correlation between Cr and other elements(Table 4;Fig.5).Cu,Zn,Cd,and Pb elements in sediments were signif icantly positively correlated(Table 4;Fig.5),andtheir contents increasedgraduallyfrom upstream to downstream,suggesting that these elements may be controlled by the same sources or processes.In fact,as discussed in the follow ing section,Cu,Zn,Cd,and Pb were all byproducts of current m ining and industrial activities,and their gradual increase downstream probablyindicated an increase of such anthropogenic input(Wen et al.2013).

Table 3 Results of sequential extraction of heavy metal in surface sediments

Fig.4 Spatialvariation ofenrichment factors(EF)ofheavymetals in the Pearl River,southern China

4.2 Partitioning ofm etals between dissolved and particu late loads

The solid-solution-interaction of heavy metals can be described in terms of partition coeff icient Kd,which is def ined as the ratio of the adsorbed and/or the total particulate concentration(Cs)to the dissolved concentration(Cw)of the same chem ical component(Benoit and Rozan 1999;Fu et al.2013),and was qualitatively used for investigating the distribution of heavy metals between water and particulate.Some researchers have described Kdas an important physicochem ical characteristic parameter of heavy metal pollution in water environment(Li et al.2011;Benoit and Rozan 1999).Under the physical and chem ical conditions of natural rivers,the partition coeff icient of heavy metals between water and particulate could potentially be used to assess them igration ability and the potential ecological effects that ensue(Chen and Zhou 1992).Though heavy metals are principally accumulated and transported in sediments in river systems(Chen et al.2014;Deng et al.2017;Xu et al.2009),dissolved metals(including colloidal form)aremore bioavailable and thus potentially display much direct environmental effect.The partition coeff icientof ametal can be calculated as follows:

where Cs(mg·kg-1)was elemental concentration of solid phase andCw(mg·L-1)was the total concentration remaining in water at equilibrium.

Themean partitioning coeff icientsof Cr,Ni,Cu,Zn,Cd,and Pb in the Pearl River were 3.4×105,9.1×104,5.4×104,4.0×105,7.1×104and 2.3×106,respectively(Table 1).The ranges of Kdfor each elementwere consistentw ith the results reported for the upper reachesof the Xijiang River(Liu et al.2017).Sim ilar f indings were also reported for the Seine River Basin in France(Chen etal.2009).The high partition coeff icients(e.g.above 105)indicated that the elementsm igrated mainly in the form of particulate phase,while the relatively lower values(e.g.below 104)would suggest that the transportation in dissolved phase is becom ing more important(Gue′guen and Dom inik 2003).A comparison of the range of the partition coeff icients for the six investigated metals was illustrated in Fig.6.The partition coeff icientof Cr,Zn,and Pb varied from 1.9×105to 5.5×105,9.4×104to 8.5×105and2.4×105to 7.0×106,respectively,indicating that these three metals were more concentrated(and m igrated)in particles,compared to the other threemetals Ni,Cu,and Cd,whose partition coeff icients ranged from 2.9×104to 1.7×105,3.2×104to8.0×104and2.4×104to 1.4×105,respectively.

Table 4 The results of Pearson correlation analysis between heavy metals in the Pearl River Basin

Fig.5 Relationship between heavy metals in the Pearl River,southern China

Fig.6 Diagrams show ing partition coeff icients of heavy metals in the Pearl River,southern China

The partitioning coeff icients showed large variations for single elements,and even some interesting trends w ith water pH(Fig.7),which may be caused by either thechange in source contribution or in situ exchange processes triggered by,for example,biological processes or adsorption/precipitation characteristics of sediments directly or indirectly controlled by pH.From upstream to downstream,the Kdvalues of Cr increased gradually as the pH decreased,show ing a signif icant negative correlation with pH,whereas Kdvaluesof Niand Cu clearly decreased with pH from upstream to downstream,indicating particulate forms were favored at high pH for those two metals.The Kdvalue of Zn,Cd,and Pb remained almost constant from upstream to downstream and showed no correlation w ith pH.These distinct variation trends suggested that,under acid conditions,Crwould be enriched in sediment toward downstream,whereas particulate Niand Cumay gradually decrease.

Fig.7 Relationship between partition coeff icients of heavy metals and the pH values in the Pearl River,southern China

From upstream to downstream,the DOC(Table 1)generally showed an increasing trend.However,Zn,as an essential nutrient for organisms,did not display a corresponding variation in its Kdvalue,thus excluding the inf luence of biological processesonmetal Kdchanges.The element partitioningmay be also inf luenced by riverwater pH.In fact,allwaters displayed a gradual decrease of the pH downstream(Fig.7),resulting from either the major lithology change from carbonate in the upper reaches to silicates in the lower reaches,ormost likely the increasing contribution of anthropogenic inputs(e.g.m ining wastes,wastewaters,industrial discharges)w ith relatively low pH(Cong and Zhao 2003).As acid condition was not conducive formetal adsorption,and under low pH condition heavymetals in the particleswould be readily released into the water,resulting in a lower Kdvalue.Therefore,the adsorptionmay impact Kdand thus partially explain the Kdtrend w ith pH,at least formetals Ni or Cu.However,the different trends illustrated by othermetals would suggest thatsuch in situ exchange effectwasnotdom inant Though further systematic study is needed to better assess the mechanisms causing the change ofmetal distribution w ith pH downstream the Pearl River,our results showed that(see the above discussion)the upstream—downstream variation ofmetal partitioning was likely derived from the contribution of different sources.

4.3 Distributions of heavy metals am ongst solid phases and the potential ecological effects

Although the contents of heavy metals in the Pearl River water were classif ied as human drinking water reference(I)according to the environmental quality standards forsurface water(GB 3838-2002),and the heavy metal contents in the sedimentwere also below the aquatic sediment quality guidelines(Chen et al.2001),heavy metals contained in sediment particles could be released into the overlying water again under certain conditions,form ing‘‘secondary pollution,''which is also likely to be potentially harm ful to the riverine ecosystem.From the above analysis results itwas clear that heavy metalswere transported in the Pearl Rivermainly in solid phase,w ithmean Kdrange from 5.4×104to 2.3×106.However,not all heavy metals contained in sediments have an environmental impact;it depended on their chem ical speciation/fraction.In the follow ing section we further discuss the distribution of heavymetals in differentsediment fractions with a goal of elucidating the proportion of adsorbed metals and their ecological effects.

In the four chem ical fractions derived from our BCR experiments,the metals in exchangeable fraction were much more active and thus easily uptaken by biota,especially when the acidity changed.The reducible fraction referred to metals in the oxides,and the elements of oxidizable fraction weremainly bound to sulf ides or organic matter.The fact that theheavymetals in these two fractions can be indirectly assim ilated by organisms indicated a relatively weak migration potential and environmental effectscompared to the f irst fraction(Hu etal.2014).In the residual fraction,metalwas considered to be a component of the crystal lattice of m inerals,and was thus not bioavailable and had a poorenvironmentalmobility(Zhang et al.2017).Therefore,the distributions of elements in variable chem ical fractions of sediments were pivotal for assessing the potential biological toxicity and environmental impact of heavy metals in aquatic ecosystems(Islam etal.2015).

The exchangeable fraction mostly referred to metals adsorbed onto sediments(Zhang etal.2017).In this study,all sixmetalsshowedanincreaseproportionof exchangeable species downstream(Fig.3),contrasted w ith thedownstreampHdecreasethat wouldstrengthen adsorption effect.This indicated that the Kdvariation was not mainly derived from riverine adsorption.Moreover,except for Cr,the exchangeable fraction of the other f ive metals was relatively low(＜20%for Zn and Cu and even＜10%for Ni,Cr,and Pb),conf irming again the limited impact of adsorption on particulatemetal budget.

Cr and Niweremainly distributed in the residual fraction,and their proportion remained quasi constant from upstream to downstream.This suggested that the distribution of Crand Niwere relatively insensitive to the variation of surrounding conditions(Nemati et al.2011).The extremely low proportion of Cr in non-residual fraction in allsamplesmay be resulted from the refractory behaviorof Cr during surfaceweathering processes and its inability to form precipitate or complex w ith carbonates or Fe/Mn oxide-hydroxide(Sundaray et al.2011).

The proportion of Cu in residual fraction decreased from upstream to downstream,whereas the proportion of nonresidual fractions increased from upstream to downstream.This suggestedthat Cubecamemoremobileand bioavailable in themiddle and downstream.In contrast to the residual fraction,the reducible fraction and both other fractions of Cu becamemore important downstream.The decrease trend of Kdwould thusmainly be controlled by the decline of Cu in residual phase.Therefore,the change of Cu Kdand its distribution among chem ical fractions were notmainly affected by the pH,butmay be inf luenced by the sources.

Sim ilar to Cu,except for the residual fraction,the reducible phase was another important Zn carrier in river sediment.This resultwas consistentw ith previous observations(Lietal.2001;Yu et al.2010)that Fe/Mn oxides might be an important transporter ofmetals.Zn(and Cu)bounded to Fe/M n oxidesmay be explained by the fact that the amorphoussesquioxideshave a greaterability to adsorb and retain heavy metals such as Zn onto their specif ic surface area(Ahdy and Youssef 2011).The similarity between the atom ic radius of Zn and Femade the above exchange possible(Ahdy and Youssef 2011).Zn in the exchangeable fraction could be interpreted by its co-precipitation with the carbonates due to similar mineral structure and ionic radius of Zn and Ca(Crocket and W inchester.1966)and Zn associated w ith calcite(either tetrahedrally coordinated Zn(II)adsorbed at the calcite surface or octahedrally coordinated Zn(II)incorporated in the calcite structure)was reported(Priadietal.2011).Parts of Zn were carried by sulf ide,which was consistentwith previous research(Qiao et al.2013).The occurrence of signif icantquantitiesof Zn sulf ide in suspendedmatterwas also proved in the anthropogenically impacted Seine River(Priadi et al.2011;Chen et al.2014).Carbonatesmay be derived from natural processes prior to joining the river system in the basin,but sulf ide was likely input by human activities(Liu et al.2017;Deng et al.2017).

Interestingly,Cd was mainly distributed in exchangeable fraction and reducible fraction,w ith the proportion of 37%—67%and 23%—40%,respectively.However,the proportion of reducible fraction decreased gradually from upstreamtodownstream,whereas the proportionof exchangeablefractionincreasedsignif icantlytoward downstream.Twomechanismsmay be responsible for this variation:(1)Cd has special aff inity w ith carbonates and would precipitate simultaneously w ith the carbonatem ineralsatalkaline environments/high pH(Ahdy and Youssef 2011),and(2)Cd has special geochem ical characteristics on the Earth‘s surface:it exists in the form of sulphide when the environmental is reduced,and in the form ofoxides in the strong oxidation environment.The Cd oxide is easily oxidized to CdSO4and then transferred into aqueous solution in the strong oxidation environment(Yang et al.2016a,b,c).In this study,the increasing percentage of exchangeable fraction was considered to be only affected by the source,because the downstream environmentwasweaker in oxidation from the perspective of the distribution of reducible fraction;therefore,Cd was easier to convert from oxide to sulf ide under these circumstances.In weak oxidation environment in them iddle and lower reaches,Cd was diff icult to oxidize from an oxidizable fraction to a reducible fraction oreven to CdSO4and then co-precipitated w ith carbonate in the form of ions.Finally,as discussed above,it was concluded that the increase in the exchangeable fraction was affected by the source rather than the transition from other fractionscaused by changes in redox conditions in the river system.The global enrichment of Cd in the surface soil layer of the whole basin would be an important contributor(Li et al.2007;Zhen etal.2016).Furthermore,the proportion of Cd in residual fraction decreased gradually from upstream to downstream,indicating that Cd in the downstream was more prone to m igrating and had thus higher pollution potential for aqueous environments.

Pb wasmainly distributed in reducible fraction,which accounted for 70%—71%of total Pb.Residual fraction accounted only for16%—19%.In fact,Pb could easily form stable complexeswith Feand Mn oxides.Pb could notonly adsorb to manganese and iron during oxide/hydroxide coprecipitation,butalso interactw ith Fe orw ith itself to form a stable and solid compound(Zhang et al.2017).Therefore,it could explain the relatively high proportion of reducible Pb.

In summary,Cd and Pb have the strongestgeochemical mobility among the six heavymetals analyzed in the Pearl River,and may bring potential ecological and biological risksonce the river conditions change.At the opposite side of the spectrum,the geochem ical activities of Cr and Ni were the weakest and have the weakest environmental effect.

5 Conclusions

In thisstudy,the distribution of six anthropophile elements(Cr,Ni,Cu,Zn,Cd,and Pb)was investigated based on the concentration measurement and the modif ied BCR leaching experimentsw ith a goal to evaluate the potential ecological and biological risks induced by these heavy metals in the Pearl River.Except for Cr,the content of other metals in water gradually increases from upstream to downstream,and the f luctuation of concentration clearly indicates that the concentration is affected by the tributary rem ittance.The fact that the contentsof Cu,Zn,Cd,and Pb in sediments increased gradually downstream and their relatively higher enrichment factors suggested an increasinganthropogeniccontributionderivedfromhuman activities,such asm ining and industry.The relative constantvariationsof Cr and Niconcentrations(and thus EFs)would indicate a main natural contribution(e.g.by weathering processes)rather than anthropogenic input.All metals displayed higher partition coeff icients(Kd)between sediment and water(in the range of 104—106),indicating that they weremainly transported in solid form.The relatively lower Kdvalues for Ni,Cu,and Cd(in the range of 104—105)implied amore potential ecological risk effectof thesemetals,as their dissolved phases aremore important.The large ranges and geographical variations of Kdfor these six metalswere likely caused by the contribution of different sources rather than in situ exchange processes.Our results showed that Kdvalues of Cr continuously increased from upstream to downstream;conversely,Ni and Cu decreased toward downstream,whereas Zn,Cd,and Pb remained almost constant.Our BCR experiments demonstrated that Cr,Ni,Cu,and Zn were mostly concentrated in residual fraction,while Cd was mainly in exchangeable fraction and Pb in reducible fraction.Except for Pb,the proportion of non-residual fractions formetals Cr,Ni,Cu,Zn,and Cd increased towards downstream,implying increasing biogeochem icalmobility and potential environmental effects of these metals in the lower Peal River reaches,a region w ith high population and important econom ic activities.Though systematic work is needed to better identify the exact contributing sources and mechanisms controlling the exchange ofmetals among different riverine phases,our study clearly demonstrated that the distribution and partitioning should be taken into account when investigating the geochem istry of metals and their potential ecological effects in a river system.

Acknow ledgem entsThis study was f inancially supported by the Natural Science Foundation of China(41561134017,U1612442,41625012,U1301231).Jianfeng Liu,YinaWang,Zuoying Yin,Ying Zhong,Yuhong Fan,and Benqing He are gratefully thanked for their logistic supports.Philippe Roux is acknow ledged for the constructive review.The anonymous reviewer is thanked for the valuable comments,which help to improve the quality of the paper.