Ecotoxicology and Environmental Safety 46, 101}107 (2000)

Environmental Research, Section B

doi:10.1006/eesa.1999.1879, available online at http://www.idealibrary.com on

Associations between Trace Metals in Sediment, Water, and Guppy,Poecilia reticulata (Peters), from Urban Streams of Semarang, Indonesia

B. Widianarko,*,1 C. A. M. Van Gestel,- R. A. Verweij,- and N. M. Van Straalen-*Faculty of Agricultural Technology, Universitas Katolik Soegijapranata, Jl. Pawiyatan Luhur IV/1, Semarang 50234, Indonesia; and

-Department of Ecology and Ecotoxicology, Vrije Universiteit, De Boelelaan 1087, 1181 HV Amsterdam, The Netherlands

Received April 19, 1999

The present study aims to answer the question whether theamount of metal in aquatic biota re6ects the concentrations inthe sediment and water, and whether the physico-chemical prop-erties of the water and sediment have any in6uence on thesuspected relationship. A study was made of 101 small streams inthe city of Semarang, Central Java, Indonesia. Data on 5shoccurrence in 63 streams of the greater Semarang indicated thatthe guppies did not avoid the highly polluted sites. No signi5cantdi4erence in body weight between sites was found. Signi5cantdi4erences were found in metal body concentrations (Pb and Zn)between 5sh collected from sites with di4erent degrees of pollu-tion. A signi5cant declining trend of Pb concentrations withincreasing organism size was observed, whereas for two othermetals, Zn and Cu, the concentrations did not depend on thebody weight. Apparently, body concentrations of these twometals are regulated and maintained at a certain concentration.For the relationships between metal concentrations in water,sediment and 5sh, water and sediment parameters, and 5sh dryweight, the presence of signi5cant multiple correlations andbivariate correlation (in 17 of 91 pairs of variables) indicatedthat, in general, abiotic parameters and body size had no in6u-ence on the metal 6ux from sediment to water, and into 5sh.Results of partial correlation analyses further suggested thatmetal concentrations in the sediments were the most importantfactor governing the metal body concentrations of 5sh. Thepresent study indicates that the guppy Poecilia reticulata fromurban streams is a potential bioindicator for urban metal pollu-tion, especially with respect to their (1) spatial distribution oversites of all pollution regimes and (2) variation in metal accumula-tion levels re6ecting the degree of pollution. ( 2000 Academic Press

Key Words: trace metals; water; sediment; Poecilia reticulata;bioindication; Semarang.

INTRODUCTION

Residues of contaminants in biota are often used to assessenvironmental quality. Measurements in biota may supple-

1To whom correspondence should be addressed. Fax: #62-24-415429.E-mail: bwid@mailcity.com.

10

ment measurements in the physical environment. For stud-ies on metals residues, several animal species have beenused, e.g., terrestrial isopods, Porcellio scaber (Hopkin et al.,1989; Dallinger et al., 1992; Hopkin, 1993); terrestrial snails,Arianta arbustorum (Berger and Dallinger, 1993); earthwormspecies (Weigman, 1991; Pizl and Josens, 1995; Kratz, 1996);Arachnida and Formicidae (Rabitsch, 1995a, b).

To provide a more comprehensive insight into the extentof metal contamination, it will be of substantial use toconnect information on the spatial distribution of metals inthe physical environment (i.e., sediment and water) withinformation from a biological matrix (i.e., "sh) in a de"nedspatial scale. The present study seeks to answer the questionwhether the amounts of metal in the biota re#ect the con-centrations in the sediment and water, and to "nd outwhether the physico-chemical properties of water and sedi-ment have any in#uence on the suspected relationship.

Previous studies on three metals, i.e., lead (Pb), copper(Cu), and zinc (Zn), revealed that the spatial distributions ofthese metals in sediments of the urban area of Semarang areas follows: (i) the highest Zn concentrations are in thecoastal grids; (ii) the highest Pb concentrations are clusteredin the center of the city; (iii) there are two hot spots, i.e., onesite in the northeast with extremely high Cu (but normal Znand Pb), and one site in the southeast with high Zn and Pb(but normal Cu). Seventy-"ve percent of the sites have lowmetal concentrations, i.e., Pb(30.2 lg/g, Zn(174 lg/g,and Cu(49.0 lg/g (Widianarko, 1997).

To classify the degree of metal contamination of eachindividual site in a single index, an evaluation based on thecalculated value of the combined contamination index,=,was proposed (details can be found in Widianarko, 1997).A site is classi"ed as unpolluted if= is less than or equal tozero (=40); a site is classi"ed as slightly polluted if= islarger than zero and less than or equal to one (04=41);a site is classi"ed as polluted if= is larger than one and lessthan or equal to two (1(=42); a site is classi"ed asheavily polluted if= is larger than two (='2). According

1

0147-6513/00 $35.00Copyright ( 2000 by Academic Press

All rights of reproduction in any form reserved.

102 WIDIANARKO ET AL.

to this classi"cation, among the 101 sites in the greaterSemarang area, 51, 36, 9, and 5 sites were categorized asunpolluted, slightly polluted, polluted, and heavily polluted,respectively (Widianarko, 1997).

The present study aimed to contribute to data on tracemetals in a tropical urban ecosystem, which until now havebeen underrepresented in global terms. There is a wide gapto be "lled, speci"cally if one considers the abundance ofdata from the industrialized world compared to the limiteddata from tropical regions (Davies, 1992). Furthermore, byincorporating data from the whole area of the city, thepresent study is taking into account not only the &&interest-ing'' areas (polluted areas) but also the unpolluted areas, inwhich usually the data are scarce.

MATERIALS AND METHODS

Sampling Sites

Sediment, water, and "sh samples were collected from 101grids (2]2 km) laid out over the whole area of Semarang,the "fth largest city of Indonesia. In each grid "ve sediment,three water, and "ve "sh (guppy) samples were collectedsystematically from the largest stream, provided the guppieswere present.

Handling and Measurement of the Samples

The "eld collection of sediment samples and its sub-sequent laboratory preparations and measurements of pHand dry matter content of these samples were described inWidianarko (1997). Prior to further analyses the sedimentpowder was kept in a polyethylene (PE) bag and stored atroom temperature. Part of this powder was used forgravimetric determination of organic matter content usinga mu%e oven with successive increases in temperature (1 h:2003C, 1 h: 4003C, 6 h: 5003C), and another part was usedfor metal analyses.

Three water samples of approximately 1 L each werecollected from each site and kept in a plastic container andtransported to the laboratory. The pH measurements ofthese samples were done using a digital pH meter (HannaInstruments, HI 8418) that was routinely calibrated usingbu!er solutions of pH 4 and 7.

Total suspended solids (TSS) content of the water wasmeasured by three successive "ltrations of 250-mL watersamples through plain "lter paper of known dry weightunder vacuum (Allen, 1988). After "ltration, the residue wasoven dried to constant weight (1053C, 30 min), put in thedesiccator for 15 min, and then weighed using an electronicmicrobalance (Ohaus AP 250 D).

To prepare water samples for metal analyses, 100 mL ofwater was "ltered using a millipore vacuum "lter and suc-cessively through plain and 0.45-mm "lter papaers. Ten

milliliters of the "ltrate was collected and acidi"ed to pH 2using concentrated HNO

3(65%) and stored in a refriger-

ator (43C).From each site three individuals of male guppy (approx-

imately of uniform size) were caught and transported to thelaboratory in PE bags with the original water. Upon arrivalin the laboratory, two body size parameters, body lengthand body weight (fresh and dry), were measured.

For the metal analyses, "sh were killed in ether andwrapped individually in aluminum foil. Dry weight deter-mination was done by putting fresh "sh samples in an oven(603C, 18 h). Prior to metal analyses these oven-dried sam-ples were stored in a desiccator.

Metal Analyses

Metal analyses were carried out at the laboratory of theDepartment of Ecology and Ecotoxicology, Vrije Univer-siteit Amsterdam. Analyses of metal concentrations in sedi-ments are described in Widianarko (1997).

Determination of metal concentrations in the water andin "sh were done using a graphite furnace atomic absorp-tion spectrophotometer (Perkin}Elmer 1100 B). No priortreatment was needed for the determination of metals in theacidi"ed water samples. Oven-dried "sh samples were diges-ted in a mixture of HNO

3and HClO

4(Ultrex grade, 7:1).

The digestion was done according to the method of VanStraalen and Van Wensem (1986). Prior to the measurement,the pellet remaining after digestion was dissolved in 1 mL of0.1 M HNO

3(Ultrex grade).

Data Analysis

Several statistical techniques were applied in the dataanalysis, including a G test, analysis of variance (ANOVA)followed by appropriate post hoc comparison of means,linear regression, multiple linear correlation, bivariate cor-relation, and partial correlation. All computations wereundertaken using two statistical software packages for PCunder Windows environment, i.e., Statistica Version 4.5(1993) and SPSS Version 6.1. (1994).

A G test of independence (Sokal and Rohlf, 1995) wasused to establish whether the presence of the guppy,P. reticulata (Peters), was dependent on the degree of pollu-tion of the streams.

A one-way ANOVA with unbalanced data (see Zar, 1984)was used to test the di!erences in body weight, metal con-centration, and metal body burden of the guppy, betweenurban streams with di!erent degrees of metal contamination(expressed in= values). To ful"ll the normality assumption,a logarithmic transformation was applied to the data priorto the ANOVA computation (Steel and Torrie, 1981).

A linear regression technique (Sokal and Rohlf, 1995) wasused to assess the allometric relationship between metal

TRACE METALS IN SEDIMENT, WATER, AND GUPPY IN INDONESIA 103

body concentration and dry weight expressed as

C"aXb or log C"log a#b log X, (1)

where X is dry weight (mg), C is metal body concentration(lg/g), and a and b are constants (regression parameters).

Three di!erent correlation techniques, i.e., bivariatecorrelation, multiple correlation, and partial correlation,were used to evaluate the relationships present among 14variables in this study. These variables include metalconcentrations in sediment, water, and "sh; pH and totalsuspended solid content of the water; pH and organicmatter content of the sediment; and "sh dry weight.

Partial correlation was used to overcome the inability ofbivariate correlation to take into account the interactions ofany of the other variables on the two variables in question.Partial correlation solves this problem because it considersthe correlation between each pair of variables while holdingthe value of one of the other variables constant (Sokal andRohlf, 1995; Zar, 1984; Nurosis, 1994).

Decisions on the degree of signi"cant correlations amongall the variables in this study were done following theprocedure in Chat"eld and Collins (1980), which is basedon the magnitude of the correlation coe$cient. Large cor-relations are those whose absolute value exceeds 0.70 andsmall correlations are those whose absolute value is lessthan 0.25.

RESULTS

From 101 sampling sites the "sh P. reticulata was presentonly at 63 sites. According to the site classi"cation based onthe value of the combined contamination index, =, thesesites fell into four categories, i.e., 30 unpolluted, 20 slightlypolluted, 8 polluted, and 5 heavily polluted sites. Results ofthe G test revealed that the presence of "sh was not depen-dent on the degree of metal pollution of the sites (Table 1).All heavily polluted sites and most of the polluted ones had"sh; slightly polluted and nonpolluted sites had "sh inaround 50% of the cases.

TABLDistribution of Fish (Poecilia reticulata) Occurrence over

Degree of meta

Fish presenta Nonpolluted Slightly polluted

No 21 16Yes 30 20

Total 51 36% Presence 58.8 55.6

a G value"8.92 (P(0.05).

Results of ANOVA indicated that there was no signi"cantdi!erence in body weight between sites, but there weresigni"cant di!erences in metal body concentration andbody burden between "sh collected from sites with di!erentdegrees of pollution (Table 2). For the metal concentrations,signi"cant di!erences were found for Pb and Zn. For thebody burdens, only Pb exhibited a signi"cant di!erence. Nosigni"cant di!erence was found for Cu presented either asconcentration or as body burden.

For evaluation of the allometric relationship betweenmetal body concentration and body weight, only "sh fromnonpolluted and slightly polluted streams were used. OnlyPb concentrations revealed a signi"cant association withthe body weight (Fig. 1a). Pb concentrations in "sh tendedto decrease with the increase of body weight. Concentra-tions of two other metals, Zn and Cu, appeared to beindependent from body weight. The linear regressionmethod did not provide an acceptable "t for the relationshipbetween the concentration of these metals (Zn and Cu) andbody weight of "sh (see Figs. 1b and 1c; the insigni"cant"tted lines were plotted as dashed lines). A highly signi"cant(P(0.01) regression coe$cient was found for Pb. For Znand Cu the estimated values of the regression coe$cientswere not signi"cantly di!erent from zero. Parameter esti-mates of the linear regressions and their standard errors arelisted in Table 3.

A signi"cant coe$cient of multiple correlation (P(0.05,R"0.61) was found for the multiple association among 14variables, i.e., Pb concentrations in sediment, water, and"sh; Cu concentrations in sediment, water, and "sh; Znconcentrations in sediment, water, and "sh; pH andtotal suspended solids content of the water; pH andorganic matter content of the sediment; and "sh dry weight.The application of the simple bivariate correlation onthese variables resulted in 91 bivariate correlationcoe$cients (Table 4). Only 17 of these 91 coe$cients weresigni"cant.

From Table 4 it can be seen that concentrations of allmetals (Pb, Cu, and Zn) in the body of P. reticulata weresigni"cantly correlated with the concentrations of the

E 1Sites with Di4erent Degrees of Metal Contamination

l contamination

Polluted Heavily polluted Total

1 0 388 5 63

9 5 10188.9 100

TABLE 3Parameter Estimates for Linear Regression between Metal

Body Concentration and Body Weight of Fish (Poecilia reticu-lata) from Nonpolluted and Slightly Polluted Streams

Parameter ($standard error)

Metal Intercept Slope coe$cient

Lead 0.84$0.19 !0.80$0.15**Zinc 2.41$0.07 !0.07$0.05Copper 0.02$0.12 0.16$0.09

**Regression coe$cient signi"cantly di!erent from zero (P(0.01).

TABLE 2Body Weight, Metal Concentration, and Metal Body Burden of Poecilia reticulata from Streams with Di4erent Degrees

of Metal Contamination

Degree of metal contamination*

Parameter Nonpolluted Slightly polluted Polluted Heavily polluted

No. of samples 150 100 40 25Body weight** (mg) 20.1$4.7 18.3$5.3 16.9$3.9 19.0$4.8Concentration (lg/g)

Lead 0.8a$0.7 1.2b$0.6 3.2c$2.8 6.9c$7.7Copper 3.4$1.1 2.6$1.0 4.6$5.8 6.4$9.3Zinc 207a$39.2 223b$43.6 343c$238 271bc$74.4

Body burden (ng/"sh)Lead 15.4a$13.7 21.3b$12.1 47.8c$34.1 135c$177Copper 69.5a$33.5 49.1b$25.0 91.9ab$150 87.4ab$94.5Zinc 4111$1055 4052$1342 6308$6548 5274$2410

Note. All values are mean$standard deviation. Di!erent superscripts show signi"cant di!erences (P(0.01), based on the LSD test following one-wayANOVA of log-transformed data.

*The degree of metal contamination of each individual site was classi"ed according to the calculated value of combined contamination index,=, basedon metal concentration in the sediment.

**Measured as dry weight.

104 WIDIANARKO ET AL.

respective metals in the sediment. Large correlation coe$-cients ('0.7) were exhibited Pb and Cu, whereas Zn re-vealed a small correlation coe$cient ((0.25). A signi"cantcorrelation between metal concentration in the water and inthe "sh was found only for Cu.

It is interesting to note that concentrations of Zn in "shwere correlated with Pb and Cu in "sh, and with Pb insediment. Zn concentrations in sediment were correlatedwith concentrations of the two other metals, Pb and Cu, inthe same medium. Pb in water was negatively correlatedwith pH of the water. Signi"cant correlations were alsopresent between Zn in water and the corresponding concen-trations of Pb and Cu. Concentrations of Cu in water,however, were correlated with Pb in water and organicmatter content of the sediment. Finally, pH of the sedimentwas signi"cantly correlated with pH of the water.

In order to identify any hidden interactions that cannotbe taken into account by the simple bivariate correlationtechnique, a partial correlation analysis was conducted. Bycontrolling sediment pH, organic matter content of thesediment, water pH, and total suspended solids contentof the water, it was found that only three signi"cant cor-relations remained, namely between concentrations of themetals in "sh (Pb, Cu, and Zn) and their correspondingconcentrations in sediment (see Table 5). In all cases, thevalues of partial correlation coe$cients were slightly largerthan the corresponding values for bivariate correlation.When the inclusion of controlled variables was done step-wise the same results were revealed.

DISCUSSION

The data on "sh occurrence in 63 streams of the greaterSemarang indicated that guppies (P. reticulata) did notavoid the highly polluted sites. Apparently, "sh are attrac-ted by the organic waste at these sites and can survive themetal pollution. This "sh species has been known to bewell-established in polluted water with a high amount oforganic waste in South East Asia (Chou and Lam, 1989).This implies that guppy "sh is a promising bioindicatorspecies. The presence of guppies in polluted sites will allowthe determination of changes in various exotoxicologicalparameters induced by metal pollution.

No simple interpretation can be made on signi"cant dif-ferences in metal concentrations (Pb and Zn) and metal

FIG. 1. (a) Regression plot of internal Pb concentrations and bodyweight of Poecilia reticulata. (b) Regression plot of internal Zn concentra-tions and body weight of P. reticulata. (c) Regression plot of internal Cuconcentrations and body weight of P. reticulata.

TRACE METALS IN SEDIMENT, WATER, AND GUPPY IN INDONESIA 105

burdens (Pb) between "sh from sites with di!erent contami-nation levels. One possible explanation might be that "shfrom polluted and highly polluted streams have developeda physiological adaptation by accumulating more metalswhen facing excessive metal concentrations in their environ-ment. Studies on metal accumulation in animals from siteswith di!erent distances from an emission source indicatedsuch a tendency (see, e.g., Hopkin et al., 1989; Dallingeret al., 1992; Berger and Dallinger, 1993; Pizl and Josens,1995).

A signi"cant declining trend of Pb concentrations withincreasing organism size was observed. This accumulationphenomenon is one of the three types of metal concentra-tion-size relationships that include increases, decreases, andno change in concentration with increased body weight asdemonstrated by several studies on aquatic invertebrates(Smock, 1983). Ray et al. (1980) reported that small indi-viduals of the polychaete Nereis virens accumulated largeramounts of Cd per unit of body weight than larger indi-viduals, in association with a larger uptake rate of Cd insmall individuals. The decreasing metal concentration withbody size might indicate that surface adsorption is an im-portant mode of accumulation (Smock, 1983).

Clearly, for the two essential metals, Zn and Cu, theconcentrations were not dependent on body weight. Appar-ently, body concentrations of these two metals are regulatedand maintained at a certain concentration. The existingtheory on the in#uence of body size on the accumulation oftrace metals in animals as exempli"ed by several bioac-cumulation studies on di!erent aquatic and terrestrial spe-cies (e.g., Moriarty et al., 1984; Janssen and Bedaux, 1989;Krantzberg, 1989; Van Hattum et al., 1991; Mersch et al.,1996) does not appear to hold for essential metals.

For the relationships between metal concentrations inwater, sediment and "sh, water and sediment parameters,and "sh dry weight, the presence of signi"cant multiplecorrelations and bivariate correlations (in 17 of 91 pairs ofvariables) indicates that, in general, abiotic parameters andbody size have no in#uence on metal #ux from sediment towater, and into "sh. It can be inferred that the signi"cantmultiple associations among 14 variables in the presentstudy were mostly dominated by the association betweenmetal concentrations in sediments and those in the "sh. Thisis in contrast with other bioaccumulation studies that haveclearly demonstrated the role of abiotic parameters in theuptake of toxicants, including metals, by animals (see, e.g.,Schrap and Opperhuizen, 1989; Van Hattum et al., 1991;Landrum and Faust, 1991; Van Hattum et al., 1993; Ankleyet al., 1994; Bervoets et al., 1996).

Based on the results of simultaneous bivariate correlationanalysis, this study demonstrated some signi"cant relation-ships that are relatively self-evident, such as relationshipsbetween concentrations of all metals (Pb, Cu, and Zn) insediment and the corresponding metal concentrations in

TABLE 4Correlation between Metal Concentrations in Guppies and Concentrations in and Properties of Corresponding Water and Sediment

Variable PbFish PbSed PbWat CuFish CuSed CuWat ZnFish ZnSed ZnWat DWFish OMSed PHSed PHWat TSS

PbFish 1PbSed 0.758* 1PbWat !0.209 !0.053 1CuFish 0.177 0.014 !0.076 1CuSed 0.348* 0.081 !0.163 0.724* 1CuWat !0.039 !0.033 0.414* 0.266* 0.232 1ZnFish 0.271* 0.343* 0.057 0.504* 0.066 0.035 1ZnSed 0.696* 0.753* !0.165 0.019 0.282* !0.068 0.243* 1ZnWat !0.017 !0.021 0.393* 0.049 0.031 0.327* !0.06 !0.03 1DWFish !0.125 0.107 0.224 0.011 !0.208 0.163 0.009 !0.04 !0.03 1OMSed !0.053 !0.064 !0.118 !0.238 0.042 !0.294* 0.156 !0.144 !0.22 !0.16 1PHSed !0.021 0.031 !0.15 !0.094 !0.116 !0.154 0.083 0.04 !0.1 0.035 0.032 1PHWat 0.018 0.114 !0.265* !0.196 !0.141 !0.145 !0.059 0.043 !0.14 !0.09 0.046 0.413* 1TSS !0.111 0.078 0.085 0.019 !0.115 !0.024 !0.113 !0.18 0.182 0.056 !0.03 0.06 0.147 1

Note. PbFish, CuFish, and ZnFish are lead, copper, and zinc concentrations in the guppy, respectively. PbSed, CuSed, and ZnSed are lead, copper, andzinc concentrations in the sediment, respectively. PbWat, CuWat, and ZnWat are lead, copper, and zinc concentrations in the water, respectively. OMSedand PHSed are organic matter content and pH of the sediment, respectively. PHWat is pH of the water, and TSS is total suspended solids.

*Signi"cant (P(0.05).

106 WIDIANARKO ET AL.

"sh, and between Cu concentrations in the water and Cuconcentrations in the "sh, or between pH of the sedimentand pH of the water. On the other hand, some other signi"-cant relationships are not so trivial, e.g., relationshipsbetween Zn concentrations in all compartments, i.e., sedi-ments, water, and "sh, and concentrations of the two othermetals (Pb and Cu) in the corresponding compartments, orbetween concentrations of the two other metals (Pb and Cu)in the corresponding compartments, or between concentra-tions of Cu and Pb in water. It seems that, although the

TABLE 5Partial Correlation between Metal Concentrations in Poecilia

reticulata, in Sediment, and in Water (Controlling for pH andOrganic Matter Content of Sediment; pH and TSS of Water)

Metal Variable Sediment Water Fish

Lead Sediment 1Water !0.03 1Fish 0.77** !0.21 1

Copper Sediment 1Water 0.23 1Fish 0.76** 0.19 1

Zinc Sediment 1Water 0.08 1Fish 0.26* !0.11 1

*Signi"cant (P(0.05).**Highly signi"cant (P(0.01).

spatial distribution is speci"c for each metal (cf.Widianarko, 1997), on the average the metals occur inassociation with each other.

Partial correlation analysis was applied, basically, toidentify and to isolate the most important confoundingvariables that possibly have some in#uence on the #ux ofmetals between the di!erent compartments in the ecosystemstudied. Results of partial correlation analyses suggest thatmetal concentrations in the sediments are the most impor-tant factor governing the metal body concentration of "sh,since none of the abiotic parameters of the sediment andwater played signi"cantly as a covariate.

CONCLUSION

Results of the present study indicate that the guppyP. reticulata from urban streams is a potential bioindicatorfor urban metal pollution, especially with respect to its(1) spatial distribution over sites of all pollution regimes and(2) variation in metal accumulation related to the degree ofpollution. Accordingly, this "sh species can be considereda suitable candidate for biomonitoring programs.

ACKNOWLEDGMENT

Funding for this research was provided jointly by the UniversitasKatolik Soegijapranata and Department of Ecology and Ecotoxicology,Vrije Universiteit, Amsterdam. The authors are thankful to Mr. FelixSholeh Kuntoro and Mr. Novi Susetyo Adi for the assistance in thecollection and preparation of the samples.

TRACE METALS IN SEDIMENT, WATER, AND GUPPY IN INDONESIA 107

REFERENCES

Allen, S. E. (1988). Chemical Analysis of Ecological Materials. BlackwellScienti"c, Oxford.

Berger, B., and Dallinger, R. (1993). Terrestrial snails as quantitativeindicators of environmental metal pollution. Environ. ¹oxicol. Assess. 25,65}84.

Bervoets, L., Blust, R., and Verheyen, R. (1996). Uptake of zinc by themidge larvae Chironomus riparius at di!erent salinities: Role of speci-ation, acclimation, and calcium. Environ. ¹oxicol. Chem. 15, 1423}1427.

Chat"eld, C., and Collins, A. J. (1980). Introduction to Multivariate Analy-sis. Chapman and Hall, London.

Chou, L. M., and Lam, T. J. (1989). Introduction of exotic aquatic species inSingapore. In Proceedings of a =orkshop on Introduction of ExoticAquatic Organisms in Asia (S.S. Silva, Ed.), Vol. 3, pp. 91}97. Spec. Publ.,Asian Fish Soc.

Dallinger, R., Berger, B., and Birkel, S. (1992). Terrestrial isopods: Usefulbiological indicators of urban metal pollution. Oecologia 89, 32}41.

Davies, B.E. (1992). Trace metals in the environment, retrospect andprospect. In Biogeography of ¹race Metals in Invertebrates (D. C.Adriano, Ed.), pp. 383}399. Lewis, Boca Raton, FL.

Hopkin, S. P. (1989). Ecophysiology of Metals in ¹errestrial Invertebrates.Elsevier Applied Science, London.

Hopkin, S. P., Hames, C. A. C., and Bragg, S. (1989). Terrestrial isopods asbiological indicators of zinc pollution in the Reading area, South Eng-land. Monitore zool. ital. (N.S.) Monogr. 4, 477}488.

Janssen, M. P. M., and Bedaux, J. J. M. (1989). Importance of body-size forcadmium accumulation by forest litter arthropods. Neth. J. Zool. 39,194}207.

Krantzberg, G. (1989). Metal accumulation by chironomid larvae: Thee!ects of age and body weight on metal body burdens. Hydrobiologia188/189, 497}506.

Kratz, W. (1996). Ecotoxicological bioindication: PAH, PCB and heavymetals studied in the natural resource monitoring programme of Berlin(Germany). In Bioindicator Systems for Soil Pollution (N. M. Van Straalenand D. A. Krivolutsky, Eds.), pp. 223}232. NATO ASI Series, KluwerAcademic, Dordrecht.

Landrum, P. F., and Faust, W. R. (1991). E!ect of variation in sedimentcomposition on the up take rate coe$cient for selected PCB and PAHcongeners by the amphipod Diporeia sp. In Aquatic ¹oxicology and RiskAssessment (M. A. Mayes and M. G. Barron, Eds.), Vol. 14, pp. 263}279.ASTM STP 1124, Philadelphia.

Mersch, J., Wagner, P., and Pihan, J.-C. (1996). Copper in indigenous andtransplanted zebra mussels in relation to changing water concentrationsand body weight. Environ. ¹oxicol. Chem. 15, 886}893.

Moriarty, F., Hanson, H. M., and Freestone, P. (1984). Limitations of bodyburden as an index of environmental contamination: Heavy metals in"sh Cottus gobio L. from the River Ecclesbourne, Derbyshire. Environ.Pollut. A 34, 297}320.

Norusis, M. J. (1994). SPSS Advanced Statistics 6.1. SPSS, Chicago.

Pizl, V., and Josens, G. (1995). Earthworm communities along a gradient ofurbanization. Environ. Pollut. 90, 7}14.

Rabitsch, W. B. (1995a). Metal accumulation in arthropods near a lead/zincsmelter in Arnoldstein, Austria. II. Formicidae. Environ. Pollut. 90,239}247.

Rabitsch, W. B. (1995b). Metal accumulation in arthropods near alead/zinc smelter in Arnoldstein, Austria. III. Arachnida. Environ. Pollut.90, 249}257.

Ray, S. McLeese, D., and Pezzack, D. (1980). Accumulation of cadmium byNereis virens. Arch. Environ. Contam. ¹oxicol. 9, 1}8.

Schrap, S. M., and Opperhuizen, A. (1989). Preliminary investigations onthe in#uence of suspended sediments on the bioaccumulation of twochlorobenzenes by the guppy (Poecilia reticulata). Hydrobiologia188/189, 573}576.

Smock, L. A. (1983). Relationships between metal concentrations andorganism size in aquatic insects. Freshwater Biol. 13, 313}321.

Sokal, R. R., and Rohlf, F. J. (1995). Biometry,¹he Principles and Practice ofStatistics in Biological Research. 3rd ed. Freeman, New York.

Van Hattum, B., Korthals, G., Van Straalen, N. M., and Govers, H. A. J.,and Joosse, E. N. G. (1993). Accumulation patterns of trace metals infreshwater isopods in sediment bioassays*In#uence of substrate charac-teristics, temperature, and pH.=at. Res. 27, 669}684.

Van Hattum, B., Timmermans, K., and Govers, H. A. J. (1991). Abiotic andbiotic factors in#uencing in situ trace metal levels in macroinvertebratesin freshwater ecosystems. Environ. ¹oxicol. Chem. 10, 275}292.

Van Straalen, N. M., and Van Wensem, J. (1986). Heavy metal content offorest litter arthropods as related to body-size and tropic level. Environ.Pollut. A 42, 209}221.

Weigman, G. (1991). Heavy metal levels in earthworms of a forest ecosys-tem in#uenced by tra$c and air pollution.=at. Air Soil Pollut. 57+58,655}663.

Widianarko, B. (1997).;rban Ecotoxicology: Spatial and ¹emporal Hetero-geneity of Pollution. Ph.D. Thesis, VU Amsterdam.

Zar, J. H. (1984). Biostatistical Analysis. Prentice-Hall, Englewood-Cli!s,NJ.