Effects of industrial metals on wild fish populations along a metal contamination gradient

ARTICLE IN PRESS

0147-6513/$ - se

doi:10.1016/j.ec

�CorrespondE-mail addr

Ecotoxicology and Environmental Safety 61 (2005) 287–312

www.elsevier.com/locate/ecoenv

Rapid Communication

Effects of industrial metals on wild fish populations along a metalcontamination gradient

Gregory G. Pylea,�, James W. Rajotteb,c, Patrice Couturec,d

aDepartment of Biology, Nipissing University, North Bay, 100 College Drive, Box 5002, North Bay, Ont., Canada P1B 8L7bDepartment of Zoology, University of Guelph, Guelph, Ont., Canada N1G 2W1cDepartment of Biology, Laurentian University, Sudbury, Ont., Canada P1E 2C6

dINRS-ETE, 490 de la Covronne, Que., Canada G1K 9A9

Received 4 July 2004; received in revised form 2 September 2004; accepted 15 September 2004

Available online 24 November 2004

Abstract

The purpose of this study was to examine relationships among water, sediment, and fish tissue metal concentrations as they relate

to fish diversity, tissue metal accumulation, and fish morphometric and reproductive condition. Fish were captured in 12 lakes near

Sudbury, Ontario, Canada, that ranged in their degree of metal contamination. In general, metal concentrations in water and

sediment decreased with increasing distance from industrial operations. However, only Cu and Ni demonstrated this trend in

sediments. Although 20 fish species were identified in the 12 lakes, only one species, yellow perch (Perca flavescens), was common to

all 12 lakes. Fish diversity was only associated with sediment metals, suggesting that short-term processes are much less important

than long-term processes for fish community recovery in metal-contaminated lakes. Multivariate characterization of water metal

concentrations resulted in three lake clusters: Group 1 consisted of reference lakes; Group 2 lakes had high alkalinity, conductivity,

hardness, pH, waterborne metals (especially Se), and sediment Cu and Ni concentrations; Group 3 lakes had high pH, waterborne

and sediment Cu, and sediment Ni, intermediate alkalinity, conductivity, and waterborne metals (except Al and Fe), and low

hardness and waterborne Al and Fe. Liver Cd, Cu, Ni, Pb, and Zn, muscle Zn, and intestinal Cd and Zn were highest, and muscle

Cu and male gonadosomatic index (GSI) were lowest, in Group 3 fish. Liver, muscle, and intestinal Se concentrations, and Fulton’s

condition factor (FCF), hepatosomatic index (HSI), and male GSI were highest in Group 2 fish. Group 1 fish had the highest muscle

Hg concentrations and female GSI. Muscle Se appeared to have an antagonistic effect on muscle Hg accumulation as a function of

distance from smelting operations. Neither Cu nor Ni, both metals of concern in the Sudbury area, was useful for predicting fish

condition, probably because of homeostatic regulatory control. Liver Cd accumulation, which was negatively related to FCF

(r ¼ �0:16; Po0:05), exhibited strong, nonlinear inhibition (r2 ¼ 0:99; Po0:0001) as a function of water hardness. Because Cd wasnot detected in water samples in this study, we suspect that branchial Ca2+ uptake may play some role in reducing dietary Cd

uptake in hard water lakes. Selenium has received relatively little attention in the contaminated systems around Sudbury, yet our

results demonstrated that tissue Se was related to all condition metrics studied. Moreover, evidence was provided that suggests that

there is a gender-specific interaction between dietary Se and Cu uptake that may contribute to decreased female reproductive

condition in wild yellow perch.

r 2004 Elsevier Inc. All rights reserved.

Keywords: Wild yellow perch (Perca flavescens); Metal pollution; Multivariate analysis; Metal interactions; Accumulation; Fish condition;

Environmental effects; Fish diversity

e front matter r 2004 Elsevier Inc. All rights reserved.

oenv.2004.09.003

ing author. Fax: +1705 474 1947.

ess: [email protected] (G.G. Pyle).

1. Introduction

Sudbury, Ontario, is home to some of the mostproductive nickel and copper mining operations in theworld. More than a century of industrial activities in the

www.elsevier.com/locate/ecoenv

ARTICLE IN PRESSG.G. Pyle et al. / Ecotoxicology and Environmental Safety 61 (2005) 287–312288

region has led to widespread acidification and metalcontamination in more than 7000 lakes, causing severeecological damage (Keller et al., 1992). Before environ-mental remediation efforts, several aquatic species wereextirpated from local water bodies (Beamish andHarvey, 1972; Yan and Welbourn, 1990). However,emission abatement and reclamation programs over thepast four decades have resulted in improved waterquality in area lakes and reestablishment of someaquatic communities (Keller et al., 1992, 1999).Although improvements have been observed, elevatedmetal concentrations in local lakes persist (Nriagu et al.,1998). Consequently, the region is thought to be in astate of ecological recovery, but not yet recovered(Keller et al., 1999).Previous studies have identified a ‘‘zone of impact’’

from industrial activities, covering some 17,000 km2

around the epicenter of mining and processing activities(Keller et al., 1992). Metal concentrations in lake waterand sediments within this zone are elevated abovebackground concentrations, and typically decrease withincreasing distance from the center of ore processingoperations (Nriagu et al., 1982; Bradley and Morris,1986). Therefore, the water bodies in the Sudbury areapresent a unique research opportunity to exploit metalcontamination gradients to study relationships amongenvironmental metal contamination, metal accumula-tion in resident biota, and biological effects associatedwith metal contamination.Most studies that examine the effects of metals in

freshwater fish use standard laboratory species, such asrainbow trout (Oncorhynchus mykiss) and fatheadminnows (Pimephales promelas). However, these speciesrarely inhabit metal-contaminated lakes in northernCanada. Yellow perch (Perca flavescens), on the otherhand, is very tolerant of metals (Taylor et al., 2003), iswidely distributed throughout North America (Scottand Crossman, 1973), is abundant where it occurs, is acommon sport fish among anglers, occupies importanttrophic positions in aquatic ecosystems (i.e., plankti-vore, benthivore, and piscivore, depending on age andfood resource availability) (Sherwood et al., 2002b), andis known to inhabit lakes contaminated by industrialwastes. Therefore, it is important to understand howthis species responds to environmental contamination inorder that environmental risk assessments provide thehighest degree of ecological relevance. This species hasreceived relatively little ecotoxicological research atten-tion in the past, although this seems to be changing(Brodeur et al., 1997; Laflamme et al., 2000; Sherwoodet al., 2000, 2002a; Rajotte and Couture, 2002; East-wood and Couture, 2002; Audet and Couture, 2003;Couture and Kumar, 2003).Recent studies by us and by others examining the

effects of metals on wild yellow perch inhabitingindustrially contaminated lakes in Sudbury and Rouyn

Noranda, Quebec, have generally focused on a few(usually two to four) metal-contaminated lakes, usingonly one or two reference lakes, and have studied onlythe ‘‘metals of concern’’, which are typically Cd, Cu, Ni,and Zn in these areas (Brodeur et al., 1997; Laflamme etal., 2000; Rajotte and Couture, 2002; Eastwood andCouture, 2002; Audet and Couture, 2003; Couture andKumar, 2003). These studies have provided importantinsights into the biological consequences of metalcontamination to wild yellow perch inhabiting thesecontaminated environments. However, it is still not clearif the effects described by these studies are general innature and extend to a wider range of contaminationscenarios, or how other environmental variables (in-cluding other metals) interact with the metals of concernto contribute toward observed trends in fish condition asthey relate to the contamination gradient. Moreover,very little is known about how industrial metalcontamination contributes to fish recolonization ordiversity.The objective of this study was to exploit a metal

contamination gradient of 12 Sudbury-area lakesto investigate fish species diversity, and multivariaterelationships among environmental metal contamina-tion, water quality, metal accumulation, and conditionin similar-sized yellow perch. We measured 20 metals(hereafter, the word metals includes metalloids) inwater, sediment, and yellow perch liver, muscle,and intestine and related those concentrations to lakewater pH, hardness, alkalinity, conductivity, andtotal dissolved solids and to yellow perch conditionfactor, hepatosomatic index (HSI), and gonadosomaticindex (GSI). Several results from this analysis haveled to suggestions for future research into the mechan-isms that may account for some rather unexpectedfindings.

2. Materials and methods

2.1. Study lake choice

Twelve lakes in the Sudbury region (Fig. 1) wereselected for water, sediment, and fish sampling based ondata from previous studies (Nriagu et al., 1998; East-wood and Couture, 2002; Rajotte and Couture, 2002).Five of these lakes formed a downstream gradient froma point source of metal input (Kelly, Mud, Simon,McCharles, and Kusk lakes); three lakes had varyingcontamination levels and were independent of the KellyLake gradient (Hannah, Whitson, and Ramsey lakes).These lakes formed a metal contamination gradient. Thefour furthest lakes (Barlow, Big Marsh, Birch, andWaubamac lakes) served as reference lakes.

ARTICLE IN PRESS

Fig. 1. Map of 12 lakes in the vicinity of Sudbury, Ont., from which water, sediment, and fish samples were collected. See Table 1 for sampling dates.

Table 1

Sampling dates in 2001, locations, and relative distances from ore

processing operations for water, sediment, and yellow perch sampling

in 12 lakes in the vicinity of Sudbury, Ont.

Lake Location Distance

from

smeltera

(km)

Date sampled

Barlow Lake 4611704400N 43 May 7

8013404100W

Big Marsh Lake 4611706000N 39 May 7

8013801200W

Birch Lake 4611404700N 47 June 2

8013301100W

Hannah Lake 4612603500N 4 May 23 and June 7

8110202400W

Kelly Lake 4612605500N 2 May 2 and June 6

8110305700W

Kusk Lake 4611902100N 23 May 17 and July 12

8112001200W

McCharles Lake 4612205100N 14 June 21

8111405100W

Mud Lake 4612402000N 7 May 16 and July 13

8110904100W

Ramsey Lake 4612803900N 12 May 22 and July 17

8015604500W

Simon Lake 4612304700N 10 June 12 and July 12

8111104800W

Waubamac Lake 4611603400N 50 May 8

8012904500W

Whitson Lake 4613502000N 18 May 24

8015802700W

aDistance calculated from Inco’s Copper Cliff smelter (4612604300N,

8010503900W) to each lake using latitude/longitude coordinate data.

G.G. Pyle et al. / Ecotoxicology and Environmental Safety 61 (2005) 287–312 289

2.2. Fish sampling

Minnow traps were baited with Styrofoam beads andplaced in littoral zones in study lakes. Traps werechecked daily. A seine net was also used to catch fish inthe littoral zones of every lake. All fish captured wereidentified to species and released. When identificationwas not possible in the field, fish were placed on ice andidentified in the laboratory. Fish sampling dates arereported in Table 1.Approximately 20 yellow perch were sampled from

each lake when possible and returned to the laboratoryeither on ice or in 23-L Rubbermaid containers with aportable aerator unit. On arrival, fish were dissected orelse frozen at –20 1C until dissection. Fork length, totalweight, and the weights of liver and gonads were recordedfor each fish. Muscle, liver, and intestine flushed withsaline were sampled and stored at –20 1C for metalanalysis. Scales were collected from the right side of eachfish, on the dorsolateral surface, immediately anterior tothe insertion of the dorsal fin, and aged by M. Gauthier(Labman Aging, Cochrane, Ont., Canada).Fish condition was assessed by calculating Fulton’s

condition factor (FCF), HSI, and GSI. FCF providesinformation about recent feeding activity, and iscalculated as

FCF ¼W t

L3� 100;

where Wt is the total weight of the fish (g), and L is thefork length (cm). The HSI, another metric commonly


used to evaluate fish condition, provides informationabout metabolic activity in the liver, and is calculated as

HSI ¼WL

W t� 100;

where WL is the weight of the liver (g). The GSIevaluates reproductive condition based on the weight ofthe gonads (WG) relative to the total weight of the fish,and is calculated as

GSI ¼WG

W t� 100:

2.3. Water and sediment collection

Water samples were taken from each study lake at 3depths: 1, 5, and 10m. When the 10m depth was notavailable, an alternative depth (3m) was sampled. Watersamples were collected with a Van Dorn bottle.Sampling containers were rinsed with lake water threetimes, filled to capacity (i.e., no head space), and placedon ice for transport back to the laboratory.Sediment samples were collected from the same

locations as water samples using an Eckman grab. Grabsampling collects approximately the top 5 cm of sedi-ment. Sediments were unloaded from the grab into aplastic tub and mixed. One sample bottle was then filledwith the mixed sediment and returned to the laboratoryon ice.Both water and sediment samples were refrigerated

for up to 12 h on return to the laboratory. Watersamples were acidified to a pH of 2 by addition of tracemetal-grade nitric acid for metal determination, thenstored at 4 1C until analysis (see below).

2.4. Water quality

Water samples (n ¼ 3) were collected from each lakeand transported back to the laboratory at TestmarkLaboratories, Ltd (Sudbury, Ont., Canada) for analysis.Conductivity of each sample was measured using anOrion conductivity meter (Model 135A with Orionconductivity cell 013610). Turbidity (in nephrelometryturbidity units, NTU) was measured with a Hach ratioturbidimeter (Model 18900). Sample turbidity wasrecorded after a 1min equilibration period. Water pHwas measured with an Accumet pH meter (Model 915,Fisher Scientific). All samples (40mL) were measured inclean beakers while stirred. Alkalinity was determinedby titration with 0.02N hydrochloric acid (HCl) to afinal pH of 4.5. Alkalinity was calculated as theconcentration of calcium carbonate present using theformula

Alk ¼V1 � N � 50; 000

V 2;

where Alk=alkalinity (mg/L as CaCO3), V1=volume ofHCl (mL), V2=volume of sample (40mL), andN=normality of HCl (0.02N). Water hardness wascalculated using the formula:

hardness ðmg=LÞ ¼ ð½Ca n 2:497Þ þ ð½Mg n 4:118Þ

where [Ca] and [Mg] are the aqueous concentrations ofCa and Mg (in mg/L) (Clesceri et al., 1998).Total suspended and total dissolved solids (TSS and

TDS, respectively) were determined by vacuum filteringa 100mL water sample through a prebaked and weighedWhatman 934-AH glass microfiber filter. The filterpaper was then placed on a prebaked and weighed dishand baked in an oven at 80 1C for 24 h. The differencebetween the initial and final weights of the filter paperyielded TSS. TDS were measured by adding the filteredwater sample into a prebaked and weighed beaker,evaporating at 80 1C for 24 h, cooling and reweighing todetermine TDS.

2.5. Sediment preparation for ICP-MS

Twenty grams of sediment from each sediment samplewas dried in an oven set at 100 1C for 24 h. After drying,the sample was cooled, reweighed, and then crushedwith a mortar and pestle. Approximately 2 g of drysample was digested in a 3:1 mixture of concentratedhydrochloric and nitric acids and boiled at 100 1C for30min. The solution was cooled and then filteredthrough a microfiber paper (Whatman grade 1) undervacuum into a volumetric flask. All equipment wasrinsed with 2% nitric acid into the flask. The solutionwas brought up to 100mL total volume with 2% nitricacid.

2.6. Fish tissue preparation for ICP-MS

Samples of approximately 10–200mg wet wt of yellowperch liver, white muscle, and intestine were dried at60 1C to complete dryness. Dried samples (ranging from1.3 to 20mg) were digested in precleaned, acid-washedTeflon vials in 2mL of analytical-grade nitric acid. Vialswere heated in a 750-W microwave oven at full powerfor 30-, 20-, 15-, then 10-s periods, with 30-s intervalsbetween heating events for sample shaking and mixing.Any tissues not fully digested were subjected to another20 s of microwave heating. Samples were allowed to coolovernight, then transferred to labeled, precleaned, acid-washed, 50-mL Fisherbrand specimen containers with18mL of reverse osmosis water. The diluted sampleswere transported to Testmark Laboratories, Ltd.(Sudbury, Ont., Canada) for inductively coupled plasmamass spectrometry (ICP-MS).

ARTICLE IN PRESS

Table 2

Results of principal component analysis of water metal data from the

12 study lakes (Table 1)

PC1 PC2 PC3

Eigenvalue (L) 2.036 0.924 0.218

Variance explained (%) 61.5 27.9 6.6

Cumulative explained

variance (%)

61.5 89.4 96.0

Eigenvectors (l)Al �0.437 0.797 0.330

As 0.176 0.071 �0.148

Ba 0.144 0.057 0.142

Ca 0.470 0.278 �0.137

Cu 0.219 �0.006 0.474

Fe 0.026 0.345 �0.313

Mg 0.290 0.145 �0.068

Mn 0.068 0.229 �0.524

Ni 0.453 0.010 0.296

Rb 0.215 0.175 0.075

Se 0.209 0.127 0.001

Sr 0.310 0.195 �0.030

Zn 0.087 0.022 0.365

Note. Eigenvalues, variance, cumulative variance explained, and

eigenvectors are reported for the first three principal components

(PC1–PC3). Components retained for further analysis were determined

by a scree analysis. The highest and lowest eigenvector loadings for

each component are highlighted in boldface.


2.7. Sample metal determinations via inductively coupled

plasma analysis

Water samples were prepared without dilution.Sediment samples were diluted one-tenth with double-distilled water. An internal surrogate with knownconcentrations of rubidium and ruthenium was addedto each sample (0.5mL). At the start of each run andafter every 10 samples, quality control samples were run.Quality assurance samples included blanks; 10, 100, and1000 mg/L standards of each metal measured, andsamples of known metal concentrations from theCanadian Association of Environmental AnalyticalLaboratories (CAEAL). Further dilutions of sampleswere made when necessary. All samples passed theCAEAL-accredited quality control standards of Test-mark Laboratories, Ltd. Fish tissues were prepared forICP-MS analysis by combining 10mL of diluted samplewith 0.5mL of internal ICP-MS surrogate in labeled,sterile, plastic ICP-MS tubes. Tubes were tightly cappedand stored until they were analyzed. National ResearchCouncil (NRC) standards TORT-2, DORM-2, andDOLT-2 as well as method blanks were preparedsimilarly to the fish tissues, and metal concentrationswere determined concurrently by ICP-MS.

2.8. Statistical treatment

Before statistical analysis, variables were tested fornormality using the Shapiro–Wilk test and homogeneityof variance using Levine’s test. Data not meetingparametric assumptions were log(x) or log(x+1) trans-formed, which usually improved distributions. Meancomparisons among variables were conducted usinganalysis of variance (ANOVA) followed by a Tukey-Kramer multiple comparison test. Statistical significancewas assumed when Po0:05: Untransformed data wereused to generate graphs for ease of interpretation, evenin circumstances where statistical analyses were per-formed on log-transformed data.A principal component analysis (PCA) was conducted

on water and sediment metal data. In the originalanalysis, 27 metals were measured in water samples and54 metals were measured in sediment samples. Each dataset was censored using a priori rules, similar to thosereported in Pyle et al. (2001) to remove variables thatwere rarely detected in water samples, and to minimizemulticolinearity among remaining variables. Metals thatwere not detected in a sample were not estimated, butwere removed from subsequent analysis. Metals thatwere not detected in 450% of samples were removed.Metals that were strongly correlated with other metalsin the data set were removed. However, metals that areknown to be associated with mining activities in theSudbury area, such as Cd, Cu, Ni, and Zn, were retained(if detected in 450% of samples), even if they

demonstrated a high degree of colinearity with othermetals in the data set. By these criteria, B, Cd, Ce, Co,Cr, Ga, Hg, La, Li, Mo, P, Pb, Ti, and V were removedfrom the water metal PCA data set, and B, Ba, Be, Bi,Ce, Cs, Dy, Er, Eu, Ga, Gd, Hf, Hg, Ho, La, Li, Lu,Mg, Mo, Nb, Nd, P, Pr, Rb, Sb, Sc, Sm, Sn, Sr, Ta, Tb,Th, Ti, Tl, Tm, U, W, Y, Yb, and Zr were removed fromthe sediment metal data set. Only the metals remainingafter censoring were used in all subsequent analyses.Metals retained in the water metal PCA were: Al, As,Ba, Ca, Cu, Fe, Mg, Mn, Ni, Rb, Se, Sr, and Zn. Metalsretained in the sediment metal PCA were: Ag, As, Ca,Cd, Co, Cr, Cu, Fe, Mn, Ni, Pb, Se, V, and Zn.All regressions and multiple regressions reported in

this study used least-squares techniques. Pearson corre-lations were used to detect significant associationsbetween pairs of variables. A partial correlation analysiswas used to examine bivariate associations of five metalsbetween compartment pairs in a five-compartmentmodel. All statistical analyses were conducted on JMPversion 5.0 statistical software.

3. Results

3.1. Water quality

As PCA was used to characterize water quality ineach lake in terms of metal concentration (Table 2).A scree analysis was conducted to identify those

ARTICLE IN PRESS

Fig. 2. Ordination plot from principal components analysis (PCA) of

water metal concentration data in the plane of PC1 and PC2. Each

point represents an optimal linear combination of 13 metals measured

in water samples (n ¼ 3) from each of 12 study lakes. Lake names are

provided for each point. This ordination identifies three lake clusters:

Group 1, Barlow, Big Marsh, Birch, and Waubamac lakes; Group 2,

Kelly, Mud, and Simon lakes; Group 3, Hannah, Kusk, McCharles,

Ramsey, and Whitson lakes. Correlations between fish condition,

water quality variables, and distance from smelting operations against

each PC are represented by vectors. Vector lengths are proportional to

correlation coefficients (r), and vector directions indicate the direction

of maximum correlation in the ordination plane. FCF, Fulton’s

condition factor; HSI, hepatosomatic index; GSI, gonadosomatic

index, *Age-corrected.

G.G. Pyle et al. / Ecotoxicology and Environmental Safety 61 (2005) 287–312292

components containing the most information, and onlythe first three components contained significant infor-mation associated with metal concentrations in watersamples. Together, these three components accountedfor 96% of the variability associated with metalconcentrations in all lakes (Table 2). The PCA ordina-tion in the plane of PC1 and PC2 yielded three distinctlake clusters (Fig. 2). Group 1 consisted of Barlow, BigMarsh, Birch, and Waubamac lakes (i.e., referencelakes). Group 2 consisted of Kelly, Mud, and Simonlakes (i.e., near-field lakes on the Kelly Lake watershed).Group 3 consisted of Hannah, Kusk, McCharles,Ramsey, and Whitson lakes, which represents two far-field lakes on the Kelly Lake chain (i.e., Kusk andMcCharles lakes), and other Sudbury-area metal-con-taminated lakes (i.e., Hannah, Ramsey, and Whitsonlakes).Aluminum had the highest negative eigenvector (l)

loading on PC1, whereas both Ca and Ni had thehighest positive loading (Table 2). Aluminum loadedstrongly on the positive side of PC2. Although Culoaded negatively on PC2, a l of �0.006 suggests that ithad little influence on the component. Manganese hadthe strongest negative loading, and Cu had the highestpositive loading on PC3.Of the limnological variables measured, water pH,

alkalinity, conductivity, and hardness were all stronglyand positively associated with PC1 (Table 3, Fig. 2).Only water hardness was related to PC2. No limnolo-gical variables were correlated with PC3. Aluminum wasthe only metal negatively correlated with PC1 (Table 3).However, As, Ba, Ca, Cu, Mg, Ni, Rb, Se, and Sr werepositively correlated with PC1, with Ni showing thestrongest correlation (r ¼ 0:97). No metals were nega-tively correlated with PC2. Aluminum, Fe, and Mn allshowed significant positive relationships with PC2.Manganese was negatively correlated, and Zn positivelycorrelated, with PC3.Median water pH was lowest in Group 1 lakes,

ranging from 6.0 to 6.8, compared with median pH inGroup 2 and 3 lakes, which ranged from 7.0 to 8.4(Table 4). Group 1 lakes also demonstrated the greatestdegree of within-lake pH variability relative to con-taminated lakes.Conductivity and TDS varied over two orders of

magnitude among the study lakes (Table 4). Waterconductivity in Group 1 lakes was about 30 mS/cm,representing the lowest conductivity among the lakegroups. Group 3 lake conductivity was higher than thatof Group 1 lakes, and ranged from 166.2 to 427.3 mS/cm.Group 2 lakes had the highest conductivity, rangingfrom 1098.7 to 1455.0 mS/cm. Water TDS followed thesame basic trend as conductivity, with Group 1 lakeshaving the lowest TDS concentrations, followed byGroup 3 lakes; Group 2 lakes had the highest TDSconcentrations.

Both alkalinity and hardness were lowest in Group 1lakes relative to the other two lake groups (Table 4). InGroup 1 lakes, alkalinity ranged from 6.3 to 8.8mg/L asCaCO3, and hardness ranged from 8.7 to 18.0mg/L asCaCO3. Group 2 lakes demonstrated the highestalkalinity, ranging from 32.1 to 33.7mg/L as CaCO3,while Group 3 lakes were intermediate, ranging from 6.3to 31.3mg/L as CaCO3. Similarly, Group 2 lakesdemonstrated the highest hardness levels(380.3–582.7mg/L as CaCO3), while Group 3 lakeswere intermediate (25.2–134.3mg/L as CaCO3).Although three of the four lakes constituting Group

1, Waubamac, Big Marsh, and Barlow lakes, hadrelatively clear water (turbidity range, 1.6–2.7NTU), afourth Group 1 lake, Birch Lake, had some of the mostturbid water (10.1NTU) of all the study lakes (Table 4).The only lake with higher turbidity was a Group 2 lake,Mud Lake (10.4NTU). Turbidity in the other con-taminated lakes ranged from 0.4 to 3.8NTU, with threeGroup 3 lakes, Hannah, Whitson, and Ramsey lakes,having an order of magnitude lower turbidity thanGroup 1 lakes (0.4, 0.4, and 0.6NTU, respectively).Except for one Group 2 lake, Mud Lake (TDS,

26.7mg/L), two of the Group 1 lakes, Birch and Barlowlakes, had the highest concentrations of TDS (21 and22mg/L, respectively) of any of the other study lakes. InWaubamac Lake, another Group 1 lake, TDS were notdetectable. The remaining contaminated lakes ranged in

ARTICLE IN PRESS

Table 3

Pearson correlation coefficients between principal components characterizing water and sediment samples, distance from smelting operations, water

quality, and metal variables

Water Sediment

PC1 PC2 PC3 PC1 PC2 PC3

pH 0.86c n.s. n.s. — — —

Alkalinity (mg/L as CaCO3) 0.85c n.s. n.s. — — —

Conductivity (mS/cm) 0.84c n.s. n.s. — — —

Hardness (mg/L as CaCO3) 0.74b 0.65a n.s. — — —

Distance from smelter (km) �0.93d n.s. n.s. �0.79c n.s. n.s.

Ag — — — 0.84c n.s. n.s.

Al �0.62a 0.77b n.s. — — —

As 0.88c n.s. n.s. 0.79b n.s. n.s.

Ba 0.90c n.s. n.s. — — —

Ca 0.92c n.s. n.s. n.s. 0.75b n.s.

Cd — — — 0.78b n.s. n.s.

Co — — — 0.75b n.s. n.s.

Cr — — — n.s. 0.91d n.s.

Cu 0.78b n.s. n.s. 0.90c n.s. n.s.

Fe n.s. 0.86c n.s. n.s. 0.77b n.s.

Mg 0.93d n.s. n.s. — — —

Mn n.s. 0.60a �0.67a n.s. 0.80b n.s.

Ni 0.97d n.s. n.s. 0.97d n.s. n.s.

Pb — — — 0.72b n.s. n.s.

Rb 0.85c n.s. n.s. — — —

Se 0.82b n.s. n.s. n.s. n.s. 0.77b

Sr 0.91d n.s. n.s. — — —

V — — — n.s. 0.85c n.s.

Zn n.s. n.s. 0.70a n.s. 0.78b n.s.

n.s., not significant (P40:05).aPo0:05:bPo0:01:cPo0:001:dPo0:0001:


TDS from 1.0mg/L in Whitson Lake to 9.0mg/L inHannah Lake.Aluminum concentrations were highest in waters

from Group 1 lakes (Table 5). With the exception ofKusk (Group 3) and Mud (Group 2) lakes, Al was notdetected in metal contaminated lakes. Arsenic concen-trations were highest in Group 2 lakes, and were belowdetection limits in most Group 1 lakes. Copperconcentrations were lowest in Group 1 lakes, and wereless than 4.5 mg/L. Although the highest water Cuconcentrations were in two Group 3 lakes, Hannah(25 mg/L) and Whitson (19 mg/L) lakes, Group 2 lakesdemonstrated generally higher Cu concentrations (ran-ging from 9.7 to 17.0 mg/L). Nickel concentrationsvaried over three orders of magnitude among the studylakes. Group 1 Ni concentrations were the lowest amongstudy lakes, followed by Group 3 lakes; Group 2 lakeshad the highest Ni concentrations. Ramsey (52 mg/L)and Kusk (60 mg/L) lakes were the only metal-contami-nated lakes that had Ni concentrations below 100 mg/L.Selenium was not detected in any Group 1 lakes exceptWaubamac Lake (Se, 0.6 mg/L), but was detected in allof the metal-contaminated lakes except Whitson Lake.

Selenium concentrations in metal-contaminated lakesranged from 1.0 mg/L in McCharles Lake to 5.8 mg/L inKelly Lake. All Group 2 lakes had Se concentrationsgreater than 4 mg/L.

3.2. Sediment quality

In general, sediment metal concentrations in lakesconstituting Group 2 and 3 were higher than in Group 1lakes (Table 6). Sediment Cu and Ni exhibited thegreatest variability between Group 1 and contaminatedlakes. Copper concentrations in Group 1 lake sedimentsranged from 58.4 to 174.7 mg/g, whereas Cu concentra-tions in contaminated lake sediments ranged from 349.4to 2672 mg/g. Hannah, Kelly, McCharles, Ramsey,Simon, and Whitson lakes had sediment Cu concentra-tions greater than 1200 mg/g. In Group 1 lakes, sedimentNi concentrations ranged from 95.6 to 210.8 mg/g.Sediment Ni concentrations were at least four timeshigher in contaminated lakes, ranging from 864.5 to4745 mg/g. Sediment Ni concentrations exceeded1000 mg/g in all contaminated lakes except Kusk(865 mg/g). On average, sediment Co concentrations

ARTICLE IN PRESS

Table 4

Water quality data for the 12 study lakes

Lake pH Conductivity

(mS/cm)TDSa

(mg/L)

Alkalinity

(mg/L as

CaCO3)

Hardness (mg/

L as CaCO3)

Turbidity

(NTU)

TSS

(mg/L)

Group 1

Barlow Mean 6.5 33.2 94.7 7.1 12.7 2.7 22.0

SEM (6.0–6.5) 1.3 22.7 0.8 0.4 1.8 21.0

Big Marsh Mean 6.0 29.2 36.0 6.3 10.2 1.6 3.3

SEM (5.7–6.3) 0.7 9.8 1.4 0.9 0.6 0.3

Birch Mean 6.1 27.6 60.7 6.3 8.7 10.1 21.0

SEM (6.0–6.3) 2.0 12.8 0.0 0.4 5.6 14.5

Waubamac Mean 6.8 33.0 61.0 8.8 18.0 1.6 0.0

SEM (6.7–7.0) 0.1 10.0 0.0 1.1 0.2 0.0

Group 2

Kelly Mean 8.4 1455.0 1085.0 33.7 582.7 3.0 9.0

SEM (8.4–8.4) 2.6 14.0 1.4 14.3 0.1 1.0

Mud Mean 7.5 1266.0 966.7 32.1 523.5 10.4 26.7

SEM (7.3–7.6) 4.9 13.2 0.8 18.8 4.3 13.7

Simon Mean 7.4 1098.7 802.7 32.1 380.3 1.1 6.0

SEM (7.4–7.5) 4.3 10.7 0.8 12.8 0.1 3.0

Group 3

Hannah Mean 7.6 427.3 211.3 16.3 46.0 0.4 4.0

SEM (7.6–7.6) 0.7 10.8 0.0 4.4 0.01 1.7

Kusk Mean 7.4 249.3 155.0 23.8 69.8 3.8 2.7

SEM (7.4–7.5) 2.8 9.1 0.0 1.7 0.4 2.7

McCharles Mean 7.5 424.7 286.0 27.1 134.3 1.3 4.0

SEM (7.5–7.5) 135.7 104.3 0.8 31.9 0.2 1.0

Ramsey Mean 8.0 378.3 202.3 31.3 52.4 0.6 6.0

SEM (7.9–8.0) 0.3 1.8 0.0 3.6 0.04 1.2

Whitson Mean 7.0 166.2 88.7 6.3 25.2 0.4 1.0

SEM (7.0–7.0) 0.2 11.9 0.0 3.8 0.01 1.0

Note. Values represent means and SEM, except for pH, where the median and range are reported. Water samples (n ¼ 3) were taken in each lake at

the same locations where sediments were collected. Highest and lowest values for each variable are indicated in boldface.aTDS, total dissolved solids; NTU, nephrelometry turbidity units; TSS, total suspended solids.


varied by approximately fourfold between reference andcontaminated lakes. In Group 1 lakes, sediment Coranged from 13.8 to 49.5 mg/g, whereas in contaminatedlakes, sediment Co ranged from 47 to 205 mg/g. Othersediment metal concentrations varied from one- tothree-fold between reference and contaminated lakeson average, and are given in Table 6.A PCA on sediment metal concentration yielded three

principal components (from a scree analysis) thataccounted for 88.3% of the variability associated withsediment metal concentrations in the 12 study lakes(Table 7). Copper and Ni had the highest positive lloadings on PC1, whereas Mn was the only metal thatloaded negatively, albeit weakly, on PC1. Calcium andCr had the strongest positive l loading on PC2, while Culoaded negatively. On PC3, Se had the strongest positiveloading, while Co and Fe loaded strongly and nega-tively.Silver, As, Cd, Co, Cu, Ni, and Pb were significantly

and positively correlated with sediment PC1 (Table 3).

Nickel and Cu were most strongly correlated withPC1 (r ¼ 0:97 and 0.90, respectively). Calcium, Cr,Fe, Mn, V, and Zn were significantly and positivelyrelated to PC2, with Cr having the strongest relation-ship (r ¼ 0:91). Selenium was the only metal signifi-cantly correlated with PC3 (r ¼ 0:77). No significantnegative relationships between metals and PCs wereobserved.The PCA ordination for sediment metals is illustrated

in Fig. 3. Group 1 lakes plotted separately from thecontaminated lakes, and were restricted to the negativeside of PC1. Lakes from the Kelly Lake chaindemonstrated an interesting ordination pattern thatclosely followed their downstream position in thewatershed (i.e., Kelly, Simon, McCharles, Mud, andKusk). Hannah and Whitson lakes plotted close toMcCharles and Simon lakes. Ramsey Lake seemed to bedistinct from all the other study lakes with respect tosediment metal concentration given its unique positionon the ordination plot.

ARTICLE IN PRESS

Table 5

Total metal concentration (mg/L) in water samples (n ¼ 3) taken from 12 study lakes

Lake Al As Ba Ca Cu Fe Mg Mn Ni Rb Se Sr Zn

MDL 50 1 1 100 1 50 100 1 1 1 1 1 1

Group 1a Barlow Mean 238.7 ND 12.7 2899 4.3 389 1314 521 9.7 2.0 ND 20.0 10.3

SEM 130.2 — 1.7 85 0.9 194 70 342 0.7 0.0 — 0.6 2.4

Big Marsh Mean 128.7 0.7 10.3 2183 3.7 421 1155 598 9.0 1.7 ND 16.3 7.3

SEM 25.5 0.2 0.7 160 1.2 96 131 340 1.0 0.3 — 1.5 1.7

Birch Mean 172.7 ND 14.3 2129 2.0 714 830 1726 7.0 2.0 ND 18.3 7.0

SEM 65.7 — 2.4 95 0.6 366 30 803 1.0 0.0 — 1.3 1.0

Waubamac Mean 80.0 ND 10.0 4203 2.0 412 1811 344 8.0 1.7 0.6 28.3 5.7

SEM 5.2 — 1.2 304 0.6 73 113 190 1.0 0.3 0.2 1.3 1.8

Group 2 Kelly Mean ND 2.4 38.4 204437 15.0 477 17545 207 338.2 19.2 5.8 365.0 11.7

SEM — 0.3 3.8 5175 1.5 35 339 22 34.7 1.7 0.9 32.4 3.3

Mud Mean 70.7 3.3 33.3 186097 17.0 653 14284 513 257.3 15.3 4.7 291.0 7.3

SEM 33.9 0.3 0.9 6537 7.5 36 632 4 7.2 0.7 0.3 8.7 0.9

Simon Mean ND 2.3 33.3 127812 9.7 422 14857 522 318.7 15.7 5.0 271.3 9.0

SEM — 0.3 0.3 4702 0.3 4 273 9 2.3 0.3 0.0 2.8 2.0

Group 3 Hannah Mean ND 2.0 23.7 11603 25.0 71 4137 251 181.0 3.0 2.3 64.7 10.0

SEM — 0.0 2.4 1150 2.0 4 377 19 44.5 0.0 0.3 6.4 3.2

Kusk Mean 46.0 2.0 14.3 22071 6.0 261 3560 328 60.0 3.0 1.2 56.7 6.0

SEM 10.5 0.0 0.3 659 0.6 9 19 13 1.2 0.0 0.4 0.9 1.5

McCharles Mean ND 1.7 19.7 43274 6.0 206 6363 332 111.3 5.3 1.0 102.0 4.7

SEM — 0.3 1.2 10589 1.0 6 1321 13 39.3 1.3 0.5 23.0 0.3

Ramsey Mean ND 1.3 14.0 14003 9.7 36 4224 72 52.0 2.0 2.0 48.7 5.7

SEM — 0.3 1.0 1030 0.3 11 240 4 2.6 0.0 0.0 2.2 0.7

Whitson Mean ND 0.7 21.3 6600 19.0 53 2111 21 155.0 2.3 ND 36.3 17.7

SEM — 0.2 3.2 981 3.5 14 317 4 24.4 0.7 — 5.2 2.8

Note. Highest and lowest metal concentrations are represented in boldface. ND, not detected; MDL, minimum detection limit.aLake groups are based on multivariate similarity as determined by principal component analysis. Group 1 represents reference lakes. See text for

details.


3.3. Fish diversity and yellow perch samples

Fish diversity was variable among study lakes(Table 8). Among all lakes, 20 species were identified.Six of these species, including central mudminnow,fathead minnow, johnny darter, largemouth bass, logperch, and muskellunge, occurred in one lake only. Allother species occurred in at least two lakes. Yellowperch was the only species that occurred in all 12 studylakes, followed by the northern pike and pumpkinseed,which each occurred in 67% (8/12) of the lakes studied.Yellow perch was the only species caught in Birch Lake.A stepwise multiple regression analysis was conducted

using the first three extracted principal components ofthe water and sediment metal PCAs as independentvariables to account for variability associated with fishdiversity. Analyses using Shannon’s diversity index (H0)as the dependent variable, and/or including pH, hard-ness, alkalinity, conductivity, and distance from smelt-ing operations, resulted in no significant relationships(P40:05). A subsequent analysis using log-transformedwater and sediment metals, in addition to pH, hardness,alkalinity, conductivity, and distance from smeltingoperations, yielded a significant relationship that couldaccount for 92.4% of the variability associated with fish

diversity (P ¼ 0:0005). This model took the form

H 0 ¼ � 0:79ð�0:44Þ þ 0:68ð�0:17Þ

� logðsediment½MnÞ þ 0:51ð�0:10Þ

� logðsediment½CoÞ þ 0:41ð�0:09Þ

� logðsediment½CuÞ � 1:37ð�0:16Þ

� logðsediment½AsÞ; r2 ¼ 0:92; P ¼ 0:0005;

where standard errors of estimates are given inparentheses, and square brackets represent concentra-tions (mg/g). Although water metal concentrations,water quality variables, and distance from smelter wereincluded in the analysis, only sediment metal datacontributed significantly to the model.Because yellow perch was the only species found in

each of the 12 lakes studied, 19 or 20 yellow perch ofapproximately the same size were sampled from each ofthe study lakes (Table 9). Although we tried to samplefish of the same size in all study lakes, there was somesize variability among lakes. Except for McCharles(86mm) and Birch (242mm), which had the smallestand largest mean yellow perch fork lengths, respectively,mean yellow perch fork length ranged from 92mm inMud Lake to 125mm in Barlow Lake. Median age

ARTICLE IN PRESS

Table 6

Total sediment metal concentrations (mg/g dry wt, n ¼ 3) from 12 Sudbury-area study lakes

Lake Ag As Ca Cd Co Cr Cu Fe Mn Ni Pb Se V Zn

MDL 1 1 100 0.1 0.1 1 1 50 1 1 1 1 1 1

Group 1a

Barlow Mean ND 13.8 3644 0.9 22.4 44.0 65.3 22728 460 122.1 23.3 24.9 40.7 132.0

SEM — 6.5 1145 0.4 8.3 10.5 27.7 4482 181 39.2 9.8 6.6 9.0 41.3

Big Marsh Mean ND 12.0 3593 0.7 49.5 52.2 58.4 37691 469 95.6 18.1 22.8 58.7 137.9

SEM — 3.2 751 0.4 27.7 9.2 27.1 14278 31 34.2 11.9 1.0 12.4 33.3

Birch Mean 0.9 33.8 5881 1.7 13.8 32.4 174.7 11024 153 210.8 67.9 70.5 47.6 158.4

SEM 0.2 4.4 978 0.4 2.3 2.6 65.0 2648 27 60.6 28.2 4.8 2.2 29.2

Waubamac Mean ND 20.6 5098 2.4 28.5 53.3 118.1 23715 528 159.2 56.2 40.5 52.5 197.1

SEM — 2.0 305 0.3 0.9 0.4 17.4 2754 44 21.9 13.8 3.3 3.1 24.7

Group 2

Kelly Mean 7.0 46.8 6665 4.6 196.8 49.7 2672.0 37032 230 4677.5 75.8 95.7 30.5 393.9

SEM 3.5 17.4 2772 1.6 72.1 19.0 1058.6 14488 89 1786.7 33.3 36.8 9.0 150.4

Mud Mean 1.5 18.1 7142 1.6 58.3 49.9 746.5 26019 206 1194.9 23.8 51.4 39.1 144.2

SEM 0.5 1.1 360 0.3 9.3 2.8 188.2 644 16 187.1 5.9 4.8 2.4 13.1

Simon Mean 2.5 51.7 7047 4.5 205.0 52.5 1658.8 34855 290 4744.8 68.2 126.6 44.6 302.0

SEM 1.0 16.3 1520 1.6 69.8 14.9 584.8 10181 47 1933.6 26.8 33.3 6.9 102.7

Group 3

Hannah Mean 3.5 87.7 2906 2.7 47.0 64.8 1450.1 34151 264 1460.7 67.1 31.8 48.4 148.0

SEM 0.6 7.6 781 0.8 7.7 5.8 283.5 1913 10 232.5 6.0 3.4 2.0 31.1

Kusk Mean 1.6 36.1 4403 1.5 80.4 46.3 349.4 29562 380 864.5 124.3 40.2 45.8 289.0

SEM 0.6 9.4 340 0.2 20.9 6.3 101.3 6352 57 234.7 42.6 2.3 6.4 77.3

McCharles Mean 1.9 79.3 5528 3.4 187.4 51.6 1287.8 33420 328 3654.3 75.7 60.7 44.9 289.4

SEM 0.9 41.6 1125 2.0 85.5 13.9 671.7 10208 74 2283.6 24.2 19.6 8.0 98.2

Ramsey Mean 1.9 67.9 2672 2.8 58.8 36.3 1606.2 26187 292 1889.6 67.8 33.4 33.2 180.5

SEM 0.7 33.3 971 1.4 29.7 14.6 495.0 11777 96 1053.2 31.6 3.3 9.1 82.7

Whitson Mean 1.8 77.5 3387 3.0 66.9 36.2 1915.5 28408 307 1302.6 204.7 38.0 42.3 176.7

SEM 1.0 34.9 495 1.2 29.9 6.5 729.5 10086 22 593.6 153.5 3.4 7.7 46.5

Note. Minimum and maximum metal concentrations are represented by boldface. ND, not detected; MDL, minimum detection limit; SEM, standard

error of the mean.aLake groups are the same as Table 4; see text for details.


among study lakes was 1–3 years, except in Birch Lake(4.5 years), where at least one 11-year-old fish wascaught. A large number of immature fish were caught inMcCharles Lake, accounting for the low fork length andweight of yellow perch from this lake.

3.4. Tissue metal concentration

3.4.1. Liver

Liver metal concentrations were generally higher infish from metal-contaminated lakes relative to referencelakes, but not always (Table 10). Pooling the data intothe three groups generated by PCA (see above), liver Cdconcentrations were lowest in fish from Group 1 relativeto those of Group 2 or 3 (F ð2;209Þ ¼ 83:5; Po0:0001)(Fig. 4). Liver Cd concentrations in Group 3 fish wereapproximately eight times higher than those in fish fromGroup 1 or 2. Liver Cd showed a very strong nonlinearrelationship with water hardness among fish inhabitinglakes in Group 2 and 3 (Fig. 5). This relationshipaccounted for 99% of the variability associated with

liver Cd in fish inhabiting contaminated lakes. Fish fromreference lakes did not follow the relationship.Among Group 1 lakes, liver Cu concentrations

ranged between 8.6 and 32.8 mg/g (Table 10). Inmetal-contaminated lakes, liver Cu concentrationsranged between 6.9 and 87.8 mg/g. Interestingly,liver Cu concentrations in McCharles Lake, which isdownstream of Simon Lake (where liver Cu concentra-tions of 6.9 mg/g were lower than in fish from anyother lake in the study), were 50.5 mg/g. In general, fishfrom Group 3 lakes had twice as much liver Cu as fishfrom lakes in Group 1 or 2 (F ð2;183Þ ¼ 4:3; P ¼ 0:01)(Fig. 4).Liver Ni was not detected in fish from Big Marsh

Lake, and was approximately 4 mg/g in fish from Barlowand Birch lakes (Group 1 lakes; Table 10). However,liver Ni was high in fish from Waubamac Lake(20.4713.6 mg/g) relative to those from other Group 1lakes. Liver Ni concentrations in fish from Group 1 and2 lakes were not significantly different, and weresignificantly lower than those of fish from Group 3lakes (F ð2;148Þ ¼ 8:5; P ¼ 0:0001) (Fig. 4).

ARTICLE IN PRESS

Table 7

Results of principal component analysis of sediment metal data from

the 12 study lakes (Table 1)

PC1 PC2 PC3

Eigenvalue (L) 0.913 0.387 0.113

Variance explained (%) 57.1 24.2 7.1

Cumulative explained variance (%) 57.1 81.3 88.3

Eigenvectors (l)Ag 0.170 0.029 �0.124

As 0.261 0.131 0.330

Ca 0.033 0.359 0.345

Cd 0.146 0.114 0.095

Co 0.291 0.259 �0.421

Cr 0.025 0.360 �0.024

Cu 0.607 �0.411 0.222

Fe 0.066 0.357 �0.400

Mn �0.071 0.333 0.125

Ni 0.564 �0.045 �0.266

Pb 0.290 0.303 0.203

Se 0.058 0.062 0.453

V 0.002 0.239 0.166

Zn 0.111 0.285 �0.018

Note. Eigenvalues, variance explained, cumulative variance explained,

and eigenvectors are reported for the first three principal components.

Components retained for further analysis were determined by a scree

analysis. The highest and lowest eigenvector loadings for each

component are highlighted in boldface.

Fig. 3. Ordination plot from PCA of sediment metal concentration

data in the plane of PC1 and PC2. Labeling conventions correspond to

those of Fig. 2.


Liver Pb concentrations were below detection in amajority of samples from Group 1 lakes (Table 10).Liver Pb concentrations in fish from Group 2 and 3lakes fell within the reference range, except for fish fromHannah (41.3 mg/g) and Kelly (17.9 mg/g) lakes. Fishfrom lakes constituting Group 2 and 3 had liver Pbconcentrations that were not significantly different, butwere significantly higher than those from Group 1(F ð2;62Þ ¼ 12:5; Po0:0001) (Fig. 4).

Liver Se concentrations ranged from 4.8 to 90.4 mg/gamong all lakes (Table 10). Selenium concentrations inliver tissues from Kelly Lake fish were at least threetimes higher than liver Se from any other lake. Liver Seconcentrations in Group 2 fish were four times higherthan in Group 1 fish, and twice the levels measured inGroup 3 (F ð2;227Þ ¼ 213:6; Po0:0001) (Fig. 4).Zinc concentrations in liver tissue varied more than

15-fold, ranging from 93 to 1441 mg/g (Table 10). Fishfrom Group 1 lakes had liver Zn concentrations thatwere not significantly different from those of Group 2fish. However, Group 3 fish had significantly higher liverZn concentrations than either Group 1 or Group 2 fish(F ð2;236Þ ¼ 28:1; Po0:0001) (Fig. 4).

3.4.2. Muscle

In general, muscle metal concentrations were lowerthan liver metal concentrations (Table 11). Muscle Cuwas not detected in any Whitson Lake fish, and rangedbetween 2.9 and 14.1 mg/g among other study lakes.Muscle Cu concentrations in Group 2 fish were notsignificantly different from those in either Group 1 orGroup 3 fish. However, muscle Cu concentrations inGroup 1 fish were 47% higher than those in Group 3fish (F ð2;132Þ ¼ 5:2; P ¼ 0:007) (Fig. 6).Muscle Hg concentrations were highest among

fish from Group 1 lakes (Table 11). Among metal-contaminated lakes, muscle Hg concentrations rangedfrom 0.10 to 0.39. Group 1 fish had significantly highermuscle Hg concentrations than fish from Group 2 or 3lakes, which did not significantly differ from each other(F ð2;151Þ ¼ 65:1; Po0:0001) (Fig. 6).Muscle Se concentrations were lowest among fish

from Group 1 lakes (Table 11). Among the metal-contaminated lakes, fish from Simon, Hannah, andKelly lakes had muscle Se concentrations above 10 mg/g,whereas fish from other metal-contaminated lakes hadmuscle Se concentrations from 2.3 to 7.7 mg/g. Group 1lakes had fish with the lowest muscle Se concentrations.Mean muscle Se concentrations from Group 2 fish weremore than double the concentrations measured inGroup 3 fish, and about 14 times higher than in Group1 fish (F ð2;182Þ ¼ 60:5; Po0:0001) (Fig. 6).Muscle Hg and Se concentrations exhibited an

antagonistic effect (r ¼ �0:50; Po0:0001) andwere related to the distance from major smeltingoperations in the area (Fig. 7). Fish caught fromlakes close to metal ore smelting operations hadelevated muscle Se concentrations but low muscle Hgconcentrations relative to those caught from moredistant lakes. Both muscle Se and Hg con-centrations were significantly, but oppositely, relatedto distance from local smelting operations(muscle Se=1.02–0.02� distance from smelter; muscleHg=�0.92+0.14� distance from smelter).

ARTICLE IN PRESS

Table 8

Fish species diversity in 12 study lakes around Sudbury, Ont., sampled in 2001

Speciesa Barlowb Big Marshb Birchb Hannah Kelly Kusk McCharles Mud Ramsey Simon Waubamacb Whitson Total Proportion

Brown bullhead + + + + + + 6 0.50

Central mudminnow + 1 0.08

Common shiner + + + + + + 6 0.50

Creek chub + + + + + + 6 0.50

Fathead minnow + 1 0.08

Five-spine stickleback + + 2 0.17

Golden shiner + + + + + + 6 0.50

Iowa darter + + + 3 0.25

Johnny darter + 1 0.08

Lake chub + + + + 4 0.33

Largemouth bass + 1 0.08

Log perch + 1 0.08

Muskellunge + 1 0.08

Northern pike + + + + + + + + 8 0.67

Pumpkinseed + + + + + + + + 8 0.67

Rock bass + + + 3 0.25

Smallmouth bass + + + + + 5 0.42

Walleye + + + + + + 6 0.50

White sucker + + + + + 5 0.42

Yellow perch + + + + + + + + + + + + 12 1.00

Total no. of species 9 8 1 2 8 6 7 8 6 11 13 7

Diversityc 0.95 0.90 0 0.30 0.90 0.78 0.85 0.90 0.78 1.04 1.11 0.85

aBrown bullhead, Ictalurus nebulosus; central mudminnow, Umbra limi; common shiner, Notropus cornutus; creek chub, Semotilus atromaculatus;

fathead minnow, Pimephales promelas; five-spine stickleback, Culaea inconstans; golden shiner, Notemigonus crysoleucas; Iowa darter, Etheostoma

exile; johnny darter, Etheostoma nigrum; lake chub, Couesius plumbeus; largemouth bass, Micropterus salmoides; log perch, Percina caprodes;

muskellunge, Esox masquinongy; northern pike, Esox lucius; pumpkinseed, Lepomis gibbosus; rock bass, Ambloplites rupestris; smallmouth bass,

Micropterus dolomieui; walleye, Stizostedion vitreum; white sucker, Catostomus commersoni; yellow perch, Perca flavescens.bIndicates reference lakes; all other lakes are known to have been affected by regional industrial activity.cShannon’s diversity index.

Table 9

Yellow perch number, gender distribution, age, length, and weight of fish sampled in 12 study lakes in the Sudbury area in 2001

Lake Immature Males Females Total n Median age (range) Mean7SD fork length (mm) Mean7SD weight (g)

Barlowa 0 6 14 20 2 (2–4) 125749 27725Big Marsha 0 10 10 20 2 (2–4) 103723 15716Bircha 0 0 20 20 4.5 (3–11) 242748 1737102Hannah 3 11 6 20 2 (1–3) 95725 1178Kelly 1 10 9 20 1 (1–3) 97714 1176Kusk 2 4 14 20 2 (1–5) 97742 22751McCharles 14 6 0 20 2 (2–2) 8677 972Mud 9 0 10 19 1 (0–2) 92731 1279Ramsey 5 2 13 20 3 (2–5) 121730 23725Simon 4 2 14 20 2 (1–2) 100717 1375Whitson 3 5 12 20 2 (2–4) 102715 1175Waubamaca 8 6 6 20 2 (2–5) 106725 16719

aIndicates reference lake.


Muscle Zn concentrations ranged from 23.5 to276.9 mg/g among study lakes (Table 11). Fish fromGroup 1 or Group 2 lakes had relatively low muscle Znconcentrations. Fish from Group 3 lakes had muscle Znconcentrations that were approximately four timeshigher than those from the other two lake groups(F ð2;228Þ ¼ 4:9; P ¼ 0:008) (Fig. 6).

3.4.3. Intestine

Intestinal Cd concentrations were below 2 mg/g in fishfrom all lakes (Table 12). Only fish from Whitson andHannah Lakes had intestinal Cd concentrations greaterthan 1 mg/g (1.24 and 1.34 mg/g dry wt, respectively).Group 1 fish had the lowest intestinal Cd concentra-tions, followed by Group 2 fish, which had significantly

ARTICLE IN PRESS

Table 10

Total liver metal concentrations (mg/g dry wt) in yellow perch from 12 Sudbury-area study lakes

Lake Ca Cd Co Cu Fe Mg Mn Ni Pb Se Zn

Group 1

Barlow Mean 3407abc 1.34de 3.7d 22.5abcd 633abcd 1108ab 31.3b 4.4bc 2.7b 6.3ef 132.4bc

SEM 648 0.18 0.4 3.5 70 59 5.8 0.6 0.5 0.7 12.4

N 20 19 12 20 20 20 20 14 6 18 20

Big Marsh Mean 1509d 1.12de 10.0abc 8.6de 839abc 1077abc 6.8def — 3.6ab 6.8ef 99.1bc

SEM 532 0.16 1.3 5.0 102 58 0.6 — — 1.1 8.9

N 20 17 8 7 20 20 20 0 1 17 20

Birch Mean 1908abcd 1.02e 7.1cd 32.8a 1163a

829d

59.0a 4.0c 2.0

b4.8

f 116.0bc

SEM 184 0.13 0.9 3.2 125 34 8.2 0.4 0.3 0.4 5.9

N 20 18 20 20 20 20 20 20 7 20 20

Waubamac Mean 1693bcd 0.78e 5.6cd 13.0bcde 919ab 1319a 13.2cd 20.4abc 12.0ab 5.0ef 93.2c

SEM 236 0.11 0.7 1.8 80 80 2.6 13.6 — 0.4 9.3

N 20 9 14 20 20 20 20 13 1 17 20

Group 2

Kelly Mean 3343ab 1.19de 7.4bcd 17.1bcde 282f 1061bc 4.7

f 5.4abc 17.9ab 90.4a 153.8bc

SEM 562 0.11 1.0 5.3 28 84 0.3 1.7 5.0 6.1 20.1

N 20 20 15 11 20 20 19 6 4 20 20

Mud Mean 3311ab 2.14cd 10.8bc 32.1abc 504cdef 916bcd 6.2cdef 5.5abc 7.8b 28.7bc 133.0bc

SEM 523 0.38 2.2 4.1 71 35 1.0 0.7 1.4 3.2 12.7

N 19 13 14 13 17 18 7 9 8 19 19

Simon Mean 1780bcd 1.83cd 17.4a 6.9e 356def 996bcd 5.8ef 3.8

c 2.1ab 19.7cd 136.0bc

SEM 219 0.14 1.2 3.1 31 44 0.9 0.4 — 1.3 14.0

N 20 20 20 8 20 20 20 19 1 20 20

Group 3

Hannah Mean 3590ab 7.49b 10.1bc 87.8a 539cdef 871bcd 6.2def 8.3bc 41.3a 31.9b 187.1b

SEM 1016 1.03 1.2 26.4 81 24 0.5 2.8 10.8 2.2 26.2

N 20 20 18 18 20 19 20 16 12 20 20

Kusk Mean 1415cd 2.90c 17.4a 12.2bcde 320f 1072abc 13.0cd 5.3abc 2.1ab 12.5def 231.3b

SEM 260 0.34 1.3 2.9 30 56 2.3 0.6 0.4 1.1 44.7

N 20 16 20 16 20 20 20 9 2 20 20

McCharles Mean 4207a 2.31c 19.5

a 50.5a 351ef 995bcd 13.9bcde 22.1a 9.6b 17.9d 179.1b

SEM 564 0.18 1.7 3.8 62 40 3.9 8.6 1.6 1.2 17.5

N 20 20 20 20 20 20 10 18 13 20 20

Ramsey Mean 2745abcd 4.20c 6.8cd 15.3cde 413def 1086bc 16.5bc 22.4ab 6.0b 15.7de 1441.0a

SEM 944 0.55 1.1 3.2 49 110 2.7 9.7 1.4 0.8 435.4

N 20 20 18 16 20 20 20 9 8 20 20

Whitson Mean 1923bcd 22.37a 12.8ab 36.0ab 758bcde 866cd 9.4cdef 8.1bc 4.2ab 22.4bcd 189.4b

SEM 430 2.20 1.2 6.2 217 28 1.4 2.3 2.6 0.5 32.7

N 20 20 20 19 20 20 20 18 2 19 20

Note. For any given metal, means sharing a similar superscript are not significantly different from one another (P40:05). Maximum and minimum

values are indicated in boldface. Lake groups are the same as in Table 4; see text for details.


lower concentrations than Group 3 fish, which hadsignificantly higher intestinal Cd concentrations thanfish from any other group (F ð2;146Þ ¼ 14:9; Po0:0001)(Fig. 8).The lowest mean intestinal Cu concentrations were

measured in fish from Simon Lake, a metal-contami-nated lake (4.070.9 mg/g) (Table 12). Intestinal Cuconcentrations in fish from Group 2 and 3 lakes did notsignificantly differ from each other. However, fish from

Group 1 lakes had significantly lower intestinal Cuconcentrations than those from Group 2 lakes(F ð2;182Þ ¼ 4:2; P ¼ 0:02) (Fig. 8).Intestinal Se concentrations were less than 2 mg/g in

Group 1 lakes (Table 12). Intestinal Se concentrationswere lowest in fish from Group 1 lakes, followed bythose in Group 3 lakes, which were 31% lowerthan those in Group 2 lakes (F ð2;180Þ ¼ 41:7;Po0:0001) (Fig. 8). Interestingly, intestinal Se was

ARTICLE IN PRESS

Fig. 4. Comparison of liver Cd, Cu, Ni, Se, Pb, and Zn concentrations in yellow perch collected from each of three lake groups identified by water

metal PCA (Fig. 2). Concentrations are reported as means7SEM (n in parentheses). Mean differences were compared by ANOVA followed by

Tukey–Kramer multiple comparison test. Within each panel, bars sharing the same letter are not significantly different from one another (P40:05).

Fig. 5. Relationship between liver Cd concentration and water

hardness in 9 of the 12 study lakes. Each point represents mean7SEM(n ¼ 9� 20) liver Cd concentration in yellow perch from each lake as a

function of mean7SEM (n ¼ 3) water hardness. Black points

represent metal-contaminated lakes (i.e., Group 2 and 3), white points

represent reference lakes (i.e., Group 1).


positively correlated with intestinal Cu in males(r ¼ 0:46; P ¼ 0:006; n ¼ 34), but negatively correlatedin females (r ¼ �0:24; P ¼ 0:03; n ¼ 80).Intestinal Zn concentrations ranged from 28.8 to

404.9 mg/g among all study lakes (Table 12). Theintestinal Zn concentrations in Ramsey Lake (Group3) fish were at least four times higher than those in fishfrom any other lake. Intestinal Zn concentrations in fishfrom Group 1 or Group 2 lakes were not significantlydifferent. However, Group 3 fish had mean intestinal Znconcentrations that were at least twice as high as thosein other lake groups (F ð2;222Þ ¼ 7:2; P ¼ 0:0009) (Fig. 8).

3.5. Partial correlation analysis

A five-compartment partial correlation analysis wasconducted using water, sediment, intestine, muscle, andliver concentrations of Cu, Cd, Ni, Se, and Zn. Thisanalysis evaluates relationships between pairs of vari-ables while holding common variance among remainingvariables constant. Copper and Ni were highlycorrelated between water and sediment compartments

ARTICLE IN PRESS

Table 11

Total muscle metal concentrations (mg/g dry wt) in yellow perch from 12 Sudbury-area study lakes

Lake Ca Cr Cu Fe Hg Mg Mn Ni Se Zn

Group 1

Barlow Mean 3913a 38.5a 7.5ab 93.1a 0.50ab 610abc 16.1ab 2.2a 1.5f 34.1bc

SEM 819 25.0 0.4 7.9 0.05 29 4.2 0.2 0.1 3.9

N 20 20 20 18 20 20 17 13 6 20

Big Marsh Mean 742cd 10.1ab 3.6cd 441.0a 0.74a 455c 7.4bc 144.5a 1.6ef 23.5c

SEM 139 6.6 0.6 296.7 0.06 29 4.2 143.3 0.4 7.0

N 20 19 11 8 19 20 8 2 4 19

Birch Mean 1423abc 2.7ab 13.3a 87.0a 0.75a 789a 19.0

a 3.2a 1.4f 39.6bc

SEM 253 0.2 2.4 12.9 0.11 22 2.6 0.7 0.1 2.9

N 20 20 20 16 20 20 20 19 12 20

Waubamac Mean 1803bc 2.6ab 6.2bc 64.9a 0.52ab 472bc 3.2c 2.5a 1.5f 24.8bc

SEM 545 0.3 0.7 6.0 0.06 22 0.9 0.8 0.1 2.7

N 20 20 19 15 19 20 16 9 7 14

Group 2

Kelly Mean 1293abc 52.9a

14.1ab

494.9a 0.18d 742a 8.9bc 40.5a 24.0

a 44.2b

SEM 155 37.0 7.0 304.5 0.02 33 5.5 30.3 1.7 5.2

N 20 20 3 9 15 20 8 12 19 20

Mud Mean 798bcd 3.6ab 5.1bcd 93.4a 0.10d 506bc 1.7bc 1.5a 7.7c 26.5bc

SEM 108 0.1 0.5 — 0.00 54 0.7 0.2 1.2 2.9

N 19 19 10 1 4 19 2 8 19 19

Simon Mean 1168bc 4.9ab 11.0bcd 99.8a 0.26abcd 712a 1.7c 5.4a 10.3b 29.5bc

SEM 212 0.9 8.0 15.5 — 21 0.3 3.7 0.6 2.1

N 20 20 7 3 1 20 9 5 20 20

Group 3

Hannah Mean 1654ab 9.4ab 11.2ab 198.4a 0.16d 735a 2.4c 6.5a 11.8b 64.2b

SEM 255 5.8 3.7 108.7 0.01 36 1.1 3.7 0.6 24.1

N 20 20 5 5 14 19 10 11 19 20

Kusk Mean 1752abc 3.1ab 2.9d 122.4a 0.39bc 756a 2.8c 1.4a 3.8d 276.9

b

SEM 514 0.4 0.1 22.3 0.04 42 0.5 0.1 0.4 169.6

N 20 20 11 12 18 20 15 4 20 19

McCharles Mean 513d 3.1ab 6.8b 96.7a 0.10

bcd228

d1.0

abc 1.4a 2.3def 24.9bc

SEM 85 0.3 0.5 31.2 — 15 — 0.1 0.1 3.3

N 20 20 20 5 1 20 1 10 20 20

Ramsey Mean 996cd 1.8b 3.1d 76.2a 0.25cd 622ab 3.5bc 6.4a 3.3de 266.5a

SEM 428 0.2 0.7 15.6 0.05 26 0.9 2.8 0.3 64.5

N 20 18 9 8 14 20 11 4 19 20

Whitson Mean 1461bc 1.9b ND 85.1a 0.15d 843a 3.2bc 1.0

a 5.6c 33.5bc

SEM 301 0.1 — 17.1 0.02 99 0.6 0.0 0.4 4.8

N 20 20 — 4 9 20 11 2 20 20




(Fig. 9). Copper displayed no clear relationships betweenenvironmental compartments and tissues. However, therewas a weak positive relationship between intestinal andmuscle Cu. Partial correlations using water or muscle Cdcould not be evaluated owing to the low concentrations inthese compartments. However, intestinal Cd was stronglyrelated to liver Cd. There were opposite relationshipsbetween Ni in yellow perch tissues and Ni in water orsediment. Liver Ni was negatively correlated with waterNi, but muscle Ni was positively correlated with water

Ni. However, liver Ni was positively correlated withsediment Ni, but muscle Ni was negatively correlatedwith sediment Ni. All relationships involving sediment Sewere weak, whereas all those involving water Sewere strong and positive. Conversely, all relationshipsinvolving sediment Zn were strong and negative, whereasthose involving water Zn were weak. Intestinal Seand Zn were strongly and positively related to muscleand liver Se and Zn, which were strongly correlated witheach other.

ARTICLE IN PRESS

Fig. 7. Relationship of yellow perch muscle Se and Hg concentrations

as a function of distance from major industrial activities in Sudbury,

Ont. Points represent mean7SEM (sample sizes provided in Table 11)

log(x)-transformed data for Hg and Se.

Fig. 6. Comparison of muscle Cu, Hg, Se, and Zn concentrations in yellow perch collected from each of three lake groups identified by water metal

PCA (Fig. 2). Labeling conventions follow Fig. 4.


3.6. Fish condition

Fish condition was analyzed on the basis of the watermetal PCA groupings using age-corrected metrics (Fig.10). Fish from Group 1 lakes demonstrated significantlylower FCFs than fish from either Group 2 or Group 3lakes (F ð2;228Þ ¼ 96:0; Po0:0001). Fish from Group 2lakes had 2.1 times higher FCFs than fish from Group 1lakes, and 1.7 times higher than those from Group 3lakes.Age-corrected HSI demonstrated trends similar to

that for FCF, in that fish from Group 1 and 3 lakes hadthe lowest HSI, while fish from Group 2 lakes had the

highest HSI (F ð2;228Þ ¼ 28:9; Po0:0001) (Fig. 9). TheHSI in fish from Group 2 lakes was approximatelydouble that of fish from either Group 1 or Group 3lakes.GSI was assessed only for mature male and female

fish (Fig. 10). In males, fish from Group 1 fish had anintermediate GSI relative to fish from Group 2 andGroup 3 (F ð2;59Þ ¼ 24:7; Po0:0001). Fish from Group 2lakes had the highest mean GSI, which was 1.8 timeshigher than that in fish from Group 1 lakes and 4.6times higher than that in Group 3 lakes. Female GSIwas highest in Group 1 fish relative to those from Group2 and 3 lakes (F ð2;125Þ ¼ 10:0; Po0:0001). Mean femaleGSI of fish from Group 1 lakes was 1.4 to 2 times higherthan those of Group 2 and 3 lakes, respectively. GSI didnot significantly differ among females of Group 2 and 3lakes.

3.7. Correlation analysis

A Pearson correlation analysis was conducted todetermine significant relationships among fish conditionand tissue metal concentration with water qualitycharacteristics (pH, alkalinity, hardness, and conductiv-ity), distance from ore smelting activities, and sediment(Cd, Co, Cr, Cu, Ni, Se, and Zn) and water (Ca, Cu, Ni,Se, and Zn) metals (Table 13). Uncorrected FCF andHSI, male GSI, Shannon’s diversity index, liver metals(including Cd, Co, Cu, Ni, Se, and Zn), muscle Cr andCu, and intestinal Co, Cu, and Ni were not significantlyassociated with any of the correlates (P40:05). All thenonmetal water quality variables (pH, alkalinity, hard-ness, conductivity), sediment Cu and Ni, and water Ca,Cu, Ni, and Se were negatively related to distance from

ARTICLE IN PRESS

Table 12

Total intestine metal concentrations (mg/g dry wt) in yellow perch from 12 Sudbury-area study lakes

Lake Ca Cd Co Cr Cu Fe Hg Mg Mn Ni Se Zn

Group 1

Barlow Mean 10781a 0.56cd 0.34abc 31.9a 18.1bc 359ab 0.27b 440ab 42.4ab 2.7ab 1.4de 73.9b

SEM 5346 0.21 0.14 20.6 3.0 109 0.04 89 15.4 0.6 0.2 18.3

N 17 11 8 16 17 17 16 17 17 11 8 17

Big Marsh Mean 8418ab 0.24cd 0.67abc 5.3ab 12.2bcd 601a 0.29b 333abc 32.5ab 5.3ab 1.3cde 55.7bcd

SEM 3976 0.06 0.25 1.3 2.6 219 0.03 74 8.3 1.6 0.1 16.3

N 14 5 9 13 14 14 14 13 14 5 4 14

Birch Mean 1826ab 0.16d 0.28bc 6.9ab 15.0bc 179bc 0.64a 386a 29.1a 5.9ab 1.5e 73.0b

SEM 292 0.03 0.03 3.0 0.8 20 0.09 21 3.1 1.8 0.1 10.4

N 20 6 17 20 20 20 19 20 20 20 15 20

Waubamac Mean 983bcd 0.15cd 0.21c 3.2bc 8.7cd 124cd 0.29b 261bc 12.7bcd 1.7b 1.4cde 31.8d

SEM 307 0.04 0.05 0.8 1.0 18 0.02 14 3.4 0.3 0.1 5.8

N 20 2 8 19 20 19 19 20 20 11 6 20

Group 2

Kelly Mean 1572bc 0.44bcd 0.34abc 2.8bc 29.0ab 140cd 0.26b 352abc 2.8f 4.4ab 8.7a 50.7bc

SEM 455 0.09 0.08 0.2 6.4 42 0.02 38 0.4 1.1 1.4 6.0

N 18 14 13 18 13 15 15 16 15 16 18 18

Mud Mean 924bcd 0.28cd 0.25c 3.4abc 11.0de 127bcd 0.21b 260cd 2.5f 2.9ab 3.2bc 62.3b

SEM 182 0.04 0.05 0.5 5.4 24 0.03 46 0.3 0.4 0.5 7.6

N 19 12 15 19 15 10 5 18 19 14 19 19

Simon Mean 1296bc 0.88ab 0.68a 2.6bc 4.0

e 78d ND 382ab 5.1def 6.1a 5.9a 76.8b

SEM 190 0.14 0.10 0.1 0.9 12 — 17 0.7 1.2 0.4 7.5

N 20 20 20 20 13 14 — 20 20 19 20 20

Group 3

Hannah Mean 1225bc 1.34a 0.51abc 5.2ab 39.5a 101cd 0.37b 354ab 8.5cde 7.6a 6.7a 71.9b

SEM 102 0.12 0.22 0.5 5.0 10 0.06 19 2.3 2.8 0.6 3.3

N 18 18 15 18 18 17 18 18 18 18 18 18

Kusk Mean 830cd 0.38cd 0.67ab 5.8bc 8.0de 107cd 0.32b 376ab 17.5abc 4.4ab 2.9bcd 97.8b

SEM 161 0.05 0.10 3.7 2.2 16 0.04 27 3.7 1.0 0.3 18.9

N 20 13 19 18 15 19 14 20 20 14 20 20

McCharles Mean 445d 0.36cd 0.52abc 11.8bc 7.3de 332abcd 0.10b

148d 3.6ef 10.3

ab 1.5e 28.8cd

SEM 85 0.04 0.13 9.6 1.0 166 — 19 0.9 6.7 0.1 4.5

N 20 20 17 20 20 4 1 12 14 18 16 20

Ramsey Mean 419d 0.49bc 0.25bc 1.5c 5.7de 90d 0.33b 346ab 14.7abc 6.1ab 3.0bc 404.9a

SEM 60 0.07 0.03 0.1 0.7 18 0.05 25 1.6 2.3 0.2 78.6

N 20 10 11 14 17 12 14 20 20 10 20 20

Whitson Mean 973bcd 1.24a 0.23c 2.5bc 5.0cde 66cd 0.18b 357ab 11.4bcd 2.7ab 3.5b 60.4b

SEM 197 0.16 0.05 0.3 1.9 9 0.02 22 1.6 0.5 0.3 4.9

N 19 18 19 19 3 4 11 19 19 8 19 19




smelting operations. Age-corrected FCF was signifi-cantly related to water quality characteristics, distancefrom smelter, sediment Co and Ni, and waterborne Ca,Ni, and Se. Age-corrected HSI was related to hardnessand conductivity. Female GSI was significantly andnegatively related to sediment metals (Cd, Co, Ni, andSe) and waterborne Se. Muscle Hg concentrations werenegatively related to water quality characteristics (excepthardness), and positively related to distance fromsmelter. Interestingly, muscle Ni concentrations were

not associated with sediment or water Ni concentrationsor water quality characteristics. Muscle Se concentra-tions were positively related to water quality character-istics, sediment Ni and Cu concentrations, andwaterborne metals, including Se. Muscle Se was alsonegatively related to distance from smelter. Muscle Znconcentrations were negatively related to sediment Znconcentrations, and unrelated to waterborne Zn con-centrations. Intestinal Cd concentrations were nega-tively related to distance from smelter, and positively

ARTICLE IN PRESS

Fig. 9. Partial correlation analysis (‘‘path analysis’’) of five metals (Cu,

Cd, Ni, Se, and Zn) in a five-compartment model (water, sediment,

intestine, muscle, and liver). Arrows represent most likely direction of

metal flow. Values are partial correlation coefficients for individual

metals between any two compartments (while holding common

variance with other variables constant). N/A, not available.

Fig. 8. Comparison of intestinal Cd, Cu, Se, and Zn concentrations in yellow perch collected from each of three lake groups identified by water metal

PCA (Fig. 2). Labeling conventions follow Fig. 4.


related to sediment and water Cu and Ni concentra-tions, but not to sediment Cd concentrations. IntestinalCr and Hg exhibited only negative relationships inindividual sediment metals (Cu and Co, respectively).Intestinal Se was positively related to pH, conductivity,sediment Cu and Ni, and all the waterborne metalstested, and negatively related to distance from smelter.Intestinal Zn had only negative associations with somesediment metals (i.e., Co, Cr), including sediment Zn.A second Pearson correlation analysis was conducted

to examine relationships between tissue metal concen-tration and fish condition, including age-corrected FCFand HSI and male and female GSI (Table 14). None ofthe condition metrics were associated with liver, muscle,or intestinal Cu or Ni (P40:05). FCF was negativelycorrelated (Po0:05) with liver Cd, and muscle andintestinal Hg, but positively related to liver Pb and liver,muscle, and intestinal Se. Liver Se had the strongestrelationship with FCF. The HSI was negatively related(Po0:05) to muscle Hg and intestinal Cd and Hg, butpositively related to liver and muscle Se. Male GSIexhibited significant negative relationships with liverand intestinal Cd and positive relationships with liverand muscle Se. Female GSI was negatively associatedwith liver and intestinal Se and intestinal Hg, andpositively related to intestinal Cu and Zn.

4. Discussion

Industrial activities have been taking place in theSudbury region for more than century, resulting in someof the most heavily metal-contaminated environments inthe world (Keller and Gunn, 1995). Consequently, there

are several studies that report on metal concentrationsin the aquatic ecosystems of the Sudbury region (Nriaguet al., 1982, 1998; Nriagu and Wong, 1983; Nriagu,1983; Carignan and Nriagu, 1985; Bradley and Morris,1986). However, most of these studies focus on theprimary metals of concern, such as Cu, Ni, and Zn. Fewstudies have looked at the range of metals that arereported in this study in water, sediments, and fish

ARTICLE IN PRESSG.G. Pyle et al. / Ecotoxicology and Environmental Safety 61 (2005) 287–312 305

tissues in an attempt to relate environmental metalcontamination to fish diversity, tissue metal accumula-tion, and fish condition.Multivariate characterization of water metal concen-

trations led to three distinct clusters of study lakes:Group 1 included reference lakes, Group 2 consisted ofthree proximal lakes in the Kelly Lake chain (includingKelly Lake itself), and Group 3 included distal lakes onthe Kelly Lake chain and other metal-contaminatedSudbury-area lakes. Group 1 lakes were characterizedby having low alkalinity, low conductivity, low hard-

Table 13

Pearson correlation coefficients for significant relationships between yellow

characteristics (pH, alkalinity, hardness, conductivity), distance from ore sme

pH Alk. Hard. Cond. Dist. Sediment metals (mg/g)

Cd Co Cr C

Dist. �0.86a �0.78b �0.63c �0.76b — n.s. n.s. n.s. �

FCFe 0.62c 0.75b 0.91d 0.90d �0.67c n.s. 0.66c n.s. n

HSIe n.s. n.s. 0.79b 0.74b n.s. n.s. n.s. n.s. n

GSIfemale n.s. n.s. n.s. n.s. n.s. �0.62f �0.63c n.s. n

Mus. Hgf �0.73b �0.60a n.s. �0.58c 0.84a n.s. n.s. n.s. �

Mus. Ni n.s. n.s. n.s. n.s. n.s. �0.61f n.s. n.s. n

Mus. Se 0.77b 0.62a 0.71b 0.81e �0.86a n.s. n.s. n.s.

Mus Zn n.s. n.s. n.s. n.s. n.s. n.s. n.s. �0.66f n

Int. Cd n.s. n.s. n.s. n.s. �0.67c n.s. n.s. n.s.

Int. Cr n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. �

Int. Hg n.s. n.s. n.s. n.s. n.s. n.s. �0.60c n.s. n

Int. Se 0.73b n.s. n.s. 0.69c �0.81e n.s. n.s. n.s.

Int. Zn n.s. n.s. n.s. n.s. n.s. n.s. �0.61c �0.60f n

Note. Correlations among tissue, sediment, and water metal concentratio

(P40:05). Alk., alkalinity (mg/L as CaCO3); Hard., hardness (mg/L as CaCaPo0:05:bPo0:01:cPo0:001:dPo0:0001:eCorrected for age.fMus.=muscle; Int.=intestinal.

Fig. 10. Comparison of age-corrected FCF, age-corrected HSI, male

GSI, and female GSI among the three lake groups identified by PCA

(Fig. 2). Labeling conventions follow Fig. 4.

ness, low water pH, low waterborne metal (except Al)concentrations, low sediment Cu and Ni concentrations,but high waterborne Al concentrations. AlthoughGroup 1 lakes did not violate the Ontario ProvincialWater Quality Objectives (PWQOs) for As (PWQO,5 mg/L), Cu (5 mg/L), Ni (25 mg/L), Se (100 mg/L), or Zn(30 mg/L), they exceeded PWQOs for Al (PWQO, 75 mg/L) and Fe (300 mg/L) (OMEE, 1994). Group 2 lakeswere characterized by high alkalinity, high conductivity,high hardness, high pH, high waterborne metal (exceptAl, which was intermediate between Group 1 and 3lakes) concentrations, and high sediment Cu and Niconcentrations. All Group 2 lakes exceeded the PWQOsfor Cu, Fe, and Ni. In fact, Ni concentrations in Group2 lakes exceeded the Ni PWQO by at least an order ofmagnitude. Group 3 lakes were characterized by highpH, waterborne Cu, and sediment Cu and Ni, inter-mediate alkalinity, conductivity, and waterborne metals(except Al, Cu, and Fe), and low hardness and water-borne Al and Fe. All Group 3 lakes exceeded thePWQOs for Cu and Ni.Sediment metal concentrations were consistently

higher than waterborne metal concentrations, which isin good agreement with other studies on Sudbury-arealakes (Nriagu et al., 1982, 1998; Audet and Couture,2003). Although metal levels were elevated in sedimentsrelative to water, only sediment Cu and Ni weresufficiently elevated to discriminate between referenceand metal-contaminated lakes, and were negativelycorrelated with distance from smelting operations,

perch condition and tissue metal accumulation and water quality

lting operations, and sediment and water metal concentrations (n ¼ 12)

Water metals (mg/L)

u Ni Se Zn Ca Cu Ni Se Zn

0.87a �0.86a n.s. n.s. �0.79b �0.87a �0.94d �0.76b n.s.

.s. 0.62c n.s. n.s. 0.89d n.s. 0.73e 0.78b n.s.

.s. n.s. n.s. n.s. 0.69c n.s. n.s. n.s. n.s.

.s. �0.69c �0.63f n.s. n.s. n.s. n.s. �0.60c n.s.

0.81e �0.85a n.s. n.s. �0.68c �0.82e �0.81e n.s. n.s.

.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s.

0.76e 0.83d n.s. n.s. 0.76b 0.81e 0.90d 0.83a 0.62c

.s. n.s. n.s. �0.70c n.s. n.s. n.s. n.s. n.s.

0.67f 0.60c n.s. n.s. n.s. 0.70c 0.64c n.s. n.s.

0.62f n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s.

.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s.

0.76e 0.76b n.s. n.s. 0.64c 0.77b 0.84a 0.80b 0.58c

.s. n.s. n.s. �0.81b n.s. n.s. n.s. n.s. n.s.

ns were conducted using log-transformed data. n.s., not significant

O3); Cond., conductivity (ms/cm); Dist., distance from smelter (km).

ARTICLE IN PRESS

Table 14

Pearson correlation coefficients describing relationships between tissue

metal concentrations and wild yellow perch condition of fish caught

from 12 lakes along a metal contamination gradient

Tissue Metal FCF HSI Male GSI Female GSI

Liver Cd �0.16a n.s. �0.38b n.s.

Se 0.57d 0.43d 0.42c �0.29c

Pb 0.26a n.s. n.s. n.s.

Muscle Hg �0.39d �0.30c n.s. n.s.

Se 0.42d 0.25c 0.52c n.s.

Intestine Cd n.s. �0.20a �0.42b n.s.

Cu n.s. n.s. n.s. 0.20a

Hg �0.22c �0.27c n.s. �0.27b

Se 0.27c n.s. n.s. �0.20a

Zn n.s. n.s. n.s. 0.23b

Note. Only significant relationships are shown. FCF, age-corrected

Fulton’s condition factor; HSI, age-corrected hepatosomatic index;

GSI, gonadosomatic index.aPo0:05:bPo0:01:cPo0:001:dPo0:0001:


which agrees with other studies (Nriagu et al., 1982;Bradley and Morris, 1986). Sediments provide a long-term record of industrial contamination in metal-polluted systems. Consequently, elevated Cu and Niconcentrations in lake sediments in Sudbury-area lakesprobably reflect the long-term deposition of these metalsthrough atmospheric deposition and surface waterrunoff into these lakes, whereas water metals probablyreflect short-term conditions.Nriagu et al. (1998) have demonstrated that industrial

activities since the late 1880s in the region have resultedin saturation of Cu and Ni in local watersheds.Although emission abatement programs have beenoperating for more than 30 years, previously depositedCu and Ni can serve as a source of contamination formany years to come. Lake sediments are the ultimatereceptors of Cu and Ni inputs, which is reflected in ourdata. Chemical and physical processes at the sediment–-water interface can potentially recycle Cu and Ni backinto the water column, thereby serving as a source ofmetal contamination in contaminated lakes (Belzile andMorris, 1995). The high positive partial correlation ofCu and Ni between water and sediment compartmentsobserved in this study (Fig. 9) supports this argument.The multivariate analysis of water metal contamina-

tion could not account for the significant variabilityassociated with fish diversity among the 12 study lakes.However, a combination of four sediment metals (As,Co, Cu, and Mn) could account for 92% of thisvariability. Regression estimates for Co, Cu, and Mnhad a positive sign, whereas As yielded a negative sign inthe relationship with diversity. Both Co and Cu can beremoved from the water column through the formation

of Mn-oxyhydroxide precipitates (Belzile and Morris,1995). Therefore, removal of Co and Cu from the watercolumn through the formation of Mn-oxyhydroxideprecipitates, coupled with a concomitant decrease inavailable As, may provide conditions suitable for thereestablishment of fish communities in recovering lakes.The fact that the water metal characterization could notaccount for fish diversity in these lakes, but sedimentmetal concentrations could, suggests that long-termprocesses are more important for the reestablishment offish species to metal-contaminated lakes relative toshort-term processes. Although this might seem intui-tive, this conclusion requires caution because ouranalysis did not account for the myriad of biotic (e.g.,predator–prey relationships) or abiotic (temperature,dissolved oxygen, etc.) processes typically associatedwith the establishment and maintenance of stable fishcommunities (Jackson et al., 2001).Yellow perch was the only species present in all 12

study lakes (Table 8). Industrial activities in the Sudburyarea resulted in severe acidification of local lakes,resulting in widespread loss of aquatic species fromcontaminated systems (Keller et al., 1992). Fish com-munity structure in many acidified systems suffered theloss of many acid-sensitive species, which allowed acid-tolerant species, such as yellow perch (Eaton et al.,1992), to establish and dominate ecological niches leftbehind by extirpated species. Consequently, the presenceof yellow perch in all 12 study lakes not only reflects itsrelative ubiquity in northern Ontario lakes (Scott andCrossman, 1973), but also reflects the effects ofhistorical acidification of local watersheds.Liver Cd, Cu, Ni, and Zn were highest in yellow perch

from Group 3 lakes relative to those from any other lakegroup (Fig. 4). Group 3 lakes were characterized ashaving intermediate water hardness relative to the otherlake groups. Hardness cations, such as Ca2+ and Mg2+,are well known to decrease aqueous metal uptake by outcompeting metal cations at uptake sites (Pagenkopf,1983; Taylor et al., 2000; Pyle et al., 2002). However,water pH can also influence metal bioavailability andsubsequent uptake by fish. At low pH, free ionic metalspecies tend to dominate total metal concentration, andat high pH, metals tend to form inorganic complexeswith hydroxides and carbonates (Stumm and Morgan,1981). Free ionic metal ions that dominate under lowpH conditions are more bioavailable, and thus moretoxic, than inorganic metal complexes (Kushner, 1993).The high pH associated with Group 3 lakes would implythat waterborne metals should have been less bioavail-able than if pH were lower. However, our data suggestthat the intermediate water hardness probably had amore important role than water pH in terms ofmediating liver metal uptake in Group 3 lakes.The role of water hardness was particularly evident

for Cd accumulation in livers of yellow perch inhabiting


contaminated lakes (Fig. 5). The contribution ofaqueous Cd concentration or speciation to liver Cdconcentration could not be estimated in our study, eventhough aqueous Cd was not detected in most lake watersin this study and was therefore not reported, itsconcentration increased in lakes near contaminationpoint sources in other studies of the same area (P.Couture and G. Pyle, unpublished results). Therefore,aqueous Cd likely also influenced liver Cd concentra-tion. Nevertheless, the relationship reported here (Fig.5) does indicate that hardness is an important influenceon liver Cd concentration in wild yellow perch.Similarly, Cd concentrations in Sudbury lake sedimentshave been reported to be elevated above backgroundconcentrations (Nriagu et al., 1998). Yellow perchprobably take up Cd through a dietary route byconsuming Cd-contaminated benthic macroinverte-brates that inhabit Cd-contaminated sediments. Audetand Couture (2003) demonstrated that yellow perchliver Cd concentrations were related to diet in fishinhabiting Whitson Lake, which is a Group 3 lake in thisstudy. Moreover, elevated liver Cd concentrations havebeen associated with physiological dysfunction andimpaired swimming performance in wild yellow perchfrom Sudbury-area lakes (Rajotte and Couture, 2002;Couture and Kumar, 2003; Audet and Couture, 2003).Zohouri et al. (2001) and Baldisserotto et al. (2004)

demonstrated that rainbow trout fed a Ca-supplementeddiet took up less waterborne Cd (to gills, liver, andkidney) than fish fed a normal diet. They reasoned thatdietary Ca may be transported to the gills where itblocks waterborne Cd uptake. We speculate that theinfluence of water hardness on liver Cd accumulation inyellow perch from contaminated lakes in this study maybe due to an effect related to but opposite that reportedby the two recent studies (Zohouri et al., 2001;Baldisserotto et al., 2004). Because waterborne Cdconcentrations were so low in this study, it is not likelythat they contributed significantly to the elevated liverCd concentrations observed in fish from contaminatedlakes; dietary Cd is a more likely source. This wascorroborated by the high intestinal Cd concentrationobserved in fish from metal-contaminated lakes (Groups2 and 3, Fig. 8) and a high partial correlation coefficientbetween intestinal and liver Cd, suggesting that liverCd likely originated from the diet (Fig. 9). It may be thatmuch more waterborne Ca is taken up through thegills in fish inhabiting hard-water lakes, thus interferingwith intestinal Cd uptake processes, analogous tothe dietary Ca interference of branchial Cd uptakereported by Zohouri et al. (2001) and Baldisserottoet al. (2004). In soft-water lakes, less Ca is availableto accommodate this putative mechanism, leadingto higher Cd uptake and subsequent accumulation inliver. Controlled laboratory studies are required toverify this speculation.

Couture and Rajotte (2003) argued that Cu metabo-lism in fish is under homeostatic regulatory controlbecause Cu is an essential element. In metal-contami-nated lakes, liver Cu concentrations are usually regu-lated below 50 mg/g dry wt. However, once thisthreshold is exceeded, Cu homeostatic control mechan-isms become overloaded and liver Cu concentrationscan increase. Data reported here support this idea. Theonly cases of yellow perch liver Cu concentrationsexceeding the 50 mg/g threshold occurred in Group 3lakes (i.e., Hannah and McCharles lakes; Table 10).Liver Cu concentrations in fish from Hannah Lakeexceeded liver Cu concentrations in fish from everyother lake by at least 42%, demonstrating the loss ofregulatory control of liver Cu in these fish. Moreover,the weak partial correlations between environmentalcompartments and fish tissues, and between liver andmuscle Cu (Fig. 9), in addition to the lack of significantPearson correlations between water or sediment Cuconcentrations and concentrations measured in tissues(Table 13), also suggest that internalized Cu is tightlyregulated.Hepatic Se concentrations were highest in yellow

perch from Group 2 lakes, which included Kelly, Mud,and Simon lakes. Selenium enrichment has beenidentified in Kelly Lake, among other Sudbury-arealakes, in previous studies (Nriagu and Wong, 1983;Chen et al., 2001). Sediment Se concentrations did notvary among the three lake groups; however, waterborneSe concentrations exceeded 4 mg/L in Group 2 lakes, andwere the highest compared with other lake groups.Probably several different factors contributed to thehigh liver Se concentrations in yellow perch from Group2 lakes. Kelly Lake has received raw sewage in the past,and currently receives treated sewage effluents (Gunnand Keller, 1995). Microbiological processes in lakesediments can convert Se into organic forms, such asmethylated selenides and selenoamino acids, for exam-ple, selenocystein and selenomethionein (Chau et al.,1976; Maier and Knight, 1994). Although organic Secompounds are generally volatile, they are readilyabsorbed by primary producers and accumulate inprimary consumers (Besser et al., 1993). Fish feedingon primary consumers can efficiently accumulateorganic Se (Bertram and Brooks, 1986), which couldcontribute to the high hepatic Se concentrationsobserved in this study.Inorganic Se, primarily in the form of selenate and

selenite, can be taken up by fish directly from the water.However, once taken up by fish, this Se is associatedwith unbound intracellular pools and is rapidly excreted(Bertram and Brooks, 1986). Dietary Se is associatedwith bound intracellular pools in the fish and is excretedmuch more slowly, resulting in greater bioaccumulationfrom dietary sources than from waterborne sources(Hodson and Hilton, 1983; Bertram and Brooks, 1986).


Accumulation of waterborne Se concentrations of2–5 mg/L in food organisms is sufficient to cause toxicityin some fish (Maier and Knight, 1994). Waterborne Seconcentrations in all Group 2 lakes of this study werewithin this potentially toxic threshold range. Theelevated intestinal Se concentrations in fish from Group2 lakes (Fig. 4), the strong Pearson correlation betweenfish muscle and intestinal Se concentrations and waterSe (Table 13), and the high partial correlation coeffi-cients between water and intestine, between intestineand liver and muscle, and between liver and muscle (Fig.9) suggest that the diet is an important source of Se inthese fish. Moreover, liver Se concentrations above10 mg/g dry wt can be associated with toxic effects(Maier and Knight, 1994), and a tissue-based Secriterion of 4 mg/g has been proposed to protect aquaticlife (Hamilton, 2002). Liver Se concentrations in yellowperch from Group 2 lakes ranged between 19.7 and90.4 mg/g, and those from Group 3 lakes ranged between12.5 and 31.9 mg/g, exceeding the 10 mg/g threshold by asmuch as nine times. Liver Se concentrations in Group 1lakes were all below the 10 mg/g threshold. This suggeststhat fish from Group 2 and 3 lakes could demonstrateSe-related toxicity. However, our understanding of Setoxicity to fish remains incomplete, and caution iswarranted in interpreting these results (e.g., Kennedy etal., 2000).In general, metal concentrations in muscle tissue were

considerably lower than those in liver tissue, which isconsistent with other studies (Bradley and Morris, 1986;Rajotte and Couture, 2002). Muscle Cu concentrationswere significantly lower in Group 3 fish relative toGroup 1 fish (Fig. 6). It is therefore possible that lowermuscle Cu may be related to the increased deposition ofCu in the liver of Group 3 fish. That is, increased Culigands in liver could result in decreased muscle Cuconcentrations.Muscle Hg concentrations were highest in Group 1

lakes and lowest in the metal-contaminated Groups 2and 3 (Fig. 6). Conversely, muscle Se concentrationswere lowest in Group 1 lakes and significantly elevatedin Group 2 and 3 lakes (particularly Group 2 lakes).Selenium has a well-known antagonistic influence on Hgaccumulation (reviewed in Goyer, 1997). Although a fewstudies have demonstrated the effectiveness of decreas-ing Hg uptake into fish by raising waterborne Seconcentrations experimentally (Turner and Swick,1983; Paulsson and Lundbergh, 1991), few studies havedemonstrated that this effect occurs under naturalconditions. Results from the current study demonstratemuscle Se concentration is negatively correlated, andmuscle Hg concentration is positively correlated, withincreasing distance from smelting operations (Fig. 7).This result demonstrates the antagonistic influence of Seon Hg accumulation, and corroborates another recentstudy demonstrating the same effect in wild yellow perch

and walleye populations from Sudbury-area lakes (Chenet al., 2001).In contrast to muscle Cu concentrations, muscle Zn

concentrations were highest in Group 3 lakes (Fig. 6)—the same lakes demonstrating the highest liver Znconcentrations. Neither water nor sediment Zn concen-trations showed appreciable variation among the threelake groups (Tables 5 and 6). However, Group 3 lakesdemonstrated intermediate hardness relative to otherlake groups, which may have contributed to the patternof muscle Zn concentrations observed. Zinc uptake isdependent on waterborne Ca2+ concentrations (Spryand Wood, 1989). Once taken up, Zn typicallyaccumulates in gill and muscle tissues (McGeer et al.,2000).Zinc is an essential micronutrient, like Cu, and is

under tight homeostatic control (Watanabe et al., 1997).Although Zn can be taken up by fish through bothwaterborne and dietary routes, branchial uptake isprobably more important (Spry et al., 1988). Homeo-static control of Zn uptake is regulated primarilythrough excretory mechanisms and by controllinggastrointestinal uptake (Handy, 1996). This is reflectedin the strong negative correlations between muscle andintestinal Zn concentrations and industrial metal con-centrations (including Zn) in sediments, and the strongnegative partial correlation between sediment andintestinal Zn (Table 13), suggesting that fish incontaminated lakes decrease dietary Zn uptake (Fig.9). Although dietary Zn uptake is reduced in fish fromcontaminated lakes, it is not eliminated, as demon-strated by the elevated intestinal Zn concentrations inGroup 3 fish (Fig. 8). Zinc that is absorbed via theintestine is strongly related to liver and muscle Zn (Fig.9). Similarly, Zn that accumulates in the liver is stronglyrelated to muscle Zn, suggesting a possible transportpathway between liver and muscle tissue (Fig. 9). Theweak partial correlation coefficients between aqueousZn and intestinal, muscle, and liver Zn are possibly areflection of the homeostatic regulation of internalizedZn. However, the low water hardness observed inGroup 3 lakes may have facilitated Zn uptake fromthe water and subsequent accumulation in muscle (andliver) without competitive interference from hardnesscations at gill uptake sites.Both sediment Cu and Ni concentrations were

elevated in Group 2 and 3 lakes (Table 6). However,intestinal Ni concentrations did not vary in yellow perchamong lake groups, and only Group 2 fish hadsignificantly higher intestinal Cu relative to Group 1fish (Fig. 8). The elevated intestinal Cu in Group 2 fishprobably reflects elevated Cu concentrations in benthicfood items in Group 2 lakes.That both Ni and Cu are elevated in sediments of

metal-contaminated Sudbury-area lakes and Ni is notelevated in yellow perch intestines in Group 2 and 3


lakes suggest that dietary Ni uptake and distributionmechanisms may be distinct from those controllingdietary Cu uptake. In a laboratory study, lake whitefishfed Ni-supplemented diets demonstrated an initialincrease in intestinal Ni concentration during the first10 days of exposure, followed by a significant decreasein intestinal Ni that persisted for the remainder of the104-day study (Ptashynski and Klaverkamp, 2002). Thissuggests that physiological protective mechanisms in theintestine may act to regulate dietary Ni uptake. Partialcorrelation analysis of our data demonstrated that Niaccumulation in muscle was positively related to waterNi and negatively related to sediment Ni. However, Niaccumulation in liver tissue was negatively related towater Ni and positively related to sediment Ni (Fig. 9).At present, the physiological mechanisms that drive Niuptake, accumulation, and metabolism are unknown.Our data, however, suggest that there may be twoseparate mechanisms driving Ni accumulation in muscleand liver tissues; that, is, sediment (presumably dietary)Ni is preferentially transported (by an unknown path-way) to liver, whereas waterborne Ni is preferentiallytransported to muscle (by another unknown pathway).Fish condition was assessed using three different

metrics: FCF, HSI, and GSI for mature male and femalefish, respectively. FCF is a measure of somatic energyreserves associated with recent feeding activity, HSI isassociated with liver energetic reserves and metabolicactivity, and GSI is a measure of reproductive potential.Condition factors can be influenced by age, in partbecause young fish have different feeding rates andmetabolic activity associated with rapid growth relativeto older fish (Farkas et al., 2003). In addition, becausethe condition factor calculated in this study used anexponent of 3.0 (see Materials and Methods), and thescaling coefficient of yellow perch is naturally above 3,smaller fish will normally have lower condition factorscompared with larger fish of the same populations (seeEastwood and Couture (2002), and Couture and Rajotte(2003), for a discussion of the scaling coefficient and ofthe condition factor). Although we attempted to samplefish of the same size in any given lake (i.e., similar age)during this study, there was some size and age variabilityin our samples, which prevented the detection ofsignificant differences in yellow perch FCF or HSIamong the three lake groups. However, by correctingboth FCF and HSI for age, significant effects wereevident. Neither male nor female GSI was corrected forage because only mature, breeding fish were included inthe analysis.Fish from Group 2 lakes had significantly higher age-

corrected FCF and HSI values than fish from the othertwo lake groups. Other studies report lower FCF andother metrics of morphometric condition in yellowperch from metal-contaminated lakes relative to fishfrom reference lakes (Brodeur et al., 1997; Laflamme

et al., 2000; Rajotte and Couture, 2002; Eastwood andCouture, 2002). However, these studies examined fish ofa narrower and more consistent size range than thosereported here, and condition calculations were not age-corrected. Moreover, the lakes included in these studieshad a much narrower range of limnological productivitythan the range of lakes considered here.There are probably a couple of factors contributing to

the elevated FCF in fish from Group 2 lakes. Forexample, Kelly, Mud, and Simon lakes, which constituteGroup 2, are very productive lakes owing to thedomestic wastewater discharge into Kelly Lake. Theelevated FCF probably reflects greater feeding activityin fish from Group 2 lakes relative to feeding activity infish from other lakes. This is further supported byelevated HSI (Fig. 10) in the same fish, reflecting storageof energetic reserves, presumably because of theincreased feeding rates. Fish can compensate formetal-induced ionoregulatory dysfunction or increasedenergetic costs of detoxification if enough food isavailable (D’Cruz et al., 1998). Nonetheless, liver Cdconcentrations and muscle and intestinal Hg concentra-tions were inversely related to age-corrected FCF(Table 14), suggesting that these metals may still leadto a decrease in growth-related condition in wild yellowperch. The positive relationship between tissue Seconcentrations and FCF (Table 14) is probably areflection of the elevated feeding status of fish fromGroup 2 lakes relative to other lakes, leading to the hightissue Se concentrations observed.Fish from Group 3 lakes had significantly higher

FCFs relative to fish from Group 1 lakes. Thisdifference probably reflects seasonal effects on fishcondition. Audet and Couture (2003) demonstrated thatyellow perch living in a metal contaminated lake duringthe summer had increased condition relative to yellowperch inhabiting a reference lake. Fish from Group 3lakes were collected somewhat later in the season thanGroup 1 fish (Table 1).The effect of metal contamination is best reflected in

yellow perch GSI. Males from Group 2 lakes demon-strated higher GSIs than males from any of the othertwo lake groups. This elevated GSI may be a reflectionof feeding status (as discussed above) resulting in moreenergetic resources being diverted to gonad developmentin postspawn animals. As with the positive correlationsbetween FCF and tissue Se, positive correlationsbetween male GSI and liver and muscle Se concentra-tions probably reflect a higher dietary Se uptake inGroup 2 fish associated with greater feeding activity.However, males from Group 3 lakes showed signifi-cantly decreased GSIs relative to males from the othertwo lake groups. This may be a reflection of metal-induced impairment of reproductive potential in theseanimals. Liver and intestinal Cd concentrations werenegatively related to male GSI, suggesting that dietary


Cd may contribute to decreased reproductive potentialin male yellow perch.Female GSI was significantly lower in fish from

metal-contaminated lakes (Groups 2 and 3) than in fishfrom reference lakes (Group 1) (Fig. 10). Female GSIwas negatively associated with liver and intestinal Seand intestinal Hg (Table 14). Also, female GSI displayeda positive relationship with intestinal Cu (Table 14).Interestingly, intestinal Se concentrations were posi-tively correlated with intestinal Cu in males, butnegatively correlated in females. This suggests thatdietary Cu and Se are antagonistic to intestinal uptakein females but not in males. This implies that theinteraction between Cu and Se in intestinal uptakepathways varies between males and females. AlthoughSe is known to be antagonistic to Cu toxicity inmammals (reviewed in Gaetke and Chow, 2003), less isknown about Cu–Se interactions in fish. Furthermore,there is no evidence suggesting that the mechanismvaries between males and females, as suggested by ourdata. Further study into the interaction between dietarySe and Cu and its effects on yellow perch reproduction iswarranted.Liver Se concentrations above 10 mg/g have been

linked to toxic effects in fish (Maier and Knight, 1994),and waterborne Se concentrations above 10 mg/L havebeen associated with reproductive impairment in fish(Hermanutz et al., 1992). Waterborne Se concentrationsnever exceeded 10 mg/L in this study; however, fish fromevery metal-contaminated lake had liver Se concentra-tions above 10 mg/g, suggesting that hepatic Se accumu-lation may be linked to potential reproductiveimpairment in fish from contaminated lakes. Despitethe antagonistic effect of Se on Hg bioaccumulation inmuscle tissue (Fig. 7), dietary Hg was associated withdecreased reproductive potential in female yellow perch.Consequently, environmental contamination, particu-larly by Se, may result in decreased reproductivepotential in female yellow perch in contaminated lakesrelative to reference lakes.

5. Conclusions

By exploiting a wide metal contamination gradient,our data provide insights into some of the complexrelationships among metals and other environmentalvariables that mediate tissue metal accumulation andcondition in wild yellow perch. Fish diversity along thecontamination gradient was associated with sedimentmetals only, suggesting that long-term processes areinvolved in reestablishing fish communities. However,our analysis did not examine other controlling factors,such as predator–prey relationships and habitat quality,which surely contribute to community structure andstability. Metal accumulation in yellow perch along the

contamination gradient is probably from branchial anddietary sources, depending on the specific metal, itsavailability, and the presence of competing factors suchas water hardness. Essential metals, such as Cu andZn, are tightly regulated in wild yellow perch, andtherefore, tissue Cu or Zn accumulation is not particu-larly useful for predicting fish condition in metal-contaminated lakes. Selenium, on the other hand, whichis also an essential micronutrient, was strongly asso-ciated with fish condition. However, caution mustbe used in interpreting the relationship between tissueSe concentration and the condition metric of interest.Because Se is primarily a dietary contaminant, laketrophic status probably interacts with Se accumulationsuch that fish having high tissue Se concentrations alsohave high growth condition factors (FCF and HSI).However, tissue Se was negatively associated withfemale GSI, suggesting that reproductive impairmentmay be experienced by fish that have accumulated highSe concentrations. The relationship among intestinal Seand Cu concentrations and male and female GSIsuggests that Cu and Se uptake dynamics maybe different in male and female fish, thereby affectingreproductive condition differently. Furthermore,muscle Se has an antagonistic relationship with muscleHg as a function of distance from industrialactivities. Selenium has received relatively little researchattention in aquatic ecosystems in the Sudburyarea, and more research is required. Liver Cd wasnegatively associated with growth condition (FCF) andreproductive condition in males (GSI). Our datademonstrated that water hardness was an importantfactor mediating hepatic Cd accumulation, and thatliver Cd accumulation is dominated by dietary sources.Consequently, we speculate that branchial Ca2+ uptakein hard water lakes probably contributes to a decrease indietary Cd uptake in wild yellow perch. Again,controlled laboratory experiments are warranted toverify this hypothesis.

Acknowledgments

The authors acknowledge G. Watson and K. Huntof INCO, Ltd., and M. Charbonneau and staff atTestmark Laboratories, Ltd., for their support andfinancial assistance. The authors also acknowledgethe following for their assistance: J. Coulas, C.Vernescu, K. Farquar, T. Champagne-Rajotte,D. Champagne, and R. Champagne. P.C. wassupported by an NSERC research grant, P.C. andG.P. by funding from INCO, Ltd., and J.R. by anIndustrial NSERC scholarship with INCO, Ltd., as asupporting partner.


References

Audet, D., Couture, P., 2003. Seasonal variations in tissue metabolic

capacities of yellow perch (Perca flavescens) from clean and metal-

contaminated environments. Can. J. Fish. Aquat. Sci. 60, 269–278.

Baldisserotto, B., Kamunde, C., Matsuo, A., Wood, C.M., 2004.

A protective effect of dietary calcium against acute waterborne

cadmium uptake in rainbow trout. Aquat. Toxicol. 67, 57–73.

Beamish, R.J., Harvey, H.H., 1972. Acidification of the La Cloche

Mountain Lakes, Ontario, and resulting fish mortalities. J. Fish.

Res. Bd. Can. 29, 1131–1143.

Belzile, N., Morris, J.R., 1995. Lake sediments: sources or sinks of

industrially mobilized elements? In: Gunn, J.M. (Ed.), Restoration

and Recovery of an Industrial Region. Springer, New York, pp.

183–193.

Bertram, P.E., Brooks, A.S., 1986. Kinetics of accumulation of

selenium from food and water by fathead minnow. Water Res.

20, 877–884.

Besser, J.M., Canfield, T.J., Lapoint, T.W., 1993. Bioaccumulation of

organic and inorganic selenium in a laboratory food chain.

Environ. Toxicol. Chem. 12, 57–72.

Bradley, R.W., Morris, J.R., 1986. Heavy metals in fish from a series

of metal contaminated lakes near Sudbury, Ontario. Water Air Soil

Pollut. 27, 341–354.

Brodeur, J.C., Sherwood, G., Rasmussen, J.B., Hontela, A., 1997.

Impaired cortisol secretion in yellow perch (Perca flavescens) from

lakes contaminated by heavy metals: in vivo and in vitro

assessment. Can. J. Fish. Aquat. Sci. 54, 2752–2758.

Carignan, R., Nriagu, J.O., 1985. Trace metal deposition and mobility

in the sediments of two lakes near Sudbury, ON. Geochim.

Cosmochim. Acta 49, 1753–1764.

Chau, Y.K., Wong, P.T.S., Silverberg, B.A., Luxon, P.L., Bengert,

G.A., 1976. Methylation of selenium in the aquatic environment.

Science 192, 1130–1131.

Chen, Y.W., Belzile, N., Gunn, J.M., 2001. Antagonistic effect of

selenium on mercury assimilation by fish populations near Sudbury

metal smelters? Limnol. Oceanogr. 46, 1814–1818.

Clesceri, L.S., Greenberg, A.E., Eaton, A.D. (Eds.), 1998, Standard

Methods for the Examination of Water and Wastewater, 20th ed.

Am. Public Health Assoc., Washington, DC.

Couture, P., Kumar, P.R., 2003. Impairment of metabolic capacities in

copper and cadmium contaminated wild yellow perch (Perca

flavescens). Aquat. Toxicol. 64, 107–120.

Couture, P., Rajotte, J.W., 2003. Morphometric and metabolic

indicators of metal stress in wild yellow perch (Perca flavescens)

from Sudbury, Ontario: a review. J. Environ. Monit. 5, 216–221.

D’Cruz, L.M., Dockray, J.J., Morgan, I.J., Wood, C.M., 1998.

Physiological effects of sublethal acid exposure in juvenile rainbow

trout on a limited or unlimited ration during a simulated global

warming scenario. Physiol. Zool. 71, 359–376.

Eastwood, S., Couture, P., 2002. Seasonal variations in condition and

liver metal concentrations of yellow perch (Perca flavescens) from a

metal-contaminated environment. Aquat. Toxicol. 58, 43–56.

Eaton, J.G., Swenson, W.A., McCormick, J.H., Simonson, T.D.,

Jensen, K.M., 1992. A field and laboratory investigation of acid

effects on largemouth bass, rock bass, black crappie and yellow

perch. Trans. Am. Fish. Soc. 121, 644–658.

Farkas, A., Salanki, J., Specziar, A., 2003. Age- and size-specific

patterns of heavy metals in the organs of freshwater fish Abramis

brama L. populating a low-contaminated site. Water Res. 37,

959–964.

Gaetke, L.M., Chow, C.K., 2003. Copper toxicity, oxidative stress,

and antioxidant nutrients. Toxicology 189, 147–163.

Goyer, R.A., 1997. Toxic and essential metal interactions. Annu. Rev.

Nutr. 17, 37–50.

Gunn, J.M., Keller, W., 1995. Urban lakes: integrators of environ-

mental damage and recovery. In: Gunn, J.M. (Ed.), Restoration

and Recovery of an Industrial Region. Springer, New York,

pp. 257–269.

Hamilton, S.J., 2002. Rationale for a tissue-based selenium criterion

for aquatic life. Aquat. Toxicol. 57, 85–100.

Handy, R.D., 1996. Dietary exposure to toxic metals in fish. In:

Taylor, E.W. (Ed.), Toxicology of Aquatic Pollution: Physiologi-

cal, Cellular, and Molecular Approaches. Society of Experimental

Biology Seminar Series. Cambridge University Press, Cambridge,

UK, pp. 29–60.

Hermanutz, R.O., Allen, K.N., Roush, T.H., Hedtke, S.F., 1992.

Effects of elevated selenium concentrations on bluegills (Lepomis

macrochirus) in outdoor experimental streams. Environ. Toxicol.

Chem. 11, 217–224.

Hodson, P.V., Hilton, J.W., 1983. The nutritional requirements and

toxicity to fish of dietary and waterborne selenium. Environ.

Biogeochem. 35, 335–340.

Jackson, D.A., Peres-Neto, P.R., Olden, J.D., 2001. What controls

who is where in freshwater fish communities—the roles of biotic,

abiotic, and spatial factors. Can. J. Fish. Aquat. Sci. 58, 157–170.

Keller, W., Gunn, J.M., 1995. Lake water quality improvements and

recovering aquatic communities. In: Gunn, J.M. (Ed.), Restoration

and Recovery of an Industrial Region. Springer, New York, pp.

67–80.

Keller, W., Gunn, J.M., Yan, N.D., 1992. Evidence of biological

recovery in acid-stressed lakes near Sudbury, Canada. Environ.

Pollut. 78, 79–85.

Keller, W., Heneberry, J.H., Gunn, J., 1999. Effects of emission

reductions from the Sudbury smelters on the recovery of acid-and

metal-damaged lakes. J. Aquat. Ecosyst. Stress Rec. 6, 189–198.

Kennedy, C.J., McDonald, L.E., Loveridge, R., Strosher, M.M., 2000.

The effect of bioaccumulated selenium on mortalities and

deformities in the eggs, larvae, and fry of a wild population of

cutthroat trout (Oncorhynchus clarki lewisi). Arch. Environ.

Contam. Toxicol. 39, 46–52.

Kushner, D.J., 1993. Effects of speciation of toxic metals on their

biological activity. Water Pollut. Res. J. Can. 28, 111–128.

Laflamme, J.-S., Couillard, Y., Campbell, P.G.C., Hontela, A., 2000.

Interrenal metallothionein and cortisol secretion in relation to Cd,

Cu, and Zn exposure in yellow perch, Perca flavescens, from

Abitibi lakes. Can. J. Fish. Aquat. Sci. 57, 1692–1700.

Maier, K.J., Knight, A.W., 1994. Ecotoxicology of selenium in

freshwater systems. Rev. Environ. Contam. Toxicol. 134, 31–48.

McGeer, J.C., Szebedinszky, C., McDonald, D.G., Wood, C.M., 2000.

Effects of chronic sublethal exposure to waterborne Cu, Cd or Zn

in rainbow trout: 2: tissue specific metal accumulation. Aquat.

Toxicol. 50, 245–256.

Nriagu, J.O., 1983. Arsenic enrichment in lakes near the smelters at

Sudbury, Ontario. Geochim. Cosmochim. Acta 47, 1523–1526.

Nriagu, J.O., Wong, H.K., 1983. Selenium pollution of lakes near the

smelters at Sudbury, Ontario. Nature 301, 55–57.

Nriagu, J.O., Wong, H.K.T., Coker, R.D., 1982. Deposition and

chemistry of pollutant metals in lakes around the smelters at

Sudbury, Ontario. Environ. Sci. Technol. 16, 551–560.

Nriagu, J.O., Wong, H.K., Lawson, G., Daniel, P., 1998. Saturation of

ecosystems with toxic metals in Sudbury basin, Ontario, Canada.

Sci. Tot. Environ. 223, 99–117.

Ontario Ministry of Environment and Energy (OMEE), 1994.

Provincial Water Quality Objectives. Queens Printer of Ontario,

Toronto.

Pagenkopf, G.K., 1983. Gill surface interaction model for trace-metal

toxicity to fishes: role of complexation, pH, and water hardness.

Environ. Sci. Technol. 17, 342–347.

Paulsson, K., Lundbergh, K., 1991. Treatment of mercury contami-

nated fish by selenium addition. Water Air Soil Pollut. 56, 833–841.


Ptashynski, M.D., Klaverkamp, J.F., 2002. Accumulation and

distribution of dietary nickel in lake whitefish (Coregonus

clupeaformis). Aquat. Toxicol. 58, 249–264.

Pyle, G.G., Swanson, S.M., Lehmkuhl, D.M., 2001. Toxicity of

uranium mine-receiving waters to caged fathead minnows,

Pimephales promelas. Ecotoxicol. Environ. Saf. 48, 202–214.

Pyle, G.G., Swanson, S.M., Lehmkuhl, D.M., 2002. The influence of

water hardness, pH, and suspended solids on nickel toxicity to

larval fathead minnows (Pimephales promelas). Water Air Soil

Pollut. 133, 215–226.

Rajotte, J.W., Couture, P., 2002. Effects of environmental metal

contamination on the condition, swimming performance, and

tissue metabolic capacities of wild yellow perch (Perca flavescens).

Can. J. Fish. Aquat. Sci. 59, 1296–1304.

Scott, W.B., Crossman, E.J., 1973. Freshwater Fishes of Canada.

Fisheries Research Board of Canada, Ottawa.

Sherwood, G.D., Rasmussen, D.J., Rowan, D.J., Brodeur, J., Hontela,

A., 2000. Bioenergetic costs of heavy metal exposure in yellow

perch (Perca flavescens): in situ estimates with a radiotracer (137Cs)

technique. Can. J. Fish. Aquat. Sci. 57, 441–450.

Sherwood, G.D., Kovecses, J., Hontela, A., Rasmussen, J.B., 2002a.

Simplified food webs lead to energetic bottlenecks in polluted lakes.


Sherwood, G.D., Pazzia, I., Moeser, A., Hontela, A., Rasmussen, J.B.,

2002b. Shifting gears: enzymatic evidence for the energetic

advantage of switching diet in wild-living fish. Can. J. Fish. Aquat.

Sci. 59, 229–241.

Spry, D.J., Wood, C.M., 1989. A kinetic method for the measurement

of zinc influx in vivo in the rainbow trout, and the effects of

waterborne calcium on flux rates. J. Exp. Biol. 142, 425–446.

Spry, D.J., Hodson, P.V., Wood, C.M., 1988. Relative contributions

of dietary and waterborne zinc in the rainbow trout, Salmo

gairdneri. Can. J. Fish. Aquat. Sci. 45, 31–41.

Stumm, W., Morgan, J.J., 1981. Aquatic Chemistry. Wiley, New York.

Taylor, L.N., McGeer, J.C., Wood, C.M., McDonald, D.G., 2000.

Physiological effects of chronic copper exposure to rainbow trout

(Oncorhynchus mykiss) in hard and soft water: evaluation of

chronic indicators. Environ. Toxicol. Chem. 19, 2298–2308.

Taylor, L.M., Wood, C.M., McDonald, D.G., 2003. An evaluation of

sodium loss and gill metal binding properties in rainbow trout and

yellow perch to explain species differences in copper tolerance.

Environ. Toxicol. Chem. 22, 2159–2166.

Turner, M.A., Swick, A.L., 1983. The English–Wabigoon river system:

IV. Interaction between mercury and selenium accumulated from

waterborne and dietary sources by northern pike (Esox lucius).


Watanabe, T., Kiron, V., Satoh, S., 1997. Trace minerals in fish

nutrition. Aquaculture 151, 185–207.

Yan, N.D., Welbourn, P.M., 1990. The impoverishment of aquatic

communities by smelter activities near Sudbury, Canada. In:

Woodwell, G.M. (Ed.), Patterns and Processes of Biotic Impover-

ishment. Cambridge University Press, New York, pp. 477–494.

Zohouri, M.A., Pyle, G.G., Wood, C.M., 2001. Dietary Ca inhibits

waterborne Cd uptake in Cd-exposed rainbow trout, Oncorhynchus

mykiss. Comp. Biochem. Physiol. C 130, 347–356.

Effects of industrial metals on wild fish populations along a metal contamination gradient

Documents