Responses of two sentinel species (Hexagenia limbata—mayfly; Pyganodon grandis—bivalve) along spatial cadmium gradients in lakes and rivers in northwestern Québec

Abstract

Specimens of the mayfly larva Hexagenia limbata and of the floater mussel Pyganodon grandis were sampled in rivers and lakes contaminated by trace metals in the Abitibi-James Bay region in northwestern Québec. Water samples were collected at each sampling site with in situ diffusion samplers and analyzed for major cations, anions and trace metals (Cd, Cu, Mn, Zn). Surficial sediment samples were also collected at each site and analyzed for Cd, Cu and Zn. In response to Cd contamination at river and lake sites, both sentinel organisms accumulated the metal and synthesized metallothionein (MT), a metal-binding protein synthesized by organisms as a defence mechanism against excess metals in the surrounding media. At the river sites, H. limbata unexpectedly maintained much higher concentrations of MT per unit of accumulated Cd than at the lake sites; this difference between lentic and lotic environments may reflect the response of the species to the more stressful hydrodynamic conditions that prevail in a river. The accumulation of Cd in the mayflies at lake and river sites decreased as a function of the ambient manganese concentration. We hypothesize that dissolved Mn protects against Cd bioaccumulation in H. limbata. The present results support the contention that one cannot extrapolate conclusions drawn from the use of a single sentinel species to a larger set of freshwater invertebrates—both the mayfly and the bivalve are promising biomonitors.

---

Environmental impact

The use of indigenous aquatic organisms as environmental biomonitors for trace metals is a well-established practice. In the present study, we have compared two sentinel organisms, both living at the sediment–water interface, and we have compared two different habitats (either lacustrine or riverine). For sites where the two sentinel organisms co-existed, the rankings of the sites with respect to cadmium contamination were different for each sentinel organism (presumably reflecting their different behaviours and different feeding strategies). Similarly, but unexpectedly, the same sentinel organism (the mayfly larva) exhibited different responses in the river and lake habitats. Metal bioavailability is clearly species-specific—from a practical environmental point of view, thus one should deploy a suite of complementary biomonitors.

1. Introduction

Metal contamination of streams and lakes is recognized as an important environmental concern, particularly in geographical areas where metal ores are extracted and refined.¹ Metals entering the freshwater environment partition among various compartments (water, suspended particulate matter, sediments, and biota). Among these compartments, sediments act as an important reservoir for metals, with concentrations reaching levels 1000–5000 times higher than those in the overlying water column.² In such environments, benthic animals can take up metals both from the water with which they are in contact and from the food that they eat.³ This dual exposure to metals constitutes a potential danger for benthic invertebrates.

Cadmium (Cd) is released into aquatic environments by mining, smelting, and refining processes.⁴ It is considered to be one of the more toxic metals, and its toxicity to aquatic invertebrates is well documented.⁵ Once accumulated by the benthic organisms, Cd can interfere with the regulation of essential metals (e.g., Zn and Cu) as well as disrupt Ca metabolism at Ca channels in gill epithelial cell membranes.⁶

At the cellular level, aquatic organisms respond to Cd contamination in part by induction of specific metal-binding proteins (MBPs). These MBPs show high affinity for group IB and IIB metal ions (e.g. Cd, Cu, Zn), and are thus able to complex these metals and regulate their intracellular speciation.⁷ Metallothionein (MT), an important MBP with numerous biochemical roles, helps protect tissues from metal damage.⁸ The use of this intracellular protein as a biomarker for exposure to elevated levels of trace metals in aquatic environments has been proposed as a tool in ecotoxicological studies.⁹ Changes at the biochemical level offer distinct advantages as biomarkers, since molecular alterations are normally the first detectable responses to environmental changes. Thus, by determining tissue concentrations of MT (with due consideration of potential season variability), one can in principle monitor changes in bioavailable levels of metals such as Cd and evaluate the biochemical state of organisms in Cd-contaminated environments.¹⁰

In previous studies we have documented strong positive relationships between Cd and MT in environmentally exposed organisms living in metal-contaminated lakes (e.g., molluscs,^11,12 insect larvae,¹³ fish¹⁴). Since anthropogenic metal releases are often directed towards running waters rather than lakes, we wished to extend this earlier work to organisms living in rivers. Compared to lakes, lotic ecosystems are primarily differentiated by a unidirectional water movement along a longitudinal gradient in response to gravity.¹⁵ In addition, more frequent fluctuations in metal concentrations and forms typically occur in small to medium-sized rivers compared to lakes.¹⁶ These differences may well influence how populations of a given benthic invertebrate species handle metals in lake and river habitats.

The filter-feeding bivalve Pyganodon grandis and the deposit-feeding mayfly larva Hexagenia limbata were chosen as sentinel organisms for this study. A preliminary study indicated that tissue levels of MT were better correlated with Cd than with other metals in these two species,¹⁷ and thus in the present study we have focused on Cd accumulation and Cd–MT relationships. We explore relationships between various environmental variables and the Cd concentrations measured in the two sentinel species. Relationships between accumulated cadmium and steady-state metallothionein concentrations are examined in the two sentinel species, and we then compare the resulting regression models for river and lake populations of the two species.

2. Materials and methods

2.1. Study areas and collection sites

Lakes. The Rouyn-Noranda study area is a major mining region in northwestern Québec (Fig. 1a). The surface is covered by glaciolacustrine plains and hills of a low elevation. Humoferric and ferrohumic podzolic soils are developed on well-drained till deposits. The region is underlain by the Abitibi greenstone belt which is the site of two geological faults rich in polymetallic deposits that have been exploited for nearly a century.¹⁸ Lakes in the area are subject to acidic and metallic contamination from mine sites, and are affected by atmospheric deposition, which principally affects areas in the quadrant SSE–NNE downwind from a smelter complex located in Rouyn-Noranda. Thirteen and twenty-two lakes situated along a metal contamination gradient were sampled for the mayfly and the bivalve surveys, respectively. Sediments, water and biota were collected at each site at depths varying from 1 to 3 m in littoral zones. Mayflies were collected in 1994 at a single station in each lake, whereas bivalves were collected in 1997 at one station in small water bodies (<1 km² surface area) and at 2–4 stations in larger lakes; samples from each station within a lake were treated separately.

Fig. 1 (a) Study lakes in the Rouyn-Noranda mining area. (b) Location of the 21 river stations sampled in the Allard and Colombière Rivers in northwestern Québec.

Rivers. Sampling was carried out in two rivers, the Colombière River close to the town of Val d'Or, Quebec, and the Allard River situated in the Matagami region (northwestern Québec) (Fig. 1b). Both rivers are affected by mining activities in their watersheds, but they are morphologically different. The Colombière River is a small stream, about 7 m wide, well sheltered by vegetation, with depths ranging between 1 and 3.5 m among sampling sites. Current speed measured at sites CO4 and CO7 can vary from negligible to 0.2–0.3 m s⁻¹ depending on rain events. In contrast, the Allard River is a larger river where the width can vary from 75 to 300 m at different sampling sites. The bathymetry of this river is very complex: maximum depths can reach up to 60 m in some reaches, but shallow waters (∼1.5 m depth) dominate and represent up to 80% of the river surface at some sites. Current speed is typically low to negligible at all sampling stations. In contrast to the Colombière River, in periods of strong wind the waters of the littoral zone in the Allard River are easily mixed.

A preliminary sampling campaign conducted at the end of October 2000 enabled us to identify potential sampling sites, representing a metal concentration gradient and characterized by the presence of at least one sentinel species. Twenty one such sites were sampled during the summer of 2001: 12 stations were established in the Allard River and 9 stations in the Colombière River (Fig. 1b). The stations were separated by a distance of about 2 km to minimize spatial autocorrelation and maximize the physico-chemical differences. Stations were chosen on the basis of their polymetallic contamination and the presence of at least one sentinel species.

2.2. Sentinel species sampling

We chose a filter-feeding bivalve, P. grandis, and a deposit-feeding mayfly larva, H. limbata, as our sentinel organisms. The selection of these sentinel species was based on the following criteria: relative abundance in the aquatic systems studied, relative immobility, ease of sampling, metal tolerance, metal bioaccumulation capacity, dynamics of metal accumulation, capacity to synthesize metallothionein, and available physiological and behavioural data. Allometric and ontogenic variations in metal accumulation and toxicity were minimized by use of a narrow range of size and/or age classes for the two sentinel species.

Adult specimens of P. grandis of similar size (lakes: 6.5–10.5 cm; N = 12; rivers: 5–6 y; 7–9 cm; N = 9 when possible) were collected by SCUBA divers before the onset of the molluscs’ reproductive cycle (late June–early July) and kept in coolers filled with site water until being processed in the field laboratory. Lake collections were performed in the whole littoral zone from 0 to 6 m depth. Gills from each bivalve were dissected within 12 h of collection, and gill tissues from three individuals were pooled (yielding three replicate samples per site). Sub-samples of the tissue homogenate were allocated for measurements of both metal and metallothionein concentrations, and for determinations of the dry to wet weight ratios. A number of bivalves collected in lakes were found to be gravid, and in these cases larvae were removed from the gills before pooling. P. grandis was found at a depth of 1.5 m at four stations in the Colombière river (CO1, CO2, CO8 and CO9), whereas it was present at ∼4.5 m depth at eight stations in the Allard River (AL1 to AL3 and AL8 to AL12).

H. limbata larvae (late instars, 1.5–2 mm, 40–60 specimens) were collected using a benthic net manipulated by divers, followed by a gentle sieving of the surface sediments accumulated in the net. Animals were allowed to depurate in river water in the field laboratory for 24 h before they were pooled into composite samples (lakes: at least 5 individuals; rivers: 10 individuals). For each site, replicate samples were prepared for metal determinations (N = 4 for lake sites and N = 3 for river sites). Similarly, three and four replicate samples were used for the metallothionein analyses. Mayflies were found at all stations in the Allard River, but they were not present at stations CO2 and CO5 in the Colombière River. The individuals were sampled at the same depth in the two rivers (∼1.5 m).

2.3. Analyses of biological samples

Lacustrine organisms. For bivalve gill samples, partially thawed tissues were gently homogenized under a nitrogen atmosphere using a motor-driven 50 mL glass tissue grinder (Duall Co., 80 rpm rotation speed). Homogenization was performed with an ice-cold 25 mM Tris buffer adjusted to pH 7.2; the buffer to tissue ratio was 3 : 1 (w/w). After centrifugation, sub-samples of the tissue homogenate were allocated for MT analysis, measurements of metals, as well as the determination of dry weight to wet weight ratios. For Hexagenia nymphs, homogenization was similar to that described above with the exception that whole animal composite samples were homogenized in 3 mL of the Tris buffer, regardless of their weight.

Bivalve homogenate (∼100 µg dry wt) was dried in an oven at 65 °C for 24 h and transferred into a Teflon digestion bomb. Ultra-pure concentrated nitric acid (3 mL) was added and the digestion was carried out in a microwave oven (700 W, ≤2 min) at a pressure of 6900 kPa. Cooled digests were diluted with ultra-pure deionized water to a final volume of 25 mL (final dilution factor applied to the dried homogenate ≈ 250 w/w) and metal concentrations were determined by flame atomic absorption spectrophotometry (FAAS, Varian Spectra AA20). Procedural blanks and two certified reference materials (TORT-1, lobster hepatopancreas, Marine Analytical Chemistry Standards Program, National Research Council of Canada, Ottawa, ON, Canada; SRM No. 1566, oyster tissue, US National Institute of Standards and Technology, Gaithersburg, MD) were analyzed during each analytical run. Blanks indicated negligible contamination (N = 8) and Cd concentrations in certified samples were within acceptable limits (oyster tissue: 83 ± 5%; TORT-1: 96 ± 3%). Insect homogenates were processed as above except for the following points. The dilution factor of acid and water to dried homogenate never exceeded 10, that is 100 µL per mg dry mass. Cadmium was determined by flameless AAS (Varian Spectra AA-30). Procedural blanks remained below Cd detection limits (N = 5) and Cd recovery from TORT-1 certified samples ranged from 81 to 95%.

Riverine organisms. For determinations of total Cd concentrations in soft tissues, homogenates (100–700 µg dry wt) of either the entire individual (H. limbata) or gills (P. grandis), prepared in 25 mM Tris buffer (pH and buffer to tissue ratio were similar to those used for lacustrine organisms), were freeze-dried for 5 days. Ultra-pure concentrated nitric acid (2–20 mL) was then added to the dried tissues and a pre-digestion was carried out for 24 h at room temperature. Sub-samples of the digests were then transferred to Teflon digestion bombs and complete digestion was carried out in an autoclave (∼3 h) at pressures up to 100 kPa. Cooled digests were diluted with ultra-pure deionized water to a final volume of 10 mL and the metal concentrations were determined by inductively coupled plasma atomic emission spectrophotometry (ICP-AES, Varian Vista Ax). Procedural blanks and a certified reference material (TORT-2) were analyzed during each analytical run. Blanks indicated negligible contamination and recoveries of cadmium were satisfactory (30.0 ± 0.4 µg g⁻¹ dry wt) and consistently close to the certified value for Cd (26.7 ± 0.6 µg g⁻¹ dry wt).

For MT analyses of riverine and lacustrine specimens, three sub-samples of the homogenate were centrifuged at 30 000 × g for 30 min at 4 °C, and the supernatant was analyzed for metallothionein with a Hg saturation assay adapted slightly from Dutton et al.,¹⁹ and described in detail by Couillard et al.¹¹ As a quality control, recovery of a MT standard (MT from rabbit liver, Sigma Chemical Co., St Louis, MO, USA) was determined with every assay. The mean recovery for 10 separate determinations was 106.6 ± 1.4% (SE) for the river collections, whereas for the lake collections the mean recovery for 19 separate determinations was 102 ± 3% (SE).

High purity water for analytical purposes (>17 Mohm cm) was obtained from a commercial system by means of mixed-bed ion-exchange, charcoal adsorption, and filtration (0.2 µm porosity) steps.

2.4. Water and sediment sampling and laboratory analyses

To evaluate habitat quality for the two sentinel species, a suite of variables was measured at each sampling site. The pH of the overlying water was measured in the field on 3–4 occasions during the summer sampling campaigns (pH meter calibrated daily). The other variables were measured twice at each site: June and September 1998 for the lake sites (Table 1); June and August 2001 for the river sites (Table 2). No water sampling other than that for pH determination was performed for the mayfly collections of 1994. Water samples were collected with in situ diffusion samplers to determine concentrations of dissolved metals, major anions and cations, humic and fulvic acids, and dissolved organic carbon. The diffusion cell consisted of a 250 mL jar (Nalgene #2118-0008) fitted with a Gelman HT-200 membrane (0.2 µm porosity). A custom closure was designed to hold the filter membrane. The diffusion cells were placed 30 cm above the sediment by divers and were left in the water for 2 weeks before collection. All cells were located in oxygenated waters. When brought to the surface, the membrane of the diffusion cell was washed with river water to remove particulates at its surface. The membrane was then pierced gently with a clean scalpel and the clean original cap was substituted for the custom-made filtering closure. Piercing of the membrane prevented suction of lake water trapped in the threads of the cap, thus preventing particles from seeping into the sample. All samples were bagged and placed on ice in a cooler and sent to the central laboratory in Quebec City within 24 h of retrieval from the stations.

Table 1 Physical features and mean summer water quality characteristics at the sites sampled in lakes of the Rouyn-Noranda area (Abitibi)^a

Lake name (code + number of stations)	Longitude	Latitude	Surface area/ha	Distance from smelter/km	Maximum lake depth/m	pH		DOC/mg L^−1	Ca/µM	Mn/nM	Cd/nM	Cu/nM	Zn/nM
						1997	1998
a Solute concentrations are reported for the dissolved phase. Each value for H^+ and Ca is the mean of the number of stations indicated in the column of the lake names. For the study with the mayfly larvae, only [H^+] is a mean estimated from a summer time series; DOC is dissolved organic carbon. b For the bivalve study, water samples were obtained on four occasions from each field station during the ice-free season (May–September). c The symbol “—” means that data are unavailable. d Baie de l'Orignal only.
Study with the bivalve P. grandis (1997)^b
Adéline (AD-1)	79° 11′	48° 12′	0.80	13.4	—^c	7.30	—	—	—	—	—	—	—
Beauchastel (BC-4)	79° 06′	48° 09′	8.23	13.0	—	7.54	—	—	—	—	—	—	—
Bousquet (BO-4)	78° 36′	48° 13′	2.39	31.0	22	6.40	6.87	12.0	105	249	0.55	62	40
Bouzan (BZ-1)	78° 58′	48° 12′	0.16	6.9	1.1	7.15	9.80	8.0	177	49.1	0.41	162	22
Caron (CA-3)	78° 55′	48° 00′	12.86	30.0	72	6.84	7.40	10.0	269	115	0.52	65	35
Cléricy (CL-2)	78° 45′	48° 18′	0.94	20.4	—	6.54	—	—	—	—	—	—	—
D'Alembert (DA-2)	79° 00′	48° 23′	1.23	14.6	—	7.05	—	—	—	—	—	—	—
Destor (DE-1)	79° 05′	48° 28′	0.62	23.9	—	7.06	—	—	—	—	—	—	—
Dasserat (DS-1)	79° 26′	48° 16′	26.86	30.7	17	7.20	7.76	6.5	210	67.3	0.29	52	78
Dufay (DU-3)	79° 28′	48° 02′	4.38	41.5	12	6.80	6.97	8.7	72.4	89.2	0.35	34	11
Evain (EV-2)	79° 17′	48° 07′	2.01	24.1	8	7.44	7.59	8.2	167	32.8	0.06	48	9
Flavrian (FL-1)	79° 12′	48° 19′	3.26	15.7	—	7.27	—	—	—	—	—	—	—
Héva (HE-1)	78° 19′	48° 12′	2.34	51.5	7	6.00	6.70	7.7	49.9	87.4	0.39	38	15
Hélène (HL-1)	79° 10′	48° 13′	0.54	12.0	—	7.73	—	—	—	—	—	—	—
Joannès (JO-3)	78° 41′	48° 11′	4.52	25.9	21	7.18	7.42	9.1	172	27.3	0.39	48	17
Moore (MO-1)	78° 57′	48° 13′	0.13	6.8	—	7.00	7.22	7.9	232	113	0.69	63	53
Ollier (OL-2)	79° 17′	48° 10′	0.84	21.9	11	7.51	7.59	6.2	309	692	0.04	46	8
Opasatica (OP-4)^d	79° 17′	48° 05′	9.68	27.4	14	7.55	7.82	6.8	207	85.5	0.06	55	9
Petit Dufresnoy (DF-1)	78° 58′	48° 26′	1.46	20.3	—	7.35	—	—	—	—	—	—	—
Renaud (RE-1)	79° 17′	48° 11′	1.0	21.2	1.5	8.55	9.29	5.9	444	16.4	0.06	56	7
Vaudray (VA-3)	78° 41′	48° 05′	7.68	30.6	30	6.51	7.25	6.8	74.9	21.8	0.57	43	32
Waza (WA-2)	79° 12′	48° 12′	0.43	15.3	—	7.55	—	—	—	—	—	—	—
Study with the burrowing larvae of the mayfly H. limbata (1994)
Adéline (AD-1)	79° 11′	48° 12′	0.80	13.4	—	7.26		5.6	197	—	—	—	—
Bousquet (BO-1)	78° 36′	48° 13′	2.39	31.0	22	6.38		14.6	105	—	—	—	—
Bruyère (BR-1)	78° 56′	48° 09′	3.88	12.7	4	7.33		11.4	344	—	—	—	—
Caron (CA-1)	78° 56′	48° 09′	12.9	30.0	72	7.11		13.3	245	—	—	—	—
D'Alembert (DA-1)	78° 56′	48° 09′	1.23	14.6	—	7.14		12.0	120	—	—	—	—
Dufay (DU-1)	79° 28′	48° 02′	4.38	41.5	12	6.55		10.6	84.8	—	—	—	—
Duprat (DP-1)	79° 08′	48° 20′	2.38	12.2	—	7.10		8.0	187	—	—	—	—
Flavrian (FL-1)	79° 12′	48° 19′	3.26	15.7	—	7.39		3.9	157	—	—	—	—
Héva (HE-1)	78° 19′	48° 12′	2.34	51.5	7	6.18		12.0	59.9	—	—	—	—
Joannès (JO-1)	78° 41′	48° 11′	4.52	25.9	21	7.22		9.4	170	—	—	—	—
Opasatica (OP-1)	79° 17′	48° 05′	9.68	27.4	14	7.39		9.8	165	—	—	—	—
Savard (SA-1)	78° 52′	48° 20′	—	13.9	—	7.15		3.4	—	—	—	—	—
Vaudray (VA-1)	78° 41′	48° 05′	7.68	30.6	30	6.56		6.3	89.8	—	—	—	—

---

Table 2 Mean summer water quality characteristics at the sites sampled in 2001 in the Colombière and Allard rivers^a

	Longitude	Latitude	pH	DOC/mg L^−1	FA/mg L^−1	Ca/µM	Mn/nM	Cu/nM
a Solute concentrations are reported for the dissolved phase. DOC = dissolved organic carbon; FA = fulvic acid.
Colombière river
CO1	77° 28′ 46.5″ N	48° 08′ 18.9″ W	6.6	12.9	5.9	317	754	24.4
CO2	77° 29′ 17.8″ N	48° 08′ 10.4″ W	6.2	14.3	5.7	185	2800	271
CO3	77° 36′ 50.8″ N	48° 08′ 32.7″ W	6.6	10.5	4.4	3890	2310	121
CO4	77° 37′ 06.1″ N	48° 08′ 42.7″ W	6.5	12.3	5.5	2260	1860	91.2
CO5	77° 37′ 38.8″ N	48° 08′ 39.5″ W	6.4	11.6	4.5	3720	2460	170
CO6	77° 37′ 45.4″ N	48° 09′ 05.4″ W	6.7	13.5	5.4	2280	1770	151
CO7	77° 38′ 05.6″ N	48° 09′ 10.0″ W	6.5	13.2	6.7	1040	1690	116
CO8	77° 37′ 18.8″ N	48° 09′ 07.8″ W	6.7	15.8	8.6	379	499	64.8
CO9	77° 37′ 33.4″ N	48° 09′ 07.6″ W	6.5	14.1	8.4	337	572	52.6
Allard river
AL1	77° 53′ 15″ N	49° 43′ 8.1″ W	7.2	15.8	7.8	379	129	<16
AL2	77° 52′ 4.5″ N	49° 43′ 53.2″ W	7.0	16.2	7.9	394	158	<16
AL3	77° 50′ 58.6″ N	49° 44′ 18.8″ W	7.2	17.4	8.3	462	164	<16
AL4	77° 50′ 27.2″ N	49° 46′ 16.4″ W	6.7	17.8	8.8	641	291	<16
AL5	77° 49′ 58.6″ N	49° 47′ 24.3″ W	7.0	18.7	9.8	1400	783	<16
AL6	77° 50′ 15.8″ N	49° 43′ 47.3″ W	6.9	24.0	13.7	259	122	<16
AL7	77° 50′ 33″ N	49° 42′ 50.8″ W	6.8	14.1	6.8	2170	886	<16
AL8	77° 49′ 31″ N	49° 43′ 30.4″ W	6.8	17.8	8.0	506	100	74.0
AL9	77° 50′ 11.7″ N	49° 48′ 24″ W	7.0	16.8	7.4	504	85.5	<16
AL10	77° 49′ 11.8″ N	49° 49′ 6.8″ W	6.9	18.4	8.3	497	45.5	<16
AL11	77° 47′ 35.7″ N	49° 49′ 33.3″ W	7.5	17.4	8.0	394	34.6	21.9
AL12	77° 47′ 38.3″ N	49° 49′ 35.9″ W	7.5	16.2	7.4	452	36.4	<16

---

When the samples arrived in the central laboratory, the diffusion cells were opened in a Class 100 laminar flow hood and sub-samples were taken for analysis. Two sub-samples were allocated for the determination of major cations (Ca, Mg, Na, K, Mn) and total dissolved trace metal concentrations (Cd, Cu, Zn). Depending on the metal to be analyzed and the concentration range, a variety of analytical techniques were used for these determinations: inductively coupled plasma mass spectrometry (ICP-MS, Thermo Elemental, Model X-7), inductively coupled atomic emission spectrometry (ICP-AES, AtomScan 25 spectrophotometer, Thermo Jarrell Ash), flame atomic absorption spectroscopy (FAAS, Spectra AA-20 spectrophotometer, Varian), and graphite furnace atomic absorption spectroscopy (GFAAS, SIMAA 6000, Perkin-Elmer). A third sub-sample was used for the determination of major anions (Cl, NO₃, SO₄, PO₄) by ion chromatography (DX-300 Gradient Chromatography Systems; Dionex). A final sub-sample was used for the determination of humic and fulvic acids (2001 only) by UV-visible spectrophotometry (λ = 285 and 326 nm: Varian, model Cary 100 Bio) and dissolved organic carbon by combustion and CO₂ detection (Shimadzu, model TOC-5000A). The quality assurance/quality control (QA/QC) methodology included the analysis of field blanks, laboratory blanks, and certified reference standards.

Free Cd²⁺ and Zn²⁺ were determined for the riverine study only, using an equilibrium ion-exchange technique.²⁰ Briefly, this technique consists of passing the filtered water sample through a sulfonic acid cation exchange resin held in a flow-through column system. When the total metal concentrations are identical in the inflow and outflow, indicating that equilibrium has been reached, the resin is rinsed, eluted with an acid solution and the eluate is analyzed for the metals of interest. The metal concentrations in the eluate are used to calculate the free-metal ion concentration in the original water sample, using the appropriate distribution coefficients as determined by calibration with solutions containing known free-metal ion concentrations (see ref. 20 for a detailed description). This technique was not available at the time the lake studies were conducted.

Surficial sediments for metal analyses were obtained using the same protocol for the riverine and lacustrine studies. Sediment core samples were collected by divers at each sampling site using hand-held corers. The cores were extruded in the boat and samples of the uppermost 0.5 cm, i.e. from the oxidized layer, were collected, pooled (two cores per replicate sample), placed in acid-cleaned polypropylene bottles, covered with river water, transported to the laboratory at 4 °C, and stored at −20 °C until analysis. Procedures for the metal extractions and analyses are described in detail by Couillard et al.¹¹

2.5. Statistical methods

Empirical relationships were established between Cd and metallothionein concentrations in the two species for each river separately, using the raw data. Normality of the data was verified using the Kolmogorov–Smirnov test. Analyses of covariance (ANCOVA) were used to test the hypothesis that the regression parameters (slope and intercept) of the Cd–MT relationships established in rivers differed significantly from the lake models for both species, and between the two rivers studied for both species. The model used for ANCOVA analyses²¹ is:

Y = β₀ + β₁X₁ + a_iC_i + d_iX₁C_i + ε (1)

where Y is the dependent variable (MT), ε is an error term, X₁ is the variable coding for the model in each test (e.g., 0 = Hexagenia in rivers and 1 = Hexagenia in lakes for this two-way comparison). C_i is the abiotic covariable in the model (Cd) and X₁ − C_i is the interaction term. Significant probabilities either for the main factor (X₁) or for the interaction term (X₁ − C_i) indicate a difference in the intercept and/or slope between regressions. Before each ANCOVA test, we verified the homogeneity of variances of the dependent variables using Bartlett's test.

Finally for lakes and rivers, multiple regression analyses with forward selection of the explanatory variables were used with [Cd]_body or [MT]_body (for mayflies) and [Cd]_gill (for molluscs) as the dependent variables. For multiple regressions, we considered the two rivers together in order to have sufficient data and a reasonable range of values for the explanatory variables. For all models, the forward selection procedure was stopped whenever the additional effect of the chosen variable had a P-value > 0.15. All data were log₁₀-transformed before analysis. The Kolmogorov–Smirnov test was used to verify that regression residuals were normally distributed and homoscedasticity of residuals was checked by examining the bivariate plots of residuals against predicted values. Regression analyses were performed using SYSTAT for Windows version 8.0.

3. Results

3.1. Physico-chemical water characteristics

Lakes. A wide range of dissolved hydrogen ion and metal concentrations was observed in our study lakes (Table 1), providing the environmental gradient required for testing the response of the two different sentinel organisms to different exposure conditions. Ratios of maximum to minimum values were 17, 4.7, and 11 for dissolved Cd, Cu and Zn concentrations, respectively. Manganese concentrations varied 42-fold along the contamination gradient. Dissolved Ca levels in the lakes chosen for the bivalve study showed appreciable inter-lake variability (max : min = 8.6), whereas they were somewhat less variable in lakes from which Hexagenia was collected (Table 1).

Rivers. Our river dataset also showed relatively variable water chemistry characteristics among stations and between rivers (Tables 2 and 3). Maximum : minimum ratios for Cd, Cu and Zn reflected the effects of point source inputs from mining operations, and were higher for the Colombière River than for the Allard River. The abnormally high spatial variability noted for Ca and Mn, with max : min values of 21 and 81, is a reflection of these anthropogenic inputs in the form of mine effluents that have been collected, treated by liming and then released into the rivers; this variability is particularly evident in the Colombière River system (Table 2). Compared to the downstream stations, the concentrations of Cd²⁺, Zn²⁺, dissolved Cd and dissolved Zn increased at stations close to the tailing ponds (Fig. 1b); the highest concentrations of these metals were detected immediately downstream from the tailing ponds (Tables 2 and 3: stations CO4, CO6; AL4, AL5).

Table 3 Mean total dissolved and free cation concentrations (nM) for Cd and Zn in the Colombière and Allard Rivers

River	2001				2002				2003^d
	Cd^2+	Cd^a	Zn^2+^b	Zn	Cd^2+	Cd	Zn^2+	Zn	Cd^2+	Cd	Zn^2+	Zn
Colombière
CO1	0.02	—^c	—	<15	0.03	0.73	37	86	—	—	—	—
CO2	0.04	—	—	<15	—	—	—	—	—	—	—	—
CO3	0.29	—	—	169	—	—	—	—	—	—	—	—
CO4	0.33	—	—	615	0.89	1.27	764	1030	—	—	—	—
CO5	0.51	—	—	793	0.85	1.10	741	1050	—	—	—	—
CO6	0.56	—	—	824	1.04	2.57	918	1230	—	—	—	—
CO7	0.25	—	—	348	0.74	1.21	739	1000	—	—	—	—
CO8	0.02	—	—	20	—	—	—	—	—	—	—	—
CO9	0.03	—	—	<15	—	—	—	—	—	—	—	—

---

a Total dissolved cadmium was below the detection limit in 2001 for the two rivers. Note that the analytical technique used to determine free Cd^2+ and Zn^2+ involved a pre-concentration step on an ion-exchange resin, yielding an analytical detection limit for Cd^2+ that was lower than that for total dissolved Cd. b Free Zn^2+ was not determined in 2001. c The symbol “—” means that data are unavailable. d The Colombière River was not sampled in 2003.
Allard	2001				2002				2003
	Cd^2+	Cd^a	Zn^2+^b	Zn	Cd^2+	Cd	Zn^2+	Zn	Cd^2+	Cd	Zn^2+	Zn
AL1	0.03	—^c	—	284	0.05	0.21	39	40	—	—	—	—
AL2	0.04	—	—	379	0.06	0.23	31	70	0.06	0.15	24	153
AL3	0.07	—	—	31	0.06	0.26	28	44	—	—	—	—
AL4	0.24	—	—	318	0.20	0.39	69	132	0.17	0.40	67	296
AL5	0.61	—	—	632	0.45	0.56	113	161	0.29	0.63	120	312
AL6	0.03	—	—	606	—	—	—	—	—	—	—	—
AL7	0.51	—	—	910	—	—	—	—	—	—	—	—
AL8	0.05	—	—	143	0.07	0.28	19	39	0.04	0.14	30	158
AL9	0.08	—	—	70	0.06	0.27	19	75	0.03	0.14	17	180
AL10	0.12	—	—	30	0.11	0.25	27	126	—	—	—	—
AL11	0.04	—	—	51	0.11	0.30	17	60	—	—	—	—
AL12	0.05	—	—	32	0.06	0.18	18	66	—	—	—	—
AL13	—	—	—	—	—	—	—	—	0.25	0.49	104	212

---

3.2. Trace metal concentrations in the sediment

Lakes. Trace metal concentrations exhibited appreciable spatial variability in the lakes, with max : min ratios of 77, 320 and 31 for Cd, Cu and Zn, respectively. These ratios were also somewhat higher in the lakes from which P. grandis was collected than in those where H. limbata was found (Table 4).

Table 4 Lake study of the bivalve P. grandis and the mayfly H. limbata. Mean cadmium and metallothionein (MT) concentrations in the tissues of the two sentinel species, and metal concentrations in sediments

Lake name	Sentinel species		Metals in sediments/nmol g^−1 dry wt
	Cd/nmol g^−1 dry wt	MT/nmol sites g^−1 dry wt	Cd	Cu	Zn
a The symbol “—” indicates either the absence of the sentinel species at this station or unavailable concentration data for the organism. b Baie de l'Orignal only.
Study with the bivalve P. grandis (1997)
Adéline	468	51	33.0	701	1450
Beauchastel	311	111	19.5	789	1850
Bousquet	910	183	10.4	143	1020
Bouzan	1300	163	188	11 100	7670
Caron	480	153	17.8	214	1930
Cléricy	271	98	12.4	266	1340
D'Alembert	778	103	54.3	1600	2660
Destor	23.1	59	28.8	549	1549
Dasserat	362	90	13.7	285	3070
Dufay	714	141	7.44	72.1	672
Evain	237	100	7.03	188	608
Flavrian	772	137	26.2	893	1530
Héva	856	313	3.02	65.9	672
Hélène	18.7	38	4.98	218	331
Joannès	1030	143	39.6	534	2060
Moore	872	91	339	21 100	10 100
Ollier	44.5	52	18.0	1460	2860
Opasatica^b	117	69	5.07	144	472
Petit	591	137	21.6	398	828
Renaud	57.8	61	16.5	2360	3570
Vaudray	2370	344	12.4	238	834
Waza	45.4	18	43.7	27 300	11 000
Study with the burrowing larvae of the mayfly H. limbata (1994)
Adéline	313	73	35.9	701	1445
Bousquet	93.7	18	33	143	1021
Bruyère	504	106	52.1	—^a	—
Caron	66.5	18	51.2	214	1926
D'Alembert	80.1	18	93.3	1595	2660
Dufay	88.1	26	9.13	72.1	672
Duprat	41.5	35	61.2	—	—
Flavrian	65.8	24	27.8	893	1534
Héva	71.1	33	10.4	65.9	672
Joannès	409	51	61.9	534	2059
Opasatica^b	165	23	4.4	144	472
Savard	207	35	—	—	—
Vaudray	265	38	38.3	238	834

---

Rivers. As expected given the local mining activities, Cd, Cu and Zn concentrations in sediments from both rivers were spatially variable, albeit less so than in lakes. The ratios of maximum to minimum concentrations ranged from 11.4 to 15 in the Colombière River, and from 5.3 to 11 in the Allard River (Table 5); in contrast, sediment Mg concentrations were much less variable. At all but two stations, the sediment trace metal concentrations decreased in the order Zn > Cu > Cd.

Table 5 River study of the bivalve P. grandis and the mayfly H. limbata. Mean cadmium and metallothionein (MT) concentrations in the tissues of the two sentinel species, and metal concentrations in sediments

	H. limbata		P. grandis		Metals in sediments
	Cd/nmol g^−1 dry wt	MT/nmol sites g^−1 dry wt	Cd/nmol g^−1 dry wt	MT/nmol sites g^−1 dry wt	Cd/nmol g^−1 dry wt	Cu/nmol g^−1 dry wt	Mg/µmol g^−1 dry wt	Zn/nmol g^−1 dry wt
a The symbol “—” indicates either the absence of the sentinel species at this station or unavailable concentration data for the organism. Further details for other metals are presented in Masson et al.^17 No mayflies or bivalves were found at CO5 in the Colombière River.
Colombière River
CO1	8.0	42.0	71.2	71.9	6.3	231	345	1380
CO2	—^a	—	62.3	64.3	6.5	220	342	1150
CO3	17.8	—	—	—	6.3	2390	388	4730
CO4	13.3	58.0	—	—	17.2	3290	448	12 400
CO6	22.2	40.3	—	—	8.9	2390	401	6360
CO7	15.1	38.9	—	—	9.2	2540	540	6310
CO8	44.5	53.2	28.5	42.4	1.5	1660	345	999
CO9	30.3	38.7	32.0	37.3	2.0	1890	342	1360
Allard River
AL1	72.1	66.1	17.8	26.1	3.8	278	605	1540
AL2	91.6	110	18.7	3.2	2.9	330	621	1760
AL3	119	120	33.8	28.3	4.6	393	662	15 200
AL4	84.6	96.9	—	—	15.3	1369	765	3240
AL5	22.2	49.6	—	—	13.7	2140	634	1350
AL6	46.3	89.6	—	—	3.1	362	642	13 700
AL8	133	173	107	36.2	4.9	551	664	3390
AL9	125	132	40.9	23.8	4.8	519	642	3960
AL10	98.8	115	48.0	28.6	5.7	441	671	3300
AL11	141	129	58.7	45.8	4.6	509	671	3490
AL12	149	145	82.7	32.9	4.8	346	662	3030

---

3.3. Metal and metallothionein concentrations in the sentinel species

Lakes. Animal Cd concentrations were more variable than those of MT with maximum to minimum concentration ratios of 12–127 for bioaccumulated Cd and 6–19 for MT (Table 4).

Rivers. In the whole body of the mayfly H. limbata, the range of Cd concentrations was wider for specimens collected in the Allard River than for those from the Colombière River (Table 5), but unlike the case for aqueous or sedimentary metals, no specific spatial pattern could be detected in either river. Total MT concentrations were higher in the whole body of mayflies collected in the Allard River than in those from the Colombière River (Table 5).

In the gills of the floater mussel P. grandis, the range of Cd concentrations was similar for specimens from the Colombière and Allard rivers. The range of gill MT concentrations was also similar in the two rivers, but the absolute values were slightly greater in the Colombière River than in the Allard River. Note, however, that in both rivers the sentinel bivalve was absent from the most contaminated stations (CO3 → 7, AL4 → 5) (Table 5).

3.4. Relationships between Cd and MT

Lakes. There was a delay of 3 to 4 years between the insect samplings (1994) and the bivalve/water collections (1997/1998). We assume here that no important environmental changes occurred in the lakes during this period, even though the Rouyn-Noranda region is slowly recovering from earlier historical contamination.²² We base our assumption on the following observations: there were no significant differences in MT levels in the gills of P. grandis specimens collected at the same six sampling stations in 1994 and 1997, and gill Cd levels differed significantly for only two of these six sampling sites.²² In addition to this, there was no significant difference between hydrogen ion concentrations for the lake sampling stations visited both in 1994 and in 1998 (paired t-test, t(6) = −2.175, P > 0.05).

Concentrations of MT were significantly correlated with animal Cd concentrations (Fig. 2a and b); in contrast, no significant relationships were obtained between MT and animal Cu or Zn (P > 0.05; results not shown). Note that this triggering role was exerted by Cd despite the fact that it was not the dominant element in the dissolved phase; molar concentrations of Cd were much lower than those of Cu or Zn in all the lakes studied (Table 1). The Cd–MT relationships for the lacustrine populations of P. grandis and H. limbata were both strong (r² = 0.68 vs. 0.74), and the increments of MT per unit of accumulated Cd were similar for both species (0.12 vs. 0.15, Table 6). Despite these similarities, marked differences are observed when metallothionein concentrations in P. grandis and H. limbata are compared for the lakes where these species co-existed. Lakes containing both species were ranked in the order of decreasing MT concentrations in these animals as follows (Table 4):

Table 6 Models for Cd–MT relationships for P. grandis (gills) and H. limbata (whole body) established in the present study, range of data for Cd–MT and dissolved Cd concentrations

Model type	Equation	Tissue MT/nmol sites g^−1 dry wt	Tissue Cd/nmol g^−1 dry wt	Aqueous Cd/nM
For H. limbata
Allard River	MT = 0.78Cd + 34.9; R^2 = 0.79	50–172	2.5–16.8	0.03–0.74 free ion
Rouyn-Noranda lakes	MT = 0.15Cd + 11.5; R^2 = 0.74	18–106	41.5–504	0.04–0.69 total dissolved
For P. grandis
Allard River	MT = 0.25Cd + 15.6; R^2 = 0.39	3.2–45.8	17.9–107	0.03–0.14 free ion
Rouyn-Noranda lakes	MT = 0.12Cd + 51.5; R^2 = 0.68	18–344	23–2374	0.04–0.69 total dissolved

---

Fig. 2 Relationships between bioaccumulated cadmium concentrations and metallothionein concentrations for the two sentinel species: (a) H. limbata, whole body concentrations for mayflies collected in the Rouyn-Noranda lakes; (b) P. grandis, gill concentrations for bivalves collected in the Rouyn-Noranda lakes; (c) H. limbata, whole body concentrations for mayflies collected in the Allard and Colombière rivers and (d) P. grandis, gill concentrations in bivalves collected in the Allard and Colombière rivers.

[MT] P. grandis: VA > HE > BO > DU ≥ JO > FL > DA > OP > AD

[MT] H. limbata: AD > JO > VA > HE > DU > FL ≥ OP > DA = BO The ranking of Cd concentrations in these species is not identical to the above sequences (Table 4), but animals with higher internalized Cd concentrations also tend to higher MT concentrations:

[Cd] P. grandis: VA > JO > BO > HE > DA ≥ FL > DU > AD > OP

[Cd] H. limbata: JO > AD > VA > OP > BO > DU > DA > HE > FL

Rivers. A previous study by our research team had indicated that differences in physical and chemical characteristics between the two rivers could affect the relationships between tissue metal and MT concentrations.¹⁷ Consequently, we established relationships between tissue metals and metallothionein (MT) for each river independently. Consistent with a somewhat lower Cd contamination than that of the study lakes, bioaccumulated Cd reached much lower levels in the sentinel organisms collected in the two rivers than in the Rouyn-Noranda lakes (compare the X-axis scales in Fig. 2a and 2c, and in Fig. 2b and 2d). Significant relationships were observed between Cd and MT concentrations in the whole body of H. limbata in the Allard River. No significant relationship was observed in the Colombière River, but the range of accumulated Cd concentrations was much smaller (<50 nmol g⁻¹) (Fig. 2c). For P. grandis, the same trends were observed in the two rivers, but the absence of bivalves at four sites in the Colombière River and the resulting low number of observations influenced the level of significance (Fig. 2d). Nevertheless, we note that the slope of the [MT] vs. [Cd] relationship was three times higher for the molluscs collected in the Colombière River than that for the bivalves collected from the Allard River. Because of the small number of observations in the Colombière River (n = 4), we cannot test the statistical significance of this result.

Within the Allard River system, where the number of sites with co-existing sentinel species was reasonably high, differences were observed between the mayfly larva and the bivalve. The Cd–MT relationship for P. grandis was less strong than the Cd–MT relationship established for H. limbata (R² = 0.39 vs. 0.79). The increment of MT per unit of accumulated Cd was significantly higher (F = 6.30, P = 0.024) for H. limbata than for P. grandis. According to our regression model (Table 6), the basal concentration of MT, when negligible Cd is present in the tissues in both species, was significantly higher (F = 23.2, P < 0.001) for H. limbata (34.9 nmol sites g⁻¹ dry wt) than for P. grandis (15.6 nmol sites g⁻¹ dry wt).

3.5. Comparison of the lacustrine and riverine Cd–MT relationships

Similar methodologies were used for the Cd and MT analyses in the river and lake studies, thus allowing us to compare the Cd–MT relationships in both environments. The model developed for lacustrine H. limbata shows a lower slope (0.15 vs. 0.78) and intercept (11.5 vs. 34.9) compared to the river model (Fig. 2, Table 6). These significant differences in the slope (F = 25.35, P < 0.001) and in the intercept (F = 17.3, P < 0.001) imply that at low concentrations of bioaccumulated Cd in the tissues, more MT is synthesized by the animals in the Allard River than in the lakes.

Because differences in the ranges of Cd and MT concentrations in P. grandis were too large between the Allard River stations and the Rouyn-Noranda lake stations (Table 6), it was not feasible to test differences between the models using ANCOVA. Although the slope of the river model was 2 times higher (Fig. 2, Table 6), we cannot extrapolate and compare these results with those from the lake model, since we do not know if the relationship derived from the Allard River specimens remains linear at higher Cd concentrations. However, according to the slope of the river model, the increase in MT per unit of accumulated Cd is more important in the Allard River for tissue Cd concentrations <100 nmol g⁻¹ dry weight than for the lake model.

3.6. Influence of environmental variables on Cd accumulation and MT synthesis in the sentinel species

Lakes. No significant relationships were observed between Cd concentrations in P. grandis and H. limbata on one hand, and Cd concentrations in sediments on the other (P > 0.05, results not shown). On the other hand, bivalve gill [Cd] did increase with increases in dissolved [Cd] (Table 7). We could detect a marginal influence of dissolved Ca and dissolved Mn, but no additional effect of the hydrogen ion on Cd bioaccumulation after Ca and Mn has been accounted for (Table 7). In contrast to P. grandis, Cd concentrations in H. limbata nymphs were not correlated with dissolved Cd alone. However, dissolved Mn exerted an apparent antagonistic effect on Cd accumulation in this insect, explaining nearly 80% of the variations in [MT] produced in response to Cd accumulation (Table 7, Fig. 3a).

Table 7 Regression models for Cd in H. limbata and P. grandis including metals in the animal as the dependent variable, and physico-chemical and metal characteristics in the environment as independent variables. The cumulative coefficient of determination (R²) of the model as well as the partial R² of the regression coefficients, standard error of the coefficient (SE), partial t-value (t), and probability of t-test (P(t)) are also listed

Model	Equation terms	R ^2	SE	T	P(t)
Log CdHL River	−8.84		2.374	−3.725	0.002
	−0.43 Log Mnd	0.79	0.074	−5.822	<0.001
	+2.75 Log Mg	0.85	0.620	4.435	0.001
	−1.48 Log FA	0.91	0.466	−3.183	0.007
Log CdHL R-N lakes	No model, too many missing data. Not significant, but negative correlations (P > 0.05) between CdHL and dissolved Mn (−0.75) and DOC (−0.76), and positive correlation with dissolved Cu (0.76).
Log MTHL R-N lakes	+1.02		0.112	9.157	<0.001
	−0.383 Log Mnd	0.87	0.097	−3.952	0.011
Log CdPG River	−0.89		0.471	−1.887	0.096
	+1.75 Log Cd	0.36	0.351	4.986	0.001
	+0.71 Log Cu	0.62	0.224	3.174	0.013
Log CdPG R-N lakes	+3.814		0.158	24.154	<0.001
	+0.890 Log Cdd	0.82	0.121	7.371	<0.001
	−0.58 Log Cad	0.89	0.197	−2.926	0.017
	−0.24 Log Mnd	0.93	0.106	−2.316	0.046

---

Concentration ranges for the variables found in the regression models (2001 sampling year):
Variable/unit	Concentration range in rivers	Concentration range in lakes
a Note: CdHL = Cd in H. limbata; MTHL = MT in H. limbata; CdPG = Cd in P. grandis gills; Mnd = dissolved Mn; Cad = dissolved Ca; Cdd = dissolved Cd; Mg = Mg in sediment; FA = fulvic acid; Cd = Cd in sediment; Cu = Cu in sediment.
Water
Mnd/µM	0.02–4.6	0.008–2.37
Cad/µM	110–7660	47–331
Cdd/nM	—	0.018–0.72
FA/mg L^−1	11–51	—
Sediment
Mg/mmol kg^−1	333–765	—
Cd/µmol kg^−1	1.5–18.9	—
Cu/mmol kg^−1	0.17–3.6	—

---

Fig. 3 (a) Partial correlation between MT in H. limbata and dissolved Mn measured in lakes around Rouyn-Noranda, holding dissolved Cd constant; X₁ = dissolved Cd, X₂ = dissolved Mn, X₃ = MT in H. limbata. Two-letter codes refer to lake names—BO: Bousquet, CA: Caron, DU: Dufay, HE: Héva, JO: Joannès, OP: Opasatica and VA: Vaudray. The partial correlation between Cd in H. limbata and dissolved Mn, holding dissolved Cd constant, is negative and has an R² of 0.42 (0.10 > P > 0.05). (b) Exponential relationship between Cd in H. limbata and dissolved Mn in water measured in the two rivers.

Rivers. The physico-chemical characteristics of the two rivers exhibited marked differences (Table 2) and we hypothesized that these differences could affect Cd uptake by the animals and thus explain the absence of relationships between tissue Cd and MT in the sentinel organisms collected from the Colombière River. Considering both rivers, and similar to the results obtained for H. limbata populations in lakes, variations in the concentration of total dissolved manganese, [Mn]_d, explained up to 80% of the variation in Cd accumulated in the body of H. limbata (Table 7), with dissolved Mn having an apparent negative effect on the accumulation of Cd (Fig. 3b). In addition, magnesium (Mg) influenced Cd accumulation positively, whereas fulvic acid (FA) had a negative influence on accumulation of Cd. These three variables together explained 91% of the variation of Cd concentrations in H. limbata. The Cd concentration in P. grandis was positively related to the concentrations of Cd and Cu in the sediment (Table 7). These variables explained up to 62% of the variability of Cd accumulation in the floater mussel. The concentration of the Cd in the sediment was the most important variable, explaining 36% of the Cd variation in P. grandis.

4. Discussion

4.1. Influence of environmental conditions on Cd accumulation and MT concentrations

Environmental factors (H⁺, Ca²⁺, temperature, dissolved organic carbon, etc.) may influence exposure–response relationships for Cd both mechanistically and physiologically. First, the uptake of either Cd or other metals (for a given [M²⁺]) can be strongly affected by the presence of competitors (H⁺, Ca²⁺), with consequent effects on MT induction. Secondly, environmental factors can modify the relative importance of metal bio-uptake pathways (diet-borne vs. waterborne metals), which can influence kinetic processes (e.g., uptake rates) involving metals within organisms. Furthermore, environmental characteristics can also affect the physiology of organisms (e.g., respiration rate, filtering rate, feeding rate, etc.), which can in turn have repercussions on basal levels of MT and on the organism sensitivity to metal inputs. These various processes are explored in the following discussion.

4.1.1. Influence of environmental variables on Cd accumulation in H. limbata. Many factors have been identified as having an impact on the accumulation of Cd by aquatic invertebrates.^3,23 Our multiple regression model, applied to the data for both rivers, identified three variables that together explained 91% of the variability in Cd concentrations in H. limbata. Dissolved manganese (Mn), negatively related to Cd, was the most important factor (79%). For lake data, partial correlation analysis also unveiled the antagonistic effect of dissolved Mn on Cd accumulation by mayfly larva.

Luoma²⁴ noted that the dependence of metal uptake on reactions with membrane transport proteins implies that major cations should influence metal availability. In the present study, we hypothesize that dissolved Mn²⁺ could act as an antagonist to Cd²⁺ accumulation, as has been observed in many invertebrates with other cations, such as Ca²⁺ (ref. 12,25) and the H⁺ ion.²⁶ These latter studies have shown that Cd tends to increase in the animal as the concentrations of calcium or hydrogen ions decrease. Alternatively, given that Mn is an essential element, the increase in Cd accumulation at low ambient Mn concentrations may reflect an acclimation to low Mn concentrations. Such an acclimation response would be expected to lead to a higher capacity for Mn uptake; if Cd is taken up via Mn transport sites, this higher uptake capacity for Mn would lead to an increased uptake of Cd. To our knowledge, the influence of Mn on Cd uptake by aquatic animals has not been reported in the literature for mayflies. However, the peer-reviewed literature does provide evidence of the protective effect of Mn against Cd bioaccumulation and toxicity in living organisms (Table 8). In particular, it was demonstrated that Cd²⁺ can enter cells of a microorganism via its cellular Mn²⁺ transport system, a situation conducive to mutually competitive inhibition between these two cations at cellular uptake sites.²⁷

Table 8 Published evidence of the influence of Mn on Cd bioaccumulation and toxicity in living organisms

Organism type	Species	Effect/observation	Reference
Bacteria	Staphylococcus sp.	Cd^2+ can enter cells via the cellular Mn^2+ transport system, a normal nutritionally required cation transport system, Mn being an essential metal.	Perry and Silver^27
Isolated hepatocytes	—	Cell injury and lipid peroxidation due to Cd were consistently reduced by Mn in the exposure medium.	Stacey and Klaassen^50
Plant	Zea mays—maize seedlings	Mn alleviated the toxic effect of Cd on root growth.	Pal'ove-Balang et al.^51
Plant	Hordeum vulgare—barley	Illustrates that Cd can be a competitive inhibitor of Mn internal transport. The authors concluded that maintaining higher Mn concentration in plant organs may be beneficial to improve Cd tolerance in barley.	Wu et al.^52
Freshwater crustacean	Hyalella azteca—amphipod	Using metal mixtures, a significant inhibition of Cd uptake by Mn was demonstrated.	Norwood et al.^53
Mammal	Rattus norvegicus—laboratory rat	Gastrointestinal absorption of Cd was decreased by Mn supplied in drinking water.	Sarhan et al.^54

---

Nymphs of H. limbata appear to be responsible for much sediment disturbance by littoral insects in lakes.²⁸ Despite this observation, this benthic species takes up most of its metal from the water column above the sediment compartment, mostly because of its irrigation behaviour which circulates large volumes of overlying water within its burrow.²⁹ In support of this interpretation, Warren et al.³⁰ determined experimentally in a field setting that H. limbata larva obtained 97% of their Cd from water above the sediments. In addition, colonization and burrowing behaviour were shown not to be affected by sediment contamination by Cd,^30,31 providing further support for the importance of the aqueous Cd bio-uptake pathway for this species.

4.1.2. Influence of environmental factors on Cd accumulation in P. grandis. Cadmium concentrations in bivalves living in lakes of the Rouyn-Noranda area varied as a function of dissolved Cd and not of sedimentary Cd. Numerous studies have established positive relationships between Cd concentrations in aquatic species collected in the natural environment and either free Cd²⁺ (ref. 12,13,32) or total dissolved Cd in water,^33,34 but relationships with total Cd concentrations in the sediment are less prevalent. For suspension feeders such as P. grandis, uptake of metal can occur from water and food.³⁵ Sediments normally act as a sink and reservoir for metals but, under certain circumstances, might serve as a source of metals to filter-feeding organisms via redistribution of sediments into the water column.³⁶

In contrast with the lacustrine bivalves, the concentration of Cd in the sediment was the first and most important factor related positively to Cd concentrations in the bivalves collected from the Allard and Colombière rivers. We speculate that individual bivalves in rivers were more often exposed to Cd-laden sediment particles than were those living in lakes. Sediment resuspension events are typically more frequent in rivers than in lakes. The Allard River presents shallow littoral waters, easily reaching 75 m in width in many river reaches, which are frequently mixed by the wind. We observed increases in velocity flow and suspended sediment load in Colombière River following an important rain event. Conversely, although some of our study lakes are subject to resuspension events (the large lakes Opasatica and Beauchastel, with appreciable fetches), quieter conditions usually prevailed in the littoral environments of the Rouyn-Noranda lakes.

The apparent positive effect of the Cu concentration in sediment on Cd bioaccumulation in river bivalves is difficult to explain; to our knowledge, no additive effect of this metal on Cd accumulation has been reported in the literature. On the contrary, Stewart³⁷ observed in a limnocorral experiment that uptake of dissolved Cd in P. grandis was reduced in the presence of a mixture of several dissolved metals (Cu, Zn, Pb and Ni). Metal mixtures may influence metal bioavailability in various ways, depending on the individual metal concentrations, the relative binding strengths of the metals for various substrates, and the physiological roles of the metals. In her experiment, Stewart³⁷ indicated that the presence of the metal mixture resulted in an increase in the residence times for Cd in the water column. Metals are thought to sorb predominantly to independent binding sites, but some sites may be mutually shared,³⁸ leading to competition among metals for these shared sites. We noted a strong negative relationship between the concentrations of Cd and Cu in the sediments (r = −0.82), in spite of which Cu in the sediment was positively related to Cd in P. grandis. The raw data indicate that the negative relationship is mainly due to the Colombière stations, which show a strong inverse relationship between the two metals. The positive or negative effect of Cu on Cd accumulation in benthic organisms will have to be studied over a wider natural gradient, or experimentally, to tease apart the metal–metal interactions and their effect on Cd uptake in molluscs.

4.2. Comparison of Cd–MT models between the two invertebrates

As has been reported for many indigenous species (molluscs,^11,12 zooplankton,¹³ fish^14,39,40), we observed a positive relationship between tissue Cd and MT for the bivalve P. grandis in both rivers and lakes, and for the mayfly H. limbata in the Allard River and lakes (the range of Cd concentrations in the mayflies from the Colombière River was too narrow to establish a relationship between Cd and MT). In the Allard River the Cd–MT relationship was stronger for the mayfly, which seems to maintain higher concentrations of MT per unit of accumulated Cd than do the molluscs. Furthermore, both species often produced very different MT levels in response to Cd exposure in the Rouyn-Noranda lakes where they co-existed.

Steady-state metallothionein concentrations in chronically exposed aquatic animals reflect the rates of MT synthesis and degradation. Indeed, differences in metallothionein turnover rates are one of the major causes of inter-specific differences in steady-state MT concentrations in invertebrates, and they have also been invoked as an explanation of why some invertebrate tissues do not show changes in MT concentration under metal exposure conditions expected to induce the synthesis of MT.⁴¹ The synthesis of MT can be induced without resulting in a concomitant increase in MT concentration in the tissues of some organisms, because of a corresponding increase in rate of MT breakdown.

In the present case, different tissues have been used for measuring Cd and metallothionein, gills for the bivalve and the entire body for the mayfly. However, this does not detract from the pertinence of comparing these species because the bivalve gill contributes a major proportion of the total Cd burden in whole animals (∼40%³²) and because MT and Cd levels in the bivalve gill are closely related to those in the whole organism (r > 0.85¹¹).

Differences between these co-existing invertebrates suggest that although they live in the same environment they accumulate Cd differently. Our results suggest that these species-to-species differences in Cd bioavailability may be related to competition between specific dissolved cations and Cd ions at the biological uptake sites of these organisms. Other possible explanations are related to the extent to which these invertebrates accumulate Cd from different environmental compartments (overlying water or suspended sediments; sediment pore water or settled particles), and to the fact that they do not necessarily live in exactly the same zones. For example, in the Allard River H. limbata was found in the littoral zone, whereas P. grandis was collected in deeper zones. On the other hand, both species were sampled in the same habitat in the Colombière River.

Steady-state MT concentrations, expressed per unit accumulated Cd, were about three times higher in molluscs from the Colombière River than in individuals collected in the Allard River (Fig. 2d). In the Colombière River, the molluscs were found at approximately 1.5 m depth, whereas those from the Allard River lived at approximately 4 m depth. During summer time, the ambient temperature can easily vary from 14 to 22 °C between these two depths. Temperature is recognized to significantly affect all biological processes, and these temperature differences may have contributed to the different MT/Cd ratios.

4.3. Comparison of Cd–MT models between lotic and lentic environments

Although impacts of metals on the community structure of benthic invertebrates have been widely studied in rivers,^23,36 we were unable to find any field study that had compared metal accumulation and detoxification for the same benthic species living in lotic and lentic environments. Physical and chemical conditions in lotic environments are typically highly variable. For example, the concentrations and forms of metals may fluctuate strongly in lower order streams and rivers, due to many factors (heavy rainfall, spring snowmelt, sporadic input of metals, fluctuation of the concentration of organic ligands, changing temperature, pH, etc.).¹⁶ Benthic organisms inhabiting such environments have to adapt to these unstable conditions and may in some cases be exposed to transitory episodes of elevated metal concentrations. We wondered whether the ability of benthic organisms living in such environments to detoxify metals might be different from that of organisms inhabiting calm lacustrine sediments. For example, after the different physiological needs of the two environments have been met, the amount of energy available for metal detoxification might well differ.

4.3.1 H. limbata. Marked differences were observed between our riverine and lacustrine Cd–MT models for the mayfly larvae (Fig. 4a). The much greater slope of the Allard River model indicates that, for low body concentrations of Cd, the animals living in the river environment synthesize and retain more MT per unit accumulated Cd. In addition, the river model revealed a linear relationship at low tissue Cd concentrations (<100 nmol Cd g⁻¹ dry weight), whereas the lake model did not detect such a linear trend.

Fig. 4 Comparison of the relationships between bioaccumulated cadmium and metallothionein for the two sentinel species collected along contamination gradients between the Allard River (this study) and different lake studies: (a) H. limbata and (b) P. grandis.

Many factors could in principle explain the differences between the river and lake models, such as the size or age of the organisms, the dynamics of the systems (lotic vs. lentic), the physico-chemical environment, and the influence of other accumulated metals on MT synthesis. The first of these factors can be discounted: the size of the H. limbata larvae is not an important variable here, since both studies used quite similar size ranges (1.5–2 cm). However, the different hydrodynamics of the two systems may be important, rivers being more dynamic systems than lakes.

Hydrological events are thought to function as disturbances for invertebrates.⁴² Giberson and Cobb⁴³ showed that mayfly communities in two different rivers could be influenced by the intensity and frequency of floods. In such dynamic environments, organisms are likely more stressed than in lakes, where quieter conditions prevail for benthic strata. Lotic species are assumed to be adapted to frequent (relative to organism life span) hydrological events.^42,44 We speculate that their response to these frequent stresses might include having higher basal levels of MT synthesis. In literature reviews on metallothioneins, Kägi⁴⁵ and Roesijadi⁴⁶ mention that MT synthesis can be influenced by metabolic states other than those resulting from metal exposure.

Metal dynamics are also different in lotic and lentic environments—river sediments are inherently unstable, being subject to frequent sedimentation and resuspension events, whereas lake sediments tend to be much more stable, with well-defined vertical redox profiles. However, it is unclear how these differences in habitat could affect the Cd–MT relationship in H. limbata.

Finally, metals other than Cd can contribute to the induction of MT synthesis.⁴⁷ Basal levels of MT are considered to be involved in the regulation of essential metals, particularly Cu and Zn, with higher MT concentrations often being associated with exposure to and accumulation of other (non-essential) metals. In the present case, positive bivariate relationships were established between MT tissue concentrations and accumulated Ag, Ni and Zn,¹⁷ but in multiple regression analysis only Zn was retained as one of the predictors of MT concentrations (inclusion of Zn as the final predictor increased the r² value from 0.85 to 0.89). This weak influence of the other metals cannot explain the markedly different Cd–MT relationships observed in the river and lake environments.

4.3.2. P. grandis. In the Allard River bivalves, the accumulated gill Cd concentrations and resulting MT levels cover a much narrower range than that observed in bivalves from the Rouyn-Noranda lakes (Fig. 4b). This narrow range of Cd concentrations limits our ability to compare Cd–MT relationships in lentic and lotic environments. Nevertheless, and in contrast to the case with the mayfly larvae, the Cd–MT model established for P. grandis in the Allard River agrees reasonably well with our earlier results, obtained for this species in lacustrine habitats, which demonstrated that gill MT concentrations were positively related to gill Cd concentrations.^11,12,48,49

5. Conclusion

The present study established that even at low ambient Cd concentrations, metallothionein is synthesized in our two sentinel species. For both species, differences in the basal levels of MT are observed in organisms collected from uncontaminated lake or river habitats. For the mayflies, this difference between lake and river environments is even more evident when Cd–MT relationships are considered along a metal contamination gradient—mayflies living in rivers maintained much higher concentrations of MT per unit of accumulated Cd than did their counterparts in lakes. Environmental conditions in rivers are much less stable than in lakes, and we speculate that in some cases the acclimation or adaptation of indigenous lotic species to fluctuating river environments may enhance their ability to deal with other (metal-related) stresses.

This study reinforces the idea that an animal’s responses to chronic metal exposure, such as metal accumulation and MT synthesis, are linked to the surface chemistry reactions that prevail at the biological membrane that constitutes the interface between the organism and its surrounding media. While this idea has gained acceptance in aquatic ecotoxicology, field results relevant to the subject remain scarce. In this regard, the identification of dissolved Mn as a factor influencing Cd uptake by H. limbata populations in lakes and rivers suggests further experimental studies on the effect of this metal on Cd internalization.

Our results support the contention that one cannot extrapolate conclusions drawn from the use of a single sentinel species to a larger set of freshwater invertebrates. In this respect, the mayfly larvae H. limbata and the bivalve P. grandis appear to be promising biomonitors of metal contamination.

Acknowledgements

This work was supported by a Cooperative Research and Development grant from the Canadian Natural Sciences and Engineering Research Council (NSERC) to PGCC, with the financial support of COREM and in-kind support from the Centre d'expertise en analyse environnementale du Québec (CEAEQ). We are grateful to Marie-Hélène Michaud, Marie-Pier Cloutier and Annick Michaud for field sampling. We also acknowledge Noranda Inc. (Matagami) and the Université du Québec en Abitibi Témiscamingue (Val d'Or), who allowed us to use their installations for laboratory analyses. Special thanks are due to the technicians from INRS-ETE, CEAEQ and COREM who helped with the toxicological and physico-chemical analyses. Yves Parisot and Dacheng Wang coordinated the organism collections in 1994. A draft version of the manuscript benefited from the constructive criticism of Robert Prairie, and two anonymous referees also helped to improve the final paper. PGCC is supported by the Canada Research Chair programme.

References

1. J. O. Nriagu, Global Planet. Change, 1990, 2, 113–120.
2. P. G. C. Campbell and A. Tessier, in Heavy Metals in the Environment, ed. J.-P. Vernet, Elsevier Sci. Publ., Amsterdam, The Netherlands, 1991, pp. 161–173.
3. L. Hare, Crit. Rev. Toxicol., 1992, 22, 327–369.
4. J. M. Pacyna, M. T. Scholtz and Y.-F. A. Li, Environ. Rev., 1995, 3, 145–159.
5. C. D. Wren, S. Harris and N. Harttrup, in Handbook of Ecotoxicology, ed. D. J. Hoffman, B.A. Rattner, G.A. Burton and E-mail: J. Cairns, Lewis Publishers, Boca Raton, FL, USA, 1995, pp. 392–423.
6. Z. Sidoumou, M. Gnassia-Barelli and M. Romeo, Bull. Environ. Contam. Toxicol., 1997, 58, 318–325.
7. O. Perceval, Y. Couillard, B. Pinel-Alloul, A. Giguère and P. G. C. Campbell, Aquat. Toxicol., 2004, 69, 327–345.
8. A. Viarengo, B. Burlando, F. Dondero, A. Marro and R. Fabbri, Biomarkers, 1999, 4, 455–466.
9. T. S. Galloway, R. J. Brown, M. A. Browne, A. Dissanayake, D. Lowe, M. B. Jones and M. H. Depledge, Environ. Sci. Technol., 2004, 38, 1723–1731.
10. O. Perceval, Y. Couillard, B. Pinel-Alloul and P. G. C. Campbell, Aquat. Toxicol., 2006, 79, 87–98.
11. Y. Couillard, P. G. C. Campbell and A. Tessier, Limnol. Oceanogr., 1993, 38, 299–313.
12. O. Perceval, B. Pinel-Alloul, G. Méthot, Y. Couillard, A. Giguère, P. G. C. Campbell and L. Hare, Environ. Pollut., 2002, 118, 5–17.
13. M.-N. Croteau, L. Hare, P. G. C. Campbell and Y. Couillard, Environ. Toxicol. Chem., 2002, 21, 737–741.
14. J.-S. Laflamme, Y. Couillard, P. G. C. Campbell and A. Hontela, Can. J. Fish. Aquat. Sci., 2000, 57, 1692–1700.
15. R. G. Wetzel, Limnology—Lake and River Ecosystems, Academic Press, San Diego, CA, USA, 2001.
16. D. A. Nimick, C. H. Gammons, T. E. Cleasby, J. P. Madison, D. Skaar and C. M. Brick, Water Resour. Res., 2003, 39, 1247.
17. S. Masson, P. G. C. Campbell, L. Hare, C. Olsen, L. Martel, B. Pinel-Alloul, G. Méthot and A. Hontela, Rev. Sci. Eau, 2005, 18, 143–160.
18. M. R. Laflèche, C. Dupuy and J. Dostal, Precambrian Res., 1991, 52, 71–96.
19. M. D. Dutton, M. Stephenson and J. F. Klaverkamp, Environ. Toxicol. Chem., 1993, 12, 1193–1202.
20. C. Fortin and P. G. C. Campbell, Int. J. Environ. Anal. Chem., 1998, 72, 173–194.
21. D. C. Montgomery and E. A. Peck, Introduction to Linear Regression Analysis, J.Wiley and Sons Ltd., New York, NY, USA, 1982.
22. O. Perceval, Y. Couillard, B. Pinel-Alloul, E. Bonneris and P. G. C. Campbell, Sci. Total Environ., 2006, 369, 403–418.
23. S. N. Luoma and J. L. Carter, in Metal Ecotoxicology—Concepts and Applications, ed. M. C. Newman and A. W. McIntosh, Lewis Publishers Ltd., Chelsea, MI, USA, 1991, pp. 261–300.
24. S. N. Luoma, Sci. Total Environ., 1983, 28, 1–22.
25. S. J. Markich and R. A. Jeffree, Aquat. Toxicol., 1994, 29, 257–290.
26. M.-N. Croteau, L. Hare and A. Tessier, Environ. Sci. Technol., 2002, 36, 3079–3082.
27. R. D. Perry and S. Silver, J. Bacteriol., 1982, 150, 973–976.
28. P. Charbonneau and L. Hare, J. North Am. Benthol. Soc., 1998, 17, 239–249.
29. C. Gallon, L. Hare and A. Tessier, J. North Am. Benthol. Soc., 2008, 27, 570–580.
30. L. A. Warren, A. Tessier and L. Hare, Limnol. Oceanogr., 1998, 43, 1442–1454.
31. A. Gosselin and L. Hare, Environ. Toxicol. Chem., 2004, 23, 383–387.
32. A. Tessier, Y. Couillard, P. G. C. Campbell and J.-C. Auclair, Limnol. Oceanogr., 1993, 38, 1–17.
33. M. Stephenson and G. L. Mackie, Can. J. Fish. Aquat. Sci., 1988, 45, 1705–1710.
34. S. A. Dressing, R. P. Maas and C. M. Weiss, Bull. Environ. Contam. Toxicol., 1982, 28, 172–180.
35. W. X. Wang and N. S. Fisher, Sci. Total Environ., 1999, 237/238, 459–472.
36. W. H. Clements, in Metal Ecotoxicology—Concepts and Applications, ed. M. C. Newman and A. W. McIntosh, Lewis Publishers Ltd., Chelsea, MI, USA, 1991, pp. 363–391.
37. A. R. Stewart, Can. J. Fish. Aquat. Sci., 1999, 56, 467–478.
38. L. Sigg, M. Sturm and D. Kistler, Limnol. Oceanogr., 1987, 32, 112–130.
39. C. Hogstrand, G. Lithner and C. Haux, Pharmacol. Toxicol. (Copenhagen), 1991, 68, 492–501.
40. P. A. Olsvik, K. Hindar, K. E. Zachariassen and R. A. Andersen, Biomarkers, 2001, 6, 274–288.
41. T. Y. T. Ng, P. S. Rainbow, C. Amiard-Triquet, J. C. Amiard and W.-X. Wang, Aquat. Toxicol., 2008, 89, 47–54.
42. V. H. Resh, A. V. Brown, A. P. Covich, M. E. Gurtz, H. W. Li, G. W. Minshall, S. R. Reice, A. L. Sheldon, J. B. Wallace and R. C. Wissmar, J. North Am. Benthol. Soc., 1988, 7, 433–455.
43. D. J. Giberson and D. G. Cobb, in Current Directions in Research on Ephemeroptera, ed. L. D. Corkum and J. J. H. Ciborowski, Canadian Scholars' Press, Toronto, ON, Canada, 1995, pp. 237–252.
44. N. L. Poff, J. North Am. Benthol. Soc., 1992, 11, 86–92.
45. J. H. R. Kägi, in Methods in Enzymology, Academic Press Inc., San Diego, CA, 1991, pp. 613–626.
46. G. Roesijadi, Aquat. Toxicol., 1992, 22, 81–114.
47. J. H. R. Kägi and Y. Kojima, Experientia Suppl., 1987, 52, 25–61.
48. A. Tessier, P. G. C. Campbell, J.-C. Auclair and M. Bisson, Can. J. Fish. Aquat. Sci., 1984, 41, 1463–1472.
49. D. Wang, Y. Couillard, P. G. C. Campbell and P. Jolicoeur, Can. J. Fish. Aquat. Sci., 1999, 56, 774–784.
50. N. H. Stacey and C. D. Klaassen, J. Toxicol. Environ. Health, 1981, 7, 149–158.
51. P. Pal'ove-Balang, A. Kisová, J. Pavlovkin and I. Mistrík, Plant Soil Environ., 2006, 52, 143–149.
52. F. Wu, J. Dong, Y. Cai, F. Chen and G. Zhang, Sci. Total Environ., 2007, 385, 228–234.
53. W. P. Norwood, U. Borgmann and D. G. Dixon, Aquat. Toxicol., 2007, 84, 255–267.
54. M. J. Sarhan, H. Roels, R. Lauwerys, H. Reyners and E. Gianfelici de Reyners, J. Appl. Toxicol., 1986, 6, 313–316.

---