Organic matter and a synthesized iron oxide, ferrihydrite, were added singly and in ... partitioning and bioavailability, whereas the synthetic ferrihydrite bound.
197
HydrobiohJ,q, ia 304: 197-207, 1995. @ 1995 Kluwer Academic Publishers. Printed in Belgium.
Geochemical partitioning and bioavailability of copper to aquatic plants in an artificial oxide-organic sediment L. A. Warren I, E M. Outridge 2 & A. R Z i m m e r m a n 1 IDepartment ofZoolog3; UniversiO, of Toronto, Toronto, Ontario M5S 1A 1, Canada 2Environmental and Resource Studies Program, Trent UniversiO; Peterborough, Ontario K9J 7B8, Canada Received 4 August 1993; in revised form 24 February 1994; accepted 2I April 1994
Key words." copper, bioavailability, toxicity, sediments, partitioning, macrophytes
Abstract This study investigated the effects of competition between binding substrates (organic matter and iron oxide) and between metals (cadmium and copper), on the partitioning of sedimentary copper and its subsequent bioavailability to an aquatic plant. Organic matter and a synthesized iron oxide, ferrihydrite, were added singly and in combination to a series of sand sediments, which were then dosed with environmentally realistic concentrations of cadmium and copper and planted with rice, O0'za sativa. Organic matter controlled copper partitioning and bioavailability, whereas the synthetic ferrihydrite bound negligible amounts of either metal, even in the absence of organic matter. As organic matter concentrations increased, operationally-defined leachable copper decreased, organic-associated copper increased and the survival of rice plants improved in an approximately linear fashion. At a nominal starting copper concentration of 5.8 F~g g dry wt -I , plant survival after four weeks averaged 0-8% in sediments without organic matter, 25% in a sediment containing 0.18% organic matter and 58% in a sediment containing 0.36% organic matter. These results suggest that organic-associated forms of copper are unavailable to plants, and that the operational definition of 'leachable' copper (extracted with dilute ammonium acetate) adequately represents the species of copper that is (are) available to plants. Our study using a well-characterized artificial sediment supports the copper fractionation patterns and correlations between copper partitioning and bioavailability reported from the heterogeneous, poorly characterized sediments of natural lake and river sediments. Introduction Aquatic systems constitute the largest sink for heavy metals mobilized by anthropogenic activities (Nriagu & Pacyna 1988). Currently, a prevalent paradigm for metal toxicology and biogeochemistry is that geochemical partitioning between solid (sediment) and solution phases is the key factor controlling the availability, and hence toxicity, of metals to biota in aquatic environments (Campbell et al., 1988; Luoma, 1989). Furthex'more, the partitioning of metals in sediments (i.e. their distribution among different geochemical substrates), is influenced by environmental factors such as pH and Eh, the abundance of organic and inorganic materials (clays and hydrous metal oxides), and the concentrations of competing cations and solution
ligands (Cahnano et al., 1988; Fulghum et al., 1988; Tessier et al., 1985). Beyond these broad generalizations, however, a number of specific questions remain unresolved. One issue concerns the relative sorptive affinities of particulate organic matter and metal oxides for different trace metals. For example, in oxidized, organic-rich natural sediments, copper is usually most strongly associated with organic matter (Warren & Zimmerman, 1993; Chen et at., 1989; Tessier et al., 1980), although this association may be less important in sediments that are relatively high in iron oxides and low in organic matter (Luoma & Bryan, 1981; Luoma, 1986). In contrast, metals such as cadmium tend to be associated only weakly with organic matter, favouring the carbonate and 'exchangeable' fractions (Warren & Zimmerman,
198 1993; Pardo et al., 1990; Tessier et al., 1980), or metal oxides (Reuther et al., 1981 ). Another primary area of interest is the nature of the relationship between metal partitioning and bioavailability. Key questions include the identity of the 'bioavailable" fo,ms of metals, how they are quantitatively influenced by sediment constituents, and the most appropriate means of measuring these forms (Luoma, 1989: Campbell & Tessier, 1989). One approach used to address these questions has been to test for correlations between various operationallydefined metal fractions in sediments and either the metal concentrations in or toxicity to sediment-associated biota such as macrophytes, bivalves and other benthic invertebrates (e.g. Campbell et al., 1985; Thompson et al., 1984: Luoma & Bryan, 1978). A positive correlation between the concentrations in a given sediment fraction and ill biota would suggest that the fraction in question is "bioavailable'. Most studies of this type have been carried out on natural systems, analysing sediments and biota collected from the field. In contrast, a second approach, which has produced most of our current theoretical knowledge regarding metal sorption onto solid phases, has largely involved laboratory model systems using well-defined "artificial' particles, predominantly metal oxides (Honeyman & Santschi, 1988). Because of the complexity of metal-particle interactions, two common features of this type of study are single metal-substrate experimental designs, and the use of syntheticallyderived goethite (a-FeOOH), a crystalline Fe oxide (e.g. Ankomah, 1992; Barrow et al., 1989). In natural sediments, however, other oxides such as ferrihydrite and lepidocrocite have been identified (Fortin et al., 1993: Tessier et al., 1993), which are more amorphous than goethite and consequently may have quite different sorption characteristics (Schwertmann & Cornell, 1991). Natural systems may also contain multiple binding substrates in potential competition for metals. Thus, the degree to which the laboratory-based literature has direct relevance to questions of metal bioavailability in natural systems remains unclear. Relatively few studies of metal-particulate associations in natural systems (e.g. Tessier et al., 1993, 1989, 1985, 1980; Warren & Zimmerman, 1994; Bendell-Young & Harvey, 1992) are available for comparison with laboratory-derived literature. This study aims to link these two different approaches, by employing a well defined artificial sedinaent that has environmentally relevant characteristics, namely, competing binding substrates and metal
cations, and realistic concentrations of metals. Tile objective of our laboratory-based study is two-fold. First, we examine the geochemical partitioning of copper in a series of sandy sediments containing varying proportions of a synthesized iron oxide (ferrihydrite) and organic matter (peat). The effect of another cation, cadmium, on the partitioning of copper was also investigated. Second, we relate tile observed differences in partitioning to the availability of copper to an aquatic macrophyte, O0'za sativa, that was grown in the sediments. Specifically, we test the hypothesis that increasing amounts of organic matter in sediments will lower the bioavailability of copper to the macrophyte.
Materials and methods E x p e r i m e n t a l design
The experimental design was an unbalanced two-way factorial analysis of variance (Table 1), set up as a sand culture experiment (Hewitt, 1966). Two different metal treatments, Cu alone or a mixture of Cu + Cd, were added to each of 4 sediment types: the sand sediment, 700 g of fine silica sand (SAND): oxide sediment, 2.5 g of ferrihydrite added to 700 g sand (OX); organic sediment, 2.5 g of peat added to 700 g sand (OM); and mixed sediment, 1.25 g each of ferrihydrite and peat added to 700 g sand (MIX). To account for any differences in partitioning due to the presence of plants, a sediment control (Cd +Cu treatment with no plants) was included for each sediment type. Similarly, to account for any differences in plant growth due to sediment type, a plant control (no metal treatment) was also included for each sediment type. Ferrihydrite binding capacity was assumed to be 0.005 tool metal per tool Fe at a pH of 6.5 to 7,5 (Dzombak & Morel, [990), so that 2.5 g of ferrihydrite should theoretically be able to complex 2.6 × 10 - 4 moles of metal ions. An initial experiment was carried out with the peat to determine its binding capacity. A 200 rnl volume of 5 mg 1- I Cu was added to a series of beakers containing fiom 0.25 to 3.0 g of peat, and porewater Cu concentrations were measured after one week. This experiment revealed that 2.5 g of peat was able to complex 2 x 10 - 4 moles of Cu, roughly equivalent to that for the same mass of Fe oxide. Both the Cu and the Cu + Cd treatments employed the same total molar concentration of metal (8.9 × 10 - s M) in each sediment type. In the Cu + Cd treatment, this was achieved by adding 3.12 × 10 -5 moles each
199 Table 1. Experimental design Sediment type Metal
Sand
Oxide
Mixed
Treatments
Organic Matter
Sediment control (no plants,
11=4
n=4
n=4
n=4
n = 6
n = 6
n= 6
n= 6
n = 4
11 = 4
n= 4
n= 4
n=4
n=4
n=4
n=4
Cu+Cd) Plant control (no metal )
Cu 5.8 Fzg g - 1
Cu + Cd 5 # g g - I Cd + 2 . 9 / z g g - t Cu
of Cu and Cd to the 700 g of sediment in each pot, giving nominal concentrations of 5 #g g dry weight-I Cd and 2.9 pg g dry wt - I Cu, while in the Cu treatment, 6.23 x 10 -5 moles of Cu was added (giving a concentration of 5.8 Ftg g - l ) . The metal concentrations added approximately an order of magnitude less cations than could theoretically be complexed by the number of particulate binding sites available, thus promoting competition between geochemical fractions for metal ions. In addition, the metal concentrations are in the range reported for sediments in lakes near Sudbury, Ontario (Tessier et al., 1993, 1985), with the Cd concentration well below the threshold toxicity level of 30 Itg g dry w t - I reported for rice (Muramoto et al., 1990). Sediment preparation Ferrihydrite was prepared by precipitation with alkali following the protocol of Schwertmann and Cornell (1991). Briefly, 1M KOH was added to Fe(NO3)3 • 9H20 dissolved in distilled water, until the pH of the solution was between 7 and 8. The resulting slurry was then centrifuged (12000 × g , 10 minutes) to remove the supernatant, dialyzed over 2-3 days to remove electrolytes, and freeze dried. The peat (Canada Brand, Annapolis Valley Peat Moss Co., Nova Scotia) was sieved to retain only particles less than 2 mm in diameter.
Oxide and organic matter in amounts appropriate for each sediment type were individually mixed with 700 g of air dried silica sand in plastic bags. The metal treatments were added in solution (50 ml volume diluted from BDH 1000 ppm Atomic Absorption Standard), and the sediment again thoroughly mixed before being transferred to 100 mm Standard plastic plant pots. Each pot was acid washed prior to use, and contained an acid-washed circle of nitex mesh (I #m pore size) at the bottom to retain sediments and allow for water movement into the pots. The pots were then kept immersed in deionized ultrapure water buffered to pH 6.5 to allow for development of a partitioning equilibrium over a two week period. An initial sediment sample was collected for fractionation prior to rice seedlings being planted in the pots at the start of the experiment (see below). Samples were collected from three replicates from each of the sand sediments (SAND column, Table 1), and from all four sediment types in the Cu + Cd treatment (bottom row, Table I ). Plant material and culture Seedlings of rice (Oryza sativa var. M 101 ) were grown in a soil/peat mixture under combined natural and fluorescent light in a greenhouse for five weeks, at which time they were approximately 20 cm high. The plants were carefully separated from the soil, and the shoots and roots trimmed to 5 cm and I cm respectively. Six plants were collected for initial dry wt and copper and cadmium analyses. Three seedlings were transplanted into each of the pots containing the experimental sediments, and placed in the greenhouse under natural light supplemented by fluorescent lights on a 14:10 light:dark cycle. Each pot was placed in a larger container that acted as a water reservoir, ensuring the sediment remained saturated during the experiment. Ambient temperatures in the greenhouse varied daily from approximately 23 to 35 °C. At intervals of four days, each pot received 40 ml of nutrient medium in its external water reservoir. The nutrient was Peter's Professional 20/20/20 Medium, added at a rate of 5 g l-I distilled water, and was adjusted to pH 6.5 with 1 M HNO3 and NaOH. Distilled water that was adjusted to pH 6.5 was added as necessary to maintain sediment saturation. Plant mortality was noted every second day, and dead plants removed. The experiment was terminated after 28 days, when approximately 10 g of sediment for metal fractionation
200 analysis was sampled from the upper three cm of each pot. All plants were then carefully excavated, washed with distilled water to remove adhering sediment particles, and dried at 60 °C overnight. After cooling, they were weighed, and digested with hot concentrated HNO3. The digestate was cooled and made up to constant volume with deionized-distilled water.
Sediment extraction All glassware, centrifuge tubes, vials and pipettes were acid washed in 10% HNO3 and rinsed with deionized water five times prior to use. A sequential extraction procedure, modified from Tessier et al. (1979), was used to extract metals in four operationally defined phases: leachable (exchangeable and carbonate associated: 1M NH4OAc, room temperature for 5 hours with continuous agitation); reducible (oxide associated; 0.04M NH~OH.HCL in 25% HOAc, heated to 96 °C., with occasional agitation for 6 hours); oxidizable (organic associated; 0.02M HNO3, 30%H202, heated to 85 °C with occasional agitation for 3 hours); and residual (resistate mineral phases; 4: l HNO3 + HCIO4 digestion, extracted with 4:1 deionized water+ warm 6M HC1). A 20:1 ratio of reagents to dry weight of sediment was used (see Warren & Zimmerman (1993) for a complete set of reagents and conditions). The supernatants from each extraction step were analyzed for Cu and Cd using flame atomic absorption spectroscopy (Laboratory Instruments Spectrophotometer, Model 351 ). Blanks and NBS reference material (NBS 2704, Buffalo River Sediment) were also run through the extraction; the total concentrations of metal recovered from the latter samples were within the certified ranges.
Statistical procedure Data on metal concentrations were log transformed, and proportional metals and plant survival data arcsine transformed, prior to statistical analysis. Plant survival was analysed by analysis of variance with repeated measures, followed by Duncan's multi-range test for differences between sediment types. All other data were analysed by ANOVA followed by Duncan's test for differences between sediment types and metal treatments. Data in figures and the text are shown as mean+standard error. Significance levels in the text are at p0.88; Fig. lb). Therefore, the relative proportions of Cu in each geochemical fraction were compared across all four sediment types in the Cu + Cd treatment. The leachable and oxidizable forms of Cu dominated in different sediment types. In the sand sediment, the proportion of leachable Cu (68%) was significantly higher than in other sediment types, while the organic sediment (OM) had a significantly higher proportion of oxidizable, i.e. organic-associated, Cu (69%). Surprisingly, reducible Cu (i.e. oxideassociated) was significantly higher in the mixed (29%) than in the oxide sediment (12%), which contained about the same relatively small amount of reducible Cu as the sand and organic sediments. The fractionation patterns in the sand sediment were not significantly different between the Cu and Cu +Cd treatments (cf. Figs la, lb; p>0.10), suggesting that the presence of Cd had no effect on the partitioning of Cu. The total concentrations of Cd recovered initially in the Cu +Cd treatment were similar to the nominal level of 5 #g ~o.--I , and showed no significant differences between sediment types (p>0.70; Fig. lc). In terms of fractionation, the leachable form of Cd was dominant. Similarly as for Cu, the sand sediment had a significantly higher leachable fraction of Cd (91%) than other sediment types, while the OM sediment had the lowest (61%). Reducible Cd was significantly higher in the oxide, mixed and organic sediments compared to sand.
Plant survival, growth and metal accumulation The concentrations of Cu and Cd used in these treatments were far more toxic to the rice plants than was expected, based on previous studies with rice. Both Cu and Cu + Cd treatments resulted in rapid mortality of plants in most sediment types, with
