Toxicity to aquatic organisms in the water column

In recent years, it has - again and again - been recognized that the toxicity, e.g. the acute median lethality, LC50, of a trace metal to a single aquatic species can vary widely between different bioassays. This has become a problem, because results of toxicity tests are often used to establish Water Quality Criteria or as inputs into environmental risk assessments, and with variable toxicity data it is difficult to apply criteria based thereon with any certainty.

Therefore, environmental chemists and ecotoxicologists have concentrated, over the last few years, on careful examinations of the causes for the variation in toxicity and then, the relative importance of the various toxicity-modifying factors has been quantified. The variation in trace metal toxicity is directly related to variations in their bioavailability and speciation which, in turn, are largely governed by water hardness, pH and the concentration of organic ligands.

In this sub-chapter examples are given of both the extremely large biological species-related variation in acute and chronic toxicity data for copper in freshwater (LC50-values ranging from 5 to 20,000 jg/l, and chronic thresholds, NOEC values, from 2.0 to 338 jg/l, as dissolved Cu) and the water hardness-related wide variation in LC50-values within one single species (of fish). Recently reported data indicate that in tests with invertebrates and microalgae, dissolved organic carbon (DOC), followed by pH, is affecting chronic copper toxicity to a much greater extent than water hardness. The variability in aquatic copper toxicity can be reduced to about a factor of 2 when the organisms' toxic response to copper in the water is estimated by means of the Biotic Ligand Model (BLM). By this method it was demonstrated that the conditional binding constants of cupric ions to the biotic ligands (BL) in gill tissue of both fish and daphnids were fairly constant (log KCuBL = 7.4 - 8.0), and that difference in sensitivity between species was reflected by the fraction of binding sites in gills that must be occupied by metal to produce 50% mortality (varying from 0.6 to 40% in the case of Cu). Other copper species, such as CuOH+ and CuCO3, do contribute to copper toxicity, since they exhibit fairly strong binding to gill BLs.

Compilation of aquatic toxicity data on nickel has provided evidence that algae and, especially, invertebrates are much more sensitive to nickel exposure than fish, among which rainbow trout larvae appear to be the most sensitive with a 96-h LC50-value of 50 jug/l in soft freshwater. Nickel toxicity to fish is greatly affected by the age of the fish followed by the DOC content and hardness of the water. The lowest chronic effect value recorded for nickel was found in a test with a marine crustacean, 22 jug/l. However, based on BLM predictions of acute toxicity, the sensitivity of freshwater fish to nickel was about 20-25 times lower than that to copper.

The lowest recently reported value for chronic zinc toxicity to fish in very soft freshwater is in the range 16-27 jug/l, a level corresponding to the range given in the previous zinc monograph (Landner and Lindestrom, 1998) for "maximum tolerable concentrations " of zinc in soft inland waters, 15-25 jug Zn/l. The conditional stability constant for binding of zinc to the BL (log KZnBL) was estimated at 5.31 and the fraction of binding sites occupied by zinc to produce 50% mortality in acute exposure of rainbow trout was found to be 14%. The fraction of binding sites corresponding to a "no-effect-level" on chronic exposure was determined to be in the range 5.1-7.7% in rainbow trout and Daphnia magna. In a series of predictions of safe zinc concentrations (or no-effect-concentrations) on chronic exposure in European inland waters, based on the authentic, prevailing water qualities, the results were: 120-1,080 jug/l for rainbow trout, 110-760 jug/l for daphnids and 23-100 jg/l for green algae. Variations are due to varying concentrations of organic matter, prone to form non-bioavailable complexes with zinc.

7.1.1 Copper Sensitivity to copper of different aquatic organisms

Because copper is an essential trace metal, several groups of aquatic organisms have developed strategies for regulating internal copper concentrations (cf. e.g., Brix et al., 2001). These physiological strategies range across a continuum from active regulation (e.g., excretion of excess copper or limitation of net uptake) to storage, i.e. sequestering excess copper in forms that are either metabolically available (e.g., metallothioneins) or unavailable (e.g., calcium phosphate granules) (Philips and Rainbow, 1989; Rainbow and White, 1989; Depledge and Rainbow, 1990). It is also necessary to consider the permeability of membranes, that can influence the sensitivity to metals. This is demonstrated by the fact that some organisms can adapt to elevated metal levels in the surroundings by reducing their membrane permeability. Finally, allometric considerations may be important, such as the amount of permeable membrane relative to absolute body size.

The capacity of organisms to regulate internal levels of copper - and other trace metals - and thereby respond to toxic effects of the metals, may be jeopardized if the organism is exposed to simultaneous additional environmental stress. For example, when the gammarid amphipod Paramoera walkeri (living in coastal Antarctic waters) was simultaneously exposed to UV-B radiation and food shortage, its sensitivity to copper increased by more than 30-fold. The added environmental stress caused by UV-B radiation obviously reduced the general fitness of the amphipods, and food shortage further increased their sensitivity to copper. The finding indicates that energy is required by the amphipods to cope with exposure to copper (Liess et al., 2001). These findings are also in agreement with the fact that starving fish get a reduced capability of excreting copper via the bile. For example, starved yearling roach showed much higher accumulation of copper in the liver than fed fish, suggesting that food-deprived fish lack the ability to regulate transfer of copper within the body (Segner, 1987).

Another important biological phenomenon that can influence the general level of copper sensitivity in aquatic organisms is the social interactions in a population. In a study of the effects of copper on the behaviour of stream-dwelling salmonid fish, as well as the effects of social behaviour on copper uptake, it was found that the salmonids' characteristic social behaviour was not influenced by sub-lethal waterborne exposure up to 30 ^g Cu/l. However, the social status of a fish determined its copper uptake, so that subordinate fish accumulated more copper (and sodium) from the water than dominant fish (Sloman et al., 2002).

It has been repeatedly mentioned in this report, e.g. in chapter 6, that water hardness is a major factor determining the speciation and, thus, the bioavailability and toxicity to fish of waterborne copper. In order to give a direct impression of to what extent water hardness will cause variation in acute toxicity to fish, it may be pertinent to cite some data compiled by Hansen et al. (2002), showing rainbow trout LC50 values (96 h) measured at different water hardness levels on juvenile fish with body weights of <5 g. Some data in very condensed form are shown in Table 7.1.

Table 7.1. LC50 values (96 h) for dissolved copper (^g Cu/l) measured with larval rainbow trout (body weight <5 g) at different levels of water hardness. Compiled from older studies by Hansen et al., 2002.

Hardness, mg/l as CaCO3 LC50 values (96 h), ^g Cu/l

9 25

* results from experimental work by Hansen et al., 2002

It is clear from the above table that acute toxicity of copper to fish varies with water hardness. Therefore, it is important to keep in mind that when copper toxicity data is expressed in this classical way, as concentrations of dissolved copper in water, the water quality (e.g. the hardness) of the water resources to be assessed must be considered. In the Scandinavian countries, lakes are quite often characterized by medium, low, or even very low water hardness, meaning that numerical values in the range 10-50 mg/l as CaCO3 are quite common (Regoli, 2003). Only in the southeastern part of Sweden (except the southernmost province of Scania), hardness levels of 50 mg/l and higher are usually found. This means that the acute toxicity data for rainbow trout that was normalised to 50 mg/l hardness (39 ^g Cu/l, Table 7.2) would in stead fall in the range 3.4 - 36 ^g Cu/l in large parts of the Swedish inland waters.

The data summarized in Table 7.1 also indicates that there are other causes of toxicity data variation, in addition to water hardness. For example, the hardness interval 31 - 41 mg/l corresponded to LC50 values (96 h) as low as 3.4, 8.1 and 14 ^g Cu/l, although the lowest LC50 value at the hardness levels 25 and 90 - 102 was 17 ^g Cu/l. No good explanation of these variations in acute toxic values was given in the paper by Hansen et al. (2002). However, it is also well-known that organic copper complexes are readily formed in freshwater and that these are generally not bioavailable nor toxic. Thus, the explanation of the variations shown in Table 7.1 may well be that concentrations of organic ligands in the different tests show a wide, but possibly undetected, variation.

De Schamphelaere (2003) has shown that aquatic organisms other than fish have a somewhat different pattern of reaction to water hardness as a modifyer of copper toxicity. In, for example, Daphnia magna and the green alga Pseudokirchneriella subcapitata water hardness did not significantly affect chronic copper toxicity. On the other hand, it was demonstrated that DOC explained about 60% of the variability and that pH explained about 15% of the variability in the chronic toxicity values in these test organisms.

In another paper, the copper complexation capacity was determined in a range of final effluents from sewage treatment plants and in the receiving waters upstream and downstream of the discharge point and the variation in copper toxicity to Daphnia magna was recorded (van Veen et al., 2002). The studied sewage effluents contained copper-complexing ligands in a three- to eight-fold excess of total dissolved copper, and the ligand concentrations were found to correlate strongly with DOC concentrations in the effluents. The toxicity of copper to the daphnids was greatly modified by the sewage-derived organic ligands. In the most extreme case, the tolerance of Daphnia to dissolved copper was quadrupled, in close relationship with the complexing capacity and the DOC concentration of the test solution.

Davies et al. (1998) developed a bioassay with Cu-sensitive bacteria isolated from tropical river water to determine Cu bioavailability. The strong linear correlation (r=0.93, p<0.005) observed between measured EC15 values and the Cu complexation capacity of the river water as determined by anodic stripping voltammetry (ASV) confirmed again the role of natural organic matter to mitigate Cu toxicity and that Cu bioavailability in freshwaters is reduced by natural organic compounds.

In their comprehensive work to develop species sensitivity distributions for different taxonomic groups of freshwater aquatic animals, Brix et al. (2001) used copper as an example. They have compiled a great number of data on acute copper toxicity from the U.S. EPA copper toxicity database and from other sources. Acute data for 86 different freshwater animal species were listed, where LC50 values were normalised to a hardness of 50 mg/l (as CaCÜ3) using the U.S. EPA's equation for hardness normalisation (U.S. EPA, 1985). Moreover, chronic data (also normalised to a hardness of 50 mg/l) were compiled for 17 species. All toxicity data are from laboratory experiments and are apparently expressed in terms of total dissolved copper. Just to give an impression of the distribution in sensitivity, acute and chronic data are presented for those species on which measured chronic data is available (Table 7.2).

Table 7.2. Examples of acute and chronic toxicity data regarding dissolved copper (|g/l), normalised to a water hardness of 50 mg/l, for freshwater animals belonging to different taxonomic groups. After Brix et al., 2001.

Main group






Daphnia magna





Other daphnids

5.2 - 69


Gammarus pseudolimnaeus




Rainbow trout

Oncorhynchus mykiss




Brook trout

Salvelinus fontinalis




Bluntnose minnow

Pimephales notatus




Fathead minnow

P. promelas




Bluegill sunfish

Lepomis macrochirus





Physa integra





Campeloma decisum





Clinostornia sp.




* Species Mean Acute Value - geometric mean of LC50 values for a given species ** Species Mean Chronic Value - geometric mean of the mean of LOEC and NOEC for a given species *** Acute-Chronic Ratio

* Species Mean Acute Value - geometric mean of LC50 values for a given species ** Species Mean Chronic Value - geometric mean of the mean of LOEC and NOEC for a given species *** Acute-Chronic Ratio

The 86 freshwater species for which (quality controlled) acute test data was available showed a range in normalised LC50 values for dissolved copper from 5.2 to 20,000 |g/l. The 17 species that exhibited high quality data on chronic toxicity to copper showed a much smaller variation, 2.8 -65.6 |g/l (Brix et al., 2001). If one "outlier", the northern pike, Esox lucius, is excluded, the range is reduced to 2.8 - 33.4 |g/l. The very low estimate of chronic toxicity to bluntnose minnow, 2.8 |g Cu/l, may furthermore merit a special comment. On closer examination of the original paper (Horning and Neiheisel, 1979), it turns out that the lowest tested concentration was 18 |ng/l, which makes the estimate of the SMCV at 2.8 |g/l very unreliable. Moreover, the estimate of chronic toxicity to the fathead minnow (same genus as bluntnose minnow), was based on 5 tests and gave a geometric mean of 11.2 (o,g/l. The acute values for the two species were not as far apart (72 ^g/l and 133 ^g/l) as the chronic values. It is known that differences in the susceptibility of individual fathead minnows - and perhaps bluntnose minnows - to Cu are genetically determined to a large extent. Genetic differences in minnow susceptibility to Cu are possibly coded by genes producing proteins involved in ionoregulation and/or accumulation and sequestering of Cu (cf. Kolok et al., 2002).

Among invertebrates, the cladoceran taxonomic group clearly is the one most sensitive to Cu of those investigated, while insects are, on average, the least sensitive one. For fish, on the other hand, it was not possible to identify any trend in relative sensitivity with respect to feeding guild or phylogenetic relationship. There was, however, a general trend towards higher sensitivity of temperate cold-water species, followed by temperate warm-water species, which in turn appeared to be more sensitive than tropical species (Brix et al., 2001).

In order to complement the database on chronic toxicity of Cu to freshwater animals, such data was estimated from the SMAVs, using a variable acute-chronic ratio (ACR), according to the following equation:

This approach was chosen because it was found that when chronic data was estimated with a variable ACR, the correspondence with measured data was much better for a variety of species, than when using a constant ACR (Brix et al., 2001). It was also demonstrated that the distribution of measured chronic data was much steeper than the acute effects distribution, which is reflected in Table 7.2. These results suggest that there may be a chronic threshold for Cu related to the ability of organisms to regulate this essential element.

Later literature surveys made it possible to increase the database of observed high-quality chronic NOECs for Cu in the aquatic environment : Delbeke et al. (2003) compiled 126 NOECs from 21 species and found them to range between 2 and 338 ^g Cu/l as dissolved Cu. In this context, it may be pertient to mention that in a recent paper (Baldwin et al., 2003), it was shown that when concentrations of total dissolved Cu were increased by 2.33.0 ^g/l over background concentration, Cu was clearly toxic to the olfactory nervous system in coho salmon (Oncorhynchus kisutch). The inhibitory effects of copper were dose-dependent and they were not influenced by water hardness.

In their recent paper, Delbeke et al. (2003) also presents an exercise of estimating a provisional PNEC value for Cu in freshwater, based on the species sensitivity distribution, the chronic NOECs, of fish and invertebrates, compiled from the literature. When the NOECs were normalised to the 5th percentile of pH, water hardness and DOC (taken from the comprehensive physico-chemical water quality database for European inland waters) by means of the chronic Cu BLM, developed for gill-breathing organisms, the combined outcome in terms of a normalised PNEC was found to be 8 ^g Cu/l as dissolved Cu. Toxicity of copper estimated by means of BLMs

In the previous section, a certain number of examples were given, all supporting the view that the toxicity of copper to freshwater aquatic organisms is dependent on a variety of ambient water chemistry parameters. This view is further strengthened by data compiled by Van Sprang (2002) showing a variation in LC50 values for one single fish species (Pimephales promelas) from 10 to 1000 ^g Cu/l when the acute toxic concentration is expressed as dissolved aqueous copper. In a similar survey presented by Di Toro et al. (2001), the LC50 values for larval stages of the same fish species ranged between 6.0 and 1560 ^g Cu/l.

This variability in toxicity due to variation in water chemistry was further confirmed and demonstrated in the work by de Schamphelaere and Jansen (2002), where they developed a BLM predicting copper toxicity for Daphnia magna. As a part of their efforts to estimate biotic ligand constants, these authors conducted a series of acute toxicity tests with D. magna, in order to assess the variation in 48-h EC50Cu2+ as a function of the chemical activity of calcium, magnesium, sodium, potassium, hydrogen and hydroxide ions. In each group of tests, one among six different water quality parameters was systematically varied while all the others were kept constant and the toxic response (EC50) was measured. In Table 7.3, the variation in EC50 values, expressed as total dissolved copper, is shown as a function of the variation in the different chemical activities.

Table 7.3. Variation in toxic response (EC50) in Daphnia magna, expressed as ^g/l of total dissolved copper, when each of six chemical activities (in mM) in the test solution was changed. After De Schamphelaere and Janssen, 2002.

Chem. activity modified Range 1, mM Range 2, mM Variation in EC50, ^g Cu/l

0.25 - 4.0 0.25 - 5.0 0.25 - 4.0 1.08 - 15.1 0.08 - 2.0 5.98 - 7.92

Modification of each of the parameters, except potassium, produced quite strong variation in toxic response of the test animals. In relative terms, manipulation of the pH caused the strongest variation in EC50 values for copper. It should be noted, however, that the DOC concentrations in these tests were assumed to be constant (0.14 mg DOC/l) and to have the same metal complexation properties as natural organic matter. Thus, in an authentic, natural surface water, the DOC - as a cause of variation in EC50 values for copper - should be added to the causes of variation related to the above investigated inorganic parameters. As a matter of fact, it was later clearly demonstrated that DOC is the most important water quality factor among all those affecting the chronic copper toxicity to Daphnia magna (De Schamphelaere, 2003).

As it has been repeatedly mentioned before, the great variability in toxic response to copper by aquatic organisms, depending on the prevailing water quality, can be almost eliminated by deriving the toxic response by means of a BLM. Van Sprang (2002) showed that the BLM-predicted EC50s differed from the observed EC50s by a factor of less than 2, while De Scahmphelaere and Janssen (2002) found this difference to be a factor of less than 1.5.

In the assessment of acute toxicity by means of BLM, the biotic ligand for fish is the gill. What is needed, thus, is to predict the amount of metal accumulated at the surface of the fish gill in order to predict metal toxicity to the fish (Di Toro et al., 2001). When juvenile rainbow trout or fathead minnows are exposed to dissolved copper, a relatively rapid increase occurs over background levels of copper bound to the gills. This rapid initial increase takes place in a few hours to a day (Playle et al., 1992). It is believed that the rapid initial increase in gill copper reflects binding to receptor sites at the gill surface that control the ionoregulatory processes of the fish. The next step in predicting acute toxicity is to quantify the relationship between mortality of the fish after 120 h of exposure and copper concentration on the gill after 24 h of exposure.

MacRae et al. (1999) performed a series of experiments with juvenile rainbow trout, using a constant total dissolved copper concentration (10 ^g/l), and the gill copper concentration was regulated by adding different organic ligands with varying affinities for copper to the test water. The gill copper concentration that caused 50% mortality was estimated to be 22 nmol/g wet weight (ww). The background gill copper concentration (at no mortality) is approximately 12 nmol/g ww, a value that corresponds well with the gill background in fathead minnow, measured by Playle et al.

(1992) in the absence of added copper. It was provisionally concluded that the gill copper LC50 (defined as the LA50, median lethal activity, value) should be approximately 10 nmol/g ww (the difference between the measured value and the background) for both fish species.

Further work by Di Toro et al. (2001), using a range of DOC, hardness and pH conditions in the tests, confirmed that the LA50 value for fathead minnow averaged 5 to 12 nmol/g ww. This finding is consistent with the previously demonstrated similar sensitivity of fathead minnow and rainbow trout to copper. It is also consistent with the finding that the two fish species accumulate copper on their gills in a similar manner and suggests that binding constants for metal-gill interactions for one species can be generalised to other fish species. The copper-gill BL binding constants for rainbow trout (7.5) and brook trout (7.2), found by MacRae et al. (1999) are quite close to the value for fathead minnow (7.4) found by Playle et al.

(1993). Thus, the BLM can explicitly account for variation in toxicity resulting not only from changes in hardness, but also from site-specific variations in DOC, pH and alkalinity.

Although the BLM has turned out to be an excellent tool to handle the great variation in acute toxicity values caused by the variable conditions for trace metal speciation and complexation in the water, and also to consider the competition between different ions for binding sites at the BL (cf. chapter 6), there is - of course - still a variation in toxicity due to differences in sensitivity between the biological species. This may be illustrated by the compilation of "critical BL concentrations" or LA50 values for different organisms shown in Table 7.4.

Table 7.4. BL binding constants, log KMeBL (M-1), site densities in gills (nmol/g ww) and critical BL concentrations or LA50 values (nmol/g ww), and fraction of binding sites occupied by metal to produce 50% mortality for copper, silver and nickel in various organisms. Sources: Di Toro et al., 2001; De Schamphelaere and Janssen, 2002 ; De Schamphelaere, 2003.

Table 7.4. BL binding constants, log KMeBL (M-1), site densities in gills (nmol/g ww) and critical BL concentrations or LA50 values (nmol/g ww), and fraction of binding sites occupied by metal to produce 50% mortality for copper, silver and nickel in various organisms. Sources: Di Toro et al., 2001; De Schamphelaere and Janssen, 2002 ; De Schamphelaere, 2003.



log KMeBL

Site density


/- 50% /MeBL

Rainbow trout






Fathead minnow




5 - 12












Ceriodaphnia dubia






Daphnia magna





0.39, 0.47






* figures in brackets are provisional estimates made by authors of the present report

* figures in brackets are provisional estimates made by authors of the present report

It should be noted that the latest constants calculated for acute toxicity of copper to Daphnia magna are based on the observation that two copper species in addition to cupric ions are bioavailable and can produce toxic effects in the animals, viz. CuOH+ and CuCO3. The log KMeBL constants for these complexes are estimated at 7.32 and 7.01, respectively, i.e. clearly lower than the constant for cupric ions, 8.02 (De Schamphelaere, 2003).

Table 7.4 indicates that binding constants, log KMeBL, for metal-gill interactions are quite similar between different biological species and for, at least, copper and silver. (It should be noted, however, that the log KZnBL for acute Zn-BLM to Daphnia magna was found to be 5.31 (Heijerick et al., 2002a)). Also the site density, i.e. the total binding site density of the biotic ligand (e.g. nmol of available sites per gram of tissue) appears to be quite similar for different species and metals. The variation in sensitivity between organisms and metals is primarily reflected in the critical concentration producing mortality, i.e. the LA50 value. This value shows a certain variation between the fish and the tested crustaceans. The fraction of binding sites needed to be occupied by the actual metal to cause 50% mortality is, consequently, also a variable.

In the development of their predictive copper toxicity model for green microalgae, De Schamphelaere et al. (2003b) demonstrated that DOC concentration and origin as well as pH of the test water had a significant effect on copper toxicity to the algae. The model was also validated by using copper-spiked European surface waters.

However, most of the BLM parameters discussed above have been determined in laboratory tests in waters with very low concentrations of DOC (or NOM, natural organic matter). It is therefore important to consider to what extent predictions of toxicity in natural waters may be biased by this lack of realistic experimental conditions. Just to give a few examples, some results given by Richards et al. (2001) may illustrate the issue: Copper was found to bind to NOM about 50 times better than to rainbow trout gills (conditional binding constants of log KCu-NOM = 9.1 and log KCu.gill = 7.4), while cadmium bound about 16 times less well to NOM than to gills (log KCd-NOM = 7.4; log KCd-giu = 8.6). Thus, NOM in the experiments conducted by Richards et al. (2001), was able to keep copper, but not cadmium, from binding to the gills. Moreover, it was found that silver binds about 10 times more strongly to the rainbow trout gills than to NOM (log KAg.gill = 10.0; log KAg-NOM = 9.0), while cobalt showed the same binding strength to the two kinds of ligands (log K = 5.1).

The conditional stability constants for sorption of copper to the marine alga Emiliania huxleyi were determined by Teresa et al. (2001) both with regard to extracellular adsorption (KCuSe) and intracellular uptake (KCuSi). The value of log KCuSe was 12.3 and that of log KCuS; was 11.0, and the constant for extracellular adsorption was compared to the corresponding values (from the literature) for two other algae, one marine (Dunaliella sp.) -8.8, and one freshwater species (Scenedesmus sp.) - 11.1.

An interesting observation made during the experiments with E. huxleyi was that copper dissolved in natural coastal seawater was taken up very rapidly into the algal cells: after 10 min. exposure, some 80% of the equilibrium amount of copper in the cells was already intracellular, while the process of external adsorption was somewhat slower. Moreover, Teresa et al. (2001) also determined the conditional stability constants for copper bound to either the exudates excreted by the algal cells as a result of copper exposure or to the organic ligands occurring in the seawater prior to copper addition. In both cases they found that these constants were practically of the same magnitude (12.2 - 12.3) as the constants for extracellular adsorption of copper to algal cells. The production of exudates from the cells was quite fast and after 10 min., the molar concentration of organic ligands in the water was close to the molar concentration of copper. It was therefore concluded that the ligands present in the seawater, mainly those released by the cells, can compete efficiently for copper with the E. huxleyi cell sites.

0 0

Post a comment