Reaction mass of zinc oxide and zinc peroxide

Endpoint summary

Additional information

No aquatic toxicity data are available for the target substance Zinc peroxide. However, extensive information is available on (ionic) Zinc, which is the main determinant of ecotoxicity. Further supporting data are available on hydrogen peroxide. A justification for read-across is attached to IUCLID section 13.

 

All cells apart from anaerobic bacteria produce hydrogen peroxide in their metabolism. Furthermore, Hydrogen peroxide is formed abiotically in the environment. To prevent oxidative damage, cells have developed the ability to decompose some amount of excess Hydrogen peroxide (EU RAR, 2003).

In the transformation/dissolution test conducted with the target source Zinc peroxide, it was demonstrated, that only low levels of hydrogen peroxide are released.

In the view of the high degradation capacity for hydrogen peroxide in many organisms, it is unlikely that the low levels of hydrogen peroxide released from ZnO2 is distributed in the organisms, and therefore the endogenous steady state levels of hydrogen peroxide are unlikely to be affected.Therefore, it is concluded, thatmainly Zn2+ determines ecotoxicity.

 

In accordance to the scientific approach followed in the EU RA process, there are two key approaches to the zinc hazard assessment:

 

The generally accepted assumption is that the ecotoxicity of zinc and zinc compounds is due to the Zn²⁺ion. Consequently, all aquatic, sediment and terrestrial toxicity data in this dossier are expressed as “zinc”, not as the test compound as such, because ionic zinc is the determining factor for ecotoxicity. A further consequence of this is that all ecotoxicity data obtained on different zinc compounds, are mutually relevant for each other. For that reason, the available ecotoxicity databases related to zinc and the different zinc compounds are combined before calculating the PNECs. The only way zinc compounds can differ in this respect is in their capacity to release zinc ions into the environment. This property is determined in the transformation/dissolution (T/D) assay and may result in different classifications.

Zinc is a natural component of the earth’s crust and present in natural background concentration in all environmental compartments. Because of the importance of the natural background, the “added risk concept” has been used in the RAR on zinc (ECB 2008). In this approach both the "Predicted Environmental Concentration"(PEC) and the "Predicted No Effect Concentration" (PNEC) are determined based on the added amount of zinc, resulting in an “added Predicted Environmental Concentration” (PEC_add) and “added Predicted No Effect Concentration” (PNEC_add­), respectively. The use of the added risk approach implies that only the anthropogenic amount of a substance, i.e. the amount added to the natural background concentration, is considered to be relevant for the risk assessment. Thus, a possible contribution of the natural background concentration to toxic effects is ignored (RAR; ECB 2008). So, for zinc, all PNECs are expressed as “added” concentration to the background.

The use of the added risk approach implies that for risk characterisation, the natural background needs to be taken into account when evaluating monitored concentrations in the environment. The correct assessment of natural background is thus important.

 

 

General considerations

 

Zinc and zinc compounds form a “data rich” substance group: a vast volume of information is available on the effect of zinc on the different ecotoxicity endpoints in the open scientific literature. This vast volume of ecotoxicological information was carefully scrutinised by the Rapporteur (the Netherlands) in the framework of the discussions on the EU risk assessment (RA) on zinc and 5 zinc compounds made under EU Regulation 793/93/EEC. In that process, the Rapporteur’s analysis of the available chronic ecotoxicity data was extensively discussed by the experts from member states and other stakeholders during the meetings of the “Technical committee on new and existing substances” (TCNES), where the data sets to be used for PNEC derivation were officially approved. The scrutiny and discussion of TCNES were focusing on the chronic data for the PNEC derivations.

 

For this reason, the data used in the RAR are the main data source for the environmental hazard assessment. Decisions on data quality and relevance that were approved by TCNES are followed and not discussed anew. Consequently, data that were considered relevant for PNEC derivation in the RAR will be used as such. Studies that were disregarded in the RAR as not relevant or not reliable are not included in this dossier.

 

The datasets from the RAR have been complemented with relevant, reliable information that became available after the closure of the RA databases.

 

Supporting data on hydrogen peroxide are also included:short-term toxicity data for fish, invertebrates and algae, as well as long-term data on zebra mussels are available. The lowest long-term aquatic toxicity test result is the NOEC of 0.1 mg/L for algae(EU RAR, Hydrogen Peroxide, 2003). As the effective values of hydrogen peroxide are higher than those of Zinc, these data further demonstrate that theecotoxicity of the target substance Zinc peroxide is determined by the release of Zn2+.

 

Due to the extensive database on Zinc, species sensitivity distribution was used to derive environmental PNECs. The values are expressed as Zn2+.

 

Aquatic toxicity: freshwater, short-term

 

Establishing the dataset

In accordance to the approach followed in the RAR, only acute data from standardised test protocols were considered in the analysis for setting the reference value for classification. This is possible because numerous data are available, and it ensures that the tests were performed under rather well defined and standard conditions.

Still, the quality and some aspects of relevancy should be checked in a critical way when using the extensive datasets from the open literature, available for zinc. It is e. g. important to know the conditions under which the organisms were tested and cultured, because these conditions may result in acclimatisation and deviating toxicity response. The information on these test conditions is often scarce in non-standardised test reports.

The short-term aquatic ecotoxicity data base for zinc was reviewed according to the following principles:

·   the data accepted for setting the acute aquatic reference value in the RA (ECB 2008, Annex 1.3.2a, table 1) were as such also accepted and used for the present analysis. Prescriptions from standard protocols were strictly followed; e. g. data from an acute Daphnia test exceeding 48 hrs were not used.

·   Data that were rejected for use in the RA (ECB 2008, Annex 1.3.2a, and table 2) were also not used for the present analysis. In this respect, data from studies that were accepted for use in the chronic database, but rejected for use in the acute toxicity database were reconsidered; this resulted in the acceptance of a few additional data.

·   In accordance to the approach followed in the RA, acute data obtained in natural waters that contained e. g. significant amount of DOC, were not used. Exceptions to this rule were data obtained on the N. -American Great Lakes waters, which were used, in accordance to the RA.

·   Fish data mentioned in the RA under “EHC 1996” were not used, since they were from a review, not from original study reports. These data are not influencing the outcome of the analysis, since they are all at the higher concentration level.

·   More recent (obtained after 1996 to the present) short-term acute toxicity data on standard organisms were included in the database.

After checking and updating the data base, the data are grouped per species as follows:

-      pH: low (6 -<7) - neutral/high (7 -8.5)

-      hardness: low/medium (<100mg CaCO3/l) and medium/high (>100 mg CaCO3/l).

If 4 or more data points were available on a same species, the geomean was calculated and used for the analysis.

 

Acute data – results

The short-term acute aquatic toxicity database covers 10 species (1 algae, 4 invertebrates and 5 fish species). The data are summarized belowtogether with the pH and hardness of the test media. A significant number of data are available at both low and neutral/high pH.

 

Acute aquatic toxicity of zinc by species as a function of pH and hardness.

species	pH	hardness	E(L, I)C50 value (mg Zn/L)	reference
algae
Selenastrum capricornutum (new name: Pseudokircherniella subcapitata	7.4	24	0.136	Van Ginneken, 1994
Selenastrum capricornutum (new name: Pseudokircherniella subcapitata	7.4	24	0.150	Van Woensel, 1994
Pseudokircherniella subcapitata	7.4	43	0.86	De Schamphelaere et al 2005
	8.0	65	0.11
	5.7	4	1.48
	5.8	2	0.71
	6.4	9	0.25
	5.7	4	2.05
	6.8	6	0.35
Pseudokircherniella subcapitata	6.6	8	0.19	Muysen et al 2007
	6.4	6	0.21
	6.6	100	1.23
	6.4	100	0.77
Daphnids
Daphnia magna	7.7 7.7	45 45	0.1 0.28	Biesinger & Christensen, 1972
	7.7	45	0.28	Cairns et al. 1978
	7.2-7.4	45	0.07	Mount & Norberg 1984
	7.2 7.2	46 46	0.259 0.131	Barata et al 1998
	7.6	50	0.330	Chapman, 1980
	7.7 7.7	91 91	1.06 0.475	Barata et al, 1998
	7.0	130	0.8	Atar & Maly 1982
	8.1	105	0.53	Chapman 1980
	7.7 8.1	179 179	0.962 0601	Barata et al, 1998
	8.2	196	0.66	Chapman 1980
	7.2	242	2.14	De Schamphelaere et al 2005
	7.8 7.8	250 250	2.91 1.83	Muysen et al 2005
	8.5	180-200	0.86	Magliette et al 1995
Daphnia pulex	7.6	45	0.5	Cairns et al 1978
	7.2-7.4	45	0.107	Mount & Norberg 1984
	6.3 7.9 7.9 7.9 7.9 7.9 7.9 7.9 7.9 7.9 7.9 7.9 7.3 8.0 7.9 7.9 7.9 7.9 7.9 7.9	84 51 69 86 46 81 84 84 84 84 84 84 84 84 104 131 122 163 241 224	0.425 0.105 0.190 0.399 0.268 0.399 0.532 0.399 0.706 0.399 0.477 0.392 0.765 0.687 0.321 0.556 0.432 0.353 1.014 0.615	Clifford & McGeer 2009
Daphnia carinata	7.5	82	0.340	Cooper et al 2009
Ceriodaphnia dubia	6–6.5 7-7.5 8-8.5	280-300 280-300 280-300	>0.53 0.36 0.095	Schubauer-Berrigan et al 1993
	6.5 7.5 7.5	44 44 44	0.413 0.2 0.155	Hyne et al 2005
	7.5	82	0.174	Cooper et al 2009
	7.3	45	0.076	Mount & Norberg 1984
	7.2-7.3	52	0.169	Carlson et al 1986
	7.8	280	0.67	Muysen & Janssen 2002
	7.8	250	0.416	Muysen et al 2005
Ceriodpahnia reticulata	7.8	250	0.937	Muysen et al 2005
Fish
Pimephales promelas	6-6.5 7-7.5 8-8.5	280-300 280-300 280-300	0.78 0.33 0.50	Schubauer-Berrigan et al 1993
Thymallus arcticus	7.1-8.0 7.1-8.0 7.1-8.0 7.1-8.0 7.1-8.0 7.1-8.0 7.1-8.0 7.1-8.0	41 41 41 41 41 41 41 41	0.315 0.142 0.112 1.58 0.166 2.92 0.168 0.168	Buhl & Hamilton 1990
Cottus bairdii	7.5	154	0.439	Brinkman & Woodling 2005
Oncorrhynchus kisutch	7.1-8.0 7.1-8.0 7.1-8.0 7.1-8.0	44 44 44 44	0.82 1.81 1.65 0.727	Buhl & Hamilton 1990
Oncorrhynchus mykiss	7.1-8.0 7.1-8.0	44 44	2.17 0.169	Buhl & Hamilton 1990

 

The lowest species values (mg Zn/l) are summarised below. When >4 data for the same species and same category of conditions for pH and hardness, geomean values are used.

 

Lowest acute aquatic toxicity data (mg Zn/L) observed

species	Low pH/ hardness≤100mg CaCO3/L	Low pH, hardness>100mg CaCO3/L	Neutral-high pH/ hardness≤100mg CaCO3/L	Neutral-high pH, hardness>100mg CaCO3/L
algae
Selenastrum capricornutum	0.576	/	0.137	/
Daphnids
Daphnia magna	/	/	0.244	1.052
Daphnia pulex	0.425	/	0.364	0.507
Daphnia carinata	/	/	0.34	/
Ceriodaphnia dubia	0.413	> 0.530	0.147	0.228
Ceriodaphnia reticulata	/	/	/	0.937
fish
Pimephales promelas	/	0.780	/	0.33
Thymallus arcticus	/	/	0.307	/
Cottus bairdii	/	/	/	0.439
Oncorrhynchus kisutch	/	/	1.155	/
Oncorrhynchus mykiss	/	/	0.169	/

 

Discussion: reference values for short term aquatic ecotoxicity

An overview of the information available for short-term aquatic toxicity for zinc is given. It can be seen that significant number of data are available at both low and neutral/high pH.

At low pH, 2 values are available for 2 daphnia species. The values are similar. They were obtained at lower hardness, where the highest sensitivity is expected, which is confirmed by the value >530 µg/L, obtained on Ceriodaphnia dubia at high hardness. Algae are usually not tested under standardised conditions at low pH, but from chronic algae data (72 hrs NOECs), it is known that the sensitivity of algae is lower at lower pH; Simulation with the biotic ligand model gives an aquatic ecotoxicity value for algae at pH 6 which is about 5 times higher than the one observed at neutral/high pH. This is confirmed by the data reported here. Fish toxicity at low pH is also not critical in this respect, so the values for the daphnids are representative for the sensitivity of organisms to zinc at low pH. The lowest value observed for Ceriodaphnia dubia is used for the classification at low pH.  

At neutral/high pH, the value obtained on the algae Selenastrum capricornutum is the lowest of the dataset. This value is taken forward as reference value for classification at this pH. This value is obtained at low hardness conditions, where sensitivity is highest. The same algae species is also the most sensitive in the chronic aquatic toxicity database (see below) so this sensitivity pattern is consistent. Among the daphnids, Ceriodaphnia dubia is also here the most sensitive, and the lowest value comes close to the one for the alga. From the paired data, it follows that the Daphnids are more sensitive at lower hardness than at the higher hardnesses. The fish are also at this pH less sensitive to zinc, although the lowest value observed on O. Mykiss also comes close to the reference value.

Overall, the lowest values among the species show also here a consistent pattern, supporting the lowest value identified.

In conclusion, the reference values for the Zn²⁺ion that are used for the aquatic toxicity hazard assessment of Zn²⁺are:

 

·   for low pH:0.413 mg Zn/L(based on single lowest value for Ceriodaphnia dubia)

·   for the neutral/high pH:0.137 mg Zn/L(based on species geomean value for Selenastrum capricornutum (=Pseudokircherniella subcapitata)

 

2. Aquatic chronic toxicity: freshwater

Chronic data - establishing the dataset

In this analysis, like in the RAR, the results of the chronic aquatic toxicity studies are expressed as either the actual (measured) concentration or as thenominal (added) concentration (Cn). The actual concentrations include the background concentration (Cb) of zinc. Because of the “added risk approach”, the results based on actual concentrations have been corrected for background, if possible. This correction for background is based on the assumption that only the added concentration of zinc is relevant for toxicity. In case both actual and nominal concentrations were reported, the results are expressed in the RAR (and in this CSR) as nominal concentrations, provided the actual concentrations were within 20% of the nominal concentrations.

Many of the reported aquatic toxicity data (either actual or nominal) represent total-zinc concentrations, i. e. the dissolved plus particulate fraction. However, the results are regarded as being dissolved-zinc concentrations, because under the conditions that were used in the laboratory tests, it is assumed that the greater part of zinc present in the test waters was in the dissolved fraction. This is especially true for the long-term studies, e. g. by using flow-through systems, in which particulate matter (suspended inorganic material and/or organic matter) was removed from the artificial test waters or natural waters. The fact that in ecotoxicity testing the nominal added concentration of zinc is very close to the actually measured zinc concentration, is also demonstrated by the many data reported in the papers of the chronic aquatic ecotoxicity database. Also in static and flow-through acute toxicity studies with several saltwater species, dissolved zinc was greater than 93% of the total zinc. Therefore, the PNEC_add values derived from the aquatic toxicity studies are considered to be relevant for dissolved zinc.

The chronic aquatic toxicity dataset for zinc was checked according to the general criteria for data quality:

-      study design preferably according to OECD guidelines or equivalent

-      Toxicological endpoints, which may affect the species at the population level, are taken into account. In general, these endpoints are survival, growth and reproduction.

-      whether or not NOEC values are considered chronic is not determined exclusively by exposure time, but also by the generation time of the test species, e. g. for unicellular algae and other microorganisms (bacteria; protozoa), an exposure time of four days or considerably less already covers one or more generations, especially in water, thus for these kinds of species, chronic NOEC values may be derived from relatively short experiments. For PNEC derivation a full life-cycle test, in which all relevant toxicological endpoints are studied, is normally preferred to a test covering not a full life cycle and/or not all relevant endpoints. However, NOEC values derived from tests with a relatively short exposure time may be used together with NOEC values derived from tests with a longer exposure time if the data indicate that a sensitive life stage was tested in the former tests.  

-      If for one species several chronic NOEC values (from different tests) based on the same toxicological endpoint are available, these values are averaged by calculating the geometric mean, resulting in the “species mean” NOEC.

-      If for one species several chronic NOEC values based on different toxicological endpoints are available, the lowest value is selected. The lowest value is determined based on the geometric mean if more than one value for the same endpoint is available.

-      In some cases, NOEC values for different life stages of a specific organism are available. If from these data it appeared that a distinct life stage was more sensitive, the result for the most sensitive life stage is selected. The life stage of the organisms is indicated in the tables as the life stage at start of the test (e. g. fish: yearlings) or as the life stage(s) during the test (e. g. eggs à larvae, which is a test including the egg and larval stage).

-      Only the results of tests in which the organisms were exposed to zinc alone are used, thus excluding tests with metal mixtures.

-      Like in the RAR, unbounded NOEC values (i. e. no effect was found at the highest concentration tested) arenotused.

-      If the NOEC was <100 µg/L, the separation factor between the NOEC and LOEC should not exceed a factor of 3.2.

-      If the EC10 was used as NOEC equivalent, the EC10 should not be more than 3.2-times lower than the lowest concentration used in the test.

-      Like in the RAR, only the results of tests with soluble zinc salts are used, thus excluding tests with “insoluble” zinc salts (ZnO, ZnCO₃), unless dissolved zinc is measured.  

 

Referring to the EU RA on zinc (ECB 2008), all the data that were accepted for deriving the freshwater PNEC in the RA (ECB 2008, Annex 3.3.2 A. part I) were as such also accepted for the present analysis. Likewise, the data that were considered not useful for the purpose of PNEC derivation in the RA (ECB 2008, Annex 3.3.2 A. part II), were not used for the present analysis.

The relevance of the long-term aquatic ecotoxicity data base for PNEC derivation was further checked in accordance to the same principles as those applied in the RA (ECB 2008). Relevance was checked

1)   related to the zinc background: results obtained in unpolluted test media (water, sediment or soil) are used, thus excluding tests that were performed in media containing high to very high background Zn concentrations, i.e. in case the control media contained zinc concentrations that are clearly above Zn concentrations normally encountered in relatively unpolluted environmental compartments. A pragmatic cut-off level of 30µg Zn/l, in accordance to decisions taken in the RA (ECB 2008) was set for this.

In accordance to the RA zinc (ECB 2008), data obtained in tests where the zinc background concentration was much lower than the natural background for EU waters, were also not used for PNEC derivation. In accordance to the RA (ECB 2008), a level of 1µg/l Zn was set as a cut-off for this.  

2)   related to test medium conditions: Zinc ecotoxicity to aquatic organisms is a function of the physicochemical characteristics of the water. Parameters such as hardness, pH, dissolved organic carbon (DOC) are well-known drivers for zinc ecotoxicity. For this reason, it was considered important in the EU RA to select ecotoxicity data that were obtained under test conditions similar to the conditions observed in EU waters. Based on information related to the parameters mentioned above in EU waters, the following boundaries for EU relevancy for pH, hardness have been used in the RA (ECB 2008) and also in the present analysis for data selection (also considering OECD test guidelines):

pH: minimum value: 6, maximum value: 9

Hardness: minimum value: 24 mg/l (as CaCO3), maximum value: 250 mg/l (as CaCO3)

As indicated above, background zinc concentration was also considered in the RA to be a factor influencing the toxicity response of organisms to zinc; to avoid influence of acclimatisation towards very low or very high zinc concentrations (not occurring in the EU waters), a minimum value for soluble zinc was also set in the RAR for data selection: “around 1 µg/l” (ECB 2008).

Data obtained under conditions failing these relevancy criteria were not used for PNEC derivation in the present analysis. For a detailed description of the relevancy criteria and their application in the RA, see the RAR (ECB 2008).

It is realised that the selected ranges of the three criteria will not cover all European aquatic systems, e. g. specific aquatic systems in the Scandinavian countries. In particularly, hardness is much lower in the Scandinavian countries, although also other abiotic parameters differ from the ‘average’ situation in European freshwaters. Therefore, a “soft water PNEC_{add, aquatic}” has been derived in the RA process, in addition to the generic PNEC_{add, aquatic}. This “softwater PNEC” should be used in situations where it has been documented that the hardness is lower then the low end of the range indicated above (24mg/l). The present analysis however relates to the development of a generic PNEC for EU waters.

In the data selection process of the RA, it was noted that the references used for the current aquatic toxicity dataset usually do not contain data on the background concentration of zinc in the test water and in a number of cases also data on pH and/or hardness were lacking. Thus, a stringent application of the above mentioned (minimum and maximum) limits for all three parameters, especially the zinc concentration, would have very strongly reduced the dataset, which was considered not acceptable from a practical point of view (RAR 2008). Therefore, the following approach was followed (EU RA zinc, ECB 2008):

 

-      When data were reported on these parameters, the above selection criteria will be used. 

-      When no data were reported on these parameters:

•      Tests that had been conducted in artificial waters were excluded when data on pH and/or hardness were lacking.

•      Tests that were conducted in natural waters were maintained, unless there were clear indications that the (above) parameters in the water strongly deviated from real environmental conditions. For example, tests in waters that received special treatment to remove zinc (and other cations such as Ca and Mg) were excluded. On the other hand, tests conducted in untreated natural United States' waters that were reported to contain a background zinc concentrations which may be considerably below 1 µg/l (depending on natural seasonal variations), such as Lake Superior water, were not excluded.

 

DOC:

The range of DOC concentrations of natural waters observed in the EU, is 2,1 mg/l (5P) < DOC < 13 mg/l (95P) (EU risk assessment, ECB 2008). When applying DOC as a relevancy criterion, a relevant range for EU waters should thus also be used for DOC, in accordance to the setting of the relevancy criteria on pH, hardness. The relevant DOC range should thus be set within these observed values. Because it is against the logic of setting relevancy criteria corresponding to the ranges observed in natural EU waters, the additional relevancy criterion of DOC <2mg/l that was introduced at a later stage in the RA process is not applied in the present analysis. This DOC range indeed does not reflect the range of DOC observed in EU waters. In fact, the cut-off of <2mg DOC/l corresponds to the lower 5 percentile of DOC concentrations observed in natural EU waters (~=2,1 mg/l; GEMS-A - Heijerick et al 2003, cited in the RAR), and as a result, data obtained under relevant EU conditions would be excluded.

Following the RAR-logic of using realistic ranges for abiotic water parameters in natural EU waters for setting relevancy criteria, a DOC criterion should actually reflect the observed range. Taking into account that artificial test media as a rule do not contain DOC, it is proposed to apply only an upper range for DOC as a relevancy criterion, i.e. the 95P concentration of the EU range. Tests have thus be considered relevant for the present analysis if DOC concentrations in the test media are between 0 mg/l and 13 mg/l.

In practice, however, the DOC criterion was applied in the RAR (2008) only to the ecotoxicity data, generated in natural waters in the conclusion (i) programme on bioavailability. Since these waters were chosen to validate the bioavailability models, their abiotic conditions are rather wide ranging, and some of the parameters, (e.g. pH, hardness) may fall outside of the range, agreed as relevancy criterion. As a result, several of the test data obtained in these natural waters were rightfully excluded from the PNEC analysis in the RAR, and are excluded also from the present analysis. However, a few results in the RAR database were not considered for the PNEC analysis because of the DOC criterion. Some of these have been included in the analysis for the reason mentioned above. In practice, this only applies to a few data on:

 

-      Daphnia magna: a few results obtained under conditions of DOC < 13mg/l are included (Heijerick et al.,2005; De Schamphelaere et al., 2005a). Since they are pooled with the numerous data entries already considered for this species, the effect of this inclusion on the geomean NOEC value for this species is limited (see below). Since moreover Daphnia is only one of 24 species in the SSD, the effect on the HC5 is insignificant.

-      For Oncorhynchus mykiss, 1 data point of the RAR database that was removed for DOC considerations was added. For this species, two entries that were in the RAR were removed because not answering the relevancy criteria of the RAR. Also for this species the effect these changes is very limited because of the numerous other data.

 

The extensive dataset on chronic aquatic toxicity in the RA (ECB 2008) was also updated with new information screened for the same criteria as those described above, for the species already figuring in the RAR (2008):

 

-      a few new data were added for D. Magna, and for O. mykiss (De Schamphelaere and Janssen, 2004; De Schamphelaere et al., 2005a). As a result, the geomean for D. Magna and O. mykiss are now 98 and 146µg/l instead of 88 and 189µg/l (RAR 2008), respectively.

-      for Pseudokircherniella subcapitata, 2 additional NOECs were obtained from a recent study (Muyssen et al 2003); as a result, the geomean changes slightly (19 µg Zn/l instead of 17 µg Zn/l in the RAR).

-      The results on Brachydanio rerio (Dave et al 1987) were not used for PNEC derivation, since the test suffered from major quality problems.

-       

In addition to the dataset of the RA, high quality chronic toxicity data were added to the dataset for six new species:Chlorellasp. (unicellular alga, Wilde et al., 2006);Daphnia longispina(cladoceran, Muyssen et al., 2003);Anuraeopsis fissaandBrachionus rubens(two rotifer species, Azuara-Garcia et al., 2006);Cottus bairdii(fish, Brinkman et al., 2005);Salmo trutta(fish, Kjallqvist et al., 2003). These data were also screened for the same criteria as those described above.

 

Chronic aquatic toxicity data freshwater – results

The 23 distinct chronic species ecotoxicity values that were used for the SSD in the present analysis are summarised below. The “species mean” NOEC values used for PNEC derivation (freshwater PNEC_{add, aquatic}), range from 19 to 530 µg/L.

 

Summary of chronic “species mean” NOEC values that are used as input values for the SSD for deriving the 5th percentile values as a basis for the freshwater PNECadd¸ aquatic. Species geomean values from the RAR (ECB 2008) that were revised are indicated in italics. New values added after the closure of the RAR database are indicated in bold.

Taxonomic groups		species		“Species mean” NOECadd values (µg/L)
Algae (unicellular)		Pseudokirchneriela subcapitata Chlorella sp.		19 48
Algae (multicellular)		Cladophora glomerata		60
Poriferans		Ephydatia fluviatilis Ephydatia muelleri Spongilus lacustris Eunapius fragilis		43 43 65 43
Molluscs		Potamopyrgus jenkinsi Dreissena polymorpha		75 400
Crustaceans		Ceriodaphnia dubia Daphnia magna Daphnia longispina Hyalella azteca		37 98 128 42
Rotifera		Anuraeopsis fissa Brachionus rubens		50 50
Insects		Chironomus tentans		137
Fish		Jordanella floridae Phoxinus phoxinus Pimephales promelas Oncorhynchus mykiss Salvelinus fontinalis Cottus bairdii Salmo trutta		44 50 78 146  530 169 112

Discussion on the SSD freshwater chronic aquatic toxicity.

 

A comparison of the database of freshwater “species mean” NOEC values with the criteria for using statistical extrapolation shows the following:

 

·        The number of species chronic NOEC entries (n = 23) meets largely the general requirement for the number of input data (minimum requirement: 10 species NOEC values, preferably more than 15 NOEC values).

·        The requirement for the number of taxonomic groups is met: Chronic species NOEC values are available for 1 unicellular algal species, 1 multicellular algal species (macro alga), 4 sponge species, 2 mollusc species, 4 crustacean species, 2 rotiferan species, 1 insect species and 7 fish species.

·        Regarding the coverage of taxonomic groups, the 8 taxonomic groups represented in the database correspond to the requirements set out in chapter R.10:

-      A fish: the database has 7 fish species

-      A second family in the phylum chordate (fish, amphibian,..): the database has 7 fish species

-      A crustacean: the database has 4 species

-      An insect: there is 1 species in the database

-      A family of a phylum other than arthropoda or chordates, e.g; rotifers, annelids, Mollusca,...: the database has 2 rotifers, 2 molluscs

-      A family in any order of insect or any other phylum not already represented: the database has 4 porifera

-      Algae: the database has 2 algae species

·        Higher plants: in the database of accepted NOEC values, data for algae are included, but data for higher plants are lacking. However, data for freshwater higher plants are included in the database of rejected NOEC values of the RAR (ECB 2008). The rejected NOEC values for higher plants are from the following studies :

-       A long-term study with four different species of freshwater higher plants (Elodea nuttallii, Callitrische platycarpa, Spirodela polyrhiza and Lemna gibba) resulted in unbounded NOEC values of > 650 µg/l for all four plant species (endpoints: survival and growth). The plants used in this study were obtained from unpolluted ditches or ponds in the Netherlands and grown in filtered ditch water with a pH of 8.0 (Van der Werff & Pruyt, 1982). The study was not used in the RAR because it had only unbounded NOECs.

-      A study with duckweed Lemna minor resulted in a NOEC of 160 µg/l (endpoint: growth) at pH 5 and hardness of 310 mg/l in artificial medium (Jenner & Janssen-Mommen, 1993). The study was not used in the RAR because the test conditions were outside the relevancy ranges for pH and hardness.

-      Tests with duckweed Lemna pauciscostata resulted in a NOEC of 5000 µg/l (endpoint: growth) at pH 4 or 5 and hardness of 700 mg/l in an artificial medium and tests in another artificial medium resulted in about 60-80% growth inhibition at 1000 µg/l at pH 6 or 7 and hardness of 120 mg/l (Nasu & Kugimoto, 1981). The study was not used in the RAR because the test conditions were outside the relevancy ranges for pH and hardness.

 

From the data for these six plant species it was however concluded in the RAR that aquatic higher plants do not appear to be very sensitive to zinc toxicity in comparison with algae or animals and thus the lack of useful NOEC values for higher plants was acceptable. Furthermore, the RAR stressed that “the database of accepted NOEC values includes a relatively high NOEC (60 µg/l) for the macro algaCladophora glomerataand macro algae resemble higher plants” (RAR 2008). 

 

It was concluded in the RAR that the taxonomic coverage requirements for applying an SSD were largely met by the RAR dataset. The present analysis has added 6 species to the one of the RAR, including an additional taxonomic group. So, a wealth of information on different species is available, and statistical extrapolation will also be used in the present analysis for PNEC freshwater derivation.

 

Statistics on the species sensitivity distribution (SSD)

Given the multitude of relevant high quality data, statistical extrapolation was used for PNEC determination. Following the RIP R.10 guidance, “different distributions may be used” for the SSD. We tested the lognormal distribution (default option), as calculated with the “ETX” software, and subsequently several other distributions with the “@Risk” software. The statistics of the curve–fitting on the chronic NOEC data are summarised below.

 

 

Summary statistics for SSD on chronic NOEC values for zinc in freshwater (N= 23).

distribution	HC5	Lower estimate on HC5	Median HC5/ lower 95% C.I.	A-D statistic	A-D significance level	K-S statistic	K-S significance level	Acceptance for PNEC setting
Lognormal (ETX)	20.6	12.3	1.68	0.89	0.01 (accepted)	0.89	p = 0.05 (accepted)	accepted
Lognormal (@risk)	20.9	Not available	Not available	0.98	0.01<= p <0.025 (accepted)	0.16	p < 0.15 (accepted)	accepted
Extreme values (@risk)	27.2	Not available	Not available	0.63	0.05<p< 0.1 (accepted)	0.16	p > 0.1 (accepted)	accepted

 

Using the Anderson-Darling (A-D)-test for normality, the default distribution (lognormal) does fit significantly at a level of 1 %. Using the Kolmogorov-Smirnov test, the lognormal is accepted at 5% level. This analysis indicates that the fit of the lognormal distribution is not very good at the lower tail of the distribution (low A-D acceptance). This outcome of the SSD-statistics is the same as the one observed for the lognormal distribution in the RAR, where it was stipulated that “the Anderson-Darling test indicated that there was only goodness-of-fit for the log-normal distribution at a low significance level (1%). The Kolmogorov-Smirnov (K-S)-test accepted both the log-normal and log-logistic distribution at a higher significance level (5%)” (RAR 2008). Since the statistical significance levels on the lognormal distribution are the same as those under which the lognormal was accepted for PNEC setting in the RAR (2008), the lognormal distribution is accepted in the present analysis also.

 

The acceptance of the lognormal distribution at the same significance level as in the RAR is expected, since the species data that were added to the distribution are quite close to the average value of the distribution (79 µg/L); they are lower in 4 cases (50, 50, 50,68), and higher in 2 cases (119,138). The relatively low goodness-of-fit is related to the lower tail of the SSD (where the input data were not changed as compared to the RAR) and where the fit with the lognormal distribution is not very good (figure 2). For the same reason, the K-S test gives acceptance at 5% level, for the revised database, too (K-S is more related to the values near the middle of the distribution).

Figure2 (see attachment). Lognormal distribution curve fitting to the freshwater chronic toxicity data for zinc (ETX graphics).

 

Other distributions fit better to the data. Of the other distributions tested with @risk, the “Extreme values” (EV)-distribution provided the best fit, as demonstrated by the lowest A-D statistic. The fit was highly significant both with the A-D and the K-S test (p values > 0.05).

 

To be consistent with the approach taken in the RAR (2008), the lognormal distribution is used to provide a basis for setting the PNEC freshwater, in spite of a better fit to the data provided by the extreme values distribution. It is noted that the HC5 resulting from the extreme values distribution is significantly higher than the HC5 calculated with the lognormal distribution (20.6 µg/l versus 27.3 µg/l).

 

It is noted also that the PNEC is a PNEC_add., i.e. to be added to the background when using monitored water concentrations.

 

The 5^th percentile value of the SSD (the HC5) is set at the 50% confidence level, using the log-normal distribution function, which results in an HC5 value of 20.6 µg Zn_added/l.

 

 

The reference values for chronic aquatic toxicity were determined:

 

-      at pH 8: from the extensive chronic ecotoxicity data available for algae, invertebrates and fish (section 7.1.1., 2.). The standard species NOEC values for each taxonomic group for which a bioavailability model is available were taken at pH 8, and the lowest of the 3 was selected as a reference value at pH 8.

-      at pH 6: the corresponding aquatic toxicity at pH 6 was calculated from the same database for the standard species for which bioavailability models were available, and the lowest of the 3 was selected as a reference value at pH 6.

 

 

The results are summarised below:

-      for algae, the NOEC of the BLM-species Pseudokircherniella subcapitata is the lowest of the SSD at pH 8 (19 µg/l). This value corresponds to a water of pH 8,0,  hardness 24 mg CaCO3 and DOC 2.0 mg/L. With the BLM, a corresponding species NOEC of 142 µg/l is calculated for this species at pH 6 (other water conditions same).

-      for invertebrates, the BLM-species Daphnia magna gives a species mean at pH 8 of 98 µg/L, corresponding to a water of pH 8, hardness 24 mg CaCO3/L and DOC 1.2 mg/L.  The Daphnia magna-BLM predicts at pH 6 (other water conditions same) a species NOEC of 82 µg/L.

-      for O. Mykiss, the species mean at pH 8 is 146 µg/L (hardness 45 mg/L, DOC 2 mg/L). Using the corresponding species BLM gives a species NOEC of 146 µg/L at pH 6 (other conditions same).

 

From this assessment, the following reference values for chronic zinc aquatic toxicity are derived:

-at pH 8.0: 19 µg Zn/L (Pseudokircherniella subcapitata)

-at pH 6.0: 82 µg Zn/L (Daphnia magna)

 

3. Aquatic chronic toxicity: marine waters

For zinc, a specific effects assessment was made and a specific PNEC was derived for the marine environment, since there is a vast dataset available on marine ecotoxicity. This specific approach is also more reflective of the toxicity of zinc in the marine environment given the different speciation and bioavailability of zinc in salt – and freshwater, and differences in physiology of saltwater organisms. Given the vast amount of available toxicity data, statistical extrapolation was used to derive the marine PNEC. This marine effects-assessment is following an added risk approach, as applied for the freshwater.

 

Establishing the dataset

Sources of data

 

This report analyses the available chronic zinc toxicity data for marine organisms. The ecotoxicological data were derived from original papers, published in peer-reviewed international journals. Literature and environmental databases, including AQUIRE (US EPA), MARITOX, ECETOC, and BIOSIS, as well as review articles covering zinc in marine waters were searched and reviewed for sources of relevant and reliable chronic toxicity data on zinc. Only original literature was used.

 

Data reliability and relevance

Selection of ecotoxicity data for quality was done according to a systematic approach as presented by Klimisch et al. 1997. Standardized tests, as prescribed by organizations such as ASTM, OECD and US EPA, are used as a reference when test methodology, performance and data treatment/reporting are considered. A detailed description of methods and conditions employed in the study should be provided. The thorough description of key requirements guarantees the high reliability (Q1) of the reported ecotoxicity data. Non-standardized test data, may also have a high reliability, but required a more thorough check on their compliance with reliability criteria before being used for deriving a PNEC. As for data relevancy, tests should be performed in media that reflect natural environmental conditions (e.g. salinity, zinc background concentrations and other abiotic conditions). A set of criteria for checking reliability and relevancy has been defined in this work and is presented below. Those criteria were used to discriminate between data accepted with restrictions (Q2) and unreliable data (Q3).

 

Test medium

Both natural or artificial sea water were accepted as test medium. In case EDTA is present in the test medium, the study is considered not reliable. Chelators other than EDTA (e.g. NTA, citrate,…) can however be added. Only the results of tests in which the organisms were exposed to zinc alone were used, thus tests with metal mixtures were not considered for this evaluation. Only the results of tests with soluble zinc salts were used.

 

Salinity

The TGD does not define the salinity range of sea water. The Water Framework Directive allows for using the saltwater EQS at salinities ≥ 5‰, which sets the limit between freshwater and brackish waters. Therefore, tests performed at salinity levels down to a value of 5 g/kg were accepted for this marine dataset.

 

Background concentrations

Adaptation to high zinc concentrations may influence the sensitivity to zinc. To ensure that test organisms are adapted to the test conditions, the culture and test conditions should be similar. In addition, organisms acclimatized to elevated background in culture media or collected at contaminated sites were not used in the analysis. Only results from unpolluted test media were used, thus excluding tests that were performed in media containing high background Zn concentrations (> 10 µg Zn/l). Low background values were not discarded. In some references, no data have been reported on the culture conditions. In those cases it was assumed that culture and test conditions were similar, as is common practice.

 

Control data

Tests were rated as not reliable if control data was not provided (Q3). Effect levels derived from toxicity tests using only one test concentration always result in unbounded and therefore not assignable data. Therefore, only the results from toxicity tests using one control and at least two zinc concentrations were retained for this evaluation. Data reporting mortality rates higher than 20% in the control were not used.

 

Test statistics

Because effect concentrations are statistically derived values, information concerning the statistics was used as a criterion for data selection. In that respect L(E)C10 values are considered as equivalent to NOEC. Studies that do not report test statistics but present trust-worthy data are rated as reliable with restrictions (Q2). Studies that do not report test statistics and do not present a dose-response analysis are rated as unreliable (Q3).

 

Test design

Data without treatment replication or with pseudo-replication were not used (Q3).

 

Type of test (duration, endpoints covered)

Only reliable endpoints from properly conducted chronic tests are being considered. Toxicological endpoints which affect the species at population level were taken into account (i.e. survival, development, reproduction and growth). Historically, chronic exposure has been defined as > 4 days for all invertebrates and fish. With respect to this assessment, the arbitrary selection of this exposure period has been reviewed in light of the sensitivity of the endpoint and the duration of the life stage under assessment. For example, early life stages tests (embryos, larvae) of 24-48 hours have been included in this assessment. Sperm cells and fertilized eggs tests of a few hours were also considered as chronic data. Indeed, abnormal development can be observed within this time frame (e.g. molluscs, echinoderms), and the continuation of these tests would derive no additional information which could provide protection for the environment. Tests performed on adults over 96-h were considered as acute toxicity tests for invertebrates and fish. For algae, and according to ASTM guidelines, data reporting growth rates < 1 in the control were not used.

 

Test species: endemic versus non-endemic species

The culture and test conditions are considered more relevant than the geographical origin of the species and the OECD guidelines recommend the use of a number of “standard” species which do not have a world-wide distribution. Moreover, using the origin of species as criterion would considerably reduce the dataset and limit the data to only a few species / taxa, which may obscure variation in sensitivity. Therefore, the geographical origin of the test species has not been used as selection criterion.

 

Measured concentrations

Zinc is a natural element with typical background levels in seawater ranging from 0.5-1 ug/L (e.g. Laane et al. 1992). Because of the importance of understanding the true exposure concentrations (including the background concentration in the culture media), any study not supported by analytical data would automatically be excluded from the high quality studies (Q1 data). Therefore, data not supported by measured concentrations are excluded from the Q1 dataset. These studies have been rated as reliable with restrictions (Q2) or not reliable (Q3), depending on the data quality as well as the availability of zinc background levels in the test media.

The results of the marine aquatic toxicity studies are expressed either as measured concentration, or usually as nominal concentration. The measured concentrations include the background zinc concentration. Because of the added risk approach, measured concentrations have been corrected for background. In case specific background concentration was not mentioned in the paper, a default background concentration of 0.5 µg zinc dissolved/L was subtracted to the total measured concentration. The nominal concentrations were used as such for the PNECadd derivation. If it is not mentioned whether the NOEC/L(E)C10 values are based on measured or nominal concentrations, they were considered as nominal concentrations.

In the reported data, zinc has been used as the test material with several salts being used as the precursor. As with other risk assessments on metals, it is generally recognized that under laboratory conditions almost all the zinc is present in the dissolved fraction, therefore these results can be regarded as being dissolved zinc concentrations.

 

Derivation of NOEC/LOEC values (methods)

The toxicological variables are estimated based on measured or nominal NOECs or EC10 values. There has to be a concentration-effect relationship. In the past, the NOEC was determined directly from the concentration-effect curve by consideration of the deviation of the control (e.g. 10%) or it was derived on the basis of ANOVA (analysis of variance) and a subordinate test (e.g. Dunett's). This method to derive the NOEC with the ANOVA is criticized. Pack et al., 1993 recommends the calculation of the ECX point as a preferable alternative. In older investigations, it may be difficult to find out how the NOEC was generated unless test reports or raw data are available. NOEC values without any information about the concentration-dependent response were excluded from PNEC analysis.

Unbounded NOEC values (i.e. no effect was found at the highest concentration tested) were not used in this analysis in accordance to the Zn RAR, 2008.

In a few cases, no NOEC or LOEC value was provided, but raw data were reported. When possible, the raw data were used to derive an EC10 using linear interpolation (LI). The EC10(LI) values were used with restrictions (Q2).

In case only a LOEC is given in the report, it was used to derive a NOEC with the following procedures according to the Zn RAR (2008):

-         LOEC ≥ 10 and < 20% effect: NOEC can be calculated as LOEC/2.

-         LOEC ≥ 20 and < 30% effect: NOEC can be calculated as LOEC/3.

-         If the effect percentage of the LOEC is higher than 30% or if it is unknown, no NOEC can be derived.

 

In aquatic toxicity the MATC (maximal acceptable toxicant concentration) is often calculated. This is the geometric mean of the NOEC and the LOEC. If in the test report only the MATC is presented, the MATC can be divided by √2 to derive a NOEC.

 

If for one species, several chronic NOEC values (from different tests) based on the same endpoint are available, these values are averaged by calculating the geometric mean, resulting in a “species mean NOEC”. If several toxicity endpoints are reported for the same species, the most sensitive one, so the lowest NOEC value is selected for PNEC derivation. The lowest value is determined on the basis of the geometric mean if more than one value for the same endpoint exists.

 

PNEC saltwater

Ecotoxicity database for zinc on species of the marine aquatic environment

The marine zinc database largely fulfils the species and taxonomic requirements for input chronic toxicity data as explained in the RIP R. 10 guidance (at least 10 species NOECs and 8 taxonomic groups). Indeed, 39 species mean NOECs based on 48 NOEC values, from 9 taxonomic groups covering three trophic levels were found to fulfil the relevancy and reliability requirements as explained by Klimisch et al. 1997. The marine zinc database includes 4 micro- and 5 macro-algae species, 4 annelid species, 6 crustacean species, 5 echinoderm species, 9 mollusc species, 1 nematod species, 1 cnidarian species and 1 fish species. The geometric mean values of the species NOECs are presented in the CSR.

The geometric mean values of the species NOECs together with their reliability scoring are presented below. Most of the effects data are ranked as reliability 2 (Q2). Most of the data were reported as nominal concentrations.

 

Geomean species NOECs of the marine zinc effects database presented by taxonomic group, species name and family name.

Taxonomic group	Species name	Family	Geomean NOECaddvalue (µg Zndiss/L)	Reliability (according to Klimisch et al. 1997)
Micro-Algae (4)	·        Asterionella japonica ·        Chaetoceros compressum ·        Nitzschia closterium ·        Skeletonema costatum	Fragilariaceae Chaetocerotaceae Bacillariaceae Skeletonemaceae	13.3 11.2 65.9 26.4	2 2 1 and 2 2
Macro-Algae (8)	·        Ascophyllum nodosum ·        Ceramium tenuicore ·        Fucus serratus ·        Fucus spiralis ·        Fucus vesiculosus ·        Macrocystis pyrifera ·        Pelvetia canaliculata ·        Ulva pertusa	Fucaceae Ceramiaceae Fucaceae Fucaceae Fucaceae Lessoniaceae Fucaceae Ulvaceae	564.8 7.8 488 613.2 100 189.7 670.8 313	2 2 2 2 2 1 2 2
Annelids (4)	·        Capitella capitata ·        Ctenodrilus serratus ·        Neanthes arenaceadontata ·        Ophryotrocha diadema	Capitellidae Ctenodrilidae Nereididae Dorvilleidae	100 100 33.3 70.7	2 2 2 2
Cnidarians (1)	·        Eirene viridula	Eirenidae	300	2
Crustaceans (6)	·        Allorchestes compressa ·        Holmesimysis costata ·        Mysidopsis bahia ·        Mysidopsis intii ·        Paragraspus quadridentatus ·        Tigriopus brevicornis	Dogielinotidae Mysidae Mysidae Mysidae Grapsidae Harpacticidae	62.5 5.6 101 101 294.5 297	2 1 2 2 2 2
Echinoderms (5)	·        Arbacia lixula ·        Asterias amurensis ·        Paracentrotus lividus ·        Sphaerechinus granularis ·        Sterechinus neumayeri	Arbaciidae Asteriidae Echinidae Toxopneustidae Echinidae	10 50 15.98 10 160	2 2 2 2 2
Molluscs (9)	·        Crassostrea cucullata ·        Crassostrea gigas ·        Crassostrea margaritacea ·        Haliotis rubra ·        Haliotis rufescens ·        Ilyanassa obsoleta ·        Mya arenaria ·        Mytilus galloprovincialis ·        Ruditapes decussatus	Ostreidae Ostreidae Ostreidae Haliotidae Haliotidae Nassariidae Myidae Mytilidae Veneridae	18.4 39.6 11.8 20.4 13.8 20.7 900 84.9 55	2 2 2 2 1 2 2 2 2
Nematods (1)	·        Monhystera disjuncta	Monhysteridae	250	2
Fish (1)	·        Clupea harengus	Clupeidae	25	2
TOTAL	39 species	27 families	39 species mean NOECs

tatistics on the species sensitivity distribution (SSD)

Given the multitude of relevant high quality toxicity data, statistical extrapolation was used for PNEC determination. As the approach taken is based on added risks, the results of the toxicity tests based on measured concentrations were corrected for background zinc concentration. Given the wealth of experimental data, no alternative method i. e. assessment factor approach was applied for the PNEC determination. 

 

Following the RIP R. 10 guidance, different distributions may be used for the SSD. Fitting of the chronic zinc toxicity data was assessed towards the log-normal frequency distribution (default distribution), obtained using the RIVM program ETX2.0 (Van Vlaardingen et al. 2004). Several distributions were subsequently calculated with the “@risk” (Palisade Inc. USA) software. The statistics of the curve-fitting on the chronic NOEC data are summarized below.

 

Summary statistics for the SSD on chronic NOEC values for zinc in saltwater (n=39).

Distribution	Median HC5	Lower estimate on HC5	Median HC5/lower 95% C. I.	A-D statistic	A-D significance level	K-S statistic	K-S significance level	Acceptance for PNEC setting
Lognormal (ETX)	6.09	3.09	2	0.49	P > 0.1	0.61	P > 0.1	Accepted
Lognormal (@risk)	6.2	Not available	Not available	0.48	P > 0.1	0.1	P > 0.1	Accepted
Best fit = Weibull (@risk)	8.5	Not available	Not available	0.44	N/A	0.08	N/A	Accepted

 

 

The goodness of fit tests reports a measure of the deviation of the fitted distribution from the input data. Preference is given to Anderson-Darling (A-D) test because it highlights differences between the tails of the fitted distribution and input data. The Kolmogorov-Smirnov (K-S) can also be used for goodness of fit purposes but it does not detect tail discrepancies very well.

 

Using the A-D test for normality, the lognormal distribution does fit significantly at all levels. The lognormal distribution is also accepted at all levels using the K-S test.The lognormal distribution results in an HC5 at 50% confidence value (HC5 5% - 95% confidence values) of 6.09 (3.09 - 10.26) µg Zn/L . The observed Q1-2 data are presented on log-scale together with their fitted normal distribution curve in figure below.

Figure 3 (see attachment). Cumulative distribution of the 39 species mean NOEC values from Zn toxicity tests in the marine organisms database. Observed data and normal distribution curve fitted on the data.

 

 

Other distributions fit better to the data. Using @risk, The Weibull distribution turned out to be the best fit with an HC5-50 value of 8.5 µg Zn/L, as indicated by the lowest A-D statistic. The difference between both distributions is however minimal.

 

To be conform with the approach taken in the Zn RAR 2008, and because the lognormal distribution provides a significant fit to the data, it is the lognormal distribution which was used to provide a basis for setting the PNEC saltwater, in spite of a better fit with the Weibull statistical distribution.

 

It is noted also that the PNEC is a PNECadd., i.e. the background concentration needs to be considered in the compliance assessment exercise. The 5^thpercentile value of the SSD (the HC5), set at 50% confidence value, using the lognormal distribution (ETX 2.0) function, results in a value of6.09 µg zinc_added/L.

 

 

 

Mesocosm studies

 

Davies and Sleep (1979) investigated the influence of zinc upon carbon fixation rates of natural phytoplankton communities present in the English Channel (near Plymouth). A series of three samples of varying biological composition (100% diatoms, 60% dinoflagellates-40% diatoms, and 60% diatoms-40%dinoflagellates) were taken in July at the same location and the tests with zinc were done in natural sea water. The zinc background concentration in seawater varied from 0.4 to 7.6 µg/L. A pre-incubation period of phytoplankton assemblages to zinc levels was designed in order to equilibrate the populations with the experimentally added zinc before measuring their carbon fixation rates. Fixation rates less than 90% of the mean control value were attributed to inhibition caused by the presence of zinc. The lowest concentrations of zinc which caused detectable inhibition of carbon fixation, i.e. rates lower than 90% of the mean control values, were in the range of 10 to 15 µg/L. From the dose-response curves, the EC10 levels were in the range of 7 to 13 µg/L. This means that the HC5 derived from the normal distribution of the log-transformed data is protective for those phytoplankton assemblages. When looking at single species toxicity tests, a number of the diatoms and dinoflagellates are among the more sensitive species in the chronic NOEC database (e.g.Asterionella japonica, Chaetoceros compressum, ...). Also in the freshwater database, unicellular algae and algal communities were among the more sensitive organisms. Davies and Sleep (1979) were already mentioned in the RAR (2008) and used among other references to derive a provisional PNEC saltwater.

 

Another study on phytoplankton communities from three different areas was carried out by Wolteret al. (1984). Influence of metal to carbon fixation rate of phytoplankton and to glucose incorporation by bacteria was determined. Water samples coming from Kiel Fjord, North Sea and a coastal area of the North Atlantic Ocean were investigated after addition of varying concentrations of metals including zinc. Surface samples were taken in Kiel Fjord during the spring and autumn plankton bloom. Zinc was added to subsamples to give concentrations in the range of 4.3 to 304.3 µg/L. The North Sea and Atlantic samples were collected during a cruise in 1981. The added zinc concentrations were 0.29 to 1.45 µg/L. The added metal concentrations were lower than in the Kiel Fjord experiments due to lower concentrations in the water compared to Baltic Sea water. In all cases the samples contained mixed phytoplankton populations which were dominated by diatoms.

From the graph and the text, zinc reduced plankton activity in the Kiel Fjord samples only at concentrations above 100 µg/L. Moreover, carbon fixation measurements carried out four and 24 hours after metal addition to the North Sea and Atlantic samples were not reduced at any test concentration, which then gives unbounded values. The same was true for bacterial glucose incorporation.

 

Results of field experiments made on phytoplankton communities coming from various natural sea waters

	Location			Endpoint
	Kiel Fjord	North Sea	Atlantic
NOEC (µg Zn/L) after 4 or 24 hours exposure	>100	>1.45 (unbounded)	>1.45 (unbounded)	C fixation rate
NOEC (µg Zn/L) after 4 or 24 hours exposure	/	>1.45 (unbounded)	>1.45 (unbounded)	Bacterial glucose incorporation

Although this study has nothing inherently wrong in the design, it has its limitations since important details are lacking (no accurate information about test organisms and test conditions, no information on statistics and on control data, …). The unbound NOEC results in North Sea and Atlantic are not useful. The results should be interpreted with care, but notably, the results obtained in Kiel fjord suggest that the HC5 level described above would be protective.

 

After the registration of the substance in 2010, a outdoor state-of-the-art marine mesocosm study was perfomed on dissolved zinc, and results were reported recently (Foekema et al 2012). In this study, an ecosystem assembly containing a.o. phytoplankton, periphyton, macro algae, zooplankton, molluscs, and annelids, was exposed to dissolved zinc concentrations between 2.7µg/l and 91µg/l, maintained at constant level by regular monitoring and spiking.

The dataset showed a consistent picture of the impact of continuous exposure of dissolved zinc on the mesocosm community during the 83 days of the test (figure 4 below). The No observed ecological adverse effect concentration (NOEAEC) resulting from the study was 12 µg Zn/l. At treatment levels 5.6, 7.5, and 12 µg Zn/l, no negative effects were observed, but indications were found of a slight stimulation of the productivity, that might be related to zinc essentiality. Since this did not result in substantial changes in the composition or functioning of the mesocosm ecosystem these effects were not considered as being ecologically adverse (Foekema et al 2012).

 

Figure 4 (see attachment). Overview of endpoints with statistical significant differences between treated and untreated (2.7 µg Zn/L) mesocosms. Effect class “A”: no effects, “B”: slight effects, “C” pronounced temporal effects, “D”: pronounced durable effects. A positive or negative difference from the control is indicated with “+” or “-“, respectively (taken from Foekema et al 2012).