Boron

Ecotoxicological Summary

Hazard for aquatic organisms

Freshwater

Hazard assessment conclusion:

PNEC aqua (freshwater)

PNEC value:

2.9 mg/L

Assessment factor:

2

Extrapolation method:

sensitivity distribution

PNEC freshwater (intermittent releases):

13.7 mg/L

Marine water

Hazard assessment conclusion:

PNEC aqua (marine water)

PNEC value:

2.9 mg/L

Assessment factor:

2

Extrapolation method:

sensitivity distribution

STP

Hazard assessment conclusion:

PNEC STP

PNEC value:

10 mg/L

Assessment factor:

1

Extrapolation method:

assessment factor

Sediment (freshwater)

Hazard assessment conclusion:

no exposure of sediment expected

Sediment (marine water)

Hazard assessment conclusion:

no exposure of sediment expected

Hazard for air

Air

Hazard assessment conclusion:

no hazard identified

Hazard for terrestrial organisms

Soil

Hazard assessment conclusion:

PNEC soil

PNEC value:

5.7 mg/kg soil dw

Assessment factor:

2

Extrapolation method:

sensitivity distribution

Hazard for predators

Secondary poisoning

Hazard assessment conclusion:

no potential for bioaccumulation

Additional information

Boron is a naturally occurring element. In order to deal with the presence of a natural background, various concepts have been developed, such as the Added Risk approach (Added RA) and the Total Risk approach (Total RA) concepts. In essence the Added RA assumes that species are fully adapted to the natural background concentration and therefore that only the anthropogenic added fraction should be regulated or controlled. The Total RA assumes that “exposure” and “effects” should be compared on the combination of the natural background and the added anthropogenic concentrations. Because of high natural background concentration of B in aquatic environments (especially in the marine compartment), the derived PNEC will be close to the natural background and therefore the added risk approach is chosen for the aquatic environment. For the soil compartment, the available data points towards a significant difference in bioavailability between the boron element naturally present in soils and added soluble B. Therefore, it was decided to use the added risk concept for assessing the risks of boron added to soils.

 

The added risk approach used in this CSR is following the recommendations of the MERAG and the REACH Guidance Document (2008)‚ Appendix R.7.13-2: Environmental risk assessment for metals and metal compounds.

 

 

Natural background concentrations of boron

The elemental boron is present in the form of water soluble borates in all inland freshwaters at concentrations of typically <1 mg/L, because of the weathering of naturally occurring borate containing rocks and soils, supplemented by contributions from anthropogenic activities. A median value of 15.6 µg/L is put forward as a typical baseline concentration for boron in European surface waters (Europe-regional scale) (source: FOREGS database). However, by far the most predominant and widespread natural source of boron is in seawater; the average concentration of borates in all oceans is 4,6 mg B/L water, but can vary from 0,5 mg/L in the Baltic Sea to 9,6 mg/L in the Mediterranean Sea (Ecetoc, 1997).

Parent material is considered a dominant factor affecting supply of B from the soil. In general, soils derived from igneous rocks, and those in tropical and temperate regions of the world, have much lower B concentrations than soils derived from (marine) sedimentary rocks, and those in arid or semi arid regions (Ho, 2000). The GEMAS data provide concentrations of Boron in grasslands spread over Europe; B-levels ranged between <0.5 and 40.84 mg/kg with 50th/90th percentiles of 2.58 and 8.33 mg B/kg.

 

Additional information on boron background levels can be found in the Background Document “Environmental effects assessment of boron”, which is attached in the technical dossier in IUCLID Section 13.

 

 

Chemistry of boron in the environment

Boron may be considered a typical metalloid having properties that are situated between the metals and the electronegative non-metals. Boron has a tendency to form anionic rather than cationic complexes (Keren and Bingham, 1985). Boron does not undergo oxidation reduction reactions or volatilisation reactions in soils (Goldberg, 1997). Boron chemistry is of covalent boron compounds and not of B3+ ions because of its very high ionisation potentials.

 

Boron oxide, B2O3 reacts with water to form boric acid, H3BO3. Boric acid is moderately soluble (4.9g 100/mL water at 20°C). It acts as a weak Lewis acid by accepting a hydroxyl ion to form the borate anion.

Aqueous boron species other than B(OH)3 and B(OH)4- can be ignored for most practical purposes in soils (Keren and Bingham, 1985). In most soils with soil solutions in the pH range 4.0 to 9.0, the uncharged B(OH)3 predominates. The borate ion is expected to form a variety of complex salts with suitable metal acceptor ions. However, there is relatively little evidence for the existence of metal borate complexes in solutions. Among the organic borates, the tendency is for boron to replace carbon or nitrogen in three-fold coordination (Keren and Bingham, 1985). In regions of low rainfall, the boron content of the soil is usually high. Boron in these soils probably exists largely as sodium-calcium borates. However, there is no information on the kinetics of dissolution of these minerals in water or on the composition of their products (Keren and Bingham, 1985). The transformation dissolution test, although not designed and very much the worst case for soil and pore water, indicates that such kinetics must be very slow and for soil probably much lower than the dilution in groundwater.

 

Freshwater compartment

The available ecotoxicity database for the effect of boron on freshwater organisms is large. Therefore, the use of the statistical extrapolation method is preferred for PNEC derivation rather than the use of an assessment factor on the lowest NOEC, as specified by the Guidance document on information requirements and chemical safety assessment Chapter R.10.3.1.3. The PNEC is based on the 50% confidence value of the 5th percentile value of the long-term effect NOEC/EC10 data (HC5,50%) and an additional assessment factor taking into account the uncertainty on the HC5,50% (thus PNEC = HC5,50%/AF). The advantage of this statistical extrapolation method is that it uses the whole sensitivity distribution of species in a collection of laboratory test data to derive a PNEC instead of taking only the lowest long-term NOEC.

 

The added LC10/EC10 or NOEC values are obtained by substracting the background B-concentrations as observed in the culture/test media (= control) from the measured LC10/EC10 or NOEC values. Derivation of a geometric mean values could be considered relevant if there was more than one set of data on the same species whith comparable endpoint, duration, life stage and testing conditions. The approach followed for determination of species-specific ecotoxicity reference values is outlined in the Background Document “Environmental effects assessment of boron”, which is attached in the technical dossier in IUCLID Section 13.

 

Seventeen reliable long-term no-effects data points for the freshwater compartment were used for the construction of a Species Sensitivity Distribution (SSD) from which a median 5th percentile was derived. This value represents the HC5,50% with 5%-95%-confidence interval. The confidence interval is calculated using a Monte Carlo analysis on the log-normal distribution that was fitted through the seventeen data points. The outcome of this analysis allows the derivation of the HC5,50% with 5%-95% confidence interval, and this value should be used for PNEC-derivation (i. e., PNECaquatic= HC5,50%/ Assessment Factor). A detailed overview of the data and calculations is provided in the Background Document “Environmental effects assessment of boron”, which is attached in the technical dossier in IUCLID Section 13. The added HC5,50% (± 95%CL) that was associated with this SSD was 5.7 mg B/L (95%CL: 3.2 – 8.3 mg B/L).  

 

The assessment factor that is applied on this HC5,50% for calculating the added PNEC for boron in the freshwater aquatic environment depends on the uncertainty analysis of the dataset. For uncertainty considerations the London Workshop 2001 recommended, for the freshwater compartment, to apply an additional assessment factor on the 50% confidence value of the 5th percentile value (thus PNECaquatic = HC5,50%/AF), with an AF between 1 and 5, to be judged on a case-by-case basis. Based on the available chronic NOEC data, the following points were considered when determining the size of the assessment factor:

 

1. The overall quality of the database and the end-points covered, e.g. if all the data are generated from “true” chronic studies using relevant endpoints, representativity of the physico-chemistry of the test media;
2. The diversity and representativeness of the taxonomic groups covered by the database;
3. Statistical uncertainties around the 5th percentile estimate, e.g. reflected in the goodness-of-fit or the size of confidence interval around the 5th percentile;
4. Comparisons between field and mesocosm studies and the 5th percentile and mesocosm/field studies to evaluate the laboratory to field extrapolation;
5. Comparison of the HC5,50% with unbounded NOEC-values.

 The boron database covered the ecologically relevant endpoints including mortality, growth, reproduction and condition. For all trophic levels, sensitive life stages and reasonably long term exposures were achieved. The B-database also fulfills the recommendations of at least 10-15 different NOEC values. Data for seventeen different species are used for calculating the species sensitivity distribution, and covers the eight different taxonomic groups that should be included in the effects database (as defined by The London Workshop (2001)).

 

The Normal distribution function (as recommended by Aldenberg and Jaworska, 2000) on log-transformed toxicity data using the ETx software (RIVM) calculated an added HC5,50% value of 5.7 mg B/L, with confidence limits ranging between 3.2 and 8.3 mg B/L. The best fitting distribution using the Andersen-Darling goodness-of-fit statistics using the @Risk software was the Logistic distribution (on log-transformed toxicity data), resulting in an added HC5,50% value of 4.6 mg B/L, with confidence limits ranging between 1.9 and 6.2 mg B/L. The larger difference between the 5th and the 95th % confidence level for the Logistic distribution (factor of 3.3) compared to the Normal distribution (factor of 2.6) reflects the higher statistical uncertainties around the 5th percentile estimate using the Logistic distribution. It must be emphasized that both software, i.e. ETx and @Risk uses a different approach to account for the sampling uncertainty.

The probability distribution of the boron dataset used for the calculations of the 5th percentile values has been checked with the Anderson-Darling (A/D) and Kolmogorov-Smirnov (K/S) goodness-of-fit tests. The A/D goodness-of-fit test highlights differences between the tail of the distribution (lower tail is the region of interest) and the input data, while the K/S test focuses on differences in the middle of the distribution and is not very sensitive to discrepancies of fit in the tail of the distribution. Based on this analysis, a better fit of the log-transformed data was achieved with the Logistic distribution function, while a better fit was obtained with the Normal distribution using the K/S goodness-of-fit statistics. However, in all cases the curve fitting functions (best fitting Logistic and Normal) fit reasonably well to the long term toxicity SSD data and none of the fit functions can be rejected at the 5 % significance level. It was concluded that the Normal distribution using ETx (RIVM) should be used for PNEC derivation.

 

No reliable experimental mesocosm data are available. However, several field studies have been reported in freshwater locations experiencing high boron concentrations, either seasonally or within a relatively small geographic range. These field studies focused on trout because early laboratory reports (Birge and Black, 1993) suggested this species was the most sensitive and would likely be affected by concentrations of 1 mg B/L or greater. These studies showed a successful establishment of a trout population, abruptly introduced into a high boron environment (Argentina), and suggests that aquatic organisms, even those considered sensitive, can tolerate such conditions.

 

 

In conclusion on the subject of the choice of the assessment factor and considering all arguments above for the derivation of the HC5,50 it is felt that the most appropriate AF would be 2.

The assessment factor 2 takes into account:

1. The extensive database of long term boron effects to freshwater organisms covering a representative range of plant, invertebrate and vertebrate species;.
2. The extensive database of long term boron effects data to freshwater organisms fulfils the requirements related to the taxonomic groups (families) mentioned in the ECHA Guidance (R.10.3.1.3).
3. The absence of a field or mesocosm study to evaluate the laboratory to field extrapolation. However, field studies have been reported which make use of naturally occurring waters with high boron concentrations. In several cases these fluctuate during the year, so they do provide a means to demonstrate that levels of boron of about 1 mg B/L and higher do not adversely affect local species, including the sensitive species rainbow trout.
4. The demonstration of a suitable probability distribution (i.e. the normal distribution on the log-transformed toxicity data) of the long-term boron dataset for the calculation of 5th percentile values.
5. The conclusion that no key species or group of species is below the HC5,50.

 Therefore, the proposed freshwater PNECadd,freshwater is based on the best fitting HC5,50% value using an AF of 2, i.e. 2,9 mg B/L. This value, based on added boron concentrations, is therefore taken forward to the risk characterization.

 

An extensive discussion on the derivation of the HC5,50 and associated AF is provided in the Background Document “Environmental effects assessment of boron”, which is attached in the technical dossier in IUCLID Section 13.

 

 

Freshwater compartment – intermittent release

According to REACH Guidance R.16.2.3 if intermittent release is identified, only short-term effects are considered for the aquatic ecosystem and no-effect levels are derived from short-term toxicity data only. Instead of using an assessment factor, it was opted to consider all available short-term toxicity data for freshwater organisms in order to derive the short term PNEC for intermittent release. The statistical extrapolation approach using all relevant/reliable short-term toxicity data was used to construct a ‘short term’ species sensitivity distribution and to derive a HC5,50% value for short term toxicity.

 

The database of short term boron effects to freshwater organisms represents 46 high quality studies among 20 different species. Acute study endpoints (L(E)C50) are available for 2 unicellular green algal species (Chlorella pyrenoidosa, Selenstrum capricornutum), 3 crustacean species (Ceriodaphnia dubia, Daphnia magna, Hyalella azteca), 2 insect species (Chironomus decorus, Allocaphnia vivipara), 4 mollusc species (Lampsilis siliquoidea, Leguma recta, Sphaerium simile, Megalonaisa nervosa), and 7 fish species (Pimephales promelas, Oncorhynchus kisutch Oncorhynchus tschawytscha, Catostomus latipinnis, Xyrauchen texanus, Ptychocheilus lucius, Gila elegans).

 

Using the above listed acute toxicity data, a species sensitivity distribution (SSD) and HC5,50% for short term toxicity have been calculated using best fitting approaches. The best fitting curve was the normal distribution function on the log-transformed short-term toxicity data. The ETX software (RIVM), using the normal distribution on the log-transformed short-term freshwater toxicity data, resulted in a short term HC5,50% value of 27.3 mg B/L (confidence limits between 12.0 and 47.8)

 

The short term HC5,50% value from the log-normal statistical approach is used as the basis for the PNEC calculation. A similar assessment factor as used in the derivation of the long term PNEC (i.e. AF = 2) was applied.

 

 

Marine water compartment

In general, there exist very limited data on acute and long term toxicity of boron to marine organisms. Three fish species (Limanda limanda, Menidia beryllina and Cyprinodon vareigatus), two crusteacean species (Americamysis bahia and Litopenaeus vannamei) are represented, one echinoderm species (Anthocidaris crassipina), and 19 algal species are represented. However, most of these toxicity data are based on nominal value and cannot therefore be considered for PNEC derivation. Marine waters contain about 5 mg B/L, so it may be expected that marine organisms are more tolerant of boron than freshwater organisms. However, the lack of a suitable database prevents direct evaluation of this expectation.

The available studies suggest that boron toxicity may vary with salinity. Li et al (2007) reported acute toxicity to the white tiger shrimp Litopeneaus vannei, to be 25 mg B/L at 3‰ and 80 mg B/L at 20‰. Thompson et al. (1976) found underyearling coho salmon more susceptible to boron toxicity (12-d LC50 of 12.2 mg B/L) in seawater (28 ‰) than in freshwater (12-d LC50 of 113 mg B/L). However, under natural conditions, coho underyearlings remain in freshwaters, so the salinity itself may have contributed to physiological stress. Similarly, Litopeneaus are reported to show optimal growth at about 20‰. Pillard et al. (2002) measured borate ion toxicity to the mysid shrimp, Americamysis bahia and reported acute values of 310, 290 and 380 mg B4O7-2/L for salinities of 10‰, 20‰, and 31‰, respectively. They concluded however, that the differences in salinity had no distinct impact on the tolerance to borate. More recent study reports from Hicks (2011) have also shown similar NOECs for a 28d test. The most sensitive endpoint was reproduction and gave a NOEC of 16.6 and 18.6 mg B/L for a salinity of respectively 8 and 20 ‰.

Addressing the concern about differential toxicity of boron as a function of salinity would require research data for organisms that occur in areas of varying salinity, to avoid the confounding factor of salinity itself being a stress on many organisms. Because of the very limited data for marine species, it is proposed that the marine PNEC is the same value as proposed for freshwater organisms. This represents a conservative proposal, given the naturally higher boron concentrations in the marine environment.

 

Based on the PNECadd,freshwater of 5.7 mg B/L derived with an SSD approach and an assessment factor of 2, it can be assumed that the PNECadd, freshwater also protects the marine/brackish environment (open sea, coastal, estuaries). Therefore, the PNECadd,marine water of 2.9 mg B/L is taken forward to the risk characterization.

 

Sediment compartment

The weight of evidence provided by the lack of partitioning (Kd estimates) and the results of the water only/whole sediment toxicity tests indicate that it is unlikely that boron will exert toxic effects via the sediment compartment and that the derivation of a PNEC sediment is not warranted for boron and could waived based on exposure considerations.

 

Detailed justification for this approach is provided in the Background Document “Environmental effects assessment of boron”, which is attached in the technical dossier in IUCLID Section 13.

 

 

Soil compartment

To estimate the HC5,50 value from the terrestrial toxicity data, the statistical extrapolation method as described by Aldenberg & Jaworska (2000) was used for calculating the median fifth percentile (HC5,50) of the best fitting distribution curve. As the available ecotoxicity database for the PNEC derivation is large, the use of the statistical extrapolation method is indeed preferred for PNEC derivation rather than the use of an assessment factor on the lowest NOEC, as specified by the Guidance document on information requirements and chemical safety assessment Chapter R.10.3.1.3.

The lognormal distribution was selected as the best fitting distribution based on the Anderson-Darling test (Figure 6). The HC5 at the 50th % confidence limit (together with 5th and 95th confidence limits) derived from the lognormal distribution, is 11,3 (8.8 – 13.5) mg B/kg (based on added boron concentrations).

 

The ECHA Guidance document recommends that the PNEC be derived using an assessment factor combined with the 50% confidence value of the 5th percentile value of the SSD (thus PNEC = HC5,50/AF), with an AF between 1 and 5, to be judged on a case-by-case basis. Based on the available data, the following criteria have to be considered when determining the size of the assessment factor:

1. The overall quality of the database and the end-points covered, e.g. if all the data are generated from “true” long term studies;
2. The diversity and representativeness of the taxonomic groups covered by the database;
3. Statistical uncertainties around the 5th percentile estimate, e.g. reflected in the goodness-of-fit or the size of confidence interval around the 5th percentile;
4. Evaluation of NOEC values below the HC5,50
5. Comparisons between field/microcosm studies and the 5th percentile to evaluate the laboratory to field extrapolation.

True long term and chronic data are available for multiple endpoints for plant species, invertebrate species and microbial processes, and for a range of soil types. The overall quality of the database can therefore be considered very high. Secondly, the database is composed of plant, invertebrate and microbial data with most data for the former. The overall quality of the database is considered close to optimal, but may be limited by a lack of reliable toxicity data for microbial C transformation in soil. With regard to the soil data treatment, the lognormal distribution has been selected as the best fitting distribution for derivation of the HC5,50. Only 1 species mean NOEC/EC10 value was found below the HC5,50 value (but was within a factor 2). Finally, no reliable field studies for the effect of boron on terrestrial organisms were available. Some field data for agricultural plant species give conflicting results, with indications for both boron deficiency and boron toxicity at the same soil boron concentration, depending on the plant species.

 

Each of these critera are discussed more in depth in the Background Document “Environmental effects assessment of boron”, which is attached in the technical dossier in IUCLID Section 13.

 

Because of the large difference in bioavailability between boron naturally present in soils and added soluble B, assessment of the risks of added soluble boron are based on added boron (i.e. the natural background was substracted in case natural soils were selected for the toxicity tests). Based on the above uncertainty analysis and the availability of studies on the effects of soil types and ageing on the toxicity of boron to soil organisms, it was decided to apply an Assessment Factor of 2 on the HC5,50,added of 11.3 mg B/kg dw, resulting in a PNECadded for the soil compartment of 5.7 mg B/kg dw.

 

The assessment factor of 2 takes into account:

1. the extensive database of long term boron effects on terrestrial organisms covering a representative range of plant and invertebrate species, microbial processes and soil conditions for Europe;
2. that there are no indications that there are some sensitive species or taxonomic groups lacking in the database;
3. the good fit of the lognormal distribution to the dataset;
4. the presence of some NOEC/EC10 values below the HC5,50;
5. the lack of good field validation;
6. the small gap between boron deficiency and toxicity for plants; and
7. information on the relatively limited effect of soil properties and ageing reactions on boron toxicity in soil.

STP-compartment

Taking into account the extensive information on the toxicity of boron to microorganism in STP, i. e.

• 4 respiration studies using activated sludge, showing NOEC between 17.5 and >10000 mg B/L,
• 1 nitrification study in a fixed bed reactor, showing no effect up to 500 mg B/L, and
• 3 studies of growth of protozoans relevant for the functioning of STPs, show NOEC data between 10 and 20 mg B/L.

Therefore, it was decided to base the PNECstp on the lowest NOEC from the available toxicity data. The lowest NOEC for the protozoan Opercularia bimarginata on growth is 10 mg B/L (Guhl. 2000). Since the studies of STP functioning showed no adverse effects at higher concentrations, no additional assessment factor was used, therefore resulting in a PNECstp of 10 mg B/L.

Conclusion on classification

Short term testing and hazard ID

A short term Transformation Dissolution test (TDp) was conducted by CIMM in 2011 on a 100 mg/l sample of Amorphous Boron at pH 6 and 8 to determine the potential for short term toxicity. The dissolution maximum was reached very fast (in terms of minutes) and all 24h B values were situated between 90-100 μg/l almost independent of the pH of the test medium.

While these results were very low, they are higher than what was expected to dissolve using thermodynamic calculations, questioning if either the Borosilicate glass would have released some B, or if the filter technique was appropriate.

The following ecotoxicity reference values for the soluble Boron were identified:

- lowest acute ecotoxicity reference value of 52.4 mg B/L for Pseudokirchneriella subcapitata

- and a long term ERV : 3.5 mg/L for Ictalurus punctatus

This allowed deriving an acute hazard ID for the environment for amorphous Boron powder:

• EU-CLP: none, because expected 7d dissolution equilibrium at 1 mg/L being 0,001 mg/l acute ERV of 54,2 mg/L
• GHS or DSD: none, because expected 7d dissolution equilibrium at 100 mg/L being 0,1 mg/l requires acute ERV of 52,4 mg/L

Chronic dissolution and hazard ID

The CLP requires also an assessment of the potential chronic environmental hazard being conducted at 28d and 1 mg/l and the pH that releases the most, for any substance which could be “classifiable in the 2 soluble form”. This is the case for the soluble Boron salts therefore requiring such Chronic test for Boron metal, to avoid a default classification as Chronic 4.

ECTX was requested to conduct this test on the sample base for both B and FeB. The protocol was equal to the one applied by CIMM with as only difference the finer filtration level (0,1μm being used instead of 0,2 μm) (but same technique).

The results for Amorphous Boron powder demonstrated high consistency between the samples, at pH 6 and 8 and over time; all showing dissolution rates 5 μg B/l (corresponding with the detection limit). The results are consequently significantly lower than those achieved by CIMM and in line with thermodynamic calculations, which (may) confirm(s) the need for refined filter techniques being applied, as conducted in this test.

The results allowed deriving a chronic hazard ID for the environment at pH 6 (expressing highest toxicity) for amorphous Boron powder:

• EU-DSD: none, because the 28d dissolution equilibrium at 1 mg B / L loading, being 0,005 mg/L chronic ERV of 3.5 mg/l
• GHS and EU-CLP: none, because the 28d dissolution equilibrium at 1 mg/lL being 0,005 mg/l below the chronic ERV of 3.5 mg/L.

The low Kd value for B (less than 2) does not justify a correction for potential rapid degradability, measured as the removal rate from the water column. The Transformation dissolution testing of Amorphous Boron powder is therefore conclusive to derive a NO CLASSIFICATION for the ENVIRONMENT (acute and Chronic) under DSD, CLP and GHS.