Terrestrial data
8 Discussion and recommendations
8.1 Introduction
In this chapter different aspects within the procedure for deriving risk limits on which Intervention Values can be based will be discussed and commented on, starting with the influence of several starting points on the derived risk limits (section 8.2). Next, some considerations will be given to the differences in the risk assessment for soil and aquatic sediment (section 8.3). Section 8.4 discusses the proposed procedure for deriving Intervention Values for Groundwater and section 8.5 and 8.6 some issues concerning the human
toxicological and the ecotoxicological risk assessment, respectively. Some considerations on the partition coefficient for soil and sediment are given in section 8.7, followed by some remarks on the use of sum values (section 8.8), soil type correction (section 8.9) and uncertainty of the derived risk limits (section 8.10). Finally, recommendations for further research and for policy are summarised in section 8.11.
8.2 Influence of starting points on the derived SRCs
8.2.1 Introduction
As described in section 1.3, several starting points are used for the derivation of SRCs. These starting points are often a consequence of the policy framework in which Intervention Values are used or related to the daily practise of soil and groundwater quality assessment. Some starting points are widely accepted and others are more subjective and can be discussed. Some of the starting points are discussed in the following sections, focussing on their influence on the level of the SRCs. A summary can be found in Figure 8.1.
8.2.2 Realistic case
The choice for parameters used for underpinning the SRCeco or SRChuman is based on a
realistic, “average” situation if possible. A -political- reason for this choice is that a consistent use of “worst case” parameters will lead to low values and an accompanying early
qualification as “seriously contaminated” soil, sediment or groundwater. In the subsequent determination of site-specific actual risks, often the predicate “not urgent” would be given as a consequence. A substantial discrepancy between generic and site-specific risk assessment is not considered desirable.
The parameters used in the derivation of SRCs all have uncertainty margins, the choice for an average situation or the more conservative choice for a worst case situation influencing the level of SRCs. In order to obtain insight into the consequences of worst case vs. realistic case, a Monte Carlo analysis should be performed. Therefore the uncertainty distributions of all underlying parameters must be known. For the most important parameters these distributions can be derived from the parameters given by Otte et al. (2001).
Deviations from realistic case
The amount of homegrown vegetables (fraction of 0.1) and the material LDPE as drinking water pipeline in CSOIL, as well as consumption of fish by anglers in SEDISOIL, do not reflect an average Dutch situation. The selected scenario of the consumption of 2 litre groundwater per day for deriving a risk limit for groundwater also does not reflect an average Dutch situation. These parameters are estimated as being more conservative because it is found important to protect also the situations in which a limited group has a higher exposure, due to a different behaviour, and in this way these potential risks can be recognised. Besides, it can be stated that the quality of soil/groundwater should be such that these uses are possible.
8.2.3 Background exposure
General
It was decided not to include background exposure of humans via other compartments than
(in)directly via the soil in the SRChuman (see section 1.4.6). As a consequence, the real human
exposure will be higher than the modelled exposure resulting in higher risks. The difference between the total and modelled exposure varies per specific situation and per compound. This information could be taken into account in the actual site-specific risk assessment, although currently it is not considered in the procedure for remediation urgency. In the risk assessment in several other countries, an average background exposure is taken into account (e.g.
Ferguson et al., 2000), and this principle was also applied for the derivation of the Dutch soil- use specific remediation objectives (Lijzen et al., 1999b).
Compounds for which background exposure forms a significant part of the Maximal Permissible Risk (MPR)
Along with the evaluation of the MPR (chapter 4, Baars et al., 2001), more data on the background exposure have come available (see Appendix 9). For metals, the contribution of background exposure is substantial, especially for barium, cadmium, lead, molybdenum and zinc. For thiocyanate, the estimated background exposure exceeds the TDI almost 7-fold. However, the kinetics and availability during uptake in the gastro-intestinal tract of naturally occurring thiocyanate in plants -contributing a large amount to the background exposure- might not be comparable to the uptake of the thiocyanate added during the toxicity
experiments on which the TDI was based. For dioxins and dioxin-like compounds, including the non-planar PCBs, background exposure via food almost equals the -ultimate- TDI (of 4
pg.kg-1bw.d-1). For β- and γ-hexachlorocyclohexane, there are also indications that a major part
of the TDI is filled by background exposure. For one fraction of aliphatic hydrocarbons (C7- C12) background exposure leads to a 5-fold exceedance of the TDI. For these C7-C12 compounds an exposure during painting was assumed, which will need further study. For all mentioned compounds, including the background exposure, the result will be an SRChuman which is significantly lower than proposed in this report.
8.2.4 Human exposure scenario
The Intervention Value is a generic value and applies to soils having various uses. The
SRChuman for soil is based upon the standard scenario “residential with garden” which includes
several exposure routes (see section 1.4.4). Situations will occur where exposure is higher or lower than in the standard scenario. If the actual exposure is lower than modelled for the standard scenario via CSOIL, this will result in the qualification “seriously contaminated but no urgency for remediation”. If, on the contrary, the actual exposure is higher than in the standard scenario (and risks are higher), this situation might not be recognised after the qualification “lightly contaminated” is given. The eventual risk in these situations should be recognised, and should be judged accordingly. Examples of these situations are allotment gardens where vegetables can cover more than the standard 10% of the total crops consumed or where the soil-user consumes meat, dairy products or eggs from animals living on the contaminated site. Another example is a situation where the depth of volatile contaminants (or groundwater table) is less than 1.25 m below the soil surface.
The choice to base the SRChuman on a lifetime exposure of 70 years, of which 6 are in
childhood (with a higher exposure to soil), has not been revised. The fact that Dutch life expectancy is currently higher than 70 years for men and women would only influence the relative importance of exposure of adults versus children. The choice for assessment of lifetime exposure leads to relatively higher exposure during childhood (and temporary
exceedance of the MPR). Only in the case of lead, is the risk assessment based on children as the most vulnerable group for lead exposure. It was recommended to consider focusing on
children or the foetus when critical for other contaminants as well (TCB, 1999b), but there were no direct indications that children or foetuses are more sensitive for other contaminants. Standard, a factor of 10 is included for intra-species variation (see chapter 4), by which also the more vulnerable humans should be considered protected.
Besides, the 6 years of “childhood” is important for the duration of higher amount of soil ingestion during childhood. Available studies deal mostly with children up to 5 years old (Calabrese et al., 1989; Van Wijnen et al., 1990) or up to 7 years old (Davis et al., 1990). The information for setting the period for the higher soil ingestion at 6 years is therefore limited; this period can be longer.
8.2.5 Exposure routes considered for ecosystems
Only direct exposure and no bio-magnification in the food chain is included in the SRCeco.
Reason for this choice is the limited surface area of individual seriously contaminated locations. Most organisms at the top of the food chain will forage in a broader area than just the contaminated site.
There are several reasons to discuss the choice to exclude biomagnification. Predators with only a small home range -e.g. mice or meadow birds- might get a large part of their prey from a seriously contaminated location. If several small contaminated sites exist in a same region, bio-magnification can be of importance. Especially in riverbeds, including river forelands, areas of serious contamination can be large, and certain predators are attached to this habitat.
Including biomagnification in deriving a SRCeco is technically possible and several methods
can be followed. Inclusion of the risks for predators after biomagnification will lower the
SRCeco for the more hydrophobic compounds (roughly Kow>5). For example for PCBs and
dioxin-like compounds, SRCeco would be seriously lowered if biomagnification were included
in the derivation. The current SRCeco is not protective for the situations in which
biomagnification plays a significant role or where cattle or food products are at risk. As for certain human exposure routes, serious risks can also occur in slightly contaminated sites, and should be considered separately.