Report of the sub-committee on the environment and health

8. Identification of environmentally harmful and dangerous pesticides, and operationalisation of the precautionary principle

8.1 Ranking of environmentally harmful and dangerous pesticides
8.1.1 Risk of groundwater pollution
8.1.2 Impacts on the environment
8.1.3 Effects inhealth
8.1.4 The sub-committee’s conclusions and recommendations
8.2 Precautionary principle and risk
8.2.1 Approaches to the precautionary principle
8.2.2 Groundwater and drinking water as a case
8.2.3 The sub-committee’s conclusions and recommendations

8.1 Ranking of environmentally harmful and dangerous pesticides

In this chapter we assess the possibility of identifying the most harmful pesticides by means of ranking methods. The chapter is based on an analysis by Lindhardt et al. (1998). The analysis covers the risk of groundwater pollution, the impact on terrestrial and aquatic environments and effects on human health.

It is possible to rank the toxicity of chemical substances, e.g. effect measurements or toxicity equivalents, and a number of intrinsic dangerous properties in connection with the substances’ hazard classification. It is also possible to rank emissions, dispersal and load, e.g. emissions in quantities or concentrations, emission measurements, or exposure in quantities or concentrations, just as a number of substances are regulated by means of limit values or threshold values. In other areas, ranking is based on vulnerability, e.g. through recommendations for pregnant women or children as special risk groups in a health context. In some areas, the problem is so complex and the underlying knowledge so limited that it is not possible to carry out a classification at all – as, for example, when a normally geologically determined soil classification is to be converted for use in ranking the vulnerability of the soil to infiltration by chemicals.

It is possible on this basis to set up different index systems for the environment field:
Risk indices, which combine the classification and the limit-value setting
Load indices, which combine the limit-value setting and the vulnerability assessment
’Pollution’ indices, which combine the classification, the limit-value setting and the vulnerability assessment.

Such indices, which are used in practice, are usually linked to some few single elements that can be ranked separately, but that unavoidably become diffuse and lose their value when many variables are combined because the uncertainty grows as a consequence of the elements’ interaction. Specifically in connection with such complex systems, there is a need to apply the precautionary principle in relation to the same uncertainties and a desire to be able to regulate in a simple way the individual elements included in the complex system. These matters will be illustrated in the following, using pesticide pollution of groundwater as an example.

International cooperation

In 1997, a conference was held, under the OECD, on risk indicators for pesticides with a view to measuring trends in the risks associated with the use of pesticides. Indicators have been developed for different purposes, one of which is to assess national risk reductions and identify those pesticides that contribute most to the total risk. Since the indicators used differ in their structure, pesticides can in some cases be indexed differently. It was also pointed out that a number of indicators would be preferable to a single indicator because there are no satisfactory methods for combining different types of risk for each pesticide. The OECD working group Pesticide Forum is now continuing work on the development of risk indicators for pesticides on the basis of the conference’s recommendations. The results of this work were presented at a workshop in June 1999.

Under the EU, work started in 1998 on a project called Concerted Action on Pesticide Environmental Risk Indicators (CAPER). The purpose of the project is to compare eight different indicators for the environmental effects of pesticides. The eight indicators have been developed for decision-support systems, for example, and for comparison of farming systems. Most of these indicators have not yet been published. It should be noted that, in EU contexts, too, there is a need to recognise the extent to which ranking models can be used to identify the most environmentally harmful pesticides (Lindhardt et al. 1998).

8.1.1 Risk of groundwater pollution

There are several methods for calculating indices that allow ranking of pesticides with respect to the load on the groundwater. Table 8.1 shows the methods that are relevant for Danish conditions and the parameters and data used in them.

Table 8.1 Look here!
Methods of ranking the potential of pesticides for leaching to groundwater. The methods are listed in order of increasing complexity and data requirements. The table also shows the types of parameters used in the methods.

GUS index

The GUS index was developed by the California Department of Food and Agriculture (Gustafson 1989). It is a very simple index for the risk of leaching to the groundwater. The index is based only on the inherent properties of the active ingredient. It divides pesticides into three categories: "probably leachable", "possibly leachable" and "probably not leachable".

Hasse diagram

The Hasse diagram is used to compare pesticides’ potential for leaching by comparing their dosage and adsorption to soil (Sørensen et al. 1997). Degradation can easily be included in the method, but there seems to be a lack of correlation between the degradation time and the actual findings in the environment in major studies. In addition, the inclusion of degradation is problematical owing to other factors, which are described under the Dutch point system, see below. For these reasons, it is proposed in this analysis only to use dosage and adsorption to soil. The strength of the method lies in the fact that less stringent model assumptions can be used to calculate the statistical probability of a pesticide’s ranking with respect to leaching. For a pesticide to be judged "better" than another, all the parameters mentioned must exhibit a smaller potential for leaching. If that is not the case, the two pesticides cannot be compared directly, but are compared indirectly through their ranking in relation to other comparable pesticides.

AF index

The AF index was also developed in the USA (Rao et al. 1985). AF stands for Attenuation Factor. It is not only based on the pesticide’s inherent degradability and ability to adsorb to soil but also includes soil parameters that describe the adsorption and degradation, together with the percolation of net precipitation. The aim of the AF index is to express the proportion of the amount of pesticide supplied that will pass a selected depth in the soil and could thus give rise to groundwater pollution. The infiltration is described by a simple, one-dimensional transport equation, with the degradation and adsorption assumed to be constant in the soil profile. The advantage of the AF index over the GUS index is that the leaching can be related to the infiltrating water, the soil parameters and the pesticide dosage.

Dutch point system

In the Netherlands, use is made of a point system comprising four main elements: the consumption of the individual pesticide; the location of the treated area in relation to important drinking water aquifers; the solubility of the pesticide, which can be used as an indirect measure of the adsorption to soil; and the degradability in soil. However, the degradability is considered separately because of a lack of data on some substances’ degradation. In addition, data on degradation under aerobic conditions can be misleading because anaerobic conditions can occur in soil environments. Lastly, some substances are broken down into more mobile metabolites, so data showing rapid degradation of the parent substance are misleading in these cases. The individual substances are indexed in relation to each other on the basis of the sum of points. The method is based on risk thinking and therefore cannot be used directly to rank pesticides because it depends on local environmental conditions.

Yardstick method

The Dutch "Yardstick" method was developed by the "Centre for Agriculture and Environment" in the Netherlands (Boesten, Van der Linden 1991; Reus, Pak 1993). It is intended for the individual farmer, who can use it to try to reduce the environmental impact of his spraying programme. The Yardstick is a point system, in which the environmental impact is calculated in the form of "environmental impact points". The system covers several environmental impacts. The model calculates the concentration of pesticides in the uppermost groundwater by means of the deterministic model PESTLA, using data for degradation and adsorption to soil. The Yardstick operates with sandy soil, which is typical for the Netherlands. The results are given for five classes of soil and spring and autumn spraying, where the temperature and precipitation vary. Since the method is based on the same simple indexing methods or on the deterministic model PESTLA, it does not produce results fundamentally different from these. Like the Dutch point system, the method is based on risk thinking and therefore cannot be used directly for ranking pesticides because it depends partly on local environmental conditions and partly on the actual cultivation practice.

Fuzzy Expert model

This method was developed in the Netherlands (van der Werf, Zimmer 1998). The risk of groundwater pollution is assessed partly on the basis of a GUS index and partly on the basis of a "leaching risk" parameter. This parameter requires supplementary considerations or model analyses. A high level of expert knowledge is needed to use the method, making it difficult to work with. The method can therefore not be used under Danish conditions without considerable input from a variety of experts. For this reason, the model has not been used here to assess leaching. However, the method also covers other environmental aspects than groundwater so its use could be considered in continuing work on environmental indices.

Expert assessment

In an expert assessment, one or more experienced experts designate substances on the basis of existing data and an integrated assessment. An expert assessment thus includes a systematic review of data and assessment elements that can be classified and ranked in the normal way, but must at the same time be combined with an experience-based assessment of lacking data and unclassifiable elements. A large number of factors are considered in the assessment of the risk of leaching to the groundwater within the framework of the authorisation scheme for pesticides. They include the inherent properties of the substances and their metabolites and degradation products, combined with the dosage, the time of application and any finds in groundwater. On the basis of experience, some substances are judged to lie closer to the critical limit for leaching to groundwater than others. This assessment determines whether a product can be authorised or not. Precise designation of the most problematical substances requires a thorough examination of the documentation and assessment of monitoring results, and possibly calculations with a leaching model, e.g. MACRO (see below). This form of assessment has been used in the assessment of many of the pesticides that have been placed on the blacklist.

Deterministic models: the MACRO model

A number of actual transport models that describe the transport of pesticides down through the root zone have been developed (e.g. LEACHM, PELMO, PESTLA and MACRO). These are so-called deterministic models in which it is assumed that the outcome under given assumptions will always be the same. The so-called stochastic (or probabilistic) models, on the other hand, take account of variations in the system and thus only talk about the probability of a given outcome. There are as yet no suitable stochastic models for ranking pesticides’ potential leaching to the groundwater. Unlike the simple index methods, both deterministic and stochastic models give a more varied description of pesticides’ movement in the soil.

The models can to a varying extent include the variations in different types of soil (root zone), special geological and geochemical conditions, transport mechanisms, climatic conditions, crops and application practice. PESTLA has been used to index the potential leaching of pesticides under Dutch conditions. The MACRO model, which can compute both the unsaturated and saturated water flow in a piece of cultivated land, is of particular interest. The model can also take account of saturated flow to drainage systems. Unlike other models, such as PESTLA and PELMO, MACRO can take account of preferential flow via macropores, and it can also handle degradation products, which PESTLA and PELMO cannot.

MACRO is gaining ground in the Danish authorisation procedure. It is used when the available documentation does not provide a clear answer to the question of whether a substance leaches to the groundwater or not. To make it easy to handle, two scenarios are described with given types of soil, precipitation events and other necessary parameters. The model calculations are one of several elements in the overall assessment. However, neither MACRO nor other leaching models have been sufficiently validated to warrant basing a registration procedure directly on the results of the model. Since the uncertainty in the model’s outcome is thus not yet known, DEPA now uses so-called "realistic worst case" data in the model.

The advantage of using numerical models to describe metabolism and transport of pesticides in soil is the possibility of testing many more combinations of types of soil, climatic conditions, time of application, plant cover, etc. than is technically and financially feasible with lysimeter and field tests. The model may therefore prove suitable for ranking pesticides’ potential for leaching.

Comparison of the methods

Of the methods mentioned, the GUS index, the Hasse diagram, the AF index and the expert assessment have been used to compare and rank 73 different pesticides used in agriculture (Lindhardt et al. 1998). The Fuzzy expert model, the Yardstick method and the Dutch point system have also been considered as a basis for ranking. However, as these methods would not give results fundamentally different from the above-mentioned methods, they have not been used. Ranking on the basis of calculations using a real transport model such as MACRO has also been considered, but has not been done for reasons of time (Lindhardt et al. 1998). Besides that, MACRO, too, is based on the same fundamental parameters for description of the mobility of pesticides as the simpler methods.

Lack of accordance between the methods

The study shows a lack of accordance between the different models. It is thus not possible to identify clearly those pesticides that have the greatest potential for leaching to the groundwater. One of the reasons for this is great variation in data, which means that the choice of data can determine how a pesticide is ranked. The rankings set up are primarily based on data concerning the pesticides’ adsorption in soil (KOC) and degradation time (DT50). In the case of the Hasse diagram, data are also included on the dosage, while net infiltration and soil parameters are included in the AF index. Major international studies indicate that pesticide consumption, in particular, and to some extent also KOC, are of significance for the occurrence in newly formed groundwater, whereas DT50 does not seem to have any significant correlation with field data. The fact that the Hasse diagram includes the dosage (consumption at field level) and adsorption to soil (measured as KOC) is an attempt to base the rankings on the results of these studies.

Setting up a gross list with a view to reviewing pesticide leaching

There is thus considerable uncertainty about the rankings set up. This is due particularly to the fact that leaching of pesticides to the groundwater depends on factors that cannot be described by simple indices – for example, extreme climatic conditions and seepage through cracks in the soil (preferential flow). However, the results of the different ranking methods can be used to draw up a gross list of the pesticides singled out as problematical using the different methods. The gross list, which comprises 35 pesticides, is shown in table 8.2. It is important to note that the list is relative, i.e., it primarily compares substances with each other and does not directly indicate substances as problematical in relation to the groundwater.

Table 8.2
Gross list of pesticides singled out in three ranking methods as potentially leachable in classes given as high (h) or low (l). Active substances with potentially high leachability in all three methods are denoted hhh, in two methods, lhh, hlh or hhl, and in one method, llh, lhl or hll. Active ingredients with low leachability are denoted lll.

   

AF indexa

GUS indexb

Hasse diagramc

  Class

log(AF)
l/kg

 

d,Koc
p factor

2,4D hhh

-1

S

0.97

clopyralid hhh

-2

S

1

metaldehyde hhh

-1

S

0.99

dicamba hhh

-3

S

1

metabenzthiazuron hhh

-3

S

1

fluroxypyr hhh

-3

S

1

isoproturon hhh

-3

S

1

bentazone hhh

-4

S

0.96

carbofuran hhh

-5

S

1

difenzoquat hhh

-5

S

1

mechlorprop-P hhh

-5

S

1

MCPA hhh

-6

S

1

metsulfuron-methyl hhl

-1

S

0.39

flamprop-M-isopropyl hhl

-2

S

0.87

metribuzin hhl

-4

S

0.88

triasulfuron hhl

-1

S

0.81

ethofumesate hhl

-6

S

0.76

haloxyfop-ethoxyethyl hhl

-2

S

0.79

imidacloprid hhl

-1

S

0.58

fenitrothion hlh

-9

M

0.99

terbuthylazine hlh

-9

M

0.94

pirimicarb hhl

-7

S

0.64

azoxystrobin hll

-8

M

0.76

isoxaben hll

-8

M

0.27

tebuconazole hll

-3

M

0.56

dimethoate llh

-22

I

0.97

metamitron llh

-19

M

0.95

chlormequat-chloride (CCC) llh

-45

I

1

triallate llh

-55

I

0.98

prosulfocarb llh

-81

I

1

mancozeb llh

-162

I

1

pencycuron llh

-84

I

1

tolclofos-methyl llh

-96

I

1

malathion llh

-289

I

0.94

dichlorprop-P llh

-13

M

0.97

linuron lll

-11

M

0.85

propiconazole lll

-11

M

0.49

phenmedipham lll

-18

M

0.76

propaquizafop lll

-26

M

0.47

triflusulfuron-methyl lll

-21

M

0.65

chlorfenvinphos lll

-26

I

0.88

esfenvalerate lll

-19

I

0.44

fluazinam lll

-32

I

0.53

propyzamide lll

-74

I

0.65

ioxynil lll

-167

I

0.7

fuberidazole lll

-163

I

0.11

prochloraz lll

-44

I

0.46

fenpropidin lll

-44

I

0.6

chlorothalonil lll

-98

I

0.84

propamocarb lll

-171

I

0.76

mepiquat-chloride lll

-209

I

0.57

bromoxynil lll

-199

I

0.64

imazalil lll

-60

I

0.07

fenpiclonil lll

-12

I

0.22

maneb lll

-300

I

1

napropamide lll

-119

I

0.46

ethephon lll

-300

I

0.43

furathiocarb lll

-300

I

0.33

fenpropimorph lll

-300

I

0.57

cypermethrin lll

-165

I

0.14

aclonifen lll

-190

I

0.79

fluazifop-P-butyl lll

-300

I

0.33

desmedipham lll

-256

I

0.2

fenoxaprop-P-ethyl lll

-300

I

0.17

pendimethalin lll

-234

I

0.55

permethrin lll

-300

I

0.2

trinexapac-ethyl lll

-300

I

0.53

tefluthrin lll

-300

I

0.18

glyphosate lll

-300

I

0.53

alpha-cypermethrin lll

-300

I

0.09

fludioxonil lll

-82

I

0.09

tau-fluvalinate lll

-300

I

0.06

lambda-cyhalothrin lll

-300

I

0.06

a = AF index: log(AF): high = greater than or equal to -10; low = less than -10
b = GUS index: high = S; low = M or I
c = Hasse diagram: high = greater than or equal to 0.9; low = less than 0.9

The cut-off values in table 8.2 are critical for the analysis and are open to discussion. Since the list is regarded as a relative comparison between substances, the cut-off values have been chosen such that each index identifies 25-30% of the pesticides as more leachable than others. The analyses accord well with each other, although the GUS index is more closely related to the AF index than the Hasse diagram is to the Gus and AF indices. That is because the dosage is only included in the Hasse diagram and because the Hasse diagram does not include the degradation, which is included in the other two indices.

The substances on this gross list should be subjected to a more detailed risk assessment in line with that used in the authorisation scheme. In addition, all the substances could be run through MACRO after it has been decided what data and scenarios are relevant.

The examination of the ranking methods also showed:
that, in the case of 9 pesticides, there was not complete laboratory data for their adsorption to soil or degradation in soil
that leachable metabolites are not included as part of the basis for ranking in the methods studied, except for the expert assessment
that, with a view to reviewing pesticides’ leaching, the gross list can be included in recommendations on substitution with less dangerous substances in the treatment situation.

Better monitoring of the groundwater

To clarify the extent to which the use of pesticides threatens our groundwater resources, a number of activities have been initiated in the last few years, partly to determine the extent to which our groundwater is affected by the pesticides authorised for use today and partly to clarify the more fundamental problems relating to the risk assessment. In order to learn more about the leaching of pesticides to the groundwater, the countrywide monitoring programmes for groundwater (LOOP and GRUMO) have been considerably expanded. The results of this will begin to emerge in the middle of 2000.

Early warning of the risk of leaching to the groundwater

The groundwater analysed is mainly more than five years old and often more than 25 years old. Feedback cannot occur before substantial parts of the groundwater resources must be presumed to be affected if the authorised pesticides or their metabolites leach to the groundwater. GEUS, DIAS, DEPA and DMU must together plan and establish a warning system that monitors young groundwater for pesticides that are used under the current rules. The warning system must make it possible to assess quickly and possibly remove authorised pesticides if, with lawful use under Danish conditions, they prove able to leach to the groundwater in concentrations that exceed the limit value. The first results are expected to become available in the second half of 2000.

Clarification of fundamental problems

Do we understand properly the transport of pesticides from the ploughing layer down to the groundwater and on to the extraction wells? As the results from the monitoring programmes have emerged over the last 10 years, showing that the groundwater is polluted with pesticides, several different hypotheses have been advanced on the reasons for this: they include preferential flow, colloidal transport or the right description of the degradation process of the individual substances. To achieve greater understanding of these problems, a number of research activities have been initiated, including The Danish Environmental Research Programme (SMP96): "Pesticides and Groundwater", which runs from 1996 to 2000, and the Interministerial Pesticide Research Programme, 1995-1998. Activities under SMP96 will shed light on the importance of preferential flow for the transport of pesticides in structured soil from the ploughing layer down to the groundwater, the transport and metabolism of pesticides in the groundwater itself, and the possibilities of describing the transport of pesticides on a regional scale by means of mathematical models. The results of these research activities can be expected to become available in the next couple of years.

Basic need for knowledge

There are four reasons why it has not been possible to identify the pesticides authorised for use today that present the greatest threat to the groundwater:

1) It is not certain that we properly understand the main processes responsible for the transport of pesticides down to the groundwater – e.g. the importance of preferential flow.
2) We have only limited understanding of the variation in a number of the main parameters (e.g. degradation and adsorption in soil) included in the description of the leaching of pesticides to the groundwater.
3) There is a lack of recognised methods for rapidly assessing the leachability of the individual substances.
4) There is a lack of data for the different models – particularly the complicated models such as MACRO.


1) Need for more knowledge about the geological factors

In the last few years a lot of new knowledge has been gained about the main processes of pesticide leaching to the groundwater. However, the results indicate that we have insufficient understanding of the spatial variation in the hydraulic parameters that govern the transport of pesticides from the ploughing layer to the groundwater. We need to translate the knowledge gained concerning the flow in structured soils into a quantitative mapping of the geographical extent of these properties.

2) Need for more knowledge about the individual pesticides

Despite the considerable resources that have gone into research activities aimed at increasing our general understanding of the transport of pesticides down to and in the groundwater, there are a number of vital problems that remain to be answered in relation to risk assessment of the individual pesticides. One of those problems is the variation in the parameters that are important in the description of the individual pesticides.
Variation in degradation paths and rates, and the substances’ adsorption in the ploughing layer, which can vary considerably, both within the individual field and between different fields.
Variation in these parameters down through the soil profile. Analyses carried out in soil from the ploughing layer do not necessarily describe the fate of the substances when they get down below the ploughing layer.
The information on possible metabolites from the individual substances is insufficient. It is not enough to know the degradation paths in the ploughing layer. In the case of relatively mobile substances, it is also important to know the degradation paths in the subsoil and possibly in the groundwater.

This information can be procured in future by tightening the requirements for authorisation of pesticides, but this method will only help in the assessment of new substances or in actual reviews. It may also be necessary to obtain EU agreement to such requirements. It will at any rate take time before the desired data are available. Before the requirements can be tightened, agreement must be reached on which data are necessary and sufficient, and a procedure must be laid down for procuring them.

3) New decision-making tools based on statistical methods

Although ranking methods are based on a very simple understanding of the process and therefore cannot stand alone in an assessment, major studies indicate that there is, in spite of everything, some relationship between relatively simple parameters and the occurrence of pesticides that is observed. Much of the observed variation in occurrence can in fact be explained on the basis of relatively simple relationships (Koplin et al. 1998; Kreuger, Tornqvist 1998). As results come in from still more studies (national and international), it is recommended that decision-making tools be developed (simple ranking index) on the basis of statistically documented relationships with a view to generalising the studies to uninvestigated areas and pesticides. The possibility of developing simple, stochastic (probabilistic) models should also be looked into.

4) Better data for the deterministic models

The question has been raised of whether one could use model calculations to rank authorised pesticides on the basis of their leaching risk. The calculation of leaching by means of MACRO covers a range of factors that are not taken into account in the simple indices – the relevant dosage for a specific crop, plant cover, dosing time, Danish climatic conditions (precipitation and temperature) and preferential flow for structured soils. However, some factors have to be clarified before MACRO can be used to rank the authorised pesticides:
Quality-assured data have to be obtained for each pesticide’s degradation and adsorption in different types and strata of soil, together with the dosage.
It must be clarified how the uncertainty on the variables describing the degradation and adsorption is to be assessed.
Limit values must be defined.
It must be clarified how representative the scenarios used by DEPA are.

Ranking by means of MACRO does not directly include an assessment of the extent to which metabolites are formed, which must be considered a serious constraint.

Even if satisfactory data are obtained and other factors can be satisfactorily clarified, ranking by means of MACRO cannot stand alone. An integrated expert assessment will also be needed.

8.1.2 Impacts on the environment

Pesticides can be ranked solely on the basis of their toxicity to different organisms. The toxicity can also be expressed as the tolerable limit value, which can be calculated on the basis of all species or groups of organisms or be expressed as a value for the most sensitive species. The intended effect of the pesticides in the terrestrial environment is direct and considerable in cultivated areas and their immediate surroundings. The effect of herbicides on the flora is obvious. The indirect effects on the fauna through impact on the primary links in the food chain are described in section 4.2.1.

Treatment frequency index as a measure of the load on the terrestrial environment

It is for mammals and birds that we have the most complete toxicity data for the terrestrial environment. However, the risk of direct poisoning of humans, other mammals and birds has been reduced considerably by reviewing existing substances and changing the authorisation scheme for new pesticides, so that all the substances in practice have a low level of toxicity to vertebrates. It is therefore not possible to rank modern Danish pesticides on the basis of their toxicity to these groups of fauna. That is because characteristic farmland fauna and, particularly, birds are considerably more affected by the indirect effects of pesticide use. These indirect effects are related to the effect of the product on the target organism and on other species belonging to the same group of organisms. Since the indirect effects are the most extensive, it is important to include them in the load and risk assessments. The dosage is determined by means of field trials, the aim being to achieve an effect on the target organisms that results in at least 90% reduction of the population density. The recommended field dosage is thus a realistic and accurate indicator of the effect of the product in the field, compared with the toxicity determined in laboratory tests. The treatment frequency index is based precisely on the recommended field dosage with a view to the direct effect on the target organisms. Since the target organisms’ related species – fungi, plants or arthropods – are also affected by fungicides, herbicides or insecticides, respectively, the treatment frequency index is also an indicator of the indirect impact on the ecosystem as a consequence of changes in the quantity and type of food in the food chains.

Critical loads for plants and animals in terrestrial natural areas

It is possible to carry out calculations for each pesticide that show the dosage that will be harmless for most faun and flora in hedges, small biotopes and other areas bordering on cultivated areas. This value is called the critical load. It expresses the total supply of chemical substance that is considered to be harmless to the environment. The calculations do not take into account the fact that there can be increased effects from applying several substances on the same area. Calculations have only been carried out for a few, selected substances (Jensen, Løkke 1998) because considerable resources are needed to collect and assess data for all relevant pesticides. Moreover, there are insufficient data for all pesticides. Calculations for four substances show that the critical load for insecticides and herbicides lies in the interval from one thousandth part to one ten-thousandth part of the field dosage. A system could possibly be developed for ranking pesticides’ load on the terrestrial environment on uncultivated land. However, the load depends primarily on how much of each pesticide is used and on spraying practice. If, however, it were found that there was no great variation in the critical load for different pesticides, the most important factors would be the consumption figure and the dispersal outside the cultivated areas. In such case, the treatment frequency index could give an indication of the load on the uncultivated areas as well.

Impact index for the aquatic environment based on acute toxicity

In the aquatic environment, the most problematical substances for fish, crustaceans and algae are assessed. Ranking solely on the basis of the substances’ toxicity is not sufficient because the effects depend on the exposure of the environment. The ranking can be based on an impact index defined as the relationship between dosage as an expression of the exposure and the acute toxicity. This ranking shows that, for fish and crustaceans, it is particularly the eight authorised pyrethroids, all of which are used as insecticides, that have the highest ranking (Lindhardt et al. 1998). In the case of algae, it is the herbicides that head the list. The method is encumbered with considerable uncertainty because it does not include exposure considerations and long-term effects. The method cannot be recommended as the only basis for identifying the pesticides that are most harmful to aquatic organisms. That requires a real risk assessment, as is done in connection with the present authorisation scheme.

Ranking on the basis of an expert assessment of the safe distance to watercourses and lakes

In the authorisation of pesticides, an expert assessment is carried out. This can lead to authorisation with a condition attached to it to the effect that the substance must not be used within a given distance from watercourses and lakes. Such distance requirements indicate that the substance (the product) is problematical in relation to aquatic organisms, and they can be used directly to rank the pesticides. Prior to the change in administrative practice in 1993, any conditions concerning distance to watercourses and lakes were based on an assessment of the hazard of the substances’ inherent properties. Since 1993, distance requirements have been based on a risk assessment in which the exposure is taken into account. If the distance requirements are to be used for ranking purposes, the substances (products) that were assessed before 1993 must also undergo a risk assessment. This would in any case be done in connection with the review required by law at intervals of not more than 10 years.

8.1.3 Effects inhealth

With respect to health effects, ADI/TDI values or the classification could be used for ranking pesticides. Classification means dividing pesticides into classes for level of hazard – in this connection, the hazard of the toxic effect – in accordance with Statutory Order on Pesticides. ADI represents a quantitative measure of the load to which one can be exposed from cradle to grave without any overt risk of harmful effects. This is a safety value calculated on the basis of the highest dose of the substances that does not produce the critical effect, i.e. the most sensitive effect in the most sensitive species of test animals or in humans if data are available on that.

The classification covers a number of undesirable biological effects resulting from exposure to a chemical substance (cancer, reproductive damage, etc.), for which there is a high probability that humans develop such diseases through contact with the substance. The assessment is a qualitative one or an assessment of evidence that a given pesticide with the biological action as an inherent property can have a harmful effect on humans.

Ranking on the basis of different health effects

Ranking can be based on a number of different biological effects. For example, pesticides can be classified on the basis of growing evidence of their ability to cause cancer. None of the pesticides authorised today is classified as carcinogenic in category 1 or 2. Pesticides classified as carcinogenic in category 3 can be authorised if a risk assessment shows that there is no risk of cancer developing with normal use. It is thus possible to rank between substances that are classified in category 3 and substances that are not classified as carcinogenic.

Ranking on the basis of the distance between ADI and the exposure

If one chose to use ADI as the basis for the prioritisation with respect to phasing out pesticides, the distance between ADI and the assessed or measured exposure to the substance could be used as the basis for ranking them. It would thereby be possible to identify the substances with the smallest safety margin with the current exposure of humans. ADI and the classification cannot be directly combined in a ranking system because the critical effect used in setting ADI is not necessarily the effect for which the substance is classified. For example, a substance that has produced skin irritation in several toxicological tests and epidemiological studies can in many cases have another undesirable biological effect that forms the basis for setting ADI because this effect is seen at lower dosages in the toxicological tests. However, it might be possible to combine the above two criteria by assigning the substances a prioritisation value for each of the criteria and then using the total ranking sum as the basis for the main prioritisation with a view to a phase-out.

8.1.4 The sub-committee’s conclusions and recommendations

The sub-committee has considered the question of ranking in the following areas:
groundwater pollution
impact on the terrestrial environment
impact on the aquatic environment
health effects

Leaching to groundwater
The sub-committee concludes that it is not possible with the existing, simple methods to rank pesticides clearly with respect to their ability to leach to the groundwater. However, by using four different methods, it is possible to set up a gross list comprising 35 authorised substances. The substances on this gross list should be more closely assessed, using mathematical models, the latest knowledge and the results of measurements.
The sub-committee recommends that, with a view to reassessing the leaching of pesticides, the gross list be included in recommendations concerning substitution with less dangerous substances in the treatment situation.

International cooperation

International fora are also working on ranking models with a view to identifying the most environmentally harmful pesticides. Both within the OECD and the EU, indicators are being developed or tested for assessing the environmentally harmful effects of pesticides. The OECD’s preliminary work was presented in 1997. It showed that many of the existing indices differed in their structure, leading to different indexing of pesticides. Within the EU, a Concerted Action (CAPER) has commenced. The object is to compare the usefulness of eight different indicators.

Knowledge-building in the groundwater field
In the years ahead, research programmes now in progress will add to our knowledge concerning the fundamental problems, and improvements that are being made in the monitoring of groundwater and early warning of the risk of leaching of pesticides to the groundwater will increase the safety level.

Need for knowledge in the groundwater field
The sub-committee wishes to point out that improved risk assessment depends on continued clarification of the processes governing the transport of pesticides to the groundwater and in the groundwater aquifers and on greater understanding of the spatial variation in the parameters governing the transport of pesticides. As results come in from an ever-growing number of both national and international studies, decision-making tools should be developed on the basis of statistically documented relationships for pesticides’ potential leaching with a view to generalising the studies to uninvestigated areas and pesticides. In connection with increased use of mathematical models (e.g. MACRO) in the assessment of the risk of pesticide leaching, the sub-committee recommends that work continue on assessing their validity. In this connection, there is a particular need for geological and substance-specific data. The possibility of developing simple stochastic (probabilistic) models should be investigated.

Terrestrial ecosystems
With respect to the terrestrial environment, it is not possible to indicate a method for ranking the direct effects, because the indirect effects and the combination of many pesticides play the biggest role. However, the treatment frequency index can be used as a measure of the impact because it is based on the biologically active field dosage and can thus be used as a simple indicator of both the direct effect on the target organisms and related species and for the indirect impact on the ecosystem as a consequence of changes in quantity and type of food in the food chains. It would also be possible to calculate an index for the dosage believed, with our present level of knowledge, to be harmless to most animals and plants in uncultivated areas that receive pesticides via spray drift or atmospheric transport (tolerable limits).

Aquatic ecosystems
For the aquatic environment, the present authorisation scheme includes an expert assessment that can result in new pesticides or products being authorised on condition that a given distance is maintained to watercourses and lakes. Such distance requirements indicate that the substance (the product) is problematical in relation to aquatic organisms and could be directly used for ranking or classifying the pesticides.

Human health
For the human-toxicological area one could use the relationship between the Acceptable Daily Intake, ADI, and the estimated exposure to the substance as a basis for ranking pesticides. In Denmark, with the present use of pesticides, the exposure at single substance level through food products is around 1% or less. Ranking would enable identification of those substances that, with the current use, have the lowest safety margin for humans. However, since the effects of the individual substances are not comparable, ranking alone would not be enough but would have to be supplemented by an expert assessment.

8.2 Precautionary principle and risk

The reasons for using a precautionary principle are as follows:
The uncertainty and variation that are always associated with the data on which decisions are based – with respect to both the individual measurement and the generalisation from limited studies to the entire environment or ecosystem and all the species and populations that are to be protected.
Some systems (e.g. meteorological and ecological systems) can in some circumstances exhibit indeterminable, e.g. chaotic, behaviour. In such cases it is in principle impossible to predict the consequences of an exposure.
Incomplete knowledge about how ecosystems are affected by and react to pesticides and about the extent to which the systems can regenerate or be re-established after a harmful exposure.
The risk associated with making a mistake – i.e. underestimating a risk with lasting consequences for humans and the environment.
The desire to protect specially exposed groups, e.g. children, even better.
The fact that mistakes could have a serious future effect, e.g. through accumulation of pesticides in groundwater or the atmosphere.

The "precautionary principle" has gained a considerable following in the environmental field (Danish Environmental Protection Agency 1998c) but has not been precisely defined. The following framing of the concept can be used:
The purpose of the precautionary principle is to provide greater certainty that damage will not occur.
The desire for more extensive use of the precautionary principle is based on the view that neither the environment nor human health should suffer as a consequence of behaviour about which there is scientific doubt or uncertainty.
The precautionary principle allows the authorities to require less evidence of a polluting effect from a given form of behaviour and to take action to regulate the behaviour if there is just some possibility of pollution.

In relation to the weighing up of risks, the main content of the precautionary principle can be construed in the form of a question: "Who is to suffer the consequences of the uncertainty that must exist scientifically concerning the polluting effect of a specific type of behaviour?" This is illustrated by the following three concrete examples of application of the precautionary principle:

1) Carcinogens (The Delaney Clause)

The best known and most frequently mentioned example is undoubtedly the American health and food legislation’s so-called Delaney-Clause, which states that colour additives, other additives or pesticide residues must not be authorised in the USA if, at any level, they have been found to induce cancer in experimental animals or man. This rule was already controversial at the time of its introduction in 1958, in part because it was expressed as a requirement of "no risk", which, scientifically, is regarded as unachievable unless the requirement is linked to a direct ban on use in the individual cases or every assessment of additives/pollution is interpreted as a requirement of negligible or "de minimis" risk. The discussion of this interpretation of the precautionary principle is still going on. In the USA, the situation is that the requirement concerning actual banning of carcinogens is enforced in the case of real additives, including food additives, whereas pesticide residues (cf. Amendment PL104-170 of 3 August 1996) are, rather, assessed on the basis of the "de minimis" rule, interpreted as a lifetime cancer risk for one individual out of 1 million (1 ´ 10-6). Viewed as the risk rate, i.e. numerically, most experts, including FDA officials, describe this – largely unchallenged – as "negligible, trivial, insignificant or even non-existent".

2) Spraying of seed-bearing and fruit-bearing crops

Right back in the first decades with growing use of spraying, i.e. from around 1955 to the mid-1970s, it was established in most countries, including Denmark, that rules for pesticides and instructions for their use should be provided in accordance with the individual countries’ agricultural needs. This was normally verified by means of controlled spraying tests carried out under the special ‘climatic and cultivation conditions’ of the country in question.

Tests of this nature were carried out in Denmark in the period 1963-1975 by the National Plant Pathology Research Institute in cooperation with the National Food Institute (now the National Food Agency) with a view to authorisation of the products by the Ministry of Agriculture’s Poisons Board. The results of the tests often showed that substantial parts of the Danish fruit, vegetable and berry production at that time could be kept free of detectable pesticide residues provided the spraying times, waiting times before harvest and similar were fixed at ‘before flowering ends’ and ‘before seed-setting’ etc. This was largely accepted as a well-defined basis for ‘Good Agricultural Practice’ and thus met the requirements and wishes already formulated at that time concerning no (or only negligible or "immeasurable") residues in crops ready for eating.

Many of the regulations and recommendations given by the Poisons Board and, later, in DEPA’s authorisation schemes, thus contained significant elements of a restriction on use that went beyond the risk assessment requirements formulated for purely health and environment reasons. A similar situation arose in the summer of 1998 in connection with the discussion on the use of the herbicide glyphosate to combat couch grass in cereal crops shortly before harvest.

3) Drinking water from water supply systems

Owing to human error, a serious pollution situation occurred in a provincial town (Fåborg) in Denmark in the middle of the 1970s. The error resulted in the insecticide Parathion (Bladan) being sucked into the pipe network, contaminating large parts of the town’s public water supply system. The unavoidable decision taken was naturally that the system was to be closed down and to remain closed ‘until the plant and pipe network were free of any measurable residue of Bladan’. In other words, the precautionary principle was applied even though, in the case in question, a higher residual content than the limit applying at that time (given as 0.1 microgramme per litre) could have been accepted on the basis of a toxicological, health assessment. As a pollution situation that attracted great public attention, this case (together with similar cases in other countries in Europe), had both direct and long-lasting consequences for many subsequent cases, leading to today’s requirement of no (i.e. ‘immeasurable’ or only negligible) pesticide residues in drinking water/groundwater.

8.2.1 Approaches to the precautionary principle

The following two approaches to the precautionary principle can be described:

1) An effect assessment/risk management approach

The effect assessment/risk management approach is based on extensive knowledge of data and other technical and scientific information. Using statistically or pragmatically fixed (un)certainty factors, one assesses the hazard of individual substances and the risk associated with their use. In that connection, dose-effect curves and lowest effect values are as far as possible established. Tools in this connection include setting of limit values, threshold values, pollution standards, etc., often supplemented by emission requirements based on health and environment considerations and, possibly, regulation of use. This approach is illustrated in figure 8.1 below.

Figure 8.1
The risk assessment approach with relationship between dose and effect for humans and for aquatic organisms and plants. C = lowest effect level for humans, B = toxicologically set zero effect level using a safety factor, A = ecotoxicologically set zero effect level for aquatic organisms using a safety factor. 0.1m g/L = the limit value for pesticides given in the Drinking Water Directive (m g/L = microgramme per litre).

(Figure text:
Effekt = Effect
Laveste effektniveau for akvatiske dyr og planter = Lowest effect level for aquatic organisms and plants
Lavest effektniveau for mennesker = Lowest effect level for humans
Økotoksikologisk risikovurdering = Ecotoxicological risk assessment
Humantoksikologisk risikovurdering = Human-toxicological risk assessmentKoncentration i vand = Concentration in water)

2) A zero value approach

In the zero value approach, basis is set by uncertainties, random variations and possible erroneous assessments, including insufficient data or a direct lack of knowledge, and greater importance is attached to a general recognition of the fact that data, documentation and/or concrete knowledge must always, scientifically, be regarded as deficient. From a desire for a "zero concentration or dose", requirements can be made concerning insignificant, possibly ‘immeasurable’ content/doses/loads etc. for exposed individuals, populations and/or environments. This is illustrated in figure 8.2.

Figure 8.2
The zero value approach with a fixed, low limit. A, B and C refer to individual pesticides with different toxicity. The curves can be the lowest toxic dose/effect curves for humans: NELA, NELB and NELC = lowest effect level for A, B and C, respectively. 0,1m g/L = the limit value for pesticides given in the Drinking Water Directive (m g/L = microgramme per litre).

(Figure text:
Effekt = Effect
Nulværdi = Zero value
Dosis = Dose)

Uncertainties and variations

Operationalisation of the approaches must accordingly be linked to the assessment of documentation and data, including – particularly, the lack of material and data, which can be concretised to a number of variations and uncertainties comprising:
Measuring uncertainty in the experimental procurement of biological, chemical and physical-chemical data. This is characterised by the fact that we can calculate both probability for effects and uncertainty for concrete biological or physical systems.
The uncertainty in the assessment in the further interpretation of performed tests, often by transformation on extrapolation of laboratory (and epidemiological) observations and measurements for use with other organisms (e.g. animals to humans) or pollution situations (e.g. aquatic environment to soil or air). This is characterised by the fact that we have some knowledge about the systems we describe, but variation and probability of risk depend on estimation via interpretation of knowledge from one system or one situation to another.
Uncertainty of knowledge or direct ignorance with insufficient or no understanding or knowledge of effects or mechanisms behind effects, including naturally also a lack of possibility of predicting effects that are insufficiently described or that have not previously been observed. This is characterised by the fact that we do not know the system and/or the effects we have to describe and that we have no basis for predicting risk, let alone calculate the probability of effect.
Uncertainty concerning use or other uncertainty is linked to indeterminacy in commercial and societal development, marketing and use of pesticides. This is characterised by the fact that we normally do not fully know the systems in which pesticide use and dissipation is taking place - and we know that we do not know them.

These four areas of uncertainty are characterised by an increasing lack of knowledge, which thus reduces the scientific basis for decisions and correspondingly increases the need for administrative or politically based decisions.

The memorandum "Discussion of caution and risk" (Sub-committee for Environment and Health 1998) gives a detailed description of different areas of uncertainty and variation.

8.2.2 Groundwater and drinking water as a case

In Ministry of Environment and Energy Order on Water Quality and Supervision of Water Supply systems, No. 515 of 29 August 1998, the following is stated in chapter 2, section 4, concerning the quality of water from water supply systems:

The water from water supply systems that supply people with water for household use must meet the limit values for content of substances in the water given as the highest permissible values in Annex 1 to this Order. However, efforts must be made to ensure that the values are lower than or within the values given as guideline values in Annex 1.

In Annex 1 to the Order, the guideline limit value for pesticides and related products is put at "u.d.", i.e. under the detection limit for a method that can measure one tenth of the highest permissible value, which is fixed at 0.1 microgramme per litre in the Drinking Water Directive. A guideline limit value for pesticides is thus 0.01 microgramme per litre for each substance.

The Drinking Water Directive’s limit value of 0.1 microgramme per litre for pesticides has been fixed on the basis of a zero value approach, which is expressed by a negligible (‘immeasurable’) limit value covering all pesticides, instead of a risk principle, which would mean setting concrete, individual pollution levels judged on a toxicological/health basis to be acceptable or permissible from the point of view of health.

Areas of variation and uncertainty in risk assessment and risk management, on the one hand, and implementation of the zero value approach, on the other, are based on concrete studies and knowledge, but oblige us also to pinpoint/define our lack of knowledge or uncertain knowledge.

Toxicological and ecotoxicological testing of a chemical substance and assessment of the possibility of this substance affecting health and the environment are illustrated in figure 8.2. A single chemical substance is tested for a number of effects in experiments on mammals (e.g. mortality, acute, sub-acute and chronic effects, and, if possible, cancer and reproductive disorders) or on other live organisms (e.g. fish, daphnia, algae). For each of these an attempt is made to establish a dose/effect curve, as illustrated in figure 8.2, although it is recognised that in some cases these effects do not necessarily have any well-defined relationship between dose and effect.

For each of the measured effects, an attempt is made to establish a so-called zero effect dose for use in setting limit values or threshold values. This serves as a basis for risk assessments, i.e. the possibility/probability of (harmful) effects occurring through exposure even though the dose is smaller than the estimated zero effect dose.

The area of ’uncertainty’ close to the zero effect value must be regarded as a grey zone within which a residual effect, an unmeasured effect or an unknown effect could occur. Such situations can be counteracted:
either by including one or more pragmatically fixed ‘uncertainty factors’ (UF) in relation to the zero effect or threshold level and as the basis for an accepted/tolerated concentration or dose (the risk acceptance model)
or by fixing a so-called negligible content/dose that is based on a zero concentration or dose and that gives greater certainty that a "grey zone" will not be entered (the zero value /precautionary model).

Of the two forms of assessment, it is the first, i.e. the risk acceptance approach that is normally practised, possibly modified by including a supplementary principle concerning ‘good treatment practice’ (see below), while the zero value approach is used for pesticide residues in drinking water. However, as mentioned above, the zero value approach is also known in other connections – for example in assessment and regulation of carcinogenic, chemical substances (cf. the Delaney Clause, see above) or as an element of a restrictive treatment practice with specific requirements concerning time limits for spraying seed-bearing, berry-bearing and fruit-bearing crops etc.

With respect to the rules on pesticide residues in groundwater and drinking water, there is a distinct difference between the health assessment of the limit value and the environmental importance of the limit value, as also illustrated in figure 8.1 above. Whereas, with respect to human toxicology, all known pesticides have up to the present time been found acceptable/tolerable with respect to the groundwater criterion of 0.1 microgramme per litre, the figure with two concentration effect curves exemplifies how the lethal effect level for aquatic organisms (measured as EC50 or LV50) can be exceeded at values below 0.1 microgramme per litre. There are several such examples, particularly among insecticides, and for an environmental point of view, the value 0.1 microgramme per litre is in this case not an expression of the precautionary principle of the zero value approach. A risk assessment of these individual substances on the basis of existing laboratory data would lead to a lower limit value than 0.1 microgramme per litre for aquatic organisms.

It should also be noted that limit values set on the basis of detection limits are not static quantities. For example, technological development, with growing use of detection by means of mass spectrometry, means that it is now possible to determine the occurrence of a number of substances down to detection levels of 0.005 microgramme (m g) per litre, i.e. concentrations up to 20 times lower than was technically possible in 1980, when the limit value of 0.1 microgramme per litre was set.

There are thus no technical obstacles to reducing the guideline limit value of pesticide residues in drinking water, and one could, in principle, continue reducing it in step with the development of analytical methods. This would mean that, as a consequence of the zero value approach’s precautionary principle, one would not only move away from the precautionary principles of the effect assessment/risk management approach, but would also, in some years’ time, approach a limit value that very few pesticides could meet.

Continued reduction of the limit values in step with the development of methods of analysis would thus ultimately mean banning all use of pesticides if one maintained that one would not accept measurable quantities.

8.2.3 The sub-committee’s conclusions and recommendations

Two different approaches to the precautionary principle can be used – here called the risk assessment approach and the zero value approach.

Use of the risk assessment approach can imply a "conservative" (= "cautious") assessment based on concrete, empirical evidence, whereas use of the zero value approach can be based on an initially value-determined quality requirement that is only deviated from after assessment based on definable protection requirements.

If the risk assessment approach could be based on a sufficient quantity of scientific data to ensure complete protection of health and the environment, then, for example, the use of uncertainty factors (UH) could be held to be a satisfactory implementation of the precautionary principle. Such an approach would thus mean that the precautionary principle did not add anything to the traditional risk assessment. The authorisation scheme thus seeks to implement the precautionary principle, using all data and taking account of negative consequences of uncertain or unforeseen exposure, incorrect use, etc. With this approach, there should be no need for greater use of the precautionary principle, although this neither solves the problem of the possible inadequacy of the uncertainty factors nor answers questions about uncertainties as a consequence of lack of knowledge or indeterminacy related to the role of pesticides in society.

Conversely, the zero value approach would not answer directly the question of the actual size of uncertainty factors, but would be based on a value-determined requirement concerning a zero risk society.

The zero value approach could tentatively be used within an authorisation scheme for pesticides in the following areas:

  1. reduction of the limit value for pesticide residues in food products to the lowest detection level at any time
  2. reduction of the limit value for pesticides in groundwater to the lowest possible detection limit at any time or authorisation only of pesticides meeting a requirement of no or negligible mobility in soil
  3. no authorisation of substances classified as Carc3, Mut3 and Rep3
  4. no authorisation of substances subject to greater distance requirements than 6 m to watercourses and lakes
  5. greater possibility of restricting use with a view to avoiding the use of pesticides directly on crops, trees and bushes in their seed-bearing, fruit-bearing and berry-bearing periods.

It must be stressed that it has not been assessed whether the zero value approach should be used for the above-mentioned examples, since that would require a dialogue between the following players:

  1. scientific experts to define the limit for what can be predicted and isolate what cannot be elucidated
  2. an administrative instance to decide what can be operationalised
  3. political decision-makers, i.e. non-experts, to make the final decision on behalf of the public, partly on the basis of trust in the scientific knowledge and partly on the basis of ethical and political considerations.

It should also be noted that in the foregoing examples use is made of limit values fixed on the basis of the currently achievable analytical detection limit. Therefore, as a consequence of continuing technological development, the limit values will regularly be reduced, ending near zero. This could ultimately lead to a total ban on the use of pesticides.