In the EFSA Guidance Document, PT is defined as “the proportion of an animal’s daily diet obtained in habitat treated with pesticide”. As a worst-case assumption, animals are supposed to find all of their food in the treated area (PT = 1). In higher tier risk assessment more realistic estimates of PT may be used (EFSA 2009).
According to decisions at the workshop in Copenhagen 8-9 May 2012, PT = 1 shall be used for assessment of acute risk. In the assessment of long-term (reproductive) effects more realistic estimates of PT may be used, if such estimates are available for the species and crop scenario in question. Because PT data are generally sparse, some read-across between structurally similar crops is acceptable. All cases of read-across between crops must be duly justified.
PT may be estimated indirectly by radio-tracking of individuals, assuming that the amount of active time spent by an animal in a given crop is directly proportional to the amount of food eaten there. In radio-tracking studies animals may be caught in general farmland or in (or in close proximity to) the crop of concern. In both cases, PT may be estimated for the whole sample of individuals tracked (“all birds/animals”) or only for the subsample of individuals that actually visited the crop of concern during tracking (“consumers”).
EFSA (2009) recommends that for focal species caught within (or in close proximity to) the target crop, PT should be estimated from the total sample of individuals – whether they used the crop of concern or not. For focal species caught in the general farmland, only those individuals proved by radiotracking to visit the crop of concern (consumers) should be included in the estimation of PT.
Irrespective of the above choice (all individuals or consumers) it is necessary to decide what level of protection is required. For example, if the first-tier PT of 1 is replaced by a median or mean value, this would suggest that the risk assessment will only relate to those 50 % individuals that fall under this PT. If the 90th percentile of the PT distribution is used, 90 % of the population will be protected, provided that no other parameters drive the risk assessment5.
At the workshop in Copenhagen 8-9 May 2012 it was agreed to follow the EFSA recommendations concerning the use of “all animals” or “consumers”. It was further agreed to use the 90th percentile of the PT distribution for the core risk assessment (Northern Zone registration report).
Refinements of PT in new studies should as a minimum be based on 10 individual animals caught within (or in close proximity to) the target crop or on a minimum of 10 animals tracked to be “consumers” in the target crop. In the available studies (referred in this GD) refinement of PT based on the "consumers" group is accepted also in cases where the number of individual animals in this group is below 10, provided the number of individuals in the "all birds" group studied was above 10.
Refinement of PT based on the “home range approach” or calculation of Jacobs’ Index are not accepted. It is considered that firm relationships between these approaches and PT estimates based on traditional (“a day in the life”) sampling protocols have not yet been established.
Dehusking of seeds may reduce exposure in granivorous birds and mammals. Regardless whether the seed has been subject to seed treatments or has been contaminated during spraying, the substance will be mainly on the outside and dehusking may thus remove the majority of the residue. Based upon experimental (manual) dehusking of seeds, Edwards et al. (1998) suggested that the reduction of exposure may be as high as 85 %. However, even in species which routinely dehusk, dehusking depends on the kind of seed and only a proportion of the seeds are dehusked (SANCO/4145/2000).
In the case of birds, dehusking is mainly observed in smaller species (body weight < 50 g) and chiefly in the specialized granivores (finches, sparrows and buntings). Larger granivorous birds (body weight > 50 g) do not dehusk as they are able to destroy even hard-shelled seeds in their gizzard. Among the small birds, species with a relatively thin bill, such as skylark, wagtails and other insectivores, do not have the capability of dehusking. Even in the small, granivorous species, dehusking is not an all-or-nothing phenomenon; certain species dehusk some but not all seed types, and in the wild the actual proportion of seeds dehusked may depend on stressors such as feeding pressure, predation risk or competition (Prosser 1999). Assuming a standard reduction of 85 % (or any other value) of the theoretical exposure in species that dehusk is therefore not justified.
For granivorous mammals, e.g. wood mouse, dehusking or cracking of seed or fruit shells is often a part of their typical feeding behaviour. Distinct anatomical features such as specialized incisors or folds of skin that prevent material from entering the mouth while being gnawed indicate that most rodents will probably minimise the uptake of husks when eating seeds (DEFRA 2005). Several older studies have demonstrated that dehusking occurs under laboratory as well as under semi-field conditions but do not provide quantitative information on the effect of dehusking. Dehusking efficiency in mice has however been quantified in two new studies.
Ludwigs et al. (2007) quantified the efficiency of dehusking by laboratory mice and wild Apodemus mice. They found that the efficiency was strongly dependent on seed structure. Dehusking of sunflower seeds where the seed coat and the fruit coat are not grown together was highly effective (≥ 90 %). Dehusking of maize seeds was less effective, 62-65 % by Apodemus mice, probably because the outer layer of the seed is firmly adhered to the rest of the kernel. Experimentally induced food shortage reduced the percentage of maize seeds dehusked from 77 % to 65 % but dehusking efficiency, as measured for those seeds where dehusking was actually performed, was unchanged.
Brühl et al. (2011) studied dehusking by individually caged wood mice and found dehusking efficiencies of 60-80 % in cereals and c. 99 % in sunflower (Table 4.4). Notably dehusking efficiency was higher in barley than in wheat and maize. Dehusking efficiency was approximately the same, no matter if the seed was treated with a fungicide or a generic pigment and no matter whether the mice were starved before the experiment or not.
Table 4.4 Seed consumption and exposure reduction through dehusking behaviour of individually caged wood mice. Results are presented separately for seeds treated with red pigment and with fungicide (prothioconazole). N = no. of mice. Source of original data: Brühl et al. (2011); 10th percentiles were estimated for the present report.
(N = 12)
(N = 11)
(N = 12)
(N = 11)
Exposure reduction (%, mean ± SD)
58.04 ± 14.55
83.95 ± 9.28
58.97 ± 13.08
98.78 ± 2.03
10th percentile* (%)
(N = 13)
(N = 14)
Exposure reduction (%, mean ± SD)
61.38 ± 15.12
79.47 ± 7.50
10th percentile* (%)
* Estimated assuming a normal distribution.
Also considering the results of Ludwigs et al. (2007), Brühl et al. recommend the use of seed-specific dehusking factors in risk assessment for granivorous mice such as wood mouse.
It is doubtful to what extent results from individually caged mice in the laboratory may be extrapolated to mice in the wild. It must be assumed that stressors such as predation risk and competitive interference will apply under natural conditions, possibly reducing the frequency and efficiency of dehusking. To account for these uncertainties it is considered that the 10th percentiles of the exposure reductions found by Brühl et al. (2011) may be used within the Northern Zone for higher tier risk assessment of wood mice feeding on cereal grain, maize or sunflower seeds. Assuming a normal distribution, the 10th percentiles have been estimated and are included in Table 4.4.
EFSA (2009) recommends that dehusking factors are not routinely applied in risk assessment. If dehusking is to be considered in a higher tier risk assessment, except for the wood mouse cases specified above, case-specific evidence must be provided that dehusking actually plays a role under field conditions for the focal species in question, and experimental data must be available for the relevant type of seed. In particular, the use of dehusking factors for weed seeds will not be accepted without case-specific experimental evidence.
It is not known to what extent dehusking is triggered solely by the structure of the seed or to what extent impalability of a seed treatment also plays a role. Also for this reason, particular care should be taken when risk assessment is performed for seed treatments with a high toxicity per single seed.
Especially for birds, a risk assessment for a dehusking species shall always be accompanied by an assessment for a second species that does not dehusk (EFSA 2009).