Welcoming the fact that environmental protection has become an issue for the ICRP and appreciating the vast amount of scientific work at the basis of this report, I have some basic comments. More general remarks §25-26: Definition of RAP is clearly individual based, not population or ecosystem based and hence the same system as for human protection is used. To me the approach taken, that of selecting Reference Animals and Plants "to provide a staring point for the assessment of radiation exposure, radiation dose, and possible dose responses" (§14) with the final aim to derive Dose Consideration Levels, is a step back from the approach taken by e.g. EDEN, ERICA, EA R&D 128, RESRAD-BIOTA also mentioned in this report. These approaches have • A far larger set of dose-effects data which were generally thoroughly screened: why not build on these and take them further • A larger set of reference organisms (rather geometries) and potential to adapt geometries (flexible dosimetry in ERICA) • More exposure situations and parameter sets for assessing transfer of radionuclides in food web (be it not over-exhaustive) • Derived benchmarks based on an approach accepted in ecotoxicology: Species Sensitivity Distribution Coming back to that last point, just selecting a few RAP, one can never use the Species Sensitivity Approach to derive benchmarks. The Species Sensitivity Distribution is also the approach taken in ecotoxicology to set benchmarks for conventional contaminants (EC Technical Guidance Documents). The SSD approach has its limitations (e.g. most of the dose effect data used in the SSD are for individuals, seldom for population, never at higher hierarchical level, seldom for real chronic exposure), but at least the radiological sensitivity (dose effect for a specific or a number of endpoints) of a series of organisms within an ecosystem is considered together and a global benchmark is derived. Therefore also, the derivation of the preliminary Dose Consideration Levels (DCL) in this document had to be derived based on several times natural background. But as the natural background at some locations (10 mSv/y and more) is not the benchmark for humans it may not be for some animal species either. Moreover, though put in Tables, in my opinion there is hardly any link between RAP (and their exposure scenarios) and background radiation levels. There are no exposure scenarios given in the derivation of DCL so how applicable are they? (moreover most data used to derive the Tables 8-11 in Chapter 6 were based on data of non-RAP organisms: therefore better put in title e.g. better amphibian instead of 'frog'.) I can understand that one would like to simplify faced with an enormous variety of living animals and plants, but by selecting a few species for assessing relation exposure to dose and dose to effect, one is directly faced with several other limitations directly linked to the approach taken: • By having detailed and traceable information on exposure to dose and dose to effect for a specific RAP, one may perhaps better understand exposure-effects link for that specific RAP, but these data are all individual based and will not learn anything on population/ecosystem effects. Moreover, if no clear indication is given how one should interpret information about RAP for another organism of interest which very likely has another life history, consumption habits and related exposure scenario. • No data available for the specific RAP: e.g. for "Duck" there are hardly any data: this questions the whole approach • Representativeness: an 'Eagle' is not a 'Duck': they have different life histories, consumption habits, perhaps another radiological sensitivity, exposed to other pathways, and it would not be very convincing if expected dose to "duck" would be below the DCL when the target bird was an eagle. Why come one step back from the 'flexible dosimetry model' used in the ERICA-tool? Moreover, geometry based dosimetry is just one source of variation. The exposure situations is another one. And another, not considered in this document, the transfer of radionuclides in the environment and through the food web which will affect external dose rate and radionuclide body burden. With information on DCFs alone, exposure cannot be assessed. Having gone through the document (some parts more in debt, others more superficially) I really wonder how this document will be taken further in the sense that it will be useful for endusers. Will there be any link with the work that has been previously done (certainly since some of the authors have been involved in e.g. ERICA). Some more specific remarks §29 stresses what RAP are not intended to be: they are not necessarily object of protection and not intended to serve as 'sentinel 'species, in the sense such that it is considered that if such species are protected then other types will also be protected. However, §12 states "Equally, they (RAP) could be used to serve as a screening mechanism for excluding the need for further detailed study." With regard to this statement, see remark on representativeness. §10 is quintessential to approach of RAP but (1) can this flow of reasoning be materialised due to lack of data?; (2) for a given ecosystem and exposure situation exposure to these RAP may be irrelevant but other organisms may be threatened because they have other life history, other radiological sensitivity, … §16 I find criteria for selecting RAP not too convincing In formulating the criteria for selection of the RAP an environmentally important one 'key ecological functioning' is missing. Why in criteria like emphasis on animals and plants relevant for fisheries, agriculture, and forestry: are these animals and plants not already affected mostly by man? §27: states: 'Thus, again bearing in mind that the primary purpose is to use the RAP to relate exposure to dose and dose to effect, it should be possible to adapt the basic data in relation to, for example, the marine flatfish that of a similar fish in an estuarine situation …. There is no indication given on 'how this can be made possible', what the limitations are, …. Table 3: How can basic population characteristics of RAP affect the guidelines? No info given on how one should go from effect on individual to (non) effect on population. I know this is not simple, but a rather detailed table is provided and its further use/implication is not given. There are some spelling errors but those will be found I expect.