Calculation of external climate costs for food highlights inadequate pricing of animal products

Although the agricultural sector is globally a main emitter of greenhouse gases, thorough economic analysis of environmental and social externalities has not yet been conducted. Available research assessing agricultural external costs lacks a differentiation between farming systems and food categories. A method addressing this scientific gap is established in this paper and applied in the context of Germany. Using life-cycle assessment and meta-analytical approaches, we calculate the external climate costs of foodstuff. Results show that external greenhouse gas costs are highest for conventional and organic animal-based products (2.41€/kg product; 146% and 71% surcharge on producer price level), followed by conventional dairy products (0.24€/kg product; 91% surcharge) and lowest for organic plant-based products (0.02€/kg product; 6% surcharge). The large difference of relative external climate costs between food categories as well as the absolute external climate costs of the agricultural sector imply the urgency for policy measures that close the gap between current market prices and the true costs of food.


1.
A) "Authors mention that they used meta-analysis to estimate the difference between the conventional and organic farming. The studies they used are listed in Table 2 and it is the only place where authors inform about the literature they used to estimate emissions from organic farming. These papers are missing in the reference list." Response: Thank you very much for your attentive comment! We added a more detailed version of table 2 in the SI section, where all literature is referenced. The missing references shall also be listed here in occurring order (and with relating reference number) in the manuscript:

B) "
Authors also do not mention the key emission values they found in the literature, how they were estimated (e.g. empirical data or models), regional coverage, to what degree the studies covered the broader and specific food categories that the authors used, and whether the studies were consistent with the boundaries (i.e. the direct and indirect emissions) selected by the authors." Response: Thank you for this very plausible notation. We have chosen the relevant papers according to our system's boundaries, but it is of course important to make the comparability between used studies, their eligibility, as well as their stated emission values clear for the reader. Therefore, we have enhanced concerning 1 The specific regional coverage was not stated in all studies. Locations are stated as precisely as possible. 2 We have excluded the in underlying study (Aguilera et al. 2011a) observed food category 'rice' for this assessment as it is an irrelevant product for the assessment of the German agricultural sector. 3 When there was more than one food category assessed in one study, we weighted them equally to not interfere with the weighting system between the studies. 4 In GEMIS 'field vegetables' constitutes a collective term describing vegetables that are grown in the open air. This form of cultivation is in contrast to the horticultural cultivation of vegetables which uses greenhouses, foil tunnels or other artificially protected areas. 5 We have excluded the in underlying study (Aguilera et al. 2011b)observed food categories 'subtropical fruit trees', 'tree nuts', and 'olives' as they are irrelevant products for the assessment of the German agricultural sector. 6 Rotation includes winter wheat, potatoes, beans, cabbage, and spring/winter barley. 7 Rotation includes potatoes, winter wheat, sunflower, winter rye, and maize. 8 Even if sunflower is irrelevant to the assessment of the German foodstuff it is, however, crucial for the underlying crop rotation and farming processes and was therefore not excludable from assessment. Reitmayr et al. (1995) (as quoted in Stolze et al. 2000, p. Tuomisto et al. (2012) and Flessa et al. (2002) explicitly state that the production of farm buildings is not considered. However, as far as it was comprehensible, all other studies have similarly not included assessment of housing production. 10 Production of fertilizer was considered; other indirect inputs for precursors like pesticides and seeds were not included as they were considered negligible; infrastructure (machines and buildings) was not included. This studies system boundaries are least in line with our assessment scope but are still comparable due to the explanation as to why certain processes were excluded. 11 As Bos et al. (2007) resports "GHG emissions per ha on the conventional dairy farms are 65% higher than on the organic model farms." (p.3). We set organic as 100% and conventional as 165%. 12 Authors refer to cradle to grave approach when introducing to the topic of LCA. They continue although with cradle to farmgate assessments of nitrogen surpluses, for example. The input data does also not include processes after farmgate. Therefore, we find this approach to be comparable with cradle to farmgate.

2.
"It is also not clear, how authors estimated the emissions from organic farming based on these studies, and what was the level of uncertainty in the estimated values. I would like to see this description very clearly because it is crucial information on which the study is founded." Response: Thank you for pointing this out. We understand that this is an important part, which we did not elaborate clearly enough in the original version. When evaluating the regarding studies the following limitations arose: Due to the varying estimation methods of considered studies, only four out of twelve papers report measures of deviation (Aguilera et al. 2015aand b, Thomassen et al. 2008, Basset-Mens, et al. 2005, Basset-Mens et al. 2005; Two studies give ranges for their found emission values (Tuomisto et al. 2012, Reitmayr et al. 1995; Four papers mention ranges or deviation for some input data but not their finally retrieved emission value (Casey et al. 2006, Flessa et al. 2002, Dalgaard et al. 2006, Haas et al. 2001; And lastly, three results are solely stated as definite values with no information about the uncertainty of these values (Cooper et al. 2011, Küstermann et al. 2008, Bos et al. 2007).
Because of this methodological inconsistency throughout the studies we found an adequate inclusion of statistical means like deviations not possible and therefore used another method of weighting the studies' results for the calculation of an average. The description of this approach is described in more detail now. The adjusted section now reads as follows: Subsection 6.2.1, page 26 f.: "As the selected studies are based on geophysical measurements and not on inferential statistics, a weighting based on the standard error of the primary study results like in standard meta-analysis 81 was not possible. We aimed for a system that weights the underlying studies regarding their quality and therefore including their results weighted accordingly in our calculations. Within the scope of classic meta-analyses 82 the studies' individual quality is estimated according to their reported standard error (SE), which is understood as a measure of uncertainty: the smaller the SE, the higher the weight that is assigned to the regarding source.
As, due to the varying estimation methods of considered studies, a majority of considered papers does not report measures of deviation for their results. These state definite values; therefore, there is no information about the precision of the results at hand. Against this background we have decided to use a modified approach to estimate the considered papers' qualities 83 . Following van Ewijk et al. 84 and Haase et al. 85 we apply three relevant context sensitive variables to approximate the standard error of the dependent variable and thereby evaluate the quality of each publication: The newer the paper (compared to the timeframe between 1968 and 2018) the higher we assume the quality of reported results. The more often a paper was cited per year (measured on the basis of Google Scholar) the higher the paper's reputation. The higher the publishing journal's impact factor (measured with the SciMago journal ranking) the higher its reputation and therefore the paper's quality. For every paper the three indicators publishing year (shortened with PY in Table 2), citations/year (CY), and journal rank (SJR) rank a paper's impact on a scale from 1 to 10, where 1 describes the lowest qualitative rank and 10 the highest. The sum of these three factors (SUM) then determines the weight of a paper's result in the mean value (WEIGHT). The papers' reported emissiondifferences between organic and conventional (diff. org/conv) are weighted with the papers' specifically calculated WEIGHTS and finally aggregated to the emission difference between both systems.
With this approach we weight results of qualitatively valuable papers higher and are therefore able to reduce the level of uncertainty in the estimated values because standard errors coulddue to inconsistencies in the underlying studies -not be used."

3.
"Authors have cited several of their previous works on which the current study is built upon. Unfortunately, it was very difficult to assess those studies. Particularly the reference 16 is only an abstract accepted for a conference. It makes it very challenging to know how the current approach is built upon the author's previous work and compare them." Response: We understand the reviewer's critique regarding this point. To respond to this comment, we have placed one of the references in another, newly formulated section. Another reference has been deleted. In detail, we proceeded as follows: In the original version, we had referred to our previous work four times:  Michalke, A., Fitzer, F., Pieper, M., Kohlschütter, N. & Gaugler, T. (2019) Response: This indeed is an aspect, that should be more clearly addressed. In the original manuscript we wrote that "all resource inputs and outputs during production up to the point of selling by the primary producer are considered." This does also include all emissions of transport along the whole value chain. Upstream supply chains (e.g. the production of fertilizer) and the emissions thereof are consistently viewed as part of the value chain. In the handbook for GEMIS (Fritsche, Schmidt. Handbuch zu GEMIS. 2008) it says about the (in this case activated) 'global switch' of considering all transport processes: "If this switch is set, then all transports (also in upstream chains) are considered in the emission calculation of GEMIS. The global switch 'non-stationary transport' now determines whether stationary (non-stationary) transport processes are included or not. Non-stationary transport processes are ship, truck, train transports etc., while stationary transport processes are e.g. power lines and pipelines" [own translation from German to English].
Additional to transport emissions, we also consider emissions linked to the preliminary building of relevant infrastructure to be part of production inputs as we quantify the categories' emission values. Furthermore, we enhanced our calculations with including emissions from land use change (LUC) in the considered data. These emissions were not considered in GEMIS data. We therefore calculated those according to the frequently used method of Ponsion and Blonk (2012).
To meet the objection of the reviewer we have now removed previous ambiguity. These additional clarifications of the system boundaries are now added in the manuscript (new text passages are underlined): Subsection 6.2.1, p. 22 ff.: "This means that we consider all resource inputs and outputs during production up to the point of selling by the primary producer ("farmgate"). This includes emissions from all productionrelevant transports as well as emissions linked to the preliminary building of productionrelevant infrastructure. We specify that for animal products, emissions from feed production, as a necessary resource input, are assigned to these animal products. Such emissions naturally should include LUC emissions. LUC emissions are of negligible proportion for locally grown products, as agricultural land area is slightly decreasing in Germany 39 . Thus, we have to focus solely on imported products; on imported feed for conventional animal and dairy products to be precise. Organic feed is not considered as article 14d of the EU-Eco regulation stipulates that organic farms have to primarily use feed which they produce themselves or which was produced from other organic farms in the same region 76 . Even stricter rules are set by many of the organic farming associations in Germany, such as Bioland, Naturland or Neuland, that ban soymeal from Latin America completely 77 . We assume that the emissions that could possibly be caused by organic farming in Germany through the import of feed constitute a negligibly small fraction of the total emissions of a product. Thus, no LUC emissions are calculated for organic products. For conventional products we calculate LUC emissions by application of the method of Ponsioen and Blonk 38 . This method allows the calculation of LUC emissions for a specific crop in a specific country for a specific year. With regards to the year, we apply our reference year 2016. With regards to crop and country one has to keep in mind that in the case of Germany, the net imports of feed are the highest for soymeal, followed by maize and rapeseed meal, making up over 90% of all net positive feed imports 78 . Maize and rapeseed meal are both imported mainly from Russia and Ukraine (93% and 87% of all imports 79 ) . Taken together, the crop area of Russia and Ukraine is decreasing by 150,000 ha/year (data from 1990-2015 was used 80 . Following Ponsioen and Blonk 38 , we thus assume that there are no LUC emissions of agricultural products from these countries. This leaves us with soymeal, of which 97% are imported from Argentina and Brazil. We thus calculate LUC emissions of soymeal for Argentina and Brazil respectively. Data is used from Ponsioen and Blonk 38 , except for data of the crop area, where updated data from FAOSTAT is used in order to match the reference year. We then weigh those country-specific emission values according to their import quantity. This results in 2.54 kgCO2eq/kgSoymeal. To incorporate this value into the conventional emission data from GEMIS, we map the LUC emissions onto all the soymeal inputs connected to the food-specific products."

5.
"As authors already mention that yield in organic production is lower than the conventional, it may be argued that more land area will need to be cultivated for the same amount of production. However, this study does not cover the emissions resulting from the land-use conversion that will be required for this. I understand that incorporation of land-use change may complicate the study. At the same time, even if we assume that no LUC occurs in Germany, I may argue that the import of food will increase. Given that emissions that occur outside Germany are excluded in the study, it might be underestimating the emissions." Response: This is a very good argument that needs to be addressed thoroughly. In the original manuscript we mentioned the alleged effect of increasing imports that might result from a shift from conventional to organic farmland. However, we did not go into detail there. To correct for this informational insufficiency, we edited part of the discussion section to emphasize that an internalization of the external costs of each food category would prevent rising emissions potentially resulting from a widespread application of organic agriculture. That is because the prices of the most resource intensive products (which are animal-based foodstuff from both conventional and organic farming) would significantly increase and thereby lead to a reduced demand of such products due to price elasticity of demand of so called "normal goods" (which also include the examined foodstuff). The associated extensive land area of these products thus would become available for organic agriculture. Furthermore, there is evidence that a shift from conventional to organic practices would indeed be beneficial for the ecosystem services and long-term efficiency provided by the particular land area (Reganold et al. (1987), Reganold and Wachter (2016)).
A scenario-analysis is, however, not part of our study, in which we solely aim at examining the status quo. We therefore do not include a hypothetical emission increase from a shift to organic farming in our calculations.
Thank you, once again, for pointing out that this aspect of our paper needed more clarification. The adjusted section now reads as follows (new text passages are underlined): Section 4, p. 16: "Further doubt towards a transition to organic farming was spread by Smith et al. 52 who rightfully addressed the potential increase of emissions resulting from a complete transition from conventional towards organic farming, given consumption patterns stay the same. These increases are thought to result from a higher amount of imported food, due to lower (regional) yields from organic farming. The financial incentives of internalization presented in our paper and the associated changing consumption patterns, however, pose a solution to these identified problems. Due to price elasticities of demand for food products (which are consistently regarded as 'normal goods' in economic literature), appropriate pricing of food would make products of organic production more competitive compared to their conventional counterparts 53 : customers would increasingly opt for organic foodstuff due to the lowered price-gap between the two options. This could potentially press the boundaries of land use for agriculture as organic practices mostly require more land than conventional systems due to lower yields 54-56 . However, our results suggest an increase in the prices of animal-based products to a significantly larger extent than the prices of plant-based products. The presumed consequential decline of animal-based product consumption would free an enormous landmass currently used for feed-production. Further expansion of area-intensive organic agriculture would subsequently be made possible 57 . Furthermore, there is evidence that a shift from conventional to organic practices would indeed be beneficial for the ecosystem services and long-term efficiency provided by the particular land area 7,58 ." Due to the reviewer commenting on the significance of emission from land use conversion we have now enhanced our calculations with data from LUC (compare response to reviewer's comment 4). However, these calculations still do not account for possible developments in the future. Our study focuses solely on the status quo of today's agricultural conditions.

6.
"The authors do not mention whether they take into account the temporal changes in emissions that can occur as the organic farming is followed on a continuous basis. It is very likely that the yield gaps between the conventional and organic farming may increase or decrease, and the same can happen with the soil-borne GHG emissions and soil carbon sequestration rates. Ideally, authors should account for uncertainties arising due to these, but if it is not possible due to the limitation of the data, these effects need to be discussed as a part of the uncertainty." Response: Thank you for the reasonable suggestion that we need to clarify this point in the original version. As already stated in the answer to the previous commentary (comment 5), we solely aim at examining the status quo of German agricultural practices and have not investigated the temporal effects. We now point to this fact in our paper (Please refer both to the response to comment 5 for details and to (the new) subsection 6.4 entitled "Dealing with uncertainties".) Following the suggestion of the reviewer, we have also added this passage (p. 19): "If one takes into account the temporal change in yield difference which would result by converting farms from conventional to organic farming, there is scientific consensus that the yield gap will decrease over time (Schrama et al. 2018, Sander andHess 2019). Comparative studies between different cultivation methods also show that organic farming has lower soil-borne GHG emissions and higher rates of carbon sequestration in the soil (Scialabba andMüller-Lindenlauf 2010, Muller et al. 2017). Soil degradation resulting from conventional systems would slow down or could even be reversed by changing to organic farming (Küstermann et al. 2008, Azadi et al. 2011." In order to further address possible uncertainties, we have decided to follow the reviewer's suggestion and also included a subsection for the discussion of uncertainties and assumptions that have been made throughout the study. Please find this as follows: Subsection 6.4, p. 32 f.: "6.4 Dealing with Uncertainties Due to the interdisciplinarity and novelty of our study we connect several methodological approaches and refer to various sources for data. Against this background we had to accept some uncertainties while assembling and using the developed framework for our calculation. The studies included in our meta-analytical approach of calculating the difference between organic and conventional emission values, for one, are not fully consistent in the methodologies each of them uses (refer to SI.1 for details). Furthermore, from the results of all included studies it is apparent that there exists a wide range of emission differences between the farming practices, depending on the papers scope and examined produce 91 . We attempted to account for this through weighting the studies according to their fit regarding the object of research (compare subsection 6.2.1). Due to insufficient availability of data for the emission differences between organic and conventional on the basis of each narrow category an average for the emission difference was used. This possibly results in imprecisions during the internalization of the external costs on the level of all narrow categories. Therefore, we focus on the aggregated broad categories as this uncertainty can be evaded here. Furthermore, the in literature reported price factor for CO 2 -equivalents is volatile over time impacting the results of this paper. It is to be expected that the external costs of GHG emissions are likely to rise in the future (compare section 2). Also, our study's scope is confined to the assessment of the current production situation within the German agricultural sector. Therefore, we do not account for future developments regarding a changing agricultural production landscape after internalization of the accounted external costs. We do, however, discuss possible effects on demand patterns as well as the environmental and social performance of the agricultural sector in section 4. Regarding the incorporated LUC emissions, there appears to be a lacking scientific consensus on a general method of calculation for such emissions 38,92-94 . We thus want to emphasize that these additional emissions should be treated with caution and are thereby displayed separately from the other data." II. Format and language 1. "The format is not according to the journal, (methods are before the results)." Response: Thank you for this reminder. The methods have now been moved to the end and can be found after the discussion/conclusion chapter. We have sensibly changed all necessary textual references according to the new format.

2.
"Authors use words like 'external effects' and external costs, both of which I understand mean externality? I will suggest using the same term throughout to make it easy for the reader." Response: Thank you very much for this advice. Indeed, we used various terms synonymously until now making it rather difficult to understand for the reader. Therefore, we now use the terms in the following nuanced manner: When we talk about volume-related externalities measured in CO2-equivalents, we now consistently use the term "externalities". If it is talked about the follow-up costs measured in monetary units resulting from CO2-eq. emission, we now refer to them as "external costs".
An exception to this is the conscious use of the term "external climate costs" in the title as well as in the abstract of the paper. The addition of the word "climate" seems appropriate to us in these two cases as the reader should be informed briefly and concisely at first glance that the article addresses climate follow-up costs.
Furthermore, as we were quality checking our work with other colleagues in the field throughout this revision process it has come to our attention that the use of the term "production systems" when referring to either conventional or organic agricultural practices can lead to some confusion. We were advised that one might think about processing steps after the farmgate when reading this term. Therefore, we have changed all mentions in the text to "farming systems" to clarify referral to all steps before the farmgate.

3.
"Page 3: what is chapter 3.2.1? I would prefer using 'section' instead of 'chapter'." Response: We agree with the reviewer's suggestion and now use the term 'section' when referring to a subchapter (e.g. 'section 3.2.1'). We refer to a "section" as such only when referring to a chapter as a whole (e.g. section 5').

4.
"Page 3 paragraph 2, authors say that it will be technically incorrect to refer to the current approach as LCA and that the term carbon footprint is more appropriate. I, therefore, suggest authors talk about carbon footprint from the beginning." Response: We understand the reviewer's critique relating to our use of the terms "Life Cycle Assessment" (LCA) and "carbon footprint" in chapter 2 (page 3). Indeed, our previous wording was not precise enough. LCA is a general approach for determining environmental impacts. This methodology can be used to determine the amount of various resources and pollutants arising during the production of a good (e.g. water consumption, SO 2 and CO 2 emissions, etc.) as well as their impact (e.g. for climate or human health). As our study focuses on the climate impact of agricultural products, which are quantified using CO 2equivalents, we apply the methods of LCA on the food-specific emission quantities of CO 2equivalents. The term "carbon footprint" is merely a measure of the amount of CO 2 -equivalents emitted. In short, by using Life Cycle Assessment it is possible to determine the carbon footprint of a product, in our case the various established food categories. Against this background, we propose that the description of LCA remains in the paper. Instead, we suggest that the expression "carbon footprint" (as the term used to describe the quantities of CO 2 -equivalent emissions) should only be briefly discussed now. According to this more precise narrative we have now adapted the text as follows (new passages are underlined): Subsection 6.1, p. 19: "The quantification includes the determination of food specific GHG emissions -also known as carbon footprints 33 -occurring from cradle to farmgate by usage of a material flow analysis tool. Carbon footprints are understood within this paper in line with Pandey et al. 63 where all climate relevant gases, which in addition to CO 2 include, methane (CH 4 ), and nitrous oxide (N 2 O), are considered. Their 100-year CO 2 -equivalents conversion factors are henceforth defined as 28 and 265, respectively 69 ."

5.
"page 8: Authors mention that they use 11 food categories but name only 10." Response: Thank you for your careful reading. We accidently forgot to list 'cereals' on the plant-based side of the eleven narrow food categories. This has been corrected and can be seen on page 9, subsection 3.1.

6.
A) " "I suggest the Authors to rewrite the abstract with a clear presentation of the main motivations, objcetives, methodologies and main inisghts (avoidind in the abstract terms as mark-up, maybe not a familiar term for the general readers of these topics).
Response: Thank you for this attentive comment. The abstract of course is crucial for the success of an article and therefore we have put careful effort into editing it according to your advice. We have now structured it along your suggested four subheadings (main motivations, objectives, methodologies, main insights) and have exchanged the unfamiliar term 'mark-up' with the more generally known and applicable 'surcharge'. For faster reference we include the edited abstract in the following:

Abstract, p. 1:
"Although the agricultural sector is globally a main emitter of greenhouse gases, thorough economic analysis of environmental and social externalities has not yet been conducted.

Available research especially lacks differentiation between farming systems and various food categories. A method addressing this scientific gap is established in this paper and applied in
the context of Germany. Using LCA and meta-analytical approaches, we calculate the external climate costs of foodstuff. Results show that external greenhouse gas costs are highest for conventional animal-based products (2.41€/kg product; 146% surcharge on producer price level), followed by conventional dairy products (0.29€/kg product; 108% surcharge) and lowest for organic plant-based products (0.02€/kg product; 5% surcharge).
The large difference of relative external climate costs between food categories as well as the absolute external climate costs of the agricultural sector imply the urgency for policy measures that close the gap between current market prices and the true costs of food." Furthermore, as we were quality checking our work with other colleagues in the field throughout this revision process it has come to our attention that the use of the term "production systems" -when referring to either conventional or organic agricultural practices -can lead to some confusion. We were advised that one might think about processing steps after the farmgate when reading this term. Therefore, we have changed all mentions in the text to "farming systems" to clarify referral to all steps before the farmgate.

2.
"The literature review is weak. I suggest improve and update significantly this part."

Response: Thank you a lot for this clear suggestion. As externality assessment in the agricultural context is a controversial topic and scientifically approached from various angles, it is indeed important to provide a profound overview of available sources and methodologies.
We have enhanced the literature review following your note. Please find the edited version here with all new inputs underlined: Section 2, p. 3 ff.: "2. Research aim and literature review […] There has been some scientific engagement previously, as Pretty et al. 14 set the scene for agricultural externality analysis at this century's beginning: they were able to record significant environmental impacts of agriculture at the overall societal level in monetary terms for the United Kingdom. This approach was translated for other regions subsequently, with calculations of agricultural external costs for the United States and Germany 8,15 . However, these first external cost assessments, with their characteristic top-down approaches, did not link specific causal emission values with said costs. Yet, a bottom-up approach for monetizing externalities of country-specific agricultural reactive nitrogen emissions was later developed 16 and subsequently used for an external cost assessment of Dutch pig production 17 . Despite, assessments concerning important agricultural emissions, which comprehensively differentiate between a variety of food categories, are yet missing. There exists a range of studies that quantify food-category-specific GHG emissions 18-21 while other studies disclose the difference of climate effects from conventional and organic practices (see table 2 for references). Monetizing such emissions, however, has been done for constituent food categories only 22 . An encompassing connection between the quantification and monetization of GHG emissions differentiated by food categories and farming systems is what seems to be lacking in the currently available literature.
Congruent to methodological differences for monetizing agricultural greenhouse gases, there are also differences in the estimation level of greenhouse gas costs, which especially in the past have been vast. Prices per tonne of emission at the stock market, for example, are as low as 3.92 to 8.33 € during this study's reference year 23 , whereas the IPCC in their last report of 2019 suggest a price between 135 and 5,500 $ per tonne of CO2-equivalents 24 . The German Federal Environmental Agencies (UBA) suggestion for the damage costs of GHG emissions also rose within the last years: in 2010 they suggested a rate of 80 € per tonne of CO2-equivalents 25 , whereas this increased to 180 € per tonne in 2019 26 . These great differences can be explained with methodological inconsistencies or a difference in approach, for example due to consideration of either damage or abatement costs. Furthermore, the price is expected to rise in the future 27 , for example, describes that it must be "exceeding $400 per tonne by mid-century" (p. 1271). […] LCA has developed as a commonly used tool for examining material and substance flows of diverse products. Its origins lie in the analysis of energy flows but it is now commonly used to assess various processes 29 . […] […] Especially during the production of animal-based foodstuff livestock related gases like methane or nitrous oxide significantly contribute to the overall GHGs emitted 4 .
[…]" Furthermore, we have clarified the use of the terms "carbon footprint" and "Life Cycle Assessment (LCA)" and put these in relation to the current scientific consensus. As our study focuses on the climate impact of agricultural products, which are quantified using CO 2equivalents, we apply the methods of LCA on the food-specific emission quantities of CO 2equivalents. The term "carbon footprint" is merely a measure of the amount of CO 2equivalents emitted. In short, by using Life Cycle Assessment it is possible to determine the carbon footprint of a product, in our case the various established food categories. Against this background, we propose that the description of LCA remains in the paper. Instead, we suggest that the expression "carbon footprint" (as the term used to describe the quantities of CO 2 -equivalent emissions) should only be briefly discussed now. According to this more precise narrative we have now adapted the text as follows (new passages are underlined): Subsection 6.1, p. 19: "The quantification includes the determination of food specific GHG emissions -also known as carbon footprints 32 -occurring from cradle to farmgate by usage of a material flow analysis tool. Carbon footprints are understood within this paper in line with Pandey et al. 63 where all climate relevant gases, which in addition to CO 2 include, methane (CH 4 ), and nitrous oxide (N 2 O), and their respective 100-year CO 2 -equivalents conversion factors of 28 and 265, respectively, are considered." Besides these adaptations we have overall embedded the paper more strongly into current scientific literature throughout the whole text, which now refers to 94 -instead of the former 75 -references. Please find the answer to comment 7 for a more specific comparison between our findings and other works.

3.
"In the current section 3 you need to support your options with other scientific papers already published. For example "....quantification and second the monetization of external effects from GHGs, visualized in...". Why this approach and not other?. Other example, "...analysis tool GEMIS (Global Emission-model for Integrated Systems) 25 is used which offers...". There are other approaches. Why not others? On other words, this section is too much descriptive in a part where you make several options." Response: We thank you for this very understandable commentary. Within the text we have now made several changes to explain in greater detail on which basis we designed the framework and why we decided on the methodological options that shape our work. As we have now rearranged the order of the manuscript according to the journal's guidelines the corresponding text is now placed in section 6. Please find according additions underlined in the following: Subsection 6.1, p. 18 ff.: "6.1 Outline of the method We differentiate between two steps within this method of calculating food-category specific externalities and the resulting external costs. These are first the quantification and second the monetization of externalities from GHGs (visualized in figure 3). We use this bottom-up approach following the example of Grinsven et al. 16 who conducted a cost-benefit analysis of reactive nitrogen emissions from the agricultural sector. This two-stepped method also allows the adequately differentiated assessment for GHG emissions of various food categories.
The quantification includes the determination of food specific GHG emissions -also known as carbon footprints 32 -occurring from cradle to farmgate by usage of a material flow analysis tool. Carbon footprints are understood within this paper in line with Pandey et al. 68 where all climate relevant gases, which (in addition to CO 2 ) include methane (CH 4 ) and nitrous oxide (N 2 O), are considered. Their 100-year CO 2 -equivalents conversion factors are henceforth defined as 28 and 265, respectively 69 . Here, the material flow analysis tool GEMIS (Global Emission-model for Integrated Systems) 37 is used which offers data for a variety of conventionally farmed foodstuffs. As GEMIS data focuses on emissions from conventional agricultural systems we carried out the distinction to organic systems ourselves.
We determined the difference in GHG emissions between the systems by applying metaanalytical methods to studies comparing the systems' GHG emissions directly to one another.
Meta-analysis is commonly used in the agricultural context, for example when comparing the productivity of both systems 55-57 or their performance 7 .
For better communicability we first aggregate the 11 food specific datasets given in GEMIS to the broader food categories 'plant-based', 'animal-based', and 'dairy' by weighting them with their German production quantities (cf. section 3.1).
[...] Figure 3: Visualization of the method of quantifying and monetizing product specific externalities; in the case of Germany, emission data was obtained from GEMIS 38 , we used production data from the German Federal Statistical Office 37 and AMI 42,64 , and calculated the emission difference between organic and conventional production based on a meta-analytical approach (see subsection 3.2.1); the category specific emission data was calculated on the basis of these input data; the emission cost rate was obtained from UBA 26 ; the category specific external costs were determined on the basis of the previously developed price-quantity-framework (see subsection 3.2.2). Source data are provided as a source data file. " Furthermore, we now elaborate why we decided to use the tool GEMIS for our approach. Please find the corresponding passage as follows with all updates underlined: Subsection 6.2.1, p. 21: "Starting with the data on food-specific emissions, GEMIS is used because of its large database of life-cycle data on agricultural products with a geographic focus on Germany.
GEMIS is a World-Bank acknowledged tool for their platform on climate-smart planning and draws on 671 references which are traced back to 13 different databases. The German Federal Environmental Agency uses GEMIS as a database for their projects and reports establishing it to be an adequate tool for the German context especially 72,73 . This tool is provided by the International Institute for Sustainability Analysis and Strategy (IINAS). GEMIS offers a complete view on the life cycle of a product, from primary energy and resource extraction to the construction and usage of facilities and transport systems." The response to the next comment (4.) also entails more detail on our use of data and methodologies, concerning subsection 6.2. We have made an effort to link our approach to other scientific literature. Please find more information in the following response.

4.
"On the other hand, the Authors need to be more specific in this section. For example, "...individual countries was accessible, EU-data is used.". What does this mean?" Response: Thank you very much for your comment. You are absolutely right, that subsection 6.2 (formerly 3.2) contains some formulations about the specifics of the data that might be confusing for readers. We clarified these passages. You can find a list of the adjusted passages below (new passages are underlined) along with a short explanation as to why we have specified the regarding passages: The following explains about our use of GEMIS: Section 6.2.1, p. 21: "Starting with the data on food-specific emissions, GEMIS is used because of its large database of life-cycle data on agricultural products with a geographic focus on Germany.
GEMIS is a World-Bank acknowledged tool for their platform on climate-smart planning and draws on 671 references which are traced back to 13 different databases. The German Federal Environmental Agency uses GEMIS as a database for their projects and reports establishing it to be an adequate tool for the German context especially 72,73 . This tool is provided by the International Institute for Sustainability Analysis and Strategy (IINAS). GEMIS offers a complete view on the life cycle of a product, from primary energy and resource extraction to the construction and usage of facilities and transport systems. In case GEMIS offered no data for Germany for certain foodstuff (this is the case for maize, milk and eggs), data from climatically comparable European countries is used. […] In this study the system boundaries for assessing food-specific GHG emissions span from cradle to farmgate. This means that we consider all resource inputs and outputs during production up to the point of selling by the primary producer ("farmgate"). This includes emissions from all production-relevant transports as well as emissions linked to the preliminary building of production-relevant infrastructure. […]" The following specifies the system's boundaries in greater detail than before: Section 6.2.1, page 21 f.: "In this study the system boundaries for assessing food-specific GHG emissions span from cradle to farmgate. This means that we consider all resource inputs and outputs during production up to the point of selling by the primary producer ("farmgate"). This includes emissions from all production-relevant transports as well as emissions linked to the preliminary building of production-relevant infrastructure. We specify that for animal-based products, emissions from feed production, as a necessary resource input, are assigned to these animal-based products." As explained in response to comment 5. we now include the emissions of LUC in our calculations. The reasoning and approach for this is explained in the following: Section 6.2.1, page 22 f.: "Such emissions naturally should include LUC emissions. LUC emissions are of negligible proportion for locally grown products, as agricultural land area is slightly decreasing in Germany 39 . Thus, we have to focus solely on imported products; on imported feed for conventional animal-based and dairy products to be precise. Organic feed is not considered as article 14d of the EU-Eco regulation stipulates that organic farms have to primarily use feed which they produce themselves or which was produced from other organic farms in the same region 71 . Even stricter rules are set by many of the organic farming associations in Germany, such as Bioland, Naturland or Neuland, that ban soymeal from Latin America completely 72 .
We assume that the emissions that could possibly be caused by organic farming in Germany through the import of feed constitute a negligibly small fraction of the total emissions of a product. Thus, no LUC emissions are calculated for organic products. For conventional products we calculate LUC emissions by application of the method of Ponsioen and Blonk 38 .
This method allows the calculation of LUC emissions for a specific crop in a specific country for a specific year. With regards to the year, we apply our reference year 2016. With regards to crop and country one has to keep in mind that in the case of Germany, the net imports of feed are the highest for soymeal, followed by maize and rapeseed meal, making up over 90% of all net positive feed imports 78 . Maize and rapeseed meal are both imported mainly from Russia and Ukraine (93% and 87% of all imports 79 ). Taken together, the crop area of Russia and Ukraine is decreasing by 150,000 ha/year (data from 1990-2015 was used 75 ). Following Ponsioen and Blonk 38 , we thus assume that there are no LUC emissions of agricultural products from these countries. This leaves us with soymeal, of which 97% are imported from Argentina and Brazil. We thus calculate LUC emissions of soymeal for Argentina and Brazil respectively. Data is used from Ponsioen and Blonk 38 , except for data of the crop area, where updated data from FAOSTAT is used in order to match the reference year. We then weigh those country-specific emission values according to their import quantity. This results in 2.54 kgCO2eq/kgSoymeal. To incorporate this value into the conventional emission data from GEMIS, we map the LUC emissions onto all the soymeal inputs connected to the foodspecific products. [...] For this weighting and aggregation step, only production quantities used for human nutrition were considered, thus feed and industry usage of food are ruled out (in contrast to emission calculation, where feed is indeed considered). Besides the German Federal Office of Statistics 36 , source for this data is the German Society for Information on the Agricultural Market (AMI) 41,70 . Only production data for conventional production is used. Thereby we imply equal ratios of production quantities across the food categories. This does not fully reflect the current situation of organic production properties but allows for a fair comparison between emission data of organic and conventional food categories. In table 3 all production data is listed, whereby total production quantities in 1,000t can be found in the right column.
Translating these into percentage shares, the column right to the narrow category's column represents the shares of the specific foods inside the narrow categories, whereas the column right to the broad category's column represents the shares of the narrow categories inside the broad categories. These shares are expressed in formula 2a and 2b (sub-section 6.3.1) by the terms , , , (share in broad categories) and , , , , , (share in narrow categories).

[table 4]
We aggregate GEMIS emission data (q b,n,i,conv ) to narrow (e b,n,conv ) and broad categories (E b,conv ) by multiplying the respective emission data with the shares from table 3 (cf. formula 2a & b, subsection 6.3.1). From these conventional emission values we derive emissions for organic production. For narrow as well as broad categories, the respective conventional emission values are multiplied with the applicable emission-differences 'D b, org/conv ' (cf. table 2). " The next section specifies the meta-analytical approach we used to calculate the differences between the systems of organic and conventional agriculture: Section 6.2.1, p. 25 ff.: "To cover a reasonably relevant period, we decided to search for studies published within the past 50 years (from 1969 to 2018) and could therefore identify fifteen relevant studies, spanning from 1995 to 2015. Four of these studies have Germany as their reference country while the other eleven focus on other European countries (Denmark, France, Ireland, Netherlands, Spain, UK; please consult SI.1 for specifics). The weighted mean of the individual study results amounts to the difference in GHG emissions between the two farming production systems. As the selected studies are based on geophysical measurements and not on inferential statistics, a weighting based on the standard error of the primary study results like in standard meta-analysis 81 was not possible. We aimed for a system that weights the underlying studies regarding their quality and therefore including their results weighted accordingly in our calculations. Within the scope of classic meta-analyses 82 the studies' individual quality is estimated according to their reported standard error (SE), which is understood as a measure of uncertainty: the smaller the SE, the higher the weight that is assigned to the regarding source. As, due to the varying estimation methods of considered studies, a majority of considered papers does not report measures of deviation for their results.
These state definite values; therefore, there is no information about the precision of the results at hand. Against this background we have decided to use a modified approach to estimate the considered papers' qualities 83 . Following van Ewijk et al. 84 and Haase et al. 85 we apply three relevant context sensitive variables to approximate the standard error of the dependent variable and thereby evaluate the quality of each publication: The newer the paper (compared to the timeframe between 1968 and 2018) the higher we assume the quality of reported results. The more often a paper was cited per year (measured on the basis of Google Scholar) the higher the paper's reputation. The higher the publishing journal's impact factor (measured with the SciMago journal ranking) the higher its reputation and therefore the paper's quality.
For every paper the three indicators publishing year (shortened with PY in Table 2), citations/year (CY), and journal rank (SJR) rank a paper's impact on a scale from 1 to 10, where 1 describes the lowest qualitative rank and 10 the highest. The sum of these three factors (SUM) then determines the weight of a paper's result in the mean value (WEIGHT).
The papers' reported emission-differences between organic and conventional (diff. org/conv) are weighted with the papers' specifically calculated WEIGHTS and finally aggregated to the emission difference between both systems.
With this approach we weight results of qualitatively valuable papers higher and are therefore able to reduce the level of uncertainty in the estimated values because standard errors could -due to inconsistencies in the underlying studies -not be used. The results of this metaanalytical approach are listed in table 2 (subsection 3.1), further details can be found in SI (SI.1). The studies considered compare GHG emissions of farming systems in relation to the crop/farm area. However, since our study aims to compare GHG emissions in relation to the weight of foodstuff, we include the difference in yield ("yield gap") between the two farming systems for plant-based products, and the difference in productivity ("productivity gap") for animal-based and dairy products. For plant-based products the yield gap is 117%, meaning that conventional farming produces 17% more plant-based products than organic farming on a given area. This gap was derived from three comprehensive meta studies 54-56 and weighted as just described for the emission difference between organic and conventional farming. For animal-based as well as dairy products the productivity gap could be determined with the same studies used for the meta-analytical estimation of the emission-differences 49-51,85-87 .
The productivity gap is 179% for animal-based and 153% for dairy products. In line with Sanders and Hess 60 the yield (or productivity) difference affects the calculation of the food-weight-specific emission difference = / between both farming systems: The yield difference is hereby multiplied with the cropland-specific emission difference . Resulting from this, the emission difference can be formulated as follows: If the yield difference would not be included, emissions from organic farming would appear lower than they actually are as organic farming has lower emissions per kg of foodstuff but also lower yields per area. With formula 1, we adjust for that." The following section explains our used monetary data in better detail: Section 6.2.2, p. 28: "For the pricing of the food categories, we determine the total amount of proceeds that farmers accumulate for their sold foodstuff in € 71 for each category ("producer-price") divided by its total production quantity. Thereby we calculate the relative price per ton for each foodstuff. We solely refer to producer prices as the system boundaries only reach until the farmgate."

5.
"Yet in the section 3, sometimes, it is hard to understand what part of the data was obtained by you and what part was obtained from other sources. For example, only in the subsection 3.2.1. we understand that "...basis of quantity-and emission-trend was conducted in order to align the data with the...". How you did this? And the scientific support?"

Outline of the method
We differentiate between two steps within this method of calculating food-category specific externalities and the resulting external costs. These are first the quantification and second the monetization of externalities from GHGs (visualized in figure 3). We use this bottom-up approach following the example of Grinsven et al. 16 who conducted a cost-benefit analysis of reactive nitrogen emissions from the agricultural sector. This two-stepped method, however, also allows the adequately differentiated assessment for GHG emissions of various food categories.
The quantification includes the determination of food specific GHG emissions -also known as carbon footprints 32 -occurring from cradle to farmgate by usage of a material flow analysis tool. Carbon footprints are understood within this paper in line with Pandey et al. 68 where all climate relevant gases, which (in addition to CO 2 ) include methane (CH 4 ) and nitrous oxide (N 2 O), are considered. Their 100-year CO 2 -equivalents conversion factors are henceforth defined as 28 and 265, respectively 69 . Here, the material flow analysis tool GEMIS (Global Emission-model for Integrated Systems) 37 is used which offers data for a variety of conventionally farmed foodstuffs. As GEMIS data focuses on emissions from conventional agricultural systems we carried out the distinction to organic systems ourselves.
We determined the difference in GHG emissions between the systems by applying metaanalytical methods to studies comparing the systems' GHG emissions directly to one another.
Meta-analysis is commonly used in the agricultural context, for example when comparing the productivity of both systems 54-56 or their performance 7 . […] [ Figure 3] Figure 3: Visualization of the method of quantifying and monetizing product specific externalities; in the case of Germany, emission data was obtained from GEMIS 38 , we used production data from the German Federal Statistical Office 37 and AMI 42,64 , and calculated the emission difference between organic and conventional production based on a meta-analytical approach (see subsection 3.2.1); the category specific emission data was calculated on the basis of these input data; the emission cost rate was obtained from UBA 26 ; the category specific external costs were determined on the basis of the previously developed price-quantity-framework (see subsection 6.2.2). Source data are provided as a source data file.

Input data
In the following, we differentiate between input data for the quantification and for the monetization of external effects. The reference year for this analysis is 2016 and the reference country is Germany, which is listed as the third most affected country in the 'Global Climate Risk Index 2020' Ranking 65 .

1 Input data for quantification
Input data for quantification includes data on the food-specific amount of CO 2 emissions during the conventional farming process from the material flow analysis tool GEMIS 38 . For the meta-analytical methods, used to translate assessed emissions to organic systems, we gather data on the difference in emissions between conventional and organic farming production systems.
[…] We specify that for animal products, emissions from feed production, as a necessary resource input, are assigned to these animal products. Such emissions naturally should include LUC emissions. LUC emissions are of negligible proportion for locally grown products, as agricultural land area is slightly decreasing in Germany 40 . Thus, we have to focus solely on imported products; on imported feed for conventional animal and dairy products to be precise. Organic feed is not considered as article 14d of the EU-Eco regulation stipulates that organic farms have to primarily use feed which they produce themselves or which was produced from other organic farms in the same region 70 . Even stricter rules are set by many of the organic farming associations in Germany, such as Bioland, Naturland or Biopark, that ban soymeal from Latin America completely. We assume that the emissions that could possibly be caused by organic farming in Germany through the import of feed even despite the EU-Eco regulation constitute a negligibly small fraction of the total emissions of a product. Thus, no LUC emissions are calculated for organic products. For conventional products we calculate LUC emissions by application of the method from Ponsioen and Blonk 39 . This method allows the calculation of LUC emissions for a specific crop in a specific country for a specific year. With regards to the year, we apply our reference year 2016. With regards to crop and country one has to keep in mind, that in the case of Germany, the net imports of feed are the highest for soymeal, followed by maize and rapeseed meal, making up over 90% of all net positive feed imports 71 . Maize and rapeseed meal are both imported mainly from Russia and Ukraine (93% and 87% of all imports 72 ) . Taken together, the crop area of Russia and Ukraine is decreasing by 150,000 ha/year (data from 1990-2015 was used 73 ). Following the methodology by Ponsioen and Blonk 39 , we thus assume that there are no LUC emissions of agricultural products from these countries. This leaves us with soymeal, of which 97% are imported from Argentina and Brazil. We thus calculate LUC emissions of soymeal for Argentina and Brazil respectively. Data is used from Ponsioen and Blonk 39 , except for data of the crop area, where updated data from FAOSTAT is used in order to match the reference year. We then weigh those country-specific emission values according to their import quantity. This results in 2.54 kgCO2eq/kgSoymeal. To incorporate this value into the conventional emission data from GEMIS, we map the LUC emissions onto all the soymeal inputs connected to the food-specific products.
For aggregation to narrow categories, we categorize every dataset from GEMIS into one of the eleven narrow food categories. The choice of separation into these specific categories is based on the categorization of the German Federal Office of Statistics 37 from which production data was obtained. According to one category's yearly production quantity, we incorporate every food product into the weighted mean of its corresponding food category.
[…] Only production data for conventional production is used. Thereby we imply equal ratios of production quantities across the food categories. This does not fully reflect the current situation of organic production properties but allows for a fair comparison between emission data of organic and conventional food categories. In table 3 all production data is listed, whereby total production quantities in 1,000t can be found in the right column. Translating these into percentage shares, the column right to the narrow category's column represents the shares of the specific foods inside the narrow categories, whereas the column right to the broad category's column represents the shares of the narrow categories inside the broad categories. These shares are expressed in formula 2a and 2b by the terms , , , (share in broad categories) and , , , , , (share in narrow categories).
[ Table 4] Table 4: Production data [q b,n,i,conv ] for food-specific products and share in broad and narrow categories for 2016 in Germany; production data was obtained from the German Federal Office of Statistics 36 and AMI 41,70 . Source data are provided as a source data file.
We aggregate GEMIS emission data (qb,n,i,conv) to narrow (eb,n,conv) and broad categories (Eb,conv) by multiplying the respective emission data with the shares from table 3 (cf. formula 2). From these conventional emission values we derive emissions for organic production. For narrow as well as broad categories, the respective conventional emission values are multiplied with the applicable emission-differences 'Db, org/conv' (cf. table 2).
With this data we aggregate the above mentioned eleven food-categories to three broad categories: 'plant-based', 'animal-based' and 'dairy'. […] As mentioned before, only data regarding externalities of conventional agricultural production is included in GEMIS and could therefore be aggregated. Nevertheless, by applying metaanalytical methods regarding the percentage difference of GHG emissions between conventional and organic production, we derive emission data for organic production for each of the broad categories (plant-based, animal, and dairy). It has to be noted that LUC emissions are consistently excluded from this procedure. To derive emission differences between organic and conventional farming, research was conducted by snowball sampling from already existing and thematically fitting meta-analysis, by keyword searching in research databases, as well as forward and backward search on the basis of already known sources.
[…] As the selected studies are based on geophysical measurements and not on inferential statistics, a weighting based on the standard error of the primary study results like in standard meta-analysis 74 was not possible. We aimed for a system that weights the underlying studies regarding their quality and therefore including their results weighted accordingly in our calculations. Within the scope of classic meta-analyses 75 the studies' individual quality is estimated according to their reported standard error (SE), which is understood as a measure of uncertainty: the smaller the SE, the higher the weight that is assigned to the regarding source.
As, due to the varying estimation methods of considered studies, a majority of considered papers does not report measures of deviation for their results. These state definite values; therefore, there is no information about the precision of the results at hand. Against this background we have decided to use a modified approach to estimate the considered papers' qualities 76 . Following van Ewijk et al. 77 and Haase et al. 78 we apply three relevant context sensitive variables to approximate the standard error of the dependent variable and thereby evaluate the quality of each publication: The newer the paper (compared to the timeframe between 1968 and 2018) the higher we assume the quality of reported results. The more the paper was cited per year (measured on the basis of Google Scholar) the higher the paper's reputation. The higher the publishing journal's impact factor (measured with the SciMago journal ranking) the higher its reputation and therefore the paper's quality. For every paper the three indicators publishing year (shortened with PY in Table 2), citations/year (CY), and journal rank (SJR) rank a paper's impact on a scale from 1 to 10, where 1 describes the lowest qualitative rank and 10 the highest. The sum of these three factors (SUM) then determines the weight of a paper's result in the mean value (WEIGHT). The papers' reported emissiondifferences between organic and conventional (diff. org/conv) are weighted with the papers' specifically calculated WEIGHTS and finally aggregated to the emission difference between both systems.
With this approach we weight results of qualitatively valuable papers higher and are therefore able to reduce the level of uncertainty in the estimated values because standard errors coulddue to inconsistencies in the underlying studies -not be used. The results of this metaanalytical approach are listed in table 2 (subsection 3.1), further details can be found in SI.
The studies considered compare GHG emissions of farming systems in relation to the crop/farm area. However, since our study aims to compare GHG emissions in relation to the weight of foodstuff, we include the difference in yield ("yield gap") between the two farming systems for plant-based products, and the difference in productivity ("productivity gap") for animal and dairy products. For plant-based products the yield gap is 117%, meaning that conventional farming produces 17% more plant-based products than organic farming on a given area. This gap was derived from three comprehensive meta studies 55-57 and weighted as just described for the emission difference between organic and conventional farming. For animal as well as dairy products the productivity gap could be determined with the same studies used for the meta-analytical estimation of the emission-differences 50-52,78-80 . The productivity gap is 179% for animal and 153% for dairy products. In line with Sanders and Hess 81 the yield (or productivity) difference affects the calculation of the foodweight-specific emission difference = / between both farming systems: The yield difference is hereby multiplied with the cropland-specific emissiondifference . Resulting from this, the emission difference can be formulated as follows: If the yield difference would not be included, emissions from organic farming would appear lower than they actually are as organic farming has lower emissions per kg of foodstuff but also lower yields per area. With formula 1, we adjust for that.

Input data for monetization
[...] Following UBA, these damage costs are analyzed in the following categories: agriculture, forestry, sea level rise, cardiovascular and respiratory disorders related to cold and heat stress, malaria, dengue fever, schistosomiasis, diarrhea, energy consumption, water resources, and unmanaged ecosystems 82 . [...] For the pricing of the food categories, we determine the total amount of proceeds that farmers accumulate for their sold foodstuff in € 41 for each category ("producer-price") divided by its total production quantity. Thereby we calculate the relative price per ton for each foodstuff.
We solely refer to producer prices as the system boundaries only reach until the farmgate.

Calculating output data
[...] Combining the input data, we are now able to quantify and monetize externalities of GHGs for different food categories.

Calculating output data of quantification: category-specific GHG emissions
For quantification we separate between the following two steps: First, the aggregation of emissions data to broader categories and second the differentiation between conventional and organic farming systems. We iterate these steps two times, once for broad categories of : Visualization of the quantification process and corresponding input and output data. GEMIS data 37 (g b,n,i,conv ) and production data 36,41,70 (q b,n,i,conv ) are combined and emission data for broad (E b,conv ) and narrow (e b,n,conv ) categories is derived for conventional production. Organic emission values are calculated by multiplication of conventional emission values (E b,org and eb,n,org ) with the emission difference (D b, org/conv ) (cf. subsection 6.2.1). Source data are provided as a source data file.

[...]"
In addition to the adjustments described above, we would like to discuss the last part of the reviewer's commentary ("...

Our suggestion is to include the data concerning linear interpolation in the SI -if the reviewer find this necessary -and thus enable the reader to understand the calculation in detail.
Following the reviewer's advice, we added "Lane 2017, p. 462" as reference in the main text (subsection 6.2.1, p. 21.) 6.
"In turn, sometimes, it is, also, hard understand the importance of each equation. For example, improve the explanations about the equation (1) and the relevance of this equation for your research. The same for the others. Another question is about the source of these equations that need to be clarified." Response: Thank you for this comment. It is very important to us that the equations enhance rather than hinder the understanding of the operations in this paper. Therefore, we improved the textual explanations of formula 1, 2, 3, 4 and 5 and, when necessary, we also added literature references (for formula 1). We also decided to remove formula 3, 4 and 5 as a sole textual explanation seemed to be sufficient in these cases. In the following, changes of the original text are underlined (for formulas 6-13 refer to reviewer's comment 7): If the yield difference would not be included, emissions from organic farming would appear lower than they actually are as organic farming has lower emissions per kg of foodstuff but also lower yields per area. With formula 1, we adjust for that."

Formula 2a and 2b:
We adjusted the formulas and indices to differentiate more clearly between conventional and organic farming; we also added direct language to emphasize that these are our own calculations: Subsection 6.3.1, p. 30 f.: "The first step of aggregation consists firstly of aggregating food specific emission data from GEMIS 'g b,n,i,conv ' to the narrow categories 'e b,n,conv ' and secondly aggregating emission data from the narrow categories to the broad categories 'E b,conv '. As mentioned before and remarked in the respective indices, all this data only refers to conventional production up to this point. For both steps the method is identical. The aggregation to narrow categories is represented in (2a) where 'e b,n,conv ' stands for the emissions of the narrow category 'n' which itself is part of the broad category 'b'. Input data from GEMIS is remarked as 'g b,n,i,conv ' whereby the index 'i' refers to the i'th element of category 'n'. It's production quantity is 'q b,n,i,conv '. 'p b,n,conv ' represents the production quantity of the narrow category 'n'. 'I' (and 'N' in formula 2b) represent the highest index of an element in a narrow (or a broad) category. The aggregation to broad categories is described by formula 2b whereby 'E b,conv ' are the emissions and 'P b,conv ' the production quantity of broad category 'b'. , = , , ∈ × , , , " Formula 3: We erased this formula as the operation is easily understandable with sole textual explanation; we also added direct language to emphasize that these are our own calculations: Subsection 6.3.1, p. 31: "In the second step, we calculate emission values for organic production by multiplying the calculated emission-difference 'D b,org/conv ' between both farming systems (cf. subsection 6.2.1) with the conventional emission values. These organic emission values are denoted as 'E b,org ' for broad categories and 'e b,n,org ' for narrow categories."

Formula 4 & 5:
We deleted these formulas, as textual description seems to be sufficient for understanding the calculations; we added direct language to emphasize that these are our own calculations; also, indices are adjusted: Subsection 6.3.2, p. 31: "To calculate the monetary cost 'C b ' of category-specific emissions, we multiply the cost rate 'P' for CO 2 -equivalent emissions with the category-specific emission data 'E b ' or 'e b,n ' (depending on whether broad or narrow categories are observed). Further, we determine percentage surcharge costs '∆ ' by setting this cost in relation to the producer price 'pp b ' of the respective food-category: ∆ = (the calculations are analogues for narrow categories).
These surcharge costs represent the price increase necessary to internalize all externalities from GHG emissions for a specific food-category. " 7. "In this section 4 you need to link more the results obtained with the equations and methodologies presented before and to compare more specifically these results with other works." figure 1, 2 (formerly table 5, 6) and figure 4 (formerly figure 2). Furthermore, we edited the formula in section 3 (formerly section 4). All formula in section 3 (formerly known as formula 8, 9, 10, 11, 12 and 13) are now only described in the text and do not appear separately anymore. The reason behind this is that a sole textual description seemed to be sufficient for understanding in all cases, whereas the excessive listing of formulas might confuse the reader. By also improving the description of all these calculations, we further tried to link the results to the methods. In the order that they appear in the manuscript, these changes are chronologically listed in the following (changes are underlined):  Table 1: Emission data for food-specific, narrow and broad categories; Emission data for food-specific, narrow and broad categories (following the classification from the German Federal Office of statistics 36 ); food-specific emission data for conventional production was derived from Global Emissions Model for Integrated Systems (GEMIS) 37 and aggregated to narrow and broad categories with German production data 36 ; differentiation between conventional and organic production was derived with a meta-analytical approach (for details refer to the methods section and the Supplementary Information (SI), SI.1); LUC data is approximated to be the LUC emissions of soymeal fodder, emissions of it are calculated with the method of Ponsioen and Blonk 38 . Emission data including LUC emissions are shown in brackets. Source data are provided as a source data file.  Table 2: Determining the emission-difference (D org/conv ) between organic and conventional production in different countries' contexts through the application of meta-analytical methods; arrows represent the yield/productivity difference for each category, this difference is then multiplied with the emission-difference per ha to derive the emission difference per kg (in bold); PY = publishing year, CY = yearly citations, SJR = SciMago journal ranking, SUM = sum of all three factors, WEIGHT = weighted sums of category. A more detailed explanation of the studies' specifics including the weighting scheme can be found in the SI section (SI.1). Source data are provided as a source data file.

Response: Thank you for this comment. To ensure that the link between the methodology and the results is clear, we added the symbols and indices (which we introduced in the methods) in all the tables 1, 2, 3 and 4 as well as
We also created another table with additional information, which we put in the Supplementary Information section (please see SI.1).  Table 3: Producer prices (pp), external costs (C) and percentage price increases (∆) for narrow and broad food categories when externalities resulting from GHG emissions are monetized; producer prices are calculated by dividing the total amount of producer proceeds for each category (in Euro) 40 with its total production quantity 36,41 ; external costs are derived by multiplying emission values from table 1 with the emission cost rate of 180€/tCO 2 eq; percentage price increases are the ratio of external costs to producer prices; in brackets are the values with LUC-emission costs included. Source data are provided as a source data file.

Formula 6, 7, 8 and 9:
We deleted these formulas and explained the calculations thereof in the text. To calculate the costs 'C b ' of category-specific emissions, we multiply the cost rate 'P' for CO 2 -equivalents with the category-specific emission data 'E b ' or 'e b,n ' (depending on whether broad or narrow categories are observed). Further, we determine percentage surcharge costs '∆ ' by setting these costs in relation to the producer price 'pp b ' of the respective foodcategory: ∆ = (the calculation is analogues for narrow categories). These surcharge costs represents the price increase necessary to internalize all externalities from GHG emissions for a specific food-category.     production data 37,42,65 (q b,n,i,conv ) are combined and emission data for broad (E b,conv  Section 4., p. 14 ff.
"As the results show, the production of animal-based products -especially of meat -causes the highest emissions. These results are in line with the prevailing scientific literature [18][19][20][21]42 .

[...]
Secondary animal-based products, such as milk and eggs, however, cause lower emissions than meat. Again, these findings are in line with other sources 21,44 . [...] The feed of organic dairy cows incorporates a significantly higher proportion of grazing (29.5% compared to 0.5%), which also avoids GHG emissions associated with the production of industrial feed compared to conventional dairy cows 47 . Moreover, the use of grassland instead of farmland leads to the preservation of CO 2 sinks 48 . However, the difference between farming practices is lower in both primary, and secondary animal-based products compared to the difference in plant farming. This may be explained with the higher land use 49-51 , living age and lower productivity of organically raised animals 47 (cf . table 2) counterbalancing or even reversing the described positive aspects of organic animal farming. [...] Due to price elasticities of demand for food products (which are consistently regarded as 'normal goods' in economic literature), appropriate pricing of food would make products of organic production more competitive compared to their conventional counterparts 53 : customers would increasingly opt for organic foodstuff due to the lowered price-gap between the two options. This could potentially press the boundaries of land use for agriculture as organic practices mostly require more land than conventional systems due to lower yields 54-56 . However, our results suggest an increase in the prices of animal-based products to a significantly larger extent than the prices of plant-based products. The presumed consequential decline of animal-based product consumption would free an enormous landmass currently used for feed-production. Further expansion of area-intensive organic agriculture would subsequently be made possible 57 . Furthermore, there is evidence that a shift from conventional to organic practices would indeed be beneficial for the ecosystem services and long-term efficiency provided by the particular land area 7,58 . If one takes into account the temporal change in yield difference which would result by converting farms from conventional to organic farming, there is scientific consensus that the yield gap will decrease over time 59,60 . Comparative studies between different cultivation methods also show that organic farming has lower soil-borne GHG emissions and higher rates of carbon sequestration in the soil 46,61 . Soil degradation resulting from conventional systems would slow down or could even be reversed by changing to organic farming 62,63 . [...] Editor 1.
To improve the quality of methods and statistics reporting in our papers, we are now asking all authors to complete an editorial policy checklist that verifies compliance with all required editorial policies. Please ensure that the checklist is completed and uploaded with your revised article. All points on the policy checklist must be addressed; if needed, please revise your manuscript in response to these points. Please note that this form is a dynamic 'smart pdf' and must therefore be downloaded and completed in Adobe Reader, instead of opening it in a web browser. Editorial policy checklist: https://www.nature.com/documents/nr-editorial-policy-checklist.pdf Reporting summary: https://www.nature.com/documents/nr-reporting-summary.pdf Reporting and materials availability requirements for Earth sciences research: http://www.nature.com/authors/policies/availability.html#requirements Response: We have filled out both the policy checklist and the reporting summary and included them in the attachments of this email.

2.
In an effort to ensure reproducibility of research data, we now also require that you provide a separate source data file. The source data file should, as a minimum, contain the raw data underlying all reported averages in graphs and charts, and uncropped versions of any gels or blots presented in the figures. To learn more about our motivation behind this policy, please see https://www.nature.com/articles/s41467-018-06012-8.
Within the source data file, each figure or table (in the main manuscript and in the Supplementary Information) containing relevant data should be represented by a single sheet in an Excel document, or a single .txt file or other file type in a zipped folder. Blot and gel images should be pasted in and labelled with the relevant panel and identifying information such as the antibody used. We also encourage you to include any other types of raw data that may be appropriate. An example source data file is available demonstrating the correct format: https://www.nature.com/documents/ncomms-example-source-data.xlsx The file should be labelled 'Source Data', with the title and a brief description included in your cover letter, and should be mentioned in all relevant figure legends using the template text below: "Source data are provided as a Source Data file." Response: The source data can be found under the following link, which is also referenced in the data availability statement: https://osf.io/e7v8x/?view_only=0bff6aa858a340df9046816c1404a51c

3.
We would like to clarify if and how the software/algorithms necessary to reproduce the results will be made available to the scientific community upon publication as required by our material sharing requirements. For more information on this please see http://www.nature.com/authors/policies/availability.html#code In order for the reviewers to evaluate the work adequately they must be able to test the software/review the code themselves. If you have not yet provided the software, we therefore request that you provide a single compressed zip file containing the software with a readme.txt file or other user manual containing complete instructions for installing and running the software. If appropriate, please also provide example data and expected output. Sufficient material should be provided for referees to directly test the performance of the software/algorithm. If the software and materials are small enough to fit in a single compressed zip file less than 6MB in size, you may email this file directly to me. If the zip file is between 6 MB and 200 MB you may upload it to our file transfer site. If necessary, a second zip file up to 200 MB in size can be used to supply the example data. Please let me know if you need to use this option and I'll send you further details. Please also fill out and return to me the code and software submission checklist that will be made available to editors and reviewers during manuscript assessment. Please note that this form is a dynamic 'smart pdf' and must therefore be downloaded and completed in Adobe Reader, instead of opening it in a web browser. https://www.nature.com/documents/nr-software-policy.pdf Response: We did not use any custom code for our study. According to the editorial policy checklist, the 'code availability statement' does not apply to our study. Emission values were derived from the publicly available material flow analysis tool GEMIS (Version 4.95), which can be downloaded here: http://iinas.org/gemis-download-121.html. User manuals, system requirements, demos, installation guides and instructions are all provided on this website.

4.
Data availability statements and data citations policy: All Nature Communications manuscripts must include a section titled "Data Availability" as a separate section after the Methods section but before the References. For more information on this policy, and a list of examples, please see https://www.nature.com/documents/nr-data-availability-statements-datacitations.pdf In particular, the Data availability statement should include: -Accession codes for deposited data -Other unique identifiers (such as DOIs and hyperlinks for any other datasets) -At a minimum, a statement confirming that all relevant data are available from the authors -If applicable, a statement regarding data available with restrictions -If a dataset has a Digital Object Identifier (DOI) as its unique identifier, we strongly encourage including this in the Reference list and citing the dataset in the Data Availability Statement.
-If a source data file is provided, please add a reference to this in the data availability statement.

5.
DATA SOURCES: We strongly encourage authors to deposit all new data and code associated with the paper in a persistent repository where they can be freely and enduringly accessed. Please note that for some data types, deposition in a public repository is mandatory (http://www.nature.com/sdata/policies). We recommend submitting the data to discipline-specific, community-recognized repositories, where possible and a list of recommended repositories is provided here: http://www.nature.com/sdata/policies/repositories If a community resource is unavailable, data can be submitted to generalist repositories such as figshare (https://figshare.com/) or Dryad Digital Repository (http://datadryad.org/). Please provide a unique identifier for the data (for example a DOI or a permanent URL) in the data availability statement, if possible. If the repository does not provide identifiers, we encourage authors to supply the search terms that will return the data. For data that have been obtained from publicly available sources, please provide a URL and the specific data product name in the data availability statement. Data with a DOI should be included in the reference list and cited where relevant. Alternatively, include the data and code in the Supplementary Information. For datasets for which mandatory deposition is not required and the data or code can only be shared on request, please explain why in your Data Availability Statement and in your cover letter. Please refer to our data policies here: http://www.nature.com/authors/policies/availability.html If you opted into the journal hosting details of a preprint version of your manuscript via a link on our dedicated website (https://nature-research-underconsideration.nature.com), it will remain on this site while you are revising your manuscript, as we consider the file to remain active. Should you wish to remove these details, please email naturecommunications@nature.com indicating your manuscript number and the link on our website that was previously sent to you. Please see our pre-publicity policy at http://www.nature.com/authors/policies/confidentiality.html For more information, please refer to our FAQ page at https://nature-research-underconsideration.nature.com/posts/19641-frequently-asked-questions Response: All relevant data of the study can be accessed here: https://osf.io/e7v8x/?view_only=0bff6aa858a340df9046816c1404a51c

6.
Springer Nature encourages all authors and reviewers to adopt an Open Researcher and Contributor Identifier (ORCID). ORCID is a community-based initiative that provides an open, non-proprietary and transparent registry of unique identifiers to help disambiguate research contributions. All authors who link their ORCID to their account in our submission system will have their ORCID published on their articles, if the article is accepted for publication. Please note that this is only possible if ORCIDs are linked prior to acceptance, that is, it is not possible to add ORCIDs at proof. Please ensure that all co-authors are aware that they can add their ORCIDs to their accounts and that they must do so prior to acceptance.
To add an ORCID please follow these instructions: 1. From the home page of the MTS click on 'Modify my Springer Nature account' under 'General tasks'. 2. In the 'Personal profile' tab, click on 'ORCID Create/link an Open Researcher Contributor ID (ORCID)'. This will re-direct you to the ORCID website. 3a. If you already have an ORCID account, enter your ORCID email and password and click on 'Authorize' to link your ORCID with your account on the MTS. 3b. If you don't yet have an ORCID account, you can easily create one by providing the required information and then clicking on 'Authorize'. This will link your newly created ORCID with your account on the MTS. If you experience problems in linking your ORCID, please contact Platform Support Dear Pieper et al, I found significant improvement in the clarity of the information presented in the paper and the approach used. My general and specific comments are given below:

Response: We created ORCID accounts for all authors of this article and linked them to our accounts in the
General comments: I find the language to be more clear yet there is scope to make it more clear and even shorter. I have indicated some specific sentences in the MS (attached) Specific comments: It is mentioned that for organic farming, EU regulations only allow the animal feed to be grown organically and in the same region. The regional context need to be clear, i.e. is it the same country or could be the entire EU, or there are other specific criteria? Also, it is mentioned later that farm associations such as Bioland put restrictions on feed import from Latin America. It is conflicting because if there is a regional restriction on feed imports, what is the relevance of banning imports from Latin America by individual organisations? Furthermore, if the feed is allowed to be imported and the restrictions are only for Latin America, does it means that imports from other parts of the world such as Africa and Asia are allowed or take place? If so, it is important to ascertain that such imports do not have LUC related emissions in those regions before taking LUC out of calculations.
In table 2, What is 'arable'? Looking in the Tuomisto (2012), I guess authors mean there is no specific crop differentiation-is it correct? This needs to be made clear. I also suggest using the citation as reference numbers as it is will be more convenient to find relevant references.
Authors give strong rationale that organic farming cause negligible LUC, but at one pojnt they say, 'Moreover, the use of grassland instead of farmland leads to the preservation of CO2 sinks 48. However, the difference between farming practices is lower in both primary, and secondary animalbased products compared to the difference in plant farming. This may be explained with the higher land use 49-51, living age and lower productivity of organically raised animals 47 (cf. table 2) counterbalancing or even reversing the described positive aspects of organic animal farming.' This is conflictng as it indicates that organic animal production has higher LUC? Please make it more clear.
In the following statement, 'In case GEMIS offered no data for Germany for certain foodstuff (this is the case for maize, milk and eggs), data from climatically comparable European countries is used.', you should provide which countries were actually used for different categories. I see that you have listed studies in Tables 2 and in the supplementary material. Are these the countries ultimately used? In such a case, I would find it difficult to compare the climatic conditions of France and Spain with Germany. Hence it is important that you mention the countries and for what food categories those countries were used giving clear rationale.
I would like to see more about working principals behind GEMIS. Since GEMIS covers the emissions related to resource extraction to its processing and transportation, these mechanisms also need to be similar to Germany and not just the climate. Do authors assume that those processes are similar across the EU?
On page 23, I am not able to relate Table 3 with the text (see the marked section in the attched MS).

Is it table 4?
On the page 23, 'These shares are expressed in formula 2a and 2b (sub-section 6.3.1) by the terms p_(b,n,conv)/P_(b,conv) (share in broad categories) and 〖q_(b,n,i,conv)〗_ /p_(b,n,conv) (share in narrow categories)', I am unable to understand whether it is only for conventional or authors assume that share of each category remains the same for both conventional and organic?
About the selection of the literature for meta-analysis, 1968 to 2018 is a very broad time range given that authors address only one year (2016) in this exercise. Certainly, the older papers such as those published more than 20 years ago would have had significantly different farming conditions and practices as well as the technological processes. Hence, there is a point in assigning greater weightage to more recent studies. Still, I wonder why authors did not select a more narrow time range for selection of literature? It is also important to pay attention to the years during which the data was collected in the respective article Rather than the publication year.
I am not trained enough in economics, hence not well equipped for commenting on cost analysis. I hope that the other reviewer can review it.

Best wishes
Reviewer #2 (Remarks to the Author): This revised version is much better. The Authors made a great effort.

Reviewer #1 (Remarks to the Author):
Dear Pieper et al, I found significant improvement in the clarity of the information presented in the paper and the approach used.
Response: We are very grateful for this feedback and are glad that our extensive efforts during the first revision phase came to fruition. Your detailed and specific feedback, as well as the feedback of Reviewer #2, were very valuable to elevate the quality of our paper. We thank you for your specific inquiry.
My general and specific comments are given below: General comments: I find the language to be more clear yet there is scope to make it more clear and even shorter. I have indicated some specific sentences in the MS (attached) Response: Thank you for acknowledging the progress we have made with the first revision process, also due to your comments and suggestions.
We do agree that the language we use can be even shorter and clearer. According to the information we have received regarding the editing process there will be an inspection of linguistic correctness by the journal later on in the submission process. We have, however, improved (i.e. clarified and shortened)  "In this paper we show a possible application of this economic instrument by calculating the surcharges for foodstuff needed for a proper internalization of external costs from GHG emissions." Section 2., p. 4: "Congruent to methodological differences for monetizing agricultural greenhouse gases, there are also differences in the estimation level of greenhouse gas costs." [shortened secntence] Section 2., p.5: "N 2 O is produced in agriculture mainly due to direct emissions from agricultural soils, mostly caused by the overapplication of nitrogen fertilizer, and indirect emissions from the production of such fertilizer 34 ." [shortened sentence] Subsection 3.1, p. 9: "As follows from table 2, with LUC emissions included, organic produced food causes fewer emissions in the broad plant-based and dairy categories, while causing slightly higher emissions in the animal category. In the narrow categories organic production performs worse for eggs, poultry and ruminants." Section 4., p. 14: "This may be explained with the higher use of land due to organic regulations prescribing a certain amount of land per animal, which is higher compared to average conventional production 46-48 , as well as a higher living age and lower productivity of organically produced feed and raised animals 60 (cf. "Thus, we have to focus solely on imported feed for conventional animal-based and dairy products." [shortened sentence] Specific comments: 1) It is mentioned that for organic farming, EU regulations only allow the animal feed to be grown organically and in the same region. The regional context need to be clear, i.e. is it the same country or could be the entire EU, or there are other specific criteria?
Response: Thank you for pointing this out. Indeed, the word 'regional' can be understood in different ways. We are, however, referring to the definition of the inspection authorities of the German federal states, who have agreed to regard regional farms as those from the same or a directly neighboring federal state or political entity ( Only in conventional production it is unreservedly allowed to import crops (as fodder) from locations outside of the regional context. This is in contrast to organic production where the majority of the fodder must come from farms from the same or directly neighboring federal states 38 .

Subsection 6.2.1, p. 22
Organic feed is not considered as article 14d of the EU-Eco regulation stipulates that organic farms have to primarily use feed which they produce themselves or which was produced from other organic farms in the same region 85 . 'Region' is understood as the same or the directly neighboring federal state.
2) Also, it is mentioned later that farm associations such as Bioland put restrictions on feed import from Latin America. It is conflicting because if there is a regional restriction on feed imports, what is the relevance of banning imports from Latin America by individual organisations? Organic feed is not considered as article 14d of the EU-Eco regulation stipulates that organic farms have to primarily use feed which they produce themselves or which was produced from other organic farms in the same region 76 . 'Region' is understood as the same or the directly neighboring federal state. Although the EU-Eco regulation doesn't completely rule out fodder imports from foreign countries, it limits its application significantly. Also, one has to consider that over 60% of organic agricultural area belongs to organic farming associations 86 . These associations stipulate even stricter rules than the standard EU-eco regulation. Examples are Bioland, where imports from other EU and third countries are only allowed as a time-limited exception 87 , Naturland, where additionally imports of soy are banned completely 88 , or Neuland, that ban any fodder imports from overseas 89 . We thus assume that the emissions that could possibly be caused by organic farming in Germany through the import of feed constitute a negligibly small fraction of the total emissions of a product. Thus, we follow standard assumptions from the literature 90-92 and calculate no LUC emissions for organic products.
3) Furthermore, if the feed is allowed to be imported and the restrictions are only for Latin America, does it means that imports from other parts of the world such as Africa and Asia are allowed or take place? If so, it is important to ascertain that such imports do not have LUC related emissions in those regions before taking LUC out of calculations. b) I also suggest using the citation as reference numbers as it is will be more convenient to find relevant references.

Response
Response: Thank you for pointing this out. We now added the citation reference numbers to table 2 and SI.1 5) Authors give strong rationale that organic farming cause negligible LUC, but at one pojnt they say, 'Moreover, the use of grassland instead of farmland leads to the preservation of CO2 sinks 48. However, the difference between farming practices is lower in both primary, and secondary animal-based products compared to the difference in plant farming. This may be explained with the higher land use 49-51, living age and lower productivity of organically raised animals 47 (cf. table 2) counterbalancing or even reversing the described positive aspects of organic animal farming.' This is conflictng as it indicates that organic animal production has higher LUC? Please make it more clear. "This may be explained with the higher use of land due to organic regulations prescribing a certain amount of land per animal, which is higher compared to average conventional production 46-48 , as well as a higher living age and lower productivity of organically produced feed and raised animals 60 (cf. table 2). This counterbalances or even reverses the described positive aspects of organic animal farming." "There is not a sufficient number of papers published describing the emission differences of organic and conventional production within the German agricultural landscape only. To broaden the database for following calculations we choose to include studies from a European background. Legislative circumstances describing organic farming are the same in all of Europe. Therefore, values declared as being of organic origin can be compared with each other as the rules and activities of this production practice are clearly defined within European boundaries. As agricultural production is influenced by climatic conditions, it is to be noted, however, that factors like precipitation or solar radiation do vary between the countries in which selected studies were conducted. However, the climatic conditions within Germany's borders vary strongly as well, especially from north to south. Some regions in Germany can therefore be more closely compared to regions of Ireland, for example, some more to regions of Spain. Since our assessment is designed to describe agricultural production of Germany in general, it is sensible that the climatic differences of the underlying studies also describe the diverse climatic conditions of Germany. Furthermore, we use the relation between the studies' reported greenhouse gas emissions of organic and conventional agriculture rather than the reported absolute values for further calculations. Climatic conditions within one study, which could result in different emission values compared to an evaluation on German ground, will likely have similar or equal impact on both the agricultural practices assessed within this one study. We therefore argue, that the ratio does not change to a considerable extent with climatic conditions. This notion is also supported by the reported values not showing particular tendencies to one or the other direction due to their origin." 7) I would like to see more about working principals behind GEMIS. Since GEMIS covers the emissions related to resource extraction to its processing and transportation, these mechanisms also need to be similar to Germany and not just the climate. Do authors assume that those processes are similar across the EU?
Response: Thank you for this inquiry. As we have pointed out with regards to your previous comment (comment No. 6) we now only use emission data from GEMIS that refers solely to Germany and no other European countries. Thus, the mechanisms behind the emission values from GEMIS are now by definition applicable to Germany, as they in fact explicitly refer to Germany. With regards to your question: "Do authors assume that those processes are similar across the EU?" we therefore want to point out that GEMIS data only refers to Germany. Thereby, as far as we can see, we don't have to make the assumption that production processes are similar across the EU.
Only when it comes to determining the emission-difference between organic and conventional production, do we use EU-data. In our answer to comment 6, we explained why this is reasonable with regards to differences in the climatic condition. With regards to production mechanisms, we also think it is legitimate to use data from across the EU, as all EU-member states abide to the same regulatory framework concerning organic production. Agriculture is one of the few industrial sectors that is almost completely regulated on the EUrather than on the national level. Thus, there is common regulation like the "Common Agricultural Policy" (CAP) and the EU-Eco Regulation. We have pointed this out in the supplementary information (cf. answer for comment 6) Still, we take your advice to include more information on GEMIS seriously and propose to add this section to the S.I.: Supplementary Information, p. 49f.
"A closer look into the mechanisms of GEMIS shall be provided with the example of beef. In GEMIS one can sort through a wide variety of processes. For this explanation, we want to look at the process of beef-production at the stage of slaughtering. This process is labeled with the code: NG-SchlachtereiDE-Rind-2010, which already gives some indications on the properties of the process. 'NG' stands for "Nahrungs-und Genussmittel" (english: food and beverage). This is followed by an indication on the stage of production and reference country, whereby 'SchlachtereiDE' stands for butchery in Germany. Then the actual product (Rind=beef) and the reference year (2010) are listed. Under this code, all inputs, outputs and the corresponding emissions are listed for one functional unit of the produce (defined as 1 kg of beef-meat). Furthermore, one can read out the compounds of this data-set. In this case, it is the electricity and process-heat for the slaughtering stage, as well as the process of animal-husbandry (including all inputs necessary to raise the animal). Reading out these data-subsets reveals their respective in-and outputs, which themselves consist of even more inputs. By repeating this recursive process, one can read out ever more fine-grained compounds, the further to the beginning of the supply chain one progresses. The following datasets from GEMIS are used for our study (more information on this can be found in the source-data file): Original GEMIS data (kgGas/kgProd) Name name (in GEMIS) 9) On the page 23, 'These shares are expressed in formula 2a and 2b (sub-section 6.3.1) by the terms p_(b,n,conv)/P_(b,conv) (share in broad categories) and 〖 q_(b,n,i,conv)〗_ /p_(b,n,conv) (share in narrow categories)', I am unable to understand whether it is only for conventional or authors assume that share of each category remains the same for both conventional and organic?

CO2eq
Response: Thank you for this comment. You have a valuable point here, as we probably didn't explain this sufficiently in the text. Yes, we assume that the share of each category remains the same for both conventional and organic. This is because doing otherwise would not enable a fair comparison between production systems on the aggregation-level of broad categories. For example, beef makes up over 50% of all produced food in the organic animal-based product category, while it only accounts for 25% of the conventional animal-based product category. As beef production produces the highest emissions of all foodstuff, these high emissions would be weighted far stronger in the organic category than in the conventional category. This would create ratios between emission values of organic and conventional broad categories that would not be representative of the ratios between organic and conventional narrow categories. We have already tried to elaborate on this in subsection 6.3.1: Subsection 6.3.1, p. 29f.: "Concerning the reasoning behind the method, the question that might come to mind is why the differentiation between farming systems happens after the aggregation and not before. This is due to the fact that the proportional production quantities of specific food as well as food categories to each other differ from conventional to organic production. Let us imagine aggregation would take place after the differentiation of farming systems: For example, beef actually makes up over 50% of all produced food in the organic animal-based product category, while it only accounts for 25% of the conventional animal-based product category (cf. production values in table 3). As beef production produces the highest emissions of all foodstuffs, these high emissions would be weighted far stronger in the organic category than in the conventional category and thereby producing a higher mean for the organic animalbased product category than for the conventional one. As can be seen from this example, the organic animal-based product category could have a higher mean of emissions than the conventional animal-based product category while still having lower emissions for each individual organic animal-based product than conventional production. Deriving GHG emissions of foodstuff before aggregating to broader categories would thus be problematic and create means not representative for the elements that make up the broader category. To prevent this problem, the chosen method in this paper is thus to first aggregate to the chosen level of granularity (broad or narrow food categories) and then to derive emissions of organic production from conventional production data." However, as this explanation is embedded in a slightly different context, we also added further explanation on our calculation logic at the passage in the text that was pointed out by you: Subsection 6.3.1, p. 23: "Only production data for conventional production is used. Thereby we imply ratios of production quantities across the food categories for organic production that are equal to those of conventional production. This does not fully reflect the current situation of organic production properties but allows for a fair comparison between emission data of organic and conventional food categories. Doing otherwise would create ratios between emission values The time lag between data collection and publication year of all studies reporting years of data collection lies between 3 and 7.5 years. There is no tendency to be noted whether older studies have a longer time lag or vice versa.
We have now revised our calculations as follows: • The weighting of the publications years is now adapted to a weighting of the years of data collection. • If a study does not declare this information it is weighted with the lowest weight in this weighting category (=0). The calculation show, that none of the results change significantly or even at all. Biggest changes are of less than 1% (organic dairy surcharge) or 0.02€ (organic ruminant external costs) or 0.11kg CO 2 eq/kg Product (organic beef emission). These changes, in our opinion, do not alter the conclusions of our study or significantly improve the quality of our results. We therefore suggest to keep the information about the years of data collection in table SI.1, enhanced according to your suggestion, but do not change the weighting process with these new years as there is no information about this available for some studies (see above).
I am not trained enough in economics, hence not well equipped for commenting on cost analysis. I hope that the other reviewer can review it.

Reviewer #2 (Remarks to the Author):
This revised version is much better. The Authors made a great effort.
Response: Thanks a lot for acknowledging the effort and significant improvements we have achieved during the first revision phase, also due to your comments and suggestions.