Overall this is an important contribution, updating one of the major datasets on global food system emissions. However the manuscript requires further work to make transparent key assumptions and issues with the data (scope and uncertainty), while the dataset itself is not sufficiently user friendly and appropriately documented in its current form. Nevertheless, I strongly support this effort and congratulate the authors on their work.

Page 1

Line 28: is it FAO or FAOSTAT (line 20)?

Line 30: “in terms of single GHG” change to “in terms of individual greenhouse gases (GHGs)”

Line 34: the time period (1990-2019) is mentioned twice, at the beginning and end of the sentence

Page 3

Line 2: typo in the first sentence, should read “as well as one of the economic sectors most at risk from it”

Line 8: EDGAR-FOOD would be another important reference to include in this sentence (https://www.nature.com/articles/s43016-021-00225-9)

Page 4

Line 31: typo in 2022-2023

Line 35- page 5 line 4: These sentences belong in the subsequent section on uncertainty.

Page 5

Line 5-11: Can the authors restructure to make a clear distinction between emissions sources that are (a) not included because they are indirect and out of scope (“upstream GHG emissions, refining, etc.”) and (b) not included because data was not available, even though they are direct and within scope?

It would be important to note in (a) whether or not indirect emissions from electricity use are also excluded, as this is generally the largest indirect source across all sectors; and in (b) how significant these sources are in estimated CO2 equivalents, and whether this is a complete list of omitted direct emissions sources.

Line 12-21: This is a relatively short discussion of uncertainty – given its importance in the context of food system emissions. As stated above, several sentences from the prior section could be brought down. Several further points could be made:

Does the estimated uncertainty range (“30—70% across many processes (Tubiello, 2019)”) also hold true for this dataset? Please be explicit.

Could uncertainty estimates be provided for sub-components of the data (e.g. by gas, or food system component)? This is critical information for the data users.

To what extent does uncertainty prevent us from making policy relevant statements on (1) total emissions levels, (2) total emissions trends, (3) the relative importance and impact of different food system components?

Does uncertainty increase with decreasing scale (global to regional to country level data)?

Page 6

Line 7: Perhaps state the denominator here too (total global GHG emissions) and its source? It is also not in Table 1. (I see that it appears in the discussion. Please move up here.) You might consider placing it in the abstract too, since the sentence appears there too.

Line 7: What would be the emissions range for the global total (± xGtCO2eq yr-1), given the previously stated uncertainty?

Page 8

Line 2-4: This is an important claim, also in the abstract. Can it be sourced? What is the measure of “national mitigation strategies”? Sector based targets within NDCs?

Line 17-22: Presumably it is also due to shifts in other sectors, e.g. all else equal, reductions in power sector emissions will increase the proportion of food system emissions in the total. And power sector emissions have been declining in most EU countries and the US (e.g. https://www.tandfonline.com/doi/full/10.1080/14693062.2021.1990831)

Line 37: The result on F-gases is surprising - and interesting. Can the authors provide a little more detail? Which are the main gases? Perhaps a link could be made to Minx et al. 2021, which corroborates F-gas growth in inventories with atmospheric inversions (Fig 2 https://essd.copernicus.org/articles/13/5213/2021/essd-13-5213-2021.html) Also, in Table 5, F-gases were 0 in 1990. Is this a data artefact? Or is it due to Montreal gases being replaced by HFCs/PFCs in the intervening decades?

Page 9

Line 1-6: The language here suggests these subcomponents are trivial sources (“only”, “mere”). Arguably 15% or even 3-4% is not trivial, so I would simply present the numbers without inferring their importance. If one wants to make a normative point, I would argue that all emissions sources should be considered important and worth policy attention.

Line 12-32: There are multiple typos and phrasing errors here that could be improved. Please carefully check. Please also consider splitting this long paragraph into smaller chunks each with a substantive point.

Other comments on the manuscript:

Table 1: Could headings be added to group these sources into their higher level categories, e.g. as in Figures 1 and 2?

Table 3: You could add the fraction of global food system emissions that the top 10 add up to, in the caption.

What global warming potentials are applied to estimate GHG emissions in CO2eq?

Comments on the dataset:

My first impression is that the dataset is too large (200mb), unstructured, and lacking important metadata. Together these make it only available for advanced users. Some specific comments:

If one opens the .csv in Excel, a warning comes up that the data is not fully loaded (too many rows). Could it be split into several files? Or could a basic user-friendly excel version be provided alongside the raw csv file, perhaps for a useful series of aggregates (e.g. global emissions by food system component, by gas, by region/country), or the full data just for high emitters/regions? Such simplified sheets would presumably be important to assist national agricultural ministries to better understand emissions along the supply chain (a claim in the manuscript).

There is no explanation of the column headings embedded in the file (What are the flags? What are the codes? Are two codes for years really needed?). For example, a basic user wouldn’t know that Area contains both countries and regions, and Element contains two separate variables for five different gases (I would personally split this in two and have a gas column).

There are no country ISO codes, which raises barriers to joining other datasets (e.g. population, gdp).

Most tricky: what is the hierarchy and structure of the “Item” column? If I filter by “World”, “2019”, and “Emissions (CO2eq) (AR5)”, the sum of Value is 228 GtCO2eq. So there is double counting among the Items. Which items do I exclude to produce the number in the manuscript – 16.5 GtCO2eq? I see already that “Energy” is included (37GtCO2eq) and shouldn’t be. How do I know which items are in and which are out of the food system account? Could you add a column for this, so we don’t have to use complicated string operations?

Can we have the GHGs in native units, so that different global warming potential metrics can be applied? (Or conversely, a column with the applied AR5 GWPs)?

The dataset is of interest but the methodology and underlying data is not described in the article.  It is described in FAO Statistics Working Paper Series working papers, but it is not acceptable to have the methodology central in the data setting not described in the article (or in other peer reviewed articles).  In particular, those methodologies are supposed to be peer reviewed, and also available (possibly as supplementary material) with a reviewed article.  The methodologies from those working papers can be shortened, but upon reading them it seems that simply copying over most of the information, maybe with a summary in the main paper and a development in a supplementary material, or all in the main paper depending on the style of the review would be good as they are well written and describe adequately the methodologies.  Another reason to bring those in the article is that there may be some additional peer review comments based on those methodologies.

It is somewhat unclear if additional data should be provided along with the main dataset.  For instance shares of the food system.  However this cannot really be discussed if the underlying methodology is not presented and discussed.

Most of the informations and the data corresponds to an already existing article, Tubiello et al., 2021a "Greenhouse gas emissions from food systems: building the evidence base". Therefore I am not sure about originality, but it may be normal as here the dataset is the focus.  It makes all the more important to describe the methodology in the data article as it would be some originality.

The dataset combines different and incompatible disaggregations and nomenclatures, which is an interesting and important point of the methodology.  There is an explanation of the relationships between the nomenclatures in figure 1, and in the https://zenodo.org/record/5615082 page.  It is badly explained in the article, only very briefly in 2.1, although describing the data should be important in the article.

For the general public, as the dataset combines different and incompatible disaggregations and nomenclatures it is not clear if it would be of interest.  Although it is important to have those informations to understand the methodology and how these data can be derived from the PRIMAP data based on the IPCC nomenclature, for a non specialist this makes a very unclear dataset.

A comparison with Crippa et al would also be welcome as it is a similar work with care to explain what is exactly the ame when crippa et al has been used as a source. It is already done adequately, as far as I can tell from my readings in the Working Paper Series working papers, but it should be in the peer reviewed article and may trigger additional comments here.

More remarks

p 4 l 33 and following, the discussion about uncertainty does not add much information, all the information is quite generic.  There is some validation done in the FAO Statistics Working Paper Series articles, theis should be presented/discussed here.

p 4 l 25 The Step 4 of imputation of missing emissions is not clear (missing how?).  It should be associated with additional data showing which data is imputed and which data is not.

p 6 l 35 3.2 Regional Trends

The numbers per regional blocks or countries are not very interesting as the populations may be very different.  Also some goods may be exported which makes these numbers also difficult to interpret.  Some emissions are directly linked with the consumption, so should be local, but it is not the case for processing, packaging and fertilizer production.

p 8 l 7 the database FAOSTAT-PRIMAP is not introduced before nor really presented.  It should be presented and even be available with this data, as if I understand well it is the data which corresponds to the methodology, the data presented is an aggregation.

A minor remark, since the data is about reorganizing disaggregated data in different categories, the comparison of nomenclatures can be of interest in term of methodology to understand the strength and limitations of each nomenclature and warn about uses.  However, this is not done at all in the article.