This study claims to provide a high resolution (4km by 4km)  dataset of phosphorus inputs, timing and application methods for the USA using interpolation methods and existing datasets at the county and state level from 1850 to 2022. They make an admirable attempt at doing this and could provide an important dataset for those interested in nutrient balances for this country. However I believe the uncertainties could be better explained in the discussion esp. related to areas not fertilized, and temporal variability in uncertainties given this is such a long term dataset taking data from multiple sources. Below I note some areas of their manuscript which could be elaborated on further to help the reader understand the uncertainty and accuracy of their estimates.

I believe the 4*4 km resolution the authors state gives a false sense of accuracy considering data at the state and county level are used for their estimates. Isn’t the accuracy of their estimates determined by their lowest resolution data (i.e. at the state level)? Given the methods employed by the authors I believe it is quite a stretch to say it has 4*4km resolution. More discussion should be made on this point, and perhaps the authors should refer to their study actually offering ‘pseudo 4*4 km resolution’ with state level accuracy. This would better indicate to users the uncertainty in their estimates and that really the accuracy is at the state level.

Lines 93-96. Please include in discussion, likely uncertainties in P consumption temporally. I assume there will be more uncertainty in the historic estimates from the 1800’s than the latter surveys?

Line 142. Where did you get data for the fertilized cropland percentage. Please add this to the method.

Line 148. How did you account for areas of wheat  not fertilized?

Line 156-157. Again how did you account for areas of other crops that received no fertilizer at all?

Line 186. You claim in your abstract that 40% of cropland has remained unfertilized in the last decade, but on line 186 you suggest you do not have spatial information to locate fertilized area. So how do you know the same 40% area has been left unfertilized? Based on line 186 it seems like you are showing average application rates across all area and not accounting for non-fertilized areas. Lines 195 onward you mention USDA-ERS data for % cropland fertilized, but how did you get % area fertilized before the USDA-ERS data before 1996?

Line 319: Can you explain the brief decline in 1980’s. Was there policy changes or a shock to the supply/demand for P that aligns to your estimates? This would be a nice way to cross check your estimates to make sure they make sense in terms of trends

Section 4.3. Discussion should be made about how uncertainty in areas not fertilized could influence average fertilizer rates.

Understanding and assessing the spatiotemporal patterns in crop-specific phosphorus (P) fertilizer management is crucial for promoting crop yield and mitigating environmental problems. This manuscript combined the top-down and bottom-up method as well as data source to rebuild the crop-specific P fertilizer inputs from 1850 to 2022. I believe it could be highly interesting for readers and related researchers. Before publication, there are still some problems as follows:

• In the line 157-165, you mentioned that “the summed P consumption of 9 major crops exceeds the state total P amount in some states”, and “the negative rates of the Other Crops were replaced by the average”. Thus, the total amount of P consumption could be increased, and do you have updated the increased parts for the county level or the state level?
• In your study, you have estimated the crop-specific P application rate, and then you have converted the unit of P use from cropland area to land area. It could be very confused. If one grid with less cropland but higher P application rate, it could be presented very low P use.
• The result section is not enough to present your studies, especially for changes and reason on the magnitude and spatiotemporal of P fertilizer use. The increase of P fertilizer use can be due to crop change or the increase of application rate, and thus I believe it could be some points on these reasons.
• The section of “Patterns of P fertilizer application timings” and “Patterns of P fertilizer application methods” are too short, and I suggested to add some interesting results.
• Figure 1 is not easy to readable, suggest to improve it.
• It is highly confused in Figure 2 (b). The light-colored bars denote the application rate on fertilized area and dark-colored bars show the application rate on all cropland. Thus, the light-colored bar should be higher than the dark-colored bar. However, the light-colored bar and the dark-colored bar are cumulative, and thus I am confused for the light-colored bar? Is it from the bottom (0 g P m-2) or the top of the dark-colored bar.
• I suggested the spatial distribution of P fertilizer application rates change to the total P consumption per grid.
• Can you add the country level of P fertilizer consumption and 9 major crops from 1950 to 2022.
• I suggest the unit P application rate should be changed to “kg P/ha”.
• There are some small errors in your manuscript as follows: - Line 224, conversely should be “Conversely”；- 3.2 Patterns of P fertilizer application methods should be 3.3