General comments: The manuscript titled “Soil carbon, nitrogen, and phosphorus storage in juniper-oak savanna: Role of vegetation and geology” is a well-written manuscript that explores how geological factors may interact with woody plant encroachment to influence soil C, N, and P biogeochemistry. The importance of this study is twofold: 1) climate change and certain types of land management are accelerating woody encroachment into grasslands and it’s important that we understand how this shift influences soil properties; and 2) it’s critical that we understand how geology and vegetation changes interact, as findings can help improve modelling efforts for soil C, N, and P dynamics in various ecosystems going forward. The authors present clear figures and include site photos, which made for a pleasurable read.

Specific comments:

Intro – Well written! May be helpful to include a brief explanation of how the interaction between soil depth and δ13C of SOC can inform us on the history of the landscape (i.e., Fig. 3a). It would help set up your hypotheses, results, and beginning of your discussion section nicely for those who are less familiar with this concept.

Line 109 – can you specify if this higher clay content is across the entire soil profile or across some specific depth?

Fig 1 – Great idea to include photos of the Edwards and Buda soils. I think it’s helps readers better understand the differences between them (i.e., depth).

Lines 127 – 129 - Would be useful to add % forage utilization in parentheses here for context on how 'light grazing' vs. ‘heavy to moderate grazing’ is being defined.

Lines 138 – 139 - Please extrapolate/clarify what you mean by ‘within the middle of each depth increment'. I believe you mean 5 cm in the 0-10 cm increment – if you specify this, it will align with the figures better.

Table 1 & Table 2 – please consider removing lines between rows in the tables. If this is required formatting, then ignore. Otherwise, I suggest removing and formatting according to journal requirements.

Lines 172 – 173 – The way this is written throws the reader off a little bit.  Please consider rewriting as: “The fraction, (f), was the proportion of SOC derived . . .” or something similar.

Lines 201 – 202 – If you weren't able to get BD measurements >20 cm for Edwards soils, how were you able to accurately make SOC predictions past 20 cm (Fig. 4a)? From what I recollect, von Haden requires BD for the input sheet and R script.

Lines 208 – 209 – can you clarify what you mean by . . . “the fact that SIC increased more strongly with soil depth beneath oak than beneath grassland or juniper vegetation”. Is this based on the slope of the lines in fig 2d? And is it pertaining to across all depths? Just glancing at the figure, it appears the biggest change in SIC between the first and second depth increment is for grass.

Table 2 – Very interesting and surprising that depth alone did not significantly affect SOC. Only the interaction between geology and depth. I suggest capitalizing Depth, Vegetation Geology in the table to make the abbreviations even more intuitive.

Fig. 3a – I like the inclusion of δ13C litter values in the same figure as δ13C soil values. However, I would add a statement indicating exactly what the dashed line on the figure indicates in the figure caption for further clarity.

Line 244 – perhaps change to … “while only oak had higher SOC and TN on Buda soils”. It reads a little easier that way.

Line 306-307 – is it plausible that higher clay content in the Edwards soil could have increased soil C relative to Buda as well? You make a point in the methods that the Buda soil has less clay content.

Discussion – towards the end of the discussion, it would be helpful to briefly address other ecological effects of woody encroachment that were not directly measured in this study (i.e., biodiversity, soil erosion, etc.). It would make sense to add this sentiment after your point about SIC loss with encroachment (line 347).

Conclusion – As is, the conclusion is quite long. Please distill and shorten where appropriate – focus on what was found and why it’s important.

General Comments

This study quantified soil properties by depth (SOC, δ13 C, nitrogen, and phosphorus) under contemporary ecological conditions (grassland, juniper, and oak) on different ecological sites (by depth and parent material; Edwards, Buda) to evaluate impact on the soil’s biogeochemistry. Results show that grass to woodland transition is relatively recent, and that vegetation transition dynamics and soil parent material uniquely condition soil nutrient stores. My main concern regards conclusions related to land use change (grass to shrub) absent of long-term data (quantification of ecological state change over time) or a more specific soil-chronosequence study (space for time substitution, better control of soil type and soil age). Here are a few suggestions for the authors to consider on revision.

The presence of a petrocalcic horizon in the Prade (Edwards) and Valera (Buda) soils infer these soils are pedogenically much older than the Eckrant (Edwards) and Tarrant (Buda) soils. Valera also has a different soil family texture class (fine) than the other three (clayey skeletal). Can the authors provide the soil taxonomy and geographic locations for the soil trenches in Table 1? This will significantly add the soil and landscape interpretation. Without this information, “shallow depth to bedrock” (L385) for the Prade and Valera soils could be confused with the “depth to petrocalcic horizon”. Petrocalcic horizons are considered ‘pedogenic’ (atmospheric additions with soil translocations and transformations) and limestone/marl is older, or ‘geogenic’. Without these data, authors could maintain some their assumptions of soil behavior (shallow vs deep), however, conclusions of geology’s role are more complicated.

Specific Comments:

L21, L398, “results have important implications for the management and conservation of these ecosystem”. Can the authors add to the discussion management suggestions and implications?

L85, is the first mention of δ13 C to test woody encroachment. Given that δ13 C is used to conclude grassland to woodland conversion (L381), consider adding to the introduction how carbon isotopes are used as proxy to identify relative abundance C3 and C4 vegetation.

L101, L104, include the taxonomy for each soil class. Eckrant is “Clayey-skeletal, smectitic, thermic Lithic Haplustolls”; Prade is “Clayey-skeletal, smectitic, thermic, shallow Petrocalcic Calciustolls”; Valera is “Fine, smectitic, thermic Petrocalcic Calciustolls”; Tarrant is “Clayey-skeletal, smectitic, thermic Lithic Calciustolls”.

L104, Valerna should be Valera.

L110, is the only mention of the Rio Diablo and Ector soil series. Are these series identified in the study? Do these soil series add any additional information to the study (Buda vs Edward)?

L107, Bkkm is a “petrocalcic horizon” (pedogenic), avoid calling it marl (geogenic).

L129, can authors provide more information regarding the historic land cover conversion dynamics at these locations? Such as the historic rate and magnitude of the grassland to woodland conversion in the study area? I suspect that the TAMU Sonora station has this data.

L136, can you add a simple rational for the depth intervals used in the study.

L146, Table 1, can you provide specific site characteristics (landscape position), taxonomic classes, and the geographic coordinates for the trenches? Site characteristics and Soil taxonomy will add to the interpretation, and coordinates will help confirm any soil-landscape relationships previously identified by soil survey.

L171, I am not familiar with what appears to be a simplified formula to determine the C4 fraction. A more precise formula uses an end member mixing model of two sources: %C₃ = [(δ¹³_Cs − δ ¹³C_C4 )/(δ¹³C_C3 − δ ¹³C_C4 )] · 100%, and then calculates from, %C4 = 100 − %C3, (Phillips & Greg, 2001, Oecologia).

L172, was C4 and C3 tissue collection completed at each trench (such as the litter in Fig 3a) for the mixing model?

L206, I advise caution interpreting how geology and vegetation impact the SIC properties. Untangling SIC complexity likely requires a detailed soil-chronosequence study with specific controls on soil type (topographic position, genetic horizon, carbonates, etc) and the timing of vegetation change.

L208, Fig 2d, I wonder if SIC increase is due to the soils being inherently different (Petrocalcic Calciustolls vs Lithic Haplustolls). The Grassland-Edwards site has a considerable amount more SIC than the other sites.

Figure 3b, any reason why Edwards-Grass is not part of figure 3b?

L342-344, Yes, but does this generate enough carbonic acid to alter (within the timeline of land type conversion) the petrocalcic horizon? I advise caution interpreting SIC results without knowing the presence/ absence of pedogenic carbon (calcic, petrocalcic) vs geogenic carbon (limestone/ marl).