General comments

The knowledge on factors controlling the fate of carbon and nutrients in subsurface porous aquatic systems (aquifer sediments) is very limited. One reason is without doubt the technical limitations in studying aquifer systems in situ in an appropriate spatial and temporal resolution. In fact, most of the experience has been obtained from lab experiments by means of flow-through sediment column or microcosm experiments. In consequence, modelling exercises are very useful for testing different hypotheses and possible scenarios. The submission by Khurana et al. in this respect is an interesting contribution. However, in its current form the manuscript is not ready for publication. Major issues are, as will explained in more detail below, an obvious lack of expert knowledge in microbiology, wrong assumptions as input parameters, and a discussion that avoids critical reflection of the overall outcome. Considerable reworking of the MS is needed before it can be accepted for publication in Biogeosciences.

Detailed comments:

Introduction

P1-L18: ‘undertook’

P1-L21: In biology we have a clear nomenclature. Conditions are either ‘oxic’ or ‘anoxic’, Organisms and processes are ‘aerobic’ or ‘anaerobic’. I suggest to use this nomenclature concisely throughout the MS.

P2-L37: Here and at many other spots

P2-L49: Papers of potential interest for the authors: Zhou et al. (2012) FEMS Microbiol. Ecol. 81: 230-242, Hofmann et al. (2020) Front. Microbiol. 11: 543567

P2-L50: Papers of potential interest for the authors: McGuire et al. (2000) Chem Geol 169: 471–485, McGuire et al. (2005) Ground Water 43: 518–530

In the Introduction section important issues such as the discrimination between ‘active’ and ‘inactive’ as well as ‘mobile’ and ‘immobile’ cells are not picked out as central points.

Methods

As highlighted in the first paragraph, the conceptual model is simplistic. While the focus is to test for spatial heterogeneity and flow velocity as steering factors with respect to carbon, nutrient and microbial dynamics, frame conditions for the model simulations are non-dynamic with steady-state flow and constant inflow concentrations of dissolved species. In this respect, I would expect an in-depth discussion of the model output. Are the set frame conditions sensitive factors? How may the results change with respect to transient input concentrations of carbon and nutrients.

The authors mention the ‘use of geochemical and geomicrobial observations from a common study site’ as basis of the conceptual model. However, I could not find any sources (papers cited) with respect to ‘values’. The concentrations of TOC, DOC, NH4, NO3, O2, prokaryotic cells (active and inactive) have been selected and based on which studies and sites.). In P6-L182 you say: ‘The concentrations of the reactive species mimicked conditions observed in the subject site’. However, in the discussion it is mentioned that there were two orders of magnitude difference in prokaryotic cell numbers. I ask the authors to carefully consider input values. Is it true that a prokaryotic concentration of mobile prokaryotic cells in groundwater of 10^9 have been found in the field? Seems very high to me.

The concept of the reaction network is simplistic, which is ok. But it should be as realistic as possible. Does all NH4 originate from DOM? Isn’t there direct input of NH4 from the surface? . As mentioned in Table A.4.2. there is a constant input of 60µM of NH4 (about 1mg/L) that cannot originate from the 10mg of DOC (800 µM).  I recall that the studied Hainich Critical Zone sites are partially located in areas with agriculture. To which of the Hainich sites does the conceptual model refer to?

I know that it is hard to collect reliable information from the literature with respect to microbial features in shallow aquifers. Having this in mind, one need to carefully select values for ‘rate constants’, ‘yield coefficients’, … The values summarized in Table A.4.1 originate from field studies and lab studies very different in nature, i.e. values derived from lab experiments with pure bacterial cultures. Are the chosen values sufficiently representative for the Critical Zone in Hainich and shallow aquifers in general? This at least needs to be critically discussed.

With respect to DOC, an contstant input concentration of 800µM has been chosen. DOC degradation in soil and in groundwater is determined not only by its concentration but more likely by its quality (degradability). Has this been considered.

There is dynamics in many aspects, including flow velocity, water retention time, activity and biomass of microbes, DOM concentration and transformation, N transformation, … Only a subset of parameters, i.e. spatial heterogeneity and flow velocity (related to residence time) has been tested. This needs to be clearly mentioned already in the Introduction section.

Captions of Tables are generally on top of the tables, not below. See all tables.

P11-L297: Is there evidence for a parallel reduction of nitrate and oxidation of ammonium?

P12-L321: Is there any evidence that the portions of active and inactive cells/species are realistic? In particular when these ratios are calculated for individual physiological guilds (nitrate reducers, ammonium oxidizers, …). See also table 4.

P15-L386: log10Da?

P15-L394: Dissolved oxygen (DO) is not a nutrient.

P16-L411: Provide a citation that supports this statement.

Discussion

P19-L459: The ‘available’ process knowledge, does it refer to the Hainich study site?

P19-L474: from carbon concentration and carbon content per cell one will not end up with gene copies per volume but cells per volume. Quantification of cell numbers in groundwater and aquifers by means of molecular tools quantifying 16S rDNA gene copies is only a rough estimation of cell abundance.

P19-L474: Does 100times less cells equal 100times less microbial activity and 100times less transformation of C and N? Please comment on that.

P20-L490 & L516: There is techniques and reports available on high-resolution sampling in aquifers. The ready should not get the impression one cannot get spatially more resolved in sampling. E.g. Ronen et al. 1987 J. Hydrol. 92, 173–178, Báez-Cazull et al. 2007 Appl. Geochem. 22, 2664–2683, Smith et al. 1991

Contam. Hydrol. 7, 285–300, Anneser et al. 2008 Appl Geochem 23:1715–1730.

P20-L501: Have the authors considered that there is different growth rates with different physiological groups within the microbes, i.e. aerobes may grow faster that nitrate reducers and sulfate reduces are extremely slow. Did I miss this information?

P20-Fig. 6: What do you mean with ‘oxic cells’. Please change.

P21-L528: Consider the review paper of Smith et al. 2018 FEMS Microb. Ecol. 94: fiy191

P22-L565: I fully agree with this statement. In many cases the contribution of the mobile fraction of microbes can be neglected in terms of ‘transformation processes’.

Findings from other studies (like the one already cited Grösbacher et al. 2018) are not discussed in comparison to the model outcome.

Summary and conclusion

P24-L630: mention at which spatial scale.

P24-L640: Can this be visualized?

General comments:

Khurana et al. used a numerical modelling approach to assess the impact of spatial heterogeneity in the subsurface on microbial activity and nutrient dynamics. The authors combined a biogeochemical process network to describe the turnover of carbon and nitrogen with reactive transport modeling and simulated scenarios that reflect different types of spatial heterogeneities in the subsurface. They propose that impacts of spatial heterogeneity on nutrient cycling can be estimated and scaled based on the residence time of solutes and the Damköhler number of reactions in the domain. 

The research question that this manuscript addresses (what are the impacts of spatial heterogeneity on subsurface biogeochemistry?) is relevant to the readership of Biogeochemistry. I believe that the authors' numerical modeling approach provides valuable insights into the relevance of spatial heterogeneity for subsurface carbon and nitrogen cycling. In my opinion, the manuscript would benefit from a more in-depth discussion of how the results of the study could be applied at larger scales or to other subsurface systems and potential caveats that would come with this. Furthermore, I think some statements in the manuscript need clarification- see specific comments and questions below. My specific comments also include some suggestions to improve the readability manuscript.

Specific comments:

L1: The term "redox dynamics" is not used in the manuscript (except once when referring to the literature) and I am not entirely sure what the authors want to convey with it.

L58: "in this microbial ecosystems...": what does "in this" refer to?

L78: "Sufficiently well" for what?

L81: Please add citations for the statement on biogeochemical reaction networks. 

L81-82: It is not clear to me how the sentence starting with "Working with ..." fits into the line of arguments here.

L84: "A straight-forward application of the soil-based biogeochemical model approaches to conditions in deeper subsurface compartments is problematic because the nature of carbon source changes as it travels into the deeper zones." I believe the authors did not specifically look into this- is there a reson for this?

L99: I think using the term "mechanisms" is not great here as I think the manuscript does not address this.

L196: "established": can you add references for this?

L196: What are variance and anisotropy values used for the base case? When I first looked at Figs. S1 and S2, I was confused because the homogeneous base case was shown in all variance:anisotropy ratios.

L250: Based on equation 2, the Da value should depend on the size of the domain relative to the size of the "heterogeneity". Have the authors looked into this?

L312: "while the removal of TOC was the lowest there...": is this trend related to microbial biomass?

L391-392: The bars in Fig. S6 are not linked to redox conditions- is it possible to do so?

L441-451: Given that the prediction of the impact of spatial heterogeneity on redox regimes is the major posit of this manuscript, I believe this section needs to be improved. A few suggestions for improvement are given in the following few comments.

L444: What is AIC and what do these values mean?

L449: Where in Fig. 5 can I see these under-/overestimations?

L452: Does this Fig. include data for different reactive species? Also, why are Da numbers given in log10 base (given that per definition Da is already the ln of the concentration ratio between outflow and inflow)? If solutes are consumed in the domain, then Cout/Cin is always < 1 and hence the Da per eq. 2 is negative- which will give a complex number when taking log10. Is there something I am missing here?

L461: I think it would be beneficial to explicitly state the range of the scenarios.

L463: "correlation length": this is not discussed in detail in the results section and I am not sure how it fits in here.

L507-508: "We establish that the persistence of microbial species in the domain is governed by the presence of the appropriate carbon source and electron acceptor, ..." I am wondering how microbial species are linked to carbon sources and electron acceptors in the model. Is species distribution independently modelled or could this finding in part be a result of the way the reaction network and model are set up?

L627: Can the authors elaborate on the significance of this results for environmental systems? For example, when and where do they expect these heterogeneities to be most significant?

L654: Or underestimate nutrient removal six-fold as stated in L443?

Technical corrections:

L18: "used" instead of "undertake"

L83: "group" instead of "groups"

L115: "attempted" instead of "attempt"

L123: "Disentangle" from what? I think describe/define would be better.

L271: I think the "removal" before "impact" should be deleted.

L424: Why are some words bold?

L446: Seems like a repetition of L 443f.

L553-555: A verb is missing in this sentence.

L609: Delete "towards".

Tables: Captions should be above tables, not below.

Figures: It would be helpful to have different symbols for the three flow regimes so that the figures are readable in black and white.

Review of Khurana et al, “Predicting the impact of spatial heterogeneity on microbial redox dynamics and nutrient cycling in the subsurface,” submitted to the journal Biogeosciences.

Summary: This manuscript presents results of numerical experiments exploring the fate of reactive compounds within spatially heterogenous subsurface environments. The conceptual model is based on in-situ observations of dissolved organic matter and microbial functional groups, and models of variably heterogeneity (ranging from homogeneous to highly heterogeneous). The numerical models use a sophisticated representation of reaction networks including biomass growth and transport as well as multi-species reactions. Overall, this manuscript is well-written and organized with only a couple points that require clarification and a few grammatical and formatting mistakes to address. The authors make a clear case for the inclusion of a scaling factor based on the Damköhler number which considers the reactivity of chemical species. The authors provided detailed explanations of the limitations of their model and dataset such as spatial variability of chemical and microbial species within the well screen, which might bias observations from the subsurface.

Specific Comments: There are a few matters that require clarification for the reader and provision of supporting information that is not in the current manuscript:

Line 170: “average hydraulic conductivity” – what average is used (arithmetic, geometric, harmonic)? I would assume arithmetic since not stated otherwise, but often the geometric mean is considered more representative of the average behavior of a heterogeneous permeability field.

Table 1: The “length scale” is given here as 0.1 meters, apparently as used to compute the Peclet number (ref. line 230).  What is that value based on?

Line 195: What is the correlation length used in the simulations? I couldn’t find it in the tables.

Lines 196-198: The outcomes may be sensitive to the assumption of a second-order stationary random field with horizontal anisotropy. Other types of heterogeneity (e.g., multipoint statistical models, geometric models, or depositional process models) could lead to different (and probably even more striking) conclusions. I don’t view this as a flaw of the study, since this assumption is conventional, but the assumption and its potential implications should perhaps be discussed.

Lines 245-246: Clarification is needed here to elucidate what is meant by “fit of the model” in defining the AIC criterion. It isn’t clear whether this is fit to actual field observations (there are some discussed in the paper but they are not described in detail) or fit to analytical solutions (e.g. Figure S8), or something else.

Line 253 (Equation 2): Is this a standard definition of Da? If so, please provide a citation. If not, please provide clarification as to how this equation represents the ratio of advective and reactive time scales.

Lines 265-270: These lines highlight a broader issue: What times were used for analysis and metrics of reactive processes? The flow field is clearly stated as being steady, but the concentration fields will be transient, and various times are mentioned in the manuscript. The breakthrough time is defined as the time at which Cout/Cin = 0.5 (for tracer), but what other times are considered for reactive species (they may not reach this value at the outlet)? This was confusing to me as a reader and should be clarified.

Technical Corrections:

Lines 58-60: awkward wording, suggest rephrasing

Lines 76-78: consider simplifying to “…representative of a system’s chemical and biological species, and second…representative of a system’s flow and transport pathways.”

Lines 102-103: suggest grouping citations at the end of the sentence

Line 153: e.g. seems out of place, consider deleting

Line 155: aerobic should be all lowercase

Line 162: necromass is one word, not two

Line 163: complete the second half of the sentence by describing how microbes become immobilized (biofilms etc.)

Table 1: “longitudinal” is misspelled

Line 270: this is “the” same (add “the” to the sentence)

Line 280: switch scale and spatial in the sentence

Figure 2: Is a title necessary? The figure caption should cover the topic of the figure.

Line 424: unbold sentence

Figure 4: Why is there a border around this figure?

Line 447: change to, “Since nutrient dynamics are..."

Line 482: Is there a difference between dormant and inactive? The term “dormant” is used sparingly within the manuscript, so I’d suggest sticking to inactive and defining that this pool includes dormant microbes to avoid confusion.

Lines 487-489: clunky sentence, suggest rewording

Figure 6: It’s a bit unorthodox to present figures within the discussion section. Why didn’t you introduce it within the results section and then reference it in the discussion?

Line 618: Change to “The regression model links the..." 

Line 631: Change to “… was considered when evaluating…” (delete “for”)