First of all, I really appreciate the effort that the authors went to in responding to my overlong review. On the whole, I’m happy with the changes that they’ve made to the manuscript. However, I do still have some concerns with a few of their responses. I’ve listed the main, overarching ones immediately below, but then give more specific comments to particular responses afterwards.
Some of the explanations for the experimental data still do not make sense (e.g. occurrence of a PO4 spike). Explanations should be consistent between the axenic and non-axenic cultures, and the authors should not be afraid to note where a feature in the data appears inexplicable (i.e. is more likely an experimental / measurement error).
The model equations are still not complete as far as I can see. There are also some cosmetic issues around the use of (unexplained) constants in the equations – it would be better to assign (fixed) conversion parameters.
The explanation of the tuning method is much appreciated. However, as it stands, it’s very dense text that’s been included. It might be better to add a supplementary section that breaks this up into bullet-pointed steps – that would be ugly for the main body, but would probably be very helpful as a supplement interested readers could consult.
While the authors have rephrased their hypotheses, they still make no use of them, nor even refer to them, elsewhere in the manuscript. Plus, and especially because of the way they’re used, these read more like conclusions of the work than hypotheses that the paper actively tests.
Overall, I think that the manuscript is now close to being acceptable once these remaining details are addressed. I would advise accept after minor revisions.
-----
Ln. 161, “The strong spike of phosphate after day 8 corresponds with the end of the exponential phase for algal growth and a spike of ammonium. At the same time bacteria abundances start increasing considerably. Thus, we explain the phosphate peak by increased bacterial regeneration (source of phosphate) and decreased algal uptake (sink of phosphate) at the same day.”
- *Qualitatively* I can see this line of argument, but it doesn't really stack up *quantitatively*. First, the timing's not quite right, with ambient PO4 jumping ~50% on day 8, when phytoplankton are at their peak and bacteria have only just got moving.
- It would be more convincing to me if there was an attempt to budget the elements to make this shift between the medium, the phytoplankton and the bacteria clearer; it remains difficult for me to shake the impression that something has gone wrong in the experiment; and this should be done for both experiments and models; among other things, if it can't be done for whatever reason, it does tend to suggest that, contrary to the authors "hypothesis 3", experiments do not provide a good testbed for ecosystem functionality
Ln. 183, “Excretion of organic phosphate by diatoms is also common in cultures with surplus orthophosphate (Admiraal and Werner, 1983), which can be another explanation of the phosphate peak after silicate becomes limiting.”
- but it only happens in the case with bacteria, so this doesn't make sense; if anything the case without bacteria should be more stressed because there's no nutrient regeneration
Ln. 193, “Figure 4, 5, B1) “…Chlorophyll a concentration in experimental cultures with a potential outlier at day 8, presumably due to photodegradation, causing a negative spike.”
- "photodegradation"? it looks like a straight error to me as it bounces right back; if you are going with photodegradation, expand on this
Ln. 202, “We added the missing equation and double-checked for any other incomplete model descriptions.”
- the equations for DOM are still inadequate; background DOM is listed as a state variable but its equation is omitted
Ln. 206, “Equation 7a) Silicate uptake”
- working through the units in this equation, using the given units for state variables and parameters, I find that there's a discrepancy of 3 orders of magnitude; I think one of the sets of units must be listed incorrectly
Vmax = (mol Si / d) / (mg C)
Sid = umol / L
Smin = umol / L
Ksi = umol / L
C = mg C / m3
dSid / dt = (Vmax * Sid * ((Sid - Smin) / (Ksi * Smin)) * C
= ((mol Si / d) / (mg C)) * umol / L * (umol / L) / (umol / L * umol / L) * mg C / m3
= ((mol Si / d) / (mg C)) * ((umol / L)^2) / ((umol / L)^2) * (mg C / m3)
= ((mol Si / d) / (mg C)) * (mg C / m3)
= mol Si / d / m3
= mmol Si / L / d
≠ Sid ≠ umol / L
Ln. 207, “Equation 7b)”
- the 14 comes from nowhere here; you should make it clear that it's a molar mass
Ln. 235, “Figure 1”
- what is meant by control here?; these lines in particular make this diagram difficult to follow; when I suggested a diagram it was to (a) make clear what the state variables are, and (b) indicate the relationships between them so that the simple food web in the experiment vessels was; this diagram is overcomplicated, while also seeming to omit detail (e.g. bacterial metabolism)
- in this diagram the bacteria - which are meant to be heterotrophic - consume only nitrate; that doesn't make sense biogeochemically
Ln. 258, “2.3 Modelling fitting”
- while thorough, this is very difficult to parse. I would suggest writing it out clearly as a bullet-pointed step-by-step process (you’ve actually begun this in your response elsewhere), and adding this description to the supplementary material. Make clear which models, which parameters, which target datasets and whether tuning was manual or automatic
- On a separate note, I don't think initial conditions used are covered; and some of the plots suggest this is not done as one might expect; e.g. Figure 6 seems to show the model starting at values quite different from the experiments
Ln. 381, “2. Diatoms continue photosynthesis under silicate limitation at a reduced rate if DIN is available”
- what does this mean for cell division?; if not here, you should certainly comment on this somewhere
Ln. 389, “Ammonium is most likely immobilized in the biofilm via adsorption to the EPS and accumulation in pockets unavailable to diatoms”
- this needs a bit of expansion - the model has less NH4 than the observations, and if it was being immobilised in the experiments within the biofilm (something that's absent in the model), one might instead expect lower concentrations, no?
- also, it's not clear which of this text has made it into the revised manuscript
Ln. 520, “In previous cultivation experiments, no efforts for obtaining axenic cultures were mentioned, which hints to bacteria contaminated cultures.”
- I take the point; from my own experience of the tedious nature of maintaining axenic cultures, it's not unreasonable to assume that a study where this isn't mentioned is probably not working with axenic cultures
Ln. 601, “Microscopy showed bacteria attached to the diatoms (mostly in the stationary phase), but mostly free-living.”
- as the audience of this paper may be unclear on this point, it would be good to add a sentence noting your microscopy results so that readers understand you are largely referring to free-living bacteria
Ln. 621, “As mentioned above, we changed the hypotheses in the following way:”
- you still don't return to them!; normally, hypotheses are stated ahead of work then reassessed at the end of the work; these are almost conclusions that are appearing in the introduction!
Ln. 754, “Thus, we implemented a labile (DONl) and refractory (DONr) DON pool with different remineralization rates (rem, remd).”
- as noted the equations for the DONr pool are not included
Ln. 760, “We do not suggest a complete stop of remineralisation …”
- OK - this sounds more reasonable
Ln. 766, “DOM C/N mass ratio”
- why use "mass ratio" when "molar ratio" is more common?
Ln. 775, “We refer to DOM as substrate for bacteria and clarified it: DOM C/N ratios….”
- Please be clear if C:N is mass or molar throughout if you're going to chop and change
Ln. 884, “Collinearity is a measure for the parameter identifiability …”
- this is good, but has it made it into the text?
Ln. 975, “due to the potential uncertainties related to immobilized ammonium”
- as noted above the immobilisation is curious given that the measurements include this extra NH4; is the NH4 only effectively immobilised for the phytoplankton such that when the vessel is stirred, it becomes apparent?
Ln. 1129, “The R code is now available at github”
- if you want to create a permanent record of the model with a DOI, consider using the Zenodo link from Github; this will preserve a version of the exact version used here
- if you do this, make sure to add the DOI to your final manuscript revision
Ln. 1154, “We argue that the large range is plausible since it is i) …”
- on point (i), outliers are usually excluded because they are suspected of being false (e.g. measurement error); on point (ii), you're talking about the size of the range of PO4 on day 14, when there's literally zero range of bacterial variability at day 14 - this doesn't make sense
Ln. 1165, “As mentioned in the results, the bacteria growing towards the end are still in so low …”
- that's OK - it's tough keeping things axenic
Ln. 1192, “we added the output of a prolongated model run in the supplement”
- thanks!
Ln. 1280, “We shortened the table slightly and explained all columns in detail in the corrected version”
- the mean, min and max columns are still not properly explained; for instance, why are there negative values?
Ln. 1313, “We changed the form of eq 14 to the same format as in eq 15”
- OK; but the use of (unexplained) numbers in these equations rather than parameters remains unhelpful
Ln. 1345, “Table A8”
- this table is still confusing; do the rows need identifying labels?
- also, what's with the strange number more than 1 million?
Figure 2
- I'd not really paid attention before, but the experiments have a large span of initial nutrient concentrations - shouldn't this be properly controlled?; or am I misunderstanding the plots?
- your plots consistently run out of x-axis; the experiments are at least 15 days long, but your axis stops at 14 days
- (and thinking back the point made previously) not only does PO4 increase inexplicably between days 6-10 (+100%), it also drops off a cliff between days 14-15 (-80%)
- Panel d runs out of y-axis (as well as x-axis)
Figure 5
- nothing about bacterial biomass and how it fares when excretion is included / omitted
- chlorophyll doesn't decline here but it should; and it does in a later supplementary plot (B1c); something doesn’t seem quite right.
Figure 6
- why do the model lines start at day = 1 and not day = 0?; the observations seem to start at day = 0 by contrast (and unlike in other plots)
- the behaviour of modelled ammonium is still confusing; it seems to have very little to do with what the experiments did; also, in the BAC- experiment, the model seems to start at a lower value than it did in the experiments
Table A2
- what does "mio." mean?; does it mean "million"?; it's not an abbreviation I'm familiar with; and, if so, it looks like bacteria reach concentrations of 60 million cells per mL, but the value for bact_max ranges 0.005-0.1 million cells per mL; this is confusing |