Review of Porz et al., Quantification and mitigation of bottom trawling impacts on sedimentary organic carbon stocks in the North Sea

General comments

Porz et al., conducted a modelling study which uses coupled 3D hydrodynamic simulations with a bioturbation model to look at trawling impacts in the North Sea. I think this may be the most holistic and up to date research on this topic that I’ve seen. Most studies on this topic have just looked at the direct impact of trawling on sediment carbon within a trawl track and some look at how that effect might evolve through time. This study looks at the potential fate of OC once resuspended and shows how trawling can lead to lower OC in some areas and more OC in other areas (something observed in field observations before as well). Furthermore, I was happy to see that in the simulations regarding fisheries closures, trawling effort was redistributed rather than assumed to have decreased. All the praise aside, I think the manuscript still has areas of improvement, most of which I’ve outlined below. My main concern is that, although, I’ve seen it mentioned throughout the paper, it was never clear to me exactly how remineralization was addressed by the model. For me, the parts about the removal of OC through resuspension and redistribution to other areas was better explained but I would like to know exactly in the methods, your assumptions on benthic mineralization after trawling as well as once resuspended. It is stated in the discussion that OC is degraded once resuspended but I would like to see further details behind this like how much mineralization decreased/increased (both in the water column and in the sediment) and why.

Kind regards,

Justin Tiano

Detailed comments

L11 There are reasons why 3D hydrodynamic models provide much needed insight on this topic but this is not yet clear to the reader. It would be nice to transition from the debate amongst scientists to something like “However, current discussions around the fate of resuspended organic matter are lacking. To help resolve this, we used 3D hydrodynamic… “

L25 is there anything that could be said about what makes these sediments vulnerable? High labile OC/biomass, low natural disturbance etc.?

L30 so does this give reason to advocate for spatial as opposed to effort management?

L49 “seabed destruction” is vague and perhaps too loaded of a term here. Can you be more specific?

L70 What I’m missing here are explicit reasons ‘why’ 3D hydrodynamic models may be useful to this topic. The fate of OC was touched on earlier in the introduction but it would be nice to make it clear to the reader here that while some OC will be mineralized by trawling, some will be resuspended and redeposited (I do see it is touched upon later on). Maybe state things about ‘how most models do not take into account resuspension and are thus not able to ‘track’ the fate of OC particles. This study aims to reconcile this through the use of 3D hydrodynamic model.’

Introduction in general: It would be helpful to have some information on potential mechanisms affecting OC dynamics. What causes trawl induced CO2 release? What causes CO2 sequestration in sediments. How does trawling potentially increase mineralization (like though O2 exposure from resuspension) or benthic mineralization? What is the role of macrofauna on these processes?

Fig 1. Instead of italics, are able to use bold or colored text? That would make the words “pop” out more and make things more obvious to readers.

Table 1. In De Borger et al., 2021 we always found reduced total remineralization (the paper is mentioned under studies that show ‘increased remineralization’). There are a few instances of increased oxic mineralization (also with relative changes) but never an increase in total mineralization in that study. Morys et al., 2021 also showed lower rather than higher mineralization (lower benthic respiration as a proxy) as well as Bradshaw et al., (2021) which was not included in the table.

Methods

L116 as the other models were named, can you please also give the name of some more details for this ecosystem model.

L158 Metiers provide more detail than gear types as they account for differences within gear types (mesh size, target species etc.). Perhaps rephrase as readers might see the first sentence (not differentiating between gear types) and become critical without understanding what a metier is.

L246 It’s not clear to me that this methodology reflects accurate mixing rates from trawling yet. Is this sentence aiming to state that the mixing coefficients are similar to measured trawled areas? The sentence talks about expected bioturbation which makes me think that the mixing coefficients are just similar to that of high bioturbation by fauna. Skeptical readers may want some more evidence stating how this method of calculating trawl induced mixing reflects real conditions. (Perhaps just rephrasing is necessary).

L305 Experts will understand the logic for looking only at oxic mineralization as this increases relative to anoxic mineralization after trawling. Many readers may not know that so perhaps add a sentence stating that and why anoxic mineralization is not taken into account.

Methods general: Perhaps I missed it but what I failed to find was how you explicitly account for changes in mineralization in sediments that were trawled. You may have a relative increase in oxic mineralization but we have found that the total mineralization decreases in sediments after trawling as OM is removed from the system. Deposition of OM from trawling may increase total mineralization as it introduces new OM to a system (though the the opposite can also happen if it smothers benthic fauna). Mineralization in the water column also increases. I see that Table 1 shows the effects considered but it's not clear to me how reduced respiration and increased mineralization are specifically incorporated.

Also, I’m not sure where you link the resuspended sediment from trawling to the hydrodynamic model to see where resuspended sediment and OC ends up. It seems like the model takes into account natural resuspension but the link with trawling resuspension is not clear to me.

Results

L355 As dead benthos end up in the pool of OC since they’re not “removed” from fishing (fishers only remove target fish biomass), how is this accounted for in the results?

L356: The loss of 14% of benthos in the North Sea can be taken out of context here. Maybe state that this is the difference between trawling and no trawling scenarios.

Use of 'REF' (reference simulation): I see why you have chosen REF/reference as it represents the status quo of trawling in the North Sea. Nevertheless, I was often confused as a reader since most of the time I see this term (reference) used, it is synonymous to 'control' conditions which are typically undisturbed such as in experimental studies. I would consider using a different term like 'SQ' for 'status quo' simulation or 'baseline' simulation (BASE) to not confuse readers in a similar manner.

L368: Trawling pressure is highest in the summer therefore their effect is also high. OC influx is also highest in the spring and summer so does that mitigate some of the trawling effect.

Discussion

L427: I imagine this may be quite difficult to incorporate to the models but how would you answer the question about how trawling leads increases in certain types of macrobenthos like benthic scavengers and more r selected species? In Tiano et al., 2020, we speculate that trawling may have led to more large infauna (sediment mixers/bioirrigators) occuring in the Frisian Front as they tend to survive trawling effects by living deeper in the sediment. Potential discussion here, though our example may be a special case.

L430: So can you say that the results suggest a greater direct effect on benthos rather than OC?

L456: As OC influx is lower in the winter, are fisheries closures less effective during these times? I was wondering that since MPA's in the low OC Dogger Bank show little effect, perhaps it's the same during the time of year when OC may be lower.

L521-528: I would specify this explanation of the depositional pattern in the models also somewhere in the methods. It is relevant here in the discussion as it explains certain model limitations but I wanted to ask questions on this topic much earlier in the manuscript so an explanation on how the model handles annual OC deposition early in the methods would be nice.

L587: I'm confused now. I thought all 'reference' simulations were the status quo trawled simulations? Please check this and make consistent throughout the paper.

L608: Perhaps you can also (re)state here how much of the North Sea is closed for the different scenarios.

L610: I agree with you guys. This is a really good study, great job!

Review of Porz et al. manuscript "Quantification and mitigation of bottom trawling impacts on sedimentary organic carbon stocks in the North Sea" by Pere Puig.

General comment:

This paper presents data from a 3D coupled numerical model that is used to quantify the major impacts of bottom trawling on organic carbon and macrobenthos stocks in North Sea sediments. The authors simulate six years of trawling activity and consider four management scenarios in which trawling effort is redistributed from areas inside to areas outside of trawling closure zones. Overall, the paper is well written, but in specific sections, the authors’ reasoning is difficult to follow because the manuscript lacks of details about how the data has been treated/modelled (please, see some examples in the specific comments). Some of the implications are well sustained by the presented data, but in some sections, the paper becomes speculative. The authors should try to modulate the conveyed messages and not oversell their ideas if they are not well supported by the presented results. Nevertheless, this is a very nice scientific contribution addressing the spatial impacts and the transport and fate of the resuspended C in the heavily trawled North Sea, which could serve as inspiration for similar future modelling exercises -currently lacking in the scientific literature- and that will help us to properly understand bottom trawling impacts at regional scales and to constrain global estimates.

Specific comments:

L 95-100: These coupled models account for Hydrodynamics (SCHISM) sediment dynamics (MORSELFE) and for interactions of OC and macrobenthos in the sediment (TOCMAIM), but what about the C remineralization? How it has been addressed? Perhaps something on this regard should be mentioned here, at the beginning of the Methods section.

L 109-110: It is weird to me to see the three different OC pools based on their bioavailability and degradation rates (fresh, semilabile and refractory) being considered as sediment classes, at the same level than inorganic particles (sand, silt, and clay), and see afterwards in Table A1 that all three OC pools, in term of their sediment dynamics properties, have been associated to the silt class. This is mentioned later in the manuscript (L 183) but to follow how the model has dealt with the OC resuspension, perhaps it should be clarified first here.

L 140: Mention here from which period the daily time series of trawling effort from the Global Fishing Watch dataset were extracted. It is introduced latter (L 150), but the reader should know it before, otherwise the paper becomes “mysterious”.

L 149: It is unclear why the authors chose the simulation period of 2000-2005 and the daily fields (of GFW trawling effort?) of 2015-2020 averaged; and it is even less clear how the scaling according to the annual historical landings of demersal fish reported in ICES was conducted afterwards. Perhaps it would have been easier and more realistic to simulate the period of 2015-2020 using the GFW trawling effort, since the simulation period of 2000-2005 based on historical landings could be spatially biased (despite the Couce et al. (2020) findings). In any case, this point should be clarified, regardless the reason behind it. Also, later in the paper, the simulation period of 2000-2005 is used as a reference (REF), but perhaps it would be good to mention this already here, and perhaps restate it or in the introduction of the Management scenarios (section 2.4).

L 183-185: In the paper, only the silt content is considered to estimate the resuspension rate of trawlers, but, generally, OC content increases in muddy sediments that are finer than silts. What about the clay fraction? Something should be mentioned on this regard.

L 256: Define here what is exactly the reference simulation (REF). This is the first time that this acronym is used.

L 340-344: The Figure 5, illustrating the change in average trawling resuspension and erosion rates, is hard to follow if previously the corresponding maps of redistribution of trawling for each scenario (similar to the map in Figure 3) are not shown. Otherwise, the authors skip the illustration of one critical step of the computing process, which is the redistribution of trawling effort. The limits outlining the areas of trawling closure zones are hard to discern, and in some maps it is difficult to identify if the areas are inside or outside the lines. Perhaps the areas should be filled with a hatched pattern. Additionally, the time constrain is not mentioned in the figure caption.

L 390-394: The same as in Figure 5. It t would be desirable to see the trawling effort maps before presenting the average differences of changes in OC fluxes in Figure 7, and also the areas of trawling closure zones should be hatched, since they are hard to discern.

L 430-432: This sentence needs a proper reference to support such a strong statement. This is an example of the speculative sentences found throughout the text.

L 444-445: Macrobenthos biomass responses are shown for the first time in the Discussion section, while they should be included in the Results section.

L 516-579: I miss the Model limitations subsection some paragraph dealing with the need to improve the computation of the sediment and C resuspension, transport and re-deposition processes, which have been treated quite simplistically in this modelling exercise. To me, this is a key aspect, since most of the distribution maps and the computation of the C fluxes caused by trawling largely depend on this parametrization. Besides, several of the listed “limitations” on this subsection are not inherent of the models used (SCHISM, MORSELFE and TOCMAIM), but instead they are aspects that could not considered or addressed in the paper using these models. Perhaps the title of this subsection could be renamed as "Model limitations and unaddressed processes/mechanisms".

L 582: Define the period during which the daily time series were generated.

L 585: Again, the period of the six consecutive years is missing.

L 706: There is no mention to the availability of the code of the MORSELFE model.

Sincerely,

Pere Puig