Dear Professor Roberts,

we would again like to cordially thank you for your highly qualified and well-founded comments on our manuscript, as well as for your kind words and benevolent résumé. In our author comment ‘AC2’ from Nov.5, we responded to your individual comments already to encourage further discussion. Since we posted that author comment, our conception of the manuscript has not significantly changed so that the present “final response” corresponds 95% with the cited earlier author comment. We only specified a few of our statements. In the following, we list our thoughts and plans referring to individual comments:

1) In the introduction please also cite some of the papers that have used 36Cl to study fault scarps in the Italian Apennines.

→ Nothing speaks against mentioning these papers in the introduction. Furthermore, we will check if the list of cited papers could be expanded by further useful literature.

2) At the end of the introduction please add a few sentences setting out the structure of the work conducted: mapping, sample collection, 36Cl sample prep, AMS, modelling of 36Cl; tectonic interpretation.

→ Certainly, these aspects should be part of an introduction so that we will readily accept this task!

3) Line 86 I think you should cite Cowie et al. 2017 which gives the most complete account of what geomorphic requirements need to be satisfied for fault plane 36Cl sample sites.

→ We will certainly do so.

4) Line 91 Typo? Do you mean 15 ±3 ka? That is what most people use.

→ This seems to be a matter of debate, as can be seen by the fact that the second reviewer suggested 20 ka in this context. We chose 18 ka in all conscience, after (from our point of view) synthesizing an ideal age from works in the surrounding areas. In section 5.1, we discuss the reliability of this age which is, e.g., also used by Papanikolaou et al. 2005, or Giraudi and Frezzotti, 1995. As our slip rate calculations based on the proposed LGM age are easily reproducible and “convertible” from the provided tables, we prefer to maintain our 18 ka age – also as a compromise between both reviewers.

5) Line 96 Please explain why you think 50 cm sample spacing is adequate. Some would argue you need to use denser sampling to identify so-called cusps (e.g. Schlagenhauf et al. 2010) whereas others (e.g. Beck et al. 2018) suggest less dense sampling is fine as long as the modelling approach takes this into account.

→ We will clarify this in the text and refer to Beck et al. 2018.

6) Line 99 Cowie et al. (2017) were the first to say that a trench at the bottom is needed, not Mechernich so cite that paper.

→ No objections. We will certainly do so.

7) Section 3.2.2 on the 36Cl modelling needs more detail. I agree that the stripes on the fault plane mean spotting cusps is unlikely to work and hence I support sampling sparsely (and within available funding/logistics constraints), at least in this initial study. However, the Schlagenhauf et al. (2010) code is usually used to spot cusps, so I think you need to write some justification of why you think it is OK to use it on your 50 cm sample spacing.

→ This will be added. 

... I think it is a good code to use as a first pass, perhaps prompting modelling with other codes in a later paper if you gain more 36Cl samples that may provide more insights (e.g. Bayesian modelling, evidence of convergence between Markov chains, iteration of variables such as colluvial densities, attenuation lengths, production rates, slip per event, age of initial 36Cl production/scarp age, etc.). But to use the Schlagenhauf code one must show/state some things that are used in that code (e.g. pre-exposure). Please state/show the following in the text or in a supplement: (1) value for pre-exposure, with some justification for why that value was chosen;

→ This value is given in Table S13. We will add a justification.

... (2) provide a data table with all elemental compositions for each sample, or at least what you have, with Ca concentration vital;

→ This was already prepared in Tables, but unfortunately not uploaded, see author comment ‘AC1’.

... (3) how you use the Schlagenhauf code if you do not try to resolve cusps, that is how you choose and propose earthquake slip histories and their implied 36Cl concentrations for comparison with the measured concentrations;

→ A quick explanation how we use the code will be added. 

... (4) how and why you model the “sliding event” in Fig. 7.

→ An explanation will be added.

8) Section 4.2.2 provides useful information, and its contents should be published because they are interesting. However, please provide more detail. Tell us exactly why you think there is a robust relationship between the slip history you propose and the measured 36Cl concentrations. In other words, explain your results and how you derived them, rather than just stating what you think the results are. How do the model results relate to the data error bars for example.

→ We will extend this paragraph by adding the methodology including further points to the uncertainty calculation and the results in more detail.

9) Section 4.2.2 should also be longer. I would expect the results section to be significantly longer, with text explaining what exactly the reader should look at in each of the “results” diagrams”, with a summary at the end explaining the overall result which would set the scene for the following discussion section.

→ As mentioned in point 8, we will expand this chapter. Good idea to include a summary highlighting the overall results of the ³⁶Cl modelling.

10) Section 4.2.2 should also perhaps discuss other possibilities for the 36Cl modelling results, stating why the chosen one is thought most likely to be correct. For example, the “result” that there is a “sliding event” (see Fig. 7) needs more explanation. Why is the 7c the “most likely” (see the caption)?; please explain. Is there geomorphic evidence for a “sliding event”? Please describe it, or if not say so.

→ We say in line 234, that there is no evidence of a “sliding event” 

... How is this constrained with the modeling? Do you mean a landslide event? If so, please clarify.

→ Thanks for highlighting that our ideas on the “sliding event” are not yet fully comprehensible. We will include further clarifications on the modelling of the “sliding event”. 

... Another example is the claim that slip commenced at 6 kyr ago (see abstract). Can you clarify why you think this? Could it not also be that slip is clustered, with a cluster starting at 6 kyrs BP, with a period of no slip before that (an anticluster), perhaps with other clusters and anticlusters in the time period before that resolved by your 36Cl data? In other words, perhaps the slip and the new tectonic regime is not so “incipient” as you claim in the title of the paper. In other words, (a) in an interpretation that considers clustering, slip did not “commence” at 6 kyrs BP, but rather long-term slip was ongoing before then, but a cluster started at 6 kyrs BP, whilst (b) in an interpretation that does not consider clustering, slip “commenced” at 6 kyrs and so the deformation is “incipient”. I think the paper would be improved if both of these possibilities were considered (a clustered interpretation and one with no clustering). I think the paper would be cited more widely if you include both. However, this is up to the authors and I do not insist on this.

→ Indeed, the “commenced at ca. 6 kyr” in the abstract is misleading and we will clarify this. We will check if it is reasonable and fitting to this ms to include a discussion on a clustering of the earthquake events.

11) Line 201 Typo? 15 ±3 ka?

→ No typo, see # 4.

12) I found the discussion section interesting and thought provoking, which is good.

→ Thanks a lot for this motivating comment! With your further constructive but more critical comments, we are stoked to further improve and put the final touches to our manuscript.

13) In the supplement, please re-organise and rotate the photographs and diagrams so that they can be viewed without having to rotate the page. Most people will read this as a pdf and having to rotate pages can be annoying.

→ We will do so. 

14)  In Fig. S7 use a linear rather than log scale for the y axis, as this is the standard approach for this type of plot.

→ We are fully aware and agree with you that this is the common way of illustrating such plots. However, we purposedly decided for a log scale since we are looking at a large range of free face (eroded and non-eroded) heights. In a linear scale, especially the (particularly interesting) non-eroded scarp heights would hardly be distinguishable. We therefore prefer to keep the present display.

15) Fig. S8 Please indicate the source of the topographic data in the caption.

→ Thanks for this valuable hint - we totally missed this. Those are TanDEM-X data which we will certainly cite.

16) Please add the rock geochemistry for the 36Cl samples to the supplement.

→ See # 17: This was part of an xls sheet that we simply forgot to upload.

17) I have a slight concern that I may have missed some supplements (apologies if this is the case), but I found it slightly difficult to be sure I had accessed all available material on the review website.

→ This was our mistake, as we forgot to upload an xls sheet that was supposed to be part of the supplement. This happened, as all other supplementary material (except for the xls file) is combined in one PDF. We immediately uploaded the missing file to the discussion. We apologize for the inconvenience and thank you for disclosing this problem.

Dear Professor Roberts,

we would again like to cordially thank you for your highly qualified and well-founded comments on our manuscript, as well as for your kind words and benevolent résumé. In our author comment ‘AC2’ from Nov.5, we responded to your individual comments already to encourage further discussion. Since we posted that author comment, our conception of the manuscript has not significantly changed so that the present “final response” corresponds 95% with the cited earlier author comment. We only specified a few of our statements. In the following, we list our thoughts and plans referring to individual comments:

1) In the introduction please also cite some of the papers that have used 36Cl to study fault scarps in the Italian Apennines.

→ Nothing speaks against mentioning these papers in the introduction. Furthermore, we will check if the list of cited papers could be expanded by further useful literature.

2) At the end of the introduction please add a few sentences setting out the structure of the work conducted: mapping, sample collection, 36Cl sample prep, AMS, modelling of 36Cl; tectonic interpretation.

→ Certainly, these aspects should be part of an introduction so that we will readily accept this task!

3) Line 86 I think you should cite Cowie et al. 2017 which gives the most complete account of what geomorphic requirements need to be satisfied for fault plane 36Cl sample sites.

→ We will certainly do so.

4) Line 91 Typo? Do you mean 15 ±3 ka? That is what most people use.

→ This seems to be a matter of debate, as can be seen by the fact that the second reviewer suggested 20 ka in this context. We chose 18 ka in all conscience, after (from our point of view) synthesizing an ideal age from works in the surrounding areas. In section 5.1, we discuss the reliability of this age which is, e.g., also used by Papanikolaou et al. 2005, or Giraudi and Frezzotti, 1995. As our slip rate calculations based on the proposed LGM age are easily reproducible and “convertible” from the provided tables, we prefer to maintain our 18 ka age – also as a compromise between both reviewers.

5) Line 96 Please explain why you think 50 cm sample spacing is adequate. Some would argue you need to use denser sampling to identify so-called cusps (e.g. Schlagenhauf et al. 2010) whereas others (e.g. Beck et al. 2018) suggest less dense sampling is fine as long as the modelling approach takes this into account.

→ We will clarify this in the text and refer to Beck et al. 2018.

6) Line 99 Cowie et al. (2017) were the first to say that a trench at the bottom is needed, not Mechernich so cite that paper.

→ No objections. We will certainly do so.

7) Section 3.2.2 on the 36Cl modelling needs more detail. I agree that the stripes on the fault plane mean spotting cusps is unlikely to work and hence I support sampling sparsely (and within available funding/logistics constraints), at least in this initial study. However, the Schlagenhauf et al. (2010) code is usually used to spot cusps, so I think you need to write some justification of why you think it is OK to use it on your 50 cm sample spacing.

→ This will be added. 

... I think it is a good code to use as a first pass, perhaps prompting modelling with other codes in a later paper if you gain more 36Cl samples that may provide more insights (e.g. Bayesian modelling, evidence of convergence between Markov chains, iteration of variables such as colluvial densities, attenuation lengths, production rates, slip per event, age of initial 36Cl production/scarp age, etc.). But to use the Schlagenhauf code one must show/state some things that are used in that code (e.g. pre-exposure). Please state/show the following in the text or in a supplement: (1) value for pre-exposure, with some justification for why that value was chosen;

→ This value is given in Table S13. We will add a justification.

... (2) provide a data table with all elemental compositions for each sample, or at least what you have, with Ca concentration vital;

→ This was already prepared in Tables, but unfortunately not uploaded, see author comment ‘AC1’.

... (3) how you use the Schlagenhauf code if you do not try to resolve cusps, that is how you choose and propose earthquake slip histories and their implied 36Cl concentrations for comparison with the measured concentrations;

→ A quick explanation how we use the code will be added. 

... (4) how and why you model the “sliding event” in Fig. 7.

→ An explanation will be added.

8) Section 4.2.2 provides useful information, and its contents should be published because they are interesting. However, please provide more detail. Tell us exactly why you think there is a robust relationship between the slip history you propose and the measured 36Cl concentrations. In other words, explain your results and how you derived them, rather than just stating what you think the results are. How do the model results relate to the data error bars for example.

→ We will extend this paragraph by adding the methodology including further points to the uncertainty calculation and the results in more detail.

9) Section 4.2.2 should also be longer. I would expect the results section to be significantly longer, with text explaining what exactly the reader should look at in each of the “results” diagrams”, with a summary at the end explaining the overall result which would set the scene for the following discussion section.

→ As mentioned in point 8, we will expand this chapter. Good idea to include a summary highlighting the overall results of the ³⁶Cl modelling.

10) Section 4.2.2 should also perhaps discuss other possibilities for the 36Cl modelling results, stating why the chosen one is thought most likely to be correct. For example, the “result” that there is a “sliding event” (see Fig. 7) needs more explanation. Why is the 7c the “most likely” (see the caption)?; please explain. Is there geomorphic evidence for a “sliding event”? Please describe it, or if not say so.

→ We say in line 234, that there is no evidence of a “sliding event” 

... How is this constrained with the modeling? Do you mean a landslide event? If so, please clarify.

→ Thanks for highlighting that our ideas on the “sliding event” are not yet fully comprehensible. We will include further clarifications on the modelling of the “sliding event”. 

... Another example is the claim that slip commenced at 6 kyr ago (see abstract). Can you clarify why you think this? Could it not also be that slip is clustered, with a cluster starting at 6 kyrs BP, with a period of no slip before that (an anticluster), perhaps with other clusters and anticlusters in the time period before that resolved by your 36Cl data? In other words, perhaps the slip and the new tectonic regime is not so “incipient” as you claim in the title of the paper. In other words, (a) in an interpretation that considers clustering, slip did not “commence” at 6 kyrs BP, but rather long-term slip was ongoing before then, but a cluster started at 6 kyrs BP, whilst (b) in an interpretation that does not consider clustering, slip “commenced” at 6 kyrs and so the deformation is “incipient”. I think the paper would be improved if both of these possibilities were considered (a clustered interpretation and one with no clustering). I think the paper would be cited more widely if you include both. However, this is up to the authors and I do not insist on this.

→ Indeed, the “commenced at ca. 6 kyr” in the abstract is misleading and we will clarify this. We will check if it is reasonable and fitting to this ms to include a discussion on a clustering of the earthquake events.

11) Line 201 Typo? 15 ±3 ka?

→ No typo, see # 4.

12) I found the discussion section interesting and thought provoking, which is good.

→ Thanks a lot for this motivating comment! With your further constructive but more critical comments, we are stoked to further improve and put the final touches to our manuscript.

13) In the supplement, please re-organise and rotate the photographs and diagrams so that they can be viewed without having to rotate the page. Most people will read this as a pdf and having to rotate pages can be annoying.

→ We will do so. 

14)  In Fig. S7 use a linear rather than log scale for the y axis, as this is the standard approach for this type of plot.

→ We are fully aware and agree with you that this is the common way of illustrating such plots. However, we purposedly decided for a log scale since we are looking at a large range of free face (eroded and non-eroded) heights. In a linear scale, especially the (particularly interesting) non-eroded scarp heights would hardly be distinguishable. We therefore prefer to keep the present display.

15) Fig. S8 Please indicate the source of the topographic data in the caption.

→ Thanks for this valuable hint - we totally missed this. Those are TanDEM-X data which we will certainly cite.

16) Please add the rock geochemistry for the 36Cl samples to the supplement.

→ See # 17: This was part of an xls sheet that we simply forgot to upload.

17) I have a slight concern that I may have missed some supplements (apologies if this is the case), but I found it slightly difficult to be sure I had accessed all available material on the review website.

→ This was our mistake, as we forgot to upload an xls sheet that was supposed to be part of the supplement. This happened, as all other supplementary material (except for the xls file) is combined in one PDF. We immediately uploaded the missing file to the discussion. We apologize for the inconvenience and thank you for disclosing this problem.

Dear Prof. Benedetti,

we would again like to thank you for the comments on our manuscript. In our author comment ‘AC2’ from Nov.5, we responded to your individual comments already to encourage further discussion. Since we posted that author comment, our conception of the manuscript has not significantly changed so that the present “final response” corresponds 95% with the cited earlier author comment. We only specified a few of our statements. Your overall comprehensive review is certainly valuable and will undoubtedly add to the quality of the final outcome. However, we were surprised to be accused of a ‘provocative and assertive tone’ of our contribution. In terms of content, a detailed look at particular points of the criticism expressed left us with the impression that our manuscript might have been (at least in parts) misunderstood (see responses below). Many aspects that are allegedly missing or misreported, can clearly be found in the manuscript. Nonetheless, we emphasise that we are in no way resentful, but highly motivated to use all of your comments to complement and re-arrange the text in order to highlight several aspects and conclusions. In the following, we are responding to individual comments:

1) The question of the surface expression of the faults affecting an area and how those are interpreted in terms of kinematics and seismotectonic of an area is crucial. However, the authors somehow avoid to thoroughly discuss this question, and very quickly interpret those faults as active and as the surface expression of an active extension, however normal faults have been observed also in compressional context (see my last comments)

→ In chapter 2 we give an outline of the seismotectonic and kinematic setting in our study area and its surroundings, including a comparison with the Italian side. We also highlight that the location of the discovered fault scarps is very close to an already known and well-approved (the according references are presented) transition between (a) extensional tectonics in Eastern Albania and (b) contractional tectonics along the coast. The reasons why we interpret the discovered features as active normal faults are detailedly presented in chapter 4.1 and further scrutinised/discussed in chapter 5.3. We are puzzled about the reproach of not considering normal faults in compressional contexts, as we repeatedly bring this explanation into play (e.g., LL 27 ff.; very clearly in LL 252 ff. and 272 ff.) Still, we take this as an opportunity to make our points even clearer and expand our elaborations on the regional tectonic setting!   

2) The tone of the paper somehow provocative and assertive […]

→ This was not our intention and we will certainly look for passages that could be ‘mitigated’. Still, we are surprised about this comment as we especially tried to formulate the more ‘daring’ statements with extreme caution (e.g. L 22 - ‘…suggesting that…’; L 24 – ‘…might be induced…’; L 47 – ‘possibly seismogenic’ etc.). Appositely, we bluntly discuss more than one possible formation mechanism of the introduced structures which (in our opinion) is anything but provocative and assertive, but rather well-balanced.

3)  On Figure 4, the colour are difficult to distinguish but the normal faults appear to correspond to the contact betwen Mezosoic carbonates and Eocene or Paleocene. This is puzzling since if there is activity over the Quaternary there should be some Quaternary deposits on the hanging wall attesting for the hanging wall subsidence […]

→ We agree that the colours of the official geological map in Fig. 4 are hard to differentiate. We will adjust the colours and add signatures to distinguish the units more clearly from each other. However, the misinterpretation that the scarp corresponds with the tectonic contact of the nappe probably goes back to the map scale. We will therefore create an inset ‘zooming into’ the direct vicinity of the fault scarps. One of the scarps (KFS) is close to a Mesozoic/Eocene thrust contact, but clearly does NOT coincide with it. The KFS is in the middle of the Triassic units as illustrated on the map. The Quaternary deposits are present (described in chapter 4.1, LL 158 ff.), but are too small to be illustrated in the chosen map scale in Fig. 4. Both aspects will be visible by adding a larger-scale map, which we will readily do.

4) Moreover I have not seen in the paper a mention about the bedding of the carbonates. It is important since hanging valleys could appear as such if the bedding is vertical and not be related to the recent fault activity. Is it possible that those are exhumed features due to active folding?

→ The bedding of the carbonates and its relation to the fault planes is clearly described in chapter 4.1, L157, and shown in the cross sections (Fig. 4 B and C). Since this specification was apparently hard to find, we will highlight it more clearly and we will add the according explanation to the legend of Fig. 4.

5) The assumption that the fault scarps are 18 ± 3 kyr supposed that all faults started to resume activity at that time which is not correct since some faults could have started to resume a seismic activity later for example 10 or 12ka ago, with a long quiescence time between 15 and 20 ka ago. The 36Cl dating is an absolute dating of the scarp exhumation whatever the cause for this exhumation, seismic or others. So I don’t think the comparison brings anything to the paper and does not make the slip-rate calculation convincing to me. You can use the assumption that those scarps are post-glacial if you have no dating but if you have an absolute dating you can discuss this assumption by mentionning that the yielded ages for the fault scarp are in agreement with an hypothesis of post-glacial exhumation but note use an age based on an hypothesis to compare a result you yield with an absolute dating. Moreover the LGM in the Appenines is probably closer to 21 kyr ago and this could be different in the Dinarides […]

→ We agree that the obtained absolute dates are the more ‘solid evidence’. Therefore, we will move this part of the study clearly to the front and present the second approach only for comparison – as you suggested. In terms of the LGM timing, we will keep our 18 ka frame for the same reasons explained in response # 4 of Prof. Roberts’ comments. First of all, the exact timing seems to be a matter of debate, as can already be seen by the discrepancy between two reviewers. We chose 18 ka in all conscience, after (from our point of view) synthesizing an ideal age from works in the surrounding areas. In section 5.1, we discuss the reliability of this age which is, e.g., also used by Papanikolaou et al. 2005, or Giraudi and Frezzotti, 1995. As our slip rate calculations based on the proposed LGM age are easily reproducible and “convertible” from the provided tables, we prefer to maintain our 18 ka age – also as a compromise between both reviewers. 

6) The identification of earthquake slip is based on qualitative observations that are very difficult to interpret in my opinion. The pictures presented do not allow the reader to actually reproduce those observations or appreciate their quality. The authors do not discuss their origin at all and interpret them as seismic exhumation. This is very discutable and should not be presented as straighforward. How are those ribbons oriented in comparison to the slope ? Could snow or others processes of erosion produced similar features ? how the 5 horizons were distinsguished ?

→ With all due respect, we disagree. As a look at chapter 4.1, LL 163 ff. and chapter 5.3 shows, it is not true that a precise description of the ribbons and a discussion of their formation mechanisms are missing. Still, we are willing to further specify and expand these aspects. To us, the constant widths of the up to 5 ribbons (also illustrated in Figs. S7 A-C) across 48 locations on the fault scarps exclude other (erosional) formation mechanisms. Especially snow (as suggested) is regarded as extremely unlikely: (a) Snow as an extremely short-lived phenomenon in this coastal climatic setting at low elevation will probably not leave such distinctly visible marks. (b) Snow would possibly never create such uniform ribbon widths, as snowdrifts would certainly yield variable thicknesses of the snow blanket. We will add some of these arguments to the manuscript.  

7) Moreover, the way they are presented in the abstract is misleading because it suggests that the slip amount and age is deduced from the 36Cl profile. While it is not possible to retrieve event on a 8 m-high scarp with such low resolution (5 samples). The way the age of the event is retrieved is not clear. Did the authors introduce the slip yielded from the ribbons observations and injected those values in the model as a direct model to yield the ages ? This has to be much better explained in the text. 

→ We agree that both approaches and their results could be separated more clearly in the text. We are aware that five ³⁶Cl samples yield a relatively low resolution, but as the other reviewer, Prof. Roberts, recognized in his review: This is only a first approach on which further studies may build upon. Even if we agree that a higher sampling density would bring "nicer" results, we do not expect any other conclusions compared to a sparser sampling, as we think the results are very clear. We will emphasize this in the text.

8) In the introduction the relations between the normal faults identified and the present day kinematic of the area is problematic to me. The presence of normal faults in the Apennines and in Albania is in agreement with the seismotectonic of the area while geodesy and focal mechanisms support no active extension in the Dinarides. So mixing those aspects in the introduction is misleading. Even more that the authors have not yet shown their observations and discuss the origin and the mechanisms underlying their observations. So I would present all this with much more caution, saying that while in the Dinarites active tectonics is driven by compression, the presence of those two normal faults is puzzling and the purpose of your pape is to understand how those features can be interpreted. First by answering the question, are those faults active or not ? The fault potential activity should be thoroughly discussed, after reading the paper as it is I am not convinced that those are active normal faults. Second, if we assume those faults have been active over the Quaternary, they could be the surface expression of flat and ramp fold as it has been described during the El Asnam earthquake in 1980. The mechanical processes is explicitated in this paper Avouac, J. P., Meyer, B., & Tapponnier, P. (1992). On the growth of normal faults and the existence of flats and ramps along the El Asnam active fold and thrust system. Tectonics, 11(1), 1-11. Such possible explanation should be added in the discussion and the bibliography concerning surface expression of folding should be thoroughly studied and discussed in that paper. It could make the paper much more appealing. If those faults are actually the surface expression of the fold an thrust affecting the Dinarides they could indeed be use to retrieve the seismic history of compressional events.

→ On the one hand, this shows us that we may need to consider a re-arrangement of the text blocks and given information or at least add aspects to the introduction. On the other hand, we are surprised about the suggestion to present the normal faults in a compressive regime as ‘puzzling’, because this is exactly (literally!) what we do in L. 41. Also, we are of the opinion that our expressions like ‘possibly seismogenic’ (L. 47) or ‘the occurrence of these newly discovered structures is still fully unexplained’ (L. 43) could hardly be more cautious. However, we will try to rephrase these sentences with even more modesty. Initially referring to the introduction it is suggested that aspects should be detailedly described and discussed which we think is better preserved in the results and discussion chapters. In exactly these chapters (especially 4.1 and 5.3 and 5.4) we describe and discuss why exactly we interpret the structures as active normal faults and which formation mechanisms could be responsible.

9) line 23-26:  all those aspects are purely speculative and should not be in the abstract, you have not proven or provide strong evidence for a kinematic change and no evidence of geophysical observations showing the upper plate of the slab is affected. 

→ Again, with due respect, this is not speculative at all, but proven by numerous (cited) publications. The transition from a (close-by) hinterland extensional (slab-tearing-induced) to a compressional domain near the coast is well-acknowledged (e.g., Dumurdzanov et al., 2005, Handy et al., 2019) for a long time already, while recent publications (Pondrelli et al., 2021) suggest that this transition could even be closer to our study area than hitherto expected.    

10) line 41: you probably mean instrumental earthquakes and not historical ?

→ True. We will change this.

11) line 47-49: please look carefully in the litterature about normal faults in active fold and thrust belt, they can also be the surface expression of contraction (see my comments below but there are probably more examples now since El Asnam).

→ See #1. We repeatedly introduce this explanation approach.

12) line 56: Extension in the Apennines is also attributed to Adria microplate rotation (see papers by D'Agostino et al. 2008, Nocquet 2012), please also cite those papers.

→ This is exactly what we say. We literally speak of both (LL 53 and 56), the Apennines and Dinarides. Furthermore, we do already cite the work of D’Agostino. Of course, we will add Nocquet as well.

13) line 63: it is not a view, this is based on evidences and before considering them obselete you should at least present your evidence and discuss the previous published ones. The tone is problematic to me, it is not an opinion paper, it is a scientific paper.

→ We are still not sure where we present an opinion: The transition between an extensional and a compressional domain in the Hellenides/Dinarides is a well-established observation backed by numerous published (and cited!) geological and seismotectonic studies – and increasingly also by geodetic ones. The same is true for the abundance and size of normal fault scarps in the mentioned regions. The current view that normal fault scarps are non-existent in Montenegro is now obsolete due to the discovery of the presented fault scarps. Although we are not conscious of any mistake, we will readily rephrase the sentence, as it apparently might have deemed you provocative, which was certainly not our intention.

14) line 87: what to you mean ? 36Cl dating is not affected by vegetation. Maybe you mean for 36Cl sampling ?

→ In fact, contrary to your statement, ³⁶Cl dating is possibly affected by vegetation according to Dunai et al., 2014 (Quat.Geochron.). But you are right that this is not really what we wanted to refer to, so we will remove the brackets and their content.

15) line 91: where does the date 18 ± 3 kyr come from ? please cite papers or discuss this date.

→ See chapter 5.1 where we discuss the origin in the discussion chapter. We will add the Papanikolaou paper as a reference here as well to prevent confusion. See also response #5 and response #4 to Prof. Roberts’ comments.

16) line 131-133: what do you mean ? not clear to me.

→ We will rephrase the sentence.

17) line 176-178: the radial pattern suggest landslide feature, why not discussing it ? could it be realated to bedding slip ?

→ Gravitational collapse is discussed (LL 258 ff.). Also, the bedding is specified (L 157). We will elaborate more on this point and emphasize our arguments against a landslide feature.

18) line 184: five horizons are very speculative, please discuss what could be their origin besides seismic slip.

→ In chapter 5.3 we discuss in detail why we exclude other formation mechanisms. Figure S7 shows a remarkably constant width of the (up to 5) ribbons across 48 locations (see also LL 183 ff.). See also response #6.

19) line 257: really not convincing, how is the bedding ? if it is perpendicular to the fault plane it is more convincing, please discuss that.

→ Again, the bedding IS specified (L. 157) and illustrated (Fig. 4).

20) line 265-268: what do you mean ? it is not clear whether you suggest those faults are an effect of the contractionnal regime and it appears in contradiction with what you said in the introduction.

→ See response #1. We are constantly discussing 2 different approaches, starting in the Abstract. Still, we see the necessity to make this clearer. We will be happy to implement this in a final version.

Dear editors,

dear reviewers,

we would like to thank the Editorial Board and editor F. Rossetti for guiding us through the complex and sometimes astonishing process of Solid Earth. We very much appreciate your comments and try to answer the raised issues point by point.

1) as observed in many cases, active thrusting can give origin to secondary features associated to it. Above all, bending moment faulting at the hanging wall of thrust faults is a typical feature secondarily connected to compressive deformation. In this term, even if the authors do not deal specifically with current activity of the thrust fault onto which the normal faults are supposed to grow, the authors themselves state that the supposed active and seismogenic normal faults under investigation occur along the coastal area, where active compressive deformation occur, and not in the hinterland, where extensional tectonics is ongoing (lines 63-67). In many cases in the Alpine chain, active thrusting give origin to dip-slip fault scarps, even some km long, on top and at the front of growing anticlines that resemble normal faulting, but which are secondary, passive, non-seismogenic features being extrados structures and large-scale gravitational sliding owing to forelimb collapse. Examples of this have been observed at different places in the central and eastern Alps, such as those investigated by Galadini et al. (2001) and Zanferrari et al. (2008) along the Mt. Baldo and the Mt. Jouf active thrust faults, respectively. On this topic see also Lettis et al. (1999).

→ Admittedly, the underlying thrust faults (blind and offshore, but well documented by seismicity) on top of which the normal faults are supposed to form and grow) are not extensively discussed in our manuscript. However, this is effectively not the focus of the present paper. From our point of view, we provide all relevant information on the regional geology, including the Dinaric thrusts. For additional detail, we refer to literature detailedly elaborating on local thrusting activity. This also includes a paper from our own group, dealing with the immediate vicinity of the fault scarps (Schmitz et al., 2020). By answering particular comments of reviewer Bendetti (e.g., her comments 1,4,8,11,17) and implementing many of her improvement suggestions in our revised manuscript, we were already able to more clearly exclude non-seismic formation mechanisms. After the third review, we will be happy to add even more arguments, particularly against a landsliding-related genesis (e.g., no internal deformation of the slide mass, no toe deformation, microseismicity, drainage networks, lithology – as we are dealing with limestone, not e.g., flysch). This combined, and with all due respect, we do not see how the frameworks presented in Galadini et al., Zanferrari et al., del Rio et al. or Lettis et al. should be comparable to our setting in Montenegro altogether. We argue that up to the present day, there is no scientifically solid argumentation available on how (deep-seated) landslides could be able to create such large-scale surface ruptures/fault planes (see above comments as well). The referenced papers cited themselves admit that no distinction of deep-seated landslides from seismically controlled surface ruptures is possible (del Rio et al.). In our paper, we present a stressable line of argument on why we assume a co-seismic origin of the presented structures (i.e., ribbons, 10 m high (!) free faces with striations, geomorphic landforms). The use and interpretation of these features is validated by numerous well-acknowledged studies that are (in contrast to Galadini et al., Zanferrari et al., del Rio et al., Kastelic et al. and Lettis et al.) very well comparable to our study sites in terms of the overall setting (including bedrock, fault scarp and slope morphology etc.). The according references are clearly cited in our manuscript. Crete is one of the best examples studied by our own group (Mason et al. 2016, 2017). We are aware of the work of Kastelic et al. (2016) in the Appenines, where we had the opportunity to visit all earthquake sites of 2009, 2016 and the surface ruptures. Those were generated by the earthquakes and are seismogenic. And of course, there is some compaction/erosion going on on the hanging walls of normal faults. However, the authors fail to present convincing evidence that the bedrock-colluvium interface has not been modified by other processes: animals, humans compacting by walking at the borderline? Wash-off by torrential rains at the hardness contrast of both materials? Plant growth and bioturbation? Remember: these faults move ca. all 500-1000 years. A modification is very likely. However, many observations show that the scarps are generated by earthquakes. A note on “deep-seated” landslides, as this is also discussed: Why do those move only in earthquakes? And not seasonally? Of course the “free faces” are modified in their many 500-1000 y long life, but according to the throw during an earthquake we do not believe that this is the case. The slip history of normal faults during earthquakes has been discussed by many (Benedetti, Roberts, our working group) by cosmogenic nuclide dating of the scarps. So, yes, we acknowledge the precautious note by the reviewer and the Kastelic et al. (2016) publication; but in our case, the experience teaches something different (Reicherter et al. 2003, 2010, Reicherter and Peters 2005; Mason et al. 2016, Mason and Reicherter 2016; Schneiderwind et al. 2016, Mechernich et al. 2018; Wiatr et al. 2013, 2014.) and of course all related publication especially of Rev 1 and 2; and many others.

2) The fact that the fault plane exposure is only due to tectonic movements and not to other non-tectonic phenomena is a critical aspect. The authors claim that fault exposure is not associated to landsliding because no indication of it is found in the sampling sites. Nonetheless, they do not provide any evidence of this assumption, such as detailed geomorphological maps of each sampling sites or pictures demonstrating long term (tens of thousands of years) slope stability. Moreover, at least sampling sites b, c and d in Figure 3 seam to coincide to visible stream incisions, testified by the white stripes (likely scree) evident in the provided picture. This appears even more evident in Figure S1, where sampling sites coincide or with stream incisions (and fault plane exposure can be simply the product of erosional exhumation) or with sectors of the slopes characterised by high topographic gradient, where gravitational component of the fault plane exposure cannot be ruled out and thus quantified. In this perspective, triangular facets and wine-glass-shaped valley are not tout court evidence of normal fault activity (lines 160-161), as stated by the authors. Indeed, formation of these supposed morphotectonic features can be due to differential erosion across the fault scarp. The authors do not demonstrate the lack of this process before claiming tectonic-related exposure.

→ We regard the comment as not appropriate here, as the shown striations cannot form in bed rock during landslides. Also, the reviewer fails to give examples of rejuvenated landslides and landforms associated. We refer strongly to the paper by Dramis and Blumetti (2005) who elaborated the “seismic landscape” concept: Dramis, F., Blumetti, A.M., 2005. Some considerations concerning seismic geomorphology and paleoseismology. Tectonophysics 408, 177-191. https://doi.org/10.1016/j.tecto.2005.05.032. And keeping erosion rates of carbonates in mind: of course triangular facets, wine-glass shaped valley are seismogenic landscape features. We have the feeling to be trapped in a non-conventional argumentation that lack geological understanding. However, from a distance we observed the discussion of tectonic vs landslide hypothesis on normal fault scarps. Judging from the comment, we would urgently recommend another detailed look at the manuscript. First of all, Figures 3 b,c and d do NOT show our sampling sites (there is also no such indication in the text or figure captions!) and Figure S1 shows that our sampling sites do certainly NOT coincide with stream incisions. This is clearly visible although the resolution at the presented scale is not even great. In the text, we clearly admit that there are large washed-out domains with stream incisions (and therefore differential non-tectonic erosional processes across the fault scarps, e.g., lines 162 f). We also clearly state that our sampling sites were selected after criteria strictly avoiding such domains (lines 85 f). To us, an incorporation of geomorphological maps for each sampling site seems somewhat excessive: As detailed descriptions and coordinates are provided, the reader can easily locate the sites on any desired additional map or e.g. Google Earth, if interested. In response to comments 6, 17 and 18 by reviewer Benedetti, we indeed do elaborate more detailedly on possible non-tectonic/seismic formation mechanisms of the fault scarps in our revised manuscript which will hopefully reduce your scepticism. We agree that this aspect possibly came up short in the first manuscript version.

3) the assumption that supposed active fault scarp exposition has a post-LGM age, since supposedly during the LGM any slope would have been uniformly regularized by erosion/deposition, is anachronistic. Indeed, erosional/depositional dynamics along mountain slopes, even during a glacial period, is a function of the global but also of the local (regional) climatic and geomorphic setting: erosional/depositional dynamics along slopes are influenced by latitude, altitude, direction of slope facing, proximity with sea/ocean, proximity with glaciers, even during global climatic forcing. This implies that the climatic morphogenic effects vary from a region to another, from a slope to another, even close to each other. Thus, assuming that the fault exposure has a post-LGM age (post 18ka) is too simplistic and, let me say, no more acceptable, because conditions that can have influenced morphogenic processes at regional and local scale do not allow to consider the assumption as reliable and robust. The above indicates that the evaluation of the fault vertical throw rate by simply performing even detailed morphological profiles across the fault scarp is based on a critical chronological assumption. Moreover, the authors do not correlate across the faults the same correlative features (such as the same deposits or landforms displaced across the fault), but they only consider local topographic offset. This is a very risky way to proceed since, for instance, the footwall may be affected by erosion, whereas deposits may accumulate at the fault hanging wall, at the base of the scrap, thus resulting in different origins and ages of the current topographic profile across the fault. This influences slip and slip rate estimates. Moreover, the total throw estimated at line 167 (200 m) is proposed only for one of the faults examined (KFS) and not for the other strands (BFSn and BFSs), and also along just one site.

→ We do not agree with the first general statement as it is not what many scientists all over the world – especially in the Mediterranean – have measured and observed. Of course there are local variations in terms of elevation, exposition etc. This is why we applied an error. Thus, we agree that slight local and temporal variations in terms of climate and erosion may likely exist, also during glacial periods. We therefore also fully agree that the applied method and assumption is indeed simplistic and partly generalized. However, we are fully convinced that the OVERALL integrated processes and erosional behaviour during glacial/non-glacial periods DO follow the pattern described in our manuscript. We believe that this attitude towards your concerns is justified, as our presented hypotheses, framework and methods are well-founded and have previously been proven and applied by many renowned (cited) authors. In that sense, your comment and concerns would possibly be more convincing if resilient scientific proof was available. In section 5.1, we discuss the possible weaknesses of our method. Here, we are ready to add a passage on possible small-scale local variations. According to comment 5 by reviewer Benedetti, we introduce the slip rate calculations based on scarp profiling as an auxiliary tool in our revised manuscript version. The fact that the method is thereby faded into the background to serve as a benchmark for 36 Cl dating certainly justifies its preservation in the manuscript. The reason why offset estimation is restricted to one profile across KFS, is simply a lack of other suitable markers. 

4) the supposed common and ubiquitous earthquake free-face exposures (drawing of most of the dashed lines in figure S6) appear very speculative in many of the showed cases. Most of them appear faint or not objectively distinguishable at all. Moreover, very critical appears lateral extent of the supposed earthquake ribbons, being up to few tens of cm long in many cases. Hence, tectonic origin is very hard to believe.

→ This comment partly corresponds to comment #6 by reviewer Benedetti. In this case, we apologize for the quality of the images. However, this is something we cannot really change, unfortunately. Many of the photos were taken in heavily forested terrain, which has an effect on both, the state of the ribbons as well as the lighting. Moreover, we are dealing with relatively narrow ribbons that are often not perfectly preserved or even defaced (probably different from what you know from Italy/Greece). We openly address and discuss this issue in the manuscript (e.g., lines 240f). To us, it is not the lateral extent and perfect discriminability per location that makes the ribbons reliable proof and markers of earthquake activity, but the constant widths of up to 5 ribbons that are traceable across 48 locations on the fault scarps. This coherence is illustrated in Figure S7 A-C, third column.

5) The Wells and Coppersmith (1994) regression allow to estimate maximum expected magnitude from fault geometric and slip parameters, only if a given fault is supposed to be a primary earthquake fault. Secondary features are not accounted in the regressions as parameters can scale differently with magnitude. In this perspective, authors do not prove that the faults the investigate are primary faults or secondary structures associated to a primary seismogenic thrust fault (see my comment at point 1). Therefore, any inference about seismic potential associated to the investigated faults must be taken and dealt with great caution at least, because the genesis of the extensional structures is not fully demonstrated, given the compressive active tectonics of the region. If the investigated extensional structures are secondary features, they only activate when the primary thrust fault activates. They do not release earthquakes by themselves but they only accommodate passively part of the overall deformation.

→ First of all, we do not fully commit ourselves to one or the other formation mechanism (primary or secondary fault/movement). The view that the faults MAY be primary features therefore fundamentally justifies an application of the Wells & Coppersmith regression, even if we follow your argumentation. Let us explain it like this: the Wells & Coppersmith publication includes a data set of c. 80 earthquake ruptures, many of those are NOT normal, the publication is 25 y old, and hence should be regarded respectfully and not as “the bible”. More importantly, we detailedly discuss the weaknesses of the technique and come to the conclusion that it is probably not too reliable for our setting anyway (short rupture length, low magnitudes; lines 242 f). Still, we regard it as a tool to at least roughly estimate the magnitudes that our structures could be related with. We are ready to further highlight this in a revised manuscript version. The Montenegro 1979 eq was a thrust event of Mw c. 7; our structures are well below this and prone to host a Mw c. 6. Once again, we like to refer to Crete, where Mw 8 (or higher) uplifted the western part, whereas onshore normal faults are much shorter but also seismogenic, more or less in the range of M 6 ± 0.5. Identically we may refer to Japan, 2011 events, and Chile 2010 and many others.

6) the sole presence of a cataclastic bend along a fault zone, not characterized in terms of microstructures, is not indicative if taken by itself of seismic slip. In this term, I would suggest to consider the work of Del Rio et al. (2021), in order to evaluate the possible origin as large-scale gravitational features of the investigates structures, as secondary structures associated to primary seismogenic thrust faults.

→ Apposite to our statements at the outset, we are not willing to promote poorly underpinned theories as a base for our work and argumentation. We absolutely respect the suggested paper by Del Rio et al. (2021) but are by no means convinced of the DGSD theory and related arguments, particularly if applied to our sites. Instead, we prefer to rely on a multitude of well-established and acknowledged comparable studies (all cited in our manuscript). Microstructural analyses would possibly add to the quality of our paper. However, this cannot really be expected as a necessity in an already multifaceted tectonically/structurally-focussed work. Detailed descriptions (particularly in chapter 4) combined with a thourough discussion (chapter 5) sufficiently justify our interpretation of the normal fault scarps as active seismogenic features. A first-time description of the normal fault scarps is combined with an introduction of different possible formation mechanisms. This makes the style of our manuscript rather open and defensive - and does not claim “the one and only” irrefutable explanation for itself.