Review of "Tectonic interactions during rift linkage: Insights from analog and numerical experiments", by Timothy Chris Schmid, Sascha Brune, Anne Glerum, and Guido Schreurs. 

This manuscript investigates the causes of faults deflection during early rift segments propagation. Numerous observations of curved fault systems are reported from early continental rift settings, yet the cause(s) of such deflections remain to be understood. Here the authors use crustal-scale analog and numerical models to investigate rift propagation and strain localisation in early rifting stages when isolated continental rift segments interact. The comparison between nature, analog and numerical models is elegant. Thanks to the high-resolution numerical modelling results the authors demonstrate the importance of transient stress rotations at the surface of early rift systems for controlling propagation of rift arms. Although this work is original and focused on basin scale tectonic evolution of propagating continental rift basins it confirms earlier findings proposed from larger scale numerical models (see point 1 below).

The paper briefly reviews published work on rift basin propagation and linkage, then introduces a set of analog and numerical models’ setups designed to investigate rift propagation dynamics in early rifting when separated rift branches interact. The different models explore a wide range of rift arms geometries (imposed using a weak seed heterogeneity at the base of the brittle upper crust) and test how these branches propagate in a sub-pristine environment before joining. Modelling results are presented in great details and highlight transient rift segments deflection prior to propagation. The numerical models analysis explains this behaviour with surface stress rotations and conclude that faults bounding rift segments do not necessary align with the regional stress field. This contribution seems well suited for EGU Solid Earth and will be of interest to the tectonics community in general. Overall, the manuscript is original, very well written, well organised, and beautifully illustrated. I would recommend accepting this manuscript after minor to very moderate revisions.

I present below some key points (mostly related to the numerical models and conclusions) for which I have some concern followed by a list of minor comments and suggestions.

1) My first comment relates to the main conclusion of the manuscript that stress re-orientations occur and change over time and with progressive deformation and the call for caution about paleo-stress measurements @ lines 131-134 & lines 735-739. I couldn't agree more! Yet, I believe that these statements are in essence what we wrote in Duclaux et al. (2020): "Our models, however, show that progressive deformation during Phase 1 extension results in rotation of the extensional shear zones to become orthogonal to the plate motion direction and control the structural style during oblique rifting. Although the stress around the active extensional shear zones rotates (Fig. 3), the progressive rotation of Phase 1 extensional shear zones during widening (Fig. 5) forces a discrepancy between σ2 direction and the strike of the structures that must be accommodated by a minor component of strike slip. Early rift structures are thus critical in controlling the final architecture of oblique-rifted margins, but because of potential rotations they must be used with caution when interpreting the tectonic evolution of passive margins." I hope I'm not biased but I believe a reference to our work here (as the paper is already cited in the MS) would be legitimate, as well as a reference to Gapais et al. (2000) paper. I do understand this original work focus on a smaller scale than ours, but the findings seem to match rather closely.

(Full ref: Gapais, D., Cobbold, P.R., Bourgeois, O., Rouby, D., de Urreiztieta, M., (2000). Tectonic significance of fault-slip data. J. Struct. Geol. 22, 881–888.)

2) Frictional softening is of primary importance to control fault localisation and propagation in the numerical models and I think a few more words should be added about it in the numerical model setup section. Lines 266-268: I understand that grid resolution varies vertically in the top part of the upper crust of the numerical models (Fig 4).  Is there a normalisation procedure in place for the softening function to account for weakening with different gris sizes (like in Lavier et al., 2000)? If not cells just below the surface will weaken faster than those deeper. That might have rather negligeable effect on the results, but it should be presented/discussed at least briefly. 

(Full ref: Lavier, L. L., Buck, W. R., & Poliakov, A. N. (2000). Factors controlling normal fault offset in an ideal brittle layer. Journal of Geophysical Research: Solid Earth, 105(B10), 23431-23442.)

3) Line 328: "the fault segments deflect and turn away from each other": Don't they just tend to form at this angle to strike orthogonal to the extension direction rather than “away from each other” as stated for the analog results (line 192). This brings me to the next point which seems worth discussing further in your work.

4) Line 504: "dip slip faults are favored over oblique-slip faults with a strike-slip component" - According to Brune et al. (2012) analytical and numerical modelling work oblique extension should be favoured. I believe this finding should be discussed in more details as it seems to contradict previous work on the subject. Is this because of the rheologies, the boundary conditions? I find this very interesting.

5) I find Figure 8 very informative and pretty well designed. It allows visualising stress deflection at the surface of the models and the surface stress regime at once. There must be an interpolation method used for the stress vectors representation as not all stress markers (one per cell cell) are depicted. Could you comment on this and how does it smooth the signal out? More importantly, I have some trouble with the location marked with "rotation jump" in Fig 8i. It seems that some of the stress markers are not resolved (non-defined in the caption), so I assume SHmax could be as depicted or be orthogonal?? How can a jump be argued in this context? I'm not arguing it doesn't take place, only that the marked region chose to highlight it is not the most suitable one... It seems to me that the overall "rotation jump" is related to the transition from compressional to extensional regions, while the gradual rotation relates to region with a transition from strike-slip to extensional. Is that correct? 

6) The comparison of rift arms propagation, symmetry, and timing between the different model geometries in the discussion would benefit some additional words related to the consequences of differential frictional softening rates resulting from the different seeds geometries. Rather than comparing time between models, you could maybe compare the amount of extension accommodated at one seed tip?

   + Lines 557-563: Because strain is distributed in the 2 arms in the early stage of the v- and y-models, this difference with the i-model is to be expected. Indeed, the models don’t have comparable strain rates, and frictional softening isn’t as effective in the y- and v- models. 

   + Lines 575-576: In the i-model case frictional strain softening rate is more effective too.

Minor comments:

+ Figure 1:  a location map for c and d in the context of the EARS would be a nice addition.

+ line 113: for the multiphase extension I would recommend adding a citation to Duffy et al. (2015) whose work seems relevant in this context. (Full ref: Duffy, O.B., Bell, R.E., Jackson, C.A.-L., Whipp, P.S. & Gawthorpe, R.L. 2015. Fault growth and interactions in a multiphase rift fault network: Horda Platform, Norwegian North Sea. Journal of Structural Geology, 80, 99–119. DOI: 10.1016/j.jsg.2015.08.015)

+ line 213: "[...] propagated minimalLY [...]" - missing the LY

+ Figure 5: Just a question: did you try a model without the random seeds to check whether surface ruptures remain symmetrical? I understand the random noise distribution will promote dissymmetry; this is out of curiosity.

+ Figure 6: While I can understand the "Curved faulting" contour line for i- and y- geometries, I struggle with v- geometries... there are plenty of faults outside the curved faulting region… and the faults within the regions do not seems to be very curved either.

+ Figure 9: In the models, main border faults are facing each other’s creating a strong asymmetry of the segments at time of propagation/linkage. On the other hand, in the natural example LT is marked as a hemi-graben with east dipping western border fault, but SV is super narrow graben and doesn’t display apparent asymmetry. Can you please comment on this significant difference?

+ Line 131, 581: I would recommend using the term "heterogeneity" or "structure" rather than "fabrics" throughout the manuscript, but this is just semantic, and I will let the authors decide whether this is the correct terminology. To me a "fabric" relates to a preferred orientation or configuration of all the elements that make up a rock. In the context of this study there is no initial fabric in this sense, but a pre-existing weak structure at the base of the brittle upper crust. A "fabric" would relate to the initial noise distribution within the upper crust region.

+ Line 621: I'm a little confused... how can a "discrete zone" be "broad"? Maybe the broad zone could be described as "distributed"?

Nice, 20/12/2022

Guillaume Duclaux

The paper by Schmid et al. presents analogue and numerical modelling of the rift segments, to investigate their propagation and interaction, and to understand how stress patterns are affected during propagation and interaction. The paper is well written and illustrated and will be a very good and useful addition to the literature on this subjecct. I have some minor comments, and one significant issue with the manuscript that are discussed in the attached word document. The significant issue is the way the deformation in Turkana is described - and in the attached document is describe the way I see the development of Turkana. Perhaps some examples form largely non-magmatic rifts outside of the EAR would be useful, since the authors note that the modelling does not consider the effects of magma on rift segment interaction. 

Chris Morley