|The manuscript “Paleomagnetic constraints on the timing and distribution of Cenozoic rotations in Central and Eastern Anatolia” presents an impressive amount of new paleomagnetic data for a very interesting and very complex area. I think their approach of obtaining as many paleomagnetic sites as possible represents the only way in dealing with such a complicated tectonic area and definitely deserves publication. It is well written, I think the geological introduction is sufficiently detailed, the structure might need improvement, especially in terms of a summary of the results, but I agree that the topic and results are quite numerous and complex.|
On the other hand, however, I think there still remains a problem in the data treatment. My only major concern is the already mentioned treatment of individual sample directions rather than site mean directions. I agree with the authors that always applying the standard paleomagnetic procedures is not always the best thing to do and new approaches are sometimes reasonable and it makes sense to explore those. However, I have the feeling that this area and the presented data might not be the perfect place to introduce a “break with paleomagnetic tradition”. In my opinion, using individual directions rather than site mean directions is reasonable when treating with inclination shallowing, or when studying different parts of a continuous section. In such a tectonic complex area, however, local rotations might vary a lot on a small scale, which would require site mean directions. Then, a site with a small number of samples and high alphas (or As) is sometimes very important and equally necessary than another site with high quality results and a large number of samples. To exaggerate, treating 2 separated sites, one with 100 directions, and the other with 10, even if they yield mean directions which are statistically different, would yield a mean direction equal to the first site. The authors talk a bit about that, but I think a much more rigorous introduction of this approach would be necessary to justify it. A way to get around this, would be the presentation of regional mean directions based on both site means and individual means. A small table 2, which shows that the two methods are comparable, but the error is smaller (?) would leave everyone with the choice what to choose. I would say it is necessary to present the standard method at first, and then the alternative. The missing propagation of errors is true, but on one hand, error propagation exists and, I think, this would be a better way to treat this problem. The presented alternative approach does not do propagation either. One thing, which is also absolutely necessary for the presented method, is a clear discussion about structural data. Which sites have which bedding? So combining individual directions is based on sites with similar bedding? This needs further discussion.
In this context, I would like an additional discussion about block rotations versus more continuous deformation? I am not so familiar with the region, but it seems that there is an orocline in the north and one in the south with continuously changing declination (figure 8), but a more blocky behavior in the center. This might well be true, but a small discussion would be nice. Also the white block boundary in figure 8 might be a bit speculative. Maybe the red stripes for coherently rotating domains should be restricted to the center?
In the end, I think the authors have to decide what they want to do. Either they present a statement in favor of their alternative statistical approach, which would require more comparisons, a more straight forward discussion for the two different approaches, and a more simple tectonic setup. Or they tackle the tectonic history of the presented area, which would require at least also to show the site mean directions and do it the classical way.
It is definitely an interesting discussion the authors present here. Also in my opinion the Deenen et al. paper is a major step forward. However, if site mean directions do not fulfil the requirements, using individual directions need to be further justified. I don’t think that increasing n and decreasing error is always better.
Two more minor points are first, I miss a discussion about inclination shallowing. Because the authors are using arguments based on inclination, some discussion including figures about shallowing would be good. The directions do not seem to show much elongation, but given the amount of results, a site by site inclination shallowing inspection might yield additional information about primary or secondary (if shallowing is present, remagnetization is less likely).
Also, it is not easy for the reader to assess the actual outcomes of the study. It would be nice to see some additional summarizing figures, like e.g. D versus time or a sketch showing blocks rotating? Most of the rotation arrows in the last figure are already in the first figure. Figure 2 and 3 show all the results (right?). So, what did change with the presented results? Is there a consistent rotation over time? What is the rotation rate?
In summary, as I said, I think this is a very interesting data set, which deserves publication after some small revision (most of my comments are rather suggestions and I think not much more modification is needed anymore). I hope my comments help to improve the manuscript.