Referee 1's review:

The manuscript presents data showing that the transcripts of ZAM and Idefix are targets of post-transcriptional silencing mechanisms that depend on the presence of the COM locus. It also shows that sRNAs antisense from ZAM and Idefix transcripts can be detected when the COM locus is present. The manuscript also shows that piwi is necessary for ZAM and Idefix silencing in the germline. Moreover, the authors show that ZAM and Idefix are also silenced in adult flies mutant for piwi. The characterization of the COM locus, its role in the production of pi-RNAs and the control of mobile elements has recently been reported. Thus, part of the results presented in this manuscript cannot be considered as completely new. However, the manuscript includes a detailed analysis of the silencing of ZAM and Idefix and contains additional data, in particular the piwi-independent silencing in adult flies, which could deserve publication in PLOS One. Nevertheless, the way the authors present and discuss the data should be improved. I would like to comment on several points:

1. The way the authors present and discuss the finding that piwi-independent silencing occurs in somatic cells is puzzling. All along the manuscript they present as a surprising result the fact that silencing in somatic cells is piwi-independent. As far as I know, the expression of piwi is mainly restricted to the germ line and surrounding tissues and thus it is not surprising that it does not participate in the silencing of adult tissues. The unexpected of this result could be that germline silencing mediated by piwi is maintained in adult tissues. For this reason it would be better to address the somatic silencing after characterising the embryogenic silencing (including the sRNA and piwi analysis).

2. Figure 1 summarises data that are not presented. What does the +++ signs represents? Have the authors quantified the expression? How? It would be better to present the original data.

3. After a detailed description of the pGFP-ZU (or pGFP-IdU) constructs no data is shown related to that. I think that it would be better to have these data in the paper, but if the authors prefer not to show it, it would be better to describe first the work on pGFP-Zenv and say that similar results were obtained when using the 5' UTR but without giving details.

4. The authors compare the expression of constructs containing sense and antisense insertions of ZAM and Idefix (p.9, l.15) shown in different Figures. These figures should be combined to allow proper comparison. In addition, it would be better to compare the same constructs in sense and antisense orientations (as well as the recombined empty vectors). It is difficult to understand why the authors have decided not to show pGFP-IdU sense data (see point 1) and they do not show pGFP-Idgag antisense data.

5. Figure 5. The scheme presenting the constructs differs from that of the previous Figures. As the constructs seem to be the same it will help the reader to have the same schemes for all the constructs. It is not clear why the authors have chosen to present only a Northern for the analysis of ZAM and only the quantification for Idefix. It would be better to present both types of data for both constructs.

6. The discussion should be more focussed and concentrate in the new findings of the manuscript. Related to this, the authors show that there is a partial silencing of ZAM and Idefix sequences in U lines, which suggests that there is a COM-independent silencing of these elements, but they don't discuss this finding in the discussion section.

Minor points

Introduction. P.4, l.7. The authors present the piRNAs as if they had not been cloned in Drosophila, while different papers cited in the article describe these RNAs from Drosophila. In addition rasiRNAs have been found to be a subset of piRNAs...this is not clearly explained in the manuscript.

P.7, l.21 What does "a single transposon silencing pathway" means?

P8, l.7. The authors state that Idefix fragments were cloned in sense orientation as previously done with ZAM transcripts. But in p.6, l.19 they state that both sense and antisense orientations were used for ZAM. The antisense constructs for both elements are then explained again in p.9. In general, the way the authors explain these experiments and the constructs used is quite confusing and needs to be carefully revised.

p12. It is strange to have the conclusions of the sRNA analysis before presenting the data

The authors should choose a single way to refer to the Flamenco/COM/20A locus throughout the manuscript.

**********
N.B. These are the comments made by the referee when reviewing an earlier version of this paper. Prior to publication the manuscript has been revised in light of these comments and to address other editorial requirements.