Thank You for Your Consideration

28 Jan 2015 Dear Dr. Lutz,

Thanks you for the reviews of our manuscript. We have revised it carefully based on helpful suggestions of reviewers. Responses to each suggestion are provided below. Please let me know if there are any further questions or concerns.

Thank you for your consideration.

Sincerely, Haibin Zhang,

Response to Reviewer #1

Over all this is an interesting study and presentation. The writing is generally clear and to the point. I do have some disagreements with the specifics of the interpretation, however, these should be easily fixed and do not necessarily alter the overall conclusions.

There is [sic] no real data supporting that these two regions are “hybridizing”. The authors use the term “intergradation” which could describe the situation here, but if the authors mean to infer that there is a blending of the genotypes in the SERP, this is not supported by their data. Most of the individuals in the SERP appear to be nearly 100% genetically like the parental individuals. Thus, the emphasis on deep- sea hybrid zones is over done and not consistent with the results. The BayesAss test we employed (Fig. 4) identifies individuals as first- or second- generation (non-parental) immigrants. We added wording to the text to clarify this matter.

The authors also refer to this as a latitudinal cline when it is more accurately described as a step cline. If this is an indication of population subdivision, then they have the difficulty of the phylogenetic relationships confounding the testing for isolation by distance. This can be dealt with by doing a partial mantel test thus controlling for distance or phylogenetic relatedness. We changed the wording to "step-cline" in the text. We added a stratified Mantel test to examine the contribution of structure due to ridge section (NEPR, SEPR, PAR) to the correlation with geographical geographical distance.

The revised the text and Figure 5 from the BayesAss analysis identifies admixed individuals along the SEPR and supports our interpretation of intergradation.

Line 101: IBD can also be caused by inadequate or sparse sampling in a structured population (see the Miriams 2012 paper you cite). We agree and have added this possibility to the text.

Line 104: Selective sweeps in mtDNA is a highly contentious issue and there have been some good arguments for indications that the view that selective sweeps are common is just an artifact of the analysis. I would not be so casual and certain about the prevalence of selective sweeps in mtDNA. See Mulligan et al. 2006 Science 314:1390, Wares et al. 2006 Science 314:1388, and Karl et al. 2012 Mol Ecol 21:4171. I also find this statement somewhat lacking. It is unclear to me how high mutation rates or maternal inheritance per se could result in a signal of IBD. We deleted the ambiguous sentence.

Line 109: This sentence is somewhat confusing. Is this a result of the current study or a previous study? If the current study, than it does not belong in the introduction. If a previous study, than it needs a citation. I find the last sentence also lacking and needs to be more specific and detailed or deleted. We clarified this point in the revised sentence and deleted the last 2 sentences of the introduction.

Line 133: This sentence is awkward. Replace “for” with “a” and rewrite so that the meaning of what was “examined for this study” is clearer. We have rewritten this sentence.

Line 140: This manuscript is somewhat table heavy. I suggest that you delete Table 2 and include the references in the text of the materials and methods. We retain Table 2 and include additional information requested below by this reviewer (line 154). We deleted Table 4.

Line 146-154: I suggest that you delete what didn’t work and rewrite this section along the line of: “To identify… Jarman et al., 2002). Inconsistent amplification…” (i.e., delete the list of the genes). We deleted references to the genes that did not work.

Line 154: You state here that “Nested PCR methods were used…” but do not detail what this entails. I presume that additional primers were used/developed and this should be detailed here. Alternately, I do not understand what you mean by “nested”. This is clarified in the text and relevant information is included in Table 2.

Line 156: Include city and state for ABI and it looks like you need a space between “at” and “10umol” Done

Line 157: “Condition” should be plural since there are several of them that follow. Done

Line 182: Delete “used” after the citation. Done

Line 185: The meaning of “a1/-, b2/-…” is unclear. You also need to provide more details on how you set the various parameters in STRUCTURE. I am also uncomfortable with the analysis of mtDNA and nDNA together. Given that one of the goals indicated in the Introduction was to look at potential differences between the mtDNA and nDNA, concatenating the data is counter productive. Also, STURCTURE works to minimize Hardy-Weinberg and linkage disequilibrium and mtDNA is not an appropriate marker to analyze in this way. Minimally, you need to justify why you concatenated the data sets and to provide some indication that this was an appropriate thing to do and to eliminate the possibility of cytonuclear disequilibrium before analysis. The paragraph gas been rewritten. We also include a BAYESASS ANALYSIS, as it does not assume Hardy-Weinberg or linkage equilibrium, which is more appropriate in the present case. Line 198: What criteria did you used to determine that the analysis had converged? You need to be more specific here on what about the output of TRACER indicated to you that the analysis had converged (e.g., large ESS.). We clarified this in the text: Convergence was determined by plotting the likelihood and prior values, and examining acceptance and mixing rates with Tracer v1.5

Line 121: HSP is figure “2C” not “2D”. Fixed

Line 230: It is not clear in the table which Fis values are significant and which are not. Consider bolding the significant values or highlight them somehow. Most of these values are very large. Assuming that most of them are also significantly different from 0.0, it would indicate that you either have non- random mating (something that you are arguing against) or significant “technical problems”. You need to address why these problems do not prohibit you from trusting the results. The linkage equilibrium between HSP and ATPsa is a single test and not a particularly strong test for random mating. We feel that the appropriate caveats are addressed. The significant values have been bolded. We hypothesized that allelic dropouts might be the cause. Allelic dropouts could be caused by a number of reasons (e.g. long-term tissue storage…), which is addressed in Lines 240-244.

Line 241: I assumed that you are going to deal with the discrepancy between ATPsa showing LD and thus admixture and HSP not showing LD and thus no admixture in the discussion, but you did not. In multi-locus studies, there often are discrepancies among the loci. These need to be addressed. We addressed this matter with the following sentence in Lines 309-311: “The significant cytonuclear disequilibrium (CND) observed between ATPsα and mtDNA genes at SEPR also indicated the admixture pattern in this region, although no significant was found at HSP.”

Line 266: I disagree with the interpretation of the STRUCTURE output. If the individuals in SEPR are admixed (i.e., each individual is a mixture of NEPR and PAR genotypes; hybrids) then why are the q values so skewed? Most of the individuals are nearly pure PAR or NEPR and thus if they are hybrids they would be the result of later generation backcrosses to the parental population. This runs contrary to your argument that the SEPR area is a sink and that there is little to no migration out of the SEPR region into the parental regions. I also would not characterize the SEPR individuals as being “thoroughly mixed” (line 269). It appears that they are either nearly 100% PAR or NEPR and very few show intermediate assignments. We have revised the sentence, now it is “Individuals from the SEPR region (S7- S17) were assigned NEPR, PAR, or as mixed genotypes.” We also add a paragraph to clarify (Line 273-285). We also added a graphic for the bayesass analysis which also shows the relative admixture.

Line 279-282: Given that there are significant Fis values for all locations, it is difficult to posit that they are due to admixture in the SEPR area but not in the PAR or NEPR. Yes, the values are lower, but are their different causes of the significance in the different areas? If so, what are the mechanisms and where is the data that support this argument? It is somewhat untenable to use data that should be in HWE in these analyses and just ignore that potential for technical problems making the data unanalyzable. We agree, in part, with this criticism and soften our interpretation in the paragraph covering Lines 288-290. Line 282: Having the SEPR area showing significant Fis values and the PAR and NEPR not would support your argument for Wahlund effect in the SEPR region. Given that there are significant Fis values in the presumed non-admixed areas, you need to argue that there are either different reasons in the various regions or that there isn’t anything particularly special about the SEPR region. Under normal circumstance, allelic dropouts are expected to be similar across samples, however; in this case the samples were collected over a 15-year period and clearly have been subjected to different handling environments. In this case, an argument could be made that there are technical reasons to have differences in allelic dropout that are associated with sample age. Differences in sample storage are not a likely reason for allelic dropouts. We note this in the revised paragraph covering Lines 292-296.

Line 290: See previous comment about the lack of intermediate assignments values in the SEPR area indicating that these individuals are not actually admixed individuals but likely pure parentals. We added Fig. 5B with results from the BAYESASS analysis, which indicated there are several admixed individuals.

Line 293: If you are going to differentiate between IBD and population structure, you need to do a partial mantel test to control for distance or phylogeny. See the Miriams 2012 paper you cite. We agree and did a partial mantel test.

Line 296: Given that these fields are discrete with uninhabitable habitat separating them, it is difficult to see how this is anything but stepping stone dispersal. I am not clear what point you are trying to make here and I might argue with the assertion that SSD could not produce the same cline across multiple, independent loci. We agree and modified the statement.

Line 330: See previous comment about lack of admixture of the individuals in the SERP region. It is also not clear to me what data support a re-invasion. How do you know that the species was there, extirpated and then re-invaded? There is no data presented to support this assertion. We added more evidence there was admixture in SEPR (Lines 276-285). We hypothesize it as secondarily based on the admixture pattern of SEPR, the instable reality of vent environment, and the previous studies of other co- distributed vent species.

Response to Reviewer #2

-The term intergradation is used throughout the manuscript, but itís not defined. The authors use it instead of hybridization or admixture. Is there a formal definition? We defined “intergradation”

-The HSP pattern is very different than the other genes. The product is 173bp and 5 instances of recombination are detected, yet none in the other nuclear gene. Is this common? Intragenic recombination is put forth as the explanation for the messy haplotype network. However, HSP is a family of proteins. Could it be possible that the primers amplify paralogs? The gene product is pretty small and double bands may not be visible on a gel. The methods donít mention checking the PCR products for clean amplification. Were the exonic products translated to make sure there were no stop codons? This seems as plausible as detecting 5 recombination instances in a 173bp stretch. Would it explain any of the other results? At least provide more of an explanation of the contrasting patterns and the justification for including it in the analyses. Four of the five putative recombination events were located in the intron region. We have added this clarification in Line 235-237.

Line 51: need citation We have added to citations: O'Mullan, 2001; Johnson, 2013.

Line 90: no “and” Removed.

Line 151: no “to” Removed.

Line 210: need Genbank accession numbers Added

Line 226, 228: Does the figure need to be referenced twice so close together? Also, the actual figure is rather obtuse/overwhelming/vertically stretched. Necessary? The figure was revised, and only referenced once now.

Line 241: “A” should be “As” Corrected.

Line 258: N=8 pretty small. And the population N11 sample size marginal across all genes. Was this taken into account? Would some of the pairwise Fst values be skewed by the low population size. We agree that N=8 is a small, so we applied rarefaction estimation of some diversity statistics. We removed pairwise Fst comparison (original Suppl Table 1)

Line 298: Unclear. SSD would not result in many independent genes showing the same pattern? Why not? A demographic process would be expected to act on the whole genome. What is meant by cline? We agree and removed the sentence.

Line 302-305: Bold statement. There are tests for selection. If they can’t be applied please explain why. We included at test for selection on each locus with the program Lositan.

Conclusions: There is no mention of Tevnia. There is evidence for admixture and immigration from 2 populations into 1, but marginal evidence for cyto-nuclear disequibrium, perhaps because it is a single species and the time of separation is more recent. Would make conclusions more specific, especially after the focus on Bathymodiolus in the discussion. The Conclusion part has been rewritten.

Figure 1: A picture of Tevnia would be nice We Included an image of Tevnia as figure 1.

Figure 3: The *hap-1 in each of the names in the legend not necessary “*hap-1” has been removed. Responses to Reviewer #3

line 134: erase “ ..for” The sentence has been rewritten.

line 154: substitute “…amply” for amplify Corrected

line 241: substitute “ A…” for As Corrected

line 268: S31 appears twice One “S31” has been changed to “S32”

line 346: substitute “…recent that” for recent than Done