Genetic Diversity and Population Structure Derived from Body Remains of the Endangered Flightless Longhorn Beetle Iberodorcadion fuliginator in Grassland Fragments in Central Europe

Round 1

Reviewer 1 Report

Rusterholtz et al. analyzed the genetic diversity and population structure of  Iberodorcadion fuliginator and tested six different hypotheses related to the phylogeography of the species. The manuscript is well-organized and well-written. It is interesting to the readers of the journal Diversity. They have used standard laboratory and analytical methods of population genetics. I feel the manuscript benefits from some amendments. 

The authors have emphasized the sampling method they employed in the title. Which may not be necessary. Rather, they do not mention the actual study area in the title. Grassland Fragments from where? The title needs to be complete. It is good practice that no destructive sampling was used in the study. But, it is not a must to highlight in the title itself. I suggest to rephrase the title.

The results of hypotheses 4, 5, and 6 are not emphasized in the abstract. It would be better to incorporate them too.

Hypothesis #5 about the odd and even years needs elaboration so that authors could understand them easily.

Figure captions are written above the figures. Please keep them below the figure. 

Line 209, and table 2: Sample size varied (12-63) in the study. It affects these measures of diversity. Did the authors check whether such effects were significant or not?

Figure 3: Please provide the regression equation also in it.

Figure 5: This NJ tree has too low bootstrap supports. This distance-based agglomerative clustering method is not considered the best for phylogenetic analysis. I suggest authors try ML or BI methods for the phylogenetic tree. 

Discussion: Results are repeated and figures and tables are cited frequently in the discussion. Please avoid redundancy.

Find the annotated PDF attached. 

All the best!

 

 

 

 

 

Comments for author File: Comments.pdf

Author Response

Please see the attachment..

Author Response File: Author Response.pdf

Reviewer 2 Report

The study aims to investigate the population genetic structure of the endangered flightless longhorn beetle Iberodorcadion fuliginator in grassland remnants in the border region of Switzerland, France and Germany. The study provides an interesting case study for the genetic effects of population decline and fragmentation because the species’ habitat has recently dramatically degraded the study area.

The study’s strengths include a non-invasive sampling approach, which is particularly important in threatened species so as not to increase the risk of local population extinction.

Limitations and weaknesses of the study include the small number of SSR loci investigated, which, in addition, did not turn out to be highly polymorphic. This does not diminish the value of the study, as the availability of markers for non-model organisms is usually not very high. In their case, any contribution to the knowledge of the species is valuable.

Another problem I see is the non-obvious way of analyzing the population structure, which was made twice on two carefully selected datasets. Lines 250-254 state that „(…) we considered the populations 2, 5 and 11 to belong to the metapopulation Blotzheim and the populations 16, 17, 19 and 20 to the metapopulation Istein/Huttingen (Figure 2, Table 1) and investigated the genetic population structure at the metapopulation scale (the isolated populations 29 and 30 in northeastern Switzerland were not considered in this analysis”. Next, another STRUCTURE analysis was performed on another dataset („Secondly, we examined the population structure at the regional scale (border region of Switzerland, France and Germany; 18 populations, data from the populations 20 and 21 were combined because of recorded dispersal; Table 1) using STRUCTURE as described above.”, lines 269-272). I don’t understand why the analysis was not performed once on the complete data. This looks like an intentional manipulation of the data in such a way as to produce the desired results.

 

Next, I cannot fully agree with the information in lines 204-207. The authors have checked the microsatellite results of each population for null alleles and misscoring using MICRO-CHECKER and have not found more significant problems with nulls or genotyping (except for one marker in one population). At the same time, there is a significant excess of homozygotes in eight of nine populations (Table 2). In these eight populations, Fis ranges from 0.223 to 0.638 (mean 0.406). Without null alleles, genotyping errors, or population substructure (Wahlund effect), such a high Fis must be due to inbreeding. Fis = 0.4 is much higher than would result from mating between siblings. It seems unlikely that the population would reach such a high level of inbreeding. It is more likely that the analysis in the MICRO-CHECKER did not correctly recognize the contribution of null alleles or genotyping failures. Since the authors included the source data, I reanalyzed it for null alleles and potential allelic dropouts using another software, INEst (https://doi.org/10.1093/jhered/esn088). Its unique feature considers the possibility of inbreeding within a population while estimating null allele frequencies. It is well known that both inbreeding and null alleles can cause an excess of homozygotes within a population. So the best strategy is to estimate both inbreeding and null allele frequencies simultaneously. That is why INEst may outperform other available packages (https://doi.org/10.1111/1755-0998.12015). For the most numerous population (16), the analysis in INEst estimates that two of six loci are loaded with null alleles. The proportion of null alleles equals  0.216 at  Dorful_001410  (95% CIs from 0.104 to 0.343) and 0.145 at Dorful_010423 (95% CIs  from 0.039 to 0.260). INEst estimates several types of models (n - null alleles, b - genotyping failure, nb - both null alleles and genotyping failure, and null - no null alleles and genotyping failure) and allows to compare them with the DIC criterion. In the case of the most numerous population no. 16, the n-model outcompeted other models. Thus, the high frequency of null alleles alone was sufficient to explain the considerable excess of homozygotes. Unfortunately, I have not similarly analyzed other populations due to lack of time. Still, it seems to me that null alleles significantly influence the calculated statistics characterizing genetic variability. This issue often represents a significant challenge in population genetic studies, as it may lead to the overestimation of population differentiation due to reduced gene diversity and also the overestimation of inbreeding. Needless to say, it may dramatically affect the entire study. For this reason, I am asking the authors to reconsider whether the allele frequencies at SSR loci could be biased due to the presence of null alleles or genotyping failure.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

The authors did a solid job responding to the criticisms from the previous round of reviews. I am in favour of accepting the manuscript in its present form.