Review of 'Species delimitation of the Dermacentor ticks based on phylogenetic clustering and niche modeling'

The paper aims to map the distribution of several species of ticks, based on several “environmental variables” that should draw these distributions. Other than molecular biology methods, that I think are adequately performed, this paper has serious flaws regarding the environmental modeling, that I would like to outline below.

1. The choice of actual records of ticks. It is obvious that the result of predictive mapping is as good as the actual distribution of the modeled organisms is captured. The authors did an effort in capturing such distribution from public repository data, such as GBIF. However, they did not any effort in the verification of the identity of the records. If these species of ticks are so widely confused, why the authors are blindly confident about the records they obtained?.  Further on this, it is widely accepted that the data on GBFIF and other public repositories are well curated, but they feed on the contributions of volunteers. Therefore, it is highly improbable that these records reflect the real distribution of these ticks. This is a very problematic issue in this paper, therefore biasing the modeling of the target species.
2. The choice of “environmental” variables. It is widely acknowledged that interpolated climate records cannot adequately map the distribution of an organism, even if they are widely used because they are easy to “download-and-integrate”. The data in WorldClim is our best estimation of the climate eon earth, beset are NOT valid for predictive modelling. There are some papers addressing this. I acknowledge I am the author of these papers. Please, if you can object our conclusions about the unreliability of WordlClim for environmental modelling, just publish and inform. WordlClim is unsuitable for predictive modeling because a serious of reasons. First, because the already accounted autocorrelation of the explanatory variables. Commonly, the authors of these studies try to remove the autocorrelation of variables by some statistical procedures that remove the variable highly correlated. My first point here is that, if autocorrelation has been already identified, why to repeat in every paper the same procedures to remove the variables that have been found already as autocorrelated? If the dataset is the same for all these papers (and it is), why it is necessary “to explore the autocorrelation of variables” if it has been already done?. Most important, how the authors know that they are not removing some variables that may have a real impact in the colonization by the tick? The blind removal of variables without any ecological criteria has been already pointed out as a very dangerous step in environmental modeling. According to this study (and many others on the topic) the authors believe that the same set of not autocorrelates variables can explain the distribution of a tick or of an armadillo. This is simply meaningless. Please if you can demonstrate that WorldClim reform better that satellite derive variables just demonstrate it. Science is discussion. We demonstrated that satellite derive variables perform much better than WorldClim dataset. I acknowledge the author may have a different post of view. Just demonstrate it. Further on this, it is well known that rainfall has no descriptive power for the distribution of ticks, but the relative humidity or the saturation deficit. However, interpolated climatologies do not include humidity estimates. I wonder how the authors expect to obtain accurate models using variables that unreliably describe the environment.
3. The real problem of interpolated variables for environmental modeling. Other than autocorrelation among variables, the real problem in this kind of studies is the spatial autocorrelation. In territories with a high density of climate recording stations, the interpolation will work relatively fine. However, in territories with a low density of these stations, interpolation is simply poor. This is not a criticism of the interpolated climatologies, it is a criticism about its use for environmental modeling. It has been repeatedly demonstrated that satellite imagery over-performs in several orders of magnitude the reliability of interpolated climate data. I honestly can not understand why researchers adhere to these interpolated climatologies once it has been demonstrate that they fail to produce adequate results for species distribution modeling. Good series of data exist from MODIS satellites, covering the complete surface of the Earth, and adequate contributions have been published explaining how to extract the components of the annual variability. The scripts to carry out such simplification of annual components of variability are publicly available. Satellite images are free to download, they can be easily transformed to obtain variables with ecological meaning. Could you please explain me why these are not used in your paper?
4. The wrong choice of methods. It is the rule of the thumb in environmental modeling that the use of a large background for estimating the distribution of the target species will increase the “reliability” of the models using the widely acclaimed AUC. If the authors want to credit the distribution of some species of ticks that colonize Far East Russia, the use of the complete Earth as background will unquestionably increase the reliability of the models. This is a very important mistake in this paper. I CAN NOT believe on the accuracy of these results.