Dear Authors,

The article "Optimising Species Distribution Models: Sample size, positional error, and sampling bias matter" successfully encapsulates a broad review and synthesis of pivotal elements affecting SDMs and ENMs. The authors reviewed an extensive list of publications and provided clear recommendations to address potential problems in occurrence records based on our collective and current understanding of these issues in SDM/ENM. My comments and suggestions are mainly to inspire more detailed and profound discussions on particular topics.

Please find my comments and suggestions below. I have added a few references that the authors may consider including.

Major comments:
1. My major concern is related to the way the authors propose to deal with sampling bias. This comment can be split into three parts:
a) I suggest the authors expand on the usefulness of target-group backgrounds. An appropriately selected target-group background can be considered not only as a way to deal with sampling bias but also as a strategy to account for sampling effort. Therefore, it can also help characterize relationships more reliably, and with that help better characterize species' environmental requirements.
b) I would like the authors to provide a more critical discussion of correcting sampling bias using a thinning in environmental space. The authors correctly explain how SDM/ENM tools help us to characterize conditions where species can thrive. The authors also mention the risk of destroying the signal of species' environmental "preferences" (i.e., the species niche is a bias in environmental space) if this type of filtering is applied. My main concern is that sampling bias is produced by the way we sample, and we sample in geographic space, not environmental space. If such bias is produced in geographic space and we more or less understand how such biases occur, what would be the justification to filter in environmental space? If the author can think of cases in which environmental filtering is appropriate please describe it with enough detail, but also describe the risks and implications of performing this process.
c) The authors state "If the geographic bias is high but the environmental bias is low, no corrections are needed, and the data can be used ‘as is’ for modelling." Other than as a coincidence, I cannot think of an explanation (biological or statistical) for trusting geographically biased data as is. I know and understand the results presented in the references provided. However, this is a critical aspect of managing sampling bias that cannot be left without further discussion. Adding an explanation of why data biased in geographic space may still be good for representing species' environmental requirements will help readers understand this argument.
2. I found the following sentence hard to understand given the context provided: Lines 470-471. "We recommend not to use bias correction methods for specialist species." I cannot think of a biological, ecological, or statistical argument that supports this recommendation as is. As stated by the authors, every species is a different case, and existing data for some of these specialist species may need to be processed to correct for sampling bias. I consider that the recommendation presented by the authors is not generally applicable and should be modified.

Minor comments:
1. Please consider the reference suggested below in the section for Positional errors. This could be a good reference for strategies to explore the effects of positional errors in the data.
2. I think the authors could expand on the implications of considering their recommendations in studies that apply SDM/ENM. At the moment the only clear reference is to conservation actions, but other examples could be added to help readers gain a more general perspective on the applicability of the ideas presented in this study.
3. Could the authors expand on how to transparently report potential biases and errors in the data used for modeling? I believe this information can be useful for a broad audience.

Additional references suggested:
General (these talk about the topic of interest from general theoretical and practical perspectives):
Soberón, Jorge, and Miguel Nakamura. “Niches and Distributional Areas: Concepts, Methods, and Assumptions.” Proceedings of the National Academy of Sciences 106, no. Supplement 2 (November 17, 2009): 19644–50. https://doi.org/10.1073/pnas.0901637106.
Peterson, A. Townsend. Mapping Disease Transmission Risk. Baltimore: Johns Hopkins University Press, 2014.
Araújo, Miguel B., and A. Townsend Peterson. “Uses and Misuses of Bioclimatic Envelope Modeling.” Ecology 93, no. 7 (July 1, 2012): 1527–39. https://doi.org/10.1890/11-1930.1.

Sampling bias:
Anderson, Robert P. “Real vs. Artefactual Absences in Species Distributions: Tests for Oryzomys Albigularis (Rodentia: Muridae) in Venezuela.” Journal of Biogeography 30, no. 4 (2003): 591–605. https://doi.org/10.1046/j.1365-2699.2003.00867.x.
Stolar, Jessica, and Scott E. Nielsen. “Accounting for Spatially Biased Sampling Effort in Presence-Only Species Distribution Modelling.” Diversity and Distributions 21, no. 5 (2015): 595–608. https://doi.org/10.1111/ddi.12279.

Positional errors:
Peterson, A. Townsend, and Abdallah M. Samy. “Geographic Potential of Disease Caused by Ebola and Marburg Viruses in Africa.” Acta Tropica 162 (October 1, 2016): 114–24. https://doi.org/10.1016/j.actatropica.2016.06.012.

Dear authors,

Thank you for your effort in responding to my comments and making changes. I consider that the manuscript has improved with the modifications included and, in my view, is ready to be accepted.