Dear John,

I think, none of them are important. These are global models, sampling domain is interesting if you use regional climate models instead of GCMs. GCMs are independent of the studied species, so the species is unimportant as well. If you want to select only one GCM, you may want to read some literatures that are about comparing the advantages and disadvantages of different GCMs. Anyway, I recommend using a multi-model ensemble, which is able to deal with the intermodel variability (i.e. a source of uncertainty).

HTH,

Ákos

Agree with Ákos Bede-Fazekas. Consider looking into the multi-model compilations done in Climgen (google "climgen") from University of East Anglia. I have used that one - It is very good.

Hi John,

Indeed, the selection of GCMs is very important as variability occurs among alternate simulations; however, keep in mind that there is no single ‘best’ model. Beaumont et al. (2008) mention: "To date there is no clear guidance on how to select the most appropriate simulations for a given application. There may be little objectivity in selecting climate models for impacts modelling, and the final decision may be influenced by data access issues or based on the untested (and probably incorrect) assumption that a climate model developed by researchers in a particular country will be the most robust model for that region".

There are alternative ways to select the most appropriate GCM. For example, by assessing their skill scores used to compare their ability to simulate observed conditions across areas (see Whetton et al. 2015). Another option is using multiple GCMs and assess consensus areas of suitability across multiple GCMs. Finally, you can use an average of multiple GCMs, but keep in mind that this average may not represent a accurate scenario because, for instance, if one GCMs projects an increase of 10% in precipitation whereas another one predicts a decrease of 10%, then the average would result as no change in precipitation, which is not correct for either GCM.

Here are some references that might be useful:

Beaumont LJ, Hughes L, Pitman AJ. 2008. Why is the choice of future climate scenarios for species distribution modelling important?. Ecology letters. 11(11):1135-46.

Beaumont LJ, Esperón-Rodríguez M, Nipperess DA, Wauchope-Drumm M, Baumgartner JB. 2019. Incorporating future climate uncertainty into the identification of climate change refugia for threatened species. Biological Conservation. 237:230-7.

Whetton P, Ekstrom M, Gerbing C, et al. 2015. CSIRO and Bureau of Meteorology 2015, Climate Change in Australia Information for Australia’s Natural Resource Management Regions: Technical Report, CSIRO and Bureau of Meteorology, Australia.

Hope this will help you.

Best,

Manuel

Try to use an ensemble of them

Dear John,

this new R+shiny application for comparing the available GCMs may be of your interest:

https://besjournals.onlinelibrary.wiley.com/doi/abs/10.1111/2041-210X.13360?af=R

HTH,

Ákos

All of your answers helped me a lot. Thank you.

Dear John,

There is a very interesting article that, although it is based on the previous CMIP5, can be very useful to choose a GCM according to the study region:

C. F. McSweeney et al. (2015). Selecting CMIP5 GCMs for downscaling over multiple regions. Clim Dyn (2015) 44:3237–3260.

Hope this will help you.

Best,