Species occurrence data
We collated and cleaned the occurrence data of the fish and odonate species of Bhutan (see Appendix 1 Table S1). We then rarefied the occurrence of all the fish and odonate species with five or more occurrences to Euclidean distance of 5km using the SDMtoolbox v2.4 (available fromhttp://sdmtoolbox.org/, Brown 2014, Brown, Bennett and French 2017). We aimed to reduce the negative influence of the spatial autocorrelation among occurrences on the species distribution models (Brown 2014, Brown et al. 2017, Galante et al. 2018), but also to retain the maximum possible number of species with at least five occurrences. This resulted in 10 fish and 28 odonate species with small occurrences (n= 5 to 21) for which we built SDMs and used these for the current study.