2.1 Presence records
We searched presence records with geographic coordinates for all sea snakes in public online repositories. These platforms were the Global Biodiversity Information Facility (gbif.org; DOI information associated with data are available in Table SM1), VertNet (vertnet.org), the Ocean Biogeographic Information System (obis.org), the Atlas of Living Australia (ala.org.au), the Online Zoological Collections of Australian Museums (ozcam.org.au) and the UNAM open data portal (datosabiertos.unam.mx). We integrated the presence records obtained from the different platforms into a single database per species. Then, we eliminated duplicate records, uncertain data (e.g., taxonomic mistakes, records outside the known distribution of the group), and obvious errors (e.g., records in terrestrial environments). Next, we eliminated occurrences that might represent sink populations through an environmental outliers’ analysis. In this process, we considered environmental variables previously identified in the literature as important to the group (see below). For this, we used theoutliers function of the “biogeo” package (Robertson et al. 2016) in R (R Core Team 2013), which searches for presence records that are located at a distance above 1.5 times the interquartile range of the data. Additionally, we employed the gridSample function of the “dismo” package (Hijmans et al. 2017) to minimize data bias towards more intensely sampled environments. These processed records comprised the complete database for each species.