First, when you talk of isolation sources, this term is not well-defined. But I assume you mean geographic sampling location or coordinates of an isolate. Now you need to look for databases that store sample or isolate information. But keep in mind that some samples may not be isolates, they may be lab strains and that bacteria could live in very many places, some other species might have an unknown origin. I think you want something like this: let's have a look at Alcanivorax borkumensis, a species where already the name tells you its origin: https://mmp2.sfb.uit.no/MarRef/records/MMP3138202/ MarRef as part of the MarineMetagenomics Portal. I think this is an excellent resource, but this only works for marine meta-genomics samples or marine species. (Still, the location is not in exact geographic coordinates)
In NCBI, every taxon that has a BioSample may have recorded the sampling location/coordinates.
In ENA, the same will be true, but look for example at this entry for Corallococcus: https://www.ebi.ac.uk/biosamples/samples/SAMN10026047
In ENA and in NCBI the location is "United Kingdom" which is pretty useless. Unfortunately, exact geo-locating biological samples has not been a thing until recently.
It is understandable that you want this information. For metagenomics projects, this is essential meta-data. Unfortunately, there are competing standards on how to store this type of information and variable record completeness. In Elixir-Norway, we have a work-package dedicated to biodiversity and we are working on improving and harmonizing meta-data standards for all sorts of studies related to biodiversity. So, for some species, you will have success and it is good to draw attention to the relevance of geographic data and meta-data in particular, but I still doubt that this undertaking is worth the effort for ALL species at the moment.