I've been reading about different methods used in linguistics, and I've been puzzled by the usage of placenames in identifying substratums in modern languages. Just because some language and its speakers exist in some area for a time doesn't imply that that language and its speakers had any effect on the colonizers/invaders that came through later. As an example, I live in Northern California, and as a result of Spanish explorers coming through this area first, many of the placenames are of Spanish origin. However, once California was annexed by the United States, it was fairly rapidly Anglicized and there has been no Spanish influence on the English spoken here (until perhaps the more recent influx of Mexican immigrants).
Placenames are NOT considered to be a valid way of identifying a substratum. You identify a substratum by examining language contact between multiple languages in an area over time. Then you need to consider similarities in the grammar, syntax, and vocabulary. For example, Louisiana French Creole and African American Vernacular English share many similarities that may point to a shared substratum (i.e., West African Pidgin English or other African languages). Why might I think these two languages share a substratum? Well, I might look at the percentage of copula-deletion in my corpus and compare the results to show how they match up to the proposed substratum. I might look at the inclusion of African vocabulary in both languages, such as the word 'voodoo'. Then you might consider the speakers, mainly African slaves and their descendants, their lack of a formal dominant language education, their need to communicate amongst themselves and their masters, and things of that nature.
Placenames simply do not offer a good strategy of deciding a substratum language and its influence. For example, in the New York area many placenames are Dutch, but you would be hard-pressed to make the argument that Dutch is a substratum of English.