Contrived microeconomic models poorly represent reality, as can be readily demonstrated by the importation of non-economic behaviour or new goods into your example model by people attempting to comment on or answer your question.
What would happen in this instance?
Given that fish is a “requirement,” it can be taken to be a necessity of Tribe B’s existence: Tribe B ceases to exist in one consumption cycle. The paucity of the model means we shouldn’t declare this to be physical liquidation, or enslavement or incorporation into A, or a reconstitution of B such that fish are no longer a necessity. Good substitution is prohibited as the only two marketable goods are necessities and non-substitutable.
Is there historical evidence in which a similar scenario has occurred?
We have good historical evidence that not only is the model posited far too narrow, but that its assumption of barter as meaningful is fallacious, and that far denser networks of trade are normal across human societies and economic forms (Graeber, Debt).
However, the question doesn’t hang on it’s curious imposition of bartering. We can rephrase with any means of exchange and deal with technologically based goods substitution making one agent in exchange irrelevant in terms of a past exchange relationship. If it isn’t technical change then the question needs rephrasing to cover lockout, strike, embargo, change in production priority, substitution, dilution, etc…
In examples like nylon parachutes, silk production transferred to other goods: agent “B,” ceased to exist, or was reconstituted on a lesser scale.
However, given the liminality of “tribal” existence labour history and labour economics supply a much more useful pool of experience. In frame knitting (a form of weaving) Agent C developed a new technique which meant that they no longer required as much (eventually any) wage labour from Agent L: frame knitting domestic producers. This is because they mechanised production, using a new body of workers, newly produced as workers. As L was no longer able to exchange wage labour at previous rates, C dictated new ratios in the relationship of wages to output. L engaged in legalistic, civil disobedience and armed action against C. C reduced L’s necessities of life, and reduced their freedom of agency (with A and B we would call this enslavement of B). C’s new production was inferior, but C was accepting of this dilution. Eventually L ceased to exist, losing elements to migration, reconstitution, but eventually due to an inability to accept inter generational reproduction at the standards of life required to continue as a domestic frame knitter.
The question would be best reasked with a complex model that doesn’t import unneeded assumptions; or as a question about technical substitution in economic history.