Starr County TX is 96% Hispanic, & 96% of Hispanics in Starr identified racially as "White" on the Census. It swung 55 pts towards Trump. In contrast, in Imperial County (80% Hispanic) only 56% of Hispanics identified as White & it swung only 17 pts Trump https://twitter.com/Millenarian22/status/1237041106971267072

So the theory would be that Trump gained by more with "White" identifying Hispanics (53% of Hispanic population) vs. "Some Other Race" identifying Hispanics (37% of Hispanics). I might run a county regression, but it would probably be better with precinct data which I don't have.

On the other hand, if Imperial County has more immigrants who are not eligible to vote than Starr (which has Tejanos who have been there long-term), the White share of the electorate might be significantly higher which would explain why Trump's swing there was much lower.

Also could be a divide based on urban-rural rather than racial identification, I know there is evidence that the swings towards Trump in urban Hispanic precincts in Phoenix were much smaller (tho they still existed) than in many rural or small town Hispanic precincts elsewhere