I'm not finding a systematic analysis of the data to address this question, but I suspect it results from the intersection of two sets of facts:
- "Today, 75 percent of African-Americans live in cities or suburbs. 15 percent live in smaller cities and towns, while just 10 percent live in rural areas, mainly in the South. (source)
- Trump won only 35 percent of the vote in urban areas but 62% in rural areas. (source)
Urban and rural voters have all kinds of strong differences in terms of general political attitudes and values. (source) So my hypothesis would be that those districts with a high density of black voters are more likely to be urban, and that the black vote follows the general voting trend of urban/Democrat vs. rural/Republican. Perhaps someone more ambitious and adept with data can test this directly.