Suwa made a good job of an impossible task, and, like all generic keys, it is very difficult to reach a satisfactory conclusion. The reason is that the genera of Anthomyiidae are characterized by the structure of the male and female genitalia. So closely related species may have different chaetotactic characters. These are the main externally visible characters. So that if one writes a key in which one couplet runs to species with an anteroventral seta on the mid tibia (a good and easily recognisable character) one will end up with (as a made up example) 2 species of Botanophila, 4 species of Delia and miscellaneous species of several genera.

The only reliable key to genera would be one to genitalia, and that would explain their phylogenetic groundplan; but means that one would have to dissect every specimen to identify it to genus (and species). And that is exactly what I do when examining material from some little known area, such as the Altai Mountains of S. Russia. When I have 2 or more specimens with the same genitalia, I then study the outward appearance of the fly, make a description, and then I may be able to separate others of the same species. Often however some later specimens turn out to be another, and new, species.

That is why my key to British genera leads to several groups for some of the genera.