Data Is Beautiful

It's quite interesting that "rural highways" is one of the categories identified, but not any other sort of improved road. The data source has a base granularity where one square is 250,000 acres (~100,000 hectares), and then additional state data is factored in for increased precision. It supposingly being USDA data, they might primarily care only about those highways used to connect farms to the national markets.

That said, I would be keenly interested in the land used for low-volume, residential streets that support suburban and rural sprawl, in comparison to streets in urban areas. Unlike highways which provides fast connectivity, and unlike dense urban-core streets that produce value by hosting local businesses and serving local residents, suburban streets take up space, intentional break connectivity (ie cul de sacs), and ultimately return very little in value to anyone except to the adjacent homeowners, essentially as extensions of their privately-owned driveways.

It may very well be in USDA's interest to collect data on suburban sprawl, as much of the land taken for such developments was perfectly good, arable land.