The Census Bureau has published a diagram explaining the relationships between the various tribal and nontribal administrative areas. This is a heavily simplified diagram that elides, for instance, the three levels of the Navajo Nation and the potential for a state to designate a reservation at the county level or the county subdivision level. It also predates McGirt. Still, it speaks to the logic of assigning admin_level=* in relation to mainstream civil administrative boundaries.
In following annotated copy, I’ve highlighted in yellow the types of boundaries we’re currently tagging as boundary=aboriginal_lands without any formal tagging distinction. Based on what I know so far, I’m concerned that multiple highlighted items sit beside each other in the diagram and that there are highlighted items at multiple levels of the diagram. We’re just throwing everything that’s somehow indigenous into the same tag. (If this is not actually a problem, I’d welcome more information.)
In practical terms, the indistinct tagging elevates the subdivisions to reservations in their own right, making for a confusing map. In the following screenshots, it isn’t possible to tell where the Standing Rock reservation ends and the Cheyenne River reservation begins, or that the latter even exists, despite its size. Instead, it looks as if the reservations have been broken up – a problematic representation, given the history.
As boundary=aboriginal_lands apparently was never designed for anything but the top tier of “American Indian Areas (Federal)/Off-Reservation Trust Lands”, a conservative solution would be to split off boundary=indigenous_administration admin_level=* for federal reservation subdivisions and state-level reservations. In OSM Carto and OSM Americana, the subdivision boundaries would temporarily disappear. This would be unfortunate, but at least it would be possible to see the reservation boundaries again. Changes to these renderers would resurface the subdivision boundaries at a higher zoom level and with dashed lines.
A long-term split between boundary=aboriginal_lands and boundary=indigenous_administration would be awkward, so we need to decide whether and when to move the top-level reservation boundaries to boundary=indigenous_administration. Only a handful of data consumers recognize boundary=aboriginal_lands, so we might be able to migrate all the administrative boundaries over to a new tag simultaneously in coordination with the developers.
An alternative would be to combine admin_level=* with boundary=aboriginal_lands. But if the presence of admin_level=* is the only thing that distinguishes a reservation or subdivision boundary from something less structured, then we’ll probably keep finding examples of data consumers that mishandle these cases. It’d be as if we tag administrative boundaries as boundary=place, relying on admin_level=* to distinguish them from other verifiable boundaries.