Aboriginal land boundaries

[1ec5]

Native American reservations and other tribal lands are an integral part of the system of boundaries in the United States. Plenty have already been mapped as boundary=aboriginal_lands relations. Many are large enough to be displayed at mid- and low zoom levels.

We should figure out a way to depict these boundaries that won’t be confused with non-indigenous state and local boundaries but still communicates their importance as delimiters of sovereignty. There was an extensive discussion in gravitystorm/openstreetmap-carto#3520 on this topic; some of the considerations may also apply to this style, although so far we’re rendering fewer things that constrain our options compared to openstreetmap-carto at the time.

For reference, most of the maps in my collection fill Indian reservations as they would fill parklands and military reservations, but this treatment doesn’t seem to be as intuitive as a boundary line, which is how Florida DOT and Best Western maps depict them:

(I find it interesting that the Oklahoma Department of Transportation’s maps omit the reservations entirely. That might be an editorial decision as much as a design decision.)

[ZeLonewolf]

Blocked by openmaptiles/openmaptiles#1296. OpenMapTiles does not currently support boundary=aboriginal_lands

[1ec5]

Another example from a Benchmark Maps atlas of Arizona, which leans into the earth-toned fill. The labels vary in size depending on the reservation’s area, with some labels as large as some of the largest cities’ labels:

Some maps also mark each reservation’s capital or tribal headquarters. OSM relies on the reservation’s boundary relation to contain the capital’s place node as a member with the admin_centre role (example).

This project has been relying on the capital key to identify capitals and county seats (#384). However, capital is supposed to be set to a numeric value matching the admin_level tag of the associated administrative boundary, whereas so far reservations have been tagged as boundary=aboriginal_lands without any attempt to fit into the admin_level hierarchy. (There’s probably a more nuanced way to distinguish between federally and state-recognized reservations anyhow.)

Arizona’s official state tourism maps mark the tribal headquarters in each reservation with a star that matches the state capital icon, indicating the tribe’s sovereignty to some extent. On the 2019 map, the only difference is the icon color:

On the 2021 map, the stars are the same color, but cities are denoted by a circle, which in the case of Phoenix surrounds the star. This puts the tribal headquarters on par with the state capital but makes clear that the state capital is a full-fledged city:

[quincylvania]

Fwiw I think this should be targeted toward the v1.0.0 milestone (given the inclusion of issues like #232). Reservations are fundamental to American mapping and, well, "Americana".

[ZeLonewolf]

Targeting this for 1.0.0 would be problematic since it's not yet present in openmaptiles yet. It would imply that we would either need to hold things up for that update OR implement custom vector layers.

I've suggested an approach to this on the openmaptiles side in openmaptiles/openmaptiles#1296 (comment) though I've not gotten any feedback on the approach. However, it's probably something ripe for working on.

[1ec5]

Targeting this for 1.0.0 would be problematic since it's not yet present in openmaptiles yet. It would imply that we would either need to hold things up for that update OR implement custom vector layers.

v1.0.0 is blocked on OpenMapTiles anyways: #216.

[ZeLonewolf]

It's possible to render "tracks, paths, and trails" in the current OpenMapTiles, even if it doesn't have all the attributes we would like. I would not consider v1.0.0 blocked by OpenMapTiles as a general rule.

[1ec5]

For reference, most of the maps in my collection fill Indian reservations as they would fill parklands and military reservations, but this treatment doesn’t seem to be as intuitive as a boundary line, which is how Florida DOT and Best Western maps depict them:

By the way, the OpenHistoricalMap project has also been grappling with how best to depict indigenous lands: OpenHistoricalMap/issues#204 (comment). Unfortunately, the solution for the Historical style probably won’t translate well to this style; a solid gray line would easily be mistaken for a road.

There’s something to be said for giving tribal administrative boundaries similar treatment to non-tribal administrative boundaries, but one challenge is that the two kinds of boundaries often cross in complex ways. That orthogonal relationship to state and county lines might be the motivation behind shading them in. If that implies the wrong level of sovereignty, then more prominent labels and a capital icon could help, as in the Benchmark atlas in #105 (comment). But we’ll also need OpenMapTiles to consider admin_centre relation roles, not just the capital key.

[1ec5]

As of today’s Americana tileset, most of Oklahoma is a verdant park:

This demonstrates why openmaptiles/openmaptiles#1489 should never have conflated boundary=aboriginal_lands with leisure=park. We can and should temporarily work around the issue by filtering out features with class set to aboriginal_lands, both in the style and in the legend code. But in the long term, the tileset needs to stop treating apples like oranges (to put it mildly).

[Scott5114]

As of today’s Americana tileset, most of Oklahoma is a verdant park:

When this is corrected, if at all possible, most of Oklahoma should probably be treated differently than other aboriginal lands so as to not confuse the user as to what is trying to be depicted. On most maps, shaded areas indicate reservation land, and this is probably what most people would associate a shaded area with.

However, in Oklahoma, the vast majority of tribes have "service areas" rather than reservations. These are more or less non-exclusive jurisdictions of tribal governments. What this means exactly is complicated, with McGirt v. Oklahoma making it more so, but the short version is that unlike a reservation (as I understand it), a non-tribal member can purchase land, live, and do business in the service area without ever interacting with the tribal government, instead only interacting with local, county, and state government agencies as usual. If you are a tribal member, there are certain services available to you if you live within the service area of your own tribe. If a tribal member commits a crime within the boundaries of any tribe's service area, they may be placed in the custody of tribal police rather than the county sheriff. (How this works exactly post-McGirt is still being ironed out.)

Therefore, while these are aboriginal lands (in some sense, at least: this is where tribes were removed to by the Jackson administration, and their actual aboriginal lands are further east), they don't function the same way as the reservation lands in, say, Arizona and New Mexico. Typical symbology for these, when mappers bother to include them at all, is normally an unshaded line feature, like a county line but with a different dash pattern.

A caveat here is that I have seen a number of maps that use the "tribal reservation" shading for Osage County. I'm guessing this means that there is some sort of legal distinction that makes the Osage lands function more like a traditional reservation, but I'm not exactly sure what the details are. (The Osage Nation government in Pawhuska might be willing to shed some light on the matter should it come down to it.)

[1ec5]

Thanks for this insight. You’re right that the boundary=aboriginal_lands tag alone doesn’t capture the different forms of government or degrees of sovereignty between these tribal areas. Nor does it seem to extend to the larger reservations’ political subdivisions. For example, the Navajo Nation’s agencies are tagged nearly identical to Midwestern townships, so that a geocoder might mistakenly think they’re subordinate to counties. The only mechanism for highlighting these distinctions is border_type, which can capture a lot of nuance, perhaps too much for a renderer to know what to do with. I don’t think OpenMapTiles looks at border_type, but if it’s already converting protection_title to something resembling keywords (openmaptiles/openmaptiles#1296), I don’t see why it couldn’t also expose border_type.

Back when the OSM community debated introducing boundary=aboriginal_lands, there was some concern about shunting these boundaries over to a tag other than boundary=administrative. In fact, there was debate among some members of different tribal nations about whether that would be a problem. But there was a sense of urgency because the preexisting tagging scheme was so much worse: boundary=protected_area protect_class=24 was based on the tagging for wildlife refuges, distinguished only by a faux IUCN category of 24 chosen arbitrarily.

In the end, what OSM wound up with was something of a political compromise for the mapping communities in other countries with vastly different relationships to indigenous peoples. But mappers periodically wonder what to do about off-reservation trust lands and, indeed, the better part of Oklahoma. Now, with this change to alias boundary=aboriginal_lands to the same layer as leisure=nature_reserve, it kind of feels like we’re back to square one with OpenMapTiles. The regression affected not only Oklahoma but also reservations elsewhere.

Personally, I think an ideal treatment would be an unfilled boundary line with an interior halo – but I’d favor an interior halo for any administrative boundary. A complete fill would introduce too much ambiguity when tribal lands have their own political subdivisions, tribal national parks, etc. We have more options for disambiguating a boundary line, such as dash patterns and labels with glosses (along the lines of #503). Unfortunately, MapLibre GL JS doesn’t support interior halos: mapbox/mapbox-gl-js#769. The renderer has some longstanding shader limitations that date back to the Mapbox days: mapbox/mapbox-gl-js#8183 mapbox/mapbox-gl-js#6816. For country and state boundaries, we’re currently faking it by plastering an opaque, preworn color band underneath the dashed boundary line, but the effect falls apart if we ever need to make the band a different color on either side of the line.