To operationalize CARR for tribal boundaries we computed area-weighted averages of the original block-group scale CARR. U.S. Tribal boundary areas published by the U.S. Environmental Protection Agency (EPA) representing American Indian reservation, tribal statistical areas, and off-reservation land trusts was used.

In addition to presenting the area weighted average CARR scores, we also independently calculated area weighted averages for each constituent component of CARR. U.S. census urban-rural categorizations and USDA ERS rural urban continuum code (RUCC) classifications were summarized by taking the mode across all block groups intersecting with a tribal boundaries.

The values of the tribal boundaries CARR scores fall within the original range of CARR scores (0-1), where a value of 0 means very urban with high concentrations of services and amenities and a value of 1 means very rural with low concentrations of services and amenities.

However, the distribution of CARR by tribal boundaries is shifted towards higher scores and the range is compressed (as should be expected from the aggregation process). The minimum score is 0.44, the maximum score is 0.99, and the mean score is 0.63. (In the original CARR the mean value was ~0.5.)

The spatial distribution of CARR across tribal boundaries is consistent with the original CARR although the irregular spacing and size of tribal boundaries makes this difficult to assess.

Like the original CARR, the index values at the tribal boundaries level indicate generally higher values of CARR in census designated rural areas but with a large degree of overlap in urban and rural CARR scores.

Notably, the tribal boundaries CARR scores show no apparent trend with RUCC, perhaps in part to the constrained CARR range and reduced sample size. Areas classified by RUCC as highly urban have about the same median CARR score as areas classified by RUCC as highly rural.

The availability component spatial distribution is consistent with the original CARR. Larger values of this component, which has an inverse relationship to the final CARR score, are associated with greater density of services and amenities and are generally found in more developed metropolitan areas. Note that this component has a highly skewed distribution and is log transformed for visualization purposes.

The three accessibility components’ spatial distributions are consistent with the original CARR as well. Larger values of the first accessibility factor are associated with greater distances to amenities. Larger values of the second accessibility factor are associated with increased distance to essential services. Larger values of the third accessibility factor are associated with greater distances to both amenities and natural resource extraction (e.g. peri-urban).

We also see that the relationships between tribal boundaries CARR and population density, distance to metro areas, and developed land cover meet expectations. CARR has a negative relationship with population density and developed land cover and a positive relationship with distance to metro areas.