This would allow the range of the covariate to vary with a grouping factor, such that the data slice doesn't result in extrapolation where for some levels of the factor, the covariate is only observed over a part of the overall range of the covariate in the data.
Suggested by @nsj3 and see the Woodbridge et al example in my materials for the BES Palaeo R workshop, November 2022
This would allow the range of the covariate to vary with a grouping factor, such that the data slice doesn't result in extrapolation where for some levels of the factor, the covariate is only observed over a part of the overall range of the covariate in the data.
Suggested by @nsj3 and see the Woodbridge et al example in my materials for the BES Palaeo R workshop, November 2022