Thanks for your questions @sariya
The only thing R version matters for when installing a package is the minimum R version specified in the DESCRIPTION file, e.g., for taxize it is R v3.2.1 taxize/DESCRIPTION at 393b71714fbace9088254dc732e28eeb0cfad5cc · ropensci/taxize · GitHub - that’s just a minimum version, so any R version greater than that is fine.
CRAN builds binaries of packages for oldrel, release and devel, which has R versions associated, right now they’re 3.6, 4.0, and 4.1. I don’t know exactly how those work, but I assume when you install a package, the install.packages code downloads the appropriate one closest to the version you have, but not sure if that;s right.
In general I’d suggest using the latest stable R version (v4.0.2), that’s what I do. Unless you have a good reason to use an older R version, for example if you have a project that you don’t want to upgrade because for example it only works with older versions of some packages.
It may or may not matter, depending on the situation. In general as I said above I’d stick with using the latest stable R version. If you need to use an older version just be aware that some things change between R versions. For example in I think R v4 and greater
stringsAsFactors in data.frame’s is by default
FALSE, whereas before v4 it was
TRUE, so that can lead to different results if you’re contributing to a package and using v3x where others are using v4x. Another example is raw strings, introduced in R v4.