Counting the number of packages isn't worth the effort.  R is used by many different communities, e.g., pharma, academia, etc.  Within these communities, linux users tend to gravitate to the same platform and packages used in that community will get attention on that platform.  Many R packages use external libraries, so user communities will insist that these libraries are packaged and usable.  Ubuntu is a very popular distribution, so can be expected to have a wider range of user communities. You may, however, find that key libraries and R packages for your subject area are not current or have unreported bugs (because they are not heavily used).   

While you are learning R, any distro should provide basic packages.   If your ultimate interest is in a specialized subject area, you need to look at the packages and support libraries being used in that field and check for packages of current versions.   For advanced R users, the biggest issue is not R packages, but the presence of workable support libraries.  If you suitable supporting libraries, it is generally very simple to install current R packages from the sources on CRAN.   

My work is in remote sensing and uses spatial statistics and images.  A "mission critical" package from a national space agency was developed on Ubuntu, so I use Ubuntu but have often had to build some supporting libraries (gdal, hdf5, netcdf4) because the distro packages for these libraries were outdated or built with stripped own options that make them unusable for my work.  This situation has improved over time, but just when I think the distro has caught up with my needs a new feature is introduced and I end up having to build support libraries from sources all over again.   Building support libraries often gets into nitty gritty distro-specific details figuring out how to ensure that your R packages use the locally compiled libraires without creating conflicts with the distro-supplied versions of the packages.