Hi!
The update process works by installing to a temporary lock directory and testing, and then renaming it to the package name, overwriting the existing package version.
Dependencies actually are user-installed, as a consequence of installing the packages that need them. The packages pane is a peek into the default user (not system) directory of R packages and it would be cumbersome to organize that directory to distinguish between the primary package and its dependencies, especially when many dependency packages often are also used apart from the primary package.
(How's the writing goin'?)