yes, the reason for this is that this plugin doesn't run a maven reactor, but rather follows the maven conventions to read artifact information from a maven repository (http://docs.codehaus.org/display/MAVEN/Repository+Metadata). It knows nothing of pom.xml and settings.xml and other maven things.
The reason I built the plugin in the first place was to avoid bootstrapping the heavy maven just to find a couple of strings. I use it in a lot of deployment jobs which themselves don't run maven, but rather wget things for deployment or update metadata about future deployments.
Basically all the plugin does is a single HTTP GET of the maven-metadata.xml file, parses it and presents the information inside in a dropdown (or has some simple logic to select one of those versions automatically) and does some simple string concatenations to generate a few env vars.
It would be a lot of work for me to read and parse settings.xml files, or to integrate a maven reactor and I wouldn't want to do it anyway, because at the time I wrote this plugin there were others who already did that.
You can try this plugin instead if you like: https://wiki.jenkins-ci.org/display/JENKINS/Maven+Repository+Client
I myself haven't used that plugin, but it seems from the description it will do what you want.
There is as strange behavior in this plugin: I used the plugin within a maven build job to present available nexus artifacts.
From the beginning, there is no artifact version presented - empty list, I see 401 errors in the log file instead.
If I click on "build" with this empty list, nothing happens - but suprisingly on the next try to build the project: the list is filled with the artifacts??