The world of plants is one of staggering variety: to date there are around 400,000 species known to science, with many more waiting to be discovered. An interesting facet to this variety is the fact that there can be incredible diversity within species themselves, such that different individuals of the same species growing side by side might not be immediately recognizable as being at all related.
Oftentimes, a variety will supercede the parent species in popularity, perhaps because it performs better in cultivation or has a more interesting appearance, and becomes used more often than the 'original' species. (Note that I am not discussing man-made hybrids or cultivars here- that is a subject for another time.) A perfect example is Dracaena fragrans 'Massangeana', which sports a pale green stripe down the center of its leaves not present in the typical D. fragrans. One doesn't even really see the boring old D. fragrans anymore, so popular has its variegated variety become.
But would the effect of 'Massangeana' not be greater if it were a single specimen amongst several of the regular D. fragrans? After all, that's how one would find such a variety in nature: they would stand out like a sore thumb, and perhaps even give taxonomists a run around thinking it might be a different species. I think it would, and so that's what I do with these types of plants.
I believe that, rather than leaving these varieties to stand alone as representatives of their species, they should be incorporated and used (sparingly) with their parent species to highlight and exemplify the fact of their origins, and to allow people a glimpse at the near endless variety of the plant kingdom.