Ants, along with bees and a few other species have a haplodiploid sex determination system, unlike the XY system in mammals. This has a variety of consequences that are hard to swallow at first sight -- males have no father, all workers in a colony are clones, and so on. This is what leads to their highly social behavior within a colony: the good of one is the good of all. I don't yet understand how supercolony behavior arises out of ant genetics, but I'm sure it somehow does.
It appears to me that this one of those evolutionary behaviors gone awry due to the rapidly changing environment. In the past, you could expect that ant colonies would migrate geographically at a fixed rate (on average); so with a constant mutation rate, you'd expect territoriality to arise from this behavior -- ants colonies in regions within (say) a 100- mile radius would be co-operative. Quite clever, actually. However, since humans are capable of transmitting the ants anywhere instantaneously (in evolutionary time), it has produced results that evolution didn't "intend."
BTW here's the original paper http://www.springerlink.com/content/u081508512u7k421/
More info. about the species in question: http://en.wikipedia.org/wiki/Linepithema_humile
Also, http://en.wikipedia.org/wiki/Haplodiploid_sex-determination_...