The Evolution of Communication

From Causes of Color:

To attract the potential pollinator to that particular blossom, availability of nectar has to be advertised to the butterfly. This is displayed in the color of the petals. The color of the nectar guide of Aesculus hippocastanum [horsechestnut] changes from yellow to red when nectar is no longer in production.

I've heard of flowers having markings to attract insects, but never of those markings changing, like some organic "no vacancy". Why would they want to do that? It seems too difficult, why not always advertise "vacancy" and let the insect figure it out themselves? This must be a sign of ID, right?

The meanings of the color can only exist if they change in the first place. This change happens to be evolutionarily beneficial: imagine a bunch of horsechestnut flowers; only a few are producing nectar. If the insect visits them at random, there is a low probability of fertilization and reproduction. However, if there is some sort of signal coupled with the production, the insect will adapt to this (possibly by learning, more likely by selection over multiple generations).

The color only has meaning when it's related to some other property. Language can only evolve when there is a definite semantics; and when this is useful, it will necessarily evolve. This makes useless and ambiguous language — nonsense and poetry — the zenith of communication.

(Update: Bruce MacLennan has some done some interesting research studying the emergence of communication within artificial species, echoing these ideas. See his more recent summary or the original from 1990.)