Ideally the AI should look to whether it is able to funnel trade into its home node from a colonial region before putting colonies into it. But it may be better and simpler to block certain countries from colonizing certain colonial regions if there's room elsewhere.
Iberian colonies in North America made some sort of sense in 1.7, when almost all trade from the Americas north of the equator was funneled into the Western Europe node, so trade from Chesapeake and St. Lawrence nodes could be funneled into Sevilla. Now it's impossible to funnel trade from Chesapeake or St. Lawrence into Sevilla, so there's no reason Spain or Portugal would put colonies in those trade nodes.
In general it seems that the AI doesn't do a proper cost-benefit analysis for its colonization targets. For example, in my current England game I stole all of Portugal's starting Atlantic islands, so it couldn't start colonizing Brazil and the Caribbean before me. Portugal promptly dropped a colony on the west coast of Africa, complete with negative annual population growth and single-digit chance of monthly population increase. Twenty or thirty years later, that colony now has 200/1000 population, +15 colonists per year, and 7% chance of monthly population increase.
So there should probably also be a hard block to prevent the AI from putting colonies in tropical provinces until 1550 or so.
The exploration mechanic seems okay. It would be neat if countries could designate claims for future settlement, and then other countries could choose whether or not to respect those claims.