There is already a planned feature to have groups found new civs after year 1, though for the life of me I can't find the quote so I don't know exactly how deep in the future to expect that behavior. Toady seemed to be thinking of a bandit camp starting Australia, but it would work just as well with a bunch of animal men settling down in a nice cave.
I do not mean that existing groups should develop into civilization. I mean that civilizations should not develop their first real site until a certain randomly determined date rather than everyone suddenly setting up all their first settlements at the same time in Yr1. The problem is site limits, there is a hardcoded and zero-sum limit on the total number of sites that may exist in a given map but I have thought about how to solve this sort of.
The solution is to reserve about a quarter or so of the total sites on the map and then
before Yr 1 create a number of camps for everyone in the correct biomes for starting off. The camps have a population already in them and this is carried over once they become settlements.
On the random date one of the camps of a given civilization turns into a settlement and then it settles the other nearby camps turning them into a settlement too (but without changing the total amount of sites on the map). Only once all camps are civilized does a civilization actually create new sites from scratch.
In order to have new civilizations actually emerge that are randomly generated from animal people or bandit groups we need to find a way to sometimes have surplus sites left over. The best way to do this to give civilizations an urbanisation number that starts ticking the moment all their camps are civilized. What this number does is increase the total maximum population of all their settlements, meaning that their existing sites grow bigger and bigger rather than them making new sites using those people.
Depending upon how many new settlements have been created in the world recently there is a probability of one of the animal men populations within one of the biomes growing into a civilization. Basically the population of the biome's animal men population is divided up among a number of camps and then exactly as with the regular civilizations these camps turn into settlements.
The new civilization's sites are restricted to available terrain and selects a random site type out of the 5 sites available. All the entity values are randomised, so the nature of the animal person civilization is a wild card. It could have any basic nature and values at all.