Brilliant! THAT'S the piece I was missing in this discussion. Your last paragraph sums things up nicely. (One of those "duh-oh" moments *smacks forehead*). Selection is local, not global. Allow me to modify your population growth equation a bit:
PG = rN[(K-N)/K]
Where population growth is restricted by the carrying capacity (K) of the local environment (r is effective growth rate or birth - death + immigration - emigration, N is current population). The term (K-N)/K represents the amount of carrying capacity remaining for population growth. When N approaches K, growth approaches zero.
What this implies is that high fecundity
may ultimately be
negative in terms of fitness for the population. IOW, density dependent factors limiting population growth in the developing world will likely become more and more critical until local K - N is zero. As K in the developing world is much lower than K in the developed world, it is conceivable (all other things being equal), that the fecundity issue that crash discusses may be self-limiting - which is disturbing from a moral standpoint. However, local economic development and technology may be able to reverse both trends - the former by reducing r, the latter by increasing K. Maybe we're not going to hell in a handbasket...