The supercritical Galton-Watson process in varying environments (Q1193398)

In a supercritical Galton-Watson process in varying environments, the problem of finding a sequence of norming constants such that the normed process converges with positive probability to a positive random variable (such a sequence being called a rate of growth) is more complex than in constant environments. For example, \textit{I. M. MacPhee} and \textit{H.-J. Schuh} [Aust. J. Stat. 25, 329-338 (1983; Zbl 0536.60084)] found an example with two essentially different rates of growth. The natural candidate for growth rate is the sequence of means, and the present paper gives some sufficient conditions for this sequence to be a rate of growth precisely on the event of non-extinction. In brief, the conditions involve controlling the means and tails of the family size distributions in a uniform manner over time. It is also shown that, under a weaker condition, if several rates of growth exist, then in a certain sense they cannot differ by very much.