Weak convergence to a Markov chain with an entrance boundary: Ancestral processes in population genetics (Q809925)

In the first part of the paper the main result is derived stating conditions under which a sequence of Markov processes converges to a continuous-time Markov chain with an entrance boundary. The form of the conditions is motivated by the genetic problems discussed later. This result is then applied to prove a conjecture of \textit{J. F. C. Kingman} [see Exchangeability in probability and statistics, Proc. int. Conf., Rome 1981, 97-112 (1982; Zbl 0494.92011)] in mathematical population genetics concerning the convergence of an ancestral process associated with a wide class of exchangeable reproductive models. An ancestral process is a process which counts the number of ancestors of a random sample. Using the main theorem it is shown that a sequence of these processes converges to a particular death process with an entrance boundary at infinity. Furthermore, it is shown that the weak convergence of the absorption times and the weak convergence of the marginal distribution of the ancestral process are equivalent leading to a Wright-Fisher diffusion approximation of the absorption times of many neutral genetic models. Sometimes genetic models consider rather the number of lines of descent without any intervening mutation than the numbers of ancestors. Finally it is shown that such corresponding population lines of descent converge weakly to another death process.