The non-Markovian nature of migration: a case study in the Aland Islands, Finland

Markov theory is applied frequently to the study of human migration processes. A Markovian migration process is one in which the pattern of migration in one generation depends only on that of the previous generation. Several Markovian migration models are discussed here. It may be, however, that the presence of ancestors in a particular place had a significant bearing on the decision to migrate, in which case the process is not independent of the past. One example of such a process may be the parish of Kökar in the Aland Islands, Finland. Parent-offspring migration data which span the last two centuries of migration in Kökar are used here to test the assumption that the migration process in this parish is Markovian. Matrices of observed parent-, grandparent- and great grandparent-offspring migration are constructed and compared to the respective matrices expected on the basis of Markovian migration, using a X2 test for independence in a contingency table. Results indicate that a Markovian model does not describe adequately the migration process in this parish. Possible genetic implications of a non-Markovian migration process are discussed, with particular reference to populations of the type presented here which are experiencing the transition to urbanization.