Cultural versus demic diffusion in agricultural expansions according to three definitions of dispersal distances

Abstract

Mathematical models of agricultural spread use distances between birthplaces of parents and their children (often called "birthplace distances"). However, the difficulty to find those distances for pre-industrial farmers has often led to the use of other kinds of distances. One example is the distance between the birthplace and the place of residence of each individual ("residence distances"). Another example is the separation between the birthplaces of parents ("mating distances"). It is poorly known to what extent the latter two distances are valid approximations to birthplace distances. In order to address this question, we have prepared a database with the three distances for a specific pre-industrial agricultural population (the Yanomamö). For the spread of the Neolithic in Europe, all three kinds of distances yield spread rates consistent with the archaeological data, as well as cultural effects below 50% (so demic diffusion was more important than cultural diffusion). The three kinds of distances also yield estimations for the percentage of early farmers who interbred with hunter-gatherers that are consistent with the corresponding estimation from genetics. There is wide agreement for the cultural and demic effects in other expansions of agriculture and/or herding (the spread of the Neolithic in Asia, Bantu, and Khoikhoi expansions; the spread of rice in Asia, etc.) and using distances measured for other populations. We conclude that estimates are largely insensitive to the kind of distances used. This implies that the conclusions drawn so far in the literature using these three kinds of distances are robust