New methods for reconstructing geographical effects on dispersal rates and routes from large-scale radiocarbon databases

摘要: We introduce a methodology for reconstructing geographical effects on dispersal and diffusion patterns, using georeferenced archaeological radiocarbon databases. Fast Marching methods modelling front propagation enable scenarios to be explored regarding barriers, corridors, favoured unfavoured habitat types. The use of genetic algorithms as optimal search tools also enables the derivation new scenarios, is especially useful in high-dimensional parameter spaces that cannot characterized exhaustively due computer runtime constraints. Model selection guided by goodness-of-fit statistics observed predicted dates. We an important additional model output, namely, modelled phylogenies dispersing population or diffusing cultural entity, based branching networks shortest 'least cost' paths. These 'dispersal trees' can used tool evaluate their degree congruence with reconstructed independently from other kinds information. We illustrate our approach case study, spread Neolithic transition Europe, database literature (Pinhasi, Fort Amerman 2005). Our find support which limited altitudinal cut-off there climate-related latitudinal gradient rate spread. This leads deceleration geodesic distance, consistent models under space competition pre-existing populations hunter-gatherers. meanwhile searched found alternative involving fast along northern Mediterranean coast Danube/Rhine riverine corridor. Both these outperformed geography-free Great Circle distance model, both another, almost geography-free, constrains land near-offshore coastal waters. adjusted coefficient determination dates first arrival supports GA-derived model; path network analysis, however, gives greater rate, since it produces are more congruent sites' clade memberships (as defined material culture).