Migration matrices and surnames in populations with different isolation patterns: Val di Lima (Italian Apennines), Val di Sole (Italian Alps), and La Cabrera (Spain)

Biodemographic methods are widely used to infer the genetic structure of human populations. In this study, we revise and standardize the procedures required by the migration matrix model of Malecot ([19501 Ann Univ Lyon Sci [A] 13:37-60), testing it in large historical-demographic databases of 85 populations from three mountain valleys with different degrees of isolation: Val di Lima (Italian Apennines, 21 parishes), Val di Sole, (Italian Alps, 27 parishes), and La Cabrera (Spain, 37 parishes). An add-on package (Biodem) for the R program is proposed to perform all calculations. Results from migration matrices are compared with those obtained from isonymic relationships. Migration and isonymy matrices are derived from 22,781 marriage records. Matrices are analyzed using a nonlinear isolation-by-distance (IBD) model and multivariate techniques (multidimensional scaling, Procrustes rotation, and cluster analysis). Microdifferentiation levels (FST) from the migration data agree with the observed inbreeding values: higher values are found in La Cabrera (F-ST = 0.0082), the most isolated population; Val di Lima (F-ST = 0.0015) and Val di Sole (F-ST = 0.0012) have lower values due to the larger parish population sizes and greater mobility. Temporal changes of FST and IBD are analyzed using the migration matrix approach. The populations show a marked decline in F-ST values in time, together with increased population mobility and emigration rates. In all three valleys, marital migration and isonymy yield similar results, suggesting that geographic distance is the most important factor structuring the populations. However, isonymy shows a lower correlation with geographic distance than migration matrices do. This difference can be attributed to the differing sensitivity of the methods for past migration events, and to genetic drift.