Fig. 3
(A) Exploratory Principal Component Analysis (PCA) computed in Adegenet and performed using the 39-STR genotypes of 56 Historical Italian wolves HWIT (dark blue dots), 56 Contemporary Italian wolves CWIT (light blue dots), 175 reference Italian wolves (WIT, blue dots), 89 Italian dogs (DIT, red dots) and 196 European wolves from 5 geographical populations: Dinaric = WDIN, Iberian = WIBE, Carpathian = WCARP, Balkan = WBALK, Baltic = WBALT. The first component PC-I explains 43.94% of the total genetic variability and clearly separates the Italian wolf population from the European wolves and domestic dogs, while these latter two are plainly separated along the second component PC-II which explains 21.44% of the total genetic variability. (B) Estimated posterior probability LnP(K) and corresponding standard deviations of the K genetic clusters from 1 to 10. (C) Bar plotting of the individual qi-values obtained through Bayesian model-based clustering procedures implemented in Structure and performed using the 39-STR genotypes of 56 HWIT, 56 CWIT and, as reference populations, the 39-STR genotypes of 175 Italian wolves (WIT), 89 dogs (DIT) and 196 European wolves from 5 geographical populations (WDIN, WIBE, WCARP, WBALK, WBALT). Each individual is represented by a vertical line partitioned into coloured segments, whose length is proportional to the individual coefficients of membership (qi) to the wolf and dog clusters inferred assuming K = 2 clusters and using the ‘‘Admixture’’ and ‘‘Independent allele frequencies’’ models.
