Network of the STR haplotypes of the main subhaplogroups of Hg J. These networks were obtained by the analysis of a subset of the samples for the following microsatellites: YCAIIa, YCAIIb (Mathias et al. ), DYS388 (Thomas et al. ), DYS19, DYS389, DYS390, DYS391, and DYS392 (Roewer et al. ), by the same procedures used for Hg E (). Apart from the YCAII system in Hg J-M267, which was considered as a stable marker in this haplogroup (see text), the STR loci were weighted according to their relative variability in Hg J. The most complex networks, J-M267* and J-M172*, were calculated by the median-joining method (ɛ=0) on the preprocessed data with the reduced-median method; the other networks were calculated by using only the reduced-median algorithm. The shaded area in J-M267* indicates the branch characterized by the YCAIIa-22/YCAIIb-22 motif. For the areas of the circles and the sectors, see . The expansion time of this branch was calculated using TD (Zhivotovsky ), which gives 8.7 and 4.3 ky, respectively, for the earliest and the latest bounds of the expansion time. The former estimate was calculated by using the variance in the number of repeats of the remaining six loci, assuming a variance at the beginning of population separation (V0) equal to zero, and thus gives an upper bound for the TD (Zhivotovsky ). The latter assumes a linear approximation of the within-population variance in repeat scores as a function of time and takes a predicted value of V0 prior to population split; because the linearity can be achieved in a case of infinite population size only and because each survived haplogroup started from one individual and could maintain small size for a long time, the linear approximation overestimates V0 and thus might be considered as a lower bound for divergence times (L.A.Z., unpublished method).