González-Pérez et al. (2010) have analyzed populations from the northern and southern shores of the Mediterranean, with Central Europeans and West Africans as external references. In the Discussion section, they admit that the inflated "Alu/STR estimate might be artefactual" and favor the estimate based on the Alu loci set alone because it's consistent with previous mtDNA, Y-chromosome and 500,000-SNP structure data.
According to the more accurate latter method, Sub-Saharan African admixture is ~13% in North Africa and "imperceptible" or noise (~ 0.01%) in Southern Europe:
