We have characterized the mitochondrial DNA (mtDNA) and non-recombining portion of Y-chromosome (NRY) variation in a sample from Senegal as well as two major groups of Philadelphians: self-identified European Americans and African Americans. These two groups comprise over 88% of the Philadelphian population (45% and 43.2%, respectively, according to the 2000 U.S. Census). We found mainly African haplogroups in the Senegalese sample, with the exception of 12.2% of Senegalese (3 Wolof, 2 Fulbe, and 1 Sahalle) carrying U6 and U5b1b mtDNA haplogroups that, although haplogroup U is of Eurasian origin, can be found throughout North Africa as a result of an ancient migration back to Africa. In Philadelphian African Americans, we observed a significant European admixture (mtDNA>9% and NRY>31%) as well as a small (<2%) contribution from Native Americans. To calculate the corresponding autosomal ancestry of self-identified African Americans, accounting for both maternal and paternal contributions, we used our data to compute mAUTO
½ mNRY 
, which was estimated to be: 78.4% African, 20.1% European, and 1.5% Native American. These calculated estimates seem to accurately reflect the autosomal group admixture, based on typing a small subset of samples using autosomal AIMs (n
31, 74.4% African, 23.7% European, and 1.9% SE Asian/Native American, Table S4
). Also, these estimates parallel previous reports, although our estimates suggest a higher European contribution, especially compared to the 12.7–13.8% autosomal and 2.8–11% low resolution maternal European ancestry found in a sample from Philadelphia reported by Parra et al. 
. For example, European contribution to NRY, autosomes, and mtDNA was estimated to be 28.46%, 19.99%, and 8.51%, respectively, in African Americans from Pittsburg, Chicago, Baltimore and North Carolina 
, or autosomal ancestry of African Americans from NY state was estimated to be 83% African, 15% European and 2% Native American 
In contrast to the admixed nature of African Americans, we observed little admixture in the European American sample (<7% in mtDNA and <1.7% in NRY). Group ancestry or uniparental admixture in European Americans has not been widely reported. However, reports of admixture using the FBI mtDNA database 
or autosomal loci have presented estimates that are consistent with our findings that European populations have contributed the vast majority of ancestry of European Americans. In our case, the calculated autosomal admixture is 95.8% European, with African and Native American contributing less than 5% (2.8% African, 1.4% SE Asian/Native American), consistent with published work (1.6% and 1.2%, respectively, in the US 
), as well as our own autosomal estimates from a subset of the samples (95.7% European).
To further characterize the ancestry of Philadelphian populations within the global context, we mined the literature for published reports of mtDNA and NRY variation, selecting Brazil and Cuba as representative of South America and Caribbean that have sufficient resolution, sampling range, and sample size to represent the country. First, we analyzed admixture in published reports that contained mtDNA and NRY data from White and African Brazilians that were comparable to the data we collected in our Philadelphia sample (). This analysis revealed directional admixture patterns. First, separating the maternal and paternal admixture shows clearly that European males contributed to the populations of America to a greater degree than European females. This is true for both African- and European-derived Americans, although less pronounced in the case of the Philadelphian European American sample. The admixture data in the general population of Cuba support this trend.
Therefore, while male admixture is dominated by European Y-chromosomes, the female admixture shows a remarkable influence of African and Native American female ancestors, the latter prominent mainly in the South American/Caribbean pool, as seen in Brazilians and Cubans. For example, both African Americans and African Brazilians have a high percentage of admixture from European NRYs and some non-African mtDNA admixture that is drawn mainly from European or Native American mtDNA pools in North and South America, respectively. On the other hand, both European Americans from Philadelphia and White Brazilians 
do not show admixture in their paternal gene pool (NRY being almost 100% European in both cases), while, as in the case of African-derived populations, the African and Native American mtDNAs contribute greatly to the maternal pool of White Brazilians. This is in contrast to the maternal pool of Philadelphian European Americans that shows <7% admixture from non-European sources, consistent with the European Americans from the FBI mtDNA database 
. Thus, there are distinct differences between North and South America in the extent of admixture from the three founding populations in the pool of New World individuals who self-identify as “black” and “white” (Table S5
To further investigate whether the patterns we observed in Philadelphian, Brazilian, and Cuban populations have a similar impact in other countries in the Americas, we compared published mtDNA and NRY frequencies of African-derived and general populations, considering the demographics of the investigated countries. Focusing first on the published mtDNA and NRY admixture of African-derived populations of the Caribbean, Colombia, Brazil, and Uruguay, it is clear that they show the same trend as the African American and African Brazilian populations analyzed in this paper (). Namely, from North to South, there is a decrease in the contribution of both maternal and paternal African ancestry, mainly due to admixture with Native American females and European males. Also, we detect the same sex-biased admixture, where more African females than males contributed to the pool of African-derived populations across the Americas. While the observed North-South trends seem to be consistent in African American populations, in order to dissect in greater depth the processes that shaped the populations of North and South America differently, we turned our focus to European-derived populations.
Published mtDNA and NRY profiles of African-descended Americans.
Since previously published data on maternal and paternal ancestry of European-derived populations are scarce 
, we studied this group indirectly by correlating the known mtDNA and NRY ancestry of African-derived and general populations with demographic information. First, although only 5% of the African slave trade arrived in North America 
, the US has the highest proportion of self-identified African Americans (~13%) out of the regions studied (with the exception of some of the Caribbean islands, such as Jamaica, which has up to 91% self-identified African-descended individuals). This implies that a significant portion of African parental variation in South America (and parts of the Caribbean) exists as part of admixed populations (e.g., Mulato and Pardo, although little African admixture was reported in Mestizos 
). To evaluate whether the genetic data are consistent with this hypothesis, we calculated the proportion of African male and female lineages that were contributed to the general population solely
by individuals who self-identify as “black” or African (Table S6
). Nearly all of African Y-chromosomes are found in individuals that self-identify as “black”, whereas less than 50% of the African mtDNAs are found in these individuals. In other words, a significant fraction of African mtDNAs are found in groups that do not self-identify as “black”. To determine which other populations these African maternal lineages significantly contribute to, we estimated the possible admixture in European-derived populations. While these estimates are only approximate, the proportion of contribution of European females to the pool of “white” individuals in the Iberian-founded Caribbean and South America is clearly lower than to European Americans in the United States, with variable proportions of African and Native American females contributing to each of these populations (Table S7
Sex-biased admixture is not a process unique to the Americas. The pattern of NRY variation documents this phenomenon on every continent. For example, the unique Y chromosome lineage spread by (males related to) Genghis Khan over the vast steppes of Asia 
or uni-directional mating of Bantu males and Pygmy females 
can both serve as examples of the history of a population being reflected in Y chromosome phylogeography in the Asian and African continents. In the Americas, European males contributed significantly to all admixed populations 
. However, the difference between North and Caribbean/South America lies both in the diverse cultural histories that categorized people of admixed ancestry either by descent or color 
, as well as the availability of European women. While individuals with any amount of African ancestry were considered “African American” in the United States (the “one drop rule”), in Brazil, where most of the first settlers were male, unions between European males and Native American/African females were common and the “skin tone” of offspring was used to define an individual's “race” 
. Therefore, in contrast to European North Americans, who have relatively low levels of non-European admixture from both male and female predecessors, individuals categorized as “White” Brazilians show higher levels of African and Native American admixture. This non-European ancestry is almost entirely derived from maternal lineages.
There are several limitations of our study. First, our estimates of admixture in European-derived populations in the Americas should serve only as approximations, since this information was mined from indirect sources (genetic data from complementary populations and demographics). Second, in spite of the advantages of using uninterrupted single locus-like information to trace maternal and paternal ancestry, the use of uniparental markers is limited to group ancestry estimates, bearing only very limited information about the ancestry of an individual. Care should therefore be exercised when interpreting our results on anything other than the group ancestry level.
We have shown that estimates of group ancestry derived from combined mtDNA and NRY admixture estimates predict average autosomal ancestry. When separated, these estimates mirror gender-specific admixture processes, reflecting diverse socio-historical demographic processes. Also, groups sharing the label of self-identified race across the Americas are often shaped by different social pressures and this will be reflected in their genome. This may add to the complexity of the population stratification issue in molecular epidemiology, which strives for enhancing the analysis by increasing the number of individuals. In the future, characterization of source European, American and, more importantly, genetically diverse African populations that contributed to the admixed pool of the Americas would enhance the present analysis.