190
J. D. Wall
substantial Khoesan ancestry in South African Bantu speakers (Henn et al. 2011),
and substantial non-African ancestry in the East African Maasai (Wall et al. 2013).
8.3
Effective Population Size
The current human population size is in excess of seven billion and is thought
to have been in the millions for all of recorded history (Tellier 2009). However,
genetic sequencing studies have found that extant humans are quite similar to each
other, with the average sequence divergence between two individuals around 0.1%
(Li and Sadler 1991; Wall et al. 2008; 1000 Genomes Project Consortium 2010,
2012, 2015). This corresponds to an effective population size (i.e., time-averaged
number of breeding adults) of around 20,000–25,000, which is vastly less than our
estimates of the recent census size. While there are reasons why the census size
might be more than the effective population size (e.g., nonrandom mating, variance
in reproductive success across individuals, unequal sex ratio, natural selection,
etc., cf. Caballero 1994), the magnitude of the difference is still surprising. In
contrast, great ape species with geographically restricted ranges generally have
effective population sizes larger than humans (Fischer et al. 2006; Prado-Martinez
et al. 2013). This suggests that despite being able to colonize virtually the whole
world, our species must have had a much smaller population size (and likely a
more restricted geographic range) for much of its evolutionary history. Researchers
have generally considered models of recent exponential population growth from a
small initial population to explain the discrepancy between census and effective
population size (e.g., Slatkin and Hudson 1991; Marjoram and Donnelly 1994;
Gutenkunst et al. 2009; Gravel et al. 2011).
The initial analyses of genetic variation in human mitochondrial DNA (mtDNA)
estimated a relatively recent time to the most recent common ancestor (TMRCA)
and observed an excess of rare variants over equilibrium expectations (Cann et al.
1987; Vigilant et al. 1991). These data were then used to estimate a time of onset
of recent explosive population growth of 60–120 thousand years ago (Kya; Rogers
and Harpending 1992). However, mtDNA does not experience recombination so
operates as a single genetic locus. As such, the observed patterns of genetic variation
are sensitive to the effects of natural selection, and inferences of demographic
parameters from mtDNA data are inherently untrustworthy. When comparable
sequence polymorphism data was obtained from multiple nuclear regions, the
skew toward rare variants was much smaller, suggesting that substantial midPleistocene population growth was unlikely (Wall and Przeworski 2000; Voight
et al. 2005; Gutenkunst et al. 2009). Current estimates based on resequencing
data from thousands of individuals suggest that population growth started in the
late Pleistocene (20–25 Kya) and that the growth rate has accelerated (i.e., superexponential growth) in the past 5 thousand years (e.g., Coventry et al. 2010; Nelson
et al. 2012; Tennessen et al. 2012; Gazave et al. 2014; Chen et al. 2015; Gao
and Keinan 2016). This is consistent with the spread of agriculture (and a more
steady food supply) being the primary innovation that enabled our recent explosive
J. D. Wall
substantial Khoesan ancestry in South African Bantu speakers (Henn et al. 2011),
and substantial non-African ancestry in the East African Maasai (Wall et al. 2013).
8.3
Effective Population Size
The current human population size is in excess of seven billion and is thought
to have been in the millions for all of recorded history (Tellier 2009). However,
genetic sequencing studies have found that extant humans are quite similar to each
other, with the average sequence divergence between two individuals around 0.1%
(Li and Sadler 1991; Wall et al. 2008; 1000 Genomes Project Consortium 2010,
2012, 2015). This corresponds to an effective population size (i.e., time-averaged
number of breeding adults) of around 20,000–25,000, which is vastly less than our
estimates of the recent census size. While there are reasons why the census size
might be more than the effective population size (e.g., nonrandom mating, variance
in reproductive success across individuals, unequal sex ratio, natural selection,
etc., cf. Caballero 1994), the magnitude of the difference is still surprising. In
contrast, great ape species with geographically restricted ranges generally have
effective population sizes larger than humans (Fischer et al. 2006; Prado-Martinez
et al. 2013). This suggests that despite being able to colonize virtually the whole
world, our species must have had a much smaller population size (and likely a
more restricted geographic range) for much of its evolutionary history. Researchers
have generally considered models of recent exponential population growth from a
small initial population to explain the discrepancy between census and effective
population size (e.g., Slatkin and Hudson 1991; Marjoram and Donnelly 1994;
Gutenkunst et al. 2009; Gravel et al. 2011).
The initial analyses of genetic variation in human mitochondrial DNA (mtDNA)
estimated a relatively recent time to the most recent common ancestor (TMRCA)
and observed an excess of rare variants over equilibrium expectations (Cann et al.
1987; Vigilant et al. 1991). These data were then used to estimate a time of onset
of recent explosive population growth of 60–120 thousand years ago (Kya; Rogers
and Harpending 1992). However, mtDNA does not experience recombination so
operates as a single genetic locus. As such, the observed patterns of genetic variation
are sensitive to the effects of natural selection, and inferences of demographic
parameters from mtDNA data are inherently untrustworthy. When comparable
sequence polymorphism data was obtained from multiple nuclear regions, the
skew toward rare variants was much smaller, suggesting that substantial midPleistocene population growth was unlikely (Wall and Przeworski 2000; Voight
et al. 2005; Gutenkunst et al. 2009). Current estimates based on resequencing
data from thousands of individuals suggest that population growth started in the
late Pleistocene (20–25 Kya) and that the growth rate has accelerated (i.e., superexponential growth) in the past 5 thousand years (e.g., Coventry et al. 2010; Nelson
et al. 2012; Tennessen et al. 2012; Gazave et al. 2014; Chen et al. 2015; Gao
and Keinan 2016). This is consistent with the spread of agriculture (and a more
steady food supply) being the primary innovation that enabled our recent explosive