9 Natural Selection, Genetic Variation, and Human Diversity
227
at hitchhiking neutral sites as long as migration rates are not too high (Fig. 9.3d)
(Slatkin and Wiehe 1998).
The connection between phenotypic similarities and population relatedness
becomes even more tenuous when discussing physical, rather than molecular,
phenotypes. One notable example is adaptation to high-altitude living (Scheinfeldt
and Tishkoff 2010). Although the concentration of atmospheric oxygen is the same
at high altitudes, lower atmospheric pressure means that the partial pressure of
oxygen is too low to induce gas exchange in the lungs. Without adaptation, longterm high-altitude living would be impossible. Populations living in the Andes,
the Tibetan Plateau, and the Ethiopian Highlands all exhibit a lack of hypoxia
at high altitudes. This shared phenotype is the result of convergent evolution and
could be expected to cause genetic similarities between these populations that
are otherwise distantly related. When the phenotype is examined more closely on
a molecular level, however, Andeans, Ethiopian Highlanders, and Tibetans may
have different hemoglobin and oxygen saturation levels (Beall et al. 2002; HuertaSanchez et al. 2013). Closer examination shows that even populations exhibiting
similar molecular phenotypes have adaptations at different genes involved in the
same pathway (the hypoxia-inducible factor 1, or HIF-1 pathway) (Bigham et
al. 2009; Xu et al. 2011; Scheinfeldt et al. 2012; Huerta-Sanchez et al. 2013).
Clearly, even detailed physiological phenotypes are not always a reliable indicator
of population differentiation due to positive selection.
While much attention has been focused on visible phenotypic differentiation
between populations (e.g., hair and skin pigmentation, hair texture, stature, and
facial features), this focus is almost entirely unwarranted. Genome-wide scans for
the targets of positive selection have detected the signatures of selection in genes
associated with visible phenotypes—notably skin pigmentation via SLC24A5 and
hair texture at EDAR—but these signals are far outnumbered by those of other pathways, such as metabolic processes, immune responses, and olfaction (Akey 2009).
While it is not entirely clear how much of human genetic variation is represented
in visible phenotypic differences, we can speculate that the majority of molecular
differences have no immediately visible physical manifestation. Therefore, the
majority of population differentiation, whether due to selection or not, is likely to
be found in specific molecular phenotypes that require detailed measurements to
detect. Furthermore, the amount of population differentiation seen at selected loci
is extreme compared to the rest of the genome and should not be used to draw
sweeping conclusions about population differences. The temptation to conjecture
about the meaning and significance of these selected population differences is
high, but there is potential for such conjectures to be misused for harmful and
discriminatory purposes (Vitti et al. 2012). Researchers should therefore interpret
population adaptations with restraint and the use of appropriate supporting data.
227
at hitchhiking neutral sites as long as migration rates are not too high (Fig. 9.3d)
(Slatkin and Wiehe 1998).
The connection between phenotypic similarities and population relatedness
becomes even more tenuous when discussing physical, rather than molecular,
phenotypes. One notable example is adaptation to high-altitude living (Scheinfeldt
and Tishkoff 2010). Although the concentration of atmospheric oxygen is the same
at high altitudes, lower atmospheric pressure means that the partial pressure of
oxygen is too low to induce gas exchange in the lungs. Without adaptation, longterm high-altitude living would be impossible. Populations living in the Andes,
the Tibetan Plateau, and the Ethiopian Highlands all exhibit a lack of hypoxia
at high altitudes. This shared phenotype is the result of convergent evolution and
could be expected to cause genetic similarities between these populations that
are otherwise distantly related. When the phenotype is examined more closely on
a molecular level, however, Andeans, Ethiopian Highlanders, and Tibetans may
have different hemoglobin and oxygen saturation levels (Beall et al. 2002; HuertaSanchez et al. 2013). Closer examination shows that even populations exhibiting
similar molecular phenotypes have adaptations at different genes involved in the
same pathway (the hypoxia-inducible factor 1, or HIF-1 pathway) (Bigham et
al. 2009; Xu et al. 2011; Scheinfeldt et al. 2012; Huerta-Sanchez et al. 2013).
Clearly, even detailed physiological phenotypes are not always a reliable indicator
of population differentiation due to positive selection.
While much attention has been focused on visible phenotypic differentiation
between populations (e.g., hair and skin pigmentation, hair texture, stature, and
facial features), this focus is almost entirely unwarranted. Genome-wide scans for
the targets of positive selection have detected the signatures of selection in genes
associated with visible phenotypes—notably skin pigmentation via SLC24A5 and
hair texture at EDAR—but these signals are far outnumbered by those of other pathways, such as metabolic processes, immune responses, and olfaction (Akey 2009).
While it is not entirely clear how much of human genetic variation is represented
in visible phenotypic differences, we can speculate that the majority of molecular
differences have no immediately visible physical manifestation. Therefore, the
majority of population differentiation, whether due to selection or not, is likely to
be found in specific molecular phenotypes that require detailed measurements to
detect. Furthermore, the amount of population differentiation seen at selected loci
is extreme compared to the rest of the genome and should not be used to draw
sweeping conclusions about population differences. The temptation to conjecture
about the meaning and significance of these selected population differences is
high, but there is potential for such conjectures to be misused for harmful and
discriminatory purposes (Vitti et al. 2012). Researchers should therefore interpret
population adaptations with restraint and the use of appropriate supporting data.