3.2
Frequencies of lactase persistence
3.2.1
Allelic and genotypic frequencies −13,910*T in all samples
A total of 437 of the 451 samples collected were successfully sequenced, of which 13 corres- pond to samples collected in the Elqui valley, 223 in the Limari valley and 206 in the Choapa valley. From the examination of chromatograms it was possible to determine absence of unrepor- ted polymorphic regions and the presence of the European −13,910*T in all three groups as the only lactase–persistence associated variant found.
The frequencies of −13,910 genotypes of these samples are shown in Table 3.2. From these, pre- dicted frequencies of lactase persistence were calculated.
Valley CC CT TT Total (n) *T freq. Predicted Digest. (%)
Elqui 10 2 1 13 0.15 23
Limarí 135 73 15 223 0.23 39
Choapa 128 65 8 201 0.2 36
Total 273 140 24 437 0.22 38
Table 3.2.Frequencies of −13,910 genotypes in Norte Chico’s Agricultural Communities.
These results are in agreement with HWE expectations, and show sufficient predicted digesters (38%) to make comparisons between persistent and non–persistent groups.
Assuming that no other evolutionary factors are in place, any selection coefficient higher than 0.094 acting over the current generation would have resulted in significant (p > 0.05) deviation from HWE given this sample size (n = 437)3. If there is any selection acting in this population in
the current generation, selection coefficient is not likely to be between 0.09 – 0.19 as estimated by Bersaglieri et al. (2004) for neolithic northern Europe based in current frequencies of LP and the allotted time since animal domestication.
3.2.2
Distributions of frequencies by groups
3.2.2.1 GeographyTo examine geographic distribution, genotype and allele frequencies were calculated for the dif- ferent locations (see section 2.2 for details). and are presented in Table 3.3.
Geographic distribution of these frequencies was evaluated by the method of kernel density es- timation (Figure 3.1), as has been done in previous studies to estimate worldwide frequencies of 3HWE power to detect selection in the current generation depends on a high selection coefficient (s) and a large sample
size (n). In a model of complete dominance, the null hypothesis is rejected only if s ≥ pχ2/n, where χ2is the value of
3.2. Frequencies of lactase persistence 88
Location CC CT TT Total (n) *T freq. Predicted Digest. (%)
Puclaro 10 2 1 13 0.15 23 Canelilla 16 6 1 23 0.17 30 Monte Patria 24 10 5 39 0.26 38 Barraza 22 16 3 41 0.27 46 Espinal 7 5 2 14 0.32 50 Calera 30 14 0 44 0.16 32 Mal Paso 36 22 4 62 0.24 42 Canela 106 48 6 160 0.19 34 Huentelauquén 22 17 2 41 0.26 46 Total 273 140 24 437 0.22 38
Table 3.3.Frequencies of −13,910 genotypes by Village.
lactase persistence variants (Gallego Romero et al., 2012; Ingram et al., 2009a; Itan et al., 2010, 2009). Interpolation was done using the R package “spatstat” (Baddeley & Turner, 2005).
0.16 0.18 0.2 0.22 0.24 Barraza Canela Canelilla Mal Paso Espinal Huentelauquen Calera Puclaro Monte Patria
Figure 3.1.Interpolation map showing distribution of −13,910*T based on the method of kernel density estimation calculated from the frequencies of this allele in each of the nine villages from which data were collected.
3.2. Frequencies of lactase persistence 89 male female 0.0 0.2 0.4 0.6 CC CT TT CC CT TT Genotype -13910 Frequency Genotype CC CT TT
Figure 3.2.Frequencies of −13,910 genotypes by sex.
cline corresponding to a increment of frequencies with longitude (see Figure 1.4). However, these correlations are not significant (p-value = 0.1) tested as a linear multiple regression model com- bining both latitude and longitude4.
The frequencies differed from location to location, but did not show differences in HWE. Only Monte Patria shows small deviation (χ2 = 4.19, p-value = 0.04), with fewer heterozygotes than
expected. Population differentiation (Fisher’s exact test) was not significant for all populations (genotypic p-value = 0.4215, allelic p-value = 0.366), nor for any pair of populations. Accordingly, Fstwas small (= 0.0007), and pairwise Fst not significant for any pair of populations. No significant
correlation was found between pairwise Fst and geographic distances between villages (Mantel
test p-value = 0.942).
3.2.2.2 Sex, age, and birth place of grandparents
Genotypic and allelic frequencies of −13,910 were compared next between sexes, age groups, and number of grandparents born outside the Coquimbo Region.
The null hypothesis of homogeneity of the groups cannot be rejected for any of these factors. Grouped by sex (Figure 3.2), there is no differentiation (Fisher’s method) in either genotypic (p- value= 0.734) or allelic (p-value = 0.727) frequencies.
Following Mead et al. (2008), comparison of age groups was used to explore historical trends acting as selective pressures in particular generations. No significant differences were found using four age groups (Figure 3.3, genotypic p-value = 0.226 , allelic p-value = 0.193, Fst = 0.003), and all
groups are in agreement with HWE expectations.
At a pairwise level, no significant differences were found for any pair of age groups using both 4digest
3.2. Frequencies of lactase persistence 90 <25 25-49 50-75 >75 0.0 0.2 0.4 0.6 CC CT TT CC CT TT CC CT TT CC CT TT Genotype -13910 Frequency Genotype CC CT TT
Figure 3.3.Frequencies of −13,910 genotypes by age group.
Fisher’s method and Fst, yet differences were slightly higher between the youngest and oldest
groups (Pairwise Fisher’s genotypic p-value = 0.076, Pairwise Fisher’s allelic p-value = 0.088, Pair- wise Fst = 0.029). However, this difference is not shown when the analysis is defined for two groups
only (born before and after 1960).
Birth places of the grandparents of each participant were recorded on questionnaires, and missing data (20%) were handled using multiple imputation as implemented by the R package “Amelia” (Honaker et al., 2011). No significant differences or significant deviations from HWE were found according to the number of grandparents born outside the Coquimbo Region (Figure 3.4, genotypic p-value = 0.735, allelic p-value = 0.731). As most people in the dataset have no grandparents born outside the region (316 out of 437), differentiation tests were also used to compare this group and all the others together (i.e. all grandparents born in the region versus at least one grandparent born outside), yet no significant differences were found.
none one two three all
0.0 0.2 0.4 0.6 CC CT TT CC CT TT CC CT TT CC CT TT CC CT TT Genotype -13910 Frequency Genotype CC CT TT