Genomic Analyses of Human European Diversity at the Southwestern Edge: Isolation, African Influence and Disease Associations in the Canary Islands

Abstract

Despite the genetic resemblance of Canary Islanders to other southern European populations, their geographical isolation and the historical admixture of aborigines (from North Africa) with sub-Saharan Africans and Europeans have shaped a distinctive genetic makeup that likely affects disease susceptibility and health disparities. Based on single nucleotide polymorphism array data and whole genome sequencing (30×), we inferred that the last African admixture took place ∼14 generations ago and estimated that up to 34% of the Canary Islander genome is of recent African descent. The length of regions in homozygosis and the ancestry-related mosaic organization of the Canary Islander genome support the view that isolation has been strongest on the two smallest islands. Furthermore, several genomic regions showed significant and large deviations in African or European ancestry and were significantly enriched in genes involved in prevalent diseases in this community, such as diabetes, asthma, and allergy. The most prominent of these regions were located near LCT and the HLA, two well-known targets of selection, at which 40‒50% of the Canarian genome is of recent African descent according to our estimates. Putative selective signals were also identified in these regions near the SLC6A11-SLC6A1, KCNMB2, and PCDH20-PCDH9 genes. Taken together, our findings provide solid evidence of a significant recent African admixture, population isolation, and adaptation in this part of Europe, with the favoring of African alleles in some chromosome regions. These findings may have medical implications for populations of recent African ancestry.

Fig. 1.

Plot of the first two principal components (explaining 62.5% of variability) from PCA of Canary Islanders and samples from reference populations from Europe, North Africa, and sub-Saharan Africa. The inset depicts a detailed view per island. Results are based on a subsample of 100,175 SNPs excluding those in high LD (pairwise r² threshold = 0.5).

Fig. 2.

ADMIXTURE estimates for K = 4 for Canary Islanders and samples from reference populations from Europe, North Africa, and sub-Saharan Africa.

Fig. 3.

Average length (Mb) of ROHs using two classifications into categories in the populations from the Canary Islands.

Fig. 4.

Average number of genome regions in ROHs with respect to the average total (top), ≤1.6 Mb (middle), and >1.6 Mb (bottom) lengths of ROHs per island.

Fig. 5.

Triangle plot of individual genomic admixture proportions in Canary Islanders as estimated by ADMIXTURE with K = 4 (red), ELAI (blue), and LAMP-LD (orange).

Fig. 6.

Inference of local ancestry by ELAI and LAMP-LD. The plot shows an example of inference in a chromosome region (one panel of each parental population) comparing ELAI (blue) with LAMP-LD (green) allele dosages.

Fig. 7.

Genome-wide Z-score scan of ELAI local admixtures in Canary Islanders for NAF (top), EUR (middle), and SSA (bottom). Horizontal broken lines (blue > |2|; red > |3|) indicate score thresholds.

Fig. 8.

Enrichment analysis on regions with large deviations for any ancestry. Top ten significantly enriched human diseases (left) and MSigDB pathways (right).