Association mapping desiccation resistance within chromosomal inversions in the African malaria vector Anopheles gambiae

Abstract

Inversion polymorphisms are responsible for many ecologically important phenotypes and are often found under balancing selection. However, the same features that ensure their large role in local adaptation-especially reduced recombination between alternate arrangements-mean that uncovering the precise loci within inversions that control these phenotypes is unachievable using standard mapping approaches. Here, we take advantage of long-term balancing selection on a pair of inversions in the mosquito Anopheles gambiae to map desiccation tolerance via pool-GWAS. Two polymorphic inversions on chromosome 2 of this species (denoted 2La and 2Rb) are associated with arid and hot conditions in Africa and are maintained in spatially and temporally heterogeneous environments. After measuring thousands of wild-caught individuals for survival under desiccation stress, we used phenotypically extreme individuals homozygous for alternative arrangements at the 2La inversion to construct pools for whole-genome sequencing. Genomewide association mapping using these pools revealed dozens of significant SNPs within both 2La and 2Rb, many of which neighboured genes controlling ion channels or related functions. Our results point to the promise of similar approaches in systems with inversions maintained by balancing selection and provide a list of candidate genes underlying the specific phenotypes controlled by the two inversions studied here.

Keywords: aridity tolerance; balancing selection; local adaptation; recombination.

Figure 1.

Survival of teneral adult female An. gambiae under acute desiccation stress. Data are stratified by In(2La) karyotype. Dashed lines represent 50% survivorship.

Figure 2.

Experimental design. Mosquitoes collected from nature as larvae were individually phenotyped for dessication tolerance and karyotyped. Pools of female mosquitoes with alternate inversion arrangements and extreme phenotypes were sequenced together in replicate. Reads were mapped to both the reference (standard for In(2La)) and pseudo-reference (inverted for In(2La)) assemblies. Alleles at each polymorphic site are counted to provide estimates of allele frequencies. Between pools with the same karyotype but different phenotypes, contingency tables were created for each polymorphic site. Association mapping using the Cochran-Mantel-Haenszel (CMH) test combines contingency tables across karyotypes and replicates in order to test this global hypothesis at each polymorphic site.

Figure 3.

Manhattan plot of the association P-values for acute desiccation tolerance inside In(2La). Horizontal red line represents the In(2La)-wide significance threshold at FDR of 5% (P < 0.00027). Tested SNPs below that threshold shown as blue dots; green dots represent significant SNPs conferring high tolerance. The SNPs with the lowest P-values are numbered, and correspond to genomic coordinates closest to the following genes: (1) AGAP006026, (2) AGAP006961, AGAP006962, AGAP006963, (3) AGAP006785, AGAP006786, (4) AGAP006633, (5) AGAP006059.

Figure 4.

Manhattan plot of the association P-values for acute desiccation tolerance inside In(2Rb). Horizontal red line represents the In(2Rb)-wide significance threshold at FDR of 5% (P < 0.00092). Tested SNPs below that threshold shown as blue dots; green dots represent significant SNPs conferring high tolerance. The SNPs with the lowest P-values are numbered, and correspond to genomic coordinates closest to the following genes: (1) AGAP002578, (2) AGAP002444, (3) AGAP002372, (4) AGAP002744, AGAP013160, (5) AGAP002487.