Experimental swap of Anopheles gambiae's assortative mating preferences demonstrates key role of X-chromosome divergence island in incipient sympatric speciation

Abstract

Although many theoretical models of sympatric speciation propose that genes responsible for assortative mating amongst incipient species should be associated with genomic regions protected from recombination, there are few data to support this theory. The malaria mosquito, Anopheles gambiae, is known for its sympatric cryptic species maintained by pre-mating reproductive isolation and its putative genomic islands of speciation, and is therefore an ideal model system for studying the genomic signature associated with incipient sympatric speciation. Here we selectively introgressed the island of divergence located in the pericentric region of the X chromosome of An. gambiae s.s. into its sister taxon An. coluzzii through 5 generations of backcrossing followed by two generations of crosses within the introgressed strains that resulted in An. coluzzii-like recombinant strains fixed for the M and S marker in the X chromosome island. The mating preference of recombinant strains was then tested by giving virgin recombinant individuals a choice of mates with X-islands matching and non-matching their own island type. We show through genetic analyses of transferred sperm that recombinant females consistently mated with matching island-type males thereby associating assortative mating genes with the X-island of divergence. Furthermore, full-genome sequencing confirmed that protein-coding differences between recombinant strains were limited to the experimentally swapped pericentromeric region. Finally, targeted-genome comparisons showed that a number of these unique differences were conserved in sympatric field populations, thereby revealing candidate speciation genes. The functional demonstration of a close association between speciation genes and the X-island of differentiation lends unprecedented support to island-of-speciation models of sympatric speciation facilitated by pericentric recombination suppression.

Fig 1. Percentage assortative mating in females and males carrying M or S-type X-chromosome islands.

A) virgin females or males from the RbMM and RbSS recombinant strains where presented with a mixture of recombinant individuals of the opposite sex and with X-islands matching and non-matching their own X-island molecular type; B) virgin females or males from the M-form Mopti and S-form Kisumu strains used for creating the recombinant lines were given a choice between equal numbers of potential mates from both strains. The number and percentage of assortative and disassortative mating recorded across 3 replicates are indicated.

Fig 2. Genomic structure of recombinant strains.

The genomes of the assortatively-mating RbMM, RbSS and parental Mopti strains were compared using F _ST estimates at ~3x10⁶ SNP marker loci (left Y-axis and red, blue and black lines). The genomic region introgressed from Kisumu into the Mopti genetic background and differing between the RbMM and RbSS recombinant strains is characterized by high F _ST values (blue shade) and extends from position ~14.5Mb to the centromere on chromosome X. The RbMM and RbSS differed at 160 protein-changing positions all of which located within the introgressed island and flanking region (right Y-axis, grey histogram bars). The pericentromeric region sharing conserved fixed differences with the field Anopheles coluzzii and gambiae s.s populations starts at position ~18.1Mb (orange shade). The position of inversions c, u and a on chromosome 2 is indicated (pink shade).

Fig 3. Field population frequency distribution of protein coding SNPs identified in RbSS.

A region covering the X-island and flanking region up to reference position 17Mbp was captured and re-sequenced in sympatric An. gambiae s.s. and An. coluzzii populations from Ghana. The frequency of alleles coding for unique protein differences in the RbSS recombinant strain was measured in the field An. gambiae s.s. population. The proportion of alleles occurring at high 0.8 (orange bars) and very high (freq >0.95: red bars) frequency increased towards the centromere suggesting a potential role in speciation whilst other alleles (blue bars) were not conserved (see text for details).

Fig 4. Genetic crossing design used for selective introgression of the X-island of divergence in recombinant strains.

Following the creation of hybrid females at the X-island diagnostic rDNA locus, 4 generations of backcrosses were used to introgress the S-type X-island into an M molecular form Mopti genetic background. Here, 'MM', 'MS' and 'SS' refer to the female genotypes at the rDNA marker locus in the X-island and 'M' and 'S' refers to the male genotype at the same locus (see methods). Thereafter 2 generations of crosses within the introgressed strain resulted in recombinants strains that shared a high genetic identity but differed at the S or M-type X-chromosome islands of divergence.