Widespread Adaptive Introgression of Major Histocompatibility Complex Genes across Vertebrate Hybrid Zones

Abstract

Interspecific introgression is a potentially important source of novel variation of adaptive significance. Although multiple cases of adaptive introgression are well documented, broader generalizations about its targets and mechanisms are lacking. Multiallelic balancing selection, particularly when acting through rare allele advantage, is an evolutionary mechanism expected to favor adaptive introgression. This is because introgressed alleles are likely to confer an immediate selective advantage, facilitating their establishment in the recipient species even in the face of strong genomic barriers to introgression. Vertebrate major histocompatibility complex genes are well-established targets of long-term multiallelic balancing selection, so widespread adaptive major histocompatibility complex introgression is expected. Here, we evaluate this hypothesis using data from 29 hybrid zones formed by fish, amphibians, squamates, turtles, birds, and mammals at advanced stages of speciation. The key prediction of more extensive major histocompatibility complex introgression compared to genome-wide introgression was tested with three complementary statistical approaches. We found evidence for widespread adaptive introgression of major histocompatibility complex genes, providing a link between the process of adaptive introgression and an underlying mechanism. Our work identifies major histocompatibility complex introgression as a general mechanism by which species can acquire novel, and possibly regain previously lost, variation that may enhance defense against pathogens and increase adaptive potential.

Keywords: MHC; adaptation; host–pathogen coevolution; hybridization; introgression.

Fig. 1.

Study design. a) Time-calibrated phylogeny of the species studied; b) the geographic extent of genome wide and MHC introgression expected under adaptive MHC introgression, along with the division of species ranges into areas classified as “near” and “far” from the contact zone; c) the approaches used to compare genome-wide and MHC introgression.

Fig. 2.

The number of MHC alleles per individual in the studied species. These results show clear differences in the complexity between MHC classes and between taxa. Note that we use term “allele” for each unique sequence variant, although these variants may originate from different MHC genes. Boxplots show the median (horizontal line), the first and third quartile (box), and the range (whiskers). Different MHC markers were used for Clethrionomys (see Materials and Methods), so estimates of MHC variation for this genus are not directly comparable with those for other taxa. Bom., Bombina; Lis., Lissotriton; Sph., Sphyrapicus; Cle., Clethrionomys.

Fig. 3.

Interspecific MHC allele sharing was elevated in the vicinity of contact but beyond the actual hybrid zone (“near”) compared to the areas further from the contact (“far”). a) Allele sharing between species near and far from the contact zone; b) SESs (Cohen's d) of the difference in allele sharing between species near and far from the contact zone and the PGLS model estimates with 95% confidence intervals; Sol., Solea; Bom., Bombina; Ang., Anguis; Cle., Clethrionomys.

Fig. 4.

Comparison of MHC and genome-wide a) geographic and b) genomic clines indicates more extensive introgression of both MHC classes compared to the genome-wide introgression. For each hybrid zone and MHC class, points show effect size estimates with 95% confidence intervals (a) or the ML parameter estimates with 2 log-likelihood ranges (b). S., Solea; B., Bombina; L., Lissotriton; T., Triturus; A., Anguis; N., Natrix; Sph., Sphyrapicus; PGLS estimates, phylogenetic generalized least squares estimates together with 95% confidence intervals; higher/lower div. sp., species in a pair exhibiting higher/lower MHC diversity. Note that uncertainty in parameter estimates for individual taxa is incorporated as weights in PGLS models, so confidence intervals of PGLS estimates should not be interpreted in the context of uncertainty in individual estimates.

Fig. 5.

Comparison of the MHC and genome-wide geographic clines. Light-colored bands show 2 log-likelihood intervals, while points correspond to populations.

Fig. 6.

Genomic clines. Light-colored bands show 2 log-likelihood intervals, while points correspond to individuals.