Parallel evolution of gene classes, but not genes: Evidence from Hawai'ian honeycreeper populations exposed to avian malaria

Abstract

Adaptation in nature is ubiquitous, yet characterizing its genomic basis is difficult because population demographics cause correlations with nonadaptive loci. Introduction events provide opportunities to observe adaptation over known spatial and temporal scales, facilitating the identification of genes involved in adaptation. The pathogen causing avian malaria, Plasmodium relictum, was introduced to Hawai'i in the 1930s and elicited extinctions and precipitous population declines in native honeycreepers. After a sharp initial population decline, the Hawai'i ‘amakihi (Chlorodrepanis virens) has evolved tolerance to the parasite at low elevations where P. relictum exists, and can sustain infection without major fitness consequences. High‐elevation, unexposed populations of ‘amakihi display little to no tolerance. To explore the genomic basis of adaptation to P. relictum in low‐elevation ‘amakihi, we genotyped 125 ‘amakihi from the island of Hawai'i via hybridization capture to 40,000 oligonucleotide baits containing SNPs and used the reference ‘amakihi genome to identify genes potentially under selection from malaria. We tested for outlier loci between low‐ and high‐elevation population pairs and identified loci with signatures of selection within low‐elevation populations. In some cases, genes commonly involved in the immune response (e.g., major histocompatibility complex) were associated with malaria presence in the population. We also detected several novel candidate loci that may be implicated in surviving malaria infection (e.g., beta‐defensin, glycoproteins and interleukin‐related genes). Our results suggest that rapid adaptation to pathogens may occur through changes in different immune genes, but in the same classes of genes, across populations.

Conservation of adaptive potential and functional diversity

“…In the case of the amphibian study, RNAseq data after experimental infection revealed the role of MHC super-types in eliciting a massive immune response in dying versus surviving leopard frogs, and directional selection in response to fungal infection was implicated in the reduction of MHC diversity in a related species (see Savage et al 2016Savage et al , 2017. For honeycreepers, genomic analyses identified candidate loci that may facilitate survival after malarial infection in the amakihi (see Cassin-Sackett et al 2018). In the special issue we include one example based on the genomic context of susceptibility of the Tasmanian devil to facial tumours (Hohenlohe et al this issue).…”

Phenotypic and genotypic parallel evolution in parapatric ecotypes ofSenecio

“…If knowledge of causal alleles is available, then one must demonstrate that a given site has been under repeated and independent selection in each replicate pair (Lee & Coop, 2017, 2019. Typically, these analyses are undertaken comparing individual replicate pairs (e.g., Roda et al, 2013b;Lamichhaney et al, 2017;Cassin-Sackett et al, 2019), although some studies detect broad, or 'global' outliers by comparing the aggregate of all populations within each ecotype (e.g., Jones et al, 2012;Kautt et al, 2012). In the absence of repeated selection on the same allele or gene, parallelism can be manifested at the functional level, such that different genes under selection participating in the same predicted biological function contribute to the pattern of phenotypic parallelism in a system.…”

“…In the absence of repeated selection on the same allele or gene, parallelism can be manifested at the functional level, such that different genes under selection participating in the same predicted biological function contribute to the pattern of phenotypic parallelism in a system. It is frequently asked whether predicted biological functions are consistently enriched across independent populations (Smith & Rausher, 2011;Kowalko et al, 2013;Roda et al, 2013b;Perreault-Payette et al, 2017;Cassin-Sackett et al, 2019). These analyses can inform us about how genotypically parallel a system is at different scales of divergence, so we can also view genotypic parallel evolution as broad-sense or narrow-sense ( Figure 1).…”

“…These results suggest that adaptation in S. lautus could be rather flexible and redundant at lower levels of organization (the nucleotide site and gene), and only parallel at the level of the biological function. Parallelism at the biological function level might be common within nature (e.g., Smith & Rausher, 2011;Kowalko et al, 2013;Roda et al, 2013b;Laporte et al, 2015;Perreault-Payette et al, 2017;Cassin-Sackett et al, 2019) given that there are fewer biological functions than there are genes or nucleotide sites (Tenaillon et al, 2012;Tiffin & Ross-Ibarra, 2014). In this regard, our study is rather conservative as we have only defined narrow-sense parallelism in function when the function was separately enriched in at least all nine, or eight population pairs.…”

“…The evolution of similar phenotypes within the same environment can arise via independent and repeated selection on the same nucleotide site or gene (reviewed in Wood et al, 2005;Christin et al, 2010;Stern, 2013). Similar phenotypes can also arise by selection on entirely different genes, although often from the same functional pathway (e.g., Smith & Rausher, 2011;Kowalko et al, 2013;Roda et al, 2013b;Laporte et al, 2015;Perreault-Payette et al, 2017;Cassin-Sackett et al, 2019). Here, different genetic routes are able to produce similar phenotypic outcomes across populations, even though selection does not act upon the same allele or gene.…”