Evolution and genetic architecture of sex-limited polymorphism in cuckoos

Abstract

Sex-limited polymorphism has evolved in many species including our own. Yet, we lack a detailed understanding of the underlying genetic variation and evolutionary processes at work. The brood parasitic common cuckoo (Cuculus canorus) is a prime example of female-limited color polymorphism, where adult males are monochromatic gray and females exhibit either gray or rufous plumage. This polymorphism has been hypothesized to be governed by negative frequency-dependent selection whereby the rarer female morph is protected against harassment by males or from mobbing by parasitized host species. Here, we show that female plumage dichromatism maps to the female-restricted genome. We further demonstrate that, consistent with balancing selection, ancestry of the rufous phenotype is shared with the likewise female dichromatic sister species, the oriental cuckoo (Cuculus optatus). This study shows that sex-specific polymorphism in trait variation can be resolved by genetic variation residing on a sex-limited chromosome and be maintained across species boundaries.

Fig. 1.. Genetic structure and mapping of female plumage polymorphism in a single population of C. canorus.

(A) Representation of female gray (top) and rufous (bottom) common cuckoo plumage morphs and sampling location at Apaj village, Hungary. Map includes sample sizes from each morph. (B) PCA using whole-genome sequencing data showing no genetic structure between Hungarian canorus morphs. Individuals are colored by phenotype, and the percentage of variance explained by each component is given in parentheses. (C) Genome-wide association analysis. Each dot represents the −log₁₀ transformation of likelihood ratio test (LRT) P values based on a 1 df chi-square distribution, per variant site. (D) Genetic differentiation between morphs across the genome. Each dot represents the fixation index (F_ST) averaged in 50-kb non-overlapping windows. In (C) and (D), the significance threshold after genome-wide Bonferroni correction (P = 2.77 × 10⁻⁹) and the top 0.1% of the empirical distribution are shown by a red dashed line, respectively.

Fig. 2.. Physical and biochemical characterization of plumage color in two cuckoo species.

(A) Photos depicting the two female plumage morphs of both species studied (common cuckoo, C. canorus; oriental cuckoo, C. optatus) (left). Feather insets show measured sites for both spectroscopic analyses (right). (B) Spectrophotometric measurements (percent reflectance) obtained from wing covert feathers of rufous and gray morphs for each species. (C) Raman spectra of the same feathers used for spectrophotometry. Diagnostic intensity peaks for eumelanin and pheomelanin identified in each spectrum of dark and light bands, respectively. (B) and (C) In rufous feathers, light and dark bands were measured separately to better account for patterning. Photo credits: Imran shah (canorus gray and rufous, licensed under CC BY-SA 2.0), ming110 (optatus gray, licensed under CC BY-NC 4.0), and Ged Tranter (optatus rufous, ML193641471). a.u., arbitrary units.

Fig. 3.. W chromosome–encoded female-limited polymorphism predates speciation of C. canorus and C. optatus.

(A) Maximum-likelihood phylogeny of the W chromosome (n = 53,712 biallelic SNPs) rooted along the truncated outgroup branch of the lesser cuckoo (C. poliocephalus), further including two specimens of another outgroup species (C. micropterus). Nodes with SH-like approximate likelihood ratio test (SH-aLRT; > 99%) support are marked with a diamond. (B) Autosomal ancestry coefficients (n = 16,031,301 biallelic SNPs) with individuals (x axis) displayed in the same order as the phylogeny. (C) Jittered sampling distribution with insets for two localities with higher sampling density. (D) First principal component of autosomal (left) and W chromosome (right) genetic variation, with PC scores scaled from 0.0 to 1.0 (outgroups C. poliocephalus and C. micropterus excluded). Because of deviations from normality, two Wilcoxon rank sum tests were conducted on each axis to evaluate significant differences between species and plumage morphs, denoted below each axis after Bonferroni correction. The eigenvalue proportion of variance explained is indicated adjacent to each PC label. *P < 0.05; ns, not significant.

Fig. 4.. Evolutionary history of the W encoded female-limited polymorphism.

(A) Cartoon depicting strategy for calculating age in generations since divergence using genetic variation within each population (π) and sequence divergence between them (D_XY) to estimate D_a in conjunction with a generational mutation rate (μ_generation = 1.01 × 10⁻⁰⁸). Divergence was calculated for both species split and plumage color contrast. (B) Relative ratio of plumage to species divergence for autosomal and W chromosomes. Mean and 95% CIs were generated by calculating the mean D_a(Plumage)/D_a(Species) with 1000 bootstrap resampling events. Dashed line indicates a value of 1.0 where divergence between plumage morphs is equal to divergence between species. (C) Autosomal D-statistics from ABBA-BABA analyses. Mean and SD D-statistics were obtained by jackknifing all autosomal data. The analysis was run once with gray canorus as the target population (P3) and again with rufous canorus as the target (P3) to examine excess allele sharing with rufous optatus (P2), using canorus from the Hungarian population. (D) Tajima’s D in 50-kb windows for autosomal and W chromosome data within the canorus Hungarian dataset, with violin plots showing all windows with 25 and 75% quartiles indicated. (E) Proportion of genome-wide topologies exhibiting majority support (≥50%) for either the Species, Plumage, or “other ILS” tree with 100-SNP windows. All topologies other than the species tree and plumage tree were categorized as other ILS trees. Bootstrapped mean and 95% CIs are indicated.