The genetic basis of the kākāpō structural color polymorphism suggests balancing selection by an extinct apex predator

Abstract

The information contained in population genomic data can tell us much about the past ecology and evolution of species. We leveraged detailed phenotypic and genomic data of nearly all living kākāpō to understand the evolution of its feather color polymorphism. The kākāpō is an endangered and culturally significant parrot endemic to Aotearoa New Zealand, and the green and olive feather colorations are present at similar frequencies in the population. The presence of such a neatly balanced color polymorphism is remarkable because the entire population currently numbers less than 250 birds, which means it has been exposed to severe genetic drift. We dissected the color phenotype, demonstrating that the two colors differ in their light reflectance patterns due to differential feather structure. We used quantitative genomics methods to identify two genetic variants whose epistatic interaction can fully explain the species' color phenotype. Our genomic forward simulations show that balancing selection might have been pivotal to establish the polymorphism in the ancestrally large population, and to maintain it during population declines that involved a severe bottleneck. We hypothesize that an extinct apex predator was the likely agent of balancing selection, making the color polymorphism in the kākāpō a "ghost of selection past."

Fig 1. The kākāpō feather color polymorphism.

(A) Green (left) and olive (right) kākāpō individuals; the photographs show the kākāpō individuals Uri (left) and Bravo (right) who are full siblings. The inserts show the standardized photography that was applied for color assessment across the extant kākāpō population (Material and Methods). (B) Manhattan plot of the mixed-model GWAS between all bi-allelic SNPs and the binary color polymorphism phenotype (Material and Methods); the horizontal line indicates the genome-wide significance line after Bonferroni multiple testing correction. The genome-wide significant hit at the end of chromosome 9 represents a single SNP, which we therefore discarded as noise. Inlet: Zoom-in of the Manhattan plot on the end of chromosome 8 (from 6.25 × 10⁷ bp to the end of the chromosome); the two most significant SNPs Chr8_63055688 and Chr8_63098195 are highlighted in orange color. All photographs are originals taken by the authors of this manuscript. The code to generate this figure can be found in https://zenodo.org/records/13302801. For data see the “Data and code availability” section. GWAS, genome-wide association study; SNP, single-nucleotide polymorphism.

Fig 2. Optical analyses of kākāpō feathers of both color polymorphisms.

(A) SEM images of 3 green (left 3 columns) and 3 olive (right 3 columns) feather barbules at different resolutions (rows); the green feathers show a smoother surface than the olive ones. (B) Photoreflectometry of near-infrared/visible wavelengths in the feather tips; the relative reflectance of an exemplary olive and green is plotted over the wavelength of the reflected light (Material and Methods). All photographs are originals taken by the authors of this manuscript. The data underlying this figure can be found in https://zenodo.org/records/13302801. SEM, scanning electron microscopy.

Fig 3. Genomic forward simulations of color polymorphism evolution.

The simulations assume (A) a positive selection model, where the novel olive phenotype had a selective advantage ranging between 1% and 20%, and (B) a balancing selection model (i.e., NFDS), where being the rarest phenotype had a selective advantage ranging between 1% and 20%. We ran 10,000 replicates per scenario: establishment of color polymorphism for a range of selection coefficients (0–0.2). (C) Demographic history of the kākāpō population as reconstructed with PSMC [25] and GONE [26] (Material and Methods). The vertical dashed line marks the time point of 40 generations when the 2 natural avian predators of the kākāpō went extinct. (D) Probability of the simulated color polymorphism being balanced as observed in the empirical data of the extant population under three competing scenarios: (1) NFDS is removed for all 1,000 generations, (2) NFDS is removed for the last 40 generations, or (3) same as scenario 2 plus genetic load accumulation. The dashed red line marks the probability threshold of 5% (Material and Methods). The data underlying this figure can be found in https://zenodo.org/records/13302801. NFDS, negative frequency-dependent selection.