Frequent Hybridisation Between Parapatric Lekking Bird-of-Paradise Species

Abstract

Hybridisation is known to occur between a wide range of taxa, including species for which strong sexual selection has led to markedly different sexual phenotypes and lek-mating behaviours. To what extent occasional hybridisation can overcome the reproductive barriers in such systems and, for example, lead to the establishment of hybrid zones is poorly known. In this study, we address this question by focusing on one of the most well-known avian radiations in which sexual selection has resulted in an extraordinary assemblage of phenotypic diversity and lek-mating behaviours: the birds-of-paradise (Paradisaeidae). We quantify the genome-wide distribution of introgression and find multiple signals of recent and historical gene flow between and within two genera of birds-of-paradise, Astrapia and Paradigalla. In addition, we present the first empirical genomic indication of a putative hybrid zone between two lekking bird-of-paradise species that differ substantially in their sexually selected traits and behaviours. Our findings are consistent with the idea that behavioural and phenotypic traits may constitute weaker pre- and post-zygotic barriers to gene flow than generally thought in lek-mating species.

Keywords: birds‐of‐paradise; hybrid zone; hybridisation; introgression; lekking; museomics.

FIGURE 1

Comparison between mitochondrial tree (left) and autosomal summary coalescent tree for 5 kbp windows (right). Individuals (individuals 21, 22, 23, 24, 25 and 26) of A. mayeri and A. stephaniae show signs of recent hybridisation and are indicated in bold. Both phylogenetic trees are rooted with the clade Lycocorax, Manucodia and Phonygammus (black dots). Clades with full bootstrap support marked with 1 in mitochondrial phylogenetic tree. Clades marked with bootstrap/gene concordance factor/site concordance factor in the autosomal phylogenetic tree. Illustrations by Szabolcs Kókay.

FIGURE 2

(A) Admixture for K3 to K7 using NGSadmix displaying signs of contemporary hybridisation. Sample identifiers are provided in Table S1. The samples of A. mayeri and A. stephaniae are ordered along a longitudinal gradient from west to east. (B) PCA of PC axis 1 and 2 calculated with PCAngsd displaying relationships between species. (C) Location of samples, higher resolution and individually marked in Figures S1 and S3. (D) Fbranch statistic is calculated using Dsuite indicating historical gene flow.

FIGURE 3

Locations and admixture proportions of A. mayeri and A. stephaniae samples. The samples are indexed after the order they appear in Figure 2. The sample locations of A. mayeri (square) and A. stephaniae (cross) with A. stephaniae samples indexed. (A) The putative hybrid zone is shown in the enlarged map and samples are plotted overlaying elevation at a resolution of 1 km². The A. stephaniae samples excluded in the enlarged map (27, 28, 29, 30 and 31) did not show any indication of recent admixture. (B) Principal component analysis of A. mayeri (square) and A. stephaniae (cross). Illustrations of A. stephaniae and A. mayeri . painted by Szabolcs Kókay.

FIGURE 4

Admixture proportions across different regions of the chromosomes in hybrids between A. mayeri and A. stephaniae resulting from at least one backcross between a hybrid and A. stephaniae . High values indicate regions of high A. mayeri content, which has been preserved after backcrossing with A. stephaniae . A blue “x” indicates a specific region of the chromosomes where the parental species' reference panel display admixture between each other. These regions can therefore not be separated from regions of high likelihood of introgression and incomplete lineage sorting in the hybrids.

FIGURE 5

Introgression trends in windows distributed across chromosomes in backcrossing hybrids between A. mayeri and A. stephaniae . (A) Median admixture proportion across the five backcrossing hybrids in 9 windows distributed across large chromosomes (> 40Mbp), (B) Median admixture proportion across the five backcrossing hybrids in 7 windows distributed across intermediate chromosomes (< 40 Mbp and > 20Mbp), (C) Median admixture proportion across the five backcrossing hybrids in 5 windows distributed across micro chromosomes (< 20Mbp). (D) Chromosome cartoon displaying relative size of chromosomes. Admixture proportion in windows, white low and red high. Windows not separable form incomplete lineage sorting filled with blue and are excluded from the box plots.