Insights into the evolution of Darwin's finches from comparative analysis of the Geospiza magnirostris genome sequence

Abstract

Background: A classical example of repeated speciation coupled with ecological diversification is the evolution of 14 closely related species of Darwin's (Galápagos) finches (Thraupidae, Passeriformes). Their adaptive radiation in the Galápagos archipelago took place in the last 2-3 million years and some of the molecular mechanisms that led to their diversification are now being elucidated. Here we report evolutionary analyses of genome of the large ground finch, Geospiza magnirostris.

Results: 13,291 protein-coding genes were predicted from a 991.0 Mb G. magnirostris genome assembly. We then defined gene orthology relationships and constructed whole genome alignments between the G. magnirostris and other vertebrate genomes. We estimate that 15% of genomic sequence is functionally constrained between G. magnirostris and zebra finch. Genic evolutionary rate comparisons indicate that similar selective pressures acted along the G. magnirostris and zebra finch lineages suggesting that historical effective population size values have been similar in both lineages. 21 otherwise highly conserved genes were identified that each show evidence for positive selection on amino acid changes in the Darwin's finch lineage. Two of these genes (Igf2r and Pou1f1) have been implicated in beak morphology changes in Darwin's finches. Five of 47 genes showing evidence of positive selection in early passerine evolution have cilia related functions, and may be examples of adaptively evolving reproductive proteins.

Conclusions: These results provide insights into past evolutionary processes that have shaped G. magnirostris genes and its genome, and provide the necessary foundation upon which to build population genomics resources that will shed light on more contemporaneous adaptive and non-adaptive processes that have contributed to the evolution of the Darwin's finches.

Figure 1

Evolutionary mechanisms for beak shape diversity in Darwin’s finches (Thraupidae, Passeriformes). (A) Molecular phylogeny of 14 species of Darwin’s finches shows a range of beak shapes in this group of birds. These species have beaks of different shapes that allow them to feed on many different diets: insects, seeds, berries, and young leaves. Species are numbered as follows: small ground finch Geospiza fuliginosa; medium ground finch G. fortis; large ground finch G. magnirostris; cactus finch G. scandens; large cactus finch G. conirostris; sharp-billed finch G. difficilis; small tree finch C. parvulus; large tree finch Camarhynchus psittacula; medium tree finch C. pauper; woodpecker finch C. pallidus; vegetarian finch Platyspiza crassirostris; Cocos finch Pinaroloxias inornata; warbler finch Certhidea fusca; warbler finch C. olivacea (phylogeny from [5]). (B1) Large ground finch (left) has a very deep and broad bill adapted to crack hard and large seeds, while the cactus finch (right) has an elongated and pointy beak for probing cactus flowers and fruits. (B2) Geospiza finch bills develop their distinct shapes during embryogenesis and are apparent upon hatching (mid-development stage 35 embryos are shown from Abzhanov et al.[12]). (B3) The cactus finch-specific expression of CaM was validated by in situ hybridization after it was identified as a candidate by a microarray screen [14].

Figure 2

Constrained sequence analyses. Frequency histograms of inter-gap segment lengths are compared against the neutral expectation (solid line) (a,b). The shaded orange area represents the total amount of indel-purified sequence shared by the species pair. Histograms are derived from (a) chicken - G. magnirostris and (b) chicken - zebra finch whole genome alignments. Results are shown for a representative G+C-fraction from the 11^th of 20 equal size G+C-bins, with the corresponding histograms from all G+C-fractions presented in Additional file 9, Additional file 10 and Additional file 11. Predicted amounts of constrained sequence estimated between three avian species pairs plotted against (c) the synonymous substitution divergence (d_S) and (d) GC content of equally populated GC bins, with data inferred from the T. guttata – G. magnirostris and G. gallus – G. magnirostris alignments, respectively. The larger amount of constrained sequence inferred for the G. magnirostris and T. guttata comparison compared to the two chicken – finch comparisons implies that there is functional sequence that is passerine-specific and thus not present in chicken.

Figure 3

Phylogeny of seven amniotic species. Branch lengths are proportional to d_S; the degree of constraint (d_N/d_S) for each terminal lineage is also indicated (values shown in red). Evolutionary rates (d_S and d_N/d_S) are median values deriving from 1,452 alignments of simple one-to-one orthologues present in each species.

Figure 4

Evolutionary rate analyses. (a) The Branch-site test models of Zhang et al.[42]. The schematic represents the alternative model that allows for positive selection. Under the null model, sites fall into site classes 0 or 1 only. The two models are compared using a likelihood ratio test. (b) The number of positively selected genes identified on G. magnirostris, T. guttata, passerine, G. gallus, M. gallopavo, galliform, and avian branches. (c) Average levels of dN/dS for the G. magnirostris or T. guttata lineages for G. magnirostris and T. guttata positively-selected genes (PSGs) and for non-PSGs inferred by parsimony. Alignment showing the candidate Geospiza positively selected codon sites (highlighted in red) in (d) POU1F1 and (e) IGF2R. Alignment visualised with the belvu software [90].

Figure 5

Gene tree showing the evolution of CCDC147 . Lineage-specific d_N/d_S values estimated for the CCDC147 gene across aminotes. The long passerine branch highlighted in red is inferred to have experienced many events of positive selection.