Chromosome-Level Genome Assembly of the Common Chaffinch (Aves: Fringilla coelebs): A Valuable Resource for Evolutionary Biology.

The common chaffinch, Fringilla coelebs, is one of the most common, widespread, and well-studied passerines in Europe, with a broad distribution encompassing Western Europe and parts of Asia, North Africa, and the Macaronesian archipelagos. We present a high-quality genome assembly of the common chaffinch generated using Illumina shotgun sequencing in combination with Chicago and Hi-C libraries. The final genome is a 994.87-Mb chromosome-level assembly, with 98% of the sequence data located in chromosome scaffolds and a N50 statistic of 69.73 Mb. Our genome assembly shows high completeness, with a complete BUSCO score of 93.9% using the avian data set. Around 7.8% of the genome contains interspersed repetitive elements. The structural annotation yielded 17,703 genes, 86.5% of which have a functional annotation, including 7,827 complete universal single-copy orthologs out of 8,338 genes represented in the BUSCO avian data set. This new annotated genome assembly will be a valuable resource as a reference for comparative and population genomic analyses of passerine, avian, and vertebrate evolution.

Significance

High-quality reference genomes of wild, nonmodel species are very useful tools to understand how organisms evolve. If genomes are annotated, so that the specific genes are identified, we can make progress toward associating specific physical or behavioral traits with the genes that code for them, and thus further understand the evolutionary process. Here, we provide a high-quality, annotated genome of the common chaffinch, a common and widespread Eurasian finch, that will be a useful resource in studies related to evolution, phylogenomics, biogeography, and adaptation genomics, among others.

Introduction

The decreasing costs of DNA sequencing, along with advances in computational genomics, are promoting a rapid increase in the availability of high-quality reference genomes of nonmodel species, which greatly improves our capacity to address a range of biological questions from a genomic perspective. Among them, the correct annotation of protein-coding genes in whole genomes allows to identify new genes involved in the process of evolutionary adaptation and provides a better understanding of the evolutionary mechanisms involved in the speciation process. Avian genomes are particularly suited for studying the molecular basis of speciation as they have a relatively simple architecture and are among the smallest within amniotes, ranging from 0.91 to 1.3 Gb (Gregory 2002). In the last decade, the number of bird reference genomes has increased dramatically (e.g. Dalloul et al. 2010; Warren et al. 2010; Zhang et al. 2012; Jarvis et al. 2014; Poelstra et al. 2014; Frankl-Vilches et al. 2015; Friis et al. 2018; Louha et al. 2020; Peñalba et al. 2020; Ducrest et al. 2020, Wang et al. 2020), providing major scientific breakthroughs in phylogenetics (Alström et al. 2018; Braun et al. 2019; Jarvis et al. 2015), comparative genomics (Zhang et al. 2014, Feng et al. 2020), adaptation genomics (Wirthlin et al. 2014; Lawson and Petren 2017), and genomic architecture (Poelstra et al. 2014; Vijay et al. 2016), among others. Moreover, the Ten-Thousand Bird Genomes (B10K) consortium has generated and analyzed over 300 avian genomes from 92.4% of bird families, providing an unprecedent genomic resource for avian comparative studies (Zhang et al. 2015, Feng et al. 2020).

The common chaffinch (Aves, Passeriformes, Fringillidae, Fringilla coelebs) is a widely distributed species, ranging from across Eurasia to the north of Africa, and has colonized three Macaronesian archipelagos in the Atlantic Ocean (Azores, Madeira, and the Canary Islands) (Collar et al. 2020). With about 15 currently recognized subspecies, the common chaffinch is an ideal system for testing hypotheses on the evolutionary process given its distribution across the continent and the colonization of several oceanic islands, recognized as excellent natural laboratories for studying evolution (Brown et al. 2013). Island systems have inspired the development of biogeographical theories (MacArthur and Wilson 1967) and are of central importance for understanding the role of area and isolation in colonization, extinction, and speciation rates (Valente et al. 2020), which are processes influencing global patterns of species richness (Losos and Schluter 2000). Species that have colonized insular environments, like the common chaffinch, are also excellent systems for the study of demographic events, such as bottlenecks leading to small effective population size (Ne) (Leroy et al. 2021), or the roles of drift and selection in the divergence process (Barton 1996). The common chaffinch has been intensively studied using molecular tools, so that the availability of a reference genome represents a valuable resource to improve our understanding of avian evolution, biogeography, and demography (Illera et al. 2018).

Results and Discussion

Assembly and Quality Control

The total length obtained by the HiRise software for the common chaffinch assembly was 994.87 Mb. Nevertheless, the estimate from k-mer metrics is 1.2 Gb. The discrepancy between these estimates could be caused by the presence of repetitive elements given the assembly strategy used, which could have been improved including long-read sequencing technologies. This final assembly consists of 3,255 scaffolds, 3,239 over 1 kb, and an N50 of 69.73 Mb (see supplementary table S1, Supplementary Material online) with a sequence coverage of 249×. The use of Chicago and Hi-C libraries provided a clear improvement in quality by increasing 917 times the scaffold N50, reducing the number of scaffolds from 38,666 to 3,255 (Supplementary table S1, see supplementary methods for details, Supplementary Material online). In fact, 98% of the total genome sequence maps in the 30 described chromosomes.

The chaffinch genome showed high synteny with the zebra finch genome (fig. 1), evidencing the completeness of the assembly, with all micro-chromosomes and the Z chromosome present in the assembly. In addition, the alignment between these genomes suggests the presence of several inversions in chromosomes 1, 1 A, 2, 3, 5, 7, 8, and 9. Several studies have documented that inversions are very common in birds (Aslam et al. 2010; Völker et al. 2010; Skinner and Griffin 2012; Zhang et al. 2014). For instance, Hooper and Price (2017) identified 319 inversions on the 9 largest autosomes combined in 81 independent clades. No putative contaminations were detected and 89.6% of the reads were mapped in the genome assembly (see Supplementary fig. S1, Supplementary Material online). The mean GC content of the assembly was 41.86% (±11 SD). The common chaffinch genome assembly included 7,832 complete copies (93.9%) out of the 8,338 BUSCO data set from avian genomes, among which 7,816 were single-copy orthologs and 16 were duplicated. Only 1.8% of the gene models were fragmented, and 4.3% were missing in the genome. These few missing gene models could represent divergent or lost genes in our species, but also could be related with putative errors during the assembly process or missing data.

. 1.

(a) Circos plot comparing the zebra finch (right hemisphere) and the common chaffinch (left hemisphere) genome assemblies. The common chaffinch chromosomes marked with an asterisk (*) show inversions with respect to the zebra finch assembly. (b) Linear synteny plots of the common chaffinch chromosomes showing inversions relative to the zebra finch generated with the R package genoPlotR (Guy et al. 2010). The zebra finch assembly (top) is compared with the common chaffinch assembly (bottom), and numbers designate specific chromosomes.

Repetitive Regions

Overall, 7.82% of the genome assembly are repeats (∼78 Mb), of which 85.4% are transposable elements (TEs). The most abundant TEs are LINEs (53.5%) followed by LTR (29.4%), DNA elements (4.1%), and SINEs (1.4%), with the remaining 11.6% unclassified. The rest of repeats (14.6%) contained simple repeats (75.4%), low complexity repeats (18.5%), satellites (4.2%) and small RNA (1.9%) (see supplementary table S2, Supplementary Material online ). The number of repetitive regions is within the expected range in birds, which is at 4–10% of the genome (Zhang et al. 2014).

A total of 111,076 microsatellites, with motif length ranging between 2 and 20 bp, were identified in the common chaffinch genome (see supplementary fig. S3, Supplementary Material online; their genomic locations are shown in supplementary file S1 in the Figshare repository). The most common k-mer sizes conforming the microsatellites were 2 (68.2%), 3 (15.9%) and 4 (8.2%) (see supplementary file S1). The most common length of the microsatellites was 10 bp (40.4%), followed by 12 bp (13%) and 15 bp (8.8%) (see supplementary file S1 for the length distribution of microsatellites). In addition, the number of microsatellites was positively correlated with the sequence length (Supplementary fig. S3, Supplementary Material online; see supplementary file S1 for the frequency of occurrence in every scaffold).

Gene Annotation and Function Prediction

Our annotation pipeline combining both de novo and homology-based predictions inferred 21,831 proteins encoded by 17,703 genes in the common chaffinch genome with a mean length of 15,818 bp (Table 1). The common chaffinch genome annotation (see supplementary file S2 in Figshare) included 7,850 complete copies (94.2%) out of the 8,338 of BUSCO avian data set used, retrieving all expected copies with a slight increase from that estimated in the un-annotated genome (see above). Among the complete BUSCO genes, 7,827 were single-copy orthologs (99.7%) and 23 were duplicated (0.3%). Around 1.9% (162) of the gene models were fragmented and 3.9% showed no significant matches (326).

Table 1

Genome Statistics and Predicted ncRNAs of the Fringilla coelebs Genome Compared with Other Similarly Sized Avian Species (Melospiza melodia, Taeniopygia guttata, Ficedula albicollis, Manacus vitellinus, and Geospiza fortis), Modified from Louha et al. (2020).

	F. coelebs	M. melodia	T. guttata	F. albicollis	M. vitellinus	G. fortis
Number of genes	17,703	15,086	17,561	16,763	18,976	14,399
Mean gene length (bp)	15,818	14,457	26,458	31,394	27,847	30,164
Number of CDSs	17,703	15,086	17,561	16,763	18,976	14,399
Mean CDs length (bp)	1,679	1,325	1,677	1,942	1,929	1,766
Number of exons	221,872	131,940	171,767	189,043	190,390	164,721
Mean exon length (bp)	165	153	255	253	264	195
Mean number of exons/gene	10.16	8.67	10.25	12.22	11.51	11.41
Number of introns	200,041	116,724	153,909	171,236	171,089	149,563
Mean intron length (bp)	1,902	1,695	2,930	3,257	3,294	2,813
Total proteins	21,831
ncRNA
tRNA	325	267	184	179
miRNA	140	166	302	510
snRNA	18	16	44	32
snoRNA	126	154	241	199
rRNA	5	8	100	22
lncRNA	17	20	908	1473

Over all predicted proteins, 19,458 (89.1%) provided positive BLASTP hits against the Uniprot SwissProt database, and 19,617 (89.9%) against the annotated proteins from the zebra finch genome. In addition, InterproScan identified 18,551 (85%) specific protein-domain signatures in the predicted peptides. The combination of the annotation from these databases allowed assigning a functional annotation with GO terms to 19,425 proteins (89%) assigned to 15,309 genes (86.5%; supplementary file S3 in Figshare).

tRNAs and Other Noncoding RNA Prediction

The search by tRNAscan-SE (supplementary file S4 in Figshare) identified 325 tRNAs in the common chaffinch genome, of which 167 decode for the standard twenty amino acids. Among all the tRNAs detected, 131 presented low scores and therefore were categorized as pseudogenes (i.e. lacking tRNA-like secondary structures). There were no suppressor tRNAs, 1 had undetermined isotopes, 25 were chimeric and 15 included introns within their sequences. One of the tRNAs was predicted to code for selenocysteine (sequences and structures of the predicted tRNAS are available in File S5 in Figshare). In addition, the search against both tRNA databases (GtRNAdb and tRNAdb) yielded positive results in many other species, suggesting that tRNA prediction in our assembly was correct. Moreover, our searches using Infernal identified 354 ncRNAs, which were classified as follows: 39 CREs, 2 Ribozymes, 7 Gene, 140 miRNAs, 126 snoRNAs, 18 snRNAs, 5 rRNAs, and 17 lncRNAs (File S6 in Figshare). The number of tRNAs predicted in the common chaffinch genome is the highest when compared with other passerine species (i.e., M. melodia, T. guttata, and F. albicollis), but the other types of ncRNAs present similar values to the M. melodia genome and lower than the other two species (Table 1), probably because we applied a strict threshold to avoid an excess of false positives.

Conclusions

We provide here a high-quality assembly for the common chaffinch, a valuable resource as a reference genome to address a range of biological questions from a genomic perspective. Moreover, our annotation provides useful information to detect candidate genes involved in adaptation and divergence processes. The combination of the Chicago and shotgun sequencing with the HiRise assembly approach lead to a highly contiguous chromosome-level genome assembly. The genome assembly size was 994.87 Mb, with the 30 chromosomes accounting for 98% of it. Although the expected length of the genome was 1.2 Gb, closer to those obtained in other avian species by flow cytometry (Gregory 2002), the BUSCO analyses showed that both the assembly and structural annotation encode 93.9% and 94.2% complete copies out of the 8,338 orthologous conserved genes in avian species, respectively. This discrepancy of the genome size could be caused by the absence of large repetitive elements in the assembly. The structural annotation predicted 17,703 coding genes, with most of them (86.5%) assigned to functional annotation and GO terms.

Materials and Methods

Sample Collection and Genome Assembly

A blood sample was extracted from a common chaffinch female captured in Torreiglesias, Segovia, Spain, in 2017 and frozen immediately in liquid nitrogen. The sample was handled by Dovetail Genomics for DNA extraction, sequencing and genome assembly using the HiRise pipeline (Putnam et al. 2016). The absence of a Z chromosome in our first assembly (GenBank assembly accession: GCA_015532645.1) led us to conduct a second assembly presented here, which includes sex-linked scaffolds used to reconstruct the chaffinch Z chromosome (see supplementary methods, Supplementary Material online, for details). Gene completeness in the chaffinch genome assembly (and in the annotated gene set) was assessed through BUSCO (Benchmarking Universal Single-Copy Orthologs) v4.0.5 (Seppey et al. 2019) by using the 8,338 single-copy orthologous genes in the Aves lineage group odb10, using chicken as the Augustus reference species.

Identification of Repetitive Regions and Gene Annotation

Repetitive regions were identified and masked prior to gene prediction. First, repeats were modelled ab initio using Repeat Modeler 1.0.11 (Smit and Hubley 2019) in scaffolds longer than 100 Kb with default options. The repeats obtained were merged with known bird repeat libraries from the RepBase database (RepBase-20181026) (Bao et al. 2015), Dfam_Consensus-20181026 and repeats from the zebra finch (obtained from B10K). The resulting repeat library was compared against the complete assembly with Repeat Masker 4.0.7 (Smit et al. 2015) and the identified regions were soft-masked. For the identification and description of microsatellites in the common chaffinch genome assembly we used GMATA v.2.01 (Wang and Wang 2016), with sequence motif length between 2 and 20 bp.

Gene prediction was conducted with BRAKER v2.1.5 (Hoff et al. 2016) and GeMoMa v1.7.1 (Keilwagen et al. 2016, 2018). First, the conserved orthologous genes from BUSCO Aves_odb10 were used as proteins from short evolutionary distance to train Augustus (Gremme et al. 2005; Stanke et al. 2006; see figure 3B in Hoff et al. 2019). The predicted proteins were combined with homology-based annotations using the zebra finch (GCF_008822105.2; Warren et al. 2010) and chicken (GCF_000002315.6; Hillier et al. 2014) annotated genes with GeMoMa pipeline, obtaining the final reported gene models. We applied a similarity-based search approach to assist the functional annotation of the chaffinch predicted proteins, using the UniProt SwissProt database, the annotated proteins from the zebra finch genome (Warren et al. 2010; UniProt Consortium 2014) and InterProScan v5.31 (Jones et al. 2014). The functional annotation, including Gene Ontology terms, was integrated from all searches providing a curated set of chaffinch coding genes (see supplementary methods, Supplementary Material online, for details).

Noncoding RNA Prediction and Identification

For the prediction and functional classification of Transfer RNAs (tRNAs) in the common chaffinch genome we used tRNAscan-SE v2.0 (Lowe and Chan 2016). The tRNA search across the genome and the identification of ncRNA (noncoding RNA) homologues was conducted using the software package Infernal v1.1.1 (Nawrocki 2014) (see supplementary methods, Supplementary Material online, for details). For comparative purposes, we added our results to those from Louha et al. (2020), which compared different genome assemblies of avian species.

Supplementary Material

Supplementary data are available at Genome Biology and Evolution online.

Click here for additional data file.