The mechanisms of genome evolution are most often considered from the perspective of individual genes or gene families; there is nonetheless increasing evidence supporting the functional role and significance of events at a chromosomal (cytogenetic) level [1]. To date, bird genomes remain relatively understudied from an overall genome organization perspective; however, the recent availability of multiple avian genome sequence assemblies [2] allows us to consider the role of chromosomal change in the evolution of Aves from their dinosaur ancestors. Chromosome rearrangements between species can cause or reinforce reproductive isolation through reduced fitness of hybrid offspring due to a compromised ability to synapse and segregate chromosomes at meiosis [3, 4]. Moreover, reduced interspecific recombination in rearranged regions is thought to promote the accumulation of incompatibility loci in such regions [5–7]. The purpose of this study was to gain further insight into the mechanism of bird evolution through the multiple comparative analyses of chromosomal segments and breakpoints.

Unraveling the mechanisms and relevance of bird karyotype evolution has hitherto been impeded by a karyotype that is difficult to define because of indistinct banding on the macrochromosomes and a preponderance of cytogenetically indistinguishable microchromosomes. Indeed, to date, only a single avian karyotype (chicken) has been fully defined using a combination of BAC/cosmid clones and chromosome paints generated by flow cytometry and microdissection [8]. Moreover, karyotypes are broadly similar in overall pattern from species to species. For instance, at a cytogenetic level, two thirds of bird species have a chromosome number of around 2n = 80 with similar numbers of macro- and microchromosomes suggesting little interchromosomal changes between species [9]. Molecular insights into interchromosomal differences between species (and the evolutionary events that have led to them) have focused mostly on the largest macrochromosomes. These studies applied chicken chromosome paints [10] to the chromosomes of numerous other species (reviewed in [11]) in zoo-FISH experiments. Such investigations have provided much insight into inter-macrochromosomal rearrangements between birds with the underlying message that the ancestral pattern has remained largely unaltered in the majority of species. Rare exceptions include significant chromosome rearrangement in Psittaciformes (parrots etc.), Falconiformes (falcons) and Sphenisciformes (penguins) [11]. There are also individual changes associated with representative orders, e.g., fission of chromosome 1 in Passeriformes (songbirds) and of chromosome 2 in certain Galliformes (land fowl) (reviewed in [11]). Studies of interchromosomal changes involving the microchromosomes are much more limited as the flow cytometry methods used to generate the chromosome paints [10] do not have the resolution to isolate individual microchromosomes.

Using chicken BAC clones, studies provide a low-resolution appraisal of intrachromosomal rearrangements between chicken and other species [12–14] (turkey, duck, zebra finch, respectively). This approach, however, is limited in its ability to identify the molecular coordinates of evolutionary breakpoints. The availability of whole assembled genomes [15–17] allows comparative genomics at a much more detailed level of resolution than can be achieved by cross-species FISH. Burt et al.[18] were the first to use bioinformatics to define inter-species analysis of whole avian chromosomes at a genomic level (chicken-human). The publication of the chicken genome sequence [15] provided more detailed information, establishing conserved synteny between chicken and human whole genome assemblies. In the ten years since, only conserved synteny comparisons have been made between the chromosomes of two [14, 19], or at most three [20, 21] avian species.

The use of whole genome assemblies to study cytogenetic phenomena has raised interest in the study of comparative cytogenetics from the perspective of evolutionary breakpoint regions (EBRs) and homologous synteny blocks (HSBs). To date, the majority of such studies have focused on mammals [22], however, analysis of other groups, such as birds, is essential in order to establish whether mammalian systems are representative of, or an exception to, general patterns observed in other animal groups. Larkin et al.[22] found that, in mammals, EBRs can lie in gene-dense regions. In the human genome EBRs also lie in regions with more zinc finger protein genes, more genes whose function is associated with environmental stimulus response, as well as more segmental duplications, CNVs, SNPs and retrotransposed genes. Such “EBR genes” appear to be related to lineage-specific biology and adaptive features [22–24]. EBRs are also frequently reused, i.e. there are regions of the genome that are prone to chromosomal breakage leading to translocations, inversions and fissions [25, 26]. Comparison of sequence assemblies in chicken, zebra finch and turkey suggests that breakpoint reuse is higher in birds than in mammals [20, 21]. The data in birds also suggests a key role for recombination-based mechanisms in the generation of chromosome rearrangements in that EBR location is consistent with elevated levels of genetic recombination at these loci [14]. This is consistent with the notion that, if recombination drives chromosomal rearrangements and assuming an evolutionarily conserved recombination landscape [27–29], EBRs might be enriched in genomic regions with elevated recombination rates. Not all species show an association of chromosomal breakage and elevated recombination however, e.g., insects [30, 31] and mammals. Indeed, in mammals Larkin et al.[22] suggested that the highest levels of recombination are located between the EBRs rather than in association with them.

HSBs have been defined in all animal species thus far examined for conserved chromosomal synteny [32]. Larkin et al.[22] argue that the continued presence of HSBs in all species may indicate a selective advantage to the retention of gene combinations in close proximity. Supporting evidence is found in the fact that multispecies HSBs (msHSBs) involving nine mammals plus chicken, unlike EBRs, are enriched in gene ontology (GO) terms for organismal development, central nervous system, and brain function in the human genome. Others argue that the idea of close proximity and any resulting correlation in expression patterns (if present) are not necessarily adaptive or required (e.g., [33, 34]). Given that around three quarters of avian chromosomes are small, cytogenetically indistinguishable microchromosomes, and that overall karyotype structure appears broadly similar between at least two thirds of bird species, a high degree of conserved chromosomal synteny is inferred [9]. This raises the hypothesis that avian karyotypes are evolutionarily static; however, for this to be tested, we would first need to establish that inter-microchromosomal rearrangements are rare or absent in most birds. If true, we would subsequently hypothesize that, like HSBs in mammals, individual whole microchromosomes are enriched for functional GO terms (regardless of any intrachromosomal rearrangements between them).

A detailed account of the chromosomal differences and changes that have occurred during the evolution of avian chromosomes is an essential prerequisite for any further insights into functional and/or mechanistic relevance. The combination of comparative analysis by bioinformatics and chromosome painting has the potential to do this, provided the appropriate tools are developed and used. The purpose of this study was thus to examine multiple avian genomes recently sequenced [2, 35], reconstruct the common ancestral karyotype and thence the evolutionary events that led to extant karyotypes. Furthermore, we tested the hypothesis that EBRs occurring in two lineages (chicken and zebra finch) are associated with elevated levels of genetic recombination and assessed the degree to which EBRs are reused in avian evolution. Finally, we tested the hypothesis that whole microchromosomes essentially constitute interchromosomal HSBs (i.e. that rearrangements between them are rare or absent) and that each microchromosome consists of functionally enriched GO terms.

The results presented here signify the most comprehensive appraisal of avian comparative cytogenetics to date. They provide a more detailed reconstruction of avian genome evolution than could be achieved by zoo-FISH analysis alone and demonstrate proof of principle from which further studies of genome evolution and comparative genomics can ensue.

We used a highly interactive avian genome dataset from the Evolution Highway comparative chromosome browser [37, 38] that, as has already been demonstrated in mammals, can be applied to compare the chromosome organization of individual or multiple species. The ultimate aim for this browser is that, in chromosomes for all avian species uploaded, HSBs will be displayed with reference to the chromosome number, as is currently the case for turkey, zebra finch and duck, or to specific scaffolds for other birds. In future, this will be achieved by a number of strategies: (a) by improved scaffold sizes, e.g., using optical mapping such as has been achieved to some degree in ostrich and budgerigar in this study; (b) by linkage to radiation hybrid (RH) maps such as was achieved for duck in this study (see also [19]); (c) by association with known linkage and other physical maps (e.g., [39, 40]); d) by use of novel algorithms to order and orient scaffolds into longer chromosomal fragments or whole chromosomes using comparative genome information and pair-end reads (reference-assisted chromosome assembly; [41]); (e) by systematic FISH mapping to chromosomes of orthologous clones derived from the individual scaffolds. We are currently concentrating our efforts on the development of FISH probes that will identify not only on which chromosomes the scaffolds lie in the species of interest, but also the order in which they appear on the chromosome. With current technology, however, even the best-assembled genomes (e.g., assisted with optical mapping) require a degree of intervention by molecular cytogenetics in order to generate a complete picture of overall genome organization. Given the efforts that have been made to sequence the genomes of the birds recently by current technologies [2], it is questionable how many of them will be re-sequenced using newer technologies that generate large scaffolds. A note of caution is relevant here: no genome assembly is “perfect” - the results reported here and elsewhere represent the state of the art in terms of what can be reasonably gleaned with the current technology available. Our future studies will focus on the systematic molecular characterization by zoo-FISH of as many scaffolds and EBRs as time and resources allow.

Earlier cytogenetic data suggested that, for the majority of bird species, karyotypic patterns are broadly similar to one another [9, 11, 14, 20]. This purportedly extends to ratite birds [42–44]; however, further analysis presented in this study challenges this notion. That is, we identified 26 interchromosomal rearrangements in ostrich compared to the ancestor. Moreover, the question of whether the conserved interchromosomal synteny seen in the macrochromosomes applies to the microchromosomes has hitherto been beyond the resolution of contemporary methodology. This study is the first to classify inter-microchromosomal rearrangements in any species; we provide evidence that interchromosomal rearrangements are nonetheless rare, except in cases (around 1/3 of species) where we already knew that karyotypes were highly rearranged [9]. Ostrich is the exception and it will be interesting to note whether this applies to other ratite birds.

Microchromosomes are not a uniquely avian feature. They are also found in some primitive amphibians (Cryptobranchidae and Hynobiidae have 14–19 pairs [45, 46]), most (but not all) reptiles (snakes have around 20 pairs [47]), but paradoxically not Crocodylia [48] – the closest phylogenetic lineage to birds. Indeed microchromosomes are typical of most amniotes (mammals and crocodilians being exceptions); however, the greatest number and smallest size of microchromosomes are typically found among birds. Burt [49] in a “fission-fusion” hypothesis suggested that most microchromosomes were already present in the common dinosaur ancestor that gave rise to birds (which probably had already evolved a small genome size and karyotype of around 2n = 60 including 20 pairs of microchromosomes) but that chromosome fission created the remainder, presumably including the smallest ones. In the current study, the similar number of chromosomes amongst most species but relatively large number of rearrangements between ostrich and all the other birds studied suggest that a basic pattern of 2n = 80 (~30 pairs of microchromosomes) became fixed before the Palaeognathae-Neognathae divergence 100 MYA but that interchromosomal rearrangement was still relatively common in birds at the time. Another alternative is that ratite birds underwent further adaptive changes that may be associated with the very different phenotypes present in this clade alone. The paucity of inter-microchromosomal rearrangements between most Neognathae (if the evidence presented here is representative, this would presumably include the 2/3 of Neognathae species where 2n = ~80) supports our hypothesis that the microchromosomes represent blocks of conserved synteny at an interchromosomal level. An absence of interchromosomal rearrangement could either suggest an evolutionary advantage to retaining this particular configuration or a lack of opportunity for chromosome rearrangement. The latter might be explained by few recombination hotspots, transposable elements or endogenous retroviruses, all of which have been associated with chromosomal change. Both inter- and intrachromosomal change can arise via these mechanisms, and thus the rapid amount of intrachromosomal but not interchromosomal change in our representative passeriform species, the zebra finch, suggest that there may be an evolutionary advantage to keeping microchromosomes numerous, gene dense, compact and evolutionarily static. Stasis in evolution can, however, arise via alternative interpretations; it may be that the mutational mechanisms underlying chromosomal changes are different in birds or that lack of adaptive value, rather than purifying selection, slows down the rate of chromosomal changes. At the time of writing no sequences have yet been associated with the very smallest of the avian microchromosomes (29–38) and this is an issue that will require rectifying in future avian genome projects using more sophisticated technologies.

The rate of chromosomal change in any eukaryotic organism, and the speciation that ultimately arises from it, is dependent on two factors: the rate of mutation and the rate of fixation [18]. The mutation rate of chromosomes is, in turn, related to the frequency of homologous sites [49]. Repeat structures in general, and transposable elements in particular, provide substrates for chromosomal rearrangement. In a genome that is constrained by size (perhaps, as has been suggested, because of the energy requirements associated with flight [50, 51]), the opportunity for mutation is reduced and only fission (or intrachromosomal rearrangement such as inversion) can occur. This would explain first why the avian genome is the most fragmented of any vertebrate genome (i.e. birds have the most chromosomes) and second why there have been few interchromosomal rearrangements in most species. There are also possible advantages of multiple chromosomes in a karyotype in terms of generating variation, the driver of natural selection. That is, more chromosomes lead to more combinations of gametes as well as an increase in recombination rate as there has to be at least one obligatory chiasma per chromosome. The absence of positive selection for much change in chromosome number is a possible explanation of why there was little fixation of any interchromosomal changes among birds although inbreeding and genetic drift may play a role [18, 49, 52, 53]. Burt [49] suggested that a higher recombination rate is another constraint that has resulted in the properties we most associate with microchromosomes (e.g., high GC-content, low repeats, high gene-density) and led to the maintenance of the typical avian karyotype with both macro- and microchromosomes and few rearrangements between them.

A constraint of overall karyotype structure does not preclude intrachromosomal rearrangements. Indeed there is a correlation between the rates of speciation and intrachromosomal rearrangement [4]. In the current study, the rapid rate of intrachromosomal rearrangement in the zebra finch would argue for a relationship between intrachromosomal rearrangement and speciation in birds given the Passeriformes represent over half of all species. Such mechanisms could be mediated through an increase in localized repeat content. Hotspots of recombination have previously been reported to also play a role [14] and in this study we tested the hypothesis further utilizing “zebra finch only” and “chicken only” breakpoints comparing them to previously reported genetic maps of each species [37, 54, 55]. In chicken, recombination rates were near identical in regions with breakpoints compared to those without. In zebra finch, the difference in rates between regions containing EBRs and regions without EBRs, although similar in magnitude to that previously reported [14], failed to reach statistical significance (at P < 0.05). This therefore casts doubt on our original findings, thereby either suggesting that our hypothesis should be rejected or that the numbers in the study were not sufficiently large to reach statistical significance. A further alternative explanation is that the available recombination maps have too low marker density (typically Mb scale) to pick up local recombination rate variation at a sufficiently detailed scale (Kb scale) to detect associations with EBRs. Study of a greater number of species in this manner using high-density linkage maps or population based recombination rate estimates may resolve the paradoxical difference between [14] and the current study.

Some avian species undergo a radical departure from the typical (2n = ~80) avian genome organization. The presence of an unusually high chromosome number in the Adélie penguin (2n = 96) and a lower than average number in the emperor penguin (2n = 72) (but both associated with high degrees of inter-microchromosomal rearrangement) suggest that similar mechanisms can act to either reduce or increase chromosome number rapidly. Evidence from the penguins and the rearranged karyotypes of the Falconiformes and the Psittaciformes suggest that these changes can happen in a relatively short time. Mammals, reptiles and amphibians with larger, repeat-rich genomes have the potential to undergo rapid intra- and interchromosomal rearrangements and the results presented here suggest that birds too can undergo similar changes in certain groups. We are not, however, aware of any evidence to suggest that highly rearranged avian genomes are especially large, or significantly more repeat-rich than other avian genomes. Comparisons of the zebra finch and the budgerigar suggest that mutation rates of chromosomes may well be similarly high in both groups but that they are features associated with exploiting evolutionary niches in certain groups that serve to fix interchromosomal rearrangements, while in others such fixation is prevented and the overall avian karyotype maintained. Such processes are, to date, undiscovered but possible clues might lie in the study of GO terms present in EBRs. In an associated study, a correlation between EBRs and specific avian adaptive features in individual species has been demonstrated. This included forebrain development in budgerigar, one of the six species focused upon in this study and consistent with this species being not only vocal-learner but having distinctive neuronal connections compared to other vocal-learners [37]. As more genomes become available with better assemblies, these analyses may well point to adaptive phenotypic features of individual orders and families.

Finally, we observed that it appears to be the chicken that seems to have undergone the fewest chromosomal changes compared to the ancestor. There are interesting parallels between this study and another study [56] examining sex chromosome evolution. While our data demonstrates that autosomes have been reorganized least in chicken chromosomes 1–5 in comparison to the common avian ancestor, Zhou et al.[56] conclude that the ancestral sex chromosome organization is observed closer to that of the Palaeognathae (ostrich and emu). Zhou et al.[56] show less degradation of the sex chromosomes and a closer synteny to the lizard. As, in this study, we only examined the Z chromosome in the Neognathae (for the reasons given), further studies will be required to establish whether sex chromosomes and autosomes preserve their ancestry differently in the different lineages. The question also arises of whether chicken and related species, having undergone the fewest chromosomal changes, have undergone the fewest adaptive changes compared to the avian ancestor. Most authors agree that the dinosaur ancestors of birds were bipedal and terrestrial, relatively small (small size being an immediate pre-adaptation to flight) and had limited flying ability, not unlike Galliformes [57]. On the other hand, the earliest known Ornithurae along the presumed direct line to modern birds were either fully aquatic or amphibious (e.g., Gansus[58]) and details of their anatomy, including webbed feet, have been likened to ducks [59, 60]. The oldest relatively certain fossil representative of Neornithes (modern birds) is aquatic, and identified as a Galloanseres (e.g., Vegavis[61]). However, the fossil record may be difficult to interpret due to geographic and depositional sampling biases, limited understanding of functional anatomy, and the uncertainty that avian ancestors were ecologically and behaviorally typical of the larger groups to which they belonged. As an independent record of the actual substance of inheritance of living birds, genomic characteristics such as chromosomal arrangement complement a fossil record that may imperfectly represent actual neornithine forebears. Thus, chromosomal rearrangements may provide information on the ecological adaptations of avian ancestors that the fossil record may never be able to establish unambiguously [62].

In summary, this study represents the most comprehensive appraisal of changes in overall avian genome structure hitherto reported. We provide further insight on previously reported roles of genetic recombination in chromosome rearrangement and on the functional significance of karyotype stability in the avian genome. Here, we establish that the chicken lineage contains the fewest number of chromosomal changes compared to the dinosaur ancestor relative to the other five species studied. At this stage it would be unwise automatically to infer that this means that the chicken has the fewest number of adaptive changes also. This will nonetheless be the topic of future study.

Michael N Romanov and Marta Farré, joint first authors.

Denis M Larkin and Darren K Griffin, joint last and corresponding authors.