Meiotic recombination is the main source of genetic variability in any population of sexually reproducing organisms. For this reason, the number of recombination events (crossovers) per genome (recombination rate) and their distribution along the chromosomes are considered as important genomic characteristics of species and studied actively in plants, fungi and animals (mostly mammals). These studies demonstrated substantial interspecies variation in recombination rate (Dumont and Payseur 2008; Frohlich et al. 2015; Dapper and Payseur 2017; Stapley et al. 2017) while within-population variation was found to be of the same magnitude across various taxa (Ritz et al. 2017). The minimum possible rate of recombination is constrained by the necessity of at least one crossover per pair of homologous chromosomes to ensure their orderly segregation in the first meiotic division. Thus, the recombination rate cannot be lower than the haploid chromosome number (n). Studies on mammals demonstrated that chromosome number (2n) and chromosome arm number (FN) are the best predictors of the recombination rate (Pardo-Manuel de Villena and de Sapienza 2001; Segura et al. 2013; Capilla et al. 2016). A well-known phenomenon of crossover interference (a low probability of crossovers to occur close to each other) is an important constraint of the maximum possible rate of recombination (Berchowitz and Copenhaver 2010; Segura et al. 2013). The smaller the chromosome, the less likely it contains more than one crossover. Thus, the total genome size and the total length of the synaptonemal complex (SC, the core structure of pachytene chromosomes) may also serve as predictors of species-specific recombination rate (Peterson et al. 1994; Kleckner et al. 2003). While molecular and cellular mechanisms controlling the recombination rate are more or less clear, the adaptive importance of interspecies differences for this trait remains a matter of discussion (Stapley et al. 2017; Dapper and Payseur 2017; Ritz et al. 2017).

Most of the birds species studied for recombination rate have common karyotypes (2n = 78–82). Only two species with reduced chromosome number (2n = 68 and 74) have been examined and they did not show a drastic reduction of the recombination rate (Lisachov et al. 2017a). More birds with atypical karyotype as Apodidae (swifts), Psittaciformes (parrots) and Falconidae (falcons) need to be studied to reveal the cytogenetic and evolutionary basis of interspecies variation in recombination.

There are two reasons why bird meiotic chromosomes are poorly studied. First, the pachytene cells suitable for analysis can only be obtained in short time windows: from testes of adult males during or shortly before the breeding season or from ovaries of female embryos before hatching or chicks soon after hatching (Scanes 2014). Secondly, the bird has to be sacrificed.

In this study, we collected materials from fatally injured Common Swift (Apus apus Linnaeus, 1758) and Eurasian Hobby (Falco subbuteo Linnaeus, 1758) males provided by the local Bird of Prey Rehabilitation Centre. These specimens represent two bird families, which show a wide karyotypic variation. The karyotype of the Common Swift remains unknown. However, the other swifts examined have reduced and variable diploid numbers (from 62 to 70) (Yadav et al. 1995; Torres et al. 2004). Falconids show a very wide variation in diploid chromosome numbers (from 40 to 92) (Christidis 1990). Eurasian Hobby has 2n = 50 (Christidis 1990; Wang and Chen 1998; Nishida et al. 2008).

In this study, we for the first time describe the somatic and pachytene karyotypes of the male Common Swift and the pachytene karyotype of the male Eurasian Hobby. We estimate the overall number and distribution of recombination events along the chromosomes of these species, using immunolocalization of SYCP3, the protein of the SC lateral elements, MLH1, the mismatch repair protein marking mature recombination nodules, and centromere proteins. This method has been successfully used to analyze recombination landscapes in fish (Moens 2006; Lisachov et al. 2015), reptiles (Lisachov et al. 2017b, c), mammals (Capilla et al. 2016) and birds (Pigozzi 2016). This approach has lower resolution than linkage analysis at the fine scale level. However, cytological approach provides more reliable estimate of total number of global recombination rate, because it does not depend on the number and location of informative markers. Immunocytological analysis of SC spreads is especially efficient to examine bird karyotypes, because it provides unambiguous visualization of all microchromosomes, while at conventional and differential stained metaphase spreads it is difficult to distinguish microchromosomes from non-specifically stained debris.

In this study, we for the first time describe the somatic and pachytene karyotypes of the male Common Swift and the pachytene karyotype of the Eurasian Hobby and estimate the overall number and distribution of recombination events along the chromosomes of these species.

The karyotype of the Common Swift consists of three metacentric, three submetacentric and two telocentric macrochromosomes and 31 telocentric microchromosomes (2n = 78; FN = 90). It differs from the karyotypes of the previously described related Apodidae species: Little Swift (Apus affinis affinis) (2n = ± 70) (Yadav et al. 1995), Fork-tailed Swift (Apus pacificus) (2n = 62), Needle-tailed Swift (H. caudacutus) (2n = 64) (Christidis 1990) and two tropical swift species in the genus Streptoprocne: S. biscutata (2n = 64) and S. zonaris (2n = 66) (Torres et al. 2004). Thus, the Common Swift has a karyotype that is typical of most Neoaves and apparently basal for Apodidae. The karyotype evolution of other swifts has probably involved fusions of some microchromosomes with each other or/and macrochromosomes and pericentric inversions of macrochromosomes (Torres et al. 2004). Apparently, the reduction of chromosome numbers occurred independently in the genera Apus, Hirundapus and Streptoprocne, which diverged from each other about 30–40 Mya (Jetz et al. 2012).

The karyotype of our specimen of the Eurasian Hobby coincides with those reported earlier for this species in the diploid number (2n = 50) but differs for the chromosome morphology (Christidis 1990; Nishida et al. 2008; Wang and Chen 1998). Its chromosome 1 and one of the microchromosomes are metacentrics, all the other chromosomes are telocentrics. Christidis (1990) referring to a conference report of X. Bian and Q. Li indicated that all chromosomes in the hobby karyotype were telocentric with exception of chromosome 1, which was acrocentric. Wang and Chen (1998) described the karyotypes of two hobby specimens from China. Both of them had 2n = 50, but differed in macrochromosome morphology. A male from Qiqihar city had metacentric chromosome 1, subtelocentric chromosomes 2, 4 and 5, submetacentric chromosome 9 and all the other chromosomes as telocentrics. A female from Dalian city differed from this specimen by the morphology of chromosomes 2, 4, and 5: they were telocentric. It is likely that this species is polymorphic for pericentric inversions or/and centromere shifts or/and the presence of additional heterochromatic short arms at some chromosomes.

It should be noted, however, that the somatic metaphase and pachytene SC karyotypes of the same specimen might differ in details or, to be precise, reveal different details of the chromosome structure. Immunolocalization of SYCP3 might misrepresent the arm ratio of some chromosomes. It is well known that heterochromatin is underrepresented in the SC (Stack 1984). Studies on mammals show that short heterochromatic arms present at some chromosomes may not be visualized at SCs (Graphodatsky et al. 2000). On the other hand, immunolocalization of centromere proteins reveals centromere position more reliably than primary constrictions at metaphase spreads. For example, the metacentric microchromosome detected in our study and absent in the description of Wang and Chen (1998) can only be identified at pachytene spreads (Fig. 2b), but not at metaphase plates.

Two centromere signals that we detected at the hobby SC1 might indicate heterozygosity for pericentric inversion or a centromere shift, or the presence of vestigial inactive centromere sequence at one or both homologs. Although pericentric inversions and centromere shifts are well documented mechanisms of bird chromosome evolution (Kasai et al. 2003; Skinner and Griffin 2012; Griffin and Burt 2014), it is unlikely that they are the cause of the double centromere signal in the hobby. The occurrence of MLH1 foci between two centromere signals disapproves inversion heterozygosity, and the difference in signal intensity disapproves centromere shift. If there had been inversion heterozygosity, no crossovers would have occurred in the non-homologously synapsed inverted region. If there had been centromere shift, both signals would have been of the same intensity and weaker than the signals of the centromeres of other chromosomes. The vestigial inactive centromere sequence is the most likely explanation of the observed feature. Inactive centromeres in yeasts, plants and mammals lack several centromere proteins, such as CENP-A, CENP-C and CENP-E (Ross et al. 2015). If that were the case with the bird chromosomes, that would explain the weaker signal of the secondary centromere at the hobby SC1.

A rather surprising result of our study is that despite an about 40% difference in 2n and FN, the swift and the hobby have almost the same number of recombination nodules per genome. Apparently, the decrease of the lower limit of recombination rate (the number of obligatory crossovers) in the hobby has been compensated by a relaxation of the higher limit: increase in SC length and decrease in crossover interference. Similar in overall recombination rate as they are, the swift and hobby have very different patterns of recombination event distribution along macrochromosomes: extremely polarized in the swift and rather even in the hobby. A relatively even crossover distribution along the macrochromosomes with moderate peaks at the chromosome ends (including pericentromeric ends of acrocentrics) has been observed in the majority of bird species examined cytologically. A polarized distribution is more typical of mammalian chromosomes (Borodin et al. 2009; Segura et al. 2013) and macrochromosomes of anole lizards (Lisachov et al. 2017c). In birds, a pattern like this has so far been described in Zebra Finch and is considered as unique (Calderon and Pigozzi 2006; Backström et al. 2010).

Our data add two more species to the list of birds in which the number and distribution of recombination nodules have been examined (Table 1). This list is short; however, it already reveals some important features of recombination in birds. It confirms earlier suggestions that recombination rate in birds is substantially higher than in mammals. At the same time, birds demonstrate rather a low interspecies variably of recombination rate compared to mammals. This may in part be due to a strong conservation of chromosome number in birds. However, our data demonstrate that reduction of chromosome number does not lead to decrease of recombination rate. Meanwhile, more data from different taxa are needed to draw statistically supported conclusions about the evolution of recombination in birds.