https://orcid.org/0000-0002-2970-4189
Figures
Abstract
Reshuffling of genetic variation occurs both by independent assortment of chromosomes and by homologous recombination. Such reshuffling can generate novel allele combinations and break linkage between advantageous and deleterious variants which increases both the potential and the efficacy of natural selection. Here we used high-density linkage maps to characterize global and regional recombination rate variation in two populations of the wood white butterfly (Leptidea sinapis) that differ considerably in their karyotype as a consequence of at least 27 chromosome fissions and fusions. The recombination data were compared to estimates of genetic diversity and measures of selection to assess the relationship between chromosomal rearrangements, crossing over, maintenance of genetic diversity and adaptation. Our data show that the recombination rate is influenced by both chromosome size and number, but that the difference in the number of crossovers between karyotypes is reduced as a consequence of a higher frequency of double crossovers in larger chromosomes. As expected from effects of selection on linked sites, we observed an overall positive association between recombination rate and genetic diversity in both populations. Our results also revealed a significant effect of chromosomal rearrangements on the rate of intergenic diversity change between populations, but limited effects on polymorphisms in coding sequence. We conclude that chromosomal rearrangements can have considerable effects on the recombination landscape and consequently influence both maintenance of genetic diversity and efficiency of selection in natural populations.
Author summary
Reshuffling genetic variation is fundamental for maintaining genetic diversity and creating novel allelic combinations. The two main processes involved are the independent assortment of chromosomes and homologous recombination. The number and size of chromosomes can influence the amount of pairwise reshuffling and local recombination patterns. However, studying this in natural populations is challenging. In this study, we used the wood white butterfly, which exhibits an extreme within-species karyotype difference. Extensive fusions and fissions have resulted in almost twice as many chromosomes in the southern populations compared to the northeast populations. This unique system allowed us to assess the relationship between karyotype differences, pairwise reshuffling, recombination rate variation and subsequent effects on diversity and linked selection. We found that a higher number of chromosomes results in higher number of crossovers, although the difference was less than expected due to multiple recombination events occurring on longer chromosomes. Both populations showed an association between recombination rate and genome-wide patterns of genetic diversity and efficacy of selection. We provide evidence that chromosomal rearrangements have considerable effects on the recombination landscape and thereby influence the maintenance of genetic diversity in populations.
Citation: Näsvall K, Boman J, Höök L, Vila R, Wiklund C, Backström N (2023) Nascent evolution of recombination rate differences as a consequence of chromosomal rearrangements. PLoS Genet 19(8): e1010717. https://doi.org/10.1371/journal.pgen.1010717
Editor: Bret Payseur, University of Wisconsin–Madison, UNITED STATES
Received: March 25, 2023; Accepted: July 17, 2023; Published: August 7, 2023
Copyright: © 2023 Näsvall et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The data used in this article are from public repositories. The genome and linkage map data for L. sinapis is available at the European Nucleotide Archive (ENA) under accession number: PRJEB58697 (genome assemblies) and PRJEB58905 (pedigree RADseq data). The population data are available at ENA under accession number: PRJEB21838. All in-house developed scripts are available on GitHub (https://github.com/EBC-butterfly-genomics-team).
Funding: This work was funded by the Swedish Research Council (VR research grant #019-04791 to N.B.). The authors acknowledge support from the National Genomics Infrastructure in Stockholm funded by Science for Life Laboratory, the Knut and Alice Wallenberg Foundation and the Swedish Research Council, and SNIC/Uppsala Multidisciplinary Center for Advanced Computational Science for assistance with massively parallel sequencing and access to the UPPMAX computational infrastructure. The project was also supported by NBIS/SciLifeLab long-term bioinformatics support (WABI; to N.B.). R.V. was supported by the Spanish government (Agencia Estatal de Investigación) through grant PID2019-107078GB-I00/MCIN/AEI/10.13039/501100011033. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Genetic variation is a prerequisite for evolutionary change and mutation is the ultimate source of novel genetic variants. However, mutation only contributes to variation at individual genomic sites and establishment of new combinations of alleles across loci is dependent on reshuffling via independent segregation and recombination. Physical linkage can lead to a reduction in genetic diversity as a consequence of fixation or loss of haplotype blocks, either by random drift, or more rapidly by selection via hitchhiking and/or background selection [1,2]. Recombination events resolved as crossovers (from here on recombination) break physical linkage and generally lead to maintenance of genetic variation by generating a higher number of segregating haplotypes in the population [3–5]. Physical linkage between variants influenced by natural selection can also result in less efficient selection than if such variants would have been segregating independently. This process is generally referred to as Hill-Robertson interference, and can lead to an increased accumulation rate of deleterious mutations in regions with reduced recombination rate [6,7]. However, reshuffling can also break up co-adapted or synergistically interacting loci and therefore hinder beneficial epistasis.
Reshuffling of genetic variation occurs by two different mechanisms in diploid organisms. First, the independent assortment of homologous chromosomes to the gametes during meiosis (Mendel’s second law) results in reshuffling of loci located on different chromosomes [8]. Hence, the higher the number of chromosomes in an organism, the higher the potential for pairwise reshuffling. However, the effect renders a diminishing return as the probability of pairwise reshuffling approaches the maximum of 0.5 [9]. Second, the exchange of genetic material between chromosome pairs during homologous recombination can lead to novel allele combinations within chromosomes [10]. Consequently, characterizing where and how often recombination events occur along chromosomes is key to understanding maintenance of genetic diversity and the efficiency of selection in different parts of the genome. Empirical data have unveiled that the recombination rate (frequency of recombination events per unit length, e.g. cM / Mb) differs over several orders of magnitude, both among species, between chromosomes and across different genomic regions [11–13]. Interspecific variation in genome wide recombination frequency could be related to differences in both genome size [14] and haploid chromosome number [12]. The number of chromosomes can have an effect since at least one recombination event per chromosome pair (i.e. an obligatory chiasmata formation) is necessary for correct segregation during meiosis in many organisms [11–13,15]. However, there are exceptions to this generalisation, in Lepidoptera females and some Diptera males, for example, where meiotic divisions are achiasmatic [16–18]. The location and frequency of crossover events along chromosomes has been investigated in several organisms. The results so far generally show that the recombination landscape can be highly heterogeneous and that location and frequency of recombination events can be affected by, for example, interference between chiasmata, presence of centromeres and telomeres, nucleotide composition and chromatin state [11,12,19–21]. Furthermore, chromosome fusions have been shown to affect both the genomic distribution and inter-individual variation in frequency of crossovers in house mice, pointing to an effect of chromosomal rearrangements on the recombination landscape and recombination-dependent evolutionary processes [22]. However, detailed analyses of links between karyotype changes and recombination are limited, in particular in natural populations where fine-scale recombination rate data have been difficult to establish.
In order to explore how karyotype differences affect the recombination rate in a natural system, we used the wood white butterfly (Leptidea sinapis), a species within the Eurasian genus Leptidea which have considerable interspecific karyotype variability compared to the majority of other lepidopterans which usually have a remarkably stable karyotype of 2n ~ 62 [23]. Within L. sinapis, there is extreme intraspecific variation in karyotype; chromosome numbers range from 2n = 56–62 in the northern (Scandinavia) and eastern (central Asia) parts to 2n = 106–110 in the south-western (Iberian peninsula) part of the distribution range [24,25], and the sex-chromosome system consists of three Z-chromosomes and three W-chromosomes [26]. Like other lepidopterans, the wood whites have holocentric chromosomes—i.e. they lack a localized centromere region and multiple kinetochores can form along the entire chromosomes [27,28]. We know from genomic data that the intraspecific karyotype rearrangements in L. sinapis predominantly have been driven by recurrent fissions and fusions [29]. Specifically, the Catalan and the Swedish population differ by at least 27 fission / fusion events and 3–5 fission/fusion polymorphisms currently segregate in each of the two populations [29]. Previous analyses in other butterfly species have unveiled an inverse relationship between chromosome size and recombination rate [30], and a positive association between chromosome count and neutral diversity across Lepidoptera [31,32]. In addition, a detailed analysis in Heliconius ssp. has shown that fused chromosomes have a reduced recombination rate and genetic diversity as compared to chromosomes not involved in rearrangements [33,34]. Here, we use a study system that allows us to investigate the effects of extensive chromosomal rearrangements, both a large set (> = 27) of fissions and fusions that have been fixed between the two wood white populations and a handful of fission / fusion polymorphisms segregating within each respective population, on the recombination landscape and recombination-dependent evolutionary processes.
The aims of the study were; i) to characterize the recombination landscape in Swedish (2n = 56, 57) and Catalan (2n = 106–110) populations of L. sinapis, two populations with highly different karyotypes, ii) to compare the genome-wide probability of pairwise reshuffling between the two populations, and, iii) to quantify the effect of chromosomal rearrangements on maintenance/loss of genetic diversity and the efficacy of selection. We hypothesized that the Catalan population, with a more fragmented karyotype, would have a higher per base recombination rate and a higher total pairwise reshuffling rate, resulting in reduced linkage disequilibrium, higher neutral genetic diversity and increased efficacy in removal of slightly deleterious mutations.
Results
Recombination rates
We used pedigree data from a previous study to estimate the recombination rate from linkage maps for the two populations. The total recombination distance was considerably longer for the Catalan (2,300 cM) than for the Swedish (1,711 cM) population (S1 Fig and S1 Table), and, consequently, the average per base recombination rate calculated in 2 Mb windows in the Catalan population was significantly higher (mean = 4.60 +/- standard deviation 3.54 cM / Mb) than in the Swedish population (3.60 +/- 4.62 cM / Mb; Wilcoxon’s rank sum test, W = 58,797, p-value = 5.04*10−03). As expected given the similar genome sizes but different karyotypes in the two populations, the recombination distance per individual chromosome was lower in the Catalan (average = 43 cM; range = 10–90 cM) than in the Swedish population (average = 59 cM; range 33–84 cM). In the Catalan population, the per base recombination rate was lower on the Z-chromosomes than on the autosomes, but this difference was not significant (Z-chromosomes = 3.11 +/- 2.16 cM / Mb; Autosomes = 4.22 +/- 3.77 cM / Mb; Wilcoxon’s rank sum test, W = 4,438, p-value = 2.19*10−01). This effect can be a consequence of the relative difference in size between the Z-chromosomes and the autosomes in the two populations. The Z-chromosomes are virtually conserved between the two populations and hence constitute the largest chromosomes in the Catalan population, except for one large autosome. In the Swedish population on the other hand, the sizes of the Z-chromosomes are comparable to the majority of autosomes and the recombination rates of the two chromosome classes were also very similar (Z-chromosomes = 3. 37 +/- 3.56 cM / Mb; Autosomes = 3.38 +/- 4.68 cM / Mb).
We observed a significant negative association between the recombination rate and chromosome size in both the Swedish (Spearman’s rank correlation, rho = -0.51, p-value = 5.1*10−3) and the Catalan population (rho = -0.44, p-value = 1.2*10−3; Fig 1C). The larger chromosomes in the Catalan population had similar recombination rates as chromosomes of comparable size in the Swedish population, suggesting that the overall difference in total linkage map length and average recombination rate between the populations is predominantly an effect of the difference in karyotype structure. We also found that the recombination rate on smaller chromosomes was close to the expected rate per base with a single crossover per meiosis (50 cM / chromosome length; Fig 1D). However, the recombination rate on the larger chromosomes was higher than expected if the number of crossovers is limited to one per chromosome, which shows that more than one crossover per meiosis can occur if a chromosome is large enough (> ~ 18 Mb). Similarly, the ‘excess map length’ (remaining after removing the assumed obligate crossover (50 cM) from chromosome specific estimates) was significantly positively associated with chromosome size and explained 75% of the variation (linear regression model, R2 = 0.75, p-value = 2.2*10−16; Fig 1D).
(A) The distribution of recombination rate estimates (cM / Mb, y-axis) in the Catalan and Swedish populations, respectively. Box hinges represent the 25th and 75th percentiles, whiskers extend to values within 1.5 times the distance between the 25th and 75th percentiles and solid dots represent outliers. (B) Regional distribution of the recombination rate (cM / Mb; y-axis) for relative positions from the center (0) to the end (0.5) of chromosomes in the Swedish (green) and the Catalan (yellow) populations, respectively. Lines represent local regression (LOESS) of large (> 18 Mb, > 1 crossover / meiosis; dashed lines) and small chromosomes (< 18 Mb, < 1 crossover / meiosis; solid lines), respectively. (C) Association between the weighted mean recombination rate (cM / Mb; y-axis) and chromosome size (Mb; x-axis) in the Swedish (green) and Catalan (yellow) populations. The dashed lines represent the expected recombination rate with one (black) and two (red) crossovers per meiosis. The Z-chromosomes are represented by open and the autosomes by filled circles. Note that the shortest Z-chromosomes in the Catalan and the Swedish population, respectively, are of the same size and circles for these chromosomes are therefore overlapping. (D) Excess map length (map length—50 cM; y-axis) as a function of chromosome size. Colors and symbols as in (C) and the regression line and statistics correspond to a linear regression model.
The regional distribution of recombination showed a bimodal pattern for the larger chromosomes (> ~ 18 Mb) with a reduced crossover rate in the center and at the ends (Figs 1B and S2). For smaller chromosomes (< ~ 18 Mb) on the other hand, the recombination rate was highest in the center (Figs 1B and S2). The decrease in recombination in the chromosome ends was less pronounced in the small chromosomes in the Swedish population (Fig 1B). This could be a consequence of the fact that several of the smaller chromosomes are involved in ‘fusion polymorphisms’ currently segregating in the Swedish population.
Pairwise reshuffling
We proceeded by estimating the relative contribution of independent segregation and homologous recombination to the total reshuffling rate in the different populations. The major mechanism for genome-wide reshuffling in both populations was segregation rather than recombination (Table 1 and Fig 2A). The male interchromosomal contribution was 106 times higher than the intrachromosomal contribution in the Catalan and 61 times higher in the Swedish population. Despite the lower recombination rate on larger chromosomes, the contribution of intrachromosomal pairwise reshuffling increased linearly with chromosome size and size explained 62% of the variation between chromosomes (Fig 2B). This is likely a consequence of the occasional occurrence of double crossovers on larger chromosomes which should increase the pairwise reshuffling rate.
(A) Genome-wide probability of pairwise reshuffling per population and sex as a function of number of independently segregating chromosomes (interchromosomal effect). The open circles represent the expected trajectory of reshuffling in genomes with equal-sized chromosomes as the chromosome number increases (x-axis). (B) The probability of pairwise reshuffling within each chromosome (y-axis) as a function of chromosome size (x-axis) in the two populations (intrachromosomal reshuffling). The line represents a linear regression model.
The total contribution of males (Total male) and females (Inter female) and sex-average reshuffling (Sex average) are also given for comparison. Intrachromosomal reshuffling corresponds to recombination between loci within chromosomes and interchromosomal reshuffling reflects independent segregation.
The analysis revealed that the proportional difference in total reshuffling is less than 1% higher in the Catalan than in the Swedish population, despite the fact that the difference in chromosome numbers is almost two-fold and that females contribute more to the total reshuffling probability in the Catalan population (Table 1). The difference in reshuffling between populations was further reduced by the higher probability of intrachromosomal reshuffling in the Swedish males—the genome-wide intrachromosomal reshuffling contribution was 75.6% higher in the Swedish population (Table 1).
Associations between recombination and genetic diversity
Given the observed variation between populations in both overall map length, average recombination rate and relative intrachromosomal reshuffling rate, we proceeded by investigating the relationship between recombination rate and genetic diversity. We found that the Catalan population had significantly higher π in all site categories (Wilcoxon’s test; p-values = 2.73*10−04–7.05*10−07), but we did not detect any difference between the populations in the ratio of heterozygosities at zero- (π0) and four-fold (π4) degenerate sites (p-value = 1.52*10−01; Table 2), i.e. the efficiency of natural selection to remove slightly deleterious variants. Consistent with the ongoing and dynamic karyotype changes in these populations, there was also no significant association between π per chromosome and chromosome length after excluding the Z-chromosomes (S3 Fig). However, the regional distribution of π estimates followed the expectations from effects of selection on linked sites [35] in both populations, with a decrease in π and an increase in π0 / π4 at the terminal ends of the chromosomes where the recombination rate is significantly reduced (Fig 3A and 3C). We also observed a lower neutral genetic diversity for the Z-chromosomes compared to the autosomes. In addition, while Z1 and Z3 had similar π0 / π4-ratios as the autosomes, Z2 had a significantly higher π0 / π4-ratio compared to the other chromosomes in both populations, driven by an elevated π0 (S3 Fig).
(A) Regional distribution of nucleotide diversity (π) in zero-fold (π0), four-fold degenerate (π4), intergenic and intronic sites. (B) Nucleotide diversity (π) as a function of the recombination rate in each population estimated in 2 Mb windows. (C) Zero-fold/four-fold diversity (π0 / π4) along the chromosome (relative position from chromosome center (0) to chromosome end (0.5)) for the Swedish (green) and Catalan (yellow) L. sinapis populations. (D) Zero-fold/four-fold diversity (π0 / π4) as a function of the recombination rate in each population estimated in 2 Mb windows. Lines represent local regression lines (LOESS). Note that the trajectories for intergenic and intronic π are largely overlapping in panels A and B and therefore difficult to distinguish, except for windows with comparatively high recombination rates.
The ranges for chromosome specific estimates are given in parentheses. The statistics (W) and the p-values correspond to Wilcoxon rank sum tests with continuity correction.
We applied linear models to investigate the association between different diversity estimates and the variation in the regional recombination rate in each population in more detail. These analyses revealed that window-based estimates of diversity at all site categories, except π0 (R2 < 0.01, p-value > 2*10−01), were significantly associated with the recombination rate in both populations (R2 = 0.05–0.13, permuted p-values < 1*10−03; Figs 3B and S4). Consequently, there was a negative relationship between the recombination rate and π0 / π4 in both the Catalan (R2 = 0.04, p-values < 1*10−03) and the Swedish population (R2 = 0.05, p-values < 1*10−03), suggesting that the overall recombination landscape has been stable enough to have a similar effect on diversity and efficacy of selection in both populations since they diverged (Figs 3D and S4). The relationship between recombination rate and genetic diversity was non-linear (Fig 3B and 3D) and mostly driven by a stronger association in low recombination regions, especially in the Swedish population where regions with lower recombination rate (= < 2 cM / Mb) showed a stronger association (R2 = 0.12–0.23, p-values < 1*10−03) than windows with higher recombination rate (> 2 cM / Mb; R2 = < 0.01–0.03; S4 Fig). This indicates that recombination rate is not a limiting factor for diversity and efficacy of selection when the recombination rate increases above a certain level (~ 2 cM / Mb).
Global and regional estimates of recombination and diversity could be affected by other covarying variables, such as different genomic features, nucleotide composition and demographic processes like population bottlenecks, expansions, subdivision and migration [36]. In an attempt to account for these variables and investigate the relative effect of karyotype rearrangements, we quantified how observed changes in recombination rate between populations were associated with changes in diversity between the populations. This analysis was performed using homologous segments in ancestral chromosome blocks–i.e., pieces of chromosomes that have not been structurally affected by the previously identified > 27 fissions or fusions [29] that underlie the difference in karyotype between the Catalan and the Swedish population (Fig 4A). We focused on ancestral chromosomal units where the fission / fusion has been inferred to occur after the split from the closest outgroup species (L. reali), and cases where incomplete lineage sorting or reuse of chromosomal breakpoints have resulted in differences in karyotype between the two populations. This analysis unveiled a significant positive association between the change in recombination rate and intergenic π (Linear regression, y = 1.04 + 0.039x, R2 = 0.12, permuted p-value = 7.99*10−03; Fig 4B and 4C), i.e. changes in the recombination rate caused by chromosomal rearrangements explained 12% of the variance in genetic diversity between populations. The regression hence shows that there is an 8% excess in the diversity level in the Catalan population when recombination rate differences have been cancelled out (at X = 1). This shows that recombination rate changes have had a limited, but still substantial influence on the levels of diversity in each respective population. There was also a positive relationship between recombination and intronic π, but this trend was not significant after correction for multiple testing. We found no associations between recombination rate change with changes in π for other site classes or with π0 / π4, possibly an effect of stronger selective constraints for those sites, or that the sample size for coding sites is smaller which gives a lower power to detect associations (S5 Fig).
(A) Chromosomal rearrangements between Swedish (green squares) and Catalan (yellow squares) population of L. sinapis, the bands representing homology are colored by fission/fusion history. Chromosomes are ordered by size in each respective population (data from [29]). (B) Diversity estimates (π), specifically for intergenic sites across ancestral chromosomal units in each population. (C) The ratio between Catalan and Swedish π, as a function of the ratio between Catalan and Swedish recombination rate for intergenic sites in each ancestral chromosomal unit with known fusion/fission history. The line represents the slope from a model II linear regression (R2 and permutated p-value on top).
To summarize, the analysis shows that chromosomal rearrangements have affected the recombination landscape and changes in the level of standing genetic variation in intergenic sequences in L. sinapis.
Discussion
General
Here we characterized the effects of extensive within-species chromosomal rearrangement on recombination and total pairwise reshuffling rate variation and combined this with analyzing the associated influences on levels of genetic diversity and the efficacy of natural selection. In summary, we found i) that karyotype rearrangements affect both the genome-wide and the regional recombination rate, ii) that the number of chromosomes and their relative size are major determinants of the recombination rate, but that karyotype changes have very limited effects on the total pairwise reshuffling rate, and iii) that recombination rate changes associated with chromosome fissions and fusions have detectable effects on the neutral diversity at intergenic sites, but not on the efficacy of selection.
Recombination rate analysis
Our linkage map data revealed that the recombination rate was higher in the Catalan than in the Swedish population. This is expected given that chromosome size previously has been shown to be strongly associated with recombination rate in Lepidoptera [30,32]. We found that both Leptidea populations had global recombination rates in parity with genome-wide estimates in other Lepidoptera species, although direct comparison here is difficult since different methods have been applied [30,34,37]. However, the effective recombination rate per gene is likely higher in L. sinapis than in many other butterflies, since it has a comparatively large genome size (686 Mb) as a consequence of a considerable proliferation of transposable elements in intergenic sequences and introns [38]. The difference in recombination rate between populations was lower than predicted from karyotype differences if the average chromosome map length would have been 50 cM. One potential explanation for this deviation is the occurrence of multiple crossovers in large chromosomes, an explanation which is supported by our observation that excess recombination (> 50 cM / chromosome) increases linearly with chromosome size. The presence of multiple crossovers per chromosome indicates that butterflies are not limited to one crossover per meiosis, as C. elegans [39], but rather that the number of crossovers is dependent on chromosome size like in other holocentric species [40]. It should be noted that experimentally induced fused chromosomes in C. elegans can have multiple crossovers during meiosis [41], indicating that the interference distance is longer than the general chromosome size in C. elegans. The increase in excess recombination (> 50 cM / chromosome) and the bimodal distribution of recombination events across large chromosomes in the Swedish population are hence consistent with a scenario where the chiasmata interference length in L. sinapis is longer than a typical chromosome in the Catalan population, but shorter than the largest chromosomes in the Swedish population. Notably, we also found that smaller chromosomes sometimes showed a lower average crossover rate than expected if crossovers are necessary for correct segregation during meiosis in males (i.e. < 50 cM / chromosome). This could potentially be a result of very closely located double crossovers or single crossovers occurring close to the chromosome ends. Such events are likely largely undetected with our marker density. Another explanation, perhaps less likely, could be that correct segregation in male meiosis might not require chiasma formation at all, as has been observed in the achiasmatic Lepidoptera females. Unpaired univalents have for example been observed during meiosis in crosses between chromosome races of L. sinapis [42]. Another factor that likely can influence the difference in recombination rate between larger and smaller chromosomes is the difference in telomeric and subtelomeric proportions where the recombination rate was observed to be significantly reduced.
We also found that the recombination landscape along many of the larger chromosomes was bimodal, with a tendency towards lower recombination rate in the center. Similar patterns have been observed in other recombination analyses in Lepidoptera [30,32,43], but typically not as pronounced as in other organisms [19,39,44]. There are two potential explanations for this bimodality in recombination frequency along larger chromosomes. First, it is likely that interference will occur when more than one chiasma are formed on a single chromosome [45]. This would likely result in chiasma formation towards chromosome ends, but not all the way towards the telomeres where recombination seems to be inhibited [46]. An alternative, albeit not exclusive explanation, is that double-strand break initiation is directed away from the center and closer towards the telomere regions, as a consequence of physical proximity when the chromosome ends aggregate close to the nuclear membrane (the ‘meiotic bouquet’) during early meiotic stages [11].
Reshuffling rate estimates
Changes in chromosome numbers as a consequence of fissions and fusions may have an effect on the reshuffling rate of genetic variants, since there are more opportunities for independent segregation with an increasing number of chromosomes. We therefore investigated the relative contribution of both independent segregation and recombination to the total reshuffling rate in the two L. sinapis populations. The probability of interchromosomal pairwise reshuffling was close to the maximum (50%) in both populations, which means that the difference in number of chromosome pairs between the Catalan and the Swedish L. sinapis (29 and 52, respectively) has a marginal effect on the difference in reshuffling rate (diminishing return when chromosome numbers exceed n = 25–30; [9]). Notably, most eukaryotes have a diploid chromosome number << 50, with a mean 2n = 17.05 [12] and reshuffling has been proposed to be beneficial for short term distribution of diversity in the offspring [47]. However, the relatively limited range of chromosome numbers in natural populations indicates that there are only minor selective advantages of increasing the chromosome number above a certain threshold. Possible explanations can be that beneficial or co-adapted allele combinations can be inherited together when there are fewer chromosomes and there could possibly be increased risks for segregation errors with higher chromosome numbers. Here we show that the reshuffling between chromosomes contributes much more (orders of magnitude) than intrachromosomal processes to the total reshuffling rate in both populations, confirming the paramount importance of independent assortment of chromosomes for distributing variation in the gametes. Homologous recombination can therefore be seen as a long-term mechanism for disruption of linkage (and Hill-Robertson-interference) between closely positioned loci within chromosomes. It has for example been shown that increasing the number of recombination events is more important than increasing the number of chromosomes for the efficiency of selection [48]. Although we found that the intrachromosomal reshuffling contributed marginally to the total rate, we also found that the pairwise intrachromosomal reshuffling was almost twice as high in the Swedish population compared to the Catalan population. This is obviously an effect of chromosome size, analogous to the positive correlation between genetic map length and intrachromosomal reshuffling rate in some plants [19]. The underlying process must be that more than one crossover can occur on larger chromosomes, and that these crossovers are located distantly enough to increase intrachromosomal reshuffling significantly (double crossovers increase the intrachromosomal pairwise reshuffling if they are far apart—two crossovers located at ¼ of the chromosome length from the terminal results in as high pairwise reshuffling as one crossover in the center). Our data suggest that recombination events are limited by interference mechanisms that both influence the distance between, and the total numbers of, chiasmata occurring on a single chromosome. Interference seems to be ubiquitous among sexually reproducing organisms, but the precise mechanism is not described for many species [45,46]. We also found that chromosomes < 20 Mb roughly have a recombination distance of 50 cM, similar to the situation in H. melpomene where all chromosomes are < 20 Mb [34,49]. These results spur interest in further investigating the mechanistic basis of recombination interference in butterflies in general. In summary, we observed a higher recombination rate in the smaller chromosomes, but this was compensated for by a higher intrachromosomal reshuffling in larger chromosomes. Hence, we proceeded by assessing if these differences and similarities in recombination and reshuffling can affect levels of genetic diversity.
Associations between recombination and genetic variation
To get more information about how chromosomal rearrangements and recombination rate differences between populations and genomic regions are associated with genetic diversity, we estimated a set of population genetic summary statistics at different site categories. Our results show that there was a significantly higher overall genetic diversity in the Catalan population than in the Swedish population. Such a difference could obviously be a consequence of differences in demographic history or other evolutionary processes that affect the populations differently. First, we know that L. sinapis populations inhabiting warmer climatic regions (e.g. Catalonia) are multivoltine with 3–4 generations per year, while the populations at higher latitudes (e.g. Sweden) predominantly are univoltine [50]. A shorter generation time leads to a higher input of novel mutations per time unit, but this should only affect the segregating neutral diversity in the populations differently if they are not at mutation-drift-equilibrium [51]. Second, the demographic histories are likely different between the two populations. While Catalonia likely represents a refugium from the last glaciation period, Sweden must have been colonized well after the ice sheet retracted. This founder event in the north likely occurred with influx of individuals from another refugial population, potentially as far east as central Asia [43,52]. Hence, both the shorter generation time and the absence of obvious founder events could have led to a higher genome-wide level of genetic diversity in the Catalan population.
To further investigate the effects of recombination on patterns of genetic diversity we therefore focused on intra-genomic variation in each population separately. We found that the levels of genetic diversity were heterogeneous across chromosomes and there was no association between genetic diversity and chromosome size after excluding the Z-chromosomes from the analysis. The latter result could be a consequence of the dynamic ongoing karyotype rearrangements in this species, and fits well with linkage-disequilibrium-based recombination rate estimates in the Swedish population [43]. The recurrent fissions and fusions that must have occurred over the divergence time of the populations should mean that both the size of chromosomes and (consequently) the recombination rate has been under constant change, which would reduce the association between current chromosome states, recombination rates and levels of genetic diversity. In addition, a time lag between changes in recombination rate and subsequent changes in genetic diversity is expected, which should dilute the signals further [53]. However, we did actually detect a significant genome-wide association between genetic diversity and recombination rate in both populations. This suggests that the split between the populations occurred a sufficiently long time ago for recombination rate differences to affect genetic diversity. Here, the association was strong at the lower end of the recombination rate spectrum and reached a threshold at moderate recombination rates, with similar levels to those previously detected in Drosophila [54]. This is also in line with theoretical predictions of the association between recombination and genetic variation [35].
The regional distributions of genetic diversity along the chromosomes were in line with the expectations [3], following the trajectory of the recombination landscape in both populations. It should be noted that the chromosomal rearrangements in L. sinapis mainly have involved fusions and/or fissions of comparatively large chromosome parts [29]. The distribution of the recombination landscape on chromosomes could hence be partly maintained even after fusion of two smaller chromosomes or a fission of a larger chromosome—i.e. unimodality in small chromosomes with recombination predominantly occurring at the center and bimodality in larger chromosomes with recombination events towards, but not at, chromosome ends. However, this relative stability obviously depends on the history of the merged chromosome units. Recent fusions, for example, would be expected to have comparatively high genetic diversity as a result of the higher recombination rate on the previously smaller chromosomes, while more ancient fusions would have a lower level of genetic diversity as a consequence of a more long-term recombination rate reduction [33]. It should be noted that fission / fusion polymorphisms are known to be segregating in the populations [29] and that the linkage maps used here are composite maps of different families which may add some uncertainty to the estimates of recombination rates. To further investigate the extent of potential individual differences in recombination rate as a consequence of karyotype variation between families within a specific population, we would need to produce individual recombination maps for males homozygous for each karyotype.
Both cytogenetics [25], linkage mapping and genome assemblies [29] have revealed the existence of an unusual sex-chromosome system in the Leptidea genus where e.g. L. sinapis has a karyotype including three Z-chromosomes and three W-chromosomes. Of the three Z-chromosomes, one is homologous to the inferred ancestral Z-chromosome (Z1) in Lepidoptera while the other two (Z2, Z3) seem to have been recruited as sex-chromosomes in the Leptidea genus specifically [25]. To understand how this relatively new evolution of sex-chromosomes has affected the recombination and diversity landscapes, we compared the different Z-chromosomes to the set of autosomes. The results from this analysis showed that the Z-chromosomes had a lower neutral genetic diversity than autosomes. This is expected given that the Ne of a Z-chromosome should be approximately ¾ of any autosome, under the assumption of equal sex ratios. A reduced level of genetic diversity on the Z-chromosome has also been observed in the Heliconius melpomene and the Monarch butterfly, but not in the moth Manduca sexta [55,56], suggesting that relative Ne of Z-chromosomes and autosomes might vary between species, or that other factors than mutation-drift balance drive relative levels of genetic diversity between chromosome classes. One factor that can contribute to discrepancies between expected and observed levels of genetic diversity on autosomes and sex-chromosomes based on differences in Ne, is the effective recombination rate [57]. Female butterflies are achiasmatic and a Z-chromosome spends only 1/3 of the time in the female germline (in contrast to autosomes that occur in the female germline 50% of time). This means that the effective recombination rate of the Z-chromosome(s) in butterflies should be 4/3 of the autosomes under the assumption that crossovers occur at similar rates for Z-chromosomes and autosomes of similar sizes. Hence, differences in operational sex-ratios, and numbers and relative sizes of autosomes and sex-chromosomes, can potentially drive differences in levels of genetic diversity across different organisms / systems [57]. An interesting result was that while Z1 and Z3 had similar π0 / π4-ratios as the autosomes, Z2 had a dramatically higher π0 / π4-ratio compared to the other chromosomes in both populations. The higher ratio was a consequence of an elevated π0, indicating reduced selective constraints on Z2 (or possibly a group of genes under strong positive selection). This result is somewhat puzzling since Z2 is the most structurally conserved of all chromosomes within Leptidea [26,29] and the observation, although outside of the scope of this article, definitely merits further investigations into the evolutionary forces underlying sex-chromosome formation and maintenance in this genus.
The efficacy of selection translates to both removal of deleterious mutations and fixation rate of adaptive mutations. Hence, under the assumption that most novel mutations are neutral or slightly deleterious, i.e. the nearly neutral theory [58], we expect an increase in the π0 / π4-ratio with reduced efficacy of selection. Since the efficacy of selection is dependent on Ne, the frequency of (slightly) deleterious mutations is expected to segregate at higher rates in regions of low recombination [59]. Our data showed a negative association between the recombination rate and the π0 / π4-ratio in both populations, an observation supporting that physical linkage between sites under selection reduces the efficacy of selection to remove slightly deleterious variants [6]. However, we did not detect any significant difference between the two populations, which would be expected given the difference in overall recombination rate (i.e. a lower Ne in the Swedish population). In addition, there was no detectable effect of the variance in recombination rate in rearranged chromosomes on the ratio of π0 / π4. This indicates that the average recombination rate is high enough to efficiently remove slightly deleterious mutations in both populations, consistent with the observations that selection efficacy can be maintained at a stable level even at relatively low recombination frequencies [60]. In the genome-wide analysis we observed that the π0 / π4-ratio approached the asymptote when the recombination rate was > 2 cM / Mb. Hence, an increase in the recombination rate above that level does not seem to affect the efficacy of selection in this system. It is also possible that the relatively short divergence time between these populations [61], in combination with the extensive chromosome changes, affects our power to detect signals of selection.
We found a positive association between the variation in intergenic diversity change and the variation in recombination rate change across ancestry blocks, with recombination rate explaining > 12% of the variation in intergenic diversity in rearranged chromosomes. A positive association between recombination rate and genetic diversity has been observed in many different taxa [62] and our data lends further support for the claim that recombination is a more potent mechanism than pairwise chromosome reshuffling for breaking linkage disequilibrium and, hence, maintenance of genetic diversity [48]. It is noteworthy that we did not see any significant association between changes in recombination rate in rearranged regions and genetic diversity at other site categories (introns, π0, π4). This is likely a consequence of temporal differences in the effects of reduced linkage disequilibrium on genetic diversity at sites under direct and indirect selection pressure.
Conclusion
We investigated the difference in recombination rate between two populations of L. sinapis with extreme difference in karyotype structure and chromosome count. We show that karyotype evolution directly affects the global recombination rate, but that recombination has a limited effect on total pairwise reshuffling, partly as a consequence of multiple crossovers in larger chromosomes. A key observation from our data was a significant difference in neutral genetic diversity (but not on the efficacy of selection) in regions where chromosomal rearrangements have affected the recombination rate. Finally, we validate the importance of recombination rate variation driving differences in maintenance of nucleotide diversity between populations at an early stage of divergence.
Methods
Linkage map
We used Swedish (n = 6 families; 186 offspring) and Catalan (n = 6 families; 186 offspring) full-sib families of L. sinapis to develop RAD-seq based linkage maps for two populations that represent the most extreme karyotype variants in the species. Details about family structure, RADseq data generation and preprocessing are available in [29]. To get recombination rate information, the linkage map markers were anchored on recently established chromosome-level assemblies for each respective population [29]. The filtered reads were mapped to the genome assembly from each population using bwa mem [63] with default options. The mapped reads were sorted with samtools sort and filtered based on quality (samtools view -q 10), and reads with multiple mapping locations were removed with a custom script [64]. Mapping coverage was assessed with Qualimap [65] and only individuals with > 100,000 mapping reads were included in the study, resulting in 184 and 178 offspring in the Swedish and Catalan pedigrees, respectively. Samtools mpileup [66] with options minimum mapping quality (-q) 10 and minimum base quality (-Q) 10 was used to identify variants, which were converted to genotype likelihoods using Pileup2Likelihoods in LepMap3 with default settings [67]; minimum coverage = 3 per individual (minCoverage = 3) and < 30% of the individuals allowed to have lower coverage than minimum coverage (numLowerCoverage = 0.3). Only markers mapping to chromosome-sized scaffolds in each respective population were retained for downstream analysis. LepMap3 [67] was used to construct linkage maps for the two populations separately applying the following steps. Informative markers were identified with ParentCall2 using default settings, except for ZLimit which was set to 2 to identify markers segregating as sex chromosomes. Non-informative markers were removed with removeNonInformative = 1. Markers showing segregation distortion, minimum allele frequency below 0.05 or absence in more than 50% of the individuals were excluded with Filtering2 (options dataTolerance = 0.00001, MAFLimit = 0.05, missingLimit = 0.5). In addition, only markers represented in at least four families in each population were retained (familyInformativeLimit = 4). As a first mapping step, markers were assigned to linkage groups with SeparateChromosomes2 using female informative markers only (informativeMask = 2). The optimal LOD-score threshold was empirically estimated to 11 for the Swedish and 7 for the Catalan population. Markers that were informative in either males or both sexes were subsequently added with JoinSingles2All with the same LOD-score thresholds. Linkage groups covering several scaffolds were split based on information from the physical assembly before ordering the markers along the chromosomes with the module Ordermarkers2. Only male informative markers (informativeMask = 1) were used for the ordering since females are achiasmatic (recombination2 = 0). We set a penalty for diverting from the order in the genome assembly with likelihood chain (usePhysical = 1 0.1) and improveOrder = 0. The ordered linkage maps were visually inspected in R and uninformative markers (extending the map ends) were removed. After manual removal of uninformative markers, OrderMarkers2 was run again with the same settings (see above) and the final genetic distances were calculated using Kosambi’s map function to account for multiple recombination events along the chromosome.
Recombination estimation
Regional recombination rates were estimated by dividing the genetic distance with the physical interval between each marker pair (i.e. cM / Mb). Weighted means for 2 Mb windows, per chromosome and per population were obtained by weighting the recombination rate at each interval by the proportion of the physical distance. Interpopulation comparisons of the weighted mean recombination rates were performed with Wilcoxon Signed rank test and potential associations between recombination rate and chromosome length were assessed with Spearman’s rank correlation coefficient, as implemented in cor.test in R [68]. To compare the observed recombination rate with the expected recombination rate for exactly one or two crossovers per chromosome per meiosis, we calculated the expected rates as 50 cM / chromosome length and 100 cM / chromosome length, respectively. Note that potential recombination events resolved as non-crossovers and / or gene conversion events associated with the double-strand break repair in recombination tracts are undetectable with the marker data used here and therefore not addressed.
Pairwise reshuffling
The probability of total pairwise reshuffling is the probability that two random loci are originating from different parental alleles during gamete formation. The total pairwise reshuffling has two components; the interchromosomal component describing the effects of independent assortment and the intrachromosomal component describing the effect of crossing over. The heterogeneity in the spatial distribution of crossovers is also corrected for. The probability of interchromosomal pairwise reshuffling was estimated by the probability of two loci being located on different chromosomes with independent assortment of the chromosomes during meiosis [9]. The intrachromosomal probability of pairwise reshuffling per chromosome was inferred using the genetic and physical distances from the linkage maps. 1,000 pseudomarkers were placed along each chromosome and the number of pairwise differences in origin was calculated based on the probability of recombination between each pair of pseudomarkers [9]. The sum of pairwise differences was divided by the total number of comparisons for each chromosome. The intrachromosomal contribution to the total reshuffling rate was estimated by summing up the contribution of each chromosome, weighted by the probability that two loci are located on the same chromosome (chromosome length / assembly length)2. The total pairwise reshuffling was calculated as the sum of the interchromosomal and intrachromosomal contribution [9].
Population genetic summary statistics
Previously available resequencing data from 10 Swedish and 10 Catalan individuals were used for population genetic analysis (Talla et al., 2019). Positions with Phred-score < 33, sequencing adapters and seven bases from the 3’-ends of all reads were trimmed using TrimGalore 0.6.1 (https://github.com/FelixKrueger/TrimGalore), a wrapper for cutadapt 3.1 [69]. Trimmed reads were mapped to the genome assemblies for each respective population [29] using bwa mem [63]. Variant calling was performed using GATK 4.2.0.0 [70]. Base quality score recalibration was performed using an initial variant calling round, keeping a set of variants with good support as recommended by GATK [71]. Subsequently, a second variant calling round was performed on recalibrated mapped reads. A final filtering step of all sites (variant and invariant sites) was applied using BCFtools filter 1.16–1 [72] to mask genotypes overlapping annotated repeats and genotypes with a coverage < 5 or > 25. The applied options for these filtering steps were—mask-file $REPEAT_GFF,—soft-filter "REPEAT", and -i FMT/DP > 5 & FMT/DP < 25’—set-GTs.
Nucleotide diversity (π) for intronic, intergenic, 0-fold and 4-fold sites were estimated in 50 kb windows with the software pixy 1.2.5.beta [73]. Coordinates for 0-fold and 4-fold sites were obtained with a custom script (modified from https://github.com/simonhmartin/genomics_general) and genomic features (intron, intergenic) were extracted using BEDTools 2.29.2 complement [74]. Potential associations between π and chromosome length were assessed with Spearman’s rank correlation, as implemented in cor.test in R [68]. We used the ratio of π0 / π4 as a proxy for selection, reflecting the efficacy in removal of deleterious mutations, since most protein changing mutations are either deleterious or neutral according to the nearly neutral theory [58].
Potential associations between the recombination rate and π were assessed by binning data into 2 Mb genomic windows for each population separately. Population genetic summary statistics were also estimated in ancestry blocks (ancestral chromosomal units), i.e. previously identified homologous regions of chromosomes in which no rearrangements have occurred between the populations [29]. We included ancestral chromosomal units involved in fissions/fusions after the split from the closest sister species (Leptidea reali), or where incomplete lineage sorting or reuse of breakpoints have resulted in a fission in one population or a fusion in the other. To investigate if observed changes in recombination rate between populations have resulted in a corresponding change in π, we compared the ratio of the Catalan and the Swedish π with the observed recombination rate ratio between the populations within each ancestral chromosomal unit. Both the recombination rate and the π estimates are random variables, so we applied a model II linear regression using ordinary least squares (OLS) and a permutation test with 1,000 permutations to determine the significance of the slopes, as implemented in the R-package lmodel2 [75].
Supporting information
S1 Fig.
Marey maps for the Catalan (n = 53, top) and the Swedish (n = 29, bottom) population of L. sinapis. Genetic distances are given on the y-axes (cM) and physical positions along each respective chromosome on the x-axes (Mb).
https://doi.org/10.1371/journal.pgen.1010717.s001
(PDF)
S2 Fig.
Regional recombination rates (cM, y-axis) as estimated in 2 Mb windows using local regression (LOESS) along the chromosomes in the Catalan and the Swedish population, respectively. Physical position is given on the x-axis (Mb).
https://doi.org/10.1371/journal.pgen.1010717.s002
(PDF)
S3 Fig.
Population specific pairwise differences (π and bottom panel π0 / π4) per chromosome (y-axis) and chromosomal length (x-axis) for each diversity class. Spearman’ rank correlation coefficient including Z-chromosomes (purple) and without Z-chromosomes (black).
https://doi.org/10.1371/journal.pgen.1010717.s003
(PDF)
S4 Fig.
a) Nucleotide diversity (π) as a function of recombination rate in each population estimated in 2Mb windows in zero-fold, four-fold degenerate sites, intergenic and intronic sites for each population. b) The ratio of polymorphisms at 0-fold and 4-fold degenerate sites (π0 / π4) as a function of the recombination rate in each population. Diversity was estimated in 2 Mb windows. Lines represent local regressions, statistics represent results from a model II linear regression, and permuted p-values for the OLS slope are given in black font. The statistics for windows with a recombination rate 2.0 cM / Mb or below and > 2.0 cM / Mb are given in blue and grey font, respectively.
https://doi.org/10.1371/journal.pgen.1010717.s004
(PDF)
S5 Fig.
The ratio between Catalan and Swedish estimates of a) diversity (Ratioπ) and b) genetic diversity at 0-fold to 4-fold degenerate sites (Ratioπ0/π4), as a function of the ratio between Catalan and Swedish recombination rate for each ancestral chromosomal unit with known fusion/fission history. The line represents the slope from model II linear regression, with the estimated R2 and permuted p-value.
https://doi.org/10.1371/journal.pgen.1010717.s005
(PDF)
S1 Table.
Mean weighted recombination rate (Rec rate) per chromosome in the Swedish (Swe) and Catalan (Cat) populations. Chromosome length in Mega base-pair (Chr length Mb), genetic map length (cM) and number of markers (Mkr no) per chromosome were also included.
https://doi.org/10.1371/journal.pgen.1010717.s006
(PDF)
Acknowledgments
The authors acknowledge support from the National Genomics Infrastructure in Stockholm funded by Science for Life Laboratory, the Knut and Alice Wallenberg Foundation and the Swedish Research Council, and SNIC/Uppsala Multidisciplinary Center for Advanced Computational Science for assistance with massively parallel sequencing and access to the UPPMAX computational infrastructure.
References
- 1. Charlesworth B, Morgan MT, Charlesworth D. The effect of deleterious mutations on neutral molecular variation. Genetics. 1993;134: 1289–1303. pmid:8375663
- 2. Maynard Smith J, Haigh J. The hitch-hiking effect of a favourable gene. Genet Res. 1974;23: 23–35. pmid:4407212
- 3. Begun DJ, Aquadro CF. Levels of naturally occurring DNA polymorphism correlate with recombination rates in D. melanogaster. Nature. 1992;356: 519–520. pmid:1560824
- 4. Burri R, Nater A, Kawakami T, Mugal CF, Olason PI, Smeds L, et al. Linked selection and recombination rate variation drive the evolution of the genomic landscape of differentiation across the speciation continuum of Ficedula flycatchers. Genome Res. 2015;25: 1656–1665. pmid:26355005
- 5. Cutter AD, Baird SE, Charlesworth D. High nucleotide polymorphism and rapid decay of linkage disequilibrium in wild populations of Caenorhabditis remanei. Genetics. 2006;174: 901–913. pmid:16951062
- 6. Hill WG, Robertson A. Linkage disequilibrium in finite populations. Theoret Appl Genetics. 1968;38: 226–231. pmid:24442307
- 7. Rodgers-Melnick E, Bradbury PJ, Elshire RJ, Glaubitz JC, Acharya CB, Mitchell SE, et al. Recombination in diverse maize is stable, predictable, and associated with genetic load. Proc Natl Acad Sci USA. 2015;112: 3823–3828. pmid:25775595
- 8.
Klug WS, Cummings MR, Spencer CA, Palladino MA. Essentials of Genetics. 9th edition. Essentials of Genetics. 9th edition. Essex, England: Pearson Education Limited; 2017.
- 9. Veller C, Kleckner N, Nowak MA. A rigorous measure of genome-wide genetic shuffling that takes into account crossover positions and Mendel’s second law. Proc Natl Acad Sci USA. 2019;116: 1659–1668. pmid:30635424
- 10.
Korol AB, Preygel IA, Preygel SI. Recombination variability and evolution: algorithms of estimation and population-genetic models. 1. ed. London Weinheim: Chapman & Hall; 1994.
- 11. Haenel Q, Laurentino TG, Roesti M, Berner D. Meta-analysis of chromosome-scale crossover rate variation in eukaryotes and its significance to evolutionary genomics. Mol Ecol. 2018;27: 2477–2497. pmid:29676042
- 12. Stapley J, Feulner PGD, Johnston SE, Santure AW, Smadja CM. Variation in recombination frequency and distribution across eukaryotes: patterns and processes. Philos Trans R Soc Lond B Biol Sci. 2017;372: 20160455. pmid:29109219
- 13. Wilfert L, Gadau J, Schmid-Hempel P. Variation in genomic recombination rates among animal taxa and the case of social insects. Heredity. 2007;98: 189–197. pmid:17389895
- 14. Lynch M, Bobay L-M, Catania F, Gout J-F, Rho M. The Repatterning of Eukaryotic Genomes by Random Genetic Drift. Annu Rev Genom Hum Genet. 2011;12: 347–366. pmid:21756106
- 15. Pardo-Manuel De Villena F, Sapienza C. Recombination is proportional to the number of chromosome arms in mammals. Incorporating Mouse Genome. 2001;12: 318–322. pmid:11309665
- 16. Morgan TH. Complete linkage in the second chromosome of the male of Drosophila. Science. 1912;36: 719–720.
- 17. Suomalainen E, Cook LM, Turner JRG. Achiasmatic oogenesis in the Heliconiine butterflies. Hereditas. 1973;1973: 302–304.
- 18. Turner JRG, Sheppard PM. Absence of crossing-over in female butterflies (Heliconius). Heredity. 1975;1975: 265–269.
- 19. Brazier T, Glémin S. Diversity and determinants of recombination landscapes in flowering plants. PLoS Genet. 2022;18: e1010141. pmid:36040927
- 20. Storey AJ, Wang H-P, Protacio RU, Davidson MK, Tackett AJ, Wahls WP. Chromatin-mediated regulators of meiotic recombination revealed by proteomics of a recombination hotspot. Epigenetics Chromatin. 2018;11: 64. pmid:30373637
- 21. Wang Y, Zhai B, Tan T, Yang X, Zhang J, Song M, et al. ESA1 regulates meiotic chromosome axis and crossover frequency via acetylating histone H4. Nucleic Acids Research. 2021;49: 9353–9373. pmid:34417612
- 22. Vara C, Paytuví-Gallart A, Cuartero Y, Álvarez-González L, Marín-Gual L, Garcia F, et al. The impact of chromosomal fusions on 3D genome folding and recombination in the germ line. Nat Commun. 2021;12: 2981. pmid:34016985
- 23. de Vos JM, Augustijnen H, Bätscher L, Lucek K. Speciation through chromosomal fusion and fission in Lepidoptera. Phil Trans R Soc B. 2020;375: 20190539. pmid:32654638
- 24. Lukhtanov VA, Dincă V, Talavera G, Vila R. Unprecedented within-species chromosome number cline in the Wood White butterfly Leptidea sinapis and its significance for karyotype evolution and speciation. BMC Evol Biol. 2011;11: 109. pmid:21507222
- 25. Šíchová J, Ohno M, Dincă V, Watanabe M, Sahara K, Marec F. Fissions, fusions, and translocations shaped the karyotype and multiple sex chromosome constitution of the northeast-Asian wood white butterfly, Leptidea amurensis. Biol J Linn Soc. 2016;118: 457–471.
- 26. Yoshido A, Šíchová J, Pospíšilová K, Nguyen P, Voleníková A, Šafář J, et al. Evolution of multiple sex-chromosomes associated with dynamic genome reshuffling in Leptidea wood-white butterflies. Heredity. 2020;125: 138–154. pmid:32518391
- 27. Melters DP, Paliulis LV, Korf IF, Chan SWL. Holocentric chromosomes: convergent evolution, meiotic adaptations, and genomic analysis. Chromosome Res. 2012;20: 579–593. pmid:22766638
- 28. Dernburg AF. Here, There, and Everywhere: Kinetochore Function on Holocentric Chromosomes. J Cell Biol. 2001;153: F33–F38. pmid:11402076
- 29. Höök L, Näsvall K, Vila R, Wiklund C, Backström N. High-density linkage maps and chromosome level genome assemblies unveil direction and frequency of extensive structural rearrangements in wood white butterflies (Leptidea spp.). Chromosome Res. 2023;31: 2. pmid:36662301
- 30. Shipilina D, Näsvall K, Höök L, Vila R, Talavera G, Backström N. Linkage mapping and genome annotation give novel insights into gene family expansions and regional recombination rate variation in the painted lady (Vanessa cardui) butterfly. Genomics. 2022;114: 110481. pmid:36115505
- 31. Mackintosh A, Laetsch DR, Hayward A, Charlesworth B, Waterfall M, Vila R, et al. The determinants of genetic diversity in butterflies. Nat Commun. 2019;10: 3466. pmid:31371715
- 32. Martin SH, Davey JW, Salazar C, Jiggins CD. Recombination rate variation shapes barriers to introgression across butterfly genomes. Moyle L, editor. PLoS Biol. 2019;17: e2006288. pmid:30730876
- 33. Cicconardi F, Lewis JJ, Martin SH, Reed RD, Danko CG, Montgomery SH. Chromosome fusion affects genetic diversity and evolutionary turnover of functional loci but consistently depends on chromosome size. Mol Biol Evol. 2021;38: 4449–4462. pmid:34146107
- 34. Davey JW, Barker SL, Rastas PM, Pinharanda A, Martin SH, Durbin R, et al. No evidence for maintenance of a sympatric Heliconius species barrier by chromosomal inversions. Evol Lett. 2017;1: 138–154. pmid:30283645
- 35. Cutter AD, Payseur BA. Genomic signatures of selection at linked sites: unifying the disparity among species. Nat Rev Genet. 2013;14: 262–274. pmid:23478346
- 36. Pool JE, Nielsen R. Population size changes reshape genomic patterns of diversity. Evolution. 2007;61: 3001–3006. pmid:17971168
- 37. Yasukochi Y, Ashakumary LA, Baba K, Yoshido A, Sahara K. A second-generation integrated map of the silkworm reveals synteny and conserved gene order between lepidopteran insects. Genetics. 2006;173: 1319–1328. pmid:16547103
- 38. Lohse K, Höök L, Näsvall K, Backström N, Programme WSIT of L, Collective WSISODP, et al. The genome sequence of the wood white butterfly, Leptidea sinapis (Linnaeus, 1758). Wellcome Open Research; 2022. pmid:37025368
- 39. Barnes TM, Kohara Y, Coulson A, Hekimi S. Meiotic recombination, noncoding DNA and genomic organization in Caenorhabditis elegans. Genetics. 1995;141: 159–179. pmid:8536965
- 40. Nokkala S, Kuznetsova VG, Maryanska-Nadachowska A, Nokkala C. Holocentric chromosomes in meiosis. I. Restriction of the number of chiasmata in bivalents. Chromosome Res. 2004;12: 733–739. pmid:15505408
- 41. Hillers KJ, Villeneuve AM. Chromosome-Wide Control of Meiotic Crossing over in C. elegans. Current Biology. 2003;13: 1641–1647. pmid:13678597
- 42. Lukhtanov VA, Dincă V, Friberg M, Šíchová J, Olofsson M, Vila R, et al. Versatility of multivalent orientation, inverted meiosis, and rescued fitness in holocentric chromosomal hybrids. Proc Natl Acad Sci USA. 2018;115: E9610–E9619. pmid:30266792
- 43. Palahí i Torres A, Höök L, Näsvall K, Shipilina D, Wiklund C, Vila R, et al. The fine-scale recombination rate variation and associations with genomic features in a butterfly. bioRxiv (preprint). 2022 [cited 21 Nov 2022].
- 44. Rockman MV, Kruglyak L. Recombinational landscape and population genomics of Caenorhabditis elegans. PLoS Genet. 2009;5: e1000419. pmid:19283065
- 45. Bishop DK, Zickler D. Early decision: meiotic crossover interference prior to stable strand exchange and synapsis. Cell. 2004;117: 9–15. pmid:15066278
- 46. Otto SP, Payseur BA. Crossover interference: Shedding light on the evolution of recombination. Annu Rev Genet. 2019;53: 19–44. pmid:31430178
- 47. Maynard Smith J. A short-term advantage for sex and recombination through sib-competition. J Theor Biol. 1976;63: 245–258. pmid:1011844
- 48. Burt A. Perspective: Sex, recombination, and the efficacy of selection—Was Weismann right? Evolution. 2000;54: 337–351. pmid:10937212
- 49. Davey JW, Chouteau M, Barker SL, Maroja L, Baxter SW, Simpson F, et al. Major improvements to the Heliconius melpomene genome assembly used to confirm 10 chromosome fusion events in 6-Million years of butterfly evolution. 2016; 14. pmid:26772750
- 50. Friberg M, Bergman M, Kullberg J, Wahlberg N, Wiklund C. Niche separation in space and time between two sympatric sister species—a case of ecological pleiotropy. Evol Ecol. 2008;22: 1–18.
- 51. Kimura M. Evolutionary rate at the molecular level. Nature. 1968;217: 624–626. pmid:5637732
- 52. Talla V, Soler L, Kawakami T, Dincă V, Vila R, Friberg M, et al. Dissecting the effects of selection and mutation on genetic diversity in three wood white (Leptidea) butterfly species. Gonzalez J, editor. Genome Biol Evol. 2019;11: 2875–2886. pmid:31580421
- 53. Booker TR, Payseur BA, Tigano A. Background selection under evolving recombination rates. Proc R Soc B Biol. 2022;289: 20220782. pmid:35730151
- 54. Castellano D, Coronado-Zamora M, Campos JL, Barbadilla A, Eyre-Walker A. Adaptive evolution is substantially impeded by Hill–Robertson interference in Drosophila. Mol Biol Evol. 2016;33: 442–455. pmid:26494843
- 55. Martin SH, Möst M, Palmer WJ, Salazar C, McMillan WO, Jiggins FM, et al. Natural selection and genetic diversity in the butterfly Heliconius melpomene. Genetics. 2016;203: 525–541. pmid:27017626
- 56. Mongue AJ, Hansen ME, Walters JR. Support for faster and more adaptive Z chromosome evolution in two divergent lepidopteran lineages. Evolution. 2022;76: 332–345. pmid:34463346
- 57. Veller C, Edelman NB, Muralidhar P, Nowak MA. Recombination and selection against introgressed DNA. Evolution. 2023;77: 1131–1144. pmid:36775972
- 58. Ohta T. Role of very slightly deleterious mutations in molecular evolution and polymorphism. Theor Popul Biol. 1976;10: 254–275. pmid:1013905
- 59. Galtier N. Adaptive protein evolution in animals and the effective population size hypothesis. Schierup MH, editor. PLoS Genet. 2016;12: e1005774. pmid:26752180
- 60. Haddrill PR, Halligan DL, Tomaras D, Charlesworth B. Reduced efficacy of selection in regions of the Drosophila genome that lack crossing over. Genome Biol. 2007;8: R18. pmid:17284312
- 61. Talla V, Johansson A, Dincă V, Vila R, Friberg M, Wiklund C, et al. Lack of gene flow: Narrow and dispersed differentiation islands in a triplet of Leptidea butterfly species. Mol Ecol. 2019;28: 3756–3770. pmid:31325366
- 62. Smukowski CS, Noor M a. F. Recombination rate variation in closely related species. Heredity. 2011;107: 496–508. pmid:21673743
- 63. Li H. Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM. 2013 [cited 23 Mar 2022].
- 64. Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, et al. The Sequence Alignment/Map format and SAMtools. Bioinformatics. 2009;25: 2078–2079. pmid:19505943
- 65. Okonechnikov K, Conesa A, García-Alcalde F. Qualimap 2: advanced multi-sample quality control for high-throughput sequencing data. Bioinformatics. 2015; btv566. pmid:26428292
- 66. Li H. A statistical framework for SNP calling, mutation discovery, association mapping and population genetical parameter estimation from sequencing data. Bioinformatics. 2011;27: 2987–2993. pmid:21903627
- 67. Rastas P. Lep MAP3: robust linkage mapping even for low-coverage whole genome sequencing data. Berger B, editor. Bioinformatics. 2017;33: 3726–3732. pmid:29036272
- 68.
R Core Team. R: A language and environment for statistical computing. In: R Foundation for Statistical Computing, Vienna, Austria. [Internet]. 2021 [cited 23 Mar 2022]. Available: http://www.r-project.org/index.html
- 69. Martin M. Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet j. 2011;17: 10.
- 70.
Auwera G van der, O’Connor BD. Genomics in the cloud: using Docker, GATK, and WDL in Terra. First edition. Sebastopol, CA: O’Reilly Media; 2020.
- 71. DePristo MA, Banks E, Poplin R, Garimella KV, Maguire JR, Hartl C, et al. A framework for variation discovery and genotyping using next-generation DNA sequencing data. Nat Genet. 2011;43: 491–498. pmid:21478889
- 72. Danecek P, Bonfield JK, Liddle J, Marshall J, Ohan V, Pollard MO, et al. Twelve years of SAMtools and BCFtools. Gigascience. 2021;10: giab008. pmid:33590861
- 73. Korunes KL, Samuk K. pixy: Unbiased estimation of nucleotide diversity and divergence in the presence of missing data. Mol Ecol Resour. 2021;21: 1359–1368. pmid:33453139
- 74. Quinlan AR, Hall IM. BEDTools: a flexible suite of utilities for comparing genomic features. Bioinformatics. 2010;26: 841–842. pmid:20110278
- 75. Legendre P. lmodel2: Model II Regression. In: CRAN.R-project.org [Internet]. 2018. Available: https://CRAN.R-project.org/package=lmodel2