Fig 1.
Mean recombination rates per chromosome (cM/Mb, log scale) are negatively correlated with chromosome physical size (Mb, log scale).
Each point represents a chromosome (n = 665). Species are presented in different colours (57 species). (A) The bold solid line represents the linear regression line fitted to the data. The thin lines correspond to the expectation of one, two, three or four COs per chromosome. (B) Correlations between recombination rates and chromosome size within each species with at least 5 chromosomes (coloured lines, 55 species) and the overall between-species correlation controlled for a species effect (black dashed line, n = 57 species). Solid bold line as in (A).
Fig 2.
Linkage map length (cM) is positively correlated with genomic chromosome size (Mb).
(A) Correlation between chromosome genomic size (Mb) and linkage map length (cM). Each point represents a chromosome (n = 665). Species are presented in different colours (57 species). The black solid line represents the simple linear regression (linkage map length ~ log10(chromosome size), adjusted R2 = 0.036, p < 0.001) and the black dashed line the fixed effect of the mixed model (linkage map length ~ log10(chromosome size) + (1 | species), marginal R2 = 0.49, conditional R2 = 0.99, p < 0.001). Species random slopes are shown in colours. Isolines of recombination rates are plotted for different values (indicated cM/Mb) as dotted red lines to represent regions with equal recombination. (B) The excess of COs (linkage map length minus 50 cM for the obligate CO) is positively correlated with the relative chromosome size (size / average size of the species). The black solid line is the linear regression across species (excess of CO ~ relative chromosome size, adjusted R2 = 0.13, p < 0.001) and the black dashed line the fixed effect of the mixed model (excess of CO ~ relative chromosome size + (1 | species), marginal R2 = 0.14, conditional R2 = 0.86, p < 0.001). Coloured solid lines represent individual regression lines for species with at least 5 chromosomes (55 species).
Fig 3.
Diversity of recombination landscapes exemplified by six different species.
Recombination landscapes are similar within species (the dashed line is the average landscape for pooled chromosomes, all recombination landscapes of the species are contained within the colour ribbon). Genomic distances (Mb) were scaled between 0 and 1 to compare chromosomes with different sizes. Estimates of the recombination rates were obtained by 1,000 bootstraps over loci in windows of 100 kb with loess regression and automatic span calibration. One chromosome per species is represented in a solid line, with the genomic position of the centromere demarcated by a dot. The six species are ordered by ascending mean chromosome size (Mb).
Fig 4.
Patterns of recombination within chromosomes (n = 665).
Relative recombination rates along the chromosome were estimated in ten bins of equal ratio of the observed genetic length divided by the expected genetic length (one tenth of total size) of the bin (log-transformed). Values below (above) zero are recombination rates that are lower (higher) than expected under a random distribution. The 57 species are ordered by ascending genome size. Each horizontal bar plot represents one chromosome. When available, the centromere position is mapped as a black and white diamond.
Fig 5.
The periphery-bias ratio is positively correlated with chromosome genomic size.
(A) Linear regression between the species mean periphery-bias ratio and the mean chromosome size (log scale) across species (n = 57 species; adjusted R2 = 0.44, p < 0.001). Points are coloured according to the classification of the CO patterns described below (orange = distal, blue = sub-distal, black = unclassified). (B) Distribution of periphery-bias ratios (n = 665 chromosomes). The mean periphery-bias ratio and its 95% confidence interval (black solid and dashed lines) were estimated by 1,000 bootstrap replicates. The red vertical line corresponds to a ratio of one. (C) Distribution of Spearman’s correlation coefficients between the periphery-bias ratio and chromosome genomic size (Mb) within species (n = 57 species).
Fig 6.
Distribution of crossover: main patterns.
(A and B) Standardized recombination rates for species (chromosomes pooled per species, n = 57 species) are expressed as a function of the relative genomic distance from the telomere in 20 bins representing the two main patterns (orange = distal, blue = sub-distal). The seven unclassified species are shown in supplementary (S7 Fig). Chromosomes were split in half and 0.5 corresponds to the centre of the chromosome. In each plot, the solid line represents the mean recombination rate estimated in a bin (20 bins) and each dot per bin represents the average of a species. Upper and lower boundaries of the ribbon represent the maximum and minimum values. (C) Distribution of chromosome genomic sizes (Mb) for each pattern (D: distal, SD: sub-distal, E: exceptions).
Fig 7.
Possible models of crossover patterns.
Schematic representation of the three competing models for the two main patterns, with an example of a centromere position at 1/3 of the chromosome. Model 3 is the best model (box).
Table 1.
Model selection for the telomere/centromere effect (n = 37 species with a centromere position, 425 chromosomes).
Three competing models were compared based on the adjusted R2, p-value and AIC-BIC criteria among chromosomes (the best supported model is in bold characters). The number of species supporting each model was calculated based on the adjusted R2 within species, for all species with at least five chromosomes. (1) ‘telomere’ model. (2) ‘telomere + centromere + one CO per arm’ model. (3) ‘telomere + centromere + one CO per chromosome’ model. d(c) is the genetic distance to the centromere. d(1) is the total genetic distance. A second model selection was done on a subset of chromosomes with at least 50 cM on each chromosome arm (n = 36 species, 333 chromosomes).
Fig 8.
Recombination rates are positively correlated with gene density (n = 483 chromosomes, 41 species).
(A) Distribution of chromosome Spearman’s rank correlations between the number of genes and the recombination rate in 100 kb windows. The black vertical line is the mean correlation with a 95% confidence interval (dashed lines) estimated by 1,000 bootstrap replicates. Colours correspond to CO patterns (orange = distal, blue = sub-distal, black = exception). (B) Slopes of the species linear regression between gene count and recombination rates are independent of the species averaged recombination rate (Linear Model, adjusted R2 = -0.02, p = 0.83). (C) Standardized recombination rates for each number of genes in a 100 kb window (centred-reduced, chromosomes pooled per species) estimated by 1,000 bootstraps and standardized within species. The gene count was estimated by counting the number of gene starting positions within each 100 kb window. The black line with a grey ribbon is the quadratic regression estimated by linear regression with a 95% parametric confidence interval (Linear Model, adjusted R2 = 0.62, p < 0.001).
Fig 9.
Gene counts patterns along the chromosome are correlated with CO patterns (n = 41 species).
Standardized gene count (centred-reduced) as a function of the relative distance from the tip to the middle of the chromosome (genomic distances distributed in 20 bins). We used the same groups as identified for the CO pattern in Fig 6; (A) distal pattern vs (B) sub-distal pattern. Same legend as Fig 6.
Fig 10.
Marey maps of six chromosomes with the relative physical distance expressed in genomic distances (black dots, position in the genome in Mb) or in gene distances (grey dots, position measured as the cumulative number of genes along the chromosome).
Marey maps are ordered by ascending chromosome size (Mb). The diagonal dashed line represents a theoretical random distribution of COs along the chromosome.
Fig 11.
Differences in genetic shuffling between estimates based on genomic distances (Mb) and gene distances (cumulative number of genes).
The difference is the genetic shuffling in gene distances minus the genetic shuffling in genomic distances. Colours correspond to CO patterns (orange = distal, blue = sub-distal, black = exception). (A) Distribution of the chromosome differences in the genetic shuffling (n = 444 chromosomes). (B) Distributions of the species difference in the genetic shuffling (n = 41 species, chromosomes pooled). (C) Species differences in the genetic shuffling are positively correlated with the averaged chromosome size (Linear Model, adjusted R2 = 0.20, p = 0.002, n = 41, 95% parametric confidence interval).