Evidence of intense chromosomal shuffling during conifer evolution.

While recent advances have been gained on genome evolution in angiosperm lineages, virtually nothing is known about karyotype evolution in the other group of seed plants, the gymnosperms. Here we used high density gene-based linkage mapping to compare the karyotype structure of two families of conifers (the most abundant group of gymnosperms) separated around 290 million years ago: Pinaceae and Cupressaceae. We propose for the first time a model based on the fusion of 20 ancestral chromosomal blocks that may have shaped the modern karyotpes of Pinaceae (with n=12) and Cupressaceae (with n=11). The considerable difference in modern genome organization between these two lineages contrasts strongly with the remarkable level of synteny already reported within the Pinaceae. It also suggests a convergent evolutionary mechanism of chromosomal block shuffling that has shaped the genomes of the spermatophytes.

Introduction

Knowledge about genome structure and evolution is a fundamental step toward understanding species adaptation and evolution. Genome evolution is based on genetic variability generated by mutation, recombination, and the acquisition of new genes. New genes can be acquired by interspecific hybridization or the duplication of some or all the existing genes of an organism (Renny-Byfield and Wendel 2014; Cong et al. 2015). Plant genomes, unlike those of animals, have evolved through frequent, rapid chromosomal rearrangements, including whole-genome duplications (WGDs) followed by nested chromosome fusions in particular (Abrouk et al. 2010; Salse et al. 2015). The sequencing of plant genomes has made it possible to construct evolutionary models for various angiosperm lineages (Murat et al. 2010; Salse 2012; Murat et al. 2015). These studies revealed that angiosperms have experienced successive common and lineage-specific WGDs, which have governed increases in genome size and shaped the genome structure and composition of extant species (Renny-Byfield and Wendel 2014). Evolutionary genome shuffling events (chromosomal fusions and fissions) have been identified during the course of angiosperm evolutionary history, making it possible to reconstruct the karyotypes of the common ancestors of eudicots, with seven protochromosomes, and of monocots, with five or seven protochromosomes (Abrouk et al. 2010; Salse 2012). However, we still know little about karyotype evolution in the other group of seed plants, the gymnosperms.

Conifers are the most abundant group of gymnosperms. Genome structure and evolution differ between conifers and angiosperms. Conifers have extremely large genomes (ranging from 18 to 35 Gb) characterized by the presence of repetitive elements (Kovach et al. 2010; Mackay et al. 2012; Neale et al. 2014). These features have hindered attempts to sequence the genomes of these plants and the recently released draft genome sequences are highly fragmented (Nystedt et al. 2013; Zimin et al. 2014; Warren et al. 2015). Consequently, the evolutionary mechanisms shaping the composition and structure of conifer genomes remain to be determined. One ancient WGD event is known to have occurred before the angiosperm–gymnosperm split around 350 Ma (Jiao et al. 2011). However, there is no evidence to suggest that other WGD events have occurred during the evolutionary history of conifers (Kovach et al. 2010; Nystedt et al. 2013), by contrast to what has been reported for angiosperms. Conifer genome size seems to have increased due to the accumulation of large numbers of retrotransposons (Morse et al. 2009; Nystedt et al. 2013). Polyploidy is exceptional in gymnosperms, with only two species from the Cupressaceae known to be polyploids, one of these species being hexaploid (Sequoia semperviens 2n = 66) and the other tetraploid (Juniperus chinensis 2n = 44). The haploid number of chromosomes in conifers ranges from 9 to 19, but karyotypes are highly conserved across species and genera, with most having 11 or 12 chromosomes (Wang and Ran 2014).

Pinaceae, the largest family of conifers, has been studied more thoroughly than other conifers, for ecological and economic reasons. Pinus and Picea, the main genera within Pinaceae, separated around 87–193 Ma (Morse et al. 2009). Comparative mapping between Pinaceae species has revealed high levels of interspecific and intergeneric synteny and macrocollinearity (Krutovsky et al. 2004; Pelgas et al. 2006; Pavy et al. 2012), suggesting a lack of chromosomal rearrangement within this family, despite the ancient nature of the divergence between some taxa. Nevertheless, the issue of the conservation of synteny across different families of conifers has not yet been addressed. Further studies of the evolution of conifer karyotypes are therefore required to determine whether it has followed a pattern similar to that in angiosperms or more similar to that in the Pinaceae. In the absence of a completely contiguous reference genome in conifers, high-density comparative mapping provides an opportunity to compare genomes from different lineages, thereby improving our understanding of conifer karyotype evolution.

In this work, we analyzed conifer karyotype evolution through comparative mapping, making use of published genetic linkage maps for two families: Pinaceae (n = 12) and Cupressaceae (n = 11). The aims of this study were 1) to analyze the degree of gene synteny and collinearity at the interfamily level, and 2) to set a likely scenario of karyotype evolution between both families.

Results and Discussion

We carried out a literature review to select high-density gene-based linkage maps for conifers for which sequence information is publicly available. We included 18 genetic maps (table 1) for six different species from two botanical families—Pinaceae (n = 12) and Cupressaceae (n = 11)—in this study. We made use of the high degree of synteny and macrocollinearity within Pinaceae (Krutovsky et al. 2004; Pavy et al. 2012) to establish a gene-based composite map for this botanical family. A stepwise strategy, from species to family level, was used to maximize the number of mapped markers common to different maps, thereby maximizing the number of anchor markers for the construction of the composite map for Pinaceae (fig. 1a). We began by constructing a composite linkage map for Pinus pinaster from 14 maps (table 1). We then generated two genus-level composite maps: 1) For Pinus sp., by combining the P. pinaster composite map generated in this study with a published map for Pinus taeda (Eckert et al. 2010); and 2) for Picea sp., based on published maps for Picea abies (Lind et al. 2014), Picea glauca and Picea mariana (Pavy et al. 2012). The composite maps for Pinus sp. and Picea sp. were then merged into a unified composite map for Pinaceae. This composite map for Pinaceae was then compared with a published gene-based map for Cryptomeria japonica (Moriguchi et al. 2012), a representative member of Cupressaceae. The strategy used to combine and compare the genetic maps is illustrated in figure 1a and supplementary figure S1, Supplementary Material online.

Table 1

Characteristics of the Genetic Linkage Maps Used in This Study to 1) Establish Composite Maps for P. pinaster (from 14 Individual Maps), and for the Pinaceae Family (combining Pinus pinaster, Pinus taeda, Picea glauca, Picea mariana, and Picea abies linkage maps), and 2) Compare the Pinaceae Composite Map with the Map of One Representative (Cryptomeria japonica) of the Cupressaceae Family

Species	Pedigree Name	Linkage Map ID	Number of Individuals	Number of Markers	Length (cM)	Mean Marker Interval (cM)	Reference
Pinus pinaster	C×L	C	106	574	1,488	2.8	Lagraulet (2015)
		L		826	1,863	2.3
	M×L	M	117	627	1,658	2.8
		L		920	1,953	2.2
	C×M	C	94	728	1,886	2.6
		M		630	1,619	2.6
	F2	F2_O	69	1,481	1,688	0.98	Plomion et al. (2015)
		F2_N	92	2,052	1,993	1.17
	G2	G2M	83	619	1,425	2.3	Chancerel et al. (2013)
		G2F		562	1,445	2.57
	C14×C15	C14	63	812	1,714	2.1	extended from de Miguel et al. (2012)
		C15		923	1,577	1.71
	Gal1056×Oria6	Gal1056	69	666	1,426	2.14	modified from de Miguel et al. (2014)
		Oria6		755	1,296	1.72
Pinus taeda	qtl	Ptaeda	172	1,815	1,899	1.1	Eckert et al. (2010)
Picea glauca	D, P	Pglauca	500, 260	2,270	2,083	1.1	Pavy et al. (2012)
Picea mariana	9920002		283
Picea abies	S21K7622162×S21K7621678	Pabies	247	686	1,889	2.8	Lind et al. (2014)
Cryptomeria japonica	YI	Cjaponica	150	1,262	1,405	1.1	Moriguchi et al. (2012)

F

Flowchart of the comparative analysis between Pinaceae and C. japonica. (a) Bioinformatic analysis developed for homologous genes identification. (b) Results of the comparative analysis between Pinaceae and C. japonica testing different confidence thresholds applied at two different steps: Sequence comparison for homolog unigne identification and comparative gene position for orthologs identification. Pathways that allowed the identification of orthologs covering the full set of chromosomes from C. japonica and Pinaceae are marked in bold.

The P. pinaster composite map comprised 3,491 unigenes of the Pine V3 Unigene set (Canales et al. 2014) as well as 182 AFLP (amplified fragment length polymorphism) or SAMPL (selective amplification of microsatellite polymorphic locus) markers (table 2 and supplementary fig. S2, Supplementary Material online). There were 3,639 unique markers, 60% of which were present on at least two component maps. Overall, we found high degrees of synteny and collinearity between all the P. pinaster component maps and the composite map (supplementary fig. S3, Supplementary Material online). The mean proportion of markers noncollinear (inversion greater than 5 cM) between the composite map and a component map was 1.3%. The large number of markers common to different component linkage maps and the high levels of collinearity observed increased the degree of certainty concerning the relative positions of the mapped unigenes. The P. pinaster composite map contributed the largest number of mapped markers for construction of the composite map for Pinaceae (tables 1 and 2).

Table 2

Details of the Composite Genetic Linkage Maps Generated in This Study

	Pinus pinaster	Pinus	Picea	Pinaceae
Nb of LGs	12	12	12	12
Size (cM)	1,721.7	1,943	2,032.9	2,094.9
Nb of markers	3,673	5,195	2,325	6,912
Nb of markers corresponding to PineV3a unigenes	3,491	4,639	1,940	5,971
Nb of unique unigenes	3,457	4,605	1,931	5,927
Nb of unique positions	1,819	2,336	2,006	3,077
Mean marker interval (cM)	0.47	0.39	0.88	0.30
Mean unique position interval (cM)	0.96	0.93	1.02	0.68

aFrom Canales et al. (2014).

The composite map for Pinaceae contained 6,912 mapped markers over 2,094.9 cM (table 2 and supplementary fig. S4, Supplementary Material online). As single nucleotide polymorphisms (SNPs) mapped in this composite map were identified from a variety of transcritptomic assays in diverse species, we considered as different gene loci only those that had an homolog in Pine V3, the gene catalog used as reference. Following this criterion, at least 5,927 different unigenes were mapped in the Pinaceae composite map (table 2 and supplementary table S1, Supplementary Material online). On average, 42 unigenes per linkage group (LG) were common to the Pinus sp. and Picea sp. maps, identifying a total of 513 othologous unigenes between both species (fig. 2 and supplementary table S1, Supplementary Material online). Only 5.9% of unigenes were nonsyntenic and 2.8% presented an inversion of more than 15 cM (supplementary table S1, Supplementary Material online). These markers were identified and removed from the Pinaceae composite map. As expected, there was a high degree of synteny and collinearity between members of the Pinaceae, providing support for the strategy followed in this study. A small fraction of mapped unigenes in the Pinaceae composite map may be originated by paralogy as revealed by the 44 unigenes mapped in more than one LG (supplementary table S2, Supplementary Material online). Finally, the P. pinaster, Pinus, and Pinaceae composite maps generated in this study are the densest linkage maps ever produced for conifers at species, genus, and family levels, respectively.

F

Comparison between the composite linkage maps for Pinus sp. and Picea sp. The Pinus sp. composite map is shown in blue and the Picea sp. composite map is shown in green. Orthologous markers are linked by black lines connecting the two maps. The number of orthologous markers is indicated at the top of each LG.

Sequence comparisons between unigenes mapped on the Pinaceae and C. japonica genetic maps resulted in the identification of 257 and 229 homologous loci depending on the e-value cutoff applied (fig. 1b). Homologous unigenes were identified for all LGs whatever the threshold considered, from 17 on LG4 to 28 on LG3 for the Pinaceae map and from 13 on LG8 to 35 on LG3 for the C. japonica map (for an e-value cutoff of 1e−30). Linkage maps were aligned using homologous unigenes as anchor points. The alignment of both genetic maps enabled the identification of common genomic regions. Dotplot representation for map alignment (supplementary fig. S5, Supplementary Material online) showed a threshold of four shared unigenes between both maps suitable for othologous LG block determination. A more stringent criterion for ortholog selection was additionally tested, which consisted in a minimum of six shared unigenes between two regions to be orthologs. A total of 12–20 othologous LG blocks were identified depending on the threshold used (fig. 1b). However, orthologous LG blocks covered the complete set of chromosomes of the analyzed species only when a threshold of four shared markers was used (fig. 1b). Consequently, this threshold was considered the most appropriate in view of the level of resolution of available genetic maps, and further discussion of the results is based on this threshold. The use of a threshold of four homolgous unigenes to consider an orthologous LG block resulted in the identification of 143 orthologous loci (i.e., 55.6% of the homologous markers) for an e-value cutoff lower than 1e−30 and 124 (i.e., 54.2% of the homologous markers) for an e-value cutoff of 1e−35. Thus, the use of a more stringent selection criterion for the identification of homologous sequences did not decrease significantly the proportion of identified orthologous unigenes.

As a result, we found that each of the LGs on the Pinaceae map corresponded to one or two different LG blocks on the Cupressaceae map, and that each LG on the Cupressaceae map corresponded to one to three LG blocks on the Pinaceae map (fig. 3a and supplementary fig. S5, Supplementary Material online). Each pair of orthologous LG blocks determined a contiguous ancestral region (CAR). Most CARs were identified whatever the e-value threshold applied with the exception of CARs #14 and #19 (table 3) that could not be confirmed using the most stringent criterion. Mean number of unigenes per CAR was 6 and 7 depending of the e-value cutoff applied. The number of orthologous unigenes per CAR was slightly reduced in eight CARs for the most stringent e-value cutoff, but the size of CARs was maintained with the exception of two CARs that were reduced by 20.1 and 48.1 cM, respectively (table 3). Thus, the number and size of identified CARs were consistent under the two different thresholds tested for homolog identification. Therefore, our results suggest the existence of 18–20 CARs that may have shaped the 12 chromosomes of modern Pinaceae species and the 11 chromosomes of modern Cupressaceae species through a different number of fusions (fig. 3b). Taking the Pinaceae map as the reference and inspecting the 20 proposed CARs (e-value threshold of 1e−30), seven C. japonica LGs displayed crossed CARs. Taking the noncrossing CARS as a measurement of collinearity, 40% of the CARs identified were considered to be collinear. Besides, high levels of collinearity were found within CARs, with only 18.1% of orthologous markers presenting an inversion of more than 15 cM within a CAR (supplementary fig. S5, Supplementary Material online). These results suggest an intense shuffling of orthologous LG blocks during the evolution of Pinaceae and Cupressaceae, but a higher conservation of gene order within these blocks.

F

Pinaceae–Cupressaceae comparative mapping. Results for an e-value cutoff of e−30 for homolog unigene identification and a threshold of at least four homologs shared between the two maps to determine an orthologous LG block. (a) Positions of the 143 orthologous unigenes mapped for representative species of both Pinaceae and Cupressaceae. Orthologous LG blocks are indicated by color-coded ribbons connecting the Pinaceae (in gray) and Cupressaceae (in white) LGs. LG number and genetic distance in centimorgans are indicated outside the circle. Pinaceae LGs are ordered from 1 to 12 and C. japonica LGs are ordered to facilitate graphical visualization. (b) Representation of the more parsimonious model of evolution of the identified orthologous LG blocks mapped on Pinaceae and C. japonica. Each orthologous LG block determined a CAR. Identified CARs are numbered from 1 to 20.

Table 3

Number of Unigenes and Size (cM) of Identified Orthologous LG blocks (CARs) at Two e-Value Cutoffs for Homologous Unigenes Identification

	e−30		e−35
CAR	Nb Unigenes	cM (Pinaceae)	Nb Unigenes	cM (Pinaceae)
1	9	10.9–155.8	9	10.9–155.8
2	10	11.9–84.9	9	11.9–84.9
3	12	0–146.5	11	0–146.5
4	7	65.3–147.5	7	65.3–147.5
5	5	27.6–69	4	27.6–69
6	5	13.5–123	5	13.5–123
7	6	12.2–70.8	6	12.2–70.8
8	7	37.4–157.9	7	37.4–157.9
9	8	42.9–83.7	7	42.9–83.7
10	6	77.2–146.4	4	125.3–146.4
11	4	27.7–64.7	4	27.7–64.7
12	4	116.6–161.8	4	116.6–161.8
13	6	30.9–69.9	5	30.9–69.9
14	5	25.9–74	0	—
15	8	59.8–132	8	59.8–132
16	10	5–163.7	9	5–163.7
17	6	6.7–41.2	6	6.7–41.2
18	9	41.9–124.7	8	62–124.7
19	4	39.5–62.6	0	—
20	12	14.9–163.3	11	14.9–163.3

Note.—Changes in the number of unigenes or size of CARs following the most stringent threshold are indicated in bold.

Previous comparative genomics studies in Pinaceae have reported high levels of synteny and collinearity for genes (Chagné et al. 2003; Krutovsky et al. 2006; Pelgas et al. 2006; Pavy et al. 2012). The conservative genome macrostructure among Pinaceae species has been interpreted as evidence that genome rearrangement events are rare in conifers (Diaz-Sala et al. 2013; Nystedt et al. 2013). The results presented here revise this view of conifer genome evolution, which was inferred essentially from comparisons of Pinaceae species. Our findings also support a new hypothesis that substantial chromosome rearrangements have occurred between families. Molecular phylogenetic studies support the splitting of conifers into two groups: Pinaceae and Coniferales II, corresponding to all conifer families other than Pinaceae (Bowe et al. 2000; Gugerli et al. 2001; Lu et al. 2014). The observed chromosomal rearrangements may have generated a reproductive barrier separating the two lineages around 290 Ma (Burleigh et al. 2012). On the other hand, Pinus and Picea display remarkable levels of synteny and collinearity despite their ancient divergence, confirming the exceptionally high degree of genome structure conservation within Pinaceae. According to Gernandt et al. (2011), conifers (Coniferales) can be grouped into six different families: Pinaceae, Podocarpaceae, Araucariaceae, Sciadopityaceae, Taxaceae, and Cupressaceae. Comparative genomics studies with representatives of other conifer families are crucial to determine whether the lack of genome rearrangement observed in Pinaceae is a feature common to other conifers. The adaptive radiation of some Cupressaceae species dates from the Oligocene (23–33 Ma), but the first fossil record of C. japonica dates from 55 to 65 Ma (Yang et al. 2012). The study of genome structure in other species from Cupressaceae with a shorter life history could provide new insight into the mechanisms and patterns of genome evolution in conifers.

The n = 12 karyotype is considered the most primitive in the Pinaceae family, based on chromosome morphometrics (Nkongolo et al. 2012). However, we were unable to reconstruct the karyotype of the common ancestor of Pinaceae and Cupressaceae in this study due to the lack of a suitable outgroup species phylogenetically close to conifers and with a well-assembled genome. The candidate species best matching these criteria is the basal angiosperm Amborella trichopoda (Amborella Genome Project 2013). A comparison between basal angiosperms and conifers should open up promising perspectives for the construction of a model of karyotype evolution. Comparative genomics and phylogenetic studies based on genome-wide comparisons with conifers will be crucial to bridge the gaps in our understanding of the evolution of plant genomes from cryptogams to flowering plants.

Conclusion

The results reported here take us a step beyond the “stasis” already described for the Pinaceae, opening up new avenues of research into the evolution of conifer genomes. We propose a possible scenario for conifer genome evolution, based on the fusion of chromosomal blocks, serving as a prelude to the modern karyotype configuration in Pinaceae and C. japonica. However, further improvements in our knowledge of basal angiosperms and gymnosperms will be required to reconstruct the karyotype of the common ancestor of seed plants.

Materials and Methods

Description of the Genetic Linkage Maps Used in This Study

The following terms were used to describe the different kinds of genetic maps used in this study, as suggested by Hudson et al. (2012): 1) Sex-averaged map: a consensus map for both parents of a pedigree; 2) consensus map: an integrated map based on segregation data from individual component maps; 3) composite map: an integrated map of different individual component maps built by a marker-merging method; and 4) component map: each of the maps used in the construction of a composite map. The graphics and the representations of genetic maps were produced with R 3.1.0 (R Core Team 2014).

Pinus pinaster

We used 14 base maps generated from seven different crosses to generate a composite genetic linkage map for P. pinaster. The first six maps were obtained from three controlled crosses (pedigrees #1, #2, and #3 in supplementary fig. S1, Supplementary Material online) between three different genotypes: Corsica × Landes (C×L), Morocco × Landes (M×L), and Corsica × Morocco (C×M). In total, 106, 117, and 94 full-sibs were genotyped with the 9k SNP-array described by Plomion et al. (2015) for C×L, M×L, and C×M, respectively. The regression mapping algorithm of JoinMap 4.1 (Van Ooijen 2011) was used to produce two maps for each parental genotype (one per cross), according to a two-way pseudotestcross mapping strategy (Grattapaglia and Sederoff 1994), using testcross markers (i.e., segregating in a 1:1 Mendelian ratio) only. The genetic maps were then combined into sex-averaged maps (Corsica, Landes and Morocco, see supplementary fig. S1, Supplementary Material online) with the function “combine groups for map integration” of JoinMap 4.1. More details on the construction of the maps can be found in Lagraulet (2015).

Four other maps from two different mapping populations were also used. The first population was a three-generation inbred pedigree consisting of an F2 population (#4 in supplementary fig. S1, Supplementary Material online) resulting from the selfing of an interprovenance tree (Landes × Corsica). The second population was a three-generation outbred pedigree (G2, #5) resulting from a controlled cross of two intraprovenance hybrid trees (Landes × Landes). The construction of these maps was described by Plomion et al. (2015) for the F2 population and Chancerel et al. (2013) for the G2 population. For the F2 population, two different sets of individuals were used to generate two maps (F2_O and F2_N) with the RECORD algorithm (Van Os et al. 2005). For the G2 population, one map for each parent (G2M and G2F) was produced with the regression mapping algorithm of JoinMap 4.1 (Van Ooijen, 2011). The F2_O, G2M, and G2F maps included different marker types: AFLP, single sequence repeat (SSR), expressed sequence tag (EST), and SNPs from different arrays (Chancerel et al. 2011, 2013), whereas the F2_N map contained only SNPs from the 9k SNP-array (Plomion et al. 2015). We made use of the large number of common markers and the high level of collinearity between the two F2 maps to construct a composite map (referred to as F2C by Plomion et al. 2015).

The last four maps were generated from two different F1 crosses: C14×C15 (#6 in supplementary fig. S1, Supplementary Material online) and Gal1056×Oria6 (#7). From the initial parental maps of the C14×C15 mapping population described by de Miguel et al. (2012), we mapped an additional set of 980 SNPs from the 12k SNP-array described by Chancerel et al. (2011) and 273 SNPs from a 1,536 SNP-array (Sáez-Laguna et al. unpublished) here to increase the number of anchor markers common to other maps. The parental maps of the Gal1056×Oria6 population used by de Miguel et al. (2014) were reconstructed in this study with the most informative individuals. For both pedigrees, we used the maximum-likelihood mapping algorithm of JoinMap 4.1 to generate individual genetic maps and sex-averaged maps. The four maps from the C14×C15 and Gal1056×Oria6 crosses contained different types of molecular markers (SSRs, ESTs, SAMPLs, and SNPs).

For all maps, genetic distances in centimorgans (cM) were calculated with the Kosambi mapping function (Kosambi 1943).

Other Pinaceae

We carried out a literature review to identify previously published high-density gene-based linkage maps for members of the Pinaceae family, for which sequence information was publicly available. Only four studies satisfied these criteria. Eckert et al. (2010) provided a sex-averaged linkage map for a two-generation outbred pedigree based on SNPs for P. taeda (accession number: TG091, http://dendrome.ucdavis.edu/cmap/, last accessed May 21, 2015). The map provided by Pavy et al. (2012) is the densest genetic map published to date for Picea. This map was a consensus of the white spruce (Pic. glauca) and black spruce (Pic. mariana) linkage maps. The white spruce pedigree was an F1 full-sib family, whereas the black spruce pedigree was a backcross representing the hybridization species complex Pic. mariana × Pic. rubens. Pic. glauca and Pic. mariana linkage maps were constructed with the regression algorithm in JoinMap 3.0 software. Lind et al. (2014) provided the most saturated and gene-rich map to date for Pic. abies. This map was a sex-averaged map of the two parents of an F1 controlled cross and was also constructed with the regression algorithm of JoinMap 3.0. A detailed list of mapping features for each component map included in this study is available in table 1.

Cryptomeria japonica

A high-density linkage map for C. japonica was incorporated into this study, as a representative species from the Cupressaceae family (Moriguchi et al. 2012). This map was constructed from an F1 full-sib family (table 1), with the regression mapping algorithm implemented in JoinMap v 3.0.

Identification of Homologous Genes within Pinaceae

The 17 maps described above were mostly constructed with SNP markers (100% of the markers for Picea sp., 98% for P. pinaster, and 90% for P. taeda). The flanking sequences of each SNP marker were compared with the most recent Unigene sets available for each species to obtain the sequence of unigenes containing the mapped SNPs: Canales et al. (2014) for P. pinaster, Rigault et al. (2011) for Pic. glauca, Pic. mariana and Pic. abies, and Lorenz et al. (2012) for P. taeda. This comparison was carried out with the BLASTn tool (the BLAST 1 step in fig. 1a). Unigenes with a percentage identity exceeding 95% with mapped SNP flanking sequences were retained for the next step.

Pinus pinaster Unigene set from Canales et al. (2014), Pine V3, was then used as the reference for the identification of homologous unigenes within Pinaceae species. A second sequence comparison (R-BLAST 2 in fig. 1a) was performed, between the mapped unigenes of each species and the unigene sequences of Pine V3. For this interspecific comparison, a stringent reciprocal tBLASTx analysis was performed. Only sequences with a reciprocal best hit with a percentage identity exceeding 85%, an e value below e−65, and an alignment of more than 200 bp were retained as homologous unigenes. Homologous unigenes between different species were considered as orthologs if they were positioned in the same LG (i.e., syntenic unigenes). Identified orthologous unigenes were used as anchor markers to construct a composite linkage map for each genus (Pinus and Picea), as a preliminary step in the construction of a composite map for the Pinaceae family including both genera.

Construction of Composite Genetic Linkage Maps

We used the R package LPmerge (Endelman and Plomion 2014) to integrate component linkage maps into a composite map without the use of segregation data. LPmerge assessed the goodness of fit of the composite map by calculating a root mean squared error (RMSE) per LG, by comparing the position (in cM) of all markers on the composite map with that on the component maps. We calculated this metric for different maximum interval sizes (parameter K in the algorithm), ranging from 1 to 10. The value of K minimizing the mean RMSE per LG was selected for construction of the composite map. This method was used for the construction of all the composite species maps reported here. Further details about the production of each composite map are described below.

Before integrating the 14 base genetic linkage maps into a single composite map, we established consensus maps (supplementary fig. S1, Supplementary Material online) based on markers common to different accessions across pedigrees (Corsica, Landes, Morocco genotypes for pedigrees #1, #2, and #3, respectively), or accessions within pedigrees (Coca and GxO for pedigrees #6 and #7, respectively), or based on the merging of different data sets of the same pedigree (F2 for pedigree #4). This process, designed to increase the number of markers common to component maps, was facilitated by the use of the same 12k (Chancerel et al. 2013) and 9k (Plomion et al. 2015) SNP-arrays for some pedigrees.

The SNP marker ID of each component map was replaced by the corresponding maritime pine unigene ID from Canales et al. (2014). This step, which was essential for the use of LPmerge (i.e., same marker name for orthologous markers), also made it possible to check the collinearity between maps. Thus, nonsyntenic unigenes between different P. pinaster linkage maps were removed from the analysis with the exception of those validated for at least two other component maps (supplementary table S2, Supplementary Material online). Finally, LPmerge was used to create the composite map for P. pinaster. Given that similar numbers of genotypes were used to obtain the base maps and the high degree of synteny between base maps, each component map was assigned the same weight in LPmerge.

Pinaceae

The SNP marker ID of each species component map was replaced by the corresponding homologous unigene ID of P. pinaster (Canales et al. 2014). We established composite maps at genus level before integrating the four genetic maps for each species into a single composite map for Pinaceae. This process was designed to increase the number of markers common to the component maps for each genus (Pinus and Picea). LPmerge was used to build these two composite maps, following the same procedure as described for P. pinaster. We discarded nonsyntenic unigenes, except for nonsyntenic unigenes validated by at least two component maps in the P. pinaster composite map, from the construction of composite linkage maps. Noncollinear unigenes with inversions of more than 15 cM were also excluded from the construction of the composite linkage map. A large inversion of a group of markers was detected in LG7 of Pic. abies (Lind et al. 2014), when the map for this species was compared with that for Pic. glauca (Pavy et al. 2008). Pic. abies LG7 (renamed LG2 after comparison with the P. pinaster reference map) was reconstructed from genotyping data provided as supplementary material by Lind et al. (2014), using the same parameters described in Lind’s article and the same mapping software (JoinMap v4.1). Two markers with a −Log10 (p) > 1 that produced a large number of double recombinants were excluded from this LG map. We were thus able to map 16 additional markers, and a much higher degree of synteny and collinearity was found between the homologous LGs of Pic. abies and Pic. glauca (supplementary fig. S3, Supplementary Material online).

Comparison with C. japonica

The available map for C. japonica consisted of 77% of SNP markers (Moriguchi et al. 2012). Sequences of unigenes containing the mapped SNPs were retrieved from the Unigene set developed by Ueno et al. (2012). Then, unigene sequences from Ueno et al. (2012) mapped in C. japonica linkage map (Moriguchi et al. 2012) were compared with those of P. pinaster (Canales et al. 2014) mapped in the Pinaceae composite map using tBLASTx (BLAST 3, fig. 1a). Different e-value cutoff for homologous unigenes identification between Pinaceae and C. japonica was tested: Lower than 1e−30 and 1e−35.

Selected homologs from the Pinaceae and C. japonica linkage maps were used for comparative mapping. We established orthologous blocks within LGs where several homologous unigenes were shared between both families. Different thresholds were also tested to consider an orthologous block within an LG: Blocks with at least four and six shared unigenes. Each pair of orthologous chromosomal blocks determined a CAR between the two families. The most parsimonious evolutionary model between Pinaceae and Cupressaceae considering the existence of the identified CARs was proposed. Circular genetic maps used in interfamily comparative mapping were drawn with Circos software (Krzywinski et al. 2009).

Data Accessibility

All the linkage maps described here are available from the Pinus portal (a European genetic and genomic resource for Pinus) through the PinusMap application (https://w3.pierroton.inra.fr/PinusPortal/index.php). Accession numbers for marker sequences used in this study are available in supplementary table S3, Supplementary Material online.

Supplementary Material

Supplementary figures S1–S5 and tables S1–S3 are available at Genome Biology and Evolution online (http://www.gbe.oxfordjournals.org/).