Registered cultivars and pedigree data

Information on the flax cultivars registered in Canada was collected from six sources: (1) the database of cultivars registered in Canada (CFIA 2015a ); (2) scientific journals — Canadian cultivars released before 1961 were usually published in the Journal of Agronomy and then in Crop Science, and since 1980, all registered flax cultivars have been published in the Canadian Journal of Plant Science; (3) the Germplasm Resources Information Network (GRIN) database ( http://www.ars-grin.gov/npgs/index.html), a part of the National Genetic Resources Program (NGRP) of the United States Department of Agriculture, Agricultural Research Service (USDA-ARS), as well as the Canadian GRIN database (GRIN-CA) of Agriculture and Agri-Food Canada’s Plant Gene Resources of Canada ( http://pgrc3.agr.gc.ca/index_e.html); (5) the database and website of the Flax Council of Canada ( http://flaxcouncil.ca/); and (6) personal communications with flax breeders. For all flax cultivars registered in Canada, we traced their parentage to the oldest known ancestors or any most remote variety or breeding line for which parentage information was available. An ancestor was defined as the ultimate variety, landrace, or breeding line from which a cultivar was derived and where further parentage was unknown or untraceable from available pedigree records. Some direct or indirect parents were also considered ancestors when complete pedigree records were lacking.

Genetic contribution of ancestors to genetic base

The GC of an ancestor to a flax cultivar was determined as the genetic proportion in the flax cultivar that could be traced to an ancestor via pedigree analysis (Gizlice et al. 1994). CP was used to estimate this contribution. The CP between two genotypes was defined as the probability that a random allele at a random locus in one genotype is identical by descent to a random allele at the same locus in another genotype (Kempthorne 1957; Malécot 1969). In flax, we assume all ancestors, cultivars, and parental breeding lines are homozygous. All ancestors are assumed to be independent, i.e., the CP between any pair of ancestors is ‘0’. A cultivar or a breeding line derived from crossing of two parents obtains half of its genome from each parent. Several breeding approaches, such as direct selection of natural variants from germplasm, artificial mutants, tissue culture-derived somaclonal variants, and transgenic technique generated cultivars, feature almost the same genetic background except for minor genetic changes in some gene loci for specific traits. To simplify the CP calculation, we assumed that the cultivars developed using these approaches obtained all genetic material from their progenitor, i.e., the CP between the progenitor and the essentially derived cultivars is equal to 1. In theory, the GC of different ancestors to a derived cultivar is defined as the relative genetic constitution of the ancestors in the derived cultivar but, in practice, it is estimated using the CPs of the ancestors and the derived cultivars. The sum of CPs of all ancestors of a cultivar must be equal to 1, i.e., where n is the total number of ancestors of a derived cultivar. The sum of the mean CPs of ancestors across all or selected cultivars must also be equal to 1, i.e., where n is the number of ancestors of all or selected cultivars and m is the number of all or selected cultivars. These two conditions can be used to validate CP calculations. The algorithm to calculate CPs is described in detail elsewhere (Cui et al. 2000). All pedigree analyses were programmed using a custom Perl script, which is available upon request.

Cluster analysis of registered cultivars

Based on the distance (1-CP) among flax cultivars, a hierarchical cluster analysis was performed using the hclust function with the Ward method in R ( http://www.cran.r-project.org).

Pedigree relations of flax cultivars

Pedigree information was sought for all registered flax cultivars; however, records for the recent CPS/Viterra cultivars WestLin 70 and WestLin 72 and for several cultivars registered before the 1950s could not be obtained and are therefore not presented. The pedigree records for several direct or indirect parents were also not available, so these parents were considered ancestors. Based on available pedigree records, individual pedigree trees for all registered cultivars were drawn and merged to build a chart with an overall Canadian flax cultivar pedigree tree (Fig. 1). Of the 82 cultivars, 72 group into a single tree, illustrating the extent of the common genetic base among Canadian flax cultivars. The other cultivars group into several smaller pedigree trees or singletons as a consequence of the unavailability of their pedigree records or because they were not used as parents for breeding.

Fig. 1.

Pedigrees of flax cultivars registered in Canada from 1910 to 2015. The major ancestors of Canadian flax cultivars are underlined.

We calculated the CPs among all registered cultivars, as well as the mean CPs among cultivars released in different years and within/among the cultivars released by different breeding programs. The mean CPs among all 82 registered cultivars and among the 46 cultivars released from 1981 to 2015 are 0.14 and 0.23, respectively, indicating the narrowness of the Canadian flax cultivars, especially those recently developed. Among the 46 cultivars released from 1981 to 2015, only one out of 946 pairs of cultivars (VT 50 vs. Vimy) has no pedigree relation; all other pairs have CPs ranging from 0.01 to 1. On average, any two cultivars share 23% of their genetic base.

Comparison of the CPs between cultivars released by different breeding programs (Table 2) shows the cultivars developed by the CDC have the closest pedigree relation (mean CP of 0.30) vs. those from the AAFC and CPS/Viterra programs (0.24). Comparing cultivars across breeding programs also indicates fairly high percentages of genetic similarity (0.16–0.25) (Table 2).

Table 2.

Mean CPs of cultivars released from 1981 to 2015 by the three major breeding programs in Canada.

A cluster analysis was performed based on the distance (1-CP) among the cultivars. The dendrogram of registered cultivars and their major ancestors (Fig. 2) clearly depicts the pedigree relation of the flax cultivars and their parents or ancestors. Cultivars merging at a 1-CP value of 0 are indicative of cultivars obtained via direct selection (e.g., Royal from Crown), tissue culture (Andro and CDC Normandy from McGregor), irradiation (Redwood 65 from Redwood), or transgenic technique (e.g., CDC Triffid from Norlin). CDC Triffid was deregistered in 2001 by the Canadian Food Inspection Agency (CFIA) due to concerns by the European market about the production of transgenic flax and the use of transgenic technology as such by European consumers. In the cluster analysis, cultivars with no pedigree relation were excluded. The remaining 64 cultivars and their major ancestors grouped into three major clusters (Fig. 2). Cluster I represents the 14 major ancestors and some early registered cultivars. Because most cultivars in this cluster are ancestors that had no pedigree relation, the mean CP among the cultivars in Cluster I is relatively small (0.04); the mean CPs of cultivars between Cluster I and the other two clusters represent the mean genetic contribution of each ancestor to the derived cultivars in Clusters II and III (0.06–0.08) (Table 3). Clusters II and III contain most of the modern flax cultivars derived from some Cluster I ancestors. The cultivars within Clusters II and III share 24% and 34% of their genetic base, respectively (Table 3). Clusters II and III cultivars share 15% of their genetic base. Three ancestors, Argentine 8C, Redwing, and Kubanksij, cluster together into a sub-cluster of Cluster II (Fig. 2), suggesting these three cultivars contribute more to the cultivars in Cluster II than Cluster III.

Fig. 2.

Dendrogram of flax cultivars released in Canada and their major ancestors (boxed) based on the 1-CP distance among cultivars. Fifteen cultivars (Cascade, Crown, Diadem, Dutch White Blossom, Novelty, Ottawa 829C, Royal, Scorpion, Stormont Motley, Victory, Viking, Weira, WestLin8 70, WestLin 72, and Winona) were not included because they have no pedigree relation with other cultivars or no pedigree records.

Table 3.

Mean CPs of cultivars across the three clusters generated by the cluster analysis in Fig. 2.

Ancestors of flax cultivars

Of the 82 registered cultivars, eight (Diadem, Novelty, Ottawa 829C, Winona, Dutch White Blossom, Stormont Motley, Scorpion, and Cascade) have no derived cultivars and two (WestLin 70 and WestLin 72) have no pedigree records, and were thus excluded from ancestor analysis. The 72 remaining registered cultivars trace back to 46 ancestors (Table 4), of which 18 are breeding lines without pedigree records. These ancestors originate from Canada (28%), the USA (15%), Europe (20%), Australia (4%), and unknown geographical origins (33%), corresponding to 17%, 40%, 15%, 3%, and 25% of the GC to the flax cultivars, respectively (Table 5). As the 15 ancestors of unknown geographical origins are non-Canadian lines, they were considered foreign germplasm, which therefore accounts for a total of 72% of the 46 ancestors and explains 83% of the genetic base of the Canadian flax cultivars.

Table 4.

Ancestors of registered flax cultivars, their number of derived cultivars, and their respective genetic contribution.

Table 5.

Numbers and mean GC percentages (in parentheses) of ancestors of flax cultivars registered from 1910 to 2015 classified by types and geographic origins.

The cultivars released in the periods 1910–1950, 1951–1980, and 1981–2000 have 16, 19, and 18 ancestors, respectively. In the past two decades, new germplasm has been introduced into Canadian flax breeding programs, such as Pacific, Windermere, and other improved breeding lines. This is reflected in the increased number of ancestors for cultivars released from 2001 to 2015, which is almost twice (36) that of the previous period. Likewise, the number of ancestors per cultivar has gradually increased, especially over the last 35 years (Fig. 3). The mean number of ancestors per cultivar released from 1910–1950, 1951–1980, 1981–2000, and 2001–2015 is 2.3, 4.1, 10.1, and 13.6, respectively. Cultivars, such as Prairie Sapphire released in 2009, have up to 20 ancestors in their genetic makeup (Table 1).

Fig. 3.

Relationship between the number of ancestors of individual flax cultivars and their year of registration.

Of the original 46 ancestors, the top 11 in Table 4 contribute 28.6%, 81.0%, 92.5%, and 69.6% of the genetic base to the cultivars developed during 1910–1950, 1951–1980, 1981–2000, and 2000–2015, respectively (Table 4). These ancestors contribute the majority of the genetic base to cultivars released between 1981 and 2015 and form the core gene pool of Canadian flax cultivars. Redwing, Bison, and Argentine 8C alone contribute 34%–46% of the genetic base to cultivars registered since 1951 and more than 40 flax cultivars are derived from each of them (Table 4).

Germplasm preservation and evaluation to assist expansion of the genetic base

Increasing reliance on relatively few cultivars as parents in most breeding programs has led to the loss of genetic diversity and narrowing of the genetic base, which may reduce tolerance to the effects of climate change, new pests, and diseases. The higher number of ancestors of recent cultivars is an indication that breeders are paying attention to the consequences of the narrow genetic base. Moreover, focused breeding efforts for brown as well as yellow linseed development, the demand from industry for higher oil and ALA content flaxseed and from flax producers for improved agronomic traits associated with yield production in the Northern Prairies have resulted in the necessity to introduce novel germplasm into breeding programs for crop improvement. It is imperative that flax breeders continue to consider suitable breeding strategies for expanding the genetic base and increasing the genetic diversity of future flax cultivars.

Of the 82 registered cultivars, 25 were directly introduced for production from foreign countries (e.g., USA, Germany, and Australia) and 72% of the ancestors originated from foreign germplasm, indicating the potential adaptability and importance of foreign germplasm as a source of genetic diversity and elite alleles. Expansion of the genetic base in recent decades is primarily attributed to imported germplasm. Thus, the introduction, evaluation, and utilization of new exotic germplasm should continue to be an efficient approach to further broaden the genetic base of Canadian flax cultivars.

There are approximately 48 000 flax accessions in the ex situ world collections (Diederichsen 2007), whereas the Canadian national plant germplasm repository, Plant Gene Resources of Canada (PGRC), preserves a collection of 3252 accessions of Linum usitatissimum L. and 76 accessions of other species of the genus Linum, originating from 72 countries and representing all historical and present cultivation regions for linseed or fibre production (Diederichsen and Fu 2008). In the last two decades, most of these Canadian collections (over 90%) have been resampled to be characterized and evaluated for morphologic, phenological, and agronomic characters such as plant height, seed color, seed weight, seed oil content, fatty acid profile, seed coat mucilage content, fiber content, adaptation to dry growing conditions, disease resistance, seed vigour, seed lignin content, and so on (Diederichsen 2001; Diederichsen and Raney 2006; Diederichsen and Fu 2008; Diederichsen et al. 2008; Diederichsen and Ulrich 2009). Molecular markers of randomly amplified polymorphic DNA (RAPD) have also been used to evaluate the genetic variation and geographic patterns of more than 2700 flax accessions and cultivars (Fu et al. 2002, 2003; Fu 2005, 2006; Diederichsen and Fu 2006). Large-scale characterization and evaluation demonstrate a wide range of diversity of economically relevant traits within and among accessions is a valuable source of genetic variation for breeding (Diederichsen and Fu 2008). This variation exists largely within geographical regions because only a small portion (8.2%) of the RAPD variation in the flax accessions can be explained by the geographical regions (Fu 2005). Comparison with Canadian cultivars indicates that the wide variation in agro-botanical characters of flax germplasm is useful for flax breeding (Diederichsen 2001). The weak correlation between fibre content and other agronomic traits in the flax germplasm demonstrates that the variation in fibre content of flax germplasm is useful for identifying germplasm relevant for breeding dual purpose flax (Diederichsen and Ulrich 2009).

Due to the redundancy in the Canadian whole flax collection and breeding demands, a subset of 381 accessions plus 26 later appended Canadian cultivars, named the core collection (which includes most of the core ancestors identified in this study, their derived cultivars, and exotic germplasm) has been assembled to include most of the genetic variation present in the whole collection of 3378 accessions and current flax cultivars (Diederichsen et al. 2013). Molecular evaluation via 448 microsatellite markers has demonstrated that the core collection has abundant genetic diversity and a weak population structure, making it suitable for association mapping of agronomic and quality traits in flax (Soto-Cerda et al. 2013b ). Important agronomic traits such as seed yield, oil content and quality, fiber traits, and disease resistance have been phenotyped at multiple locations and years (unpublished data). Thus, this core collection provides a realistic and useful germplasm for expanding the genetic base of future flax cultivars.

Pale flax (Linum bienne Mill.) is the wild progenitor of cultivated flax (Diederichsen and Hammer 1995; Fu and Peterson 2010). Pale flax and cultivated flax share the same chromosome number (n=15) and their crosses generate fully fertile offspring, thus, it represents cultivated flax’s primary gene pool (Diederichsen 2007). Genetic variation studies of the two species indicate that strong genetic differentiation exists between pale and cultivated flax and that pale flax is more genetically diverse (Diederichsen 2007; Uysal et al. 2010, 2012; Soto-Cerda et al. 2014a ) and possesses favorable alleles with a potentially positive effect on improving seed yield through yield components (Soto-Cerda et al. 2014a ). Therefore, pale flax provides another potential source of genetic variation to broaden the narrow genetic base of Canadian flax cultivars.

The genetic base of Canadian flax cultivars

The genetic base of all Canadian flax cultivars was estimated at both the population (all flax cultivars) and at the individual level (individual cultivars) using the number of ancestors and the genetic contribution of ancestors to all flax cultivars or individual cultivars included in this study. We identified 46 ancestors from all registered flax cultivars, of which 36 are from cultivars released in the last 35 years. The top 11 ancestors contribute as much as 70%–93% of the genetic base of modern flax cultivars. Although the genetic base of individual cultivars has greatly improved in the last two decades, flax cultivars share an average of 8.8 common ancestors, resulting in a mean CP of 0.23. These results suggest that Canadian flax cultivars have similar backgrounds, a relatively narrow genetic base, and low genetic diversity. This corroborates the results of Soto-Cerda et al. (2013a) that show all 31 Canadian cultivars present in a germplasm collection of 407 core collection flax accessions cluster together on the basis of 414 neutral DNA markers. While the entire core collection has 2202 alleles, the Canadian cultivars only have 1081. Even adjusted for sample size difference, the Canadian cultivars only comprise 2/3 of the genetic variability of the core collection, substantiating our findings regarding their narrow genetic base.

Ancestors of Canadian flax cultivars to assist parent selection

Flax is an autogamous crop plant and its breeding involves creating genetic variability via crossing, selecting the best recombinants, and fixing the desirable traits or genes by inbreeding (Hall et al. 2016). Crossing followed by selection is the predominant approach in Canadian flax breeding. Most cultivars (86%) released in the last 35 years were developed via this approach. Parental selection and crossing schemes to broaden the genetic base and generate variability in the breeding program material are key factors to the success of conventional hybridization breeding methodology. We identified 46 ancestors and 11 major ancestors that constitute the core gene pool of Canadian flax cultivars. These ancestors possess elite and adaptive alleles for economically important traits in flax. The genetic structure of Canadian flax ancestors, if known, could reveal loci conditioning characters for improvement. For example, the majority of Canadian flax cultivars have the wilt resistant cultivar Bison as an ancestor. Understanding the major loci conditioning Bison’s resistance to fusarium wilt would allow flax breeders to identify early generation lines with resistance loci, thereby increasing the efficiency of selection. Optimally, the selection of genetically diverse parents with complementary characteristics for the development of segregating populations should be chosen from two different clusters based on their CPs, the number of ancestors, and their common ancestors (Fig. 2).

Genomics-assisted breeding to assist expansion of the genetic base

Plant breeding pyramids elite complementary alleles from distinct parents using different approaches, such as conventional, mutation, transgenic, and, more recently, gene editing (Sauer et al. 2016) breeding techniques for crop improvement. Molecular breeding incorporates DNA molecular markers associated with specific traits to improve selection efficiency. Flax has a small to medium size genome of ?370 MB (Wang et al. 2012) that has been sequenced, assembled, and sorted to generate a chromosome-based pseudomolecule physical map based on CDC Bethune (Wang et al. 2012; You et al. 2015). With the rapid advances and application of next generation sequencing and high-throughput genotyping technologies, it is possible to affordably sequence and genotype sufficient numbers of individuals and readily use molecular breeding strategies for cultivar development, such as marker-assisted selection (MAS) and genomic selection (GS) (You et al. 2016). MAS and GS are strategies that use DNA markers to predict phenotypes, thus increasing precision and efficiency and reducing the cost of offspring selection. Based on MAS, Peleman and van der Voort (2003) propose the “breeding by design” concept, which aims to bring together superior alleles for all genes of agronomic importance from potential germplasm resources to generate superior cultivars. This might be achieved through genome-wide high-resolution allele detection via precise quantitative trait loci (QTL) mapping of potential parental resources. In flax, some QTL associated with seed yield, oil content, ALA, and other important agronomic traits have been identified based on simple sequence repeat and single nucleotide polymorphism markers with bi-parental mapping populations (Cloutier et al. 2011; Kumar et al. 2015) and the Canadian flax core collection (Soto-Cerda et al. 2012, 2013c , 2014b ). Utilizing the outcomes from the Total Utilization Flax Genomics (TUFGEN) project, the high-resolution chromosome haplotype map of the core collection has been generated (unpublished data). This information with the phenotyped traits have been used and will be further applied for precise assessment of the genetic basis of economically important traits, detection of allelic variations at these loci, and identification of superior genotypes for breeding. Thus, genome-wide genetic dissection of elite germplasm and modern genomics-assisted breeding strategies provide the knowledge to widen the genetic base of future flax cultivars not only through selection of multiple diverse elite lines as parents but also through the identification of recombinants with favourable alleles for the development of innovative, improved, and better adapted flax cultivars. Recent innovations in crossing schemes such as multi-parent advanced generation intercross (MAGIC) (Huang et al. 2015) may generate parental inbreds with a much broader genetic base if the constituent parents are chosen carefully from the pedigree tree (Fig. 1). Recently, an eight-parent derived wheat MAGIC-population was developed (Mackay et al. 2014) for further genetic enhancement and similar strategies can be applied in flax breeding using the diverse ancestors identified from the pedigree chart.