New Interspeciﬁc Brassica Hybrids with High Levels of Heterosis for Fatty Acids Composition

: Winter oilseed rape ( Brassica napus L.) is the most important oil crop in Europe. Optimizing the profile and quantity of fatty acids in rapeseed is critical for maximizing the value of edible oil. Although the utilization of crop heterosis for hybrid breeding in rapeseed is limited by the relatively narrow genetic basis of adapted germplasm, an up-to-date significant effort has been made to broaden the rapeseed gene pool using different strategies. The present study was aimed to estimate heterosis for oil quality of the newly developed Brassica interspeciﬁc hybrids, using selected parental lines. For this purpose, ﬁve parental genotypes and twenty-two interspeciﬁc cross-derived Brassica lines were evaluated in a randomized complete block design with three replications in the Greater Poland region during 2009, 2010 and 2011. Generally, the variation among genotypes was evident for most of the tested fatty acids mean values, but the di ﬀ erences between genotypes were not always statistically signiﬁcant when based on individual fatty acids (FAs). However, the highest number of signiﬁcant heterosis e ﬀ ects was observed for behenic and lignoceric acids and for Brassica hybrid line H1. Based on obtained results it was possible to select one genotype—the hybrid line H5, which is recommended for further inclusion in the breeding programs.


Introduction
Heterosis (hybrid vigor) has been successfully utilized in order to increase the productivity in several crop species such as maize, sugar beet, sunflower, forage crops and grasses. Several authors reported that hybrid vigor is evident also in seed yield of F 1 hybrids for both winter and summer types of oilseed rape B. napus [1,2]. Although, in Brassica oleracea, heterosis is the most efficient tool providing stimulus to the hybrid vegetable industry, in oilseed rape, heterosis has not been extensively exploited, because of the lack of an effective pollination control system for commercial production of F 1 hybrid seeds [3,4].
Rapeseed is considered as an important source of oil for industrial use, as well as for edible purposes. The objective of modifying oil quality is to develop oils with enhanced nutritional and functional properties. The industrial utilization of rapeseed oil requires a specific fatty acid composition, e.g., a high oleic acid content [5,6]. This type of fatty acid composition is also of interest as a cooking oil. The range of genetic variability for oleic acid content was thus greatly extended, which created the possibility for selection. The change of the fatty acid composition in interspecific crosses between different Brassica species is commonly known [7].
Fatty acids profile plays a key role in the use of Brassica oil by humans. Vegetable oilseed because of higher proportion of 16 and 18 carbons unsaturated fatty acids, mostly monounsaturated fatty acids (MUFAs) are utilized principally as a source of edible oil [8][9][10]. For this purpose, for example, linolenic acid (C18:3) is undesirable, because of reducing the durability of the oil, although it is an essential dietary fatty acid. Moreover, erucic acid (C22:1) contains nearly 50% of total fatty acid, which is undesirable for human consumption, as it is reported to lead to myocardial lipidosis [11]. Although natural forms of rapeseed and mustard contain high levels of erucic acid (over 40% of total fatty acids), levels in rapeseed cultivated for food use are typically below 0.5%.
Heterosis is agronomically important in the use of F 1 hybrid cultivars in many crops and vegetables, especially due to the superior performance, which can appear as biomass, yield and abiotic and biotic stress tolerance.
For obtaining a new hybrid plant, as a first step, some information on the genetic background of important characters should be collected. Next, these details are used to combine desirable traits in a hybrid. Therefore, especially for plant breeders it is essential to have detailed information about the desirable parental combination in hybridization programs, which usually resulted in a high degree of heterotic response. Furthermore, although interspecific hybrid lines do not meet the elite rapeseed standards, they are a valuable source for hybrid breeding due to their large distance from present breeding material and their high heterosis when combined with European rapeseed cultivars. Most heterosis research has been focused on increasing various quality traits in Brassica and other crops [12]. According to Pal and Sikka [13] heterosis is a quick, cheap and easy method for increasing crop production. To improve quality and quantity traits in Brassica and other crops, heterosis is one of the most effective methods. Owing to heterosis, breeders can increase plant production in a short time by utilization of less input [13].
Therefore, in the present study, an effort was made to identify promising Brassica genotypes to be used as a donor parent for the production of low erucic acid canola lines.
In the present studies, heterosis (better-parent) was estimated for the fatty acid composition in F 1 -F 3 generations of Brassica interspecific hybrids, using the five parent cross experiment.

Plant Material
Plant material for the fatty acids composition analysis consisted of 27 Brassica genotypes including five parental genotypes, i.e., Brassica rapa ssp. pekinensis, Brassica rapa ssp. trilocularis cv. Yellow Sarson, B. carinata, B. juncea and the male sterile line of an F 8 generation of B. napus (MS8), as well as twenty two interspecific cross-derived Brassica lines. The MS8 line was selected from resynthesized oilseed rape (B. rapa ssp. chinensis × B. oleracea var. gemmifera) using in vitro cultures of isolated embryos [14] ( Table 1). F 1 -F 3 generations of tested lines, as well as parental genotypes, were selected from the rapeseed breeding program of the Department of Genetics and Plant Breeding, Poznań University of Life Sciences (PULS).
Brassica genotypes were tested in the 2009, 2010 and 2011 growing seasons. Field trials were performed in three replicates of a randomized block design at the PULS experimental station Dłoń (51 • 41 23" N, 17 • 04 10" E) located 100 km south of Poznań, (Greater Poland, Poland). The field experiment in Dłoń was conducted on typical heavy soil of III quality class of good rye complex [14]. The sums of precipitation during the vegetation season of Brassica genotypes in 2009, 2010 and 2011 were respectively: 605.1, 753.9 and 465.5 mm, but the mean temperatures during the vegetation season were respectively, 9.1, 7.9 and 9.8 • C. As shown in Table 2, these three seasons differed slightly in their climatic conditions. The field trials were arranged in a randomized complete block design with three replicates. Sowing dates were between the 24th and 28th of August. Each genotype was grown in a three-row plot of 9.0 m 2 , with a 0.30 row distance and a sawing density of 60 seeds/m 2 .

Seeds Sampling and Analysis of Fatty Acids Composition:
For each genotype, 10 single seeds, sampled from three to five plants, were analyzed and the average data were put to statistical analysis. Seeds were harvested at physiological maturity (BBCH 89). Fatty acid content was expressed as per cent of total fatty acid content.
A gas chromatography-flame ionization detection (GC-FID) method was used for direct quantitative analysis of seventeen fatty acids in seeds of analyzed Brassica genotypes in the Department of Biochemistry and Food Analysis of PULS. All the measurements were performed using a chromatograph Hewlett-Packard model 5890 series II with a flame ionization detector (FID), equipped with HP INNOWAX capilarry columns (30 m × 0.32 mm × 0.15 mm) and a split-splitless injector (Agilent Technologies, Inc., Cheshire, UK). Analyses of fatty acids were done at three replicates, and the arithmetic average was adopted for the results.

Statistical Analysis
Analysis for parental forms and hybrids were made independently. Firstly, the normality of the distributions of the studied traits were tested using Shapiro-Wilk's normality test [16]. A multivariate analysis of variance (MANOVA) was performed. The canonical variate analysis was applied in order to present a multi-trait assessment of similarity for the tested genotypes (parental forms and/or hybrids) in a lower number of dimensions with the least possible loss of information [17]. This makes it possible to illustrate variation in genotypes in terms of all the observed traits in the graphic form. The Mahalanobis distance was suggested as a measure of "polytrait" genotypes similarity [18], the significance of which was verified by means of critical value D α called "the least significant distance" [19]. Mahalanobis distances were calculated. Next, the heterosis effects of all observed traits were tested for each year of study. All the analyses were conducted using the GenStat 18 statistical software package (VSN International Limited, Hemel Hempstead, HP2 4TP, UK).

Results and Discussion
Nowadays, selecting high quality rapeseed, with good yield and increased oil content and improved edible oils with a modified fatty acid composition, seems to be one of the most important B. napus breeding goals [20]. Furthermore, palmitic, stearic, oleic, linoleic, linolenic, eicosenoic and erucic acids are the most important fatty acids in rapeseed, which determines the flavor and nutritional quality of B. napus [21]. However, reports in the literature on the performance of B. napus interspecific hybrids, with respect to the potential of heterosis for seed quality traits in rapeseed, are relatively rare [22]. In order to improve breeding efficiency, the selection of desirable genotypes is based mostly on a selected trait, i.e., the fatty acid composition regarded as nutritionally favorable [23]. Therefore, the multivariate approach to evaluating the fatty acid composition of Brassica hybrids seed oil can be useful in the assessment and selection of genotypes with the fatty acid composition advantageous from the human nutrition point of view.
Individual traits are of different importance and have a different share in the joint multivariate variation. A study on the multivariate variation for parental forms also includes an identification of the most important traits in the multivariate variation of parental forms. The analysis of canonical variables is a statistical tool making it possible to solve this problem [24][25][26]. Results of the analysis of canonical variables for investigated parental forms are presented in Figure 2. The first two canonical variables explained jointly 97.16% total variation between parental forms ( Figure 2).
The greatest variation of parental forms in terms of all the traits jointly (measured Mahalanobis distances) was found for MS8 and B. carinata (the Mahalanobis distance between them amounted to 143.8). The greatest similarity was found for B. juncea and B. pekinensis (30.4). Mahalanobis distances between MS8 and other parental forms were statistically significant (Table 4).    Individual traits are of different importance and have a different share in the joint multivariate variation. A study on the multivariate variation for parental forms also includes an identification of the most important traits in the multivariate variation of parental forms. The analysis of canonical variables is a statistical tool making it possible to solve this problem [24][25][26]. Results of the analysis of canonical variables for investigated parental forms are presented in Figure 2. The first two canonical variables explained jointly 97.16% total variation between parental forms (Figure 2). The greatest variation of parental forms in terms of all the traits jointly (measured Mahalanobis distances) was found for MS8 and B. carinata (the Mahalanobis distance between them amounted to 143.8). The greatest similarity was found for B. juncea and B. pekinensis (30.4). Mahalanobis distances between MS8 and other parental forms were statistically significant (Table 4).  Table  5). The number of the significant heterosis effects for individual hybrids ranged from eight (for H18) to 28 (for H1) ( Table 5). Considering observed traits, the lowest number of statistically significant heterosis effects was obtained for C16:1 (14), however, the highest was for C22:0 and C24:0 (62) (Table  Generally, 458 significant heterosis effects were obtained: 148 in F 1 , 151 in F 2 and 159 in F 3 ( Table 5). The number of the significant heterosis effects for individual hybrids ranged from eight (for H18) to 28 (for H1) ( Table 5). Considering observed traits, the lowest number of statistically significant heterosis effects was obtained for C16:1 (14), however, the highest was for C22:0 and C24:0 (62) ( Table 3). The obtained results indicated statistically significant heterosis effects in all three generations for 122 out of 242 cases. In 61 cases, we observed non-significant heterosis effects in each of the three generations. Statistically significant heterosis effects for individual hybrid had the same sign in particular generations, except H6 for C22:0, H11 for C20:1, H14 for C16:0, C18:1, C22:0 and C22:1, as well as H19 for C18:1 and C22:1 ( Table 4). The hybrid H5 is recommended for further inclusion in the breeding programs because it has the highest positively heterosis effects for C18:1, and negatively for C18:2, C18:3 and C22:1 in all three generations.    Individual traits are of different importance and have a different share in the joint multivariate variation. A study on the multivariate variation for treatments also includes identification of the most important traits in the multivariate variation of treatments. Canonical variables analysis (CVA) is a statistical tool making it possible to solve this problem [27,28]. The results of the analysis of canonical variables for investigated hybrids analyzed in three years are presented in Figure 3. The first two canonical variables explained jointly 66.37% total variation between hybrids. The first canonical variable accounted for 38.97% of the total variation, while the second canonical variable accounted for 27.40% (Figure 3). Individual traits are of different importance and have a different share in the joint multivariate variation. A study on the multivariate variation for treatments also includes identification of the most important traits in the multivariate variation of treatments. Canonical variables analysis (CVA) is a statistical tool making it possible to solve this problem [27,28]. The results of the analysis of canonical variables for investigated hybrids analyzed in three years are presented in Figure 3. The first two canonical variables explained jointly 66.37% total variation between hybrids. The first canonical variable accounted for 38.97% of the total variation, while the second canonical variable accounted for 27.40% (Figure 3). The most significant, positive, linear relationship between the first canonical variables was found for C16:0, C16:1, C18:0, C18:1, C18:2 and C18:3, while it was negative for C20:1, C22:0 and C22:1. The second canonical variable was significantly positively correlated with C16:0,18:1 and C18:2, while it was negatively correlated with C18:3, C20:0, C22:1 and C24:0. The greatest variation in terms of all The most significant, positive, linear relationship between the first canonical variables was found for C16:0, C16:1, C18:0, C18:1, C18:2 and C18:3, while it was negative for C20:1, C22:0 and C22:1. The second canonical variable was significantly positively correlated with C16:0,18:1 and C18:2, while it was negatively correlated with C18:3, C20:0, C22:1 and C24:0. The greatest variation in terms of all the 11 fatty acids jointly (measured Mahalanobis distances) was found for H3 in F 2 and H6 in F 1 (the Mahalanobis distance between them amounted to 19.919). The greatest similarity was found for H18 in F 1 and H18 in F 3 (0.150). According to the literature data, canonical variable analysis and Mahalanobis distances are statistical tools, which may be confirmed by their extensive application by breeders and geneticists [29][30][31][32][33][34].
Our results fully confirmed that assumption and proved that in order to identify the best Brassica genotypes in respect of heterosis for fatty acids requirements, the multivariate analysis of variance conducted for the eleven analyzed acids is a very useful statistical method of high importance.
Some authors reported that, most probably, it will soon be possible to change the profile of fatty acids content in Brassica seed oil, according to industry expectations [35,36].
Moreover, interspecific hybridization between appropriate Brassica species have a great potential in creating lines with a modified fatty acid composition [37,38]. In particular, as our results showed, the development of lines with a high oleic acid content seems promising. According to the magnitude of the predicted breeding values in this study, it is possible to select desirable breeding materials (hybrid line H5), which show the proper spectrum of the fatty acids in oil.

Conclusions
All analyzed parental genotypes as well as Brassica hybrid lines show statistically significant multivariate diversity for eleven fatty acids. The greatest variation of parental forms in terms of all the traits jointly was found for B. napus MS8 line and B. carinata. These two genotypes would be the best as parental forms in future breeding programs.
Analysis of canonical variables is a good tool for multivariate analysis of relationships among genotypes. The first two canonical variables explained jointly 97.16% and 66.37% total variation between parental forms and hybrids, respectively.
The highest number of significant heterosis effects was observed for behenic and lignoceric acids and for Brassica hybrid line H1.
The hybrid line H5 is recommended for further inclusion in the breeding programs, because it has the highest positively heterosis effect for oleic acid, and negatively for linoleic, linolenic and erucic acids in all three generations.