Abstract

The potential usefulness of AFLP markers for cultivar distinctness was investigated on a collection of 83 rapeseed cultivars. A total of 324 markers were revealed by 17 primer combinations. Two primer combinations (60 markers) allowed to uniquely idendify all the 83 cultivars even a pair of cultivars different for one nanism gene. Three structure levels of the genetic diversity were tested with AMOVA and were significant : type of cultivars (spring vs winter type), country of origin (French vs German cultivars) and breeder (five different breeders). The a posteriori structure exhibited by a factorial method led to similar groupings. Three classical genetic distances (Nei & Li, Jaccard, Sokal & Michener) and a Jaccard distance weighted to take into account frequency of markers were compared for their ability to represent genetic relatedness between cultivars. The four genetic distances were highly correlated, all the pairwise coefficients were over 0.945. Dendrogams drawn on the basis of dissimilaritiy matrices led to define groups consistent with the genetic origin of the cultivars. Computed with a bootstrap procedure, the mean coefficient of variation for the four genetic distances was 8 %.

The results showed the great power of dicrimination of AFLP markers and their ability to represent the genetic relationships between cultivars consistently with their genetic origin. The four genetic distance estimators led to similar groupings of cultivars with a good precision of estimation which provided good statistical tools in the context of plant registration and protection.

Introduction

Since the notion of essential derivation was introduced in the last UPOV convention, the assessment of the genetic relatedness between cultivars has become a crucial issue. At present and for most species, registration or/and protection of a new variety only relies on morphological traits for the establishement of Distinction, Uniformity and Stability (DUS). Many of the traits considered are complex and their expression are influenced by the environmental conditions. Morever, for some crops like rapeseed, only few discriminant characters (about ten) are available for DUS testings and alternative descriptors are required. Because of the high degree of discriminant information, molecular markers have been widely and successfully applied for cultivars identification in many species. In particular, AFLP method (amplified fragment length polymorphism) easily provides a large number of markers on a single gel without requiring sequence information for their development.

The importance of reference collections in DUS testings is a problem where molecular markers could be hepful. At present, a new variety has to be compared with all the cultivars previously registered which implies the use of wide area to make comparisons. Molecular markers could be useful to permit a pre-screening of the closest varieties for direct comparisons with new varieties.

In the framework of plant protection, statistical tools are required to evaluate the utility of molecular data for assessing genetic proximity and dependance between cultivars. For this purpose, genetic distances seem to be a suitable approach.

The aim of this study is to (i) evaluate the utility of AFLP markers for the identification of rapeseed cultivars, (ii) apply AMOVA to test levels of structure of a collection of cultivars and (iii) compare several genetic distance estimators in the way they reflect true genetic associations between cultivars.

Materials and methods

Plant material

Eighty-three rape seed cultivars were studied including both spring and winter types from various oprigins. Among the collection were three pairs are near isogenic lines : Darmor and Darmor Nain are different for one nanism gene, B ms and E ms are the mâle sterile forms of B and E, respectively. Three synonimous cultivars (Apex, goeland and Lady) were also included in the collection. (Table 1).

DNA extraction and AFLP assays

Total DNA was extracted using a CTAB protocole from thirty seedlings per cultivar. AFLP analysis was performed according to the procedure described by Vos et al. (1995), and radiolabelling Eco RI to detect bands. 17 primer combinations involving five EcoRI and seven MseI primers provided clear interpretable patterns. AFLP bands were scored as absence (0) or presence (1).

Statistical analysis:

Structure of the genetic diversity

The a priori structure of the genetic diversity of the collection was tested with AMOVA (Excoffier et al. 1992). This method provides us with an estimate of the fraction of between-population diversity (i.e. Φ_st ), and allows to test the significance of structure levels. This method has been applied to test the structure (i) among winter and spring cultivars (type of cultivars), (ii) among countries of origin for winter cultivars and (iii) among breeding companies. The results were compared with a posteriori structure revealed by a principal component analysis (PCA).

Comparisons of genetic distance estimators

Three common distance estimators were computed between each pair of cultivars: the Jaccard’s distance (J) (1908), the Nei & Li’s distance (NL) (1979) and the Sokal & Michener’s distance (SM) (1958) as,

[1]

[2]

[3]

where n₁₁ is the number of bands shared by the cultivars x and y (positive matching), n₁₀ is the number of bands present in x and absent in y, n₀₁ the number of bands present in y and absent in x, and n₀₀ the number of bands absent both in x and y (negative matching). We also computed a weighted Jaccard’s distance (WJ) to take into account the frequency of each marker in the calculation of the distance.

Results

Polymorphism and power of dicrimination revealed by AFLP markers in rapeseed

17 primer combinations revealed a total number of 324 polymorphic bands. The number of markers per primer combination ranged from 12 to 30, with an average of 19.1. The most polymorphic primer combinations were E-AAC+M-CAA, E-AAC+M-CTT and E-AAG+M-CTT, which revealed 30 markers. The power of discrimination of each primer combination was estimated by computing the number of distinguished cultivars. This number ranged from 30 to 77. Only two primer combinations were required to distinguish all the cultivars: E-AAC+M-CAA and E-AAC+M-CTT or E-AAC+M-CTT and E-AAG+M-CTT

Structure of the genetic diversity (a priori and a posteriori)

The two factors (type of cultivar and country of origin) tested with AMOVA were both highly significant (Table 2). 32.9% of the molecular variance is distributed between winter and spring cultivars. Among winter cultivars, 11.4% of the variance is due to the partition between French and German cultivars. For the third factor, the genetic variance was significantly structured between each pair of breeders, except between breeders 2 and 3 who are French and German, respectively (Table 2).

Comparaisons between genetic distance estimators

Jaccard distance, Nei & Li distance, Sokal & Michener distance and weighted Jaccard distance ranged from 0.070 to 0.758, from 0.036 to 0.610, from 0.038 to 0.599 and from 0.166 to 2.541, respectively. For the four types of distance, Apex – Goeland and Goeland – Lady are the closest cultivars and the biggest distance was between a French winter cultivar and a Canadian spring cultivar. The three pairs of near isogenic lines were also included in the 16 nearest couples (Table 3). The simple coefficients of correlation calculated between the four genetic distances were very high for the six pairwise comparisons and were over 0.945.

Discussion

Polymorphism of AFLP markers and discriminatory power of the primer combination

Our results show that AFLP is a powerful tool to dicriminate rapeseed cultivars. 324 markers are revealed by 17 primer combinations, with an average of 19.1. This number is not very high in comparison with other intraspecific genetic diversity analyses using AFLP - 60 for sunflowers accessions (Hongtrakul et al. 1997)-. This result reflect the relative low genetic diversity in the european genetic pool. AFLP markers are higly polymorphic and only two primer combinations (60 markers) are required to distinguish the 83 cultivars.

Structure of the collection

The significant a priori structure of the collection (in term of genetic type, country of origin and breeder) assessed with AMOVA is consistent with the a posteriori structure revealed by PCA (Fig. 1). PCA exhibited a clear split between the spring and winter cultivars and high genetic relationships between cultivars from the same breeder. In our study, we can separate most of the French winter cultivars from the other European winter cultivars, showing the genetic originality of this group. These results are similar to associations revealed with RFLP markers reported in Diers et al. (1994) or with RAPD markers in Mailer et al. (1994). AMOVA seems to be a good tool to test significance of groupings.

Comparison between genetic distances.

Jaccard and Nei & Li are very highly correlated (0.996) and lead to identical rankings of genetic distances. This was expected, because NL can be easily expressed as an increasing function of J. The high correlation between J and SM (0.983) was not obvious. The difference between these distances (formula [1] and [3]) came from negative matches which are taken into account in the denominator of SM distance. One explanation, supported by Peltier et al. (1994), is that in a case of intra-specific study, an allelic relation exist between presence and absence of a band and a negative matching is an indication of similarity leading to same kind of results with SM and J in our study. The weighting of Jaccard (WJ) distance by the inverse of the PIC provided similar relationships between cultivars to Jaccard ones. Because of the struture of the marker frequency between spring and winter cultivars, WJ lead to take the most different cultivars away from each other.

Choice of a genetic distance estimator for essential derivation

In the framework of plant protection, the choice of the genetic distance is crucial for determining the level of relatedness between cultivars. For the distinctness and without any genetic consideration, J and NL are independent of the cultivar samples because only bands present in x and/or in y are considered. For SM, negative matches are counted and if a new cultivar carries a new band absent in the cultivars previously registered, this becomes a new negative matching for these cultivars and the distance will change. The stability of genetic distance is a very attractive quality for breeders because a distance between two cultivars is constant when the number of cultivars in the reference collection increase.

A contrario, the disadvantage of J results in the difficulty of finding statistical distribution of this distance which is important to calculate a confidential interval. This difficulty comes from the denominator, which is not a constant but a random variable. It is easier to work with euclidian distances like SM or Rogers distance. They can be modelled as a binomial variable and their statistical properties are well known (Dillmann et al. 1997).

To conclude, this study shows the great discriminatory power of AFLP makers and their capacity to well represent the genetic relationships between rapeseed cultivars consistently with their genetic origin. AMOVA can be successfully applied to test significance of groups in reference collection but methodologies to define groupings must be investigated. The different genetic distance estimators are very correlated and lead to very similar associations between cultivars. However, the rankings are a bit different especially for near cultivars which can lead to different decisions about the genetic dependance between varieties. Then, the choice of a genetic distance in the context of plant protection has important consequences and further investigations are needed about the precision of their estimations and the conditions of their applications (in term of type of molecular markers, genetic structure of the cultivars, diversity of reference collections, breeding programs).

References

1. Diers BW, Osborn TC (1994) Genetic diversity of oilseed Brassica napus germ plasm based on restriction fragment length polymorphisms. Theor Appl Genet 88:662-668

2. Dillmann C, Charcosset A, Goffinet B, Smith JSC, Dattée Y (1997) Best linear estimator of the molecular genetic distance between inbred lines. In: Krajewski P, Kaczmarek Z (eds) Advances in biometrical genetics. Proceedings of the tenth meeting of the EUCARPIA section biometrics in plant breeding, 14-16 may 1997, Poznan, pp 105-110

3. Excoffier L, Smouse PE, Quattro JM (1992) Analysis of molecular variance inferred from metric distances among DNA haplotypes: application to human mitochondrial DNA restriction data. Genetics 131:479-491

4. Jaccard P (1908) Nouvelles recherches sur la distribution florale. Bull Soc Vaud Sci Nat 44:223-270

5. Mailer RJ, Scarth R, Fristensky B (1994) Discrimination among cultivars of rapeseed (Brassica napus L.) using polymorphism amplified from arbitrry primers. Theor Appl Genet 87:697-704

6. Nei M, Li WH (1979) Mathematical models for studying genetic variation in terms of restriction endonucleases. Proc Natl Acad Sci USA 76:5269-5273

7. Peltier D, Chacon H, Tersac M, Caraux G, Dulieu H, Berville A (1995) Utilisation des RAPD pour la construction de phénogrammes et de phylogrammes chez Petunia. In: Techniques et utilisations des marqueurs moléculaires. Coll Les colloques INRA

8. Sokal RR, Michener CD (1958) A statistical method for evaluating systematic relationships. Univ Kansas Sci Bull 38:1409-1438

9. Vos P Hogers R, Bleeker M, Reijans M, Lee T van de, Hornes M, Frijters A, Pot J, Peleman J, Kuiper M, Zabeau M (1995) AFLP: a new technique for DNA fingerprinting. Nucleic Acids Res Vol 23, No 21: 4407-4414

10.

Table 1 Origin of the plant material with, in parenthesis, the number of cultivars

Type of cultivar	Country of origin	Breeding company
Winter (68)	France (26) Germany (26) 4 other European countries (15) Japan (1)	Breeder 1 (14) Breeder 2 (9) 3 other French breeders (3) Breeder 3 (5) Breeder 4 (7) Breeder 5 (9) 3 other German breeders (5)
Spring (15)

Table 2 Analysis of molecular variance of three factors on molecular data matrix

Factor	Levels	sample size	Φst
Type	Winter Spring	68 15	0.329 ***
Country of origin	France Germany	26 26	0.114 ***
Breeders	1	14		1	2	3	4
	2	9	2	0.112 *
	3	5	3	0.205 **	0.090 NS
	4	7	4	0.255 ***	0.196 ***	0.202 **
	5	9	5	0.207 ***	0.087 **	0.094 *	0.110 **

^NS: non significant, *: P<5%, **: P<1%, ***: P< 0.1%

« « «

Table 3 Genetic distances between pairs of related cultivars

Pairs of cultivars	Nei & Li distance		Jaccard distance		Weighted Jaccard distance		Sokal & Michener distance
	value	rank *	value	rank *	value	rank *	value	rank *
Apex – Goeland Goeland – Lady Apex – Lady Darmor – Darmor Nain A ms – A B ms – B	0.036 0.044 0.065 0.069 0.060 0.082	1 3 8 11 5 14	0.070 0.084 0.121 0.129 0.112 0.152	1 3 8 10 5 14	0.166 0.254 0.343 0.296 0.312 0.466	1 3 9 4 6 16	0.038 0.045 0.067 0.076 0.061 0.093	1 3 10 12 4 14

* rank among the nearest distances

Fig. 1 principal component analysis of 83 rapeseed cultivars based on 324 AFLP markers