Analysis of a database of DNA profiles of 734 Hybrid Tea Rose

varietiesM.J.M. (René) Smulders, D. Esselink, R.E.

Voorrips and B. Vosman

Rose

Most important ornamental crop More than 25,000 varieties of modern rose

(Cairns, 2000) More than 10,000 hybrid tea varieties Rose list 2002: 13,000 varieties in

commercial trade Large collection of roses in “common

knowledge”

This large number may cause problems in Distinctness, Uniformity and Stability (DUS) testing How to compare to all varieties in common knowledge?

• Or to, e.g., varieties that can be grown in similar climatic zones? How to obtain the reference varieties?

• A reference collection is expensive, and susceptible to diseases• If the examination offices request reference varieties from the

breeders, the identity of the material should be verified

For this aspect of quality assurance molecular markers are ideally suited, as they are highly discriminating and can be assayed rapidly

DUS testing

Options for use in DUS testing

Within UPOV the Biochemical and Molecular Techniques Working Group (UPOV-BMT) discussed three options1. Molecular characteristics as predictor of traditional

characteristics a) the use of molecular characteristics which are directly linked to

traditional characteristics (gene specific markers)b) the use of a set of molecular characteristics which can be used reliably

to estimate traditional characteristics; e.g. quantitative trait loci

2. Calibrate threshold levels for molecular characteristics against the minimum distance between two varieties for traditional characteristics

3. Clearly distinguishable differences based on molecular characteristics as threshold levels for judging distinctness

The study: a database of rose variety

DNA marker profiles

Large set of microsatellite markers developed (TAG (2003) 106: 277-286)

For this study 11 markers were used (in 4 multiplex assays)

Data on 734 varieties, including duplicates

Topics

Discriminative power of the markers Reproducibility of the results Usefulness for identification purposes Usefulness in DUS research

option 2 option 3

Microsatellite analysis of rose (3 loci)

Original variety 1

Mutant of variety 1

Variety 2

Variety 3Variety 4

Variety 5

Dominant scoring (1/0) of microsatellite alleles

Allelic phenotypes for SpaGeDi analysis

Marker characteristics

Locus

Number of

alleles

Number of allelic

phenotypes

PIC value based on

allelic phenotypes

Frequency of most common

allelic phenotype

Number of different alleles in

allelic phenotype

with highest frequency

RhAB15 6 28 0.72 0.29 2 RhAB201 4 15 0.67 0.23 2 RhAB22 7 23 0.52 0.31 2 RhAB40 9 79 0.76 0.19 2 RhB303 6 37 0.76 0.12 3 RhD221 6 32 0.67 0.31 2 RhE2b 7 32 0.54 0.37 1 RhEO506 6 34 0.72 0.20 2 RhM405 4 9 0.73 0.4 4 RhO517 5 27 0.77 0.12 3 RhP519 6 32 0.71 0.22 3

Based on 407 genetically different varieties

Occurence of allelic phenotypes

Discriminative power

All seedling varieties had a unique DNA profile

Pairwise genetic similarities (Jaccard) of seedling varieties was < 0.9

Mutants had a genetic similarity of 1 with original variety

Reliability of the database

Number of assays that needed to be repeated Repetitions of multiplexed PCRs: 3-4%

Error rate: 1/1000 alleles(based on duplicated samples and mutants)

Genetic structure within the set of varieties

Analysed using Fst (population differentiation) Between years (Fst=0.0007 +/- 0.0005) Small differences among breeders (Fst= 0.0056

+/- 0.0011) PCA also shows no structure

Conclusion: All breeders use largely the same gene pool

For variety identification purposes the database can be a very efficient tool

When DUS testing stations would completely abandon living reference collections and obtain plant material for comparison from the breeders, they could easily check the identity of the material

A DNA fingerprint made when plant material has been submitted for DUS testing would be sufficient to spot such mutants or mutant groups right away, and based on that result one could include these mutant varieties for comparison

Marker data in relation to DUS testing

We have analysed the correlation between molecular and DUS

characteristics (option 2) whether or not candidate varieties would have

been granted PBR when only markers are used to show distinctness (option 3)

Genetic versus overall morphological similarity

Correlation with flower color

Most important distinguishing trait UPOV color grouping (classes 1-19, 34, 40,

46-47, 50) Question considered:

Does a higher genetic similarity between two varieties increase the probability that these varieties are in the same color group (have the same color)?

Correlation of genetic similarity with flower color class

Bold = significantly higher than the background level

(Mantel test, p<0.001)

Genetic similarity (Jaccard) above

Total number of pairs of varieties

number of matches (same color class)

% matches in the same similarity class

0.90 0 0 0.85 4 0 0 0.80 16 4 25 0.75 98 17 17 0.70 504 82 16 0.65 1957 216 11 0.60 4805 484 10 0.55 10899 921 8 0.50 14609 1181 8 0.45 21951 1626 7 0.40 14062 944 7

total 82621 6377

Option 2 (threshold level)

We found no correlation between genetic similarities based on morphological characters and molecular characters

except for genetic similarities above 0.7, but this only refers to 0.8% of the pairs of varieties

Option 3 (use molecular differences)

All seedling varieties showed a unique DNA fingerprint When granting of PBR would have been based solely on

molecular markers, the decisions made would have been identical to those based on traditional DUS testing on seedling varieties

Mutant varieties have a fingerprint that is identical to that of the variety they were derived from Evaluation of the mutant varieties for morphological

distinctness can be very efficiently done as the related mutants or mutant groups are readily identified using the molecular markers

Conclusions

Markers show a high discriminative power All seedling varieties can be uniquely identified Mutants are identical to original variety

Reliability of the data stored in the database is high Error rate 1/1000 alleles Multiplexing reduces errors

No correlation between similarities based on morphological and molecular characters An option 2 approach is not realistic for rose But option 3 could be implemented easily