Tree Genetics & Genomes

, 13:14

Unravelling genetic diversity and cultivar parentage in the Danish apple gene bank collection

  • Bjarne Larsen
  • Torben Bo Toldam-Andersen
  • Carsten Pedersen
  • Marian Ørgaard
Original Article

DOI: 10.1007/s11295-016-1087-7

Cite this article as:
Larsen, B., Toldam-Andersen, T.B., Pedersen, C. et al. Tree Genetics & Genomes (2017) 13: 14. doi:10.1007/s11295-016-1087-7
Part of the following topical collections:
  1. Germplasm Diversity

Abstract

Characterization of apple germplasm is important for conservation management and breeding strategies. A set of 448 Malus domestica accessions, primarily of local Danish origin, were genotyped using 15 microsatellite markers. Ploidy levels were determined by flow cytometry. Special emphasis was given to pedigree reconstruction, cultivar fingerprinting and genetic clustering. A reference set of cultivars, mostly from other European countries, together with a private nursery collection and a small set of Malus sieversii, Malus sylvestris and small-fruited, ornamental Malus cultivars, was also included. The microsatellite markers amplified 17–30 alleles per loci with an average degree of heterozygosity at 0.78. We identified 104 (23%) duplicate genotypes including colour sports. We could infer first-degree relationships for many cultivars with previously unknown parentages. STRUCTURE analysis provided no evidence for a genetic structure but allowed us to present a putative genetic assembly that was consistent with both PCA analysis and parental affiliation. The Danish cultivar collection contains 10% duplicate genotypes including colour sports and 22% triploids. Many unique accessions and considerable genetic diversity make the collection a valuable resource within the European apple germplasm. The findings presented shed new light on the origin of Danish apple cultivars. The fingerprints can be used for cultivar identification and future management of apple genetic resources. In addition, future genome-wide association studies and breeding programmes may benefit from the findings concerning genetic clustering and diversity of cultivars.

Keywords

Malus domestica cultivars Germplasm Pedigree Genetic structure Microsatellites Ploidy 

Introduction

Genotyping of plant genetic resources is a crucial task for gene banks if they are to ensure preservation of maximum genetic diversity. Germplasm collections of long-lived field-cultivated material that is expensive to maintain, such as the genus Malus, are especially important in this context, and genotyping should have high priority, because in such collections, there is a risk that accessions lose their identity and that duplicates have accumulated under different accession names. This can happen during renewing of a collection that has experienced a series of procedures, such as grafting, labelling and planting. Cultivar fingerprinting is an essential tool that complements classical phenotyping to verify that accessions are true to type after repropagation of a collection. Also, where germplasm has not been systematically collected, genotyping is a useful tool to pinpoint areas of diversity that are ‘overrepresented’ and discovers unknown diversity within a collection.

The largest collection of Danish apple (Malus domestica Borkh.) cultivars is kept in the gene bank collection at the Pometum (University of Copenhagen, Denmark). It contains around 300 named accessions originating from all over the country except the northern-most and western-most regions, where climatic conditions are marginal for apple cultivation. The Pometum collection was founded in 1863, and many of the accessions were collected and recorded during the early part of the twentieth century, when cultivar trials were performed (Pedersen 1925). Extensive pomological description works (Bredsted 1893; Matthiessen 1913; Pedersen 1950) provide data on approximate year and locality of origin for many of the cultivars. Only rarely, however, do we know their parental origin since the great majority of the cultivars originated from random pollination events. Here, genotyping is useful to reveal parentages, e.g. by showing common origin such as parent-offspring or sibling relationships. Pedigree reconstructions are also important in population-based genome-wide association studies (GWAS), and knowledge of genetic diversity is of major importance in breeding programmes for selection of genetically distinct breeding partners.

Simple sequence repeat (SSR) markers have proven to be a powerful tool and have been the major choice of markers for describing the germplasm in various apple collections (Evans et al. 2011; Fernández-Fernández 2010; Garkava-Gustavsson et al. 2008; Gasi et al. 2010; Gross et al. 2012; Guarino et al. 2006; Lassois et al. 2016; Patzak et al. 2012; Urrestarazu et al. 2016; Urrestarazu et al. 2012; van Treuren et al. 2010). A standardized set of SSR markers was proposed by Evans et al. (2007) to access, fingerprint and evaluate genetic resources. This set has proved to be very efficient for genotyping (Potts et al. 2012) and enables comparisons of genetic resources across collections and countries (Lassois et al. 2016; Urrestarazu et al. 2016). Since triploids are relatively common in apple gene banks, with a frequency of between 10% (Gross et al. 2012) and 24% (Urrestarazu et al. 2012), the ploidy level should be considered in relation to parental analysis.

Here, we study a gene bank collection of 287 apple accessions of Danish origin. As references, we included 86 accessions originating primarily from other European countries but which have been cultivated in Denmark. An additional set of 75 accessions from a private apple tree collection was included to check if they were identical to known cultivars in the collection or contained additional variation. Finally, we included 36 accessions of Malus sieversii, Malus sylvestris and small-fruited, ornamental cultivars. The goal was to decipher the genetic structure and diversity, identify duplicates, determine ploidy levels and obtain more information about the origins of Danish apples.

Material and methods

Plant material

Plant material (Online Resource 1) was collected as young leaves from vigorously growing shoots at the Pometum’s gene bank collection (University of Copenhagen, Taastrup, Denmark). In addition, material from the private nursery collection of Assens Planteskole, Funen, Denmark (Assens) was included. A few accessions were provided from other sources (see Online Resource 1). Leaves were freeze dried for 48 h. Ten milligrams of dry leaf material was used for DNA extraction using DNeasy® 96 Plant Kit (Qiagen®, Hilden, Germany), following the manufacturer’s protocol.

Ploidy level

The ploidy level was determined for all local, Danish cultivars including few reference accessions. A 2-cm2 area of a freshly harvested young leaf from each individual was chopped for 1 min in a petri dish in 400 μL WPB buffer (0.2 M Tris-HCL (pH 7.5), 4 mM MgCl2, 2 mM EDTA Na22H2O, 86 mM NaCl, 1% PVP-100, 1% (v/v) Triton X-100) and left for 5 min. The solution was filtrated through 20-μM filter and stained with 800 μL 4′,6-diamidino-2-phenylindole (DAPI) solution (137 mM NaCl, 2.7 mM KCL, 10 mM Na2HPO4, 2 mM KH2PO4, 1 μg/mL DAPI) for 10 min in darkness. The ploidy level was determined using flow cytometer BD FACSAria™ Illu counting 5000 events. Selected cultivars with known ploidy level were used as standards, such as the triploids ‘Belle de Boskoop’ and ‘Gravensteiner’ (syn. ‘Graasten’) and the diploids ‘Ingrid Marie’ and ‘Filippa’ and others. On this basis, two fully separated flow cytometry graphs were created and used as standards for diploid and triploid accessions, making it possible to unambiguously distinguish between diploid and triploid accessions.

Self-incompatibility alleles (S-RNase alleles) were determined in a separate study (Larsen et al. 2016a).

PCR amplification

The PCR mix contained 1× key buffer (10× key buffer, VWR® International, Radnor, PA, USA), 0.5 mM MgCl2, 0.2 μM dNTP, 0.05 μM forward primer, 0.2 μM reverse primer, 0.25 μM fluorochrome-labelled M13 primer, 0.5 U VWR® Taq DNA polymerase and 10 ng template DNA in a final volume of 20 μL. For the primers CH02c06 and CH03d07, twice as much MgCl2 was used.

The PCR reaction was carried out with a ‘touch-down’ programme initiated at 94 °C for 2 min, followed by 18 cycles of 94 °C for 1 min, annealing for 30 s starting at 64 °C falling 0.5 °C for each cycle and prolonged for 1 min at 72 °C, followed by 20 cycles of 94 °C for 1 min, 55 °C for 1 min, 72 °C for 1 min and finally storage at 4 °C.

Amplification products were labelled with FAM, VIC or NED using the fluorochrome-labelled M13 primer (CACGACGTTGTAAAACGAC) which was incorporated in amplification products during initial PCR cycles as described by Schuelke (2000). We mixed 3 μL of each of three different fluorochrome-labelled products with 31 μL water. From this solution, 2 μL was added to 12 μL loading buffer (1 mL 0.1× TE buffer and 40 μL of an internal size standard ranging from 58 to 508 bp).

Initially, we tested 18 SSR markers listed by Potts et al. (2012) plus the marker CH-Vf1 (Vinatzer et al. 2004). However, the primers NZ05g8, CH04f10 and CH-Vf1 did not give satisfactory amplification and were not included in further analysis. For SSR genotyping, a set of 15 SSR markers (Table 1) was used.
Table 1

SSR marker information for 18 markers

SSR name

Forward primer

Reverse primer

Sequence repeat

NA

Size range (bp)

Hi02c07

AGAGCTACGGGGATCCAAAT

GTTTAAGCATCCCGATTGAAAGG

GTT

17

114–153

CH02c09

TTATGTACCAACTTTGCTAACCTC

AGAAGCAGCAGAGGAGGATG

GA

20

237–277

CH04e05

AGGCTAACAGAAATGTGGTTTG

ATGGCTCCTATTGCCATCAT

GA

27

171–251

GD147

TCCCGCCATTTCTCTGC

AAACCGCTGCTGCTGAAC

AG

20

133–171

CH04c07

GGCCTTCCATGTCTCAGAAG

CCTCATGCCCTCCACTAACA

GA

21

111–162

CH02c11

TGAAGGCAATCACTCTGTGC

TTCCGAGAATCCTCTTCGAC

GA

28

218–263

CH01f02

ACCACATTAGAGCAGTTGAGG

CTGGTTTGTTTTCCTCCAGC

GA

30

167–241

CH01h10

TGCAAAGATAGGTAGATATATGCCA

AGGAGGGATTGTTTGTGCAC

GA

22

99–157

CH02d08

TCCAAAATGGCGTACCTCTC

GCAGACACTCACTCACTATCTCTC

GA

24

214–275

CH05f06

TTAGATCCGGTCACTCTCCACT

TGGAGGAAGACGAAGAAGAAAG

GA

22

182–209

CH01f03b

GAGAAGCAAATGCAAAACCC

CTCCCCGGCTCCTATTCTAC

GA

19

141–246

CH01h01

GAAAGACTTGCAGTGGGAGC

GGAGTGGGTTTGAGAAGGTT

AG

22

110–177

CH02c06

TGACGAAATCCACTACTAATGCA

GATTGCGCGCTTTTTAACAT

GA

26

214–288

CH03d07

CAAATCAATGCAAAACTGTCA

GGCTTCTGGCCATGATTTTA

GA

28

180–250

GD12

TTGAGGTGTTTCTCCCATTGGA

CTAACGAAGCCGCCATTTCTTT

CT

19

157–208

CH04f10

GTAATGGAAATACAGTTTCACAA

TTAAATGCTTGGTGTGTTTTGC

GA

nd

189–288

CH-Vf1

ATCACCACCAGCAGCAAAG

CATACAAATCAAAGCACAACCC

AG

nd

142–191

NZ05g8

CGGCCATCGATTATCTTACTCTT

GGATCAATGCACTGAAATAAACG

GA

nd

nd

NA number of alleles per locus, nd not determined for markers not used for genotyping

Fragment separation and detection were carried out using ABI 3130xl DNA analyser (Applied Biosystems, Foster City, CA, USA). Subsequent analysis of fragment sizes was done by the software GeneMarker® v. 2.2.0 (Softgenetics® LLC, State College, PA, USA), based on an internal size standard added to each sample. Afterwards, each band was manually checked.

Data analysis

Data was imported to the software R version 3.2.2, where a distance matrix was computed using the package Polysat version 1.4 (Clark and Jasieniuk 2011) following the guidelines in ‘POLYSAT version 1.4 Tutorial Manual’, by Clark (2015, https://cran.r-project.org/web/packages/polysat/vignettes/polysattutorial.pdf) to perform PCA analysis on all genotypes. The number of rare alleles (NB), effective number of alleles (Ae), expected heterozygosity (He) and observed heterozygosity (Ho) were estimated with the software SPAGeDi ver. 1.5 (Hardy and Vekemans 2002). The software was also employed to calculate the pairwise Fst differentiation levels between the five putative groups defined by STRUCTURE using ANOVA approach. For diploid samples, parentage analysis and identification of duplicate genotypes were performed with the software ML-Relate (Kalinowski et al. 2006). In order to check parent-offspring relations the software FRANz (Riester et al. 2009) was employed to predict parentages and could confirm parentages generated by ML-Relate. Parentage analysis for triploid samples was performed with the software Colony version 2.0.5.9 (Jones and Wang 2010) after converting all microsatellite data to binary data (Rodzen et al. 2004). Parent-offspring relations were considered for accessions sharing one allele in each of the 15 SSR loci and were checked manually afterwards. All accessions with duplicate genotypes were also checked manually to confirm an identical profile.

Genetic structure analysis of unique diploid and triploid cultivars was performed using the software STRUCTURE v. 2.3.4 (Pritchard et al. 2000) admixture model, where the number of clusters were determined with K = 2–12 over 20 runs. The best number of clusters (K) was determined as recommended by Evanno et al. (2005) using the online programme STRUCTURE HARVESTER (Earl and vonHoldt 2012). Accessions were assigned to the group to which they had the highest coefficient of relationship, and we considered a probability of assignment of 0.80 or greater to be strong assignment to a group. After evaluation of delta K graph (Online Resource 2) and cultivar clustering, a non-significant putative structure of five groups is proposed since it is consistent with PCA and parental analysis such as clustering of Cox’s Orange and derivatives.

Results

Diversity and ploidy

We found between 17 (Hi02c07) and 30 (CH01f02) alleles per loci among the 484 accessions studied in total. For the 448 M. domestica cultivars, between 12 (Hi02c07) and 26 (CH01f02) alleles per loci were amplified, with an average of 19 (Table 2). The 15 SSR loci amplified a total number of 284 alleles of which 193 were rare (frequency < 5%). There were 42 unique (private) alleles that were identified in only one accession. The proportion of genotypes with unique alleles was larger in the Assens collection (19%) than in the Danish collection and reference collection with 7 and 8%, respectively. The He and Ho were 0.81 and 0.78, respectively, and the observed heterozygosity was slightly higher for the Danish cultivars (0.79) than for the reference set of cultivars (0.76).
Table 2

Estimates of genetic diversity at four levels: all studied cultivars, local Danish cultivars, reference cultivars and ‘Assens’ collection

 

All cultivars (n = 448)

Set of Danish cultivars (n = 287)

Reference cultivars (n = 86)

Private collection Assens (n = 75)

NA

NB

Ae

He

Ho

NA

NB

Ae

He

Ho

NA

NB

Ae

He

Ho

NA

NB

Ae

He

Ho

Hi02c07

12

7

3.32

0.70

0.45

11

6

3.29

0.70

0.44

6

0

2.65

0.62

0.35

8

3

4.25

0.76

0.60

CH02c09

15

9

5.94

0.83

0.87

12

6

5.78

0.83

0.86

8

2

5.16

0.81

0.92

14

8

6.77

0.85

0.84

CH04e05

23

18

5.46

0.82

0.32

19

15

4.97

0.80

0.37

9

3

5.37

0.81

0.24

14

8

6.37

0.84

0.23

GD147

16

11

6.5

0.85

0.91

16

11

6.52

0.85

0.92

13

8

5.51

0.82

0.84

14

8

6.99

0.86

0.97

CH04c07

15

7

6.25

0.84

0.88

14

6

6.24

0.84

0.86

12

5

5.96

0.83

0.92

11

5

5.85

0.83

0.89

CH02c11

24

14

9.87

0.90

0.91

21

11

9.71

0.90

0.92

14

5

8.65

0.88

0.91

13

4

10.34

0.90

0.85

CH01f02

26

19

9.11

0.89

0.92

23

16

9.2

0.89

0.91

16

10

6.92

0.86

0.97

18

9

10.04

0.90

0.91

CH01h01

16

10

6.88

0.85

0.86

14

8

6.65

0.85

0.87

12

6

6.84

0.85

0.80

10

3

7.08

0.86

0.90

CH02c06

24

15

9.65

0.90

0.87

21

13

9.51

0.89

0.90

13

5

7.55

0.87

0.82

17

10

9.37

0.89

0.82

CH03d07

24

17

7.39

0.86

0.88

19

12

7.07

0.86

0.89

12

5

7.24

0.86

0.80

17

9

8.48

0.88

0.91

GD12

16

12

2.18

0.54

0.54

13

9

2.14

0.53

0.52

10

7

2.51

0.60

0.67

10

7

1.95

0.49

0.47

CH01h10

16

13

3.17

0.68

0.76

14

10

3.5

0.71

0.80

11

8

2.3

0.56

0.61

13

10

3.18

0.69

0.77

CH02d08

20

14

6.04

0.83

0.86

17

12

5.97

0.83

0.89

14

8

5.21

0.81

0.84

17

11

6.28

0.84

0.74

CH05f06

22

16

6.16

0.84

0.89

18

12

6.45

0.85

0.88

15

10

4.79

0.79

0.90

14

9

5.68

0.82

0.91

CH01f03b

15

11

4.57

0.78

0.81

12

8

4.16

0.76

0.80

11

7

5.24

0.81

0.79

13

9

5.26

0.81

0.87

Mean

19

13

6.17

0.81

0.78

16

10

6.08

0.81

0.79

12

6

5.46

0.79

0.76

14

8

6.53

0.82

0.78

No unique alleles

42

    

21

    

7

    

14

    

NA number of alleles per locus, NB number of rare alleles that are less than 5% frequent, Ae effective number of alleles, He expected heterozygosity, Ho observed heterozygosity

Of the 393 accessions whose ploidy levels were determined by flow cytometry (Online Resource 3), 17% were triploids and the remainders were diploids. In the Danish cultivar collection, 22% triploids were identified. However, the exact chromosome number can well vary from 34 and 51 in cases of aneuploidity. Since such minor variation in chromosome number cannot be detected by flow cytometry (Chagné et al. 2015), cultivars were categorized as ‘triploid’ (polyploid) or ‘diploid’. Ploidy levels determined by flow cytometry were confirmed by SSR data where diallelic or triallelic loci were found for diploid or triploid genotypes, respectively.

Duplicate genotypes

Analysis of the 15 SSR loci among all 448 M. domestica accessions revealed a unique genotype for 344 accessions (Fig. 1a). The remaining accessions (69 + 35; Fig. 1b) were duplicates, where two or more accessions had an identical genotype, and among the duplicates, 23 accessions were from the private tree collection Assens. However, since the duplicate genotypes also contain skin colour mutants which are phenotypically distinctive, a subdivision of the duplicate genotypes was needed, so we divided them into different groups (Fig. 1b; Online Resource 4). These were either subclones, such as phenotypically distinctive colour mutants that the markers were unable to separate, or duplicates with name synonyms originating from (1) previously reported synonyms where the name often indicates that cultivars are associated, e.g. ‘Broholm’ and ‘Broholm Rosenæble’; (2) not previously reported synonyms where the most likely scenario is that the identity of an apple tree has been lost and it has been given a new name, e.g. ‘Grevinde Ahlefeldt’, which is identical to ‘Laxton’s Superb’ or (3) incorrectly named accessions. For accessions with duplicate genotypes, except colour sports, we compared fruit morphology and tree stature, which in all cases supported the conclusion that accessions were duplicates.
Fig. 1

Distribution of unique genotypes and duplicate genotypes. a These are divided into diploids and triploids, respectively. b Duplicate genotypes were classified as sports (not actual duplicates) and different name synonyms (actual duplicates). Duplicates counted as surplus

Pedigree reconstruction

The SSR data allowed us to study putative relatedness such as parent-offspring relationships. Historical information regarding where and when cultivars originated has in many cases allowed us to point out the putative parent and offspring in the relationship (Fig. 2 and Online Resource 5). S-RNase alleles (Larsen et al. 2016a) also supported these parent-offspring relationships.
Fig. 2

Network of first-degree relationships among included apple accessions. Arrows point from parent to offspring. Information given: cultivar name, accession number, ploidy level, S-RNase alleles and approximate geographical origin and year of origin. Further parent-offspring relations are shown in Online Resource 5

The two cultivars Cox’s Orange and ‘Pigeonnet blanc d’hiver’ (syn. ‘Weißer Wintertaubenapfel’) have by far the highest number of putative offspring relatives in the collection. These were followed by, e.g. Filippa, ‘Elmelund’, ‘Rød Nonnetit’ and ‘Maglemer’ in addition to the non-Danish cultivars ‘James Grieve’, ‘Transparante Blanche’ and ‘Golden Delicious’.

In some cases, a triploid cultivar originated from two diploid parents. We found that Golden Delicious, for example, most probably contributed an unreduced 2n gamete, which fertilized a normal reduced n gamete from ‘Jonathan’ resulting in the triploid ‘Jonagold’, as previously reported by Gianfranceschi et al. (1998). Similarly, Cox’s Orange (2n) gave two sets of chromosomes to Holsteiner Cox (3n). Apparently, triploid cultivars have given rise to triploid offspring: Orleans Reinet (3n) probably gave a 2n gamete to Blenheim (3n), whereas Borgherre (3n) has probably given one set of chromosomes to Mormors æble (2n; Online Resource 5).

Population structure

On basis of population STRUCTURE, we show a putative assemble of five genetic clusters (Fig. 3a). No more than 21% of the total number of genotypes, however, has strong assignment (>0.80) to a specific group with a similar trend in each group (Table 3). Even though this putative assembly was not opposed by either PCA analysis (Fig. 3b) or parental analysis, there was no support to conclude that a population structure exists. This is supported by the Bayesian model-based approach and delta K values in a very low and tight range (delta K < 1.5) with a similar number of genotypes assigned to each group and low genetic differentiation among the five groups (Table 4).
Fig. 3

Population structure. a The STRUCTURE analysis of 344 unique M. domestica genotypes with five non-significant, putative groups. b PCA plot of all studied accessions. Danish Malus domestica cultivars (grey ellipse), reference collection of cultivars with other geographical origins than Danish (green ellipse), M. sieversii (blue ellipse), M. sylvestris (black ellipse), small-fruited ornamental Malus cultivars (Malus sp.) (orange ellipse). Green solid cloud indicates Cox’s Orange with first-degree relatives. Variance explained by PCA1 = 7.6%, PCA2 = 6.8%

Table 3

Average measures of genetic diversity for five subgroups identified by STRUCTURE analysis (n = 344)

 

No. acc.

No. strongly affiliated acc.

AR

NAe

Unique alleles

He

Ho

Group 1

80

16

14.73

6.41

18 (0.08)

0.82

0.79

Group 2

60

17

10.53

5.13

2 (0.01)

0.78

0.79

Group 3

72

15

11.60

5.21

9 (0.05)

0.77

0.79

Group 4

84

15

12.20

5.95

11 (0.06)

0.81

0.76

Group 5

48

10

9.60

4.88

2 (0.01)

0.77

0.75

No. acc. number of accessions assigned to group, No. strongly affiliated acc. number of accessions with strong affinity to group (>0.80), AR allelic richness, NAe effective number of alleles, Unique alleles number of alleles that are unique for the group and their frequency within the group in parenthesis, He expected heterozygosity, Ho observed heterozygosity

Table 4

Pairwise estimates of Fst for five putative groups

Group

1

2

3

4

5

1

    

2

0.030

   

3

0.035

0.033

  

4

0.023

0.027

0.032

 

5

0.047

0.056

0.052

0.044

Among the five putative groups (Fig. 3a), the largest genetic differentiation was found between groups 2 and 5 (Fst = 0.056) (Table 4). The genotype with strong assignment to group 2 is Pigeonnet blanc d’hiver with derivatives, and strongly assigned members of group 5 are derivatives of Cox’s Orange. The lowest Fst value (0.023) were found between group 1 that is composed of several relatively early season cultivars, especially those of Danish origin, and group 4 with several cultivars that have been less frequently grown in Denmark. Group 3 where Golden Delicious and derivatives are strongly affiliated shares the lowest Fst value (0.032) with group 4. The two later groups consist of several cultivars that have been less commonly cultivated in Denmark. The most distinct group is group 5 (Fst values ranging from 0.043 to 0.056), whereas group 4 (Fst values ranging from 0.022 to 0.043) is least distinct from the other four groups. Groups 1 and 4 are the most diverse groups (allelic richness 14.73 and 12.20, respectively), whereas group 5 is the least diverse (allelic richness 9.60) (Table 3). PCA plot (Fig. 3b) indicates a cluster of Cox’s Orange and first-degree relatives and no distinction between Danish cultivars and the reference collection. On Malus species level, it indicates that M. sieversii, M. sylvestris and small-fruited Malus cultivars are intermixed with M. domestica cultivars.

Discussion

The high level of polymorphism with 12 to 26 alleles per locus among cultivars (Table 2) and complex genetic structure (Fig. 3) demonstrates a substantial genetic variation. The numbers of alleles per locus are within the range found in previous studies (Garkava-Gustavsson et al. 2008; Gasi et al. 2010; Lassois et al. 2016; Potts et al. 2012; Urrestarazu et al. 2016; Urrestarazu et al. 2012) and supports the assumption that the chosen set of SSR markers has the power to make accurate cultivar fingerprints and precise germplasm characterizations. The set of 10 markers used by Potts et al. (2012), who designated 10 SSR markers in the set to be sufficient for genotyping, is included in the 15 markers used here, except for the marker CH04f10. The number of markers used here also gave strong predictions of parentages (Al-Atiyat 2015) since they are distributed over different linkage groups and are highly polymorphic.

We identified 42 unique alleles, which is higher than the number found by Urrestarazu et al. (2012) but comparable to the number found among 21 SSR loci by Lassois et al. (2016). In the Danish collection, we found 7% unique alleles (Table 2), pointing to the fact that the collection contains germplasm unique within European material. However, the highest percentage of unique alleles (19%) was found in the Assens collection that suggests a valuable diversity in this assemble.

Duplicate accessions

We found 23% duplicate genotypes (Fig. 1a) which is in accordance with previous studies where between 21% (Gross et al. 2012) and 32% (van Treuren et al. 2010) dublicate genotypes have been found. The highest number of identical genotypes was found for Gravensteiner (15 subclones) followed by Almindelig Pigeon, Filippa and Cox’s Orange (all with 6 subclones), all of which have been among the most frequently planted cultivars in Denmark for decades. It is important to note, however, that these also include distinctive skin colour mutants such as Ribston and Ribston Red, which the SSR markers are not able to separate (Gross et al. 2012). This needs to be considered, since it in many cases is reasonable to keep a number of these subclones in a collection. The subclones mainly comprises red-coloured mutants, well known from several commercial cultivars such as Ingrid Marie, Belle de Boskoop and Elstar, where the red-coloured mutant today is cultivated and marketed to a larger extent than the original clone. The presence of such subclones can thereby in many cases be justified in a gene bank collection, whereas duplicate accessions hidden under various synonyms can be exposed by DNA markers and should be eliminated from the collection.

Looking apart from the duplicate genotypes in the Assens collection (Online Resource 4), 10% of the identified duplicate accessions were found in the Danish apple gene bank collection. Thus, we have demonstrated the potential for using SSR markers to reduce the number of accessions in the collection without losing genetic variation. Such screens of germplasm collections can therefore pay off when they result in a reduction in the number of accessions that needs to be maintained and allowing a more efficient use of resources for the conservation of maximum genetic diversity.

The ‘Cox’s’ apple types

The majority of the cultivars originate from open, uncontrolled pollination events. In some cases, one parent has been suggested in the literature, which we were able to confirm in many cases. The cultivar Ribston has long been assumed to be parental to both Cox’s Orange and Cox’s Pomona (Pedersen 1950). This could be confirmed for Cox’s Orange, whereas Cox’s Pomona did not have a parent-offspring relation with Ribston.

In the middle of the nineteenth century, Cox’s Orange came to Denmark and was the most commonly planted cultivar during early twentieth century (Pedersen 1950). The high planting frequency and many bee-facilitated pollination events made it ancestor to a cluster of cultivars (Fig. 2). This cluster is strongly affiliated to group 5 (Fig. 3), and their close genetic relationship makes it the least diverse group (Table 3). In comparison, groups 1 and 4 hold the highest diversity, and they are composed of many unrelated cultivars that frequently lack close affiliates in the studied material. The derivatives of Cox’s Orange (group 5) form the genetically most distinctive group (Table 4).

The cultivar Ingrid Marie became highly popular in Nordic countries during the twentieth century. Earlier, it was thought to be an offspring of Cox’s Orange (Pedersen 1950), which we could confirm. However, contradictory to previous findings made on the basis of RAPD markers (Garkava-Gustavson and Nybom 2003), we unambiguously identified the other parent of Ingrid Marie to be Cox’s Pomona. We conclude that Ingrid Marie is the result of the hybridization event Cox’s Orange × Cox’s Pomona.

The ‘Pigeon’ apple types

The cultivar Pigeonnet blanc d’hiver has been cultivated in Denmark at least since the latter part of the eighteenth century (Bredsted 1893) and might well be an important ancestor to the gene pool, although ‘Pigeon fra Juellinge’ is more related to ‘Guldborg’, and Almindelig Pigeon is putatively an offspring of ‘Dansk Rosenhäger’. The group comprises eight distinct genotypes in the Danish collection. There were three groups with duplicate genotypes (Online Resource 4) that contain Pigeon fra Juellinge, ‘Pigeon Hvid Sommer’ and Almindelig Pigeon, respectively. However, colour sports certainly exist in at least some of these groups. In addition, there were five unique genotypes, ‘Pigeon Langeland Hvid’, ‘Pigeon Ildrød Dronningmølle’, ‘Pigeon Rød Vinter’, ‘Bødker Pigeon’ and ‘Gul Pigeon’.

Genetic structure

Globally, the Danish cultivars do not represent a distinct cluster within the European germplasm. Previous report for Swedish cultivars has also been reported to be genetically intermixed with cultivars from other European countries but generally genetic distinct from Finnish cultivars (Garkava-Gustavsson et al. 2013). Finnish cultivars that are adapted to cold climate with a short season may be more closely related to Russian cultivars, even though Urrestarazu et al. (2016) reported no clear genetic distinction between Swedish and Finnish cultivars. We conclude that Danish and Swedish cultivars comprise many unique genotypes adapted to local climate conditions, even though a somewhat shared gene pool with cultivars from other European countries exists.

On the basis of genetic clustering (Fig. 3a, b), we find no genetic structure in the studied material. We have shown a putative clustering of five groups which are in accordance with the average number of clusters found in other studies evaluating population structure of apple germplasm (e.g. Lassois et al. 2016; Urrestarazu et al. 2016) where much higher delta K values were found (delta K > 100) (Liang et al. 2015; Urrestarazu et al. 2012) compared to present study (delta K < 1.5). A high number of accessions in admix is reflected in the high number of genotypes (79%) without strong affinity to a specific group and is probably a major reason to the lack of genetic structure. This is also reflected on the PCA plot (Fig. 3b). The Danish apple collection consists of cooking and dessert apples including many intermediate types and lacks distinct apple types such as bitter cider apples which might have given distinct clustering. The ensemble of cultivars in the Danish collection descents from a narrow geographical region composed of many islands relatively far from the distribution centre of the cultivated apple. It probably differs from most of the studies performed at European level previously outlined, which represents larger geographical regions on the European continent where a more dynamic exchange of genotypes from various has been possible. However, our results are consistent with studies performed on other cultivar ensembles from the Nordic Region where the lack of genetic clustering likewise has been reported in Swedish and Finish cultivars (Garkava-Gustavsson et al. 2008; Garkava-Gustavsson et al. 2013).

The M. sieversii accessions are to a large extent genetically intermixed with M. domestica (Fig. 3b), which is in agreement with the previous assumptions that the species is an ancestor of M. domestica (Juniper and Mabberley 2006; Velasco et al. 2010). This is not in accordance with previously published STRUCTURE analysis performed at the Malus species level (Cornille et al. 2012). Our results, on the other hand, support a genetic intermix of M. sylvestris with the cultivars (Fig. 3b). Even though the studied assemble of M. sylvestris represents eight Danish populations, they were too few (nine accessions) to be genetically representative for these populations and therefore not sufficient for drawing conclusions. Previous report of genetic distinction between Danish M. sylvestris populations and M. domestica cultivars of Danish origin is concluded on the basis of a larger selection of M. sylvestris (Larsen et al. 2006). The accession Malus 43, which was collected from a wild M. sylvestris population, is most likely a hybrid with between M. sylvestris and the cultivar ‘Pederstrup’ (Larsen et al. 2016b). This suggests gene introgression from M. domestica cultivars into M. sylvestris populations, as previously reported for apples (Coart et al. 2006; Cornille et al. 2013), and questions whether genetic introgression exists in the M. sylvestris accessions included here.

Diploids and polyploids

The triploid cultivars pose a special problem, since they are expected to share one third of their alleles with one parent and two thirds with the other parent, which could be a triploid itself or a diploid contributing an unreduced gamete. Allelic frequencies should be taken into account so that rare alleles are weighted more than common alleles. Unfortunately, most software does not include analysis of polyploids, hindering the identification of parent-offspring relations. This could be one reason why we found a relatively low proportion of parent-offspring relations among triploids. In addition, not all possible parental genotypes were included in the analysis. Another possible explanation is that triploids produce fewer offspring compared to diploids because they usually produce a much lower proportion of viable pollen (Crane and Lawrence 1930).

Unreduced diploid gametes can either be homozygotic or heterozygotic in the SSR loci, depending on the exact mechanism leading to the unreduced status. In Citrus, most unreduced gametes were the consequence of second division restitution, which results in homozygotic SSR loci for those close to the centromeres but may result in heterozygotic SSR loci due to recombination between the centromere and the locus (Cuenca et al. 2015). The predominant mechanism leading to unreduced gametes in apple is not known, however.

Offspring populations from a triploid × diploid apple cross have been show to segregate into aneuploids, diploids and polyploids, and among the polyploids, both triploids and tetraploids have been observed (Bergström 1938; Druart 2000). The material studied here included several progenies from triploids, and consequently, aneuploids with a chromosome number close to 51 are likely to occur among the triploids (Chagné et al. 2015).

The triploid cultivar Gravensteiner has been cultivated in Denmark for more than three centuries and has been frequently planted for decades. To our surprise, we did not find any parent-offspring relations for this cultivar. In his pomology, Bredsted (1893) suggests the cultivars Filippa and Arreskov to be seedlings of Gravensteiner, which is clearly not the case according to the present results. To our knowledge, it is not known whether the cultivar is able to produce viable offspring (Kobel 1927) although it has been reported that a low percentage of its pollen are able to germinate (Florin 1926; Kvaale 1926).

New techniques for accessing genetic resources

Genetic diversity has been studied in a number of apple collections (e.g. Lassois et al. 2016; Urrestarazu et al. 2016), primarily using SSR markers from the core set described by Evans et al. (2007). This core set provides sufficient resolution power for cultivar identification, for revealing duplicates and for identifying parent-offspring associations, giving an overall picture of the genetic diversity within a collection. Single-nucleotide polymorphism (SNP)-based techniques such as SNP arrays (Bianco et al. 2016; Bianco et al. 2014; Chagné et al. 2012) or genotyping by sequencing (Gardner et al. 2014) have recently been developed. They are useful for QTL mapping by high-resolution linkage mapping or GWAS where thousands of markers are needed (Ingvarsson and Street 2011). Such high-throughput SNP genotyping techniques will complement and perhaps replace the use of SSR markers to characterize gene bank resources and could be used to confirm cultivar parentages found here. Comparing the two approaches, SSR markers are probably more precise for predicting parent-offspring associations, whereas the other similar SSR markers can yield information about ploidy. Furthermore, to fully exploit the genetic diversity present in collections, high-throughput phenotyping techniques may be useful. Such approaches can facilitate an increased awareness and accessibility of germplasm resources and allow identification of genes underlying important traits through GWAS.

Future perspectives for local apples

There is increasing awareness of the value of old, local cultivars. Attention has been focussed on characteristics and qualities of specific cultivars for processed products (apple juice, fruit wine, etc.), but there has also been focus on the historical origin of local apples. In this light, we need more knowledge about cultivars in germplasm collections. A previous study of Danish cultivars has demonstrated the large diversity in more than 50 chemical aroma compounds in apple juice samples (Varming et al. 2013) and showed that the collection contain traits that are desired in local apple production and modern breeding programmes.

In summary, the present genotypic data add to previous phenotypic data showing that the Danish apple gene bank is a valuable resource for important quality parameters. These parameters are rooted in a diverse gene pool and have been developed over the last two centuries, during which cultivars were imported from abroad. The great majority of cultivars came from European countries south of Denmark and a substantial number from England. The continuous import of new cultivars created the basis for a considerable diversity that was facilitated by countless, random bee-facilitated pollinations. This has created the current gene pool, which has been adapted to local climate conditions and is now a valuable and available resource of genes of interest for future breeding programmes.

Acknowledgements

A generous grant from Foreningen PlanDanmark and a PhD scholarship from the Department of Plant and Environmental Sciences, University of Copenhagen made this work possible. We are thankful to Jacob Weiner for valuable comments on the manuscript, to Charles-Eric Durel for a fruitful discussion and to Stefan Morberg for skillful technical assistance on flow cytometry. We thank Prima Plant (Sabro, Denmark), Tuse Næs Gårdmosteri (Holbæk, Denmark), Karen Syberg and Louise De Bang for providing plant material.

Data archiving statement

A full list of accession names, numbers and origin are given in Online Resource 1.

Supplementary material

11295_2016_1087_MOESM1_ESM.pdf (75 kb)
Online Resource 1List of accessions and accession numbers. Accessions originate from the gene bank collection, “Pometet” (Taastrup, Copenhagen Region), University of Copenhagen, unless otherwise specified. (PDF 75 kb)
11295_2016_1087_MOESM2_ESM.pdf (10 kb)
Online Resource 2STRUCTURE analysis for 344 M. domestica cultivars. The graph shows delta K vs. K for K = 2–12, tested over 20 runs. (PDF 10 kb)
11295_2016_1087_MOESM3_ESM.pdf (60 kb)
Online Resource 3List of ploidy level for Malus domestica cultivars and M. sieversii, determined by flow cytometry. (PDF 60 kb)
11295_2016_1087_MOESM4_ESM.pdf (37 kb)
Online Resource 4Each horizontal row contains accessions with identical genotype profile of 15 SSR-loci. The duplicate genotypes are divided into four categories: previously (beige) and not previously reported synonyms (blue), subclones such as colour sports from an original genotype (red) and accessions from the private nursery collection Assens (green). (PDF 36 kb)
11295_2016_1087_MOESM5_ESM.pdf (57 kb)
Online Resource 5Network of first degree relationships. Arrows point from parent to offspring. Information given in each box: cultivar name, accession number, ploidy level, S-RNase alleles and approximate geographical origin and year of origin. (PDF 57 kb)

Copyright information

© Springer-Verlag Berlin Heidelberg 2017

Authors and Affiliations

  • Bjarne Larsen
    • 1
  • Torben Bo Toldam-Andersen
    • 1
  • Carsten Pedersen
    • 1
  • Marian Ørgaard
    • 1
  1. 1.Department of Plant and Environmental SciencesUniversity of CopenhagenFrederiksberg CDenmark

Personalised recommendations