Autochthonous Austrian Varieties of Prunus avium L. Represent a Regional Gene Pool, Assessed Using SSR and AFLP Markers

Sweet cherry production faces new challenges that necessitate the exploitation of genetic resources such as varietal collections and landraces in breeding programs. A harmonized approach to characterization is key for an optimal utilization of germplasm in breeding. This study reports the genotyping of 63 sweet cherry accessions using a harmonized set of 11 simple sequence repeat (SSR) markers optimized in two multiplexed PCR reactions. Thirty-eight distinct allelic profiles were identified. The set of SSR markers chosen proved highly informative in these germplasm; an average of 6.3 alleles per locus, a PIC value of 0.59 and above-average expected and observed heterozygosity levels were detected. Additionally, 223 amplified fragment length polymorphism (AFLP) markers derived from eight selective primer combinations were employed to further differentiate 17 closely related accessions, confirming the SSR analysis. Genetic relationships between internationally known old cultivars were revealed: SSR fingerprints of “Schneiders Späte Knorpelkirsche” and “Germersdorfer” were found to be identical to those of the standard cultivar “Noire de Meched”, among others, whereas four accessions known as “Hedelfinger Riesenkirsche” and four known as “Große Schwarze Knorpelkirsche” showed allelic differences at various loci. The genetic diversity of locally-grown cultivars worldwide might be currently underestimated. Several autochthonous Austrian sweet cherry germplasm accessions were genotyped for the first time and their genetic relationships analyzed and discussed. Interestingly, seven Austrian sweet cherry landraces were shown to be clearly genetically separated from international and modern varieties, indicating that Austrian germplasm could include valuable genetic resources for future breeding efforts.


Introduction
Autochthonous varieties and their wild relatives are valuable genetic resources for the breeding and development of any domesticated crop [1][2][3][4]. Global climate change is a key driver for some of the challenges that sweet cherry production must address, such as double fruits [5] or fruit cracking [6] stem. The availability of well characterized genetic resources from different climatic zones and growing regions will continue to gain importance, especially since r modern sweet cherry breeding programs have focused on few selected genotypes as parents causing a severe genetic bottleneck [2]. Future breeding programs aiming to support sustainable sweet cherry cultivation will need adequate and diverse germplasm to answer to these challenges. The need for a harmonized approach to genetic characterization of national and international germplasm pools to maximize their usefulness as resources for breeding and research has been highlighted in various

Plant Material
Plant material for SSR-analysis was collected at two locations. (1) Sample Set STB: Stoob, Burgenland, AT (n = 29) representing old landraces and probably regional selections of known cultivars in the traditional high-stem meadow orchard growing system; and (2) Sample Set BOK: University of Natural Resources and Life Sciences, Vienna (BOKU) germplasm collection, Vienna, AT (n = 23) including Austrian landraces and some modern varieties. Additionally, ECPGR reference genotypes from East Malling Research, United Kingdom (GB) (n = 5) and United States of America (US) reference genotypes (n = 6) were included in Sample Set IS. These reference genotypes were used to standardize the SSR data as described by [7]. The individual samples were given a code (A01-A63) and named according to their phenotype-based identification of a known variety or cultivar, respectively, or using the accession name in the germplasm collection. Varieties that could not be reliably identified were referred to by their tree numbers (TN). In this study, the term "accession" is used to address the individual plant in the collection or sample set, respectively. Tree numbers identify the individual tree in the collection (Table 1).
AFLP analysis was conducted to further investigate the genetic identity and relationships of certain accessions showing identical SSR fingerprints. With AFLP analysis it is possible to detect polymorphisms between samples, covering the whole genome without prior knowledge of the DNA sequence [32]. Therefore, it is considered a suitable technique to differentiate between very closely related individuals of the same species [18,28]. We tested a subset of the BOKU germplasm accessions against the ECPGR standards "Noire de Meched" and "Noble" as well as landraces from two Austrian sweet cherry growing regions.
Plant material for the AFLP-analysis was collected at three locations: (1) Leithaberg, Burgenland (n = 4) (2) Scharten, Upper-Austria (n = 5) and (3) BOKU germplasm collection (n = 8), Vienna. ECPGR reference genotypes "Noire de Meched" and "Noble" were included in the analysis ( Table 2). Samples from Stoob were not included in this analysis. For simplification of the graphs, samples are coded by their TN instead of variety names.

DNA Extraction
For SSR analysis, genomic DNA was extracted from fresh leaves or winter buds with the DNeasy Plant Mini Kit (QIAGEN, Chatsworth, CA, USA) and diluted to a concentration of 2.5-4 ng/µL. For the reference accessions, lyophilized leaf-samples were extracted and diluted accordingly. The genomic DNA for the US reference accessions was kindly provided by Dr Amy Iezzoni's team in Michigan State University.
For AFLP analysis, DNA was extracted from lyophilized leaf-samples according to [33] with minor modifications.

SSR Analysis
Twelve primer pairs recommended by the ECPGR [7] and suitable for multiplexing were combined in two multiplex (MP) PCRs depending on the fragment sizes of amplified PCR products: MP1: large (173-261 bp) and small (98-187 bp), MP2: medium (120-208 bp). Primers were labeled with fluorescent dyes (Table 3).   LG4 [35] Primer concentrations in PCR reactions were optimized to produce similar intensity of chromatogram peaks to facilitate scoring. The final volume of PCR reactions was 13 µL containing 5-8 ng genomic DNA, 2× Type-it Multiplex PCR Master Mix (QIAGEN, Hilden, Germany), 1.25-5 nmol of primer, the exact quantity depending on each primer (Table 1).
PCR was started with a denaturing step at 95 • C for 5 min, followed by 10 circles of touchdown-PCR: 95 • C for 30 s, 60 • C for 1.5 min (−1 • C per cycle) and 72 • C for 30 s, followed by 18 cycles: 95 • C for 30 s, 50 • C for 1.5 min and 72 • C for 30 s, with a final elongation step of 60 • C for 30 min.
The amplicon sizes were measured against a LIZ 500 standard with an ABI 3130 Genetic Analyzer (Applied Biosystems, Waltham, MA, USA). Subsequent scoring was done with GENESCAN ® and GENOTYPER ® (Applied Biosystems, Waltham, MA, USA). Scored fragment sizes were harmonized with those of the reference genotypes using the allele sizes published by [7].

AFLP-Analysis
AFLP analysis was performed following the protocol by Vos et al. with the following modifications: Genomic DNA (0.3 µg) was incubated with 3.6 U Tru1I and 45 U EcoRI in 25 µL Tango-Buffer (Fermentas, St. Leon-Rot, Germany) for 1 hour at 37 • C, followed by two hours at 65 • C and 15 min at 85 • C. 5 µL of the adapter-ligation-solution as described by Vos et al. [40] were added and incubated overnight at 20 • C followed by a 1min-step at 65 • C to inactivate the T4 Ligase. For preamplification the template was diluted 1:5 and 2.25 µL were incubated in a 15 µL-PCR reaction with Taq DNA polymerase (5 U/µL) and 10× Taq buffer (Fermentas, St. Leon-Rot, Germany), 1.5 pM of each AFLP-primer without selective extensions, 3 mM MgCl 2 , 0.2 mM of dNTPs. The pre-amplified template was diluted 1:10 and 4 µL were incubated in a 10µL-PCR reaction with 0.5 pM of each of the selective primers. For this final selective amplification, 18 different primer combinations were tested on nine varieties. The eight most promising combinations were selected for the analysis: ATC/ATC, ACC/ATC, ACC/CAG, AGG/AGT, AGG/CAG, ATA/ATC, ATA/AGT, ATA/CAG. AFLP-fragments were run on a LI-COR (NEN Model 4300) analyzer (LI-COR Inc., Lincoln, NE, USA) and scored for presence (1) or absence (0) by hand with SagaTM (Version 3.0) (LI-COR inc., Lincoln, Nebraska USA). For all samples, except two, biological replicates (separate leaves, DNA extractions, digest, ligation, preamplification and selective amplification) for estimation of clonal variation were done and run side by side. Additionally, a technical control was run for four samples (same DNA extraction, separate digest, etc.). Three types of negative controls, one from the start and one for each of the PCRs were included. Only clearly visible bands between 90 and 400 bp of length were scored.

Data Analysis and Statistics:
For SSR data and allele frequencies, the identification of unique alleles was done in GenAlEx version 6.5 [41,42]. polymorphism information content (PIC)-values were subsequently calculated in Excel according to 43 [43].
The calculation of the genotype association curve (Figure 1), frequency based diversity estimators, allelic richness, observed heterozygosity, expected heterozygosity, the distance matrix, the dendrogram with Nei's distance as well as the genetic diversity indices for the clusters resulting from Discriminant Analysis of Principal Components (DAPC) were done in the statistics environment R version 4.0.3 [44] using packages poppr version 2.3.0 [45,46] and adegenet version 2.1.3 [47,48]. The DAPC was done in R, package adegenet version 2.1.3 [47] for SSR -data. In contrast to the Principal Component Analysis (PCA) or Principal Coordinate Analysis (PCoA), this analysis doesn't focus on the global genetic variation resp. diversity of the dataset, but instead optimizes the discriminant functions which show differences between groups, minimizing variation within clusters [49].
For AFLP-scores the PCoA was done in GenAlEx [41,42]. This analysis displays the overall genetic diversity present in the dataset.

Method Evaluation of SSR-Analysis
Eleven of twelve SSR markers gave unambiguous results in the multiplex approach, which could be standardized. Scores of BPPCT034 [39] could not be standardized, because the allele lengths of the reference accessions were found to be ambiguous. Data for this marker were excluded prior to analysis. The remaining 11 markers were amplified in two multiplexed reactions, resulting in a fast and easy-to-use fingerprinting system. The calculated genotype association curve ( Figure 1) showed that the number of loci obtained with these markers was sufficient to cover the genetic diversity present in the sample set. Other studies used similar numbers of makers for genetic fingerprints in Prunus avium, obtaining reliable results [7,19].
The number of alleles per locus ranged from 3 to 9, with an average of 6.3; PIC values ranged between 0.22 (EMPa017) and 0.78 (BPPCT037), with an average of 0.59. The observed and expected heterozygosity ranged between 0.06 resp. 0.23 (EMPA017) and 0.89 resp. 0.81 (BPPCT037) depending on the SSR marker (Table 4). The genotype association curve ( Figure 1) reveals that with five loci sampled already 100% of the Multilocus Genotypes (MLGs) across the sampling set tested could be detected. We conclude that the method using 11 SSR markers has adequate power to discriminate between the unique individuals in our dataset and adding more markers would not reveal many additional genotypes.
The DAPC was done in R, package adegenet version 2.1.3 [47] for SSR -data. In contrast to the Principal Component Analysis (PCA) or Principal Coordinate Analysis (PCoA), this analysis doesn't focus on the global genetic variation resp. diversity of the dataset, but instead optimizes the discriminant functions which show differences between groups, minimizing variation within clusters [49].
For AFLP-scores the PCoA was done in GenAlEx [41,42]. This analysis displays the overall genetic diversity present in the dataset.

Method Evaluation of SSR-Analysis
Eleven of twelve SSR markers gave unambiguous results in the multiplex approach, which could be standardized. Scores of BPPCT034 [39] could not be standardized, because the allele lengths of the reference accessions were found to be ambiguous. Data for this marker were excluded prior to analysis. The remaining 11 markers were amplified in two multiplexed reactions, resulting in a fast and easy-to-use fingerprinting system. The calculated genotype association curve ( Figure 1) showed that the number of loci obtained with these markers was sufficient to cover the genetic diversity present in the sample set.
Other studies used similar numbers of makers for genetic fingerprints in Prunus avium, obtaining reliable results [7,19].
The number of alleles per locus ranged from 3 to 9, with an average of 6.3; PIC values ranged between 0.22 (EMPa017) and 0.78 (BPPCT037), with an average of 0.59. The observed and expected heterozygosity ranged between 0.06 resp. 0.23 (EMPA017) and 0.89 resp. 0.81 (BPPCT037) depending on the SSR marker (Table 4). Eight different genotypes (four genotypes from Stoob, two from the BOKU collection and two international standards) show unique alleles in up to three markers. The best markers in terms of detecting unique alleles in different genotypes were CPPCT022 and EMPaS10, with unique alleles amplified in six and five genotypes, respectively.

Variety Identification and Verification of Trueness to Type
The genotypes obtained by SSR fingerprinting allowed us to confirm or reject the morphological identification of landraces and germplasm accessions and revealed the occurrence of homonyms and synonyms in otherwise well-known varieties. Some phenotyping, grafting or labelling errors were also detected (e.g. "Burlat VG" which is not identical with the cultivar "Bigarreau Burlat"; "Lambert" and "Stella Spur" which surprisingly showed the same fingerprint). Four trees in Stoob were phenotyped as cultivar "Große Schwarze Knorpelkirsche"; only two showed the same genotype and all of them differed compared to the reference accession in the BOKU germplasm collection. Two of three trees phenotyped as "Hedelfinger Riesenkirsche" and collected in Stoob had the same fingerprint, but all of them differed from the two accessions of the same name in the BOKU germplasm collection, which also turned out to show two distinct genetic profiles. Some clonal variants were also identified such as the samples phenotyped as landrace "Butterkirsche" and the two types of "Kritzendorfer Einsiedekirsche". Six of the sampled Austrian landraces showed to be identical to several germplasm accessions and standard cultivars from different origins. This group of seemingly identical genotypes includes the accessions "Germersdorfer", "Schneiders Späte Knorpelkirsche", the ECPGR reference cultivar "Noire de Meched" and the US-reference "Schneiders", as well as traditional Austrian cultivars like "Melker Riesenkirsche" and "Horitschoner Herzkirsche". For simplification purposes this group of accessions will be mentioned as the "Schneiders-Group" throughout this study.
An AFLP-analysis including further samples from two other sampling sites was conducted to confirm if these cultivars are true clones i.e., synonyms (see below).

Genetic Diversity
The Discriminant Analysis of Principal Components (DAPC, R, adegenet) was used to identify groups of genetically related individuals in all three sample sets. This analysis has been developed and is suitable for clonal or partly clonal populations, since it does not rely on Hardy-Weinberg or linkage disequilibria [50]. The varieties group into three defined separate clusters ( Figure 2); each cluster contains members of the three sample sets: BOK = BOKU germplasm collection, IS = International Standard, STB = Stoob. Part of the samples showed to be admixed according to the estimated probability of group membership, which is depicted in the membership probability plot (Figure 3).
has been developed and is suitable for clonal or partly clonal populations, since it does not rely on Hardy-Weinberg or linkage disequilibria [50]. The varieties group into three defined separate clusters ( Figure 2); each cluster contains members of the three sample sets: BOK = BOKU germplasm collection, IS = International Standard, STB = Stoob. Part of the samples showed to be admixed according to the estimated probability of group membership, which is depicted in the membership probability plot (Figure 3).
Cluster 1 consists of nine multi-locus-genotypes (MLGs) and contains landraces from Burgenland ("Butterkirsche", "Sämling von Sauerbrunn" and "Donnerskircher Blaukirsche") as well as the rootstocks "NY54" and "F12/1" ( Table 5, Table 1). Only one of these samples ("Kritzendorfer Einsiedekirsche") shows to be admixed with cluster 3. The other two clusters share more admixed genotypes (Figure 3). In cluster 2, 18 samples or MLGs are found; some of the modern cultivars, all samples of the two widely distributed cultivars "Große Schwarze Knorpelkirsche" and "Hedelfinger Riesenkirsche" and some landraces denominated by tree numbers (TN). Cluster 3 comprises of the rest of the modern cultivars and the "Schneiders-Group" which was included in the analyses represented by one sample of this genotype: TN39 (Table 5). Cluster 1 consists of nine multi-locus-genotypes (MLGs) and contains landraces from Burgenland ("Butterkirsche", "Sämling von Sauerbrunn" and "Donnerskircher Blaukirsche") as well as the rootstocks "NY54" and "F12/1" (Tables 1 and 5). Only one of these samples ("Kritzendorfer Einsiedekirsche") shows to be admixed with cluster 3. The other two clusters share more admixed genotypes (Figure 3). In cluster 2, 18 samples or MLGs are found; some of the modern cultivars, all samples of the two widely distributed cultivars "Große Schwarze Knorpelkirsche" and "Hedelfinger Riesenkirsche" and some landraces denominated by tree numbers (TN). Cluster 3 comprises of the rest of the modern cultivars and the "Schneiders-Group" which was included in the analyses represented by one sample of this genotype: TN39 (Table 5).      According to Simpson's diversity index (lambda, Table 6), DAPC-cluster 2, which includes many accessions of the germplasm-collection, is the most diverse, with 0.944, followed by cluster 3, with 0.90. Cluster 1 is the least diverse, with lambda 0.889; this cluster mostly consists of the landraces from Burgenland, which, unsurprisingly, were shown to be very closely related to each other. The expected heterozygosity (Hexp) of all three clusters were very similar, with values ranging from 0.595 (cluster 2) to 0.63 (cluster 1).
The dendrogram (Figure 4) was calculated based on Nei's distance, bootstrapping (10,000) and UPGMA (unweighted pair group with arithmetic means). Variety A11 TN46 had to be excluded for this calculation, because the algorithm based on Nei's distance cannot process missing data, and the SSR marker CPSCT038 did not amplify any PCR products for this genotype (see also Supplement-File S1). The dendrogram shows four main clusters of sweet cherry varieties. Though based on the same data, the clusters are composed differently compared to the DAPC (Figure 2, Table 3), as they are calculated using a different algorithm.
DAPC-cluster 1 (red) is represented on the upper end missing "Kritzenorfer Einsiedekirsche", which groups with the varieties of the cluster below. NY45 groups together with Goodnestone Black (uppermost end), and F12/1 appears separated from all other varieties on the bottom end of the dendrogram. The other two DAPC clusters appear to be admixed, whereas notably all genotypes of "Hedelfinger Riesenkirsche" appear in one cluster and all genotypes of "Große Schwarze Knorpelkirsche" in another. A separate cluster consists of "Hybrid 222", i.e., "Burlat VG". Unlike DAPC analysis, the dendrogram shows the rootstock "F12/1" as more genetically distant from the rest of the varieties (Figure 4). This might be a result of the clustering algorithm UPGMA assuming the occurrence of a hierarchical structure between the individuals and rooting this structure in one sample, which might not be an accurate assumption for our data set. The low bootstrap-values on the left-side nodes indicate that the separation of the main clusters as shown in the dendrogram is not very well supported by the data. This is probably due to the limited amount of processed data, combined with the fact that the sweet cherry varieties in this study are generally very closely related.
An AFLP analysis with eight selective primer combinations and 17 samples of four different origins resulted in 223 markers, of which 64 (28.7%) were polymorphic.
The error-rate was moderate, with 1.36-1.79% of difference in band-occurrence between technical replicates. Biological replicates differed up to 6.73%. The samples of "Noble" could clearly be separated from the other samples with PCoA ( Figure 5). The percentage of variance explained by the first two axes was 32.7% and 44.17% by the three main coordinates. This low power of explanation of variance is probably due to the low number of samples and the generally low variation between the tested genotypes. Therefore, these results should be interpreted with caution.

Discussion
The optimized multiplex-PCR approach to SSR genotyping was broadly successful; 11 of the 12 SSR-markers were easily scored and standardized against published data for five reference accessions (data not shown). This gives us confidence in the quality of the data generated which should prove straightforward to compare with similarly standardized data for other germplasm collections in the future The number of markers was comparable to that of similar studies [3,4,23,27,51] and showed to be sufficient to reveal the genetic diversity of MLGs expected to be present in the samples analyzed. Furthermore, with an average of 6.3 alleles per locus and PIC-value of 0.59 as well as the above-average expected and observed heterozygosity-levels, it is clear that the set of SSR-markers chosen was highly informative in the sweet cherry varieties in this study. The method of charac-

Discussion
The optimized multiplex-PCR approach to SSR genotyping was broadly successful; 11 of the 12 SSR-markers were easily scored and standardized against published data for five reference accessions (data not shown). This gives us confidence in the quality of the data generated which should prove straightforward to compare with similarly standardized data for other germplasm collections in the future The number of markers was comparable to that of similar studies [3,4,23,27,51] and showed to be sufficient to reveal the genetic diversity of MLGs expected to be present in the samples analyzed. Furthermore, with an average of 6.3 alleles per locus and PIC-value of 0.59 as well as the above-average expected and observed heterozygosity-levels, it is clear that the set of SSR-markers chosen was highly informative in the sweet cherry varieties in this study. The method of characterization proved effective and provided solid results that can be reliably harmonized with future studies.
The chosen primer combination was found to be effective for differentiating between most of the Austrian sweet cherry landraces tested. Regarding the accurate identification of duplicates (i.e. synonyms) and homonyms (i.e. genetically heterogenous groups phenotyped as the same cultivar), both could be detected by SSR-analysis, providing essential information for the optimal management of the germplasm collection and future breeding approaches.
In DAPC cluster 2 (Figure 2), there are two heterogenous groups. In five samples, all phenotyped as "Große Schwarze Knorpelkirsche", four distinct genotypes could be detected. This is understandable; this cultivar dates back to the 16th century and has been one of the most important in central Europe [52].
Four of five samples phenotyped as "Hedelfinger Riesenkirsche" showed differences in allele sizes. The cultivar "Hedelfinger Riesenkirsche" originates from Hedelfingen in Germany where it was selected from seedlings and taken to Hohenheim (Germany) around 1850. Thereafter, this cultivar has been widely distributed by tree nurseries [8] and is found in many places around the world nowadays often referred to as "Hedelfingen".
"Große Schwarze Knorpelkirsche" as well as "Hedelfinger Riesenkirsche" were assigned to the same DAPC cluster; both have been referred to as "population cultivar", indicating that the cultivar consists of several phenotypes [53]. This could be due to mixed vegetative and sexual reproduction combined with selection for a certain fruit morphologywhich makes these cultivars relatively easy to phenotype-over the past decades. Another possibility is clonal variation, i.e., the vegetative propagation of sport mutants.
On the other hand, three not previously identified genotypes of distinct phenotypes found in Stoob showed the same fingerprint (TN96, TN120, TN142; Figure 4). In that case, it has to be considered that the discriminating power of the limited set of SSR markers might not be adequate for these genotypes.
The identification of synonyms or duplicates helps to reduce the number of accessions and therefore the running costs in conservation efforts such as germplasm collections [54]. On the other hand, mutations and thus clonal variation occur, especially in long-lived tree species [55,56]. Certain clones potentially harbor desirable superior traits such as yield [27], climatic adaptation or tolerance to pests and diseases, and thus the identification of true clonal variation within germplasm collections is essential.
A group of accessions showing the same genetic SSR fingerprint includes presumably different late-ripening, heart-shaped, dark-red, firm cultivars with considerable fruit size. Surprisingly, very well-known, Europe-wide distributed cultivars like "Germersdorfer" and "Schneiders Späte Knorpelkirsche" as well as typical Austrian cultivars like "Melker Riesenkirsche" and "Horitschoner Herzkirsche" (named after Melk and Horitschon-two towns in the eastern part of Austria) are found in this group, along with the ECPGR reference cultivar "Noire de Meched" and the US-standard "Schneiders". Although these findings could be due to limited SSR-marker resolution, suspicions that some of the cultivars could be clones and their names therefore synonyms have been raised before. Braun-Lüllemann and Bannier [52] found that in Germany the cultivar denominated as "Germersdorfer" in the past decades is morphologically identical to the cultivar known as "Schneiders Späte Knorpelkirsche".
This finding is especially intriguing since this latter cultivar was found to be comparable in fruit size to widely distributed modern cultivars like "Regina" or "Kordia", and thus represents a potential candidate for breeding. Considering that maximizing fruit size still is one of the most important objectives in sweet cherry breeding [57], it is essential to find out whether the studied varieties are in fact clones. The verification of this hypothesis would probably reduce the number of suitable high-fruit-sized parent-candidates for future breeding programs. Differences in the phenotype due to clonal or epigenetic variation would have to be studied in appropriate trials, to select for the best clones. Moreover, for a cost-effective rationalization of germplasm collections, it is important to know if those Austrian cultivars are in fact all duplicates.
To our knowledge, until now there has been no such study. To gather additional evidence on the correct genetic identity of the above-mentioned varieties which have identical SSR profiles, an AFLP analysis was conducted.
We compared the ECPGR standard "Noire de Meched", accessions of the Austrian germplasm collection "Germersdorfer", "Schneiders Späte Knorpelkirsche", "Melker Riesenkirsche" and "Horitschoner Herzkirsche", as well as samples from two different sites consisting of several landraces identified as one of the just mentioned cultivars. Three genetically and phenotypically different accessions were included for comparison.
Based on the results of the AFLP analysis, the tested varieties in general show low genetic variation; only about 29% of markers were shown to be polymorphic. This is in agreement with reported polymorphic rates of 21% for AFLP markers in sweet cherry [19]. Technical error rates are as high as 1.79%. For biological replicates (same tree, different branch), differences of up to 6.73% were recorded. These differences could be explained by clonal variation resp. sport mutation, and therefore it is also probable that morphologically identical or very similar varieties are clones of one and the same widespread cultivar. As expected, samples of the cultivar "Noble" could clearly be separated from the rest by PCoA ( Figure 5). Nevertheless, samples of the other two very different phenotypes "F12/1" and "Rainkirsche" appear close to or inside the cluster of "Schneiders-Group" cultivars. This could be due to the proportion of technical errors combined with the comparatively small genetic distance between Prunus avium varieties, which puts more weight on such technical errors. To sum up: based on the results shown in this study genetic differences among the tested varieties exist. These genotypes might represent valuable resources for future breeding efforts, if they have superior traits, e.g., disease resistance or superior fruit size.
The genetic diversity of the Austrian landraces evaluated was subsequently compared to that of international standard cultivars based on the results of the DAPC and calculated diversity indices. The DAPC sorted the samples into three clusters. In cluster 2, all samples of "Hedelfinger Riesenkirsche" and "Große Schwarze Knorpelkirsche" group together with "Goodnestone Black", "Napoleon", "Stella Spur", "Lambert", "Lapins", "Ulster", "Tavriczskai" and "Sarga Dragan". "Chelan" groups with "Noble", "Early Burlat", "Burlat", "Jaboulay" and "Früheste der Mark" in DAPC cluster 3, whereas DAPC cluster 1 comprised only varieties from Burgenland, i.e., landraces not mentioned in the available literature. These autochthonous varieties seemed clearly distinct from the other groups and may therefore constitute a valuable germplasm for breeding as members of a regional gene pool. Details on valuable phenotypic and physicochemical characteristics of these varieties such as unique taste, low susceptibility to rain-induced cracking or high content of polyphenols in the fruit have been recorded in prior studies [11,12,58]. The landraces are probably admixed with the wild cherry population of this specific region. Interestingly, rootstocks "NY45" and "F12/1" are also assigned to this cluster.

Conclusions
A successful method of fast and easy-to-use multiplex SSR analysis for international harmonization of sweet cherry accessions was presented. The investigated collection of autochthonous Austrian sweet cherry landraces is highly diverse and could constitute a valuable germplasm for future breeding programs, since Austrian landraces were shown to represent a regional gene pool. They exhibit interesting traits that might be valuable for breeders (Table 1) and are most probably adapted to the local climate and environmental factors, since they comprise ecotypes that have been cultivated in the same region for decades. Furthermore, they might harbor certain traits like tolerance to fruit cracking or tolerances to diseases and pests, which has to be evaluated in further studies.
Concerning the various genotypes of "Große Schwarze Knorpelkirsche" and 2Hedelfinger Riesenkirsche", marker assisted selection (MAS), field trials and cultivar evaluations should be conducted to identify the most valuable of the clones for breeding purposes.
It would be interesting to compare the genetic diversity of Austrian landraces with those from the French collection described by Mariette et al. [2]. Does it comprise a different gene pool? What is the influence of the wild cherry population in Austria and how is this gene pool different compared to the French wild cherries?
Phenotype-based surveys on sweet cherry diversity have been conducted for some Austrian regions [10][11][12][13], and Austrian landraces were shown to bear valuable characteristics such as a high content of polyphenols in the fruit [58]. While important first steps have recently been taken to preserve and protect these landraces in the future, considerable gaps of knowledge still need to be filled to effectively preserve the Austrian sweet cherry diversity.
Part of these gaps could effectively be filled by genetic evaluation, as has been shown in this study. Homonyms, synonyms and labeling errors were detected. The genetic data help to evaluate the genetic diversity, identity and trueness to type of Austrian sweet cherry accessions and thus serves as an important and valuable tool for the management of Austrian germplasm collections.

Data Availability Statement:
The data from SSR-and AFLP-analysis is provided as supplemental files (Supplement S1, Supplement S2) to this manuscript.