Molecular characterization of Spanish Prunus avium plus trees

Aim of the study: The Breeding Program of wild cherry ( Prunus avium ) developed by Lourizán Forest Research Center (NW aims for the creation of the Main Breeding Population, that is formed by a large number of plus trees and for obtaining an Elite Population generated from controlled crosses of a number of plus trees selected by, at least, one trait of economic importance. The aim of this study was to genotype 131 accessions of Prunus avium plus trees, included in the breeding program. Material and methods: Prunus avium plus trees were genotyped with nine microsatellites. Several genetic parameters were calculated. Genetic data were analyzed with STRUCTURE and the genetic distance between plus trees were calculated. Main results : A total of 122 multilocus genotypes were detected. Several accessions with the same genotype were identified, which could be due to clonality or to labelling errors. The nine microsatellites are useful for identifying individuals because the combined probability of identity was low (PI = 5.19 × 10 –9 ). Bayesian methods detected two genetic clusters in the sampled plus trees. Research highlights : The unique genotypes identified in this work are suitable for being included in the elite breeding population for economic traits.


Introduction
Wild cherry (Prunus avium) is a noble hardwood species of economic importance which is being used in clonal plantations in order to produce high quality timber. In 1998, the Lourizán Forest Reseach Center, located in Galicia, in the Northwest of Spain, started a phenotypic selection of P. avium plus trees. These trees were propagated by grafting to stablish clonal seed orchards. In addition, a clonal selection of other 30 individuals was developed to study their rooting ability and to select clones to be used in commercial plantations. Presently, the Innovation and Forest Tree Breeding Plan of the Galician region, aims to obtain long-term genetic gains in several traits of interest for wood production. For P. avium, the plan defines two different populations. On one hand, the Main Breeding Population contains plus trees from the North and Northwest coast and Central Spain, phenotypically selected on the basis of their value for timber produc-tion. On the other hand, the Elite Population has the best individuals selected by at least one of the following breeding traits: growth, resistance to Blumeriella jaapii, straightness and propagation f itness. These individuals are being crossed in a half-diallel mating design to obtain high multi-trait genetic gain.
The main objectives of this study were to genotype, with nine nuclear loci, 131 accessions of P. avium plus trees that are being used in the Main Breeding Population and in the Elite Population and to detect a clonal and genetic structure. Several analyses were performed to know the necessary number of loci to distinguish all the multilocus genotypes and its discrimination power for individual fingerprinting.

Plant material and laboratory methods
The samples were collected from the seed orchards of Areas and Sergude, the clonal trial of Bos, the germplasm collection of Mantequera and from several micropropagules located in in vitro laboratory. Up to 173 samples were collected from 131 accessions of P. avium that were classif ied as plus trees in different field prospections in the North of Spain (Figure 1). Individuals were classified into five populations: Navarra-Basque country, Asturias-León, Eastern Galicia-Asturias, Atlantic Galicia and Central Spain. A total of 42 replicas were sampled from 38 accessions in order to verify the clonal fidelity.

Statistical analysis
The presence of null alleles was determined using Micro-Checker ver. 2.2.3 (Van Oosterhout et al., 2004). Loci with estimated null allele frequencies higher than 0.19 were excluded from further analysis because from this threshold, the underestimation of the expected heterozygosity due to null alleles is signif icant (Chapuis et al., 2008).Several standard measures were calculated with GENCLONE v2.0 (Arnaud-Haond and Belkhir, 2007) in order to detect the presence of potential clones in the defined populations: the number of samples (N), the number of multilocus genotypes (MLGs), the number of repeated MLGs (MLG r ), the number of unique MLGs within each population (MLG l ) (Ellstrand and Roose, 1987), the number of multilocus lineages (MLLs), and the modified index of genotypic richness (R) (Dorken and Eckert, 2001). The probability that two individuals with the same MLG were originated from different sexual reproductive events, Psex, was also calculated.
Once the number of MLGs was established, the probability of identity (PI) and the combined Molecular characterization of Spanish Prunus avium plus trees 121 probability of identity was calculated with SPAGeDi v1.4 (Hardy and Vekemans, 2002) in order to know whether the set of loci are useful to estimate the real number of multilocus genotypes (MLGs). The probability of identity (PI) represents the average probability of a match for any genotype. The number of alleles (n a ), the effective number of alleles (n e ), and the number of privative alleles were calculated with SPAGeDi v1.4 (Hardy and Vekemans, 2002). STRUCTURE version 2.3.4 software (Pritchard et al., 2000;Hubisz et al., 2009) was used to assign the defined MLGs to different genetic clusters. Two independent analyses were performed with and without LOCPRIOR model. In both cases, a model without admixture and with correlated frequencies was assumed. A burn-in period of 50,000 iterations followed by 100,000 Markov Chain Monte Carlo (MCMC) iterations was used for K values from 1 to 10. Ten independent runs were tested for each K value. The L(K) non parametric test (Pritchard et al., 2000) and ΔK approach (Evanno et al., 2005) were used to identify the most likely number of clusters (K).
The number of shared alleles was used to calculate the genetic distance between clones. The NEIGHBOR package of PHYLIP software (Felsenstein, 1989) was used to construct a dendrogram following the UPGMA method and it was displayed with FIGTREE 1.4.0 (http://tree.bio.ed.ac.uk/software/figtree/).

Flow cytometry
Prunus avium leaves of trees suspected to be triploid were sent to Centro Nacional de Biotecnología (CNB) in Madrid, Spain. A flow cytometry analysis was performed using Citomics FC500 (Beckman Coulter, Fullerton, California, USA). The intensity of the fluorescence of the cell nuclei of the putative triploid was compared with LU23 clone that was used as diploid control.

Results and discussion
The analysis with MICRO-CHECKER detected deviations from Hardy-Weinberg equilibrium for EMPA015 in Navarra-Basque country population due to the presence of null alleles. However, EMPA015 was maintained in further analysis because the estimated null allele frequency was 0.08.
The amplification with nine loci showed one or two alleles in all samples except for the accessions SA4 and SA12 (Supporting information 1).
Although the samples SA4 and SA12, were coded as different accessions, they belong to the same plus tree because they display the same genotype, the same three alleles in loci EMPaS01, EMPaS10 and EMPaS12 and the same doubled alleles in the remaining six loci. Flow cytometry revealed that the fluorescence intensity of SA12 was nearly 1.5 times higher than the control LU23 and confirms that SA12 and, therefore SA4, are triploids. Natural triploids of P. avium are not very common in nature. Nevertheless, one individual was detected in Germany (von Schelhorn, 1947), another one in Belgium (De Cuyper et al., 2005) and eleven in France (Serres-Giardi et al., 2010). In general, they are phenotypically superior trees. Triploids have significantly better height and circumference growth than diploids and therefore, they can be suitable for wood production (Serres-Giardi et al., 2010).
A total of 122 MLGs were detected out of 131 accessions (Table 1). A clonal lineage (MLL) was also found; PO34 located in Areas was different from the micropropagated PO34 in one allele in locus EMPA015 (Supporting information 1) probably due to somatic mutation or scoring errors. There were 5 MLGs with at least two different accessions in the same population and one MLG, LU40-PO36, with the accessions generated from different populations. The psex values of the accessions LU40-PO36, LOURIZÁN 1-LU47, NA7-NA12, NA8-NA9-NA10-NA11, NA22-NA26 and PO33-PO34 suggest that they were produced by asexual reproduction because the probability of the genotype to be present once or more times as the result of different reproductive events are quite low (Supporting information 1). In addition, a labelling error can explain the MLG identity of LU40 and PO36, coming from different populations.
The nine loci used in this study are useful for f ingerprinting because the probability of another random and unrelated individual with the same genotype is very low (Combined probability of identity = 5.19 × 10 -9 ) (Supporting information 2).
The genetic analysis of the 131 accesions of Prunus avium revealed that all microsatellite loci were polymorphic. A total of 84 alleles were detected. The number of alleles ranged from 17 in EMPA015 to 4 in EMPA014 (Supporting information 2). Nevertheless, the effective number of alleles decreased significantly Supporting information 1. Geographic coordinates and allelic profiles of 131 accesions of Prunus avium plus trees genotyped with nine loci. Note that SA4 and SA12 are different accessions that belong to the same plus tree and PO34/IV is different from PO34 in locus EMPA015. Several accessions share the same geographic coordinates because it represents the parish or the council where the accessions are located. Accessions in bold belong to the same MLG and show the Psex value which means that only a few alleles have high allelic frequencies in each locus. The number of privative alleles ranged from 12 in Navarra-Basque country population to 1 in Asturias-León population. No privative alleles were detected in Atlantic Galicia and Central Spain populations (Data not shown). The allelic range of the loci overlaps or partially overlaps with the results of other works (Guarino et al., 2009;Tanceva Crmaric et al., 2011;De Rogatis et al., 2013). The minimum combination of loci to distinguish all MLGs is EMPA015, EMPaS12 and EMPaS06. However, there still were 58 MLGs that were different When LOCPRIOR model was disabled, STRUCTURE did not find any population structure. The LOCPRIOR model takes into account the sampling locations and it is suitable for detecting a genetic structure when the signal is too weak (Hubisz et al., 2009). The least negative value of LnP(K) with the lower standard deviation was at K = 2. All runs at K = 2 displayed the same result. From K = 3, LnP(K) decreased and the standard deviation in the different K values was very high. In addition, the Evanno method also detected two main clusters. The 122 MLGs could be classified into two genetic clusters (Figure 2). The f irst group 126 J. Fernandez-Cruz et al. / Forest Systems (2014)   comprised individuals sampled in Navarra, the Basque country, Asturias, León and Central Spain. The second genetic pool contained individuals from Atlantic and Eastern Galicia-Asturias. No clear assignation is achieved for individuals AS0402/3, AS1, AS19 (Asturias-León), OU42-PO28-PO29 (Atlantic Galicia). Notably, the triploid SA4 = SA12 is different from the other individuals of the Central Spain population and it was assigned to the Asturias-Galicia group. The two genetic clusters detected in this study can be explained by population fragmentation and reproductive isolation. The information provided by STRUCTURE can be useful for the selection of stands for the production of seeds for conservation plantations. A dendrogram constructed according to nine nSSR data (Supporting information 3) of 122 MLGs and 1 MLL divided them into two main clusters. The accession PV2 and PV21 have the most different genotypes. The highlighted individuals are being used in the elite population for crosses. They are genetically quite different among them and it is expected that the crosses between these parental lines will produce individuals with a superior expression of economic traits.