4.3. Data Analysis

The frequencies of observed microsatellite alleles were measured using the GENALEX 6.41 software [27]. The polymorphism information content (PIC) was calculated as:

where P_i and P_j are the population frequency of the ith and jth allele [28] in MS Excel. Expected (H_e) and observed (H_o) heterozygosity values of each microsatellite and frequency of rare (less than 5% of the accessions) and unique (less than 1%) alleles were calculated using the GENALEX 6.41 software [27].

Dice coefficient measured in PAST 3.16 software [29] was used for genetic similarity estimation, and to visualize genetic relationships among the studied accessions by an UPGMA (unweighted pair group method with arithmetic mean) clustering method, using MEGA 7 [30]. PAST 3.16 software [29] was used to carry out principal coordinates analysis (PCoA).

Genetic structure analysis of the collection was performed using Structure v.2.3.4 software [31]. From 1 to 15 clusters (K) with 30 replicates for each K were tested. The number of possible clusters was found as the result of 200,000 iterations of Markov chain Monte Carlo, taking into account genetic admixture and correlated allele frequencies. The first 20,000 generations were eliminated (burn-in). The optimal number of clusters was determined as recommended by Evanno et al. [32] using the online program Structure Harvester [33].

Differentiation of accessions depending on geographic origin (North American and NBG (other countries were represented by a small number of accessions)) and fruit traits (clingstone–non-melting, clingstone–melting, semifreestone–melting, and freestone–melting), was investigated with Analysis of Molecular Variance (AMOVA) in the GENALEX 6.41 software [27]. The threshold for statistical significance was determined by running 999 permutations. Pairwise F_ST estimates for groups of accessions were calculated using GENALEX 6.41 software [27].