AFLP-Based Analysis of Variation and Population Structure in Mutagenesis Induced Faba Bean

: Genetic diversity enrichment is urgently necessary to develop climate-resilient faba bean cultivars. The present study aimed to measure the enrichment of genetic diversity and changes in the population structure of faba bean, following induced mutagenesis. 120 samples, including 116 M 2 mutant plants, generated by exposing the ILB4347 accession to four mutagen treatments (25 and 50 Gray gamma radiation and 0.01%, and 0.05% diethyl sulfate) and four reference genotypes were characterized using 11 ampliﬁed fragment length polymorphism (AFLP) primer combinations. The AFLP markers generated 1687 polymorphic alleles, including 756 alleles (45%) that were detected infrequently (f ≤ 0.1). The total allele count of the mutant plants ranged from 117 to 545. We observed a wide range of banding patterns and counts among the mutant plants, showing the high genetic diversity induced by mutation. Mutations also changed the population structure, by altering 31.78% of the total membership coe ﬃ cient (Q). Although mutations changed the population structure, Nei’s genetic distance showed that the mutant population remained closely related to its control parent. This is the ﬁrst report examining genetic diversity and population changes in faba bean mutant populations and, thus, could facilitate the application of induced mutagenesis during faba bean breeding.


Introduction
Modern plant breeding and agronomic techniques have increased global food production by an average of 32 million metric tons per year, between 1961 to 2007. However, the Food and Agriculture Organization (FAO) has predicted more than an additional 70% of food will be necessary to feed nine billion people in 2050, which will require an increase in food production of 44 million metric tons per year, annual incremental increases of 37% [1,2]. Future crop production is likely to face other challenges, such as the impacts of climate change. Therefore, crops that are resilient to changes in the environment, including biotic and abiotic stresses, are urgently necessary to adequately feed the growing population.
The development of climate-resilient crops depends on the availability of genetic diversity. However, many studies have reported that crop improvements have been accompanied by reductions in genetic diversity [3][4][5][6]. Liu et al. [7] explained that domestication, the oldest breeding method recognized by human history, has reduced genetic diversity by changing the pattern of gene expression. Another breeding method that has been associated with reduced crop genetic diversity is intraspecific

AFLP
The AFLP plant mapping protocol from Applied Bio-systems (ABI, Waltham, MA, USA) was adopted for this study, with some modifications. Samples containing 500 ng genomic DNA were added to 0.2 mL tubes, 5.5 µL of the digestion-ligation mix was added to each sample, and the samples were incubated overnight, at room temperature. TE0.1 (90 mL) was added to each tube and stored at 4 • C. Pre-selective amplification was performed using primers that matched the adapter sequence and included one additional 'selective' base (EcoRI + G and MseI + C). In a 96-well PCR plate, the PCR mix was formulated with 10 µL 2× PCR Master Mix (Promega, Madison, WI, USA), 0.5 µL each of EcoRI-G and MseI-C, 2 µL diluted digestion-ligation product, and 7 µL double-distilled water, for a final volume of 20 µL. The PCR parameters were as follows: 72 • C for 2 min, 30 cycles of 94 • C for 30 s, 56 • C for 30 s, 72 • C for 2 min, followed by 60 • C for 10 min, before holding at 4 • C until use. The product was diluted 10 times with TE buffer.
Selective amplification uses the diluted, pre-selective amplification product as the new template. The primers for this reaction had the same sequences as the pre-selective primers, with two and three additional selective bases in EcoRI and MseI primers, respectively. The EcoRI +2 primers were labeled with a fluorescent dye (6-FAM) that could be detected with the ABI 3130xl Genetic Analyzer. In a 96-well PCR plate, the PCR mix was formulated, containing 10 µL 2× PCR Master Mix (Promega, Madison, WI, USA), 1 µL each of the EcoRI +2 and MseI +3 primers, 2 µL diluted pre-selective amplification product, and 6 µL double-distilled water, for a final volume of 20 µL. The PCR parameters were as follows: initial denaturation at 94 • C for 2 min, followed by 10 cycles of 94 • C for 30 s, 66 • C for 30 s, decreasing by 1 • C per cycle, and 72 • C for 2 min, followed by 20 cycles of 94 • C for 30 s, 56 • C for 30 s, and 72 • C for 2 min, and then 60 • C for 30 min, followed by holding at 4 • C until further use.

Running the AFLP Gel on the ABI 3130xl Genetic Analyzer
Two microliters of the selective amplification product were mixed with 0.15 µL GeneScan 500 LIZ size standard (Applied Biosystems P/N 4322682) and 14.85 µL Hi-Di Formamide (Applied Biosystems Diversity 2020, 12, 303 4 of 14 P/N 4311320). The mixture was denatured at 94 • C for 5 min and directly placed on ice, prior to loading into the 16-capillary system of the Applied Biosystems 3130xl Genetic Analyzer. A 36 cm capillary array (Applied Biosystems P/N 4315931) and 3130 POP-7 polymers (Applied Biosystems P/N 4352759) were used.

AFLP-Fragment Analysis
Eleven out of 36 selective combination primers were selected to assess the genetic diversity among the mutant populations. The primer combinations were selected based on gel electrophoresis observations. The AFLP adapter/primer names and sequences used are shown in Supplementary  Table S1. AFLP fragment analysis was performed using GeneMapper Analysis Software v3.7 (ABI), and the data were assembled in binary format, with 1 representing the presence of an allele and 0 representing the absence of an allele. The threshold for allele calling was set to 100 relative fluorescence units (rfu). Peaks at 100 rfu or higher were assigned a value of 1 and those with lower rfu values were assigned value of 0. Fragment analysis was performed for allele sizes in the range of 100-500 bp, as recommended by Paris et al. [30]. To avoid biased parameter estimates, the loci or alleles generated from all primers in all samples were checked and corrected, based on Lynch and Milligan's [31] critical value of (1+ where N = the number of individuals sampled. Based on this estimation, loci with fewer than four total fragments across all samples were deleted from the analysis. The total number of alleles, total number of bands, average bands per sample, and average bands per allele for each primer were calculated to evaluate the primer combinations. The discrimination power (DP) was calculated by dividing the total number of polymorphic alleles amplified by each primer by the total number of polymorphic alleles obtained for all primers [32].

Genetic Diversity Parameters and Analysis of Molecular Variance (AMOVA)
The binary matrix displaying the presence/absence of each allele was analyzed further, using the GenAlEx 6.503 complement for MS Excel [33]. The total number of different alleles (Na), the total number of effective alleles (Ne), Shannon's information index (I), the expected heterozygosity (He), the percentage of polymorphic loci (%P), and the number of private alleles per population were calculated using this approach. The PAST (Paleontological Statistics) v. 3.20 program [34] was used to evaluate pairwise Jaccard genetic similarity, to assess the genetic diversity among the 116 mutant plants and the four reference genotypes. An analysis of molecular variance (AMOVA) was performed, using GenAlEx 6.503 [33], with 9999 permutations. AMOVA was performed by partitioning the genetic variations between five populations which were the reference genotypes and four mutant populations, based on the mutation treatment applied (25 or 50 Gy of gamma radiation or 0.01% or 0.05% of DES).

Population Structure
The mutant population structure was evaluated using the STRUCTURE 2.3.4 program [35][36][37][38]. The STRUCTURE analysis was performed using an admixture model, with 10,000 burn-in periods and 20,000 Monte Carlo Markov Chain (MCMC) simulations. Ten independent runs were performed for each K value, ranging from 1 to 10, to evaluate the best K. The K value was determined using Evanno's ∆K, based on Structure Harvester software [39]. Nei's genetic distance among the mutant populations was calculated, using GenAlEx [33]. Cluster analysis was performed based on the unweighted pair group method, with arithmetic averaging (UPGMA) and principal component analysis (PCA), using the Euclidean similarity index, in the PAST (Paleontological Statistics) v. 3.20 program [34].

AFLP Polymorphism
Eleven AFLP primer combinations were used to assess the genetic diversity among the mutant plants. The characteristics of 11 AFLP primer combinations are presented in Table 1. A total of 1687

AFLP Band Abundance
The frequencies of the 1687 alleles detected from among the mutant plants and reference genotypes are presented in Figure 1. The allelic frequency ranged from 0.01 to 0.99, with an average frequency of 0.18. A total of 756 alleles, or 45% of the total alleles detected, were detected at a low frequency (f ≤ 0.1), whereas only 15 alleles, or 1%, were detected at a high frequency (f > 0.9), which indicated that allelic abundance and richness were generated by mutations. ILB4347, the control parent, generated 195 alleles across 11 AFLP primer combinations, whereas the total number of alleles detected in mutant plants ranged from 117 to 545, which revealed that mutations could both generate new alleles and reduce or lose alleles. The mean number of AFLP bands observed throughout the mutant plants was 327.79. Whereas the mean number of AFLP bands observed on reference genotypes was 334.50. The standard deviation of mutant plants and reference genotypes were 108.22 and 161.21, respectively. A wide range of standard deviation on mutant plants and reference genotypes indicated high variation of banding pattern among individuals. Moreover, increasing the level of mutagen treatments increased the mean performance, standard deviation, and range of AFLP bands observed as shown in Table 2.

Genetic Diversity Parameters and Analysis of Molecular Variance (AMOVA)
The genetic diversity parameter analysis is presented in Table 3. The comparison analysis was divided into three groups. The first group compared the genetic diversity between the mutant population and the reference genotypes. The second group compared the genetic diversity among the reference genotypes and two mutant populations, divided based on the mutagen used (gamma radiation or DES). The third group compared the genetic diversity among the reference genotypes and four mutant populations, divided based on the levels of mutagen used (25 or 50 Gy gamma radiation or 0.01% or 0.05% DES). The genetic diversity parameters revealed that the mutant population generated from ILB4347 was more diverse than the reference genotypes, as indicated by the higher values observed for all the genetic diversity parameters. The number of alleles (Na) and the number of effective alleles (Ne) of mutants and the reference genotypes were 2.000 and 1.008, respectively. The mean values of Shannon's information index (I) were 0.283 and 0.234, respectively. The expected heterozygosity (He) values were 0.164 and 0.150, and the percentages of polymorphic loci (%P) were 100% and 49.38%, respectively. A total of 820 private alleles were associated with ILB4347 mutant plants, whereas the reference genotypes did not have any private alleles.

Genetic Diversity Parameters and Analysis of Molecular Variance (AMOVA)
The genetic diversity parameter analysis is presented in Table 3. The comparison analysis was divided into three groups. The first group compared the genetic diversity between the mutant population and the reference genotypes. The second group compared the genetic diversity among the reference genotypes and two mutant populations, divided based on the mutagen used (gamma radiation or DES). The third group compared the genetic diversity among the reference genotypes and four mutant populations, divided based on the levels of mutagen used (25 or 50 Gy gamma radiation or 0.01% or 0.05% DES). The genetic diversity parameters revealed that the mutant population generated from ILB4347 was more diverse than the reference genotypes, as indicated by the higher values observed for all the genetic diversity parameters. The number of alleles (Na) and the number of effective alleles (Ne) of mutants and the reference genotypes were 2.000 and 1.008, respectively. The mean values of Shannon's information index (I) were 0.283 and 0.234, respectively. The expected heterozygosity (He) values were 0.164 and 0.150, and the percentages of polymorphic loci (%P) were 100% and 49.38%, respectively. A total of 820 private alleles were associated with ILB4347 mutant plants, whereas the reference genotypes did not have any private alleles. Two subpopulations were analyzed, according to the mutagen used, which revealed that the highest genetic diversity was identified in the mutant population induced by DES (Na = 1.697, Ne = 1.256, I = 0.284, He = 0.171). The lowest genetic diversity was recorded for the reference genotypes. The %P ranged from 49.38%, for the reference genotypes, to 99.84%, for ILB4347 mutations induced by gamma radiation. The number of private alleles ranged from 0 to 164. Based on the levels of mutagenic treatments, the highest and lowest genetic diversity values were recorded in the mutant population induced by 50 Gy gamma radiation and the mutant population induced by 25 Gy gamma radiation, respectively. The %P ranged from 49.38%, for the reference genotypes, to 93.06% for the mutant population induced by 25 Gy gamma radiation. The number of private alleles ranged from 0 to 30. These results also revealed that when higher doses or levels of mutagenic treatments were used, broader genetic diversity was generated, regardless of the type of mutagen used. The AMOVA result revealed that 89% of the variation was concentrated within populations, whereas the variation among populations contributed to the remaining 11% of the variation ( Table 4). The pairwise Jaccard's genetic similarity among the 116 mutant plants and the four reference genotypes ranged from 0.06 to 0.67, with an average value of 0.23 (Supplementary Table S2).

Population Structure
The population structure analysis indicated that the faba bean mutant population could be divided into five groups, based on the highest Evanno's ∆K value of K = 5 (Figure 2A). Figure 2B shows the results of the STRUCTURE analysis for the ILB4347 mutant populations. The group or cluster assignment was based on the proportion of admixture within a single mutant plant. In this study, a 70% probability value was used to group the mutant plants. Mutant plants with a probability value below 70% were considered to be intermixed. The grouping distribution of mutant plants, based on results from the STRUCTURE software analysis, is presented in Supplementary Table S3. Based on this grouping, none of the mutant plants were assigned to a group, except for group 1 (red).  Figure 3 shows that the red portion, which coincides with the genetic background, of all mutant plants was 68.22%, indicating that 31.78% of the population structure changed due to the mutation process. The mutant population induced by 0.05% DES displayed the largest change to the population structure, followed by the mutant population induced by 50 Gy gamma radiation, 0.01% DES, and 25 Gy gamma radiation, in that order. The color or group proportions affected by population change also differed among the mutant populations. In the mutant populations induced by 0.01% DES and 25 Gy gamma radiation, the green portion of the population structure greatly changed compared with the other colors. Whereas in the mutant population induced by 50 Gy gamma radiation, the pink and the blue portions of the population structure were affected more strongly. In the mutant population induced by 0.05% DES, the green and the pink portions showed greater population structure changes.

Genetic Relatedness
Four reference genotypes, representing three different regions, were selected to provide the genetic diversity baseline. A total of 867 polymorphic alleles from 11 AFLP primer combinations were recorded across these reference genotypes. Hassawi 3 had the highest total alleles count, with 515 alleles, or 59.4% of the total alleles, followed by Misr 3, with 426 alleles, Hassawi 2, with 202 alleles, and ILB4347, with 195 alleles. The wide range of the selected genetic diversity baseline is shown in the scatter biplot in Figure 4A. The two principal coordinates, based on Jaccard's similarity index,  Figure 3 shows that the red portion, which coincides with the genetic background, of all mutant plants was 68.22%, indicating that 31.78% of the population structure changed due to the mutation process. The mutant population induced by 0.05% DES displayed the largest change to the population structure, followed by the mutant population induced by 50 Gy gamma radiation, 0.01% DES, and 25 Gy gamma radiation, in that order. The color or group proportions affected by population change also differed among the mutant populations. In the mutant populations induced by 0.01% DES and 25 Gy gamma radiation, the green portion of the population structure greatly changed compared with the other colors. Whereas in the mutant population induced by 50 Gy gamma radiation, the pink and the blue portions of the population structure were affected more strongly. In the mutant population induced by 0.05% DES, the green and the pink portions showed greater population structure changes.  Figure 3 shows that the red portion, which coincides with the genetic background, of all mutant plants was 68.22%, indicating that 31.78% of the population structure changed due to the mutation process. The mutant population induced by 0.05% DES displayed the largest change to the population structure, followed by the mutant population induced by 50 Gy gamma radiation, 0.01% DES, and 25 Gy gamma radiation, in that order. The color or group proportions affected by population change also differed among the mutant populations. In the mutant populations induced by 0.01% DES and 25 Gy gamma radiation, the green portion of the population structure greatly changed compared with the other colors. Whereas in the mutant population induced by 50 Gy gamma radiation, the pink and the blue portions of the population structure were affected more strongly. In the mutant population induced by 0.05% DES, the green and the pink portions showed greater population structure changes.

Genetic Relatedness
Four reference genotypes, representing three different regions, were selected to provide the genetic diversity baseline. A total of 867 polymorphic alleles from 11 AFLP primer combinations were recorded across these reference genotypes. Hassawi 3 had the highest total alleles count, with 515 alleles, or 59.4% of the total alleles, followed by Misr 3, with 426 alleles, Hassawi 2, with 202 alleles, and ILB4347, with 195 alleles. The wide range of the selected genetic diversity baseline is shown in the scatter biplot in Figure 4A. The two principal coordinates, based on Jaccard's similarity index,

Genetic Relatedness
Four reference genotypes, representing three different regions, were selected to provide the genetic diversity baseline. A total of 867 polymorphic alleles from 11 AFLP primer combinations were recorded across these reference genotypes. Hassawi 3 had the highest total alleles count, with 515 alleles, or 59.4% of the total alleles, followed by Misr 3, with 426 alleles, Hassawi 2, with 202 alleles, and ILB4347, with 195 alleles. The wide range of the selected genetic diversity baseline is shown in the scatter Diversity 2020, 12, 303 9 of 14 biplot in Figure 4A. The two principal coordinates, based on Jaccard's similarity index, distributed four genotypes into three quarters, supported by the value of Jaccard's genetic similarity index, which ranged from 0.14 to 0. 26. distributed four genotypes into three quarters, supported by the value of Jaccard's genetic similarity index, which ranged from 0.14 to 0.26.
The matrix of Nei genetic distance and identity shows the relatedness among mutant population and the four reference genotypes (Supplementary Table 4). The values showed that the mutant population closely related with its parent genotypes, ILB 4347, followed by Hassawi 2, Misr 3, and Hassawi 3. It illustrated by principle component analysis (PCA) of Nei's genetic distance ( Figure 4A). Further investigation was conducted among four mutant populations and four reference genotypes based on Nei's genetic distance. The analysis showed that the mutant populations gathered in one cluster near to the ILB 4347 and Hassawi 2 and far from Misr 3 and Hassawi 3 ( Figure 4B).

Relationship between Molecular and Phenotypic Data
The previous study [17] was conducted to assess genetic diversity of M2 population at the morphological level based on 12 qualitative traits, including inflorescence, flower, seed, leaf, and stem traits. The summary of mutant variation on different populations is presented in Table 5. Table  5 showed that increasing the level of mutagen treatments increased the number of phenotypic variation per sample. The highest phenotypic variation was recorded in 50 Gy gamma radiationinduced mutant population with 2.76 variations per sample, followed by the mutant population induced by 25 Gy gamma radiation, 0.05%, and 0.01% DES with 1.87, 1.80, 1.27 variations per sample, respectively. The correlation between phenotypic and molecular variation showed positive and no significant correlation with r = 0.801.  The matrix of Nei genetic distance and identity shows the relatedness among mutant population and the four reference genotypes (Supplementary Table 4). The values showed that the mutant population closely related with its parent genotypes, ILB 4347, followed by Hassawi 2, Misr 3, and Hassawi 3. It illustrated by principle component analysis (PCA) of Nei's genetic distance ( Figure 4A). Further investigation was conducted among four mutant populations and four reference genotypes based on Nei's genetic distance. The analysis showed that the mutant populations gathered in one cluster near to the ILB 4347 and Hassawi 2 and far from Misr 3 and Hassawi 3 ( Figure 4B).

Relationship between Molecular and Phenotypic Data
The previous study [17] was conducted to assess genetic diversity of M 2 population at the morphological level based on 12 qualitative traits, including inflorescence, flower, seed, leaf, and stem traits. The summary of mutant variation on different populations is presented in Table 5. Table 5 showed that increasing the level of mutagen treatments increased the number of phenotypic variation per sample. The highest phenotypic variation was recorded in 50 Gy gamma radiation-induced mutant population with 2.76 variations per sample, followed by the mutant population induced by 25 Gy gamma radiation, 0.05%, and 0.01% DES with 1.87, 1.80, 1.27 variations per sample, respectively. The correlation between phenotypic and molecular variation showed positive and no significant correlation with r = 0.801.

Discussion
The present study was conducted to examine genetic diversity and population structure changes in faba bean M 2 populations. The study used 11 AFLP primer combinations to evaluate 116 M 2 mutant plants, representing 116 mutant families that were generated by four treatments (two kinds of mutagen and two kinds of intensity), using four reference genotypes as a genetic diversity baseline. AFLP was used due to the high reproducibility of this technique and the power to detect point mutations throughout the genome. The high reproducibility was demonstrated by the average number of polymorphism alleles detected, which generated 153.36 alleles. In total, the 11 AFLP primer combinations generated 1687 polymorphic alleles from 120 sample plants. The number of alleles obtained exceeded those described by previous studies examining genetic diversity in faba bean populations, such as the 302 alleles in 255 samples identified using SSR [25], the 104 alleles in 187 samples identified using nuclear SSR [40], the 657 SNPs identified in 45 samples [24], the 1036 alleles detected in 58 samples using SRAP [23], the 209 alleles detected in 802 samples using ISSR [22], the 71 alleles identified in 34 samples using ISSR [41], and the 289 alleles examined in 28 samples using RAPD [19].
A total of 39,362 bands (data points) were obtained, with an average of 23.33 bands per sample, per primer combination. These results returned more bands than those obtained using SRAP markers, during which 10,700 data points were obtained, with an average of 13.17 bands per sample per primer combination [23]. These results validate the power of the AFLP technique to assess genetic diversity, due to the high reproducibility of this technique compared with other types of molecular markers. The efficiency and effectiveness of AFLP markers for the evaluation of genetic diversity can be explained by the high polymorphism rate and polymorphism information content (PIC) values. All of the primer combinations utilized in this study exhibited 100% polymorphism, with an average PIC value of 0.99%.
Enriching genetic diversity, in terms of molecular-based diversity, simply increases the number of polymorphic bands or new alleles detected. Our study identified many new alleles within the mutant population, in which greater than 88% of the alleles detected were classified as new alleles compare with the control plant. Moreover, various banding patterns were also identified in every mutant plant. The total allele count in mutant plants ranged from 117 to 545, whereas 195 alleles were detected in the control plant, which indicated that induced mutagenesis resulted in random mutations that both generated new alleles and resulted in the loss of existing alleles. These results agree with the sequencing data for 1504 rice mutants, in which a wide range of mutations was observed across the mutant plants [42]. Moreover, genome-wide comparisons using resequencing was performed by Yang et al. [43], to compare four mutant rice plants induced by 250 Gy gamma radiation. The results showed that the individual mutants had different total numbers of mutations and proportions of mutation types. The banding pattern analysis also revealed that the higher doses of mutagens resulted in a larger range of banding patterns, which were detected as shown in Table 2. Comparisons among mutant populations, according to the treatment used, showed that increasing the dose or concentration of mutagen treatments also increased the mean performance, standard deviation, and range of AFLP bands observed, indicating that higher mutagen doses can result in larger changes in the DNA arrangement and structure.
Genetic diversity analysis comparing between the mutant populations and the reference genotypes revealed that the mutant populations were more diverse than the reference genotypes, due to the high AFLP band polymorphism rate found in the mutant populations. The mutant population induced by DES at selected doses showed higher genetic diversity than the mutant population induced by gamma radiation. Similar results have previously been reported, when Kurowska et al. [27] reported that N-methyl-N-nitrosourea (MNU) at doses of 0.5-1.5 mM was more efficient for inducing point mutations in barley than gamma radiation at doses of 150-210 Gy. The AMOVA model compared four mutant populations and the reference genotypes. The AMOVA showed that the maximum genetic variation was found within the populations. These results are in line with the results of other studies performed in faba beans [44,45]. A total of 89% of the variation was concentrated within populations, and only 11% of the variation was observed between the populations. The 11% variation between populations may be due to the different genetic makeup relative to the reference genotype or differences in the mutagen used to generate different mutation types.
The population structure analysis showed that a total of 62% of the mutant plants were characterized as intermixed. This high proportion of intermix far exceeds the population structure analysis of faba bean accessions, based on several geographical areas. Göl et al. [25] evaluated 255 accessions from 30 different countries, using SSR, and reported that only 3.9% of the accessions were intermixed. Previously, Wang et al. [22] evaluated 802 faba bean accessions, worldwide, using ISSR markers, and were able to separate Chinese accessions from non-Chinese accessions. They concluded that China was an independent center of diversity for faba beans, due to being reproductively isolated from the rest of the Asian, African, and European gene pools. Furthermore, population structure analyses between cultivated and wild relatives of faba beans, conducted by Oliveira et al. [40], showed grouping between the two populations, indicating a very limited amount of gene flow between cultivated and wild faba beans. Therefore, the high proportion of intermixed populations indicated by the population structure analysis in the present study demonstrated the effects of random mutations induced by gamma radiation and DES.
The Evanno test showed that the mutant population could be divided into five groups (K = 5). However, only one group had members with coefficient memberships greater than 0.70. The other mutant plants had coefficient memberships lower than 0.70, explained by the high mutation effects on individual mutant plants, which indicated that ILB4347 mutant population belonged to a single genetic background, which is represented by the red portion in Figure 2. The portions represented by the other colors of each plant genome in the population structure analysis could be attributed to the diversity generated by the different types of mutations or to the locations of mutations within the genome. Unassigned members to inferred groups, generated by the STRUCTURE program, were also identified in a population structure study examining Tunisian faba bean landraces, conducted by Babay et al. [12]. The authors tested 29 faba bean landraces and determined that these populations could be divided into three groups/subpopulations, as inferred by the STRUCTURE program. However, none of the landraces belonged to these subpopulations, due to high levels of cross-pollination and human-mediated seed flow. Altogether, mutation, cross-pollination, and gene flow due to human-mediated seed flow can influence population structures.
The coefficient membership means of the mutant populations suggested that higher doses of mutagenic treatments caused higher changes in the coefficient membership proportion. Mutant populations induced by 0.05% DES presented larger changes in the coefficient membership proportion, with a 40.59% alteration, compared with the 36.53% alteration observed in the 0.01% DES-induced mutant population. Whereas, the 50 Gy gamma radiation-induced mutant population modified 40.09% of the coefficient membership proportion, which was 15.07% higher than that of the 25 Gy gamma radiation-induced mutant population. Interestingly, the changes in the coefficient membership of these mutant populations were significantly and positively correlated (r = 0.953) with the value of Nei's unbiased genetic distance (Supplementary Table S5).
Genetic relatedness among the mutant populations and the four reference genotypes showed a close relationship between the mutant population and its control parent. The four reference genotypes represented three different regions. The PCA analysis separated the reference genotypes into three out of four quarters, indicating wide variations among the four reference genotypes. Cluster analysis, based on Nei's genetic distance, showed a distinct cluster, consisting of the four mutant populations. This mutant cluster was near ILB4347, which was the control parent, and far from the other reference genotypes, indicating that although induced mutagenesis changes the banding pattern of the individual mutant plant, it is not sufficient to completely alter the identity relative to the control parent.
The relationship between phenotypic and molecular variation showed that the higher level of mutagen treatment increased both phenotypic and molecular variations. However, there is no significant correlation between phenotypic and molecular data due to the different order of treatment that gave higher variation. The order of phenotypic variation was 50 Gy > 25 Gy > 0.05% DES > 0.01%, whereas the order of molecular variation was 50 Gy > 0.05% DES > 0.01% > 25 Gy. The difference can be explained due to the difference between the number of alleles screened and morphological traits evaluated. A total of 1687 alleles were screened in 116 mutant plants compare to 12 morphological traits evaluated. Moreover, any mutation at the DNA level did not express at the morphological level as mutation could be changed at non-coding DNA sequences.

Conclusions
The present study revealed that induced mutagenesis enriched genetic diversity and changed the population structure of faba beans. The abundance of new AFLP bands and the high number of infrequent alleles (f ≤ 0.1) that were observed in the mutant population reflects the high degree of genetic diversity induced by gamma radiation and DES. Higher doses of mutagens generated increased genetic diversity; thus, high doses of mutagenic agents could be further applied to faba bean mutation breeding. The structure of the mutant population changed with increased mutagen levels; however, a close genetic relationship continued to be observed between the mutant population and its control parent, relative to the other reference genotypes.
Supplementary Materials: The following are available online at http://www.mdpi.com/1424-2818/12/8/303/s1. Table S1. Adapter and primer sequence used for AFLP analysis, Table S2. The pairwise Jaccard's genetic similarity among the 116 mutant plants along with four reference genotypes, Table S3. The membership coefficient (Q) of individual mutant plant at K = 5, Table S4. The value of Nei's genetic distance and identity among mutant population and four reference genotypes, Table S5. The value of Nei's unbiased genetic distance among mutant populations and four reference genotypes.