Genetic Diversity among Selected Medicago sativa Cultivars Using Inter-Retrotransposon-Amplified Polymorphism, Chloroplast DNA Barcodes and Morpho-Agronomic Trait Analyses

Alfalfa (Medicago sativa L.) is a major forage crop of family Fabaceae and is frequently cultivated in Egypt. The present study is concerned with the genetic discrimination of fifteen alfalfa cultivars from three different countries (Egypt, Australia, and USA) using two molecular approaches: inter-retrotransposon-amplified polymorphism (IRAP) markers and two chloroplast DNA barcodes matK and the trnH in addition to the analysis of fifteen morpho-agronomic traits. The genetic relatedness, based on analysis of IRAP marker polymorphism and produced using eleven primers by clustering via principal component analysis (PCA) and multivariate heatmap biostatistical methods differentiated the two Egyptian cultivars EGY1-Ismailia1 and EGY2-Nubaria1 from the three Australian and seven American cultivars, with some distinction of the cv. USA6-SW9720 and cv. AUS4-SuperFast. The results were also supported by the sequence analysis of the matK and the trnH genes on the genetic relatedness between eight cultivars. Moreover, it might be suggested that breeding lines from M. sativa cultivars may provide novel insights and a better understanding of the domestication of M. sativa genetic diversity. The classification of the eight cultivars, as revealed by morpho-agronomic traits, confirmed the close genetic relationship between the two Egyptian cultivars and indicated some resemblance between them and the AUS2-Siri Nafa, whereas the two American cultivars, USA1-Super supreme and USA4-Cuf101, were clearly isolated from a cluster of other three cultivars USA7-SW9628, USA8-Magna901, and USA9-Perfect. The results are useful sources of genetic information for future breeding programs in crop development and open new possibilities of producing M. sativa lines harboring high forage quality, productivity, and resistance to biotic and abiotic stresses.


Introduction
Alfalfa (Medicago sativa L.), known as alfalfa and lucerne, is a major important cultivated forage legume crop that was originated in the Caucasus region, and plays a significant role in agricultural sustainability [1]. It is one of the worldwide planted legume crops for forage sources cause of its high indications of genetic divergence present in populations and assists in comparative investigations of population diversity [28]. In recent years, DNA barcoding have become a useful tool for investigating and monitoring biodiversity and in phylogeny reconstruction and plant evolution studies [29] although alignment of the trnH-psbA spacer region needs careful attention [30]. Combined universal chloroplast and nuclear DNA sequence targets was applied to barcode the major Mediterranean leguminous crops [31]. The combination of universal chloroplast and nuclear DNA sequence targets was applied to barcode the major Mediterranean leguminous crops [32].
Genetic diversity of M. sativa has been frequently characterized using morphological traits [33], sometimes in combination with molecular markers [16]. Warburton and Smith (1993) suggested that there were at least six regional germplasm groups of non-dormant alfalfa among the North African, Arabian, and Indian germplasm [34]. Canonical discrimination functions showed that 75% of total morphological variability among accessions of 18 non-dormant alfalfa was strongly influenced by leaflet shape, stipule shape, and the peduncle, that is, the petiole length ratio [16]. The development of improved alfalfa cultivars may have altered some of the traits of ancestral populations that were associated with inherent yield and could have contributed to the relatively narrow genetic base that is now available for breeding [35][36][37].
The objective of the current study is to use the IRAP markers to differentiate 15 selected cultivars of M. sativa. In addition, analysis of fifteen morpho-agronomic traits and DNA barcoding, using the two chloroplast genes matK and trnH are used to assess the genetic relatedness and ancestry of the selected alfalfa cultivars.

Plant Material
Fifteen M. sativa cultivars were kindly assessed by ICARDA, research program of crop genetics, currently at the Agricultural Genetic Engineering Research Institute (AGERI), Agricultural Research Center (ARC), Giza, Egypt. The samples include two cultivars from Egypt, three from Australia and ten from the United States of America as listed in Table 1. Table 1. Origin, codes, and names of the alfalfa cultivars as recorded by ICARDA and the Genebank deposited accession numbers for matK and trnH genes for eight cultivars of Medicago sativa.

No.
Origin Code # Cultivar Name matK Genebank Accession High molecular weight genomic DNA was isolated from 50-100 mg ground and freeze-dried seed meals of M. sativa cultivars based on the procedures of Murray and Thompson (1980) using DNeasy Plant Mini Kit (QIAGEN, Hilden, Germany). The quantity and purity of extracted DNA were assessed spectrophotometrically using the ND-1000 system (Nano-Drop Technologies, Thermo Fisher Scientific Inc., Waltham, MA, USA). DNA manipulation and analysis procedures in terms of quantification and purity analysis were performed according to the Molecular Cloning Laboratory Manual [38].

IRAP Primer and IRAP Marker PCR Amplification
Eleven IRAP primers of those designed by Kalendar and Schulman (2014) were used ( Table 2). The primers were synthesized by HVD Vertriebs-Ges. m.b.H. (Vienna, Austria) in 10 nmol stock concentration, delivered in a lyophilized form, rehydrated in sterile water to reach the 100 µM final concentration, and finally stored at −20 • C until used. IRAP-PCR amplification was carried out in 25 µL reaction volume containing 1× PCR master mix buffer supplemented by 1.5 mM MgCl 2 , 25 µM primer, and 75 ng genomic DNA. The amplification process was performed in a Perkin-Elmer/GeneAmp ® PCR System 9700 (PE Applied Biosystems, Waltham, MA, USA) programmed to initial denaturation cycle for 3 min at 94 • C followed by 35 cycles, with each cycle consisting of a denaturation at 94 • C for 1 min, annealing at 55 • C for 1 min, and elongation at 72 • C for 1.5 min. The primer extension segment was extended to 7 min at 72 • C in the final cycle. Visualization and documentation of the PCR products were executed by using Bio-Rad ChemiDoc TM MP gel documentation and image system (Cat. no. 1708280). Underlined values denote the minimum and maximum value of each estimated parameter.

DNA Barcoding of mat K and trn H Plastid Genes
For the matK and trnH barcoding, the forward and reverse primers sequences given in Table 3 have been used. The PCR amplification of the two gens was performed at initial denaturation of 5 min at 94 • C followed by 40 cycles, each consists denaturation at 94 • C for 30 s, annealing step at 45 • C for 30 s, and elongation at 72 • C for 30 s. The primer extension segment was extended for 7 min at 72 • C in the final cycle. PCR specific products were subsequently electrophoresed on 1.5% w/v agarose, stained with EtBr (final concentration 100 µM/L, Sigma-Aldrich ® , St. Louis, MO, USA) in 1× TBE buffer and visualized as described for the IRAP PCR products. Fractionated PCR amplicons of matK and trnH amplified fragments were recovered from agarose gel using QIAquik ® PCR PURIFICATION KIT (Qiagen Inc., Venlo, The Netherlands, Cat. no. 28106) according to the manufacturer's instructions. Selected and purified amplicons were then cloned into pGEM cloning vector prior to sequencing. Table 3. Primer codes and sequences for barcoding the matK and trnH genes.

Primer Code
Sequence Product Size Reference trnH-F 5 -GTTATGCATGAACGTAATGCTC-3 700 bp [31] trnH-R 5 -CGCGCATGGTGGATTCACAATCC-3 [32] matK-472F 5 -CCCRTYCATCTGGAAATCTTGGTTC-3 700 bp The DNA sequence was determined by automated DNA sequencing reactions performed using ABI PRISM Big Dye Terminator Cycle Sequencing Ready Reaction Kit (PE Applied Biosystems, Waltham, MA, USA) in conjunction with ABI PRISM (3100 Genetic Analyzer, Macrogene DNA sequencing services, Seoul, Rep. of Korea). DNA sequencing was executed using the Gene Amp 2400 Thermal Cycler. The reaction was performed in 20 µL including 7 µL of terminator ready reaction mix, 0.5 µg of plasmid DNA, and 3.2 pmol M13 universal forward primer. The sequence cycling was programmed to 95 • C for 10 s, 55 • C for 5 s, and 65 • C for 4 min for 30 cycles with thermal ramping according to manufacturer's recommendations. The sequence of nucleotides was detected automatically by the electrophoresis of the sequencing reaction product on 3100 Genetic Analyzer. The result was provided as fluorimetric scans and revealed nucleotide sequence was assembled using Gblocks software version 0.91b [39,40].
The relationship among the eight M. sativa cultivars, used for phylogenetic relatedness using the DNA barcoding of matK and trnH plastid genes, were also investigated using variation in 15 morpho-agronomic traits as described by Moawed [41]. A list of these traits and their states and description in the eight cultivars are given in Table 5. These traits were regarded as two state characters and scored as a binary matrix and analyzed by the Principle Coordinates Analysis (PCA) to calculate the relatedness of the examined cultivars and determine the impact of the morpho-agronomic traits on the plotting of cultivars in a PCA biplot.

Data Analysis
The clear distinct reproducible bands generated as IRAP markers were scored as (1) for presence or (0) for absence. The capacity of IRAP primers to discriminate between examined genotypes were analyzed by means of calculating the polymorphic information content (PIC), Resolving power (Rp), Marker Index (MI), and effective multiplex ratio (EMR) values. PIC value for each primer, representing the reflection of the degree of detecting polymorphism, was calculated according to Ghislain et al. [42]. Rp was calculated following the formula of Gilbert et al. [43]. Rp = Σ IB, where IB represents band informativeness which was calculated using: where P is the frequency of accessions that harbor bands. EMR and MI values were estimated according to Powell et al. [44]. The IRAP markers matrix were used to calculate the coefficient of genetic similarity matrix and construct a distance tree displaying relationships among cultivars using the unweighted pair group method with arithmetic mean (UPGMA) using PAST, ver. 4.02 software [45]. In addition, principal component analysis (PCA) scatter diagram was constructed based on Dice coefficient genetic similarity matrix by using PAST, ver. 4.02 software [45]. The multivariate analysis was performed through constructing a Heatmap matrix using the module of R software.
Computational analysis of matK and trnH DNA barcoding sequences was carried out. Analysis and assembling of matK/trnH gene sequences for every cultivar was performed via BioEdit software version 7.2.5. Latter sequences were then compared using BLAST function (www.ncbi.nlm.nih.gov/ bastn) with all accessible sequences in database. matK/trnH gene sequences were submitted to GenBank (BankIt). Alignment of multiple nucleotide sequences were performed by online ClustalW2 (https://www.ebi.ac.uk/Tools/msa/clustalw2/) and double checked by using MEGAX. Gaps positions were checked and assessed in Gblocks Version 0.91b [39,40]. Phylogenetic analysis was performed in MEGAX software using UPGMA algorithm. Confidence of the clustering was executed in SEQBOOT Plants 2020, 9, 995 6 of 17 and followed then by running the data set for 500 replication times in each of the previous programs.
The bootstrap values were written on the nodes of generated phylogenetic trees. In addition, PCA biplot based on the morpho-agronomic data matrix was constructed by the multivariate analysis of PAST ver. 4.02 software [45].

Molecular Characterization and Genetic Relationships as Revealed by IRAP Markers
The IRAP fingerprinting profiles generated by the eleven primers targeting inter-retrotransposons in 12 alfalfa cultivars are shown in Figure 1. The polymorphism produced by the same 11 IRAP primers is summarized in Table 2. These primers generated a total of 97 amplicons; 85 of these were polymorphic (87.6%). The number of total bands ranged from 3 for primer IRAP-4351 to 21    The IRAP marker data for 12 alfalfa cultivars were used to produce a genetic distance tree based on Dice's genetic similarity matrix ( Figure 2). In this tree, the two Egyptian cultivars, EGY2-Nubaria1 and EGY1-Ismailia1, were clearly separated from the other cultivars, while the USA6-SW9720 and the AUS4-SuperFast cultivars were distinguished as two branches. The other two Australian cultivars were then differentiated from the five USA cultivars; AUS2-Siri Nafa and AUS3-Siriver clustered together and USA10-Super10 was distinct as one branch. USA1-Super-supreme and USA2-Nafa Extra clustered together and the other three cultivars clustered. USA1-Super-supreme and USA2-Nafa Extra clustered together and the other three formed another cluster. The similarity matrices among the 12 M. sativa cultivars as computed using Dice's coefficient based on IRAP molecular marker polymorphism are given in Supplementary Table S1.   The PCA analysis reflecting the strength of the IRAP markers to classify the examined cultivars in a PCA scatter plot by plotting of PC1 and PC2, clearly discriminated the two Egyptian cultivars ( Figure 3) and revealed the characteristic grouping of the EGY1-Ismailia1 and EGY2-Nubaria1 cultivars. In the same context, the Australian cultivars AUS2-Siri-Nafa, AUS3-Siriver and AUS4-SuperFast were grouped together, but USA10-Super10 was grouped with USA3-Grasis II. On the other hand, the rest of the American cultivars, USA1-Super-supreme, USA4-Cuf101, USA5-Supreme forager and USA6-SW9720, were scattered with some distance to each other.
Multivariate heatmap analysis was also used to construct a heatmap using the heatmap module of R software. The twelve cultivars clustered, as shown by the columns, into five clusters of two or three cultivars (Figure 4). The first cluster grouped the Egyptian cultivars together adjacent to USA10-Super10. The cultivars AUS4-SuperFast, USA2-Nafa Extra and the cultivars AUS2-Siri Nafa and AUS3-Siriver are differentiated as two neighboring pairs of cultivars. The other five American cultivars appeared as two neighboring clusters: one of USA1-Super-supreme and USA3-Grasis II and the other of USA4-Cuf101, USA5-Supreme forager and USA6-SW9720.  Multivariate heatmap analysis was also used to construct a heatmap using the heatmap module of R software. The twelve cultivars clustered, as shown by the columns, into five clusters of two or three cultivars (Figure 4). The first cluster grouped the Egyptian cultivars together adjacent to USA10-Super10. The cultivars AUS4-SuperFast, USA2-Nafa Extra and the cultivars AUS2-Siri Nafa and AUS3-Siriver are differentiated as two neighboring pairs of cultivars. The other five American cultivars appeared as two neighboring clusters: one of USA1-Super-supreme and USA3-Grasis II and the other of USA4-Cuf101, USA5-Supreme forager and USA6-SW9720.  Multivariate heatmap analysis was also used to construct a heatmap using the heatmap module of R software. The twelve cultivars clustered, as shown by the columns, into five clusters of two or three cultivars (Figure 4). The first cluster grouped the Egyptian cultivars together adjacent to USA10-Super10. The cultivars AUS4-SuperFast, USA2-Nafa Extra and the cultivars AUS2-Siri Nafa and AUS3-Siriver are differentiated as two neighboring pairs of cultivars. The other five American cultivars appeared as two neighboring clusters: one of USA1-Super-supreme and USA3-Grasis II and the other of USA4-Cuf101, USA5-Supreme forager and USA6-SW9720.

DNA Barcoding Loci of matK and trnH Sequencing
Two DNA barcoding loci matK and trnH from the chloroplast genome were compared for their amplification, sequencing, and successful use in measuring the genetic diversity and phylogeny among eight alfalfa cultivars marked by an asterisk (*) in Table 1. These include the two Egyptian cultivars, AUS2-Siri Nafa, and five USA cultivars (USA1-Super-supreme, USA4-Cuf101, USA7-SW9623, USA8-Magna901 and USA9-Perfect). PCR amplification products of matK and trnH regions showed 100% amplification success across all eight cultivars and agarose electrophoresis results demonstrated  Figure S1).
The produced DNA fragments recorded a size of a 700 bp for both matK and trnH. To ensure that the matK and trnH sequences, generated in the present study, were of M. sativa species, a BLAST function detected that all the sequences strongly matched Medicago sativa and Medicago truncatula matK and trnH sequences. Pairwise distances were obtained and assessed from the sequences of the conserved matK and trnH sequence. Additional information on the genetic variation estimates of DNA barcoding regions of matK and trnH loci in eight cultivars of M. sativa is given in Table 4, particularly the aligned length, undetermined characters, missing percent, variable sites and the proportion of variable sites, and parsimony informative sites. The evaluation of sequence estimates of DNA barcoding locus of matK locus is given in Supplementary Tables S2 and S3. A phylogenetic tree generated based on matK sequence variation using the UPGMA algorithm clearly discriminated between the eight alfalfa studied sequences ( Figure 5). To manifest the accuracy and efficacy of the generated tree, five NCBI-extracted matK sequences belonging to four M. sativa (NC_042841.1) and its subspecies x-varia (HM159582.1), glomerate (MK46595494) and caerulea (HM159580.1) as well as M. truncatula (NC_003119.1) were used as outgroups. The tree has two major clades: One includes the Egyptian cultivars EGY1-Ismailia1 and EGY2-Nubaria1 along with M. sativa NCBI-extracted matK and its subspecies members and USA9-Perfect. The second includes the five USA cultivars, USA4-Cuf101, USA7-SW9623, USA8-Magna901 and USA1-Super-supreme grouped with M. truncatula, which is used as an outgroup member. The bootstrap values are very high (98-100%), confirming the validity of the tree branching. A PCA scatter plot based on the of matK barcoding (Supplementary Figure S2) confirmed the classification of the examined cultivars as indicated in the phylogenetic tree generated by the matK locus sequence. This figure revealed the characteristic grouping of EGY1-Ismailia1 and EGY2-Nubaria1 cultivars and the distinction of AUS2-Siri Nafa and USA4-Cuf101. A multivariate heatmap analysis of the matK sequence generated using the heatmap module of R software (Supplementary Figure S3) also confirmed the distinction of the Egyptian cultivars EGY1-Ismailia1 and EGY2-Nubaria1 and AUS2-Siri Nafa from the USA cultivars.
On the other hand, similarity matrix revealed a variation between trnH-generated sequences among the M. sativa cultivars. A phylogenetic tree was constructed based on this variation in eight alfalfa cultivars using trnH of M. sativa (NC-042841.1; M. falcata (NC-032066.1) and M. truncatula (NC-003119.1) (Figure 6) as three outgroups.
The tree clearly isolated the three outgroups from the major clade of the eight alfalfa cultivars used for trnH barcoding. In this clade, AUS2-Siri Nafa and the two Egyptian cultivars sequences EGY1-Ismailia1 and EGY2-Nubaria1 were clearly isolated from the five American cultivars. A phylogenetic heatmap analysis of the trnH sequence generated using the heatmap module of R software (Supplementary Figure S4) also confirmed the distinction of the Egyptian cultivars EGY1-Ismailia1 and EGY2-Nubaria1 and AUS2-Siri Nafa from Australia.
Plants 2020, 9, x; doi: www.mdpi.com/journal/plants the phylogenetic tree generated by the matK locus sequence. This figure revealed the characteristic grouping of EGY1-Ismailia1 and EGY2-Nubaria1 cultivars and the distinction of AUS2-Siri Nafa and USA4-Cuf101. A multivariate heatmap analysis of the matK sequence generated using the heatmap module of R software (Supplementary Figure S3) also confirmed the distinction of the Egyptian cultivars EGY1-Ismailia1 and EGY2-Nubaria1 and AUS2-Siri Nafa from the USA cultivars.  The tree clearly isolated the three outgroups from the major clade of the eight alfalfa cultivars used for trnH barcoding. In this clade, AUS2-Siri Nafa and the two Egyptian cultivars sequences EGY1-Ismailia1 and EGY2-Nubaria1 were clearly isolated from the five American cultivars. A phylogenetic heatmap analysis of the trnH sequence generated using the heatmap module of R software (Supplementary Figure S4) also confirmed the distinction of the Egyptian cultivars EGY1-Ismailia1 and EGY2-Nubaria1 and AUS2-Siri Nafa from Australia. In addition to the genetic diversity among the alfalfa cultivars, the variation in 15 morpho-agronomic traits (Table 5), applicable only to eight M. sativa cultivars, was dissected. A distance cluster tree, illustrating the classification of M. sativa cultivars, was constructed based on the Euclidean distance of variation in the studied morpho-agronomic traits using the UPGMA algorithm (Figure 7). Table 5. List of 16 morpho-agronomic traits and their state in the eight alfalfa cultivars used for phylogeny of these cultivars using matK and trnH gene sequences. from the other alfalfa cultivars was most influenced by plant height (PH) and stem number (SN). The distribution of the American cultivars in the biplot was most influenced by the following traits: lateral leaflet width/length (LlW/L), stipule length (SL), lateral leaflet length (mm) (LtL), lateral leaflet width (mm) (LtW), lateral leaflet length (mm) (LtL), and peduncle length (PdL). On the other hand, the distinction of AUS2-Siri Nafa was mostly impacted by the terminal leaflet width (TlW) and terminal leaflet width/length (TlW/L).  The clustering of the eight cultivars confirmed the close genetic relationship between the two Egyptian M. sativa cultivars. However, AUS2-Siri Nafa was assigned close to these two cultivars, whereas the two cultivars USA1-Super-supreme and USA4-Cuf101 were clearly isolated from a cluster of the other three cultivars, namely, USA7-SW9623, USA8-Magna901 and USA9-Perfect. A PCA biplot, constructed by blotting the PCA 2 and PCA 4, showed the best illustration of the impact of the studied morpho-agronomic traits on the discrimination of the examined cultivars ( Figure 8). The affinity of the two Egyptian cultivars and their distinction from the other alfalfa cultivars was most influenced by plant height (PH) and stem number (SN). The distribution of the American cultivars in the biplot was most influenced by the following traits: lateral leaflet width/length (LlW/L), stipule length (SL), lateral leaflet length (mm) (LtL), lateral leaflet width (mm) (LtW), lateral leaflet length (mm) (LtL), and peduncle length (PdL). On the other hand, the distinction of AUS2-Siri Nafa was mostly impacted by the terminal leaflet width (TlW) and terminal leaflet width/length (TlW/L).

Discussion
In this study, IRAP markers and the chloroplast DNA matK and trnH barcodes were used to assess genetic diversity in some M. sativa cultivars currently cultivated in Egypt, USA, and Australia. Genetic diversity has been assessed using 97 markers produced by 11 IRAP primers; with 79.73% polymorphism and an average of 7.72 markers per primer although this number ranges between 3 for IRAP 4351 and 21 for IRAP 4340. Five primers (IRAP 4334, 4370, 4340, 4342, and 3471) produced 100% polymorphism indicating highly variable regions in the genome (Mansour 2008). IRAP fingerprinting also resulted in high polymorphism in different species and reflected intraspecific variations in Mastic tree [46]. The high percentage of polymorphism detected in the present study by IRAP may also indicate a high insertional activity for the assessed RTNs in the genome of M. sativa cultivars as reported for wheat [22][23][24]47].
The value of genetic diversity parameters revealed by IRAP markers was used to calculate the genetic diversity of 12 M. sativa cultivars using multivariate clustering, PCA, and heatmap analyses indicated that the two Egyptian cultivars and AUS2-Siri-Nafa are distinguished from other cultivars. Close affinity is also clear between the Australian cultivars to each other and the American cultivars to each other with some resemblance between some cultivars as indicated in Figures 2-4, in particular between USA10-Super10 and USA3-Grasis II; USA4-Cuf101 and USA5-Supreme forager, as indicated by cluster analysis (Figure 2); and between USA10-Super10 and USA3-Grasis II, as shown in PCA analysis ( Figure 3); between USA10-Super10 and EGY1-Ismailia1; and EGY2-Nubaria1 and AUS4-SuperFast and USA2-Nafa Extra (Figure 4). The differentiation of the studied M. sativa cultivars of different countries may be accounted for by former differences in climatic conditions, which might contribute to increasing the genetic variation between cultivars. The findings based on IRAP marker analyses might be attributable to the instability of RTN insertion events as well as to the environmental conditions [48,49].The high level of IRAP markers polymorphism could relate to the outcrossing and the tetraploid nature of alfalfa [20,50,51].
The present study revealed that matK is more variable than trnH in sequence variability among the tested Medicago cultivars. It has been reported that matK, due to its conservative mode of evolution, is widely used in many phylogenetic studies of flowering plants. It was used to distinguish medicinal plants and validate many herbal products [52,53]. In the current studies, eight alfalfa cultivars were differentiated based on matK sequence variation using five outgroup sequences ( Figure 5). The phylogenetic tree generated using five NCBI-extracted matK sequences representing four M. sativa subspecies and one of M. truncatula supported the outstanding finding of separating the two Egyptian cultivars EGY1-Ismailia1 and EGY2-Nubaria1 along with M. sativa NCBI-extracted matK of M. sativa and its subspecies members as well as USA9-Perfect which was not included in the IRAP experiment. However, EGY2-Nubaria1 was separated with the M. sativa NCBI-extracted matK sequence of the subspecies caerulea, indicating sequence homology between them. In the second clade of four American cultivars, USA4-Cuf101, USA7-SW9623, USA8-Magna901 and USA1-Super supreme grouped with M. truncatula and AUS2-Siri Nafa, and the latter two cultivars shared the two most similar matK sequences.
The phylogenetic tree constructed based on the trnH sequence showed close homology of the eight examined cultivars as all were clearly isolated from the three outgroups ( Figure 6). This may be due to the limited sequence variability in trnH sequence in the tested cultivars. In this clade, AUS2-Siri-Nafa and the Egyptian cultivars were clearly isolated from the five American ones in agreement with their clustering based on the morpho-agronomic traits. The phylogenetic heatmap analysis of the trnH sequence generated also confirmed the distinction of the Egyptian cultivars EGY1-Ismailia1 and EGY2-Nubaria1 and AUS2-Siri Nafa from Australia.
Despite criticisms of using morphological traits to identify plant cultivars or study their genetic variability, they have proven useful in evaluating genetic diversity in alfalfa [33]. The classification of eight cultivars in the current study confirmed the close genetic relationship between the two Egyptian cultivars and AUS2-Siri-Nafa, whereas the two cultivars USA1-Super supreme and USA4-Cuf101 were clearly isolated from a cluster of other three American cultivars: USA7-SW9623, USA8-Magna901 and USA9-Perfect. Morphological variability among accessions of 18 non-dormant alfalfa was strongly influenced by leaflet shape, stipule shape, the peduncle, and the petiole length ratio [12]. In the current study, the PCA biplot classification of the studied alfalfa cultivars confirmed the affinity between the two Egyptian cultivars, EGY1-Ismailia1 and EGY2-Nubaria1, and their distinction from the other cultivars, but also indicated resemblance to AUS2-Siri Nafa from Australia. The distinction of the two Egyptian cultivars EGY1-Ismailia1 and EGY2-Nubaria1 agrees with the clustering together of local accessions in Saudi Arabia and their differentiation from other origins. Clusters of local accessions at high similarity sometimes correlated with their collection site and morphological traits [16].
Morphological characters are the external expression of the genotype. The plant habit, stem, leaves, flower, fruit and seeds have been used for classification of alfalfa cultivars based on morphological variation [34,54]. Some of the examined cultivars were characterized by morphological traits [41]. The American cultivar USA8-Magna901, originating from DairyLand, America, has a high forage production, a fall dormancy rating of 9, and good growth in winter with a high resistance to nematodes [55]. USA4-Cuf101 is characterized with a high resistance to cold stress, Fusarium root rot, and root nodule nematodes. Hereby, USA4-Cuf101 is favored for its high forage production, a dormancy rating of 10, and dry climate resistance. Moreover, USA4-Cuf101 is characterized with a high tolerance to Fusarium and Pythophtora [55]. USA4-Cuf101 and USA8-Magna901 showed closer affinities to other American cultivars and a distant relationship to the Egyptian and Australian cultivars. Among the Australian cultivars, AUS2-Siri Nafa is characterized by drought tolerance and resistance to root diseases [55].

Conclusions
The presented results of genetic variation between the studied alfalfa cultivars, based on IRAP, DNA barcoding and morpho-agronomical traits, are considered as potentially useful sources of genetic information in developing future breeding program(s) and opening up new possibilities of producing M. sativa lines that have high forage quality, productivity and resistance to biotic and abiotic stress conditions. For these objectives, collection and characterization of genetic resources are required for the development of new cultivars.