Morphological and Isozyme Variation in Natural Populations of the Genus Medicago L. Prospected in Northern Algeria

As part of the evaluation and enhancement of genetic resources, morphological and isozyme variability within and among 169 accessions, representing 14 species of the genus Medicago L. collected in northern Algeria, was assessed using twelve quantitative traits and two enzymatic systems. Phenotype frequencies were scored in six enzyme zones to determine isozyme variability within and among populations. The data analysis resolved a high level of genetic diversity. Ten morphometric characteristics contributed to the discrimination of the species. The relationship between the collection site environment and phenotypic characteristics was also studied. Esterase (EST) enzyme system was more polymorphic than glutamate oxaloacetate transaminase (GOT) system. Were scored 2 zones with 10 bands and 21 phenotypes for GOT (glutamate oxaloacetate transaminase) and 4 zones with 22 bands and 71 phenotypes for EST (esterase) Polymorphism index and Jaccard’s genetic distances revealed the existence of a high genetic diversity within and among the studied populations. The annual species M. polymorpha presented an intraspecific polymorphism index of 0.57, which was higher than all other species indices. Clustering of the species based on isozyme markers was in agreement with taxonomic criteria and showed no significant correlation with morphological characteristics. Conservation programs should take into account the level of genetic diversity within and between populations revealed by isozyme markers.


Introduction
The Medicago genus (medics) is part of the Galegoid phylum that includes a number of important legume (Fabaceae) crops such as pea, chickpea, clover and lentil. It consists of approximately 87 annual and perennial species of herbs and shrubs widespread from the Mediterranean to central Asia (Small and Jomphe, 1989;Small, 2010). It includes the widely cultivated major forage crop and weedy species M. sativa L. (alfalfa, lucerne) and the legume model species M. truncatula Gaertn. (Bell et al., 2001). Medicago has been one of the most widely studied legumes because of numerous agriculturally important and domesticated species (Small, 2010). In addition to alfalfa, which is the most cultivated legume, other species are used as medicine, human food (honey production, sprouts), green manure, sources of industrial enzymes in biotechnology (Lewis et al., 2005), model genomic species (Cannon et al., 2006) and model systems for the study of nitrogen fixation (Bailly et al., 2007). The genus contains a great variety of morphological characteristics (spiny and spineless pods, presence of hairs, corolla colors etc.), of ploidy levels and basic chromosome numbers (Bena et al., 1998). Algeria represents one of the richest zones in genetic diversity and contains 22 annual and perennial Medicago species (Quezel and Santa, 1962). Morphological and agronomic characteristics such as seed and fruit size are often used for estimating genetic variation. Despite criticisms of using these markers, that are usually polygenic and influenced by environmental conditions (Kongkiatngam et al., 1995), they have proven useful in many studies on the genus Medicago L. (Al-Faifi et al., 2013;Bonnin et al., 1996;Bullita et al., 1994;Cherifi et al., 1993;El Hansali et al., 2007;Radwan et al., 1980;Tucak et al., 2008;ivković et al., 2012). Isozymes as codominant markers provide a useful clue for the estimation of population structures and phylogenetic relationship (Ferguson and Robertson, 1996;Malaviya et al., 2008) because zymographic patterns reflect directly a particular gene system. It has been well demonstrated that genetic diversity is correlated with genetic distance as measured by allozyme variation (Brown and Weir, 1983). Nevertheless, only a limited number of enzymes can be detected in electrophoresis, which limits their usefulness in estimation of genotypes variability ( ivković et al., 2012). However, isozymes have been extensively used in the studies of Medicago genetic variation (Bullitta et al., 1994; were scored: pod size: length (LP), width (WP); number of seeds per pod (SP); seed size: length (LS), width (WS); weight of 30 pods (WTP); weight of a thousand seeds (WTS) and the ratio weight of seeds/weight of pods (RW); seedling traits: Emergence date of cotyledons (DCOT), length of the first leaf (LFL), breadth of the first leaf (BFL) and Area of the first leaf (AFL=LFL*BFL). DCOT was measured from the moment the seedlings were placed under light. LFL, BFL and AFL were measured after six days of germination. Each trait was scored as a mean value of ten randomly chosen individuals for every accession.

Isozyme analysis
A total of 169 accessions represented by 14 species were assessed using two enzymatic systems: Esterase (EST, E.C. 3.1.1.-) and glutamate oxaloacetate transaminase (GOT, E.C. 2.6.1.1) (Weeden and Wendel, 1990). Isozymes were extracted from seedling leaves using the procedure of Bendiab et al. (1993). Polyacrylamide gel electrophoresis (PAGE) was performed according to Singh et al. (1991) on 10 % polyacrylamide gels for the analysis of EST, while GOT system required 7% polyacrylamide gels (Bendiab et al., 1993). The analysis was carried out at 40 mA for 3 hours at 4 °C. Gels were stained according to the methods given by Bendiab et al. (1993).

Data analysis
The morphological characteristics were analyzed using the principal component analysis (PCA). Mean, standard deviation (s.d.) and coefficient of variation (CV) of the morphological traits for the 13 studied species were computed. The correlation between morphological variability and bioclimate was examined by another PCA including the most important ecological parameters (Alt., Pm, Tm, TM and Q2). Phenetic relationships among species were evaluated by UPGMA (Unweighted Pair-Group Method using Arithmetic Averages) cluster analysis based on Euclidean distances. All analyses based on morphological markers were performed using Statistica version 6.0 program. For each enzyme system, Phenotypic polymorphism for each isozyme locus and species was estimated using the Pj index utilized by Kahler et al. (1980), Al Mouemar and Gasquez (1983), Warwick and Black (1986) and Bendiab et al. (1993): Pj = Pi (1 -Pi) = 1 -Pi 2 where Pi is the frequency of the ith phenotype and n is the number of phenotypes observed per locus for each species. The weighted average amount of phenotypic polymorphism over all observed enzyme zones is given by where Nj is the total number of phenotypes observed (collection wide) per jth locus for k loci. To estimate the isoenzymatic variation among the species UPGMA dendrogram was constructed based on Jaccard's index (Jaccard, 1908) using XLSTAT version 2014.1.09 software package. The reliability of the generated dendrogram was also tested using SPSS version 20.0 and Statistica version 6.0 Software programs. The correlation between the morphological and isozyme distances was tested with Mantel Z-statistic (Mantel, 1967). Chandra et al., 2011;Chérifi, 1996;Haddioui et al., 2012;Jenczewski et al., 1999aJenczewski et al., , 1999b.
Medicago genus is distributed worldwide and grown in highly contrasting environments. This extensive geographical adaptation promotes genetic variation and gives us the opportunity to use diverse gene pools. In this investigation, it was aimed to assess genetic diversity within and among natural populations of the genus Medicago L. collected in Northern Algeria at morphological and enzymatic levels. This will be helpful in directing conservation and breeding programs for the improvement of Medicago species.

Plant materials
The study was conducted on 169 accessions representing 14 species of the genus Medicago L.: M. truncatula (29 acc), M. polymorpha var. polymorpha (28 acc) and var. vulgaris sativa (1 acc), M. arborea (1 acc) and M. arabica (1 acc). These species were collected from 47 different geographic Algerian sites from 15 May to 15 July 2013. The collecting sites extended from the coastal region to the Atlas Tellian mountains, including the coastal plains and the steppic highlands, along a gradient of increasing aridity from North to South. The studied populations cover a wide range of bioclimatic stages ranging from the humid to the arid in Northern Algeria (Fig. 1). The prospected sites were characterized by their altitude (Alt), the classical Emberger's coefficient (Q2), the average annual rainfall (Pm), the average of the minimum temperature of the coldest month (Tm) and the average of the maximum temperature of the hottest month (TM) ( Table 1). Mature seeds of each individual were extracted from pods, scarified, then sterilized with 3% (v/v) sodium hypochlorite solution. Sterile grains were germinated in the dark for 2 days in petri dishes containing 1% water-agar (w/v) at room temperature, then kept under normal daylight for 6 days.

Morphological analysis
Morphological measurements were implemented on only 13 species of the collection. For each accession, twelve quantitative traits related to pod, seed and seedling described by Issolah and  * The Mediterranean bioclimatic coefficient of Emberger, Q2, is calculated according to Stewart (1974)

Morphological markers Morphometric characteristics analysis
Mean, range, s.d. and CV values of the morphological characteristics studied are shown in Table 2. Area of the first leaf (AFL) was found to be the most variable characteristic with CV of 80.79 %. Forty-five accessions showed AFL greater than the overall mean. It ranged from 6.17 mm 2 to 94.55 mm 2 with an average of 27.19 mm 2 across all species. Followed by SP with CV value of 5.61 %, whence the lowest SP was 1, while the highest was 28. Seed size characteristics (LS and WS), DCOT and BFL displayed the lowest CV values with 0.18, 0.06, 0.20 and 0.33, respectively. Fig. 2a shows the principal component analysis performed on the twelve quantitative traits analyzed. According to Kaiser Standard, we elected the first three axes, which accounted for 76.47 % of the total variation with 50.84 %, 16.68 % and 8.94 % for axis 1, 2 and 3, respectively. The component values of each variable were projected in the two first principal components (PC1 and PC2). The characteristics were distributed along PC2 and in the negative values of PC1, except RW, which was negatively correlated with all other characteristics (Table 3). The traits: LP, WP, WTP, LS, WS, LFL, BFL and AFL displayed a relative correlation between them (Table 3) and showed the highest contribution to PC1 which differentiates the majority of species. PC2 was influenced essentially by SP characteristic. WTS was misrepresented in the plan 1-2, but contributed highly to the construction of PC3.

Correlation between morphological characteristics and ecological parameters
In order to ascertain the relationship between morphological traits studied and bioclimate, another PCA (Fig. 2b) was performed taking into account both morphometric characteristics and the five ecological parameters (Alt, Pm, Tm, TM and Q2). The first four axes explained 74.22% of the total variation (36.40%, 19.08%, 11.74% and 6.99%, respectively). All the characteristics formed a group strongly contributed to the formation of PC1, excluding SP and WTS that showed high loadings in relation to PC3 and PC4, respectively, whereas climatic parameters displayed another group along PC2. This distribution of variables indicates a very low correlation between morphological characteristics and bioclimate. Whence, the closest relationship was found between TM and WTS with r = 0.26 and the strongest negative correlation was shown between Alt and WP with r = -0.19 when p < 0.05 (data not shown).

Cluster analysis based on morphological characteristics
Euclidean distances values based on the twelve quantitative traits were represented in a UPGMA phenogram (Fig. 3). The topology of the phenogram subdivided the species at a distance level d = 33.09 into two major clusters. Cluster I included six species with highest values of morphological characteristics (M. ciliaris, M. rigidula, M. intertexta, M. murex, M. italica and M. arborea) and was especially characterized by the highest WTS. The second cluster is subdivided into three main groups at a high distance level d = 24.17. All species presented intermediate

Isozyme markers
The revealed zymograms displayed a number of variable bands according to the system and analysed species. Examples of zymograms for the two enzymatic systems studied (EST and GOT) and their diagrammatical representations are shown in Fig. 4. A total of six isozyme loci with 32 bands were detected and 92 isoenzyme phenotypes were recorded for the two enzymes analyzed. Frequencies of isoenzyme phenotypes are given in Table 4.

Phenotype frequencies
Ten bands were scored based on the GOT enzyme system, attributed to 2 loci with 9 and 12 types of zymograms for Got-1 and Got-2, respectively. The number of bands varies from 1 to 3 at the 2 loci (Fig. 4b). The Got-1 isozyme type A, which is characterized by only one allozyme (band 4) was the most common and present in all the fourteen species analyzed with 0.77 frequency (Table 4). All other isozyme types of this locus were considered to be rare. At the Got-2 locus, the most frequent isozyme type was Got-2A (encoding the allozyme 9) with 0.36 frequency. This one was absent in five species but present in all accessions of M. truncatula, M. littoralis and M. rigidula (1.00), followed by isozyme types D and F with frequencies 0.17 and 0.14, respectively. The lowest frequency (0.01) was observed for the isozyme types C, I, K and L, which were only present in one accession of M. polymorpha, variaties M. polymorpha var. vulgaris and M. ciliaris, respectively.
Four zones of activity considered as product of four EST loci were observed among all 169 accessions. In total twenty-two EST bands attributed to 6, 20, 23 and 22 isozyme types for Est-1, Est-2, Est-3 and Est-4, respectively were recorded. All species represented by several accessions showed intra-specific variation for esterase system. The Est-1 locus was only observed in the two M. polymorpha varieties and M. intertexta. A and B phenotypes represented the two varieties: M. polymorpha and M. polymorpha var. vulgaris with frequencies 0.09 and 0.11, respectively. M. intertexta was characterized by the presence of the isozyme type C, which was also present in M. polymorpha var. polymorpha at a low frequency. The Est-1 phenotype F was very rare; hence, it was present in uniquely one accession of M. polymorpha. The loci Est-2, Est-3 and Est-4 were present across the collection with a high phenotypes number and variability. The isozyme types: C of Est-2, C of Est-3 and F of Est-4 were the most frequent with 0.15, 0.21 and 0.27 frequencies, respectively. The Est-4 phenotype (F) was characteristic of M. polymorpha and M. polymorpha var. vulgaris and it was present with high frequencies: 0.68 and 0.96 in the two varieties, respectively. A considerable number of rare and species-specific phenotypes with relatively low frequencies were recorded in these three loci (Table 4); as an example, phenotypes P, E and N of Est-2, Est-3 and Est-4 respectively, were only observed in two species, M. ciliaris and M. intertexta, hence, the phenotypes E and N exhibited two bands (11 and 22, respectively), which were species-specific to these two species.

Phenotypic polymorphism
The amounts of phenotypic polymorphism for each isozyme locus in each species (Pj) are reported in Table 5. The two GOT enzyme system loci (Got-1 and Got-2) were monomorphic in some species and polymorphic in others, presenting average phenotypic polymorphism values ( ) of 0.18 and 0.21 respectively. However, these loci were highly polymorphic within the two varieties M. polymorpha and M. polymorpha var. vulgaris followed by M. minima with lower values of phenotypic polymorphism (Pj), but considered monomorphic in the four species M. truncatula, M. orbicularis, M. littoralis and M. intertexta with values of phenotypic polymorphism (Pj) ranging from 0.00 to 0.10. The Est-1 isozyme had the lowest polymorphism value because it was present only in the two M. polymorpha varieties, but with a high level of genetic variation in each. The highest polymorphism indices ( ) were recorded in the three EST loci: Est-2, Est-3 and Est-4 with values of 0.39, 0.44 and 0.38 respectively, since all species were polymorphic for these EST loci with relatively high levels of variation, with Pj values ranging from 0.07 to 0.75 (Table 5). Based on the polymorphism values for each locus, differences between species could be observed. Generally, all the species displayed more or less important genetic diversity. The two varieties of M. polymorpha species showed the highest intraspecific variation across all the loci with of 0.57 for each, followed by M. italica and M. ciliaris with of 0.44 and 0.41, respectively. The three species M. sativa, M. arborea and M. arabica disclosed a null phenotypic polymorphism amount ( =0.00) because they were represented by only one accession in the study.

Cluster analysis based on isozyme variability
Interesting information in interspecific variability is obtained with cluster analysis. The UPGMA dendrogram based on distances matrix of the species wise pooled zymogram (Fig. 5) agrees with taxonomic criteria and showed two major clusters. The first one was composed of two species belonging to the subsection Leptospireae of the section Spirocarpos, namely M. polymorpha and M. arabica. The second cluster was the largest and embodied the remaining species belonging to five sections namely Spirocarpos, Orbiculares, Medicago, Lupularia and Dendrotelis. This cluster is subdivided at high close distances into four sub-clusters (A, B, C and D) each having three species only (Fig. 5) Finally, Mantel Z-statistical analysis based on Pearson correlation showed no significant correlation (r = 0.092, p = 0.395) between morphological and isozyme distance matrices.

Discussions
In this screening of genetic divergence in natural populations of the genus Medicago L., morphological and isozyme markers were compared for fourteen species represented by 169 accessions collected in wide range of climatic stages in Northern Algeria. The importance of morphological variation among the examined species is demonstrated by the PCA analysis on twelve quantitative traits and the topology of the dendrogram based on Euclidean distance that showed the discrimination of the species under study.
The majority of morphological characteristics analyzed in this study were important in the differentiation of the species. Pod size (LP and WP), WTP, seed size (LS and WS) and seedling traits (DCOT, LFL, BFL and AFL) participated strongly to the discrimination of all the species into two mean groups: a group included species with high measured values of morphological characteristics (M. 0.06 0.14 -   The importance of these traits is evidenced by its high loadings in the first principal component. Seedling traits: LFL, BFL and AFL had large genetic correlations among them (Table 2); this finding was also observed by Bonnin et al. (1996) when analyzing Medicago truncatula population structure by quantitative genetic markers. The characteristic SP which contributed strongly to PC2 differentiates the two species (M. orbicularis and M. lupulina) from one another and from the other species. Hence, it is known that M. orbicularis is characterized by a high number of seeds per pod, while M. lupulina is characterized by only one seed per pod (Lesins and Lesins, 1979). Another characteristic with great contribution to PC3 was WTS, it allowed the ingathering of species with high seed size, e.g. M. ciliaris, M. intertexta, M. murex and M. rigidula. Lesins and Lesins (1979) noted the relationship between these traits and reported that the weight of 1,000 seeds give a priori a fair idea about the size of the seeds.
The even large differentiation among and within these Algerian wild pool for quantitative characteristics is probably due to the vast geographic area that is represented, and thus the large heterogeneity among habitats encountered. Nonetheless, this hypothesis is rejected by our finding about the relationship between environment and phenotypic features, whence no significant correlation was appreciated. Various studies on the genus Medicago L. revealed positive correlation between morphological traits and climatic factors (Abdelguerfi and Laouar, 1999;Al-Faifi et al., 2013;Zivković et al., 2012). Prosperi et al. (2006) when evaluating morphologic and agronomic diversity of wild genetic resources of Medicago sativa L. in Spain found low significant correlations between the two sets of variables (morphology and bioclimate) and noticed that the environmental characteristics that influence more the phenotypic traits of the natural populations studied were the altitude, summer rainfall and latitude.
Because morphological traits only represent a part of the genetic diversity, and because the organization of this diversity is highly subject to natural selection and environmental factors, knowledge of the evolution and dynamics of populations requires analysis of more neutral markers such as isozymes that should help to specify the genetic structure of populations. The two enzymatic systems (EST and GOT) revealed high genetic diversity within and among the species studied, hence, six loci with 32 bands and 92 phenotypes were detected.
This result is in agreement with findings of many enzymatic variation studies on Medicago populations (Bullitta et al., 1994;Chandra et al., 2011;Cherifi, 1996;Haddioui et al., 2012;Jenczewski et al., 1999aJenczewski et al., , 1999b. This important enzymatic variability could be the result of the fact that the majority of these species are annual reputed to be autogamous (Hamrick and Godt, 1990).
EST are a complex and heterogeneous enzyme group with multiple substrate specificity (Scandalios, 1969) and one of the enzymatic systems with a high polymorphism in plants (Weeden and Wendel, 1990). In the current study, the estimate showed highest variability for EST isozymes, recording four loci with 71 types and all 22 bands contributed this difference. A lower genetic diversity was detected based on GOT enzymatic system, thus, two loci and 10 bands attributed to 21 phenotypes were scored across the collection. However, in addition to EST enzyme, high intra-specific variation was observed in the two varieties of M. polymorpha for GOT enzyme too. Similar results were found by Bullitta et al. (1994), Hannachi (1998), Paredes et al. (2002), suggesting that the high levels of autogamy in this species result in a strong pattern of among population genetic differentiation.
The topology of the dendrogram based on Jaccard's index generated from the isozyme binary data matrix was logical and the mean Jaccard's indices of the fourteen species ranged from 0.18 to 0.59 with an average of 0.37. This wide range of similarity indices reflects a high degree of polymorphism at isozyme level. This type of clustering was used also by Chandra et al. (2011) who estimated a high level of similarity among accessions and species of the genus Medicago L. The most similar species were M. truncatula and M. littoralis. This is in agreement with taxonomic criteria; hence, this two species belong to the same subsection (Pachyspireae). Valizadeh et al. (1996) when estimated genetic distances among nine species of Medicago based on RFLP markers, showed that M. truncatula and M. littoralis were the most similar, that makes isozyme markers as important as molecular ones. Strong similarity was observed also between the two M. polymorpha varieties (polymorpha and vulgaris) which were characterized by the presence of band 4 and band 5, respectively, based on EST enzyme system. M. ciliaris and M. intertexta showed also a high similarity between them, since belonging to the subsection Intertextae (Small and Jomphe, 1989). The close isozyme relationship among these two species was already elucidated (Cherifi, 1996).
The Mantel test (r = 0.092, p = 0.395) indicated no significant correlation between the matrix based on morphological characteristics and that derived from the enzymatic analysis. Therefore, clustering of accessions based on morphological characteristics was inconsistent with that derived from the isozyme markers analysis. The result suggests that the two marker systems give different estimates of genetic relations among populations. A large number of studies have been published that detail the comparative divergence of phenotypic and allozymic characteristics (Bonnin et al., 1996;Jenczewski et al., 1999a;Long and Singh, 1995;Podolsky and Holtsford, 1995;Spitze, 1993;Yang et al., 1996). Most of these have reported stronger differences with analyses of quantitative traits data than with electrophoretic ones and have suggested that natural selection acting upon the quantitative traits is responsible for the different patterns of differentiation revealed by neutral and quantitative traits. 93 In conclusion, the present study gives an insight into the rich variability and relationships present within and between Algerian Medicago species at morphological and enzymatic levels. This may allow greater understanding of the evolution of morphological and biochemical characteristics in this genus. In addition, the studied accessions can prove to be a good genetic source for further improvement programs. Moreover, certain species were having specific zymograms, hence, could serve as markers for species identification. In order to define more the genetic diversity of this Medicago collection, it is necessary to include protein and molecular analysis to have a complete understanding of the starting material for breeding purposes.