Barberry, as a medicinal plant, has been known and used for a long time in Iran and many other ancient civilizations around the world (Zargari, 1990). This plant belongs to the Berberidaceae family which contains approximately 15 genera and 650 species found in temperate regions of the northern hemisphere (Bottini et al., 2002). Seedless barberry (Berberis vulgaris C. K. Schn. Var. asperma Don.) is one of the few unique crops grown only in Iran (Tehranifar, 2003). Due to salinity of water and soil, large cultivated areas in the eastern parts (32.5-34.5º N. Latitude) of Iran are not suitable for the growth of most crops, hence, in such areas especially during the last 20 years, the seedless barberry has been introduced as a major crop (Balandari and Kafi, 2001). Approximately 95% of the total cultivated areas and production of this plant in Iran is located in these regions (Handbook of Agriculture Statistics, 2004). Moreover, according to the capability of this compatible and tolerant crop to grow in mountain valleys and river sides, its important role in conservation of water, soil and vegetation is of considerable significance.
So far, all studies have been focused on medicinal properties of barberry, but the evaluation of variation and genetic structure of its populations in Iran has yet not been investigated.
Despite the fact that classical studies based on botanical and systematic principles have attracted much interest, there is still a great deal of debate and several questions that makes it necessary to implement new methods and devices to reveal a phylogenetic relationship between the Berberis and Mahonia genera. Several systematic studies on Berberis and Mahonia using chromosome number (Dermen, 1931), wood anatomy (Shen, 1954), floral anatomy (Trabayashi, 1978) and serology (Jensen, 1973) have reported them to belong to the same genus. On the other hand, embryological studies and morphological differences between these two plants, such as the existence of thorn and simple leaves in Berberis and their absence in Mahonia, represent them as two distinct genera (Ahrendt, 1961). Therefore, according to recent progress in the area of genome and molecular techniques, DNA fingerprinting together with the aforementioned information, can be a suitable method to identify this relationship. Kim and Jensen (1994), by preparing a chloroplast DNA (cpDNA) library of Mahonia have managed to obtain its exact mapping by means of restriction enzymes. The results are in agreement with previous chromosomal, morphological and serological data showing a close phylogenetic relationship between the two genera.
By using the AFLP technique, Bottini et al. (2002) have evaluated genetic variation of 13 Berberis species and the relationship within diploid and polyploid populations growing in southern Argentina and Chile. The dendrogram of DNA fingerprinting has shown that in general, populations of the same species form closely related groups with high coefficients of similarity. Furthermore, they have compared AFLP, morphological and seed protein data by means of the mantel correspondence test. The correlation between AFLP and morphological data is rather low and no significant correlation has been found between AFLP and seed protein data.
In this study, by using AFLP molecular markers, the genetic structure and variation of seedless and native Berberis populations growing in eastern Iran along with two ornamental Berberis species and Mahonia aquifolium could be assessed. Data were also compared with morphological traits to assess the identification key of the Berberidaceae family in Flora Iranica.
MATERIALS AND METHODS
Plant materials: Plant material used in this study consisted of 8 native populations of Berberis species collected from different locations of the Khorasan provinces in eastern Iran, namely: Shomali (north), Razavi and Jonubi (south), together with two species of ornamental Berberis and one sample of Mahonia aquifolium (Table 1). The plants were identified by the herbarium of the Research Center for Plant Science at Ferdowsi University of Mashhad. Selected morphological traits from the identification key in Flora Iranica (data not shown) were used for assessment of morphological variation in samples of the Berberidaceae family (Heidary, 2008).
DNA isolation: Total genomic DNA was isolated from young leaves using the cetyl trimethylammonium bromide (CTAB) method of Saghai-Maroof et al. (1984). The quantity and quality of DNA were evaluated by a UV- Spectrophotometer (JENWAY, UK).
AFLP procedure: In this study, AFLP protocol developed by Vos et al. (1995) was performed with minor modifications and double stranded adapters were ligated to the fragments. The EcoRI adaptor consisted of the combination of two primers: 5´-CTCGTAGACTGCGTACC-3´and 5´-AATTGGTACGCAGTCTAC-3´. Similarly, the Tru1I adaptor contained two primers: 5´-GACGATGAGTCCTGAG-3´ and 5´-TACTCAGGACTCAT-3´.
The digested and ligated template DNA was preamplified using EcoRI+1 (5´-GACTGCGTACCAATTCA-3´) and Tru1I +1 (5´-GATGAGTCCTGAGTAAC-3´) primers. AFLP fingerprints were generated using pair of EcoRI+3 and Tru1I+3. 40 primer combinations were tested from 5 EcoRI primers (EcoRI+AGG, EcoRI+AAC, EcoRI+AGC, EcoRI+ACT, EcoRI+ACG) and 8 Tru1I primers (Tru1I+CGG, Tru1I+CTC, Tru1I+CAG, Tru1I+CTG, Tru1I+CAA, Tru1I+CCT, Tru1I+ CGA, Tru1I+CCG). Four primer combinations producing strong and reproducible bands were selected and exploited to detect AFLP polymorphism among the various genotypes. The primers were EcoRI+ACG/Tru1I+CGG, EcoRI+ACG /Tru1I+CAG, Tru1I+CCT/EcoRI+AGC and Tru1I+CAA /EcoRI+ACT. The amplification products were separated on 6% polyacrylamide gels (acrylamide-bis-acrylamide [20:1], 7.5 M urea, 1X TBE buffer). Electrophoresis was performed for 2 h in 1X Tris-Borate ethylenediaminetetraacetic acid (TBE) at 1200 volt. Bands were visualized by silver nitrate staining (Sanguinetti et al., 1994).
AFLP data analysis: AFLP fragments (Fig. 1) were scored as either present (1) or absent (0) across all populations. Only distinct, well-resolved fragments were scored. Binary matrix was used to estimate genetic similarities between the pairs by employing the Dice index (Nei and Li, 1979). These similarity coefficients were used to construct a dendrogram using the unweighted pair group method with arithmetic averages (UPGMA) and employing the sequential, agglomerative, hierarchical, and nested clustering (SAHN) from the numerical taxonomy and multivariate analysis system (NTSYSPC), version 2.02 program (applied biostatistics) (Rohlf, 1990). Genetic coefficients and indices such as number of polymorphic loci, percentage of polymorphic loci, observed number of alleles (no), effective number of alleles (ne), Nei’s gene diversity (h), and Gst Factor were analyzed using the POPGENE program version 1.32 (Yeh et al., 1997). Principle coordinate analysis (PCoA) (Huff et al., 1993) was implemented by GenAlEx version 6.1 (Peakall and Smouse, 2007). Using the same software, an analysis of molecular variance (AMOVA) was performed to partition the total genetic variation within and among populations (Huff et al., 1993; Excoffier et al., 1992). Differences among populations were quantified using Wright’s inbreeding coefficient (Fst).
Fst = (HT-HS)/HT
Where, HS is mean heterozygosity within populations and HT is total heterozygosity between populations. Fixation index Fst and indices population specific Fst were calculated by Arlequin ver. 3.0 (Excoffier 2005). Finally, the estimated Nm of gene flow, as the number of migrants entering a population in each generation, was calculated according to Wright (1931).
Nm = (1-Fst)/4Fst.
Morphological data analysis: In addition, raw data from 39 morphological traits were analyzed by Statistica V5.5A, distance matrix and cluster analysis. PCoA analysis and the Mantel correspondence test were also performed with GenAlEx (Heidary, 2008; Peakall and Smouse, 2007).
A total of 223 AFLP bands consisting of 204 (>90%) polymorphic types were detected when 4 different primer combinations were tested.
Grouping analysis: The dendrogram generated from the AFLP results showed that the populations are divided into two main groups with a similarity coefficient of 0.48 that separated the two Mahonia and Berberis genera. Group 1 contains Mahonia aquifolium from the Berberidaceae family. Group 2 was formed by the populations of the Berberis genus that contains B. gagnepaini in subgroup 1, B. thunbergii in subgroup 2 and Berberis integerrima from population 7 in subgroup 3, with a similarity coefficient of 0.77. The populations of B. integerrima and B. vulgaris were very closely grouped and showed a similarity coefficient of 0.79-0.92 (Fig. 2).
Cluster analysis based on the UPGMA (Sneath and Sokal, 1973) generated from morphological data, showed two main groups, the Berberis genus and the Mahonia genus. Two ornamental species were also separated from other Berberis species in two distinct subgroups. The results of these three species were similar to AFLP cluster analysis. But other samples (B. integerrima and B. vulgaris) in morphological cluster against AFLP cluster analysis were not clustered in separated groups as well.
Principal coordinate analysis: A 2-dimensional graph plotted for grouping and establishing the relationship among these samples showed considerable distance between M. aquifolium, B. gagnepaini, B. thunbergii and samples belonging to Calat region (B. integerrima) from the other samples (Fig. 3).
Molecular analysis of the genetic structure of Berberis population: Gene variation (h), an index of variation magnitude, was 0.181 among the tested samples, whereas in each population the h value is 0.1 or less. The smallest h value (0.008) and polymorphism (2.96) were seen in cultivated barberry and populations of population 3, pop.4 and pop. 7 had the highest gene variation and polymorphism (Table 2).
Pair-wise comparisons of present populations in the Nei similarity matrix revealed their genetic similarity rate. The dendrogram of Figure 4 properly showed the relationship of populations and various species existing in eastern Iran. it showed that populations of B. integerrima are very similari to each other (>0.9 similarity coefficient). Seedless barberry (B. vulgaris) is more similar to B. integerrima than pop.7 and there is not high genetic distance between B. vulgaris and B. integerrima.
In order to study the structure of Berberis populations, Fst and Gst indexes were calculated. Gst was 0.5909 and the total Fst value was 0.4. The highest Fst value was observed in the cultivated barberry population, while the lowest one was monitored in pop. 3, pop. 4 and po. 7 (Table 3).
Analysis of molecular variance: The total amount of genetic variation detected has been partitioned into its components due to the subdivision between and among the populations. Based on the reasons mentioned above, this analysis was made only on pop.1 to pop.8. The results of AMOVA (Table 4) showed that a large and significant amount of genetic variation (40% of the total) was due to differences among populations and the other significant amount (60% of the total) was due to differences within populations. This result may be due to a slight difference in the Fst value of the two populations namely pop.5 (Fst = 0.39) and pop.6 (Fst = 0.38) from the total Fst (0.4) that has elevated part of the intra-population variation as compared to inter-population variation.
Mantel test: By means of the mantel correspondence test (Mantel, 1967), the dendrogram was constructed using AFLP data and was compared with another dendrogram previously obtained using morphological traits (Heidary, 2008). The correlation between all of samples was rather low, although significant (r = 0.47). In case of 30 sample belonging to eastern Iran there was no significant correlation (r = 0.13).
PCoA and cluster analysis and the genetic similarity based on Nei coefficient, revealed that Mahonia aquifolium lies in a completely distinct group with a long genetic distance from the other species. Although Mahonia genus according to morphological studies has recently been separated from Berberis, however their phylogenetic relationship is not yet clear. This study showed that Mahonia can be a distinct genus different from Berberis.
Despite the long geographical distance between seedless samples, they have very low h value (0.008). Despite a long period of cultivation of this crop in different orchards, due to vegetative propagation, they have not diverted much. Furthermore their near genetic distance with B. integerrima despite of the fact that they are two different species, is interesting. This finding may help us to find the origin of the seedless barberry.
Based on studies carried out in our herbarium we found that the, B. integerrima is a predominant species in eastern Iran. Furthermore, a considerable molecular variation within the B. integerima populations was in agreement with morphological variation (Balandari and Kafi, 2001). Although the populations namely, pop.1, pop.2, pop.3, pop.4, pop.5, pop.6 and pop.7 belong to the same species, geographical separation together with a high percentage of self-pollination has probably caused a great difference among the members of this species.
In this study, a high Gst and a high Fst values indicate that, individuals within a population are relatively similar but populations are significantly different (Wright 1978; Chai 1976; Nei 1975; Wright 1969). The high Fst values in wild Berberis populations indicate that they have a low gene diversion, with their gene frequencies is getting further reduced (Hamrick et al., 1991; Hamrick and Godt 1990). Genetic variation helps organisms overcome environmental changes. As a result of low heterozygosity reproduction and survival of organisms is reduced (Jones and Luchsinger, 1986). Species such as barberry that form sporadic small populations in some parts of Iran and show a high percent of self-pollination, are exposed to homozygocity and genetic erosion (Jones and Luchsinger, 1986). Assuming that under similar conditions, species with wide gene diversion that conserve gene frequency should have low Fst and the populations should be similar (Crow 1986), while calculation of Fst index in all populations and its comparison to Fixation index Fst (0.4) shows that intra-population variation is less than that of inter-population. The h value mentioned in table 2 also confirms these results.
Based on the results of the herbarium in the Research Center for Plant Sciences at Ferdowsi University of Mashhad, natural habitats of Berberis species are at risk of being converted to farmlands. For example, in the region which lies between Maravetape and Birjand that is nominated as natural habitat of Berberis khorasanica in eastern Iran (Rechinger, 1975), barberry is no longer found.
Results of the present study show that morphological markers have lower efficiency than AFLP markers for grouping and systematic studies of the Berberidaceae family. Although morphological markers could distinguish the samples at the genus level, but this was reduced considerably at the species level. The taxonomic treatment of species in the Berberis genus based on external morphology is still a matter of debate (Bottini et al., 2002). Morphological traits of the Berberidaceae family in the identification key of Flora Iranica are often vegetative characteristics which are under the influence of environment.
Clustering of the samples in pop. 7 is one of the major difference of AFLP and morphological markers in this study. In term of morphological traits, these samples are similar to other B. integerrima samples, eventhough they are grouped differently based on AFLP analysis. The morphological traits present in the identification key of Berberis not only have a lower efficiency compared to AFLP markers, but also are inappropriate for identification and classification of Berberis. Based on this study, it is suggested that molecular markers are more accurate and applicable for further analysis of the systematics of the Berberidaceae family than morphological markers. These results could be helpful in revision of traits present in the identification key described for the flora of each region.